^{page 41 note 1} Boon, G. C., Carms. Antiq. vii (1971), 136–8Google Scholar.

^{page 41 note 2} Mr. Jones wrote after his retirement to say that the finder had misled him as to the place of discovery and that the brooch could have come from anywhere. But we had understood that the of what had been an unrecognizable lump of corroded matter, to which the rosette-headloop had not been considered to belong, was Mr. Jones himself. And without wishing to enter more deeply into the matter, we affirm our belief that the circumstances of discovery as originally communicated, such as they were, were correct.

^{page 41 note 3} The Archaeology of Roman Britain (1930), pp. 251–4Google Scholar.

^{page 41 note 4} Notably by Atkinson, Donald, Wroxeter 1923–7 (1942), p. 207Google Scholar.

^{page 42 note 1} Britannia, i (1970), 270–1, fig. 2Google Scholar.

^{page 42 note 2} By Messrs. Minton, Treharne and Davies Ltd., Analysts and Chemical Consultants, of Cardiff; and likewise for the detection of mercury, below, p. 44.

^{page 43 note 1} Cope, L. H., Num. Chron. (1967), 113Google Scholar, tabulating the 1908 results of Hammer, J.; B.M.C., Coins of the Roman Empire, iii (1936), p. xxiGoogle Scholar.

^{page 43 note 2} Strabo, iv, 199; cf. Boon, G. C., ‘Aperçu sur la production des métaux non-ferreux …’, Apulum, [Alba-Iulia] ix (1971), 454–6Google Scholar.

^{page 43 note 3} Caley, E. R., Orichalcum and related ancient Alloys (Num. Notes and Monogr., 151, 1964), pp. 81–4Google Scholar, tables.

^{page 43 note 4} Zinc-rich copper ores occur, e.g. at Nant-yr-arian (Cards.), but most of the zinc would have been vapourized in smelting. The zinc in the Tal-y-llyn bronzes was probably incorporated by cementation (packing copper billets with ore and heating): cf. Caley, ibid., pp. 11–12; Savory, H. N., B.B.C.S. xxii (1968), 101–2Google Scholar.

^{page 43 note 5} Ibid. 96–7; and ibid. xx (1964), pl. 4, 1, left for the tumbler-lock keyhole plate.

^{page 43 note 6} Lead models are very uncommon. One for an unspecified kind of ‘fibula’ was found at Brough (Westm.), P.S.A., 2nd ser., vii (1876–8), 19Google Scholar, probably the best evidence there is for broochmanufacture there, otherwise highly deficient, as at the other oft-quoted sites of Kirkby Thore (river-votive deposit?) and Traprain (enamelling?). The existence of a very close match, virtually a duplicate, for the Segontium brooch at Wroxeter (see Appendix) implies the existence of a lead model, Part of a lead model, for a sixth-or seventh-century penannular brooch, has been published by Alcock, L., Dinas Powys (1962), pp. 120–2Google Scholar, fig. 23.

^{page 44 note 1} Method, H. Maryon, Metal-work and Enamelling (1954 ed.), p. 159.Google Scholar Cf. Vitruvius, vii, 8, 4, writing of mercury (argentum vivum): ‘neque argentum neque aes sine eo potest recte inaurari’. See further Forbes, R. J., Stud. Ancient Technology, viii (1964), 173–5.Google Scholar See also below, n. 8.

^{page 44 note 2} Smith, R. A., Archaeologia, lxi (1909), 341Google Scholar, fig. 9.

^{page 44 note 3} May, J., Antiq. Journ. l (1970), 243, pl. 32BGoogle Scholar.

^{page 44 note 4} Hawkins, E., Arch. Journ. viii (1851), 39Google Scholar, facing.

^{page 44 note 5} Fox, Sir Cyril, Pattern and Purpose (1958), pl. 41DGoogle Scholar.

^{page 44 note 6} (Sir) Evans, Arthur, Archaeologia, lv (1897), 179–96Google Scholar, fig. 9. For a note pointing out that this ‘hoard’ was probably buried in the late third century, see Charlesworth, D., Antiq. Journ. lii (1972), 354Google Scholar, and further her paper ‘The Aesica hoard’, Arch. æl. ser. 5, i (1973), 225–34Google Scholar, with fine new photographs.

