Sei sulla pagina 1di 279

ELEMENTS

Infinitesimal Calculus.

by

JOSEPH BAYMA, S.J.,


Generated on 2015-06-01 01:01 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

professor of

Mathematics

in Santa Clara

San Francisco

A.

S.J., Santa Clara, California.

College,

WALDTEUFEL,
737

MARKET STREET.
1889.

CONTENTS.

Introduction,
PART.

^-DIFFERENTIAL

Section I. Rides of
Differentiation.
Algebraic functions of one variable,

CALCULUS.

Transcendental functions of one variable,


Functions of two or more variables,
Implicit functions,
Section

Generated on 2015-06-01 01:02 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

27
36
40

II.

Successive Differentials.
Maclaurin's formula,
Taylor's formula,

46

40

De Moivre's formulas,

Maxima and minima,


Exercises on maxima and minima,
Values of functions which
assume

Section

SI

55

66
63

an indeterminate form,

79

III.

Investigations about Plane


Curves
Tangents, normals, etc.,
Direction of curvature.

Singular points, .
Order of contact, osculation,
Measure of curvature,
Evolutes,
Envelopes,

....
.....

Elements of arcs, surfaces, and


volumes,
5

85
90
98
101
104
108
113
117

CONTEXTS.

Differentials with polar co-ordinates,


Spirals,

123

PART II. INTEGRAL CALCULUS.


Section

I. Various Mel hods for Finding Integrals.

Integration of elementary forms,

Reduction of differentials to an elementary

form,

Integration by parts,
Integration of rational fractions,
Integration of binomial differentials,
Integration by successive reduction,
Integration of some trinomial differentials,
Integration by series,
Integration of trigonometric expressions,

Integration of logarithmic differentials,


Integration of exponential differentials,
Integration of total differentials of the first order
Integration of the equation Mdx + Ndy =

0,

Generated on 2015-06-01 01:03 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Integration of other differential equations,


Integration by elimination of differentials,
Double integrals,

Section

II.

Application of Integral Calculus

to Geometry.

1"

Rectification of curves
Quadrature of curves,
Surfaces of revolution

204

Solids of revolution,

210

Other geometrical problems,

213

208

Problems solved by double or triple integrals,

Section

III.

Application of Integral Calculus

to

.223

Mechanics.

Work

231

Movement uniformly varied,

237

Movement not uniformly varied

238

CONTENTS.

Composition

PAGE

and decomposition of forces

241

Moments,

245

Virtual moments,

249

Attraction of a sphere on a material point,

Generated on 2015-06-01 01:03 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

251

Centre of gravity,

256

Moment of inertia

263

Curvilinear movement,

272

INFINITESIMAL

CALCULUS.

Generated on 2015-06-01 01:04 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

INTRODUCTION.
1. The Infinitesimal Calculus is exclusively con
cerned with continuous quantities ; for these alone
admit of infinitesimal variations.
A variable
quantity is said to be continuous, when it is of
such a nature that it cannot pass from one value to
another without passing through all the intermedi
ate values.
All the parts of a continuous quantity
are continuous : and, as all continuum is divisible,
every part of a continuous quantity, how small so
ever it be, is still further divisible.
In other terms,
the division of continuum can have no end.
2. Infinitesimal quantities are sometimes con
ceived as resulting from an endless division of the
finite. But this is not the real genesis of infini
tesimals ; for, in the order of nature, it is the infini
tesimal itself that gives origin to the finite. Thus,
an infinitesimal instant of duration does not arise
from any division of time; for it is the instant
itself that by its flowing generates time. In like
manner, the infinitesimal length described by a
moving point in one instant of time does not origin
ate in any division of length ; for it is the actual
infinitesimal motion of the point itself that by its
continuation generates a finite length in space.

Generated on 2015-06-01 01:04 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

10

IXFIXITESIMAL CALCULUS.

Hence infinitesimals of time and of length are not


mathematical fictions.
They are true objective
realities. Had they not a real existence in nature,
neither the origin nor the variations of continuous
movement would be conceivable.
For the same reason we must admit that continu
ous action cannot produce acceleration except by
communicating at every instant of time an infini
tesimal degree of velocity : and speaking general
ly, all continuous quantities develop by infinitesi
mal moments. Hence the branch of Mathematics
which investigates the relations between the con
tinuous developments of variable quantities, has re
ceived the name of Infinitesimal Calculus, and its
method of investigation has been called the infini
tesimal method.
This method has been used by the best mathe
maticians up to recent times. Poisson, in the intro
duction to his classical Traite de Mecanique (n. 12),
says : "In this work I shall exclusively use the in
finitesimal method. . . . We are necessarily led to
the conception of infinitesimals when we consider
the successive variations of a magnitude subject to
Thus time increases by de
the law of continuity.
grees less than any interval that can be assigned,
however small it may be. The spaces measured by
the various points of a moving body increase also
by infinitesimals ; for no point can pass from one
position to another without traversing all the iiP
termediate positions, and no distance, how small
soever, can be assigned between two consecutive
positions. Infinitesimals have, then, a real exist
ence: they are not a mere conception of Mathe
maticians."

INFINITESIMAL CALCULUS.

11

it

it

is

it

Generated on 2015-06-01 01:04 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

is,

3. Modern authors often define the infinitesimal


as the limit of a decreasing quantity.
This defini
tion we cannot approve.
For the divisibility of
continuum has no limit, and therefore cannot lead
to a limit.
This is so true, that even those authors
confess that the limit the absolute zero can never
be reached.
On the other hand, infinitesimals, in
the order of nature, do not arise from finite quanti
ties: it is, on the contrary, these quantities that
arise from them.
The origin of infinitesimals is
dynamical ; for they essentially either consist in, or
depend on, motion : and as motion has no other
being than its actual becoming or developing, so
also infinitesimals have but the fleeting existence of
the instant in which they become actual. It is for
this reason that Sir Isaac Newton conceived them
as fluxions and nascent quantities ; that is, quanti
ties not yet developed, but in the very act of de
we believe, the true notion of
veloping. This
the infinitesimal, the only one calculated to satisfy
a philosophical mind.*
remains true
So long as
that line cannot be drawn except by the motion
of a point, so long will
remain true that an infini
tesimal line
the fluxion of a point through two
consecutive positions.
An infinitesimal change may be defined,
change
which is brought about in an instant of time.
Now, the true instant is the link of two consecutive
terms of duration
is obvious that between
and
two consecutive terms of duration there is no room
for any assignable length. Hence the fleeting in
stant has
duration less than any assignable
On the modern doctrine and method of limits see the note appended to No. 23.

Generated on 2015-06-01 01:04 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

12

INFINITESIMAL CALCULUS.

length of duration, that is, it has a duration strictly


infinitesimal. And in the same manner, every other
infinitesimal change is a link of two consecutive
terms, or of two consecutive states ; for it takes
place iu an infinitesimal'instant.
4. But here the question arises : How can an in
finitesimal quantity be intercepted between two
consecutive points? -Consecutive points touch one
It
another and leave no room between them.
would seem, then, that what we call "an infinitesi
mal" is not a quantity, but a mere nothing. We
answer that a point in motion has always two con
secutive modes of being in space ; for it is always
leaving its last position, and always reaching a fol
lowing position which cannot but be consecutive to
Now, it is plain, that if the
the last abandoned..
actual passage from the one to the other were not a
real change, the whole movement would be without
change ; for the whole movement is but a continu
ous passage through consecutive
But
points.
It is
movement without change is a contradiction.
therefore necessary to concede that between two
consecutive points there is room enough for an in
finitesimal change, and accordingly for an infinitesi
mal quantity.
As a further explanation of this truth, let us con
ceive two material points moving uniformly, the
one with a velocity 1, the other with a velocity 2.
Their movement being essentially continuous, there
is no single instant in the whole of its duration, in
which they do not pass from one point to a con
secutive point, the one with its velocity 1, the other
with its velocity 2. But the velocity 2 causes a
change twice as great as that due to the velocity 1.

INFINITESIMAL CALCULUS.

13

Generated on 2015-06-01 01:04 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Therefore the ratio of the two movements is, at


every instant, as 2:1. But two absolute nothings
cannot be in the ratio 2 : 1. Therefore the move
ments comprised between two consecutive points
are not mere nothings, but are real quantities,
though infinitely small. They are, in fact, fluxions,
or nascent quantities, or, as the Schoolmen would
say, quantities in fieri.
Nor does it matter that these infinitesimals are
sometimes represented by the symbol 0. For this
When
symbol has two meanings in mathematics.
it expresses the result of subtraction, as when we
have a a = 0, it certainly means an absolute
nothing: but when it expresses the result of di
vision, it is a real quotient, and it always means a
quantity less than any assignable quantity : but
because it has no value in comparison with finite
quantities, it is treated as a relative nothing, and
is represented by 0. Thus, in the equation

the zero represents an infinitesimal quantity.


This
For it is only continuous
can be easily proved.

quantities that admit of being divided in infinitum :


and, when so divided, they give rise to none but
continuous quotients, because every part of con
Now, the ab
tinuum is necessarily continuous.
solute zero cannot be considered continuous.
Therefore the absolute zero caji never be the quo
And in this sense, it
tient of an endless division.
is true, as the theory of limits affirms, that a de
creasing quantity may indefinitely tend to the
limit zero, but can never reach it. On the other

14

INFINITESIMAL CALCULUS.

hand, the above equation gives


a=0 X

oo ;

and this does certainly not mean that the finite


quantity a is equal to an infinity of absolute noth
ings.

5. We have said that infinitesimals have no value


A few years
as compared with finite quantities.
ago. an American writer* was bold enough to
maintain that this fundamental principle of infini
The principle,
analysis is not correct.
however, has been admitted by the greatest mathe
maticians, and its correctness will not be doubted
by any one who understands the real nature of in
finitesimals. The principle, says Poisson (loc. cit.),
"consists in this, that two finite quantities which
do not differ from each other except by an infini
tesimal quantity, must be considered as equal ; for
between them no inequality, how small soever, can
be assigned'''' ; because the infinitesimal is less than
any assignable quantity.
Again, it is plain that the infinitesimal is to the
finite as the finite is to the infinite. Now, the infi
nite is not modified, as to its value, by the addition of
Therefore the finite is not modi
a finite quantity.
fied by the addition of an infinitesimal. That the
infinite is not modified by the addition of a finite
quantity, can be assumed as an evident truth : but
Thus, it is shown in
it can also be demonstrated.
Trigonometry that between the angles A, B, C of a

Generated on 2015-06-01 01:05 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

tesimal

* Mr. Albert Taylor Bledsoe in his Philosophy


of Xathematic/t, where he strives
We arc afraid that
to prore that the infinitesimal method should be abandoned.
philosophical readers will not consider his effort a success.

INFINITESIMAL CALCULUS.
plane triangle there is the relation
tan A + tan B + tan O = tan A tan

15

B tan C /

A = 90, 5 = 45, C =

and this equation, taking


45, gives
oo

-f-2 =

co ;

which shows that the addition of a finite quantity


does not modify the value of the infinite.
We may draw from Arithmetic a still plainer
Dividing 1 by 3 we obtain
proof of our principle.
5

= 0.333333

and multiplying this by


1

we obtain

3,

= 0.999999

Generated on 2015-06-01 01:05 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

In this last equation, if

the second member be


understood to continue without end, the difference
between the two members will be an infinitesimal
fraction viz., unity divided by a divisor infinitely
Now, we can prove, that, notwithstanding
great.
this infinitesimal difference, the equation is rigor
For, let the second member of the
ously true.
equation be represented by x; then

x = 0.999999

Multiply both members of this by


10a;

= 9.999999

10
9

then

+ x;

and from thi?, by reduction, we have


925

9,

x=l.

This clearly shows that the equation 1=0.999999


is rigorously

It

. . .

is plain, therefore, that an


infinitesimal difference has no bearing on the
value of a finite quantity, and that no error is com
true.

16

INFINITESIMAL CALCULUS.

is

is,

mitted by suppressing an infinitesimal by the side


of a finite quantity.
6. The notions above developed may suffice as a
first introduction to the infinitesimal calculus. We
have shown
1st. That infinitesimals are not nothings, but ob
jective realities :
2d. That infinitesimals are not limits of decreas
quantities in
ing quantities, but fluxions that
the act of developing, or more briefly, nascent
less than any assignable
quantities, whose value
value of the same nature
3d. That infinitesimals may have different rela
tive values, and form different ratios
4th. That an infinitesimal, whether added to, or
taken from, a finite quantity, cannot modify its
value.

which we shall have to speak throughout our treat


ise, we have here simply to state the fact, that in
finitely great, and infinitely small quantities are
capable of degrees, so that there may be infinites
and infinitesimals of different orders, each infinite
of a higher order being infinitely greater than the
lower order, and each infinitesimal of
infinite of
a higher order being infinitely less than the infini
How this can be, one
tesimal of a lower order.
may not find easy to explain, because both the
infinite and the infinitesimal lie beyond the reach of
human comprehension nevertheless we know, not
only from Algebra and Geometry, but also from
rational philosophy, that such orders of infinites
We know
and of infinitesimals cannot be denied.
that the species ranges infinitely above the indi
:

Generated on 2015-06-01 01:05 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

As to the different orders of infinitesimals, of

INFINITESIMAL CALCULUS.

17

vidual, and the genus infinitely above the species.


Substance extends infinitely less than Being, animal
infinitely less than substance, man infinitely less
than animal. From this it will be seen that the
notion of an infinite infinitely greater than another
infinite, is not a dream of our imagination, but a
well-founded philosophical conception, familiar to
every student of Logic, and admitted, implicitly at
least, by every rational being.
Let us, then, write the following series :
.

x , x , x,

1,

- ,

Generated on 2015-06-01 01:05 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

If

we assume x = m, it is plain that the first term


will be infinitely greater than the second, the second
infinitely greater than the third, and so on. The
middle term 1 being finite, all the following terms
are infinitesimal, and each is infinitely less than
the one that precedes it. Hence infinites and
infinitesimals are distributed into orders.
Thus, if
x be an infinite of the first order, x> will be of the
second order, Xs of the third, etc. ; and in like

manner

-x will

be an infinitesimal of the first order,'

-5 of the second order, CO of the third, etc.

OCT

7. The problems whose solution depends on the


infinitesimal calculus, are generally such that their
conditions cannot be fully expressed in terms of
finite quantities.
Hence a method had to be
found by which to' express such conditions in
infinitesimal terms. The part of the Calculus
which gives rules for property determining such
infinitesimals and their relations, is called the
Differential Calculus. As, however, none of such

18

INFINITESIMAL CALCULUS.

Generated on 2015-06-01 01:06 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

infinitesimals must remain in the final solutions,


rules were also to be found for passing from the
infinitesimal terms to the finite quantities, of which
they are the elements ; and to effect this, a second
part of Calculus was invented under the name of
Integral Calculus.
Of these two parts of the infinitesimal calculus
we propose to give a substantial outline in the pres
ent treatise : and we shall add a sufficient number
of exercises concerning the application of the Cal
culus to the solution of geometric and mechanical
questions ; for it is by working on particular ex
amples that the student will be enabled to appreci
ate and utilize the manifold resources of this branch
of Mathematics.

PART I.

Generated on 2015-06-01 01:06 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

DIFFERENTIAL CALCULUS.
8. Our object in this part of our treatise is to
find, and to interpret, the relations which may exist
between the infinitesimal changes of correlated
quantities varying according to any given law of
continuous development.
Such a law is mathe
matically expressed by an equation between the
variable quantities ; and it is, therefore, from some
such equation that the relative values of the infini
tesimal changes must be derived.
An infinitesimal change is usually called a differ
ential, because it is the difference between two con
secutive values, or states, of a variable quantity.
The process by which differentials are derived from
given equations is called differentiation, and the
equations themselves, by the same process, are said
to be differentiated.
Hence this part of infinitesi
mal analysis received the name of Differential
Calculus.
Differentials are expressed by prefixing the letter
d before the quantities to be differentiated. Thus,
dx = differential of x, d(ax') differential of ax2.

9. When an equation contains only two vari


ables, arbitrary values can be assigned to one of
them, and the equation will give the corresponding
19

INFINITESIMAL CALCULUS.

20

values of the other.


The one to which arbitraryvalues are assigned is called the independent vari
able, and the other, whose value depends on the
yalue assigned to the first, is said to be a function
of the same. Thus in the equation of the parabola,
y' = 2px, if we take x as independent, y will be a
function of x.
When an equation contains more than two vari
ables, then all the variables but one can receive
arbitrary values, and are, therefore, independent,
whilst the remaining one will be a function of all
the others.

Functions are often designated as follows


V

F(x, y\

<p

(x, y, z)

Generated on 2015-06-01 01:11 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

the first and second being explicit functions, the


third implicit. Different characteristics, as
F, <p,
denote equations having different forms. In these
equations, besides the variables, there are usually
some constant quantities, though, for the sake of
brevity, the variables alone are expressly pointed
put.

This part of our treatise will contain three

sec

tions. In the first we shall explain the rules of


In the
differentiation for all known functions.
second we shall consider the successive differentials,
and their extensive use in mathematical investiga
In the third we shall show the bearing of
tions.
differential expressions on the solution of problems
regarding curves.

INFINITESIMAL CALCULUS.

21

SECTION I.

RULES OF DIFFERENTIATION.
10. The function of which we have

to

find the

differential, may be either algebraic or transcen


dental.
It is algebraic, if it is formed of expres
sions obtained by the ordinary operations of
algebra, as addition, subtraction, multiplication,
division, and the formation of powers with con
stant exponents, entire or fractional.
It is tran
scendental when it contains logarithms, circular
functions, or exponentials.

Generated on 2015-06-01 01:11 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Algebraic Functions.
11. Polynomials. An algebraic sum of func
tions constitutes a polynomial.
Let

y=s

at -f- bu v + c

(1)

polynomial, in which s, t, u, v are functions


of x, and a, b, c constant quantities. When x ac
quires an infinitesimal increment dx, the functions
s, t, u, v acquire the corresponding infinitesimal
increments ds, dt, du, dv, and y acquires its incre
ment dy. Hence the equation (1) becomes
be a

y-\-dy =
s + ds
a(t + dt) + b(u + du)-(v + dv) + c.
Subtracting (1) from (2) we shall have

(2)

INFINITESIMAL CALCULUS.
dy = ds

aclt

-+-

22

&<Zw

(3)

c,

b)

is

is

which
the differential of the given polynomial.
Comparing (3) with (1), we see, first, that the dif
the algebraic sum of the
ferential of polynomial
differentials of its terms second, that constant fac
remain unchanged
tors (as a and
third, that
isolated constant terms, as
disappear in the pro
cess of differentiation.

12. Products. Let us now have

st,

(1)

ds)(t+dt).

y
+

dy =

(s

and being functions of x. When x acquires the


increment dx, these functions acquire the corre
will become
sponding increments ds and dt, and
+ dy. Hence
(2)

dy = tds + sdt -f dtds.


But the term dtds, as being an infinitesimal of the
second order, has no value by the side of the
other terms of the equation, which are of the first
order.
And therefore, suppressing that term, we
shall have simply

dy, or d(st) = tds -f sdt.

(3)

Hence, to differentiate the product of two functions


of x, we multiply each by the differential of the
other, and take the algebraic sum of the results.
If in (3) we assume = uv, we obtain by this rule
ds = udv

-f-

Generated on 2015-06-01 01:12 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Subtracting (1) from (2) and reducing, we obtain

vdu;

INFINITESIMAL CALCUL US.

23

and this substituted in (3) gives


d(uvt) uvdt

+ utdv + vtdu ;

which shows that the differential of the product of


three functions is obtained by multiplying the dif
ferential of each function by the product of the
other functions, and by taking the algebraic sum
of the results. And the same rule holds when the
functions are more than three.
13. Quotients.

Assume

y=\,

(i>

and t being functions of x. When x becomes


x + dx, then s, t, and y become s + ds, t-{- dt, and
y + dy. Accordingly
s

Subtracting (1) from (2) we obtain

dy.

s-\-ds

t + dt

ids sdt
i~ P + tdt
s

'

But the term tdt has no value by the side of V.

jfs\

d(j)

= tdssdt

dy, or

and we obtain
^

Hence we suppress

it,

/o\
(3)

Therefore, the differential of


quotient of two
functions of the same variable
equal to the de
nominator into the differential of the numerator,
minus the numerator into the differential of the
denominator, divided by the square of the denom
is

Generated on 2015-06-01 01:12 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

inator.

INFINITESIMAL CALCULUS.

24

14. Powers and Roots. Let us have


(1)

s being a function of x, and m a constant.


x becomes x + dx, this equation becomes

y + dy =

(s

When

+ ds)m,

or, developing by the binomial formula,

y + dy =

sm

+ ffls"-1

<fe

+ !!^pV-

dsa

. .

(2)

Subtracting (1) from (2), we obtain

dy = msn-lds + '!*T11sm-*ds* +
which, by suppressing all the terms that transcend
the first order, reduces to

Generated on 2015-06-01 01:12 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dy, or d (sm)

= ms91-1

ds.

(3)

Hence, to obtain the differential of any power of a


function, diminish the exponent by 1, and then
multiply the result by the original exponent and

by the differential of the function.


As the binomial formula is true for all expo
nents, whether positive or negative, integral or frac
tional, equation (3) holds good for all possible
powers and roots.

Thus, making m

y
the differential

will

then

and

tys = s,

be
1

If n = 2,

= ^,

ds

ds

INFINITESIMAL CALCULUS.

= dVs =

25

ds

2Vs

that is, the differential of the square root of a


quantity is equal to the differential of the quantity
divided by twice the radical.
The preceding rules are sufficient for the differen
tiation of any algebraic function of one variable.

Examples. It is of the utmost importance that

ax'

(a*+xj-b,

dy = (d'+x,)xdx,

ax?+^,

dy=(2ax

a-\~x

a^-bx

1)

a(ax)4-

dx,

ax*

(a-xy

a(x'-2ax +
(x

af

xdx

dy=--^a,

* =

x) dx

va>-(b-xy

'

Va-(b~

dy = 6ax

,
dx,

b)

(^7'

8.y=l/a*-x*,

y=

2adx

ax

-^dx,

(x

y
^

5.

dx,

dy=&=xj,

= 2ax' Sax\

b)

ac, dy = (3ax'

(b

bx

-f-

A-y=a-=x'

6-

Generated on 2015-06-01 01:12 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

3.

2.

y
y

1.

the student should at once familiarize himself with


the above rules of differentiation, and test, by ex
Let him
amples, his practical knowledge of them.
work out the following :

INFINITESIMAL CALCULUS.

26

._

/a'x'

(a*-\-2x')dx

7,

2a'xdx

=-

13.

y=(a'-x*) Va'+af, dy

14.

y=

15.

y=

16.

y =x(l+x') VT=x~\ dy = 1+^|g' dx,

(2aa?

18.

^x+

Va'+x\ dy =

y=-p^=

x) (2ax

- x')* dx,

Va'+a?
l/x+
=-dx,

Sx'dx

^ = (3-^,

a?

19-

dy-6(a-

a;')')

/
Generated on 2015-06-01 01:12 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

<tfo

y-a.+ vTH^' ay~2x(1-af)+ VT^?'

/l Vx

^^Tl-^'
.

flte

^ = 2(1 + ^) ^a?"
7

The differentiation of complex functions is


often simplified by the use of auxiliary functions.
Thus our twelfth example, by assuming
Note.

Va*

x*

= s, and

t'a

+ a!a=

t,

INFINITESIMAL CALCULUS.

27

becomes
s

y=p

whence

ids

ay

sdt

On the other hand, the differentials of the auxiliary


functions are
,

xdx

xdx

and these values, and those of s and t, substituted


in the expression for dy, give the differential in
terms of x.
Transcendental Functions.

15. Logarithmic Fuxctioits. Let


tion of x, and let

s be a

y = \ogs.

func
(1)

When x receives the increment dx, this equation

y + dy = log (s + ds).

(2)

dy = log

(s

+ ds)-logs = log *^? = iog

(l +

*p\

Subtracting (1) from (2), we have

But we have from Algebra

Hence, substituting, and suppressing all the infini


tesimals of the second and higher orders, we shall
have

=M

dy, or d. (log

s)

Generated on 2015-06-01 01:22 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

becomes

(3)

INFINITESIMAL CALCULUS.

28

The factor M is the modulus of the system of


logarithms. In the common system, whose base is
In the Napierian
10, we have M = 0.43429448
system, whose base is

...

= 2.718281828459

M=l.

The logarithms used in the Cal


we have
culus are always Napierian, or hyperbolic, if no
warning be given to the contrary ; and the differen
tial of the logarithm becomes simply

d(log*) = .
16. Exponential Functions. Let a be a con
stant quantity, and * any function of x. The ex
ponential
y = a'
(1)

will
Generated on 2015-06-01 01:15 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

be readily differentiated by the following pro


cess.
Take the logarithm of both members of (1),

differentiate, and reduce.


log y
whence dy

Thus,

log a,

= log a. ds,

= y log a.ds, or
d(a") = log a.a'ds.

If

a be changed into
have simply

e,

d(f) =

then, since loge


<*ds.

EXAMPLES.
1.

y=

log (a + bx),

= l,

we

ftp =

X-+-a

ftp =

Hp =

.H

ftp

-L

ft

-8

ft,

ft

ft

Sot '^-i
7) X

.a

+a+X)%01 H.X+XaZA fiv

x)So\ +
1

2x/[

\v

xpq
v txq

xpvz

xp

xp

xp
xvzfi + ^x
xp

-dp
Hp

x So\ lx

Hp =

'(xSo\(3o\

dp

xp

Xx~\(x
-f-

.ft

Soj v( \xq

1)

Generated on 2015-06-01 01:15 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

-g

-f

'z

93

(xSo\ 'xp

xp

txSo\x

'\\ fl = \\-x)x9

Hp = 'xpx^a

-gi d =
'xSo\a3

ftp =
tip =

-fi /i

= ixx

'xp^+xSo^
xpf
,'(,-+)

Hp = \)x%

+ So[ 'xp(x

INFINITESIMAL CALC UL US.

30

17. Trigonometric Functions. Differential of


sine. Let 5 be a function of x. The equation

y sin s,

(1)

when x receives the increment dx, becomes

y-\-dy = sin (s +

ds),

(2)

and subtracting (1) from (2),

dy = sin (s + ds) sin s


= sin s cos ds + cos s sin

But cos ds

1,

and sin ds = ds.

sin s.

Substituting, and

reducing,

dy = d (sin
Differential of cosine.

s)

= cos s ds.

When y = cos

s, we

shall

dy

cos
(s

+ ds)
= cos

cos s
s cos

which, because cos ds

ds

1,

sin s sin ds cos s ;

and sin ds ds, reduces

to
d?/

=d

sin s ds.
(cos s) =

Differential of tangent.

When y = tan

s,

shall have

cos s d (sin s) sin s d (cos s)

\cos s/

'

that

cos" s

is,

Generated on 2015-06-01 01:15 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

find

dy

we

INFINITESIMAL CALCULUS.

31

or, reducing,

When y = cot

Differential of cotangefd.

s, we

shall have

, /cos *\
sin * d (cos s) cos

=
=
dl
-.
dy
v-i s
)
v
sin
\sin si

d (sin s)

+ sin'

sin

cos'

ds,

dy =

is,

that

Differential of cosecant,

ds

li

cosec

sin

^-j,

(sec

= sec

dy =

we

we shall have
cos ds
sin
s

dy =

(vers

sin

then

s)

cos

ds

vers

If

s)

(cosec

Differential of versed-sine.
s,

dy =

Generated on 2015-06-01 01:17 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

shall have

^-5

sm

If

Differential of secant.

's)

(cot

dy
a

or, reducing,

INFINITESIMAL CALCULUS.

32

of

Differential
1

sin

If y =

versed- cosine.

covers 5

then

dy=d

cos s ds.*
(covers s)

mm,
xdx,

EXAMPLES.
m

,
dy =

2.

y = sin' x,

3.

y = sin nx

4.

y=\og sin(a x), dy

cos

dy = 3 sin' x cos xdx,


cos nx, dy

= n (cos' nx

cot
(a

_
~

cos

(cos a;

sin5

V-ls\nx), dy dxV

1.

a?

a;

'

sin'

aVcos'

In all these equations represents the length


For. since
able angle in circle whose radius =
linear
the latter as well as the former must be
all the circular functions to be hereafter subjected
a

ab.dx

dx,

1.

= ecos

a?,

a;aa;,

sin

x)

sin

= log (cos x

eC08

aV

'

=tana; cot x, dy
,y

-y

a-b tan

dy = sin x cos

a+^tan^a;

_ log
~

a;

cos*

12.

a;

cos a;'

10.

cos'

dx

(|-|),

6.
8.
9-

11.

= log tan

<&r,

*7
1

dx sec*
,
Vl-a;, dy- -

?/= tan

Generated on 2015-06-01 01:17 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

7.

sin' nx) dx,

sin a; a; cos x, dy x sin ayZ.r,


.

x,

a;)

5.

1.

v = sm

of the arc that measures any vari


dy and dg must be homogeneous,
element. This remark applies to
to infinitesimal analysis.

INFINITESIMAL CALCULUS.

33

Circular Functions.

18. Inverse

We

have

hitherto regarded the trigonometric lines as func


tions of arcs. If we now regard the arc as a func
tion of one of its trigonometric lines, we shall have
an inverse circular function. The inverse functions
are designated thus,
sin-1 y, cos~ly, tan-1^, etc.,

and are read respectively, arc whose sine is y, arc


whose cosine is y, arc whose tangent is y, etc. The
equations
s

= sin-1 y, s cos^y, s

= tan-1

y, etc.,

are nothing else than the equations

y = sin s, y

cos s,

y = tan s,

etc.,

Hence, it is from these


presented in a new form.
latter that we shall derive the differentials of the
Generated on 2015-06-01 01:17 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

former.

From y = sin s we have found dy

cos s ds

hence

ds =
and, since s =

dy
cos s

dy

Vl-sinV

sin-ly, therefore
dismay) =

Vl-y

From y = cos
hence

s we have

found dy
dy =

sin s ds ;

1NFINITESIMAL CALCULUS.

34

and, since s = cos-1 y, therefore

ds = di/xos1 s =

or

= cot

(tan-1 y) =

COS

tan' *

hence

found dy =

we have

From

<?

ds

found dy =

sin'

'

s we have

2/'

From y tan

d(coa-'y)=

hence

*=-l +

ds--=-dySm

-coVg>

whence

cos'

sec

4/sec"

dy
5

Vl

sin

y)
:

2/'

dy

cos's
s

dv

<2y

have found

we

cos!
ds = dy
sin

= sec

From

<Zs

1'

hence

cos ds
sin"

'

= cosecs we have found

tV-

From

d(sec

2/

whence

Generated on 2015-06-01 01:17 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

<Z(COt-'

INFINITESIMAL CALCULUS.

ds

sdy-sin

cos s

35

Vl

sina s

cosec s V coseca s

'

whence

d (cosec-' y) =
From y = vers

Vy'-i

found dy = sin s

s we have

<2s

hence
ft ?

dy
~

. -

Vl

sin s

dy
*

dy

VI (1

cosa s

vers

sf

V2y

= covers

we have found

<Zy

hence

dy
cos

y)

Vl

sina

ds=

ds

dy

which reduces to

a (covers

y)~

EXAMPLES.

V2y-y>

covers s)a

'

Vl

(1

Generated on 2015-06-01 01:17 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

cos

Finally, from

d(vers~x

2/a

which reduces to

ISFIXITESIXAL

36

c,

3?

CALCULUS.

3. s

=8\a-'(2x VT^af), ds.

4. s

= cos-1
=

5.

0.

7.

2dx

2r7

x7

adx

ds

adx

tan-'0)>
vers-'/^Y
,-

= Bin-

ds

as

2dx

Generated on 2015-06-01 01:17 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

9. s

= tan-1

,,->
m

10. s = sin

A ~ 00s
| l-f cos ^
a

a;

/a
j,
i/
^
T V-x'
*

<Zs

= <to,

ds =

xdxVb'a'

(b'-x')

Va%-x'

Functions of Two or More Variables.


10. In the preceding pages we have given the
rules by which any function of one variable can be
We must now extend the same
differentiated.
rules to the differentiation of functions of two or
more variables.

liW

H=/(T,),)
be a function of two independent variables.

(1)

When

INFINITESIMAL CALCULUS.

x and y

are made to acquire their respective


ments dx and dy, the equation (1) becomes

u + du =f (x + dx,y + dy).
Hence, subtracting (1) from (2),

du =/ (x + dx,y + dy)

-f

(x, y).

37

incre

(2)

(3)

The meaning of this last equation will be better un


derstood if we add and subtract the term/ (x + dx, y)
in its second member, which we then put in the
form

-f

du f (x + dx, y)
(x, y)
(x + dx, y).
+/ (x + dx, y + dy)

Generated on 2015-06-01 01:24 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

It

-f

(4)

is obvious that the difference


'(x -j- dx, y)
fix, y) represents the differential of the function
with regard to x alone ; for this difference arises ex
clusively from the increment dx given to x. Nor is
it less obvious that the difference./' (x -\-dx, y -\- dy)
f(x-\-dx,y) represents the differential of the
function with regard to y alone ; for this difference
arises exclusively from the increment dy given to
y, the other increment dx being common to both
terms, and showing that the differentiation with re
gard to x has already been performed.
It follows that the total differential of a function
of two variables must consist of two parts, which
are obtained by differentiating the function first
with regard to x, considering y as constant, then
with regard to y, considering x-\- dx (or merely x)
The total differential is represented
as constant.
thus,

INFINITESIMAL CALCULUS.

38

where

^ dx is

the

partial differential of

tion with regard to x,

and^dy

the func

its partial differ

ential with regard to y.

