skip to content

Greco-Coptic exercises in Hebrew script: T-S Ar.35.109

Julia G. Krivoruchko


געדענקט זשע, קינדערלעך, געדענקט זשע טייערע,

וואס איר לערנט דא;

זאגט זשע נאך א מאל, און טאקע נאך א מאל:

קמץ־אלף: אָ!

M. Varshavsky, Oyfn Pripetshik


T-S Ar.30.109 recto

T-S Ar.35.109 recto


The fragment T-S Ar.35.109 was characterized by Baker & Polliack 2001 as ‘possibly from a magical text’. The fact that it contained alphabetic material was discovered by G. Bohak, who presented his findings on the seminar of Divinity Faculty, University of Cambridge, 16.01.2008 as ‘A Greek Primer from the Cairo Genizah’.1 I am grateful to Gideon Bohak for permission to publish the fragment, and for placing his preliminary transcription thereof at my disposal.


0.1.    Codicology and palaeography

The fragment T-S Ar.35.109 is a paper bifolio 20.5 x 24 cm, ‘mutilated, stained and rubbed’ (Baker & Polliack 2001, ibid.), once part of a quire. Sewing holes are visible in the middle; the position of the bifolio in the quire is unclear. Since the actual direction of folding/reading is uncertain, I will call the side labelled ‘Front’ recto, and ‘Back’ verso.2

As follows from the palaeographic description in the footnote 2, the bifolium was reused as writing material.

Edna Engel (private communication, Bohak 2008) described the writing of the main Hebrew-script sections as ‘non-professional Oriental hand of the eleventh century’.

The Judeo-Arabic texts (JA) on recto and verso are written in phonetic orthography (Blau & Hopkins 1987). They seem to belong to independent textual units discussing respectively lexicological and grammatical topics.3 Since engagement with Arabic lexical semantics presupposes an appropriate language competence, it is sensible to presume that the scribe was a native — or at least proficient — speaker of Judeo-Arabic. 

Below I will deal with three sections of T-S Ar.35.109, namely the list of letter names (1), the syllabary (2) and the abecedaria (3). The description I give here is solely introductory, aiming to attract the attention of relevant specialists. The transcription is based on Bohak 2008. Uncertain readings are marked with double underline, lacunas with square brackets, scribal deletions with strikethrough, additions between the lines with angular brackets, and non-identifiable graphemes with question marks.


0.2.    Text typology

Teaching reading and writing in pre-modern societies was a highly traditional activity that consisted of long-established types of exercises following in established order. In the words of Dionysius of Halicarnassus, 

τὰ γράμματα ὅταν παιδευώμεθα, πρῶτον μὲν τὰ ὀνόματα αὐτῶν ἐκμανθάνομεν, ἔπειτα τοὺς τύπους καὶ τὰς δυνάμεις, εἶθ’ οὕτω τὰς συλλαβὰς καὶ τὰ ἐν ταύταις πάθη, καὶ μετὰ τοῦτο ἤδη τὰς λέξεις καὶ τὰ συμβεβηκότα αὐταῖς, … ὅταν δὲ τὴν τούτων ἐπιστήμην λάβωμεν, τότε ἀρχόμεθα γράφειν τε καὶ ἀναγινώσκειν, κατὰ συλλαβὴν <μὲν> καὶ βραδέως τὸ πρῶτον …

When we are taught letters, we first learn their names, then their shapes and phonetic values, afterwards syllables and their variations, then words and their properties … Only having learnt these, we begin to write and read, syllable by syllable, at first slowly … (De compositione verborum XXV, ll. 249–257, ed. Radermacher & Usener).4

Novelties were not welcome, and offenders not tolerated. Protecting the ages-old educational practices was recognized as a necessary part of maintaining social order, a notion common across very divergent cultures, cf.: 

tamen si quempiam ludimagistrum audiremus, conantem docere puerum syllabas, quem prius litteras nemo docuisset; non dico ridendum tamquam stultum, sed vinciendum tamquam furiosum putaremus, non ob aliud, opinor, nisi quod docendi ordinem non teneret.

If we heard of a teacher trying to teach syllables to a boy who did not already know his letters, we would not just laugh at him and consider him foolish, we would think that he was totally out of his mind, because he did not follow the sequence of teaching (St Augustine, De Ordine II 7 (24), transl. by Cribiore 1996:139). 


ודכר איצא //אן//תם בעץ אלמעלמין ואנה יעלם [אל

צביאן בלא הגא ולא נקט והדא מא לא יגוז לאן אצל אלתעלים אלהגא ואל[נקט

והו אלדי ינדר אללה בה ובה יסתקים אמור אלמתעלמין ועליה אלמ[עוול

פאן אלקראה ללצבי בלא הגא ולא נקט מא תפיד שיא ולא ינתפע בהא

You mentioned in your letter that there is a teacher here who instructs children without the alphabet and the vowels. Such a thing is forbidden, because the basis of all knowledge is the alphabet and the vowels. It is God who commanded us to carefully follow the alphabet, so the pupils should rely upon it, because it is the foundation of everything. Teaching a child to read without the alphabet and the vowels is useless (T-S 13J23.20, a responsum by the judge Isaac b. Samuel ha-Sephardi in the hand of Ḥalfon b. Manasseh, 12th century, transl. by Goitein 1962:41, cf. 1971:83–110).

