The Palaeography of the Madaba Map in the Light of Recent Discoveries
A Preliminary Analysis

by James Russel

Appendix: Computer Techniques Employed in Scanning (by Gail Fenwick)

From the moment of its first appearance a century ago the Madaba Map has stimulated intense scholarly interest on a broad range of topics and, as the programme of the Centenary Conference clearly demonstrates, this interest shows no sign of diminishing. One subject, however, is conspicuous by its absence in the vast bibliography of research on the Madaba Map that has accumulated over the years since its discovery. There has never been any serious attempt to analyse the form of the script employed by the mosaicists for the numerous explanatory texts inserted to identify the places illustrated in the Map. This neglect was perfectly understandable in the early stages of research on the Map, which naturally concentrated on its content and its significance in elucidating the topography of the Holy Land. As the scope of scholarly interest widened to examine other aspects of the Map, however, one might reasonably have expected some attention to be paid to the palaeography of the inscriptions, especially when comparative material began to appear at Madaba, Mt. Nebo and at other sites in the vicinity in the increasing numbers of new mosaics that contained lengthy inscribed texts. Yet this has not occurred, and the lack of any detailed treatment of the letter-styles of the Madaba mosaics generally, and of the Map in particular, remains a striking omission. When the subject has been mentioned at all it is usually treated in cursory fashion. Avi-Yonah, for example, dealt with the script in the briefest of phrases thus: "The script used in the map is the revived oval script," with a reference to one of C. B. Welles' categories of Byzantine alphabets for the inscriptions of Gerasa added by way of clarification.1 Detailed comment is confined to one single remark that notes the occasional appearance of "alpha with a broken middle stroke", a letter-form in fact nowhere to be found in the Map. By contrast, considerably greater space is devoted to the relatively trivial subject of scribal spelling errors in the Map, of which there are but a handful.

The Madaba Mosaic Map character breakdown (sample)

