The Cartographic Context
of the Madaba Map

by Paul D. A. Harwey

According to the late Oswald Dilke, to whom we owe so much of our knowledge of maps of the classical period, the mid-sixth-century mosaic map at Madaba is "probably the best known example of Byzantine cartography". 1 This is undeniably true; but it is also an understatement. The Madaba mosaic is in fact the only detailed map we have from the Eastern Empire before the thirteenth century, the date of the earliest surviving manuscripts of Ptolemy's Geographia that are accompanied by maps. Other Byzantine maps known to us are drawings showing the whole world in schematic form or, in one case, a mosaic at Nicopolis in Greece that may represent Paradise surrounded by water. 2 But none of these present topographical detail, and to see the Madaba map in its cartographic context we have to look not to comparable Byzantine maps - they simply do not exist - but to maps from the earlier Roman Empire of the West.
These do exist; and to the few that actually survive we can add others known to us through medieval copies or from contemporary references. Just how far maps were an everyday part of life in imperial Rome, i.e., how far their use was generally understood, is a matter of current debate. Dilke considered that the later Romans were generally familiar with maps and used them widely for administrative and other purposes. 3 More recently, scholars have reacted against this view, and have suggested that the use of maps was restricted to particular occasions. Thus, in a thorough critique of Roman mapping, Kai Brodersen shows that literary passages that have been supposed to refer to maps should in fact be interpreted in other ways, while Richard Talbert suggests that Dilke's work, far from proving a widespread use of maps in the classical world, actually "strengthens the impression that the majority of Greeks and Romans had only the most limited use for maps". 4
This difference of opinion is an important one, with significant implications for our understanding of the development of cartography in general. It has significant implications too for our view of the Madaba map. All the same, it is a difference only of interpretation and emphasis. All would agree that the Roman maps we know are no more than a few stray survivors, the tip of an iceberg; what is disputed is how large that iceberg was, with the corollary that the sheer number of maps produced must point to the degree of map-consciousness, i.e., how far cartographic concepts were generally understood in the Roman world. Of all the maps drawn in the Roman period, the proportion known to us today, whether the originals or copies, must be tiny or indeed, if Dilke is right, infinitesimal. However, it is reasonable to suppose that those we know offer us a fair sample of the types of map that were produced, and give a true picture of the nature of Roman cartography even if they fail to reveal its extent.
It has long been held, from surviving examples, that the idea of drawing maps to consistent scale from measured surveys was known in the Roman empire. Brodersen now questions whether the surviving maps really prove this, 5 but their form and style point strongly to the concept of consistent scale, whether or not it was successfully achieved. Certainly, if the idea was known to the Romans it was then lost, and the scale-map had to be reinvented in the late Middle Ages. We know of two other societies that discovered the scale-map and also failed to pass the idea on: Babylonia in the second millennium B.C., and Han dynasty China, contemporary with imperial Rome. 6 That the Romans should have been technically and conceptually capable of drawing maps to scale is hardly surprising. Some achievements of Roman surveyors are well known: their work throughout the empire in laying out roads that run dead straight for many miles and in planning centuriated estates that cover vast areas of land with grid-pattern fields. Whether or not they actually drew maps to scale, the surveyors had unquestionably mastered one important aspect of scale-mapping: they drew maps in strict outline plan at ground level, resisting the temptation to show features in perspective or elevation.
The maps known to us that are based - whether to scale or not - on careful surveys are mostly carved in stone, which may have been the most durable medium that was used, but which is unlikely to have been the most usual. Two sets of fragments come from large maps that must have been set up for public display. One group, of over 400 fragments from Orange in Provence, comes from three maps dating from the first and early second centuries AD; they show the rectangles of centuriated fields, roads, drainage ditches and other streams which in places form a complicated pattern of channels, meanders and islands. 7 The other group consists of nearly 700 fragments of one of the most remarkable artefacts to survive from the Roman period: a map of the entire city of Rome, arguably intended to be at the scale of 1:240 throughout, and measuring, when complete, about 13 metres high and 18 metres wide. Produced between AD 203 and 211 on rows of marble tablets, it was affixed to the wall of the Forum Pacis at Rome and shows individual buildings, monuments, fountains, columns, possibly trees, all in outline ground-plan except the aqueducts, which appear in elevation. It is an extraordinary testament to the cartographic skill and sophistication of the Roman surveyors. 8

