Sei sulla pagina 1di 222
IWR. THE DETERMINATION OF ORBITS A. D. Dubyago Translated from the Russian by R. D. Burke G. Gordon LN. Rowell FT. Smith The RAND Corporation THE MACMILLAN COMPANY, NEW YORK, 1961 © The RAND Corporation 1961 All rights reserved no part of this book ma reproduced in ony farm without potesce te writing trom the publisher, excopt by @ reviewer who wishes to quote brief possages in connee magazine or newspaper, First Printing Librory of Congress catalog card number: 61-13309) The Macmilton Company, New York Brett-Mocmillan Ltd., Galt, Ontario Printed in the United States of America TRANSLATORS’ PREFACE ‘This text is concerned primarily with classical methods of de- termining heliocentric orbits of celestial bodies. The Russian edi- tion was published in 1949, making it much more recent than simi- Jar books in English on the subject, all of which are currently out ‘of print. Tho author clearly explains each method covered, and he includes convenient summaries of formulas and a number of exam- ples worked out in detail. In addition, the pitfalls associated with each method are discussed. Dr. Dubyago presents some original work, particularly in Chapter 12, as well as some Russian varia- tions of the classical methods of orbit determination. Since this book was published, a number of man-made objects havo been placed in orbits around the earth and soveral in orbits around the sun, As a result of these events many people who formerly had little interost in celestial mochanics have become interested in orbit-determination techniques. This text provides much useful introductory background material that is pertinent to the determination of orbits of these man-made objects. Such topics as astronomical coordinate systems, the treatment of observations, the determination of preliminary orbits, the numerical calculation of perturbations, and the application of differential correction tech- niques to the determination of definitive orbits are covered. Cowell's and Encke's methods of integrating equations of motion have beon used in several machine computation programs for cal- culating space vehicle orbits. Differential correction techniques are useful in the synthesis of guidance systems and in performing error analyses as well as for orbit correction. The determination of the orbit of @ ballistic racket, except for the possible use of radar observations, differs little in principle from the determination of the orbit of a celestial hody. ‘The Russian text has been followed as closely as was consist- ent with good translation practices. The only modifications are the replacement of the spelling of trigonometric abbreviations and the abbreviation for logarithm, by those more consistent with American usage, In addition, commas indicating decimals have “i Tronslotors' Preface been replaced by decimal points, Several typographical errors here and thoro have been tacitly corrected. The tanslators have omitted certain tables that appear in the Russian text because equivalent or better tables are available in The American Ephemeris and Nautical Almanac ot in the Planetary Coordinates, 1960-1980. Tho tables omitted are numbered I through XIII, and XXI through XXV in the Russian version. A bibliography of references published during the last ton yoars: has been apponded to the original list of references. Dr. Alexander Dmitriyevich Dubyago was born in 1903, and holds a doctorate in physico-mathematical sciences. He was a member of the Kazan State University from 1941 to 1950, and was & professor there from 1944 to 1950. In 1943 Dr. Dubyago became a member of the Institute of Theoretical Geophysics, USSR Academy of Sciences. From 1948 to 1948, he was also « member of the Kazan Physico-Technical Institute, an affiliate of the USSR Academy of Sciences, and since 1948 has been a member of the Commission for Minor Planets, Comets, and Satellites of the Inter national Astronomical Union. In 1950 he joined the Engelhardt Astronomical Observatory of the Kazan State University. ‘The most significant of the problems Dr. Dubyago has worked on are: the movement, structure, and disintegration of comets; the estima tion of accuracy of observations with variometers; the question of the integration of gradients of gravity; the secular acceleration of comets; the determination of orbits; the structure of comet nuclei; ‘and the formation of meteor streams. The translators wish to express their gratitude to Dr. Alexander Pogo, of the Mt. Wilson and Palomar Observatories, for his careful editing of the translation. Dr. Pogo road tho ontire manuscript, collated it with the Russian toxt, and made many recommendations which have greatly improved the quality of the translation, The translators also wish to thank Me, Neil Horgan, who did the preliminary editing on five of the twelve chapters, and Mrs. Marie Saunders, who typed and retyped the manuscript many times during the process of its evolution to final form. This translation was prepared as part of the continuing program fof research undertaken for the U.S. Air Force hy The RAND Corporation. R. 0. BURKE G. GoRDON LN. ROWELL, FLT. SMITH PREFACE In this book I have set as the goal the presentation of guiding principles for the solution of the fundamental problem of theoreti cal astronomy, that is, the problem of determining orbits of coles- tial bodies (minor planets, comets, and meteors) revolving around the sun. For the determination of preliminary orbits many methods have been suggested, hut in practical applications to the ever increas- ing number of minor planets and comets, only a few justifiable methods have gradually crystallized igto complete computational schemes froquently used by astronomers. The other methods, how- fever interesting some of them may be from a theoretical point of View, remain almost without application. For the determination of orbits from three observations, it is almost always possible to confine oneself to the classical methods of Gauss and Olbers. It appeared natural, therefore, to present these methods in thoir moder form characterized by the introduction of rectangular equa- torial coordinates and suitable for calculation with an arithmometer. Only in this form is the whole process of solution of the problem clearly revealed to the student; moreover, this form of presentation results in certain advantages of a purely practical nature, for ex- ample, in relation to the calculation of parallax. But [do not wish to imply that it is necessary completely to abandon logarithms for calculating orbits. It should be added that some experienced computers find it less fatiguing to use logarithms than a calculating machine. Often, in performing parallel calcula: tions, I have found that, for a parabolic orbit, the calculations are faster with logarithms than with an arithmomoter. Therefore, 1 felt that it was necessary both to set forth the logarithmic form of the method of Olbers, and generally to devote that attention to the de- termination of a parabolic orbit which this problem deserves be- cause of its importance. On the contrary, for an elliptical orbit the calculations are faster on the whole with an arithmometer than with logarithms. Being guided by this, I have found it possible to omit a presentation of the method of Gauss in the well-known form vii Profoco given it by Encke, since at prosent it has only historical signifi- ennee. On the other hand, it occurred to me as expedient to dwell fon the methods for the calculation of orbits in the so-called ex- coptional cases when an elliptical orbit cannot be calculated from three observations, or when a parabolic orbit should not be te- termined by the method of Olbers. In the same way, I have singled Out a chapter for determining an orbit by two observations, and in connection with this the problem of identifieation of minor planets and of comets is considered. Omitted in its turn is the theorem of Lambert which gives the connection between two radius vectors of 8 celestial body, their closing chord, and the somimajor axis of the orbit, on the one hand, and the period of time, on the other. Even Gauss thought that the application of this theorem does not simplify the solution of the problem of determining orbits, and that romains correct up to now, if we do not consider certain (suffi- ciently rare) cases of calculating orbits of long-period comets. The caleulation of perturbations with the help of numerical ine tegration and the obtaining of improved and dofinitive orbits are considered in the second part of the book. Here, in particular, it is assumed that the reader is familiar with numerical methods of integrating differential equations and with the method of least squares. ‘The prosentation of these mathematical problems could not he included in the plan of my book, Among the methods devised for calculating perturbed motion and the methods of correcting pre- liminary orbits, I have considered only a few that have really stood the test of practice. T have considered it my duty to be careful in the choice of ma- terial, selecting only that which in the final result has essential value for the calculation of orbits, remembering in this connection that the study of problems that have no useful applications is al- ways burdensome if one is concerned with the solution of practical Problems; the determination of orbits is one of them, Mastering the technique of calculating orbits is not easy and requires time. Apparently this is so hecause beginners who are approaching this subject are not yet skilled in the production of long astronomical calculations, whose successful execution re= quires patience and work of a systematic character. In time, one acquires systematic habits, computer skills, and an ability for presenting clearly the geomotrical side of the separate stages of the determination of an orbit; this gaometrical insight permits one to compute more confidently and to avoid gross miscalculations. It is also necessary to prosent the calculations in a clear and Prefece ie convonient form (howover, pedanity inthis connoction is uncalled tor), Every computer, having noquited experience, welects scheme tn keeping with his own tastes. and habits, and does not rely ene tirely on another's instructions. However, for tho tnnofit of those who bogin to study theoretical astronomy, Tflt it mi duty to give 2 suffciont number of summaries of working formulas and of numer: Teal examples. Certain examples (on the teatment and preparation of observations, on tho calculation of definitive orbit) are given Jn great dotal, and thoy concern matters which are not as sipte 25 thowe found'in the majority of existing manuals. The concluding chapter 1s devoted to the determination of the cxbits of moteors. It is Included in accordance with the wishes of the reviewer, Professor Ne, Melson. Tam taking this opportanity to express my gratitude to hin for his valuable advice with respect imany sections of my werk. ve Aan incrotuction, a brie essay on the history ofthe study of the determination of orbits is offered. ‘Tho tablos appended al the en anawor almost all needs which arise during the calculation of orbits and ephemerides nceording to the mothods presented here. IC'is hoped. that this. book will prove useful not only to the students Studying theoretical astronomy ae the State universities, ut also o astronomer specialists who, in particular, may notice that the method ven by me for corectng nestly parabolic ebits, when rectangular coordinates are sed, is new. A.D, DUBYAGO KAZAN STATE UNIVERSITY octoseR, 1948 CHAPTER TWO; THE PROBLEM OF TWO BODIES ....++++ CHAPTER THREE: GEOCENTRIC MOTION «+... TABLE OF CONTENTS TRANSLATORS! PREFACE sseesessesseeseesees PREFACE. CHAPTER ONE: THE PROBLEM OF THE DETERMINATION OF ORBITS AND ITS HISTORY ese eseeoe 1. The Orbits of Minor Planets, Comets, ond Meteors... 2, The Pret Pretontonlan Attempts ot the Determinationet Comet fy Oriltsccseceseeseasserceeeseesess 4. The Epoch of Newton ond Halley.sevesee 2. The Wort of Evler, Lambert, Logrange, ond Laplace scscsss+e 13. The Methods of Olbere ond Gauss sous &. Sebsaquont Development ofthe Thooty of Orbit Determinations» 1. The Low of Universol Grovitationvevsevee+s &, The Equotions of Relative Motion 8 The Invegralt of Arsar ond Kepler's First Low. 10. The Integral of Vie Vivo and Kepler's Second Low ie Thitd Lew and Gaves's Conator mination af the Farm of an Orbit vss fon af am Orbit in Spocesseeseeeeieese 1A. The Elements of on Orbit pees 15: Motion Along on Ellipse ond o Circle TE. Motion Along o Pordbole srsevssseseeceesonses 17, Motion Along o Hyperbe 18, The Determination of the Heliocentse Coordinot 19, The Connection Between Heliccenire and Geocentic Coordinates 20. Concerning, the Influence of Precession on the Elements of on ‘Orbin and the Vector Elements. 21 The Celculotion at Hellocenteie Coordinates or Eilipte ond Circular Orbits «+++ 22, The Colevletlon af Hellocentie Coordinotes fer @ Porbolie Orbit 25. Calculotion of Heliacentrie Coardinotes. far Neorly Porsbotte Chita ssoscssseseeneeeses Pa 24, Finding, Seperate Geacentrie Positions and the Calculation of Ephemerides vesesreeeessee= 25, Obtaining on Ephemerie by the Mathod of Numerical Integration += 36, The Apparent Motion of Minar Planets gad of Comets vos sse+s00 Toble of Contents Table of Contents xi (CHAPTER FOUR: THE REDUCTION OF THE OBSERVED POSITIONS OF m2 PTER EIGHT: THE DETERMINATION OF A PARABOLIC ORBIT .+~ MINOR PLANETS AND COMETS « COMPARISON WITH EPHEMERIS « caarre ona Orit sevsees 202 Nonwae Pusces ieee tee eee 57 Geral Remo Abou the Conptaion of Come OF 2 SS eimeolten aston cscs : i oeeeeeeafe nan cree ea ace ae fee genera eae Le Their Positions..o+ 0020+ tees . 36 40. The Determination of Geocentrie Distances ssres arity 28, The Moments of Observations. ateaee eee 8 2a 29. Observed Coordinates ond Their Corrections for Precession, oe 218 Nctton, sod Silay Abenon, ad Ae or he tapes Peston nd ie lgrevense : Ingen The Resosenerion othe Mite Postion ofthe Coparzon Sine fa erin see ane bo. The Coentaron a the nance sf Plenty Abowonn. 10s 4 beep! Cote inthe Boeriouton fw Basle Gates: 220 31. The Calculation of Porolloxsrsss++sue eeciceresse 106 65. Summory of Formolos + ze 32 Trensioaton” ot Equal Cootdnoer tats Eig a el eereceeere 27 Mncotisnscrscetet ness eee ag er cmumess aneeienicice sania “Bo 22. The Compatson of ihe phonons igh Obvenaions cosets We 6b, Sonmery of Ferutos svossers 20 34. The Selection of Observations and Thele Control sone (oof Exenplafsesecrsre ees 245 CHAPTER FIVE: THE GENERAL PROBLEM OF DETERMINING AN CHAPTER NINE: NUMERICAL METHODS FOR THE CALCULATION OF Oneit 3 THREE OBSERVATIONS essen RuNNE AN ay HAOT ER NINE: NUMERICAL METHODS FOR THE A An 38 25, The Shomer af he Problem, osveveessoeverecccaeenceseey 1% medion cree sosoee Las 3, The Equations of the Plone ne... oa Wr ee nerbed of Cowell oa eee 233 37 The Expression ofthe Aone af he Tiangoe¥y Vineseccllls Ty Wen of Eeootos ta hid a Ewa 3 BE The Fie Apponmaton fer Hallogaone Seca 3 73. Ewomyle meen ms 39. The Analysis of the Fundamental Equations and Lombert"s Thee. 74, ‘The Method of Bond-Encke ssccuscssesse nn 68 rem Concerning the Curvoture of the Apporent Orbitssevsseses 133. 75. Summary of Formulos for the Method of Bond-Encke..essseeeeee 273 40. The Improvement of the Results of the Flrst Approximation sscss. 139 78. Example « ea a6 lavig| eles iameet a sete is asiace rns meee a 7 Fie led ci Veiaion of Elmers mo 42. The Second and Further Approximations..ssceveseeersesesseses 47 8. Summery wf Formulas for the Method of Voriotion of Elements... 292 43, The Determination of the Elements veresccusvsswesevsevevsere 148 IB. Example vesssseseeeeneerensecoseets ae 27 Mt Sunny of Pomdlens vesntrsecicceeeeetinecee ee Ie alee aera sesernteesea nececercacieese cd CHAPTER TEN: THE PRELIMINARY IMPROVEMENT OF AN ORBIT .... 301 ve 30 CHAPTER sit: THE DETERMINATION OF AN ORBIT BY FOUR OB: £0. Metts of Inposing Obitteneev ees = SERVATIONS vvsosrssncrrresrsr torsion ee 168 Becht aie an Soni son 45, Concerning the Use af Four Oberetions far the Detention eee eee cin of Werialon of Gaasnhic of on Orbits ssesseee . 168 OS erorees Peer feeeeeees 207 47. Basie Equotions...+++ 189 we Vorlation of the Ratio of Geacentric Distoncetscvccscvssees 308 48. Semon Poston : te ee eer cone a 49, Example «+++-+00 . . 19 7 TERMINATION OF A DEFINITIVE ORBIT 319 CHAPTER SEVEN: THE DETERMINATION OF AN ORBIT BY TWO O8- CHAPTER ELEVEN: THE DETERMINATION ‘nb Cowers nee PET es SRE finial Eatin Coan the onan. 318 50. Essence of he roam svvssessusvenveescsvoncnecccey 89 86. 4 hth forte Carecton of on Orit Based on the Aaplcaion 51, The Determinotion of a Circular Orbit ss. ssscsevseeenseeeusy TBA Te ake aeareg 27 52. A Summary of Fermules for the Computation of @ Civculor Orit oto ear on : m3 and an Example «vss. sssosverers : wees 186 A Method of Improving an Orbit Using Coordinates Referred to 52, Titwond'e Ctean ire Bocibiny fs Greaon OAH. WS Geena sec me ee ne 54 Determination of Elliptic Orbirs foo 192 SN oor iene apenas econ) 58. Shnny af Fermalos and on Crome oo es t2 sen a amin cc cesigicncnennnencnen SE, Cina of obsorotene Beorglag soo Sri ra Sh The Aponte Cove oe 94, Example sssssesecerevssesseieseses xiv Toble of Contents METEORS «seeseseses CHAPTER TWELVE: THE DETERMINATION OF THE ORBITS OF 95. Introdu een 7 96. Celevlotion of @ Metoor Path inthe Atmospheres 97, The Influence of the Earth's Airoctien and Met 498, The Determination sf the Orbital Elements 99, Summary of Formulas 100; Example 101. The “Calculation of Elements 102. Example ++. TABLES vs esseseeee Description of the Tables - ‘The Differenc Eceen Motion Along a Parabolossssessstsissessreevesvs The True Anomaly for Nearly Parabolic Orbite ss. The True Anomaly for Nearly Parabolic Orbits. s..ssssses ‘The True Anomoly for Neorly Parabolic Orbits (Hyperbole ‘ood Ellipae) = oe A F Toble VI. The Time of Perihelion Passage ior Nearly Porabelic fe and Maon Anomalies Tobte Vil. ‘Toble Vill. Of the Aves of @ Sector to 2 Teiongle lo of the Area of Sector to « Triangle, Units of the éth Decks Teble IX. Solution of Euler's Equati Ein REFERENCES v1... PUBLISHED RAND RESEARCH-. 356 356 357 363 a7 374 381 ae 395 399 399 401 404 aa ag 40 a2 a3 44 425 ar 42 Chapter 1 THE PROBLEM OF THE DETERMINATION OF ORBITS AND ITS HISTORY 1. The Orbits of Minor Planets, Comets, ond M ‘The study of the motion of celestial bodies, and also of their figures, is the object of celestial mechanics. Among the numerous and difficult problems in this province of science, the problem of dotermining orbits of minor planets, cohets, and meteors, is one of the most simple for which a solution ean be obtained. Tts practical significance to astronomy is due to the fact that each year dozens of new asteroids and new comets are discovered, and one must be ablo to determine their orbits in order to keep track of them suc~ cessfully. It is equally obvious that there is a need for the com- utation of the orbits of metoor ‘The bodies of the solar system move under the influence of the force of universal gravitation acting according to the law of New: ton. The bodies themselves may be assumed to be material points. "There is no doubt that these two statements reflect reality only approximately. Is it legitimate to limit oneself only to the force of Newtonian gravitation? The corrections to the law of Newton re- sulting from the theory of relativity are very small, and they have no significance in almost all of the cases of interest to us. ‘There- fore, we shall not touch upon the difficult problem of the extent to which the introduetion of relativistic corrections to the gravita- onal theory, in any form, is justified, However, ather forces bo- sides gravitation influence the motion of celestial bodies, comets in particular; the most significant of these forces, originating in the sun, is light pressure, Besides this, the intonsified heating of comets during their approach to the sun gives rise to an outflow of gases from their nuclei; the mass of the comet must gradually de- crease through the ejection of gaseous and also of solid particle: the comet breaks up gradually. For these reasons, and perhaps for others, not yet known, the motion of many comets deviates significantly from that predicted by a theory based only on the law of universal gravitation. Such deviations, of which the secular acceleration of Encke’s comet fone, are widely known and mentioned in all courses of astronomy. They are an interesting problem of both celestial mechanics and astrophysics. But, besides this, the shape, the inner structure, and the rota tion of celestial bodies directly influence their translational tio- tion along their orbits. Strictly speaking, only for spherical and homogeneous bodies or for bodies consisting of regular concenteic spherical layers of difforont densities are we permitted to assume that all of their mass is concentrated at the center and that they Altract one another as material points. ‘The fact that comets som times possess no spherical symmetry at all (for instance, recall the cuses of the division of comets) may, to a certain extent, mani- fest itself in anomalies of theit motion. However, the distances between the bodies of the solar system are almost always very large compared with tho dimensions of these bodies, and we usually have the right to accept small planets and comets as material points. As a first approximation, it is as~ sumed that they move under the influence of the attraction of the sun only, and are not subject in any way to influences from the Iarge planets. The force of mutual gravitation between minor planets and comets can always be distegarded because of their very small masses. In the same way, the attraction of the stars is imperceptible because of the enormous distances involved. In more careful investigations of orbits it is necessary to take into account the perturbations due to the attractions of large planets. ‘These perturhations in the vast majority of cases are small, since the mass of the largest planet Jupiter is 1,047 times smaller than the mass of the sun. This overwhelming peeponderance of tho central mass over the combined mass of all the other bodies is a guarantee of the stability of the solar system, which under other conditions (for example, if the sun were a double star with con- siderable distance between its components) would hardly be able to exist in its present form. To he sure, the stability of the solar system, as woll as any stability, is relative, Actually, some com- ets sometimes come so close to Jupiter that its attraction begins to predominate over that of the sun and gives rise to radical changes in the character of the comet orbies; however, this is very uncommon. Having established the general character of the problem of The Problem ond Its History a ‘orbit. determination, we shall enumerate the gradual stages of its solution: 1. The dotermination of the eloments of a preliminary orbit, ‘without taking into consideration the perturbations: in other words, the solution of the problem of two bodies. For this purpose, three and occasionally four observations of the body under investigation fre taken, covering an interval of several days ot weeks or, under special circumstances only, of several months. It is desirable to obtain the preliminary orbit of @ planet or a comet as Soon as pos- sible after discovery to permit a quicker prediction of its future positions for further observations. 2 The improvement of a preliminary orbit, which is done after the accumulation of a longer series of observations. It is also possible in the majority of calculations of improved orbits to neg- oct perturbations. 3. Computation of tho definitive orbit. ‘Tho definitive orbit is the one which agrees, in the most probable manner, with all the obtained observations. This applies,in particular, to comets for which, after the end of the period of visibility, it! is possible to collect all of the observations made at the various observatories of the world. In the determination of a dofinitive orbit, it is almost always necessary to take into account the perturbations, at least those that are due to the principal planets. If the orbit of a celestial body is found, thon it is not difficult to compute its ephemeris, i.e., to give the positions of this body fon the geocentric celestial sphere for a series of equidistant mo- ments. The ephemeris can be calculated for a series of years, thus providing the possibility of a future recovery of some minor planets and periodic comets. In the latter case, it is usually necessary to tako into account. the perturbations ‘enused by the lanets. vin The difforontial oquations of the porturbed motion corresponding to the problem of three or more bodies does not admit a solution in finite form, unlike the problem of two bodies. ‘These equations are solved approximately, either with the aid of series or by numerical integration. The solution in the form of series giving an analytic expression for the perturbations offers a complete reprosentation of the character of the motion of the body during a reasonable length of time. However, the application of the analytical method to the majority of minor planets and to almost all the comets is rather difficult. Numerical methods for ealculating pertuehations are simple and reliable, but they have one important deficiency. Tho valuos of the porturbations can he found only within the time 4 The Determination of Orbits limits included in our calculations; if it is required to find portur- bations for a series of years, the work can become extremely burdensome. In spite of this, ‘numerical methods for caleulating perturbations constitute an important part of the theory of the mo- Gon of minor planets an! of comets. In regard 10 meteors, there can be no possibility of improving their orbits, since observations can be mavte only during tho short time of their flight in the earth's atmosphere, Orbits calculated for individual meteors oF for meteoric showers are not very precise hhocause of the essential errors to which visual observations are subject. Greator precision (still not comparable with the precision ‘of cometary and of planctary orbits) can he obtained for orbits eal- culated from photographie observations of meteors, but up to now only a few such orbits exist. Th astronomy, we also encounter the problem of the determina- tion of orbits of other celestial bodies. For example, we have available extremely long series of observations of principal plan- ots, many satellites and, first of all, the moon; these observations should be, as exactly as possible, represented by the theory. Here is the principal field of application of analytical methods of celes- tial mechanies, but we shall not examine these questions since they go heyond the range planned by us. Again, the problems of finding the orbits of newly discovered satellites of planets or the orbits of double stars—visual, spectroscopic, ani photomotr are, it is truo, similar to the determination of orbits of comets and asteroids, in that we encounter in all of these cases the problem of two bodies. Nevertheless, the methods of solution for each of those problems aro too diverse, and they are now usually treated separately. Thus, we shall limit ourselves to bodies moving around the sun, i.0., to minor planets, comets, and meteors. 2. First Pre-Newtonian Attempts ot the Determination of Cometary Orbits The views of the ancients on the nature of comets were rather vague and contradictory. The only helief that prevailed for a long time was the opinion of Aristotle that comets were not colestial hodies but that they represented masses originated from terrestrial evaporations and ignited in the air. One who took exception to this view was the Roman philosopher, Seneca, who was convinced that comets are celestial hodies hut are distinguished from plane! by the character of their orbits; they are visible only when they approach the earth, One should not really be surprisod that Aristotle and Ptolemy did not associate the study of comets with the domain of astronomy. The Problem and Its History 5 With the naked eye one may observe only bright comets, and they appear comparatively rarely and are not visible for very long—at best, for a few months only. In the motion of the comets in the sky, tho regularity and periodicity that is revealed by observations of the major planets is absent. But the Greeks considered invariabil- ity and eternity as the basic properties of colostial bodies, as well 1s the circular form of their paths, and comets conspicuously lack these qualities. Only in the epoch of the Renaissance, shortly before the Co- pernican discoveries, the Nuremberg astronomer Regiomontanus Set out to prove, for the first time, that comets actually are heay- enly bodies, He absceved the comet of 1472 for the purpose of finding its daily parallax and thus to determine its distance from the earth, Since observations convinced him that the parallax of the comot is not perceptible, Regiomontanus concluded that comets must he farther away than the moon, which has a parallax of about 1°. This deduction received its final confirmation a hundred years Inter when Tycho Brahe concluded, om the basis of his more pre- cise observations, that the comet of 1577 did not have a percepti- ble parallax. Tycho was so sure of the correctness of his observations that he was the first in the history of astronomy to attempt to find the orbit of this comet. First, he assumed that the comet moved around the earth and he computed geocentric orbit for it, but this did not satisfy him so he searched further for a heliocentric orbit. Ap parently, these calculations wore made by him in the style of the Ptolemaic theory of epicycles, which could not be readily applied to comets (because comets are observed for only a short time). It is no wonder that the work by Tycho Brahe on the orbit of the comet of 1577 did not yield an accoptable solution. Nevertheless, it is interesting to notice Tycho’s conviction that the circles being described by comets did not, in fact, exist in the sky, but that these circles wore merely mental aids for the calculations, What is more, ho says in passing that perhaps the omat, does not describe a circular path about the sun, but an clongated path of egg-shaped form. ‘This was a courazcous thought for that tine—to break with the concept of the circular form of the orbits of celestial hodios. Of course, this thought could not be proved by Tycho Brahe and it remained an unverified supposition. Copernicus and Kepler achieved success in explaining the true character of the mation of planets. In particular, the three laws of Kepler must have led, sooner or later, to the discovery of the form of cometary orbits, but this did not happen all at once. Thus, for 6 The Determination of Orbits instance, Kepler himself assumed that the trajectories of comets were straight lines. From Kepler's point of view this was a logi- cal conclusion because Kepler thought that comets did not return periodically, and of all the forms of orbits of colostial bodies, ho knew only ellipses, which are closed curves. Later Borelli and Hevelius introduced the thought that comets move along parabolas and hyperbolas, but still no one thought of placing the sun at the focus of the orbits, At last, shortly before the publication of Newton's Principia, the work of Doerfel on the comet of 1680 appeared, in which for the first time the supposition (only a supposition!) was stated that the trajectory of a comet is a parabola with the focus in the sun. Unquestionably, at this point Doerfel was able to follow the anal- ogy with Koplor's second law, but nevertheless it was not so easy for him to he convinced of this analogy. In fact, what is the main difficulty in the problem of finding the orbits of bodies in the solar system? If we were to obsorve tho directions to the heavenly bodies marking for clarity their pos tions on the celestial sphere, and if we learned simultaneously the distances of these bodies from the earth, thon of course it would be easy to construct the true orbits in space and to find their form. However, we do not know the distances to the comets and, as a matter of fact, an observer on earth, who considers himself at the center of the celestial sphere, moves together with the earth around the sun, so that the apparent shift of a celestial body in the sky consists of the motion of the body itself and of the motion of the arth in space, Without using calculations, Doerfel must therefore have imagined a sketch in space of the observed diroc- tions to the comet, and from this sketch he must have deduced its parabolic teajectory. Only after the laws of motion of material bodies wore clearly understood, and, as a result of the discovery hy Newton of the law of universal gravitation, Kepler's laws had hocome meaningful and had acquired their final form, was it possible to devise rules for the determination of orbits of comets, using * few observations encompassing a small interval of time. The Inhorious methods, used by Kepler to establish the Inws of motion of che planets, were based on numerous observations accumulated through the years, permitting the derivation of the periods of revolution of the planots with great precision, These methods were not applicable for determining the orbits of comets. Only with Newton did theo- retical astronomy begin, or, speaking more precisely, the theory of The Problem and Its History 7 the determination of the orbits of comets, and later on, of minor planets. 3. The Epoch of Newton ond Halley In his Mathematical Principles of Natural Philosophy, Newton states that the problem of determining the orbit of a comot is & traordinarily difficult and that he did not find its solution im- mediately, The original method of Newton, as sketchily outlined by him and, apparently, nover tested in practice, was unsuccessful. It was based on the old Keplerian assumption that during the time of observation the comet moves along segment of a straight tine with constant velocity, What is more, if we were to analyze New- ton’s course of reasoning, it would show that he accepted a similar motion even for the earth. It is true that he considers the displace- ment of the comet and of the earth for only a two-hour duration, but this leads to new errors. : Apparently Newton was aware of the insufficiency of his first method, but he was not stopped by the difficulties. He proceeded from an absolutely different premise and worked out in the third book of the Principia another, entirely reliable, method for the de- termination of cometary orhits from three observations. First of all, let us notice that the principal pact of the solution by this, method was accomplished by Newton with the aid of graphical con- structions. ‘The accuracy of the elements of orbits found in this way was usually sufficient, if we remember that at that time the measurements determining the positions of comets wore fairly rough and chus frequently produced errors in coordinates up to 15 minutes of arc, and even more. However, later on, Newton indi ated how certain computational techniques could make tho clo- ments of the orbit, obtained by graphical means, more precise and could achieve any required precision. Three observations of a comet determine three directions to- ward it deawn from three positions of the earth at the moments of the observations. Constructing on the sketch the projections of these directions on the plane of the ecliptic, Newton solects, at first. arbitrarily, a certain position of the comet at the middle one ‘of these directions, Considering the projection of the ecliptic of the radius vector of the comet at the moment of the second obser- vation and the chord between the first and third positions of the comet, Newton at first assumes that the radius vector divides the a The Determination of Orbits chord into segments which are proportional to the intervals of time between the first and second, and between the second and third observations, respectively. That is, this seems to imply that the point of intersection of the radius vector and the chord moves Along the chord with constant velocity; this does not quite accu- rately correspond to reality, but it is, in any case, much better than the assumption that the motion of the comet itself is recti- Tinont and uniform. However, Newton was not satisfied by this and vory skillfully found a new point on the chord in which the div sion proportional to intervals of time is accomplished much more accurately than by the first supposition, In addition, the length ‘of the chord must correspond to a dynamic condition resulting from the motion of the comet along a parabola. The fulfillment of these requirements can be achieved by means of trials, by varying the position of the comet on the direction line of the second obsorva lion. After this, the construction of the orbit itself and the de- termination of its elements prosont no particular difficulties. As fan example, Newton gives (very briefly) the principal stages of the search for the orbit of the bright comet of 1680, the same comet, that was observed by Doerfel. ‘Academician A. N. Krylov has given a detailed analysis of the method of Newton as presented in the Principia [8]. He has shown how it is possible to translate the geometric arguments of Newton into the language of formulas and, by the calculation of a few or- bits, he has established that Newton's method gives good ngrec- ment with the observations, A. N. Krylov has explained the very condensed arguments of Newton, which in his time were so diffi- cult to understand that of all the contemporaries of Newton only Halley was able to master completely his method and to apply it to the calculation of orbits of those comets for which there was a sufficient number of observations. ‘The first list of the 24 comets observed between 1337 and 1698 was published by Halley in 1705. An enormous amount of effort was spent by him on these calculations because of the imperfoc- tion of the calculation technique of the period. The discovery of the first periodic comet, which Inter was to bear Halley's name, was his reward, By s comparison of the orbital elements of the comets of 1531, 1607, and 1682, Halley was able to notice their Similarity and to conclude that in these years one and the same comet appeared, revolving around the sun in a closed elliptical orbit with a period of about 75 years. Halley found later that this comet appeared also in 1456, In our time, as a result of investien tions by Cowell and Crommolin, and by M. A. Vilyev, there was The Problem and Its History 9 found in ancient, chiefly Chinese, chronicles an entire series of earlier appearances of Halley's comet. In his remarkable work on the earliest recorded appearances of Halley’s comet, M. A. Vilyev suggested that the comet observed in China in 487 B.C. was prob- ably Halley's comet, and so this would be its very earliest appear- ance known to us [3]. Halley was so convinced of the validity of his results that he decided to predict the next appearance of the comet for the end of 1758. This prediction, confirmed and made nore precise by the investigations of Clairaut, who calculated the perturbations of the comet due to Jupiter and Saturn, came true in P’prilliant manner and appeared to contemporaries as a veritable ttiumph of the law of universal gravitation, a clear confirmation of the correctness of the materialistic view of the universe. 4. The Wor ‘The first purely analytical method for the computation of comet ary orbits was proposed by Buler in 1744, However, Euler himself was soon convinced that this method bas a series of deficiencies, jn that this method utilizes four observations, i.e., a groator number than is essentially requited by the problem, and the calcu- Tations are long and tiresome. It is more important that to Euler belongs the discovery of the equation connecting two radius vec tors and the subtended chord of the parabola with the interval of time during which the comet doseribes the corresponding arc. It not necessary to know for this the elements of the orbit. Tt is true that Newton knew the geometric relation which was absolutely equivalent to the equation of Euler and which Newton applied in his method, but only in the formulation of Euler did this theorem actually become fruitful for the theory of the determination of parabolic orbits. A detailed account of the history of this remark- able theorem is piven by A. N. Krylov. Lambert, in his works of 1761 to 1771, established a series of positions which essentially facilitated the problem of the orbits of comets, bringing out clearly its principal factors. He asserted that the most important thing is to reduce the prob- tem to one unknown, any unknown for which it is possible to find fan approximate value quickly, and then to give a method for the Correction of this approximate value. He indicated that it is of ‘considerable advantage to project the parabolic orbit not onto the plane of tho ecliptic, as was done by Newton, but onto another plane, perpendicular to the radius vector at the moment of the second observation. Lambert gave a generalized formulation of the theorem of Euler of Evler, Lambert, Lagrange, and Laplace 10 The Determination of Orbits for the case of elliptical and hyperbolic orbits. He also estab- Vishod an interesting theorem concerning the curvature of the ap- parent trajectories of bovlies of the solar system on the celestial sphere. This theorem indicates (in principle) direct way for Finding the distance of celestial bodies from the sun. However, Lambert failed to combine the results of his research into a single whole, and he did not create a completed method for the determination of cometary orbits. Lambert was first of all a geomelrician and he was not inclined to an analytical development ‘of a method. However, only analytical developments could lead to success; it camo to Lambert's successors. ‘Tho investigations of Lagrango, published in 1778 and 1783, after the death of Lambert, paved the way for the solution of the general problem of determining the orbit of a celestial body without first_ making suppositions about its eccentricity. Lagrange, like Lambert, did not succeed in adapting kis methods to Lhe demands of practical calculations of orbits, but nevertheless his conclu- sions have fundamental significance for theoretical astronomy. Lagrange subjected the general problem of the determination of orbits to analytical treatment, and his analysis was profound and elegant. Lagrange started from the condition that all throe ob- servations of the position of the body are situated in one plane with the sun, Having solved the obtained equations, he gave an expansion according to a power series in time for areas of triangles formed by three radius vectors and the orbit chords between them. Finally, Lageange reduced the problem of finding the radius vector ‘at the moment of the second observation to the solution of an equa. tion of the eighth degroo. The results obtained by Lagrange in his first memoirs acquired 8 completed form in the method of Gauss. In the second memoirs, Lagrange approached the problem in a different way by utilizing the coordinates and tho velocity components at the momont of the second observation as the desired constants in the intograls of the differential equations of motion. A series of authors, the first fof whom war I. A. Vostokov (5], tried to make the second method of Lagrange suitable for practical calculations; they were succe: ful in this to a significant dogroe, but novertheless this method has not boon widely used. On the other hand, Lageange occupied himself with the problem of finding the perturbations which comets undergo because of the major planets. It is truo that before him this problom was studied by Clairaut. What is moro, Buler had already utilized for this pur- pose the method of variation of arbitrary constants. But only The Problem ond Its History u Lagrange succeeded in revealing all the possibilities of this method, and that is why it carries his name, If we examine only the problem of tivo bodies, then the differen- tial equations of the relative motion (for example, of a comet around the sun) are solved in finite form, and in the solutions there will be six arbitrary constants, which could be, in particular, the usual elements of the orbit. If we take into account the action of the third body, then additional terms appear in the right-hand side of the differential equations, and the motion will no longer proceed according to the laws of Kepler. However, it is possible, as be- fore, to satisfy the equations of motion by the general formulas of unperturbed motion, if we assume that the elements of the orbit fare not constant, but that they vary with time. One can calculate the derivatives of the eloments with respect to time for any moment, after which it is possible, by means of numerical integration, to find the values of the elements for any epoch, and that solves the problem. The method of the variation of orbital elements was worked out in detail in tho 19th century and has found wide appli ation. Beginning with Newton, the problem of the determination of orbits was always so stated that its purpose was the finding of a system of elements satisfying as precisely as possible three properly chosen observations of the celestial hody. ‘That is the ‘most natural statement. of the problem, and later developments have fully justified the formulation of the problem. Laplace started from a completely different premise. In his method, proposed in 1780, the number of observations boing. uti- zed was not limited, and from them asa set the coordinates of the object on the celestial sphoro, and theie first and second de- rivatives with respect to time, are determined for some moment (usually for the moment of one of the observations). Laplace has shown how from these six quantities it is possible to derive, in a simple and rigorous manner, all the orbital elements for any form of conic section described hy a celestial body around the sun. From the point of view of analysis, it is hardly possible to find 1 more complete solution than that obtained by Laplace. However, it has been repeateilly emphasized that the problem of the determi nation of orbits doos not exclusively belong to the domain of analy- sis, but that it is first of all a practical problem. Practical con- siderations concerning the precision of the results and the conveni- cence of the calculation process have a decisive significance. The method of Laplace did not justify itself in practice, because it is very difficult, almost impossible, to obtain from the observations, n The Determination of Orbits with sufficient precision, the first and second derivatives of the Fight ascensions and the dectinations of the object. ‘The funda- mental thought of this method is so alluring that there were re- peated attempts to modify and adapt it to practice. Leuschner, in particular, applied much effort in this direction. In the end he suc- cessfully obtained agreement with the observations, but only by means of a long and involved process of improvement of the ele- ments obtained by direct solution (which, as a rule, poorly satis- fies the observations). ‘Thus, the mothod of Laplace did not find followers in practical astronomy. The Methods of Olbers and Gauss ‘Through the work of Lambert and Lagrange, the way was com pletely cleared for the successful solution of the problem of the determination of paraholie orbits. In 1797 the work of Olbers was published undor the title, dn Essay on the Basiest and Most Com venient Method of Calculating the Orbit of a Comet. To this day the method of Olbers justifies this title. In the course of time many changes have been introduced in it; however, they do not affect the basic concept of the method. Assuming that the chord between the first and thitd positions of the comet in its orbit is divided by the second radius vector in the ratio of the intervals of time between the observations, as was done by Newton in his first approximations, Olbers obtains an equation connecting the first and thied geocentric distances of the comet; the second necessary relationship between them is given by Euler's equation. In this lies the essence of the solution, be- cause after finding the distances of the comet from the earth, the calculation of the elements of orbit no longer presents a difficulty. Olbers wrote that he had carefully studied the works of his predecessors; of them all, he was undoubtedly most indebted to Lambert, Olbers made a reality of what Lambert had merely indi cated; Olbers gave a practical and convenient method for the solu- tion of equations for the goocentric distances by means of trinls. In the method of Olbers the required level of precision was well established, ‘The convenience of the method of Olbers depends to fa considerable extent on the fact that it is not entirely exact. But the etror in the basic assumption of Olbers will not, in the majority of cases, have an effect excoeding the influence of the errors in the observations. Moreover, there always exists the possibility of correcting the otbit and of obtaining, if necessary, better agree- ‘ment with the observations, The Problem ond Its History ‘The Kiev astronomer Fabricius has shown that a method virtu- ally identical with the method of Olbers was already contained in the work of Duséjour, 1779, but that Duséjour himself did not at tach much importance to his derivation, Olbers road the memoir fof Duséjour but apparently not too attentively, because in his book he speaks of him only in passing. ‘The discovery of new planets increased the incentive to solve the genoral problem of the determination of orbits. Titius found an empitical rule for distances of planets from the sun which pointed Ou an evident gap between Mars and Jupiter. ‘The discovery of Uranus by Herschel in 1781 increased the number of planets known from ancient times, and concurrently the rule of Titius appeared well confirmed. Many astronomers had not doubted that between Mars and Jupiter there revolved still another, undetected, planets in autumn of the year 1800, on the initiative of Zach, it was de- cided to start a search for it; the ecliptic region was divided into 24 sections, and each of them was assigned to one of the partici- pants in Zach's plan. . ‘Among the prospective participants was Pingzi, but before he had received the formal notification he already discovered a new planet on January 1, 1801. Piazzi observed it until the middle of February, and when information about the planet reached other ob- servatories, it had already disappeared in the rays of the sun. ‘The planot was due to become visible again in the autumn of that same Year, but the search for it would have been very difficult if the cal- Culation of its orbit had not been undertaken by Gauss, who found the system of elliptic elements satisfying the observations of Pinzzi and gave the ephemeris of the planet. Using this ephemeris, Olbers rediscovered the planet on January 1, 1802, exactly one year after its first discovery. Piazzi gave the new planet the name Cores. Gauss's work, Theory of Motion of Celestial Bodies Moving “Around the Sun in Conic Sections, was published in 1809. Its prin- Gipnl contents is the method worked out by Gauss for the determi- nation of elliptic orbits; in his method Gauss considerably supple- mented and developed the ideas uf Lagrange and worked out « complete and practical solution of the problem. He has shown that it is extremely difficult to reduce the problem to the evalua tion of one unknown, and that it is much more convenient to find the values of two unknowns. He succeeded in giving to Lagrange's fundamental equation of the eighth degree a simple and elegant (rigonometric form. {le gave a method of finding, with any required precision, the ratios of the area of the sector of « conic section 4 The Determination of Orbits between two radius vectors to the triangle formed by the same radius vectors and the chord joining thom. It was very significant that Gauss was not only @ theoretician but also a practical com- puter, a profound stulent of the technique of the calculation proc- fess, and a founder of its rational basis. Mis book contains much that is instructive from the purely computational point of view. 6. Subsequent Development of the Theory of Orbit Determination After Olbers and Gauss the theory of orbit determination reached ‘a certain degree of completeness, but its development did not stop there. ‘The efforts of some of the most prominent mathematicians fof that epoch, all of whom were also astronomers, established the fundamental principles, but it was possible to work them out in many different directions. Many new methods for the determination of othits (based on already well-known principles) were offered; not only did they not supplant the methods of Others and Gauss, but one may say that the now methods did not stand the practical tosts of comparison. For example, attempts were made to increase the precision of the first approximation as compared to the one given by the method of Gauss. ‘This turned out to be possible and led in some casex to a lesser number of approximations than with Gauss’s method, but the formulas became so complicated that they resultod not in a saving but in a waste of working time, However, it is impossible to deny that very interesting specific results were found, such as Gibbs's elegant oxprossions which give the ratio of the areas of the triangles hetween the radius vectors with a high ogre of precision. On the basis of these formulas, Fabricius worked out an original method of orbit calculation [16]. Tho method of Gauss has the virtue that i€ permits to improve at will the precision of the first approximation without too much work and thus to achieve full ageeoment with the observations on which the calculation of the orbit is based. Thousands of orbits of minor planets anil of comets, calculated according to the method Of Gauss, Inve confines its convenience and effieionay. With regard to this, one should note that M. P. Subbotin has made an intensive investigation of the Euler-Lambert theorem which represents a generalization of Euler's formula for a parabola in the case of arbitrary eccentricity, and ho has shown that using the theorem as a starting point itis possible to create new method of orbit determination differing in principle from Gauss's mother. Somewhat less attention was devoted to the theory of parabolic The Problem and Its History 6 orbits. It is possible to encounter, in the solution of this problem, the case where the positions of a comet on the celestial sphore are so distributed that the method of Olbers does not give reliable results, The determination of an orbit is possible even in this case, but the computations must be hased on an equation between the geocentric distances which differs from Olbers's equation, This problem occupied Gauss and, more in detail, Oppolzer. A similar exceptional case in the arrangement of observations may take place also for an ellipse. In theory, it oceurs when tho orbit lies in the ecliptic plane. However, the orbit is calculated tunteliably if its inclination to the ecliptic is small but not strictly equal to rero, Three observations are insufficient for the determi- nation of an orbit with a small inclination, and it is then necessary to use four observations. This was already clear to Gauss, and he gave the corresponding formulas for the derivation of an orbit by using four observations. Te also developed a method of determining a cireular orbit, Of course, there are no exactly circular ogbits, hut their existence is ‘8 useful assumption for a computation if only two observations of ‘a minor planet were made, which are sufficient for finding four elements characterizing a circular orbit, but certainly insufficient for an ellipse. This problem is rather simple and was solved in different ways both before and after Gauss. ‘The problem of orbit determination does not always have one solution, although multiple solutions are found in practice almost oxclusively in the calculation of orbits of comets appearing at short distances from the sun. ‘A. A. Yakovkin investigated the possibility of multiple solu- tions for a parabolic orbit, by using an elegant geometric method [17]. In the determination of a parabolic orbit, it is necessary to find only five elements, because the eccentricity is assumed to be ‘equal to unity. Therefore, the second observation enters into the caleulation incompletely; for the construction of the fundamental correlation between the geocentric distances, it is sufficient to take a cortain great cirele passing through the second observed position of the comet. In Ulbers’s method this circle also passes through the second position of the sun on the geocentric celestial sphere. Considering only the first and third position of the comet as having beon given, Yakovkin examined all the possible posi- tions which a comet may occupy, when moving in a parabolic orbit, at the moment of the second observation. These positions lie on ‘8 cettain curve whieh Yakovkin called the isochrone. If the iso- ‘chrono intersects the fundamental circle of Glbers's method in any % The Determination of Orbits point, we shall have an orbit satisfying the initial conditions. TThero may be one a three of these intersections; the problem of TiC dotormination,mecording to Olbers’s method will havo the hme number of solutions. It is ovon more intaresting that in the Skeoptional caso when the mothod of Olbors is unsuitable, and it iipmeemeary to lace in another manner a great cielo through the Sacond position of the comet, thoro may be five solutions: In- Sestisadne the isockronous curve in more detail, Yakovikin mado important deductions concerning tho precision of the calculation af purnbolic ort elements. Much may De expected fron Yakorkin's suggestion to use a planinotor of orginal construction for mensuring the neea of the ttle sector Retween two radiys voctore, This offers the possi- bility of replacing by simple mechanical mensurements the Iabori- tus procest of calculating the area ratio of a sector to a tangle, Chick plays an essential volo in tho detormination of an olliptic orbit. ‘The investigations of Schiapntelli published in 1868 established the connection between meteoric showers and comets. In this way, tot only was a stop forward taken in the knowledge of the coamie nature of meteors, on which By F Chladni insisted in 17M, but a Stimulus wae given to developing methods for detormining orbits of meteors. "An essential simplification in the ealoulaion of orbits Of molcoes arises trom the fact that the. position of meteors in Space may be considered as known (coinciding with cho position of the earth), anlan oxhit can be obtained if tho observations give the velocty of the meloors, in magnitude and direction. Besides, the Value af the velocity may often be considered equal to tho arahoe velocity at adistanee of the eaeth from the sun. Tlaving. created the theory of tho origin of meteoric swarm, FA pradichin also studied their orbits, and his work shed Tigh, in particular on the fact that motoors of ono swarm can bo observed foc averal days in succession: they do not move in one oebit but tna bundle of evits, all itersocting in one point (21). v2 Shnin investigated the perturbations oxperionced by the most important meteoric showers [35]. K. P. Stanyukovich and E.'S, Anaporich developed methods for calculating the teajectory OF aimatacr in the ataouphere on tho basis of observations from tho" diferent tations on the tomestrin! surface, Previously, this troblom usunlly wae solved on the basis of simplified and unoatis- fhetoryeesumptions, an the practical results obtained were of low tality Inthe recent decals, the whole Lochnique of ealeulations hes i ' ' The Problem ond Its story v changed. Previously, astronomical calculations were carried out with the aid of logarithms; this required, in particular, a reduction of tho working formulas to an appropriate form, sometimes by means of rather artificial transformations, for the most part trigonometric, through which it was not easy to watch the development. of the basic’ idea of this or of that method. In connection with this, it was customary to use ecliptic coordinates which have the advan- tage thal it is possible to neglect one of the coordinates (Iatitude of the sun) due to its smallness. In our day widesprend use of fast and reliable ealeulating ma- chinos has come about, which in wany calculations, including tho determination of orbits, began to supplant logarithms. In using an arithmometer, it appeared advantageous to preserve the equatorial coordinates as they are obtained from observations. ‘This elimi- nated th necessity of transformation of coordinates to the ecliptic system, it made easier the calculation of parallax, and the basic formulas of orbit computation acquired a simple and intelligible form, thanks to the use of rectangular coordinates and tothe aboli- tion of trigonometric transformations. Despite all these advantages, there are problems, for example, tho determination of a parabolic orbit, where logarithms are still used to bring about a quick solution. Considerable successes were also achieved in working out methods of calculation of perturbations. Besides the method of variation of constants, which gives the perturbations of orbital clomonts, methods for calculating perturbations of the coordinates of celestial objects were created. The mothod proposed by Hansen for the calculation of perturbations of polar coordinates was in- tended primarily for periodic comets moving along orbits with con siderable eccentricity; incidentally, it was applied infroquently. For small intervals of time it was found to be very advantageous to compute perturbations in rectangular coordinates. In place of six elements there appear only three coordinates, and the formulas, published almost simultaneously by Bond in 1849 and by Encke in 1851, are simple and aro easily adapted to calculations with the aid of a computing machine. In 1909 Cowell and Crommelin presented a method in which, by moans of numerical integeation, they obtained not the perturba. tions of coordinates, i.c., differences. between perturbed and unperturbed coordinates, but the perturbed coordinates themselves. This method, well justified in practice, is based on an almost di- rect utilization of the differential equations of motion written in their simplest form—in rectangular coordinates, 8 The Determination of Orbits The determination of improved and of definitive orbits was the subject of numerous papers appearing in the 19th and 20th centuries. We will point to the remarkable concepts of M. A. Kovalski, connectod with the invention of the method for correction of an orbie in coordinates referred not to the equator or to the ecliptic but to the plane of the orbit being corrected [28]. A simi- lar method, subsequently perfocted, found wide application in the determination of orbits of minor planets and turned out to be very simple in practice. S.A. Kazakov obtained important rosults in the theory of correction of cometary orbits [26]. Our scientists not only played a prominent role in the development of the study of the determination of orbits but produced also fundamental works of profound intorest devoted to the detailed study of the motion of individual planots and comets. We already have seen above that Eulor, who was a member of the Petersburg Academy of Sciences, worked out the first analyti- cal method of the determination of orbits; famous formula, basic for the determination of a parabolic orbit, bears his name," Under his guidance, another Petersburg academician, Lexell, made in- vestigations and calculations of the elliptic atbit of the comet of 1770 and dotermined its period. This comet proved to be the first short-period comet, for up to then the only known periodic comet was Halley's comet, with a much longer period. After the investi~ gations of Lexoll, it appeared surprising chat his comet had not been observed either before or after the year 1770, although its period is only 5% years. Laplace explained why observers failed fo find the comet in its subsequent returns, In 1767 the comet closely approached Jupiter, and the atteaction of Jupiter changed its orbit to such an extent that the comet began to move along an cllipse with a short period; hut it described such a path only up to 1779, when @ new approach to Jupiter once more changed its orbit radically, so that the comet ceased to be accessible to ob- servation. The comet of Lexell (so named after its first investi- gator; the comets of Halley and Encko were also named after their Inventigators, not aftor their discoverses) ia alao remarkable, be- cause in its appearance in 1770 it came closer to the earth than any other body of the solar system, with the exception of the moon, the tiny asteroids Adonis and Hermes, and the meteors. Lexell also computed the orhit of Uranus, discovered in 1781, and confirmed that it was a now planet moving far beyond