² Directed by Ken MacGowan and Mike Stone for the Passmore Edwards Museum. Information from Pamela Greenwood, who submitted the stone for inspection. For two stamped tiles from the same site; see below Nos. 42 and 43.

³ During excavation for the Eccles Excavation Committee (see Arch. Cant, lxxxvi (1971), 29, with fig. 1) directed by Mr A. P. Detsicas, who made it available to RSOT. This entry is based upon ‘A late-Roman defixio (curse tablet) from the Eccles villa’ in Arch Cant, cii (1985), 19–25.

⁴ This is the only British curse tablet to be written boustrophedon, but other inversions (e.g. of letter-order) are found. The A is unusual (found also in Britannia xiv (1983), 339, No. 4 and 341. No. 7), there are two Ds (one upper-case), two Es (one of earlier form), and two Ss (one of earlier form).

⁵ Butu seems to be a thief whose health is cursed until he returns stolen property to ‘the house of God’. This phrase may have Christian significance. The syntax is confused and the text contains formulas and Vulgarisms (e.g. sanetate for sanitate(m) ).

Commentary

Side (a)

Five letters only, all but the T elongated, like the address on a letter. The first three resemble 5 in sanetate (b₃). The horizontal line drawn across them afterwards suggests an abbreviation, like the double S in Britannia x (1979), 344, No. 3 (Uley) and JRS liii (1963), 122 ff. (Ratcliffe-on-Soar), deo s(upra)s(crip)to. The third S remains unexplained.

Side (b)

1. donatio: the stolen property, or the thief himself, is commonly ‘given’ to the god (dono, donai, etc.). This is the first instance of donatio.

diebus: properly the dative plural of dies (day), but since curse tablets are usually addressed to a god, diebus is probably an error for dibus (‘to the gods’) or a ‘Vulgar’ spelling of the variant form diibus.

quo: its antecedent is presumably donatio (f.), and it is thus a solecism for qua.

2. peril: contracted form of the perfect indicative of pereo (perish), the intention being apparently to state as an accomplished fact what it is hoped the curse tablet will achieve. One would have expected the present subjunctive pereat.

Butu resque: this seems the likeliest word-division, since stolen property is decribed as res, e.g. Britannia x (1979), 343, No. 3 (Uley), donatita ut exsigat istas res quae s(upra)s(scrip)ta(e) sunt. The repetition of que might be an error, but it seems better to take the first as a conjunction and the second as a ‘Vulgar’ spelling of qu(a)e.

Butu: cf. CIL xiii 10010, 373, BVTV F (a samian potter), but this may be an abbreviated form of the well-attested But(t)ur(r)us. Butto is found in Pannonia and Buttus in Noricum as a cognomen.

3. fu[…]: corrosion-damage makes reading and restoration uncertain, but perhaps fu[rat] (for furatur (‘he steals’), since deponent verbs disappear from Vulgar Latin).

3–4. nee ante sa│netate nee salute: the division of sanetate between lines means that the text was written, and intended to be read, in the line-sequence followed in these notes. Sanetate is a ‘Vulgar’ spelling of Sanitate. The ablative case is unexplained: one should either understand fruatur (‘let him not enjoy health …’) or, more likely, see it as an example of the ‘Vulgar’ confusion between case-endings which leads to the omission of the accusative -m. A formula seems to be intended like Britannia x (1979), 342, No. 2 (Uley), erogat deum Mercurium ut nee ante sanitatem habeant nissi …

5. nesi qua(m): the dotted letters are damaged, and there is an unexplained diagonal below N (not however suggesting D), but this reading seems reasonable. Nesi is a ‘Vulgar’ spelling of nisi found elsewhere, followed by qua(m) with ante as its antecedent.

in do[m]o dei: there is a trace of what may be M. Domus is used with the god's name (domus Veneris, etc.) in the sense of ‘temple’, and British curse tablets often require the return of stolen property to a fanum or templum, but domus dei is a common phrase in Christian authors in the sense of ecclesia, whether meaning the Church (or Christian community) or a church (see TLL s.v. domus, 1970, 17ff.). The polytheism of diebus (i.e. dibus) does not preclude, though it makes less likely, a Christian interpretation of in domo dei: Britannia xiii, (1982), 404, No. 7 (Bath) invokes Sulis, but also alludes to Ch(r)istianus. The Anglo-Saxon place-name Eccles, rare in the south-east and hinting at a sub-Roman Christian community (see A.C. Thomas, Christianity in Roman Britain to AD 500 (1981), 262ff.), makes it tempting to give in domo dei a Christian significance when it occurs in a local fourth-century document. But this must remain a hypothesis.

6–7. Damaged by corrosion. There seems to be a repeat of the formula in 3–5, so in do[mo dei] is a reasonable restoration, but the space is cramped.

7–8. Corrosion and other damage have left only isolated letters recognizable, and no restoration seems possible.

⁶ By Mrs Barbara Bagge. Information, squeeze and photograph from A. P. Garrod of the Gloucester City Museum Excavation Unit.

⁷ For an earlier find of a ‘centurial stone’ from Gloucester, see Britannia iii (1972), 353 No. 5 = Ant. Journ. liv (1974), 9, 10 + PL. I (a). This stone, however, did not mention the legion, only a cohort, and, presumbly, century. The present inscription emphasises the connection between Legion XX and Gloucester cf. RIB 122, the tombstone of a serving soldier (now lost), and Britannia xv (1984), 344, No. 1, the tombstone of a veteran of the legion. It is possible that it records work on a stone building in the predominantly timber fortress first constructed on the later coionia site in the 60s A.D., after the redeployment of legions on the withdrawal of Legion XIV in 67. However, a subsequent phase of building activity is of course also possible.

⁸ By the Committee for Rescue Archaeology in Avon and Gloucestershire, directed by Ann Ellison, who provided a drawing (by Joanna Richards) and full details. For other epigraphic material from Uley see Britannia iv (1973), 324, No. 2; x (1979), 340–5, Nos. 2–4; xi (1980), 411–12, No. 36; xii (1981), 370, No. 5 and below, No. 50.