Their other surviving maps are much smaller, but show the same techniques and conventions; all are outline ground-plans. They include plans of an estate from Urbino, of three houses from Perugia, of workshops from Ostia, and some others, all carved on stone. However, the latest known plan in this tradition is drawn on parchment. It is a plan of a monastery, probably an ideal layout rather than a practical proposal, that was drawn at Reichenau early in the ninth century and sent soon after to the neighbouring abbey of St Gall, where it survives to this day. Like the third-century plan of Rome, it shows walls in simple outline, with the same tiny transverse lines at doorways. 9 After this the idea of drawing maps in outline ground-plan, let alone to scale, was lost for some five hundred years.
Some copies of the account of Bishop Arculf's journey to the Holy Land in 670 include outline plans of the Church of the Holy Sepulchre, the Church of Sion, the Church of the Ascension and a building at Nablus; their style is that of the Roman surveyors, but none of the manuscripts that include the plans is older than the ninth century; though they are not drawn to scale, they could well derive from originals that were, copied perhaps at several removes by draftsmen unaware of the concept. 10 One plan in the surveyors' style is a mosaic, of uncertain date and surviving only in part, that comes from Rome and shows the layout of a set of baths. 11 But though surveyors' plans are thus known and might take the form of a mosaic, there is nothing on the Madaba map to connect it with these maps. Features above ground are not shown in outline plan, it was certainly not meant to be drawn to scale and none of the conventions of a surveyor's map appears. But other kinds of Roman map are known, and it is to these that we must look for connections with the Madaba map.
One kind of Roman map we know only by deduction, for no specimen or fragment survives from the Roman period. This is the map of the world and its regions, drawn to scale from geographical co-ordinates. We have, certainly, the Geographia of Claudius Ptolemy, written at Alexandria in the second century AD, which lists co-ordinates for many places, but we have no copies of his book earlier than the twelfth century. While earlier origins have been argued, the sets of maps that accompany many of the later manuscripts seem to have been compiled at Constantinople about 1300 though, being based on his text, they differ significantly from other world maps of the eighth to fourteenth centuries. 12 Given that neither the administrative nor the cartographic techniques were available to draw up a new world map at that time, those other medieval world maps must derive from maps compiled in the Roman period. Whether they derive ultimately from a single survey or from more than one we do not know, nor whether there were other separate regional surveys on which maps were based. There are references to a survey of the world ordered by Julius Caesar in 44 BC and completed thirty-two years later in the reign of Augustus. But those references are late and untrustworthy, while the supposed map of Agrippa mentioned by Pliny and Strabo is more likely to have been a written text. 13 The surviving world and regional maps of the Middle Ages are the only convincing evidence we have for the existence of Roman maps of the world.
At first sight they offer an unflattering view of Roman world maps. Even those that are not purely diagramatic present the merest travesties of the geographical outlines that are familiar to us. But we must remember that they are probably the result of many successive copyings by draftsmen to whom the outlines were not familiar at all, who knew nothing of the concepts of scale or of co-ordinates, who at best were simply copying the shapes of the exemplar and who would have no objection to altering these shapes to fit a circular frame or a narrow page or just to make a vastly simplified agreeable pattern. In view of this, it is amazing that any medieval world map bears any relation at all to geographical reality. We can assume that any true outline to be found on these maps has somehow survived these processes unscathed from the ultimate Roman original; and that this original thus contained all the accurate features found on any of these derivatives.
It follows that the Romans had access to a world map that may well have been impressively accurate, at least for the parts of the world that lay within or near the empire. Even on the Hereford world map of the late thirteenth century, mostly a jumble of unrecognisable shapes, the triangular outline of Sicily is still preserved. On the Cotton map of the early eleventh century much more can be recognised: the south-west peninsula of Britain, the coasts of Flanders, Frisia and Jutland, the Adriatic and the Peloponnese, the Egyptian coast with the Nile delta. 14 Neither conceptually, technically nor administratively was the eleventh century remotely capable of surveying a continent or a region to produce a map. Insofar as these features correspond in any way to reality, this is because through sheer good luck they have survived, more or less unscathed, many successive copyings of classical maps by draftsmen who knew nothing of the principles of cartography and even less of the actual shape of these coastal outlines.
What inland features on medieval world maps derive from Roman maps can only be guessed, but we can be fairly certain that rivers and mountains do, and probably at least those towns that were important in the empire. Can we place the Madaba map alongside these much later medieval maps and see its outline, perhaps its principal inland features, as based on Roman world or regional maps, drawn to scale from tolerably accurate surveys? Did the Madaba map owe anything to maps of this kind? It may well have done, but its coastal outline is all but totally lost and the shapes of the Jordan and the Nile show that if this was the origin of its rivers, the process of distortion through inaccurate copying was already well advanced when it was designed.
If, however, we turn from the content of the map to its style we can speak with much more assurance in linking it with the third kind of map that we know was drawn in imperial Rome, the picture-map. Late-Roman treatises on surveying were illustrated not with reproductions of the maps that the surveyors themselves drew, ground-plans like the third-century carved plan of Rome, but with maps of real or imaginary countryside that show all features above ground level pictorially, as though viewed from the side or from a height: they include actual pictures of mountains, towns, bridges, milestones and so on. There is no attempt in the maps at consistent scale or proportion - features are simply placed broadly in relation to one another. 15