^{page 44 note 7} We are much indebted to Mr. May for generously placing full details of the Dragonby brooch, and the two bronze brooches of similar type and likewise from pre-conquest levels of the site, at our disposal.

^{page 44 note 8} Strong, D. E., in I. M. Stead, Archaeologia, ci (1967), 20–2Google Scholar; A. E. Werner and H. Barker, ibid. 23. In view of Vitruvius' statement (above, n. 1) it is very curious that mercury-gilding was not employed.

^{page 44 note 9} Personal inspection, June 1973, of the cups published, Archaeologia lxiii (1911-1912), pl. 2Google Scholar, suggested this was so, and gilding has now been proved by analysis at the British Museum. Mercury again does not appear to have been employed (inf. kindly sent by Dr. Ian Stead).

^{page 45 note 1} This wire ring exhibits a minute longitudinal ridge, which is consistent with the wire having been drawn. For a note on this technique in antiquity, as applying to gold, a still more ductile metal, see Carroll, D. L., Amer. Journ. Arch. lxxvi, 3 (1972), 321–2.CrossRefGoogle Scholar Drawn wire, mostly brass, Roman, see Epprecht, W. and Mutz, A., Jahrb. schweiz. Ges. f. Ur- u. Frühgesckichte, Iviii (1974-1975), 157–61Google Scholar.

^{page 45 note 2} Bretz-Mahler, D., La Civilisation de la Tène I en Champagne (23. Suppl. à Gallia, 1971), 20–1Google Scholar; 35; pl. 5, 1 etc.; Krämer, W., ‘Silberne Fibelpaare aus dem letzten vorchrl. Jahrhundert’, Germania, xlix (1971), 111–32Google Scholar [cited hereafter as Krämer].

^{page 45 note 3} e.g. the Chesterford, Great and brooches, Folkestone Warren, Fox, , Pattern and Purpose, pl. 40BGoogle Scholar, Krämer Taf. 24; Allen, J. Romilly, Reliquary and Illus. Archaeologist, N.S. vii (1901), 197Google Scholar, fig. 1.

^{page 45 note 4} Ibid. 198, fig. 2, here reproduced as fig. 2, 4.

^{page 45 note 5} Collingwood, R. G., Antiq. Journ. xi (1931), pl. 16Google Scholar.

^{page 45 note 6} Cook, N., Finds in Roman London 1949–52 (Guildhall Mus.), fig. 9Google Scholar; personal inspection.

^{page 45 note 7} e.g. Ferri, S., Arte romana sul Danubio (1933), figs. 50, 70, 129Google Scholar (Klagenfurt); Wild, J. P., Bonner Jahrbücher, clxviii (1968), 207Google Scholar, fig. 26.

^{page 46 note 1} Bretz-Mahler, , op. cit., pl. 31 etcGoogle Scholar.

^{page 46 note 2} Archaeologia, lxxx (1930), 46Google Scholar; cf. M, B.., Early Iron Age Antiqs. (1925), p. 96Google Scholar

^{page 46 note 3} Allen, J. Romilly, Celtic Art in Pagan and Christian Times (1904), p. 103Google Scholar, of the Chorley brooches: ‘the knob, which is the survival of [the La Tène II-III] collar… has here assumed a highly ornamental form resembling two floriated capitals of columns placed together.’

^{page 46 note 4} Archaeologia, lxxx (1930), 46–7Google Scholar; quotation partly from here and partly from The Archaeology of Roman Britain (1930), p. 253.Google Scholar (Though awkwardly placed for his theory, he might possibly have cited the Chester Composite capital, which is strikingly like the moulding, if one must see things so: Chester Arch. Journ. N.S. vi (1899), pl. facing 277Google Scholar; see now ibid. xxxviii (1951), 36, fig. 16.)

^{page 46 note 5} Archaeologia, lxxx (1930), 49Google Scholar, on fig. 6a; but the Beckhampton brooch mentioned is of a different type.

^{page 46 note 6} Jacobsthal, P., Early Celtic Art (1944), pl. 38Google Scholar, no. 44; pls. 40–1, no. 47; pls. 128–9, nos- 214, 216; Bretz-Mahler, , op. cit.Google Scholar, passim.

^{page 46 note 7} Cf., however, Smith, R. A., B.M.Q. V (1931), pl. 7Google Scholar; Jacobsthal, pi. 39, no. 45. Some others have enamel studs moulded as flowers, with central decorative metal rivets for attachment.

