The Lost Model Codices of the Masorah

Medieval manuscripts of the Hebrew Bible often name other manuscripts in their margins. A note beside a word may state that a particular vowel or a particular spelling is found “in the Hilleli”, “in the Zambuki” or “in the Muggah”, and in doing so it appeals to a copy of the Bible that was regarded as an authority by the scribes and Masoretes who wrote the note. These authoritative copies are known in modern scholarship as model codices or standard codices. None of them survives. They are known only through the citations that later manuscripts, grammarians and commentators made of their readings, and through a small number of historical reports about their fate.¹

This article describes what can be learned about these lost books. It begins with the earliest reports of standard copies kept in the Temple court, then considers the principal model codices cited in the Masorah one by one, and finally examines a modern study of five Spanish manuscripts that cite them. Throughout, the evidence is indirect, and the identity, date and even the existence of some of these codices as single physical books remain matters on which scholars have disagreed. The account given here follows the sources closely and indicates where they conflict.

Standard Copies in the Temple Court

The practice of correcting copies of Scripture against a recognised exemplar is older than the Masoretes. Christian David Ginsburg, whose Introduction to the Massoretico-Critical Edition of the Hebrew Bible appeared in 1897, began his discussion of the subject with a passage from the Jerusalem Talmud, tractate Ta’anit, which reports that three copies of the Torah were kept in the court of the Second Temple. The passage names them as the codex Me’on, the codex Za’atute and the codex Hi, and it records three places in which one of the three differed from the other two. In each case the reading of the two was accepted and the reading of the single copy was rejected.²³ The same three copies are named in the article on Bible manuscripts in The Jewish Encyclopedia, which treats them as the earliest recorded examples of authoritative exemplars.⁴

Ginsburg drew two conclusions from the Talmudic report. The first was that the redactors of the text at that period collated manuscripts and followed the majority reading. The second was that they did not destroy the minority reading, so that its existence was remembered and could still be discussed. He also observed that the Temple copies were not the only manuscripts of standing in antiquity, and he cited the tradition, transmitted by Josephus and by later Jewish sources, that a copy of the Law was carried from Jerusalem to Rome after the capture of the city, and that a list of its variant readings was preserved in later manuscripts.⁵ Whatever the historical value of each of these reports, they show that the idea of a standard copy against which others could be checked was deeply rooted in Jewish tradition before the period of the Masoretes.

The medieval model codices belong to a later and different world. They were not kept in a single sanctuary, and they did not owe their authority to an institution. Their authority came from reputation, from the learning of the communities that owned them and from the accuracy that later scribes believed them to possess. Ginsburg described them as the standard manuscripts “in the possession of the different communities”, distinguished by special names because of their excellence, and he noted that the Masoretes cited them both in support of a reading they had adopted and in order to record an alternative reading.⁶

How the Masorah Cites a Model Codex

The citations occur in the Masorah, the body of notes written around the biblical text in medieval manuscripts. The short notes in the side margins and between the columns are known as the Masorah parva, and the longer notes in the upper and lower margins as the Masorah magna.⁷ A typical reference to a model codex is very brief. It may say only that a word is written with a particular vowel “in the Hilleli”, or that it is spelled in full “in the Muggah”, without further explanation. The reader was expected to know which codex was meant and why its testimony mattered.

These notes formed part of the work that produced the text now called the Masoretic Text, and they concern the same details that occupied the Masoretes in general: the spelling of words with or without vowel letters, the vowel signs, the accents and the divisions of the text.⁸ Because the model codices themselves are lost, the notes cannot be checked against their originals. A modern reader therefore depends on the accuracy of each copyist who transmitted a note, and on the assumption that the same name always refers to the same book. Ortega Monasterio, whose study is described below, emphasised that the model codices “have not come down to us” and are known only through the references in which they appear.⁹

The notes also vary in form from manuscript to manuscript. Some copyists recorded the citations in a full and explanatory style, while others reduced them to a few words in the manner of the ordinary Masorah parva. In certain manuscripts the citations were added by a second hand after the text was complete, which suggests that a later reader or owner compared the manuscript with a copy of a model codex, or with a list of its readings, and entered the results. These differences in presentation are important for understanding how knowledge of the model codices was transmitted, and they are discussed further in the section on the Spanish manuscripts.

The Codex Muggah

The name Muggah is a Hebrew word meaning “corrected”, and a sefer muggah is a corrected book. Ginsburg regarded the Codex Muggah as the earliest model codex quoted by the Masoretes as far as he could trace the evidence. He gave as his first example a rubric of the Masorah in the British Museum manuscript Oriental 4445 on Exodus 39:33–43, a passage in which the object marker is repeated many times and is sometimes preceded by the conjunction and sometimes not. The rubric lists the pattern verse by verse and ends with the statement that this is the order according to the Codex Muggah.¹⁰ Ginsburg explained the purpose of such a rubric as a protest against variants in other recensions, for manuscripts and ancient versions differ in this passage in almost every verse.

