Readers of the above history will have noticed the near absence of discussion of the Notitia Dignitatum (ND). The ND is a document that was found originally bound together with a number of other texts all pertaining to Roman government and the army, and usually referred to as the Codex Spirensis. The ND itself purports to be a list of all positions in the administration of the Roman Empire towards its end, but assessments as to when exactly it was compiled differ: with dates ranging from 395 to the 430s, depending on the area covered in each chapter and whether the text was updated or left unchanged. For a long time, the British section of the Notitia has been seen as a document showing the disposition of troops close to the end of Roman Britain. A mention in the ND suggested that a fort remained in use until late in the sequence, and for much of the island, including the Saxon Shore, it is our only evidence for the military structure of the late Roman provinces. The proposed dates of composition thus focus around 408 (when it would represent the situation of the Roman army at the very end of Roman rule). But even if other suggestions are taken into account, for example that the system represents the situation in 395, and may have been updated until the 430s in places, there are difficulties in matching it with the archaeological record. There are a number of forts in the north that clearly remained in use to the end of Roman Britain, which are not listed by the ND, and the reverse is also true. Hodgson (2009, Northern Britain 28) has recently compiled a list of such problems in the North of England, showing that Ribchester and Housesteads are listed in the ND, but have archaeology that suggests they were abandoned well before 395, while Manchester and Lancaster remained in use, but are not listed. In addition it has long since been noted that the ND omitted any mention of Wales, something that is sometimes explained away as the loss of a page. But such problems are not restricted to Britain; they form a general failing of the document.
Kulikowski (2000) has recently argued that rather than portraying an actual state of play, the ND might have been a document created as a memorandum or wish-list of the government in Ravenna, an image of what a restored Empire might look like, drawing on old records, but not laying claim to current accuracy. If this were the case, it would fit well with another document in the Codex Spirensis, the ‘de re militari’, substantial parts of which appear to consist of great ideas that ought to be implemented at some point in the future. Kulikowski’s argument will no doubt be bemoaned by many, as it robs Late Roman archaeology of a heretofore useful source, but it would certainly go a long way towards explaining the numerous problems of the text. For the moment it might be safest to suggest that if a site is documented in the ND, it may have existed at some point in the fourth century, but it may be well worth checking against the archaeology whether the information can be deemed accurate.