The total differential du


is the sum of the partial differentials.
Had we added and subtracted the term f(x, y -\-dy)
in the second member of (3), we might have put the
equation in the form

du =/(x, y + dy)-f (x, y)

Generated on 2015-06-01 01:25 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

+f(x + dx,y + dy)

-f

(x,

y + dy).

(6)

Here the difference f(x, y-\- dy)-/ (x, y) repre


sents the differential of the function with regard to
y alone, and the other difference f(x-\-dx, y -f- dy)

f(xi V + (ty) represents the differential with re


obtained after the differentiation with
gard to
regard to y has been performed. Now equations
The total differential is
(4) and (6) are identical.
therefore the same, whether we differentiate the
function hist with regard to x, then with regard
to y, or first with regard to y and then with regard
to x. In other terms, the result does not depend
on the order of differentiation.

20. We can show in the same manner that the


total differential of a function of three independent
variables is equal to the sum of the partial differen
tials obtained by differentiating the function with
Let
regard to each of the variables in succession.

u = <p(x,y,z)
given function.
obviously be

be the

Its total differential will

INFINITESIMAL CALCULUS.
du =

f (x + dx,

y + dy,

dz)

<p

39
(x, y, z).

By adding and subtracting the terms <p (x + dx, y, z)


and <p (x +
+
2) in ^ne second member, we
may put this equation under the form

du =

z)

(a?

(x

<p

(x + dx, y, z)-<p (x, y, z)


dx, y,
p
+ <p (x + dx, y +
2)
(#+6te, y+dy, z),
+
?/+ dy,
<p

dy-(ly

du

+ d-zdz.

du ,

It

is

would be easy to prove, in a manner quite


similar to that followed in the preceding case (No.
independent of the
19), that this final result
order of differentiation.

1.

u = xy,

2.

u=

3.

EXAMPLES.

u = x'

4.

u = sin (x

du = ydx + xdy,

x
-,

y,

,
ydx
xdy
-,
du

(dx

8xydy,
dy),
r

du cos

4y')-dx

(^x2

y)

y),

du =

4xy",

(x

Generated on 2015-06-01 01:31 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

du=dxdx

du ,

is

where the second member consists of three differ


ences.
The first of these differences exhibits the
differential of the function with respect to x alone
the second represents the differential of the func
tion with respect to
alone
and the third repre
sents the differential of the same function with re
The total differential
spect to alone.
therefore,
according to our notation,

INFINITESIMAL CALCULUS.

40

B.u

^,

du = yzdxxzdy-xydz '
2'

S.u = x\ogW

du = dx\ofr?y +

xzdy-ydz.

Implicit Functions.
21. An implicit function is one whose value is
only implicitly given in an unsolved equation.
Thus y' 2xy = a' is an implicit function of x;

whereas, if we solve the equation, we shall have


the explicit function

When the function becomes explicit, its differen


found by the rules already given but, as
some equations cannot be readily solved, the func
tion may remain implicit, and its differential is
then to be found by the following process. Let
;

f(x,?/)
dy)

-f

(x,

(x-\-dx, y),

dx,y

dy)

-f(x + dx,

0.

+f(x

y)

y)

-f(x,

f(x + dx,

y)

or, by adding and subtracting the term

0,

dx,

y)

Its differential will be

given function.
(x

be the

tial

is>

Generated on 2015-06-01 01:31 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

y x Va?-\-x'.

INFINITESIMAL CALCULUS.

41

The tirst of these two differences expresses the differ


ential of the function with respect to x, and the
second exhibits its differential with respect to y.
T J*

Denoting the first by - dx, and the second by

-~ dy, we have
dx

'

dy "

Hence, the differential of an implicit function


(x, y) = 0 is obtained by differentiating it first
with respect to x, as if y were constant, then with
respect to y, as if x were constant, and making the
sum of the results = 0.
a'b' = 0, we
Thus, from the equation a'y' + 6

shall obtain

-j- dx = 2b'xdx,

dy = 2a'ydy,

Generated on 2015-06-01 01:31 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and

2b'xdx+2a'ydy =

0.

22. By a reasoning analogous to the above it


may be shown that the differential of an implicit
function of three variables, as
9

fo

2A

z)

>

will be expressed by the equation

42

INFINITESIMAL CALCULUS.

SECTION II.

SUCCESSIVE DIFFERENTIALS.

Generated on 2015-06-01 01:31 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

23. When a function y=f(x) is differentiated,


its increment dy is its first differential. The differ
ential of dy (which is written d.dy, or d'y) is its
second differential.
The differential of d'y (which

is written d.d>y, or d'y) is its third differential ;


and so on.
The differential of the independent variable, inas
much as it is the fixed standard with which the
successive increments of the function are com
pared for determining the rate of development, is
Hence it does not admit
always assumed constant.
of further differentiation.
If the differential of the function be divided by
the differential of the variable, the quotient will be
the first differential coefficient of the function.*
*

The authors who use the metliod of limits conceive the differential coefficient
limit towards which a certain variable ratio if indefinitely approaching.
Their theory is as follows.
Let a certain magnitude y depend for its value on

as the

some variable magnitude x, and suppose the relation between the two magnitudes
to be expressed by the equation
y = x'.

y\

takes the increment h, and if the corresponding value of y be represented by


we shall have
y' = (x + W = Js + Hxh + h',
(3)

and subtracting (1) from (2),

whence

"\

(1)

If x

y'

-y = 2xh + h',

v-^-

=2x +

h.

(3)

INFINITESIMAL CALCULUS.

43

The differential coefficient of the first differential

coefficient will be the second differential coefficient


of the function : the differential coefficient of the
In this last equation, the term 'Zx being independent of A, this increment may
undergo any change of value without affecting 2x. Let, then, h continually decrease
y
y
till it becomes 0. The expression for the ratio
will then be simply 2x.
Hence 2x is the limit toward which the ratio

approaches as h is diminished

which limit the ratio cannot reach until h becomes zero. Such is the process by
which differential coefficients are determined in the theory or method of limits.
This theory, though still fashionable in France and elsewhere, labors under
We remark, first, that when h 0, then also y' y = 0.
great radical defects.
Hence, at the limit, the first member of the equation (8), though represented by
~- in order to keep a trace of the variables, would really
' be%0 Now, to assume
dx
that hy dxy and dy are absolute zeros is to assume that the limit has been reached,
whereas the theory Itself teaches that h = 0 can never be reached, inasmuch as the
If, then, the as
hypothesis h = 0 would exclude all idea of change or continuity.
sumption h = 0 can never be true, how can we assume h = 0, and accept equations
based on such an assumption?

We must also remember that from

in which expression,

if dx

^
d.r

= 2b we derive

were an absolute zero, what would y be but a mere sum

Generated on 2015-06-01 01:31 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

of nothings?
represents, as is well-known,

Again, the symbol

the trigonometric

tangent of

the angle that an clement of the curve makes with the axis of abscissas, the element
itself being represented by (Is = \ dx"1 + dy*. Now, this element is not an absolute
zero; for the absolute zero, or a mere point at rest, cannot form any particular
angle with the axis. Hence d* must be a real quantity; and, if so, dx and dy are
Accordingly the theory which considers them as limits of de
also real quantities.
is
not consistent with itself.
quantities
creasing
Prof. Todhunter, a follower of this theory, to eschew objections, declares that he
considers the symbol

dv

as a whole, and does not assign a separate meaning

to dy

di"/ dx. though he knows that- the student will very possibly (and very reasonably,
too) tugpect that some meaning may be given, to dx and dy which will enable him
dy
as a fraction. The student, however, might humbly remind the Pro
to regard
fessor that the ratio

represents only a particular

state of the ratio

and

that, as this latter, so also the former is a result of division; and, therefore, that dy
And as to the separate meanings of
is a real numerator, and dx a real denominator.
dy and ete, it is not difficult to see that, if x and y be considered as co-ordinates of
a point in motion, dx and dy will represent the developments which x and y are ac-

INFINITESIMAL CALCULUS.

44

second differential coefficient is tine third- differen


tial coefficient of the function ; and so on.
For example, let y = an?. The successive differ

entials of this function will be the following,

quiring in a given infinitesimal instant dt : and if r and v' be the velocities


which they develop in that instant, then dx = vdt and dy = v'dt ; whence
dy

dx

v'dt
vdt

_
~

with

4/
v

Thus it is plain that dy and dx have their separate meaning, and that the ratio
is the ratio of the velocities

with which

x and y are developing at

a given

in

stant. It is evident, therefore, that dy and dx are real and distinct quantities,
which correspond to h infinitesimal, and not to the limit h = 0.
" when h = 0 " are
Perhaps it will be said that, in the theory of limits, the words
only an abbreviation, for the words " when h is continually diminishing towards
zero." This is, indeed, what Professor Todhunter explicitly teaches (Diff. Cole.
But it is obvious that, if h is only diminishing towards zero and never reaches
S 9).
the limit zero, the theory of limits remains without object, and virtually abdicates
in favor of the old doctrine of infinitesimals.
For infinitesimals, as defined by the
advocates of the theory of limits (Duhamel, Diff. Cole.), are just such, variable, mag

This is against nature and against reason. The variable x, when continuously
increasing, does not change suddenly into e-|-A, but it changes directly into x-\-dx.
is necessary to travel
What is then the use of travelling the whole distance
within immediate
back again? Is
natural, when dx presents itself and
reach, to wander away from it. with no hope of finding
again, except perhaps
Is
not more natural, and therefore more
after
long and circuitous journey?
The infinitesimal increment dx pre
reasonable, to pass from x to x -f- dx directly?
hence to derive dx from
cedes, in the order of causality, the finite increment
the limit of
continually decreasing quantity,
and to say that dx
to overturn
the limit of
continually
the order of causality, just as
we said that the acorn
the limit of
con
decreasing oak, or that the most rudimentary human embryo
tinually decreasing baby.
It will be said that all writers on differential calculus, whatever be their theory,
and assume
to be a finite quantity. Yes; all authors, in
always begin with x-\finite value, because infinitesimals cannot be
making their diagrams, give to
finite increment
drawing; but, though they mark out
represented or shown in
it

h,

is

is

visible to the student, they do not mean that the increment


mean that there
an increment, and they declare that what ap
mere infinitesimal, or
the diagram ihust be considered as
Hence they have no need of going to and fro in search of
but they immediately take hold of the nascent quantity dx,

is

it

in order to make
finite: they simply
pears as finite in
nascent quantity.
limit unattainable,

h,

is

if

is

is

h
;

it

it

it

is

if

ft,

able.

it

Generated on 2015-06-01 01:31 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

nitudes as tend to the limit zero, though they never reach such a limit.
To this doctrine of limits we must object another serious defect regarding its
The theory needlessly starts by giving to the independent
method of working.
variable x the finite increment h, thus creating for itself the strange necessity of run
ning for ever after the limit h = 0, which limit the theory declares to be unattain

INFINITESIMAL CALCULUS.

45

dp = 3ax'dx, d'y = Qaxdx', d'y = Gadx', d'y =


.

and the successive


'

dx

0,

differential coefficients will be


'

dx'

dx*

dx*

'

^ = - sin
<lp

= a* log a'

= cos x,

dx'

1.2

-5-^.

(x 4-

'

= a"

ax"~\

a?,

d'y

= sin

1.2

= cos

4.

1)4

= sin x,

= a* (log a)''

^(loga)',g

= ^(logay,

is

which is, in fact, the immediate increment of the variable x, and which
the only
quantity required by the procesB of differentiation.
Prom these remarks, and from right conception of the nature of infinitesimals
as laid down in our Introduction,
the reader will understand how we were logically
compelled to adopt the infinitesimal method: and we are confident that even the
advocates of the theory of limits,
they examine the subject in truly philosophical

if

Generated on 2015-06-01 01:32 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

3.

dx'

Fl-1,g=-(Air,)
l)a

_~

dty

-1) ax-\

= log(a;+l),g =
(x

2.

d'v
rJL = n(n

= n (n

2)

j = nax"-\

ax",

(n

y=

1)

1.

||[

EXAMPLES.

spirit, will not be loath to recognize that our doctrine enjoys the great relative ad
vantage of being able to account for itself without any of those inconsistencies that
mar the substance, or at least the exposition, of their favorite theory.

+ 3)c-,

Maclauriri

(a;

ISFIXITESIMAL CALCULUS.

46

-Jl=[x + n)<*.

Formula.

24. By the aid of successive differentials many

it

...

A + Bx+Cx*+Dx' + Exl +

(1)

*Ex* +

= B + 2Cx

C,

D, . . . will be determined
the constants A, B,
by successive differentiations. Differentiating (1),
and dividing by dx, we find
.

(2)

Differentiating (2), and dividing by dx, we find

3.4JElr'+

= 2<?+2.3Ar

(3)

Differentiating (3), and dividing by dx, we find

Generated on 2015-06-01 01:32 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

if

if
y

is

functions have been developed into series, from


which the functions themselves can be easily calcu
lated.
The formulas most used for such develop
ments are Maclaurin's and Taylors formulas.
By Maclaurin's formula the function of one vari
able
series arranged according
developed into
to the ascending powers of that variable, the co
efficients being constant.
Thus,
is
function
of x, and
admits of being developed in the
form

2.3Z>

+ 2.3.4ir +

(4)

INFINITESIMAL CALCULUS.

47

and so on. Now, as x may have any value con


sistent with the convergency of the series, assume
x = 0, and let
<*".

what y,

be

sis.

Then

dx

(i). my

dx'

become in this hypothe-

W7

whence

V<W'

2.3

V^V

is

dy

= sin

a;.

d'y

We
.

To develop

EXAMPLES.
have
.

is

it

it,

In using
This is Maclaurin's formula.
necessary, of course, that the values attributed to *
be such as will make the series convergent.
If function not susceptible of development by
this formula, the formula itself will give notice of
the fact for, in such a case, some of its constant
factors will become infinite.

1.

Generated on 2015-06-01 01:32 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and, substituting these values in (1),

INFINITESIMAL CALCULUS.

48
hence

= (IH, (H (S)-1.--

and substituting in the formula (5),


sin

-a;

a;

3.3

1.2 3.4.5

1.2.3.4.5.6.7+

This series, being differentiated and then divided


by dx, gives
cos
2.

To develop y =

y = (l+*),

-j

- j-g +
(1

+ a;)".

2 3 4 5 6

We have

gU^l+aO"-1,

g=
(^

-1)

fa

n(n-l)(n-2) (l+)-8,

Generated on 2015-06-01 01:32 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

hence

^-w(w-l)(/i-2),

\dx>)
and therefore

(l + aOn =

3.

n(n 1)

To develop y = log

1)

(1

+ x).

( 2)

We have

y = iog(i+a!),g=-iA_,

^|/_

r^5-

_i

1.2

<fos_(l+tf)a'

'

'

INFINITESIMAL CALCULUS.

49

hence

=.='. (SO
and therefore

=-].(SH----

log(l + aO = tf- 2"+g


CC*

4.

To develop y = a".

V = a"

>V9

We

CC^

--5 +

have

2 = a* log a' S= a" (log a)''


^ = a*aoga)',

we make a
we have

whence log a

log

",

^rf+^rf
e

If

e,

(loga)* +

. .

then

Taylor's Formula.
25. Let

w ./(#) and

if

is

it

Considering
a; and
as two arbitrary parts of
certain line,
obvious that,
the line receives an infinitesimal in
crement, the result will be the same, whether the
increment be attached to the part x or to the part h.
In other terms, the result will be the same whether
the function u'=f(x-\-h) be differentiated with
A

Generated on 2015-06-01 01:32 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

*=

and therefore

= dog

1,

(SO

.-.

whence

INFINITESIMAL CALCULUS,

50

regard to x, considering 7t as constant, or with re


In the first
gard to h, considering x as constant.
case, the differential coefficient of the function will.
be

in the second

da;

case

it will

be

dlv

and there-

fore we shall have

du'

_ du'

dx~~

dh'

'

f1

' '

This equation will afford us the means of de

P 4- Qh

Rl?

4- Sh*

+ TV +

u' =

ar-Ti,

veloping the function u' =f(x-\-7i) into a series


ranged according to the ascending powers of
with coefficients that are functions of x alone.
Let us assume a development of the form

(2)

h' 4-

h' 4-
dx

ft1

4-

and divid

dx

then, differentiating (2) with regard to

'

(3)

>

dx

'

"

4-

h,

dx'4- dx

'

dx

2Rh + SSh'+iTh* +

||

ing by dh,

(4)

Now, by (1), the first members of (3) and (4) are


equal hence their second members are also equal,
and the coefficients of like powers of
in those
second members are equal.
Therefore
h

Generated on 2015-06-01 01:32 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Q,

R, . . . are functions of x alone.


in which P,
Differentiating (2) with regard to x, and dividing by
dx, we shall find

INFINITESIMAL CALCULUS.

51

P is

the value u of the function when h = 0


and therefore dP du; hence we shall have

But

_
r~U)
'

_ du

M-2-

d'u

d*/t.

*-2~d- d&*

dx~*'

__
~

'

d'u
dx*

' '

"

'

These values substituted in the equation (2) give

^2^4+

'

(5)

This is Taylor's formula. The values attributed to


x and h must be such as will- render the series con
vergent.

_ ,

1N

rca*-1

7i)

7i

(x

hence

1)

= n(7i-l)(7l-2)a!-

whence

(Pu_2_

(&cs

x'

d'u

(a;

+ 7i); and
2.3
a;*

'

.r

then ' = log

'

ate'

2.
3

+
x

'

dx

d'u _

' 1

du_l
~

'

we have

= log a?/

xn-* h*

n(w-l)(n-2)flf.. y+

Let

and we have

d*u

h)n

2.

Generated on 2015-06-01 01:32 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

du

-f-

Let u = x* ; then u' = (x


,

1.

EXAMPLES.

INFINITESIMAL CALCULUS.

52
/

Let u =

3.

ex

; then

v! =

A*

e?+h

,
, d'u
'
dx'~ '
dx~

du

A*

A*

; and we have

d"u
' ' '
<ta2_e

whence

"'

e",

e*(I+"+f + 2-3 + 244+

In this equation make x = 0 and

'

Then we

1.

have

or
e

2.3T2.3.4

= 2.718281828459

Which is the basis of the hyperbolic logarithms.

Let u = sin x, and u' = sin (x + y)

4.

then we

d'u
du
-r- = cos x, jzi =
dx'
dx

sin

d'u

a;,' -=-=

dx?

= cos a\

a;

'

^-

we shall have also

into

= sin

sin rc

rf47ocOSa'-

cos .t

+ |_c0sa;

sin

2i4Sm^

sin

cos rr

we change

y)

If

-2^cOS^

-\-

y)= smx

u'= sin (x

-\-

whence

(a;

Generated on 2015-06-01 01:32 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

have

a;

2-^_Sllia.+

INFINITESIMAL CALCULUS.

53

Scholium. Taylor's formula may be used for the development of a


function u = f(x, y) of two independent variables.
If we begin by giving to x an increment h, we shall have
.

fix + h,y)

= u +

du.

^h

cPu h?

Wl^

d3u

h3
.

^K-3+

we give to y an increment k, the lirst


+ h, y + k), and in the second member u

and in this equation, when


member

f(x

become

become

du.

U+dyk

*'"~f"
+
~5r
dy

dy'

become

(d*u +

dSMli

dy*

-will

)h,

\dx^~

dx*

2+d^2^

become

(du
d'u K'

k3

dx

d3u

dy~'

'h will

d'u

will

will

/2"'

__+

d3u k3

dxh + djT

du

=+

ajk

*)

'

(A)

We might have begun by giving to


this case we would have directly obtained

f(*,y

Mfc

dy

<fa''

VdW

d'

~W

dC^)
+

d/^

u+^k+ 5-,-

dw K'

<Pu W

the increment A;; and in

d3u k3

W 2+d?0

+,

du,

'

f(* + h,S +

k)

Substituting these values in the above equation, we shall have

Generated on 2015-06-01 01:32 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and so on.

1ZFIS1TESIMAL CALCULUS.

54

In this equation, when x receives the increment h, the first mem


ber becomes f(x + h, y + k), as before, and in the second member u
becomes

du

dy

becomes
d

(du

\dy
d'u

h?

(d*\

d*(du)

\dy)

dx

\dy)

h*

dx\

)*'

^rk

d'u

d?u h?

k? .

3-5 becomes
dy*

dx

KdyV 7t'
.
" T TJi
IT
dx*

ft

\dyV
and so on.

Substituting these values in the above equation,

dx

f--\
KdyV hk*
2

dy*

d'u W

h?k

dx*

(B)

d'(dU)
\dy/

=f(x,

y), the equations

(A) and (B) may

be

-j

hk +

dxh

'

du

Jj/k

du

-f(x,

V)

k)_f{x,

Txh

as (A) and (B) are identical, we


efficients of hk are identical whence
;

and,

h,y

du,

du

Tyk+

\dy'

, ,

-^-hk

/du\
,

y)

f(x

h,y

k)

.,

f(x

Generated on 2015-06-01 01:38 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Remembering that u
written as follows

7i3

dx

ay

+ . .

d*u

h*

\dy/

du

d'u

du,

we shall have

...

conclude that the two co

INFINITESIMAL CALCULUS.
,

(du\

55

d fdu'\

\dxJ

Uy>

(0.

dx

dy

When h and k are infinitesimal, the equations (A) and (B) reduce
to

du

du ,

which is the total differential of the function (No. 19). It is, there
fore, the property of such a total differential that the differential
coefficient of
coefficient

of

^ taken
^

with respect to y is equal to the differential

taken with respect to x.

De Moiwe's Formulas.

26. We have found (No.


cos^ =

i_|, + t5|4-2^)T;+

- a3;6fi7 +

smx=x-^ + ^g-j-g
If we

change x into x V
shall have

x3

(1)

P)

in this last series, we

2.3

2.3.4.5

#4

^2.3.4
(4)

If, on the other hand, we multiply both members


of (2) by V 1, and then add the result to (1), we

cos

shall find that the sum

-f-

Generated on 2015-06-01 01:38 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

24) the three series

sin x

INFINITESIMAL CALCULUS.

56

will give a series identical with that of equation

(4).

Therefore
cos x

If

we change

x into

cos

a;

V^l

v 'sin x e* '.

(5)

x, we shall have also

x V

And now let


cos mz

+V

sin

a;

= e-*v-l.

(6)

= W2, to being any number

sin mz = e~v'ri =

then

(e**rr,)m,

that is,
= (cos 2+ ^

+ V

sin

cos mz V

sin mz = (cos

cos tos

TO2

Generated on 2015-06-01 01:38 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

as also
2

V7

sin

2)m,

(7)

sin

2)m.

(8)

These are De Moivre's formulas. Each of them


is equivalent to two ; for, after developing the
second members, the real terms will form an equa
tion among themselves, and the imaginary terms
will form another, which being divided by the com

mon factor V 1, will express another real rela


tion. Thus, making m = 3, formula (7) will give
cos 32 = cos' 2 3 cos 2 sin' 2,

sin

32

cos' z sin z sin' 2,

Maxima and Minima.


37. A function is a maximum when it reaches a
value greater than the values immediately preced
ing and immediately following, and it is a minimum,
when it reaches a value less than the values im

mediately preceding and immediately following.

INFINITESIMAL CALCULUS.

57

To determine if a function u=f (x) has a maxi


mum or a minimum for a certain value of x, the
following plain rules are laid down.
If the function, for a certain value of x, becomes
a maximum, it must reach that maximum by a last
Hence the
increase, and then begin to diminish.
differential du must he positive immediately before,
and negative immediately after the maximum.
If the function, for a certain value of x, becomes
a minimum, it must reach that minimum by a last
diminution, and then begin to increase. Hence
the differential du must be negative immediately
before, and positive immediately after the mini
mum.

When du changes from + to


so

Generated on 2015-06-01 01:38 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

as

or from to +,

.
And,
^a quantity subject to the law of continuity can

also does the differential coefficient

change its sign only by becoming zero, or infinity,


no value of the variable will give a maximum or a
minimum value to the function, unless it reduces

^to

zero or to infinity.

Hence the roots of the

equations

du

-.du

will give all the values of x which can possibly


make u a maximum or a minimum.

Let a be one of such roots. It is yet necessary to


ascertain whether x = a corresponds to a maximum
or to a minimum.

If in

the expression of

we

put first a dx, then a + dx, instead of x, and if

ISFIXITESIMAL

58

CALCULUS.

the first result be positive, and the second negative,


it is plain, from the preceding considerations, that
the value x a will correspond to a maximum ; if
the first result be negative and the second positive,
the value x = a will correspond to a minimum ; but
if both results be of the same sign, there will be no
maximum and no minimum for x = a.
Assume, as a first example, the function

(a-xf + b,

g =
which gives

'
f=^2(a-x).
ax
a.
Putting
(a x) = 0, we find x

a
dx, then a + dx, instead of x, the expression
2
(a x) becomes successively
2

(>i

(a (a
dx)) = 2dx,
2
2dx
dx) =

(a

Making

b.

is

mum, which

?/

x)

(a

= x gives xco,
The hypothesis
which cannot verify the conditions of either
maximum or minimum.
"
Assume, as second example, the function

= 2ax

x%

from which we have

dy

0,

x)

Making

(a

=^
dx

Generated on 2015-06-01 01:49 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

is

and, since the first result


negative and the second

a corresponds to a mini
positive, we see that x

V2ax-x'
we have

x = a.

Putting

INFINITESIMAL CALCULUS.

59

x)

= dx, (a

(a

becomes

dx)

dx.

dx)

dx) =

(a

(a

whilst the expression

dx) -\-dx,

(a

x)

(a

-f-

(a (a

a dx, and then a + dx, instead of x, the expres

becomes
sion

a.

= a, and the mini

The maximum

mum

is

sign of

Hence x = a corresponds to a maximum and to a


minimum at the same time, owing to the double

The hypothesis

fl*
.

co

gives 2ax x'=0,

is

0,

be
whence x
or x = 2a. But in both cases
no other maximum
comes =
and thus there

or minimum.

and minima can be made to depend entirely on its


successive differential coefficients.
Let the function
(x) be at its maximum or

and u"

=f

h)

(x

u' =f

iif
h)

minimum, and let

(x

Generated on 2015-06-01 01:49 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

28. When a function of x can be developed by


Taylor's formula, the determination of its maxima

values of u immediately before and immedi


ately after the maximum or minimum. By Taylor's
formula we have
be the

INFINITESIMAL CALCULUS.

60
whence

'*(-+

+...)

u'

u" U Jl

/ du d'n
V"dx + dx'
.

h
2

dhi

+ dx*

2.3

If u

is a maximum, then u>u', and u>u" ; hence


If u is a
the two series must be both negative.
minimum, then u<u', and u<u", and the two
series must be both positive. Now, the two series
cannot have equal signs unless the terms

-jj- disappear;

^ and

for, when h is very small (as it

Generated on 2015-06-01 01:49 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

must be in the present case), the sign of each series


is the same as that of its first term. Hence we
must have

and the roots of this equation will give the values


of x for which the function u will be a maximum
or a minimum.

Let a

be

one of these roots.

As

pears, the two series reduce to

<-*-'().+().&+.
and their signs will be those

If

this

ISFISITESIMAL

CALCULUS.

61

term be negative, ua will be a maximum


positive, ua will be a minimum.

If x a

were to give also

(^^j =0,

if it

be

then the two

series would reduce to

u~u-

- \a&)a
.

+
h'

ld'u\

/d'u\
2.3.4

and thus they would again have different signs.


Hence there could be no maximum and no minimum,
unless

=0-

In this

begin by the term (^J;)

case the t wo series

^\

which,

if

would

positive,

Generated on 2015-06-01 01:49 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

would give a minimum, and if negative, a maximum.


If this term also were to become = 0, we would

have to proceed as before with regard to the subse


quent terms of the series. From all this we may
draw the following conclusion :
the first differential coefficient which does not
become = 0 is of an uneven order, the two series
have opposite signs, and there is no maximum or

If

minimum.

If

the

first differential
= 0 is of an even

which does not become

coefficient
order, the

have equal signs, and there will be a


maximum when their sign is negative, and a mini
two series

mum when the>r sign is positive.

29. The investigation of maxima and minima


may often be simplified.

Thus a constant factor

INFINITESIMAL CALCUL

62

that affects the whole function can

in the differentiation

VS.

be

suppressed
=

for, since we have

the result is independent of any such factor.


So also, if we have a function of the form
Va*x
<

0,

y=

- bx\ we can make y' = n = a'x bx', whence


a1

2bx,

we have x

a'

= -rr

2b

2b.

and
.

Making

0,

This value makes u a maximum

hence it makes if a maximum, and consequently


also y a maximum.
By this artifice we can differ
entiate the function without taking notice of the

radical sign.
And, again, we may simplify operations by tak
ing the logarithms of the quantities to be differen
tiated. Thus, if we have a function

_(x-l)

Generated on 2015-06-01 01:49 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

y~(x + l)

(x 2)
( + 2)'

passing to logarithms, and making log y = u, we


shall have
w=log [x l)+log (x-2) log (x-\-l) log (x-\-2),
and

_\_
dx~ x -1+ x - 2
du_

x+

1_

x + 2~

'

from which we obtain x = V 2.


It is obvious
that, when this value of x makes u a maximum or a
minimum, yalso will be a maximum or a minimum.
When the first differential coefficient is a pro
duct of two or more factors, and one of these fac
tors becomes = 0 for a value of x corresponding to

INF1NITES1MA L

OA LC'UL US.

63

maximum or to a minimum, the second differen


tial coefficient can he obtained without differen
tiating the other factors, as in the following ex

ample.

Let

=PXX*,
P,

Q, and

being functions of x.

The regular dif

ferentiation would give


dx

dx

dx

'

Generated on 2015-06-01 01:49 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Now, if the factor i2, for instance, becomes = 0 for


a value x a, it is evident that the differential will
reduce simply to

Hence it will suffice, in such a case, to multiply the


other factors by the differential coefficient of the
factor which becomes = 0.

Exercises on Maxima and Minima.

30. The application of the preceding principles


to the solution of problems is not difficult, though
the student may, at times, experience some difficulty
in finding out the mode of expressing the particular
function which is to be worked upon. A few ex
amples will show how the difficulty may be prac
tically overcome.
I. Required the dimensions of the maximum

cylinder that can


cone.

be

inscribed in a given right

INFINITESIMAL CALCULUS.

64

Let A VB (Fig.

1) be the cone,

and suppose

cylinder inscribed. Let VC


= h, AC=r, VO = y, DO
x. The volume V of the
cylinder will be expressed
by

V=nx'(7>-y).

But from the similar tri


angles A VC and D VO we
have

y : : AC

VC ::r

h, and

hx
Substituting

this value of y in the preceding ex

pression, we have

V= Tx^(r

x).

Generated on 2015-06-01 01:49 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Hence

Placing

^
d'V

and x =

makes it positive, the second makes it nega


Hence x

tive.

2r

V=

find the roots # =

The first value placed in the expression of

ar

0, we

=0

corresponds to a minimum, and

to the maximum

27

of its base is

required, which will be

Its altitude is

2r
-5-

\o

whilst the radius

INFINITESIMAL CALCULUS.

65

II. In
tres

the line CC (Fig. 2) which joins the cen


two spheres, to find the point from which the

of

greatest portion of spherical surface is visible.


Let J. be the point

Draw

sought for.
the tangents

= r,

AJ/and
radii CM

the

CM'

r', and

c'<

AM',

MP, M'P'

the lines

perpendicular to CC.
Make CC = a, and

AC=x.

is

The portion of surface visible from A on the


one-half of the sphere, minus
zone
right hand
of the altitude PC, or
2nr*

2zr X

PC;

PC r::r

x, and

x PC -r'

+ rn

P'C r'::r'

2n (ra

But

the total visible surface, we have

Calling

W - 2w' x P'C.

P'C).
a

x;

= 2-

whence

P'C=
r"

a x

and therefore

- x- - a-?)xl

and

(r'
1
+

7iC=

hpnce

Generated on 2015-06-01 01:49 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

is

one-half
and the portion visible on the left hand
of the surface of the other sphere, minus a zone of
the altitude P'C, or

INFINITESIMAL CALCULUS.

66

ds
dx
and making

0,

we shall find

VP

xa

+ Vrn

Such is, then, the distance from the centre C to the


point A that satisfies the
condition expressed in the
enunciation of the problem.
III. Through a point P
(Fig. 3) a straight line is

drawn meeting the axes OX


and OY at A and B respec
tively. Find, the least length
that this line can have.
Let OMa, and PM=b be the co-ordinates of

Generated on 2015-06-01 01:49 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

the given point

P,

and make the angle

PA0

Then

PA
hence, making

b
^ ,'

sin

AB =

and

PB

a
cos #

u, we shall have

b
'

sin &

'
cos #

whence

du
Making

'

cos

&

sina i?

= 0we shall find

a sin
cos' i?

'

&.

INFINITESIMAL CALCULUS.

67

From this last equation we obtain


sin

cos

&

and therefore

This is the minimum required.


IV. To find the least cone that can be circum
scribed abou ' a given sphere.
Let SCh (Fig. 4) be the altitude of the cone,
AG It the radius of its
base. Then the expression of
its volume V will be

Generated on 2015-06-01 01:49 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

V=^rtR\
Draw the radius OP r to
the point P where the element
BS of the cone touches the
The similar triangles
. sphere.
SCB and SOP give us

OB

BS:: OP

that is,

B:

OS,

VB* + h*::r

:h-r;

whence, by developing and reducing, we find

R-JYr-

This value of

R substituted

gives
'

whence

in the above equation,

2r

'

INFIX ITESIMAL

68

CALCULUS.

Ttf (h-2r)2h-h
dV_
~

dh

'

(h-2ry

'

from which we obtain


h = 4r, and therefore B? = 2r'.
The volume of the minimum cone will therefore
be

V=

-g

As

surface of this cone is

the

8 87tr',

we see that as its volume is twice that of


the sphere, so also is its total surface twice that of
the sphere.
V. A man being in a boat a miles distant from

the nearest point of the beach, wishes to reach in


the s7wrtest time a place b miles from that point
along the shore. Supposing that he can walk m
miles an hour, but pull only at the rate of n miles
an hour, required the place where he must land.