Maps of this kind were drawn in most literate pre-industrial societies. 16 From imperial Rome we have only a few, but in a variety of forms. Some are known only from later medieval copies. The earliest manuscripts of the treatises on surveying date from the sixth and ninth centuries, and alongside them we can place some map-based illustrations to the Notitia Dignitatum, an account of the administrative structure of the empire in the fourth century. 17 More striking is the magnificent Tabula Peutingeriana, a twelfth- or thirteenth-century copy of a map of the road system of the entire empire. It takes the form of an elongated strip, some seven metres long, so that its regions are drawn neither to scale nor to shape. However, the roads and places along them are all in correct relationship, so that any particular route can be followed accurately, while the towns, staging-posts, mountains and forest are all shown by tiny pictures; most are conventional, but some, like the port at Ostia, are drawn from life. 18 Written itineraries or, indeed, a route-map were the probable source of the non-biblical names on the Madaba map, but these are set in a more orthodox cartographic framework than the Tabula Peutingeriana. 19
The few original picture-maps to survive from the Roman period are mosaics. One that is of a real place is in the Forum of the Corporations at Ostia and dates from the second or third century AD. It shows a river dividing into three to form a delta - or, possibly, flowing in the other direction, joined by two tributaries - with, across the main stream, a pontoon bridge on three ships. It may represent the mouth of the Rhône, as the mosaics around the forum relate to the merchants who occupied the buildings and one specifically for those from Narbonne is close by, but merchants from Alexandria were there too, so it may perhaps be the mouth of the Nile. 20 Cities of Palestine shown on eighth-century mosaics at Umm al-Rasas in Jordan, are all viewed from the side, but in the fifth- and sixth-century mosaics in the church of Santa Maria Maggiore at Rome there are more map-like bird's-eye views of Jerusalem and other cities, mostly shown as hexagonal walls with buildings inside. 21 Other picture-maps on Roman mosaics are more likely to be of imaginary or mythical sites. One of the second century at Conimbriga, in Portugal, is a plan of the minotaur's labyrinth; around it is a wall with eight towers, all in elevation. 22
Whether what they show was real or imaginary, whether they survive from the Roman period or only in later copies, in these picture-maps we are at once in the world of the Madaba map. Its outline of coasts and rivers may or may not have derived from geographic maps of the world or its regions, but within this framework everything is shown pictorially. Mountains, towns, villages, all appear as if seen from the side or from a height, and although they are broadly in the same relation to each other as on the ground there is no attempt at consistent scale. The pictorial effect is enhanced by the scenes added to the map by way of local colour: the trees, the fish in the river, the figures in a boat, the deer pursued by a lion. In all this the Madaba map is comparable with the picture-maps not only of imperial Rome but of many other societies as well. The way that a double horizon is introduced into its picture of Jerusalem, to show in elevation not one but both sides of the principal street with its rows of arcades, is a technique for which there is no known Roman parallel but which appears in many picture-maps from other sources: from Assyria in the seventh century BC, from fourteenth-century Japan, from fifteenth-century France, to give only a few examples. 23 It is to the widespread genre of the picture-map that the Madaba map belongs.
Roman maps of any kind are few in number, but they have been well studied and are indeed a subject of current debate. When we look for the relatives of the Madaba map not among Roman maps but among medieval maps of Palestine the position is very different. Whatever the final verdict on the use of maps by the Romans, we can be certain that they were far more familiar with them than people in medieval Europe. The Middle Ages were unfamiliar not only with local plans or geographic maps drawn to scale but with picture-maps and other maps of every kind. Certainly there was a fashion for encyclopaedic world maps in the twelfth and thirteenth centuries, and from the late thirteenth century maps of the Mediterranean and the Atlantic coasts - now called portolan charts - seem to have come into use for navigation. However, no other regular use of maps is known before the fifteenth century; when one was drawn it will normally have been an innovative expedient for a particular occasion.