Saturn. Ho tried to explain the deviation of the motion of Uranus from his calculations hy the influence of an unknown planet located at an ‘enormous distance from the sun, Of course, the data which Lexell The Problem and Its History w had available wore totally insufficient for establishing his hy pothesis. Nevertholoss, the deviations of Uranus from the theory, built by taking into consideration the actions of the known plancts only, led to the discovery much later of Neptune. Having touched upon the discovery of Neptuno, it is impossible not to mention that M. A. Kovalski thoroughly investigated its mo- tion. He constructed the first tables of this planet which were used for a number of years in the calculation of the ephemerides given in astronomical yourbooks. Among the numerous works devoted to the study of the motion of individual minor planets, one should mention the investigation hy A. A. Ivanov of the orbit of the planet 122 Gerda, based on ob servations made in the course of many decades. ‘Tho profound theoretical investigation of M. A. Vilyev touched upon the most diversified problems connected with the determina- tion of orbits. Vilyov liked computations and he know how to com- pute; not limiting himself to theoretical studies, he computed a great number of orbits of minor planéts and of comets and his ‘ophemerides have rendered invaluable services to observers. After the untimely death of Vilyov, his work was continued by a group of astronomers of the State Institute of Computation, founded in 1920 in Leningrad, later reorganized into the Astronomi- eal Institute, and again (1943) into the Institute of Theoretical Astronomy. One of the basic tasks of this Institute was the calculation and the publication of the Astronomical Yearbook of the USSit, which has appeared without interruption since 1922. This yearbook was continuously improved; much was done in this respect by N. I. Idelson who gave it in 1941 its prosont form. ‘The Astronomical Yearbook of the USSR is a necessary tool for all Soviet astrono: mets who Compute orbits and ephemerides. In the Institute of Theoretical Astronomy, the work on minor planets has boon considerably expanied. Apaet from the calcula: tion of absolute perturbations for a series of planets, perturbations for many planets were computed in apinl courdinates. by 1 nu merical mothod especially worked out in the Institute for minor planots. This mothod, formerly called the method of extrapolation, may offor certain advantages provitled that extensive auxiliary tables, compiled especially for this purpose, are used, and pro: vided that the calculations are carried out on a large scale, In recent years the Institute of Theoretical Astronomy, under the guidance of M. F. Subbotin, started to provide, as an int national center of work on minor planots, partial (1046) and then 0 The Determination of Orbits complote (1948) calculation and publication of ephemerides for minor-planet searches, ‘These ephomerides are calculated without perturbations for the majority of planets, but for many minor planets absolute perturbations are taken into account or the numerical values of the pertuthations are determined. Tho oxtensive contri- butions of the Institute to the study of absolute perturbations of minor planets remain heyond the limits of our subject. In the Institute interesting investigations wore carried out on the motion of periodic comets; in this connection we shall mention the work of N. I. Tdelson and M. M. Musselius on Comet Mechain= Tuttle (1790 11 = 1926 IV}. Many of our astronomers have worked on the determination of definitive cometary orbits. They were M. A. Vilyev, 8. A. Kazakov, ALA. Mikhailov, I. F. Polak, and others. Often these investiga tions are notable for the groai care with which thoy were made and for the striving to increase, as far as possible, the precision of re- lus. Tho investigations take very carefully into account the per- turbations and derive the most likely positions of the comparison stats by the use of many catalogs, taking into consideration theit systematic corrections for the reduction to a fundamental system; and tho observations of the comets are subjected to attentive study. ‘The investigations by G. N. Neuimin of the motion of the second of the periodic comets discovered by him are of special interest. But the most extensive investigations, profoundly conceived, wore conducted at the Pulkovo Observatory on Comet Encke, which has the shortest period of revolution around tho sun. As a result of many years of calculations and of theoretical investigations by B, Asten, 0. A. Baklund, L. L. Matkovich, and N. I. Melson, the niotion of this comet was studied in a most thorough manner, cover- ing an enormous length of time, starting with the year 1819, when this comet was discovered by Pons and when Encke established its periodicity. ‘The detailed investigation of the secular acceler ation of Comet Encke was accomplished by Baklund, who has shown that this acceleration was moro or less constant from 1819 to 1858 and then sharply decreased; moreover, its value experi~ fences frequent oscillations from one revolution of the comet to another, ‘The study of the motion of minor planets and comets requires a methodical approach and patience during the performance of the long, monotonous, and Inborious calculations. Many prominent theorists have devated decades of strenuous labor to such investi~ gations, These efforts were not wasted: as an example, we may take once again the investigation of the secular acceleration of The Problem ond Its History a Comet Encko, s0 evidently showing that not all the motions in our solar system are entirely governed by the law of universal gravita- tion. “More than 100 years have passed since the time of its di covery and Cont Encke has returned to the sun tens of times, but the reason for its secular acceleration is still awaiting a satistac- tory explanation. This one fact shows that though the methods of theorotieal astronomy have attained a certain completeness, theit practical application to bodies of the solar system promises many Interesting discoveries in the future which will also have signti- cance for cosmogony. ‘On the eehor hand, calculation technology is continuously por- fected. The application of the newest analog computers may’ ex- tremely shorten the most laborious processos of the calculations that ate of interest to us, as, for example, the nuerieal deter nation of the perturbations in rectangular coordinates and the c lation of ophomerides. "In recent. yoars outstanding successes have beon achioved in this field and already at the present tine all the ephemerides of unperturbed minor planets are cateulated inthe Institute of Theoretical Astronomy withthe aid of anelog computers. Those are undoubtedly only the first steps, and in the near future tho new techniques will pocmit ws to solve, in the field of orbit calculation, voluminous problems which were, until now, simply beyond hunan endurance. Tm our country, all conditions have been erented for the flower- ing of seience, Soviet astionony develops rapidly and one may sty with confidence that in tho field ing examined by us, me trill soon oceupy first place. Chapter 2 THE PROBLEM OF TWO BODIES 7. The Low of Universal Gravitation Nowton's law of universal gravitation is formulated in the fol owing manner: Any two material particles attract each other with a force di- rectly proportional to the product of their masses and inversely pro- portional (a the square of the distance betveen them Tn other words, the magnitude of the force of attraction is ex- prossed by the formula en whore F is the force of attenction, m) and mg are the masses of oth particles, r is the distance between them, and &# is the factor of proportionality (the constant of universal gravitation). ‘According to the Law of equality of action and reaction, the force with which the first particle attracts the second is equal in magni- tude and opposite in direction to the force with which the second particle attracts the first. Both of these forces act along a straight line connecting the given particles. It is evident that the factor of proportionality is numerically equal to the magnitude of the force of attraction of two particles with masses equal to unity and at a unit distance from one another. The choico of the units of mensurement is arbitrary. The centimeter gram-second system of units used in physics is not convenient for the study of the motion of astronomical bodies, where mass and distance aro expressed by huge numbers in these units. Gauss proposed to use the semimajor axis of the earth's orbit for the unit of length, the mass of tho sun for the unit of mass, and the mean solar day for the unit of time. By tho same token, the nu- merical value of the constant of gravitation £? is determined (which 2 orn A RNG The Problem of Two Bodies a in physies is designated differently because of solection of other units). ILis shown in the theory of potential how bodies of finite di- mensions attract one another. Formulas derived there find applica~ tion in theoretical astronomy when it is necossary to consider the motion of bodies during their immodiate approach to one another (for exampla, the periodic Comet Brooks If during its approach to Jupiter in 188), but such cases are raroly mot. In general, however, one may consider that bodies of the solar system attract one another mutually as material points. It has al- ready been mentioned above that this is justified for two assump- tions: either if the attracting bodies have spherical form and con- sist of concentric spherical layers of different densities, or if the distance hetwoen the bodies considerably exceeds their dimensions. For the sun and planets, these conditions are observed in suffi- ciont measure. Certain planets have satellites, but this makes little difference. Even the moon has a small mass compared with the earth, and we are able to consider these bodies together as a point mass situated at their common center of inertia. To a still proater oxtent, this is permissible with the remaining planets and their satellites. 8. The Equations of Relative M: We observe the motions of planets and comots in the solar sys- tom as relative motions. We refer their motions to the center of the sun, i.8., to a point whose position we may know directly at any moment. It is natural and convenient to treat them in this way be- ‘cause the apparent motion of the sun dotermines the position of one of tho fundamental planes in astronomy, namely, the plane of the ecliptic. From this is determined the point of the vernal equinox, from which are measured celestial longitade and right ascension. In addition, to study the motion of a minor planet or comet with re~ spect to the common center of inertia of it and the sun has, in fact, no paint, since wa may properly diseopard its insignificant (and moreover unknown) mass. ‘Therefore, the problem of two bodies that we shall consider here assumes the origin of the coordinates at the center of the sun, The mass of the sun $ wo take as unity, and the mass of the second body P (a planet or a comet) we denote by m, in fractions of the solar mass. Assuming that both bodies aro material points and, in equation (2-1) Letting m; = 1, and mz =m, we obtain the ac 4 The Determination of Orbits coloration of body P by dividing the magnitude of the force by the corresponding mass Horo 7, as before, denotes the distance, SP, oF the radius voo~ tor of point P. ‘The accoleration of exch body is directed toward the othor body; consequently, the acceleration of body P with re- spoct to tho sun is given by the sum of both accelerations jrith piss, (22) If wo donote tho Cartosian coordinates of point P by z, 9, 25 thon the projection of acceleration j on tha coordinate axes will be ee em S dee fy aa 2. ote -armS es) eye 2 Tee ems. ‘The last expressions are obtained by multiplying the magnitude of tho acceleration (equation (2-2)) by its direction cosines: cos (j, 2) = ~ €08 (7, 2) = = 08 (j, y= - 08 (We = 4 cos (j, 2) = - cos (7, 2) = ~ 2. On the supposition that r does not vanish, the system of dif- forontial equations (2-3) admits a rigorous solution in which the coordinates of body P are represented in the form of functions of time and arbitrary constants. The number of the Inttor will be equal to the order of the system, i.e., six. Tho arbitrary constants in The Problem of Two Bodies 25 every specific case must he determined from observations; this de- termination may be made by different methods, and, thorefore, the selection of the constants will differ. 9. The Integrals of Areas and Kepler's First Low Tho usual mothod of intogration of the system of equations (2-3) is the following: multiplying these equations respectively by the following groups of factors Olrel-y ~2{| Ofer syl-2] 0 and adding them together properly by pairs, we obtain, whoneo, having carried out the transformation ds ty df de dy Yue? dee ae a ‘and the analogous transformations of the left parts of the other two equations, wo obtain the first intograls of tho difforential equa tions (2-3) da Vue ae. a eH where ¢:, e2, and cy aro arbitrary constants. We multiply further the first of equations (2-4) by 2, the second by y, the thied by 2, % The Determination of Orbits and add them together. We will find ein tony 4 eg2= 0. 3) This, as is known, is the equation of a plane passing through the origin of the coordinates. Since this plane passes through a defi- nite point of space, it is completely determined by two parameters. In fact, having divided equation (2-5) term by term by one of its co- officients that is not equal to zero, say by cy, we obtain : erosytoie=0, (26) where ef = ea/e1 and o§ = es/eu i Returning to equations: - . tare) (2-4) ne clatty dere mare ing. Multiplying both sides of each by dt, wo obtain yde~adyexdt, ade ede ~cydt, dt. Figs Let the moving body at the moment £ be at point P with coordinates 2, y,2,and abmomenté + dtbe at point P’ with coordi- nates 2 +de, y+ dy, 2+ da (Rig. 1). Thon, dy ~ y dz expresses twice the area of triangle SQQ’, which is the projection of triangle SPP” on plane 2Sy. Therefore, the aroa SQQ” equals the area SPP” multiplied by the cosine of the angle hotween the plane of motion of the body ASA’ and the plane Sy. This angle is equal to the angle (q,) between the normals to both planes, and the area of tri- angle SPP’ may be considered equal to the area de of the elemental sector SPP’ of the trajectory. Hence, we obtain dy ~yde aedy ~ yde =2 cos (n,2)do = endt, and analogously yde ~ ady = 2c0s (n,2)do = crdt, 2dr ~ 2dz =2 e058 (n,y)do = cadt, and further 2 cos (m2) 2 x0, The Problem of Two Bodies a de 2 €08 (ny) 42 = ea, a we e (cont.) 20s (ny2) 22 = ca, a whence ode. VT r ane, es) whieh aftr intogration gives Bon otre, es» Wo come to Koplor’s First Law: Eaoh body revolving around the sun moves in a plane passing trough the center of the sun; moreover, the area being described by the radius vector af the body changes proportionally with time. 10, The Integral of Vis Vive and Kepler's Second Low We return to equations (2-8). Utilizing the fact that the motion of body P lies in a plane, we combine the coordinate plane 28y with the plane of this motion. ‘Thon the coordinate 2 and its de- rivatives will be equal to zero, and we aro loft with only the first ‘ovo of equations (2-8) (2-10) de zi ware 2, of cours, has the provous value (since 08 (8) inthe last oquation of (2-1) is now equal to unity). For finding further integrals we multiply the first equation of de dy 2-10) by 2“, the second by 2 and add: (2-10) by 2, th by 2.57 and ada de 82 dy dy (de dy of Se, EY) oa imale Ss + j wa a) POM aaa =e, eu 28 The Determination of Orbits Sinco we have af de ara ‘Transforming also the lft part, we obtain d [fde\ {dy\* afi AY ("1 ona om 2 (2). allay: Gy 20-926) Intograting, we will find (ae)? (ay? _ 2620+ md nt v () (2)- BEM oy. ee) ‘This retation is called the integral of vis viva. On the left it con- tains the square of the velocity of the body, ¥?. The magnitude is independent of the selection of the coordinate axes, and, therefore, the constant cg will be independent of the coordinate system. ‘Transforming to polar coordinates by means of the substitutions wercosu, y=rsinu, we transform equations (2-11) and (2-12) to the form adu dt aod oo, 13) aa (dr)*, ,afdu?_oetem) . v -(%) (2) HEM. aH Substituting in (2-14) ae at we obtain ‘Thus, we have excluded time and found the differential equation of the trajectory of the body. Having performed the following sim- ple transformations: meee RET | The Problem of Two Bodies » (a) -EeG ae and introducing for simplification of notation whieh gives @-18) we find the following intogeal sare cost 5 + deem b= A cos (ue). @16) Substituting hore for A and A, we have opine YEE 05 008 w= 00) and re Ose) 5 ey 1a Yhe eg Shprcon te e0 ‘This is tho equation of a conic section with the focus at the origin of the coordinatos, an equation that Ts usually writen inthe form ? ete 38 T¥e cos v" Ce whore p is the parameter, ¢ ts the eccentricity, and 2 is the angle Towwecn the radius vector rand the polar axis direction along the mnajor axis toward the pertelion, wherein the given case 30 The Determination of Orbits Thus, we come to the generalized form of Kepler's Second Law: Bodies moving around the sun describe conic sections, at one of the foci of which is found the sun. 11, Kepler's Third Law and Gouss’s Constant During the integration of the equations of motion we introduced six arbitrary constants, the meaning of which is now to be ‘established. From the first equation (2-19) we find kViv mvp. (2-20) Substituting this exprossion for ¢ in equation (2-0): Bonk Tempe tes. (21) Let us assume that the motion takes place along an ellipse. Thon after a lapse of a period of revolution P, the body returns to the initial point; moreover, after that time the radius voctor de- seribes the whole area of tho ollipso, i.c., for the moment ¢ + P we will have on Vervepvo? 2( + nab) VIF VP + P)+ ey, where a and 5 are the semiaxes of the ollipse. Subteacting from this equation (2-21) wo obtain nab kit m ypP. propertics of an ellipse it follows that But from the goomet beavi-e?, p=a(l-e%). Substituting this, we have ano ak VIvmP. 222) ‘This equation applies to every planet and periodic comot in the solar system. Thus, it is possible to write it in the following form: feta? Antat Pin) “PFA em) where a,41,..2)P,Piseesmymisere fofer to all bodies revelving around the sun in elliptical orbits. From here Keplor's Thied Law is obtained in its generalized form: sek, @-28) eames eat emREEEATION een The Problem of Two Bodies a For bodies of the solar system, the squares of the periods of revolution multiplied by the sum of the masses of each body and the sun are proportional to the cubes of the semimajor azes of the orbits. However, if the masses of the planets are neglected in compati- son with the mass of the sun, we obtain the usual formulation of Kepler's Thitd Law. This law permits us to find the numerical value of the constant of gravitation: it is sufficient to know, for one of the planets, the mass, the period of revolution, and the semimajor axis of the orbit. For the earth, Gauss took m= 1/354710, the length of the sidereal year P ~ 365.2053835 mean solar days, and a= 1, and from (2-23) there follows the formula: (2-25) Gauss computed . = 0.01720209895, 255814414 - 10. log k The values adopted by Gauss for P, and particularly for m, are inaccurate, However, it would be very inconvenient to change now the value of the Gaussian constant (it would be necossary to ro- calculate all auxiliary tables in which the value of this constant enters). However, a change in & is not really necessary, bocause it is possible to renounce the assumption that the semimajor axis of the torresttial orhit is strictly equal to unity and to determine it, con- sidering that & is given, from equation (2-22), namely, o- ERP) . (2-28) ‘Thus, the value of the basic unit of length in the solar system, the so-called astronomical unit, is determined through the Gaussian constant: a planet, having the mass and period of revolution as- sumed by Gauss for the earth, would revolvo around the sun (in & eireular orbit) at the distance of one astronomical ‘According to Newcomb, the sum of the masses of the earth and the moon m= 1/329390, and P = 365.256386042 monn solar days (for the year 1900); from which log @ = 0.000000013, i.e., differing from zero only in tho eighth decimal place. 2 The Determination of Orbits 12, The Determination of the Form of on Orbit From equations (2-19) the constant es can be found Gee am? et) Bee) . e ? Be G+m Subatiting this vl in (219) wai rovers (2), eas ‘The last equation permits one to judge the form of the orbit by the magnitude of the velocity at a given position, From equation (2-28) it is soen that the semimajor axis of the orbit really determines only the magnitude of the velocity at the corresponding position for a known or zero mass m of the body, and for a given magnitude of the radius vector r of the body. Thus, we have: Pony 22.029), 95.9 (Onis an allipso), por? 2228) ge (onitis apartol, | 29) For v2 > 24° Cm) 4 90°, then the motion is called retro grade. The inclination cannot exceed 180° or the ascending node loses its meaning. ‘Let us consider the spherical triangle nasi, in which ail = 90°, zit= il, and Zesin = 90° ~ i, From it we find cos (m2) = sin sine 34 The Determination of Orbits Turning to triangle myst, in which ysl= 90° i and Zystn 90° + i, we have cos (a, y) == sin Fos Ste Finally, because (172) = 608 (n, 2) = €08 f After this, equations (2-4), (2-7), and (2-8) in connection with (2-20) sive us 6 sin isin Siek VT4m Vp sin isin &, de dt a de = =o sini cos st=~ hk yivm yp sini cost, } (2-81) =e cos i=k V1 +m yp cosi. ‘This form of the integral of areas establishes the connection between the coordinates of body P and thoir derivatives on one side, and the quantities p, # 2, replacing our previous constants of integration ¢:, ¢2, ¢3 (a8 well as ¢ = VeF on the other. It remains to determine the orientation of the axis of the orbit in its plane, In the coordinate system introduced in Section 10, in which the plane zy coincides with the plane of tho orbit, the direc- tion of the a-axis remained arbitrary. We combine it (Fig. 2) with the line Si and note on the projection AA” of the orbit on the ce- loatial sphoro the point If on the direction to the perihelion of the orbit and the point P on to the body. ‘Then, in the last of equations (2-19), » denotes the angle SiSP, and » denotes the angle ISP, while the angle ee 18 then evidently equal to the angle {1SI1, It is customary to call u the argument of latitude of the body, » tho true anomaly, and cy, which is desige nated by the symbol «, the argument of perihelion. Instead of the indicated equation, we consequontly shall write o (2-82) In Fig. 8 the orbit together with the new system of coordinates is represented in the plane of the drawing. It is easy to see that 6 can take any value from 0° up to 860° in the direction of motion of the body, and of course this applies not only for the elliptic or- ee ee The Problem of Two Bodies 35 Fige 3 bit represented in the drawing but also éor the other forms of conic sections. If the celestial equator is taken as the fundamental plane, then the longitude of the node, the inclination of the orbit, and the argue ment of perihelion can be designated by S15 i% 0% 14, The Elements of an Orbit The last constant that still remains for us to consider is eg. In equation (2-9) o represents the area described by the radius vector, starting from some initial position. It is easiest to take as initial that position in which the radius vector is directed toward the porihelion of the orbit, We shall designate the moment of passage ff the body through perihelion by T, then at that moment o = consequently, cone, (23) and equation (2-9) will give Qonc(t-T)=k ITM yp-T) ean ‘Tho six constants of integration are now expressed by six other constants, the elements of the orbit. These new constants ate the followin 4 the inclination of the orbit to the ecliptic. 5 the longitude of the ascending node, the argument of perihelion (the angular distance of pei- helion from the asconding node). 36 .¢ three elements give the position of the plane of the orbit ice and the direction of the major axis of the orbit, ‘They de- pend upon the selection of the coordinate system. If these ele- ments are referred to the plane of the equator, their designations are i’, 5 0". a= the semimajor axis. @ = the eccentricity. ‘The size and the form of the orbit are determined with the aid of these elements. T ~ the moment of passage through perihelion. It fixes the position of the body in the orbit at a definite mo- Certain of the enumerated elements can be replaced by others. Thus, ina parabola, the major axis hecomes infinite and in place of a the perihelion distance 9 is introduced; in other words, the distance from the vertex of the parabola to the sun. Tt is also con- venient to make this replacement for noarly parabolic orbits. It is also possible to introduce other elements for an ellipse in place of 4, e, and T. We shall mention further that among the orbit elements being determined, it should be necessary, strictly speaking, to con~ sidor a seventh quantity, the mass of the body. But only the masses of tho major planets play an essential role, and it is possible to neglect the masses of the rest of the bodies of the solar system in comparison with the mass of the sun. 15, Motion Along an Ellipse ond « Circle Conic sections possess many interesting geometric properties. However, for the study of the motions of celestial bodies, only the general equation of these curves is required. ‘This equation can be iziven in different forms; for example, in polar coordinates with the origin at the focus: (a5) ‘To this equation we add the intogral of areas (2-13), but pre~ liminarily wo replace du by dv on the basis of (2-19) and substitute the value of e from (2-20) 2 do Jivmi ype kytvm yatta. ee byT po kyiem yale). (2-36) The Problem of Two Bodies 37 It is evident that equations (2-85) and (2-28) are sufficient to obtain the dependence of the coordinates r and e on time; however, it will be convenient to find separately the solution for each kind of conic section. We begin with the ellipse, and in this case we introduce, for facilitating integration, the auxiliary variable E de- termined in the following way. Making use of the system of Cartesian coordinates € and 7, the axes which are directed along the principal axes of the ellipse (Fig. 4), then in place of (2-85) we shall have the equation ee. . Gen (231) We reduce this equation to parametrie fom assuming f-acosk, yebsink 38) ‘Transferring the origin of the coordinates from the center o to the focus S, preserving the direction of the" axes, and designating the new coordinates by 2 and y, we shall have the following formulas for transformation of coordinates: te f-ae=a(cosk~e), y=nebsing. Now passing again to the polar coordinates r and » we find 1 cos » = a(cos fi ~ e), eas rsinv=bsin£ =a yi—e? sin &. aed Squaring these equalities, adding them, and then exteacting the root, we obtain rea(l~e cos £). (2-40) ‘The geometeical significance of F is the following (Fig. 4). We escribe about the center of the ollipse o a circumference with ra dins ofl = a, Thon r = SP, 9 = ZIISP,£ = MOP’, where the point P* jes on the intersection with the circumference of the perpendicular to the major axis passing through P. It is also evident from tho Figure that 1 cos v= SK = OK ~ 0S = (cos & e). 38 The Determinotion of Orbits If in equation (2-87) we make will be the equation of the circumference circumserib- ing the ellipse. This ex presses one of the well- known properties of an el- Tipse, namely, that all of its ordinates are related to the Flas 4 corresponding. ordinates of the cireumseribed circle as boa KP OB _b 2 OB barra. 06a K But, because KP =r sine, KP’ ~ asin E, we have the depend~ rsinv-ayI-e? sink. We supplement equations (2-80) and (2-40) by certain relations ‘among r, ®, and £ following from them, which can be useful in cer~ tain eases. Combining first by addition and thon by subtraction ‘equation (2-40) and the first of equations (2-89) we obtain #(1 +c03 a) =a (1 ~e) (1+ 008 F), 1(L = cos 2) =a (1 +¢)(1~ cos Fy for, extracting the root Vi cos $ p= ya(T=e) cos $ . ” et) vi sind e «ya te) sin $8, and further (ee) 1 tan The Problem of Two Bodies 9 Now it is possible to begin the integration of equation (2-36). Differentiating equation (2-£2) we find do five ak cost Le cos? whence, using the first of equations (2-41) we obtain cs dua yIzarae, Substituting this expression, and also the expression (2-40) for + in equation (2-88), we have 290. 4? TeT — ¢ c08 £) eo -hytvm vat), (1-6 00 By a6 YET™ ay, whence, after the integration, it follows that B-esing ASM Gn, (2-43) whore Tis an arbitrary constant, which is equal to the time of the passage of the body through perihelion, because at porihelion B = 0, If we now substitute the value of # from (2-28) into (2-43), we obtain the equation Breone-Sa-repe-n, 40 called Kepler's equation. In it 4 designates the mean daily motion of the body, or the angle described in a moan day by an imaginary point moving with constant velocity in a citcle around the sun and accomplishing a full revolution in P days, the period of revolution of the body But then it is evident that ¢ (¢- 7) is the angle described by our imayiinary point from the moment of its perihelion passage, Ks accepted designation is the lottor M, and it is called the mean anomaly. As for the angle B, it is called the eccentric anomaly. We can write Kepler's equation also in the form EnesinE=w(t-T)=W=Mo+u (eto), (2-45) 0 The Determination of Orbits where fo denotes a cortain initial epoch, and, consequently, My = hello ~T) is the value of the mean anomoly at epoch fo. The quantity My ean thus replace T as an clement of an elliptic orbit. Precisely in the same way the element »» can serve in place of a, because from a ‘comparison of (2-13) and (2+44) i¢ follows that (2-46) and conversely : A -( vie) ae en Finally, instead of the eccentricity ¢, the angle of eccentricity 6 is often introduced, being defined by the equality (2-48) As can be seen in Fig. 4, ¢ is the angle at point B, situated at the end of the semiminor axis 5, and subtended by the segment OS = ae. Consequently, in view of b = a yT-e3, Ss . sin = sing SB = VBR Va? a = We are, therefore, able to replace in equations (2-39), (2-40), 2-41), an (242), for the radius yeetor and tho true anomaly, i= e by cos 6, Vie by y2 sin (45° + Nea), VI ¢ by V8 eos Use +465), and YI we/yt ~e by tan (45° + ed), which is some- imes convenient. vm Geplar's envition is transcendental. Certain methods frit s0- lution will be indicated below. Let us prove that Kepler's equa- tion always has one root. We shall writo it in the form ME) = B =e sin B= M0, then it is evident that {(-=) =~, f(s) > +=. At the same time the derivative age) ak is positive since ¢ <1. Consequently, the function /(F) mono- tonically increases and thus has one root. The Problem of Two Bodies a ‘The special case of the ellipse for e=0 is a circle. In this case e it follows from equation (2-85) that epee a are dope \ equation (2-45) and equation (2-42) give v= EM. Motion is accom A a lished with constant angular ve- locity, and sinee the direction of the major axis cannot be estab. lished, the initial point for the measurement of the angles on the orbit romains arbitrary. Tt is pos- sible, in particular, to take for the initial point the node Si and thus to introduce, in place of v, the argument of latitude u (Fig. 8). Kep- let's equation (2-45) becomes =o + w(t ta), (49) where is given as before by equation (2-46). Fla. 5 16, Motion Along @ Porabola The eccentricity ¢ of a parabola equals unity, and the semi- major axis q is infinite, In consequence of this, as equations (2-46) tand (2-45) show, the mean daily motion y, the mean anomaly M, and the eccentric anomaly H vanish, However, here it is completely unnecessary to introduce these quantities, hecause the problem is solved by direct integration. From equation (2-85) it follows that ro P_ P (2-50) where ¢ is the perihelion distance equal to p/2, since += when v=0. The radius vector for a parabola at v= 180° becomes in- finite. The branches of the parabola do not close at aphelion, and thorofore the true anomaly ix measured from —180° tos 180° Equation (2-36) assumes the form 22. eVian POT VEk VE om Vi ani on the basis of (2-50) we find Bk yinw 2 ‘The Determination of Orbits ‘Tho left side of this equation can be expressed differently; j 2 fled) a ft eat eR WEVETA fa, and this inmediately gives 2 (tan bes Etant Lo) 57a yarn au aps alsin P In the given case, T is again the moment of perihelion passage, for which \ both parts of the equation vanish. By analogy ith on ellipse, the quantity (t= 1/0", proportional to the time In elapsed from the moment of perihelion $7] passage, is also denoted by M, although, of course, it now has quite another meaning. Equation (2-51), a cubic in relation to (tan 2/2), can be solved by direct means. In the same way as Kepler's equation, it gives one solu- tion for an arbitrary time, ¢. written Having 2 4) Land La) = was (meio a) we ato that 6(-180°) = == and «(+ 180°) = 2-82) grrcalst ew fand that roves our assertion, In Fig. 6, Meg, 5 eae The Problem of Two Bodies a 17. Motion Along @ Hyperbole It follows feom the formula p= a(t ~ e%) that a is negative for hyperbola, because > 1. The major axis lies on the other side of the curve from the focus and is contained hetween the vertices of both branches of the hyperbola (Fig. 7). Of course, under the attraction of the sun, the body deseribes only one of these branches, that which is concave towar the sun. The other branch of the hy- perbola could be described by a point repelled by the sun aecord- ing to the law of inverse proportionality to the square of the dis- tance. This caso will be of no interest to us here. From equation (2-85) it is possible to soo that + bo- comes infinite for voto (2) Designing cos (2) by we conclude that during motion along a hyperbola, the true anomaly changes between the limits of (180° ~ ¥) and 180° ~y. For values of » lying outside these limits, we obtain a negative radius vector. In this ease, the exten- sion of the radius vector beyond the sun, in the opposite direction, intersects the second hranch of the hyperbola, which, of course, will no longer correspond to real In order to determine the dependence of r and » on time we can apply the same method as for the ellipse, since all the equations of Section 15 keep their same form for the hyperbola, and in exactly the same way the parameter £ may be introduced that appeared as the eccentric anomaly in the case of the ellipse. In place of equation (2-87) we will have Fig. 7 4 The Determination of Orbits with the axis situated as shown in Fig. 7. But it is more conven- ient to keep equation (2-87) unchansed, considering b as an imagi- nary number, and determining the semiminor axis by the same for- mula as for the ellipse, beavi ‘Thon from the parametric equations (2-88) it follows that sin & is also an imaginary quantity, because it is necessary for m to re~ main real. However, it is not possible to keep imaginary quanti- ties in the final formulas, otherwise those formulas would not be suitable for calculations. ‘Therefore, hyperbolic functions are used hy introducing the new real argument It Heik, sinhif-isinB, coshH=cosk, (2-54) and the absolute value of the somimajor axis |@|. Tt is not difti- cult to introduce this roplacement in all the formulas for elliptic rrotion; obtainod consecutively for the hyperbola are: reos v= a(eosh I~ e) =| al (¢ ~ cosh H), reine! a] VE=e sinh 58) reali —e cosh M)=[aj (ecosht=1), @80) 7 L re 1 tan bowl YESS tanh (251) In place of Kepler's equation we now find M_esinb Ht kyIFH yp e sinh HH + ™(t—Ty=v(t-T)=N, aie tho quantity » heing dotormined by the equality (2-59) By analogy with the ellipse, » is also sometimes called the mean daily. m Nis the mean anomaly for a hyperbola, although now, of course, these quantities ean in no way be connected with the period of revolution, which does not exist for open hyperbolic The Problem of Two Bodies “5 orbits. Notice, incidentally, that formulas (2-48) and (2-44) give an imaginary P and for nogative a, ‘Turning once again to Fig. 7, and designating the coordinates with the origin at the focus by 2 and y, we notice that for a transi- tion from and 7 to z and y it is necessary to transfer the origin fof the coordinates to the left along the axis of the abscissa by the length ae, Hence, we have z=reos v= £~ae -acos ae = al (e~ cosh Hi), yersinv=n= being =|alt-e? sin E = |a|Ve™=1 sinh H, in agreement with equations (2-55). Tt is necessary to remember that here we make use of the same equations (2-88) for £ and 1 a5, in the case of the ellipse. In Fig. 7, moreover, r= SP, the peri- holion distance q = SI = a(1~e) =|al (e - 1), and Z1ISP = We shall not dwell more in detail on hyperbolic motion. Rqua- tions (2-55) through (2-59) give all that is necessary for calcula- tions. When desired, other expressions for r and » can be derived from them of from the formulas for an ellipse. It is also rather easy to introduce, in place of hyperbolic functions, trigonometric func tions dependent on a certain parameter. However, this leads to less convenient formulas. Certain comets move close to the sun along nearly parabolic hyperbolas. ‘The cases of nearly parabolic cometary orbits require special attention, he they ellipses or hyperbolas. Hyperbolic or- bits with an eccentricity noticeably different from unity are ex- tremely rare in the solar system, and only occur if some periodic comet approaches so close to Jupiter that its attraction begins to dominate over that of the sun, Such comets describe hyperbolic trajectories around Jupiter. For the calculation of the coordinates of such a comet it is sometimes necessary to make use of our for nla, 18, The Determinotion of the Hellocentric Coordinates of Body It now remains to express the heliocentric coordinates of a body by the constants of integration, i.e., by the elements of the orbit, and thereby to complete the solution of the problem of two bodies. In Fig. § the centor of the sun is taken for the origin of the co- ordinates, the z-axis being directed toward the point of the vernal 4 The Determinotion of Orbits ‘equinox , the y-axis toward the point of the ecliptic with a longi- tude of 90°, and the 2-nxis toward the pole of the ecliptic. AA is the orbit, and P is the position of the body determined by the Le G LX Le Figs 8 radius vector + and the angle u, We introduce first the auxiliary axes #1, Ys, 21, directing axis 2, along the line of nodes and lo- cating the axis y, in the corresponding manner in the plane of the ccliptic, Further, let the spherical coordinates of the body be de- noted by 7, 1, 6 where I is the holiocentric longitude measured from 0° to 360°, and 5 is the heliocentric Latitude of the body measured from -90° to +90°. Thon the longitude of the hody measured from the line of nodes Soi will be 7 ~ 2, and by the usual formulas of transformation of coordinates we obtain #1 =F €08 D608 (I~ 2), ay = c0s b sin (I~ 52), arersin 5 But these same quantities can be expressed with the aid of + and 1. We note on the celestial sphere (Fig. 9) the intersections of the axes z (point T), y, 2, of the directions Sil, and SP (points {i P), and the intersection of the sphere by the plane of the orbit (great ciole 4A’). We draw the spherical perpondicular PK to the The Problem of Two Bodies a Fig. 9 ecliptic BE’. Then, from the spherical triangle SPR, NP =u, AK 1 il, PRB, LRP = 3, we find cos w= cos B cos (2-2), in whieh sin u cos i = cos 8 sin (1~ 2), sin usin é= sind, consequently’, 21 = 1008 } cos (1-2) =rcos u, yr =reos sin (I~) =r sin wos é, (2-60) rind a aE , the application of the derivative F = ST, In the first approxi tion, let & be the value taken; with it M; is caleulated hy Kepler's equation, so that By —esin Ey =a Then, neglecting the higher orders of = M;, we will have BaP Le FQ M)y (-28) where, evidently, @-29) ire co ‘The value of E obtained is substituted again in Kepler's equa- tion, and if it is not satisfied, the method is repeated. It is not worth while computing P according to equation (3-29); in the first place, it can he tabulated (191, and secondly, it is easy to take it from tables giving E according to the argument M, because the dif- feronce of successive values of E, divided by the interval of the argument, is exactly equal to F. Example. Given ¢ = 0.627411, iM = 296° 18°40’'6. We find the approximate value of B from the table of M. F. Subbutin [13]. For interpolation and other auxiliary calculations, it is convenient to use a slide rule. Bocauso the table does not give E, but £-M, to find P it is necessary here to take the dif- ference of adjacent values of F ~ M increased by a unit. We cal- culate first with the aid of an arithmomoter e”” =208264.81 ¢ = 12911978, Wo find E = -3524T » ~35°28", F Subsequent schome of solution: 0.91. The following is the Geocentric Motion 4 1 2 3 260° 51° -60°49'2470 260° 49°24779 sin 0.987275 = 0.987201 = 0.987201 a 19778670 = 19775074 Beni =85° 90°96 BOP29'1G74 98° 29°18%"4 296°20°2670 290784074 298° 18°4076 MM “Vt +072 o Fy 12670 4002 ‘The quantity of F anomaly Example ofa calculation with logarithms. Given ¢ = 26° 9°5/47, w= 86°40'19771. bo the sought for value of the eecentrie sing 9.6441888 irae” 5.9144251 + 21°98" ea 4,9586139 E Bes BBP OrSBT0 58° 0°52720 sin £ 9,9990504 9,9280816 9.928617 ev sinE 4.8876043 4.875752 48875756 E—M= 21° 2648736 91°31 (MW) = Ber4yi1764 36°40" Ma (M)= 51798 + fl CNl= 6870 + 21° 26°827'59 36° 4019771 0 OF all the formulas for the calculation of rand », it is usually convenient to make uso of equations (2-39): cos $ sin F = ayia oF sin B, Mia tiacatoinsitageeetod 6 and subsequently, to find the heliocentric coordinates with (3-18) and (3-18), where in tho right-hand part of (2-18) it is possible 10 substitute (3-80), which will give: 2 aP, (cos F~e)+acos d Qe sin F, y= aPy (cos Be) + acon $ Qy sin By (at) AP, (cos Ee) +acos f Qz sin E. Designating, Ag=aPe, By = 40056 Gey Ay-aP,, By =ac0s 4 Q, (8-39) A,=aPs, Be =acosd Qz, a The Determination of Orbits we can write: #2 A, (cos F =e) + Be sin By y= Ay (eos B= e) + B, sin 2 =A, (cos B= e) + Bs sin (3-83) In turn, for calculation with logarithms itis possible to derive similar forpas in-place. of (16), and by the Same token to c0- placer ante by E, but iis considerably less convenient lo use thom than (3-23) Tr great provision is not required, and also if the oceonttcity of the orbit is not vory promt, the cieuation of the holiocenteic coontinates, can he slightly shorianed chrough the use of special tables giving directly ceriain expressions doponding on rand sith arguments Mande (or 8). Yror‘computations with # calculating machino it 18 possible to use tablos by Innes, containing fivesplace values of © cos v and % sin » (losignated by X and Y) forall yaluos of ¢ fom 0,01 to 1.00 {BEI “Phe table of Stracke gives these same quantities with four figures, but only up to @ = 25°. Tho tables of Peters, intended fee caiculation with fourplace logarithns, give = and log F ap to ga24? (Bly 38h see also tho tables of Stracke with three figures (371). Making use of the quantities Ca=eose, 8 fsinw, ay one may on the basis of (8-30) rewrite (9-81) in such a form: 2 AO + BLS, ye AC 4 BSS, (8-35) 20d, 04 BIS, where Ags aPy, BE=aQs, A,=aPj, By =aQy, (30) A, =aP,, Be=aQe ‘The time saved with the aid of these tables should not be over- costimated hecause interpolation in them is rather disagreeable. ‘As to citcular orbits, they serve, as a cule, for the ealculation of positions with only limited precision. Equations (2-46) and Geocentric Motion a (2-49) permit the finding of x In equations (3-16) it is necessary, ‘of course, to replace r with @ In equations (8-18) and (3-20) the angle @ is not determined, and one may take advantage of this arbitrariness later on for conveniently obtaining the vector ele- ments; computing conditionally, then, v from a certain point TI, we will have: 2 = aP, cos v4aQ, sin 9, = oP, cos ¥ 4a, sin», @ean 224P, cos v4 aQ, sin v. 22. The Calevlation of Heliocentrie Coordinates for 0 Porabolic Orbit Equation (2-51) Ly Luge by). Eat (ton $0 Stan fo). pM (38) can, certainly, be solved for tan $v by algebraic means, but it is for simpler to use tables giving v with the argument or, some- limes, vice versa. ‘The most detailed and convenient table for use, especially if the calculations are made with a greater precision, is the table of Oppolzer (301, which gives Mf or log M for values of » from 0° to 176° in increments of 10°. ‘There are many other, less dotailed tables contained in all collections of tables for theoreti- cal astronomy; sometimes they aro called Barker's table, because ‘one of the first tables for parabolic motion was published in 1757 by Barker. ‘The radius vector is found by equation (2-50) ra gscc? Jv = g(1 + tan? 40); 39) after obtaining r one may make use of equations (9-18). However, it ig convenient, without calculating r, to write them in the form ze qsinasin(4+ 0+ 0) 00? 1, y=qsin 8 sin (D+ w+ v)s00% du, (3-40) sin osin(C 4a +2) sec? du, whore the values q sin a, ete, are found for a for all. For calculations with an arithmometer, it is better to make use of tan $2 in place of 2. Designating ven orbit once and o=tngo @-41) 64 The Determination of Orbits and turning to (8-89), we see that eos v= q 8007 49(2 cos* 0-1) a(t 0%), rein v= 29sec? du sin Jv cos fv = Consequently, in place of (3-18) we have: Fe a,(1~07) + deo, y= ay (1-02) + bo, @-2) 2ea,(1 0%) +b, 0, aye gay de 24 Qey ay 9gly, by = 290, (0-43) p= gP bea 2G0n Tho quantity o is given with argument in the tables added to the texts of M. F. Subbotin and A. Y. Orlov (18, 10], and in other collections of tables. We observe that M is designated by M. F. Subbotin by B. At tho ond of this book are given tables of M with argument o (Table II): such organization of the tables allows values of Mf to he given with sufficient (seven-place) precision for the entire range of the argument. The third difference as seen from the form of equation (3-38) is constant and will not have to be taken into account by the user of the table. Near the end of the table one may find Mf, or conversely o, with more than seven decimals. Such cases will he found for comets with a small perihelion dis- tanco; the factors @,, b,, ete., will be small; on the contrary, a, and particularly 1~, are larue. For finding the coordinates 2, 1y, # with seven-place precision it will be necessary to know @ more precisely than (o seven decimals. The table also moots this re- quirement. If» approaches 180°, any tables giving the connection between ‘» and M hecome inconvenient, since ¥ in this range bogins to in- crease very quickly, and for y= 180°, 4. Simplo methods have been proposed to avoid this difficulty, but it is haedly worth while to pause for them, because such cases may be encountered very rarely. 23. Calculation of Heliocentric Coordinates for Neorly Porabolic Orbits Many comets move along elliptical of hyperbolic orbits with an Mricity differing insignificantly from unity, The formulas wo have derived for finding rand v become inconvenient in their usual form. Indeed, if in Kepler's equation (2-43), Geocentric Motion 65 ee dye-ne ¢ tends toward unity, and @ toward infinity; the left-hand side bo- cones a small difference between two similar quantities and that is always disadvantageous for the solution of equations. Heinrich Bruns has shown [18] that one may avoid this ineon- vonience by representing Kepler's equation in the form W=B~sinE+(.-e)sin If the means are available for rapid and precise finding of B sin E, loss of precision does not occur, because (1 ~ e) sin E is always small and may be calculated with complete roliability. Nevertheless, during this effort one can hardly avoid laborious computations. Almost all the other existing methods, such as tho classical rrothod of Gauss presented in particular in the text of A. Y. Orlov and B. A. Orlov (10], the ingenioys method of M. A. Vilyew for nearly parabolic ellipses [2], and also the method proposed by M. F. Subbotin (13, 14], immediately give v or tan x, which is also Sufficient for finding r. ‘These methods proceed by means of suecessive approximations and quickly converge. In othor methods the true anomaly is obtained by direct ealeulation, but on tho other hand it does not always provide the necessary precision. Tho direct mothod af Oppolzer is quite preciso and convenient for cal- culations, He requires sufficiently extensive tables [30], and at the ens of this book they are given in abbroviated form. The dori- vation of the necessary formulas by Oppolzer is quite long, and we shall only state here its principal features. From oquations (2-35) and (2-36) p re Pde ey pee ive cose ve ke a 2" J as cos where on the right a constant of integration is still implied. Also as before, suppose 9 = tan de, then we shall have follows ey) 65 Tho Determination of Orbits Introtucing o in place of » as tho variable of integration, we shall obtain (9-44) in the form A (Ls od rp J itreva-oaP We replace p by g, the perihelion distance, pra(t~e%)-g(l +e) @-45) and introduce for abbreviation the designation toe fea: @-46) Afier multiplication of both sides by (1 + )® we shall write the previous integral as Mylve ii do odo [es/itstels| (ame co men [noncome 7) 2q™ (i+ 07)? (1+ 07) on I: is possible to carty out the integration of this exprossion in closed form, and this would give us the equations of motion along an ellipse and hyperbola, but they are unnecessary now. llowevor, it wo place © ~ 0, that 18 ¢ ~1, the sual cubic equation for ¢ in parabolic motion is immediately obtained. Since the parabola can how be consiored asa special caso, wo shall integrate (8-41), expanding the expressions under the inceprals in & series in powers of eo?, which gives deyt ee peu (Jeet sg ctot— > ge } (48) We consiiler the constant of integration equal to zoro, and here £ Will be measured from the moment of porihelion passage, when it is evident that o = For the convenient solution of this equation with respect to Oppalvor introduses two now unknowns, + and f, determined by means of two arbitrary conditions. In the first placo, they satisfy the equation MyTTe Nags, Mio (49) If we multiply both patis of this equation by f, it will have the form of the usual parabolie oyuation (3-38), only in place of tan bv Geocentric Motion a we shall enter fz, and in place of if wo shall onter @ and if fis knowa, one may employ Barker's table for finding f2. Secondly, the following relationship between o and e must exist: onal + Aven? + Age?ats...), (8-50) in which 4,, 4a, ..-are functions of € only. Replacing the left-hand part of (3-48) with the terms of (9-49), ‘and on the right substituting for a the expression in x given by the condition of (8-50), we shall find erdpareeset (A,-erd 4a? (Ay 2A 4B e2 + (Ar -Re} a... (51) whence with the aid of the method of undetermined coefficients we shall obtain fP-14Be(Ay 2); (3-52) by the samo means after extensive and complicated computations, which Oppolzer states in detnil, tho coofficients 4,, da, ... can be found in a form of series in powers of £: Gi (3-53) By the same token the problem is solved in principle, hecause now with the aid of (8-52) and (8-53) the quantity / can be tabu- ated as a function of €; then, as wo already saw above, fx is found in Barker's table. After that there remains the use of equa- Giun (8-50) to obtain o = tan Lv, according to the determined value of 2 Tis, however, impossible to do that directly, because in practice this would require excessively Inrze tables with two entries in = and €, This stage of solution of the problem requires additional transformations. Using the expressions of (2-58), we shall rewrite equation (3-50) in the following Form: 68 ‘The Determination of Orbits ones Ajeet + Ageteta. sft +(2- 3 +) eet se (6-54) here wo will neglect for simplicity all terms with e* and higher. Let us assume pea, 2 (8-55) ee ay nt reetate array ee sina ove(ued .). introtcing 2, a Ga14 2 ne Sonreeny (3-56) 5 175 i ) whore the coefficients, excopting the first, are the constant terms of the expansions of (3-53), one may write =e, (3-57) where @ is a function of n only. The immediate utility of this ex- pression is that, with a very high degree of approximation, it de- termines the value a and, for the majority of practical eases, gives fa result correct up to seven figures. But if we write, for full rigor, a=20H, (3-58) where H is a function of # and €, both variables entering in the initial equation (3-50), or else (it makes no difference) of mand &, then # will differ very little from unity. I is rather difficult to find an explicit expression for H, and Oppolzer prefers to obtain H from (8-58) hy comparing the many values of ¢ calculated, on the one hand, according to equation (-57) and, on the other, according to the exact equation (3-50). ‘AL the end of this book are added tables giving f and D for the argument ¢ (Table Ill), @ for the argument n (Table IV), and H for the arguments © and n (Table V), where log is expressed in units of the seventh decimal, Those tables include all cases in which it is inconvenient to compute v with the usual formulas; they were borrowed from Oppolzer and shortened: namely, the Geocentric Motion 0 limits in & are here —0.03(e = 1.10) and 4 0.15(e = 0.74), and the limits in n are ~0.15 and +0.15. Supposing that @, H, D approximately equal unity, we see that n= Dea? = etan? fv = tan? 4h, whence it follows that the upper limit taken by us corresponds to 15, B= t4e°, can? 2B 1 L Fe 3.8 but under these conditions one may already use Kepler's equation in its usual form without groat difficulties. The principal difficulty in its solution arises because of too largo values of F creating an uncertainty in the search for the precise value of E according to the given M. This difficulty, if ono did not resort to tables with a large number of figures, could he removed only by artificial trans- formations. We have seen that this difficulty no longer arisos for values of F for which cos # is appreciably less than unity, but it is worth noticing that for nearly parabolic orbits this may exist only for a moment very romote from the perihelion date, ‘And so, we shall have tho following final formulas. Given e, a t, we find f ra B=eD. (8-59) These quantities are constant for a given orbit. For each posi- tion we calculate Meat (8-60) and from Barker's table we take », designating it w because here it doos not represent the true anomaly. ‘Then fe tan 2 4 stan 5 na Bat, 7 1 auton ye een a H. (01) Por the radius vector we shall have according to (3-45) and (3-46) ase) ase) Trecos® teed le 7 The Determination of Orbits One may also find g(l~ 07) 1+ (8-63) The calculation of rectangular coordinates 2, y, 2 from rand 9 is done with the aid of the Gaussian constants, in the same way as for an ellipse. Por calculation with an arithmometer, equations (3-03) and the fundamental equation (3-18) are used. Example, Comet 1942 f (Tevzadze 2). Given 1a1499, 0.28183 Tasiest fassso8 © _9,79000 rye oposa7 Ltseald 4, 68°25'3 e° Tasiaza eoeeir70 a? "9.gatay0 = 00039204 9.790080 eo? T.76314 7 9.990816 o.7o0Ts¢ 1 + 0%, 0.140488 2 ‘gages a'ss1468 14 eo? 0.000851 rigM pisoa7s2 Tanta r” oatis6 s.a9914s o.0014088 a.s01216 0.000061 Tsaa7s ° The converse problem--to find the time of perihelion passage, knowing ¢, 9, and v—is much easier to salve, We shall rewrite -48) in the Following form Jor Pros Prot @st) Here P, and Py are functions of ea and can be tabulated for this argument. That was done in Table VI, also borrowed trom Oppolzer. now £ is measured not from the instant T of perihelion passage but in the usual sense, then in the left part of (2-84) in place of ¢ will appear ¢—T, and we can ensily find T: A (Pro Peo (0-68) Vive and, evidently, it is necessary first to calculate €0%. Geocentiie Motion n For the preceding example, the calculations will be arranged thus: Given ¢ = 0.992172, log 7 = 0.131499, v = 63° 25/5872. Ts94524 eo? T.TeaLA eT L.988808 oug724s 202 = +0.0015008 f= T - 00.28140 bsis9668 >, 2ousot4 0.047884 Py 1saTs46 8:790095 Pig 1.856000 91581990 Pros 0.960581 9.572985 = 1.908014 24. Finding Separate Geocentric Positions ond the Calculation of Ephemerides Almost everything required for the calculation of geocentric positions of minor planets and comets is already contained in the previous paragraphs. Depending on the problem to which we apply ourselves, the discussion may concern the calculation of soparate positions ‘of the celestial body or an entire sequence of such posi- tions separated by equal intorvals of time, of a so-called ephemeris, where, however, there will be no essential difference in caleuln- tions. Given bolow is a summary of formulas and certain practical in- structions that ean be useful for the calculation of ephemerides or of isolated positions. If the Gaussian constants or the vector elements are not given, necessary, first of all, to find them from the elements «, {, 4 of the orbit. ‘This operation is porformed once for a given orbit and is independent of the number of positions to be calculated. Those calculations require some time and they must be well con trolled. If it is necessary to find only three or four positions of the celestial body, it is possible, by using ecliptic coordinates fand equations (3-8), to dispense with the calculation of the Gaus- sian constants For the catculation of separate positions control is absent. In view of this it is necessary to verify all calculations carefully. Calculating an ephomoris, wo can apply control at any stage, eheck- ing by the differonees how smoothly all quantities entering in the calculation change from one momont to another. The separate geocentric positions and tho ephemerides are calculated either for the purpose of comparing them with observa tions, or of showing observers beforchand tho apparent path of a colestial body at a future time, thus facilitating the search for it (search ephemerides). In the first ense, the precision must be a justed to the precision of the observations, i.e., it is usuaily nec- n The Determination of Orbits essary to conduct the calculation with six or seven decimals; in the second case, four or, as a last resort, five decimals are suf- ficiont, corresponding to approximately 1° on the celestial sphere. In beginning the calculation of an ephemeris, it is necessary to der the choice of the interval between separate dates. For precise (six-figure) calculations for minor planets, four-day inter- vals are taken; for comets the interval usually must be reduced to two days; but sometimes—for fast geocentric motion, or in the ap- proach to a pole—it is necessary to calculate with much shorter intervals. For intervals of over 4 days, the short-period lunar in- equality in the longitude of the earth’ begins to manifest itself noticeably, thereby disturbing the smooth trend of the geocentric coordinates of the ephemoris. For objects approaching close to the earth, or if higher precision is requirad (for a definitive orbit), the calculations are made with seven decimals. For search ephem- crides giving less precision, the interval often can be increased up to 8 days (as a cule for minor planets) and even up to 16 days. However, even here it is occasionally necessary to take small intervals for comets. ‘A. Auciliary quantities. For calculations with an arithmometer the vector elements are usually borrowed from the calculation of the orbit. If calculations are to be conducted with the aid of the tablos of Innes of Stracke, the quantities 4 and B” are necessary; for a parabolic orbit a and } are needed. The vector elements, as well as o, Sl, i and the coordinates of the sun X, ¥, Z refer to one system of coordinates, tied oither to the mean equinox at the bo- ginning of the year of the observations or to the normal equinox of 1925.0 (up to 1987) and 1950.0 (from 1938). Tf these quantities are not given, they are calculated accord- ing to the following formulas &y =sin Qsin w, Px = 2 ~ a1 c08 i, ag =sin Q cos w, Py = (a2 + B1 00s f) cos €- yx Sin & Ba (2 4 B1 cos 4) sine + ys cos & 008 sin w, Ps Ba = cos 1 co8 w, Qs == Bi ~ 42 008 4, yi=sin isin w, Oy = (a: + B2 cos i) cos €~ ya sine, & (D ya =ain § cos w, Qe =(-a1 + fia cos i) sin € + ya cose, Ay = aP x, By = 0608 b Oxy BL = Oa, Oe = aPey be = 2.9085 Ay = aPy, By = 0008 6Qy, By = aQy, ay = gPys by =29Qy Ay = aPr, Br = 6008 Qs, BE ~aQry @e = 9Pey b: = 29Qe, Geocentric Motion B Control: PLs PRs Pat, +O} 4 QF =1, Pe Qu + PyQy + PrQe = 0. Als Abs AD = a2, B24 BB) BE = 02 cos? 4, B24 B24 Bead, ALBy + AyBy + AyBs = 0, ALBi + Ay BS Ae BE «0. a2 4a} + al = 92 OF sob + od = 4g 0, Dg + By by + Gu be = 0. Concerning the corrections for precession see Section 20; the formulas given there are applied without changes for any three vector elements. Tho Gaussian constants ato ealeulated if logarithms are to be nsin N= sini, sin } sin B = sin SL cos & ncosN =cosicos 2, sin dcos B = ncos (N+ 8, sin asin = cos St, sinesin C=sinSsine, f sin a.cos A= ~cos ¢sinQ, sin coos C = nsin (N +e), Aw Ate, Bia C=Cto Control: sindsincsin@-0)__1.,; sin a cos A B. Eulipticat orbits. The ephemeris givos n positions My = Mo + w(tr ~to)y Man Mi em toe Control: a Ma = Mo + 1(tn ~ to). ‘The subsequent values of M are obtained from those preceding by means of adding wz, where w is the interval of the ephomeris in days. The last My is also calculated independently. m™ The Determinotion of Orbits E-esinE=M, any €° = ST.Q9578e = [1.7581226e, e”” = 206264.81¢ = [F.144251Ie, For the first two of three dates, the approximate values of Fy are taken from Ostrand’s tables: or lacking these tables, they are taken from any other tables giving F with the arguments Mf and ¢. ‘The corrected value of F is My-By-esin ky, B= 8, =F(W~ Mh). Concerning the finding of # soe Section 21, If necessary, the approximations are repeated until Kepler's equation is satisfied exactly. For the thied, fourth, and subsequent moments the values of Fare given approximately by extrapolation of the values already ‘obtained, taking into consideration the higher differences, The smoath trend of these differences gives control of the calculation, of Be For lculation with an arithmomoter pcos 6 cos t= A, (cos Ee) + Basin E+ X, pcos dsin a ~ A, (cos Fe) + By sin E+ Y, ay psin§ -A,(cos Ee) +B, sin E+ Z. If the tables of Innes or Stracke are used, it is not necessary to find the eccentric anomaly. For tho arguments ¢ or y and M the tables give @ and S. Aftor that the following formulas are applied: pcos Bcos a= A, 4 BSS +X, peosdsin x= ALC 4 Bs Ss ¥, av) psin 6 © A,CuBiSs 2% Fer caleulation with logarithms rain v =a cos g sin F, Va Fos v= a(cos E- ey; } sae pcos 8 cos «= rin asin (A 6 0) +X, pcos sin w= rsin bsin (B40) + ¥, (va) pain 6 =rsinosin (0% +0) + If the tables of Peters are available, we take e - Mand log 7 with the arguments and and we find, without calculating the ooonttic anomaly, = M+ (@ =i), al rw a2. Goocentric Motion 5 ‘The a and 5 obtained are controlled by the differences. It is often useful to know the value of p, which is found with not more than theee or four decimals. If the ephemeris is to he compared with very numerous observations, it is possible, by successive application of tho formulas of interpolation to halved intervals, to find a and 8 for each day, which facilitates comparison. However, there is no direct necessity for this. ©. Circular orbits. The precision is to 4 decimals, and the in- terval is 8 days; the ephemeris can be useful only for a short time after the discovery of a new planet while the deviation from the true orbit is not too appreciable. Concerning tho calculation of the vector elements see Section 19. In the Gaussian constants wo have a = 0 Ha Uo + w(t ta), v= to + KE to) a For calculation with an arithmometer pcos 6 cos X= A, cos 0+ Bisinw + Xy pcos 6 sin a= A, cos v+ BZ sinw+ ¥, amy psin 3 =A,cos vy Bsino4 Z. Catculating with logarithns, we have pcos cos & asin asin (Av) +X) Cpe ener (lay psin Basin esin (C4 w+ 2 D. Parabolic orbits. (oe v is taken from Barker's table for the argument M, or else tables giving o = tan 2» are utilized. For calculations with an arithmometer p c08 5 e085 X = 4,(1~0%) + a +X, pcos Bsin t= 0,(1 0%) + bya Vy any psin 8 =a,(1-0%)+ 0+ 2 For calculation with logarithms r= qsec? Jn, (allay % pcos 5 cos x =r sin asin (A + 0) +X = gsin asin (A +0) sec? e+ X, pcos Bsin x7 sin b sin (BY + 0) + ¥ = q sin b sin (BY + v) sec? e+ Y, (vay pain Sarsinesin (C4 w+ = gin eain (C" + 2) sec? Lou 2. If the radius voetor is unnecessary, the right-hand terms are used, and ysin a, gsin b, @sin e are calculated beforehand. The control of & and 6 is hy the difforences. E. Nearly parabolic orbits. At first the calculation is con- ducted with logarithms, ‘The ailditional auxiliary quantities are (a) Tre According to the tables of Oppolzer of to Table Hl, fand D are taken for the argument ©. (a) For every position Maat (lay From Barker's table » is takon, designated here by 1: tan gu, n= Ast «attay Table Il will give 2 directly. In Tables IV and V (or in the tables of Oppolzer) G is found (with argument n) as well as H (with argu ments mand €): ontan do aGH. (vay Continuing the calculation with an arithmometer, we have: av) resin ve eae) Seat pcos 8 cos x» Pyrcos + Qursin y+ Xy pcos Bain = Pyros v + Qyrsin + Y, sin 8 =Pprcos v4 Qersin vs Zs Geocentric Motion n In calculating with logarithms (vay pcos 6 cos X= rsinasin(A’ +) +X, pcos 5sin a=7 sin bin (BY + 2) + Y, (Va) psin 8 =rsinesin (0 +0) +Z. Control of ct and 8 is by differences. Here are some observations that may be useful. For the calcu- lation of separate positions for the purpose of comparison with ob- servations, it is necessary to find the coordinates of the sun for the moments of observations. In the interpolation of X,Y, Z with six decimals, it is possible to apply any interpolation formula. In those rare cases when it is necessary to have seven decimals, it is better to make use of Bessel's formula, because then the third differences will not influence the result, For control of the coordi- nates of the sun (an operation demandifg great attention, because the errors in interpolation are not infrequent) one may also interpo- late the radius-vector and make use of the formula Xt 4 84 22 RY One can also apply the mothod of interpolation of Willis which is presented by M. F. Subbotin [14]. However, it does not save much time, if, for ordinary interpolation, tables of prepared correc~ tions for second differences aro used. During comparison of an ephemeris with observations it is nec- essary to consider the influence of aberration and parallax. This will be discussed in the following chapter, as will he consii tions on the scope of search ephomerides for comets and minor planets. A precise ephemeris, having as its object the comparison of an orbit with observations, must certainly include the entire period of observations, with the addition of one or two spare in- tervals at the beginning and at the end of the ephemeris in order that the subsequent interpolation of the ephemeris data at the mo- ments of the first and last observations may not be found to be unreliable, If che observations were made with large interruptions, thon the corresponding omissions can also he in the ephemer Search ephemerides are not always calculated with fully reli- able elements, and often they fail to lead at once to the discovery fof the object (not to mention the search that fails entirely). ‘The teatest questionable factor is the position of the celestial hody in the orbit, because an error in the determination of j influences the position proportionally to time. In order to facilitate the problom we The Determination of Orbits for the observer and to make a more certain identification of any suspicious object, the so-called variation is calculated, ‘The mean anomaly for minor planets or the time of perihelion passage for comets is varied. If the coordinates of the body Hints Yanty 24-1 Fofoering to the moment q-) are combined with the coordinates of the sun Ya, Ya, Ze for the date tx, this will be equivalent to a change of T by +, where w is the ephemeris interval; moreover, it is necessary to complete the calculation, ‘until « and 8 are obtained, Designating by cx and Bx the coordi- nates of the celestial body for the moment ¢y and its changed co- ordinates by a, and 8,, we shall obtain (for comets) vara = 2528, var eo (3-66) the varintion of «t and 5 corresponding to a change of T by +1 day. Those variations are calculated over equal intervals for tho entire ophomeri For minor planets the variation is ealeulated differently. It is designated vs and the formula for it is ore Orne Gray ‘The meaning of this variation is that it gives the change of 6 corresponding to change of «in +1". Te is calculated for the be- inning and for the end of the ephemeris, or even only for its middle, Seatch ephemerides are often supplied with indications of the expected brightness of the object. For minor planets reflecting sunlight the brightness J changes in inverse proportion to the square of the distances from the sun and from the eatth, ie ae (3-88) 7p @-67) ‘Therefore, for r= 1 and p=1 tho brightness of the planet Designating the corresponding stollar magnitute by 9, ing to Pogson's law la 2 0 (wo ey ee 3-70) m= +5 (log r+ lox p). ae Geocentric Motion n For the determination of g we enlist. the obs brightness of minor planets (usually these are rough estimations). Every observation gives m, while g is found by equation (3-70). From all of the determinations the mean is taken. If g is known, ‘equation (3-70) is used for the peecalculation of m. For the general characterization of the brightness of the planet, mo is introduced, its magnitude in the so-called moan opposition, for which r= a, p=a-l. This gives mg = 9 +5 (log a + log (a ~ 1). any Tho change in the brightness of eomets doos not obey a genoral law. By no means always does the maximum brightness, reduced for a change of distance to the earth, coincide with perihelion. It is often assumed that the brightness of a comet changes approxi- mately according, to the Inw: m=genlogrs5 log p, @ 73) whore n jointly with g aro derived from Sbsorvations separately for fovery comet. Tho moan valuo of n according to Bobrovnikoy is equal to 8.8, while according to S. K. Vsekhsviatski, it is about 4.0, However, the separate valuos aro vory diverse and among them cean bo found nogative values indicating that the brightness of the comet decreased as it approached perihelion. In addition, other changes of brightness in comets are also obsorved. For poriodic comets, 9 noticeably increases (brightness decreases) in the course of a series of revolutions, and many comets, in the course of time, have become totally invisible to us. To prodict tho brightnoss of a comet in a future appearance is an unreliable matter, and itis bottar to restrict ourselves to the indication of rand p. Example, Tho calculation of an ophemeria of « minor planet with the aid of Steacko's tables, Planet 1935 NA. Blomonta 1933 U.P, VII, 27.0 VILL, 4.0 VIM, 12.0 VIII, 20.0 VII, 28.0 fo= 1998 July 27.0 U.T. | -M 19°92 189312 17°30%3 20° 1528 22°37: Mom 13°02 CHSOTL 47930 +7768 47581 +7075 pa859%4 S 42018 45108 $3588 $3908 4 4427 10687252 X—se01 “ees — 7625 ~sia7 ~ 9095 Aen 0.213 Y 47769 +7008 +6121 +5122 +4050 Ay~ = 2.0788] 24897) +3040 +2855 +2022 +1748 A= 0.7590] | peonBeowa +2487 +2599 42892 +2500 42T38 B,=+2.2006[ "pcan B sin t—8482 ~8798 ~9248 — OTT ~ 1.0397 By s0.21T8 pain 8—2107 —2922 2488 — 2686 — 2915 Blo s0a1g peosB 1 A791 + 911T +9552 4 1.0082 + 1.0005 80 The Determinction of Orbits aohs.7i9tom9 8tss"0 8 -1eR 1a 1830 0.006 a 2nB"26" 285715" 28430" 1933.0 O41 0.987 pple. Comet 1946 b (Pajdus ‘The lements are P= 1946 May 1141722 UT. 2° 18°38. 