² Information and photo from J. Maloney, who directed excavations for the Museum of London, Department of Urban Archaeology. For the site see Britannia xi (1980), 379.Google Scholar

³ Excavations supervised by Mrs J. Norton under the direction of B. Hobley for the Museum of London, Department of Urban Archaeology. Information from Michael Rhodes, Finds Officer at the Museum.

⁴ Information from A. C. King, joint director of the Hayling Island Excavation Project. For the site see Downey, R., King, A. and Soffe, G., The Hayling Island Temple: Third Interim Report (London, 1979).Google Scholar

⁵ l. 1. The name may have been Na]evian[us, which is statistically the most likely of several possibilities.

⁶ Full information and photograph from R. J. Briggs. Excavations by the Wimborne Archaeological Group directed by A. G. Giles. The stone is now at the Priest's House Museum, Wimborne. For the site see Britannia xi (1980), 391–2Google Scholar, and for a graffito from the site see below No. 31.

⁷ Some of the letters are provided with exaggerated serifs. A cursive T has been employed in the word et in l. 2.

⁸ The nomen is found similarly abbreviated to three letters in RIB 344 (Caerleon) and in full in Britannia x (1979)Google Scholar, No. 20 (Binchester). It is formed from the common cognomen Cupitus (Cupitus/a is found 118 times in CIL, of which 79 examples come from Celtic areas).

⁹ Several Celtic, but no purely Latin cognomina begin with the letters Vep. e.g. Vepogenus, RIB 191 (Colchester). If, however, the third letter is to be read as R, the name may have been the common cognomen Verinus/a. For this name see below under No. 49 (from Ribchester).

¹⁰ A.D. 258. This consulship occurs on another inscription from Britain, RIB 1589 (Housesteads). See ILS iii 2, p. 946 (index xvii) for examples of the year and/or date of death or burial inscribed on funerary monuments (mostly from Rome).

¹¹ Excavations for the Department of the Environment and the Committee for Rescue Archaeology in Avon, Gloucestershire and Somerset directed by Dr Ann Ellison who provided full details. For the excavations in 1978 see Britannia x (1979), 323. See also Ann Ellison in ed. Rodwell, W., Temples, Churches and Religion in Roman Britain, BAR British Series 77 (1980), 305–19Google Scholar, especially 314 and the plan, fig. 15.3.

¹² For the names of the sculptor and his father cf. Suarigillus Vasilli fil. CIL xiii 4433Google Scholar, Metz (not in Holder, Alt-celtischer Sprachschatz) derived, like Suarica, ibid. 5532, Dijon, from Suarix,–igis.

¹³ Lover(n)ios, ‘son of the fox’ was the name of the father of Bituitos defeated by Rome in B.C. 121 (Strabo iv. 2.3). The name Lovernianus derived from it, is already attested in the province (JRS lix (1969), 239Google Scholar, no. 23) on a pewter platter from Appleford, Berks, (now Oxon.).

¹⁴ Excavations for the Bath Archaeological Trust were directed by Professor B. W. Cunliffe. See Britannia xi (1980), 387–8.Google Scholar

¹⁵ After conservation in the laboratory of the Institute of Archaeology, Oxford, the curses were examined by both of us independently, before producing agreed, but provisional, texts, although in No. 9 below the initial reading of the first and last lines was by RSOT alone. The drawings of texts Nos. 6 and 7 and the form in which they are here published are due to MWCH and those of Nos. 8 and 9 to RSOT. 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, and the use made by us of Ian Marriott's word and letter-group Index of Curses. This shows, by the absence of parallels from other provinces, the strongly local character of the Bath texts (cf. Britannia x (1979), 342, note 11).

¹⁶ Analysis of this and the other tablets by Dr A. M. Pollard of the Oxford University Research Laboratory for Archaeology and the History of Art shows that the metal is 79·7% tin and 20·3% lead. The proportions of the metals in the rest are: No. 7 = 57·3% lead: 42·6% tin (o·1% copper); No. 8 = 26·1% lead: 73·8% tin (o·i% copper) and No. 9 = 40·6% lead: 59·1% tin (0·3% copper).

¹⁷ (a) 1. The diagonal stroke between A and v of DONAVI existed before either stroke had been cut and is not part of the text.

1–2. arge]ntiolos the only likely restoration to be indicated in O. Gradenwitz, Laterculi vocum Latinarum (1904) (reverse dictionary). This rare word is the diminutive of the adjective argenteus, itself used officially of a specific denomination of silver coin under the tetrarchy (see K. Erim, J. Reynolds, M. Crawford, ‘Diocletian's Currency Reform: A New Inscription’, JRS lxi (1971), 171–7, and compare the use of both nummus and nummulus. Argentiolus occurs with the specific meaning of a silver coin in the gloss on the phrase concisum argentum in Juvenal, Satires v, 14,291, in argentiolos sive nummos (ed. P. Wessner, Scholia in Iuvenalem vetustiora (1931)). The scholia on Juvenal were, according to Mommsen (‘Zeitaltes des Scholiasten Juvenals’, Ges. Schr. vii, 509–11Google Scholar) probably composed in Rome about A.D. 400. In HA Aurel 9.7 and 12.1 the phrase argenteos Philippeos minutulos occurs; for this and other numismatic and related terms in the Historia Augusta, see RSOT ‘Fairy Gold: Monetary History in the Augustan History’, in ed. King, C. E., Imperial Revenue, Expenditure and Monetary Policy in the Fourth Century AD, BAR International Series 76, (1980), 255–79.Google Scholar

3. nominibus. The word nomen is used in a transferred sense of creditor or debtor from the practice of writing the name of the person concerned at the head of the page that contained his account, cf. Oxford Latin Dictionary s.v. nomen 22(a).

4. exactura, for the classical exactio, is not included in the Thesaurus Linguae Latinae. The earliest examples of the word used in this sense in J. F. Niermeyer, Mediae Latinitatis Lexicon Minus (1976), date from the eleventh century. For the phrase a nominibus … exactura as a whole, cf. Cicero, Att. v 1.2 extrema exactio nostrorum nominum.

5. For the spelling of Senecianus with the letter 1 cf. RIB 306, the Lydney curse, and CIL vii 1305, the gold ring from Silchester. For the confusion between the two letters, see J. C. Mann, ‘Evidence of Spoken Latin in Britain’, Britannia ii (1971), 218–24, esp. 220–1.

