CHAR

What it does

CHAR(number)

CHAR maps a small integer to the single character that the integer designates in some character set, and returns that character as a one-character text value. The whole content of the function is the question "which character set?", and the answer is not a mathematical constant: Microsoft's article states that the character returned comes from the character set used by your computer, and names two — the ANSI character set on Windows, the Macintosh character set on Macintosh.

So the function is best read as a two-step composition, not a one-step map:

1. the argument is truncated to an integer and checked against the admissible range 1–255;
2. that integer is interpreted as an index into a host-dependent single-byte character set, and the resulting character is delivered as Text — which, in the Handbook's value model, is a sequence of UTF-16 code units (see the value universe).

Step 2 is where implementations can legitimately differ, because the two steps only agree trivially on the ASCII subrange 32–126, where every candidate encoding coincides. Above 127 the identity "code number 128 means the character U+0080" and the identity "code number 128 means the Windows-1252 character €" are different functions, and the argument range 1–255 spans both.

Arguments

number — required, the code number. Microsoft's article states it as a number between 1 and 255 specifying which character you want. The argument is numeric, so the ordinary to-number coercion of chapter 02 applies: a logical, or text that reads as a number, converts before the range check.

There is no second argument and no encoding selector. A formula that needs a specific encoding cannot express it through CHAR; that is the reason UNICHAR exists.

Try it

Quirks

Result and edge cases

The return kind is Text, always exactly one character long when the call succeeds.

Two boundaries are specific enough to state:

• Non-integer arguments. The admissible range is documented as a range of numbers, not of integers, so some truncation or rounding rule must exist. Which one Excel applies — truncation toward zero, floor, or round-half-away — is not stated in Microsoft's article and has not been checked here.
• Excel for the web. Microsoft's article carries an explicit platform restriction: on Excel for the web only CHAR(9), CHAR(10), CHAR(13) and CHAR(32) and above are supported. This is a documented per-surface difference, not a quirk of one build, and any implementation claiming to match "Excel" must say which Excel.

Array arguments, empty cells and omitted arguments follow the shared call model; nothing about CHAR overrides it.

Errors

Error inputs propagate rather than convert, per the universal rule in chapter 02.

How to implement

Notes for implementers

The interesting decision is the 128–159 window. In Windows-1252 (the encoding usually meant by "ANSI" on a Western Windows install) that window holds printable characters — the euro sign, curly quotes, the en and em dashes — mapped to Unicode code points scattered far outside 128–159. In a "code unit equals code number" implementation the same window holds the C1 control characters. A program that treats CHAR as chr() over UTF-16 code units and one that treats it as a Windows-1252 decode agree everywhere except here, and disagree visibly on the most commonly typed non-ASCII characters in English text.

A second decision follows from the first: "the ANSI character set" is not one encoding. The system ANSI code page differs by Windows locale (1252, 1251, 1250, 932, …), so a faithful host-dependent implementation is locale-sensitive in a way the function's axis chips do not currently record.

Rust code

1 char_fn "CHAR" (1, 1) [V "number"] |s, a, sv| {
    let n = num(s, a, 0)?;
    out_text(k::char_from_number(n))
}

/// CHAR: the character with code `n` (truncated) of code page 1252, 1 to 255.
pub(crate) fn char_from_number(n: f64) -> Result<ExcelText, WorksheetErrorCode> {
    let n = n.trunc();
    if !(1.0..=255.0).contains(&n) {
        return Err(WorksheetErrorCode::Value);
    }
    let code = n as u16;
    let unit = if (0x80..=0x9F).contains(&code) {
        CP1252_HIGH[usize::from(code - 0x80)]
    } else {
        code
    };
    Ok(ExcelText::from_utf16_code_units(vec![unit]))
}

No implementation has been published with a Handbook test suite for this function.

Version differences

No comparison between specific Excel builds is recorded here yet. This does not establish that all builds behave alike.

Alternatives

Relationships

• CODE is the inverse direction — character to number — and shares the same host-character-set dependency and therefore the same open question.
• UNICHAR and UNICODE are the Unicode-explicit pair. They are the functions to reach for when the code point matters and the host must not: UNICHAR takes a Unicode code point rather than a byte-sized index. CHAR is not deprecated in favour of them; both pairs are live.
• Readers routinely reach for CHAR(10) to embed a line break in a formula-built string, and for CHAR(34) to embed a double quote. Neither use is a special case in the function; they are just the two control characters that formula text cannot easily carry literally.

Named variants

No named variants are recorded yet.