² By a local resident in her garden, and reported to Robert Wallace, director of the Culver Archaeological Project, who sent photographs and other details. The project excavates locally in the Ouse valley, its sites including the Romano-British roadside settlement at Bridge Farm a mile away, which would seem a likely provenance.

³ IMP would suggest some case of imp(erator) (‘emperor’) abbreviated, but this would usually be the first word of a building inscription, not towards the end, and nothing can be made of NI which follows. There is a faint suggestion of a diagonal below the curve of P, as if for R, but this letter-sequence would be even more difficult. There is no sign of any ligaturing, which would be unlikely in what seems an early and well-spaced inscription. […]COLA at the very bottom is followed by a space, as if for a word-ending, and although there are other possibilities, such as Publicola and Silvicola, much the most likely name is [Agri]cola. This would suggest a reference to the governorship of Cn. Iulius Agricola (a.d. 77–84), which the quality of the lettering would support.

⁴ By R.E. Hooppell, the excavator of South Shields, who reported his reading to Watkin, W.T. (The Archaeological Journal 42 (1885) 145)CrossRefGoogle Scholar. This reference was noted by Lindsay Allason-Jones, who tracked the fragment to its present location and sent photographs. She will publish it fully in CSIR 1.11, but it is not known why it was not included in RIB, even if only ‘sought in vain’.

⁵ There is apparently some trace of the initial M. Another damaged fragment of an altar dedicated to the Matres was found in 1866 at Skinburness (RIB 881), 11 miles to the north.

⁶ Portable Antiquities Scheme (PAS) ref. ESS-1B508F.

⁷ Despite there being no medial line, this is probably a ‘Chnoubis’ amulet like the finger-rings from Thwing (Britannia 40 (2009) 355, no. 99) and Clare (no. 29 below). For SSS (barred), compare the Woodeaton gold amulet (Britannia 1 (1970) 305, no. 1, published with commentary by Kotansky, R., Greek Magical Amulets 1 (1994) 13–15CrossRefGoogle Scholar, no. 3) and the Eccles lead tablet (Britannia 17 (1986) 428, no. 2, face (a)).

⁸ During excavation by AC Archaeology (Britannia 49 (2018) 390), in the fill of the defensive ditch and in a context dated by pottery to a.d. 60/65 to c. 120, which would suit the Castleford fragment (see next note) with its context of a.d. 71/74–86. Denise Allen sent a photograph and other details, and Hilary Cool discussed the reading.

⁹ Hilary Cool has seen that it matches the fragment of a moulded bottle base from Castleford (no. 30 below) and has made a composite drawing (fig. 4b). The space between FE[CIT] and [V]ERECVNDA of about five letters would have contained Verecunda's nomen, and IVLIA (although it cannot be certain) fits perfectly. Iulia Verecunda is well attested as a name and would suit a woman in a Celtic-speaking province whose family had been Roman citizens for some generations. Moulded bases name two other female glassmakers: Sentia Secunda (AE 1999, 1214 (a), (b)) and Ennia Fortuna (CIL 15.6993 with Archäologischer Anzeiger 2 (2019) 149–56).

¹⁰ In the same excavation as the previous item, but in a demolition layer dated to a.d. 140–200. Naomi Payne sent a photograph and other details. There are three other sherds from this jar, one of which carries the end of two long diagonal strokes with space to the left, at about the same height on the wall, but they are unlikely to be the beginning of the graffito since they are quite different in character from its small, neat capitals. Presumably they belong to another graffito.

¹¹ The first letter is a short vertical stroke set above the end of a downward diagonal, which is not an incomplete N to judge by the penultimate N. The two strokes are possibly an incomplete V, but more likely to be I above the end of another letter: this might be L (with a diagonal second stroke), M or R. Cognomina in -enus are uncommon, but perhaps this is an error for -ianus.

¹² According to Bloomsbury Auctions (Dreweatts) in the description of Lot 9 in their online catalogue of 8 July 2020, when it was sold with other items from the Schøyen Collection, in which it was MS 2344. The cataloguer, Tim Bolton, sent a photograph and other details. It was bought from Lennox Gallery in 1996, but its previous history is unknown. It is now in a private UK-based collection.

