Sei sulla pagina 1di 12

June 1927.

On Two British Earthquakes.


TABLE V.--continued.
Values of

483

&.
7

sins

, $ =
00.

. ' 5 . 1

30".
'000

45O.
'000

6. 0 '
'000

75O.
'000

goo. '000

'0

'000
'000
'000

'000

'I

'2

- .I05 - .072

--.208

-.306
-.251

'3 '4

'000
'000
'000

.005

-.I55 -.038 .109 .264 .408


'522

-'390 -.346 -.198


'000

-.108 '077
.275

-.446 -'418 -.275

-.46fj

-.444

-.305
-.109
'111

'5 6
'7
4

.077 .163
,242 '303 '332 .300
'000

-'077
'I43

.216 '424

'000
'000

.462 .612 691 636


'000

'595
.693 ~649
'000

'000
'000

378
'525
'000

'9
1'0

'357 '537 .646 618


'000

'327 .5IO
6 2 2

600
'000

'000

On Two British Earthquakes. By Harold Jeffreys, M.A., D.Sc., F.R.X. (Plates 11-13.)
(Received 1927 May
12.)

I . Intensive discussions of near earthquakes, as distinguished from world-shaking earthquakes, have hitherto dealt mainly with those occurring in Central Europe. The earthquakes that took place near Jersey and Hereford in the summer of 1926 have provided an opportunity of comparing British and continental conditions directly. Further, it has seemed worth while to examine in relation to them some questions suggested, but not decided, by previous discussions. A definite pulse of type 8, (a distortional wave propagated in the granitic layer) was recognized in the records of the Oppau explosion, but has received little attention in true earthquakes. The pulse found by Gutenberg in the South German earthquakes of 1911and 1913, and by Conrad in the Tauern earthquake of 1923, and called P* by Conrad, also required further investigation, since Gutenberg's results were hardly enough to determine its velocity, and it is too important to be dealt with adequately by a single series of data. Records were therefore collected from most of the chief observatories of Britain and Western Europe. Of those to be mentioned, Uccle, Zurich, and Hohenheim sent photographic copies. The others lent original seismograms. Measurements were made with a millimetre scale under a lens and in a strong light. The records were read to the nearest second, though several of them would probably have been capable of greater accuracy.

484

Dr. Harold Jeflreys,

I 9, .

2. The Jersey Earthquake of 1926 July 3o.-The epicentre provisionally adopted was a t St. H6lier. On reading the most definite pulses recorded a t the nearest stations, it was quickly found that most of the records were incomplete, only some of the waves already known to exist being large enough to be distinguishable. In the first place attention was therefore concentrated on the beginning of the largest pulse, which was very sharp and was soon found to travel out with a velocity of a little over 3 km./sec. From the close agreement of its velocity with that found from the Oppau explosion it was identified with S,. When the times of arrival were plotted against the provisional epicentral distances, the times a t the British stations, in comparison with continental ones, were found to be systematically late by 4-5 s. The stations were all in the north-east quadrant or just outside it, and the greater part of the difference could be accounted for by supposing the epicentre to have been in some direction between east and south of St. HBlier. Acting on a note in Nature for August 7, I assumed the epicentre to be 15 km. east of St. HQlier and revised the distances accordingly. The records were found to show six recognizable pulses, as shown in the following table. The situation of the epicentre is 49' 11' N., I' 42' W.

TABLE I.
Station.
Type.
A(km.).

Times.
A

0-0.

Kew G. Paris W. Uccle W. Bidston M.S. Stonyhurst M.8. De Bilt W'. Strasbourg 19-ton Heidelberg W. Edinburgh 111.5. Zurich Q.P. Hohenheim Ma. Ravensburg Ma. Dyce M.S. Hamburg W. Wien 17 4.

282 297 469 59 474 519 586

..

..

P.

P".

..
70
74(?)

P,.

R.

s*.
lo(?)

s .,

P.

44

.. ..
..