⁹ Similar to one from Niederbieber, cf. in general J. M. C. Toynbee, ‘A Londinium votive leaf or feather and its Fellows’ in Bird, J., Chapman, H. and Clark, J. (eds.). Collectanea Londiniensia: Studies presented to Ralph Merrifield (1978), 129–47Google Scholar , and for the Niederbieber example 136–7, No. 16 and fig. 5.2.

¹⁰ During excavations for the Bath Archaeological Trust (see Britannia xi (1980), 387–8) directed by Professor B. W. Cunliffe, who made it available to RSOT. It has been chosen for inclusion here as one of the most interesting of the texts which await publication in the Final Report. The tablet has been analysed by Dr A.M. Pollard of the Oxford University Research Laboratory for Archaeology and the History of Art: 64.1% lead: 35.9% tin (0.1% copper).

¹¹ Commentary

1. Docimedis: only the vertical strokes survive of both Ds, but the name can be restored with certainty from its occurrence in Britannia xiv (1983), 339, No. 4. The name is not found in TLL Onomasticon, and is probably

‘Celtic’ like others in Doci-, but Greek derivation is also possible. It has been written in cursive letters (M and 5 being of New Roman Cursive form), with the same instrument as the rest of the text in capitals, and clearly belongs to it. A few other texts from the spring (e.g. Britannia xiii (1982), 400, No. 4) combine cursive with capitals, but not in this way.

2. [p]erdidi(t): perdidi (‘I have lost’) could of course be read, but the scribe had the ‘Vulgar Latin’ habit of omitting the final -t of the 3rd sing, verb forms (cf. involavi(t), destina(t)).

2–3. manicilia dua: the word manicilium, a diminutive of manica (‘sleeve’), is virtually unique, being attested only (in the spelling manicillium) as a gloss of the Greek word χεΙρΊδΙον (C. Gloss. Lat. II, 476, 24). This may mean ‘glove’ in a medical writer quoted by Oribasius (vi 18.5, used for massage). In the words of DrWild, J. P. (Bonner Jahrbiicher lxviii (1968), 227)Google Scholar , ‘Gloves were unknown in classical antiquity and long sleeves (manicae) served instead to protect the hands against cold’. The Elder Pliny's secretary may only have worn long sleeves to protect his hands when writing, though mittens would seem more likely (Pliny, ep. iii 5.15, cuius manus hieme manicis muniebanlur). But leather gloves, not sleeves, were certainly worn to protect the hands against thorns in rough farmwork: Odyssey xxiv 230, χειρἲδάζ τ'επἱ χερσΊ βάτων ἑνεκ, and Palladius, i 42(43), 4 (farm equipment), tunicas vero pellicias cum cucullis et ocreas manicasque de pellibus, quae vel in silvis vel in vepribus rustico operi et venatorio possint esse communes. It was a glove like this that the raven stole from St Columbanus: tegumenta manuum, quos Calli uuantos uocant, quos ad operis labore solitus erat habere (Ionas, vita Columbani i 15). The biographer evidently thought it would be something unfamiliar to his readers. Manicilia, forming a pair detached from any garment, are more naturally taken as ‘gloves’ than ‘sleeves’, but the question must remain open. Perhaps it should be added that St. Columbanus Cursed the raven – it would not rear its chicks (cf. Britannia xii (1981), 376–7, No. 8, nee natos nee nascentes) -unless it returned the stolen property. The bird brought the glove back.

4. illas: a solecism for ilia unexpected after the hypercorrect dua. ‘Vulgar Latin’ tends to treat the neuter plural as a feminine singular, but that is not the case here; rather it seems to be analogous with the Italian idiom of coupling the femine plural article le with the plural in -a for some nouns signifying pairs of things (e.g. le labbra, le braccia). involavi(t): see note to perdidi(t).

5. ut introduces an indirect command dependent upon a verb like rogo in ellipse.

mentes sua(s): this is certainly what was written. For the curse, cf. RIB 7, Tretia(m) Maria(m) defico et illeus vita(m) et me(n)tem, etc. The plural mentes perhaps anticipates the plural otoculos (7). The V of sua(s) is curtailed because the stilus struck the edge of the tablet which had already been folded over (see FIG. 3); A is therefore the final letter and S has been omitted, either because the scribe confused feminine and neuter plural endings (see note to illas), or because of the ‘Vulgar’ tendency exemplified in RIB ₇ of dropping the final consonant.

6. perd[at]: the sequence of strokes (the diagonal succeeded the two verticals) makes it clear that the scribe wrote N, not AT; there is no trace of a vertical stroke to accompany the cross-bar of rand, despite damage here, it seems that no such stroke was ever made. Perdnt(l) thus seems to be a copying error by a scribe who would have written perdat as perda(t) without the final T, just as he wrote perdidi(t) and involavi(t). The verb occurs in other tablets as to ‘lose’ property (Britannia, xii (1981), 371, No. 6, cf. xiv (1983), 339- No. 5 (confused)) and to ‘lose’ one's own life (Inv. no. 668, ani(m)am pe(r)d(e)re sui). The author has used the two senses in echoing perdidi(i) with perd[at] (‘tit for tat’), a rhetorical touch in keeping with the pretentious ending (8–9) of his text.

7. oculos: for other curses on sight, which reverse the healing power of the spring (cf. Solinus, Collectanea Rerum Memorabilium 4, 6), see Britannia xiv (1983), 338, No. 3 (understanding luminibus) and Inv. no. 691, non Mi permittas oculos nisi caecitatem.

su[o]s: the final horizontal stroke of the second S survives. The identity of terminations evidently saved the scribe from writing suo(s).

8–9. in fano ubi destina(t): understand Sulis. For the spelling destina(t) see note to perdidi(t) (2). The formula does not occur in any other Bath tablet, and may be the author's own literary invention. No other Bath curse specifies that the breakdown in health is to occur ‘in the temple’. This idea is found at other sacred springs (e.g. ps.Aristotle, Mirabilia, 152 = Philostratus, Life of Apollonius i 6), but the author seems to have been more immediately influenced by British curses which require the return of stolen property ‘to the temple’ (e.g. RIB 3061 Britannia xii (1981), 376–7, No. 8; x (1979), 343, No. 3)

¹² By Mr J. Hull with a metal-detector. He has given it to the Woodspring Museum, Weston-super-Mare, where Jane Evans made it available to RSOT.

