² By Mr N. Chappie, from whom it was bought by the Museum of London, where Christine Jones made it available to RSOT.

³ This reversal is known from Britain only in a lost curse tablet from Norfolk and (in cursive) in Britannia xv (1984), 337, No. 4 (Bath, Tab. Sulis 61).

⁴ This can be deduced from the position of DA TVS (twice), which was clearly written from the R. margin, and from the way that some rightward-tending letters crowd the letter to the right, e.g. CI in SILVICO- and AI in VASIA-. Various reversals of letter-sequence are found among the Bath tablets; similar to side (b) are Britannia xv (1984), 334–5, No. 2 (Tab. Sulis 44), ibid., No. 4 (Tab. Sulis 61), and below, No. 3 (Tab. Sulis 62).

⁵ Commentary, side (a)

1. tibi: dative, a solecism for the accusative (te) which is correctly used in 7.

1ff. rogo u(t) m(e) vendicas: the letters omitted suggest careless copying, like the omission of T in 8 and the corrections on side (b). For the formula Cf. Tab. Sulis 35, rogo sanctissimam maiestatem tuam ut vindices, and Audollent Defixionum Tabellae 122 (Emerita), per tuam maiestatem te rogo oro obsecro uti vindices quot mihi furti factum est.

Metunus: nominative for vocative, like Mercurius in Britannia iv (1973), 325, No. 3 (Old Harlow), the vocative being obsolete in Vulgar Latin. A deity is clearly being addressed, but no god Metunus is known. The obscure priapic god Mutunus can be excluded, likewise the little-attested Matunus (RIB 1265). A ‘Vulgar’ pronunciation of Neptunus is probable, the initial nasal being confused (m for n) and the labial being dropped before a dental, as in Italian Nettuno (‘Neptune’); British Celtic had no -pt- and the same change is found in Latin loan-words taken into Celtic (e.g. captivus > Welsh ceithiw, French chétif. see Jackson, K.H.. Language and History in Early Britain (1953), 394).Google ScholarDominus Neptunus is invoked in unpublished curse tablets from Brandon, Suffolk, and the Hamble estuary. The discovery of this tablet in the Thames near the site of the Roman bridge supports the identification: the god may have had a shrine there, as on the Tyne bridge at Newcastle (Cf. RIB 1319).

2. vendicas: this ‘Vulgar’ spelling (for Classical vindicas) may be the first instance of a change found in some, but not all, Romance languages: Cf. Italian vengiare, French venger, Spanish vengar, but Sardinian vindicare, Portuguese vingar (see W. Meyer-Lübke, Romanisches Etymologisches Wörterbuch (1968), s.v. vindicare).

3. de iste numene: iste (nominative for ablative) is for isto, by attraction from numene, knowledge of the case-endings being weakened in Vulgar Latin. Numene (unstressed í>ě is typically ‘Vulgar’) is probably a confusion for Classical nomine (‘name’): vindicare de (ab is more usual) means ‘to avenge (oneself) on someone’, so numen (‘divinity’) is inappropriate, unlike nomen, Cf. Britannia xii (1981), 370–2, No. 6 (Tab. Sulis 8), a nominibus infrascriptis deae exactura est. This confusion seems to occur in Tab. Sulis 102, numen furti for nomen furis, and may have been easy for British Latin-speakers: Cf. the variant transcriptions of a Celtic god's name, Nodenti (RIB 306) and Nudente (RIB 307), and the shift in the pronunciation of Londinium (see A.L.F. Rivet and Colin Smith, The Place-Names of Roman Britain (1979), s.v.).

5. ante q(u)od: the omission of V (as in 10) is ‘Vulgar’, the result of the same reduction of the sound [kw] to [k] before a back vowel that produced com[o](do) for quomodo in RIB 154 (Taft. Sulis 4). The conjunction is a ‘Vulgarism’ for Classical antequam, attested here for the first time in the Roman period (see O. Prinz (ed.), Mittellateinisches Wörterbuch (1967), s.v. ante).

6. ven(iant): conventionally restored as present subjunctive (of a subordinate clause within an indirect command), but ven(ient) (future) or ven(erint) (future perfect or perfect subjunctive) are also possible. The same omission is made in 11, and may have been due to the author's uncertainty about tenses and verb-endings, but more likely by haplography with the ANT of ante and by optical confusion between VEN()DIE() and VENDICAS.

