Select review from JRA 14 (2001)


Sitting down with the Barrington Atlas

Susan E. Alcock, Hendrik W. Dey and Grant Parker

RICHARD J. A. TALBERT (ed.), BARRINGTON ATLAS OF THE GREEK AND ROMAN WORLD (Princeton University Press, Princeton). Atlas: pp. xxviii + 102 (i.e., 204); Map-by-map directory: 2 vols., pp. xxi + 1383 or 1 computer optical disk. ISBN 069103169x; 0691049459. $325.


Clown: But for me, I have an answer will serve all men.

Countess: Marry, that’s a bountiful answer that fits all questions.

Clown: It is like a barber’s chair that fits all buttocks, the pin-buttock, the quatch-buttock, the brawn buttock, or any buttock.

(All’s Well That Ends Well II.ii.13-19)


This pungent Shakespearean quotation was employed by Canadian novelist Robertson Davies (in his Salterton trilogy) to evoke the manifold rôles and multiple requirements laid heavily upon a newspaper. Atlases cannot be far behind in the varied expectations and ever increasing demands placed upon them. Precisely what should be mapped? Which categories of data included? What size, what format, what audience, what price, what purpose? Each question raises its own hydra-headed assemblage of options, each with its vociferous advocates and caustic critics. These fearsome logistics must explain, at least in part, why mapping the classical world passed into such a parlous state in the 20th c. The American Philological Association’s ad hoc Committee on Basic Research Tools pulled no punches in their 1980 report:

We come, finally, to an area of extremely great importance, where the state of our tools is utterly disastrous, cartography. There is hardly anything more important to understanding ancient history than a clear conception of the terrain on which its events took place. But the best available maps ... are virtually unavailable, and nothing really useful has become available for most areas in the last few decades ...(FN 1)

Blunt acknowledgement of the situation’s severity gave birth to the work reviewed here, the Barrington Atlas of the Greek and Roman world, edited by R. J. A. Talbert.

*               *                *

In his preface and introduction to the Atlas, Talbert reviews in detail the history of the project: his first involvement (in 1988), its subsequent twists and turns, its technological challenges and choices, its many heroes (villains, in the form of defaulting contributors, go charitably unnamed). The final product actually appears in two parts: the Atlas volume itself (with a striking color cover adorned with N. Purcell’s rendering of Latium–Campania, Vesuvius mons very much to the fore), and a Map-by-map Directory, available both in CD-Rom format (included with the Atlas volume) and in a two volume hard-copy version (c.1400 pages). The Directory contains a list of every name and feature on each map, with basic information about its period of use, modern name equivalents, and some brief, key bibliographic information. The Atlas (with an internal gazetteer) appears intended to be able to stand alone, although users wanting detailed information about any particular locale will soon find themselves turning to the Directory.

The Atlas contains over a hundred maps (102, to be precise) at a limited number of different scales (3 at 1 : 10,000,000; 6 at 1 : 5,000,000; 47 at 1 : 1,000,000; 43 at 1 : 500,000; 3 at 1 : 150,000). Talbert is admirably clear about where the base maps came from, their advantages and disadvantages, and their conditions of production (xx and, on Mapquest Production Data, xxvii-xxviii). The extent of coverage was determined, quite simply, by taking in “all regions for which penetration by Greeks or Romans can be documented, thereby highlighting the vastness and diversity of their world” (xx). Decisions about representational scale were clearly fraught, and in the end the ‘core’ zones of Greece, Italy, and the Mediterranean proper receive more detailed attention (1 : 500,000), while ‘broader-brush’ depictions (1 : 1,000,000) were felt appropriate for the NW provinces, N Africa, and the eastern reaches of “Mesopotamia and beyond” (as M. Roaf’s bailiwick is poetically described). Athens, Rome and Constantinople (here, Athenae, Roma and Constantinopolis) and their respective hinterlands are given close-up treatment (1 : 150,000). To provide an overarching geographical sense, the Atlas begins with 6 large-scale overviews (at 1 : 5,000,000), one of which is a pull-out map of the entire Internum mare (map 1). At the end appear three different ‘versions’ of the Roman empire, showing provincial divisions at the death of Trajan; dioceses and provinces according to the early 4th-c. Verona List; and dioceses and provinces according to Hierocles in the late 5th/early 6th c. These last three, incidentally, are the only overtly ‘historical’ (or ‘political’) maps to be found in the Atlas.

