¹ Thanks go to Professor Donovan and to Dr M.L.K. Curtis and Mr R. Clark of the Geology Department at Bristol City Museum for examining the stone (Donovan in litt. 20 Sept. 1985). Professor Donovan notes the possibility of a source further east, about Nunney, or on the main outcrop north of Mendip, about Radstock; but ‘the nearer to Bath, the more likely (one would think) that they would have used Bath stone.’ It cannot be excluded that a Dundry source supplied the stone, just across the valley of the Chew, for Dundry Hill is also Inferior Oolite; but no bed of such coarse stone is known there, and Dundry freestone is very different in appearance (Donovan in litt. 15 Jan., 18 March 1988).

² Miss Jennifer Stewart, Keeper of Archaeology at Bristol City Museum, enabled me to inspect the sculpture on several occasions. The photographs and excellent drawings by Miss Ann Linge were contributed by the Museum through the good offices of Mr Nicholas Thomas. N.B. – A wrong figure of overall length has been given in the past.

³ Rahtz, P. and Harris, L.G., Proc. Somerset Arch. & N. H. Soc. ci/cii (1956–1957), 15–51.Google Scholar A retrospective paper, adding the results of further excavation around the well, by Professor P.A. Rahtz, with a reappraisal of the famous glass vessel by Professor V. Evison, is forthcoming in Arch. Journ.

⁴ E.M. Jope, ibid, xcvi (1951), 137–42.

⁵ Ralegh Radford, C.A., Med. Arch. ii (1959), 111.Google Scholar For examples of heraldic figures see e.g. Garner, T. and Stratton, A., Domestic Architecture during the Tudor Period (1911), pl. cxlvGoogle Scholar, etc. – I am grateful to Dr Radford for discussing the Pagans Hill dog with me (in litt. 10 Sept. 1982, 28 Nov. 1987) and in the latter signifying his agreement with the Roman dating of the sculpture. This paper has otherwise benefited from his observations.

⁶ Lewis, M.J.T., Temples in Roman Britain (Cambridge, 1966), 45–6Google Scholar, omitting the Lydney goddess (R.E.M., and Wheeler, T.V., Lydney Report (1932), pl. xxiv)Google Scholar, perhaps a water-goddess (cf. Schauerte, G. in Matronen u. verwandte Gottheiten (1987), 61–2).Google Scholar

⁷ Reinach, S.. Répertoire de la statuaire grecque et romaine ii.2 (Paris, 1909), 759.Google Scholar nos. 2, 5; Vatican piece, also in J.M.C. Toynbee, Animals in Roman Life and Art (1973), fig. 44. Cf. Waywell, G.B.. The Lever & Hope Sculptures Monumenta Artis Romanae xvi (Berlin, 1986), 90.Google Scholar

⁸ The tail is an important diagnostic feature. When looped over the haunch as e.g. in the marble fragment from Maiden Castle (CSIR Gt. Britain i, 2 no. 98) one may be sure that a dog is not in question, but a lion or panther.

⁹ CSIR Gt. Britain i, 2 no. 100. This relief has suffered since discovery: see the first photograph reproduced by Haverfield in Roman Britain in 1913 (Oxford, 1914), 49, fig. 20. or in his Romanization of Roman Britain (Oxford, 1923 ed.), fig. 26. The piece shown on the right of the photograph (in Bristol City Museum with the main fragments but not included in CSIR) should not have been placed so close, leaving too little room for the leg. See further, notes 57, 83.

¹⁰ CSIR Gt. Britain i, 5 no. 60.

¹¹ Perfahl, J., Wiedersehen mit Argos (Mainz, 1983)Google Scholar , colour cover-picture.