¹³ There is no spacing in the original except between D and D in line 5. The dedicator's name is restored from RIB 1077, where it is spelt as Auffidius Aufidianus. In the present item, AVFIDI in AVFIDIANVS was omitted by oversight. Otherwise, letter-style and layout are the same, except that in RIB 1077 D for d(ono) and d(edit) is inscribed within each ansa, which in the present item are blank except for a dot-punched outline. The other British dedications to Victoria Augusta are RIB 842 and 843 (Maryport) and 1138 (Corbridge), for which cult see the note to RIB 1138+add. AVG is abbreviated as usual, but since this plaque is paired with one dedicated to MARTI AVG, Marti Aug(usto), VICTORIAE AVG should be understood here as Victoriae Aug(ustae), not Aug(usti).

¹⁴ With the next two items during excavation by Cotswold Archaeology, and made available by Jacky Sommerville for publication in the final report.

¹⁵ It looks as if there was no letter after V, but not quite enough of the surface survives to be sure. It might then be a numeral (‘55’ or more, presumably a batch-total) or the potter's name abbreviated, perhaps Silu(anus). When this is the name of the god Silvanus, it is quite often abbreviated to SILV, in RIB 972 for example.

¹⁶ The intersection of the strokes shows they were made from left to right as drawn (fig. 8), by drawing the stylus down each time from near the centre to the circumference. The first stroke is not quite touched by the second, and might be read as I, but names in Im(…) are so rare that it was more likely combined with the first stroke of M to make an ‘open’ (unbarred) A, especially as the two strokes are angled towards each other, whereas if the first were I, it should have been more or less parallel. There are many names in Am(…) and some are quite common, for example Amabilis, Amandus and Ammonius, all of which have been found in Britain.

¹⁷ The first of the three letters is now only the tip of a downward diagonal which would best suit A, but might be R or even L. Possible names include Iasus, Celsus and Ursus, all of which have been found in Britain. Iasus is usually spelt Iassus (see note to RIB II.8.2503.277, IɅSI), but Celsus and Ursus are both quite common.

¹⁸ During the excavation published in A. Woodward and P. Leach, The Uley Shrines: Excavation of a Ritual Complex on West Hill, Gloucestershire: 1977–9 (1993), in which it is noted on p. 130 as no. 83. It will be published with fuller commentary in the final report being prepared on the inscribed lead tablets, but is included here as a tailpiece to the tablets published last year (Britannia 51 (2020) 480–3, nos 12, 13) which also relate to sheep theft.

¹⁹ Line-by-line commentary

1. [de]o. It is most unusual not to name the god, but he is undoubtedly Mercury.

1–2. Pecti[ll|u]s. The petitioner's name is restored from its only other instance, RIB 307 (Lydney).

2–3. ue[rbe]|ces. The restoration is certain, since the word is repeated in line 8. For these ‘wethers’, see the note to line 8.

3. perd[idit]. ‘Pectillus’ is the subject, the verb perdere being used by British curse-tablets in the sense of to ‘lose (by theft)’.

4. DITIT. There would have been enough space for the whole of perdidit in line 3, so DITIT must be superfluous; perhaps it is a would-be correction. The repeated letters in the perfect tense of perdere bothered scribes, for example in Uley 4.3 (Woodward and Leach, op. cit. (n. 18), 125), where perdidit was written as PEDIT.

4. don[at]. ‘Pectillus’ is again the subject, the verb donare being used by British curse-tablets in the sense of ‘giving’ the thief or the stolen property to the god, and by extension (as here) in the sense of getting the god to work the curse. The space available suits don[at] better than don[auit].

5. ut illum. VT is preceded by a ‘cross’ which is presumably the crossing-out of T repeated by mistake. illum (‘him’) must be the thief, with ut introducing the ‘prayer’ implied by don[at]. The incomplete letter after ILLVM might well be Q, for the relative pronoun q[ui], but the substance of its clause has been lost in the next line.

6. This line is too damaged by the fold to be recovered but must have ended in IN for [in]|uolauerunt. The plural is a solecism: petitioners, thinking in formulas, sometimes became confused whether to refer to the unknown thief in the singular or the plural.

