This is a preliminary report on some of the fifteen papyrus commentaries and annotated texts that will be included in a volume of Commentaria et Lexica Graeca in Papyris Reperta devoted to Euripides (see the accompanying table). All are on papyrus. Twelve are book rolls, two codices, and the nature of one text is debated. Only two are hypomnemata, both very late in date. Seven of thirteen annotated texts are from the high Roman period, five from late Antiquity, and one was copied in the late first century BC. Plays of the « Byzantine triad » are numerous : three texts contain Orestes, two Phoenissae, and one Hecuba. Five of the rest are from canonical plays ; the other two, both Roman in date, contain lost tragedies. In this paper I offer insights arising from examination of seven of the less-studied texts in the group. Although this fresh look at the texts yielded only a little new information, other facts have emerged from the research of others and from the practice of a little so-called « museum archaeology » which are worth recording. Furthermore, the seven texts presented here represent, reasonably well, the state of the rest of the material¹.

In a few texts, new readings or possible new readings emerge, but their significance is generally slight. In MP³ 415.01, for example, which is a crudely written second-century copy of Phoenissae, there are two interlineations above line 57 (κόραс τε δiссάс · τὴν μὲν Ἰсμήνην πατήρ | ὠνόμαсε). The first editor, Michael Haslam, read the note above διссάс as δύ]ο followed by a diagonal stroke. The second, above Ἰсμήνην, he transcribed as θ̣       ρ̣ and suggested θυγατερα had been written, with a diagonal stroke preceding it. I believe a form of this word was intended, but I am not sure so many letters can be crowded into the space available, despite the scribe’s small handwriting. In fact, the last trace visible appears to be suprascript, making θυγατε, i.e. θυγατε(ρα), an option. It would be a normal way to abbreviate the word, and the traces of the last preserved letter conform to the shape of the bottom left of the writer’s epsilons ; the top left portion, which the writer habitually begins a little to the right of the end of the crossbar, has broken off at the edge of the papyrus with the rest of the letter.

MP³ 379.2, a second-century copy of Andromache 346-368, includes an unpublished scrap written in a smaller hand than that of the main text and surrounded by sufficient blank papyrus to show it comes from an intercolumn or margin. It is possible to read ἐ]κτελεῖ δ[, plausibly part of a gloss on 352 πορсύνειν or 365 ἐξετόξευсεν, or possibly part of a metaphrase of the gnomic observation in lines 368-369.

In the case of MP³ 406, an Antinoopolis copy of Medea of the fifth/sixth century, four scraps with remnants of notes turn out not to have been published at all. Only letters, not words, are legible, but once they are available, we may hope that someone may find sense here.

MP³ 402, a P.Oxy. « distributed » papyrus of the fifth century in Williams College Library, illustrates typical technical problems in the Euripides texts I examined. It contains copies of Medea and Orestes written in corrosive brown ink. In the editio princeps, Grenfell and Hunt printed ἄλ]λ̣ο̣ ἡ̣μ̣ι̣χ̣(όριον), a speaker note, beside Or. 1260. Where they saw ink, however – and the published plate confirms this – only a horizontal crevice remains, stained above and below where that ink has eaten through the fibers. It is impossible now even to verify the reading, which in any case inspired little confidence in the editors². Like this papyrus, many of the texts I examined received the best possible consideration from the best papyrologists. As time has passed, however, ink has flecked off or eaten through papyrus, and edges have been lost, so that in some cases the best hope for extracting any more information is intuition and a good attitude.

Another papyrus, MP³ 405, has yielded a few more secrets which although not exegetical at least untangle some of the modern history of the original book. A nicely written copy of Medea from the Roman era, it is divided now among three different British collections. The most substantial part is in Birmingham, among the Rendel Harris papyri now in the University Library. Another large fragment belongs to the Fitzwilliam Museum in Cambridge. These two have been associated and dissociated over the years. The 2001 editors of three additional tiny scraps among the Oxyrhynchus papyri in Oxford identified them as part of the same manuscript as the Harris and Fitzwilliam pieces, apparently on palaeographical grounds. I have found evidence that secures the link and helps to explain how the three pieces reached their present locations³.