^{page 47 note 1} Pattern and Purpose, pl. 26B: a La Tene I type, cf. Jacobsthal, pl. 33.

^{page 47 note 2} Cf. the tore in the British Museum cited above, p. 46, n. 7.

^{page 47 note 3} e.g. from Boscoreale, , Strong, D., Greek and Roman Gold and Silver Plate (1966), pl. 38Google Scholar A: notice the open-flower rosettes. Krämer mentions parcel-gilt brooches (dissimilar pair in a La Tène II grave, Dühren), 130, Taf. 30, 2–3.

^{page 47 note 4} Hawkes, C. F. C., Antiq. Journ. XX (1940), pl. 54B, 348Google Scholar, fig. 3 (after Curie).

^{page 47 note 5} Fox, , Pattern and Purpose, pl. 40AGoogle Scholar.

^{page 47 note 6} Krämer, 121, Abb. 4, nos. 3–5. By Allen's dating (‘The origins of coinage in Britain’, in Frere, S. S. (ed.), Problems of the Iron Age in Southern Britain [1960], pp. 297–8Google Scholar, following Colbert de Beaulieu) these brooches are provided with a terminus ante quem of 56–51 B.C.; but Krämer, 128, thinks that this is too early; and that because there are other Jersey hoards of Armorican coins which end with Roman coins of 39 and 32 B.C. respectively, Le Catillon hoard may well be post-Caesarian also.

^{page 47 note 7} Proc. Univ. Bristol Spelaeological Soc. 1. 3 (1921-1922), pl. 24Google Scholar, poor drawing; personal inspection, January 1973, by courtesy of Prof. E. K. Tratman, F.S.A. Though this brooch survived the 1940 air-raid which destroyed the Society's museum, it is now very frail, and the other La Tène III brooch on the same plate perished.

^{page 47 note 8} Bushe-Fox, J. P., Swarling Report (1925), p. 13, etcGoogle Scholar.

^{page 47 note 9} M, B.., Antiqs. ofRoman Britain (1922), fig. 57Google Scholar; Allen, Romilly, Celtic Art, pl. facing p. 102Google Scholar, top two.

^{page 47 note 10} Thompson, F. H., Antiq. Journ. xliii (1963), 290Google Scholar, fig. 2 (silver-inlaid Rii with early moulding). (To his list of this type add Gardner, W. and Savory, H. N., Dinorben (1964), pp. 133–4Google Scholar, fig. 16, 1.)

^{page 47 note 11} Brewis, W. Parker, Arch. Ael.ser. 3, xxi (1924), pls. 7, 7 and 9, 2Google Scholar.

^{page 47 note 12} M, B.., Antiqs. of Roman Britain (1922)Google Scholar, fig. 62 (from Hawkins, , Arch. Journ. viii)Google Scholar; inadequate in 1951 ed., fig. 9, 17. The Backworth hoard is obscure. The best account is Haverfield's, in A Hist, of Northd. ix (1909), 26–32.Google Scholar Of 280 denarii, the latest was said to be of A.D. 139, but no others are specified, and two sestertii of Pius were not dated nor their condition given. For a step towards at least a kunsthistorisch dating of the brooches, compare the similar style of engraving on the Romanizing umbo from Peter-borough, , M, B.., Later Prehist. Antiqs. (1953) flg. 25, 2Google Scholar.

^{page 48 note 1} Boon, G. C., Arch. Camb. cxvii (1968), 45Google Scholar, fig. 9–poor; better, the brooch, Polmaise, P.S.A.S. lii (1917-1918), 27Google Scholar, figs. 1–2.

^{page 48 note 2} Allen, Romilly, Celtic Art, facing p. 102Google Scholar, lower two.

^{page 48 note 3} Green, C., Proc. Prehist. Soc. XV (1949), 188, fig. 1CrossRefGoogle Scholar.

^{page 48 note 4} Krämer, Taf. 23, Lauterach and Manching.

^{page 48 note 5} Saalburg-Jahrbuch, xiii (1954), 34Google Scholar, Abb. 5, 9, and 58, Abb. 3 show curious eastern European La Tene III brooches with internal chord; they have a ‘reversed La Tène II’ foot-construction, the upper strut of the catch-plate being wrapped round the bow. They are thus the exact opposite to what the vestigial markings seem to imply.