6. The name Anniolus/a is a diminutive formed from the nomen Annius cf. CIL xiii 1396 Annia Anniola, and as such should be spelt with double N though in at least one other case (CIL viii 10481) it is found with only one.

7. C(h)arta picta. Literally ‘painted sheet’. Such a phrase would be an appropriate description of the draft of the Diocletianic inscription painted on papyrus found at Oxyrhynchus (Pap. Oxy. 1972 xli, Section iv, No. 2950 and pl. iii). If drafts were written for the Bath curses (cf. note below on No. 8 l. 2), perhaps in ink on wood like the Vindolanda tablets, then it is conceivable that the phrase could be applied to them, but there may be some other technical meaning perhaps connected with accounting, cf. the note on nomen (l. 3) above and the word perscribere, ‘write out in full’, which is used both of copying and of writing a draft for a sum of money.

¹⁸ 2. The name occurs five times in PLRE i (A.D. 260–395) and twice in PLRE ii (A.D. 395–527), the earliest dated example being Domitius Dracontius, magister privatae rei Africae attested A.D. 320–1, although Aurelius Agapitus Dracontius, eq(ues) R(omanus), is assigned a third-century date. For other examples see Diehl, Inscriptiones Latinae Christianae Veteres Index i, and Dr W. Smith, Dictionary of Greek and Roman Biography and Mythology (1870). Kajanto, The Latin Cognomina, 115, shows that cognomina ending in -ius tend to be late (cf. Innocentius l. 4 and Simplicius l. 5), but there appear to be no obviously fourth-century letter forms.

3. Kajanto, op. cit., 135, 252 shows that Innocentius was a late cognomen favoured by Christians. Innocentia, for example, occurs in one of the inscriptions from the Water Newton treasure, Britannia vii (1976), 385 No. 32.Google Scholar

5. For the spelling of Senicio with 1 in the middle syllable cf. Senicianus No. 61. 5 + note. The name is included in A. Holder, Alt-Celtischer Sprachschatz, but Kajanto, op. cit., indicates no predominance in Celtic areas in contrast to Seneca and Senecianus.

6. The second stroke of the initial c instead of being written above the first stroke is written next to it. The two strokes are linked at the bottom.

7. The initial letters si are omitted, perhaps by haplography due to the final s of Candidianus in the preceding line. Simplicius, though it can be a nomen (as in RIB 1546) will have been a cognomen with the late suffix -ius, cf. Kajanto, op. cit., 253, who shows that it was common among Christians.

8. The letter B is written in an identical way to the letter D, cf. the way in which these two letters are written on the curse from Caerleon (RIB 323). Belator is normally written with double L as in Flavius Bellator, RIB 674 and Iulius Bellator, EE ix (1913)Google Scholar, No. 1331. Nine out of twelve examples of this name in CIL come from Celtic areas, and it is probably the Latinized form of the Celtic name Ballatorix (see Kajanto, op. cit., 16, 17, 361).

9. Suril(l)us/a is included by A. Holder in the Alt-Celtischer Sprachschatz, twelve of his seventeen examples coming from Danube provinces.

10. Austus is apparently unattested, although the nomen Austius, perhaps formed from it, is found (CIL ix 2869Google Scholar, 2870), cf. the cognomen Aust-olus included by Holder in Alt-Celtischer Sprachschatz.

This list of eleven names includes two, Belator and Surilla, that are probably Celtic (cf. also note on Senicio l. 5). Three of the other names have the late Roman suffix -ius. The nine men and two women listed will have been the enemies of the person for whom the curse was written, compare the defixio found at Bath in 1880 (RIB 154), which lists nine suspected rivals, and the inscribed sheets of lead from Leintwardine listing thirteen names (JRS lix (1969), 241, No. 31).

¹⁹ The tablet was not rolled up, but at least one nail was driven through it, creating the crux in l. 10. The writing is so regular that, where traces remain, it has not usually been necessary to indicate any doubts about which letter to restore.

2. Bruceri: the name is not attested, and Bruceti is a local patronymic (RIB 105, 151), but the R is certain. The scribe perhaps mistook a cursive T for R in the text he was copying (which would throw some light on the circumstances of composition); cf. RIB 306 (petmittas), another text in capitals, where T has been written for R.

5. devoveo: VE is the only ligature in the whole text, and looks like a correction of a scribal error due to the repeated E_P, V and O.

6. caracellam: the correct spelling is given in l. 16.

10–12. Meaning and restoration uncertain. The regularity of spacing between letters and of left-hand margin suggest that two letters have been lost in 10 and one in 12 (where adigat seems unavoidable). The expression letum adigere is not exactly paralleled, but its usage with vulnus would suggest it meant ‘inflict death upon’ (someone, dative); thus maximo, and not […]um, would seem to be the recipient. In this case, maximo should be taken as a personal name (‘upon Maximus’), and […]um either as a conjunction (although dum or cum, the obvious candidates, would not fill the space available) or as an adjective oddly removed from letum, say suum in emphatic position (‘her own particular brand of death’ or, less grammatically, ‘his own particular death’, just possibly a reference to suicide). The syntax of curse tablets is commonly based upon an indirect command introduced by ut (e.g. Britannia x (1979), 342, No. 2, erogat deum Mercurium ut nec ante sanitatem habeant …), which would guarantee ut in 10 as a separate word (not part of utrum e.g., even if that were grammatical) governing permittat and perhaps adigat. (Subordinate clauses in indirect command in British curse tablets are sometimes, but not always, found in the indicative). The chief difficulty in reading Maximo (proper name), apart from the illogicality of supposing that the thief can return the cloak after he has been put to death, is that it caps an elaborate curse on ‘whoever’stole the cloak with a casual reference to the thief by name. The unpublished tablet (Small Find No. 671) cited in note 21 below, which may refer to this very theft, does not name the thief. These objections, however, may be too literal-minded; curse tablets should be treated as magical texts, not the legal documents they sometimes superficially resemble. It should be noted that the writer of RIB 306 informs Nodens that he has ‘lost’ a ring and requires him to curse inter quibus nomen Seniciani until the ring is returned.

