² By Norman Owen with a metal detector, who sold it to Hampshire County Museums and provided details. The tablet is now in Fareham Museum, where David Allen and Kay Ainsworth made it available.

³ For these New Roman Cursive letter-forms see Tab. Sulis, p. 94. The scribe wrote both forms of E and, exceptionally, K in line 8. He separated words by means of a space or by extending final S or T. He made transcription errors which are noted below, e.g. tb diio (2) for tibi dono.

⁴ A commentary on the Latin text follows. Tab. Sulis cited here is R.S.O. Tomlin, Tabellae Sulis: Roman Inscribed Tablets of Tin and Lead from the Sacred Spring at Bath (1988), reprinted from B.W. Cunliffe (ed.), The Temple of Sulis Minerva at Bath, II: The Finds from the Sacred Spring (1988).

1, domine Neptune. The other three British ‘curse tablets’ addressed to the water god Neptune have also been found in rivers: the Thames at London Bridge (Britannia xviii (1987), 360, No. I, Metunus), the Little Ouse at Brandon (Britannia xxv (1994), 293, No. 1, domino Neptuno), and the Tas at Caister St Edmund (Britannia xiii (1982), 408, No. 9, Neptu(nu)s).

2, t(i)b(i) d(o)no. n is written with two vertical strokes but without the linking diagonal, almost as if to supply the missing i i in tibi. This n and the omission of i and o must be errors in transcription, like ide(o) in 5, since dono tibi is correctly transcribed in 8. For dono in the sense of ‘giving’ the thief to a god, i.e. cursing him, see Tab. Sulis, p. 64.

2, (h)ominem. The author for the sake of rhetorical variation refers to the thief as qui involavit (2-3, 13, 16), qui decepit (6), qui conscius fuerit (11, 14), and the antecedent of qui variously as (h)ominem (2), nomi(n)a(5), sanguem eius (10, 16-17), animus (12), furem (15). The reading of the first two antecedents is not a problem, but their interpretation is. Ominem is presumably hominem unaspirated (thus Italian uomo < homo), rather than a garbled nomen or omnem, especially since ‘name’ is apparently used in 5 (see below).

3, (solidum). The letters are undamaged and as drawn in FIG. I, with a diminutive o, then a figure which resembles t and s combined, but is more like f, or even k (compare Niske in 8); then the sequence Idmu. The context requires that this is the object of theft and, since the text continues et argenti[olo]s sex, a reference to money is very likely, gold before silver. The solidus (a gold piece first struck by Constantine at 72 to the Roman pound) was current at the palaeographical date of this text. Unless the reading incorporates an unknown symbol for solidus, the scribe wrote an anagram for solidum (understand unum), perhaps for secrecy's sake, but more likely in keeping with his transcription errors in 2 and elsewhere; for comparable anagrams see Tab. Sulis 6.

3-4, Muconi. In the context this is the owner's name in the genitive case; its nominative ended in -onius. The beginning is uncertain because of the damage to the end of 3: initial Mis likely, but the two short parallel strokes after it do not admit e or ec, which excludes the known Celtic name Meconius < Mec(c) o. There is a Thracian name Muc(c) o, but this would be unlikely in a south-British civilian context.

4, argentifobjs. Argentiolus, a ‘small silver coin’, the diminutive of argenleus, occurs twice in the Bath tablets: see Tab. Sulis 8.1-2 (with note) and 54.4. The capital letters of Tab. Sulis 8 cannot be dated, but the Old Roman Cursive of 54 is probably third-century. The New Roman Cursive of the present text belongs to the fourth century; but since fourth-century silver coinage is not common until after c. a.d. 350, and especially in the reigns of Valentinian I and his sons (364-92), this tablet should be dated to the second half of the century. By now the Empire was officially Christian, but for another instance of the cult of Neptune co-existing with Christianity see the Frampton mosaic inscription (RIB II.4, 2448.8).

