XXV .
A new Method of investigating the Sums of infinite Series .
By the Rev. S. VINC E , A. M. of CAMBRIDGE , in a Letter to Henry Maty , A. M. Secretary .
Read June 6 , 1782 .
SIR , HAVING lately discovered some very easy methods of investigating the sums of certain infinite series , I have taken the liberty of requesting the favour of you to present them to the Royal Society .
I have divided the subject into three parts : the first contains a new and general method of finding the sum of those series which DE MOIVRE has found in one or two particular cases ; but whose method , although it be in appearance general , will , upon trial , be found to be absolutely impracticable .
The second contains the summation of certain series , the last differences of whose numerators become equal to nothing .
The third contains observations on a correction which is necessary in investigating the sums of certain series by collecting two terms into one , with its application to a variety of cases .
I am , &amp;c. Cambridge , May 3 , 1782 .
PART I. LE M. I. Let r be any whole number , and then the fluent of x/ 1 + xr can always be exhibited by circular arcs and logarithms ; but when x= 1 , the fluent of the same fluxion will be expressed by the infinite series 1 1/ r +1+ 1/ 2r +1 1/ 3r +1+ &amp;c. the sum of this series therefore can always be found by circular arcs and logarithms .
LE M. II .
To find the sum of the infinite series a/ 1 .
~r +1a+ b/ ~r + 1 .
~2r +1+ a+ 2b/ ~2r + 1 .
~3r +1 &amp;c. Assume 1 1/ r +1+ 1/ 2r +1 1/ 3r +1+ &amp;c. .
= S ; therefore , ( A ) 1 1/ 1 .
~r +1+r+ 1/ ~r + 1 .
~2r +1 2r + 1/ ~2r + 1 .
~3r +1+ &amp;c. ... = S. In the first series , add together the 1st and 2d , the 2d and 3d , &amp;c. &amp;c. terms , and the resulting series will evidently be equal to twice that series minus the first term ; therefore , ( B ) , r/ 1 .
~r +1 r/ ~r + 1 .
~2r +1+ r/ ~2r + 1 .
~3r +1 &amp;c. ... = 2S 1 .
Now ( A/ r ) 1/ r 1/ r/ 1 .
~r +1+1+ 1/ r/ ~r + 1 .
~2r +12+ 1/ r/ ~2r + 1 .
~3r +1+ &amp;c. ... =S/ r , or 1/ ~r + 1 .
~2r +1 2/ ~2r + 1 .
~3r +1+ &amp;c. ... = S/ r ; + 1/ r 1/ r/ 1 .
~r +1+ 1/ r/ ~r + 1 .
~2r +1 &amp;c. ... Now the sum of the lower series , omitting the first term , is equal to B divided by r2 , or = 2S 1/ r2 ; hence , by transposition , and , multiplying both sides by b , we shall have , b/ ~r + 1 .
~2r +1 2b/ ~2r + 1 .
~3r +1+ &amp;c. . .
= bS/ r + 2bS ~r + 1 .
b/ r2 ; also by multiplying B by a/ r we have a/ 1 .
~r +1 a/ ~r + 1 .
~2r +1+ a/ ~2r + 1 .
~3r +1 - &amp;c. ... =2aS a/ r ; subtract the last equation but one from the last , and we shall have a/ 1 .
~r +1a+ b/ ~r + 1 .
~2r +1+a+ 2b/ ~2r + 1 .
~3r +1 &amp;c. ... = ~2ra~r + 2 .
bxS ra + ~r + 1 .
b/ r2 Cor. 1 .
Hence it appears , that the sum of this series can never be exhibited in finite terms , except a : b as r+ 2 : 2r , in which case the sum is equal to a/ r + 2 .
Hence , if a= 3 , b= 2 , then r= 1 ; ~3/ 1.2 5/ 2.3 + 7/ 3.4 - &amp;c. ... = 1 ; if a= 1 , b= 4 , then r= 4/ 3 ; ~5/ 3.7 9/ 7.11 + 13/ 11.15 17/ 15.19 + &amp;c. ... = 1/ 6 ; if a= 4 , b= 3 , them = 6/ 5 ; ~ 4/ 5.11 7/ 11.17 + 10/ 17.23 13/ 23.29 + &amp;c. ... =1/ 20 ; Cor 2 .
Put a=c b , and we shall have , after transposition , c/ ~r + 1 .
~2r +1c+ b/ ~2r + 1 .
~3r +1+ &amp;c. ... = ~3r + 2 .
b 2rc XS ~2r +1b+ rc/ r2 + c-b/ r + 1 .
PRO P. I. To find the sum of the infinite series m/ 1 .
~r + 1 .
~2r +1+ m+n/ ~2r + 1 .
~3r + 1 .
~4r +1 m+2n/ ~4r + 1 .
~5r + 1 .
~6r +1+ &amp;c. Every series of this kind may be resolved into the following series a/ 1 .
~r +1 a+b/ ~r + 1 .
~2r +1+ a+2b/ ~2r + 1 .
~3r +1 a+3b/ ~3r + 1 .
~4r +1+ &amp;c. for if we reduce two terms of this series into one , it will become 2ar b/ 1 .
~r + 1 .
~2r +1+ 2ra +~2r 1 .
b/ ~2r + 1 .
~3r + 1 .
~4r + 1 .
+ 2ra + ~4r 1 .
b/ ~4r + 1 .
~5r + 1 .
~6r +1+ &amp;c. where the denominators being the same as in the given series , and the numerators also in arithmetic progression , we have only to take a and b such quantities that the respective numerators may be also equal ; assume , therefore , 2ra b= m , 2ra + ~2r 1 .
b=m+ n ; therefore , b= n/ 2r , a= 2rm + n/ 4r2 , which substituted for a and b in LEM .
2 .
gives m/ 1 .
~r + 1 .
~2r +1+ m+n/ ~2r + 1 .
~3r + 1 .
~4r +1+ m+2n/ ~4r + 1 .
~5r + 1 .
~6r +1+ &amp;c. ... = 2rm~r + 1 .
n/ 2r3 xS+ ~2r + 1 .
n 2rm/ 4r3 .
Let r= 1 , and we have m/ 1 .
2 .
3+m+ n/ 3 .
4 .
5+m+ 2n/ 5 .
6 .
7+ &amp;c. ... = ~m n. S+ 3n 2m/ 4 .
If m= 1 , n= 3 , 1/ 1 .
2 .
3+ 4/ 3 .
4 .
5+ 7/ 5 .
6 .
7 &amp;c. ... = 7/ 4 2S ; m=1 , n=0 , 1/ 1 .
2 .
3+ 1/ 3 .
4 .
5+ 1/ 5 .
6 .
7+ &amp;c. ... = S-1/ 2 .
Let r= 2 , and the series becomes m/ 1 .
3 .
5+m+ n/ 5 .
7 .
9+m+ 2n/ 9 .
11 .
