skip to content

Cambridge University Library

Geṭ groundwork from the Cairo Genizah: practising writing a Jewish divorce document

Divorce deed dated 1181 CE, T-S 10J2.34
Marc Michaels
Tue 30 Apr 2024

As a Sofer STa”M (scribe) one of the tasks I undertake is to write giṭṭin – Jewish divorce documents – sadly, usually one or two a month for the last 25 years.

There are many giṭṭin in the Genizah and, as dated documents,1  with mention of the husband and wife and the ʿedim (witnesses) they provide a window into the society, bringing real people to the fore. However, whilst looking through these, one fragment caught my attention as it was not about the individuals, but more about the training for the scribe. 

Having previously written a FOTM about a trope trainer,2  here we have what is described as a ‘practice or model divorce document’ for the scribe to familiarise himself with how the document is written. This is key since any substantive error in this legal document would require it to be written again, since no corrections are permitted. Also, the geṭ would be written at the instruction of the head of the bet din (court) in the presence of the parties and thus had to be prepared swiftly to avoid inconvenience to all present – though obviously not too swiftly that there would be errors or illegible writing. The scribe needed to be familiar with the text.3

T-S 10J2.34 is dated Monday 20th Elul 1492 according to the dating mechanism for a šṭar (formal legal document), and was likely written by a scribe, as I discuss below, as a training exercise.4 His writing is quite cursive and rushed and poorly spaced in many places. It is a single leaf 9.9 cm wide and 11.8 cm high written on kelaf (parchment). There are 12 lines of text written on one side, which is the prescribed amount of lines for a geṭ today. However, this is likely just a coincidence, since this is quite rare in the Cambridge Genizah corpus. There are in fact only six other examples with 12 lines, and the numbers of lines of the rest vary from 9 to a staggering 35 (a geṭ written in the margins of another legal document).5 15–19 lines are typical, but it seems clear that no standard was set or expected in Fusṭāṭ.6

My opinion, that this is not a model, but rather a practice geṭ, is more likely because:

a)    There is no space left for the witnesses and a model would probably have shown where these were to go.

b)    The date given would correspond to the Hebrew Year 4941. Thus, in the Julian calendar this would be 2 September 1181, which was a Wednesday and not a Monday. Since this is a training piece the day does not really matter, though it is odd the scribe would not just have used the day he actually wrote it on.7 That said, the year given is likely accurate.

c)    The scribe has used a particular date and place and river, whereas a model would be likely have these as בכך (on such and such) for dates and as בפלוני (in such and such) for places as per Maimonides’ model text in the Mishne Torah, Hilkhot Gerušin 4:12.8

d)    Our scribe makes an error in spelling (see below) and this would have been corrected in any model.

e)    A model would likely have better spaced between words, and perhaps neater. That said many of the actual giṭṭin in the Genizah are just as poorly written. 

f)    Our scribe does not adhere to all of the traditions that were defined in the Talmud.

A transcription and translation9 is given below, the correction added in { } and the elongated letters underlined:


1. בתרי בשבה דהוא עשרין יומי לירח אלול דשנת

2. אלפא וארבע מאה ותשעין ותרתין שנין ל לשטרות

3. בפסטאט מצרים דעל נילוס נהרא מאתבה אנא פל׳

4. בר פל׳ וכל שום וחניכה דאית לי צביתי ברעות

5. נפשי בדלא אניסנא ותרוכית ופטרית ושבקית

6. יתיכי ליכי אנת פל׳ בת פל׳ ובכל שום וחניכה דאית

7. ליכי דהוית אנתתי מן קדמת דנה וכדו תרוכית

8. יתכי די תהוייין דשאה {רשאה} ושלטאה בנפשיכי למהך

9. להתנסבה לכל גבר די תצוייין ואנש לא ימחה

10.       בידיכי מן שמי מיומה דנן ולעלם והרי את

11.       מתרת לכל אדם ודן די יהוי בידיכי מני

12.       ספר תרוכין וגט פתורין ואגרת שבוקין


1.    On the second day of the week [i.e. Monday], which is the twentieth day of the month of ʾElul of the year

2.    one thousand, four hundred and ninety-two years according to the (Era of) Documents, 

