LVI .
An Explication of an obscure Passage in Albert Girard 's Commentary upon Simon Stevin 's Works ( Vide Les Oeuvres Mathem .
de Simon Stevin , a Leyde , 1634 , p. 169 , 170 ) ; by Mr. Simson , Professor of Mathematics at the University of Glasgow : Communicated by the Right Honourable Philip Earl Stanhope .
PUIS que je suis entre en la matiere des nombres rationaux , j'adjousteray encore deux ou trois particularitez , non encor par cy devant practiquees , comme d'expliquer les radicaux extremement pres , &amp;c. ' ' Read Dec. 20 , 1753 .
The first thing Albert Girard gives in this place is a method of expressing the ratio of the segments of a line cut in extreme and mean proportion , by rational numbers , that converge to the true ratio .
For this purpose he takes the progression 0 , 1 , 1 , 2 , 3 , 5 , 8 , 13 , 21 , &amp;c. every term of which is equal to the sum of the two terms that precede it : and says , any number in this progression has unto the following the same ratio [ nearly ] that any other has to that , which follows it .
Thus 5 has to 8 nearly the same ratio , that 8 has to 13 ; consequently , any 3 numbers next one another as 8 , 13 , 21 , nearly express the segments of a line cut in extreme and mean proportion , and the whole line ; so that 13 , 21 , ~ .
B. 13 is wrong printed for the second number ~ of 21 ) constitute stitute near enough an isosceles triangle , having the angle of a pentagon ; i. e. whose angle at the vertex is subtended by the side of a pentagon in the circle described about the triangle .
Now this will be plain , if it be shown , that the squares of the numbers in this series are alternately lesser and greater by an unit , than the product of the two numbers next them upon each side .
Thus , in the four numbers , 5 , 8 , 13 , 21 , the square of 8 is an unit lesser than the product of 5 and 13 ; but the square of 13 that next follows 8 , viz. 169 , is an unit greater than 8 times 21 , or 168 ; and so on constantly .
Case 1 .
If a , b , c , be such numbers , that 1 .
a+b=c 2 .
ac = bb +1 Then , if d be taken , so that d= b+c ; then shall bd +1=c c. Because d=b+ c ; bd +1 shall be = bb + bc + 1= ac + bc [ 2 ] which is = a+b xc= cc [ 1 ] : Ergo bd+1= cc .
Case 2 .
If a , b , c , be such that 1 .
a+b=c 2 .
ac +1= bb Then , if d be taken , so that d=b+ c ; then shall bd = cc + 1 .
Because bd = bb + bc = ac + bc +1 [ 2 .
] = a+b xc+1 = cc +1 [ 1 .
] Problem .
Having given the number a , in Case I. to find b and c , i. e. having given a to find b such that bb +1= ( ac = ) aa+ab ; then is bb ab = aa I : and therefore b=a+~ 5aa-4/ 2 .
Whence , to make ba rational integer number , 5aa-4 must be a square ; which it will be , if a= 1 ; and then b will also be 1 , and c will be 2 : and having continued the series , every number will have the properties mentioned .
The second thing which Albert Girard mentions , is a way of exhibiting a series of rational fractions , that converge to the square root of any number proposed , and that very fast .
He tells nothing about the way of forming it , and only gives the two following examples ; viz. He says , ~2 is equal nearly to 57 7/ 4 0 8 : or , if you would have it nearer , to 139 3/ 9 8 5 .
His other example is of ~ 10 , which , he says , is nearly equal to 35335 3/ 3 2877 6 ; i. e. to 103968 1/ 3 2877 6 .
And these are the fractions your lordship has turned , at first sight , into continued fractions of the same value * .
The way of making a series of rational fractions , which converge to the square root of any number proposed , in such a manner , that the square of the numerator of any of them being lessened by an unit , or , in some cases , increased by an unit , the remainder , or sum divided by the square of the denominator , shall be exactly equal to the number proposed , depends upon the following propositions : Prop. * N. B. That the continued fraction here alluded to for expressing the square root of 10 was 1/ 6 x 19-1/ 3 8 1/ 3 8 1/ 3 8 1/ 3 8 , &amp;c. ad infinitum .