¹³ Information from Mr Hull. No tablets were found when the temple was excavated. Of unknown dedication, it was built c. 340 and demolished c 390: A. M. ApSimon, Proceedings of the University of Bristol Spelaeological Society 10, No. 3 (1965), 195–258. These dates and the handwriting of the tablet, whose closest parallel among the Bath curse tablets is RIB *2349, fit together.

¹⁴ A deity is being addressed, perhaps as domina, which would imply a goddess. A thief is being cursed in formulas found in other British curse tablets. Damage makes much of the text uncertain. It is difficult to distinguish A, N, and V. There are at least two copying errors, which were corrected, and probably others (see Commentary), but the state of the text makes this difficult to determine.

Commentary

1. This seems to be the original top line, but the centre was lost when a corner broke away from the folded tablet, and the other letters are all damaged. Analogy with other curse tablets would suggest it contained an address to a deity (described as sanctus?), followed by dono tibi or similar.

2. caricula: damage to the end of 1 makes it impossible to know whether this is the end of a word or complete in itself. That it is a neuter plural is implied by the succeeding quae and by the restoration of ilia in 6; that it is the object(s) stolen, not the name of the petitioner, for instance, is implied by its position in the text: cf. Britannia xii, (1981), 371, No. 6 (text restored), deae Suli donavi argentiolos sex quos perdidi, etc. But cariculum and the word of which it is the diminutive seem to be unknown. Carica (a dried fig) or a blunder for caracalla can probably be excluded. W. Meyer-Lübke, Romanisches etymologisches Wörterbuch (3rd ed., 1935), deduces *caraculum (a small stake or pole) and ^*car(r)icare (to load), *carricum (a load), from romance derivatives; cognate with the latter, apparently, are the Celtic-derived words carrus/carrum (a wagon) and carruca (a carriage). Is car(r)iculum the diminutive? And if so, what has been stolen? (Wheelbarrows are said to be a medieval invention).

2–3. si servus si liber and (repeated in 7–8) si baro si mulier are common formulas in the Bath tablets, and can safely be restored here. The ‘Vulgar’ spelling ser(v)us, also found at Bath, is required by the space available. Baro for vir is also ‘Vulgar’.

4. qui must introduce a relative clause with the sense of ‘who has stolen’, but there is no obvious restoration. VT may be ut introducing an indirect command typical of curse tablets, but should perhaps be taken with the preceding A as a blundered termination,…av(i)t, of the verb dependent on qui. EVS suggests [d]eus, but the reading is too uncertain.

4–5. [d]omina: the first letter looks like I and cannot be N (thus precluding any formula involving nomina); the second letter might be V; something resembling T or C has been written above MIN, perhaps to continue it (cf. the end of 7). Thus the restoration of domina is uncertain, if attractive: cf. Britannia xiii (1982), 404, No. 7 (also fourth-century), where Sulis is addressed as tu, domina dea; and RIB 323 (Caerleon), which invokes dom(i)na Nemesis.

5. No obvious restoration.

5–6. facias sic [i]lla (re)dimat sa(n)guin[e s]uo: for the formula (and the restoration), cf. Britannia xii (1981), 378, No. 9, facias ilium sanguine suo illud satisfacere; ibid., xiv (1983), 339, No. 5, sanguine et vitae suae illud redemat, and 340, No. 6, hoc donum non redemat nessisangu(i)nesuo; RIB 323 (emended), non redimat ni(si) vita sanguine sui. British curse tablets sometimes mis-spell sanguine(m), but by a ‘Vulgar’ contraction of the second syllable (cf. Portug. sangue, Sp. sangre, Fr. sang); the omission of the first N seems to be a simple error, like dimat for redimat. The otiose sic occurs, again in an obscure context, in Britannia xiii (1982), 404, No. 7.

7. si baro si mulier (repeated from 3) can be restored with certainty. MV at the end of the line requires LIER, which is hard to find in the traces at the beginning of 8, and seems instead to have been inserted above, at the end of 6 after suo. The first letters of 7 are obscured by earlier(?) scoring and re-writing, but the first word may be et.

8. No obvious restoration. This seems to be the last line of the original, but its length is unknown.

¹⁵ Excavations directed by Susan M. Davies and Peter J. Woodward for the Trust for Wessex Archaeology. Full information and photograph from Susan Woodward. For the site see above p. 417

¹⁶ It is uncertain what, if anything, is missing at the end of the first line. On the milestone from Margam, RIB 2255, Postumus' titles and names are abbreviated to IMP(ERATORI) C(AESARI) M(ARCO) C(ASSIANIO) L(ATINIO) POSTVMO AVG(VSTO), and it seems likely that C(AESARI) at least was included here.

¹⁷ During excavation for Braughing Excavation Committee directed by Dr T. W. Potter (see Britannia iv (1973), 299–300). Mr S. D. Trow made it available to RSOT.

¹⁸ A detailed account will be published in Dr Potter's Final Report. The surface is badly worn and corroded, and most of the first line and a half is lost, RAFVSA can be recognized at the end of 2, perhaps fusa or the end of a personal name, and at the beginning of 4 what seems to be a denarii sign as found in other curse tablets (JRS liii (1963), 121 ff.; Britannia xiii (1982), 403, No. 6; Britannia xv (1984), 339, No. 7). For three other items from the site see below Nos. 51–3.

¹⁹ By Daryl Douglas and the Rev. Ian Stuart-Hargraves. Information from John Boyden who submitted the stone to MWCH for inspection.

²⁰ The title pius felix Augustus was first used by Commodus, but did not become a normal part of imperial titulature until the time of Caracalla, who used it occasionally before, and regularly after, AD 211. It is unusual to have patri patriac written in full, though a near parallel exists in an inscription of Caracalla from Chesterholm (RIB 1705). This inscription is unlikely to refer to that emperor, since there is insufficient space for the appropriate titles between pius felix Aug. and patri patriae (as indicated for example on the Chesterholm inscription cited above).

In 1. 2 the apparent ligature between O and P is presumably a stone-cutter's error. In this line it is just conceivable that PART(H)I║[CO, is to be restored, but if this were intended one would expect a reversed R ligatured to the left side of the upright of the T (as for example twice in RIB 1234), rather than a normal R attached to the right-hand side of the T.

