Ancient History & Civilisation


Putting the World in Order: Mapping in Roman Texts

The scanty survival of genuinely contemporary cartographic artifacts from Roman antiquity thrusts the simple textual description of geography into the limelight. While there is no denying the existence and public display of the grand cartographic visualizations discussed in the previous chapter, the dearth of surviving maps and diagrams frequently forces us to rely upon the examination of purely textual sources for evidence of “worldview,” whether explicitly expressed by them or embedded in their structure. Indeed, the paucity of direct evidence for visual mapping has led some modern scholars to downplay the role of graphic illustration in the description of geographic information by the Romans.¹ Such a view is encouraged by the prominence of itinerary lists in the surviving literature, giving rise to the idea that the Romans’ perception of the world was primarily linear, organized as so-called hodological space.² However, the use of itineraries to record or present routes for travel need not be incompatible with the consultation of maps, map-based thinking, or general “map consciousness” in other contexts.³

This chapter sets out to examine a wide range of texts from Roman antiquity (in Greek and Latin) both for signs of map consciousness and for evidence of its nature. Specifically, the descriptive logic of the texts will be examined for evidence of the sort of picture of the world that the educated Roman carried around in his or her head. For example, on what was it centered? In which direction was it orientated? Was a cartographic image of the world even common at all? Obviously the answers to these questions will have been different at different times and locations across Roman history. They are likely to have differed between public and private contexts and varied according to personal education. They may well have differed, too, depending upon whether the world was being imagined on a global, regional, or simply local scale. Although the survey will encompass literary works, the emphasis will be upon less self-conscious technical, subliterary, and documentary material. The theme has been explored before. The material, despite being scattered, is relatively plentiful and well known, so that this will be an exercise in sampling rather than a comprehensive examination. The primary contribution of this fresh analysis will be to consider the literary and subliterary material together over an extended time frame stretching from the third or second century BCE to the sixth century CE. Treating these disparate materials together may appear an undisciplined and incoherent approach but, in accordance with the principles of Brian Harley (2001), the analysis will attempt to treat each within its proper social and cultural context.

Cultural and Chronological Parameters

“Roman texts” constitute, of course, a rather broad category. It requires some further definition, but its openness is deliberate, as we shall be ranging widely across works of very disparate character. They vary greatly in genre, authorship, and language—literary and documentary, official and private, Greek and Latin. They survive via a wide range of media—inscriptions on stone and bronze, even bone, and writings on papyrus, as well as the more familiar medieval manuscript tradition. The only group largely and deliberately set aside here is that of scripture and scriptural exegesis (whether Jewish or Christian). The rise and predominance of Christianity as a cultural force by the end of the period under consideration are undeniable phenomena, and the inclusion of religious texts within the survey could no doubt only enrich its findings. However, it would also threaten to distract from the purpose of elucidating traditional, secular Roman cultural practice rather than Judeo-Christian perspectives. As it is, the texts discussed span a long period, during which the Roman world grew from being merely one of the powers in the western Mediterranean to becoming the supreme power there and in continental Europe from the first century BCE to the fifth CE, only to retreat once again to being a primarily Mediterranean power. At the same time the res publica Romana developed from a pagan city-state to a Christian empire. Over such a broad spectrum of social and cultural development, it would be unreasonable to expect a consistent worldview to be expressed. Nor is it necessarily reasonable to expect the same practices to be applied to the different purposes of general description of the world (geography), description of discrete regions or local areas (chorography), and the compilation of way-finding itineraries. Nevertheless, it might be an a priori assumption that “Roman” authors would tend to view the world relative to the standpoint of the capital. In fact, as we shall see, this is surprisingly uncommon.

A complicating factor is that the “Romanness” of the mapping encountered in texts emanating from the Roman world cannot be taken for granted. First of all, as in many areas of Roman culture, in matters of geography there was an acknowledged debt to Greek models. Nevertheless, for those texts that are the output of Roman authorities, whether central or provincial, their Romanness is relatively unproblematic. However, in the case of literary works, the matter is less straightforward. It is not so much one of language (Latin versus Greek) as of political and cultural identity. In the Greek east, because the acquisition or inheritance of Roman citizenship was frequently not accompanied by knowledge of the Latin language, let alone by any great familiarity with Italian culture, many individuals who considered being Roman one component of their identity were no different culturally from the mass of noncitizens around them. Conversely, certain authors of some significant surviving works of geography written within the Roman world were not just non-Roman by citizenship, but also would probably have shuddered to find their work branded as a product of Roman culture. Before the extension (in 212 CE) of Roman citizenship to most free subjects who did not yet possess it, non-Roman (mostly Greek-speaking) writers active under Roman rule varied greatly in their engagement with the contemporary political framework.

The point applies as much in the realm of geography as it does in others. At one end of the spectrum stands Claudius Ptolemaeus (Ptolemy), an Alexandrian Greek—and to judge from his name Claudius almost certainly a Roman citizen—of the later second century CE, who produced a mathematically plotted cartographic description of the contemporary world discussed in chapter 4 above. At the other end stands his slightly older contemporary, Pausanias, an Ionian Greek whose description (Periegesis) of Greece (an area he called by the name Hellas, not the contemporary Roman provincial designation Achaia) largely eschews mention of the Roman context. Somewhere between the two stands the Cappadocian Greek author of the first quarter of the first century CE, Strabo of Amasia, who explicitly acknowledges contemporary political reality in hisGeography (see chap. 3 above) without necessarily being straightforwardly philo-Roman.

We need to keep this spectrum in mind when assessing the extent to which there might be an identifiably Roman approach. Nevertheless, there is some coherence to the range of texts surveyed: all are Roman in the sense of being produced within the varying extent of the dominion of Rome, all engage more or less with that Roman context, and most are demonstrably by Roman citizens, even if not natives of the metropolis or of Italy. Of the various standpoints represented, that of members of the educated public is best represented (in literary works). By contrast, the standpoint of the Roman government (insofar as it is legitimate to think of such an entity) is least well represented, glimpsed only sporadically through chance documentary survivals or embedded in the official naming of political units.

The texts examined exhibit a range of relationships with maps. Some of those that survive may originally have accompanied maps. The most conspicuous example is Ptolemy’s Geography, written in Alexandria in the late second century CE. Its text provides the method and data to construct both overview and regional maps, and it formed the basis for the Renaissance atlas tradition, but no illustrated manuscript survives from antiquity. By contrast, although it has been argued that the verse Periēgēsis (Guide around) or Periodos gēs (Route around the earth) by Ptolemy’s fellow Alexandrian Dionysius is intended as an ekphrasis (explanation) accompanying a graphic illustration, the evidence is ambiguous, and the maps found in the medieval manuscripts do not match the text.¹⁰

Even so, more generally there is strong circumstantial evidence that works akin to modern atlases did circulate, whether or not Ptolemy’s work carried illustrations. Two enigmatic Latin texts, entitled the Dimensuratio (better Demensuratio) provinciarum (Measuring out of the provinces) and the Divisio orbis terrarum (Division of the world), are best understood as comprising the captions (commentarii) that once belonged to a series of individual regional maps.¹¹ Both works are made up of short entries, each devoted to named regions and ending with figures in Roman miles for maximum longitude and latitude, which would have provided the reader with an idea of the relative scale of each region, since they were no doubt depicted on columns of equal width (if a scroll) or on pages of equal size (if a codex). Although there is a great deal in common between the data provided by the two works, they differ in the number and order of captions, so that they are witnesses to two different Latin atlas traditions, both now lost.¹²

Conversely, as with Dionysius Periegetes, some of the maps and diagrams that do illustrate medieval manuscripts of classical works are clearly not original to their design.¹³ Moreover, graphic visualization of geographic data does not necessarily equate to cartography by modern definitions, as the example of the Dura parchment illustrates (fig. 7.1 and plate 9). This piece of parchment, used as shield decoration, was preserved in the dry conditions of the Roman fort of Dura on the Euphrates in Syria. Part of the design depicts a maritime itinerary along the coast of the Black Sea, the surviving portion running clockwise from Odessos (modern Varna in Bulgaria) at the top to Arta (the Straits of Kerch, or Cimmerian Bosphorus, between the Crimean peninsula and Russia) at the bottom.¹⁴ However, while indubitably Roman by date, archaeological context, and textual content (the distances are indicated in mil(ia), i.e., Roman miles, not Greek stades), as an artifact it is arguably decorative rather than cartographic. Without illustrations, how can we diagnose “map consciousness” in texts? It might be characterized by the organization of geographic data in a manner that is coherent in terms of spatial awareness. Occasionally a text may display this cartographic quality because it has actually been derived from a map, but in most cases it is likely that textual descriptions were simply composed according to a mental visualization of the world that authors could reasonably expect their readers to share.