Prop. 1 .
Let a be any number proposed , and b/ c be such a fraction , that bb-1/ cc = a , i. e. bb = acc + 1 , then , if two other fractions be taken , one of which is b/ ac , the first divided by the proposed number a , and the other is c/ b , the reciprocal of the first fraction ; then the fraction bb + acc/ 2bc , whose numerator is the sum of the products of the numerators , and of the denominators of the fractions b/ c and b/ ac ; and its denominator the sum of the products of the numerators , and of the denominators of the fractions b/ c and c/ b , shall have the same property with the fraction b~ei.e.~bb+acc2/ 2bc2-I=a , Because bb = acc +1 bb acc = 1 , and squaring b4 2 ab2c2 + a2 c4 = 1 .
And adding 4ab2c2 b4 +2 ab2c2 + a2c4 = 4ab2c2 +1 .
Whence ~b2+acc2-1/ ~2bc2 = a. Prop. 2 .
If b/ c be such a fraction , that bb+1/ cc=a , i.e. bb+1=acc , all other things remaining as in Prop. I. ; then shall the fraction bb+acc/ 2bc , formed as there described , be such , that ~bb+acc2-1/ ~2bc2 = a , Because bb+1=acc , then acc-bb=I ; and squaring b4-2ab2c2+a2c4=1 .
Whence , as in the foregoing , it will follow , that ~bb+acc2-1/ ~2bc2 = a. Prop. 3 .
Let the fraction b/ c be such , that bb-1/ cc = a , i. e. bb = acc + 1 ; also let d/ e be another fraction , having the same property with b/ c , i. e. such , that dd = aee + 1 .
Then , if from the fraction d/ e , and the two others mentioned in Prop. 1 .
viz. b/ ac , and c/ b , a new fraction be formed , in the same manner as the fraction bb + acc/ 2bc was formed from b/ c , and the same two b/ ac and c/ b , which fraction will be bd + ace/ cd + be ; this new fraction shall have the same property with the other two b/ c and d/ e , i. e. ~bd+ace2-1/ ~cd+be2=a .
Hypoth .
1 .
bb = acc +1 2 .
dd = aee +1 3 .
ac2 d2 = a2 c2 e2 + ac2 [ 2 .
] 4 .
b2 d2 = ab2 e2 + b2 [ 2 .
] 5 .
b2 d2 = ab2e2 + ac2+1 [ 4 , 1 .
] 6 .
b2 d2 + a2 c2 e2 = ab2e2 + a2c2e2 + ac2+1 [ 5 .
] 7 .
b2 d2 + a2 c2 e2 = ac2 d2 + ab2 e2 +1 [ 6.3 ] 8 .
b2d2+2abcde+a2c2e2=ac2d2+2abcde+ab2e2+1[7 ] .
i. e. bd+ace2 =a~ cd + be2 + 1 .
9 .
bd+ace2-1/ cd+be2=a .
Prop. 4 .
The same things being supposed as in Prop. 3 .
except that bb , instead of being equal to acc+1 , as there , is equal to acc-1 , or bb+1=acc ; it will follow , by the like steps as in Proposition 3 .
that bd+ace2+1/ cd+be2=a .
Prop. 5 .
If likewise d2 be equal to aee-1 , as well as b2= acc-1 , all other things remaining as in Proposition 3 .
then shall bd+ace2 =a~ cd+be2 + 1 , i. e. bd+ace2 1/ cd+be2=a .
Hypoth .
1 .
b2+1=acc 2 .
d2+1=aee 3 .
b2d2+b2=ab2e2 [ 2 .
] 4 .
ac2d2+ac2=a2c2e2 [ 2 .
] 5 .
b2d2+ac2=ab2e2+1 [ 3 , 1 .
] 6 .
b2d2+ac2+ac2d2=ac2d2+a2be2+1 [ 5 .
] 7 .
b2d2+a2c2e2=ac2d2+ab2e2+1 [ 6.4 .
] 8 .
b2d2+2abcde+a2c2e2=ac2d2+2abcde+ab2e2+1 [ 7 .
] i. e. bd+ace2 =a~ cd+be2 +1 .
9 .
bd+ace2 -1/ cd+be2 = a. Prop. 6 .
