² In the garden of a house which was being renovated, where it was apparently re-used to make a step into the house. It was acquired by ArtAncient, London, where Costas Paraskevaides reported its discovery and made it available. It is now in private hands, but the owner may place it in a public collection on loan.

³ Letter heights: line 1, 143 mm; 2, 116 mm; 3–5, 110 mm; 6–7, 105 mm; 8, 120 mm, with faint traces of horizontal setting-out lines. The increased height of line 8 suggests that the text was coming to an end, with the formula T F I and the heir's name (and description?) balancing the first two lines. It is only the second tombstone from Dorchester, but the other (RIB 188) is also a Purbeck marble slab of similar width (0.71 m) and lettering.

⁴ Abbreviations are marked by a leaf-stop (hedera) or triangular medial point, with a medial point also after EX. In line 8, T is preceded by a medial point and followed by a leaf-stop; there is uncertain trace of a second leaf-stop after F, and a third may be inferred after I. In line 1, the second I was inserted within the second D to save space; and likewise in line 6, the numeral II with suprascript bar was inserted within the G of LEG. All that remains of line 9 is the upper edge of the first two letters, which resemble CL at the end of line 3. The first letter is C or G, the second is I or L. SI (for sibi) cannot be read. The two letters are presumably the beginning of the name of Bassus’ heir. C may have embraced a stop, suggesting C I[VL …] for G(aius) I[ul(ius) …], but this would be rather cramped, and CL is an easier reading. The absence of a praenomen would then imply either a non-citizen or a woman; and if the woman were a citizen, then Cl(audia) […].

⁵ Bassus’ age at death (55) is ‘age-rounded’, but would suggest that he had been discharged about ten years earlier. He came from Heraclea Lyncestis in Macedonia (enrolled in the tribe Fabia), to which Forni (Reclutamento, 165, 175, 183) attributes ten pre-Hadrianic legionaries including Gaius Saufeius of the Ninth, buried at Lincoln (RIB 255). His nomenclature, like that of Bassus, is full and formal, but notably is not preceded by Dis Manibus unabbreviated (let alone D M), features shared by the early (Claudio-Neronian) epitaphs of legionaries at Colchester, Lincoln, Gloucester and Wroxeter. A similar date may thus be proposed for Bassus’ tombstone, which the lettering strongly supports. As an early veteran of the Second Legion, he is the counterpart of Valerius Geminus at Alchester (RIB III, 3121), but is probably the earlier. However, he was buried at Dorchester. In a very full discussion of Geminus’ tombstone (Britannia 36 (2005), 101–33, esp. 102), Eberhart Sauer emphasises that ‘all known legionary veterans in Britain were buried at the main base of their own legion or a colony’, and deduces that Alchester was chosen by Vespasian (the future emperor) as the first base of his legion. Previously Sheppard Frere had suggested that the legion was first based at Dorchester or Lake Farm in Dorset, a thesis developed by Mark Hassall in R.J. Brewer (ed.), Roman Fortresses and their Legions (2000), 51–65, who suggested Silchester in a.d. 43–49 and then Dorchester or Lake Farm in a.d. 49–55. Dorchester is now confirmed as the legion's early (if not also its first or only) base, but the question remains of whether it was concurrent with Alchester, or the two were consecutive.

⁶ By a metal-detectorist, as part of a hoard of copper-alloy scrap including the figure of a running hound, first published as Adams, K., Henig, M. and Pearce, J., ‘A founder's hoard from near Gloucester’, ARA 24 (2018), 53–9Google Scholar. It was made available by Sally Worrell through the Portable Antiquities Scheme (PAS ref. GLO–BE1187).

⁷ In the upper left-hand corner is part of a tilted serif and, in the broken edge below it, the edge of the diagonal of V. In the upper right-hand corner, the square-edged break may mark the end of T, but there is no definite trace of the letter. However, the broken edge excludes B or P with extended serif, for conla[bs…] or conla[ps…].

This fragment is almost unique in Britain as part of a ‘monumental’ inscription inscribed on bronze, which is not a votive plaque like RIB 191 and 194 (Colchester), 662 and 663 (York), and those in RIB II.3, 2432. Civic inscriptions cut in bronze, although quite frequent in Italy and other provinces, are very rare in Britain: see RIB 197 (Colchester) and RIB II.3, 2433.16 (Canterbury). John Pearce, who will publish the inscription more fully, sees a reference here to public ‘contribution’, with –VM being a genitive-plural termination: compare ex voluntaria civium con[latione] (CIL viii 23863). For such collationes, compare RIB 69, 70 and 71 (Silchester).

⁸ During excavation by Wardell Armstrong Archaeology directed by Frank Giecco, who made it available.

⁹ The letter is incomplete, but would have been 100–110 mm high. The width of the first line is unknown, but it seems unlikely to have accommodated Dis Manibus unabbreviated.

¹⁰ With the next two items during excavation by Wardell Armstrong Archaeology directed by Frank Giecco, who made them available.

¹¹ There seems to have been a space to the right of the letter, suggesting that it was the end of a word, and thus excluding D and d[eo]; also the other possible letters, B, P and R.

¹² The Severan empresses, especially Julia Domna (mother of Caracalla) and Julia Mamaea (mother of Severus Alexander), are often honoured with their sons as ‘mother of the army, senate and country’, so the sequence [ma]tri … [c]astro[rum … senatus … p]atria[e] can be restored here with certainty, but not the connective words (whether ac, et or -que). The separation of [ma]tri from [c]astro[rum] shows that she is being expressly honoured as the mother of the Emperor, described as sanc[tissimi] (‘most holy’), the first instance from Britain, unless it be restored in RIB 1282. But there are too many variations in formulation to restore the text in full, or even to decide between Julia Domna and Julia Mamaea, both of whom are honoured locally, Julia Domna at Netherby (RIB 976), Julia Mamaea at Carlisle (RIB 949) and Old Penrith (RIB 919). The remnant of MA at the end of line 1 suggests that Julia Mamaea is the honorand, with her cognomen divided between two lines as in RIB 919. This would be understandable if the inscription were quite a long one, with Severus Alexander being honoured first, thus making the layout more cramped; as indeed is suggested by the line-endings SANC and [C]ASTRO.

