A further Attempt to facilitate the Resolution of Isoperimetrical Problems .
By Mr. Thomas Simpson , F. R. S. A BOUT three years ago I had the honour to lay before the Royal Society the investigation of a general rule for the resolution of isoperimetrical problems of that kind , wherein one , only , of the two indeterminate quantities enters along with the fluxions , into the equations expressing the conditions of the problem .
Under which kind are included the determination of the greatest figures under given bounds , lines of the swiftest descent , solids 3 of Read April 13 .
of the least resistance , with innumerable other cases .
But altho ' cases of this sort do , indeed , most frequently occur , and have therefore been chiefly attended to by mathematicians , others may nevertheless be proposed , such as actually arise in inquiries into nature , wherein both the flowing quantities , together with their fluxions , are jointly concerned .
The investigation of a rule for the resolution of these , is what I shall in this paper attempt , by means of the following GENERALPROPOSITIO N. Let Q , R , S , T , &amp;c. represent any variable quantities , expressed in terms of x and y ( with given coefficients ) , and let q , r , s , t , &amp;c. denote as many other quantities , expressed in terms of x and y ; It is proposed to find an equation for the relation of x and y , so that the fluent of Qq+Rr+Ss+T t , &amp;c. corresponding to a given value of x ( or y ) , may be a maximum or minimum .
N p ' ' M d p ' L r c AE P ' F P ' ' G Let A E , A F , and A G , denote any three values of the quantity x , having indefinitely small equidifferences E F , F G ; and let E L , F M , and G N , ( perpendicular to A G ) be the respective values of y , corresponding thereto ; and , supposing OF ( = FG = x ) to be denoted by e , let cM and dN ( the successive values of y ) be represented by u and w. Moreover , supposing P'p ' and P''p ' ' to be ordinates at the middle points P ' P ' ' , between E , F and F , G , let the former ( P'p ' ) be denoted ~ , and the latter ( P''p ' ' ) by ~ ; putting A P ' =a and A P ' ' = b. Then , if a and ~ ( the mean values of x and y , between the ordinates EL and F M ) be supposed to be substituted for x and y , in the given quantity Qq+Rr+Ss+T t , &amp;c. and if , instead of x and y , their equals e and u be also substituted , and the said ( given ) quantity , after such substitution , be denoted by Q'q ' + R'r ' + S 's ' + T't ' , &amp;c. it is then evident , that this quantity Q'q ' + R'r ' + S 's ' + T't ' , &amp;c. will express so much of the whole required fluent , as is comprehended between the ordinates EL and F M , or as answers to an increase of OF in the value of x. And thus , if b and ~ be conceived to be wrote for x and y , e for x , and w for y , and the quantity resulting be denoted by Q''q ' ' + R''r ' ' + S' 's ' ' + T''t ' ' , &amp;c. this quantity will , in like manner , express the part of the required fluent corresponding to the interval F G. Whence that part answering to the interval EG will consequently be equal to Q'q ' + R'r ' &amp;c. + Q''q ' ' + R''r ' ' &amp;c. But it is manifest , that the whole required fluent cannot be a maximum or minimum , unless this part , supposing the bounding ordinates E L , GN to remain the same , is also a maximum or minimum .
Hence , in order to determine the fluxion of this expression ( Q'q ' + R'r ' &amp;c. Q''q ' ' + R''r ' ' &amp;c. ) which must , of consequence , be equal to nothing , let the fluxions of Q ' and q ' ( taking ~ and u as variable ) be denoted by Q~ and q~ u ; also let R~ ~ and r~ u denote the respective fluxions of R ' and r ' ; and let , in like manner , the fluxions of Q ' ' , q ' ' , R ' ' , r ' ' , &amp;c. be represented by Q~ ~ , q~ w , R~ ~ , r~ w , &amp;c. respectively .
Then , by the common rule for finding the fluxion of a rectangle , the fluxion of our whole expression ( Q'q ' + R'r ' &amp;c. + Q''q ' ' + R''r ' ' &amp;c. ) will be given equal to Q ' q~ u+ q ' Q~ ~+ R ' r~ u+ r ' R~ ~ &amp;c. + Q''q~ w+ q ' ' Q~ ~+ R ' ' r~ w+ r ' ' R~ ~ &amp;c. = ~ .
But u+w being =GNE L , and ~a= GNE L/ 2 ( a constant quantity ) , we therefore have w= u , and ~= ~ : also u being ( = 2r p ' ) =2~ 2E L , thence will u=2 ~ : which values being substituted above , our equation , after the whole is divided by ~ , will become 2 Q ' q~ + q ' Q~ +2 R ' r~ + r ' R~ , &amp;c. 2 Q''q~ + q ' ' Q~ 2 R ' ' r~ + r ' R~ , &amp;c. = ~ ; or , Q''q~ Q'q~ + R''r~ R'r~ &amp;c. = q ' Q~ + q''Q~/ 2 + r ' R~ + r ' ' R~/ 2 , &amp;c. But Q''q~ Q'q~ , the excess of Q''q~ above Q'q~ , is the increment or fluxion ( answering to the increment , or fluxion , x ) arising by substituting b for a , ~ for ~ , and w for u. Moreover , with regard to the quantities on the other side of the equation , it is plain , seeing the difference of q ' Q~ and q ' ' Q~ is indefinitely little in comparison of their sum , that q ' Q~ may be substituted in the room of q ' Q~ + q ' ' Q~/ 2 , &amp;c. which being done , our equation will stand thus : Flux .
