^{page 1 note 1} Common abbreviations, pp. 80–1.

^{page 2 note 1} JRS. xliii, 89.

^{page 2 note 2} Hants, IV, sheets 12 and 16. Where outside the walled town, the relevant parcel numbers are marked on pl. 1. Inside, the Roman detail is too close.

^{page 2 note 3} An interim report appeared in PHFC. xviii, 1, 8–21.

^{page 4 note 1} The normal shades, light for bank and dark for ditch, are reversed in this photograph, owing to the rapid germination of weeds after a period of prolonged drought in July 1957. Note the appearance of the streets and their central drains.

^{page 4 note 2} A slight hollow here runs into the stream-valley in which the public baths were situated, and it is possible that a subterranean stream, or a drain inserted when the boundaries of Chalk Meadow were modified, was responsible for the proton magnetometer and resistivity results, Cf. H. Maclaughlan's plan, ArchJ. 1850, A. lix, 363 ff. and a vignette of Isaac Taylor's Map of Hampshire, 1759.

^{page 4 note 3} When Chalk Meadow was put down to wheat in 1961, the growth was unfortunately so excessively lush that no crop-marks appeared to decide the question.

^{page 5 note 1} Pale upon first exposure, this substance (hydrous ferrous phosphate) darkened later to a bright blue.

^{page 6 note 1} A. xcii, 153, fig. 11, nos. 19–20.

^{page 6 note 2} Ibid. 137.

^{page 9 note 1} Cf. Cam. form 113, c. A.D. 10–61.

^{page 9 note 2} Under the street in this test-section occurred the linch-pin (fig. 7b, no. 7, p. 50) published AJ. xxvi, 221, and pottery (fig. 13, nos. 127–8, p. 69).

^{page 9 note 3} RS. 182.

^{page 9 note 4} F. H. Thompson, Roman Cheshire (1965), 75, fig. 19, Fio.

^{page 9 note 5} RS. 51.

^{page 10 note 1} A. lxi, 476 ff., pi. 83.

^{page 10 note 2} A. lxii, 322–3, fig. 4.

^{page 10 note 3} A. xcii, 129, 133. Probably late second century rather than c. 160–70 as suggested loc. cit.: see Frere, Britannia (1967), 250–1; cf. RS. 75.

^{page 10 note 4} A. xcii, 133. Near the north gate, however, it was left open (A. lxii, 325).

^{page 10 note 5} A. lxii, 322 and fig. 3, lowest.

^{page 12 note 1} RS. 206–7.

^{page 12 note 2} Kindly identified by Mr. B. R. Hartley, F.S.A.

^{page 12 note 3} AJ. xlvii, 280 ff.

^{page 12 note 4} Another square shaft appeared in the south side of H(4). A third square cut, in the north side of H(1), seemed to be recent.

^{page 13 note 1} Found at Theale (Berks.), TNDFC. viii, fig. 21, 37 (the text mentions two, but there is only one in this series at Reading Museum). Except for Selsey in the farthest south (AJ. xiv, 49), the Hurstbourne Tarrant aberrant urn (ArchJ. lxxxvii, 307, fig. 32, no. 1) apparently completes the list, since Mr. Wm. Manning found none in his recent excavations at Ufton Nervet near Theale.

^{page 14 note 1} May, pl. 76, no. 8. Claudio-Neronian samian and an intrusive Lezoux 33 in the filling above this group of ten.

^{page 14 note 2} A. xc, 24–5. See further, p. 36.

^{page 14 note 3} Ver. 153 ff.

^{page 14 note 4} Ditch and bank are about the same width (p. 4). See p. 11 for possible addition of material.

^{page 14 note 5} Hill-Forts of Northern France (1957), 8 ff.

^{page 14 note 6} As suggested (in litt.) by Professor C. F. C. Hawkes.

^{page 14 note 7} Cf. now PPS. xxxi, 287–8 (Ann Birchall) and Ant. xxii (1968), 10 (C. F. C. Hawkes).

^{page 14 note 8} A. xc, 6.