²¹ By Mr A. D. Phillips. The stone remains in situ.

²² The stone is too thick to be part of a tombstone, and too ample in its layout to be part of a centurial stone. An altar is possible, but much less likely than a monumental inscribed slab. EIVS (1.2) might be the end of a personal name (Pompeius, Cocceius, etc.), but the pronoun eius (‘his’, etc.) is more likely. This suggests the third-century formula (with variations) devotus numini maiestatique eius or, less likely, since the legate's name would precede it, legatus eius. Since York was a legionary fortress, and then the capital of Britannia Inferior, one would expect to find the legate's name on a monumental dedication. It is therefore tempting to identify Gaius Octavius with Octavius Sabinus (praenomen not stated), who is attested at Lancaster (RIB 605) as legate of Postumus in 262/6. The Lancaster inscription records the rebuilding of the fort bathhouse and basilica; contemporary rebuilding at the fortress of York is not unlikely.

²³ During excavation for Carlisle Archaeological Unit (see Britannia xiv (1983), 290–2) directed by Ian Caruana. Mr Caruana sent full details of this and the next eight items.

²⁴ Not to be identified with any of the three persons of this name already attested in Roman Britain. For similar graffiti, cf. RIB 1007–15, 2031.

²⁵ Perhaps an unrecorded personal name, not ni(ui)bus contracted. Mr Caruana reads AVIBVS (‘for the birds’, a reference to augury?), pointing out that the diagonal stroke of ‘N’ is unusually thick and changes direction as if V has been superimposed on A. This is possible.

²⁶ This and the next three items seeem to have been freshly cut for the Flavian II structure (in the period 79-c. 105). The numerals were incised after the timbers had been cut to shape but, since this and the next item did not form part of the final structure, the numerals probably did not relate to final assembly.

²⁷ The two strokes do not quite meet, but II is less likely. The coincidence of numeral with the previous item from the same building suggests that numerals were batch-numbers or site-numbers.

²⁸ ‘…at Carlbury on the North of the brook amongst old foundations of houses near Peirsebridge’, according to a note inserted by George Allan of Darlington (for whom see Birley, E., Research on Hadrian's Wall (1961), 21)Google Scholar into his copy of J. Horsley, Britannia Romana (1732). Horsley (p. 486) noted that the houses on the Tofts field, Piercebridge, were called the Bury or Carlebury. Allan copied this note, and others, from notes in another copy of Horsley which had been made by Christopher Hunter of Durham (for whom see Birley, and Rogan, J. in Arch. Ael. 4th ser. xxxii (1954), 116–25).Google Scholar Allan's copy of Horsley is now owned-by Dr G. Webster, who made it available.

²⁹ The transcript (FIG. 7) suggests that most of the L. margin survived, but that the top and bottom and R. edge were lost. It is uncertain how many letters have been lost from the R. side, but since lines 2 and 6 need only one letter to complete a word, and lines 5 and 7 end with a complete word, the chances are that the original ended here; this would indicate a width of line like RIB 1026, a contemporary tombstone from Piercebridge. The smaller fragment evidently conjoined, since nothing has been lost in the middle of PIENTISSIM; but it is uncertain to which line CONIVNX belongs, although 5 seems likely, and it is structurally improbable that a fragment which broke away from one corner of a rectangular slab would itself be a rectangle. The transcript drawing is therefore probably schematic, but trustworthy, since it reproduces ligatures without understanding them, and words which can have made no sense, without trying to rationalise them.

The presence of a vexillation at Piercebridge in 217 drawn from the Sixth legion and both Rhine armies (JRS lvii (1967), 205, No. 16), containing ordinati (i.e. centurions) from Germania Superior (RIB 1022, 1026) is already well attested, but was of course unknown in 1750. Analogy with RIB 1026 suggests that the heading D(is) M(anibus) has been lost, and that EXN… is the name of the deceased. Like Gracilis (RIB 1026) he was a legionary from Upper Germany. Aetate triginta (cf. RIB 183) is a possible restoration, but seems to be unparalleled. Aetate usually occurs in the formula aetate prima (‘in the prime of life’), but aetatis is used to express age at death: see G. N. Olcott, Thesaurus Linguae Latinae Epigraphicae (1904), s.v. If posuerunt is restored in 7, an additional subject to coniunx pientissima must be sought, and it is hard to see where it could fit. Line 8, if vixit has been correctly restored (which makes the restoration of triginta in 3 less likely), will have noted age at death, and perhaps also length of military service.

³⁰ By Mr B. Dodds, son of the owner of Mains Farm, Ebchester, Mr W. Dodds. The stone is now preserved in the museum on the site, where Mr A. H. Reed made it available.

³¹ Information from Carolyn Wingfield of Bedford Museum, who supplied a drawing and submitted the brooch for inspection.

³² During excavation by the Carlisle Archaeological Unit directed by Ian Caruana, who sent full details of this and the next thirteen items.

³³ For another sealing of this unit, see EE ix 1296d (Corbridge).

³⁴ By Ian Caruana with a metal-detector, in spoil from modern sewer renewal.

³⁵ TVD is found on the reverse of sealings of Coh. VI Thracum (Cumb. Westm. xxxvi (1936), 121). Coh. VII Thracum (ibid., 119), Ala II Asturum (EE ix 1296d), and Coh. I Aquitanorum (Britannia vii (1976), 386, No. 36 with corrigendum). Perhaps an abbreviation for t(ut)ud(it) (‘struck’).

³⁶ For other sealings of the Twentieth Legion, see Britannia vii (1976), 387, No. 40 (Leicester), and xiv (1983), 347, No. 39 (Rudchester).

³⁷ During excavation by the Carlisle Archaeological Unit directed by Ian Caruana.

³⁸ The letters are blackened, and there is a double impression of the first line. The suprascript bar in line 2 may be complete, implying that Leg(io) XI is meant, but this is not certain.

³⁹ During excavation for the Carlisle Archaeological Unit directed by Ian Caruana.

⁴⁰ Also found at Brough-by-Bainbridge and Housesteads: see Britannia xi (1980), 408, No. 13, with note ad loc.

⁴¹ During excavation by the Carlisle Archaeological Unit directed by Jan Sewter (Castle Street, 1981), by J. A. Dacre (Crown and Anchor Lane, 1982), and in spoil from modern sewer renewal (Fisher Street, 1983) examined by Alan James and Ian Caruana.

