² The line-drawing was made in the same way as those of the Bath ‘curse tablets’ (cited here as Tab. Sulis). with the help of an enlarged photograph and with constant reference to the original. The photograph is published for comparison. It will be appreciated that no photograph, however good, can catch all the lines on a ‘curse tablet’. See further Tab. Sulis, p. 85.

³ At the sale by auction of the effects of a local collector deceased. Language and handwriting are easily paralleled at Bath and Uley, but not the four nail-holes and absence of folding. If it was intended for display, this would be unusual in Britain (see Tab. Sulis 15. with note, and compare RIB 7 and 8), most tablets being thrown into water or rolled up before burial. The find-spot was evidently a temple of Mercury, but not one already known to have produced ‘curse tablets’ (Uley. and probably Pagans Hill); the patination. which is very light, is in any case different. The present owner, who bought it from the original purchaser, made the tablet available.

⁴ For this rather ‘clerical’ style of writing, often found in British ‘curse tablets’, see Tab. Sulis. pp. 84-93.

⁵ Commentary

The left margin survives of 4-8 and the right margin of 3-6. making it possible to estimate how many letters have been lost elsewhere.

1. deo (in enlarged letters) guarantees the restoration of Merc[urio] and also determines the end of the previous word. For donatur compare Tab. Sulis 16.5: and for the formula of ‘giving’ the thief to the god. compare ibid., pp. 63-4. 70. The R is peculiar: now damaged, it seems to have been of capital form (compare Tab. Sulis 76).

1-2. [si] q[u]is involaverit: the tail of Q survives under 3. For this formula compare Tab. Sulis 63.2 and pp. 66 and 69-70.

c…lam: evidently the object stolen. If res [p]ictor[i]a[s] (9) alludes to it. perhaps it was ‘writing things’, but calamum (‘pen’) cannot be read. Traces of the four missing letters suggest ARAM, but there are other possibilities. There is no obvious restoration.

3. [.]icinnum: the context suggests that this is a second object stolen, but there seems to be no word of this form. Possibly [r]icinium was intended, a hood worn by women: see Tab. Sulis, p. 192, note s.v. mafortium. Caracallam (‘hooded cloak’), however, cannot be read in 2.

nee non alia minutalia: this long-winded equivalent of ‘etc’ does not occur in any other ‘curse tablet’. Its sense is the same as et resculas plurimas in a list of stolen property in an unpublished text from Uley (interim report. No. 72).

4. Tocitami(?): the first letter is partly lost, but what survives suggests an enlarged initial T. The last two letters could also be read as RVS. but the scribe seems to have lifted his point for the ascender of M (compare M in curam). producing a three-stroke M like that in some stilus-tablet texts. Tocitami would be the owner's name in the genitive case. No such name is attested, but for the first element compare Togodumnus and Tocca > Toccius; the suffix -tamus is also a Celtic name-element.

4-5. si baro … si servus: for these formulas, mutually exclusive alternatives defining the victim of the curse, see Tab. Sulis, pp. 98-9. The usual sequence si puer si puella has been reversed, perhaps for the sake of chiasmus. (The descender of the first L can be seen to the right of the hole at the end of 6.) Si ingenuus si liber is a previously unattested variant of the frequent si servus si liber.

5-6. n[o]n an[t]e eum laset quam: O can be restored in the worn space between the N s (for the spacing, compare non (3)); for the formula, compare Britannia x (1979). 343Google Scholar. No. 3 (Uley). non ante laxetur. (For laxet > laset, see Colin Smith, ‘Vulgar Latin in Roman Britain’. ANRW 11.29.919.) One would have expected a reference to the return of stolen property (‘no peace until …’), but instead another curse apparently follows.

6-7. mimbra (?ra)pi manu di\em mortis: it is hard to see what to make of this. Mimbra may be for membra (‘limbs’ cf. manu, ‘hand’), a seribal error, since a ‘Vulgar’ change of stressed e to i is unlikely: if so. the scribe may have intended a catalogue of parts affected by the curse (‘limbs’ and ‘hands’). But even if (ra)pi is correctly restored - the scribe having omitted RA by haplography - a passive infinitive (‘to be torn’) has no obvious place here. Nor is there obvious sense to be made of manu (‘by hand’); perhaps manu(s) (‘hands’) was intended, by analogy with mimbralmembra. Finally diem mortis (‘day of death’): perhaps (ad) diem mortis was intended, ‘until the day of his death’. The scribe seems to have been copying formulas ungrammatically and without comprehension.