¹³ There are faint traces of setting-out lines above and below the letters, which in the bottom two lines are 35 mm high. The leaf-stop (hedera) is cut within a crudely scribed oval. There is trace of the curve of P in the broken edge, and the reading is confirmed by the following letters and the milliary symbol. It is the first stone inscription to attest the ala Petriana at Stanwix, although one of its prefects dedicated RIB 957 in Carlisle. In appearance the stone might be part of an altar die above a moulded base, but since […]VM LX surely refers to the age at death (‘60 years’), it must be part of a tombstone consisting of a panel recessed within a moulded border. The dedicator evidently belonged to the ala Petriana, but it cannot be seen whether the deceased was also a member of the unit, or only related by blood or marriage.

¹⁴ During a community excavation by Berkshire Archaeology Research Group and Boxford History Project, supervised by Cotswold Archaeology. Lindsey Bedford sent details and photographs.

¹⁵ The fragment is about one-third the width and height of the original (280 mm square), so it preserves only a small part of the inscription. The break is just too close to the fourth letter in both lines to be sure whether it is I (which is visually the easier reading) or A (with rather an upright first stroke). Thus the personal name in line 1 is probably Lucius or a derivative such as Lucianus, not Lucanus, which is much less common. But in line 2, peda[les] with a numeral, an itemised batch-total like those in RIB II.5, 2491.1–8 (but with no actual instance of pedales), is less likely than pedi[cat] or similar, with the obscene verb p(a)edicare (‘bugger’) in its usual ‘Vulgar’ spelling, for which compare RIB II.5, 2491.157 (with note).

¹⁶ With the next item by Simon Hall with a metal-detector, who made them available. Wendy Scott, Finds Liaison Officer for Leicestershire and Rutland, sent a drawing and photographs of the spearhead (PAS ref. LEIC-35D01B).

¹⁷ If these are numerals, two are not of standard form: ‘7’ consists of seven successive strokes, as if a tally, rather than VII; ‘8’ is not VIII, but V (‘5’) enclosing three short incisions. IXI is repeated on the next item (No. 9), where if taken by itself, it might be seen as decorative. In view of the dedication to Silvanus, it is tempting to see these numerals as tallies of game bagged by a huntsman. Iron spearheads have been found at temple sites in Britain and Gaul, including Uley (where the god is variously identified as Mercury, Mars or Silvanus) and Lamyatt Beacon. They have been seen as votive, but this (copper-alloy) spearhead is the first to be found in Britain with a votive inscription.

¹⁸ The crucial first line consists of the lower loop of S (with just enough of it to exclude D or O); a vertical stroke with bottom-serif like the top-serif of I at the end of line 3; another vertical trending left with a single dot to the right, acceptable as a narrow L; and three diagonals, acceptable as V and an incomplete (or perhaps ligatured) A. For SILVANO thus in dot-punched capitals, compare RIB 194 (Colchester). In the missing blade-tip above, there would have been enough space for DEO. The dedicator's name, Cunomi|nus, is unattested, but there are many Celtic names formed from the element *cuno-s (‘hound’). In the next line (5), VO is certain, but the expected VOTVM or VSLM cannot be read. The third letter is D, not T, but rather cramped by the shaft-tip; then R, and either A or the first half of M. This sequence is unexplained, but perhaps the dot-puncher misunderstood guide-letters which had been sketched for him with brush or pen in the awkward space either side of the shaft-tip.

¹⁹ Two diagonal strokes intersecting at right-angles, bounded by two strokes which can be read horizontally or vertically. Taken on its own, this combination might be seen as space-filling decoration, but comparison with the previous item (No. 8) suggests that it is another ‘milliary’ IXI.

²⁰ By a metal-detectorist (PAS ref. LVPL-942D3C). Teresa Gilmore, Finds Liaison Officer for East Staffordshire and the North West Midlands, sent a photograph and details. At present they are in Congleton Museum, which hopes to acquire them, together with two other pieces still in the finder's possession but not yet seen. Ian Doughty of the Museum sent more photographs and further details.

²¹ Like the Shavington brine-pan (Britannia 29 (1998), 436, no. 11), published with commentary by Penney, S. and Shotter, D.C.A. in Britannia 27 (1996), 360–5CrossRefGoogle Scholar. See also Shotter in M. Nevell and A.P. Fielding (eds), Recent Archaeological Work on the Roman Salt Industry in Cheshire (2005), 41–6. The Shavington pan, like the present item (No. 10), incorporated a moulded inscription either side of the lifting-hole; compare also RIB II.2, 2416.2, 3 and 5.

²² The first reading (in roman capitals) is the literal reading from left to right of the two sheets as photographed (fig. 8), and is apparently a personal name between the numerals CCC (‘300’) and XXX (‘30’), but the inversion of M and T implies that the mould-maker was actually working the other way up, without realising that the inscription would then be upside-down when the pan was used. It should rather be seen as XXX MOHTO ↃↃↃ (with CCC reversed), but this is probably not quite what the mould-maker intended. OTHOM is unacceptable as a name, in view of its impossible termination and because -th- was foreign to Celtic and Latin, but MOHTO can be explained as a mistake for MOTTIO: the mould-maker was working retrograde and misplaced two horizontal strokes so as to make TTI into HT. Mottio is a Celtic name (CIL xiii 4257) which was ‘romanised’ as Mottius, for example by the British potter Mottius Bollus (A.R. Birley, The People of Roman Britain (1979), 132). In the spelling Motio (with one T) it has already occurred in Britain (RIB 281).

