MS Bodley Or. 621 as a “Study Psalter” for Christian Hebraists

Abstract

Because of its diminutive size, unconventional decorative scheme and total lack of built-in translations, MS Bodley Or. 621 is an anomaly among the nine “Hebrew Psalters for Christian Use” listed by Raphael Loewe. It has received scant attention principally because, as a book that appears to have been intended for Jews only to be later appropriated by Christians, it seems to be of lesser relevance to discussions of thirteenth-century English Hebraism than those psalters that were unambiguously custom-made for gentile scholars.

This article challenges this premise and this conclusion. By re-examining and synthesising the paleographical and codicological evidence presented by the text proper, I suggest that the psalter may indeed have been commissioned by a Christian according to his specific needs. By considering the form, content, and distribution of the marginal annotations, especially those that contain not only Latin and French translations but also Hebrew roots, I highlight the remarkable and unique method of learning Hebrew adopted by one of the scholars, which is not unlike the modern notion of learning a language through immersion. By approaching the all-Hebrew psalter with an imperfect grasp of the language, he transforms the experience of reading into a learning process, and the psalm text into a type of textbook. This agrees well with the short and fragmentary nature of surviving Hebrew grammars from thirteenth-century England, and perhaps explains why there was no need for more comprehensive works.

A mong the nine surviving Hebrew psalters from twelfth-and thirteenth-century England that bear unmistakable signs of Christian patronage or use—whether these are in the form of Latin and French marginal annotations, a superscriptio (word-to-word translations placed above each line of the Hebrew), or complete parallel texts, Oxford, Bodleian Library, MS Bodley Or. 621 (hereafter Or. 621), is clearly the odd one out.¹ Its small dimensions (14.5 × 9.7 cm) and unusual decorative scheme aside, it is the only psalter that contains neither superscriptio nor parallel text and the only one in which the Hebrew is written not in square script, but in semicursive.² Another unique feature of Or. 621 is the presence of a group of marginal glosses that include, in addition to Latin and French translations, the derived (sometimes inaccurately so) Hebrew roots, written in an untutored non-Jewish hand. Combined, these factors have led scholars to conclude that Or. 621 was made for ordinary Jewish use and later appropriated by Christian Hebraists, one of whom went so far as to treat the psalter as an exercise book on which notes on Hebrew vocabulary could be casually jotted down.³

In this article, I suggest that the early history of Or. 621 may not have been quite so arbitrary, that it was most likely made with a Christian user in mind, and that the marginal glosses containing Hebrew roots are not random, unconnected notes taken during a lecture but the product of a coherent, if somewhat eccentric, method of independent study. Since my analysis mostly draws on paleographical, codicological, and textual evidence, I am unable to dwell on the historical and cultural context of Or. 621 at any great length; instead, readers interested in these broader concerns are directed to Caroline Gruenbaum’s recent article on the same manuscript.⁴

In revising the current consensus about the production and early reception of Or. 621, I hope to demonstrate that this little psalter fills two important gaps. On the one hand, it contests the assumption that a book of “Hebrew scripture for Christians” must contain in-built reference tools, such as a superscriptio or a facing translation.⁵ The possibility that certain Christian Hebraists would commission Hebrew-only volumes such as Or. 621 means that apart from the ready availability of a full or partial translation, there may have been other factors dictating the preferred format of these books, including portability, cost and time of production, the patron’s proficiency in Hebrew, and the type of books already available to him. On the other hand, the trilingual marginal glosses give a much-needed illustration in situ of the ways in which some Christian Hebraists in thirteenth-century England used Hebrew books. Unlike a modern student reading a foreign-language novel, who proceeds in a linear fashion and glosses all the unfamiliar words, the Or. 621 glossator (largely because of his familiarity with the content of the psalms) used his Hebrew psalter more as a reference book to be read selectively. And unlike many other medieval compilers, annotators, and users of “Hebrew psalters for Christians,” whose scholarly activities were directed primarily toward making sense of the disagreements between different versions of the Latin psalter, the Or. 621 glossator took a much more active and expansive interest in the language of the psalms.⁶ I believe that the origin of this interest is traceable, and that the apparently distinct types of glosses in Or. 621 in fact represent one scholar’s journey from the common textual-critical approach to a glossing practice that is linguistically oriented.

Planning a Pocket-Sized Psalter

The nine “Hebrew psalters for Christians” form a three-tier hierarchy according to their content: those with Latin parallel texts, and in most cases also a superscriptio (Oxford, Corpus Christi College, MSS 10 and 11; Cambridge, Trinity College, MS R.8.6; London, Westminster Abbey, MS 2; Leiden, Leiden University Library, Or. 4725); those with only a superscriptio, supplemented by marginal notes (Warminster, Longleat House, MS 21; London, Lambeth Palace, MS 435; Paris, Bibliothèque nationale de France, hébreu 113); and those that lack in-built Latin translations altogether (Or. 621). The formats of these psalters diverge accordingly: those of the first type tend to be large books, measuring around 30 × 22 cm; those of the second type are medium-sized (approximately 20 × 15 cm); and Or. 621, the only one of its kind, is roughly a quarter the average size of the Type I psalters, and half that of the Type II psalters.⁷

Portability, in turn, required scribal frugality. For example, Or. 621 has just 51 folios to squeeze the same amount of text that BnF MS hébreu 113 spreads across 138 folios, meaning that it must fit twice as many words in half the writing space on each page. Both the size of the letters and the space between lines in Or. 621 were therefore kept to a minimum, leaving an interlinear space (after the addition of vocalization marks) of 1 mm and an average height of 1.5 mm for the letters themselves. Bearing these constraints in mind, it is easy to see why most of the glosses were not written as superscriptio, as squeezing them between the lines would severely affect the legibility of both text and gloss. It is equally understandable why the semicursive script was chosen instead of the more calligraphic square script, the default script for the biblical text in Hebrew manuscripts from medieval Ashkenaz.⁸ The simplified ductus and the slim dimensions of the semicursive letters, taller than they are wide, enabled the scribe to maintain legibility and elegance while saving as much space as possible, whereas the square script, with its horizontal splay, would have been both more time-consuming and technically challenging to write at such a small scale.⁹

Figure 1.