^{page 48 note 6} Clarke, R. R., P.P.S. xx (1954), 27–86Google Scholar.

^{page 48 note 7} Owles, E., Antiquity, xliii (1969), 208–12CrossRefGoogle Scholar.

^{page 48 note 8} Brailsford, J., P.P.S. xxxviii (1972), 219–28Google Scholar.

^{page 48 note 9} About twenty years' difference in the date of Le Catillon (p. 47, n. 6) would not make much difference to this point.

^{page 49 note 1} Fox, , Pattern and Purpose, p. 33Google Scholar.

^{page 49 note 2} Ibid., pl. 19c.

^{page 49 note 3} Clarke, , loc. cit., pl. 4Google Scholar.

^{page 49 note 4} M, B.., Later Prehist. Antiqs. (1953), pl. 15, 1Google Scholar.

^{page 49 note 5} Cf. Boon, G. C., Apulum, ix (1971), 454–6Google Scholar.

^{page 49 note 6} Jacobsthal, , op. cit. (above, p. 46, n. 6), pp. 97–103Google Scholar.

^{page 49 note 7} Brailsford, , loc. cit., fig. 3Google Scholar.

^{page 49 note 8} Clarke, , loc. cit., fig. 12Google Scholar.

^{page 49 note 9} Fox, , Pattern and Purpose, pp. 49–51Google Scholar.

^{page 49 note 10} ibid., pl. 45B.

^{page 49 note 11} Spratling, M., Current Archaeology, ii (1970), 188–91Google Scholar, figs. 1–2.

^{page 49 note 12} Raftery, J., Prehistoric Ireland (1951), p. 190, 219Google Scholar.

^{page 49 note 13} e.g. Ch. Boube-Piccot, , ‘Une phalère de harnais à décor de trompettes’, Bull. d'Arch. marocaine, V (1964), 183–200Google Scholar.

^{page 49 note 14} Hawkes, C. F. C., Antiq. Journ. XX (1940), 338–57Google Scholar.

^{page 49 note 15} Beds. Arch. Journ. V (1970), 13–14.Google Scholar Upon personal inspection of the piece, we should add that the statement by Green, Charles, P.P.S. XV (1949), 189Google Scholar, followed later by Fox, that the ornament consistsof ‘caststrips rivetedon’ seemstotallyerroneous: what we have are rivets for attaching enamel studs (traces of red enamel, though now gone, are mentioned by Smith).

^{page 50 note 1} See above, p. 44, n. 6; Fox, , Pattern and Purpose, pl. 41cGoogle Scholar.

^{page 50 note 2} Ibid., pl. 12D.

^{page 50 note 3} Megaw, J. V. S., B.M.Q. XXXV(1971), 145–54, pl. 59A-BCrossRefGoogle Scholar.

^{page 50 note 4} It is not known whether, or to what extent, the Welsh sources of argentiferous lead were worked before the Roman period; but prima facie there is no reason why they should not have been worked, as the Mendip deposits almost certainly were.

^{page 50 note 5} Archaeologia, lxxx (1930), 45–6Google Scholar, and Curle, J., A Roman Frontier Post … (1911), pp. 321–2, pl. 85, 8Google Scholar.

^{page 50 note 6} Archaeologia, lxxx (1930), 44Google Scholar.

^{page 50 note 7} Ibid., 48.

^{page 50 note 8} Cf. for example the remarks on brooches, Silchester, Archaeologia, xcii (1947), 144–5Google Scholar, fig. 8.

^{page 51 note 1} Above, p. 47, n. 12.

^{page 51 note 2} Archaeologia, lxxx (1930), 45Google Scholar.

^{page 51 note 3} Ibid., 50.

^{page 51 note 4} (1930), 251–4; (1969), 296–7.

^{page 51 note 5} Wheeler, R. E. M., Y Cymmrodor, xxxvii (1926), fig. 58Google Scholar, nos. 1–6, of which no. 3 is the possible Rii, and no. 5, not a trumpet-brooch but a head-stud brooch, is from a stratified context with first to second-century pottery.

