² Martyn Owen, who has inspected the fragments, tells us that it is undoubtedly one of the Lincolnshire Limestones from the Inferior Oolite (Middle Jurassic) strata of the Northants-Rutland-Lincolnshire area and identifies it as almost certainly Ancaster Rag (or Weatherbed) from the quarries of the same name near Grantham.

³ During rescue excavations directed by J. Maloney for the Department of Urban Archaeology of the Museum of London. For the bastion see Britannia xi (1980), 379. Information and details from Mr Maloney and Francis Grew.

⁴ For D.M. memoriae perpetuitati, compare the phrase perpetuae securitati in RIB 612 (=JRS lix (1969), 236-7, No. 6) and the parallels there cited.

⁵ Excavations for the Bath Archaeological Trust were directed by Professor B. W. Cunliffe. See Britannia xi (1980), 387-8.

⁶ After conservation in the laboratory of the Institute of Archaeology, Oxford, four of the curses were examined by both of us independently, before producing agreed, but provisional, texts. The drawings of curses Nos. 2, 3 and 4 and the form in which they are published are due to MWCH and that of No. 5 to RSOT. The texts, drawing and commentary of curses No. 6 and 7 are the work of RSOT alone. We would like to acknowledge the help of R. L. Wilkins of the Institute of Archaeology, Oxford, and P. Dorrell and S. Laidlaw of the Institute of Archaeology, London, for supplying photographs of the curses.

⁷ Analysis of this and the other curses by Dr A. M. Pollard of the Oxford Research Laboratory for Archaeology and the History of Art shows that the metal in the present example is 90·5% lead: 9·4% tin (01 % copper). The proportions of metals in the rest are: No. 3 = 74·3% tin: 25·5% lead (03 % copper); No. 4 = 55·0% tin: 45·0% lead; No. 5 = 98·3% lead: 1·6% tin (0·1% copper); No. 6 = 66·1% tin: 33·9% lead and No. 7 = 73·9% lead: 26·0 % tin (0·1 % copper).

⁸ As on the defixio from Bath published in Britannia xii (1981), 372-4, No. 7, only the names of those cursed have been given.

1. The first name Brit(t)uenda occurs in one of the as yet unpublished Bath curses.

3. For Memorinus/a, Kajanto, The Latin Cognomina, cites only three examples of which two come from Roman Germany, C1L xiii 5183 and 8150. The name Memor (61 examples in CIL) is already attested at Bath as the cognomen of the haruspex, L. Marcius.

⁹ The general form of the curse is similar to that of No. 5 below, but neither object stolen, side (a) 1. 3, nor the verb used to describe the theft, side (b) 1. 1, can be determined.

¹⁰ Of the eleven persons mentioned on side (a) of the larger fragment, only two, Victoria and Minervina, have Latin names, those of the rest being Celtic. This contrasts with the eleven names listed on the Bath curse published in Britannia xii (1981), 372-4, No. 8 where the proportion of Celtic to Latin names is reversed. The explanation of this is probably to be found in the social status of those named, rather than any chronological difference, those given in the present tablet being of humble origin (three are described as servi). Side (a):

1-2. PETIO | ROVE TE-, Pet(it)io Rovet(a)e, ‘the Petition of Roveta’. Compare the heading in one of the Uley curses, Commonitorium deo Mercurio a Saturnina muliere (Britannia x (1979), 343, no. 3). The word petitio, is already attested in the epigraphy of Roman Britain, occurring in a commercial context in one of the wooden writing tablets from London (JRS xxi (1931), 247, N. 2). For the interpretation of Roveta as a personal name, it is necessary to assume that the stops in I. 2 are decorative only and, possibly, that an extra stop has been left out after the first two letters, RO(.)VE.TE, cf., among many examples that could be cited, the graffito TA.MII.SV.BV.GVS.FII.CIT. (Britannia v (1974), 468, No. 50). Against this interpretation is the fact that there is another stop definitely separating words, or rather personal names, in 1. 3, and that there are no clear parallels for Roveta as a name; note however, Roveca (masculine), found on Gallic coins and Roveos CIL v, 8758, both included by Holder in his Alt-Celtischer Sprachschatz and also the name Roveri (genitive), Macalister, CIIC No. 343 Brecon.

2-3. Vindocunus: cf. Cunovindus, Britannia ii (1971), 292 No. 14 and Cunovendus, Britannia x (1979), 347, No. 20.

6. Minervina: Kajanto, the Latin Cognomina, shows that names derived from Minervina are all comparatively rare, and the presence of the name at Bath, the city of Sulis Minerva, may not therefore be coincidence. Compare Sulinus son of Brucetus (RIB 105 Cirencester, RIB 151, Bath) and Sulinus son of Maturus (RIB 150, Bath), both of whom have theophoric names.

ussor: for uxor, cf. 1. 8. For the spelling compare J. C. Mann, ‘Spoken Latin in Britain as evidenced in the inscriptions’, (Britannia ii (1971), 218-24, especially 223, -s- for -x-, three examples.

