time. They did so, even during the time of the Roman occupation of Britain. The work of Theodosius in 368 was to drive back the Picts and Scots from Lothian and the country immediately north of the Southern Wall, and this he successfully accomplished. Then again, in 383, there was a devastatio of this region by the same tribes, and they were again repelled by Roman aid. Other similar inroads continued after the Roman evacuation of the island, and led finally to the withdrawal of the Britons from this district altogether. They were, in fact, driven westwards from the plains of the Tweed and its tributaries, and sheltered themselves in the higher hills of the district. It occurred to me in these circumstances to suppose that this line of earthwork had probably been constructed by the Picts of the east against incursions from the Britons on the west; it may be both as a boundary-line and a line of protection for the lower valleys of the rivers flowing to the Eastern Sea. This, I still think, is in itself not an improbable hypothesis; and it has the confirmation, so far as it is worth anything, of the fact that the line of the earthwork had the name in last century of the Picts Work, or Picts Work Ditch, all along the border of Selkirkshire to a point on the Borthwick Water where it enters Roxburghshire. But, as I have said, I do not attach great importance to this designation as a proof of the Pictish construction of the work. The Southern Roman Wall was for long persistently called The Picts Work. It appears as such on nearly all the older maps, yet nobody would think of attributing the construction of it to them. Their connection with it was only attack and attempted destruction.

The main objection that has occurred to me to the Pictish theory is, that the line of the defensive work is on the eastern side of the hills. Had it been a work of the Picts, set up against foes on the west, and they had had possession of the hills, unquestionably they would have placed the defensive line either on the summit of the ridges or on the western side, facing their foes. Clearly the people who constructed this earthwork had possession of the hill-line. And as the work runs on the eastern and north-eastern slopes, I am constrained to think that it was carried out by a tribe holding the western area, and this could be only the Britons, as they were being gradually dislodged and driven backwards.

The probability of this conclusion is confirmed by the fact that, as a rule, the forts on the line of the earthwork are on the upper or higher side, towards the skyline of the hills across which it runs. This shows that the holders of the defensive line were people living generally to the west or north-west, on which side their forts were naturally a strength to them and a protection against aggression on the earthwork itself.

It was, I think, the Cymri or Britons of what was afterwards known as Strathclyde,—the Strat Clud Wealas, -who actually constructed this earthwork, and constructed it as a defensive boundary against their Northumbrian assailants, call them Saxon, Angle, or Pict. In fact they were often all three, for the Saxons, the Frisians of the Mare Frenessicum or Firth of Forth, coalesced with the Angles of Bernicia and Deira in their mood of aggression and conquest, and the latter frequently combined with the Picts. The diggers of

this fosse, the heapers of the mounds, the builders of the forts at the commencement, along the line, and at its termination, were one and the same people, and these were the hardly pressed and back-driven Cymri of Strathclyde.

Its

A turf wall, a vallum, was clearly regarded by the Britons as a means of defence, and used by them as such. It was, in fact, all they could construct. insufficiency does not disprove its purpose. But it was not, perhaps, so futile a defence as we may now inconsiderately suppose. For the work was in all probability from the first much more than a ditch and mounds. The Catrail was clearly, like other lines of the same sort, a palisaded line of defence. The statements of Cæsar in dealing with the Celts of the Continent point expressly to palisading the mounds of their camps as a characteristic feature of those works. And a very formidable means of defence such a device often proved.1 But we have further the express testimony of Bede to this fact of palisading. "Severus," he tells us, "thought

1 The very term vallum, as from vallus, a stake or pale, actually means an earthwork or rampart set with palisades,-a palisaded rampart or entrenchment. (See numerous authorities given in White and Riddle.) Murus is literally a protecting wall, but it too was originally of turf or turf and stone. Varro tells us : Aggeres qui faciunt sine fossâ eos quidam vocant muros.” — (R. R., i. xiv. 3.) This is how an oppidum of the Aduatici was fortified: "Quod cum ex omnibus in circuitu partibus altissimas rupes despectusque haberet, una ex parte leniter acclivis aditus in latitudinem non amplius ducentorum pedum relinquebatur; quem locum duplici altissimo muro munierant: tum magni ponderis saxa et præacutas trabes in muro conlocabant."-(De Bell. Gall., ii. 29.) Then "postea vallo pedum xii. in circuitu xv. milium crebrisque castellis circummuniti oppido sese continebant."-(Ibid., ii. 30.) The "præacutas trabes "—that is, the stakes pointed at one end against the foe-form an essential and characteristic feature of the defence.

fit to divide the island, not with a wall (murus), but with a rampart (vallum); for a wall is made of stones, but a rampart, with which camps are fortified to repel the assaults of enemies, is made of sods cut out of the earth, and raised above the ground all round like a wall, having in front of it the ditch whence the sods were taken, and strong stakes of wood fixed upon its top. Thus Severus drew a great ditch and strong rampart, fortified with several towers, from sea to sea.'

Whether the statement about the work of Severus as a separate thing from the Roman Wall is correct or not, this at least is clear, that the common form of defence of the time was a ditch with palisading. Bede's description of the Vallum is exactly applicable to the Catrail. This latter work is only the Roman Vallum on a weaker scale-an imitation, in fact, at a later period of what the Romans had shown at its best. As the Britons were taught to imitate the weapons of defence of the Romans, when Constantine left the island, so were they asked to follow the lessons of mural defence. Both of them they did very imperfectly. Hence perhaps, partly, the ready breaks in the work of the Catrail.

2

The Catrail, no doubt, generally runs on the slope of the hill considerably down from the summit line. This is very much the line of the Devil's Dyke, as described by Mr Vere Irving, who speaks of it as running "midbrae," and not rising to the top of a spur of hill, across which it runs. It has been argued from this that such a line could not have been intended for a military defence; for a divergence of a few yards would have 1 Historia, i. v. (Giles' edition). 2 Gildas, Historia, xiv.

enabled the defenders of the rampart to command the approaches from the valley, instead of thus leaving themselves open to surprise from the other side.

This argument seems to me to be of but slight force. It assumes, in the first place, that the vigilance of the defenders is located wholly in the rampart or dyke itself. It forgets, in the second place, the shelter and security from storm which a position "mid-brae" would give as compared with one on the exposed summit of the hill. Besides, it is found that the rising ground along the Catrail is universally to the west, or westwards of the rampart—that is, within the territory to be defended. The ditch is at the foot, so to speak, of the defending host, and in front of the invader; and the place sloping upwards to the summit of the hill is exactly what would afford standing-room for a band of men ready to fight for their territory and stop advance across their frontier line.

An objection to a line such as the Catrail or the Devil's Dyke being intended for a defensive military work has been taken on the ground of its extent. The Catrail is close on fifty and the Dyke close on sixty miles in length, and it is argued that the amount of force needed for guarding it would be so great as to preclude the idea of military defence. But let us suppose that the people lived on the line, as they usually did high up on the hills-suppose it was an enceinte round their course of frontier forts-then we can understand how they might be able readily to defend the whole line.

The general relations of Angle and Cymri in this