[ my homepage (in Dutch) | Contact]

Ptolemy's World: index map

( Click on map to zoom in. )

World map Europe West Asia East Asia Africa W. Indian Ocean E. Indian Ocean

This map shows the world map from Antiquity designed by Alexandrian geographer Claudius Ptolemaeus (Ptolemy, ca. 160 CE).
In his book Geographia he gave geographical positions (longitude and latitude) for some 7,800 cities, villages, peoples, mountains, rivers and other geographical landmarks.

The numbers he gives for geographical positions can be used as input data to create computer generated maps of the Antique world as Ptolemy saw it.
This is how I made sets of 6 small-scale maps, 24 medium-scale maps, and 96 larger-scale maps, on a uniform scale, that together show Ptolemy's world. Where necessary to show all the detail, I will also prepare partial sets on a still larger scale.

The maps presented here are only an unpolished first draft. They still have hardly any geographical names on them, and the coordinates used for input need checking [note 1].
In time, I also hope to add larger-scale maps to show particular areas in greater detail.

I also would like to make a coherent set of large-scale maps that can be glued together to make a single, detailed, large-format wall map, such as Agrippa, a contemporary of emperor Augustus, is said to have installed [Berggren, 47]. Given that in some areas Ptolemy has up to twenty names per square degree, a total map size of even 3.6 meters wide and 1.6 meters high (which would correspond to a scale of 2 centimeters per degree) would hardly be large enough to accomodate all names. [note 2]

Density of named items on Ptolemy's world map.
Above: World, absolute numbers in 1 × 1 degree boxes [note 3]. Below: close-up of Europe (½×½ degree boxes, with slightly different colour coding).

item density
item density

It might also be a nice idea to add the (Roman) roads network, as shown on the Late Classical Tabula Peutingeriana, and specified in itineraria like the Itinerarium Antonini (text), or the Parthian Itinerarium by Isidorus of Charax [note 3a]. The land and sea routes to China and East Africa, described in book I of the Geography, might also be included.

Specimen of a Ptolemaic map with Roman roads indicated.
Note that Atuatuca (today's Tongeren) and Nemetacum (Arras = Atrecht) had to be moved from their Ptolemaic positions because otherwise no sensible road map could be constructed.
map with roads

Structure of the Geographia

Ptolemy's Geographia consists of eight books. The middle six books are the atlas. The contents is as follows:

Some manuscripts have maps with them, most of them a set of 27 corresponding to the 26 maps ('pinakes') described in the text, plus a world map. A few manuscripts have a set of 68 maps for which there seems to be no clear basis in the Ptolemaean text. This set of 68 consists of 63 regional maps (corresponding to either one chapter in the text, or a combination of a few short chapters), plus one world map and four 'quadrant' maps (Europe, Asia, North Asia, and South Asia). [note 4]

Text tradition

The extant manuscripts are not the documents written by Ptolemy himself (his 'autograph'), but are later copies. In this process of copying a text so full of numbers and unfamiliar names it was inevitable that scribes introduced errors. Sometimes this was by accident (to an Alexandrian or Byzantine scribe the name of a Western European village would not have been familiar). Errors were also introduced in attempts to "correct" errors and "improve" (emendate) the text.

Today, some fifty manuscripts are known. Only a dozen or so are really important. An analysis of the extant manuscripts (codices) shows that they fall into two 'families' (or three), the Xi and Omega traditions. [note 5] (Note that major manuscripts are conventionally indicated by Latin or Greek capitals). The oldest extant manuscripts in each group were written about 1300 CE or slightly earlier.

1. The so-called Xi text tradition ('recensio Ξ')

X reflects a very ancient text. [note 5ab] Apparently, at one moment this text was deliberately revised by scholars to solve inconsistencies: in many geographical regions (for example, Spain, Crete) the coordinates in the Xi tradition differ systematically from those in the non-Xi, or Omega, tradition. This revision must have taken place before 300 CE, probably in the then 'scientific capital' Alexandria. [note 5b] It is possible that Ptolemy himself already started the revision. [note 6] It may also have been later: Mittenhuber [p. 93] links the revision to a third century transition from papyrus scrolls to parchment books, which allowed larger maps to be included in the texts.