7. concrutiat: for concruciat, like the ‘Vulgar’ condicio > conditio, although this seems to be the first instance from Britain. The verb is a variant of the usual excruciat which is otherwise found only in Lucretius III, 143. For another such almost unique variant, presumably intended for literary effect, see _excomesis in Tab. Sulis 97. 6. The subjunctive might have been expected (compare laset (6)); the apparent indicative may be due to a ‘Vulgar’ confusion between a Ist-conjugation verb in -io (-iare) and the 4th-conjugation -io (-ire).

qu(q[u])[i]: QV seems to have been repeated, as in 9. (The tail of the second Q can be seen.) This was presumably a transcription error, rather than a blundered quicumque or similar, and makes it easier to posit other errors in 6-7 etc.

7-8. se[curam: the reading of CVRAM is certain, and there is sufficient trace of SE. It does not seem to be an adjective, as there is no noun for it to qualify; since it is apparently the object of involavit (8). it would seem to be something stolen, pehaps _{securam < securim} (‘axe’), by a ‘Vulgar’ reduction of a 4th-declension noun to the 1st.

8. [.]nnoris: if securam is correctly understood as ‘axe’, this must be the owner's name, although it is puzzling that there has been no previous mention of the item or its owner. No obvious restoration presents itself, and it is possible that the scribe wrote [lu]nnoris in error for Iunioris (the cognomen Iunior being fairly common); another instance of NN for NI may occur in 3 (ricinium?).

8-9. After involavit there are only uncertain traces. The first five letters of 9 are well preserved, although PR (the likeliest reading) is oddly made; nec which follows means that PR must be a word-ending. The context does not help.

9. qu(qu)i: QV has been repeated, a transcription error presumably, as in 7 res [p]ictor[i]a[s]: there is possible trace of the initial P (rather than V for a personal name), and sufficient trace of A, but the meaning is obscure. Picta is used in the sense of ‘written’ in Tab. Sulis 8. 6 (see note ad loc.), and it may be that the property stolen here was not ‘painting things’ but ‘writing things’. ‘Curse tablet’ texts sometimes conclude with a general reference to stolen property already specified (e.g. istas res in Tab. Sulis 32. 15), but this cannot be assumed here.

₆ By the local farmer, the late George Greggdins, who gave it to the Netherhall Collection in the Senhouse Museum, The Battery, Maryport. The Hon. Curator, Lt. Cdr. Brian Ashmore, made it available.

⁷ For another example of the same pose, see Britannia vii (1976), pl. XXXVIII BGoogle Scholar.

⁸ There is a shallow vertical stroke before IIX, which does not seem to be deliberate. The bottom stroke of E is incomplete: F could be read, but is excluded by the context. Before it only the bottom half of a vertical stroke with serif survives, appropriate to H. I or N. The format of the stone suggests a record of building work by a military unit, but too little survives for confident restoration. IIX is presumably a numeral or part of one, perhaps preceded by P(edes) (compare RIB 1820. etc.). ER is almost certainly the end of an abbreviated word, one possibility being Ner(viorum) (compare RIB 418. CHO I NER, and 1683. COH II NER). Cohors III Nerviorum is recorded by the Notitia as the garrison of Maryport (ND Occ. 40.53).

⁹ With the next two items during excavation by Carlisle Archaeological Unit directed by Ian Caruana, who made them available. See Britannia xxi (1990), 320Google Scholar.

¹⁰ The first letter, which resembles the tail of L, is cut by the edge of the stone; this and the manner of incision suggest that the inscription was made at the quarry-face.

¹¹ The thinness of the slab suggests that it may have been a ‘centurial stone’ with a scratched inscription, now lost, for insertion into turf-work. This was presumably one of the fort ramparts on the site (see Britannia xxi (1990), 321. fig. 10)Google Scholar.

¹² The marks can be read horizontally or vertically. They resemble mason's marks, but this is not an ashlar; perhaps it identified a quarry-batch, or was intended (like the previous item) for insertion into turf-work.

¹³ By Alan Rae during excavation directed by Tony Wilmott for Cumbria County Council.

¹⁴ The letters are of cursive form (note the A and G. and the ligaturing of C and G with the succeeding letter). The centurion's name could be read as CIAGISI, but the initial sequence CIA- is un-Roman and in any case very rare. It seems better, thèrefore, to read CLA- by understanding the second letter as a cursive ‘long’ L (see A.K. Bowman and J.D. Thomas. Vindolanda: the Latin Writing-Tablets (1983), 64). Claudius is usually abbreviated to CL in RIB, and the cognomen Agisus is locally attested in Noricum (CIL iii 5542) and Pannonia (CIL iii 10883). There is a unique Thracian name Clagissa is unknown.

¹⁵ By Tony Wilmott during excavation directed by him for Cumbria County Council. The stone remains in position.