6–7. die(s) novem: the omission of the final consonant (repeated in more extreme form in 11) is ‘Vulgar’. For the formula, Cf. AE 1929, 228 (Carnuntum curse tablet), infra dies nove(m), and No. 3 below (Tab. Sulis 62), ante dies novem.

8. (t)u: T omitted by haplography with VT.

9. mi: correctly spelt in 4, a ‘Vulgarism’, Cf. simper for semper (RIB 2311). The first stroke of the M was omitted, for no apparent reason.

11. n[o]vem: the O was written as D by mistake. Commentary, side (b)

The text seems to have been inscribed in the sequence of the transcription, anti-clockwise round the tablet, as would be natural when writing from right to left. 1–5 was certainly written before 8–12, since the second A of Vasianus (10) overlies the Toi datus (5). 6 was written after 1–5, since it does not begin at the edge of the tablet; a line was drawn to distinguish P from the end of 1–5. 8–12 was written after 7, since its right-hand margin takes account of 7. The text, except for 5, perhaps 6, and 12, seems to be a list of personal names ‘given’ (datus) for punishment to the deity.

1. xuparanti: probably a mis-spelling of the personal name Exsuperantius, common in Christian (and thus late) texts (see I. Kajanto, The Latin (Zognomina (1965), 277). It is not clear whether the nominative or genitive was intended.

2. silvielesatavile: presumably one or two garbled personal names. The first might be Silvia or more likely its rare diminutive Silviola (Kajanto, 168; Britannia xii (1981), 392, No. 90), since that would leave satavile, which might be a diminutive of the common Celtic name SattolSatta, i.e. Sat(t)avilla. The intermediate name Sanava is known (CIL V 3605), although a closer parallel is Satavacus (Britannia vi (1975), 294). Again, the case of both names is unclear; perhaps genitives in the ‘Vulgar’ spelling –(a)e. 3. xsuparatus: a mis-spelling like that in 1, for the cognate personal name Exsuperatus. It is widely distributed (see A. Mócsy, Nomenclátor); in Britain it is found at Chesterholm (Britannia ii (1971), 301, No. 74) and twice at Caerleon (RIB 369; JRS xlvi (1956), 148, No. 7).

3–4. Silvico/le: a ‘Vulgar’ spelling of Silvicolae, genitive (or dative) of the masculine cognomen Silvicola (attested in CIL xiii 2016 (Lugdunum), Valerius Silvicola, derived from his mother's cognomen Silvina). Perhaps the patronymic of Exsuperatus, ‘son of Silvicola’.

4. Avitus: a common Latin cognomen, already attested in RIB 492; the derived Avitianus is found in Britannia xiii (1982), 404–5, No. 7 (Tab. Sulis 98).

4–5. Melusso: this seems to be a variant spelling of Melissus (found in Britain as Melisus, RIB 619), a personal name of Greek etymology which Holder suggests may also be Celtic. Why the dative(?), ‘to Melussus’, is obscure.

5. datus: Datus is a common Latin cognomen (see Kajanto, 298), but the word's symmetrical position at the end of both groups of personal names suggests that it should be taken as datus (est), ‘is given’. Many of the Bath tablets ‘give’ thieves and stolen property to the goddess (donare is the verb used, but an unpublished tablet from Uley begins Biccus dat Mercurio); Cf. also Audollent 97, data nomina ad inferos.

6. perucitibi: perhaps intended for pervici tibi, ‘I have prevailed upon you’, or possibly concealing a personal name cognate with Pervica (RIB 1747) and Pervinca (RIB 1620).

7. Santinus: for the Latin cognomen Sanctinus, well attested in CIL xiii; in sub-Roman Wales, see Nash-Williams, ECMW No. 83 (correctly spelt). The derived nomen Sanctinius is found at Caerleon (JRS xlvi (1956), 148, No. 7). The ‘Vulgar’ spelling with omission of the velar [k] is common (V. Vaananen, Introduction au Latin Vulgaire (1981), 62); in Britain Cf. defuntus (RIB 369) and the borrowing into Celtic of santus (Jackson, LHEB, 406; discussed further by Hamp, E.P. in Britannia vi (1975), 155–6)Google Scholar.

8. Mag[2–3]etus: the reading is difficult. 8 may begin with A, which is ambiguously placed for 8 or 9, but there seem to be no names known in Amag[…], whereas ‘Celtic’ names beginning Mag[…] are common. The S is oddly placed, and seems to have been inserted to replace another letter; it is simplest to suppose the name ended in -us. There is no obvious restoration.

