Unicode Demystified

Endian-ness and the Byte Order Mark

Most experienced programmers are familiar with the concept of byte order, or “endian-ness.” Most personal computer architectures of today evolved from old machine architectures that had an 8-bit word length. While their modern counterparts have longer word lengths, they still allow access to individual 8-bit fragments of a word. In other words, successive memory-address values refer to successive 8-bit fragments of words, not successive whole words. A 32-bit word on most modern machines thus takes up four memory locations.

How a 32-bit value is distributed across four 8-bit memory locations (or how a 16-bit value is distributed across two 8-bit memory locations) varies from machine architecture to machine architecture. ...