¹⁶ Too little survives of the irregular letters in line I for identification; CH[…] for C(o)h[otrtis…)] and LI for L(egionis) I[I Aug(ustae)] are both possibilities. The centurion's name could be read as COISE[…], but no such name is known and the sequence of letters is un-Latin. Instead of IS, therefore, it is better to read a cursive V (note the cursive form of E) and to suppose the centurion bore a name of Celtic etymology cognate with Coventina: compare CIL xiii 6028, Iulius Coventi (filius).

¹⁷ Information from Ian Caruana. For similar stones, usually thought to mark quarry-batches, see Britannia xxi (1990), 366Google Scholar, No. 5 with note 8. It closely resembles Britannia xix (1988), 495 with pl. XXXII BGoogle Scholar.

¹⁸ During excavation by Carlisle Archaeological Unit under the general direction of Mike McCarthy: see Britannia xiii (1982). 82. Tim Padley made it availableGoogle Scholar.

¹⁹ The ‘obverse’ is so called because it was the natural face to inscribe first, with the attachment hole out of the way in the bottom left-hand corner. When the tag was re-used, another text was inscribed on top. It is now difficult to distinguish the two (the distinction is exaggerated by the line-drawing, FIG. 4), but the different sequences of letters can be recognized, and the secondary text was incised more sharply and deeply; it also turns out to be the same as the reverse, though not apparently by the same hand. The second line of the reverse is illegible; it may have included a numeral.

The purpose of such tags is obscure: see Britannia xx (1989), 334–5Google Scholar, note 17. Primary and secondary texts were evidently a personal name in the genitive case, and must have identified someone's property. The secondary name. Luciavus. seems to be unattested. (Lucianus cannot be read.) It can be understood as one of the many cognates of Lucius popular in Britain and Gaul because they incorporated a Celtic name-element.

²⁰ With Nos 11-18, 21-23, 26-27 and 29, below, during excavation by Carlisle Archaeological Unit directed by Ian Caruana, who made them available.

²¹ For other Sealings of the legion see RIB 2411.42-67. The I on the reverse has a slight rightward extension, but not really enough for L (and Fl(avius)). This reverse is presumably an abbreviated nomen and cognomen. Fi(…) Flo(rentious) or similar.

²² When the centurial symbol ‘7’ occurs on the reverse of a legionary or cohortal sealing, with the possible exception of RIB 2411.162-7, it follows the abbreviated name of the centurion. The symbol should be understood as ‘centurion’, not ‘century’, since similar abbreviated names (of centurions, hardly of centuries) sometimes occur on reverses without the symbol. In particular, RIB 2411.120 and 121, Aet(i) Comini, is clearly equivalent to 122. Comini 7. In the documents collected by R.O. Fink, Roman Military Records on Papyrus (1971), it will be seen that where ‘7’ precedes the name it means ‘century’, but where it follows the name it means ‘centurion’. That is the usage here.

The centurion's name is in the genitive case. i.e. ‘(sealing) of so-and-so, centurion’, like the genitive applied to the witnesses' names in a Roman legal document. The case is explicit only in RIB 2411.122; and in 98 with Addendum et Corrigendum (f) below, and in Britannia xix (1988). 499. No. 5. Optati D(…) 7, which imports confusion by reversing the order of nomen and cognomenGoogle Scholar.

²³ For other sealings of the legion see RIB 2411.68-75 and Addendum et Corrigendum (f) below.

²⁴ For another sealing of this unit, the governor's mounted bodyguard, see RIB 2411.91.

²⁵ The die seems to be too small to restore [LX]X, [L(egionis) X]X.

²⁶ The meaning of S is unclear. It cannot be s(emis), ‘half (one uncia)’, since the weight is almost the theoretical weight of one uncia (27.125 gm).

²⁷ In theory 109.15 gm. The small discrepancy can be explained by surface corrosion, but is quite typical: compare RIB 2412.74-7.

²⁸ During excavation by Carlisle Archaeological Unit directed by Mike McCarthy, now published in M.R. McCarthy, A Roman, Anglian and Medieval Site at Blackfriars Street, Carlisle (1990). This and the next item are fully published by Ian Caruana, ibid., 135, Nos 97 and 98, with fig. 119.

²⁹ This is the first instance of this Capuan bronzesmith from Britain, but Ansius Epaphroditus is already well attested (RIB 2415.5-7).

³⁰ See above, note 28.

³¹ The stamp is otherwise unattested, and it is not certain that it is complete and that the medial points are associated with the letters, although both seem likely. If so, the maker's name was perhaps G(aius) I(ulius) A(…).

³² The maker's name; probably Auctus. Compare RIB 2441.6, a bone comb stamped DIGNVS.

³³ During excavation by Carlisle Archaeological Unit before a lift shaft was installed. The director of the excavation. Ian Caruana, made it available. Four of the fragments, including this one, carried traces of cursive writing on the inner (recessed) face. On three only isolated letters were legible, but the fourth fragment, part of a loan agreement dated 7 November 83 between two soldiers of the Twentieth Legion, will be published in next year's Britannia.