44 48 79

64 66
105

75
82 130 133 148 165 198
212

74(?)
120(?)

.. .. .. 1-2
o
+I

P".

P,.
-I

s.
+I

S".

5r
0

o
o
-I

o
-I
-I

o
-I
+2

+2
-2 -1
+I

80 109 125 7 7 ( ? ) .. 113 .. .. 100 128

.. ..
+2

..

..

+3 o

+4

..
-2

-4 -4
o
-I
-2

..
+2

..
o

-3
0

689 8 9 ( ? ) 104 737 .. .. 748 776 784 841


889

121

..
98

..
..

..
.. ..

168 195 137 177 .. .. 138 178 .. 136(?) 150 190 2 2 0 118

..

.. ..

156 182 166 ..

219 223 226 246 262 279 391

,. .. .. .. .. ..
o

..
-I

-2 +2
-2 -2

+I

,.

..

942 1300

.. ..
I .

..

..
215

..
248

..
..

292

352

.. + 7 + I - I -I .. .. .. .. .. .. .. .. + I - I .. .. .. -6 +6
S. "
%. 7

.. ..

+I

+2

+3 +6

The zero of time is 1926 July 30, 13h 2om 08. above comparison are
P. P".

The-calculated times used in the

pa.

9.

-I+A/7.8

-5+A/6.3

-1-kA15.4

-2+A/@35

-6+A/3'7

-IO+A/3.3

The initials under " Type " refer to the instrument used, thus: G. =Galitzin, W. =Wiechert 1000-1500 kg., W'. =Wiechert 200 kg., M.S. =Mihe-Shaw, Ma. = Mainka, Q.P. =Quervain-Piccard 21,000 kg.

June 1927.

On Two British Earthquai&.

485

On inspecting the columns of residuals we see that they are very satisfactory. The readings and computations having been carried out only to the nearest second, residuals of f I s. are not significant. In every column except the last the majority of the residuals are in this range. In the last column (8,) the majority lie between - z s. and + 2 s. inclusive, and they are not systematic. Since this last pulse was the only one used in finding the epicentre we may feel confident that the adopted epicentre is correct within a few kilornetres, and that the adopted formuls for the times of arrival are not systematically wrong by more than a fraction of I s. for distances under goo km. Some of the readings, however, were rendered very difficult by microseisms. This refers especially to P a t Strasbourg, and P* a t Bidston and Stonyhurst ; wxile I think it probable that the records a t De Bilt, Edinburgh, and Hamburg would have been more complete had microseisms been absent. In a homogeneous body a sudden shock within a finite volume should theoretically generate disturbances of P and S type, each of which would travel out with its characteristic velocity and take a finite time to pass over a given point. When a pulse has passed a place the matter there should return to rest. This is not the case in an earthquake. Each pulse, instead of coming to an end and allowing the seismograph to return t o rest, is followed by a long train of waves comparable in extent with the pulse itself. These may diminish in amplitude or remain nearly constant ; but in any case no disturbance except the first starts from rest. When a later pulse is sharp and much larger than the residual movement left by the previous ones, there is no difficulty in reading its time in spite of the oscillation ; but if not it may be impossible to distinguish it. As a rule, judging from this earthquake, P, Po, S, and S, can be read clearly whenever they are large enough to be visible and microseisms do not complicate the problem, though S a t Paris presented some difficulties through being of nearly the same size as P,. I?* and S* are more difficult, not being much larger than P and S respectively, and the identification of S* a t Kew, Paris, and Uccle is not quite certain. On the Zurich record the large waves following S made it impossible to identify S*definitely, though this was actually one of the two most beautiful records received. The relative amplitudes of the phases are exhibited in the following list, based on the stations with the most complete records. The first maximum displacement in each phase is given (not the first maximum amplitude). The unit is I p .
TABLE11.
P.

Paris
Uccle Bidston (N.S.)
I

..

