² In a modern (c. 1800) deposit of rubble in the grounds of Muncaster Castle (SD 103 963). It was not associated with any building, and was evidently brought to the castle as building material from elsewhere in the estate; in view of its text, it came from the Roman fort. Information from Kurt Hunter-Mann, who excavated the findspot. It remains at the castle, where Peter Frost-Pennington made it available.

³ Letter-heights: line 2, 65 mm; line 3, 72 mm. There is trace of a setting-out line below M in both lines. In lines 2 and 4, the word-divider is a short vertical chisel-cut after M and I respectively; and there would have been one in line 1 after L, as the spacing suggests, but the surface is damaged here. In line 1, the base survives of both diagonals of A; the next letter is either E or L, followed by just enough of L to exclude E. The fourth letter looks too wide for O or Q, and is either C or G. At the end of line 2, the triangular cut in the edge is casual damage, not part of R; in fact the letter is I with damage immediately to the right of its top-serif which has probably removed the tip of the top-serif of the next letter, which would be N (see next note). In line 3, a small part of the curve of O survives in the broken edge. In line 4, the first letter must be an incomplete A, since M would not have been followed by R. The letter after R is almost certainly I, since there is not enough space for L. The exact division between lines 2 and 3 cannot be certain, but the dedication dis deabusque and the nominative case of coh(ors) can be deduced from RIB 1579 (Housesteads). See next note.

⁴ The letter-sequences are incomplete but quite sufficient to show that this is the second British example (after RIB 1579) of a dedication ‘to the gods and goddesses according to the interpretation of the oracle of Clarian Apollo’. Eleven such dedications, all but one from the Latin West, are collected by C.P. Jones in JRA 18 (2005), 293–301 and 19 (2006), 368–9, and he argues convincingly that they were prompted by Marcus Aurelius’ response to the outbreak of the Antonine Plague in c. a.d. 165. He notes that only one names the dedicator, that of cohors I Tungrorum at Housesteads. This does not mention the commanding officer, nor (it would seem) did the Ravenglass inscription, but they were both dedicated by an auxiliary cohort.

This new evidence confirms that cohors I Aelia Classica was stationed at Ravenglass, as Paul Holder deduced from the discovery here of a lead sealing of the unit (RIB II.1, 2411.94) and the diploma of a.d. 158 (Britannia 28 (1997), 464, no. 28 = RMD V, 420) issued to one of its veterans. He also deduced, from the Notitia Dignitatum (Occ. 40.51) and the Ravenna Cosmography (107, 3), that the ancient name of Ravenglass was Itunocelum (as emended), not Glannoventa: see Holder, P., ‘Roman place-names on the Cumbrian coast’, in Wilson, R.J. and Caruana, I.D. (eds), Romans on the Solway: Essays in Honour of Richard Bellhouse (2004), 52–65Google Scholar, esp. 54–9.

⁵ During excavation by Albion Archaeology before development. Holly Duncan made it available.

⁶ Line 2 is indented with respect to line 1, and if [D]EO[R]V[M] is correct, it would seem that line 2 was centred below line 1, as if together they formed a heading. There is no sign of any text above or to the left, but the broken edge below is too close to be quite sure there was not a third line. The sequence LAGON is quite clear, with L and O both made rather narrow (like ET in line 2), and A with a short diagonal cross-stroke. This cross-stroke is now missing from the second A which follows N. A is followed by a vertical stroke, which is probably a narrow E like that in ET below. The resulting lagonae might be either genitive or dative, but is taken here as a ‘dative of purpose’, the intended use of the flagon. After lagonae there is a space, then trace of what looks like IC (but with a hint of the horizontals of a narrow E), another space, and then the right-hand tip of V, which in view of line 2 (the names of two deities in the genitive case) can be restored as [D]EO[R]V[M]. A photograph taken by infra-red did not reveal any more than this. Line 2, after the initial IOVI, is a little faint and rubbed, but the reading is certain.

