Infinite Series by Osgood

/v\.
i! tt.
STAT.
UBRAIt
INTRODUCTION
TO
INFINITE SERIES
BY
WILLIAM
F.
OSGOOD,
Pir.D.
ASSISTANT PROFESSOR OF MATHEMATICS IN HARVARD UNIVERSITY
CAMBRIDGE
b$ Ibarparfc
1897
"Ulniversitp
Copyright, 1897, by
HARVARD UNIVERSITY.
fft
fl/ffti
PKEPACE.
TN an introductory course on the
Differential
and Integral Calculus
the subject of Infinite Series forms an important topic.
presentation of this subject should have in view
first to
make
The
the
beginner acquainted with the nature and use of infinite series and
secondly to introduce him to the theory of these series in such a
that he sees at each step precisely
what the question
never enters on the proof of a theorem

actually requires proof.
till
at issue is
way
and
he feels that the theorem
Aids to the attainment of these ends are
(a) a variety of illustrations,
taken from the cases that actually arise
in practice, of the application of series to computation both in pure
and applied mathematics (b) a

meaning and scope of the more
;
full
diagrams and graphical illustrations
The pamphlet
that follows
the kind here indicated.
The
is
and careful exposition of the
difficult
theorems
(c)
the use of
in the proofs.
designed to give a presentation of
references are to Byerly s Differential
Calculus, Integral Calculus, and Problems in Differential Calculus ,
and to B. O. Peirce
Ginn
&
Short Table of Integrals;
all
published by
Co., Boston.
WM.
CAMBRIDGE, April 1897.
F.
OSGOOD.
LNTKODUCTIOISr.
1.
Consider the successive values of the variable
Example.
sn
for
1 -f-
r2
-f-
-f~
rn
~1
Then
Let r have the value J.
1, 2, 3,
s2
*3
If the values
-|-
=
=
+J
- +4+i
1
be represented by points on a
S,
line, it is
Sa
FIG.
1J
if
easy to see the
S,
S4
2.
1.
law by which any s n can be obtained from its predecessor,

lies half way between * H _ 1 and 2.
namely
Hence it appears that when n increases without limit,
:
The same
result could
formula for the sum
-f-
s n of
ar
1, r
tl
=.
2.
have been obtained arithmetically from the

the first n terms of the geometric series
-\-
ar 2
-\-
a (I
-
When
/t
Lim
Here a
sn _
.s
-\-
ar n-1
r)
J,
w increases without
and hence as before Lim
SH
limit,
=. 2.
^ approaches
as
its
limit,
INTRODUCTION.
2.
any
an
Definition of
2.
Let M O w 1? w 2
Infinite Series.
Denote the sum of the
Allow n
.....
+M +
2
n terms by
first
sn
be
series
(1)
Then
to increase without limit.
.....
,
and form the
set of values, positive or negative or both,
either a) sn will
approach
U:
a limit
Lim
n
or b) s n approaches no limit.
Infinite Series, because n
is
SH
U;
co
In either case we speak of (1) as an

allowed to increase without limit. In
case a) the infinite series is said to be convergent and to have the

In case b) the infinite
or converge towards the value U.
value*
series is said to
be divergent.
series above considered
The geometric
vergent
is
an example of a con
series.
1
1
.....
.....
+ 2 + 3+
+ 1-1
are examples of divergent series.
Only convergent
series are of use
in practice.
The notation
uo
is
+u +
.....
often used for the limit
U
*
C7,
inf. (or to infinity)
or simply
+u +
.....
often called the sum of the series.

But the student must not forget
not a sum, but is the limit of a sum. Similarly the expression "the sum
of an infinite number of terms" means the limit of the sum of n of these terms,
that
as
is
7 is
n increases without
limit.
I.
a)
3.
ALL OF WHOSE TERMS ARE POSITIVE.
SERIES,
Let
Example.
CONVERGENCE.
it
be required to test the convergence of the
series
where n\ means 1-2 3
....... n
moment
Discarding for the
the
first
and
read
is
factorial
term, compare the
sum
n".
of the
next n terms
19ft
IZIZO
1.9
9
lÔ
with the corresponding
ft
sum
n
^L<2
(Cf.
Each term
of
term in
and hence the sum
S.n ,
a- n
or, inserting the
after the first
factors
1).
two
is
less
than the corresponding
discarded term and denoting the
terms of the given series by
sum
of the
first
SM ,
no matter how large ?i be taken. That is to say, s n is a variable

that always increases as n increases, but that never attains so large
a value as 3.
To make these relations clear to the eye, plot the
successive values of s n as points on a line.
CONVERGENCE.
3, 4.
= +
= + + 1-
s2
*3
=2.
=2.5
l-fl
$4
*5=l
+ jL + JL
+ +
l
*6
+ f7
111
^j-
j _i_
]_
_i
s,
2.667
2.708
-9717
718
When
~4T ~5l
~2~!~^3~!
n increases by 1, the point representing s H + 1 always moves

but never advances so far as the point 3.
Hence there
to the right,
must be some point

(i.e. e
<C
3),
which
e,
sn
3 or
either coinciding with
approaches as
its
lying to the
left
of 3
limit, but never reaches.
To
the value of e to
8
judge from the values computed for s 1? s 2
three places of decimals is 2.718, a fact that will be established
<s
later.
4.
The reasoning by which we have

e, although we do not as yet
FUNDAMENTAL PRINCIPLE.
here inferred the existence of a limit
know how to compute the numerical value of that limit, is of prime

importance for the work that follows. Let us state it clearly in
general form.
If s n
is
a variable that
1)
always increases wlien n increases:
but that 2) always remains less than some definite fixed number,
s tt
for
all
values of n, then
sn
<
approaches a
Lim
fs
lt
limit,
U.
U:
A:
CONVERGENCE.
4, 5.
This limit, U,
is
not greater than
U<A.
s3 s 4 U
H+H
sa
s,
1
The value
A may be
A
h
the limit itself or any value greater than the
limit.
Exercise.
creasing, but
State the Principle for a variable that is always de

is always greater than a certain fixed quantity, and
draw the corresponding
figure.
Direct Comparison.
On the prin
based the following test for the
First Test for Convergence.

of
the preceding paragraph
ciple
convergence of an infinite series.
5.
series
test.
If a
less
in the series to-be tested (a), then (a) is

does not exceed that of the series ((3).
to
than the corresponding terms
a convergent
and
series,
its
value
.....
<V-i,
Lira
n
it
a)
let
Then
of positive terms, the convergence of which it is desired

series of positive terms already known to be convergent
can be found whose terms are never
For
.....
+u +u +
Let
be
is
Sn
since
follows that
and hence by
<^
sn
A.
QO
and
sn
Sn
<^
sn
<C
Sn
approaches a limit and this limit
is
not greater
than A.
Remark. It is frequently convenient in studying the convergence

of a series to discard a few terms at the beginning (m, say, when
is & fixed number) and to consider the new series thus arising.
That
the convergence of this series is necessary and sufficient for the

convergence of the original series is evident, since
CONVERGENCE.
5, 6.
u is constant and hence s n will converge toward a limit

and conversely.
Examples. Prove the following series convergent.
.....
+ +r + +
++-^
^
^
5
3
r
Write
<
does,
1
<
3-4
2-3
1-2
Solution.
r 16
if s n _ m
in the fo-in
+!
then
Lim
4.
1^
T^
22
A New
6.
g2
Test-Series.
^2
.....
+ +
l+
o-6
8 4
sn
>
has just been seen that the series
It
converges when the constant quantity

also converges whenever
inequalities is at once evident
that
it
>
>
We
2.
The
will
I+1
2?
III
4;
I_~ I-~
P
P
T~
5p
~r
QP
"^
3*
2
~
^ _ = J_
2
<-
2*
!<-!^^ 4/
7
,
;>
now prove
truth of the following
i<r^P
/)
JL
^-i
"-
(5,
CONVERGENCE.
7.
Hence, adding
of these inequalities together,
I+QP +
+ /O
-.
OP
rt
1
1
<
^-*-
\/>
"
1
-I
}V)
by
O^J
1
^>
converges toward the limit
many terms
Ap
Qp
L- 41
1- H
J
then, since
get
^ -f -JL +
+)p
_i_
Denote 1/2
we
0, r
r -f r 2
I
and the
series
<^
r 3 4-
-4I
Consequently ho matter how
of the series
2~P
+ 3; + 47 +
be taken, their sum will always be less than
and
this series is
4.
therefore convergent, by the principle of
Series (3) is useful as a test-series, for many series that could not
be shown to be convergent by the aid of the geometric

For example,
be so shown by reference to it.
1
7.
f)
Divergent Series.
The series (3) has been proved convergent

Thus the series
^>
2^2
a convergent series, for
7
p
r
4
1.01.
Now
consider what the nu
merical values of these roots in the denominators are
In fact
^ 2 = 1.007, ^ 3 = 1.011, ^ 4
^ 100 = 1.047 and ^ 1000 = 1.071;
thousand terms of the

last
term
is
can
+_l_
+ ^_+
+ _J_
9
OA/O
4:y4:
zyz
for every value of
is
series,
1.014.
that
is,
when a
have been taken, the denominator of the

a
number
so slightly different from 1 that
multiplied by
series
significant figure of the decimal part appears only in the

second place. And when one considers that these same relations will
be still more strongly marked when p is set equal to 1.001 or 1.0001,
one may well ask whether the series obtained by putting p
1,
the
first
1.
is
not also convergent.
CONVERGENCE.
This
is
11
I
n+l
since each of the
we can
For
however not the case.
?i
n terms, save the
last, is
111
^V.
+2n
7, 8.
+ n^
f^\
2n~~2
greater than 1/2 n. Hence

add a definite number of
strike in in the series anywhere,
terms together and thus get a sum greater than

this as often as
we
and we can do
For example,
please.
i+\>\
j+i+t+t>j
"=
Hence the sum
++
the
of
first
+>
n terms increases without
limit as
increases without limit,
Lim
or
The
series (4) is called the
How
sn
oo
oo
harmonic
series.
the apparently sudden change from convergence for p

1 to be accounted for?
in series (3) to divergence when p
is
^>
The
When p is only slightly greater than 1,

explanation is simple.
series (3) indeed converges still, but it converges towards a large
value, and this value, which
is
of course a function of p, increases
without limit when p, decreasing, approaches

limit exists,
8.
and the
Test for Divergence.
Exercise.
namely
When p
5 for convergence,
Let
o
+w +
of positive terms that is to be tested for divergence.

series of positive terms already known to be divergent
be
no
1,
Establish the test for diver
gence of a series corresponding to the test of

:
1.
series is divergent.
series
a)
If a
(ft)
can be found whose terms are never greater than the corresponding
terms in the series to be tested (a), then (a) is a divergent series.
Examples.
,
J_4.
j_
+
^
CONVERGENCE.
8, 9.
4_ i
_i_
i _L I
JL
_L_
This last series can be proved divergent by reference to the series
Let
""
""
"""
111
+ + +
246
sn
+ v~
2n
series in parenthesis is the harmonic series, and its sum in

hence S H increases without limit
creases without limit as n increases
The
and the
series is divergent.
The
Second Test for Convergence.
9.
Test- Ratio.
Let the series
to be tested be
+ Ul +
U
and form the
When n
definite
Then
test-ratio
increases without limit, this ratio will in general approach a

fixed limit (or increase without limit).
Call the limit r.
if r
the series is convergent,
1
<^
\ there is
+ Un+
no
test
"
+1
un
*>
<^
"
^>
"
First, let r
1.
<^
to the values of
ifr^>
1,
it
is
divergent, if
Then
as
1,
Convergent ;
Divergent;
No
Test.
n increases, the points corresponding

about the point r, and hence if
u, + 1 /u n will cluster
l
FIG.
y
4
4.
a fixed point y be chosen at pleasure between T and 1, the points

u n + l /u n will, for sufficiently large values of n, i.e. for all values of n
equal to or greater than a certain fixed number m,
of y,
and we
shall have
lie
to the left
CONVERGENCE.
10
n =-
Adding p
<"
The sum
"H+3
"
>
-j- z,
<\
we
of these inequalities,
2
2
y 4- y 4- y
^w + 3
^ u
m + iy
um y
O
i
get
4- y p )
-I-
<^
w
1
beginning with w w + i never, therefore,

V rises as high as the value u m Hence the ^-series converges.
The case
and may be
If r
the terms w,
of
that T
^>
1 (
or T
treated in a similar manner,
oo ) is
an exercise for the student.

there is no test.
For consider series
left as
(3).
The
test-
ratio is
-
UH
and hence T
1,
(n-f- 1)
may have. But when p
no matter what value p
^>
1,
Thus it appears
1, (3) diverges.
converges; and when p
that T can equal 1 both for convergent and for divergent series.
Remark. The student will observe that the theorem does not say
(3)
<^
that the series will converge
than
when n
u u + i/ u n snall b e
series this ratio
if
u + l /u
lt
increases, but that

I
Thus
GSS than 1.
n/(n
-f- 1) is
less
lt
it
becomes and remains less

demands that the limit of
in the case of the
than
But the limit is not less than 1.