13 + &amp;c. ... = 4m 3n/ 16 xS+ 5n 4m/ 32 .
If m=1 , n=1 , 1/ 1 .
3 .
5+ 2/ 5 .
7 .
9+ 3/ 9 .
11 .
13 + &amp;c. ... = S/ 16 + 1/ 32 ; m=1 , n=0 , 1/ 1 .
3 .
5+ 1/ 5 .
7 .
9+ 1/ 9 .
11 .
13 + &amp;c. ... = S/ 4 1/ 8 .
Let r= 5 , and we shall have m/ 1 .
6 .
11 +m+ n/ 11 .
16 .
21 +m+ 2n/ 21 .
26 .
31 + &amp;c. ... = 5m 3n/ 125 xS+ 11n 10m/ 500 .
If m= 1 , n= 1 , 1/ 1 .
6 .
11 + 2/ 11 .
16 .
21 + 3/ 21 .
26 .
31 + &amp;c. ... = 2/ 125 xS+ 1/ 500 ; m= 1 , n= 0 , 1/ 1 .
6 .
11 + 1/ 11 .
16 .
21 + 1/ 21 .
26 .
31 + &amp;c. ... = S/ 25 1/ 50 .
Cor. If 2r : r+ 1 : : n : m , the sum of the series can be accurately found , and will be equal to m/ 2r .
r+ 1 .
Let therefore m=r+ 1 , and then n= 2r , consequently 1/ 1 .
2r +1+ 1/ 2r + 1 .
4r +1+ 1/ 4r + 1 .
6r +1+ &amp;c. ... = 1/ 2r ; which is also known from other principles .
PRO P. II .
To find the sum of the infinite series m/ r + 1 .
2r + 1 .
3r +1+ m+ n/ 3r + 1 .
4r + 1 .
5r +1+m+ 2n/ 5r + 1 .
6r + 1 .
7r +1+ &amp;c. This series resolves itself into c/ r + 1 .
2r +1c+ b/ 2r + 1 .
3r +1+c+ 2b/ 3r + 1 .
4r +1 &amp;c. ; for by reduction , as before , it becomes 2cr r+ 1 .
b/ r + 1 .
2r + 1 .
3r +1+ 2cr +r 1 .
b/ 3r + 1 .
4r + 1 .
5r +1+ 2cr + 3r 1 .
b/ 5r + 1 .
6r + 1 .
7r +1+ &amp;c. where the denominators are the same as in the given series , and the numerators in arithmetic progression ; assume therefore 2crr+ 1 .
b= m , 2cr+r 1 .
b=m+ n , hence b= n/ 2r , c= 2rm +r+ 1 .
n/ 4r2 , which , substituted in cor .
2 .
LE M. 2 .
give m/ r + 1 .
2r + 1 .
3r +1+m+ n/ 3r + 1 .
4r + 1 .
5r +1+m+ 3n/ 5r + 1 .
6r + 1 .
7r +1+ &amp;c. ... = 2r + 1 .
n 2rm/ 2r3 xS+ 2rm 3r + 1 .
n/ 4r3 + 2rm r 1 .
n/ 4r2 .
r+ 1 .
Cor. 1 .
In prop. 1 .
substitute a for m , and 2b for n , and we have a/ 1 .
r+ 1 .
2r +1+a+ 2b/ 2r + 1 .
3r + 1 .
4r +1+a+ 4b/ 4r + 1 .
5r + 1 .
6r +1+ &amp;c. ... = ra r+ 1 .
b/ r3 xS+ 2r + 1 .
b ra/ 2r3 .
Also in this prop. substitute a+b for m , and 2b for n , and we have a+ b/ r + 1 .
2r + 1 .
3r +1+a+ 3b/ 3r + 1 .
4r + 1 .
5r +1+ &amp;c. ... = r+ 1 .
b ra/ r3 xS+ ra 2r + 1 .
b/ 2r3 + ra + b/ 2r2 xr+ 1 .
Subtract this latter series from the former , and a/ 1 .
r+ 1 .
2r +1a+ b/ r + 1 .
2r + 1 .
3r +1+a+ 2b/ 2r + 1 .
3r + 1 .
4r +1 &amp;c. ... = 2ra r+ 1 .
2b/ r3 xS+ 2r + 1 .
b ra/ r3 ra + 2b/ 2r2 xr+ 1 .
Let r= 1 , and we have a/ 1 .
2 .
3a+ b/ 2 .
3 .
4+a+ 2b/ 3 .
4 .
5 &amp;c. ... = 2a 4b xS+ 11b 5a/ 4 .
If a= 1 , b= 0 , 1/ 1 .
2 .
3 1/ 2 .
3 .
4+ 1/ 3 .
4 .
5 &amp;c. ... = 2S 5/ 4 ; a= 1 , b= 2 , 1/ 1 .
2 .
3 3/ 2 .
3 .
4+ 5/ 3 .
4 .
5 &amp;c. ... = 17/ 4 6S .
Let r= 3 , and we have a/ 1 .
4 .
7a+ b/ 4 .
7 .
10 +a+ 2b/ 7 .
10 .
13 &amp;c. ... = 6a 8b/ 27 xS+ 53b 33a/ 216 .
If a= 1 , b= 0 , 1/ 1 .
4 .
7 1/ 4 .
7 .
10 + 1/ 7 .
10 .
13 &amp;c. ... = 2/ 9 S 11/ 72 ; a= 1 , b= 1 , 1/ 1 .
4 .
7 2/ 4 .
7 .
10 + 3/ 7 .
10 .
13 &amp;c. ... = 5/ 54 2/ 27S .
If , instead of substituting in prop. 1 .
2b and a for n and m , we had substituted two other quantities , as 2r and s , and then proceeded as above , a series would have been formed , the numerators of whose alternate terms would have formed each a 2 separate arithmetic progression .
If the latter series had been added to the former a series would have been formed whose terms would have been all positive ; but as I purpose , in the second part of this paper , to give a general method of summing all such series , I shall not stop here to apply this method of investigation .
Cor. 2 .
In proposition 2 .
substitute a for m , and 2b for n , and we shall have a/ r + 1 .
2r + 1 .
3r +1+a+ 2b/ 3r + 1 .
4r + 1 .
5r +1+ &amp;c. ... = 2r + 1 .
b ra/ r3 xS+ ra 3r + 1 .
b/ 2r3 + ra r 1 .
b/ 2r2 .
r+ 1 .
Also in prop. 1 .
write ab for m , and 2b for n , and there results a+ b/ 2r + 1 .
3r + 1 .
4r +1+a+ 3b/ 4r + 1 .
5r + 1 .
6r +1+ &amp;c. ... = ra 2r + 1 .
b/ r3 xS+r+ 1 .
b ra/ 2r3 a b/ r + 1 .
2r + 1 .
Subtract this latter series from the former , and we shall have a/ r + 1 .
2r + 1 .