Generated on 2015-06-01 01:50 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Let

AB = a

(Fig. 5),

BC= b, BD = x, D

being

the point where the man must


land. The time employed in
rowing from A to D will be

and the time em


ployed in walking from D to
Va*

B will

be 4(6

x);

so that

the total time T employed in


the journey will be

T_

Va' + x'

b-x
m

Differentiating, we find for the minimum

INFINITESIMAL CALCULUS.

69

This value of x gives the distance BD, and the

point D where the man must land.

will

be

^,_nb
Make a 4,

T=2h 40m.

VI. A triangle

-f a Vm'

n'

mn

m 5, n

8,

then x

and

and a perimeter 2p.


What must its second and third side be, that the
triangle be a maximum
Let the second side be denoted by x the third
will then be 2p x. Its area will then be ex
pressed by the equation
b

has a base

'S,

time

The shortest

isosceles.
VII. To

the triangle being

the dimensions of the maximum


solid cylinder whose total surface is
the alti
Let x be the radius of the base, and
Its total surface will then be
tude of the cylinder.

S.

find

= 2nx*

expressed by
&

2nx.y,

and its volume by Vr.x^.y, or, eliminating y, by

4/I

|/|

x=

us - ft-D =

whence
o,

Generated on 2015-06-01 01:50 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

a maximum when x =

x p).

(b

found by the ordinary process that

(p

x)

be

Vp (p

b)

is

It will

INFINITESIMAL CALCULUS.

70

or y=2x. Hence Fis a maximum when its alti


tude is equal to the diameter of its base.
VIII. To find the greatest isosceles triangle that
can be inscribed in a given circle.
It will be found that it is an equilateral triangle.
IX. To find the smallest isosceles triangle that
can be circumscribed about a given circle.
It will be found that it is an equilateral triangle.
X. To find the greatest rectangle that can be in
scribed in the ellipse whose semi-axes are a and b.

Its base will

XI.

be a V% and its
the greatest cone

altitude b V2.
that can be cut from

To find
a sphere.
If r is the radius of the sphere, the altitude of the
Ar
2r
cone will be .=- , and the radius of its base V2.

Generated on 2015-06-01 01:50 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

XII.

To find the greatest segment


that can be cut from a right cone.
If It is the altitude of
the cone, and r the
radius of its base, then
the axis of the segment

will

be

XIII.

| Vh' + r\
To

find

of a parabola

the

parabola
maximum
that can be inscribed in
an isosceles triangle
having the altitude h
and the base b.
Let VH= x, and PIT
= y (Pig. 6) be the co
ordinates of the point P, where the parabola
touches the side CA of the given triangle. The
area of the parabola will be A = * YD x DK.

INFINITESIMAL CALCULUS.

AsCF

=
other hand,

VH= x,

DK* PIT':: VD
:

VD

we have

VB, or

DK'

=.

71

- x.
h

y* ::

On the

-x

x,

and therefore

To eliminate
we have from the similar triangles
^IDCand PBC

PB: AD:: Off:

CD,
or

?/

:: 2a;

A,

or y =
-^-,

hence

Accordingly,

the equation of the problem becomes


46

Generated on 2015-06-01 01:50 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Hence

_ 26

<M

and x =

(h

- a;)3

3a;

Consequently FZ>

(h

- x)' _

and

XIV.

DK=

To find the minimum parabola that can be


circumscribed about a circle.
Let r be th<^ radius of the circle (Fig. 7), VB=x,
PH= y the co-ordinates of the point
of contact,
VA = a, and AB = b the terminal co-ordinates,

72

and 2p the parameter of the parabola.


will be A = $ab.

The area

Now, a = VH+ HO +
OA x -\-p + r, and b' =

2pa. On the other hand,


the triangle POH gives
p' = r' p* = 2px, whence

and therefore

2p

r)*

r)a,

or

=p

3p(j? + r)'- (j? + r)'_


:0,

and consequently 2p = r.
3r

in the expres

'

^4 =
dp

whence

j9a

9?'
Hence a

and

'

What must
cone has a total surface S.
its dimensions be that its volume be a maximum
the ra
Let
be the altitude of the cone, and
= nr*
Its total surface will be
dius of its base.
nr V7?

XV. A

Generated on 2015-06-01 01:50 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Substituting these values of a and


sion of A, we have

r.

&

also

(p

2p

-\-p*
~~

(p

2pr
r'+
~

-f-

"=-2f+r+P

/'

-f

2p

2?a

T*

'

r\

whence

INFINITESIMAL CALCULUS.

73

On the other hand, the volume will be

Hence

dV

S*

2S7tr*

2SttT1

XVI.

To find the maximum cylinder that can be


cut from an oblate ellipsoid of revolution, whose
semi-axes are a and b.

will

and the altitude

make u

a minimum?

x x'
l+x~^x'

-f-

What value of x

ai/s

XVII.

The function is a minimum when x =

XVIII.

Through the focus of an ellipse two


chords are drawn at right angles. Find when
their sum will be a maximum, and when a mini

mum.

mum when x =

XX.

the least ellipse


scribed about a given rectangle.
To

find

x)

-f-

The solution will be reached through the polar


equation of the curve.
XIX. Show that u sin x (1 cos
is a maxi
5

Generated on 2015-06-01 01:59 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

The radius of the base =

that can

be de

INFINITESIMAL CALCULUS.

74

Let 2a and

(Fig.

2b

of the given

8) be the sides

rectangle, and let

Ay + B,x* = A2B'
equation of the ellipse.
In all ellipses we have
be the

y*

-x'

yn :: A2

A"

- x"

if

52

b'::A'

A'

5=

a', or

The area of the ellipse is


for B its value,

Ab

hence

dE Tib%2A(A2

E-AB.

- -

A*
=
\y (A" - ay

-j-.

dA

which gives A

=a

in the expression of
the area of the

V2.

a*)

a?

Substituting

n
0,

And this value substituted

B gives Bb

ellipse will

E = tm V2

VA*

A'
VA' a

E = r.b
Generated on 2015-06-01 01:59 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

we take y B and
J, we shall have x = 0.
a, and

hence,

V2.

Accordingly

be
b

V2 2nab.

Scholium. The theory of maxima and minima, as above developed


(No. 28), can be extended to the investigation of the maxima and
minima of any function of two independent variables.
Let the function

u=f(x,
be at

y)

its maximum or minimum, and let


u'

=/(*

h, y k),

and u"

=f(x

+ h, y + Jc),

INFINITESIMAL CALCULUS.

75

values of u immediately before and immediately after the


maximum or minimum. Then by Taylor's theorem, as extended to
two variables (No. 20, Scholium), we shall have
be the

j-h - du
-r k
dy

du ,
dx

=u

,,
u = "

j-

du ,
du ,
A + j- k
dx
dy

which may be written

(dih
/du

-;-;

dx'

diu 1i?
-r-i

dx>

du

dHi

+ -7

j- hk

dxdy

/dHi ,

du.\

_
+ 2

dyk)+2 (

d'u , ,
d-u
-jj-hk +3-5-5- ~
dxdy
dy' 2
du

k>

+ -y-3 -^
dy2 2

etc.

+ etc.

as Hollows :

u' ~ =

~ /du

d>u 7i'

l/<78w,

d-u

17

dxdyKk

d2

d-u ,\

..

*")

"et0-

e?!if,\

>

>

dyk) &nd+{dIh

du

/du

du
+

-(dxh

dJ,k)

necessary to have

du ~
-Ydx

and

du
-5- =
dy

0.

separately

we must have

(1)

which corre

These two equations will give the values of x and


maximum or minimum of the function.
spond to

T-k

dy

du

are independent of each other,

and

and since

du ,

dx

is

is

and
are very small (as they must be in the
for, when
whole
the
of
the
series
the same as that of its
case),
sign
present
existence
of
maximum
or of
Hence
for
the
minimum
first term.

disappear

it

Generated on 2015-06-01 01:59 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

(du ,

be

if

', and
maximum, then u
u", and the two
If u be minimum, then u <u', and
series must be negative.
But the two series can
u < u", and the two series must be positive.
not have equal signs unless the terms
Now,

Since the first differential coefficients


first term of both series will now

be the

-=-

and

trinomial

-5-

disappear,

the

INFINITESIMAL CALCULUS.

70

which, if negative,
minimum. Making

will

the trinomial

/efu,.

d'u

will indicate

,,

d!

,,\

a maximum, and,

if

positive,

take the form


+

2Bhk +

CW,

or

or, by adding and subtracting

Now, since

h and

cannot assume

JJ2

-j. and factoring,

7 ,'

and therefore

As

to the second term,

cannot

the first term


be
,

0,

f^

JB, we
+

and is always posi-

it is easy to prove that it

For as the sign of (2) must remain unchanged


cannot be negative.
for all the small values of the arbitrary constants h and k, it follows
that the value of the expression (3) must not pass through zero. But,
if we had AC B2 < 0, we might choose for 7t and k such arbitrary
values as would give
Kk

2)

m-AC

would pass through zero.


Hence the assumption
inadmissible; and we must therefore have either
or AC B* = 0; and thus in both cases the factor
within the brackets will be positive.
Hence the sign of (2) will be
the same as that of the other factor A.
It follows that the existence of maximum or minimum cannot
be inferred from (2) unless either AC> BP, ox AO = B*
that
is

(2)

unless

is,

AC B2<0
AC B? >

0,

that

is,

Generated on 2015-06-01 02:00 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

of the factor within the brackets


tive.

and independent of A and

k are arbitrary,

INFINITESIMAL CALCULUS.
'

dx*

d>f

>

'

\dxdy)

dxl

'
"dy1

11

\dxdy)

'

'

The values of x and y which ought to satisfy either the one or the
other of these two conditions must, of course, be taken from the
The conditions

and

maximum, and when

dx*

d?u

dx

is

as

the function u

negative

that of

will

positive the function u will be

be

when

always the same

and as

is

it

rplain that

is

2
,

or of the trinomial,

(3),

is

of

the sign

must always be either both negative or both positive

ay

-r-;
dx*

that the differential coefficients

(3) show

'

d^u

jjjj

equations (1).

inini-

must then be determined by the sign of the fourth differential


after having ascertained that the third differential coeffi
cients, which have opposite signs, reduce to zero, as the theory (No.

it

but

0,

But let the student remember that, although the existence of


maximum or minimum cannot be tested by (3) when the trinomial
=
yet even in this case there may be a maximum or minimum

is a

mum.

coefficients,
38)

requires.

A cistern which

depth.

The total surface u will bo

u = xy +

a%

a3

=:

a3

xy

+
3

therefore

minimum.

Differentiating first with regard to x, then with regard to y, we find


3a3

du

2a>

dy

x2

du
dx

x'y =
and therefore

3o3

= y*x, and

the base must be


a3

a3

xy

square.

ay'3

hence

Generated on 2015-06-01 02:00 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

is

to contain

is

I.

certain quantity of
water
to be constructed in the form of
rectangular parallelopipedon.
Determine its form, so that the smallest possible expense shall
be incurred in lining the internal surface.
breadth, and
Solution. Let a3 - its content, x = length,
Exami>le

The depth will be

INFINITESIMAL CALCULUS.

78

the depth must be equal to half the length or breadth.

and therefore
Since

_~ 4,i* _~ 4s

d?u

dx*

tPu
'

dy*

_~

iPn
'

dxfiy

_
~

'

the first of conditions (3) is satisfied, and u is a minimum.


Example II. Find the values of x and y which shall make the

function

it x* + y* boxy*

maximum or a minimum.
Solution. Here we find

~ =

4x3

iay- =

= Aif

0,

8axy 0 ;

hence

- 2a?y/2 =

a?y/~8,

's positive,

dj;' and

ay5

the first condition (3)

12/

ay/%.

8ax = 16a? V2,

satisfied

a a/2 and

and

as the

a//8
v

sign of

make

24a5,

is

and therefore

find

dx*

and

dy1

0.

And

these values

are also satis

of

and

we

since in this case the third differ-

ential coefficients are


d3a

With

and

(-,

a;

and

0.

fied by taking

the equations
0

example,
x

In this

minimum.

Generated on 2015-06-01 02:00 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

^ = 12x*

And again,

y =

if

ay/2,

Zax, x* = %a?x, x* 2a?

x =

y' =

ay'-,

f"r

x3

d'u,

and the fourth differential coefficients

INFINITESIMAL CALCULUS.
are positive,

19

we conclude that the values x 0 and y = 0 correspond

to another minimum.

Remark.

When u is a function of three independent variables,

the conditions of its maxima and minima are determined by a process


analogous to the preceding, but which is based on the extension of

Taylor's formula to a function of three variables, and is too long to


be inserted here.
The result, however, of such an investigation is
If a function
simple enough.

u=f(x,
has any maximum or minimum,


~
dx

it must


dy~

0'

y, z)

0'

give

dz

_~ q

'

and the values of x, y, z found from these equations

must satisfy the

condition
\ dxi

'

\dxdy)

dy*

'

d'u

dht

\dydz

dx*

'
dx'1

\dxdz)

dz*

tPu\*
dxdz)

dxdy

The function will bo a maximum if the two factors within brackets


in the first member of this inequality are negative; but a minimum
Generated on 2015-06-22 21:33 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

if

they are positive.

Values of Functions which assume an Indetermi


nate Form.

31. It sometimes happens that in giving to the


variable a certain value, the function assumes one
of the forms
0

oo

0, w

Thus the fraction


|j

0
^
X

~
~,
oo , x

ar1

- ~, OXqo
oo ,

when x1, takes the form

though its real value is

2.

The real values of

INFINITESIMAL CALCULUS.

80

functions that assume the form

can be found by

the following process.

Let z and y be functions of x, and let u =

be-

come r when x = a. Clearing of fractions, and differ

entiating, we have

udy

-f ydu

= dz ;

but when x = a, the term ydw disappears.


(udy)a = (dz)a, or
M.

Accordingly,

Generated on 2015-06-01 02:00 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

j|

Hence

=().

the value of u, when it takes the form

for x=a, will

be

found by differentiating sepa

rately the numerator and the denominator of its ex


pression, and substituting in the resulting fraction
the value x = a.

If

were again of the form

we would

ap

ply again the same process of differentiation, and we


would obtain

if

necessary, we might continue the same pro


cess until a determinate value is reached.
Thus
and,

the fraction

sin
,

cc

becomes

when x = 0 ; but by

the process just explained we successively obtain

INFINITESIMAL CALCULUS.
Fx

sin

Tl

x~\

cos

81

("sin afl

afl

Tcos

af|

or the form

go

this re-

Thus the function

not needed.

which takes the form

Sometimes

when x =

oo

lo"
-
x a

will give

is

that

easily obtained

The reduction

above explained.

by remembering

and then applying the process


is

to the form

duction

oo

its real value may be found by first reduc-

ing

-
it

^r

or the form

sumes the form

|,

When the function, for a certain value of x,jls-

im

The form

For let

oo

oo

may also be reduced to the form

and w be two functions of x, which

jj.

for certain value of x become infinite.


Then the
function u = w becomes oo oo for that value
of x. But we have

[l--J= -
V

u = v-w =

provided we have

oo

the form

and w = co, the function will take

and when

Generated on 2015-06-22 21:32 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

mediately

0.

If

this

INFINITESIMAL CALCULUS.

82

condition were not fulfilled, then the equation u

=v

(l

would make the function infinite.

Assume

u
when x

~,

\ sec x x tan x ;

the function

the form

takes

oo .

But we may write

and from this we shall obtain (u)n

1.

Generated on 2015-06-22 21:32 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

32. The indeterminate forms

0, oo, 1", can be


the
reduced by
Let v and w be
following method.
x
of
of
such
two functions
a nature that, when
=
the
cause
x a, they
expression u = vw to assume

one of the forms 0, oo , 1.


the base of the Napierian
Now, the
have " = ewlogu.
each of the three proposed

Ox,

Since v

= &ogv

(e

being

logarithms), we shall
exponent w log in
cases takes the form

which can be reduced to

plained.
EXAMPLES.

as we have ex

Generated on 2015-06-01 02:00 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

io.
=1
r-V-i
up
Jo

11

\a

8.
'

[^]0

cot

<r

X*

_
'

0,

=
4

tan x
~
+ tan x\l

~"|

[a;

a;]0

[tanglog(2-5)] =

\2x- V5x'-aya~
.

Ll r

log

'

5.

[(i_aOtanf]=?,

6.

3.

&H

INFINITESIMAL CALCULUS.

;T"

tan,-.
=1,

a;

=log-log&.

83

INFINITESIMAL CALCULUS.

84

III.

SECTION

INVESTIGATIONS ABOUT PLANE CURVES.

33. Let y = f(x) be the equation of a plane curve


MPL (Fig. 9), and let x = OA and y = PA be the
co-ordinates of a point
Fig.
P of the curve. Draw
PT' tangent to the
curve at P, and PX'
parallel to the axis
OX; and make the
angle T'PX' = &.
Let the point
be
consecutive

P.

the

The inflnitesimal increment PQ


of the curve entails an increment AB of the ab
scissa x, and an increment QR of the ordinate y.
If then we represent by s the portion MP of the
curve, then PQds, whilst AJ3 = dx, and QR
= dy. Now, we have
point

AB = PQ

cos &,

QR = PQ sin #, PQ' = PR*+ QR',

dx = ds cos

dy = ds sin

&,

hence
j?,

ds = \fdx'

+ dy',

and

dx
ds

tan

&.

Generated on 2015-06-01 02:00 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

to

INFINITESIMAL CALCULUS.
As the tangent

85

PT is

but a secant which meets


the curve at two consecutive points, it follows that
the tangent and the curve have a common infini
tesimal element PQ, and that the angle which the
element PQ of the curve makes with the axis OX is
identical with the angle # made by the tangent at
P with the same axis.
The trigonometric tangent of the angle # is taken
as a measure of the slope of the curve at the point

P,

and, as tan # =

the differential

coefficient

of the ordinate of any point of the curve is the


measure of the slope of the curve at that point.
Tangents, Normals,

etc.

34. The equation of a straight

line

Generated on 2015-06-01 02:00 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

passing
of
a
whose
co-ordi
curve,
through two given points
nates are x', y', and x", y", is

v"v'
When the two points are consecutive, as

y' = dy', and x"


(Fig.
the equation becomes
9), then y"

P and

x' = dx'; and

This is the equation of the tangent to the curve at


the point P, whose co-ordinates are denoted by

y\

x'.

Making y =

we find for the point T, where the


tangent meets the axis OX,
0,

INFINITESIMAL CALCULUS.

86

xX

.y

,dx'

dp"
The subtangent A T is evidently = x'

x= y

,dx'

35. The normal being perpendicular to the tan

dy'

instead of +
1

-j.

7^7

gent at the point of contact, its equation can be de


rived from that of the tangent by substituting

dx'

y-y'=

Hence
dx

3?

0,

is

the equation of the normal to the curve at the


point (x', y').
we find for the point ilT, where the
Making =
normal meets the axis OX,

AN

dx'
evidently equal to x x'

dx'

to

is

36. An asymptote to a curve


line that con
tinually approaches the curve and becomes tangent
a

//

is

T,

10

it

at an infinite distance. Such


line will,
1y
of course, cut either one
or both the co-ordinate
axes at
finite distance
from the origin.
Let the straight line
A T (Pig. 10) be an asymp
tote to the curve LP.
Since A
when infinitely
prolonged,
tangent to
the curve, its equation will be of the form
Fig.

'

Generated on 2015-06-01 02:00 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

The subnormal

is

x --- x'-\-y

INFINITESIMAL CALCULUS.

dj/

(a?

y-y

dx'

87

x'),

0,

y'

x' and
being the co-ordinates of the point of con
tact infinitely distant from the origin.
This equa
tion, when x
gives

it

0,

and, when

OB;
gives

OA.

If

if

dv'
determined by the value of ~~

If

both OA

it

and OB are infinite, the curve has no asymptote, as


no place can be found for
in the plane of the co
ordinate axes.

37. Let us inquire, for example,

whether the

We find by differentiation

-f 3x ~
_
y

_
dx~

dy

,3x'

+ 3x
s/2x + 3x'

'

has any asymptote.

= 2x +

y*

curve

Generated on 2015-06-01 02:06 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

be

if

is

the values of OA and OB obtained from these


equations are both finite, the asymptote will inter
sect both co-ordinate axes
one of the two values
infinite, the asymptote will be parallel to one of
the axes
both values are zero, the asymptote
will pass through the origin, and its direction will

This value being put in the expressions for OA and


OB will give, after reduction,

88

INFINITESIMAL CALCULUS.

OA=-

+ az'

And now make x

OA

OB =

V2x +

3x'

Then

~~
3' OB=

V2

Hence the curve has two asymptotes,


intersects the axes at the distance x

x=

and the other at the distance

=
^,
O

one of which
1

5,

y-

y-

J_

Vr6

~V3

Again, let us inquire whether the curve


V = log x
has any asymptote.

Here we have

dx

hence

Generated on 2015-06-01 02:06 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

OA x

(1

- log

'

OB = log x
1.
gives both OB and OA infi

x),

The assumption y = co
nite, and cannot correspond to an asymptote ; but
the assumption y = 00 gives x = 0, whence we
obtain (No. 31)
~]g'
log an
OA = [>(!- log z)]0=rl

[1

Jo
1

= Mo = 0

1^

and therefore the curve has an asymptote which


passes through the origin, and makes with the axis

OX an

angle whose tangent is -~- =


ace

-=

cc

00 ;

that is,

INFINITESIMAL CALCULUS.

89

the asymptote is at right angles to OX, and co


incides with the axis O T.
Let us inquire also whether the curve

y=

tan

Here we have

has any asymptote.

hence

OA =
Make y

ten-iy-T?L-> OB=tanx--^--

Then

co.

to the

Hence the curve has an asymptote parallel

axis 07, at a distance 5 from the origin.

also to x

corresponds

^~

not only to

= ^

. 71

but

it follows that

^
an
endless series of asymptotes, all
there will be
07, and all at a distance it from
the
axis
parallel to
each other.
To find whether the parabola y* = 2px has

let us put its differential coefficient

asymptotes,

ax

-'m

<^-

the expressions for OA and OB.

We

co

Making

we shall have also x =

oo

shall find, after reduction,

Generated on 2015-06-01 02:06 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

value y=cc

As the

and thus

90

INFINITESIMAL CALCULUS.

both OA and OB will be infinite.

Hence this curve

has no asymptotes.
In the hyperbola represented by the equation

ay

we have

&

V=

dy _ Vx
~
dx
a'y

- a'b'
'

hence, by substitution and reduction,

OA =

-a'

OB=--.
y
h*

y=co,

we shall have also x 00;


and therefore OA 0 and OB = 0. Hence the hy
perbola has two asymptotes passing through the
origin of co ordinates, and making with the axis
OX two supplementary angles, of which the acute
one corresponds to x and y affected by equal signs,
and the obtuse one to x and y affected by opposite

Making

Generated on 2015-06-01 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

signs.

Direction of Curvature.
38. Let M, iV, P (Fig. 11) represent three con

secutive points of a plane


If we
curve y = f(x).
draw the chord MP, and

if

Fig.

11

the point
lies above
the chord, the curve is
said to turn its concavity
downwards ; if the point
were to fall beneath
the chord, the curve would be said to turn its con

vexity downwards.
Let ?/, ?/, y" be the equidistant ordinates of the
points Iff, iV. P. Since y' is consecutive to y, and
y" consecutive to y\ we have

jyFIXITESIMAL

CALCULUS.

91

y' y + dy, y" = y' -\-dy' = y + dy + dy'.


If the curve is concave downwards, as it is in our
But BN=y', and BQ
diagram, then BJSf> BQ.
therefore

%(y

+ dy)>y +

<

0,

<

and consequently

If the

dy + dy'

dy > dy', or

and, by reduction,
<fy

ov2y'>y + y",

P"),

dy <

0,

V'>$(P +

y
+

that

or

0.

i (y + y")
is,

or d'y >

0,

0,

is

curve convex downwards, then BN < BQ,


and consequently dy < ^y'
hence dy' dy>
and

> 0-

Accordingly,

39. If, for

d'x
g

ferential coefficient

is positive or negative.

certain value of x, the first differen(Iff


and
tial coefficient -~ becomes =
the second
be negative,

that value of

maximum.; whereas

oVy

if

x will answer to

differential coefficient

if

0,

Generated on 2015-06-01 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

whenever the second differential co


efficient of the function is negative, the curve is
concave downwards
and whenever the same co
efficient is positive, the curve is convex downwards.
If, in the equation of the curve, we take
as in
reasoning similar to
dependent, we can show, by
the above, that the curve turns its convexity or its
concavity to the left, according as the second dif-

CLX

be

INFINITESIMAL CALCULUS.

92

positive, that value of x will answer to a minimum,


as we have seen (No. 28).
It is plain, then, that
any maxima and minima can be graphically ex
hibited as ordinates to the culminating points of a
curve.

^ = 0,

When

the tangent

curve at the

to the

go

that is,

the tangent may also be parallel to the

y'

axis of y.
Thus in the ellipse (Fig. 12) whose equation

is

g- =

for

maximum or a minimum for


0,

be a

-J^

point of the maximum or of the minimum is par


allel to the axis of x. As, however, there may also

^,(2ax-x'),

we have
dec

If
and

_
-M-

x)

b(a
a V2ax x*

0,

y=b,

then x

a,

value
-\-b being a maximum, and the value
the

6a

mini--

mum.

If

oo,

then

^=0,

and 2ax x* = 0; and

or x = 2a, the first value being


accordingly x =
minimum and the second
maximum.
The double sign which affects the expression for
equally

affects

the expression for

0,

Generated on 2015-06-01 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dy

and

INFINITESIMAL CALCULUS.

93

therefore while one half of the curve turns its con


cavity downwards, the other half turns it upwards.

The value

expresses

gents at the points

OX, and the value

^=

the fact that the tan

and B' are parallel to the axis


00

expresses

the fact that

the tangents at O and at A are perpendicular to


OX, that is, parallel to O T.

Singular Points.

Generated on 2015-06-01 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

40. A singular point is a point at which the


curve presents some peculiarity not common to
other points. Four classes of singular points are
"particularly remarkable ; points of inflection, cusps,
multiple points, and conjugate points.
A point of inflection is a point at which the cur

vature changes from concave downwards to convex


downwards, or vice-versa. This change requires that
the second differential coefficient of the function
should change from to + , or from + to ; hence,
at a point of inflection, the second differential co
efficient must be either = 0 or = 00 . If, then, the
co-ordinates of the curve are such that, for values
immediately preceding and following, the second
differential coefficient has contrary signs, they are
the co-ordinates of a point of inflection.
Thus, the curve

gives

a)'
V o + (x

When x = a, then

0.

When x < a, then

INFINITESIMAL CALCULUS.

94

d'y
<
dx'i

0,

and when x > a, then

x=a

there is a point of in
flection, whose co-ordinates
are x = a, y = b ; and since.
at this point,

= 0,

d'y.

^>

Henoe

0.

at

we have

passing
through it is parallel to the
axis of abscissas, as in Fig. 13
The curve
the

tangent

= a

2/

gives

dy

_a

dx~

Generated on 2015-06-01 02:13 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

When x = \tx, then


j^i >

(l

x
c

'

cos

d'y
a
x
=
cos dx
c
c

d'y

= 0.

When x <

\ zc, then

then
and when x > h~c,
'

< 0. Hence at
dx
x = \~c there is a point of inflection, whose coor
dinates are x = $ nc and y = a.

dx

0,

The slope of the tangent through it is

f~=-,

and is a maximum. The curve has a minimum or


dinate (y = 0) at the origin, and a maximum ordi
nate (y 2a) at the point where x xc.
The curve (Fig. 14) extends indefinitely to the
right and to the left of the origin ; for, making

x = nnc, the values

n = h = t,

will

=i

give a series of points of inflection,


whilst the values

P,Q, ft,

INFINITESIMAL CALCULUS.
n=

will give a

1,

=3, =5,

95

= 0),

series of minimum ordinates

the values

2,

=4, =6,

and

will give a series of equal maximum

ordinates

= 2a).
Fig.

The slopes of the tangents at the points of inflection


will be alternately

Generated on 2015-06-01 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

41. A cusp is a point at which two branches


of a curve terminate, being tangential to each other.
There are two species of cusps. In the first species
the two branches lie on opposite sides of a tangent
drawn at their common point; in the second species
both branches lie upon the same side of the tan
gent.

Thus, the curve of the equation y =


will have a cusp of the first
M
species (Fig 15) ; for we find
Y

and

(x a)

INFINITESIMAL CALCULUS.

96

and when x a, we have =

0,

and

co

The curve has, therefore, two branches, PM and


PiV, the one convex, the other concave, downwards,
which meet tangentially at the point P, whose co
ordinates are y = b, x a, and which do not extend
to the left of P, because, when x < a, y is imagi
nary.
The curve of the equation y x' x? will have a
cusp of the second species
Fig. IS
(Fig. 16) ; for we find

fx

2xlat,

and

^V

+iAxi-

Generated on 2015-06-01 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and, when x = 0, we have

for both branches

is then positive, both


^Js
brandies at the origin are convex downwards. The
two branches stop at the origin, as x cannot be
of the curve.

Now, since

made negative.

If
x=

we make
64
;

dx'

.. 0,

and therefore this value of x corresponds

to a point of inflection.

if

we

we find for the lower branch

make

0, we

In

the same lower branch,


10

find x = ~, which value cor-

25'
dx
responds to a maximum ordinate. This lower branch
cuts the axis OX at the distance 1 from the origin.
From the preceding consideration it is plain that

INFINITESIMAL CALCULUS.

97

there is a cusp whenever the differential coefficient


has two values which become identical at the point
where the curve stops.

42. A multiple point is a point where two or


more branches

If

other.

of a curve intersect, or touch each


intersect, '

the branches

will

ax

have as

many values at that point as there are branches


they are tangent, these values will be equal.
Thus in the curve

y=

VI

if

x',

we have

dy

and when x = 0,' then


Generated on 2015-06-01 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

x'

dx

1,' which shows that

the curve has two branches


intersecting at the origin

(Fig.

17).

ceed 1,

^{Flg'17

As x cannot ex
the curve is limit

ed in both directions.

The

maximum and minimum ordinates correspond to

whence

fill

= 0, in which

case we have

Vl-x*- VI -x'
_^

and the corresponding ordinate will then be

y=h

INFINITESIMAL CALCULUS.

98

In

the curve y x' Vx

dy
dx

-\- 4x

5a?
2

VaJ+l'

0, then 5r = 0.
When x =:0,
then^

Hence the origin O

(Fig. 18) is a multiple point,


where the two branches are
tangent to each other and
to the axis OX. The curve
cuts the axis at the point
a? =

In

1 ,

where

-f-=

<

Fig.

x' + 2x'y

-^>_

- y' =

V-yyVy

x=

18

the curve

we have

we have

+ 1,
1

Vy

Vy +

y
2

becomes either

o
o

and

0,

can

'=
=

0,

which, when x =

*+*

when

the equation of the curve

(I)

be found by putting
under the form

But the real values of

according as we take the upper or the lower

signs.

0,

3y

This expression, when


or =

Vy

-j-

*"

dy _
"'

0,

Generated on 2015-06-01 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

whence

may be written thus

INFINITESIMAL CALCULUS.
it being obvious that
ratio

= 0.

when

99

is nothing else than the

these quantities are becoming each

We thus find

These values show that one branch of the curve


is tangent to the axis
OX at the origin (Fig.
while a second
19),
branch, at the origin,
makes with the same
axis an angle whose

12, and a
third branch, at the
origin, makes with the
same axis an angle
Generated on 2015-06-01 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

tangent is

whose tangent is = V2. Hence the curve has a


triple point at the origin, with two symmetrical
loops under the axis of x.
Its lowest ordinate is y = 1, corresponding to

x=

1,

for which

we

have

= 0.

curve, at the corresponding points


parallel to the axis OX.

43. A conjugate point is

Hence

the

and N, is

point whose co-ordi


nates satisfy the equation of the curve, but which is
isolated, and therefore has no consecutive points.
Thus the equation
y1 x' (x a)
a

is satisfied by the co-ordinates x

0,

y=

0,

but no

INFINITESIMAL CALCULUS.

100

other value of x less than a gives a point of the


Hence the point x = 0, y 0 is an isolated
curve.
point. The differential coefficient of the equation,
which is
3x 2a
dy

dx~

Vx-a

'

shows that the curve (Fig. 20) has two branches cut
ting the axis OX perpen
Fig. to
dicularly at the point x = a.
It shows also that when x=0,

is imaginary

which ex

presses the obvious fact that


a solitary point can form no
angle whatever with the axis

Generated on 2015-06-01 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

of x.
The second differential co
efficient
of the equation,
which is

d*y_
dx,~

(x

- V(x a)

3x

2a

a)'

shows that the curve has two symmetrical points of


and Q, corresponding to the abscissa
inflection,

4a

If

a curve has a series of conjugate points, this


series is called a branche pointillee, or a dotted
branch. For example, in the curve
y*

=x

sin' x,

every positive value of x will give two values for y;


but when x is negative, y becomes imaginary, ex

INFINITESIMAL CALCULUS.

101

cept when # is a multiple of k, in which case y 0.


There will then be an indefinite series of conjugate
points lying on the axis of x at equal distances, and
forming a dotted branch.

As a second example, take the curve


y = ax* Vx

(1

cos

x).

Generated on 2015-06-01 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Here for every positive value of x we have two val


ues of y, and consequently two points, except when
cos x = 1, in which case the two points reduce to
one.
Hence the curve, owing to the periodic co
incidence of such points, will consist of a series of
loops not unlike the links of a chain, having for a
diametral curve the parabola y = ax*.
But for all negative values of x the ordinate is
i maginary, except when cos x = 1 ; and in these
cases we have a series of isolated points situated on
the left branch of the diametral curve.
Order of Contact,

Osculation.