Greek Heritage

Anyone who has opened a textbook on Roman history will be familiar with illustrations in which the political divisions of the empire are mapped onto a modern cartographic projection of Europe and the Mediterranean basin (for example, fig. 7.2). As it happens, Rome appears here more or less central horizontally as well as vertically, just as London (or more precisely Greenwich) does, longitudinally at least, in modern projections of the world. London’s placement is, of course, the product of the British chauvinistically placing themselves as the initial reference point and eventually convincing everybody else (even the French) to follow suit. However, in the Roman case such central placement was essentially fortuitous. For, as already noted, the Romans’ understanding of geography was influenced by Greek culture, to which the Romans themselves were peripheral. As Georgia Irby demonstrates in chapter 3 above, by the classical period a view of the “circle of the world” was common, in which a landmass divided into three continents (ēpeiroi in Greek) surrounding, but not entirely enclosing, the Mediterranean was itself surrounded by the outer sea (Ōkeanos). The Pillars of Hercules (Straits of Gibraltar) provided the natural demarcation of Europe from Africa (Libyē in Greek), while the rivers Tanais (the modern Don) and Nile were generally considered to mark the transitions from Europe to Asia and Asia to Africa respectively. Advances in knowledge and science meant that by the Hellenistic period the earlier “world” had been reduced to just one portion of the entire globe. This known part came to be termed the oikoumenē, “habitable zone,” one of two such zones that were postulated north and south of the equator. Dicaearchus imagined that this oikoumenē was intersected by a principal east-west axis (diaphragma) that ran from the Pillars of Hercules to the Levant and then along the line of the Taurus Mountains, which were thought to extend on into the Caucasus and the Himalayas, as seen in the Peutinger map; Alexandrian scientists added a north-south axis through Byzantium, Rhodes, Alexandria, and Syene. It was this image of the world, glimpsed by us in the surviving fragments of Eratosthenes’s writings, that educated Romans of the second century BCE inherited when Rome emerged as the dominant power in the Mediterranean.¹⁵ They identified the Greek concept of the oikoumenē with the Latin notion of the “globe of the earth” (orbis terrae or terrarum).

FIGURE 7.1 The Dura parchment (Paris, Bibliothèque Nationale de France ms. Supplément grec 1354², V). Insert illustration bound with F. Cumont, “Fragment de bouclier portant une liste d’étapes,” Syria 6, no. 1 (1925): 1–15.

Following Hellenistic tradition, knowledge of geometry (“earth measuring”) remained a standard part of the formation of an educated Roman, as late antique educational manuals demonstrate. Both Martianus Capella¹⁶ in the early fifth century and Cassiodorus in the mid-sixth include geometry, alongside arithmetic, music, and astronomy, as part of the advanced curriculum (the quadrivium).¹⁷ While the ars gromatica (art of surveying) was an area in which Roman science seems unusually to have been genuinely independent of Greek learning, the large scale geometry of the earth remained dominated by Greek authors, albeit through translations.¹⁸ Cassiodorus specifically prescribed among “cosmographers to be read by monks” (cosmographos legendos a monachis) the “map” (pinax) of Dionysius Periegetes and, for more advanced students, the “book” (codex) of Ptolemy.¹⁹ Even though Dionysius had not been Roman by citizenship (unlike his compatriot Ptolemy), his description of the world in 1,186 lines of hexameter verse, written under Hadrian in the early second century CE, is likely to have been more influential generally in forming the worldview of educated Romans.²⁰

FIGURE 7.2 Map of the regions of Italy and provinces of the Roman empire ca. 200 CE, including the disposition of the legions as listed on the Maffeian column (fig. 7.13 below). Adapted by the author from R. J. A. Talbert, ed., Atlas of Classical History (London: Croom Helm, 1985), 170–71.

As Cassiodorus’s instructions show, Ptolemy was reserved for further study despite (or because of) his scientific superiority. By contrast, Dionysius’s Periēgēsis became a staple of the later Roman classroom in the Latin west as much as in the Greek east, being translated into Latin twice, once in the fourth century by the pagan senator Rufius Festus Avienus and again in the sixth by the Christian grammarian Priscian of Iol Caesarea (modern Cherchel in Algeria).²¹ The work’s popularity no doubt derived from its mapping of mythology onto a contemporary understanding of the world, making it a perfect companion to the study of epic poetry. However, unlike Ptolemy later, who incorporated the latest knowledge of the Far East, Dionysius perpetuated the traditional oikoumenē of Eratosthenes that extended only as far as India. Nevertheless, an awareness of the Chinese (Seres) did penetrate to the most basic level of instruction, as exemplified by the Liber memorialis of Lucius Ampelius, written in the second or third century CE.²² It functioned as an elementary compendium of universal knowledge, perhaps for learning by rote. Astronomy and geography are included among the preliminary matter. Here, after treating the cosmos, constellations, stars, and winds, Ampelius includes a section on the orbis terrarum (sec. 6), before embarking on the bulk of the work, which is devoted to a digest of history down to the time of Trajan.

Textual versus Graphic Mapping

It is perhaps best to start with some observations on the fundamental differences between graphic and textual representation of the world. Although a labeled map encourages reading from left to right (at least in Greek or Latin), which might in turn equate to reading from west to east depending upon orientation, a purely textual description of geographic data imposes many more constraints on how they are to be read. We can compare, for example, the Roman governor Arrian’s Circumnavigation of the Black Sea, written in the 130s CE,²³ with the fragmentary Dura parchment from the mid-third century CE mentioned above. Both describe stations along the Black Sea coast (fig. 7.3). The viewer of the parchment can begin at any point and “read” in either direction. As an author, however, Arrian had to make choices and the reader is compelled to follow—in this case anticlockwise from the Thracian Bosphorus. The need for him to choose a starting point, an order of description, and an ending point has useful consequences nevertheless. Unless they are completely arbitrary, his choices can offer insight into authorial viewpoint at least, if not into more widely held understandings of the proper organization and hierarchization of space.


In order to assess the significance of the choices made by the writers or compilers of our texts in describing the world around them, we need first to consider the range of possible options. Above all, it must be remembered that there are alternatives to the organization of intrinsically geographic data according to a spatial principle. Given his intended purpose, the fourth-century scholar and bishop Eusebius of Caesarea naturally organized his gazetteer (Onomasticon) of places in the Holy Land as an alphabetical lexicon,²⁴ and Stephanus of Byzantium in the sixth century did the same for his guide to the proper forms of the names of peoples.²⁵ In his great encyclopaedia, the Natural History, the Roman eques (“knight”), Pliny the Elder, appears to have utilized digestiones in litteras—alphabetically organized statistical lists—of the communities of individual provinces and regions, compiled by the Roman authorities since at least the end of the first century BCE.²⁶

Indeed, alphabetical order can be found lurking in unexpected places. The final chapters of the tetrarchs’ edict on maximum prices from 301 CE state maximum rates for the transport of goods by sea along a network of routes spanning the length and breadth of the Mediterranean and into the Black Sea.²⁷ These routes are organized into a series of significant departure points that comprise a mixture of specific ports and more broadly defined regions. In Latin, which was certainly the language of the authoritative version of the edict, the order of the place-names—Alexandria, Oriens, Asia, Africa, Roma, Sicilia, Nicomedia, Byzantium—does not respect a consistent geographic principle. Even so, it can be read convincingly as reflecting a recognized hierarchy of importance from the viewpoint of a shipper based in the eastern Mediterranean, where the edict as a whole was certainly compiled.²⁸ Leaving aside Nicomedia, whose otherwise unexpected prominence is comprehensible given the presence there of the emperor Diocletian’s court, all the others have traditional claims to be ranked among the most significant hubs in maritime transport and commerce. However, if the list is translated into Greek, the natural language of maritime matters especially in the eastern Mediterranean, it emerges that alphabetical order—Alexandreia, Anatolē, Asia, Libyē, Rōmē, Sikelia, Nikomēdeia, Byzantion—may be a simpler explanation; in addition, the data for the routes may have been recovered from an alphabetized filing system.²⁹ If so, then both Nicomedia and Byzantium are exposed as addenda to the core list—Nicomedia for the reason just suggested. Byzantium may bring up the rear out of alphabetical order, having been considered outside the Mediterranean group and signaling rather the gateway to the Black Sea.

FIGURE 7.3 The circumnavigation of the Black Sea as recounted by Arrian, Periplus Maris Euxini, in a firsthand report (1–11) and a secondhand description (12–25). Adapted by the author from A. Liddle, ed. and trans., Arrian, Periplus Ponti Euxini (London: Bristol Classical Press, 2003), 136–39, maps 1–2.

As well as alphabetical or numerical order, a hierarchy of status may override purely geographic considerations. Several documents expose the habits of the Late Roman administration in this respect. The first is the famous Notitia Dignitatum, which preserves a version from around 400 CE of the list of senior government offices maintained by a palace official, the primicerius notariorum. It exhibits a basic binary division into eastern and western parts, reflecting the political division of the empire since 395 between the emperors Arcadius and Honorius.³⁰ Priority is given to the east over the west, no doubt because of Arcadius’s seniority to his brother in the imperial college. As the lists of contents of the two halves show, the provincial administration is described primarily in order of the rank of the posts and then only within each rank with reference to geography. In the eastern list, within each of the grades (praetorian prefects, proconsuls, vicarii, comites rei militaris, duces, consular governors, and ordinary praesides and correctores) the posts are listed by prefecture and their constituent administrative dioceses (groups of provinces) in an anticlockwise order from Libya (Cyrenaica) to the Danube: Aegyptiaca, Oriens, Asiana, Pontica, Thracia, Macedonia, Dacia. Its western counterpart is not quite so consistent, but the predominant order of dioceses is Pannoniae, Italiae, Africa, Hispaniae, Septem Provinciae-Galliae, Britanniae. Within each diocese, there is often some geographic logic to the order in which provinces are listed; however, this is subordinated first to ranking by dignity of governor and, in the case of subdivided provinces, to numerical sequence. For instance, the list of the provinces of the diocese of Galliae is headed by Lugdunensis Prima, under a consularis, followed by Germania Prima and Secunda, Belgica Prima and Secunda, Alpes Poeninae et Graiae, Maxima Sequanorum, Lugdunensis Secunda, Tertia, and Senonia, all under praesides (Not. Dig. Occ. 3.14–31).