15. nascentes: there is damage due to corrosion, but enough survives of the doubtful letters to be certain. For the meaning of nec natos nec nascentes (and thus confirming the reading), cf. CIL vi 8063, xii 3702, apatronus provides a future burial place for himself and his freedmen natis nascentibus.

15–16. donec: the position of the left-hand margin suggests two letters have been lost to corrosion (a trace survives of C). Do (‘I give’) is precluded by the syntax of the rest of the sentence, and the formula is paralleled in RIB 306, donee perferat usque templum Nodentis.

²⁰ The regular contrast between thick and thin strokes, and occasionally the double (‘W’) profile of the scratches, suggests that a bronze pen with cleft nib was used, like that illustrated in R. E. M. Wheeler, London in Roman Times (1930), 58. This example of ‘rustic capitals’ is characterised by the long descender of the L and, in particular, the s being made in three unconnected strokes.

²¹ The ‘curse’ is of standard type: a petitioner asks the deity to curse a thief until he returns the stolen property to the temple, cf. RIB 306 (see above note 19 on lines 15–16) and Britannia x (1979) 343, No. 3. The si vir si femina (etc.) type of formula is peculiar to British curse tablets, where it is common. An unpublished tablet (Small Find No. 671), for example, curses the thief of a hooded cloak, furem qui caracallam meant involavit si servus si liber (etc.). If this refers to the same mishap as the present tablet, it raises interesting questions about how the tablets were composed and inscribed, since it is written not in ‘rustic capitals’, but in cursive. For the expression nec natos nec nascentes, see n. 19 on l. 15: it would seem to be another of the quasi-legal formulae found in British curse tablets. For the periphrasis ad templum sui numinis, cf. phrases in unpublished tablets, dono numini tuo (Small Find No. 621) and templo Sulis (Small Find No. 615), and Solinus’ use of Sulis Minervae numen (Coll. 22, 10, emended: see n. 24). The chief interest of this tablet, however, may be that it attests the deity's name in the nominative case for the first time: Sulis, not Sul, is the correct form, as the late Professor Tolkien divined (see note to AQUAE SVLIS in RIB, p. 42), and *Sulis need no longer bear an asterisk.

²² 1. The left-hand corner is crumpled, the upper part of the middle of the line has been lost, and there is damage due to corrosion at the end, but the reading accords with the surviving traces.

4. a has been repeated in error after nomina. (In view of the similar error in 5, hyper-correction of the e of eorum is not likely).

5. qui iuraverunt has been repeated in error.

7. The reading is facias (not faciat), and the subject is presumably the goddess of the spring into which this tablet was thrown, but the syntax is confused. Periurare takes an accusative of the deity perjured, satisfacere a dative of the party compensated, so that deae Suli should be taken with satisfacere; this means that, grammatically, the subject (‘you’) of facias is in the dative case. Perhaps the writer intended an explicit invocation on the lines of quicumque illic periuraverit, dea Sulis, facias ilium tibi sanguine suo satisfacere, rather as an unpublished tablet (Small Find No. 622) seems to invoke tu, domina dea in mid-text without being formally addressed to the goddess; this invocation may have been obscured, however, by the vocative dea Sulis absorbing the tibi and being attracted into its case, at the cost of grammar and sense.

²³ Dr J. D. Thomas, with whom we have discussed this question, points out that the letter-forms with one exception can be found in the Vindolanda Tablets; the exception is the E formed by two hooked strokes, which is never found in Latin papyri before c. 200. (It can be seen, with other late-Roman letter-forms, on the inscribed brick from Binchester, Britannia x (1979), 347Google Scholar, No. 20 and fig. 24). But dated texts in Latin cursive outside Egypt are too uncommon for any certain dating to be possible.

²⁴ For the translation of the second sentence see the first note. There are several points of interest. The tablet was not rolled up or pierced with a nail like most ‘curse tablets’, and does not in fact fall into any of the usual categories: instead of cursing a thief (e.g.), it protects an oath with a sanction against perjury. (The notion of ‘paying with one's own blood’ has already occurred on a curse tablet from Britain: sanguno suo solvat, JRS xlviii (1958), 150, No. 3).Google Scholar The oath-takers are a family group defined by their relationships to Uricalus: his wife, his son and daughter, his brother and presumably his sister-in-law. The oath itself is not specified, but since it was sworn by two brothers and their families, it may have regulated the division of inherited property. The name Uricalus seems to be unattested, but Calus is known (e.g. CIL vii 1336.Google Scholar 216) and names in Viro- and Viri- are common (of which Uri- would be a by-form, like Urilucolus, Britannia viii (1977), 444, No. 101)Google Scholar; for Alogiosa cf. Alogiosus (CIL xiii 1331) and Aloiosa (ii 4976. 39)Google Scholar. Uricalus' parents gave one of their sons a ‘Roman’ name, the other one a ‘Celtic’; and Uricalus married a woman with a ‘Roman’ name which was adapted for their children: this intermingling of ‘Roman’ and ‘Celtic’ names is typical of the Bath tablets as a whole. The date and place of the oath are recorded. The 12th of April may have had a local religious significance, since it was not thought necessary to specify the year, but this is only conjecture. (At Rome it was the first day of the Ludi Cereales). This is the first mention of the fons Sulis from which Aquae Sulis derived, but the name could have been inferred from Solinus' well-known reference to the burning of coal in the temple at Bath: (There are in Britain) fontes calidi opiparo exculti apparatu ad usus mortalium: quibus fontibus praesul est Minervae numen, in cuius aede perpetui ignes numquam canescunt in favillas, sed ubi ignis tabuit vertit in globos saxeos (Collectanea Rerum Memorabilium 22, 10). This metaphorical use of the rare word praesul (‘dancer at the head of a religious procession’) is hard to match, and the coincidence of the syllable sul tempts one to modify an emendation printed by Holder to read quibus fontibus praeest Sulis Minervae numen.

²⁵ Information from A. Fitzpatrick. Excavations for the Essex County Council Planning Department, Archaeological Section and the Department of the Environment directed by Robin Turner. For the site see Britannia xi (1980), 378–9Google Scholar and R. Turner, Ivy Chimneys, Witham, Excavations 1979, Interim Report, Essex County Council (1980), with illustration p. 8, fig. 6.1.