¹² Cunliffe, B. and Davenport, P., The Temple of Sulis-Minerva at Bath (Oxford, 1985) i.1, pl. lxxii.Google Scholar The dogs accompanying the Meleager and Diana statues in the Vatican have similar bound collars (Amelung, W., Die Sculpturen des Vat. Mus. i (1903)Google Scholar , Taf.15, no. 108; ii (1908), Taf.12, 10). In general cf. Toynbee, J.M.C., Latomus xxxv (1976), 272–3.Google Scholar Stephen Bird of the Roman Baths Museum kindly sent photographs.

¹³ I have met no other example of a jewelled necklace on a dog, but glittering and gem-set necklaces are mentioned by Ovid (Met. x, 113) and Calpurnius Siculus (Eel. vi, 41–5) in connexion with stags, and by Virgil (Aen. vii, 278) and Statius (Theb. ix, 689) in connexion with horses.

¹⁴ de Ridder, A., Cat. sommaire des bijoux antiques (1924), pl.xGoogle Scholar , no. 503.

¹⁵ Marshall, F.H., B.M. Catalogue Jewellery, Greek, Etruscan and Roman (1911), pls. lvi–lxiGoogle Scholarpassim.

¹⁶ Greifenhagen, A., Schmuckarbeiten in Edelmetall, i (1970), 73Google Scholar , Taf. 54; Farbtaf. 6, 4.

¹⁷ Hettner, F., Rhein. Mus. xxxvi (1881), 446–7Google Scholar , note 2 (detail omitted by von Massow, W., Die Grabmäler von Neumagen (Berlin, 1932), i, 216).Google Scholar

¹⁸ Cod. Theod. xvi, 10.2, 4 (cf. 6).

¹⁹ My list in Rahtz, , Proc. Somerset Arch. & N. H. Soc. xcvi (1951), 134Google Scholar, no. 60. A valuable map of traces of the fourth-century pagan revival appears in C. Thomas, Christianity in Roman Britain (1981), fig. 48.

²⁰ Cod. Theod. xvi, 10.2, 7. Cf. Frend, W.H.C., Journ. Brit. Arch. Assn.³ xviii (1955), 1–2Google Scholar, who, however, seems to err in placing closure as early as 341.

²¹ e.g. the restorations of the Temple of Diana at Caerleon c. AD. 250, of a locus religiosus at Bath, 3rd century (?), and of a column and statue at Cirencester set up in the time of the prisca religio (RIB i, 316, 152 and 103). Archaeologically, besides Lydney, the main Caerwent temple of c. A.D. 335⁺ was restored after 364 (Britannia xvi (1985), 260; xviii (1987), 308). Repairs to statues are in ILS ii, i passim; archaeologically, note repairs to the Silchester Tutela and eagle (Boon, , Britannia iv (1973), 108Google Scholar, note 11, and idem, Silchester: the Roman Town of Calleva (Newton Abbot, 1974), 119–20).

²² Muthmann, F., Statuenstützen und dekoratives Beiwerk an griech. und röm. Bildwerken (Heidelberg, 1950)Google Scholar , Taff. xiii, xvi, xxi etc. for struts.

²³ Henig, M. in Rodwell, W. (ed.), Temples, Churches and Religion BAR 77 (Oxford, 1980), 321–3Google Scholar , fig. 15, 5.

²⁴ Hammerson, M.J. in The London Archaeologist iii.8 (1978), 206–12Google Scholar ; Merrifield, R. in Henig, M. and King, A. (eds.), Pagan Gods and Shrines of the Roman Empire (Oxford, 1986), 85–92.Google Scholar I am grateful to Mr Hammerson for his help in finding the photographs reproduced here, and likewise Mrs Jenny Hall of the Museum of London. This sculpture was either part of a larger group or was set in the coping of some sort of low wall or, perhaps, pool; it was intended to be seen from all sides. There are grooves for cramps on either side of the plinth.