101 16.28.65 $1946.0 £= 169 33.82.81 49 1.018122 sin tan sin W 9.258200 Reos 8:T0S0340, i N= 160°277266 ® 0.011609 Ns e188 54 18 sin (+ €) 88830140 a 9.733802 cos (W +E) 8.99809 in sin @ ain 4 9.715985 sin a cos A 9921587 Am 1as°11'3774 ain a 9,091727 9.007709 2° 0.008900 2 = [0.002826] sin (170°38"14772 + ») smc? 156" nein’ ask eage ster 1.044 kové-Rotiart), sin be sin 2 9.8049828 sin 8 cow B 9.732700 = 288°25" 169 sin 6 9.978709 sine-sin € 8,831570n sin © com © 45668180 = 9897453675 sine 9.534110 sin b sin © 9.512009 sin (8 ~ 0) 8.677108n sin a cow A 9.924457 li ie 9.265460 ta fy 9.265460 sin? ‘a = 0,976010 Sin? b= 0.908950 sin? roo 909909 be y= [9.086598] sin (287°40°8877 + 9) soo? bv 2 = (8.541909) sin (280°41" f= 1946 June 18.0 PE ag.ss278 >> 1.88860 Ww asaT16 » assy soo? by 061414 + ayacet te a= - 0.638490 Fe 4 0.118613 y= ~ 0.964642 Fo + 0.995532 2 =~ 0.208674 2» + 0.401400 eae) Siena pom Boose” “wien dana eo! po ovatante poem Sana ssviaien weet) Ssbetn punk Stans gain b sec? Ly 0.048012 peosS 9.721954 : se oleoses fe: he pe MTA Geocentric Motion 81 All the other positions of the ephemeris are calculated in the same way. 25. Obtaining on Ephemeris by the Method of Numerical Integrotion The method of numerical intogration can be applied not only to the solution of the problem of perturbed motion, as mentioned above, but also to the construction of an unperturbed ephemeris, whero, however, its value is not so great. Its use for the latter purpose belongs to the current contury (although certain particular ways wore also shown earlier), in spite of the fact that the fundamental iden of the method of numerical integration has long been known, and its theory was worked out by Gauss with sufficient completo: ness. The formulas of Gauss ate the most precise and are now most ftequently used, but other methods of numerical intogeation also exist, and among them ought to ,be mentioned the original method of Cowell. ‘The foundations of the method of numerical integration can be found in the works of a number of authors, in particular, in that of M. F. Subbotin [14]. Here only the final formulas will be cited. ‘The solution of the differontial equations of the problem of two hodies is our task. Lot us assume in them m= 0. This will give, in place of (2-3), ae ae ay ae (@-73) Before starting on the solution of this system of differential equations, we hall give a summary of tho necessary dosignatio and formulas. Let /(¢) be a continuous function of timo and Let its value be known for a series of equidistant moments: ..., a ~ 20, a-w, a, a+w,a+2u, We compile the table shown on the following page. This table ean be continued as far as necessary upwards, downwards, and to the right. Each of the quantities enter- ing into the table is equal to the difference between the two quanti- ties standing, in the nearest left-hand column, directly bolow and 82 The Determination of Orbits 3 zl le] jal |= 3 on] faa} [om] fon ae 2! jel ls} |s 2} }a] jt a) ls 8 a| [a| jel |= ge eo} foe} fom} POR) = =] /e/ IS) IS | j2| |e 3| |2 z a| \e a 5 cis} [aml [am] fom] | ' ato fe +]: = =} S| [SIS ee 3| |? B | |=} (3 =| |5 £ a 8 i a el cli} yet fe) fs] fs Bll s 3 Geocentric Motion 8 above this quantity. Thus, Y(a- gw) = "Y(a) = "Yla- wv), 1a) = a+ bu)~ "fla ~ be), and so forth. All columns containing the differences of different orders can be filled immediately, according to the given values of the function f(t). However, for finding the values of the sums, the valuos of one of the first sums and of one of the second suis must be known, The table can be completed by inserting in the empty places of the table the half sums of the quantities standing in that same column, in the immediate neighborhood of the place being filled; thus, fas zw) = Zlf(a)+ fas wl, Plas) EYMa+ du) far gudl, and so forth. S Let it be required to solve the differcntial equation dz (3-78) =105 a where the values of the function 2 and its derivative 2’ are given corresponding to the moment of time ¢=a. We shall designate these values by (a) and 2’(a), respectively. Our problem is finding the values of the function 2(a) for a series of equidistant values of the argument. We shall begin with completing the table of (3-74). As was altoady stated, for filling of the first two columns of this tablo, it is necessary to know the value of ono of the first sums and of one of the Second sums. For the finding of these values [ram the given values 2(a) and 2’ (a) can servo the following formulas: Yer fe) ett yas 5 Po- + sey G76) M4(a) = 2(@)- Lays tay - AL pr (a) + ¥(a) = 2(a)— Ff) + see) ~ FaegG PO + - Mf our initial given values aro 2(a~ $12) and x’(a~ $10), then ‘he formulas (8-76) can bo replaced by the followins: a4 The Determination of Orbits @7) AT (9 781(a = w) + Pa ~ stg Ma =) PMO)] if V (a= w) + 2/1 (al eee + BR) +20 @) Alter forming the table of (8-74), we ate able to calculate a series of values of the required function #(¢), using the formulas: sto tna fle tos fet) sha + bo) + Lp(a + bao) = afee(e Ba] =e fire (i Be] 18) det [ee( + Eo] + Herrero se] = Bter [es (ee Re] +}. Lot us romark that when the values of "fare known for two con- socutive moments, itis not necessary to calculate tho fest sum ae- Cording to the formulas, because it is found as the difference of the values of Hf. ‘Tho solution of the differential equation of the fist ordor ue eto is found according to the Formulas stata) <1 [ees tm 2 tab) _ 1911 (a 4 hw) + (a+ kao) — FEES 1Y (a + kee) + « off Dofeefrfe fo (es A)e] tafe (ee Ho] C28 rfes(e=3)e]---} For finding 1 it is possible to make use of the first of equa- tions (3-76) or (3-77), whore in this case instead of e’(a) wo must write 2(a). For tho solution of the system of (3-78), the integration of all three equations is done simaltancously. The initial data may be taken as, (a), y(a), 2(a) and 2(a-w), yla-w); 2(@- 0); instead of the theee latter quantities it is possible to take 2’(a), y'(@)y 2’{a). If the coordinates for te moments selected by us ‘wore not determined earlier, they can be calculated by the formulas of (3-18): 2= Per case + Qer sin vy 3-50) ete, For example that will give, in the case of an elliptic orbit, z= aP,(cos £ ~e)+acos 6 Q, sin E, st) ete. For finding the values 2’, y’, and a” we shall write first on tho basis of (2-18) and (2-30) 1 +e cosy dv ip eS ee ard ' P irae ela whence it follows. L dr _esiny dv andi ieee ar Ce d(¢cos ») _ ke sin eos» a we ipoose ee tee costy , ve 4+ (1 @ 008 ») cos w 86 The Determination of Orbits Substitatiny these expre: ‘equations (3-80) by differontiation with respect. to time, we find: the equations obtained from Py sin v + Q. (cos » + eI, K(-P, sin» + Q, (cos v + e)], (8-82) ve Ls ve In the case of an elliptic orbit this will give with the aid of (2-39) Py sin v + Qz (cos v +e). ve agers sin E+ a cos 0, cos F), k ; ; : ¥~ Hal aP, sin +a c05 60, e088), PGB) ee ghee. sin E +a cos Q, cos F). It is not difficult to derive these equations directly from (3-61). Similarly, for a parabola, we make use of the substitations » = 2¢, oa tan 2, req suc? 2, and wo shall find after simple tanstonna- tions: x Wee Cty0 Qe) = EE Pos Oo, y «peg 0+ Be moe a, (3-84) apt e Qa)» . Ve P04) ‘The following relationship ean serve for verification of the calculations in the ease af an elliptic orbs Oc yy’ a ae’ « ker sin a ; 1 Baar ey sae BEEBE pe Yasin ®. (8-85) The calculation of the ephemeris can he done by two methods. It is possible in the first process to calculate the coordinates 2, ys 2 for two moments @ and @ + 1, making use of the usual for- mulas for tho caleulation of ophomeridos, let us say oquations (8-81). If the interval w of the ephemeris is taken for the unit of Geacentric Motion a7 time, then instead of & we can write mk, and correspondingly in- stead of &?, w?k?, Finding 2, y, 2, we calculate for those same moments yee, and also [on -wtker’, fy = why Jen wae (6-86) Because of our selection of the unit of timo, equations (3-78) will not contain the factor w2. There are tables giving r”* {10, 32) or directly 7k? 1"? with the argument 1%; for example, the table of M, F. Subbotin [18] for the interval w = 8 days. The calculations are often conducted in units of the seventh decimal; then for w= 24 — 107m? k* 1886.49, for w= 44 107 wk? = 47348.95,, for w= 84 107 w2k? = 189988.81.. Then "¥(a) and "Y(a +1) are determined in accordance with (8-76), where in the first approximation it is necessary to take F(a) and f(a + w) equal to z0r0. In order to ealeulate further, we must find a(a+ Bw) = Mla + Qe) + Epa 4 200) making use of (9-78). ‘The second tarm to tho right contains f(a + 21). ‘This quantity can be found by linearly extrapolating f(a) and f(@ + 12), that is, assuming Has 2) =2f(e +1) ~ f(a) = f(a + w) + Lf(a + w) - fla). With the dotormined values of x(a + 2r), y(a + 21), 2(a + 2), the equations (3-86) aco calculated, and the search for 2(@ + 21), ote., is repeated with the improved value of f(a + 2), and then the iniGal terms of the series of sums are improved hy correcting for the influence of the second differences. This process is repeated tntil the approximation hecomes sufficient. After this, in the samo way, the calculations are extended to the successive moments: Tho necded values of /(¢) are found by extrapolation; after the do- termination of several values of this quantity the extrapolation pro- cy The Determination of Orbits coods with sutficient certainty, at lonst under ordinary conditions. But one must romember to wateh whethor the exizapolated values of /(0, agroe in sufficient measure with the values obtained from the calculations for each interval, and to make certain that a possible deviation does not. call for a repetition of the calculation. The process of approximations may be shortoned if initial eo- ordinates aro calculated not for two hut for several moments. ‘This will permit the differences of higher orders to be found, and finally the initial terms of the sums to be determined. Also it will give conttol of the initial coordinates by the differences. However, hore the work noticeably increases. During the calculation every error in the determination of the value of /(@ appears in the first sums, and in the second sums it appears already multiplied by the number of tems of the series of first sums; consequently, its influence quickly increases according © the increment of time from the initial moment. The simplest con- siderations, based on the theory of errors, show that in the series of the first sums the error increases in proportion to n'/2, where n is the number of intervals of the ephemeris. But in the series of the second sums, the increase will be proportional to n3/2,‘There~ fore, it is desirable to conduct the calculation with two or three re= serve figures beyond the number of significant figures required to ensure precision of the final rosult. At loast ono roserve figure is, necessary in any case. See the work of Brouwer (22) concerning the accumulation of errors during integration Consequently, if a precise ephemeris is calculated (6 or T deci- mals are raquired), it is necessary to conduct the calculation with 8 or 9 decimals, If many-valued tables aro not within casy reach, Ghon it is difficult to insure this (incidentally, purely formal) pro cision of the initial coordinates. In such a case, it is more con- venient to turn to another version of the initial data in which, for one initial moment, one computes #(a), y(a), 2(a) and 2° (a), y” (a), 27(a), whoro tho values of the derivatives aro found with a larger number of figures than the values of the coordinates, for example, with 8 decimals. This is attainable even by the use of tables with @ smaller num- or of figures, because the velocities are expressed by small num bors. The computation begins as follows. Aftor the initial values of the coordinates and their derivatives have been obtained, f(a) is calculated by equations (3-86), and then t/(a + 41) by equations (8-76), neglecting the terms with f(a) and (a). Now it is possi ble to'find Mya +m) and MY(a =o). After this x(a tw), y(a £m Geocentric Motion cy 2(a +1) are calculated, taking f(a +) = f(a). In the second ap- proximation, first f(a £ w) is ealeulated and the improved values of (a+ bw) and "(2) aro found by equations (3-78). If the adopted intorval is not too great, we shall obtain, caleulating (a ©) and so forth in the second approximation, the valuos of f(?) of such pro- cision that a new approximation will Leave things unchanged, and it will be possible to begin the caleulation of further intervals, each time extrapolating f(d), a8 mentioned above. Continuous control of calculations is ensured by observation of the regular trond of the differences of higher orders. Any error in the determination of /(¢) will be indicated by jumps in the teend of differonces and thorefore it will be easy to detect. Errors can also originato in the additions during the formation of the first and sec- ond sums. It is necessary to watch for these errors by applying the usual verifications of additions. For final control the coordinates for the extreme moments of the ephemeris are calculated directly from the elements. This will also reveal the magnitude of the eetor in the determination of coordinates which accumulated because of tho unavoidable errors during the calculation of the values of f(t). For the mitigation of the influence of the indicated errors it is ad- visable to place the initial moment in the middle of the ephemoriss this will not complicate the work, because the calculation is con- ducted with the samo ease in oth directions with respect to time. The beginning of the intogration process is rather titing because of the necessity of finding the hase coordinates, and possibly also the velocities, and of carefully controlling their determined values. Aside from this, the process of approximations is unpleasant. How- ever, when the calculations have left behind the initial stages, they ate continued with exse. From this it is evident that the method of numerical intogration ean prove to be advantageous for the calcula: tion of long ephemerides, but on the condition that. the computor watches for the accumulation of exrors. Remembering that this ace cumulation proceeds proportionally to n*/, and having discovered that tho control coordinates gave certain residuals on the ends of the ephemeris, it 18 possible to improve the rosults somewhat hy distributing the residuals among the intervals. If the number of intervals from the initial up to the conteal moment ism, then for the 0 jntorval the correction will he a(t), where A is the discovered rosidual in the given coordinate, 0 The Determination of Orbits Lot us note, too, that during the usual calculation of an ephemeris the laborious part is the finding of the eecontric and true anomaly. Whore this calculation is most difficult, namely, in the case of a nearly parabolic orbit, the method of aumerieal integration can be especially recommended. Example. Comot 1942 f (Tovandze 2). It is roquired to find for the ephemeris the coordinates of the comet from February 25 to May 16, 1943, with a precision of 6 decimals. Elements 1943 February 6.71867 U, = 39°49" 1071 SL = 100°1°22"7 1948.0 42°50 1.853628 o.921t2 1. (soe Section 28) Lot us find the vector elements according ta the formulas af Section 24. sin ff + 0.984738 2 + 0.630897 Pe ~ 0.727818 om fl = 0.740: ay + O-T86364 Py + 0511056 sin ¢ 4 0.887325, By = oninis2 Pr + 0.457878 fom #4 0.941988 Ba ~ 0.183077 @, ~ 0.600581 Sia + 0.640870 yi 0.216018 5 = 0.19707: 08 «0 + 0.188058 Ya + 0.259088 2, = 0.089288 ‘sin © + 0.307506 oly + By con t+ 0.851424 ow & + 0.917431 na) + Ba cos t~ 0.756489 The values of ¢ and » are calculated according to the formulas fof the same section. The necessary auxiliary quantities are hor rowed from the example in Section 23. ‘Tho calculations are done with logarithms. 2 1943, 11, 25.0 V, 16.0 = 18.2813, os.gaiaa : iaea008 ‘ona401 Kange2e Ennass7 anipazas = m= 1B ea % 05°5258 70 tehw gstaseTt b.s19808 B oct615s 9.820490 2 Rlangatn a:610080, a 5.585950 Tasse20 nh « s0lo000T89 + 0.001716 @ ~ “o‘ooo018 9.000298) Geocentric Motion ” Hu o 0 ° 9.146168 ‘.s20788 O = + 00140018 407663803, rat, 2" Vorseog Tas6102 Fyco®= Logoort root Pes ss003s, ioaaaay ol apeye aes at orato The coordinates and volocitios ate calculated next, whore for control equation (3-85) is utilized; » = 44, The seventh decimals Of tho valuos of the cootdinaton =,'n, » aro taken as qual lo rer. ¢ 1943,11,25.0 V,16.0. pe g(t se) 2.696660 ine + 0414612 40030610 Vp 1et2i81 fone TOuIstT tose Weve Bona wh cose +0519 = 0.0115079 ve rsiny + 0.970022. + L.Ta8836 roos y+ 1.926001 + 0.780201 + (oon w +e) + 0,0818636 ye =~ LasaTe40 - 1.626549 + y+ 0.3768950 — LoaT30s 2, ~0.5820460 + 0.234069, w — Lo.oto7942 y= 0.0711895 wkereine 4 oorsts72 : Pane ~ 00104448 ax yy’ 00" + 0.018872 Ic is convenient to conduct the numorical integration of all throo coordinates simultaneously on one Large shoot (oF on three different sheets). The first approximation, and also the extrapolated values of /(0), ean be inserted in pencil. To save space we give the cal- culations for the first four momonts only. 1943 7? 10TH ee gee tl an a 121 1.87646 18419 =1.180809 —1.1510762 + 21198, “418665 +280 25 1.90456 18014 ~ 1.192764 —L.s920490 vaust ra Ssorts ant +16 MH 1 1.09940 7599 —1.on2481 — 1.999608 sareo8' | ase -aiss77 ag 5 1.98110 16980 — 1.270030 — 1.2702188 21568 y 1913 oy uf 4 fof ope 21 404eTer1 +0.4477396 ~ 9246 = T0750 +7 25 +0.878895 + 0.8700516 — 6789 -2 714869 +1438 28 2 1048 y poo pe M1 +0,805440 -415 +1390 5 +o.gsa440 1943 aoc My c t 1p Wet +0.502858 + 0.5920401 10920 99156 +419 25 FO.S8201 + 0.5890555 = 10501 109657 4475 ML 1 +0.5TI98¢ + 0.5720878 10026 = 11968 +517 5 +0.500020 +0.s600905, 9500 By such means an ephemeris of heliocentric coordinates was brought up to 16 May 1943, and for this dato = ~ 1.626548, y= ~ 1.087812, 2= + 0.234070, wore obtained. Comparing them with the control values found formerly, we find the residuals Aye =O, Aya ods Ay Sinco the ephemeris contains 20 intervals, we have n3/? = 89, and the accumulated residuals can be considered as reasonable in spite of the fact that the calculations wore made with seven decimals. Tho rosiduals can be distributed among all tho intermediate values of the coordinates and for this it is convenient to make use of the upper scales of a logarithmic slide rule. Then it will be possible to trust the corrected coordinates within one ot two units of the sixth decimal. For obtaining greater precision it would be neces- sary to conduct the calculations with eight decimals. 26, The Apparent Motion of Minor Planets and of Comets For the determination of the conditions of visibility and of the possibility of observations of minor plancts and of comets, their position rolative to the sun and tho earth plays a groat role, Minor planets move only with direct motion along their orbits, which, as a rule, do not deviate too much from circles. With very taro exceptions, thoy are always farther from the sun than the earth. Geocentric Motion 2 We distinguish for them two principal apparent positions relax tive to tho sun. The first of them is opposition in longitude, when the longitude of the planet diffors feom the longitude of the sun by 180°. Instead of this it is also possible to talk about opposition in right ascension, when the difference of the right ascension of the sun and of the planet amounts to 124, In tho vicinity of opposi- tion (theoretically this is not obligatory and, indeed, for cortain planets with small values of the semimajor axes and considerable ‘eccentricities, this is not fulfilled) the planet usually moves with fa retrograde motion; in tho romaining time the motion of the planet is direct. In this way, for a considerable extent of time the apparent path of the planet will form loops, the shape of which depends upon the oloments of the orbit and the position of the planet in the orbit. Usually the loops are closed, but it is not difficult to see that, if the planet during its motion slong the loop intersected the ecliptic, then the loop will appear to be open. Indeed, whon tho planot passes through a loop completely and returns to tho initial longi- tade on the apparent. celestial sphere, it will by this time be in the other hemisphere relative to the ecliptic, and its apparent trajec tory will not have a point of intersection. The planct occupies its second principal apparent position when its longitude is equal to the longitude of the sun. It is said that the planet is in conjunc~ tion with the sun. Except for those very rare cases when the planot penetrates inside the earth's orbit, minor planets can have only & superior conjunction, i.c., the sun is between the earth and the planet. In opposition a planet is closest to the earth, of, more exactly, the moment of the closest approach is not far from the moment of opposition. Evidently, when the planet is in opposition in right ascension with the sun, it culminates at midnight. This time is, therefore, the most favorable for its observation, especially if the opposition occurs whon tho planot is not far from the perihelion of its orbit. The average time, in the course of which minor planets are observed, usually does aot much exceed one or two months, and to find Faint planets after the first or discovery observation is often impossible. The search ephemerides of minor planets are usually calculated for a duration of 40 days in the vicinity of opposition, except for specially interesting planets, for which more extensive and more preciso ephomorides aco given. Tho intorval of an ephemeris, as a rule, is taken equal to 8 days, and moreover opposition must occur in the third or middle interval 04 The Determination of Orbits ‘The synodic period of the planet, or the period hetween its op- positions with the sun, is found according to the simple formula 60. 60, om H (en) whoo jig = & = 3548” is the mean daily motion of the earth. The synodic periods can be taken with a rough approximation from the little table, borrowed from Stracke [36], which is given below: # Po 2 Po Boor 12.7 montha 90077 Tési montha 800" is 100077 16.8 007 BY 11007 we 500 wo 8 1200/7 wa” 80077 ws > 300% 10" 700" 150 14007, i390 500" ws "| i500” 20.8 ‘These figures are only for the average intervals of time betweon oppositions and therefore in reality can give an error, which in- creases with the eccentricity of the orbit. For the determination of 8 more precise date of opposition it is possible to calculate the geocentric position of the planet for 20 days preceding the moment Cf opposition and for 30 days following this moment, and by Tineae invorse intorpolations to find the moment when the difference of the right ascensions of the sun and of the planet becomes 124, It is possible to treat this more simply and to find for these same Gatos tho true anomaly, and by it to determine the heliocentric longitudes of tho planet by the formulas: tan (7, ~ 2) = tan ws cos i, tan (ly ~) = tan ny cos é, which follow from equations (2-60). The moment of opposition in longitude will be that date for which the differonce of the longitude of the planet and of the geocentric longitude of the sun will be equal to 180°. Wo shall find tho exact date of opposition in right ascension by applying the mothod cited above to the right aseen- sions of a planet's ephemeris. The orbits of comets in the vast majority of eases are consider ably more eccentric than the orbits of minor planets, and many comets, including almost all bright, comets, have the peribelions of their orbits within the orbit of the earth. For comets, both superior Geacentric Motion 95 and inferior conjunctions, as well as oppositions with the sun, can fexist, but the latter do not play so important a role as for minor planets. The most advantageous time for the observation of comets is not the moment of opposition but the epoch of greatest bright- nos coinciding with the greatest proximity of the comet to the sun find to the earth. However, here the angular distance of the comet, from the sun must not be too small, otherwise the comet will be concealed in the solar rays. Depending on the relative position of the comet and the sun in the sky, certain comets can, for the most, part and even exclusively, be observed only in observatories of the horthern hemisphere of, the other way around, of the southern. Cor- tain comets can be successfully observed for a short time only, but in the majority of eases they are observed longer than minor planets, for several months, and oceasionally over a year. As an exception from the general rule, we shall mention the periodic Comet 1925 11 (Gchwassmann-Wachmann 1), the orbit of which is almost circular ‘and which is observed overy yoar as are the planets. For any body of the solar system Lambert's theorem concerning the curvature of the apparent path is observed. The path of the body is convex toward the sun ifr > and the celestial body is farther from the sun than the eacth; if, however, r < R, the path of the body is concave toward the sun. ‘The proof of this theorom will be given in Section 39. Chapter 4 THE REDUCTION OF THE OBSERVED POSITIONS OF MINOR PLANETS AND COMETS * COMPARISON WITH EPHEMERIS ¢ NORMAL PLACES 27. Designation of Minor Planets and Comets and Observations of Their Positions ‘The initial data on which every investigation of the motion of minor planets and of comets is based are the determinations from observations of their positions on the celestial sphere, Data of a similar kind ace communicated in astronomical journals and circu- lars and also in the publications of astronomical observatories Although the theory of these observations pertains to astrometry, their results aro published in such a form that they require some additional reductions before they can be used for the calculation of orbits. Every published observation, or series of observations, con- tains the conventional designation of the object to which it per- tains, with the exception of tolepraphic communications, in which only the character of the object is indicated—a comet or a minor planet. However, telegraphic communications are now made only about those newly discovered minor planets whose unusual charac ter of motion in the sky suggests an especially intorosting orbit, Up to the last decade of the nineteenth century, the discovery of minor planets was done visually and the number of discoveries per yenr was small, Bach planet, immediately after discovery, ro- ceived a number and a name. AU first these names were borrowed from ancient mytholoxy. lator thoir choice hecame arbitrary. This system of designations has been proserved, but it is applied to those planets only for which sufficiently reliable orbits have al- ready heen computed. Provisionally, however, a planet receives a designation consisting of the yoar of its discovery and of two let tors of the Latin alphabet. Up to 1925 the letters were given in the following order: A, By...,%, AA, AB,..., 22, alter which the series was repeated, but in 1925 this system was replaced by nother in which the designation contains an indication of the time %6 The Positions of Minor Planets and Comets ” of discovery of the planet. ‘Tho first letter gives the moment of discovery ta within a half-month (A—from the 1st up to the 15th of January, B—from the L6th up to the Bist of January, ete.). The second letter denotes the order of registration in a given half- month interval (AA—first, AZ—-25th, AA;—26¢h planet, ete.). Newly discovered comets are designated by the year of dis- covery and a Latin letter in the order of discovery, and also by the name of the discoverer (with the exception of periodic comets, which in new appearances keep their own provious designation). If the comet is discovered almost simultaneously by two or more observers, as offen happens, the discoverer whose report reaches the Central Astronomical Bureau first receives priority. However, as a rule, the names of the other observers who discovered the comet independently are added, The order of the discovery of comets does not necessarily coincide with the order of their peri~ helion passage. After the lapse of several years every comet re~ ceives instead of the preliminary designation a final one, in which the letter is replaced by a Roman nufteral indicating the order of perihelion passage in a given your. Positional observations of planets and comets are made at the prosent time either visually by means of the direct determination fof the difference in cight ascension and in declination of the ob- served object and of a star with known coordinates, o by photo- graphic means. The position of the object relative to two, three, ‘and more comparison stars is determined by measurements made on the plate, For each observation the moment is indicated at which it was made and to which the observed cooedinates % and 6 of the celestial body belong. ‘Tho precision with which a and 5 of minor planets and comets is determined is difforont depending on the sizo and quality of the instrument and the difficulty of the observation of the object. The errors in the positions of the comparison stars are added to this, with the exception of those rather eare cases when the posi- tion’ of the object are observed with the meridian circle. For minor planets, the errora ace, ax a rule, of the order of from 0"%6 up to 2”. For comets the errors might reach 5-10’ in the case of furzy images of objects, or even more, if the instrument doos not excel by its optical power. In conformity with the rule of a spare decimal, % and 6 are usually published with an accuracy of 0*.01 and 07.1, If, however, the observations do not possess sufficient precision, the coordinates are given with an accuracy of 0.1 and 1’, 98 The Determination of Orbits In viow of the groat number of minor planets not all plates are moasured with complete precision. Usually approximate coordi- nates of minor planets are published with an accuracy of 0.1 and 1’, Tho samo applies sometimes to the first observations of comets, if the discoverce had no means for the precise deterni- nation of the coordinates of the comet. It is sufficient to give the moment of observation with an ac curacy of 0.00001 (or up to 1°}, because that provides a precision of 07.1 even for rapidly moving comets. For minor planets it is often sufficient to give the moment of observation with a £0 times smallor precision. In each opposition a few observations of « minor planet aro rade at considerable intervals of time, unless it is a particularly interesting object. Comets are observed much more assiduously than minor planets, and observations are accumulated in great num- hor, particularly soon after the discovery. For some comets hun- dreds of observations have been obtained in the course of one ap- pearance (sometimes more than a thousand, although this docs not happen any more in our time). Usually each comet is observed not less than several dozen times, except whon for some reasons only a fow observations can be made, sometimes not even enough for the calculation of an orbit, In conclusion we shall notice that the published results of ob- servations are not always free from gross errors. This is especially teue with rogaed to the first observations of new objects, when the observers have no timo to chock properly theit reduction ealeula- tions. Thetefore, blind trust in observations is not justified, 28. The Moments of Observations At the prosent time the moments of observations aco indicated in Universal Time, which we briefly denote by <> and which is, as a matter of fact, Groonwich Civil Time. Up to 1924, inclu: sively, Mean Astronomical Time was used which, as it is known, is reckoned from noon. Provious usage was to indicate time ac- coring to the local meridian; now, in view of its unsuitability, it has been abandoned. The difference between Universal Time and Greenwich Mean Astronomical Time amounts to 12*. Thus, 12* Greenwich Mean Astronomical Time, 31 Decomber 1924, coincides with 0% Universal Timo, 1 January 1025. ‘The coordinates of the sun and of the planets in the astronomical annuals are referred to Universal Time. The Positions of Minor Planets ond Comets 9 However, even now the calculation of parallax may require local sidoroal time with a precision not exceeding 0".1, ‘The teansfor- mation of Universal Time into local sidereal time is accomplished particularly easily with the help of special tables, Necessary for this purpose is che longitude of the observatory from Greenwich, which is sometimes indicated by the observers themselves in the publication of the observations. This information is also contained both in astronomical annuals and in tables on theoretical astronomy. Sometimes in the published observations (in addition to Universal Time) the moments of observations in local sideroal time are given. In such a case, one should not overlook the possibility of checking the transformation of time, because this can reveal errors in the primary processing. Generally speaking, it is difficult to verify the moments of the observations but if there are many observations, their comparison may occasionally reveal that a given observation deviates from the others hy a shift of the object in the sky corresponding, let us as. sume, to the lapse of 1 hour (such an effor is encountered in mis- roading clionometers, ete.). permissible to introduce a cor- responding, correction, and the observation keeps its usefulness, 29. Observed Coordinates and Their Corrections for Precession, Netotion, and Stellar Aberration, and also for the Improve~ ment of Positions of the Comparison Stars During the last decades, a rule was gradually established and approved subsequently by the Intornational Astronomical Union, to refer the observed coordinates of minor planets and of comets only to the mean equinox at the beginning of that year in which these ‘observations were made, or else to the normal equinox of 1950.0 (up to 1937 inclusively, to 1925.0). The coordinates of the sun— L, B, and X, Y, Z—given in astronomical annuals are referred to tho same coordinate systems. ‘The observers determining the position of the celestial body by visual means usually give, besides a and 8 of the celestial hndy itself, the differences Aa and AS between the coordinates of the object and the comparison stars, and also a, and 5, of the com- parison stars, referred to the same equinox as the coordinates of the observed celestial body. It is evident that Hao e Aa, 8H 58 +A‘: (1) however, this control yields little. 100 The Determination of Orbits It it should be necessary to use observations made long ago fone must take inte account that in earlior times the apparent posi- lions were given, and for finding the mean positions it is neces- sary to add the reduction from the apparent position to the mean = I+ ho sin @ + ay tan d+ 4 Asin (P+ x) see 3], Bed = Bane = ~ |g cos (G+ x) + Acos (H+ 4) sin 8 + icos 8], where f, 9, @ contain the effects of precession and nutation, and A, H, {contain the aberrations of tho stars. ‘These quantities are functions of time and are tabulated in astronomical annuals for each day. In the ordinary teentment of the positions of minor planets and ‘of comets, the observed differences in right ascension and in declination must be corrected for rofeuction (the observer himself always makes that), and since these differences are obtained with fan instrument for the apparent coordinates, they must also be cor- rected for the differential precession, nutation, and aberration, before they can be added to the mean coordinates of the compari son stars. In case of a reduction to the beginning of the year of the observations (but not to a normal equinox!) the indicated re- duetions in the majority of cases are insignificant, for example, if the differences in coordinates do not exceed 2" for & and 10° for 5, and if the object is not too close to the colestial pole. In gen- eral, however, the necessary reduction quantities will he found in some asteonomical annuals. Photographic observations, which are always referred to the mean equinox by the very method of their treatment, do not give, ‘of course, the differences in coordinates. On the other hand, factors of dependence of the coordinates of the object on the co- ordinates of the comparison stars (the so-called dependences) are fraquently published. Sometimes even the positions of the com- parison stars are given, but more often merely the sources whence they were taken are indicated. Tn the transfer of the moan positions from one equinox to an- other, the precession is considered in the formulas: perc scarn Sena as) Be Bosncos ti (t~ to), The Positions of Minor Planets and Comets 101 where, as in Section 20, the quantities without indices cofer to the epoch’ t, those with index 0 refer t epoch to, and those with index 1 refer to the epoch ¢, = © to i J and for this last epoch the precession quantities m and n are taken ftom the tables, In those vory rare cases when the transfer must he made for a very large dif- ference of epochs, the above formulas considering terms of second ‘order in the expansion of the precossion in powers of time will be insufficient and it will be necessary to apply more precise expres- sions. If, on the contrary, the difference of epachs is very small, as when the observations of the colestial hody were made in the course of two consecutive years and it is merely necessary to refer the observations of one year to the equinox of the other, then i stead of (4-8), it is possible to write at once = 9 + (9.078 + 1*.936 sin a tan 8) (¢~ Co), B50 +207.08 cos a(t ~ to). | oo Prepared values of corrsctions for provession greatly simplit the calculations, if it is possible to make use of them. veimeuy For reduction to a normal oquinox, the formulas given for the first time by Ristenpart are more convenient: doa 4 Ay Ay tan 8 Aa tan? 8, Bo =5+D+D, tan B. @) Auxiliary quantities 4, 4’, ote., occur in astronomical annuals and in tables (32, 94]. Sometimes it may become necessary to caleulate the precession for ecliptic coordinates. This purpose is served by the following formulas (the meaning of the indices wes given previously): A=Ao + [p+ mtan By cos (I ae] B =o +n sin (lh ~d1) (~t0)- Gy The determination of quantities p, #, II is given in Section 20. These same formulas, of course, are applicable also to the coordi- nates of the sun, but because the latitude of the sun B is small they acquiro simplified form: Lelo+p(t-ton } B= Bo +n sin (= Li) (¢~ to). on wz The Determination of Orbits Latitudes of the sun, referred to the equinox at the beginning of the year, are so small that it is often possible to neglect them completely. In the system of a normal equinox, however, they Amount to Several seconds, i.o., highly perceptible quantities. Aside from the superficial control mentioned above, one rarely succecds in subjecting the results of the observations toa more profound verification. It is possible, of course, to control the positions, if they are given, of the comparison stars, by caleu- Inting them anew according to the data in the catalog. Since in the final determination of orbits (if there is roason to expect a significant precision from the results) it is desirable to do everything possible for decreasing the errors in the observed coordinates, sometimes an improvement of the positions of the comparison stars will be undertaken, For the deduction of their tight ascension and declination, several star catalogs are con- sulted (oven all the available star catalogs, at least the latest ones). The star positions given in various catalogs should be reduced to one and the same fundamental system, because for a lacge number of observations the systematic errors—rather than the accidental errors—of the coordinates of the comparison stars hocome important. Unfortunately, this is not the place for a de- tailed discussion of this important. question of astrometry; a suffi- ciontly complete presentation of tho matter in modern textbooks does not exist, apparently, It is, therefore, necessary to study it in special works, and s0 we shall make only brief remarks. There exist indexes of star positions (24, 38] which help one to become oriented in the numerous star catalogs. Nevertheless, the consulting of the catalogs and the exteaction from them of the star positions remains a time-consuming task. The copied posi- tions of the stars are reduced by moans of the corrections for pro- ‘cession to the equinox at the beginning of the year of the observa~ tions of to & selected normal equinox. Here the special tables of precession [34] are very useful. As a fundamental system, it is possible (o recommend the system of the ‘General Catalogue” of Boss [20], becuuse it includes Uhe faint stars which usually seeve ‘as comparison stars in observations of planets and comets. The tables of systematic corrections of meridian catalogs are used for the reduction of catalog positions (for example, the tables ap- pended to the catalog of Boss, usually designated as GC). Weights assigned to the catalogs are taken from these tables, and the most probable star positions are derived, taking into consideration these weights, by the method of least squares, Often it is necessary to use catalogs for which there are no rendy systematic corrections: The Positions of Minor Planets ond Comets 103 (photographie catalogs). ‘The computer must sometimes himself de- termine the systematic corrections for these catalogs; this requires skillful approach. With a considerable number of catalog posi- tions, it is possible to obtain stellar proper motions, and the de- rived positions can, therofore, he reduced not only to the chosen equinox but also to the epoch of observations of the object under Giscussion. For orientation in these questions one can use an article by B. I. Rak {11} which, it is true, is already somewhat. antiquated. If we apply some corrections da and d5 to the coordinates of the comparison stars, then in the case of visual observations the coordinates of the observed object receive, of course, those very. same corrections d& and dé. In the case of photographic observa- tions each position of the object depends on several comparison stars, Let them all receive corrections 4, and dé,, then the cor- rection to the position of the object will be ¥ D, da,'and © Di ai, where D; are factors of dependence, the sum being taken over all the comparison stars. . It is useful to note that 3D, = 1, 430, The Coleulotion of the Influence of Planetary Aberrotion ‘The motion of the observer together with the earth and the mo- tion of the observed celestial body bring about a twofold influence of the aberration. As a consequence of the first motion, all celes- tial hodies are displaced on the geocentric celestial sphere hy certain angle, moasured from the geometeic direction to them to- ward the apex of the motion of the earth (stellar aberration, or an= nual aberration). As a consoquence of the second motion, we do na see the celestial body in the direction in which itis at the present moment, but in the direction in which it was when it emit ted the ray of light being observed by us (planetary aberration). Wo alroady presented the formulas permitting us to take into ne- count the influence of the annual abereation. The equations then selves do not reflect the small torm arising from the eccentricity of the earth's orbit. It is a function of the coordinates of a celes- tial body and for a given sta it is almost constant; therefore, it is customary to include it in the coordinates of the stars and there is no need io take it into account in any other way. Turning to planetary aberration, we shall notice that light trav- cls one astionomieal unit in 498*.72 or 0.005772. In the der tion of this quantity, which we denote by 4, we accept the follow- ing constants: the equatorial semi-axis of the earth a = 6878.388 km, the parallax of the sun po = 8.80, the volocity of light in @ vacuum 04 The Determination of Orbits = 299776 km/sec. Bul, if we accept the determination of the solar parallax by Spencer Jones according to the observations of Eros in 1981, namely, po= 87.79, then A = 499°.29 = 0.005779 is obtained. It is possible also to obtain A through the aberration constant and the elements of the orbit of the earth, but such a calculation would hardly give a more reliable result. If we denote by ¢” the moment when the light ray observed by us loft the celestial body, then t= ¢- Ap =¢- 04005772 p, (4-8) where £is the moment of observation, and Ap is the travel time of the Light ray from the celestial body to the observer. For minor planots and for comets this time rarely exceeds half an hour; however, not only for such an interval, but even for the duration of several hours we ean consider their motion as being rectilinear and uniform and to as- sume tho same for the motion of the Figs 10 arth. Let the celestial body move in a straight line PoP and be at Po at. the moment, ¢°, of the emission fof the ray of light (Fig. 10). At that moment the objective lens of the telescope of the observer is at the point T> and moves recti- linoarly together with the earth in the direction ToT. ‘The ray of light falls on the objective lens at. the moment ¢1, when the objec- tivo Ions is situated at the point T;, but the eyepiece and the eye of the observer are at the point O;. Finally, when the ray of light arrives at the eyepiece at the moment f, the eyepiece is at 0, and the objective lens is at T. The angle 0:70, as it is known from spherical astronomy, is called stellar aberration ant OT, is the true direction to the celestial body. 0; T; is the apparent direction. Pots ‘The ratio 7" of tho sexments of the ray of light, evidently, is Kio Fog OF tho Sout f the ray of lig! nee ‘equal to the ratio of the corresponding intervals of time, th As a consequence of the accepted uniform moti n of the earth, the TM ratio GE of the paths travelled by the earth is equal to the ratio of thoso intorvals of times hence, The Positions of Minor Planets and Comets 105 Pots ToT TO TT and triangles Py ToT and 7) OT ace similar, having equal angles contained between proportional sides. Consequently, the straight linos OT and Ty Pa are parallel. ‘The first of thom gives the apparent direction to the celestial body at the moment ¢, and the second is the true direction ta the celestial body at the moment ¢°. For a clear understanding of the resulting rules for taking into consideration planetary aberration it is necessary to keep in mind that the coordinates of the comparison stars, to which the observa tions of moving objects in the solar system are referred, are now taken for the mean equinox (see Section 29). Consequently, they alrealy are corrected for stellar aborration. We ean say that sork= ing in a similar manner we obtain the true direction “to a planet as to a fixed star.” Such coordinates of a planet or a comet are sometines eallod asteographic and the nulmber of the year indicat ing the equinox is added to them. After this we establish the first rule: if we utilize the mean (astrographic) position of a celestial hody, i.0., already corrected for stellar aberration, we obtain the true direction to the celestial body, considering that the earth is in the position referring to the moment t, while the position of the celestial body refers to the moment ¢, The second rule: if to the mean (astrogeaphic) position of the celestial body we add the stellar aberration, we shall obtain the apparent direction to the celestial hody. This direction is also the true geometric direction for the moment ¢° to which refer the posi- tions both of the earth and of the celestial body. In ordor to find ¢°, it is necessary to know p. This quantity is given in the ephomorides, and in the comparison of the observation with the ephemeris it is more convenient to apply the second rule. In this caso, it is necessary to add, with its sign, the stellar aber- ration to the moan (astrographic) position of the minor planet ot the comet, and then to take from the ephemeris tho coordinates for the moment ¢ = t= Ap. In the case of any significant number of observations an equiv- alent procedure is more convenient: not to introduce the stellar abettation into the observed coordinates but compensating for this {0 intcoduce it with inverse sign in all the ephemeris positions. For this purpose it is sufficient to compute the aberration for every 106 The Determination of Orbits 8 slays (at shorter intervals for rapidly moving objects, at 10-day intorvals for slowly moving objects). ‘The results of the caloula- tions can he controlled by differences. In the calculation of an orbit, p is not known at tho beginning. Therefore, hore it is necessary to make use of the first rule: the coordinates of the earth are calculated for the moment ¢ and the imean (astrographic) place of the celestial body is taken as it is published by the observer. When in the process of the calculations, the distance » becomes known, the moments of the observations are corrected for the quantity dp and later on also the coordinates of the celestial body will be considered as referring to the moment £° ~~ dp, but the coordinates of the earth (refereing to the mo- rent £) romhin therealter unchanged. 31. The Calculation of Porallax In order that equations of transformation of heliocentric coordi- nates into geocentric (8-11) hold, the obvious requirement must be Satisfied that the observed positions of the celestial body and the coordinates of the sun are raferred to one and the same origin of coordinates. Howover, observations actually aro made'not from the center of the earth to which the coordinates of the sun given in astronomical annuals are referred, but from its surface. In order to remove this hindrance, it is possible to act in two ways: either to ‘obtain the geocentric coordinates of the celestial body, having corrected its observed position on the celestial sphere for the in: fluence of parallax, of to leave its position unchanged but to take the topocentrie coordinates, i.e., the coordinates of tho sun referred to the position of observation. It is clear that the first method will have to be used for a com: parison of the observations with an ephemeris giving the geocentric coordinates of the object. In spherical astronomy the following, formulas of corrections for parallax in a and 5 are derived: Evens . Pa = 75 TPP cos 4 sin (4 a) see 3, (4-9) PaP2 tyin a” cos 8 ~ co oi ° whore py is the noocentrie railius vector, 6’ is the geocentric lati- Lurie of the place of observation, po is the constant of solar paral- lax, # is the sidorenl time, and % and d are tho coordinates of the colostini hovly. Hore 7, i8 obtained in soconds of time, and ps in Seconds of are. Usually p is known beforehand, and the course of : sin 6 cos (9 ~ «)), The Positions of Minor Planets and Comets 107 the computation is somewhat changed. Thus, instead of computing tho influence of parallax one seoks the so-called parallactic fae~ tors by using the following formulas suitable for logarithmic lations: tan y = tang” (5 =), y < 180°, 1 e Popo cos sin (#~ a) sec B, py = 22F, b4-10) Pah = a5 ° ' : Pap = pono tind sin (y 8) cose y, re = 22 whieh, obviously, are equivalent to (4-9). The quantities tan 3’, +k poPo 608 4’, poPo sin ’ are constants for a given point on the terostrial surface, and they are given in the lists of observa- tories in astronomical annuals and in all collections of tables in theoretical astconomy. ‘Thore are tables even more convenient (and even more extensive} [30], where the same formulas ace utilized Somowhat, dfferetly. Its clo that pa and py are found when ¢ In general, if p is not known, it is impossible to reduce the ob- served coordinates of the celestial body to the center of the earth, However, it is possible to refer X, Y, Z of the sun to the position of observation. In the equatorial system of coordinates the posi- tion of the place of observation is determined hy the quantities Por #, and tho sidereal times, showing the angle between the plnne of the local meridian and the plan of the hour circle of the vernal equinox. The coordinates of the place of observation £=p9.008 6" 008% 7 = po C08 6" sin 8, C= py sing’, taken with the reverse sign will be the corrections to the coordi nates of the sun X, ¥, Z. We express these corrections in astro- romical units, chen AN ==po are 1” & =PoPo ME 3 COS "COB 8 © Agy COB 3, ar eee me eet eats Pee aa) AZ = =p ate 1 = =popo are 1” sin 3. ‘The quantities Agy and AZ are constants for a given observatory; thoy are given in the above-mentioned tables. In the calculations of orbits with logarithms usually the ocliptie system of coordinates is applied, and then, if it is possible, one 108 The Determination of Orbits should first correct the equatorial coordinates for parallax. Tt is impossible to do this for unknown distances to the celestial body, hut by a Gaussian method it is possible, nevertheless, to take parallax completely into account, by extending the line of sight to the celestial body to its intersection with the plane of the ecliptic. The point of intersection is called the fictitious place of the earth (Jocus fictus). Simultaneously the condition is ful- filled that the latitudes of the sun are strictly equal to zero, which is useful in the calculation of orbits. However, the intro- duction of lacus fietus is not done so simply, and it is better to void it, even to the axtent of completely neglecting the parallax. If the parallax in @ and 8 has been considered and the observed coordinates are referred to the center of the earth, it is not diffi- cult to reduce the Intitude of the earth to zero, by drawing the per- pendicular 7p from the conter of the earth T (Fig. 11) to the plane of the ecliptic. In the diagram the angle Ty ST is the latitude of the oarth, or the latitude of the sun (given in the astronomical an- nuals) with the inverse sign (-B), and it is easy to see that be- cause Afi and B aro very small angles, we have approximately BR cos 8 __ B cos f ° ° This correction is introduced, as alrendy mentioned, if the ob- sorvations ate corrected fur paallax, and consequently, p io knowns otherwise it is neglected. OF course, in the correction for latitude of the sun, & and L of the earth remain unchanged. ae (4-12) 1 Coordinotes Into 32. Transformation of Equator Ecliptic Coordinat Such transition is encountered in the calculation of an orbit with logarithms, where ordinarily instoad of the observed o and 6 The Positions of Minor Planets and Comets 109 of the colontat body, A nnd fare used. Sphotical astronomy gives the following formulas for this purpose: ve cos f cos A 108 3 eos a, cos 8 sin A= cos sinacose +sindsine, b (4-19) sin 8 =-cos 8 sin w sin © sin 8 eos ©. It is more convenient not to apply them directly in this form, mak- ing use of logarithms of addition and subtraction, but to introduce an auxiliary angle, Indeed, if wo write ain Win 8, } mcos M = cos § sin a, (ut) thon we get £08 8 €08 A= cos 6 05 a, cos B sin A = m cos (M~ €), (415) sin B = main (Pe). |A good control of the obtained a and 5 is achieved by the cal- culation of the following equations which are derived from (4-14) and (4-15): ein led gain e-), sin @~p)=s0e 46 +8) msin Seas (w- §)- sin (Qa) =2 cos 4 seo 8 msi (4-18) But, in addition, the possibility of a simple and often useful con- trol results from the fact that the obtained sin B and cos 8 must correspond to one angle. By careful attention to the signs of the trigonometric functions, it is often possible even to limit oneself to this verification, if the angle A is not small (and not close to 20°), because otherwise cos (Sin f) changes slowly and only ommarativaly in the caloulation of the angle can he 33. The Comparison of the Ephemeris With Observations In the comparison of the ephemeris of a minor planet or a comet with observations it is necessary to obtain the ephemeris positions for the moments of the observations and to take into account the influence of precession, nutation, aberration, and parallax. M0 The Determination of Orbits 1. If the observations give the apparent coordinates, which may be encountered only in the use of old observations, it is necessary to reduce them to the beginning of the year and to obtain the mean coordinates according to the formulas (4-2), omitting, however, tho aborration terms: Bee Bs If the ephemeris refers Lo another (normal) equinox, the ob- served cootdinates still must be reduced for precession to that equinox according to the formula (4-3) oF (4-4), of (4-5). 2. The observed « and 8 must be corrected for parallax. For this purpose p is interpolated from the ephemeris for the moments of the observations (three significant figuros are sufficient) and we compute 17) a eee meee 9 008 (0+ 2). (4-18) where pap and psp are usually given by the observers [hut it is reconmensle to verify them according to the formulas (4-10)). 3. If in this way, of by othor means, more precise positions of the comparison stars were obtained than those which were used by the observer, the observations are corrected for the difference be- tween the improved positions and the positions of the comparison stars accepted by the observer; see Section 23. 4. In cases when the observations have been reduced according to (4-17) they can be immediately compared with the ephemeris. If, however, as is now customary, they are referred to the mean equi- nox and consequontly were corrected for stellar aberration, then it is host to enter beforehand this aberration with the sign reversed jn tho ephomaris positions. 3. The quantity dp = 0%,0057T2 p is subteacted from the mo- ments of the observations and from the ephemeris are interpolated, for the corrected moments, the coordinates of the celestial hotly, Alenoted by 2. and B-. Depending on the interval of the ephemeris land the magnitude of the differences, the interpolation is performed with second, third, and even fourth differences. Any interpolation formulas will suffice but in using an arithmometer it is easiest to take Newton's formula (the designations are given in Section 25) The Positions of Minor Planets and Comets m 1a #0) = f(a) + aft ( + 3s) 4 AO=D peat wy 4 M=Da-d 2-5 which is especially convoniont if the fourth differences are in- sensitive and the thicd differences do not exceed 60 units of the last significant figure, because then in Bessel's formula it is pos sible to confine oneself to the first three terms. It is useful to write out all interpolation coofficients heforohand; moreover, fot the second and higher differences they can be taken from tables with tho argument of the fraction of the interval n= (0? = to)/, where ¢° is the moment of observation corrected for aberration time, to is tho precoding ephemeris monont, and » is the ephemeris intorval. 