¹ Session 1915-16 (henceforward cited as T.C.S.), pp. 28-69.

² ibid, pp. 1-28.

³ In short, this is : S. and W. section 1285-91 ; S. and E. section 1295-1301 ; N. section 1315-22.

⁴ e.g. Caernarvon Castle Official Guide, which has passed through two editions and many printings, and Sidney Toy’s Castles (1939), p. 195.

⁵ See the present writer’s paper in History, Oct. 1950, ‘ The Birth of Edward of Caernarvon and the Beginnings of Caernarvon Castle ‘ (Historical Revision, No. CXVI).

⁶ The point is discussed on pp. 45-6 of Dr J. G. Edwards’s Rhys Memorial Lecture, ‘ Edward I’s Castle-building in Wales ‘ (Proc. Brit. Acad., XXXII, pp. 15-81), which appeared after this paper was written. Dr Edwards shows that, if due allowance is made for certain missing rolls of accounts, there is room for supposing that the stonework of the castle may well have proceeded much more rapidly than has hitherto been realized. That this was indeed the case is suggested by the fact that already in the third building season (1285), when Peers thought the main castle construction had scarcely begun, lead was being sent by sea from Bristol for roofing the castle buildings : in the Liberate Roll of 14 Edward I there is note of the payment of £31 7s 8d to Peter de la Mare, constable of Bristol castle, for 14 loads of lead bought and sent by him usque castrum nostrum de Kaerenarvon ad domus nostras eìusdem castri inde cooperiendas anno regni nostri terciodecimo (P.R.O., Liberate Roll, no. 62).

⁷ P.R.O., Exchequer Accounts, 5/18, no. 11. The whole document has now been printed by J. G. Edwards, loc. cit., pp. 79-80 ; its implications are examined ibid, pp. 46-7.

⁸ T.C.S., p. 12 : ‘ The new work was evidently started in 1295, and by the end of the year had progressed as already described ‘.

⁹ ibid, p. 10.

¹⁰ ibid, p. 12, note 1.

¹¹ Works expenses at Beaumaris were £8555 7s 10d for one year (18 April 1295—7 May 1296) (Pipe Roll, no. 158), as compared with £4393 13s 7½ at Caernarvon for four years (11 Sept. 1295—28 Sept. 1299) (ibid, no. 146, m. 24).

¹² J. E. Morris, Welsh Wars of Edward the First, pp. 242, 253.

¹³ Pipe Roll, no. 136.

¹⁴ T.C.S., pp. 11, 12.

¹⁵ I have to thank Mr W. J. Hemp for calling my attention to this photograph and allowing me to reproduce it here. Plates II and IVa are from photographs by Mr G. B. Mason of the National Buildings Record, to whom thanks are due for permitting their reproduction. Plates I and III are from photographs by the author. The plan (p. 31) has been redrawn from a survey in the possession of the Ministry of Works.

¹⁶ Pipe Roll, no. 158.

¹⁷ Indeed the word is actually used of the Caernarvon castle ditch in a record of June 1283 : ‘ ... pro fossur’ et apparatu noue mote de Caernaruan . . . ‘ (Exch. Accts., 351/9).

¹⁸ T.C.S., p. 12, note 1.

¹⁹ Exch. Accts., 486/29 : ‘ Henrico de Elreton cementario pro duodecim perticatis Kaye ville de Kaern’ ad tascam faciendis de petra et calce sumptibus suis propriis, percipienti pro perticata continente xxv pedes viij.li, ... in partem solucionis iiij.xxxvj.li. quas percipere debet pro tasca predicta—lx.li. ‘ Also ‘ Ricardo de Wykombe pro quinque perticatis, dimidio rod’, tribus pedibus Kaye predicte cum arrena inter murum ville et dictam kayam obstupandis, qualibet perticata continente xxv pedes . . . xl.s. ‘

²⁰ Possibly a long perch was allowed because a water wall presented additional difficulties. Dr. J. G. Edwards draws my attention to the fact that calculations based on payments made in July 1279 ^for building the side of the castle ditch at Flint (Chancery Miscellanea, 2/2/17) point to a perch of 25 feet. This was probably a wet ditch and similar considerations may have applied.