²⁵ Colouring of architectural details in Britain, Blagg, T.C.F., Britannia vii (1976), 171.Google Scholar There was no window-glass at Pagans Hill, and none was found at Brean Down where the cella-windows were thought, on the basis of a reconstructed arch, to have been very large (though this may have been only a relieving-arch over a smaller embrasure), cf. ApSimon, A.M., Proc. Univ. Bristol Spelaeol. Soc. x.3 (1965), 201.Google Scholar Glass may have been scavenged in sub-Roman times for re-melting to make beads, etc. Window-glass occurred at Nettleton but is not mentioned in connexion with the ambulatory embrasures, of which the sills remained; cf. Wedlake, W.J., The Shrine of Apollo at Nettleton, Wilts. (1982), 46–8, 153.Google Scholar

²⁶ Anderson, F.W. in Rahtz, P.A., The Saxon and Medieval Palaces at Cheddar BAR 65 (1979), 226Google Scholar; Colchester, L.S. (ed.), Wells Cathedral (Shepton Mallet, 1982), 18, 89Google Scholar; significantly similar date at Doulting church itself, Donovan, D.T. and Reid, R.D., Proc. Somerset Arch. & N. H. Soc. cvii (1962–1963), 60–71.Google Scholar

²⁷ Doulting stone is not mentioned by Williams, J.H. in his article on Roman building-materials, Trans. Bristol & Glos. Arch. Soc. xc (1971), 95–119.Google Scholar

²⁸ Margary, I.D., Roman Roads in Britain (1976 ed.), 125–7, 103Google Scholar , 139–40.

²⁹ Ward-Perkins, J.B., Proc. Brit. Acad. lvii (1973), 137–58Google Scholarpassim; Dworakowska, A., Quarries in Roman Provinces (Wroclaw, 1983), 56–9, 81.Google Scholar

³⁰ Hope, W.H. St. J., Archaeologia lx (1907), 162–3Google Scholar, figs. 2–3; Boon, op. cit. (note 21), 186.

³¹ Boon, , Britannia iv (1973), 107–14CrossRefGoogle Scholar ; op. cit. (note 21), 116–7, 166.

³² Kähler, H., Röm.-germ. Forsch. xiii (1939), 27Google Scholar, 32; Bath, , JRS xlv (1955), pl. xxiiiGoogle Scholar; Caerwent, CSIR i, 5, no. 82.

³³ RIB i, 149, 151, cf. 105.

³⁴ Ward-Perkins, loc. cit. (note 29), 146–7.

³⁵ RIB i, 69–71, with Boon, op. cit. (note 21), 154, 172.

³⁶ RIB i, 309.

³⁷ Wright, R.P., Britannia xvi (1985), 248–9CrossRefGoogle Scholar , for an improved reading.

³⁸ AÉ 1908, no. 11; CIL xi, 4123 shows that offerings might go towards the adornment of the temple – bronze doors, etc.

³⁹ RIB i, 153 and 320 (Bath and Caerleon).

⁴⁰ Many in ILS record the donor but not the circumstances.

⁴¹ BL Add. MS 33, 717, f. 153^v.

⁴² Haverfield, F. in VCH Som. i (1906), 309.Google Scholar Mr S. Minnitt of Somerset County Museum helped in connexion with this accession and gave me a copy of Dr Anne Robertson's list of the coins. Unfortunately ‘cleaning’ at some distant date has made it impossible to distinguish between the component pieces on grounds of surface appearance, as may often be done with contaminated deposits.

⁴³ Tomlin, R.S.O., Britannia xv (1984), 336Google Scholar and notes at 351, n.17; and further in litt. 14 Nov. 1987 and 3 Jan. 1988.

⁴⁴ My list in Rahtz and Harris, loc. cit. (note 3), 33 no. 97, redated in accordance with modern views.

⁴⁵ The curious run of denarii might be explained if they were merely the residue of a very large hoard, all the better and more interesting pieces having been abstracted, and replaced by various modern casts and the single semis, which was probably unrecognised.