Yet, along with the illustrations that they describe, the texts are surely the dominant feature of the Map. With well over 2,000 characters preserved in whole or in part, the cumulative sum represents one of the largest concentrations of mosaic writing surviving anywhere in the Byzantine world. Moreover, if the supposition is correct that the Map was intended to instruct pilgrims in the biblical geography of the Terra Sancta, then we must presume that the words had a significance transcending that of most other texts to be found in the Christian churches of the region. The intrinsic importance of the subject, therefore, provides justification enough to submit the palaeography of the Madaba Map to detailed examination. Moreover, the appropriateness of attempting this is much enhanced by three recent developments which combine to reinforce the benefit to be gained from analysing the Map's letter-forms at this time.
The first of these three factors is the comprehensive photographic documentation of the Map made available since 1977 through the publication of Volume I (the Tafelband) of H. Donner and H. Cüppers, Die Mosaikkarte von Madeba. This splendid collection of plates must now be considered, second only to autopsy of the Map itself in situ or the replica of the Map newly created by the students of the Madaba Mosaic School, as the primary source for the study of the Map.2 The second factor, already mentioned, is the substantial increase in recent years in the number of mosaic floors excavated in Madaba itself and at other centres of the diocese, notably at Mt. Nebo and at Umm al-Rasas by Fr. M. Piccirillo and his colleagues. The admirably prompt and well illustrated publication of these discoveries in preliminary reports and their integration into broader treatments of the mosaics of the region as a whole in works such as Chiese e Mosaici di Madaba, The Mosaics of Jordan and, most recently, the publication of the mosaics of the complex of St. Stephen at Umm al-Rasas, provide the necessary corpus of comparanda for a detailed analysis of the mosaics of the region in general, including the writing-styles of their inscriptions.3 As a consequence, it is now possible to compile a comparative set of alphabets to illustrate the letter-styles of individual mosaic floors along the lines of those prepared by Bradford Welles for the Byzantine inscriptions of Gerasa and by Doro Levi for Antioch.4
The third factor, of a somewhat different order, is the recent development through computer technology of the process known as digital scanning. This makes it possible to scan photographs of the individual letters of inscribed texts at a high enough resolution to create extremely detailed images by highlighting the letters and reducing or eliminating extraneous background. The task of devising a system suitable for scanning photographs of the mosaics of the diocese of Madaba and creating the accompanying tables of alphabets (Tables 1-4) was accomplished by Gail Fenwick whose assistance and collaboration I acknowledge with profound gratitude.5 Her detailed account of the technical processes employed in scanning and the difficulties encountered in creating suitable images appears as an appendix to this paper. Some general observations on the efficacy of computer imaging in creating alphabets are nevertheless worth making at this point in the discussion.
While digital scanning probably offers no appreciable saving in time over the traditional hand-drawing of letters, its advantage lies firstly in its absolute, indeed one might say its mindless objectivity; secondly, in its capacity to produce an image that only an artist's tracing can match for accuracy, but coupled with the ease by which the images can be endlessly manipulated to any scale and even perspective that may be desired. These qualities can be best appreciated when images are presented at a larger scale than is possible in the accompanying tables, but even here the precise accuracy of the representation may be noted in the apparent fuzziness in the outline of many of the letters reproduced. This is not the consequence of any defect in the printing process, but is actually how the irregularities of the tesserae that define the shapes of individual letters appear when reduced to the small scale at which the alphabets are presented in the tables.
For the Madaba Map itself a total of 1233 characters was scanned, accounting for approximately two-thirds of the complete letters surviving in the text. Except for rare letters such as xi and psi, with only one occurrence in each case, the sample more than suffices to define the typical shape or shapes of each letter of the alphabet as represented in Table 1. The standard ligature ? for omicron-upsilon and the abbreviation K for kai are also included at the end of the alphabet because of their frequency, although these forms display little variation amongst the mosaics of Madaba. Since this study is primarily concerned with letter-forms, however, no consideration is given to other abbreviations or any forms of punctuation or diacritical marks such as the double dots above iota to indicate the consonantal version of that letter.
In reviewing the Madaba Map, we find that for about half of the letters, including some of the commonest, such as gamma, eta, iota, omicron, pi, rho and tau, there is no appreciable variation in shape throughout the entire Map. The remaining letters, however, do display sufficient differences in shape and proportion to justify classification into two, or even in five cases, into three distinct styles. A few of these variants exhibit ornamental touches, such as delta resting on a base that projects beyond the triangle, mu with diagonal bars terminating in a diamond or triangular wedge, omega composed of two separate hooks in place of a single central hasta, and the introduction of a wedge-shaped serif at the centre. The statistics that accompany the alphabet on the right of the table provide a convenient breakdown of the number of times each letter appears in the sample and the frequency of percentages of each variation of individual letters where these exist. At a glance it is obvious that the vast majority of letter-forms have a distinctly uniform and functional appearance that eschews any form of fussy ornament. As will become apparent from comparison with other alphabets, this stands in sharp contrast to the decorative letter-styles to be found in the inscriptions of some other churches of the Madaba diocese. Indeed the restraint and even austerity practised by the mosaicists responsible for the lettering of the Madaba Map seem singularly appropriate for advancing the presumed didactic purpose that the designers had in mind when planning the Map. Certainly the simple format, easily legible by those who viewed them, is well suited to the explanatory texts that accompany the vignettes and landscape features. The overall uniformity of the lettering, however, cannot conceal evidence of individual taste in the writing of the legends, probably reflecting the work of different hands. This is especially evident in the contrast between the cursive forms of alpha, (types 1 and 2), delta (type 2) and lambda (type 2), in each case the right leg projecting beyond the upper end of the left and the more square forms of the same letters; in the alternative forms of theta in which the cross-bar may be contained within the oval frame, or less frequently, project beyond it on either side;and in the more decorative forms of mu (types 1 and 3) and omega (types 2 and 3), already mentioned.

