1Thousands of ostraca excavated in the Roman forts of the Eastern Desert attest the importance of written communication for their inhabitants.1 “It seems fairly clear that the distances involved forced a good deal of people who would not normally have needed to write letters, to do so”,2 whatever educational background they may have possessed. Although in general the level of literacy may well have been higher in the Roman army than among the civilian population in Roman Egypt,3 occupants of the desert forts comprised people of various origins, social standings and careers, whose ability to read or write could differ greatly. Even illiteracy was not a homogenous phenomenon: J.‑L. Fournet speaks of “analphabétisme total ou partiel” when exploring strategies by which those without schooling altogether or in a certain language could get by in the world of written communications, for example, by asking others to do the reading, writing, or correcting for them or by copying a muster.4 The ultimate and most efficient means of addressing the illiteracy, however, was through acquisition of literacy. Ostraca found on the Koptos to Myos Hormos road, in particular those in the forts of Maximianon and Krokodilo, give witness to occasional learning taking place there, a product of what Fournet calls “l’enseignement de fortune et de circonstance”.5 In what follows, I will look at evidence for training in literacy from the fort of Didymoi located on the Koptos to Berenike road, where a sizeable number of non-documentary, that is literary and paraliterary, texts have been found by the French archaeological mission to the Eastern Desert during the three campaigns conducted between 1997 and 2000.

2The fort of Didymoi was built in 76/77 as part of a project aimed to equip the route from Koptos to the port of Berenike with a network of fortified stations with wells and cisterns.6 Military garrisons installed there provided security of travel and ensured the passage of official communication, while the local population was involved in supporting the daily needs of the inhabitants. In the middle of the 2nd century, however, the course of life in Didymoi was interrupted: apparently due to the collapse of the well, the fort was temporarily abandoned. Activities there did not resume until after reparations were carried out in 176/177 and continued, accompanied by various further modifications to the buildings, for about another century.7 Because of the interruption of occupation in the middle of the 2nd century, the chronology of Didymoi naturally falls into two periods. The first spans the time from 76/77 to ca 150, the second encompasses the century after 176/177 and until the final abandonment of the fort, which now has been shown to have occurred in ca 270.8 With the help of stratigraphy of the site established by J.‑P. Brun, the editors of the inscribed material found there, mostly documentary ostraca, were able to date them with an unusual degree of precision and to distinguish between ostraca from the two chronological periods. This has allowed them to outline some significant differences.9 Thus, while there is a slight decrease in the overall number of ostraca in the second period (380) as compared to those of the first (552), the number of private letters decreases dramatically in the second period (264 in the first period vs 42 in the second).10 On the other hand, all but one ostracon (O.Did. 27) published under the rubric “Official correspondence” and most of the smaller documents associated with the transmission of such correspondence date to the second period. H. Cuvigny is inclined to believe that the latter peculiarity is due to the chance of preservation because ostraca from other forts of the same network to which Didymoi belonged furnished plentiful evidence for the circulation of official correspondence throughout the desert under Trajan or Hadrian.11 The decrease in private correspondence in the second period, however, cannot be similarly explained and is likely to reflect real changes in this kind of communication.12

3An uneven chronological distribution is also observable in the literary and paraliterary material from Didymoi, but first a few words about the numbers of such texts are in order. All together 43 sherds and two papyrus fragments can be identified as either literary or school texts, and another dozen contain texts of indeterminate nature, possibly literary. When compared with numbers reported for Krokodilo, 8+1? out of 772, and Maximianon, 34+10? out of 1,549, the ratio of such texts in Didymoi to all inventoried ones (970) is about twice higher than in Maximianon and four times that of Krokodilo.13 Since the south-east part of the fort in Didymoi was not excavated, this disparity might be partly due to the chance of preservation, that is, the overall number of ostraca may have been significantly higher had the entire fort been investigated.