P".
0'2

PQ

5 6
2

S.

s4.
9 8
2

I
I0

so. 18
20

..
0.1(?)

(?)

3
4
12 '
I

12

Strasbourg (E.W.) Zurich (E.W.)


Hohenheim

0.06
*.

03 ' 0.15

0'5
0.4

..
.I

Wien

..

..

..

..

15 6 5
9

<0*5

486

Dr. Harold Jeffreys,

1. 9,

The maximum amplitude was usually one and a half to three times that of S,. The natural interpretation to put on the results is that three fairly uniform layers of plutonic rock underlie the sedimentary layer ; the sedimentary layer is not itself concerned with the transmission of the principal pulses for much of their paths. The uppermost plutonic layer is granitic, and contains the foci of a11 near earthquakes pet studied. P, and S, are the direct compressional and distortional waves from the focus to the observing station. P and S are indirect waves travelling mainly in the deepest layer, while P* and S* travel mainly in the intermediate one. The three compressional waves having been recognized, the corresponding distortional ones were t o be expected. S is, of course, familiar in distant large earthquakes, but has been less studied in relation to near ones. S,, as has been remarked, has hitherto attracted less attention.? The reason for this is, I think, that its large size suggests a n analogy between it and the large surface waves of distant earthquakes, especially as the extent of the vibration following i t usualIy increases to the largest displacement on the record ; this displacement is usually read and denoted by M, but the impulsive beginning of the movement leading up to it is the true S,, and occurs some tens of seconds earlier. The intermediate wave S* is here given that name for the first time. It is, however, probably the same as that called i L by Gutenberg in his discussion of the South German earthquakes ; for he defines iL to mean the beginning of the large amplitudes and determines its velocity as 3.75 km./sec. Conrad finds velocities near 3 - 3 km./sec. and may have been dealing with Sg. The most curious fact about the f o r m u h adopted for the times of arrival of the pulses is that the constant terms in the expressions for P, and S, are unequal. The distortional wave seems to have started on its way 3 s. before the other. I believe that this difference is genuine. I we deliberately suppose that P, started at - 10s. we shall have to f reduce its velocity to about 5 . 2 km./sec. to fit t h e Kew and Paris observations, while no previous discussion has given a velocity under 5.4 km./sec. On the other hand, if 5, is supposed to have started 3 s. later we must assume a velocity of 3.45 km./sec. for it to fit Kew, and there will be a residual of 5 s. a t Zurich and more at more distant stations. The hypothesis of systematic errors in reading the records seems equally unlikely. The time scales range from 8 to 60 mm. to the minute ; any systematic error would therefore probably vary considerably from one station to another, and hence should show in the residuals ; further, it is difficult to see how any such error in reading could affect P, and S, by amounts differing on an average by 3 s. Supposing, then, that the difference is real, we notice that transmission through the sedimentary layer will not explain it, for i t would delay a distortional wave more than a compressional one and would therefore act in the wrong direction. The facts would, however, fit the hypothesis that the primitive wave sent out from the focus was wholly distortional or at any rate included no compressional part

j Cf.Gutenberg, Lehrbuch der Geophysik, 1926, p.

279.

June 1927.

On Two British Earthquakes.

487

large enough to be recorded, and that compressional movement is generated only b y reflexion at the top of the granitic layer or at t h e outer surface.t There is really something to be said for such a view. It is known that when a distortional wave strikes a free boundary at any angle of emergence between 5 4 O and 75 more than half the energy is reflected in a compressional uave. Hence, even if there was no P movement t o begin with, reflexion would generate a P, wave carrying a moderate fraction of the energy of the original disturbance. Further, it would show a delay in starting depending on the time the original S, wave took to reach the surface ; that is, on the depth of focus.$ On the other hand, if the original disturbance was partly compressional, additional compressional movement would still be generated by reflexion, and two different P, wavcs would arrive a t every station. There is no sign of this.

s
FIG. 1.-Diagram of the Probable Paths of the Six Pulses observed in Xear
Earthquakes.
Broken lines indicate waves propagabed along or near to boundaries The horizontal scale is, of course, much smaller than the vertical ; the angles are approicmately correct. aa, Sedimentary layer ; bb, granitic layer ; cc, intermediate layer (basalt or diorite) : dd, lorn-erlayer (dnnito, peridotite. or eclogite). The compressional waves may actually be generated a t thc upper surface, and not a t the bottom of the sedrnientary layer, as shonu in the diagram. The transitions are a k a probably gradual and not sudden.

