Malachi Beit-Arie THE CODICOLOGICAL DATA-BASE OF THE HEBREW PALAEOGRAPHY PROJECT: A TOOL FOR LOCALIZING AND DATING HEBREW MEDIEVAL MANUSCRIPTS When in 1965 the Hebrew Palaeography Project, established by the Israel Academy of Sciences and Humanities and the Institut de Recherche et d'Histoire des Textes of the C.N.R.S. (France) forged tools for the systematic and detailed description of all the dated manuscripts as a basis for creating a typology of handwritten Hebrew books and providing criteria for localizing and dating the many thousands of manuscripts without colophons, we decided to study and record all the physical, technical, graphic and aesthetic features involved in producing handwritten books, including scribal formulas. This decision was initially motivated by methodological and theoretical reasons, following the demand voiced by recent Latin palaeographers calling for the expansion of the scope of palaeography, usually confined to the study of the script, as the sole criterion for dating, and the introduction of a codicological discipline which no longer regards manuscripts merely as vessels for transmission of texts or of old scripts, but as physical objects and cultural products reflecting the social context and the technology and aesthetics of their time and place.1 In the questionnaire we composed for the purpose of describing all the dated manuscripts, we included as many aspects and features as we could think of, based on our own experience, on sporadic studies of some codicological elements by Latin and Greek palaeographers, and on consultations with a few distinguished non-Hebrew palaeographers. Thus, in studying the dated manuscripts we recorded in detail all the visible data concerning writing materials, inks, quiring, means for ensuring the right order of the quires, sheets, leaves or columns, pricking and ruling techniques, format and layout, density of letters, devices for producing even left margins, graphic para-scriptural elements and auxiliary signs, substitutes for the tetragram, decorations and illuminations, scribal formulas at the beginning and end of copying, and formulations of colophons, including rendering of dates, names etc. We did not and could not know in advance which of these codicological elements might be determinant, conditioned by time and place. We therefore preferred to record all of them and to find out at the processing and classifying stage which of them - if any - would emerge as typical practices of the time and area of the production of a manuscript and consequently serve as criteria for palaeographical identification. Already as our field work was progressing we realized that not only was our guiding theoretical approach justified, and our assumptions materialized, for we saw that many codicological practices are indeed grouped according to geo-cultural areas, and part of them also chronologically,2 but that the recording, classifying and exploiting of these data in identifying manuscripts are vital for Hebrew palaeography even more than for palaeography of other scripts, because of the singularity of Jewish history. The extraordinary historical circumstances of the Jewish people, the great mobility of groups and individuals, the frequent expulsions of communities or entire populations and migration by choice or out of economic necessity, undermined the effectiveness of script typology in identifying the origin of manuscripts, since immigrant professional scribes and copyists retained their homeland script. On the other hand, we realized that books produced by immigrant scribes and copyists reflect entirely or at least in part, codicological practices of the geo-cultural regions of their new localities,3 in particular the writing materials, composition of quires and ruling techniques, as the scribe, who had to be provided with or buy the local writing materials, probably got as ready-made ruled quires.4 Thus the codicological features of a Hebrew manuscript are essential not only to the comprehensive characterization of a handwritten book, and as support in identification of its provenance and date on the grounds of script, but may constitute the sole grounds for identifying the area of production, unlike the script, which may reflect the origin of a scribe and not necessarily actual location. After several years of studying dated manuscripts and recording their codicological data we realized that due to their quantity and complexity, the only way to process, classify and retrieve them would be by computerization. We had to convert our questionnaires into coded forms and key most of the information into an electronic data-base. Having started computerization many years ago, in the age of punched cards, we paid a high price for our pioneering. We created a complicated and rigid data-base, which had to be stored in and generated by a main-frame