square fort, now almost wholly destroyed, in area about 7½ acres, with rounded corners, walls of rubble concrete, faced and bonded with local sandstone, and, as it seems, external bastions at one gate. It has yielded a few early coins, and much that is datable to the fourth century. There can be little doubt that it was a fort in the fourth century, and its name identifies it with Branodunum. Some of its structural details, in particular its shape, have suggested that the site was first occupied and the ramparts built at an earlier time. But we know too little of the remains to be able to speak with confidence on this point.⁸

   (2) Burgh Castle, near Yarmouth, situated close to the junction of the rivers Yare and Waveney and the landlocked expanse of Breydon Water. In shape and size it is an irregular oblong of some five acres, planted on a cliff 40 ft. above the river, but extending down the slope to marsh level, that is, to the water’s edge. Except on this side its walls still stand high and strong. They are constructed of concrete rubble with flint facing and bonding tiles, strengthened with external circular towers. The remains found here belong almost wholly to the fourth century, and thus connect it with the Saxon Shore, while the river name Yare and other evidence justify its identification with Gariannonum.⁹

   (3) Felixstowe. The Roman site here, Walton Castle, has long been swept away by the sea. But the accounts and drawings of it which survive show that it was defended by a wall of concrete rubble with bonding tiles, strengthened apparently by external bastions, while abundant remains, found especially in a neighbouring cemetery, indicate an occupation during the fourth century and perhaps earlier. We cannot positively prove that it ever was a fort. But all we know of it agrees with such a theory. As to its Roman name we have no evidence.¹⁰

   (4) Bradwell, on the Essex coast, in a sheltered position near the mouths of the Blackwater and the Colne and opposite to Mersea Island. Here was a rectangular fort, girt with a wall of the usual type—concrete rubble core, bonding tiles, round external towers. Much of it has perished by the encroachments of the sea, but its one perfect side is 500 ft. long, and it probably covered about five acres, like Burgh Castle. The coins found here date from A.D. 260—400; those of the fourth century are commonest. The name of the place in early English days, Ithanceaster, identifies it almost certainly with Othona.¹¹

   (5) Reculver, on the north coast of Kent, the Roman Regulbium: p. 19.

  ⁷  Short accounts of the Saxon Shore are given by Lewin, Arch. xli, 421—52, C. R. Smith, Coll. Ant. vii, i 52—69, and others, and individual forts have been described by various writers named below. In general, they have tended to neglect the historical side of the subject. An excellent summary of the facts, as known in 1914, is given by Haverfield in Pauly-Wissowa’s Real-Encyclopädie, ii, A (s.v. ‘ Saxonicum Litus ‘). For a brief sketch of later discoveries, see G. Macdonald in Funfundzwanzig Jahre Romisch-Germanische Kommission (Frankfurt, 1929), 107 ff.
   ⁸ V.C.H. Norfolk, i, 303—5. See further below.
   ⁹ Harrod, Norfolk Archaeology, V, 146; Fox, Arch. Journ. xlvi, 348; lvii, 120; Ives, 
        Gariannonum (ed. 2).
  ¹⁰  Fox, Arch. Journ. lvii,115 and Y.C.H. Suff. i, 287.
  ¹¹ Lewin, Arch. xli, 439; M. V. Taylor, Essex Arch. Soc. Trans. xvii; Roy. Com. Hist. Mons. 
          S.E. Essex, xxxviii and 13.