The traditional sequence of basic teaching procedures has been uniform to the degree of being universal: Indo-European Greek, Semitic Hebrew or Afro-Asiatic Coptic, RTL or LTR, alphabet or abjad — in the initial stages the structural differences between the languages and writing systems almost did not matter. Pupils were required to learn the nomenclature of letters (i.e., to know their names and identify them by sight), to learn the phonetic value of letters, and to learn to combine these phonetic values into syllables - again, identifying the phonetic value with its grapheme. When required to write, the students proceeded from alphabets to syllabaries and then to larger text units. The classical monographs of R. Cribiore (1996; 2001) illustrate how these stages were passed through in Hellenistic and Roman Egypt. Her work became foundational for researching education after the Byzantine and Muslim conquests, thus contributing to our understanding of medieval Egyptian multilingualism.5

The Cambridge Genizah abounds in similar exercises: Hebrew letter practice in every order (random sequences, complete alphabets, atbash, coupled atbash, etc., see folder T-S K5), letters with vocalization (e.g., T-S K5.2, T-S K5.4, T-S K5.19, T-S K5.55), separate words, copies of complete sentences and lengthy biblical passages, etc.6 As a set of educational materials, T-S Ar.35.109 is standard in some of its aspects and unusual in others. It opens with the list of letter names (1), τὰ ὀνόματα of Dionysius of Halicarnassus. Such lists were performed by/intended for those who already could read, which makes them much rarer than the pupils’ alphabetically or otherwise ordered scribbles. Our list presumes Hebrew literacy.

Then follows the syllabary (2) opening with what I will call a ‘pattern’: the names of vowels followed by their respective sound values: Vn [Vn]. Further it continues with the following sequences:

C1V1[C1V1], C1V2[C1V2], …, …, C1V7[C1V7], 

C2V1[C2V1], C2V2[C2V2], ..., …, C2V7[C2V7], ..., ..., up to [C17V7],

where C=name of a consonant, V=name of a vowel, [CV]=sound value of the syllable. Below I will call the syllables with the same consonant a ‘series’.

Fortunately, Greek literature has preserved to us a recording sui generis of how such syllabaries sounded when practised. Athenaeus of Naucratis, a Greco-Egyptian intellectual of the 2nd-3rd century AD, tells us about the play composed by Callias the Athenian. In its prologue the actors perform an alphabetical poem:

<ἄλφα>, βῆτα, γάμμα, δέλτα, θεοῦ γὰρ εἶ,

ζῆτ’, ἦτα, θῆτ’, ἰῶτα, κάππα, λάβδα, μῦ,

νῦ, ξεῖ, τὸ οὖ, πεῖ, ῥῶ, τὸ σίγμα, ταῦ, <τὸ> ὖ,

παρὸν φεῖ χεῖ τε τῷ ψεῖ εἰς τὸ ὦ.

Each actor pronounces their letter/s, and they repeat it in reverse order. Then the chorus of women sings the following:

βῆτα ἄλφα βα, βῆτα εἶ βε, βῆτα ἦτα βη,

βῆτα ἰῶτα βι, βῆτα οὖ βο, βῆτα ὖ βυ, βῆτα ὦ βω,

the next strophe being

γάμμα ἄλφα, γάμμα εἶ, γάμμα ἦτα, γάμμα ἰῶτα,

γάμμα οὖ, γάμμα ὖ, γάμμα ὦ … .   (Deipnosophistae, book 10, § 79, ed. Kaibel, cf. Athenaei dipnosophistarum epitome, vol. 2:2, p. 46, ed. Peppink).

They continue with the remaining combinations until the end of the alphabet.

I believe that Kallias’s drama is a parody on the schooling process to which all educated Greeks were subjected in their childhood. It must have been significantly standard to provoke a predictable response from the audience. For our purposes, let us note that the syllabary is is ἔμμετρος ἅμα καὶ μεμελοπεποιημένος, a chant ‘with meter and melody’.7

It is important to realize the difference between the Kalias’s theatralized classroom and T-S Ar.35.109 on one side, and the bulk of Genizah / papyri / ostraca / wood tablets / otherwise preserved syllabaries on the other: the latter do not contain the names of the letters, neither reflect phonetics, they are just grapheme combinations.8 While they evidence for studying writing, ours aims in teaching reading: it is an ‘oral syllabary’. 

Finally, the T-S set is complete with a standard writing exercise of the abecedaries and separate letters (3)


T-S Ar.35.109 verso

T-S Ar.35.109 verso


1.    Letter names in alphabetical order

Text (ll. 1–3v left)


[     ]טא: ג'מא: דילטא אֵי זיטא [    ]

[  ]וטא: קמבא: למדא: מִי נִי אכסי או פי רו

טאוה: הוה פי: שי: אפּסי: אוטומיגה

The alphabet does not include stigma or other numerical symbols. Presuming the standard Greek order, the names for Η and Θ at the end of the first line and for Σ in the second line are missing.



The names of individual letters are separated through either spacing or colon, in one case through a raised dot (mid-line after gamma).9

A circlet with a dot inside marks the end of the alphabet.10 The line is drawn under the alphabet to signify a completed exercise section as in Boak 1921:190, Husselman 1947:130–131, etc.11

What looks like dagesh in ט must be the end of the downward stroke, the rest of which was flaked off. Unclear, whether the spot above pei is a diacritic or an accidental ink splash. 



● The use of rafe with ג is irregular: it is present in ג'מא but absent in אוטומיגה, although both correspond to /ɣ/ in Greek (the fricativization of /g/ was complete by the 4th century AD, see Gignac 1976: vol. 1, 64; Horrocks 2010:170). 

From the viewpoint of the Judeo-Greek (JG) spelling, it is not unusual: in T-S NS 122.126 the word λέγουν is spelled with diacritic on 4r l. 13 but without on 3v l. 12. Only thoroughly executed texts, such as T-S Misc.28.74, are more consistent. Given that upon fricativization /g/ became rare, rafe turned out to be superfluous in the absolute majority of cases, and therefore optional.12

● אֵי accurately reproduces the historical Greek name of the letter epsilon, which has been also inherited by Coptic. The ninth-century grammarian Georgius Choeroboscus, explaining the persistence of ει in the declension paradigm of passive aorist participles, wrote: 

καὶ αὐτὸ δὲ τὸ εἶ τὸ ὄνομα τοῦ στοιχείου διὰ τοῦ ε καὶ ι γράφεται καὶ ἐκφωνεῖται – τούτου χάριν ἡ ει δίφθογγος ἄτρεπτός ἐστιν·

Even the very name of the letter ει is written and pronounced (emphasis mine — JK) with ε and ι, that is why the diphthong ει is unchangeable (Prolegomena et scholia in Theodosii Alexandrini canones isagogicos de flexione verborum, p. 51, ed. Hilgard).