After MOHTO, the mould-maker failed to reverse CCC. It is unclear whether he intended the numeral (a note of the pan's capacity) to be CCCXXX by addition (‘330’) or XXXCCC by subtraction (‘270’, a much less common form, but compare XXXC for ‘70’ in RIB 63). ‘270’ is more likely, assuming that he meant the numeral to be read, like the name, from left to right. When the other two pieces are available (see above, note 20), it may be possible to calculate the pan's capacity and thus to resolve the question, but meanwhile compare the Henhull brine-pans (RIB II.2, 2416.2 and 3): they measure 1.0 m by 1.0 m/0.95 m and are 0.13 m deep, with a capacity of c. 130 litres which in sextarii (546 ml) would be 238.

²³ During several seasons of excavation by Oxford Archaeology North now being published. Stephen Rowland made it available (sf 3387).

²⁴ There is just enough uninscribed space to the left to suggest that the graffito is complete. The second stroke of M extends downward, and the cursive S is exaggerated vertically, as if to mark it as the final letter. Two-letter ownership graffiti are uncommon, and likely to be an abbreviated cognomen like RIB II.7, 2501.438 (PR) or 524 (SV), but this cannot explain MS. It might be expanded to MVS, the initials of tria nomina, by ligaturing V to M and reducing S to a downward stroke, but this would be rather forced. More likely, the owner simply reduced his name to the initials of his nomen and cognomen.

²⁵ By Philip Cracknell at Garth Lea, during a watching-brief for Cumbria County Council. Ian Caruana sent details at his request.

²⁶ With items Nos 4–6 above, during excavation by Wardell Armstrong Archaeology directed by Frank Giecco, who made it available together with fragments of six other bessales bearing the same stamp. Many unstamped bessales were also found.

²⁷ This stamp is widely distributed in Carlisle: see RIB II.5, 2483.

²⁸ During excavation by Trent and Peak Archaeology after the evaluation noted in Britannia 45 (2014), 343–4. Kate Smart and Gwladys Monteil made it available (sf 979). Three other samian graffiti (each 1–3 incomplete letters) will be published in the final report.

²⁹ The incomplete letter after T is either A or I. It slopes forward slightly, which would suggest A and thus Cupita. But a man's name is inherently more likely, the slope then being due to the graffito following the curve of the vessel or even to being a downward continuation of the cross-stroke of T. This name would either be Cupiti (genitive), as in RIB II.5, 2491.78 and Britannia 28 (1997), 468, no. 40, or the derived Cupiti[anus], as in RIB 1988. Both names are quite common, and the same ambiguity is found in RIB 344, Cup(…), RIB II.7, 2501.152, Cupit[…] and Tab. Sulis 77, Cupit[…].

³⁰ With the next sixteen items and five sealings already published (Britannia 20 (1989), 337, nos 30–3, and Britannia 22 (1991), 302, no. 33), by the divers Bob Middlemass and Rolfe Mitchinson. Die-duplicates are treated as a single item, the reading being that of the best-preserved example. There are 35 epigraphic sealings in all, ordered here as in RIB II.1, 2411 (Lead Sealings). They will be published more fully as part of the final report on the Piercebridge deposit now in preparation by Philippa Walton and Hella Eckardt.

³¹ Unlike the die-duplicates below, these dies although similar seem to be different. Six other sealings with the legend D N have been found in Britain.

³² With some exceptions, noted below, the initials are those of a Roman citizen's tria nomina. They are probably centurions, which in two instances (Nos 17 and 20) is made explicit. None has been identified with a centurion known elsewhere. In this item (No. 16), unusually, E was added to the A of the cognomen, as if to reduce the number of possibilities; it was probably Ae(lianus) or Ae(milianus).

³³ Only two sealings are completely preserved (fig. 14), but all five are die-duplicates. Another of the legion's sealings (Britannia 29 (1998), 439–40, nos 28 and 29) carries the initials L S S, suggesting that both centurions may have been a L(ucius) S(eptimius), but there are other possibilities.

³⁴ The same legend as RIB II.1, 2411.70 (South Shields), which is probably a duplicate. Since OVA is impossible as the initials of tria nomina, and there is no name attested in Ova(…), the reverse should be read retrograde as AVO. RIB expands this to a(la) Vo(contiorum), but the two sealings explicitly of this unit (RIB II.1, 2411.90 and Britannia 47 (2016), 393, no. 6) abbreviate its name as AVOC. There is no independent evidence to associate this ala directly with the legion, so as to cause them to issue joint-sealings, nor any other legionary sealing that associates an ala with a legion, so AVO is more likely to be another tria nomina.

³⁵ The praenomen Gaius was conventionally abbreviated to C, but G is occasionally found instead: in Britain, on stone (RIB 373, 812) and in clay (Britannia 40 (2009), 321, no. 14, the man's own signature). When tria nomina were reduced to their initials, this convention of C for Gaius evidently became less general: in the index of ‘probable initials of tria nomina’ in RIB II, there are 10 instances of G as the initial of a praenomen, as against c. 20 of C. Cf. No. 25 below.

³⁶ The first sealing of the ala Hispanorum Vettonum, which is attested at Binchester (Vinovia), the next fort to the north from Piercebridge, by RIB 1028, 1035 and III, 3260; compare item No. 26 below. The decurion has only two initials, as in RIB II.1, 2411.84 and 88, but this was probably to save space; the praenomen is omitted of a fourth decurion (RIB II.1, 2411.90) who was explicitly a Roman citizen.