6. Denote the observed coordinates of the colestial hody by dy and 5o after all the necessary reductions and form the differences Ad = to-do, Ab =30- Be. It is desirable, and if the celestial body was observed near the celestial pole, it is necossary to express Aa as an arc of a great citele Aa C08 5 = (ito ~ 4) cos 3. (21) ‘The quantities Ac cos 5 and AS, takon in tho sense here indi- cated, are corrections to the ephemeris, although it. should be re- membered that they are burdened with unavoidable errors of ob- servations, If there are many observations, which happens only for comets, then from the agreement between the individual correc tions it is possible to judge the precision of the observations and further, by the method of least squares, to determine the correction of the ephemeris as a function of time by the formulas Aa cos 8 = a+ 6+ et?, AB wa’ +s Bt + ots Ce) ination of Orbits Ww ‘The Determination of 01 with plas and ine ephemeris date (with pl jaken inl Mpecively, forward and backward in time) minus Sie vce such an inzerval of tine for which the formar a ae eatifontapiroximation vo the ral rn of the, cor, as 20) Shay, hot nsw Tove curt posse eon. Moe er ing by the construction of graphs inwhick be Ne tea and Aa cos Bo” on tho ‘Detarmining the deviations of hore ¢ is taken initially from Roea the moan errors of one observation (separately In % an VE ve of observations, and k is the number of oe Ae amen (22), Here at the se ine oh seevations ‘errors are revealed. It isnot difficult, Sons Wh Bey neorvations made at different dbsotvs: with TAFE parents neparaly the precision of tho series of SReorvatio ech observatory. But in view of the possibility 2a a ete in the chservations of comets, the results Il not be sulticient ey he vorrction for the ephemeris for the initia ate expressed Dy! Nueces Bea, AD e leuations leading to che obtaining of the seeipct toa very thorough verification, ‘hel eollaberetion of wo computers 15 ite gocensary to ty (0 find an explant: ations excooding tho possible eandon sponse, tovweite to observers 08k where a is the num! sorvati AMT stages of tho 0 ephonorie eorections mnometetly spenking, Beales Tn paticulay on for tho ind ine t cl procesaing of their ohservations. J thom to chock the oes rpecrvations must not be taken into conn, Heats seu down he bounds of wnsile ores and wha ¢ ith respect to comets, it is necessary ence and Me esr obsorver sn order to jue, depend some enon vTon of the appearance of the comet —the prosance ee of a nucleus, the brightness, the fuzziness, e a ‘tude oF absence ol nom during the observation, the presence of the The Positions of Minor Planets ond Comets 13 in the sky and other conditions—which observations one should recognize as burdened with errors at the time they were made or during the processing. As some approximate criterion it is usually possible to take 1* in & and 10” in 5 as limits of admissible er- rors for comets, but sometimos it is necessary approciably to widen these limits. By adding the Aa and Ad to the ephomeris values wo got the so-called normal places. They replace the individual observations and will be significantly more precise. Thus, a large number of observations made of the comet during its visibility will be re- duced to a small number of normal places, each of which represents 8 combination of observations pertaining to a given segment of the ephemeris. Tho more accurate the elements on which the ephemeris calcu- Ietions are based, the longer will he the segments in each of which the formulas (4-22) will be true, the fewer will bo the normal places, and the easier it will be to compare them with theory. If the ephem= ris is in good agreement with the obsérvations, the coefficients 8, B and especially 0, o” are very small. In that case it is possi ble to forego their determination and directly to take the mean of the separate Aa and AB and add them to the ephemeris coordinates for the nearest. date, If the observations are being compared with an approximate ephemeris giving positions, Iet us assume, up to 0-1 and’, it is meaningless to introduce corrections for aberration and for ‘paral- lax and it is sufficient to ascertain that the observations and the ephemeris data refer to one and the seme equinox. After this the ephemeris positions are interpolated for the moments of the obser- vations, and it is usually sufficient to limit oneself to an approxi- mate estimate of the influence of the second differences; the in- terpolated positions are then compared with the observed coordi~ nates, and this yields the correction of the ephemeris, Aa and AS. This correction can be of use, in the future, for a more confident search for the object with the same ephemeris. Example. Tt was necessary to campare with the ephemoris the observation of the periolic Comet Brooks 1911 T, made on 28 Sep- tember 1910, at the Lick Observatory, by Aitken and Wilson. In View of the fact that, in this appearance of the comet, no further observations were obtained, the calculations were made with the greatost possible care in order to insure the reliability of « normal position based on this unique observation. Because doubts were advanced that this observation belonged to Comet Brooks, a complete duplicate of the original notes of the 4 The Determination of Orbits ahservers at the telescope was used for the redueton a the ale ulations. Ibis necessary to have recourse to a similar procedure aly inexceptiont eaves, bit its not always possble. ye the ferences btroen the coordinates of the comet ad yo eetarizon star wore again caleulted and corected for eens” toa "taing the mation ofthe conot in 8, hnown fom the ephen tri, tho mean of the differences Ady having boon obsored SePH Stn fom ay na reduced to ie mean omen of observation of Re or, Those apparontdiferences wete ih ome thot validity, me etering to the equinox of 1910.0, Because the Siti edution to the texoaing of the yan wow of he oa af the chrono, the moment ofthe observation c= ‘ he calculation of the We smallne the conection of the chton ; ere cal Sidcreal tine was sed fr comtiR ie actary and hen converted into mean Pertin tines uote those oaleuation Chron, 2th 45" 26" BE Mg et PT a ag HS lft TH MRR Rae AE AEE eka * =07.4 — s(Berlin; 5 Ae Egle og Se ‘Chron. athasm™2e® Aa OME gy aa ik cos 9.669 ArReomece® a PREP at) MS ag 8 ae erty yes eat ‘Thus, the following results were obtained: Mount Hamilton 36- inch rofeactor. Wilson and Aitken Aa | Ad omits] —o'as" [Comparison os parla rat es a Monn Berlin Time os pa? |r OTD, Sept. 2.76213, In the enlculation of the final orbit the results of the process iti of all the observations of the given object are usually reiuce to Such a form. In the column ‘Comparison’ the numbors of micro. metric comparisons (in & and in 6) of the comet and of the s| The Positions of Minor Plonets ond Comets us given. In column * is given the number of the comparison star, in 8 general list of stars for the given work. Comparison star No. 2, to which the comet was immediately linked, is faint (about 14th magnitude), and it is not contained in star catalogs. Therefore, it was immediately attached by the observers to a neighboring star which we shall designate No. 1. This micrometric attachment, having been processed like the ob- servations of the comet itself, gave the differences Aa and AB be- tween hoth stars, referred to 1910.0 (it is not worth while to give here the detailed calculations). For the determination of the posi- tion of star No. 1, a series of star catalogs were examined and its coordinates were found in Gould's zone catalog, catalog & of the Cordoba Observatory, and in the astrographic catalog of the same observatory. For abbreviation, these catalogs aro respectively designated GZ, Cord B, and GC Cord, in agreement with intor- national conventional designations of catalogs [24,27]. Systematic corrections for the reduction to the system of Boss were added to the positions taken out of the catalogs, ahd reductions for proces- sion from the equinoxes of the catalogs to 1910.0 were made using Schort's tables [34]. The data were insufficient for the derivation of the proper motion and therefore the final coordinates of star No. 1 were obtained as the weighted mean of the separate results, woighting the catalogs according to Boss. By the addition of the observed differences Aa and AB, the coordinates of star No. 2 were found. However, a direct determination of its position was also obtained by Jeffers from a phologeaphie plate, taken with the Crossley reflector at Lick Observatory. A simple mean of both ro- sults was taken. We quote briefly the course of these calculations (it is quite typical), omitting the transformation into equatorial co- ordinates of the rectangular coordinates given in the photographic catalog of the Cordoba zone. From the sufficiently precise posi- tion of the star for 1910.0, the precession (annual), its secular variation, and the third term of the precession were calculated: N disse Pre VES OM Bisioo Pre VS. ID ‘Thon, the calculations based on the catalog data were carried out, No.1 Catslos Baulnox « Gz rstigo¢ 1575.0 1 toke5mar# 3a Cort $3052 1900.0 2 194710707 CC Cord =28252403 1900.0 2 19%T™10.04 = 20251864 6 The Doterminotion of Orbits Syatomatic precession a ‘ Corrections Z 1 42M 9863 IQMATMETE IG 02 1873.7 12 40 a7 108 AT AIT 0.8 1896.0 ree 0107.05 471 Lo 1918.6 ToT a8 6 1 =28°7'40"4 ks'ts7.0 =9899'98.5 02 183.7 2 98 S57 8 +130 8 26 19 04 1896.0 2-28 387 6 4130 16 aT tL 1918.8 og 228.8 No. 2 “ 8 ‘ Microm. 19447™47#,09 ~28°7'17.1 1910.7 Photo. 47 94 49 2 1995.9 To 47 47.96 -28 750 Here, after the designation of the star in the catalog, is given the equinox of the catalog, the number of observations of the star, the a of the catalog, its systematic correction, and the reduction for precession to 1910.0; then the right ascension for 1910.0, the weight of the position depending for a given catalog on the number Of observations of the star, and, finally, the mean epoch of the ob- Servations as it is given in the catalog. Then are given similar calculations for the declination. By using the rosults obtained for Star No. 1, the coordinates of star No. 2 were found; they aro givs jn the first row; below them ate copied the coordinates according, to Jefter's determination (without any changes), The final posi- tions of the comparison atars aro given in the following form (if the proper motion of one or moro of the stars were determined, it would have been given here and its influence would have been taken into considerntion in deriving the coordinates of the comparison stars, Which naturally must be referred to the epoch of observation of our object): N Hya00 — Bis10.0 Source 1 Cop —25°.16950 1947474.19 =28°9'26".9 GZ, Cord Bh, CC Cord 2 Anonyma 1947 47.96 =28 750 2 The mean of the micro- rnetrie link with Now t Mount Hamilton (1910 Sept. 28) and of the phetographie position fom plate taken swith the Crossley reflector (4935 Now. 14) The Positions of Minor Planets and Comets uw The following positions of the comet are borrowed from the ophomeris: Monn Berlin ee vera 8 vera los 1910 Sept. 25.5 104455418 =2929°91"%5, 0.1966 “+48850 0 st9 26.5 4550-98 418s 8 OT FG wa $47.66 610.8 275 M6 AS DL F185 | BH BER A $49.81 4615-7 28.5 AT IBIS 414 38 TOI. Hho $51.85, +6206 29.5 1819.50 FL 8 AIRS dS 0.2016 “158.19 4605-4 30.8 49 19.69 W278 581 First, p is taken for the moment of the observation (uncorrected for aberration). We find the aberration time and the corrections of the coordinates of the comet for parallax, and then we interpolate a and 5 of the ephemeris according to Bessel’s formula (because the third differences are insignificant, it is possible to use any interpolation formula). ‘The observed coordinates are reduced to the apparent position which in the given case is immediately com- pared with the true ephemeris position. p 0.2081 py + 0*-16A, 04.0021 A TTS py 447 Interp. Cooft 10 1 1 ae Be, ato, Sept. 28.7522 0.25292 ~0.0044 19847"%6%.60 -28° So's 198474 9%.25 3, 280837 o-c Aa + 108.60 AB 4277.1 te varn 19447408.06 Be vera 28°9°3"".8 @ nonin” geod 00ST f+ 21°78 a 20655 sin(# +a) 9.5508 Ma + 8,08 a 8H 21 wo 13M Mat 7.73 asa 11 cos ea) 9.9694 Ag + 35.44 waa 16 “sind 9.6737 Ag+ 25.30 tend 9.7284n Mg, OLSRBD I~ 18D sin@ +a) 990001 Ms O.BITIn is 8.26 g 1.0582) i o.oo Wey TAS cos (@ +4) MOTin cos 9.4BS AD 2S a 0.7805 My 0.8544 Ts 0.1286 us The Determination of Orbits Now we find the normal place for the nearest ephemeris date, 1910 Sept. 28.8. It is possible with good reason to assume that the correction to the ephemeris is invarinble for the short interval fof time between the observation and the ephemeris date; moreover, ft is impossible to judge its variation, The true coordinates aro gain changed to mean coordinates referring to 1910.0. Besides sand 8 themselves, their mean errors are given, consisting of the ‘arrors of the micrometer settings, derived from their internal agroe- ont, and of the errors of the coordinates of the comparison star. 1 1910 Sept. 28,5, iy vera, 19847"28*.15 Be vorn ~28°10'S9"741 ‘aa 10.05) “AS yard Bo vera 19 AT B8.73 Bo vera ~28 10 12.0 a 00°58" on 0-T8OL @ age 1d Tp 0g}30 ea 268 9 r $317.65 tb gsm Me 46.03, sin@e +c) 9.996 AK + 2.8 2 0548, Aa 418.85. cos (dea) 9.0Tésn AB = ‘The normal place of Comet Brooks in its appearance of 1910: Mean Berlin Time 1910.0 Bisie.o facos Tot0 Sept. 26.8 296°51'14".0 =28°1010.6 1727 ETAT In conclusion, we shall note that in view of the extreme diffi culty of the observations (the comet was very faint) the microme ter settings “jump”; nevertheless, they show a distinet trend with time consistent in both coordinates according to sign (and not too vasn deviation according to value) with the real motion of the comets That shows that the object is undoubtedly real (not men- fioning. the fact that bwo observers have seen it) and it is indeed Comet’ Brooks. 34. The Selection of Observations and Their Control Verification of the observations by means of their comparison with a good ephemeris permits the spotting of erroneous obsorva- tions confidently only when there are many observations. With fow observations, it is not clear what caused the large deviation—the Grror of the observation or defects of the ephemeris. If, moreover, re have to deal with a new object, for which an ephemeris often Joes not exist, the peoblom becomes very difficult, And yet, in the first. determination of an orbit it is especially important that the The Positions of Minor Planets and Comets ns observations used should not contain errors. It is, t not contain errors. It is, therefore, use- ful to dopond on observations mide by well-known obaevers nd with good instruments, but even that does not i 8 not give a guarantee, ways justify thomselves in practice. It is. geatifying, if there are observations separated by small intervals of time of the order 0.i day or loss. In such a case, we might try to see how the difference of the observed coordinates agroos with the motion of the object for the given interval of time. 1a almost always possible to onoulle that 1 set of even a modest number of observations exists, moro ‘of leas uniformly distributed in timo, it is possible to detect the existence of gross errors by examining the trend of the divided dif- ferences, as recommenied by Stracke (36). ‘To bogin with, the first difforonces of tho moments and coordi- nates are formed, denoted hy f,, fa, ff. The divided differences Agefysf, M8 = fale (4-28) rofer to the moments fq, representing the arithmetic moans betw the corresponding . : * : betweon ‘Tho process is repeated for Aa and AB so that Aa’ w tha fig AB= Thy thie (4-24) Example. The following observations of the planet 1983 NA vot fioss0 ff 1 1933 July 1.90002 1oh2s"2F.28 15.895, 2 W895i 13 18 28 6.991 8 aasiseo $0.00 4.027 4 arsv209 56.00 2018 5 29.59118 19 3 48.85 18.081 6 48.8564 16 57 21 36 9.989 t 58.8120 18 59 18.05 120 The Determination of Orbits te Big fgg ARE costes 808 oat oer 909.16 20850 tt an 199.09 20880 aT gs ~ Sts 88s 10.430 “on 420.54 41.96 a cuameaaa MATS mee +3135 42.16 tira 58.7 sas Chapter 5 THE GENERAL PROBLEM OF DETERMINING AN ORBIT BY THREE OBSERVATIONS 35. The Statement of the Problem ‘The motion of a planet or of a comet around the sun, in the first approximation, i.e., without taking inlo account the auraction of the other bodies of the solar system, is determined by means of six constants entoring into the solution of the differential equa- tions of the problem of two bolies. It is clear and often more con- venient to take, as these constants, the elements of the orbit, but of course this choice is not unique. ‘Thus, for example, in place of the three elements, w, 1, %, determining the position of the plane of the otbit in space and of the line of apsides in this plane, the Gaussian consiants 4, B, C, a, 5, 2, ot the vector elements P and Q, can be introduced; these represent, by virtue of the existence between them of identical relations, ‘only three independent quantities. Also, it is possible to replace all Keplerian elements of the orbit by three rectangular coordinates at a selected moment and by the three corresponding velocities. Such a solution gives a satisfactory representation of the real motion of a planet or a comet during a relatively short interval of time, as long as it is possible to neglect the action of the other planets, ‘Therefore, the methods of determination of preliminary orbits are built on the supposition that the number of observations is small and that they aro separated by short intervals of time, In further deductions, one assumes that the value of the ec- contricity of the arhit heing datermined is not subject to any kind of restrictions, and the general mothod given below applies equally to orbits of any form (if one is not concerned with details), But in practice ono has to deal with orbits whose eccentricity is not much greater than, say, 0.7 (for periodic comets), because in the ease of strongly eccentric comet orbits, the initial ealcula- tions are slways made under the assumption that they are parabolas. The solution for our problem consists of a series of steps which we consider in their natural order. In many textbooks, some of the 121 12 The Determination of Orbits pertinont questions are singled out and considered earlier, but this merely Teads to a bass natural order of presentation, 11 is not difficult to convince oneself that three obsoralions anade at the instants ¢1, fay t9 are theoretically fully sufficient for finding tho six elements of the orbit. These three observations frive six independent. quantities, the geocentric spheriea aa ee ay a heliocentric coorinats ofthe ody ar ro ated to each other by nine equations (B-11) p, cos 8, c03 a, see 4 Xi pcos Brsin ain yse Me ¢ =1,2,3) (1) pysin 8: = 214 Ze ‘are known from the observations and Xi, Yes zi in i oa Mz ae ahr oy Nifieance for the theory of the problom, we can for the sake of a ax man ae to ba sul nes an aera tag ath wean en of divi ion, “it is evident that we obtain six equations with six the eric) ‘of the ‘earth (determined relative to the center xen I a the coordinates X;, Yi, Zi, with inverse sign)—draw th ree rites ul double the areas of the sectors between the ing, of the foens), then Determination of an Orbit by Three Observat ns 123 first and second radius vectors (7 r2) and between the second and third radius vectors (zs), rospectively, must, in agreement with the second and thied Jaws of Kepler, be equal to Cet (rata) =k VP (ta ~ ta). With the aid of the conditions shown hy these wo equations, tho position of tho plane of the orbit can be determined, since it is completely characterized by (wo parameters, for example, the inclination to the fundamental plane and the longitude of the node. AAs soon as this is done, the points of intersection of the observed directions to the celestial body with the plane of its orbit are de- termined, i.e., the coordinates 2,, y,, 2, in equations (5-1) and at the same time p; are found, The first patt of tho solution consists in finding from (5-1) the valuos of p, which satisfy (5-2). This can be done by means of successive ‘approximations. In equations (5-2) the parameter p of the orbit is unknown, and we are not able to find the double areas of the sectors (rss) and (rzra) on the left side. We denote the double arcas of the triangles between the radius vectors and the chords of the orbit by [ria] and [ras]. Assuming that p, is known, we can oblain the heliocentric coordinates of the body’ ,, y,, 2, from (5-1) and consequently [rs ra] and [rors]. Further let 9 be the ratio of the area of a sector to the area of the corresponding ti- ‘angle; knowing this ratio we can ovidontly obtain the needed areas of the sectors. The introduction of 7 in the thoory of the determi- nation of orbits proved to be particularly fruitful, since the ap- proximate value of this quantity can be given beforehand. Furthor- mote, Gauss has shown that it is possible to find 7, for a given Position of the body and for given intervals of time, with any de- sited precision without the knowledge of the elements of the otbit, and, in particular, without knowing the value of p. Thus, taking some approximate value for 7 it is possible to find p, from equa- tions (5-1) and (5-2), then with these to obtain the heliocentric coordinates of the body and by means of thom to find an improved value of j. Repeating this process it is possible to attain in the fend (usually it comes soon) initial and final values of 7 agreeing with each other. This will show that the obtained values of p; satisfy the conditions of the problem. In this lies the basic diffi= culty in the problem of determining an elliptic orbit, and the method based on the idea of successive refinements of j carries with good reason the name of the method of Gauss. (5-2) 124 The Determination of Orbits The second part of the problem consists in obtaining the or- ital elements from the heliocentric coordinates of the celestial body and from the ratios of the areas of the sectors to the areas of the triangles, and it does not contain any fundamental difficulties. In its practical development the Gaussian method, in its en- tirety, has been adapted to the peculiarities of logarithmic calcu- lation. In connection with this, in place of the observed equa~ torial coordinates of the body, the ecliptic coordinates were intro duced, because the system for which the fundamental plane is the fcliptic and in which the third coordinate of the sun, the latitude, tan be taken as equal to zero has some advantages. For calcula tions with an acithmometer these advantages disappear, and it proves to be more convenient to koep the equatorial system and (0 fnake Tull use of roctangular coordinates; this leads to a simplifi- cation of the formulas and to the shortening of the calculation process. Gradually the calculating machine has completely sup- planted the logarithms for solving these probloms, and in our time hardly anyone is going to compute an elliptic orbit in its entirety with the aid of logarithms. Maybe it is reasonable to proceed in the manner recommended in Merton's work [20] devoted to a modi- fication of the Gaussian method in the applications to arithmo- netric calculations, namely, to have recourse to logarithms only jn certain stages of the work. However, this combined method of ‘operation seems not to have obtained wide recognition. We restrict here our exposition of the Gaussian methoil to its present form, 36. The Equations of the Plane Let Aas By + Ce =0 (6-3) bo the equation of the plane passing through the origin of the heliocentric coordinates at the center of the sun, and let at least ‘one of the coefficients 4, 8, @ not equal zero. Substituting the coordinates #;, yi, 21 (7 = 1, 2, 8) of three positions of the body in (6-3), we shall have dy « Byy +021 « 05] day Bys + C29 =0, es fay 4 yn 6 C25 =O ‘Tho condition for the existence of @ plane consists in these equations giving a non-trivial solution for A, B, C. For this, as Determination of an Orbit by Three Observations 125 is known, the determinant of the coefficients must be equal to ei yay ea ys 2 (5) ts ys 25 Expanding the determinant by elements of the columns, we ob- tain the equations (5-5) in throe forms: . #1 Lys 2s ~ anys) ~2alys aa ~ 21 ysl + ealys 22 ~ 21 yal = 0) ¥ilea%5 ~ 2929] ~ ya ler zy = 2129] + ys l21z9 ~ 21 20) = 0,7 (5-6) eles ys ~ yaa) - 2aleiys ~ yi es] + afer yo ~ yr 20) = OY which are distinguished from one another by cyclic permutatio Of tho Tere, ie shall intzoduce, tn placo of the expressions Standing in the brackets, the teas of the tianglos hy usm tions of the type an a Ya2s ~ 2245 © [rata] cos (1, 2), where (n, 2) is the angle between the normal to the : 08 0 the plane of the orbit, and the x axis. By cancellation of the direct ion cor i the normals we obtain: d f the direc snes of Urgry] 2 Erp ryl ey + Cry ral ty = 0, Urgrgl yy = Cry ral ya # Urs ral yy = 0, 1) Urgrg] 2, — [ry ry] 29 4 [ry rg] 2g = 0. Wo shall divide these equations by (r) r5] and suppose frarad (rsral Be wee (5-8) This gives + nga = 0, MyTy~ Vat My Vy rf (oa) Xt wo now consi: and ng to be known in equations (6 which wo cnn do only by wsing tho conltions associated withthe Cine intervals hetwon observations, then the expressions. (59) hecone indepondent equations among 2, ¥p 2p In equations (Ge) we shall denote for brovity tho direction coninos ofthe line of sight co tho celestial body by 126 The Determination of Orbits aj = €05 5; 608 i, (10) and oxpress 21, yi, and 2: by them: #,=a;p)- Xo a2 bei Yo (1) tee ca Atwor substitution in (5-9) wo shall have: my py ~ Agha + 5g Py = MX, = Xa + me, wn Yy-¥otmaYarp (92) Bymyprm Papa * Patera 0, ypy = Cady + Ca Mas = M124 ~ Za Ms Fae ‘a wns, My, My, and p;. Tt is ‘those equations contain five unknown, my, Ras tnd py ovina that it oquaions (5-12) are solved fer py, then the 9 are areas iy known quanition and by my and ae_ If approsima, eee att na are taken, we find. the approximate py 8% tthe appeximatons of mandy will give improved oy vals Pr fhe determination of py for oxampley it fs mero aia tageus to eliminate spy andy py from (52) elght away rather than p and py. This gives -Doxed (53) where ay Xr Xa sma Xa a jas Xi ay er a Zi~ZaeneZs es 1 Zi eal jay Xa as ay Xs aa i Ya bs nf Ys Bl emeinse mt Jer Za es jer Zs ca cay a. a a5 a: Xy ay e126 er fies Jay Xa as a, Xs as dy =|bi Ya bal, da =] be Yo “| Jey Za es er Zs es errr rere serceeneeneenrccer er ran POS RE RERPETE Determination of an Orbit by Three Observations W In practice the solution is not carried through by means of de- terminants, but by numerical elimination of nyo, and nyp,. We note, by the way, that the determinant D expresses six times the volume of the teteahedron with vertices at the origin of the coor nates and at the observed positions of the body on the geocentric celestial sphoro with radius equal to unity. In order to go further it is necessary to introduce approximate values for ny and ns. 37. The Exprossion of the Areas of the Triangles by Time Assume for the sake of simplicity that the plane of the coordi nates ay coincides with the plane of the orbit. Then the equations of motion (8-73) assume the form ee ay ae oe ae (518) In what follows it is convenient to choose the unit of time so as to make & equal to one, by introducing in place of ¢ the new variable Take (5-16) Thus, our new unit of time will be 1/k = 58.134087 mean days and equations (5-15) will be replaced by tho following: ae 2 ee (5-17) a mw Let, at the initial momont, when 7 = 0, the body have the co- ordinates 2 and y and the radius vector r, and at the moment T let. the corresponding quantities be 2", y’, +. It is required to deter- mino the area of the triangle between the radius vectors 1, 1” and the chord of the orbit closing them, considering r and its deriva- tives with respect to time as known. We assume that r does not become zero, so that, for intervals of time that are not too large, the coordinates of the body ean he ‘expanded in a convergent Taylor's series in powers of T: re a ce ama oo a (18) wi a 1e Determination of Orbits io The Determination of Orbit Determination of an Orbit by Three Observotions 129 Differentiating equations (5-17) we can obtain expressions for tho higher derivatives of 2 (and y) in terms of T, which containy de hosides rand its derivatives, only # and besidi its Lives, only a aHis- tay tea k(a~ ty Tas bla -t), (524) lacing 2” by 2, and 25, res} tively), because the second and higher-order derivatives of x can fi Sama # nds, rospectively (similarly for y), we must in be eliminated by the aid of (5-17), namely, ; (5-21) subottae -y and Ta» tespecively, for T which gives ws ‘G1, Bs, Gs. ATtor this we shall have in place of (5-20) sede A de A 5 oS seafinso(#) etna) iT), at) (5-19) | oy ‘ (6-25) nePy 0, (& a y i % (2); vn Pant o(2). | tone, Having introduced these exprossions in (5-18) we find \ int ft (418) } bile eum 79; =-0, 4% ae { Ugryl= 2545 ~ Yate = 0p VB 628) oa | Uryra stays ~yita = F103 ~ 4, Fe) VPs where (21) on Consequently we can, from (5-20), determine the required double area of the triangle: Consequently, dovblo the areas of the triangles by the equations the triangles are determined ay z (5-22) wary et a(e Gare hut the parenthesis on the right is tice the sector velocity a6 a ja coon from (2-11); since = 1, it will be, according to (2-20), 7 equal to vp, which gives a staf uit a) : | tel Gv (5-28) ha lah toy (5-28) Wie apply this rosult to the three positions of the body with eo. ins Bt dor tafe ma coedinates 2, yyy 2; and radius vectors re(i = 1,2, 5) Ne denote ja, as it is usually done, by es the intervals of 130 The Determination of Orbits [As for the ratios of the areas of the triangles previously desin~ nated by us as n; and ng they will be expressed as follows: ay = ated af} ttast 6 re “Tard” Ta ‘These formulas are correct for any conic section with the focus at the center of the sun, as is obvious from their derivation. 38. The Fiest Approximation for Heliocentric Coordinat ‘Tho expressions obtained for mand ng are only approxima- tions, and the question arises as to which of the terms it is neces sury to keep for the calculation of tho orbit. *We find with the nid of expressions (5-28) the aren of the tri- angle with vertices at the ends of the radius vectors, rr, r2, and rs: titata rural trand= inl avF gee te (80) It is ovident, if wo neglect the terms in (8-28) containing +} in tho denominator, that the expression (5-80) gives zero for the arca ‘of the indicated triangle, which is only precise for motion along a Straight line. ‘This, of course, cannot be allowed, even as an ini- dint approximation, in the determination of the orbit. Consequently, ic is nocossary to keop the terms with 1/3, Further, terms with (22), ae onto n finding the et apron at), to neglect them. ‘Thus, we shall set If, however, wo proceed differently and do not take py OF tay which is the same, as the basic unknown, but p, and py aS pro- Determination of an Orbit by Three Observations 1 posed by Oppolzer, or all three geocentric distances as in the method of Gibhs, then the precision of the fiest approximation is considerably raised, but at the expense of a substantial complica- tion of the formulas. ‘Thus, in practice, thore is an advantage in the Gaussian method. : Formulas (5-31) wore given by Encke, and they are more exact than those which Gauss himself used, We shall substitute in them (5-39) and wo shall have monte 3 ce) 1 Iisteoxpromsions: may be ubetitutod tn (6-4) and (6-18) which give ayy + dyyy Dog = dn? ~idy + dgng + 1, (Bat) (5-38) where by Gant nda + dant divi tdava 9 = SE toe SEO (6-8) Thus we have expressed p, not by two unknowns, m1 and 29 but by one unknown only, 12. But betweon p, and ra there exists, in addition, a vory simple geometrical rolation resulting from the consideration of the triangle formed by the sun S, the earth T, and the body P (Pig. 12). If the exterior angle at vortex T is denoted by 02, then FPR} 6 2g 0g 008 94 4 93s (537) 132 The Determination of Orbits wwe have RE eXD YE +24, My cos 0, ~~(tgXz Pa Yo + ea %2) 68) . 7 clarified, If it is divided term- ‘The last equality may easily be el m wise by Ra, then the right side represents the proxtuet of the dite Mee oy aia of the vectors. 7S and TP takon with inverse sign. enguations (G11) aro squared and added together, we obtain b-4'7) and (6-38), which replaces the geometric reasoning. Fa an eo (pat) porn ut to fad pa ae Ione of Chece unknowns is oliminated, an equation of the eighth dogree ix tained for the other. OF course, it is not necessary to fo io instead, bath equations are solved by successive approxi; eat a ttnty convonent when a tble giving 1/* with argument r? is available. . t acuument "gown the equations (8-35) and. (G87) can be re, duead co one ranseondental equation easily solvable by use of dgacthins, This equation played # major role in the classical tee eye Gaussian method. M.A. Vilyey has devoted an im featant,momoit to its investigation (4), and T. A. Banachiowicr por artylished tablos {11 permitting one to find its golution almost ane epsteulations- For calculations with an acithmomoter it i ithowt venient vo use equations (5-35) and (5-87) withovt trans Gaussian equation. From triangle STP in Fig. 12 in which the angle at ver- tox P is denoted by 22, ¥6 . find ‘, Pa To Re Ite cay ands Sing’ SRF It the values of py and ra Figs 12 found trom these relations . fat na nm 8), emit ne Ie a aation (9-27) hocause the geometric dependence between Pp Ft sy imac, After @ sal psfornation we find ° fea sin a2), Wem 2, ain a BE in® Oo = Ra sin 0g cos 22 + (Ihy COS 02 4 ko) sin 22 = Determis ion of an Orbit by Three Observations 133 Here we shall set ke in g= Ra sin Oa, 00s g= Rs cos 24 ko, m= yy using =Rz sin G2, «00s = Ry cos Oy + & Wa and wo shall have sin (29 =) = m sin‘ 2a. ‘This is tho equation of Gauss. 39. The Analysis of the Fundemental Equations and Lambe! Theorem Concerning the Curvoture of the Apparent Orbit We raise an important question: Does the system of equations (5-85) and (5-87) always have a solution, how many solutions are there, and are all of these solutions suitable for finding orbits? Returning to (5-13) and (5-14) we see that a unique exceptional case requiring special consideration will be the case of D = 0. Considering the direction cosines aj, bi, 01 a8 the coordinates of ‘a point on the celestial sphere of radius 1, we find that the de- terminant D being equal to zero represents tho condition chat all throe positions lie in one plane passing through the origin of the coordinates—complotely analogous, for example, to condition (5-5) found previously. The apparent positions of the celestial body will lie on one great circle. In this unique exceptional case, unless p, is infinitely groat, d must also be equal to zero, whence it follows that either at least two of the determinants d,, da, ds differ from zero or all three determinants equal zero (because if two of them equal zero, then it is evident that the third one also equals zero). In the first case we can admit that at Ieast one of the magni- tudes of dy and ds doos not equal zero. After that, writing on the basis of (5-84) diy + dove dint = da +dyng + eon it is possible for us to determine r2 and then to find p4 by moans of (5-97). However, this determination will be vory unreliable due o tho smallness of the coefficient of 1/r3, If one of the determi- nants di equals rero, it is possible to conclude, by analogy with the preceding deduction, that the corresponding position of the sun lies in the samo plano with the position of the body and the 134 The Determination of Orbits contor of the earth, in other words, on the same groat circle of the olestial sphere. In this connection, if, in particular, da = 0 (the Second position of the sun is on the same great circle as the cor- Tosponding position of the body) and, moreover, if ts = Tay we Shall have no 2-22 and vy © vs, and in place of (5-89) we obtain which implies a; + dy = 0, because nf and v7, are positive. The Squntion for +2 becomes the identity 0 = 0 and it is impossible to find Fo. [Lat us turn to the second case in which dy = da = ds = 0, and in the boginning let us assume that neither one of the positions of tho celestial body coincides with the other on the celestial sphere. Thee. all three positions of the colostiat body lie on one great irelo, and all three positions of the sun are on it, too, since this fallows from the written conditions. Consequently, this great circle is the ecliptic. Equation (5-39) becomes the identity 0 = 0 find it is again impossible to determine ra. ‘Among the exceptional cases may be those in which two of the throe positions or all three positions of tho body coincide (the first can happen if the celestial body deseribes a loop during the ine of the observations, and che second only under special cit: cumstances whieh will not occur in actual orbit ealculstions). If the first and third positions coincide, all three determinants dy, de, dy vanish and our previous conclusion concersing the impossi bilhty of caleulating the orbit remains valid, but this time on an- other basis. Tot us notice, for the sake of completeness, that to this group holongs also the ease of xo positions that are diametrically op- posite on the celestial sphere, corresponding to 5 = as, b1 = eee, Ca ees. Turning to tho initial system of equations (5-12), Jo oo' that itis impossible to find mp, and np, separately, be- MNase theie coe'fictonts are numerically equal (with the same oF Cah opposite signs). Thus, only their combination mp; + ra 09 Tan be ebtained from the equations. However as far as p is com Cemed, it ean be found, because in this caso it is sufficient to take two of the three equations (5-12). ‘One is not likely to encounter a set of observations of « minor planet of of a comet strictly fitting one of the exceptional eases vensiored, but it is clear that if any of the enumerated conditions Geo observed, even only approximately, great uncertainty would be introduced in the determination of the orbi (5-40) Determinati 1n of an Orbit by Three Observations 135 In order to avoid disadvantageous conditions for the determina- tion "of tho bit, it in neconsury to roplace tho unsuccensfally chosen observations with others (if other observations exist). However, if all three observed locations of the body lie on the ecliptic, it will bo impossible to romovo the exceptional cireum- stances’ in this way. ‘The planes of the orbit of the earth and of the celestial body coincide, and we shall never be able to calcu- late the orbit by throe observations. In fact, two of the elements of the orbit are eliminated, since i= 0, and § loses its meaning. The remaining elements, T, x, ¢, and e, cannot be found by using the three remaining independent observational data, ists, Ary Aa, As. In order to solve the problem, it is necessary to take into con- eration a fourth longitde andy therfore, four observations are required to calculate an orbit with a small inclination to the ecliptic. Irvin we ata show ho, fom he route we Ma htaineds te \eoren of Lambert. concerning the curvature of an apparent orbit can be deduced, * fare of an apparent exe If the observer were to change position in space strictly ac- cording to Kepler's lnve, then py = O would havo to aalisty oqun tions (5-35) and (5-87), because the orbit of the earth would satisfy all requirements of the problem (if one does not consider perturba: tions and neglects the mass of the earth). In reality, these condi- tions are not observed exactly, but the deviations are small. Therefore, assuming p, = 0, we find from (5-37) ra = Ra (which is evident), and from (5-35) we obtain key = tgs, (an fn equation which can serve for che approximate control of the cal- citation of Fp and Zq. In place of (535), it is now possible to write (5-42) Equation (-42) contains Lambert's theorem in latent form: If a celestial body at the mon ment of observation is located farther from the sun than the earth, then its apparent trajectory on the sky is convex toward the sun; if, L however, the celestial body is a closer to the sun than the earth, then the apparent trajectory is Fig. 13 concave toward the sun. Indood, Ecpie 136 The Determination of Orbits it rq» Ray it follows from equation (5-42) that Zo > 0, heeanse ao 0. ty howover, 23 0, because vir va» And Ry are positive, Consequently, f2> Ry, and wo obtain the first part of Lambert’s theorom, If the signs of the doterminants feo different, the porimoters are traversed in the opposite direction {us shown in Fig. 18), whonce it follows that the apparent path of (he eotestin! body tums its concavity toward tho sun. Here, to < 0 fand rz Ra, which gives the second part of Lambert's theorem. When ry passes Ukrough the value equal to Hy during the motion of Determination of an Orbit by Three Observations 137 tho colestil body, tho apparent obit a be Js tho apparent orbit has 9 point of inflection Lanier’ theorem says that the ; exceptions! case considers above taton place actly whon ra becomes cual to if more a the celestial body is not moving in the plane of the ecliptic. ow Tot aa take equation (S13) and ein to (©3T) voile) _ (6-41) TB = BE 20, pp 008 05 + 63. Wo shall set 15:2, = 2, pis ) Ig: R§ =, and cos In place of (5-47) will he By Tes RE = Dean 08 Oo ( -a) (18) a als doys y%, (5-49) 2>0, y>0, -18 Eliminating y from these equations, we have f(z) = 28 — (14 Rel 4) 284 216 4 2-12 0. (5-50) This is the so-called equation of Lagrange for r2, on which the in vestigation of multiple solutions of our problem is based. ‘The de- froo of tho equation can bo reduced at onco to the soventh if the root z=1 is excluded, correspond , corresponding to the terrestrial orhit (ra zh ‘The second eooffcien in aleays negatives because Le dels Pel -e%)e (e+)? Lot 1> 0. Then accor t0(6-48,.1 = 3, > 0, han, 2 dan Yet He 0, thon 1+ > O56 < 0 and ot os ay, then om atel pce + y? it follows that in order that 2 > 1, it is neceaty for yo 26°) or y> 20% Bat then 7> Bo and Co ininy Poo Mare 13 On aoe fe +) > 138 The Determination of Orbits L If <0, then according to (5-48), 1 = Z, <0 ore <1. Whence, from (B49) it follows that y? 1 Bey <0, ies, ¢< 0, which gives ee Toads to the conclusion that jn (5:50) there exis tree changes of the sign. Consequently, the number of positive roots an be wither 1 or 3. However, onc of the roots corresponds to the orbit of the earth and, therefore, in addition to it, for ench netunlly observed celestial hody, there exist two valid roots of Bf the basic equation of the orbit for ra, But this still does not mean that they both represent the solution we seek. Ie is nocessary also to observe tho condition py > 0, expressing the tnct that the observed positions lie on the line of sight, an nat omits extension backwards, We shall obtain two usable solar ons only in that case where the two roots for ra are. simultane- ere oa simatanoously salon than uni Hf Poweye, fone of the roots is groater than unity and the other is smaller than nity, then one of them gives, according to (5-48), y < 0, because the wansition of z though 1 for a given Zchanges the sign of Ye In order to clarify what conditions determine the double solx tion, we shall oxclude the root x = 1 from (5-50), representing /(#) in the form f(@)=(@-No@s (5-51) ap: whore g(z) has two positive roots. From this equation, it is parent that g0)=~F() =P, 90) > 0. Let us find g(t). Differontiating (5-61) we find ae age) 1) gyre SO, whence afte vo-[Fh: Aftor a small calculation we obtain from equation (5-50) g(t) =2 (1 8el). Determination of an Orbit by Three Observotions 139 If g(1)> 0, then either both or neither of the roots of g(z) lies botweon O and 1. Consequently, the roots of g (2), i any case, multaneously either greatet or smallor than unity, and we have a double solution for the orbit. If g(t) < 0, then one of the roots of g(2) lies between 0 and 1, the other is greater than unity, and the orbit is uniquely determined. We find, therefore, on the basis of (5-52), that a double solution is possible if 1- Bet > 0, or (5-53) Blo cos 02 < RS, which is the criterion found for the first time by Oppolzer. If there exist two solutions satisfying the given three observations, cach of them gives a system of elements of an orbit and only further ob- servations are able to indicate which orbit is the real one. ‘There exist tables (18) simplifying the solution of the problem concorning the roots of Lagrango's equation, if the nocessary ini- tial quantities are given. 40. The Improvement of the Results of the First Approximation As soon as rz becomes known, the equations (8-38) give us ni and ng, and roturning to equations (5-12) we shall obtain p, and py with least difficulty from the equalities resulting from the process of liminating the unknowns. Next, the heliocentric coordinates yy Yin 2, ate easily found from equations (8-11) and acteyleet, dasbeyteek (5-54) In place of this the equations may be utilized, which result from triangles analogous to the one represented in Fig. 12, r= R242R,p, cos 0, + ps 12 = R2 + 2Ryp, cos 05 + p2, (6-55) where RKB yt S23, RE AXES V24 23, Ry cos 8, ==(a1Xy + BLN +012 ay (as Xs + bs Ys + 082s). Knowing 9, 92, and py, we are able to take into consideration the noglectod ‘influence of planotary aberration on the moments of Ry cos 64 vo The Determination of Orbits observation. The corrected moments on the basis of (4-8) will bo Ap Ga ty~ Apa Ba ty- dog 80) and it is necessary to substitute thetn in place of che initial fin are oe iGalation of the formplas in the second approximation. Mae eis possible without Wiffiulty to ealeulute the higher terms in the expansions of my and ns by powers of lime, But such eth of approximation is a concept which goes back to Lagrange see itt her eumersone. Gnuss has shown how it is possi- le ta find completely rigorous valwos for those quantities. Lae tern (para) (ers) be twice the area of the sectors be- cee he redGtive radius vectors to the body. We shall desig. ae owes the ratio of the areas of the sectors to the areas of the corresponding triangles by J,. In other words, let us suppose 0 if 2f < 180°, since this follows from the ex: woo rce for m ant T, i is ensior to make wse of tables siving revenge for the negument A And So the solation js achieved in Determination of an Orbit by Three Observations us the following way. The quantities m and f are found by formulas (5-68), and noxt (5-77) gives A. Here at first it is assumed that & = 0, because from (5-T4) it follows that is a quantity of the fourth ‘order of smallness. Having found y by (5-18) or from the correspond- ing tables, we find x according to (5-75), which pormits ¢ to be cal- culated by using (5-T4), or again the tables (Table VIII) can be utilized for this purpose. Then the calculation of fi is repeated and with tho new value of %, @ and & are determined until the approximations ageee. AS a rule it is not necessary to go further than the second approximation. For small values of 4 Hansen's method is also convenient. It does not require tables and is as follows Lot us determine A from the equation (5-78) and lot us sot which gives (8-79) But since Arara(ir s ) (He) 7 it follows from (5-78) that Bak io‘) * 1007 ‘Tho last torm on the left side is very small, and for small in- tervals of time it can be discarded. Then we obtain 6 The Determination of Orbits (5-80) This is the continuod fraction of Hansen, which is very easily calculated either with an arithmometer oe with logarithms. Here, too, it is possible to neglect the quantity £ in equation (8-77). Conse- quently, we obtain (6-81) Oppolzer has shown, assuming that the eccenteicity of the orbit is small, that the error in the formulas of Hanson is less than one unit in the seventh decimal for 9f < 18° yon sixth ap< ar? hone ifth wow pega ‘Thus, the method of Hansen is entirely suitable for the preliminary determination of planetary orbits, because here 2f is usually con- siderably loss than these limits, Encke, Tietjen, and other authors also gave their versions of the formulas for finding the ratio of the area of the sector to the area of the triangle. Let us note that in cases of parabolic and citoular orbits this ratio does not play such an important role as for the ellipse. It romains to show how to introduce the rectangular coordinates of tho colestial body into the calculations. Let 2, y, 2 and 2", y’, 2 be given for the instants ¢ and ¢’. Starting from rr 008 Of = 2" + yy’ + 22 and introducing the notation 22 4 ae" 4 y's 28") = drt’ cos? f, (5-82) we shall find from (5-68) and (5-81) 2 Ifrew we ele) a (5-83) Determination of an Orbit by Three Observotions ur For the determination of the orbit, the quantities 9, (= 1, 2, 8) are introduced according to (5-87). Cortespondingly we shall have i, mi, [iy hee In each of these quantities will enter the coordi- nates Of tho body and the radius vectors with uw subscripts differ- ing from the corresponding value of i. 42. The Second and Further Approximations By making use of the quantitios J, it is possiblo to reduce tho equntions for the determination of p, to the same form a5 bofora. we have ‘Thus, equations (5-85) and (5-817) remain im are solved anew; n,p, and ngp, are again obtained from tho basic equations, and then p, and py. The heliocentric coordinates are calculated by formulas (5-11). It is possible to verify and, if nocossary, to correct for the aberration time (if the p, wore notice ably erroneous). After this the y, are calculated anew. ‘The proc- 88 of approximations is continued until the values of my and ns coincide in two consecutive approximations, after which it is pos sible to pass on to the determination of the elements of the orbit. Because nm; and ng are now known, it is possible to find from equations (5-9) ye mye i mgeyy Ya MY # Mg Vay and then, using (5-10) and (5-11), to obtain from the equations: Py C08 By C08 ay = Xy 4+ 2yy Py C08 By Sin chy = Py + Yay (6-86) py sin 8, 24 + ayy us The Determinotion of Orbits the coordinates of the celestial body a, and 64. Comparing them with the observations gives control of the calculations. 43. The Determin The results obtained by the last approximation must. serve for Finding the elements of the orbit, From equations (5-60) we obtain for the parameter of the Elements _ tuts sin (02-23) — Ts; ae From these three equations it is ordinarily suff use of the first one which gives the groatest accuracy because its numerator and denominator are larger quantities than those in the two following equations. In order to calculate the sine of the difference of the true anom- lies one could make use of the formula (y~%) = VU AY Gy 8 Aad + He VF a)” 88) Indood, the square of twice the aren of the triangle contained between r, and ry equals the sum of the squares of the areas of its projections on the coordinato planes. Similarly riry 008 (y=) = Fy ey + Vs Yat 212 (5-89) Prom (5-88) and (5-89) we find sin (vs ~ 01) and cos (vs ~ 2 ‘They must correspond to the same angle, However, formula (5-88) is rather inconvenient for calculations. ‘Therefore, another method was recommended for finding r:r3 sin (a = 04). Lot us set Pity t Vita ty 2a MTe Os (vg ot 74.608 (8 ~ 91) (5-90) and let us introduce the notations Byetyn om, Yon Yan he } oo rhehe yb eet. Determination of an Orbit by Three Observations ve From here it follows that Pha rB—20(e 25 + yy + 22g) 4071 == 084 213 sin? (0, -0,) and furthermore Fits sin (6 — 95) = rio. (5-92) For calculations by equation (5-92) it is possible to use for- mula, (5-88) for control. For determining the true anomalies we return to the equations r= ts ‘" Tyee’ which give e cos v, =a, and ecos vy = (6-98) ‘Tho Inst equation can he represented in the form € 008 ¥, cos (0 ~ ¥,)-€ sin v, sin (Ys —¥,)= dys whence 4, 008 (05 ~ ¥4)- %5 esiny, = {1008 a= t= (5-94) sin (2, ~ v4) Equations (5-93) and (5-94) permit. us to find ¢ and »,. Conse- quently, v5 =; +(e» ~ 01) is also obtained. ‘Then we calculate ¢ and a by the formulas sing =e, a=psect p= (5-95) or, if the orbit is nearly parabolic, ? os (5-06) For elliptic orbits with moderate eccentricity we find the oc- centric anomalies by the formulas (2-42) ine 1 1 tan tey an dy, Br Vayetr a ton Ley 2 1 tan ys, (64 Tre tt gts 5-07) 150 The Determinotion of Orbits ‘and then we apply Kepler's equation giving the mean anom MyeF,-esinEi, My-Es—esin Es, (5-98) whore £ is expressed in degrees or in seconds of arc. The second ‘equation of (5-63) can serve for a partial control of the calculations a.cos d sin $(Es ~ Ex) = rita sin d(es ~ v1). (5-99) Caloulation of the mean daily motion is made in two ways: My = Mr Bo geh” Cae wee (6-101) n= Aa Aa both values of yw must coincide within the limits of the accuracy of the ealeulations. If the orbit is nearly parabolie, wo find T through gy ¢, 01, and vy with the aid of the formulas ‘and instructions of Section 24. FFor the determination of an orbit in the general case it is not nec: essary, in practice, to take into consideration the possibility of obtaining parabola or a circle, Therofore, we do not dwell on these cases, as there exist other methods of calculating orbits for thom. Now, the vector elements and also, if necessary, the usual ecliptic elements i, 2, and o must be obtained. Etarting from the expressions for the heliocentric coordinates y= Perr cos vy + Qari sin Ory by = Perscos vs + Quta Sin Day eeey we find ve cristae a ” ven @anee ey gg = Eategue te = zite con ne Firs sin (va — 0s) For the determination of preliminary orbits the angle vy ~ 01 is almost always small, and the numerators as well as the denomi- ators in these expressions are small quantities. We determine the voctor elements at two close heliocentric positions; the inaccuracy js therefore essentially unavoidable. Because it is impossible to taise the accuracy with which the vector elements are calculated, Determination of an Orbit by Three Observations 11 it would appear most natural to calculate them by tho unchanged formulas (5-102). But then we are deprived of the possibility of confidently controlling the results by the formulas (8-21) which express the condition that the vector olemonts represent n system of direction cosines for two orthogonal vectors, If we wish to ob- serve these relations exactly, we must assume that coordinates 23, Vas 24> Zgs Var 29 aro perfectly exact, and using this assump- tion perform the calculations, using an additional decimal place, if necessary. However, it is hetter to do this somewhat differently, by using the auxiliary quantities zo, yo, Zo introduced above. Equations (5-102) give after the substitution of vy = vs + (vs ~ 21) and a small transformation: cos v — fey — zz cor (es =v] sino mn fy sin (va = 01)" in the notation of expressions (5-90), (5-91), and (5-92) Prey (6-108) (S108) 152 The Determination of Orbits According to what has been said, we must compute 29, Yor 20 with a sufficient numbor of significant digits, Since these values aro caleulated by formulas (5-01), also as differences of small quantities, we must (supposing that the heliocentric coordinates fice porfectly exact) know with high precision; and this refers also to 2125 +9, Ys +2125 and to 73, Thore is no noed to turn to trigonometric tables’ for those calculations. All refinement is re- duced to copying, with excess decimal places, the results obtained ‘on an arithmometer, and this presents no difficulty. Computing furthor Ayn aPy, Brea cos Qe, ete, we ate able at once to control the results by the formulas AD, Abs Ata a?, BY4 82482 =a? cos? g =a2(1 ~ 6%), A,B, +4,B, + 4,B, =0. There is no direct necessity for finding the ecliptic elements, bot nevertheless they are usually computed in view of their clear ness and usefulness for the identification of various objects. From formulas (3-20) it follows that sin isin o =P, cos €~ Py sin, sin feos @ =Qz 60 © - Q, sine, sin = (Py cos w ~ Qy sin a) see &, (6-105) cos N= Py c08 = -Qz Sin oy cos i =-(Py sina + Ox cos w) cosee 2. By applying all the written equations, we sufficiently control the calculation of the three quantities #, 2, a. It is ovident that sin 2 and cos {1 must correspond to the same angle, and the same rofers to hoth sin i and cos i. 44, Summary of Formulas In the gonoral cae for the calculation of orbits, observations are used giving the coordinates accurately to within 0.1". Cot respondingly, six-place trigonometric tables are used in the calcu- Intion. Calculations are conducted with the aid of an arithmometer, but a€ certain stages an experienced computer could find it con- venient to resort to logarithms, Not all parts of the calculations are verified hy control formulas; a cloxe approciation of the extent of these verifications is very important for the calculation of an Determination of an Orbit by Three Observations 153 orbit. Moreover, the results of the computation and the verifica- tion may be in agreement, while the quantities entering the cale Jation may have errors (which have but an insensible influence on the result of the control formula). On the other hand, the calcula- tions may be correct, while the control does not ageee well, par- ticularly if small quantities are involved. A, Initial Data They are fo Wy By Xi Va Say tay Oa, Ba, Xa, Ya, Za, © fay hay Bay Xsy Yay Za It tho orbit of the body is already preliminarily known, the mo- ments of the observations, expressed in Universal Time, are cor- rected for aberration by means of the subtraction of the quantities Ap,(i=1, 2, 3), where A= 04005772, The coordinates of the colestial 'body a; and 5; are referred to the same equinox as are Xs, Yu, Zi 10s, to the boginning of the year in which the observa: tions were made or else to the normal equinox. Xz, Yo, Z; ato taken for momonts of time uncorrected for aberration. ‘The a, and By are corrected for parallax by formulas (4-10). If an initial ocbit is calculated, the corrections for aberration are not introduced, and in place of correcting the observations for parallax, the cor” rections AX =A,, cos 3, AY =A, sina, az are introduced in the rectangular coordinates of the sun. In select- ing observations, one should consider tho possibility of the excep- tional cases enumerated in Section 39. With this purpose in mind, as a preliminary step, plot the observations on a graph and roughly estimate the aren of the spherical triangle hetwoen tho points taken for the calculation of the orbit; it is proportional to the magnitude of the coefficient D in (5-18), and it characterizes the precision with which the geocentric distances and the orbit itsel? are ob- tained. From the available observations thase are selected which fare separated by the greatest intervals of time. Equality of the intervals is desirable but not obligatory. If an unsuccessful choice of observations is not noticed at once, it will become apparent 154 The Determinotion of Orbits later, anyway, especially in determining py. It is useful to com- bine observations that are very close in time into meat Every essential error in the quantities (1) is dangerous, he- cause it will not be dotected until the end of calculations. There- fore, the initial data must be subjected to the most careful veri- fication. B. Auciliary Quantities os 3; sin af (= 1,23) (ly Control: a? +5? 40? = 1 4; ~ ¢¢ sin ay = €08 (21 + 5:)s by + 6, 608 a4 ~ sin (a + 51), cay 20, cos 0; = ~2(arXe + Bi Ye + cx Zady RB aXP+Y2 423 Control (X; ~ a)? + (¥) - ba)? + (Zs = RF + 2K; cos 6; + 1, ym, py = aR a + OaMaPy =MiXy ~ Ky HMA gs Bymy 01 Pa Pp + ByyPy =, V, — Yat my Mos avy 2m 1M Py = Cy Pa + Ea y Ps From these equations n,p, and agp; are successively elimi- nated. As a result the equation for p, is obtained in the form =D pg = dyn, ~ dy + 45mg, whore the quantities Dy dy, da, ds may differ, from their values de- duced above, by a certain’ multiplier (usually the reciprocal to the value of one of the minors of determinant 0). ©. Pinat Approzination Trekla-ty Tek - ty Te kG-o) k = 0.01720210, ett Men Ge Zonteal: ny yng = 1 Determination of an Orbit by Three Observations 155 dnt - dy + dns dw, 4d kya Miz te t dans dy -D : D wy Contsol: ky = yf (cough approximation). R2 + 21, p, 008 0, + p3- wy The last two equations aro solved by means of successive ap- proximations. In the first approximation it is possible to take Ps > ky for minor planets and to substitute it in the second equse tion, Which gives rs. With this r2, one obtains, anew from the first equation, ete. For faster work it is possible to have recourse to interpolation. Let the initial value of p, in the first approxima: tion be p and the final p’’ = p+ Apj. In the second approxima: tion it will be p% and py + Apy, respectively. If Ap, depends Tinearly on pz and p, is the tue value, thon ean es | Mey pees) Robe Pawns + eB, Paes AP Npp- Awe which ean bo taken for the thitd approximation. For minor planets the probability of a double solution, partiou- larly close to opposition, is small. For comots one may’ approxi- mate a graph of the curve f(p,) = py ~kq + lo/r3 by a sorios of points for various values of p,. This shows the possibility of double solutions and gives the approximate value of the roots. For this purpose tables can bo used {13]. Usually p, and r, ate dotorininod with small accuracy in view of tho smallness of D: Tho last docimals of p, will, therefore, be fictitious, but it is advisable to keep them, Because this assures uniformity with the calculations that follow. IF thore is peoliminary orbit, + and v, aro calculated in place of», and vy, and then oF #4, yjy 2, ate found and then ti Va cos g a = * Wear = at, Vals + Bae) which gives », and vy hy the formulas (XID): 156 The Determination of Orbits win wations (IV) and those which were obtained in the process of climinating the unknowns, give n,p, and nypy- The control con- sists in the substitution of x ps p25"? in all of the equations (IV): m1? Maes a EL, py SP, 2 = R2 + 2Ryp, cos 0, + ps (IX) R2 + 22,p, cos 0, +ph # z= a0, 4 > be; ~Y, ee el ate yP eat ark, iY, + Made D. Improvement of the Results of the First Approimation If planetary aberration has not yet been taken into account, the cortacted moments are found by e #005772. ~ Ab: If, howover, this correction was mado in the hoginning, the ro- sults are votified and, if necessary, corrections aro mado. With the Correctod tnomonts of the observations the values of Tees +MY» cau aro calculated for a control. E. Determination of the Elements ty + Ua eats ; yy Yo = Va OV» seadayd rade Determination of an Orbit by Three Observations 159 The quantities r,, 2425 + yyy + 2,255 and o are calculated with fan increased number of decimal places, so that z9, yoy 29 are ob- tained with a full set of significant figures (six decimal places). Control (it ean be omitted): (rata)? = (Wa29 ~ 2449)? + 2s 8s 2324)? © yyy — 1409)? sin oaye®, or, Byty + + 242; oy) cos (oy =,) = WE = a tte fs, Control: Tho sino and cosino must bo forthe samo angle. ween, 6 008 #, = ay3 vp 4, 208 (05 ~9))~ 4 sin, = 0) vy ars y= v1) Control: ¢ ¢08 vy = qq sing, a= 1 =e tay L tan By = tan bo, = cat (45° 5 2 Ive "2 ( 4) tan Zon, & xv L eee Controt: a cos g sin (Es ~ £1) ~ Vrirs My int (vs — m). EyesinEy, My = Bye sin Bs, BT.Q0STRe, —e”” « 90B26E.8 Te, My Ws vin shat k? = 0.985008, 548.188. 160 The Determination of Orbits Control: Both values of x must agroo. For the selected epoch 0 Mg = My + lta = 01) = Ma + allo 4) Ife is close to unity, after (XVI) calculate _2 "Tee and if iis desirable, a7 2, p=a°/? (period in years), (XVIla) ey stan boy, op «tant (xvita) ystan bos) oy = tan 5 Path E01 + Prot) == Tie os» Pyodd- 17 ire Vite P, and Ps are taken from tables with arguments €03, 863, and £. Conttol: Both values of T must agree. + Yo *, Pye BE ay AE, Qe ME MET, FID Age aPy, By=a cos $Qx, A,=aP,, By =acos $Q,, A,=aP,, B,=acosdQe+ Cont Abed? ¢A?