44. Let yr=<p(x) and y, =f(x) represent two


curves.
If x be an abscissa common to both curves,
and if this abscissa receive an
ment h, we shall have (No. 25)
/

in

dy

-l

d'y

infinitesimal incre

d'y h'

Now, if in these equations we have y=ylf the


curves will have a common point : if we have also

^i

they will have two common points, i.

e.,

INFINITESIMAL CALCULUS.

102

will
the

their contact being of

be tangent to each other,

first order.

If

we have also

f-

the

curves will have three common points, and their


contact will be more intimate than in the preceding
case: it will be a contact of the second order. If
,

we had also

d'i/ =

oJyx

gj1

would have a contact

d'y

d'i/,

,.

, . .

of the third,

the curves

fourth

order.

Generated on 2015-06-01 02:08 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

The order of the contact corresponds to the num


ber of successive differential coefficients which are
common to both lines.
Any line which at a given point of a curve has a
higher order of contact than any other line of its
own species, is called an Osculatrix.
Thus, an os
cillatory circle is a circle that has a higher order of
contact than any other circle.

45. If a curve y=f(x) has an osculatrix of the


ft* order, then the osculatrix, at the point (x', y') of

osculation, must have the same co ordinates x', y\


and n successive differential coefficients identical
with those of the curve ; in other terms, it must sat
isfy n + 1 conditions. Hence if a curve of a certain
species is to be made an osculatrix of the nth order,
its general equation must contain ft + arbitrary
constants, by the determination of which the
conditions may be satisfied.
For example, since the general equation of a
straight line is

y = ax-\and contains two constants,

b,

the straight line can be

INFINITESIMAL CALCULUS.
made an osculatrix
the function y = ax

103

of the first order.

b,

For when

and its derivative

= i

are assumed respectively equal to the functions y'

and

a
^7

of the given curve at the point

(#',

y'), we

have

and thus the equation of our straight line will be


come

dy'

dy'

that is,

y-y'=%
which is the well-known equation of a tangent at a
point (x', y') of the curve. This tangent is an os
Thus also, since the general equation of the circle

(x-ay + (y-p)' = l?
contains three ai-bitrary constants, the circle can be
For, when
made an osculatrix of the second order.
the function y drawn from this equation, and its
derivatives

'

dxi~~

dxx

d'y

/?'

x
dx~~ y
dy

^J^j

are assumed respectively equal to the functions


,

Generated on 2015-06-01 02:15 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

culatrix.

y\

of the given curve at the point (x\ y'), we

INFINITESIMAL CALCULUS.

104

have the equations

x'

-a _

dy'

dx"

y'-p

_ dy
dx"

'

from which we obtain

dx"
and finally

Generated on 2015-06-01 02:15 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Such is the expression of the radius of the osculatory circle. The co-ordinates of its centre are

dx'1

dx"

Measure of Curvature.

46. Curvature is the ratio between the change of


direction and the length over which such a change

takes place. Thus, if # be the angle that an ele


ment PQ of a curve (Fig. 21) makes with the axis
AX, and if d& TQT' be the change of direction
which takes place in passing from PQ to QJR, then,
denoting by ds the element on which the change

>

INFINITESIMAL CALCULUS.

105

takes place, the curvature of the curve, at that


element, will be expressd#

If

Fig..Sl

a circle be made to

pass (as is always pos


sible) through the three

points P, Q, R, its curva


ture will coincide with
that of the given curve
at the same points ; and
therefore, designating its
radius CQ by R, we shall

A~

have

ds Rd&,
and consequently the expression of the curvature

will

be

d#_

d&

R'

Now, if x and y be the co-ordinates of the point

P,

we have

+0\dxl
= Vdaf + di/ =

hence
d&

dx^/l +

d'y
dx

ds

(p

On the other hand

Generated on 2015-06-01 02:15 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

ds~ Rd&~

INFINITESIMAL CALCULUS.

106

and therefore

Mi)?
d>y

da?

that is, the radius of curvature is the radius of


the osculatory circle (No. 45); whilst the curvature
is the reciprocal of that radius.
When the curve bends upward, d& is positive,
and the value of R is positive ; but when the curve
bends downward, d& is negative, and then R is
negative.

We have hitherto assumed that y was


Remark.
the function of an independent variable x ; hence
dx was constant.
If both x and y were functions
of a third variable I, then both dx and dy would be
variable ; and then

Generated on 2015-06-01 02:15 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

a
d&

dy
dx

dxd'y

dyd'x
~

dx'

dxd'y

dyd'x

and consequently
(dx' +
R = dxd'y
dyd'x
dy*)*

or, introducing into this value the differential


the new independent variable,

/dx'

B=

\dl' +
dx d'y

*'

dy'\*
";
i

dy d'x

dl dt'~W d

of

INFINITESIMAL CALCULUS.

107

Examples. To find the radius of curvature of


the parabola.
The equation y*

= 2px

dy_p

dx

gives

&y

'

dx*

p%

'

y*

these values in the expression for R,

Substituting
we find

p*

y \ then -# = 13.888 . . .
To find the radius of curvature of the ellipse.

Assume

p=

The equation of the ellipse being a'y2


we have

dy ~
_
dx

b*x

'

a'y

d'y

&'

dx*

a'y'

-f&V
'

Generated on 2015-06-01 02:15 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

whence, by substitution and reduction, we find

(ay + b'x*)l

R=
When y =

0,

'- , which is
and

,../,.

a'b*

we have x*

= a\

minimum.

When x = 0, then y'V,

becomes =

a'
-

and

reduces to

which is a maximum.

To find the radius of curvature of an equilateral


hyperbola referred to its asymptotes.
The equation of the curve being xy = m', we
have

dy ~
_
dx

x'

(Py ~ 2y
dx"

x*

'

INFINITESIMAL CALCULUS.

108
whence,

by substitution

and reduction, we shall

find
~

2mV

'

involutes.

If

we draw osculatory circles to all the points


of a curve, and conceive that a line is drawn through
the centres of all these
circles, this latter line will
be the so-called evolute of
the given curve, and this

Generated on 2015-06-01 02:15 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

47.

given curve will then be


called its involute.
Thus
abed . . . (Pig. 22) is the
evolute of which ABCD
. . . is the involute.
The radii of the oscula
tory circles, being normal
to the given curve, have convergent directions, so
that each radius produced is intersected by its con
It is evident that the distance between
secutive.
two consecutive points of intersection constitutes
an element of the evolute, and that, therefore, the
radius of the osculatory circle is at the same time
normal to the involute, and tangent to the evolute.
The evolute may be constructed by drawing nor
mals to the curve, and then drawing a curve tan
gent to all of them ; the closer the normals, the
more accurate will be the construction.
The involute may be constructed by wrapping a
thread around the evolute, holding it tense, and
then unwrapping it. Every point of the thread will

INFINITESIMAL CALCULUS.

109

/?

/9.

describe a curve, and every one of these curves will


be an involute of the given evolute.
Since the evolute passes through the centres of
curvature of the given curve, the co-ordinates of
the evolute will be those of these centres, which we
have designated (No. 45) by a and
The equation
of the evolute will therefore be found by combining
the expressions for a and with the equation of the
given curve, so as to eliminate x, y, and their differ

Wy~'
dx'

parabola

shall have to make

dx

we shall find
'

= x + {f

whence

ft

p-v

whence


~.=
dx

and

---.

p\.

ft

and

values of x and
being substituted in the
equation of the curve, we shall find
These

or

Generated on 2015-06-01 02:15 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and

2px, we

be the

y'

the given curve

y,

da?

If

do?

W]f dx'P-y^

<W

+
^

';

+ dJt
^
dx' dy

y* .

_.
a-X

have (No. 45)

We

entials.

INFINITESIMAL CALCULUS.

110

which is the equation of the evolute of the para


bola.

the given curve were the ellipse cCif + 6V


= a'P, we would find for its evolute the equation

= (' -&*)*

If

(6/3)*

= (a"

a)*

the given curve were the hyperbola aly* Vx*


aW, we would find for its evolute the equation
(a

(&#)*

(a

a)*

If

6J)i

is

is

is

it

The length of an arc of the evolute is, in all cases,


equal to the difference between the two radii of
curvature by which
intercepted. For, as the
elements of the evolute are differentials of consecu
a sum of
tive radii, any finite arc of the evolute
measured, therefore, by
such differentials, and
the difference between the intercepting radii.

48. There is an interesting curve, the cycloid,


D

whose evolute
its involute.

C
a'

is t

^%

A/

Generated on 2015-06-01 02:15 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Fig. ts

another cycloid exactly equal to

Ill

INFINITESIMAL CALCULUS.

The cycloid is a curve generated by a point in the


circumference* of a circle, when the circle is rolled
along a straight line.
Let AB (Pig. 23) be the straight line, and A the
starting point. When the circle has reached the
point N, the generating point will have described
Take the origin of
the portion A M of the curve.
co-ordinates at A, and let AH = x, MH y, LN
= 2r. Then

- NH,

AH=AN

MN AJYbj
r& r sin #.

or, representing the arc

x=

On the other hand, y =


= r r cos #. Hence
cos

*=

sin

Generated on 2015-06-01 02:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

r&,

MH= PN = ON

cOS-1

- OP

J */2^?5

Substituting these values in the expression for x, we


have

X:

cos-1

7^

- V2ry - y\

which is the equation of the ascending portion of


the cycloid.
From this equation we obtain

dy

V2ry

y*

whence
da;'

2r
^

_ j'

and

dty
dx dx

_ _ 2r^/
y*

'

INFINITESIMAL CALCULUS'.

112

that is,

d*y

dx'

'

To find the equation of the evolute, these values


must be put in the general expres

of -# and

sions given above (No. 45)

T dx'
,

dx*

ft 2y,

Hence

y=

dy

da>

dx*

The result will be

x=

- 2 V2ry

S+

y V

cos-1-^+ V-2rft

- p.

If

AE=a = 7zr

if

now we produce DC to A' making CA' = DC


= 2r, and we then transfer the origin to A', the new
axes being A'X' and A'C, and the new co-ordinates
and ft', we shall have for any point, as M'
a'

a',

EM'

-ft = 2r-ft',

and substituting these values in the last equation,

Tcr-a! =

cos-1

(^-^-~-) +

But
cos"1

(-

Generated on 2015-06-01 02:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

evolute

ft

ft,

2rft /?.
and x in the above
Substituting these values of
equation of the cycloid, we shall obtain for the
a

T-=A)

=*- cos-1

V2rft'

ft".

(^A)'>

INFINITESIMAL CALCULUS.

113

and therefore
tit a' nr

cos

and consequently
a:'

cos

which is the equation of the evolute referred to the


new axes. This equation is of the same form, and
contains the same constant r, as the equation of the
involute. Hence the two curves are equal.
Envelopes.

Generated on 2015-06-01 02:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

49. We have

seen

how

e volutes

are formed

by

the intersections of consecutive normals drawn to


There are other
each point of their involutes.
questions whose solution depends upon the inter
section of lines varying in position, or in length, ac
cording to some given law. The simplest method
for answering such questions is the following.
Let F(x, y, a) = 0 be the equation of a curve, in
which a is an arbitrary constant, or, as it is called,
a parameter variable at will. For every particular
value of a we shall obtain a particular curve of
the class represented by this equation. If then a
and a-\-da be any two consecutive values of this
parameter, the two corresponding curves will be re
spectively represented by the equations

F(x, y,

a) 0, and

F(x,

y, a + da) = 0

and the co-ordinates of the point of intersection of


the two curves will be the values of x and y derived
from these equations.
But the second of these equations is equivalent to

INFINITESIMAL CALCULUS.

114

F(x,

y, a) + da

F(x,

y, a) =

0,

a alone being supposed to vary. And since we have


already the term F(x, y, a) = 0, we shall have also
da F(x, y, a) = 0 ; and the solution of the case will

wholly depend on these two equations.

The point of intersection given by these equations


changes its position for every different value given
to a, and will describe a continuous line when a
If, then, we eliminate
varies without interruption.
a between the same equations, the resulting equa
tion <p (x, y) = 0 will be the equation of the curve
formed by the successive intersection of all the
curves derived from the equation
(x, y, a) = 0,
The
when a is supposed to vary continuously.
curve <p (x, y) = 0 will touch all the curves repre
sented by the equation F(x, y, a)=0, and
therefore, called the envelope of these curves.
Examples.
To find the envelope of a series of
== a (x
parabolas whose equation
a).
Differentiating the given equation with regard to
a, we shall find a = \x. This value substituted in
the equation gives

\x\ or = \x.

y'

if

The envelope will, therefore, consist of two straight


lines.

projectile is discharged from the point


(Fig. 24) with a constant velocity, but at different
inclinations to the horizon. To find the envelope
2.

it

describes.
of the curves
We know by mechanics that

ax-(l + a')

the path of a pro


parabola represented by the equation

jectile

is

Generated on 2015-06-01 02:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

is

1.

is,

INFINITESIMAL CALCULUS.

115

in which x and y are the co-ordinates of the curve,

a the trigonometric tangent of the angle of projec


tion, and h = O V the height due to the velocity of

projection.

Differentiating

this equation with re-

2h
gard to a, we shall find a =

and this value sub-

CO

stituted in the equation will give


x*

= 4h (h

- y)

for the equation of the envelope.

A VX

This envelope is

having its focus at the point O


and its vertex at V, its parameter being = 47i.

Generated on 2015-06-01 02:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

a parabola

Given the area of a right-angled triangle, to


find the curve to which the hypotenuse is always a
tangent.
Let the given area be

2m% and let AM= x, MP


3.

= y (Fig.

25) be the co-or

dinates of a point P com


mon to the hypotenuse and
If AT a
to the curve.
be the parameter by the
variation of which the tri
angle changes its position,
we shall have

\\

Fig-

S5

VI

INFINITESIMAL CALCUL US.

116

AB =

(Xi

y = (a

x) tan

ATB = (a

x)

,
Q*

whence

a'y

4m'

(a x).

Differentiating this equation with regard to a, we

find a =

and this value

equation gives

substituted in our

xy = to' /

and therefore the curve is a hyperbola referred to


its asymptotes.

50. The above theory may also be applied to the


solution of problems in which the parameter de

In this
on the variation of two quantities.
case, the data of the problem must give rise to two
equations. Let us give an example.
pends

The straight line PQ


(Fig. 26) slides between
Generated on 2015-06-01 02:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

two

Find

Y'

rectangular axes.

the envelope, or

the locus

tions.

of its intersec

For

the solution, let


AX, A The the co-ordi
nate axes. Let AP = a,

AQ =

PQ =

Then
the equation of the line
b,

PQ will

c.

be

which, for greater


thus :

convenience,

may be written

INFINITESIMAL CALCULUS.

5+f".
and in which a and

are subject to the condition

&

a"

117

&"

= c\

(2)

Differentiating (1) and (2) with regard to a and


we obtain
a

whence, by eliminating da and

xyfi,
,

a'

Substituting,
obtain

b"

or

a-=-sf
b",

, y
b'x
and , = T
a'
b

in succession, these values in

(1), we

a',

and xc* = a'

b',

yc' =

a x^c',

whence

= ylcl.
b

in equation
Substituting these values of a and
(1), and reducing, we finally obtain the equation of
the envelope, which
is

xt

= c&.

+
yt

Elements of Arcs. Surfaces, and Volumes.

51. We have

if

ds = Vdx' + dy\

is

ds
an in
(No. 33) that
finitesimal element of the line referred to rectan
gular axes, we have
seen

Generated on 2015-06-01 02:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

tfy

we shall find

y (a2 + b') V and x (a" + 62) =

or

b,

INFINITESIMAL CALCULUS.

118

the values of dx&nd dy being drawn from the equa


tion of the line. This equation gives the differen

tial of any arc.

The differential of an
comprised between two
secutive ordinates of
curve. Thus, if AP
BQ (Pig. 27) are two
secutive ordinates, the

area is the infinitesimal area


con
the

and

con
area
APQB will be the differen
tial of the area comprised
between the curve and the

OX. Now APQB=


Hence, denoting APQB by dA,
ydx + \dydx.
axis

and neglecting the term dydx, we shall have

dA = ydx
expression for the differential of a plane area.
If the element PQ of the curve be made to revolve
about the axis OX, it will generate an infinitesimal
element of a conical surface.
Denote it by dS. Its
expression will be

Generated on 2015-06-01 02:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

as the

dS =

2zy + 2n (y + dy) \ Vdx' + dy',

or, reducing,

dS = 2nydx' + dy';
and this is the differential of a surface of revolu
tion.
If the area APQB be made to revolve about the
axis OX, it will generate an infinitesimal element of
a conical volume. Call it dF.
Then

dV=\

ny' + (y + dyy + ny(y + dy)

dx,

INFINITESIMAL CALCULUS.
or,

reducing,

119

dV = ny'dx;

and this is the differential

of a volume of revolution.

Differentials with Polar Co-ordinates.


52. Let
curve

PL

pf(f)

be the polar equation of any


(Fig. 28), the pole being at O, and OX

being the initial line.


At the point
we have p = OP, and

Generated on 2015-06-01 02:21 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

<p

POX.

Draw the tangent PB, the normal PC, and through


the pole O draw BQ perpendicular to the radius
vector. Then OB will be the subtangent, and OC
the subnormal, while OA perpendicular to the tan
gent will be the polar distance of the tangent.
Let ds = PQ be an infinitesimal element of the
curve.
Draw the radius vector OQ, and, with O as

INFINITESIMAL CALCULUS.

120

centre, the arc PR.


may be considered
at R. Hence

The infinitesimal triangle PQR


rectilinear, and right-angled

PR

PQ? =
+ IIQ\
But PQ = ds, PR = pd<p, RQ dp ; hence, substi
tuting and extracting the root,
ds = Vdp' +

p'd<p\

Such is the differential of the arc.


Let dA be the differential of the area swept over
by the radius vector. The infinitesimal area POQ
may be considered as the area of a rectilinear tri
angle having the base OQ and the altitude PR.
Hence

i OQ x PR =

dA =

Q>

+ dp) pdf,

or, as dp disappears by the side of p,

Generated on 2015-06-01 02:22 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dA =

p'df.

Such is the differential of the area.


Let V denote the angle OPB of the tangent with
the radius vector.
This angle and the angle RQP,
may be considered equal, as they differ only infinitesimally. Now

therefore
tan

Let OA =p.

V=

?p
dp

We have OA

- OP

sin

therefore

tt
p=PsmV,

or

tan

PVl + t&n'V

OP A ;

INFINITESIMAL CALCULUS.

121

Substituting the value of tan V, and reducing, we


have

+ p'df

VdP*

'

Such is the polar distance of the tangent.


Let S. T denote the subtangent.
We have

S. T OB=OP. tan V,
or

S.

T-p

dp

Let S . iV denote the subnormal.

OC: OP

OP

::

gives

OB
Let PCf=i? (Pig.
radius of
curvature of the line
PL at the point P.
Drawing OM perpen
dicular to PC pro
duced, and joining
the points O and C,
we have from the
triangle POO

Generated on 2015-06-01 02:22 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

29) be the

OC' = OP' +

The proportion

OB

_ f

p<d.l

dp

Fig.

S9

PC

-2 OP. PC cos OPC.

But OP cos OPC =

PM= OA=p.
Therefore

OC

= p' +

J? -2pfi.

dp

df

INFINITESIMAL CALCULUS.

122

Now, if we pass from P to Q, the side OC and the


radius R will remain unchanged, p and p alone
being subject to variation. Differentiating, then,
and considering O0 and R as constants, we iind
o

= 2pdp 2Rdp, whence

R = ^-

Spirals.
53. A spiral is a plane curve generated by a
point moving on a straight line while this straight
line is uniformly revolving about a fixed point or
pole. The portion of the curve generated during
one revolution is called a spire. The law according
to which the moving point advances along the re
volving line determines the nature of the spiral.
Denoting by p the radius vector, and by <p the angle
that the radius vector makes with the initial line,
and considering p as a function of , the general
equation of a spiral will be

=/(?).

The most remarkable spirals are the spiral of


Archimedes, the parabolic spiral, the hyperbolic
spiral, and the logarithmic spiral.

54. The spiral of Archimedes is generated by a


point moving uniformly along a straight line uni

/>'

-,

radius of a measuring
=2tt.
correspond to

tp'

as the

<p',

circle (Fig. 30), and let

whence

p'

AC=

Let us take

::

tp
:

p'

<p'

formly revolving. If p and p' be two radii vectores,


and <p and
the corresponding angles, by the law
of uniform variation we shall have
P

Generated on 2015-06-01 02:22 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

INFINITESIMAL CALCULUS.
Then p =
= a,

TW

^
p

or, making for greater simplicity

= af .

(1)

Generated on 2015-06-01 02:22 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Such is the equation of the


spiral of Archimedes. Its
form is the same as that of
the equation y = ax of a
straight line passing through
the origin of co-ordinates.
This spiral may be con
structed by dividing the cir
cumference into a number of
equal parts, and the radius
into the same number of
equal parts, and then taking
as many parts of r for the
radius vector as there are
corresponding parts taken
on the circumference.

Differentiating (1) we have


Substituting this
dp = ad<p
in the general formulas of
No. 52, we shall find
ds =

ad<p

V\

dA = fafy'df,

tanV=

-\-

t =w,r'
a

8 T=cup\

123

_
S.N=a.

Fig.

SO

INFINITESIMAL CALCULUS.

124

It will

remarked that the subtangent CB cor


responding to = 2n is equal to the circumference
of the measuring circle ; for a<p% becomes a(27r)'
be

Hence the area of the

triangle

= tit' = area of the circle.

55. The parabolic spiral (Fig.


f>%

cause its equation

31)

is so called be
(2)

2a<p

2ax of the

Fig. SI

^ will

Differentiating

revolution

give

and * CA =
2a X

(2), we have pdp

r,

2n,

be

and the
r*

equation

we shall have

x
-

then

becomes . Hence after

i'

and

comes

radius of the measuring circle

be the

Generated on 2015-06-01 02:22 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

is

of the same form as the equation


It may be con
parabola.
structed by first constructing
the parabola and then laying
off from A to D along the
circumference of
measur
ing circle any assumed ab
scissa x, and drawing from
the centre
towards D
the corresponding ordinate
Let
as a radius vector.

y"

(2s-)* = 2nr.
ABC is \r x 2r.r
=

or o = .

ad<p.

tuting this in the general formulas of No.

Substi
62, we

shall find

^ = *fyM!V}. dA = afdf,

tan

V=^=2f,

INFINITESIMAL CALCULUS.
2a<p

'

125

a(l+V)'

~~

(3

+ 4f> ") V'2ap(l+V)

'

56. The hyperbolic spiral is so called because its


equation
P<p=a

(3)

is of the same form as the equation xy m' of the


hyperbola referred to its centre and asymptotes.
From (3) we obtain

hence the radii vectores are inversely proportional


to the varying angles.

Differentiating (3), we have dp ^

Substi

Generated on 2015-06-01 02:22 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

tuting this in the general formulas of No. 52, we


shall find
ds =

^ VT+7',
a

dA =

?0,

V=

tan

- f,

S.T=-a, S.N=--2

<p

The equation of this spiral shows that when <p = 0,


then p = oo ; which means that the curve begins at
an infinite distance from the pole. On the other
hand, p cannot become = 0 unless <p becomes infin
ite ; and this shows that the curve can make an in
finite number of spires in approaching the pole.

INFINITESIMAL CALCULUS.

126

Again, when <p = 0, the curve and its tangent are


parallel to the initial line CX, and the equation of
the polar distance of the tangent gives p = a.
Erecting at C (Fig. 82) CD = a, perpendicular to
CX, and drawing BR parallel to CX, the line DH,
if prolonged to infinity, would be the tangent in
question.
The equation tan V <p suggests the following
manner of constructing the curve by points. With
S3

Cas centre and CD=za as radius, describe a circle.

Draw the indefinite line CJ/:ind the radius CT per


pendicular to it. The arc AB, which measures the
has
angle A CB =
On CT take
length a<p
CE a<p = AB, and draw EB then from draw
TM parallel to EB. The point M will be a point
of the curve. For, by our construction, we have

<p,

EC=a

tan

CBE =

a tan C'MT =

a<p

= AB,

CMT' =

CMT'

or tan

therefore

CMT

<p

tan

tan

CMT
;

But tan CMT=

tan

<p.

hence

Generated on 2015-06-01 02:22 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Fig.

= tan

V;

INFINITESIMAL CALCULUS.

127

hence the point M belongs to the curve.


Other
points may be determined by the same method.

57. The logarithmic spiral is represented

by

the equation
<p

in which

<p

= a log p,

(4)

is considered as a function of p.

Differentiating (4), we obtain

= -

d<p

Substi

tuting this in the general formulas of No.


shall find
ds = dp VT+-a% dA =
ap

Generated on 2015-06-01 02:22 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

We

Vl + a'

%~

>

tan

52,

a>

"

'

we

that the angle formed by the radius vec


tor and the tangent is constant, and that the trigo
nometric tangent of this angle is equal to the
modulus a of the logarithms used. In the Napier
ian system we have a I, and consequently
see

V = 45.
Since

- is

the Napierian logarithm of p, the arith

metical progression
a

will entail

a geometric
A

,6,6
1a

progression
2*

A*

,6 a

for the corresponding radii vectores.

Hence,

if

we

128

INFINITESIMAL CALCULUS.

describe a circle with the radius

a'

and divide its

e"),

circumference into equal parts, and if then from its


centre we draw through each point of division in
definite straight lines, and on these lines we take
such lengths as are required in order to form a
series of radii vectores in geometrical progression

Generated on 2015-06-01 02:22 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

we shall determine any num


(the ratio being
ber of points belonging to the spiral.

PART II.
INTEGRAL CALCULUS.
58. The object of the integral calculus is to ex
plain how to pass from given differentials to the
functions from which they may be derived by dif
The functions thus obtained are
ferentiation.
called integrals, and the operation by which they
are found is called integration.
To express that a function x is the integral of dx,
we write

Generated on 2015-06-01 02:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and the sign/ prefixed to the differential is called


the integral sign.

Integration and differentiation are inverse opera


tions. It follows, first, that, since the differential
of a polynomial is the algebraic sum of the differ
entials of its terms, the integral of a differential
polynomial must be the algebraic sum of the integ
rals of all its terms.
Secondly, since constant factors are not subject
to differentiation, they are not subject to integra
tion, and may, therefore, be written without the
sign ; thus, if a is a constant,

adx = a

I dx ax.

Thirdly, since a constant term disappears by dif


ferentiation, a constant must be added to the in
tegral obtained, to make it complete. We deter189

130

INFINITESIMAL CALCULUS.

mine the value of such a constant, in each particular


case, by making it agree with the conditions of the
problem proposed ; but, as it is capable of being

determined by various arbitrary conditions, it is


commonly considered as an arbitrary constant.
Before any value has been assigned to the con
stant the integral is said to be indefinite; when
the value of the constant has been determined so
as to satisfy a particular hypothesis, the integral
is said to be particular ; and when, moreover, a
definite value has been assigned to the variable,
the integral is said to be definite.

59. The following example will show how defin


ite integrals can be obtained.
Let Sx'dx be the
function to be integrated. We already know that
d (x3) = 'ix'dx. We have therefore

Generated on 2015-06-01 02:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and this is the incomplete

integral.

Add to it a

constant C ; then

and this is a complete integral

but it is still in

definite.

Assume now that the particular problem under


consideration requires the integral to be =0 when
x a, that is, that our integral begins at x=.a.
In this case we shall have
0= a' + C,
and consequently C= a'.

Hence we shall write

INFINITESIMAL CALCULUS.

131

the letter a placed at the bottom of the sign


in
dicating the inferior limit of the integral, viz., the
place where it begins ; and the letter x at the top
of the same sign indicating the. superior limit
where the integral ends.
But, as x is still vari
able, this is a particular integral. If our problem
requires the integral to end at x b, we shall have
at last

and this will be the definite integral.


In general, if we have

Generated on 2015-06-01 02:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

assigning to the integral the limits required by the


conditions of the case, say a and b, the definite in
tegral will be expressed by

The present treatise will contain three sections.


In the first we shall explain the various methods of
integration ; in the second we shall apply the in
tegral calculus to many questions of Geometry ;
and in the third we shall solve by it various
mechanical problems.

132

INFINITESIMAL CALCULUS.

SECTION L

VARIOUS METHODS FOR FINDING IN


TEGRALS.

Generated on 2015-06-01 02:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

60. Before we proceed further, let the student


be warned that, although in the differential cal
culus we can always pass, by a uniform method,
from a given function to its differential, in the in
tegral calculus we have no general method for pass
ing in all cases from a given differential to its
integral ; and for this reason we are not unfrequently compelled to make use of devices of
various kinds, the choice and employment of which
is not always exempt from difficulties.
We shall start from the fundamental formulas
which regard the simplest cases of integration ;
after which we shall explain various processes used
for the reduction of other less simple cases to forms
integrable by the same fundamental formulas.

Integration of Elementary Forms.


61. The fundamental formulas of the integral
calculus are found without labor by simply revers
ing the corresponding formulas obtained in the
differential calculus.
Thus,

since the expression

ax+i

has ax"dx for

INFINITESIMAL CALCULUS.

133

its differential, we immediately see that the integral


of axndx is r

n-\-l

/**>

Therefore

% +

0.

(1)

This formula is general, whether n

be

positive or

The only case


negative, integer or fractional.

to be excepted is n =
1, because in this case
+
Cl&n '

r
n+1

becomes

7; ,

0'

and gives
no differential.
6

On the other hand, ax-ldx is not the differential


of a power, but of a logarithmic function ; for

ax-ldx = a

ad log x ; and thus

fax^dx =

I = a log x +

C.

(2)

The expression ax log adx is the differential of

Generated on 2015-06-01 02:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

a" ; hence

&
yfa'dx = ,-^log a +

(3)

'

By reversing

the differential trigonometric for


mulas of No. 17, we shall find the following :

/cos xdx sin x +

/sin

#dfo

C.

(4)

= cos x + C.

(5)

cos' a;
VfJ^.=i2inx+c.

(6)

'

= _cota; + e.
y/l^_
sin* #

(7)

INFINITESIMAL CALCULI'S.

134

^sin
^cos

xdx = vers x +
a^.r

covers x

fsmxdx = secx
/
cos1 a;
i =
/cos^rafa;
sin x

C.

cosec

(8)

+ C

(9)

c
~

x +1 6.

(v

'

(11)
v

In like manner, if we reverse the differential


formulas of No. 18, we shall find the following :

Generated on 2015-06-01 02:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

f^r=f

-*-'y

+ c-

(13)

/w^=vers~^+a

^f=i

In

these

latter

cosec~1

formulas,

y+

make

(16)

y=

<19)

for

then

INFINITESIMAL CALCULUS.
dy

=^

135

dx; and by substitution and reduction

we

shall obtain the following:

/iFFW-a^T+a
C

dX

f^

Generated on 2015-06-01 02:30 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dx

J vers-1

-+

=-\ covers-1 - +
_1
+ C.
sec-1

f~=t =-\ cosec-1


By these formulas

C.

^+a

(24)

C.

(25)

(26)

(27)

great number of differentials

can be integrated. We give here a few examples


for the exercise of the student.
EXAMPLES.
1.

dy xdx,

2.

,
adx
dy

3.

dy = ax'dx,

yz=.\x*.
,

y=
y=

2a Vx.

INFINITESIMAL CALCULUS.

4.

dy = 2x^dx,

5.

dy =

y = fr*,
y = ^x-^,

\x-idx,

n-n

6.

di/=x-n dx,

y=

7.

dy = 2x-l dx,

y = log

8.

dy = 1 2x

dx,

y=

x'- log x.

9.

*
dy
cos
y = atln

axfcr,

y
"=

y=

sin 2x.

?/=

2 tan

dy = 2

11.

dy =

12.

<Zt/

13.

dy = x

cos (2a;)

(|)

CO,'

dz,

Ida;,

y=

d#

16.

dy=0^{,

dy =

(x>).

log a

vers (a#)-

covers

x
^

= isin-12ir.

= cos-1

(a;').

= itan-'(^.

- -^j=^i

15.

J/

i/f-3^'
,

14 tfy =

cos

,, sin x

y= -a

= sin (a#) (te,

10.

Generated on 2015-06-01 02:29 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

136

INFINITESIMAL CALCULUS.
dec

1Tdy=m^,

2/

=*

2xdx

19-

2<&r

y=

=ia

sin

A^^

dft

22.

=-3-,
dy
* = ,
1 + sin a;

2S.dy =

cos

~.
1

cl

/#a

V5\

/a;

y = *os"
*V

a;

21-

cofl2-

2dx

-to.

^=~4+^'

nA

vers

137

(2 cos

...
(sin

^
tan-1

'

tan 'a1.
2/=:,
log a

In

Generated on 2015-06-01 02:29 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

the above examples we have omitted the arbi


trary constant, as it had no bearing on the work
proposed.

62. When the fundamental formulas above given


are used for the successive integration of differ
entials of the second, third, or any higher order, a
constant is to be added at each integration.
Thus, if the expression

d'v

^ = ax is to be

integrat

first multiply both its members by dx, which


gives us
ed, we

= axdx, or rather d

= axdx.

138

INFINITESIMAL CALCULUS.

Integrating this, and adding a constant, we have


as?
d'y _
~
+C2
da?

Multiplying this first integral by dx, we have

and integrating, and adding a new constant,

dx

Multiplying this second integral by dx, integrating,


and adding a third constant, we shall at last obtain

is,

is a general rule that every complete integral


must contain as many arbitrary constants as there
have been successive integrations performed for
of course, to be de
obtaining it. Each constant
termined by taking the integral between the limits
required by the particular conditions of the problem
to be solved.

