² During excavations for the Colchester Excavation Committee directed by Rosalind Dunnett (Mrs B. R. K. Niblett). For the site see JRS lv (1956), 214.Google Scholar Information and copy of a drawing by J. Calloghan from P. J. Drury.

³ The temple of the imperial cult at the provincial capital of Tarraco, was, like the temple of Claudius at Colchester, octastyle as is known from representations on coins, but the columns (155 m in diameter) and the frieze (similar in size to the present one) which are attributed to it do not seem to be of the correct proportions to go together. For a reconstruction of the temple and a photograph of the frieze see Étienne, R., Le Culte Imperial dans la Penninsule Ibérique (Paris (1958) Pls. iv 1–3, and vi 1.Google Scholar

⁴ Found by Mr M. Carlisle. The leaf has been acquired by the British Museum, see Potter, T. W. et al. ‘The Fenlands in Roman Times’, Britannia xii (1981), forthcomingGoogle Scholar. For a full discussion of votive leaves and feathers from Britain and abroad see Toynbee, J. M. C. ‘A Londinium votive leaf or feather and its fellows’ in eds. Bird, J., Chapman, H. and Clark, J., Collectanea Londiniensia: Studies presented to Ralph Merrifield (London, 1978), 129–146.Google Scholar

⁵ The omission of letters is paralleled by the inscription on one of the Barkway leaves RIB 220, which reads NV(MINI) V(O)LC(AN)O below a representation of the smith god.

⁶ By Mr Geoffrey Eyre of Mill Farm, Brough, who reported the find to Mr R. W. P. Cockerton, FSA. We are grateful to Mr Cockerton for providing full details and to the Director of the Sheffield City Museums, Mr J. Bartlett for supplying a squeeze and photographs. The altar, which remains the property of Mr Eyre is currently on loan to the Weston Park Museum, Sheffield.

⁷ Lines 7–9 read from the squeeze by R.P.W. For line I we have adopted the reading and restoration of Professor J. J. Wilkes who suggests HERCVLI[AV]G. Hercules Augustus has not been previously attested from Britain but is found in inscriptions from other provinces. 1. 2: the phrase occurs in CIL xi 1545 = ILS 3084 a dedication to the CaPltoline Triad set up to record the restoraton of a municipal capltolium and in CIL iii 789 set up to Fortuna on the restoration of a bath building; 1. 3: the two letters comprising this line seem to be a later insertion; 1. 3–4: though the restoration and completion of bath buildings are sometimes recorded on altars set up in them these are invariably dedicated to Fortuna (e.g. RIB 730 and 1212), and it seems impossible to read Deae Fortunae in 1. I and a word for shrine or temple should therefore be required as on RIB 1396 and Britannia vii (1976), 378–9Google Scholar, No. 2, both recording the restoration of temples. It is tempting to suggest A SO[LO] at the end of 1. 4; 1. 5: this line will have contained the dedicant's praenomen, cognomen and, perhaps, filiation. 1. 7: praef(ectus) perhaps of Cohors I Aquitanorum known from RIB 283 to have been at Brough under the governor Julius Verus (A.D. 155–8) but by the third century at Brancaster (Britannia vi (1975), 288, No. 25Google Scholar and ibid, x (1979), 354, No. 44). For the unit in general see Birley, E., Derbyshire Archaeological Journal xcviii (1978), 1980, 59–60.Google Scholar

⁸ Found by Mr H. S. C. Hodgeson by whom it has been donated to the National Museum of Wales. For other tombstones from the vicinity see RIB 359, 363, 371–5 and 377 and Britannia viii (1977), 429Google Scholar, Nos. 14 and 15. All will have come originally from one of the cemeteries attached to the legionary fortress at Caerleon. Full information from R. J. Brewer of the National Museum of Wales who has discussed the inscription with us (see Monmouthshire Antiquary forthcoming).

⁹ In 1. I R. J. Brewer reads M and points out that the letter is preceded by a blank space. He suggests DIS] M[ANIBVS.

1. 2: the combination of letters is hard to explain unless they form the end of the deceased's nomen in the genitive case followed by a cognomen beginning with the letter M.

