² During excavations by the Kent Archaeological Rescue Unit under the direction of Brian Philp, who submitted details and a photograph to RPW.

³ See RIB 1001, Davies, R. W., Cumb. and West. Trans. 2nd. ser. lxviii (1968), 22.Google Scholar

⁴ Analysis of this and the other tablets by Dr A. M. Pollard of the Oxford University Research Laboratory for Archaeology and the History of Art shows that the alloy is 54·2% lead: 44·6% tin (1·2% copper). The proportion of the metals in the others are: No. 3 = 22·8% lead: 77·1% tin (0·2% copper); No. 4 = 65·8% lead: 34·1% tin; No. 5 = 0·3% lead: 997% tin; No. 6 = 60·5% lead: 395% tin; No. 7 = 67·8% lead: 32·2% tin.

⁵ During excavation for the Bath Archaeological Trust directed by Professor B. W. Cunliffe (see Britannia xi (1980), 387–8). Other tablets from the deposit have already been published inGoogle ScholarBritannia xii (1981), 370–9, and xiii (1982), 396–407. As before, we are grateful to Mrs S. Pollard for conservation and information, and to Mr R. L. Wilkins for photographs, both of the Institute of Archaeology, Oxford; and to Ian Marriott for his word and letter-group Index of Curses. Unlike the case with previously published tablets from Bath where we have both been involved, work on the six curses here published is by RSOT alone.Google Scholar

⁶ Alphabets and part-alphabets are a common graffito on instrumentum domesticum. Probably the closest parallel is the part-alphabet cut on a leaden cap from Holt (JRS lvi (1966), 221, No. 22) perhaps, it was suggested, for magical protection. Since the present text was deposited with overt curse tablets, its writer may have intended the break in alphabetical sequence as a covert allusion to cursing: ABC def(i)x(io).Google Scholar

⁷ The date is suggested by the name Basilia (see below) and by the letter-forms, especially A, B, E, M and P. The hand is closely related to that of another fourth-century text, Britannia xiii (1982), 406, No. 7, which is, however, that of someone else: note the differences in A, B, D, G and N, for example. The hand is even closer to that of No. 5 below, which is, however, written more loosely with a finer point, and uses more ligatures; certain letters (for example A, G, N and QV) are formed slightly differently. Both may have been written by the same person, perhaps some years apart, but it is probably best to regard them as ‘contemporary’.Google Scholar

⁸ 1. Basilia: a personal name of Greek origin hitherto unattested in Roman Britain and only twice in CIL xiii (both Christian). In the Latin West, like Basilius, it is unlikely to be earlier than the fourth century. donat: for the formula of ‘giving’ stolen property to a god who is required to ‘exact’ it or to punish the thief, cf. RIB 306 (Lydney), and Britannia xiii (1982), 406, No. 6 (with note ad loc.).Google Scholartemplum Martis: hitherto unattested in Roman Bath, but cf. RIB 140, an altar dedicated to Loucetius Mars and Nemetona found near RIB 152, which records the restoration of a locus religiosus, apparently a temple precinct distinct from that of Sulis. 1–2. anilum: the L was written twice, almost on top of itself. This ‘vulgar’ spelling of anulum also occurs in the Lydney curse (RIB 306). 2. si servus si liber: a formula peculiar to British curse tablets, in which it is common; cf. Britannia xiii (1982), 406, No. 6; xii (1981), 372, No. 8, etc.Google Scholar 2–3. (ta)mdiu: there is room for the missing (ta) at the end of 2, but no sign that it was ever written there or lost by damage. 3–4. siluerit…noverit: the perfect subjunctive in a subordinate clause, the mood dictated by an indirect command explicit or implied, is a favourite usage in British curse tablets (see Britannia xiii (1982), 401, n. 11). See further below, 7–8.Google Scholar 4. ut: translated here as if introducing an indirect command dependent on an elliptical et rogat (or similar), but could be seen as introducing the condition on which the ring is ‘given’ to Mars. sanguin(e) et liminibus: the G was omitted in error and inserted above the N; the & was omitted by haplography; liminibus for luminibus may be another ‘vulgar’ spelling of 1 for v (cf. anilum), but is more likely to be another omission due to haplography, since in this hand the ligatured LI is similar to the v. Catalogues of the parts to be affected are common in curse tablets, but the mention of ‘blood’ seems to be peculiar to Britain. ob: sense requires another et, but ob (not ab) seems to have been written. In late Latin ob is sometimes found with the ablative case, but the usual sense of ‘because of’ is inappropriate here. 5. configatur: a variant hitherto unattested in curse tablets for the usual defigatur. 6. excomesis: this rare compound of comedo occurs only in late Latin, notably in Gildas, De Excidio 85.2, animae viscera excomedens (hunger for the Gospel eats out the bowels of the soul). (om)nibus: (om) was apparently omitted in error. habe(at): apparently another omission in error. As at the end of 2, there is room for the missing letters, but no sign that they were ever written there or lost by damage. The syntax would in fact be more orderly without habe altogether, since intestinis (etc.) would then depend upon configatur like membris (etc.), instead of standing in the ablative(!) case as the direct object of habe(at). 7–8. involavit…fuerit: involavit has been written in error for involaverit; for the perfect subjunctive see above, 3–4. 9. Primurudem: apparently a personal name in the accusative case, which seems to be unattested, though its formation recalls the titles borne by the two leading members of a gladiatorial establishment, the summa rudis and secunda rudis.

