XIV .
A Method of finding the Value of an infinite Series of decreasing Quantities of a certain Form , when it converges too slowly to be summed in the common Way by the mere Computation and Addition or Subtraction of some of its initial Terms .
By Francis Maseres , Esquire , F. R. S. Cursitor Baron of the Exchequer .
ARTICLE Ist .
Let a , b , c , d , e , f , g , h , &amp;c. ad insinitum , represent a decreasing progression of numbers , so that b shall be less than a , and c than b , and d than c , and so on of the following numbers , ad infinitum .
Read Feb. 13 , 1777 .
And 2dly , let these numbers be so related to each other , that they not only shall form a decreasing progression themselves , but that their differences , a-b , b-c , c-d , d-e , e-f , f-g , g-h , &amp;c. shall also form a decreasing progression , so that b-c shall be less than a-b , and c-d than b-c , and d-e than c-d , and so on of the following differences ; and likewise , that the differences of these differences ( which may be called the second differences of the original numbers a , b , c , d , e , f , g , h , &amp;c. shall form a decreasing progression ; and that the differences of those second differences , or the third differences of the original numbers a , b , c , d , e , f , g , h , &amp;c. shall also form a decreasing progression ; and in like manner , that the differences of the said third differences , or the fourth differences , of the original numbers a , b , c , d , e , f , g , h , &amp;c. and the fifth and sixth differences , and all higher differences , of the same numbers , shall also form decreasing progressions .
And 3dly , let x be a quantity of any magnitude not greater than unity .
Upon these suppositions the value of the infinite series a-bx+cxx-dx3+ex4-fx5+gx6-hx7+ &amp;c. ( in which the second , fourth , sixth , and eighth , and every following even term , is marked with the sign - , or is to be subtracted from that which immediately precedes it ) may be determined in the following manner .
Art .
2 .
Compute the first , second , third , fourth , and other subsequent differences of the co-efficients of the powers of x in this series , that is , of the numbers b , c , d , e , f , g , h , &amp;c. as far as shall be convenient .
These differences will be as follows .
First differences , b-c , c-d , d-e , e-f , f-g , g-h , &amp;c. Second differences , b-c-c-d , c-d-d-e , d-e-e-f , e-f-f-g , f-g-g-h , &amp;c. or , b-2c+d , c-2d+e , d-2e+f , e-2f+g , f-2g+h , &amp;c. Third differences , ~-2c+d-c-2d+c , c-2d+e-d-2e+f , d-2e+f-e-2f+g , e-2f+g-f-2g+h , &amp;c. or , b-3e + 3d-e , c-3d + 3e-f , d-3e + 3f-g , e-3f + 3g-h , &amp;c. Fourth differences , b-3c + 3d-e-~c-3d + 3e-f , c-3d + 3e-f-~d-3~ + 3f-g , d-3e + 3f-g-~~-3f + 3g-h , &amp;c. or , b-4c + 6d-4~ + f , c-4d + 6e-4f + g , d-4e + 6f-4g + h , &amp;c. Fifth differences , b-4c + 6d-4e + f-~c-4d + 6e-4f + g , c-4d + 6e-4f + g-~d-4e + 6f-4g + h , &amp;c. or , b-5c + 10d-10e + 5f-g , c-5d + 10e-10f + 5g-h , &amp;c. Sixth differences , b-5c + 10d-10e + 5f-g-~c-5d + 10e-10f + 5g-h , &amp;c. or , b-6c + 15d-20e + 15f-6g + h , &amp;c. Let the first difference of the first order , to wit , b c , be called DI ; and the first difference of the second order , to wit , b2c+ d , be called DII ; and the first difference of the third order , to wit , b3c+3d e , be called DIII ; and the first difference of the fourth order , to wit , b4c+6d4e+ f , be called DIV ; and the first difference of the fifth order , to wit , b5c+ 10 d 10 e+ 5f g , be called DV ; and the first difference of the sixth order , to wit , b6c+ 15 d 20 e+ 15f 6g+ h , be called DVI ; and in like manner let the first differences of the seventh , eighth , ninth , and tenth , and every following order of differences be denoted by DVII , DVIII , DIX , DX , &amp;c. that is , by the capital letter D , with a Roman numeral figure annexed to it , expressing the order of differences to which it belongs .
These things being supposed , the aforesaid infinite series abx+cxxd x3 +e x4 f x5 +g x6 b x7 + &amp;c. will be equal to the following differential series , to wit , ab x/ 1 +x DI x x/ 1 + x~2 DII x3/ 1 + x~3 DIII x4/ 1 + x~4 DIV x5/ 1 + x~5 DV x6/ 1 + x~6 DVI x7/ 1 + x~7 &amp;c. ; in which series all the terms after the first term a are marked with the sign - , or are to be subtracted from that term .
Art .
3 .
If we insert the differences themselves instead of DI , DII , DIII , DIV , DV , &amp;c. in the foregoing differential series ( which it may perhaps sometimes be convenient to do ) that series will be as follows : ab x/ 1 +x ~b cxx x/ 1 + x~2 ~b 2c+dx x3/ 1 + x~3 ~b 3c + 3d ex x4/ 1 + x~4 ~b 4c+6d4e+fx x5/ 1 + x~5 ~b 5c+ 10 d 10 e+ 5f gx x6/ 1 + x~6 ~b 6c+ 15 d 20 e+ 15f 6g +hx x7/ 1 + x~7 &amp;c. ad infinitum .
Of the convergency of the foregoing differential series .
Art .
4 .
The foregoing differential series will always converge with a considerable degree of swiftness , so that six or eight of its terms will give the value of the whole ( and consequently that of the original series abx+cxxd x3 +e x4 f x5 +g x6 h x7 + &amp;c. to which it is equal ) exact to several places of figures , even in the most difficult cases : for if x is =1 ( which is its greatest possible magnitude ) 1+x will be =1+1 or 2 , and consequently 1+ x~2 , 1+ x~3 , 1+ x~4 , 1+ x~5 , and the following powers of 1+ x , will be equal to 4 , 8 , 16 , 32 , and the following powers of 2 ; and the powers of the fraction x/ 1 +x will be equal to the powers of 1/ 2 .
Therefore the series ab x/ 1 +x DI xx/ 1 + x~ DII x3/ 1 + x~3 DIII x4/ 1 + x~4 DIV x5/ 1 + x~5 DV x6/ 1 + x~6 DVI x7/ 1 + x~7 &amp;c. will in this case be = to a b/ 2 DI/ 4 DII/ 8 DIII/ 16 DIV/ 32 DV/ 64 DVI/ 128 &amp;c. the terms of which decrease in a greater proportion than that of 1 to 2 , because the numerators a , b , DI , DII , DIII , DIV , DV , DVI , &amp;c. form a decreasing progression , and the denominators increase in the proportion of 2 to 1 .
Of the investigation of the foregoing differential series .
Art .
5 .
The foregoing differential series was investigated by first , supposing the original series abx+cxx d x3 +e x4 f x5 +g x6 h x7 + &amp;c. to be equal to another series whose terms should involve the same powers of x as the former , but in which every power of x should be multiplied into the same power of the fraction 1/ 1+x , in order to accelerate their convergency , and then inquiring what would be the co-efficients of the terms of such a series , if such a series is possible , and what would be the signs to be prefixed to them , or in what manner they would be connected with the first term , whether by addition or subtraction .