⁴⁶ P.A. Rahtz and E. Greenfield, Excavations at Chew Valley Lake (1977), passim.

⁴⁷ Boon, G.C. and Rahtz, P.A., Arch. Journ. cxxii (1966), 28Google Scholar , 34.

⁴⁸ Tomlin in litt. 14 Nov. 1987.

⁴⁹ Turner, E.G., JRS xlvi (1956), 116.Google Scholar – His rather elaborate suggestion that the land mentioned in the document may have been outside Britain because there was mention of the jus Italicum, and because the tablet was of larch, a foreign wood (many are) is superfluous.

⁵⁰ Rahtz and Greenfield, op. cit. (note 46), 64, and Boon, ibid., 296.

⁵¹ Tomlin, loc. cit. (note 43), 351 note 19, on line 3.

⁵² RIB i, 306; the ring, Boon, op. cit. (note 21), 133 with fig. 18, 4; cf. Toynbee, J.M.C., Journ. Brit. Arch. Assn.³ xvi (1953), 20–1.Google Scholar

⁵³ Growth of inflation in the 3rd century, cf. Carson, R.A.G., Rev. num. (1965), 225–35.CrossRefGoogle Scholar Papyrus, Lafaurie, J., Rev. num. (1975), 80.Google Scholar – Burnett's, A. account in Coinage in the Roman World (1987), 108–31Google Scholar provides a good background.

⁵⁴ Robertson, A.S. in Casey, J. and Reece, R., Coins and the Archaeologist (2nd ed. 1988), 18Google Scholar ; of some 1,600 hoards, not many more than 120 contained over 1,000 coins, only about 1 in 12.

⁵⁵ ILS 4908 (lex), 5440–1.

⁵⁶ Wright, R.P., JRS lii (1962), 191Google Scholar and note 8.

⁵⁷ J.M.C. Toynbee in Wedlake, op. cit. (note 25), 135–8. R. Merrifield (loc. cit. (note 24), 91) suggests an Apollo citharoedus with long robe ‘as on the mosaic at Littlecote’, but the legs are clad in a woman's stola with palla over: for Apollo draped, see LIMC ii, nos. 135, 145, 477, etc. Diana as huntress is unlikely to be in question either, for (in the west at least) her dress, like Miss Lindsay's in the song, was generally kilted up to the knee (ibid. 849). The separate fragment (note 9) may show the end of a torch held by the left hand, unusual (but cf. ibid. fig. 338), though that Diana has the short gown. See further note 83.

⁵⁸ Tomlin, , Britannia x (1971), 342Google Scholar; Mercury, Henig, loc. cit. (note 23) and idem, Religion in Roman Britain (1984), 57–8, figs. 18, 69; idem in Ellison, A., Excavations at West Hill, Uley, 1977 (Bristol, 1978), 37Google Scholar, figs. 3–4, and cf. ibid., 40 for reference to votive spears.

⁵⁹ R.E.M., and Wheeler, T.V., Lydney Report (1932), 41.Google Scholar

⁶⁰ Votive bronzes and coins down to Arcadius, from a temple (?), found near Abergele (Clwyd) and to be published by R.J. Brewer, include several dog-figurines, two similar to the Woodeaton piece, Henig, op. cit. (note 58), 65, fig. 22.

⁶¹ Wheeler, op. cit. (note 59), 87, no. 113a-b, fig. 21; 90, nos. 123–4 and 126, pls. xxvii-viii. Cf. Bathurst, W.H. and King, C.W., Roman Antiquities at Lydney Park, Glos. (1879), 40, pl. xiii.Google Scholar – At Biglis near Barry (S. Glam.), a lozengular brooch of British make shows what may again be this figure (Webster, J. in Robinson, D.M. (ed.), Biglis, Caldicot and Llandough BAR 188 (1988), 53Google Scholar no. 7, fig. 21). The site, on low hill, overlooks the Severn Estuary. – On the piscatorial side of Nodens, cf. William of Malmesbury, Gesta Pontificum Anglorum, Rolls ed. p. 292: fluvius Sabrina…quo nullus antepiscosior. Several Severnside churches are dedicated to fisherman-saints, as DrCottle, Basil points out, Trans. Bristol & Glos. Arch. Soc. cvi (1988), 6Google Scholar, cf. J.G. Wood, ibid. lviii (1936), 212.

