Reconstructing a Book Collection Through the Identification of a Copyist: Reused Rolls in Context

1 Introduction

In an interesting 1983 article, W. Clarysse raised the question of whether it is possible to relate the fragments of Greek literature that have come down to us through papyri to the people who actually lived in Egypt and produced documents. He concluded that “quite often […] the survival of individual literary texts among the documents is not a matter of mere chance” but tells us something about the interests of their owners, most of whose books, however, may have been dispersed.^[1] Indeed, contextualising literary texts – that is, setting them in connection with their readers/consumers – is a challenging task not only because they lack any internal indications of their context, but also because the archaeological information on their discovery is often poor.^[2] The matter becomes even more complicated with purchased papyri, for which the circumstances of discovery are largely unknown.

In this paper I will focus on a group of papyri dated to the second half of the second century BC that have been scattered among the Berlin, Paris and Manchester collections as a result of their respective purchases on the antiquities market by O. Rubensohn in 1901 and 1902, by Th. Reinach in the winter of 1901/2, and possibly by A.S. Hunt around the same years.^[3] Three of these papyri, namely BKT V.2 (1907), p. 123–128, BKT II (1905), p. 55 + P.Rein. I (1905) 5, and P.Ryl. I (1911) 18, have already been connected to each other, being attributed to the same scribe. In the introduction to P.Ryl. I 18, Hunt wrote:

Among the papyri published by M. Th. Reinach is a small fragment (No. 5) which contains a few lines relating to κιθάριϲιϲ, and which, judged from the facsimile, is in the same handwriting as 18; […] according to information kindly supplied by Dr. Schubart, P. Reinach 5 is in the same hand as the two pieces published in Berl. Klassikertexte, ii. 3 (= P. Berlin 9869), which evidently deal with a similar topic. That the Berlin and Reinach fragments have a common origin is therefore clear; but it becomes increasingly difficult to establish any connexion between them and 18, and it appears preferable to suppose that the latter, though written by the same scribe, belongs to a different treatise. Additional colour is lent to this theory by a further similarity of the script of P. Berlin 9869 to that of Berl. Klassikertexte, v. ii. 20, which contains part of a poetical florilegium, the resemblance being so marked that Schubart would explain it by referring the two texts to the same writer. Apparently, then, a singular chance has preserved parts of three distinct manuscripts from the pen of one copyist.^[4]

However, since then, these papyri have never been studied as an ensemble. I will therefore demonstrate how a careful material analysis of them can provide the basis for the reconstruction of a book collection, albeit a small one: first, I will show how two other papyri from Rubensohn’s purchase, BKT X (2012) 15 and BGU III (1903) 1011, are to be assigned to the same scribe; secondly, by investigating whether and how these papyri were reused, I will try to connect them with a documentary context; and, finally, I will attempt to draw some provisional conclusions about the nature and purpose of the book collection. A more in-depth study of the texts will provide further evidence on these matters.^[5]

2 One Scribe for Five Manuscripts: the Handwriting

BKT V.2, p. 123–128 (TM 62570), inventoried as P.Berol. 9772, consists of two non-contiguous fragments (a: cm 16.5 × 15; b: cm 51 × 15). The distance separating them cannot be estimated. Both are complete in height and the second also on the right; they yield the remains of an anthology of texts on marriage from comedy and tragedy.^[6] As has been noted since the ed.pr., the papyrus is a palimpsest on the recto:^[7] after a piece had been cut out from a taller and longer roll and the previous text had been very roughly washed out, the new text, of which six columns have survived, was overwritten along the fibres.^[8] At an unclear stage, the same hand added another column on the verso, in correspondence with the second column on the recto of the second fragment, with the top the same way up as on the recto.^[9] Three kolleseis at intervals of approximately 21 cm are visible on the larger fragment. Some damage occurred at the second one, causing the left-hand kollema to tear. This must have happened before the reuse of the roll, because the scriptura superior at that point runs along the fibres of the right-hand kollema (Fig. 1).