8. uerbeces. This noun is repeated from 2–3, where it is the object of ‘loss’, a conclusion reinforced here by the complementary verb [in]|uolauerunt (‘have stolen’). The form uerbex is well attested as a ‘Vulgar’ variant of the Classical ueruex (‘wether’, a castrated male sheep eaten as mutton), but should not be seen as evidence of local confusion between [b] and [v], since in the Acts of the Arval Brothers at Rome, uerbeces (thus spelt) are specified as objects of sacrifice, for example in ILS 5047. Sheep and goats dominate the very rich animal-bone assemblage at the Uley temple, prompting the estimate that 150 were sacrificed every autumn for 50 years, a reflection of ‘the wealth of the region’ (Bruce Levitan, in Woodward and Leach, op. cit. (n. 18), 300). For other stolen sheep, see Britannia 51 (2020) 480–3, nos 12, 13.

9. [s]ine sangu[i]n[e]. When British curse-tablets refer to ‘blood’, they require the thief to make payment ‘with his blood’, but the restoration of [s]ine here is compelling, which would imply that the wethers were to be returned unharmed, ‘without blood(shed)’.

10. The traces are illegible but were probably a verb such as reduca(n)t, requiring the thief to ‘bring back’ the stolen wethers unharmed. This verb is used in Tab. Sulis 64.4 for the ‘return’ of a stolen cloak.

²⁰ With the next item during excavation directed by David Wilson for the University of Keele, and noted with drawings in his third interim report in Glevensis 22 (1987) at 57, fig. 12c and 12f, respectively. See further, n. 23 below.

²¹ Inscribed by the same hand as the third Wortley tile (no. 13 below), which would suggest that, like Iuliis, Octobres was not abbreviated here. In the second line, the upper stroke of D is unusually flat, as if to avoid the numeral in the line above. It is apparently separated from CLV[…] by a space, which makes it less likely that the sequence is a batch-total, ‘655 (or more)’; but a further reason to doubt this is the tile-stamp DCLVI found locally at Berkeley (RIB II.5.2489.54). Its meaning is unknown, but a numeral (‘656’) would be inappropriate in a tile-stamp; perhaps it is an abbreviated name such as D(ecimus) Cl(audius) Vi(talis).

²² Only N is complete. It is not AL, since there is no descender for L, but, if taken with the two incomplete strokes in the broken edge to its left, it looks very like KAL on the other two Wortley tiles (nos 11, 13). The difficult area to its right can be explained as two attempts to make the loop of D with a double stroke, as in no. 11 above, by supposing that the missing diagonal was not flat, but conventionally diagonal or upright. It would seem that no. 12 is by the same hand as the other two, and names the tile-maker. In written Latin, initial C was quite often replaced by K: see Britannia 50 (2019) 516, no. 36 (with note).

²³ Later in the series of excavations noted for the two previous items. All three were deposited in the Museum in the Park, Stroud, from where Alexia Clark sent photographs and other details. They are noted with drawings in D. Wilson et al., Report on the Excavation of a Romano-British Site in Wortley, South Gloucestershire (2014) 115, fig. 32, as nos 287, 291 and 290, respectively. A fourth tile with graffito, found in 1991, is noted there with a drawing as no. 289, but it has not yet been located and no photograph is available.

²⁴ As noted above, this was written by the same hand as the first graffito (no. 11). The published drawing (see previous note) omits the lower tip of S in the right-hand edge.

²⁵ Later still in the series of excavations noted for the three previous items (n. 23 above), where it is no. 294 (with fig.). It is now in the Museum in the Park, Stroud (2007.26/37), from where Alexia Clark sent photographs and other details. Although described in Wilson et al., op. cit. (n. 23), as a ‘pedestal footring’ probably from the Severn Valley kilns, its profile suggests a lid like that in P. Tyers, Roman Pottery in Britain (1996) 199, fig. 253.27, although this is twice the diameter.