As noted previously, Piccirillo's recent publications, with the wealth of photographs and plans they contain, provide ample scope to place the Madaba Map within the larger context of the mosaic pavements discovered in Madaba and the surrounding countryside during the past century. Indeed it is probably true to claim that, second only to Antioch, the diocese of Madaba has produced the largest accumulation of mosaics with a well documented context from anywhere in the Roman-Byzantine world. Unlike Antioch, however, where the mosaics span nearly five centuries, the Madaba mosaics are concentrated within a period of around three centuries at most, with the large majority falling within the hundred years that extend from the reign of Bishop Cyrus at the beginning of the sixth century through that of Bishop Leontius who was in office during the first decade of the seventh century. By far the greatest advantage that the Madaba mosaics offer for the purposes of comparative study over those of other cities is the high proportion that can be dated reliably from the chronological information contained in the accompanying inscriptions. These dates can be very precise when the text includes, as it frequently does, not only the year in which the mosaic was executed, but also the month and indiction year. Even without such precise information, however, the prosopographical references to the bishop or priest in whose period of office the work was accomplished that appear in many dedicatory inscriptions can frequently be dated within a decade or two at most. The advantage of dealing exclusively with the actual texts that supply the chronological information rather than with the structures that housed them must also be stressed, for only rarely can there be any reason to question whether the lettering of a mosaic inscription actually dates from the time stated in the text.6 The structural relationship of the building with the dated mosaic that forms its floor, on the other hand, is frequently uncertain, since the date may refer to the addition of a mosaic floor at a later date or the replacement of an earlier mosaic removed to make way for its successor. This uncertainty can occasionally apply to the individual mosaic itself, not all of which may necessarily be of uniform date throughout, but may contain a panel added subsequently. While such additions are often detectible through inconsistencies in style, only rarely can an addition be dated precisely by the text of a dedication, as in the extension to the already existing nave mosaic of the Chapel of the Theotokos in the Wadi 'Ayn al-Kanisah at Mt. Nebo.7
These favorable circumstances have made it possible to compile a series of alphabets derived from mosaic inscriptions from various parts of the diocese of Madaba. Produced by the same scanning technique as employed for the Madaba Map, these alphabets are presented in Table 3. Of the nineteen alphabets selected, thirteen can be dated to the precise year when the mosaic was laid. In the case of Madaba, the date mentioned in the text is normally based on the establishment of the Province of Arabia on 22 March, AD 106.8 Eleven of the alphabets dated by a specific era are recorded in this fashion. A further two texts are dated by the Byzantine world era which fixed Christ's nativity on March 21, 5508 years after the creation of the universe. The use of this era was coming into fashion in the eighth century, which explains its appearance in only the two latest examples of the series, the Chapel of the Theotokos in the Wadi 'Ayn al-Kanisah at Mt. Nebo and in the latest mosaic floor of the Church of the Virgin at Madaba, dated AD 762 (anno mundi 6270) and February, AD 767 (anno mundi 6274) respectively.9 The remaining six alphabets can be dated approximately through the naming in the text of the bishop of Madaba in office at the time when the floor was laid. Since their names are also found in some of the texts dated to precise years through reference to an era, the span of the episcopal reigns at Madaba during the sixth and early seventh centuries can be determined within a decade or so.10 The examples selected for inclusion in Table 3 by no means exhaust the number of inscriptions suitable for inclusion, but some dated examples are omitted because the range of letters in the surviving text is too limited to produce a reasonably complete alphabet.11 A few others, including notably the latest inscription from the Church of St. Stephen at Umm al-Rasas, are omitted because the correctness of the date is suspect.12
The compilation of a series of dated alphabets as copious as this, especially when concentrated within such well-defined geographical and chronological limits, will undoubtedly inspire the hope that it can provide guidelines for dating the letter-styles of inscriptions for which chronological information is lacking in the text or in the subject-matter of the mosaic itself and its archaeological setting. The Madaba Map, the date of which continues to rouse controversy, is of course an obvious candidate for the application of palaeographical considerations.13 Before proceeding further, however, it is necessary to enter a word of caution. Anyone familiar with the palaeography of Greek and Latin inscriptions will readily concede that, with a few notable exceptions, letter-forms as a means of dating inscriptions must be assessed with great care, for, as Woodhead rightly notes, "the evidence is far less precise and secure than is popularly supposed."14 Yet scholarly literature abounds with the confident assignment of dates to inscriptions, largely, if not exclusively, on the basis of letter-shapes. It is easy to be deluded by the relatively rare instances where letter-styles can offer a reliable criterion of date, as in some of the epichoric alphabets of archaic Greece, which are now well enough documented to fix a date within fairly narrow limits; or when the adoption of a passing fashion in letter-styles can supply dates with remarkable precision, as is the case with official inscriptions containing examples of the new letters introduced into the Latin alphabet in AD 47-48 by the Emperor Claudius, but which were abandoned by the end of his reign in 54.15 These examples are by far the minority, however, and in many instances the chronological indications provided by letter-forms, even when there is a well documented wealth of data available for regionally homogeneous centres such as Athens and Gerasa, offer little prospect of dating a text to a period much narrower than a half-century or so. For this reason, it is wise to adopt
Woodhead's cautious attitude and recognize that "this criterion, so often used as a first resort, is much better left as a final refuge. ..... A robust conservatism cannot be too strongly recommended."16