He also found the Muggah cited eight times in the Masorah of the Saint Petersburg codex of the Latter Prophets, dated 916, which he regarded as the next oldest Masoretic witness. The passages are Jeremiah 6:10 and 51:46, Hosea 1:7, 2:21 and 11:9, Joel 1:12, Amos 5:2 and Habakkuk 1:5. With the exception of Jeremiah 51:46, he observed, the codex of 916 cites the Muggah in support of the reading of the received text.¹¹ This manuscript, sometimes called the Codex Babylonicus Petropolitanus, is one of the oldest dated codices of any part of the Hebrew Bible, and its citations of the Muggah are therefore among the earliest dated references to a model codex of any kind.¹²

According to Ginsburg, the Muggah was afterwards cited as an authority in almost every important manuscript, either by its full title or simply as “the Muggah”. In a manuscript in Cambridge University Library, Add. 465, he counted several hundred references to it. He printed in the third volume of his Massorah a list of variations between the Muggah and the Hilleli covering the whole Bible, which he had found in a manuscript at Munich, and he concluded that the Muggah had been copied by the heads of schools in different communities and different ages, since later manuscripts sometimes distinguish between the earlier and the later copies of it.¹³

This interpretation has not been universally accepted. Ortega Monasterio quoted the view of Aron Dotan that “the term Sefer Muggah does not apparently refer to a specific book”, that is, that a note citing the Muggah may refer to any corrected copy and not to a single famous codex.¹⁴ Her own study of a fifteenth-century Bible at El Escorial produced evidence that supports Dotan’s caution. In a note on Numbers 21:24 the Masorah of that manuscript states that a certain word is pointed with qamets “in the Sefer Muggah”, but that “in another Sefer Muggah” the writer had found it with patah. The phrase “another Sefer Muggah” makes sense only if the term could be applied to more than one book.¹⁵ The same manuscript records at Exodus 37:24 that a word is pointed with patah in the Sefer Muggah, whereas the Masorah published by Ginsburg records the opposite reading for the Muggah at the same place. Such contradictions may result from errors of transmission, but they may also show that different copyists had different corrected books in mind.¹⁶

The two positions are not entirely incompatible. It is possible that the term began as the name of a particular famous codex and was later used more loosely for any copy that had been corrected against a good exemplar, or that it was a general designation from the beginning and only came to be treated as a proper name by later readers. The evidence currently available does not allow a firm decision, and the citations of the Muggah are best treated as references to a corrected text whose identity may have varied from one note to another.

The Codex Hilleli

The most famous of the model codices is the Hilleli. Its history is known from a report in the chronicle of Abraham Zacuto, the Spanish Jewish astronomer and historian, who compiled his work about 1500.¹⁷ Ginsburg translated the passage as follows. In the year 4957 of the Jewish era, on the twenty-eighth of Av, corresponding to 14 August 1197, there was a great persecution of the Jews in the Kingdom of León “from the two Kingdoms that came to besiege it”. At that time the twenty-four books of the Bible were removed from there. They had been written about six hundred years earlier by Rabbi Hillel ben Moses ben Hillel, and were therefore called the Hilleli; the codex was “exceedingly correct and all other Codices were revised by it”. Zacuto added that he had seen the two remaining parts, containing the Former and Latter Prophets, which had been brought by the exiles to Portugal and sold at Bugia in Africa, the port now called Béjaïa, where they still were.¹⁸

Zacuto’s statement that the codex was written about six hundred years before 1197 implies a date around the end of the sixth century. Modern scholars have not accepted this. David Stern, in an essay on the Hebrew Bible in medieval Spain written for the Biblioteca Nacional de España, describes the Hilleli as reputed to have been written around 600 but more probably written around 1000 in León.¹⁹ The earlier date would place the codex before the development of the Tiberian system of vowel signs and accents, which is difficult to reconcile with the nature of the readings attributed to it, many of which concern exactly those signs.²⁰ Stern’s account also differs from Zacuto’s in its description of the event of 1197. Where Zacuto, as translated by Ginsburg, refers to a persecution in León caused by two besieging kingdoms, Stern writes that the original was still in existence in 1197 when the Almohads attacked the Jewish communities of Castile and Aragon and carried away at least part of it.²¹ The two accounts agree that the codex was lost or broken up at the end of the twelfth century, but they describe the circumstances differently, and the historical details of the event cannot be settled from the sources used here.