4About twenty Didymoi ostraca are likely to be a product of learning, though not necessarily school texts. This distinction is important because learning to read and write could be achieved not only by means of systematic schooling, but also by an ad hoc acquisition of immediately applicable skills. The basic steps of an elementary curriculum in Greco-Roman Egypt are well known from papyrological evidence. Instruction started with the letters of the alphabet, which one practiced writing in conventional order and in various permutations. It proceeded to syllabaries, which were composed in a mechanical way, with a given consonant combined with each of the seven vowels in alphabetic order and forming thereby sets of seven syllables. Along with syllable series, lists of words were introduced, in which multisyllabic words were divided into syllables by means of spaces or dots. The lists, usually arranged alphabetically or thematically, would comprise words, or groups of words, with a given number of syllables. At some point, students were charged with copying sentences from a model provided by the teacher or by dictation. There followed passages for reading, including commentaries or glosses.14 These steps and their progression are attested from the earliest surviving school handbook, P.Guéraud – Jouguet, dated to the 3rd century BC (Arsinoite?),15 through the late antique period, and they are marked by the persistent appearance of the same elements, including words or personal names that are not likely to have been common outside of school texts.

5In addition to those individuals who followed the curriculum methodically, there were always people who aimed to learn quickly whatever it was that they needed to write down on a regular basis, very often one’s name.16 Since these skills were acquired ad hoc, they can be referred to as functional literacy.17 A particular situation could put even further demands on an individual’s learning objectives, such as a need to write or read in another language or alphabet, which one could learn perhaps by copying the necessary bits of information.

6In Didymoi, there is evidence for both systematic and ad hoc learning, but its distribution is chronologically uneven. Most, if not all, examples of the former date to the latest phases of occupation of the fort. There is no evidence in the ostraca of the first period for a school curriculum, with a possible exception of one poorly preserved sherd that appears to have parts of a word list.18 For this reason, let us first consider evidence for ad hoc learning, which, while spanning both chronological periods, is also not homogenous.

7The earliest piece that might be a product of learning is a little sherd inscribed with the name of Pakoibis first in Greek and then in Latin. The shapes of the strokes suggest that both were written by the same person (fig. 1).

Fig. 1 – TM 968593, O.Did. inv. D809, Phase 5, SU 76, 5x6 cm, discarded ca 88‑96.

8The writing is fluent in both lines, but the Greek letters look perhaps a little more confident than their Latin counterparts. The sherd may have served as a kind of a tag written by a person literate in both Greek and Latin to the benefit of a reader comfortable only with one of them. It is also possible that the writer, perhaps the Egyptian named Pakoibis, was learning to write his name in the two languages used by the inhabitants of the fort.

9Another sherd inscribed with a name in two languages is likely a product of a writer familiar with Latin but learning to write his name in Greek (fig. 2).

Fig. 2 – TM 968594, O.Did. inv. 607, Phase 9b, SU 1519, 13x8.5 cm, discarded ca 120‑125.

10Although the difference in fluency between the Greek and the Latin may point to two different hands, the similarity of the first letters in both lines makes it somewhat likelier that they were the work of one writer who was more proficient in Latin. At first glance, it looks like the name was misspelled in Greek as Ἀπολιλάριος, with lambda in place of nu. Comparison with the ductus of other letters, however, suggests that the letter after the first iota is in fact nu followed by an awkward Latin-like alpha made of two strokes. The writer may have had trouble making the Greek alpha distinct from the Latin a.

11Conversely, the writer of O.Did. inv. 45 seems to be more confident in Greek. He first inscribed the first four letters of the Latin alphabet in uneven capitals, then added the name Lucius, perhaps his own, in Greek. The Latin a and c in line 1 look very much like the Greek lambda and sigma of line 2, respectively (fig. 3).

Fig. 3 – TM 968595, O.Did. inv. 45, Phase 9a, SU 141, 12.8x9 cm, discarded ca 115‑120.

12Several ostraca in which the writer appears to be learning to write such necessary bits of information as a personal name or military affiliation are attested from the second period, too, but they display a curious difference. Whereas the earlier ostraca seem to be written by people trying to handle a second language or script, the later ones are rather products of those simply struggling to get control over writing their names. Ostracon O.Did. inv. 570 illustrates the learning process (fig. 4).

Fig. 4 – TM 968596, O.Did. inv. 570, Fort SW, SU 12903, 7.4x8.5 cm, 3rd century.

13Hand 1: “Apoll-, soldier … of the Ituraeans”. Hand 2: “Apoll-, soldier … of the Ituraeans”.