Figure 1

On the verso some strips of papyrus that were used to repair damage are present, but it is not possible to establish whether they were attached before or after the new text was written.^[10] In either case, the scribe not only wrote on reused material but was even untroubled by the presence of damage on it.

On the recto, the handwriting, which is readable albeit rapid (most letters are written with the fewest constituent strokes), appears to be strongly influenced by the contemporary “scritture di rispetto”: the highlighting of the upper line is achieved by ligatures (especially with the right-hand stroke of alpha, the horizontal stroke of gamma, the middle bar of epsilon, the upper stroke of sigma and the cross-bar of tau) and linking strokes between letters (after kappa, eta, nu, pi, ypsilon and chi), while bilinearism is systematically broken at the top and bottom (epsilon, kappa, rho, tau, phi and psi).^[11] As far as the structure of the letters is concerned, the following should be noted: alpha is almost always drawn in one movement and sometimes in two movements with the first and the second strokes linked by a loop; beta in two movements has a very small upper loop; delta, with the first and the second strokes drawn in one movement, is mostly symmetrical and raised above the baseline; epsilon is drawn in three movements with its middle bar always protruding beyond the semicircle; kappa, with a long vertical sometimes hooked at the ends and with arms which are mostly detached from it, is drawn in one movement; mu, in two or, more rarely, three movements, has middle strokes forming a shallow saddle that touches the baseline; nu, often in two movements, has an almost horizontal second stroke, which merges into the third with a loop; tau has the split cross-bar; the arms of ypsilon meet in a shallow cup, hooked at the left end, and a stem descending to the right, hooked at the lower end as well; phi has a very long vertical and a very small loop; omega’s first loop is reduced to a concave curve to the right and the second, mostly in two movements, is detached from the first and starts on top of it.

All these characteristics are also found on the verso. Here, the use of a more rigid and vertical script, which led R.M. Piccione to think that a second hand was at work, could be due to the greater difficulty of writing on this side of the papyrus, where the direction of the fibres impedes horizontality, or more simply because less attention was paid to the graphic rendering.^[12] Indeed, the same rigidity and verticality can be observed in the last column of the recto, where, in spite of its general appearance, which looks more careless, the letters exhibit the same design as in the other columns: see, in particular, alpha and mu in the cursive forms, tau with the split cross-bar and the ligatures ει and εϲ.

BKT II, p. 55 and P.Rein. I 5 (TM 65634), already recognised as belonging to the same roll by W. Schubart, were republished jointly by F. Della Corte, who identified them as parts of a treatise on music.^[13] BKT II, p. 55 consists of two pieces, inventoried as P.Berol. 9869, a (5.9 × 5.3 cm) and b (8.6 × 11 cm), written on the recto along the fibres with a blank verso. The former is from the upper part of a column and preserves an upper margin of 3.3 cm, while the latter joins directly with P.Rein. I 5 (8.3 × 6.7 cm) to yield the remains of the lower part of two columns and a lower margin of 3.3 cm. A kollesis is visible at about 2 cm from the left edge of BKT II, p. 55, fr. b. There are neither material nor textual grounds for establishing the relative position of BKT II, p. 55, fr. a and BKT II, p. 55, fr. b + P.Rein. I 5. But if they really belong to the same roll, they must come from different sections of it, unless one assumes that the scribe’s control of the writing decreased as he proceeded towards the end of the column:^[14] for, while BKT II, p. 55, fr. a exhibits a more calligraphic script, with the letters drawn with the greatest number of movements, BKT II, p. 55, fr. b + P.Rein. I 5 were written more rapidly, but still more slowly than BKT V.2, p. 123–128. For example, alpha is mostly in two movements, mu always in three. Akin to the habit of contemporary bookhands is kappa with the vertical stroke contained within the bilinear space. Moreover, particularly in BKT II, p. 55, fr. a, the use of ligatures and linking strokes is more moderate, while serifs are added at the ends of vertical (gamma, nu, pi), horizontal (tau) and diagonal strokes (lambda, mu).^[15]