The roughly contemporary Notitia Galliarum, which lists just the provinces of the two dioceses of Gaul (see fig. 7.4), is much more successful in blending respect for political hierarchy and numerical sequence with geographic coherence.³¹ Here, although geographic contiguity is still subservient to the other criteria, the provinces of Galliae and Septem Provinciae are listed in clockwise and anticlockwise sequences respectively. So Galliae, for instance, is described in the order: Lugdunensis Prima, Secunda, Tertia, and Senonia, Belgica Prima and Secunda, Germania Prima and Secunda, Maxima Sequanorum, and Alpes Graiae et Poeninae (Not. Gall. 1–10).

Another document comparable to the Notitia Dignitatum is a schedule (notitia) of fees to be paid by those appointed to various provincial commands within the praetorian prefecture of Oriens that is appended to Novel 8 of the emperor Justinian, dated 15 April 535. Here the offices are again grouped according to grade in order of descending dignity, rather than by physical proximity, from the vir spectabilis comes Orientis, through the consular governors, to the ordinary praesides and correctores.³² Similarly, as in the Notitia Dignitatum, within this framework there is some evidence of organization according to administrative dioceses in an anticlockwise geographic order from Africa to Europe (Aegyptus, Oriens, Asiana, Pontica, Thracia), but it is not consistently applied.

FIGURE 7.4 The Roman provinces of Gaul in the fifth century CE. Map by the author.

The discipline of these official lists contrasts with the much less organized presentation of the same sort of material in the mid-fifth-century Laterculus of Polemius Silvius, a retired palace official, who groups his lists of provinces thus: Italia, Galliae, Africa, Hispania, Illyricum, Thracia, Asia, Oriens, Pontus, Aegyptus, Britannia.³³ However, in contrast to the dry listings in official documents that do not intend to be geographically helpful, Polemius does provide his readers with some occasional geographic context of a very basic kind. For example, in Gaul, where he was writing, he notes, under Belgica Secunda (2.9), “It is from here that the crossing to Britain is made” (de qua transitur ad Britanniam), while both Germania Prima and Secunda (2.10–11) are noted as “on the Rhine” (super Rhenum), and Lugdunensis Secunda and Tertia (2.13–14) as both “on the Ocean” (super Oceanum). At the end of the Spanish section, he records that Mauretania Tingitana (4.7) is “across the strait, which flooded by the ocean divides the lands between Calpe and Abinna” (trans fretum, quod ab oceano infusum terras intrat inter Calpem vel Abinnam). In Illyricum, Dalmatia (5.1) is “on the sea” (super mare), while Noricum Ripense (5.9) is “on the Danube” (super Danuvium). In the east, Cilicia (8.5) “abuts the Taurus mountains” (iuxta montem Taurum), and Mesopotamia (8.7) is “between the Tigris and Euphrates” (inter Tigrim vel Eufraten). All of these points, if they are to be any help, assume that the reader has a basic grasp of the physical geography of the Roman world.


Another form of list particularly associated with the Roman mindset is the itinerary.³⁴ These are found used for planning and recording routes, and even for commemorating individual journeys. As in later periods, these simple one-dimensional enumerations of stops with intervening distances served a practical purpose for Romans in describing routes or recording specific journeys.³⁵ Maritime itineraries (periploi) have a long history in the Greek world, and they continued to be produced by and for Hellenophone shippers under Roman rule.³⁶ Most famous is the anonymous description of the trade route from Egypt to India via the monsoon winds known as the Periplus of the Red Sea, written in the mid-first century CE.³⁷

However, it is itineraries for land travel that predominate in the Roman tradition. Two notable texts, recording journeys made by individuals, happen to survive from within little more than a decade of each other in the early fourth century CE. First, among the memoranda generated by the journey of one Theophanes from Hermopolis Magna in Egypt to Antioch in Syria and back some time in the period between 322 and 324, are his records of the stops made and distances covered.³⁸ Second, the account of an anonymous traveler from Bordeaux to the Holy Land and back again, dated to 333, has been transmitted via the medieval manuscript tradition.³⁹ Here the bare list of stops and distances has been annotated by the traveler to indicate his or her use of places, mostly as either an overnight stop (mansio) or a change of mount (mutatio); the section about Palestine, however, expands into a full-blown tourist narrative.⁴⁰ Three centuries or so earlier, several travelers from Spain to Rome had their itineraries of just over a hundred stages from Gades (Cadiz) to the capital vividly commemorated by being engraved on four silver beakers (see, for example, fig. 7.5). These would appear to have then been ritually deposited, perhaps in thanksgiving for safe completion of the trip, because they were retrieved in modern times from the waters of a spring at the Bagni di Vicarello on the Lago di Bracciano in South Etruria.⁴¹ The framing of the columns of the text on each beaker by imitations of architectural elements suggests that the design may be modeled on a monument, perhaps one that stood in Gades displaying the itinerary to Rome.

FIGURE 7.5 One of the silver beakers from the Bagni di Vicarello: Corpus Inscriptionum Latinarum XI 3281, headed “ITINER | ARIVM | A GADES | ROMAM.” Rome, Museo Nazionale delle Terme, inv. No. 67497. Berlin-Brandenburgische Akademie der Wissenschaften, Corpus Inscriptionum Latinarum archive. Reproduced with permission.

The existence of such public displays of itinerary lists is demonstrated by the fragment of an octagonal column from Tongeren (Atuatuca Tungrorum) in modern Belgium (fig. 7.6).⁴² It presents lists of stations along several routes in the region of northern Gaul in the same fashion as the Vicarello beakers and the archive of Theophanes, that is, as simple names with scarcely any contextualizing references to physical or political geography, let alone explicit statements about direction. For example, in the seventh line of the central column the entry Isara l(eugae) XVI (Isère 16 leagues) does not indicate that this is the name of a major river. Moreover, when the column’s routes are mapped out, with attention paid to the direction of each, it becomes clear that they are not described as radiating out from Tongeren or even from any single consistent viewpoint (fig. 7.7). The routes through Antunnacum (Andernach) and Borbetomagus (Worms) in the first column, through Durocortorum (Reims) and Samarobriva (Amiens) in the second, and from Bagacum (Bavay) to Nemetacum (Arras) in the third might all conceivably depart from Tongeren, but then the route from Castellum Menapiorum (Cassel) to Nemetacum (Arras) clearly cannot. Such lists of itineraries might cumulatively describe a network of routes; without indications of direction or explicit signaling of junctions, however, they do not facilitate the “mapping” of the space traversed by their readers. Nevertheless, this method of point-to-point description, presenting the perspective of a traveler rather than a map reader, predominates in the remarkable geographic digressions of the late fourth-century historian Ammianus Marcellinus, despite the fact that he cites Ptolemy on at least one occasion (22.8.10).⁴³

FIGURE 7.6 The pillar from Tongeren, Belgium: Corpus Inscriptionum Latinarum XIII 9158 = XVII/2 675. Brussels, Musées Royaux du Cinquantenaire. Berlin-Brandenburgische Akademie der Wissenschaften, Corpus Inscriptionum Latinarum archive. Reproduced with permission.

It was the major achievement of the anonymous compiler(s) of the collection of itineraries known as the Itinerarium Antonini, or Antonine Itinerary, which took its final form around 300 CE, to knit regional subcollections of itineraries together to form a network covering the length and breadth of the empire.⁴⁴ The seemingly chaotic agglomerations of itineraries, exhibiting switches in viewpoint of description reminiscent of the Tongeren pillar, and the variability in the nature of the material between sections strongly suggest that the collection was built up from disparate pieces of publicly circulating material rather than having been excerpted from central archives. Nevertheless, using such material, the compiler of the Antonine Itinerary collection was able to patch together fifteen distinct regional collections to form an ensemble that describes the Roman world in an anticlockwise direction from the Straits of Gibraltar to Britain, that is, from the western extreme of Africa to the northwestern fringe of Europe (see fig. 7.8). The result produces a superficial impression of coherence at the “oecumenical” level, even where individual “chorographic” components run counter to the general geographic tendency.⁴⁵

FIGURE 7.7 The route network of the Tongeren pillar. Map by the author.

Although the Antonine Itinerary collection might seem confusing at a provincial level, one of the texts inscribed on a monument erected in honor of the emperor Claudius at Patara in Lycia in 45 or 46 CE demonstrates that it was possible to map the layout of a province using itineraries according to a consistent presentation.⁴⁶ Besides the dedication to the emperor on the narrow (front) face, the long left side (seen in fig. 7.9) and the opposite right side are covered with a remarkable Greek inscription listing sixty-five separate stretches, making up thirty itineraries, linking twenty significant places throughout the newly annexed province. The monument stood on the dockside at Patara, greeting new arrivals from abroad. The careful structure of the route network is best appreciated in the diagram in figure 7.10, where Arabic numerals label the thirty itineraries and letters indicate the twenty significant nodes.⁴⁷ Maintaining the overall primacy of Patara (the chief city of the province and location of the monument), the network is described by proceeding along the most westerly route to the next node and so on until the path is exhausted. Then, jumping back to the last significant node, the next route out of town clockwise is followed, and so on. By this method the territory of the province is traversed from the west coast in a clockwise arc until it hits the east coast at Phaselis and Korydalla. Then, rather than double back to Patara, the description returns to the chief node and begins again with the next route anticlockwise out of Patara.