Key-maps in the end papers are essential to the volume’s use. At the front of the book is the quick ‘flip reference’ to where a particular map sheet lies (inevitably the maps do not proceed in a strictly contiguous order, so one needs to check). The back end-papers delineate modern national boundaries (as of August 1999), with the organization of map sheets replicated atop of them. Talbert refers in passing (xxiii) to the inequities of available evidence — some countries are simply better explored and more accessible than others.  FN 2 Such ‘boundary effects’ on patterns of distribution have long been familiar to archaeologists, and users of the Atlas may want to keep an eye on where, in modern space and time, they are, as they assess the level of detail on display.

The Atlas strove, very consciously, to achieve a high degree of uniformity in presentation and to offer a “comprehensive vision” (xxi). Hence the restricted variation in map scales, the uniform choice of contours and colors, and the limited number of symbols employed; hence also some underlying, if not fully articulated, rules about just what to map. There was clearly a worry about ‘clutter’: “compilers were instructed to present significant features only, resisting the temptation to overwhelm their maps with less important data that could not be meaningfully accommodated” (xxiii). “Significance” emerges as the yardstick by which peoples, places and things were measured as meaningful, and thus mapped. Obviously, some filter had to be applied (no map can be all-embracing), but “significant” is a bald word that may stick in some craws. In the end, the same sorts of features are represented from map to map — physical geography, major settlements and monuments, roads, aqueducts, centuriation — a conformity which contributes to the strong programmatic sense of the Atlas. We return to some of the omissions from this “comprehensive vision” below.

In terms of labelling, it is clear that the Atlas had particular preferences: “So far as possible, the atlas limits itself to naming only those features (physical or cultural) to which ancient names from the time period covered can be attached” (xxv). For those sites where no original name was known, but which were still adjudged “significant”, modern names are used in a less conspicuous, sans sérif font. All in all, Latin is the language of choice, and users must internally translate their needs: for Lake Ochrid, for example, try Lychnidus Lacus.

Policies on symbols and periodizations are explained cogently by Talbert (xxiv-xxvi), although additional specific examples, cross-referenced between that text and particular map pages, might have clarified things slightly more quickly. Again, a concern about visual overload emerges. Only one symbol (out of some 20-odd possibilities, including settlement, tumulus, church/monastery, well, etc.) marks any specific locale (“normally for whatever is considered the most important of the possible choices”, xxv), even for those places which may have changed function dramatically over time. One site-identity thus rules over, or subsumes, all others. As for chronology, the Atlas works with 5 periods: Archaic, (FN 3) Classical, Hellenistic, Imperial, and Late Antique (each with its own color code: e.g., green for Archaic, purple for Hellenistic). Further subdivision would undoubtedly have been unwise. Indications of chronological activity on the maps themselves take a somewhat counter-intuitive, if internally logical, form. For any site securely identified as active in one period only (or if a name is attested in one period only), the name is underlined in a period-specific color. Thus, the Iceni (map 8) are underlined in bright Roman red, as is Claudianon, at Mons Claudianus (map 78).(FN 4) Conversely, for any site where activity is agreed to extend over more than one period, no additional information is presented (except in the Directory). In essence, chronological data in the Atlas proper are included in inverse proportion to the original importance or longevity of a site — the more we know, the less we see. Given the multi-period occupation typical of “significant” sites in the Mediterranean world, this certainly works to minimize the amount of labelling and lining on the map sheets, but it also raises serious issues for recognizing patterns or for charting chronological change in the landscape.

The Atlas extended its brief to features not invariably included, but very welcome, in ancient atlases. Road-systems pop up clearly; their integration into maps with contour lines and settlements (rather than flat and empty representations) is illuminating. The same holds true for aqueducts. Seeing centuriated patterns, in all their checker-board splendor, is outstanding and, again, an eye-opener when presented as one element in a more general mapping of terrain. A glance, for example, at map 44 (Latium–Campania) or map 67 (Antiochia) clearly demonstrates how regions could be significantly altered through the imposition of roads, aqueducts or centuriation. The contextualization of these and other ‘special features’ (including mines/quarries, water wheels, salt-pans, tunnels, reservoirs and lighthouses) is a real treat, and a genuine contribution to our understanding of the ancient landscape.(FN 5)

