Just to word pme's response a little more simply: Some decimal numbers can be represented by IEEE-754 exactly, others cannot. For lots more information see the article linked above, or search the Wikipedia page Floating point for the term "exact".