lagona (variously spelt) means ‘a bottle with a narrow neck, flask, flagon; esp. a wine-flask, bottle of wine’ (OLD). In Britain it is applied to a wine-amphora (RIB II.6, 2492.7), a storage jar (II.8, 2502.4) and a jug (2503.111), but the only parallel for its use in a religious context seems to be the jug from Le Rondet (Switzerland) which carries two separate but related graffiti, lagona and Is[i]di: ‘a flagon for Isis’ (L. Bricault, Recueil des inscriptions concernant les cultes isiaques (2005), II, 609/021, illustrated in Epigraphica 71 (2009), 400, fig. 7). There is evidence from Britain for the cults of Jupiter and Vulcan being combined, in a silver plate (RIB 215) and a stone altar (899) dedicated to them jointly; RIB 835 may be another such altar. The purpose of this flagon can only be guessed, but perhaps it held wine to refresh a guild of worshippers. Such collegia are attested for Apollo (RIB 611), Mercury (2102, 2103), Minerva (1268) and Silvanus (1136+add., II.3, 2422.52).

⁷ By a metal-detectorist (PAS ref. BH-584656). Full details from John Pearce, who is studying moulded mirror-frames found in Britain with their continental parallels, on which Giulia Baratta of the University of Macerata is completing a book.

⁸ The sigma encloses a medial point, a combination read by the PAS database as theta. The text can be restored from complete examples, but has only occurred once before in Britain, in Latin at Vindolanda (Britannia 42 (2011), 452, no. 17 with note).

⁹ PAS ref. BERK-1271A1. This and almost all the other items from the PAS database were communicated by Sally Worrell.

¹⁰ The lettering, which is damaged in places, is best seen in the bottom line of fig. 4. The same die was used each time, but it was inverted for the second line. G is a very full C-like loop incorporating a diagonal stroke, which the PAS database reads as Q, and S is reversed. The cognomen Geminus is quite common, and was presumably the name of the owner or merchant, but there is no instance in RIB II.2, 2412 of a metal weight being stamped with a name; a stone weight (Britannia 42 (2011), 456, no. 24) is inscribed with two letters which may be an abbreviated name. Paul Booth comments that the notional weight was probably half a (Roman) pound, 163.73 gm, and attributes the shortfall to a hole now drilled in the back, and the fact that most recorded weights seem to fall short of the notional weight. Only two of the seven explicitly half-pound weights in RIB II.2, 2412 are more or less correct, whereas 2412.20 and 80, like this weight, weigh less than 150 gm.

¹¹ With sherds of Roman pottery during field-walking. They are thought to have been deposited in Peterborough Museum in 2005, but the sealing has not been located there. The only evidence for its appearance is a rough sketch made at the time (fig. 5). Information from Geoffrey Dannell.

¹² The obverse looks as if it may have been impressed by the same die as the obverse or reverse of the Chi-Rho P(rovincia) M(axima) C(aesariensis) sealing found at Silchester (RIB II.1, 2411.38), but this cannot be confirmed.

¹³ With the lead sealings published in Britannia 49 (2018), 437–41, by the divers Bob Middlemass and Rolfe Mitchinson. It will be sf 400 in the final report on the Piercebridge deposit in preparation by Philippa Walton and Hella Eckardt, who made it available.

¹⁴ Finger-rings incised with the three initials of a Roman citizen's tria nomina are quite common, but the only other instance of the initials being preceded by a centurial sign seems to be RIB II.3, 2422.51, ϽMAV. There is no instance in CIL xiii, 10024, nor in F. Henkel, Die römischen Fingerringe der Rheinlande (1913). The purpose of the inscription is unclear. Since it is rectograde, it can hardly have been used to sign lead sealings like those found with it. The ring is rather small for a man's finger, so it might have been attached to something jointly owned or dedicated by the century (compare RIB 446, 447 and 944), but most likely its owner was simply asserting his membership of the century.

¹⁵ During excavation by Gloucestershire County Council Archaeology Service. Information from Hilary Cool.

¹⁶ Hilary Cool notes that IIIII for ‘5’ is rare, with RIB II.3, 2440.186 being the only other instance of IIIII in the rim of a bone roundel. The purpose of such numerals is unknown.

¹⁷ During archaeological excavation before development. Information from Hilary Cool and Paul Nichols of Gloucestershire County Council Archaeology Service. The item will be deposited in the Corinium Museum, Cirencester.