9. apidimis: assuming the A to belong to 9, the reading is hardly in doubt. It is possible, however, that S belongs with the next word: this would become Santoni, a Gallic tribal name (perhaps used as a cognomen, or an unattested personal name cognate with it), and APIDIMI could be explained as a mis-copying of Apiciani, a Latin cognomen. However, Antonius is so common a Latin name that it is difficult not to read it.

10. Santus: the same ‘Vulgar’ spelling (of sanctus, Cf. RIB 1039, etc.) ^as found in Santinus (above, 7; see note). Vasianus: properly Vassianus, one of several personal names derived from the Celtic ^*uasso- (‘servant’, hence ‘vassal’); Cf. Vassinus (RIB 215), Vassicillus (Britannia xv (1984), 339, No. 7).

11. Varasius: the first S is awkwardly formed and might be G. The name is not attested, unless it is cognate with the nomen Cavarasius (CIL v 3710).

12. datus: see note to 5.

⁶ Excavations for the Department of Greater London Archaeology, Museum of London, directed by D. Beard. Information and drawing (by S.A. MacKenna) supplied by Mike Hammerson.

⁷ The first line could read …]MPI[…, with the last letter theoretically any vowel or consonant with an initial straight member. Although the presence of two preceeding consonants would rule out certain other consonants, the letters E, H,L or R would all be possible. The fact that this inscription is cut on marble invites comparison with the lengthy marble inscription from the site of Winchester Palace, also in Southwark, Britannia xvi (1985), 317–22Google Scholar, No.1. The size and style of lettering, and the space of 28 mm separating the two lines are all similar to those on the heading of the Winchester Palace inscription, although the greatest thickness for the marble sheeting used for that inscription was only 24 mm, compared to 34 mm in the present case. If either line is part of a personal name, the most likely possibilities would be, depending on the reading, Domitius, Comitius, Pompeius, Campester, Simplex or Sempronius or cognates in line 1 and Floridus or cognates in line 2.

⁸ During excavations for the Bath Archaeological Trust (see Cunliffe, B. and Davenport, P., The Temple of Sulis Minerva at Bath, I (1985))Google Scholar directed by Professor B.W. Cunliffe, who made it available to RSOT; RSOT will shortly be publishing all the inscribed texts in The Temple of Sulis Minerva at Bath, II, where this will be Tab. Sulis 62. It is included here as one of the more interesting texts not to have appeared in Britannia.

⁹ The same reversal to achieve a ‘secret’ text is found in Britannia xv (1984), 334–5, No. 2 (Tab. Sulis 44).

Other reversals are also found, e.g. in No. 1 above (London). The scribe worked from a ‘straight’ text as usual, as can be seen from his mistakes:

(i) failure to reverse the ligatured ER of puer (5);

(ii) need to correct letters in 4, 5, 7 and 8, esp. / (4) which was due to writing SI twice;

(iii) space left at the end of lines (i.e. the left margin) had to be filled with letters of exaggerated size. There was probably a space at the (left) end of 9.

¹⁰ Commentary

1. [c.?]eocorotis: apparently the petitioner's name, perhaps cognate with a unique title of Mars, Deo Marti Corotiaco (RIB 213).

perdedi: variant oiperdidi by analogy with the stem dare (dedi), also found in RIB 306 (Lydney). The form of D is uncommon, and is not repeated in 2ff. (donavi, etc.). For the formula, Cf. Britannia xvii (1986), 432, No. 5 (Tab. Sulis 5), Docimedis perdidi(t) manicilia dua; hvAperdedi is probably first-person, since there is no other tendency in this text to omit the final consonant.

2–3. Nothing has been lost in the Y-shaped crack between the two fragments of the tablet, but the main fragment (see FIG. 3) has become distorted; the G of sagum thus fits between A and V, and the IS of [S]ufc is complete.

1–2. la[enam / pa”lleum sagum: Dr J.P. Wild comments that they are all plain rectangular cloaks, whose physical differences are unknown: ‘But socially they have quite different overtones. The sagum is the Gallic native/military cloak par excellence while the pallium is the Greek national equivalent, carrying the aura of intellectual life and the stage into the Roman world. The laena is an old-fashioned term for an Italian cloak (Greek χλαίνα).’ Since the author can hardly have lost three different cloaks, he must be attempting to define the cloak he has lost in a bookish, quasi-legal way which will both impress Sulis and ensure that his cloak is fully identified. Sagum is now attested in the Vindolanda tablets (JRS lxxvi (1986), 121). Pallium (of which palleum is a ‘Vulgar’ variant) is the term used by most petitioners at Bath (five instances, against two caracallae and a mafortium). Like bicycles in Oxford, it was a classic object of theft: Cf. Catullus xxv, 6, cinaede Thalle… remine pallium mihi meum quod involasti. Martial uses pallium and laena indifferently, of the dinner guest who leaves with two cloaks: viii 59, 9–10, lapsa nec a cubito subducere pallia nescit / et tectus laenis saepe duabus abit.