Tab. Sulis 98, a New Roman Cursive text from Bath which alludes to Christianity, was prompted by the theft of s(e)x argente[o]s, so argenteus and argentiolus were evidently both used of the silver coinage current in the second half of the fourth century. Possibly they served to distinguish the so-called miliarensis from the lighter siliqua (the two terms are modern), but more likely they were both used collectively, and in practice referred to the siliqua, since this was much more common than the miliarensis. The siliqua was struck at 144 to the Roman pound of silver; since a pound of silver was equated with four solidi, six argentioli would have been worth one-sixth of a solidus. This is the third theft of ‘six’ silver coins (the others are Tab. Sulis 8 and 98), but the number is probably a coincidence; there are also two thefts of ‘two’, in Tab. Sulis 54 and in 36 (restoring duo de[narios]).

5, ide(o). Another transcription error (as in 2), since ideo dono is written correctly in 8; the scribe's eye slipped from -deo to do-.

nomi(n)a. See note to 2, (h)ominem. The ‘name’ (nomen, nomind) of a thief is often cursed, if his identity is unknown: see Tab. Sulis, pp. 65 and 95. The syntax is faulty here, as in the parallel clauses introduced by animus (12) and furem (15), because the authors of ‘curse tablet’ texts often know the formulas but not how to link them together grammatically.

6, decepit. The first instance in a British ‘curse tablet’; it is not found in Audollent either. Although a reference to ‘deception’ would be understandable here (it is the verb's usual meaning, and for fraudem facere see Tab. Sulis, p. 64), the author instead is using decepit as a synonym for involavit (3, 13, 16) and even couples decipias with consumas (17-18, compare 15). So he must understand the verb literally as if de-cepit, ‘took away’. This is not a Classical usage, but in epitaphs decipere is used in the sense of ‘to rob (of one's hopes)’, i.e. ‘to bereave’; and note Vita Nicetii Lugdunensis episcopi (MGH SS Merov. III, 521-4), 11, omnium membrorum vigore deceptus.

6-7, si mascel si femina. Variations of this formula are frequent (see Tab. Sulis, pp. 67-8); this one is not found at Bath, but occurs in JRS liii (1963), 122 (Ratcliffe-on-Soar) and Britannia xiii (1982), 408, No. 9 (Caistor St Edmund).

7-8, si puuer si puuella. This formula is also frequent: see Tab. Sulis, p. 68. The second u is a glide [w] inserted between a back vowel and another vowel in hiatus: compare Tab. Sulis 31.5, suua, with Adams in Britannia xxiii (1992), 10, who cites CIL xi 6289, puuer.

8, Niske. The surface is undamaged here, and the reading is certain. The fourth letter is differently formed than f in this text, and is a very rare example of k. (There is just one at Bath, in Tab. Sulis 52.(b)i.) Niskus like Neptune is evidently a masculine divinity who can seriously damage the thief's health, but he is hitherto unattested. He might be the river itself, but there is no suitable candidate in the Ravenna Cosmography's list (see Rivet and Smith, The Place-names of Roman Britain (1979), 212-13) and one might have expected *Moina/Meon (ibid., 419). A more attractive idea is that Niskus is the masculine counterpart of Niska. The lead tablets found in the principal hot spring at Amelieles-Bains, Pyrenees-Orientales, invoke the Niskas (feminine accusative plural) who are said to be aquiferas and domnas, evidently water-nymphs. (Water-nymphs are addressed in Audollent No. 129, aquae ferventes siv[e v]os Nimfas, and in Habivi (1975), 125, dom(i)na Fons; compare Tab. Sulis 94.5, ad fontem deae Suli(s).) The Amélie-les-Bains tablets were lost soon after their discovery in 1845, but they were carefully drawn (see Revue Archeologique iv (1847), pl. 71); they have been republished by J. Coromines, ‘Les plombs sorothaptiques d'Aries’, Zeitschrift für Romanische Philologie xci (1975), 1–53. Since neska means ‘girl’ in modern Basque, Coromines argues implausibly that the word niska is archaic Indo-European, but he does not discuss the ‘water’ element and in particular the Germanic evidence: although English ‘nixie’ is a modern borrowing from German, as nicker (etc.) the water nymph or demon is found in Anglo-Saxon and other early Teutonic languages. (See The Oxford English Dictionary s.vv. nicker, nix, nixie.) Whether or not this is cognate with niska, it does look as if Niskus was a Celtic water-god, either local, or having a wider jurisdiction which would equate him with the Roman Neptune. Unfortunately there is no evidence from Britain of what Celtic deity was identified with Neptune, and very little elsewhere.