^{page 14 note 9} PPS. xx, 6.

^{page 14 note 10} Celticum (Suppl. à Ogam-Tradition Celtique no. 73–5, 1961), 104–5.

^{page 15 note 1} A. lxii, 317. But Col. J. B. P. Karslake, F.S.A., found it (with an entrance—or later disturbance?) in The Beeches, parcel 126, on our line, in 1911, PHFC. vii, 43–4.

^{page 15 note 2} ArchJ. viii, facing 227.

^{page 16 note 1} A. xcii, 137–8.

^{page 16 note 2} Kindly undertaken by Dr. Ian Cornwall. From c. 0.080 mgm./gm. in the core, and 0.096–0.092 in the overburden, to 0.320 mgm./gm. at the line of junction.

^{page 16 note 3} A. xcii, 140.

^{page 16 note 4} A. xcii, 137–40, pl. 38.

^{page 17 note 1} Ancient Wilts, ii (1821). The Roman Æra, facing p. 56.

^{page 17 note 2} A. lxii, 317–9; uselessly small generalized profiles, fig. 1 (upper).

^{page 17 note 3} The sites of the 1909 cuttings are taken from a 25 in. Ordnance Plan in the library of the Society of Antiquaries.

^{page 18 note 1} A. lxii, 317.

^{page 18 note 2} A. lxii, 318, 330. The pot mentioned is Antonine and is in the Calleva Museum.

^{page 18 note 3} A. xcii, 137, pl. 38.

^{page 19 note 1} RS. 176–7, 192–3, and refs. Of particular interest is the vast deposit of ox-jaws, calculated to represent about 2,520 beasts, found in Insula VI and dated to the first century A.D.: A. lx, 157–8.

^{page 19 note 2} The excavation of this site was supervised by Mr. A. M. ApSimon, to whom my thanks for a careful record are due.

^{page 20 note 1} This was one of the shards stated in the interim report, p. 19, to have been found on the base of the ditch (our no. 131). It was, in fact, found where stated in the present report, at the base of the filling, not that of the silting. The other shard indeed occurred on the base of the ditch, but in black filling where the silting had been removed: see next paragraph.

^{page 21 note 1} Nos. 2219–2314. Constantine II, 2259.

^{page 21 note 2} Monumenta Britannica, MS. Bodl. Top. Gen. C. 24, f. 211.

^{page 21 note 3} The Ministry of Public Building and Works began the total clearance and conservation of the town wall in the autumn of 1966, and if bastions exist, they should at last now come to light.

^{page 22 note 1} Locus classicus, ArchJ. lxxxvii, 291 ff. (Hawkes and Dunning). Most recent, S.S. Frere, Britannia (1967), 40–1.

^{page 22 note 2} Allen, 117–8, gazetteer, 199–203.

^{page 22 note 3} Allen, gazetteer, 208.

^{page 22 note 4} Site described, RS. 49–50. One of the few to discern the excellent tactical situation of Calleva is E. W. Gilbert, in H. C. Darby (ed.), An Historical Geography of England (1936), 50.

^{page 22 note 5} Listed, RS. 218, note 27.

^{page 22 note 6} Seen and photographed; in private hands, 1951. Lost.

^{page 22 note 7} Keller, J., Das keltische Fürstengrab v. Reinheim, I (1965), Taf. 26a, 26b, 1–2Google Scholar. Rare.

^{page 22 note 8} Allen, 113–16; gazetteer, 163. The British Q, gazetteer, 200. Picked up by Capt. J. S. Eyton from the surface of a ride in the plantation west of Pond Farm, and still (1968) in his collection at Silchester.

^{page 22 note 9} SU 627631. Described, FAH. 387, plan facing 386. About 4 acres.

^{page 22 note 10} WAM. xlvi, 579–624.

^{page 22 note 11} Professor Hawkes has kindly suggested further comparative sites to me, of which the much smaller Knap Hill (Wilts.) ‘plateau enclosure’ of WAM. xxxvii, 42–65, is undoubtedly Belgic, and of ‘defensive’ rather than ‘agricultural’ type.