7. Cunitius: cf. Cunittus on the defixio from Leintwardine, JRS lix (1969), 241, No. 31; and Cunitti (genitive) on an Irish Ogham inscription, Macalister CIIC No. 149.

Ser(v)us: cf. 1. 9, 10. For the omission of -v-, see Mann, op. cit., eight examples.

8. Senovara: the first element occurs in a number of Celtic names including Senovirus, CIL xiii 3584 and 5569.

ussor: cf. 1. 6, n.

9. Ser(v)us: cf. 1. 7 and n.

10. Mattonius: derived from the name Matt-o, -onis, see Holder, op. cit., s.v. It occurs as a gentilicium in CIL xiii, 2018, Lyon. For the alternative reading Mationius (taking the secondt as an T with an exaggerated serif at the top), cf. the name Mattio, AE 1897, No. 114.

Ser(v)us: cf. 1. 7 and n.

11. Catinius (rather than Latinius). There are identical names in Latin and Celtic and, accordingly, it is included both by A. Holder, op. cit. and Schulze, Lateinischer Eigennamen. Here it will certainly be Celtic, cf. Cattinius, Macalister CIIC Nos. 153 and 157.

exsactoris: Cf. No. 6 1. 5 exsigatur. The spelling -xs- for -x- is not included in Mann op. cit., but does occur elsewhere in Britain, e.g. in the cognomen of Q. Neratius Proxsimus, JRS lii (1962), 192, No. 8 and frequently in the name Maxsimus as is shown in I. Marriott's Analytical Index of Personal Names in RIB i and CIL xiii.

13. Methianus (rather than Metell(i)anus): The -th- may represent the barred Celtic D, cf. Carathounos and Caraddouna where the letters D are given cross bars, Holder op. cit., s.vv. If so the same name may be identical with Medianius CIL xiii 2895, two other examples only, not necessarily with the same derivation, being cited by Kajanto, op. cit., s.v., from Spain and Mauretania Caesariensis.

side(b):

2-3. These lines are probably garbled since the reading is fairly clear; they may be intended for f(a)cias (V) (san-)|g(u)ine (suo) Min(erva) Su|l(is), cf. facias ilium sanguine suo illud satisfacere on one of the previously published curses from Bath, Britannia xii (1981), 375 ff., No. 8.

4. igienunus. Possibly a garbled attempt for Ingenuinus. For Ingenuinus|a see RIB 123, 358 and 1716(?).

¹¹ The text is complete, but donatur (cf. No. 3 above) is perhaps to be understood.

2-3. destrale: ‘vulgar’ spelling of the late-Latin word dextrale, which also survives in modern Italian as ‘destrale’.

3-4. involaverit: as in Britannia xii (1981), 372, No. 8 (Bath) and JRS xlviii (1958), 150, No. 3 (Kelvedon). This perfect subjunctive in a relative clause, the mood dictated by an indirect command explicit or implied, is often the finale of a curse, e.g. Britannia x (1979), 342, No. 2 (Uley), postulaverit, and No. 7 below, inrogaverit.

¹² I. Docca: CIL xiii 10010.794 (followed by TLL) takes ‘Docca’ as an abbreviation of ‘Doccalus’, both forms occurring on stamped samian; the present text enables one to separate them.

2 f. dono… exsigatur: for the formula of ‘giving’ stolen property to a god who is required to ‘exact’ it, cf. Britannia x (1979), 343, No. 3 (Uley), deo… donat ita ut exsigat (incidentally the same ‘vulgar’ spelling, cf. exsactoris, No. 41. 11 and n. 10 ad loc.

2. numini tuo: cf. Britannia xii (1981), 375, No. 8 (Bath), with n. 21 ad loc.

3. [a]misi: cf. Britannia x (1979), 343, No. 3 (Uley), amisit. (denarios): the usual sign, a variant of which is found in JRS liii (1963), 122 (Ratcliffe-on-Soar). Other sums of money lost at Bath are expressed in what is probably fourth-century currency, argentiolos (Britannia xii (1981), 371, No. 6) and argenteos (No. 7 below).

4. [earn involaveri]t: cf. Britannia xii (1981), 372, No. 8, and No. 5 above with note 11 ad loc.

4-5. si ser[vu]s s[ix liber] etc. This type of formula is peculiar to British curse tablets, where it is common; cf. esp. Britannia xii (1981), 372, No. 8, si vir si femina si servus si liber.

¹³ The casual reference to Christianity, perhaps the most interesting feature of this tablet, can hardly be earlier than the fourth century in Britain. Gentilis in the sense of ‘pagan’ and argenteus absolutely as ‘silver piece’ (not quasi-adjectivally ‘silver’, of coins) are late-Latin usages. The letter-forms are also late, in particular the E formed by two hooked strokes, not found before the third century, and the A, B, M, N and P, which are all closer to modern forms than to those of ‘classical’ cursive (e.g. No. 6 above).