3r +1a+ b/ 2r + 1 .
3r + 1 .
4r +1+a+ 2b/ 3r + 1 .
4r + 1 .
5r +1 &amp;c. ... = 2r + 1 .
2b 2ra/ r3 xS+ ra 3r + 1 .
b/ r3 + ra r 1 .
b/ 2r2 .
r+1+a b/ r + 1 .
2r + 1 .
PRO P. III .
To find the sum of the infinite series m/ 1 .
r+ 1 .
2r + 1 .
3r +1+ m+ n/ 2r + 1 .
3r + 1 .
4r + 1 .
5r +1+m+ 2n/ 4r + 1 .
5r + 1 .
6r + 1 .
7r +1+ &amp;c. This series resolves itself into a/ 1 .
r+ 1 .
2r +1a+ b/ r + 1 .
2r + 1 .
3r +1+a+ 2b/ 2r + 1 .
3r + 1 .
4r +1 &amp;c. for by reduction it becomes 3ra b/ 1 .
~r + 1 .
~2r + 1 .
~3r +1+ 3ar + ~4 1 .
b/ ~2r + 1 .
~3r + 1 .
~4r + 1 .
~5~ +1+ 3ar + ~8r 1 .
b/ ~4r + 1 .
~5r + 1 .
~6r + 1 .
~7r +1+ &amp;c. where the numerators are in arithmetic progression , and the denominators the same as in the given series ; assume therefore 3ra b= m , 3ra + ~4r 1 .
b=m+ n , hence b= n/ 4r , a=~+ n/ 12r2 ; substitute these values into cor .
1 .
prop. 2 .
and we have m/ 1 .
~r + 1 .
~2 + 1 .
~3r +1+m+ n/ ~2r + 1 .
~3r + 1 .
~4r + 1 .
~5r +1+ m+ 2n/ ~4r + 1 .
~5r + 1 .
~6r + 1 .
~7r +1+ &amp;c. ... = 4rm ~3~ + 2 .
n~/ 6r4 xS+ ~3r + 1 .
n 2rm/ 6r4 rm + n/ 6r3 .
~r + 1 .
Let r= 1 , and we have m/ 1 .
2 .
3 .
4+m+ n/ 3 .
4 .
5 .
6+m+ 2n/ 5 .
6 .
7 .
8+ &amp;c. ... = 4m 5n/ 6 xS+ 7n 5m/ 12 .
If m= 1 , n= 1 , 1/ 1 .
2 .
3 .
4+ 1/ 3 .
4 .
5 .
6+ 1/ 5 .
6 .
7 .
8+ &amp;c. ... = 2/ 3 S 5/ 12 ; m= 1 , n= 1 , 1/ 1 .
2 .
3 .
4+ 2/ 3 .
4 .
5 .
6+ 3/ 5 .
6 .
7 .
8+ &amp;c. ... = 1/ 6 1/ 6 S ; m= 7 , n= 5 , 7/ 1 .
2 .
3 .
4+ 12/ 3 .
4 .
5 .
6+ 17/ 5 .
6 .
7 .
8+ &amp;c. ... = 1/ 2 S ; Let r= 2 , and we have m/ 1 .
3 .
5 .
7+m+ n/ 5 .
7 .
9 .
11 + m2n/ 9 .
11 .
13 .
15 + &amp;c. ... =m n/ 12 xS 19n 16m/ 288 .
If m= 1 , n= 3 , 1/ 1 .
3 .
5 .
7+ 4/ 5 .
7 .
9 .
11 + 7/ 9 .
11 .
13 .
15 + &amp;c. ... = 41/ 288 1/ 6 S ; m= 1 , n= 0 , 1/ 1 .
3 .
5 .
7+ 1/ 5 .
7 .
9 .
11 + 1/ 9 .
11 .
13 .
15 + &amp;c. ... = 1/ 12 S 1/ 18 .
m= 19 , n= 16 .
19/ 1 .
3 .
5 .
7+ 35/ 5 .
7 .
9 .
11 + 51/ 9 .
11 .
13 .
15 + &amp;c. ... = 1/ 4 S. Cor. If n : m as 4r : 3r + 2 , the sum of the series can be accurately had ; let therefore n= 4r and m= 3r + 2 , and we shall have 3r + 2/ 1 .
~r + 1 .
~2r + 1 .
~3r +1+ 7r + 2/ ~2r + 1 .
~3r + 1 .
~4r + 1 .
~5r +1+ &amp;c. ... = 1/ 2r .
~r + 1 .
If r= 1 , 5/ 1 .
2 .
3 .
4+ 9/ 3 .
4 .
5 .
6+ 13/ 5 .
6 .
7 .
9+ &amp;c. ... = 1/ 4 * ; r= 2 , 1/ 1 .
3 .
5 .
7+ 2/ 5 .
7 .
9 .
11 + 3/ 9 .
11 .
13 15 + &amp;c. ... = 1/ 96 ; r= 6 , 5/ 1 .
7 .
13 .
19 + 11/ 13 .
19 .
25 .
31 + 17/ 25 .
31 .
37 .
43 + &amp;c. ... = 1/ 336 .
PRO P. IV .
To find the sum of the infinite series m/ ~r + 1 .
~2r + 1 .
~3r + 1 .
~4r +1+ m+ n/ ~3r + 1 .
~4r + 1 .
~5r + 1 .
~6r +1+m+ 2n/ ~5r + 1 .
~6r + 1 .
~7r + 1 .
~8r +1+ &amp;c. This series resolves itself into a/ ~r + 1 .
~2r + 1 .
~3r +1a+ b/ ~2r + 1 .
~3r + 1 .
~4r +1+a+ 2b/ ~3r + 1 .
~4r + 1 .
~5r +1 &amp;c. for by reduction it becomes 3ra ~r + 1 .
b/ ~r + 1 .
~2r + 1 .
~3r + 1 .
~4r +1+ 3ra + ~3r 1 .
b/ ~3r + 1 .
~4r + 1 .
~5r + 1 .
~6r +1+ 3ra + ~7r 1 .
b/ ~5r + 1 .
~6r + 1 .
~7r + 1 .
~8r +1+ &amp;c. where the denominators are the same as in the given series , and the numerators also in arithmetic progression ; put therefore 3ra ~r + 1 .
b= m , 3ra + ~3r 1 .
b=m+ n , hence b= n/ 4r , a= 4rm + ~r + 1 .
n/ 12r2 , which , substituted in cor .
2 .
prop. 2 .
give m/ ~r + 1 .
~2r + 1 .
~3r + 1 .
~4r +1+m+ n/ ~3r + 1 .
~4r + 1 .
~5r + 1 .
~6r +1+ &amp;c. ... = ~5r + 2 .
n 4rm/ 6r4 xS+ 2rm ~4r + 1 .
n/ 6r4 + 2rm ~r 2 .
n/ 12 .
r3 .