The Birmingham component was published first, in 1936, as P. Harris I 38. John E. Powell dated it to the second century. Two years later, in 1938, Denys L. Page published the Cambridge fragment in Classical Quarterly. Evidently Powell knew its existence before it appeared, for in the intervening year he is reported as saying it was part of the same manuscript as P. Harris I 38⁴. Page was evidently unaware of this claim at the time of his own edition of the play the following year, and in any case he assigned his papyrus an earlier date. When the Oxford fragments finally appeared as P.Oxy. LXVII 4550 in 2001, the editors, D. Hughes and A. Nodar, accepted Powell’s assertion and dating. Thus the Cambridge and Birmingham fragments acquired a provenance.

Significantly, the inventory number of the Oxyrhynchus fragments indicates that they were found in the winter of 1903-1904, during Grenfell and Hunt’s third season at Behnesa⁵. The excavators will have sent them to Oxford with the rest of the winter’s finds at the end of that season. In fact, although they were last to be published, they were actually the first to reach England : the Birmingham and the Cambridge portions figure neither among the Oxyrhynchus « distributa » nor did they emerge in the Italian excavations of 1909-1914.

The first notices these two pieces receive are in 1922-1923, so many years after the excavators left Egypt that we are almost forced to conclude that they emerged from the ground separately. Pieces so large – large at least by comparison with the tiny pieces of MP³ 405 that were actually recorded during the 1903-1904 season – would have come to the attention of Grenfell and Hunt if they had emerged while they were in Egypt, for substantial fragments of ancient literature were precisely what they were after⁶. It is theoretically possible that someone associated with the British – or the later Italian excavations in 1909-1914 – found these pieces, held them until 1922-1923, and then offered them for sale : texts did disappear from Egypt Exploration Society digs⁷. But this is not a likely explanation. The market value of papyri was already well known in 1903-1904. The excavation report for that year notes that « the dealers are at length turning their attention to Behnesa, and papyri continue to command extravagant prices in the market »⁸. Since papyri were good for ready cash even then, there was no particular advantage in holding on to them for nineteen or twenty years. All things considered, it seems most likely that the Birmingham and Cambridge texts turned up during the sebakh-mining of the early 1920s, probably not long before they entered the Harris and the Fitzwilliam collections in 1922-1923. But there are many other possibilities, and we shall probably never know the exact circumstances of the discovery of these two fragments. A little information about how they probably arrived in British hands has, however, come to light.

Powell’s preface to The Rendel Harris Papyri discreetly reveals only that the Birmingham portion of MP³ 405 was acquired « privately in Egypt » in 1922-1923. Sixty years later he was more loquacious. To an interviewer for Times Higher Education, he described how in 1932 the Biblical scholar Harris, an old man by that time, related to Powell that « as curator of the Rylands Library in Manchester, he [Harris] (...) made his purchases surreptitiously and then used the hat boxes of his female companions to smuggle the papyri out of the country »⁹. Where Harris met the seller and whether he acted on his own or through an agent is not known. Circumstances suggest the transaction occurred at Oxyrhynchus, where many papyri traded hands that year. We know this from the reports of Petrie, who as Director of the British School of Archaeology in Egypt was conducting an archaeological (not papyrological) excavation of the ancient city that season. He records the vigorous activity of hundreds of sebakh diggers who during the war, with British and Italian excavators gone, started systematically mining the rubbish mounds where Grenfell and Hunt had made their finds. A new railway line built for the purpose facilitated their work. The sebakh-mining turned up papyri, « many » of which were « purchased from the diggers and dealers » that season¹⁰. We do not know whether Harris, in Egypt at the time, was on the site himself, but the circumstances at Oxyrhynchus were clearly ideal that winter for the Birmingham Medea fragments to have changed hands.

Furthermore, the accession record of the Fitzwilliam Museum fragment lists it as a 1922 gift of the British School of Archaeology in Egypt and thus leads right to Petrie himself¹¹. Yet Petrie’s own report of donations made that year by the School contains, interestingly, only summary descriptions of sculptures and inscriptions and no reference to a Greek papyrus. This may be an oversight by the director of a specifically archaeological mission. But since we know – from the inventory number of the P.Oxy. fragments – that the Medea manuscript lay for centuries in a mound outside Oxyrhynchus, it is not likely that Petrie’s team came across it while excavating within the city. I suspect, rather, that sebakh diggers found the fragment and that a dealer brought it within Petrie’s orbit. The British School’s arrangement with the Egyptian government permitted export only of items excavated by his team. He may therefore just have quietly omitted reference to the papyrus in his report.