FIGURE 7.8 Schematic diagram of the structure of the Itinerarium Antonini. By the author.

There is a neatness and economy to this method that is not found in the Antonine Itinerary, but severe limitations to its usefulness as a practical guide to the traveler need to be recognized. The method results in the atomization of conveniently direct long-distance routes (Patara to Kaunos, say, or Patara to Kibyra) into relatively short and disjointed stretches. Had the compiler of the Antonine Itinerary subjected his component material to such ruthless editing, its utility would have been greatly undermined. This application at Patara of a rigorously consistent method of description at the expense of practical utility suggests that—as with the route network on the Peutinger map—the list of places on the monument was intended to impress by its extensiveness. Moreover, the beauty of the Claudian monument is that its physical location is identical with the viewpoint of the text inscribed on it.

FIGURE 7.9 Three-dimensional rendering by Fatih Onur of the left face (face B) of the Claudian monument from Patara, Lycia, bearing the list of the first thirty-nine stages of the route network. Reproduced from S. Şahin and M. Adak, Stadiasmus Patarensis: Itinera Romana Provinciae Lyciae (İstanbul: Ege Yayınları, 2007), p. 25. Reproduced with permission of Fatih Onur.

How the list on the monument was produced is an intriguing question. Since most of the distances are demonstrably calculated in Roman miles, it is likely that the data are not a pooling of preexisting information, but rather in large measure derive from a recently conducted survey of the new province.⁴⁸ Interestingly enough, it is easier to imagine the particular structure of the list being generated by reading off a sketch diagram of the network of routes showing the relative position of each town in two dimensions; to generate this structure from conventional itinerary lists, by contrast, would require a considerable effort in splicing and recompiling. Thus, the Patara monument is perhaps now a stronger candidate for a text derived from consultation of a map than any other previously proposed.

FIGURE 7.10 The route network of the Claudian monument from Patara. Map by the author.

A less strong candidate in this regard is the list of marine transport costs in the tetrarchs’ maximum prices edict. Keith Hopkins had diagnosed it as an example of map-based thinking on two main grounds: first, the attribution of the same maximum price to a route from one port to another regardless of direction implied that no account was taken of the contrary effects of prevailing winds and currents depending on the direction of travel; second, inconsistencies of cost per unit of distance between routes could be attributed to inaccurate maps that overestimated the length of the western Mediterranean.⁴⁹ However, a recent alternative analysis by Pascal Arnaud concludes that there is some sensitivity to the different conditions between routes, and that the apparent artificiality of the pricing structure does not derive from reading a map but through simplification. The basic factor taken into account in calculating the cost was duration rather than distance. Direct routes are cheapest per unit; those that are segmented (i.e., pass through points at which transshipment is likely to have taken place) are more expensive.⁵⁰

As the example of the Patara monument demonstrates, the one-dimensional itinerary, or periplus, had severe limitations when it came to describing the contours of a region. Accordingly, as noted in chapter 3 above, both Hellenistic and Roman writers were accustomed to communicating the two-dimensional outline of physical geographic regions by comparison with abstract geometric forms or the shape of everyday objects.

Mapping the Oikoumenē: The Continents

As we have seen, educated Romans took over from classical and Hellenistic geographers the concept of a habitable world (oikoumenē) divided into three continents (Asia, Europe, and Africa) grouped around the Mediterranean Sea (mare nostrum—“our sea” in Roman terms), through which ran the central east-west axis (diaphragma), with the whole surrounded by the outer ocean. This identification of the oikoumenē was an easy one to make when the existence of India (and later China) was tacitly ignored. By the first century CE Roman power had sufficiently encompassed the Mediterranean world that it was easy enough for Romans and their subjects to pretend that their empire was synonymous with the entire orbis terrarum.⁵¹ When Augustus (Rome’s first emperor) died and his own account of his achievements was posthumously published, whoever posted up our most complete surviving copy in the province of Galatia headed it with the summary “An exemplar of the things achieved (res gestae) by the divine Augustus, by which he subjected the orbis terrarum to the authority of the Roman people, and of the expenditure that he made for the state and the Roman people.”⁵² On the monument from Patara discussed above, Claudius, famous as the first emperor to extend Roman power across the ocean to Britain, is celebrated as “the emperor of the oikoumenē.”⁵³ The Jewish historian (and recent Roman citizen) Flavius Josephus, writing soon after 70 CE, has the Jewish king (and Roman citizen from birth) Agrippa II, speaking in 66, say “Throughout the oikoumenē all are Romans,” even while referring to a Jewish population outside the Roman empire “beyond the Euphrates” (Bell. Iud. 2.388).⁵⁴ Aelius Aristides, a Greek orator from Pergamum, in his mid-second-century panegyric To Rome, similarly identifies the boundaries of the Roman empire, defended by a ring of fortifications, with the boundaries of the civilized world.⁵⁵

Given such identification of Roman dominion with the oikoumenē, it seems surprising that Rome itself is not more often made the center or starting point for geographic descriptions. Even those authors most closely associated with the glorification and cataloging of the extent of Rome’s dominion do not routinely appear to consider the city the umbilicus orbis, or geographic center, of the oikoumenē, as Greeks had once considered the national shrine at Delphi its omphalos. Strabo, writing in the early first century CE, comes close to such a formulation. Near the end of hisGeography (17.3.4) he conceptualizes the Roman empire and the entire world as spreading in concentric circles around Rome: first Italy, then the regions around Italy in a circle (kyklos), and the three continents (Europe, Libya, and Asia). In a similar fashion, according to Richard Talbert in the previous chapter, the original scheme of the Peutinger map put Rome at the center not only vertically but also horizontally.

There were various potential options for subdividing descriptions of the entire Roman world according to differences in physical, political, or cultural and environmental geography. In physical terms, leaving aside grouping by continents for a moment, the Romans did on occasion categorize places as loca maritima (facing the sea), media terrena or mediterranea (inland),⁵⁶ or ripensis (facing a major river). However, writers of geographic treatises or of geographic digressions in historical works evidently did not make these differentiations. We do find one official legal source, the edict of an anonymous emperor, distinguishing Italy on the one hand from the provinciae transalpinae et transmarinae (provinces usually reached by travel over land or overseas) on the other.⁵⁷ This edict, preserved on a papyrus in Egypt, was clearly of interest to the inhabitants of that provincia transmarina because it established a differential in the time permitted for the referral of appeals from criminal trials to the emperor; those from Italy were permitted nine months, and those from the transalpine and transmarine provinces eighteen months. Such a ruling clearly assumes an emperor primarily based in Rome, a situation best suited to the conditions of the first or second century CE. By the time that the papyrus copy was made (some time in the early third century perhaps), the ruling may indeed have been somewhat anachronistic.

In grouping descriptions, we might also expect reference to the fault lines of cultural geography that might be observed running through the Roman empire. If describing the same region of the world today, we might naturally organize countries into groupings of shared religious culture or of cognate languages. However, the first would not really work for the Roman world because, with the exception of the Jews and the increasingly ubiquitous Christians, its entire area was characterized by one form or another of polytheistic cult. Even so, there were two cultural fault lines that did run through the Roman world; both bisected it, one primarily east-west, the other north-south.⁵⁸ The horizontal divide was an environmental one separating the (mostly Mediterranean) provinces, whose staples were wheat, wine, and olive oil, from those in which the last two were replaced by beer and butter. The vertical divide was a linguistic one between the Latin “west” and the Greek “east” (a convenient shorthand much used by modern scholars), or to put it more precisely, between those areas in which the language of Roman government was Latin and those in which it was Greek. On the North African coast this divide separated the eastern and western arms of the Gulf of Sidra (in modern Libya), dividing Greek Cyrenaica from Latin Tripolitania. On the northern shore of the Mediterranean the Latin-Greek divide ran roughly east-northeast through southwestern Europe from the Adriatic to the Black Sea, dividing the Latin provinces of Dalmatia and Moesia from Greek Epirus, Macedonia, and Thrace. Real as these divides might be, we do not find them echoed in the geographic or historical literature, whose writers preferred to group descriptions according to continents or political units or both.

At this point, then, we should consider the role of the names of the three continents in Roman political toponymy or, as Ptolemy would no doubt term it, “choronomy.” Of their three names, both Asia and Africa also served emblematically as labels for the provinces occupying the portions of those two continents nearest Rome—indeed, those areas where in the second century BCE Rome had gained its first intercontinental footholds. Such labeling arose from the nature of Roman provincial commands, which originated as conceptual theaters of operation allocated annually to Rome’s senior magistrates. Only when these theaters came to be administered as permanent possessions did they take on concrete terrestrial form. This helps to explain why the term “Europe,” despite its currency as a geographic concept, did not feature in the Romans’ political geography as either a provincial or regional designation.⁵⁹ Given the physical location of Rome and Italy on the European continent, the term could not usefully serve to distinguish any sphere of responsibility from the empire’s heartland. A rare example of the use of Europe to designate a group of territories in an official Roman context comes from a grant made by the masters of the Roman state in the mid 30s BCE, the triumvirs Mark Antony and Octavian (“Imperator Caesar”). The text survives (in Greek translation) inscribed on marble in honor of its beneficiary at his hometown Rhosus, an ancient port south of modern Iskenderun in the Gulf of Issus.⁶⁰ It grants Roman citizenship to a naval captain, Seleucus, in gratitude for his contribution to the defeat of the triumvirs’ former rival Sextus Pompeius. Various other privileges are granted too, including freedom from customs dues on imports and exports to or from “a city or territory of the provinces of Asia and Europe” (line 48). The omission of Africa from the grant—for whatever reason—is interesting.