At about the same level of formality as BKT II, p. 55, fr. b + P.Rein. I 55 is P.Ryl. I 18 (TM 65622), which preserves on the recto the remains of the upper part of two columns (with a 1 cm upper margin) of a prose text that has been identified as a historical work or a work on the Seven Sages; the verso is blank.^[16] Striking here is the polymorphism of the letters, which vary from a more calligraphic to a more cursive form. For instance, alpha alternates between the two- and three-movement form, while kappa, mainly bilinear, also occurs in the form in which the vertical stroke breaks bilinearism.

To the same scribe who made these copies I have been able to attribute two other Berlin papyri, BKT X 15 (TM 154388) and BGU III 1011 (TM 44037), hitherto considered unrelated to each other and to the rest of the group. Both were generically dated to the second century BC, but they can now be more accurately assigned to the second half of the century.

The first one, inventoried as P.Berol. 9872, consists of two fragments, a (11.5 × 12.7 cm) and b (1.6 × 2.3 cm), which have been published together as belonging to the same roll.^[17] The larger fragment bears, on the recto, the remains of a column of a text that the editor, W. Luppe, identified as a work of oratory on Odysseus. However, his interpretation is severely compromised by his failure to recognise that the column is complete not only at the bottom and on the left, but also, at ll. 8–13, on the right. Unlike fr. b, which yields only a few letters on the recto and is blank on the verso, fr. a is also written on the verso, upside down in relation to the recto text, but the writing is so faded that it is barely legible.^[18]

The overall appearance of P.Berol. 9872 is that of a skilled bookhand. It is mainly bilinear, except for rho, tau and phi; some cursive elements persist, such as ligatures (with the middle bar of epsilon and the cross-bar of tau) and linking strokes (after kappa, eta, nu and pi), but letters are mostly distinguished from each other. Moreover, the ornamentation, achieved through curvatures (rho, tau, phi), hooks or blobs (iota, kappa and ypsilon) and serifs (gamma, iota, kappa, pi and tau) at the end of letter strokes, is much more pronounced than in BKT V.2, p. 123–128, BKT II, p. 55 + P.Rein. I 5 and P.Ryl. I 18. Although the structure of the letters is the same as in all these papyri, as can be observed in the table below (Tab. 1), the scribe in this case prefers a slower and more formal execution: one may notice, for instance, that alpha is never written in one movement, while nu almost always occurs in its three-movement form.

BGU III 1011 (P.Berol. 9796) is a fragment of a roll that, to judge from the regularity of the edges, may have survived in its original height (32.7 × 24.3 cm): both the upper and lower margins are extant, both of 2.5 cm.^[19] Of the four columns the papyrus preserves on the recto, only the second is complete; the others are badly mutilated. A kollesis is visible 17.5 cm from the left edge. The papyrus is mounted on cardboard, so the verso is assumed to be blank.

Schubart published the fragment as “Amtliche Korrespondenz”, while noting that “scriptura medium fere obtinet locum inter diligentiam libri et manus cursivas, sed angustia columnae, paragraphi haud omissae, emendationes magis sapiunt officinam librarii quam scribae cuiusvis indocti”.^[20] In fact, although ligatures and linking strokes between letters are present, as in BKT X 15 the scribe uses a formal script, rich in ornamental elements (curvatures, hooks and serifs), and draws letters with the greatest number of constituent strokes: for instance, alpha and nu are always in two and three movements respectively; noteworthy is mu, always in three movements, with the addition of a serif on the right-hand diagonal, which makes it identical to the calligraphic mu of BKT II, p. 55, fr. b + P.Rein. I 5.

