MATHEMATICAL
MISCELLANY
Edited by
Charles K. Robbins
Let us know (bricfly) of unusual and successful programs put on by your
MathematicsClub, of nes uscs of mathematics, of famousproblems solared, and so
on. Brief letters concerning the MATHEMATICSMAGAZINEor concerning other
"matters mathematical" will be welcome. Address: CHARLES K. ROBBINS,
DepartmBntof Mathematics, PurdueUlliversity, Lafayette, Indiana.
On the Danger of Induc t ion
In pointing out the danger of jumping to general conclusions from a
few particular results the Euler polynomial f(x) = x2+ x+ 41 which is
prime for x = 1,2,t*,39 (but composite for x = 40) is often cited. We
will gire here some other examples of false theorems which are true for
the first few cases, and which seem somewhat less artificial than
Euler's result.
Let f(n) be the number of regions determined in the inside of a
circle by joining n points on the circumference in all possible ways
by straight lines, no three of which are concurrent inside the circle.
It is easy to check that f(n) = 4,2,4,8,16, for n = 1,2,3,4,5; and it
seems natural to suppose that f(n) = 2n-l . Howeverif we go to n = 6 we
find that f(6) = 31. We leave it to the reader to show that the correct
result for n > 1 is actually f(n) = j2O(nJI]
CXrnext example deals with prime representing functions. Consider
g(n) = 1 + E $(j), where $(j) is Euler's totient function. For
i g 1
n = 1,2,3,4,5,6, this yields 2,3,5,7,11,13, which are the first 6
primes. Ihis together with the fact that $(n) is an arithmetic function
connected with the theory of prime nwmbersand the fact that both g(n)
and Pn have only a single even element, make6 the conjecture g(n) Pn
plausible. However g(7) = 19 shatters the illusion. g(8) - 23 and
8(9) = 19, so we might still suspect that g(n) is always prime, but
this too is false since g(10) = 33.
Finally there is the somewhat humorous experimental 'proof' that
"Every odd number is a prime!" A prime is a number divisible only
by 1 and itself. Certainly ehis is true for n = 1. We proceed step by
step and Ln each case proec primality by dividing the odd numberby all
numbers less than it. In this way we find that 3,5, and 7, are primes.
Apparently 9 = 3x 3 but this would spoil the theory which has covered
the facts perfectly so far, so we ascribe the discrepancy to exper-
imental error and continue undaunted. We find the theory to hold for
11 and 13 and then we test a few odd numbers chosen at random like
23, 37, 41, *-w. Ihus the theorem has been proved.
lJnierersi ty of North Caro1 ina . Leo Moser
This content downloaded from 194.27.18.18 on Mon, 15 Feb 2016 16:10:32 UTC
All use subject to JSTOR Terms and Conditions