⁴² Cumb. Westm. 1st ser. xv (1899), 483, No. 47.

⁴³ JRS xliv (1954), 109, No. 32 (Rickerby Park).

⁴⁴ Cumb. Westm. n.s. xvii (1917), 85, pl. xviii, 3.

⁴⁵ Boon, G. C., Lalerarium Iscanum (1984), 14, fig. 5.Google Scholar

⁴⁶ During excavation by the Carlisle Archaeological Unit directed by J. A. Dacre (see Britannia xvi (1985), 271). Published by Caruana, Ian in Cumb. Westm. n.s. lxxxv (1985), 63–4.Google Scholar

⁴⁷ Only the top of the first letter survives, but it has the distinctive serif of Wright type 5 (Britannia ix (1978), 380).

⁴⁸ When they were given to Carlisle Museum by Sir Fergus Graham, and recorded as having been found at Netherby (information from Colin Richardson of the Museum, and Ian Caruana). This tends to confirm the provenance of the Leg. VI vexillation building-stone RIB 981, which has been doubted.

⁴⁹ By metal detector, information from Wendy Huddle of the City Museum Sheffield, who sent a slide and cast.

⁵⁰ The third letter could be L. The N. if read correctly, is retrograde. Alternatively the character might be M or vs. If this is a military seal then it might emanate from either Conors VI Raetorum or VI Thracum, which are respectively abbreviated as CVIR and CVIT or CVITR on seals from Brough under Stainmore, , Trans. Cumb. West. xxxvi (1936), 104–25.Google Scholar In these sealings the unit's name does not stand alone, but is followed by other abbreviations, and in one case certainly a personal name. Another possibility will be Cohors VI Nerviorum, stationed at Bainbridge (Yorks.) in the third century. (RIB 722, JRS li (1961), 192, No. 4 and Not. Dig. Occ. xl, 56). However, traces of a fourth letter do not really seem compatible with R, T, or N and the seal could have named an individual C(aius) Vi… cf. the lead sealing from Baldock, which reads C.VIC, for C(aius) Vic (…) on one side, Britannia xvi (1985), 327, No. 23.

⁵¹ Directed by J. R. Collis for Exeter Archaeological Field Unit for the City and University of Exeter. For the site see Britannia iii (1972), 344 and for the object itself Charlesworth, D. in P. Bidwell, T., The Legionary Bath-House and Basilica and Forum at Exeter (1979) 222Google Scholar + fig. 70.1. The piece is discussed with other analagous vessels by G. C. Boon (who first recognised that it was inscribed), in Journal of Glass Studies xxvii (1985), 16–17. For a similar, handleless vessel from Caerwent, see below No. 85.

⁵² With other metal objects by Gary Parkin, Consett, using a metal-detector. Miss L. Allason-Jones sent a drawing.

⁵³ Directed by Ken MacGowan and Mike Stone for the Passmore Edwards Museum. Information on this and the following item from Pamela Greenwood, who submitted the stamp for inspection.

⁵⁴ For the expansion of abbreviations on these stamps, see Wright, R. P., Britannia xv (1985), 193–6.CrossRefGoogle Scholar The retrograde stamps do not appear to be followed by the letters LON.

⁵⁵ By the Colchester Archaeological Trust. Martin Henig, who drew our attention to the inscription, dates the intaglio stylistically to the late first century A.D‥

⁵⁶ Excavations by the Great Chesterford Archaeology Group directed by Mr T. E. Miller, who provided full details and a drawing. There is a nail hole on the edge of one of the breaks.

⁵⁷ In 1. 2 the V overlies and replaces an S which had been cut in error.

⁵⁸ A Mócsy's Nomenclátor gives Tertiolus as the only possibility in 1. 1. Venedus is not precisely matched but is presumably the Latinized equivalent of the Celtic Venextos (where the x represents the Greek Chi) cited by Holder, alt-celtischer Sprachschatz, and found on coins of the continental Parisi.

⁵⁹ Information from Chris Going.

⁶⁰ Either the end of a feminine nomen formed from a cognomen with a stem ending in -NT, as Valentia from Valens, -entis, or a cognomen ending in -NTIANVS. based on such a nomen, as Velentianus, but there are many possibilities in either case.

⁶¹ By R. Phillips, see his account of the work in Clevensis No. 19, 31–5, fig. 6 and plan fig. 2. See also McWhirr, A. and Viner, D., ‘The Production and Distribution of Tiles in Roman Britain with particular reference to the Cirencester region’, Britannia ix (1978). 359–77,Google Scholar sp. fig. 4 for the TCM stamp.

⁶² Directed by C. J. Guy for the Western Archaeological Trust. For the site see Britannia xv (1984), 314. Information and drawing from C. R. Wallace of the Gloucester Post-Excavation Project.

⁶¹ By the committee for Rescue Archaeology in Avon and Gloucestershire, directed by Ann Ellison. For other epigraphic material from the site see above No. 4. and n. 8.

⁶⁴ Excavations for the Department of the Environment directed by T. W. Potter. For the site which lies on the west side of Ermine Street, south of the main settlement on Wickham Hill, Braughing, see Britannia iii (1972), 329. Information on this and the following two items from S. D. Trow. For a curse tablet from the same site see above No. 8.

⁶⁵ For the excavation, directed in 1972 by T. W. Potter, see Britannia iv (1973), 299.

⁶⁶ Cf. JRS xli (1951), 143, No. 14 (Sammico) and Hi (1962), 192, No. 7 (Trenico) and RIB 1543 (Venico).

⁶⁷ Johns, C. M., Potter, T. W., Ant. Journ. lxv (1985), 212–53Google Scholar , Cat. No. C.3 (pp. 313, 316), illustrated fig. 3 and pi. XLVib, with general discussion pp. 328–30. For another ingot from the treasure see the following item and for two spoons, also possibly from it, Nos. 80 and 81 below.

⁶⁸ A full Roman pound weighs 327.45 g. On analysis, the metal was shown to consist of 91.7% silver, 7.4% copper, 0.3% gold, 0.7% lead.