14Two hands can be clearly distinguished. The first, confident and neat, inscribed the name of Apoll-, a soldier in one of the Ituraean cohorts, likely the cohors III Ituraeorum.19 The other, awkward and uncertain, copied it under a horizontal line, which separates the two entries. There is little doubt that the copy was made by a person who was learning to write his name following the model provided by the first writer.

15Repeated efforts of a certain Valerius Domitianus to learn writing his name are attested by two sherds, on one of which, O.Did. inv. 255, he made two attempts (fig. 5).20

Fig. 5 – TM 968597, O.Did. inv. 255, Phase 12, SU 216, 5.2x4.3 cm, discarded ca 230‑270.

16Although altogether the number of ostraca attesting ad hoc acquisition of literacy is not large, it might be more than a coincidence that examples from the earlier period bear signs of more than one language or script and thus involve crossing linguistic boundaries, while later ones show no sign of it. This can hardly be explained simply by the multilingual environment of the fort in the first period, because Latin speakers and writers are well attested later, too, but perhaps rather by educational practices and the background of the writers. In the earlier period, those who write Greek and Latin do not seem to be learning it in any systematic way in the fort. They may have acquired whatever degree of literacy they possess elsewhere, while at the fort they occasionally opted to familiarize themselves with the “other” common script, that is, either Greek or Latin, which for some reason they would have needed to use. The “function” of this literacy can be seen, for example, in the bilingual water receipts from Berenike from approximately the same period, where writers switch, for reasons which are often not clear, from one language to the other.21

17In the second period the situation in respect to the educational environment is different. We see people like Valerius Domitianus, who can hardly form letters, trying to learn to write their names, but there also appears plenty of evidence for systematic school curriculum, from letters of the alphabet to lists of syllables, and from word lists to moralizing sententiae. The following two ostraca illustrate well the conventional character of educational practices in Didymoi in that period. The first, O.Did. inv. 166+310, is a list of words arranged alphabetically and with syllabic divisions indicated, a true educational staple (fig. 6).

Fig. 6 – TM 968598, O.Did. inv. D166+310, Phase 12, SU 212, 9.5x8.5 cm, discarded ca 230‑270.

18Col. 1: “Zenon; Zethos; belt; warrior’s belt; custom”. Col. 2: “Theon? god? …”.

19The letters, especially in the first column, are carefully formed capitals, although their sizes differ, suggesting that we have a pupil’s work. The format of the list finds a close parallel in P.Genova II 53 (unknown provenance, 1st century), which preserves an almost complete set of 24 tetrads of alphabetically arranged disyllabic words.22 P.Bouriant 1 (unknown provenance, 4th century), f^o 1‑5 also lists disyllables, trisyllables, and tetrasyllables in groups of four words for each letter.23

20As to the choice of words, one notes that Ζήνων, a common name in Egypt and elsewhere, appears among several other school texts.24 More curious, however, is the name Ζῆθος: it might be a mythological hero, since Zethos was the son of Zeus by Antiope and the twin brother of Amphion,25 or a personal name. Although the mythological character seems to be minor and the personal name is not common,26 the name must have been a curriculum staple. It occurs in at least four other school word-lists on papyri or ostraca that date from the 3rd century BC through the 5th century.27

21Had the ostracon been found not in controlled excavations, it would be very difficult to date it since the hand lacks peculiar features and both the arrangement and the choice of the words find parallels in papyrological evidence spanning more than half a millennium. Stratigraphy established for the exterior dump in Didymoi, in which the sherd was found, however, indicates that it was discarded in the middle or the third quarter of the 3rd century, that is, during the last phase of occupation of the fort.28

22Another curious piece, O.Did. inv. 549 contains a variation of a well-attested exhortation to a student to work hard, or to pay attention, φιλοπόνει.29 The expression often stands alone, but can also be accompanied by a circumstantial participial construction, e.g. φιλοπόνει γράφων, “pay attention when you write!”, or νέος ὢν φιλοπόνει, “since you are young, work hard!”. On a school tablet now in Berlin, ÄM 13234 (Hermopolis? 3rd century), the exhortation turns to admonition. The student is tasked to copy several times a sentence written by his teacher, φιλοπόνει, ὦ παῖ, μὴ δαρῇς, “work hard, kid, lest you get a beating!”.30 The same sentiment is expressed in a Menandrian sententia, Ὁ μὴ δαρεὶς ἄνθρωπος οὐ παιδεύεται, “without a beating, one learns nothing” (Men. Sent. 573 Jäkel), and it is to these variations that the Didymoi example comes closest (fig. 7).

