OK, so this question is perhaps somewhat weird, but I have no idea where to start, so here I am. Let me give some introduction.

Me, languages, and Greek

Let's start very far back. As my blog illustrates, I like fiddling with languages. At high school, I had Latin classes, but in my high school there were different curricula, one of which, called "Classical Lyceum", featured Ancient Greek, so I decided to pick my mother's grammar up and start on Ancient Greek. Year 1, textbook Greek translation into Latin. Year 2: Latin author excerpts from my textbook, translation to Greek. Summer of year 2: authors. And that summer, an appendix on dialects popped up, so I thought, "I wanna be able to understand any dialect", and selected texts for that. The appendix had Homeric, Herodotian (if that is the English of Erodoteo), Aeolic, Doric. Since Homeric essentially seemed like a weirder version of Herodotian, I decided to gobble up tons of Herodotus, and then some Homer. Summer ended, and only two passages of the Iliad were translated. March or May of year 3, OK, I've done enough Homer, time for Aeolic.

The time of Sappho

Who wrote in Aeolic? Sappho and Alcaeus, that I know of. Pick something from both? Nah, just choose one and pull all the texts from Greek Wikisource, ready to translate. The choice fell on Sappho, and I went on to discover how my assumption on the easiness of getting the texts was complete and utter bull… err, tauricopria. My blog illustrates that very well. While researching Sappho, I came across a number of sources.


One of those was a document, I mean a Word document, called safopoemas.doc. I have already mentioned this document here. The document is now no longer available. It resurfaced some time ago as a pdf, but my Firefox blocks it as harmful, so I don't even know if it's there.

In any case, the contents of this document were "reminiscent of a Spanish edition of Sappho". I use quotes because I can't tell if this was what was published, or if a lousy OCR had something to do with this. I sure hope at least the horrid corruptions were an OCR's fault, otherwise the team who directed this should never have started. The horrors of this document are beyond description.

  • It's not just the weirdest of typos, like παΤ for παῖ or ϊγω for ἔγω or words that aren't even recognizable, oh no.
  • The line breaking is horribly skewed, a fact I originally attributed to better matching the Spanish verso libre translation, which was bad per se, because if it's libre it should have been fitted to the Greek, but then I saw an example where the line breaking was wrong in the Greek but the Spanish matched the correct line breaking.
  • In some places, the lines are even scrambled!
  • There is even a fragment in two distinct numbers, precisely 73 and 159!

In any case, that horror actually gave some interesting restorations, as that post I linked to before exemplifies.

Striving for completeness on the blog

Fast forward a number of years, fifth year of university, I open my blog, and Sappho starts to go up on it. Fast forward another year, and I am preparing "The rest of Sappho". To be sure I have anything that any of my sources attributes to Sappho, I start readying a huge comparative numbering table between them all.

One of the sources is safopoemas, so I give it a first pass, and locate all but 14 fragments on Lobel-Page. In the meantime, I realize that fragments 40 and 88 are actually collections of P.Oxy. 1231 and P.Oxy. 1787 fragments respectively, each fragment being started by its number and ended by ||, with the line breaks all gone in the Greek. That number "14" accounts for that as well.

So I give those 14 a second pass. Of those that aren't papyrus fragments omitted by Lobel-Page, of which there are about 8, I identify, by locating them in other sources, all but one.

The mystery fragment

Fragment 207 reads:

ώράνα· χελ^όνων όροφή

and is translated as:

. . .la hermosa golondrina, en el tejado. . .

that is:

…the beautiful swallow, in the tiled roof…

I have figured out that ὀροφή means "roof", so it matches "tejado". I would assume it should be ἐν ὀροφῇ, "in the roof", and then we have three short syllables and a non-Aeolic form, so we would need ἐν δ' ὀρόφᾳ if this was Sappho, which would make it Ionic a minore or a maiore in meter. χελ^όνων may be a corruption of χελίδων, "swallow", which would require three syllables to be lost after it to fit the meter, so χελίδων uu– ἐν δ' ὀρόφᾳ. ώράνα doesn't match the translation at all, however. Hermosa from the translation could be integrated in the lacuna getting χελίδων u κάλα ἐν δ' ὀρόφᾳ, and the missing syllable could be γε, to intensify κάλα (maybe). Amending the last Greek word we could reconstruct ὄρραννα χελίδων γε κάλα ἐν δ' ὀρόφᾳ u–x, a ionic a maiore tetrameter. Of course, this all supposes this is Sappho. If this were another author, we could just write… um, ὠράνα with an omega seems to have to be Aeolic, so maybe a gloss? ὠράνα· χελίδὼν ἐν ὀροφῇ?

Can anyone identify this fragment and/or provide a source for this?


In case you're wondering, after over a year and a half of struggling with Sappho, I decided to leave Doric alone and proceed with Chinese and Japanese at the end of August of year 4 :).

safopoemas online?