¹⁴ The writer has failed to reverse the an of Candidina. Seu gens seu Chistianus requires correction, and quaecumque is probably an error for quicumque. Other difficulties in the text may be due to transmission errors. (See further, next note.) The first tablet ever found at Bath (RIB 154) was also written with a reversed text.

¹⁵ Lines are numbered according to the transposition, not as transcribed.

1-2. seu gen(tili)s seu Ch(r)istianus: the expansion is certain since a contrasted pair is implied both by seu… seu and by analogy with the three utrum… utrum pairs which follow. Gentilis is the usual late Roman Christian term for ‘pagan’ (see A. Blaise, Dictionnaire Latin-Francais des Auteurs Chretiens, s.v.), gens being so used only once where demanded by metre; it was not, of course, used by non-Christians to describe themselves, but it would be unwise to conclude that the writer, though a believer in Sulis, was also a Christian, since the word gentilis is being used here as the formulaic alternative to Christianus (‘whether Christian or not’). The fact that both words have been misspelt suggests that the formula was unfamiliar; it is certainly unique in curse tablets, and is also the first occurrence of the word Christianus in the epigraphy of Roman Britain. The three utrum… utrum pairs are typical of British curse tablets, but not found elsewhere, and it is tempting to imagine that the novel gentilis|Christianus pair was added as a tribute to the universal power of Sulis by someone who knew not only the traditional formulas, but also the words of St Paul: ‘… there is neither bond nor free, there is neither male nor female: for ye are all one in Christ Jesus’ (Gal. 3.28).

2. quaecumque: would imply that the thief is known to be female, whereas the neutral quicumque is required by the vir|mulier, puer}puella formulas, and by the qui in 9.

4-5. annianomantutene: the tablet is damaged (see FIG. 32), but the reading is fairly certain. Since the unmistakable anni yields no sense in isolation, it seems best to take this as the beginning of a personal name (cf. Anniola, Britannia xii (1981), 372, No. 6, with n. 17 ad loc.) in apposition to mihi, for which the o in the next line provides a dative ending. Thus Annian|o, word division like this between lines occurring elsewhere in the text, mantutene is then best seen perhaps as a misspelling of the rare matutine (‘in the morning’) which, Priscian notes (lnst. ed. Hertz II, 137, 7), is derived from mane.

5. bursa: in classical Latin used only in its primary sense of ‘ox-hide’, but in medieval Latin the sense of ‘purse’ dominates; the present text proves that the latter sense was also current in the Roman period.

s(e)x argente[o]s: cf. Britannia xii (1981), 371, No. 6, argentiolos sex, with n. 17 ad loc.

6. tu, d[o]mina dea: cf. RIB 323 (Caerleon), which invokes dom(i)na Nemesis. Sulis is invoked as, by implication, by facias in Britannia xii (1981), 375, No. 9, with n. 22 ad loc.

6-7. perexi[g]e…: the verb perexigo is not attested, but is plausible as an intensified form of exigo (for which see the note to line 2 of No. 6 above). It is uncertain whether to restore the imperative perexige (cf. dona in 8), or perexigeris (for perexegeris) which better suits the trace of the next letter and the space available.

7-8. indepreg[.]stum: the end of 7 is corroded and the beginning of 8 is worn (see FIG. 32), so that the reading is difficult. An object of dederit is required, but it is uncertain whether to read something like indeprehensum (‘undiscovered’) or to separate inde (‘thence’) from an obscure technicality like pr(a)egestum (‘previous action’), pr(d)egustum (‘foretaste’), etc.

8-9. The syntax is obscure and the text (if correctly transcribed) probably corrupt. The purpose is apparently apotropaic (to make any counter-spell by the thief rebound upon him?), and eputes possibly conceals a verb equivalent to solvat (sanguinem suum); for where ‘his blood’ occurs in British curses it is usually with the idea that the culprit should pay it (cf. Britannia, xii (1981), 375, No. 9; JRS xlviii (1958), 150, No. 3). Another possibility is to restore (m)e putes by haplography, but the sense remains elusive.

9. inrogaverit: for this usage see note to No. 5 above s.v. involaverit.

10 f. Side (b) is a list of eighteen personal names, the first eight in pairs, the suspects presumably of the theft on Side (a). They are the usual mixture (e.g. Britannia xii (1981), 372, No. 7) of ‘Roman’ (or Greek-derived) names and ‘Celtic’ names. The ‘Roman’ (Latin) names, with one exception, are well-attested cognomina (not that this distinction mattered by the fourth century), some of them derived from a simpler cognomen.

10. Postum[ianu]s: Postum[inu]s is also possible; derived from Postumus (e.g. RIB 1651).

Pisso: apparently unattested, but probably ‘Celtic’ rather than a version of Piso.

11. Locinna: apparently unattested, but cf. Locirnus/Logirnus (CIL xiii 10010.1152).

Alauna: what survives of the first two letters resembles the -ul- of ‘Gunsula’, but a and u are often similar in this hand. ‘Alauna’ is a common place-name, and plausible as a personal name: see A. L. F. Rivet and C. Smith, The Place-Names of Roman Britain (1979), 243.