Fig. 7 – TM 968599, O.Did. inv. 549, Fort NW, SU 12302, 31x22.5 cm, discarded ca 230‑270.

23“Work hard … Work hard, lest you suffer! Work hard, lest you suffer! Work hard…”

24Although the ink is faint, which makes it difficult to assess the quality of the handwriting, it seems that the sentence was first inscribed in the middle of the sherd in even and well-formed letters and then repeated by a student, who tried to imitate the teacher’s handwriting in letters of different sizes.

25All in all, there are over a dozen ostraca from the second period, the content of which can be associated with systematic education. They come from different areas, though almost all were found in stratigraphic units dating to the last stages of occupation of the fort, ca 230‑270. The hands, when they do have any peculiar features, appear to differ, indicating thereby different writers.

26One may wonder whether this evidence for systematic learning finds reflection in private correspondence in the second period. Certainly not in the volume of it, since private correspondence is sparse. Perhaps the Greek of the later letters is overall somewhat more normative than in the earlier period – there are no writers like Philokles31 – but since the volume of evidence is significantly smaller, the argument risks becoming circular. And just as in the earlier period, poor spelling does not keep some writers in the 3rd century from composing letters, as is exemplified by the monomachos Eukylistros of O.Did. 44‑45 or Beryllos of O.Did. 464.

27Unsurprisingly, however, evidence for systematic learning and thus teaching is accompanied by evidence for literary interests and activities in the forts of the Berenike road. Two ostraca found in Xeron have texts that would be of special interest to the inhabitants of the fort or to those travelling through it, which suggests that they may have been composed if not at the fort itself then probably for those living there. One is a poem about the waters of Xeron (O.Xer. inv. 48)32 and the other is an account, in hexameters, of travel from Koptos to Berenike (O.Xer. inv. 995).33 Part of a travel account along the same route may have featured on an ostracon from Didymoi, O.Did. inv. 445, though too little survives to ascertain its content or meter.34 The piece is remarkable in that it combines literary aspirations with an apparent lack of familiarity with normative spelling. The larger part of the text is a monologue addressed to a country dweller and probably meant to be performed by a woman; its rhythmical arrangement and the repetition of the refrain (lines 1‑3 and 9‑11) may indicate that it was meant as a song. The second text, whether it relates to the first or not, seems to narrate a journey from Koptos, possibly mentioning the oasis of Phoinikon and almost certainly the station of Aphrodite of the Desert, the stops immediately before and after Didymoi on the road to Berenike. I reproduce this text, edited elsewhere, with an approximate translation:

	ἀγρότα⸍ ἄπαγέ μαι μετ’ ἐσοῦ ⸍ ἐν ἀγρῷ χα-
	ταμένιν⸍ μή μαι λυπήσῃς⸍ πάντα σοι
	ποήσο⸍ ὥσα δεῖ χορεικῆν⸍ γυναῖκαν⸍
4	αὐτόπλεκτον καλύβην ἐς̣⸍ βρώδ̣ον ποήσο
	στρόσο⸍ δὲ πατικῶς⸍ αγρότ̣α ̣ ι̣κ ̣φ̣ ̣ βοη⸍̣
	ο⸍κ’ωποσ⸍ο μετ’ ἐσοῦ⸍ ἐ̣μ̣ὴ̣ τ̣ρυφὴ γενήσε-
	ται⸍ ἀγέλας⸍ ἐλάσο⸍ δέν̣δρη⸍̣ κλαδεύσο⸍ ‵ ̣ ̣ ε̣α̣ν ἀνατρε̣ί̣σο⸍′
8	συρίνγει πέξω⸍ μόνη ωλυκον⸍ μώνον ἄ̅-
	παγέ με μετ’ αἰσοῦ⸍ ἐν ἀ{ν̣}γρῷ καταμένιν⸍
		‵μή με λυπήσῃς⸍′
	βοῦτα⸍ πάντα σοι πωήσω⸍ ὅσα δῖ χο̣ρεικῖν̣⸍
	γυναῖκαν ὡς ἀπὸ Κόπτ̣ο̣υ̣ . . . . [
12	τ̣ι̣αν⸍ ἐπέβην⸍ φ[
	καλὸν ὕδωρ⸍ ἐχ̣ε[
	οὐκ ἔμελλέ μοι[
	κοὐκ ἐστεξάμην̣[
16	ἔπια πικρὸν ἁ‵λ′μυ̣[ρ
	εἰς Ἀφρωδείτην⸍ [
	]αν⸍ ηκα̣[
	] ̣[
	- - - - - -