¹⁸ This might be a mark of identification, but is probably a numeral locating the spoon within a set.

¹⁹ With the next two items during excavation directed by Michael Fulford (sf 7889, 7723 and 6574 respectively). Emma Durham made them available. They will be published in the final report (now in preparation) with five other graffiti of less than three letters.

²⁰ The upper C is now incomplete, but was evidently repeated on a smaller scale by the letter below it, C rather than L. Just enough of the base survives to suggest that the graffito was centrally placed and consisted of only three letters, an abbreviated name. The nomen Cocceius is quite common, but (since it implies enfranchisement by Nerva) is unlikely on such an early vessel; also an abbreviated nomen without cognomen is unlikely. Much more likely is the Celtic name Cocus and its variants, which occurs in Britain in the feminine form Cocca (RIB II.8, 2503.230; compare 496).

²¹ There is enough space to the left and probably to the right to mark the beginning of the name and its end, but the constituent letters are now incomplete except for V in the middle. They are also damaged and obscured by the horizontal scoring which was intended to delete them, marking a change of ownership. The space after the first letter suggests it was F, P or T; and the second letter(s) might be A with a medial downstroke, or perhaps LI. After V, the letters look most like I and S, but with a space between them. The sequence –avi– would suggest Suavis or Flavius, both quite common names, but they are not supported by the other traces.

²² The first figure, to the left of the four digits, might be read as X (for the numeral XIIII, ‘14’), but with its long horizontal stroke at right-angles to the first digit, cut by a shorter vertical stroke which is parallel to this digit, it looks more like a vertical ‘cross’; a mark of identification introducing the numeral which follows. This numeral's significance is unclear, but it may have numbered the jar within a set of vessels, or have indicated its weight or capacity. Numerals of uncertain reference are quite common on coarseware (RIB II.8, 2503.34–95).

²³ During the excavations directed by J.P. Bushe-Fox, but not published; no contextual information survives. It was found by Kathryn Bedford in the English Heritage store at the Stone Hut, Dover Castle (acc. no. 96003916), during a cataloguing project. She sent full details including a photograph, and suggests it may relate to the pieces of bronze sculpture associated with the Monument described and discussed in B.W. Cunliffe (ed.), Fifth Report on the Excavations of the Roman Fort at Richborough, Kent (1968), 70–1.

²⁴ During the excavations directed by J.P. Bushe-Fox (sf 5639), and now stored in Dover Castle (acc. no. 88396114). It was found in Area XVII (a row of shops south of Watling Street within the walls), probably in a context of c. a.d. 80–95, but quite likely to be residual; there is no need to see it as ‘military’. Information from Philip Smither, who noticed the graffito while studying the Richborough small finds, and sent photographs and other details.

²⁵ The graffito is now very worn, but seems to be a ‘star’ formed by three intersecting lines, its horizontal extending rightwards to a short downward stroke, followed by X. This does not look like an incomplete personal name, but the extended horizontal may have been meant to mark IX as a numeral (‘9’). Although this ‘star’ resembles a denarius sign, the graffito cannot have been a price, since this would have been far too high.

²⁶ Fully published with a drawing and commentary by Mullen, A. and Tomlin, R.S.O., ‘More from the Romano-British poets? A possible metrical inscription from East Farleigh, Kent’, Britannia 50 (2019), doi:10.1017/S0068113X19000084Google Scholar.

²⁷ PAS ref. LEIC-5FC533. Fully described by Sally Worrell in Britannia 39 (2008), 361, no. 10, with reference to earlier discoveries of Chi-Rho rings, but not noted in that year's ‘Inscriptions’.

²⁸ PAS ref. LIN-2AE922.

²⁹ Many silver TOT rings have been found, mostly in Lincolnshire, ever since the first five recorded in RIB II.3, 2422.36–40. For a catalogue of 68 such rings, see Daubney, A., ‘The cult of Totatis: evidence for tribal identity in Roman Britain’, in Worrell, S. et al. (eds), A Decade of Discovery: Proceedings of the Portable Antiquities Scheme Conference 2007 (2010), 109–20Google Scholar.

³⁰ PAS ref. LIN-B7D304.

³¹ Capital O is enclosed by two narrow capital Ts which are little more than ‘I’.