= 02, BE4B2 4 B2 =a? cos? 4, A,By + AyBy + A,B, = 0. sini sina = P, cos © -P, sin ey sin feos w= Q, cos € ~Qy sine, sin Q = (P, c08 ~ Q, sin @) see €, (xx) cos fl = P, cos w ~ Q, sin a, 08 i = ~(P, sin a + Q, c08 w) cosee 1. Determination of an Orbit by Three Observations 161 Gontrol: sin {Q and cos fl must belong to the same angle, and the same applies to sin { and cas i. F. Representation of the Observations by the Elements ‘The moment of the second observation or the moments of all the mused observations, if any are left, are corrected for aberration time which is found most easily by interpolation between values oc- curring in D. Then, for all observations with the exception of the first and third basic observations, the following quantities are calculated: M~Mo + u(t? ~ to), E-esing @M, pcos 5 cos xm A, (cos Fe) +B, sin +X, > (XI) pcos 8 sina = Ay (cos Ee) +B, sin E+ Y, psind =A, (cos fe) +B, sin +Z. ‘The coordinates of the sun are taken for the moments of ob- servation uncorrected for aberration and are reduced for parallax according to (4-11). Or, in place of this, the observed coordinates of the colestial body are corrocted for parallax by using parallatic factors for which it is necessary to find p by interpolation between known values. For nearly parabolic orbits (see Section 23), ee feat (liar qe VEE Aa eo. meas. ‘The quantity v, denoted here by w, is taken from Barker's table. tan be : . n= fot, omtan Darah, 290 rsinv = te, p05 5 cos X= Per cosy + Qursino +X, b (XXla) pcos 6 sin u=P,rcosv+Qyrsinu +, p sin 5 =P,r cosy +Qzrsiny + It is desirable to ropresent all available observations, except the first and third of those on which the calculation of tho orbit was based. The representation of the second basic observation gives a control for the correctness of the work: the differences 162 The Determinotion of Orbits 0s BAG = (Hp = aa) e058 3 LAB = 80-8 o-¢ taken as the observed minus the computed values must not exceed V" for a six-decimal calculation. The remaining observations (if they exist) must be represonted within the limits of their probable errors; this will be the criterion that the orbit is practically cor- rect. If such an agreomont does not occur, we must assume that either the unused observations or the hasic observations are errone= fous. In the Lattor case, if the doviations are great, it may be neces- sary to discard certain of the hasic observations and to calculate the orbit anew, selecting for this purpose new observations in place of the used erronoous observations, 45, Example For the planet 1933 NA we have the following observations ob- tained by G. N. Neuimin at Simeis (Crimea): T No, UT. Aiga8.0. Frgsa.o Is 1923 July 128430 July 1.96042 19A2e™ |F98 — 19°50"7.3 2 1720 and 1785514 19 13 18.28 5 2320 18 assis to 8.00, 4 aT 40 STL 2ts7298 19 6.00 5 2021 28, 2089118 193 43.85 5) Aug, 1T20 7-8 8.85264 18 87 21a v 2720 126 8.84204 18 80 13,08, Asterisks mark the basic observations selected for calculating the initial orbit. These observations have received the subscripts 1, 2, 3 in the following tabulations 1 2 3 cy 16969 0.600418 ~0.908350 oO +0.919683 —+0.750090 0.408195, @ +0.398695 0.525727 + 0.178746 As ash ens atk26™5 —aokiemg 5 Wak 20 aR 8 LT L 2600 a Wer 207.6 0 ae ~i6 =21 ay 26 +22 az 50 30 x 0.600429 ~0.905171 y +0.910710 — ¢0.751018 0.405220 Zz +0.a98605 — ¢o.828607 | OL TSTIB Determination of an Orbit by Three Observations 163 1 2 3 W cong #0.8740T0 —¢0,97¢809 — +0.258892 con 0.070847 So.96a7o1 + 0.064818 sin TogaTiz1 “o.961885 — “o.968808 ° So.sessa5 4.280216 0.248531 5 “o.o00083 “oloassae — “o-aazre7 ° “olsa9so7 “0.243037 0.262020 WI cos (+5) F0L1416068 F0.001650 “0.00767 sin (2+) yo.oago23— o.eng49 So-o00071 2R oon + 1.070388 + 1.870286 1.206899 e iosst6i + 1.050619 1.030217 npr pa napa my as TV +0.263835 0.266216 + 0.216583 = 0.160709 + 0.600429 ~0.908371 “olnooona + O:oB6a6 “ocanete7 = x O.aiaTIO Toctsto16 + 0.408320 Tolsnono7 “ol2isoa7 —olg6a0a0 = + O.20sa6s —OL2aue0T + OL1TEI8 363885 + 0.376940 0.377080 = + aTITeL —o.30387S + OL168708 Folasaeor “o!24ssa0 + 0:248403 = ~ 0.244021 + 0.199007 ~0.107911 +0.110783 ~0.190819 = + 0.202085 + 0.296894 ~ 0.744573 *oioads09 “o%orag27 = + 0.138040 0.125700 + 0.067503 Tolorases 0.014897 = ~ Oogats— —O.on3040 + 0.080873 0.016724 = + 0.191485 0.188740 +0.150677 1 2 3 v 18,95000 55.8816 28,93078 = 098018 O.978484 0.480468 2 o.s0007 — E-xyzy 0.030880 nf! o.401033, vy 0060178 vy 0.050482 vw pa = LoHTHi0~ Lovers 2. FB = 1.080810 + 1.870286 py + p2 VII py, 4 0.936638 0.020100 0.919738 0.910738 73 3.000005 9.668302 3.60634 3.604977 3.604060 Trg 0.190346 0.142944 0.146017 0.346090 0.148100 py 0.986883 0.921831 0.010816 0.910733, A’ exse7 A” — 1312 vit ny 0.517759 ng_——0.499720 mie: OASTOGS—mgpy—OLSEETIE Pr onsasad "py" 1.110852 1 2 3 IX 2 o.ssesi4 91973110852 2 sisesos3—aeoanea 604154 s0unisi6 — ¥0.848976 + 1482200 y Titasss | Tieos700 "Lav dos 2 ~o.Bi0ss0 — Lorss005¢ L487 702 , 11386513 1.808676 1.922018 164 x Ap mee XI xi EN From calculations of the second and third approximation ac- cording to formulas VI through XH, the principal results are given below. vl va vil 1X XI xl + vl 2 vil Ps The Determinotion of Orbits 1 0.00811 1.95831 2794880 049707 0.508940 44041 0.082064 0.268071 0.248001 0.0045 152 0.002887 0.008401 1005063 1.017868, 0.517900 0.061198 Loon = 1.020619 0.917407 0.517982 0.889382 1 +0.490750 1.713986 = 0.610889 1esst400 1.005989 0.517989 0.061245 soezit1 1.080619 0.017258 2 0.00881 20,88587 56,85036 0.978462 15,95208 0.019280, 0.267013, 0.957588 0.0184108 0.010178 0.021882 1.028687 ovzeaa 0.007894 {6.000003 o.02 15281 1.028636 yy =1.c0007 e.sT0280 72 5.50325 sn, 0490628 po tense + 0.844057 “1.606591 Folsaa4s7 1596305, 1.028738 tones 4 + 1.879256 py PB a.sest 3 0.00841 38.80867 2709086 0.480804 0.491040 14.19833 0.0186015 0.265588 0.280846 0.043149 0.002272 0.051688 1.008702 1.01732 0.499796 0.059982 +03 3 4 1.181370 a 1438152 = 0.486874 onss2a 1.005727 ny 0.499884 5 0.059976, +08 Determination of an Orbit by Three Observations va m4 ms 1K x y z x a XI m 0.517941 ny 0.499536 oseN234 py 1.107353 1 2 ougoseT 0.844618 =1113803 1.08392 Folg10a3t = 0.549450 1.854400 1.596252 1.005990 1.028738 0.517941 165 3 1.181319 “taaT98a 01400820 19186846 1.005728 ig 0.499886 NIM ni21 tg7s + 0.844818 ayy: 4 mays ~ 1.60659 XV oo o.oate2T1 eo +0.7219982 Yo +0.1689501 XV sin (vs = 01) 0s (09 = 91) tan (v8 = 21) 121 tnaes ~ 0.549450 0.380075 0.920825 $,0.428508 2g 40.1054487 anon 22°57 1077 "3 o-ssoss30 re artasatos rita 114008863, : XVI 1aTsos oy 1sr30".8 2uteest ‘150269 écos ty ag) SO1BATES oy 0° 54"477.8 $h OUIOT—@cotey 0.184070 esinos Tolmer" @ so50'25" 4 XVI 1- o.srsss0 cot (ise + 3) 0.854225 : ° 2.280802 tan 0.009878 wnt? 0Teson & E tan + 0.059455, tans .zae201 B souso".s | Ee 20°34"40".2 cos! O98 deepen) 11828°38"4 2.008 zpos singivg v1) 0.108406 W@a-Fy) 9°53"18".2 Vranas 1.901867 sing(Es—81) 0.171729 acon gein}ies=s) O.ts07 YFyFesindoos—w1) ocTato7 XVI sin Ey so.ssis) sin Epp 0a7439 esin fy T8aaso ein By ab eae ET me EEE MyM) 164982 2 Va 1493490 Ms — My sose2.2 at 3.30845, z To0s.257 104s" 282 4 21.0 toned 25.01168 ty Mo 18° 910.7 Mo 16 The Determination of Orbits XIX cos 0») 40.90036 sine 0.198886 eos ty:h 0.525600 sin ryiro — +0.078496 Ssinvyiry “0.185087 cos e1ire — +1.828027 Pe yorsa20 Oy 4 0.990701 Py Horas 2, +0.007689 =0.340308 Or SoonTis a 02798 By v2ana7 a, =2078T70 By Sotsi8s ar 0759000 +0.208619 XX sinisin@ —+0.058071 sin] = 0.725910, Sin feos @ — +0.048029 03 SL o.gsTTSs ‘sin f 0.07886, so°ai'as”. 4 os sosstiat S226 an 40 7 sino so77sT05 42055 3 cose 0-63 Blements Epoch 1988 July 27.0 U.P. My 13° 971078 $0 4145 ot Babe §240 3 roo. 14 208s 8 85098 4 068.252 A, ¢0.2TTI98 By +2.180447 Ay Lelorsr70 By 4 018155 4 es teh GD f BG re) } A similar expression is obtained for ng if we substitute the ex- pressions, which were written above, into the second equation of (5-29), ‘The result will be analogous to (8-7) if in it t, and t, ex change places, and likewise r, and r,, namely, x 4 pat? act? + — -—t+ ti tees (6-8) sf 3G, Ger } . Wy nay Ti, and T, will appear in the formulas for ry. ty» tai there will be no other changes. The radius vectors r, and r, are dotermined by using (5-85) i 12 = R242, p, cos 0, +p, 2 +229, 008 8, + p3. (6-9) nm The Determination of Orbits whore REG NPA? 422, RPA N24 V2 22, By cos Oy = ~(aiXi +b ¥s + eZ) (G10) Rg C08 04 = ~(agXe + BeYe + C424) Let us assume 1 (en) (6-1 ” ‘Thon from equations (6-7) and (6-8) and the corresponding equa- tions for the second group of triangles, we shall have mot 4% (y= te eat? ens AEE Ee AT En z (6-12) tt (ye eat Pen, 1, u (tye TE ATT Ene Insorting this into (6-4), wo obtain Py =(@HHE+TEn)py +K+ LE + M egw Pp, +Qy joes Pen (@ HE AV Endy, +HE + LE + WEN Pp, +O where Batt, Bat Fa$ut Fedtita, G=ae, a= are’, H=F(A-@), WeP (A -6), (eu feat, rettiay, eee ala ee Cnc L=P@-C+D-K), LaF (B04 DK), MAACPB STC), We =ACPB ET TIC). The four oquations of (8-9) and (8-18) contain p,, pyr, and ry as unknown quantities, and they are solved by means of successive approximations. If the observations of a minor planet are separated Determination of an Orbit by Four Observations wm by small intervals of time, the terms with y are very small and can cofton be simply discarded. If thore are no indications for the values of the geocentric distances (for comets it is necessary to make use of the results obtained in the calculation of a parabolic orbit), then in the first approximation for minor planets it is possible to take Py wre = 2, iO. E = 0.008, 9 = 0. ‘As soon as the process of solution gives the values of p, and Pg, it is possible to proceed to the calculation of the elements, which is conducted entirely as in the Gaussian method set forth above. 48. Summary of Formulas A. Initial Data ty Oy Bry Kay Vay Zay tay ay 82, Xay Yay Zay fay tay Bay Xsy Yar Zay a ay Bay Kas Yor Zae In relation to theso quantities the remarks which were made in Section 44 romain valid. o B. Invariable Auzitiary Quantities 4; = 008 5; €08 aus by = cos 8, ain 44, a eq sin By (= 1,2,3,4), Control: +P e eta, i =e; Sin ay = cos (a; + 5), by +e; cos ay = sin (a, + 5)- DR, 08H ~~ 2X1 + Ys + eZ. 2Ry cos = - 2(aaXy + bi Yat C474), RP =XT4Y?24Z?, RRa x? 42423, am Control: (X, =a)? +, —5,)7 + Z, = 0)? = RF + BR, cos 9, +1, (Ke = ay)? + (Fy =)? + Z, ~ 04)? = RP + 20, cos , +1. 174 DE ecm ete Kemet Bam yey BaP + Batam = iF = Vat melee Crips ~ Cate + CaM Pe = Mas ~ Za t mys avy Aimer = agPy + Mere = MX, ~Xy 4m X, Dou Baka * Paming = MV =e emir WNiPLn Pay UM Pg =m Ze = Ey mye From these equations p, and py. are eliminated by numerical moans. From both groups are selected in twos the equations In wthich tho coofficients for p, and p, show the greatest change, Usually the fist two equations in onch group prove to be such, "As t rosult two equations are obtained of the form ” where the coefficients 4... ,D’ are found numerically. Tak lytA te = kets yn ky = hs Tekst) k= 0.0x720210. Control: T= T+ T= ET betiy, Bett, Fegut, retna, Ga ab, earn’, H=PF(A-@), WPA Oy wh T2424, Veatia, KeE(B+C)+C+D, Ko sR (BHO) + Oo DY, L=FQ@-C+D-K), L' = FB 0" + DT -K), MAACIB ST, 1,0), MO = ACTER + T1507). In the case of minor planots, with small intervals of time and moderate eccentricity, it is possible not to caleulate the quantities ATM, and Determinction of an Orbit by Four Observations 175 Control: G+ HAE +F)-FO, OW AE AP) PO, AM ~ BI = 4407, 7, AW ABV @ SACU T, LeFIBQ-£)-CQ+8), Le FB (L-8)- O'R EN. ‘These equations check rather superficially the calculations of group (VI). It is quite useful to chock the results somewhat dil ferently, for example, by having the quantities of (VI) calculated by a second computer. ©. Determination of Geocentric Distances 1 ees ss - : Vit) (re ray® ie rere ey PeG@+Héslén, Pea@+WEs En, aoe ae QaK UL E+ Men, (vit a= oop Paw Pot Om Pr te In tho ease of minor planots, with small intervals of time and with moderate eccentricity, it is not necessary to calculate 7: Berea ace 0 12 R23 +2R,p, cos 0, +p} 3) Uf thre is a prliminary orbit, ono can borow from it the values of tho radiun vectors or the momenta of observations and thom iad tho inal £ and no For totally unknown orbits of minot planota, {= 0.006 and y =O aro takon in the fist approximation, and then from (VIM) p, and p, are found, and from (IX), r, and r, + From oquations (Vil ¢ and fate obtained and th deseribd proc tov is roponed until th final and y aro founds For tho thd ap- froximation fin posible to resort tothe mathod of interpolation Minhas given in Soctign 44 (©) and to take Ag? fo 00+ SEE where £, {a the initial value af tho frst approximation and A, and Ag, aro tho differences of the initial and final values of ¢ in the fick ana socond approximations. Por lv is poselilo to procood in 1 sinter manors 16 The Determination of Or D. Derivation of Elements “ Control: ateyleat= aleylesdar. If the aborration time was not takon into account, Haty-dey aq - App 4 = 0F008772, =kGp- 89, etait rnin t 880s hee aay Jy is found in tables or is calculated by the formute «mp op «20 ~ Leva, 7 = 0.909001. aie t yy +22 Yo=¥a- 2%» cam rBacdey2+ a2. Control: (ryra)® = 124 ~ 24)? + 24 ~ 217g)? + ry = Es Determination of an Orbit by Four Observations Ww sin, - »,) c0s (v, = 94) = Control: The sine and cosine must pertain to the same angle. «vy 2080 = ay KV) asin v, = 108M = M1) = 4 ‘ 7 sin(o,-v,) ” 1 (Og %)> Control: ¢ 008 ¥, = 74. esindy a pha, 1 © tan t eyt 1 tan 5B = Y/ Posten boy = oot ts + Ea) andy, «vy 1 T=? ia, } Bee 1 tan b= Tifton by = oot (ts +43) un boy. Control: @ cos $ sin $ (Ey ~ Ey) = yirite sin $ (vy ~ 04) 7 = 206264.8¢, e° = ST.29578 6, M,=E,-e sinEy, My Ey -e sin Ey, Me- Me (xvity OR k= 0.985608. 178 The Determination of Orbits 2+ lo 0) = Ma lla ~ 2) cose sine sine) | cos o Py wey SEO ay oa my csv pen An aPs, Br= 008 $0x, Byn acca dQy, B,= 0008 60s. Control: ADS ADS AZo? B24 B24 MP =a? cos? g ABs + AB, 6 ABs = sin i sino =P, cos © ~P, sin &, sini cos a cos & ~ Qy sin ey 7 7 (XIX) sin =(P, cos w = Q, sin w) sec ey eae cos {l= Py cos a = Qs sine, cos i = ~(P, sin w + Q, cos a) cosee 2, Control: sin { and eos {2 must correspond to each other, and so must sin i and cos i. E. Representation of Observations Pat-Ap, where dp is found by interpolation betwoon the values Ap, and Ap. Ma Mo +n = ts Besin EM, p08 5 cos = Ay(cos Be) +By sin +X, ~ (XX) pcos 8 sina =A, (cos Fe) +B, sin E + Y, psin 5 = A,(cos Ee) +B, sin +2. Apropos of the second and thied observations used in the caleu- lation of the orbit, it is necessary to notice that complete ageeo- { { i Determination of an Orbit by Four Observations 179 ment may not rosull. If, for example, for the deduction of equations (V) the fiest two correlations in every group of equations (IV) wore stilized, then right ascensions must. be accurataly represented while in dectinations there may be divergencies of the order of errors of observation. If right ascensions are represented to the limit of the precision of the calculations, it is possible to conclude that there are no errors in the calculations, and the series for the ratios of the areas of the triangles have yielded sufficient accuracy. i, simultancously, accord is ackioved for the declinations, this indicates the reliability of the observations of the orbit itsolf, be- cause the declinations entered the calculations only indirectly, through cos 8. Under those conditions, the observations not used in the determination of the orbit (if there are no errors in them) usually are also satisfectorily represented. If the declinations aro represented badly, thon there are errors either in the observations or in the calculations. If, in addition, the right ascensions also ato represented unsatisfactorily, then this, apart from errors in the observations, may indicate that the bases of the method—expan- sions in sories—are insufficiontly precise for the given case. Then it is necossary to rosort to the methods of improvement of Initial orbits that aro given below. 49. Example For the planet 1934 TF there are the following observations which were made by G. N. Neuimin in Simeis: No. UT. cisae.o Dyess 3 1934 0c, 5 20h 15.6 OMMB™EBSO8 + nna’ 8.5 2 92244-1043 31.38 +5 298101 3 22 87 0.39 2098 48 as 87’ @) 2918 545028 800344 BOE Be Now TIT BESO 28 a5 Bok aA ‘The observations marked by asterisks were used for the calcu lations of dhe orbit; in what follows they received the indices 1, 2, 3, 4. Omitting the reducing calculations, we give their results: t a 5 x y z 5.84416 11°44’81".2 45°R9" 88 0.978311 —0.188890 —0.081959 8.94729 10 6250 {7 +5 2951 <1 —o.960178 —0.25i866 —0.100195 wiogg7: (9 5114 7 $5 1857 15 Locsaitae gorms7ae T9044 "4 50 14 “3 —o-80085S S8T4818 5 87 6 lo 4 4343 18 —ost0R0TE 180 The Determination of Orbits 1 2 3 4 com H0.0T9OTE +O.9R5RAT + 0.992224 +0,994610 feos 5 0.905336 0.095835 + 0.996438 1 0.908596, Sin @ $0,203504 + 0.571140 401124489 4-0. 108608, a Fo.aTa50s So.9aI1Ai +0.988690 0.99129 (7) 5 Tolaoasss Torito Locisioas tol roaaet 8 Loloveas7 0.001200 + ol0s439T +oLoK2440 cos (a +6) F0.954878 +0:965538 +0.978193 + 0.982675, ia (a +5) 0.297008 0.260250 0.207607 0.188876 2 +0,999489 +0.981688 (4) 2R cond Ti.a9007s Piersaaa mp, Pa np, a my +0.974508 ~0.981141 +0.901124= ~0.978811 + 0.981736 ~0.705078 0.202885 —0. 170437 +0.108841= ~ 0.188890 + 0.828602 ~0.688580 «jy “otasoats $0:1704aT ~oLi721Ta= + 0.169930 ~Oc161848 +0.122474 $0l0s3280 01068835= ~ 0.018050 401183087 ~0.516108 iP Ps Pe x my +0,974508 ~0,988690 + 0.991224~ ~ 0.978511 + 0.508855 ~0.705078 0202555 —0:124026 + 0.10834 1= —0: 188590 + 0.581040 — 0.658550 Tola22247 4 01124026 + 0.124344 +0.122724 ~ 0.101218 + 0.088448 0.080305 “o%021003= “oLoges66 + 0.120824 *0:850132 a m 1 + 0.483454 p, + 0.278812“ _ 9.951750 + 7.407400 ni ni ” py = +3.s2204s Lp, + 3.180812 “90, 460868 2 + 26.103022 ty 0409814 ta 0.805984 = 0.156170 = o.tsa102 goats 9.167948 vi oo2amaT By 2ieness & so.ataiat F 0.300266 Fr 40.116202 @ + 1.268657 @ +1430582 (VI) He ~0,242835 We so.aTs204 1 +0.834780 1, +0.388200 K 0.411996 Ko Toctao959 L + 8.267844 bo s5.s7512 Mw ~0.42aTTa w =as06548 € 0.006000 0.049081 0.0047551 0.047659 no Olooaets 0.002061. 0.002804 2 0 =0.0000169 ~0.0000142 — 0.000138, P +1.267200 + 1.287460 + 1.267490 + HaETAD (VII) Pr vilasaas + 1ag952 41431008 + 1.491904 @ ~o-ss2339 0.095050 0.306353 “o.aveai2 @ Tortie — 0.721019 721755 0.721858 py 1.982676 + 1.976820 + 18TOIIA + 1.970EA8 pe 2.082042 21108987 auniaise "21112526 (yr rh s.risos 5.586148 savas &.874TT4 7B s.snese2 5.751566 8.770898 8.772753. | i i Determination of an Orbit by Four Observations 11 ry 2.959085 2.975028 20TsT22 2.979086 re 2.982008, 2.988355 2io61486 2.961850 rer Sissiz0 5.05425 5.94021 5.94081 (IX) € 0.040081 0.047851 0.047708 0.004760 7 0.008446 —0.00201 ——o,002002° ~0.002806 Ag im0 Ag 18 Any—485 Any —59 1 4 LoToKes 2.139696 = ¢2.907200 + 2:790062 y — $0.s89897 4 0.856801 . z }og7a9%0 — todsti62 © A “eetarrso “airrassrs + a'97a0se1 29518803 4g 0.01013 = 0.1219 ©” senate '38.73300 {oq 82.9016 ye osesoTH aD =i 0.820828 8.17956 Denog. 174.0819 5.98124 dy 0.018408 ¢xTTy 4.04753, a, 1.002040 In the further course of the calculations there are no essential differences from the example in Section 45, and it suffices to give only the elements of the orbit and the reprosentation of the observed positions. Elomonts Epoch 1984 Oct. 10.0 U.T. My 23° S/1a".4 @ 484282 01 Q 10.8136 ‘sp 1984.0, i ospa1 ca @ 44820 ‘0 626074 Ag +LTSTE81 Be = 2.687486, 4, talaigon7 oy Falazeraa AP +¥agsa2 yt roasiae 2 3 4 a +i 0% 04 one fRB Lore WE Here are given the ropresentations hy the elements of the orbit of all the above-mentioned observations of the planot 1934 TF, with the exception of the first and last or anchor observations. vez The Determination of Orbits ‘The agreement of the observed and computed coordinates leaves nothing to bo desired and indicates that neither in the observations nor in the calculations aro there serious errors, Nevertheless, the elements of the orbit are based upon the small are described by the planet in a little more than a month, and it is not possible to consider them as particularly reliable. In this case, tho planct was near its node during the time of ob- servations, and it would havo beon difficult to calculate the orbit by three obsorvations on account of the smallness of the goocentric latitudes (not to mention the unfavorable distribution of observa- tions in time); the inclination, taken by itself, is not so small— about 10 degrees. Chapter 7 ‘THE DETERMINATION OF AN ORBIT BY TWO OBSERVATIONS AND THE PROBLEM OF IDENTIFYING MINOR PLANETS AND COMETS 50. Essence of the Problem On the basis of wo observations of a colostial body, it is im- possible Lo find all six of the elements of its orbit. If we have to caleulate an orbit from two observatiogs, we must first introduce some assumptions about the motion of the bovy, in place of the missing observational data. It is not possible to verify the cor- rectness of these assumptions until additional observations are ob- tained, Thus, every orbit calculated by two observations must necessarily deviate from roality and possibly even to a great de- bree. Nevertheless, it is sometimes entirely expedient to solve such a problem, and whoro furthor observations could not be obtained it is anyway impossible to do anything else. What, then, aro the as~ sumptions that must he made in place of the missing third ob- servation? Allowing, for example, that the orbit of a minor planet is a circle (lor comets, with very rare exceptions, this does not hold), we do not have to determine two of the elements, the eccentricity @ and the argumont of perihelion «, while it is possible to caleu- Tate with two observations the remaining elements, a, Sl, i, and uo, che argument of latitude of a planet for a selected opoch. ’ Such an orbit does not correspond precisely to the real motion. Tt usually characterizes rather well the position of the plane of the orbit which may be of possible use afterwards in the identification of the same planet if it is observed again in some later opposition. On the other hand, for a newly discovered planet it is not possible to caleulate immediately an elliptical orbit. If there are three ob- servations separated by too small intervals of time, then the orbit ‘obtained from these observations will be very inaccurate. On the other hand, a circular orbit is quickly calculated and can serve to 183 184 The Determination of Orbits obtain the ephemeris of the planet for the following weeks. In ad- dition, a circular orhit gives an approximate idea of the distance of the planet from tho earth, Thus it permits one to correct, with satisfactory precision, the observations for aberration time and for parallax in the subsequent calculation of the elliptical orbit, It is not always possible to satisfy two observations of a minor planet by a circular orbit; in some cases, it does not exist. On the other hand, it is possible, starting from two observations, to obtain an elliptic orbit of a minor planet; but two elements in it will remain indeterminate, and for thom it will be necessary to assign arbitrary values, which, by the way, may he used to simplily the calculations, Such an elliptic orbit can serve about the same purposes as a circular orbit. 51. The Determination of o Circular Orbit Observations made at the moments ¢, and ¢2 give two directions to the celestial boly, and if a circular orbit passes through them, the radius of this orbit determines the values of the geocentric distances p, and p, for the moments of both observations. But the radius of the orbit must be such that the arc traversed by the planet in the orbit corresponds to Kepler's laws. Observance of this con- dition can be obtained by moans of tests if we vary the radius of the orbit, a. Knowing a, it is not difficult to find p, and p» and the heliocentric coordinates of the planot. Aftor at the other elements of the orbit are determined. Having for the moments ¢, and t2 the observed coordinates os, 5, and a, 5a, we suppose, as usual, a; = 008 Bi 608 Hy Bim cos Sisin ash (= 1,95 (rt) in 8: furthermore we have ore aipe~ Key “Hi, (ay hie If the cadius of the orbit is equal to a, from equations (7-2) or hy using Fig. 12 applied to both positions of the planet, we find at = R26 2p, 1, e088, + p2, (78) Determination of an Orbit by Two Observations 185 where hbaxP eee ze, Rpe0s Oe» —(arks + BY; + 612). Finding», from (7-8) we obtain y= Val -B? sin? 0, - R, cos 9. (7-8) By analogy with an ellipse we denote the heliocentric are be- vween both positions of the planet by vz ~ v1 =2f. Moreover, for fa circle, the measuring of » may begin at an arbitrary point of the orbit, We shall have the obvious relationship sy sin? fa Ey tes 20+ (yp wa) + (an e0 Ph (78) following from the fact that 2a sin f is the length of the chord be- teen the heliocentric positions of the planet. Tn addition to the above equation, Kepler's third law must be observed, which gives for uniform circular motion Pf = wlte ~ tr) =a (ts = ta), wn The problem is solved by the selection of the quantity a, satis- fying equations (7-2), (7-8), (7-8), and (7-7) which contain nine un- knowns besides a, namely, six coordinates of the body, its two geocentric distances, and the are f. Having found 2, we obtain waka = 2f/(ty — 1) (8) ‘and the equatorial heliocentric coordinates from equations (7-2), if they were not obtained befoco. As to the determination of the vector elements, it is convenient hore to take advantage of the fact that » may be measured from an arbitrary point, and to place it at a point of the orbit located mid- way botwoen the positions of the body at the moments ¢; and fz. Thus, 0 =—f, v2 = fe From equations (3-18) for the given ease we have tre Pe acos/—Qrasinf, 22-2, acos {+ Qe asin, and so forth; consequently, aPl=4(a,+ag)secf, a Qe 42-2) ese fy aP, dy + yo) sec fy @Q,=Eye-¥yese f,> (7-9) @P, 32, + 22) 800 fy @Q, = a(n ~ 24) 86 fy 186 The Determination of Orbits and, in general, ae aP, cos» +aQy inv rineactagaeo} (7-10) =aP, cos v+aQ, sine, whore va nlt-to to= dtr sta) (ru) 52. A Summary of Formulos for the Computation of @ Circular Orbit and an Example A. Initial data fy ay By Ky My 2a fy an 82 Xa, Yo Za, W) a, 81, Xu, Y4, Zi ato eoforred to the samme equinox. The calevla- tion is cattied out with five or even with four dociwnal placoss thre: fore, it is possible to neglect the influence of aberration and rallax. B. Constants 4; = C08 8; COs Hy Bem cos bisin ay b (f= 1,2). a cpa sin 5 Control: abe bP acta. Bra e:cos y= ain (01+ 8), mais Ye4 0120) ay Ea ND YP Zh (sin 997 = #2 ~ (Ry cos 0/99. Ry cos C. Finding the radius of the orbit ope Va RT STD, ~ B, 008 0, = 0,0;-Xy eed avy) 2 ei Ba Determination of en Orbit by Two Observations 187 1 5 ; ain) oer net etare (ote yaar (etesen at eo gis Ue 2 P+, -n) © fqn 4h(a tr) 7%. The quantity a is varied until the geometrical value of the arc, designated by f,, becomes equal to its dynamical valuo fy. In the first approximation, it is possible to take a=2.7, a? = 5.29, and ava =44368. For the second approximation it is possible, de- pending on the value of f, ~ f,, to change @ in one way or another by an amount of the order of 0.2 to 0.8, and further to have recourse to interpolation. D. Derivation of the elements =kam 21H =), } om) 0.98561,” = 3548.19, 2,2 a40)- Xp Ye ayoe~ Vy (VIE) 21-00, ~ Ze at eyte steals yds edna = 4 (a, +22) sec f, By = a0, = 4(2_ ~ #4) ce f, Hy + ya) see f, B, = a0, = Flys ~s)ese fp (VIN) Fe, + 2,) see f, B, = aQ, = 4 (2, - 2,) ese f. Control: Al 4 424 Ab B24 B24 BB =a, A,B, + A,B, +4, B, = 0. sin isin =P, cos ¢- Py sine, sin icos w =@, cos €-Qy sine, sin {l= (P, cos w - Qy sin w) sec £, x 08 R= P, cos w ~ Qx sin oy “ cos i =-(P, sin m+ Qx cos «) csc 2, E, Representation of observations to Matty v= all=toh cos 6 cos = A, cos v4 Bs sine +X, . a B. x, «® pcos 5 sin d= A, cos v4 By sino + Y, psin 5 =A, cos v4, sings Z. 188 The Determination of Orbits Generally speaking, the representation of the observations by circular orbit will he tho loss satisfactory, the farthor these ob- servations are from the basic observations ‘and the greater the eo- contricity of the actual orbit. Therefore, it is possible to limit oneself to the representation of the two basic observations. The dovintions must be within the limits of the precision of calculations. As an example of the calculation of a circular orbit we shall again take the planet 1934 TF and make use of the first two ob- servations listed in Section 49, namely, ur. ans Bisa T1934 Oct. 5 20415".8 Got, 5.844 O86" 587.08 45°89'8"".8 Baad 1 OOAT OAS 31 3B +5 B4SIC1 in 1 2 Xx -o.a7ss 0.0602 y o.rssy 0.2515, 2 =o.0si9 “0.1091 cou 40.9701 + 0.9820, cos 5 +0,9083 +0.9956 sini $0235 40.1887 2 $0075 +0.97TT % so208 — +0.1879 2 0.0968 +0.0081 cos +5) 0.9548 +0.9809, sin (% 43) 40.2970 40-2803, m rk? oan95 0.9971 Rcog d — +0.8098 —+0,0083 RX eint 8 40,0005 +0.0045 v = 2.7000 3.0000 3.1420 et T2900 91000098729 17001 2.0008 2.t428 Tiros = 30m8 2480 218353 42.077 45.0882 folsis2 — $0.5060 +0.8235 oust F027IT + 0.2884 falesno 2.0184 +8.057% 4O.sTI5 40.0783 0.6545 Tous — Foz7e — F0%3109 v 0.00007300,0000487 9.0000402 0.00854 0.00098" 0.00884 peas 24".00 2150 ers 0.1924 0,196 28.86 2H 50) +199 40.64 ° vI ears Determination of on Orbit by Two Observations 189 vi faces 0.5000 eosee/ 18.84 ap) esopis 0.6008 i) Fosssn ay S382 12 Yozosn at +18 PL Foot OE 2208 PL 02082 ay o.7954 PL Foe Oe 40.8646 IX sinésin@ 40.0007 —siné— + 0.2016 Sinfoos@ 402015 cost 40.9773 Sin 40.0383 w= up 12547.3 cos +0,0095 shane? 5cr sinfl 40.1925 §11e89%0 con ¢0.gRIE Elements Epoch 1934 Oct. 7.896 UT. to 1884 Tn oF 1994.0 ¢ 1s] neat 1 2 5 ‘ iste ogi, 35,748, » 02s 407218 +8278 cos @ +1.0000 © $1.0000 0.9985 sine “0.006 $0.0068 0.0983 peon8cona —+2.0879 + 2.0972 42.2769 pecs Ssin% — +0.4387 $o.s031 +0.2851 paind §—— +0.2008, + 02019 + 0.1901 poosS 42.1824 +2:1356 +2280 & 1154.0 10°82".7 5°5K'.0 Bes 313 +50 HH aT oc {at 40's Ont an AB $0.8 fo Xo Besides the two observations used in the determination of the orbit, the last of the observations listed in Section 49, made about a month Inter, was compared with the elements. While the first wo observations agree with the cloments within the limits of pre- cision of a calculation made with four significant figures, the latter observation deviates from them. However, the deviation is So insignificant that it would by no means interfere with a scarch for the planet, even by visual means. This shows the usefulness of a circular orbit. 53. Tisserand’s Criterion for the Possibility of e Circular Orbit A real value a, for which the equality f, = f, is observed, does not always exist, and thus not every two observations of a planet 190 The Determination of Orbits can be satisfied by a cireular orbit. A complete analysis of this question would be rather complex, but it is hardly required, be- cause the problem of the determination of a circular orbit is not of particular importance, Tisserand's criterion of the possibility of a circular orbit is Aerived from the simplified assumptions that the earth movos along circle of radius 1, and that the planet at the moment of the first observation is exactly in opposition with the sun and, therefore, also at the node of its orbit. For the derivation of Tisserand’s rule we shall take the equations (8-9) and conduct the reasoning in ecliptic coordinates, since here this will he more natural. At the moment of the first observation, according to what has boon said, A Qy B=0, w20, rae andk Also, it is possible to write A= L = 180° +, but this changes nothing in the results. Tho first equation of (8-8) gives prant, which, incidentally, is obvious. We shall take ¢s —¢; as & small quantity and during this interval we shall assume that the coordi- nates of the planet vary linearly with time. In that case we shall find the condition of the possibility of a circular orbit by differ- entiating the equations (8-9) and hy replacing next the coordinates of the planet by the values given above. This gives tp Fo, a du_ db a= 1) Beg eog 4 MH. th, - (@-1) Fa acon i BT, (7-12) Bat (7-18) Hence, wa have Row a =(a-1) Baa, a i ie (7-14) o-1) #8 dt” Determination of on Orbit by Two Observations 191 or, eliminating #, 2-2) (2 ] ate Bsn The root a= 1 corresponds to the orbit of the earth; therefore, we can divide by @~ 1, which gives 2 f(A) (28)"] afar (2)? (48)7] 20. ras “(ay -Gy)a-G)-@ oy In order tat bo con tis noconsary thatthe dscrininent of (2-2) -(@)T ea) «(ay poo At opposition the motion of a minor planot in a circular orbit is always retrograde. This is also evident from the equations (7-14) because with a> 1 . ever eet BK (a) co can Inthe motion along a rea rl act min nar opposition can aeur for soma plasty forex fo (8) Bro, wo is sour poeion Ts To expaned bythe fee tte elo of shed nan lpcel i ett at prilion thn nn ces ir thi pasing ngugh the seas pot in syeny ae my bo soon to he eyaions 38) anim nt conde ett {ean eis inposilefor cl tion of Uw gleet net op pasion. After having expanded th noni (18) 0 sal tn (2) 9G) 2] da)* | (aa), -10(8) (2) 20. cram (1B)? on? 92 B_ (BY) + a4 er + 2% o ww The Determination of Orbits 4B ox 04 _(O) op ferra B®, @ (BY see wh (BY oon fire Bao) (2) oe Yee, ‘The first caso is very improbable, heeause for such a large mo- tion in latitude the planet must be very close to the earth. Thus, in practice, it is sufficient to apply the inequality (7-20). How: over, if ~dd/dt = k/2, both values of the root coincide, and if = dv/idt > k/2, a real root does not exist; thus for ~dh/dt > #/2, a circular orbit is possible for any dp /de. Wo have seen that our investigation is not rigorous, and we cannot, therefore, fully rely on the derived inequalities. However, in case of doubt, the solution of the fundamental equations of a circular orbit will always indicate whether such an orbit exists that satisfios the available observations. As shown by experience, it often is found to be impossible, (7-20) 54. Determination of Elliptic Orbits With but two observations, it is possible to obtain an elliptic orbit only if two of the parameters determining the solution to the problem have been given arhitrarily. Such an orbit will have no ad~ Vantage over a citcular orbit in the sense of indicating the true trajectory of the body. It is worth while, therefore, to ealeulate fn elliptic orbit either if it is impossible to find a circular orhit. ot if the calculation of the elliptic orbit, instead of being more diffi- cult, is easier than tho calculation of & circular orbit, As the arhitrarily given parameters, it is possible, for example, to take both of the geocentric distances of the planet. ‘This im= mediately gives the heliocentric coordinates for the moments of the observations. After that, tho elements of the orbit can be derived accoriling to well-known rules. Instead of p, and p, IL 1s possible to give beforehand the value of the eccentricity ant the position of perihelion and, by using those assumptions, to derive the remaining elements of the orbit. It is probably simplest to solve the problem hy the method pro posed by Viisilla.' He assumes that the planet, at ono of the moments of observation (the second of them), is at perihelion. This seems entirely plausible, because in our time mostly faint 'A mesentation of this methost is given by A. Ys Orlov and B.A. Ontoy in their texthook [101 Determination of an Orbit by Two Observations 193 planets are being discovered, and they are more accessible to ob- servation just whon they are near the perihelion of their orbits. As for the second arbitrary condition, the geocentric distance of the planet at the moment of the second observation is taken more or Tess at random and usually tho value taken is used, without any changes, for the calculation of the elements of the orbi Thus, preserving our previous notations, we shall suppose Ay <2 C08 8g. Having taken a cortain value of Ay, we shall find: A, cos a, -X3) Yq = Ay Sin Og ~ Yay fea We use equations (5-25) for finding, by means of 2, yg, 2g, the coordinates of the planet at the moment of the first observation. Wo can weite nse 0,(2), noFisneos (2), om) Fie, +4(F). The quantities F and @ are determined by equations (8-21), in which we can omit the terms with dr/dt, heeause at the moment of the second observation the body is at perihelion, where dr/dt = 0. Hence, we obtain with an accuracy up to T* inclusively, (7-23) where T= (tr ~ 2); the geocentric coordinates of the planet, in theie tuen, will be é = 8,005 0, = Fray +04 ( n (raty 194 The Determination of Orbits where \, =p, cos 8,. Hence, after multiplication of the equations by 25, yg, and 2, respectively, and by addition, we derive Fyre Nays Vive ty 20 8" eee a, + yy 8 2, land, (7-95) eee (#), Sm ; Rquation (2-28) makes it possible to calculate the semimajor axis a. Noting that dt = £ df, and neglecting, of course, the mass re ele tan and (728) For the determination of the eccentricity it is possible to make uso of the circumstance that at perihelion rs = a(t ~ e); conse- quently, (7-29) The time of perihelion passage is T= ¢z, according to our basic assumption. It is not necessary to calculate the vector ele- ments, because the ephemeris for the immediate future ean be cal culated even without thom, according to the formulas Determination of an Orbit by Two Observations 195 acre a(t pom dcosa oF, x0( 22) (2) pewodsinaaPee(Z) er} am whore the exprossions for F and @ are given in (7-23). However, it is not difficult to find the vector elements. Be cause v= 0, dr/d< =0 at the moment of the second observation, it follows from equations (8-18) that Platyityy, Pla 8) ay (rsty It is even possible to obtain the last-thtee equations from the consideration that if at perihelion the velocity is perpendicular to tho radius vector, thon it must coincide in direction with the vec- tor Q. The elliptic elements o, fi, i are found as usual, Moreover, the node and the inclination, as also in the calculation of the cir- cular orbit, must not differ greatly from the true values. parallax and aberration, it is possible to conduct the calculation with five or four decimals. ay B. Finding the coordinates and the velocities The quantity A, is taken arbitrarily (usually from 1.0 to 2.0, depending on the large or small apparent daily motion of the planet) and with it are caleulated a 196 The Determination of Orbits 1 1 sap Bapa Takin), Fy-t-At% G,= amy Fyrbe Nyty Vata + Zit F008 & + yy sin wy v5, tnd, — Bre, Em (A, cos oy ~ Frag -X,)20,) vem (A, sin X,~Fryy— Vy) Gy, av) 25m (Ay tam 3, Fag - 24): Conteot Peeh + Ya Ys + 2224 = 0. C. Derivation of the elements Viewty a 28, Lei. W) Control enrVPa1 If a is slightly largor than ra, for example, 0S a ~r2 $0.5, the orbit is suitable, Otherwise, it is nocessary to change the value of A, accordingly and Lo repeat the caleulations. ay Py = Vaitar Pom 2a a vn 1 Qyaugl Qn 2h:Ve o Control: P24 P24 PPA O24 Q2 4 QP a1. sin éain w= Ps cov €= Py sin 6 sin {008 w = Q2 cos = Qy sin & sin l= (Py cos @ ~ y sin a) see &, wm c08 N= Py cos @ = Qe sina, cos i = (Py sin © + Qx 608 a) ese 8. Control: sin and cos must pertain to the same angle, and Similarly sin i and eos i. Determination of an Orbit by Two Observations 197 D. Cateulation of the ephemeris Takt=ta), k= 0.017202, Fel-At?, Gat~Br, pcos 8 cost =F 2, 4G25 4X, peos 8 sin X=Fy+@yseY, psin B= Fay4@25+Z, ‘The ephemeris can extend for 1 to 1% months and even more from the basic observations. In conclusion, let us calculate, as an example, the orbit of the planet 1934 TF according to the same observations as in Section 52. vt. Biss0 B19. T1934 Oot. § 20415".6 Oct. 5.844 O848"58.08 + 5°2"8.5 Bap 4d to Sloat 043 31 98 +8 an sr 1 1 2 = -0.0188 ~0.9802 y — Zoltss9 Zo358s Telesis Zolt091 A 1697 1 con +0:0701 + 0.9820 Sina toaoas so.t8e ten 80.0980 tolnats 72 42.6286 a “Tats % t0Ta8 Ho ao0e fogesr onsen & loost2 1 =, -0.7058 F, 40.0099 wv +0,0405, G, — =0,07058 ” 0035 Numer. Ais t.stt6 Donom A) S18 wv 0,24 +0.01060 2-015 yy, -0.08796 7 + 0.5378 @i2h Soman a2 Soaso3 v outain 2 “oso ooniee % O00 1497 Sine ar lies within admissible limits, it is possible to dis- continue further calculation and to weite the formulas for the cal- culation of the ephemeris F = 0.017202 (+ - Oct. 9.947), F=1-0,002527 17, @=t - 0.00842 T°, 198 The Determination of Orbits pcos 5 cos = 42.6296 F 0.1515 7 +X, pcos 5 sin A= 40.5728 F 4 0.8978 G+ Y, sin = 40.2607 F 40.9903 G + According to those formulas, we calculate a representation of the same observations as in Section 52. 1 2 5 ‘ 5.5i4 dott 3.146 + ~0.07088 ° + 0.40510 a +0.00195, ° 4024542 a 0.00058, 0 0.12158 F ¥0.9909 1.0000 +0,0938 o Zo.oT08 0 4 04bat peor Soot 41.867 41.6698 L8H peo Sain td 40454 08205 — +0,1960 sind 0.1615 ¢0.1608 + 0.1543 peos8 41.8073 + 1.7000 L898. & 1844.7 1082.7 6a.0 5 +5300 45.2400 +4 447 Aa ~02 +04 5.9 ono fh BIE Although the result for the last. observation is somewhat less satisfactory chan that given by the circular orbit, the deviation is, nevertheless, sufficiently small, ‘The calculation of the clomonts, carried out necording to equa tions (V), (VI, and (VID, gave the following values: Bloments T 1984 Oct. 9.07 UL. @ scar St soeaartt 1984.0 Test, pn 6ae”9 ® o.1no P. 40.9722 Q,~ 0.9834 Py 0.2118 Qy + 0.8286 Py 40.0097 Q, + 0.5089. Comparing this system of elements and the circular orbit eal culated in Section 5 with the elements found in Section 49 by four observations, we see that the elliptic orhit just obtained and the circular orbit deviate only slightly and in approximately the same Determination of an Orbit by Two Observations 199 amount from the elements in Section 49. At any rate, such char- acteristic elements as q, Ml, and ¢ are obtained with’a precision completoly sufficiont for identification of the planct 1934 TF with some later discovered object, if a question of identity were to Comparing both methods, the calculation of the circular orbit ‘and the method of Vista, ‘one ought to acknowledge the advan- tage of the latter, if the saving of the computer's time is being considered, if the elements as such aro not required, and merely fan epheneris of the planet is needed. Otherwise, the expenditure of work is almost the same, and the elements obtained by both methods will probably bo approximately equal in their reliability. 56. Criteria of Observations Belonging to an Orbit For answering the question, whether a given observation of a minor planet or of a comet refers to a body whose orbit is already known from previous appearances, one*may establish a criterion based on the requirement that the observed direction toward the body must intersect the plane of the orbit near the orbit itself. More is not required, since the calculated position of the hody ob- served carlier in its orbit may be very unreliable for a given mo- ment, particularly if the orbit refers to @ long-past epoch. Thus, in order to obtain the required criterion we shall use the equations @=PLE+ Qc n =p 008 6 cos 4 —X, ¥=Py£+Qyn=p cos bin a-¥, (7-32) 22 PL E+ Qenaosin 5-2, in which the vector elements, cooedinates of the sun, and the ob- served coordinates of the celestial body are known (and must be referred to the same equinox). AS we know, the quantities € and 7 ace the eectangular heliocentric coordinates of the body relative to axes in the plane of the orbit, whore the € axis points in the direc tion of perihelion. Tt is clea that it is possible to find from equations (7-32) all three of the unknowns, £ 1, and p, by using determinants, or by simple aumerieal calculation. Howover, for our purposes it is sufficient to obtain only ¢=1 cos =a (cos E =e) and =7 sin = 2 03g sin B, alter which it follows that 4 woos 6 200 The Determination of Orbits If cos F and sin & belong to the same angle, the body is situ- ated on the orbit and, solving further Kepler's equation, it is pos- sible to obtain the mean anomaly of the ody and the time of peri- helion passage. OF course, one should not demand too much pre~ cision in the use of this criterion, because the elements of the orbit change with time as a consequence of perturbations. Tn prac- tice, a certain amount of caution is required, ZK So Figs M4 It is more convenient to apply a similar criterion to ecliptic co- ordinates. For this purpose it is necessary first to transform the observed a and 5 into A and B. On the heliocentric celestial sphere (Fig. 14) the observed geocentric position of the celestial body P’ is indicated according to the coordinates A and , and the position of the earth by T, determined longitude L.” In Fig. 15 the sun S, the earth 7’, and the celestial body P are represented. Figs 15 The directions ST and SP” in Fig. 15 correspond to the points T and P” in Fig. 14. The heliocentric direction SP to the celestial Determination of an Orbit by Two Observations 201 boty ties in the plane of the orbit and evidently in the plane TSP” also, i.e., at the point of intersection P of the great circle TP* andthe projection of the oxhit (Fig. 14). Tho are PP’ is equal to the angle J, and we have sin 9sin 1 =sin 8, sin 9 cos [ «cos i sin (AL), (7-84) 08 9 = cos f cos (k= L). We designate the angle at the planct P by 2 (Fig. 15); we notice that the heliocentric are TP = ZTSP = @~ 2. The triangle SLPP is solved by using Napior's analogies: 1 1 si tan 5 us (0-2) wtan 5 (L-) sU4i 1 1 (7-38) tan 5 [u~(0~ 2)] # tan 5 (l= 2) Having found wand 6 ~ 2, we obtain 2, and ftom Fig. 15 we see that pe Rsing eae (7-36) But, on the other hand, p Pp “Tye conv” Tye cos (way a Since p, e, and w are known from the elements, both values of r, given in equations (7-36) and (7-87), can he calculated and must, coincide. We recall, however, the reservation made above con- cerning the application of this criterion in practice. I it is assumed that tho observation of the comet refers to ® certain previously observed comet, for which there exists only a parabolic orbit, and if it is necessary to verify the identification, fone must sot e'= 1 in equation (7-87). That gives, re qsec?d(u-o), (7-88) and equation (7-36), of course, still applies. However, in this ease it is necessary to make the identification with still greater caution. Chapter 8 THE DETERMINATION OF A PARABOLIC ORBIT 57. General Remarks About the Computation of Comet Orbits The orbits of comets can have the most diverse form. ‘The ma- jority of comets move wlong parabolas, but many comets return periodically to the sun, describing more or less elongated ellipses. In certain very rare cases comet orbits tumed out to be nearly cireutar The rapid motion of the comets near perihelion, where they aro most often discovered, and their changes in brightness and in form can make the work of observers difficult, if an ephemeris ia not at hand. Consequently, the calculations of preliminary comet orbits and ephemerides usually must be carried out without del A parabola always serves as a good first approximation to comet orbits. Frequently, it is possible to caleulate a completely satisfactory parabolic orbit from observations separated by a few Uays or even only by 24-hour intervals. It must not be forgotten that usually the speed of the calculations plays a greater role than tho achievement of the greatest possible accuracy. It must be recognized that the mothod of Olbers leaves little to be desired in the overwhelming majority of eases, It is at least as convenient for calculation with logarithms as for work with an arithmo@eter. The course of reasoning for the determination of a parabolic orbit is briefly as follows. A parabolic orbit is determined by only five elements, because the eccontricity is equal to unity. This restriction is reflected in the existence of a relationship between two radius vector of the comet and their closing chord, om 01 side, and the interval of time during which the comet moves from the first to the second position, on the other side. This relation- ship is expressed in the so-called Euler's theorem, If we have three observations of a comet, this gives us an equation connect: ing the first neocentrie distance with the third, Olbers deduced a second equation between these quantities from the equations of the plane, eliminating the middle geocentric distance. Finding , and p, from tho indicated basic relations is 202 Determination of @ Parabolic Orbit 203 not done directly, but by reponted trials. As a result the helio- conttic coordinates of the comet hecome known, from which it is possible to determine the elements of its orbit. We see that not all the observed coordinates enter into the solution of the problem, because from the six data of the observations it is necessary to obtain only five elements of the orbit. Later on we shall show chat the two coordinates for the middle position of the comet give ‘only one condition and not two. Moreover, it should be noted that the method of solution differs in essence from the reasoning fol- owed by us for the ellipse. While for the elliptic orbit the values of two basic parameters (n; and ns) were sought by means of sue~ cossive approximations, for the parabolic orbit everything is re- duced to one parameter, the ratio of the extreme geocentric dis- tances. 58. The Equation of Olbers Wo shall uso tho samo basic notations as for tho elliptic orbit, and also the equatorial systom of coordinates, as it was used thero. Let there be given, for the three moments of observation f(é= 1, 2, 8), tho observed coordinates of the comet, ay and 51, and the coordinates of the sun, Xi, Yi, Zu Equations (5-10), (5-11), and (6-12) aro rowritton without change: 4, = 608 3; 008 4%, by = cos Sin ae, (8-1) a= a1; ~Xp n= bp. ~ Von 2) a= 6401 Ban Qr mpi ~ ay pg + 4yMgpg = MX, ~ Xa t My Xay Bimer— Papas basen =Mi¥i~ Fat maYar (88) ey My Py ~ Cha + Ca MgPy = MZ, ~ Za t Mg Zae Wo must eliminate p, from these equations and obtain two equa- tions connocting g, and p- Cloarly, this elimination can be e- complished by various moans. For examplo, itis possible to mul- tiply oquations (8-3) by three factors, A, 8, C, respectively, for tho present arbitrary, but satisfying the condition Aaz + Bb2 + Coa = 0, (8-4) and then to add the equations. 204 The Determinction of Orbits Lot us introduce the notations Aa; + Bb + Cor = Se (E0149), AN; + BY + 021 =; (€=1,2,8), “ and we shall have in place of (8-3) 111 + 8yyng = 01m, = Oy 4 Og nye (66) Moreover, we can afterwards subject A, 2, @ to still another con- ition besides (8-4). OF course, in addition, we can multiply these three quantities hy any factor, but this will not in the least affect equation (8:6). From this equation we can find p, in the form: a= Mey em en where (8-8) To begin with, we must introduce the ratios of the aroas of the triangles into these equations, At first, of course, we must be satisfied with thoir approximate expressions. Most convoniont for four purposes are the equations (6-12), in which it is only neces sary (0 replace the subserint ¢ by 8. They give, with the aid of (ot), $2) pg on (8-9) where Takata) Tes lst), Te kG ty). (810) ‘The last terms in (8-9) are usually insignificant (if the radius vectors are not very small). In any case, in the first approximation they must he discarded, because of lack of knowledge of r5 ~ r3. ‘The accuracy of the determination of the orbit essentially de- ponds upon the accuracy with which we shall know the values of Determination of a Porabolic Orbit 205 Wand m. But the accuracy with which M and m are found depends in turn upon the accuracy of the ratios, 44/85, 1/4, my/ngy and 1/ny, We dwell first on «,/4, and 1/é.' We can consider a, B, Ca components of a certain vector which, according to (8-4), is, perpendicular to the second observed direction toward tho comet, nd is, therefore, directed to the pole of one of the great circles fon the celestial’ sphere passing through the second geocentric, position of the comet. ‘The selection of this citele plays an im- portant role in the theory of the determination of a parabolic orbit, find we shall call it the fundamontal eirele. In Fig. 18, Ky, Ka, Ks ate tho observed positions of the comet, P iis the pole of the fundamental circle, and Ly, La, Ly are the heliocentric positions of the earth dotermining its longitude (the antipodes of the geocentric positions of the sun). If we join Ky and K's with P by ares of great circles, then the valuos of 3 and Ty will be proportional to the cosines of the ares PK and Pits, or to the sines of the spherical perpondiculges KrQ1 and Ks Qs drawn from tho first and third positions of the comet on the fundamental circle, Similarly, ®,, ©, ©, will bo proportional to the sines of the spherical peependiculars to the fundamental citcle LiMi, LeMay and La May multiplied, respectively, by Ri, Ra, Ry where Bw VXPO VES Clearly, since we have at our disposal an additional condi~ tion for A, B, C, wo can al- ways draw tho “fundamental circle in the most. advanta- goous way for the determina- m, tion of and m, and this will Figs 16 occur when J and, ac- tyuire Che maxierumn values Tn such a case, as is ensy to see, the point P lies on the great cireles passing through Ky and Ks and the fundamental citclo is porpendicular to these circles. It is evident also that it is im- possible to lay out the fundamental circle so that it passes near points K and fs, because then M will contain the hard-to-determine factor 4/4), and m will contain tho greater factor 1/<)y As for the quantities m, and ng, they ean be found by the proc- fess of successive approximations with arbitrary accuracy, as in P. 206 The Determination of Orbits the computation of an elliptic orbit. I is true that this is some- what more difficult here. It is vory desirable to simplify this process and as far as possible to introduce at once sufficiently accurate values of n; and ng. For the preliminary calculation of a comot orbit the intervals of time between observations are usually small; therefore, in the first approximation we ean make use of the smallness of T, and ty. The corresponding displacements of the ‘comet in the sky will usually also be small, as will the ares K's Q; and Ks Qs (if the comet is not very close to the earth; a close ap- proach can lead to rapid apparent motion of the comet). By neglecting the last terms in (8-9), we obtain the value of nix/ng with accuracy to the third order of smallness, if we replace it by 2%, provided that t, #1. If, howover, T, is only close to T,, the error will be of the second order, but in any case it will not be large. In the denominator of the expression for m is a small quantity of the first order, 4’, and the error in m is determined mainly by the error in the factor 1/ng standing in beackets. If we assume 1/ny = t,/Ty, the error in m will be of the first order, and it does not vanish for any ratio of intervals of time. However, by a suitable selection of the fundamental circle we can entirely eliminate the quantity m from the equation for py. Namely, let us introduce, as a supplement to (8-4), the condition ©, = AX, + BY, + CZ, =0, (8-1) for which the fundamental circle passes through the second geo- centric position of the sun (or, what is the same thing, through the second heliocentric position of the oarth). Then the equation for a will be (e-12) If we turn to equation (§-7), we notice that it holds for every body moving in a plane passing through the center of the sun, and in particular for the earth, or strictly spoaking for the eontor of ine crtia of the earth and moon. And so, we shall set p, =p, = 0, and we obtain from (8-7) and (8-12) 26, where Ny and Na are the ratios of the areas of the triangles ba- twoon the radius vectors of the earth at the moments of observa - Determination of « Porabolic Orbit 207 tion. Subtracting this relation from (8-12) we obtain If we apply equations (8-9) to the motion of the earth wo will have ns to ty a ole; Tae ES al ee o, ‘This exprossion for m is particularly convenient. Now ®,, as can be seen in Fig. 17, is a small quantity of the same order as 4, and, consequently, m itself is a small quantity of the second order, which moroover vanishes for equal intervals of time. Thero- fore, if we ass ume (e414) omitting quantities of tho second order in the first approximation, then by the samo reasoning we oan completely omit m as well and write instead of (8-7) Ps= Mp, (618) This choice of the fanda- mental citele which passes through the second positions of the comet and of the sun was made by Olbers, In view of this, the fundamental circle is often called Olbers’ circle. The simplification achieved by this choice is very advan- tageous, and only if tho fun- damental circle passes close to the points Ky.and Ky is it nocessary, ns wo havo. soon thove, 10 give up the simpli- fication. For such an excep- Figs 7 tional cage spectal troatment is required, We add here that from the fact that 4) and 4/, aro proportional to the sines of the spherical perpendiculars K,Q, and Ka Qa, it 208 The Determinotion of Orbits also follows that sin y sin Ks Q Sin ysin Ky? and therefore, (S16) Moroover, it should be noted that the second observation enters the calculation of the orbit only as far as it is necessary tosatisty equation (8-4). Consequently, the only condition made use of is that the second position of the comet must lie on the fundamental circle; im other words, it must be 90° from its pole. It romains to find the coefficients A, B, C in explicit form, and to express through them /and ©, and also M and m. From equa: tions (8-4) and (8-11) are obtained the ratios of the quantities 4, Band @: baa 4 CO" aYe BR caNXp—aa Ze Because an arbitrary not concern us, it is pos ‘ommon factor in these quantities docs Ie to write simply Aw ba Za—e2¥ay Be eqNXa-aZa, (an C= 0, ¥2-b2Xa, and then ee ie Tg dag + by Ces" ae or, a5 a first approximation, on the basis of (8-14) tua ta day 4 BBs 5 Cos -- 258 a (19) 42-0)" day + Bhs + Coa” Determination af o Parabolic Orbit 209 Similaely, we shall have for m ne iets tha (a Any + Bba + Ces \ns for, a8 a first approximation, m ~ 0. 59. Euler's Equation If we have two positions of the celestial body, for the moments and £’, givon, for example, by two radius vectors r and 7” and by the angle 2/ between them (or, what ia equivalent, by the chord s of the orbit), then, as we have previously soon, theso data are suf- ficient for the determination of the elements of the orbit, not con nected with its position in spaco. Howover, if ono of these elements is given, for example, tho eccentricity’ or the major axis, then not all of the quantities 7,1”, 8, and "= ¢ can have arbitrary values, but there must exist a re= lationship between them by moans of wfich it is possible to ox- pross one of the quantities in terms of the remaining. Such a relationship is Lambert's equation giving ¢~¢ if the threo sides of tho triangle formed by the radius vectors and the chord of the orbit are known, as well as the semimajor axis, This equation must be satisfied by the heliocentric coordinates of the celestial boty obtained in tho process of the caleulation of the orbit. Buler's equation for a parabola is obtained, as a special case, from Lambert's equation, if it is assumed that 2 = In spite of its theoretical value, Lambert's equation is rarely applied in the calculation of orbits. As Gauss has justly remarked, it doos not simplify the calculations in comparison with the usual moans based on the wlilization of the ratio of a sector to a triangle. Therefore, we shall not derive it. On the other hand, Euler's equation plays a most important role in the determination of a parabolic orbit. Academician A. N. Krylov has doscrihod in detail the history of this remarkable theorem, in- dlieating that its origins can bo traced ax far bacl: az Nowton {9} A. N. Krylov has shown how Euler's equation can be derived from the tonth lomma of Nowton's mothod for the dotermination of parabolic orbit. He analyzed Euler's original deduction, and pro- sented in conclusion tho simplest and shortest derivation of Euler's equation belonging to A. N. Savio. It is this derivation that we aro going to follow, in its general form; following Euler, Savich based his derivation On the properties of parabolic motion. 20 The Determination of Orbits From equation (8-38) we have for the moments ¢ and ¢” ae t)r 8b. 2) ct where o = tan dv and o” = tan Jy", From this, letting = k(t’ ~ 2), we find B(o ~ 0) vo" ~ In addition, equation (3-89) gives (40%), realy o) We introduce the chord s of the orbit and we express by it and by the mdius vectors tho angle v=» included between the radius vectors. Wo denote pete and we obtain according to the formulas of plane trigonometry cos F(u— nya t The minus sign arises if v” - > 180°, and, in such a case, this angle is the supplement to 360° of the internal angle of the triangle formed by rr’, and 8, We develop the ‘cosine of the difforence of the angles, using (21) 1 om Mra =1229, oan tao at 2G-9) (8-23) 7 Equation (8-21) can be transformed to the following form: B(@ - 9) 4 0°? ~ 0 w (0 ~ a)(3 +0? +0"? 4 0") = Vive? Five 30 ~B Foot 02s lees theo") Detemineton of Poreolie Orit m But, boenuso e172 3 (2 A, rae EET ar 7 = A WEF vVE-8) 12 + @- 9) VEE—o)- BL, pa® VIE + @~ 8) # VEE “a= 8), this is reduced to the equation az sara ts We substitute here for ¥ and simultaneously multiply by 2 ¥ (rare AF e 9) a6. (624) Here, the upper sign is taken if 0° ~» < 180°, and the lower sign if e” = v> 180°, This is Euler's equation, Buler's equation gives the connection betwoen the sum of the radius vectors, the chord, and the times but in ordinary cases of the calculation of parabolic orbits it is not particularly convenient because the small quantity s is determined as the difforonce of to terms on the left which are close in value. But it is not difficult to modify Euler’s equation for a small value of s in such @ way that this inconvenience is eliminated. Assuming, as it will always be in practice, that v” ~ » < 180° (otherwise the transformation is not noeded), and therefore taking only the minus sign in equation (8-2), we shall seok the solution of this equation for the chord, which ean be obtained by various methods. We confine ourselves to one of those mothows and write the ex- pansion given by Lambert: enero lfis 8 sernef “} (8-25) We note that a vory interesting deduction of this same equation is iven by N.E. Zhukovski:"Concerning a New Proof of the Theorem of Eambert™ (Complete Collection of Works, Vol. IX, p. 07, Moscow, 1081) 22 The Determination of Orbits at wr ++ (8-26) s\?__1a.s fs \s Ter) ores io lee ‘The notation » is introduced for brevity, From this it is possi- ble to find the inverse expansion for s/(r +r’) by powers of 1. Supposing. piper tan sagan. and substituting this in (8-26) we have Dam asn" 4 ayn? from which are found the coefficients ai, and the series for will be aedateghate sa (rite) nay (8:27) stem neta deaths Soate ne 29) It can be tabulated by the argument 7[19, 86]. In certain tabula- tions 1 is given at once, dosignated by a [10] (see also Table IX). It is also very simple to solve equation (8-24) without the aid of series and tables, but this hardly ever appears necessary in practice. Determination of @ Porabolic Orbit 218 60. The Determination of Geocentric Distonces ‘The problem consists in choosing such a value of p, that, de- ducing 94 from it by means of (8-15) and dotermining both the ra~ dius vectors and the chord, one should obtain a value of t on the right side of Euler's equation (8-24) coinciding with the observed value of We write equations (8-2) with the aid of (8-15) in the form Fy ayey— Ky tg = ay Mpy~Xyy Ve=b04- Yay Ya = ba Mey ~ Yay (829) Beier Z yy 25 = CM, Za From this we find 12 = R24 2R, py cos 0, + 93, } festh rh R34 2R, Mp, cos 6, + Met, where : RUONDSYPe 73, REX 4 V24Z3, 2R, cos 0, =-2(0,X, +b, Y, +0, 21), (st) BR, cos 0, =-2(a,Xy +b Vy + 0525): Furthermore, 88 = (ey 24)? +, — Ms? + (@y ~ 21)" ‘or by substituting the exprossions (6-29), 82-97 4 2ghp, cosy +h" pgs (6-82) where = (hy Ng)? + (V1 ~ Fs)? + (21 - Zs), 2gh cos yx =2 (as M~ a1)(X1 ~ Xs) (838) + (aM ~ by)(Va ~ Fa) + (co = 02) (Za ~ Za)hs 2 = (ag M ~ a4)? + (0g M = D1)? + (eg M = 04) Tho quantities introduced by us have an interesting geometrical interpretation, but we shall not concern ourselves with it, We note that it is not worth attempting to substitute these expressions in Euler's formula, In practi¢e, we choose a certain value of p,, and with it we determine ry, ray and s from the above equations." The value of the chord s found by this (geometric) method is denoted 24 The Determination of Orbits by 5,. Caleutating 9 by (8-26), we Find j, and from (8-27) we find the Chord, which we dosignnte by sy, because it was obtained from dynamic contitions expressed by Euler's theorem. ‘The values of p, must be modified until the equality 8, ~ sg is obtained. This may require several trials or, as they are som Limos called, hypotheses. By simplifying somewhat the equations of the problem, Banachiowier gave a method by which it is possible to find ap- proximately a value of p,, satisfying the stated conditions, and also to reveal multiple solutions for p,, if they exist, Without stopping for a complete prosentation of this problem, to which many investigations have heen devoted, let us notice that in the nothod of Olbers three solutions are possible, although in rare cases. However, not all these solutions give an accurate or even fin approximate agreement of the observed and tho computed co- ordinates of the second position, because this position does not fully enter the determination of the orbit. Therefore the solutions Which do not correspond to real motion of the comet can sometimes bbe separated from the true solution only after the exponditure of much work on calculations that are unnecessary for the final « sult. Under particular circumstances (the middte position of the comet. is oxactly in opposition with tho sun), all throe solutions ccan give an accurate representation of the ‘middle position, as Banachiewiex has shown in a fictitious example. It should not he overlooked that, because of the simplifications made, all these discussions aro, strictly speaking, correct only for infinitesimal intervals of time between observations. 