Oxford, Bodleian Library, MS Bodley Or. 621 (145 × 97 mm), fol. 2v. Use of this image is governed by the terms and conditions of the CC BY License.

Whether the scribe saw these space-saving measures as a shortcut or an inconvenience, they certainly made reading more challenging.¹⁰ Besides the usual difficulties associated with reading tiny, tightly written letters, the rarity of the semicursive script in Hebrew manuscripts known to have been consulted by medieval Christians means that some additional training or assistance was most likely needed before the Christian reader was able to decipher the text, due in large part to the ligatures frequently formed between strokes belonging to adjacent letters, supporting Loewe and Smalley’s suggestion noted above that Or. 621 was most likely intended for normal Jewish use.¹¹

The Order and Time of Production

At the same time, however, a number of features cannot be explained by the identification of Or. 621 as an ordinary Jewish book of psalms. Before turning to an examination of these features, it would be useful to consider the circumstances of and the people involved in the manuscript’s production.

The consonantal text of the psalms was written by a single hand in dark brown ink that gradually fades into a chestnut-golden hue. The hand is elegant, idiomatic, and remarkably steady despite the fact that the individual lines were not ruled (hence the variation of the number of lines on each page, even on different sides of the same folio, between 22 and 25). Given the ease and fluidity of the script, the scribe was clearly a Jew. The most telling piece of evidence is his distinctive method of line management.¹² He wrote out as many letters of the last word as possible before reaching the end of the text block, then canceled the unfinished word fragment by placing a dot to its top left, and finally rewrote the same word, in full, on the next line. This peculiar way of articulating the shape of the text block at the expense of efficiency and readability, though utterly alien to Latin codices, was the most common method of line management for Jewish scribes in medieval Ashkenaz.¹³

The text was then furnished with vocalization and cantillation marks by another Jew, who also amended some of the scribe’s errors in a less calli-graphic, though equally fluent Ashkenazic semicursive script. This step was followed by red and blue penwork decorations that take three forms: strings of circles to fill the space between the shorter lines and the edge of the text block, slightly larger drawings (whether abstract, calligraphic, or figural) to fill the gaps between psalms, and faintly drawn red frames around the punctuator’s corrections and around the square-script opening words of each psalm, written in the side margin (for all three types, see fig. 1).¹⁴ The last of these, the square-script opening words, deserve special mention. Although they are in almost the same teal ink as that used for the penwork decorations, they belong to an earlier stage in the production process—that is, after the copying of the main text and before the vocalization.¹⁵ As such, they were most likely written by the scribe himself. The disparity in the types of script is by no means an argument against this attribution, since medieval Jewish scribes were known to code-switch between different registers of script.¹⁶

The marginal glosses in Latin and Old French, written in several hands that most likely belong to the second or third quarters of the thirteenth century, postdate the decorations, which is indicated by the fact that they are always forced out of their usual alignment whenever a drawing protrudes from the text block.¹⁷ A slightly later hand with more developed features of Anglicana, most likely from the late 1200s or the early 1300s, added the Latin first words and Arabic numbering of the psalms on the lower margin of each page and inserted a Hebrew alphabet, complete with lexical explanations taken from Jerome’s Epistles, at the beginning of the psalter.¹⁸ Interestingly enough, both the original glossators and the annotator responsible for the Latin initial words and the alphabet made use of abstract tie-marks made up of simple combinations of circles, short strokes, and zigzags (see fig. 1). These marks, used for linking marginal glosses to their lemma in the text proper, are common in English-made Latin manuscripts dating from the twelfth to the fifteenth centuries that are of a biblical, exegetical, or homiletic nature, though in the thirteenth century they would have been recognizable as part of Robert Grosseteste’s system of concordantial symbols.¹⁹ Another Hebrew psalter that contains an abundance of these symbols is CCC MS 11; the connection between these two manuscripts, and between them and Grosseteste’s system, requires further study. All other inscriptions and annotations date from the fifteenth century or later.²⁰

From this brief overview, it may be concluded that the three main phases of the psalter’s production—the copying, the pointing, and the decoration— probably all took place before the last quarter of the thirteenth century. A terminus post quem may be tentatively placed at around 1220, since both the use of plummet for ruling (visible on most folios) and the adoption of non-square scripts in Ashkenazic books are only attested in manuscripts produced after this date.²¹ This window of roughly fifty years does not guarantee, however, that the three parties necessarily collaborated. Since several scholars have cited the decorations in Or. 621 as evidence for the identity of its patron, it seems to me that the case must first be made about whether the penwork decorations were part of the original project or added by a later owner.²²

The Place of the Penwork Decorations

Compared to other Ashkenazic manuscripts of the psalms from the thirteenth or fourteenth centuries, Or. 621 has a peculiar way of signaling the transition between psalms. The simplest and most traditional method of marking division was to leave a midline gap, approximately the length of two words, between the last word of one psalm and the first word of the next.²³ Another method, heavily indebted to contemporary Latin codices, required the first letter of each psalm to be enlarged and often decorated.²⁴ To accommodate the larger initial, two or more lines of the body text were typically indented or set apart to produce square-shaped clearings.²⁵ A third “hybrid” method called for the augmentation and embellishment not of the initial letter, but of the entire opening word. Large, rectangular gaps were left at the beginning of each psalm for this purpose (fig. 2). In all three cases, the shape and size of the transition gaps remained stable from the end of Psalm 1 until the beginning of Psalm 150. The same cannot be said for Or. 621.