There is also some a priori reason in favour of t h e suggestion. The original movement in a n earthquake is a fracture, and requires shearing stress to produce it. Simple compression would never cause a n earthquake. The fracture relieves distortional stresses, and therefore the wave sent out may be expected to be mainly distortional. I n sliding on a fault plane, for instance, the movement will be wholly distortional except for compressional movements of adjustment near the ends of the displaced blocks. I d be the depth of the focus, a the velocity of P,, and fl that of f S,, the angle of emergence of a n S, wave capable of being reflected as a P, travelling horizontally is given by cos e
= /3/u
*

(1)

I n view of the importance of waves transniitted or reflected at discontinuities with change of type, they seem to require a distinctive name equivalent to the German Wechselwellen. I propose transformed waves. Thus, when a P or SV wave is incident on a boundary, there are in general four derived waves ; each medium conveys two, one of which is in the original type and the other a transformed wave. $ The possibility of a P wave originating in this way and showing such a delay is proved by H. Nakano, Japanese Journ. of Astronomy and Geophysics, 2, 1925,
1-94.

488

Dr. Harold Jeflreys,

1 9, .

while the distance from the focus to the point of reflexion is d cosec e, and the time of transmission of S, thither is (d cosec e)//3. But a P, would reach that point from the epicentre in time (d cot e)/u, and therefore a P, generated by reflexion will show a delay in starting equal to

'With our results for the velocities, and taking the delay to be 3 s., we find the depth of focus to be about 1 2 km. The initial delay of P* in comparison with P, is z s., and that of P in coniparison with P* is 4 s. Assuming that P, originates a t the top of the granitic layer, the indirect waves travel through this layer twice. By the method of a former paper t we find that the delays determine the depth of the granitic layer as 10km., and that of the intermediate layer as 21 km. The former is rather less than the depth of focus, which is of course inconsistent with the hypotheses, but as none of the f times is certain to less than I s. there is no real contradiction. I P, originates a t the outer surface (in this case the sea bottom) part of the initial delay is caused by transmission through the sedimentary layer. We can now compute what the initial delays of S* and S should be, on the supposition that the focus was near the base of the granitic layer. That of S* in comparison with S, should be equal to
I ( ;

-py. .

(3)

where h is the depth of the granitic layer, and p' is the velocity of the distortional wave in the intermediate layer. With our results this is 1 . 3 s. That of S in comparison with S* is given by

(4)
with an obvious extension of the notation. The factor z has been restored since S must pass twice through the intermediate layer. We find that this is equal to 6 s. The total delay of S in comparison with S, should therefore be between 7 s. and 8 s. Referring back to the formula? for the times of arrival we find that the delay found from observation was 8 s. The observed time of S* is about z s. later than the calculated time ; but as this was the most difficult wave to measure accurately the difference is not serious. On the whole we may say that the delays of the distortional waves are consistent with the results found from the compressional ones. We may remark that the passage of S, into the lower layer should also give a transformed P wave. But for an S, wave to give a transformed P wave travelling horizontally its angle of emergence must be about cos-l 3.317.8 or 65", and at such steep incidence the transformed waves are probably small. This angle, on the other hand, is near that required to transform S movement completely into P movement when

M.N.R.A.S. Geophys. Suppl., 1, 1926, $35-402.

June 1927.

On Two British Earthquukes.