Other medieval writers knew the Hilleli or part of it. Ginsburg cited David Kimhi, who stated in his grammar, in a comment on Numbers 15:4, that the Pentateuch of the Hilleli was extant in Toledo.²²²³ Ortega Monasterio quoted a note of Jedidiah Solomon Norzi, the author of the Masoretic commentary Minhat Shai, on a word in Deuteronomy 12:11. Norzi wrote that in correct codices the letter dalet in this word has a dagesh, but that he had seen in the margins of Spanish manuscripts that in the Hilleli it is rafeh, and that the Mikhlol, citing Jacob ben Eleazar, says the same of “the Sefer Hilleli that was in Toledo”.²⁴ If Kimhi’s and Zacuto’s statements are combined, the Pentateuch of the Hilleli seems to have been in Toledo in the early thirteenth century, while the Prophets were later carried by exiles to Portugal and sold in North Africa. The fate of the Writings is not recorded.

The readings attributed to the Hilleli were collected in lists. Ginsburg printed a list of plene and defective spellings throughout the Pentateuch taken from what he called the Merzbacher manuscript, and he noted that Jacob Saphir had printed a similar list in the second volume of his travel account Eben Sappir, published at Mainz in 1874.²⁵ Ortega Monasterio states that Saphir’s list came from a quire bound into an eleventh-century Bible in the Paris library.²⁶ Ginsburg also found two places, in 1 Kings 1:18 and Jeremiah 5:6 in British Museum Add. 15251, where the codex is called “the Hilleli of León”, a designation that agrees with Zacuto’s location of it.²⁷

A Toledo Copy of 1241

One manuscript has been described as a direct copy of the Hilleli. It is a Pentateuch now in the library of the Jewish Theological Seminary of America in New York, MS L44a. An online exhibition of the Grolier Club describes it as copied in Toledo in 1241 by the scribe Israel ben Isaac ben Israel from the Codex Hilleli, and it illustrates the manuscript open at the Song of the Sea, written in the traditional brickwork layout of alternating blocks of text and space.²⁸ Stern likewise states that a copy of the Pentateuch section of the Hilleli was completed in Toledo in 1241 and survives as JTS MS L 44a. He adds that the Hilleli was notable for recording the decorative crownlets known as tagin and certain peculiarly shaped letters, the otiyyot meshunot, and he takes this as evidence of how carefully the codex was consulted and copied.²⁹

The identification has not been accepted by every scholar. Ortega Monasterio notes that the Hilleli has often been identified with the New York manuscript, but that Moshe Goshen-Gottstein, writing on biblical manuscripts in the United States in the journal Textus in 1962, held that the two should not be identified.³⁰ The disagreement may turn partly on the meaning of identification. A manuscript can be copied from a model codex without reproducing all its features, and a later copyist may have worked from an intermediate copy rather than from the original. The New York Pentateuch is nevertheless the closest physical link with the Hilleli that has been proposed, and it gives a sense of the kind of large, carefully written Spanish codex that the Hilleli may have been.

The Zambuki and the Jerusalem Codex

Ginsburg described the Zambuki as another standard codex often cited in the Masorah parva. He suggested that it took its name from the community of Zambuki on the Tigris, which would make it an Eastern codex, although he presented this as a probable explanation rather than a certainty. Its readings, he observed, are frequently cited side by side with those of the Hilleli, especially in the manuscript Oriental 2626–2628 in the British Museum, and like the other standard codices it is known only through quotations.³¹ The Jewish Encyclopedia adds that the Zambuki was praised by Menahem de Lonzano, the author of the Masoretic treatise Or Torah.³²

The Yerushalmi, or Jerusalem Codex, was another lost standard manuscript frequently quoted in the Masorah. According to Ginsburg, it was used extensively by the grammarian and lexicographer Jonah ibn Janah, whom Kimhi described as having constantly cited it as his authority. Kimhi also stated that it had been for many years in Saragossa, the city now called Zaragoza. In the Masorah the Jerusalem Codex is often quoted as showing a different orthography from that of the Hilleli.³³³⁴ The name may indicate an origin in Jerusalem or an association with a Jerusalem tradition, but the sources used here do not establish where the codex was written.

The appearance of an Eastern codex and a codex associated with Jerusalem in the Masorah of Spanish manuscripts shows that the authorities cited by Spanish scribes were not exclusively Spanish. The Masoretic tradition of Spain drew on books and lists that had travelled from the East, and Spanish scholars such as Ibn Janah consulted manuscripts that were themselves regarded as heirs of Eastern traditions. In this respect the model codices are evidence for the movement of texts and learning across the Mediterranean world in the tenth to twelfth centuries.