2. paxsam: this word is also found in Tab. Sulis 32, paxsa(m) ba(ln)earem, found absolutely in 63 as balniarem. For its meaning see Wild, J.P., ‘Soft-finished textiles in Roman Britain’, CQ n.s. xvii (1967), 133–5.CrossRefGoogle Scholar It seems to be a ‘Vulgar’ spelling oipexa (‘combed’), for (tunica or vestis)pexa, a soft-finished woollen garment or shirt; in Roman Britain it may have meant no more than tunica, judging by the Middle Welsh loan word peis (‘tunic’), modern Welsh pais (‘petticoat’).

do[navi]: enough survives of the last four letters to guarantee the restoration. The ‘gift’ of stolen property to the deity is a regular formula, e.g. Britannia xii (1981), 370–2, No. 6 (Tab. Sulis 8), deae Suli donavi argentiolos sex quos perdidi.

3. [‥]./[S]uto: the one letter of which trace survives is too far from V to be the initial S of Sulis; it is C, E or S, which excludes tibi, fano, or dea. Sulis could be nominative, vocative or genitive, but nominative is unlikely, since Sulis, who did not have a rostrum (5), so far as we know, can hardly have been the subject of déferai (6).

ut: introduces the terms of the ‘gift’, as in Britannia x (1979), 343, No. 3 (Uley), deo s(upra)dicto tertiam partem donai ita ut exsigat (etc.).

hoc: object of déferai (6), the neuter pronoun being used loosely to refer to the four nouns in 1–2 variously feminine or neuter in gender.

ante dies novem: a ‘magical’ time-limit, as in No. 1 above, ante q(u)od ven(iant) die(s) novem; AE 1929, 228 (Carnuntum), infra dies nove(m). The last two letters in the line (EM) are the wrong size because the scribe miscalculated the space remaining; he was evidently trying to finish the line with the end of a word.

4. ser(v)us: this ‘Vulgar’ spelling is common in the Bath tablets.

si [li]bera si serva: si liber si servus is the most common such pair of mutually exclusive alternatives found in British curse tablets, but this feminine variant is unique; si anelila is found in Britannia xv (1984), 338, No. 5 (Tab. Sulis 52). The final A is exaggerated to fill the space at the end of the line.

5. puer: the scribe failed to reverse ER (the same mistake is found in Britannia xv (1984), 334–5, No. 2 (Tab. Sulis 44)) because he wrote E in its correct place and automatically ligatured R to it, which is why R is misplaced; except perhaps for N1(7) under the O of rostro (5), this is the only ligature in the text, which of course did not admit of ligatures since it was written letter by letter in reverse sequence.

5–6. [i]n rostr[o] s[uo] defera[t]: the restoration is guaranteed by the repetition in 9. Déferai does not mean ‘return’ (of stolen property), strictly speaking, but other compounds oí fero (pertuierit, attulerit) are used in this sense in other British curse tablets. The typical formulas of 4–5 (repeated in 7–8) strongly suggest that a thief is being required to bring (back) stolen property. The problem is in rostro suo: the word means ‘beak’, usually of birds, but occasionally of other animals. (It is also applied to beak-like objects, notably a warship's ram, but none of these seems applicable.) Must we translate the Jackdaw of Rheims to Roman Bath? St Columbanus cursed the raven which stole his gardening glove (Ionas, vita Columbani i 15), but marauding birds can hardly have been a hazard at the baths of Sulis. It is easiest to take rostrum as an abusive reference to the mouth or ‘snout’ of the (human) thief, rostrum being sometimes thus transferred to the human anatomy in colloquial Latin: see Adams, J.N., ‘Anatomical Terms Transferred from Animals to Humans in Latin’, Indogermanische Forschungen lxxxvii (1982), 90–109Google Scholar, esp. 103.