³² PAS ref. LIN-2AADFD.

³³ The two capital Ts are plain, without serif, and the central O is a simple pellet.

³⁴ PAS ref. LIN-9D01AA.

³⁵ Each capital T has an exaggerated bottom serif so as to enclose the lower-case O. Incised diagonally above and below them is a pair of ‘palm-branches’. The bezels of most TOT rings are integral with the hoop, but for two other detached circular bezels, see Britannia 48 (2017), 465, nos 12 and 14.

³⁶ During excavation by MoLA of the site of Bucklersbury House before redevelopment (Britannia 45 (2014), 370). Julian Hill made it available (sf 10388) and provided a photograph.

³⁷ In CORNILII, the descender of L follows the second ‘I’, not the third, an error for CORNIILI or CORNIILII. Cornelius was presumably the producer or shipper. To the right of his name, before the numerals II and XV, the surface is damaged, suggesting that C has been lost here, for [C]II, ‘102’. By comparison, the Drapers’ Gardens barrel-head (Britannia 40 (2009), 337, no. 36), almost the same in diameter (0.76 m), is annotated CV XIS (105 modii, 11½ sextarii). The lettering of A MIISSORII is larger and less well-formed, clearly by another hand; its first strokes (of A and M) and the last (of II) are now rather faint, and R has no loop. Messor (‘harvester’) is a common cognomen; presumably Messor ‘signed’ the barrel because he had made it.

AMIN for Amin(eum) is well attested by the dipinti on Gallic amphoras published by Bernard Liou, six from Fos (Archaeonautica 7 (1987), 72–6, F103, F104, F105, F106, F107, F111) and two from Lyon (ibid., 141–2, L1 and L2); the name is usually abbreviated, but F103 gives it in full as AMINNEVM. As Liou notes, this was a famous grape-vine variety: Pliny gives it first place (Hist. Nat. 14.21), principatus datur Aminaeis (vitibus). Previous evidence of its export to Britain is the amphora graffito AMINE from Caerleon (RIB II.6, 2493.12).

Amin(eum) is qualified as MASSIL, which must be Massil(iense), ‘from Massilia’ (Marseilles). This is paralleled by the Fos dipinto (F111) which reads MAS | AMIN, although Liou understands it as Mas(sicum) Amin(neum) since it was found with another (F108) which explicitly reads MASSICVM (a famous Campanian wine). But previously, since all these MAS dipinti were found quite near Marseilles, Liou had understood MAS as Mas(siliense) (Archaeonautica 2 (1978), 147, nos 31 and 38; 149, no. 40). The wine of Massilia was outstanding, according to Athenaeus (Deipn. i.27c, Μαϲϲαλιήτηϲ), and it was exported to Rome (Martial, Epig. 3.82, 10.36 and 13.123, who calls it Massilitanum). Since MASSIL, not MASSIC, is undoubtedly the reading on the London barrel-head, it would now seem that Liou was right first time, when he expanded MAS to Mas(siliense). A further complication is that Damian Goodburn has noticed that the inner face is coated with black pitch, not the yellow resin used to seal wine barrels, and suggests it was a wine barrel re-used to import pitch used for caulking.

³⁸ With the next item during excavation by Pre-Construct Archaeology (Britannia 49 (2018), 382). Lucy Whittingham made them available. They will be published in the final report with eight other graffiti on pottery, either ‘crosses’ and marks of identification, or literate graffiti of less than three complete letters.

³⁹ A photograph by infra-red revealed no more than fig. 16, which was taken by sunlight. From its position, this is titulus β (see RIB II.6, pp. 1–2, fig. 1) which names the shipper(s) in the genitive case. The first letter is usually exaggerated in height, as it is here: apparently a long descender tending right (\), with a short diagonal just to its left near the bottom; this would suit either M or A, but M is suggested by a second such long descender. To the right of these two descenders (\\), the next letter is represented by three very short horizontal strokes in line vertically, the remains of E. This is followed by what looks like another pair of descenders (\\), much shorter than the first pair, but taller than the letters which follow; perhaps a second M. The traces thereafter are illegible, except for what looks like V towards the end. The CEIPAC database (ceipac.ub.edu) lists many names in titulus β which begin with M for the praenomen M(arci), but none in which M is followed by E for a nomen in E–. The only possibility is the label of the two Memmii which is attested by four examples from Monte Testaccio in Rome (CIL xv 3965–8), but neither CIL nor Rodríguez Almeida (in his ‘I mercatores dell’ oleo della Betica’, MEFRA 91.2 (1979), 873–975) provides a drawing which would allow close comparison. This firm is dated by two of its sherds (CIL xv 3965 and 3966) which also carry a titulus δ dated by the consulship of a.d. 149, a date which suits (905), the context of sf 258, which has been independently dated to a.d. 120–140/60.