³⁴ Britannia xix (1988), 496, Nos 31-35Google Scholar. The writing instrument was probably a metal nib loaded with ink, leaving a series of indentations across the grain of the wood linked by discolouration due to bruising, which have been represented in the line-drawing (FIG. 5) as continuous lines. Horizontal strokes made with the grain tend to be lost.

³⁵ Something may have been lost of both Ls in the horizontal grain, but enough remains of the curve to guarantee the reading. (They were the cursive ‘long’ L (see above, note 14), not the form with a descender.) There is possible trace (not drawn) of the second stroke of P in the horizontal split running across the tablet, and the space before L also implies it. I after S is certain, its cramped position probably due to the binding cord being in the way.

The cognomen Albanus also occurs as a witness' name in a text from Castle Street (Britannia xix (1988), 496, No. 30)Google Scholar. Albanus is the first instance of a seplasiarius from Britain, but the term is well attested in Gaul and the Rhineland: ILS 7606 (Cologne), negotiator seplasiarius; ILS 9310 (Paris); CIL xii 5974 (Narbonne); xiii 6778 (Mainz), seplasiar(ius) in leg(ione) I Ad(iutrice) (and thus of Flavian date). Etymologically it derives from Seplasia, part of Capua known for its unguentarii (Lewis and Short, Latin Dictionary s.v.), which is why OLD like other dictionaries translates it as ‘dealer in ointments’. Ancient glosses suggest a wider application by identifying it with the Greek term παντοποώλης (‘general dealer’): seplassarius (sic), negotiator qui multa venundat, quem Graeci pantopoles vocant (CGL v 244, 13. compare ii 182, 20; 393, 48-9). There seem to be only two references, both late, to what one actually bought from a seplasiarius. (1) Vegetius, , de mulomedicina iv. 3–6Google Scholar, panacem a seplasiariis comparas (a medicinal herb). (2) A sub-Roman sermon on the issue of blood (Mark v. 25ff.) published by Turner, C.H. in JTS xxi (1920), 3 ff.Google Scholar, voices the woman's despair of human medical prescriptions: ut quid mihi dictata inaniter quaerere pigmenta, ire per simplassarios (sic)? These references suggest that seplasiarius was a specific term, and it is tentatively translated here as ‘pharmacist’ leaving unresolved the question of whether Albanus was a private individual (compare ILS 7606 above, negotiator seplasiarius) or was serving in a military hospital (compare CIL. xiii 6778 above, seplasiar(ius) in leg(ione) I Ad(iutrice), like this tablet a Flavian text). In view of the legionary presence (see above, note 33), the latter is an attractive possibility. The cognomen Albanus is common, but of the two instances in RIB, one (RIB 524) can be attRIButed to the Twentieth Legion in the Flavian period. The ‘mason's tools roughly outlined’ on his tombstone could be anything; they suggest only that he was a specialist of some kind.

³⁶ During excavation by Carlisle Archaeological Unit directed by John Timperley, under the general direction of Mike McCarthy. For related excavations see Britannia xiii (1982). 79–89Google Scholar. These are not the tablets descRIBed on pp. 83-4.

³⁷ Compare the previous item, and note 34, above. The line-drawing (FIG. 6) represents only the indentations left across the grain of the wood.

³⁸ This may be the last word of the ‘address’, compare Britannia xix (1988). 497, No. 31Google Scholar. Rigianus is the cognomen of the addressee. Only R and G are really certain, and one might have expected REGINO, but there is no trace of II = E (nor space for it) and there is. in spite of damage, sufficient trace of the A before N. The cognomen Rigianus seems to be unrecorded, but it must be related to the unique nomen Rigius (CIL xii 5091) and the unique masculine personal name Rigia (CIL v 4594). They belong to a group of names of common Celtic etymology best represented by the cognomen Reginus (see D. Ellis Evans. Gaulish Personal Names (1967). 373. s.v. Regenos). In Britain this probably occurs in ECMW No. 141 and, as Regina, in RIB 1065 (a Catuvellaunian). The British tribal name recorded by Ptolemy (ii. 3, 13) as Regni or Rigni must also belong to this group: see Rivet and Smith, PNRB, 445-6, s.v. Regini.

³⁹ In the same context as No. 24, above.

⁴⁰ See RSOT. ‘The Roman ‘carrot’ amphora and its Egyptian provenance’. Journal of Egyptian Archaeology lxxviii (1992)Google Scholar. forthcoming.

⁴¹ By the dealer from whom the Ashmolean Museum bought it in 1991. Michael Vickers provided a photograph and made the object available to RSOT.