Jericho, Sinai and the Great Mahzor

Several other model codices are known by name. Ginsburg described the Codex Jericho as having apparently contained only the Pentateuch, since it is sometimes called the Jericho Pentateuch, and he printed in his Massorah a list of its readings collected from the Masorah parva of the British Museum manuscript Oriental 2696.³⁵ The Jewish Encyclopedia notes that Elia Levita cited the Jericho Pentateuch in his work Shibre Luhot.³⁶³⁷

The Codex Sinai is cited in the Masorah parva of British Museum Arundel Oriental 16 on six occasions. Ginsburg listed them as Joshua 21:36, where it is quoted to justify the omission of verses 36 and 37, 2 Kings 6:25, 23:31 and 25:11, Jeremiah 39:1 and Amos 5:6.³⁸ In the printed Masorah parva the codex is quoted twice more, both times in Exodus 18 and both times on matters of accentuation, and Levita had concluded from these two printed notes that the Codex Sinai contained only the Pentateuch and concerned itself only with accents. Ginsburg replied that the manuscript citations from Joshua, Kings, Jeremiah and Amos showed beyond doubt that it had contained the whole Bible.³⁹

The name Sinai also gave rise to a different interpretation. Ginsburg recorded that Jacob ben Isaac of Zousmir, in a treatise on the Masorah first printed at Amsterdam in 1649, held that Sinai was the name of one of the redactors of the Pentateuch, and that Joseph Eshwe, a commentator on the Masorah, went further and described Sinai as one of the sages of Tiberias who had disagreed with the Masorah on an accent. Ginsburg rejected these views on the ground that the manuscripts give the full form “Sefer Sinai”, which can only mean the Sinai Codex, just as “Sefer Yerushalmi” means the Jerusalem Codex.⁴⁰ The episode illustrates the uncertainty that surrounded the model codices even among early modern Jewish scholars, who had only the brief notes of the Masorah to work from.

The Great Mahzor was a codex whose name derived from the term mahzor, which Ginsburg explained as originally meaning a cycle of readings and hence a book arranged according to the cycle. He argued that the Great Mahzor exhibited the recension of Ben Naphtali, since the readings attributed to it in the Masorah of British Museum Add. 15251 at Deuteronomy 31:21 and 1 Samuel 22:17 agree with the readings attributed to Ben Naphtali.⁴¹ If this is correct, the Great Mahzor was a witness to the rival of the Ben Asher tradition, which later became standard.

Ginsburg also mentioned a Codex Ezra, cited in the Masorah parva of a manuscript in his own possession that professed to be a copy of it. The notes appealed to the Codex Ezra at Numbers 21:14 and Deuteronomy 32:6, in each case on a question of word division.⁴² The name associates the codex with Ezra the Scribe, but no evidence connects it with the historical figure, and Ginsburg made no such claim. It is best understood as one of the names under which a respected exemplar was known.

The Babylonian Codex and Other Eastern Exemplars

Among the model codices, Ginsburg gave particular attention to the Babylonian Codex, which he regarded as a witness to the Eastern recension of the text. He collected eleven readings from it, nine of them in the Latter Prophets, from several British Museum manuscripts and from a Paris manuscript dated 1286. Examples include Numbers 26:33, where the Paris manuscript states that the Babylonian Codex reads the name Tirzah without the conjunction. The note is written in the condensed mnemonic manner typical of the Masorah, giving the order of the five daughters of Zelophehad by their initial letters, once for the Western readers and once for the Babylonian Codex, and Ginsburg reproduced it as an example of the cryptography that a reader of the Masorah had to decipher.⁴³ Another example is and Isaiah 27:8, where the Masorah parva of British Museum Oriental 2201, a manuscript dated 1246, states that the Babylonian Codex reads “a rough spirit” without the possessive suffix, whereas the codex of 916 reads “his rough spirit” with the received text.⁴⁴

Some of the Babylonian readings concern the division of words rather than their spelling. In 1 Kings 20:33 the Babylonian Codex, according to Ginsburg, divided a group of letters in the way that the received text gives only as its written form, while the Western tradition read them with a different division in the margin. Ginsburg used the British Museum manuscript Oriental 1478 as his evidence for this case and noted that here, unusually, the received text exhibits the Eastern recension in its written form. In Jeremiah 23:18 he observed that the codex of 916 had originally written “his word”, and that its Masorete altered the text to “my word” and placed “his word” in the margin as the reading to be read, thus bringing the manuscript into conformity with the Western recension.⁴⁵ Cases of this kind show how the citations of a lost model codex could be used to trace the process by which surviving manuscripts were adjusted to an accepted standard.

Ginsburg used these readings for a specific argument. The Saint Petersburg codex of 916 had been described as a Babylonian manuscript because of its vowel signs, and it had been assumed to contain the Eastern recension. Ginsburg found that in ten of the eleven readings that the Masorah attributes to the Babylonian Codex, the codex of 916 agrees instead with the Western text, and he concluded that the designation of the Saint Petersburg manuscript as a Babylonian codex was incorrect so far as its textual recension was concerned.⁴⁶ The argument is of interest here because it shows how a lost model codex could still be used, through the citations made from it, as a measure against which surviving manuscripts were tested.