In order to this inquiry , I denoted the unknown co-efficients of the assumed series by the capital letters P , Q , R , S , T , V , &amp;c. and wrote down the terms of it near each other , without prefixing to them either of the signs + and - , but separated them from each other only by a comma ; so that the fundamental equation , from which I derived the differential series above-mentioned , was as follows : a-bx + cxx dx3 + ex4 -fx5 + gx6 hx7 + &amp;c. is = P , Q x/ 1 + x , Rx x/ 1 + x2 , S x3/ 1 + x3 , T x4/ 1 + x4 , V x5/ 1 + x5 , &amp;c. By necessary deductions from this equation it appeared that P would be equal to a ; and that all the following terms , of the assumed series , to wit , Q x/ 1 + x , Rx x/ 1 + x2 , S x3/ 1 + x3 , T x4/ 1 + x4 , V x5/ 1 + x5 , &amp;c. must be subtracted from the first term P , or a ; and that Q would be equal to b-c , or DI ; and R=b-c-c-d , or b-2c+d , or DII ; and S=b 2c +d c-2d+e , or b-3c+3d-e , or DIII ; and T=b-4c + 6d 4e + f , or DIV ; and V=b 5c + 10d 10e + 5f g , or DV ; and so on of the following co-efficients , to wit , that every new co-efficient of the assumed series is equal to the first difference of the next order of the differences derived from the original co-efficients b , c , d , e , f , g , h , &amp;c. And from hence I concluded that the series a bx + cxx dx3 + ex4 fx5 + gx6 hx7 + &amp;c. was equal to the series , a bx/ 1 +x DI xx/ 1 + x2 DII x3/ 1 + x3 DIII x4/ 1 + x4 DIV x5/ 1 + x5 DV x6/ 1 + x6 DVI x7/ 1 + x7 &amp;c. Art .
6 .
The thought of supposing the original series a bx + cxx dx3 + ex4 fx5 + &amp;c. to be equal to the series P , Q x/ 1 + x , Rx x/ 1 + x2 , S x3/ 1 + x3 , T x4/ 1 + x4 , V x5/ 1 + x5 , &amp;c. containing the powers of x multiplied into the same powers of the fraction 1/ 1 +x in order to accelerate their convergency , occurred to me in consequence of reading the late Mr. THOMAS SIMPSON'S Mathematical Dissertations , p. 62 , 63 .
concerning the summation of serieses , in which he makes a supposition of a similar kind .
Yet there seems to be a considerable difference between his proposition and that which is the subject of these pages ; for he seems to suppose his quantities p , q , r , s , t , &amp;c. ( which answer to a , b , c , d , e , &amp;c. in the notation made use of in the above serieses ) to form an increasing progression of terms , and accordingly subtracts p from q , and q from r , and r from s , and s from t , and so on ; and he seems also to suppose the differences q-p , r-q , s-r , t-s , &amp;c. to form an increasing progression , and every subsequent order of differences to form likewise an increasing progression , and accordingly subtracts q-p from r-q , and r-q from s-r , and s-r from t-s , and so on ; whereas in the foregoing series a-bx + cxx dx3 + ex4 fx5 + gx6 hx7 + &amp;c. the numbers a , b , c , d , e , f , g , h , &amp;c. are supposed to form a decreasing progression of terms , as they are most commonly found to do in the serieses that occur in the solution of mathematical or philosophical problems .
Examples of the usefulness of the foregoing differential series in finding the values of infinite serieses whose terms decrease very slowly .
Computations of the lengths of circular arcs by means of infinite serieses derived from their tangents .
Art .
7 .
It is well known , that if r be put for the radius of a circle , and t for the tangent of any arch in it that is not greater than 45~ , the magnitude of the arch whose tangent is t will be expressed by the infinite series t t3/ 3rr + t5/ 5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + t13/ 13r12 t15/ 15r14 + &amp;c. This series converges with great swiftness when the tangent is much less than the radius ; but when the tangent is nearly equal to the radius , it converges exceeding slowly ; and when it is quite equal to the radius , or the arch is equal to 45~ , the decrease of the terms is so slow as to make the computation of it in the common way , by computing the value of its initial terms , absolutely impracticable .
For Sir IS A AC NEWTON has observed concerning this series in that extreme case ( which then becomes equal to r r/ 3 + r/ 5 r/ 7 + r/ 9 r/ 11 + r/ 13 r/ 15 + &amp;c. ) and another series that is almost as slow as this , that to exhibit its value exact to twenty decimal places of figures , there would be occasion for no less than five thousand millions of its terms , to compute which would take up above a thousand years .
See Sir ISAAC NEWTON'S second letter to Mr. OLDENBURGH , dated October 24 , 1676 , in the Commercium Epistolicum , p. 159 .
In these cases therefore it will be convenient to make use of some artifice to discover the value of the series t t3/ 3rr + t5/ 5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + t13/ 13r12 t15/ 15r14 + &amp;c. ; and we shall find the application of the differential series abovementioned to be a very proper artifice for this purpose .
Art .
8 .
In order to make this application , we must consider the series t t3/ 3rr + t5/ 5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + t13/ 13r12 t15/ 15r14 + &amp;c. as being the product of the multiplication of t into the series 1 tt/ 3rr + t4/ 5r4 t6/ 7r6 + t8/ 9r8 t10/ 11r10 + t12/ 13r12 t14/ 15r14 + &amp;c. and must substitute x instead of tt/ rr in the terms of this last series , by which means it will be converted into the series 1 x/ 3 + xx/ 5 x3/ 7 + x4/ 9 x5/ 11 + x6/ 13 x7/ 15 + &amp;c. This series is of the same form with the original series above-mentioned , a bx + cxx dx3 + ex4 fx5 + gx6 hx7 + &amp;c. the numeral co-efficients 1 , 1/ 3 , 1/ 5 , 1/ 7 , 1/ 9 , 1/ 11 , 1/ 13 , 1/ 15 , &amp;c. of the powers of x in the former series answering to the literal or general co-efficients a , b , c , d , e , f , g , h , &amp;c. of the same powers in the latter series .
And these numeral co-efficients evidently form a decreasing progression , as the co-efficients a , b , c , d , e , f , g , h , &amp;c. are supposed to do ; and we shall find , upon examination , that the differences of these numeral co-efficients , of the several successive orders , also constitute decreasing progressions , as the several successive orders of differences of the co-efficients a , b , c , d , e , f , g , h , &amp;c. are supposed to do .
Consequently the series 1 x/ 3 + xx/ 5 x3/ 7 + x4/ 9 x5/ 11 + x6/ 13 x7/ 15 + &amp;c. will be equal to the differential series a bx/ 1 +x DIxx/ 1 + x2 DIIx3/ 1 + x3 DIIIx4/ 1 + x4 DIVx5/ 1 + x5 DVx6/ 1 + x6 DVIx7/ 1 + x7 &amp;c. if we suppose the letters a , b , c , d , e , f , g , h , &amp;c. to be equal to the numbers 1 , 1/ 3 , 1/ 5 , 1/ 7 , 1/ 9 , 1/ 11 , 1/ 13 , 1/ 15 , &amp;c. and DI , DII , DIII , DIV , DV , DVI , &amp;c. to be the first differences of the several orders of differences of those numbers , beginning from the second term 1/ 3 .
Now the values of these numbers , 1 , 1/ 3 , 1/ 5 , 1/ 7 , 1/ 9 , 1/ 11 , 1/ 13 , 1/ 15 , &amp;c. and of their differences of the several successive orders , beginning from the second term 1/ 3 , will , when expressed in decimal fractions , be as follows : 1 is = 1.000,000,000,000 ; 1/ 3 =.333,333,333,333 ; 1/ 5 =.200,000,000,000 ; 1/ 7 =.142,857,142,857 ; 1/ 9 =.111,111,111,111 ; 1/ 11 =.090,909,090,909 ; 1/ 13 =.076,923,076,923 ; 1/ 15 =.066,666,666,666 .