Crete, X version Crete, UKVR version

Ptolemaic map of Crete, X version (left) and Ω version (right).
The differences are obvious. Data from Stückelberger. (Larger pictures in note 7a.)

From the revision originated the second, or Omega text tradition. This text tradition consists of two (closely related) families of texts: NKU and VRCW. By the way, not all differences between the Omega and the Xi tradition are due to the revision: some differences will simply be due to copying errors (for example, lines of text inadvertently lost in one or the other traditions).

2. The so-called Omega text tradition ('recensio Ω')

The Omega text tradition divides into two families:

The split between the Delta and Pi groups is dated to the fifth or sixth century CE. [note 7b] It may have been the work of Byzantine scholars (the so-called 'Byzantine revision'). [note 7ab]

It should be noted that there was no absolute separation between the three families: Some manuscripts were partly copied from one manuscript and partly from another, while other manuscripts have been corrected by comparing them with a manuscript from another family,

The three groups share certain remarkable errors. For example, in England they all have 'Iamesa' instead of 'Tamesa' for the river Thames. Another example is 'Belticè' in stead of correct 'Belgicè' as the name of the northern part of Gaul (cf. 'Belgium'). It seems unlikely that Ptolemy had these names wrong, but it is very probable that an Alexandrian or Byzantine scribe, who was not well acquainted with the local topography of Gaul, mistook a letter Γ (Greek gamma, 'G') for a T, or a T for an I. Such common errors strongly suggest that all three groups derive from a single common ancestor, the so-called archetype [Cuntz, pp. 15, 45].

Around the year 1405 Jacopo d'Angelo (Jacobus Angelus) made a Latin translation of the Geography, which was published in 1475. It took amother 58 years before the text of the Greek original was published. This was the first of only six complete Greek text editions that have ever appeared, five of them long ago (Erasmus 1533 and reprint 1546, Montanus 1605, Bertius 1618, and Nobbe 1843-1845).

It is important to note that these Greek editions were not yet based on a careful comparison of the various manuscripts, or a proper understanding of their mutual relations. In fact, Erasmus and Montanus based their text on only one manuscript, Γ (which was a copy of v). Bertius did slightly better by also using Z. However, this was not much of an improvement because, as we have seen, both Γ and Z are comparable: both manuscripts are intermediate between the X and VRCW families. A diferent source text was only available in the guise of d'Angelo's Latin translation, wich was based on the manuscripts Σ and B (B was formerly called Φ). [note 8]

Besides the complete Greek editions that have been mentioned, twice an attempt was made to publish a complete and "critical", that is scholarly sound, edition (Wilberg & Grashof 1838-1845, Müller & Fischer 1883-1901), but these two initiatives remained unfinished. However, there have been critical editions of various smaller parts of the Geography, in particular Germania (Cuntz 1923, among others) and the Far East (Renou 1925).

Until recently, it was unfortunate that there was no recent scientific text available, especially because to nineteenth century text editors only part of the manuscripts were available, and their mutual dependance was not well understood. It was therefore impossible to properly "weigh" the significance and reliability of the various manuscripts. For example, where manuscripts showed mutual differences Nobbe's policy often seems to have been to simply choose the majority version. [note 9] From the introduction to Wilberg's edition it is clear how difficult it was for him to get access to any manuscript outside Paris. Even at the brink of the twentieth century Müller's edition, admittedly "the best available critical text" (apart from the chapters that were later treated by Cuntz), wrestled with the problem: "An inadequate classification of the manuscripts resulted in [...] citing readings from unimportant copies while omitting many important variants in the principal ones." [Berggren, 52]

Until recently the best text editions were:

Wilberg and Müller also present the Latin text; for books VII and VIII one may consult a Mercator edition of 1555 (Google Books).

