It made more sense in an era when computers hadn't settled on 8-bit bytes yet. A better idea (not mine) is to separate the variable type from the storage type. There should be only one integer variable type with the same width as a CPU register (e.g. always 64-bit on today's CPUs), and storage types should be more flexible and explicit (e.g. 8, 16, 32, 64 bits, or even any bit-width).