7. caballarem: this word can be restored in Tab. Sulis 49, but otherwise as a substantive is unique. Caballus is the ‘Vulgar’ word for ‘horse’ (Classical equus) which has entered the Romance languages (French cheval, etc.). Caballarius is used as a substantive (‘horseman’), the derived adjective is caballinus (‘of a horse’). Caballaris seems to be attested, as an adjective, only in AE 1906, 138, homo caballaris (‘man on horseback’, contrasted with homo pedester, ‘pedestrian’). But caballarem here is a substantive. Since it is an object of theft, the meaning ‘horseman’ is excluded; the cloaks and tunic of 1–2 suggest a textile context, perhaps a horse blanket like the τάρηςκαβαλλαπικóρ (Latin term lost) of Diocletian's Prices Edict, xix 33. ‘Horse blanket’ is likely to have been used loosely for a rug or blanket used in the baths; Cf. Tab. Sulis 6, which refers to the loss of a stragulum (‘(bed)spread’).

7–8. Formulas repeated from 4–5, but with serva and libera in reverse order; the restoration seems certain.

9. defer[at]: as a verb probably the last word of the text. If, as seems likely, it was followed by a space, this would imply a primary text which was transposed line by line.

¹¹ Work for Cambridge County Council as a Manpower Services Commission scheme directed by Mark Alexander. Information from Alison Taylor who submitted the objects for inspection.

¹² During excavation for the York Archaeological Trust directed by Mr R.L. Kemp. Mr A.D. Hooley made it available and provided full details.

¹³ The obvious parallel is RIB 674, the sarcophagus of the first recorded decurión of the colonia Eboracensis. (Another decurión is named in Britannia i (1970), 308, No. 14, also a sarcophagus front.) The formulation is unclear, but it looks as if the decurión, whose name is lost, was buried by two men, one named […]cus and the other M(arcus) Aurel(ius) Ir[‥]us. (Aurel(ius) is followed by what resembles a small open A, but the text is otherwise well spaced, and such inserted letters – see RIB 1091, 1092 for example – are not found as initial letters; it seems better to take it as a medial point abbreviating Aurel(ius) like that after the preceding M(arcus).) The cognomen Ir[‥]us is unlikely to have been of Latin etymology, and there is no obvious restoration. The superlative in line 3 is likely to have been, not carissimus (which was almost exclusive to wives and daughters), but piissimus or pientissimus, which could have been a conventional description of either the deceased or his mourners; since an adverb (-e), a dative singular (-O), or nominative plural (-i) is likeliest, only one letter has been lost after M; there is space for two before the conjectural F of fecerunt, but the word was evidently generously spaced, judging by -SSIM-.

The use of inhumation, the nomen Aurelius, and the lettering, all suggest at least a third-century date, and thus the text is of no help in dating the colonia Eboracensis. Like RIB 674 it confirms that Eboracum had become the standard spelling (see A.L.F. Rivet and Colin Smith, The Place-Names of Roman Britain (1979), s.v. Eburacum).

¹⁴ During excavation for the Tyne and Wear Museums Service directed by Mr P.T. Bidwell, who made this and the next two items available, and provided full details.

¹⁵ The traces are not enough to identify the goddess concerned. The first letter is a diagonal stroke (A, M or V?), the second might be E, F or H, the third C, G or O.

¹⁶ There are many altars to ‘Vitiris’, all crudely inscribed, and only once is the deity identified with another (RIB 971, Mogons), whereas the epithet sanctus occurs several times. ANSV may therefore be a dyslexic sancto or san(c)to. The name of the dedicator, perhaps abbreviated, is concealed in the last line.

¹⁷ This is the third building stone of the Sixth Legion from South Shields, the others being JRS lii (1962), 193, No. 13 (the same formula of legion and century, not cohort and century as found in ‘centurial stones from Hadrian's Wall) and RIB 1061, which Mr Bidwell notes was originally part of one of the piers surrounding the forecourt of the first stone headquarters building, now dated to c 163. Stylistically an Antonine date is appropriate: Cf. RIB 2209, a Second Legion building stone from the Antonine Wall, the text inscribed within a wreath held by two standing nude winged Genii.

Lines 3 and 4 have been erased with a small pick or point, perhaps the result of casual or wanton damage, since it extends to part of VIC and the centurion's nomen. P(ia) F(idelis) may have been lost from 3. The I of Severi seems to have been ligatured above the R. The distance built, 102 feet, is several times the ‘centurial’ pedaturae recorded from Hadrian's Wall (see RIB Index, p. 107), but in line with the 112-feet lengths of fort wall built at Carvoran (RIB 1818, 1820).

¹⁸ By the cousin of the farmer of Hill Head, Mr J. Reay, who allowed access. Mr P.S. Austen provided information.