~r +1+ 4rm ~2r 1 .
n/ 12r2 .
~r + 1 .
~2r + 1 .
* Vide DEMOIVR E ' S Mis .
Anal. page .
134 .
Cor. 1 .
In prop. 3 .
write a for m and 2b for n , and we have a/ 1 .
~r + 1 .
~2r + 1 .
~3r +1+a+ 2b/ ~2r + 1 .
~3r + 1 .
~4r + 1 .
~5r +1+ &amp;c. ... = 2ra ~3r + 2 .
b/ 3r4 xS+ ~3r + 1 .
b ra/ 3r4 ra + 2b/ 6r3 .
~r + 1 .
Also in this prop. write a+b for m , and 2b for n , and we have a+ b/ ~r + 1 .
~2r + 1 .
~3r + 1 .
~4r +1+a+ 3b/ ~3r + 1 .
~4r + 1 .
~5r + 1 .
~6r +1+ &amp;c. ... = ~3r + 2 .
b 2ra/ 3r4 xS+ ra ~3r + 1 .
b/ 3r4 + ra + 2b/ 6r3 .
~r +1+ 2ra + b/ 6r2 .
~r + 1 .
~2r + 1 .
subtract this latter series from the former , and we have a/ 1 .
~r + 1 .
~2r + 1 .
~3r +1a+ b/ ~r + 1 .
~2r + 1 .
~3r + 1 .
~4r +1+ &amp;c. ... = 4ra ~3r + 2 .
2b/ 3r4 xS+ ~3r + 1 .
2b 2ra/ 3r4 ra + 2b/ 3r3 .
~r +1 2ra + b/ 6r2 .
~r + 1 .
~2r + 1 .
Let r= 1 , and we have a/ 1 .
2 .
3 .
4a+ b/ 2 .
3 .
4 .
5+a+ 2b/ 3 .
4 .
5 .
6 &amp;c. ... = 4a 10b/ 3 xS+ 83b 32a/ 36 .
If a= 1 , b= 0 , 1/ 1 .
2 .
3 .
4 1/ 2 .
3 .
4 .
5+ 1/ 3 .
4 .
5 .
6 &amp;c. ... = 4/ 3 S 8/ 9 .
a= 3 , b= 1 , 1 1 .
2 .
4 1/ 2 .
3 .
5+ 1/ 3 .
4 .
6 &amp;c. ... = 2/ 3 S 13/ 36 .
Let r= 3 , and we have a/ 1 .
4 .
7 .
10 a+ b/ 4 .
7 .
10 .
13 +a+ 2b/ 7 .
10 .
13 .
16 &amp;c. ... = 12a 22b/ 243 xS+ 1027b 156a/ 3 .
81 .
56 If a= 1 , b= 1 , 1/ 1 .
4 .
7 .
10 2/ 4 .
7 .
10 .
13 + 3/ 7 .
10 .
13 .
16 &amp;c. ... = 871/ 3 .
81 .
56 10/ 3 .
81 S. a= 4 , b= 3 , 1/ 1 .
7 .
10 1/ 4 .
10 .
13 + 1/ 7 .
13 .
16 &amp;c. ... = 13/ 72 2/ 27 S. Cor. 2 .
If a : b as 3r + 2 : 2r , the sum of the series can be accurately found ; take ~a= 3r + 2 , and b= 2r , and we shall have 3r + 2/ 1 .
~r + 1 .
~2r + 1 .
~3r +1 5r + 2/ ~r + 1 .
~2r + 1 .
~3r + 1 .
~4r +1+ &amp;c. ... = 1/ ~r + 1 .
~2r + 1 .
If r= 1 , 5/ 1.2.3.4 7/ 2.3.4.5 + 9/ 3.4.5.6 &amp;c. ... =.1/ 60 ; r= 2 , 2/ 1.3.5.7 3/ 3.5.7.9 + 4/ 5.7.9.11 &amp;c. ... = 1/ 15 ; r= 3 , 11/ 1.4.7.10 17/ 4.7.10.13 + 23/ 7.10.13.16 &amp;c. ... = 1/ 28 .
PRO P. V. To find the sum of the infinite series m/ 1.r+1.2r+1.3r+1.4r+1 + m+n/ 2r+1.3r+1.4r+1.5r+1.6r+1 + m+2n/ 4r+1.5r+1.6r+1.7r+1.8r+1 + &amp;c. This series resolves itself into a/ 1.r+1.2r+1.3r+1 a+b/ r+1.2r+1.3r+1.4r+1 + a+2b/ 2r+1.3r+1.4r+1.5r+1 &amp;c. ; for by reduction this series becomes 4ra-b/ 1.r+1.2r+1.3r+1.4r+1 + 4ra+br-1.b/ 2r+1.3r+1.4r+1.5r+1.6r+1 + 4ra+12r-1.b/ 4r+1.5r+1.6r+1.7r+1.8r+1 + &amp;c. where the numerators are in arithmetic progression , and the denominators the same as in the given series ; assume therefore 4ra-b=m,4ra+6r-1.b=m+n , hence b= n/ 6r , a = 6rm+n/ 24r2 , which values being substituted in cor .
1 .
prop. 4 .
give m/ 1.r+1.2r+1.3r+1.4r+1 + m+n/ 2r+1.3r+1.4r+1.5r+1.6r+1 + &amp;c. ... = 2rm-2r+1.n/ 6r5 XS+ 4r+1.n-2rm/ 12r5 6rm+9n/ 72.r4.r+1 2rm+n/ 24r3.r+1.2r+1 .
Let r= 1 , and we have m/ 1.2.3.4.5 + m+n/ 3.4.5.6.7 + m+2n/ 5.6.7.8.9 + &amp;c. ... =2m-3n/ 6 XS+ 25n-16m/ 72 .
If m= 1 , n= 0 , 1/ 1.2.3.4.5 + 1/ 3.4.5.6.7 + 1/ 5.6.7.8.9+ &amp;c. ... = 1/ 3S 2/ 9 ; m= 1 , n= 1 , 1/ 1.2.3.4.5 + 2/ 3.4.5.6.7 + 3/ 5.6.7.8.9+ &amp;c. ... = 1/ 8 1/ 6 S ; m= 4 , n= 2 , 1/ 1.2.3.5 + 1/ 3.4.5.7 + 1/ 5.6.7 9+ &amp;c. ... = 1/ 3S-7/ 36 .
m= 25 , n= 16 , 25/ 1.2.3.4.5 + 41/ 3.4.5.6.7 + 57/ 5.6.7.8.9+ &amp;c. ... =1/ 3S .
Let r= 2 , and we have m/ 1.3.5.7.9 + m+n/ 5.7.9 11.13 + m+2n/ 9.11.13.15.17 + &amp;c. ... = 4m-5n/ 192 XS+ 59n-44m/ 24.120 .
