A Simple Approximation for π

Author(s): M. G. Gaba
Source: The American Mathematical Monthly, Vol. 45, No. 6 (Jun. - Jul., 1938), pp. 373-375
Published by: Mathematical Association of America
Stable URL: http://www.jstor.org/stable/2304030
Accessed: 04-02-2016 03:38 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Mathematical Association of America is collaborating with JSTOR to digitize, preserve and extend access to The American
Mathematical Monthly.

http://www.jstor.org

This content downloaded from 209.175.73.10 on Thu, 04 Feb 2016 03:38:59 UTC
All use subject to JSTOR Terms and Conditions
1938] QUESTIONS, DISCUSSIONS, AND NOTES 373

A SIMPLE APPROXIMATION FOR 7r

M. G. GABA,Universityof Nebraska

John Wallis is given credit for being the firstmathematician to express 7r

as the limit of a sequence of rationalnumbers.The earliergeometricalmethods
of inscribingand circumscribingregularpolygonsin or about a circle led to ap-
proximationsinvolvingsuccessive extractionsof square roots. It has been gen-
erallybelieved that the methodsof the "Newer Analysis" introducedduringthe
latter half of the seventeenthcenturyare necessaryin order to evaluate 7ras a
limit of rational numbers.That this is not so will be shown by provingby ele-
mentarymeans that
X = lim 7 k,

where
4k- 2 16k3 =k 1
4k2 + 1 4k2 + 1 =1 n2 + k2

Let ToTk be a quadrant of arc of a unit circle whose center is C and ToO a

T.~~~~~~~~~~~~~~t

T. c T' O

diameter.The inverseof this quadrant with respectto the unit circlewhose cen-
ter is at 0 will be the line segment To Tk'. Obviously OTJ = T' Tk' = '. Divide
To'Tk into k equal parts and call the points of division TV, T2', , T'-,. Let
the inversesof these points on the arc ToTkbe T,, T2, T3, , Tk-l.
The areas of trianglesT'_1OT' are all equal to 1/(8k). If we call this area
A and that of the triangleTn_1OTnis called An, then
An OTn_l'OTn
A OT'_ OTn'
But since

This content downloaded from 209.175.73.10 on Thu, 04 Feb 2016 03:38:59 UTC
All use subject to JSTOR Terms and Conditions
374 QUESTIONS, DISCUSSIONS, AND NOTES [June-July,
(n)21
2 =

and OTn' and OTn are reciprocalsof each other,we have

An

A
1

(OT' l)2(OT' )2 [(n-1)2

1

1 [(n)2 +
1

2k3
= [(n - 1)2 + k2] [n2 + k2]

The area of the polygon 0ToTjT2 * * Tk iS

n=k n= k

n=
En== 2k0Z
n1 [(n - 1)2 + k2][n2 + k2]
Since
1 1 2n +3 2n+ 1-
[(n - 1)2 + k2][n2 + k2] 4k2 + 1 [(n-1)2 + k2 n2 + k2]

and
n=k -2f + 3 1 +2k nk - 2n + 1

n= (n-1)2 + k2 2k2 n=1 2 + k 2

we have
n=k =
4k 1 ?k(1?+2k).
n=1 = 4k2+ I nl + k'2 4k2 + I
The area of the polygon CToT1T2 * Tk iS that of the polygon 0ToT1T2
* Tk less the area of the triangle OCTk. If we let 7rk/4 be the area of
CT0T* Tk, then
7rk n=k 1 4k3 n=k 1 k -
_ = sE n --= E +
4 n= 2 4k2 +1I n1 n2 +k 2 4k2+1I
The area of the quadrant CToTk is r/4. But we see that as k tends to infinity
the area of the polygon CToT1 * Tk approaches the area of the quadrant
CToTk as a limit,and therefore7r=limk.7rk, as we proposed to prove.
Computation shows that 7r70=3.1413; the sequence is therefore seen to be
not very rapidly convergent.
Note by theEditor. It readily followsfromthe foregoingshort article that
k k
7r
(a)
kw-hcoo nm
4 = limZ=:n2 + k2
+
whichformulais seen to be closelyrelated to

This content downloaded from 209.175.73.10 on Thu, 04 Feb 2016 03:38:59 UTC
All use subject to JSTOR Terms and Conditions
1938] QUESTIONS, DISCUSSIONS, AND NOTES 375

(b) -
4
7r
- r o1
1 dx
+X2

ITERATED AND DOUBLE LIMITS

ARTHURSARD,Queens College

1. Let s(x, y) be a functionwiththepropertythatthelimits

(1) Si(y) = lims(x, y), s2(x) = lrn s(x, y)
X-$oo v-go

We shall be concernedwith the iterated

existforall positivey and x, respectively.
and double limitsof s(x, y):
S12 = lim sl(y), s21 = lim s2(x), s = rim s(X, y)-
Y-+oo x- +00 (x,Y)-(x00 oo)

It is evident that if thedouble limit,s, exists,theiteratedlimits,S12 and S21,

exist; and all threelimitsare equal. And it is well known that a sufficientcondi-
tion for the existence of s is that either one of the limits (1) be uniformlyap-
proached over the entirerange of the remainingvariable.* We shall prove
THEOREM A. A necessaryand sufficient conditionthatthedoublelimit,s, exists
is thefollowing:Given any sequence {y1} which tends to infinity,thens(x, yj)
-*>sl(yj)uniformlyover {yi} as x tendsto infinity.
We may express the conditionbrieflythus: s(x, y)- -s1(y) uniformlyover se-
quences {y}.
A. Pringsheimhas made an extended investigationof the relation between
the iterated limitsand the double limitof a double sequence.t He introducesa
generalizationof the concept of uniformconvergencewhich permitshim to dis-
pense with the assumptionthat the limits(1) exist,However, none of his results
has the simplicityof Theorem A. Theorem A shows exactly the r6le played by
uniformconvergencein our problem.
2. To prove Theorem A we firstconsider
THEOREMB. Let {xi}, {yj} be twosequences, eachofwhichtendstoinfinity.
Then thedoublelimits =lim(i,j),(oo,,) s(xi, yj) existsif and onlyif s(xi, yj-*sj(yj)
uniformly.
Proof: The sufficiency of the conditionis well known.
To prove the necessity,observe that if s exists,S12 also exists and equals s.
Let e >O be given. Then thereis an integer,N, such that
* See, forexample, W. F. Osgood, Funktionentheorie,Leipzig, 1928, I, pp. 617-623. The ideas
of the present paper occurred to the author after study of this portion of the Funktionentheorie.
t A. Pringsheim, Zur Theorie der zweifach unendlichen Zahlenfolgen, Mathematische An-
nalen, vol. 53, 1900, pp. 289-321; and Vorlesungenuber Zahlen und Funktionenlehre,vol. I, part 1,
pp. 269-292.

This content downloaded from 209.175.73.10 on Thu, 04 Feb 2016 03:38:59 UTC
All use subject to JSTOR Terms and Conditions