^{page 23 note 1} Such a wide area is exactly what the Camulodunum excavations have led us to expect of Belgic oppida: Cam. 8 ff., 50–1, and pl. 1.

^{page 23 note 2} Allen, 105.

^{page 24 note 3} The Flex Ditch is described by Williams-Freeman, 7 Mack, nos. 107–8, Allen, gazetteer, 213. Two gold

^{page 24 note 1} These four, Allen, gazetteer, 174, 200, 244, 172.

^{page 24 note 2} Aspects of Archaeology (1951), 105–7, figs. 22–3; cf. also Fowler in WAM. lix, 55–6. Once more I owe the suggestion to Professor Hawkes.

^{page 24 note 3} The Flex Ditch is described by Williams-Freeman, FAH. 407, curiously as a ‘modern digging for soil’, but as Crawford realized (note on O.S. 6 in. map) it is an ancient earthwork. Called Flax Ditch in the Tithe Award, presumably because the ditch, dammed in effect by the passage, on the east, of the Roman road to Sorviodunum, was used for steeping flax. I have been unable to discover the authority for a statement in Miss Florence Davidson's manuscript ‘History of Silchester’ (1914), copy in Reading Public Library, that ‘Flex comes from a pond or stream that comes and goes according to the season’, although the pond does this.

^{page 24 note 4} When Allen wrote his great 1944 paper on the Belgic dynasties of Britain and their coins (A. xc), he was inclined to believe that the Commius of the coins was not the person mentioned by Caesar and Frontinus. It is more probable that he was, in fact, the same person.

^{page 24 note 5} Aspects of Archaeology, 338 f.; Hawkes in Bag. 46.

^{page 24 note 6} Mack, nos. 93–5.

^{page 24 note 7} Mack, nos. 107–8, Allen, gazetteer, 213. Two gold from Sussex, two silver from Wallingford (Berks.) and Windsor; the silver coin in Copenhagen has no provenance (information from Dr. O. Mørkholm).

^{page 24 note 8} A point also noted by Frere, JRS. lii, 273, and Britannia, 42.

^{page 24 note 9} Allen, gazetteer, 209.

^{page 24 note 10} Refs., RS. 33.

^{page 24 note 11} A. lxi, 215–18. RIB., no. 70.

^{page 24 note 12} A. xc, 7 f., table, 44–5.

^{page 24 note 13} Mack, no. 97.

^{page 24 note 14} Two out of ten examples in the British Museum read C B. Mr. Allen recognizes two c a and one c b dies (information in litt.). Evans's c f (Coins of the Ancient Britons (1864), pl. 2 no. 4) is presumably a mis-read c b.

^{page 25 note 1} For similar marks in the Roman Republican series, see BMCRR. i, 251 ff., including c a. The system was however long obsolete by the time of Tincommius, and was only used for very large issues. Allen (in Hit.) points to possible parallels in the British series: the mark c o of Mack, no. 288, and various devices on Dobunnic and Icenian coins. The plain or dotted ca mv of Mack, nos. 203, 210–13 may also be interpreted as evidence of a control-system.

^{page 25 note 2} AJ. xxxiv, 68 ff.; two more fragments, near Forum, 1961.

^{page 25 note 3} Cam. 132, table. No exact comparison between metals absorbed by a clay mould and those present in the alloy of a coin is possible.

^{page 25 note 4} Cam. 142.

^{page 25 note 5} S. S. Frere, Roman Canterbury (3rd ed. 1962), 6.

^{page 25 note 6} Allen, gazetteer, 276, 277, 279, 278, 172.

^{page 25 note 7} May, pl. 2, I am much indebted to Professor Howard Comfort for much help and extensive comments on the Arretine from Silchester.

^{page 26 note 1} This Provinzialbetrieb is curiously indistinguishable in fabric from genuine Arrezzo products, as I have myself tested. Why should this be so?—A study of the Ateius question: E. Ettlinger, R.C.R.F. Ada iv, 27–44, with 12 maps.

