Θαβωρής, Α.Ι. 1959. Τα εκφραστικά μέσα προσδιορισμού του χρόνου του ημερονυκτίου στην αρχαία μεσαιωνική και νέα ελληνική. (Επιστημονική Επετηρίς Φιλοσοφικής Σχολής, Παράρτημα αρ. 2.) PhD Dissertation, University of Thessalonica. Thessalonica: University of Thessalonica.

Antonis Thavoris was Linguistics Professor at the University of Ioannina. His 1959 PhD thesis is on expressions for the time of day throughout Greek; I’m going to mention bits of it from Modern Greek that strike me as being of interest.

• p. 19. In dialect, if you want to say that someone goes to bed early, you say that he goes to bed at “7, 9, or 10 at midday”—where the hours given are merely conventional.
• p. 27. Similarly, “7, 9, 12 at midnight” for “very early in the morning”
• p. 28. The twelve hours of daytime were set by the Babylonians, and known to Herodotus; the extension to a 24-hour day was done by Hipparchus in the 2nd century BC.
• p. 31. In the Koine, hours were given with ordinals, and started counting from sun-up: “the ninth hour” was 3 pm. (That is the language used in the New Testament, and has been carried over in traditional translations.)
• p. 32. That ordinal naming continued throughout the Middle Ages; “from the first hour” occurs in the Chronicle of Morea (so 14th century.)
• p. 32–33: The West started counting hours not at sun-up, but at midnight. So in the early modern period, Greeks started having the option of telling time Western-style or Eastern-style. Since the Ottomans continued Roman and Byzantine practice, Eastern-style was called by Greeks, in Italian, alla turca, and Western-style alla franca. Thavoris gives an example as late as 22 December 1912, from the Athens newspaper Akropolis, when the Greek king visited Kozani, newly added to the Greek state: “at two in the afternoon [Western time, which is what the Athens press used], the herald shouted from the base of the enormous bell tower of St Nicholas: ‘at 11 o’clock alla turca our King is arriving, five o’clock alla franca.’”
• In the monasteries of Mt Athos, Eastern time is still followed.
• p. 32. The Modern Greek way of telling time does not use an ordinal: it uses a number + i ora “the time”, e.g. efta i ora “seven the time = seven o’clock.” But the older vernacular pattern (which Thavoris does not date, but he does cite from dialect sources) looks like English military time: “ten hours at night” (stsi deka ores tis nixtos), “hours 11 of the day” (is tas oras 11 tis imeras), “four hours of night” (os tis teseres ores tis niktos), “one hour, the seventh hour according to the older generation” (mia ora, i evðomi ora par arxeoteris).

There’s a deeper story there: ordinals were on the way out of the vernacular, which either made them more transparent (efta “seven” > eftatos instead of evðomos “seventh”), or outright dropped them: contemporary ordinals like tesarakostos ðefteros “42nd” are completely revivals in Puristic. As those old ways of telling time show, Greek wasn’t comfortable with saying “seventh hour”, like Koine and Mediaeval Greek did, and it wasn’t comfortable with saying “hour number seven”, with an abstract number, as the contemporary colloquial language might (ora numero efta): it could only use 7 as a cardinal number, counting hours in the plural.

In fact, Triantafyllidis recorded that the early Greek migrants to New York found themselves in the same discomfort with ordinals, and resolved it in the same, now obsolete fashion. The did not refer to 42nd Street as tesarakosti ðeftera oðos, with a Puristic ordinal, nor as oðos saranta ðio “Street 42”, as the contemporary vernacular might have attempted: they used the expression stus saranta ðio ðromus “at the forty-two roads”.

Evangelos Lolos, remember how we were wondering how the old vernacular would have said something like sistima ɣuintouz “Windows System”, with its current French word order (Système Windows)? The old Greek hours don’t give us the answer, but it does show that that French ordering, which you might see in ora efta “hour seven” or ora miðen“hour zero”, is not how the vernacular used to do it.