¹⁰ During excavation for the Department of the Environment directed by Mr P. R. Scott, who provided details including notes by Mr W. Dodds. They were examined in the Department of Archaeology, Durham, but will go to the Bowes Museum.

¹¹ Above the D is a vertical split-baluster moulding, and above that an incised border or perhaps the lower half of two letters (1, F, P or T, followed by B, D, E or L). There were other fragments of dressed stone among the chippings.

¹² During excavation for the City of Newcastle directed by Miss B. Harbottle. It is now in the Museum of Antiquities, Newcastle, where Dr D. J. Smith made it available. It will be published in Arch. Aeliana (forthcoming) by Mr C. M. Daniels, with whom we have discussed reading and restoration in detail.

¹³ The dowel will have secured a statue of Julia Domna, or perhaps a lintel, if the stone is the upper half of a door jamb, as R.P.W. suggested for RIB 976, a closely comparable text.

¹⁴ Pro Pletate ac devotione communi, a phrase apparently unique to Britain and the governorship of Marcus, which survives complete in RIB 1202 (see note ad loc.) and can be restored in 928, 976(?), 1235, 1278 and 1705. Marcus' name has been erased in 1202 and 1235, but apparently not in 976 (cf. 977 and 2298, where it survives). Julia Domna shared the dedication of 1235 with Caracalla (cf. 590, different governor), but not of 1202, 1278, 1705 (cf. 2298), and received 68 (different formula and undated) and 976 in her own right, like this new inscription. In the numerous dedications to her from all over the EmPlre she is regularly entitled Iulia Augusta and, with minor variations, mater castrorum et senatus et patriae.

¹⁵ The cohort is already attested on Hadrian's Wall by RIB 1524, an altar dedicated in its name by one Aurelius Campester at Coventina's Well, which may not prove that it was the garrison of Carrawburgh. It was already in Britain by A.D. 103 (CIL xvi 48) and shortly afterwards gained the titles by which its name can be restored here, presumably during fighting in the north (CIL xvi 69 with E. Birley, Roman Britain and the Roman Army (1961), 24). It also took part in the advance into Scotland under Antoninus Plus (RIB 2313 with Britannia iv (1973), 336–7).Google Scholar By the time of the Notitia Dignitatum it had ceased to be the garrison of Newcastle and is not recorded elsewhere.

¹⁶ The original width of the inscription (about twice what survives) can be established from the restoration of lines 8–11 in the light of the unit title (see previous note) and the similarity of formula shared with RIB 1202. Since variations of abbreviation and ligature allow considerable latitude (see RIB 590 for example), the proposed restoration of lines 2-6 is not certain in every detail. Of these, lines 2–3 present the greatest problem. Enough survives of the o in line 2 to preclude D, but to its left is what appears to be a ligatured N. Mr Daniels, to whom we are indebted here and elsewhere, compares RIB 976 (now lost), where Camden read an isolated M in line 2 after luliae Au. in line I, and Huebner conjectured n(ostrae); Mr Daniels suggests NO[STRAE] (‘Our Empress’), but we have not found any parallel. Matri certainly followed, whether in line 2 or 3, and a simplified form of Caracalla's imperial titles: imperatoris Caesaris or domini nostri, variously abbreviated and ligatured, are the most likely; the compression of the surviving portion of Aureli Antonini in line 4 suggests the missing letters were also ligatured, but even so, there does not seem to have been room for Aug(usti) before ac. Line 5: Julia Domna is sometimes mater castrorum alone, but in inscriptions after her husband's death the castrorumj senatus/patriae formula is usual; thus CASTRORUM desPlte the boldness of its initial letters must have been abbreviated, to make room for an abbreviated senatus (cf. AE 1937, 239, mater Aug. et castr. et sen. et pair.) probably with the ligatures TR and EN. Line 7: there seems to have been no room for communi after devotione, unless of course both words were drastically abbreviated, which is perhaps unlikely in what was apparently a novel formula; it may be noted that communi is actually attested only in RIB 1202 and 1705, and must be restored elsewhere. RIB 976 at least seems to have diverged from the ‘standard’ formula.