¹⁹ Like RIB 1406 (0.28 by 0.13 m), ‘this stone is rather thin in comparison with facing stones in Hadrian's Wall; it may well have come from the Vallum’. The cow byre from which it comes was built on the Vallum. The Vallum was the work of auxiliary units (RIB 1365). This unit must be identified with Coh. I Aelia Classica, first attested in Britain by the Chesters diploma (CIL xvi 93) of A.D. 145/6, and later (in abbreviated form) on a leaden sealing (discussed by DrShotter, D.C.A. in Potter, T.W., Romans in North-West England (1979), 73–4).Google Scholar It is not the Coh. I Classica which is attested in early inscriptions and by A.D. 80 was serving in Lower Germany (AE 1948,56), where another diploma records it in A.D. 158 (M. Roxan, Roman Military Diplomas 1954–1977 (1978), No. 52): a contemporary career-inscription of c. 161 (CIL xiv 5347, Cf. AE 1955, 179) names the Coh. I Aelia Classica, which is clearly a distinct unit. The omission of Aelia by the present stone is a difficulty. It is possible, but seems rather unlikely, that there was a Coh. I Classica in both Britain and Lower Germany in the reign of Hadrian, one of which was subsequently granted the title of Aelia by him (see CIL xiv 5347: the title existed before his death). It is more likely that the (British) Coh. I Aelia Classica drastically abbreviated its name for an ephemeral building record that would soon be swallowed by the turf of the Vallum. This seems to have been done by the only other auxiliary unit recorded from the Vallum, Coh. I Aelia Dacorum, which also omits Aelia from its name (RIB 1365). Professor E. Birley has suggested (in Arch. Ael. 4th ser. xiv (1937), 236–7) that the latter cohort may have received the title Aelia at a later date, but the reading of the other inscriptions which omit it (RIB 991, 1289) is doubtful, and the conventional dating of the dynastic title to the moment of enrolment is probably to be preferred.

The abbreviation of cohortis to CH (not COH or CHO) is rare, being found in RIB 850 and Britannia vi (1975), 288, No. 25 with x (1979), 354, No. 44. The centurion's name is one of the cognomina derived from Augustus: Augustalis, Augustanus and Augustinus are all attested in Britain.

²⁰ By Mr A. Griffiths. Mr P.S. Austen and Mr D. Sherlock pointed it out.

²¹ By the farmer, Mr Dalrymple, in whose possession it remains. Mr P.S. Austen provided information and, with Mr D. Sherlock, provided access.

²² The same centurion is already attested in Britannia iv (1973), 329, No. 8 (west of T 29a), RIB 1572 (sector 34a-35) and 1668 (sector 4115–42). Thejwo latter stones do not record the cohort, and spell his name correctly Ph-. However, the first stone reads: COH > GELLI PILIPPI. The difference of cohort is due to Philippus having been transferred (not promoted), like Flavius Noricus, who is recorded in the ninth cohort (RIB 1664) and the tenth (RIB 1812, JRS 1 (1960), 11(a). The ‘Vulgar’ spelling P(h)ilippi is due to the aspirated consonant not being native to Latin. Greek ø in loan-words was at first transcribed by simple p, but ph became Classical and p a Vulgarism.

²³ During excavation for the Bath Archaeological Trust (see Britannia xiii (1982), 381) directed by Professor B.W. Cunliffe, who made it available to RSOT.

²⁴ Other batch-totals, including ‘1,119 [or 319] (tiles)’ from Cirencester, are collected in A.D. McWhirr (ed.), Roman Brick and Tile (1979), 233-6.

²⁵ During a watching brief by the Cumbrian and Lancashire Archaeological Unit on a trench being dug on the south side of the fort. Dr D.C.A. Shotter sent the details.

²⁶ Excavations by Buckinghamshire County Museum directed by M.E. Farley who supplied details and a photograph.

²⁷ If this is part of a personal name, the most likely restoration would be Sedatus, or a cognate. For a probable example of the same name, also on a tile, see Britannia ii (1971), 303, No. 92 (Holt).

²⁸ Excavations conducted by Professor K. Branigan and Martin Dearne on behalf of HBMC (for the site see above p. 321.

²⁹ Brough has already produced tile stamps reading COH[… and CH[…, CIL vii 1245 (a) and (b). These could be tiles made by Cohors III Bracaraugustanorum whose products have been found at Manchester and Melandra Castle, the next fort west of Brough or, less likely, ones produced by Cohors iv Breucorum at Grimescar, whose tiles occur at Castleshaw and Slack.