3.    in Fusṭāṭ, Egypt, which is situated on the river Nile [Nilus],10 I , ‘so and so’ 

4.    son of ‘so and so’, and all the names and nicknames that I have, do desire of my own

5.    free will, without coercion,11 to divorce, to release, to let go12  

6.    you, ‘so and so’ daughter of ‘so and so’, and all the names and nicknames that you

7.    have, who have until now been my wife. Now that I divorce 

8.    you, you have the licence and power over yourself 

9.    to be married to any man that you wish, and no-one will interfere 

10.  on my account, from this day forth and forever, and behold you are

11.  [hereby] permitted to any man, and this will serve for you as 

12.  a document of divorce, and a bill of release and a letter of leaving.

The text of modern giṭṭin is slightly different to this. Indeed, there are slight differences between different countries and synagogue movements. However, none are significantly different to elements that we see discussed by the rabbis in various places in the Mishna and Gemara for Masekhet Giṭṭin.

Also, given the time and location of our fragment, we are indeed fortunate to have Maimonides’ model text in his Mishne Torah to compare our FOTM against. Making it all the more relevant, this key halakhic work was completed a mere year earlier than the date of our geṭ also in Fustat.13

On first glance, we might assume that there is an error on the second word, which in Maimonides’ model and modern giṭṭin is given as בשבת, the days of the week that follow being counted from then. However, our scribe has written בשבה with a he, and this is the case in the vast majority of other actual giṭṭin in the selected corpus. From the period, we also see בשבא with an ʾalef on others, though less common. In the whole corpus we see שבתא three times and only twice with בשבת. בשבה with a he or an ʾalef means ‘week’ as well as ‘Shabbat’.14

Modern giṭṭin use Hebrew numbers for the dates, but here we see all the numbers are in Aramaic. Maimonides offers a choice for the scribe, that the dating is ליצירה או לשטרות according to ‘Creation’ or an administrative dating for ‘legal documents’. This later method is mentioned by Maimonides earlier in Hilkhot Gerušin 1:27 as relating to Alexander the Great.15  Our scribe has mentioned that this is לשטרות and has, accordingly, employed that method of dating since, from the corpus examined, this was clearly the standard. Modern giṭṭin only ever use the date from Creation.  

As well as the name of the place, it is traditional, that as well as the city or town name, to include reference to the nearest river. For example, when I write giṭṭin, in London, I generally use the phrase על נהר טאמס דמתקריא טאמישיש  (‘on the river Thames that is also known as Thamesis’).16 Similarly the Talmud speaks of one example בִּשְׁוִירֵי מָתָא דְּעַל רָכִיס נַהֲרָא (‘in Sheviri the city, which is on the Rakhis river’)17 to ensure that if there were two such cities then it is the right one. Here our scribe has included the river נילוס (Nilus), the Nile.18 Interestingly Maimonides’ model does not include any reference to having to make note of the nearest river.

Since this is not a real geṭ, our scribe has added פל׳ בר פל׳, shorthand for ploni bar ploni. This device to mark ‘so and so’ hails from the book of Ruth where Boaz’s potential rival for Ruth’s hand is so unimportant to the plot that he does not even receive a real name, just פְּלֹנִ֣י אַלְמֹנִ֑י.19 Later, our scribe also brings plonit bat ploni relating to the wife. 

It is vitally important to ensure that the date, location and the names of the individuals on the geṭ are correct as if they are not, then the divorce is invalid. These together with certain phrases (see below) are substantive elements and no errors are permitted. Indeed, the geṭ must be written specifically for the individuals party to the divorce, particularly the woman, and with them in mind.20 Much halakhic ink has been spent on discussions over doubts and errors and the potential impact. The modern practice is to write all of the names that a person may be known by; the Hebrew/Yiddish names of the husband and wife (with their fathers’ – and for some movements, mothers’ – names) and, given that most people do not use their Hebrew names, it is also customary to add their English names in phonetic Hebrew and any nicknames by which they are generally referred to.21 Seder ha-Geṭ 19:3 explains ... יש לכתוב תחלה שם העיקר ואח׳׳כ שם הטפל (... one writes first the šem ha-ʿiqqar – ‘root name’ – and then šem [or šemot] ha-ṭofel – ‘the secondary name[s]’). This inclusion of all the names is an Ashkenazi tradition, in line with the opinion of the Rama in ʾEven ha-ʿEzer 129:1. 

