The Greek alphabet currently has twenty-four letters, and this has been standard for millennia now. However, three extra symbols are used for numbers, and other answers mention letters like tsan used only in specific dialects.

How many letters are known to have been used in the "Greek alphabet" over the years? (And, if possible, what are they all?)

(Note: by "Greek alphabet" I mean all the varieties used outside of Italy, up until the Latin alphabet took over. So Latin and Etruscan don't count, and Cyrillic doesn't count, but pretty much any variety of Greek does.)

2 Answers 2


Incomplete answer while I do more research! Here are all the ones I've found:

  • Letters that survived as numbers:
    • Digamma/waw (Ϝϝ) was originally used for /w/, and left traces of itself all over Homer's writing, but the phoneme disappeared from Attic quite early so the letter never caught on there. It remained alive in Aeolian, so we find the letter used in Sappho (at least in the papyri), and also travelled west to become Latin "F" via Etruscan. In mainstream Greek, it survived only as the number 6.
    • Qoppa (Ϙϙ) came from Phoenician qoph (/q/), and was used for /k/ before back vowels, where it was somewhat more retracted. Eventually the Greeks realized that /k/ was just a single phoneme and dropped it, though Corinth kept it as sort of a local symbol (calling their city Ϙόρινθος and branding their horses with a Ϙ, as Aristophanes mentions in the Clouds). In mainstream Greek, it survived only as the number 90.
    • Sampi (Ͳͳ) was used only in Ionian, possibly for /ʃ/. It's unclear if this letter came from Phoenician or was borrowed from Carian, and might be the same as san (below). It survived better as the number 900 than as any sort of letter.
    • Pamphylian also had a sibilant letter that looked vaguely like Ͳ or Ψ, distinct from psi. It's not clear how it got there, but sampi might have been more widespread than the inscriptions imply, and might have been the first letter in Sappho's name.
  • Extra sibilants:
    • San (Ϻϻ) was one result of the Greeks confusing and rearranging all the Phoenician sibilants. Abecedaria included both sigma and san for a long while, but each individual dialect generally chose one or the other for /s/ and left the other unused. In the end, Etruscan found a use for both of them, but none of the Greek dialects did, and san was unceremoniously disposed of.
    • Sho (Ϸϸ) was never used in Greece itself: it was invented for the Bactrian language (written in the Greek alphabet). It was almost certainly used for the /ʃ/ sound in "ship".
    • Tsan (Ͷͷ) appears in exactly one inscription in Arcadia, but it's used several times in that inscription, always where Attic has /t/ and Cypriot has /s/. So it was probably a sibilant or affricate.
  • Miscellaneous:
    • Pamphylian digamma/vav (Ͷͷ) looks identical to tsan and was unified with it for Unicode purposes, even though it was a completely separate letter. According to Brixhe, it was a re-borrowing of Semitic waw to represent allophony: Ϝ was used for /w/ [v], while Ͷ was used for /w/ [w]. Confusingly, Ϝ is transliterated w, while Ͷ is transliterated v, for historical reasons.
    • Heta (Ͱͱ) was a Heracleian modification of eta used specifically for the consonant sound, distinct from the vowel. Sometimes it looked like a box with a line through the middle, other times like half an H; the latter was eventually adapted into the rough-breathing mark.
    • Ei (not in Unicode, but sort of like ᔑ) was used in Corinth, in places where Attic would use beta—since the Corinthians repurposed beta for something totally unrelated. Attic ε η ει correspond to Corinthian β β ε.
    • Zigzag (not in Unicode, but sort of like ⥌) was used in both Greek and non-Greek languages as a high front vowel distinct from iota. Or it might have been an early chi. There's not really enough data to tell, and most people just treat it as a glyph variant of iota.

Honorable mention:

  • The letter "yot" (ϳ) gets Unicode codepoint U+03F3 despite never actually being used by the actual Greeks. In all the fonts that I've checked, it looks exactly like a Latin j. Why this exists, I cannot say.
  • Just as a side note, the "confusion of the sibilants" theory which you allude to under San is IMO wrong, though it often gets put forward as fact in discussions of the transmission of the alphabet.
    – TKR
    Commented Apr 13, 2019 at 18:42
  • @TKR Oh? I admit I've only seen it put forward as fact, I didn't realize it was contentious.
    – Draconis
    Commented Apr 13, 2019 at 18:43
  • 1
    It's often taken for granted, largely I think because Jeffery favored it in LSAG, but I think the evidential basis for it is very slight.
    – TKR
    Commented Apr 13, 2019 at 19:29
  • 1
    @TKR Added a new question, since I'm curious now; I'd be quite interested to hear any other explanation, since "the Greeks mixed up all the letters and their sound values and their order" does seem a bit strange to assume without solid evidence.
    – Draconis
    Commented Apr 14, 2019 at 0:18
  • 1
    The principle is separation of scripts: Because it is weird to have a single Latin letter in a string of Greek letters, it was given the status of a Greek letter as well. The principle has also led to the cyrillic letters Ԛԛ and Ԝԝ for Kurdish with cyrillic script. Commented May 14, 2019 at 16:21

Here is an addition to the excellent list by Draconis:

• apparently in origin a cursive form of digamma
• the name “stigma” originally applied to a medieval sigma-tau ligature,
  whose shape was confusably similar to the cursive digamma
• used as a symbol with a numeric value of 6
→2185 ↅ  roman numeral six late form

→03C2 ς  greek small letter final sigma

(quotes are from the Unicode standard, and I have nothing to add).

  • Thinking of ligatures, the Greek ligature omikron-upsilon also has some special status and I have seen it to be proposed as an additional symbol for use in engeneering (see Mylonas and Whitney (1993), TUGboat vol 13, p.39, tug.org/TUGboat/Articles/tb13-1/tb34mylonas.pdf ) Commented May 14, 2019 at 16:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.