9-10, vitam, valitudinem, sanguem. This exact combination is new, but similar combinations are found at Uley: see Britannia xxvii (1996), 439, No. 1, sanguem [e]t sanitatem, with note. The contracted Vulgar form sanguem is found there also and in Tab. Sulis 44.5-6.

11 and 14, qui conscius fuerit. (fueris in 11 is a transcription error, probably because the preceding word and the next three words all ended in -s.) This formula is new, but compare Tab. Sulis 97.7-8, qui medius fuerit, where the same sense of ‘guilty knowledge’ is required. It is not clear that the author distinguished the conscius from the actual thief, but if he did, he cursed him perhaps as an accomplice, certainly for not revealing the thief's identity.

12, animus. The syntax of this sentence wavers between animus eius ut decipiatur and animum eius ut decipias, and misses them both; the author intended a curse parallel to that in 15-19: ‘take away the animus of the thief. A thief's mentes are cursed in Tab. Sulis 5.5-6, and his anima in 103.2 (with note of other instances).

14-15, ut eum decipias. ut as often can be taken either as dependent on dono understood, or as the ‘free-standing ut introducing a wish’ noted by Adams in Britannia xxiii (1992), 6. decipias (repeated in 17-18) is chosen because it echoes decepit (6) and deceptionis (12): the thief is repaid in kind. This incidentally confirms that decipere means ‘to take away’ (see note to 6, above).

17, eiius. The syntax breaks down again as in 15 (eum) with the insertion of a redundant demonstrative pronoun. Since eius is correctly written in 10 and 11, the repeated i may be an oversight here; but the Vulgar eiius, in which the semi-vowel i in hiatus is reinforced by [y], is quite common. (Dessau ILS iii, p. 822 collects examples, to which RIB 601 can be added.)

consumas. Only the second instance from Britain: see Tab. Sulis 54.8-9, sanitatem consumas.

18-19, domine. The scribe actually wrote an incomplete m followed by i i.

⁵ Letters are transcribed as in the original, with word separation where detectable. Dotted letters are incomplete or ambiguous.

⁶ Words have now all been separated, the initial letters of proper names capitalized, and modern punctuation added. Two transcription errors in 2 and 18 have been tacitly corrected, letters omitted in error are supplied as (abc), and the anagram in 3 has been similarly resolved. Lost letters are restored as [abc]. Consonantal v has been distinguished from u, but Vulgar forms like sanguem have not been normalized.

⁷ With the next item during excavation for the National Museum of Wales and the Glamorgan-Gwent Archaeological Trust directed by Dr E.M. Evans, for which see Britannia xviii (1987), 307 and 305, fig. 4. A fragment of a rectangular sandstone panel 0.04 m thick, 0.23 by 0.20 m, was also found; it carried a simple incised double border, but none of the inscription survives. The stones are now in Newport Museum, where Bob Trett and Joyce Compton made them available.

⁸ The restoration of some case and gender of Ingenuus is almost certain, the name and its cognates being so popular in Britain (for Caerleon compare RIB 358, Ingenuinus), but the rare cognomen Diligens is just possible. Above it, to the left of M, there is a rosette or ‘wheel’ like those in the gable of RIB 369, suggesting that D (is) was separated from M(anibus) by a crescent and two rosettes within a gable. This gives the approximate width of the stone, and shows that a name has been lost before [I]NGEN[…], whether it was a nomen gentilicium for [I]ngen[uus] or the name of the deceased with [I]ngen[ui…] as his/her patronymic.

⁹ Line 2 presumably began where the chamfered gable met the line scored below D M, implying the loss of 1-2 letters; they were the same width as F(?) below them in line 3. This apparently began with FIL and a medial point, which would imply that line 2 contained the names, probably non-Roman, of the deceased and his/her father.