⁶² O'Rahilly, T.F., Early Irish History and Mythology (Dublin, 1946), 495–6Google Scholar, note 2; he could not accept Tolkien's Germanic derivation (Wheeler, op. cit. (note 59), 132–7).

⁶³ Nennius, Historia Brittonum (ed. Lot, 1934), cap. 68, refers to the ‘two kings of Severn’ (duo rig Hafren), and this must mean the parting of the bore by sandbanks and the subsequent tumultuous coming-together of two waves: cf. Rowbotham, F.W., The Severn Bore (Newton Abbot, 1970)Google Scholar and remember William of Malmesbury's description, loc. cit. (note 61). The connexion with Nodens was a typical touch of SirRichmond's, Ian (Roman Britain (1955), 139).Google Scholar No bore occurs below Sharpness, except very rarely, and its size is inconsiderable until the river narrows at Frampton. The effect is lunar, and grandest at equinox; the lunar effect was well-understood in the case of the Garonne, cf. Sidonius Apollinaris, Carm. xxii, 105–13. Cf. too Buckland, F., Log Book of a Fisherman and Zoologist (1891), 287–309Google Scholar: at 288 he describes the plight of a boat coming in front of the bore or ‘… wheels of the chariot of the water king, who was racing up with his foaming steeds:’ this without reference either to Bathurst and King's Lydney or any apparent acquaintance with the discoveries at Lydney. At 287 he states ‘the roar was like nothing I had ever heard before, and never before did I so fully understand the…expression ‘the voice of many waters’.

⁶⁴ The only other British trace of Nodens, in that case Mars-Nodens, is a pair of bronze statuettes, since lost. found before 1718 at Cockersand (Cockerham) Moss south-west of Lancaster (RIB i, 616–7). Rivers flowing into Morecambe Bay narrow rapidly and so provide conditions favourable for a bore to develop. The Lune itself is weired, and no bore is known; the Kent, on the north side of the bay, does have a small bore (I am grateful to Professor J.H. Johnson, of Lancaster University, for this information). It seems likely that the Cockerham Moss statuettes reflect a worship of the deity parallel with, and similar to, that at Lydney, being based on tidal phenomena.

⁶⁵ Lydney dog, Wheeler, op. cit. (note 59), 88, no. 114, pl. xxv; others, poorer, Leyden and Cologne. Dog in mundus, 89, no. 120, pl. xxvi. Orifice, 28, pl. xixa, cf. Bathurst and King, op. cit. (note 61), 21, pl. vii, where cubes should have been drawn in. – Both Richmond, I.A. and Crawford, O.G.S., Archaeologia xciii (1949), 40Google Scholar, and Rivet, A.L.F. and Smith, C., The Place-Names of Roman Britain (1979), 424Google Scholar, note that the ‘Metambala’ of Ravennas (ibid., 207) is a Nemet– name; but the latter dismiss the former's notion that –ambala = ‘navel’. They restore Nemetobala, ‘grove-hill’, which would suit Lydney; but perhaps the mundus was the navel in question. The place must be in west Glos, or in Mon.

⁶⁶ Wheeler, op. cit. (note 59), 89, no. 119, pl. xxvi, without comment; Bathurst and King wrongly call it a ‘Bacchic panther’. The attitude is curiously reminiscent of two figures widely scattered in time – the ‘sorcier des Trois Frères’ and an ithyphallic demon at Abson church, south Glos. (Dobson, D.P., Man, Jan. 1930, 10–11CrossRefGoogle Scholar, fig.). On the chthonic quality of dogs, cf. Gourevitch, D., Mél. École franç, de Rome lxxx (1968), 270–1Google Scholar, in an important paper on ‘le chien thérapeutique’. Cf. also Jenkins, F., Latomus xvi (1957), 66Google Scholar, 68.