⁴⁰ Because of the break, it is uncertain whether the series is complete or not. Each figure consists of a vertical stroke topped by two shorter strokes meeting at right-angles. The first two (from the left) are less neatly made, but the same figure seems to be intended throughout.

⁴¹ During excavation by Archaeology South-East at Centre Buildings, London School of Economics and Political Science, before redevelopment. Isa Benedetti-Whitton sent a photograph and other details. The fragment will be stored by London Archaeological Archive and Research Centre.

⁴² N is incomplete, but is not A or M, since the first stroke would have been diagonal. For other dated bricks and tiles, see RIB II.5, 2491.9–15. The original size of the brick is unknown, but if it were a bessalis c. 200 mm square, it would not have been wide enough to write either Nonas or Iunias unabbreviated. The more likely reading is therefore III N[on(as)] | Iu[lias], ‘5 July’, since it would have been just wide enough for Iulias, but not for Iunias unabbreviated. But this cannot be certain, and there may also have been a third line, for a name or batch-total.

⁴³ With the next two items during excavation by Pre-Construct Archaeology (Britannia 47 (2016), 341–2). Victoria Ridgeway made them available. They will be published in the final report with 19 other graffiti on pottery, either ‘crosses’ and marks of identification, or literate graffiti of less than three complete letters.

⁴⁴ Only the leftward tip of T survives. The first surviving letter (a vertical stroke) is also incomplete, but with its slight curve looks more like part of A than I or II (=E); and there is no well-attested name in –ilent– or –elent–. The name is thus probably [V]alent[is] in the genitive case, or a cognate name such as [V]alent[inus]. Compare RIB II.7, 2501.558 (also London): VALIIN[…].

⁴⁵ In the broken edge to the left of N is the tip of the previous letter; probably E made with two downstrokes (II). The downstroke of L trails off to the left, and the letter is completed by a short diagonal second stroke, detached and misplaced at mid-letter height. The sequence –nial– is almost unique to Genialis, which is a common name and abbreviated thus to GENIAL in RIB II.7, 2501.208 (Caerleon) and 209 (Brecon).

⁴⁶ For other instances of the Christian monogram on coarseware, see RIB II.8, 2503.132–6. The constituent letters do not quite coincide, which might suggest instead a note of weight of the form p(ondo) X… (see 2503.25–29), but in these the numeral is quite separate from the P. In any case, a mortarium is unlikely to have needed a note of weight, and this mortarium (Young type C100) is fourth-century, when a Christian graffito would not be unusual.

⁴⁷ PAS ref. LON-B37E05. AVCISSA brooches were imported in great numbers during a.d. 43–61, probably from Gaul, and are often but not exclusively found on military sites (J. Bayley and S. Butcher, Roman Brooches in Britain (2004), 151). There are 41 examples in RIB II.3, 2421.

⁴⁸ PAS ref. WMID-746CF7.

⁴⁹ A love-token, with a message for the recipient to complete, discussed with parallels by John Pearce in his draft paper ‘si amas: emotional literacy in Roman Britain’. A possible expansion (from Britain) is offered by the copper-alloy brooch from Richborough which reads: SI Λ|MΛS | EGO P|LVS, ‘If you love (me), I love (you) more’ (RIB II.3, 2421.50, with a Gallic parallel in CIL xiii 10027.150a); other possible expansions are offered by Gaul and the Rhineland, notably the copper-alloy ring published by Fontaine, P. in Annales de la Société Archéologique de Namur 86 (2012), 22–4Google Scholar, which reads: si amas pignus amoris escipe, ‘If you love (me), accept (this) pledge of love.’