The earliest account of the continents to survive in Latin literature is that by the late republican senator Sallust. In his historical monograph on the Numidian king Jugurtha (written in the late 40s BCE), he sets the scene with a geographic and historical introduction to Africa as a continent (Bell. Iug. 17–19). It opens: “In dividing the orbis terrae, the majority put Africa as the third part (pars tertia), though a few recognise only Asia and Europe, putting Africa in with Europe.” This initially rather surprising statement is echoed much later in the geographic introduction to theHistoriae adversus paganos by Paulus Orosius, a Christian priest from northwest Spain. He wrote this history, in Africa as it happens, around 417 CE as an apologetic response to the pagan claim (prompted by the sack of Rome in 410) that Christianity had brought disaster to the Roman empire.⁶¹ He explains (1.2.1) that “our ancestors conceived of the orbis of the whole terra, surrounded by the periphery of ocean, as tripartite (triquetrus) and called its three parts Asia, Europe, and Africa, although some conceived of two, that is, Asia and then Africa subsumed in with Europe.”

Why might Africa be the third part, and why might it be merged with Europe? To those of us accustomed to thinking of these divisions as continents in the modern sense, the idea of merging Asia with Europe or with Africa on the basis of contiguity seems perfectly reasonable. By contrast, a merging of Africa with Europe—separated as they are by Asia at one end and the Straits of Gibraltar at the other—seems highly counterintuitive. However, the meaning of Sallust and Orosius can be illuminated by reference to the entry De orbe in the encyclopaedic dictionary that Isidore, bishop of Seville, compiled some time between 615 and 630 (Etymologiae 14.2.1–3).⁶² He divides the orbis into three parts but explains that they are not equal: Asia extends from north to east, Europe from north to west, Africa from west to south. Accordingly Europe and Africa, divided into two regions by the Mediterranean, together occupy half the globe, Asia the other half by itself. So, if you divide the globe in two parts, the east and the west, Asia is in one, Europe and Africa in another. These, of course, are roughly the proportions found earlier in Eratosthenes as well as later in the so-called T-O tradition of early medieval maps, in which the circular landmass of the oikoumenē is divided unequally into the three continents in a fashion resembling a T placed in its bottom half.⁶³ However, as will emerge below, Orosius’s viewpoint is slightly different.

Although six possible combinations exist for the order of the continents, only four are regularly attested: (1) Asia, Europe, Africa; (2) Europe, Asia, Africa; (3) Africa, Asia, Europe; and (4) Europe, Africa, Asia. The phraseology of Sallust and Orosius suggests that the first order had the authority of tradition, and it is found again in Isidore, among others.⁶⁴ I do not think that its coincidence with the alphabetical order of their names in Greek (Asia, Eurōpē, and Libyē) is of any great significance. The priority of Asia is easy enough to explain as both the largest and most easterly continent, nearest the rising sun. However, the general priority of Europe over Africa is puzzling. It runs counter to clock- (or sundial-)wise order, and if based on a viewpoint from the east it also runs counter to the left-right pattern of reading in Greek and Latin. Alternatively, if viewing from the east, as Orosius does (see below), the order Europe-Africa might reflect a preference for right over left, or from an objective viewpoint for north over south. Certainly the latter preference is seen in Julius Caesar’s famous opening description of Gaul in his Gallic War, written at the end of 52 BCE. Here, after declaring “The whole of Gaul is divided into three parts,” he describes its three constituent elements (the Belgae, Galli, and Aquitani) from northeast to southwest (1.3–1.7), which is certainly not the order in which he first encountered them.

Confirmation that Orosius’s anticlockwise treatment of the world (starting from the east) represents a common tradition emerges from observation of the organization of a somewhat earlier work, the Expositio (E), or Descriptio (D), totius mundi et gentium, that is, “Explanation” or “Description of the whole world and peoples,” composed in the late 350s.⁶⁵ This work is non-Christian (i.e., secular, if not explicitly pagan) and its author now anonymous, probably from the general region of Syria. Although originally written in Greek, it survives only in two Latin translations. It begins with the statement “We must first say what peoples are established from the east (oriens) to the west (occidens)” (D 2). It then proceeds to describe the peoples of Asia from the extreme east to Bithynia (D 4-E/D 49), then Europe from Thrace to Spain (E/D 50–59), and finally Africa from Mauretania to Cyrenaica (E/D 60–62), with an appendix on major islands east to west from Cyprus to Britain (E/D 63–67).

The second order—Europe, Asia, Africa—is also well attested. Its arrangement may be explained by its indebtedness to the Greek periplus tradition, as exemplified by Hecataeus.⁶⁶ Thus, Strabo’s Geography describes the world clockwise from Spain through Europe to the Black Sea, then southward through Asia to finish in Africa.⁶⁷ Similarly, Pliny the Elder begins the geographic books of his Natural History at the Straits of Gibraltar and treats the continents in the order Europe, Asia, Africa. However, he differs slightly in pursuing an anticlockwise description of Europe along its Mediterranean and Black Sea shores, before heading up the Danube and down the Rhine, followed by an anticlockwise periplus of the southern and eastern coasts of the Mediterranean and Aegean around to the Black Sea and across to Persia, followed by a periplus of Oceanusclockwise back around to west Africa. The same order of continental description is found in the Divisio orbis terrarum and Martianus Capella. In this company, the otherwise derivative Iulius Solinus stands out. His Collection of Memorable Things, written soon after 200 CE and almost entirely cribbed from Mela and above all Pliny, deviates from Pliny in starting at Rome.⁶⁸ His other significant innovation betrays a further unconventional perspective. While most Romans tended to view the world from a maritime viewpoint and hence used the term mediterraneus to refer to inland regions, Solinus is the first to invert this perspective and refer to the “Mediterranean” as just that, “the inland sea.”⁶⁹

The third order—Africa, Asia, Europe—is found in the Chorographia of Pomponius Mela, a Roman from southern Spain writing in the 40s CE.⁷⁰ Again eschewing Rome as a starting point, his work is organized as an anticlockwise periplus of the Mediterranean from the Pillars of Hercules (Straits of Gibraltar), followed by a clockwise periplus of the outer ocean back to the Pillars. His starting point may reflect a local perspective (as was common in Greek periploi of the Roman period),⁷¹ but it also has its own logic. As noted above, an anticlockwise periplus of the Mediterranean from Mauretania to Spain is found in the Antonine Itinerary too. The third order also occurs in one Greek writer with some pretensions to high style. The boundaries of the Roman empire are described according to that order by Aelius Aristides in his panegyric To Rome (82).

The fourth order—Europe, Africa, Asia—is best attested among Greek writers. As already noted, Strabo cites it toward the end of his Geography (17.3.4). More complex, however, is the organization of the geographic preface to the history written in Greek by the Roman citizen and equesAppian in the mid-second century. He manages to maintain an anticlockwise direction consistently throughout his account of the limits (horoi) of Rome’s empire. Like his close contemporary and fellow Alexandrian Ptolemy, he begins in Rome’s northwestern extreme (Britain), and by making two overlapping circuits he ends up back at his starting point. After Britain and the Atlantic coast of Gaul and Spain, he describes the provinces surrounding the Mediterranean in an anticlockwise periplus, and then the landward (desert and river) frontiers of Africa, Syria, and the Danube and Rhine provinces. The most famous proponent of the fourth order is, of course, Ptolemy. Even so, we should not allow the fact that he represents the pinnacle of Greco-Roman geographic science to seduce us into assuming that his choice was typical.

Orientation and Viewpoint

It is evident from the Peutinger map and from the geographic digressions in several Latin authors that Eratosthenes’s oikoumenē was the model onto which many educated Romans mapped their understanding of the world. However, since most of these authors were careful to use objective terminology (north, south, east, west) rather than subjective (left, right, up, down), it is seldom possible to determine whether they shared the northern orientation preferred by the theoretical geographers and seen, for instance, in the layout of the Peutinger map. Today, of course, we are accustomed to orientation toward the north and (in languages written in scripts derived from Greek or Latin) to reading from left to right. Other things being equal, these norms ought to favor description from north to south and west to east. Indeed, with the extra refinement of grouping coverage by continent (in the fourth order above, Europe, Africa, Asia), this is exactly what we find in the ordering of Ptolemy’s twenty-six regional maps, which begin with the British Isles and end with Taprobane (modern Sri Lanka). The general arrangement suits the left-to-right reading order of a papyrus roll. It is also uncannily similar to the organization adopted by Richard Talbert for the Barrington Atlas (2000) when he faced exactly the same challenge.

However, despite seeming so logical, this arrangement cannot be safely assumed to represent the contemporary norm in antiquity. After all, Ptolemy was deliberately attempting to produce an objective description of the known world—hence his decision to abandon the prime meridian through Alexandria (used previously for his celestial mapping in the Almagest) for one through the furthest westerly point he knew, the Fortunate Isles (modern Las Canarias).⁷² Those with less lofty aims might be more influenced by their everyday experience of the natural world. Accordingly, if they looked to the sky, they might reasonably imitate the sun’s daily trajectory, so that descriptions would start with the rising sun (the east) and end with the setting sun (the west). In addition, we might expect a southerly orientation and, perhaps, a clockwise order of description, following the arc of the sun (as viewed from the perspective of the Northern Hemisphere) through the southern sky. This orientation we do see in Egypt, where the sun and the river Nile, rising in the south, both had enormous religious and cultural significance. On the other hand, if they looked to the night sky, then the revolution of the constellations around the North Pole might suggest a northerly orientation, as followed by Ptolemy, who came to geography through astronomy. If they looked to the ground, it would be equally reasonable to follow the trajectory of shadows, as on a sundial, clockwise from west to east through a northerly arc.

