² During the same excavation by Pre-Construct Archaeology (LLS02, 972 sf 184), but not reported at the time. Victoria Ridgeway, Publications Manager of PCA, made it available. It is much thinner than RIB III, 3014, which ranges from 25 to 33 mm. Kevin Hayward notes that it bears some resemblance to the finer white marbles from the eastern Mediterranean.

³ The surviving text is too slight to be restored, but in view of its quality is no later than Antonine, and surely ‘official’.

Since line 1 is almost twice the height of line 2, the sequence AV strongly suggests AV[GUST...] and an imperial dedication of some kind. The content of line 2 can only be guessed, but in view of the likely religious context, the possibilities include [c]oll[egium] (a guild of worshippers) or (like RIB III, 3014) a dedication to the Numen Augusti or Numina Augustorum coupled with another deity, for example [deo Ap]oll[ini] as in RIB 611 and III, 3191. Less likely would be [deo S]oli or [Iovi D]oli[cheno].

⁴ In ploughing, and noted on the PAS database (LANCUM-273C82) with commentary by Shotter, David and Noon, Stuart, who will publish it and the next item in Trans. Cumb. Westm. 3rd ser. 13 (2013)Google Scholar, forthcoming.

⁵ The letters (c. 50 mm high) are rather tall in proportion to their width. A is cut indifferently with or without a cross-bar. The medial points are leaf-stops (hederae).

⁶ Unusually, this is not an individual tombstone but evidently marked a mausoleum or communal burial plot: the mixture of adults and children, and the related names Marinus and Marianus, suggest a family group like RIB 594 (Ribchester) for example. The cognomina Marinus, Marianus, Vitalis and Saturninus are all quite common, but Pannonius is unusual as a personal name, Lupia is an unusual variant of Lupus / Lupa, and Toisia seems to be unique.

⁷ In ploughing, a month after the previous item, and also noted on the PAS database (LANCUM-277A53). The finder has donated both stones to Tullie House Museum, Carlisle, where Tim Padley made them available.

⁸ In rather square, well-drawn letters c. 90 mm high. A is cut with an apex like that of M, but shorter. The cross-bar of T is rather narrow. A scored setting-out line (not drawn) is visible at the foot of line 4.

⁹ In line 1, the beginning of the third stroke of M survives, confirming it is not A. This is the age at death, whether or

not annos, menses and dies were abbreviated: ‘... 19 (or 29, 39, etc.) years, ... months ... days ...’ In 2, the sequence MACV suggests the phrase uixit ... sine ulla macula (‘lived without any blemish’) as in RIB 1828 (Carvoran). There it is suggested that it has ‘a Christian flavour’, but one of the examples cited (ILS 7518) is of a man ‘whose spirit has been received among the gods’ (cuius spiritus inter deos receptus est); obviously a polytheist, not a Christian. In 3, ATVR suggests the cognomen Saturnina or Saturninus (as in the previous item, but the name is quite common). In 4, although ONIV could be part of Antonius, it is surely part of coniunx (‘wife’ or ‘husband’), whether in the nominative or dative case. In 5, DVMQ is a more likely sequence than DVMO, and [...]dumq[ue] suggests the end of a gerundive such as faciendum or ponendum; this would be an elaboration of the formula titulum ponendum curavit (‘saw to the erection of this tombstone’) which has already been found in the same cemetery (RIB 909, compare 912). Thus, although the name of the deceased is lost and other details are elusive, this is part of quite an elaborate text in which Saturninus commemorated his wife, or perhaps Saturnina her husband.

¹⁰ With another altar, not inscribed, during excavation of a room which was probably the apodyterium, where it was still standing in one corner. David Mason directed the excavation by Archaeological Services, Durham University, on behalf of a partnership of that university, Durham County Council Archaeology Section, the Architectural and Archaeological Society of Durham and Northumberland, and Vinovia. See Mason, D., ‘New excavations at Binchester: results of season 5 (2013)’, Archaeology County Durham 9 (2014), 60–9Google Scholar. The altar is now stored by the Bowes Museum, where Jane Whittaker made it available.

¹¹ In neatly incised rather square letters 30 mm high, carefully massed to fill the space available, at the cost of some irregular spacing; thus the first three letters of REDVCI (2) are spaced out, and in line 7, which is inscribed on the base, the spacing is VS LL M. There are four medial points, after EX (5), ALAE (6), and in 7 after V and the first L, but none elsewhere.