⁵⁰ By Bowman, A.K., Thomas, J.D. and Tomlin, R.S.O., ‘The Vindolanda writing-tablets (Tabulae Vindolandenses IV, Part 3): new letters of Iulius Verecundus’, Britannia 50 (2019), doi:10.1017/S0068113X19000321Google Scholar.

⁵¹ With the next eleven items (except for No. 38, which was found in 2012) during excavation directed by Andrew Birley, who made them available. It is the second diploma to be found at Vindolanda, the first being RIB II.1, 2401.9.

⁵² Three names which date the diploma since, as Paul Holder points out, they are those of the fifth, sixth and seventh witnesses known to have signed in that order between 13 December 156 (CIL xvi 107) and 8 July 158 (CIL xvi 108). By 27 December 158 the witnesses had changed.

⁵³ Lines 3 and 4 are damaged by a break. The fragment will be published by Paul Holder, who comments that the script on the inner face is typical of the reign of Antoninus Pius in being angular and irregular in size with large serifs, making it difficult to distinguish between the letters I, L and T; it is also unclear whether all the letters were fully inscribed. But he can identify the descriptum et recognitum formula which copied the text from the original bronze tablet, [ex tabul(a)] aer(ea), posted at Rome near the temple of Minerva, [ad Mine]rv[am]. Above it would have been inscribed the names of the recipient, his unit and his commanding officer, but these are lost; however, line 2 may be the end of his place of origin.

⁵⁴ Inscribed above a horizontal line drawn across two-thirds of the diameter. The loops of P and R were not completed, and S is only a long vertical stroke (but I is precluded by the preceding V). Underneath, within the foot-ring, is the natural place for an ownership inscription, so Primus was probably the first owner, especially since the horizontal scoring suggests that his name was subsequently deleted.

⁵⁵ A was made with a short vertical third stroke, and the loop of R was not closed.

⁵⁶ Lines 1–2 must be the nomen and cognomen of the man who ‘made’ this brick, the subject of [f]ecit in 3; there is a space after this verb, which may have been centred as if to end the sentence. There is no obvious nomen in …anius, but many possibilities; Vindicatus, although rare, is almost the only possible cognomen. Coincidentally, Canio and [Vi]ndicatus occur together on a quarry-face graffito in the area (RIB III, 3223).

Line 4 introduces a second nomen, probably [Calp]urnius, since it is much more common than Liburnius. It suggests that 4–6 were written by a second person, especially since the lettering is smaller and more cramped, and there are differences in letter-form, especially A, V and S. But [Calp]urnius is not followed by another cognomen, to judge by 5, which would end in [h]omo (‘man’), if the short horizontal stroke is not damage but the remains of H. Nor does 6 look like a verb or numeral: it may begin with a long horizontal Q, followed by what looks like deis s[ui]s (‘to his gods’) if the next letter is a vertical D quite unlike that in 2. But none of this makes much sense, and 5–6 are really too damaged and incomplete to be legible.

⁵⁷ The capacity of a Dressel 20 is usually expressed as a numeral after (or implying) M for m(odii), but unusually, this is a note of weight like those inscribed on coarseware vessels using the TP PP formula (RIB II.8, 2503.11, 14, 16, 20, 22, 28). T is ligatured to P, as in 2503.16, and P to P. Each combination is followed by a numeral, now incomplete, which expressed the weight (in pounds) empty and full. The letter after ligatured PP is cut by the broken edge, but is apparently a rather small C (‘100’); to read the upper half of S would be difficult, quite apart from the difficulty of explaining a numeral which began with s(emis) (‘half’). This may be the second such note of weight from Vindolanda: compare the amphora RIB II.6, 2494.161(a), with its unexplained graffito PP. Two numerals are accepted as notes of weight by RIB II.6, 2494.89 and 90, but for a complete example (with P and PP) see Britannia 36 (2005), 484, no. 16 (Stonea).

