Historically, there is a tendency to place the unattested civitates of the Ravenna Cosmography proximate to preceding or following entries. Much of this derives from the long held, but unjustified conviction that an itinerary from a Roman road book or map was related. That conviction has lead too many to overlook the most basic aspects of the cosmographer's method.
I chose to first examine the cosmographer's methodology. His own words indicated his source was a map. His choices and their order in the manuscript were certainly influenced by his source and its presentation. Once his method was characterized it could be used to qualify past and future assignments. In absolutely known circumstances, what did he do? Was he repetitive? What elements were common to his choices? What was the geographical relationship between the choices?
To characterize the methodology, the reliability of historical place name assignments must first be judged. Some are quite absolute, confirmed by a variety of archaeological, linguistic and literary remains. Some lack critical elements in their proofs, but can be roughly located in relation to other recognized sites or areas. The remainder must be set aside for further consideration. The most certain assignments, once categorized, can be plotted and their geographical relationships ascertained. Jones and Mattingly* Map 2:14 illustrates the absurd path taken by the Rivet & Smith* assignments. (See their accompanying text and criticisms on page 33.)
The
first observation was a simple one. All of the known sites listed prior to Nauione
Brough-on-Noe (R&C 106) are south of
the latitude [~53° 20' 27"N]
of Aquis Arnemeza Buxton (R&C
107),
those that follow are north. Next was the stunning realization that the area
south of the Buxton line could be divided into fairly uniform rectangular areas
that contained discrete sequences of the text. (This was thoroughly
unexpected, a chance observation while considering Quadrant 8.) These areas are something less
than two standard degrees in breadth and one degree in depth, but significantly
greater than the degree of Posidonius employed by Ptolemy. The area north from the
Buxton line to Hadrian's Wall is also divisible, east from west, into two
discrete sequences twice the depth of the southern areas.Outlines of latitude and longitude? Demarcation of sectional maps? Cadastration? Any remains to be proven, though I am heavily inclined toward the first and believe the second to be correct, also. Far from the finish, there appears to be one more level of division in the sequences. I believe that the areas, henceforth 'quadrants', can be further divided into four 'quarters' of the same proportions - NE, NW, SE and SW. This is not so readily demonstrated, but nevertheless is a component of my working hypothesis.
Can we expect an absolutely definable and predictable methodology? I don't believe it is prudent to expect anything more than to identify the cosmographer's habits. Short of recognizing his pattern of inventory in another situation, we must presume that his method was an adaptation suitable to the problem at hand.
Anyone that believes the cosmographer was erratic in his listings should pick up a 500 year-old road-map (the Gough Map will suffice) and inventory a wide-ranging sample of place-names with the expectation that your method of choice and the order of presentation will be perfectly understood more than a thousand years later, and that you will not be criticized for being unpredictable in the sequence of your record.