The differences of these numbers , beginning from the second term,.333,333,333,33 , are as follows : First differences .
Second differences .
.133,333,333,333 ; .076,190,476,190 ; .057,142,857,143 ; .025,396,825,397 ; .031,746,031,746 ; .011,544,011,544 ; .020,202,020,202 ; .006,216,006,216 ; .013,986,013,986 ; .003,729,603,729 ; .010,256,410,257 ; &amp;c. &amp;c. Third differences .
Fourth differences .
.050,793,650,793 ; .036,940,836,940 ; .013,852,813,853 ; .008,524,808,525 ; .005,328,005,328 ; .002,841,602,841 ; .002,486,402,487 ; &amp;c. &amp;c. Fifth differences .
Sixth differences .
.028,416,028,415 ; .022,732,822,731 ; .005,683,205,684 ; &amp;c. &amp;c. Therefore DI is =.133,333,333,333 ; DII =.076,190,476,190 ; DIII =.050,793,650,793 ; DIV =.036,940,836,940 ; DV =.028,416,028,415 ; DVI =.022,732,822,731 .
Therefore the series 1 x/ 3 + xx/ 5 x3/ 7 + x4/ 9 x5/ 11 + x6/ 13 x7/ 15 + &amp;c. is equal to the series 1 -.333,333,333,333 , x x/ 1 +x -.133,333,333,333 , x xx/ 1 + x2 -.076,190,476,190 , x x3/ 1 + x3 -.050,793,650,793 , x x4/ 1 + x4 -.036,940,836,940 , x x5/ 1 + x5 -.028,416,028,415 , x x6/ 1 + x6 -.022,732,822,731 , x x7/ 1 + x7 &amp;c. ; and consequently the product of this latter series into the tangent t will be equal to the product of the former series 1 x/ 3 + xx/ 5 x3/ 7 + x4/ 9 x5/ 11 + x6/ 13 x7/ 15 + &amp;c. into the same quantity , that is , to the product of the series 1 tt/ 3rr + t4/ 5r4 t6/ 7r6 + t8/ 9r8 t10/ 11r10 + t12/ 13r12 t14/ 15r14 + &amp;c. into the tangent t , or to the original series t t3/ 3rr + t5/ 5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + t13/ 13r12 t15/ 15r14 + &amp;c. which expresses the magnitude of the arch of which t is the tangent .
Computation of an arch of 30 degrees .
Art .
9 .
Now let t be the tangent of 30~ , which is =rx 1/ ~3 .
Then will tt be = rr/ 3 ; and tt/ rr , or x , = rr/ 3rr , or 1/ 3 .
Therefore 1+x will be =1+ 1/ 3 = 3/ 3 + 1/ 3 = 4/ 3 , and x/ 1 +x will be = 1/ 3/ 4/ 3 = 1/ 3 x 3/ 4 = 1/ 4 .
Therefore xx/ 1 + x~ will be = 1/ 16 , and x3/ 1 + x3 = 1/ 64 , and x4/ 1 + x~ = 1/ 256 , and x5 = 1/ 1024 , and x6 = 1/ 4096 , and x7 = 1/ 16384 .
Consequently the differential series will in this case be equal to 1 -.333,333,333,333 , x 1/ 4 -.133,333,333,333 , x 1/ 16 -.076,190,476,190 , x 1/ 64 -.050,793,650,793 , x 1/ 256 -.036,940,836,940 , x 1/ 1024 -.028,416,028,415 , x 1/ 4096 -.022,732,822,731 , x 1/ 16384 &amp;c. =1 -.083,333,333,333 , -.008,333,333,333 , -.001,190,476,190 , -.000,198,412,698 , -.000,036,075,036 , -.000,006,937,506 , -.000,001,387,501 , &amp;c. =1 -.093,099,955,597 , = 0.906,900,044,403 .
Therefore the series 1 x/ 3 + xx/ 5 x3/ 7 + x4/ 9 x5/ 11 + x6/ 13 x7/ 15 + &amp;c. or 1 tt/ 3rr + t4/ 5r4 t6/ 7r6 + t8/ 9r8 t10/ 11r10 + t12/ 13r12 t14/ 15r14 + &amp;c. is in this case = 0.906,900,044,403 , or ( neglecting the latter figures after the sixth place of figures , because we are sure they are not exact ) 0.906,900 .
Therefore the product of the series 1 tt/ 3rr + t4/ 5r4 t6/ 7r6 + t8/ 9r8 t10/ 11r10 + t12/ 13r12 t14/ 15r14 + &amp;c. into the tangent t is equal to 0.906,900 , xt= 0.906,900 xrx 1/ ~3 = 0.906,900 , xrx 1/ 1.732,050,8 = 0.906,900 xr x.577,350,2 = 0.523,598,8 x r ; that is , the series t t3/ 3rr + t5/ 5r5 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + t13/ 13r12 t15/ 15r14 + &amp;c. ( which expresses the magnitude of the arch of which t is the tangent ) is in this case = 0.523,598,8 x r , or an arch of 30~ is equal to 0.523,598,8 x r. Art .
10 .
This value of an arch of 30~ is exact in the six first places of figures , and errs only an unit in the seventh figure , which should be a7 instead of an 8 , the more exact value of that arch being 0.523,598,775,598 , &amp;c. And thus by the help of only eight terms of the differential series a bx/ 1+x DIxx/ 1+x~2 DIIx3/ 1+x~3 DIIIx4/ 1+x~4 DIVx5/ 1+x~5 DVx6/ 1+x~6 DVIx7/ 1+x~7 &amp;c. we have obtained the value of the series t t3/ 3rr + t5~5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + t13/ 13r12 t15/ 15r14 + &amp;c. in the case of an arch of 30 degrees , exact to six places of figures .
This degree of exactness is the same with that which we should attain by computing twelve terms of the series t t3/ 3rr + t5/ 5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + t13/ 13r12 t15/ 15r14 + &amp;c. itself , as will appear from the following calculation .
Art .
II .
The series t t3/ 3rr + t5/ 5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + t13/ 13r12 t15/ 15r14 + t17/ 17r16 t19/ 19r18 + t21/ 21r20 t23/ 23r22 + &amp;c. is =tx the series I tt/ 3rr + t4/ 5r4 t6/ 7r6 + t8/ 9r8 t10/ 11r10 + t12/ 13r12 t14/ 15r14 + t16/ 17r16 t18/ 19r18 + t20/ 21r20 t22/ 23r22 + &amp;c. = , in the case of an arch of 30~ , to rx 1/ ~3 into the series 1 1/ 3x3 + 1/ 5x9 1/ 7x27 + 1/ 9x81 1/ 11x243 + 1/ 13x729 1/ 15x2187 + 1/ 17x6561 1/ 19x19683 + 1/ 21x59049 1/ 23x177,147 + &amp;c. =rx 1/ ~3 x the series 1.000,000,000,000 , -.333,333,333,333./ 3 +.111,111,111,111 , / 5 -.037,037,037,037 , / 7 +.012,345,679,012 , / 9 -.004,115,226,337 , / 11 +.001,371,742,112 , / 13 -.000,457,247,370 , / 15 +.000,152,415,790 , / 17 -.000,050,805,263 , / 19 +.000,016,935,087 , / 21 -.000,005,645,029 , / 23 + &amp;c. =rx 1/ ~3 x the series 1.000,000,000,000 , -.111,111,111,111 , +.022,222,222,222 , -.005,291,005,291 , +.001,371,742,112 , -.000,374,111,485 , +.000,105,518,624 , -.000,030,483,158 , +.000,008,965,634 , -.000,002,673,961 , +.000,000,806,432 , -.000,000,245,436 , + &amp;c. 1 =rx 1/ ~3 x ~1.023,709,255,024 , -.116,809,630,442 , + &amp;c. =rx 1/ ~3 x.906,899,624,582 , =rx 1/ 1.732,050,8 x.906,899,624,582 , =r x.557,350,2 x.906,899,624,582 , = ( if we neglect the six latter figures of 906,899,624,582 , which we know to be not exact ) r x.577,350,2 x 906,899 , =r x.523,598,319,029,8 ; of which the first six figures.523,598 , are exact .