¹⁷ Found during excavations by the Manshead Archaeological Society. Full information and drawings from the director Mr C. L. Matthews and Elizabeth North. Report Beds. Arch. J. forthcoming. We would like to thank a number of scholars including Dr Jaczynowska, Professor J. J. Wilkes and Dennis Saddington for discussing the text with us.

¹⁸ Reading VE(R.) REGILLINVS, the suggestion of Professor J. J. Wilkes, rather than VEREGILLINVS, neither Veregillinus or Regillinus are attested as personal names but the latter is a possible formation from the cognomen Regillus.

¹⁹ For a general account of dendrophori see the article in PW and for a more recent account Ladage, D., Stadtische Priester und Kultamter im Lateinischen Westen des Imperium Romanum zur Kaiserzeit (Köln, 1971) 126–131Google Scholar, Die Dendrophoren im Kult der Magna Mater; Laguerre, G., Inscriptions Antiques de Nice-Cimiez, (Paris, 1975), No. 71 with discussion pp. 115–8Google Scholar, see also Fishwick, D., ‘The Cannophori and the March Festival of Magna Mater’, TAPA 97 (1966), 193–202.Google Scholar

²⁰ The procession may be represented on a relief from Bordeaux, see Esperandieu, E., Recueil général des bas-reliefs, statues et busies de la Gaule romaine (Paris, 1908)Google Scholar, No. 1096, also illustrated in Brogan, O., Roman Gaul (London, 1953), Fig. 42a.Google Scholar

²¹ Amsoldingen: CIL xiii 5153, Heddernheim: Germania xl (1962) 73Google Scholarff. = AE 1962 No. 232 + Germania li (1973), 173 ff.Google Scholar

²² Kastel: CIL xiii 7317 = ILS 7095 and CIL xiii 7281 = ILS 5 3805 add.; Cologne: CIL xiii 8184. See Fishwick, D., ‘Hastiferi’, JRS lvii (1967), 142–60.Google Scholar

²³ Wheeler, R. E. M., Verulamium: A Belgic and Two Roman Cities (Oxford, 1936), 113–20, Pl xxxiv.Google ScholarE., and Harris, J. R., The Oriental Cults in Roman Britain (Leiden, 1965), 97Google Scholar, express doubts on the identification of the temple.

²⁴ Cf. especially CIL v 81, 5296, ix 939, 1463, x 445, 8100, 8107 f.

²⁵ Excavations directed by Davi d Allen for Buckinghamshire County Council. For the site see above p. 373.

²⁶ Excavations by the Milton Keynes Archaeological Unit directed by Mr R. J. Zeepvat who provided full details.

²⁷ In 1. 1 Felicianus is statistically the most likely restoration of the first name, since, according to Kajanto, The Latin Cognomina, it occurs 140 times in CIL and is much more common than other possibilities indicated by I. Marriott's reverse index of personal names, all of which occur less than 30 times. Note, however, that Senicianus (20 times in CIL) was particularly frequent in Celtic provinces.

In I. 2 Marcellianus, Vitellianus and Gemellianus (15, 13 and 7 times respectively in CIL) are all possible. For similar exhortations expressed on tiles see EE vii 1139 (=RCHM York vol i, p. 114, No. 24 Pol(l)io col(l)egio felic(i)ter and CIL vii 1256, a list of names ending with the word feliciter.

²⁸ Foun by T. E. Ward of the excavation section of the Grosvenor Museum. Information and rubbings from P. Carrington and Margaret Bulmer of the Museum.

²⁹ Fo r a list and distribution map of names with the stem Alt.-, in the Germanies and Belgica, see Alfoldy, G., ‘Epigraphisches aus dem Rheinland II’ in EPigraphische Studien 4 (1967), 10–16.Google Scholar Names with this stem occur most frequently in the Cologne-Bonn region.

³⁰ Perhaps of Dressel form 20.

³¹ Or if inverted IISI[….

³² During excavation for Carlisle City Council and the Department of the Environment directed by Mr M. R. McCarthy, who sent a drawing and other details. For the site see Britannia x (1979), 281.Google Scholar We would like to thank Dr Margaret Roxan and Ian Caruana for discussing the stamp with us.