However, both Maimonides and our scribe preserve the older tradition from the Talmud where Rabban Gamliel the Elder stated that you do not need to record all of the other names and instead employ the phrases וְכׇל שׁוּם שֶׁיֵּשׁ לוֹ or וְכׇל שׁוּם שֶׁיֵּשׁ לָה (and all names that he/she has).22

Indeed, Maimonides states that this also applies to other names by which the father or place are known, and there are many examples in the Genizah where this happens. Our scribe, however, considerably shortens this, referencing only ‘other names and nicknames’ of the person, either the wife or the husband, but these secondary names are not actually written out. This is perhaps surprising as many at that time would have had an Arabic name or kunya as well as their Hebrew name. Likely because this is a trainer, the scribe employed the shorter form rather than invent some more names to include.

Our scribe appears to have actually erred in line 8 confusing a dalet with a reš and writing דשאה instead of רשאה (licence). Also, though not errors, דנה and מיומה are given here with a he, but end with an ʾalef in Rambam’s model and modern giṭṭin. In the corpus, דנה is much more common than דנא. However, our scribe is the only one spelling מיומה with the he, with the most common renditions as מן יומא or מיומא. Similarly מני is spelt מנאי  in Maimonides and modern versions, but the vast majority are מני. We also see מיני quite often and on two occasions מינאי, though never as Maimonides. Thus, looking at other giṭṭin of the period from the Genizah, the spellings we see in our practice geṭ are fairly typical – reflecting the mixed nature of Aramaic orthography at this time – though not in line with Rambam’s master. 

The eagle-eyed amongst you may have noticed some letters waw in the image of our FOTM are elongated. Indeed, because of the cursive monoline nature of our scribe’s practice geṭ, these, to a modern reader, now more resemble a nun sofit. This is particularly noticeable with one section on line 6 that could easily be read as וכדן תרן כית because of the poor spacing. His normal nun sofit does have a bent roof so there is a distinction. However, it is mildly ironic since the reason given in the Talmud, and repeated in Maimonides, for lengthening the waw is to prevent them being read as a letter yud

Our scribe has only added this long waw three times, in כדו and תרוכית in line 6 and then in תרוכין in the final line, but has omitted doing this for שבוקין. Other giṭṭin in the corpus add this in פטורין too, and this later became standard, though Rambam does not call for this. Our scribe has ensured that דן is spelt without a yud, and that תהוייין and תצוייין are each spelt with three letters yud. He has also ensured למהך has no yud after the lamed and the he is clearly a he and not a het. However, technically, according to a strict interpretation of Maimonides because of this one omission of an elongated waw, the geṭ would be pasul (invalid),23  though very easy to rectify with a drop of ink. In a modern geṭ, in the Maimonides model and mentioned in the Talmud, these scribal idiosyncrasies must be present, and they are each furnished with a reason.24

It is particularly striking that in all otherwise Aramaic giṭṭin, one phrase is always given in Hebrew והרי את מתרת לכל אדם. This is explained in Mishna Giṭṭin 9:3, גּוּפוֹ שֶׁל גֵּט הֲרֵי אַתְּ מֻתֶּרֶת לְכָל אָדָם (the essential element [lit. its body] of a bill of divorce is: ‘You are hereby permitted to marry any man’). No doubt its central importance dictated that it was included in Hebrew.

This Mishna continues רַבִּי יְהוּדָה אוֹמֵר וְדֵין דְּיֶהֱוֵי לִיכִי מִנַּאי סֵפֶר תֵּרוּכִין וְאִגֶּרֶת שִׁבּוּקִין וְגֵט פִּטּוּרִין לִמְהָךְ לְהִתְנְסָבָא לְכָל גְּבַר דְּתִצְבַּיִן (‘Rabbi Yehuda says [also essential is] sentence: “And this that you shall have from me is a scroll of divorce, and a letter of leave, and a bill of dismissal to go to marry any man that you wish”’). 