¹⁰ During excavation of a medieval chapel and its vicinity by Chester Archaeology and the Department of Archaeology, Liverpool University, as reported in Rescue News 70 (1996), 4, with photograph.

¹¹ During excavation by Carlisle Archaeological Unit for the Department of the Environment and Carlisle City Council, directed by M.R. McCarthy: see Britannia id (1980), 360; xii (1981), 325; xiii (1982), 344. They will be fully published in the Lanes report, Volume 2 forthcoming, edited by T.G. Padley, who made them available. This report contains further annotation, some graffiti which are very fragmentary or which contain less than three letters, and X-like marks of identification. For improved readings of three items already published, see below, Addenda and Corrigenda (a), (b) and (c).

¹² Barrel staves bearing an abbreviated name branded on the inside face, where it would have been invisible after the barrel had been assembled, are known from London (RIB II.4, 2442. 5, 6, 14 and 23) and Silchester (17), but the best parallels are two staves from Vindolanda published in E. Birley, R. Birley and A. Birley, Vindolanda Research Reports, New Series, Volume II, The Early Wooden Forts: Reports on the Auxiliaries, the Writing Tablets, Inscriptions, Brands and Graffiti (1993), p. 78, LCMA; and p. 82, QVB [but the drawing reads QVF].

¹³ In line 1, the letters are too damaged and incomplete to be legible, but they do not look like part of a consular date. Line 2 is a possible day-date, assuming that [pr]ide is by synizesis for pridie.

¹⁴ Possibly a day-date incorporating ‘October’, but note RIB II.6, 2493.77, the Dressel 20 potter's signature [V]ictoris (as the transcription should have been printed), ‘(product) of Victor’.

¹⁵ Possibly [Imp(eratore) Hadriano III] et Rus[tico co(n)s(ulibus) | […], a.d. 119, but for discussion of this uncertain reading and expansion see the Lanes report cited above in note 11.

¹⁶ The scratches are too ambiguous for any firm conclusions, but if there were two names here and the first ended in […] vani then it was probably [Sil]vani, ‘(Property) of Silvanus’.

¹⁷ This reading suits the surviving traces better than a numeral, VII or VH[I].

¹⁸ The numerals VII and VIII are frequent as measures of capacity scratched on Dressel 20 amphoras. See RIB II.6, p. 33, and below, No. 47 (Caerleon).

¹⁹ M was apparently first scratched in two halves as A A, then scratched a second time, correctly. For these measures of capacity, and the suggestion that the number of sextarii might be scratched on the handle, see RIB II.6, pp. 33-4.

²⁰ For another example of ATT see RIB II.7, 2501.72 (Chester). Atto and derived names, including ‘Roman’ names like Atticus which recalled the same name-element, were popular in Gaul (especially Gallia Belgica) and the tw o Germanies, and so in Britain: see G. Alföldy in Epigraphische Studien 4 (1967), 10-16.

²¹ For other instances of CAN and IAN as abbreviated personal names, see RIB II.8, 2503.217 and 276 respectively.

²² Lucius and Lucanus and kindred ‘Roman’ personal names were popular in Gaul and Britain because they recalled the Celtic name-element *leuco- or *louco-, ‘bright’, ‘shining’. For examples see RIB II.7, 2501.298-307 and II.8, 2503.305-7.

²³ The scribe first inverted the dish, and incised each letter in turn on the side away from him, so that they face out from the centre of the dish and are thus inverted with respect to each other. If this arrangement were only intended for symmetry, it would follow that they were meant to be taken together: XV, ‘fifteen’, perhaps the number of this vessel in a set.

²⁴ Dedications are known to the Italian goddess Marica (e.g. ILS 9264), but the absence of DEAE in this graffito, and in particular the localization of her cult to the Campania n coast near Minturnae, exclude this possibility; Marica here is a woman, not a goddess, and of non-Italian origin. There are five other examples of the name Metrica/Maricca: two in south-eastern Noricum (CIL iii 5257, Maricca Atemeri f(ilia), and iii 11647, Marica Blendonis f(ilia)), two from the frontier zone of Noricum or northern Pannonia (AE 1978.599, Vextila liberta Marices, and CIL xv 175, Maricca Curi f[il(ia)]), the homeland of the Gallic rebel Mariccus e plebe Boiorum (Tacitus, Hist. 11.61); and a fifth from north-eastern Italy (CIL v 6850, Aosta, Maricca Namicifil(ia)), where there was also a tribe of Marici near Ticinum (Pavia). So it is a Celtic personal name, but not necessarily British; Marica may have been an immigrant from Noricum or its neighbourhood.

