^{page 215 note a} Archaeologia, lii. 754–757Google Scholar.

^{page 218 note a} The actual dimensions of the rooms in this house are as follows: (1–4) each 17 feet 9 inches in length, by 14 feet 8 inches, 15 feet 9 inches, 14 feet 6 inches, and 16 feet 3 inches in width, respectively; (5) 13 feet by 11 feet; (6) 23 feet by 15 feet 9 inches; (7) 13 feet 3 inches by 12 feet 1½ inch; (8) 49 feet 3 inches long by 7 feet 6 inches wide; (9) 29 feet 6 inches by 27 feet 6 inches; (10) 28 feet 6 inches by 12 feet 3 inches. The remaining walls were 2 feet thick.

^{page 219 note a} Archaeologia, liii. 276, pl. xxiiiGoogle Scholar.

^{page 224 note a} Many of the pieces of wood found in the wells and pits at Silchester are found, after removal, to bear in parts an incrustation or deposit of a bright blue colour. This has been identified by Professor Church, F.R.S., F.S.A., as the mineral vivianite, and he has obligingly communicated the following note on a specimen found in 1894:

“I have examined the blue substance from the remains of the iron-bound wooden bucket found at Silchester. It proves to be, as I suspected, the mineral vivianite, a hydrous ferrous phosphate, which if pure would contain the following percentages:

Its formula is Fe₃ P₂ O₈, 8 H₂O. Under the microscope the earthy look of the compound is replaced by a distinctly crystalline appearance, indeed the substance is seen to be made up of felted crystals, oblique prisms. The occurrence of vivianite has been noted in peat bogs in Shetland and Isle of Man, accompanying sometimes the horns of the elk and deer. Dana says the mineral has been found ‘near an old slaughter-house in Edinburgh.’ ”

An extensive deposit of vivianite seems to have occurred throughout a barrow known as Roylow, in Derbyshire, which was opened in 1894. See Proceedings, 2nd S. xv. 426.

^{page 225 note a} The construction of this pavement was as follows: (1) a bed of coarse yellow mortar, 8 inches thick, the upper surface of this was represented over the channels of the underlying hypocaust by the top of a bridging tile, 1½ inch thick, resting upon other tiles at each side, with mortar joints 1 inch thick; (2) a bed of white concrete, 3 inches thick; (3) red cement, 1½ inch thick; (4) fine tesserce embedded in pink cement, with a united thickness of 1½ inch. The thickness of the floor over the hypocaust channels was 7½ inches. The furnace channel was arched over for most of its length by six overlapping layers of tiles, with a seventh across the top. From the floor of the hypocaust to the under side of the bridging tiles was 3 feet 2 inches.

^{page 226 note a} The construction of this pavement is as follows: (1) 1 inch of pink cement with tesserce set in it; (2) 3½ inches of drab-coloured cement or mortar; (3) 3 inches of white cement containing small pebbles; (4) 5 inches of loose gravel, beneath which was (5) a bed of mortar 1 inch thick.

^{page 226 note b} This pavement was laid as follows: (1) 1 inch of pink cement with tesserce set in it; (2) 2½ inches of white mortar or cement; (3) 4½ inches of hard gravelly concrete; (4) a layer of burnt clay 1½ inch thick; (5) 7½ inches of mould or earth; (6) ½ inch of chalk; and (7) 1½ inch of yellow mortar, very irregular.

^{page 227 note a} The construction of this floor was as follows: (1) red tesserce, 1 inch thick, set in ½ inch of pink mortar; (2) 3 to 4 inches of white cement; (3) 11 inches of gravel and builder's rubbish; (4) a bed of hard yellow mortar 5 inches thick. The whole rested on earth containing layers of clay and gravel.

^{page 227 note b} This pavement, which was very evenly and well laid, was constructed of the following layers: (1) 1 inch of pink cement with tesserce set in it; (2) 4 inches of white cement; (3) 7 inches of gravelly concrete; (4) 8 inches of gravelly yellow mortar. The whole rested on a bed of flint.