Computation of an arch of 45 degrees .
Art .
12 .
Now let the tangent t be equal to the radius r , or the arch ( whose magnitude is expressed by the series t t3/ 3rr + t5/ 5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + t13/ 13r12 t15/ 15r14 + &amp;c. ) be an arch of 45~ .
This series will , in this case , become equal to r r/ 3 + r/ 5 r/ 7 + r/ 9 r/ 11 + r/ 13 r/ 15 + &amp;c. of which the first eight terms will give the value of the whole exact to only one figure , as will appear by the following computation .
These terms are equal to rx the eight terms 1 1/ 3 + 1/ 5 1/ 7 + 1/ 9 1/ 11 + 1/ 13 1/ 15 , that is , to rx the eight terms 1.000,000,000,000 , -.333,333,333,333 , +.200,000,000,000 , -.142,857,142,857 , +.111,111,111,111 , -.090,909,090,909 , +.076,923,076,923 , -.066,666,666,666 , =rx 1.388,034,187,034 , -.633,766,233,765 , =r x.754,267,943,269 ; which agrees with the value of the whole series r r/ 3 + r/ 5 r/ 7 + r/ 9 r/ 11 + &amp;c. only in the highest figure 7 , the more exact value of that series being.785,398,163,397 , &amp;c. But , if we compute eight terms of the differential series which is equal to the series r r/ 3 + r/ 5 r/ 7 + r/ 9 r/ 11 + &amp;c. , we shall thereby obtain its value exact to three places of figures ; which is as great a degree of exactness as would be attained by computing about five hundred terms of the series r r/ 3 + r/ 5 r/ 7 + r/ 9 r/ 11 + r/ 13 r/ 15 + &amp;c. itself .
The computation of the eight first terms of the said differential series is as follows .
Art .
13 .
Since t is in this case = r , tt will be = rr , and consequently tt/ rr , or x , will be = 1 .
Therefore xx , x3 , x4 , x5 , and all the other powers of x , will in this case be equal to 1 , and 1+x will be equal to 1+ 1 , or 2 , and the powers of 1+x to the powers of 2 .
Therefore the fraction x/ 1 +x and its powers will be equal in this case to the fraction 1/ 2 and its powers .
Therefore the general differential series in art .
8 .
to wit , 1 -.333,333,333,333 , x x/ 1 +x -.133,333,333,333 , x xx/ 1 +x2 -.076,190,476,190 , x x3/ 1 +x3 -.050,793,650,793 , x x4/ 1 +x4 -.036,940,836,940 , x x5/ 1 +x5 -.028,416,028,415 , x x6/ 1 +x6 -.022,732,822,731 , x x7/ 1 +x7 &amp;c. will become in this case equal to 1 -.333,333,333,333 , x 1/ 2 -.133,333,333,333 , x 1/ 4 -.076,190,476,190 , x 1/ 8 -.050,793,650,793 , x 1/ 16 -.036,940,836,940 , x 1/ 32 -.028,416,028,415 , x 1/ 64 -.022,732,822,731 , x 1/ 128 &amp;c. =1 -.166,666,666,666 , -.033,333,333,333 , -.003,174,603,174 , -.001,154,401,154 , -.000,444,000,443 , -.000,177,600,177 , &amp;c. =1 -.214,474,414,470 , =.785,525,585,530 .
Therefore the series 1 x/ 3 + xx/ 5 x3/ 7 + x4/ 9 x5/ 11 + x6/ 13 x7/ 15 + &amp;c. or 1t t/ 3 rr+ t4/ 5 r4 t6/ 7 r6 + t8/ 9 r8 t10/ 11r10 + t12/ 13r12 t14/ 15r14 + &amp;c. is equal to .785,525,585,530 ; and consequently the series t t3/ 3 rr+ t5/ 5r4 t7/ 7 r6 + t9/ 9 r8 t11/ 11r10 + t13/ 13r12 t15/ 15r14 + &amp;c. is in this case =t x.785,525,585,530 , =r x.785,525,585,530 ; that is , the length of an arch of 45~ , in a circle whose radius is r , is =r x.785,525,585,530 ; which number is true to three places of figures , the more exact value of that arch being r x.785,398,163,397 , &amp;c. Art .
14 .
It has been asserted in art .
12 .
that in order to obtain the value of the series t t3/ 3 rr+ t5/ 5r4 t7/ 7 r6 + t9/ 9r8 t11/ 11r10 + t13/ 13r12 t15/ 15r14 + &amp;c. exact to 3 places of decimal figures by the mere computation of its terms , in the case of an arch of 45~ , we must compute at least 500 of its terms .
This may be proved in the following manner .
The indexes of the powers of t in that series are the odd numbers 1 , 3 , 5 , 7 , 9 , 11 , 13 , 15 , &amp;c. in their natural order ; to which if we add an unit , the numbers thereby produced will be the even numbers 2 , 4 , 6 , 8 , 10 , 12 , 14 , 16 , &amp;c. in their natural order , which are the doubles of the natural numbers , 1 , 2 , 3 , 4 , 5 , 6 , 7 , &amp;c. Therefore the number of terms of that series from the beginning of it to any given term in it , including the said term , is always half the number that is produced by adding an unit to the index of t in the said term .
Thus , if we take the term t11/ 11r10 , and add 1 to 11 , which is the index of the power of t in it , the sum will be 12 , the half of which is 6 , which is the number of terms in the series from the beginning of it to the term t11/ 11r10 , including the said term , that term being the sixth term in the series .
If therefore we take the term t999/ 999r~ , and are desirous of knowing its place in the series , or the number of terms from the beginning of the series to that term inclusively , we must add 1 to the index of the power of t in its numerator , which will increase it to 1000 ; and half this sum , to wit , 500 , will be the number of terms from the beginning of the series to the term t999/ 999r998 inclusively ; or , in other words , this term will be the 500th term of the series .
To arrive therefore at those terms of the series in which the indexes dexes of the powers of t are greater than 999 , or 1000 , or in which the numeral co-efficients of the terms ( which , by the law of this series , are equal to 1 divided by these indexes ) are less than 1/ 999 or 1/ 1000 , it is necessary to compute 500 of its terms .
Now when t is = r , and consequently the literal parts of the terms of this series do not converge at all , it is evidently necessary to carry the computation as far as those terms in which the numeral co-efficients of the terms are less than 1/ 999 or 1/ 1000 , in order to get the value of the series exact to the 1/ 999th or 1/ 1000th part of the radius r , or to the place of thousandths , or the third place of decimal figures .
Therefore , when t is = r , or the arch is = 45~ , it is necessary to compute at least 500 terms of the series t t3/ 3rr + t5/ 5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + t13/ 13r12 t15/ 15r14 + &amp;c. , in order to obtain the value of it exact to three places of decimal figures , that is , to the same degree of exactness to which we attained in art .
13 .
by computing only eight terms of the above-mentioned differential series .
Q. E. D. Art .
15 .
But the best way of applying the aforesaid differential series to the investigation of the value of one of these very slow serieses , is to compute a moderate number of the first terms of the slow series in the common way , and then apply the differential series to the computation of its remaining terms .