Since the opening word of each psalm was omitted from the semicursive text of Or. 621, which is exactly what the third method required, one expects the transition gaps to contain the complete and decorated opening words.²⁶ Instead of conforming to an invariable oblong template, however, the transition gaps in Or. 621 assume an erratic range of shapes and sizes. While the majority are the height of one line and the length of two or three words, others are several lines high or in the shape of oblongs, rectangles, or squares of various dimensions. Though the majority of the larger gaps are found in the first seven folios, even there the larger and smaller gaps regularly alternate.²⁷ It seems that the design of the transition gaps was never meant to follow a set pattern and was instead determined by location—for example, if a transition gap happens to be found next to the right edge of the text block, its being made multiple lines high would prevent the risk of confusing it with regular indentation—and by the whim of the scribe.²⁸ As a result, the opening words could not have been written into the gaps in the manner of figure 2. The gaps are either too small for the text to make a visual impact or are of the wrong shape altogether (fig. 3). The implication is twofold: the decorated opening words had to be accommodated on the side margins, and the irregular transition gaps within the text block were not intended for any sort of text. Considering the scribe’s fastidiousness in his justification of the left margin, it would be out of character to create these asymmetrical gaps only to leave them blank.

Figures 2 and 3.

Oxford, Bodleian Library, MS Laud Or. 174, fol. 2r; MS Bodley Or. 621, fol. 4r. Use of both images is governed by the terms and conditions of the CC BY license.

The scribe copied the text in such a way as to make decoration essential. The scribe and the decorator, and by extension the punctuator, must have collaborated—unless by an extraordinary and highly unlikely turn of events, the psalter was accidentally left undecorated until an early owner saw the unappealing gaps and had them filled by a meticulous artist who went so far as to copy the exact shade of the teal ink.

The Identity of the Decorator

The penwork decorations provide potential evidence for the original use to which Or. 621 was put. To begin with, the similarity of the abstract patterns that occupy the smaller transition gaps to line fillers in thirteenth-century Latin psalters from England is remarkable:

Figures 4 and 5.

Oxford, Bodleian Library, MS Bodley Or. 621, fol. 12r; London, British Library, Additional MS 62925 (285 × 205 mm), fol. 30v. Use of both images is governed by the terms and conditions of the CC BY license.

Figures 6 and 7.

Oxford, Bodleian Library, MS Bodley Or. 621, fol. 10r; London, British Library, Additional MS 50000 (300 × 195 mm), fol. 142v. Use of both images is governed by the terms and conditions of the CC BY license.

Figures 8 and 9.

Oxford, Bodleian Library, MS Bodley Or. 621, fol. 20v; London, British Library, Royal MS 1 D X (350 × 240 mm), fol. 48v. Use of both images is governed by the terms and conditions of the CC BY license.

Figures 10 and 11.

Oxford, Bodleian Library, MS Bodley Or. 621, fol. 22r; London, British Library, Additional MS 70000 (175 × 120 mm), fol. 30r. Use of both images is governed by the terms and conditions of the CC BY license.

Figures 12 and 13.

Oxford, Bodleian Library, MS Bodley Or. 621, fol. 30v; London, British Library, Additional MS 49999 (150 × 125 mm), fol. 76v. Use of both images is governed by the terms and conditions of the CC BY license.

It may be safely assumed that the decorator either had at his disposal physical specimens of decorated Latin psalters or was sufficiently acquainted with them to draw from memory. Conversant as he was with the general motifs, however, he was much less adept at imitating the quality of their execution. He was not able to achieve the symmetry, the uniformly thin outlines, and the finer details of the line fillers that characterize decoration in Latin or French manuscripts. In figure 8, for example, the red scribbles within the blue rectangle merely suggest the diamond shapes typically associated with line fillers. Similarly, instead of drawing delicate diagonals that, filament-like, stretch evenly into all four corners of the diaperwork squares in figure 10, the decorator could muster only a few crabbed x-shapes. Considering the extremely small scale of these drawings (measuring little more than 3 mm in height), the decorator’s skills are more impressive than suggested by the enlarged reproductions above. Nevertheless, the clear lack of precision and balance and of uniformity and control in the thickness of strokes suggests that he had no professional training.²⁹

The combination of an explicitly “Christian” vocabulary of abstract patterns and the hand of a non-professional is somewhat surprising. If we are dealing with a Jewish user-producer embellishing his own book, then it is worth asking how he came by his exemplars, and why he followed the patterns so closely.³⁰ If we are dealing with a Jewish user-producer entrusting a Christian atelier with the decoration of his book, then the amateurish execution of the penwork is surprising. If we are dealing with a Jewish scribe collaborating with a Christian non-artist, since the skills required for penwork drawing and calligraphy are largely transferable, why was collaboration even necessary?³¹

Of these three possibilities, the first may be completely ruled out if the decorator’s Hebrew writing is taken into account. In a handful of cases throughout the psalter, he filled the space between psalms with a stylized rendition of the opening word.³²

