^{page 93 note 1} A preliminary note on this is in Antiq. Journ. xi, p. 66.

^{page 94 note 1} Lloyd, History of Wales, ii, 429. Cydweli, the Welsh form, is in this paper retained for the commote, the English spelling, Kidwelly, being used when only the castle or the Norman settlement is meant.

^{page 94 note 2} Monasticon Anglicanum, iv, 65: ‘Haec donatio facta … fuit in domo castelli de Kadweli.’ Among the witnesses is ‘Edmundus qui tune castellum de Cadweli custodiebat’.

^{page 94 note 3} Giraldus Cambrensis, Itinerarium Cambrense, vi, 79 (Rolls Series): ‘Kedwelly … a Mauricio Londoniensi, loci illius tune domino et viro egregio Gaufrido praesulis Constabulario.’

^{page 94 note 4} Bruty Tywysogion, 236 (Rolls Series).

^{page 94 note 5} Annales Cambriae, 62 (Rolls Series), 1101: ‘Maredut filius Resi … a Francis de Kedweli. occisus est.’

^{page 95 note 1} Bruty Tywysogion, 284.

^{page 95 note 2} Royal Letters of Henry III, 1, 176 (Rolls Series); cf. Lloyd, op. at, 658, note 17.

^{page 95 note 3} Lloyd, op. cit., ii, 712, note 110.

^{page 95 note 4} Brut y Tywysogion, 318.

^{page 95 note 5} Cal. Pat. Rolls, 1232–47, 289.

^{page 95 note 6} Excerpta e Rotulis Finium, Henry III, 1, 410 (Edition 1835).

^{page 95 note 7} Annales Cambriae, 92 and 96 (Rolls Series): ‘Postea apud Kedweli Walenses castrametati sunt, domos et villas praeter castrum Kedweli combusserunt.’

^{page 95 note 8} I.P.M., 42, Henry III, no. 417. Cal. Pat. Rolls, 1258–66, 32.

^{page 95 note 9} I.P.M., 2, Edward I, no. 51.

^{page 95 note 10} Cal. Pat. Rolls, 1266–72, 440.

^{page 95 note 11} I.P.M., 7, Edward I, no. 310.

^{page 95 note 12} Ibid., II, Edward I, no. 477.

^{page 95 note 13} Cal. Pat. Rolls, 1281–92, 464.

^{page 96 note 1} Demolished in 1932.

^{page 96 note 2} Grants of murage are recorded in 1280–81 (Cal. Pat. Rolls 1272–81, 418 and 427). Leland in the sixteenth century speaks of Kidwelly as a walled town (Itinerary in Wales, p. 59. Edition, Toulmin Smith).

^{page 98 note 1} The boulders were almost certainly of local origin. Similar material is still obtained at Mynydd y Grug only two miles away from the castle.

^{page 100 note 1} See p. 106.

^{page 101 note 1} Cal. Pat. Rolls, 1401–5, 319 and 502.

^{page 102 note 1} Leland, Itinerary in Wales, p. 59. Edition, Toulmin Smith.

^{page 102 note 1} The reason the tail of this bank is consistently of shale when the subsoil of the castle knoll is clay, is that at a certain depth hereabouts the shale is reached (as can be seen in the re-excavated moat). These lower deposits were doubtless cast inwards over the top of the bank by the builders (following the ordinary technique of mound-building) and so reached the position in which we find them.

^{page 106 note 1} A similar fade-out can be seen by the northern gate, where the line of the bank has survived as a slight difference in the level of the surface.

^{page 106 note 2} Hedingham and Norwich, both of c. 1140, are built on mottes, but these are very largely natural mounds, and the foundations probably go down into undisturbed soil. Guildford, c. 1170, is only partly on the motte. The more usual type built on mottes is the shell keep of late twelfth-or early thirteenth-century date, e. g. Cardiff.

For curtain walls on top of a rampart the earliest we can date accurately is Exeter. Here the bank is dated 1068, the curtain 1206/10. Cf. Hamilton Thompson, Medieval Military Architecture, 128.

^{page 108 note 1} Archaeologia, xlvii, plate xx: a date in the reign of Stephen for this fortress is almost certain.

^{page 108 note 2} Somerset Arch. Soc, xlix, 1903, p. 46.

^{page 108 note 3} Antiq. Journ., xi, 1931, p. 257.

^{page 109 note 1} Third Report on the Excavation of the Roman Fort at Richborough, 1932, plate xlii, no. 362.

^{page 109 note 2} As yet unpublished. Saxon date is indicated by body form, by impressed ornament, and by a technique peculiarly Saxon—the use of the trimming-knife on the wet clay.

^{page 109 note 3} e. g. Paterswolde in Holwerda, Nederlands Vroegste Geschiedenis, 273, fig. 90. F. G. Walker found a cooking pot with sagging base in situ on its hearth at Barton, Cambs., surrounded and overlaid by deposits which contained datable relics. He judged it to be of the Roman period, but this is not correct, as the latest datable associated objects were of the Anglo-Saxon period. The only difficulty one is faced with in accepting this example as evidence of the development of the type in Saxon times lies in the completeness of the evolution; it is a typical ‘medieval’ cooking pot (Proc. Camb. Antiq. Soc, 1908, vol. xii, 304, plate xxii).

^{page 114 note 1} The body form of this jug is, however, not that commonly met with in imitations of ‘polychrome’ jugs.