However, now a comprehensive scholarly edition has been published:

A supplementary volume with in depth studies has also been published: Klaudios Ptolemaios. Handbuch der Geographie. Ergänzungsband. Mit einer Edition des Kanons bedeutender Städte. Alfred Stückelberger, Florian Mittenhuber (eds.) (Basel: Schwabe, 2009). [review]

The new Geography text is primarily based on the five best manuscripts, UKVR from the Omega text tradition, and X from the Xi tradition. The running text reflects the majority of the four Omega manuscripts. Where the X text differs, its reading is added between parentheses. If needed secondary manuscripts are also taken into account, viz. A E Z v (semi-Xi text tradition) and B C D F L N O (Omega tradition). Apparent copying errors common to all manuscripts have been corrected (contrary to Cuntz).

A full translation in German is given, that also offers modern equivalents of place names and other toponyms. Finally the book has a set of 26 beautiful regional maps as well as a world map in three projections, based on the newly constituted text, but drawn in the tradition of the medieval manuscript maps.

A cd-rom accompanies the book, offering a searchable database (freeware) of Ptolemy's Geography. Scans of the manuscripts KVRXA should be available at www.philoscience.unibe.ch/ptolemaios (however, I get error 404).

It is only fitting that the Stückelberger edition was published in Basel, the city where Erasmus in 1533 published the first Greek Geography edition ever. The publishing company Schwabe rightly deserves praise for this new initiative (p.8)!

For my maps I have generally used Wilberg's and Nobbe's editions (available on Google Books, though a few pages are missing). Where available (parts of Western Europe) I also used Cuntz.
However, it is my plan to update my data using the Stückelberger edititon.


Various notes

Spelling. — The maps give geographical names in their conventional Latin form, as given in the Wilberg and Müller editions. Among standard transcription conventions (Greek to Latin) are
-ει- → -i-; therefore final -εια → -ia (thus Alexandria, Antiochia), but sometimes -ea (Caesarea, Apamea);
-οι- → -oe-;
final -η → (mostly) -e, but sometimes -a (Roma, Corduba);
final -ον → -um;
final -ος → -us (thus: Lemnus), but with river mouths the Greek genitive -ος corresponds to Latin -is;
-ου before vowel → -v-.
In a few cases transcription rules for the Latin speaking West differ from those of the Greek speaking East, for example Greek phi = "f"(West) or "ph" (East); initial rho = "r" (West) or "rh" (East). In cases where a Greek word has another gender than the Latin equivalent, this has of course consequences for name endings (Greek ’ορος ...ον → Latin Mons ...us). Where in the Greek text people's names are given in the genitive (-ων), this may correspond to a Latin nominative ending in -i, -es, or even -ae; I have followed the traditional Latin rendering. Evidently corrupted names I have felt free to restore (Belgica, Alba Fucens).

Measures of distance. — From book VII.15 and elsewhere it follows that Ptolemy assumed the circumference of the earth to be equal to 180,000 Greek stadia, that is, one degree corresponds to 500 stadia. In the Ancient world several stadia coexisted, each with a different length. This causes some ambiguity, but modern scholars generally agree that a typical length of 185 meters is a good approximation. This would make one degree equivalent to 92.5 kilometers, which is about 17 percent short.
The Roman mile (mille passus or '1,000 steps') was 1478.7 meters, which makes it almost exactly equal to eight Greek stadia. The Gallic mile (or leuga) was 1.5 Roman miles, or 2.22 kilometer.
The Egyptian schoenus and the Persian parasang were both reckoned to be equal to 30 stadia, or 5.55 kilometers (roughly, one hour of walking).