489

the boundary is a free surface, and hence the important P is probably derived from P,, the latter itself being a wave transformed at the outer surface or perhaps a t the bottom of the sedimentary layer. 3. T h Herefordshire Earthquah of 1926 August 14.-This was a decidedly smaller earthquake than the Jersey one, and accordingly was not so well recorded a t great distances. On the other hand there were five observatories within 200 km. of the epicentre, all of which got useful records. It was found by trial that the times of arrival of So were in accordance with an epicentre near Leominster, which proved to fit also the observations of Po. As for the Jersey earthquake, Po was found to show a slight systematic delay, in this case about 2 s. The zero of time adopted in the following table is gh. 58m. 0s.
TABLE 111.
Times.

Station.

Type. A(km.).
P.

West M.S. 63 Brommich Oxford 1 1 s 114 1,. Bidston 1II.S. 135 Stonyhurst M.S. 181

. . . .

p. "

Po.

s.

25 35 39 47

. . . .

S".

. s .,
30
46

O-C.

, P.

.....

PO.

Pg. o
+I
+I
0

s.

. . . .

so.

s,.

.
0 0

. . . . . . . .

. . . .

..

42

.... ....

52 65
68-70? 133 165 166 167

. . . . . . . . .. - 2
. . . .

. . . . . . . .

. . . .

- 1

Kew

G.

187

. . . .

Edinburgh M.S. Uccle Paris De Bilt

414 504 508 522

. . . . . . . . . .
87 94(?) 92

. . . . . . . . . .
+3

- 4
I

w.
W.

. . . .

W. '

..

I05 135 144 106 137 .. 112(?)144 .. 165t


211

-3 . . . .

-2
-I

I I

- 7

+8

..

+z

+ 4 ..
0

+ .. +

- a

Strasbourg 19-ton 8 2 7 133 Hamburg Zurich Wien W.


848

{ ;; }

240 (?) 264

+8(- +I1 } - I I

+ 3

+ 3

. . . . . . . . . .
149

289(??) .
+8

. . . . . . . . .
-27

+zr

Q.P. 945

W. 1380

.... . . . . . .
PO.

237 (:~~)302-308t 311

..
S.

436

. . . . . . . . . .
.S '

.. + I

The calculated times are


P. PO.
SO.

19-tAi7.8

15+Aj6.3

13+A/5'4

20+A/4.35

14+A/3.7

II+A/3.3

Minute-gap.

The position of the epicentre is 52' 14' N., '2 44' W. On inspection of the residuals we see that those for P,, S, and 8 , are as satisfactory as for the Jersey earthquake. As the delay of Po in starting is about 2 s. instead of 3 s., the depth of focus was probably slightly less, somewhere about 8 km. This is supported to some extent by the fact that the delay of S in comparison with S, is now 9 s. instead

490

Dr. Harold Je$reys,

of 8 s. But one must not insist unduly on the significance of differences of I s. even in the means of reasonably accordant observations. The three more difficult waves to observe, P, P*, and S*,give much larger residuals. In the case of P the reason is that this wave in near earthquakes is always an emrsio, and if small will inevitably be read too late. It may be remarked that the actual displacements on the records in this phase never exceeded 0.05 mm. For P*, the readings do little more than to confirm its existence ; movement waq definitely going on somewhere near the time of its arrival, and the amount was too large to be attributed to the train of waves headed by P. Readings were made specially difficult a t Edinburgh and De Bilt by microseisms. In the case of S*, there was always movement going on about the time when it would be expected, but it began too gradually to be satisfactorily timed. The amplitudes of the first displacements in the various phases were as follows. The unit, is again I p.
TABLE IV.
P. P'. Po.

e.

West Bromwich Bidston (N.S.) Stonyhurst (E.W.) Edinburgh Uccle Strasbourg Zurich (N.S.) Wien

.. .. ..

'.
..
0'2

7
2

..
..

S*.

..

.. .. ..

eg.
20

..
0'1
0.05

..

