Sappho · two papyri, centuries apart · the guesses in between

The Other Half Turned Up

For eighty-two years Sappho's poem on old age survived only as the right-hand ends of its lines, on a papyrus from Oxyrhynchus, and scholars wrote the left halves themselves. In 2004 a papyrus four or five centuries older supplied them. Below, set the boldest of the old guesses against the letters that turned up, one line at a time, and then try a line yourself.

What the rubbish heap kept

In 1922 A. S. Hunt published a scrap of a papyrus book of Sappho found at Oxyrhynchus in Egypt, now P.Oxy. 1787. Its first fragment keeps only the right-hand ends of its lines. Hunt read the poem as one long piece, “in which Sappho dwells on the advance of age and the inevitable approach of death”, and of the fragments in general he was plain: “restoration cannot be attempted with any real chance of success.” He supplied only a few letters here and there.

Six years later J. M. Edmonds, in the revised Loeb Lyra Graeca, supplied all of them. He printed the result marked e.g., for the sake of example, with a warning of his own: “the restoration is the merest tour-de-force”. What he could claim for it was a fit: “supplements all fit gaps”. His Sappho is old and cannot help it, and where the missing halves run out she has turned to a myth: Death, who would not give Orpheus his wife.

“[and even as he would not give his] beloved wife [to Orpheus, so he ever] thinks to [keep prisoner every] woman that dies”

J. M. Edmonds, Lyra Graeca I (1928), p. 437; his brackets mark the words he supplied

In the same note he recorded another reader's hunch, that the reference in these lines “may well be, as L suggests, to Tithonus”. L is presumably Edgar Lobel, whose edition of Sappho had appeared in 1925: a reading Edmonds's notes give to L, ἔσχατα in line 10, is the one Hammerstaedt credits to Lobel.

What the mummy case kept

In 2004 M. Gronewald and R. W. Daniel published pieces of a papyrus roll acquired by the Cologne papyrus collection, P.Köln inv. 21351 + 21376. It had been recycled into painted mummy cartonnage, and its first hand is dated to the early third century BC: “by far the earliest manuscript of Sappho which we now possess”, in the words of Jürgen Hammerstaedt, who checked its disputed readings on the original. The Oxford copy is dated to the second or third century AD, so the two are four or five centuries apart (their scholars disagree on the exact dates, and the sources below give each). It came onto the market in 2002 among a private collector's papyri, and where it had been before that is not on record: “We do not know how and when the papyri became his property.”

The Cologne pieces carry twelve lines that the Oxford fragment also has, but the other end of them. In Dirk Obbink's words, “the two manuscripts overlap for a thin strip of several centimeters in the middle.” Together they give most of a twelve-line poem, and its myth is not Orpheus. It is Tithonus, the mortal whom Dawn loved. In the standard version of the myth, as Richard Janko puts it, “the goddess omitted to ask for her lover to have agelessness as well as immortality”, and he aged for ever.

Letters are compared stripped of accents and breathings, and each guess is lined up with the real half by the alignment that needs the fewest changed, added or dropped letters, anchored at the join. Only letters the papyri actually preserve are scored, spelled as the papyri spell them. Letters an editor restored in a hole (in square brackets) and the five letters of line 10 whose reading is disputed count for nothing either way.

By that count Edmonds's halves share 62 of the 176 letters the papyri now preserve left of the join, which sounds like more than it is. The same twelve halves placed against the wrong lines, all 132 pairings, share 27.1 percent of their letters (525 of 1,936). Greek in one dialect and one metre shares a lot of letters with itself: vowels, endings, little words. Most of his 35.2 percent is that. Counting whole words, his halves contain two of the words that turned up: an elided μ’ (“me”) in line 5, and εἰς (“to”) in line 10, which the ending of that line, εἰς ἔσχατα γᾶς, “to the ends of the earth”, all but asks for.

The diagonal is not uniformly the best of its column. His worst line is the one that matters most. Line 9 is where Sappho names Tithonus; Edmonds had starry Night following rose-armed Dawn there, and ten of his other eleven halves match that line's letters better than his own does. Where he scored best, the half word at the join had done most of the work. Line 4 is his only perfect line, three letters of three, and the letters that earn it belong to ἐγένοντο, “became”, from Hunt's supplement of 1922, λεῦκαί τ’ ἐγένο]ντο, which Edmonds kept. Line 1 scores two of three because the -πων that Oxford kept invites a compound in -κόλπων, “-bosomed”; he chose βαθυκόλπων, “deep-bosomed”, and the papyri have ]οκ[.]λπων, confirming Stiebitz's ἰοκόλπων, “violet-bosomed”. Line 10's εἰς follows from ἔσχατα.