1 l. με, l. καταμένειν 2 l. με 3 l. ποιήσω, l. ὅσα, l. χορηγεῖν, l. γυναῖκα 4 l. βρόδον, l. ποιήσω 5 l. στρώσω, l. παθικῶς 6 l. ὃ καὶ ὁπόσο(ν)? 7 l. ἐλάσω, l. κλαδεύσω, intr. l. ἀνατρήσω? 8 l. συρίγγι, l. ὦ Λύκων, l. μόνον? 9 l. μετʼ ἐσοῦ, l. καταμένειν 10 l. ποιήσω, l. δεῖ χορηγεῖν 11 l. γυναῖκα 12 l. or ν̣αν?, Φ[οινικ-]? 14 l. οὐκ ἔμελέ μοι? 17 l. Ἀφροδείτην

Country-dweller, lead me away to live with you in the country. Do not let me down! I will do everything for you, whatever a woman is to provide. I will transform your hut into a rambling-rose and will make a bed obediently (?),35 country dweller, […] whatever and however much with you, will become luxury for me. I will drive herds, I will cut off wooden sticks and those that I make hollow (?), I will fasten into a panpipe, all alone, o Lykon, only lead me away to dwell with you in the country! Herdsman, do not let me down! I will do everything for you, whatever a woman is to provide […] When from Koptos […] I came down … Ph[oinikon?…] good water […] I did not care to […] (or, “he/she/it was not to […] to me”?) and I have not covered (?) […] I drank bitter and salty […] in Aphrodite […].

28The question that presents itself is what circumstances could be associated with the proliferation of school ostraca attesting systematic education as well as literary texts in Didymoi in the 3rd century. It has been noted that archaeological record indicates “a very significant evolution in the rhythm of daily life” in the forts of the Berenike road in the middle of the 3rd century.36 The internal spaces of the forts were divided in smaller sections of irregular plans and continuously reconstructed, whereas some rooms were not used anymore for habitation, but converted to dumping grounds. In Dios, small pens for livestock and/or poultry were constructed, a possible sign of diminished mobility and trading between the forts. Archeological evidence also reveals an increase in bread production, made manifest by the multiplication of ovens in Xeron, Dios, and Didymoi, and the number of silos and millstones at the later.37 The large volume of bread production has recently been associated with the establishment of a “working” relationship between the garrisons and the desert dwellers, with the inhabitants of the forts enlisting help of the local population in securing the safety of the desert, for which they supplied them with wheat and bread in return.38 Could it be that life in the garrisons became somewhat more stable, settled, and less dependent on faraway centralized powers than it was under the High Empire, creating the environment more conducive to educational and literary activities? Could it also be that the higher volume of bread production meant an increase in the female population of the forts and thus more children living there, who would be a natural object of systematic schooling? Unfortunately, it is impossible to determine whether school texts in Didymoi were penned by children or adult learners, nor can we be sure that children were likelier to be taught more systematically than adults, even if it intuitively seems so. Be that as it may, the evidence from Didymoi shows that interest in literary activities and schooling changed significantly over the fort’s two hundred years of existence, apparently culminating in the last phase of its occupation in the middle of the 3rd century. It also makes it clear that any discussion of literacy in the forts of the Eastern Desert, and probably in Egypt in general, ought to be more diachronically nuanced.