Although cases of documents written in careful literary hands are of course known,^[21] the attribution of BGU III 1011 to a scribe who both copies books and is able to use a ‘documentary’ script (deliberately not used here) makes it possible that this papyrus is also a literary work, such as an epistle (within a historical work?) in which an officer gives instructions to a colleague.^[22]

As a summary, I show in the following table the similarities between the most significant letters as they appear in each papyrus.

3 One scribe for five manuscripts: layout, errors and linguistic peculiarities

Although I attribute all of the examined fragments to the same scribe, the lack of continuity in content as well as differences in layout mean that they are unlikely to derive from the same roll. In the following table (Tab. 2), I outline the main characteristics of each papyrus’ layout, i. e. upper and lower margins, column height and width, intercolumn and line spacing. An asterisk indicates where the preserved margin, due to the continuity of the edge, may match the original margin size.

In BKT V.2, 123–128, the layout is quite irregular. The columns are not perfectly aligned at the top, while their width obviously depends on the verse length. In the prose texts of the other papyri, however, one can observe the scribe’s habit of making the letters smaller and more compact at the end of the line, in order to avoid encroaching upon the intercolumn. As for BKT X 15, fr. b, it should be noted that the line spacing (cm 0.3) is narrower than in fr. a, which could also mean that it comes from a different roll than the larger fragment.

Aids to the reader such as paragraphoi and vacua are systematically used by the scribe. In BKT V.2, p. 123–128, paragraphoi mark the end of each passage in the anthology,^[23] while in the other papyri they seem to indicate the transition from one sentence to another. This is clear in P.Ryl. I 18 and BGU III 1011, where paragraphoi are combined with vacua.^[24] In BKT II, p. 55, fr. b + P.Rein. I 5, by contrast, there is only one paragraphos between ll. 5 and 6 of col. ii, whose function, judging from the presence of the particle δή at l. 6, is likely to have been to mark this transition as well.^[25] In BKT X 15.11 a vacat without accompanying paragraphos is presumably intended to signify a shorter pause.

The scribe for the most part writes correctly. Minor mechanical errors are found both in BKT V.2, p. 123–128 (particularly in the last column, where the handwriting becomes faster and more careless) and in P.Ryl. I 18. These are corrected in scribendo by writing the right letter in place of the wrong one; in P.Ryl. I 18 and in BGU III 1011, the addition of omitted letters in the interlinear space could indicate that the text was also re-read and revised by the same scribe. Similar revision may not have occurred for BKT V.2, p. 123–128, where some errors remain.^[26] There is systematic assimilation of the preposition ἐκ to ἐγ before voiced stops, a very common phenomenon in documentary papyri during the Ptolemaic period; in BKT V.2, p. 123–128, by contrast, there is no assimilation of the nasal before velar and labial stops in composition.^[27]

4 Reusing Discarded Documents. Towards the Identification of a “Schreibort”

Since, as was shown above, it is impossible to place more than one of the main fragments in the same roll, we must conclude that parts of at least five manuscripts written by the same copyist have come down to us. I have already pointed out a first peculiarity of this scribe, namely that the formality of his handwriting varies from roll to roll. However, a second aspect should also be noted.

As was mentioned, BKT V.2, p. 123–128 has been recognised as a palimpsest on the recto since its first edition. The scriptura inferior, a document written in a fast and minute cursive, was washed out so roughly that it is clearly visible across the entire extent of the two pieces. However, closer analysis of the other papyri has revealed the presence of ink traces under the main texts there, too.

In BKT X 15, illegible traces are visible here and there, while between ll. 8 and 9 I discern the symbol for talents followed by a number, probably a stigma, and possibly other numbers (Fig. 2). Similarly, between ll. 21 and 22 of BGU III 1011, col. ii, one can easily read (γίνονται) (τάλαντα) with a number, perhaps a xi (“total 60 talents”) (Fig. 3), from a previous text that was otherwise washed out with more care than the one in BKT V.2, p. 123–128. Therefore, both BKT X 15 and BGU III 1011 were copied on reused rolls that originally contained documents.^[28]

Figure 2

Figure 3

Regarding BGU III 1011, it is also worth noting that the scriptura superior was here and there written on the vertical fibres, which means that the papyrus, just like BKT V.2, p. 123–128, had already been damaged before it was re-written; it is possible that such damage was caused by the washing out of the scriptura inferior (Figg. 4a–4b).