The advantage of this method of proceeding will be manifest , if we apply it to the foregoing example of the series t t3/ 3rr + t5/ 5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + t13/ 13r12 t15/ 15r14 + &amp;c. in the case of an arch of 45~ .
Compute therefore the first twelve terms of this series in the common way .
These terms will be as follows : t=r=rx 1.000,000,000,000 ; t3/ 3rr = r3/ 3rr = r/ 3 =r x.333,333,333,333 ; t5/ 5r4 = r5/ 5r4 = r/ 5 =r x.200,000,000,000 ; t7/ 7r6 = r7/ 7r6 = r/ 7 =r x.142,857,142,857 ; t9/ 9r8 = r9/ 9r8 = r/ 9 =r x.111,111,111,111 ; t11/ 11r10 = r11/ 11r10 = r/ 11 =r x.090,909,090,909 ; t13/ 13r12 = r13/ 13r12 = r/ 13 =r x.076,923,076,923 ; t15/ 15r14 = r15/ 15r14 = r/ 15 =r x.066,666,666,666 ; t17/ 17r16 = r17/ 17r16 = r/ 17 =r x.058,823,529,411 ; t19/ 19r18 = r19/ 19r18 = r/ 19 =r x.052,631,578,947 ; t21/ 21r20 = r21/ 21r20 = r/ 21 =r x.047,619,047,619 ; t23/ 23r22 = r23/ 23r22 = r/ 23 =r x.043,478,260,869 .
Therefore the twelve terms t t3/ 3rr + t5/ 5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r~ + t13/ 13r12 t15/ 15r14 + t17/ 17r16 t19/ 19r18 + t21/ 21r20 t23/ 23r22 are rx 1.000,000,000,000 , -rx.333,333,333,333 , +rx.200,000,000,000 , -rx.142,857,142,857 , +rx.111,111,111,111 , -rx.090,909,090,909 , = +rx.076,923,076,923 , -rx.066,666,666,666 , +rx.058,823,529,411 , -rx.052,631,578,947 , +rx.047,619,047,619 , -rx.043,478,260,869 , =rx 1.494,476,765,064 , -rx.729,876,073,581 , =r x.764,600,691,483 .
Having thus found the value of the first twelve terms of the series t t3/ 3rr + t5/ 5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + &amp;c. to be r x.764,600,691,483 , we must apply the differential series to the discovery of the value of the remaining part of this series , which is the series t25/ 25r24 t27/ 27r26 + t29/ 29r28 t31/ 31r30 + t33/ 33r32 t35/ 35r34 + t37/ 37r36 t9/ 39r38 + &amp;c. ad in~ finitum .
Now this series is equal to the product of t25/ r24 into the series 1/ 25 tt/ 27rr + t4/ 29r4 t6/ 31r6 + t8/ 33r8 t10/ 35r10 + t12/ 37r12 t14/ 39r14 + &amp;c. ~ ( putting x , as before , = tt/ rr ) to the product of t25/ r24 into ~ series 1/ 25 x/ 27 + xx/ 29 x3/ 31 + x4/ 33 x5/ 35 + x6/ 37 x7/ 39 + &amp;c. which is of the same form with the series a bx + cxx dx3 + ex4 fx5 + gx6 hx7 + &amp;c. Therefore , if we put a= 1/ 25 , b= 1/ 27 , c= 1/ 29 , d= 1/ 31 , e= 1/ 33 , f= 1/ 35 , g= 1/ 37 , h= 1/ 39 , and so on , and compute the differential series a bx/ 1+x DIxx/ 1+x~2 DIIx3/ 1+x~3 DIIIx4/ 1+x~4 DIVx5/ 1+x~5 DVx6/ 1+x~6 DVIx7/ 1+x~7 &amp;c. thence resulting , the number thereby obtained will be the value of the series 1/ 25 x/ 27 + xx/ 29 x3/ 31 + x4/ 33 x5/ 35 + x6/ 37 x7/ 39 + &amp;c. or 1/ 25 tt/ 27rr + t4/ 29r4 t6/ 31r6 + t8/ 33r8 t10/ 35r10 + t12/ 37r12 t14/ 39r14 + &amp;c. This computation is as follows : Here a is = 1/ 25 =.040,000,000,000 ; b= 1/ 27 =.037,037,037,037 ; c= 1/ 29 =.034,482,758,620 ; d= 1/ 31 =.032,258,064,516 ; e= 1/ 33 =.030,303,030,303 ; f= 1/ 35 =.028,571,428,571 ; g= 1/ 37 =.027,027,027,027 ; h= 1/ 39 =.025,641,025,641 .
The differences of these numbers ( beginning from the second number 1/ 27 , or.037,037,037,037 , ) are as follows : First differences .
Second differences .
.002,554,278,416 ; .000,329,584,311 ; .002,224,694,104 ; .000,269,659,891 ; .001,955,034,213 ; .000,223,432,481 ; .001,731,601,731 ; .000,187,200,187 ; .001,544,401,544 ; .000,158,400,158 .
.001,386,001,386 .
Third differences .
Fourth differences .
.000,059,924,420 ; .000,013,697,010 ; .000,046,227,409 ; .000,009,995,115 ; .000,036,232,294 ; .000,007,432,265 .
.000,028,800,028 .
Fifth differences .
Sixth differences .
.000,003,701,894 ; .000,001,139,044 .
.000,002,562,850 .
Therefore DI is =.002,554,278,416 ; DII =.000,329,584,311 ; DIII =.000,059,924,420 ; DIV =.000,013,697,010 ; DV =.000,003,701,894 ; DVI =.000,001,139,044 .
Consequently the differential series a bx/ 1+x DIxx/ 1+x~2 DIIx3/ 1+x~3 DIIIx4/ 1+x~4 DIVx5/ 1+x~5 DVx6/ 1+x~6 DVIx7/ 1+x~7 &amp;c. is=to~ .040,000,000,000 , -.037,037,037,037 , x x/ 1+x -.002,554,278,416 , x xx/ 1+x~2 -.000,329,584,311 , x x3/ 1+x~3 -.000,059,924,420 , x x4/ 1+x~4 -.000,013,697,010 , x x5/ 1+x~5 -.000,003,701,894 , x x6/ 1+x~6 -.000,001,139,044 , x x7/ 1+x~7 &amp;c. But , since t is in this case = r , tt/ rr , or x , will be = 1 , and consequently x/ 1+x will be = 1/ 1+1 or 1/ 2 .
Therefore the foregoing differential series is in this case equal to .040,000,000,000 , -.037,037,037,037 , x 1/ 2 -.002,554,278,416 , x 1/ 4 -.000,329,584,311 , x 1/ 8 -.000,059,924,420 , x 1/ 16 -.000,013,697,010 , x 1/ 32 -.000,003,701,894 , x 1/ 64 -.000,001,139,044 , x 1/ 128 &amp;c. =.040,000,000,000 , -.018,518,518,518 , -.000,638,569,604 , -.000,041,198,038 , -.000,003,745,276 , -.000,000,428,031 , -.000,000,057,842 , -.000,000,008,898 , &amp;c. =.040,000,000,000 , -.019,202,526,207 , &amp;c. =.020,797,473,793 &amp;c. Therefore the series 1/ 25 x/ 27 + xx/ 29 x3/ 31 + x4/ 33 x5/ 35 + x6/ 37 x7/ 39 + &amp;c. or 1/ 25 tt/ 27rr + t4/ 29r4 t6/ 31r6 + t8/ 33r8 t10/ 35r10 + t12/ 37r12 t14/ 39r14 + &amp;c. is in this case =.020,797,473,793 .