Are the following series convergent or divergent?
the series diverges.
Examples.
harmonic
for all values of w, and yet
i.?.
+ 5^3
+
3^3
i.
11
100
3
7
28
2-2
A + Jl +
2ioo
3100
2*
2S
4.
i
4100
+ J*!
+ j*L +
100 n 100
2
.....
CONVERGENCE.
10, 11.
10.
Further Test-Ratio Test.
gence and divergence

If
omitted
is
11
The following
sometimes useful
test for
conver
the proof of the rule is
approaches a
converges, if
limit, let this limit
be denoted by
a-.
Then
the series
CONVERGENCE.
12
The following
series
11
serve as an example.
may
+ ~ +
Let
Proof.
and plot the points

points s l5 s 3
s5 ,
st , s2 ,
s3 ,
.....
*2/
..... Then we shall show that the

..... always move to the left,
+u
i-
FIG.
but never advance so far to the
5.
left as s 2 , for
4 they approach a limit,
the principle of
Lira s 2m +
m = QO
t/i
Hence by
example.
t/i.
.....
.....
s 2/)i ,
Similarly, the points s 2 , s 4 , ,s 6
always move
to the right, but never advance so far to the right as s 1? for example
,
hence by the same principle they also approach a
Lim
.j
m.:
oo
Finally, since
2w
lim
WI
but lim
1(2,,,
=n
,s
2ra
+1
"T~
s 2m
m
a
*II
2ra
+ lim
m
?6 2m
<x>
here the third hypothesis of the theorem comes
hence
__
-
Ul
Thus
m=oo
QO
TT
over
=:
+ 1 z= lim
into play for the first time
or simply U.
2//t
limit,
U2
approaches a
- --
s rt
its limit.
TT
limit,
llll
FIG.
C7,
continually springing
vHi^
1
<
6.
It remains to supply the

the reasoning of the proof.
this reasoning depends.
on
which
establishment
of
the
facts
analytical
Such
is
-2
2m _
and the parentheses are
all
2w
_l
positive (or null)
Next
s zm
The proof
13
CONVERGENCE.
11, 12.
is
now
+l
>
and
s2
s 2w
sx .
<
complete.
Examples.
log 3
log 2
Iog4
12. The Limit of Error in the Alternating Series.

Suppose it be
required to find the value of series (5) correct to &, say, to 3 places
of decimals.
For
this
purpose
it
is
not enough to
know merely
that the series
converges, and hence that enough terms can be taken so that their
sum s n will differ from the limit
by less than .001, for n might be
so great, say greater than 10,000, that it would be out of the question
to compute s n
And in any case one must know when it is safe to
.
stop adding terms.

The rule here is extremely simple. The sum of the first n terms
of series (5), sn differs from the value of the series, U, by less than
,
the numerical value of the (n -\- l)st term.

stop adding terms as soon as we come to a
In other words, we may

term which is numerically
smaller than the proposed limit of error.

The transition from
For, consider Fig. (6).
s n to s n + 1 consists in
the addition to sn of a quantity numerically greater than the distance
from s n to U. This quantity is precisely the (n -{- l)st term of the
series.
Hence the
For example,
rule.
let it
be required to compute the value of the series
I-I.I+ .I- 4 .1+

1
32
(7 )
34
33
correct to three places of decimals.
=
=
=
=
(i)*
()
(W
.3333
i (|) 3
5
I (^)
.0123
.0000
\
4
\ (i)
6
| (|)
.0008
0556
.0031
.0002
.3464
.3464
or, to 3 places, the value of series (7)
*
The 4th place is retained throughout the

place in the final result.
is
.0589
.2875
.288.*
work
to insure
accuracy in the third
14
CONVERGENCE.
Show
1.
Examples.
12, 13.
that the value of the series

_
~"~
4 2*
to three places of decimals is .405.

2. How many terms of the series
-4+i-J
would have
to be taken that the
sum might represent
the series correct to 3 places of decimals
13.
any
convergent series
General Theorem.
the value of
Let
.
be
of positive,
and negative
Lim w n
More
oo
generally,
Lim [X
71
where
Then
terms.
is
any
The proof
+!
.....
+ W-i] =
CO
integer, either constant or varying ivith n.
of this theorem flows directly out of the conception of a
Let
limit.
and plot the points
we say
"s
which the
.....
s 1? s 2 , $,,
approaches a limit
U"
is
Then what we mean when
that there
is
a point I/ about
as n increases.
This does not necessarily re

quire that (as in the series hitherto considered) s n should always come
s n s cluster,
Thus s 3 may lie further away

steadily nearer to U, as n increases.
than s 2 does. But it does mean that ultimately the s s will
from
-- U -a
s4
U-(?
1
sa U+rf
Mill
cease to deviate from
/t
s,
U by more
than any arbitrarily assigned quan

In
other
words, let 8 be taken at pleasure
tity, 8,
an interval extending to a dis
and
off
(= 1/1,000,000, say)
lay
then for the
tance 8 from U in each direction, (U
8, U -f- 8);
n
all
of
values
of
more
values
for
n,
larger
greater than
precisely,
however small.
a certain fixed
number m,
s n will
lie
within this interval.
can be stated algebraically in the following form
8
<
sn
<
(7+8,
when n
>
m.
This
CONVERGENCE.
13, 14.
Having thus stated what
we now
If
n
^>
The
is
meant by
turn to the proof of the theorem.
m, both
sn
and s n+p
15
sn s
approaching a limit U,
The sum
wi\l lie in the interval
distance between these points
is
(U
8,
therefore less than 2
U-{-
8).
Hence
8.
-28
no matter what value p may have. But if a quantity depends on n
and can be made to remain numerically as small as is desired by
Thus
as its limit, when n
increasing ?i, then it approaches
the proposition is established.

It is to be noticed that while the condition
if
the series
to converge,
is
is
it
in
no wise
Lim u n
==.
is
<x>
necessary,
sufficient for the conver
series (4) the general term approaches

The harmonic series
the series diverges.
however does not satisfy the more general condition of the theorem
gence.
Thus
harmonic
in the
as its limit, but
still
for if
we put p
n,
and does not converge toward
new proof
It
may
as its limit.
This fact affords a
of the divergence of the harmonic series.

be remarked that the more general condition
Lim [u n
where
p may
vary with n in any
ivise
u e choose,
dition for the convergence of the series.
14.
Convergence.
The General Case.

ô
be any series and
+u +
is
a sufficient con
See Appendix.
Let
.....
( a)
let
%+
.....
*>!+
denote the series of positive terms,
the series of negative terms, taken respectively in the order in which

For example, if the it-series is
they occur in (a).
CONVERGENCE.
16
then the v-series
is
i
and the
14.
w-series
+p+p+
is
~_
03
T5
Let
Then, whatever value n may have,

Sn
Here
etc.
When n
limit,
sn
can be written in the form
Vm
TP
denotes the number of positive terms in S B

m their sum,
increases without limit, both
increase
without
andp
<r
and two cases can arise.

I.
Both m and rp approach
Case
<r
Lim
tn
a-
limits
Lim
T7
",
QO
rp
TF;
cc
so that both the ^-series and the ^y-series are convergent.
Hence
the ^-series will also converge,
Lim
n
and
U,
oo
U=
W.
The above example comes under
this case.
Case
I will
be of
principal interest to us.
Case
II.
At
least
one of the variables
For example, suppose the

3,
1
As
it
-r
w-series
_.
3
<r
m , TP
approaches no
limit.
were
_.
-5
1
these examples show, the u- series
33
T^
may
7
I
then be convergent and
may be
divergent.
Exercise.
Show that
if
the w-series converges and one of the
w-series diverges, the other must also diverge.

Let us now form the series of absolute values* of the terms of the
v-,
*
By the absolute value of a real number is meant the numerical value of that
3 is 3; of 2^ is 2.
number. Thus the absolute value of
Graphically it
means the distance of the point representing that number from the point 0.
CONVERGENCE.
14, 15.
and write
^-series
uH
If
will
we
it is
this series as
be a certain
un
v, if
is
positive
a certain w,
if
un
is
negative.
set
clear that
= *. +
<
From
is
17
this relation
a convergent
we deduce
at
*,-
once that in Case I the w -series
series.
Conversely, if the u -series converges, then both the v-series and
and ice have Case I.

For both the v-series and the w-series are series of positive terms,
and no matter how many terms be added in either series, the sum
Hence by
cannot exceed the limit U toward which s n converges.
the w-series converge,
4 each of these series converges.

whose absolute value series
the principle of
are convergent
said
to
be
u
are
-series converge)
whose
absolutely or
are
series
said to be
other convergent
unconditionally convergent
Definition.
(i.e. w-series
Series
The reason
not absolutely convergent or conditionally convergent.
and
for the terminology unconditionally
conditionally convergent will
appear in
Since the it-series surely converges

and
then absolutely convergent
Test for Convergence.
15.
if
34.
the
tt
-series
is
it
converges
is a series made up exclusively of positive terms,

the tests for convergence obtained in I. a) can be applied to the
tt -series and from its
convergence the convergence of the it-series can
since the ^ -series
The series that occur most frequently in elementary

analysis either come under this head and can be proved convergent
in the manner just indicated, or they belong to the class of alternat
thus be inferred.
ing series considered in
The
test of
9 can
11.
be thrown into simpler form whenever the
test-
the rule being that when t is

approaches a limit,
numerically less than 1, the series converges absolutely; when t is
numerically greater than 1, the series diverges ; when t is numerically
ratio u n + \/u n
equal
to 1, there is
no
test
(
*
t
<^
or
<^
<^
^>
Convergence;
or
Divergence
No
Test.
CONVERGENCE.
18
15, 16.
For, the test-ratio u n + l /u n is numerically equal to the test-ratio of

the series of absolute values, u n + i/u n
Now if a variable f(n)
a
n
when
its
numerical
oo,
limit, H,
approaches
value, being the
.
distance of the point representing /(n) from the point 0, approaches

a limit too, namely the numerical value of
from 0)
(distance of
Hence
lim
*>
u n+1
un
where T equals the numerical value of t. If then

follows that r
1 and the u -series converges.
<^
t
<^
The
<^
1, it
^-series is
then absolutely convergent.

The second part of the rule will be proven in the next paragraph.
Example. Consider the series
~
Xs
X2
2"""
X*
~~
""
"3
n
n
Hence the
e. when
JU
oo
converge when x
series will
__
_
--
Lim
n
xn
is
numerically less than
i.
-iO<i.
When
x =.
or
this test fails to give
any information concern
ing the convergence of the series. But it is then seen directly that in
the first case the series is convergent, in the second case, divergent.
16.
To
Divergence.
establish the divergence of a series
ô
with positive and negative terms,
divergence of the
-series, as the
shows.
It will
however
.....
+%+
1
it
is
not enough to establish the
example of the
series
+ 3 ~~ 4
suffice to
show that the terms do not approach
as their limit, and this can frequently be done most conveniently

by showing that the terms of the ^ -series do not approach 0.
Thus
and
if
or
>1
uw+1
.
>
1,
*<--!,
when
then
1
>
>
m.
1(5,
CONVERGENCE.
17.
Hence
"
that
the
is, all
quantity p
limit,
if
<
>
uu
or
^w
>
>
u n and u n cannot approach
as their
oo.
In the series of
Example.
= m on are greater than a certain positive
from
u m and hence
when n
19
15,
a;;
hence
verges for all values of x numerically greater than

may be represented graphically as follows
this series di
These results
Divergent
Divergent
Convergent
For what values of x are the following series conver

what values divergent? Indicate these values by a diagram
Exercise.
gent, for
similar to the one above.
x*
"
.
|
V~2
V3
Ans.
a;
<C
10 x
17.
-\-
x5
X*
10 2 a 2
+ +
x
THEOREM.
1
<^
>.2
-fa;
Conv.
^>
1,
1,
<^
Div.
1
,
10 3 x 3
-|-
a;
Ze^
absolutely convergent series; p
pt
.....
,
/o 2
Then
quantities not increasing numerically indefinitely.
a ^2/
se ^
the series
converges absolutely.
For,
let
an p
,
be the absolute values of a

>t
/a rt
respectively, // a
and form the
M,
positive quantity greater than any of the quantities p

series
20
CONVERGENCE.
The terms
17, 18.
of this series are less respectively than the terms of the
convergent series
Ha
and each
.....
+ Ha\ + //a +
2
made up exclusively of positive terms.

converges and the series
Hence
series is
first series
the
Examples.
The
series
since
sin 3
~~^~
converges absolutely for
all
32
converges absolutely and sin

If
-f-
-f- a-!
and
cos x
6 X sin
nx
never exceeds unity numerically.

an( l
^1
^2
.....
are
series, the series
+a
-f- 6 2
a?
series
.....
+ I-T
5
any two absolutely convergent

a
For the
x.
.....
tf
~~5*~
values of
I. -1
2.
sin5#
a;
"
cos 2 x
SU1 2a; -f-
.....
.....
-f-
converge absolutely.
3.
Show
that the series

e"*
cos x
-f-
e~ 2;r cos 2x
.....
-\-
converges absolutely for all positive values of x.