Projection. — The original map projections described by Ptolemy are relatively simple to draw by hand, but have the disadvantage of quite serious distortion. I therefore decided to use more modern projections for my maps: the conical projection with one true parallel for the large-scale maps, and the Bonne projection for the small-scale maps.

Language. — The question is, should the maps be presented in Greek, Latin, or English? Greek dropped out because the writing system would be unreadable to most people. Initially I used Latin designations, but it gradually became clear that these might obscure simple meanings. For example, Mare Rubrum is the well-known Red Sea, and Accipitrum Insula simply means Hawks' Island. As the Latin designations are themselves literal translations of the Greek, I felt free to use English equivalents, the more so as the Latin rendering of Greek words varies quite illogically. For example potamos (river) is often translated as fluvius, but also as flumen or amnis. And polis (city) becomes either civitas (Roman city) or oppidum (settlement), dependent on the interpretation of the translator. I therefore chose to use English descriptions. When in doubt, I give both the English and Latin name.

Major cities. — Major cities are emphasized on the maps. Ptolemy has a list of major cities in chapters VIII.3-28. I have assumed that cities marked 'metropolis' (except in Egypt, where this indicates a provincial capital) or 'royal' (Greek basileion, Latin Regia) were major too.

A comparison with modern maps. — Below, Ptolemy's map is shown superposed on a modern world map. To get the best fit, Ptolemaic longitudes have been reduced by 29%. Red 'nails' or arrows show residual displacement between the two maps (blue circle = Ptolemaic position, red end = modern position).
It is remarkable how well longitudes of prominent features in the entire longitude range correspond to modern positions (Spain, Italy, Greece, Cyprus, Persian Gulf, Sri Lanka, Malaysian Peninsula). Latitudes however are not seldom off by five or more degrees (Algerian coast, Black Sea coast, Ireland, India). This is remarkable, since one would expect latitudes to be relatively easy to measure experimentally. It is also remarkable that inner seas are often drawn too large (Black Sea, Caspian Sea, Persian Gulf, Western Mediterranean). Far-off areas, such as West Africa, Scotland, or China, are less well fitted, but this is of course as expected; red nails here are scarse and not very reliable due to the difficulty to identify Ptolemaic features. [note 10]

Ptolemy's Map Superimposed on Modern World Map

 


Notes

1. [ ^text ] For some parts of the map I initially used the positions given in a version of the Geography that was printed in Venice, 1511, pictures of which are available online. However, I then discovered that the authors of this edition had "corrected" nearly all the positions as given by Ptolemy - the printed numbers therefore reflect early 16th century geographical knowledge. For example the shape and position of the British isles is very different from Ptolemy's original data.
Of course I should have been warned: the full title says it all - Claudii Ptholemaei Alexandrini liber geographiae cum tabulis et universali figura et cum additione locorum quae a recentioribus reperta sunt diligenti cura emendatus et impressus [= The geography book by Claudius Ptolemaeus the Alexandrian, with maps and a picture of the world, and with the addition of positions recently found, accurately corrected and printed].

2. [ ^text ] Berggren (p.108) notes that such a "monumental size" of the map would also be implied by the size of a text that Ptolemy seems to have meant as a "summary caption" to his full world map: three full pages in print (Book VII.5; Berggren, 108-111). - Compare Strabo (Geographica, II.5.10), who proposed that a three dimensional Earth globe ought to be at least ten feet in diameter (!) to "properly contain all the regions of the habitable earth, and present an accurate view of them" [Stückelberger, I, p. 113 n. 128].
[add Stückelberger, Ergänzungsband, pp. ...].

3. [ ^text ] At the equator the surface area of a 1 by 1 degree box is one square degree. At higher latitudes the area of such a box is less, and therefore the effective density per square degree of surface area is higher than shown in the graph. This is because towards the poles the distance between meridians decreases as a cosine, cos (φ), where φ denotes the latitude. - As might be expected, the area with the highest density of named features coincides with the Roman Empire: the iso-line of 1.0-1.5 features per square degree is a fair approximation of the border of the Empire.