Therefore the series t25/ 25r24 t27/ 27r26 + t29/ 29r28 t31/ 31r30 + t33/ 33r32 t35/ 35r34 + t37/ 37r36 t39/ 39r38 + &amp;c. is in this case equal to t25/ r24 x.020,797,473,793 , that is , to r25/ r24 x.020,797,473,793 , or r x.020,797,473,793 ; that is , the remainder of the infinite series t t3/ 3rr + t5/ 5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + &amp;c. after the first twelve terms , is =r x.020,797,473,793 .
But we before found those first twelve terms to be =r x.764,600,691,483 .
Therefore the whole series t t3/ 3rr + t5/ 5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + &amp;c. ad infinitum is in this case =r x.764,600,691,483 , +r x.020,797,473,793 , =r x.785,398,165,276 , which is true to eight places of figures , the more exact value of that series being r x.785,398,163,397 , &amp;c. ; so that the value here found for this series , by the help of only eight terms of the differential series , differs from its true value by less than an unit in the eighth place of decimal figures , that is , by less than an hundred-millionth part of the radius r , which is a degree of exactness that could not have been attained by the mere computation of the series t t3/ 3rr + t5/ 5r4 t7/ 7r6 + t9/ 9r8 t11/ 11r10 + &amp;c. itself without computing fifty millions of its terms .
There cannot be a stronger instance of the utility of that differential series .
Computation of the series which express the time of the descent of a pendulum through the arch of a circle .
Art .
16 .
As another example of the utility of the foregoing differential series in finding the value of a series that converges very slowly , I will now apply it to the series which expresses the time of descent of a heavy body through a circular arch of 90~ , which decreases creases almost as slowly as the above-mentioned series rr/ 3+r/ 5-r/ 7+r/ 9-r/ 11+r/ 13-r/ 15+ &amp;c. , which expresses the magnitude of a circular arch of 45~ in a circle whose radius is r. Art .
17 .
If a heavy body , or a pendulum , be supposed to descend by the mere force of gravity through any arch of a circle not exceeding the arch of a quadrant , or 90~ ; and the motion be supposed to begin from a state of rest , and to continue till the bob of the pendulum , or the heavy body , comes to the lowest point of the circle ; and the radius of the circle be called r , the perpendicular height , or versed sine , of the arch through which the descent is made , be called v , and the right sine of the same arch be called s ; and ~ be put for the number 1.570,796,326,794 , &amp;c. which expresses the semi-circumference of a circle whose diameter is called 1 ; and the time of the fall of a heavy body through the versed sine v , or the perpendicular altitude of the arch through which the pendulum descends , be denoted by v ; the time of the descent of the pendulum through the said circular arch , corresponding to the versed sine or altitude v , to the lowest point of the circle , will be expressed by the product of ~x rv/ s into the series 1 1.1/ 2.2 x w/ ss + 1.1.3.3/ 2.2.4.4 x v4/ s4 1.1.3.3.5.5/ 2.2.4.4.6.6 x v6/ s6 + 1.1.3.3.5.5.7.7/ 2.2.4.4.6.6.8.8 x v8/ s8 1.1.3.3.5.5.7.7.9.9/ 2.2.4.4.6.6.8.8.10.10 x v10/ s10 + &amp;c. in which the law of the continuation of the terms is very manifest~ every new term being derived from the preceding term by multiplying it into the fraction w/ ss , and likewise into a numeral fraction , whose denominator is the square of the index of the powers of v and s in the new term , and whose numerator is the square of the odd number that is less than the said index by an unit .
Art .
18 .
Let the numeral co-efficients of the terms of this series be denoted by the capital letters of the alphabet , A , B , C , D , E , F , G , H , &amp;c. in their natural order , so that A shall be equal to 1 , and B shall be = 1.1/ 2.2 , and C=1.1.3.3/ 2.2.4.4 , and D=1.1.3.3.5.5/ 2.2.4.4.6.6 , and so on of the rest .
And we shall have B=1.1/ 2.2 A , C=3.3/ 4.4 B , D=5.5/ 6.6 C , E=7.7/ 8.8 D , F=9.9/ 10.10 E , G=11.11/ 12.12 F , H= 13.13/ 14.14 G , and so on ; and consequently the series 1 1.1/ 2.2 x w/ ss + 1.1.3.3/ 2.2.4.4 x v4/ s4 1.1.3.3.5.5/ 2.2.4.4.6.6 x v6/ s6 + &amp;c. or A Bvv/ ss + Cv4/ s4 Dv6/ s6 + Ev8/ s8 Fv10/ s10 + Gv12/ s12 Hv14/ s14 + &amp;c. will be A 1.1.Avv/ 2.2 ss + 3.3.Bv4/ 4.4 .
s4 5.5.Cv6/ 6.6 .
s6 + 7.7 Dv8/ 8.8 .
s~ 9.9 .
Ev10/ 10.10 .
s10 + 11.11.Fv12/ 12.12 .
s12 13.13.Gv14/ 14 14 .
s14 + &amp;c. or A 1Avv/ 4ss + 9Bv4/ 16s4 25Cv6/ 36s6 + 49Dv8/ 64s8 81Ev10/ 100s10 + 121Fv12/ 144s1~ 169Gv14/ 196s14 + &amp;c. or , if we convert the co-efficients of the terms into decimal fractions , 1 -.250,000,000,000 , xvv/ ss +.140,625,000,000 , x v4/ ~ -.097,656,250,000 , xv6/ s6 +.074,768,066,406 , x ~/ s8 -.060,562,133,788 , xv10/ s10 +.050,889,015,196 , x v12/ s12 -.043,878,793,714 , xv14/ s14 + &amp;c. The co-efficients of these terms decrease so slowly ( especially after the first twelve or fourteen terms ) that , when the versed sine v is very nearly equal to the right sine s ( as is the case when the arch through which the heavy body descends is nearly equal to 90~ , or the arch of a whole quadrant of a circle ) it would be necessary to compute a vast number of the terms of the series in order to obtain its value exact to seven or eight places of figures ; and , when v is quite equal to s ( as is the case when the arch , through which the descent is made , is exactly equal to 90~ ) the computation of the value of the series to that degree of exactness in that direct manner becomes wholly impracticable .
But by the help of the differential series abovementioned its value may be found , even in this case , to that degree of exactness without much difficulty ; more especially if we compute the first twelve terms of the series in the common way , and then apply the differential series to the investigation of the remaining part of it in the same manner as in the last example .
This we shall now proceed to do .
Art .
19 .
The co-efficients of the first twelve terms of the series A 1Avv/ 4ss + 9Bv4/ 16s4 25Cv6/ 36s6 + 49Dv8/ 64s8 81Ev10/ 100s10 + 121Fv12/ 144s12 169Gv14/ 196s14 + &amp;c. are as follows : A=1= 1.000,000,000,000 ; B= 1/ 4 =.250,000,000,000 ; C= 9B/ 16 =.140,625,000,000 ; D= 25C/ 36 =.097,656,250,000 ; E= 49D/ 64 =.074,768,066,406 ; F= 81E/ 100 =.060,562,133,788 ; G= 121F/ 144 =.050,889,015,196 ; H= 169G/ 196 =.043,878,793,714 ; I= 225H/ 256 =.038,565,346,037 ; K= 289I/ 324 =.034,399,336,434 ; L= 361K/ 400 =.031,045,401,131 ; M= 441L/ 484 =.028,287,235,328 .
But when v is = s , as it is in the case of an arch of 90~ , ~v/ ss and all its powers will be = 1 , and the twelve terms A Bvv/ ss + Cv4/ s4 Dv6/ s6 + Ev8/ s8 Fv10/ s10 + Gv12/ s12 Hv14/ s14 + Iv16/ s16 + Kv18/ s18 + Lv20/ s20 Mv~/ s22 will be equal to their co-efficients A-B+C-D+E-F+GH+I-K+L-M .