^{page 26 note 2} PPS. 1954, 8.

^{page 26 note 3} May, pls. 1, 3, 4, 5, Stamps, pl. 81 a.

^{page 28 note 1} May, pl. 81a, nos. 1, 3, and 15 are Montans ware 16 is Lezoux; p. 198, no. 16a, is also Gaulish. Pl. 81a, no. 3, on form, is evidence for Apronius well before Oswald's suggestion (Potters’ Stamps, ad loc.) of ‘Domitianic’.

^{page 28 note 2} Cam. esp. 189–91. Tardo-Italic ware is not taken into consideration in the present remarks.

^{page 28 note 3} JW. 124–5. Pottery, ibid., AJ. xiii, 58–9, ascription, Comfort, Coll. Latom. lviii, 454; AJ. xviii, 275

^{page 28 note 4} A. lxxviii, 74 ff. Ascriptions of decorated ware, Stenico, Revisione critica (1960), 38, nos. 439–43: P. Cornelius, signed and two others not: M. Perennius Crescens (A. no. 10) and tardo-Italic (A. no. 12). The lack of corresponding evidence of pre-Conquest occupation at London has caused most modern writers to view the Arretine with caution, e.g. Wheeler in Roy. Comm. Roman London (1928), 24–7. Merrifield (Roman London (1966), 29–32) re-emphasizes the wide distribution of the ware, the same as that of Claudio-Neronian pottery.

^{page 29 note 1} The position at Leicester seems to be the same. All five in the recent excavations came in Roman contexts (JW. 9, 43–4).

^{page 29 note 2} No Arretine at Richborough, none at Valkenburg, three scraps only at Hofheim (Ritterling, Hofheim 1912, 201). In Britain, the Margidunum piece (AJ. xx, 284) seems the only certain Roman introduction, plus probably the extraordinary Trent Vale (?) piece in Hanley Museum (Comfort, Coll. Latom. lviii, 448–56).

^{page 29 note 3} Cam. 190.

^{page 29 note 4} Bag. 210–11, fig. 45, rightly omitting Hengistbury and Oare (Gallo-Belgic and south Gaulish), but also Leicester and Heybridge (Essex) with its P. Hertorius stamp noted by Oswald and Pryce, Introd. to T.S. (1920), 5 and correctly read by Comfort (Am. Journ. Arch, xlvi, 92, note 15); not mentioned in VCH Essex iii (1963), 146–7.

^{page 29 note 5} The phrase is added because of the Period IV picture (p. 36). The Dr. 27 from Southampton (G. B. Rogers in Gallo-Roman Pottery from S. (Southampton Museum Publn. no. 6, 1966), no. 2) is tardo-Italic and may be discounted.

^{page 29 note 6} This section cannot be closed without reference to the M. Perennius Tigranus bowl from Lake Nemi (Comfort, R.C.R.F. Ada ii, 6, Coll. Latom. lviii, 455) kept with the Nemi ships material in the Terme (inv. no. 594). As such it implies a Neronian date for much high-quality Arretine of typologically earlier appearance. Professor Comfort however informs me (in liit. 2 April 1967) that although listed with the ships material in the edition of Le Navi di Nemi accessible to him when he wrote these papers—wherein a late date is suggested for Arretine in Britain—another edition clearly states that the bowl was found elsewhere in the lakeside; he adds that Miss Ann Brown of the Ashmolean Museum drew his attention to the discrepancy.

^{page 29 note 7} Cam. 179–80.

^{page 29 note 8} May, pl. 2, no. 10 (borderline); pl. 11, nos. 1 and 2; no more in reserve.

^{page 29 note 9} May, pl. 81a, nos. 1, 3, 15; pl. 81c, nos. 1, 2, 3, cf. Oxé, Arch. Anz. 1914, 70, Abb. 3, nos. 8–10.

^{page 29 note 10} AJ. xlvii, 30–1, vessels nos. 1, 1b, 7.

^{page 29 note 11} Déch., i, 135, mentions the commonness of Montans ware at Bordeaux, the likely exporting centre for Britain.