Raphael Loewe has already described these stylized opening words as the work of a “non-Jew” on the basis of several characteristics.³³ Among these characteristics and pertinent to Or. 621 is the frequent willingness on the part of the scribe to compress Hebrew letters horizontally, as seen in figure 14. While this technique is scarcely ever attempted by Jewish scribes, it is a distinguishing feature not only of the Or. 621 artist, but also of the Christian decorator of another Hebrew-Latin psalter, now Leiden Or. 4725.³⁴ Another unmistakable sign of non-Jewishness is the peculiar ductus of the Or. 621 artist. Without delving too deeply into Hebrew orthography, I will take the letter ל (lamed) as a case in point.³⁵

In both the square and semicursive modes, ל is made up of three basic strokes: the upper vertical stroke or the mast, the horizontal stroke or the roof, and the right downstroke, though the square-script ל often also contains a small ornamental serif or flag on top of the mast, as may be seen in figure 14.³⁶ Ashkenazic book scripts tend to make the roof the thickest stroke, gently tapering off both upwards and downwards into the much thinner mast and downstroke; this effect is achieved by keeping the nib perpendicular to the horizontal strokes. The decorator of Or. 621 is far less scrupulous about the direction of the nib; as it happens, the mast becomes the thickest stroke, giving the letter a distinctly top-heavy appearance. The upwards widening of the mast, often to the point of forming bizarre wedge shapes, is reminiscent of the treatment of ascenders in contemporaneous Latin scripts. In addition, the red-letter ל in figure 14 is noticeably written in two, rather than three, strokes.

Figure 14.

Oxford, Bodleian Library, MS Bodley Or. 621, fol. 38v. Use of this image is governed by the terms and conditions of the CC BY license.

Similar misunderstandings about the number, shape, relative size, and thickness of strokes are demonstrated by other letters written by the Or. 621 decorator. I will not enumerate them here. Instead, I would like to point out one extreme case, also visible in figure 14, in which misunderstanding leads not only to infelicity but also to downright illegibility. In blue, the decorator wrote the opening word (happy is) in imitation of the scribe’s handwriting to its left. The result is indecipherable because of the crooked proportions of all four letters, because of the near-indistinguishability between the last two letters, ר (resh) and י (yod)—the former narrowed and the latter lengthened—and, above all, because of the bizarre appearance of the letter ש (shin). While the letter normally consists of three downstrokes, one to the right and two to the left, the decorator’s version of it contains four, including a thicker stroke to the right and three hairline “branches” growing out of it, as well as a small “bud” just above the middle “branch.” The bottom two “branches” overlap unnecessarily, the top “branch” is much too high, and the serifs at the end of the “branches” are smudged together. It appears that the decorator tried to correct himself, only to make it worse. What he lacked was not professional scribal training, but basic orthographic competence.³⁷ The decorator evidently did not have a Jewish education. We must also ask why someone would imitate the scribe’s handwriting in such a clumsy, persistent, and completely superfluous way? The curiosity and enthusiasm of a non-Jew offers a much more satisfactory solution to the problem than boredom or irreverence on the part of an amateur Jewish artist who lacked even the ability to write Hebrew normally.

Figure 15.

Detail from figure 14. Oxford, Bodleian Library, MS Bodley Or. 621, fol. 38v.

Non-Jewishness, however, does not necessarily imply total ignorance of the Hebrew language. In fact, the subject matter of the decorator’s figural drawings, of which fourteen examples exist throughout the psalter, often appears to be connected thematically to the adjacent text.³⁸ It would be useful to note briefly the resemblance, once again, between the human heads drawn by the Or. 621 decorator and those used as line fillers in English psalters of the early and mid-1200s. The three-quarter faces in figures 16 to 18 share a number of distinctive features, including the gracefully arched eyebrows, the large almond-shaped eyes, and the full, squarish jaws. The solemnity of the facial expression is set against the whimsically abstract treatment of hair and headgear.³⁹ On the other hand, the Or. 621 decorator’s relative inexperience is indicated by his confused application of both red and blue ink to the same area that results in a murky brownish-blue; for example, the pupils, eyebrows, nose, and mouth of David in figure 16 confirm his in-experience in the practical aspects of pen-flourishing.

Figures 16, 17, and 18.

Oxford, Bodleian Library, MS Bodley Or. 621, fol. 2v; London, British Library, Lansdowne MS 420, fol. 41v; Additional MS 62925, fol. 29v. Use of all three images is governed by the terms and conditions of the CC BY license.

Out of the fourteen instances of figural decoration, ten involve human heads and the other four depict the heads of animals or monsters. The latter two categories have been grouped together because the artist’s tendency toward abstraction makes it rather difficult to draw the line between the fantastic and the real. While several of these, like the head of King David in figure 16, are of such a general nature that one cannot sensibly ascribe any local meaning to them, others stand out in their unusual choice of imagery, such as the only profile (as opposed to three-quarter-view) human head, which appears on folio 3v (fig. 19):

Figure 19.

Oxford, Bodleian Library, MS Bodley Or. 621, fol. 3v. Use of this image is governed by the terms and conditions of the CC BY license.

Next to the square-script second word of Psalm 11, (of/to David), where another bust of David would have been very appropriate, we find instead the profile of a bald man with a feathery forelock.⁴⁰ His head is turned away from the text, his mouth open. Since the use of the profile form in Western medieval art usually served to signal involvement in action or moral depravity, the one-off appearance of this profile head should not be taken for granted.⁴¹ The first line of the new psalm ends with the word (wander [imperative]), immediately next to the red frame to which the profile head is attached. The decorator’s odd choice of subject matter, I suggest, may have been inspired by this word as the literal interpretation of the adjacent text, a practice that can be amply attested in contemporary Latin psalters.⁴² The connection between and the bald profile is further supported by the marginal gloss to the left of the head, (an incorrectly derived version of the root [to move to and fro, wander]), accompanied by its Latin translation moveo (I move).⁴³ The horizontal dash above moveo that points at the nose of the bald figure, clearly written in the same ink as the rest of the gloss, is significantly not an abbreviation mark. It must be either a superfluous stroke or a linking device that identifies the gloss with the breathless fleeing figure. The glossator almost certainly saw the connection between word and image and appropriated it as a type of mnemonic.⁴⁴ There may also have been a link between the sharp, beak-like nose of the bald man (not to mention his feathery forelock) and the reference to birds in the same verse: “Flee to Your mountain like a bird.”