³⁰ Excavations for Winchester Archaeology Office directed by Ken Qualmann. Information from Kim Holmes who also supplied a photograph and drawing.

³¹ Information and drawing from Pamela Irving. She suggests that the sherd may come from a Camulodunum Type 185b or 186. The find is not certainly provenanced but may be the “flagon with painted script” whose find spot is indicated in a published section drawing from the northern end of the villa. If so it would be associated with the third-century demolition of Room 8 (see Meates, G.W., Lullingstone Roman Villa, vol. 1 (1979)Google Scholar, fig. 15c, and, for the position of the section, fig. 14, Cf. the general plan fig. 6).

³² We are grateful to Professor R. Rodriguez-Almeida for the reading (from a drawing) and the suggested significance of the text. He tentatively restores Peregrino et Aemiliano cos. Armenius Peregrinus and Fulvius Aemilianus were the ordinary consuls for A.D. 244.

³³ Directed by A.M. Snell for the Trust for Lincolnshire Archaeology. Photograph supplied by Jenny Mann who supplied full information on this and the following item.

³⁴ See M. Henig and J. Ogden, Ant. Journ. forthcoming, where references to other rings inscribed with these three letters from the Lincolnshire area are given and their significance discussed.

³⁵ Excavations for the Trust for Lincolnshire Archaeology directed by C.J. Guy. For the site see Britannia xvii (1986), 389.

³⁶ Stamps consisting of the letters LVL in relief followed by an A, D, E, or F are recorded from several sites in Lincoln. For references to these with a discussion as to their significance see most recently Bogaers, J.E., Britannia viii (1977), 275–8.CrossRefGoogle Scholar

³⁷ Excavations by the Department of Greater London Archaeology, directed by Peter Mills. Information from P. Mills and Ken Whittaker who submitted the tile for inspection.

³⁸ For the expansion of the text see Wright, R.P. ‘Official tile stamps from London which cite the Province of Britain’, Britannia xvi (1985), 193–6.CrossRefGoogle Scholar Several other variations of the text are found normally, unlike the present version, within a rectangular border.

³⁹ Excavations for the Museum of London's Department of Urban Archaeology directed by John Schofield. Information on this and the following item from Michael Rhodes of the Department. See Dyson, T. (ed.), The Roman Quay at St Magnus House London, London and Middlesex Archaeological Society, Special Paper 8 (1986), 214Google Scholar, and fig. 7.7.

⁴⁰ T. Dyson (ed.), op. cit. previous note, 108, fig. 1. 46.

⁴¹ Maros/us (Celtic for ‘great’) is a very common element in Celtic compound names (see Holder, Alt-celtischer Sprachschatz s.v.) and is occasionally found on its own as a name (for examples, see Mócsy, Nomenciator), so there is no need to assume the omission of an I as would be the case if Marius in the genitive were intended, though this is of course also possible.

⁴² Excavations by Northamptonshire County Council, Archaeology Unit, directed by Brian Dix, who supplied full details and a drawing.

⁴³ Excavations by the Upper Nene Archaeological Society directed by Roy Friendship-Taylor who supplied full information on this and the following two items, and made the objects available for inspection.

⁴⁴ Excavations by the Chichester Excavation Unit for Chichester District Council, directed by J. Magilton, who provided details and submitted the sherd for inspection.

⁴⁵ During excavations funded by Rescue Archaeology, HBMC and the Metropolitan Borough of North Tyneside, directed by Mr CM. Daniels. Mr P. Moffat made it available and provided full details.

⁴⁶ This is the first sealing of Coh. IV Lingonum, but the restoration seems fairly certain. It was the third- and fourth-century garrison of Wallsend, but its location in Britain during the second century, and the second-century garrison of Wallsend, are not known. LCF on the reverse is presumably the initials of the issuing officer. The second line is unfortunately corroded; it is tempting to read LXX7, L(egionis) XX (centurio), but the damaged letter looks more like a star (also found on sealings) than X.

⁴⁷ During excavation of a cross ridge dyke for the Department of the Environment directed by (now) Professor P.A. Rahtz, who sent a tracing to RSOT.

⁴⁸ The beginning of a name of Celtic etymology (^*cuno- ‘dog’).

⁴⁹ By Mr J.G. Hunt with a metal detector. He intends to deposit it in the site museum, Aldborough, where there are other objects found by him. Dr Jennifer Price sent it for study.

