circles were more irregularly, but generally more closely, set. All these uprights (except a few on the north side of the inner circle) were linked together by horizontal beams having square-cut holes, through which the upper ends of the piles passed. The horizontal beams were arranged in two ways. Some lay along the circumference and bound together all the uprights in the same circle to each other, while others took the radial position and connected each circle together. Some of the latter were long enough to embrace three circles, and when this was the case I have noticed that the upright in the middle circle was sometimes firmly caught in a deep cut in the transverse, instead of passing through a mortised hole (see Fig. 190). Although the uprights in the inner circle were not linked together circumferentially along the whole course of the horizontal beams, the particular construction of the log pavement on the north side rendering this unnecessary, every one of them had a radial beam, directed from within outwards, which kept it from yielding to lateral pressure. This purpose was equally well served in several ways; sometimes the inner end of the radial beams pressed tightly against the upright, at other times the former projected half-way into the log pavement, where its end was firmly fixed by a thick pin passing through it into the under structures of the island, and its middle contained either a notch or mortised hole for holding the latter in position. The external ends of these radial beams were occasionally observed to be continuous with additional strengthening materials, such as wooden props and large stones. The main object of the whole of this elaborate structural system was to give stability to the island, afford fixed points on its surface, and prevent the superincumbent pressure of whatever buildings may have been erected over it from causing the general mass to bulge outwards--objects which appeared to have been most effectually attained. The piles in the outer circle were merely round posts, smaller and more closely placed than those in the inner circles, being sometimes only a few inches apart, and appeared to have been bound together by a transverse rail, into which their tops were inserted after the manner of a hurdle. Portions of these rails, pierced with holes, were found at the south-east side, though none actually in position; so that the inference that the outer stockade was intended as a fence or bulwark seems quite legitimate. In support of this view I may state that nowhere along its course were the piles connected together by horizontal beams, either circumferentially or radially, nor did they penetrate deeply, so that for giving stability to the island the outer circle would be of little use.