Q'q~ + R'r~ &amp;c. = q ' Q~ + r ' R~ &amp;c. But q ' Q~ + r ' R~ &amp;c. represents ( by the preceding notation ) the fluxion of q ' Q ' + r ' R ' &amp;c. ( or of Qq +Rr &amp;c. ) arising by substituting ~ for y , making ~ alone variable , and casting off ~ .
If , therefore , that fluxion be denoted by ~ , we shall have flux .
Q'q~ + R'r~ &amp;c. = ~ , and consequently Q'q~ + R'r~ &amp;c. = ~ .
But Qq~ + R ' r~ &amp;c. ( by the same notation ) appears to be the fluxion of Q'q ' + R'r ' &amp;c. ( or of Qq+Rr &amp;c. ) arising by substituting u for y , making u alone variable , and casting off u. Whence the following GENERALRUL E. Take the fluxion of the given expression ( whose fluent is required to be a maximum or minimum ) making y alone variable ; and , having divided by y , let the quotient be denoted by v : Then take , again , the fluxion of the same expression , making y alone variable , which divide by y ; and then this last quotient will be = v. When y is not found in the quantity given , v will then be = 0 ; and , consequently , the expression for v , equal to nothing also .
But if y be absent , then will v= 0 , and consequently the value of v=a constant quantity .
It is also easy to comprehend , that , instead of y and y , x and x may be made successively variable .
Moreover , should the case to be resolved be confined to other restrictions , besides that of the maximum or minimum , such as , having a certain number of other fluents , at the same time , equal to given quantities , still the same method of solution may be applied , and that with equal advantage , if from the particular expressions exhibiting all the several conditions , one general expression composed of them all , with unknown ( but determinate ) coefficients , be made use of .
In order to render this matter quite clear , let A , B , C , D , &amp;c. be supposed to represent any quantities expressed in terms of x , y , and their fluxions , and let it be required to determine the relation of x and y , so that the fluent of Ax shall be a maximum , or minimum , when the cotemporary fluents of B x , C x , D x , &amp;c. are , all of them , equal to given quantities .
It is evident , in the first place , that the fluent of Ax+bBx+cCx+dD x , &amp;c. ( b , c , d , &amp;c. being any constant quantities whatever ) must be a maximum , or minimum , in the proposed circumstance : and , if the relation of x and y be determined ( by the rule ) , so as to answer this single condition ( under all possible values of b , c , d , &amp;c. ) it will also appear evident , that such relation will likewise answer and include all the other conditions propounded .
For , there being in the general expression , thus derived , as many unknown quantities b , c , d , &amp;c. ( to be determined ) as there are equations , by making the fluents of B x , C x , D x , &amp;c. equal to the values given ; those quantities may be so assigned , or conceived to be such , as to answer all the conditions of the said equations .
And then , to see clearly that the fluent of the first expression , A x , cannot be greater than arises from hence ( other things remaining the same ) let there be supposed some other different relation of x and y , whereby the conditions of all the other fluents of B x , C x , D x , &amp;c. can be fulfilled ; and let , if possible , this new relation give a greater fluent of Ax than the relation above assigned .
Then , because the fluents bB x , cC x , dD x , &amp;c. are given , and the same in both cases , it follows , according to this supposition , that this new relation must give a greater fluent of Ax+bBx+cCx+dD x , &amp;c. ( under all possible values of b , c , d , &amp;c. ) than the former relation gives : which is impossible ; because ( whatever values are assigned to b , c , d , &amp;c. ) that fluent will , it is demonstrated , be the greatest possible , when the relation of x and y is that above determined , by the General Rule .
To exemplify , now , by a particular case , the method of operation above pointed out , let there be proposed the fluxionary quantity xn ym yp / xp-1 ; wherein the relation of x and y is so required , that the fluent , corresponding to given values of x and y , shall be a maximum , or minimum .
Here , by taking the fluxion , making y alone variable ( according to the rule ) and dividing by y , we shall have p xn ym yp-1 / xp-1 = v. And , by taking the fluxion a second time , making y alone variable , and dividing by y , will be had m xn ym-1 yp / xp-1 = v. Now from these equations to exterminate v , let the latter be divided by the former ; so shall my/ py=v/ v ; and therefore a ym/ p =v ( a being a constant quantity ) .
From whence ym/ p y=a/ p~1/ p-1 ~ xx n/ p-1 ; and consequently p/ m+pxy m+p / p=a/ p~1/ p-1 x p1/ pn1xx p-n-1 / p-1 .
Let there be now proposed the two fluxions xn ym x and xp yq y , the fluent of the former being required to be a maximum , or minimum , and that of the latter , at the same time , equal to a given quantity .
Then the latter , with the general coefficient b prefixed , being joined to the former , we shall here have xn ym x+ bxp yq y. From whence , by proceeding as before , b xp yq = v , and m xn ym-1 x+qb xp yq-1 y= v. From the former of which equations , by taking the fluxions on both sides , will be had pb xp-1 yq x+ qbxp yq-1 y ( =v ) = mxn ym-1 x+ qbxp yq-1 y. Whence pb xp-1 yq =m xn ym-1 ; and therefore pb yq-m+1 = m xn-p+1 .
And in the same manner proper equations , to express the relation of x and y , may be derived , in any other case , and under any number of limitations .