Cf. the reflections of the twelfth-century scholar Eustathius of Thessalonica:

Ἰστέον δὲ ὅτι τὸ ε στοιχεῖον εἶ ἔλεγον οἱ παλαιοὶ προστιθέντες τὸ ι, ἵνα τῇ διὰ διφθόγγου ἐκτάσει δύνωνται περισπᾶν καὶ αὐτό, καθὰ καὶ τὰ ἄλλα στοιχεῖα. Τοιοῦτον δὲ ποιοῦσι καὶ ἐπὶ τοῦ μικροῦ ο. Καὶ ἐκεῖνο γὰρ διὰ τὴν αὐτὴν αἰτίαν οὖ λέγουσι.

You should know that the letter ε was called by the ancients ει, with the addition of ι, in order that through lengthening it will be able to receive circumflex like other letters. The same is done with the small ο, which is called ου for this very reason (Commentarii ad Homeri Iliadem, vol. 2, p. 1, ed. van der Valk).

Since our או lacks vocalization, it may reflect either [o] or [u].13 Remarkably, the scribe did not opt for further specification, although the Hebrew writing system provides him with suitable diacritics. 

● There is no phonological reason in Greek, why [p] in κάππα should suffer voicing – but it would be normal for Coptic and Arabic, cf. the Coptic letter name transliterated into Arabic as كَبّا in the MS dated to the 13th century (Monferrer-Sala 2015:282, cf. Prince 1902:292). Writing down [b] in Greek would call for the spelling κάμπα, which transliterated into Hebrew letters would give קמבא

● The letters μ and ν have not preserved their classical Greek names μῦ and νῦ, coinciding instead with Coptic ⲙⲏ, ⲙⲉ, ⲙⲓ and ⲛⲉ, ⲛⲓ (Crum 1939: s.vv. Ⲙ and N; Kasser 1991:40).

● A prothetic vowel is not required by Greek phonology in אכסי  < [ksị] and אפּסי < [psị], but would be not out of place in Bohairic and Arabic, cf. اكسى , ابسى (Crum 1939: s.vv. Ⲝ and Ⲯ). 

● The final in the name טאוה is unusual. By analogy with אוטומיגה = ὦ τὸ μέγα, one would expect something like *[taua/tauwa/tawa], since inside the section and seem to be interchangeable for final [-a]. In JG orthography this choice is often a matter of taste: e.g., T-S K24.14 spells every final [-a] with aleph, but once with hei, and once with both. I do not know any name for the letter τ with an [a] Auslaut in Greek or Coptic. 

JG spelling has an option of writing final [-e] with hei in monosyllabics, probably in imitation of Hebrew זה. This may mean that the letter-name can be read as *[taue/tave], but I could not verify this variant either.

It is perhaps worth mentioning that some names of Coptic letters (e.g.  مية,  نية etc., see Stern 1880:418, et al.) have been spelled in Arabic with ta marbuta, for which ה could be a suitable equivalent. The questions as to why ta marbuta was used in transcribing Coptic letter names, how old this tradition is, and whether it is evidenced for the particular letter, should be addressed by specialists.

● The initial hei in הוה  demonstrates that the letter upsilon, more familiar under the name ὖ, could have been also called ὗ, cf. Coptic letter name ϩⲉ (Kasser 1991:40). Vav shows that the vowel still retained its historical rounded quality (Horrocks 2010:274). Coptic variants with back vowel are also known, e.g., او (Crum 487 s.v. Ⲩ(ⲞⲨ)). For the final hei cf. Coptic names ⲩⲉ, ⲩⲁ (Kasser, ibid.). 

● שי  may reflect [ʃi], a realization of the classical letter-name χεῖ, Hellenistic χῖ following the palatalization of the velar before high front vowel. The phenomenon is known in the South-Eastern Greek dialect area, see Holton et al. 2019: vol. 1, 194–198. For a similar pronunciation of the Coptic letter name reflected in Arabic script compare شيه (Monferrer-Sala, ibid.). More on this below in (2), under series. 


Source/s the list of letter names 

The list seems to have been based on a Greek alphabet of professional grammatical quality using precise names, such as אֵי, אוטומיגה = ὦ τὸ μέγα vs. simple τὸ ὦ. Its transmission by the speakers of Coptic and possibly (Judeo-)Arabic led to the transformation of the letter names in accordance with their native phonology. 


2.    Syllabary 

To facilitate the analysis, the material is presented in the table. For comparison, the column and the row marked ‘name’ contain the names of the letters inherited from the previous section. 


Text (l. 4v left – l. 14r right)14

title of image here


Notes on the pattern (ll. 4–5v left)

Inside the pattern, the scribe does not divide the name of the vowel from its phonetic value, using instead scriptio continua.

● אלפאָה corresponds to [ˈalfa.a], with further synaeresis a+a > ā, kamatz (and hei) being used to mark the length/combined origin of the resulting vowel. 

● Comparing אַיא to the relevant data of the previous section, we should divide it as *[ai.X], X being the sound value of epsilon, to which this pattern refers. Several questions emerge: is patah a mistake for tzere (1) or a conscious choice of the scribe, i.e., should we presume that the name was pronounced differently in (1) and (2)? why was an unvocalized alef chosen to write down a mid vowel, if it was indeed a mid vowel? The pattern is unclear, but its applications in the series, see below, are slightly more informative. 

● Both אִיחטַיִי and אִיוֺטַאיִי exhibit prothetic vowel [i]. The consonantal Anlaut of the name eta surfaces in its Coptic equivalents: (ϩ)ⲏⲧⲁ, ϩⲁⲧⲉ (Kasser 1991:40); hida and other variants with initial h (Prince 1902:292). As a result of completed itacism, the phonetic value of the vowel became /i/, cf. ⲏⲓⲧⲁ (Kasser, ibid.), which expectedly appears in the series below. Hence, the name of the vowel eta can be reconstructed as *[iˈhita]. 

Double yod does not imply a phonetic value [ji] but is a graphic device to separate the names of the letters from the phonetic value [i], which is a separate — albeit attached — entity. 

