^{page 199 note a} Insula IV.

^{page 202 note a} Like those used in the corridor of House No. 2, Insula I.

^{page 204 note a} Vol. L., pp. 268–271.

^{page 204 note b} The dimensions of the chambers are as follows: 1, 15 feet 1 inch by 17 feet 3 inches; 2, 20 feet 6 inches by 15 feet 8 inches; 3, 19 feet 6 inches by 15 feet 6 inches; 4, the same by 9 feet; 5, the same by 17 feet 8 inches; 6, the same by 7 feet 9 inches; 7, the same by 6 feet 6 inches; 8, 31 feet 1 inch by 24 feet 1 inch. The width of the corridor was 9 feet 2 inches, and the thickness of its outer wall 1 foot 9 inches. The main walls of the house had an average thickness of 2 feet, though the partition walls of the chambers were less.

^{page 206 note a} These had all been removed.

^{page 207 note a} These measurements may be compared with those of the well-known circular temple at Tivoli, the cella of which has an internal diameter of 23 feet 11 inches with a width of 5 feet 8 inches for its peristyle. The internal diameter of the cella of the equally well known circular temple in the Forum Boarium at Rome is 28 feet, the width of the peristyle being 7 feet 5 inches (Taylor and Cresy, Architectural Antiquities of Borne, 1874). This latter edifice there is reason to believe was dedicated to Hercules (See Professor Middleton's Ancient Borne, p. 378, 1st Edit.), and it would be a curious coincidence if the one at Silchester had the same dedication. Round temples occur in Gaul, notably one at Périgueux (M. de Taillefer, Antiquités de Vesone, torn. i. p. 328 et seq.), and other examples will be found at Jublains (Mayenne) (De Caumont, Rudiment d'Archéoloqie ère Gallo-Romaine, p. 237), one south of Jublains (Bulletin Monumental, 9, 1843, pp. 633–4), and in the forest of Beaumont-le-Roger in Normandy (Meinoires de la Société des Antiquaires de la Normandie, 1831, p. 168 et seq.), and others elsewhere. Circular buildings of the Roman period are rare in Britain, and the examples found come under the head of tombs rather than temples. A circular tomb occurs in a Romano-British cemetery at Lockham Farm near Maidstone (Archaeologia Cantiana (1883), xv. 81). At Keston in Kent are the foundations of a circular building having an internal diameter of 23 feet 6 inches. As it stands in a cemetery in which sarcophagi have been found, and its doorway is of very small proportions, it was probably a tomb-house built to contain such sarcophagi and also sepulchral vases (see Archaeologia, xxii. 336 et seq.) There is also a record in Cartwright's Rape of Arundel, 257, of the foundations of a circular building of considerable dimensions on a farm between Wiggenhall and Pulborough, Sussex, but the Roman character of these remains has not been definitely ascertained.

^{page 209 note a} Corpus Inscriptionum Latinarum, vii. No. 6.

^{page 209 note b} Freeman, Edward A., English Towns and Districts (London, 1883), 163Google Scholar.

^{page 219 note a} The measurements of the different rooms of this house are as follows: From 1 to 5 inclusive are the chambers found in the western corridor. 1, 12 feet 9 inches by 6 feet 8 inches; 2, 10 feet 6 inches by the same; 3, 8 feet 2 inches by the same; 4, 15 feet 8 inches by the same; 5, 11 feet 6 inches by the same; 6, 13 feet 5 inches by 14 feet 10 inches; 7, 5 feet by the same; 8, 14 feet 6 inches by the same; 9, 15 feet 8 inches by the same; 10, 9 feet by 23 feet 6 inches; 11, 9 feet 5 inches by 18 feet 8 inches; 12, there are faint traces of a dividing wall here, which might show the existence of a chamber 9 feet 7 inches by 7 feet 3 inches.