61. The Second Approximation IF tho intervals between observations are short and, moroover,, roar equality, and if the tative vectors aro not too smal, thes Usually © second approximation is not needed. If, however, tho Contitions for the Wetermination of the erbit aro. unfavorable. in this sense, it is useful, without completing the calculations, to improve tho adopted valves of the ratios of tho areas of tho te angles. [Nt the start of this problem, itis possible first to correct the moments of observation for aberration; the correction forthe mid= dle moment is obtained by interpolation hetweon its valuos for tho extreme moments. Following this, “! and 4 aro calculated by Determination of a Porabolic Orbit 25 equations (8-9); the value of the middle radius vector is not re: aad, hom 2 at a ma fom! by mens af the gre expression! (6) ad (8) that do nt consi hn went Snera iene he uth sat fn int estat the ula a} ns Sere po-Morsme He A )e = Woy (6-84) Py where p,, standing in the parentheses, is taken from the rosults of the preceding approximation, and we arrive formally at the same ‘equation (5-15). In consequence of this, expressions (8-82) and (8-33) remain valid, and it is merely necessary to recalculate, with the new value (if), the quantities 2h cos y and 47. Having prepared the initial quantities, we can return to the de- tormination of p, and p5- : If the calculations are not for a preliminary orbit—in other words, if the elements of the comet, although not very precise, are al- teady known—we can determine at once the geocentric distances and the radius vectors for the moments of tho observations, and then proceed with the computation of the orbit according to the more precise formulas given above. Another method of improvement is based on the representation of the middle position of the comet, after the elements of tho orbit have been obtained; it will he briefly described in Section 63. 62. The Determination of the Elements ‘The course of the determination of parabolic elements does not diffor in prineiple from the corresponding derivation for an ellipse. Using the values p,, py, 7, and ry, we find the heliocentri coordinates, and then we calculate, according to (5-90) and (5-81), os Sos Yor #os Ad ro. Through equations (5-82) we determine sin 8f = sin (vy - v1) = 2, cos (v3 = v= SL (8-35) For the derivation of the true anomalies we turn to equation (8-39), whieh gives ry qsec? gor, t= gs00? dvs. 26 The Determination of Orbits From hero, it follows that cos do. YE a°t "1 cos 3 vy =cos(S 0,41) (8-30) =c0s Fy, cos ¢— sin} oy si = cos tv, cos /—sin Fo, sin f or tan bay cin f— 2 1 cos by, sing (837) After this, we have dopa dore : BU g th ovatan des, (838) oto 8-39) Teal ive a) Through os and vs are found hand ay BY using Barker's table, From equation (3-38) we obtain m Tad ou, Mae My (8-40) Finally, the vector eloments are determined in tho usual way, and from them o, 2, and i . 63. The Representation of the Middle Position and the Improvement of the Orbit Having obtained the elements of the orbit, we can use thom to represent the middle position (as woll as other observations, if they exist) according to the formula: (8-41) Determination of a Parabolic Orbit a « is found in Barker's table, According to (8-42) and (3-43), p cos § cos 4 =a, (L-07)+ bro ¥ X, pcos B sin «=a, (1-0) + dy or Yy (62) psin 8 =a, (1-0) +b,04Z. However, in conteadistinction to an elliptic orbit, it is not al- ways possible to require that the parabolic elements represent ox- acily (within the limits of computational errors) the observed co- ‘ordinates of the second position of the comet. Divergencies can arise for a numbor of reasons. First of all, the second observation enters in the calculation of the orbit only by means of the condition that the observed posi tion must lie on the fundamental circle, In order to check if this requirement is fulfilled, one must substitute in equation (8-4) the values of a2, b2, e2 obtained from the coordinates ca and 5a calculated with the eloments. If agreement is obtained, the calcu lation of the orbit is in order, and if the dovintions Aa’ and Ad do not exceed the errors of observation, the results can be considered as satisfactory. Indeod, if the comet moved exactly along a parab- ola, and the observations wore not subject in any way to errors, the calculated second position must coincide with the observed. But, because the observations aro not completely accurate, a divergence can arise as a consequence of errors of observation. If, however, the representation shows, as is not infrequent, con- siderable divergence of the observed and calculated coordinates, then this can be for three reasons: noticeable errors of the ob- servations used, an essential deviation of the real orbit of the comet from a parabola, or conditions theoretically disadvantageous for the derivation of the orbit. from the observations selected for this purpose. If, moreover, equation (8-4) is not satisfied with sutficiont ne- curacy, this may indicate, besides computational errors, that the values ‘obtained for ni/ng and 1/ns were of insufficient accuracy. Tn order to clarify the causes and chaencter of the deviations in the representation of the Second position of the comet, we turn to the following considerations expressed by A. A. Yakovkin [17]. We shall consider that we are given the first and third posi- tions of the comet, and we shall assumo that tho coordinates c1, 34, cay By aro invarinble, As we have seen, the second observa: tion is’ used only for the Finding of the value M = p4/p,. Conse quently, having the extreme positions of tho comot and Knowing M, wwe can calculato the orbit by which the position of the comet 218 The Determination of Orbits at the moment of the middle ob- servation can be found. The set of such positions caloulated for all possible values of sf repre: sents a cortain curve on the eo- lestial sphere which \. A, Yakov- kin called the isochrone, and the second position of the comet must be at some point of this curve. Meanwhile, instead of the second position, a great circle passing through it is used for the determination of the orbit. Ealiotic If, as a consequence of errors of observation or deviations of the orbit from a parabola, the second observation does fall exactly fon the isochrone (Fig. 18, on which Ly, La, La are the posi- tions of the earth, Hy, Fa, K's are the positions of the comet, and the curve KK’ KY Ks is the isochrone), we obtain, instead of the observed coordinates of point 2, the calculated coordinates of of the point A% situated at the intersection of the fundamental circle LyX and the isochrone. If the fundamental circle inter- sects the isochrone at an acute angle, as is shown in Fig. 18, the feviation of the observed coordinates from the calculated coord; nates, characterized by the segment 2 K's, may be rather sizeable This, however, merely indicates that the fundamental circle is Inid out disadvantageously for the determination of the orbit. Evidently, the very bost representation of the second observation of the comet will give an orbit to which corresponds point K'' of the isochrone lying at the shortest distance from 2, i,e,, at the end of the seg- mont KK’ perpendicular to the isochrone It is not easy to have an accurate idea of the form of the iso- chtone in every concrete case. Therefore, the most convenient, method for correcting 4 parabolic orbit, according to three given observations, consists simply of empirically modifying M. Namely, we calculate, in addition to the orbit already obtained, another system of elements in which the initial data are the samo, but the value of Mf is somewhat changed, for example, by an amount of the order of 0.001. Let Aa’ an Ad” he the deviations of the observed cooriinates from the calculated coordinates of the second position, in the case of the first orbit obtained with the value M= Mi. For the second orbit, in the calculation of which M = Mz Determination of @ Porabolic Orbit 29 was adopted, let these deviations be Aa” and AS’, respectively. Then, assuming that those equantities change linearly with a change in M, we find the most probable value of M= My +2 by equations cos 5 (Aa” ~ Aa’) a” cos 5, i (ns’=A5”") AB 7 If both equations are solved by the method of least squares, thoy give 7) Aa’ 4 M8” (A8" Bayes Oe aa (43) Now we caleulato the orbit with te value M= My +2. The calculated second position will correspond to the point Ky’ on the condition that z is small and that our assumption concerning the linearity of the changes is justified. In such a ease, it is possible to assert that the paraholic hypothesis is quite unable to give a better representation of the second position than that obtained. Consequently, if great deviations in the coordinates of the second position of the comet romain even after this, they can only be the consequence of errors of the observations or of an actual Geviation of the orbit from a parabola. This can be verified by making use of other observations and by determining the orbit anew. ‘An analysis of the form of the isochrone was carried out by A. A, Yakovkin, and he obtained, for comets loss than 90° away from the sun on the celestial sphere, a line similar to that repre sented on Fig. 18. Moreover, in certain cases, the isochrone can have an isolated oval degenerating, under certain conditions, into loop of the basic eurve. ‘The number of solutions in the method of Others is determined by the points of intersection of the funta- mental cirelo with the isochrone because all of these points satisfy tho conditions of the problem. Consequently, as seen from Fig. 18, there can be one of three solutions, depending on the existence ‘of an isolated oval Q of the isochrone and on the existence of intersections of the oval with the fundamental circle. Of course, in the case of three solutions, eack root gives its values of the devi- ‘tions of the second position from the observed coordinates. For roots not corresponding to the actual orbit these deviations can be very sizeable, Besides, as mentioned above, even all three solu- 220 The Determination of Orbits tions can give an accurate reprosentation of the second position— of course, under exceptional circumstances. A. A. Yakovkin obtainod for the first time the intorosting result that if the fundamental circle is drawn, not as done by Olbers and as shown in Figs. 17 and 18, i.e., through the second position of the sun (and the earth), ut somehow differently (for example, as a perpendicular to the indicated direction) then even five solutions can be obtained for comets which pass close to the sun at peri- holion, 64. An Exceptional Cose in the Determination of a Parabolic Orbit The determination of an orbit by the metho of Olbees in an exeaptional case becomes unreliable, because one does nol suc- cood in determining the values of the ratios | ,/44 and 1/0, with sufficient accuracy, Tt has been shown above that the fundamental Circle then passes near che first and thitd positions of the comet. It is ensy to establish on the basis of the analysis by A. A. Yakovkin givon in Section 63 that the most advantageous position of the fundamental cirele is obtained if it intersects the isochrone at right angles. This condition woul! have been satisfied by us boforehand if the form of the isochrone was known. In the majority of cases the isochrone does not deviate too much from a preat circle passing through the First and thied positions of the comet, whence it follows that this great citcle must intersect the funda” mental circle at an angle as close as possible to a right angle, and in any case not at too acute an angle. The isochrone may have, as shown by A. A. Yakovkin, a rather complicated form. ‘This will occur for comets close to the sun for which, in consequence of the smallness of the radius voctors, the iletermination of the orbit ropresents in general a difficult problem; as presented, our analy- sis, based on expansions in series of expressions for n,/ma and Vn, would not be sufficiently rigorous. Jn practice it will be satisfactory to choose the coefficignts in equation (8-4) in such a manner that the values of oy and «'y ob- tained are not too small. For this it is simplest to write equations (8-3) in numerical form, and then to eliminate 9, from them, choos ing such a pair of oquations that the coefficient of p, "will be small after the elimination of py. Thus we find equation (8-6) and from this we obtain, through division by the coofficiont for py, equation (8-7), also in numerical Determinetion of @ Parabolic Orbit 221 form with the exception of the quantities ny/ng and 1/ng. It is possible to take as a first approximation for these My tant 1 tents ay nt oe (8-44) aftor which the values M and m contained in equation (8-7) are ealeulated. For the radius vectors of the comet we shall have in place of (s-B0). Pe RB a p36 2p, cos By, 0) FE a R24 (Mp, +m? +28, (Mp, « m) 608 6, Furthermore, y= aype— Nyy tym agilp + aym—Xyy brat 05m —Vyy (5-48) Oper ~Ziy 25 = CyMoy + Cgm— Za Mya bror- Vos vam be ‘The cocfficients of formula (8-82) for # are given by the fotlow- ing expressions: g? = (agms Xy—Xa)? + ams Vy Ys)? +(esm4 21 ~Zs)%, Xa) 2gh cos x = 2llay M ~ a1)(osm + Xs + (ba M = bi)(bam+ Yu = Fs) + (CsM=e1)(cym+ 21 = Zas BE m= (ag M = ay)? + (Ba M = by)? + (og =), (8-47) The solution of the hypotheses for p, and the determination of the elements does not offer anything now. However, it may be reconmended, in the majority of cases, to carry out a'second ap- proximation as presented In Section 61. Indeod, # contains In the denominator a small quantity of the fiest order, J 4, exprassed by the sine of the spherical perpendicular from the third position fof the comet on the fundamental citcle. Formulas (S-44) have an error of the second oriler in comparison with (8-9). ‘Thus, when (S44) is used, errors of the first order might remain in the goo- centric distaneos and this might lead to noticeably inaccurate colemonts of the orbit, 222 The Determination of Orbits 65. Summary of Formulas A. Initial Data try ty Bu, MG, Yay Zay fa, 2, Ba, Xa, Yay Zay o ty, a, Bay Nay Yay Zs. ‘Tho coordinatos of the comet and of the sun must be referred to the same equinox. If the elements of the orbit are already known, it is necessary to cortect the moments of the observations for aber= ration and the coordinates of the comet for parallax. If prelim nary orhit is calculated, parallactic corrections are introduced in X,Y, Z only if it is assumed that a second approximation for ni/ng ant 1/ng is to be calculated. Otherwise, it is possible to Aisrogard tho parallax. A parabolic orbit is usually calculated with five decimals, and only in rare cases with six decimals. Often one is obliged to make use of observations of restrictod accuracy, and then it is sometimes possible to be satisfied with four decimals. B. Constant Elements 2 cea sin 5 08 6; c08 «ts cos 5; sin cy = 12,3). a Control: a}+ D2 4c? =1, a,~ 0, sin a, = 60s (0, + 5), by + 0,008 a, = sin (0, + 8,6 RE aXtayte 23, Rpts E22, an QR, cos 0, = -2(a,%, +5, ¥,4e,2,), QR, cos 0, = -2(a,Xy + by Vy + €5 25), Control: RY 428, cos 0, + RZ 4 af, cos 0, + =X, =a)? + (F, - 8 +Z,~ ey), Ay = ag)? + (Vy ~ Bg)? + (Zy — 05) Determination of a Parabolic Orbit me ©. First Approzimation A= baZa—cnYa, BecgXz-aaZa ayy Cw ay ¥2~baXa Control: ag + Bby + Con =0, AX: +BYa + 0Za ty ty Aas + Bbs + Cor tate Alay + Bb + Coy” 9? = (Ky = Xa) + (= Yay? 4 (21 Za) 2gh cos y = 2M(agM = ai)(X1 ~ Xa) + (ba M = Br)(Ya = Ys) + (ea 01) Zr ~ Za) BB = (ag May) + (Bg M— B1)? + (ea He). Control: . 4 29h 008 y 448 = [Xs — Xo) + (aM =a)? +01 = Ys) + aM = 9)? + (Zs ~ 23) + (es M- es))- Taking @ value of p, at random (with some experionce it is possible to judge the distance of the comet from its daily motion in the sky), or taking 9, from a previous determination, we cal- culate Me- Ww) P= Mos ri R44 22,9, cos 0, + 02, Fe R24 2i 4 cos 03 + 03, a} a9? + 20ho, cosy + AP7, wD 2k(ts ~ 63) ray whore is taken from the tables ((10, 19, 36] and Table IX). The values of p, are modified until s, equals sg. After the second hypothesis one may use interpolation. ga (ry tre) ny 2k = 0,0844082, Cy, The Exceptional Case If tho numerator and denominator of M aro very small, the equa- tions are written in numerical form 224 The Determination of Orbits AML Maps + Oy Mypy =m X,—Xy + ny Xyy Bymyoy— Papa + Patapg emi Vy —Ya+ng¥y f (Va) SEM 1 CoMy + Cg Mypy = M12, ~ Zz + ng Zye From two equations of this group p, is eliminated, which gives Sime, + Yo mypy Oy + Ogn5, a= Mp, +m, 9? = (as m4 Xy~ Xa) 4 (byms Ya ~ Ya)? + (ey ms Zi ~ 2s), 2yh cos y =21(ayM- a1)(ax m+ Xe - Xs) + (By M—Bs)(Ba m4 Ys ~ Ya) + (cal er)(oam+ Zy~ Zs), WP = (ay M~ ay)? + (by M = 51)? + (ca M~ 04) (Va) Control: 97 + 29h cosy +h% = (ag M— ay + agme Xs Xs)? +s br 4 byme Ys ~ Ys)? +(csM~ cr eegms Zi ~ Za). D. Second Approximation Usually it is possible to get along without a second approxima tion if the Intervals of timo are small and nearly equal, and also if 7: and ry are not very small. If a second approximation is needed, then we calculate: Baty Api =~ og, = ty-Apy, A= 04.0579, where Ap, is found by interpolation between Ap, and Ap,. Tre kUS ED Tee R= 02), Te= RE =H), VID Determination of o Parabolic Orbit 25 4 4a a NTE aA ben in 1b sheet), myean) ns 8 +t) (retry) 7” wet Ae BOs 4 Coe mm day + Bos + Gos 1 a Boa + Cos om x [on +BY: 402s) 2+ AX + BYs +03]. In the exceptional case, ni/ng and 1/ns ate substituted in the formulas for I and m of group Ca. Py= (py, (= Me a (p, is from the first approximation). te (VI)-(VIl) again, taking (¥) in place of Next cal E. The Determination of the Elements Pas Mp, oO py Mp om B= a;0,-X, veo bev Ve a oe CS) ata yPs ater? Control: The values r; and rs must agree with the results of the Inst hypothesis. Fie tM Ys ter «ny Peaks vated, sin of = 72, 226 The Determination of Orbits Control: oy stan «mp 1 Tye isot Conical: Tho two values must agree. Mand Ms are taken from tables with argumonts of o1 and os (or thant). Tatham aay a P.=2, = zt Pyen ae 1 Oy aM a) Pyne, Ot Qa : A= aPe an 9Pyy a, =P, d= 290, b= 290), = 270 Control: a2 4a eat= 93, be + be + 62 = 49%, a,b, +4,d, 64,0, =O. sin isin @ =P, con 6- Py wins, sin {cos w = Q, cos €~Q, sin f, sin 2 = (P, cos © ~ Q, sin a) see «ivy cos = Py cos @- Qe sina (Py sin @ + Q, 008 @) ese 8 Control: sin 1 and cos § must belong to the same angle, and the same applies to sin i and cos 7. Determination of a Parabolic Orbit 2 F. Representation of the Middle Position r 1 Saw? Oa tan 5 ae yy 2 (or 2) is taken feom tables with the argument Ma 4 C08 By C08 a = ay (1-03) +b, 02+ Koy 608 By sin dy =a, (1-08) +3, 0,+¥a,> (XVID) py Sin By =a, (L-03)+b, 044 Zy- Concerning the differences between observed and computed co- ordinates, soe Section 63. 66. Exomple For the Comet 1946d (PajduSikové-Rothart) there are the fol- lowing observations: . T No UT. a Fiaeeo 1 1945 June 6.87526 18431"10*.75 +88°56"41"".9 2 8.87900 12 58 30 189 +28 4734 8 a 10.88198 1230 16 “66 425 1198.6 made June 6 and 10 by A. D. Dubyago at Kazan and June 8 by D. J. Martinov at the Engelhardt Astronomical Observatory. The calcula- tion of the corrections of the coordinates of the sun for parallax are omitted here; only the corrected coordinates are given. 0 1 2 3 x -0.253601 0.220828 + 0.187794 Y 0.901602 909093 + 0.918485 2 Tolsgoo7s ~—Folsgazns + 0.396900 cos 2 0.901890 0.067582 ~ 0.086350 cos 0.829574 +o.sTe367 0.904286 sing Zo'seraat = Lo.z5a587 0.170544 @ — o.rea7s4 = 0.847957 0.8060 Togas = Lowzaiaaa = Lontsaazs © sosssso7 | F0.aBIRas F026 cos (a=3) o.teiiz3 ——Torsna600 904228 Sin (43) FO.1NSGBE ——FO.BtABOT 0.268078 m ae 1.030064 cH 1.030958, 2Rycos 8 0880812 «2M con Oy +0.27D502 W 4 4o.s25t18 fs-ta 198198 8 40552 ana 2°00473 & — For%220%0 tin o.988652 Numerator $.0.143180 ¥ 's23600 Denominator 092001 228 The Determination of Orbits Vaya; 0.503707 Nya +0.085807 bs; 0.086260 r= ¥a “0.013883 cay +0.090136 Zi~2a = 0.006020 2 0.0035595 tg 8.98667 2ohoosy — = 0.0516823 Dyn) 0.137155 » + 0.3651602 Because 2 second approximation was not planned, correction for aberration has been introduced in the process of calculating the hypothesis, after the socond hypothesis, when tho value of p, has been determined with sufficient accuracy. vi 0.250000 9.260000 ate1i7s 0.261140 7 0.062500 0.067000 0.0880189 _0.0881955 "J O.orioN4 —ownsa2ar 00088520 0.005404 r} assoc —.2ase7s——L.a36014 1.280884 py 01880922 o.8aur80 ——ONSaTISL—O.n9TOTO rp 1.282558 1208657 1.300584 300537, fF} Vtesois —atiteo) | iv2ien—tcit2iag fs tutaasot 1 ta9580 © igora —"igoatt ruzg szsoui9 —a'2stigs a asane7 2 aaaned (rsray% oznsnes 0.200060 20578 0.298700 7 0.010028 0.010607 0.040561 0.040502 mi 0.040087 G.010610——B.0keses 0.040868. 4a 0.001682 o.oataai——o.oa1ate —oc0ntaTs 84 0.054025 0.000879 —ovoaiaya —0.u1375. speed 1037 a2 S10 ° tov00. r42 178 vu Py O61 bs o.a0s ap, 0.00151 ae 0.00230 #2 6.7378. 1 10,85963, 4-6 3.9888 katy) OtdTIAL x 1 3 = -o.saa24 ~ 0.542586 > ~onsssat To.oretee Paereraty T0223 r? 1.236886, 1.300535 > vainist paren) XI © 1.022854 ain (oy = 01) +0.077102 fq ~ 00701262 cen (oy 1) $0.897023 ¥y 0.030676 eye, ASaS TE 2 2h soloa20003 7 21239 6 12 o.oorma126 ro wsroare Xt im Determinction of o Parabolic Orbit 29 oanssa0 ote Fi o.asrsa2 pos 10° 6's" etn f 25.9011 ay +0807 mcosoe s fH ~25.s972 ° .o18i2s a1 +0.503900 ‘ 1.018120 xm 28.7388 4 1.018122 20.8009 aia 1008020 aoutsete 7 1027508 so.t4ns r aime r tats 1 Moy 11aTe2 0.850907 sin vy + 0.556822 Torus sine sry $0.g00801 75.338070 con vino, +9-449000 BS onaoss 2," “otasss PS Lolssoxs0 3 20817 Pl osa8690 2 Fo.o8sas0 ae + 0.164961 be a hastane 3) tosaera91 8 Toiaae a2 0334656 sr Fontspara XVasin énin oy +0,088049 ain Q_—~ 0.854600 sin feos s0167TIs cas 0.81943 sino F0aTSsIE @ 22°1s'36"5 cas +0,005873 got i638 se Sint }0usti2t é 169 $3.32 18 cost “0.988441 Elomonts T 1946 May 1141722 U.T. a 22° 18'36".6 f sor teas ‘} 1040.0 f 169 33.09 ¢ ors XVI App 0.00190 pg cas 8g con ay —0.277787 * —so.sts00 py con By sin 4, 0.079500 '$- 7 +25eosr pasin 3, +O.187786, Lorton pcos 82 + 0.287000 + oa2b 85 33 tweear’as + 0.804054 3, sae AT a5 ka a8 Ome Tas ea 230 The Determination of Orbits The deviations do not exceed the errors of the observation, ad- missible in the case of rapidly moving and fuzzy objects such as this comet. Further improvement of the orbit could be only formal, ‘These elements represont other observations made on neighborins, datos with errors of the same order, and we ought to consider the orbit reliable. How accurately the computed position of the comet lies on the fundamental cirele was determined as a verification of the orbit; the obtained discropancy does not exceed the limits of possible errors in the ealeulations: 08 «ae ~ 0.967589 a, ~ 0.847959 cos Dae + 0.876363 5, - 0.221310 sin ye ~ 0.252532 c, + 0.481651 Aa, + Bb, + Cee ~ 0.000006. 67. The Me 1d of Olbers in Ecliptic Coordinates In the majority of parabolic orbit computations, the introduction of equatorial rectangular coordinates, combined with the use of a calculating machine, does not yield those advantages realized in the determination of an elliptic orbit. The formulas remain quite unwieldy, and whether it is possible to transform them succes: fully is doubtful. At tho same time it has greater meaning for the Gotermination of a parabolic orbit to introduce ecliptic coordinates and to make use of logarithms; this will almost always be the quickest approach to the goal. Therefore, one should study the method of Olbers in ecliptic coordinates, i.6., in the form in which it was appliod all through the nineteenth century. The method oes not differ in principle from the presentation given above; it is, therefore, possible to refer repeatedly to the preceding de- duetions. Tho initial data horo aro: the moments of the observations t, (é=1, 2, 3); Av and By obtained by formulas transforming the co- edinates fram the observed %, and 84; leu the longitulos of the arth; and &;, tho radius vectors of the sun (in the calculations Appear their logarithms only). We use directly the basic equations of Olbers, i.e., (6-14) and (8-15), which give the connection between p, and p, in the form es- Mer (48) where (see Fig. 17) Determination of o Parabolic Orbit BI tanta Yi ts-ta cos PRy tat)" cos PRs tate sin KiQs tyaty in Ka Qs" But in triangles PEK) and PEXs between the pole P of the fundamental circle, the pole E of the ecliptic, and the positions of the comet, K's and K, respectively, we have PE = f, EK = 90° - Bay EKy = 80° ~ Bay ZPEK, = 90° 4 A, = Lay LPEKg = 90° 4 As La consequently, J, cos Py sin Bs cos [= cos 81 sin I sin (Ar - La) WV, cos PR, ~ sin Ay cos 7 cos fs sin I'sin (As ~ La) (3-49) (s-50) From the triangle LK’, gq, in which Ky 92 = Bay! = Az ~ Las we obtain for the angle 7, Lyi, =1 tan Bas Sin Qa — Lay After substitution of (8-50) and (8:51), equation (8-49) acquires the form tan] = (6-51) t-te sin By etn I~ cos By sin (1 ~ La) tq =) 005 By Sin Qy ~ La) — sin fis ota J This is the basic equation of the mothod of Olhers. It can also be written in a more elegant form, but one that is less convenient for calculations. In the exceptional case the numerator and denominator of the expression for M become very small. In order to avoid the rosult~ ing inaccuracies, i¢ will be necessary to have recourse to the rig- orous equations (8-7) ant (8-8): (5-52) (58) Calculation by these formulas with logarithms is already less convenient than with an arithmometer, Thorefore, if it is evident from the start that an exceptional case exists, it is useful at once 22 The Determinotion of Orbits to hegin the calculations with an atithmometer, in equatorial co- ordinates. In order to decide this question it is usually sufficient to mark the first and thied positions of the comet on a star map, and then to mark the second position of the earth or sun (by equa torial coordinates taken from the astronomical yearbook), depend- ing on the closeness to the positions of the comet, If the fest and thied positions of the comet lie approximately on a great circle with the sun (or the eatth), it means we are faced with an excep- tional ease. If this is not so, one may proceed confidently with tho calculations by the method of Olbers. As was said, in the ex- ceptional case itis bottor to calculate with an arithmomoter (which will bo preferable, particularly for correction of the ratios s/s and 1/ng in the second approximation). But this is not absolutely necessary, and here an altornate calculation designed for use with logarithms’ will be given. ; In equations (8-53) the symbols J, and JJ, ropresent quantities proportional to the sines of the spherical porpendiculars on the fundamental citcle from the corresponding positions of the comet. We now mako the factor of proportionality oqual to unity. ‘Turning to the hasic equation (6:5) we note that for this it is necessary that A, B, and C represent direction cosines of the vector passing through the pote of the fundamental circle, The second formula of (8-5) shows that ©, represent the sinos of the spherical perpen- iculaes from the positions of the earth on the fundamental circle multiplied by the corresponding R= yX? + ¥?4Z?, , The choice of the fundamental circle; must be such that J and &, become as great as possible in value. As was already shown in Section 58, this can bo achieved when the funda- mental circle is laid out perpendicular to the great circle drawn through the first and third positions of the comet. Lat (Pig. 19) the fundamental circle be drawn hy. the indicated method through the second position of the comet, Ka, inversect- ing the ecliptic in point M1. There is no particular need to confine oneself exactly to the indicated choice of this point, because some Determination of o Parabolic Orbit 233 displacement of it does not cause s noticeable reduction of acew- racy in the determination of |, and 4, Instead of taking the fun damental circle perpendicular ‘to the ereat circle passing through the first and third positions of tho comot, it is possible with cho same success to draw tho fundamental citclo perpondicular to Otbers* circle passing through the second position of the comet and the arth. In tho exceptional caso, it reduces to almost the Same thing. Phen, using Fig. 19, we find tan Las = tan (Aa ~ La) soe 1, tan (Il ~ La) = tan Laks sec l, (854) tan (I~ Lg) = tan (Ap = La) see? L If the presence of an exceptional ease was not noticed at the start of the calculation of the orbit, it will reveal itself during th determination of the ratio '/,: 4, when the numerator and denomi nator obtained turn out to be very small. Then we find / by equa ion (8-51) and calculate M1 according (0 (8-54). The equations used for obtaining J and M are completely analogous to (8-50) and (6-51), only in place of the second position of the earth will ap- pear the point 1. Consoquently, they will be tan Ba tan J = — (3-55) in Bs otn 08 Ba sin (Aa But it is apparent that in the given case (see Fig. 19) Oy = Re sin Las = Rp sin (Il ~ La) sin J, © ey sin Lyoy «Ry sin (1 ~ Ly) sin J, (est) ©y = Rs sin Laos = iy sin (fl ~ Ls) sin J, (56) whence we have £2 angie Ryn te 7 1 + Ra sin (1 ~ ba) cos By sin (Ag =f) = sin Bs ind fy sin ht) Leys tb +s sin (I-Ls)|. (858) 24 The Determination of Orbits Finally, equations (8-56), (8-57), and (8-58), with the sub- stituions J = and Ml = Lz, give the basic equations of the method fof Others. The second member in brackets in the expression for 1m drops out, a5 we have seen previously in Section 58. AS a first approximation in che exceptional ease, we ean take, a5 in tho equation of Olhers, my tants mg teats" (559) aftor which it is possible to find 4. As far as m is concerned, it is not difficult to show that it is « small quantity of the first order. Reasoning in the same way as for the deduction of equation (8-18), we shall have for the earth Jpn 1 [Re as sin (= ta) = je Ra sin = La) +85 sin tt —La)} Subtracting this from (8-58), we chtain Ry sin (VL) -(¢ - i) a an at] (6-60) The differences of the areas of the triangles standing in the brackets are small quantities of the second order, as follows from ‘equations (8-9), while in the denominator stands sy, a quantity of the first order with respect to the intervals of time, which proves ‘our assertion. From equations (8-49) it is possible to conelude that the first rand thied positions of the comet lie on different sides of the funta- mental cirele; otherwise the value of 4 obtained would be nega tive. If the fundamental circle which in the method of Olbers passes through the second position of the comet and the earth is, in the exceptional case, close to the circle joining the first and thied positions of che comet, then we may conclude that the curva ture of the apparent path of the comet must be small, and it will he the smaller the more precisely ensues the exceptional case. But from Lambert's theorem concerning the curvature of the ap- parent path of the comot it follows that this takes place when ra Determination of @ Parabolic Orbit 235 is close to Rs. Therefore, the numerical value of m must dacrease Still more, as follows from equations (8-8). In fact, if rs and Ra aro equated there, only terms with the difference of the radius vectors remain, which in the majority of eases are known to bo small. Thus, it is possible to discard m in the first approxima tion. Howaver, in the second approximation m must be taken into account. We now return to the method of Olbers and pass to the caleula~ tion of the auxiliary quantities. For this purpose we first make up ‘expressions for the rectangular ecliptic coordinates of the comet, using formulas (3-5) and letting B = 0 in them. Thus, in the method of Olbers it is assumed that the earth moves strictly in the plane of the ecliptic for a selected equinox. This does not create errors exceeding the limits of the errors of observations of comets, pro- vided that the equinox is taken at the beginning of the yoar of the observations. This is generally the case; but if the coordinate system is referred to the normal equinox, and if this equinox is far removed from the year of the observations, it is necessary to re- mombor that latitudes of the earth can reach perceptible magni- tudes, and this can lead to inaccuracy of the results of the caleu- Intion of the orbit. And so we obtain 2, =p, 008, cosh, +, cos L ¥,~p, cos B, sin A, +R, sin L Fy =p, sin By 81 2 = Mp, 60S fy C05 Ay + Ry COS Lg, a Ya = Mp, C08 fly sin Ay + Ry sin D 21 = Mp, sin By. From this it is easy to derive 12 RE 4 9i jp, c08 0, +03, } 500 = RE 20 Mp, cos 9, + HP 0%, a where si som 846 Qs = Loh} (53) 608 03 = cos fla cos (As ~ La). Those formulas ace evident, if it is taken into account that 0; and 3 are the angles hetween the continuations of the radius vec= 236 The Determination of Orbits tors of the earth and the corre- sponding geocentric directions to the comet. From cos 0; and cos 02 it is possible to find tho sines of these angles. If, however, the cosines are close to unity, the determination of the sines will be uncertain and it is better to use eaiome S the following formulas, resulting is from Fig. 20, where the notations Fig. 20 aro obvious: sin 8¢sin w,=sin ® Sin 0; cos 1; = cos Bi Sin (Ai = Li) } eum een Lot us notice that these equations can also he obtained referring to the figure. If the parameter w is eliminated from eq) tions (864) by squaring the equations and adding thom together, the result appears as a simple consequence of equations (8-63). However, the angles 1 are not really needed. For the determination of the chord wo shall take the differ- ences of the geocentric coordinates of the comet according to equations (8-61) and sot, Mp, cos Bis C08 Ay Mp, cos @y sin Ay 1 008 By €08 A, = Ap, c08 08 Hl, Py cos By sin Ay Mo, sin By ~p, sin Rs c05 ba ~ Ri cos bi = Ry sin ba ~ Risin b hp, cos sin Ml, ho, sin ¢, (8-05) g cos @, sin @ We also obtain, as before, 8? = 92 + 29hp, cos Leos (= G)+ 3292, (8466) ‘and for the auxiliary quantities we introduce formulas more con- voniont than (1-05), namely J cos ¢ cos (H 24) = M cos fix 60s (Ay ~ Ar) ~ C08 Bay cos ¢ sin (H— Ay) = Meas fig sin (Xa ~ As), Asin f= Msin By ~sin As, (8-67) 9 £08 (@ = Ly) = Rs cos (les = Ls) = Ray gin (G ~ Li)= Ry sin (La ~ bie Determination of a Parabolic Orbit 27 In conclusion, we assume cos x = cos ¢ cos (Hl ~ 4), (6-88) and also sin x sin w=sin ¢, ain y cos w= cos {in (HG), (8-69) and we shall have a? = 9? + Q9hp, cos y + A¥p2. (8-70) Furthermore, we replace (8-62) and (8-70) by these expressions: Fa (+ Ry cos 0)? + RE sin? 8, = (0, + 1)? + 12 a (Ip, + Ry 005 04)? + RF sin? Oy = MI (py + f4)® +131, 33 = (Ap, +9 008 x)? + 9? sin? y = Allo, +)? + 2, where Ry cos A= fi, Risin di = ly Ry ud 4 eT “7.008 Oa = fay Ge sin O3 = tay en) g g foo =f Zsiny #1, fund we introduce the auxiliary angles as T matty Thon, evidently, 1, sec Hy, ty = My 800 yg, 8, - Asoc Ye. (873) tan yy = tan y =P") (Beta) We have obtained as the result of our transformations formulas, particularly conveniont for logarithmic calculations. The determination of p, is carried out by means of hypotheses as in Section 60, Taking a value p,, we calculate r,, 1g, and 5,5 the geometric value of the chord, Bosides this, we have according, to (§-26) and (8-27) 2k (ta = t1) 1 yea ETH STM (8-74) which gives the dynamic value of the chord. Wo must find a value of p, for which 8, = 84+ 238 The Determination of Orbits If it is necessary to make a second approximation, the ratios of the areas of the triangles aro improved and m is introduced (it it has not alrondy been introduced). This procedure is carried out as indicated in Soction 62._If the solution does not need a second approximation, it is possible to calculate directly tho elomonts of the orbit, Knowing , we find p, by one of the formulas Ps= Mor oF ps~Ma, sm (8-75) Next, from equations (8-10) we have 7, 608 8, c08 (I, ~ Li) cos B; cos (Ay - L,) + Ry 7, c08 b, sin (1, ~ L,) fa 1,8). (8-76) = 9,008 i, sina, -Lo ae 1, sin b, = p,sin B, For determination of the position of tho plano of the onbit, first of all we notice that if, =2;> 0, the inelina- tion #< 90°, ‘and if Ty 1, <0, i> 90°. 7 From equations (2-00) it follows tan ésin (ty - 9) ER @ Eien a tan Bi, $17 Ban nisin ay POTD = tan, I ig also possible to obtain these equations from consideration of the triangle on the heliocentric celostinl sphore (Pig. 21) bo- twoon the position of the comot K, tho nado Sof tho orbit, and point 0. Expanding in tho second equality of (8-11) ly == (Is =) + 2) =, we have tan By = tan sin (21 =) 08 (ly = 1s) + tan i 60s (J; = 2) sin (is = 1) = tan by cos (Is ~ 21) + tan feos (2) ~ 2) sin (ly ~ 1) Determination of a Parobelic Orbit 29 From this it follows that tan By ~ tan by cos (la sin (fy =) Equations (8-77) and (878) give i and Sl. The same formulas (2-60) of Fig, 21 permit us to write tan i c0s (ty ~ 2) = (8-78) tan by tan uy = tan (2; ~ Sl) sec = ——-——_____, ce =i” | og ay) tan By tan uy = tan (1s 2) see f= (8-80) Hla =e) 4s =m 1 a ese 5(us = ui)i (6-61) aye etm 5 (us ~ ua) ~ according to equations (8-80) and (8-81) we calculate v1 and 9. Next, we have ato, ws sus— a (s-82) Knowing %; and vs, We take HM, and Ms from Barkor’s table and by (8-40) we find P. This completes the determination of tho orbit. For calculations of the representation of the middle observation of the comet by the elements just found, we apply equations (2-51), (2-50), (2-32), and @-10). With regard to the representation of the middle position, the re- marks in Section 63 remain valid, To these remarks we further add what follows. 240 The Determination of Orbits If the middle position is not represented quite satisfactorily, it is possible, denoting the computed values of the coordinates of the comet by Axe and Bae, to find tan Bae Sin Qa, = La) and to compare Je with the value of J calculated according to (8:51). In the ease of noticeable divergence it is possible to apply ‘1 method of empirical correction to M, indicated by Carlini, namely, to uso as the basis of the calculation of the corrected orbit a value of M which is found by changing log tan I by the value of the divergence with opposite sign: tan Je (8-83) log tan fx = log tan I+ (log tan I= tog tan I). (8-84) Using 1,, we calculate M by the expressions given above, presorving Without change the quantities not depending on M. If the deviation in J is due to the inaccuracy of the approximations in the method of Olbers (and not as a result of calculating errors) the method of Carlini can, in certain cases, considerably decrease the divergence in J, and it will generally indicate how one should change M in improving the orbit. However, wo recall that even though the computed second position of the comet may fall right on the fundamental circle, this still does not guarantee irreproach- able consistency of the observed and computed coordinates, or particular accuracy of the elements of the orbit. The accuracy of the elements should be judged only by the representation of the observations that did not enter into the ealeulation of tho orbit. 68. Summary of Formules Ay Initial Quantities thy day Bay Lay Bay tay doy Bay La, Ray a fay Agy fay bay Rae The coordinates of the comet and the earth must be reforred to the same equinox, preferably to the beginning of the year of the observations and not to the normal equinox. If the distances of the comet from the earth are known from the preliminary orbit, it is possible to correct the moments of the observations for aberra- tion and the coordinates of the comet for parallax, according to instructions in Section 1. The coordinates « and 5 are replaced Determination of a Parabolic Orbit 2a by A and 2, according to the formulas in Section 82. A correction for the Intitude of the sun is introdueod in tho latitudes of the comet according to equation (4-12). If an initial orbit is being calculated, it is possible to neglect all of these corrections. Refer to Section 65 concerning the accuracy of the calculations. B. Auitiary Quantities 1©08 0; = cos C08 (4 ~ £4), C08 Oy = C08 f2g C05 (Ay ~ Ls), 0, < 180" and 45 < 180°. If the values of cos 6; and cos 95 are close to unity, we caleu- lato according to the formulas: ay sin 6; sinw,=sin 8, — sin 03 sin ws = sin Bs, sin 6, cos wy = cos Bi sin (Ai ~ Ls)s sin 03 cos wa = cos Ba sin (iy ~Ls)- ‘Tho values of the angles 1 are not neodod. Control: sin 8 and cos @ must correspond to the same angle. 9.008 (@ ~ Ly) = Ry cos (Ly ~L1)- Ray gsin (G ~Ly)=Rs sin (La ~Li), ay 9> 0, tan Ba ne en a _ av) tan l= Ss = La) i) C, First Approximation u ta sin Bs etm im tq — 0) 60S By Sin Qs ~La)— sin fy ota If an exceptional case is presont (numerator and denominator of the trigonometric part of M are very small), we calculate tan (I = La) = tan (ha ~ Lea) sec? Fe It is possible to take either of the two values of TI. tan Ba A tnd ava) ye. 2 f2, sin Bs ctr J~ cos Bisin@i=M yay tq= ty" cos By sin (As ~ Il) sin fia ctn J” 24 The Det notion of Orbits feos & cos (H=Ay eos M cos Bs C05 (As Ay) = 008 Bi, in (=A) = cos By sin (As = Ar) Asin ¢ = Hsin By ~sin Bi, wp cos y = cos ¢cos (H = @), A> 0, £< 180° and y < 150°. Ir cosy is close to two ealeulato according tothe formulas: Comtol: sin y and cos y must correspond tothe same angle. ey r=fc08 01, to = 8 00005, f= Zoot xs (VI), T= Rysindy, tee Bin oy, 22 2 sin pe Ry sind, ton Bin ay, 22 Zain x. The fist value ofp is taken at random, Next are calculated att tang = (rs +79) ny, log 2k = 8.586611. Wo obtain y with tho argumont 7 from tables (2, 10, 19]. p, is modified with the application of methods of interpolation until Pa = Moy ax) D. Second Approximation If the intervals of tine are short, and if ry and ry ate not too small, itis possible to get along without a second approximation. If, howover, it is necessary, we calculate: =Apy B=te-Apg B=ty- Ad A= 04005712, log A = 7.7644 ice © Ap, is obtained by linoar interpolation according to the time be- tween Ap, and Apy. Determination of o Parabolic Orbit 248 TREES 63) Tee RS A) Te RE ey 4 Una 4) Aten) : ay Pea Cen ty) In the excoptional case the value of I/ny is also calculated 4 nth) Auten) -$ ARG, (ta) ~———___ 1 fesny cos By sin (As ~ La)~ sin Bs etn [Banos by hat In the exceptional case 7 yt, sin Ba etn J - cos 81 sin (Ai ~ M1) ing 008 By Sin (Ay —M)— sin By ctn J 1 © com Ba sin (ts ~ II) — sin By etn J (Il) EE By sin (= L,)~ Ly sin (0 ~ La) + Ry sin (~ wa) In the second approximation (1) = M+ m/p, is taken in place of M, where p, is the result of the preceding approximation. Next (VI) ~ (X) aro calculated. B, Derivation ofthe Rlements +, 008 5,008 (I, = b,) = 9,08 f, cos(a, = Li) + By fata, QI riteoa Staion) piece tag) (pe aio 1, sin}; = 9, sin, Control: The values of ry and ry must coincide vith the results of the last hypothesis 244 The Det ination of Orbits tan i sin (Fy —St) = tan br, tan éc08 y= = 22.88 tan 81 cos (Is sin (@ =) Conteol: tan i sin (ly ~ St) = tan By Wty — 2, > 0, 4< 90", and if fy = 1) <0, > 90°. 1) p XIV) tan 8, tan us aa sin cos ()— 2) an (=) 800 to G=1,8 KY) It is possible to make use of either part of the formula, according to which is groater in absolute valwo—cos i or sin i If b> 0, 0 Neg lecting the mass of the body m in formulas (9-3), we shall write the ‘equations of perturbed motion in the form (9-12) Numerical Caleulation of Perturbations 9 and analogously for y and 2. In what follows, we shall write the equations for the coordinate » only. For unperturbed coordinates, in accordance with (3-78), we havo consequently, the differential equations for the perturbations as- ‘sume tho form wen Be | @E dr deo we” a” a + (18) since In these equations, m; is the mass of the perturbing planet, furthermore Panteys a3, ph = (2, ~ 2)? + (y, 9)? #2, - 2)? rBaadey2s 22, Ory rPeadey2+ 23, ‘The first corm in the right-hand part of the equalities of (9-13) is the same as that in the equations in Cowoll’s method. ‘The perturbed coordinates 7, y, 2 enter in it, but, just as there, they fire multiplied by small factors of planetary masses. Therefore, as ‘first approximation, the unperturbed coordinates 2, yoy 2) cam be introduced. The Second torm represents the difference of two large quantities, difforing from each other only by a quantity of the same order of sinallness as the perturbations themselves, as it is not difficult to see. For the calculation of this torm it is neces- sary to know the values of the perturbations. In ordor to reduce it to a more convenient form, we shall write the expression in brackets in such a form: -9)-J. ow Since = (By + 6)? + (yg +n)? + (25 + 0 Br Olle, + LEVEL (H+ Ennely + SLIT, a0 The Determination of Orits wo can write Berea, (8 where gu COL EOE (iy + (8-17) consngesy Beeagrtetaty 3.5.7 eve tata, (18) ras where (9-19) and this quantity can be tabulated according to the argument @ [32]- On the basis of formulas (9-15) and (9-18), equations (2-13) as- sume the form: Be (fqe- 6). (@-20) For numerical intogration it is moro convoniont to carry out a multiplication by w?, where w is the interval of integration. The result, writton for all perturbing planets whose action is taken into account, will be thus: no Bette, oat 4) + A(fge~€), (8-21) PE " where eae (9-22) For the calculation of perturbations by this method, the unper: turbed coordinates 25, Yo, 4q of the body ace First found by using well-known formulas, and with them the first terms of equations (0-21) are caleulated. Neat the opoch of oseulation one can neg: lect the perturbations in the first approximation, since here they m ue alvaye sll, Fo yay tp = Es ee to take advantage of the data in the “Planetary Coordinates” or the “*Asteonomical Yearhook of the USSR." If for some roason the: data are not available (for example, for calculations referring to a sufficiently remote past), it will be more convenient to work, not in equatorial, hut in ecliptic coordinates, obtaining the coordinates of planets from the obvious expressions 2,71, 005 b, 008 2, a1, £08 B, sin I, 0-28) 2,=1, sin by by using the heliocentric coordinates r, 1, 6 of the perturbing planets given in astronomical annuals. Those calculations are performed for the dates a-2w, a~w, a, a+, a+ 21, which aro sym metrically disposed relative to the epoch of osculation a. Thon, by the samo formulas as in Cowell's method, wo intogeato pro- visionally, on a separate sheot, the equations (9-21), neglecting the second terms, which gives us the values of the perturbations for the mentioned dates. Hero, in the formulas for the initial terms of the series of sums, & £% ete., appear instead of 2, 2’, etc., and they will, of course, be equal to zero for the epoch of osculation. With & 7, ¢ found, formulas (9-21) aro calculated in a second approximation, this time with the second terms, and the integration is ropeated which usually brings the final result and further ap- proximations are not required. After this it is possible to calcu- Inte, one after another, the following (or procoding) intervals; the values of & n, ¢ necessary for obtaining the second terms of (9-21) aro each time obtained by extrapolation. Oppolzer proposed a method by means of which one can elimi- nate the extrapolation of £, 7, ¢ and reduce the calculation almost to a direct process. Howover, this method appreciably lengthens tho work, and in ordinary ensex it eannot be recommended for the simple reason that it makes sense if the extrapolation of & ns ¢be- comes unreliable, and it is precisely this that should be avoided in every possible way. Indood, if the values of the perturbations fare extrapolated unreliahly, this means that their differences change irregularly, and that should not be pormitted, because a smooth trend of the differences constitutes the most roliable guar- anteo of the correctness of the calculations. 22 The Determination of Orbits If the calculations cover a sizable interval of time, the pertur- hations start to increase strongly and with them the second terms in (9-21). Sooner or later the necessity arises for a transition to the osculating olements for some new epoch, which requires the expenditure of Labor and produces a discontinuity in the calcula tions favoring the appearance of orrors. ‘Therefore the method of Bond-Eneke can be recommended only for a rolatively short caleu- lation, for example, for a single appearance of a comet. Hero, it tums out to be most convenient in practice, since the caleulations of perturbations can be conducted, with the same absolute pro cision with a far smaller number of decimals than the calculations of the perturbed coordinates themselves in Cowell's method. This last method is also inconvenient if it is applied to a periodic comet, that comes rather close to the sun (nearer than 1.5 astronomical units). During the approach of a comet to perihelion it is useful to abandon Cowell’s method and to replace it with the calculation of porturbations of the rectangular coordinates, for which a transition to osculating olements is required. The perturbations £9, ¢ which are obtained by the method of Bond-Eneke are usually introduced in the form of corrections to the unperturbed rectangular coordinates given by the ephemeris, after which the ephemeris can he compared with observations. If it is rable, one can determine the influonce of the perturbations directly on the geocentric coordinates « and 8. According to the well-known formulas (3-11) for geocentric coordinates in which & m, and £ have a meaning different from that which was attached to them in this section, we obtain P COs B eos d= zy HK +E, py COS By 005 Hy =2y + Xs peosdsina=y,+ Van, po cos 5, sin tg mye +¥,b (9-24) psin=2,+24, py sin By = 29 +2, where «i, and 5, are the unporturbed and c and 6 the perturbed co- ordinates. From this we shall find C08 5 c08 x ~ p. cos 5, cos X= £,|-sin x | ~sin 5 cos a pcos 8 sina —p, cos d, sin ay =9,| + cos a} - sind sin a psind—p,sind,=¢.| 0 |+eoss Multiplying by factors written to the Fight, adding, and assum- ing da = 4~%, d5=5~8, whore those quantities may usually be considered a8 differentials, we shall have Numerical Calculation of Perturbations 73 Po C08 8 Sind =~ Esina +n cos 4, Po (cos 8, sin 5 cos da ~ sin 3, cos 8) Bsin a+ cos 6 = py sin dd = ~ Esin 8 cos a ~ 9 si ot, with sufficient precision, =£ sin an cos a eos 5 da e pare Y ; ae 5 88) gar 2 Ein cos a—y sin 8 sina + ¢ cos 8 peel If the calculations were conducted in ecliptic coordinates, the perturbations should be transferred to the equatorial system ac- cording to the obvious formulas eee, weose-¢’ sine, (9-28) San’ sine +l’ cose, where £’, 1's ¢” ate the perturbations of ecliptic coordinates. Jn those cases when it is necessary to calculate now osculat ing elemonts, this can be dono just as in the corresponding section on Cowell's method, where aeaysd, sensed, (9-27) ete. The unperturbed coordinates and velocities are obtained from the initial elements by moans of the usual formulas, and £9, ¢, £7,115 ¢7 aro found by the formulas of numorical integration. ‘The calculations according to the method of Bond-Encko aro most conveniently conducted with an arithmometer, although it is also possible to calculate with logarithms. The distinction will be ‘only in the expressions being utilized for obtaining ty Yo, Zo: Here it is sufficient to refer to Section 24, whore all the necessary instructions were given. 75. Summary of Formulas for the Method of Bond-Encke A. Initial quantities As a rule they are the elements of the orbit; however, the un- perturhed coordinates 24, yo, 2, can also be given directly, if the contemplated interval of time was included in the calculation of the unperturbed ephemeris. It is almost always sufficient to con- duct the calculations with five or even with four decimals. The m4 The Determination of Orbits interval for minor planets is usually equal to 40 days, for comets it is 20 days, and only in rare cases (relative closeness to the sun oF to one of the perturbing planets) is it shorter. For an ellipse: MaMy + u(t-t)aE-esinE, e7 = 206265, 6° = 57.2958-6, 29 =A, (cos F ~e) + By sine, a Yq = Ay (cos Ee) + By sin Ey 29 =A, (cosh ~e) +B, sin BE. For a parabola: «ris takon fom tales with tho angament M. ty =a, (1-07) + bo, Yo - 4, (1-07) + bya, (a) 29 + a,(1-0%)+ bo For nearly parabolic orbits, see the instructions in Section 24. Bude 33. Control (for a parabota): ro = q(1 +02). raat ayty at, Pp = (ar 2)? + (yy ~ y)? + (2, ~ 2)? B, Fundamental equations oe 1.20? SE are ndae- Os a Got FOE+ Wot eMat Got FOE with the argument g is taken from tables (13, 15, 19, 80, 92, 26). Numerical Colculotion of Perturbations 25 "hae - Lh + eh AY (a) + oo nee 7) ay 1 La 10(a) 4 2D yy (ay fielos SAM) - Be pays ME pra... In tho first approximation xo is substituted here in place of 2, etc., and these calculations are made for the dates a ~ 210, a~ 1, 4, 2+ w, a + 2m, neglecting the second terms of /,. In tha second (and usually the final) approximation the & m, ¢ found according to (V) permit the second torms of fy, fy, fe to be calculated and to proceed toward the final construction of the integration schemes according to the same rules that wero given for the method of Cowell. The preliminary integration is mado on a separate shoot. The valuos of n, ¢ necessary for the caleulation of subsequent datos are found by extrapolation. C. Perturbations of coordinates and of velocities = a(a sha) + FE fy(a + kw) 4 fa + kw (a + fae) = 6 agg e+ Ho) + gag + be) se avy f= "tela + bo) LA (a+ be) 11 60480 + AM a+ kw) - $Y (a + bo) + vee If the calculations were conducted in ecliptic coordinates and the teansition to an osculating orbit is not to he made, ca enone -(osine, “ Can sine+0/ cose are calculated, whero ¢%, 9/, ¢ designate the perturbations in ecliptic coordinates. D. Deduction of elements Usually the osculating elements are not calculated, but if they are needed, they can be found by means of: 26 The Determination of Orbits wk, % (B, cos E ~ A, sin E), ava . wk B, cos B= Ay sin £), wn B, cos E - A, sin E). "Va These quantities are calculated according to the initial elo ments. Using them wo find Beep ky area ee, YRYotn, Ym vo tas aaayed, se aye ce The velocities ace expressed in units of time equal to w days, and they must bo divided by wk before the determination of ole- ments, which is carried out according to formulas (VI)=(X) cited in Section 72. If the perturbations of the ecliptic coordinates were calculated, it is necessary to set €= 0 in the indicated formulas whore the vector elements will be roferred to the ecliptic. How- ever, it is possible to make use here of formulas adapted to loga- rithmic calculations in which the vector elements do not enter. In view of the rarity of a similar case we shall not consider it. E, Perturbations of geocentric coordinates = fsin «+7 cos a p= 10" aro 1” Essin 5 cos a= sin pe i0" are 1 107 sare 1” = 48.4814, cos 5 dat” = vi a+ ¢ cos B oa if the perturbations, as is usual, are given in units of the seventh decimal. 76. Example For the petiodic Comet Brooks it is necessary to calculate the perturbations of rectangular coordinates for the period of visibility of the comet from June 6, 1939, up to February 1, 1940, taking into account the action of Venus, the Earth, Mars, Jupiter, and Saturn. The following system of elements is takon as tho osculating orbit: Numerical Calculation of Perturbations am Epoch and osculation 1939 Sept. 24.0 U.T. Mo 1913.05 Qo 20 4 'er i 5 ‘5100 As $3.5424 B, ~0.7375) 4S oor 8} Fa'e4so} 1050.0 Ay force BY Folssas Oise eo The unperturbed coordinates of the comet were calculated at ‘once in 20-day intervals for the whole required duration. Here, the calculations will be given only for five datos. Note that only for the dates most remote from the epoch of osculation do the porturba- tions have an influence on one unit of the last, the fourth, docimal of the coordinates. Therefore in subsequent calculations, as this is often possible in similar cases, the unperturbed coordinates can be used in place of the perturbed coordinates in the first part of the formula for f,. At first, as shown above, the perturbations were integrated for the four datos @ ~ 2w, a~"w, a, a+ w, where the ex- pressions (3-77) were taken for the determination of the initial terms of the series of sums, since the epoch of osculation falls on 2-11, Thon the second torms of fy were obtained for these dates and tho calculations were continued in the preceding and the fol- lowing intervals. Dato 1939 Ang. 25.0 Sept. 14.0 Oct, 4.0 Oct. 24.0 Nov. 13.0 I w 012.89 299781 sPaTisa seatira e 0253 5°57 1048.8 15 56.7 sin E 0.1034 0.0074 + 0.0888 0.1853 + 0.274 conf 40.9948 + 1.0000 to.9960 $0:0830 + 0.9618 com#—e 40.5085 0.5190 +0.8000 + 0.4060 0.4754 % S1S7I8 418350 1.7408 + 1.6250 + 14815 u $0109 $0'3990 Solera Fols408 S Lt027 2 $0,066 Sol1396 Fo.at04 +0.2074 | Lo.370 3 as873 8.5027 B.s277 8M 8.7619 mous 0.18035 0.15255 0.16098 0.14853 0.13700 ’ 0.0177 ocoxene H.av7RR o.0i77D 0.01628 Soft vont agg +1458 1S 410 608.2 4 20.2 : $4108 44.93 48.60 FaT.IT 4 86.54 1 3.0000 "3:0000 "3/0000 "5.6000 "3.6000 fe 418.2 412,70 41M 48L8 4178.8 Jary-€ 4188 Bek Agazy-€) 42.72 4027 40.20 41.10 +120 x 473.08 $60.60 462.26 +5208 + 43.81 h $75.75 $60.8T 48245 LOS 44TH 778 The Determinotion of Orbits Date 1939 Aug. 25.0 Sept. 14.0 Oct. 4.0 Oct. 24.0 Nov. 18.0 MW favy—n 2 46 +9 +91 +380 AGauy =n) $0.39 F017 0.18 BL + 4.08 ? fain frost“ teaa 19.82 iy Soo Ses? Saae1 sa “15.56 faeo-o 48 + +3 430 8 AUqzg~ 2) 40.18 40.05 40.05 4053+ 18 Zz e308 Sse “set 6.31 he AM 588-8. 8d 408 For these dates wo shall sive the calculations of the perturba- tions by Venus: Wo aye 245 2.58 nae naw 4027 20.38 ass Fors 0.08 = 0.68 61 66 8 on OT 0.20 047 0.48 0.20 $452 $5.00 44.92 42.76 -0.16 +405 481T FA.8 0.45 $0.05 -0.08 = -017 0.28 0.37 7206 0.04 +248 ded + 5.00 -2.91 0,90 421 kts + 468 $0.03 0.01 0.05 0.09 0.14 164-048 40.80 184 42.28 2161-049 F075 FETS RE ‘Those calculations could also be carried out for all the planets and for the whale considered duration immodiately after obtaining the unperturbed coordinates of the comet. We shall also give the integration scheme for z and the calculation of the perturbations in geocentric and 5 for a single date. 1939-1940 "te “te fe GP TILA June 6.0 +1146 + 1138.19 + 75.56 =388.46 + 4.89 26.0 + 787 + 750.73 + 80.75 3.48 ~s0171 +t Numerical Calculation of Perturbations a 1939-1940 Rh % tf ff TLAV July 16.0 4 450 + 443.02 +82.16 295.55 1.92 Aug. 8.0 4 294 4 9ITAT 80.04 = 145.31 = 40 28.0 4 78 + TRE $78.18 = 09.56 ~ 5.88 Sopt0 + 8 + 260 +00.87 + 081 ~ 1a oct 40 4 8 + 281 60.46 + ant = sat 20 + 10 + 08.68 + 54.08 + 116.82 = 9.04 Now. 13.0 + 186 + 162.50 eam 4161.83 = 10.36 Dec, 9.0 + 947 + 344.08 sad.as +018 + 195.88 10.18 28.0 + 642 + 599,01 wou 41.35 +220.08, = 8.89 Jon. 19,0 + 761 + 750.06 +15. 420t +288,00 = 6.86 Fob. 1.0 4 996 + 905.85 7 + 8.08 1940 Jan. 4.0 va é sore 1.5118 280 aout g on 73.8 sina 0.8704 Ta cosa OSH lait sind FOL5186—umer, 199 fos8 —¥0,9928—danom. 78,3, sin 8 cos 2 = Yee sia 5 sin The differentials da and d5 ean bo added to a» and 8 as given in the precise ephemeris, after which the calculated perturbed positions can be compared with the observations. This is the usual purpose of the calculations of the perturbations according to the method of Bond-Encko. In tho given example, the trend of the differences is rot sufficiontly rogular. ‘This is connected with the uration of dhe interval which is a Tittle Tong for Ue two inner planots. However, it is not worthwhile taking a shorter interval, bearing in mind that the perturbations were sought for a duration of less than a year. The precision of the calculations, 0.01 of a unit of the seventh decimal, does not completely hold everywhere. Sometimes an error two or threo times greater was tolerated as may be ascertained by analyzing the figuros of the example. To pursue absolute precision did not make much sense, since the errors in- troduced in rounding off the sums of the numerous quantities enter- 280 The Determination of Orbits ing in f., ote., can give almost the same resulting error. All of this had no importance for this problem. 