4. What can you say about the convergence of the series
1
+r
cos
cos 2
.....
-f
18. Convergence and Divergence of Power Series.

ascending integral powers of a;,
where the
called a
aj,
but
others.
coefficients
power
it
series.
will in general
In the
a l5 a 2
Such a
series
series of
may converge
for
all
of #,
is
values of
converge for some values and diverge for
latter case the interval
distances in each direction

Divergent
.....
..... are independent
a o 4~ a i x H~ a x2
<2
from
of convergence extends equal

x =. 0, and the series con-
the point
Direr gent
Convergent
verges absolutely for every point x lying within this interval, but not
necessarily for the extremities of the interval.
CONVERGENCE.
18.
21
Let x be any value of x for which the

a
terms of the power series n x^ do not increase without limit a n a?
Then a n aj is less than
the absolute values respectively of a M x
The proof
as follows.
is
"
some fixed positive quantity (7, independent of n,

a,
the power series may converge and
For #
all
it
may
Let h be any value of x numerically less than x

value.
Then
where r
7i
the power
/a;
values of n.
for
diverge.
its
numerical
h.
For
Hence the terms of the absolute value
series
series converges absolutely for x
<C
+ a\h + a
.....
are less respectively than the terms of the convergent geometric series
C + Or
and the
.....
Or 2
series
-\-
a2 h 2
From this theorem it follows that
.....
-\-
if
the power series converges for
converges absolutely for all values of x within an interval

from
to x and reaching out to the same distance on the
stretching
and if diverges for x
other side of the point x
x^ it diverges
x =.
a*
it
x lt If
for all values of x lying outside of the interval from x t to
now the series ever diverges, consider the positive values of x for
which it diverges. They

x
r, where r in general
a region extending down to a point

and such that the series
greater than
fill
is
converges absolutely for all values of x numerically less than r

this is what was to be proved.
and
simpler proof of this theorem can be given for the special case
For then
oo
that a n + 1 /a n approaches a limit, L, when n
Lim
n
or
=. Lx.
lutely for
all
absolutely
is
GO
u ni __
a
Hence when
values of x
when x
is
lim
n
L=
a
a xn
co
0, the
while
power
if
L ~\-
series
converges abso
0, the series
converges
15)
(
numerically less than 1/Z/, and diverges when x
;
numerically greater than 1/L.
This proves the proposition.
II.
SERIES AS
a)
19.
One
A MEANS OF COMPUTATION.
THE LOGARITHMIC SERIES.
most important applications of
of the
infinite
series in
analysis, and the one that chiefly concerns us in this course, is that
of computing the numerical value of a complicated analytic expres
sion, for
example, of a definite integral
like
when
the indefinite integral cannot be found.

In fact, the values of
the elementary transcendental functions, the logarithm, sine, cosine,
Let us see how a table
etc., are computed most simply in this way.
of logarithms can be computed from an infinite series.

series for the function log e (1 -f- li) can be obtained as follows.
Begin with the formula
The function
(1 -|-
can be represented by the geometric series
a?)"
Integrate each side of this equation between the limits

fly.
/h
_
1
a-
f*h
/Vt
/
clx
f*h
xdx
Jo
Jo
+ Jo
x*dx
and h
log
(l+/
7i
In deducing the above formula
h
+
it
-=---
.....
Evaluating these integrals we are led to the desired formula

2
.....
(8)
has been assumed that the theo
integral of a sum of terms is equal to the sum of the

Now
integrals of the terms can be extended to an infinite series.
an infinite series is not a sum, but the limit of a sum, and hence the
rem that the
extension of this theorem requires justification,
v.
39, 40.
SERIES AS A MEANS OF COMPUTATION.
19, 20.
23
Obtain the formula
Exercise.
1
tan"
Hence evaluate
.....
+ ---
the series
-J+4- .....
12 the value of series (8) was computed
J and h
J and it thus appears
20. In the examples of
to three places of decimals for h
that
log 1J
To
find looO 2
.287 (5),
we could
log 2
But
U=
log
.405 (5).
the value h
substitute in (8)
\ /
this series is not well
+k
.....
adapted to numerical computation.*
In
fact to get the value of log 2 correct to the third place of decimals, it
would be necessary to take 1000 terms.
simple device however
Write
makes the computation easy.

2
= 14
and then take the logarithm of each side
= log | + log
= .287 (5) + .405
Hence, to three places, log 2 = .693.
log 2
Next, to find log

different
way, for
5.
.693 (0).
series must be applied in still a

h be set equal to 5, h =. 4, and the series
therefore set
Here the
if 1 -f-
We
does not converge.
(5)
= 4 + l=4(l.+ J),
log 5 = 2 log 2 + log 1J
= 1.386 (0) + .223 (2) =
5
where log 1J
From
is
computed
directly
from formula
1.609 (2),
(8).
the values of log 2 and log 5, log 10 can at once be found.
log 10
log 2
+ log 5 =
.693 (0)
1.609 (2)
2.302 (2)
or to 3 places,
log 10
of great importance, for its value must be

compute the denary logarithm from the natural
This latter logarithm
known
*
in order to
The formula
We
is
2.302.
is
nevertheless useful as showing the value of a familiar
could not find by direct computation the value of this series

say, seven places, because the work would be too long.
series, (6).
to,
24
20, 21.
logarithm.
By the formula for the transformation of logarithms
from the base c to the base 6,
"
Iog c 6
we have
Hence
logio^
for
log,
log, 10
example
Compute
Examples.
log 20,
Iog 10 20,
log 9,
Iog 10 9,
log 13,
Iog 10 13.
pu pose well when only a

Suppose however we wished to
know log 2 correct to 7 places of decimals. Series (8) would then
In fact, it would require 16 terms of
give less satisfactory results.
Series (8) is thus seen to serve its
21.
few places of decimals are needed.
the series to yield log 1 J to 7 places.

From (8) a new series can be deduced as follows.
Then
(8)
Let h
x.
becomes
log (1
Next replace h
x)
in (8)
log(l
by x
+*)=:
x-1
xs
+ *_| +| -
Subtracting the former of these series from the latter and combining
the logarithms we get the desired formula
:
We have subtracted on the

We have not we have limits
;
in
right hand side as if we had sums.

of sums.
This step will be justified
35.
We
now apply series (9) to the determination of log 2 to seven

x must be so chosen that
will
places,
zn 2,
i.e.
11_
=$
lill
_ + __ +
and
21.
25
The advantage of this series over (8) is twofold first, it suffices to

J, and
compute the value of the series for one value of &, x
second, the series converges more rapidly than (8) for a given
value of JB, since only the odd powers of x enter.
:
(-i)
Q)
3
5
(I)
=
=
=
(|)
9
(|)
11
(I)
13
(|)
15
(|)
=
=
=
.333
333
33
.037
037
04
.004
115
23
.000
457
25
.000
050
005
81
000
000
63
.000
.000
.000
The term TV
(|)
8
(|)
(i)
7
823
05
32
.000
065
.000
005
65
11
.000
000
51
.000
000
05
.000
000
00
.346
573
59
(i)
-
(i-)
=
=
13
TV
T^
07
68
.000
33
(i)
|
TV
65
333
345
=
=
|
i
.333
:= .012
(|)
15
(i)
16
has no effect on the eighth decimal place.

But
not enough to justify us in stopping here.
must show
that the remainder of the series from this point on cannot influence
(-J)
We
this is
Now
this place either.
the remainder
A ~ JL
J_
15 3 15
The value
nor
is
_!_
-1
19
3 19
--
3 15
15
_! ~
17 3 17
is
17 3 2
19
the series
~
34
of the series in brackets cannot be readily determined

it is obviously less than the value of the
;
that necessary, for
from it by discarding the coefficients if, ^|,

than the geometric series
series obtained
i.
e.
1,1,
+
3*
32
and hence the remainder
1
=
~*
in question is less
r^î
~~
s
than
119
15
3"
and so does not
We
affect the eighth place.

obtain then finally for log 2 the value
.346
573
or to seven places
log 2
5(9)
.693
.693
147
147
2.
1(8)
etc.,
26
Show
Examples.
that
=
=
log 1J
log 5
Compute
log 2
by
log 2
.223
143
(4)
1.609
437
(8).
aid of the formula
log 2
Knowing
21, 22.
log
= -- log |
and log 5 we can

log 10
log J.
find log 10
2.302
585.
Compute
Example.
log
Iog 10 9
to six places.
Series (9) is thus seen to be well adapted to the computation of

If y denote any positive number and x be so deter
logarithms.
mined that
1+3 =
r-
"
*=FFT
1 and -\- I and series (9) converges

then x always lies between
towards the value log y. For values of y numerically large the
convergence will be less rapid and devices similar to those above
explained must be used to get the required result.

In the actual computation of a table, not all the values tabulated
few values are computed

are computed directly from the series.
in this way and the others are found by ingenious devices.
b)
THE BINOMIAL SERIES.
(a
22. In elementary algebra the Binomial Theorem

exponent
l
m(
m
l a m-* b *
am
ma m -^b
is
established.
integral
+ b)
for a positive
_|_
(to
-\- 1
terms)
Consider the series
If
/A
is
a positive integer, this series breaks off with ^

this point on, each numerator contains
then, from
Thus
if /A
1 -\-
2,
-^-
terms, for
as a factor.
we have
-"2x
-\- 1
1*2
x2
-)
x8
1
-\- etc.
(subsequent terms
all
0),
SERIES AS A MEAXS OF COMPUTATION.
22.
or simply
2x
-J-
-f-
x2
In this case the series
son with the binomial formula (a

value (1
a;)**:
is
x,
27
seen by compari
=.
to
/A)
have the
any number not a positive integer (negative num

never breaks off, i.e. it becomes an
infinite series.
Let us see for what values of x it converges, for
The general term of
only for such values will it have a meaning.
however
If
/x
is
ber, fraction, etc.) the series
the series
is
?T~
~.
Hence
/ji
UH + I
1)
(/A
(/A
(n -\- 1
Ti-l-1)L
(M
1)_
/u,
1)
-j-
n
i**(i*<
1
(/x,
x*
fji/n
and
Lim
71
00
all values of x numerically less

For
x
the
values
1 special investiga
1,
15.)
(
which
we
will
not
into
here.
necessary,
go
Consequently the series converges for

than unity.
tion
is
We may
Divergent
__
Convergent
note in passing that when

series, a fact that
_ _
Dirergenl
<^
becomes an alternating
is
is
<^
the series finally

when the series
useful
used for computation.
Toward what value does the series converge when x lies between
1 and
1 ?
The answer to this question is as follows For all
values of x for ivhich

(1
the
binomial series converges,
its
value
is
*)":
(1
The proof
III).
Let us
a.-)"
==
+ M* +
^- +
**
(10)
of this theorem will not be considered here

(v. Chap.
first see whether the series is of
any value for the pur
poses of computation.
Example I. Let it be required to compute
places.
.....
^J
35 correct to
five
28
We
by the
must throw the radicand into a form adapted to computation

series.
We do this as follows. Since 2 5 32 we write
^
The second
(i
22.
35
factor can be
off,
series.
+ f^^ (A) +
703 + .000 040
.000
(A)
.000
003
1.018 08(4)
breaking
+ ^)s.
(1
.018 750
and
Show
we
Exercise.
computed by aid of the
+ *)* == + i A +
1
35
2.036 17.
that in the above computation

did, with the fifth term.
as
we
are justified in
ffi
would be com
Example II. Find -\J 15 to five places.