²⁶ Details together with a copy of a drawing and squeezes submitted by D. Zienkiewicz who directed excavations for the Welsh Office (Ancient Monuments Branch). For the excavations in 1978, see Britannia xi (1979), 273–4.Google Scholar

The inscription will have been similar to the Trajanic S.E. gateway inscription from York (RIB 665) where both the title Dacicus and the phrase tribuniciae potestatis were written out in full.

²⁷ During excavation for Carlisle City Council and the Department of the Environment directed by Mr M. R. McCarthy, who made it and the next item available.

²⁸ N is preceded by a serif which presumably belonged to A or 1, although H, M or R are epigraphically possible. The two AS are linked by two parallel lines and other strokes, which cannot be read as ligatured letters; perhaps an error was remedied by cutting out a roughly parallelogram-shaped panel and plugging it with mortar, before incising new letter(s). That this letter was N is suggested by a deeply-incised vertical. A possible restoration is then [ANTON]INIA[N]A. The third letter, of which a trace survives in the previous line (c, o or s), might be part of an auxiliary regiment's name, […]O[RVM]. There seems to have been no concluding formula (VSLM, etc.).

²⁹ The combination of standing figure and altar suggests a Genius (for local examples see R. P. Wright and E. J. Phillips, Roman Inscribed and Sculptured Stones in Carlisle Museum (1975), Nos. 18 (RIB 944) and 174, and Britannia viii (1977), 376Google Scholar with pl. xxiii(a)). Unfortunately the original width of the panel cannot be deduced from what remains of the figure above. Mr I. Caruana has suggested the reading [GE]N[I]O [ET M]A[T]R [IB]VS, which is attractive but open to objection. The combination of genius and matres can be paralleled (RIB 130, 1334), but not so baldly; it is, for example, usual to specify the genius (esp. loci or of a military unit). In line 1 there is no trace of 1 ligatured to the N; in line 2 the 1 does not resemble T; and more than two letters must be lost from line 3, which precludes the restoration [IB]VS. However, the diphthong -ai- is un-Latin (its only occurrence in Romano-British inscriptions is in Belatucairus, the name of a local god, but any restoration on the lines of Deo sancto Belatucairo is implausible); a word-division here is unlikely (the second word would have to be ira or one of its cognates, or part of an exotic name like Cocceia Irene, RIB 507); and it is possible the serifed 1 is an error for T, cf. RIB 1334 for a comparable confusion, MATRIRIBVS for Matr(ibus) tribus.

³⁰ During consolidation directed by Mr R. E. Birley, who made this and the next two items available.

³¹ cf. RIB 1708 (Chesterholm), a building stone of Leg. XX VV with boar facing R. Vexillations of the Twentieth in the Wall area are already attested by RIB 1014 (R. Gelt quarries, perhaps A.D. 207, cf. RIB 1009) and 1149 (Corbridge, A.D. 163).

³² Of the first letter only the top of a curve survives, ligatured to the top half of E (F can be excluded); of the M only the first apex, but enough to preclude A and make N unlikely. The first line of a dedication slab almost always consists of the name and titles of a god or emperor, but none of them seems to fit. Possible combinations of letters are found in RIB 89 (I.O.M. in hono REM domus divinae…) and JRS li (1961), 191, no. 1Google Scholar (aeDEM principiorum cum basilica sub… name of legate). At Vindolanda, a fort with an elaborate principia rebuilt several times, the second possibility is more attractive.

³³ Only the R. margin survives. The approximate number of letters missing is calculated on the assumption that D(is) M(anibus) was centred in the panel, as is suggested by the space still surviving to the left of D(is). Line 2 probably read [Iu]liae or [Ae]liae, and line 3 a female cognomen ending in -iae.

³⁴ By Mr B. Carter of Stroud (Glos.), who left it with the tenant of Walltown Farm, Mr Stewart. (Information from Mr C. M. Daniels and Mr Stewart). It is now in the Museum of Antiquities, Newcastle, where Miss L. Allason-Jones made it available.

³⁵ The centurion is otherwise unknown. Two other centuries of Coh. V are already attested at Walltown (RIB 1766) and nearby at Allolee (MC 44, RIB 1754); that it belonged to the Twentieth Legion follows from the discovery of building stones of that legion's Coh. X at Turret 44b (RIB 1762) and only 25 ft. west of Turret 45a (JRS l (1960), 237, no. 11(a)).Google Scholar

³⁶ It was noticed by Mr C. Richardson of Carlisle Museum. The century, presumably of the Twentieth Legion, is the most prolific on the Wall: this is the sixth of its building stones to be found (RIB 1499, 1506, 1681, 1859, 1861). For the identification of the legion, and the Gallic origin of the centurion, see Birley, E. in AA 4th ser. xvi (1939), 234–5.Google Scholar

³⁷ During excavation for the Bath Archaeological Trust (see Britannia xi (1980), 387–8) directed by Professor B. W. Cunliffe, who made this and the next five items available. Another (fragmentary) silver patera with a punched inscription was examined, but the letters were too worn to be legible. This, and the following five items are reported on by RSOT. For four inscribed tablets from the reservoir, see Nos. 6–9 above.

³⁸ This variant of the usual scripula symbol (like a c reversed) also occurs on two pieces from the Hildesheim hoard: E. Pernice and F. Winter, Der Hildesheimer Silberfund (1901), 45–6. The patera now weighs 189 gm, in Roman terms nearly 7 oz. (1 libra (12 oz.): 327 gm), but the weight may have been noted before it was repaired; in any case, the exact number of scripula (24: 1 oz.) has not been added to the total.

³⁹ Excavations by the Milton Keynes Archaeological Unit directed by J. Barnbrook for Milton Keynes Development Corporation. Information and drawing from Ros Tyrell. For the site see Britannia x (1979), 303.Google Scholar

⁴⁰ For similar greetings from Pompeii see CIL iv Suppl. 2, index iii s.v. ave.