Figure 4a

Figure 4b

Sparse and very faded ink traces under the main text are also visible in BKT II, p. 55, fr. a and fr. b + P.Rein. I 5 (Figg. 5a–5b) as well as in P.Ryl. I 18 (Fig. 6),^[29] but I have not been able to identify any letter. It is therefore difficult to establish whether we are here dealing with offset traces or the remains of a scriptura inferior, but the latter hypothesis is consistent with my assignment of these fragments to the output of a single scribe.

Figure 5a

Figure 5b

Figure 6

It is remarkable that the scribe chose to copy the books on the recto, washing out the pre-existing writing, even though the verso of at least four of the five rolls was blank.^[30]

Of course, since these papyri have reached us via the antiquities market, we have no reliable data on their provenance. Nevertheless, once it has been established that the rolls reused by our scribe originally contained documents, one may wonder whether these documents can provide any clues to the context in which the copying of the literary texts took place.

No attempt has hitherto been made to decipher the scriptura inferior of BKT V.2, p. 123–128, but a closer inspection, facilitated by a digital microscope, has allowed me to read something. Just to the right of the last column of the scriptura superior, one can make out the beginning of 16 or 17 lines of the underlying document.^[31] Some of these (ll. 2, 4, 6, 11, 13) are indented, which may indicate the individual paragraphs of a list or register, while five of the others certainly begin with toponyms. Two of these, Ψυχιϲ and Τοου, appear twice (ll. 7 and 9, 14 and 16 respectively) and can be read with certainty (Figg. 7a–7b; 8a–8b), while the deciphering of the third, Ψωβ̣θ̣..[ (l. 1), is more uncertain (Fig. 9). Although these are common toponyms, the only districts to which all three can currently be traced are the Heracleopolite and Hermopolite nomes.^[32]

Figure 7a

Figure 7b

Figure 8a

Figure 8b

Figure 9

The expert hand that wrote the text of the scriptura inferior, the use of a medium-thick pen, the width of the columns and line spacing, as well as the toponyms at the beginning of the line which presumably introduce individual entries, lead me to believe that this is an official, administrative document.^[33] The large amounts of money read in the underlying texts of BKT X 15 and BGU III 1011 point in the same direction.

That said, can we find anything that would favour one of the two districts as the place of origin of the papyri? It is true, on the one hand, that only in the Heracleopolite nome are all three villages known since Ptolemaic times, but this may be a function of the extant documentation. On the other hand, it seems noteworthy to me that the papyri purchased by Rubensohn in 1902 include a substantial group of documents dated to the second or mid-second century BC, mostly administrative documents relating to state auctions of houses and land, which certainly came from the Hermopolite nome.^[34] This observation is not decisive in identifying where the literary papyri were written, because we do not know whether Rubensohn acquired them together with the documents. However, it should be added that in his Tagebuch 1901/1902, on 10 December 1901, Rubensohn wrote:

Von Mohammed Adim in Ashmunen erstehe ich für 5 £ einen Kasten Papyri, in dem mehrere litterarische Fragmente. Dann Besichtigung des ausgedehnten Koms. Im N. desselben werden nach Mansurs + seiner Freunde Aussage noch griechische Papyri gefunden. Der ganze Kom wimmelt von Sebbach-Gräbern. Einem derselben kaufe ich ein paar Fetzen Pap. ab, die er gerade gefunden hat (1 P.E.). Dann zu dem im Lauf dieses Jahres ausgegrabenen Tempel, der am östlichen Teil des Koms liegt. […]

Dann in das Haus von Abd el Al, die Ezbet el Adara ist ein großes Dorf, am Nord-Ostende des Koms. […] er hat ein kleines Fragment, wie es scheint, litterar Papyrus, da es vielleicht zu den von Borchardt erstandenen gehört, kaufe ich es für 25 P.E.

