Finding something worth knowing…

Science

A byte was not always eight bits, and its spelling is a deliberate typo

Werner Buchholz coined the word byte in June 1956 while designing IBM's Stretch computer, spelling it with a y so nobody would mistype bite as bit. For years afterward a byte could hold six, nine or other numbers of bits, and machines using sizes from 1 to 48 bits all existed.

A byte began as the bits needed to encode one text character, which is why it became the smallest chunk of memory many computers can address. In the 1960s six-bit and nine-bit bytes were common, packed into words of 12, 18, 36 or 60 bits, and some of those legacy systems survived into the 21st century. Before byte caught on, engineers sometimes called such bit groups syllables or slabs. Louis Dooley claimed a separate coinage for four-bit groups while working on the SAGE air-defence system at MIT Lincoln Laboratory in 1956 or 1957.

Character codes drove the change. Early machines used four-bit decimal codes and six-bit military sets such as the Army's FIELDATA. In 1963 the American Standard Code for Information Interchange, ASCII, stretched this to seven bits, adding lowercase letters and control signals like line feed, and replaced clashing teleprinter codes across government and universities. At the same time IBM gave its System/360 line an eight-bit code called EBCDIC. The popularity of that system made eight-bit storage universal, even though the two codes differ in detail. AT&T's long-distance digital phone lines of the early 1960s also used eight-bit encoding.

Donald Knuth captured the transition. In the first volume of The Art of Computer Programming in 1968, his imaginary MIX computer used a byte of unspecified size holding between 64 and 100 values, six bits on a binary machine. He later noted that since roughly 1975 the word had come to mean exactly eight binary digits.

Eight bits is a handy power of two, giving 256 values from 0 to 255, and standards such as IEC 80000-13 now codify it. Modern processors use 32-bit or 64-bit words of four or eight bytes. Its symbol is a capital B, but network specifications like the Internet Protocol prefer octet, symbol o, which is unambiguously eight bits and never confused with the bel.

Source: Byte

Related

More in Science · All topics