⁴¹ Excavations by the Excavation Section of the Grosvenor Museum directed by T. J. Strickland. Information from P. Carrington of the Museum. For the site, in the dexter side of the retentura of the fortress, see Britannia vi (1975), 240.Google Scholar

⁴² Excavations conducted by D. J. P. Mason of the Excavation Section of the Grosvenor Museum for the Department of the Environment and Chester City Council. Information from P. Carrington of the Museum. For the site see D. J. P. Mason, Excavations at Chester, 11–15 Castle Street and Neighbouring Sites 1974–8, Grosvenor Museum Archaeological Excavations and Survey Reports, No. 2) 1980, and Britannia viii (1977), 387.Google Scholar We are deeply indebted to Professor Emilio Rodrigues-Almeida for interpreting the inscription for us.

⁴³ See M. H. Callender, Roman Amphorae (1965), 20–2 for a brief discussion of tituli picti on globular Spanish amphorae.

⁴⁴ This formula is attested for the third century when the south Spanish export trade was taken over by the state. This is usually linked with the presumed confiscation of the estates of adherents of Clodius Albinus in 196. (See Callender op. cit. 22). Professor Almeida, however, gives A.D. 117–260 as the extreme date range for the sherd (implying some shipping by imperial agents from the beginning of Hadrian's reign) and, within that period, A.D. 220–26 as the most likely date of the vessel.

⁴⁵ Professor Rodriguez-Almeida reads … CI. for the first two surviving characters. For FSA he suggests either f(ecit) Sa(…), or f(iglino) Sa(…). The only figlina to be suggested by I. Marriott's analytical index of dipinti on amphorae is that at Saxum Ferreum in the Corduba area (cf. CIL xv 4171Google Scholar, … Fig. Saxo Ferreo, with a consular date of A.D. 14). One might however have expected the abbreviation FSF as on CIL xv 2830Google Scholar and 2852 and EE ix, No. 424, 8 d. For Saxum Ferreum see the article in PW. Professor Rodriguez-Almeida notes faint traces of other letters below, and to the left, of the surviving line.

⁴⁶ See Cumb. Westm. n.s. lxv (1965), 133–68Google Scholar. It is now in the possession of Mrs W. Davies, Kirkbampton, Cumbria. Information from Mr J. Taylor, who made it available.

⁴⁷ The B is written as a capital. The fourth letter in line 1 could have been A, M, N, R or s. The sherd is too fragmentary for it to be determined whether there were other lines of writing above or below what has survived.

⁴⁸ During excavation for Carlisle City Council and the Department of the Environment directed by Mr, M. R. McCarthy, who made it available.

⁴⁹ Recorded by K. Jarvis and I. Horsey of the Poole Museums Archaeological Unit. Information, photograph and drawing from K. Jarvis. For the site see Britannia iii (1972), 346Google Scholar and ibid. iv (1973), 316.

⁵⁰ Excavations by the Wimborne Archaeological Group directed by A. G. Giles who provided details, photograph and rubbing. For an inscription of the same site see above, No. 4.

⁵¹ During excavation for the Department of the Environment directed by Mr Rainbird, J. S. (Britannia iii (1972), 309).Google Scholar It was not published with others from the excavation (ibid., 355, No. 20) but was made available by Mr I. M. Ferris of the Bowes Museum, where it now is.

⁵² The bottoms of most letters have been lost, and some alternative readings are given in brackets: 11 (IL); s (c, G); B (P, R); A (N, but unlikely after C). There is a space as indicated, and perhaps another after DVBII, which suggests one should read personal names in the genitive case: perhaps […] Ves[u]vii Dubii Ca[ndidi etc..

⁵³ With a metal detector. The person responsible took it for identification to the Museum of Antiquities, Newcastle, where Miss L. Allason-Jones had a drawing made.

⁵⁴ The standing eagle is well represented among signet types: see M. Henig, A Corpus of Roman Engraved Gemstones from British Sites 1978), Nos. 689–710. The regiment is attested in Britain by the diploma of 122 (CIL xvi 69)Google Scholar as ala I Pannoniorum Sabiniana, but as ala Sabiniana on a third-century (?) tombstone at Halton Chesters (RIB 1433). It has been located at South Shields on the evidence of a lead sealing of Coh. V Gallorum found there with the reverse ASA (AA 4th ser. xi (1934), 102, no. 19, which corrects earlier attempts at a reading in EE iii 202a, EE iv 706a (wrongly cited as a tile stamp in Cumb. Westm. n.s. xxxix (1939), 213, 221), and R. E. Hooppell, On the Discovery and Exploration of Roman Remains at South Shields in the years 1875–6 (1878), 39, No. 2), but ASA is surely an abbreviated personal name. The combination of abbreviated unit-title and personal name is common among the lead sealings from Brough-under-Stainmore (Cumb. Westm. n.s. xxxvi (1936), 118 f.)Google Scholar; an example from Sewingshields is published below (No. 104). Such unit-titles have been conventionally expanded in the nominative case; this is the first sealing to indicate a case-ending – in the genitive. Since lead sealings of military units occur at military sites, the find-spot is also of interest.

⁵⁵ Excavations for the Beauport Park Excavation Committee directed by G. Brodribb and Henry Cleere. Information from the former who supplied the sherds for inspection.

⁵⁶ Among other possibilities are the common Celtic name Matugenus/a, see below, No. 101, and Maturus/a, found 74 times in CIL.

⁵⁷ Excavations by the Colchester Archaeological Unit, directed by Carl Crossan for the Colchester Excavation Committee. For the site see Britannia ix (1978), 451. Information from Nina Crummy.

⁵⁸ Found by Mr B. Pateman and presented by him to the Saffron Walden Museum. Information from Jenny Hall of the Museum of London to whom the object was submitted for identification.

⁵⁹ Information from M. Millett who traced the graffito from a rubbing by Malcolm Lyne. The name of the finder and present whereabouts of the sherd was not recorded. Mr Lyne points out that the last letter looks like an o and that its appearance as Q (sic) may be fortuitous since the ‘tail’ has a different character to the rest of the letter. It is possible that o was originally written and this was then mistakenly corrected to a Q (rather than a p). For British examples of the slogan see RIB 289 and 930 (both bases) and 2308 and 2314 (both milestones). For a milestone of Julian from near Verona on which the phrase is abbreviated as here, including the omission of nato see CIL v 8035.Google Scholar

⁶⁰ With the aid of a metal detector by a Mr Atwell, in whose possession it remains. The find-spot is said to have been outside the scheduled site, for which see T. Davies Burlton in Trans. of the Woolhope Club (1883–5), 340–2. The fragment was made available by Mr R. Shoesmith of the City of Hereford Archaeology Committee, who provided details. It was analysed by means of X-Ray Diffraction by Dr M. Pollard of the Oxford University Research Laboratory for Archaeology and the History of Art, and proved to be an alloy of lead (62%) and tin (38%).