To avoid sinking too deeply into speculation, I will consider another example. Psalm 51 ends with the sentence (Then they will offer on your altar bulls). In the transition gap after the last word (bulls), the artist drew the profile heads of two animals, both with drop-shaped eyes, long snouts, and wide-open mouths. The animal on the left, with its pointed and droopy ears and narrower snout, looks rather dog-like, while the animal on the right is distinctly bovine (fig. 20). Since this is the first appearance of animal figures in a transition gap, the artist’s decision to draw these long-faced beasts was likely to have been prompted by the word directly to their right.

The question of whether the decorator sought external help in translating individual words, or at least in matching the Hebrew verses to those in the Vulgate, is of secondary importance. That he made any effort at all to coordinate text and image despite not being under the least obligation to do so is extraordinary enough, judging by the unambitious nature of the manuscript as a whole and the varied and haphazard nature of its decorative scheme. This effort is consonant with what we already know about the decorator from his stylization of the opening words and from his involvement in the making of such a book. His incompetence as a draftsman was compensated by his personal interest in, and quite possibly knowledge of, the Hebrew language. While it is not impossible that he may have been a good-humored Christian friend of the Jews who was incidentally asked to decorate one of their own books, it is far more likely that the psalter was produced in the context of an informal Hebrew “school,” possibly attached to a cathedral or monastery, where Christian scholars and Jewish instructors regularly collaborated over the production of biblical volumes that were used not for devotion but for study.⁴⁵ In the making of a “study text,” economy and efficiency would have been as important as structural transparency effected through a decorative scheme that highlights all of the major divisions within a text. The seemingly self-contradictory format of Or. 621, with its bewildering combination of small size, informal script, complex system of decorated opening words, and embellished transition gaps, fits into this model perfectly. The most plausible theory for the origin of Or. 621, therefore, is that it was made by two Jewish informers (whether converts to Christianity or not) and one Christian Hebraist, who may well have been the first owner (if not one of the annotators) of the psalter.

Figure 20.

Oxford, Bodleian Library, MS Bodley Or. 621, fol. 17v. Use of this image is governed by the terms and conditions of the CC BY license.

What remains to be addressed is the question of accessibility. It may well be that a Christian patron is implicated by internal evidence. Considering the lack of evidence as to whether any Christian Hebraist in thirteenth-century England was able to read Hebrew books unaided, the assertion that one scholar would so willingly subject himself to such an unnecessary ordeal, eschewing the more student-friendly options such as a bilingual psalter or at least one designed for extensive interlinear glossing, is plainly absurd.⁴⁶ For the remainder of this article, I will consider the problem of accessibility and describe some of the surprising ways in which the psalter was used by its earliest readers.

The Type I Glosses

There are, broadly speaking, two types of marginal glosses in Or. 621: those of the first type are found on all but the last nine folios; those of the second type are largely limited to the first three and last six folios of the psalter. Both types are shown in figure 21.

Glosses of the first type (hereafter “Type I glosses”) are situated close to the text block on the left and right margins and are accompanied by either signes de renvoi (here in the form of z-shapes, though the lexicon of symbols becomes increasingly diverse throughout the book) or short dashes that connect them to their Hebrew lemmata in the text. They are written with a very fine nib in a neat and meticulous hand. The short Latin and French glosses are inflected in accordance with the grammatical context of the Hebrew. In Psalm 5:8, for example, inclinabo (I will bend, incline) in the first-person singular future translates (I will bow down), which is in the first-person singular imperfect.⁴⁷ The rendition of the Hebrew imperfect as the Latin/French future (and the Hebrew perfect as the Latin

Figure 21.

Oxford, Bodleian Library, Bodley MS Or. 621, fol. 2r. Use of this image is governed by the terms and conditions of the CC BY license.

perfect or French preterite) is also common practice in the superscriptio psalters.

In their brevity, legibility, and careful replication of grammatical detail, the Type I glosses resemble the complete superscriptio translation as preserved in CCC MS 10, CCC MS 11, and Trinity College MS R.8.6. However, the Latin translations given by the Type I glossator (not to mention his frequent use of French instead of Latin) rarely agree with those in the superscriptio psalters, a fact already noted by Beryl Smalley.⁴⁸ In Psalm 5:4 (fig. 21), for example, the Hebrew (and I will look forward to, hope for, expect) is translated as [et] sperabo ([and] I will hope, look for, expect) in Or. 621, and as et speculabor (and I will watch, examine) in the superscriptio psalters. Just one line below in Psalm 5:6, for the Hebrew (mad, foolish ones), the Or. 621 glossator has derisores (mocker, scoffer) and the superscriptio psalters, errantes (straying, erring ones).⁴⁹ It may be assumed that the Type I glosses were not taken from the superscriptio Lincolniensis but from some other source.