Tried opening the link with Safari, and:

enter image description here

Aw, butts. But Scribd seems to have it. And here is the edition… with the Greek completely lost! And here it should be again. Bonus: «Safo Poemas BAJO LA DIRECCIÓN DE MANUEL DE EZCURDIA CON LA COLABORACIÓN DE TERESA SILVA TENA Y CARLOS TRILLAS SALAZAR Safo Introducción, traducción directa y notas de Carlos Montemayor Edición completa de los fragmentos sáficos EDITORIAL TRILLAS Las ilustraciones han sido tomadas de: Dictionnaire de la civilisation grecque, Fernaud Hazan editeur, París, 1966.». Oh yep, it's the one. Enjoy the horrors of this :).

  • "Who wrote in Aeolic? Sappho and Alcaeus, that I know of." And Julia Balbilla, though her Aeolic is pastiche. Commented Aug 27, 2018 at 7:56
  • @NickNicholas Who is that :)? PS I would be on that list as well, but only after I started Sappho, I think. I wrote some Sapphic and Alcaic stanzas in Aeolic, plus a few "warning lines" for Sappho's poems (stuff like "There's one or more lines missing here, the last of which may have ended with 'Ilíō'" or "This exists only by my phantasy", but cfr. my blog post "All of Sappho").
    – MickG
    Commented Aug 27, 2018 at 8:09
  • en.wikipedia.org/wiki/Julia_Balbilla Commented Aug 27, 2018 at 8:14
  • You have a lot more to work with with Doric than with Aeolic. But that's an argument for sticking with Aeolic, in fact: more tractable :) . In going with Sappho instead of Alcaeus, you avoided the surprise of θᾶς: hellenisteukontos.opoudjis.net/aeolic-θᾶς-until Commented Aug 27, 2018 at 8:18
  • @NickNicholas Lucky that appendix had Aeolic before Doric then :). Otherwise I'd have Pindar on the blog instead of Sappho, probably :).
    – MickG
    Commented Aug 27, 2018 at 8:42

2 Answers 2


Hesychius ω 302.

ὦ ’ράνα χελίδων· ὀροφή

In Lobel & Page, this is fr. 135

τί με Πανδίονις, Ὤιρανα, χελίδω ...;

—based on a less corrupt transmission of the verse in Hephaestion: τί με Πανδιονὶς ὤραννα χελιδών "Why, O Irene, lovely swallow, Pandion's child, dost thou [weary] me?"

Hesychius somehow thought the verse needed to be glossed as "roof", or put "roof" as part of the verse.

The commentary in the latest edition of Hesychius is:

in ὀροφή nonnulli censent verbum Sapphus latere (ὀροφαία Fick, ὀροφῇ Schm.); ἐν (cf. H) ὀροφῇ coni. M.L. West ("grammaticus ὠράνα ab οὐρανός derivare conatus est"); hanc esse veram derivationem censet Schwyzer IF 14, 1903, 27–8, i.e. ὠράνα: ὠρανός ut στεφάνη: στέφανος etc., et interpr. [ut iam HSt.] "arundinum tectum", neque ad Sapph. refert.

Which I think means:

Several scholars think the word "roof" goes back to Sappho ("of the roof" Fick, "in the roof" Schm.); M.L. West conjectured "in the roof": "the grammarian tried to derive ὠράνα from οὐρανός 'sky'" [instead of "O Irene"]. Schwyzer, Indogermanische Forschungen 14 (1903): 27–8 thought this was the true derivation of the word, i.e. ὠράνα: ὠρανός like στεφάνη: στέφανος, and interpreted it (as Henricus Stephanus had done) as "roof of reeds" [where a swallow would nest?], and not as referring to Sappho. [?]


Story time continues. Hit post button. Fast forward 5ish-10ish minutes, and I think, "Maybe this gloss is in the critical notes of some fragment? Perhaps the Pandionis one [Lobel-Page 135]?". Well, bedtime it was.

Fast forward 20ish minutes, and I'm like "Ya dum-dum, why not just google ὠράνα?". And in the dead of night, around 1am, I pick up my mobile, turn it back on, painstakingly compose that word without a proper polytonic Greek keyboard (or any Greek keyboard at all), google… um, yeah, I don't want Modern Greek "ώρα να", better wrap the query in double quotes… and I land on this, which I initially mistake for Bergk, but it is not. Let's read the critical note to precisely the fragment I had thought of:

Fr. 88. Hephaest. 66: Καὶ ὅλα fisv οὖν ᾄσματα γέγραπται ἰωνικὰ. . . . παρὰ Σαπφοῖ· τί με κτλ. — ὠ ῤ́αννα Is. Vossius, vulgo ὠράνα, E ὠραινα, Turn. in calce ὡραῖα. — χελιδὼν, Gaisford χελιδοῖ, fort. χέλιδὸν. Ceterum conf. Hesych.: Ὠράνα · χελιδόνων ὀροφή, fort. Ὤραννα χέλιδον · ὑπωροφία.

OK, Hesychius gloss then. Let's also pick up Lobel-Page (~2am by then). Suggestion: ὀροφαία to start l. 2. Hmm. This leaves me with questions:

  1. To amend, or not to amend?
  2. Where is the mention of the roof in the first part of the gloss? I mean, ὠράνα has to come from ὤρανος "sky" somehow, right?
  3. What the hell is up with that Spanish translation?