computer, devoid of direct access and flexible retrieval system. Recently, after many attempts to adopt more modern and friendly technologies, the data-base was installed on an IBM-AT computer, upgraded to a 25 mega-herz 80-386 system, with 4mb, RAM, CDC 28ms, 70mb hard disk. The data-base was converted into d-base format. We developed a user-friendly system, and a large number of application programmes to match our own specific requirements and to perform various statistical analyses and modes of search, data correlations and combinations. The benefits and potentials of this data-base and its retrieval system will now be described. The data-base now contains the codicological features and also the textual information (such as names of scribes, owners, localities and countries, and various scribal formulas) of most of the dated manuscripts studied by the Hebrew Palaeography Project. It includes the data of 2122 manuscripts dating up to 1540, which were recorded in situ. However, as one-tenth of these manuscripts were copied by more than one scribe, and because each of these scribes was studied and described in a separate questionnaire, the data-base at this stage in fact holds 2412 dated palaeographical units, or manuscripts. In addition we created a separate file, in which the bibliographical data - dates, localities and their geographical identifications, names of scribes and original owners - of some 800 unstudied dated manuscripts were recorded, and can be retrieved together with the codicological data of the studied manuscripts. These manuscripts comprise those which have not yet been studied by us, or are inaccessible, and those which do not exist any more. The accessible manuscripts were recorded on the basis of microfilms in the Institute of Microfilmed Hebrew Manuscripts of the Jewish National and University Library, while the others were recorded according to catalogues, handlists and references in literature. The data-base now holds elementary data on 600 unstudied existing and 165 lost dated manuscripts. The sum total of the dated manuscripts now stored in the data-base is 3177. In addition we have started to add data of some 1500 medieval manuscripts with undated colophons which contain scribes' names or localities, out of which some 500 have been fully studied, and the rest recorded on the basis of microfilms. Therefore, the data-base will shortly comprise about 4700 Hebrew medieval manuscripts. Following are a few statistics yielded by the computer which have significant implications for the history of the Hebrew book and the intellectual and social history of the Jewish people in the late Middle Ages, as well as for the validity of the data-base and its probability limitations in serving the typology of the handwritten book and as a tool for dating and localizing undated manuscripts. Of the 2122 studied dated manuscripts, 185 were copied by several (two to nine) scribes together, i.e. only about 9%. In Ashkenaz, particularly in the thirteenth and fourteenth centuries,5 and in Byzantium, the rate of multi-scribe copies is a little higher and amounts to 13%, while in Yemen such a phenomenon is rare and is found only in 3.5% of the dated manuscripts. The low rate of multi-hand manuscripts demonstrates one of the fundamental differences between the ways of producing Hebrew handwritten books and transmitting and disseminating texts in Hebrew script and those used in the case of Latin handwritten books and texts. While Latin books from the late seventh until the mid-thirteenth century were produced mainly in institutional copying centres of monastic multi-copyist scriptoria, and were later reproduced by university stationers employing the pecia6 system, and in the late Middle Ages to a large extent in commercial urban and lay ateliers,7 medieval Hebrew books were not produced and disseminated by the intellectual establishment or upon their initiative,8 whether in religious, academic or secular institutional copying centres, but privately and individually.9 The individual nature of the production of Hebrew books is clearly demonstrated by the statistical data provided by the data-base with regard to the destination of the dated copies. Of the manuscripts whose colophons state explicitly or implicitly for whom they were made, 65% were written by hired scribes for owners, and 35% by copyists for their own use (or, to a small extent, for members of their own family).10 The rate of non-professional scribes was in fact higher, since many of the multi-hand manuscripts were copied for one of the copyists, and all the other copyists were most likely not professional scribes but rather his relatives or students. Furthermore, it seems that most of the manuscripts whose colophons do not state their destination (37% of the existing dated copies) must have been owner-produced books, and only a small part