⁴² The legend may be incomplete, since in antiquity the die was trimmed to its present size, probably with shears. It was presumably intended for official use, but its purpose is unknown. The metal would be unsuitable as a brand (e.g for wood or leather) and must have been intended for impressing on something softer than itself, e.g. clay or lead. There is no sign of any attachment-ring on the back, or mark of hammering, so perhaps it was once mounted in a metal or wooden matrix. There is no close parallel in RIB 2409 (dies), but one possibility is that the die was intended for lead ingots from the Derbyshire Field centred on Lutudarum (see note to RIB 2404.39). Lead ingots from the Mendips Field carry incuse legends (see RIB 2404.2-13), although none is closely comparable; compare also RIB 2405.1, a tin ingot stamped DD NN, d(ominorum) n(ostrorum), and RIB 2409.26, an iron die reading MPBR. presumably m(etalla) p(rovinciae) Br(itanniae).

⁴³ During excavation for the Trust for Wessex Archaeology directed by Davies, Susan and Farwell, D. (see Britannia xxi (1990). 350–2)Google Scholar. Dr A.P. Fitzpatrick made this and the next item available to RSOT.

⁴⁴ The medial point(?) does not seem to be a letter. On the reverse are two pairs of scratched lines intersecting at right-angles; and superimposed on them two triangular figures resembling two forms of ‘A’, not necessarily letters at all.

⁴⁵ By the divers Bob Middlemass and Rolfe Mitchinson (for whom see Britannia xx (1989). 337. note 29)Google Scholar. Mr Mitchinson made it available, with another sealing from the same site; this bears the impress of a rectangular die. obverse and reverse, now illegible.

⁴⁶ Barred B (understood here as ligatured BF) as a symbol for beneficiarius occurs elsewhere in Britain only in Britannia xix (1988). 490Google Scholar. No. 4 (see ibid., note 11). The same obverse and reverse legends, but from different dies, are found in RIB 2411.246 and 267. whose obverses can now be read as BFC and [B]FC respectively.

⁴⁷ During excavation directed by M.C. Bishop, who made it available.

⁴⁸ The stamp is not recorded, and it is just possible the letters are a graffito made before firing. Their meaning really lies beyond conjecture, but the stamp NCON is found at Binchester (Britannia ix (1978), 477. No. 25)Google Scholar, perhaps N(umerus) Con(cangensium); if so, symmetry might suggest NV[IN], N(umerus) V[in(oviensium)] here.

⁴⁹ During excavation by the Central Excavation Unit directed by Fachtna McAvoy. Jan Summerfield made the bowl available to RSOT.

⁵⁰ The second stroke was made at a slight diagonal, like the first stroke of the V, as if the writer intended V but incorporated the following stroke into L instead. VL would give lulius, which occurs as a personal name without cognomen (e.g. Britannia xxi (1990), 365, No. 4Google Scholar and RIB 1950). Elius could be a ‘Vulgar’ spelling of Aelius, but it is questionable whether this imperial nomen gentilicium was used on its own as a personal name. The other possibility is the Greek name Helius (compare RIB 251) unaspirated, which is occasionally found: see H. Solin. Die Griechischen Personennamen in Rom: ein Namenbuch (1982), 368-70.

⁵¹ By A. Brading on behalf of the Isle of Wight Archaeological Service. Dr D.J. Tomalin made it available to RSOT. Despite its provenance (see Falsa, below), the graffito is almost certainly genuine. It looks rather fresh, but the letter forms are Roman, and in manner and content it is quite unlike the modern forgeries found here.

⁵² Presumably an abbreviated personal name, such as Lucius and its cognates. These would be of Roman appearance but Celtic etymology; however, purely Roman names like Lucretius are also possible.

⁵³ By A. Yule on behalf of the Isle of Wight Archaeological Service. Dr D.J. Tomalin made it available to RSOT. Despite its provenance (see Falsa, below), the graffito is genuine; but it was mis-read as RR because of acasual scratch, and prompted modern forgeries from here.

⁵⁴ During excavation in Marvel Lane directed by Dr D.J. Tomalin, who made it available to RSOT. It has been published, with the reading ALANCCA (see next note), in D.J. Tomalin, Roman Wight (1987), 34, with pl. 3.

⁵⁵ Given the close spacing of the letters, enough survives to left and right to indicate that the graffito is complete. T has been ligatured to V, N being precluded both by its unlikely form and by the sequence -NCCA- that would result; but the reading ALANCCA was followed by modern forgeries from Fishbourne (see Falsa, below). Alatucca is presumably a personal name hitherto unattested, cognate with the Celtic name Tucca (whence the nomen Tuccius); compare also Paltucca at Bath (Tab. Sulis 98. 21). An ala would be an unlikely subject for a graffito, and there is no reason to seek it here.

⁵⁶ By a visitor from Northern Ireland ‘while walking near High Rochester’, in whose possession the sherd remains. Details from Robin Birley. who sent a drawing.

