Floating values only store a specific precision. I don't know the details, as they differ from compiler to compiler.