● The part of the pattern referring to omicron, upsilon and omega appears damaged and/or misunderstood by the copyist. It is possible that some text was lost at the end of l. 4, and some scrambled – which, given the abundant mistakes in the series, would not be surprising. The passage should have consisted of three Vn[Vn] units, and contained three values, namely [o], a development of classical upsilon (> [y]/[u]/[i]/[?]), and [o] – provided the writer’s system had such vowel as [o], and it was not closed into [u]. Trusting the preserved text, at least one of the vowels should have had either the phonetic value [u], or the name [u].

I can offer no convincing explanation of the passage: several Coptic letter names can be matched to the currently visible text, but none would allow reconstructing all the three units without assuming miscopying, dittography, and the like.


Notes on the series (l. 5v left – l. 14r right)

Just as there is no spacing between Vn and [Vn] in the pattern, there is no spacing between the name of the second letter and the resulting syllable in the series. Counterintuitively, the text is mostly spaced not as Cn | Vn | [CnVn] or CnVn | [CnVn], but as Cn | Vn[CnVn], which may be conditioned by stress assignment rules and/or metrical structure. 

More evidence about metrical structure can be probably derived from the spelling. Mid-word, alpha can be written fully as אלפא (cf. unexpected קמבאאלפאקא), defectively as (אלפכסא) אלפ, or even as לפ in (תיטלפתא). Eta is normally חטא, but חט when combined with δ, ξ, ψ. Similarly, iota is normally יוטא, but יוט in the series of δ, θ, ξ, ψ. This distribution may mean that the above consonants were pronounced as clusters with facilitating vowels, whose length balanced out the missing [a] of alef

It seems unlikely that the metre of the syllabary was flawless, and even unlikelier that it could be reconstructed from a source as imperfect as this, but some hints on its existence are present.



● The letter sigma is called sima in typically Coptic fashion, cf. ⲥⲏⲙⲙⲁ, ⲥⲩⲙⲙⲁ, ⲥⲓⲙⲁ, ⲥⲙ︥ⲙⲁ (Kasser 1991:40).

● Voiceless stop /p/ everywhere (π, ψ) appears as /b/, as in Coptic and Arabic. 

● Since the assibiliation /x/ > /ʃ/ before the front vowels does not appear anywhere apart from the letter-name itself, it does not seem to be a bona fide phonological process, but rather a phenomenon limited to the particular lexical item. One wonders whether שי as a letter-name has not been influenced by the name of ϣ, with whom it shared its value /ʃ/. 

● The following table summarizes the changes in spelling of consonants, triggered by vowel addition. Omicron and omega do not trigger any alternations.15 Only those cells, in which changes occur, are populated.


title of image here

* Evident mistake. 

** See Palaeography above.


Provided the spelling is of Judeo-Greek type, this table shows that the vowels ε, η, ι and possibly also υ trigger fricativisation of the preceding consonants: /t/ > /θ/, /k/ > /x/ (there is not enough data to establish, whether /d/> /ð/). If the oral syllabary is supposed to provide a reading rule – perhaps not universal, but at least statistically useful – we would end up with very strange Greek indeed, because sweeping fricativization would preclude such syllables as /ke/, /ki/ or /te/, /ti/ that never ceased to exist in the Greek language. Similarly, there is no reason in Greek why perfectly normal clusters should be simplified /ksy/ > /sy/, /psy/ > /bsy/ > /sy/ or an open front vowel should trigger affricativization (unless it is pseudo-data, see ‘Mistakes’ below). 

Also, writing the syllables with /a/ identicaly to those with /e/ would create a dysfunctional writing system, since such syllables constitute a significant proportion of an average Greek text. Phonologically, an idiom that neutralizes these vowels in all the positions suggested by the syllabary is unknown to Greek linguistics. Similar problems have already surfaced in the second pattern above.

A logical conclusion here would be that we are dealing with the imposition of some foreign phonological system/s on top of Greek. To analyse T-S Ar.35.109 correctly, one should revise the normal Judeo-Greek values of Hebrew characters and diacritics taking into account the impact of this system, if we are to achieve more realistic results. 

The section (3) seem to point in the direction of Bohairic (or, much less probably, Fayyumic) as a possible overlapping system – provided the abecedaria were written/added in the same environment/region as the syllabary. The primer does not teach Greek, it teaches Copto-Greek.


Mistakes – or not?

Unfortunately, the state of the fragment and its uniqueness does not allow for easy differentiation between scribal mistakes and linguistically informative details. The text is often inconsistent: spellings like ניווטא ,מיווטאמי could result from vav/yod confusion during mechanical copying. A syllable that should appear as [ri] ends in vav. Yods may be omitted, e.g., from the name of a vowel (אכסי אכסא, expected אכסי איכסא) or a consonant: (שאלפכא, expected שיאלפכא). In זהטלפאזה the first hei appears for yod and the second for aleph, etc.


Sources of the syllabary

Given the differences in the letter names, sections (1) and (2) should have originated from different sources. 

The syllabary has plenty of visual errors: dittography at the end of verso and l. 4r, confusion of dalet and resh in l. 3r, possibly also of vav and yod in ll. 20 and 21v.16 They would have been impossible in a dictation, so the text is an obvious copy of a written source. Quite another question is how this written prototype was generated, and here everything points to either dictation or recording of the chant.

The inferior quality of the text and its model is remarkable: not even two series among seventeen follow the pattern in a consistent manner. Omissions and miscopying are expected from absolute beginners, such as children or illiterate monks, but our scribe was a rapid writer, and his mistakes could be due either to carelessness or a faulty model (a model in bad condition or a model with errors).

It is commonly believed that Greek persisted in Egypt until the end of the 8th century, perhaps longer.17 Whatever Greek-speakers remained in Egypt would gradually succumb to the pressure of official Arabic and/or domestic Coptic, and a good-quality Greek education would be increasingly difficult to acquire locally. The 11th century, the era of victorious Arabization, must also have been a period of decline in Coptic language competence.18 Under such circumstances, well-preserved models of ‘proper Greek’ or even ‘proper Coptic’ could have been no longer obtainable in the scribe’s environment. Yet the very existence of such models shows that our scribe had predecessors in his pursuit of Greek/Coptic literacy.