12. Materna: already attested in Britain (RIB 1181).

Gunsula: apparently unattested, but possibly derived, like Sulinus, from Sulis.

13. Candidina: written Cnadidina in error; derived from Candidus, which is common in Britain, cf. Candidianus (Britannia xii (1981), 372, No. 7).

Euticius: properly Eutychius (from the Greek), but cf. Euticianus (Britannia iii (1972), 352, No. 2, Cirencester).

14. Peregrinus: already attested at Bath (RIB 140).

15. Latinus: likewise (RIB 158).

16. Senicianus: likewise (Britannia xii (1981), 371), No. 6) and, in the same spelling (not Senecianus), on the Lydney curse (RIB 306).

17. Avitianus: derived from a cognomen found in Britain (e.g. RIB 188), and common elsewhere.

18. Victor: one of the most common names in Roman Britain.

19. Scu[tri]us: the surviving traces (see FIG. 34) suit this restoration, even though it is a nomen, and not a common one (e.g. CIL vi 220, Rome).

20. Aessicunia: apparently unattested, but perhaps formed from the war-god Aesus (like the place-name Aesica) and *cunos (dog).

21. Paltucca: also apparently unattested, although Tucca is known.

22. Calliopis: transliteration of a Greek personal name also found as Calliopus (late-Roman and rare, according to TLL Onomasticon).

23. Celerianus: derived from Celer, which is common in Britain.

¹⁶ During excavations by the Colchester Archaeological Unit directed by Philip Crummy. Information and drawing supplied by Nina Crummy.

¹⁷ Found by Timothy Sapwell together with a tear-shaped lead tag 50 by 22 mm which had been pierced by a nail. We are most grateful to Mr Sapwell for providing full information on the context of his find and for lending it to us, and to Cathy Giangrande of the Institute of Archaeology, London for her work in unrolling and conserving the tablet. It remains the property of Mr Sapwell.

¹⁸ 1. There is no obvious restoration of the name, Kajanto, The Latin Cognomina gives only Nasennianus.

2. eve(h)it. For the omission of -h- see J. C. Mann, ‘Spoken Latin in Britain as Evidenced in the Inscriptions’, Britannia ii (1971), 221 (eight examples). The alternative reading eve(h)itur is attractive but an active verb seems to be required. The name of the thief is very uncertain. It may begin with the element *brocco-, badger, cf. JRS xliii (1953), 131, No. 21, Vrocata. It is curious to find the thief named when the phrase si mascel, sifemina, si puer, sipuella is included (lines 6-7). For another curse where a similar formula is used, although the thief may be named, see Britannia xii (1981), 372 ff., No. 8 (Bath), especially p. 374 n. on lines 10-11.

3. armi[lla]|s cf. l. 10.

4. 11. cape(t)olare. The T(hesaurus) L(inguae) L(atinae) sv. 2 cites Isidore, Origines 19, chapter 31 de ornamentis capitis feminarum: diadema, nimbum, capitulum et mitra. In Section 3 he defines capitulum as quod vulgo capitulare dicunt. Mann, op. cit, gives no examples of the omission of -t- but it is left out in the first two occurrences of the word et in l. 10.

4. spectr[um. The reading is fairly certain and, if correct, may be used for speculum, ‘mirror’, but we can find no examples recording the use of the word with this meaning in late or medieval Latin. Alternatively the letters might conceal a synonym for amictus mentioned below, l. 10.

5. cufia(m). See TLL sv, cofia. The word coif is derived from it. For the spelling cufia, see ed. G. Goetz, Corpus Glossariorum Latinorum, 5, 584, 8, where the word is given, along with six others, including tiara and mitra as a synonym for cidaris. Note that the final m has been omitted as often on Romano-British inscriptions. See Mann, op. cit., 222 (twenty examples) and note on (i)llu(m), l. 9.

6-7. Similar phrases occur on eight British curses, the words si mascel sifemina being exactly paralleled on the defixio from Ratcliffe-on-Soar, Notts., JRS liii (1963), 123, side A, lines 8-9, and the whole phrase on an unpublished curse found in the estuary of the river Hamble, Hants, cf. No. 7 above, side (a), transposition lines 2-3.

8. ocrisa, for ocrias by metathesis cf. l. 5 for a previous occurrence of the word.

8-9. vulleris for volueris. For the change of person from the second to the third, requerat, compare the Bath curse (Britannia xii (1981), 375 No. 9).

8 and 11. sanguine (suo) occurs on several British curses including that from the Hamble estuary cited in the note to 1. 6.7.

9. (i)llu(m). The initial i omitted by haplography. For the omission of the final m see note on cufia(m), 1. 5. requerat. For the use of -e- for -i- see Mann, op. cit. above, 16 examples.

In Neptu(nu)s, the -p- replaces -u-. Neptune is addressed on unpublished curses from the Hamble estuary and Brandon, Suffolk. It is possible that the god was invoked as the deity of the river waters near which the curse was found.