The instruction to give a geṭ is midʾorayta (from the Torah) and not a rabbinical ordinance. It derives from Deut. 24:1 which states that if a man wishes to divorce his wife then וְכָ֨תַב לָ֜הּ סֵ֤פֶר כְּרִיתֻת֙ וְנָתַ֣ן בְּיָדָ֔הּ וְשִׁלְּחָ֖הּ מִבֵּיתֽוֹ (‘he writes her a bill of divorcement [lit. document of cutting] and gives it into her hand and sends her from his house’). ספר כריתת is the difficult phrase here and the extant Targumim all translate this differently and in a way that are strikingly familiar given the above Mishna and the text of the modern geṭ. Onqelos translates as גַט פִיטוּרִין (‘a bill of release’). Targum Neofiti, however, renders אגרת שׁיבוקין (‘a letter of leaving’) and finally Targum Pseudo-Jonathan refers to this as ספר תירוכין (‘a document of divorce’). That the Mishna refers to all three suggests that for this verse at least, all three of these targumic variants were in circulation at the time.25

Given the order presented in the Mishna, we might expect our practice geṭ and Maimonides’ model to finish in this way,26  except that they do not. The order is different, with the last two elements swopped over, reading instead וגט פתורין ואגרת שבוקין. Indeed, examining the Cambridge corpus, this order is the standard. Only three – T-S 10J2.8, T-S 13J6.28 and T-S 10J3.2 (a very partial fragment) – do have the order of the Mishna and modern texts. Conversely three others, T-S 10J2.37, T-S 10J3.11 and T-S NS J282, jumble the order further, bringing instead גט פתורין וספר תירוכין ואגרת שבוקין.


detail from MS T-S 13J6.28

Detail from T-S 13J6.28, the neatest written divorce deed that most resembles modern examples27



detail from MS T-S 10J2.37.1

Detail from T-S 10J2.37 showing the ‘jumbled’ order28


It is clear that, despite the Mishna’s direct instruction, variants circulated. This is even in evidence in what is held to be the oldest extant geṭ – Mur. 19 – found in Wadi Murabaʿat29 and dated (by Benoit, Milik and De Vaux) to the Roman period, 111 CE. It provides the most ancient text of an actual geṭ that we have. The writing was written in an extreme cursive by an unskilled scribe and is hard to decipher with its frequent allographs for the same letter and the many holes in the papyrus fragment.


MS Mur 19

A detail from Mur. 19 including lines 7 and 8


Both Milik and Yardeni have made attempts at decipherment, though with some elements of disagreement. But the section we are interested in is clearer. In lines 7–8 we have spr trkyn wgṭ šbqyn and similarly in 20–21, s[p]r trkyn wgṭ šbq[yn],30  i.e. in this ancient geṭ we have ספר תרכין וגט שבקין, with a jumbling of the Targumic renditions.31

Whilst our Genizah geṭ is 12 lines, one large difference between its layout and the modern interpretation (see below) is the spacing between lines, the lengthening of the ascender of the lamed, and the creating of an elongated letter nun. This is designed to ensure that no one can insert any extra text in between the lines, in what is a legal document. When first learning this, I always wondered why they insisted on expanded line spacing when normal line spacing would have achieved the same and obviated the need for the long lamed and nun. The giṭṭin from the Genizah clearly ignore this expanded spacing.


An example of a modern get

An example of a modern divorce deed that I wrote a few years ago (with the names digitally removed)32 


Additionally, modern giṭṭin end with the phrase כדת משה וישראל (according to the law of Moses and Israel) stretched across one full line of the 12 permitted. Neither our scribe nor Maimonides seem to have any use for that, as it was clearly obvious and possibly redundant. This is therefore excluded from our FOTM. However, this absence is, it seems, also rare, since where extant, this phrase is only absent in a further four giṭṭin; the norm in the Genizah corpus I have examined was to include it.33  One geṭ, rather interestingly, does provide a slightly different ending, T-S 10J2.3 instead includes כדת משה ויהודאיים (according to the law of Moses and the Jews).34


detail from T-S 10J2.3

A detail from T-S 10J2.3


It would thus appear then that the text of giṭṭin was not as fixed as one might be led to believe, though our anonymous scribe was largely in step with the practice of the period.

One hopes that ‘practice made perfect’ for him, and that he was able to serve the community in this important task of writing giṭṭin – but, since ‘the altar sheds a tear’ when a marriage ends,35 hopefully not too many. 