⁵⁷ This personal name occurs as the name of a samian potter (see Oswald. Index s.v.), but the derived nomen Atinius is more common; compare Britannia xv (1984). 348Google Scholar. No. 62 (Newstead), t(urma) Atinia. Arguably it is cognate with the Alt- names widely attested in Belgica and the Germanies (collected by Alföldy, G. in Epigraphische Studien 4 (1967), 10–16)Google Scholar which presumably incorporate a Celtic name-element.

⁵⁸ With the next eight items, during excavation directed by M. Cooper, Birmingham University Field Archaeology Unit, and Dr A.S. Esmonde Cleary. Birmingham University Department of Ancient History and Archaeology. Iain Ferris made the items available to RSOT. who will publish them and some slighter graffiti with line-drawings in the final report. For other graffiti from Rocester, see Britannia xvii (1986), 448–9. Nos. 70-74; and below. Addendum et Corrigendum (j)Google Scholar.

⁵⁹ A signature, presumably the mould-maker's. The reading is by Brenda Dickinson and Brian Hartley; Peter Webster contributed to this note. The name is not recorded at La Graufesenque, although it is fairly common in Narbonensis and Aquitania, and Allard Mees has noted a (different?) instance from Le Rozier. The script is anomalous, no doubt because the writer lacked formal training. [P]H are capitals, but the remaining letters are only about half the height. H and V resemble the cursive forms in contemporary stilus tablets, but B resembles the much later cursive B in New Roman Cursive. There is a downstroke with a hook at the bottom; a loop; and above it, possible trace of a third stroke. It would seem to be a simplified capital-letter form: compare R. Marichal, Les Graffites de La Graufesenque (1988), 25, fig. B 4.

⁶⁰ L is at an unusual angle, and was scratched without lifting the instrument. A distorted C is just possible, but the combination BC is not likely. BL would be an abbreviated personal name; there are various possibilities, but the most likely are Blaesus or Blandus (and its cognates).

⁶¹ A personal name Diatus is not recorded either by Kajanto, Cognomina, or Mócsy, Nomenclator, or TLL, Onomasticon. It may be identical with the cognomen DIATVA recorded by an early drawing of the lost RIB 888 (see ad loc), particularly if S was mis-read as A. Compare also the name DIANTVS inscribed on a samian dish from Ospringe: see Research Reports Soc. Ant. viii, Ospringe (1931), 67–8Google Scholar with pl. LIV.

⁶² Presumably the genitive of a personal name in -acus or -acius. It is not possible to tell whether E was the first letter or not. Mocsy, Nomenclator, does not note any name Elsac(i)us or -elsac(i)us.

⁶³ The second A is oddly placed, with a space after it, and an apparent space before it. which may have been filled by a V just above the line of the other letters. It so. the scribe must have repeated VA in error, since the name lanuarius can hardly be doubted. It is a common cognomen already attested at Rocester (Britannia xvii (1986), 448, No. 70)Google Scholar. The penultimate letter, R. is of cursive form and distinguished in this hand from A and N by its long diagonal descender.

⁶⁴ Unless the letters were oddly spaced, the graffito is in fact complete, and thus would have been intended to be the first two letters of the owner's name. There are many possibilities.

⁶⁵ Similis, ‘like’ (his father) and thus a delicate compliment to the mother, as A.R. Burn puts it (The Romans in Britain (1969), 35). is a common personal name in Britain and Gaul.

⁶⁶ Nothing survives of the tail of the R, but it is implied by the space before the succeeding H. The otherpossibilities. B or P, can be excluded because of the rarity of the sequence VEB or BEP. and because Verecundus is such a common cognomen. Like Venrus it is good Latin, but notably frequent in Britain, Gaul and the Rhineland. so that in common with other Ver- names it probably contained a Celtic name-element: see Alföldy, G. in Epigraphische Studien 4 (1967). 16–19Google Scholar.

⁶⁷ Information from Dr P.H. Robinson, curator of Devizes Museum, who provided a drawing. The ring remains in the finder's possession. Dr Martin Henig brought it to the notice of RSOT.

⁶⁸ The ring is the latest member of the group noted in Britannia ii (1971). 300Google Scholar. No. 64 with note 62. from Caistor St Edmund, Yatton (Henley Wood) and Owslebury. now published as RIB 2422.53-6. See note to RIB 2422.53.All five rings are of identical form, and each bears a different permutation of the same seven letters, with an eighth ‘letter’ as a spacer.

⁶⁹ During the excavations directed by (Professor) Wacher, J.S. for the Ministry of Works (see JRS I (1960), 217–8)Google Scholar and the Department of the Environment (see Britannia iv (1973), 279–80)Google Scholar. The items were made available by Peter Wilson of the Central Excavation Unit. For other graffiti from the same excavations, see Britannia xxi (1990), 372–6, Nos 40-67Google Scholar. They will all be published in the forthcoming final report on the excavations, to which reference should be made for line-drawings.