3.    Two abecedaria and other glyphs

Text (ll. 15–19r right)

The graphemes are presented in a table for ease of comparison.




Not being a Greek/Coptic palaeographer, I would not venture to date the abecedaria, neither to extend to them the dating suggested by Engel. A gap of a generation between writing and reusing the bifolio is not improbable, and it can bring us into the twelfth century.

It has been suggested (N. de Lange 2008, seminar discussion) that List 2 served as an example for List 1, yet it is impossible, since they differ in ductus of letters and their number. Compared to the main Hebrew hand with its stable letter proportions and uniform serifs, the hand/s of abecedaria look less consistent, List 1 being particularly striking for its hugely disproportionate graphemes.19 List 2 is more elegant, but still shows irregular alignment even considering that it could have been squeezed on the top of the Hebrew.

For all the differences between the lists and their awkward penmanship, both represent (distinct varieties of) cursive minuscule. All the letters have an upright axis, rounded shapes, and few elongated hastas. These forms ultimately derive from the ‘stylized chancery script’ (Cavallo 2011:136) that spread in Arabicized Egypt and eventually developed into a book handwriting and further into standard Byzantine minuscule. Close parallels to the letters of both lists can be found on the plates of Stegemann 1936 dedicated to Coptic documentary hands of 9th–11th centuries.

Both lists include ‘extras’, i. e. glyphs that should not have appeared in their position presuming the standard order of the Greek alphabet. Below they will be called ‘after-x’, where x=preceding letter. Theoretically, an ‘extra’ can be a variant of the preceding letter, a variant of the following letter, a letter from elsewhere in the alphabet, or a letter from another alphabet. Of these options, the insertion of a random letter in an arbitrary position seems to be the less likely since both lists consistently adhere to an alphabetic order. The exercises containing random sequences of letters are well-known, e.g., Pernigotti 1985:96–97 (Greek & Coptic), but they do not contain lengthy ‘correct’ sequences.

Apart from the abecedaria, the gap between ll. 3 and 4r contains several attempts to write two letters, the last being theta. Its ductus differs from the thetas inside the lists.


Notes to List 1

Several letters are penned with prominent ligature lines (alpha and beta from below, epsilon from the middle, gamma and omicron from the top). The dot inside alpha may be a sign denoting the beginning of the unit (cf. the use of dotted circlet above) and/or a decorative element. Delta descends below the base line. Theta must have originated from an open-type cursive variant. Iota is of generous height, as is tau. Lambda consists of a vertical stroke with a hand joining at a right angle. Nu may stem from ‘maiuscolo corsivo inclinato’. There is a supralinear stroke above it, characteristically appearing above syllable-forming sonorants and preposed morphemes in Coptic script. Since the letter is both the beginning of the word and an independent item, such marking would be appropriate, cf. similar marking in Coptic alphabetic exercises (Schenke 2010:293). Rho is formed with a rounded hooked leg, while the descender of mu is straight. Omega is of open type.

The grapheme appearing in place of psi is not part of Greek minuscule, but a Coptic janja, cf. Cromwell 2020:365–366 (alphabetic list, 7th-8th cent.).

Abecedarium 1 includes two extra glyphs, ‘after-fi’ and ‘after-psi’ (or rather ‘after-janja’, as above). Since ‘after-fi’ does not look like fi or khi, it should be a misplaced ksi. The appearance of ksi in the twenty-third place in the list may be not random but motivated by the considerations of grammatosophy (Ar. علم الحروف). There is a possibility that the scribe or his prototype were familiar with the circle of ideas preserved in the Copto-Arabic MS of the 14th century (Hebbelynck 1900–1901), whose mid-6th century Greek prototype has been recently discovered. There we read: 

Τούτων τῶν κβ’ τοῦ Χριστοῦ πραγμάτων εἰσὶ τύπος τὰ κβ’ ἔργα τῆς κτίσεως, ἅ ἐποίησεν ὁ θεός: ὁμοίως καὶ τὰ κβ’ γράμματα ὡς προεῖπον τῆς ἀλφαβήτου. … Τὸ γὰρ Ξῖ καὶ Ψῖ ὕστερον προσετέθησαν τῇ ἀλφαβήτῳ ὑπό τινων φιλοσόφων.

The archetype of these twenty-two acts of Christ are the twenty-two works of creation performed by God. So too, as I said, are the twenty-two letters of the alphabet. … Ksi and psi were inserted into the alphabet later by certain philosophers (On the Mystery of Letters, pp. 112–114, ed. Bandt; cf. p. 170).

‘After-psi/janja’ may be an open variant of Ϩ hori (cf. Stagemann 1936, pl. 22, 966 AD).


Notes to List 2

Several graphemes in this abecedarium are markedly different from their counterparts above: alpha is a dotless loop; gamma does not have an angular rising connector; delta descends in a curvy stroke, not hook; theta is of closed type with a slim body and relatively long crossbar. A diaeresis above iota may occur both in Greek and Coptic writing. Omicron is closed and lacks a connector; pi has a separate hat; the upper part of the lunate sigma is longer than the lower; a decorative curl adorning fi is offset to the left, while psi is cross-shaped with a small connector.

The lack of rho may be due to random omission or subconscious avoidance by a speaker of a rho-less dialect, such as Fayyumic.

Three extra glyphs of this abecedarium are ‘after-epsilon’, ‘after-khi’ and ‘after-psi’. ‘After-epsilon’ may be an episemon for 6 (Gr. stigma, Copt. soou) in its mirrored version similar to the majuscule zeta (Ꙅ > Ꙁ). Its presence may indicate a theologically oriented source, since the number six, being divisible by two and three, was symbolic of Christ (see On the Mystery of Letters, pp. 174, ed. Bandt).