Fast forward to next morning, with an answer to this question coming in. I find the gloss in Campbell too, not in Bergk, and I find out that Hasychius's work is a Lexicon called Συναγωγὴ πασῶν λέξεων κατὰ Στοιχεῖον, "Collection of all words in Alphabetical Order", and the entry is just the quote, so the idea that "hermosa" could come from the neighborhood of the gloss is incorrect.

It also seems that the tradition is not concordant on this gloss, because I have seen like three different forms up till now, plus the link's proposed amendment:

  1. <ὦ 'ράννα χελίδων>· ὀροφή, from Greek Wikisource;
  2. Ὠράνα · χελιδόνων ὀροφή, from the link;
  3. ὦ ’ράνα χελίδων· ὀροφή, very similar to 1, in the answer here;
  4. Ὤραννα χέλιδον · ὑπωροφία, the link's proposed amendment.

I think the most likely form of the safopoemas fragment is number 2, with the translation just being bogus. Or perhaps that form but adopting M.L. West's conjecture mentioned in the other answer, making it ὠράνα· χελίδων ἐν ὀροφῇ, so that "golondrina, en el tejado" is found in the original, though again "hermosa" isn't, unless they went fantastically creative and read ὠράνα as "beautiful".


Quoting @NickNicholas's comments on this answer:

(1) is the previous edition of Hesychius (1961?); (3) the new edition. [(2) is] The edition of Hesychius before that. From the app crit of the current edition: Ὠράνα χελιδόνων ὀροφή is in fact what is in the manuscript. So (2) is what is in the manuscript of Hesychius, (1) is (Schmidt's?) edition, and (3) is Hansen & Cunningham's.

The Spanish edition at hand is based on Reinach's Alcée, Sapho, Texte établi et traduit par Théodore Reinach, dating 1966, and itself dates 1982, so one would assume the 1961 was available to Reinach and that our Spanish edition would have text (1). Maybe they just decided to follow the manuscript, which was (I assume) reported in the magnífico aparato crítico of Reinach?

The weird thing about this Spanish edition is that it makes it look like Reinach did not have the Ostrakon Florentinum at his disposal, which was published in 1937, as the Idyll with Aphrodite fragment (Lobel-Page 2) on that ostrakon is present in the form of two quotations as found in Bergk and Edmonds, one quote being stanza 2, one being stanza 4 (last stanza), respectively from Hermogenes (Bergk 4 Edmonds 4) and from Athenaeus (Berg 6 Edmonds 6), whereas the Ostrakon text is among the no recopilados, that is the fragments not present in Reinach's edition, sitting at number 209. If Reinach didn't have the ostrakon, I would assume he established his text way before the publication, when the ostrakon wasn't well-known yet. I mean, Lobel-Page had the ostrakon, and it's a 1955 edition… So maybe Reinach actually had the manuscript text only, and the Spanish edition stuck to that. Maybe the date 1966 is actually a reprint?

Look at this! Alcé, Sapho, by Reinach and Puech, dated 1938! Seems the 1966 one was a reprint after all! So that 1938 one definitely had the older edition with the manuscript text, and the Spanish edition seems to have followed it. The question obviously is what the hell is going on with the translation which doesn't match that text. Time to date that M.L. West conjecture, I guess.

Not so fast

Oh well, that was an article actually… but this book has a biblio entry with a 1937 Reinach&Puech Alcée, Sapho: would that be yet another article? The biblio doesn't give coordinates within a journal, if "Les Belles Lettres" is a journal, so I would assume this is an edition of Sappho and "Les Belles Lettres" is like "Oxford University Press". Which means that edition definitely didn't have the ostrakon, and very likely only had the manuscript quote of the gloss at hand.

Also, French Wikipedia says:

Alcée, Sapho, texte établi et traduit par Théodore Reinach, avec la collaboration d'Aimé Puech, Les Belles Lettres, Paris, 1937, rééd. 1981. Édition incomplète, de nouveaux textes ayant été retrouvés depuis sa publication.

Which proves my point. And adds a second reedition in 1981, the year before the Spanish one.

  • 1
    (1) is the previous edition of Hesychius (1961?); (3) the new edition. Commented Aug 27, 2018 at 10:42
  • @NickNicholas What about (2)?
    – MickG
    Commented Aug 27, 2018 at 10:44
  • 1
    The edition of Hesychius before that. From the app crit of the current edition: Ὠράνα χελιδόνων ὀροφή is in fact what is in the manuscript. So (2) is what is in the manuscript of Hesychius, (1) is (Schmidt's?) edition, and (3) is Hansen & Cunningham's. Commented Aug 27, 2018 at 10:49
  • 1
    Sorry, I was wrong about the date. 1961 is when the modern edition started, but it didn't get to omega until 2009. So (3) is 2009, (1) is presumably Schmidt in 1868, and (2) is the ms reading; whoever used it was rejecting Schmidt's emendation. Commented Aug 27, 2018 at 13:53
  • 1
    French publishers are notoriously bad at indicating the date of publication. The date you see on the title page is normally that of the last reprint.
    – fdb
    Commented Aug 28, 2018 at 12:21

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.