63. When the proposed differential

is

Reduction of Differentials to an Elementary


Form.
not in the

it

form required for integration,


may often be re
duced to a proper form by some simple algebraic
Thus the differential
process.
(a'

Generated on 2015-06-01 02:29 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

It

xjdx

INFINITESIMAL CALCUL

will

139

US.

reduced to an integrable form by perform


ing the operation indicated by the exponent of the
We shall then have
parenthesis.
{a,

+ x,ydx =
[a'

2a

V + O dx = a\z + ^

Thus also the differential


CL

SO

will

\x\

be properly

As
a

'

'

4- ax 4- a*

=
ax

(x'

reduced by division.

a' +

x/J

dx,

be

hence

_=-_-_-'*- log (a-s).


(ojj-J)x)dx
a'+x7
we might

'

we had a differential of the form

into its parts by writing

adx
a?

a' +

bxdx
a"

+ ar

ic

whence

/(a
In like

+ bx) dx

Generated on 2015-06-01 02:29 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

If

manner the differential

xdx
V2ax x'

will

be

reduced to two integrable terms by simply

IXFISITESIXA L CALCUL US.

140

adding and subtracting the quantity a in its nume


rator. We obtain thus

adx

xdx

V2ax

V2ax x"

(a x)dx
V2ax x7

_
x7

'

whence

/xdx
j=

V2ax x'

= a vers

.
1

x
-

V2ax

x7.

The differential
dx Vaf a'

x
may be reduced to an integrable form by multiply
ing its numerator and its denominator by Vx' a'.
We thus obtain
dx

vV

a'

x'-a'

Generated on 2015-06-01 02:29 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

x Vx*

- tf

,
dx

xdx
Vx*

a'dx
a'

- a'

x Vaf

'

whence

dx

Vx* a" a sec-1

64. Sometimes a differential is reduced to an in


tegrable form by the introduction

variable.

Let
dp = x2

If

i a -4- xdx.

we make a + x = z, whence dx

we obtain

dp =

(z

of an auxiliary

af zi

dz,

dz, x, = (z a)',

INFINITESIMAL CALCULUS.

141

which, being developed, gives

dy =

(z*

2az%

y=

\A

- \a + \ a'zi,

and

or, replacing

zby a-\-x,

dz,

and reducing,
9/7* i

4./7

+ a'zl)

Again, let it be required to integrate the differen

tial
,

dx

whence

\x + z)dx
dx
z

(dx + dz),

d
_~ dx + dz _
~ (x + z)

x-\-z

x-\-z

Therefore

that

rdix + z)

is,

/dx

Generated on 2015-06-01 02:29 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Adding
Assume x' a' = z*, whence xdx = zdz.
zdx to both members of this last equation, and
factoring, we have

formula of great use in calculus.


If we had

dx

INFIXITESIMAL CALCULUS.

142
we might

reduce

the denominator

to

the form

V (a-\- x)' a*, and the numerator dx to the form


d(a + x) ; whence by formula (28),

dx

fit

= l0g ia

+ X+ V2ax +

*'}'

(29)

which is another very useful formula.

Integration by Parts.
65. We have found (No.

that the differential


of the product uv of two variables is
12)

d (uv) = udv + vdu ;


and therefore

Generated on 2015-06-01 02:29 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

whence
(30)

By this formula

we are enabled to integrate the


expression udv whenever we can integrate the ex
pression vdti. This method of integration is called

integration by parts.
As a first example, let

dy = x'dx Va'

x'.

To integrate this expression, make


x' = u, and xdx Ya* x* = dv;
then

du = 2xdx, and

= $

Y (a'

a?)'.

INFINITESIMAL CALCULUS.

143

Substituting these values in formula (30), we have


x'dx Vd'

- xl =

VW=3? +f2jT SW=*)\

and, by integrating the last term of this equation,

y=

fx'dx Va'-af=

- 1 V(di-x'y- ft V

(a'

- xj.

As a second example, let


dy

x log xdx.

Make log x u, and xdx = dv.


x*
;

-g

/'

and

a?

log x

-J -grxdx

==

^'o; log

a*&c

log

a;

-g

is,

that

third example, let


dy

Make

du~

of
=

dx Yd1
dx = dv.

=
;

Generated on 2015-06-01 02:30 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

x log xdx =

As

and formula (30) will give

^a

x\
Then

a?,

and formula (30) will give

^/
On the other hand,

tt

Then du =

Va*

- x'

and

INFINITESIMAL CALCULUS.

144

dx Val x* =
a' x*

_ r

a'dx

afdx

hence, adding together these two equations,

J" dx Va'

x' = x Vc?

/.

-x' -f

J'

-f====

>

whence

a*

,x-

dx Va' x' = g Va' x' + g sin-1

Integration of national Fractions.

Generated on 2015-06-01 02:30 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

66. When a fraction whose terms are rational


and entire is to be integrated, the variable must
have a lower degree in the numerator than in the
denominator. When the degree is too high, it is
reduced by division. The quotient will then con
sist of an entire rational part, plus a remainder
divided by the denominator. Thus

x'dx
As there will

no difficulty in integrating the


entire part of the quotient, we have only to show
here how any remaining fractional part can be in
be

tegrated.

There are four cases. The first regards a fraction


whose denominator has real unequal roots, and can
be decomposed into unequal binomial factors of
The second regards a fraction
the first degree.
whose denominator has some real equal roots, and

INFINITESIMAL CALCULUS.

145

contains some equal binomial factors of the first


degree. The third regards a fraction whose denom
inator has unequal imaginary roots. The fourth
regards a fraction whose denominator has some
equal imaginary roots.

All

67. First case:

real and unequal.

roots

Let
ri

(2a; 5)

dx

y^(x-1)(x-2)(x -3)'

Assume

(x-l)(x-2)(x

3)

+x

"+P

'

A,

B, and C are quantities to be determined.


Clearing of fractions, we have

where

Generated on 2015-06-01 02:30 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

2x-5 =
A(x~2)(x-3)+B(x-l)(x-3)+C(x-l)

(x-2).

Performing the operations indicated, and equating


the coefficients of the like powers of x, we might
obtain three equations for determining A, B, and C.
But, as our equation must be true for all values of
x, we can adopt a simpler process.
Make

x=l

then 3=2J,

and therefore

Make x=2; then 1 = B, and therefore


Make a;=3

then

= 20, and therefore

(7

A=

1.

f.

Putting these values in the assumed equation, and


multiplying both its members by dx, we shall have
(2x 5) dx

(x-1) (x-2) (x-3)

_~ _ 3
2

dx

x-1

dx

+ x^2 +

dx

^3

'

INFINITESIMAL CALCULUS.

146

and integrating,

_
-2)
log (*-!)+
,

(x

2)

or finally

Vx

2)

(a?

3)

I<>g

log

(x

dx

(x

5)

(a?

1)

(x -

(2x

3),

-3

EXERCISES.
1

2/

log|(.r-l)'(.T

rt

a-4-a:

= 2^l0g^=

= log

'

,,

dy =

j'JLf

2-

<fr

3.

^tf^.
tZa;

Or*

2)sS.

ar).

(x

The existence of equal roots in the denominator


entails the existence of some factors of the form
a)n, n being the number of the equal roots, or
Now, in the decomposition of
Of the equal factors.
the given fraction, we must obtain as many partial
fractions as there are factors in the denominator.
If we were to assume, as we have done in the first

that

these

Ji

fractions

would

coa

therefore,

to obviate this, we shall assume the frac

tions

single fraction

lesce into

evident

fractions of the form

is

case,

it

Generated on 2015-06-01 02:30 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

68. Second case: Some roots real and equal.

and

INFINITESIMAL CALCULUS.

'

'
a)n' (x a)n-v (x a)n-%

(#

147

'

"

x~^~a

Let the given differential be

+ x) dx

(x-i)

Assume

(x

(a;

+B

+x=A

(x

x2

we find

z-l
0

By dealing of fractions,

2)'

ic-g

'

1)

{x

- 1)
2)' (x

(a>

'

x^+x _

(x

dy'~(x-2y

(a?

2)

- 2)\

J. =

2,

1,

This equation being true for all values of x, we make


x and we find C=
then we make x =
and
6

1.

2)

dx

x)

dx

(a-

(x

(x' +

6dx
-l)"-(aj 2)"

2dx

a;-2^a?

log

(a

log

2)

ar-2

Or

~
0

(a:

2)'

rite

1)

a;)

-+ -

(a?2

(x

Hence

-1),

1)

= logvi

and
#

^(flJ-l)'

2/~

2x +
2/r-f
x>

a;

dx
(x'~
[x* -2)dx
2)

EXERCISES.

Generated on 2015-06-01 02:30 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

0,

and replacing A
lastly we make # =?

=
by their values, we find B
Patting these values in our equation, and multi
plying by dx, we have

find
and

x-1

<fcr

(a?

2)

INFINITESIMAL CALCULUS.

148

x-2

69. Tliirdcase: Unequal imaginary roots.

We

In this third

and x = a

1,

x = a -\-b

b*

(x

know from Algebra that imaginary roots are found


only in pairs, and that for each pair we must have
factor of the second degree which, when placed
equal to zero, will give the imaginary roots. Thus

= gives the two imaginary roots,


a)'
V

1.

is

and to determine

and

- a)' +

(x

A + Bx

by the ordinary process.

Let, for instance,

xdx
(*

+ l)(a?+l)
+

1)

l)(x' +

Bx
x' +

_A

Assume

(x

Cxi

C.

(A+B)x + A

x = Bx,

Clearing of fractions, we find

Generated on 2015-06-01 02:30 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

it

instead of assuming
partial
fraction for each imaginary factor of the first de
more convenient to assume for each pair
gree,
of such factors (that is, for each factor of the second
degree) a fraction of the form
case,

INFINITESIMAL CALCULUS.

149

of the like powers of x,

Equating the coefficients


we find

B+C = Q, A + B = 1, A + C=0,
whence

i, = i,

C=-h

Substituting these values in the assumed equation,


and multiplying by dx, we have

xdx

(x +

xdx

dx

dx

l) (x*+i)~ 2\x' + i^ x*+i

x+

xdx
(x

+ l)

(x*

l)-

tan-1 x

(tan-1

i log
a;

1)

log (x +

+ log -^qr^)

tan-

1.

1)'

-^i-1)

log

(a?

= log

tan-1^

4)

(3

1)

2)

(x'

(x

dx

1)

{x

(x

dp

1)

EXKRCISES.

2.

Generated on 2015-06-01 02:30 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

or

1),

and

ISFIXITLSIMAL

150

70. Fourth case

CALCULUS.

Equal Imaginary

roots.

In

this case, if n be the number of the equal roots, we


assume n fractions of the form

A + Bx
'
a)' + 6' J"
[ (x

-C+Dx
af +

[ (as

ft']"-1

'

"

" (xM+Nx0'
-af +

of the second and

thus combining the methods


third case. Let us have

Assume

_x'
(af

+ iy

A + Bx
(x' + l)"

'
+ V+_Dx
af + 1

Clearing of fractions, we obtain


af =

A + Bx +

+ Daf + C+ fix,

Caf

Generated on 2015-06-01 02:30 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

which gives

A=0,

-1,

C'

0,

i) = l.

Substituting these values in the assumed equa


tion and multiplying by dx, we have

x'dx
(x7

xdx

+ iy

(af

1)'

xdx

siF+i

'

whence

or

x'dx
(a'

+ !)'

2'

iil+*l0S(a;, + 1)'

(rf +

1)

'

+ log

V + 1.

INFINITESIMAL CALCULUS.

151

EXERCISE.

dy =

(x

2)

2 (a?

dx

2)

^2 tan-1

+ log Vg+2
a; 1
=

We conclude that all differentials which

are ra
tional fractions can be integrated, provided the
factors of the denominator can be discovered.
Their integrals will depend on one or more of the

forms

Integration of Binomial Differentials.


71. Every binomial differential can

be

reduced

Generated on 2015-06-02 01:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

to the form

in which m, n, r, and

whole numbers, and n


positive. The reduction to this form is made by
different means, according to each particular case,
especially by the use of an auxiliary variable.
The factor A may be omitted during the integra
tion, as it represents a constant.
There are three cases in which the integration of
a binomial of the above form can be easily per
formed.
s are

r
The first case is when -e is a whole number.

1NFINITESIMAL

152

CALCULUS.

Then the binomial is rational, and can be integrated


by some of the processes already explained.
The second case is when is a whole number.

For, in this

case,

if

we assume

a + bxH =

or z

s*,

= (a -f

bx*y,

we shall have
i

whence, differentiating, and dividing by m,

and therefore

Generated on 2015-06-02 01:22 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

(a +ba?y>
which

x-1 dx=

expression

dz,

{^^Y

is rational

(31)

and integrable,

if

is a whole number.

The third case is when the sum


number.

For, in this
a -f bx" =

we

shall have

case,

z* of1,

if

I- n

is a whole

we assume

or z =

INFINITESIMAL CALCULUS.

153

of" an (z" b) n,

and

whence, differentiating, and dividing by m,


of1'1

dx

a*

(2*

&)-n-

z"-1

dz ;

and therefore
(a

r
bxnj' of1-1 dx =

-s.

a*~' (z>

dz,

(32)

which expression is rational and integrable when

h -T is a

771

whole number.

Generated on 2015-06-02 01:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Integration by Successive Reduction.

12. When the proposed binomial does not fall


under one of the three cases just mentioned, and
when its integration can be made to depend on
that of a simpler expression, we can, by a suc
cessive repetition of the process, make it finally de
pend on an expression integrable by some funda
mental formula.
This is the process by which
irrational differentials are integrated when they do
not allow of immediate and direct integration. It
is called the process of successive reduction.

Let

(a

+ baffzm-1 dx

be an irrational binomial differential, p being a


fractional exponent, and m and n any whole num
bers. The difficulty of proceeding to its integration
arises from the value of the exponents, which may
In the former
be either too great or too small.

INFINITESIMAL CALCL'L US.

154

cause the integration to depend on


lower exponents ; in the latter we must try to make
it depend on higher ones. Let us come to particu
case we must

lars.

First

case: To lower the exponent m

variable.

of

the

We have identically
(a

+ bxny

of1-1

tlx = x

(a

-f bxny

x "-1 dx.

Making
= u, and (a + bx*Y x*~l dx = dv,

xm~n

we shall have

Jo*)"

xm-1

dx

-ST,

and

^ nb(p+l)JX ^

ax-

consequently

Xp

a?"-1

A *

n6(jp+l)

(fe = -7-7 i-^p6 (p+

1)

Transposing the last term to the lirst member, and


reducing, we get

&(/i

x
l)J*
X'**"
!)

nb(p

Generated on 2015-06-02 01:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

or, making, for greater simplicity, a

(et

and, integrating by formula (30),

axa

(m-n)
nb(p +

r*

l)J

INFINITESIMAL CALCULUS.

155

and dividing by the coefficient of the first member,

y~ X" xm-1dx =

Xp+1 xm-n

r?

-\

_
.
of1-"-1
/r Xp

\
rr -n)
a(m

,_.

dx.

(33)

of Xp af"-1 dx is
made to depend on that of X" xm-n-1 dx, where the
exponent of the variable is diminished by n units
at each application.
The formula fails when
np + m 0 ; but then the integration can be made
by formula (32).
Second case : To lower the exponent p of five bi
nomial.

By this formula

the integral

We have identically
(a

+ bxn)p = a(a +

bxn (a

bxn

X" \

bat1)"-1

+ bx*y-\

or

Generated on 2015-06-02 01:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

X' = aX"~1

wlience

fxp xm-1 dx=ajxp-1 xm-1 dx-\-b J"xp

x*"

dx.

Now, the last term of this equation can be reduced


by formula (33) ; for, changing m into m + n, and
p into p 1, and multiplying by b, that formula
gives
b

fxp~l

x"+n-1

dx=z

Xp xm
np + m
Substituting

am

np + m - fxp-1x-1

dx.

this for the last term of the above

INFINITESIMAL CALCULUS.

156

equation, uniting the similar terms, and reducing,


,
we obtain

fxpxm-'dx=

(34)'
np-\-mJ
np-\-m +7^fx>-lxm-1dx.

the integral of Xp x-1 dx is


made to depend on that of Xp-l xm-l dx, where the
exponent of the binomial is diminished by unity
at each application. When np + m = 0, this for
mula fails ; but then the integration can be made

By this formula

by formula (32), as already remarked.


Third case: To increase the exponent of the
variable.
When m is negative we tnay need to diminish
it arithmetically or increase it algebraically.
To
'
do this we proceed as follows.
Reversing formula
(33), we obtain

fxp of1-n-1

dx =

Xp^
a (m
n)

Generated on 2015-06-02 01:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

bjnp + rn)
a(m

n)

This equation, by changing m into

J/>*-<&.
m

become

PX^'x

+ n, will

Xp x-n-1dx =
am

+ n)rXPx.m^dx
^(np-m
am
1

By this formula

the integral of

that of Xp

Xp

x-m-1

dx is

x-m+n-1 dx, where the


made to depend on
exponent of the variable is increased by n units at
each application.
Fourth case : To increase the exponent of the

binomial.

IXFINITESIMAL CALCULUS.

157

we proceed as follows.

it,

When p is negative, and we need to increase

Reversing formula (34), we

obtain

1>

into

This equation, by changing

'

X" a--1 dx.


fx*' af-1 dx= - ~P-anp + VBf
anp
will

become

m-\-nnp
an (p1)

of1

an (p

1)

I-"'1

or-1 dx.

(36)

the integral of X-p of1-1 dx


made to depend on that of X-"+1 xm-1 dx, where the
increased by unity
exponent of the binomial
at each application.

is

is

By this formula

V2rx

x'

dx (2rx

is

- x')-i=(2r-x)-i a9-* dx.

Now, representing (2rx)-i by


with formula (33), we have
a = 2r,
whence

b=-l,

n = l,p =

X-i,

and comparing
m =

x3

i,

can be made to depend on that of


like expression,
in which the exponent
diminished by unity.
We have identically

Generated on 2015-06-02 01:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

73. The integration of the expression

i;

ISFINITESIMA L

158

(np + m) =

p-}-l

a (m n) =

q,

CA LCUL US.

i,

m n

=q

(2q 1),

i.

Substituting these values in (33), we find

afldx

If

of-1 dx

V2rx-x7

(gg

"

a?

1)

V2rx

x*-1

x*

Jf

V2rx

that

is,

X-l&-ldx =

V2rx

In like

vers

manner the integration of the expression


x9

dx

\'2rx-\-x'
is

formula ought to be changed.

is

can be made to depend on that of a like expression


in which the exponent
diminished by unit.y.
For the only difference between this case and the
now positive instead of
that
preceding one
which
and therefore
(np + m) =
negative,
shows that the signs of the second member of the

is

Generated on 2015-06-02 01:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dx

positive, by a successive appli


cation of this formula, we ultimately arrive at the
form
be entire and

Accordingly

INFINITESIMAL CALCULUS.

/a?

dx
~
V2rx + x'

x-1

_ r (2g

V2rx-\-x'
?

1)

159

dx

x*-1

'

'

V2rx+x*

Formulas (33), (Si), (35), (36), (37), (38) are called


formulas of reduction, and have a wide range of
application.
EXAMPLES.
1.

To integrate
afdx

dy =
Here we have

X= ra

- x\ a r',

By formula

Generated on 2015-06-02 01:16 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

fx

(33) we have,
^

x' dx=

1,

2,

for this

, m

3.

case,

- zj/ ^ dx,

that is,

/x"

dx

r'

dx
a

or

/3?
2.

cLo?/

)C

7*

4 Ob

sin-1 -

To integrate

dp = afdx Vr*

x\

Here we have

X= r' x\

= r\

1,

n=

2,

p=

3.

INFISITESIMAL CALCULUS.

160

Formula

(33) gives,

fx\

dx =

x*

that is,

fx* dx

in this case,

Vt^x* =

^-

X* dx,

- xV^-^' + r' Jdx Vl^:

To integrate the last term of this equation we


use formula (34), in which we shall make

X= i*

x\

a r', n = 2,

p = ,

m = 1.

Substituting

this in the

VF~-^x,

sin-1

preceding equation, we

have

To integrate
r7.r

d,y.

7i

2,

1,

X=

Here we have

m =

2.

sin-'

1.

3.

-f

V(f=F?+1^ VF=tf

dx Yr' x' =

2,

Jaf

Generated on 2015-06-02 01:17 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dxVY~^x* =

12

Xidx = *i? + ?fx-idx,

that

is,

We obtain

INFINITESIMAL CALCULUS.
By formula

161

(36) we have

that is,

or

tan

xfl5.

Integration of some Trinomial Differentials.


74. A trinomial differential of the form
xm

dx

can be made rational in terms of an auxiliary vari


When p
able, when p and m are whole numbers.
is even the expression is already rational ; when_p
is odd, then, makings = 2n-\- 1, the expression be
comes
(a

bx

z*f (a-\-bx

a?"

dx,

x*

Va + bx

x,

bx

z*

from which we shall get


2zx,

-f-

bx
x'.
in which the only irrational part is Va
positive we obtain a rational form by
When x'
assuming

is

Generated on 2015-06-02 01:17 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

(a-\~bx x*f

162

INFINITESIMAL CALCULUS.

and

y/a

"

+ bx-\-x' =

6-

'

Tims the given expression will become rational in


terms of z, and therefore integrable. After the in
tegration it only remains to substitute for z its value
x + Va -\- bx + x\

and

i (x h) (x k) = (x

z,

a?

are the roots of the equation


a?

bx a =

0.

where

bx

ia+

7i)

When x' is negative let us assume

= (x

Kf

*>

(k

x)

h)

bx x*

(x

We then have

h)

x =

(x

or
z'.

and

^)'

'

2(k-h)zdz
(1

aX~

Thus the given expression will become rational


and integrable in terms of 2. After the integration
will only remain to substitute for its value
z

it

Generated on 2015-06-02 01:17 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Hence

Ik

- .t

INFINITESIMAL CALCULUS.

163

EXERCISES.

Integration by Series.
75. When an expression Xdx, in which X is a
function of x, is to be integrated, it is often con
venient and useful to develop A' into a series by
any of the known methods, and then to integrate
each term separately. This is called integration
If the series obtained be convergent for
by series.
any particular value of x, we shall obtain the ap
proximate value of the integral for that particular
value of x.
Thus, given

dx

we may, either by the binomial formula or by mere

division, obtain
^

+x

X*

+ X2

x'-\-

xdx -f x*dx

x'dx + x'dx

Whence
dx

\+x

= dx

and

/dx
1+^

a?

a?

xK

^-2+3-4

which is the expression for log

(1

+
-J-

Generated on 2015-06-02 01:17 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

x'

5-

x).

ISFIXITESIMAL

164

Again, given

dy =

CALCULUS.

dx
+ x'

'

we may, by simple division, obtain the series

jJL-^i
whence

-r

\-\-x

= dx

x'dx +' x'dx

+1 x'dx

x*dx

and

/dx

x'

x7

x1

x*

which is the expression for tan-1 x.


In a similar manner, if we have
dx

Generated on 2015-06-02 01:17 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

we may obtain,

by the binomial formula,

(!-*')-* =
+
whence,

+ 2.4 +

2.4.6

2.4.6.8

multiplying by dx, and integrating,

find

/ax
W=7
dx

+ 2.3 +

=
3x^

2.4.5

&5af_ +
+ 2.4.6.7

3.5.7a;'

2.4.6.8.9

which is the expression for sin-1 x.

'

we

INFINITESIMAL CALCULUS.

165

76. Let us have to integrate the general expres


sion

dy = Xdx,

X is a function

where
we have

y = Xx

of x.

-JxdX

Integrating by parts,

-J
dX_ f&
J

dX

Xx

xdx.

dx

This last term, again integrated by parts, will give

rdX
dx

xdx_^_
"

dx

'

d'X
dx

This last term will give, by the same method of in


tegration,

Generated on 2015-06-02 01:17 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

rx' ' d'X_


~
2

dx

rx'
2

(PX_a^
dx*

~2.3

&X_
da?

r
2.3

'

d?X
dx'

'

and continuing to integrate in the same manner, we


shall find

J^Xdx

dx '-2 + da?

'

2.3

da?

'

2.3.4

'

'

This elegant

formula, by which the integral is


obtained through successive differentiations, was
discovered by John Bernouilli, and bears his name.
.

Integration of Trigonometric Expressions.


77. Trigonometric expressions can be reduced to
integrable forms by suitable transformations based
on the correlation of trigonometric lines.

INFINITESIMAL CALCULUS.

166

1.

-doc

dy
* = sin
a;

Given

we have

from trigonometry

sin x =

sin

cos

&r == 2 tan

\x

cos* &c

whence

dx

dx
sin x

tan $x cos" $x

tan
and therefore

V=f
2.

tfcg

cos" %x

a;

<Z

(tan frg)
tan $a;

~ = log(taniaO.

Given

cos

(39)

a; '

Generated on 2015-06-02 01:22 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

we have cos x = sin (90 x). Hence, substituting


90 x for x in (39), we shall have

y=f hv = ~ log [tan (4 _ **)]3.

y
^x
*
tan x

Given

we have

dx
tan

a;

cos xdx

sin x

(Z

(sin x) ,
'
sin a;

and therefore
dx

y=/ta*

= log(sina:)-

(40)

'

INFINITESIMAL CALCULUS.

167

we have

dx
cot x

_
~

sin xdx
cos

d (cos x)

cos

,
'

whence

5.

Given

dx

dy:
we have

sin x cos x = sin

2#.

Substituting, and integrating,


sin x
/<&c

cos

6.

sin

..

(tan
#).
v
'

Given

dy =
Generated on 2015-06-02 01:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

a?

.
jr- = log
= //* -2dx

2a;

it will

be convenient

sin

cosn axftr,

to transform this differential

into an algebraic expression by assuming sin x =. s,


whence cos x = Vl z% and
dz

dx--

Vl

-z'

'

Thus we shall obtain the differential in the form


dy =

(1

z>)

3m

ds,

which can always be integrated when m-\-n is a


whole and even number (No. 71).
If m-\-n is a whole but uneven number, then
the integration may be made by formula (33) or

INFINITESIMAL CALCULUS.

168

by which the exponents will be gradually


reduced till we reach some simple and elementary
(34),

form.

(33)

will

At i- zry*\m

dz =

Formula

give us

(l-W*1

.
1

m-j-n

_ tfr
Ai
m-\-nJ
1n

dz,

or
#<fa:

sin"-1 x cos"+1 #
m
w

if

sin"

we obtain

sin"-1 # cos

a;

Formula

y"(i

(34)

- 2')^

/*. _ .

/sir^^xdx.

will give us

2m <fe

+n

^m +

nj

rj

atffcr

0,

from which,

/*. __,'
_
xcosTxdx;
/sin"
m-\-nj
m 1

az>

or

x cos*

sinm+1 # cos"-1

Js'mm

.n

Generated on 2015-06-02 01:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

-f-

cos"

H,

sin"

- .

INFINITESIMAL CALCULUS.

if

from which,

7"

cosn

m = 0, we obtain

,
sin x
xdx =

cosB_1

.
'

n~l
n

169

__,*

/cosn

xdx.

Integration of Logarithmic Differentials.


dy =

78. Let
be proposed

and af-1 dx

(log x)n dx

a?n-1

for integration.

= dv,

we obtain
&

<#&

Assuming (log x)n u,

dx,' and =
m

Hence, by formula (30),

ix"11 (log x)n dx =


1
.

(log

xf

fx1 (log xf-1 dx.

(43)

The formula fails


log # by 1 at each application.
for m 0 ; but then the integration is obvious.
If we reverse (43), we obtain
(log x)n-l dx =
XT

if

we change n

dx

into n, we shall have

/of1-1
(log x)n

a?
(log

1)

xm-l (log x)n dx,

xf-l

n-ljr
m

of-1 dx
(log x)-1

....
>

and,

(log a;)"

X**-1

'

(to

and by this formula we are enabled to reduce


the exponent of the denominator by
at each ap
1

Generated on 2015-06-02 01:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

By this formula we can reduce the exponent of

170

JNFINITESIMAL CALCULUS.

plication.
when to =

The
0

formula fails when n = 1


and n 1, the integral becomes

/dx
#

log x

d.logx

-ilog

rc

dx
To integrate dy

[^77^.?

= 2,

whence

And,

as

e*

a;

= e*, dx =

+ 2 + 3. +

but

= ,log
n (log
v
6 rr).
1

it suffices to make log x


e*

and dy =

dz ;
.

<r^ ~H

tUe

t^z
inte~

gration is obvious.

Integration of Exponential Differentials.


dy xmaxdx
79. Let .
proposed for integration.

be

dx = dv,

ax

we obtain du mx*

and formula (30) will give


Generated on 2015-06-02 01:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Assuming

yCxm a" dx = log a - log a

fxm

At each application of this formula


x will be lowered by 1, till we reach

= u,
ax

tfrr,

xM

log a

ax dx.

'

;
(45)
v

the exponent of
the simple form

dx.

When to is negative, by reversing the formula


we

first obtain

y\.

, ,

x-1 ax

dx =

xm ax

to

log
a
S
to

r _ a:

/af

dx.

then, replacing to 1 by to, and reducing,

dx

/axxm

_~

(to

ax
1) xm-1

log a raxd,x

m^lJ

xm-1

'

)}

INFINITESIMAL CALCULUS.

171

By this formula

the exponent of the denomina


tor is lowered by 1 at each application.
But the
formula fails when m = 1 ; and in this case the in
ax

tegral of

dx

is obtained by changing ax into its

sc

development

l+(loga):r+(loga)'|V(logfl)'|^+

X being a function

of x.

have

JfXa*

dX.
^- - J/12
log a

dx =

log a

Generated on 2015-06-02 01:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dx,

dX' = X"

ax

X'ax
dX_ (log

log a

r(loga" aj

and so on.

a)'

dX,
aJL

dX" = X'"

dx,

we obtain

Jf

Integrating by parts, we

we take the successive differentials of


place

dX X'

dy = Xaxdx,

80. Let

If

-_ X"a*

(log a)'

J
_ r
J
r

Hence, by successive

ax

(log a)'
ax

(log a)'

X,

and

dx,

dX,
UjSl '
dX

substitutions, we

shall find

a.

Xax dx =
(

log a

rcf

TJ

X'

(log ay +

dX(n)

(loga)n+1-

X"
(log a)'

' '

'

X(n>

(log

)n+1
.

^47)

INFINITESIMAL CALCULUS.

If X is

...

such a function that one of its differential


coefficients X', X",
is constant, then the
next differential coefficient will be =0, and the

dXw will
r^i

integral will be exact.

case the

have

dy = (a?

X= a?

If)

vanish.

In

Thus,

we

if

such

-rj

/a"

dx,

ex

- h\ dX= 2xdx, X' =

X"

2,

2x, dX' = 2dx,

dX" = o. On the other hand, log = 1.


Substituting in (47), we find
then

= (7(x,-h'-2x + 2).

Integration of Total Differentials of the First


81. The total differential of

Order.

function

in which

-=-

ce

dxdx

du ,

du

dydy>

dx and 3- dy
partial differen
a are the *

dy
tials of the function. We have also

seen

whence

follows that

Ht) "Q
dx
dy

'

(No. 25,
Scholium) that the existence of an exact total dif
ferential of such
function entails the existence of
the relation

it

Generated on 2015-06-02 01:24 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

du =

is, as we have seen (No. 19),

u=f(x, y)

differential of the form

du = Mdx+ Ifdy

INFINITESIMAL CALCULUS.

will

be an exact

are such as

173

M and N

total differential when

will satisfy

the condition

^=^^;

for, in such a case, we evidently have M^j^

If

and

dy

the condition is satisfied we may immediately


integrate the partial differential Mdx and write

u=

Y exhibiting

Mdx+

Y,

a function of y which is to be deter

mined so as to satisfy the condition

= JY.

Ac

we now differentiate our integral with


respect to y, and, dividing by dy, we have

cordingly,

Generated on 2015-06-02 01:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

du_dfMdx

dy

dy

dY=N
dy

'

whence

_ d/Mdx

~
dY=
dy

and

Tand finally

u=

dy

J ('-V->*
C

/ 'Mdx V

Ir Mdx+ Ir lN~ dfMdx\


dy'~)

V'

(48)

INFINITESIMAL CALCULUS.

174

EXAMPLE.

Let
du = (2axy

3bx'y) dx

+ (ax'

bo?)

dy.

Here we have
2axy 3bx*y,

N=ax*

bo?,

whence

dx

dy

Hence our total differential is exact and immedi


ately integrable. Integrating the partial differen
tial .3/dx, we have
(2axy

3bcr\'/)

T= ax'y

bafy-\- T.

this integral with respect to y, we

Differentiating
hnd

f
dy

= ax'-bx' + d*=N,
dy

whence

|- = N-

(ax*

- bx') =

0,

and

Y-C.

bx'y

C.

u = ax'y

-f-

And therefore

du=x^^,
x'
+

O.
C.

u=

dy,

(2y-^)

u x'y*
X-

2afydy,

y*

du=d^

du = 'Sx'y'dx

y'

2.

1.

EXERCISES.

3.

Generated on 2015-06-02 01:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dx +

= tM- xV+C.

ISFIXITESIMAL

CALCULUS.

175

Integration of the Equation Mdx + Ndy=0.

82. The implicit function

where

ax

and

-i-

differential (No. 21)

ay

(x, y)

=0

has for its

are the differential coefficients

the function taken with respect to x and y.


by iV,

the differential

the form

Mtfx

iV^y =

Now, this equation, whenever we have

will

Repre

0.

will take

by 31, and

senting

of

differential (No. 81), and may be


integrated by formula (48) but, as we have here
the integral will be n C.
du =
When by any transformation the equation can be
placed under the form

Xdx

Ydy=0,

X being a function

of x alone, and Fa function of


alone, the integral can be found by taking the
Thus
sum of the integrals of the two terms.

Ydy =

C.

