Any discussion of the Cosmographer’s methods and sources must turn to Ptolemy for comparisons as he and the Cosmography alone of our toponymic sources deal with the short-lived Flavian forts of Scotland. This allows us to assume that they (or their sources) treated the common material at their disposal in different ways. For Britain north of (and including) Hadrian’s Wall, the Cosmography lists 79 names of ciuitates, while Ptolemy has 21 names of πόλεις; the Cosmography shares 16 names with Ptolemy. The Cosmographer thus includes 76% of the names given by the earlier writer, although these shared names account for only 19% of the combined total of 84 names.
Of the names given by the Cosmographer for places in the north of Britain, the sixteen shared with Ptolemy must belong to sites occupied during the Flavian conquest of the north, some of which can be identified with sites only occupied at this time; in addition, he lists twelve of the forts on Hadrian’s Wall and ten of those on the Antonine Wall. Of the remaining forty-one names, it is not certain that they should be taken as representing forts all occupied contemporaneously. For instance, Habitanco (Risingham) does not seem to have been founded before the reoccupation of the Lowlands in AD 139-141 and is thus part of the Antonine system, while the majority of forts listed north of the Antonine Wall can only date from the Flavian invasion of the 80s, the Severan campaigns in the early third century or the Constantian campaigns almost a century later.
The map source, then, was evidently a composite of various dates; the original layer of information must have been Flavian and derived from the same source as Ptolemy’s (or Marinus of Tyre’s) information. Whether we should envisage a single official map being updated with new information in the Hadrianic period and again in the Antonine period without erasing the old information or hypothesise an intermediate compiler in late Antiquity using maps or lists relating to three separate phases of military activity in the north is not clear. However, this activity almost certainly took place before the Cosmographer acquired his material as the work of integration was too systematic to have been his own.
One period at which this could have taken place was during the Severan invasion of Scotland, particularly of reoccupation of the Antonine Wall was intended. This is the model preferred by Rivet and Smith (1979, 196). On the other hand, the poorly documented campaigns of Constantius I in northern Britain also provide the context for the compilation of a new map. In both these cases the map would have been prepared as a prelude to invasion to give background information on previous military dispositions in Scotland rather than incorporating the results of the new campaigns. Thus we lack the names of tribes encountered by Severus – the Maeatae and the Caledonii – and the main adversaries of Constantius I, the Picti. A date during the fourth century is perhaps more convincing for the composite map, as the name Augusta applied by the Cosmographer to London as if an honorific epithet was probably conferred during the visit of Constantius I in AD 305 (Frere 1987, 199).
Rivet and Smith (1979, 438) have seen the ethnonym *Pecti (or a territorial name *Pectia) behind the name <Pexa> of the Cosmography 10753. I have argued (above, 00) that this is mistaken and that the Cosmographer intended the name *Dexa (written with Greek -c- rather than Latin -x- at 10753 and at 1088 as <Decha>) to refer to the fort at Camelon. Sir Ifor Williams (in Richmond & Crawford 1949, 31; 43) discusses this name and suggests that it is in origin a stream name. The use of -c- to represent the guttural British sound now spelled ch in Welsh does not imply a Greek source: it is attested in the TEXTOVERDORVM (for *Tectouerdorum) of RIB 1695, a local Romano-British inscription. Even if we are forced to accept Rivet and Smith’s emendation, the appearance of *Pecti or *Pectia on an early fourth-century map would occasion no surprise.
For the island of Britannia and the seas surrounding it, then, the Cosmographer seems to have used only one source, a map probably dating from the early fourth century, but compiled from materials of earlier date, including information relating to the various short-lived military occupations of northern Britain. He took names from the map in a vaguely systematic manner, dealing with discrete regions in turn, but with no logic in his division of regions or necessarily in the ordering of names within them. There is no suggestion, disappointingly, that the regions were based on civitates or provinces. There are many names in his list which do not occur elsewhere but which can, on the basis of his ordering, be assigned to specific Romano-British sites or, where we are less certain, be suggested for a number of alternative sites within a relatively restricted area.