²⁵ The bird's significance to Marica is not explicit; a rebus might have been expected, like the graffito TAVR I accompanied by the drawing of a bull (RIB II. 5, 2491.124), but no Roman bird called a marica is known.

²⁶ The letter was incised just to the left of centre and, although one can be certain only of the uninscribed space to its right, it seems to have stood alone.

²⁷ Only a drawing is available, but the dish was apparently Drag. 18/31R.

²⁸ The graffiti, now damaged, are more fully discussed in the Lanes report (see above, note 11). N in (i) is by a different hand than N in (ii) (a), and over it was inscribed S when the new owner wrote his name. To confirm the change of ownership, he also wrote his name underneath. The name Sanctus is frequent in Gaul and Britain.

²⁹ Only the upper left-hand corner survives of the letter after M, the tip of a diagonal down-stroke appropriate to A (but M or N would also be possible). There would have been space for only one letter before and after MA, so the graffito is likely to have been a personal name abbreviated to three letters, for example MA[R] (see RIB II.7, 2501.322-5; II.8, 2503.322-4) or MA[T] (ibid. 2501.353-9; 2503.330-2).

³⁰ RIB identifies Ravenglass as Glannaventa, for reasons followed by Rivet and Smith in The Place-names of Roman Britain (1979), 367, s.v. Glannoventa (and thus in RIB II), but with reservations (ibid., 171). The discovery at Ravenglass of a lead sealing of cohors I Aelia Classica, which is the Notitia garrison of Itunocelum (for this emendation see Rivet and Smith, s.v.), first cast doubt on this identification: see RIB II. 1, 2411.94 with note. Holder (see next note) now suggests that, if the cohort was at Ravenglass in a.d. 158 and was still there when the Notitia list was compiled, Ravenglass should be identified as Itunocelum.

³¹ By James and Howard Meadowcroft with a metal-detector. Since the fragments were found on Crown Estate land, they were presented to Manchester Museum (accession number 1996.201) and the finders were rewarded. The discovery was noted in Britannia xxvi (1995), 389-90. The diploma is now published with full commentary by Paul Holder, who sent an advance copy of his text. See P.A. Holder, ‘A Roman military diploma from Ravenglass, Cumbria’, Bulletin of the John Rylands University Library of Manchester 91.1 (spring 1997), forthcoming.

³² Julius Verus is attested in Britain in a.d. 158 by RIB 2110, but A. R. Birley, The Fasti of Roman Britain (1981), 118-21, has already argued that he must have arrived some while earlier. His reasoning is now confirmed: if Verus was already there in February 158, he must have arrived in the previous year (157) at least.

³³ This is the first evidence that the cohort was equitata. It is first attested in Britain in a.d. 146 (RIB II. 1, 2401.10), but a Hadrianic prefect is known (CIL xiv 5347), though not necessarily in Britain. It is unusual for a Roman colonia to be the origin of a non-Roman auxiliary veteran, and other instances are due to recruitment in time of war. This veteran from Heliopolis, if he served the usual 25 years, would have been recruited in a.d. 132/3, presumably in response to the Bar-Kochba rebellion in Judaea; and since Sex. Iulius Severus was transferred from Britain to crush it, it is possible that he took reinforcements with him, including cohors I Aelia Classica in whole or part. Two other instances have been suggested by E. Birley, The Roman Army; Papers 1929-1986, Mavors iv (1988), 110 and 216, but see A.R. Birley, The Fasti of Roman Britain (1981), 124-5 with note 4.

³⁴ During excavation by Keith Scott, who made it available.