Here we have a choice between the expressions
15
and
15
In the
first
case (1
8+ 7=2
= 27 12 = 3
(1+S)
zz
in the
-(- J)fc,
(1
second
-*)
(1
puted by aid of the series. In practice however there is no question

as to which expression to use, for the second series converges more
rapidly than the first.
1.
Examples.
2.
Show
that
Complete the computation of ^ 15.

2.080 09
and
$9
^/ 2000
3.
Compute
then by writing
first
by
->/
letting
^ =.
4.
Find
5.
Obtain from (10) the following formulas
i,
by any method.
2 to five places
l+x
1
2x
1
X2
3.T
,
"
2^1
2-4
~l~
1-3-5
2-4-6
2-4-6
zz=
2.961 94.
i
;
23, 24.
29
The Computation of
23. Series for sin~ l li and tan~ l h.
The method set forth in 19 is applicable to the representation
1
1
^ (v. Exercise,
and
19) by series.
sin"
dx
T^
.tan- 1
/*
This
these series the value of

set
1,
we
is
h*
TT
If in series
not well adapted to computation.
obtained by putting h
(12)
can be computed.
i in
series (11)
-4
~~
6
.....
4
o
get the equation
series, like series (6), is
better series
h8
+ 23+^4
=/i
From
we
of
tan"
/*-
(12)
TT.
This series yields readily three or four places of decimals
but
if
more elaborate methods are necessary.

262
(v. Jordan, Cours d Analyse, Vol. I,
1893).
Exercise.
If the radius of the Earth were exactly 4000 miles, to
how many places of decimals would you need to know ir in order to
is
greater accuracy
desired,
compute the circumference correct to one inch?

number of places by Jordan s method.
24.
The Length of
Arc of an
the
the ellipse be given in the form
Then
will
sin
Determine
TT
to this
Let the equation of
Ellipse.
<
6 cos
<
the length of the arc, measured from the end of the minor axis,
be
/*
a
Jo
where (a 2
b 2)/a 2
& si
The integral that here presents itself

1.
as an Elliptic Integral and its value cannot be found in the
usual way, since the indefinite integral cannot be expressed in terms
is
<
known
of the elementary functions.

last
sin 2
can however be obtained by

for x in the
substitution of
esin<
22 gives the formula
example of
1
Its value
The
the aid of infinite series.
<
sin 2
<
-i-
sin 4
<
30
Hence
40)
(v.
24, 25.
<
/"0
sin 2
e
<<i4>
/*0
I
Jo
"i
sin 4
d>(7<f>
These integrals can be evaluated by the aid of the formulas of IV of

Short Table of Integrals.
In particular, the length of a
will
S
be
found
quadrant
by putting
^ TT and using the formula
240
of
the
(No.
Tables)
Peirce
<
sm"<ad>
1-3-5
2-4-6
Jf
The
elliptic integral
1)
(?i
z= -
_
n
TT
97,
an even integer,
then becomes the integral known as the Complete

Second Kind it is denoted by
Elliptic Integral of the
ft
Jo
(No. 248 of the Tables).
sm 2
<
d<f>
Hence
a#.
the ellipse reduces to a circle and S

^-n-a.
Examples. 1. Compute the perimeter of an ellipse whose major
axis is twice as long as the minor axis, correct to one tenth of
If e :=
one percent.
2. A tomato can from which the top and bottom have been removed
is bent into the shape of an elliptic cylinder, one axis of which is
Find what size to make the new top and

twice as long as the other.
If the original can held a quart, how much will the new
bottom.
can hold ?
25. The Period of Oscillation of a Pendulum. It is shown in

Mechanics (v. Byerly s Int. Cal., Chap. XVI) that the time of a
complete oscillation of a pendulum of length
=4*
is
given by the formula
where a denotes the
initial inclination of the
pendulum
to the vertical.
known
as the Complete Elliptic Integral of the First Kind and

The substitution of fcsin< force
its value is computed as follows.
is
in the series for (1
2
o?
)~5 gives the formula (v. Exs.,
22).
31
25, 26.
Integrating and reducing as in
we
24,
obtain the formula
If the angle through which the pendulum oscillates is small, an

sufficiently accurate for most purposes will be
approximation for
obtained by putting k
the usual
Then
0.
K z=
ITT
and
pendulum formula.
Exercise.
less
Show
that
if
a
<
5,
this
approximation
is
correct to
than one tenth of one percent.
APPROXIMATE FORMULAS IN APPLIED MATHEMATICS.
c)
26.
It is often possible to replace a complicated
mathematics by a simpler one which
is
still
formula in applied
correct within the limits
of error of the observations.*
The
of Expansion.
Coefficient
sion of a solid
is
meant the
By
the coefficient of linear
expan
ratio
V
I is the
length of a piece of the substance at temperature
The coefficient of cubical expansion
the length at temperature t
is defined similarly as
where
n__V
~
P
where F,
V stand for the volumes
-V
-
at temperature
respectively.
Then
as
is
we
at once clear if
of an edge being
is as follows.
at t.
13
consider a cube of the substance, the length

The accurate expression for a in terms of ft
^/
(3
/3
P+
See Kohlrausch, Physical Measurements,
1-6.
32
26, 27.
the error made by

Since (3 is small,
usually less than .0001,
neglecting all terms of the series subsequent to the first is less than
the errors of observation and hence we may assume without any loss
of accuracy that
a =:
j^ /:?
/?
:= 3 a
Double Weighing.
Show that if the apparent weight of a body
in one scale pan is p lt when placed in the other scale
when placed
pan, p 2 (the difference being due to a slight inequality in the lengths

of the arms of the balance) the true weight p is given with sufficient
accuracy by the formula
,
iCPi +P*)In an experimental determination of
27. Errors of Observation.
a physical magnitude it is important to know what effect an error in

an observed value will have on the final result. For example, let it
be required to determine the radius of a capillary tube by measuring

the length of a column of mercury contained in the tube, and weigh
From the formula
ing the mercury.
W
where
of the
column
TTl^lp
denotes the weight of the mercury in grammes,

in centimetres, p the density of the
the length
mercury (== 13.6),
and r the radius of the tube, we get

I
Jw
TT
pi
to
I
Now
the principal error in determining r arises from the error in

I
Let I be the true value, I
observing /.
-f- e the observed value
of the length of the column

puted value of the radius.
from the error of observation
assumed
negligible.
e,
the error in observing
Since
is
first
small
term
iv
being
Hence
x
only the
r -\r the true value, r

the com
is the error in the result arising
Then
we
get a result sufficiently accurate
and hence, approximately,
by taking
27, 28.
33
Thus
for a given error in observing Z, the error in the computed value

inversely proportional to the length of the column of mercury
a result not a priori obvious, for r itself is inversely propor
used,
of r
is
tioned only to
An
Exercise.
1"
surveys a
engineer
field,
using a chain that
is
of one percent of its length.

Show that the
error thus arising in the determination of the area of the field will be
two tenths of one percent of the area.
by one tenth
incorrect
clock regulated by a pendulum is

28. Pendulum Problems.
If it is carried to a
located at a point (A) on the earth s surface.
h
feet
above
the
level
of
point
neighboring
(-B),
(A), show that it
will lose
^T
seconds a day,
li
i.
e.
one second for every 244 feet of
elevation.
The number
of seconds
N that
the clock registers in 24 hours is

T of the oscillation of the pen
inversely proportional to the period
dulum.
Hence
25)
(cf.
N
where the unprimed
to (B).
letters refer to the location (-4), the
If the clock
was keeping
where
(+
117.)
If h does not
letters
86,400.
/t
)a>
(Cf Byerly s
.
Hence
Z
first
N=
denotes the length of the radius of the earth.
Diff. Cal.,
the
primed
true time at (A), then
exceed 5 miles, li/E
<
NR
.001. h 2 /fi 2
+h
<
.000 001, and
N correct to seconds
term of the series gives N
N N = th h
Examples.
the sea level.
1.
The summit
How many
of Mt.
Washington
is
6226 feet above
seconds a day will a clock lose that keeps
accurate time in Boston Harbor,
if
carried to the
summit of the
Mountain ?
2.
pendulum that beats seconds on the surface of the earth
is
observed to gain one second an hour when carried to the bottom of a

How deep is the mine?
mine.
34
29.
Exercises.
1.
Show
29.
that the correction for expansion and

is given by the formula
contraction due to heat and cold
where a denotes the
43, 200 a
coefficient of
t,
linear expansion,
the rise in
temperature, and n the number of seconds lost in a day.

For brass, a
.000 019, t being measured in degrees centigrade.
Thus for a brass pendulum n
.82
and a rise in temperature of
causes the clock to lose a
over 4 seconds a day.

man is standing on the deck of a ship and his eyes are h ft.
2.
above the sea level. If
denotes the shortest distance of a ship
little
away whose masts and rigging he can

measured
to him, 7^ the height,
the water,
show
If h
1.23
(V
whose
hull is invisible
which the hull
rises out of
can be neglected,
that, if refraction
see, but
in feet, to
miles
>i)
16 ft,,
10 miles (nearly).
that an arc of a great circle of the earth, 2J miles long,
recedes 1 foot from its chord.
h,
3.
Show
4.
Assuming that
the sun s parallax
tance of the sun from the earth

5.
earth
6.
Show
is
is
76, prove that the dis
about 94 million miles.
is
8".
that in levelling the correction for the curvature of the

How much is it for two miles ?
8 in. for one mile.
The weights
of an astronomical clock exert, through faulty

when the clock
construction of the clock, a greater propelling force
has just been wound up than

increase the amplitude of the
when
it
has nearly run down, and thus

2 to 2 4 on each side
pendulum from
Show that if the clock keeps correct time when it

has nearly run down, it will lose at the rate of about .4 of a second
a day when it has just been wound up.
of the vertical.
7. Two nearly equal, but unknown resistances, A and jB, form

two arms of a Wheatstone s Bridge. A standard box of coils and
a resistance x to be measured form the other two arms.
A balance
is obtained when the standard rheostat has a resistance of r ohms.
When however A and B are interchanged, a balance is obtained
when
the resistance of the rheostat
proximately,
x
8.
The
focal length
/of
b(r
a lens
is
is
+r
).
PI
ohms.
Show
given by the formula
--- + /
P.
that,
ap
29.
p and p 2
where
35
denote two conjugate focal distances.
simpler approximate formula for
that will answer
Obtain a
when p and p z
are nearly equal.
ranchman 6 feet 7 inches tall, standing on a level plain,

How much must
agrees to buy at $7 an acre all the land in sight.
he pay? Given 640 acres make a square mile." Admission Exam,
"
9.
in Sol.
Geom., June, 1895.
Show
assumed the altitude of the zone

ranchman s eyes above the
ground and had made no other error in his solution, his answer would
have been 4 cents too small.
that
the candidate had
if
in sight to be equal to the height of the
Show
10.
that for small values of
approximately correct (h
li the
following equations are
be either positive or negative)
may
+ h) m =
= + 2h
(1
Hence
(1
+ h)
+
(1
7i,
&,
+ 70
Z,
p,
(1
.....
fc)
(1
are
all
mh
1
+h= +
l>
+h
numerically small, then, approximately,
.....
= +H+ + +
i
= +h+k
1
(1+p)
V
(i
V
If
1 -f-
lc
.....
,
TAYLOR S THEOREM.
III.
30.
It is not the object of this chapter to
prove Taylor s Theorem,

done satisfactorily in any good treatise on the Differ
ential Calculus
but to indicate its bearing on the subject under con
sideration and to point out a few of its most important applications.
It is remarkable that this fundamental theorem in infinite series
since this is
admits a simple and rigorous proof of an entirely elementary nature.

Rolle
Theorem, on which Taylor
of the
Mean
From
Rolle
lie
Theorem follows
Theorem depends, and
the
Law
at the very foundation of the differential calculus.
at once the theorem contained in the
equation
J
(13)
This latter theorem

with the
is
frequently referred to as Taylor
R =/
Remainder
lt
(x
Theorem
It includes the
-\-OJi)
Law
Mean
of the
/(a-o
+ h)
/(.TO)
hf
(x
Oh}
(14)
as a special case and thus affords a proof of that Law.