Xdx

Ydx

Xdy =

0,

When the equation can be placed under the form

or under the form

XYdx + X'Y'dy =

Q,

Generated on 2015-06-02 01:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

0,

be an exact

INFINITESIMAL CALCULUS.

176

the variables can be separated by division.

dx

dy

'

Y'

Thus

-grdx+y- dy = 0,

and when the variables are thus separated the in


Hence the separation
tegration becomes possible.
of the variables has been one of the main objects of
study on the part of mathematicians.
EXAMPLES.
1.

Given ydx

-^y =

0,

xdy
log

Divide by xy.

0.

Then

x- logy = C= log c;

and therefore

x
-=
y
Generated on 2015-06-02 01:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

2.

Given xy'dx

c,

or x cy.

+dy = 0.

Divide by

if

x'y =

2(Cy+l).

Then

whence

- ^,
____

Given (1
a?) ydx
vide by x'y. Then
3.

or

a?

(1

- y')

x*dyz=0.

,
1-__ dy-0,

Di

INFINITESIMAL CALCULUS.

177

whence
~

~ X ~

log V +

or

I=

equation Mdx + Ndy = 0 is


homogeneous with regard to the variables, that is,
when the sum of the exponents of the variables is
the same in Mas in N, the variables can be sepa
rated by the aid of an auxiliary variable. Let

83.

When

the

x'dy

y (x + y) dx

This equation being

0.

we

homogeneous,

y = zx, and

assume

therefore dy = zdx-\-xdz.
Substitut
ing these values in the equation, we have

x'zdx-j-x'dz
Generated on 2015-06-02 01:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and, dividing by

+ x'z) dx =

(z'x'

0,

x\

xdz z'dx =

and

0,'

=
-r
Z'
X

:'

whence

log x =

In like

1-

C= C
'

;'

and y .

C log x

manner, the equation

x'
4-xt/
!

dy
''

=0
ydx
u

being homogeneous, we assume

zx, dy

= zdx + xdz,

and we find, by substitution and reduction,

INFINITESIMAL CALCULUS.

178

(1

z)

dz + 2z'dx = 0, whence

and

log^=|-ilog^ = ^-log //l+C.


-f-

= dx Vx'

axfy

y'

Again, the equation

we assume

= zx, dp

zdx

xdz,

-f-

being homogeneous,

and we find, by substitution and reduction,


^o;

rZ2

s;

whence

,v

= dx

c,

j)

yi

(I

vi + 2') = log

an integral which, freed from transcendentals and


radicals, reduces to

x' = 2C1y + Cl\

-f

84. When Mdx

Ndy =

is

being an arbitrary constant.


0

xydy

#')

it
C.

a?

(.!

becomes an exact differential


2x, and gives the integral

if

?/) dx

(1

it

is

it

ferential,
possible to reduce
ferential by an integrating factor.

not an exact dif


to an exact dif

Thus
0

Cl

Generated on 2015-06-02 01:23 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

log x = log

(*

Accordingly

be multiplied

by

INFINITESIMAL CALCULUS.

179

The multiplier 2x is termed an integrating factor.


The same equation might have been made an exact
differential

by

the

factor

integrating

'

g~jr~*\

whence

5 + r^ =

0'

and

log* + ilog(l + y*) =

6V

result identical with the preceding one, though


under a different form.
The expression Mdx + Ndy can be written as
follows :

+*(!&"- Ny)

\ (Mx + Ny)

*jf)

Since

log

y,

dy

, dx
xv
log
and
,y
s

(a)

-ity) d. log

Ny

happens to be identically =0,

Mdx + Ndy
Tr
at
Me ivy

,
=i log .

a;
;

if

Now,
Mx-\we shall have

d. log a>y+(Jfo

(Jfe + Ny)

we may write also

that is.

jx

an integrating factor.

is

is

and because the second member of this equation


an exact differentia], the first member
also one

is

Generated on 2015-06-02 01:34 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dy- d.

'

dx

CALCULUS.

LXF1N1TES1MAL

180

on the contrary, Mx
identically 0, we shall have

If,

Mdx + Ndy

Ny

happens to be

is

is,

where, because the second member is an exact dif


ferential, the first member is also one ; that

j^rj^Vty

then an integrating factor.

0,

if

is

iden
neither Mx-\-Ny nor Mx Ny
equation (a) divided by Mx-\-Ny
tically =
But,

gives

,1 MxNy

Mx+Ny =idAeW+2MxTX

lg

y'

Mdx4-Ndy

(b)

case, then, the

y
as

is

is

it

or

a homo0,

of the degree

first member of

perfect differential, and the inte

grating
8 factor will be
6

Mx + Ny
Mx Ny

will also be

(b)

that

if

y,

x
-

geneous function of x and

In such

function of log
is,

Ny

function of

Mx

is

-ry- rsM
-f-

if

generally

differential,

<p

nor Mx Ny
neither
identically =
equation (a) divided by Mx Ny

is

-f

0,

Again,

if

,
x
+
id-lZy
=2~Mx^Ny-d]0SXy
,

'Mx-Ny

Mx+Ny ,

Mdx-\-Ndy

gives
1

Generated on 2015-06-02 01:35 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and the last term of this equation will be an exact

(c)

INFINITESIMAL CALCULUS.

181

and the first term of the second member

of this
I ' _ZV?y

ATcc
will be an exact differential if , , '
equation
H
Mx

_y

Ny

is a function of log xy, or generally of xy, that is,


if it is a homogeneous function of x and y of the
second degree, as is <p (xy). In such a case the first
member of (c) will also be an exact differential, the
integrating factor being

Mx\_ Ny

To give an example of this process of integration,


let
(x' + y*) dx xydy = 0.
Here we have

M= x' + y%

and JY xy.

Consequently

Mx

- Ny = x' + 2xy',

Mx + Ny = x\

Generated on 2015-06-02 01:35 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and

Mx Ny _ x' + 2xy' ~
Mx + JSTyx*

(yV '

Hence, by formula (b),


(a?

-f- y")

_~

x'

i<Uogay +
i(l+2 ~)d.\og
Integrating this expression, we have

Jloga^ + ilog5+y|i tf.log


But, making

zi we

^1K^

|=

0.

182

INFINITESIMAL CALCULUS.

r/.log

2
2a

2za

And, therefore, the whole integral will reduce

U/

'

to

As the proposed differential was homogeneous,


we might have integrated it by assuming p = zx,
according to the method explained above (No. 83).
We would thus have found dx = zxdz, whence
,
,
dx
= zdz,
and

Generated on 2015-06-02 01:35 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

a result identical with the preceding one, as was to


be expected.

Integration of other Differential Equations.


85. We have seen in the preceding pages that
our success in the integration of a differential fre
quently depends on our ability to give it a simpler
form. Though we have pointed out many such
cases, many others would have to be examined, if
we had to give an adequate idea of the resources
But, as this is an elemen
offered by the Calculus.
tary work, we must content ourselves with giving a
few examples of a certain number of other pro
cesses frequently adopted by analysts for the trans
formation of differentials not directly integrable.
I. Let

dp

aydx =f(x) dx.

(1)

INFINITESIMAL CALCULUS.

If

183

we make y = uv, we have dy udv

+ vdu,

udv

-J-

by substitution,
vdu auvdx

=/(x)

dx.

As one of the two quantities u and


trarily assumed, take

and,

can be arbi

vdu = auvdx, or du = audx.

(2)

Then the above equation will be reduced to


(x) dx.

udv

Now, (2) gives log u = ax, or u =


substituted in (3), gives
,

eax

(3)

and this,

fix) dx

II. Let

rf(x)dx

~\

the integral required.

a)

and

assume the two equa

|-^

0,

fx-Ky

z,

To simplify this expression,


tions

an auxiliary variable, whilst


and
to be determined.
Differentiating
first of equations (2), we have

is

where
constants

- = = k'z
dx
dx

'

^M-k
dx*

-1

Generated on 2015-06-02 01:35 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

which

is

ff(x)dx

Whence

(2)

are
the

IXFIXITKSJMAL CALCULUS.

184
or, since z

ky, substituting and reducing,

which, compared with (1) gives k-\-k' a, and


kk' b.
Now, from the second of equations (2) we have

= k'dx,' and
z

z = e*'x+c, or z C&'*:'

hence, substituting in the first of equations (2), and


multiplying by Ux,

dy

kydx =

Ce*

dx,

Generated on 2015-06-02 01:35 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

an equation of the same form as the one which we


We
liave integrated in the preceding example.
shall have, therefore,

But

fCe**dx

a,

M.IV

7W

Ce<*

-*

therefore

and C, being two arbitrary constants.


This same integral could be easily obtained by
another method which deserves special notice,
owing to its simplicity and the range of its applica
tion. In the equation
Ct

INFINITESIMAL CALCULUS.

185

(1)

make y = C.e**.

Then, differentiating, we have

and therefore, substituting in (1),

-4b

*,

_ a-

0.

Va' 46
2

'

Va'
2

1._a +
*~

ak

'

the

rejecting

k*

factor, 6Y,t*z,
This auxiliary equation gives
or,

-f-

-ak + b)=0,

(k'

2,

make a

Cj*x.

3,

(1), we

= C^'x

If, in

we shall find

The complete integral of differential equation of the second order must not
only satisfy that equation, but also contain two arbitrary constants (No. 62). The
= Cxek'*, and
= Cae*, satisfy equation (1) for they give
particular integrals

(?,*(*'- a*

0,

= <);

single arbitrary constant.


y

but each contains only

and

b)

- ak'

(*''

C,<?*'-

The integral

= C,e"'*+ C,e**,

Hence

= Cje*''

+Cfe"

is

our complete integral.

0.

- ak

b)

it

(*

(C,'

is

the sum of these particular integrals, contains two arbitrary constants, and
equally satisfies equation (1) for
gives

which

Generated on 2015-06-02 01:35 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

b,

are connected by the re


and these two values of
as in the preceding
lations
-\-k' = a, and M' =
solution. We have, therefore, two particular in
= Olekx, and
= C,ekx, which, added totegrals
*
gether, will give us for the complete integral

INFINITESIMAL CALCULUS.

186

k' =

1, &

= 2; and

y=Cle"+C,t*.

III.

integral

the complete

will

be

Let

differentiating, and substi

Assuming ?/C'emx,
tuting in (1), we find

Cemx (m*

+ 13) =

4rn

0.

Solving the auxiliary equation m' 4m +


we obtain
m'

= 2 + 'dV

-1,

m"

= 2 3V

13

0,

l;

and the complete integral will be expressed by

This integral, by referring to De Moivre's formulas


(No. 26), may take the form

Ce*x

(cos 3a;

+V

e**
1

+
(C + C.) ete cos

3a;

"'^

sin

3a;)

(cos 3a;

C.e2*

+ (C

C.e**

C.)

e~**

vri

- V~\

^"^1

sin

e2"

3a;

-f-

or, replacing C + C, and (C C,) V


bitrary constants,

y = Ae^ cos

Be2x

sin

sin 3a;)

3a;,

by new ar

3a;.

the form of the integral, when the auxiliary


This
equation has two unequal imaginary roots.

is

Generated on 2015-06-02 01:52 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

y=Ceu.

IV. Let

dx'

dx'

dx^y

K>

INFINITESIMAL CALCULUS.

187

m' to

0,

to3

-j-

Assuming y = Ct"'x, and differentiating, we find the


auxiliary equation

1.

is

O,

-j-

1
,

1,

1,

1,

to, =
to, =
the roots of which are m, =
we shall have in the
Owing to the root to, =
to, =
integral the term Ce~x. The roots to, =
will give the terms
+ C,^- But these two terms
coalesce into the single term (C,
Q) e*, where
equivalent to a single arbitrary constant,

to,

by

whereas the complete integral of (1) must contain


To remedy the defi
three arbitrary constants.
ciency, let us begin by supposing to, to differ from

small quantity h.

Then

and, developing by the exponential

formula (No.

must become less than any assign


in order to verify the equation
to, = to yet the product CJi may remain finite,
we assume
greater than any assignable quantity.
be a new arbitrary constant.
Let, then, CJi =
The complete integral will be

and, in our case, wi th to,

(C,

Csx),

to, to,

1,

Cem'x

1,

C,

Ct

if

Now, though
able quantity

Generated on 2015-06-02 01:52 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

24),

= Ce-*+e*(Cl+C,x).

V. Let
(1)

auxiliary variables.
sumed equations,

dx =

zef d&,

e*

dz

we have

(f

dy

xz e?z,
and being two
Differentiating the two as

and

Assume x =

c*,

IXFIXITESIJIA L CALCUL US.

188

d&,

whence

d&

d^y_
~e
dx'

dz\

_2y

W ~rd#)~af'
((Vz

'

(ty_dz_
~

dx

and suppressing the com

and (2) becomes


Cem

(m*

+m

Then

Cem'.

0.

gives to,

4/
,

But

and

e*

2.

The auxiliary equation m' -\-m 2

Hence
m2 =

Generated on 2015-06-02 01:52 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

To integrate this, assume

= x.

Therefore, finally,

1,

therefore, substituting,
mon factor e~',

_ 2z _

2ez

2)

2y

But

INFINITESIMAL CALCULUS.

189

VI. Let

To get rid of the term mx, we shall assume

y = z + Px+Q,

P and

Q being constants to be

(2)

suitably determined.

Prom (2) we obtain

dy

_ dz

dx~

dx

p'

d%y

dx'

_ d'z

dx'

'

These values substituted in (1) give

^a<gaP

+ bz + bPx +

Now, let bP m, and


be reduced to
Generated on 2015-06-02 01:53 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

d'z

bQ=TaP.
dz

bQ = n,x. (3)

Then (3) will

and (2) will become

y^+jST-gr-

(5)

Equation (4) will be easily integrated by the


method which we have followed in examples II
and III. When z has been found, the integral
of (1) will be known by equation (5).

VII. Let

(1)

INFINITESIMA L CAL CUL US.

190

To get rid of the term

e*, we

shall assume

y = e*(z+P%

being a constant to
From (2) we obtain

be

(2)

suitably

determined.

Substituting these values in (1), and cancelling the


common factor ex, we have

<g-2(*-l) |g+(*_l)'s + P(*-l)' = l.


Hence,

if

(3)

we make

P(*-l)'

l, orP

(T^a,

Generated on 2015-06-02 01:53 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

we shall have

y=e.(2+__L_).

and

now make z emx, the integration of the last


differential equation can be made to depend on the
auxiliary equation

If we

m'

2m (k

which gives m.k


of example IV,

1)

+ (k

=m*.

1)'

= 0,

Hence,

by the rule

INFINITESIMAL CALCULUS.

191

and

y e'.e*-1** (C+

Ctx)

(k-iy(k-iy

In this
d'z

j-j

(1)

1,

example,

and

if k

equation (3) reduces to

1,

P disappears.

In this

case,

equations

and (2) will become

S-2i+^
Hence,

and

integrating the equation

d'z

1,

we shall

have for this case,

Generated on 2015-06-02 01:53 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

|+

Cfc

C;,

and y =

e*(^ +

Cx+c).

VIII. Let

^^rey
Assume y = z +

cos ax.

where h is a constant to

be determined in view of a future


We shall have

dy

dz

d'y
dx

d*z

a*

dar

3^3
dx
dx

(1)

sin ax,
T
h

simplification.

INFINITESIMAL CALCULUS.

192

Thus (1), by substitution, will become


d>z

j-z

dx

a'
,
r- cos ax 4h

n'

nz +
.

7- cos ckt

'

cos ax,

or
d'z ,
-y-i
dx +'
Take h = n'

n'

= cos ax

a%

and the equation will redu

7i

To integrate this, assume


find, by differentiation,
m'em*

a'

,
nz

+ raV" =

from wl

emx,

or i' +

0,

and therefore, m = n V 1.
method followed in example III,

tf =

0,
1

J.

cos w#

-\-

is,

Hence,

sin

wa;,

and

and

being arbitrary constants.

cos nx 4-

sin nx

+.

cos ax
-5

;
,

Integration by Elimination of Differenti

,,-M^

86. Tliere are differentials of which the inu6iai

of

can be found without direct integration.


They are
those whose form admits of
ready elimination
the differentials themselves.
We shall show by
few examples how this method can be utilized.
a

Generated on 2015-06-02 01:53 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

that

cos ax.

INFINITESIMAL CALCULUS.

193

I. Let

Making

^=p,

dypdx, the equation

whence

becomes

a;

Differentiating

=a+

(2),

we have

7ip*.

(2)

dx = mdp

whence

or dy = mpdp

-f Znp^dp,

3np'dp,

and

y = bmp*+frip< +

C.

(3)

If we

can eliminate
between (2) and (3), we shall
have the expression of the complete integral.

n. Let

Generated on 2015-06-02 01:53 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

-)+
Making

^=p,

(2)"

the equation becomes

y=

mp*

+ 2np\

(2)

Differentiating (2), we have

dy = 2mpdp + 6np*dp =pdx ;


whence

dx = 2mdp + Qnpdp,
and therefore

x = 27rep +

37ip'

From this equation we have

C.

(3)

INFINITESIMAL CALCULUS.

194

This value of p being placed in (2), we shall have


the complete integral of the proposed equation.
III. Let

+=
Dividing by

^ =p,

'

we obtain

y = xp +

(2)

Differentiating, dividing by dx, and reducing, we


have

m\

This equation must be satisfied either by

bya;- =

0.

If

0,

then

0,

or

and (2) be

Generated on 2015-06-02 01:53 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

comes
I
/7t
y=Cx
+ ^.
c
n

If

0,

then

- 1

p=

(4)

which

value sub

stituted in (2) will give


y*

= 4mx.

(5)

any arbitrary constant, that is,


solution ; * whereas equation (4) gives

an integral without

a singular
*

By lingular solutions we mean such results

as nrc not contained in, or cannot


whatever particular value be assigned to the
arbitrary constant.
They can only be obtained by differentiating the complete in
tegrals with respect ,to the constant alone, considering it as a variable parameter, as
we have done above (Nos. 49, 50), and substituting its value, in terms of the vari
be obtained from, the complete integrals,

INFINITESIMAL CALCULUS.

195

the complete integral, and represents the tangents


by whose intersections the parabola y' = 4mx is
generated.

IV. Let

We write, as usual,

x-\-py

ap'

=0

(2)

(2) we have

-^

From

0.

ydp -\-pdy 2apdp =

dx

-f-

whence
(3)

Substituting

this

in (3) and reducing, we find

-x

xdp

Vl +p'

,
a

adp

Vl+p*'

of the integrals.

^/

alone, we find

ables,, in the expression

V1 +p% will take the

is

the integral of which

= apdp,

Thus, differentiating

value which will reduce (4) to

(4)

with regard to

-{p*

pdx Vl

(1

+p')

which, being divided by


form

= 4mx.

It not our purpose to discuss the relations existing between the complete in
We will simply state that, owing to such rela
tegrals and the singular solutions.
tions, as interpreted by Geometry, the latter are often called envelopes of the former.
This subject has been very deeply treated in the excellent work of Mr. George Boole
is

Generated on 2015-06-02 01:53 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dx

on Differential

Equations.

MFISITESIXAL

196

Lt!

= a log

CALCULUS.

p+

1/1

C,

or

x=

(c+

a log CP +

(4)

Now, from (2) we have

_
This value of

>,

*V + 4aa?

placed in (4), will give the com

plete integral.

V. Let

ydx
Dividing by

da:,

xdy = n Vdx* +
and making

dy2.

^.=P,

(1)
as usual, we

shall have
Generated on 2015-06-02 01:53 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

y=px+n Vl+p\

(2)

whence

dy =pdx +

axZp

~=2

>

or, since dy=pdx,

dp(w+-^L=)

Q.

(3)

=
This equation must be satisfied either by dp
by x A

^ = 0.

If

dp = 0, then

(2) becomes

y = Ox +

7i

Vl +

6>",

0,

or

C: and

INFINITESIMAL CALCULUS.

197

an integral without any arbitrary constant, repre


This solution is the
senting a singular solution.
locus of the intersections of all the straight lines
obtained by varying the constant C in the complete

integral.
Double Integrals.

87. When an infinitesimal

rectangular
sion is the infinitely small rectangle of which dx
and dy are the sides ; and thus dA = dxdy.
To find the area A, we must integrate between
proper limits with respect to each variable in suc
This double integration is indicated by
cession.
writing the sign of integration twice before the
quantity to be integrated. Thus we have the
double integral

dydx.
of these integrations can always be per
formed, so that we may have either

Then,

if

area be

xdy +

O.

or

C,

ydx

-f-

One

the equation of the line that bounds the


=/(%), or x f(y), we shall have either

Generated on 2015-06-02 01:53 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

area is referred to
co-ordinates, its most general expres

198

INFINITESIMAL CALCULUS.

f(x)dx+C,

or

<p(y)dy

C,

that is, a function of a single variable, to be in


tegrated within proper limits, as usual.
In like manner, an infinitesimal volume may be
expressed by the product of its infinitesimal di
mensions dx, dy, dz. Hence
d V dxdydz, and V

J^J*f dxdydz,

a triple integral, to be taken between proper limits


with respect to each variable in succession. The
If it
order of the integrations is wholly arbitrary.
is found convenient to integrate first with regard to
z, the integral, between z = 0 and z z, will be

dxdy.z,

Generated on 2015-06-02 01:53 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and if z = f(x, y) be the equation of the surface


that bounds the volume, then we write

and we proceed to the double integration, as we


Some examples of this
have just explained.
method of integration will be given in the following
section.

INFINITESIMAL CALCULUS.

SECTION

199

II.

APPLICATION OF INTEGRAL CALCULUS


TO GEOMETRY.

88. Our object in this section is to find definite


values for the length of some curves, for their areas,

of revolution, and volumes of revolution.


The operation by which we determine the length of
a "curve, is called rectification ; no curve, however,
is properly said to be rectified, unless its length be
expressed by a finite number of algebraic terms, to
the exclusion of transcendental s. The operation by
which we determine areas and surfaces, is called
quadrature ; and the operation by which volumes
are determined, is called cubature.

Generated on 2015-06-02 01:53 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

surfaces

Rectification of Curves.
89. The arc
axes

will

of a curve referred to rectangular

be found by integrating between proper

limits the formula (No.

51),

ds = Vdx'+dy*.

Example I. Let the curve be a parabola y' = 2px.


Thenete =

*.
P

formula, we have

Placing this value of dx in the

INFINITESIMAL CALCULUS.

200

and integrating by formulas (34) and (28),

or, taking the integral between y 0 and y = y,

*-_ y Vy'+p* +2p


2^

II. Let

Example

(No. 48), dx =

+
l0g (y

+^'\

'

Then

the curve be a cycloid.

y&y^

V2ry-y*

Placing this value of

dx in our formula, and reducing, we have

Hence
s

= V2r.2 V2r

y+C;

and taking the integral from y =

Generated on 2015-06-02 01:53 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Make y =

(- 2

2r;

^27-

0toy

= y,

-iy + 2 V^r).

then * ;= 4r = one-half of the cy-

cloidal curve.

Example

III.

Let the curve be a circle.

equation of + y* = r' gives

Vr'

x*

stituting this value in our formula, we get


ds dx

Make r =

1.

\f

Then

s=J dx(l-x')~$ =

j =

Its
Sub-

INFINITESIMAL CALCULUS.

201

and integrating from x 0 to x = x,

1.3.5

1.3
x +
S~x + 2.dX , +, 2.4.5 ^, +
2.4.6.7
1

90. Let

tan-1

ential will be

Its differ

be a circular arc.

a;

dx

or developing the second member of this expression


ds = dx

x' + x> x* + )
This, integrated from x = 0 to x = x, will give
s

= tan

Generated on 2015-06-02 01:54 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

When x =

(1

x=x

i}

then s =

1,

+ 5
.

"y

Hence this series may

serve to calculate the value of jr. However, as it


converges but slowly for = 1, the calculation has
been facilitated by the following artifice :
By trigonometry we have

tan a + tan 5
tan (a 4- b)'
1
tan a tan 6

a = tan-1 m,

Make
and

(a

6)

'

= tan-1 n,

tan-1 z = tan-1 m

tan-1 n.

Then
3

= tan ( +

But tan a =

m,

&)

and tan

m-\-n
z = - !
1
mn

tan
a+
tan & !
1
tan a tan o
b

n ; and therefore

whence =

l-\-mz

INFINITESIMAL CALCULUS.

202

Now, assuming z =

m = , we find ?i = $ .

1,

There

fore
tan-1

Assuming

= tan-4 +

tan-1

= t, w = i, we find

f.

T\.

Therefore

tan-^^tan-H + tairSV
Assuming z =

^,m

tan-1

Assuming z =
fore

= h we find = A-

^=

tan-1

ft,m = h

tan-1

A=

tan-1

we find n =.

tan-1

Therefore

A-

- *h-

There

tan-1 ^y.

Hence, by successive substitutions, we at last find

Generated on 2015-06-02 02:06 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

tan-1 =

4 tan-1

tan-1 -fa.

Thus, if we make x = \ in the formula, we find


the value of tan-H, and in like manner, if we make
we find the value of tan-1 ^. The calcula
x=
tion of six terms for tan1!, and of two terms for

tan-1 ^b)

will

Slve tne vale o* 2 UP to tne tenfcn

decimal.

91. The differential of an arc, when expressed in


function of polar co-ordinates, is (No. 52)
ds

.=

Vdff + rW,

being the radius vector, and <p the angle which />
makes with the initial line. The integration will
be made as follows.

INFINITESIMAL CALCULUS.

203

Example I. To find an arc of the spiral of


Archimedes. The equation of the curve is p = a<p,
a =

where

Hence

d<p

~'

and

ds = ^dp Va' + p\
Integrating by parts,

fdP

voir?

-f-^f

and the last term, integrated by formula (33), gives

Hence

Generated on 2015-06-02 02:06 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and taking

from p =

the integral

to p = r,

we

have for the first spire


S

2^r^+7+2

r
or, sinoe a =

r + VV + r"
5

To obtain the length of n spires, - it suffices to


take the integral from p = 0 to p = nr.
Example II. To find the length of the curve
traced by the end B of a tense string AB (Pig. 33),
whose other end A is fixed on a circle around which
the string is being wrapped.

INFINITESIMAL CALCULUS.

204

= I, and AC=r, and let E be a point


Draw ED tangent to the circle, and
on the curve.
draw the radius CD.
Fig. 33
and B
Make A CD =
the
arc
consequently,
AD = r&. Then the in
finitesimal arc ds =
described by ED will be
equal to ED x d&. But,
by the nature of the
Therefore
case, ED = AB AD = l-r&.

Let

AB

EF

ds = (l r#) d&.
Integrating from

=0

to

lS- br\

#,

we have

As a particular case, assume I = 2nr / then the


whole curve will be described when &=2it. Then

Generated on 2015-06-02 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

= 2aV, or s r=

Quadrature

of

Curves.

92. The differential of the area of a curve re


ferred to rectangular axes is (No. 51)
d A = ydx.

This equation, integrated between proper limits,

will

give us the area intercepted between the curve,


the axis of x, and the limiting ordinates.
Example I. Let the curve be a parabola y' = 2px.

Then dx =

whence

dA =

&
P

INFINITE SUIAL
Integrating from y =

pjo
II. Let

Example

CALCULUS.

205

to y y, we have
y

Sp

Then

the curve be a cycloid.

dx Jj^iL whence

\2ry

y*

V 2ry y'

Integrating by formula (37), we have

_ _ y V2ry - y' ^ Sr r ydy


2
2J Y2ry-y''
i/2ry-y
tfdy

and again, by the same formula,

ydy
Y2ry

Generated on 2015-06-02 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

y*

- V2ry

'
tf
+ r
'

vers-1

rf

Substituting this value in the last term of the pre


ceding equation, and taking the integral from y = 0
to y 2r, we have
,

-4

3r'

.2r

-rr- vers-1
2

3w"

This is the area of a semi-cycloid.

The area of the


whole cycloid is therefore 3w', or three times the
area of the generating circle.
Example III. Let the curve be a circle. In this
case, since

Vr* x', we have

dA - dx
whence (No. 73)

Vr*

INFIMTESIMAL

206

CALCULUS.

Taking this integral from x =

0 to

x = r, we shall

have

and therefore 4J. = ct-' for the

for a quadrant,
whole circle.

Example IV. Let the curve be the logarithmic

y = log x.

Then

dA

log xdx.

Integrating by parts, we have


log xdx x log x

I dx = x (log x

Taking this integral from x =

to x x

= x log x x +

l)-\-C.

we have

Generated on 2015-06-02 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

for the area extending above the axis of x, so long


as we take x > 1. But if we take the integral from
x = 0 to x = 1, then we find

[x (log x

1)

]0

that x log x x = 0 when


x = 0. Thus A = 1 is the area extending to infin
ity beneath the axis of x from x = 1 to x = 0, that
is, from ?/ = 0 to ?/ = oo .
Example V. Let the curve be the equilateral

for we know (No.

hyperbola

#3/

37)

Then y

= m'.

and integrating from

a;

to

A=m? log

hence

x = x,
x,

INFINITESIMAL CALCULUS.
When m =

207

log x. Thus the Napierian


logarithms, whose modulus is 1, exhibit so many
areas taken in the hyperbola xy = 1. Hence they
are also called hyperbolic logarithms.
1,

then

93. When the curve is referred to polar co-ordi


nates, then (No 52)

dA =
In

the

and dp =

fyi'dy.

spiral of Archimedes, in which


ad<p ,

a<p ,

we have

=
dA = $a'<f*d or dA = ^ or
p*dp. dA

^ p*dp.

Integrating the first expression from <p = 0 to


2n, and the second from- p = 0 to p = r, we
<p

Generated on 2015-06-02 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

have

-, both expressions reduce to


and, as a
the parabolic spiral, in which p' =
pdp ady, we have

In

A = iw\
2a<p,

and

dA =ay<d
and integrating from

f 0 to
J.

But,

as

a =

r*

j-

(No.

f>

2tt,

2a?ra.

55),

this

surface

becomes

INFIS1TESIMAL CALCULI'S.

208

In
da,

spiral, in which

the hyperbolic

p<p

= a, and

we have

And integrating from

=p to

= a,

a being the radius of the measuring circle.


In the logarithmic spiral, in which <p = a log p,
and

d<p

we have

dA~2 pdp>

Generated on 2015-06-02 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and integrating from p =

to p = p,

Surfaces of Revolution.

94. The area of a surface of revolution will be


found by integrating

the

differential

expression

(No. 51)

dS = 2ny

Vdx>

Example I. Surface of a

-f dy\

The convex sur


face of a cone is generated by a straight line y = ax
revolving about the axis of x. We have, in this
case, dy = adx; hence
cone.

INFINITESIMAL CALCULUS.

and integrating from x = 0 to x =

l;
7i,

dS = 2naxdx Va' +

209

2nr.

which

is

then a =

be the radius of the base

Let

S=7tah' VaF+l.
and

the common expression

dS = 2ny

hence

dx*

a;

d-ra

= 27rr<fo;

to x = 2r,

S=4xr\
the surface of the sphere.
Example III. Surface of a paraboloid of revolu
= 2px of the parabola,
tion. From the equation

V2px

dx' = 2ndx V2px -\-p'

and integrating from x =

2n

hence

dS =

we have dy
9 =

y'

which

is

Generated on 2015-06-02 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and integrating from

^'

(r

dx;

dy = - (r

x)

in Geometry.
Example II. Surface of a sphere. From the
= V2rx x* of the circle, we have
equation
of

to x

= x,

INFINITESIMAL CALCULUS.

210

Example IV. Surface generated by a cycloid

revolving about its base.

In

V2ry

the cycloid we have

y'

hence, by substitution and reduction,

dS=2nVr

ydy
\/2r y

y=2r z\

making 2r y z\ whence

or,

dy=

2zdz,

dS =
and

S=

An

4tt

V2r (2r

V2r (2rz

and

dz,

z*)

- |)

or

Generated on 2015-06-02 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

S = 4nV2f(2rV2r-y-$V (2r
Taking the integral from y =

-y)*) +

C.

y = 2r, and re

0 to

ducing, we get
Hence the whole surface generated

will

be =

V w1-

Solids of Revolution.
95. The volume of a solid of revolution will be
found by integrating between
differential (No. 51)
d

proper limits

the

V= Tty'dx.

Example I. Volume of the frustum of a cone.


The cone is generated by the revolution of a line

INFINITESIMAL CALCULUS.

211

y= ax about the axis of


y' = aV. Substituting this

x.
Hence we have
value of y% and inte
grating from x W to x = h, we find for the volume
of the frustum

V= n %

(h>

- hn) =

no?

(V + hh' + h"),
^-^'
o

where h 7a' is the altitude of the frustum, na?h? its


lower base, na'h" its upper base, and na'hh' a
mean proportional between the two bases, accord
ing to a well-known theorem of Geometry. Mak
ing h' = 0, we find for the volume of the whole
cone,

V
T7-

Tty

Example

A
75

II.

Volume of the sphere. The equa


tion of the generating circle being y' = r' x% our
differential formula becomes

dV=n(r,-af)dx;

Example

to

x = r,

III.

We

have

hence
it

and integrating from

V=

(a"

#')

<&r,

x= atox

v_ 4nb'a

= a,

'

about its transverse axis.

?/'

Volume of the prolate spheroid,


which is generated by the revolution of an ellipse

Generated on 2015-06-02 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

whence, integrating from x =

(a'x');

212

Example

ISFISITEMMAL

CALCULUS.

IV.

Volume of the oblate spheroid,


which is generated by the revolution of an ellipse
about its conjugate axis. In this case, the differen
tial of the volume is d V = xx*dy. Now

*- =

|!(&'-y');

hence

dV=x

(b'-y>)dy,

and integrating from y = b to y =

V=

b,

tea'b
3

Example V. Volume of the paraboloid of revolu

tion. The equation of the revolving parabola being


y1 = 2px, we shall have

Generated on 2015-06-02 02:07 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dV= 2npxdx,
and integrating from x =

to

x = x,

V=xpa? = ny*.ji,
which is equal to the volume of a cylinder having
the same base, and half the height of the parabo
loid.
Example VI. Volume generated by the revolu
tion of a cycloid about its base. Since, in this case,

dx=

-A

v2ry

the differential becomes

INFINITESIMAL CALCULUS.