The problem confronting the student analysing the Greek scripts of Late Antiquity is actually more acute than for earlier periods. As Mango has demonstrated in his study of Byzantine epigraphy, the script employed throughout the Byzantine world was remarkably uniform, remaining essentially unchanged for over six centuries, from the fourth century until around the year 1000.17 The Madaba alphabets are no exception. Despite their apparent variety, all of the alphabets illustrated in Table 3 conform to the palaeographic koiné of the contemporary Byzantine world. Nevertheless, within this uniformity of style, it is possible to detect evolving fashions at Madaba, especially in the overall shapes and proportions of the letters. To illustrate these trends in action, it may be useful to take the case of a single church with several well-defined building phases, with each of which is associated a dated inscribed mosaic text. The example chosen is the Memorial of Moses at Mt. Nebo with alphabets derived from each phase presented on the left side of Table II. They cover a lengthy time-span, the earliest perhaps belonging to the late fourth century and the latest, the Theotokos Chapel from the reign of Bishop Leontius, to the beginning of the seventh century. Nevertheless, a sharp distinction may be drawn between the earlier two alphabets, which cover the fifth and early sixth centuries, and the later two, which belong to the very end of the sixth and beginning of the seventh century. Several developments may be noted; first the compression of curving letters such as epsilon, theta and sigma from broad rounded shapes to longer ovals; second, and related to the first, the narrowing of certain letters, accompanied by a general trend towards elongation. This is especially notable in gamma, lambda, mu, nu, and especially omega. Third, a revival of square capitals for alpha, delta, lambda and mu in the later scripts, at the expense of the cursive forms encountered in the two earlier alphabets; and fourth, an increasing penchant for adorning letters with decorative flourishes, a feature especially noteworthy in alpha, upsilon, phi and omega of the New Baptistery Chapel of 597/598, and in the alpha, upsilon and omega of the Theotokos Chapel dated in the first decade of the seventh century.
Ornamental accretions, however, need to be treated with caution, for they tend to affect only a limited group of letters and appear in too wide a range of variants to suggest any uniformity of taste or style. The entire range of decorative features can be seen in Table 3. They include relatively modest touches such as the triangular pendant wedges already noted that adorn alpha with broken bar and mu, the drooping serifs terminating the horizontal bar of tau, the application of a cross-bar through the hasta of upsilon, gamma, tau, phi and psi, the hooked outward curves that terminate the extremities of upsilon, psi and omega, the oblique stroke of nu interrupted by a circlet at mid-point, the decorative flourishes that adorn the cross-bar of epsilon and the linking of the loops of omega with a variety of ornaments.
More exotic examples include a second shorter cross-bar in tau and upsilon, the "bespectacled" phi with lozenge-shaped lenses in the New Baptistery Chapel of Phase 3 at Mt. Nebo, and a seemingly limitless variety of flourishes adorning omega.18 Indeed, it is hard to escape the conclusion that these features are largely matters of taste, reflecting rather the artist's wish to exploit to the fullest the potential for creating interesting ornaments that is possible when composing shapes with mosaic tesserae. When artists resort to decorative flourishes, therefore, the results more likely reflect personal idiosyncracy, the equivalent in fact of an artistic signature. Even in executing a single commission the individual mosaicist occasionally demonstrates considerable variety in applying decorative flourishes to his letters. This is evident by comparing the alphabets in the panels flanking the altar of the Church of St. Stephen at Umm al-Rasas. While a comparison of the alphabets on the left and right, illustrated in Table 2 (right), leaves little doubt that they are the work of the same artist, a far greater degree of decorative refinement is evident in several letters on the left-hand panel compared to those on the right of the altar. Note, for example, the prominent wedges and apices in epsilon, kappa, mu, nu, pi, rho, tau, upsilon, phi, psi and omega, the cross-bar in gamma, the curling extremities of sigma, upsilon, psi, and the inverted wedge that crowns the peak of theta. Several explanations for the contrast suggest themselves, lack of time necessary to maintain a consistently decorative style throughout the entire commission, a loss of interest in sustaining an elaborate quality, the need to accommodate the wishes of the patron for a simpler style of lettering, or the difference between the master's work on the left and that of an apprentice on the right. The ornamental touches in this case are incidental to the basic style of lettering. Only the underlying shape of the letter stripped of all decorative accretions can serve to define this artist's inscriptional style.
To set the writing of the Map in the larger context of the mosaic scripts of Madaba, the larger series of alphabets covering the entire diocese illustrated in Table 3 may be consulted. Here we note the same distinctive trends from broad, splayed and rounded forms in the earlier alphabets to flatter oval forms with a distinct tendency for letters to become narrow and elongated, especially marked by the end of the sixth century. This is coupled with a marked predilection for fancy flourishes, some verging on the exotic, beginning sometime after the middle of the sixth century. A summary of these trends, based on representative examples is illustrated in Table 4 where four alphabets are presented, though in fact the last two are variants of a single third category.19 The first two categories, which I have defined as "Broad-rounded" and "Oval-unadorned", correspond loosely with two of Welles's scripts from Gerasa, the Round and Oval alphabets respectively, but the third Madaba script, the "Elongated" with variants varying from simple to adorned, has no close parallels at Gerasa.20 Within this scheme the Madaba Map appears to conform closest to alphabets from the time of Bishop John, especially the Chapel of the Martyr Theodore in the Cathedral at Madaba of 562, which stands about the point when the transformation from the "Broad-rounded" to the "Oval-unadorned" alphabet is virtually complete, but without the slightest hint of any of the decorative flourishes that became fashionable during the reign of Bishop Sergius towards the end of the sixth century. This indicates a date fairly close to the generally accepted terminus post quem for the laying of the mosaic around the mid-sixth century.