⁶⁷ In general see McMullen, R., Paganism in the Roman Empire (New Haven & London, 1981), 32–3Google Scholar, etc.; medieval times, Finucane, R.C., Miracles and Pilgrims (1977), esp. 130–51.Google Scholar

⁶⁸ RIB i, 306; for the others see now Tomlin, R. in Cunliffe, B., The Temple of Sulis Minerva at Bath, ii: The Finds from the Sacred Spring (Oxford, 1988), 64–7Google Scholarpassim.

⁶⁹ Dehn, W., Germania xxxv (1941), 104–11.Google Scholar Excellent, in brief, with plans and notes of other curative centres including Epidaurus, Krug, A., Heilkunst und Heilkult, Medezin in der Antike (Munich, 1985), 176–7.Google Scholar See also Thévenot, E., Divinités et sanctuaires de la Gaule (Paris, 1968), 107–12.Google Scholar

⁷⁰ Wheeler, op. cit. (note 59), 61, etc. A whole class of bent bone bangles was omitted (as also by Bathurst and King). Specific votives include the female figurine in bone (ibid., 89, no. 122, pl. xxvi) with umbilical hernia, and the model arm in bronze with koilonychiae (spoon-nails) indicative of gross iron-deficiency (readily remedied at Lydney with its ferrugineous marl and waters) (ibid., 121, pl. xxvi, see Hart, G.D., Bull. Hist. Medicine xliv (1970), 76–9Google Scholar, a paper drawn to my attention by the Dowager Lady Bledisloe, kind hostess on many visits ten years ago). Magic, note the leaden strip, later re-used, inscribed with ABCDEFG|HIK|L… (Collingwood, ibid., 101 no. 5), paralleled in Britain at Bath (Tomlin, loc. cit. (note 68), 108 no. 1); for the apotropaic, magical and oracular powers of the alphabet, cf. Merlat, P., Rép. des inscriptions et monuments figurés de Jupiter Dolichenus (Paris & Rennes, 1951), 102–3Google Scholar, for refs.

⁷¹ Collingwood, in Wheeler, op. cit. (note 59), 102 no. 6; Espérandieu, CIL xiii, 3.2, no. 101; J. Voinot, Inventaire des cachets d'oculistes gallo-romains (Conférences lyonnaises d'Ophtalmologie, no. 150, 1981/2, Annonay [1984]), no. 107.

⁷² E. Espérandieu, Recueil général des bas-reliefs etc. de la Gaule romaine, no. 2067. In the main I follow here Thévenot's view of this piece in Latomus ix (1950), 426, and op. cit. (note 69), 67–8. Benoit, F. (Latomus xii (1953), 77–84)Google Scholar offers a metaphysical explanation which does not, perhaps, take sufficient account of ordinary pagan religiosity. Hatt, J.-J., ‘Apollon guérisseur’ (in Pelletier, A. (ed.), La médecine en Gaule (Paris, 1985), 216–7)Google Scholar, is wrong in saying that the seated personage is a god; he would not have needed medicaments to cure. The intermediacy of a sanctified human agent is typical: also in a case of ophthalmia, see the healing abbot in Bib. Troyes MS 2277 (Ariès, P. and Duby, G., Hist. de la vie privée (Paris, 1985), ii, 63Google Scholar, fig.).