12. Carta s(upra) s(cripta). For the 55 abbreviation compare the abbreviation 55 dictas, ss dictum, ss dicto and s(uper) s(cript)ta(e) on one of the Uley curses. Britannia x (1979), 343, No. 3 and s(uper) s(crip)to on the curse from Ratcliffe-on-Soar, JRS liii (1963), 123 side B, 1. 5. For the whole phrase compare the Bath curse, Britannia xii (1981), 370-2, No. 6, 1. 6 read as carta picta perscri[pta, where the second word should perhaps be read dicta; ‘what was spoken, has been copied onto the sheet’.

¹⁹ For the treasure see Britannia xii (1981), 347-8, and 389-93 (the inscribed objects). See also below p. 421 under Addenda. We are gratefulto A. K. Gregory for making the object availableto us for study.

²⁰ Excavations for the Lincoln Archaeological Trust directed by K. Camidge. For the excavations in 1980 see Britannia xii (1981), 336. Full information from the director of the Trust, M. J. Jones.

²¹ l. 2 presumably to be restored as VIX(IT) A]NN(OS) followed by a numeral at the beginning of line 3. In ll. 3-5 it is tempting to restore PILI|[O INNO]CEN|[TISSIMO, the words applying to the deceased named in the preceding lines, but it is then hard to explain the plural masculine names in lines 5 and 7 where one would expect one only, that of the father. Hence the interpretation adopted here with fili(i) qualifying personal names which follow. In l. 4 CEN is probably part of the name Crescenti(a)nus, Innocenti(a)nus or Placenti(a)nus. Censorinus seems less likely as it is then hard to explain the lacuna at the beginning of the line, probably too short for another complete name. ll. 6-7 – possibly MINER|[VALI]VS, although the name Minervalis is much commoner than its longer cognates.

²² During excavations for the Department of the Environment directed by Frank and Nancy Ball who supplied full information.

²³ For the omission of -u- in eques, see RIB 108 and 109 and in equitis RIB 1064 and Britannia ix (1978), 479, No. 49, where it is perhaps used as a personal name, possibly as here.

²⁴ During excavation for Carlisle City Council and the Department of the Environment directed by Mr M. R. McCarthy. Mr I. Caruana provided a photograph and other details.

²⁵ Most likely to be a dedication either to Victoria (Aug., etc.) or by a legionary vex(illatio).

²⁶ The ring was found by Mr Eric Compton, together with two Roman coins and other miscellaneous objects. It was subsequently declared treasure trove and acquired by the British Museum. We are most grateful to Catherine Johns of the Department of Prehistoric and Romano-British Antiquities for providing full information about the find. See her note in Antiq. Journ. lxi (1981), pt. 2, 343-5.

²⁷ Catherine Johns draws our attention to two parallels, both from Corbridge, one inscribed TIOAEMIOY OIATPON and the other Aemilia zeses, sees D. Charlesworth, ‘Roman Jewellery in Northumberland and Durham’, AA 4th. ser. xxxix (1961), 1-36, Finger Rings Nos. 6 and 7. She dates rings of this type to the second or third century.

²⁸ During excavation for Exeter City Museums Archaeological Field Unit directed by Mr C. G. Henderson. Mr P. T. Bidwell made this and the next two items available to RSOT, and provided information.

²⁹ The third stroke of A is vertical. The dotted x includes a third stroke linking the other two strokes.

³⁰ During excavation for Exeter City Museums Archaeological Field Unit directed by Mr C. G. Henderson.

³¹ The third stroke of a is vertical. Possible restorations include [Cant]abri (cf. RIB 1568) and [L]ab(e)ri (cf. Britannia vii (1976), 381, No. 10).

³² Excavations by the Wimborne Archaeological Group directed by A. G. Giles who supplied details and a photograph.

³³ Information and rubbings from Nick Wickenden.

³⁴ CIL iii, 11383 and EE ii p. 434 and cf. Britannia xi (1980), 413-4, No. 53 and n. 73.

³⁵ Found with the following two items during excavations for the Department of the Environment, Essex County Council and Braintree District Council, directed by B. R. G. Turner who supplied details and drawings.

³⁶ Or possibly ABES, abes, ‘you are not here’. For the final letter of this retrograde graffito to be an s, it would be necessary to assume that it had not been reversed, unlike the two preceding letters.

³⁷ During excavation (see JRS lviii (1968), 198) for Birmingham University Department of Extra-Mural Studies directed by Dr G. Webster, who made it available to RSOT. The other inscribed object found was a building stone marked FIRMINI (ibid., 206, No. 5), but it seems to have had no connection with this item.

³⁸ By Mrs V. Smith in whose possession the intaglio remains. Information and photograph from M. J. Watkins of the City Museum and Art Gallery, Gloucester. For parallels to the intaglio Martin Henig draws our attention to P. Steiner, Xanten (Frankfurt a. M., 1911) p. 135, No. 174, from the colonia, and No. 175 from the fortress site.