If in (13),
when n increases indefinitely, n converges towards
as its limit,
the series on the right hand side of (13) becomes an infinite power
series, representing the function f(x -\- Ji) throughout a certain
region about the point XQ
f(x
+ h) =
This formula
is
/(**>)
known
+ f (x )h
as Taylor s
h2
+f"(x )
Theorem and
.....
(15)
the series as Taylor s
Series.
The value X Q is an arbitrary value of x which, once chosen,

The variable x is then written as x -\- h. The object
fast.
is
as follows.
is
held
of this
It is desired to obtain a simple representation of the
terms of known elements, for the purpose of com

One of
of
value
the function or studying its properties.
the
puting
with
known
series
coefficients.
is
a
forms
such
of
the simplest
power
function f(x)
in
TAYLOR
30, 31.
Now
it is
by one and the same

and even when this is possible, the
a:,
converge rapidly enough for large values of the argu

Consequently we confine our
series will not
ment
37
usually impossible to represent f(x)
series for all values of
power
THEOREM.
to be of use in computation.
attention to a limited
domain of values, choose an x
in the midst of
domain, and replace the independent variable x by
this
The values
XQ ~\-
of x for the
values of h will be, h
chosen that /(o),/
(o?
ll
7i,
XQ.
in question may not be small, but the

XQ
If x is so
corresponding to x
ad inf. are all finite, then
),/"(ie )
i.
e.
for values of h
numerically small, will usually* be given by Taylor
example
will aid in
making
Let
f(x)
Then it is at once
Theorem for x
Theorem.
above general statements.
clear the
log x.
clear that f(x) cannot be developed by Taylor s

oo
It is just at this
log
for/(0)
0,
we have
point that the freedom that

good stead; for if we take x
will all
/"(XQ),
where
domain
the value of f(x) for values of x near to x0l
An
7i,
be
in the choice of
greater than
finite
and/(a;
-f-
x stands us
in
x
0, then f(x ),
( o),
can
be
developed by
h)
Theorem, the series converging for all values of h lying

Taylor
x
The proof is given for x
between x and
1 in the Diff.
s
Thus we have a second proof
130.
CaL,
log (1
31.
7i),
Two
(formula (8) of
(13).
Any
series.
19).
Applications of Taylor
This theorem,
it
will
of the development of
Theorem with
be observed,
is
the
Remainder,
not a theorem in infinite
function whose
expressed in the
first n derivatives are continuous can be

form (13), while the expression in the form (15)
requires the proof of the possibility of passing to the limit
qo
when
Thus
(13) is a more general theorem than (15) and it avoids the

It is because of the applica
necessity of a proof of convergence.!
tions that (13) and (15) have in common, that it seemed desirable to
treat
*
some applications of (13)
Exceptions to
this
rule,
here.
though possible, are extremely rare in ordinary
practice.
f It is desirable that (13) should be applied much more freely th;m has
hitherto been the custom in works on the Infinitesimal Calculus, both because
affords a simple means of proof in a vast variety of cases and because many
proofs usually given by the aid of (15) can be simplified or rendered rigorous
by the aid of (13). The applications given in this section are cases in point.
it
TAYLOR
38
THEOREM.
31.
First Application: Maxima, Minima and Points of Inflection;

Curvature.
Let it be required to study the function f(x) in the
=X
+/
neighborhood of the point x
+ h) = f(x
f(x
Plot the function as a curve
Q.
(*o)
J/"(a>
Oh)
li\
*
:
and plot the curve
The
latter
curve
nates, y l and
?/ 2
is
Consider the difference of the ordi-
a right line.
y*
i/"(a
0ft)
ft
Hence it appears that ?/!

?/ 2 is an infinitesimal of the second order.
This property characterizes the line in question as the tangent to the
curve in the point oj and thus we get a new proof that the equation
,
of the tangent
is
)+f (x )(x
y=f(x
).
Next, suppose
Then
f(x
The equation
(2
it
"
== f(x
of the tangent
will in general
(cc)
is
+ h)
positive
at
this
is
+ f^
(x,
+ Oh) -^-^
now
# and
be continuous near the point #
it will therefore be positive in the
point
neighborhood of this point and hence

ya
Ik
>
both for positive and for negative values of 7i, i. e. the curve lies
x
above its tangent and has therefore a minimum at the point x
it
can be shown that
Similarly
vatives vanishing, f(x) has a
Lastly,
(xo)
*
if
(2
"
(a; )
maximum
<
0, all
in the point
let
=0,
The student should
/ (2
illustrate
(ô)
0,
each case in
/(
this
the earlier deri
+ 1)
(*o) =j= 0.
by a figure.
TAYLOR
31.
Then
{2
y,
+ 1)
"
will
(aj)
- =
y,
THEOREM.
+
(.*
/<*
39
fc2n+l
ft)
(2n+1)
be continuous near a?
x and it will
same sign for small values of 7i, positive or
changes sign with h. Hence the curve lies on
tangent on opposite sides of the point x and
in general
therefore preserve the

+1
but
negative;
opposite sides of its
this is then a point of inflection.
7r"
Exercises.
1.
Show
that the condition for a point of inflection not
parallel to the o;-axis is

/"(%)=0,
"
+ 1)
)=b
/->(aR D
being continuous near
(#)
35
a?
a
1
"
/<
+1
>(o
b )=|=
0,
Show
that a perpendicular drawn to the tangent from a point

is an infinitesimal of
infinitely near to a point of inflection
2.
.....
higher order than the second.

The osculating circle was defined (Diff. Col. 90) as
Curvature.
a circle tangent to the given curve at
and having its centre on the
We
inner normal at a distance p (the radius of curvature) from P.

and a
be taken infinitely near to
will now show that if a point
perpendicular
FM be dropped from P
P
the osculating circle at P", then

of the third order referred to the arc
Let
on the tangent at P, cutting

in general an infinitesimal
is
P"
PF
as principal infinitesimal.
be taken as the origin of coordinates, the tangent at

being
and let the ordinate
the axis of x and the inner normal the axis of y
Here
y be represented by the aid of (13).
;
0,
x=h,
y =
and
The radius
/(0)
$f"(0)x*
of curvature at
P is
and the equation of the osculating

*
Hence the
circle is
2
(y
p)
P-
lesser ordinate y of this circle is given
y r= p
*?
P (1
*
by the formula
:
Instead of the infinite series, formula (13) might have been used here, with
But we happen to know in this case that the function can be developed
4.
by Taylor
Theorem
(15).
TAYLOR
40
and
THEOREM.
31, 32.
=
)
K*^*>:-*?From
this
result
follows
that
y )/x approaches in general

a finite limit different from 0, and hence that ?/
y is an infini
tesimal of the third order, referred to
x as principal infini
tesimal.
PM
But
proposition.
Exercise.
Show
(y
PM
and
PP
are of
the
same
order.
that for any other tangent circle y
Hence the
y
is
an
infinitesimal of the second order.
Second Application : Error of Observation

Let x denote the magni
tude to be observed, y
the
f (x)
magnitude to be computed from
the observation.
Then if .r be the true value of the observed magni
.
tude, x =. x -\- h the value determined by the observation, h will be

the error in the observation, and the error // caused thereby in the
result will be (cf. (14))
H = f(x + h)
f(x
= f (x
+0h)h.
In general / (#), will be a continuous function of x and thus the

value of f(x
Oh is
Oh) will be but slightly changed if a?
replaced by x.
Hence, approximately,
H = f (x)h
and
this is the
formula that gives the error in the result due to the
error in the observation.
The Principal Applications of Taylor s Theorem without the

Remainder, i. e. Taylor s Series (15) consist in showing that the
1
fundamental elementary functions e x sin a, cos a-, logo;, o^,
SB,
32.
sin"
1
tan"
a;
can be represented by a Taylor
explicitly the coefficients in these series.
Series,
It is
and
in
determining
shown in Ch. IX of
the Diff. Cal. that these developments are as follows.*
2!
xs
z5
5!
3!
COS
99=1
These developments hold for

*
The developments
x4
x*
all
values of x.
for sin" a; and tank s are to be sure obtained by in

tegration; but the student will have no difficulty in obtaining them directly
from Taylor
Theorem.
TAYLOR
32, 33.
log*
-i
l
tan~"
a?
log(l
* 4.
"
+ h) =
XS
2^4
THEOREM.
1-
-J-
These developments hold for all values of 7i

last two formulas, of a?) numerically less than
Exercise.
Taylor
-4-
-I-
2 3
41
Show
(or, in the
case of the
1.
that sin a; can be developed about any point X Q by

that the series will converge for all values of h.
Theorem and
Hence compute
sin 46
correct to seconds.
33. As soon however as we pass beyond the simple functions and

Theorem, we encounter a difficulty that is
In order namely to show that f(x) can be
usually insurmountable.
expanded by Taylor s Theorem it is necessary to investigate the
general expression for the n-ih derivative, and this expression is
try to apply Taylor s
To avoid this difficulty recourse

usually extremely complicated.
methods
of obtaining the expansion.
more
or
less
indirect
is had to
For example,
let it
be required to evaluate
ra
/
The
indefinite integral
do;.
cannot be obtained and thus we are driven to
develop the integrand into a series and integrate term by term. Now
x
e~ r )/x^
if we try to apply Taylor s Theorem to the function (e
the successive derivatives soon
ever proceed as follows
become complicated.
We
= -x+
X1
and hence, dividing through by
.T,
X
- 3T
+
s
we have
3l
5l^
.....
can how
TAYLOR
42
dx=2
Examples.
Do
(l
THEOREM.
+ 3 37+5^7+
33.
made up
functions, into a
502-
the examples on p. 50 of the Problems.
General Method for the Expansion of a Function.

function /(a?),
= *.H4
power
To
develop a
manner out of the elementary

the general method is the following.
in a simple
series,
The fundamental elementary functions having been developed by

32, we proceed to study some of the simplest
Taylor s Theorem,
operations that can be performed on series and thus, starting with
the developments already obtained, pass to the developments de
sired.
ALGEBRAIC TRANSFORMATIONS
IV.
OF SERIES.
has been pointed out repeatedly (
19, 21, 24) that since
not a sum, but a limit of a sum, processes appli
if applicable, this
cable to a sum need not be applicable to a series
34.
an
It
infinite series is
fact requires proof.

For example, the value of a
which the terms are added.

terms be extended to series?
value
Its
less
is
terms as follows
i
The
than
-)-
(12)
+ * - i + i + i- * -M + A -J +
general formula for three successive terms
if
the result
it is
is
its
Rearrange
4 A;
.....
(i
2k
+ I) - i + (i + A)
an alternating
(-8)
is
each pair of positive terms be enclosed in parentheses
+ i) - i +
(i
For
4k
and
sum is independent of the order in

Can this interchange in the order of the
Let us see. Take the series
(y)
series of the kind considered in
11).
easy to verify the inequalities
Hence the
converges toward a value greater than (1 -)- ^)

of the first n terms of (/?) differs from a properly
chosen sum of terms of (y) at most by the first term of a parenthesis,
series (y)
The sum
J z= J.
a quantity that approaches

as its limit when n =.
Hence
the series (/?) and (y) have the same value and the rearrangement of
<x>
terms in
from
(a)
has thus led to a series
(/?)
having a different value
(a).
In fact
it is
possible to rearrange the terms in (a) so that the
have an arbitrarily preassigned value, C. For, if

positive, say 10 000, begin by adding from the positive terms
series will
new
is
ALGEBRAIC TRANSFORMATIONS OF SERIES.
44
enough have been taken so that
till
their
sum
34, 35.
will just
exceed C.
This will always be possible, since this series of positive terms

Then begin with the negative terms
diverges.
and add just enough
to reduce the
sum below
As
C.
soon as this
has been done, begin again with the positive terms and add just
enough to bring the sum above (7; and so on. The series thus
obtained
value
is
is
the result of a rearrangement of the terms of (a) and its
C.
In the same
way
it
can be shown generally that
"o
"i
+w +
2
if
.....
any convergent series that is not absolutely convergent, its terms

can be so rearranged that the new series will converge toward the pre-
is
Because of
assigned value C.
such series are often called
this fact
35 justifying the denoting

conditionally convergent, Theorem 1 of
of absolutely convergent series as unconditionally convergent.
There is nothing paradoxical in this fact, if a correct view of the
For a rearrangement of
nature of an infinite series is entertained.
terms means a replacement of the original variable s n by a new vari
able s n in general unequal to S H and there is no a priori reason why
,
these two variables should approach the same limit.