Appendix: Computer Techniques Employed in Scanning and Creating the Alphabets of Mosaic Inscriptions.
By Gail Fenwick

The alphabets were produced by digitally scanning black and white photographs of the individual mosaics and manipulating the resulting images on the computer. The scanning was done on a Umax single pass scanner and the resulting images were acquired by Adobe Photoshop 3.0 running on a Pentium-133 MHz PC with 40 MB of RAM. Photoshop is a highly capable but very resource-intensive software package, and although it can run with as little as 16 MB of RAM, the minimum recommended is 32 MB. Ideally 64 MB should be available, as well as scratch space on the hard drive of up to ten times the size of the largest scanned image file. During the course of this project, it was found that at least 200 MB was necessary, but better results were achieved when up to 400 MB scratch area was available.
When scanning the image, several choices had to be made which would affect the quality and usefulness of the resulting file. The most important was the choice of image resolution. Resolution refers to the amount of information stored for any image, measured in pixels per inch (ppi). The resolution and the dimensions of the image determine the file size of the document. The resolution of an image is critical in determining the quality of printed output. Images should be scanned at the highest resolution capable by the printer to be used for the final output. If the resolution is too low, the PostScript language used by the printer will use a single pixel's colour value to create more than one half-tone dot. This results in pixelization, or very coarse looking output. If the resolution is too high, the file will contain more information than the printer needs and processing time by both the computer and the printer will be increased appreciably. It is important to know at the start of the project how the images will be presented at the end, since resolution can vary from as little as the 83 ppi (6,889 pixels in a square inch) capable of a standard computer screen to as much as the 2,400 (5,760,000 pixels in a square inch) or higher, which is the capability of high-end image setters. As the number of pixels recorded in the computer file is the square of the resolution, and the size of the file is directly related to the amount of information being stored, a wrong choice in image resolution can have negative effects throughout the entire project. The choice of 600 ppi for this project was based on the expectation that the final output would be printed on a HP LaserJet 4 printer with a maximum resolution capability of 600 lpi.
The scanned photographs were saved in grayscale colour as TIF files. The choice of file format is relatively arbitrary. Images can be acquired from the scanner in any standard file format, such as TIF, GIF, BMP, JPG, etc. During manipulation, Photoshop creates a new image in a native file format (i.e., readable only by the program which produced it) called a PSD file, which can be quite large. After modifications are made, the file should be saved in a compressed format, such as JPG.
The letters were then individually clipped out from the overall image and the background was removed by using Photoshop's "magic wand tool" which selects portions of an image based on the colour similarity of the adjacent pixels. The tolerance of the tool can vary from 0-255 in grayscale mode which uses up to 256 shades of gray to represent the image and each pixel has a value ranging from 0 (black) to 255 (white). A low tolerance selects colours very similar in value to the pixel clicked on, while a high tolerance selects a broader range. Experimentation with different tolerances resulted in the correct one to choose the majority of the desired letter while eliminating areas of the mosaic which in the photograph may have appeared to have the same colour as the letter, but in fact did not belong to it. The tolerance had to be repeatedly readjusted depending on the individual mosaic being worked on. The letters were made consistent with each other by filling the space selected by the "magic wand tool" with black and clearing the area outside of the space selected. The letter was then cropped and resized for consistency across all alphabets. When an image had to be enlarged (in this case to make the letters of inscriptions of differing original sizes consistent with each other), Photoshop creates additional pixels by the process of interpolation. There are three kinds of interpolation. For this project, bicubic interpolation was used. This produces the most precise image but is very slow. For this reason it was more advantageous to scan images which it was known would have to be enlarged at a higher resolution and then resize the letter in the same step as the resolution was reduced, thus producing the least amount of arbitrary creation of information by the program.
The use of Photoshop's "magic wand tool" to distinguish the letter is very objective. The computer is unable to distinguish intuitively those parts of the image that belong with the letter and those that do not belong. It can only distinguish the brightness value of the pixels. Depending on the quality of the original image, this sometimes resulted in a letter being chosen incomplete if the tesserae composing the letter were of inconsistent quality. It also resulted in the choice of pixels which did not belong with the letter when tesserae around the letter or the cracks between the tesserae appeared to be of a similar colour as a result of wear or deterioration of the mosaic. Some retouching of the letters was required in order to correct the "over-objectivity" of the computer.
In summary, while digital scanning offers little, if any saving in time over the traditional hand-drawing of letters, its advantage lies in its unequivocal objectivity and its capacity to create an image comparable to an artist's tracing in its authenticity. In addition, however, digital scanning has the superior merit of enabling the images to be adapted with relative ease and in limitless variety to whatever uniform scale and perspective may be required.