²⁶ B and F are well drawn, F with even a bottom-serif, but the next two letters are simply two angled strokes. The second is surely L, not C, since the letter B and the decoration show that the potter could draw a semi-circle. Wilson et al., op. cit. (n. 23), suggest it may have been a bungled copy of a samian stamp, with FECI[T] instead of F and the potter's name ending in B, but this is difficult. Other possibilities are [… omni]b(us) feli[citer] or [Fa]b(ius) Feli[x …].

²⁷ By metal-detector, PAS ref. LEIC-62F9AF. Details from Sally Worrell, who also sent comments by John Pearce.

²⁸ Each letter or digit was made with one or two intersecting blows of a chisel, so that even C is angular. The first four digits are quite clear (CCCX), but the others become progressively more faint. One of the Carsington lead pigs (RIB II.1.2404.56, with footnote) is incised CCX in similar digits for ‘210 (pounds)’. It now weighs 65.3 kg, which is almost 210 librae (68.7 kg). Other pigs with numerical annotations corresponding quite closely with the weight are RIB II.1.2404.69 and 71, and this is the likely explanation of ‘336’, but it would imply that about one-third of the ingot has been cut off, since 336 librae would be 110 kg. This weight would be exceptional according to RIB II.1, p. 38, since British pigs average c. 74 kg but ‘many weigh over 80 kg and one [RIB II.1.2404.16] weighs as much as 101 kg’. But the ingots of British origin in a Brittany wreck include at least one with similar digits amounting to more than ‘300’ (CCC): see M. L'Hour, ‘Un site sous-marin sur la côte de l'Armorique: l’épave antique de Ploumanac'h’, Revue Archéologique de l'Ouest (1987) 113–31, fig. 13, where the numeral is not transcribed but is evidently CCC … V.

²⁹ With fragments of six other tablets and a recessed label like Tab. Lond. Bloomberg 183, but not inscribed, during excavation by Museum of London Archaeology (MOLA) (Britannia 26 (1995) 361), which made them available with full details.

³⁰ This looks like an ‘address’, but the genitive is difficult and the crossing-out on the other face suggests a loan-note which has been ‘cancelled’ after payment. It may have been identified by the outer text, which certainly ends in VɅNI. This would suggest the end of a cognomen in the genitive case, probably Siluanus (‘of Silvanus’), but SIL should not be read immediately above it, since this line would have extended across the width of the tablet. However, it probably ended in SIL. A possible explanation of the genitive [Sil]uani would be that a cavalryman was identified as belonging to ‘the troop of Silvanus’: TVR SILVANI, for tur(ma) Siluani. This already identifies two of the witnesses in Tab. Lond. Bloomberg 62, which is only dated by its archaeological context as c. a.d. 80–90/95 or earlier residual, but probably refers to one of the Rhineland units that reinforced the British army after the Boudican revolt (a.d. 61). This would suit the context of no. 16, which is infill associated with Neronian quay construction (a.d. 63–64).

³¹ With the next item and six other fragments with almost no trace of text, during excavation by MOLA (Britannia 32 (2001) 364), which made them available with full details. Since both items (nos 17 and 18) came from the same context and are the same in width, it is possible they were once paired together.

³² The graffito is complete, the surrounding surface showing no sign of any letters. It is too slight to have served as an ‘address’ but is probably an abbreviated name identifying the owner of the tablet or perhaps the text within. There are a few cognomina in Ven(…), but the most likely is Venustus. A similar graffito, VIIN, was found not far away at the Bank of England, as the ownership-inscription underneath a samian platter (Dr. 18) dated by its stamp to c. a.d. 70–90 (RIB II.7.2501.581). Since no. 17 was found in well-fill dated to a.d. 120–200, it is probably later in date, but might refer to the same man.

³³ The phrase is formulaic in loan-notes and acknowledgements of debt, as part of a promise to pay the money to the creditor ‘or to whom the matter will concern’. For other examples from London, see Tab. Lond. Bloomberg 44.8 with note. Since it occurs at the top of the ‘page’, this must have been the second ‘page’. The first ‘page’ may have been the recessed face of no. 17 above (see n. 31), but this is now blank except for a few incisions and an s. If they did belong together, the document they contained was written only once, not duplicated in a three-leaf ‘triptych’.