V2ry

By formula

(37)

213

y*

applied three times in succession,

we shall find

V=
i
7t

^ *
- 2y'+5ry-\-15r'
V
0

l/s
V2rp-yt+
.

Taking the integral from

2/

5r*

-j vers-1

0 to

y
*

C.

y = 2r, we have

for one-half of the volume

hence the whole volume

will

be

5Ttr'

X nr.

Other Geometrical Problems.

96. Problem I. To find the curve whose sub-

Generated on 2015-06-02 02:08 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

tangent is constant.

The expression for the subtangent is (No. 34)


Hence

if

a be the constant,

ydx
,
dy

_ a.

Consequently,

ay

and x = a log
"y 4- ft

This equation shows that the curve is a logarithmic.


Problem II. To find the curve whose subnormal
is constant.
The expression for the subnormal is (No. 35) ^~

INFINITESIMAL CALCULI'S.
Hence

if a

be the constant,

ydy =
a
dx

ydy = adx, and

= 2ax

C.

axis of x.

Problem

is

therefore a parabola whose parame


2a, and whose vertex lies anywhere on the

III.

To find the curve whose normal

constant.
The expression for the normal

is

is

ter

is

The curve

y1

-\-

Consequently,

= a, whence dx =

j/l

a be the constant,

circle having the radius a.


To find the curve whose tangent is

constant.
The expression for the tangent

is

therefore

Problem IV.

= a'.

The curve

- Cy

y*

(x

or

x Va' -y' +

C,

and integrating,

is

Generated on 2015-06-02 02:08 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Hence,

if

ds

INFINITESIMAL CALCULUS.
Hence
V

if

\/

215

a be the constant,

= a, whence dx =

Multiplying and dividing by


dy

dx=

a*-jf_
Va'-y*

'

2/

_T

vV

Va'

- y\

y', we find

y^y

a'V

Va'-y*

Va'-y*'

and

2aV

Va

dz

3*

a"

then

a-z/

2a

And therefore
dV
~-

yVa'-y*

Consequently

_ a w
log

Va*

a*

a*-z*

1, a Va,
= - log +.

a+

-1,log6

ga-2~2a10g a"-2*
1

2a

yt

But

Jf

Generated on 2015-06-02 02:08 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

/dy

whence ay =

'

- y' =

Make a'

z%,

a;

INFINITESIMAL CALCULUS.

216

x=Va'-y'^a log

SL+f . +

The curve of this equation is called the tractrix.


Problem V. To find the curve in which the
square of the arc is proportional to the ordinate.
The equation of condition will evidently be
s* = ay, a being a constant.
Differentiating, we
2sds ady,

have

ady

ady

whence

dx = dy

y -14p

Making a = 8r, we shall have

Generated on 2015-06-02 02:08 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dx-. dy

y/2r -

whence

dy
<fo

y
v^ry

Now, this equation differs from the differential


equation of the cycloid in this only, that we have
-rr-

dx

instead of -=-

dy

It

follows that our equation

belongs to a cycloid whose vertex has the axis of x


for tangent. Hence the curve is a cycloid gene

rated by a circle whose radius is

r %.
o

INFINITESIMAL CALCULUS.

217

Problem VI. To find


the curve whose evolute is
a circ'e.

Let

AB = r

(Pig. 34) be
the radius of a circle, and
let the curve start from
the point B. Let M be
a point in the curve,
whose co-ordinates are x?
and y'. Draw MT tan
gent to the circle, and
therefore normal to the
curve ; then MT\v\\\ be the radius of curvature at
the point M, as is plain from the theory of evolutes
(No. 47).
Now, the equation of the normal is (No. 35)

since the line passes through the point T,


whose co-ordinates we shall designate by a and
we shall have
assuming x = a and

ft,

ft,

(i)

MT

tangent to the circle at the point


(a, ft), we have also from the general equation of
the tangent (No. 34)

But, as

is

Generated on 2015-06-02 02:08 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and

Therefore
da/
dy'

da

INFINITESIMAL CALCULI'S.

r\

ada

therefore

and consequently

-idfi =0,

-\-

/? =

or, omit

and

a'
%
;

since

Now,

218

ting the accents, and clearing of fractions,


ady pdx =

0.

(2)

(2)

and (3) we obtain

_~

r'dx
xdx ydy

-f-

reduces

r\

(3)

r'dy
#<&r

'

'

From

-j-

ax

to

a*

and this, owing to the relation

r\

On the other hand, equation (1) will become

ydy)* = r2 (dx'

/?

(axfcc

dy').

= a?+?/,
be the radius vector AM. Since
BMs,
we shall have pdp = xdx -{-ydy; also,
we shall have da;" + <%* =
Therefore, substi

if

Let

/>

tuting, and extracting the square root,


pdp rds

(4)

When

0,

then

= 2rs+

p=r.

C.

P*

whence

Therefore

s=

2rs

-f-

accordingly
p*

Generated on 2015-06-02 02:33 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

-(-

Squaring these equations, adding them together,


and changing a*
into r\ we find

r\

C=rt;

and

INFINITESIMAL CALCULUS.

MAB

Let the angle

<p ;

and (4) becomes


pdp

219

then ds = Vdp* + p'dy'

= rYd[?+p'd9\

whence
rd<p

- Vp* t*

and multiplying and dividing this by


integrating,

f=
When

f 0,

Vp2

r\

and

~ i*
+ sin-1 - + C.

we have p =

r;

hence C =

*
;

and

therefore

Generated on 2015-06-02 02:33 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

<p

Vp*

r'

cos-1

(5)

Such is the polar equation of the curve required.


To find a corresponding equation with rectangu
lar coordinates, we may remark that equation (5)

gives

r=

cos

QVp'-r'-y),

or

r-

cos

<p

cos - Vp*

- r" + sin

<p

1
sin - Vp*

But
p*

+ y\

cos

<p

sin

<p

r".

Therefore

r')

sin

Vaf+y'-r'}

The radical Va?


co-ordinates.
presents the radius of curvature.

y'

the equation of the curve with rectangular

Such

tV +

cos

is

r=x

2/

INFINITESIMAL CALCULUS.

220

r'= TM

Problem VII. A ship starting from

re

it

have

cos

AB AO'.d
r

d&,

BC = BO.dl = rdl;

CAB

m* rW\
AB BC tan

cos'

we have

(1)
a,

Since

ds*

AC2 =

r'

whence

a,

Generated on 2015-06-02 02:33 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

is

it

it

it

it

given lati
and
given longitude #0, cuts at
tude
given
a
meridian
which
every
through
passes.
angle
be when
has reached
In what latitude will
And where will
be when
has
longitude
travelled statute miles
Let us assume, for greater simplicity, that our
perfect sphere and that the ship finds
globe
no obstacle in its way.
Let OP = (Fig. 35) be
the radius of the globe.
AVhen the ship reaches
place A in latitude
let AC
and longitude
be the direction of its
in
course.
Taking
we
shall
finitesimal,

that is,

INFINITESIMAL CALCULUS.

221

cos ).d& rdk . tan a.

(2)

This being squared, and substituted in

= fdX

eW

+1 tan' )'

(1
x

(1), we have

7-W

-
cos

hence

,
<#s

As

=0

rdl
cos a

when

(2)' we

cos a

.14- C.

cos a

(X-*o).

(3)

cos a

= tan a

obtain

cos

whence

<7.

(|

tan a log tan

hence
-*o)

then

kj

#0

#0,

|
/0

But, when

= tan a log. tan


#

-+-

(No. 77)

l)

(2

tan
from which we obtain

j ~

= tan a log

tan

and therefore

&

Generated on 2015-06-02 02:33 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

From

we shall have

0
and therefore

(4)

INFINITESIMAL CALCULUS.

222

tan
2

tan-1

and

*=6-*)-

tan
tan-1

This value substituted in

(3) gives

tan
2 tan-1

= 45,

J?

rr:

0.

r= 0,

2tt.

Make

tan-1

^=z.

tan-1

4-

Then e* = cot \z

Then the ship starts from the equator in the direc


tion north-east or north-west, and traverses all the
meridians. Then we have

and tak

0
.

283 85

434294

2
.

=
X

2n.\og

^z

log. cot

ing the logarithms,

Generated on 2015-06-02 02:33 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

(?

and the problem is solved.


As a particular case, assume

728749,

INFINITESIMAL CALCULUS.

of 06'25".2.

which corresponds to the cotangent


Therefore
2

= 0

12' 50* .4,

and

223

90 0 12' 50". 4.

The ship will therefore be in latitude 89 47' 9'. 6.


Calculating the value of s and assuming the radius
of the earth = 3,960 miles, we shall find * = 8,774
miles.

The assumption

gives

oo ;

hence

the

ship can never reach the pole. Indeed, if its course


constantly makes an angle a with the meridian, it
is plain that whilst the meridian traverses the pole,
the ship must always be at either side of it.
Problems

soloed by Double

or Triple Integrals.

97. A few examples will show how geometrical


problems may be solved by double or triple inte
Generated on 2015-06-02 02:33 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

gration.

Problem I. A circle AMB (Fig.

the axis OZ.


To
find the surface gene
rated by the circumfer

36)

revolves

about

ence

of that circle.

Let

CM=r, CN=x,

NM=y,

and OG=R.
The distance of the point
M from the axis OZ will
be R-\-x; hence the in
finitesimal element ds of the circumference, while
describing an infinitesimal angle dd, will generate
a surface

dS= (li+x)ddds.

INFINITESIMAL CALCULUS.

224

Therefore

which is in
The integral from

We integrate first with respect to


dependent both of s and of x.
d = 0 to # = 2x will be

S=2x

&,

x)ds;
J%(R +

= 2*/- (jS sin-1

- i2

between
sin-1

x=

V?

0.

integral

1)

This

sin-1
the

taking

is

and

= Inr .nR.

the area of the surface generated

by the

is

The surface of the


semi-circumference AMB.
whole ring
therefore =.2zr .2ttR.
Problem II. The axes of two equal right cir
cular cylinders intersect at right angles. To find

of the solid

common to both.
Let us take the origin of co-ordinates at the in
tersection of the axes, the one being the axis of x,
and let
the other of
be the radius of the base

of the cylinders.

will

be

the volume

Generated on 2015-06-02 02:33 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

And

= 2w

flf

that is,

a;')

or, since

The equations of their surfaces

INFINITESIMAL CALCULUS.

225

whilst the volume of their common part will be ex


pressed by

^fffdxdydz.

Integrating first with respect to x, and reflecting


that the integral must begin from x = 0, we have

V=

ffxdydz,

or, substituting for x its value Vr* z',

V=JYdydz

Vr'

- z\

Integrating now with respect to y from y =


y = |V z1, we have

to

Generated on 2015-06-02 02:41 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

V=fdz (r' - z').


Integrating finally with regard to z, from z = 0 to
z = r, we have
3

This is one-eighth of the intercepted solid.


whole is = -=-

Problem

The

III.

To find the volume of the ellip


soid with unequal axes, whose equation is

* X.tji-i-l

INFINITESIMAL CALCULUS.

226

The general formula

V=

fff

dxdydz

being integrated from 2 = 0 to z z, then substi


tuting for z its value, we have

Integrating with respect to y, we have

Generated on 2015-06-02 02:33 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

/}? (a*

f?

Ir ft'te'-aQ

_ X*)

a*

_ b2(a'-x2)

y"dy
5

av
aV

?/

6' (a' x")

And, taking the integral from y =


b

Va'

0 to

x'

and reducing,
dir.

INFINITESIMAL CALCULUS.
This, integrated from x =

_ n be
I' a*

'

227

to x = a, will give

_ it

2a'

~3~2'

dbc
'

and, as this volume is only one-eighth of the whole


ellipsoid, the entire volume will be
3

Generated on 2015-06-02 02:33 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

If a b

'

= c, the ellipsoid becomes a sphere, and

228

INFINITESIMAL CALCULUS.

SECTION III.

APPLICATION OF INTEGRAL CALCULUS


TO MECHANICS.

Generated on 2015-06-02 02:33 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

98. Before attempting the solution of any me

chanical problem, it is necessary to give an exact


definition of Force. This word Force is often pro
miscuously used as synonymous .with quantity of
action, quantity of movement, quantity of pres
sure, quantity of acceleration, etc.. and a great
deal of confusion has been created by this loose
employment of the word.
When an agent exerts its power to produce or
modify movement, it acts; and the quantity of its
action is a dynamical force, which is measured by
the change that would be produced in the quantity
of movement if the action lasted for a second of
time.

A quantity of movement is
force; it is measured by the

kinematic or kinetic
product of the actual
velocity of the movement into the mass in motion.
A quantity of pressure is a statical force ; it is
measured by the quantity of movement that the
pressing mass would acquire in the unit of time if
all obstacle to motion were suppressed.
Though a body in movement can do work, as we
shall explain, yet movement, as such, is not a
a

INFINITESIMAL CALCULUS.

229

force, but only a change of position due to an exer


tion of power. Velocity is the act or form of
movement ; hence, when a body moves with a con
stant velocity, its movement is said to be uniform.
Velocity has intensity, but the space measured by
Time is the actual
the body has only extension.
ity, or duration, of movement ; and if the movement
be uniform, time is the ratio of its extension to its
intensity, that is, the ratio of the space measured
to the velocity with which it is measured.
Velocity, on the other hand, is measured and ex
pressed by the length which it causes to be meas
ured in the unit of time ; in other words, the
intensity of the movement is measured by its ex
tension in the unit of time.
Hence, if the velocity
v remains unaltered, the space s measured in a
number t of seconds, will be s vt.

Generated on 2015-06-02 02:34 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Velocity is always gradually acquired, or grad


ually lost, through a series of infinitesimal incre
ments or decrements corresponding to the series of
infinitesimal instants during which the body is
acted upon. This is true even in the case of the
so-called instantaneous forces, v. gr., in the com

munication of velocity by impact. Even in this


case, the action is really continuous ; and the only
reason why it is called instantaneous is, that its
continuation is too short to allow us to value ex
actly its duration. This short duration we often
call an instant, though it is a finite length of time,
and comprises a series of infinitesimal instants. It
follows, that all velocity is gradually acquired, or
lost, by infinitesimal degrees, through some con
tinuous exertion of power. It also follows, that a
constant continuous action, all other things being

230

INFINITESIMAL CALCULUS.

equal, is proportional

to the

length of time during

If, then, an agent


which it is allowed to continue.
by its continuous action is competent to impart to
a body a velocity v in the unit of time, the same
agent, all other things remaining equal, will in the
instant dt impart a velocity vdt ; for 1 : dt : : v : vdt.
Let us conceive a free material point, which
under the continuous action of an agent A meas
ures a space a; in a time t, and at the end of this
time has a velocity v. If left to itself, the point
will, with this velocity, measure in the following
infinitesimal instant dt an infinitesimal space dx ;
and as movement during an infinitesimal instant
cannot but be uniform, we must have

Generated on 2015-06-02 02:34 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dx = vdt,' and

^.

dt

But, if the agent A continues to act after the time


t, then the velocity v in the instant dt will undergo
an infinitesimal change dv. Let, then, a represent
the intensity of the action at the end of the time t,
Since this action will in the instant dt produce the
change dv, we shall have
adt = dv, or a

dv : : 1

dt

which means that the actions, all other things


being equal, are proportional to their duration.
From adt = dv, we have
a=

dv

m,

or a =

d?x

w,

which is the general expression of the action that


modifies the velocity of the movement.
Its effort
is called acceleration.

INFINITESIMAL CALCULUS.

231

If, instead of

one point, we have a mass contain


of points, and if the agent A acts
ing a number
equally on each of them, its total action will be

,.-

-,r

dm

Ma=Mm

,r

d?x

Mw>

and this is called a dynamical force, or better,


quantity of the action after the time t. It is
product of the acceleration into the mass of
body acted on.
The quantity of movement of the mass M at
end of the time t, is evidently

Generated on 2015-06-02 02:34 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Mv, or

M^

the
the
the
the

and this is a kinetic force. It is the product of the


mass into its velocity.
The quantity of pressure, or the statical force, is
expressed by Mv like the kinetic force, v represent
ing the velocity which the mass M would acquire
in a second of time, if no obstacle existed. It is
the product of the mass into its virtual velocity.
Work.

99. A body moving under a continuous resist


ance is said to do work. The work is by so much
the greater according as a greater mass measures a
greater space under a greater resistance. Hence
the unit of work will be the work done by the unit
of mass, measuring the unit of space, under a unit
of resistance ; and, if m be any mass, R the resist
ance,
have

8 the

space measured,

W m.R.

S.

the work, we shall

INFINITESIMAL CALCULUS.

232

are going to show that, if u be the initial


velocity of the mass m, the total work which this
mass can do, is always = \mv?.
When the moving mass, after a time t, has
measured a space x, its velocity may be expressed

We

dice
;

and the resistance, which is an action modi

fying that velocity, and opposed to

product

m, dx, and

of the three factors,

work

will be the

to the instant dt,

corresponding
o

dW

Hence the infinitesimal

pressed by

d,x

will be ex-

it,

by

d?x

^r

dW=

mdx

Therefore

dt

Integrating this, we have

we

Therefore C= \mv?.

W=
When

the

total

have

n
-im (dx\
{dt)+a
.

W=

and

cLcc

^7

= u.

Hence
%mu?

km

work has been

Therefore the total work

done,

then

is

0,

when

But,

0,

d'x

W=-mJ df-di

0.

%mu\

The total work of which a mass


capable is
often designated by the name of energy, or work
is

Generated on 2015-06-02 02:34 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

rdx

stored up in the mass m.

INFINITESIMAL
Example

CALCULUS.

233

I.

Work under constant resistance.


A body m is thrown up vertically with an initial
velocity u. The resistance is here the action of
gravity, which we represent by the letter g. Hence
we have

Multiplying both members of this equation by dx,


and integrating,

When t =

0,'

dx

we have x 0,' and -jn u.

dt

There-

Generated on 2015-06-02 02:42 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

fore C = $u*; and

But, when the whole work is done,


and x=S.
by m,

Therefore

gS=\u\

m.g.S=
Example

II.

or,

then

dx

0,

multiplying

kmu*.

Work tinder a resistance varying


in the inverse ratio of tlie squared distances. Let
a mass m with the initial velocity u recede from a
centre of attraction whose action at the unit of dis
Let a be the
tance we shall designate by A.
original distance of m from the centre of attrac
tion.
Then, every particle of the mass m, after a
time t, will be subject to a resistance or retardation

INFINITESIMAL CALCULUS.

234

A
dt' ~(a + x?

d'x

'

Hence

Adx
dx d'x _
dt ~dt~~ (a + x)'
When x =

\dt)

= u:'

we have

0,'

(dxY ~

'

hence

dt
Substituting, and reducing,
1

(dxY

\dt)

_ iU,

.
n
+
a+x

C=

'a

Ax

a(a+x)

When the whole work has been done,' then


and x = S.

sr =

dt

0,

Hence

iu

~a(a + S)

'

Generated on 2015-06-02 02:42 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and multiplying by m,

a(a + S)
where the resistance

between the initial resistance


sistance

:^
(a + sy

-.

Examplk

is a geometrical mean
.

ft

and the final re

III.

-Res/stance proportional to the


velocity of the movement. In this case, if a be a
constant, we shall have

d'x

dx

d'x

INFINITESIMAL CALCULUS.

j7 =

dt
x=

When

then

0,

ax +'

u;

235

G.

hence

Cu;

and

therefore
'

dt
whence

As 2 =

gives

a;

0,

hence

log .

There

fore
1

ax

Generated on 2015-06-02 02:42 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

When the whole work has been done, then


x= 8=

and t =

It

oo .

^-

would, therefore,

0,

take

infinite time to measure the finite space S, and to


exhaust the velocity u.

Introducing the value

into the general

equation
find R = \au. Thus the mean resistance is here
an arithmetical mean between the initial resistance
au and the final resistance zero. Therefore the
mass m would, under a constant resistance \au
we

measure the space

S=

INFIMTESIMAL

236

Example

IV.

Resistance proportional to the


In this case, if a be a constant,

space measured.
we have

_
_-ax,

d*x

When x =

0,'

CALCULUS.

dx d'x -_

dtat-

then -^-.u.

dt

Hence

ucdx,

0=u\

There-

fore

When the whole work has been done,' then


x=

=.

Va

8.

dt

= 0,'

Hence

m.S.R=m Generated on 2015-06-02 02:42 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dX

-37

Va

R = %mu*

and therefore R \u\fa. This mean resistance is


an arithmetical mean between the initial resistance
zero and the final resistance u Va.
It may be observed that the

/dx\*

{dl)=u-ax

which, when x =

integration

glves

S=

Va

reduces

to

of

INFINITESIMAL CALCULUS.

237

sin_1l

2Va

Va

This shows that the time employed in doing the


whole work would be independent of the initial
velocity, and of the space measured.

Uniformly Varied.

Movement

100. When the velocity of

a mass m increases

or decreases uniformly, its movement is uniformly


accelerated or retarded ; which implies that the
action, under which the movement varies, must be
constant. Assume m = l, and let the accelerating
action be g. Then the equation of the movement

will

be

-37-,

= g ; whence
and s = igt' + Ct +

C,

being the space measured at the end of the time


ds
t. The constant C is what the velocity
becomes
-^
s

t.

is

0.

is

that

= 0;

t,

gt

is,

the initial velocity.


will be =
Hence, if there be no initial velocity,
=
what
The constant C
becomes when
that is, 0" is the initial space, or a space already
measured before the beginning of the time
Hence, when the space is reckoned from the begin
will be =0. Supposing
ning of the time
therefore (7=0 and C =
in the above equations,
the velocity
acquired in the time will be

when

Generated on 2015-06-02 02:42 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

g =gt + 0,

(1)

INFIXITESIMAL CALCULUS.

238

and the space traversed in the same time


s

= igf.

(2)

This last equation, being multiplied by


2gs = g't\ or
= 2gs, and v = V2gs,

(3)

is,

0'

2g, gives

is
s

is

s,

is

that
the velocity acquired or lost by a body
while measuring a space
with uniformly accele
rated or retarded movement,
equal to the square
root of twice the product of such space into the
accelerating or retarding action. Such velocity
whilst
styled the velocity due to the space
itself
called the space due to the velocity v.
Movement not Uniformly

Varied.

in general, varies inversely as the squared distances


of the bodies attracted. Let be the radius of the
g

the intensity of its attraction at its surface,


earth,
the intensity of its attraction at a distance s.
g'

r'

-,.

whence

g'

r*

::

Then
g'

We shall have, therefore,


d's _

r'

Generated on 2015-06-02 02:42 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

101. The attraction of the earth, and of planets

where the second member has the negative sign,


because the action tends to diminish the dis
tance s.

INFINITESIMAL CALCULUS.
Multiplying

this by

and integrating,

2ds,

239
we

have

Let

=h

when

i=0

then

ds

0,

and

hence

(S'-wG-i)-

Such is the velocity acquired by the body while


falling from the original height h to the height s.
Let h = oo , and * = r. Then equation (1) be
comes

Generated on 2015-06-02 02:42 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and, as

gr

= 2gr,

hence

32.088 feet,

r=

= V2gr,

20,923,596

feet, we shall

find
v

36,644 feet.

Thus a body falling upon the earth from an in


finite distance would reach its surface with a veloc
ity of nearly 7 miles a second.
attraction of the sun at its surface is
g = 890. 1 6 feet, and its radius is r = 430,854 . 5 miles.
With these data, we find that a body falling upon
the sun from an infinite distance would have a final
velocity of about 381 miles per second.
From equation (1), extracting the root, and tak
ing the negative signs (as ds diminishes when dt
The

240

INFIX ITES IMA L

CALCULUS.

increases), we obtain

dt--J

dsVs

a/

which may be written thus,"


h

Hence
,

2s

Generated on 2015-06-02 02:42 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

But, when t =

0,

we have s = h.

Therefore

And therefore

This equation, when s = r, will give the time re


quired for a body to fall from a distance h to the
surface of the attracting body.
The mean distance of the moon from the earth is
The time re
60r, r being the radius of the earth.
quired for a body to fall from the moon to the
earth would be t = 417,881 seconds, that is to say,
4 days, 19 hours, 56 minutes, and 21 seconds.

INFINITESIMAL CALCULUS.

241

Composition and Decomposition of Forces.

102. Before we proceed further, we must say

Generated on 2015-06-02 02:42 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

few words on the composition and decomposition


of forces.
Dynamical forces are usually repre
sented by lines proportional to the velocities which
they impart in the unit of time ; kinetic forces by
lines proportional to the spaces measured in the
unit of time ; statical forces by lines proportional
to the spaces that would be measured in the unit
of time, if no obstacle existed.
Let two equal forces P and P (Fig. 37) be applied
to a point A, and let them make with each other an
If we
angle PAP = 2x.
can find a single force R,
which applied to A will
produce the same effect as
the two forces P, this new a
force will be called the re
sultant of the forces P,

whilst the forces

P will be

The resultant will evidently lie in


its components.
the plane of its compcnents ; and when the two
components are equal, it will bisect their angle.
The value of the resultant depends both on the
intensity of its components, and on the angle at
which they meet. Hence the resultant R of the
two equal forces P may be expressed by

R = 2Pf(x),
f(x)

being a trigonometric function of the angle x.


This function is easily determined. For we know,
that when x 0, the resultant is R 2P. Hence
/(0) = 1. We know, also, that when x 60, the

INFINITESIMAL CALCULUS.

242

resultant is R = P. Hence /(60) = J. We know


in like manner, that when x = 90, the resultant is
R 0, as the two forces then neutralize each other.
Hence
Now, the only trigonometric
(90) = 0.
function which can satisfy these conditions is the
cosine of x; for

cos0

= l,

cos60 =

cos 90 = 0.

Therefore the resultant of the two equal forces

R = 2P

is

cos x.

which will intersect AR at right


angles in D, we have AD = P cos x ; and there
fore R = 2AD AR. Thus the resultant is repre
sented as to its intensity -and direction by the
diagonal of the rhombus constructed on the two
equal components.
Drawing

PP.

Generated on 2015-06-02 02:42 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

103. Let now two unequal forces


at
applied to the point
right angles (Fig. 38), and
let x and y be the angles
which the resultant shall
make with P and Q respec
tively.

If P

be

conceived

as

and Q be
Fig.

SS

resultant of two equal forces


p that make with it an angle
x, one of these components will lie in the direction
of the resultant, and the other in the direction Ap.
If, in like manner, Q be conceived as a resultant of
two equal forces q that make with it an angle y,
one of these components will lie in the direction of

INFINITESIMAL CALCULUS.

243

the resultant, and the other in the direction Aq.


And we shall have (No. 102)

P 2p cos x,

2q cos

y.

Now, evidently, the resultant is the sum of the two


forces p and q which lie in its direction, while the
other two forces p and q which lie in the direction
Ap and Aq, are directly opposite, and must neu
tralize each other. Therefore p =q, and R 2p2q.
Hence
cos x

or

cos x =

P,

cos y

R cos y =

Q.

Squaring these equations, adding them together and


remarking that in our case cos y = sin x, we obtain

Generated on 2015-06-02 02:43 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

R' =

P' +

Q',

that is, the resultant of two forces meeting at a


right angle is the diagonal of the rectangle con
structed on its components.

If

and Q do not form a right


the forces
as the diagonal
angle (Fig. 39), then considering

104.

P
_

of a rectangle of which
n
one side Ap lies on the
Q
resultant, and Q as the
!
diagonal of a rectangle
!
of which one side Aq
Q
lies on the resultant,
1
then the resultant will fi
be the sum of Ap and
Aq, and the other components Ap' and Aq' will be
opposite and destroy each other. Now, if x and y

INFINITESIMAL CALCULUS.

244

be the angles

that

and Q make with the result

ant, we shall have

R = Ap +
Squaring,
1

sin* x,

Aq

cos x 4- Q cos y.

and changing cos" x and


and 1 sin3 y, we have

2? =

P'-f #'-P"

But,

as

sin' y+2PQ cos

sin'

Ap' = Aq', we have P/i =

cos' y into

sin x Q sin

cos

or

Qq,

y/

hence

P" sin' x= PQ sin

a?

sin y =

Q* sin2

and

P" sin'

a: +<2"

sin" y

= 2PQ

sin

a;

sin

/.

Theref ( >re, substituting,


-f-

2P$

(cos x cos

sin x sin y),

+ 2P

cos (x

P2 = P2

2/)

or finally

which
the expression for the diagonal of
paral
contain the angle
lelogram whose sides P and
x -f- y. Hence the resultant of two forces meeting
at any angle is the diagonal
the parallelogram
constructed on those forces as sides.
When the forces are more than two, their resxiltant is obtained by successive compositions. On the
other hand, every resultant can be resolved into
for the same line
components, in many manners
may be the diagonal of many different parallel;

<

is

Generated on 2015-06-02 02:43 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

R* = P" -j-

ograms.

INFINITESIMAL CALCUL US.

245

Moments.

105. Let

rigid rod AB (Fig.

40) be acted on

by a force P applied ;it A, and by a force Q ap


plied at B. If the direcFig. 40
tions of PA and QB con
verge towards a point M,
we may, without altering
the system, apply the two
forces
at M, and find
their resultant MN. Pro
longing MN towards R,
and taking OR = MN,
/
line
the
Pi
OR will represent
the resultant of P and Q
applied to the rod at the
point O, and producing
the same effect as the
forces P and Q applied at

Generated on 2015-06-02 02:43 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and B.
To determine the posi
tion of the point O on the rod, draw NH and Op
perpendicular to AM, also NK and Oq perpendic
ular to BM. The similar triangles NPH and NQK

will give

NH: NK:: NP

NQ :: Q

P.

On the other hand

NH: NK::
therefore

Op

Q :: Oq

Oq

Op

or making Op =p, and Oq q,

::

p, whence Pp =

Qq.

INFINITESIMAL CALCULUS.

246

If

and Q were parallel, we could,


without altering the system, apply at the extremi
ties A and B of the rod (Fig 41) the equal and op
posite forces AX and BX, which being com
pounded with P and Q respectively, would give the
two converging resultants
AP' and BQ'. These re
sultants would now be
transferred along their
direction to M, and de
composed there, the first
into P and MX, the sec
ond into Q and MX; and
the forces

MX

de
there

since the forces


stroy each other,

Q,

:
:

Q'

P'.Op' = Q'.Oq'.
But the similar triangles AOp' and

P:

P'r.Op' OA,
:

Generated on 2015-06-02 13:18 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Q.

is,

would remain only


and Q; that
the resul tant would be P +
Prolonging MP towards

B, and taking OB =
P + the force OB ap
plied to the rod at the
will produce the
point
same effect as the two forces P and
applied
at A and B.
To determine the position of the point
on the
rod, draw Op' perpendicular to AM, and Oq' per
pendicular to BM. Then, by the preceding de
monstration, P'
and
Oq' Op'

PAP'

give

INFINITESIMAL CALCULI'S.

247

and the similar triangles BOq' and QBQ' give

Q:Q'::Og'

OB;

therefore

P.OA

P.Op',

and Q.OB = Q'.Oq'.

Hence we have also

P. OA = Q.OB,
or, making

OA=p, OB

= q,

Pp=Qq,

Generated on 2015-06-02 13:05 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

as in the preceding case.

The products Pp and Qq are called moments.


Hence the moment of a force with respect to a
point O is the product of the force into the perpen
dicular drawn from O to the line of its action. The
moment is the expression of the effort with which
a force tends to turn, the rod about the point O,
when this point is fixed. In this case, the fixity of
the point O will neutralize the resultant of the ac
tions ; and the equality of the moments Pp and Qq
shows that the efforts of the forces P and Q balance
each other, and that the rod is in equilibrium.
When the forces P and Q are parallel, their re
sultant is parallel to them, as is evident from the
mode of its construction ; and the same is true of
the resultant of any number of parallel forces.

106. When two forces, or more, act in the

same

plane on the same point, the moment of the re


sultant taken with respect to a fixed point is equal
to the algebraic sum of the moments of its compo
nents with respect to the same fixed point.

INFINITESIMAL CALCULUS.

2-48

P and

(Fig. 42) be two forces acting on the


point A, and let R be their resultant. Let AO be
a rigid rod which can re
volve about the fixed point
0. This point O will then
be styled the centre of mo
Draw Op, Oq, Or
ments.
perpendicular to AP, AQ,
AH respectively. Denote
the angle BAP by a, RAQ
(or its equal ARP) by
we have from the
Since PR =
and RAO by
PR the equations
triangles APJY and
Q

Q,

sin

/?

cos

RQ

/3

If

-4-

P sin
<p

(sin

or

/?)

(p

sin

R sin =

f)

/?

cos
cos a sin a cos

P sin

(#>

+ sin

cos

/?

(sin

<p

<p

sin

a).

a)

sin

(p

sin^=40'

>

sm(f +

Or

3*

But

Or =

Op

=p,

Oq =

Rr = Pp

and reducing, we have


Qq,

q,

substituting in the preceding equation, making


r,

Generated on 2015-06-02 13:05 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

<p,

a.

and the second by


and adding together the results, we find

Multiplying the first by sin


cos

cos

a,

<p
.

/3,

Let

p)>

WFINITESIMAL CALCULUS.

249

as was to be proved. And, if we regard Q as a re


sultant of two forces, and these again as the
resultants of other forces, our conclusion can be
extended to any number of concurrent forces acting in the same plane, and thus we may write in

general

Rr = Pp + P'p' + P"p" + P"'p"' +

within the angle formed


by the resultant and any of its components, then
the moments of such components are negative ; for
When the point

O lies

sin (f a) is then negative.


Hence the above sum
of moments is algebraical, not arithmetical.

