The counter-argument, used by non-mathematicians (and, I must admit, me, when I first heard this a few years ago) is ``but each step has the net effect of adding an integer number of balls, so surely there should be an infinite number of balls after an infinite number of steps''.

The problem with this argument is that it misuses mathematical induction. The argument is basically:

  1. After step1, there are 9ยท1 = 9 balls.
  2. Assume that there are 9k balls after step k. Then, after step k+1, there will be 9k + 10 - 1 = 9(k+1) balls.
  3. Thus, by induction, for any natural number n >=1*, after n steps there will be 9n balls in the urn.
  4. Therefore, after an infinite number of steps, there will be an infinite number of balls in the urn.
When stated this way, it becomes relatively obvious that the error lies in going from step 3. to step 4---it holds for every natural number, yes, but `infinity' is not a natural number.

In general, not to sound like a pompous-arsed academic, but your intuitive notions about the nature of the infinite are probably wrong. At least, they do not lead to a consistent mathematical system, and are thus useless to mathematicians.

*: thus sidestepping the question of whether zero is a natural number.


Think there are an infinite number of balls left in the urn? Okay, show me a single ball that is in the urn after a countably infinite number of steps. What is its number?

Putting in nine balls and setting one aside is, when you are dealing with infinite quantities, not the same as putting in ten balls and removing the lowest-numbered one. Again, you cannot use common sense when dealing with infinity, at least not if you want an internally consistent system of mathematics.