The above example illustrates the impossibility of extending a
Most of such
priori to infinite series processes applicable to sums.
processes are however capable of such extension under proper restric

and it is the object of this chapter to study such extension for
some of the most fundamental processes.
tions,
35. THEOREM
In an absolutely convergent series the terms can
1.
be rearranged at pleasure without altering the value of the series.
suppose
First,
sn
all
the terms to be positive and let
=u + u +
.....
_!
After the rearrangement
U.
lim s n
co
let
For s n always in
approaches the limit U when n
creases as n increases but no matter how large n be taken (and then
held fast), n can (subsequently) be taken so large that s n will include
Then
sn
all
the terms of
>
and more too
therefore
<x>
35.
45
no matter how large n be taken,
or,
U.
,
<
Hence
We
ated
n,
approaches a limit
may now
U.
turn things about and regard the i/-series as gener

of the u -series, and the above
by a rearrangement of the terms
reasoning shows that
<^
U =
hence
U.
q. e. d.
The second step in the above proof was abbreviated

Exercise.
by an ingenious device. Replace this device by a direct line of
reasoning.
Secondly,
let
the series
.....
+U +U +
2
be any absolutely convergent series and
Let
let
.==.*-*
u v w.
+ u\ + u .....
S n
V and
1
25
22
28
THEOREM
W;
hence
+ 12
let
U.
series
TX
Find the value of the
III
21
"
M/
u
Exercise.
14)
be the series after the rearrangement and
But
(Cf.
I+
29
2*
J_
2 11
2^
2.
F=
V,
.....
are any two convergent series, they can be added term by term, or
If they are absolutely convergent, the third series will also be abso
lutely convergent and hence its terms can be rearranged at pleasure.
Let
Then
S
n+
U
<
46
When
n =.
oo
hand side converges toward
the left
U+ V = (u +
It
Ul
+ vJ +
V\ hence
.....
may be
remains to sliow that the parentheses
U -\-
35.
dropped.
shown in the same way as in the case which arose in 34.

The proof of the second part of the theorem presents no
and may be left to the student.
Exercise.
Show
that
U
is
any convergent
THEOREM
3.
This
is
difficulty
if
series, c
?*!
.....
+u +
z
any number,
If
U=
V=
u
v
+n +
+v +
l
?<
.....
.....
r 2 -f
are any two absolutely convergent series, they can be multiplied together
like sums; i.e. if each term in the first series be multiplied into each
term in the second and the series of these products formed, this series
toward the limit UV. For example
will converge absolutely
U Vl
"Mô
+U
VZ
%!?!
+U
VQ
.....
This theorem does not hold for series that are not absolutely con
vergent.
Let
.....
then
lim
t
it
n_
UV.
oo
The terms
of the product s n t n are advantageously displayed in the

scheme.
following
They are those terms contained in a square n
terms on a side, cut out of the upper left hand corner of the scheme.
+.*&
I
The theorem asserts that if any series be formed by adding the

terms of this scheme, each term appearing in this series once and
35.
47
for example, the terms that lie on the oblique lines, the
only once,
successive lines being followed from top to bottom
:
Vo + Vi +
+V +
MIÔ
.....
(a)
converge absolutely toward the limit UV.

show that one series formed in the prescribed way
from the terms of the scheme, for example the series formed by fol
lowing the successive boundaries of the squares from top to bottom
this series will
It is sufficient to
and then from right to
namely the
left,
series
converges absolutely toward the limit UV. For any other series
can then be generated by a rearrangement of the terms of this series.
Let
Sy denote
First suppose
the
Then,
positive.*
sum
of the first
N terms in
(/?).
the terms of the w-series and the v-series to be
all
if
n2
Hence
<
lim
N=
(n
JV<
I)
UV.
oo
the ^-series and the v-series are any absolutely con

vergent series, form the series of absolute values
Secondly,
if
The product
But
of these series
is
the convergent series
this series is precisely the series of absolute values of (a),
therefore (a) converges
value toward which
it
UV.
is
converges
and
remains to show that the
It
absolutely.
Since
$v
approaches a
increasing, passes through all integral values, Sx will

continue to approach a limit, and this will be the same limit, if
limit
when N,
2
passes only through the values n
lim
7V=
But
tlt
Sy
=
n
co
lim
and
s n tn
tl2
oo
lim
n
/S, l2
UV.
oo
This proves the theorem.

*
The
case that
the double sign
some of the terms
in
are
must not however be excluded
the inequality below:
S.y
fC s n +
it, l
+ i.
hence
48
For example,
35, 36.
let
+ a^x +
f(x)
be two convergent power
a.2
x2
.....
-\-
x any point lying at once within the
series,
Then
region of convergence of both series.

series is given by the formula
the product of these
This formula can be used to give the square, or by repeated appli

cation,
any power of a power
Thus
series.
it
gives as the square of
the geometric series
+ x + x* +
+ 2x + 3x* +
l
the series
l
.....
.,
a result agreeing with the binomial expansion of (1 -}- #)~ 2

Exercise.
Find the first four terms in the expansion of
e~ x s\ux COBX
V
Square the
expansion of
x
series for e
los (I
and
2
and show that the
x)
-\-
+v
result agrees with the
*.
36. One more theorem is extremely useful in practice.

would carry us beyond the bounds of this chapter.
Its proof
Let
be any polynomial in y and

series in x
:
Then
the powers of y
+ c^x + a
2
y be given by the convergent power
let
?/
x2
.....
8
,
.....
n
y can be obtained at once as
x by repeated multiplications of the ^-series by itself,

power
then formed by multiplying these
the terms of the polynomial
(?/)
the
cofficients 6, and the polynomial
series
power
respectively by
series in
<
/;
<f>
n (//)
thus represented as a power series in x by the addition of these
terms.
Suppose however that instead of the polynomial

infinite series
Under what
<(//)
power
as a
we had an
</>(?/)
restrictions
series in
power
x?
series in
can
the
above process of representing
x be extended to representing
as a
<f>(y)
06.
One
restriction
immediately obvious.
is
49
Since a power series
represents a continuous function (v.

38) the values of y corre
small
values
x
will
to
of
lie
near
to
and thus the point a
sponding
must surely lie within the interval of convergence (
r
y
r) of
Suppose a
now our theorem
the series
<f>(y).
And
satisfied.
condition
is
<^
<^
is
then this condition
is
always
precisely this, that no further
necessary.
THEOREM
If a
4.
the above process
0,
no further restriction is necessary; i.e.

(y) as a power series in x is always
of representing
<f>
applicable.*
Remark. The point of the theorem just quoted is this. We know

from 35 that each term in the y series can be expressed as a power
series in x
bn y
=f (x) = a w + Oi
and hence that
<
(?/)
#(30
+ ajt* +
.....
can be expressed in the form
=/o (*).+/i ()+/(*)+ .....
remains to prove (and it is precisely this fact that the theorem as

a fact not true in general of a convergent series of the form
serts,
It
+ /i(a)-h/t(*) +
/o(*)
where
fn (x)
series all the
denotes a power series) that if we collect from these

terms of common degree in x and then rearrange them
form of a single power
in the
and secondly,
Examples.
its
series, first, this series will converge,
value will be
<j>(y).
1.
Let
Let y
powers of x.f
*(y)
.....
e>
it
be required to develop e* sinx according to

x sin.r. Then
=l+y + \tf + ^
But the theorem

is the one that usually arises in practice.
The case
an
r
holds provided only that
r, the only difference being that the
Cf.
coefficients in the final series will then be infinite series instead of sums.
still
<^
<^
25.
Stolz, Allgemeine Arithmetik, Vol. I, Ch. X,
f Even when it is known that a function can
be developed by Taylor s
Ch. Ill; Diff. Cal., Ch. IX; Int. CaL, Ch. XVII) it is usually
simpler to determine the coefficients in the series by the method here set forth
than by performing the successive differentiations requisite in the application of
Theorem
Taylor
(v.
formula.
The example
in
hand
illustrates the truth of this statement.
50
36.
Find the first 4 terms in the expansion of sin

Obtain a few terms in the development of each of the following
functions according to powers of x.
2.
3.
log cos ft.
Let cos x
Suggestion.
and
log cos x
2x
1 -|-
x2
cos
(9
-^ x
-f
a;
then
iV ^
^/ i
A:
sin 2 ft
Theorem 4
gives no exact information concerning the extent of

the region of convergence of the final series. It merely asserts that
This deficiency is supplied by an elementary
there is such a region.
theorem in the Theory of Functions.*
But for many applications it is not necessary to know the exact

For example, let it be required to determine
region of convergence.
the following limit.
log cos a? 4-
== *
Mm
+^
Both numerator and denominator can be developed according

powers of x. The fraction then takes on the form
to
s
J x -\- higher powers of x
x 8 -j- higher powers of x
.1.
Cancel x s from numerator and denominator and then

as its limit.
The
limit of the fraction is then seen to
let
x approach
be
3.
The
usual method for dealing with the limit 0/0 is applicable here, but
the method of series gives a briefer solution, as the student can
readily verify.
Example.
Determine the
Va
lim
x
limit
--ft 2
l
An
Va*
ft
is
to the proof of the
cos x
important application of Theorem 4

following theorem.
* Cf. Int.
Cal.,
220; Higher Mathematics, Ch.VI, Functions of a
Variable, by Thos. S. Fiske John Wiley & Sons.
;
Complex
36.
51
THEOREM. The quotient of two power series can be represented

as a power series, provided the constant term in the denominator series
is
not
CQ
+az +
a^
Cj
C% X
if
show
It is sufficient to
0.
_|z
that
1
a{ x
>
2 .t
can be so represented, for then the power

can be multiplied into the numerator series
Let
1
o
_ 1
+y
al x
a2 x2
~~
+4
-\-
series that represents it
-\-
<V
provided y/a is numerically less than unity, i. e. y numerically less

Thus the conditions of Theorem 4 are fulfilled and the func
than a
.
4~ y) can be expressed as a power series in x by develop

n
n
ing each term (
l) y /a n + l into such a series and collecting from
these series the terms of like degree in x.
tion l/(a
COROLLARY.
denominator
7i
OQ-\-
/y
j
(>
If the
of the first m powers of x in the

can be expressed in the form
coefficients
series vanish, the quotient
jt
n
0_
/y.2
U.2
X-\-
-f-
I*n
nj
co
r ^
^
^1
r Xo
^2
t-
"T~
-\1
i
l
Ml
ri
+ l_l
-z=r ~r
-f-
-^-1
-r
T"
For
-|- 6 t
/7
Cv
I
"
/>.
a?
x2
-|- bz
/7
^
m
/vim
-4-
b l -|- 5j
-\-
+1
/yiW
r,
~Tn (C
JU
and
it
only remains to set
Show
Examples.
tana;
tt
-m
,,
C1
0,
+C
an(^ divide
=x+1
.^
a*
1
-a;
xs
X2
1 "
a?
into each term.
be given in
37.
-4-
and develop sec x and esc x to three terms.

A more convenient mode of determining the
expansions
("
that
ctna^ :=
will
-4-1
coefficients in these
CONTINUITY, INTEGRATION AND

DIFFERENTIATION OF SERIES.
V.
AVe have had numerous examples in the fore

Continuity.
Is the
functions
of
continuous
represented by power series.
going
converse true, namely, that every power series represents, within its
37.
That this question

interval of convergence, a continuous function?
is by no means trivial is shown by the fact that while the continuous
functions of ordinary analysis can be represented (within certain
limits) by trigonometric series, i. e. by series of the form
a trigonometric series does not necessarily, conversely, represent a

continuous function throughout its interval of convergence.
Let us
first
function.
<
put into precise form what is meant by a continuous

be continuous at the point x if
(x) is said to
lim
X
i.
e. if,
and y
a belt being
<f>
e,
(#<,)
<
=X
marked
where
off
e is
(a?)
<t>
(>
e
bounded by the lines y
) ~h
an arbitrarily small positive quantity,
<(#
FIG.
an interval
8,
(.*
when x
.T
O -(-
8), 8
^>
0,
can then always be found such
within this interval,

will lie within this belt.
These conditions can be expressed in the following form
that,
lies
<f>(x)
*(*o)
<
<t>
(x)
<
4>Oo)
e,
<
*
<
+ 8;
CONTINUITY, INTEGRATION, DIFFERENTIATION.
37.
53
or
A simple sufficient condition that
+ U, O) +
U (X)
represent a continuous function
THEOREM
1.
V (x)
is
series
the series of continuous functions
.....
given by the following theorem.
is
If
+u
.....
(x)
<
<
ft
of continuous functions convergent throughout the interval
f (x) represented by this series will be con