All the same, we have several maps of Palestine as a whole, others just of Jerusalem, from the twelfth to the fifteenth century, some of them known in more than one copy or version. Though they have mostly been published and discussed as individual items, there has been little analytical or comparative work on these maps as a group - work which is badly needed and which would probably tell us a great deal about medieval cartography in general. It would explore the relationships, close or distant, between these various maps and also between them and the depiction of Palestine on contemporary world maps. This work, when it is done, will undoubtedly take the Madaba map as one of its principal points of reference. Almost certainly it will point to links with these later maps that only detailed investigation can reveal. All that can be done here is to mention a few outstanding medieval maps of Palestine which, however indirectly, are likely to be related to the Madaba map.
One is the Jerome map, so called because it is on the last page of a manuscript containing works by St Jerome - one a translation from Eusebius - on names in the Bible. On the other side of the folio is an accompanying map of Asia. No other of the many surviving manuscripts of those works contains maps, and there is no reason to suppose that Jerome himself had anything to do with them. The manuscript was written at St Martin's Abbey, Tournai, probably late in the twelfth century, and the maps were drawn at the same time, presumably to fill a blank folio in the book with interesting and relevant illustrations. The maps may have been copied in the first place from existing regional maps, but if so, they were closely related to world maps of the time and it may well be that the Jerome maps were copied directly from a world map. Soon after they were drawn the same hand altered both maps extensively, apparently to bring them in line with a different world map. But besides those successive versions of the maps of Palestine and Asia, there seems to have been an earlier version of the map of Palestine; it was erased to clear the page for the first version of the map of Asia, but sufficient traces can still be seen to identify it and to show that it differed significantly from the later - but only slightly later - successive maps of Palestine on the other side of the folio. On the Jerome maps, like the Madaba map, much work is needed before their relationship to other maps can be established. 24
Other maps of Palestine were drawn in the mid-thirteenth century by Matthew Paris, monk of St Albans Abbey, England. He was a distinguished chronicler and a more than competent artist, with a wide-ranging and inquisitive mind that led him to an interest in diagrams and maps. He produced two different maps of Palestine. Of one there are three versions in his own hand, all included in the manuscripts of his chronicles; its basis was probably a world map - we may suppose that his two-humped camel outside Acre is related to the two-humped camel outside Jerusalem on the Ebstorf world map - but with the addition of other drawings and long descriptive notes in French. The other is known from only one manuscript and is a quite different map, possibly the only medieval map of Palestine with north, not east, at the top; it has only a few short topographical notes, in Latin, but has many more towns and other features than the other map and includes the distances in day's journeys between some of the places along the coast. 25
Another extraordinarily interesting map of Palestine was compiled about 1320 by Pietro Vesconte of Venice to illustrate the Liber secretorum fidelium by Marino Sanudo; seven copies are known. The whole map is covered with a grid of squares, each, as a note explains, one league or two miles long. A list is given identifying the square in which each town is placed - as far as we know, an entirely novel idea at that time. The way Vesconte drew the mountains on the map is oddly reminiscent of the Madaba map. 26 Though a direct connection seems unlikely, it is a reminder that the antecedents of all these medieval maps of Palestine have yet to be elucidated and that the Madaba map may well prove an important element in the analysis. The various medieval maps of Jerusalem are likely to be part of the same picture, though here we can say with conviction that none has any apparent connection with the representation of Jerusalem on the Madaba map.
But here we are moving into uncharted waters, a challenge to some future researcher prepared to undertake a more complex piece of work than might appear at first sight. One final point should be made. In its design as well as in its execution the Madaba map is an elaborate and intricate work, involving, one would suppose, a good deal more preparation than a more conventional pictorial or patterned mosaic. Though we can only guess, it seems unlikely that once the design had been produced it was used only on a single occasion, for a single mosaic; and still more unlikely that that single mosaic would have been at Madaba. Far more likely, what we have at Madaba is a copy of a mosaic that was designed and produced at some other centre. Where this might have been leads us still further into speculation, but Jerusalem or even Constantinople comes immediately to mind. But whatever its origin, whether it stood alone or is the sole survivor of a whole family of mosaic maps, it is far more than a local curiosity; it is an important fragment of our cultural heritage from the ancient world.