^{page 29 note 12} Although none of the decorated Tiberian sigillata of Cam. pls. 21–2 was found in Period I levels at that site, there is a suspicion that since none of so early a date is published from Richborough or Valkenburg (Cf. Richb. iii, 94 ff., pls. 22, no. 1, 2; pl. 23, no. 1; pl. 24, no. 1, Richb. iv, 160 ff., nos. 1, 3, 6, 9, for the earliest, Tiberio Claudian/early Claudian; Glasbergen in 25–8 Jaarverslag … Terpernonderzoek, Bijlage iv, 206 ff., has even fewer pieces of similar date: Afb. 55, no. 3; Afb. 59, no. 1, and Afb. 60, no. 6, are all that were dated as early as ‘vroeg Claudius’), some may have reached Sheepen before the Conquest, to be, evidently like very much of the Arretine, carefully kept and treasured.

^{page 30 note 1} 237–8, 215–21; add Casterley, WAM. xxxviii, 103, pl. 8, 31–2; Crookham, B.M. unpublished (information from Prof. Hawkes); Cunning Man, BAJ. lvi, 46–53 (Reading, Berks.).

^{page 30 note 2} May, pl. 82a.

^{page 31 note 1} A. lv, 548; May, 186–9, correcting pit number.

^{page 32 note 1} May, pl. 77.

^{page 34 note 1} A. lvii, 97, fig. 3.

^{page 34 note 2} A. lxi, 210–11, May, pl. 76.

^{page 34 note 3} May, pl. 78.

^{page 34 note 4} Westf. Mitt, v, 286.

^{page 34 note 5} Cam. 238.

^{page 34 note 6} It may be added that no. 3 (cf. May, type 152, but plain; cf. PPS. xxxi, 343, no. 174, Colchester Museum) is similar to a vessel containing a cremation-burial, mentioned by Radford (PPS. xx, 22) from the site (exact location unknown). The graffito on the base of no. 3, read by May as X Θ A, is X Φ A. This graffito is of interest in connection with Caesar's statement that the Gauls used Greek letters (Bello Gallico vi, 14, but contrast v, 48 where a letter is written in Greek characters to confuse the Gauls), since the pot cannot well be later than the Roman invasion and may be considerably earlier. The only other pre-Roman graffiti known occur at Camulodunum (Cam. 284 f.), but are not in Greek letters; there is one possible Roman-period example, ibid., no. 30.

^{page 34 note 7} J. G. Joyce ‘Journal of Exc.’ sub Oct. 27, 1870. Illus., Ant. v, pl. 2, no. 5.

^{page 34 note 8} Pattern and Purpose (1958), 76. But the question is not argued.

^{page 34 note 9} Cf. Strabo, IV, 6, 3.

^{page 34 note 10} Inlay: A. lv, 430. Cf. Glass from the Ancient World (1957), no. 117, etc. Vessel: cf. Haltern, Westf. Mitt, ii, Taf. 33, no. 2, p. 172, and Cam. pi. 87, no. 37, p. 297, pre-Roman like ibid. 46 and 47a. The Silchester pieces will be fully published in my account, for Reading Museum, of the glass from the site, 1864 to date.

^{page 35 note 1} Monum. Ancyr. (ed. Hardy), 145.

^{page 35 note 2} Described FAH. 322–4, 406–7, and well, except for a perverse interpretation of the main lines as Roman roads.

^{page 35 note 3} Isaac Taylor's Map of Hampshire (1759) marks it as ‘Roman road to Winchester with a division ¾ mile off to Old Sarum’. This ‘division’ is probably part of our Dyke C. Colt Hoare (Ancient Wilts, ii. 51) also regarded the dyke as a Roman road and his plan shows a branch ‘off the Winchester line’ in the wood adjacent to Rampier Copse. There is no fork visible now, and Maclauchlan saw none (ArchJ. 1850, 232, note 5). 1 have traced the dyke to the slope of O.S. parcel 154, and beyond there are suggestive indications as far as Elliott's Farm, where a declivity begins. The length, about 5 furlongs.