⁹ The date is suggested by the A, B, M, and (in No. 7 only), the P (cf. the R). The handwriting of both tablets is virtually identical and includes a distinctive and unusual A. Moreover they are the only texts from Bath to be written on such ‘blobs’ (which have almost the same composition) and have similar layout and content. So the unity of authorship seems certain.

¹⁰ The names of this and No. 7 are the usual mixture from Bath of ‘Roman’ (or Greek-derived) personal names and ‘Celtic’ personal names: see Britannia xiii (1982), 407, n. 15 at 10 f.Google Scholar, and cf. Ibid., 398, No. 4 and 406, No. 7; Britannia xii (1981), 372, No. 7; and RIB 154, for other such lists of enemies or suspects.Google ScholarCunsa: presumably the feminine form of Cunsus (below), but cf. the potter's name Cunis(s)a (F. Oswald, Index of Potters' Stamps on Terra Sigillata (1931), s.v.). Cunsa and Cunsus seem to be unattested (they are not in the Thesaurus Linguae Latinae Onomasticon), but initial Cun(o)- is a common Celtic name-element. Docimedis: also not in TLL Onomasticon, but initial Doci- is a common Celtic name-element; however, the termination -medis (for -medes) perhaps suggests a name of Greek derivation. Sedebelia: apparently unattested, but cf. Sedia (CIL xiii 2492) and Belia (CIL xiii 3469). Maria: the feminine form of the Roman nomen Marius, already attested in a curse from London (RIB 7). Vendibedis: apparently unattested, but initial Vind- is a common Celtic name-element. Cunsus: unless the flourish(?) after N is an attenuated letter, this is presumably the masculine form of Cunsa (see above). Severia(a)nus: the writer seems to have begun a second A by mistake, started to write N over it, and then wrote a second N instead. The name is derived from the common Roman cognomen Severus, like Severinus also on a Bath curse (RIB 154), and Severiana (RIB 563). Seni(i)la: the 1 was written twice, presumably in error. The name is attested as a samian potter's (Oswald, Index s.v.) and in CIL viii 13133.

¹¹ The hand is closely related to that of No. 3 above and of another fourth-century text, Britannia xiii (1982), 406, No. 7: see n. 7 above.Google Scholar