1 M. Avi-Yonah, The Madaba Mosaic Map (Jerusalem, 1954), 27, n. 78.

2 The replica, constructed from the same local materials as the original, is a meticulously accurate copy of what remains of the Map. It was exhibited on the opening day of the Conference in the Balle Hall of the Royal Cultural Center, Amman. The raised platforms and excellent lighting conditions provided viewing conditions far superior to those that the visitor to the original Map encounters in the present church of St. George at Madaba.

3 M. Piccirillo, Chiese e Mosaici di Madaba (Jerusalem, 1989), The Mosaics of Jordan (Amman, 1993); M. Piccirillo and E. Alliata, Umm Al-Rasas: Mayfa'ah Vol. I, Gli Scavi del Complesso di Santo Stefano (Jerusalem, 1994).

4 C. H. Kraeling, Gerasa, City of the Decapolis (New Haven, 1938), 355-368, especially figs. 14-16; D. Levi, Antioch Mosaic Pavements 2 vols (Princeton, 1947), 627-629.

5 Ms. Fenwick's participation in this and other projects under my direction has been funded by a generous research grant from the University of British Columbia Hampton Fund.

6 A notable example is the dedicatory inscription in front of the bema steps in the Church of St. Stephen at Umm al-Rasas. See note 12 below.

7 M. Piccirillo, Le Due Iscrizioni della Cappella della Theotokos nel Wadi 'Ayn Al-Kanisah - Monte Nebo, LA 44 (1994), 521-530.