³⁵ Taking the first (broad) stroke to belong to this graffito, which would then be inscribed in a semi-circle in large capitals: P[ACA]TI or similar.

³⁶ The first letter could be I or V. Then the only two complete letters, both of them resembling C or L; but if the second one continues with the short diagonal stroke cut by the first of the next two vertical strokes, a cursive P can be read. But Ulpii, the genitive of a nomen rare until the reign of Trajan, is difficult here.

³⁷ Someone has noted on it in ink: ‘M'L. 25.5.30 Old Jewry’. Jenny Hall at the Museum of London has not been able to locate this more exactly, but the sherd must be one of the hundreds of pieces of decorated samian found in the City during the 1920s and 1930s, and exported to North America: see Macdonald in J. Bird, M. Hassall, H. Sheldon (eds), Interpreting Roman London (1996), 248. It was bought recently in the U.S.A. by the booksellers Bernard Quaritch Ltd, one of whose Directors, Richard Linenthal, made it available.

³⁸ The letters have cut through the glossy slip and exposed the underlying fabric, but sometimes the slip is banked up one side as if it were plastic at the time, and at the extremities of letters, where the pressure was least, the stilus has not only sometimes cut part-way, but quite often the slip seems to have fused over again. So the graffiti seems to have been made before firing, and thus to belong to the manufacturing process. In RIB II.7 (graffiti on samian) there is no example of this script, which resembles that of Flavio-Trajanic waxed writing tablets from Britain. For [P]hoebus[s] as a South-Gaulish (Drag. 37) mould-maker's name, see Britannia xxii (1991), 304, No. 40 (Rocester), with note 59.

³⁹ With the next two items during excavation for the Central Unit directed by P.R. Wilson, for which see Britannia xxii (1991), 239-40. They will be published with other graffiti in the Roman Catterick monograph (1997, forthcoming).

⁴⁰ After CAIISI there is an uninscribed space wide enough to exclude the nomina CAIISI[INNYVS…] and CAIISI[IRNIVS…], and almost certainly wide enough to indicate that the word ended here. Caesius is a common nomen gentilicium, which would imply that a cognomen has been lost after it, although Caesius itself is occasionally found as a cognomen. There was a firm of south-Spanish olive oil exporters, Quintus Caesius Caesianus and Quintus Caesius Macrinus, known from painted inscriptions on Dressel 20 amphoras: QQ CAESIORVM CAESIANI ET MACRINI (CIL vx 3797, dated a.d. 149; also xv 3798 and 3799). There may be a connection here, but a direct link between potters’ dating-graffiti and the painted inscriptions on Dressel 20 amphoras remains to be proved.

The rest of the graffito is evidently the date of manufacture. For other examples see below, Addenda and Corrigenda (a) and (e). These graffiti are sometimes found on the base of a Dressel 20 amphora, but they are always fragmentary and their purpose is not fully understood: see R. Rodriguez-Almeida, Il Monte Testaccio (1984), 256-7. They usually contain three successive elements: (1) a consular date in the mid-second century; (2) the day and month; (3) a personal name in the genitive case, whether the potter's name or that of the firm.

⁴¹ The now incomplete O is much smaller than the succeeding letters, whose vertical height has been exaggerated, but this is acceptable in such a crude graffito. In Latin the letter-sequence OPT is likely to be the beginning of a word (e.g. optimus, optio), and the cogonomen Optatus is common.

⁴² This is the likeliest reading, but the first letter might be R and the second I; none of the three letters in certain.

⁴³ During salvage operations during redevelopment. It was sent by the County Durham Archaeologist Niall Hammond to J.N. Dore at the Museum of Antiquities, Newcastle, but it will probably go to the Bowes Museum.

⁴⁴ The letters are now incomplete except for -LIO-, but even O is uncertain; it might be D or P. The first letter is apparently an exaggerated C or S.

⁴⁵ During weeding of the granary/barrack block C₁₂ after consolidation. Inventory number SS96, IM71. Alex Croom made the three sealings available, and sent photographs and other details of them and of the coarseware graffito (No. 38).