I suspect, in fact, that Arthur Hunt was an active participant in the disposition of both the Birmingham and the Cambridge fragments. The evidence, all circumstantial, is as follows. First, an established link already existed in 1922 between Hunt and the Fitzwilliam, for all the early Oxyrhynchus texts that Grenfell and Hunt distributed to Cambridge University have Fitzwilliam inventory numbers. Secondly, we know from Petrie that Hunt was at Oxyrhynchus, evidently to buy papyri, in 1922-1923 : « At Oxyrhynkhos many papyri were purchased from the diggers and dealers ; a few were retained for publication by Professor Hunt, the bulk were sent to Washington University, St. Louis, and others to Ann Arbor University, Michigan. »¹² Third, Hunt made a transcription of the Fitzwilliam fragment, which Page consulted for his 1938 edition¹³. Finally, if the Harris and the Fitzwilliam pieces both changed hands at Oxyrhynchus in 1922, which seems likely, then Hunt, whose palaeographical acumen is legendary, may have seen them both, spotted the connection, and commented on it to Harris, who – as we saw – may also have been present¹⁴. This would explain Powell’s confident assertion, even before Page published the Fitzwilliam portion, that the two pieces came from a single manuscript.

Throughout the story, in any event, Hunt is a presence in the background, and I suspect was instrumental in the acquisition not only of the Fitzwilliam fragment but perhaps also, as agent or advisor to Harris, of the Birmingham piece. What dealers separate, however, we can virtually reunite, and some useful insights emerge from the exercise. The Birmingham pieces include the largest and most complete single fragment, which occupies most of a column and its surrounding margins. Because it includes the top and bottom lines of a single column of 34 lines, it demonstrates that the entire play occupied about forty columns, and that the three sets of fragments fall close together at the end of the play.

Two previously unnoticed facts reinforce the – presumably palaeographical – original identification of the Oxyrhynchus and Harris fragments. First, the final lines of two formerly adjacent columns, Medea 1312 (P. Harris) and 1346 (P.Oxy.), confirm the number of lines per column already known from the Harris text. Second, there is a previously unnoticed link between P. Harris fr. 1 and P.Oxy. fr. 1. The former contains the ends of Medea 719-723 and part of the adjacent right intercolumn, which extends upward for the space of several lines. High in this intercolumn are the barely legible remains of the names of Aegeus and Medea, written at the level that lines 748 and 749, spoken by those characters, must have occupied in the next column. Part of lines 748-749 survives, in fact, in a tiny Oxyrhynchus piece, but it comes from the middle of the column.

A technical point of interest in this manuscript, finally, is its lection marks, which are numerous in the lyrical portions of the Birmingham fragment. Several, added by a second hand, resemble a large, distinctly written apostrophe. In Roman texts, the apostrophe typically marks elisions, or separates pairs of vowels or consonants within a word. It does not separate words ; this is the job of the diastole. Nor does its function ordinarily vary within a text. In P. Harris I 38, however, both these conventions are undone, for the use of the apostrophe varies according to nature of the passage. In iambic trimeters, and six times in choral lyric, it marks elision, as usual. It is commonest in the choral song, however, where it marks word separations. Evidently in passages where the meter is complex, the diction elevated, and the scriptio continua more than usually difficult to parse, its role has been adapted and extended, perhaps to make recitation or memorization easier.

I end with two frequently discussed papyri, both very late commentaries with unsolved difficulties. P. Oslo inv. 1662 (MP³ 429, fifth century) consists of nine lines of historical background explaining Euripides’ use of the epithet Παρνάсιοс in Trojan Women 9. Although no lemma survives, the mode of discourse is exegetic, and the learned comment corresponds closely to the scholia. A key question is the original form and the purpose of the text. Because the back is blank, the first editors assumed implausibly it was a remarkably late fragment of a papyrus roll, but this idea is now generally rejected¹⁵.