In the medieval tradition, the east (the location of Paradise/Eden)⁷³ is frequently given pride of place at the top of mappae mundi. Earlier, among the classical Latin geographers, Sallust’s use of objective terminology in his account of the continents is unrevealing, but we are fortunate that Orosius is less objective. He explains (1.2.3): “This part [i.e., Asia] facing west touches on its right (a dextra) Europe, beginning below the north pole, while to the left it stretches out to Africa.” So Orosius is inviting his reader to imagine looking from an eastern perspective along thediaphragma, with Europe to the right (north), Africa to the left (south). This perspective is far from universal, however. Not only did Ptolemy and other theoretical geographers favor orientation to the north, but it is found too in the Expositio/Descriptio totius mundi et gentium, whose anonymous writer explicitly locates Egypt “to the left (de laeva parte) of Syria” (E/D 34) and Arabia “to the right (de/a dextris) of Syria” (E/D 38). By contrast, another late antique pagan author and Orosius’s approximate contemporary, Martianus Capella, in his digest of geometric knowledge shares the eastern orientation of the later medieval Christian tradition: he unequivocally puts Europe to the right of the Pillars of Hercules and Africa to the left (6.624). The fact that he offers his observations as part of a synopsis of traditional learning in an entirely secular context suggests that the orientation of the later T-O and mappa mundi traditions may represent a continuity with pagan antiquity (even if a coincidental one), rather than a new departure, and that Orosius is idiosyncratic in his westerly orientation.


The importance of this east-west axis (Levant to the Pillars of Hercules) to Greco-Roman geographic thinking was emphasized in a recent paper by Glen Bowersock (2005), where he offers a useful starting point but not, as he admits, the last word on the subject. Further evidence for his central thesis can indeed be found in an utterance placed in the mouth of the Roman emperor Justinian by his speech-writer (quaestor), Tribonian. This passage comes from the preamble to a new law on senatorial membership (Novel 62) issued in December 537; although meant as a historical statement, it has to be read in the context of the newly confident empire, resurgent after the reconquest of Africa and initial successes in Italy. Tribonian produced a neat formulation to describe the basic divisions of the Roman world into east and west and northern and southern shores of the Mediterranean: “Roman power extended not only to the risings and settings of the sun (non solum ad ortus solis et occasus) but also to both flanks (sed etiam in utrumque latus) of the orbis terrae.” Given the immediately preceding contrast with east-west, by latera (flanks) he clearly means the northern and southern halves of the oikoumenē, lying to either side of the diaphragma.

This said, I think that Bowersock pushes his evidence too far in claiming that the pivotal importance of the Mediterranean, or of a maritime reference point more generally, lies behind the use of the terms superior (upper) and inferior (lower) in Roman political geography. These terms were used in the imperial period to distinguish two parts of previously single provinces, as happened in Germany, Moesia, Pannonia, Britain, and eventually Cyrenaica (see fig. 7.2). This usage has sometimes been explained with reference to elevation from sea level, and that certainly works for the riparian provinces Germany, Moesia, and Pannonia, where the superior province is upstream from the inferior one. However, the explanation clearly does not work for Britain, where there is no obvious riverine reference point and the superior province lay in the lowland south, the inferior in the upland north. Nor will it work in terms of relative military force. While Pannonia Superior and Britannia Superior ended up with greater numbers of legions than their inferior counterparts, the first such provinces to be divided, Germany and Moesia, had equal establishments. It is tempting to imagine that superior might reflect proximity to the central diaphragma, but the stumbling block to this explanation is the order of the Pannonias.

Instead, Bowersock prefers a new nuance on the first (“upstream”) theory just mentioned. He would have inferior indicate proximity to the sea of reference, be it the Mediterranean, Black Sea, or German (i.e., North) Sea. Ingenious as his nuance is, he still struggles to explain the order of Britain and especially of Libya Superior and Inferior (the subdivided Cyrenaica), where superior lies to the west, inferior to the east, though both are equally Mediterranean. I contend that the real explanation is to be found in the history of the naming of Roman provincial divisions. In the republican period this was done with explicit reference to Rome as the viewpoint. When Spain was divided into two commands, it was natural that the eastern portion should be citerior (nearer) and the western ulterior (further), likewise that Gaul be thought of as being divided into one portion on the side of the Alps closer to Rome (Cisalpina) and another beyond (Transalpina). With regard to the question of the superior-inferior distinction, therefore, suffice it to observe that in every case the superior province is geographically closer to Rome than the inferior.⁷⁴ Such a Rome-centered perspective is hardly surprising in official terminology. It also helps to explain the phasing out of this system of distinctions during the fourth century CE. By then, the further subdivision of traditional provinces, combined with the multiplication of imperial courts (none of which might be based in Rome) culminating in the foundation of Constantinople, rendered the traditional binary divide obsolete. Superior and inferior were superseded by the more neutral and flexible prima and secunda, which allowed for the creation of a tertia or otherwise-designated extra subdivision, as in Britain (Britannia Prima and Secunda, Flavia and Maxima Caesariensis). This transformation in terminology had begun, but was not yet complete, by the time of the compilation of the so-called Verona List (discussed below), which may be dated to 314.⁷⁵


Aside from the republican and early imperial provincial designations, a Rome-based standpoint can also be detected in the ordering of two sets of official territorial units: the thirty-five voting tribes of the Roman republic and the nine regions of Italy in Augustus’s time. It is well known that the four “urban” and thirty-one “rural” tribes created up to 242 BCE represented territorial subdivisions of the populus Romanus, spanning central Italy from south Etruria to Campania and from the Tyrrhenian Sea to the Adriatic coast. It is also well known from scattered allusions that the tribes had a particular order. But this could not be established in its entirety until Michael Crawford recognized the significance of two bone tesserae in the British Museum and was then able to reconstruct it in a recent paper (2002). On these tesserae (or counters) the standard abbreviations for the names of two tribes—ROM for Romilia and OVF for Oufentina—are twinned with the numerals five and sixteen, respectively (see fig. 7.11). Putting all the evidence together, Crawford realized that the rural tribes had been numbered according to their positions along the major public roads leaving Rome, from nearest to furthest, and in an anticlockwise direction starting from the Romilia on the road to Ostia, Rome’s port at the mouth of the Tiber and first-ever colony, around to the Arnensis on the coast of south Etruria.

The location of the starting point southwest of Rome in this instance finds a counterpart in the southeast orientation of the marble plan of the city made around 200 CE and discussed in the previous chapter. It took the alignment of the Via Latina in the direction of the sanctuary of Iuppiter Latiaris in the Alban hills as its point of reference. This alignment corresponds with the starting point for the anticlockwise numeration of the fourteen regions of the city of Rome, established by Augustus in 7 BCE.⁷⁶ Crawford has suggested that the same anticlockwise sequence also conditioned the numbering of the regional subdivisions of Italy that were created in the first century BCE. The nine of these in peninsular Italy were numbered in an anticlockwise direction from Regio I, Latium et Campania, to Regio IX, Liguria (see fig. 7.2).⁷⁷

On a much more local level, a further paper by Crawford (2003) has noted the same tendency to anticlockwise description by those engaged in the definition of community boundaries, one of the jobs of the agrimensores (surveyors), a profession discussed by Michael Lewis in chapter 5 above. A treatise by Hyginus preserved in the late antique collection of their technical writings (the Corpus Agrimensorum) suggests the types of place where boundary markers (termini) might be placed, but it does not specify a particular direction to follow.⁷⁸ A famous early example of the definition of territory in this way is the bronze tablet dug up at a village called Pedemonte, north of Genoa in Liguria, in the early sixteenth century.⁷⁹ This inscription, known generally as the Sententia Minuciorum, preserves a judgment delivered in 117 BCE by the Roman commissioners Quintus and Marcus Minucius Rufus concerning the dispute between the Genuates (the Genoese) and their inland neighbors, the Langenses Viturii. Their territory straddled the Via Postumia, the Roman highway that ran northward from the coast over the Apennine watershed into the upper Po valley. The record contains an outlining (determinatio) of the locations where termini should be placed. It demarcates two concentric circuits: an inner one (indicated by the letters a to d in fig. 7.12) delimiting the ager privatus (private fields) of the Langenses, and an outer one (indicated by the numbers 1–15) fixing the borders of their ager publicus (i.e., common pasture and woodlands). In recent times two of the boundary markers have been identified, still very close to the sites dictated by the text (at nos. 13 and 14). Both determinationes begin at the south of the territory and follow an anticlockwise path.

FIGURE 7.11 The bone tesserae ROM/V and OVF/XVI (British Museum). From M. H. Crawford, “Tribus, tessères et régions,” Comptes rendus des séances, Académie des Inscriptions et Belles-Lettres 2002, p. 1127, fig. 1.

FIGURE 7.12 The approximate boundaries of private land (a–d) and public land (1–15) within the territory of the Langenses Viturii, as described in the Sententia Minuciorum: Corpus Inscriptionum Latinarum I² 584 = V 7749. Adapted by the author from M. H. Crawford, “Language and Geography in the Sententia Minuciorum,” Athenaeum 91 (2003): 208, map 2.