As one leafs through the Atlas, and turns to the lists provided in the Directory, one is struck by a solemn awe at the amount of sheer hard work that went into the project. Talbert does well to enumerate the many individuals involved, as well as the structure of command that made the effort possible. Below the editor (aided by his project team and a supportive University of North Carolina) were 10 carefully selected “vicars”. These men are clearly better envisioned in terms of Late Roman administrators than the clergy of Church of England parishes, for they controlled wide tracts of territory and no doubt had to rule with iron hands over the many compilers (73 in all, not to mention additional helpers) who actually generated data. The list of those involved is a veritable ‘Who’s who’ of ancient historians and archaeologists (indeed, it has been a standing joke that to find uninvolved individuals to review the Atlas would be no easy task). As a further quality-control, Talbert arranged a detailed reviewing process (again, an impressive collection of 95 people). In other words, reliable people worked on an atlas checked by reliable people in a commendably collaborative endeavor.(FN 6) One feels in good hands; you believe in the dots on the maps of the Barrington Atlas, and so you should.

Within the parameters set for it, the Atlas is a highly successful achievement. It is clearly and systematically laid out; it is, by its own lights, comprehensive and comprehensible. Beauty may be in the eye of the beholder, but many will find the color schemes and design of the volume very attractive. It is undeniably a handsome object and, while expensive, it remains affordable(FN 7) — for which thanks are due to the supporters of the project, both individual and institutional (though note there is no ‘Mr. Barrington’ — the eponymous community of Great Barrington, Mass., is the home of a principal patron). Talbert set such accessibility as one of his initial objectives, and — as with so many of his other goals — he has met the target. Without question, the field of ancient history and classical studies owes the editor a debt for this monumental labor of love.

*               *                *

On one level then, all buttocks will sit and fit in this barber’s chair, and be happy. But why are they sitting there? What is the Barrington Atlas FOR? Talbert speaks to this question relatively little: “This Atlas and its accompanying Directory finally provide an essential tool that everyone with an interest in classical antiquity has sought in vain for more than a century” (xvii); the Atlas and Directory “aim to provide every branch of classical studies and related fields with an essential tool that has been missing for over a century” (xix).(FN 8) The genesis of the project was apparently self-evident: to understand the ancient world, we need to have a sense of its geographical extent and of ‘where things were’; with tools currently in hand, we lack these badly; hence the Barrington Atlas.(FN 9) If you want to know where precisely the mines lay on Mount Pangaion; if you want to check the position of Artemis Brauronia in relation to Athens; if you have forgotten precisely where Fishbourne lay; if you wonder whether we have located the Tripyrgia of Xenophon, Hell. 5.1.10 (we have not), then the Atlas has the answer: in other words, it reliably locates significant places and puts them in a geographical and (to some extent) human context. This reference/research function is not to be sniffed at, and for many scholars it will suffice, especially when backed up by the more detailed chronological information and bibliographic references available in the Directory.

Others, wanting to do more with such a potentially powerful resource, might squirm a bit in the barber’s chair. The natural home for the Atlas is clearly the library or the office, not in the field or the large classroom. While the choice of size (13 x 181/4 inches) and of a bound volume (rather than multiple loose sheets) was sensible, it is not a readily luggable item. Nor is it the final answer to teachers’ needs for slides to show in their classrooms.(FN 10)  Although the Atlas will be tremendously helpful in one-on-one or small group instruction, the quest for good ‘teaching maps’ will continue.

Apart from such practical or functional concerns, more worrying conceptual issues arise. Maps have, of course, been exposed in recent years as non-transparent and problematic creations; all maps — in one way or another, aggressively or unconsciously — ‘lie’ to us. Using maps requires thinking about their agenda and their silences, and that holds true for the Barrington Atlas, the decisions taken in its formation, and the manner in which people will employ it. Can one use the Atlas to get an accurate sense of overall human occupation in a region? Is it helpful for obtaining a ready sense of changing human occupation in a region? The answer to these more open-ended questions is, on the whole, no. We note immediately that these are not questions the Atlas set itself to answer, involving data it did not set itself to master.(FN 11) Yet the decisions taken about those questions, and those data, do have potential consequences which must at least be acknowledged. In other words, there are hazards in unreflective uses of the Atlas — invisible dragons in its margins.