⁴⁶ The other possibility is al(ae) S(ebosianae), compare RIB II.1, 2411.88, bu t another sealing of the ala Sabiniana has already been found at South Shields (ibid., 86).

⁴⁷ During excavation of a granary extended into the barrack block C16 (see Britannia xxvii (1996), 408-9), directed by Dr N. Hodgson and Mr P. Bidwell for Tyne and Wear Museums and Earthwatch. Inventory number SS96, IM₇₄.

⁴⁸ During excavation of the barrack block Coo (see previous note). Inventory number SS96, IM₇₃.

⁴⁹ The central N is badly damaged by the channel of the binding-cord. The Severan sealings from South Shields, where legible, read only AVGG like the previous item, i.e. before the proclamation of Geta as Augustus. The only instance of AVGG G from Britain is RIB II. 1, 2411.19 (London). The date must be winter 210/11, strictly speaking between about October 210 and 4 February 211: see A.R. Birley, The African Emperor: Septimius Severus (1988), 218.

⁵⁰ During excavation of the barrack block C₁₅ (see above, note 47). Inventory number SS96, IM₇₅.

⁵¹ As Minna this Celtic name occurs in RIB 694 (where see note); for the derived name Minnius Holder cites CIL v 1892; 7034; cii 1871.

⁵² It remains in the finder's possession, but was examined by Paul Robinson at Devizes Museum (reference DB 1972). He sent details and a drawing, observing that it is the first stamped tile from Verlucio. It probably derives from the Minety kilns: see RIB II.5, p. 72, introduction to 2489.44.

⁵³ During excavation for the National Museum of Wales and the Glamorgan-Gwent Archaeological Trust directed by Dr E.M. Evans, for which see Britannia xviii (1987), 307 and 305, fig. 4. They are now in Newport Museum, where Bob Trett and Joyce Compton made them available. Marks of identification only, and incomplete graffiti of less than three letters, have been excluded.

⁵⁴ Three other lead tags were also found, but they are heavily corroded, and there is no sign of any text. The inventory numbers of the others are 79/001. 200 (No. 40), 79/191. 1239 (No. 41), 79/600. 424 (No. 42), and 79/600. 299 (No. 43).

Lead tags were apparently used to identify various classes of object, and the brevity and allusiveness of their texts, which are also often damaged and difficult to decipher, increase the problems of understanding them. A large group from Noricum has been well published by Elizabeth Römer-Martijnse, Römerzeitliche Bleietiketten aus Kalsdorf, Steiermark (1990), whose bibliography is now extended by Cristina Bassi in Epigraphica lviii (1996), 207-16. For Britain Bassi cites tags published by M.W.C. HASSAL-R.S.O. TOLMIN in 1975 and 1977, but others will be found in RIB II.1, 2410 and II.8, 2504.7-9; Britannia xxii (1991), 297, No. 9, and xxv(1994), 304, No. 35.

⁵⁵ Line 1 is certain, but line 2 is complicated by casual damage or the remains of an earlier text; i seems to have been inscribed at an angle. This name, developed from the cognomen Aeternus, is quite well attested, bu t this is the first British example.

⁵⁶ The first two or three letters of line I have been crossed out, and there are traces of an earlier text between the lines. The cognomen Cupitus is very common. In line 2, the initial A has been inscribed by a broader point and on a different alignment from cres, which may therefore represent a n abbreviated Crescens, itself a very common cognomen.

⁵⁷ Like Ar(r)untius on the reverse, apparently, Calpurnius although a nomen gentilicium is used on its own, as if it were a cognomen; there may be another example in RIB 1619, if the transmitted CAPVRIVS does not conceal a Germanic name. This usage is found with two very common nomina gentilicia, Iulius quite often and also Valerius (e.g. RIB 1577). Calpurnius was probably a legionary cavalryman, but the ala I Thracum is a possibility (compare RIB II.2, 2415.39).

⁵⁸ E overlies another E, followed by a diagonal appropriate to Q, so the previous inscription may have named another eques. The traces underlying CALPVRNI look like a name crossed-out, or even two names.