‘After-khi’ could be a Coptic Ϥ fai, cf. Stegemann 1936, pl. 23, ca. 1013 AD. ‘After-psi’ looks like Ϧ khai, cf. Stegemann 1936, pl. 19, narrow style bookhands, 933 AD. Note that the letter is Bohairic; the scribes of List 1 and 2 could have been speakers of/familiar with different Coptic dialects.


4.    Author/s and purpose/s 

While the main Judeo-Arabic text of the bifolio and sections (1) & (2) point to a scribe with linguistic interests, much less can be said about the authors of the rest. Greco-Coptic and Arabic additions could have been made by children/pupils/heirs or other individuals who came in possession of the bifolio. It is difficult to establish the exact number of hands involved in its production: the notion of ‘cross-script hand’ (a hand in multigraphic products) has not yet been sufficiently developed, and even if it were, the non-Hebrew parts are too short for confident judgement.20 It is therefore safer to dwell on the characteristics of each section, not the whole bifolio or manuscript.

The list of letter names (1) and the syllabic exercise (2) do not include Coptic graphemes, but neither can they be characterized as Greek, because they include phonological phenomena typical of Coptic and occasionally Arabic, as well as Coptic lexemes. Whatever the precise identifications of the glyphs in section (3), their presence cannot be explained from Greek, hence it would be better characterised as Greek-Coptic or Coptic-Greek.

While the purpose of the third section should be to learn Coptic, one can still conjecture that the purpose of the first two sections was learning some form of contact-modified Greek, and if all the sections were penned by the same individual (of which one cannot be sure), it is unclear, which language was the intended target. One cannot exclude dual purpose, i.e., studying both Greek and Coptic, or a non-committal track, where the decision about the specific language would be postponed to a later stage.

It remains to be established, whether the main motivation for the study undertaken by the scribe of (1) and (2) was a practical need of dealing with old legal and administrative documents, interest in mysticism or sheer intellectual curiosity. Given the content of the Judeo-Arabic sections, he could have been attracted (or prompted) by the works of 10th-century Copto-Arabic grammarians (see Sidarus 2005).

11th-century Greco-Coptic exercises accomplished by a Judeo-Arabic speaker are a rare discovery. Chronologically close educational materials come from completely different socio-cultural settings, such as Bohairic-speaking monastic communities, whose object of study was their native language and very familiar liturgy. The scribe of our fragment would have had different inspirations and passed through different study tracks. Publishing the remaining Judeo-Arabic parts of T-S Ar.35.109 should help to put the above passages into their proper context, including the Jewish context.



1 During that period G. Bohak was funded by the Israel Science Foundation (Grants no. 725/03 and 635/08).

2 Recto: on the right side, the main text in dark ink, fourteen lines in total, starts from the very top of the leaf. The writing line of the main text frequently diverts from the horizontal; only the right margin is aligned. There is a cleave-like gap between the third and the fourth line into which Greek and Arabic letters in brownish ink have been squeezed. Below the main text there is a sequence of non-Hebrew letters, three lines long, in blackish ink from a thick writing tool (further – List/Abecedarium 1), followed by a divider and two more lines in paler ink of the same shade but neater pen (further – List/Abecedarium 2). The three lines above the divider are crossed out, possibly by the hand that continued with List 2. Finally, at the lowermost part of the page there are three more lines of Hebrew script, in brownish ink. 

On the left side, the main text in Judeo-Arabic twenty-five lines long, is written in the same fluent but inaccurate hand as the main text on the right. The side margins are minimal, without alignment or justification. A different hand inserted on the upper margin a basmallah in Hebrew letters. The main text ends mid-line with three vertical dots. Further in the same line another hand penned אשר in darker ink and thinner writing tool. Leftmost there are several words in Hebrew script, in brownish ink, and an extra line, possibly in yet another hand. Some glyphs in this line may be non-Hebrew. A line dividing the folio is drawn below, followed by two incomplete lines in Hebrew script. Judging from the colour of the ink, the divider and (at least) the last two lines may belong to the same hand as the text above. 

Verso: on the right, the main text, 14 lines long starts from the very top of the page. The mid-page gap is followed by a smaller section of 2.5–3 lines, a divider and five more lines in brownish ink. Inside the gap, there are at least 4 lines of brownish jottings perpendicular to the main text. The first and the second sections of horizontal writing are overwritten with pen trials or scribblings in black ink. Two glyphs on the left resemble the Heb. letter ט, and Greek/Coptic γ and ν from the List 1 scribbled upside down at ca. 15˚ to the base line. The left side opens with three lines in thick black ink followed by a divider and another twenty-one lines in the same hand. In continuation of the last line several glyphs (או?) were penned in dark brown ink with a thinner pen. Below there is a line of accurate smaller letters and another with larger ones in discoloured brownish ink similar to that of the recto and the opposite side of the bifolium.

3 Many thanks to N. Vidro and A. Ashur for consulting me on the JA texts.

4 Here and below the translations of primary sources are mine, unless stated otherwise.

5 See also Cribiore 2007 for Coptic.

6 For an introduction see Olszowy-Schlanger 2003.

7 More on Kallias’s play, its dating, suggested titles and interpretation see in Smith 2003 with earlier bibliography. 

8 Of course, phonetic data can be extracted from them also — through secondary reasoning, such as observing mistakes, etc. The combinatoric nature of such exercises is particularly evident in complex letter combinations, which hardly occur in real speech and have no reality as linguistic units.

9 By default, the names of the letters here and below are given in their Greek versions.

10 The practice of using circlets as dividers is known also in Judeo-Arabic writing.

11 Cf. Cribiore 1996:76. 

12 Cf. Blau & Hopkins 1985:440 on the irrelevance of the corresponding rafe in early JA.

13 The phonetic transcriptions here and below are by necessity approximate and should be adjusted once the described language variety is better understood. 

14 The pattern and the syllabary are executed as a uniform text block; no underlining or empty lines separate between the two. 

The first units Vn[Vn] are separated with colons, which are then dropped after iota. Inside a series, each CnVn pair is separated from the following with a colon. In the beginning it is consistently placed, then omitted after delta and dropped altogether after kappa, re-appearing in rho and continuing inconsistently. In one case (after קמבא יוטאכי) the colon looks inserted into the text upon writing.