⁶¹ The penultimate letter(s) are not clear. The final s precludes a genitive case-ending (except perhaps illlus or ullius) of a personal name after c(un)no, a possibility suggested by the nature of the drawing. Instead one should probably read a personal name in Cuno…, perhaps C(u)novill(iu)s, which is apparently unattested. That it may have been a Celtic feminine like Vellaunius, cf. RIB 369 with note ad loc.

⁶² Excavations directed by I. M. Stead for the Baldock Excavation Committee. Information from Jennifer Foster.

⁶³ Information and photograph from G. R. Burleigh. For excavations conducted by Mr Burleigh in another part of the field at the time see above p. 345.

⁶⁴ During excavations by the Welwyn Archaeological Society directed by A. Rook, who supplied details and information.

⁶⁵ In (a) the first letter could just possibly by an 1 preceded by the tail of an R, or part of an x. If v is the correct reading the name could be restored as Post]umin(a)e.

⁶⁶ Excavations directed by J. Lunn. Information, drawing and rubbings from the Deputy Keeper of Archaeology, Verulamium Museum, Stephen Greep. On the sherd itself is written ‘gully at E.W. road’.

⁶⁷ Of the first letter two horizontal strokes only survive, and it could be an E with shorter central stroke which does not appear on the sherd. Since the name… cianus is in the nominative, rather than vocative as might have been expected, it may have been the subject of a verb such as inquit. The final R might be part of a second exhortation such as REPLE, REMISCE or REPETE, all of which are found on Rhenish motto beakers, cf. M. Bös, ‘Aufschriften auf rheinischen Trinkgefässen der Römerzeit’, Kölner Jahrbuch für Vor- und Frühgeschichte iii (1958), 20–5.

⁶⁸ It was acquired in 1685 as part of the Tradescant Collection by the Ashmolean Museum, Oxford, where Mr P. D. C. Brown provided full details and made it available. RSOT.

⁶⁹ The name is well attested as a samian potter's stamp, in which form it has been found in Britain, but chiefly occurs at Trier and in Upper Germany. For the practice of inscribing a cinerary jar with the name of the deceased, cf. Thalius Vassu at Colchester (EE ix p. 674Google Scholar No. 1351(a) = T. May, Catalogue of the Roman Pottery in the Colchester and Essex Museum (1930), 245, No. 3), and the goblet found in an inhumation burial at Verulamium with the freshly incised graffito Maurusi (R. E. M. Wheeler, Verulamium (1936), 138, with pl. cxv).

⁷⁰ Excavations for the Canterbury Archaeological Trust directed by T. Tatton-Brown, with site directors Kevin and Marion Blockley. Information from N. Macpherson-Grant and Marian Green on this and the following three items. Rubbings from Joanna Bird.

⁷¹ For the name compare CIL xii 2632 (Geneva)Google Scholar, Segellius, a nomen which will have been formed from the peregrine name Segellus/a.

⁷² A nomen Turranius or Turpilius, perhaps abbreviated to three letters, seems more likely than one of the cognomina beginning TVR, all of which are comparatively rare.

⁷³ Agat(h)opus (which also occurs 144 times in CIL vi), and the rare Syntropus are the only names to be suggested in I. Marriott's analytical index of names in CIL xiii and RIB i.

⁷⁴ By Master M. Halliwell while bird watching and retained by him. Information from Mr K. Parfitt.

⁷⁵ Taking the MA as an abbreviation for the owner's nomen and the MO as the beginning of his cognomen which may have been either Moderatus or Modestus, the two most frequently found cognomina beginning with these two letters.

⁷⁶ Excavations for Lancaster University and the Department of the Environment directed by J. H. Witherington who provided MWCH with details. For the site see Britannia x (1979), 291–2.Google Scholar

⁷⁷ The identification of the stamp is due to Mrs K. Hartley. Iunius worked at Mancetter and Harthill probably during the third quarter of the second century.

⁷⁸ For other examples of the name in Britain see RIB 447 (Chester, and Britannia viii (1977), No. 101 (Ilchester).Google Scholar

⁷⁹ Excavations for Lancaster University and the Department of the Environment directed by J. H. Witherington, who provided MWCH with details on this and the following five items from the Access Road site. For the site see Britannia ix (1978), 429.Google Scholar We are grateful to Felicity Wild for supplying rubbings of the samian and to Mr Witherington for sending photocopies of drawings by Paul Gibbons of the graffiti on the coarse pottery and to Mr C. J. Chandler for answering further queries about the graffiti.

⁸⁰ The personal name Tunger/Tungra is not apparently attested, but Kajanto, The Latin Cognomina 201, lists a number of other personal names derived from tribes in Gallia Belgica and Germania Inferior (as here) eg. Batavus, Trever, Remus, etc.

⁸¹ RIB 577 (Manchester, a centurion in Cohors I Frisiavonum. Compare Anavo, a woman, CIL xiii 4270Google Scholar (Luxemburg), Sedavo, a Batavian, ibid. 7025, (Mainz), and Haldavvo, a Nervian, ibid., 8340 (Cologne).

⁸² Compare JRS xlix (1959), 138Google Scholar, No. 20 a samian bowl (Drag. 31) Candidi sum, and ibid. xxiii (1933), 216, No. 13 on a flagon, Gai sum peculiaris. For the restoration f[eliciter] cf. the exhortation on a graffito on a cooking pot from Chester, Britannia viii (1977), 436Google Scholar, No. 46. The word is also found with personal names on tiles, see EE vii 1139 and CIL vii 1256.Google Scholar

⁸³ It is possible also to read…]AXI, but I. Marriott's analytical index of personal names of RIB i and CIL xiii gives only one name Hannaxus, CIL xiii 323.Google Scholar

⁸⁴ Excavations for the Cumbria and Lancashire Archaeological Unit directed by Adrian Olivier. See above p. 331. The site includes the defences of the newly discovered early fort but lies to the north of the defences of the well known later fort. Excavation revealed that the area immediately north of the later fort's defences remained unoccupied but also located the southern limit of the extramural vicus. We are grateful to Felicity Wild for sending rubbings of the graffiti and to C. J. Chandler for supplying details of the contexts in which the sherds were found, MWCH.