Virtual Moments.
107. When two forces
extremities

and Q applied at the

AB

(Pig. 43) are


in equilibrium about the
fixed point O, if the equi
librium be disturbed or
by an ex
endangered
traneous force, the points
A and B will describe, or
tend to describe, similar
arcs A A' and BB' about
the point O ; and we shall
have the proportion
the lever

Generated on 2015-06-02 13:05 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

A and B of

Fig, is

A A1 BB'r.OA
:

OB.

Projecting A A' on the direction of P, and BB' on


the direction of Q, we shall have also

Aa

Bb-.-.AA'

BB'

INFINITESIMAL CALCULUS.

250

and therefore

Aa: Bbr.OA
and because OA

OB::Q

P,

Aa: Bbv.Q: P,

OB;

therefore

P.Aa = Q.Bb.

or

The lines Aa and Bb represent the so-called


virtual velocities of the forces P and Q. Not that
these forces have any velocity at all, but because
Aa and Bb are the measure of the virtual velocities
which rule the ascent or descent of the points of
application of the forces. The former, Aa, which
falls on AP is considered positive ; the latter Bb
which falls on the prolongation of BQ is taken as
The products P.Aa, and Q.Bb are
negative.
called the virtual moments of P and Q.
The virtual moments of P, Q, and their resultant
R (Fig. 42) are P.Ap, Q.Aq, R.Ar. Resuming
the equations

Q cos

/?

sin

/3

multiplying the first by

P cos a,
P sin a,

f,

the second by sin


and subtracting the second from the first, we find

(cos

sin

cos

sin

sin

00s

ylr

2o'

cos W

cos

+/?)

An
2lo

'

But

-f-

(y>

cos

=P

a)

cos

a)

or

f)

f + sin

/9

(cos a cos

tpy

9?)

y>

/9

tp

cos

(^

cos

Generated on 2015-06-02 13:05 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

INFINITESIMAL CALCULUS.
cOS

251

(f + 0 = -^.

Substituting in the preceding equation, and reduc


ing, we have
R.Ar P. Ap + Q.Aq;
that is, the virtual moment of the resultant is equal
to the algebraic sum of the virtual moments of its
This conclusion extends to the re
components.
sultant of any number of forces that lie in the same
plane.
The virtual velocities Ar, Ap, Aq, etc., are
usually represented by 3r, dp, Sq, etc. Hence we
may write, in general,

Generated on 2015-06-02 13:05 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

fidr = Pdp + P'dp' + P"8p" +

where the character d denotes an infinitesimal vari


ation corresponding to an infinitesimal instant dt.
The point O with regard to which the virtual
moments are taken, is the centre of the virtual
moments.
When the algebraic sum of the virtual
moments of the forces is =0, the system evidently
is in equilibrium about the point O.

Attraction of a Sphere on a Material Point.


108. Let C(Fig. 44) be the centre of a spherical
shell of infinitesimal thick
and

a material
point at some distance
from the shell.
If the
mass of the shell be de
noted by m, the mass of
the infinitesimal zone AB,
whose centre lies at a dis
tance CD x from the centre
ness,

of the shell, will

INFINITESIMAL CALCULUS.

252

be expressed by dm =

being the radius of

the shell. The


upon the point

action of this infinitesimal mass


P is the resultant of the equal ac
tions of all its particles, and is equal to the sum of
such actions multiplied by the cosine of the angle

APD
Let

(No. 102).

AP = z.

By the Newtonian law, if the ac


particle at the unit of distance be = 1, its

tion of a

action at the distance z will be =


action

cos

Generated on 2015-06-02 13:05 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

PD

D71 =
APD
-rr> =
AP
.

where a = CP.
PAC we have
z1

on the point

mdx

But

A P1 =

AB

of the zone

d<p

CP-

Hence the

will

be

jtn

CD

a-x

'

On the other hand in the triangle

- 2ar

= a' + r*

cos

A CD

= a* -j- r* 2ax

therefore

to
2/-

(a

x) dx

V (a' + r'

'

2ax)'

or
,

to

xdx

adx

Now
adx
V

(aa

+ r*

2ax)3

V a' +

r*

2ax

INFINITESIMAL CALCULUS.

253

and

x
a Va? +

V a' +

2ax

a'

-2ax~^~

r2

'/'"

xdx

Hence, substituting, and reducing,

(a

r)

-\-

a
+
a-\-r

(a

2ra*

TO

and taking the integral from x =


reducing,

to x =

r,

- 2ax + 0;
and

_m

a'Va2 + r*

r)

2r

ax

to

centre.

109.

If

the point

were
placed anywhere within the
spherical shell (Fig. 45), the
resultant of all the actions of
would con
the shell upon
stantly be =sO, and therefore the point

it

For, in this case,


remain in equilibrium.
ferential equation would be

a) dx

'

r'

(o^

2r

(x

df

'

Generated on 2015-06-02 13:05 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

is

if

it

if

is

that is, the total action


the same as
the whole
mass of the shell were concentrated in C. The same
being true of all the shells into which solid sphere
follows that the action of a
may be decomposed,
sphere formed of homogene
Fig. 45
the same as
ous shells
its
mass were concentrated in its

~2axJ

would

the dif

INFINITESIMAL CALCULUS.

254

which differs only by the sign from the equation


of the former case.
Hence the integral will be

r'

"

ax

taken from x =

if

and this,

2ra* \

_l c

to x

= r, will give

- a) _ r (r + a) _
a
r -j-a

r (r
r

>

And this is true of all the shells whose radius is

greater than CP. Accordingly, if the point P were


placed within a solid sphere, it would be attracted
as if the shells beyond CP had no existence ; that
is, it would only feel the attraction of the nucleus

PQIt.

If an

opening were made along one of


the diameters of the earth, and a body allowed to
fall through
the body (abstraction being made
from the resistance of the air) would be urged to
wards the centre by an action varying as the simple
distance from the centre.
the density (sup
For, let be the radius, and
Its mass will then be
posed uniform) of the earth.

r,

g::s

=
g'

centre

g'

distance

whence

or

^, and

s T

action of the nucleus at

g= -

the earth at its surface

PQR will

Now, the action of

being its radius.

p,

the mass of the nucleus

g'

4ttS*

whilst

is

be

p;
-5o

4ivr'

is

Generated on 2015-06-02 13:05 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

it,

Corollary.

the

from the

INFINITESIMAL CALCULUS.

255

Hence the equation for the movement of a point


approaching the centre, is

d's

gs

dJ'~~V
Multiplying by 2ds, and integrating, we have
gs'

(ds\'_
Making

-^

when

= r,

we

have

C=

Therefore

()'=*=? *'-'>

-w

When the body reaches the centre, then

0r

the maximum velocity.


When the body has reached the centre, its velocity
will carry it further on, and. as s changes its sign,
the motion will be retarded instead of accelerated,
until v reduces to zero when s= r. Then the
body will fall again towards the centre, and meas

and

Generated on 2015-06-02 13:28 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

v=

Vg/-,

which

is

ure backward the same diameter, and perform a


continuous series of oscillations of the same kind.
Extracting the root of equation (1) and taking
the radical negatively, because ds and dt have op
posite signs, we find

Vf'-s'

V g

and this integrated from s =


y

to

s=

r, gives

This is the time of one entire excursion.

This time

256

INFINITESIMAL

OA

LUULUS.

is equal to that in which a body would measure the


semi-circumference r.r with a uniform velocity

The time t of the excursion is independent of the


distance from which the body begins to fall. For,
since we have

r g::s
:

g',

we can replace the radical

without

altering the value of t. Hence all the excursions


will be isochronous, whatever may be their simplitude.
But these results would be greatly modified
by the resistance of the air, which we have
neglected.

Generated on 2015-06-02 13:18 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Centre

of Gravity.

110. The centre of gravity of a body is a point


within the body, through which the resultant of
the actions of gravity on each particle of the body
All these actions are directed
always passes.
towards the centre of the earth ; yet they may,

without error,

be

considered parallel.

Hence their

resultant is their sum (No. 105).


It is obvious that the centre of gravity of a
straight line is at its middle point ; also, that the
centre of gravity of a plane figure is in that plane,
and if the figure has a line of symmetry, its centre
of gravity is on that line. In like manner, if a
solid has a plane of symmetry, its centre of gravity
is in that plane.
The centre of gravity of a homogeneous body

INFINITESIMAL CALCULUS.

257

does not depend on the intensity of gravity or on


the density of the body. Its position depends only
on the form of its volume. We may therefore sub
stitute volumes for masses and weights, and con
sider only the relative position of the elements of

which the body is composed.


Let M (Fig. 46) be a homogeneous

body of any
form. Draw rectangular axes, and let the plane
be horizontal.
The action of gravity will be
parallel to the axis AZ. Let to be an element of
the body, and let its co-ordinates be x = qn, y pny
If dv be the
z = mn.
< :/
volume of thtj element
to, its moment with re
spect to the axis A Y
will be xdv. Every other
element of the body M
will give a similar mo
ment, the value of x
Generated on 2015-06-02 13:06 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

XY

varying between
the
limits of the body. Hence
the sum of the moments
of all the elements with
respect to the axis

A Fwill

be

I xdv.

Let now O be the centre of gravity of the body,


and x0 = DC, p0 = BO, z0 = OG, its co-ordinates.
Since the resultant of the actions of gravity passesthrough O, the moment of the resultant with re
spect to the axis A

will be

xn

the theory of moments (No. 106),


x

/dv.

I do = I xdv ;

Hence, by

INFINITESIMAL CALCULUS.

258

and therefore
xq

If
axis

the moments

AX,

we

xdo

were taken with respect to the

would find in like manner


y<i=

7^

and if the figure were turned about so as to make


the axis of x vertical, we would have also

Generated on 2015-06-02 13:06 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

z0-

j* zdv

7^

Such are the values of the co-ordinates of


centre of gravity of the body.

the

111. Centre of gravity of a circular arc.

Let

the axis

OX

(Fig. 47) bisect the arc


ABC. Then OX will be a line of
symmetry, and the centre of gravity
will lie on OX. Let AC=c be the
chord, and OA = r the radius of
the circular arc. If O be the origin
of co-ordinates, the equation of the
circle will be

tt =

ra

-x';

and therefore
dv = VbW+dy1 = dy

dy
x

rdy

Vr*-tf

'

INFINITESIMAL CALCULUS.

259

Substituting in the above expression for

x0,

we have

J'rdy

/rdyVr*

y*

and integrating from y = \c to y = $c,

Xo~

rc
c\\ arc. ABO'

rc

(sin- 2- sin-^
I

.(

?,

= 2r sin
we have
Making the angle AOX =

whilst the arc ABC


2r& whence we get also

-~ J~
Vr'y

xdv = 2y'dy;

dv = 2ydx =

/2y2dy
Vr'

and integrating from


(33),

yl

'

whence

Generated on 2015-06-02 13:06 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

112. Centre of gravity of a circular segment.


Referring to Fig. 47, where the segment ABC is
bisected by the axis OX, we find

to

yy,

by formula

_ sin

Cos

<?

"

sin*

"

_2r

T0

substituting and reducing,


&

sin

<?,

and, since y
we have

&

INFINITESIMAL CALCULUS.

2C.0

113.

O,

Centre of gravity of a circular sector.


Referring again to Fig. 47, let us conceive that the
sector OABC has been divided into equal infini
tesimal sectors, every one of which may be looked
and an
upon as
triangle having the vertex in
The centre
infinitesimal base on the circumference.
of gravity of every one of these sectors, or triangles,
as can
will be at
distance \r from the centre

In

114.

ABC~

sin

'

arc

_ 2r
3

ABC~

'

arc

2r
3

Xa

x
jg

Ill,

of gratity of a parabolic area.


= 2px, we
parabola, whose equation is
Centre

hence

V2p. x^dx,

xdv =

\^2p

x*dx

= 2ydx =

dr)

y*

the
have for an area comprised between the curve and
a double ordinate,
2

Generated on 2015-06-02 13:06 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

it

be shown by a simple geometric construction.


Hence the whole series of these centres of gravity
will determine an arc of circle similar to the arc
ABC, but having a radius fr. As the matter of
each elementary sector can be considered concen
trated in its centre of gravity,
follows that the
centre of gravity of the whole sector is the same as
the centre of gravity of said arc. Hence, applying
to our case the result of No.

INFINITESIMAL CALCULUS.

261

x dx
x dx

Let a be the terminal abscissa of the area in ques


tion. Integrating between the limits x 0 and

x = a,

we shall have

a value independent of the parameter of the curve.

115. Centre of gravity of a paraboloid of revo


lution. In this case
dv

ny^dx,

xdv = ny'xdx,

that is,
dv = 2n.pxdx,

xdv = 2jt .px'dx.

Hence
&o

= Jx'dx
-r

/ xdx

Generated on 2015-06-02 13:19 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

and integrating from x=.


Xq

to x

a, and reducing,

8^5

a value independent of the parameter of the curve.

116. Centre of gravity of


Let O (Fig.
a right pyramid.

and OA = h
the height of the pyramid, o"
Calling b the base, any section
parallel to it at a distance
48) be the vertex,

x from the vertex, will be =


tesimal volume will be

, -

x1

and its infini-

Hence

IXFIXITESIMAL CALCULUS.

262
dv

= i2 afdx,

therefore
x0

and xdv

J/

x*dx;

afdx
;

x'dx

and integrating from x =

to x = h, and reducing,

x0= \h.
For the frustum of a pyramid, the integral is to
be taken from x = h' to x = h. This would give
Xq

_ 3 '

hl
h'

- h"
h'1

If

Generated on 2015-06-02 13:19 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

the base of the pyramid becomes a circle, the


above equations will give the centre of gravity of
the cone, and of the frustum of a cone.

117. Centre of gravity of a spherical zone.

We

have from Geometry

dv 2zr . dx, and xdv

= 2nr xdx ;
.

whence

xdx
dx

or, integrating from x a to x r, and reducing,

If

0,

we have

ity of the surface of

#0

for the centre of grav

a hemisphere.

INFINITESIMAL CALCULUS.

263

118. Centre of gravity of a spherical

sector.
The spherical sector may be conceived as a sum of
equal infinitesimal pyramids having a common
vertex in the centre of the sphere, and an infini
tesimal base on its surface.
Each such pyramid
has its centre of gravity at a distance fr from the
centre of the sphere ; so that we may consider the
surface passing through all such centres of gravity
as forming a spherical zone with a radius fr.
Hence the centre of gravity of the whole spherical
sector will be found by substituting fr and fa for
r and a in the result of No. 1 17. Therefore
o

f (r + a).

a = 0, then x0 = f r will be the distance of the


centre of gravity of a solid hemisphere.

Generated on 2015-06-02 13:19 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

If

Moment of Inertia.
119. Let OA r (Fig. 49) be the radius of a
If its axis be hori
cylinder having a mass ra.
zontal, and a weight P be attached
Fig' 49
to the cylinder by a string wrapped
around its surface, the cylinder will
be caused to revolve about its axis.

Let be the angular velocity im


parted to every element of the mass
at the time t. Then any element
dm, whose distance from the axis is

OC,

will

have a velocity x

d&

-57- ,

and its quantity of movement will be dm.x

d&

IXFIXITESIiTAL CALCULUS.

264

P on

The accelerating action of

will be, at this instant, dm.x


with respect

such an element
and its moment

d'&

axis will be dm.x'

to the

Hence

on the whole mass


the moment of the action of
m will be the sum of all such moments (No. 106),
that is,

(afdm

dt

+ x'\lm' +

x"2d?n"

),

or, briefly,

dt
But the moment of the action of
pressed by

-nGenerated on 2015-06-02 13:20 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dt

Pr ;

J/

is also ex

hence

x dm = Pr, and

The quantity

rtdm

x'dm is called the moment

inertia of the mass m with respect


passing through its centre of gravity.
forming the integration we find

J* x'dm = ink',

of

to the

If

axis
by per

(1)

k is called the radius of gyration, inasmuch as the


movement will be the same as if the whole mass m
were collected together at the distance k from the
axis of rotation.

120. It is

remarked that the moment of


inertia varies in the same body according to tne
to be

INFINITESIMAL CALCULUS.

265

position of the axis of rotation. To investigate the


law of its variation, let ABGD (Pig. 50) be a section
of the mass m by a plane
Fig. 50
perpendicular to the axis of
rotation, O the point where
the axis is cut by this
plane, and O the point where
a
axis passing
parallel
through the centre of grav
ity of the mass is cut by the
same plane.
Considering an element dm of this
mass occupying any position E, and denoting OE
by x, OE by z, and OO by p, the triangle OOE will
give us the equation
a?

? + p*

2pz cos

OOE.

Substituting in (1), and separating the terms, we


have

Generated on 2015-06-02 13:20 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

mk*

P fdm +

p'dm 2

P pzdm

or, since the distance p is constant,

m&'=

J* z'dm + mp'

2p

and

cos

OOE,

Pdm = m,

dm.z cos OOE.

But z cos OOE OH the lever arm of the mass


dm with respect to the axis passing through the
centre of gravity of the body.
Hence
zdm cos OOE
is the algebraic sum of the moments of all the par
ticles of the body with respect to the axis passing
through its centre of gravity ; and this sum, by the

INFINITESIMAL CALCULUS.

266

of moments (No. 106) must be = 0, be


cause the moment of their resultant is also = 0.
principle

We have, therefore, simply


mk' =

Now

J*

z%dm

J'

z'dm-\-mp'.

is the moment of inertia of the mass

m with respect to the axis passing through the


centre of gravity.
Denoting it by m?02, we shall
have
mro

mk%

+ mp'.

(2)

Therefore, the moment of inertia of a body with re


spect to any axis is equal to the moment of inertia
with respect to a parallel axis through the centre
of gravity of the body, plus the mass of the body
into the square of the distance between the two axes.

Generated on 2015-06-02 13:20 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

121. Moment of inertia of a rectangle.

Let PQ

(Fig. 51) be an axis passing through the centre of


gravity of a rectangle, and
J7
lying in the plane of the
p
0
to
rectangle perpendicularly
its length AB. Let m be the
mass of the rectangle, and A
B
D
Q
AB = 2a. The infinitesimal
element CD = dm placed at any distance x from
the axis, will be found by the proportion

2a

dx : : m

dm ': or dm = ^ dx.
2a

Substituting in (1), we have

INFINITESIMAL CALCULUS.

267

and integrating from x = a to x = a, and reduc


ing,

Hence the radius of gyration is here k

This result is independent of the altitude of the

Hence considering the straight line AB


rectangle.
as a rectangle having an infinitesimal altitude, its
moment

of inertia will also be m

a?

-~

m denoting

the mass of the line.

Generated on 2015-06-02 13:20 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

122. Moment of inertia of a circle, when the


axis coincides with a diameter AB.
ss
Let OC r (Fig. 52) be the radius of
the circle, CD dm an element of
/^-^~~\o

its area parallel to the axis, OEx


its distance from the centre O. We /
I
shall have
itr*

2ydx v.m: dm, or dm

^ ydx.

Substituting in (1) and remembering


that y is = Vr' x', we have

By formula

(33) we have

and by formula (34),

/_

/\

Ey

268

INFINITESIMAL CALCULUS.

- arfdz = i Vr' -

(f

sin-1

x-

Substituting, and taking the integral from x =


to x = r, we have
mk' = m

-4

i*

Hence the radius of gyration is here k = g

133. Moment of inertia of a circle, when the


axis through the centre is perpendicular to the
With a
plane of the circle.
radius OC=x (Fig. 53) describe
circle, and give to its circum
Then 2nxdx
ference a width dx.
will represent an element of the
area, and thus

Generated on 2015-06-02 13:20 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

nr'

2nxdx::m

dm;

or
dm = r xdx.
1*

Substituting in (1), and integrating from x = 0 to


x = r, we have
mk1

= m

r*
g-

This formula is independent of the thickness of the


circular plate ; hence it will be true for any thick

ness. It therefore expresses the moment of inertia


of a solid cylinder of any length, revolving about

its axis.

INFINITESIMAL CALCULUS.
124. Moment of inertia of a circular ring with
Let
respect to an axis perpendicular to its plane.
r and r' be the extreme radii of the ring, and m its
mass.
Taking x between r and r', we have

* (r*

r")

2nxdx

: :

/ and dm

pi

Substituting in (1), and integrating from


x = r, we find
'

mk =

m r*
-7T

2 r*

r"

75

r"
m r*
+jr
2

a;

x^x-

= r'

to

This formula, being independent of the thickness


of the ring, will be true for a hollow cylinder of
any length.

and let BD be
an element perpendicular to
the axis of the cylinder, at
a distance OC = x from the
axis of rotation. Let r be
the radius of the cylinder,
and ft the mass of the element BD. The moment
of inertia of this element with respect to one of its
Q

cylinder,

r'

diameters would be (No. 122) = ft j-

but its mo

will be //

^-

ment of inertia with respect to the axis PQ parallel


to that diameter, and placed at a distance OC=x,
x'^j

Generated on 2015-06-02 13:20 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

125. Moment of inertia of a cylinder with re


spect to an axis perpendicular to the axis of the
cylinder. Let the axis PQ
(Fig. 54) be taken through
the centre of gravity of the

as we have

shown (No. 120).

INFINITESIMAL CALCULUS.

270

2a, or

/-<

m::dx

fi

Now, let 2a be the length, and m the mass, of


We shall have
the cylinder.
= ~- dx;

sion

2 ( J"

x'f

and therefore the moment of inertia of the whole


cylinder will be found by integrating the expres
d& between x a and x

a.

Integrating, and reducing, we shall find


nik'

=m

(?+!)-

126. Moment of inertia of a sphere.

Let the
axis pass through the centre of the sphere. Let

radius of the sphere and


the mass of an
elementary segment perpendicular to the axis,
placed at distance x from the centre, and having
radius y=Yr'x'.
The moment of inertia of
this element with respect to the axis
(No. 123)

Now,

if

01

'

11

is

m be the mass of the sphere, we have

whence

in the above ex
and substituting this value of
pression, we shall have for the moment of inertia
fi

Generated on 2015-06-02 13:20 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

ft

be the

INFINITESIMAL CALCULUS.

271

of the elementary segment


3m

8r

(r'-xjdx.

Integrating this from x


we get, after reduction,
mfc1

to x = r,

2mr'

for the moment of inertia of the whole


sphere.

Generated on 2015-06-02 13:20 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

127. Problem. As an application of


the theory of the moments of inertia,
let R and r (Fig. 55) be the radii of two

to

solid cylinders having the same hori


zontal axis, M being the mass of the
larger and m that of the smaller ; and
let the weights p and P be applied to
them respectively by a thread wrapped
on their surfaces.
If the two cylinders
are so connected that they must rotate
together, and if the weight p acts in a
direction opposite to that of P, what
will be the movement of these weights
after a time t ?
Solution. Adapting to our case the
formula (No. 119)

Pr

y x'dm

and reflecting that the moments of inertia of the

INFINITESIMAL CALCULUS.

272

two solid cylinders

M R

are

and m respect-

ively, and that the moments of the accelerating


forces are Pr and pR, we shall have

_ Pr-Rp

d'd
~
dV

Let now

MIt'

mf

Pr-Rp

MR

mr%

'

M'

and m' be the masses of the weights


P and p respectively. Then g being the action of
gravity, we have P = M'g, p m'g. And there
fore

dW_
~
dV

- Rm'
MR M'r

mr*

and integrating from t = 0 to


d&

dt

_~ 0,
9

M'r

Rm'

MR' -mr*

'

t= t,
9

'

MR -

M r - Rm'
M'r
mr*

'

Generated on 2015-06-02 13:29 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Curvilinear Movement.
point M (Fig. 56) cannot move in a
curve, unless its direcFig.se
tion
be
continually
changed by an action pro
ceeding from some other
Let us con
direction.
ceive this action decom
posed into two, X and
Y, respectively parallel
to the co-ordinate axes
OX and OY. Draw MN
normal, and M T tangent
to the curve at the point M.
Resolve X into MT

128.

A free

INFINITESIMAL CALCULUS.

273

tangential, and Mp normal, also Y into MT' tan


Then the
gential, and Mq normal to the curve.
forces X and Y may be replaced by a tangential
force TMT MT', and by a normal, or centri
= Mp -j- Jl/i?.
petal force
Now, calling # the angle that the element ds of
the curve at M makes with the axis of abscissas,

If

we have

MT- X cos
Mp =
Hence

0,

X sin

MT' = Y sin
Mq = Y cos

&,

#,

&.

T= Xcos#-Fsin#,

N= Xsintf + Fcostf.
But, according to our usual notation,

Y__ dty
dt'

W
d'x

Generated on 2015-06-02 13:29 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

sin
Therefore

dy

,,

dx

cos v -*-

__ d!#

<fa;

ds

&

= ~r

ds +
~

d'y dp

df

ds

'

df ds'

The first of these equations can be reduced to

_ df

rp_ d (dx' + dtf) ~ d ids') ~


_ 2dsd's
dt'. 2ds

-ids

2dsdf

_ d's
df

'

The second, being multiplied and divided by ds\


becomes

274

INFINITESIMAL CALCULUS.

."-'W
dtf

- d'xdy

tfydx

ds'

and this, according to the remark made by us on


the expression for the radius of curvature when t is
the independent variable (No. 46), will become

_ _ dS

Generated on 2015-06-02 13:29 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

df'R~

B'

Such is the expression of the centripetal accelera


tion. Hence the centripetal force is equal to the
product of the mass into the square of the tangen
tial velocity divided by the radius of curvature
The moving point, while obeying the centripetal
action, always keeps its tendency to follow a
straight line, that is, the tangential direction, and
thus to recede from the centre of curvature. This
centrifugal tendency, so far as counteracted and
thwarted by the centripetal action, is usually called
the centrifugal force, and its intensity is measured
by that of the action by which it is thwarted.
Hence the centrifugal force is equal to the centri
petal, and directly opposed to it. The centripetal
and centrifugal forces are commonly called central
forces.
(Fig. 57)

Y be

roll down a curve OMC


under the action of gravity.
Let OX and

129. Let a point

the co-ordinate axes, and let the ordinates


downward be positive. "When the moving point
has reached any position M, the action of gravity
MO, by which its movement is accelerated, may be
O

decomposed into MN normal, and MT tangential


to the curve.
The first will be destroyed by the

INFINITESIMAL CALCULUS.

275

resistance of the curve, which we assume to be in

variably fixed, and which is


thus playing the part of a
The sec
centripetal force.
ond will have its whole effect.
Let # be the angle that the
curve at M makes with the
axis OX. Then

MT =

g sin &

--

ds

Fig.

57

>

or
d's
~ y
dt'
ds
Hence
2ds d's

Generated on 2015-06-02 13:29 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dt

dt~

- %gdy, and

(!)"=*

This velocity is the velocity due to the height y


(No. 100). Therefore the velocity acquired by a
body rolling down a curve under the action of
gravity, is equal to that which it acquires by falling
freely through the same vertical height.
This result is true not only when g is constant,
but also when g varies according to a fixed law ;
for, even in ihis case, g may be regarded as con
stant from element to element, inasmuch as the
same law of variation applies to the elements of the
curve and to those of the vertical line. Hence a
body falling toward the sun on a spiral line will
have the same final velocity as though it had fallen
directly towards its centre.

130: The simple pendulum.

material point

INFINITESIMAL CALCULUS.

276

(Fig. 58) suspended from a horizontal axis by


a rigid line AO without
weight, and free to oscil
late about that axis, con
stitutes a simple pendu
lum.
Let AO = I, the angle
AOC^a, and the angle
When
the
MOC = &.
point A under the action
of gravity reaches the point
M, it will have acquired a
ds
=
velocity
^ V2gy (No. 129), y being = BE. But
ds -- ld&, and y

OE OD

I (cos &

cos a)

therefore

Generated on 2015-06-02 13:29 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

= V2gl

(cos # cos a)

whence

dd

V cos

J? cOS

Prom Maclaurin's formula we have


cos,? =

008

l- + or4-

0=1

-I+OT4-

'

'

hence, if the arc J. 6T be small enough to allow us


to neglect all the terms of the series after the sec
ond, we shall have cos # cos a = $ (a' #'), and

INFINITESIMAL CALCULUS.

Integrating from

a to

&

a,

277

we shall have

This is the time of one excursion, or semi-oscilla


tion, of a simple pendulum, when the amplitude of
the excursion is small; that is, not exceeding 9 or
10 degrees.

131. Planetary orbits.

By

the first of Kepler's

laws, the orbits of planets are ellipses, of which


one focus is occupied by the
SUn.

PMA

Let

Generated on 2015-06-02 13:30 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

Jlg..59

(Fig. 59) be a
the
planetary orbit, and
focus occupied by the sun.
Let the planet, at a given pr
v.
instant dt, be moving from
M towards the perihelion P.
Make FM = p, the angle

MFA =,

OQ

= x, MQ =

y
^

y,

OF

= c,

and let <p be the solar action at the unit of distance.


We shall have, from the inspection of the figure,

FQ

= p cos v = FO + OQ = c

MQ =

-f x,

sin v = p,

whence
cos

X
C-\p

sin

IXFISITESIMAL

278

CALCULUS.

The action of the sun at a distance p being ^


have also

_
d"x

_^co6B

dv

<c

we

(c + x) ,

-?-7

Multiplying the first of these equations by dx, and


by dy, adding the products, and re
marking that we have
the second

we find

dx (Fx
dy d'y _
(c + x) dx -+- ydy
dt ~df+dtW~~<f % | (c + xY+y^'

'

and integrating,

Generated on 2015-06-02 13:30 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

dx' +

or, since

dif
-f-

2y

<7y*

= JfiV"'

2y

ds*,

the velocity of the planet at the perihe


Calling
we have
lion P, and making PF =

Therefore

()'=--HH)-

(2)

INFINITESIMAL CALCULUS.

279

Calling V" the velocity of the planet at the aphe


lion A, and making FA p", we have

Therefore

and eliminating

ds

between (2) and (3), we obtain

V"' = V-2f

g>-~).

(4)

Now, by the second of Kepler's laws, the areas


described by the radius vector are proportional to
the times in which they are described.
Hence two
areas described in a second by the planet must be
equal ; and therefore V'p' = % V"p", whence
Generated on 2015-06-02 13:31 GMT / http://hdl.handle.net/2027/mdp.39015063880952
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

V=
This value of

V"

Py

substituted in (4) will lead to

But, if a and b be the major and minor semi-axes


of the orbit, we have
P'

+ P" =

2a,

p'p"

(a

c) (a

Substituting in (5),

ft'

+ c) = a'

& = V.

IyFISITESIMAL

280

CALCULUS.

Comparing equations (3) and (5), we obtain

(ds\ _

oir

4_

which, owing to (6), will be reduced to

(S'-*G-s)=>-!!erj-

*>

And now, from the general equation


ds'

= (pdvy + dp'

we have

by con

from which we shall easily eliminate

Generated on 2015-06-02 13:31 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

sidering that, since the area described by the


radius vector in the time dt is ty'dv, the area described in the unit of time will be ip*

But we

have already found that this area can be expressed

byiV'p".

Hence

{pdcy

\dt)
or substituting
and reducing,

_-

V^fT

the value of

p~

'

V from

equation (6),

INFINITESIMAL CALCULUS.
This being substituted in

(ds\_f
V#/

~ff

Now, eliminating
(7),

ds

we have

(8),

'

281

fdpy

\dt)

between

'

this equation, and

we shall find

\dt)

a'

p*

whence

-vl

(p*+a)dp

a)'

Vc'

If

-a)')

is,

c,

/>

A,

we

when

we reckon the time from the aphelion


=
= a+
that
shall have
when
a = c; and then C0.
Therefore

Vc*-(p-ay

(p-a)dp

adp

Vc'-(p-

- a/

'

(P

f\j

V&-(p-af

of which the integral

is

Such
the expression for the time taken by the
planet in measuring any portion AM of its orbit,
being the radius vector of the point M.

Generated on 2015-06-02 22:20 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

_ _ ./a

pdp

is

f'

dt=-\/~-

<p'

b'=a'

V2ap p' b'

c',

or, since

dt.

INFINITESIMAL OALCUL

282

US.

The time T employed in measuring the semiorbit


will be found by taking

AMP

(>

Then

(9)

=a

c,

or p a =

c.

gives
(10)

This equation shows that the squares of the times


of the revolutions of two planets are to each other

Generated on 2015-06-02 22:20 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

directly as the cubes of the transverse axes of their


orbits. This is the third among Kepler's laws.
The same equation shows also that the time of a
revolution is independent of the minor axis b of the
orbit. Assume b so small, that the ellipse may
sensibly be reduced to a double straight line.
Then the focus will sensibly coincide with the ex
tremity P of the transverse axis, and T would be
the time taken by the planet in falling directly
upon the sun from the distance AP = 2a.

From

(10) we have

T
/a- =

Hence (9) may

be transformed into

and if

be the eccentricity of the orbit, we may

substitute

for

and reduce the equation to the

form

INFINITESIMAL CALCULUS.

283

Lastly, making

COS #

(11)

(& +

sin

ft'

the equation will take the form


(12)

will

e5

cos

(13)

a~l-

is

The formulas (11) and (12), and the polar equa


tion of the ellipse, which

to determine the time taken by the


reckoned
planet in measuring any given angle
from the aphelion, when the eccentricity of the
v

suffice

is

is

2.

is

Generated on 2015-06-02 22:20 GMT / http://hdl.handle.net/2027/mdp.39015063880952


Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google

is

orbit
known.
The eccentricity of the terrestrial
Scholium.
=
orbit being
0.016833, and the earth, during the
tropical year, measuring only 359 59' 9".8 around
the sun, the longitude of the aphelion (which
reckoned from the vernal equinox, and which on
the 1st of January, 1800, was 99 30' 8*. 39)
in
The length of the
creasing every year by 50".
= 365".242256. Will the student,
tropical year
with these data, and with the aid of the last three
formulas, try to determine the length of the four
seasons for some given year?