⁷⁰ C is of unusual form, the downstroke not curving to the right. For the cognomen Calamus (‘(reed) pen’), see Kajanto, Cognomina. 342.

⁷¹ Only the bottom survives of the first two letters, but although LS or LO would be possible reading. Mocsy. Nomenclator, records no names of this form, whereas Nestor is well attested: sec H. Solin. Die Griechischen Personennamen in Rom: ein Namenbuch (1982). 503-4.

⁷² An abbreviated personal name, perhaps Velox (compare Britannia xxi (1990), 374, No. 60)Google Scholar. but there are other possibilites like Velius and rare names of Celtic etymology.

⁷³ The surviving letters are all incomplete, and too irregular to be restored with confidence. The concluding HS, unless it is a numeral, suggests a name of Greek etymology in -es.

⁷⁴ The graffito is complete, presumably the female name Iunia rather than an abbreviated Iunianus. The only other possible female name from Catterick is Sennia (see Britannia xxi (1990), 373, No. 49)Google Scholar.

⁷⁵ The last four strokes are more closely grouped to each other than to the preceding stroke, as if they formed a distinct numeral. (The sequence VIIIII would be unlikely in any case). The graffito may be a note of (liquid?) capacity, like others incised on the rims of amphoras and storage jars.

⁷⁶ Most of them by K. Trott on behalf of the Isle of Wight County Archaeological Service, where Dr D.J. Tomalin made them available to RSOT. They have already been condemned by Arthur ApSimon and George Boon.

⁷⁷ It was sold by the Duke of Northumberland at Sothebys (London) sale of antiquities, 13-14 December 1990, as Lot 87. For an illustration in colour, see the catalogue. Information from Lindsay Allason-Jones.

⁷⁸ By courtesy of the British Museum: see Britannia xix (1988), 485Google Scholar. note 4. It will be republished with other tablets in RSOT's interim report in Dr Ann Woodward's report on the excavation (1991).

⁷⁹ Commentary

1.M(a)rti Mercuri[o]: A was omitted by haplography with the second half of the preceding M. This is the only Uley tablet to couple Mars with Mercury, but Mercury has displaced Mars Silvanus in Britannia x (1979), 343Google Scholar. No. 3, and two unpublished tablets (interim report, Nos 24 and 84) are addressed to Mars.

2.anulus aureus: the gold ring presumably had been stolen, compare Tab. Sulis 97 (a silver ring), RIB 306 (agold ring), but the nominative case instead of the usual accusative is puzzling; perhaps donalur (‘is given’) has been lost in 1.

de hos[pitiolo…]’. the characteristic bottom stroke of S survives; for the restoration, compare Addenda et Corrigenda (c) and (d) with note, below. The next word would have been the possessive pronoun meo or suo.

3. erit: the end of the word is damaged, and oddly inscribed (perhaps E was written twice). Probably a verbalending (e.g. involav/erit) rather than the future of esse.

et pedica ferre[a]: an iron fetter is hardly an object of theft; perhaps a rhetorical conceit, one metal circle for another, ‘an iron fetter to punish the theft of a gold ring’. It seems to be unparalleled.

4. S is certain, followed apparently by another letter, perhaps E or I (whence s[i] qui for si quis).

qui fraudem feci[t]: the last surviving letter seems to be a serifed I rather than E (for fece[rit]), With the addition of mihi after qui, the phrase also occurs in Tab. Sulis 32. 5 (see note ad loc, and p. 64) and in another Uley tablet interim report, No. 78).

5. deus inveni[a]t: for this phrase compare Tab. Sulis 44, qui rem ipsam involavit deus inveniat, and ibid. 99, qukumque r[es] deus ilium inveniat. Here it seems to be preceded by r[es] or similar. After inveni[a]t there is an uninscRIBed space, too brief for certainty, which suggests that this was the end of the text.

⁸⁰ This corrected reading by RSOT was prompted by Dr J.N. Adams' observation that (ta)mdiu is an adverb which cannot be treated as if it were a conjunction. What was read as SILV is in fact SFV, the two strokes of the F being rather widely separated, as can be seen by comparison with lines 7 (end) and 8, where the phrase is repeated. Si… m(e)dius fuerit is thus used in tandem with si … aliquid de hoc noverit, and medius ought to carry the sense of ‘knowledge’ whether licit or not.

⁸¹ For this reading compare the next item (with note 82). Although line 2 of Tab. Sulis 2 (ii) apparently supplies part of the restored text, its line 3 (which seems to begin DEOS, just possibly for de (h)os[pitiolo]) cannot be fitted in, and so the two fragments must belong to different tablets.