^{page 35 note 4} Williams-Freeman (FAH. loc cit.) failed to find a connection between Dykes C and D, and Crawford (note on O.S. map) thought that the present termination of Dyke D (Byes Lane) was original, founded perhaps on dense vegetation. Dyke C is about 11 furlongs in length. Williams-Freeman noted that the surface of the dyke, in O.S. parcel 158, contained gravel, and observed that the ploughed-out section north of Early Bridge Copse was marked by a trail of flints and gravel foreign to the surroundings. It may have been the destruction of the dyke at this or another spot which led Ward (Phil. Trans, abr. ix, 1744–9 (1809), 601) to remark that ‘from the south gate towards Winchester has lain a military road, which, when broken up, appears to have been pitched with flints.’ I have never seen any part of the dyke under cultivation, and so cannot confirm the metalling; but Miss Murray has observed that one of the Chichester dykes possessed a bank probably deliberately cobbled with very large flints (SAC. xciv, 139–43, Devil's Ditch, S. of Lavant House.). Stukeley (Itin. Curios. (1724), 171) evidently muddled his notes on these dykes: Dyke C was never called Grimesdike, the name of the Padworth dyke (fig. 1, f) but may, I suppose, have been called Longbank.

^{page 36 note 1} Allen, 221, 230–1, 237.

^{page 36 note 2} Dio, lx, 19.

^{page 36 note 3} Allen, 238.

^{page 36 note 4} AJ. xiv, 49, fig. 5, no. 14 (M. G. White).

^{page 37 note 1} Tacitus, Agricola, 14.

^{page 37 note 2} RIB., no. 91.

^{page 37 note 3} No. 7 of the British section.

^{page 37 note 4} Tacitus, Annales, xii, 31.

^{page 37 note 5} In the same passage, the historian refers to attacks on the agri sociorum (of the Dobunni in the west, presumably, and by Caratacus). It is unlikely that the same word would be used in a different sense a few lines apart, See Lexicon Taciteum (edd. Gerber and Greef, 1903) for other instances of Tacitus’ use of the words socius, socialis; they were ordinarily employed to denote ‘allies’ whether or not of client status.

^{page 37 note 6} Brigantes, Seneca, Apocolocyntosis, 12; Dobunni, Hawkes, Bag. 66 f.

^{page 37 note 7} Cf. Stevenson, Roman Provincial Admin. (1939), ch. 2.

^{page 37 note 8} S. C. Claudianum, Dessau, Inscr. Lat. Sel. no. 212, col. 1 ad fin. A. Momigliano, Claudius (2nd ed. 1961), 54 f.

^{page 38 note 1} Roman Britain (1955), 23. But possibly the added civitates, as suggested in the present text, were ruled by him as legate and not as king.

^{page 38 note 2} Annales, xii, 31. Bradley, Collect. Papers (ed. Bridges, 1928), 241–4. For ‘direct rule’, cohibere, cf. Tac. Hist, i 11, (provinciae) procuratoribus cohibentur.

^{page 38 note 3} Agricola, 14.

^{page 38 note 4} Prosopogr. Imp. Rom. iv. 3 (1966), 108, no. 14.

^{page 38 note 5} I am indebted to Professor Birley for this additional information.

^{page 38 note 6} The alternative explanation, that these appointments were occasioned by Agricola's prolonged absences in the north, is less compelling. Although he was, in 80–1, engaged in the conquest of novas gentis (Agr. 22), these and other campaigns occupied only the summer season (cf. Agr. 21, sequens hiems saluberrimis consiliis absumpta), leaving the winter for civil work. Furthermore, as Birley points out, the two juridici were very senior and eminent men, unusually so, in fact (Prosopogr. loc. cit.).

^{page 38 note 7} Cf. Furneaux's Agricola (ed. Anderson, 1922), p. xix, for date of Tacitus' birth, which is not exactly known.

^{page 38 note 8} For Agricola's attitude to legal work, see Agricola, 9.

^{page 38 note 9} Prosopogr. loc. cit.: leg. Leg. III Aug., 83; consul suffectus, 86.