(a, /3),
tinuous throughout this interval, if a set of positive numbers,
Q
2
independent of x, can be found such that
then the function
.....
M^
1)
u n (x)
=0,1, i, .....
.....
+ M, + M, +
3/
2)
it
a<x<p,
\<Mn
a convergent series.
We have to show that, x being any point of the interval, if a posi
tive quantity e be chosen at pleasure, then a second positive quantity
8 can be so determined that
is
()
/)
\f(x)
Let
s n (x)
u (x)
f(x)
s
\
if
+ U! (x)
will
n( x)
/(ar
(x)
\x
<r
+;+
/()=:.(*).+
Then
We
e
<
(x
|<8.
_!
(aj),
*.(*)
}
rn (x ).
rn (x)
show that the absolute value of each of the

sn (x
o)\,
chosen and
x
|
M (
:tl
XQ
)j
r n ( xo) is
less
From
8.
\
<^
tlian
this
if
quantities
is
F
properly
follows that the absolute
*,
value of f(x)
f(%o) is less than e; hence the proposition.
Let the remainder in the Jtf-series be denoted by n
and
let
n be so chosen that
<^
ê,
and then held
fast.
Then,
since
UW
it
^n,
follows that
I
<
I
The
(*)
absolute value of a quantity
<
shall
R.
from now on be denoted by
.
\
54
for

all
37,38.
values of x at once,* or
I
rn (x)
Kit-,
a<x<p.
sn
(x) is the sum of a fixed number of continuous functions,
a continuous function and hence 8 can be so chosen that
Since
it is
*()
Kit,
*(so)
a>-3
|<s.
Hence
and the theorem
is
proved.
Show
Exercise.
that the series
since
sin3#
"I*"
~~32
sin5cc
I
52
converges and represents a continuous function.
38. The general test for continuity

power series
just obtained
can be applied
at once to
THEOREM
2.
power series represents a continuous function
interval of convergence.
The function may however be
discontinuous on the boundary of the interval.
within
come
its
Let the
series
be
convergent when
+a
f(x)
<^
<^
a 2 x2
-\-
and
.....
,
let (a, /?)
be any interval con
tained in the interval of convergence, neither extremity coinciding

with an extremity of that interval.
Let
be chosen greater than
either of the quantities
a
less
but
than r. Then
(3
and the
<*
|<
X-,
a<x<(3;
series
a
|
converges.
Hence
if
a,
we
X+
Theorem
1 will
continuous throughout the interval
By
a2
X + .....
2
set
J.=
the conditions of
.!*,
be satisfied and therefore f(x)
is
(a, /?).
the aid of this theorem the following theorem can be readily
proved.
* It is
just at this point that the restriction on a convergent series of con
tinuous functions, which the theorem imposes, comes into play. Without this
restriction this proof would be impossible and in fact, as has already been
pointed out, the theorem
is
not always true.
55
38.
THEOREM. If a power series vanishes for

a certain interval about the point x
+ a^x + a
a2
values of x lying in
all
.....
I
<
<
then each coefficient vanishes:
First put x
0,
0,
then a
.....
and the above equation can be
written in the form
=
From
this equation
x (a l
satisfied
=j=
a2 x
al
-{-
.....
.....
but it does not follow that this last equation is

=. 0, and therefore a l cannot be shown to vanish by
here as in the previous case.
Theorem 2 furnishes a
;
when x
/!()
since /j (x)
is
=a +a
1
lim /! (x)
But
lim
x
/i
Let
difficulty.
2
.....
a;+
by that theorem a continuous function of x

x
By
az x
putting x
convenient means of meeting this
Then
-j-
follows that
it
=
provided x
-\-
(x)
=Q
./I
(0)
a,
a x =:
.-.
repeating this reasoning each of the subsequent coefficients can

0, and thus the theorem is established.
be shown to be
If two power series have the same value for
COROLLARY.
values of x in an interval about the point x

are respectively equal:
=&
&i,
all
0, their coefficients
etc.
Transpose one sei-ies to the other side of the equation and the
proof is at once obvious.
The Determination of the Coefficients c. It was shown in

quotient of two power series can be represented
that the
power
By
36
as a
series.
the aid of the theorems of this paragraph a more convenient

of determining the coefficients c can be established.
mode
each side of the equation by the denominator series
Multiply
56
=. a
c -f- (î c -f-
Hence
Ci)
ac
x -\- (a 2 c
-)-
%c
-|-
38
c<
c2
simple mode of solving these equations for the successive c s is

furnished by the rule of elementary algebra for dividing one poly
nomial by another.
Quotient:
a 3 ^8
fl
(&o-
a2 c
(& 2
ajCi
c2
)a?
+(6
a3
a2 cl
c-
a^)x^-\-
etc.
The equations determining

the
first
the c s are precisely the condition that
term in the remainder shall vanish each time.
For example,
to develop tan
aj,
divide the series for sin x
by the
series for cos x.

Quotient:
/y
_i_ 1 /^3 _j_ J2_

1
3
I
15
I
I
*
X
T5^
etc.
Hence
tan x
This method
lary,
is
-\-
s
$x
-\-
5
2
T ^- x
-\-
applicable even to the case treated in the corol
36.
Exercise.
Develop
+x+x
csc
aj
39.
57
39. The Integration of Series Term by Term. Let the continuous

function f(x) be represented by an infinite series of continuous func
tions convergent throughout the interval (a, ft)
:
u 9 (x)
The problem
is this
f(x)
given by
+ Ul
(x)
i.
e. to
<
<
(3.
(A)
when
the integral of f(x) will be

of
the terms on the right of
integrals
to determine
the series of the
equation (A)
.....
determine when
f
C *u
(x )dx
(B)
i) a
The right hand

a true equation.
the term by term integral of the ?<-series.
will be
member
of (J5)
is
called
Let
Then
r&
r&
(*B
I
f(x)dx=
sn
(x)dx-\-
tja
tJ a.
f(x)dx= Pu,(x)dx+ Pu,(x)dx+

e/a
rn
(x)dx
tJ a.
.....
t/a
+e/aPu
_ l (x)dx
(*B
+t/a
Hence
isthat
the necessary
and
r (x)dx.
lt
sufficient condition that
^^^ Jaf
is
a true equation
r.OOdtes-O.
To
obtain a test for determining

plot the curve
rn
FIG.
The
(B)
lim
when
this condition is satisfied,
9.
area under this curve will represent

geometrically
r n (x)dx.
58
Draw lines
through the highest and lowest points of the curve parallel

The distance p n of the more remote of these lines
to the ie-axis.
from the
maximum
aj-axis is the
value that
r n (x)
attains in the
a belt bounded by the lines y

p a and y
p
the curve lies wholly within this belt and the absolute value of
interval.-
Then
Lay
off
tl
the area under the curve cannot exceed the area of the rectangle
bounded by the line y
This area will converge
p n or (ft
a) p n
as its limit if *
toward
lim
and thus we
if
(B)
we
have a
shall
tl
0,
QO
sufficient condition for the truth of
establish a sufficient condition that the
maximum
equation
value p n
in the interval (a, ft) approaches

rn (x)
when n
Now
GO
in the proof of Theorem 1 that if the series (A) satisfies the
conditions of that theorem,
of
we saw
r n (x)
<
|
HJ
<
<
Hence any such
Let the
arise in ordinary practice.
f(x)dx
c$P
/
ya
t/a
test be
most of the cases that
formulated as follows.
>
Ul
(x)dx+
t/a
a true equation, if a set ofpositive numbers

independent of x, can be found such that
|
u n (x)
<
|
3/
2)
is
i. e.
CP
(x)dx+
will be
1)
0.
oo
Series (A) can always be integrated term by term,
3.
r$P
I
can be integrated term by term and we have
series
in this result a test sufficiently general for
THEOREM
R =
lim
ft,
<
<
ft,
M\,
0, 1, 2,
.,
+ M + Mt+
a convergent
series.
The form
which the
been deduced is restricted to real

But the theorem itself is equally appli
complex variables and functions. It is desirable therefore
in
test has
functions of a real variable.

cable to
to give a proof that applies at once to both cases.

*
This condition
restrictions
is
not satisfied by
hitherto imposed on (A).
all
series that are subject
Not every
series
merely
to the
of this sort can be
See an article by the author A Geometrical Method

Treatment of Uniform Convergence and Certain Double Limits, Bul
letin of the Amer. Math. Soc., 2d ser., vol. iii, Nov. 1896, where examples of
series that cannot be integrated term by term are given and the nature of such
series is discussed by the aid of graphical methods.
integrated term by term.
for
the
Keeping the notation used above, the
CP
CP
I
Ja
59
39, 40.
f(x)dx
CP
(x)dx+
relation
+JaCP u _
u l (x)dx-\-
Jo.
Jo.
+ JCP r
I
(x)dx
(x)dx
holds and the proof of the theorem turns on showing that the
hypotheses are sufficient to enable us to infer that
still
Then
it
Rn
37 by
Let the remainder of the Jf-series be denoted as in
follows, as in that paragraph, that
r n (x)
<
\
Now
/fl
/fl
rn
Ja
(x)dx
<
Ja
dx
rn (x)
<
n l
the second integral being extended along the same path as the first,
and I denoting the length of the path. But lim (R n l)
hence
when n
(x)dx converges toward
rn
oo
and the proof
is
i/a
complete.
of
40. We proceed now to apply the above

some of the more common forms of series.
test to the integration
A power series
First Application: Power Series.

term by term throughout any interval
of convergence and not reaching out
Let the
series
let
a
|
|/8
I
and
if
a^x
X be chosen greater than

,
we
can be integrated
contained in the interval
to the extremities
of this interval:
be written in the form
f(x)
and
(a, /3)
but less than
<
\a n
set
|
+a
^2
the greater of the two quantities
Then
r.
X",
a<x</3,
an
X =M
1
the conditions of the test will be satisfied.
60
Ill
particular
C* h
+a
a h
Joo
when x
If
40.
then h
r or
may be
h2
+a
h3
.....
li
|
<
the series for f(x) converges absolutely,

r.
If however the series for
?*,
taken equal to r or
r or
r, it
f(x) does not converge absolutely or diverges when x
may nevertheless happen that the integral series converges when
r or
r.
In this case the value of the integral series will still
h
Thus
be the integral of f(x).
+ x~
diverges when x =
the series
holds
still
when h
The proof
but the equation
of this theorem will be omitted.
Let
be
Second Application : Series of Poicers of a Function.
a continuous function of x and let its maximum and minimum values
<(#)
between
lie
converge when
and
r
<^
\x\
when a
Then
r.
the series
<^
can be integrated term by term from a
Let the power series
ft.
to ft
ft
t/a
f(x)-d=a9 f^dx+ai
e/a
t/a
For
P<t>(x)dx
if
value of
Y be so taken that
greater than the numerically greatest

<
tw
+ |i ^+
i)
2)
converges
it is
(x) in the interval a
<
and
if
we
<C
but less than
)8,
r-,
<
|"
set
an
Y
|
the conditions of the test will be satisfied.

Thus the integrations of
24, 25 are justified.
r,
then
40, 41.
Third Application.
If the function
+
satisfy the
!?/
same conditions as
2?/
+ a^(x)
a ^(x)
the series
<f>(x)
.....
in the preceding theorem
any continuous function of x, then

f(x)
and
<fr(x)
61
and
if
\ff(x)
is
the series
+ a^(x)
O(V>P
.....
can be integrated term by term.

The method of proof has been so fully illustrated in the two pre
ceding applications that the detailed construction of the proof
be left as an exercise to the student.
This theorem
Cauchy
needed in the deduction of Taylor
is
may
Theorem from
s Integral.
I.
Examples.
Compute
/
x*e~ x dx,
2.
Show
-v/
sin a? da?.
t/O
J"!
that
X5
.
Hitherto the limits of integration have always been the limits of

the interval considered, a and j3.
It becomes evident on a little
reflection that if
the interval
(a, (3)
For,
all
any other
<
limits of integration,
are taken,
<
Theorem
still
ft,
<
aj
a;,
hold
<
lying within
ft.
the conditions of the test will hold for the interval
they hold for the interval
throughout the interval
any point of
(a?
x)
(a, ft).
Then
Term
by Term.
the derivative
the interval by the series
Let the function
f (x)
ivill
of the derivatives
be given
provided the series of the derivatives
u
satisfies the conditions
(x)
if
(a, ft).
41. The Differentiation of Series

f(x) be represented by the series :
at
3 will
+u
(x)+
.....
of Theorem 1 throughout the interval
(a, ft)
62
Let the
We
be denoted by
latter series
<
41.
(a;)
wish to prove that
^(x)=f
By Theorem
the function
Theorem
<j>(x)dx
2 the series
+ Jo.
u (x)dx
ul(
t/a
.....
+
W +
= /(*)-/().
MO()}
{<>(*)
r*x
f*x
f*x
\
Jo.
represented by the î -series is con

can be integrated term by term
(x)
<f>
tinuous and by
(x).
<X)
Hence, differentiating,
Show
Exercise.
q. e. d.
(x),
4>(x)=f
that the series
cos a?
cos 3
cos 5 a;
a;
~^
~1~
can be differentiated term by term.

