Zeroeth powers and zero factorial
To see what led mathematicians to define 0! as 1, let’s consider the similar
proposition that x0 = 1 whenever x is a nonzero real number. You might
at first have thought that raising a number to the zeroeth power should give
zero, and that might be okay if assigning a value to x0 were an isolated issue,
but it’s not: we want to define xn for all integers n (positive, negative, or
zero) in a way that makes them fit together well. One way the powers of x
with positive exponent already fit together is via the rule xm+n = xm × xn ;
we’ll break that rule unless we set x0 equal to 1, x−1 equal to 1/x, x−2 equal
to 1/x2 , and so on. Putting it differently, we could take the proposition that
xn+1 is equal to xn times x and turn it around, getting the proposition that
xn is equal to xn+1 divided by x. If we want this to continue to hold even
when n isn’t positive, x0 “should” be defined as x1 divided by x, which is
1; x−1 “should” be defined x0 divided by x, which is 1/x; x−2 “should” be
defined to equal x−1 divided by x, or 1/x2 ; and so on.
It makes no literal sense to say that 1/100 is “what you get when you
multiply −2 10’s together”; when we define 10−2 we have to let go of the
interpretation of xn as “what you get when you multiply n x’s together”
and adopt a more abstract definition of exponentiation. What do we get in
return? The very handy modern system of scientific notion for numbers, for
starters!
A similar letting-go is required when we try to figure out what 0! “should”
mean. Instead of thinking of values of n! in isolation, we have to think of
how they relate to one another. One way the existing factorial numbers 1!,
2!, etc. already fit together is via the rule that (n + 1)! is equal to n! times
n +1. Turning that around, we get the proposition that n! is equal to (n +1)!
divided by n + 1 whenever n is a positive integer. Plugging in n = 0 (not a
positive integer, but close) we find that 0! “should” be 1! divided by 1, which
is 1. This suggests that we let go of the definition “n! is the product of all
the integers from 1 up to n” and make a special rule that 0! equals 1.
What do we get from this special proviso? For one thing, the binomial
theorem! This theorem (discussed in section 2.4) expands (x + y)n as a sum
n!
of n + 1 terms of the form k!(n−k)! xk y n−k (with k running from 0 to n). For
2! 2 0 2! 1 1 2!
instance, (x + y)2 = 2!0! x y + 1!1! x y + 0!2! . This simplifies to the familiar
2 2
x + 2xy + y , but only if 0! is taken to be 1. In particular, if we were to take
0! to equal 0, then (x + y)2 would expand as 02 x2 + 21 xy + 20 y 2 , which makes
no sense (since 2/0 is undefined).