²¹ It is not practicable to base any arguments on the heights to which the wall is described as standing in 1296, as the depth of the foundation from which these were presumably taken is unknown.

²² T.C.S., pp. 13, 59.

²³ The revetting suggests the possibility that the original intention may have been not to remove the motte, but rather to put a heavy building on it. A tower in this position would have been the crown of the whole design.

²⁴ The three lower storeys of the Eagle Tower, which are earlier than the fourth storey and surmounting turrets (T.C.S., p. 46), may represent the general height reached by the south curtains and buildings ; the fact that the hall had been built to gable height before the time of Madoc’s revolt (cf. ibid, p. 36) shows there must have been at least equivalent height in the section of the curtain against which it was built. The uppermost part of the curtain between the Eagle and the Queen’s towers, the beginning of which is marked by a string course, was probably added when the Eagle Tower was heightened, i.e. after 1295.

²⁵ There is in any case no particular virtue in this date. Work was certainly going on in 1304-5 (see below, p. 33). I have shown elsewhere (Trans. Caernarvonshire Hist. Soc., 1948, pp. 16 ff.) that it was also in progress in 1309, and that it was in that year, not 1315, that Walter of Hereford died and his deputy Henry of Ellerton succeeded him as master mason at Caernarvon.

²⁶ This reading of the building is at any rate partially supported by documentary evidence for the completion of the heightening of the Eagle Tower and another tower in about 1316, for the erection of the statue over the King’s Gate in 1320, and for work on the hall over that gate in 1321-2 (T.C.S., pp. 16-18).

²⁷ Pipe Rolls, nos. 131, m. 26d, and 136, m. 28.

²⁸ The dates are those of accounting periods within which payments were made. There is evidence that at Conway and Harlech works were in progress earlier than the commencing date of the account, while at Caernarvon the figure here given actually only covers the work from August 1284 onwards although it had then been in progress for more than a year.

²⁹ Cf. the present writer’s note, ‘ Harlech Castle : the dating of the outer enclosure ‘, in Journal of the Merioneth Hist, and Record Soc., 1951, pp. 202-3.

³⁰ Pipe Roll, no. 146.

³¹ The error arose from a wrong dating in the printed P.R.O. List of Exchequer Accounts, Various (1912), where the roll (486/1) was given as 16-17 Edward I ; in the Search Room copies this has been corrected to 32-33 Edward I. Cf. Edwards, loc. cit., p. 49, note 2.

³² Cf. note 25 above.

³³ Calendar of Chancery Warrants, I, p. 63.

³⁴ See my note on ‘ The Cloister of Vale Royal Abbey ‘ in Journal of the Chester and North Wales Archaeological, Architectural and Historical Society, 1949, pp. 295 ff.

³⁵ ibid, note 4.

³⁶ English Historical Review, LXV, pp. 441-2.

³⁷ Hilda Johnstone, Edward of Caernarvon (Manchester U.P., 1946), p. 7, note 1. The survival of the Caernarvon-born prince and the death of his elder brother barely four months later must be regarded as fortuitous ; but one is inclined to wonder whether tradition may not preserve a genuine memory of a policy, conceived with the passing of the young Alphonso, of making capital out of the fact that the eldest surviving son was born in this ancient heart of Gwynedd.

³⁸ Flores Historiarum (Rolls Series), III, p. 59 : (s.a. 1283) Apud Kaernervan, corpus maximi principis, patris imperatoris nobilis Constantini, erat inventum, et rege jubente in ecclesia honorifice collocatum. As the late Dr Charlesworth noted (The Lost Province, p. 28, note 2), Nennius had already recorded the existence at Segontium of an inscribed tomb of Constantine. Cf. also Mabinogion (Everyman edn.), p. 86, and C, E. Stevens in Archaeological Journal, XCVII, p. 134.