³⁴ During excavation by MOLA (Britannia 47 (2016) 338), deducing the date from the site code SGA12. The fragment of another tablet was found, but there is no trace of any text.

³⁵ The adjective primas (‘first’, feminine accusative plural), reading m despite two extraneous strokes. In stylus-tablet financial texts from Pompeii primas often qualifies the ‘next’ date (in kalendas Augustas primas, for example) as the deadline for an action such as the repayment of a debt.

³⁶ During excavation by the Department of Urban Archaeology, Museum of London, on the site (MFI87) of the Docklands Light Railway Monument Street station. It was in the backfill of a well of Flavian date abandoned in the late first century which was soon filled with domestic refuse. It is now in the Museum of London, from where Mark Burch, Richard Dabb and Francis Grew provided a photograph and other details.

³⁷ The association of an apotropaic phallus with ‘luck’ and ‘happiness’ is also explicit in the stone panel from Pompeii that depicts a phallus with the legend HIC HABITAT FELICITAS (CIL 4.1454; for which see C. Johns, Sex or Symbol: Erotic Images of Greece and Rome (1982) 65, fig. 47).

³⁸ With the next item during excavation by MOLA (site BJG10) and made available by Michael Tetreau. They will be published in the final report with three other graffiti, a ‘star’ and two literate graffiti of fewer than three letters.

³⁹ All three letters are incomplete, but the diagonal stroke of N cuts its first stroke, showing that the graffito was written from left to right, as drawn (fig. 21). The downward axis of the writing would also suggest there were only three letters below the rim, the space to the right of N marking the end of the word. Since names ending in –on are uncommon, this is probably an abbreviated name. All that remains of the first letter is an upward stroke trailing off in the direction in which it was made; this would suit C, if it were made with two strokes, but F, G and S are also possible. However, it is difficult to envisage a name abbreviated to FON, GON or SON, whereas CON is quite acceptable: compare the graffito CON underneath a goblet in the Traprain Treasure (RIB II.2.2414.16) and the Birdoswald centurial stone (RIB III.3451), which abbreviates the centurion's cognomen as 7 EPPI | CON.

⁴⁰ The first letter is now only the top of a vertical stroke. It is too close to A to be L for lago[nam] (‘jug’; compare RIB II.8.2503.111) and is probably too vertical for V, but it might be H, I or N. These would all suggest a Greek personal name such as Pythagoras, Diagoras or Athenagoras, the most common being Athenagoras; but none of them is yet attested in Britain.

⁴¹ Unstratified (sf 95340) in the excavations noted by V. Crosby and L. Muldowney, Stanwick Quarry, Northamptonshire: Raunds Area Project: Phasing the Iron Age and Romano-British Settlements at Stanwick, Northamptonshire (2011). It will be published more fully by Historic England, from where Rachel Cubitt made it available with details from Vicky Crosby.

⁴² The letters are angular because they were made with straight strokes, vertical or diagonal. Thus D, now incomplete because of the abraded surface, was made with three strokes intersecting to form a triangle, an angular version of the lower-case form like D in one of the Uley tablets (see fig. 23c for comparison). To its right is E, also made with three angled strokes; there is no upper stroke for ‘capital’ E, since this too is ‘lower-case’. The incomplete letter to the left of N is apparently another such E; the second nick in the broken edge looks deliberate, which excludes V.

The graffito is presumably the owner's name, that of a woman, possibly Iulia Britiuenda or Cunouenda. The sequence […]endeiu[…] does not suggest a personal name, but the second E may represent the feminine name-ending –ae in the genitive case, ‘(property) of […]a’, with ‘Vulgar’ e for ae since the pronunciation was the same. Latin names in –endus/a are uncommon, making a Celtic name more likely, for example Britiuenda (Tab. Sulis 2, 3) or the feminine form of Cunouendus (RIB II.5.2491.78; III.3477 and perhaps 3232; Britannia 40 (2009) 327, no. 21). To the right of E, IV would then be the beginning of the name, and perhaps IV[L] as an abbreviated Iul(iae), which would imply a family with Roman citizenship still using Celtic cognomina.