The end of the pattern is marked with a circlet. Further on, circlets are used to separate series. A circlet also occurs after the first unit of beta, showing that the scribe either initially misunderstood the division practice, or changed his mind re what division signs to use as he was writing. The use of circlets starts as systematic, but soon becomes chaotic, and then they are dropped altogether. There is a circlet on the margin at the end of l. 16v, possibly supplying the missing circlet between the kappa and lambda series. The end of the syllabary is marked by circlet with a dot inside.

Dittography in the last line of verso is deleted through supralinear carons, and above the fourth line of recto through dotted strokes. The number of cancelling strokes does not correspond to the number of deleted characters. 

It is unclear what was intended in the last complete word of l. 4v. There is a line under both vavs and hei, possibly deleting a previously written hirik, tzere or patah, or perhaps even the letters themselves. Also unclear is whether the dots above vavs were intended as vocalization [oo] or deletion signs; perhaps there are also two deleting dots above hei

Starting from the mi series, the ligature aleph lamed is used to write the letter name alpha

Supralinear diacritic called here rafe for the sake of uniformity, may occasionally have a shape of a dot. At times it is difficult to distinguish it from an accidental spot, e.g. in ד׳, l. 9v.

The series are characterized by sparser use of diacritics than the pattern and section (1). It is unclear, whether the dot between the last lines of recto is a ḥolem belonging to the last line (sc. אֺבסי) or a ḥireq belonging to the preceding line (sc. אִבסי), cf. the position of ḥireq under the left leg of aleph on l. 4v.

15 A dagesh appears in bet preceding omicron but is absent in all other bets. I treat it as accidental. Similarly, I treat the data re gamma as insufficient, pace the presence of rafe before omicron

16 Provided it is a spelling phenomenon, not phonetic one. 

17 E.g., “Greek continued to be used several decades after the Arab conquest: in Egypt, the latest explicitly dated papyri in Greek are from 796–7 CE. In Fustat – the main administrative centre – Greek was used longer than in rural areas which adopted Coptic, before Arabic became the official language of administration. Jewish minorities maintained the use of Greek for an even longer period” (Olszowy-Schlanger 2014:297); 

“Sometime in the ninth century Greek stopped being used as a language of administration, and Coptic, although continuing longer, especially for private communications, finally disappeared as a living language – at least as far as our textual evidence is concerned – in the fourteenth century” (Sijpesteijn 2016[2010] :106), cf. Bucking 2007:242. For more analysis of the place of Greek in the Egyptian socio-linguistic landscape see de Jong 2015; 2022, Mavroudi 2014.

18 See Richter 2009:417–419, 426–434; 2017:44; Parker 2013:224–225; Papaconstantinou 2007:291–292.

19 In terms of Cribiore, L1 is almost ‘zero-level’ (1996:112). Was it somebody’s first attempt to write Coptic?

20 Cf. “the dynamism of the Arabic hand is somewhat similar to that of the Greek, although it is very problematic to compare the same handwriting in two different kinds of scripts” (Berkes & Khaled 2012:98).



Baker, Colin F., and Meira Polliack. 2001. Arabic and Judaeo-Arabic manuscripts in the Cambridge Genizah collections : Arabic old series (T-S Ar.1a–54). Cambridge: Cambridge University Press.

Bandt, Cordula. 2007. Traktat "Vom Mysterium der Buchstaben": kritischer Text mit Einführung, Übersetzung und Anmerkungen. Berlin: De Gruyter.

Berkes, Lajos, and Khaled Younes. 2012. "A Trilingual Scribe from Abbasid Egypt? A Note on CPR XXII 17." Archiv für Papyrusforschung und verwandte Gebiete 58: 97–100.

Blau, Joshua, and Simon Hopkins. 1987. "Judaeo-Arabic Papyri." Jerusalem Studies in Arabic and Islam 9: 87–160.

Boak, Arthur Edward Romilly. 1921. "Greek and Coptic School Tablets at the University of Michigan." Classical philology 16 (2): 189–194.

Bohak, Gideon. 2008. A Greek primer from the Cairo Genizah. Paper presented on the Hebrew, Jewish and Early Christian Studies Research Seminar, Faculty of Divinity, University of Cambridge.

Bucking, Scott. 2007. "On the Training of Documentary Scribes in Roman, Byzantine, and Early Islamic Egypt: A Contextualized Assessment of the Greek Evidence." Zeitschrift für Papyrologie und Epigraphik 159: 229–247.

Cavallo, Guglielmo. 2011. "Greek and Latin Writing in the Papyri." In The Oxford Handbook of Papyrology, edited by Roger S. Bagnall. 101–148. Oxford: Oxford University Press.

Clackson, Sarah, and Arietta Papaconstantinou. 2010. "Coptic or Greek? Bilingualism in the Papyri." In The Multilingual Experience in Egypt, From the Ptolemies to the Abbasids, edited by Arietta Papaconstantinou. 73–104. Farnham: Ashgate.

Cribiore, Raffaella. 1996. Writing, teachers, and students in Graeco-Roman Egypt. Atlanta, Ga.: Scholars Press.

---. 2001. Gymnastics of the mind: Greek education in Hellenistic and Roman Egypt. Princeton: Princeton University Press.

---. 2007. "The Coptic school exercises in the collection of Columbia University." In Akten des 23. Internationalen Papyrologen-Kongresses (Wien, 22.–28. Juli 2001), edited by Bernhard Palme. 127–130. Vienna: VÖAW.

Cromwell, Jennifer. 2020. "Writing Exercises from Wadi Sarga: O.Sarga II 1–14." Archiv für Papyrusforschung und verwandte Gebiete 66 (2): 359-376.