¹² The reading ELTAOMINVS (3) is certain, but this name seems to be unattested, and even its elements unparalleled except for the termination -minus. The military architectus was an immunis (Digest 50.6.7), but not necessarily a legionary, despite the implication of the note to RIB 2091, even though they are most often attested in the legions and Praetorian Guard. They are also attested in the equites singulares Augusti at Rome, who were seconded from provincial cavalry units (M.P. Speidel, Die Denkmäler der Kaiserreiter: Equites Singulares Augusti (1994), nos 9, 223, 321, ?580), but this seems to be the first instance of one in a cavalry ala. The other instances from Britain are at Carrawburgh (RIB 1542) and Birrens (RIB 2091 and 2096), where auxiliary cohorts were stationed, but they do not specify their unit; for a possible instance in the Fleet, see RIB III, 3036.

¹³ During excavation by the Vindolanda Trust directed by Andrew Birley, who made the stone available.

¹⁴ The words and abbreviations are separated by medial points, with a redundant point within initial D; there is no point after the final M, presumably because it was unnecessary. This is only the second instance of the uotum rettulit formula from Britain, the other being RIB 1523 (Carrawburgh), but it occurs a few times elsewhere.

¹⁵ Re-used as a gatepost, but adjacent to the presumed line of the Roman road (Margary 7c) north from Ribchester to Overborough. Patrick Tostevin sent a photograph and other details. It has been published as Shotter, D. and Tostevin, P., ‘A newly-reported Roman milestone from Ribchester’, Trans. Lancashire and Cheshire Antiquarian Soc. 107 (2011), 117–18Google Scholar. It remains in the landowner's possession.

¹⁶ There is no sign of any further lettering. It is unclear whether Maximian was honoured jointly with Diocletian, as in RIB 2256, or alone, as in RIB 2301, which are his only other British milestones. He was western Augustus during a.d. 286–305 as Diocletian's colleague (but Britain was controlled by Carausius and Allectus c. a.d. 286–93), and again in a.d. 307–8. Rather less likely is Diocletian's successor as eastern Augustus, Galerius Maximianus (a.d. 305–11), although one British milestone (RIB 2293) is dedicated to him.

¹⁷ With the next two items (the samian identified by Margaret Ward) during excavation for Cheshire West and Chester Council and English Heritage directed by Garner, D. and Wilmott, T., as noted in Britannia 36 (2005), 420–2Google Scholar, and 37 (2006), 401–2. They will be included with some minor graffiti in Wilmott's final report on the Chester Amphitheatre Project.

¹⁸ All the letters of the alphabet are found singly or in pairs on bone roundels, probably as marks of ownership (RIB II.3, 2440 (a), discussed on pp. 105–8). For an example of K from Caerleon, see RIB II.3, 2440.46.

¹⁹ An abbreviated personal name, probably Blaesus or Blandus. B is of capital-letter form, not cursive. The reading BA is just possible, but the second letter looks more like a slipshod L, its second stroke diagonal, not horizontal. This form is very common, and the writer was probably hampered by the foot-ring, which is why graffiti in this position are often inverted in respect to the vessel, since they were then easier to inscribe.

²⁰ The abacus is described in RIB II.3, p. 105. It was a board inscribed with two rows of parallel columns, up and down which counters were moved to make calculations.

²¹ With the next item during excavation by Northern Archaeological Associates. Alex Croom made them available.

²² The graffito is almost certainly complete, although some potters' signatures on Dressel 20 include a date, for example item No. 14 (Snodland). A name on its own, in the genitive case, may be that of the ‘firm’ rather than the individual potter: Rodríguez Almeida (Il Monte Testaccio (1984), 253–60) notes Quinti, Fortunati, Palladi and Vitalis as typical, but gives no references.

²³ The graffito is incomplete, and its meaning unclear. XV[…] would be a numeral from ‘15’ (XV) to ‘19’ (XVIIII), whether it numbered the jar in a sequence or was a measure of capacity. Personal names in Lug- are very rare, but LVG is a regular abbreviation of Lug(u)dunum (Lyon), and may have been extended to a British place-name incorporating the same element, whether Luguvalium (Carlisle) or Lugudunum (A.L.F. Rivet and C. Smith, The Place-Names of Roman Britain (1979), 401; quite possibly South Shields, see Add. (a) below). An abbreviated personal name such as L(ucius) V(alerius) G(...) is also possible.

²⁴ Unstratified, in a molehill beside the area of the fort displayed. Information from David Mason, who made it available.

²⁵ The lower left-hand portion of the letter after V survives in the broken edge, but looks more like M (accusative) than S (nominative), since there is no sign of the leftward extension of S. The initial et implies that at least two persons were named (and in the accusative case apparently), which means this cannot be a potter's signature. Perhaps a third party is greeting (salutat, or similar) two or more named persons. Eprius is a Latin nomen, but is quite uncommon; it has not occurred in Britain before.

²⁶ During excavation by the Archive and Archaeology Service, Worcestershire County Council (HSM51615 sf 6377), as part of the Yazor Brook flood alleviation scheme. Laura Brook made it available.

²⁷ The name of the potter, for example Lucius. The letters below are incomplete, and their reading not certain. There is room for one more, which might be E for a locative ending -ica[e]; S for a feminine noun in the accusative plural, -ica[s]; or T for a verbal ending, -ica[t]. There is too little room for M, and the word is not part of a date.

²⁸ In demolition/make-up layers (context 617, sf 248) during excavation by Archaeology South-East (UCL Institute of Archaeology) directed by Clive Meaton. Information from Louise Rayner, who made the sherds available.

²⁹ The surface to the left of line 1 is unmarked except for a casual scratch, so there is only a single digit (i) in the date-numeral; exceptionally, the scribe wrote i Idus as an alternative to the usual pr(idie) Idus. Iules is for Iulias, probably because post-consonantal i in hiatus was confused with e (as in Iuleas for Iulias in CIL iv 814) and the a was then omitted. Confusion may have been easier because five out of twelve month-dates ended in -es, not -ias. In line 2, the tip of initial l survives in the broken edge, also the end of the first stroke of u. The second stroke of o was extended downwards and to the left, so as to provide the first stroke of ligatured n. There is another instance of Lucrionis inscribed on a Dressel 20 before firing (CIL xv 3614, Monte Testaccio), but the hand is not the same, and the content of the graffito is different.

³⁰ The TOT and TOT-related rings which follow, under Leicestershire (1), Lincolnshire (7) and Nottinghamshire (3), have all been recorded in the PAS database and in the catalogue made available by Adam Daubney, Finds Liaison Officer for Lincolnshire; see further A. Daubney, ‘The cult of Totatis’, in S. Worrell et al. (eds), A Decade of Discovery: Proceedings of the Portable Antiquities Scheme Conference 2007 (2010), 201–7. Four unprovenanced rings (nos 78, 84, 87 and 88) and six very recent discoveries (nos 82, 83, 85, 86, 89 and 90) have not been included here.

³¹ PAS ref. WMID–63F643 (Daubney 75).

³² T is inscribed rather like I with exaggerated horizontal serif top and bottom, presumably to read the same either way up.

³³ By the late husband of the present owner, who brought it to The Collection, Lincoln, for evaluation (PAS ref. LIN-9C11F6). Its present location is unknown since she declined their offer to buy it, choosing instead to approach a dealer. Information from the Access Officer (archaeology), Antony Lee.

³⁴ The lettering does not clearly distinguish between L, E, F and I, but is confirmed by the context. The first F is inverted, the second has a redundant third horizontal bar. C of fecit is indistinguishable from L of felix. There is also an improper space between F and E of felix. Lettering and decoration are identical to those of the Bishop Norton tank (RIB II.2, 2416.7) found a few miles to the north, which must derive from the same workshop. Although the fragment is now curved, it is uncertain whether this feature is original, and thus would imply the tank was cylindrical.

The Bishop Norton tank reads [...]DO FECIT FELIX [...], the break after felix leaving it unclear whether the inscription ended here. Probably not, for if felix qualified fecit, the adverb feliciter would have been more natural; but the addition of sis or vivas would complete the phrase nicely. Compare the Caistor tank from the same area, which reads (RIB II.2, 2416.4): Cunobarrus fecit vivas. As already noted by RIB, the maker's name was Celtic and ended in -edo; it can now be understood as Malledo, which is well attested as the name of a late Antonine samian potter at Lezoux ( Hartley, B.R. and Dickinson, B.M., Names on Terra Sigillata, V (2009), 225–8Google Scholar, s.v. Malledo (Malledus)). The letter A was probably omitted by confusion with M, unless it was thought to be contained within M, but there is no sign of any cross-bar.

³⁵ PAS ref. LIN-156706 (Daubney 69).

³⁶ The usual O of TOT is represented by a central pellet.

³⁷ PAS ref. SWYOR-9C5DA6 (Daubney 70).

³⁸ Both Ts are inscribed with exaggerated bottom serif formed by two diagonals.

³⁹ PAS ref. LIN-A884E7 (Daubney 72).

⁴⁰ PAS ref. LIN-C79A47 (Daubney 76).

⁴¹ PAS ref. LIN-CB8401 (Daubney 77).

⁴² Reported by autopsy as ‘the letters T.T’. Compare item No. 17 (Marton).

⁴³ PAS ref. NLM-F594B5 (Daubney 79).

⁴⁴ PAS ref. NLM-AA76B4 (Daubney 81).

⁴⁵ During excavation by MoLA (site director Sadie Watson, senior archaeologists Jessica Bryan and Michael Tetreau) of the site of Bucklersbury House for Bloomberg in advance of redevelopment. 200 iron writing-stili were also found. The tablets were made available by Julian Hill, senior post-excavation manager, and will be published as a Museum of London monograph. In deference to the sponsor's wishes, none of the texts already drawn and transcribed can be presented here.

⁴⁶ During excavation (LYD 88) by MoLA. Michael Marshall made it available.

⁴⁷ This die is not found in RIB II.1, 2411 (lead sealings), in which there are many instances of tria nomina abbreviated to their three initials. Some are directly associated with military units, and must be the initials of officers. The initials LAF are those of a cavalry decurion engaged in the building of Hadrian's Wall (RIB 1445), but this may only be a coincidence.

⁴⁸ During excavation (CID 90) by MoLA published as J. Hill and A. Woodger, Excavations at 72–75 Cheapside/83–93 Queen Street, City of London (1999). Michael Marshall made it available.

⁴⁹ The reading can be restored from RIB II.1, 2411.266 (London), a complete example of the same die. Various abbreviated personal names of Greek derivation are possible, including Eustathius.

⁵⁰ PAS ref. NMS-DE40B2.

⁵¹ The most likely god is Mercury (to whom other silver rings are explicitly dedicated; see RIB II.3, 2422.20, 29 and 30), but Mars and Minerva are also possible. DE is difficult as an abbreviated de(o) when D would have been sufficient, but easier as a ‘Vulgar’ reduction of de(ae), which means that Minerva must be considered. However, IN DE for in de(o) is found on the Silchester gold ring (RIB II.3, 2442.14) because space ran out.

⁵² PAS ref. NMS-0B5BB1, identified by Adrian Marsden. It was declared Treasure and acquired by Norwich Castle Museum (acc. no. NWHCM:2014.13).

⁵³ Only S is correctly reversed; N, D and E are not. NT is ligatured. The Christian legend VIVAS IN DEO is engraved rectograde on gold finger-rings from Silchester (RIB II.3, 2442.14) and Brancaster (ibid., 15), but a silver finger-ring and a silver disc both from Silchester provide better parallels of a bezel engraved with a head facing right and retrograde legend (ibid., 25 (IVL BELLATOR VIVAS) and 42 (VIVAS)).

⁵⁴ With the next nine items during excavation by the Vindolanda Trust directed by Andrew Birley, who made them available with other graffiti of less than three letters, which are not included here.

⁵⁵ RIB II.2, 2412 (at p. 1) accepts a figure of 327.45 gm for the Roman pound (libra), which would make this example 4.5 per cent underweight, but it compares closely with ibid., 81 (153.38 gm), 83 (157.83 gm) and 84 (157.21 gm).

⁵⁶ RIB II.4, 2460.48, assuming the diagnostic middle stroke of E has not registered completely. The die has already been found at Vindolanda, ibid. (vii). Two fragmentary stamps of Leg VI V were also found in 2012 (sf 16633 and 16663), and two more in 2013 (sf 17357 and 17633).

⁵⁷ There is apparently a third stroke between the two surviving strokes of M, but it is probably casual, since numerical notes of capacity in m(odii) are often found here on Dressel 20. They mostly range from seven to eight modii, with or without a fraction: see RIB II.6, 2494, p. 33. One modius was equivalent to 8.754 litres.

⁵⁸ There is quite a good trace of the downstroke of T in the broken edge. I widens towards the foot, and the corner of the next letter survives in the break. It looks more like V than S, although the name Vitalis is very common. Possibilities include Natalius and Vitalius.

⁵⁹ The first two letters are incomplete, but the angle of the first suits V better than I or N. The sequence VSPO suggests the end of one name and the beginning of another, i.e. nomen and cognomen.

⁶⁰ Line 2 is metrical, and it may be conjectured that the graffito is part of a poem perhaps amatory. The verb in 2, if present indicative, could be second-person plac[es], or third-person singular plac[et] or plural plac[ent]. The adjective sola would thus be feminine singular or neuter plural, but an attractive restoration is [tu mi]hi sola plac[es], ‘you alone please me’. The verb in 3 is active exting(u)unt, or passive exting(u)unt[ur].

⁶¹ The letters are quite carefully formed capitals and well-preserved, and the reading looks certain, but there is no obvious restoration.

⁶² PAS ref. SWYOR-B1D197 (Daubney 71).

⁶³ TOT abbreviated. The inscription is complete and fills the bezel.

⁶⁴ PAS ref. SWYOR-D86B87 (Daubney 73).

⁶⁵ PAS ref. DENO-8EAFE4 (Daubney 74).

⁶⁶ For the same legend, see RIB II.3, 2422.22 (Benwell), where a dedication to Mars Toutatis is regarded as likely, but two other possibilities are suggested. They should now be withdrawn.

⁶⁷ During excavation by Border Archaeology in advance of a water pipeline. Information from Jane Timby, who sent a photograph.

⁶⁸ The owner's name, but there are too many possibilities to guess what it was.

⁶⁹ In controlled metal-detecting during excavation by Surrey Archaeological Society, directed by David Calow, who sent full details. For the site, see Britannia 43 (2012), 349–50Google Scholar. The find-spot was subsequently excavated, showing that the fragment had been found in an occupation deposit containing Roman pottery and tile, over the undisturbed fill of a ditch which had previously been excavated 10 m away, where it contained pottery dated between a.d. 250 and 400.

⁷⁰ There may be trace of E before X, but this is uncertain. The stop after PATR is very faint. LIB is a common abbreviation for libertus (‘freedman’), but cannot be linked to a conjectural [e]x patr(ono) (‘former patron’). The expansion suggested here is more likely, although there is no British example of the phrase ex patrimonio (suo): it occurs in dedicatory inscriptions in Lusitania and Baetica. lib(ens) would then suggest a variant of the u(otum) s(oluit) l(ibens) m(erito) formula, and imply that the vessel was dedicated to a deity. The site is within 2 km of the Roman temples at Wanborough, but no direct connection has yet been established.

⁷¹ By metal-detector, PAS ref. NMGW-C4CF26. It looks Roman, but is too fragmentary to be identified.

⁷² During excavation by York Archaeological Trust directed by Peter Connelly (see Britannia 40 (2009), 237 Google Scholar). A drawing and full details were sent by Hilary Cool.

⁷³ It is not quite possible to tell from the spacing of the letters, which is slightly irregular, whether LISSO is the beginning of a word, but in other respects this is most unlikely. O is incomplete, but the other possibilities, C, G or Q, cannot have followed SS.

⁷⁴ The sequence LISSO suggests the end of a place-name or masculine personal name in the dative or ablative case, but a place-name like Porolissum seems unlikely, whereas Melissus, Hyalissus or Ilissus are quite well-attested personal names. The difficulty is that salutations, for example on ‘motto beakers’, naturally address the recipient in the vocative case; there is just one instance (AE 1982, 750) of the genitive Melissonis, which would imply a nominative (and vocative) form Melisso, whence a possible [Me]lisso [viva]s (‘Long life to you, Melisso’). Much less common is the formulation with the adverb feliciter and the dative, whence a possible [Me]lisso [feliciter] (‘Good luck to Melissus’), which would not account for the detached S. In default of good parallels, it is not possible to reconstruct the text.

⁷⁵ By Bidwell, Paul, ‘The Roman names of the fort at South Shields and an altar to the Di Conservatores ’, in Collins, R. and McIntosh, F. (eds), Life on the Limes: Studies of the People and Objects of the Roman Frontiers presented to Lindsay Allason-Jones on the Occasion of her Birthday and Retirement (Oxford, 2014), 49–58 Google Scholar.

⁷⁶ See previous note. The digital scan (ibid., 54, fig. 7.3) strongly supports the sequence LVG, and suggests it was preceded by two letters, one of which was N. The final M, of which there is now no trace, was read by Lister; it is possible that what he saw was actually ligatured .

⁷⁷ As noticed by A.R. Birley in commenting on RIB III, 3117, at JRA 2011, 687, n. 32. RIB follows CIL vii 742 in reading VICS[...], which it understands as uics[it ...], but the medial point was recognised by Collingwood in his drawing. In his note to CIL vii 742, Huebner suggested VICo, the diminutive O being his own interpretation of the medial point, but Collingwood Bruce had already recognised it for what it was in LS 287.

⁷⁸ By A.R. Birley, citing G. Köbler, Germanisches Wörterbuch (2007), s.v. ansu.

⁷⁹ *ansuz is generally accepted as a pre-Christian Old German term for ‘god’, whatever its etymology ( Palomé, E.C., ‘L’étymologie du terme germanique *ansuz “dieu souverain”', Études Germaniques 8 (1953), 36–44 Google Scholar); Gothic ansis, which Jordanes (Getica 13.78) glosses as semideos (‘demigods’), although an i-stem, is probably related ( Bammesberger, A., ‘Gotisch ansis und urgermanisch *ans(u)’, Beiträge zur Namenforschung 31.3 (1996), 231–40Google Scholar). The element has been recognised in the names of two uniquely-attested goddesses, Vihansa (CIL xiii 3592) and Hariasa (CIL xiii 8185), by S. Gutenbrunner, Die germanischen Götternamen der antiken Inschriften (1936), 101–2, but has not been found in any dedicatory formulas. Dedications to ‘Veteres’ are virtually confined to northern Britain, more than sixty, a concentration which led Eric Birley to conclude that it was a native British deity, not German (ANRW II, 18.1, 63). They are typically sub-literate, and SVACNV in RIB 1458 (Chesters) is apparently only part of the dedicator's name.

⁸⁰ The inscription is north-facing, and in the shadow of overhanging rock, but the photograph in RIB III was taken by available light. It has now been photographed with the aid of remote flash.

⁸¹ Henig, M., ‘A miniature altar from Waltham Villa, Whittington’, Glevensis 46 (2013)Google Scholar or 47 (2014), forthcoming. It carries the same motif of a circle bisected by a transverse line and quartered with a saltire cross. Henig compares two miniature altars from Chedworth (CSIR i.7, nos 127 and 128) which carry, respectively, a circle bisected by one and two lines at right-angles (compare RIB III, 3448), and a saltire cross between horizontal lines. The ‘wheel’-like motifs may have been solar symbols.

⁸² Dana, D., ‘Un nom féminin thrace dans une épitaphe de provenance obscure (RIB III 3533*)’, ZPE 190 (2014), 162–3Google Scholar, shows that DEMETRVS is not an error for Demetrius but the Thracian feminine name Demetrus, and that SEPETESVCV is not a place-name but a distortion of the Thracian feminine name Eptesucu. NAT THRAEEC is for nat(ione) Thraec(ia).

⁸³ Sold at auction by Bonhams (London) in their Antiquities sale of 3 April 2014 (Lot 70), with details of type and provenance. Despite containing cremated human remains, it would be regarded as a typical ‘Grand Tour’ import, were it not for the provenance suggested. Similar chests have been attributed to the city of Rome in the Claudian period by F. Sinn, Stadtrömische Marmorurnen (1987), 26–7 and 105, nos 54 and 55, with pl. 19.