⁷³ Espérandieu, no. 4465. Cataract rightly doubted by Feugère, M., Künzl, E. and Weisser, U.. Jahrb. d. Röm.-Germ. Zentralmuseums xxxii (1985), 479Google Scholar, cf. Taf.67: the patient would not be standing. – Collyria in ancient times were sticks of medicament to be rubbed down in wine, water, milk, etc. and applied as prescribed. The Lydney stamp (note 71) refers to a honey medium, to drops, and to a tincture to be applied by brush. Unfortunately it does not name the preparations in a manner which would enable their effectiveness to be judged.

⁷⁴ Caesar, De bello gallico vi, 17. Cf. Jullian, C., Hist. de la Gaule (Paris, 1920–1926) vi, 35–6Google Scholar, 62–4.

⁷⁵ Scholz, H., Der Hund in der griech.-röm. Magie und Religion (Berlin, 1937), 46–8Google Scholar, 51, 53. Apollodorus, Bib. iii, 10, 3.10 for Aesculapius' education by Chiron the Centaur in the arts of medicine, surgery and cynegetic; cf. Xenophon, Cyneg. i, 1.1–6.

⁷⁶ Reinach, S., Rev. arch.³ iv (1884), esp. 77–8.Google Scholar Dog-licking occurs in the folk-medicine of many lands; cf. e.g. H. Gaidoz, ibid., 218 for the rural French saying ‘langue de chien | sert de médecin’.

⁷⁷ See a letter from Sir Francis Avery Jones, The Times, 20 Dec. 1986.

⁷⁸ Aelian, De natura animalium xi, 3 and 20, referring to temples in Sicily.–‘Drunk', see an interesting detail in Pausanias, Descr. Graec. ii, 27.3.

⁷⁹ Radford in Rahtz, loc. cit. (note 19), 123–6.

⁸⁰ Wedlake, op. cit. (note 25), 37 fig. 20, etc.; Wheeler, op. cit. (note 59), 26, fig. 3.

⁸¹ On Mars see Jullian, op. cit. (note 74), 28–9, 32–3; there was in Gaul a strong link between Mercury and Mars, who seem between them to express the nature of Teutates, who was god of the peoples. Cf. Thévenot, op. cit. (note 69), 62, etc.; Lambrechts, P., Contrib. à l'ét. des divinités celtiques (Bruges, 1942), 146.Google Scholar

⁸² Jullian, op. cit. (note 74), 35–6, makes this distinction between Neptune and Apollo, whose sphere was the warm waters. Inland Neptunes: CSIR Gt. Britain i, 6, no. 88 for a Housesteads fountain with reclining Neptune and nymphs; at Frampton and other villas, beside which a stream ran, Neptune appears on mosaics, cf. Smith, D.J. in Rivet, A.L.F. (ed.), The Roman Villa in Britain (1969), 83Google Scholar; Frampton also in CIL vii, 2.

⁸³ Wedlake, op. cit. (note 25), 54–7, fig. 34 and pl. iiia-b. Little is recorded about the excavation of Building I. The relief was found when a limekiln was built on the site. It appears to have occupied a position in the centre of the back wall of the temple, really a revetment of the natural rockface. Cf. Grey, G.J., Som. Arch. & N. H. Soc., Proc. Bath & Dist. Branch (1909–1913), 177–8Google Scholar, and pl.; the position of the sculpture, however, depends on the findings of excavation in 1939. On the sculpture see notes 9, 57. If indeed of Diana, for this goddess and waters cf. Jullian, op. cit. (note 74), 40, citing CIL xiii, 1495, a massy bronze ring devoted by Dianenses, found at Vichy.

⁸⁴ See note 69.

⁸⁵ Thevenot, op. cit. (note 69), 207–8. Albius = Phoebus, and Phoebus ‘was den reinen und den Reiniger bedeutet’ (Fehrle, E., Die kultische Keuschheit im Altertum (Giessen, 1910Google Scholar, repr. Berlin, 1966), 224–5). Apollo-epithets and names, Le Roux, F., Ogam xi (1959), 216–26Google Scholar; J.-J. Hatt, loc. cit. (note 72), 209–12.