3a. [ ^text ] See Wikipedia lemma Isidore of Charax; English translation and commentary; Greek text with Latin translation and commentary by Karl Müller, Geographi Graeci Minores (Paris 1853) I, 244-256.

4. [ ^text ] Five manuscripts have the set of 68 maps: codex O (formerly called Ω) (Flor. Laur. XXVIII 49; 14th century), and four daughter manuscripts of O (Ω). The position of this group in the manuscript family tree is not clear. Diller (1966) finds the source of O/Ω "uncertain". Berggren & Jones do not mention the manuscript at all. Cuntz has claimed that it was derived from U. Stückelberger (Ergänzungsband, p. 22) takes O to be offspring of both U and a manuscript from the Xi tradition. But if so, why was the 27-set in U changed into the 68-set in O? Berggren & Jones suggest [p. 46] that this was done "for the convenience of those wanting to fit the Geography in manuscripts with smaller page dimensions, which necessitated more maps representing smaller areas." However, the way this was implemented would not solve the problem for maps for which the problem was most acute, namely one-chapter maps with the most names crammed unto them, like Italia, or India West of the Ganges. The real solution to fit overcrowded maps into smaller-size sheets would have been subdivision of these maps into several partial maps (for example, Northern Italia, Central Italia, and Southern Italia), but this was not done.
If each chapter in the Geography would be given its own map, the total number of maps would be 84-89 (there is some ambiguity whether Thule, Vindelicia, Euboea, Meroe, and Armenia Minor should be included in the chapter count). In the 68-set therefore about one third of all single chapter maps are combined with another map. A list in Stückelberger, Ergänzungsband, pp. 53-55, shows the combinations made: Raetia + Vindelicia + Noricum; Pannonia Superior + Inferior; Italia + Corsica; Iazyges + Dacia + Moesia Superior & Inferior; Thracia + Chersonnesus; Asia Propria + Lycia; Galatia + Pamphylia; Cappadocia + Armenia Minor + Cilicia; Colchis + Iberia; Syria + Palaestina; Arabia Deserta + Babylonia; Parthia + Carmania Deserta; Hyrcania + Margiane; Bactriana + Sogdiane; Aria + Paropanisades + Drangiane + Arachosia; India extra Gangem + Sinae.
At the beginning of the eighth book of the Geography yet another number is mentioned: 73 'countries' (χῶραι) - 22 in Europe, seven in Africa, and 44 in Asia. Those numbers are reached when chapters are combined where the region's 'first name' is the same, for example Gallia (Aquitania + Lugdunensis + Belgica + Narbonensis), Pannonia (Superior + Inferior), and also a few other plausible combinations are made (Albion + Hibernia + Thule, Thracia + Thracian Chersonesus, Achaia + Peloponnesus + Euboea) [Stückelberger, p. 767, note 1].

5. [ ^text ] Stückelberger, Ergänzungsband, pp. 10-25. Cuntz, pp. 1-41 (with supposed family tree, p. 14), saw two families, assigning U (and Z) to the VRW-group. With increasing knowledge of the N and K manuscripts the NKU-group was recognized to be a separate family: Diller (1966), pp. x-xv; Berggren & Jones, pp. 43-45. - The present lettering system for manuscripts (Paul Schnabel, 1938; Stückelberger, Ergänzungsband, pp. 12-20) goes back to Wilberg. It may be noted that manuscripts denoted by a letter from the beginning of the alfabet (A, B, C,...) are virtually absent from the set of important manuscripts. This reflects the fact that the manuscripts most easily available to early editors are now known to be of only secondary importance.

5a. [ ^text ] "eine recht unsorgfältige Abschrift einer wertvollen Vorlage" {Stückelberger, 39; also pp. 27, 33).