³⁹ During excavations for Gorhambury Excavation Committee supported by the Department of the Environment, directed by D. S. Neal who supplied full details.

⁴⁰ Belatonus is the only cognomen to be suggested by 1. Marriott's analytical index of personal names in CIL xiii and RIB i. It occurs on two inscriptions, both from Reims, CIL xiii 3283 and 3284. the nomen could be either lanuarius, or the cognate Ianuarinius, which appear from the index of CIL xiii to have been equally common.

⁴¹ During excavations for the Canterbury Archaeological Trust under the overall direction of T. Tatton-Brown. For the position of this, and other sites in Canterbury mentioned below, also excavated under Mr Tatton-Brown's overall direction, see the map reproduced in Britannia xii (1981), 367. Full information on the Canterbury material was provided by Pan Garrard, Marian Green and Nigel Macpherson Grant. It has been arranged in the order referred to in n. 1 except for graffiti on coarse pottery, which have been arranged initially by site within the town.

⁴² For another example see CIL vii 1250 = RCHM, London, 176, No. 57, and C. Roach Smith, Illustrations of Roman London (1859), 114 (with figure). The design is not included in A. W. G. Lowther, A Study of the Patterns on Roman Flue-Tiles and their Distribution (Research Papers of the Surrey Archaeological Society No. 1) or in Johnston, D. and Williams, D., ‘Relief patterned Tiles – a Reappraisal’ in McWhirr, A. (ed.), Roman Brick and Tile, BAR International Series 68 (Oxford, 1979), 375–93Google Scholar.

⁴³ This combination of letters do not occur in I. Marriott's Analytical Index of Personal Names. It might conceivably be part of the rare Greek name Naumachus (IG xii 8 No. 277, 1. 119, Thasos) cf. Naumachios, fourth century A.D., PW, S.V.

⁴⁴ The only name to be suggested by I. Marriott's Analytical Index of Personal Names is Minervalis, but the first letter seems definitely not to be an N.

⁴⁵ Probably to be restored as Vitalis or Natalis, but there are other possibilities.

⁴⁶ The first letter is very uncertain. The second letter could be a P but the DD combination is more common in Celtic, cf. the river Addua (Adda) in Gallia Cisalpina. Note also Aduatuca (Atuatuca), Tongres and the tribal name Aduatuci (Atuatuci).

⁴⁷ For the site see S. S. Frere, P. Bennett and S. Stowe, Archaeology of Canterbury vol. ii: Excavations on the Roman and Medieval Defences of Canterbury (1982), 83 with figs. 38-40.

⁴⁸ There are several possibilities: Saxa is an attested Latin name, though omitted by Kajanto in the Latin Cognomina, while Holder includes Saxanus (and variants) as Celtic in his Alt-celtischer Sprachschatz. A third possibility is that it could be the name Saxo (‘Saxon’) or cognates. For names derived from ethnics in the N.W. provinces see Kajanto, op. cit., and for examples from Britain cf. Remus (Britannia iii (1972), 356, No. 24) and Tungra (Britannia xii (1981), 388, No. 51.

⁴⁹ Excavations by the Leicestershire Museums Archaeological Field Unit with the support of the Department of the Environment, directed by Jean Mellor, and J. N. Lucas. For the site see Britannia vii (1976), 327-8. Full information, rubbings and drawings from R. J. Buckley.

⁶⁰ Excavations by the Leicestershire Museums Archaeological Field Unit, with the support of the Department of the Environment, directed by R. J. Buckley who supplied details of the find, rubbings and a drawing.

⁵¹ Excavations by the Museum of London, Department of Urban Archaeology, with the support of the Department of the Environment, directed by Gerald Clewley. Information and photograph from Michael Rhodes of the Department of Urban Archaeology. Mr Rhodes draws our attention to a close parallel in technique, a second ring, also from London, in the Museum (Accession No. IV-1-395, EE iv 716 = RCHM London (1928), 176 No. 49. Like the present example it is also of iron, decorated with a strip of copper alloy and the words VITA VOLO. The apparently early date of the New Fresh Wharf ring makes the possible Christian significance of the ring seem less likely.

⁵² During excavations conducted for the Museum of London's Department of Urban Archaeology by G. Milne. Information from F. Grew of the Department.

⁵³ Excavations directed by John Schofield, for the Museum of London's Department of Urban Archaeology. Information, drawings and photographs from M. Rhodes of the Department.

⁵⁴ Excavations directed by Peter James for the Museum of London's Department of Urban Archaeology. Information from F. Grew of the Department.

⁵⁵ The stamp of Vitalis (i) of La Graufesenque, die 6j c. A.D. 50-65, identification by Brenda Dickinson.

⁵⁶ During excavations directed by Gus Milne for the Museum of London's Department of Urban Archaeology. Information from F. Grew of the Department.

⁵⁷ During excavations directed by Jenny Norton for the Museum of London's Department of Urban Archaeology. Information from F. Grew of the Department. For the site see Britannia xii (1981), 351.