³⁷ The first sealings of the Fourth Cohort of Gauls which, after being stationed on the Antonine Wall and at Risingham, became the garrison of Vindolanda in the reign of Septimius Severus, where it remained.

³⁸ This is the second sealing of a beneficiarius consularis from Piercebridge, the other being Britannia 22 (1991), 302, no. 33. The same reverse legend (EX) is associated with a beneficiarius consularis at Brough under Stainmore (RIB II.1, 2411.246 and 267), and must be a variant of EXP, which is associated with legions (RIB II.1, 2411.44–54 and 75). Its expansion is uncertain, but RIB accepts Richmond's ex(pedivit) and exp(edivit), ‘… has despatched’, which is relevant. A general formula, even if self-evident, is more likely than a specific, limited category such as exp(ensus) (‘expended’) or exp(editionalis) (‘for a military expedition’).

³⁹ The Roman name of Binchester written as V(i)nov(ia), with i reduced after u, or simply omitted. It was the posting of a beneficiarius consularis (RIB 1030 and 1031), and a sealing from Buxton (RIB II.1, 2411.307) reads VIN | OEN, perhaps Vino(vi)en(ses), for its garrison or inhabitants. The ala Vettonum (compare item No. 22 above) was stationed there.

⁴⁰ Duplicate of Britannia 20 (1989), 337, no. 33, also from Piercebridge, in which the ‘zigzag’ above the tria nomina was read as letters. By collating the two examples (fig. 17), it can now be seen that the ‘zigzag’ is continuous, and is more like a snake advancing right. Compare RIB III, 3257, a building stone from Piercebridge, on which the initials of tria nomina are accompanied by a similar zigzag figure representing a snake, noted there as ‘a good symbol associated with a man's genius’.

⁴¹ OPH cannot be the initials of tria nomina, whether retrograde or not, so it must be an abbreviated cognomen; it would be distinctive, since names in Oph(…) such as Ophelimus are very uncommon. For the reverse die, which is EX like the next item (No. 29) but differently formed, see above, note 38.

⁴² This sealing is so similar in appearance to the previous item (No. 28) that the obverse should also be seen as an abbreviated cognomen, especially since Optatus is quite common. Two sealings from Kirkby Thore (Britannia 19 (1988), 499, no. 51; RIB II.1, 2411.98 with Britannia 22 (1991), 311 (f)) attest a centurion of the Sixth Legion called Optatus, but he need not be the same man. This may even be a beneficiarius consularis, if the EX reverse (see above, note 38) were exclusive to them.

⁴³ The format of this sealing is unparalleled at Piercebridge and in RIB II.1, 2411: it is long and narrow, too narrow for the edges of the dies to register. The pattern of initials suggests two persons who shared the same nomen, Gavius or Gabinius for example, but bore different cognomina, I(…) and C(…); most likely father and son, brothers or fellow-freedmen, but by implication business-partners (socii). On the reverse, M is separated from LGF by a medial point, suggesting that LGF is the initials of tria nomina; and that, with the nomen G(…), this man was related to the partners. M can be understood as an abbreviation for m(anu) (‘by the hand of’), as in samian potters’ stamps.

⁴⁴ There is no sign of a loop or handle, but otherwise it resembles the bronze strap-ends with the same legend (utere felix) in incised letters filled with enamel, similar in size but triangular in section, which have been found at South Shields and Chester (RIB II.3, 2429.13, 14 and 15).

⁴⁵ This is the reading as the tube is rotated, but (2) and (3) are inverted, and (3) and (4) are retrograde. II represents E in (1) and (4), but E is used in (2) and (3). In (2) the mould-maker did not complete the first E, which remains as I. Opposite faces must be taken together, that is (1) with (3), and (2) with (4). This reconstructed reading is illustrated by fig. 18, which is a composite photograph taken with the faces rearranged in the sequence (1), (3), (2) and (4). (2) and (3) have also been turned the right way up, and (3) and (4) reversed so as to appear rectograde.

⁴⁶ By a metal-detectorist (PAS ref. SUSS-BA3CBE).

⁴⁷ The expected K for K(alendae) was probably just too far to the left to survive, but there is no sign of the initial I of Id(us), which was presumably omitted by mistake. The Kalends were 1 August; the Nones 5 August; the Ides 13 August. This fragment resembles the fragment for September found at Vindolanda (Britannia 41 (2010), 461, no. 65), except that the holes are smaller and the letters incised, not dot-punched. Alexander Jones saw that both fragments resemble the inscribed rim of the water-clock in Frankfurt Archaeological Museum published as D. Stutzinger, Eine römische Wasserauslaufuhr (2001), a copper-alloy bowl which marked the twelve hours of day or night by the outflow of water, and was calibrated to take account of the varying length of an ‘hour’ according to the date. See now A. Meyer, ‘The Vindolanda calendrical clepsydra: a re-examination of the Vindolanda “Calendar”’, in Britannia forthcoming. (Information from Kevin Birth and Alexander Meyer.)

⁴⁸ During excavation by Maidstone Area Archaeological Group, and reported in MAAG Blog 2017. Information from Alex Mullen.

⁴⁹ Another possibility would be [VIV]AS (‘long life to you’), but the usual formulas do not continue with A[…], whereas capital letters of just this style are used to caption mythological scenes engraved on glass bowls from the Rhineland. They include a theme popular in Roman art, the musical contest between Marsyas and Apollo which ended in Marsyas being flayed alive: for a British example, see RIB II.2, 2419.41 with note; and compare another, but with quite a different, convivial legend, illustrated in Journal of Glass Studies 2 (1960), 66, fig. 55.

⁵⁰ By a metal-detectorist (PAS ref. FASAM-F55282).