I’ll add that the current way of adding minutes to the hour, efta ke ikosi “seven plus twenty = twenty past seven”, efta para ikosi “seven minus twenty = twenty to seven”, is not that old—although Thavoris does not dig further into it. When I was researching how to render contemporary notions into Ancient Greek for a picture dictionary I collaborated on (Textkit Greek and Latin Forums), I found an 1832 naval manual trying to revive the old Koine ordinals, by using “15 minutes of the 10th hour, 45 minutes of the 7th hour, 20 minutes of the 9th hour” (πεντεκαίδεκα λεπτὰ τῆς δεκάτης ὥρας, πέντε καὶ τεσσαράκοντα λεπτὰ τῆς ἑβδόμης ὥρας, εἴκοσι λεπτὰ τῆς ἐνάτης ὥρας).

https://books.google.gr/books?id=uARfAAAAcAAJ&pg=RA1-PA54&lpg=RA1-PA54&dq=ογδόη+ώρα+και+λεπτά&source=bl&ots=Owe1FwRp8V&sig=49N5Xc-7JyOEJUCIFoeUFxn9qjc&hl=el&sa=X&ved=2ahUKEwjrhIf-l8PfAhUO1hoKHc3DCIEQ6AEwB3oECAQQAQ#v=onepage&q=ογδόη%20ώρα%20και%20λεπτά&f=false

• p. 39. In one of the more colourful expressions of Modern Greek, sunset is referred to as o iljos vasilevi, “the sun reigns”, presumably because the sun then looks at its most magnificent. But in Oinoe (Ünye) in the Pontus, “the sun reigns” is used in the more predictable meaning of “it is midday, the sun is at the zenith”.
• p. 46–50. The rising and setting of constellations and stars was used to tell time at night: the Pleiades (pulja), Orion (aletropoði “the Ploughshare”), the Morning Star (avɣerinos) and Evening Star (aposperitis), Andromeda (stavros “the Cross”), the tail of Scorpio (galonomja, “the milking sheep shepherd’s stars”).
• p. 52. The Ancients told the time by how long their shadow was—Aristophanes refers to seven-foot shadows and ten-foot shadows; p. 53 shepherds still did so in Crete in 1918: if their shadow was eight feet long, it was time to take the milking sheep in for milking.
• p. 56–58. The Homeric amphilykē “twilight” survives in the Koine lykophōs “twilight”, which was revived in learnèd Modern Greek. Thavoris doesn’t go deep into the etymology, and apparently there is still controversy around it, but it seems to me lyk– is the same word as Latin lux “light”, which also turns up in the epithet of Apollo as lykeios. So amphi-lykē “both-light”, “either side light” could refer to amb-iguous light; the etymology of lyk– forgotten, lyk– was reinterpreted as meaning just “twilight”, and re-elaborated as lyko-phōs “twilight light”. The word is first attested in Aelian, and Aelian already thinks the lyko– refers instead to wolves (lykos), because twilight was when wolves came out; Eustathius of Thessalonica, a millennium later, reports that “vulgar erudition” instead distorts it to ɣlikofos, “sweet light”.
• p. 99. Times of eating have been used consistently to name times of the day; it’s useful to be reminded not just that Modern apoɣe(v)ma “afternoon” is literally “from-meal”, “after-meal”, but that ɣjoma, an old term for “noon”, is just a vernacular form of ɣe(v)ma “meal”.
• p. 103. The ancients routinely referred to bedtime as the “first sleep”, peri prōton hypnon; the expression survived into Mediaeval Greek as proθypnion, and in Modern dialect as proto(i)pnin. Thavoris does not seem to make the explicit connection, but as I’ve seen mentioned once by someone on Quora, people who in olden days got to bed at sundown used to wake up around midnight, and then get back to sleep for a second sleep; apoipni“from-sleep” in Symi is glossed as “second, morning sleep”.
• p. 107. The Roman notion of watches of the night (=guard duties), lasting three hours each, survived into Byzantium; they persisted into the pre-WWII navy, and I’ve reported here that the 2–4 am watch in the contemporary Greek army is called “German” (ɣermaniko), because you have to be a Schwarzenegger to endure it.