¹² 1. execro: hitherto unattested in curse tablets, but cf. defico (RIB 7) and devoveo (Britannia xii (1981), 375, No. 8) in much the same sense.Google Scholarinvolaverit…perdiderit: grammatically the indicative mood would be correct (as in No. 6) but the writer seems to have half-recalled the formulaic perfect subjunctive in subordinate clauses where the mood is dictated by an indirect command (see n. 8 above s.v. siluerit). 2. The whole of this line, but only this line, has been written with the sequence of letters reversed, like the whole of Britannia xiii (1982), 406, No. 7 (see previous note) and of RIB 154.Google ScholarDeomiorix: this Celtic personal name is apparently unattested, but initial Deio-/Dio- and the termination -rix are both well attested name-elements; cf. also CIL iii 10324,….]amiorix. The nominative is a solecism for whatever oblique case(s) would be appropriate after involaverit and perdiderit. 2–3. de hos(i)pitio suo: in classical Latin hospitium almost always means ‘lodgings’, but is probably used here in its derived and ‘vulgar’ Latin sense of ‘house’. 3–5. quicumque re(u)s deus ilium inveniat: res, which is what seems to have been written, is obscure, and the syntax is faulty, but the general sense of the tablet requires this to be a reference to the thief. Although res in British curse tablets means ‘property’ (JRS xlviii (1958), 150, No. 3;Google ScholarBritannia x (1979), 343, No. 3), it makes better sense to see it as a mistake for reus, cf.Google ScholarBritannia xiii (1982), 401, No. 5. The writer seems to have been uncertain whether to subordinate to deus inveniat an indirect question (quicumque sit reus) or an accusative (quemcumque reum), and blundered by combining elements of both.Google Scholar 5–6. sanguine et vitae suae illud redemat: this, like qui involaverit, reads like a half-recalled formula, cf. No. 6 below, hoc donum non redemat nessi sangu(i)ne suo, and RIB 323 (Caerleon), non redimat ni(si) vita sanguine(i) sui, where in the light of these two parallels the preferred reading Sanguinei (‘;Blood-red charger’) must now be abandoned. illud: the two parallels just cited suggest that this must mean ‘that (gift to the god)’, whether of the property stolen or, more probably, of the thief himself. vitae suae: genitive written ungrammatically for the ablative case. redemat: ‘vulgar’ spelling for redimat, as in No. 6.

¹³ The letters are of ‘classical’ form except for the E made with two curved strokes, which is usually regarded as post c. 200.

¹⁴ This is the tablet cited in Britannia xii (1981), 375Google Scholar, n. 21, and compared with Ibid., No. 8, which also curses the thief of a caracalla. Despite two similarities of language (involavitjinvolaverit and si servus si liber), both formulaic and easily paralleled in other Bath tablets, it would seem that the tablets refer to two separate incidents. (Theft of a caracalla cannot have been unusual in a bathing establishment). The fine ‘rustic capitals’ of Britannia xii (1981), 375, No. 8, are arguably earlier in date than the present tablet; the differences in formula are striking: note the different titulature of Sulis, the unusual si baro si mulier of the present tablet, its cursing of the thief without reference to the possible return of the caracalla.Google Scholar 1–2. Minervae de(ae) Suli: elsewhere, when Sulis is identified with Minerva, her name precedes that of her Roman counterpart. 2. de(ae): the (ae) was omitted perhaps by haplography, in view of the similarity in sound between e and ae. donavi: it is more usual to ‘give’ the stolen property to the god (e.g. Britannia xii (1981), 370, No. 6, donavi argentiolos sex), but cf.Google ScholarIbid., 375, No. 8, devoveo eum (etc.), and Britannia xiii (1982), 398, No. 3, and 401, No. 5, both of which, as it were, give Sulis the nomne of the thief.Google Scholar 3 f. furem qui caracallam meant involavit: cf. Britannia xii (1981), 375, No. 8, eum qui caracellam meam involaverit.Google Scholar 6–7. si servus si liber: this formula, which also occurs in Britannia xii (1981), 375, No. 8, is common in British curse tablets, but peculiar to them.Google Scholar 7–8. si baro si mulier: a variant of the more usual si vir sifemina formula which also occurs in an unpublished curse tablet from Brandon (Norfolk). In classical Latin baro always means ‘fool’; only in medieval Latin does it gain the meaning ‘man’. The present text, which contrasts baro with mulier and thus guarantees that it means ‘man’, is evidence for the first time that this meaning was already current in the spoken, ‘vulgar’ Latin of the Roman period. 9–11. hoc donum non redemat nessi sangu(i)ne suo: for the formula of a ‘gift’ which the victim ‘recovers’ at the cost of his blood, cf. No. 5 above and RIB 323. redemat: ‘vulgar’ spelling for redimat, as in No. 5 above. nessi: ‘vulgar’ spelling for nisi, cf. Britannia x (1979), 342, Nos. 2 and 3 (Uley), nissi for nisi.Google Scholarsangu(i)ne: for the spelling cf. JRS xlviii (1958), 150, No. 3 (Kelvedon), sanguno.Google Scholar

¹⁵ See n. 9 above.

