THE NORTHERN FRONTIER AFTER A.D. 369

Abstract

The starting point must be Bury's article of 1920, which suggested (albeit somewhat equivocally, when it came to Hadrian's Wall) that the Notitia Dignitatuni represents an up-to-date statement of imperial dispositions at a date after 425 a.d., and therefore that Britain was still a part of the empire at that time'1*. This flatly contradicted the evidence available from other sources. In seeking to reply, Collingwood found that the most coherent base from which to launch a counter-attack was the line of Hadrian's Wall, directly attested in the Notitia in the section per lineam vaiii of the command of the dux Britanniarum. Since he could find no coins minted later than the reign of Gratian from any Wall structure, he argued that Hadrian's Wall was not held after the usurpation of Magnus Maximus'2). He suggested that it was in fact the latter who had withdrawn its garrison, and that Hadrian's Wall no longer constituted a frontier line after about 383. With a firm date thus apparently established, 383 was henceforth accepted for the end of the military occupation of Hadrian's Wall. The date could thus be allowed, without discomfort, to round off the chronology derived from the 1929 excavations at Birdoswald'3). But of course it left a lacuna in the defensive system of northern Britain, between 383 and 410. Attempts were made to fill this lacuna in various ways. Collingwood himself was driven to suggest that, since the Wall was given up in 383, the section per lineam valli of the command of the dux Britanniarum had become out-of-date by the early fifth century, but that the rest ofhis command, consisting of equites and numcri stationed on the roads leading north from York, was still intact, forming a defensive screen which survived to the end of the Roman occupation'4). The suggestion was not very convincing, since it left completely without defences the whole of the western flank, that area precisely which lay open to attack by the Scoti of northern Ireland. A different solution was proposed by Richmond'5) and elaborated by Hunter Blair'6*. In the king-lists of two of the British kingdoms which could be located in southern Scotland, a number of names of apparently Latin form were noted. The individuals could be approximately dated to the late fourth century. In the king-list for the kingdom of Strathclydc, the names were Cinhil and