Ginsburg also found references to other Eastern standard manuscripts in the Masorah parva of Add. 15251, which cites a Codex of Baghdad and a Codex Sharki, the latter name meaning eastern. The readings he quoted include the spelling of personal names in Kings, a consonant in Isaiah 5:2, a dagesh in Isaiah 51:10 and the placing of an accent in Ezekiel 4:16.⁴⁷ These codices are even more obscure than the Hilleli or the Zambuki, but their names confirm that Western Masoretes were aware of Eastern authorities and recorded their readings.

The Model Codex of Maimonides

The most widely discussed model codex in modern scholarship is one that is not usually counted among the lost codices of the Masorah, because a manuscript identified with it survives. In his code of Jewish law, the Mishneh Torah, Maimonides explained that, because he had seen great confusion in the scrolls he examined concerning open and closed sections, he had relied in writing his list of them on a book renowned in Egypt that contained the twenty-four books of the Bible. This book had been in Jerusalem for several years and had been used to check scrolls, everyone relied on it, and it had been corrected by Ben Asher, who had worked on it for many years.⁴⁸⁴⁹ The Ben Asher named here is generally understood to be Aaron ben Moses ben Asher, the last and most celebrated member of the Tiberian family of Masoretes.⁵⁰

The Jewish Encyclopedia notes that the Ben Asher codex served as a model for Maimonides, and that the attribution of the codex now known as the Aleppo Codex to Ben Asher rests on an inscription whose authenticity was questioned in the nineteenth century, for example by William Wickes.⁵¹ Modern scholarship has generally identified the codex described by Maimonides with the Aleppo Codex, which is now kept in Jerusalem and of which a large part was lost in the twentieth century.⁵² The case is instructive for the study of the lost model codices because it shows what such a book was: a complete Bible of exceptional accuracy, kept by a community, consulted by scholars and used as a standard for checking other copies. The difference is only that the Aleppo Codex survived, at least in part, while the Hilleli, the Zambuki and the others did not.

The Evidence of Five Spanish Manuscripts

The most detailed modern study of the citations of model codices in particular manuscripts is María Teresa Ortega Monasterio’s article “Los códices modelo y los manuscritos hebreos bíblicos españoles”, published in the journal Sefarad in 2005. She examined the Masorah of the Pentateuch in five Spanish biblical manuscripts and recorded every reference to a model codex. Three of the manuscripts are in Spanish libraries and two in the British Library.⁵³

The first is El Escorial, MS G-II-8, a fifteenth-century Bible that came from the collection of Benito Arias Montano and was kept in the convent of Santiago de la Espada in Seville until it was taken to El Escorial in 1602. It contains the whole Bible, written in a Sefardic square script with illumination in gold and colours, and it carries notes in red ink by the scholar Alfonso de Zamora. Its Masorah cites the Sefer Muggah seven times, five of them for a variant in pointing, and it cites the Hilleli four times in Genesis and once in Exodus.⁵⁴

The second and third are two Bibles in the library of the Complutense University of Madrid, known as M1 and M2. M1 was copied in Toledo in 1280 according to its colophon and served as the basis of the Hebrew text of the Complutensian Polyglot Bible. It contains few references to model codices in its Pentateuch, four in Leviticus and five in Deuteronomy, and all of them were written by a second hand.⁵⁵⁵⁶ M2 was copied at Tarazona in 1482 by Yom Tov ben Isaac Amarillo for his own use. It too carries notes by Alfonso de Zamora and was used for the Hebrew text of the Complutensian Polyglot, and among the model codices it cites only the Muggah, which it names sixteen times.⁵⁷

The fourth manuscript is British Library Or. 2626–2628, a Bible copied in Lisbon in the month of Kislev 1483 according to its colophon. Its Pentateuch contains 108 references to model codices, the largest number in any of the five manuscripts. The Muggah is cited 28 times, the Hilleli 33 times and the Zambuki 55 times. Of the variants, 75 concern vowels, 20 concern accents and 3 concern plene or defective spelling, while ten concern other matters. In six places more than one model codex is cited for the same variant, and at Deuteronomy 30:7 the Hilleli, the Zambuki and the Muggah are all cited together.⁵⁸ This is the same manuscript in which Ginsburg had noted the frequent pairing of the Hilleli and the Zambuki.⁵⁹

The fifth is British Library Or. 2201, copied in Toledo in 1246 according to its colophon and therefore the oldest of the five. Its margins contain only four references to model codices, two in Genesis, one in Numbers and one in Deuteronomy, and all four cite the Hilleli. Three concern vowels, and the fourth records that the Hilleli has no section break at Deuteronomy 31:1, a point that Norzi also discussed.⁶⁰