^{page 51 note 6} Roman Britain and the English Settlements (1936), p. 243Google Scholar, n. 1; (1937), ibid. The continental occurrences of trumpet-brooches are listed by Böhme, Astrid, Saalburg-Jahrbuch, xxvii (1970), 5–20Google Scholar, in common with other types of British brooch, There are indeed no trumpet-brooches from the Odenwald Limes, to which an inscription of 145–6 refers British garrisons (Schonberger, H., J.R.S. lx (1969), 161, 167Google Scholar, Map B, forts nos. 98–104). There is every possibility of earlier arrivals on the Continent in any case, not to mention other units returning from Britain. The actual dating-evidence for Rii-Riv brooches in continental contexts maybe mentioned. Jacobi, H., Saalburg-Jahrbuch, iii (1912), 165Google Scholar, referring to the Saalburg finds, says ‘nach der Fundstelle unter dem Wehrgang auf der Sohle [sie] gehören … in die Zeit nach dem Erd-kastell, also etwa in den Anfang oder die Mitte des II. Jahrhunderts’. The Erdkastell was occupied until c. 135/9, 139 being the latest date at which the Cohrs II Raetorum, c.R. arrived as garrison of the much-enlarged fort (Hartley, B. R., Saalburg-Jahrbuch, xxvii (1970), 30Google Scholar; Schönberger, , loc. cit., p. 165Google Scholar and note 164). By ‘unter dem Wehrgang’ Jacobi meant the mound behind the fort wall. But as Dr. D. Baatz has kindly informed us, there is no question of a useful terminus ante quem of 139 for so developed a type, because the fort wall was later rebuilt and the mound is associated with the second building, which dates to the second half of the second century. Other continental occurrences are in graves at Thuin (Namur province), with a coin of 15 5–6, and at Rheinzabern, with a sigillata plate stamped ABBO F, also Antonine (Böhme, , loc. cit., pp. 10, 16Google Scholar, no. 38; p. 18, no. 44—Ludowici's, illustration in Cat. iv (1912), p. 176Google Scholar, of the second brooch is poor, but it looks an Rii).

Other Rii-Riv brooches have been found at Nijmegen, three in the area of G. M. Kam's Cemetery ‘S’, which was mostly in use before the Batavian rising of c. 70, though unpublished material carries occasional use into Flavian times (van Buchem, H. J. H., De Fibulae van Nijmegen (1941), nos. 1202–4Google Scholar, pl- 14 nos- 10, 13)-Unfortunately the material from ‘S’ (like that from G. M. Kam's other cemeteries ‘E’ and ‘O’) was all collected by workmen and others when the area–‘S’ is on the west of the fortress, more or less where the Rijksmuseum G. M. Kam now stands–was ‘developed’; and as Prof. J. E. Bogaers has explained to us, there are no grave-inventories and all must be treated merely as stray finds, possibly indeed derived from the second-century occupation of the fortress in some cases. Bogaers himself has suggested that the British brooches were brought back by elements of Legio IX Hispana, in garrison during the earlier 120s (Studien zu den Militärgrenzen Roms (1967), p. 74).Google Scholar We show here as pl. XIIIc one of the ‘S’ cemetery brooches (no. 1202, Kam Mus. Nr. 651) by courtesy of the Rijksmuseum Kam and of the late Dr. M. H. P. den Boesterd. Typical of continental finds of Rii-Riv, it is clearly of a much more advanced date than those discussed here and shown to belong to Neronian or Flavian times. There are in fact rather better brooches with a terminus ante quem of c. 140 from Newstead (Curie, , op. cit., pl. 86, 13–14)Google Scholar stratified at the base of the little rampart surrounding the bath-house of the equites.

^{page 52 note 1} Loc. cit. (p. 41, n. 4): inter alia, ‘there is no evidence that Ri is earlier than Rii’.

^{page 52 note 2} Wheeler, Sir M., The Stanwick Fortifications (1954), p. 50Google Scholar, fig. 15, 2; cf. the section, fig. 10, layer 4c: layer 6, which preceded it, contained a fragment of Dr. 18 which might be Vespasianic (fig. 7, 4). Two second-century Dr. 31 shards are mentioned, ibid. 32 and fig. 7, 7; fig 11, 30 is late third or fourth century.

^{page 52 note 3} Mackreth, D. in Hobley, B., Trans. Birm. Arch. Soc. lxxxiii (1969), 106Google Scholar, fig. 19, 9. A Silchester brooch (Boon, G. C., Roman Silchester (1957), fig. 17Google Scholar, 4 = Silchester, the Roman Town of Calleva (1974), fig. 19Google Scholar, 5) is very close, and together with other southern examples may tell us something ofthe true origins of the enamelled ‘graceful northern’ trumpet variety, e.g. Curie, , op. cit., pl. 85, 11Google Scholar.