⁸⁵ For Dida cf. JRS liii (1963), 161Google Scholar, No. 8 (c) and RIB 1365, with a note that the name is Thracian.

⁸⁶ For examples of the relatively common name Venustus/a from Britain see CIL vii 1331Google Scholar, 191 and JRS xliv (1954), 107, No. 25.Google Scholar

⁸⁷ I. Marriott's analytical index of personal names in RIB i and CIL xiii gives no instance of the combination of letters ROQ so it seems best to take the stop after the second letter as intentional. If the letters are part of a name it could either be in the dative case or, more likely, in the nominative, since names ending in -ro exist in both Celtic (eg Danomaro) and Latin (eg Macro). The Q could then be an abbreviated praenomen or the first letter of a nomen, either way coming at the beginning of the text, with 13–15 letters to be supplied between it and the RO.

⁸⁸ Excavations by the Department of Urban Archaeology of the Museum of London supervised by G. Milne under the direction of B. Hobley. Information from M. Rhodes.

⁸⁹ The letter B was originally cut as an R.

⁹⁰ By the late Mr A. Brooks and his wife. The hoard was declared Treasure Trove in February 1981 and is now in the British Museum. We are grateful to Dr T. W. Potter and Miss C. M. Johns for providing full details and making the objects available for study. For the rest of the contents of the treasure see above p. 347.

⁹¹ Compare the two spoons from Mildenhall inscribed in the bowl PAPITTEDO and PASIENTIA JRS xxxvii (1947), 181 Nos. 11(a) and (b).Google Scholar

⁹² For the references see E. de Ruggiero, Dizionario Epigrafico di Antichita Romane, s.v. Faunus.

⁹³ cf. P. Lambrechts, Contributions a I'Etude des Divinites Celtiques, (1942), 123.

⁹⁴ Note, however, the colour-coated vessels made at Colchester showing scenes of curious figures harvesting grapes, or of diminutive hooded figures engaged in conflict. M. R. Hull, The Roman Potters' Kilns of Colchester (1963), fig. 53, 13 and 9.

⁹⁵ Compare the Collegium dei Silvani, No. 36 above. The worship of Faunus in the late Roman period is implied by the continued observance of the festival of the Lupercalia at Rome on the 15 February. It is mentioned in both the Philocalus Calendar A.D. 354, and the calendar of Polemius Silvius A.D. 448/9. Pope Gelasius, A.D. 492–496, wrote a tract against it, Adversum Andromachum etc. For the connection between the Lupercalia and Faunus see G. Pomares (ed.), Gelase I^et. Lettre contre les Lupercales et dix-huit messes du sacramentaire Léonien (1959), 31 ff.

The question arises whether the names on the spoons which seem particularly appropriate to devotees of Faunus or the Fauni, particularly Agrestius, Ingenua, Primigenia and Silviola (see Notes 102, 104, 105 and 106 on these names) were not the actual names of the users but nicknames (signa), used particularly at club banquets. One objection to this, the commonly held view that members of a club all used the same signum, has been disposed of by I. Kajanto, Supernomina (1966), 50–2. In favour of the idea is the use of the word vivas, acclamations of this type being commonly found with signa when detached from the personal names of their bearers (cf. RIB 812, the acclamation on the side of an altar set up by the tribune C. Cornelius Peregrinus at Maryport, Volanti vivas, ‘Long life to you, Volantius!’ – the only example of a signum certainly attested from Roman Britain). One of the names, Auspicius, is actually used as a signum, CIL viii 16406.Google Scholar

⁹⁶ For permission to hold monthly meetings granted to military collegia see Dig. 47.22.1 and for two silver spoons with figured scenes in gilt inscribed Eusebiorum dignitas found near Aquileia, perhaps, but not necessarily belonging to a club, the Eusebii, see CIL v 8122.10 (see Kajanto op. cit. (note 95) 48). The date of the Faunalia, the nones of December, is given by Horace, Odes iii. 18. 10.

⁹⁷ The number in brackets following each entry refers to the catalogue in the forthcoming publication on the Thetford Treasure by Dr Potter and Miss Johns. This publication will also include a discussion on the Celtic epithets by Professor Kenneth Jackson.

⁹⁸ It is tempting to see GRA, as an abbreviation for GRANVS, a variant of CRANVS, cf. the abbreviation AN., BLO. and MED. for Andicrose, Blotugus and Medigenus. If this is correct then Grannus, the Rheinland god equated with Apollo, would be a possible parallel. For Apollo Grannus see A. Holder Alt-celtischer Sprachschatz.

⁹⁹ More usually spelt Medugenus, see the following item. For an example from Lusitania of this as a personal name, see CIL ii 162.

¹⁰⁰ For examples of the personal name see Holder, op. cit. (note 11). A Dea Naria is also attested CIL xiii 5096.Google Scholar

¹⁰¹ Saternio for Saturnio, appropriate since Saturn was the father of Picus the father of Faunus, unless this is really a Celtic descriptive epithet.

¹⁰² AGRESTE probably a mistake for AGRESTI, the vocative of Agrestius, rather than, as it formally appears, the vocative of the unattested name Agrestus. Kajanto, The Latin Cognomina, cites three examples only of the name Agrestius, two dated A.D. 384 and 433. The suffix -ius is typical of late Roman cognomina. The name, from agrestis, ‘countryman’, ‘rustic’, is appropriate for a devotee of the fauns, cf. Virgil, Georgics i, 10, vos, agrestum praesentia numina, Fauni, and Ovid, Fast1. iii, 315, (Fauni) di summi agrestes.

¹⁰³ Kajanto, op. cit. (note 102) gives four examples only of the cognomen Auspicius/a, one strictly a signum. Like the preceding example it has the -ius suffix typical of late cognomina.

¹⁰⁴ Ingenuus/a is a common cognomen probably because of its secondary meaning of ‘free born’. Its primary meaning, ‘native’, ‘indigenous’, would again be appropriate in the context of the fauni, who, with the nymphs, are described by Virgil as indigeni (Aeneid viii 314).