⁶⁹ See Johns and Potter, op. cit., Cat. No. C. i (pp. 212–3), illustrated fig. 2 and pi. xlvia, with general discussion 328–30.

⁷⁰ See note 68 above. On analysis the metal was shown to consist of 98.6% silver, 0.4% copper, 0.7% gold and 0.4% lead.

⁷¹ pus(t)ulatum is abbreviated to PS and PVS on two ingots from Dierstorf near Hanover, for which see the catalogue of silver ingots published by Painter, K. S. in Ant. Journ. lii (1972), 86–91,Google Scholar Nos. 31 and 30. It means literally ‘blistered’ but comes to have the sense ‘purified’ and is found both with argentum explicitly mentioned (Dig. 19.2.31, Suet. Ner. 44) and, as here, only implied (Mart 7.86.7. Hispani…libra pustalati). The name Leo occurs on a stamped ingot from Paspoel, Koninksem, Belgium, Painter op. cit., 84, No. 1.

⁷² By Mr L. Claringbold of Faversham, subsequently acquired by the British Museum. First published by Tatton-Brown, T. W. T. in Kent Archaeological Review, No. 62 (Winter 1980), 40–1Google Scholar and subsequently by Painter, K. S. in Arch. Cant, xcvii (1981), 201–7Google Scholar , together with the unprovenanced Kent find stamped EX OFF │ CVRMISSI (= Britannia ii (1971), 299, No. 59, see below under Corrigenda).

⁷³ See note 68 above.

⁷⁴ Other ingots with the same stamp have been found at Richborough (E.E. ix 1257) and at Balline, Limerick, Co., JRS xxxv (1945), 91Google Scholar , No. 5a. For the name see CIL ii, 21.

⁷⁵ Now in the possession of Dr M. Mitchiner. Catherine Mortimer of the Research Laboratory for Archaeology and the History of Art, University of Oxford, made it available to RSOT.

⁷⁶ For this, the more correct version of the tribal epithet, often written ‘Coritanorum’, see the note by RSOT in Ant. Journ. lxiii (1983), 353–5, cf. Britannia xiv (1983), 349–50.

⁷⁷ Directed by J. A. Danieli for Leicester City Museum. Information on this and the following two items from Richard Pollard who provided rubbings and submitted two of the sherds for inspection.

⁷⁸ Connus/a is not apparently attested but Connius/a, both as a nomen and cognomen, occurs. It is derived by Holder from Celtic Connos, a name which occurs on Gaulish coins, and which would, in its closest Latinized form, become Connus/a.

⁷⁹ Directed by Jean Mellor for Leicester City Museums.

⁸⁰ Directed by Jean Mellor for Leicester City Museums. Foir the excavations in 1965 see JRS lvi (1966), 203.

⁸¹ For the name see CIL xiii 6221 (Worms), Amandus Velugni f(ilius) Devas (i.e. from Chester) and ibid. 3632 (Bastogne, Belgium), Velugnius.

⁸² Directed by S. Riviere for the Department of Urban Archaeology of the Museum of London. For the site see Britannia xvi (1985), 297.

⁸³ Directed by A. Thompson for the Department of Urban Archaeology of the Museum of London. For the site see Britannia vii (1976), 345. Information from Beth Richardson of the Department.

⁸⁴ Information and drawing (by Nick Griffiths) from Christine Jones of the Museum of London.

⁸⁵ For a ring of very similar form from the same context see Henig, M. and Chapman, Hugh, Ant. Journ. lxv (1985), 455–7.Google Scholar The authors, who cite a number of other parallels, date the type to the early third century.

⁸⁶ It is rather unusual to find filius written virtually in full, rather than abbreviated to a single letter, but the alternative: G. Fl. G. (/) Filis, where the last word represents a cognomen, perhaps Felix or Felis (cf. CIL ix 1253 Feles) seems, on the whole, less likely.

⁸⁷ In contractors' spoil by Mr S. Cornelius, in whose possession it remains. He made it available for study by RSOT through Christine Jones of the Museum of London.

⁸⁸ There is no address to a deity nor any sign of the usual formulas. The place-name heading and the numeral (see the transcription and the next note), like the find-spot itself, all suggest that it served some commercial purpose. It does not have a hole punched for attachment, like the leaden labels with notes of quantities etc. published by Solin, H. in Aquileia Nostra 48 (1977), 146–63Google Scholar, which were found at Concordia. A closer parallel may be the leaden tags which were found wrapped round the handles of amphoras recovered from a wreck off the Algerian coast: Lequément, R., ‘Etiquettes de plomb sur les amphores d'Afrique’, MEFRA 87 (1975, 2), 667–80.CrossRefGoogle Scholar Each carried the name of an officina, presumably where the contents were manufactured. These tags, however, are twice the width of the Billingsgate one. (It is hard to say whether or not it was ever rolled.)

⁸⁹ Notes of the transcriptions

There is some corrosion, especially on (b), but the chief obstacle is the multitude of small casting blemishes. The N and succeeding letters of (b)2 were inscribed over a deep scratch (not drawn), which is probably to be linked with the diagonal stroke below N.

(a) i. vico lovio: inscribed in ‘capitals’ (really enlarged cursive) as if a heading. This place-name (‘Jupiter's Village’) seems to be unattested, but is plausible. Vicus Minervius (CIL v 4450 4451 ) and Vicus Herculeus (v 4488) are found in the territory of Brixia, and Vicus Venerius (v 5804) in the territory of Milan. Ad loven is attested in Narbonensis (Ilin. Buráig. 551, u), lovis Pagus in Moesia (ibid., 565, 2), and lovia twice in Pannonia (Itin. Ant. 264, 8: 130. 2). Such a colourless place-name as Vicus Iovius is unlikely to have been unique, nor was it necessarily in Britain.

(a)2. The dotted letters are uncertain. The first R is of ‘capital’ form, unlike the second one and those in (a)3.

(a)3 Possibilities for the first two(?) letters include IVL. NI. AL. The concluding numeral (duo) does not seem to be part of a date. If DARI is the correct reading, the genitive (?) case of the preceding word suggests (duo) is a note of quantity.

⁹⁰ By G. de la Bedoyere, who sent details, together with a rubbing and drawing.

⁹¹ It is tempting to restore this as MAXI[MVS, but the first X seems quite clear.