^{page 52 note 4} Hull, M. R. in O'Neil, H. E., Trans. Bristol and Glos. Arch. Soc. lxxxi (1962), 37–8Google Scholar, fig. 5.

^{page 52 note 5} Similarly-decorated central mouldings appear on brooches from Corbridge and the Derbyshire caves (Arch. Ael., ser. 3, vii (1911), 181Google Scholar, fig. 11; V.C.H. Derby, i (1905), figs. 35, 39).Google Scholar This ornament occurs on terrets of the first century B.C, cf. Spratling, M., Antiquity, xlvii (1973), 122Google Scholar, fig. 3, no. 8, of which examples from northwest Suffolk (?), Christchurch (Hants), and Gussage All Saints (Dorset) are mentioned. On a brooch-moulding, it may be a reminiscence of La Tène II, cf. Dechelette, , La Tène, 1253Google Scholar, fig. 535, 2 (Dühren, Baden).

^{page 54 note 1} To whom we are also indebted for enabling us to explore the stratigraphical associations through the medium of the records of the Usk excavations of 1968.

^{page 54 note 2} Distribution kindly communicated by Mr. Hull.

^{page 54 note 3} The quotation is from Thomas Sheppard's original account in Hull Museum Puilns. no. 38 (1907), p. 260.Google Scholar The brooches are illustrated in line by S. C. Hawkes, ibid., no. 214 (1963), 27, fig. 2. The Coritanian coins, being uninscribed, would imply a date long before c. 49 for the deposit of the brooches, if indeed all the components of the group belong together. On the coins, cf. Derek Allen, ibid., p. 33; and his Coins of the Coritani (Sylloge of the Coins of the Br. Is., no. 3, 1963), pp. 11Google Scholar, 33 for the Lightcliffe and Honley hoards, which contained inscribed issues, and Roman imperial coins down to Gaius and to Vespasian respectively.

^{page 54 note 4} e.g. trumpet-brooches from Poltross Burn milecastle (before c. 140) and its close match, noted by Collingwood, from Charterhouse-upon-Mendip: Gibson, J. P.et al., Trans. C.W.A.A.S. N.S. xi (1911), 441Google Scholar, no. 2, fig. 20, cf. pl. 2, section E-F, for layering; F.C.H. Somerset, i (1906), facing 338Google Scholar, fig. 92, centre. Cf. also the trumpet-headed Sii class, Antonine, Richardson, K. M., Antiq. Journ. xl (1960), 200–13CrossRefGoogle Scholar, fig. 1.

^{page 55 note 1} Mr. Hull drew our attention to this brooch, and Mrs. S. H. M. Pollard, the excavator, gave us very full details and lent us the brooch for study. The dating is based on the accumulation of sigillata in the ‘midden’, studied by Dr. Grace Simpson for Mrs. Pollard: the latest piece is somewhat earlier than 150. Cf. Pollard, Sheila, Proc. Devon Arch. Soc. xxxii (1974), 78–9, 138–9Google Scholar, fig. 22, 2; Samian, 111 ff.

^{page 55 note 2} Somewhat similar shackles are very well known, good parallel occurs on a trumpet-brooch (Ri) from Elton, of which Mr. Hull sent us a drawing.

^{page 55 note 3} Cf. Aileen, , Fox, Lady, ‘A decorated bronze mirror from Holcombe’, Antiq. Journ. liii (1973), 21–37Google Scholar.

^{page 55 note 4} Ibid. 23.

^{page 55 note 5} Ibid., pls. III-V.

^{page 57 note 1} Webster, G. A. in Wacher, J. S. (ed.), Civitas Capitals ofRoman Britain (1966), p. 35Google Scholar.

^{page 57 note 2} Both Arthur Evans and Parker Brewis emphasized the native character of the Backworth type, see Archaeologia, lv (1897), 181–2Google Scholar, and Arch. Ael. ser. 3, xxi (1924), 173Google Scholar: ‘the brooches are not Roman but British, their decoration being not of classic but of Keltic character’.

^{page 57 note 3} Clifford, E. M., Bagendon, a Belgic Oppidum (1961)Google Scholar, with a report on over 70 brooches by M. R. Hull; end of occupation, ibid., p. 21.