⁵⁸ For examples of the name from Britain see ibid. 389, No. 61 and No. 86.

⁵⁹ Excavations directed by G. Milne for the Museum of London's Department of Urban Archaeology. For the site see above p. 373. Information from F. Grew of the Department.

⁶⁰ We are grateful to Professor E. Rodrigues-Almeida for reading this inscription and the following two items for us. For the rare name Acerronius, previously unattested on painted inscriptions on amphorae, Prof. Almeida draws our attention to one of the consules ordinarii for A.D. 37, Cn. Acerronius Proculus.

⁶¹ See previous note.

⁶² Both METT and MAT occur as painted inscriptions on amphorae from Rome (CIL xv 4505 and 4624).

⁶³ See n. 59 above.

⁶⁴ Professor Almeida tentatively suggests s(ocii) or s(ocietas) as an expansion of the first letter.

⁶⁵ Excavations directed by J. Maloney for the Museum of London’s Department of Urban Archaeology. Information from Jo Groves of the Museum.

⁶⁶ Excavations by Sarnia Butcher for the Ministry of Works. For the site see the Trans. London Middlesex Arch. Soc. xxxiii forthcoming. Information from G. Parnell.

⁶⁷ Excavations directed by G. Milne for the Museum of London's Department of Urban Archaeology. For the site see above p. 374. Information from F. Grew of the Department.

⁶⁸ Excavations directed by Louise Miller for the Museum of London's Department of Urban Archaeology. For the site see Britannia xi (1980), 380, and xii (1981), 351. Information from F. Grew of the Department.

⁶⁹ Excavations directed by John Schofield for the Museum of London's Department of Urban Archaeology. For the site see Britannia vii (1976), 347. Information from Michael Rhodes of the Department.

⁷⁰ The graffito perhaps recorded the owner's nomen in the nominative followed by his cognomen, one of the rare cognates of the common Asper, as Asprianus, Asprinus or Asprio. Alternatively, and perhaps more likely, -us may be the ending of a word for vessel, such as caccabus (a deep segmental bowl), catinus (a shallow bowl), or urceus (a jar), followed by the name Asper in the genitive. The names of other types of vessel ending in -us are less likely in view of the form of the pot, see Hilgers, W., Lateinische Gefdssnamen, Beiheft der Bonner Jahrbiicher, 31 (1969)Google Scholar, Taf. 1 which shows various pot shapes with the appropriate Latin names.

⁷¹ Information from A. K. Gregory who submitted the object for inspection.

⁷² It is tempting to expand the stamp as C(ohors I) B(…) fec(it). However, one would probably expect the numeral to be included, cf. C(ohors) I Aq(uitanorum) on a lead sealing from Leicester, Britannia vii (1976), 386 No. 36, and tile stamps from Reculver C(ohors) I B(aetasiorum). The expansion C(lassis B(ritannica) fec(it) also seems unlikely, since on tile stamps at least, this form of abbreviation is, with one exception, not found, CL BR being the norm. See Brodribb, G. ‘A further survey of stamped tiles of the Classis Britannica’ in Sussex Arch. Collect, cxviii (1980), 183–96Google Scholar; with B. Philp The Excavation of the Roman Fortsofthe Classis Britannica at Dover 1970-1977 (1981), 129, Tile Stamp Bl = fig. 28 No. 6, the sole die to read C(iassis) B(ritannica).

⁷³ By metal detector. Miss L. Allason-Jones provided RSOT with a drawing and other information.

⁷⁴ During excavation for the Joint Management Committee of the Department of the Environment and the Vindolanda Trust directed by Mr P. T. Bidwell, who made this and the next four items available. Other graffiti were examined, but are too fragmentary for inclusion here.

⁷⁵ Only the tail of the first letter survives; its slope suggests R, but M is possible.

⁷⁶ The stamp was read by Mr B. R. Hartley, who dates the bowl to the late second or early third century. The name Talio is already attested in Britain as a foreign bronzesmith's stamp (EE vii 1168, cf. CIL iii 1640, 4); for the genitive case, see CIL iii 5473 (Noricum).

⁷⁷ IV resembles N, but the sequence NM can be excluded. Possible restorations include[c]ivium and [conv]ivium (in the sense of ‘dining club’, cf. CIL iii 14242, 14250 convivium veteranorum).

⁷⁸ Information from Dr Graham Webster who directed excavations by the Wroxeter Training School. See above p. 358.

⁷⁹ Taking Q as the initial letter of the nomen rather than the abbreviated praenomen Quintus. On quite a number of oculists’ stamps the nomen is similarly abbreviated, with or without the praenomen, cf. CIL xiii part 3.2, p. 602-3 (index of nomina on stamps).

⁸⁰ penecille, a type of medicament perhaps applied with a sponge, penicillum. For a discussion of the precise meaning of the word see Nielsen, H., Ancient Ophthalmological Agents (Odense U.P., 1974), 83–8Google ScholarPubMed. The word ad regularly introduces the disease that the ointment was intended to cure; it is included, as here, without the disease being mentioned on a number of stamps, AE 34. 119541. 85; 45. 14 (twice). In fact medicaments described as penicille were most often used for an ailment described as lippitudo, see CIL xiii, part 3.2 p. 606, index of collyria. Lippitudo is inflammation or watering of the eye. cf. H. Nielsen, op. cit. p. 91.