What the Citations Record

Ortega Monasterio drew several conclusions from her survey. The Spanish manuscripts she studied cite only three named model codices in the Masorah of their Pentateuchs, namely the Hilleli, the Zambuki and the Muggah, in addition to anonymous codices. The notes mostly concern plene and defective spellings and pointing, and very few concern other kinds of variant. In the two oldest manuscripts, M1 and Or. 2201, all the references to model codices were written by a second hand, as were some in M2, whereas in G-II-8 and Or. 2626–2628 they belong to the original Masorah. Finally, the text of a manuscript was sometimes corrected so as to agree with the information in its own Masorah parva.⁶¹

These conclusions suggest a particular history for the knowledge of the model codices in Spain. In the thirteenth century, readings from the model codices appear to have been added to finished manuscripts by later hands, perhaps by owners or scholars who had access to a list of readings. By the late fifteenth century, they had become an integral part of the Masorah written by the original Masorete of a manuscript, as in the Lisbon Bible of 1483. The correction of the text to match the Masorah shows that the readings of the model codices were not merely recorded as curiosities but were treated as authoritative and acted upon.

The emphasis on vowels and accents is also significant. The model codices were valued not chiefly for their consonantal text, which was already highly stable in the medieval period, but for the precise details of the vowel signs, the accents and the spelling of individual words. These were the areas in which manuscripts most often differed and in which a scribe most needed an authority. A model codex in this sense was a reference work for the finest details of the text, and its citations record the questions that medieval scribes found most difficult.

Throughout her article Ortega Monasterio compared the readings cited in her manuscripts with the opinions of Lonzano and Norzi, the two early modern scholars whose works most often discuss the testimony of Spanish manuscripts, and she found them frequently in agreement. Norzi, as she quotes him, appeals on occasion to the reading found “in all the codices of Sefarad”.⁶² Through such works the authority of the lost model codices was carried into the period of printed Bibles.

The Reputation of Spanish Model Books

The prestige of the model codices was part of a wider reputation enjoyed by Spanish biblical manuscripts in the Middle Ages. Stern observes that the masoretic Bible in Spain sometimes served as a sefer muggah or tikkun, a codex written with special care as an exemplar for scribes. He cites the praise of Meir of Rothenburg at the end of the thirteenth century for “the superior and exact books of Sepharad”, and the report of Menahem ben Solomon Meiri of Perpignan, who described a German rabbi travelling to Toledo to acquire a codex of the Pentateuch copied from a Torah scroll written by Meir ha-Levi Abulafia, so that it could serve as a model for writing Torah scrolls in Germany.⁶³⁶⁴

These reports show that the model codex was a living institution as well as a memory. The famous lost codices of the Masorah were the most celebrated examples of a broader practice in which particular books, or copies made from them, were sought out and used to correct other books. A scribe in Germany or Italy who wished to write an accurate Torah scroll might depend on a codex copied in Toledo from a scroll written by a Toledan authority, just as an earlier Masorete in Lisbon depended on readings that came, directly or indirectly, from the Hilleli and the Zambuki.

The Problem of Identification

The evidence for the model codices raises a number of problems that cannot be resolved with the sources currently available. The first is the problem of date. Zacuto’s report places the writing of the Hilleli around 600, but Stern regards this as a reputation rather than a fact and prefers a date around 1000, and no surviving part of the codex is available to decide the question.⁶⁵ For the other codices there are no reports of date at all. The earliest dated citation of a model codex is in the codex of 916, which gives a terminus for the existence of the Muggah, if the Muggah was a single book, but tells nothing about its origin.⁶⁶

The second problem is that of identity. As Dotan’s remark and the evidence of the Escorial manuscript show, some names may not refer to single books at all.⁶⁷ Others may refer to copies rather than originals, as the distinction between earlier and later copies of the Muggah implies.⁶⁸ A note citing the Hilleli in a manuscript of 1483 may therefore be citing the original codex through a chain of intermediate copies or lists, and errors may have entered at any stage. The contradictions between notes in different manuscripts, which Ortega Monasterio documented, are what one would expect if the readings were transmitted in this way.

The third problem concerns the purpose of the notes. Ginsburg interpreted many citations as protests against variant readings in other schools, and Ortega Monasterio showed that they could lead to corrections of the text. But the notes also preserve alternative readings without always deciding between them. In this they resemble the Talmudic report of the three Temple copies, in which the minority reading was rejected but remembered. The Masorah of the model codices records disagreement as well as authority.

These uncertainties do not diminish the importance of the model codices. They show that the medieval Hebrew Bible was not copied in isolation but within a network of reference copies, lists and notes, in which particular books were valued above others and their readings were preserved even after the books themselves had been lost. The study of these citations, carried out by Ginsburg at the end of the nineteenth century and continued by scholars such as Ortega Monasterio, remains one of the ways in which the history of the Masoretic tradition between the tenth and fifteenth centuries can be reconstructed.