of them were made by professional scribes who prepared copies of popular texts in advance for chance buyers or book dealers.11 Therefore, one may assume that at least half the medieval Hebrew manuscripts were personal user-produced books, copied by the scholars12 who were going to use them, and only half, or less, were impersonal, but not institutional books, produced by hired scribes, either professional or occasional. The high rate of the personal production of Hebrew books and the individual nature of copying and disseminating texts single out and characterize Jewish society and Hebrew palaeography, and reflect the individual mode of learning and the extent of literacy and education, unconfined to religious or academic centres. These factors also shed light on the particular ways of transmitting Hebrew texts. The fact that so many books, mainly those which were less popular, were copied by scholars for their own use, must have affected their versions. The hired scribe probably avoided critical and conscious interference in the transmission, his copying is more loyal to his model, but he is more vulnerable to unconscious changes and mistakes conditioned by the copying mechanics, while the scholar-copyist may deliberately interfere in the transmission, revise his exemplar, amend and reconstruct the text, add to it and modify it according to his knowledge, memory, conjecture or other exemplars, and regard copying as a critical editing and not as mere duplicating.13 Localities of production were explicitly indicated only in 1429 out of 3012 existing dated palaeographical units stored in the data-base, and a similar rate of about 48% also prevails in the studied manuscripts. Indications of the place of copying are more frequent in Italy (about 56%) and Yemen (about 60%), and less so in Ashkenaz, where they are found only in some 21% of the manuscripts. In all areas the rate increases from the second half of the fourteenth century onwards. The geo-cultural14 distribution of the manuscripts in the data-base, those with explicit locality and those identified by us, is as follows (the numbers indicate palaeographical units, as multi-hand manuscripts were split into different units; the numbers in parenthesis indicate the number of actual codices): area known dated mss extant dated mss studied dated mss Sefarad 689 (618) 659 (588) 539 (470) Ashkenaz 455 (393) 443 (381) 388 (329) Italy 1076 (987) 1047 (958) 834 (749) Sicily 23 ( 20) 23 ( 20) 20 ( 17) Byzantium 284 (245) 275 (236) 217 (180) The Orient 305 (278) 301 (274) 249 (222) Yemen 160 (156) 152 (148) 121 (117) Unidentified 185 (179) 112 (106) 44 ( 38) DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD Total 3177(2876) 3012(2711) 2412(2122) Since the survival of dated manuscripts was surely accidental, their geographical division cannot be representative. This distribution does not necessarily indicate the rate of book production (in relation to the size of the Jewish population) or the intellectual level of each area, but it does reflect the different historical conditions of the Jewish communities which resulted in differing rates of survival of medieval books. The fact that more than one-third of the surviving manuscripts were produced in Italy, whose Jewish population in the late Middle Ages is estimated to have constituted about one-third of that in the Iberian peninsula and Provence, while the number of surviving manuscripts from the areas of the Sefardic book tradition (which included North Africa) amounts to hardly a quarter of the total surviving manuscripts, or, again, the small quantity of Ashkenazic dated manuscripts (about one-seventh of the total number) does not necessarily reflect greater literacy among Italian Jewry, but rather different historical circumstances which affected the physical survival of their books.15 Perhaps the extent of use in each area also affected the preservation of books. Yet, these statistics may contribute additional information to Jewish history. For instance, the relatively large number of Byzantine manuscripts (about 10%), which is close to that of the Oriental, may attest to the intellectual level and perhaps the size of the Jewish population in an area where our knowledge of the Jewish communities is not sufficiently wide.16 However, the main significance of the geo-cultural division relates to the validity of the data-base and its limitation in identifying manuscripts without colophons and setting up a typology of the handwritten Hebrew book. The chronological distribution of the manuscripts regardless of their provenance is as follows: extant dated mss studied dated mss units mss units mss 10th century17 10 10 4 4 11th century 38 31 28 21 12th century 54 53 44 43 13th century:1st half 58 51 50 43 13th century:2nd half 213 180 184 151 13th century (total) 271 231 234 194 14th century:1st half 292 253 