⁵⁰ This is the first leaden sealing from Aldborough, although Mr Hunt has since found another. The die is unrecorded, and not explicable as the initial letters of tria nomina. BF is the regular abbreviation of b(ene)f(iciarius) and, if so, V might identify (obscurely) his superior officer; there would be a comparable sealing from Brough-under-Stainmore, which reads BFC (RPW'S corrected reading of CIL vii, p. 313 (Additamenta), 1269.35).

⁵¹ During excavation for the York Archaeological Trust directed by Mr N.F. Pearson. Mr A.D. Hooley made this and the next fifteen items available, and provided full details. He compares this piece with the very similar tent-piece illustrated in Cumb. Westm. n.s. xxxiv (1934), 71, fig. 6 (bottom right-hand corner).

⁵² Presumably a personal name in the genitive. The letter before T resembles II, but is too cramped for N; two letters are possible, but T cannot be excluded. Perhaps a Celtic personal name in -onius, such as Mattonius or Vattonius (not attested, but easily deduced from Vatto, Cf. RIB 610, Vatta), both of which suit the trace surviving before A.

⁵³ The centurion can be identified with Marcus Sollius Iulianus of the Third Cohort, who is named on a building stone from Hare Hill between Milecastle 53 and Turret 53a (Britannia vii (1976), 382, No. 14), that is from the rebuilding of the first stretch of Turf Wall towards the end of Hadrian's reign. He is now confirmed as a centurion of the Sixth Legion, which is already well attested in this stretch.

⁵⁴ This elaborate stamp seems to be unrecorded, and its interpretation is unclear. It ought to be the maker's name, like L(ucius) Aeb(utius) Thales T(iti) f(ilius) on the Vindolanda sandal (Britannia iv (1973), 332, No. 28 and PL. XLI B), but only QF (for Q(uinti) f(ilius)?) suggests this. The shoe looks like a more elegant version of the Vindolanda sandal but, as Mr Hooley observes, its construction is different: there seems to have been a layer of padding, possibly cork, enclosed by a leather strip, between inner and outer soles; its outline is unusual, if not unique, and it was stamped on the wear surface instead of the insole location usual for sandals. It was evidently a high-quality shoe not intended for heavy wear.

⁵⁵ The angle of the incision suggests that an inverted C should be read rather than two inverted Ts, but it is difficult to understand it as a centurial sign since T is unlikely to be an abbreviated centurion's name.

⁵⁶ Perhaps the end of a masculine personal name in the genitive case.

⁵⁷ References by number are to Wright, R.P., ‘Tile-Stamps of the Sixth Legion found in Britain’, Britannia vii (1976), 224–35CrossRefGoogle Scholar; and idem, ‘Tile-Stamps of the Ninth Legion found in Britain’, Britannia ix (1978), 379–82.

⁵⁸ During excavation for the York Archaeological Trust directed by Mr P.J. Ottaway. Mr A.D. Hooley made this and the next two items available and provided full details.

⁵⁹ During excavation for the West Yorkshire Archaeology Service directed by Mr P. Abramson. Mr J.C. Clarke made this and the next nine items available. They will be published, with other graffiti of less than three letters, in the final report on the Castleford excavations.

⁶⁰ The X is more finely incised and may be a later addition. The graffito is presumably the owner's name.

⁶¹ The first letter could be open A or M, but these look less likely. Between the / and the centurial sign is a vertical stroke of different character, which may only be casual damage. Since personal names in -orus are comparatively uncommon and usually of Greek etymology, it may be that the centurion's name ended in -orius and that its genitive was contracted to -or(i)i.

⁶² The personal name Ixus is once attested (CIL v i868), in north-east Italy, but of uncertain etymology since borne by a freedman.

⁶³ This is a common Latin cognomen, already attested in Britain (RIB 720). The sherds were found in a late context, where they were probably residual, since another sherd with graffito of the same fabric (which seems to be / imitating terra nigra ware) was found in a late-Flavian context.

⁶⁴ Excavation by the West Yorkshire Archaeological Service directed by John Hedges. Information from Quita Mould who made the object available for inspection (by MWCH).

⁶⁵ Excavations for the Glamorgan-Gwent Archaeological Trust directed by J.D. Zienkiewicz. Information and rubbing from Dr. Evans. For the site, lying across the river from the legionary fortress, see Britannia xvi (1985), 258–9.

⁶⁶ Information from Keith Blood of ‘Duffenfoot’, Greenhead.

⁶⁷ See Caruana, I., ‘The Provenance of RIB 2028’, Cumb. Westm. n.s. lxxxvi (1986), 259–60.Google Scholar