key of the roads to the north. One is reminded of various analogies along the Rhine—of Moguntiacum (Mainz) lying on the left bank directly over against the mouth of the Main, and of Vetera (Xanten) on the same bank confronting the entrance to the valley of the Lippe. And it is worth observing that to-day two railway lines converge within the very area of which we have been speaking—the main line from England and the branch that runs past Earlston and eastwards through the Merse towards the sea.

It is interesting to note that the position of the great camp must have conformed to the rules laid down by Roman military writers. It stood upon a slight rise, thus permitting the general to survey the whole interior; and it was near at river, which ensured an ample supply of water, but yet so high above it as to make inundation impossible.[1] No doubt the site was to some extent overlooked from the heights on the further side of the Tweed, but it is not improbable that the necessary precautions were taken to counteract this weakness. On the top of the hill beyond the river is a small oval fortified enclosure which may have been a Roman outwork. It occupies the edge of the declivity, pitched high enough to look down on the Tweed and the opening of the Leader valley. Its situation suggests that its purpose was to hold in check any movement from the northern uplands that might have been directed against a force passing over the river. In the spring of 1909 some trenches were cut across this enclosure, when it was ascertained that it had been defended by a single ditch, and that it had had two entrances, one on the east side, the other on the west. The ditch was V-shaped, but towards the bottom for a depth of ten inches the sides became perpendicular. This was noted in a portion cut through the rock to the south of the east gate. There were no traces of buildings in the interior; the only relic found was a small piece of pottery of an orange yellow colour, which came from the ditch at a depth of four feet. It was much finer in character than the ordinary native pottery, and although the surface was much injured by the action of the soil, there can be little doubt that it belongs to the Roman period.

The plan of the great camp, and more particularly the character of its gateways, would be sufficient to prove its Roman origin, even were other evidence awanting. It differs entirely from those annexes which are so commonly attached to Roman forts in Scotland and elsewhere. It must have been constructed for the accommodation of a large, probably a legionary,

1 Hygini Gromatici Liber de Munitionibus Castrorum, ed. Domaszewski, p. 29, § 56.