³³ The stamp is already attested at Carlisle, as well as at Housesteads and Brough-by-Bainbridge (Britannia vii (1976), 390, no. 58Google Scholar, with n. 74), but its expansion remains conjectural. If a personal name, then cf. CIL xiii 10010 (samian stamps), 1025, IM; 1026, IMPETRATVS (etc); 1027 IMPRITO. If ‘imperial’, then cf. Inscriptiones Daciae Romanae iii. 1, no. 282 (Szeged), a tile stamped IMP; and J. Szildgyi, Inscriptiones Tegularum Pannonicarum (1933), 103, no. 76 (Sopianae), also a tile stamped IMP. Other Pannonian tile stamps are explicitly ‘imperial’ (ibid., nos. 75b, 77, IMP N; no. 80, IMP ANTO). Tha t IMP by itself could mean ‘imperial’ in Roma n Britain is suggested by an intaglio from Colchester, which labels two imperial(?) busts with IMP: M. Henig, A Corpus of Roman Engraved Gemstones from British Sites (1978), no. 785.

³⁴ During excavation for Carlisle City Council and the Department of the Environment directed by M r M. R. McCarthy, who sent rubbings and other details. For the site see Britannia x (1979), 281.Google Scholar

³⁵ For the name Tarinus cf. CIL xiii 10001, 438 (Lyons), TARIN F stamped on a lamp.

³⁶ During excavation for Carlisle City Council and the Department of the Environment directed by Mr M. R. McCarthy, who made this and the next item available.

³⁷ During excavation for the Bowes Museum and Durham County Council directed by Mr I. M. Ferris, who made it and the next six items available, and by Mr R. F. J. Jones.

³⁸ For a similar sealing from York, not the identical die, see JRS lv (1955), 146Google Scholar, no. n = CW² liv (1954), 102–3Google Scholar, fig. 1; with photograph, R.C.H.M., Eburacum (1962), 133, inscr. no. 144 (Pl. 65).

³⁹ The die seems to be Wright 44 (Britannia vii (1976), 228, fig. 3)Google Scholar, already recorded from Ebchester and elsewhere. RIB 1038 attests building work by the Sixth Legion at Binchester.

⁴⁰ For the stratigraphical description of this and succeeding items, see Britannia ix (1978), 425, and x (1979), 284.Google Scholar

⁴¹ Since the two letters touch, they could be read other-way up as N with a tilted first stroke. Numerals(?) like this seem to have marked quarry-batches or to have served as control-marks of some kind: see Britannia vii (1976), 381, no. 8, and 389, no. 49Google Scholar; and the Newcastle inscription published above.

⁴² Like the previous item, this could be read as N. A broken edge occurs immediately to the right (thus…]iv), but presumably the same numeral(?) was intended as for the previous item.

⁴³ The reading…]AVXL would also be possible, since the cross-bar of the T cuts the upright two-thirds of the way up, and the final 1 has a short diagonal second stroke, but the genitive of a name in -aut(i)us (Cautus, Tautius, etc.) seems more likely.

⁴⁴ Information and rubbings from P. J. Drury who directed excavations for the Chelmsford Excavation Committee.

⁴⁵ Found during excavations directed by Carl Crossan for the Colchester Archaeological Trust. Information on this an d Nos. 26-28 and 30-32 from Nina Crummy.

⁴⁶ Excavations directed for the Colchester Archaeological Trust by Phillip Crummy.

⁴⁷ Excavations for the Colchester Excavation Committee directed by Max Hebditch.

⁴⁸ Excavations directed by Howard Brooks and Philip Crummy for the Colchester Archaeological Trust.

⁴⁹ Or possibly…]iv LEN[…

⁵⁰ Perhaps the nam e Ursicinus/a (which occurs 14 times in CIL as opposed to Ursico, twice, an d Ursicius once).

⁵¹ Found during excavations conducted for Colchester Corporation and the Society of Antiquaries directed by Dr M. A. Cotton; see M. R. Hull Roman Colchester (1958), 180-9 and for the sherd itself p. 186 No. 4 with fig. 95 No. 4 (loop of rho omitted) where analogies to the form of the vessel of late first to second century date are cited. The form in fact persists into the late Roman period and need not be regarded as residual when found in late Roman contexts so that the interpretation of the graffito as a Chi Rho as proposed by P. J. Drury is acceptable. The angles of the arms in relation to each other suggest that the symbol may originally have taken the form of a simple cross with a loop at the end of one of the arms (‘monogrammatic cross’), a third straight stroke being subsequently added to make a Chi Rho of conventional form. The ‘monogrammatic cross’ is found as a graffito on a red colour-coated bowl from Richborough, Kent, see the note with discussion by Greene, K. in Britannia v (1974), 393–5CrossRefGoogle Scholar ( = Britannia viii (1977), 442 No. 92)Google Scholar and for a further example, from Gatcombe, also cut as a graffito on a potsherd ibid., p. 444 No. 100.

⁵² Found by Mr M. Maillard in whose possession it remains, reported with details by David Viner.

⁵³ The second letter could be an R and the third a Q.

⁵⁴ Found during excavations by the Kingscote Archaeological Association directed by E. J. Swain. See the full discussion of the object by Dr Henig, M., Ant. J. Iviii (1978), 370–1 + plsGoogle Scholar. LXXIV and LXXVA, and for an up to date plan of the building, the bulletin on excavations 1975-9 published by the Kingscote Archaeological Association. Dr Henig prefers the identification with Fausta.

⁵⁵ For Roman weights and their metric equivalents see Hultsche, F., Griechische imd romische Metrologie, 2nd. edn. (Berlin, 1882)Google Scholar, Table xiii. A libra is 32745 gm and so 3 librae 982·35 gm; an uncia is 27·288 gm so that if the interpretation given above is correct, the total weight indicated would be 1009-638. Dr Henig points out that the accuracy of any steelyard depends on the calibration of the bar, so that it is not necessary for the counterpoise to represent an exact unit.

⁵⁶ Information from D r An n Ellison who directed excavations for the Committee for Rescue Archaeology in Avon, Gloucestershire and Somerset (CRAAGS) and provided details and a drawing. Fo r the excavations in 1977 see CRAAG S Occasional Papers No. 3 and Britannia ix (1978), 457.Google Scholar

⁵⁷ The first line presumably began with the name of the deity, perhaps Mercury, to whom the object had been given followed by the names of one of the dedicators, probably FL]ORV[S. The second line will have contained the name of a second dedicator followed by a word describing the object presented, perhaps SACR]VM FECERVNT ‘made the offering’ (cf. RIB 151), though other expansions are possible, e.g. VOT]VM or DON]VM.

⁵⁸ The Post Office site was among those in Baldock excavated for the Department of the Environment by Ian Stead. Information on this and the following six items from Jennifer Foster.

⁵⁹ Only part of the bottom right of the first letter survives and it could theoretically be s while the last letter could be an A but is more probably part of an N. The name given either in the nominative or genitive was probably Belleninus/a, formed from the attested nomen Bellenius (cf. CIL viii 4695: Madaura, Numidia Proconsularis). See also A. Holder, Alt-celtischer Sprachschatz s. v. Belenius for cognates.

⁶⁰ Either a personal nam e Donatus, or cognates, or donavit, ‘dedicated’.

⁶¹ Excavations for the Welwyn Archaeological Society directed by A. G. Roo k who submitted the sherd for inspection.

⁶² Melitine occurs 23 times in CIL vi. For a Greek name, Eutyches, already attested at Welwyn by a graffito, see Britannia x (1979), 349, No. 26.Google Scholar

⁶³ Excavations for the Canterbury Archaeological Trust directed by T. Tatton-Brown with site directors Kevin and Marion Blockley. Information and drawing from Gill Hulse and T. Tatton-Brown. The brooch is tentatively dated by D. J. Mackreth to the second half of the second century.

⁶⁴ For this inscription on fibulae see CIL xiii 6734 an d 10027. 146, 148 an d 169; also Saalburg Jahrbuch xxix (1972), 29Google Scholar (three examples, two illustrated abb. 1, 6 an d 7). We owe this reference to Mr Mackreth.

⁶⁵ Excavations for the Canterbury Archaeological Trust directed by T. Tatton-Brown with site directors Kevin and Marion Blockley. Information on this and the following five items from Ti m Tatton-Brown and Nigel Macpherson-Grant. For the location of the site see above p. 400 f., FIG. 13, opp. p. 375.