Another factor that distinguishes the Type I glosses from the super-scriptio Lincolniensis is the former’s partial nature; only a small fraction of the Hebrew text was translated. As I suggested earlier, the limited marginal space in Or. 621 allows no more than two or three words to be glossed in each roughly ten-word line. Yet considering that much empty space is left on the side margins in most folios, the decision to gloss the psalter sparingly must have been a matter of choice rather than necessity. Caroline Gruenbaum, apparently puzzled by this decision (and by the fact that the same word was often re-translated), describes the Type I glossator to be “indiscriminatingly translating at whim.”⁵⁰ I suggest, however, that a clear and consistent program may be discerned in his activities.

The glossator’s tendency to gloss Hebrew words whose Latin equivalents in the Gallicana and iuxta hebraeos versions happen to disagree cannot have been purely accidental. This intention to make sense of the differences between the two Latin translations is sharply defined near the beginning of his glosses. Out of the eleven Type I glosses for Psalm 5, for instance, eight serve to explain the most significant lexical discrepancies between Jerome’s Gallican and “Hebrew” translations. The first gloss, parabo (I will prepare, provide) for (I will prepare, arrange), addresses the difference between the Gallican Psalter’s adstabo (I will stand by) and praeparabor (I will prepare) in Jerome’s final translation. Later in the same psalm, the Hebrew (from their counsels) is translated by the glossator as studiis suis (from their studies). There is a disagreement between the Latin psalters at this point, where the versio Gallicana has cogitationibus suis (from their thoughts, meditations) and the versio iuxta hebraeos has consiliis suis (from their deliberations, counsels). In short, there is nothing whimsical or directionless about the Type I glosses, nor are they strongly personal; instead, they were driven by a need, widely felt at a time, to note and explain the variant Latin readings without necessarily committing oneself to a “correct” replacement translation.⁵¹

From this evidence, several conclusions may be drawn about the glossator behind the Type I glosses. His interest was predominantly textual-critical, and his choice of words to gloss was unrelated to his familiarity with these words. He might skip over words that he does not recognize or gloss a word despite knowing its meaning or despite having already seen it several times, since the purpose of the Type I glosses was not so much to supply meanings as to label certain words in the psalter as especially worthy of scholarly attention. Indeed, because his method was predicated on the assumption that the Latin versions must have been more or less faithful to the Hebrew when they did not disagree, he was under no obligation to “read” the complete Hebrew psalter at all. With his (presumably) excellent knowledge of the Gallican psalter, he only needed to supply the Hebrew originals underlying the variant readings before considering himself to have mastered the Hebrew psalter as a whole.⁵² Because of the special relationship between the glossator and his text and the specificity of his aims, it would be inappropriate to apply modern notions about reading and language-learning to his activities, since he saw the Hebrew psalter more as a tool of his trade—almost dictionary-like in its capacity to be consulted ad hoc— than something to be read from cover to cover. The condensed and seemingly forbidding format of Or. 621 was perfectly suitable for this purpose.

As for the conditions under which the glossator worked, the arbitrary alternation between Latin and French from gloss to gloss suggests that the glosses may have been added in a spontaneous and hurried fashion, quite possibly because these glosses were not copied from another book but dictated by a Jewish teacher sitting by his side.⁵³ Whichever the case, it is unlikely that the glossator supplied the translations himself, and his own proficiency in Hebrew does not emerge with any clarity from his annotations. However, in order to make any use of the glossed psalter, he must at least have known enough Hebrew grammar and vocabulary to tell which word was which.

The Type II Glosses

The Type II glosses are situated not only on the side margins, but also above and below the text block (fig. 21). In the first three folios, where they coexist with the Type I glosses, their awkward positioning indicates that they were added later. Only some on the upper and lower margins are accompanied by signes de renvoi, though these only identify the lines in which the lemmata appear, and not the lemmata themselves. The glosses on the side margins are placed, wherever possible, on the same horizontal plane as their respective lemmata; again, their precise locations are not indicated.

A noticeably broader nib was employed, and there is much less uniformity in the size and placement of the letters. The result is considerably more chaotic than the Type I glosses, though if we remind ourselves of the psalter’s dimensions, these flaws are excusable. Each gloss tends to give multiple meanings, which are almost always in Latin.⁵⁴ The inflections have been standardized so that all verbs are in the third-person singular perfect, and all nouns are in the nominative singular. In the case of verbs, the meanings for different binyanim or verbal stems are lumped together under the same root, with no distinction being made between them.⁵⁵ For the root (fig. 21), for example, the Type II glossator proposes three meanings: appretiavit, composuit vel paravit (he has estimated the price of, he has put together, or he has prepared).⁵⁶ The first of these, appretiavit, translates the hiphil stem (he valued, estimated) of the root while the second and third translate the qal stem (he arranged, prepared), which is the form attested in Psalm 5:4: “In the morning I will arrange [my prayers] to you.” It is characteristic of the Type II glossator that he deliberately avoided selecting one meaning out of many based on context, and unlike the Type I glossator, his choice of lemmata was not limited to words that had caused disagreements in Latin, which account for less than half of his total glosses.⁵⁷

This wider and more diffused interest in the language of the Psalms is confirmed by the Type II glossator’s inclusion of Hebrew roots in addition to the translations. The letters are clumsily written, and vocalized according to a simplified system followed by many Christian Hebraists at the time, the basic principle of which was the drastic reduction of the number of vocalization marks so that, like in Latin, each of the five vowel sounds was represented by one mark only.⁵⁸ The consistent use of this simplified system, the choice of Latin as the principal language, and the ordered way in which different meanings are listed within one gloss suggest that the Type II glosses were unlikely to have been supplied by a Jewish teacher, and most likely derived, instead, from some kind of written source, quite possibly a dictionary or a biblical concordance.