..
0.6
I

s
1'4

..
3

..
12 '

..
0 1 .

5
1'3
01 .
0.05

0.3

0.7
0.06
01 '

0'1

..

.. ..

..
..

0.06

(M-0.6)

..

These amplitudes do not vary very regularly with epicentral distance.

It wac hardly to be expected that they would. Those of P, and S,


are more nearly inversely proportional to the epicentral distance than to any simple power of i t ; the same applies to those given for the Jersey earthquake and to those tabulated by Gutenberg for the South German earthquakes. Now a pulse spreading out in three dimensions in a uniform medium according to the laws of elasticity has an amplitude proportional to the reciprocal of the distance from the origin. This is very good as far as it goes. For waves of type SH it is almost the whole story ; for when these waves are reflected at a free surface the reflected wave makes the same contribution to the motion of the surface as the incident one, and the law that the amplitude of the motion of the surface is proportional to [/A will hold. The waves that follow S, are mostly of type SH. P and SV waves, however, produce no motion of the ground a t grazing incidence, and a t small angles of emergence the motion, for a given amplitude of the emerging wave, is proportional to the angle of emergence. At large epicentral distances this angle varies as A when the curvature of the earth is taken into account, and as I/A if it is not.

June 1927.

On Two British Earthquakes.

491

In either case its variation spoils the I/A law for the motion of the surface. Lamb's discussion? of the waves generated by an internal source within an elastic solid with a plane boundary gave the law 1/A2 for both P and SV, as we should expect from the above considerations. It seems to me that the amplitude in the bodily waves in the granitic layer must really be nearly proportional to I/A, but that these waves are brought up to the surface a t moderate angles of emergence by ref fraction a t the (possibly irregular) base of the sedimentary layer. I so, the angle of emergence at the outer surface will on the whole be nearly independent of A, and the amplitudes will be proportional to I/A, as the observations seem to require. The above discussion concerns only P, and S,. It re-opens the question why the amplitude of P is a moderate fraction of that of P,, instead of being very small, as the ordinary laws of refraction indicated. Diffraction appeared to explain the facts $ when the crust was considered as of two uniform layers, but left both amplitudes proportional to 1/A2. I we assume that the sedimentary layer cancels f a factor I/A for P,, something must do the same for P ; and it cannot be the sedimentary layer this time, because the angle of emergence of P at the top of the granitic layer is moderate to begin with. We may, perhaps, look to the increase of velocity with depth in the lower layer to confine the wave to a limited depth. 4. The Energy.-An estimate of the total energy in the waves dent out in an earthquake would be very difficult, but we can simplify the problem by noticing that a moderate fraction of the original energy would probably be in waves of type SH, which can b y no means be converted into P or SV waves The records a t Uccle, De Bilt, Paris, and Strasbourg, which gave both horizontal components, made it clear that in the large waves following S, most of the observed horizontal movement was a t right angles to the line joining the station to the epicentre. The energy in the SH waves of this phase alone would therefore be a moderate fraction of that of the earthquake as a whole. From the observed velocity of the ground we can find the kinetic energy per unit volume ; on an average the potential energy would be the same. The length of the wave train is the product of the mean velocity of the waves composing it and the time it takes to pass, and the energy per unit volume may be taken t o be the same throughout the granitic layer. Finally multiplying by ZTA we get the volume of the region occupied by the wave train when its middle was passing the station considered. The total energy can then be found. The following data refer to the Jersey earthquake. The periods and amplitudes are mean values for the largest waves, which come after So.
A.

Amplitude ( p ) .

Period (sees.).

Duration (secs.).
20

Paris Uccle Zurich

291 469 I16

90
60

1.6
I

30
20

1'5

Phil. Trans. A . , 203, 1904, 1-42. $ Proc. Carnb. Phil. Soc., 23, 1926, 472-81.