^{page 115 note 1} It may be noted that the finer wares of class b occur at Dyserth Castle in Flintshire (destroyed a.d. 1263), but they are very rare—2 sherds out of about 200 preserved.

^{page 116 note 1} e. g. Cheam, Surrey Arch. Coll., xxxv.

^{page 116 note 2} Rackham and Read, English Pottery, plate III.

^{page 116 note 3} Argnani, Ceramiche arcaiche faentine, plate III.

^{page 116 note 4} Bode, Die Anfänge der Majolikakunst in Toskana, 9.

^{page 118 note 1} e. g. Graves at Andernach: Bonn. Jahrb., vol. 105, pp. 103 ff. and plate xiv.

^{page 118 note 2} See p. 109 of this paper.

^{page 118 note 3} Antiq. Journ., xi, 259, fig. 8.

^{page 118 note 4} ‘Anciens Vases à Bee’, Bulletin Monumental, Ixiv, pp. 3–62. We owe this reference to Mr. G. C. Dunning.

^{page 118 note 5} A. Conil (Revue des Musées, 1926, p. 209, fig. 4, 31 and 36) also shows an example from a Merovingian grave in Dordogne.

^{page 118 note 6} A possible connexion between Central Italy and Aquitaine at this period is suggested by the migrations of the Papal Court. Boniface VIII's seizure at Anagni and his death in 1303 were followed by the short pontificate of Benedict XI, after which the struggle of the factions caused a long vacancy. This was ended in 1305 by the election of Bertrand de Got, son of a Gascon nobleman and Archbishop of Bordeaux. After his coronation at Lyons the new Pope, Clement V, resided for about two years in Bordeaux and western France. It is known that the craftsmen whom Boniface VIII had attracted to Rome were scattered after his death. It is, therefore, possible that this polychrome technique may have been brought to Aquitaine by a displaced potter, following some patron attached to the Papal Court.

^{page 119 note 1} In wood technique, as with a spokeshave.

^{page 119 note 2} Rich brown, and yellow, as well as the green characteristic of Kidwelly.

^{page 120 note 1} Now in National Museum of Wales. See Arch. Camb., 1915, 47.

^{page 120 note 2} Hawkes, Myres, and Stevens, St. Catherine's Hill, Winchester, 1930, p. 236.

^{page 124 note 1} Identified by Mr. G. C. Dunning.

^{page 125 note 1} Information from Mr. Bernard Reckham.

^{page 125 note 2} Information from Mr. John Charlton.

^{page 127 note 1} British Museum Catalogue of English Pottery, p. 72, B 12, fig. 63; Arch. Journ., lix, 10, fig. 17,

^{page 130 note 1} Trans. Essex Arch. Soc, N.S. xii, 181, fig. 7, no. 13.

^{page 130 note 2} Guildhall Museum Catalogue, p. 177, no. 3, pl. LXVI, 11.

^{page 130 note 3} The same arrangement occurs on two puzzle-jugs found on the sites of the Angel Inn and the Town Hall, Oxford (B.M. Catalogue of English Pottery, p. 57, B 6. Burlington Fine Arts Club, Illustrated Catalogue of Early English Earthenware, 1914, p. 6, pl. III, 22). On these jugs a tubular handle communicates with a false bottom which is emptied by a spout in the form of a stag's head. The upper part of the vessel is emptied by a separate spout on the other side, in the form of a human bust. The jugs are green-glazed, and probably date from the early fourteenth century. In form and the use of human and animal figures, these jugs appear to be local copies of the Exeter jug or a very similar prototype.

^{page 131 note 1} List in Archaeological Journal, lix. 6. The pottery aquamaniles were clay models of the fine metal ewers in use at the same period, see Antiquaries Journal, xii, 446 ff.

^{page 132 note 1} A similar violin is being played by a figure on horse-back in pottery, found in the fort of Sainte-Marie-les-Anvers, in the Cinquantenaire Museum at Brussels, Bulletin Monumental, lxxiii, 63, fig. 27.

^{page 132 note 2} On the restored vessel, two fiddlers have been added on this side, modelled on the one remaining figure. There is an indication of a second figure on the other side, but it is not certain that the four figures were identical, or all playing the same instrument.

^{page 132 note 3} F. Jännicke, Grundriss der Keramik, 1879, i. 226. See also M. L. Solon, The Ancient Art Stoneware of the Low Countries and Germany, i. 67.

^{page 133 note 1} Sussex Arch. Coll., lxxiv, 51, pl. xiv.

^{page 133 note 2} Ibid., pl. xv, 1–2.

^{page 133 note 3} Mémoires de la Societé des Antiquaires de Normandie, xxiv, 6.

^{page 134 note 1} Abbé Cochet in Archaeologia, xxxvii, 399 ff. and xxxix, 117 ff.; Cochet in Bulletin Monumental, xxii, 329–63, 429–46; Dervieu in Bulletin Monumental, lxxiii, 44–74.

^{page 134 note 2} Compare the proportions in the following list of domestic stock belonging to early fourteenth-century South Wales noblemen—4 plough oxen, 4 bulls, 8 oxen, 264 cows, 153 steers and heifers, 70 yearlings, 244 sheep, 77 lambs, 114 swine, 15 mares, 6 foals, 188 goats (Cardiff Records, iv, p. 61).

^{page 135 note 1} See Dr. Jackson's report in Arch. Camb., 1915, pp. 77–82.

^{page 135 note 2} Another from Segontium is 223 mm. long, with an index of 69.

^{page 136 note 1} Cf. Cossar Ewart's report on the Newstead Equidae.