⁹² Photograph and information from A. K. Gregory of the Norfolk Archaeological Unit.

⁹³ Information, and drawing (by H. Spalding) from Daniel Gurney, who directed work for the Norfolk Archaeological Unit.

⁹⁴ Found by M. De Bootman. Information from A. K. Gregory of the Norfolk Archaeological Unit, who submitted the sherd for inspection.

⁹⁴ By the Upper Nene Archaeological Society directed by R. M. and D. E. Friendship-Taylor. Mr Friendship-Taylor kindly supplied a slide and drawings.

⁹⁶ This is a more elaborate variation of the Langton Down type and is so called from the stamp Nertomarus which (along with others) is found on the spring case.

⁹⁷ i.e. so as to be read with the brooch in position as worn with the spring at the bottom.

⁹⁸ See G. Behrens, in G. Behrens and J. Werner (eds.), Reinecke Festschrift, 1–12 with the supplementary list by Noll, R., ‘Römerzeitliche Fibeln-Inschriften’, Germania xxx (1952), 395–9.Google Scholar

⁹⁹ During excavation for the Stafford Community Programme Agency and Staffordshire County Council directed by J. Symondsand L. Watson. Ian Ferris made this and the next four items available to RSOT and provided details.

¹⁰⁰ Ianuarius is a common cognomen, and AG (for Agricolae, Agrippae?) seems to have been added to distinguish it.

¹⁰¹ The layout of the graffito, and the fact that / seems to cut the tail of S, suggests that the graffito was inscribed in the sequence KIINSONNI, for reasons unknown. K is a rare letter, usually found as the intitial letter of Kal(endae) or Carus and its cognates, but K, not C seems to have been intended by the writer.

¹⁰² During excavation directed by M. Cooper of Birmingham University Field Archaeology Unit and Dr A. S. Esmonde Cleary of Birmingham University Department of Ancient History and Archaeology.

¹⁰³ Hardly to be understood as perii (‘I have perished’). More likely an abbreviated personal name Pere (…). say Peregrinus.

¹⁰⁴ During excavation for the Tyne and Wear Museums Service directed by Messrs. P. T. Bidwell and R. Miket. Mr Bidwell made this and the next two items available.

¹⁰⁵ During excavation for the Manpower Services Commission directed by Mr S. Cracknell (see Britannia xv (1984). 295). Helen Maclagan made it available to RSOT through Dr G. Webster. It was unrolled by Mrs E. Cameron of the Institute of Archaeology. Oxford.

¹⁰⁶ Mr Cracknell informs us that occupation of the site was probably second-century. The object seems to be an ownership tag. but there is no hole punched in it for attachment.

¹⁰⁷ During excavation and observation directed by Messrs. T. G. Manby and L. P. Wenham: see Britannia i (1970). 280. Dr Heywood made it available.

¹⁰⁸ See Johns, C. M. and T. Potter, W.. Ant. Journ lxv (1985). 32–52.Google Scholar This spoon is Cat. No. Ui. p. 344. with pi. 1.11(a). Its twin, certainly from the treasure, is Cat. C.s (p. 316–8) = JRS liii ( 1963), 163, No. 21. The present spoon on analysis proved to be 93.6% silver. 4.9% copper, 0.8% gold and 0.8% lead. For a second spoon, perhaps also from the Canterbury treasure, see the following item, and for two silver ingots certainly from it, see Nos. 54 and 55 above.

¹⁰⁹ See Johns and Potter, op. cit. (note 108). This spoon is Cat. No. U.3, p. 344, with pi. 1.111 b. On analysis it proved to be 96% silver, 2.3% copper, 0.7% gold and 0.7% lead. In the inscription one would expect pie iorpium.

¹¹⁰ By Mr J. W. Elliott, who sent a squeeze and a photograph.

¹¹¹ Not an obvious unit title, but perhaps a personal name (Artorius, Sertorius, etc.). Legionary tiles stamped with an (officer's?) name are known from Britain: Boon, G. C.. Laterarium Iscanum (1984), 22,Google Scholar 35. But stamped tiles from north of Hadrian's Wall are very rare.

¹¹² During excavation for the Scottish Field School of Archaeology directed by Professor J. J. Wilkes. Professor S. S. Frere provided details and made it available.

¹¹³ During excavation for the National Museum of Wales (see Britannia xv (1984), 269) directed by Mr D. Zienkiewicz. He provided RSOT with a photograph and allowed access to the original while it was being conserved. It has not yet been possible to photograph it by infra-red, which may improve the reading, especially of lines 9–10.

¹¹⁴ There is no closely comparable hand among the Vindolanda writing-tablets, but the use of interpunct dates it securely to the first century: see Bowman, A. K. and Thomas, J. D., Vindolanda: the Latin Writing Tablets (1983). esp. 68–9.Google Scholar Writing across the grain is confined to documents (ibid, 40). To this monograph and to R. O. Fink. Roman Military Records on Papyrus (1971) (cited below as RMR), RSOT is much indebted.

¹¹⁵ Apart from formulaic expressions of readiness (quod imperatum fuerit faciemus and ad omnem tesserum parati erimus), there is only one instance of a future tense in RMR, the summary of the day's orders (admissa) at 47 i 16. By analogy, therefore, the present document with its petiluri sunt. [se] referent (or similar), and extent (although two of these readings are uncertain, all three are future tenses), is a summary of orders given to one or more bodies of soldiers. The interlineation of sunt (4a) and the space which seems to follow 8 suggest that petiluri sunt and referent mark sentence-endings. The text is too fragmentary to say whether these represent successive orders to the same body of soldiers, or to different bodies, but the former seems more likely, since personal names (of the officers or NCOs responsible?) occur only once, in 2–3.

¹¹⁶ These points are discussed in the Commentary which follows. The document is the bureaucratic counterpart to the sort of activity in south Wales a generation earlier, when a praefectus castrorum and eight centurions were killed while building forts (extruendis apud Siluros praesidiis, Tacitus, Ann. xii 38). The praefectus castrorum was responsible for timber-working tools (ferramenta quibus materies secatur vel caeditur. Vcgetius, Epit. ii 10), although the specialists responsible were directed by the praefectus labrum (ibid‥ II)

Commentary

1. [ad opin]wnem peten(dam): the restoration is not certain, but is attractive since the phrase similarly abbreviated heads a column of a strength report of the cohort at Dura (RMR 66. b ii 1). For its meaning, sec R. W. Davies. Historia xvi (1967). 115–8. He argues that opinio means the estimates which preceded the payment of stipendium. which were taken to headquarters by a military escort which returned with the cash; the rosters of 219 and 222 (RMR 1 and 2) suggest that this escort, for a milliary cohort, might be about thirty men.