⁴³ With the next item, presumably in an early excavation. They are now stored with other amphora sherds from Housesteads (but these with only one or two letters each) at Corbridge Roman Site, from where Frances McIntosh sent photographs.

⁴⁴ Since Housesteads was garrisoned by Tungrians, Agad[…] may be an unattested Germanic name, but an attractive possibility is that it represents the name-element shared by Tagadunius and Tagadianus. These are both found at the shrine of Nehalennia on the Schelde, notably in AE 2001, 1452, the altar dedicated by Tagamas Tagadiani (filius). The Tungrian name Tagamas/Tagomas is well attested at Vindolanda (Britannia 34 (2003) 377, no. 37 with note). Immediately to the left of the first Ʌ in the Housesteads graffito are two faintly incised lines meeting at right-angles like ‘T’. Although they may be only casual, perhaps they represent a correction or a note of where to add the missing T.

⁴⁵ Atticus and its cognates, as Géza Alföldy notes (Epigraphische Studien 4 (1967) 10), are typical of Gallia Belgica and Upper Germany. They are well attested in Britain: for Vindolanda, see Tab. Vindol. III.586 (with note); for Corbridge, see RIB II.7.2501.74; Britannia 48 (2017) 466, no. 16. The second name (in the genitive) was probably Acceptus or Receptus, but there are other possibilities. The most likely relationship between the two names is father and son, peregrine nomenclature at an auxiliary fort, but Atticus might have been a Roman citizen who included his nomen and identified himself by his century (‘the century of […]ptus’).

⁴⁶ During excavation by the Vindolanda Trust directed by Andrew Birley and made available (sf 19407) by Phil Mills with eight other inscribed sherds (amphora and samian), but these all with only one or two letters.

⁴⁷ The incomplete first letter in the broken edge includes a vertical stroke intersected by a diagonal, at the same angle as those of the next letter, N, except that the diagonal extends further; almost certainly N, not Ʌ. Alfred Holder (Alt-Celtischer Sprachschatz 2 (1904) 19, s.v.) notes -icco- as a Celtic name-suffix Latinised as -iccus, and the sequence -nniccu- suits Congenniccus already attested at South Shields by RIB 1053, which notes CIL 12.4883 (Narbonne) as another instance.

⁴⁸ Like the next item during excavation by South Oxfordshire Archaeological Group and illustrated in 3D on its website.

⁴⁹ For other beakers with this motto, see CIL 13.10018, 106a–e.

⁵⁰ There is no beaker with this motto in CIL 13.10018, nor apparently elsewhere.

⁵¹ PAS ref. SF-9B8965.

⁵² Probably a ‘Chnoubis’ amulet like the Maulden ring (no. 3 above, with note).

⁵³ During excavation by West Yorkshire Archaeological Service published in three volumes as Roman Castleford Excavations 1974–85 (1998; 1999; 2000). In volume 1, Small Finds (edited by H.E.M. Cool and C. Philo), it is no. 288 in Part 3, ‘The vessels and objects of glass’ by H.E.M. Cool and Jennifer Price, at pp. 150, 174, fig. 61.

⁵⁴ The reading can now be restored by comparing the Exeter fragment found in 2017 (no. 4 above), since it preserves another part of the same inscription. The mould-maker failed to reverse the letter N.

⁵⁵ By Tom Shurlock with a metal-detector, who sent it with a photograph to Martin Millett. Unfortunately it was stolen in transit, and the photograph is now the only record.

⁵⁶ According to the photograph, the legend concluded with a bold L, its second element a downward diagonal. To the left of this, only faint traces: perhaps the lower curve of C for c(ohortis) followed by three or four verticals for the numeral. Sealings are known for the Second, the Third (perhaps) and the Fourth Cohort of Lingones in Britain, but all with a right-angled L: see, respectively, RIB II.1.2411.108, 106 (with M.C.W. Still, Roman Lead Sealings, unpublished PhD thesis, University College London (1995) 353, no. 457) and 109.

⁵⁷ PAS ref. YORYM-A9E79C.

⁵⁸ For many previous discoveries of TOT rings, see Britannia 50 (2019) 504, no. 16 (with note); 51 (2020) 485, 494, nos 17, 29.