^{page 229 note a} The dimensions of the rooms are as follows, the first measure being from east to west, the second from north to south: (1) 11 feet 9 inches by 14 feet; (2) 11 feet 1 inch by 7 feet 6 inches; (3) 12 feet 3 inches by 7 feet 6 inches; (4) 13 feet 2 inches by 5 feet 3 inches; (5) 3 feet by 5 feet 3 inches; (6) 5 feet 10½ inches by 14 feet 6 inches; (8) 7 feet 11 inches by 7 feet 7 inches; (9) 14 feet 8 inches by 7 feet 4 inches; (10) length uncertain, 7 feet 4 inches wide; (11) 7 feet 7½ inches by 17 feet; (12) 19 feet 9 inches by 17 feet; (13) 17 feet 2 inches by 17 feet 6 inches; (14) 17 feet 9 inches by 12 feet 8 inches; (15) 17 feet 9 inches by 15 feet 8 inches; (16) 17 feet 6 inches by 22 feet 4½ inches; (17) 17 feet 6 inches by 20 feet; (17a) and (17b) each 17 feet 6 inches by 7 feet 6 inches; (18) 17 feet 6 inches by 20 feet; (19) 22 feet 6 inches by 18 feet 3 inches; (20) 23 feet by 18 feet; (21) 8 feet by 15 feet 9 inches; (22) 22 feet 3 inches by 23 feet 4 inches (mean); (23) 22 feet 4½ inches by 20 feet; (24) 22 feet 4½ inches by 18 feet 4½ inches; (25) 22 feet 5 inches by 6 feet 3 inches; (26) 22 feet 6 inches by 15 feet 4 inches; (27) 22 feet 6 inches by 20 feet 1½ inch; (28) 9 feet 8 inches by 15 feet 4½ inches; (29) 21 feet 2 inches by 19 feet 6 inches; (30) 13 feet 3 inches by 12 feet. The thickness of the walls varies considerably. In the older part of the house the average is 18 inches, but in the eastern wing the main walls are 2 feet thick.

^{page 233 note a} The following is the description given De Ctesibica Machina by Vitravius: “Insequitur nunc de Ctesibica machina, quæ in altitudinem aquam educit, monstrare. Ea fit ex æere, cujus in radicibus modioli fiunt gemelli paulum distantes, habentes fistulas, furcillaæ sunt figura, similiter coheerentes, in medium catinum concurrentes: in quo catino fiant axes in superioribus naribus fistularum coagmentatione subtili collocati, qui præobturantes foramina narium, non patiuntur exire id, quod spiritu in catinum fuerit expressum. Supra catinum penula, ut infundibiilum inversum, est attemperata, quæ etiam per fibulam cum catino cuneo trajecto, continetur et coagmentatur, ne vis inflationis aquæ earn cogat elevare: insuper fistula, quce tuba dicitur, coagmentata, in altitudine sit erecta. Modioli autem habent infra nares inferiores fistularum axes interpositos supra foramina earum, quee sunt in fundis: ita de supernis in modiolis emboli masculi torno polito et oleo subacti conclusique regulis et veetibus convolvuntur, qui ultro citroque frequenti motu prementes aerem, qui erit ibi cum aqua, axibus obturantibus foramina, cogunt et extrudunt inflando pressionibus per fistularum nares aquam in catinum, e quo recipiens penula spiritu exprimit per fistulam in altitudinem. Et ita ex inferiore loco castello collocato ad saliendum aqua subministratur.” De Architectura, lib. x. cap. xii.

^{page 234 note a} The identity of the Silchester contrivance with the Ctesibica machina of Vitruvius was first pointed out to us by Mr. J. T. Micklethwaite, F.S.A.

^{page 234 note b} A pump of similar construction to that described by Vitruvius, but of bronze, was found some years ago at Bolsena, in Italy, and is now in the British Museum. Another example of the same kind, found in 1795 at Castrum Novum, near Civita Vecchia, is engraved in Smith's, Dictionary of Greek and Roman Antiquities, 3rd edition (London, 1890), i. 570Google Scholar, s.v. Ctesibica Machina. Although neither of these resembles the Silchester specimen, the principle on which they are constructed is identical. A description of the Bolsena pump is appended to this paper.

^{page 237 note a} Such chapels are occasionally found in the houses of Pompeii. A good example of such a one may be seen in the house of Epidus Rufus in that town, where an ala of the atrium is devoted to this purpose. It contains an aedicula for the sacred images to which this at Silchester must have had considerable resemblance.

^{page 238 note a} For such chambers see a trace in House 1, Insula I.; House 4, Chamber 3, Insula VII.; House 1, Chamber 6, Insula VIII.; also a projection from the western corridor of the Hospitium in the same insula, and a room projecting from the centre of the corridor of a small house called by Mr. Joyce, Block I. (Archaeologia, xl. pl. xxiii.*)

^{page 239 note a} Now in the possession of W. G. Mount, Esq., M.P., of Wasing Place, Berks. See Proceedings, 2nd Series, xvi. 74.

^{page 239 note b} Edmond Tudot, Collection de Figurines en Argile de l'Époque gallo-romaine, 1860.

^{page 239 note c} Revue Archéologique, troisième serie, iv. 285. Ed. Flouest, Archéologie Gauloise. Deux stèles de Laraire.

