Explain xkcd

Date: May 17th, 2011
From: Dan Hoey
Subject: Number Line

Chrisl wrote:

  Nope, its easily possible to create a number smaller than
  that with single-precision floats.
  MIN_FLOAT is 1.4E-45, and the difference (1.0f - 0.9999999f)
  (the longest 0.99... float for which this difference is
  bigger than 0.0f) evaluates to 1.1920929E-7.

Is there a float between 1.0f and 0.9999999f?  What's 1.0f-x, where x
is the largest single-precision float strictly less than 1.0f?