^{page 221 note a} The following are the measurements of the various chambers of House No. 4: 1, 14 feet 7 inches by 16 feet; 2, 4 feet 11 inches by 16 feet; 3, 9 feet 7 inches by 16 feet; 4, 13 feet 6 inches by 16 feet; 5, 6 feet by 16 feet; 6, 17 feet by 16 feet. (This is the ruined part of the older house.) 7, the winter room, 8 feet by 10 feet; 8, 12 feet 4 inches by 10 feet 3 inches; 9, 22 feet by 22 feet 3 inches; 10, 10 feet 1 inch by 11 feet; 11, 10 feet 3 inches by 8 feet 2 inches; 12, 10 feet 6 inches by 11 feet 9 inches; 13, 22 feet by 9 feet 6 inches. The vestibule attached to the western side of these chambers (from 9 to 13) was 34 feet long by 7 feet 6 inches wide. The north and east walls of the addition to the older house had an average thickness of 2 feet 3 inches, the walls of 7 and 8 were 1 foot 9 inches, and the rest, from 1 foot 5 inches to 1 foot 6 inches. The wall between 11 and 12 appears to have a thickness of 2 feet 9 inches, but in reality was only 9 inches, the other 2 feet showing the width of a dresser of masonry in chamber 11.

^{page 221 note b} We are indebted to Professor Roberts-Austen, C.B., F.E..S., for the following note on a sample of the lead found during the excavations and on this residue:

“One of the specimens was found to be metallic lead containing nearly five ounces of silver to the ton, and is probably lead obtained by the direct smelting of lead ores, and the metallic lead has probably not been desilverised.

The second specimen is of considerable interest. It is evidently a product of smelting lead and copper ores, mixed with a considerable quantity of sulphide of lead with oxide both of lead and copper. The metallic portion contains 67·64 per cent, of copper and 12·42 per cent, of lead. There is also silver, the amount of which is no less than 45·87 ounces to the ton of the material.

The mass was probably obtained by smelting complex argentiferous lead and copper ore, but it may have been the result of melting together argentiferous lead and argentiferous copper with a view to desilverize the copper. Cakes or discs of such an alloy would be placed on a sloping hearth and heated to a temperature well above the melting point of lead with a view to ‘liquate ’ or sweat out the lead from the copper, which would be left behind as a sponge. The lead would carry away the silver it originally contained, as well as any silver present in the copper.

Elaborate accounts of this process were published in the sixteenth century, but it is interesting to find indications of its use in late Roman times, and the specimens certainly point to the possession of considerable metallurgical knowledge by those who conducted the smelting operations which yielded these products.”

^{page 222 note a} Archaeologia, LII. 752.

^{page 224 note a} de la Croix, Le Père Camille S. J., Mémoire Archéologique sur les découvertes d'Herbord dites de Sanxay, 1883Google Scholar.

^{page 225 note a} Rev.Bathurst, W. H., M.A., Roman Antiquities at Lydney Park, Gloucestershire. London, 1879Google Scholar.

^{page 227 note a} Archaeologia, 1. 277.

^{page 227 note b} ibid. plate XVII.

^{page 228 note a} These sides are now broken down to about 1 foot from the bottom.

^{page 235 note a} I wish, to modify what I said in a former letter (Proceedings S.A.L., 2nd S. xiv. 183) about Vepogeni, as I have since come back to the view that it is Celtic. What has happened to it is that the Picts, adopting the name, sooner or later came to treat the en of Vepogen as their own genitive termination, so that they next inferred Vepog, the Vipoig of the list of Pictish kings. The association of this Caledo with the Campsie Fells is countenanced by Skene, so locating a Caledonian wood as to take the forests of Selkirk and Ettrick in. Lastly it may be worth while considering whether most of the ancient references to Caledonii, Caledonii Britanni, and Silva Caledonia prior to Agricola's campaigns were not made to the same comparatively southern region. See Skene's Celtic Scotland, i. 40–2.

^{page 237 note a} Archaeologia, xl. 416.