8 For a convenient discussion and table of equivalents for the era of Bostra (province of Arabia) for the period 312-613, R. E. Brünnow and A. von Domaszewski, Die Provincia Arabia, vol. 3 (Strassburg, 1909), 303-307.

9 The much disputed interpretation of the enigmatic first digit in the date of the dedicatory inscription of the Church of the Virgin at Madaba has been convincingly resolved through the appearance of the same character in an inscription discovered in a monastery at Ramot near Jerusalem where it was recognized as ?, the letter stigma, equivalent to 6, with a diacritical appendage signifying a multipland of 1,000, the figure thus representing 6,000. A date of this magnitude could only refer to a chronology based on a creation era. In the case of Ramot the creation era employed was that of Alexandria, based on the date of Christ's birth at 5492 years after the creation. The date specified, 6254, was thus equivalent to AD 762. This coincides with the fifth year of the indiction cycle also mentioned in the text and is further corroborated by the reference in the text to Theodorus, the patriarch of Jerusalem, known to have held office around that date. (R. Arav, L. Di Segni, A. Kloner, An Eighth Century Monastery near Jerusalem, LA 40 [1990], 313-320).
Of the two Madaba examples employing a creation era, the case of the Chapel of the Theotokos in the Wadi 'Ayn al-Kanisah on Mt. Nebo is unequivocal. The date ? CO, i.e. 6270, is associated in the text with the fifteenth year of the indiction cycle, a coincidence accommodated by the Byzantine creation era, but not the Alexandrine. The corresponding date in the Christian era would be 762. The correctness of this date is reinforced by reference in the text to the reign of the Bishop Job, who is known from an inscription in the Church of St. Stephen at Umm al-Rasas to have held office in 756, only six years before. (L. Di Segni in Piccirillo, LA 44 [1994], 531-533, see n. 7 above). The discovery of this inscription and the one from Ramot confirms the significance of the same character in the dedicatory inscription of the Church of the Virgin as equivalent to 6,000, and thus the opening digit in a date based on a creation era. The resulting date of 6074, whether AD 567 [Byzantine] or 582 [Alexandrine], however, does not correspond with the fifth year of the indiction mentioned in the text, and the reigning bishops in 567 and 582 were John and Sergius respectively, and not Theophanes named in the Church of the Virgin. Di Segni offers a plausible solution by suggesting that the digit sigma with the value of 200 has been omitted in second place in the date and that the correct date should read ? COD, i.e. 6274. Applying the Byzantine creation era apparently preferred at Madaba, to judge from the al-Kanisah text, we obtain a date of 767, which brings the Church of the Virgin in line with the other two eighth century inscriptions, produces the required correspondence with the fifth year of the indiction cycle, and places Theophanes as successor to Job as reigning bishop at Madaba (L. Di Segni, The Date of the Church of the Virgin in Madaba, LA 42 [1992], 251-257).

10 To the bishops of Madaba listed by Piccirillo (Chiese e Mosaici di Madaba, 319-323) may be added the name of Bishop [Ma]lechios which appears on the mosaic of the central church at Mekawer. His date is unknown, but see note 19 below. M. Piccirillo, Lo Scavo della Chiesa del Vescovo Malechios a Mekawer, LA 40 (1990), 466-468; Le Antichità Cristiane del Villagio di Mekawer, LA 45 (1995), 297-311.

11 One noteable omission from Table 3 is the inscription at the centre of the courtyard of the Cathedral at Madaba which is securely dated to year 470 of the province of Arabia (AD 575-576) in the reign of Bishop Sergius. At the time when the alphabets were being composed the only photograph of this text available was not of suitable quality for digital scanning, P-L. Gatier, Inscriptions de la Jordanie, vol. 2 (Paris, 1986), Pl. 27, no. 135.