Ifm=1 , n=0,1/ 1.3.5.7.9+1/ 5.7.9.11.13+1/ 9.11.13.15.17+ &amp;c. ... =1/ 48S-11/ 24.30 ; m=3 , n=2,1/ 1.5.7.9+1/ 5.9.11.13+1/ 9.13.15.17+ &amp;c. ... =1/ 96S-7/ 24.60 .
Cor. If n : m as 2r : 2r + 1 , the sum of the series can be accurately found ; assume therefore n= 2r , m= 2r + 1 , and we have 1/ 1.r+1.3r+1.4r+1+1/ 2r+1.3r+1.5r+1.6r+1+ &amp;c. ... =1/ 6.r.r+1.2r+1 .
If r= 1 , 1/ 1.2.4.5+1/ 3.4.6.7+1/ 5.6.8.9+ &amp;c. ... = 1/ 36 ; r= 3 , 1/ 1.4.10.13+1/ 7.10.16.19+1/ 13.16.22.25+ &amp;c. ... 1/ 504 .
Having thus far explained the method of summation of such series as I proposed to treat of in the first part of this paper , I trust it is not necessary to say any thing further , as the same method of proceeding will manifestly continue the series to any proposed number of factors in the denominator ; I shall therefore conclude with pointing out a remarkable property of those series whose sum can be accurately found : that when the number of factors in the denominator is even , the numerator is always equal to the sum of the two middle factors ; and when the number of factors be odd , the numerator will be equal to the middle factor , and consequently will take it out of the denominator , and leave a series whose numerators are unity , and whose denominators want the middle factor .
The method of summation of series here made use of may also be applied in investigating the sums of a great variety of other series ; but as a further application of this method would carry us beyond the limits to which this paper must be confined , I shall re-assume the subject at some future opportunity , and proceed immediately to the second part .
PART II .
PRO P. To find the sum of the infinite series p/ n.n+m ... ..n+rm + q/ n+m ... .n+r+1.m+s/ n+2m ... .n+r+2.m+ &amp;c. when the last differences of the numerators become equal to nothing .
Assume a+nb+n.n+m.c+n.n+m.n+2m.d+ &amp;c. to any number ( r ) of terms ; then , if for n we write n+ m , n+ 2m , n+ 3m , &amp;c. successively , there will result a series of quantities whose rth difference is = 0 ; substitute , therefore , this series of quantities for p , q , s , &amp;c. respectively , and the given series becomes a+ nb + n. n+ m. c+ &amp;c. / n. n+ m ... .
n+ rm +a+n+ m. b+n+ m. n+ 2m .
c+ &amp;c. / n + m ... . .
n+r+ 1 .
m+ a+n+ 2m .
b+n+ 2m .
n+ 3m c+ &amp;c. / n + 2m ... .
n+r+ 2 .
m+ &amp;c. which manifestly resolves itself into the following series a/ n. n+ m ... .
n+ rm + a/ n + m ... .
n+r+ 1 .
m+ a/ n + 2m ... .
n+r+ 2 .
m+ &amp;c. b/ n + m ... .
n+ rm + b/ n + 2m ... .
n+r+ 1 .
m+ b/ n + 3m ... .
n+r+ 2 .
m+ &amp;c. c/ n + 2m ... .
n+ rm + c/ n + 3m ... .
n+r+ 1 .
m+ c/ n + 4m ... .
n+r+ 2 .
m+ &amp;c. &amp;c. &amp;c. &amp;c. where the number of series is r , the sum of each of which being taken by a well known rule , the sum of the given series becomes a/ n. n+ m ... .
n+r 1 .
m. m. r+ b/ n + m ... .
n+r 1 .
m. m. r1+ c/ n + 2m ... .
n+r 1 .
m. m. r2+ &amp;c. where the law of continuation is manifest .
CASE 1 .
To find the sum of the infinite series 3/ 1 .
2 .
3 .
4+ 6/ 2 .
3 .
4 .
5+ 10/ 3 .
4 .
5 .
6+ 15/ 4 .
5 .
6 .
7+ &amp;c. Here n= 1 , m= 1 , r= 3 , and the third differences become = 0 ; therefore a+b+ 2c = 3 , a+ 2b + 6c = 6 , a+ 3b + 12c = 10 , consequently a= 1 , b= 1 , c= 1/ 2 , and therefore the sum sought will be 1/ 1 .
2 .
3 .
3+ 1/ 2 .
3 .
2+ 1/ 2 .
3= 11/ 36 .
CASE 2 .
To find the sum of the infinite series 1 .
2/ 1 .
3 .
5 .
7+ 2 .
3/ 3 .
5 .
7 .
9+ 3 .
4/ 5 .
7 .
9 .
11 + &amp;c. In this case n= 1 , m= 2 , r= 3 , and the 3d differences become = 0 ; therefore a+b+ 3c = 2 , a+ 3b + 15c = 6 , a+ 5b + 35c = 12 , consequently a= 3/ 4 , b= 1/ 2 , c= 1/ 4 , and hence the sum of the required series becomes 3/ 1 .
3 .
5 .
2 .
3 .
4+ 1/ 3 .
5 .
2 .
2 .
2+ 1/ 5 .
2 .
4= 1/ 24 .
CASE 3 .
To find the sum of the infinite series 1/ 3 .
4 .
5 .
6 .
7 .
8+ 2/ 4 .
5 .
6 .
7 .
8 .
9+ 4/ 5 .
6 .
7 .
8 .
9 .
10 + 12/ 6 .
7 .
8 .
9 .
10 .
11 + 31/ 7 .
8 .
9 .
10 .
11 .
12 + &amp;c. Here n= 3 , m= 1 , r= 5 , and the 4th differences become = 0 ; therefore a+ 3b + 12c + 60d = 1 , a+ 4b + 20c + 120d = 2 , a+ 5b + 30c + 210d = 4 , a+ 6b + 42c + 336d = 12 , consequently a= 54 , b= 47 , c= 12 , d= 5/ 6 , therefore the sum of the given series is -46/ 3 .
4 .
5 .
6 .
7 .
5+ 47/ 4 .
5 .
6 7 .
4 12/ 5 .
6 .
7 .
3+ 5/ 6 .
7 .
2 .
6= 61/ 50400 .
CASE 4 .
To find the sum of the infinite series 1/ 1 .
4 .
7 .
10 + 9/ 4 .
7 .
10 .
13 + 25/ 7 .
10 .
13 .
16 + &amp;c. Here n= 1 , m= 3 , r= 3 , and the 3d differences are = 0 ; therefore a+b+ 4c = 1 , a+ 4b + 28c = 9 , a+ 7b + 70c = 25 , consesequently a= 1/ 9 , b= -8/ 9 , c= 4/ 9 ; ~ the sum of the given series will be 1/ 1 .
4 .
7 .
3 .
3 .
9 8/ 4 .
7 .
3 .
2 .
9+ 4/ 7 .
3 .
9= ~ .
CASE 5 .