492

Dr. Harold Jeffreys,

At Paris the speed of the vibration was +/set., and the maximum velocity 360 p/sec. Taking the density as 3 gm./cm., we have the mean energy (kinetic and potential) = 4 x I O - ~ e r g / ~ m . ~The depth was about 1o6 em., the length of the train 60 km., and znA = 1800 km. Hence the energy was about 4 x 1 0 ergs. Uccle ~ ~ similarly gives g x 10l8 ergs and Zurich 1 d 7 ergs. The true values were, of course, a few times greater. For the Herefordshire earthquake we have
A.

Amplitude.
I2 I0

Period.
2

Duration.
15 I5 I4

Energy.
1016 2

West Bromwich Stonyhurst

63
I81

1.6
1.8
2

1016

Uccle Zurich

504

4
0.4

g x 1016
2

915

I5

x10~4

I n each case there is not much sign of loss of energy up to a distance of 500 km. The great loss is in the next stage. This tends to show that loss of energy in transmission is not great in most regions, but that in some it is very considerable ; this supports the view expressed elsewhere that the loss of energy in surface waves is mainly due to reflexion. We may remark that granite has a Young's modulus of about 7 x 1d1 dynes/crn.z, and a crushing strength of about 109 dynes/cm2. The extension when on the verge of crushing is therefore about 1-4x I O - ~ , and the strain energy 7 x 105 e r g s / ~ m . ~ The whole energy of the Jersey earthquake could therefore have been supplied by, say, 1013 C.C. of granite on the verge of giving way, or a cube of zoo metres edge. The size of the focus of an earthquake may then be rather small. Superposed on the large waves all stations show a smaller movement of under half the amplitude, but continuing for some minutes longer, while gradually falling off. This must carry energy comparable with that of the main waves. The amplitude of the largest waves usually increases gradually to a maximum, and then they suddenly end, leaving the other movement just mentioned. This behaviour is characteristic of dispersion of waves when there is a minimum group velocity, which on this interpretation would be about 10 per cent. less than that of S ., But the fact that the time of passage of the large waves is not proportional to A argues against this view. 5 . Summary.-The existence of the main waves found by continental investigators of near earthquakes has been confirmed, and a wave S* that may be the distortional analogue of P* has been found. The velocities found are as follows : P, 7.8 ; P*, 6.3 ; P,, 5.4 ; S, 435 ; S*, 3.7 ; S,, 3.3, all in km./sec. All are probably trustworthy to within I per cent. or a little more, except S*, which is rather less certain. There is no sign of a difference between Great Britain, the North Sea, the Channel, and the Continent in this respect. P, shows a delay of two or three seconds in starting,in comparison with S,, which can be explained by saying that the original movement was wholly distortional and that compressional movement arises only by reflexion a t the top of the granitic layer or a t the outer surface. The depth of the granite

June 1927.

On Two British Earthquakes.