Try one yourself

Edmonds had the right-hand halves, the metre, and his whole knowledge of Greek poetry. You have the right-hand half and a counter. Pick a line and write what you think stood before the join. Only preserved letters score, so a guess of the right length helps, and the count of letters in the real half is given.

The first four lines are still guesses

The Cologne pieces are broken too. In lines 1 to 4 they add only a few letters to what Oxford had, so the opening words of the poem are on neither papyrus. Every modern text you read supplies them, and Obbink's apparatus lists four different openings: ὔμμες πεδὰ (West), ὔμμες τάδε (Janko), γεραίρετε (Di Benedetto) and φέρω τάδε (Gronewald and Daniel). West's English version, the one reprinted with that edition, begins “Pursue the violet-laden Muses’ handsome gifts, my children”, and “pursue” translates his own supplements, πεδὰ in line 1 and σπουδάσδετε at the start of line 2. In the join above, lines 1 to 4 show how little the new papyrus added there: eleven letters in all.

Four restorations of the Oxford fragment did turn out right, and Hammerstaedt names them: ἰοκόλπων, “violet-bosomed”, by Stiebitz in line 1; ὄρχησθ’, “to dance”, by Edmonds himself in line 6 (his 1928 text has the compound ἐσυνόρχησθ’); ἔσχατα, “the ends”, by Lobel in line 10; and ἀθανάταν, “deathless”, by Stiebitz in line 12. Each fills a hole of a few letters in a word that was already half there. The last one is the whole difference between the two poems. Edmonds read the same last word, ἄκοιτιν, “wife”, and filled the half word before it as κεδνοτάταν, “most beloved”, the wife of Orpheus. The letters -άταν belong to both. What the papyrus had was ἀθανάταν, the wife who cannot die.

“For people used to think that Dawn with rosy arms
and loving murmurs took Tithonus fine and young
to reach the edges of the earth; yet still grey age
in time did seize him, though his consort cannot die.”

Lines 9 to 12 in M. L. West's restoration and translation, as reprinted by Obbink (2011)

Where the poem ends

So the two copies disagree about what comes after the twelve lines, and scholars disagree about what that means. Janko holds that the poem “is proved to be a complete poem by the fact that the two papyri place different verses before and after it”; Hammerstaedt writes, “I still believe that in fr. 58.23 a new poem begins.” Others have argued that the song circulated in a shorter and a longer version (Janko names that idea in order to reject it). Edmonds, with only the Oxford copy, ran the lines on as one poem, through the two lines Athenaeus quotes and beyond, which is how Hunt had read it too.

The Orpheus in the Cologne roll is a coincidence and nothing more: Edmonds had printed his Orpheus by 1927 (his note cites his own article of that year), and the anonymous poem that names him was published in 2005. Nothing in the method could have led him there. It is only a reminder of what a guess at a lost half is made of: the other poems the guesser knows.

Sources

1. A. S. Hunt, “1787. Sappho, Book IV”, in B. P. Grenfell and A. S. Hunt, The Oxyrhynchus Papyri Part XV (London 1922), pp. 26 to 46; text pp. 28 and 29. Internet Archive.
2. J. M. Edmonds, Lyra Graeca vol. I, revised and augmented edition (Loeb Classical Library, London and New York 1928), Appendix, fr. 118, pp. 434 to 437. Internet Archive.
3. J. Hammerstaedt, “The Cologne Sappho: Its Discovery and Textual Constitution”, Classics@ 4 (2011), Center for Hellenic Studies. Dates the Oxford papyrus to the second or third century AD.
4. D. Obbink, “Sappho Fragments 58–59: Text, Apparatus Criticus, and Translation”, Classics@ 4 (2011), Center for Hellenic Studies. The modern text scored here, the apparatus, and M. L. West's restoration and translation reprinted as an appendix. Dates the Oxford papyrus to the later second century AD and calls it “four centuries later”.
5. D. Yatromanolakis, “P. Colon. inv. 21351+21376 and P. Oxy. 1787 fr. 1: Music, Cultural Politics, and Hellenistic Anthologies”, Hellenika 58.2 (2008), PDF. Dates the Oxford papyrus probably to the third century AD: “Five hundred years or so separate P. Oxy. 1787 from P. Colon. 21351+21376.”
6. R. Janko, “Tithonus, Eos and the cicada in the Homeric Hymn to Aphrodite and Sappho fr. 58” (De Gruyter, 2017), author's PDF.
7. First edition of the Cologne papyrus: M. Gronewald and R. W. Daniel, “Ein neuer Sappho-Papyrus”, ZPE 147 (2004) 1 to 8, and “Nachtrag zum neuen Sappho-Papyrus”, ZPE 149 (2004) 1 to 4 (cited through Hammerstaedt and Obbink; not read here).

Greek set in GFS Didot (SIL Open Font License), served from this page's own folder.