⁵¹ Both names are Celtic, but difficult to parallel. Names in –dacus (or –dagus) are found occasionally, for example Bitudacus (RIB II.7, 2501.108) and Cundacus (Britannia 43 (2012), 402, no. 12), and Daccus is found independently. [M]andacus is an attractive restoration in view of mannus (‘pony’) and the names in Mandu–, but they suggest that this name is really a variant of *Manduacus, which is unattested but would combine two well-attested elements, the second being –aco-s. The patronymic is probably related to the feminine name Matta, and is thus a variant of Mattaus (CIL xiii 6003, in the genitive MATTAI), the second u being inserted as a ‘glide’.

⁵² By a metal-detectorist (PAS ref. NLM-7632F8).

⁵³ O is square, and formed by four connected strokes. The bezel is bordered by separate short strokes facing inward, three at top and bottom, and one on either side. For other TOT rings, see Britannia 45 (2014), 443–4, nos 15, 17–23; 46 (2015), 404, nos 31–4; 48 (2017), 465, nos 12–15, and item No. 46 below. In general see Britannia 45 (2014), 443, n. 30.

⁵⁴ During a community excavation by Old Sleaford Heritage Group supervised by Archaeological Project Services (Britannia 48 (2017), 357), in a Roman layer containing pottery of third/fourth-century date. Alex Beeby made it available.

⁵⁵ In Britain, alphabets are inscribed before firing on tiles (RIB II.5, 2491.135–45) and coarseware (II.8, 2502.6–8), but this is the first on a colander. Apart from two silver strainers (RIB II.2, 2414.3 and 21), it is only the second inscribed colander to be found in Britain, the other (RIB II.8, 2503.399) being just an abbreviated owner's name. This alphabet, being inverted and near the base, would have hardly been visible when the vessel was used; only when it was upside-down on the shelf. So it was hardly ‘decorative’ or even didactic, but perhaps the potter was simply practising his literacy or showing it off.

⁵⁶ During excavation by Pre-Construct Archaeology (Britannia 47 (2016), 334), to be published in Trans LAMAS. Eniko Hudak made it available.

⁵⁷ M is barred, identifying it as an abbreviation, as in RIB II.6, 2494.44 (South Shields, a Dressel 20 graffito), m(odii) VII s(emis) | […]. For such notes of capacity on Dressel 20, see RIB II.6, pp. 33–4: VII is the most usual numeral, often followed by semis (‘half’) or a second numeral for sextarii. The usual place is the lip or handle, but for another on the shoulder see RIB II.6, 2494.46 (also London).

⁵⁸ In the same excavation by MoLAS as the three pewter tablets published as Britannia 30 (1999), 375, no. 1 (list of valuables); 44 (2013), 390, no. 21 (charm against plague); 46 (2015), 406, no. 42 (phylactery). Jenny Hall made it available from the Museum of London (VRY 89 <101>). Still to be published is a fifth tablet <254> referring to recurrent fever (tertiana, cottidiana, etc.).

⁵⁹ The left edge is broken. The top, right and bottom (cut) edges are largely original, but now ragged as a result of folding and corrosion. The larger fragment includes two conjoining scraps (top-right corner). The surface everywhere is crumpled and corroded, and abraded from cleaning.

⁶⁰ Line 1 apparently continued below (extreme left), and was followed by a space to the right. Only two of the ‘characters’ are more or less complete, the ‘ring-letter’ at the top left, and the ‘supine beta’ in the middle. They resemble two of the ‘characters’ in the upper half of the phylactery cited above in note 58.

⁶¹ None of these readings is quite certain, since the sample is small and there is no recognisable word. The sequence nemiml is acceptable, but meaningless; after it, d might be b; p is capital-letter in form (or late Roman cursive); k might be f; the second e is different from the first, but both forms are found together in Old Roman Cursive; however, it might also be a late Roman r.

⁶² These readings too are uncertain, but P, M, B and E are acceptable. Despite the poor state of preservation, the suprascript bars seem to be consistent. Perhaps they were intended to turn Roman capitals into ‘characters’, like the Greek ‘barred lambda’ and ‘barred omega’ in line 2 of the phylactery cited above in note 58.

⁶³ During excavation by Durham University Excavation Committee, judging by the attached label which reads: ‘No. XLVII | FORM 36 | (Small) | Owner's Name | AVDAX | [….].’ It was noticed by James Gerrard in a boxful of plain samian forms passed to him for teaching purposes by the Great North Museum, and the Corbridge provenance is deduced from the reference to ‘No. XLVII’ (Site 47). Other material from the Committee's excavations at Corbridge and other sites on Hadrian's Wall went to Newcastle when its Corbridge hut was demolished to build the Roman Site Museum, for example the stones cited below in Change of Location (e).

⁶⁴ Audax is quite rare as a Latin cognomen, since the adjective (‘daring’) is used in a pejorative sense, but as a personal name it is well attested in Gaul and Britain: Holder suggested it might represent the ‘Gallic’ name Audagus (RIB 774), which however is probably German. The name Audax has already occurred at Corbridge as the owner of an amphora (RIB II.6, 2494.104, AVDAX), as well as in the Vindolanda Tablets (Tab. Vindol. 186, 590, and WT 2017.15 not yet published). As Audax or Audacius it is also found at Caerleon (RIB II.1, 2409.10) and London (RIB II.7, 2501.83, RIB III, 3002, Tab. Lond. Bloomberg 64 with note); compare the centurion Audac(…) Romanus at Carvoran (RIB 1779).

⁶⁵ Like the next five items during excavation directed by Andrew Birley, who made them available. The graffiti were all made after firing, and thus relate to ownership and use.

⁶⁶ There is a medial point after both S and F, confirming that they are abbreviations and in fact the first two initials of tria nomina. V is neatly finished with a serif on the second stroke. The most likely nomen is Flavius, but many cognomina are possible, such as Verecundus, Verus, Victor, Virilis and Vitalis. But compare Tab. Vindol. 802 (with note), the heading of a letter from [F]lavius Vindex.