^{page 57 note 4} Information from Mr. Hull. A northern outlier is that from Thirst House Cave, near Buxton: V.C.H. Derby, i (1905), fig. 35Google Scholar, lower right.

^{page 57 note 5} Almgren, O., Studien über nordeurop. Fibelformen (Mannus-Bibliothek 32, 1923), Taf. 4, Type 65Google Scholar; cf. Kovrig, I., Die Haupttypen der kaiserztl. Fibeln in Pannonien (Diss. Pann. ii. 4, 1937)Google Scholar, Group VIII.

^{page 57 note 6} Evans, , loc. cit.Google Scholar; Hawkes, , Antiq. Journ. xx (1940), 495Google Scholar; The Archaeology of Roman Britain (1969), pp. 296–7Google Scholar.

^{page 57 note 7} See Hull, M. R. in Riciborough, v (1968), p. 83Google Scholar, list; pl. 29, nos. 37–9. There is a useful note on the type as represented in the cemeteries of the Unatal, by Marié, Z., Wissenschaftliche Mitteilungen, i A (Sarajevo, 1971), 49–50Google Scholar.

^{page 58 note 1} See below for north Italy; Stradonice, Filip, J., Keltové ve středni Europe [The Celts in Central Europe] (1956), pl. 126Google Scholar, for a series; named by Bren, J., Trišov, a Celtic Oppidum in S. Bohemia (1966), pp. 78–9Google Scholar.

^{page 58 note 2} Cf. the Schüsselfibel type, Krämer, Taf. 23, 4 and 27, 1, Manching, Stradonice.

^{page 58 note 3} P.P.S. xxxi (1965), 241–367Google Scholar.

^{page 58 note 4} Agostinetti, Paola Piana, Documenti per la protostoria della Val d'Ossola (Centro Studi e Documentazione sull'Italia Romana, Monogr. a suppl. degli Atti, i, Milano, 1972)Google Scholar.

^{page 58 note 5} Ibid. 34–5, fig. 10. (Cf. Bianchetti, E., I sepolcreti di Ornavasso (1895), pl. 10, 15Google Scholar, and Moberg, C.-A., Acta Arch. [Copenhagen], xxi (1950), 108Google Scholar, fig. 9, for the same; headloop and tip of foot lost by the time it was drawn for Agostinetti.)

^{page 58 note 6} Ibid. 27–8.

^{page 58 note 7} Boon, G. C., Antiq. Journ. xli (1961), 13–31CrossRefGoogle Scholar.

^{page 59 note 1} Agostinetti, , op. cit., pp. 36–40Google Scholar, fig. 12, 1–2.

^{page 59 note 2} The original publication by Evans, , Archaeologia, lii (1890), 381Google Scholar, fig. 18, misses this detail, but cf. M, B.., Early Iron Age Antiqs. (1925), p. 127Google Scholar, fig. 138.

^{page 59 note 3} Krämer, 128–30.

^{page 59 note 4} This was a point made to one of us (G. C. B.), in connnection with a La Tène I brooch from Silchester, by Mr. A. M. ApSimon.

^{page 59 note 5} Swarling and Deal, Bushe-Fox, J. P., Swarling Report (1925), pl. 12Google Scholar, 3 (one of a dissimilar (?) pair); pl. 15, 16. Arundel, , Hawkes, , Antiq. Journ. XX (1940), 492–5Google Scholar.

^{page 59 note 6} One from Mont Terrible (Jura Bernois), Lerat, L., Cat. des Colls, archéol. de Montbéliard, Les Fibula galio-romaines de Mandeure (1957), pl. 9, no. 164Google Scholar; one from Presles-Saint-Audubert (Aisne), Birchall, A., loc. cit. (above, p. 58, n. 3)Google Scholar, 355, no. 282, ‘one of the few with internal chord’; one from Alizay (Eure), Hawkes, C. F. C. and Dunning, G. C., Arch. Journ. lxxxvii (1930), 197Google Scholar, fig. 11, no. 5–but the middle of the bow, and any mouldings there, missing.

^{page 59 note 7} The Deal brooches may be compared with those from Lički Ribnik (Croatia), all but identical except in the matter of their external chord, apparently datable to c. A.D. 6–9 (Krämer, 131, Taf. 30, 4–5).