The same practice can be observed in a determinatio (or horothesía in Greek) issued in 100 CE by the governor of Moesia Inferior, Laberius Maximus, to the city of Histria (near the mouths of the Danube), as part of the settlement of its dispute with the imperial contractor who collected the customs dues of the “Thracian shore” (portorium ripae Thraciae).⁸⁰ Here, although the demarcation starts at the north, the order is again anticlockwise; a southerly start would in any case have been precluded by the coastline. It may be that with the Roman tribes, Italian regions, and outlines of the territories of the Langenses and Histriani, it was the major watercourse (the rivers Tiber, Porcobera, and Danube, respectively) that dictated the starting point in each case.

Moving from the local to the provincial scale, two principles of organization can be observed in the description of the province of Asia found in the Roman customs law for the province.⁸¹ Although this law only survives in a Greek version, inscribed at Ephesus in 62 CE or soon after, it is an indubitably Roman document (the preamble makes clear that the text is derived from official archives at Rome) comprising an accretion of regulations going back at least as far as 75 BCE. To this earliest layer belongs a definition of the area covered by the customs regulations (lines 7–8 §1), that is, the coast and the borders of Cappadocia, Galatia, and Bithynia, an order implying an anticlockwise description beginning and ending in the north. This order is subsequently confirmed by the detailed listing of maritime customs stations from the mouth of the Black Sea to Side in Pamphylia (lines 23–26, §9). A more subjective viewpoint is taken in a later section, dating from 17 BCE, that enumerates the “dioceses” (administrative subdistricts) of the province of Asia (lines 88–91 §39). Here, Ephesus, the chief city of the province, is taken as the primary point of reference, as the districts are listed first in a southerly direction (Ephesus, Miletus, Halicarnassus), then in a northerly one (Smyrna, Adramytium, Hellespontus), and finally in a westerly one (Sardis, Cibyra, Apamea, Synnada, Lycaonia). This enumeration may offer us a glimpse into the organization of the formula provinciae (register of the province) maintained in central financial archives at Rome, a category of source material generally assumed to have been exploited by Pliny the Elder around 70 CE for the geographic sections of his Natural History.⁸² If Pliny is any guide to these formulae, then they would appear to have organized different categories of material according to varying logic. For example, in his description of Gallia Narbonensis (modern Provence), Pliny lists the regions (regiones) occupied by tribal groupings in geographic order clockwise from the Pyrenees to the Alps, Roman colonies by date of foundation, and then towns of Latin status alphabetically.⁸³


On an empire-wide scale, we gain another glimpse into the geographic world-view of the Roman filing clerk with the so-called Maffeian column (fig. 7.13). Now preserved in the Vatican’s Gallerie lapidarie, this column originally stood in the Basilica Aemilia on the north side of the Roman Forum.⁸⁴ On it, under the heading nomina leg(ionum) (names of the legions), are inscribed the names of the thirty-three legions as they existed around 200 CE, disposed in three equal columns of eleven names. Each legion in the Roman imperial army always had its own number and epithet, both of which might be shared by other legions but not in the same combination (e.g., Prima Adiutrix, Secunda Adiutrix, Prima Italica, Secunda Italica). So, in principle, it would be possible to draw up a list in combined numerical and alphabetical order, as indeed the early third-century historian Cassius Dio does (55.23). In fact, embedded in the final layout of the column is a geographically organized document drawn up around 170 (see table).

Here the names of the three “Parthian” legions created in the 190s during the Parthian campaigns of Septimius Severus (two stationed in Mesopotamia and one at Albanum, south of Rome) have been added across the bottom (cols 1–3, line 11), while the names of II and III Italica, raised for Marcus Aurelius’s wars against the Marcomanni in the 170s (based in Raetia and Noricum), appear at the end of the third column (lines 9–10). Otherwise the remaining twenty-eight legions in fact form a coherent geographic sequence from II Augusta at Caerleon in South Wales (col. 1, line 1) to VII Gemina at León in Spain (col. 3, line 8). The legions are grouped by province, delineating the frontiers of the empire in a clockwise sequence circling Rome and the Mediterranean (as indicated in fig. 7.2): Britain (col. 1, lines 1–3), Germany (lines 4–7), Pannonia (col. 1, lines 8–10; col. 2, line 1), Moesia (lines 2–6), Dacia (line 7), Cappadocia (lines 8–9), Syria (col. 2, line 10; col. 3, lines 1–2), Judaea (lines 3–4), Arabia (line 5), Egypt (line 6), Numidia (line 7), Hispania Tarraconensis (line 8). As with the Claudian monument from Patara, the physical location of the column is in harmony with the focal point of the list displayed on it. Within each province the legions are listed upward by number, rather than according to any geographic sequence. So, for Britain, II Augusta (Isca/Caerleon) precedes VI Victrix (Eburacum/York) and XX Valeria Victrix (Deva/Chester). However, as is clearest from the ordering of the legions in the divided province of Germany, the logical priority of “Superior” over “Inferior” has been allowed to disturb the geographic logic, so that the army of the more southerly Germania Superior (VIII Augusta, XXII Primigenia) is listed before that of its northern neighbor Germania Inferior (I Minervia, XXX Ulpia).

FIGURE 7.13 The Maffeian column listing the Roman legions around 200 CE: Corpus Inscriptionum Latinarum VI 3492 = 32901. Vatican, Musei Vaticani, Galleria lapidaria inv. 7460. From P. Cosme, “Les légions romaines sur le forum: Recherches sur la colonnette mafféienne,” Mélanges de l’École Française de Rome—Antiquité 106 (1994): 172, fig. 1a.

Nomina leg(ionum)

The list’s starting point—Britain—is strikingly reminiscent of the near-contemporary Geography of Ptolemy. But influence from it is unlikely because the Maffeian column takes Asia and Africa in the order opposite to Ptolemy’s and, in fact, seems deliberately to ignore organization by continent when this principle could quite easily have been respected. Continental coherence had been similarly disregarded in the account of the disposition of Roman military forces in 23 CE by the senatorial historian of the late first and early second century Cornelius Tacitus (Annals 4.5). Although he began at an equivalent point (Germany; Britain still lay outside the empire), he chose to describe legionary deployment in an anticlockwise manner: from Germany to Spain, across to Africa, and round to Egypt, Syria, and Cappadocia, before crossing back into Europe with Thrace, the Danube provinces, and Dalmatia, to finish in Rome with the urban and praetorian cohorts. In contrast to both Tacitus and the Maffeian column, a Roman theoretical geographer might have preferred to begin with III Augusta at Lambaesis in Numidia, follow the legions around anticlockwise through Asia, and finish in Europe with Spain or Britain.


Official documents reflect the political fragmentation of the empire during the later Roman period in multiple viewpoints. A prime example is the Verona List, so called not because it was written in Verona but because it survives uniquely in a manuscript preserved in the chapter house library there. This list plausibly derives from an official record generated at one of the early fourth-century imperial courts or regional administrative centers. It comprises a list of the names of all the provinces of the empire around 314, organized according to the twelve regional groupings of them called “dioceses” (as mapped out in fig. 7.14).⁸⁵ However, although the twelve are presented here in a single list, they are not ordered in a single geographic sequence, but rather in two separate eastern and western groups, reflecting the respective spheres of responsibility of the emperors Licinius and Constantine during the period between Licinius’s defeat of Maximinus in 313 and his own defeat in his first civil war with Constantine in 316–317. Licinius’s (eastern) group of dioceses—Oriens, Pontica, Asiana, Thraciae, Moesiae, Pannoniae—precedes that of Constantine: Britanniae, Galliae, Viennensis, Italiae, Hispaniae, Africa. The split is apparent from the discontinuity midway in the list between the dioceses of Pannoniae and Britanniae. The eastern half of the list circles the Mediterranean neatly anticlockwise from south to north or, in continental terms, from Africa, through Asia, to Europe. The arrangement of the western half is less tidy, though it is approximately anticlockwise from north to south or from Europe to Africa, since the diocese of Hispaniae in fact spanned the Straits of Gibraltar to include the province of Mauretania Tingitana.

Within the dioceses of Licinius’s portion the provinces are listed according to geographic sequence, though not according to any consistent pattern. The internal organization of Constantine’s portion is less tidy geographically, but there is little evidence that geographic sequence has been subordinated to political hierarchy, as seen in the Notitia Dignitatum and the schedule to Justinian’s Novel 8, both discussed above. The priority given to the eastern half of the list contrasts with the Maffeian column (which starts in Britain) and is all the more unexpected in view of the difference in status between the two emperors at the time. As contemporary inscriptions from Licinius’s portion of the empire demonstrate, even though Licinius had once considered himself senior to Constantine, in this period he now acknowledged him as his senior.⁸⁶ In the Verona List, as we have it, the inversion of the expected order suggests that a special significance is being accorded to the eastern half of the empire, therefore, perhaps reflecting a place of writing or editing within Licinius’s realm. In fact, the list’s structure becomes far more comprehensible if the two halves are reversed. So either our copy derives from a version that suffered accidental dislocation in transmission; or it may preserve a deliberately manipulated “eastern” redaction of a document that in its original state did respect the contemporary political hierarchy as well as geographic logic. With a west-east (Constantine-Licinius) order restored, the list flows from Britain in a single anticlockwise sequence through western Europe, Africa, and Asia, around to eastern Europe. This result is decidedly more elegant, even if it no more respects the continental divisions than did the Maffeian column.