⁶⁷ The graffito is inverted as to the vessel, so the scribe was leaning over it. There is nothing to the left, but the broken edge is too close to tell whether there was anything above. The numeral ‘9’ is sometimes found on amphoras (RIB II.6, 2494.64–8), but as a measure of modii it is rather high for a Dressel 20 (see above, note 57); so it may have specified sextarii below a numeral for modii now lost.

⁶⁸ Quietus is a common Latin cognomen, found also in Essex (RIB II.8, 2503.393) and at Corbridge (Britannia 48 (2017), 468, no. 19).

⁶⁹ Only the tip of T survives, but it underlies the stroke of I, showing that the vessel was inverted when the graffito was made. The end of a masculine personal name.

⁷⁰ F is made with a downstroke and a single second stroke, a short upward diagonal. Exactly this form is found in Tab. Lond. Bloomberg 57.6 (facere), and in these tablets generally the letter is a sinuous downstroke followed by a single short diagonal; this is quite like the ‘very remarkable form’ noted in Tab. Vindol. II, p. 50, ‘in which the top stroke is entirely missing’. R and L are also of cursive form, but E is capital. The name Freio is Tungrian: it occurs in Tungria itself (CIL xiii 3614) and further north at the shrine of Nehalennia (AE 1997, 1164); note also CIL xiii 7036 (Mainz), Freioverus Veransati f(ilius) cives Tunger.

The sherds have two small-find numbers because they were found in two separate contubernia of a Period IV cavalry barrack block after demolition, so the mortarium must have passed out of Freio's possession by then, since he was an infantryman in a century. The century of Tullio is already attested in Tab. Vindol. 184.31, ‘the only major ink document from period V’ (Vindolanda Research Reports II, caption to pl. XVII), when it must have been residual.

⁷¹ The sequence ANCC is impossible in a personal name, unless it is an error for something like Tancinus or Tanicius. The alignment changes between […]TAN and CCII[…], suggesting that CCII[…] might be a numeral (‘102’ or more); compare RIB II.8, 2503.1 (also Vindolanda), enumerating coriander seeds.

⁷² By a metal-detectorist (PAS ref. DENO-494B61).

⁷³ O has a pellet in the centre. For TOT rings, see item No. 35 above (with note).

⁷⁴ During excavation by Oxford Archaeology. Paul Booth sent details and a photograph.

⁷⁵ The graffito is complete. An abbreviated personal name, but with many possibilities.

⁷⁶ During the excavations by Oxford Archaeology summarised in Britannia 47 (2016), 253–61, by Paul Booth, who sent a photograph and other details. He notes that, except for […].R scratched on another black burnished sherd, this is the only sherd with a literate graffito, of c. 60,000 sherds recovered. But 55 sherds carry marks such as knicks or ‘crosses’ which can be interpreted as marks of identification, a ratio of 1:1049.

⁷⁷ There are some marks which look casual, but also a horizontal stroke above II which looks deliberate, and the first vertical has been repeated. Laetus (‘happy’) is a popular cognomen, which suggests that the scribe anticipated T by writing LATI by mistake, before correcting it to LAII, and then repeating T. Finally he extended I downwards, to mark it as a genitive termination. Despite the broken edge, this elongation and the location of the graffito within the circle of the base make it possible to exclude one of the less frequent derivatives of Laetus such as Laetianus (compare RIB 1851).

⁷⁸ Outside the scheduled area by a metal-detectorist, who brought it to Alex Bliss, Finds Liaison Officer for Suffolk. He read qui perierunt and saw that it was a curse tablet against theft, and made it available (PAS ref. SF-BA1337). The finder has now donated it to Suffolk County Council Archaeological Service.

⁷⁹ The scribe did this without inverting the tablet, unlike the Eccles tablet (Britannia 17 (1986), 428, no. 2) which was written in cursive letters boustrophedon, by inverting it each time to write the next line. In line 2, si was twice reversed as IS. mulier and baro were written with letters inverted, facing right, except for B which was reversed, facing left. The first letter (V) of line 3 was inverted.

Such protective ‘encipherment’ is found in other curse tablets: whether written from left to right in capitals, but reversing the letter-sequence in each word (Tab. Sulis 4); or from right to left in mirror-image capitals (Britannia 18 (1987), 360, no. 1; Britannia 24 (1993), 310, no. 2; Britannia 25 (1994), 296, no. 2; Britannia 39 (2008), 380, no. 17); or from right to left in cursive, either reversing the whole letter-sequence (Tab. Sulis 98) or line by line (Tab. Sulis 44 and 62); or from right to left in mirror-image cursive (Tab. Sulis 61). A Latin text from Uley (no. 52, not yet published) is even transliterated into Greek letters.

⁸⁰ Except for [a]nuli, the text is nothing but formulas, as if stolen property could be recovered simply by copying out disconnected but familiar phrases like a magic spell. Elements of more literate texts are missing, such as the name of the deity, the name of the petitioner, an explicit reference to the thief (qui involavit, etc.), a request that he be punished or compelled to make restitution. But the tablet is physically complete: there is no sign that it is a fragment, or part of a longer text. Line-by-line commentary follows:

1, [a]nuli. There is no sign of A to the left of N, but the surface here is corroded, and there has probably been some loss to the upper left-hand corner. The restoration is acceptable since gold and silver rings were a convenient store of value, and quite often feature in curse tablets: in Britain RIB 306 (Lydney), anilum perdedit; Tab. Sulis 59.2, anulis or anuli s[ui]; 97.1–2, anilum argenteum … qui anilum involavit; Uley 3, anulus aureus; Uley 50, anulli quat(tu)or; note also DTM 7.2 (Mainz), quisquis nobis sustulit … anulos aureos; Kropp 4.3.2/1 (Dax/Landes), quicumque levavit anulum.