Therefore in this case the first twelve terms of this series are 1.000,000,000,000 , -.250,000,000,000 , +.140,625,000,000 , -.097,656,250,000 , +.074,768,066,406 , -.060,562,133,788 , +.050,889,015,196 , -.043,878,793,714 , +.038,565,346,037 , -.034,399,336,434 , +.031,045,401,131 , -.028,287,235,328 , which are = 1.335,892,828,770 , -.514,783,749,264 , =.821,109,079,506 .
Art .
20 .
The remaining part of this series is Nv24/ s24 Ov26/ s26 + ~v~/ s28 Qv30/ s30 + Rv32/ s32 Sv34/ s34 + Tv36/ s36 Vv~/ s38 + &amp;c. or 23.23.Mv24/ 24.24.s24 25.25.Nv26/ 26.26.s26 + 27.27.Ov28/ 28.28.s28 29.29.Pv30/ 30.30.s30 + 31.31.Qv~/ 32.32.s32 33.33.Rv34/ 34.34.s34 + 35.35.Sv36/ 36.36.s36 37.37.Tv38/ 38.38.s38 + &amp;c. or .025,979,075,500 , ~ v~/ s24 -.024,019,115,661 ~ v26/ s26 +.022,334,101,169 , ~ v28/ s28 -.020,869,976,759 ~ v30/ s30 +.019,585,984,048 , ~ v32/ s32 -.018,450,810,232 , ~ v34/ s34 +.017,440,001,955 , ~ v36/ s36 -.016,534,184,679 , ~ v~/ s38 + &amp;c. ; which is = v24/ s24 ~ the series .025,979,075,500 , -.024,019,115,661 , w/ ss +.022,334,101,169 , v4/ s4 -.020,869,976,759 , v6/ s6 +.019,585,984,048 , v~/ s8 -.018,450,810,232 , v10/ s10 +.017,440,001,955 , v12/ s12 -.016,534,184,679 , v14/ s14 + &amp;c. or , if we substitute x in this last series instead of w/ ss , = v24/ s24 ~ the series .025,979,075,500 , -.024,019,115,661 , x +.022,334,101,169 , xx -.020,869,976,759 , x3 +.019,585,984,048 , x4 -.018,450,810,232 , x5 +.017,440,001,955 , x6 -.016,534,184,679 , x7 + &amp;c. Now the value of this last series may be discovered by the application of the differential series a bx/ 1+x DIxx/ 1+x~2 DIIx3/ 1+x~3 DIIIx4/ 1+x~4 DIVx5/ 1+x~5 Dvx6/ 1+x~6 DVIx7/ 1+x~7 &amp;c. in the manner following : Here a is =.025,979,075,500 ; b =.024,019,115,661 ; c =.022,334,101,169 ; d =.020,869,976,759 ; e =.019,585,984,048 ; f =.018,450,810,232 ; g =.017,440,001,955 ; and h =.016,534,184,679 .
Therefore the differences of b , c , d , e , f , g , and h , of the several successive orders , are as follows : First differences .
Second differences .
bc is =.001,685,014,492 ; .000,220,890,082 ; cd =.001,464,124,410 ; .000,180,131,699 ; de =.001,283,992,711 ; .000,148,818,895 ; of =.001,135,173,816~ .000,124,365,539 ; fg =.001,010,808,277 ; .000,104,991,001 .
gh =.000,905,817,276 .
Third differences .
Fourth differences .
.000,040,758,383 ; .000,009,445,579 ; .000,031,312,804 ; .000,006,859,448 ; .000,024,453,356 ; .000,005,078,818 .
.000,019,374,538 .
Fifth differences .
Sixth differences .
.000,002,586,131 ; .000,000,805,501 .
.000,001,780,630 .
Therefore DI is =.001,685,014,492 ; DII =.000,220,890,082 ; DIII =.000,040,758,383 ; DIV =.000,009,445,579 ; DV =.000,002,586,131 ; DVI =.000,000,805,501 .
Consequently the differential series a bx/ 1+x DIxx/ 1+x~2 DIIx3/ 1+x~3 DIIIx4/ 1+x~4 DIVx5/ 1+x~5 DVx6/ 1+x~6 DVIx7/ 1+x~7 &amp;c. is=to .025,979,075,500 , -.024,019,115,661 , ~ x/ 1+x -.001,685,014,492 , ~ xx/ 1+x~2 -.000,220,890,082 , ~ x3/ 1+x~3 -.000,040,758,383 , ~ x4/ 1+x~4 -.000,009,445,579 , ~ x5/ 1+x~5 -.000,002,586,131 , ~ x6/ 1+x~6 -.000,000,805,501 , ~ x7/ 1+x~7 &amp;c. This is the general value of the said differential series , whatever may be the value of x , or w/ ss .
But in the case here supposed of an arch of 90~ , the versed sine v is equal to the sine s ; and therefore w/ ss , or x , is = 1 , and x/ 1+x = 1/ 1+1 , or 1/ 2 .
Therefore the foregoing differential series is in this case equal to .025,979,075,500 , -.024,019,115,661 , ~ 1/ 2 -.001,685,014,492 , ~ 1/ 4 -.000,220,890,082 , ~ 1/ 8 -.000,040,758,383 , ~ 1/ 16 -.000,009,445,579 , ~ 1/ 32 -.000,002,586,131 , ~ 1/ 64 -.000,000,805,501 , ~ 1/ 128 &amp;c. =.025,979,075,500 -.012,009,557,830 , -.000,421,253,623 , -.000,027,611,260 , -.000,002,547,398 , -.000,000,295,174 , -.000,000,040,408 , -.000,000,006,292 , &amp;c. =.025,979,075,500 , -.012,461,311,985 , &amp;c. =.013,517,763,515 , &amp;c. Therefore the series a-bx+cxx-dx3+ex4-fx5+gx6-hx7+ &amp;c. , or .025,979,075,500 -.024,019,115,661 , x +.022,334,101,169 , xx -.020,869,976,759 , x3 +.019,585,984,048 , x4 -.018,450,810,232 , x5 +.017,440,001,955 , x6 -.016,534,184,679 , x7 + &amp;c. is in this case =.013,517,763,515 , &amp;c. Therefore v24/ s24 ~ this last series , or v24/ s24 ~ the series .025,979,075,500 , -.024,019,115,661 , w/ ss +.022,334,101,169 , v4/ s4-.020,869,976,759 , v6/ s6 +.019,585,984,048 , v8/ s8-.018,450,810,232 , v10/ s10 +.017,440,001,955 , v12/ s12-.016,534,184,679 , v14/ s14 + &amp;c. is in this case = v24/ s24 x.013,517,763,515 , &amp;c. ; that is , the series .025,979,075,500 , v24/ s24-.024,019,115,661 , v26/ s26 +.022,334,101,169 , v28/ s28-.020,869,976,759 , v30/ s30 +.019,585,984,048 , v32/ s32-.018,450,810,232 , x ~/ s34 +.017,440,001,955 , v36/ s36-.016,534,184,679 , v38/ s38 + &amp;c. , or N v24/ s24 O v26/ s26 +P v28/ s28 Q v30/ s30 +R v32/ s32 S v34/ s34 +T v36/ s36 V v38/ s38 + &amp;c. is in this case = v24/ s24 x.013,517,763,515 , &amp;c. = ( because v is in this case = s , and consequently v24/ s24 is = 1 ) .013,517,763,515 , &amp;c. But we before found the value of the first twelve terms of the series ABv v/ s s+C v4/ s4 D v6/ s6 +E v8/ s8 F v10/ s10 + &amp;c. to be in this case = .821,109,079,506 .