2. Ofillio: this nomen. variously spelt, is quite common in the Italian volumes of CIL. but has not been found in Britain before. The praenomen would have preceded it. and et suggests that there was another name in 3. There is no Flavian consul of this name. The ablative case suggests that -is in 3 is also an ablative (plural) ending. Since two or three named soldiers are far too few to be sent ad opinionem petendam. and the ablative case is difficult, it is a reasonable guess that a party was sent [cum …] Ofillio ft […] in charge. […] is being their rank or position.

4. [im]pensam: the tail of the M survives, and it is hard to see what other word can be restored. (There is no sign of the cross-stroke of X. which excludes expensam: in any case, unless it is a past participle, the word is used in the plural). The interpunet has been added as an afterthought above the final M. Impensa often means “expenditure”, and may have some technical meaning in the context of opinio which has gone unrecorded; but it can also mean ‘building materials’ (see TLL s.v.) and. in view of the references below to materiae, this may be its meaning here. The sense seems to be that the party which has been sent ad opinionem petcndam will go next to fetch impensam.

4a. sum has been inserted between the lines, and is presumably to be taken with petituri.

5. mate-: cf. n. [ma]teriarum. clearly a reference to building timber (maleria). Mutinous legionaries complained of collecting timber and firewood (Tacitus. Ann. i 35. materiae lignorum udgestus). Vegetius refers to the special tools required (Epit. ii 10. ii 25. dolabras secures ascius serras, quibus muteries ac pali dcdolantur clique serrantur). the need for specialists (ii 11) and for training (iii 4). Inscriptions from the early third-century record working parties of Leg. XXII felling trees (CIL xiii 11781) and active in the legionary sawmills (ibid., 6618. 6623. and 40 her. RGK (1959). 179, No. 151).

6–8. in […e]t in praelori[um?]: there are at least two possibilities here, depending on whetherpraetorium is used concretely of ‘the legate's residence’ (at Caerleon) or abstractly of ‘headquarters’. Both senses are found in RMR. concretely in 82. 7 (from the list of naval supplies expended), claui in praetorium jer(ri) p(ondo) (‘nails for the praetorium. iron, so many pounds’), and abstractly in the common phrase ad praetorium. variously abbreviated (see RMR. p. 14 and Index 9 s.v. praetorium): e.g. 47 ii 7. reuersi u(uondam) d(e)p(utaii) adpraet(orium) praesidis cum epistulis (‘Returned, previously detailed to the headquarters of the governor with letters’). If 8 (sec below) is correctly restored as [se] referent, the abstract sense of in praetorium (‘to headquarters’), and incidentally the restoration of [urn] not [o]. should be preferred. The working party will do something (5–7) ‘and will return to headquarters’.

8. [se] referent: reading and restoration have been suggested by Dr J. D. Thomas. The first surviving letter seems to be a bold R which has lost part of its cross-stroke, and the third letter a similarly damaged F(cf. the Fof Ofillio). Just below and slightly to the right of NT. visible on the original but not in the photograph, are the capitals NEA they are not in alignment with the rest of the text, and seem to have been deliberately erased.

9–10. The reading is difficult because the writing has faded or been rubbed. Examination or photography by infra-red would be desirable.

9. emans(ii): ‘was absent without leave’, an abbreviation found in RMR 1 xxii 8.

10. extent: the sequence NT · Q is certain, and thus a verbal form is to be expected; the reading offered here fits the rather blurred letters, and assumes that X was formed like A and N with an exaggerated descender. Exit (‘has left’) is frequent in the documents of RMR. balanced by r(edit) or r(eversus): extent (‘will be away’) is thus in keeping with the future tenses of this document.

11. [ma]teriarum: the tail of the first A survives, and this restoration seems certain in view of mate- in 5. The genitive implies some less-literary equivalent of Tacitus' materiae adgestus (Ann. i. 35). e.g. cura. or perhaps prepositional causa.

12. Enough space survives at the bottom to guarantee that II is the last complete line, but the verb required by qui probably belongs here.

¹¹⁷ Directed by R. J. Brewer for the National Museum of Wales.

¹¹⁸ See Boon, G. C.. Journal of Glass Studies xxvii (1985). 11–17Google Scholar ; figs. 1–6.

¹¹⁹ By the Glamorgan-Gwent Archaeological Trust, directed by A. G. Marshall. For the excavation in 1984 see Britannia xvi (1985), 256. Information and rubbing from D. R. Evans of the Trust.

¹²⁰ Composed by the author of the Augustan History Life of Trajan, Professor A. R. Birley. Information from Dr A. Tyler. The new inscriptions supply the name of the missing praepositus of RIB 576: it is unusual to find a vcxillation commanded by a centurion of the provincial army to which it has been transferred, but perhaps the original commander had died or been posted elsewhere.

¹²¹ See Johns, C., Ant. Journ. lxv (1985), 461–3 with fig. 7.Google Scholar

¹²² Painter, K., Ant. Journ. lii (1972), 84–92,CrossRefGoogle Scholar esp. 85–6, and Arch. Cant, xveii (1981), 201–7. esp. pp. 204–5.

¹²³ Mattingly, H. and Pearce, J. W. E., Antiquity xi (1937), 39–45.CrossRefGoogle Scholar esp. p. 44.

¹²⁴ Elmer, G. in Numismatičar. Belgrade 1935. p. 19.Google Scholar

¹²⁵ By Mr A. H. Reed, who sent photographs and full details. It was still at Little Grcencroft farm in 1902 (photograph), but was removed soon afterwards, apparently intended for Lanchester church (cf. RIB 1074). How-it reached Broomshields Hall is not known. The present owner. Mr Jennings, allowed access.

¹²⁶ During excavation for the Vindolanda Trust directed by Mr R. E. Birley. who sent full details.