¹⁰⁵ Primigenius/a, a common cognomen, probably often used of first-born children, but see the note on Ingenua above.

¹⁰⁶ Silviola: Kajanto, op. cit. (note 102), 168, refers only to a single Christian inscription from Rome. It is a diminutive of Silvius/a, a name from the legendary past of Latium derived from silva, a forest, and on both counts again suitable in the context of Faunus (described as silvicola, ‘inhabiting woods’ in Virgil Aeneid x, 551). The cross following the name is noteworthy.

¹⁰⁷ Restitutus/a is a common cognomen. For an example from Britain, a graffito, see CIL vii 1336, 1276.Google Scholar

¹⁰⁸ cf. CIL xiii 3262 = 10027.72, perseverate on a bronze vessel with a scene of (named) gladiators.

¹⁰⁹ A variant, using an infinitive, uti, instead of imperative utere, of the common motto utere felix, see JRS xlvii (1957), 232Google Scholar No. 25, for VTI FEL on the bronze handle of an iron scraper from East Winterslow, Wilts.

¹¹⁰ cf. VIRIBONISM on a silver spoon from Canterbury JRS liii (1963), 163 No. 20, there interpreted as the genitive of a personal name, Viribo, followed by M, but in view of the present example probably to be interpreted as viri boni s(u)m, ‘I belong to a good man’. Vir bonus is a standard expression for an upright man.

¹¹¹ Martin Henig refers us to H. B. Walters, Catalogue of Gems and Cameos in the British Museum (1926), No. 1759 and pl. xxiii, for a close parallel cut on a plasma and for a discussion of the type, derived from the statue by Eutychides of Sikyon (cf. Pausanias vi, 2, 7), to Gesztelyi, T., ‘Die Tyche von Antiochia auf einer Gemme von Debrecen’, Ziva Antika xxv (1975), 274–84.Google Scholar

¹¹² For two examples from Britain see G. C. Boon, Silchester: the Roman Town of Calleva (1974), 171–2 and fig. 24, 1 and Henig, M., ‘The Aesica amulet and its significance’, Arch. Ael.⁴ l (1972), 282–7Google Scholar, pl. xxiii, fig. 1. Henig cites Goodenough, E. R., Jewish Symbols in the Greco-Roman Period 11 (1953), 250–1Google Scholar, who argues that ‘Iao’ was the name of the figure.

¹¹³ Excavations by the Norfolk Archaeological Unit directed by A. K. Gregory who sent drawings and full details.

¹¹⁴ During excavation for the Vindolanda Trust directed by Mr R. E. Birley, who made it and the next three items available. Graffiti on ten more sherds of samian and coarse ware were examined, but were too fragmentary for inclusion here.

¹¹⁵ This seems to be its first occurrence in Britain, but the derived nomen Attonius is known (RIB 768, 1024); for the distribution of Atto and related names in Att- in Belgica and the two Germanies, see Alföldy, G. in Epigraphische Studien 4 (1967), 10–16.Google Scholar

¹¹⁶ Capital letters, but the E has been represented, as sometimes in cursive, by a thick vertical from which a thin diagonal stroke diverges to the right; N has two diagonal strokes; and only the top of the s survives. The name is already attested in Britain as a maker of mortaria at Brockley Hill, Middlx. (JRS xlvi (1956), 149, No. 13)Google Scholar, cf. Britannia vi (1975), 286, No. 10, Matugena.Google Scholar

¹¹⁷ Presumably Felix or a cognate personal name (e.g. Felicianus), but possibly feliciter, utere fetix, etc.

¹¹⁸ During excavation directed for the Department of the Environment by Mr M. Savage. It seems to be the first such sealing to be found in a milecastle. Miss L. Allason-Jones made it available at the Museum of Antiquities, Newcastle, where it now is.

¹¹⁹ Obv. C(ohors) (Prima) T(hracum) or T(ungrorum). Rev. probably Ae(lius) Sec(undus), but there are other possibilities. Both cohorts are attested in Britain from at least 122 (CIL xvi 69).Google Scholar During 205/7 Coh. I Thracum was at Bowes (RIB 740), of which it was evidently the garrison (cf. RIB 730, 733, 734); at the same time a Coh. I Thracum c.R. was active at Birdoswald (RIB 1909). RIB 1323 attests activity near Newcastle, but is not evidence that it was ever in garrison there. Coh. I Tungrorum was the third- and fourth-century garrison of Housesteads less than two miles from Sewingshields, which is inconclusive, since goods carrying lead sealings usually travelled considerable distances. The date of this sealing is in any case unknown.

¹²⁰ During excavation for the York Archaeological Trust directed by Mr R. A. Hall.

¹²¹ LAT could be read, at least in part, as a numeral; but this seems less likely, even if the restoration late(res) (‘bricks’) can only be a guess. The s is missing the second stroke that would make the reading certain. A diagonal ‘stroke’, perhaps due to a flaw in the fabric, has obscured the final L.

¹²² By Mr G. Bevan, 22 Burton Drive, Wrexham, who made it available. It remains in his possession.

¹²³ If the reading is indeed Coccei followed by a cognomen, this would seem to be a tile-maker's signature; but although the first letter in l. 2 with its long descender could be M, the diagonal scratches which follow do not support a reading MANV. RSOT.

¹²⁴ Information on this and the following four items from the Fortress Baths from D. Zienkiewicz who directed excavations for the Welsh Office, Ancient Monuments Branch.

¹²⁵ Though written in Latin letters, this is probably a Greek personal name beginning with the prefix EU-.

¹²⁶ The foot of the F is provided with a serif and Mr G. C. Boon who has discussed the graffito with us, has suggested it may be intended for an E in which case F]elicii would be a possible reading. For the nomen Felicius, formed from the cognomen Felix, see RIB 690 York.

¹²⁷ Up to five letters are missing and there are several possible restorations of the name.

¹²⁸ The upper diagonal stroke of the K has been omitted. This is only the third certain example of a date on tile from Roman Britain, the other two being from London (Britannia xi (1980), 413–14Google Scholar, No. 53 and Silchester, EE ix, 1294.Google Scholar

¹²⁹ They were noticed by Mr R. A. H. Farrar of the Royal Commission on Historical Monuments, who sent details including photographs, RSOT has visited the site.