12 The inscription occupies much of the width of the nave in front of the bema steps. The date mentioned in the last line of the text, XP (680) of the era of the province of Arabia, i.e. AD 785, however, does not correspond with the second year of the indiction cycle also mentioned in the text. The explanation for this discrepancy lies in the fact that the letters of the date had suffered serious damage at some point and were subsequently replaced. It is evident from the erroneous repair of the damaged text in the preceding line that the craftsman responsible for the repair was both illiterate and ignorant of the original text. This accounts for the meaningless letters in the restoration of the damaged portion of the previous line, and reinforces the likelihood that the date in the last line was also restored wrongly in the course of its repair. R. Schick has proposed the date XGI, 613 of the province of Arabia, as the date that appeared in the original text, which does correspond with the second year of the indiction cycle, being equivalent to 718-719 of the Christian era. While there is general consensus that the 785 date recorded in the text is a scribal error, the certainty of Schick's proposal of 718-719 as the date originally recorded in the text before the damage occurred, however attractive, cannot be verified. For this reason this text is not included amongst the alphabets in Table 1. R. Schick, Christianity in the Patriarchate of Jerusalem in the Early Abbasid Period, 132-198/750-813, Bilad al-Sham During the Abbasid Period: Proceedings of the Fifth International Conference in the History of Bilad al Sham, 4-8 March, 1990 (Amman, 1991), 63-80, especially 75-78; The Christian Communities of Palestine from Byzantine to Islamic Rule (Princeton, 1995), 472-473; Piccirillo and Alliata, Umm Al-Rasas I, 244-246.

13 For a brief summary of the range of opinion on the Map's date, Gatier, Inscriptions de la Jordanie, 179. A terminus post quem of 542, date of the consecration of the Emperor Justinian's Basilica of the Nea Theotokos in Jerusalem, which is almost certainly depicted in the map, is generally accepted. The majority of scholars appear to favour a date in the second half of the sixth century, but there is support for a date in the first half of the seventh century. For a recent discussion proposing an early seventh century date, P. Doncel-Voute, La Carte de Madaba: Cosmographie, Anachronisme et Propagande, RB 95 (1988), 519-542.

14 G. Woodhead, The Study of Greek Inscriptions (2nd ed., Cambridge, 1981), 62.

15 Quintilian, Inst Or, 1.7.26; Tacitus, Ann 11.13.3; 11.14.5; Suetonius, Vita Claudii 41.3.

16 Woodhead, Greek Inscriptions, 62, 66.

17 C. Mango, Byzantine Epigraphy (4th to 10th centuries), in D. Harlfinger and G. Prato, Paleografia e Codicologia Greca: Atti del II Colloquio Internazionale (Berlin-Wolfenbüttel, 17-21 October 1983) 2 vols. (Alessandria, 1991), 242-243.

18 In some cases the decorative omega in the abbreviated form of osiw(tato) includes the mark indicating the abbreviation.

19 The existence amongst the mosaic inscriptions of Madaba of another category of alphabet not represented in the list presented in Table III may be noted. It is distinguished by its preference for squared capitals, especially epsilon, theta, omicron, sigma and omega with straight rather than rounded configuration. No example of this alphabet has been included because none of the texts displaying these features contains sufficiently precise chronological information. It appears in the south aisle of the reconstructed basilica of the Memorial of Moses on Mt. Nebo (S. J. Saller, The Memorial of Moses on Mount Nebo, 2 vols. [Jerusalem, 1941], 264-265, Pl. 118.2; Piccirillo, Chiese e Mosaici di Madaba 160), in the monastic complex of ed-Deir at Ma'in (Piccirillo, ibid., 242-245) and in the Church of Bishop Malechios at Mekawer (Piccirillo, LA 45 [1995], 297-311, see n. 10 above). The quality of execution of the scripts represented in these three buildings varies considerably, from a neat and careful script at Ma'in to the clumsy and irregular lettering at Mekawer. Even from a superficial inspection, it is clear that they have little in common except for a common predilection for squared letter-forms. They probably differ in date also. Piccirillo's proposal of an early date for Mekawer, second half of the fifth or first half of the sixth century, is plausible. Internal evidence from the text suggests a mid-sixth century date for ed-Deir, and the text from Mt. Nebo presumably dates from the reconstruction of the basilica dated around 600.

20 Kraeling, Gerasa, 366-367, figs. 15-16.

This contribution was first published in: The Madaba Map Centenary, Jerusalem 1999, 125-133.

logo logo

Created Saturday, December 16, 2000 at 11:29:07
by Eugenio Alliata ofm in collaboration with Stefano de Luca ofm
Webmaster: John Abela ofm - Space by courtesy of Christus Rex
copyright - Studium Biblicum Franciscanum - Jerusalem 2000