We can begin by returning to that word “significant”: not normally mapped, Talbert states, are the “findspots of coins, inscriptions, milestones, or pottery; battlefields; fords; kilns; movements of people; sea routes; shipwrecks” (xxvi). Subsumed in this list, of course, is the vast amount of new evidence from regional surveys in the Mediterranean world. Look at the territory of Metapontum (map 45), or of Messenia (map 58), or of the ager Tarraconensis (map 25) (just three better known out of the innumerable regions surveyed across the area mapped).(FN 12) All are here depicted as relatively empty spaces; Messenia especially is strikingly (and unnervingly) more or less a map of Pausanias’s travels.(FN 13) Yet each of these areas is known to have contained dozens, even hundreds, of places where people (of the relevant time-periods) lived, farmed, were buried, worshipped; in them we possess some sense of the local patterns of people’s lives. These maps should look as if afflicted by a case of the measles — and in that ‘clutter’, of course, lay the problem.(FN 14)

The disappointment is not so much with the decision to skip the ‘small stuff’. The Atlas is merely remaining consistent with its own goals, and it is true that to replicate survey data to scale in hard copy would be an ordeal (electronic formats, of course, will have no such excuse). What is disappointing is the manner in which the silence is explained or excused (or not). Presumably, “findspots of pottery” were simply considered insignificant, begging the question of just what is significant, and to whom.(FN 15) Many archaeologists and historians will smile wryly at parts of the Atlas, but there is danger here: since no real sense is given of just how much known human activity fails to be reported and depicted, users not already ‘in the know’, and who might take the Atlas at face value, will see the Greek and Roman world as a largely urban, ‘big site’ phenomenon. (FN 16) What a retrograde step that would be.

The limited presentation of chronological data in the Atlas proper also hampers certain types of analysis. One cannot glance at a map and gain an impression of regional change through time, since multi-period sites declare themselves only as multi-period (obviously, cross-referencing to the Directory would be possible here, if somewhat unwieldy). The restricted use of symbols also prevents nuanced pattern recognition (again, unless one turns to the Directory). One of the more remarkable things about the diachronic history of the ancient world is how places alter in character and function, yet such developments are lost in the decision to select one representative symbol per locale. It is interesting that while settlement types receive a certain degree of ‘breakdown’ (into settlement; estate/villa; fort/tower), many types of ‘special purpose’ site are conflated into one broad ‘catch-all’ category “temple/sanctuary/shrine/monument/tomb”. Given growing interest in Mediterranean ceremonial, mortuary and sacred landscapes, this is unfortunate, and one wonders why it was necessary.

Finally, there is, it is fair to say, a strong ‘classical’ bias to the Barrington Atlas of the Greek and Roman world. It is unashamedly self-declared: in its preference for ancient names, in the scales selected for representation, in the area covered (with a long extension stretching out to capture Ai Khanum [or Alexandria Oxiana? as it is called]), in the very title of the Atlas itself. What receives shorter shrift, inevitably, are the indigenous worlds “penetrated” by the Greeks and the Romans. Talbert explains: “That is not to say that, for the period of contact or involvement, only activity by Greeks or Romans is to be recorded, or only toponyms in Greek or Latin marked. Far from it. Rather, in order to give users an appropriate understanding of the landscape in which Greeks and Roman moved, the attempt has been made to record activity by other peoples at the same period, as well as names known only in their languages, although in a purposely selective fashion” (xxiv).(FN 17) Quite how to interpret “purposely selective” is not clear, but any perusal of the Atlas leaves little room to doubt just who and what receives the lion’s share of attention. Again, in logistical terms and given the disciplinary allegiances of the scholars involved in the project, this bent is understandable and (at least in some zones) unavoidable. But, as with the missing evidence of survey, that relative absence should at least be more prominently recognized — and recognized as undesirable. Otherwise, and without this at all being the work’s intention, the Atlas could feed a Eurocentric, colonialist view of the ancient world.

Its very professionalism, good looks and authoritative nature contribute to making the Atlas the admirable resource it is, yet these qualities cannot blind us to what it does and does not do, where it ‘lies’ to us. In other words, for anything beyond a straightforward ‘spot check’ of an ancient toponym, the user has to meet the Atlas half-way and with caution — as, of course, he or she should approach all maps. There can be no doubt that the Atlas will make a lasting impression, if perhaps in ways unanticipated by the project itself. The onus is now on us to become its intelligent users.