⁶⁶ As Victorinus (769 examples in CIL), Severinus (275 times), Marinus (255) and Peregrinus (204); and other names are also possible.

⁶⁷ The combination of letters occurs in a number of names, all rare, and the name in this instance cannot therefore be restored.

⁶⁸ Of the final letter part only of an upright stroke survives. If it were at then the reading would be…]VIROT[…as on e word.

⁶⁹ Excavations directed by T. Tatton-Brown: site director, John Driver.

⁷⁰ Information from Mr J. Bradshaw. For five earlier inscribed sealings from the same site of which three are virtually identical to the present example, see Britannia x (1979), 350–3, No. 33.Google Scholar

⁷¹ Excavations directed for the Department of Urban Archaeology of the Museum of London by D. Perring. Information from Beth Richardson.

⁷² For the expansion m(anu) cf. the closely analogous text on a tile from Silchester vi K. Oclo manu ice. where the ice is surely a batch total.

⁷³ The average lengths of Romano-British tegulae is 416 mm (information from Gerald Brodribb), which means that there is space for not more than three characters before nonas. This space was probably occuPled by a number only, since the letters AD (for ante diem) were usually left out in dates written on tiles. Since the nones of August fell on the 5th, the date will have been either the (AD) HI, the third day before the nones, (i.e. the 3rd of August) or (AD) IV (i.e. the 2nd), PRI(DIE), 'the day before the nones (i.e. the 4th) is also just possible but on grounds of space perhaps less likely.

The number in the final line, if it represents the total day's output of one man was probably cc]xxv since 220 seems to have been the regular daily output of tile workers at Siscia (cf. CIL iii, 11383 and EE ii p. 434). This is only the second certain date on a tile from Britain, the other being the tile from Silchester cited in n. 72. For a comprehensive account of graffiti on Roman bricks and tiles in Britain see R.S.O.T. in ed. McWhirr, A., Roman Brick and Tile, BAR International Series 68 (1979), 231–251.Google Scholar

⁷⁴ During excavations for the Southwark and Lambeth Archaeological Society directed by Peter Marsden. Information and copy of drawing from Mrs M. A. Michael.

⁷⁵ If the text were inverted the third letter, v, could be interpreted as an L, with diagonal foot. The inscription would then read]VIILVS[.

⁷⁶ Found by Mr C. Marshall in whose possession it remains. Information and copy of drawing from A. K. Gregory. For the site see Norfolk Archaeology xxvi (1937), 123–163.Google Scholar Full publication and discussion by Mr Gregory in Norfolk Archaeology forthcoming. Mr Gregory argues that the belt may be a copy of a military cingulum. The grave of a soldier from Lyon, a victim of the battle of A.D. 197, included a belt of bronze letters, originally thought to have been mounted on a leather belt, which may have spelt VT]ERE FELIX on one side and…A]LAE(?), on the other (Gallia viii (1950), 146–7 with fig.).Google Scholar

⁷⁷ There will have been enough plates to have included the owner's nomen and praenomen (abbreviated to a single letter).

⁷⁸ Found by Mr J. Mayes and given by him to the owner of Woodcock Hall, Fit. Lt. R. A. Brown, who provided information and submitted the ring for inspection.

⁷⁹ By Mr G. H. Bye and presented by him to Ipswich Museum. Information and rubbing from Mr C. J. Balkwill of the Museum.

⁸⁰ Ian Marriott's index of personal names shows that this is either the end of the relatively rare Postumina (10 examples of Postuminus/a in CIL) or the end of a more common name such as Septimina, Decimina or Maximina, all of which may be written with a u instead of 1 in the penultimate syllable of the stem.

⁸¹ During excavation for Doncaster Corporation directed by Mr M. J. Dolby. Dr P. C. Buckland gave full details and made the sherd available. It is now in Doncaster Museum.

⁸² There would be just space enough before the broken edge for part of T after the R, but not of c; since, however, there is no trace of any letter, a cognomen in Marc… (Marcianus, etc.) is more likely than one in Mart… (Martialis, etc.).