Further Study

The fullest account of the model codices in English remains the eleventh chapter of Ginsburg’s Introduction, which is freely available in digitised form and should be read together with his collections in The Massorah.⁶⁹ Ginsburg’s career and the reception of his edition are discussed in the article on Christian David Ginsburg on this site.⁷⁰ Ortega Monasterio’s study is available as an open-access article from the journal Sefarad and contains a complete list of the passages she examined.⁷¹

For the wider setting of the Spanish model codices, Stern’s essay for the Biblioteca Nacional de España provides an introduction to the forms of the Hebrew Bible in medieval Spain.⁷² The relation between model codices and the Masorah as a whole can be followed in the articles on the Masorah parva and Masorah magna, on the Aleppo Codex and the Tiberian standard, and on Norzi’s Minhat Shai, and the Leningrad Codex offers a surviving complete codex of the Ben Asher tradition against which the readings of the lost codices may be compared.⁷³

Notes

1. María Teresa Ortega Monasterio, “Los códices modelo y los manuscritos hebreos bíblicos españoles”, Sefarad 65:2 (2005), pp. 353–383, doi:10.3989/sefarad.2005.v65.i2.502. *
2. Christian D. Ginsburg, Introduction to the Massoretico-Critical Edition of the Hebrew Bible (London: Trinitarian Bible Society, 1897), chap. XI, pp. 408–410, digitised copy at the Internet Archive. *
3. “Jerusalem Talmud”. *
4. “Bible Manuscripts”, The Jewish Encyclopedia (New York: Funk and Wagnalls, 1901–1906). *
5. Christian D. Ginsburg, Introduction to the Massoretico-Critical Edition of the Hebrew Bible (London: Trinitarian Bible Society, 1897), chap. XI, pp. 408–410, digitised copy at the Internet Archive. *
6. Ginsburg, Introduction, chap. XI, pp. 428–429, Internet Archive. *
7. “The Masorah Parva and Masorah Magna”, Masoretes. *
8. “Masoretic Text”. *
9. María Teresa Ortega Monasterio, “Los códices modelo y los manuscritos hebreos bíblicos españoles”, Sefarad 65:2 (2005), pp. 353–383, doi:10.3989/sefarad.2005.v65.i2.502. *
10. Ginsburg, Introduction, chap. XI, pp. 429–431, Internet Archive. *
11. Ginsburg, Introduction, chap. XI, pp. 429–431, Internet Archive. *
12. “Codex Babylonicus Petropolitanus”, Masoretes. *
13. Ginsburg, Introduction, chap. XI, pp. 429–431, Internet Archive. *
14. María Teresa Ortega Monasterio, “Los códices modelo y los manuscritos hebreos bíblicos españoles”, Sefarad 65:2 (2005), pp. 353–383, doi:10.3989/sefarad.2005.v65.i2.502. *
15. Ortega Monasterio, “Los códices modelo”, pp. 356–357, on El Escorial G-II-8. *
16. Ortega Monasterio, “Los códices modelo”, pp. 356–357, on El Escorial G-II-8. *
17. “Abraham Zacuto”. *
18. Ginsburg, Introduction, chap. XI, pp. 431–432, citing Zacuto’s chronicle in the edition of Filipowski (London, 1857), p. 220, Internet Archive. *
19. David Stern, “The Hebrew Bible in Sepharad. An Introduction”, bilingual essay for the exhibition catalogue of the Biblioteca Nacional de España. *
20. “Tiberian vocalization”. *
21. David Stern, “The Hebrew Bible in Sepharad. An Introduction”, bilingual essay for the exhibition catalogue of the Biblioteca Nacional de España. *
22. Ginsburg, Introduction, chap. XI, pp. 431–432, citing Zacuto’s chronicle in the edition of Filipowski (London, 1857), p. 220, Internet Archive. *
23. “David Kimhi”. *
24. Ortega Monasterio, “Los códices modelo”, p. 367, quoting Norzi’s note on Deuteronomy 12:11. *
25. Ginsburg, Introduction, chap. XI, pp. 431–432, citing Zacuto’s chronicle in the edition of Filipowski (London, 1857), p. 220, Internet Archive. *
26. Ortega Monasterio, “Los códices modelo”, pp. 357–359, on the Hilleli, with reference to M. Goshen-Gottstein, “Biblical Manuscripts in the United States”, Textus 2 (1962). *
27. Ginsburg, Introduction, chap. XI, pp. 431–432, citing Zacuto’s chronicle in the edition of Filipowski (London, 1857), p. 220, Internet Archive. *