There will be users and there will be users, as Shakespeare’s Clown would have understood. The years in which the Atlas has been in preparation have brought with them ever more specialized needs on the part of classicists, and hence greater demands for complexity and detail. Whereas the APA report of 1980 spoke of an atlas in relation to the needs of history, landscape and mapping themselves have become hot topics in various guises. No longer is a map merely required to provide historians with a “clear conception of the terrain on which … events took place”: instead, more and more scholars have tended to see the environment as an active player in the events of the past, and particularly in the longue durée. Recent work, most notably P. Horden and N. Purcell’s The corrupting sea, has presented ecologizing visions of the ancient (and mediaeval) Mediterranean world, even as it fights the bugbear of environmental determinism.(FN 18) From this point of view it is impossible to underestimate the potential value of the Directory in making it possible to consider various facets of topography and ecology. It is in this thicket of details, rather than in the visual presentation of the Atlas itself, that we must recognize that topography is not merely the handmaid of (political or military) history; rather, it is a rich topic in its own right.

These considerations make one wonder to what extent this Atlas — indeed, to what extent any map — is a means to an end or an end in itself. To a significant degree, this is in the eye of the beholder: to focus on the aesthetic aspects of, say, the Vatican Map Gallery is no less important than to focus on the late 16th-c. power-relations in which its frescoes are imbricated.(FN 19) The choice depends, it might be argued, on one’s point of view. In the case of the Barrington Atlas, it is worth emphasizing that different users will find different atlases. The most obvious difference is that between the main map volume and its larger form which includes the Directory. It is the maps themselves that have the aesthetic element which allows the Atlas, for a wide range of users, to be an end in itself — an object just as attractive in a university library as on a suburban coffee-table. But it is the Directory that makes possible further inquiry on any detailed aspect of an area of the Mediterranean. Whereas the maps themselves have limited means by which to indicate change over time, it is the Directory that presents and nourishes further lines of historical inquiry. Text and image thus stand in an even more complex relationship than is usually the case in maps.

In general, maps are a certain kind of formula for presenting information about space by visual means, often combined with verbal. This Atlas is a product of the Information Age, insofar as its compilers were faced with far too much potentially “significant” information to incorporate into the maps themselves. Its current bipartite form surely makes it something of a transitional species, taking its form (for most readers) in a combination of hard copy and CD. In some ways, the hyperlink format makes it easy to present detailed information in a series of choices, with the possibility of zooming in and out at will. Yet the materiality of the book is still very much alive in our age, as Barnes and Noble Bookstore will be happy to tell us. The Barrington Atlas, in its current form, may thus be seen as the late blossoming of an older kind of hard-copy atlas, yet one which, through its partly digital preparation and partly electronic directory, looks forward to different maps of the future.

Maps will never grow less than complicated but, as we make more and more of them, load more data onto them, and have more ways to view them, the more they will become good servants, not bad masters. Talbert says himself that the Barrington Atlas is “only a beginning” (xviii), and he is absolutely correct. The timing of the Atlas project was in some ways unfortunate (though no ‘perfect time’ ever exists), spanning as it did so many early stages in electronic mapping and the digital revolution. It is impressive that the Atlas has stayed as current with technological developments as it has, and is continuing this advance (see the project web site at What this means, of course, is that there now exist (or are in the process of being made) digital base-maps for much of the ancient world, maps that individual scholars could rework and recast to their own ends (see already the ‘Interactive Ancient Mediterranean’ project: All the “open-ended” (xxi) possibilities that lay beyond the means, time or interest of the present Barrington Atlas could be pursued: the addition of sea routes, survey datasets, battlefields or trade routes — not to mention the basic generation of period-distribution maps. Above all, maps could be generated to specific purposes, responding to sharp questions, going beyond this brave attempt at a comprehensive vision, at a barber’s chair for all. What will truly revolutionize the mapping of the classical world will be the availability of these images in digital and interactive form — at prices we can afford, without let or hindrance. We are already far better off with the Barrington Atlas than we were without it — as Talbert says, it is “a giant step forward” (xviii). But even better days are on the horizon, as we rise and stand on the solid foundation the Barrington Atlas has provided.