Dann mit ihm zum Fundort der Papyri. Er zeigt nicht ein einzelnes Zimmer sondern einen ganzen Kom von allerdings nicht sehr großer Ausdehnung.^[35]

Rubensohn’s reference to the papyri purchased by Borchardt may be significant, since a few days earlier he had acquired a lot of literary papyri in Cairo with an alleged Hermopolite provenance.^[36] On the other hand, it is not possible to identify either the several literary papyri Rubensohn bought from Mohammed Adim or the small one he bought from Abd el-Al. Interestingly, however, BKT V.2, p. 79–84, whose inventory number (P.Berol. 9771) immediately precedes that of BKT V.2, p. 123–128, is the only literary papyrus among those purchased by Rubensohn in 1901 for which El-Ashmunein is recorded in the Berliner Papyrusdatenbank as “Fundort”; there must be others whose whereabouts were not recorded.

5 Conclusion

Let us now try to draw some more general conclusions on the basis of the analysis so far.

In the second half of the second century BC, perhaps in the Hermopolite nome,^[37] a single scribe, professionally trained to write in a bookhand of differing levels of formality, copied at least five books of various contents by reusing documentary rolls of roughly contemporary date, regardless of their material appearance. Except in BKT II, p. 55, fr. a and fr. b + P.Rein. I 5 and in P.Ryl. I 18, which are difficult to assess due to the small size of the surviving fragments, he gave little attention to washing out the previous writing and, in at least two cases (BKT V.2, p. 123–128 and BGU III 1011), he reused papyri that had already suffered some material damage.

Comparison with other Ptolemaic palimpsest papyri shows that, some examples of careful work notwithstanding, reused material of this type was mainly intended for school exercises or for copying texts for personal use.^[38] Regarding BKT V.2, p. 123–128, an interesting parallel is PSI II 120, which has recently been identified by F. Maltomini as a palimpsest. This papyrus holds a collection of moral maxims, dated between the second and first century BC, and written by two different cursive hands, the first of which is quite rapid. The roll is only 16.6 cm tall and was cut from a taller one, and although the recto document was washed out, the verso text, a register of land inspection from Oxyrhynchus (P.Pintaudi 20), was left in place. As was noted by G. Messeri, the literary text was copied, if not in the same room, then at least by the same people who drafted the register, and was perhaps a model prepared by an office manager to train his grammateis in reading and writing.^[39]

As far as our papyri are concerned, we do not know whether they were found in the same place or in different locations in the same city, and illuminating their background and purpose more precisely will only be possible with a new study of the texts they preserve. At any rate, I believe that the link with the administrative context should be stressed, not only on account of the underlying documents that have now been recognised, but also due to the strong influence of the “scritture di rispetto” which is clearest in the handwriting of BKT V.2, p. 123–128. Indeed, it cannot be excluded that we are dealing with material intended for the training of professional scribes. It is true that such training is supposed to have been based mainly on practice in the language of Greek bureaucracy, but we still know too little about the educational programme followed by these individuals. Moreover, BGU III 1011, as we read it now, could have been a valuable tool to learn the unusual, abstract and colourful vocabulary that bureaucratic and legal writing required.^[40] But another scenario is also possible. On the assumption that all the papyri in question share a common origin, they could belong to the private library of a learned reader, perhaps a member of the administration who would have had easy access to discarded material: he himself might have copied some books of interest to him or might have commissioned a professional scribe in the service of the same administration to do so. In this scenario, the level of formality in the writing, layout and even language may have been specified by the reader/consumer, based on the function of the individual manuscript – i. e. whether it was a copy (certainly not a luxury one, but still of a good standard) intended to be kept in his personal library, or a text to be used on different private occasions.^[41]