But if b2= acc+1 , and d2=aee-1 , all other things remaining as in Prop. 3 .
Then shall bd + ace2 +1= a~ cd+be2 .
i. e. bd+ace2 +1/ cd+be2=a .
which may be shown , as the rest were .
Now , let a be any number proposed , and let the fraction b/ c be such , that either bb-1/ cc = a , or bb+1/ cc = a , and take the fractions b/ ac and c/ b , before described ; then the series of fractions converging to ~a will be as follows : c/ b , b/ ac b/ c= the first term of the series .
bb+ace/ 2bc=d/ e the second term .
Every term is formed from the preceding ; and the 2 bd+acc/ cd+be=f/ g the third term .
fractions b/ ac and c/ b , in the bf+acg/ cf+bg=h/ k the fourth term .
same manner as the second from the first , and &amp;c. in infin .
these fractions .
And from the foregoing propositions it follows , 1 .
That if bb-1/ cc = a , then every fraction of the series shall be such .
That if from the square of its numerator be taken an unit , the remainder , divided by the square of its denominator , shall be equal to a. For , by Prop. 1 .
the fraction d/ e shall be such ; and by Prop. 3 .
the next fraction f/ g shall likewise be such ; and so all the following terms .
Example .
Let a=2 ; then the first fraction , i. e. that in the smallest numbers , b/ c , that makes bb-1/ cc = 2 , is when b =3 , and c=2 ; so that c/ b. b/ ac b/ c are And the terms following the first 3/ 2 2/ 3 .
3/ 4 3/ 2 are 17/ 12 .
99/ 70 .
577/ 408 .
3363/ 2378 .
&amp;c. 2 .
But if bb+1/ cc=a , i. e. if the first fraction b/ c of the series have the square of its numerator an unit less than acc , the multiple of the square of its denominator by the number a ; the second term shall have the square of its numerator an unit greater than the said multiple of the square of its denominator ; and the third term shall have the said square an unit lesser , and so on alternately .
For , by Prop. 2 .
the second term d/ e shall be such , that dd-1/ ee=a : and therefore , by Prop. 4 .
the third term f/ g shall be such , that ff+1/ cc=a .
And by Prop. 5 .
it follows , that the next term h/ k shall be such , that hh-1/ kk=a ; and so on alternately , by Prop. 4 .
and 5 .
Example .
Let a=2 ; then the first fraction b/ c that makes ~b+ 1/ ec = 2 , is when b= 1 , and c= 1 .
So that a/ b. a c/ b~ b/ c And the following terms are ~ 1/ 1 are 3/ 2 .
7/ 5 .
1 7/ 1 2 .
4 1/ 2 9 .
9 9/ 7 0 &amp;c. 1/ 1 .
1/ 2 ~ But if a be 13 , then the first fraction will be 1 8/ 1 5/ 1 8 .
1 8/ 6 5~1 8/ 5 .
64 9/ 1 8 0 .
2338 2/ 6 485 &amp;c. 3 .
But if the fraction b/ c be such , that bb-1/ cc=a , and if the fractions b/ a c , c/ b , be taken , from which the series is to be formed , as has been described ; then , if the first fraction of the series be made not b/ c , but some fraction d/ c , such that dd+1/ ee=a ; then shall every term of the series be such as the fraction d/ e , i. e. the square of the numerator being increased by an unit , and the sum divided by the square of the denominator , the quotient shall be equal to a. For , since bb = acc + 1 , and dd = aee 1 , by Prop. 6 .
it follows , that the next term f/ g shall be such , that ff+1/ gg=a ; and so on for every term .
Example .
Let a=2b= 3/ c = 2 ; then will b/ a c= 3/ 4 , and c/ b = 2/ 3 , and let d/ e =1 1 ; then c/ b. b/ a c~ d/ e are 2/ 3 .
3/ 4~ 1/ 1 And the other terms are 7 / 5 .
41/ 29 .
239/ 169 .
1393/ 985 .
&amp;c. To find b/ c such as makes bb I acc , i. e. acc + I= bb , recourse must be had to Lord Brouncker 's method in Dr. Wallis 's Commercium Epistolicum .

