Underscore

As early output devices (Teleprinters, CRTs and line printers) could not produce more than one character at a location, it was not possible to underscore text, so early encodings such as ITA2 and the first versions of ASCII had no underscore. IBM's EBCDIC character-coding system, introduced in 1964, added the underscore, which IBM referred to as the "break character". IBM's report on NPL (the early name of what is now called PL/I) leaves the character set undefined, but specifically mentions the break character, and gives RATE_OF_PAY as an example identifier.[4] By 1967 the underscore had spread to ASCII,[5] replacing the similarly-shaped left-arrow character, ← (see also: PIP). C, developed at Bell Labs in the early 1970s, allowed the underscore in identifiers.[6]

Underscore predates the existence of lower-case letters in many systems, so often it had to be used to make multi-word identifiers, since CamelCase (see below) was not available.

Underscores inserted between letters are very common to make a "multi word" identifier in languages that cannot handle spaces in identifiers. This convention is known as "snake case" (the other popular method is called camelCase, where capital letters are used to show where the words start).

An underscore as the first character in an ID is often used to indicate an internal implementation that is not considered part of the API and should not be called by code outside that implementation. Python uses this to indicate protected member variables of classes and a double underscore to indicate private variables. In Dart, all private properties of classes must start with an underscore; this usage is also common in other languages such as C++ even though those provide keywords to indicate that members are private. It is extensively used to hide variables and functions used for implementations in header files. In fact, the use of single underscore for this became so common that C compilers had to standardize on a double leading underscore (for instance __DATE__) for actual built-in variables to avoid conflicts with the ones in header files. Python uses double underscore to "mangle" a private id to make it much harder to refer to it, and "PHP reserves all function names starting with __ as magical."[7]

A variable named with just an underscore often has special meaning. $_ or _ is the previous command or result in many interactive shells, such as those of Python, Ruby, and Perl. In Perl, @_ is a special array variable that holds the arguments to a function. In Clojure, it indicates an argument whose value will be ignored.[8]

In some languages with pattern matching, such as Prolog, Standard ML, Scala, OCaml, Haskell, Erlang, and the Wolfram Language, the pattern _ matches any value, but does not perform binding.

HTML has a presentational element <u> that was originally used to underline text; this usage was deprecated in HTML4 in favor of the CSS style {text-decoration: underline}.[9] In HTML5, the tag reappeared but its meaning was changed significantly: it now "represents a span of inline text which should be rendered in a way that indicates that it has a non-textual annotation".[9] This facility is intended for example to provide a red wavy line underline to flag spelling errors at input time but which are not to be embedded in any stored file (unlike an emphasis mark, which would be),

The elements may also exist in other markup languages, such as MediaWiki. The Text Encoding Initiative (TEI) provides an extensive selection of related elements for marking editorial activity (insertion, deletion, correction, addition, etc.).

Unicode has a free-standing underscore _ at U+005F, which is a legacy of the typewriter practice of underlining using backspace and overtype. Modern practice uses the combining diacritic "combining low line" at U+0332 ◌̲ that results in an underline when run together: u̲n̲d̲e̲r̲l̲i̲n̲e̲. Unicode also has the combining macron below, a single letter diacritic.[10]

• single underline: a̲b̲c̲d̲e̲f̲g̲h̲i̲j̲k̲l̲m̲n̲o̲p̲q̲r̲s̲t̲u̲v̲w̲x̲y̲z̲0̲1̲2̲3̲4̲5̲6̲7̲8̲9̲
• double underline: a̲̲b̲̲c̲̲d̲̲e̲̲f̲̲g̲̲h̲̲i̲̲j̲̲k̲̲l̲̲m̲̲n̲̲o̲̲p̲̲q̲̲r̲̲s̲̲t̲̲u̲̲v̲̲w̲̲x̲̲y̲̲z̲̲0̲̲1̲̲2̲̲3̲̲4̲̲5̲̲6̲̲7̲̲8̲̲9̲̲

In plain-text applications, including plain-text e-mails, where emphasis markup is not possible, the desired emphasis is often indicated by surrounding words with underscore characters. For example, "You must use an _emulsion_ paint on the ceiling".

Some applications will automatically add emphasis to text manually bracketed by underscores, either by underlining or by italicizing it (e.g. _string_ may render string or string).

Underline (typically red or wavy or both) is often used by spell checkers (and grammar checkers) to denote misspelled or otherwise incorrect text.