Whichever type of written source he consulted, one important observation to be made about these roots is that to obtain them from the inflected forms in the text, the glossator must constantly exercise his knowledge of Hebrew grammar, especially since Hebrew roots take on prefixes as well as suffixes, not to mention changes within the root itself. On folio 2r, for example, he mistakenly derived the root (he spoke, said, uttered)— translated into Latin as locutus est (he has spoken)—from the adjective (weak), which is in fact a derivative of the root (to be weak, miserable). This mishap reveals a cogent train of grammatical thought: having observed that contains the letter א—a first-person singular imperfect marker— followed by three more letters, and that the next word is (I), the glossator concludes that must be the first-person singular imperfect form of a verb. This leaves as the three-letter root; a subsequent consultation with his source yields (he spoke).

The Type II glossator also had a tendency to group together Hebrew words that look or sound similar. The first three folios of Or. 621 contain numerous groupings of words that differ from one another either only in vocalization or in the last letter: for example, and on folio 2r (see fig. 21). Of these five words, only (a misspelled version of [anger, fury])—translated as ira vel furor (anger or fury)—is found in Psalm 6 on the same page, while two of the four “false friends” introduced here— , a misspelled version of (cream, butter) translated as butirum (butter) and , an incorrectly vocalized version of (mother-in-law) translated as socrus (mother-in-law)—do not appear even once in the psalter. While Judith Olszowy-Schlanger considers these groupings to be a learning aid through which similar-looking words may be memorized together, it could equally be likely that, confronted with these possibilities in his alphabetically arranged source, the glossator wrote them all down simply out of uncertainty as to which is the best way to translate (in your fury) in Psalm 6:2.⁵⁹ In doing so, he was again willfully turning a blind eye to the absurdity of talking about the “butter,” let alone the “mother-in-law” of God. This eagerness to take all possibilities into account, without giving much thought to the resulting shifts in meaning, is entirely consonant with his treatment of the multiple translations within one single gloss. While the Type I glossator was anxious to establish the precise meaning of the Hebrew psalter, the Type II glossator seems much more at ease with equivocation. The frequency of his misreadings, most of which can directly be traced back to visual ambiguities in the scribe’s semicursive handwriting, testify both to the originality of these glosses since they are unlikely to have been copied from elsewhere and, paradoxically enough, to the glossator’s possession of a rough-and-ready command of Hebrew since no Jew assisting him would have made such mistakes.⁶⁰

The scope of this article does not allow me to consider the issue of sources, specifically the connection between the Type II glosses in Or. 621 and the only Hebrew-Latin-Old French dictionary to survive from medieval England, now Part IV of Longleat House MS 21, in great detail.⁶¹ By way of a conclusion to this section, I will briefly note the overwhelming affinity between the Type II glosses and the entries in the Longleat dictionary, not only in verbal pattern—such as in the frequent use of the phrase pro quo habemus (for which we have) to introduce Vulgate translations for a given Hebrew word—but also in the translations themselves, where ubiquitous overlaps suggest that the Type II glosses could very well have been abbreviated from dictionary entries that were very much like those in LH 21. Since the dictionary entries vocalize and translate the verb roots as second-person masculine singular imperatives, while the Type II glossator used the third-person masculine singular perfect, the relationship between the two manuscripts was most likely indirect. There is little doubt, however, that they originated from the same milieu. Much could be gained from their detailed comparison.

Two Glossators, One Glossator?

While the first three folios of the manuscript are useful in illustrating the two extremes of glossing practice in Or. 621—one tidy, impersonal, and textually motivated, the other messy, private, and of a quasi-linguistic nature—further investigation into the contents of the manuscript reveals that the distinction is not nearly so secure. For instance, the Type II glosses are not entirely absent from folios 4–42. In fact, glosses of this type can be

Figure 22.

Oxford, Bodleian Library, MS Bodley Or. 621, fol. 21v. Use of this image is governed by the terms and conditions of the CC BY license.

found in as many as seventeen of these folios, relatively evenly scattered throughout this range.⁶²

On the top and bottom right corners of folio 21, for example, and on the left side of its lower margin, we find the familiar arrangement of Hebrew roots accompanied by Latin translations. The untrained Hebrew script, the simplified system of vocalization, the presentation of the translations in the third-person singular perfect, and the positioning of these annotations in accordance with the location of the lemmata in the text block are all typical of the Type II glosses. The same principle of grouping like with like informed the gloss for (name of a mountain): idem quod quod est infernus ( the same as which is hell). (great darkness), which does not appear in the psalm text at this point, was introduced to explain the unfamiliar word. Even more tellingly, the glossator responsible for these scattered entries had a tendency to place a silent shewa beneath word-final consonants that do not require it, as in . The same habit is attested in the Type II glosses, where the consonants בד and ז are regularly vocalized with a shewa at word-final position.⁶³

These Type II glosses appear, however, to come from the hand that wrote the Type I glosses on the same folio. The letters, apart from being of the same shape, size, ink, and nib width, are also written with the same level of currency, which vacillates between linked and disconnected strokes. This proves that the glossator to whom I have attributed the Type I glosses would occasionally also produce Type II glosses such that they resembled those found in the first and last folios of the psalter. I would go even further and suggest that the two glossators were the same person.

In another example, figure 23 shows an ordinary Type II gloss; in figure 24, the Type I glossator made a one-off attempt to derive the root (he acted, did)—causavit [sic] (he has pretended, he has pleaded) from (deed, act)—causationem (excuse, apology). In normal, idiomatic Ashkenazic Hebrew scripts (right side of figs. 24 and 25), the letter ע is roughly on the same level as the lower half of ל. In the handwriting of both glossators, however, the letter ע has been pushed up to the point where its right arm touches the middle of the right downstroke of ל. The long and narrow dimension of ל, the ogee by which its lower turning is represented, and the round hook on top of the right arm of ע oddly situated above rather than slightly below the top of the other arm, are all highly atypical features that, judging from surviving examples, were by no means representative of the handwriting of Christian Hebraists.⁶⁴ The most plausible explanation for this striking resemblance is that there was only one glossator.

