A binary unit of information equal to 8 bits. Unfortunately, the storage of binary numbers in computers
is not entirely standardized. Because computers store information in 8-bit
bytes (where a bit is a single binary digit), depending on
the "word size" of the machine, numbers requiring more than 8 bits
must be stored in multiple bytes. The usual FORTRAN77 integer size is 4
bytes long. However, a number represented as (byte1
byte2 byte3 byte4) in a VAX would be read and interpreted as (byte4 byte3 byte2 byte1)
on a Sun. The situation is even worse for floating-point
(real) numbers, which are represented in binary as a mantissa
and characteristic, and worse still for long (8-byte)
reals!

The naming of large multiples of bytes follows standard SI prefixes, as summarized in the following table.

Unfortunately, there is some ambiguity in the meanings of the prefixes kilo-, mega-, etc., when applied to units of information. This arose historically out of the fact
that , so "kilobyte"
was used to mean 1024 bytes, "megabyte" to mean bytes, etc. However, such usage is now deprecated
in favor of the usual SI unit prefixes, and a special set of prefixes have been invented
for binary powers of information units, summarized in the table below.

Barrow, B. "A Lesson in Megabytes." IEEE Standards Bearer. p. 5, Jan. 1997.IEC 60027-2. Letter Symbols
to Be Used in Electrical Technology--Part 2: Telecommunications and Electronics,
2nd ed. November 2000.National Institute of Standards. "Prefixes
for Binary Multiples." http://physics.nist.gov/cuu/Units/binary.html.