254 215 14th century:2nd half 365 330 309 276 14th century (total) 657 583 563 491 15th century:1st half 529 479 457 410 15th century:2nd half 1015 910 763 663 15th century (total) 1544 1389 1220 1073 16th century (until 1540) 438 414 319 296 DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD Total: 3012 2711 2412 2122 As one would expect, the chronological distribution of the dated manuscripts is far from being balanced: the number of surviving dated Hebrew books increases chronologically, reaches a peak in the second half of the fifteenth century, and naturally decreases in the first half of the sixteenth (until the year 1540, the chronological limit of the Hebrew Palaeography Project),18 which overlaps the spread of Hebrew printing. The number of manuscripts dating from the second half of the fifteenth century comprises more than one-third of the total number, and that of the entire fifteenth century more than half of it. The number of extant dated pre-1200 manuscripts is meagre; of these still fewer belong to the tenth century. Indeed, more than one hundred pre-1200 manuscripts are known to have survived, but many of these are fragments from the Cairo Geniza which may represent the oriental book script, but hardly the codicological practices. Furthermore, the geographical distribution of the early codices is not at all balanced as is shown by the detailed table below, which presents the geo-cultural division (chronologically). Only 21 known units (15 manuscripts) dating before 1200 were clearly produced outside the Middle East (of which 18 units = 12 manuscripts have been studied): The earliest book from the "Sefardic" areas is Ms Leningrad, Saltykov-Shchedrin Public Library, II Firkovitch B124, a biblical manuscript copied in 946 in Ga[bes] (Tunisia); Ms Leningard, II Firkovitch B115, a Bible (only the last 36 leaves have survived) was written probably in North Africa or Spain for Joseph the Spaniard ben Isaac in 994; Ms Oxford, Bod. Heb. b.1, fols. 10-20 (Cowley's Catalogue no. 2673), a treatise of the Babylonian Talmud copied in 1123 by a scribe originating from Libya; three localized manuscripts from Spain dating from late twelfth century, the earliest being Ms Hamburg, Staats- und Universitatsbibliothek Cod. Hebr. 19 (Steinschneider's Catalogue no. 165), a Talmudic copy produced in Gerona in 1184. From Ashkenazic areas only four books have survived, all of them late twelfth century and unlocalized. The earliest is dated 1177 (Ms Florence, Biblioteca Nazionale Centrale II-I-7 of Talmudic treatises). From Italy two manuscripts (written by eight copyists) have survived from the late eleventh century, the oldest dated 1072/73 (Ms Vatican Ebr. 31) and two from the first half of the twelfth century; all of them are unlocalized. One manuscript of the beginning of the twelfth century is probably from Byzantium (Ms Leningrad, I Firkovitch A187, copied in 1111/12.19 One may expect and hope that full access to the rich Leningrad collections, whose dated manuscripts are so far only known from partial catalogues and handlists, references in literature and some microfilms (partly studied by the French team of our Project) will reveal more dated books. However this may be, the chronological deployment of the extant dated manuscripts confines medieval Hebrew palaeography and codicology, which like all palaeographies is based on the typology of dated codices, to the late Middle Ages, a time when Jewish culture had already become crystallized. An exception to this general rule is the Middle East, where earlier dated findings from the formation period are extant, as well as pre-tenth century undated fragments and books belonging to the beginning of the adoption of the codex form in the Arabic period.20 The geo-chronological distribution of the manuscripts shows that their chronological deployment is not even in all the areas. For instance, until 1350 there are more Ashkenazic manuscripts extant than Italian, while in the fifteenth and first half of the sixteenth centuries the number of manuscripts produced in Italy is much greater than that in Ashkenaz. The chronological distribution of the Oriental manuscripts is completely different to that of other areas, and unlike the others, it does not show significant growth. Both in the Orient and Yemen the number of extant manuscripts decreases in the first half of the fifteenth century. The statistics regarding manuscripts written by immigrant scribes and copyists are of great interest. These manuscripts are differentiated by the computer according to the script, which differs from that employed in the area of production. They amount to about 19% of the extant or studied palaeographical units. As might be expected, the only area where non-local types of script are found to a considerable extent is Italy. Manuscripts copied by immigrant scribes hardly exist in