Figures 23, 24, and 25.

Oxford, Bodleian Library, MS Bodley Or. 621, fols. 2v (left), and 4r (middle and right). Use of these images is governed by the terms and conditions of the CC BY license.

A Brief History of the Glosses

Why, then, would someone change both his method and his handwriting so dramatically in the course of annotating such a small book? To answer this question, I posit the following scenario.

The primary glossator began by translating the contentious Hebrew words and occasionally other words that he found interesting in the Latin translation. The result is the Type I glosses. The primary glossator almost certainly had by his side a copy of Jerome’s translation iuxta hebraeos, and very possibly, a Jew who would provide, when asked, translations for the Hebrew words in French and (naturally) inflected in the same way as the original. In some instances, the glossator would spontaneously translate into Latin to facilitate comparison with Jerome. This process created a lengthy project that lasted through many sessions, as may be seen from fluctuations in the neatness of the glossator’s handwriting: a general decline is interrupted by occasional attempts to write tidily again, presumably at the beginning of each new session. Another implication of the episodic nature of this arrangement is that the psalter was presumably in the possession of the Christian scholar when glossing was not actively taking place; this might explain the glossator’s occasional experimentations with the Type II glosses, some of which already imply the use of a dictionary or concordance.⁶⁵ An important difference between these rare and scattered Type II glosses in folios 4–42 and those in the first and last folios is the former’s clandestine and subsidiary nature. These glosses are always written as far away from the text block as possible (see fig. 22)—indeed, so close to the edge of the page that many were trimmed away during a subsequent rebinding. It seems that while the glossator already had an interest in etymology and the technicalities of the Hebrew language, he did not allow himself to be distracted from his primary purpose.

Something changed at folio 41, where the Type I glosses in alternating Latin and French cease abruptly. It seems that from this point on, the glossator was no longer obtaining the translations from his Jewish informer. How this change came about can only be surmised. Considering the increasingly dire circumstances under which the Jews lived in the time of Henry III and Edward I, especially the successive regulations on Jewish settlement, this sudden interruption is not surprising.⁶⁶ As a result, the final eight folios witness something of the birth of the Type II glosses proper. The use of French becomes rare. The glosses without Hebrew roots increase in length, and the glosses containing Hebrew roots migrate from the distant corners of each page to the side margins and the middle of the upper and lower margins. From folio 43 onwards, the “Type II glosses” as we know them dominate the page.

The changes in the format of the glosses, then, reflect the changed ways in which they were obtained. Rather than being provided French translations, the glossator now had to excavate the meanings for himself by analyzing the morphological structure of the Hebrew, by looking up the roots in a dictionary or concordance, and quite possibly also by selecting, condensing, and rearranging the meanings given by his source. I will leave the reader to decide whether it was out of laziness or diligence that the glossator chose to copy out as many options as possible. In either case, there was clearly a shift of focus. The glossator had passed from a systematic scholarly scrutiny of contentious words in the Psalms to a more leisurely and good-humored exploration of the Hebrew lexicon, led perhaps by curiosity, perhaps by confusion, most likely by both.

As for the changes in his handwriting, I have already remarked that the Type I glosses become progressively more carelessly written in the course of the book. I would also suggest that the experience of writing Hebrew shaped his Latin handwriting in very specific ways. As a result of more pressure being exerted at the nib, the Latin letters became larger, more angular, much less graceful, and in short, more similar to his Hebrew writing.

The Type II glosses in the first three folios are in fact the glossator’s last contributions to the psalter. After reaching the end, he returned to the beginning. Understandably unsatisfied with his former approach, he revised some of the existing translations and glossed many more new words of interest. The final development in his glossing practice was the organization of the Type II glosses into veritable “glossaries” on the upper and lower margins of folio 3.

I have no explanation for the sudden cessation of the Type II glosses at folio 3r, at a point where their presentation had only just become formalized; nor is the account above anything more than a simplified, generalized, and possibly mistaken version of what may have happened. What may be stated with some confidence about Or. 621 is that it documents the progress of a Christian scholar whose initial concern with the accuracy of the Latin translations eventually led him to a profound appreciation of the Hebrew language itself. After all, the most significant development in his glossing practice was the increasing frequency with which he wrote Hebrew letters. I cannot help but think that like the decorator (if the two were different people), he derived some pleasure from that simple and humbling act.

Conclusion

The last word has not been said about Or. 621. The Type I glosses need to be studied in greater detail. The Type II glosses need to be compared more closely with the Longleat dictionary to reveal the degree of their affinity. The possible connection between Or. 621 and CCC MS 11 requires further research. The specific dialect of Anglo-Norman French used in the glosses should be examined by scholars of that field, since it may help to localize the manuscript more precisely. Finally, many gaps are yet to be filled in our knowledge about the scale and operation of medieval English “Hebrew schools” and their scriptoria.

Ultimately, the importance of MS Bodley Or. 621 lies in what it has to say about cultural exchange between Christians and Jews in pre-Expulsion England. Its pages remind us that the Hebrew language, quite independently of its usefulness for scriptural exegesis, intrigued and fascinated medieval Christians. At a time when other, more public forms of cross-cultural contact were increasingly circumscribed, biblical scholarship was one of the few pretexts under which this fascination might be given concrete expression.