This is because binary floating-point values can only approximate common decimal numbers. A value of 0.1 (one tenth), for example, requires an infinitely recurring binary pattern of 0s and 1s.