The chest has been at Ockenden House, Cuckfield, since the eighteenth century, and a British provenance has been deduced from the entry by Timothy Burrell of Ockenden House in his journal for 5 April 1703 ( Sussex Arch. Coll. 3 (1850), 120 Google Scholar) of the discovery on Highbridge Hill of vessels of figured samian with urna sepulchralis Romana cineribus et ossibus humanis repleta (‘a Roman sepulchral urn full of human ashes and bones’). But the editor of his journal, R.W. Blencowe, comments: ‘This must have been a Roman urn with Samian pottery, such as has lately been found on the Downs at Barmer.’ The implication is that Blencowe in 1850, whether or not he actually knew of the ash-chest, did not identify it with Burrell's urna. Even if the chest could be described as an urna, which would be contrary to ancient usage at least, it is still more unlikely that such a sophisticated object would have been imported from Rome into Britain just after the invasion to commemorate a person of Greek (and ultimately Egyptian) name. So it must be supposed that Burrell did indeed record a cremation jar (urna), and that his notice somewhat later (after 1850, it would seem) was taken to refer to another, much grander but unprovenanced, funerary container in the collection at Ockenden House.

⁸⁴ The next six items were noted by Scott Vanderbilt when collecting accession-numbers in the Great North Museum for the on-line Roman Inscriptions of Britain, but he did not have time to check every location.

⁸⁵ RIB 1406 and 1407 belonged to Mr V.L. Benson of Rudchester, who also owned RIB 1400. This too may have been given to the Hancock, but an accession-number has not been found.

⁸⁶ The next five items were donated to the Vindolanda Trust by the executors of the author Rodney Legg, from his collection of Romano-British inscribed or carved stones. They are now stored by the Trust at Vindolanda, and the Director, Andrew Birley, made them available. They resemble many informal but genuine small inscriptions, but they have no provenance and show signs of improvisation and implausible details of lettering and content.

⁸⁷ There is some sign that it was intended to cut away the irregular end of the stone, so as to centre the panel, more or less. The third line of text, if it ever existed, was lost except for the tip of F (or E) when the stone broke away along the edge.

⁸⁸ The inscription seems to have been improvised from a stone which had already been used. Its layout, and in particular the indented second line, suggests an attempt to supply the loss of RIB 1204 (Whitley Castle), but if so, the choice of red sandstone was inept.

⁸⁹ The letters of lines 1 and 2 are c. 25 mm in height, but those of 3 only c. 10 mm, so as to fit them into the die rather than postpone them until the base. D, the first letter, has been ‘pecked’, but the other letters are incised. This peculiarity is shared by (c), despite the two stones being geologically quite distinct. The inscription must also be condemned for its use of the nominative case, the clumsy insertion of line 3, the form BRIGANTA, and the impression it gives of being improvised. It may have been inspired by two altars now in Leeds City Museum, RIB 630 which resembles it and appears to begin with DEΛE | BRIGΛ, and 628 which appears to be dedicated to BRIGANT|A.

⁹⁰ D, the first letter, has been ‘pecked’, but the other letters are incised: compare (b). M is not of Roman form. The ligaturing of A and R is clumsy. A ‘medial point’ divides T from I, although they are linked by a redundant horizontal stroke. I is followed by further redundant ‘pecking’.

⁹¹ S has been cut twice, and the final O ostensibly lost to damage. L looks rather sharp, and E is not altogether Roman. The cult of Sucellus is well attested in Gallia Belgica and Upper Germany, but the god is represented as holding a mallet (G. Webster, The British Celts and their Gods under Rome (1986), 61–2), not as a left-handed warrior. There are only two instances of his name being spelled with one L; one is CIL xiii 6224 (Worms), the other is RIB II.3, 2422.21 (York), incidentally the only evidence of the cult in Britain, a silver finger-ring found in 1875 which is inscribed DEO SVCELO.

⁹² It resembles two small altars found at Vindolanda in 1972, RIB III, 3338 and 3339, the formulation VETERIBVS POS being confined to Vindolanda (as also in RIB 1699, now lost; the verb is spelled out in RIB 1606 at Housesteads). The two dedicators identify themselves by cognomen, LONGI|NVS and SENILIS, unlike the dedicator of (e) who apparently abbreviates his praenomen and nomen, but omits any cognomen, that is if L | CATIV is meant to be L(ucius) Catiu(s). Such a nomenclature is improbable.