University of Michigan


Much of our thinking in this review grew out of a graduate seminar jointly taught by Alcock and Parker at the University of Michigan (Space and Place in the Greco-Roman World), and we would like to acknowledge the other members of that class: B. Anderson, J. Davis, C. Hernandez, M. Hiers, D. Kertai, D. Ng, D. Shoup, and D. Wilburn. Above all, we would like to thank R. J. A. Talbert, who visited the seminar and discussed the Atlas in detail with us, as well as delivering an excellent lecture to the University of Michigan community and the Michigan Map Society. His generosity with his time and thoughts was much appreciated, and he kindly responded to both our praise of the Atlas and our reservations about some of its aspects.



1         R. S. Bagnall (ed.), Research tools for the classics: the report of the American Philological Association’s ad hoc Committee on Basic Research Tools (Chico 1980) 27; quoted by Talbert in the introduction (xix) to the Barrington Atlas.

2         As Talbert notes, “Of course, if we were only to allow ourselves to embark upon producing a comprehensive atlas of the Greek and Roman world once we considered our knowledge of the whole to be uniformly even, then in practice we might as well abandon all hope of ever beginning. Meanwhile, while it is true that some areas have been mapped in more detail than the present atlas attempts, the great majority have not, and certainly no comprehensive rendering has been achieved” (xxiii).

3         Talbert spells out the rough “start date” for the Archaic period (it varies across the span of the Atlas) on xxiv. This information should probably have been repeated in the key, where it merely says “pre-550 BC”.

4         This can lead to some difficult ‘calls’. For example, Talbert notes (xxvi) that aqueducts are presented as only belonging to the Roman period, even though the structure may have survived in later times. It is unclear why Constantinopolis and Thessalonica — with their notable post-Roman history — are both ‘Roman red’ cities, while Nikopolis (or indeed Roma itself) are not. In other cases, what seem to be errors creep in. For example, monumental basilical churches constructed outside Rome (S. Paulus, S. Petrus, S. Agnese) are listed in the Directory as both “Roman” and “Late Antique”, and thus receive no period indication on map 43 proper. Yet these churches (indicated by the church symbol) were clearly built after 300 — the cut-off point for the beginning of the Late Antique period. The apparent suggestion that Christians might be building large public churches, just outside of Rome, prior to Constantine is misleading. On this same map, the Aurelian Wall (constructed over roughly a 10-year period beginning in 271-272, with major renovations after 300) is marked specifically as Late Antique in date (see L. Richardson jr, A new topographical dictionary of ancient Rome [Baltimore 1992] s.v. Muri Aureliani).

5         The desire to depict “the physical landscape in its ancient aspect” is also noteworthy, though perhaps even more difficult to present (diachronically) than the Atlas suggests (xxiv). But certain categories of information given here are very helpful; for example, coastline change is an especially vital element in understanding settlement histories (see, for example, map 61, with its inset detail of Ephesus).

6         Acknowledgment is also due to the cartographic team involved at, Inc.

7         This was particularly the case during the widely-promoted pre-publication offer.

8         In his very helpful review article in JRA 5 (1992), “Mapping the classical world: major atlases and map series 1872-1990,” Talbert had remarked: ‘It is plain that a major classical atlas would be invaluable for reference, research and teaching’ (30).

9         One thing it was clearly for, of course, was to supplant previous classical atlases. Talbert’s evident awareness of past efforts in historical cartography is thus both necessary and commendable. Since editorial decisions concerning the Barrington Atlas seem often to have been made in response to features of previous projects (e.g., xxi: “this restriction is a deliberate reaction to the practice of TIR and TIB …”, referring to the Tabula Imperii Romani and the Tabula Imperii Byzantini, respectively), it is helpful to comment briefly on other past projects, to situate the Atlas in a cartographic tradition, and to gauge the extent of the void addressed in the APA report on Basic Research Tools.