Marc Michaels

Marc Michaels is a scribe and a PhD candidate at the Faculty of Asian and Middle Eastern Studies at the University of Cambridge.

This article first appeared as the Fragment of the Month for April 2024. If you enjoyed reading it, other Fragments of the Month can be found here.


1 As dated documents they could also provide a useful tool for palaeographers, just as a colophon does. The same would apply to ketubbot (marriage documents). A search on CUDL for geṭ yields some 104 results in the Cambridge part of the Genizah. 85 of these are complete or partial giṭṭin, dated between 1024 and 1399, most of which are from the 1100s. Four of which are not Rabbanite (three are identified specifically as Karaite) and so present a very different text. I have used the remaining 81 to define what is more common in the period, though not all elements are extant in each fragment. 

2 See Michaels, Marc, FOTM, T-S Misc.22.264 – a trope trainer from the Cairo Genizah (http://

3 The halakha surrounding giṭṭin is extensive because of the dangers that would ensue if it were invalid, and the parties were not properly divorced – with the potential impact on offspring of any remarriage of the woman. Deriving from an entire complete tractate of the Talmud devoted to the subject, Giṭṭin, the basic halakha lemaʿase (practical application of law) is defined in the Shulḥan ʿArukh ʾEven ha-ʿEzer Hilkhot Giṭṭin sections 119–154, and in particular Seder ha-Geṭ which lays out in order the proceedings from start to finish. However, Seder ha-Geṭ itself is expanded on and brought more up to date (for example dealing with elements of the proceedings via the postal system) in the ʿArukh ha Shulḥan version of Seder ha-Geṭ, which includes other stringencies not in the basic text, and also of lenience given modern-day practice. The Qav Naqi (1868) is a compilation of the laws of Giṭṭin, considered a valuable reference source for batey din. Some of this is discussed in English in Klein, A Guide to Jewish Religious Practice, JTS, 1969, pp 476–508, and also in the Jewish Catalogue.

4 In the Cambridge corpus there is one other that employs Ploni as opposed to actual names, though is undated, T-S NS 292.14 – see

5 T-S 28.6 – see

6 This is significant as one of the popular explanations for the name גט is that it has the numerical value of ‘12’. Another popular explanation is that these two letters are never found together in Tanakh. Maimonides makes no reference to a set number of lines.

7 My thanks to Ben Outhwaite and Nadia Vidro for their calendrical expertise to uncover this.

8 For an excellent edition with comprehensive notes, see Touger, Eliezer, Maimonides Mishneh Torah Hilchot Gerushin (Laws of Divorce), Moznaim, 5755. Rambam’s model is discussed on pp.66-69.

9 My thanks to Ben Outhwaite, particularly for his constructive comments on the translation and earlier drafts of this FOTM.

10 In other giṭṭin in the Cambridge corpus, we usually see the word spelt מותבא.

11 i.e. of sound mind with no constraint or force applied. A geṭ must be freely given by the man.

12 Technically, all of the words could be translated as ‘to divorce’. תרך has a more active sense of movement, being sent away or dismissed, so ‘banish’ could also be a valid translation here. However, as Ben Outhwaite notes, ‘[divorce] was already its specialised meaning by the time of Wadi Murabaʿat, and throughout the Babylonian Talmud, rather than more euphemistically as ‘banish’ or ‘release’’. Sokolof, Michael, A Dictionary of Judean Aramaic, Bar Ilan University Press, 2003, p. 87, gives ‘to divorce’, noting also its JPA meaning ‘to chase away’. פטר has a sense of openness, to be exempted from, to be freed. שבק has more of a passive sense of leaving behind, abandoning, or forsaking. See Jastrow also for all three roots.

13 The Mishne Torah was compiled between 1170 and 1180 CE. It is difficult to say when Gerušin was completed and whether it was circulated independently earlier than our practice geṭ.