⁵⁹ With the next four items and also Britannia 42 (2011) 459, no. 33 and 44 (2013) 393, no. 24 (for which see the addendum below), during excavation by Cardiff University and University College, London, directed by Peter Guest and Andrew Gardner for Cadw and the Roman Legionary Museum, Caerleon (Britannia 42 (2011) 323). Peter Guest made them available. They will be published more fully in the forthcoming final report, together with a tag (sf 3213) used twice and now illegible.

⁶⁰ The nomen Iulius is frequent among legionaries, and the cognomen Iucundus is quite common, but this may well be the same man as the next item (no. 34), although the reading there is not certain.

⁶¹ The tag was folded twice, stressing the surface which is also corroded. Three letters have been lost, but the cognomen is likely to be Iucundus or Secundus (in the genitive), both common; in view of the previous item (no. 33), it is more likely Iucundus. The handwriting of both tags is different, which implies that they were written for Iucundus by someone else, either because he was illiterate or (more likely) because they did not assert ‘ownership’ but identified something ‘for collection’ by him.

⁶² IVL is quite well preserved and separated by a space from the cognomen. This is much fainter but begins with a damaged V and an incomplete E (not L, since the L of IVL is quite different); then space for two letters, the second perhaps I; and ends in ɅTI. There is no obvious restoration, but possible cognomina are Vediatus or Veliatus derived from the nomina Vedius or Velius but not attested, or Veniatus which is barely attested.

⁶³ This graffito is apparently complete, and from its position unlikely to be an unfinished name such as IVL or IVCVNDI.

⁶⁴ The first letter is a long vertical stroke cut by a short horizontal stroke at top and bottom. This might suggest capital E, but there is no mid-stroke, and a more likely reading is I, enlarged and emphasised as the initial letter. It is followed by a cramped xe, not ue (in view of how u is written elsewhere), and then by distinct letters fluently written that become cramped towards the edge. This is a Celtic personal name, to judge by the termination -magulus which incorporates the element magu- (‘youth, slave, vassal’; for which see D. Ellis Evans, Gaulish Personal Names (1967) 221); in Britain it terminates the name Taximagulus (Caesar, Gallic War 5.22), but the preceding ixesoro- remains obscure. In the second line, the un-Latin sequence aurlus can hardly describe this man; nor can it be his patronymic, since its termination is nominative, not genitive. But perhaps it is a bungled Aur(e)l(i)us, for his nomen.

⁶⁵ Only S is certain; perhaps followed by I and an incomplete N. Line 2 apparently ends with D or CI, the latter suggesting a name in the genitive case. The horizontal incisions suggest it has been deleted.

⁶⁶ The graffito is clearly a genitive cognomen in -ia|ni, but M is incomplete and R has lost its loop. The sequence MR is impossible, but the intervening Ʌ was presumably omitted by confusion with the second half of M. The next letter(s) resemble modern ‘W’, but can easily be read as I conjoining N. Varinia|ni would be possible by supposing that the first letter is a small V ligatured to Ʌ (and sharing a diagonal); but this would be unlike the ample Ʌ which ends the line, and the cognomen Varinianus is very rare although VɅRENIɅN has been read in another lead tag from Caerleon (RIB II.8.2504.7). Marinianus, however, is well attested and has already occurred in Britain, at Bath: Tab. Sulis 30.5, Marinianus Belcati (filius), where it is noted that Nig(…) Marinianus, a cavalryman seconded to the Guard in Rome (eques singularis) and buried there, was of British origin, natione Britanicianus (CIL 6.3279); his son was also called Marinianus.

⁶⁷ The third letter consists of a downstroke tending to the left which was originally read from photographs as N, but the surface to the right is still intact and denies trace of further strokes. The area above is too damaged to exclude the possibility of an upward stroke for S. Lais is quite a common woman's name, but there is no man's name that might be abbreviated to Lai- except as the initials of tria nomina.

⁶⁸ 21 High Street, Barkway. Its rediscovery was reported by MacKay, William in Britannia 41 (2010) 469Google Scholar, add. (a), to whom it has now been given by the previous owner.