Crum, Walter Ewing. 1939. A Coptic Dictionary. Oxford: Clarendon Press.  Reprinted by London: Sandpiper Books Ltd, 2000.

de Jong, Janneke H. M. 2022. "A Changing Position of Greek? Greek Papyri in the Documentary Culture of Early Islamic Egypt." In Egypt and the Eastern Mediterranean World: From Constantinople to Baghdad, 500–1000 CE, edited by Jelle Bruning, Janneke H. M. de Jong and Petra M. Sijpesteijn. 379–407. Cambridge: Cambridge University Press.

de Jong, Janneke H. M., and Alain Delattre. 2015. "Greek as a Minority Language in Early Islamic Egypt." In The Late Antique World of Early Islam: Muslims among Christians and Jews in the East Mediterranean, edited by Robert G. Hoyland. 37–62. Princeton: Princeton University Press.

Gignac, Francis T. 1976. A Grammar of the Greek Papyri of the Roman and Byzantine Periods. Milano: Goliardica.

Goitein, S. D. 1962. Jewish education in Muslim countries: based on records from the Cairo Genizah. Jerusalem: Ben Zvi Institute.

---. 1971. "Side Lights on Jewish Education from the Cairo Geniza." In Gratz College Anniversary volume. 83–110. Philadelphia: Gratz College.

Hebbelynck, Adolphus. 1900–1901. "Les mystères des lettres grecques. Texte copte, traduction, notes." Le Muséon 19, 1900, pp. 5–36, 105–136, 269–300; 20, 1901, pp. 5–33, 369–414.

Hilgard, Alfred. 1894. Grammatici Graeci, vol. 4.1. Leipzig: Teubner (repr. Hildesheim: Olms, 1965).

Holton, David, Geoffrey Horrocks, Marjolijne Janssen, Tina Lendari, Io Manolessou, and Notis Toufexis. 2019. The Cambridge Grammar of Medieval and Early Modern Greek. 4 vols. Cambridge: Cambridge University Press.

Horrocks, Geoffrey. 2010. Greek: A History of Language and its speakers. 2 ed. Chichester: Wiley-Blackwell; first ed. London: Longman 1997.

Husselman, Elinor Mullett. 1947. "A Bohairic School Text on Papyrus." Journal of Near Eastern Studies 6 (3): 129–151.

Kaibel, Georg. 1887–1890. Athenaei Naucratitae deipnosophistarum libri xv, 3 vols., Leipzig: Teubner (repr. 1–2:1965; 3:1966).

Kasser, Rodolphe. 1991. "Alphabets, Coptic." In The Coptic encyclopedia, edited by Aziz S. Atyia. vol. 8: 33–41. New York: Macmillan.

Mavroudi, Maria. 2014. "Greek Language and Education under Early Islam." In Islamic Cultures, Islamic Contexts: Essays in Honor of Professor Patricia Crone, edited by Behnam Sadeghi, Asad Q. Ahmed, Adam Silverstein and Robert Hoyland. 295–342. Leiden: Brill.

Monferrer-Sala, Juan Pedro. 2015. "Dos tablas inéditas con alfabeto copto y cifras coptas cursivas insertas en un códice del s. XIII del Monasterio de Santa Catalina, Monte Sinaí." Collectanea Christiana Orientalia 12: 279–286.

Olszowy-Schlanger, Judith. 2003. "Learning to read and write in medieval Egypt: children's exercise books from the Cairo Geniza." Journal of Semitic Studies 48 (1): 47–69.

---. 2014. "On the Hebrew script of the Greek–Hebrew palimpsests from the Cairo Genizah." In The Jewish-Greek Tradition in Antiquity and the Byzantine Empire, edited by James K. Aitken and James Carleton Paget. 279–299. Cambridge: Cambridge University Press.

Papaconstantinou, Arietta. 2007. "“They shall speak the Arabic language and take pride in it”: reconsidering the fate of Coptic after the Arab conquest." Le Muséon 120: 273–299.

Parker, Kenneth S. 2013. "Coptic Language and Identity in Ayyūbid Egypt." Al-Masaq 25 (2): 222–239.

Peppink, Simon Peter. 1939. Athenaei dipnosophistarum epitome, vol. 2.2. Leiden: Brill. 

Pernigotti, Sergio. 1985. "I papiri copti dell'Università Gattolica di Milano: I." Aegyptus 65 (1/2): 67–105.

Prince, J. Dyneley. 1902. "The modern pronunciation of Coptic in the Mass." Journal of the American Oriental Society 23: 289–306.

Radermacher, Ludwig and Herman Usener. 1929. Dionysii Halicarnasei quae exstant, vol. 6, Leipzig: Teubner (repr. 1965).

Richter, Tonio Sebastian. 2009. "Greek, Coptic, and the ‘Language of the Hijra’. Rise and Decline of the Coptic Language in Late Antique and Medieval Egypt." In From Hellenism to Islam: Cultural and Linguistic Change in the Roman Near East, edited by Hannah M. Cotton, Robert G. Hoyland, Jonathan J. Price and David J. Wasserstein. 402–447. Cambridge: Cambridge University Press.

Schenke, Gesa. 2010. "Zwei Schultexte der Bodleian Library in Oxford: Das Gebet eines frommen Sünders, ein Psalmenvers und Fragen zur Reihenfolge des koptischen Alphabets." Archiv für Papyrusforschung und verwandte Gebiete 56 (2): 290–293.

Sidarus, Adel Y. 2005. "Medieval Coptic Grammars in Arabic: The Coptic Muqaddimat." Journal of Coptic Studies 3: 63–79.

Sijpesteijn, Petra M. 2016. "Multilingual archives and documents in post-conquest Egypt." In The multilingual experience in Egypt, from the Ptolemies to the Abbasids, edited by Arietta Papaconstantinou. 115–136. ???: Routledge.

Smith, Joseph A. 2003. "Clearing up Some Confusion in Callias' Alphabet Tragedy." Classical Philology 98 (4): 313–329.

Stern, Ludwig. 1880. Koptische Grammatik. Leipzig: Weigel.

van der Valk, Marchinus. 1976. Eustathii archiepiscopi Thessalonicensis commentarii ad Homeri Iliadem pertinentes, vol. 2. Leiden: Brill.





If you enjoyed this Fragment of the Month, you can find others here

Contact us: 

The manuscript in this article is part of the Cairo Genizah Collection in Cambridge University Library. To see more items from this collection visit: