382 
these are alternate, and outwardly curved, so that the ground 
plan presents the outline of scales. (See section and plan.) 
In this fashion the work is made to support itself on the slope 
of the hill, and also perhaps to offer standing points for slingers 
or other defenders. Very many sling-stones are found on these 
stages. 
It is worth while to observe that the plan of the work is 
generally curvilinear, and there are no salient angles. Where the 
ground affords a free scope instead of leading the lines, the plan 
approaches a circle, as at the Eastern stronghold. This resembles 
the Cornish fastnesses, which are built of dry granite masonry, 
as I pointed out in a paper read to the Som. Arch. and Nat. 
Hist. Society, in 1852. Such a plan is not to be wondered 
at, as the difficulty of constructing an outward angle in 
masonry of rough stone without mortar would be considerable. 
About 60 feet to the west of the great entrance there is, just 
within the rampart, a quadrangular oblong excavation about 17 
feet in length by 13 in breadth, faced within on the north, east 
and west sides with dry masonry of careful construction. Of the 
use for which this was intended no very valid conjecture has been 
formed. 
The rampart westward to the point of the hill overlooking 
Birnbeck is now a continuous ruined heap of shapeless limestone 
blocks. 
Below it on the southern slope of the hill, two parallel trenches 
ran, with interruptions here and there, from nearly opposite the 
great south entrance to almost the end of the rampart westward. 
To the northward was a sort of private entrance for the 
defenders, covered by flanking works down the steep side of the 
hill, to the perennial dripping spring at the point called Spring 
Cove, opposite Birnbeck, which, after so many ages, has at last 
within a few years past been lost by a landslip. This was the 
nearest supply of water. The way led through a cave or cleft 
of the limestone cliff now stopped with stone or earth. The 