FIGURE 7.14 The provinces and dioceses of the empire according to the Verona List, ca. 314 CE. Map adapted by the author from original made by the Ancient World Mapping Center, University of North Carolina, Chapel Hill, and reproduced with permission.

Geography and History

The ancient writers who attempted to blend history with comprehensive accounts of geography set themselves a much tougher task than those that simply described snapshots of the world at a single moment, or those literary authors, such as Sallust, Appian, and Orosius, that confined their geography to discrete sections. A famous example is the speech (mentioned above) that was put in the mouth of Agrippa II by Josephus. Since the king had corresponded with Josephus during the writing of his account and had received a presentation copy of the Jewish War when it was finished, Josephus’s version of the speech may be more faithful to the original than is usual in Greco-Roman writing. In attempting to dissuade the Jews from revolt, Agrippa reminds the crowd in Jerusalem of the vast scope and military resources of the Roman empire. He begins his historical account of Rome’s expansion with the conquest of Greece and Macedonia, when Rome first penetrated the sphere of the Hellenistic kingdoms to which Judaea had also been subject, and thus when Romans are first likely to have impinged on Jewish consciousness.⁸⁷ He then traces a rather undisciplined quasi figure eight around the oikoumenē, describing Asia Minor, the Balkans, Gaul, Spain, Germany, Britain, and then jumping to the North African provinces before ending closer to home on the Euphrates.⁸⁸

In very similar fashion, a geographic tour of the Roman world is related as part of the personal history of a military veteran of the tetrarchic period, Aurelius Gaius, on the tombstone that he erected for his wife, Iulia Arescusa, near Cotiaeum in Phrygia around 300.⁸⁹ Having listed his advancement through the ranks, he says that he had “circled the empire” (tēn hēgemonian kykleusas), a claim that he then expands upon with a list of at least twenty-three provinces, two cities, and four regions beyond the empire’s borders: “Asia, Caria [Lycia?, Phrygia?], Lydia, Lycaonia, Cilicia [Isauria?, Armenia?], Phoenicia, Syria, Arabia, Palestine, [Egypt?], Alexandria, India (i.e., Ethiopia), [- - -], Mesopotamia, Ca[ppadocia, Pontus?], Galatia, Bithynia, Thrace, [- - -], Moesia, Carpia, [- - -], Sarmatia four times, Viminacium, [- - -], Gothia twice, Germa[ny, - - -, Darda]nia, Dalmatia, Pannonia, [- - -], Gaul, Spain, Mauretania, [- - -].” The list starts close to home, taking a clockwise sweep to the southernmost point (India, which probably means the Red Sea coast of Ethiopia), then an anticlockwise sweep passing through the easternmost point (Mesopotamia) to the northernmost (Germany), before beginning anew with an east-to-west account from the central Balkans to the westernmost points in Spain and Mauretania. As the references to repeated forays into enemy territory demonstrate, this is not a simple chronological itinerary, but represents a consolidated account that avoids repeated mention of any single locality. Rather, the list is artfully constructed to emphasise the extent of his travels in relation to his own starting point (the text gives his birthplace as Pessinus, also in Phrygia), which is close to that of the location of the audience for the tombstone in the territory of Cotiaeum. In this context, it is understandable that the toponymy of more familiar nearby regions (Caria, Lydia, Lycaonia) defines smaller units than that of distant places, which sometimes names dioceses rather than individual provinces (Gaul, Spain). The choice of the term “India” also suggests an attempt to emphasise the exotic.

Much more sophisticated is the often underrated Breviarium dedicated by the Roman official Festus of Tridentum to the inadequately educated emperor Valens around 370.⁹⁰ This pamphlet is nothing less than a crash course in geopolitical history from the foundation of Rome to the emperor’s current Persian campaigns. Despite its brief compass, Festus manages to weave an account of the rise of Roman power seamlessly into a roughly clockwise and extremely economical tour of the Roman world that ends appropriately with Valens’s current theater of operations.⁹¹ In retrospect, the work’s only fault was to encourage the emperor in his thinking that the recently defeated Goths were a spent force, a judgment that was to be proven dramatically wrong at the battle of Hadrianople (Edirne) in 378, with immediately fatal consequences for the emperor and (it might be argued) chronically terminal ones for the empire.

This survey of mapping in Roman texts closes with a work that may seem preternaturally unprepossessing as well as archetypally Byzantine rather than Roman in a traditional sense: the Synekdēmos of Hierocles, written in Greek and most probably in Rome’s younger sister and rival, Constantinople. It is a documentary text, comprising lists of cities organized by provinces and dioceses as they existed in about 500.⁹² It is offered as a description of “that part of the Roman world under the dominion of Constantinople,” but no further explanation of its purpose or organizing principle is ventured. Mapping out the data of this gazetteer (as in fig. 7.15) is not necessarily immediately enlightening. Given its relatively even quality throughout, the information plausibly derives from an official source of some kind, as did the earlier Verona List. The work has also been carefully executed.

In fact, the arrangement of the Synekdēmos reflects a synthesis of political and theoretical geography. Although Hierocles notes the status of each province’s governor, he still prefers a geographic sequence. Within each diocese, the provinces are consistently presented in a roughly clockwise order, where the Verona List had no consistent pattern. Moreover, within each province, after the metropolis, the cities are listed according to a recognizable geographic logic. At a higher level, too, the order of description is independent of the imperial government’s official understanding of the world. In the two centuries since the Verona List had been compiled, an extra tier of organization had been superimposed. In consequence, as already seen in the Notitia Dignitatum, the dioceses had by now been grouped into prefectures, each headed by one of the emperor’s praetorian prefects. In the later fourth century three such territorial prefectures had embraced the empire between them: (from east to west) the praetorian prefect of the east (Oriens); the praetorian prefect of Italy, Africa, and Illyricum in the center; and the praetorian prefect of the Gauls (Galliae) in the west. However, the division of the empire at the end of the fourth century, followed by the implosion of its western half before the end of the fifth, had left its eastern half divided into two very unequal prefectures, that of Oriens (based at Constantinople and controlling the diocese of Thrace as well as the four dioceses east of the Bosphorus), and the much smaller, two-diocese prefecture of Illyricum (based at Thessalonica). The division between these two prefectures is marked by the double line on the map.

FIGURE 7.15 The provinces of the eastern Roman empire ca. 440–500 CE, as cataloged in Hierocles’s Synekdēmos. Map by the author.

The prefect of Oriens was considered senior to that of Illyricum, and his name always took precedence in official protocols.⁹³ Accordingly, we might expect Hierocles to describe the whole of Oriens before turning his attention to Illyricum. However, after beginning with the province of Europa and then the rest of Thrace, he turns his attention to the prefecture of Illyricum before resuming the description of Oriens. Here, moreover, he does not start with the provinces of the Pontic diocese immediately over the Bosphorus from Constantinople, but in Asiana with the province of Asia.⁹⁴So not only has he transcended contemporary political geography by treating the dioceses of the continent of Europe together, but he has also sought symmetry by beginning the description of each continent with its eponymous province. The prioritization of Europe over Asia was no doubt influenced by the physical location of the capital on the west side of the Bosphorus. By understanding Hierocles’s rationale, it is possible to appreciate his work as a sophisticated blend of political and theoretical geography rather than a mere catalog. Given the cartographic nature of its arrangement, it is hard to imagine that his description was not influenced by consultation of a map.


The analysis of often superficially unsophisticated material, nestled in the interstices between more visually striking artifacts, can be utilized to reveal some underlying themes in the Romans’ perception of the world around them. Of course, what this comparison of a wide range of documents and authors reveals is not a single Roman worldview, but a range of them, reflecting variations in cultural background, personal agenda, and political developments over time. Despite the common use of the itinerary to describe sequences of places, even in this category of material there is some circumstantial evidence for familiarity with cartographic images, either in informing the organization of the material or for facilitating its comprehension. Within the documentation emanating from generations of Roman clerks, there is clear evidence of sensitivity to the actual spatial relationships between places cataloged and to the physical geography of the real world in the organization of data, even if such sensitivity often becomes subservient to respecting hierarchies of status.

Despite the variations, some consistent features emerge. One is the acknowledged significance of the sea (specifically the Mediterranean and the outer ocean) as the chief reference point for many textual surveys of the Roman world (including the Antonine Itinerary), even though starting and ending points and directions of description might vary. The fortuitous positioning of Rome and Italy near the center of the oikoumenē surrounding the Mediterranean meant that Romans did not necessarily have to adjust their Greek models to satisfy their own sense of self importance. Moreover, long after the publication of Ptolemy’s work with its more complete image of the oikoumenē, it is clear that the predominant model of the inhabited world onto which Romans mapped their understanding remained that inherited from Eratosthenes. Another consistent feature, despite varying preferences in orientation, is a noted tendency to describe space in an anticlockwise direction, regardless of whether the scope be as wide as the continents or much more limited. This tendency runs counter to expectations based on the direction of the shadow from a sundial’s gnomon and, in combination with the northerly orientation of astronomy, on the direction of reading in Greek and Latin. Because it occurs already in the ordering of the rural voting tribes, there is the temptation to suggest that it is a tendency conditioned by the priority accorded to the tribes along the road to Rome’s earliest colony at Ostia, as well as by a preference for describing the area to the south and east of the Tiber (traditionally Latin territory) before the area to the north and west of the river (traditionally Etruscan and hostile). For certain, such an anticlockwise perspective can be traced remarkably far back in the history of the Roman republic.

If you find an error please notify us in the comments. Thank you!