14 See CAL. For the spelling a he vs ʾalef, see Friedman, M.A., Jewish Marriage in Palestine: A Cairo Geniza Study, The Ketubba Traditions of Eretz Israel, Vol. 1, JTSA, 1980, pp. 54–55, where he notes that ‘Kutsher considered use of he a definite sign of a reliable GA [Galilean Aramaic] text not tampered with by Babylonian copyists’ and that among the ketubbot in the Genizah he examined he predominates, though some also have ʾalef. Whilst Friedman’s research concerns ketubbot, there is much overlap of vocabulary with giṭṭin, as covered in chapter 1 (pp. 48≠47) of Vol. 1 on the Aramaic employed. There are clearly mixed Babylonian and Galilean influences in the Egyptian legal documents.

15 Mishna Giṭṭin 8:5 notes other possibilities for dating, such as the era of the Medes, the Greeks, after the building of, or destruction of the Temple, but not all are acceptable. The dating from the Seleucid Era, ‘the Era of Documents’, comes from Babylon into Egypt since ‘Egypt had never been part of the Seleucid empire and its dating systems had always been quite different’. The ‘introduction of the Seleucid era in Fusṭāṭ … is visible in the Genizah from the second half of the 10th Century onwards (though may have been earlier)’, Krakowski, Eve & Stern, Sacha, The ‘oldest dated document of the Cairo Genizah' (Halper 331): The Seleucid era and sectarian Jewish calendars, JRAS Series 3, Cambridge University Press, 2021, p. 629 and p. 630.

16 This more ancient name for the river is likely proto-Celtic, tamēssa, possibly meaning ‘dark’, likely referring to the dark waters – which, thanks to Thames Water, have now returned with a vengeance. The Middle English spelling was typically Temese and the Brittonic form Tamesis with alternative spellings of Tamesa or Thamesis.

17 Babylonian Talmud Giṭṭin 27a.

18 As Outhwaite notes, ‘the Nile, as the major feature of Egypt, is almost always mentioned. Only when the city/town is on the coast, then ‘the Great Sea’ or ‘the Salty Sea’ or something is used instead … and it is widespread in legal documents of all types in the Genizah’ (personal correspondence 21/9/23). See Vol. 2 of Friedman, M.A., op. cit., which covers the ketubba texts. Indeed, amongst other terms, it is referred to in one ketubba in Hebrew as שעל נהר מצרים (‘which is on the river of Egypt’) – my emphasis, see p. 171.

19 Ruth 4:1. But also see 1 Sam. 21:3 and 2 Kings 6:8 where this refers instead to a place name. 

20 See Mishna Giṭṭin 3:2.

21 This potentially caused me some difficulty in a geṭ involving a transgendered individual who transitioned between the civil and religious divorce and theoretically would have five names. This is covered in detail in my article in Review of Rabbinic Judaism (forthcoming).

22 See Mishna Giṭṭin 4:2.

23 See Hilkhot Gerušin 4:14.

24 See Hilkhot Gerušin 4:13 and Babylonian Talmud Giṭṭin 85b for discussion of those reasons.

25 Whilst Targum Onqelos is generally accepted to be early and Targum Neofiti similarly, Targum Pseudo-Jonathan is always consigned to a much later date because of later layers – but some of the elements in this Targum such as this were also early. It is also quite usual for Targum Pseudo-Jonathan to be in accord with Targum Onqelos but this is not the case here. 

26 This is also the order presented in the earlier halakhic summary, Halakhot Gedolot, which general scholarly opinion considers was written by Simeon Kayyara at the end of the ninth or the beginning of the 10th century (though there is much debate over authorship). See and (both accessed 24/9/23).

27 See

28 See

29 See the papyrus manuscript at

30 See Milik, DJD 2, 1961, pp.104-109 and Yardeni, Textbook of Aramaic, Hebrew and Nabatean Documentary Texts from the Judaean Desert, 2000, pp.131–133.

31 Yardeni notes that ‘in any, the formulation of these clauses in the deeds of marriage and of divorce found in the Judaean desert differs from their formulation in later literary and documentary sources’, ibid.

32 One is obliged to ensure that you differentiate the script from the ketav Ašurit with tagin that is used in kitvey ha-qodeš (holy writings), thus this is a simpler less ornate version of my normal script.

33 This is also present in the giṭṭin master text brought in Halakhot Gedolot.

34 See

35 See Babylonian Talmud Giṭṭin 90b, though technically this only applies to his first wife in context.

Add new comment



Share this post