^{page 240 note a} Lysons, Samuel, Reliquiœ Britannico-Romanœ (London, 1817), iii. pl. xx. figs. 2, 3, 4.Google Scholar

^{page 242 note a} See for an example of this arrangement a floor found at Scamton in Lincolnshire in 1795, figured in Fowler's Mosaic Pavements.

^{page 244 note a} Such an arrangement of concentric lines is to be found in the remains of the floors of the corridors of a large Roman villa in the Greetwell fields near Lincoln. Archaeological Journal, xlix. 258Google Scholar.

^{page 245 note a} A little panel of rich work in a large plain field is occasionally found in the mosaic floors of triclinia in Pompeian houses. It was evidently there laid down as a guide for the placing of the table and couches at the dinner hour. An example occurs in the triclinium of the great Fullonica in Pompeii, and other instances are to be found in various houses in that city.

^{page 245 note b} The dimensions of the rooms in the house were as follows, the first measure being from east to west, the second from north to south: (1) 16 feet 2 inches by 22 feet 4 inches; (2) 12 feet 9 inches by 7 feet 9 inches; (3) 7 feet 10 inches by 12 feet 9 inches; (4) 20 feet 4 inches by 8 feet 10 inches (mean); (5) 20 feet 4½ inches by 13 feet 9 inches; (6) 20 feet 10 inches by 18 feet 9 inches; (7) 20 feet 10 inches by 9 feet; (8) 20 feet 10½ inches by 15 feet; (9) 18 feet 11 inches by 10 feet 8 inches; (10) 20 feet 10½ inches by 9 feet 10 inches; (11) 6 feet 11 inches by 8 feet 7 inches; (12) 20 feet 7 inches by 13 feet 1½ inch; (13) 11 feet 8 inches by 8 feet 7 inches; (14) 54 feet 2 inches by 12 feet 3 inches; (15) 6 feet 6 inches by 6 feet 10 inches, and afterwards 13 feet 10 inches by 3 feet 6 inches; (16) 13 feet 10 inches by 10 feet 4½ inches; (17) 14 feet 6 inches by 11 feet 7½ inches; (18) 21 feet 8 inches by 25 feet 3 inches; (19) 21 feet 6 inches by 20 feet 4 inches; (20) 22 feet by 18 feet; (21) 13 feet 10 inches by 8 feet; (22) 6 feet 6 inches by 8 feet; (23) 9 feet 4 inches by 28 feet 7 inches; (24) 9 feet 4 inches by 7 feet 9 inches; (25) 8 feet 3 inches by 8 feet; (26) 8 feet by 8 feet 2 inches; (27) 8 feet square; (28) 15 feet by 26 feet 10 inches. The average thickness of the walls is 2 feet.

^{page 246 note a} The statement sometimes made that foreign materials (more especially Palombino for the white grounds) were employed in British mosaic floors is erroneous. It is possible that an occasional example of the use of foreign marbles may have been noted in London, for instance, but doubtfully anywhere else.

^{page 247 note a} A good example of fretwork mixed with panels of braidwork is to be found in the principal pavement of a villa at Box Moor, Herts. (Account of Excavations on the site of two Roman Villas at Box Moor, Herts, by John Evans, F.S.A., from Archaeologia, xxxv. 56–69.) Another may be seen in a pavement discovered in the Blackfriars at Leicester in 1754, figured in Nichols's History of Leicestershire, vol. i. part i. plate viii. A far finer one occurs in the southern division of the triclinium of the Roman villa at Chedworth, in Gloucestershire, which, so far as we are aware, has never yet been published.

^{page 248 note a} Lysons, Samuel, Beliquiœ Britannico-Romanœ (London, 1817), ii. pl. 1.Google Scholar

^{page 249 note a} See for a good example of this class of decorative painting, Artis, E. T., The Durobrivae of Antoninus (London, 1828), pl. 32.Google Scholar

^{page 250 note a} Insula II. Decumanus major.

^{page 252 note a} Professor Church, F.R.S., F.S.A., has kindly communicated the following noce on this discovery: “My conjecture as to the nature of the orange-red crystalline substance found at Silchester proves to be correct. The material is realgar, arsenic sulphide (in chemical formula As₂ S₂). It is the Σανδρáκη of Theophrastus, the Sandaracha of Pliny. It occurs native (as a mineral) not only in India and China but in many European localities. Amongst the latter the most interesting to you are the Solfataras near Pozzuoli, Tolfa near Civita Vecchia, and Vesuvius. It was certainly employed in ancient paintings, but I do not recall any other use. I feel sure that it has never before been recognised (even if found) on any Roman site in Britain. Realgar has long been prepared artificially, but I am inclined to think the Silchester material is the native mineral.”

^{page 255 note a} It may be noted that the plunger socket of one barrel has dropped to a lower position in the barrel than it would ever reach when in work, and has there become fixed by corrosion.