To find the Sum of the infinite Series 1/ 1 .
3 .
5 .
7 .
9+ 4/ 3 .
5 .
7 .
9 .
11 + 10/ 5 .
7 .
9 .
11 .
13 + 20/ 7 .
9 .
11 .
13 ~ 15 .
In this case n= 1 , m= 2 , r= 4 , and the 4th differences become = 0 ; therefore a+b+ 3c + 15d = 1 , a+ 3b + 15c + 105d = 4 , a+ 5b + 35c + 315d = 10 , a+ 7b + 63c + 693d = 20 , consequently a= ~ , b= 3/ 16 , c= 1/ 16 , d= 1/ 48 , and hence the sum of the given series becomes 5/ 1 .
3 .
5 .
7 .
2 .
4 .
16 + 3/ 3 .
5 .
7 .
2 .
3 .
16 + 1/ 5 .
7 .
2 .
2 .
16 + 1/ 7 .
2 .
48 = 1/ 384 .
This proposition may also be applied to find the sum of all those series whose numerators being unity , the denominators shall be deficient by any number of corresponding terms , however taken : for as the product of all such factors must form a progression , whose differences will become equal to nothing , if such products be assumed for the numerators of the given series having its factors completed , another series will be formed equal to the given series , whose sum can be found by this proposition .
CASE 1 .
To find the sum of the infinite series 1/ 1 .
2 .
4 .
6+ 1/ 2 .
3 .
5 .
7+ 1/ 3 .
4 .
~ .
~+ &amp;c. By completing the factors in the denominators , and multiplying the numerators by the same quantities the given series ~ 15/ 1 .
2 .
3 .
4 .
5 .
6+ 24/ 2 .
3 .
4 .
5 .
6 .
7+ 35/ 3 .
4 .
5 .
6 .
7 .
8+ &amp;c. in which case n= 1 , m= 1 , r= 5 , and the 3d differences become = 0 ; therefore a+b+ 2c = 15 , a+ 2b + 6c = 24 , a+ 3b + 12c = 35 , consequently a= 8 , b= 5 , c= 1 , and therefore the sum of the series required is 8/ 1 .
2 .
3 .
4 .
5 .
5+ 5/ 2 .
3 .
4 .
5 .
4+ 1/ 3 .
4 .
5 .
3= 211/ 7200 .
CASE 2 .
To find the sum of the infinite series 1/ 1 .
5 .
11 + 1/ 3 .
7 .
13 + 1/ 5 9 .
15 + &amp;c. This series , when completed , becomes 189/ 1 .
3 .
5 .
7 .
9 .
11 + 495/ 3 .
5 .
7 .
9 .
11 .
13 + 1001/ 5 .
7 .
9 .
11 .
13 .
15 + 1755/ 7 .
9 .
11 .
13 .
15 .
17 + &amp;c. where n= 1 , m= 2 , r= 5 , and the 4th differences are = 0 ; therefore a+b+ 3c + 15d = 189 , a+ 3b + 15c + 105d = 495 , a+ 5b + 35c + 315d = 1001 , a+ 7b + 63c + 693d = 1755 , consequently a= 96 , b= 48 , c= 10 , d= 1 ; and hence the sum of the given series is 96/ 1 .
3 .
5 .
7 .
9 .
2 .
5+ 48/ 3 .
5 .
7 .
9 .
2 .
4+ 10/ 5 .
7 .
9 .
2 .
3+ 1/ 7 .
9 .
2 .
2= 487/ 18900 .
CASE 3 .
To find the sum of the infinite series 1/ 1 .
4 .
13 .
16 + 1/ 4 .
7 .
16 .
19 + 1/ 7 .
10 .
19 .
22 + &amp;c. This series resolves itself into 70/ 1 .
4 .
7 .
10 .
13 .
16 + 130/ 4 .
7 .
10 .
13 .
16 .
19 + 208/ 7 .
10 .
13 .
16 .
19 .
22 + &amp;c. where n= 1 , m= 3 , r= 5 , and the 3d differences = 0 ; therefore a+b+ 4c = 70 , a+ 4b + 28c = 130 , a+ 7b + 70c = 208 , from whence a= 54 , b= 12 , c= 1 ; therefore the sum of the given series is 54/ 1 .
4 .
7 .
10 .
13 .
3 .
5+ 12/ 4 .
7 .
10 .
13 .
3 .
4+ 1/ 7 .
10 .
13 .
3 .
3= 227/ ~ .
By this proposition we may also investigate the sum of the series when there are any number of deficient terms in the denominators , and where the last differences of the numerators become equal to nothing ; for if the factors in the denominators be completed , and the numerators be multiplied by the same quantities , their differences will still become equal to nothing .
CASE 1 .
To find the sum of the infinite series 1/ 1 .
3 .
4 .
6+ 3/ ~ .
4 .
5 .
7+ 6/ 3 .
5 .
6 .
8+ 10/ 4 .
6 .
7 .
9+ 15/ 5 .
7 .
8 .
10 + &amp;c. This series , by completing the factors in the denominators and multiplying the numerators by the same quantities , becomes 10/ 1 .
2 .
3 .
4 .
5 .
6+ 54/ 2 .
3 .
4 .
5 .
6 .
7+ 168/ 3 .
4 .
5 .
6 .
7 .
8+ &amp;c. in which case n= 1 , m= 1 , r= 5 , and as the 5th differences are = 0 ; ~a+b+ 2c + 6d + 24e = 10 , a+ 2b + 6c + 24d + 120e = 54 , a+ 3b + 12c + 60d + 360e = 168 , a+ 4b + 20c + 120d + 840e = 400 , a+ 5b + 30c + 210d + 1680e = 810 , from whence a= 0 , b= 0 , c= 1 , d= 0 , e= 1/ 2 , consequently the sum of the given series is = 1/ 3 .
4 .
5 .
3+ 1/ 5 .
2= 17/ 180 .
CASE 2 .
To find the sum of the infinite series 1/ 1 .
3 .
7 .
9+ 5/ 3 .
5 .
9 .
11 + 11/ 5 .
7 .
11 .
13 + 19/ 7 .
9 .
13 .
15 + &amp;c. By proceeding as before this series becomes 5/ 1 .
3 .
5 .
7 .
9+ 35/ 3 .
5 .
7 .
9 .
11 + 99/ 5 .
7 .
9 .
11 .
13 + &amp;c. where n= 1 , m= 2 , r= 4 , and the 4th differences = 0 ; therefore a+b+ 3c + 15d = 5 , a+ 3b + 15c + 105d = 35 , a+ 5d + 35c + 315d = 99 , a+ 7b + 63c + 693d = 209 , consequently a= 1 , b= 3/ 4 , c= 1/ 2 , d= 1/ 4 , therefore the sum of the given series = -1/ 1 .
3 .
5 .
7 .
2 .
4+ 3/ 3 .
5 .
7 , 2 .
3 .