Marco Stroppa recently pointed out a serious objection to identifying it as a codex, however. The scribe’s writing, which runs across the fibers, also runs parallel to a kollesis¹⁶. In a typical book, whether roll or codex, one would expect script to run either along the fibers and across the kollesis (if the front of the papyrus is used), or across the fibers and the collesis both (if the writing is on the back). Here, it is at right angles to the direction expected. Two explanations come to mind : first, the text was part of a sheet from a squarish codex and it was written upon the wrong way by mistake. If so, it is part of a lefthand page, and its back is blank because the text on the previous page concluded higher up. Second, and more likely, the text was copied onto a loose sheet of papyrus.

Jean-Luc Fournet’s study of the physical characteristics of letters and documents of late Antiquity provides good reasons for preferring this option¹⁷. In the fifth century, as he shows, it became customary for letter-writers to write across the fibers, along the length of a sheet turned 90° from the orientation of a typical letter in earlier centuries. This may be in imitation, he suggests, of the practice of writing documents transversa charta, which was revived in the fifth century and became the rule by the sixth. I expect that P. Oslo was written by someone familiar with this practice, for whom it was natural to orient the papyrus as he would for a letter.

Another feature supporting the idea that the text was intentionally written transversa charta is the relative state of repair of the two sides. The side with writing is largely intact and in good repair ; the back, by contrast, is poorly constructed, with gaping spaces between strips of material¹⁸. Although its condition could be the result of later damage, it is also possible that this side, never intended to receive writing, was left somewhat unfinished in appearance.

As to why a writer would bother in the first place to copy learned material onto a loose piece of papyrus, we can only speculate. Its scholarly quality suggests it was to be put together with similar fragments to create a compiled commentary or, perhaps, scholia. As such, it will have served the same purpose as the thousands of « slips » from which James Murray compiled the Oxford English Dictionary¹⁹. But a simpler and less controversial explanation may be the likelier one, namely, that it is an odd note that a reader or teacher copied out and intended to slip inside his text of Trojan Women, or into a commentary on the play²⁰. As such it will have served the same purpose as the 3x5-inch cards with useful notes and corrections that Herbert Youtie left behind in books in the papyrology collection in Ann Arbor²¹.

The last text is a leaf from a papyrus codex in Würzburg that contains a substantial part of a sixth-century commentary on Phoenissae (MP³ 419). Written in variable, sometimes terrible, handwriting, it is in many parts scarcely legible. The codex is significant not only because it is one of the last examples of a scholarly tradition reaching back many centuries but also because its factual content, like that of the Oslo piece, is unexpectedly high in quality by comparison with most late exegesis. Other noteworthy features are its brevity (the single surviving page covers about 750 lines of the play and averages about one comment per thirty lines), the absence of any discussion of long stretches of text, and the disorder of some lemmata.

The text has occupied a succession of scholars from Wilcken, the first editor, to Maehler to Athanassiou to Stroppa, who has moved enquiry forward considerably²². He observes that such disorder is actually not unique : four of the ten surviving late hypomnemata in codex form show some such non-trivial disarray, and they tend, like the Wurzburg commentary, to be brief and selective. This is a small sample but it is a curious one. The reason for the disorderliness and the brevity of these four late works is unclear, and in truth there may be four or more different explanations. Collectively, however, the texts suggest something not considered before Stroppa’s suggestion : that there may have been readers in late Antiquity (and also earlier ?) for whom a set of succinct but not necessarily orderly comments on particular parts of a text – in lieu of the thorough and orderly marshalling of information typical of the most comprehensive ancient exegesis – was acceptable and useful²³. If we avoid the easy assumption that P. Würzburg 1 and other disordered late commentaries are simply defective hypomnemata and entertain the idea, instead, that they may represent a different kind of reader’s aid, we may improve our understanding of the teaching and transmission of classical scholarship in late Antiquity. But four papyri is a weak foundation for any strong argument on the subject, of course.

This recent work on Euripidean exegesis has enabled me to corral details that went unrecorded at time of publication and to throw a little light on alterations needed in published texts, although many of the indicated changes involve only small alterations in the good work of our predecessors. The net effect of the research done so far, in fact, has been to confirm the impressively high quality of the work of the earliest editors of papyrus texts. It is a humbling experience.

Euripides Papyri Containing Commentary or Annotation