⁸³ During excavation for Doncaster Metropolitan Borough Council and the Department of the Environment directed by Mr J. R. Magilton: see Britannia x (1979), 290.Google Scholar Dr P. C. Buckland provided a drawing and other details. The fabric is grey, hard and gritty, and possibly local.

⁸⁴ Only part of the first stroke of the N survives, so that A or M would also be possible, but Latinus (and its cognates) is much more common than the only other likely candidate, Latiaris (etc.).

⁸⁵ During redevelopment, as noted in the (unpublished) minutes of the Doncaster Scientific Society Annual Report, 1905–6. Information from Mr M. J. Dolby sent by Dr P. C. Buckland, who also provided a drawing and other details. The sherd, which is now in Doncaster Museum, is provisionally dated by Mrs K. Hartley to the second century.

⁸⁶ The Celtic name Vindil(l)us an d its derived nome n Vindilius occur mostly in Noricum : see the index to CIL iii, and vi 32521, a soldier of the Urban Cohorts from the provincial capital, Virunum. Examples elsewhere tend to be legionaries whose origin is unstated.

⁸⁷ During excavation for the Department of the Environment directed by Dr D. J. Breeze, who provided photographs and full details including the next note.

⁸⁸ Of the first letter in each line only th e final serif survives, so that R is a possible reading, though unlikely. If [M]artial[is] is restored in line 2, then tw o letters have been lost before and after ASIN in line 1. A conjectural restoration is [Gn(aius)] Asin[us]/[M]artial[is]. Miss A. J. Price, when consulted by Dr Breeze, found Dr D. B. Harden in agreement with her that the bottle is very unusual, if not unique. She compares two moulded bottle bases from Spain: a quadruped from Italica (A. Garcia y Bellido, Colonia Aelia Augusta Ilalica (1960), 162 and fig. 59.1) labelled…]inus, presumably a rebus for Asinus; and from Ampurias (M. Almagro, Las Inscripciones Ampuritanas Griegas, Ibericas y Latinas (1952), 185–6, no. 148), in a circle of which a third is lost, the name Cn[…]us Martialis. But it is far from certain that the two are related, or that Asinus (not Asinius) can be a nomen.

⁸⁹ J. Clarke, The Roman Fort at Cadder (1933), 81, fig. 19.5. The brick is now in the Hunterian Museum, where Dr L. J. F. Kepple brought it to our attention and provided a photograph. We have discussed the reading with him: see next note.

⁹⁰ Part of the T is lost, but since the brick was presumably square, the graffito seems to be otherwise complete. Tarcanis is apparently unattested, but initial Tar- and final -cams both occur in Celtic personal names. Clarke (see previous note) read TIIRCARVS i.e. Tercarus (also unattested, and implausible), but this was because the diagonals of the first A do not meet and he did not read the short vertical (third) stroke which is omitted in his drawing.

⁹¹ During excavation directed for the Gwynedd Archaeological Trust by Mr R. B. White, who sent details including a drawing and photograph. It is now in the National Museum of Wales, where Mr G. C. Boon confirmed our reading. For the site see Britannia ix (1978), 406.Google Scholar

⁹² For the nomen Gellius cf. RIB 1572, 1668 (the centurion Gellius Philippus). For other examples of leather stamped with personal names, see CIL vii 1329 with CW² xxxiv (1934), 86Google Scholar; and for continental parallels, esp. CIL xiii 10034 and Bonner Jahrbucher cliii (1953), 123–4Google Scholar, see W. Groenman-van Waateringe, Romeins lederwerk uit Valkenburg Z.H. (1967), 22–3. R.S.O.T.

⁹³ During excavation for the Department of the Environment directed by Mr P. J. Casey, who made the sherds available, and Mr J. L. Davies.

⁹⁴ The L is separated from the first o by two vertical strokes which were incised on the lip before firing with the same tool that was used to decorate the wall of the pot. It is possible the writer intended to incorporate this Plece of decoration as II (i.e. E), but it seems unlikely, particularly in view of the rarity of names in Leo… compared with those in Lo… He may have intended something like (centuria) Lo(ngini) Ammo(n)ius. For an Ammonius of about this date, cf. RIB 2213. R.S.O.T.