Sefarad, Ashkenaz and Yemen, but in Byzantium and the Orient such manuscripts, written mostly by Sefardic copyists, constitute a significant proportion (about 16% of the studied Byzantine units, and 14% of the Oriental ones). In general, most of the emigrating scribes were Sefardic, though quite a considerable part of them were Ashkenazic, active mostly in Italy. There are no emigrant Oriental scribes, and hardly any Italian or Yemenite ones. The chronological deployment of these manuscripts in Italy is illuminating. Apart from a few manuscripts written in Sefardic and Ashkenazic scripts in the second half of the thirteenth and the first half of the fourteenth centuries, Sefardic and Ashkenazic scripts make a considerable and consistent appearance in dated manuscripts produced in Italy from 1396 onwards. Almost half the manuscripts dated from and after that year were written by immigrant Sefardic, Ashkenazic, and to a much smaller extent, Byzantine hands. Of 662 studied manuscripts produced in Italy between 1396 and 1500, 306 were written in non-Italian scripts; 204 of them in a Sefardic, 93 in an Ashkenazic type of script. In the fifty years preceding the expulsion of the Jews from Spain, the proportion of the studied manuscripts produced in Italy in Sefardic script grows to 34% of the total, while after the expulsion in 1492 it drops to 23%. Apart from the historical implications of these data on the intellectual activity of the Sefardic and Ashkenazic communities in Italy, their size and nature, or on the role of immigrant copyists within local Italian society (according to manuscripts written by immigrant scribes for Italian owners), these and similar data completely modify the proportions of the geo-cultural classification of our data-base as far as the types of script are concerned: To the manuscripts written in a Sefardic type of script should be added 351 extant manuscripts produced outside Sefardic territories (equal to more than half the extant manuscripts produced in Sefarad!); or 312 studied (almost two-thirds of the total studied). To those written in an Ashkenazic type of script, should be added 147 produced outside Ashkenaz (amounting to about one-third of the extant manuscripts produced in Ashkenaz); or 116 studied. From the total of Italian manuscripts 388 (322 studied) should be subtracted. Consequently, the quantitative relationship between Italian, Sefardic and Ashkenazic manuscripts is modified: The number of manuscripts written in a Sefardic type of script turns out to be the largest (1010 extant units, 851 studied); the number of manuscripts written in Italy in Italian script decreases from 1047 to 659 (512 studied); the number of the manuscripts written in Ashkenazic script approaches that of Italian (590 units, 504 studied); and to the Byzantine manuscripts 25 manuscripts are added (21 studied). Further, illuminating statistics are provided by the computer with regard to the subjects of the copied texts in dated manuscripts, (in the following table the total percentage in each area is larger than 100%, because one manuscript may contain texts belonging to more than one subject). subject distribution of the studied mss (percentages) total Sefarad Ashkenaz Italy Byzantium Orient Yemen Bible 26 31 28 24 21 23 31 Bibl.commentaries 14 14 8 17 20 14 5 Halakha & Midrash 28 22 46 19 23 34 59 Liturgy 10 4 16 14 2 10 3 Sciences,Linguist. 14 17 6 14 20 9 14 Philos.& Kabbala 21 25 8 25 36 12 2 Literature 2 1.5 0.3 2 3 3 0 Varia,Compilations 4 4 3 5 4 5 2 Though some subject distributions are similar in all areas, such as biblical manuscripts, there are significant differences between areas in other subjects, such as philosophy and Kabbala or Halakha and Midrash. The chronological deployment of these statistics reveals some interesting data. In all regions the proportions are different in the earlier periods. The number of biblical manuscripts is much greater up to 1300, when they constitute 62% of the studied Ashkenazic manuscripts, 50% of the Sefardic and 48% of the Italian.21 The ratio of Halakhic and Midrashic books in Ashkenaz before 1300 is only 21%, while the ratio of philosophical and Kabbalistic books in Italy in the sixteenth century (until 1540) increases to 36%. In Yemen the percentage of Halakhic and Midrashic copies in the fourteenth century comprises 80% of the total studied, while that of biblical commentaries in fourteenth century Oriental manuscripts reaches 32%. The data-base will serve and promote our project in three ways: 1. Yielding lists of dated manuscripts, scribes, original owners, localities of copying, and scribal formulas, as well as searching facilities according to such textual data and their combinations. 