⁸⁶ Cf. note 82; and Jullian, op. cit. (note 74), 40.

⁸⁷ CSIR Gt. Britain i, 2, no. 3, cf. Cunliffe, , Roman Bath (1969), pl. lxiv.Google Scholar

⁸⁸ The only known details are in litt. D. Darwish to Professor P.A. Rahtz, 2 Aug. 1984: cf. Rahtz forthcoming. Deities with Phrygian caps, Toynbee, J.M.C., Britannia xi (1981), 2Google Scholar, pls. ii-iii. Maponus is another, cf. Merrifleld, loc. cit. (note 24), 90, now CSIR Gt. Britain i, 6, no. 2.

⁸⁹ Attis, Henig, op. cit. (note 58), 113, but cf. Hepding, H., Attis: seine Mythen und sein Kult (Giessen, 1903Google Scholar, repr. Berlin, 1973), 100–1. Röscher Ausfürl. Lexicon der gr. u. röm. Mythologie ii.2, 2868, suggests confusion with Adonis.

⁹⁰ Merrifield, loc. cit. (note 24), figs. 4–5; RCHM Roman London (1928), pls. 12, 1 and 14, 4; Toynbee, J.M.C., Art in Roman Britain (1962), pl. 79Google Scholar, better in Goodburn, R., The Roman Villa, Chedworth (Nat. Trust, 1972), pl. 9.Google Scholar

⁹¹ CSIR Gt. Britain 1,2 no. 99. The animal is manifestly a stag, but contrary to the first reports (Poulett Scrope, G., A Hist. of Castle Combe (1852), 6Google Scholar; Gent. Mag. 1860-ii, 159), the relief does not represent a figure spearing the stag. The attitude is as a Diana striding to right, holding a bow (rather straight, probably seen front-on, as in the Bevis Marks statue, but now weathered extremely) and drawing an arrow from a quiver at the shoulder, as Toynbee saw, Jocelyn (Art in Britain under the Romans (Oxford, 1964), 162).Google Scholar What gave the appearance of a spear was the very heavy fold of the cloak across the right breast. The left hand emerges from the same fold to hold the bow low across the thighs, not in line with the upper folds of the cloak. No dog; otherwise very like the Chedworth deity. The findspot is probably correct in VCH Wilts, i.i (1957), 55, as far as record goes; doubt remains, the coins said to have been found with the sculpture about 1826 being ‘not a hoard’ and mixed with extraneous material, as Dr Paul Robinson kindly informs me (in litt. 16 June 1988). The sculpture, now in the Ashmolean Museum (where Mr A. McGregor kindly showed it), is one of the ‘two’ reliefs stated by Haverfield (Romanization, 73) almost without qualification to have been found at Nettleton.

⁹² Koethe, H. (23. Ber. d. Röm.-Germ. Komm., 1933 (1934), 16)Google Scholar observed that where there was evidence of dedication, Mercury and Apollo are the most frequent in the round and polygonal temples; but added that the figures merely reflected the general popularity of those deities.

⁹³ Paris, R., Rev. arch. de l'Est xi (1960), 169.Google Scholar

⁹⁴ Gildas, De excidio Britanniae i, 4: patriae portento ipsa diabolica … quorum nonnulla lineamentis adhuc deformibus intra vel extra deserta moenia solito more rigentia, torvis vultibus intuemur.

⁹⁵ Registrum Joh. de Pontissara Ep. Wintonensis (ed. Deedes, C., Canterbury & York Soc. xix, 1915), 238Google Scholar: set nec lapides, ligna, arbores sive fontes propter somnium aliquod venerentur ut sancta, quoniam ex hujusmodi animabus fidelium multa pericula credimus evenisse. This admonition was neither unique nor novel: Gildas, for instance, goes on from the passage quoted (note 94) to refer to the veneration of mountains, hills, rivers and springs.