71. The Method of Veriation of Elements Studying the solution of the problem of two bodies, wo have seen that the motion of any body relative to the center of the sun is detormined by moans of six arbitrary constants which specifi- ally can be the elements of a Keplerian orbit. At any instant of timo tho elements of an orbit uniquoly determine the coordinates of the body and the components of its velocity. Conversely, the ole- ments of an orbit. can be found if the coordinates of a body and the velocity components aro known for a certain moment. If, besides the action of the sun, the attraction of the planets of the solar system is taken into account, the motion of the body will no longor take place according to Kepler's laws. Neverthe- loss one can consider that the body moves along an orbit which is conic section, the elements of which continuously change with time. At any determined moment (the moment of osculation) we shall have a system of otoments determining an osculating orbit for which tho coordinates of the body and the velocity components, fat the moment under consideration, aro the same as those in the actual motion. ‘Thus, one can say that the osculating orbit is in contact with the true trajectory of the body at the moment of oscu- lation, The method of variation of elements which was proposed by Lagrange consists of replacing the elements by variable quantities to satisfy the differential equations of perturbed motion of the body. The differences between the values of the elements of the ‘orbit for two differont epochs of osculation can be called the per- turbations of the elements. If it is assumed that, starting with the moment of osculation, tho perturbing forces ceased to act on the body, its motion will take place along the osculating orbit. Since the coordinates of the body, and the velocity components, in perturbed motion, have at the moment of oseulation those same values which arise for the osculating orbit, then, evidently, every relationship appearing as a consequence of the solution of the problem of two bodies, among, the coordinates of tho body, the velocity components, and the elements of the unperturbed ‘orbit, will also exist for perturbed motion. Any element of the orbit (in general, any constant in the solu- tion of tho problom of two bodies) is a function of the coordinates Numerical Coleulation of Perturbotions 21 of the body and theie derivatives with rospect to time for a certain moment. For example, = ale, ys 22", y's 2"). (6-28) In perturbed motion, on the basis of this equality, we have dada, da), 9a, da ds’ da dy’ aa da” ae” a" * ay" * a2" * ae ae” By Ge * ae" ae Designating the coordinates of the body in unperturbed motion along the osculating orbit by 25, yoy 29, we oblain from (9-28), since @ is constant, (0.29) on ae yg ag 28 8 an G27" ay %* ag °* Gat wa oO) According tothe dfnton of he osclating ott, 29 = 2, ot wes treter ten ° de’ dx peace way where j, is the projection of the perturbing acceleration on the z-axis, and correspondingly for y and 2. Therofore, subtracting (9-80) from (9-29), wo shall find da_ a, , da, | da G7 batt * She * Sooke (0-32) This shows that in the search for the variations of elomonts of the orbit produced by tho perturbing accelerations, the coordinates 2, y, 2 of the body should be considered invariable, and differenti tion with respect to them is unnecessary. We orient our system of rectangular coordinates in the follow- ing manner. We shall diract the z-axis along the radius vector of the body at the moment of osculation ¢, the y-axis in the plane of tho osculating orbit in the direction of motion of the body; and the z-axis toward the north pole of the osculating orbit. We shall now tuen to the integeals of areas (2-81). Let us as- sume the mass of the body m in them is equal to zero; we shall dosignate by » the inclination of the perturbed orbit to tho plane of the orbit which is osculating at the moment ¢, and by II the longitude of the node of the perturbed orbit on the osculating orbit, moasured from the direction of the z-axis. We shall have 282 inotion of Orbits = ere kyp sino sin, oA cg = =kyp sino cos Ml, (0-33) cg = kypoose. We differontiate these equalities, and then we assume, as fol- lows from our definitions, that 2 =*,y=2=0,0=0,l=0. This will give Fi (aay iy Je we te. (9-35 ae = ‘Then, on the hasis of the second oquation of (9-34), we shall find da ae (0-38) We seo that, on account of tho porturbing accoloration por pondicular to the plane of the orbit, this plane rotates around the radius vector. In Fig. 22, representing the heliocentric celestial Numerical Colevlotion of Perturbations 283 sphere, P is the position of the body, T is the ecliptic, GP is the projection of the orbit, da is the angle of rotation of the por- turbed orbit, &P is tho argumont of latitude of the body, w=» +o On account of the rotation of tho plane of the orbit, the node is displacod to point WV’. Thus, 8" = dK, moreover Z2H'P = 180° = (644i), UP = udu. Brom the triangle S15UP we derive sin (+ di) sin di = sin w sin do, 0s (i + di) = cos § 08 de sin i cos w sin do, 08 (uw + du) = c08 u cos 421 + cos i sin u sind, or, after simplification, an dix cos nde, sin w ese i do, du = ~ cos idil ~~ sin w ctnido. Substituting (9-36) hore, we shall obtain, an. wt ey sin u ese iW, a ai Fer eos uly, (9-31) du de dog, du te, 48s sin ws otnaW,. dt dt” dt cee ‘The third equation of (9-34) gives in connection with (9-35) dp Ee (9-38) Now we make use of the vis viva integral (2-28): from which it follows 1.204 (2) (y (2) Nae ae) *\aa) J We shall differentiato this equation, after which we assume on kVp dz tho basis of (0-08) 2. EVP #2 9, and on the basis of (3-85) aa 284 The Determination of Orbits te dr k sin g sino seus (0-39) aa we ‘SUI bearing in mind (9-35), we shall find SE nat (xin sin ess + 211) (940) a 7 From this, in conaequonco of k us a - the equation aoe , de 8 bing sin os, +27) o-4y = Eline si rn) oan) is obtained. But since p = a cos® ¢, then dp ga Pecot et ag Basing cos gS. in aa Daveraiag ton is “2, we have by moans of (238) at (9-40) i at « foes # sin 05s “€ on a) rn]. Having taken advantage of the relationships 2 iy sin g cos, sin gcosB, (9-42) {8 a cond lsin 98; + (cos » + e08 B)Tile — OH43) From the first equation of (9-42) it follows that as a dy 1 cong conv = sing sin 2%, rade cose Ang sine 1 dp pad tn & etn ee es aoe de Fsingainy at" Numerical Coleulation of Perturbations 285 cos? » ~ cos v cos H ) (9-44) For simplification, we shall apply the first equation of (9-42) and Feos v= a(cos £ ~ sin ¢) (cos E ~ sin §); this will give I~ c08 v cos Fo t= TOO POT on cosy ? = (1+ sin g cos v)(1~ sin g cos v) -cos* g cos v] = "sin? ve ? . P Aftor that in place of (9-44) we shall have a ; Se = gh ple cones, - (+ p)sin vm. (048) the last equation of (9-87), we find rep + TAP sin vr, =r etni sin wits. (0-46) sing 7 in 46) Finally, we shall take the equations Ma My + u(t to) =8 sin g sin Ey r= a(i~ sing cos), where fo is tho initial epoch for reading off the mean anomaly (generally speaking, not coinciding with the moment of osculation 1), and we shall differentiate them, considering the coordinates of the body as constants in the same way as the time ¢ entoring ox- plicitly. We shall find fiM\ dM dw + (de dg) a) ae HG tay = 2 () - cos g sins 2, (@ dt dt ay @ a) Ms dt ds SOM 1 da a6 0 easing sine (8) -acospoost EB casing sine (1) -2eose 286 The Determination of Orbits where (2!) and a time, entoring only by means of tho elements of the orbit which change as a result of the action of perturbing forces. We shall and then by means of (+40), (049), (2-48), and fae i GE) designate the derivatives with respect to Gi) ma rs atininate ( at equations: rainy = a-cos d sin, +08 v = a(cos & ~ sin 6), ‘we shall perform furthor transformations: = etn J (etn E - sing cos ctn sin sine) a non PaaS FT wetn g cos goin o ome a = =@r eos gp otn g cos 0), - 24 (2 _ cos? w = c08 » cos E)T i. We shall substitute hore a, 1- cos» cos & = sin? vy P eee teat e zi ‘The complote derivative of M with respect to Lime is = @r cos & ~ p etn 6 008 v)Sy (9-48) =r nome sin ely. aM ae” Numerical Calculation of Perturbotions 287 Integenting any moment ¢ is exprossion, wo shall find the value of if for eres f(Bt) ae fee amy Here one should inteoduce the expression for p, which is given wate note—ead+ ff (36) a en) te hee While all the rost of tho olemonts will bo found by simple ine logration of the expressions obtained above, M will contain the double integral depending upon the pecturbations of jw. We can con- sider the single intogral ontering in (9-50) as a term determining the perturbations of the mean anomaly of the epoch (since it does not roflect an explicit dependence of M on time) and in agraoment With this we can weite honeo egrets oes fi fie, wan snl in maa (0-48) v0 an cnt singly 2 ptacnot (4) This formula is inconvenient for small eccentricities, and it is thorofore useful to calculate, in place of M, the perturbations of the mean longitude L = M + and in agreement with this to intro duce the perturbations =o +2. Atfieat we shall find feom (9-87) and (9-46) dn FE =~ pose d cos v8) +(r= posed sin yy ee r + rtan Fi sin wWy, (9-52) fand then feom (9-48) and (9-52) 288 Tho Determination of Orbits 1 7 cos gi + p tan 4 cos v} Sy 1 1 +O sp) tan 5g sin eT) er tan Si sin uW,. (9468) In this oxprossion tan 4.4 was entered in placo of etn }, which will decrease the influence of the corresponding torm for @ < 60°. dn ‘Tho last term in 5 js similarly changed, and here also for ¢ < 60° Gp is similarly changed, and here atso for é < 60 dos we shall obtain an advantago in comparison with However, the terms with ese @ in (9-52) remained. One can eliminate them by calculating the perturbations for the quantities ain g sin « and sin § cos w or sin g sin x and sin cos z. Ibis not difficult to ako the corresponding transformation, but the application of the resulting formulas is loss convenient. Therefore, itis recommended to rosort to the elimination of g and a (or w) only in the case of a very small ecconteicity, and we shall not discuss this operation. Similarly, it is not necossary to dotive the formulas for the perturbations of the elements g, ©, 1, which are used for nearly parabolic orbits, since for comets moving along similar orbits, itis much more advantageous to calculate the perturbations in co- ordinates. We shall show how to find the components of the perturbing ac ‘colorations je, jys jz» In our system of coordinates 2, y, 2, the c0- ordinates of the perturbed body will be r, 0, 0; we shall designate by is nis & the coordinates of the perturbing planets (there can be several of them). Then on tho basis of equations (9-3), we can weite (9-54) Tho heliocentric coordinates of the perturhing plants, ri, 25 44, referred to the ecliptic and to the normal equinox, can serve for Numerical Colculation of Perturbations 289 the caloulation of £1, mus gi They are given in astronomical year= books and in “Planotary Coordinates.” Let Li, Bj be the corresponding polar coordinates referred to the plane of the othit of the body whore the longitude is measured from the a: conding node of this orbit on the ecliptic. We shall turn to Fig. 23 in which P; is the position of one of the perturbing planets, "Pl is the ecliptic, SUC is the projection of the orbit of the body, P is the pole of the ecliptic, and @ is the pole of the orbit. We find, considering tri- angle PQP): cos By cos L; = cos b; cos (I; ~ Sl), sin B= cos i sin b; ~ sin cos b; ain (% -9)-J The polar coordinates of the perturbing planet in the system yy 2 will be L, = u, By, thoreforo, f=: 608 B; c08 (L; ~ 0), ne =r; cos B, sin (Ly =), (@-58) en sin By. The distance 9, ean be calculated by the formula ala (G an andeGrerter?-2rk, (987) convenient with the use of an arithmometer. If logarithms are used, one can solve the following systen: p, cos 8, 008 8, = &- 1, 1p, 608 8, sin 8, = 1,5 (-58) p, sin B,= G5 where the angles 9 and 0 will not be required in what follows. 290 The Determination of Orbits In the final formulas the following notations are used for al breviation: (0-59) a et (0-00) EP ae pain 7 Wy ie Rp we" where the factor ——! of the perturbations of the elements in seconds of are, bearing in mind that the calculation of perturbations is over an interval of 1 days. By moans of equations (9-54) and (9-89) we find (« -) is introduced in order to obtain the values = ° gre 17 Ener) erie (0 Tho factors wk’*m, are given in soveral tables [13, 15, 19, 32] for the perturbing planets. For the integration of perturbations, the formulas which wore given in Section 25 ate applied to the First and secund integrals. In boginning the work, the values of the perturbations for several (usually four) moments enclosing the epoch of osculation are caleu- Iated by means of the unpertutbed elements. Near the epoch of osculation tho perturbations are small and the necessity for a sec- ‘ond approximation almost never arises. Later on the values of the perturbations Aé, etc., found by the formulas of numerical intogra- tion, are added to the initial osculating elements, and in this way the ‘perturbed elements for all successive moments are obtained. Numerical Calculation of Perturbotions an Proceeding to the calculations for the next moment, it is necessary to know the perturbed elements beforehand. Under ordinary cireun= stances (not too close an approach to the porturbing body) the elements can easily be exteapolated not only for one but also for two or three intorvals without any loss of reliability of tho rosults. It romains to be statod that for the ealculation of L. the following formula is used: Leternt-tos Ate efftfae, (ay In this way, the perturbations are calculated by the method of Variation of elements, if it is required to obtain the greatest pre- cision possible. However, the method of variation of elements is equally effective in the approximate calculations of perturbations, which are used in extensive works on minor planets or for pro- calculations of the returns of periodic comets. Apart from certain simplifications in the formulas, in these cases one can make use of the invariable elements over a long period of time, which is especially essential. Such a method leads to the accumulation of second-order errors as far as the perturbing masses are concerned. For higher requirements for precision one should change the ele- monts in the course of the calculations each time, for example, each year, when the perturbations reach appreciable values. This ‘causes jumps in the trend of differonces, extremely harmful for the conttol of calculations by differences. Stracke [36] advisos the following procedure. First, calculate approximately the porturba- tions from the most important planets (Jupiter, also Saturn, in an extreme case). ‘Then, make use of the perturbations obtained in order to conduct precise simultaneous calculations for the entire required duration, calculating the soparate quantities along lines, fas is done in the calculation of an ephemeris. This is much more convenient than to calculate one interval after another. Moreover, tha double calculation of perturbations provides a certain conteol ‘on the largor errors. Thoro are also other ways to facilitate the calculations by the method of variation of elements. One can introduce as the indo- pendent. variable some other quantity in place of time, for example, the eccentric anomaly, altering corresponcingly the expressions for the derivatives of the elements of the orbit. This achieves a more uniform distribution of positions of the hody along its orbit, which is important for strongly eccentric orbits of periodic comets; for equal intervals of time, the positions near perihelion arc far distant from oach other, as a rosult of which the perturbations 292 The Determination of Orbits change with insufficiont rogulatity; on the conteary, near aphiolion suecessive positions are disposed too densely which leads to un necessary work. Moreover, with the eccentric anomaly chosen as independent variable, itis possible to tabulate for integeal values of E the soparate coefficients depending upon the position of the body in its orbit, and oceuering in the expressions of the perturba- tions of elements. Such a tabulation can he made especially for ‘every object. General tables also exist for cortain coofticionts according to the arguments of the mean anomaly and eccentricity. For the use of £ or las the independent variable, the positions of the planets will he needed for dates other than those for which they are given in the ephomorides, and the necessity of intexpola- tion noticeably lessens the advantage of tabulated coefficients. Heretofore, it has heen recommended, oceasionally, to calcu- late the perturbations from each perturbing planet separately, i.e., to find S,, 7, W separately for each planet and not to sum thom. ‘This was justified in that for such a method one can subsequently correct the adapted values of the planetary massos, if it becomes apparent that the adopted masses are insufficiently precise. To do lis, it is sufficient to multiply the perturbations obtained by the ratio of the new mass of the perturhing planet to the adopted mass. One can also attompt to find a corroction of the planetary masses froin the comparison of a long series of observations of the body vith calculations of its perturbed motion. At the present time the masses of the perturbing planets aro known much more reliably than in the last century (although not with all of the desirable pre- cision) and it is practically usoless to search for their corrections by starting from the perturbations of comets. Minor planets can sorve for this purpose only in special casos: if the orbit is such that the perturbations are significant, if there aro many good ob- servations, etc. This problem goes beyond the limits of our sub- ject. ‘The method of varintion of elements can be used, in almost the identical form both for the calculations with an arithmometer and with logarithms, but the character of the formulas is such that they sue easier to calculate with logarithms. 78. Summary of Formules for the Method of Variotion of Elements AL Initial data The initial osculating elements of the orbit must be reduced to the normal equinox for which the coordinates r4, 5, B, of tho por- Numerical Caleuletion of Perturbations 27 turbing planets are given in yearbooks and in ‘Planetary Coordi- nates." The dates must coincide with those which are adopted in these ephomorides, i.e., the whole part of the Julian day must be divisible by 10, 20, 40, depending on the interval of calcula- tions. It is recommended to take Julian days divisible by 40 for the epochs of osculation. In precise calculations, tho valuos of the perturbations are computed with a precision of 0.001 for all olements except 4 for which the perturbations are found with a precision of 0”.0001. The derivatives of the elements and the perturbing accelerations from Jupiter aro calculated with five deci mals, while for the rest of the planets, bearing in mind the pres sion with which thei massos are known, four decimals are suf cient (with the exception of an unusually close approach to tho perturbing planet). For minor planets it is usually sufficient to calculate the perturbations from Jupiter and Satur, if the highest precision is not attempted and if tho planet does not have an un- usually eccentric orbit. Here one can take 40-day intervals. If the effect of inner planets is taken into account, which, in particular, is advised for comets, then the interval must be shortened to 20 days or less. The effect of Mercury, and also of Uranus and Nop- tune, is little noticeable in the majority of cases. In approximate calculations, four and even three decimals are sufficient, and only the perturbations from Jupiter, or from Jupiter and Saturn, are taken into account. Usually the calculations are conducted with loga- rithms: MeL-eaLo-ntplt~to)s E-esinE =, = 200265-6, o”* = (5.814425) sind, 5T.2958- 6 [1.788123] sing, om eer eel ia rsinv=acosd sink, 7 cos »=a (cos £ ~ sin ), ? sonven-2y cos? 4. For the calculation of £, vy and ¢ one can make use of tables ((10, 18, 15, 18, 19, 25, 81, 87, 38] and Table 1). 294 The Determination of Orbits B. Perturbing accelerations 08 By cos Ly = 608 By C08 (1s = 2), cos By sin Ly = sini sin 5, + cos i cos d sin (e-2), (Ml) sin B; = cos i sin by ~ sin ¥ cos 6, sin (ty - 2). In the approximate (three-decimal) calculations one can assume cos b= 1, and iff is not great (less than 8°) and eos By = 1, ise, beele= p, 08 9, cos 8, = 1; 008 B, cos (Ly ~u) =r = Gry p, cos B, sin 8, =r, cos B, sin (LL, = 1) = M54 foe p, sin 3, sin B, ay C. Calculation of derivatives of the elements with respect to time =roos ul, ae W Fer sin v ose i, ae i w FP a cos gf sin v-S + a cos g (cos + cos F)-T, ayy dn w FE ~~ pose g cos v-S + (rp) exe 6 sin v-T 1 srtan te sinus 2 ip Numerical Coleulation of Perturbations 295, dt i volt. (27-005 6 xp tan Ea osx) He. (2 cos 8 +p tan £6 cos») ste pytan bin wT er tan Ee sin wel, 2 0 (vy nthe aie con 2 de sing sind Par, a ae” Ve w= 204, Jog Bkw = 0.018733, w= 404, log Sku = 0.914763. In approximate calculations one can often neglect the Last term de alo on the right in the formulas for w 5” and w " Or ae Calculations according to these formulas are carried out for four dates: a-2w, a~ 1, a, a + to, surrounding the epoch of oscu- aa dy Bw by the comm dipeeanae on symbol { for the function, we have, if @ is the epoch of osculation: ai lation. Designating in general —, ignating in general 1 mS - Bl@+ Fame But, ifthe epoch of osculation is @ ~ $1, w 1 a(a-bw). Jt xt 2 ) Broo AL a 7(a- w+ Mal + p Mla 0) ~ greg 2PM — wy + MCA) + os For all successive moments, (I-(IV) and the perturbations are calculated by the following formulas, which also serve for finding the new osculating elements for the termination of the work. In the process of calculations, the values of the perturbations must be extrapolated forward by one interval, which in usual conditions is performed with sufficient certainty. It is even possible to exteapo- late the perturbations for 2 to 4 intervals, which permits the calcu- lation of several dates jointly. 296 The Determination of Orbits If the now opoch of osculation is a +.ne, Ai, AR, AS, Aw, ALg, WA uw ¥(a+nw)~ 4 fas nw) + AE M@ + nw) (a+ nw) ~ ja + nw) PM (ae nw) sees ce A 1 Ste AFC 4 my) f(a 4 me) = se a 4 me) +. 1 new epoch of osctation is 2+ (n+ 2), r (3 AL, AQ, AS, Aw, AL, An ofon(ee dq eve Lm Lg + uglt = why + Halt = ty) + ALy + AL ys =a. The calculations are controlled by the trend of the differences. Ic is advisable to verify the differences not only for the derivatives of the olements but also for several of the other quantities, in p: ticular for r and L;~w. One should beware of errors in the cone stant and tho little-changing factors entering into all the intervals. Also it is necessary to verify carofully the initial eloments, the factors depending on the masses of the planets, and all ealeula- M=L-7, o Numerical Celevlation of Perturbotions 297 tions connected with finding the initial terms of the series of sums, with the integration of the perturbations, and with the determina tion of the new osculating elements. 79. Example For Comet Brooks, according to the same initial elements as in the example of Section 78, it is required to calculate tho perturba tions for the same duration, i.e., fcom 1939 Soptomber 24.0 U.T. up to 1946 Angust 28.0 U.T., taking into account the effect of Jupiter and Saturn. The object of the work was the precalculation fof the return of the comet in 1946, Therefore, the calculations ‘were conducted with four decimals, and, in conformity with jon of 07201 and w St au the perturbations were obtained with a prov with a precision of 07.001. There wore no other derivations feom the precise scheme. The interval was 40 days. Initial Elements Epoch and osculation 1989 September 24.0 U. T. Mo 118" 3.78 we tt oc trast i} i800 ee $a tt co t “s10%eait0 ‘The calculation of tho doivatives of tho olomonts was conducted on saparata sheets; similarly, the cateulation of the perturbing com ponents from each planot and the integration of pottarbations wore Conducted on correspondinaly dosiznated shects. Wo reproduce tho alcalation® for n single date 1030, 1K, 24.0. # si0%e rin 0329 L sears ‘sine 8.9882 : 13 22.9 cos v + conf 0.2865 a 117 2.3, rep 0.6706 + 5528 poose 04a BT rp) sine 0.3585 2.7081 at 8.6852 0.2806 9.4130 298 ain E 208 P cos E cos B= sin @ i sl a 1-2 ain = "008 By coe @1 =) sin by in ain by cos f008 y sin (1) - 0) cos Bh sin La 209 By eon Ls cos sin by msin cos by sin tr =D ‘in By % 9.2646 0.5028 9.9998 9.e903 9.7674 0.2576 9.9784 91488 1755.4 19 540.6 21 538.0 9.92068, 9.74300, ola7ag) 0.4443 % 515.5 117 42. 116, 188 38.2 BATION “1 9.9951 8.9852 S.5Thn 3.9980 731260 orT250 9.17994 9199500 Percy 1600 TOL 0.0948 ‘0.0000 1841.3 0.4348 9.2049 0.4890 o.80TIn S.185%n 0,509.38, ar cos @ poo ste hp sin p sine a Sku rye é a @ @ @ ic w ie we a ») ~) s 1) 3) nr) » ») rn 3) ry we by ain ey =) ry cos B cos Li =») wdride 0.5217 9.8866 9.1748 0.03420 0.1651 021995 1.03700 9.9910 0.7893 o.t361 oua7a9 s.707e 0.60325, 9.5797 9.2090" 019930 0.9292 = 24 54,7 9.62450, 0.6948, 9.9578 0.6524 0.2702 0.81938 0.4139 5.80939, 0.8218, 9.4782 84346 7.9188 8.2780 #0304 BT1a8n 8.5245 5.99758 6.8873 1.9098 19.88, = 11.08, 218.67 = 9158 0.00 =25.25 Numerical Calculation of Perturbations 29 wy, 8.70719 a 11.84 ¥ T1929 ee ua 0 8.78620 wd grat 12.15 desde +0710 ce = 0.486 walde 40 67 Bo + 5.112 = $282 wpe + eas ‘Tho calculations of the components for Saturn aro conducted according to the same scheme as for Jupiter and hore the results face given only in the form S, Tay Ws. On the corresponding sheots every dato occupied ono long column. We give also the heginning of the integration schomes for and L (the integration of the pee- turbations of the rest of the eloments is easter) and the elements for tho now epoch of osculation. In the given example the initial osculation epoch fell on a ~ ‘10 (a rare case) and for the deterni- nation of the initial terms of tho series of sums it became nocos- sary to resort to the costficionts tabulated for similar needs in “Planetary Coordinates.”” Date 19891 14.0 ~ X 24,0 + 0.083 44 XI a0 + 6.49 “4 +10 .268 1940 1 12.0 + 16.417 +3 414.206 H 21.0 +20.623 a2 417 033 IV 1.0 +47.656 at +18 «880 ¥ 140 +68.806 +t if fwd Oe +rer oo 048 17818 0.0 802 -1.020 +0 998 -o.T | F0.2 897 40 101 +167 40.4 =1 010 sit 40.242 418 404 1m 049 +20 +05, Dato 4 ft fi PP Blo ff Lo+nolt—to)b 1939 1X 14.0 1978 $071 O40 1871172 18H X 24.0” 19.75)" 40.64 0.2 0.0 18 51.7 18 51.5 =17.96 4127 XIL a0 = Bs” 1642 —0.8 40.1 28 92.1 24 51.9 ; 38 1940 1 12.0 40.87” W176 -0.440,5 30 12.5 30194 . 09 21.0 $e 14d 0.5 40.5 35 52.9 35 591 x 5 Ww 10 4 Tar P13 0.9 40.8 41 084 41 340 62 HO eso aT AT 300 The Determination of Orbits Epoch and osculation 1946 August 28.0 U. T. Mo 0°27" 6.8 198 36.17 Qo tare ay 1950.0 i 5 3228 4 6 285919 8 509727538 These figures deviate insignificantly from the results of the example in Section 73, which in the first place depends upon the disrogarded porturbations of tho inner planets. The precision ob- tained is sufficient for ensuring the search for the comet, Chapter 10 THE PRELIMINARY IMPROVEMENT OF AN ORBIT 80. Methods of Improving Orbits In the methods considered by us for the determination of pre- liminary orbits by threo or four observations, it was assumed that these observations are separated by small intervals of time, since otherwise the process of approximation which forms the basis for obtaining ratios of the areas of the triangles between the radius vectors of the body would become difficelt or completely impracti cal. As a consequence of the small intervals of time, the peel nary orbits can be computed without the calculation of porturba- tions. As alrondy indicated in Section 70, one doos not usualiy succeed in achieving at once the final improvement of the prelimi nary orbit, because for the computation of the definitive orbit a knowledge of the perturbations is necessary; on the other hand, the perturbations can be found in the majority of cases only by using & sufficiently reliable orbit. Only whon all tho existing observations are confined to a few weeks of months (for example, if we have to doal with a briefly observed nonperiodic comet) can the influence of the perturbations be so small that it is possible either not to take them into account at all, or to find thom with a certain reli- ability by using the preliminary elements of the orbit, On the other hand, an intermediate stage between the calcul tions of the preliminary and definitive orbits is often necessary in viow of the fact that the final improvement of the elements is ob- tained by finding the differential corrections to the approximate values, The validity of this process can be justified in its tum only if the corrections sought are small, ies, if the elements sub- ject to correction aro alroady sufficiently neat to the true ones. Therefore, it is natural to refer the method of differential cor- tection of the elements to the problem of the determination of the dofinitivo orbit (although it is possible to obtain an approximate improvoment of the orbit in this way) and at first to consider the problem of the provisional correction of the preliminary orbit. 301 302 The Determination of Orbits As far as the observations which may be used for preliminary inprovement of the orbit are concerned, there exists an essential difference howween minor planets and comets. For a new minor planet the observations during the first opposition are usally not very numerous and rarely cover more than two to three months. To compare the preliminary orbit with all the available observations presonts no difficulty. If disagreements should arise, it is possi- ble to attempt to eliminate them by computing a now orbit with ‘thor hasic observations. If we succeed in finding the planet at one of the following oppositions, it is possible to rosert to the im- provement of the orbit by the method of variation of geocentric distances set forth holow, and when the planet will have boon ob- served at a minimum of four oppositions it will he possible to pro- ccoed with Finding the differential corrections to the elements, tak- ing into account the perturbations, hoginning with Jupiter. For comets, on the other hand, one usually computes not one but_a whole series of preliminary orbits covering ever ineroasing intervals of time as post-discovery observations accumulate. Here, it is often rovonied that the assumption about parabolie motion is not ontiroly justified. If there are many observations it is possible to combine groups of observations that are near to one another in time into normal places, and to operate with normal places as with observations of higher roliability and precision. For the derivation of an improved erhit it is possible to use the method of vatiation of geocentric distances which in suitable eases will indieate the de- vation of the eccentricity of the orbit from unity. Usually at the end of the period of visibility of the comet the definitive (or somi- definitive, taking into account not all but a majority of the obser- vations) orbit is calculated by che method of differential correc- ions of tho elements. For a nonperiodie comet this closes the cease, but for a periodic comet such an orbit will help to calculate the perturbations up to the next approach to the sun and will give hoth the elements and the search ephemeris. Every new appear- ance of a poriodie comet provides additional material for the cor- rection of the orbit, hut this is a very difficule problem, because the motion of comets, as a rule, docs not agree completely with tho gravitational theory, and tho observed deviations may reach considerable values and they are not of an entirely regular charac- tet. ‘The cause of similar ircegularities, even now, is not clear. Ofien it is found that a new minor planet is identical with some proviously discovered planet, If considerable time has elapsed botween these discoveries, it is impossible to expect that the elements of the orhit for one of the presumably identical planots The Preliminary Improvement of on Orbit 303 represent satisfactorily the observations of the other, To achieve agtecment, and, by the same token, to solve the problem of the identification, it is sufficient to correct only the most unreliable of the elements—the mean anomaly containing a term proportional to time. If, correcting Mf, both coordinates are successfully repre- sented, even for a single observation of the other planet (all the more so, if for several observations), then it is possible to con- sider the identity as established. The empirical correction intro- duced in M in order to achieve agreement with the observations will also permit one to correct correspondingly the mean daily motion js For comets a similar situation is also possible, but it occurs more rarely, because long unobserved periodic comets have, ap- parently, decayed to such an extont that ono should not count on their rediscovery. Only separate cases connected with a particular set of coincidences, such as unfavorable conditions of visibility during the intervening approaches of the comets to the sun, can be discussed. : ‘The number of methods suggested for the improvement of pre- liminary orbits is very great. But at the present time by no means all of them may be considered as of equal worth from the point of view of their effectiveness. Below, only those methods will be investigated which are applied most often and best prove them selves in practice. 81. The Correction of the Mean Anomaly Let the elements of the orbit of a minor planet, derived from observations obtained during the time of one opposition, be avail- able. In one of the following (oF provious—if an object having been obsorved previously is being discussed) oppositions one ot Several observations aro obtained, giving tho coordinates of the planet cy and dq, while the elements of the orbit give, for the moments of these observations, the computed coordinates a. and Se. Depending on the magnitude of the differences ao — as and Bo ~ Be, let us give to Ma larger or a small increment AM, after which we shall obtain with the changed M, for these same moments, new values of 2% and 52. Assuming that the differoneos of the coordinates and the vari- ations AM are not very large, so that the connection between them has a linear character, we shall find (l= ae) t= bo = Me, (BE- Be) T= Bo Bey (10-1) 304 The Determination of Orbits where M+ 2AM represents that value of M which must give agree- ment with the observations. If n observations were made, we shall have 2n equations (10-1) for the determination of 2, which may be solved by ono mothod or another, in particular, by taking the simple mean of the different values of 2. One should not expect complete agreement, because inaccuracy in the rest of the elements of the orbit, and ‘fisregarded perturbations can cause some residual di vergence. Therefore, identification acceeding to one observation always remains not entirely reliable, But if there are at least two observations, the residual divergences in c must. agree between themsolves, and the same applies to the divergences in 3. In that ease tho identification bocomes much more roliable. Furthor it is possible to find the correction to the moan daily motion by the formula (em fo) Awa AM, (10-2) where ¢ is the moment of observation and fy is the epoch of the cloments. If necessary, it is possible to apply the same method for tho periodic comets. If instead of if, the moment of perihelion pas- sage Tis given, then it is necessary to vary this quantity. Simi- lar calculations are usually mado not in the identifications of comets but in ® discovery of an expected periodic comet. The first observation already gives a correction to T and by the same token offers the possibility to improve the data of the search ephemeris. Beating this in mind, computed values of the varia- tions of the coordinates for « variation of T (for instance, by one day) are often given in the search ephemerides, as was already shown in Section 24. 82. The Variation of Geocentrie Distances Gauss has shown that the problem of the determination of an orbit cannot be reduced to finding one unknown, but on the other hand the introduetion of two appropriate unknowns reduces tho process of the solution to a comparatively simple one. For a pre~ liminary orbit such unknowns are a; and na, but in the improvement, fof orbits it is hardly possible to make a better selection than by taking two geocentric distances. The method of variation of geocentric distances is widely applied to minor planets as well a3 to comets. The observations used for the improvement of the orbit can be significantly remote from each other. Consequently, if necessary, the perturbations of The Preliminary Improvement of an Orbit 305 the coordinates or of tho elomonts of the orbit must bo taken into account. However, if the observations do not encompass too large an interval of time, it is possible to neglect the peeturbations. In tho mothod of variation of geocentric distances two entirely reliable observations are first selected, or, bottor still, two normal places that are neither too close together nor too far apart, i.c., belonging to one opposition of the minor planet or to one appear tance of tho comet. The problem is, in fact, to select for these ob- servations such values of the geocontric distances that, together With the observed coordinates, would determine a system of elements of tho orbit agreoing in the best possible way with tho remaining observations. As for the latter, it is not worthwhile to take all the existing observations, if there are many of them, bocause the solution will not be final, anyway, because the basic positions must be reprosented accurately, oven to the projudice of the remaining positions. It is usually sufficient to take the obgorvations referring to the other opposition with limited precision, because the perturbations calculated with preliminary olemonts will necessarily differ from the true perturbations. Having calculated the geocentric distances (;) and (p,) for the moments of hoth basic observations ¢; and ta According to the preliminary elements of the orbit, or having taken them from the ephomeris, we find by using them and the observed coordinates 1, 81, ca, 62 a system of orbital elements ET which procisoly ropresents both basic observations. Generally speaking, complete agreement is not obtained for the remaining observations. In addition, we calculate two other systems of elements EM and TIT using the same basic coordinates and the geocentric distances (©,) + Ap, and (p,); and (p,) and (p,) + Apa, respectively. The basic assumption of the method is the linese character of the con= noetion hotween the increments of the geocentric distances and the variations of tho coordinates. Therefore, Ap, and Ap, must not be too large, and yet they must not be too small either, because then it can turn out that the corrections 0 (p,) and (p,), found as a result of the calculations, will be several times lafger than Ap, and Apa- In ordinary cases, it i most convenient to take 0.001 for the Variations Ap, and Apy. ‘The coordinates X}, 51, a%, 5%, 2M, 51 are calculated with the systoms of olomonts £1, £11, EIIl for the moments ¢, of tho utilized observations. If necessary, one must take the perturbations into account. Denoting by (p,) + Mp, and (p2) + y Ap, such values of the geocentric distances for which the agreement of the computed with the observed coordinates «; and 8, is achieved, we can write 306, The Determination of Orbits fe ae | neal @P~8) z+ OP - 3) y= 8, -5) From this system of equations the most probable values of nd y may be found by the method of least squares, and then by using (p,) +2 Ap, and (p3)+ y Avg the improved orbital elements E, which will give the deviations of the observed coordinates from the computed coordinates, equal to the residuals of equations (10-3) after the substitution of the values for x and y into them. If there is no such agreement, it can bo traced to two causes, In the first place, the assumed variations, ‘\p, and Ap, or the ob- tained corrections of the geocentric distances were too great. Thon it is possible to have recourse to a procedure which is often used in similar complications in the application of the method of least squares. Namely, it is necessary to substitute the final sys- tom of oloments E in'place of the initial elements El; i.e., to insert the deviations of the observed from the computed coordi- nates obtained for the systom F in tho right-hand sides of equa- tions (10-8), after which the now solution of these equations gives corrections to the values of the geocentric distances of the system E in the same way as these corrections for the initial system ET wore first obtained. Here, a new determination of the coefficients in the lefichand sides of (10-8) and consequently, the recaleula- tions of the elements E1, KIL, EIN are usually not required. Secondly, errors could have crept into the calculation of ele- ments EI, FMl, FI or into the roprosentation of tho observations by the elements. A good control of the whole computing process is obtained if we find one more system of elements EIV having as its basis (p) +p, and (p,) +Apq. Then, having calculated for it a ands), we shall have, assuming that the variations of the geocentric distances are connected linearly with the variations of the coordinates «, and 8, af -a Maat oh 8 -5M= 5% sh, (10-4) and similar equalities will exist for any intermediate value playing f part in the calculations, so that the error will he easily revealed. If there is no confidence in the results of the calculations, it is worthwhile to calculato the system KIV hefore the solution of equations (10-8). If a parabolic orbit of a comet is being improved, the method of the variation of goocentric distances will give the most probable The Prel inary Improvement of an Orbit 307 value of the eccentricity. Consequently, if the orbit turns out to he fn ellipse, the corresponding values of the semimajor axis as well as the period of revolution will he obtained. Sometimes it is rec- ‘ommended to complete the transition from a parabolic to a nearly Parabolic orbit, by varying the ratio of the geocentric distances M = p/p, and introducing as the second defined unknown @ quan- tity equal to tho inverse of somimajor axis, 1/a. These caleula- tions are inconvenient and rathor laborious; the results thus ob- tained are the same as the results obtained hy the method of vari- ation of geocentric distances, hecause the existing observations aro satisfied by one orbit only, in the sense of the requirements of the mothod of least squares. How this orbit is found does not matter. Therefore, the opinion that the introduction of 1/¢ as a working unknown gives some gain in precision is based on a misunder- standing. 83. Summary of Formulas for the Method of Variation of Geocentric Distances A. Initial data ty Oy By Mu Yu Zn fay day Bay Kay Yay Zo w) In regard to the reduction of these values, soe Section 44. 2 = 608 5; 008 ai by cos 5; sin a; (@é=1,2) (dy ei sin 5, Control abe OB sof =1, ai ~ersin a cos (ti + 54s Bes ercos a; = sin (a) + 5). B. Derivation of elements Using the initial quantities it is nocossary to calculate the elomonts BL with (0,) and (93), PIL with (p,) + Ap, and (2), BUI with (p,) and (p2) + Mpg and, if desired, the olemonts FIV with (0) + Ap, and (93) + Avy 308, The Determination of Orbits by formulas (IX), (X), (XIV), (XV), (XVD, (XVI), (XVID, (XIX), and the representation of the observations (except the basic obsorva- tions) by formulas (XXT), Section 44. If the elements EIV were calculated, it is possible to control the results, by taking the differences of the computed coordinates, for which (xx) must be observed. Similar equalities occur for any of the intermediate values in the calculation of a system of elements. ©. Determination of geocentric distances and the derivation of improved elements (a= alyes (ap (P= 5) 2+ @P cee P| (xxi) ~ 5D y=5,-8} From these equations z and y are determined by the method of Teast squares. mee) (xxiv) #2 > (02) + ¥ Sea With (p,) + #Ap and (p,) + yApy the elements of the orbit and the representation of the observations are calculated again accord- ing to the formulas from (IX) through (XXI), Section 44. Control: The deviations of the observed from the computed co- ordinates must equal the residuals of equations (XXIII) after the substitution of x and y. If agreement is not obtained, see Section 82. If, in the presence of agreement, the residuals remain significant and have a system atic trend, this can point to an unnoticed ertor in one of the basic observations or to an unsatisfactory ealeulation of the perturbations. 84. The Variation of the Ratio of Geocentric Distances In the calculation of a parabolic orbit of a comet, by virtue of the conditions of the problem and the characteristies of the method of Olhers, it is not always possible to expect complete agreement in tho roprosentation by the orbital elements of the second position. Thorefore, in Section 63 it was shown how to improve the elements of the orbit and to obtain as noo! a representation as possible of the second position of the comet by means of the variation of the The Preliminary Improvement of on Orbit 309 ratio of geocentric distances, denoted by Wf. It is evident that similar to the method of variation of geocenteie distances, the variation of their ratios can be applied also not only to three ob- servations, but toa lnrgor number. The course of the arguments here is so similar to the foregoing (excopt that ono quantity is varied, and not two) that there is no need to develop it in detail, and it is sufficient to give a summary of the formulas cited earlier in Sections 63 and §2. A, Initia data In view of the fact that the number of the utilized observations can be significant, it is advisable to conduct the calculations in the equatorial systom in order to avoid transition from the observed coordinates to \and 8. The calculations are made with six deci- mals. ‘The corrections for parallax and aborration are introduced beforehand. thy ay By Kay Yay eal @ ay thay Bay Kay Yay Zoe 41 = 08 5; 008 a bj x c0s bi sin ap (F= 1,2). cj = sin 3, Control: oP + 62 402 a; 6; sin a» 608 (21 + 80, 8 +e, 008 0 = sin (4 + 3) Rp XP4¥2 423, } a BR; cos 01 = =2(ai Ki 4 BAY + 12d Controls RP £22, cos 0,41 Xi ~ a)? + (¥,- 6)? +, ~0)% B. Derivation of elements Using these quantities it is necessary to calculate the elements: RL with (M), ML with (M) + AM, where AW can be taken of the order of 0.001, according to formulas (WV), (VD, (X), (XD, (KM, KMD, (XIV); and also to carry out the representation of the observations (except for the basic observa- tions) according to the formulas (XVI), Section 65. 310 The Determination of Orbits C. Determination of the mtio of the geocentric distances and the derivation ofthe improved element (aB-adre Cree From these equations x is determined by the method of least squares. «vi Me (iye 2am, «vm With 4, the elements of tho orbit are again caleulated and tho rep- rosentation of tho observations is carried out by formulas from (V) through (XVI), Seetion 65. Control: The deviations of the observed from the computed co- ordinates 1pust equal tho residuals of the equations after a substi- Lution of the value of # into them. If such agreement is not obtained, it is possible to suspect a nnistako in the ealeulated elements or in the ropresontations of the positions of the comet, and jn some eases the nonlinear dependence fof tho coordinates on i. If, in tho prosence of agreement, the de- Viations of the observed from the computed positions remain sig- nificant and have a systematic trend, it is possible to assume an terror in one of the basic observation’ or # noticeable deviation of tho orbit from a parabola, In tho latter ease it is possible to advise the following method! take wo of the three systems of elements obtained and, having akon tho value of p, obtained for one of these systems, as wellas the value p, obtained for another system, calculate « now (non parabolic) ‘orbit with them according. to instructions, Section 83. ‘Then it is possible to consider these three systems of elements as the elements £1, FIl, and EIll, Section 83, hecause, here as there, wwe shall have four differont values for p, and py, and considering the last system obtained as ET, we shall find, by comparison with tho two other systems, with what values of Ap, and Ap, we are operating. Thus, it is possible to calculate an improved orbit, by ‘sing the instructions in Section 83-C. 85. Exomple For the comet 1946n Timmers there exist the following observa- tions, the first two of which were made by D. J. Martinoy in the Engelhardt Astronomical Observatory, and the remaining ones by A. D. Dubyago in Kazan: The Preliminary Improvement of on Orbit an No UT. roe Bres.0 1 1 1946 Feb. 10.08530 9433™25,08 4519211071 2 11,02559 9 29 50.64 +52 2042 18 a 26.73077 8 40 35.07 405 571K 6 4 March 1160453 § 28 58.86 +67 48.21 (7 5 2igso7o $24 2.05 S68 20 9 18 6 AT1347 8 16 23.87 $60 28 5 14 7 5.6838 8 12 1744 $69 87 18 13 5 TADILL 8 3 15.74 470 5649 13 Those moments of observations are already corrected ration time according to the provisional elements. ‘The coordinates of the sun were found for the uncorrected moments, as this is more convenient to do if normal places are not used. The extreme ob- servations were taken as the basic ones and with tho values (If) = 1.189436 and (I) + Aif =(if) ~ 0.000800 = 1.188836 the systems of elements were calculated Fl ou T 1946 Apr. 19.28978 —13.49484 U. T. a 12788 a 981105 ay =0.052073 - 0.058880 a #1485591 + 1.435325 oF +1.298979 + 1.300409, a “siiosos *a:104075 a +O.T436568 +0.738666 and the corresponding representations of the positions 2, 8, 4, 5, 6, and 7, from which follow the equations (Ac cos 8 being taken in view of the proximity of the comet to the pole): aw ” 8 » FLTe=-5.5 3% -06 z= 401 40% hegz=-98 9 11 <8 0 hei6e=208 41 3 408 HaTe-t4a 4418 4015 jagestoa to's ta 3 aabes-et At 9 m4 From these we derive the normal equation: 281.3 2-92.91, the solution of which gives =~ 0.099. 312 The Determination of Orbits With this value of x the residuals were obtained, given in the column », next to the equations of condition. Comparing these residuals with the right-hand sides of the oquations we soo chat the calculation of a thied system of elements is superfluous, because for a six-decimal precision it would not give any real improvement, fof the roprosontations; thus, the system &1 accidentally turns out, to be s0 close to the most probable orbit that there is no neod to proceed further. Having calculated for FI the ecliptic elements a», 2, and é according to the formulas (XV), Section 65, we shall have tho following oebit: 1946 April 18.28978 UT. Ba°1T 837.3 e128 5445 ‘8b 1940.0 i= 7 5920 2 92 LTS ‘Tho doviations of the observed coordinates from the computed coordinates aro rather significant (at least, some of them), which may be explained by the diffuso appearance of the comet. " Nover- theless, the elements of the orbit are sufficiently near to the true, and with diserepancies of the same order they represent the ob- servations hoyond the limits of the basic observations, beginning with 5 February through 1 April 1940, as shown by the calculations made later. Chapter 11 THE DETERMINATION OF A DEFINITIVE ORBIT 86. The Concept of « Definitive Orbit ‘The groater the number of the observations used, the longer the duration of time over which these observations extend; and, of course, the greater the accuracy of these observations, the more accurately is it possible to calculate the orbit of a celestial boy. ‘The dofinitive orbit is the completion of the process of approxima- tions applied to the determination of thg orbit of a minor planet or comet, Such an orbit must he basod on all the observations mado, and the calculations for it must tke into account the exact per: turbations from those major planets which show any noticeable ac- tion on the motion of the body under study, It is evident that every new opposition of a minor planet or every new appearance of a periodic comet can provide additional observational material for the improvement of the orbit. Consequently, in such cases the concept of # definitive orbit is relative, But evon when it is im- possible to count on future observations, as, for example, of non- periodic comets or of periodic comets which as a result of breaking up cease to be visible, the possibility often exists for further im provement of the definitive orbit in connection with a now diseus- sion and reduction of observations, with a groater accuracy in the determination of positions of comparison stars hy the use of the newest star catalogs, with new calculations of perturbations based fon more precise values of planetary masses, ete. Thus, the defini tive orbits of certain comets calculated in the first half of the last century were recognized to be absolete and have been redetermined ‘There is no need to be embarrassed by these considerations. In tho domain of the determination of orbits there are so many pos- sibilities for work on insufficiently investigated objects that cach caroful calculation of the motion of a minor planet or « comet pr sents interest and value, if the purpose of the work is well thought out. 313 3M The Determinetion of Orbits The dofinitive orbit may also be called the most reliable orbit in the sense that i must represent the observations by the principle of least squares. If wo denote the deviations of the observed from the computed geocentric coordinates by AG and AS, we must have Liha cos oy? + D (892 = mi This is equivalent to a statement that the sum of the of the distancos of tho observa positions feom the computed. posi- tions must be a minimum, TE the quantities A cos 8 and AS are so small that they can be explained by random errors of the observations, the orbit obtained is considered satisfactory. Howover, this is not always the caso. Ohon tho deviations of the observations trom theory noticeably exceed those values which are obtained for them according to the formulas of the theory of errs on the basis of random errors of observations, Whence, it follows that the premises. which have Served for the correction of the orbit are insufficient. A discus Sion of this froblom enn give vory intoresting rasults, but ie can Also present great difficulties, since the causes of the discrepan- cies ean bo very diverse. Some of these causes are insufficient calevlations of perturbations (in particular, if tho body passes wory close to tho perturbing planet, ot if the perturbations are cal- Culatod with imperfect eloments), emors in the coordinates of the earth (which can hecone noticenble if the twdy approaches the earth), systematic errors of observation of comets (Gopending. on their appearance and the presence or absence of a nucleus), devi- tions of the motion of comets from pare gravitational theory caused by forees whose mature and mechanism of action are still not eloar, ote. Bocause of the nonlinear dependence between the elements and the geocentric coordinates, tho condition of upplicability of the method of least squares is that the corrections sought to the ele- ments being, improved must be so small that Uhele spuaros and higher powers can be neglected. In those eases when the orhit being iin: proved does nol satisfy Hime requirements, it will be impossible to obtain the definitive elomonts of the orhit in a first attempt at Impenvone nt I is expeilient to apply the condition (11-1) to all the separate observations only if the observations are few in number. Other- wise, the volume of calculations will he excessively great. The solution of the problem is made easier if we use, instead of the observations, the normal places; for this the observations must be compared with an ephomeris calculated with elements that are al Determination of o Defin Orbit 315 ready sufficiently close to the correct values. The formation of normal places was considered in Section 33 and there remains little to add to what was said there. First, it is evident that the number of normal places must not he less than three, Second, how- fever, for making calculations easier, the number of normal places should not be too great, and the clustering of normal places should be avoided. Usually the observations of comets (tho problem of normal places has meaning for comets only) are grouped according to lunations, separated by intervals without observations near the full moon. If the comet has been observed for a sufficiontly long time, it is desirable to have so smooth a trend of corrections for the ephemeris that it should be possible to combine observations uring one Iunation into one normal place. Finally, third, the question remains conceming the weights of the normal places. Theoretically, it is not difficult to caleulate the weight of each place knowing the mean square error of one observation and the number of observations entering into th given normal place. How- ever, it has already been indicated above that in many cases the doviations of the observations from the definitive orbit. eannot be explained solely by the random errors of observation and that the deviations significantly exceed their possible limits. Therefore, itis possible to recommend, as the most correct, now the most often used, and the simplest way, that all Aa cos # and AB of the same weight be considered, disregarding the number of observations in each normal place. In the search for the most probable corrections to the elements according to the rules of compensation, it is necessary to establish what will be the changes in the geocentric coordinates caused by the changes in the elements of the orbit. The approach to this problem can be most diverse, and we shall consider only those methois which ate sufficiently often applied and which justify themselves in practice. 87. The Fundamental Eau ions for Correction of the El Let there be a system of orbital eloments F1, Ba, ..., Bay which must be utilized in obtaining an improved orbit by means of 8 series of normal places representing the observed coordinates of the body c, and 5. Let cc and 3, be the coordinates of the normal places calculated with the elements of the orbit under im- Provenent. We assume that the differences of the coordinates %y~ te and 8, ~8e, as woll as the required corrections to the elements, are so small that their squares may be disregarded. 316 The Determination of Orbits Thon, restricting ourselves 1o the first powers in tho Taylor's series expansion, we can write the following expression for the increment of the geocentric coordinates depending on the changes in tho orbital elements an ax = beet SE AES 2) : OEY a ok, es) and analogously forthe declinations. Accordingly, every separate hserved coordinate gives its own equation for corrections of the Slonents in one ofthe to forms: = +608 8 2% By = (24g a¢) cos 8, os 52% A, + cos SE AE, +. fe (ua) If we have three normal places, there will be just six such equa- tions, and from them will bo found the corrections of the elements AE ,.++;AEq- Much more froquontly the number of normal places is greater than three. Then equations (11-3) form an inconsistent system and must be solved by the method of least squares. ‘Theoretically, it is not difficult to find expressions for the coof- ficients of equations (11-8) which aro tho derivatives of the coordi natos with respect to the elements. In essence, the whole problem is reduced to this, because, as far as the right-hand sides of the equations are concerned, the deviations of the observed coordi= ates from the computed coordinates are obtained during the forma- tion of normal places, and tho solution of the equations is carried out according to well-established schemes. Nevertheless, the calculation of the coefficients is considerable work, in particular, if there are many normal places. Thoroforo, it is desirable, as far ‘as possible, to simplify the final working formulas. ‘This may be achieved by judicious choice of parameters by means of which these coefficients ara expressed, and often it is also worthwhile to change from the equatorial system of coordinates to a system con- nected with the plane of tho orbit. We tura now to the general equations (3-11) pcos cos ware, poos fainany+¥, ay psinb=24Z, Determination of Definitive Orbit 37 which aro the formulas of transformation of rectangular heliocen- tic coordinates to polar geocentric coordinates. From them it, follows: cos 5 cos adp ~ pcos 8 sin adu ~p sin 5 cos ad8 edz, cos 5 sin dp + pcos 5 cos ada -p sind sin add = dy, sin 5dp + pcos 5d5 = dz. Were in the differentiation we, of course, must consider time (given by the moments of the normal places) as constant. Further, wo obtain = sin adz + cos ady e a gs = in 8 008 adz~ sin 8 sin ady + 008 Bde - We do not need the analogous formula for dp because dp has an influence only on aberration and parallax; moreover, with the ex- coption of particular casos, the offect is nogligible. The differ- entials dz, dy, d2, expressed by changes of the orbital elements, must be substituted in equations (11-5) which give, in particular, for dz cos 5a (ans) an Sethe Oe (11-0) Eckert and Brouwer have shown that the above derivatives of the coordinates 2, y, 2 with respect to the orbital elements are very simply expressed by these coordinates if, in addition, the time derivatives of the coordinates are introduced, i.e., the components of velocity [23]. ae. dee edb @8. A Method for the Correction of an Orbit Based on the Application of Rectenguler Coordinates With the purpose to simplify in every possible way the calcula: tion of tho differential coefficients in equation (11-6) and in tha two equations analogous to it, it is possible to make use of the fact that it is not obligatory to introduce as unknown the corcec- tions of the ecliptic clomonts «, Si, i, but that it is possible to re- place thom by three other quantities from which the corrections of ©, 2, ¢ ate uniquely found. 318 The Determination of Orbits Though, of course, for the derivation it makes no difference to what systom 2, y, # aro referred, we shall assume that they are foquatorial coordinates as is always the case in practice. Every change in the otiontation of the orbit in space, resulting trom fcroments in the inclination, the longitude of the node, and the argu- tent of petihelion, can be tepresented as the consequence of three folations by angles dq, diy, and dy, around the axes 2, y> 2. We shall consider those olomontary rotations az the required unknowns. ‘To thom will correspond, as it is ensy to soe, the elementary ine ctemonts of the coordinates daa 2ddy— yey dy = 209, ~ ede, arn de = yt, -2dyy. From which at oneo wo find gE 20, #- ey Nis Ws ay ay ae Mas, 11.8) We Oe aed as as ee #8 20. a od, It is possible to substitute Py, Py, Pz and also Qs Qys Qs in formulas (14-7) in place of z, y, 2- Consoquontly, after die, dys ‘diy, ace found, we can obtain the changes of the direction cosines. For the determination of the increments of the usual elements da, aii, di in equatorial or ecliptic coordinates, the simplost way is t project the angles of tho elementary rotations on the corresponding axes, insofar as it is known from kinematics that infinitesimally small rotations are teansformed in the same manner as the projec: tions of vectors [12, 14}. However, it is just as easy to get the same results in the following way. Differentiating the first two formulas of (5-105), wo have, using (11-7) sin i cos ada 4 cos f sin wdi = cos edP, = sin edP, = (Py cos © +P; sin €)d Ug =P, (cos Edily + sin ed.) | cos @| sino = sini sin odo + 08 { cos w di = cos edQ, = sin €dQ, = (Q, cos & + Qs sin £) dye = Q,(cos Edy, + sin edd,). |-sin o| co w Determinction of o Defi itive Orbit ag Making simple transformations and making use of the samo equations of (5-105) as wll as (3-20), we find sin # de = sin Rady, ~ cos (cos diy + sin € dig ~ C05 (COS Ey 4) are) Ai = 608 Laiig + Sin R(cOs Edy + sin Edy). Differentiation of the fourth equation of (5-105) gives us =sin Sdil= =(Py sino + Q, 008 w)das + 008 wd? = sin adQy = 008 isin Side + (Py cas w~ Qe sin w) dd, = (P, 008 w = Qy sin a), hones dQ = ~cosidw sin edy, + cos ede. (11-10) Formulas (11-9) and (11-10) oxpross the changos of the ecliptic elements; that is what is usually required. ‘The changes of tho equatorial elements can be found from,here, if we sot cos € = sine =0. For tho dotormination of the derivatives of the coordinates with respect to the romaining orbital olements let us write the expres- sion for tho velocity component along the coordinate 2. The formu las for the other two coordinates are obtained by means of the simple replacement of lottors. Namely, we shall have wade ds dt du” ou ade’ since it is possible to put x = 2(¥, a, e, @, , #); consequently, # depends on time only by means of M, which in turn is determined hy the formula Wea + Hea) uy has, we find ao" and de 2" deat sig OMe ‘ y Therefore, for the partial derivative of = with respact to Moy the formule or de Mast artsy OMe” AM BM,” aM” will result. 320 The Detorminetion of Orbits Further, the known expression (3-31) for the rectangular helio- centric coordinates A ooeareo | wilt give dea e OE nde ME 2 oe MH ay see b+ Se Ge a We Ge de at GH ae On the basis of (11-11) i¢ will be WY OW de 8 da” On da 8 a because - dy 8 de zkavea, BLS 7 7 2@ Making further use of the formula (11-12) we shall have finally s (3) Seen. ax = PG oP, sink +aVi=e7Q, cos#). (17) # Here “js oliminated by the use of Kepler's equation, namely a ene i (nis) de" om de 7 Further, from (11-14) az ae is obtained. Bringing up again Kepler's equation and (11-12) we find a2 _ ae aM 26 | _ OM, asin E aE aM ab eae Determination of @ Definitive Orbit 321 Consequently, ae. de This equality is quite suitable for calculations if there exists (as is usual) an ephemeris computed with the elements being cor- rected, i.e., if sin & and also P and @ are known. Nevertheless, all these quantities can be eliminated. For this purpose, first eliminate P and @ using equations (11-14) and (11-17). We shall have soos = Esin B= aP,((cos E ~¢) cos & + sin? Fl = LaP,, avi-eQ, sin =x Zeos (cos f ~e) + 7sin E (cos E ~e), B20 oop 2% ging ~ 264 Mae cOSE (Cos E - &) ama # r- r=?) _ te sin E (cos E ~ 6) ele) cos E , 2(L=e%) sin B~e sin E(cos 6) «a= e%) ~~ OSE te, eos F as 1 in E But because cosh 22, peal 6%), (14-20) ‘wo can weita ae , e Sneek, (ee) where rep-2a rep si a ERT, th SOE x90) ? > 322 The Determination of Orbits ‘The elimination of sin # can be achieved if the first equation of (11-20) is differentiated and (11-18) is applied: Thus (1-28) rr aaa’ + yy’ + 22% This solves the problem of finding the derivatives of the co- ordinates with respect to the elements for elliptic orbits. But it is not difficult to see that equations (11-21), (11-22), and (1-28) do not give a definite solution if the orbit being corrected is a parab- ola. And also for nearly parabolic orbits the result will be unre liablo because of the factor a in the numerators of Hand K For- mally, the obtained equations remain valid for a hyperbola; however, in practice all hyperbolic orbits being corrected are so close to pa rabolas that one does not have to use formulas (11-21), (11-22), and (11-23) in this case. Removing a from those formulas, and also climinating it from the elements being improved, is not, in the gon- eral ease, a very easy task, the solution of which we shall now discuss. For nearly parabolic orbits the otements 1, g, and ¢ aro used. From the formule Maple) ares) wo find, as above, using (11-12), #2 ay, (125) tn oqutton valid for any ecconticity. fhe tunity gas), ad loo Cae EF ce Sal ve Introducing in place of a suming, in agreement with £e-n). et 2 aL Fee-7)- (14-26) Determination of o Definitive Orbit 323, For finding tho derivative with eespeet to ¢ we can make use of formula (11-21), but here it is necessary to remember that now a de- ponds one, Thus, denoting 2 obtainad according to (11-21) by (2) we shall have for the required derivative the following exe pression: Be _ (aa) , 9 Be de, da de” Noticing that for a constant ¢ me a ton), ea ize)™ Res Se, (1-27) Tho coatfictent fis reduced to the following sinple expression: Qa® ri p-2a+ae+ac® Rel 11-28) —y P @ Grerarey OS if we introduce, as in formulas (3468), (8-61), and (8-62), 1 1 Lee! eS ertan ge rag TAS. (Le) In tho coofficient $ we express all quantities entering in it by 4, ¢, and &, using Kepler's equation and the relationships Pratl-e%), wa kaw, r= a(1~e cos B), We obtain 7 cab eon ces aie 2A ~ 608 EY sin 6 22 8) sn =3(1 + eB sin BY) = mena +3e4¢2)sinE B(L +e) E-2e sin Ecos FI We introduce here € and o and assume for simplification t 20%. We can write 324 The Determination of Orbits tan? FE at 60%, Saretm yEm2vt (0-5 E -Fe), (a1-30) 1 Is After this it remains to carry out the following transformations: oa Garp [tte sees =2e0-9) a? ve 2 2 2 ogy elt (te) r- 8d eee > colt 77 Base wacdeoasey f-34¥ a. Mot eo%)* r <[-o- Ht E eri hers} oo ‘Tho series in the brackets is a hypergeometric series as is easily discovered when it is represented ina somewhat different form. It can be written down in the form of hypergeometric function, and thus our solution has been reduced to finding the value of this transcendental function: ayn z Ber GAH r(. Finally, we shall have for tho derivative with rospect to cecon- tcicity for nearly parabolic orbits an expression like this: % kes se’ tate 3 op D. 325 Here = $F (1, 4, J, -£0%) and the values of © are given below in the table for those arguments €0? which may be encountered in practice during the correction of the orbits under consideration. The chosen limits of €a? are the same as those in Table VI (see Section 23). eo eo et of 015 1.68618 0.05 1.61183 0.08 1.88876 283 226 aa 0.14 63060" 0.01 60027 | +0.00 5805 2st aaa 2 0.3 63109") 0.03 60803"! s0.07 -seaae 280 aaa air ~0.12 62659"? 0.02 60160 40.08 ss29 ner 230 ait not 92012" 0.01 .s0280 "7° 0.09 .ss002 26 230 ate =0.10 1.92866” 0.00 1.60000 +0.10 1.57758 ae 238 au no.09 2129 soon sora =) son s7st4 20 2 a 0.08 61880 40.02.5056” 40.12.7302 2a 225 aie 0.07 61639" 40.03 59381) +018 57180 230 22a 209 =0.00 01400"? 