The city wall seems to have been uniformly built throughout its circuit of small stones, 6 or 7 inches square on the face, bonded about every sixth course with two or three courses of large flat tiles nearly 18 inches by 12 inches, and 1-1/2 inches thick. The core was rough rubble; it was about 8 to 10 feet thick and probably 20 to 25 feet high. FitzStephen (_c._ 1180) describes it as "the high and great wall of the city having seven double gates and towered to the north at intervals; it was walled and towered in like manner on the south, but the Thames has thrown down those walls." There is evidence for a square Roman wall-tower having existed in Houndsditch, and for others, semicircular in form. It would always have had, as we know it had at a later time, a walk all round, a parapet, and battlements. A part of the late wall which still shows the walk and battlements is yet in London Wall. The turrets (of the later wall at least) were higher than the wall. [Illustration: FIG. 20.--FROM THE COMMON SEAL. REVERSE, ENLARGED, 1224.] According to Stow, the ditch of the city wall was begun in 1211, and the same writer, speaking of the Walbrook entering the city, as mentioned in the Conqueror's charter, adds "before there was any ditch." This is a mistake, for notices of Houndsditch appear before 1211, and the name is used in the _Liber Trinitatis_ in a way that infers its existence before 1125. A few years ago an excavation at Aldersgate exposed a complete section of the ditch outside the wall. It was 14 feet deep, 35 feet wide at bottom, and 75 feet wide at the top of the sloping sides. The top of the inner slope was 10 feet from the wall. This is drawn and described in vol. lii. of _Archæologia_, and a comparison subsequently made with the ditch at Silchester showed that, like it, it was certainly of Roman work. In each there was found a raised foundation in the bed of the ditch for a trestle bridge crossing from the gate (Fig. 21). After the ruins of the fire (of five or six years ago) at Cripplegate were cleared away, it was evident that the basements of the houses in the street running north and south outside the west end of St. Giles's churchyard, by the angle bastion of the wall which still stands there, were built in the old ditch. A length of embanked stream which fed the ditch ran by the east of Finsbury Circus.[75] It is shown in the so-called Aggas plan. [Illustration: FIG. 21.--SECTION OF ROMAN WALL AND DITCH.]