28. Grolier Club, “Iberia (Spain and Portugal)”, in the online exhibition Jewish Worlds Illuminated, describing JTS MS L44a. *
29. David Stern, “The Hebrew Bible in Sepharad. An Introduction”, bilingual essay for the exhibition catalogue of the Biblioteca Nacional de España. *
30. Ortega Monasterio, “Los códices modelo”, pp. 357–359, on the Hilleli, with reference to M. Goshen-Gottstein, “Biblical Manuscripts in the United States”, Textus 2 (1962). *
31. Ginsburg, Introduction, chap. XI, pp. 432–433, Internet Archive. *
32. “Bible Manuscripts”, The Jewish Encyclopedia (New York: Funk and Wagnalls, 1901–1906). *
33. Ginsburg, Introduction, chap. XI, p. 433, citing Kimhi, Mikhlol, p. 184b (Fürth, 1793), Internet Archive. *
34. “Jonah ibn Janah”. *
35. Ginsburg, Introduction, chap. XI, pp. 433–435, Internet Archive. *
36. “Bible Manuscripts”, The Jewish Encyclopedia (New York: Funk and Wagnalls, 1901–1906). *
37. “Elia Levita”. *
38. Ginsburg, Introduction, chap. XI, pp. 433–435, Internet Archive. *
39. Ginsburg, Introduction, chap. XI, pp. 433–435, Internet Archive. *
40. Ginsburg, Introduction, chap. XI, pp. 433–435, Internet Archive. *
41. Ginsburg, Introduction, chap. XI, pp. 435–437, Internet Archive. *
42. Ginsburg, Introduction, chap. XI, pp. 435–437, Internet Archive. *
43. Ginsburg, Introduction, chap. XI, pp. 437–441, Internet Archive. *
44. Ginsburg, Introduction, chap. XI, pp. 437–441, Internet Archive. *
45. Ginsburg, Introduction, chap. XI, pp. 437–441, Internet Archive. *
46. Ginsburg, Introduction, chap. XI, pp. 437–441, Internet Archive. *
47. Ginsburg, Introduction, chap. XI, pp. 441–442, Internet Archive. *
48. Moses Maimonides, Mishneh Torah, Laws of Tefillin, Mezuzah and the Torah Scroll 8:4, text and translation at Sefaria. *
49. “Maimonides”. *
50. “Aaron ben Moses ben Asher”. *
51. “Bible Manuscripts”, The Jewish Encyclopedia (New York: Funk and Wagnalls, 1901–1906). *
52. “Aleppo Codex”. *
53. María Teresa Ortega Monasterio, “Los códices modelo y los manuscritos hebreos bíblicos españoles”, Sefarad 65:2 (2005), pp. 353–383, doi:10.3989/sefarad.2005.v65.i2.502. *
54. Ortega Monasterio, “Los códices modelo”, pp. 356–357, on El Escorial G-II-8. *
55. Ortega Monasterio, “Los códices modelo”, pp. 361–364, on Complutense M1 and M2. *
56. “Complutensian Polyglot Bible”. *
57. Ortega Monasterio, “Los códices modelo”, pp. 361–364, on Complutense M1 and M2. *
58. Ortega Monasterio, “Los códices modelo”, pp. 365–367, on British Library Or. 2626–2628. *
59. Ginsburg, Introduction, chap. XI, pp. 432–433, Internet Archive. *
60. Ortega Monasterio, “Los códices modelo”, pp. 373–374, on British Library Or. 2201 and the conclusions. *
61. Ortega Monasterio, “Los códices modelo”, pp. 373–374, on British Library Or. 2201 and the conclusions. *
62. María Teresa Ortega Monasterio, “Los códices modelo y los manuscritos hebreos bíblicos españoles”, Sefarad 65:2 (2005), pp. 353–383, doi:10.3989/sefarad.2005.v65.i2.502. *
63. David Stern, “The Hebrew Bible in Sepharad. An Introduction”, bilingual essay for the exhibition catalogue of the Biblioteca Nacional de España. *
64. “Meir of Rothenburg”. *
65. David Stern, “The Hebrew Bible in Sepharad. An Introduction”, bilingual essay for the exhibition catalogue of the Biblioteca Nacional de España. *
66. Ginsburg, Introduction, chap. XI, pp. 429–431, Internet Archive. *
67. Ortega Monasterio, “Los códices modelo”, pp. 356–357, on El Escorial G-II-8. *
68. Ginsburg, Introduction, chap. XI, pp. 429–431, Internet Archive. *
69. Ginsburg, Introduction, chap. XI, pp. 428–429, Internet Archive. *
70. “Christian David Ginsburg”, Masoretes. *
71. María Teresa Ortega Monasterio, “Los códices modelo y los manuscritos hebreos bíblicos españoles”, Sefarad 65:2 (2005), pp. 353–383, doi:10.3989/sefarad.2005.v65.i2.502. *
72. David Stern, “The Hebrew Bible in Sepharad. An Introduction”, bilingual essay for the exhibition catalogue of the Biblioteca Nacional de España. *
73. “Leningrad Codex”. *