⁵⁸ The rightward diagonal of X is faint towards the end, but V cannot be read. The next letter might be either A or R, but there is a hint of a third stroke for N. Finally a vertical stroke for I, unless it is an incomplete II (for E). The initial sequence DAX would suggest Audax, which is well attested at Vindolanda, but there is no sign of V to the left of D. The best possibility seems to be the rare Celtic name Daxna (or a variant), which is found as a potter's name in Gallia Belgica (Carte Archéologique de la Gaule 59.2: Le Nord, Bavay (2011), 120 and 239; perhaps with 347, an amphora graffito).

⁵⁹ Like all seven of these samian graffiti. 16 other samian graffiti were found in 2018, but they are either ‘stars’ or other marks of identification, or literate graffiti of less than three complete letters.

⁶⁰ K is not one-half of the ‘star’ often found on samian vessels within the foot-ring as a mark of identification, but a single letter, the initial letter of the owner's name. This was probably K(arus); see further, the next item. The left-hand half of the base is lost, but it is most unlikely that it carried another letter to balance K.

⁶¹ There is apparently the tip of a third letter (R, presumably) in the broken edge to the right. An area of damage above K raises the possibility that this letter is an angular R, but K looks more likely. Also compare the previous item. In written Latin, initial C was quite often replaced by K: indeed, it was the rule according to ancient grammarians (J.N. Adams, The Vulgar Latin of the Letters of Claudius Terentianus (1977), 32). At Vindolanda, Cerialis’ correspondent Claudius Karus, probably an equestrian prefect like him, wrote his cognomen as Karus (Tab. Vindol. 250 and 251, with note); and in the letters, the form karissimus is much more common than carissimus. For instances of Karus for ‘Carus’ elsewhere, see RIB II.7, 2501.281, 282 and Britannia 43 (2012), 417, no. 39.

⁶² The cognomen Levis is almost unknown, and the only other possibilities are non-Roman, whether a derivative of Celtic *leuco-s (whence the Gallic tribe of Leucones and the nomen Leuconius) or the Germanic name Leubasnus. The latter is attested in Belgica and the lower Rhineland, and is an attractive restoration here because of RIB 1619 (Housesteads), the tombstone of Hurmius son of Leubasnus, mil(es) coh(ortis) I Tungror(um). His cohort was previously based at Vindolanda, and Leubasnus was evidently a Tungrian personal name.

⁶³ Primus (with its derivatives) is a common name, but for other instances from Vindolanda, see the pewter bowl above (No. 31), the next item, and RIB II.7, 2501.448, Tab. Vindol. 180.28 and 181.6.

⁶⁴ The loop of R was made twice. M is incomplete, but although the sequence –ria– is conceivable, the name Primus is so common (see previous note) that M is almost certain.

⁶⁵ There is just enough space to the right to suggest the graffito is complete. R is awkwardly made with an extended vertical and a square ‘loop’, but can hardly be P or ligatured RI. The cognomen Varus is quite common, but (again as VAR) has already occurred at Vindolanda as a graffito (Britannia 40 (2009), 353, no. 96).

⁶⁶ R is coarsely incised, but can be distinguished from B. VIIR is conceivable as an abbreviated cognomen like VAR (see previous item), but there seems to be the tip of another letter in the broken edge, which looks more like a diagonal for V than a vertical for I or I[I]. The cognomen Verus is very common.

⁶⁷ By a metal-detectorist (PAS ref. LANCUM-6F9EF1). Information and photograph from Richard Hobbs and Lydia Prosser, Finds Liaison Officer for Lancashire and Cumbria.

⁶⁸ The initials of tria nomina, in which the praenomen Gaius is quite often abbreviated to G, not C, as already noted for the Piercebridge lead sealing G M A (Britannia 49 (2018), 439, no. 21 with n. 35). But nomina in B(…) are not very common, and it is conceivable that the owner abbreviated his nomen to G(…) and his cognomen to Br(…) so as to distinguish it better.

⁶⁹ PAS ref. SF-4A1C36.

⁷⁰ In Britain, the abbreviation D M can be attributed by other evidence to Mars (RIB 305, 617, 1024, II.4, 2448.3, 3500) or to Mercury (RIB 952, 1304); only in RIB 1597 is it ambiguous, although Mars is likely, since it was found in a barrack-block. This ring was probably dedicated to Mars or Mercury, therefore, but there are other possibilities such as the goddess Minerva, d(eae) M(inervae).