^{page 38 note 10} It has been claimed that the apparent slightness of information. occupation at Chichester before Flavian times suggests a change in status, viz. the creation of the civitas capital, then: see A. E. Wilson, SAC. xciv, 102 f., where Professor Birley's suggestion regarding the juridici also appears; and J. Pilmer's analysis of the pottery finds, ibid. 111 ff. But the existence of substantial inscriptions of the regal period, RIB. nos. 91 and 92, disproves it.

^{page 39 note 1} See now Frere, Britannia (1967), 73–5.

^{page 39 note 2} A. xcii, 124–5.

^{page 39 note 3} S. S. Frere, Antiquity 1964, 103–7; Univ. Lond. Inst. Arch. Bulletin 4, 61–71.

^{page 40 note 1} A. Clarke, PHFC. xxi, 82 and fig. 1; RS. 206–8.

^{page 40 note 2} Cunliffe, B.W., Winchester Excavations 1949–60, i (1964), 1–24Google Scholar.

^{page 40 note 3} A. lix, 340 ff. A tile with a Neronian stamp from the cess-pool is often taken to indicate an early date for the erection of the baths; see RS. 69, 108, 226, note 65. The fate of the tile said to bear the same inscription and found at Little London near Silchester is unknown (it was not among Col. Karslake's collection which formed the basis of the Calleva Museum, established in 1951) and no illustration of it has ever been published. (AJ. vi, 75; tile in PHFC. xvi, 59 is the Silchester tile.) I have found only 17th/18th cent, tile-débris on the site, but a little Roman material is in Basingstoke Museum.

^{page 40 note 4} Antiquity, 1948, 172 ff.

^{page 41 note 1} Insula I, House 1: see RS. 36, 229, note 36. IX, 3 had Antonine samian in a pit below one (two?) floors: A. liv, 347.

^{page 41 note 2} Frere, Bulletin, loc. cit. 67. Buildings of wood on flint foundations, however, certainly existed earlier, as Lady Fox showed.

^{page 41 note 3} RS. 203–6.

^{page 41 note 4} For the sake of the simplicity of the narrative, it is left to this footnote to point out that the exact production of the London road alignment takes its course 15 yds. N. of the intersection of the two western alignments. The London road was aligned upon the little spur where the E. Gate of the town was later to arise, just N. of the parish church; the east entrance of the Inner Earthwork could not have been discerned from farther east (see fig. 1). There may thus have been a very slight deflection intended west of this spur.

^{page 43 note 1} Journal, sub. July 16, 1867. Stratification not certain. Reading Museum. Fora tend to be of this date in southern Britain (Winchester, ArchJ. cxix, 153–5, Verulamium, Cirencester).

^{page 43 note 2} A. lx, 157–8. See p. 19.

^{page 43 note 3} H. Kähler, Röm.-germ. Forsch. xiii, 32, Type c. See RS. pi. 10.

^{page 44 note 1} The alteration in the position of the main east entrance, placed 5 ft. south of the axial line, is presumably due to the formation of Insulae V and VI. The entrance was designed as a monumental archway: see A. liii, 542–3. On the street, see RS. 222, note 14.

^{page 44 note 2} A. liii, 268, Ins. VIII, Pit 5; RS. fig. 11. A. liv, 430, fig. 6, RS. fig. 11. Exact parallel at Vechten, Archaeol. Traject. iii, pi. 2, no. 2, ‘silver’: ours may be tinned bronze or base silver.

^{page 51 note 1} The important recent publication of the Valkenburg leather by Dr. W. Groenman-van Waateringe (Nederlandse Oudheden, 11, 1967) allows the suggestion that these fragments are more probably parts of tents; cf. fig. 63, 1 or 2 for pieces like the triangular fragment and fig. 24, etc. for panels sewn like the other piece. The leather used was mostly goat-skin.

^{page 80 note 1} For our no. 209 type, cf. Bonenfant, loc. cit., 522 fig. 9 no. A, 525 fig. 11 no. 2.