4+ 1/ 5 .
7 .
2 .
2 .
2+ 1/ 7 .
2 .
4= 3/ 140 .
By a method similar to that made use of in this proposition may any number of factors be taken from the denominators of those series delivered in part the first , and also from a great variety of others ; but as the examples here given must be sufficient to point out the method of proceeding in all other cases , we will proceed to the third part .
PART III .
THE sum of every converging infinite series , whose terms ultimately become equal to nothing , may always be exhibited by the sum of another series formed by collecting two or more terms of the former series into one .
This is not true , however , where the terms of the infinite series continally diverge , or converge to any assignable quantity , and are affected with the signs + , - , alternately : for instance , the series 1/ 2 2/ 3 + 3/ 4 4/ 5 + &amp;c. if we collect two terms into one , beginning at the first term , will become 1/ 2.3 1/ 4.5 1/ 6.7 + &amp;c. If we begin at the second term it becomes 1/ 1.2 + 1/ 3.4 + 1/ 5.6 + &amp;c. ; neither of which gives the sum of the assumed series ; but in this , and every other case of the like nature , a correction will be necessary : to determine the value of which , and from whence the necessity thereof arises , is the subject of this third part .
LEMM A. If r be any quantity whatever , then will 1/ 2r = 1/ r 1/ r + 1/ r 1/ r + &amp;c. ad infinitum .
For 1/ 2r = 1/ r+r = ( by common division ) 1/ r 1/ r + 1/ r 1/ r + &amp;c. ad infinitum .
Cor. 1 .
Hence 1/ 2r = 1/ r + 1/ r 1/ r + 1/ r &amp;c. ad infinitum .
Cor. 2 .
Hence also x/ 2v = z/ v z/ v + x/ v z/ v + &amp;c. ad infinitum ; and z/ 2v = z/ v + z/ v z/ v + z/ v &amp;c. ad infinitum .
PRO P. I. If rn + m/ n be the general term of a series formed by writing for n any series of numbers in arithmetic progression , and whose signs are alternately + and - ; then if a series be formed by col5 lecting lecting two terms into one , beginning at the first term , the sum of the series thence arising will be less than the sum of the given series by 1/ 2r .
If a series be formed by beginning at the second Term , the sum thereof will be greater than the sum of the given series by 1/ 2r .
For let n/ rn+m n+a/ n+a .
r+m beanytwosuccessivetermsoftheseries , which , if we begin to collect at the first term ( the first term being + ) will be two terms to be collected into one , and which will therefore give am/ rn +mXn+ a. r+m for a general term of the resulting series .
Let us now make n infinite , and then the denominator of this term becomes infinite , and the numerator finite ; therefore the terms of this latter series at an infinite distance becoming infinitely small , the series will there terminate .
Now , by making n infinite in the given series , the two successive general terms at an infinite distance become 1/ r 1/ r ; consequently this series is still continued after the other terminates ; and the terms of such a continuation will be ( as they begin with n/ rn +mn+ a/ n + a. r+m by making n infinite ) 1/ r 1/ r + 1/ r 1/ r + &amp;c. which will also be continued ad infin .
and whose sum by the lemma is 1/ 2r ; consequently the given series exceeds that which is formed by collecting two terms into one , beginning at the first , by 1/ 2r ; hence the sum of the latter series + 1/ 2r will be equal to the sum of the former .
If we begin to collect at the second term , then will n/ rn +m+ n+a/ n+a .
r+m be the two successive general terms of the given series to be collected into one ; consequently the continuation of the given series when n becomes infinite will be 1/ r + 1/ r 1/ r + 1/ r &amp;c. ad infinitum whose sum , by cor .
1 .
to the lem .
is 1/ 2r ; in this case , therefore , the sum of the given series is less than the sum of the series formed by collecting two terms into one , beginning at the second term , by 1/ 2r ; hence the sum of the latter series 1/ 2r will be equal to the sum of the former .
CASE 1 .
Let the given series be 1/ 2 2/ 3 + 3/ 4 4/ 5 + &amp;c. Here r= 1 , n= 1 , 2 , 3 , 4 , &amp;c. and m= 1 .
Now , if we begin to collect at the first term , the series resolves itself into 1/ 2.3 1/ 4.5 1/ 6.7 &amp;c. and the correction , to be added , being 1/ 2 , we have 1/ 2.3 1/ 4.5 1/ 6.7 &amp;c. + 1/ 2 for the sum of the given series .
Now 1/ 2.3 1/ 4.5 1/ 6.7 &amp;c. is well known to be equal to -1 + hyp .
log .
of 2 ; consequently the sum of the given series is = 1/ 2 + hyp .
log .
of 2 .
If we begin to collect at the second term , the series become .
1/ 1.2 + 1/ 3.4 + 1/ 5.6 + &amp;c. and the correction , to be subtracted , being 1/ 2 , we have 1/ 1.2 + 1/ 3.4 + 1/ 5.6 + &amp;c. 1/ 2 for the sum of the given series ; but 1/ 1.2 + 1/ 3.4 + 1/ 5.6 + &amp;c. is equal to the hyp .
log .
of 2 ; therefore the sum of the given series is = 1/ 2 + hyp .
log .
of 2 , the same as before .
CASE 2 .
Let the given series be 1/ 3 2/ 5 + 3/ 7 + 4/ 9 + &amp;c. Here r= 2 , n= 1 , 2 , 3 , 4 , &amp;c. m= 1 .
Now , if we begin to collect at the second term , the series becomes 1/ 1.3 + 1/ 5.7 + 1/ 9.11 + &amp;c. and the correction , to be subtracted , being 1/ 4 , we have 1/ 1.3 + 1/ 5.7 + 1/ 9.11 + &amp;c. 1/ 4 for the sum of the given series ; but 1/ 1.3 + 1/ 5.7 + 1/ 9.11 + &amp;c. is equal to a circular arc ( A ) of 22 o1/ 2 , whose radius is unity ; therefore the sum of the given series =A 1/ 4 .
PRO P. II .
If x+ nz/ w + nv be the general term of a series formed by writing for n any series of numbers in arithmetic progression , and whose terms are alternately + and - ; then if a series be formed by collecting two terms into one , beginning at the first term , the sum of the series thence arising will be less than the sum of the given series by z/ 2v .
If a series be formed by beginning at the second term , the sum thereof will be greater than the sum of the given series by z/ 2v .
For let x+ nz/ w + nv x+n+ a. z/ w +n+ a. v be any two successive termsof the given series , which , if we begin to collect at the first term , will be the two general terms to be collected into one , and will therefore give axv awz/ w + nv Xw+n+ a. v for a general term of the resulting series .
Let us now make n infinite and then this term will vanish , and consequently the resulting series will terminate at an infinite distance .
