The recent excavations conducted by Macridy Bey have thrown light on several important features not previously determined. From the plan which he publishes[589] it would seem that the frontage to the approach, on the left side at any rate (E, F), is really the outer wall of the gate tower and external to the main wall. We are thus confirmed in our conclusion that the entrance was flanked on either side by extra-mural towers, as later well known in Roman military forts and mediæval architecture of Europe. Unfortunately the excavators did not carry on their inquiry to ascertain (as might have been done with little difficulty) the line of frontage of the main wall of the whole building or enclosure. This we suspect would be in line with the nearer monoliths, though from a suggestion upon the plan it may have been a little nearer the interior—a position which from several reasons would not be probable—and, indeed, such a wall must have been much stouter than anything marked upon the plan. From the foundations preserved it would seem that each tower was designed with chambers. Two other strong oblong chambers or vaults may be noted in the thickness of the wall, and there seems to be indication in the plan of a continuation to the series. The excavators were troubled by the fact that no door openings were found to these rooms.[590] As no section is given by them to show the relative levels of the different walls, it is not possible to form an opinion as to whether they are at all preserved above the lower courses. It is in any case interesting to observe that in military Roman forts of later days (in the earlier system of the first and second centuries) the chambers on the lower floors of such guard-rooms were often merely vaults or cellars, gained from the upper floor (which was nearly level with the sentry-go upon the wall) by means of an internal ladder.