           Among the several examples cited by Talbert in his preface (xx-xxi), the Teubinger Atlas des Vorderen Orients (Wiesbaden 1977- ; hereafter TAVO) and N. G. L. Hammond's Atlas of the Greek and Roman world (Park Ridge, NJ 1981; hereafter AGRW) represent opposite ends of the spectrum in terms of the size, scope, and level of detail which can reasonably be employed in a major historical atlas. The TAVO is a collection of over 400 large-format loose sheets (together comprising two exceptionally bulky volumes) covering the geographical extent of the Near East from the Neolithic to the present. Its sheer scale alone allows presentation of a vast range of information, not least because it permits the inclusion of chronological map sequences for all regions covered (thereby largely eliminating the problem of representing profound diachronic change on a single image). By contrast, the AGRW includes the full geographical sweep of the Mediterranean littoral, as well as representations of select battles, political alliances, administrative boundaries, and so on, in some 30 pages of maps. This results in a rather simplified standard of physical topography, largely confined to the presentation of basic contours, hydrological features, roads and settlements, all at a necessarily minuscule scale. The latter work unquestionably has its merits, many of which relate to issues of accessibility. The AGRW is easily within the price-range of most prospective purchasers, and for matters of quick reference it is a far less daunting prospect than, for example, the two massive volumes of the TAVO — as long, that is, as it happens to contain the desired information. On the other hand, the TAVO is unquestionably the superior scholarly resource, stimulating (as well as answering) questions about the areas within the scope of its coverage. Issues of cost and ease of handling aside, there is almost nothing in the AGRW which is not contained, nearly always far more comprehensively, in the TAVO, to say nothing of the sheer volume of data unparalleled in the former, or indeed in any other atlas of the ancient world. The Barrington Atlas sits, in both physical and conceptual terms, somewhere between these two extremes.

10       Shooting slides from a large, bound volume is inevitably not ideal.

11       Talbert acknowledges the plenitude of options available: “In whole or in part, the Greek and Roman world could be mapped in a multitude of ways. Naturally enough, this atlas is no more able than any other single book to fulfill every conceivable need in its field. It is, rather, a balanced, practical and relatively speedy attempt to begin overcoming a prolonged period of inactivity, and to rethink traditional approaches” (xxii). Omissions such as city plans, experiential views, or nautical maps may “prove cause for disappointment”, he concedes, but “such additional cartography would be just too different in nature, and too open-ended, in relation to meeting the need for a comprehensive vision” (xxi).

12       Metapontum: J. C. Carter, “Sanctuaries in the chora of Metaponto,” in S. E. Alcock and R. Osborne (edd.), Placing the gods: sanctuaries and sacred space in ancient Greece (Oxford 1994) 161-98; The chora of Metaponto: the necropoleis (Austin, TX 1998). Messenia: J. L. Davis (ed.), Sandy Pylos: an archaeo-logical history from Nestor to Navarino (Austin, TX 1998). Ager Tarraconensis: J.-M. Carreté, S. Keay and M. Millett, A Roman provincial capital and its hinterland: the survey of the territory of Tarragona, Spain, 1985-90 (JRA Suppl. 15, 1995).

13       The compilers of the map of the Peloponnese, J. Camp and G. Rogers, are explicit about this privileging of textual evidence: “No sites known solely from survey are marked here, and likewise very few known only from excavation... the map’s primary goal is to mark the location of the mass of places and features mentioned in the literary sources and inscriptions” (Directory, s.v. map 58).

14       This problem is not restricted to surveyed regions, of course. In Cappadocia (map 64), the practice of plotting only sites attested by literary references or significant architectural remains results in a map nearly devoid of human traces. The disparity between the number of points plotted and the actual distribution of population is revealed by the presence of cemeteries/catacombs in otherwise empty areas. For other discussions of this region, see W. Gwatkin, Cappadocia as a Roman procuratorial pro-vince (Princeton 1930) esp. 17-18; D. French, “The definition of territories: Cappadocia,” in B. le Guen-Pollet and O. Pelon (edd.), La Cappadoce méridonale jusqu’à la fin de l’époque romaine (Paris 1991) 49-59.

15       It must be said that claims to the use of up-to-date, “new research” (e.g., xviii) or to be “exploiting all available historical, epigraphic, and archaeological data” (xxiii) ring rather hollow.

16       Publicity flyers produced by the publisher for the Atlas are especially suspect on this point: “The Barrington Atlas recreates the Greek and Roman world in its entirety spanning more than a thousand years... .”

17       Interestingly, it is outside the principal area of Greek and Roman settlement that findspots not normally recorded (of coin hoards, treasures, and so on) are mapped — using the modern place-name associated with the find, not necessarily its actual location (xxvi).

18       P. Horden and N. Purcell, The corrupting sea: a study of Mediterranean history (Oxford 2000), with review in the present issue of this journal by B. D. Shaw.

19       For contrasting approaches to the study of maps see, for example, D. Woodward (ed.), Art and cartogra-phy: six historical essays (Chicago 1987), and D. Gregory, Geographical imaginations (Cambridge 1994).