⁷¹ During field-walking by Coventry Historic Environment Group. Information from Ken Battersby, who sent photographs and a report by Ray Wallwork in 2015 describing the 60+ kilos of brick and tile fragments found in ‘Field 109’.

⁷² The cross-stroke of T is slighter than the other strokes, and detached from its vertical, but it seems to be intentional, which would exclude the name Iallus. The surface is damaged in the right-hand corner after TALLI, making it uncertain whether a letter has been lost here (for a variant spelling of Talio, since *Tallis is not attested), but Talli is quite acceptable as the tile-maker's name in the genitive. TALLV occurs as a samian graffito from London (RIB II.7, 2501.533), and is taken to be Tallu(s), a name for which there is a little evidence; however, it is probably a variant spelling of the well-attested Thallus. This has not yet occurred in Britain, but compare Thalius (RIB II.8, 2503.157).

⁷³ During excavation by Glamorgan-Gwent Archaeological Trust (Britannia 44 (2013), 280) before rebuilding by Monmouth School. The sherd was identified by Edward Biddulph and details, including photographs, were sent by Joyce Compton.

⁷⁴ The woman's name is probably Accepta or Recepta. Both are quite common, and (in the masculine form) are attested in Britain: Ac(c)eptus at Vindolanda (Tab. Vindol. 864) and Receptus at Bath (RIB 155). Despite the broken edge, the downward extension of the final letter suggests that Marci (‘of Marcus’) is complete, but his relationship to […]cepta might then have been made explicit, whether she was his daughter, wife or slave.

⁷⁵ In the demolition layers during excavation directed by Geoffrey Bailey, who made it available (Falkirk Museum, acc. no. 2018-10-39).

⁷⁶ In scratching the letters, the scribe abraded the surface quite widely, but their general form is clear. The drawing emphasises the surviving traces of incision. The outline of the broken surface to the left of C suggests the diagonal tip of V rather than the straight stroke of E or II, but there is no actual incision. Iucundus is thus the most likely name, but Secundus or even Verecundus remain possible.

⁷⁷ This item and the next were noticed by Fraser Hunter, who made them available in a re-examination of the graffiti which will form part of a new publication to mark the centenary of the Treasure being found.

⁷⁸ The first letter is a triangular D which leads into the exaggerated lower loop of E. The scribe then lifted his stylus and made a second, angular loop at right-angles, forming the two-hooked E typical of New Roman Cursive. Within it, he made a shallow V, which led into a square O made with two right-angled hooks. The next letter is apparently R made with a repeated downstroke, leading into an angular M. Finally a short horizontal stroke leads into a short downstroke, presumably for I. These letters were distorted by being made awkwardly in a cramped space, and their reading and interpretation remain uncertain. Presumably a personal name in the genitive case, but there is no obvious candidate.

⁷⁹ This corrigendum and the next were noted by Scott Vanderbilt.

⁸⁰ By Paul Holder in C. Ambrey et al. (eds), A Roman Roadside Settlement at Healam Bridge: the Iron Age to Early Medieval Evidence, II (Artefacts) (2017), 156–61. He sent full details, including the next note.

⁸¹ The date is certain because Antoninus Pius is still COS III (his fourth consulship was in a.d. 145), the formula ipsis liberis posterisque eorum was replaced on auxiliary diplomas at the end of a.d. 140, and siqui caelibes essent disappeared from auxiliary diplomas at the end of a.d. 140 except for one of a.d. 145 (CIL xvi 91). But a new fragment for the auxiliaries in Dalmatia which was issued by Antoninus Pius as COS IIII (and thus between a.d. 145 and 161) is dated by the suffect consuls Q. Antonius I[sauricus] and L. Aurelius Flaccus. See Eck, W. and Pangerl, A., ‘Zu den Konsulnfasten der Zeit des Antoninus Pius. Konsequenzen aus einem neuen Militärdiplomen’, Studia Europaea Gnesnensia 16 (2017), 91–101Google Scholar, who argue that Isauricus must have been consul in a.d. 156 or 157. The Antonius of the Healam Bridge fragment thus remains unidentified.