It will have been noticed that the diameters of the circles have evidently been planned according to a geometrical scale, of which the unit seems to have been equivalent to ten feet of our measure. A reference, for example, to the plan of Arbor Low will again demonstrate this point. The average diameter of the circle of the stones is one hundred and fifty feet, the width of the fosse is twenty feet, and that of the vallum on the ground level is thirty feet, and its height above the _excavated_ fosse is ten feet; the total diameter of the monument is two hundred and fifty feet, and Gib Hill, its pointer, stands one thousand feet away south-west by west. But the stones at Arbor Low, and, indeed, those of all the other examples, do not form a true circle; there is always an elliptical variation. At Arbor Low this variation is about ten feet; at the Wet-Withens it is only three or four feet. At the former there are in the centre three fallen stones, which in all probability formed a dolmen, of which the capstone measures fourteen feet in length; it may be assumed, therefore, that its supporters occupied a space of about ten feet. At the Wet-Withens we read that there was originally a single large stone in the centre, which we may assume was not more than three or four feet in diameter. If, therefore, the central cromlech was first erected, and the radius of the circle of stones measured from its outside walls instead of from the true centre, we have the probable explanation of the elliptical variation in every case. The variation, in turn, should give us some idea of the central cromlech when, as in so many instances, it has been destroyed. This suggestion is supported by another distinctive feature in the plan of stone circles, of which, also, no explanation has been offered. Nearly every circle has two entrances, or an entrance and exit, cut through the mound, and when a fosse is present it is broken at the causeways; but these entrances, although on opposite sides of the circle, and usually towards the north and south, are never directly opposite each other. If, therefore, the central cromlech was the dominant purpose, the roadway would pass alongside it, and not have to deviate around it, as it certainly would if it truly bisected the circle.