The Wandsdyke always seems to have been regarded as a barrier erected to stop the incursions of the Welsh into the southern counties, and that part of it extending from Savernake forest westward, for ten or twelve miles, seems at some comparatively recent period to have been raised and strengthened[108] (either by the Belgæ or Saxons) to make it more effectual for that purpose. According as an army is advancing northward from Winchester, or Chichester to the Severn valley, or is marching from Gloucester or Cirencester towards the south, the rampart either protects or bars the way. In its centre, near the head-waters of the Kennet, the Saxons advanced in 557 to the siege of Barbury Castle, and having gained that vantage ground, they again advanced in 577 to Deorham, and fought the battle that gave them possession of Glewanceaster, Cyrenceaster, and Bathanceaster.[109] What they then accomplished they seem to have attempted unsuccessfully thirty-seven years earlier, and to have been stopped in the attempt by Arthur at Badon Hill. If this is so, there can be very little difficulty in determining the site of the Welsh gate as that opening through which the road now passes 2½ miles south of Silbury Hill, in the very centre of the strengthened part of the Wandsdyke. If this is so, the Saxons under Cerdic must have passed through the village of Avebury, supposing it then existed, on their way to Cirencester; and if we assume that they were attacked on Waden hill by Arthur, the whole history of the campaign is clear. If we may rely on a nominal similarity the case may be considered as proved. Waden is the name by which the hill between Avebury and Silbury is called at the present day by the people of the country, and it is so called on the Ordnance survey sheets, and etymologically Waden is more like Badon than Baydon, or Badbury, or any other name in the neighbourhood. The objection to this is that Waden Hill is not fortified, and that Gildas speaks of the "Obsessio Montis Badonici." It is true there is no trace of any earthworks on it now, but in Stukeley's time there were tumuli and earthen rings (apparently sepulchral) on its summit, which are represented in his plates; but no trace of these now remains. The hill was cultivated in his day, and in a century or so beyond his time all traces of ramparts may have been obliterated, supposing them to have existed. The true explanation of the difficulty, however, I believe to be found in Jeffrey of Monmouth's account of these transactions. He is a frail reed to rely upon; but occasionally he seems to have had access to authorities now lost, and their testimony at times throws considerable light on passages of our history otherwise obscure. According to him there was both a siege and a battle; and his account of the battle is so circumstantial and so probable, that it is difficult to believe it to be a pure invention. If it is not, every detail of his description would answer perfectly to an attack on an