Therefore the value of the whole series ABv v/ s s+C v4/ s4 D v6/ s6 +E v8/ s8 F v10/ s10 + &amp;c. ad infinitum is in this case =.821,109,079,506 , +.013,517,763,515 , &amp;c. = 834,626,843,021 , &amp;c. of which the first eight figures.834,626,84 are exact , the error being in the ninth figure 3 , which ought to be a2 instead of a 3 , as would have appeared if we had computed another term or two of the differential series .
Art .
21 .
Since the series ABv v/ s s+C v4/ s4 D v6/ s6 +E v8/ s8 F v10/ s10 + &amp;c. , or 1 1.1 .
Av v/ 2.2 .
ss+3 3 .
Bv4/ 4.4.s4 5.5 Cv6/ 6.6 .
s6 + 7.7.D v8/ 8.8.s8 9.9.E v10/ 10.10.s10 &amp;c. is , in this case of an arch of 90~ , =.834,626,843 , - &amp;c. , or somewhat less than.834,626,843 , the product of that series into ~x rv/ s , will be = ~xrv/ sx.834,626,843 , - &amp;c. = ( because v is in this case = s ) ~xr x.834,626,843 , - &amp;c. = 1.570,796,326,794 , &amp;c. xr x.834,626,843 - &amp;c. =r x 1.311,028,779 , - &amp;c. , or something less than rx 1.311,028,779 ; which is exact to nine places of figures , the more exact value of this quantity being rx 1.311,028,777,146 , &amp;c. as appears by a computation made by Mr. STIRLING , in his admirable Treatise on the Summation of Serieses , p. 58 .
Art .
22 .
This value of the product of 1.570,796 , 326,794 , &amp;c. x rv/ s into the series 1 -1.1.A v v/ 2.2 .
s s+3.3.B v4/ 4.4.s4-5.5.C v6/ 6.6.s6+7.7 D v8/ 8.8 .
s8-9 9.E v10/ 10.10.s10+ &amp;c. , found by the foregoing processes in this extreme and most difficult case , to wit , 1.311,028,779 , &amp;c. x r , exceeds its true value , 1.311,028,777,146 , &amp;c. xr by only.000,000 , 002 , x r , or two thousand-millionth parts of the radius r ; which is indeed a most minute difference , and shows the great exactness and utility of this differential series .
Art .
23 .
Of the nine figures to which the number 1.311,028,779 , found by the foregoing process , is exact , the last eight are owing to the differential series .
For if we were to multiply the value of the first twelve terms only of the series ABv v/ s s+C v4/ s4-D v6/ s6+E v8/ s8-F v10/ s10+ &amp; c , or 1-1 1.A v v/ 2.2.s s+3.3.B v4/ 4 4.s4-5.5.C v6/ 6.6.s6+7.7.D v8/ 8.8.s8-9.9 E v10/ 10.10.s10+ &amp;c. , to wit , the number.821,109,079,506 , into ~x r , or 1.570,796 , 326,794 , &amp;c. x r , the product would be only 1.289 , &amp;c. x r , which is true to only one place of figures , the second figure being a2 instead of a 3 .
This therefore is an eminent proof of the utility of the said differential series .
Art .
24 .
In an arch of 90~ the versed sine is equal to the radius of the circle , that is , according to the foregoing notation , v is = r. Therefore by art .
17 .
together with the foregoing computation , it appears , that the time of the descent of a pendulum , or other heavy body ( moving freely from a state of rest by the force of gravity only ) through the arch of a whole quadrant of a circle is to the time of the fall through the corG g2 respondent respondent perpendicular altitude , or the radius , as 1.311,028,779 , - &amp;c. xr is to r , or as 1.311,028,779 , - &amp;c. is to 1 .
Art .
25 .
Hence we may determine the proportion of the time of descent of a pendulum through an arch of 90~ to the time of its descent through an infinitely small arch at the bottom of a quadrant , or rather ( to speak correctly ) to the limit of the time of descent through a very small but finite arch at the bottom of the quadrant , to which the said time continually approaches nearer and nearer as the said small arch is taken less and less , and to which it may be made to approach so nearly , by taking the said small arch sufficiently small , as to differ from it by less than any given quantity .
For this latter time , or limit , is known to be to the time of the fall of a heavy body through half the length of the pendulum , or half the radius of the circle , as the semi-circumference of a circle is to its diameter , that is , as the number 1.570,796,326,794 , &amp;c. is to 1 .
But the time of the fall of a heavy body through half the radius of the circle is to the time of the fall through the whole radius as 1 to ~2 , or 1.414,213 , &amp;c. Therefore , ex aequo , the said limit of the time of descent of a pendulum through a very small arch of the circle at the bottom of the quadrant , is to the time of the fall of a heavy body through the radius of the circle , or the whole length of the pendulum , as 1.570,796 , &amp;c. is to 1.414,213 , &amp;c. But we have seen in the last article that the time of the fall of a heavy body through the radius of the circle is to the time of descent of a pendulum through the arch of a whole quadrant as 1 to 1.311,028,779 , &amp;c. Theresore the limit of the time of descent of a pendulum through a very small arch at the bottom of the quadrant is to the time of descent through the arch of the whole quadrant as 1.570,796,326 , &amp;c. x1 is to 1.414,213 , &amp;c. x 1.311,028,779 , &amp;c. , or as 1.570,796,326 , &amp;c. is to 1.414,213 , &amp;c. x 1.311,028,779 , &amp;c. , that is , by art .
21 .
as 1.570,796,326 , &amp;c. is to 1.414,213 , &amp;c. x 1.570,796,326 , &amp;c. x.834,626,843 , &amp;c. , or as 1 to 1.414,213 , &amp;c. x.834,626,843 , &amp;c. , or as 1 to 1.180,340 , &amp;c. , or , in smaller numbers , as 1 to 1.180 , or as 1000 to 1180 , or as 100 to 118 , or as 50 to 59 .
Art .
26 .
This proportion of the times of the descent of a pendulum through an infinitely small arch at the bottom of the quadrant , and through the arch of the whole quadrant , agrees pretty nearly with that assigned for them by Mr. HUYGENS in the preface to his admirable Treatise on Pendulum-clocks , or De Horologio Oscillatorio , which is that of 29 to 34 .
For 50 is to 59 as 29 is to 34.2 , or 341/ 5 ; or ( neglecting the fraction 1/ 5 ) as 29 is to 34 ; Mr. HUYGENS meaning , probably , in that place , not to express this proportion as accurately as he could , but only as nearly as it could be expressed by small whole numbers .
However , the numbers 50 and 59 express this proportion rather more accurately than 29 and 34 , and with pretty much the same degree of simplicity , and therefore , upon the whole , are somewhat to be preferred to them .
Art .
27 .
I have endeavoured to find another differential series , similar to that above described , for the purpose of investigating the value of an infinite series of this form , to wit , a+bx+cxx+dx3+ex4+fx5+gx6+hx7+ &amp;c. ( in which all the terms are marked with the sign + , or are added to the first term a ) when the co-efficients b , c , d , e , f , g , h , &amp;c. decrease very slowly , and x is very nearly equal to 1 , and the terms of the series decrease consequently so slowly as to make the summation of it in the common way , or by the mere computation and addition of its terms , almost impracticable ; but my endeavours have not been attended with success .
I may therefore , from my own experience , subscribe to the truth of what is asserted upon this subject by the very learned and ingenious Mr. JAMES STIRLING in his Treatise , entitled , Summatio Serierum , p. 17 .
to wit , that Series quarum termini sunt per vices negativi et affirmativi , sunt magis tractabiles quam alterae , ubi de Summatione agitur ; though at first sight one would be apt to imagine the reverse of this proposition to be true .