40,08 59095 40.14 soe 287 pea 209 0.058 1.01163” 40.05 1.56876 Fo.1S 1.56732 With this tab the esteaton of 25 for erie with an ecco tricity close to unity will not be too difficult. However, more often it will be necessary to be concerned with a special ease of the problem at hand, namely, with the correction of the initial parabolic elements, where de cepresents the deviation from unity of the ec contricity of the definitive orbit. ‘Then formulas (11-81) and (11-82) are simplified. Suppose ¢ = 1, = 0, p= 24, r= 9(1 + 0%). The series in equa- tion (11-81) receiving here the value 4, is reduced to the first rr term, and the hypergeometric function F becomes equal to unity. As a result, we find ro EE 1 \ 2 ats 5 y =r cos B sin, p(11-42) aarsinB. Bearing in mind that in the left-hand patts of equations, (21-39) stand the dovia- ns of the observed from the computed coordinates of the colestial body, we must consider that 2, y, 2 afer to the computed posi- tion of the body, i.e., in expressions (11-42) "we must assume B= 0. After this we obtain: rercosu, de ~ cos udr~rsin udu, yersinu, dy=sinudrereosudy,p (11-48) 220, do =rdB, Let the corrected orbit be inclined to the plane zy of the orbit boing corrected by the angle 1. Further, in Fig. 26, representing part of the heliocentric colestial sphere, let P he the heliocentric Position of the body (in agreement with the observation), PP’ be a perpendicular drawn from P onto the plane zy, and K be the point of intersection of the corrected orbit with this plane. Assume SK = V and KP’ =U; hence, u=V +. From the elementary triangle KPP” wo find dB=lsinU, de=rlsin(u-V). (ey) Bvidently also KP = KP’ =U, and, if we desice to take the are sument of latitude according to the corrected orhit, we mnst mans- ture it not from point 2;, as one apparently should, but from point (G), appearing at the foot of the perpendicular drawn from $2 onto the corrected orbit, since then ()K =f =V and u=VsU. Correspondingly, we write u= 0+ 01, where ay is measured from point (52) to perihelion. It will be necessary for us to take this into ‘account in determining the correction of the argument of perihelion da» Because da is formed from two different incromonts, depending fon the change of position of the periholion point as woll as of the 338 The Determination of Orbits point of the node, wo shall have da = deny ~ (QS = des, ~ 608 ide (11-48) ‘Tho latter equality follows from the elementary triangle 2(9)% (Fig. 26), Finally, we shall have, from the elementary triangle ‘U0, K, after small simplifications, Fin V = 2(8) = sin ids, 1 cos V = sin cos V = di, (11-46) da =rsinudi-r eos usin id. Let us now introduce the results of (11-43) and (11-46) into equations (11-88). This gives 08 gd = =I sin (G —u)dr +r c08 (G - dul, dg = ~ 2 feos (@ - u)dr +r sin (G = w)dul aan) + SOE Goin ude 005 w sin id) Horo it is necessary to assume ar any, a arm Fredo dus Seas, A aa (1-48) due 2 anys Maes Mags dor. aM Opt ag ie It is necessary to express the partial derivatives entering here in explicit. form, which we work out only for the ease of an ellipse, making uso of tho known formulas for elliptic motion: r= cho Bin gain B= My + ple tu) P 7 ) sin ¢ cos » tan v= tan (45° + £4) an be rein v= a.cos ¢ sin Ey F008 » = a(cos & ~ sin J). Determination of a Definitive Orbit 339 Differentiating, in order, the first four formulas (tho first and the last of thom logatithnically) we find, respectively, da 2 de a” 3 yt (A= sin 4 cos E)AE ~ cos g sin Edg = dMo + (¢~ to)duy dr = Lda asin 6 sin E48 ~ a c08 6 008 BS, dv dg a Sine” cos @ * and from here aE F dle + 2(¢-to)dy + sin vdg, dw tang sin octtos [eto)atand fin v= 27a = 2 cos cos udp, , atts) £802 ay + acos 6dE cos & # i £288 aay 6 89) HSS ay + (p +2) 800 $ sin vdg. Let us assume ar Brg” 2 tang Sino = My sin Ma, ar $i = -2 008 4 008 9-6, 8in bn 12h ESE E cos ita, ay 1 a a ta) HEF oy, 005 segs 340. The Determination of Orbits 13E nm (p44) 800 g sin v= $, 608 ba, M, cos (P= 0+ My) = Mey Hy 008 (0 ~u+ Ha) = Hey #008 (= us da)= der Feos (F =u) = wey M, sin (G = + Ma) = My, ay 8in (GO ~ 0+ p= Hy Sin (=u ¥ ba) = bys rain (@ = Altor this the equations of condition for the correction of the colomonts assume tho following form: Deena eter eoesict sin g EM dMo + Hedi + edG + eden) (11-50) (© sin wat ~r 00s u sin idl) = dg. Similar equations must be eonsteuectod for all the normal places and solved by the method of least squares. Certain remarks about this have aleeady hoon made in Section 89, but the solution by the prosont method will he simpler. In the first place, the equations for cos gd@ do not contain the anknowns di and «il at all; therefore, the solution of the normal equations is correspondingly simplified. In the second place, the four remaining unknowns enter in the equations for dg with the co- fl ot ing which dos nt oxened nf whore isthe ew dua voce of the ouch. Tndod, fom formulas (1-96) and (14-43) uu because 7 is condiate of de earth, coesponting to te axis perpendicular tothe plane of de suit ho bay incvieul ane meter nf in Tao Gro he eqatons fr gw have toceconl fconce ome serminain of he Tec foe lemenia oe cannetad wih te thoncinon of eh ois emalh sil boi th mary ot Tinting diy dy tal dry Ae tay the vals Tod thoveinkaovas cubated ia th omntons oe dy hich wl Sewve Torte dteninaion of di ant te Thue plac a te it follows that sin g | Determination of o Definitive Orbit 341 system of 2m equations of condition with six unknowns, we shatl have to solve two systems of m equations, one with four unknowns and tho other with two unknowns. The method presented ean be successfully applied to the ma- jority of minor planets and of periodic comets. Comets moving long nearly parabolic orbits often have a considerable inclination For thom the main advantage of the method presented is lost, and, therefore, we shall not examino the formulas pertaining to tho el ments Tq, and ¢ of such orbits. 92. Summary of Formulas A. Initiat Data For the series of observations or normal places there are tA, By (Ho ae) C08 8, By ~ Ber a) For preciso calculations, &o— i, and*B, = 8 must be obtained from the calculation of perturbations. The elements of tho orbit, being corrected must be roferrod to the samo oquinox as x and 5. Calculations are conducted with four or five decimals. B. The Transition to Equatorial Elements co Ji sin f(W +0) = sin 4 cos 4(é- 8), 008 Ji cos 4(2’ +0) = coe 3 cos H(i + Oy a sin $i’ sin 4(W - 0) = sing sin d(é-&), sin $1" cos }(U = 9) = cos 42 sin f(i+ 6). ©. The Transformation of Coordinates O=Gs0, cos y c08 @ = cos 8 cos (# ~ 2), cos g sin G jin i sin 8 + cos i cos 5 sin (x ~ 2"), sin 20s i” sin 5 — un # cos 5 sin (a ~ 0’), ay cos g sin F = sin cos (x ~ 9), cos 7 cos F = cos é cos 6 + sina” sin 8 sin (x -2"), cos gd@ = cos F cos Suu + sin Fd, dg = ~sin P cos 3d0 + cos Fab. 342 D. The Derivatives of the Coordinates with Respect to the Elements M, sin My ~atan¢ sin», a? 008 M, cos My Sin py = (t= fg) @ tan g cos @ ~ Hy 008 py = (2= ty) ES, av) $, Sin dy = -a-co8 J cos », a 008 gy = (Pp +1) sec 6 sin % Mo =M,cos(G-u+My), My =H, sin(@@-u+ My), hey 008 (G= U4 4g), ay =H, ING H+ Hy), b, 26, 005 (G-u+d,), G,- 5, 8in(G-u+ by), w, = 1 cos (GW), wo, =7sin(@-), 2 (1.58208): n° = 5.138981: 2". 3a Control: tan My = tan g sin B, He = Mell = to) + SE tin G—ws ay = Malt = to) = Fc08 (G = wp $e = alsin E cos (G ~u) + cos 6 sin (G ~ wl, 4, = alsin E sin (G ~u) ~ cos $ cos (@ ~ u)). Me Me be ®e Feats + Mans Sides dey = cos gat . Sudan 2 6,46 - 2 oder pW) 10039 reos 9 sin ude = cos usin id= dg. = a= dg. E. Determination of Corrections of the Elements ‘The solution of equations (V) is performed in accordance with instructions in Section 89 and at the end of Section 91. After find: ing das and @2, da = da, ~ cos idl (vd is calculated. Determination of @ Definitive Orbit 343 With the corrected elements the heliocentric positions of the celestial body are calculated for the moments of the normal places. ‘The deviations found for cos 8a and dS must bo oqual, within the limits of the accuracy of the calculations, to the quantities cos dit = cos F cos gd@ - sin Fedg, 5 = sin F cos gd@ ~ cos Fig, where cos gd@ and dy are the residuals of the equations of eondi- tion after the substitation of the values of the unknowns. In other respects refer to Section 89. 93. The Approximate Correction of an Orbit If great accuracy is not roquired in the correction of an orbit, the calculations can be considerably simplified. ‘The number of minor planets is so groat that it is impossible to follow with complete accuracy the motion of all planets, even if sufficiont numbor of observations were available. However, if use is made of a preliminary orbit, the divergencies of the ephemeris from the observations can reach such values in a few years that the planet can be lost, as indeed often occurs. The method pro- sented above can bo applied for rapid correction of orhits of minor planets, and the elements obtained are so reliable that U guarantee the confident calculation of perturbations and the of the planet for a number of years. ‘AL the moment of opposition, if it is considered in respect to the plane of the orbit, we have @=u. If the inclination is not great, this will be almost true also for opposition in right ascen- sion. Selecting from the observations belonging to different op- positions (the formation of normal places in a similar case is mean- ingles) those observations which aro the nearest to the momonts of the oppositions, wo can assume @ =. In addition, for moderate inclinations it is possible to consider cos g = 1. It is almost al- ways possible to limit ourselves only to the equations for cos gd, fand the equations for dg are not worked out, because usually the corrections for i and 2 are not needed. In fact, if these eloments should appear to contain sizeable ‘errors, it is doubtful whether the planet could have been observed at four different oppositions (for the correction of an orbit, ob viously, observations are roquired at not less than four oppositions). In consequence of the simplifications made, it is possible to labulate many of the necessary quantities, and that has boon carried 344 The Determinotion of Orbits out by M. F, Subbotin [13]. Tt is assumed that the calculation is carried out hy using these tables; however, oven without the tables the calculation prosents no difficulties. The values of the cquatorial elements i” and iY can be taken from the tables with the argumonts i and S. Here an accuracy of 1° is sufficient. Letting cos y =1, we see that sin F in the first equation of (11-41) is determined only by means of i” and x - 9) which permits tabulating sin # and cos F by these arguments. The majority of cases can be treated still more simply. Take cos F = 1, and, as a consequence of this, it is possible to take directly cos 5d: in place of cos gilG, and the necessity for finding the equatorial elements and the angle F is oliminated. The coofficients of the equations of condition are also simpli« fied and are calculated by the following formulas (the unknowns and the righthand parts are expressed in dozrees): a, be pat (ats) ‘The functions fg, fg, and f depend on only wo arguments, and the tables of M. F. Subbotin give them by arguments and gy The equations of condition are set up in the form M, 8, 4 ®, Edy + dps dg sda = cos gd 52) a pes Bdge gag (11-88) and solved hy tho mothod of least squares. If four oppositions are used for the correction of the orbit, oquation (11-52) is solved immediately. Such a method for the correction of an orbit hy four oppositions is often applied in practice. For control, the positions of the planet. are ealoulated with four decimals far the momonts of the obsorvations used. If the values of a ato represented satisfac~ torily, the same usually applies t the values of 5. For large do- viations in a (if they are not due to mistakes in observations) it is possible to try to improve the result by a secondary substitution of the obtained doviations in the equations of condition and by solv ing them anow. If, for a satisfactory roprasentation in a, there ro main noticeable discrepancies in 6, it is necessary to improve + Determinotion of o Definitive Orbit 345 and %. For this one may resort to the formulas for dg introduced above, of pass on to the method of variation of roocenteic distances. However, similar work will not often be necessary. For a calculation of the coefficients (11-51) more than three decimals are not required. 94. Exemple For the periodic Comet Brooks, 34 position observations were obtained during its appearance in 1925, 52 observations during its appearance in the years 1932-1993, and 21 observations during its appearance in the years 1939-1940. All observations were proc~ ‘essed anew as far as possible for the determination of a definitive orbit from this material. In addition to control of tho reducing, calculations (parallactic factors), the position and proper motions ‘of the comparison stars were found anew in the system of the “Gon: eral Catalogue’ (20), with all accessible star catalogs used for this purpose. Photogeaphic observations were for the most part published without indication of the coordinates of the comparison stars. In such cases, systematic corrections of the star catalogs used hy the observers wore mado, and in particular for certain Photographic catalogs of the Carte du Ciel the cocrections to the GC system were found by diract comparison of positions of a series of stars in these catalogs and in the GC; those comparisons were made especially for the given purpose. All the observations made in different appearances of the comet were compared with accurate ephemerides and finally the following normal placos wore formed No. OF ULT. x 8 Equinox 11928 Sept 23 «54971010 — "48" 398 1925.0 2 Oct 13348 58.33 ‘a8 8 35 oo 3 Nove 12 382317 43 _ 6 27 43 ‘oa 4 Dec. 101 8713 [50 = 3 a8 25 “Ba 5 1992 Oct 11k 3.88 gT 28 20 “05 8 Oct, aT = 9. aD 38 «51.80 92 t Dec 8 4 4023 1) FO 338 a8 & 1993 Jans 2028-22008 T9756 96 9 Mar 22564055 (34115 5046 aa 301999 Sune 203 d's IGF BO Sy 1950.0 0 ‘Aug. 19 27 3281-39 $10 52.52 181 2 Sep 8-303 $10 35 52 [30 irs Oct 12 3a Bran 2 fT 5030 12 ut Nov. 5 313140 01} 5 849-06 15 Dee 134 1259-24} 4 4342 oT 18190 dan: 445-827 + 6 51aT 130 346, The Determination of Orbits No. Perturbation cos Bd 46 8 a3 Zo ‘on 0 00 =0 98 0 100 -0 Zo ai =o 4 Ea Zo lot iM ue na fer For the comparison of the observations with the ephemeris, parallax and aberration wero, of course, taken into account, the latter in the following way: From the moments of the.observations the aberration time was subtracted, after which the observations wore compared with the ephemeris coordinates ftom which stellar aberration was previously subtracted (see Section 30). The ob- served & and 6 of the normal places were reduced for procession to tho normal equinoxes. In the table are given the values of the perturbations of the coordinates from the nearest asculating epoch to the moments of the normal places, calculated in rectangular co- ordinates with intervals of 20 days for each respective period of visibility of the comet, and the action of Venus, Earth, Mars, Jug tor, and Saturn wero taken into account. Making use of those values of the perturbations, the observed coordinates were compared with the coordinates calculated with the osculating elements under cor- roction. The results of the comparison arc also given in the table of normal places. The elements boing corrected are as follows: Elements 1 Bpoch and Osculation, U. 7. 1925 Nov. 9.0 1982 Oct. 3.0 1989 Sept. 24.0 Me 1° yo0%3s 380° 424"%60, 1910'12"750 o 195 4214 T1195 49.52 30195 40.88 70 a 1772510 111177 2055 13377 42.23 83 4 5531 51 53254 173 5 52.47 28 3 209477) 2858 aT 29458 a2 » 12780814 5EL.16517 ‘510.6037 Equinox. 1925.0 1925.0 1950.0 Determination of o Defi ive Orbit 37 ‘These clements roprosent the samo system and differ by tho values of the perturbations between corresponding epochs of oscu- ation, The perturbations were calculated by the method of varia- tion of elements, with the same intervals and from the same planets ‘that are mentioned above. We give also the precession of the ecliptic elements from the equinox of 1925 to 1950.0 Aw - 67.86 AN 6 2173745 Ag = 11075. Tho derivatives of tho coordinates with respect to the eloments of the oxhit wero ealeulated with certain deviations from the formulas deduced by us. In place of the corrections of the ecliptic alements dlotermining the position of tho orbit in space, we found the rota tions dp, dq, ds of the orbit about axes directed, respectively, to the poritielion of the orbit being corrected, to the point of the orbit whore 7 = 0°, and to the pole of the ofbit. Ie is easy to soe that ds dos, Further, we assume that in Fig. 26, the point K coin cides with perihelion. Thon wo ean write I= dp, 7 =o in the formulas (21-46), which give dpsina=sinidl, dp cos a =di- Assuming now that at point K the true anomaly is equal to 90°, wo find dqcoso=sinid®, -dq sino =di. The complete change of the ecliptic elements { and 2 will con sist of the influonce of both rotations, dp and dq. Finally, we got di = dp cosa -dqsinw, sin idQ = dp sin w + dq cos o, (11-58) do = ds ~ cos ids. If expressiuns (13-38) ure substituted in the last worms of the equation for dg (11-80), these terms are transformed to the follow ing form: . cos 9 dg meet (sin vdp ~rcosvdq). (11-84) Having dotermined dp, dg, ds we can find, using (11-58), the valuos of di, a8, das. 348 The Determination of Orbits This replacomont of variables (which is often applied) has an almost entirely formal character and another supplement ta the usual formulas is much more essential. Elements I are obtained from the appearances of tho comet in 1925 and 1932-1933, and for this period of time the observations aro satisfactorily represented, However, they significantly deviate from the observations in 1939- 1940, and it is useless to soarch for such a system of six ele ments which could represent all three appearances of the comet. This confirms an oartior established fact that Comet Brooks has a secular acceleration of motion (as does also Comet Encke). Thus, a3 a first approximation to the unknown Law of motion of the comot, it was assumed that the mean daily motion and angle of eccen ticity change proportionally to time. In consequence of this, we shall have the equations, $= b548'C~b), = ty 2H fo), Mo My + Hot = to) + (Cte)? ‘The first equation of (11-50) can now be weitton as: ae ae ag, aa wo tng 22.» 22 ine 3 ses ag aG aa a (2 da ae al “ », OG ad da in 5 * aM ae) OE 3d, The values of certain partial derivatives can be substituted from tho procoding equations, after which 30 gy, {24 a0, cos 9G = cos e a ea Mo -(# au? + pened sag is obtained, Comparing both expressions for cos gd@, we find easily the simple formula permitting new coefficients to he caleulated from those already known: Determination of o Definitive Orbit 349 ae cos 9 ae “ a6 y, +08 9 M8 46 (t= t9) 008 9 6 mes [An analogous expression is found for dg. Wo give the caleula- tion of the coefficionts of the equations of condition for one of the normal places. Certain initial values were borrowed from the calcu lations of the ephemeris. The initial date is f) = 1932, October 3.0, U. T. The calculations wore made simultaneously by two computers, and, therefore, the control formulas (which check the calculations only superficially) were not used. © = 2882667 con ky ain LO’ + 0) 9.9046 sins” 9.4876 i 592.8 con ti’ cos LO +) 8.2571 cons” 9.0875 Q ATT 424 gin be’ win LQ’ 0) 9.1918 20,8618 © 195 40.9 sin Ey cos LG —o) 1.8097 08 9.9615 sin bt 9.0000 EGU +o)= 885216 tang 9.7452 cos sh sane 2-0) == 88 9.9 2008 6 0.5028 sin FG +6) 9.6165 b= 17 1.9 @ tan d 0.3065 cos FG) 9.9859 sin be 9.1920 P0688 sin $@~e) 8.19189 cos be 9.9947 2.1081 cos 4 (= €) 9.9947 ve 115110 2 2.4302 V= 04s oa ow. nas These calculations are made once for all normal places belong- ing to a given appearance, Further separate calculations are cone ducted for each normal place. ‘They are mado simultaneously for all normal places, in parallel columns, on several sheets, aecord- ing to order. £1940 Jan. 4.0 (eto) My sin My 9.6001 foto 2649.0 Bea tsa: tot, 4.4230 ~ aye BTR rine 0.1995 Hy sing 8.5836 feos 0.1479, iy cos jr 4.1052 ve 4T56! bys 18° 8.8 Fr o2ig mtv 2 23 d= a8°as.8 eosiig + Gu’) 9.9901 B= 6 52.0 4.1767 350 The Determination of Orbits 0.1795 anh. aisxe sind 9.0776 conS 9.9968 singa=it) 8.8 sing 0.82850 cosa= 1) 8.0141 ico d, 0817S 1 s.s852 Gye 2718.3 uu s:taa5 bye elle aA Sing + @%—u') 9.82820 He 41668 He 3.50198 pir 0.6886 Sogo @ vant cong nO" ow) peat = tees0 Bt 0888 Mm = 9.0561 sin(ds + =u") 9.89990, HN 324i0 oleae sing S182 So.st8en cong 9.9992 cos (7 =u) 9.9560 ¥ 910754 sin(@’ = Ww) 9.6816 vi 8.3323 ors cos gsinF 94017 8.95360 cos gcos F 9.0849 o'stia in F 914035 3.9873 cos F 9.9852 4.2883 cos 54a +178".26 Ce tg) con gd dln 8.0044 ab + 48 ah = 910800 com Bue 228i cos gd@ say’ 74030, 10888 Gens vm 212307 : vin 1.0913 3.6057 cos gd 2.2607 0.0984 165350 s.8082 Levas 8.6059" 0.3485 2.10510 0128048 ae 2.4081" ean H(t ta)dg idle 2.0289 sa°30.1 %2.1608n <5 21.1 Agidi’ 5.59250 oar Agiddby — 9.1T16n ota | @o:d?— 2.8046n 2 dgide 3.556% s.o0a2 soagip 0.8107 ors77 dgidp 0.0122 9.24800 do:dq 9.98760 0.7800 oco02t Thus the coefficients of all 32 equations of condition were calculated. Next, taking for all equations identical weights, new variables were introduced, as well as the right sides, by moans of the replacements: 2 = 0.86dMo, 2 =[4.24}duo, 24 =(T-651,’, Determination of a Definitive Orbit 351 24 = (0.85]dg0, 29 = (3.76)4%) 29 = [0.84)de, 27 = [0.021 dp, 4 = [0.34] dg, 1 = (2.52) 1”. After this wo have the following equ tions reduced to numerical homogeneity: elev cee ree tec sy ee cae 40.0108 -0,0872 +0.9800 ~0.0865 + 0.9882 40.0567 Folss03 “olgite 4o.ai1e “o.sozs 4 0.toe7 +0.5686 Thvgrat =0.7298 4 0.TI88 ~0.3966 + 0.8K + 0.6683 ‘rol54s3 “0.8761 0.8670 ~o.0xa8 40.0525 +0.5206 Tolaray 0.0003 "0.0000 =0.1340 0.0001 0.9808 Tolsost vosdozt oco000 +0l0283 +.0,0002 +0.9080 Foleua? +0.0087 0.0001 +0.387% + 0.0087 + 0.0240 Toleasa Toloaat Yolsesr so.oue8 40.4880 foarte Toloooe T6"74a8 so.0492 40.3688 coaiss Foluaee “elaa28 “04030 Toleas Tolerss Solteso sos16e2 tora Tolsans +o:scos +014000 Tolan Tolosae Lolnabe $26 Yo'ss90 0.0505 +olsa0s Torsrre }o.sss0 +o.0275 Tolesas Joltts2 solrier Lolss02 tolsesd 0.7009 To'stes Solssab +015803 ‘olsene 20.8783 Tolssst 0.0007 + 0.0006 ~0,0008, 10.0054 “o3p004 ~0.8428 —a.0128 “o.oo. Toler 4 o:o108 —o:0008 ‘oleate Terno9s —o.30e8 ~o-RbIs ~0.0011 Tolnans $0202 Tolo18s o.otst + o.ore4 —o-0220 +0.1308 “0.7428 +0.0021 Toloies $0.0188 oloies ~o.0723 + 0.0700 —o.0206 +0.302 “olsa22 +0.0003 {9.0002 “o‘000e —o:0000 Zolos18 “o-0000 +o.c00T “o.1431 —0.0788 “0.0087 0.0006 ~0.0030 0.0000 =oi0000 “o-01a2 So.zt90 To:p143 +.000T Toleied “os0019 0.0001 Thleoat Telnza6 Talee4s Tol6iTa +0.0088 Toloos? “0.0009 “o%0001 o:0419 “0.0018 —o.0i09 So:re70 FolserD +0:0001 ‘2.0000 0.0000 0:0000 0.0012 = 0.0001 0.0008 0.7482 ~0.1598 0.0082 =0i00¢9 ~0-0082 ~0:0052 0.0488 0.0486 0.0130 Zolasaa “00140 Chloiar “elotg “olor rolosaT 20.0838 ~O0Me Tolease “e-ores “olo0ae ~e-o0ar ooase +o'0%60 +0-0188 0.0039 Thlsaaa —6.0070 Toloore +6:0070 0.0088 “0.0823 ~0.0818 Tois2ss + 6.0080 ‘10.0026 + 0:0008 “0:00%6 —0.1038 Toissss To.o07 “hloors “elotgo “olo1at 0.0098 “oiooso Toleose “00073 Zo:0088 —o‘06ss oer From these equations of condition, the normal equations were obtained by the usual mothod. Verification was made by using the summation equation (it is not reproduced here). For the first 18 ‘equations of condition the corresponding, normal oquations will be: a * ty % on te a $8.2097 41.9001 46.2586 4 14927 44,7371 +8.as22 +3.8136 TerTo79 + 1.4008 +5.0883 + 1.0718 + 1.9850 44.0940 FOTOST + 1.0454 +5.0871 6.4228 + 4.0072 FOOTT 4155925 1.6670 + 3.0445, $s0001 24.0986 + 2.0758 3816620 + 3.9377 ‘The values of the unknowns and the inverse weigghts were found from them: 352 sy =0.06199 211 105,088 2 #0.81511 a. 0.84929 * 52205, sx 1.28288 By ~0.01802 Gis Onta097 =) —0.02500, Oss 1.16488 qe +0.06186 te 108.00 Tho unknowns were substituted in all the equations of condi- n. For the first 16 equations the sum of the squares of the ro- sidual deviations will be [vv] = 0.000241. For comparison we give farther [11] « 4.139840 and [7g] = 0.000236. The lattor value agrees With [ve] within the limits of tho accuracy of the calculations. AS ‘result of the substitution of the unknowns in the second xroup of 16 equations, the equations of condition were obtained for x and 24, for which it is enough to give the right-hand parts because Gheir coofficionts are already given above. ‘These now right-hand paris will be (according to columns): 0.0019 ~0.0038 0.0082 0.0008 0.0013 0.0008 0.0058 +0,0041 0.0023 +0.0040 0.0021 + 0,003 0.0087 $0,0005 “0.0027 + 0.0182 We write out the solution of the normal oquations for zy and rs, which was porformed according to the complete (so-called “astro. nomioal”) scheme: * : &) co) = 19342 +0.0950_| + 1.0000. ~orsasTs, +0.17877| san07s 0.0000 yocacsn =0.3488 Ss yales0s estas To000 salnaos a05e Sonesee oteser 0.5488 | + 1.0000 z Socotser| —ost200, 8 ta rosie L000 1 hesst ross 2 1:0000 ‘He obtain tr + 0,00472, Orr 0.19442, {21} 0.000375 5 ++ 0.00071 Oss 0.19081, (tq) 0.000258 Tos} 0.00258 The weighted coefficients wore calculated according to the rule of N. 1, Ielson (71; the quantities being multiplied in the columns (#:) and (e4) aro marked off to the right. Returning to Determination of @ Definitive Orbit 353 the previous unknowns and to seconds of are, we find the eorree- tions of the elements and their mean ereors. For finding the mean errors necessary to know the mean error of unit weight. Tt equals +17.499 and its derivation is given helow. The mean er- rors of «, {2, # are obtained from the formulas. £2 = e2 cos? w + ef sin? o 6 = ose? i(e2 sin? w+ €8 a 62+ ef cos? i, ® cos? a)y Wo have do ~2°.88 2% as do + 9%.38 6.08 ang 50 ‘ooag2 + 0 Lovq00 de FL 80 20 Ces weave io fageio® = dg 3010-0 28 dbo | = 08 £0.88 do 41k ET 39 oe -ua9.10® teenie = dQ = 5 ato OF a 1 $0 ‘60 After this, the corrected olemonts of the orbit of Comet Brooks may be given. Elements I Epoch and Osoulation, U. T. 1925 Nov. 9.0 1982 Oct. 8.0 1989 Sept. Mo 1° 1’25".e9 359° 4’21".77—1919%s9" o 195 4229 26 195 50 6 (88 195°41 11 Q 117 25 4 (89 177 2050 (31.177 42 18 i 53810 10 5 32.58 5 9245 4 29949 br 29 29 452 A 5127-79841 3 10”, ia 43978" s faa « Equinox 1925.0 1950.0 ‘The final control of the calculations is the representation of the normal placas hy the carrected elements. For the ealewlation of tho roprosentations it was necessary to koop in mind that now we do not have the usual Keplerian orbit, but that x and ¢ are sub- ject to secular variations. For the moment of each normal place M, (2), and g must be calculated according to the formulas which are given above. Therefore, it was convenient not to calculate A,, » B, for every normal place, but to make use of the quantities P and Q, multiplying them hy a (cas E ~ e) and by @ cos @ sin FE, respectively. Tho residual doviations of th equations of condi- 354 The Determination of Orbits tions, after substitution of the values of the unknowns, were con: verted to seconds of arc and reduced to equatorial coordinates. As a result we have a representation of the normal places accord. ing to the equations of condition and directly by the elements of the orbit; in the latter case, of course, the perturbations of the coordinates, given above, were Laken into account. By Equations of Normal Condition By Final Bloments Places cos5dx dé cos bu. i +0715 407.93 4022 40.69 2 sos to tr fo tg to Be 3 PS eremmret $0 1831 Coe 4 “23 to 2 ls fo las 5 sits oY ‘38 1 laa 6 0 “oe To +005 © Lo lor 7 sist Th 4118 Fo [9 & tage fo 2 0 ue 3 <0 4s I tos Ty lor 10 sito To un Lo ‘58 ul oso | To 20 20 lar 2 so Zo un =o 139 13 $0 =o to Tom 4 a zo Sia 20 115 ts =) Zo =1 =o lee 6 40 f a to: +3 lo The sum of the squares of the deviations obtained di- rectly amounts to 58.92, from which, according to the formula cow 2 » Where m is the number of equations, and n is the number of unknowns, follows the value of the moan error of unit weight which was already used above. Between both series of deviations thore is no complete agreement. However, the revealed systematic differencos in declination in the third appearance of the comet, though undoubtedly real, are so insignificant that further correction of the elements is not worth while, and the orhit obtained can be considered as definitive. The cause of the indi- cated divergence is evident: The deviations of the orbit heing corrected from observations were too great in the third appearance of the comet, As a consequence of this, the connection botwoon the corrections of the elements and the changes of the coordinates was not sufficiently linear. In addition, the right-hand sides of the equations of condition (the initial differences of the coordinates) should have been, strictly speaking, also reduced to the normal equinoxes because the comparisons of observations with the Determination of @ Definitive Orbit 355 ephemerides were made in coordinates referrad to the oquinoxes of tho years of the observations, I is possible to assert that the orbit satisfactorily agrees with the observed positions as shown by the small mean error of unit weight, i.e., of one normal coordinate, since all of the normal places were entered with a woight equal to unity. The more notice- able differences in the declination found for tho 9th and 16th noemal places are explained by the fact that these places were formed in each case from only two observations at the same observatory, where they fell on the very end of the observing period when the comet was close to the 16th magnitude. Another eause may be this: the dependence of the elements and ¢ on time, accepted hy us, undoubtedly only vory crudely reflects reality, and the error of our assumptions will become conspicuous for the normal places corresponding to extreme dates (of the respective appearances). Hore it is possible to repeat the remark made hy N. I. Melson in 1934 concerning Comet Encke [6], namely, that when we introduce secular acceleration, we cannot affiem, strictly speaking, that. the unperturbed motion occurs according to Kepler's laws. Conse- quently, we do not behave very logically by making use of the formulas of elliptic motion in their usual form, This, of course, happens from necessity, because the law of action of the force causing the doviations of the cometary motion ftom gravitational theory is unknown to us. However, it is possible to show further that whatever this unknown force may be, if it acts near the pori- helion of the orbit (which is very probable), then between the values x’ and §” there exists a certain dependence, which is sufficiently well satisfied by our calculated values, But we shall not investi- gate this question here. Chapter 12 ‘THE DETERMINATION OF THE ORBITS OF METEORS 95. Introduction At the present time it is considered as vory probable that the overwhelming majority of motoors observed by us belongs to the solar system. The determination of the orbits which meteors, and also bolides, deseribe around the sun, depends on the observations of tho small part of their trajectory which passos through our at- mosphere, helore their existence as independent celestial bodies comes to an end. A meteor is abserved hy us for a vory short timo, fand the corresponding segment of its path can be considered as rectilinear. If we aro able to obtain the magnitude and direction of the velocity vector of the mateor from observations, the problem of finding its orbit will be solved, because the coordinates of the meteor with respect to the sun are known. They are determined by the position of the earth at the moment of the observation. ‘The dix rection of the motion of the moteor with respect to the earth may be given by the coordinates of the point of tho radiant on the celestial sphere. Namely, the radiant roprosonts tho intersection with the celestial sphere’ of the rectilinear path of the meteor continued backward. If a shower of mateors is observed, the radiant appears ‘as the paint of intersection of the apparent paths of the separate meteors extended backwards. Tho point of the radiant ean be found ‘even in the ease when only one meteor is observed from two sepa- rate places on the earth's surface, because then the sought for point is yiven by tl intersection of the two apparent teajectorio: ‘The determination of reasonably reliable values for the velocity of moteors from visual observations is extremely difficult (if aot im- possible) because of systematic errors peculiar to this type of ob- servation. Only photographic observations with special devices for tho registration of velocity give reliable results. Because of the attraction of the earth, the unperturbed trajectory of the meteor is bent toward the center of the earth, and a corro- 356 eset tim sm aster ce a ce haa Determinetion of Meteor Orbits 357 sponding corraction must be inserted into the coordinates of the radiant. On tho other hand, as a consequence of air resistance, the motion of the moteor is roturded, and its determined velocity be smaller than it was prior to the entrance of the meteor into the terrestrial atmosphere. The calculation of this correction is diffi- cult because of the scarcity of reliable velocity observations and fon account of the theoretical complexity of the problem. According. to some data obtained from photographic observations, it is possi- ble to assume that the influence of the braking of moteors in tho atmosphere is significant for the determination of tho original velocity. IE the observations do not permit the determination of the veloc- ity of the meteors, their orbit can be calculated novertholoss. In fact, if we considor that the somi-axis of the orbit is given before hand, there is no need to know the absolute valuo of the velocity. The ‘semi-axis of the orbit can be considered as known whon the meteor shower is related to one or another petiodic comet. In the absence of similar data, it is possible to consider the orbit as a parabola. 96. Celeulation of « Meteor Poth in the Atmosphere For finding the trajectory of an individual meteor or bolide in the terrestrial atmosphere, it is possible to make use of the ob- servations of its apparent path ftom two or mote points. Precise rosulte can be obtained only from photographic observations. Visual determinations are much less reliable and are often burdened with large systematic errors. The criterion for the identity of a meteor observed from different points on the terrestrial surface is the simultancity of the observations. If the points aro not far from each other, the apparent paths of the metoor will not diffor too much. In addition, for the identification of a meteor, still other indications such as brightness, color, the presence of a trail, ote., can be of In the majority of casos, it ix difficult to establish which points of the apparent paths of a motoor observed from two points correr spond to each other, or, in other words, are simultaneous. There fore, the method of determination of the true trajectory must be so devised that it would not be necessary to rely on the simultaneous points of the apparent trajectories. Let £1, my fy and &5y nar Gy be the geocentric rectangular co- ordiaates'of the points Py and Pz at which the motecr was observed, 358 The Determination of Orbits in the equatorial system, in which the € axis is directed along the Groonwich meridian, We can write f= 94 608 87 608 Ly, Ey-= pon 605 Bf C08 Ls nimsow sfsinc) qos so gisietsy }a29) =o, sin oi, La" bon 8in 33, J where poss 9% and poas 8% are the geocentric radius vectors and the geocentric latitudes of the points P, and Pa, and Ly and Ly aro theit longitudes. ‘These coordinates, for the given points, are, of course, constants and can be calculated once and for all.” On the meteor" apparent path observed from the first station, we shall select two points (romoved as fat as possible from each other; in other cospects their choice for the determination of the trajectory doos not matter) with the coordinates 41, By, and aya, Bya. We shall make a similar choice on tho meteor"s apparent path observed from the second station and we shall designate the coordinates of the chosen points by diay, 84, and das, San We note, for clarity, that in this way we shall take, gonorally speaking, four different points on the true trajectory of the metaor. Tho direction cosines of all the points takon by us on the celestial sphere are expressed by the equations ip = 608 Bi C08 (S ~ eeu) Buy = 008 By sin (S~ ain), PsR=1, 2% (122) ca = sin Ban where 5 is tho Greenwich sidereal time of the observation of the moteor. Lot the radiant of the meteor have the coordinates A and D on the celestial sphere to which correspond the direction cosines 2 = 0s D cos (S~ A), b= cos D sin (8-4), (za) c=sinD. Wo shall consider the motion of the meteor to be roctilinoar dur- ing the time of observation. Thon, ovo directions obsorved on the apparent path of the meteor from the first station and the direction to the radiant must fie in one plane passing through the point of observation and the trajectory of the meteor. ‘The same conclusion must be made also for observations at the second station. Analytically, this ean he weitten as: | ' i Determination of Meteor Orbits 359 ab 6 { clio) ay bn eul= aay bar cay Qn bis oie a2. baa con Hence, the equations follow dat mb+ne=0 laat mgb+nge=0 aay bur eur cu ay an by Ee eee tee ee [ere ec bia ea fei ain a2 bia a tge [ot om [ex ax a1 bes 7 baa caal” - 22 @aa|" mee 22 baal - From these equations the ratios of avo unknowns to tho third can be found by simple numerical methods, for example, a” = © and 3 b¢ = 2. ‘The solution is impossible if the trajectory of the meteor lies in the plane passing through both points of observation. This is evident geometrically, because the great circles representing on the celestial sphere the path of the meteor as it appeared from bath stations, will coincide, while their intersection must give the point of the radiant (a graphic method, using a gnomonic-projection map, is, similarly, the most rational way of finding the coordinates of the radiant for visual observations, which do not have as much pre- cision). In the numerical solution, the coincidence of the planes is expressed hy the proportionality of the coefficients of the equa- tions (12-4), i.e., by the proportionality of the minors designated by the samo subscript letters in (12-5). From tho condition that a, , © are the dire te, oe Ys, oe tL Varia Veta T! OO Va whence, if it is necessary, wo can calculate the right ascension A and the declination D of tho radiant. It should not be forgotten that all the previous equations refer also to the anti-radiant. There- fore, it is useful to watch carefully that the coordinates of the radiant should correspond to that hemisphere of the sky from which the metoor moved. 1 cosines, we find (22-8) 360 The Determination of Orbits Tho expression of the rectilinear trajectory of the meteor in parameteic form will be fn fy -a9, 19-53, Se ky- 08, where the paramoter s is the path described by the meteor along the tajectory. For all four of the observed directions, it is possible to write similar formulas in the following form, if we set ¢ = 0 for tho first observation and designate the distances from the stations to the motoor by d: far hitandus farm Gxt Cardy baad Tem + Beds May = + Baytay= 15> bays Sua Sit Cadiy Sar= Set Corday S43 C8 ay fia Ext Giada 611~ 28;9> bae~ bat Caatlan™ E1)- 98000 un Ban (12-7) Mage M+ Brshia=My,~ P8495 Taam M2 boat Sia Cyt Canta Cusm CSaa1 Coa Ca + Coatlan = C4, ~ 28 a2" The solution of these fundamental equations gives all of the un- known quantities dary Qyay Gaur daar Ex tary Say Siar Says Saar @ 8, © if we add the condition that a? + 574¢?= 1, Let us make sure that oquations (12-4) follow ftom (12-7). Eliminating €45 944s C1 from the left group of equations (12-7), we obtain Sieae aude = adi, Saab = Basdes ~ Badia ses endiyn ead. Multiplying those equations, respectively, by 1, ms, ny» and auding, wo find the first equation of (124); sinew the coefficients for di: and diz aro determinants with two identical rows, both of these determinants equal zero. By analogous means it is possible ‘ obtain the second oquation of (12-4) from the right-hand group of equations (12-7). ‘nd so, me consider now a, b, ¢ 8 known, and for finding the tomaining unknown we combine the leftchand upper group of equa tions (12-7) with any group of three equations from the right. From the six equations we first eliminate &,,, 2, 43 then we solve numerically the romaining throe equations, determining diy, dass 821 | | Determination of Meteor Orbits 361 OF din, dans gaae Aftor this it is easy to find £yy, my Cy, by the simple substitution of the value of ds. The left-hand lower group of three equations (12-7) will serve for finding dy and size For this purpose, strictly speaking, it is sufficient to take only two equations. ‘The third must he satisfied identically (as a conse- ‘quence of the fact that wo havo introduced the known values of a, 3, ¢, which must satisfy our system of equations). This will give 4 good means of control of the calculations. In exactly the same way the last unused group of equations to the right will give daa and sap oF dar and 21, while one of the equations will serve for control. Thus it is possible to consider the problem of determination of the moteor trajectory as having been solved. If the observations were made not from two but from three or more stations, as often happens in the case of bolides, ox if the apparent paths of the meteor were obtained not by two but by a greater number of points, it would bo nocessary to supplement theepreceding reasoning only slightly. Namely, it will be necessary to do the calculations c= cording to formulas (12-1) and (12-2) for all of the stations and for all of the observed points. For the determination of the radiant, it is possiblo, theoretically, to combine as many pairs of equations of the type (12-4) as it is possible to take pairs of points on the Apparent paths of the meteor for all of the stations where the meteor ‘was observed. However, this would be quite cumbersome, and it is simpler to limit oneself to obtaining the coordinates of tho radiant according to a few rather reliable points, with the possible excep- tion of certain particular eases. Equations (12-7) must bo written for all of the observed directions (this also applies to the ease whon the observations were made at two stations). Rach addi tional direction will give three new equations from which will ap- pear the corresponding di, and sj. but the control, generally speaking, will no longer yield exact agreement. Thus, it is possible to find, for any point of the apparent path of the meteor characterized by a certain value s, the geocentric coordinates , 7, ¢, after whick, returning to (12-1), we can do termine the geocentric latitude ¢” and the radius vector p of this point, as well as its tongitude L. We obtain the height abovo soa level from the equation A= (p= pg) cos (¢ - 4°), where py is the geocentric radius vector of the point on the terrestrial ellipsoid with coordinates ¢” and L. The factor cos ( - ") arises because the difference p ~ p, must be projected onto the direction 362 The Determination of Orbits of the plumb line, forming the angle g ~ ¢” with the geocentric radius vector. Since ~’ < 12’, this factor can he neglected and tho hoight will be simply hap =D (22-8) If the height obtained is unusually great (too small a height may, sometimes, also he questionable), this in itself can place under suspicion the results of calculations or the observational data, ‘To find the point whero tho trajectory of the motoor intersocts the terrestrial surface, we write the equations of this trajectory f= yn os aay ds, Sa hyn em From this we derive 9? ~2(ak,, + bn, + c0,,)8 + (ER, +9, + SF) = ph = 0, (12-9) where p, is the geocentric radius vector of the terrestrial ellipsoid at the point of intersection we seek. Its value is not known be- forchand and in essence the problem should be solved by succes sive approximations. However, p, does not. change very much with latitude, and it is not difficult to nssign lo it a sufficiently correct value immediately, if the latitude of the point of intersection can be estimated approximately. Moreover, an error of several minutes will be nogligible. Two solutions are obtained from equation (12-9). ‘One should take the smaller root for s; the second root corresponds: to the exit of the continued trajectory of the meteor from under the surface of the earth and it has no real meaning. If the observations are accompanied by procise marks of time, for example, if the teaek of the meteor on a plate was intortupted during the time of exposure by known intervals of time, and if the ends of the separate segments of the track wore usod in the meas turoment, then the velocity of the meteor ean he calculated. In out rethod Usis is accomplished quita simply. For example, let he interval of time between points on the trajectory with values of = equal t0 8; and siz bo tir ~ tay Then the velocity is oun wy. Hai, 12419 an tie . y Such esleulations of velocity ean be done rapidly for any sepa: rate souments ofthe path of the meteor, and the eosults wil show the variation of its velocity asa consequence of ai resistance. Determination of Meteor Orbits 363 97. The Influence of the Earth's Attraction and Motion Perturbations, introduced by the attraction of the earth, cause noticerble change in the motion of a meteoe when it approaches the terrestrial surface, Thus, they must be taken into account even in the initial determination of the orbit (whieh, for meteors, usually hhappons to bo also the final determination). In immediate proximity to the earth, the motion of the meteor is assumed to be rectilinear and uniform, which simplifies the problem. Proceeding in this man- nor and assuming the earth to be motionless, we do not take into ac- count the attraction of the sun. This does not cause significant ef- tor, because the motion of the meteor is considered during such a short time. Moreover, the effect of the attraction of the earth, which we seck, is slight and thoreforo can be found independently from the motion of the meteor around the sun. As a result of this, we conclude that the meteor describes a conic section around the earth, namely, a hyperbola, because it arrives from infinity (having a finite velbeity at an infinite dis- tance). We refer to Fig. 27, on which 0 is the center of the earth, P is the observer, BUIIC is the hyperbolic trajectory of the meteor, where sM is the end of its visible path, and {1 is the point of peri helion (located inside the earth). For simplicity, we assume that he meteor was observed at the observer's zenith. If this is not true, one should relocate the point P correspondingly on the ter- resttial surface. This does ‘not influence our further discussion orar in view of the smallness of the / mentite ieee A Fig, 27 364 The Determinotion of Orbits (this anglo, because the motion of the meteor around the earth is planar, lies in the plane of the drawing which is simultaneously a vertical plane for the observer). In order to obtain the position of the radiant, corrected for tho attraction of the earth, it is neces- sary to add to the observed zenith distance of the radiant the ze- nith attraction, expressed by the angle Moreover, the attraction of the earth will change the velocity of the meteor. ‘Let us take equation (2-28), in which wo shall sot the mass of the earth equal to m and neglect the mass of the meteor reeetd) But since the acceleration of the attraction of the earth (which hore we can freely consider as a sphere) is expressed by the for- mula in oe (az-t1) ri wo con write, instead of the preceding, 2 gp2(2_2 w? a 902(2- 2), 124 9o*(2-2) (12-12) Hore it is evident that ¢ =, + A is the distance of the meteor from tho conter of tho carth, which it is possible to consider ap- proximately equal to G470 km, taking for the mean radius of the earth py = 6370 km and for the height of the meteors = 100 km. Usually’ the correction for the attraction of the earth is calculated under tho assumption that the moteor is at the terrestrial surface. This, of courso, is incorroct, and the existing tables of corrections for zonith attraction require, therefore, some corrections. Assuming the motion of the meteor, unperturbed by the eatth, to be rectilinear and uniform, let us suppose that its velocity under this assumption is equal to U, which at the same time evidently will correspond to the right-hand side of (12-12) if there we take pao. This gives us (12-18) (a is negative) and WU? 429m (az-14) Having taken the value 0.00952 km/sec? for g, we shall find, for the distance of the meteor from the center of the earth assumed Determination of Meteor Orbits 365 by us, that 29p = 198.2 km4/see®, Using this, we can calculate U, if the value of Wf was obtained fcom the observations. ‘The re- sult is expressed in those units which are used by the observers of the meteors. However, when the value of U is to be used for tho calculation of the heliocentric orbit of the meteor or the swarm, it will be necessary to make a transition to astronomical units and to days. For finding the zenith attraction we turn first to the integral of arens (2-36): 0? 2 aa Vp = kya Vale (az-asy a But because » 2° = sin 2, then 0 a W sin a, Um Vin VB = pW sin, (12-16) whonce by using (1211) it follows that (12.17) (12-18) In Fig. 27 we designate the angle 10D betwoon the major axis fand the asymptote of the hyperbola by (180° ~ ¥) in agreement With Section 17. Then we shall have =180° - ys, (22-19) where ZIOP =~», since the motion of the meteors ends before porigeo is reached. From the equation of the conic section we have Further, using Fig. 27, chat cine = =P, (2-21) 366 The Determination of Orbits Lot ws assume, for sbirviation, that 1.1, a then we oan wrta on the basis ofthe preceding and of Section 17 ecose = tei sin? 2-1, » esiny =~ eins Hsin 2 cos 2, ° (22-22) cos b= 1, = u esinyavetie Yee sing. Ye i ae According to (12-19) we find: tan Lo = 2" A180" - y+ 2) ~ tan ds 2°" Tyan gas" —v+yunge = OP) But (12-22) gives tan Sao" - yoy wotn Leo) = coedteass Hsin? 2 __, Wsine > (12-24) TW cosa after which we have 14. Wain 2 - (0 +N cos 2) ta . 2°° Ty Weos 34 W sine tan ja sina cin} 2-U-W cos 2 ‘Weos 2+Wsins tan ye 1 an 5, or, by simplifying, 1 1 oo tan =p e pin ge (12-28) This formula, which is usually derived somewhat differently, gives, on tho basis of the geometric properties of the hyperbola, a correction which must be added to the observed zenith distance of the radiant. For its practical application, it is necessary to find W from observations, and then to determine U from (12-14). If it is Determination of Meteor Orbits 367 impossible to derive the velocities of the meteors from abserva- tions, it is possible to assume that they are moving around the sun with parabolic velocity. This will pormit the calculation of the geocentric velocities U and WW. The vis viva intogral gives V? = 22/1 for the parabola. Taking fone astronomical unit = 149680000 km, and the equatorial semi- axis of the earth @ = 6378.4 km, we find for the parabolic velocity the following expression: v- B14 im/s00, (22-26) ve where is the radius vector of the eneth, oe, interchangeably, of the motoor. In observations of meteor showers, the equatorial coordinates of the radiants are usually sought. In order to introduco into thom the correction for the zenith attraction it is necessary either to make use of the differential formulas commecting the variation of the horizontal and equatorial coordinates, ot (it the correction is large) to perform a transformation of coordinates twice. It is ox tconoly desirable in large-scale calculations to simplify somehow those procedures. Because great precision is not required here, it is possible to make advantageous use of tablos or nomograms constructed for the latitude of the observer. Furthor, it is necessary to take into account the influence of the motion of the earth, because the velocities of the meteors ob- tained feom observations ‘re relative, ‘Tho rotation of tho easth causes an effect quite similar to the daily aberration of colostial bodies, and we can apply tho well-known formulas from spherical astronomy: 360° py 1 Aw cos 5 2 — cos ¢ cos - a s6ie4 ee 28 con y 0s (8- , 22m 360° py 1 Abe 2 cos g sin (S~ a) sin 5 86164 Ce 28° 62 W cos g sin (S~ a) sin 5, in which the velocity of the meteor is ontered instead of the ve- locity of light, and the result is obtained in dogroes. ‘This correc~ 368 The Determination of Orbits tion can turn out to be quite noticeable, and in more precise ob- servations it is necessary to take it into aecount. > Fig, 29 We denote the velocity of the meteor with respect to the sun by V, the velocity of the earth hy Vo, the angle hatween them by 180° N, and the angle betwoon U and Vo by 180° ~n. Then we obtain, according to Fig. 28, Usin n= ¥ sin Ny } (29-28) Yo sin n = V sin (V~n). Let points 4, R*, and & (Fig. 29) represent the traces of the directions Vo, U, and V on the celestial sphero. ‘The three men- tioned points’ aro situated on one great citcle, because tho three corresponding directions are situated in one plane. The point 4 toward which the orbital motion of the earth is directed is called the apex. It is situated on the ecliptic and we can find its helio- contric longitude Lo. If L and R denote the heliocentric longitude of the earth and its radius vector, and © = 180° + L is the longi- tude of the sun, given in astronomical yearbooks, then the com- ponents of che velocity of the earth aro found by the formulas a ‘9 008 Lo = 2 (R cos L) = Yo cos Lo = £( ) Vo sin (Lo ~ 9) i | { | | | Determination of Meter Or xe Vo cos (Lo - 0) =~ 28, ae tn (Lo ~ 9) = 4 a8 Ra Then wo have Re oll Tye cos (L =n) 0s (Oa) where ¢ is the eccentricity and m is the longitude of perihelion of tho torrestrial orbit. Because ¢ is small it is possible to negloct its square, after which we obtain Ra14 6005 (O-n), ak Ferre sin ~m), etn (Lo ~9) =e sin (9 ~ 8), o ) © -A), tan (90° + Lo ~@) =e sin @ =a). Replacing the tangent by the angle, we find finally Lye 9-90" 4 are if 3 sin (@ =»), (12-29) where it is possible to pot (for the equinox at the boginning of the yonr of the observations) 7» 102°5’ + 1’.03 (¢ - 1950), —* wee oT We designate the ecliptic coordinates of the point R’ by A” and A’, of the point R by A and A, and the inclination of the great circle AR‘R to the ecliptic by y. From the triangle ABR’ (Fig. 29) we obtain sin asin y= sin 8’, sin n cos y = cos A” sin (A ~ Lo), (12-80) 08 n = cos A” cos (N’ - Le). ‘The corresponding triangle ACR gives 08 B cos (A Lg) = cos N, 605 8 sin (A Lo) = sin N cos y, (az-s1) sin 8 = sin N sin y. 370 The Determination of Orbits In order to use these formulas, it is necessary to calculate with the aid of the equation Yo sin (N =n) = 2 sina (19-39) Lot us agroe to consider that NS 180°, ies, sin N20. ‘That is possible if y takes on any value from 0° to 380°. For the determination of ¥ we apply the formula derived fcom Pig. 28: V2 U4 V3 ~ 90V, c08 n. (12-83) Thus, if tho value of the volocity of the meteor is obtained fro tho observations, we calculate x and y by equations (12-30), after which formulas (12-83), (12-82), and (12-81) give the coordinates of tho true radiant. If the velocity of the meteor i unknown, it is necossary to proceed differently. Let is take the oxprossions for the velocity of the earth and of the meteor 2. a(2 2242/2 2 . ° eG } al ) from which follows (2-34) Equation (12-34) requires the value of a, but this quantity can be found only in that case when the time of revolution of the meteor swarm is known from observations of the intensity of the swarm in different yoars or from considerations about its connection with some periodic comet. Then it is possible to apply the equation a= pr, (12-35) where P is the period of revolution in years, In the remaining cases we have to assume that the orbit is a parabola, although this as- sumption is hardly valid since calculations according to photo gtaphic observations show that swarms are often encountered with very short periods. In the case of a parabola equation (12-3) be comes simpler: (12:38) Determination of Meteor Orbits an Knowing the coordinates of the true radiant and the heliocentric volocity of the meteor, we can begin to determine its orbit. 98. The Determination of the Orbital Elements Within the limits of precision with which the orbits of meteors can he calculated, it is necessary to assume that the observer and the meteor were located in the plane of the ecliptic at the time of observation and that their position is given by the coordinates of the center of the earth. These coordinates are as follows: 2 = Reos L =-R cos ©, ye Rsin L=-Rsin , 200. Tho radiant indieatos to us the point on the celestial sphere from which the meteor moved. Therefore, the components of the velocity of the Inter with respect to the sua will be: de & __V cos cos A, Gs 108 8 cos d, ay ae 7 7Y cos B sind, de & .-V sin B. 7 in 8. Applying formulas (2-81), we find yd 22 2. pypcos i= BV cos f sin (X- 9), o -vG- ae 105 B sin (A ~9), oft, sin écos 0 = RV sin B cos , > (12-81) a ae Ying in © Ge = ky sin i sin Q = RV sin 8 sino. ‘The last two equations give (since sin ¢ > 0) Qo, eeeeal 12-88 Q= 180° +o, sin 8 <0. \ ’ If the radiant is situated toward the north of the ecliptic, the earth passes the descending nodo of the orbit of the meteor at the moment of observation. If tho radiant lios to the south of the ecliptic, the earth crosses the ascending node. 32 The Determination of Orb Substituting (12-88) into (12-7), we obtain LYP c0s i ~ RV cos fi sin (d -™} Ayp sin ¢= + RV sin p, (12-39) Whore the plus sign in the Intter equation is taken if sin > 0, and tho minus sign corresponds to the case of sin f <0. Honce, we dotermine i and yp. For finding a (if it was not assumed before hand) we use the equation arising from the vis viva integral: tev aR (12-40) after which we proceed to find ¢ and vs On tho basis of (12-20) we can write PB eosin y= pH, a *® OF at sine EHF 8 da ae” eR ae ae te, 2 _ | esin v= etn A=) cos ¢, ne ecosun 24 “ a q=a(l—e) (12-43), If the meteor was located at the ascending node of the orbit, w= 0°, and if at the descending node, v = 180°. Consequently, in agreement with the statoments made above, os uno = 180° =, peal eer % B&O oD In the case of an elliptic orbit we find tho time of perihelion passage (which is usually of no interest to us and can he omitted) by using the well-known formulas: t { | a =tes (15° - fe) an Eo, (125) 2 Baesin ate Ege ) For the hyperbola, analogous formulas can be borrowed from Section 17: 5 aieeet tanh 2H = VE=2 tan do, 47 anh SH = Vat tant (12-47) A @-T) (12-48) e sinh =e Taw If the eccentricity is close to unity, jt is necessary to make use of formulas in Section 93. Equation (3-65) givos (Pros Peo), (32-49) Ay antan de, (22-50) If a parabolic orbit is being determined, tho formulas mentioned above permit important simplifications. First of all, let us take the obvious relationships: ° pcos? bo, ye ot a- $= Roos? 30, ak (2-81) Substituting these expressions into (12-89), we find cos 4 cos i= cos A sin (A~0), feewiven sn-O) Gay cos Ju sin i= |sin Al, ftom which i is uniquely determined, because cos ty>0. The sign of v can be obtained by using equation (12-41), according to which sin © has the same sign as cos (A = ©) (because all of the remaining quantities are positive). Hence, the rule: v>0, if =90" 0. Control: sin n anil 608 n must be consistent ¥2002 4 V2 a0¥, 008m, “© Determination of Meteor Orbits 379 Control: U sin n= ¥ sin N. 08 B cos (A ~ lio) = cos Ny 05 B sin (A = Lo) = sin N cos y, xu sin B =sin N sin y. Control: sin @ and eos must be consistent. If the velocity of the meteor is not obtained from observations with sufficient accurney, but the period of revolution of the swarm is well known, the calculations are begun, after finding the initial data, with the determination of the velocity a- Ps, (ib) If tho period of the rovolution is not Rnown, the orbit is takon to be a parabola: (ate) If it is desirable to introduce the corrections for the daily aber- ration and for the attraction of the earth (that eannot be done with- out the values of velocities of the meteors), Vo is calculated by using (XI), and then (VII) and ©08 Do cos Ay = cos Lo, 05 Do sin Ao = cos € sin Lo, sin Do =sin € sin Lo, cos n= sin D sin Do + cos D cos Dp cos (A ~ Ao). After this (XII) is calculated and U2 = V2.4 V2.4 ov» on N and formulas (Il) ~ (VII) and (IX) are successively calculated. However, it is not necessary to compute formula (X), and an al- ready known value must he taken for ¥, Next, (XII) and (XIII) aro calculated. If the corrections for the attraction of the earth and for the daily aberration aro not introduced, after (IIb) or (Ile), one calcu- Intes (VI) = (IX), (XI), (XIN. 380 B, Determination of Elements For an elliptic o hyperbolic orbit: ‘The velocity must be expressed in units derived from the astro- homies it and tom sani of tne onan to 3 any; fortis he value of V, expressed in kilometers per second, is multiplied by 1 km ze 86400 faetroanutisal(anitge 0.03856 [8.5258] With the preservation of tho provious designation V. Qed, if B>0, cera Titec) om f/placsie= kUjeoslafsin (i “> XV Wisin 1=RY [sin gl, oe 12 ys 1.2 y, avy if as not takon as known boforhands pe eee (XV) ecos ae 2 ao Control: p= a(1 ~ 2) = R?V? {1 - cos? f cos? (A ~O)}. g=e(e), @ = 180° - if B>0, (XVII) wane, if p0, arr +o, wacad | 0s $0 cos # = cos A sin (A~ ©), cos fv sin i= |sin Al, (XVa) tan do = etn (= 9) c08 Control: tan Jv and cos 4» must correspond to the same angle. = Beos? 40, @ = 180° - », if B > 0, or egal xv) For the calculation of T the formula vo/. 1 4) ,1\ e-7 fen joe geen? Se) = 2 «xva) is used, and the tables which replaco it. 100. Exomple Telescopic meteors of the Scorpionid shower wore obsorved by A.M. Bakharoy in Stalinabad during five evenings between June 12 and 22 of 1947, Several radiants, with sufficiently good agroement among themselves, wore obtained, and their average was taken. The Universal Time of observation refers to the mean moment of all the observations, but the sidereal time was taken as the atith- metic mean of the sidereal moments for each of the days, i.e., without respect to the calendar dates. Thus, it does not have to agree with Universal Time, It is assumed that the orbit is para- bolic. The initial data are: 1947.0 6 ‘ eee ee ee logiai ol 1389.6 1947 June’ 16,17 1699" 2519.4 ~18°.7 1.016 0.007 842.2 382 The Determination of Orbits He 1-28 9492 xt2-2 9.993 vn Sy o.o8e 20.60 LaTd sin(@®~n) 9850 ¥e Jf 9.866 Vo 1407 5 sin @-m) = 08. On m0 =-5 8 Zoan6 a1 Mb eos Lo 9.098 IVb sind 9.508XI_sinn 9.991 sin Ly 91028 cos D 9.878. sin =) 9.887 cos € 8.983 cos (4g) 9.389" Wa tits sine 9.600 1 s.1g2 208 Do con dy 9.998 1 6'3280 cos Dosindy 8.9802 conn 9.209 sin Dp me tones con De i Po vb Y 1001 Tsing 9.705 IV W = 26.0 ave 3389 cong onde coh 9913 Sake w-u 0:380 BV cosN==2004 sin(S—A) Baton WU LTT = 1746 cos(S-4) 0.000 quotient — 8.669 vie 359 T9399 mda 9.ta7 t= 601 Tort tan 3h 8.406 0 Wem HI 9.201 Ba2%9 WV aiana| ¢= 602 sin ssina 84090 sins cosa 9.028 ‘ins v sing 9.938 VIcos(S—4) 0.000 VII cos A’ 9.519 cos 5.80) Bim A} BaKEn Sin A’ 8. TEm sing 807m B662) TAB 1 9.1660 cone 0.000 ® 1430 Tt 9.0060 T8589 26.827 9/908 Mm 8'300" uta ‘Aa =—088 W953 meer AD = "0.6" cos eos A B:tatn . W983 fa’ = 250.1 coe B sin N 9.070 cos D’sin(S a’) Badan SPCC A arg sin’ 8.046 cas D’eca(S— A") Sana con 8’ 0.000 in D?8.860n Xa 250% cos DY 9.068 Seata-127 402 2515 pee a8 career wee oe 383 IX etoasere Nt ost XI ain o.ts8 See cme ees coh =L9) Valin sinW =n) 9187 coe Bain Lq) 0.1800 cinnainy 8 Wa tise sins 7.809 sinn cosy con 8 0.000 coe h eo = 214°.8 ‘sion R208 siny pasos eos y XIV Q=s4%2 XVIa v=-69%.0 Elements XVa_ sind =) 9.018 o= 9.0 @ M98 cin(h 2) 0.88% @ ase Ra ae2 b 1047.0 coa-de con 9.9160 te 04 cos }e sin 7.809 log g = 9.809) cosy 9.916 4084 cos # 0,600 tande 9.8370 Tho radiant is located very near to the ecliptic; therefore the inclination of the orbit obtained is very small and tho elements and w are determined very inaccurately (although this inaccuracy is not noticeable during the calculations). It is possible to be more or less certain only with respect to r= +S, i and g. Fur ther, it is clear that the Scorpionid shower is observed annually for & rather long time without giving a sharp maximum for the num- her of meteors. They move almost parallel to the ecliptic, and they can meet the earth for the extent of a significant segment of its orbit. Of course, hore it will be necessary to assume that the meteors met by the earth are moving not along one orbit but along 8 bundle of orbits, to a cortain dogroe different one from tho other, ut lying approximately in one plane, in the present case almost coinciding with the plane of the ecliptic. It is not necessary to assume that the stream has a significant extension in the direction perpendicular co the plano of the orbit. The calculations mentioned above can bo significantly short- ened by the use of special auxitinry tables or graphs. We notice further that the logarithms, repeatedly occurting in the calcula- fons, were written out on a strip of paper in order to avoid addi~ tions’ and subtractions of quantities widely separated feom each other, of the frequent copying of them. Great simplification is also ‘obtained if the corrections for the zenith attraction and the

Potrebbero piacerti anche