2. Assistance in setting up a precise and detailed codicological typology of the handwritten Hebrew book. In my book Hebrew Codicology I attempted to provide such a typology with regard to part of the codicological practices, but I emphasized the tentativeness of that effort in the subtitle of the book. The typology presented there was based only on partially automated data, and on manual processing. For the first time it is now possible to present a typology according to the entire classified data of the dated codices and their complex combinations. To be sure, the computer will not automatically provide us with a comprehensive typology. Luckily it will not relieve us from the need to analyse, understand, explain and expound processes, developments and conditionings, but it will assist us in processing and classifying the vast quantity of data which has accumulated and in revealing correlations. 3. We have created a so far unique and pioneering tool for scientific identification of the provenance and period of undated and unlocalized manuscripts. If one wishes to identify a Hebrew medieval book palaeographically, one records a selection of its determinant codicological features, such as sort of parchment, composition of quires, pricking and ruling techniques, layout devices, scribal formulas, etc., and requests the computer to sort out from the data-base all those manuscripts which share the same combination of features. The list will be immediately displayed on the screen, or printed out; any listing will include elementary data (date, locality and geographical identification, the scribe's name and whether he copied for himself or for somebody else, the type of script when it differs from that employed in the area of production, quiring and writing material). In addition, the inquirer will be able to compare the script of the manuscript with the natural-size bromide reproductions of selected pages of the manuscripts retrieved by the computer and to establish the identification on the basis of shared codicological practices and similarity of script. Before presenting the searching and retrieving possibilities of the data-base, I should like to point out two methodological difficulties which cast a shadow over this tool and its application. In this project, we in fact preceded and realized what Carla Bozzolo and Anzio Ornato in recent years have called "experimental" or "quantitative" codicology.22 In fact, years before Latin palaeographers, we applied sociological methodology in studying handwritten books, regarding human beings not as individuals, but as members of a group, to a large extent conditioned by the social structure of their surroundings. Like sociologists we observe groups of individuals that possess common features, and as in experimental procedure, we measure and record all the variants of phenomena in different populations. This social and quantitative methodology raises questions with regard to the extent of representativeness and the validity of exceptions or irregularities. Our data-base covers most of the extant dated manuscripts and will eventually include all of them, yet it should be regarded as representing a small sample. Indeed, while dated manuscripts constitute probably about one-tenth of the surviving Hebrew late medieval copies, they surely represent only a very small part of all the Hebrew books produced over the period of some 650 years of the extant dated codices.23 Furthermore, less than half the manuscripts in the data-base are localized, while the geo-cultural provenance of the rest was suggested by us on the grounds of the type of script and the codicological typology which has been derived from the localized manuscripts. If we are in danger of being trapped in a vicious circle of identification on the basis of false assumptions, we will have to reduce the data-base as a tool for palaeographical identification and exploit only the dated and localized manuscripts. Moreover, as the statistics have demonstrated, the number of manuscripts dating before 1250 is rather small, and their geographical distribution is not even.24 Thus, to what extent do our data represent the many-faceted reality? Are we permitted to draw up a firm typology or identify manuscripts on the basis of such a sample?25 Despite the limited sample we can safely identify or even attribute validity to a law when a certain practice is found to occur only in manuscripts produced in a certain area or period, particularly when the number of these manuscripts is large. Surely we are entitled to do so when such an exclusiveness is found in a combination of several codicological practices. But what about a practice which is employed by most of a certain population and by a very small part of another one? Undoubtedly we can use such a practice for characterization, but not so much for decisive identification. The power of statistics is undermined when we face even one exception, and one may not apply probability methods in palaeographical identification, since