By the aid of this general theorem we can
at once
prove the follow
ing theorem.
THEOREM. A power series can be

any point within (but not necessarily
its
term by term at
a point on the boundary of)
interval of convergence.
Let the power series be
+ax+
f(x)
x
|
al
Then we want
-\-
x
|
<^
to prove that
(x
?*,
<^
a,
r
<^
<^
out the interval
r,
3 a 3 x2
-\-
a:
3a 8 a;
converges.
series of the derivatives

if
con
be so chosen that
the conditions of the test will be fulfilled through

can prove this as follows. Let x
X, X).
We
<
r,
(
<
X"
\
a, 1
+2
a,
as-
+8
<^
<^
?%
series
.....
+
a.
r)
The
r.
It will serve as a test-series for the

I
.....
for in that case,
X
+| % \X +\a
series of the derivatives
?*,
<^
2tt g
.....
-\-
if
.....
^3
and form the
be any value of x within the interval

x
be so chosen that
let
show that the
It will be sufficient to
verges when
2^
convergent when
differentiated,
at
convergence of
2
I
|
and
41.
if it
63
can be shown that
n
from some
an
"~
I
a;
definite point,
J
<
This
m, on.
x
<
\a n
will
tn
be the case
n
>
if
But the expression on the left approaches

when n
oo
for
x
n
is
of
and
less
than
1
the
limit
can
independent
/
therefore be obtained by the usual method for evaluating the limit
Hence the condition that the former series may serve as
oo
0.
test- series is fulfilled and the proof is complete.
From
Exercise.
the formula
obtain by differentiation the developments for

1
1
2
(1
a)
(1
z)
(1
z)
and show that they agree with the corresponding developments given
by the binomial theorem.
APPENDIX.
A FUNDAMENTAL THEOREM REGARDING THE EXISTENCE
OF A LIMIT.
It was shown
when n increases
13 that
in
when p
approaches the limit
And
stated that
is
was
when n
will
approach a
limit.
n.
it
the variable s n approaches a limit
if
indefinitely, then
oo
constant or varies in any wise with
is
conversely, the limit of this expression

no matter how we allow p to vary with n, then sn
if,
This latter theorem
is
important in the theory
however only a special case of a theorem

regarding the existence of a limit, which is of fundamental impor
of infinite series.
It
is
tance in higher analysis.
THEOREM.
Let f(x) be any function of x such that

Urn [/(
)/(")]
when x and
regarded as independent variables,
Then f(x) approaches a limit when x
GO
nite.
05",
We will begin
f(x )
/(#")
by
stating precisely
approaches the limit
both become infi
what we mean by saying that

when x and
regarded as
a?",
is
independent variables, both become infinite. We mean that if
taken as an independent variable that is allowed to increase without
limit
and then, corresponding to any given value of X, the pair of
values (V,
both x and
chosen arbitrarily subject only to the condition that
is
aj")
x"
are greater than
X (or at least as great),
the quantity
In other
as its limit.
then converge towards
/(# )
/(#")
an
small
let
e
denote
words,
positive quantity.
arbitrarily
can be so chosen that *
will
/(a;
We
/(")
proceed
values that
now
<
on any
creasing and approaching the

1, J,
X by Xu X X
2,
e,.
l/i.
set
limit
and
x"
>
X.
e 1? e 2 , e 3 ,
for
steadily de
the values
example
Denote the corresponding values of
Then
>
Let us choose for the successive
to the proof.
to take
is
if
Then
For the notation cf
in general these latter values will

.
foot-note, p. 53.
APPENDIX.
steadily increase, and
do always increase.
we can
Begin by putting e
/f(^
Jf(i\
(x )
(X
Assign to x the value
any case choose them so that they
in
a/
/>
ei
<^
/(XO
e.
^>
^_ ^\i
v"
^>
7?
^Y!.
Then
le
/(XO-/W
i.
65
Cl
<f(x)
for all values of x greater than
|<e l5
</(XO
Cl
The meaning
tion can be illustrated graphically as follows.
of this last rela
Plot the point /(Xj)
Then
on a line and mark the points /(Xj)
^ and f(Xt ) -f- t
the inequalities assert that the point which represents f(x) always
lies within this interval, whose length is 2e^ provided x
lt
.
^>
/j )-i
/PL
/CEjH-4*
/<?*>+*
in
FIG.
10.
e
of this interval by a 1?
Denote the left hand boundary f(X\)
the right hand boundary /(Xj) -|- ^ by ^^
Then, to restate con
cisely the foregoing results,

i
Now
</()
<
/(X
(a)
/(X )
e2
both of these points
in this case, let a 2
other point
/(X )
2
denoted by fa
(c)
val
the value
e2
ai
/(X
2)
/(X )
2
+
lie
/3 2
04
e2
Then
Plot
Mark
three cases can arise
in the interval (a x , fa)
let
the
them be
that/(X2 )
e2
falls outside the inter
be taken coincident with
04
will lie in the interval (a 1? yS^
so near to
lies
denoted by a 2
a2
and
ttl
shall
the
be
in this case, let
other point
e2
Cl
</(x
and /(X2 )
,
ft
denoted respectively by a 2
near to
(6) /(X2 ) lies so
val
Xi
denotes any value of the variable x greater than

/(X2 ) this point lies in the interval (a l5 /5J.
the point
points
>
e2
2)
a;
a?
repeat this step, choosing for
i.e.
where
if
/? 2
e2
/? t
that/(X2 )
e2
falls outside the inter
be taken coincident with

will lie in
the interval (a,, fa)
fa
and
fa
the
shall be
APPENDIX.
66
In each one of these three cases

a2
</(OJ)
<
if
ft
>
a2
ft
2e 2
<
The remainder of the proof is extremely simple. The step just

described at length can be repeated again and again, and we shall
have as the result in the general case the following
*^
if
Now
a.<2e.
consider the set of points that represent a 1? a 2

a,,
in general toward the right as i increases,
.
They advance
never recede toward the
they
but no one of them ever advances
left,
so far to the right as ft.

Hence, by the principle* of
4, they
opproach a limit A. Similar reasoning shows that the points repre
And since
senting ft, ft, ... ft, ... approach a limit B.
<
ft
a,
<
e,
A=
B.
these limits must be equal
From this it follows that f(x) converges toward the
:
same
limit.
For
ft,
,</()<
and
if
when x
nitely at the
*>*;;
increases indefinitely,
we
allow
same time, but not so rapidly as
to increase indefi
to invalidate these in
we see that/(#) is shut in between two variables, a and ft,

each of which approaches the same limit. Hence /(a;) approaches
that limit also, and the theorem is proved.
equalities,
In the theorem
above quoted n
in infinite series
is
the independent
the expression s n+p

variable x, S H the function /(V)
^corre
l thus that theorem is seen to be a
x
an(
to
sponds
f(
/(# )
;
")
The domain of values for

special case of the theorem just proved.
the variable x is in this case the positive integers, 1, 2, 3,
Another application of the present theorem is to the convergence
.....
of a definite integral
when
the upper limit becomes infinite.
f(x)=
Then
/(x )
/(")
= f%
Ja
(a?)
Let
P<l>(x)dx.
dx
C*
Ja
V (x) dx
$ (x) dx.
Jxii
Xx $ (x)dx =
/
This principle was stated, to be sure, in the form S,,t

but it obviously continues to hold if we assume merely that
*
^>
n.
S
,/
if
>
n
5,,
1
^>
n;
when
67
APPENDIX.
as independent variables, both
when x and x \ regarded

finite, the integral
become
in
/oo
<f>(x)dx
s.
The domain of values for the variable x is in this
is convergent.
case all the real quantities greater than a.
In the foregoing theorem it has been assumed that the independent
The theorem can however be
variable x increases without limit.
indefi
readily extended to the case that x decreases algebraically
sides.
nitely or approaches a limit a from either side or from both
In the
case, let
first
y;
in the second, let
a H-y
if
is
always greater than
its limit
let
= a -1
y
if
is
than a.
less
always
Then
f(x)
if
we
<
set
(;y)
theorem when
will
a
and
limit.
hence
approach
Finally, if x
(y)
f(x)
in approaching a assumes values sometimes greater than a and some
times less, we may restrict x first to approaching a from above,
and the function

y
-|- oo
<
(y) satisfies the conditions of the
<f>
secondly from below.
In each of these cases
that f(x) approaches a limit,
it
has just been seen
and since
lim
may now be taken the one above, the other below a,

these two limits must be equal.
are thus led to the following
more general form of statement of the theorem.
where x and
x"
We
THEOREM.
Let f(x) be such a function of x
that,
when x and
regarded as independent variables, approach the limit
a from above or from below or from both sides, or become positively or
Then f(x) approaches a limit when x approaches
negatively infinite.
1
x",
the limit
a from above or from

or negatively infinite.
belotv or
from
both sides, or becomes
68
APPENDIX.
A TABLE OF THE MORE IMPORTANT FORMULAS.

The heavy
line indicates the region of convergence.
- = +x+ + +
1
JL>
1,5^x H
101
bx
-\
x*
x*
a2
b*
=
a3
xz
+a
.
.....
X B,
r
101
1-2
(1+x)
_ 2x + 3x
1-2-3
= a numerically.
.
,,
APPENDIX.
Vi
11-31T +
= +oX +
1
tf
69
X*
/v.2
~
*
X4
2^4
2-4-6
=
o
sin
a?
=x
a;
3
,
x*
-j-
2!
-3
/y.4
"
e*
3- 5
101
o
4!
o
6!
rt6
*
APPENDIX.
70
sec x
= +
J X*
101
xs
2 3
ic
+ â
3 x5
^ 2^~4
2-4-6 7
101
x5
a^
________f_
(a;
r-V
Jo
5 x1
""
5~
APPENDIX.
71
For small values of x the following equations are approximately

correct.
/(a
)==./(>
+ / (a)
=
Vi +
If
a;,
y, 2, 10,
+ x)
(1 + x)
(1
are
all
= +x+y+z+
+ y) (1 + z)
(1 + y)
_
x
A a;
numerically small, then, approximately,

1
(1
sin
2*
RETURN
Astronomy/Mathematics/Statistics/Computer Science Library
TOO Evans Ha
LOAN PERIOD
7
DAYS
642-3381

Infinite Series by Osgood

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Infinite Series by Osgood

Загружено:

Авторское право:

Доступные форматы

/v\.

ASSISTANT PROFESSOR OF MATHEMATICS IN HARVARD UNIVERSITY

and Integral Calculus

the subject of Infinite Series forms an important topic.

presentation of this subject should have in view

what the question

never enters on the proof of a theorem

he feels that the theorem

Aids to the attainment of these ends are

(a) a variety of illustrations,

taken from the cases that actually arise

in practice, of the application of series to computation both in pure

and applied mathematics (b) a

diagrams and graphical illustrations

the kind here indicated.

and careful exposition of the

designed to give a presentation of

references are to Byerly s Differential

Calculus, Integral Calculus, and Problems in Differential Calculus ,

Short Table of Integrals;

Consider the successive values of the variable

Let r have the value J.

easy to see the

law by which any s n can be obtained from its predecessor,

formula for the sum

have been obtained arithmetically from the

and hence as before Lim

Denote the sum of the

to increase without limit.

and form the

set of values, positive or negative or both,

In either case we speak of (1) as an

case a) the infinite series is said to be convergent and to have the

are examples of divergent series.

series are of use

often used for the limit

inf. (or to infinity)

often called the sum of the series.

ALL OF WHOSE TERMS ARE POSITIVE.

be required to test the convergence of the

where n\ means 1-2 3

Discarding for the

term, compare the

with the corresponding

and hence the sum

or, inserting the

after the first

than the corresponding

discarded term and denoting the

terms of the given series by

no matter how large ?i be taken. That is to say, s n is a variable

n increases by 1, the point representing s H + 1 always moves

must be some point

either coinciding with

limit, but never reaches.

The reasoning by which we have

here inferred the existence of a limit

know how to compute the numerical value of that limit, is of prime

always increases wlien n increases:

not greater than

the limit itself or any value greater than the

State the Principle for a variable that is always de

draw the corresponding

First Test for Convergence.