The map of the Holy Land by Marino Sanudo (drawn in 1320)


1 The History of Cartography, p. 264.

2 ibid., pp. 260-4, 269.

3 ibid., pp. 201-57; Dilke 1985.

4 Brodersen 1995, pp. 261-87; Talbert, review of Dilke 1987, p. 211. Similar doubts are expressed by C. Radding in a review of History of Cartography, ed. Harley and Woodward, vol.1, in Annals of the Association of American Geographers 78 (1988), pp. 565-7.

5 Brodersen pp. 232-6.

6 Harvey 1980, pp. 122-5, 133-6.

7 Piganiol 1962.

8 Carettoni et al. 1960.

9 Fernie 1978, pp. 583-9; Horn and E. Born 1979; Parsons 1981, pp. 259-65.

10 History of Cartography 1, pp. 466-7, and sources cited there (p. 466, n. 9).

11 Carettoni et al., p. 209.

12 Polaschek 1959, pp. 17-18, lists the works where authors argue that the maps did, or did not, originate with Ptolemy himself; Dilke 1987 pp. 154-60, clearly sets out the evidence for thirteenth- and fourteenth-century origins of the Ptolemaic maps we know.

13 Brodersen pp. 261-87.

14 Harvey 1991, pp. 26, 29.

15 History of Cartography, i, pp. 217-20; Dilke 1967, pp. 9-29.

16 Harvey, pp. 48-120.

17 History of Cartography, 1, pp. 244-5; Alexander 1976, p. 11-25.

18 History of Cartography, 1, pp. 238-42; A. and M. Levi 1967.

19 M. Avi-Yonah 1954, pp. 28-30; Brodersen p. 150.

20 History of Cartography, 1, pp. 246-7.

21 Piccirillo and Alliata 1994, pp. 165-230; Oakeshott 1967), pl. IX. XIV, 49, 50, 53, 55, 56.

22 Bairrao Oleiro 1951, pp. 47-52.

23 Harvey, pp. 53, 81; Nihon Shoen Ezu Shusei jo (Painted Maps of Japanese Manors, vol.1.

24 Traces of the earlier version can be clearly seen in the detail from the map of Palestine reproduced in History of Cartography, 1, p. 329; the map of Asia is reproduced in Harvey, Medieval Maps, p. 72. I intend to discuss these maps in a forthcoming article.

25 Vaughan 1958), pp. 241, 244-7, pl. XVI, XVII; Harvey, p. 91.

26 Harvey, 1, pp. 475-6; Harvey, p. 79.

This contribution was first published in: The Madaba Map Centenary, Jerusalem 1999, 103-107

logo logo

Created Saturday, December 16, 2000 at 11:18:01
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