Now , by making n infinite in the given series , the two successive terms ( as they begin with x+ nz/ w + nv x+n+ a. z/ w +n+ a. v by making n infinite ) become z/ v z/ v ; this series , therefore , is still continued after the other terminates ; and the terms of such a continuation will be z/ v z/ v + z/ v z/ v + &amp;c. ad infinitum , and whose sum by cor .
2 .
to the lem .
is z/ 2v ; consequently the given series exceeds that which is formed by collecting two terms into one , beginning at the first , by z/ 2v ; hence the sum of the latter series + z/ 2v will be equal to the sum of the former .
Now , if we begin to collect at the second term , then will x+ nz/ w + nv +x+n+ a. z/ w +n+ a. v be two general terms of the given series to be collected into one ; consequently the continuation of the given series , when n becomes infinite , will be z/ v + z/ v z/ v + z/ v &amp;c. ad infinitum , whose sum by cor .
2 .
to the lem .
is z/ 2v ; in this case , therefore , the sum of the given series is less than the sum of the series formed by collecting two terms into one , beginning at the second term , by z/ 2v ; hence the sum of the latter series z/ 2v will be equal to the sum of the former .
CASE 1 .
Let the given series be 7/ 3 11/ 5 + 15/ 7 19/ 9 + &amp;c. Here x= 3 , z= 2 , w= 1 , v= 1 , n= 2 , 4 , 6 , 8 , &amp; e. Now , if we begin to collect at the first term , the series becomes 2/ 3.5 + 2/ 7.9 + 2/ 11.13 + &amp;c. and the correction , to be added , being 1 , we have 2/ 3.5 + 2/ 7.9 + 2/ 11.13 + &amp;c. +1 for the sum of the given series ; but if A=a circular arc of 45~ whose radius is unity , it is well known that 2/ 3.5 + 2/ 7.9 + 2/ 11.13 + &amp;c. =1 A ; therefore the sum of the given series is 2 A. CASE 2 .
Let the given series be 16/ 1 27/ 2 + 38/ 3 49/ 4 + &amp;c. Here w= 1 , v= 1 , x= 16 , z= 11 , n= 0 , 1 , 2 , 3 , &amp;c. Now , if we begin to collect at the first term , the series becomes 5/ 1.2 + 5/ 3.4 + 5/ 5.6 + &amp;c. and the correction , to be added , being 11/ 2 , we have 5/ 1.2 + 5/ 3.4 + 5/ 5.6 + &amp;c. + 11/ 2 for the sum of the given series ; but 5/ 1.2 + 5/ 3.4 + 5/ 6.7 + &amp;c. is equal to 5X hyp .
log .
of 2 , consequently the sum of the given series is equal to 11/ 2 +5X hyp .
log .
of 2 .
Because axv awx/ w + nv Xw+n+ a. v , the general term of the series formed by reducing two terms into one , has its numerator independent of the value of , n , it is manifest , that the numerators of that series will be all equal .
Now , if a series be assumed , the numerators of whose terms are unity , and in every other respect the same as the series in this proposition , that is , if 1/ w+ nv 1/ w +n+ a. v be two successive terms of a series , it is manifest , that if every two terms of this series be reduced into one , the general term of the resulting series will be av/ w + nv Xw+n+ a. v , where the numerator is a constant quantity av ; consequently the sum of the series whose general term is avx awz/ w + nv Xw+n+ a. v is to the sum of the series whose general term is av/ w + nv Xw+n+ a. v as vx wz to v , or in a given ratio ; whenever , therefore , the sum of the latter series can be found , the sum of the former can be found , and consequently , after proper correction , the sum of the series in this proposition can be found .
Hence , therefore , in the two cases given above in whatever arithmetic progression the numerators may proceed , the sum of the former can always be expressed by circular arcs , and the latter by the hyp .
log .
of 2 .
Hence also , as it appears from lem .
1 .
part the first , that the sum of the series 1/ 1 1/ r +1+ 1/ 2r +1 1/ 3r +1+ &amp;c. can always be expressed by circular arcs and logarithms , it is manifest , that if the numerators form any arithmetic progression , the sum of such series may be found by this proposition , and will always be exhibited by circular arcs and logarithms .
Besides the series contained in the foregoing propositions , a great variety of other series might be produced where a correction is necessary , after collecting two terms into one , in order to exhibit the true value of the given series .
As the proper correction , however , may always be found from the principles delivered in the above propositions , that is , by considering what the terms of the given series become at an infinite distance , I shall only add one or two instances more , and conclude what I at present intend to offer on this subject .
EX .
1 .
Let it be required to find the sum of the infinite series 3.4/ 1.2 4.5/ 2.3 + 5.6/ 3.4 6.7/ 4.5 + &amp;c. This , by resolving two terms into one , becomes 16/ 1.2.3 + 24/ 3.4.5 + 32/ 5.6.7 &amp;c. ; and as the terms of the given series continually approach to unity , the correction , to be added , is 1/ 2 , consequently 16/ 1.2.3 + 24/ 3.4.5 + 32/ 5.6.7 &amp;c. + 1/ 2 is equal to the sum of the given series ; but by prop. 1 .
part I. the sum of the series 16/ 1.2.3 + 24/ 3.4.5 + 32/ 5.6.7 + &amp;c. is equal to 8S 2 ( S being the hyp .
log .
2 .
) consequently the sum of the given series is 8S 1 1/ 2 .
EX .
2 .
Let it be required to find the sum of the infinite series 1.2/ 1.3 2.3/ 3.5 + 3.4/ 5.7 4.5/ 7.9 + &amp;c. This series , by resolving two terms into one , becomes 4/ 1.3.5 + 8/ 5.7.9 + 12/ 9.11.13 + &amp;c. and as the terms of the given series continually approach to 1/ 4 , the correction , to be added , will be 1/ 8 , therefore 4/ 1.3.5 + 8/ 5.7.9 + 12/ 9.11.13 + &amp;c. + 1/ 8 is = to the sum of the given series ; but by prop. 1 .
part I. the sum of 4/ 1.3.5 + 8/ 5.7.9 + 12/ 9.11.13 + &amp;c. is equal to 1/ 4 S+ 1/ 8 ( S being a circular arc of 45~ , whose radius is unity ) hence the sum of the given series is 1/ 4 S+ 1/ 4 .
This method is not only applicable to those cases , where the given series resolves itself into another , whose sum is either accurately known or can be expressed by circular arcs and logarithms , but also to those cases where we want to approximate to the value of the given series , as it must , in general , be necessary first to render the terms of the series converging , by collecting two into one , before the operation of approximation begins , and consequently a correction of this latter is necessary in order to exhibit the value of the given series .
Page .
Line .
72 .
2 .
for comunicated read communicated 77 .
12 .
for mercury removed without causing this to rise read remove without causing the mercury to rise 81 .
9 .
for smaller towards the middle read larger towards the middle 391 .
the last line , for read + 403 .
12 .
for 54 read 46 408 .
3 .
from the bottom , for rn + m/ n read n/ rn + m~