1, perierunt. The verb pereo (‘perish’) serves as the passive of perdo (‘lose’), as in Uley 80 (Britannia 27 (1996), 439, no. 1, with note on p. 441), which was prompted by the ‘loss’ of gloves, manecilis qui per[i]erunt.

2, si muli[e]r. The trace of I to the left of S, in the broken and corroded left-hand edge, is uncertain; but its restoration is confirmed by IS (si reversed) being repeated midway, before baro. IS (thus written twice) is equivalent to SI before ingen|(u)us at the end of the line, when the scribe had stopped inverting or reversing his letters. E has been lost in the corrosion, and R was written without closing the loop.

2, si baro. As already noted, si was reversed as IS, not inverted. But baro was inverted, except for B, which was reversed. The scribe may have thought that if he simply inverted B, it would still look like B; so reversed it instead. But he miscalculated the position of the first, vertical stroke, by putting it too close to the preceding S. He then corrected his mistake by repeating the vertical stroke, and adding to it the double loop of B (reversed, facing left), on top of the first vertical stroke. Next he inverted A and R, probably failing once again to close the loop of R, but the surface is too corroded here to be certain.

The reading si mulier si baro (‘whether woman or man’) is guaranteed by the frequency of this formula and its variants (Tab. Sulis, p. 67), even though the reverse sequence ‘whether man or woman’ is usual; but compare Tab. Sulis 44.b1–2, si mul[ie]r si ba(ro).

2–3, si ingen|(u)us si [s]er(v)us. The scribe was naturally unable to reverse or invert O, the last letter of baro, which may have prompted him to revert to conventional capitals. He ended line 2 with SI INGEN, his G being little different from S, and completed the word in line 3. Here he inverted the first letter, V (the cuts now widened by corrosion), but seems to have written S conventionally, although it is now almost entirely lost in the corrosion. The subsequent letters are conventional.

In writing si ingen|(u)us si [s]er(v)us, the scribe twice dropped u before a second u: the form serus for servus is a frequent ‘Vulgarism’ (Tab. Sulis, p. 266). si ingen(u)us si ser(v)us is a rare variant of the very frequent formula si servus si liber (‘whether slave or free’), which has occurred once before, in the unprovenanced tablet published as Britannia 22 (1991), 293, no. 1, si baro si mulier si puel[l]a si puer si ingenuus si servus. In garbled form it can also now be recognised in Tab. Sulis 101.3, si igeuns(!).

3, last 2–3 letters. After s[e]r(v)us, there are two downstrokes linked by a horizontal stroke above, as if to write TI. The surface to the right is damaged by the fold, but probably retains the trace of a diagonal stroke. Further to the right, the surface is quite well preserved, and there is no sign of any more text. These traces after s[e]r(v)us remain unexplained.

Below the first (inverted) V in line 3, there is what appears to be a deliberate vertical stroke, as if for I. But there is no trace of any letter either side, and it seems to be unrelated to the rest of the text.

⁸¹ With the next item during the excavation by A.W.G. Lowther published in Surrey Arch. Coll. 38.2 (1929), 132–48. This includes drawings, but not readings, of ten ‘Graffiti on pottery and tile fragments’ (ibid., 146, fig. 9) which were overlooked by RIB II. Some are quite slight, and not all have been traced. After reassessment and further excavation by David Bird (Britannia 45 (2014), 389–90), Joanna Bird sent rubbings, photographs and other details.

⁸² This graffito (Guildford Museum, acc. no. AS 421) was drawn by Lowther (see previous note, 146, fig. 9.5), but he misinterpreted the broken edge at the top as part of the loop of capital R in line 1. In line 2, T is taller than V, which suggests it may have been the first letter of a word, but it is too close to the broken edge to be certain. Lowther offered no reading of the whole, but the sequence TV might have suggested a reference to tu[bi] (‘box-tiles’), as in RIB II.5, 2491.1, 2 and 6. This would have been attractive, since they were a feature of his excavation (RIB II.5, 2490.1, with introduction). The broken letters in line 1 may then be part of the tilemaker's name, but they are too incomplete to be read; one possibility is cursive […]mar[…].

⁸³ Guildford Museum, acc. no. AS 4399. The first surviving letter might be Λ, but its long descender suggests R which has lost a small loop. Short diagonal strokes follow: they hardly differentiate Λ, M, N and V, but an abbreviated name is more likely to have ended in N than V; there is no sign of any further letter. The most likely name is [Ge]rman(us), but [Fi]rman(us) is also possible.

⁸⁴ ARV07.330, during the excavation by David Bird, above note 81. Joanna Bird sent a photograph.

⁸⁵ There is just enough space after R to identify it as the final letter, and the graffito as an abbreviated name; the tip of M can just be seen in the broken edge to the left. For other examples of MAR as an ownership-graffito see RIB II.7, 2501.322–5, 332; II.8, 2503.322–4. The most likely name is Mar(tialis), but there are other possibilities.

⁸⁶ During excavation of a ritual shaft by Charles Warne (Proc. Soc. Antiq. London 2 ser. 1 (1861), 309–13), and now in Bourne Hall Museum, Ewell (acc. no. Z 141). Like the next item, it was overlooked for inclusion in RIB II.7, 2501, but was rediscovered by Joanna Bird in the course of recording stamped samian from Surrey. She sent rubbings and photographs.