⁵⁹ Between T and R are two intersecting strokes (not drawn), perhaps intended to cross out an earlier text. A numeral is to be expected after the denarii sign here, as on the other TR* tag from Caerleon (RIB II.8, 2504.7), but on neither is it beyond doubt. The meaning of TR is unknown, but Hassall conjectures tr(iticum) in Britannia xx (1989), 342, note 75.

⁶⁰ The final I resembles S, which was apparently inscribed in error. For optionis compare eq(uitis) on the obverse. A name Aruntius is not attested, but presumably Arruntius was meant; for the use of a nomen gentilicium alone, compare Calpurnius on the obverse.

⁶¹ Inscribed over traces of an earlier text. The surface is heavily corroded.

⁶² Finitus is a notably common name in Noricum. At least eight British legionaries from Noricum are known, including Aurelius Candidus of legio II Augusta (CIL iii 5476): see E. Birley, The Roman Army: Papers 1929-1986, Mavors iv(1988), 285-9.

⁶³ This and the next ten items were identified by Peter Webster.

⁶⁴ Iustus is more often found (e.g. at Caerleon, RIB 322) than Iustinus, but both names are common.

⁶⁵ VII is a common graffito on Dressel 20 amphoras: see above, No. 14 (Carlisle) with note 18.

⁶⁶ Mis[ce mi(hi)] or Mis[ce] are also possible, but less likely. Compare RIB II.6, 2498.16 (with note) and 17.

⁶⁷ M is thicker and somewhat less irregular, and may belong to a previous graffito.

⁶⁸ An abbreviated personal name; compare RIB II.7, 2501.466 and 467.

⁶⁹ There are two forms of A, the first with a horizontal cross-bar, the second with a vertical. Martia[nus] is just possible, but it is a much less common name and not yet attested in Britain, as against more than twenty examples of Martialis. It was probably preceded by a nomen gentilicium, perhaps Valerius. A contemporary mortarium of local manufacture is inscribed Val(e)rius Mar(t)ial(i)s before firing (RIB II.6, 2496.6), which implies that vessels were being made to order; compare the next item. But Valerius Martialis is a common name and, despite the coincidence of place and date, identity is not certain.

⁷⁰ The surface is very worn, but the rounded cross-section of the lettering-strokes and the easy circularity of the O both suggest that the clay had not been fired, but was leather-hard, when the graffito was inscribed. Like the previous item, the vessel was presumably made to order. The nomen gentilicium most likely began with A or M.

⁷¹ The second I is very short and slight, and so was presumably the final letter. Derived names like Lucianus are quite common, so this is a possibility, but the praenomen Lucius is very often found on its own: compare RIB II.7, 2501.2988ff.

⁷² This and the next two items have been re-examined for the report on the graffiti from the Lanes, Carlisle (see above, note 11).

⁷³ The initial letters are of exaggerated size. In line 1, the enlarged loop of G is finished with a small hooked tail, the bold diagonal stroke being part of an enlarged A; L is made with a short horizontal stroke.

⁷⁴ a.d. 150, either 2 March, 2 May, 2 July, o r 2 October. For these dating graffiti see above, note 40.

⁷⁵ The line-drawing in RIB II is inaccurate. E has been incised over the second half of M in thicker strokes, suggesting that the first half was now understood as A. Over A in (i) has been incised a zig-zag, perhaps a cursive L.

⁷⁶ The size, style and alignment of the letters change between NVI and TIIR, and T is notably enlarged, as if it were the initial letter of a name. The line-drawing in RIB II, which was made from a rubbing, omits the diagonal stroke of N and misrepresents R, whose vertical stroke apparently springs from a short preliminary downstroke.

⁷⁷ Ingenuus is the only common name in -nuus, and was popular in Britain. The other name, probably in the genitive also, is likely to have been Terentius, Tertius or Tertullus. Much the commonest of these is Tertius, which is already attested at Carlisle (Britannia xix (1988), 496, No. 32 with xxii (1990), 378) and at Stanwix (RIB II.7,2501.540).

⁷⁸ Owners often scratched their names on a vessel while holding it upside-down for greater stability. Geminus is a common cognomen.

⁷⁹ a.d. 149. For these dating graffiti see above, note 40.