¹⁶ The left-hand column was completed first: two of the final Si were overwritten by the right-hand column.

¹⁷ See n. 10 atiove.

Victorinus: a popular name derived from the Roman cognomen Victor, one of the commonest names in Roman Britain.

Talipieinus: apparently unattested.

Minantius: apparently unattested but a plausible Roman nomen formed from the present participle minans.

Victorianus: like Victorinus formed from Victor, but apparently unattested in Roman Britain. Compe(pe)dita: the writer seems to have carried the name over to a second line and to have repeated -pe- by mistake. Apparently unattested, but cf. Compedia (CIL ii 5035, Lusitania).

Valauneicus (?): the E is unlike the E which appears elsewhere on these two tablets; it is written over both the N and the 1. The name is apparently unattested, but is probably cognate with the Celtic personal name Vallaunius (RIB 369).

Belia (?): the first two (?) letters are uncertain; the word seems to begin with an A like that in Victorianus, but B was written next, the bold diagonal stroke being added later. While x and s (though elsewhere of different form) are both possible, this stroke should perhaps be seen as a crossing-out, Belia being a possible name (CIL xii 3469, Nimes).

¹⁸ The letters are not barred but are presumably intended as a numeral. They were noticed by Mr P. A. Barker in about 1979 and brought to the notice of RSOT.

¹⁹ During excavation for West Yorkshire Metropolitan County Council directed by Mr A. B. Sumpter (see above p. 295), before a new floor was laid in the tower to support the Saxon standing stones from the church-yard, with which this fragment is now displayed.

²⁰ The text is too worn and fragmentary for restoration to be possible, but a few comments may be made:

2 invicto s[…]. The epithet is applied to emperors (invictus Aug.) and to various gods, notably Sol, but in the sequence deus Sol invictus (the deo invicto Soli of RIB 1272 being anomalous). s[acrum] is just possible, cf. RIB 1041, Silvano Invicto sacr(um).

4 Perhaps [fortissimo A[ug(usto)] (cf. RIB 1282) or [invicti]ssimo A[ug(usto)] (cf. RIB 450).

5 suorum suggests the phrase pro salute sua et suorum, but ob conservatam salutem suam suorumque (RIB 649) occurs, and perhaps there was something of the kind here.

7 [f]elicis[…] suggests the sequence of imperial titles pii felicis Aug(usti), dependent perhaps on some pro salute formula (cf. RIB 905, 1272, 2066), but the surviving traces of (6) do not support this.

8 Presumably […]e v(iro) djarissimo) legiato) A[ug(usti/orum) pr(o) pr(aetore)]. If the preceding E has been correctly read, which is far from certain, the legate's name was in the ablative implying a dating formula (e.g. sub or curante). Since only one emperor seems to be mentioned, Alfenus Senecio can be excluded, but there are other possibilities, e.g. Claudius Xenophon (A.D. 223). Language and lettering suggest the inscription is Severan or later.

9 Perhaps the formula [devot]issimus nu[mini maiestatique eius/eorum], cf. RIB 2066, etc.

What remains of the inscription suggests a dedication to two or more gods for the well-being of the dedicator, probably an official or officer, and of the emperor, dated by governorship; but there are no obvious parallels to suggest more than this.

²¹ During excavation directed for the York Archaeological Trust and the Department of the Environment by Mr P. J. Ottaway, who made it available and provided a drawing and other details, including a detailed descrip- tion by Mr B. R. Hartley.

²² See Eburacum: Roman York (Royal Commission on Historical Monuments, 1962), 121a, no. 68, for a relief of Mercury from York associated with money bag, cockerel and stag.Google Scholar

²³ If the graffito was centred, 3 or 4 letters have been lost after N; part of a vertical stroke survives of the first, suggesting Martini or Martiniani.

²⁴ During excavation for the West Yorkshire Metropolitan County Council directed by Mr A. B. Sumpter, who made it available and provided a drawing and other details. It will go to Wakefield Museum, unless a place for the finds is possible at Castleford itself.

²⁵ Mr Sumpter notes that there was a spring at the cemetery of Roman Castleford, as the name suggests, in Holywell Wood (see Yorks. Arch.J. xlii (1971), 3); this may explain why a dedication to the Nymphs resembled a tombstone.Google Scholar

²⁶ By Mr R. Charlton in a garden wall. It has been deposited by him on loan in the Museum of Antiquities, Newcastle-upon-Tyne, where Miss L. Allason-Jones made it available and provided information.

²⁷ This pair of vexillations is unattested, but at Risingham there is a notable dedication slab of the Coh. IIII Gallorum (RIB 1227) and a building stone of a vexillation of a Coh. Nerviorum (numeral lost) (RIB 1240); vexillarii of Coh. II Nerviorum are also attested at Carrawburgh (RIB 1538).

²⁸ Excavations directed by Kevin Blockley and Marion Day for Avon County Council and the Manpower Services Commission (see above p. 317–19). Information and drawings of this and the following item from Kevin Blockley.

²⁹ By D. Allen, who supplied details and rubbings. For subsequent excavations on the site of the villa in 1979 see Britannia xi (1980), 373.Google Scholar

³⁰ Excavations by P. J. Woodward and A. Graham for the Wessex Archaeological Committee and the Department of the Environment. Information and photograph from P. J. Woodward. For the site see above p. 324–6.

³¹ By Mr H. J. D. Bennett. Information and rubbing from Mrs K. A. Rodwell.

³² Excavations for the Cirencester Excavation Committee directed by Alan McWhirr (Antiq. Journ. liii pt. ii (1973), 203–5). Information on this and the following seven items from Linda Viner.Google Scholar

³³ Excavations for the Cirencester Excavation Committee directed by Alan McWhirr.

³⁴ Excavations for the Cirencester Excavation Committee directed by Alan McWhirr (Antiq. Journ. lviii (1978) pt. i, 66).Google Scholar

³⁶ Excavations for the Cirencester Excavation Committee directed by Alan McWhirr (Antiq. Journ. lviii (1978) pt. i, 73–7).Google Scholar

³⁶ See n. 35.

³⁷ For the expansion M(ensura) see Britannia viii (1977), 436, No. 44 and 442, No. 88. The actual unit of measurement here is uncertain.Google Scholar

³⁸ Excavations for Cirencester Excavation Committee directed by Alan McWhirr. For the site see Antiq. Journ. liii (1973) pt. ii, 202–3.Google Scholar

³⁹ By Mark Maillard. Information an d drawing from David Viner. We would like to thank Dr R. M. Reece for reading the legend on the obverse.

⁴⁰ During excavations by the Oxford Archaeological Unit directed by D. Miles and S. Palmer. D. Miles provided details. The numeral consists of two verticals with serifs at to p an d bottom, instead of a single superscript ba r above both verticals as one might have expected, but enquiries would seem to rule, out a modern hoax. For the site see Britannia xiii (1982), 377–8 an d above p. 314.Google Scholar

⁴¹ During excavations for the Gloucester City Museum Excavation Unit directed by C. M. Heighway. For the site see Britannia vi (1975), 273. Information on this an d the following four items from C. Ireland.Google Scholar

⁴² The most likely restoration to be suggested by I. Marriott's analytical index of personal names in RIB i and CIL xiii, is Am]and(us) but other restorations are also possible.

⁴³ A. Holder, Alt-celtischer Sprachschatz, includes both Epius an d Eppius. It derives from epos, the Celtic word for horse.

⁴⁴ Found by Mr Laurie Draper. Information form Dr W. R. Trotter. The sherd was found about 100 m N.E. of the findspot of the Blackmoor coin hoard.

⁴⁶ Excavations directed for the Department of Urban Archaeology of the Museum of London by D. Perring. We are most grateful to the Department and to Dr M. M. Roxan for allowing us to see the text of her report on the diploma prior to the final publication of the site.

⁴⁶ The placing of the surviving fragment in relation to the restored text is to some extent arbitrary since the binding holes ar e not preserved between lines 1 and 2. In FIG. 41 to o much width has been allowed for the restored outline of the diploma.

⁴⁷ See Morris, J. and Roxan, M. ‘The Witnesses to Roman Military Diplomata’, Arheoloski Vestnik xxviii (1977) 299–333, Table 2 (p. 309).Google Scholar

⁴⁸ Information and drawing from Jane Cowgill of the Museum of London. Julian Bennett, whom we have consulted, thinks this is unlikely to be the end of apilum murale because: 1. it has a round and not a rectangular section, 2. where inscribed as in the examples from Oberaden, with the names of centurions, these are given in the genitive and are cut near to the central hand grip, and 3. pila muralia are normally of oak. See Bennett, J., ‘The Great Chesters pilum murale’, AA fifth series x (1982), 200–5.Google Scholar

⁴⁹ For the site see Merrifield, Roman London 201, No. 44. Drawn by R. P. W. in 1953 in Guildhall Museum. Now in the Museum of London; Mrs J. Hall provided a cast to confirm the reading. The cutter omitted the foot of the L in Aelius on the rim of the footstand. He found it difficult to incise the names on the slope of the footstand, omitted one 1 in the second name and made letters of varying size, H may have been the terminal letter.

⁵⁰ See Oxyrhynchus Papyri xiv, No. 1731,1. 9.

⁵¹ Excavations for the Department of the Environment directed by A. E. Brown who supplied a photograph and full information.

⁵² With a metal detector by Mr K. Clarke, who has donated it to the Museum of Antiquities, University of Newcastle-upon-Tyne, from where Miss L. Allason-Jones sent a drawing and details.

⁵³ Another sealing of the Twentieth Legion, with a different reverse, is known from Leicester (Britannia vii (1976), 387, No. 40). Unlike the Second and Sixth Legions, this legion is not represented among the sealings from Brough-under-Stainmore.Google Scholar

⁵⁴ During excavations for the Joint Management Committee of the Department of the Environment and the Vindolanda Trust directed by P. T. Bidwell. We are most grateful to Dr Roxan for allowing us to see her report in advance of the full publication of the site, MWCH.

⁵⁵ A. K. Bowman and J. D. Thomas, Vindolanda: The Latin Tablets No. 30.

⁵⁶ Roxan, M., ‘The Distribution of Roman Military Diplomas’ Epigraphische Studien xii (1981), 265–86 (p. 279–83).Google Scholar

⁵⁷ Chesters unlike Vindolanda lists the alae and cohortes by name on the inner face, but this is not a conclusive argument for a different date since such scribal inconsistencies d o occur in the diploma texts.

⁵⁸ During excavation for the University of Newcastle-upon-Tyne an d the Department of the Environment directed by Messrs C. M. Daniels and J. P. Gillam. M r P. Moffat made this and the following items available, with some too fragmentary for inclusion here, which will all be published in the excavations (1974-81) report.

⁵⁹ Probably Vitalis, but Vi[tali]anus is found at Housesteads (RIB 1592).

⁶⁰ Probably a ‘Celtic’ name-ending, e.g. Sacrovir(os), Senovir(os), etc.

⁶¹ There are a few Latin adjectives in -antivus and -entivus, but apparently none in -untivus; just possibly a theophoric personal name from Vanauns (or Vanauntes), a god worshipped by Tungrians (RIB 1991).

⁶² This ‘Celtic’ personal nam e is usually spelt Bellicianus (e.g. CIL vii 1255, Caerwent), but the present spelling is also found (e.g. RIB 375, Caerleon).

⁶³ It occurs elsewhere only on an altar from the shrine of Nehalennia in the East Scheldt estuary (P. Stuart and J. E. Bogaers, Deae Nehalenniae (1971), No. 30 = AE 1975, 644) and, as Neutto, in the territory of the Tungri (CIL xiii 3628). Coh. I Tungrorum garrisoned Housesteads in the third and fourth centuries, but since it had served in Britain since at least the Flavian period (Tacitus, Agricola 36.1, etc.), the survival of this personal name more than a century later is remarkable.

⁶⁴ Paiatius, unless a mistake (e.g. for Pacatius, cf. RIB 1599 (Housesteads), Pacatianus), seems to be unattested; it is apparently formed from the very rare ‘Celtic’ personal nam e Paius (CIL v 1956, cf. AE 1976, 392).

⁶⁵ The first letter could be G. Presumably the owner's initials, cf. RIB 1389, a centurion called S £(?).

⁶⁶ During excavation directed for the Department of the Environment by Dr G. Webster (see above p. 302–3), who made it available to RSOT.

⁶⁷ The graffito is complete, and is presumably an abbreviated personal name (Albanus, Albinus, etc.).

⁶⁸ During excavation directed for the York Archaeological Trust and the Department of the Environment by Mr P. J. Ottaway, who made it and the next item available.

⁶⁹ The die does not seem to be one of those collected by RPW in Britannia vii (1976), 224–35.Google Scholar

⁷⁰ The stamp is apparently unattested, but the mortarium stamp AGRIPP has been found at York: see R. M. Butler (ed.), Soldier and Civilian in Roman Yorkshire (1971), 138–9.

⁷¹ Excavation directed by W. H. Manning for the Department of Archaeology, University College, Cardiff and the Ancient Monuments section of the Welsh Office. Information from I. R. Scott.

⁷² The name Fidus/a is attested 90 times in CIL – no other name is suggested in I. Marriott's Analytical Index of personal names in CIL xiii and RIB i.

⁷³ The National Trust's Administrator, Mr C. R. A. Olliver, allowed access. The inscription was condemned on epigraphic and historical grounds which need not be repeated here, since it was subsequently learnt that it was cut by Mr R. Dower (for whom see Britannia vii (1976), 381, n. 23) in 1959.Google Scholar

⁷⁴ By Mr L. P. Wenham, among the Hutton (of Marske in Swaledale) papers in the Hailstone Collection. They include a later copy of ‘A copie of the ingraven wordes & letters in one of the two ancient columnes found interred in Thornborough in June 1620’. (Camden does not note the discovery of two ‘columns’, and gives the date as 17 July 1620).

⁷⁶ Hutton's autopsy is proved both by his failure to recognize the second (miniature) A in ARAM and by his recognition of two miniature Os and a ligatured 1 in APRONIANO; he also transcribed the ligatured ET in line 9 correctly, and his EF in line 6 is nearer the true reading than Camden's; but he ‘corrected’ the EM ligature in line 2, and failed to divide V1TALIS. His transcription of medial points is not altogether reliable, but his insertion of a point between 1 and R in line 3 is of interest since it would make the restoration T(itus) Ir/das still more difficult.

⁷⁶ By RSOT with Miss L. Allason-Jones, who is publishing the finds from the Well.

⁷⁷ ‘Gabinius Saturninus (son of) Felix’, apparently, but ‘Gabinius (son of) Felix (and) Saturninus’ is possible, cf. RIB 1531, from which Gabinius and Saturninus are restored, where it is not clear whether one man or two is meant. Without RIB 1531, Saturni(ni) might have been preferable. Felicis is a patronymic: both Is have been symmetrically displaced; note that Gabinius has also been blundered and that all the Si are reversed. Felix and Saturninus are both common names in Roman Britain, but Gabinius otherwise occurs only in RIB 488, the third-century tombstone of a soldier of Leg. II Aug. at Chester(!) called Gabinius Felix.

⁷⁸ By RSOT at the request of Dr G. Webster.

⁷⁹ Fo r the reading offered here (with which RP W concurs), and the identification of the civitas Corieltauvorum, see more fully RSOT, ‘No n Coritani sed Corieltauvi’, Antiq. J. lxiii (1983), forthcoming. The aid of Professor A. L. F. Rivet and Professor Colin Smith, and their Place-Names of Roman Britain (1979), is gratefully acknowledged.Google Scholar

⁸⁰ The B resembles that on the two incense-burners from Carrawburgh (RIB 1530, 1531) and seems to be transitional between the classical and late-Roman forms; the N is made with two strokes.

⁸¹ The names are all, except Litorianus, attested in Roma n Britain. Since they all begin with L and include two cognate -us/-ianus pairs, they may only be writing practice.

⁸² cf. Senovara, , Britannia xiii (1982), 398, No. 4.Google Scholar

⁸³ Mrs P. Colman, who was editing the manuscript of John Aubrey, Monumenta Britannica (now published in two volumes (1980-82), ed. J. Fowles), sent a photograph of Aubrey's transcript (Ibid., vol. 2, p. 663) of an inscription ‘found in the rebuilding of Arundel House in the Strand, July 1677’, which was identified by RSOT. The transcript is identical with that in CIL, except for MENSIBVS (instead of MESIBVS) in 1. 4, which is presumably a scholarly correction of what the stone read. Fo r the Arundel Collection and its dispersal, see D. E. L. Haynes, The Arundel Marbles (1975). RIB 2319* and 2320* are probably also strays from the Collection.