⁸⁷ Warne (see previous note) read the graffito as […]DINI, possibly [Secun]dini, but there is space to the left of his ‘D’, which should be read the other way up as CI, with the final I typically elongated to mark it as a genitive termination. A Greek name such as Callinicus seems unlikely, and Dominicus, being Christian, is too late. For the name Minicus, which is probably Celtic despite its similarity to Latin Minicius / Minucius, compare the tile graffito MINICI (RIB II.5, 2491.109), and on samian, MINICAE and names in MINV– (RIB II.7, 390, 391–5).

⁸⁸ During a local society excavation with several directors, published by C. Orton in Surrey Arch. Coll. 84, 89–122.

⁸⁹ E coincides with two or three vertical strokes, as if it was first written as II. There is just enough space to the right to suggest the graffito is complete, an abbreviated personal name such as Ser(enus) or Ser(vandus).

⁹⁰ Published by Martin Henig in P. Ellis (ed.), Roman Wiltshire and After (2001), 122 with 123, fig. 6.14, and in J. Gerrard and M. Henig, ‘Brancaster type signet rings’, Bonner Jb 216 (2016), 225–50 (esp. 233), where it is cat. 9.

⁹¹ During excavation by MAP Archaeological Practice Ltd after the evaluation noted in Britannia 45 (2014), 332–3. Philip Mills sent details and made the sherd available.

⁹² Potters in the mid-second century ‘signed’ Dressel 20 near the basal knob, often with two lines of cursive consisting of (1) the consular date, (2) the day and month followed by their name. Examples in RIB II.6, 2493 are fragmentary, but Rodriguez Almeida illustrates four which are complete (Il Monte Testaccio (1984), 257, fig. 91), including one from Nijmegen of 24 June 158 with just the same form of u and k in the month-date. If line 1 began not much to the left of line 2, […]uero would be the end of the first consul's name, either Vero or [Se]vero, making the consulship of a.d. 155, Severo et Sabiniano, more likely than the other possibilities, those of a.d. 146, Claro II et Severo, and a.d. 161, Marcus Aurelius (III) and Lucius Verus (II).

⁹³ During the excavation by Northlight Heritage noted in Britannia 43 (2012), 283. Fraser Hunter, who is publishing the vessel in the final report, sent details. It is Type V,5c in R. Petrovsky, Studien zu römischen Bronzegefässen mit Meisterstempeln (1993), 79–84.

⁹⁴ Petrovsky (see previous note), 255–6, Taf. 40, notes two vessels of just this type stamped by Dioratus, a Gaulish maker whom he dates to the early Flavian period (a.d. 70–85).

⁹⁵ By a metal-detectorist. Now in Falkirk Museum (acc. no. 2000-4-7), from where Geoff Bailey sent photographs and other details. Fraser Hunter also commented on it.

⁹⁶ The weight is almost exactly 4 Roman ounces (unciae), one-third (triens) of a Roman pound (327.45 g), but there is no instance in RIB II.2, 2410 (Weights) of this being expressed in such a way. Only four digits or dots would be usual. The lone digit here is smaller than the other four, the first of which is cut by a diagonal blow. If this was intentional, it may have been intended to ‘cancel’ the digit or make it into an uncia symbol. But even so, the lone digit below would surely have been the obvious choice.

⁹⁷ Information from Kevin Hayward.

⁹⁸ Hargrove MSS 1777–1825 (York Record Office manuscript GB 192 HAR), III, p. 89 with a transcript and sketch map. Information from Martin Millett.

⁹⁹ Information from Scott Vanderbilt.

¹⁰⁰ Neal, D.S., ‘The “Helicon” mosaic’, in Johnson, S. and Neal, D.S. (ed. Snape, M.), ‘The re-excavation and study of the Helicon mosaic, Aldborough Roman town’, Yorks. Arch. J. 74 (2002), 113–34Google Scholar, esp. figs 7 and 8, pls 1 and 2.

¹⁰¹ Tomlin, R.S.O. and Pearce, J., ‘A Roman military diploma for the German Fleet (19 November 150) found in northern Britain’, ZPE 206 (2018), 207–16Google Scholar.

¹⁰² Information from Tom Welsh.

¹⁰³ Information from Fraser Hunter, who sent photographs.

¹⁰⁴ Inv. NEWMA: 1928.17 (now on display) and 1928.18 respectively, each illustrated on the Museum's online database. Information from Scott Vanderbilt.

¹⁰⁵ Illustrated on the Museum of London's online database (object 723605), as noted by Scott Vanderbilt, who suggests that it was transferred after the Cuming fire in 2013. Whether RIB III, 3015, 3020 and 3023 were also transferred then from the Cuming is unknown, since his inquiries have not been answered.

¹⁰⁶ Inv. NEWMA: 1956.435 and 397 respectively, each illustrated with a photograph on the Museum's online database. Information from Scott Vanderbilt.

¹⁰⁷ By a metal-detectorist (but the landowners do not confirm this), according to a record card made by Retford Museum when it was shown to the museum. It was then sold to a dealer, and the new owner, after offering it to the Collection, Lincoln (which was unhappy about the provenance), sold it again through TimeLine Auctions. It has been published online by Lincs to the Past (ref. MLI 98578). Information from Adam Daubney, Finds Liaison Officer for Lincolnshire, and Antony Lee, Collections Development Officer.

It looks genuine, even if the legion is not given its full title of p(ia) f(idelis), but a British provenance is unlikely. This legion, which moved from Dalmatia to the upper Rhine, and then to the Danube, never came near Britain; nor is there any evidence that it ever contributed a vexillation to the British army. Except very occasionally by a centurion about to take up another post, there is no altar in Britain dedicated by a soldier who claims to belong to a non-British legion.

¹⁰⁸ The setting-out lines show that something has been lost above MARTI, without indicating whether it was only DEO (‘to the god’) or a link to another deity. The second V in POSVIT was either repeated inadvertently, or inserted as a ‘glide’.