there is always the possibility that the identified manuscript is similar to the exceptional dated one.26 Nevertheless, we can obtain relief from this limitation by basing our identification on shared combinations of practices, and also by adopting negation and elimination procedures in searching and identifying. The following are examples of the searching and retrieving facilities of the data-base, and some illumating demonstrations concerning identification. Plate 1 presents the main menu and some searching screens. Plates 2a-2c show samples of a full record of a studied manuscript. Plate 3 shows samples of a statistical search divided into centuries. Plate 4 shows a print-out of a search for manuscripts produced in one locality (Toledo, Spain). Plate 5 shows the results of a search for the occurrence of a rare name ("Yehosef") among scribes and owners. Plates 6a-6b show the results of a search of the occurrences of the scribal ending formula k'vodha H'. One realizes that this formula was employed only in Italy, and not before 1280.27 Plates 7a-7d present the overwhelmingly convincing results of a search for the employment of a codicological feature, the ruling technique of drawing the horizontal lines in ink and the vertical boundary lines in pencil: The technique, employed in 104 manuscripts (until 1500), was not practised before 1421, and is confined to Italy, where it was mainly employed in the north (according to the localized manuscripts). The data-base reveals that this codicological practice has an exclusive palaeographical significance, both chronologically and geographically, as it is attested by a considerable number of manuscripts.28 Plates 8 and 9 are examples of retrieval of combinations of codicological and scribal practices. Each result demonstrates an exclusive provenance and a limited duration of the occurrences of these specific combinations. Plate 8 shows the retrieving of manuscripts sharing the folllowing combination of ten book features: A. Writing material: 1. Paper. 2. No laid lines. 3. No chain lines. 4. No watermarks. B. Quiring: 5. Quires constructed of five sheets (ten leaves). C. Means for ensuring the correct order of the quires: 6. Signatures in Hebrew letters. 7. Placed at the beginning and at the end of each quire. D. Ruling 8. Ruling executed with a ruling-board, leaf by leaf. E. Lay-out Devices: 9. Last words which are bound to protrude over the left boundary line are written diagonally. 10.Downwards. The retrieved manuscripts represent an exclusive geographical provenance - that of Yemen, and a limited, though quite long, chronological duration between 1338 and 1509. Plate 9a shows the list of manuscripts which share the following requested combination of seven features: A. Writing material: 1. Parchment. 2. Flesh-side and hair-side completely equalized. B. Quiring: 3. Quires of four sheets (8 leaves). C. Ruling technique: 4. Pricking for horizontal lines in both inner and outer margins, folded quire by folded quire. 5. Pair of guiding prickings for special lines which are usually drawn across the entire page. 6. Ruling by pencil (plummet). D. Devices of layout: 7. Writing exceeding last letters at a distance in the margin. The result of such a complex search shows that this particular combination occurs only in manuscripts produced in Ashkenaz (according to the few localized ones - in Germany) between 1307 and 1468. Thus, the geo-cultural occurrence of this combination is unequivocally Ashkenazic, and its chronological range spans about a century and a half.29 Plate 9b demonstrates the possibility of producing a more detailed record of any list by printing out the full description of any selected field. This plate presents the full recorded data on each of the fields of the features searched in the above combination (with the addition of the subject of each manuscript). The application of the data-base for palaeographical identification is thus proved in many cases to be a precise tool for localizing the production in geo-cultural areas, a vital component in the identification of Hebrew manuscripts, while its effectiveness for dating is found to be limited, as can be expected. The wide or unlimited ranges of dating can be reduced or considerably limited by adding further features to the combinations, and, particularly, by comparing the script of the undated manuscript to the scripts of the manuscripts retrieved. Such a comparison which should also include those personal para-scriptual elements, such as graphic shapes of line-fillers, pen decorations, forms of tetragrams, etc., 30 not included in the data-base but recorded in the questionnaires may even reveal that the undated manuscript was produced by the same hand as one of the dated manuscripts which display the same codicological practices, and lead to a precise and complete palaeographical identification.