⁸² This reading was suggested by an unpublished text from Uley (interim report. No. 72) which complains of theft de hospiliolo meo. (The same text incidentally contains the phrase iteratis precibus rogo. which was hitherto unique to this Pagans Hill tablet.) The phrase de hospitiolo meo can be restored in another Uley text (see above. Addendum et Corrigendum (a)) and in Tab. Sulis 12 (i) (see above, ibid. (c)). In Britannia xxi (1990). 381Google Scholar, Dr J.M. Reynolds saw no compelling reason for taking de hospitio suo in Tab. Sulis 99. 1-2 as ‘house’ rather than ‘lodging’. The sense of ‘lodging’ is indeed possible at Bath (as noted ibid., p. 235). that is, assuming the petitioner was living there (the cult of Sulis being more than local), but this sense is unlikely in the countryside served by the temples at Uley and Pagans Hill: at Uley, in the first text cited (interim report. No. 72). property lost from the hospitiolum included two wheels and some cows.

The phrase de hospitio or de hospitiolo occurs locally often enough for it to have been a formula with a constant meaning. For it to have had the sense of ‘house’ in southern Britain need not be surprising. Hospitium in the sense of ‘house’ is already found more than a century before Tab. Sulis 99, in an official letter written at Rome in A.D. 193 (ILS 5920).

⁸³ V is ligatured to L (compare [at]tulerint in 10); the trace(?) below it was misinterpreted as the descender of an A. The context requires a verb with the sense of ‘steal’ and the rare verb peculor (‘embezzle’) fits very well. Strictly speaking a deponent verb, it is used here as if active: for this ‘Vulgar’ usage compare Tab. Sulis 41, 5, adsellare (with note ad loc).

⁸⁴ By Nigel Peart, who was consolidating buildings excavated under the direction of Tony Wilmott.

⁸⁵ C is visible at the beginning of 2, and enough survives of D to be sure that this is the reading, not O; what survives of the next letter would suit either C (thus Haverfield) or O. The tribune's name is therefore Domitius Honoratus, not Octavius Honoratus (Haverfield). (Domitius is not the only nomen in DO-, but it is so common, and the others so rare, that its restoration here is almost certain). The wording of the text is not in doubt, given the long series of official dedications to Jupiter in almost identical terms by tribunes of Coh. I Aelia Dacorum at Birdoswald (RIB 1874-1893), but variations of abbreviation and line-division are possible. However, the dedications are always headed I O M and the next line begins COH IA EL (thus abbreviated); Dacorum may be abbreviated to DAC or divided between lines as DACORVM; the tribune's nomen is sometimes abbreviated, but never his cognomen; tribunus is always the last word, quite often abbreviated to TRIB and more or less centred. The restoration proposed is based on these generalisations, and implies a die originally 0.44 m wide.

Neither the tribune's nomen nor his cognomen is distinctive, but in conjunction they are rare in the indexes of CIL. The only equestrian officer of this name actually attested is L. Domitius Honoratus, prefect of Egypt in the reign of Severus Alexander, when he was promoted to praetorian prefect (CIL iii 12052). The difficulty of reconciling the evidence of P. Oxy. 62 recto with CIL ix 338 makes the year uncertain. His previous career is unknown but must have included posts in the militia equestris. The identification would imply that the cohort was already at Birdoswald in the reign of Septimius Severus, which is probable in any case; on some of the other altars it bears third-century dynastic titles including Antoniniana (RIB 1892, i.e. Caracalla/Elagabalus).

⁸⁶ By Mr Michael Finlay. A letter from (Sir) Ian Richmond to Miss E. Cumpston (as her surname should be spelled) confirms the identity of the items. Information from Ian Caruana, to whom Colin Richardson made sealings and letter available.

⁸⁷ Information from Ian Caruana, who provided the drawings.

⁸⁸ Mócsy, Nomenclator, records no personal name of this form, and (e.g.) [p]allor or [nisi f]allor seem out of place. An abbreviated genitive plural is more likely, the form Gallor(um) being frequent in military diplomata and inscriptions. The various alae Gallorum attested in Britain were usually known by a familiar cognomen (e.g. Petriana), so a cohort would be appropriate here. Since cavalry-trappings have been found at Rocester (Britannia xviii (1987), 323)Google Scholar, a cohors equitata might be expected, such as Coh. II Gallorum and Coh. III Gallorum, which are listed with Coh. V Gallorum (whether this was equitata is not known) on the diploma of AD. 122 (CIL xvi 69). The graffito may therefore have named the garrison of Rocester, after the owner's name presumably. There does not, however, seem to be a comparable instance (as against graffiti which name a century); but note RIB 2415.39 (Caerleon), a bronze vessel stamped ala I Th(racum). which bears the owner's name in graffito.