5ab. [ ^text ] A papyrus fragment from the early third century fits in with the X text tradition, not the Omega tradition (papyrus Rylands 522, only fifty years later than Ptolemy himself): Stückelberger, Ergänzungsband, p. 143.

5b. [ ^text ] Mittenhuber in Stückelberger, Ergänzungsband, pp. 34-108; see also the family tree ibidem p. 22. The dating is based on a careful comparison of text manuscripts, accompanying maps, and references by various Classical and Medieval authors that are evidence for the coexistence of several traditions at an early date.

6. [ ^text ] Cuntz has suggested that the split between X and the revised text goes back to Ptolemy himself, who after publication of his book would have continued to revise his data: "scheint es mir jedoch fraglos, daß der Archetypus auf das Handexemplar des Ptolemaeus zurückgeht. [...] Kurz: diese Varianten stammen von keinem andern als Ptolemaeus, sie standen in seinem Handexemplar" (To me it seems evident that the archetype goes back to Ptolemy's working copy... In short, these variants go back to no one else but Ptolemy himself, they were in his working copy). [Cuntz, p. 15-16] Stückelberger (p.24) agrees there was a revision, but thinks it is not settled whether this revision was Ptolemaic or later: "ob diese [Überarbeitung] noch von Ptolemaios in Angriff genommen worden ist, bleibt allerdings ungewiss" (whether this revision was undertaken by Ptolemy, remains unclear").

7. [ ^text ] Cuntz [pp. 4-5, 42] dated the manuscript (which he called Ur, not U) to the 11th century ("nicht später", not later), but such an early date is no longer accepted.

7a. [ ^text ] Ptolemaic map of Crete, X version:

Crete, X version

Ptolemaic map of Crete, Ω version (manuscripts UKVR):

Crete, UKVR version

7ab. [ ^text ] 'Byzantine revision': Berggren, p. 43.

7b. [ ^text ] Mittenhuber in Stückelberger, Ergänzungsband, pp. 34-108.

8. [ ^text ] Σ and B (formerly called Φ) (Laur. XXVIII 9 and 38; Florence) are two very similar texts ('twin texts'), copied from a hotchpotch of sources. Book I-II.11 was copied from X. The remaining part of book II, chapters 11-16, was copied from G (an incomplete manuscript, now in Paris, which itself was a copy of Z). The source for the next three books (III-V.18) is not known; Diller suggests that it may have been a part of G that is now lost. The final part, books V.19-VIII, was copied from Ω, a Florentine manuscript that itself was based on three different sources: an unknown manuscript (V-VII.4), N (VII.5-VIII.25), and Z (VIII.25 till the end).

9. [ ^text ] This is suggested by his use of source marks 'Z' (= 'text common to all five printed editions of the Greek text') and 'Y' (= 'text common to four printed editions, excepting Wilberg'; this should not be interpreted as 'four out of five', but is used where the Wilberg edition had not yet been published) [Index criticus, III, 191]. However, a common text hardly bears weight when all editions are based on the same manuscript. Even where many manuscripts are inspected, as Müller did, a majority of lower-quality manuscripts does not outweigh a single manuscript of superior quality.

10. [ ^text ] For more than 3300 Ptolemaic features I collected a corresponding modern position. Using these pairs of positions, a least squares fit yields the following relation:

This amounts to a reduction of Ptolemy's longitudes by 28.8%.
Usually, maps comparing Ptolemy's map and a modern world map are drawn without reduction of the Ptolemaic coordinates, see e.g. Stückelberger III, 267,
Sheglov (in Russian). Each red arrow on my map is an average over a five by five degree area. In Italy, Greece, or Asia Minor averages are calculated over many features (often more than fifty), while in far-off areas averages usually represent only one single datum and may therefore be far less reliable.
Up to date tools for identifying Ptolemaic sites are the Barrington Atlas, Pelagios (info), and Pleiades.
The modern map was created using a Versamap-4 geographical database.


References

Online resorces


(19 December 2011; last modified 5 July 2017)