⁸¹ Excavation for Warwickshire County Council directed by J. Magilton and M. Carver. For the site see Britannia x (1979), 300, and xi (1980), 369-70. Information and drawing from Mr Magilton.

⁸² By Atherstone Archaeological Society while foundations were being excavated. It was made available to RSOT by Mr K. Scott, and will go to Warwickshire Museum.

⁸³ The graffito is too damaged for the reading to be certain. the name Belalucius is apparently unattested; the initial Bel- is common in Celtic personal names, and Alucius or Al(l)uquius occurs in the Iberian peninsula (see Holder s.v. Alluquius).

⁸⁴ With the next two items by the late Mr K. Croft Andrews. A squeeze was provided by Mr L. P. Wenham, who also made the next two items available. All three are now in the Richmondshire Museum, Richmond.

⁸⁵ The die appears to be Wright 44 (Britannia vii (1976), 228), already found at Chesterholm, Corbridge and Ebchester.

⁸⁶ Two attempts to cross thet were made, the second also crossing the following 1; of the L only the curving top survives. The personal name Martiola is already attested in Britain (RIB 933) and six times in CIL xiii.

⁸⁷ It is now in Doncaster Museum. Dr P. C. Buckland provided a drawing and other details.

⁸⁸ During excavations conducted by D. Zienkiewicz for the Ancient Monuments Branch of the Welsh Office, who provided a drawing and details of this and the following four items.

⁸⁹ The suggestion of Mr G. C. Boon who cites as a parallel the small disc bearing the numeral xiii found in the excavations of the Caerleon amphitheatre and possibly an admission ticket.

⁹⁰ For a full discussion of the strigil see Boon, G. C. in Antiq. Journ. lx (1980), pt. ii, 333–7Google Scholar.

⁹¹ Mr Boon shows that the six labours and the order in which they are here depicted, curiously, do not correspond either to the first, or to the last six of the labours (in an ordering that was established in the second century A.D.), but to the central six adventures. the missing strigil will then have had the first three labours on one side of the handle and the last three on the other. This anomaly can be explained by a consideration of the way in which the strigils will have been viewed - together and as a pair, first with the handles with one side uppermost, to show the first six labours, and then turned over to show the other sides and the final six labours. Among several parallels, both in Latin and Greek, to the inscription on the butt of the handle and that on its missing pair, the most striking are the two inscriptions on mosaic panels at the Baths at Anemur on the southern coast of Turkey which read καλω∈ λoνσαι ‘have a nice wash’ and καλω∈ ελóνσoν ‘you did wash well’, Russell, J., Anatolian Studies xxiv (1974), 95–102CrossRefGoogle Scholar.

⁹² For similar representation of Diana Venatrix on gemstones from Britain see M. Henig, A Corpus of Roman Engraved Gemstones from British Sites, BAR 8 (ii), Oxford 1974, Nos. 254-6, Gloucester, Chesters and Caernavon.

⁹³ Compare the suggested reading of the letters on the side of the counter reported under the following item. On another counter from the Fortress Baths site, JRS lvi (1966), 220, No. 12, interpreted as ILLAI|ILL(AI)VII, the inscriptions could be interpreted as ILLIV(S), ‘his’, and a blundered version of the same word.

⁹⁴ For the reading compare the preceding note.

⁹⁵ George Boon suggests that the first letter R, is the end of an abbreviated nomen.

⁹⁶ By thermoluminescence tests carried out at the Oxford University Research Laboratory for Archaeology and the History of Art. Information from G. C. Boon.

⁹⁷ So also W. J. Rodwell, ‘Some Romano-Saxon Pottery from Essex’, Antiq. Journ. 1, pt. ii (1970), 262-76.

⁹⁸ Information from Christopher Going.

⁹⁹ We are grateful to Martin Henig for discussing the gem and its significance with us. For a full discussion (by Mary Whiting) see the forthcoming publication on the Thetford treasure by T. W. Potter and C. M. Johns. The epithet Sabaoth ‘of hosts’, (found in the Hebrew phrase for (‘Lord God) of Hosts’, as on the gold amulet from Caernavon, RIB 436, lines 1-3), is often attached to Iao, the divinity mentioned on the front of the Thetford gem. For its use with Abraxas, as here, Dr Henig draws our attention to a very close parallel in a gem from Xanten, A. Krug, ‘Antike Gemmen in römisch-germanischen Museum, Köln,’ BRGK 61 (1980), 182, No. 47. For the most recent discussion of the god, by M. le Glay, see ‘Abraxas’ in Lexicon Iconographicum Mythologicae Classicae I, (Zurich-Munich (1981), 2-7).

¹⁰⁰ Information from Dr P. C. Buckland.