493

layer is probably about 10 km., and that of the intermediate one 20 km., while the foci mere probably a t depths of about 10 and 8 km. for the Jersey and Hereford shocks respectively. Some of these interpretations are obviously dependent on hypotheses concerning the effect of the sedimentary layer, but without more definite knowledge of this it does not seem worth while to attempt greater definiteness. The amplitudes of the first movements of P, and 5 are roughly pro, portional to the inverse distance. The energy of the Jersey earthquake was of the order of 1 d 9 ergs, and that of the Hereford one 5 x ~ol~ergs. 6. Instruments.-As this is the first occasion when a direct comparison of the records of several Milne-Shaw machines with those of machines of the older types has been carried out for a single earthquake, it may be worth while to put on record some impressions concerning the relative success of the instruments. They naturally refer mainly to the registration of near earthquakes. The 19-ton pendulum a t Strasbourg and the 21-ton one a t Zurich, with magnifications between 1000and 2000 and time scales of about I mm. per second, are of course in a class by themselves. Next comes the Wiechert with a mass of about a ton, a time scale of about I mm. to 4 s., and magnifications ranging from about IOO t o 2 5 0 . The Milne-Shaw machine, with a magnification of 150 or zjo, is slightly less successful. It records photographically instead of on smoked paper. This makes it possible to use a much smaller mass to overcome friction, but the record is not quite so clear. A well-adjusted Wiechert machine gives a white trace not more than 0.05 mm. wide, with perfectly sharp edges, and dis'placements of this amount on the record can be read easily and accurately under a magnifier. The broader trace and the slight loss of definition unavoidable with optical recording make the reading of these very small pulses more difficult. The time scale, again, is I mm. to about 8 s., so that the time itself cannot be read quite so well. On the other hand the Milne-Shaw machine can be arranged so as to give a faint trace throughout the minute gap, which often prevents the exact beginning of a pulse from being missed, as it would be by a machine that gives a complete gap. For this reason there might be a definite advantage, where machines recording on smoked paper give both components, in arranging the minute gaps on the N.-5. and E.-W. components to come a t different times. The Galitzin machine is, I think, very definitely less satisfactory. My main reason for this view is that its record bears no obvious resemblance to the motion of the ground. The magnification of the Wiechert and Milne-Shaw machines tends to a finite limit, the nominal magnification, when the period of the disturbance of the ground becomes indefinitely short, and their traces represent fairly closely the actual movement of the ground when this has a period not exceeding a few seconds. But the magnification of the Galitzin instrument is zero for both very long and very short periods, and rises to an enormous maximum for an intermediate one, which happens to be about that of the large waves from a distant earthquake. Thus these waves receive a great magnification, which they do not need, and which usually serves

494

On

Two British Earthquakes.

I. 9 .

only to make the record underexposed and to make the traces overlap, while the P phase, which is much smaller, is magnified much less, In near earthquakes, where the motion usually involves very rapid oscillation, the Galitzin machine gives only a small magnification, and its records lack the distinctness of those of the simpler instruments. If, for instance, the ground starts to move with a finite velocity, the indicator of a Wiechert or Jlilne-Shaw machine will do the same, but that of a Galitzin will begin with a finite acceleration, and the trace is rounded instead of having a sharp angle. For the present inquiry De Bilt and Strasbourg sent Galitzin records, but these were quickly laid aside in favour of those from the Wiechert zoo kg. and the Ig-ton machines at the respective stations ; the ordinary Wiechert record from Strasbourg was also much superior to the Galitzin. The Kew machme was a Galitzin, and gave useful results, but in the case of the Jersey earthquake it was distinctly less informative than Paris, while Stonyhurst excelled it in the other case, the epicentral distances of the places compared being about equal. In the Jersey earthquake the , Kew record of the waves following S consisted of a single large displacement and recovery, lasting 40 s. ; the Paris instrument was a t the same time performing regular oscillations with a period of 1.6 s., which are shown on the Kew record only by a dotted appearance of the trace. Accordingly I consider that, so far as the material available in the present inquiry has been able to show, the Milne-Shaw machine is not quite so good as the Wiechert, but both are definitely preferable t o the Galitzin.
Notes

on the Figures (Plates

11, 12,

13).

the Paris record, P*, which is a very small and rapid oscillation, too small to reproduce clearly, has been emphasized by thickening the trace by hand on a reproduction. On the Strasbourg one, alao, it has been found desirable t o ink in the trace from 2 2 m. to 24 m. FIG.3.-The amplitudes of P, P*, and P g are greater on the E.W. component than the N.S. one, but this relation is reversed for the S movements. This indicates that the former waves are longitudinal, and the latter mainly of type SH. P is most easily read on the vertical component, probably owing to its rather steep emergence. FIG. 4.-Uccle, Strasbourg, and Zurich all show an unidentified displacement between Pg and S. This may be a compressional wave in the sedimentary layer. FIG.5.-The indications of phases on the Stonyhurst record, shown above the trace, are mine. Those given below the trace were made a t the Observatory.

FIG.2.-On

Potrebbero piacerti anche