Degree Symbol
The degree sign, the prime marks that get used in its place, and the precomposed Celsius and Fahrenheit characters, each with its real codepoint and Unicode name. The notes cover what normalisation and a latin-1 channel do to each one.
Tap any tile to copy it. Nothing is sent anywhere — the copy happens in your browser.
One degree sign, and the marks that are not it
° U+00B0 DEGREE SIGN, category So, is the character for degrees of temperature and degrees of arc. ′ U+2032 PRIME and ″ U+2033 DOUBLE PRIME are category Po — punctuation, not symbols — and mark minutes and seconds of arc, or feet and inches. A coordinate written 51° 30′ 26″ uses all three, and none of them is the ASCII apostrophe or the ASCII double quote, which is why a coordinate typed on a keyboard and one pasted from a reference rarely compare equal.
The double prime is not a stable character
NFKC rewrites ″ U+2033 into two copies of ′ U+2032. ‴ U+2034 TRIPLE PRIME becomes three. The reversed forms ‶ and ‷ do the same to reversed primes. Afterwards the text looks identical and the string is longer: one character in, two out. A character limit, a fixed-width column, or an exact-match comparison all notice that; a reader does not. If a value has to match something stored earlier, decide which form is canonical before it is stored, not after.
℃ and ℉ are single characters, until something normalises them
℃ U+2103 DEGREE CELSIUS and ℉ U+2109 DEGREE FAHRENHEIT are one character each, category So. NFKC decomposes ℃ into two characters, U+00B0 followed by U+0043 — the degree sign and an ordinary capital C — and ℉ into U+00B0 and U+0046. So the choice between pasting ℃ and writing °C is only a real choice somewhere that leaves text alone. Where normalisation happens, the two converge, and the single character is the one that changes underneath you. In a field that normalises, the two-character form is the one that will still be there when you read it back.
Only one of these survives a latin-1 channel
U+00B0 sits below U+0100, so latin-1 and cp1252 carry the degree sign as a single byte. ℃, ℉, ′ and ″ are all above that range and have no representation in either encoding. If a value passes through something genuinely latin-1 — an old export, a mail header, a legacy column — the degree sign is the one that arrives, and the primes and the precomposed unit signs are the ones that turn into a substitution character or vanish.
Three characters that look like a degree sign and are not
U+00BA MASCULINE ORDINAL INDICATOR is a letter, category Lo, used in Spanish and Portuguese ordinals; NFKC turns it into a plain lowercase o. U+02DA RING ABOVE is a modifier letter, category Sk, meant to sit over another letter. U+2070 SUPERSCRIPT ZERO is a raised nought. Each is drawn as a small raised circle, none of them is U+00B0, and a unit parser or a search looking for the degree sign matches none of them.
Width
°, ′, ″, ℃ and ℉ have East Asian width Ambiguous; ‴, ‶ and ‷ are Neutral. So in a monospace column rendered with ambiguous characters wide, a double prime and a triple prime are not the same width, and a table of coordinates padded out with spaces stops lining up.
The wing index is the maths and science symbols list. The raised digits that sit beside a degree sign in a formula are on exponent symbols, and the operators are on math symbols.