² Excavations for the Chichester Excavation Committee directed by A. G. Down who sent details, rubbings, drawings and a photograph.

³ The ditch lies to the south of St Pancras (Stane Street).

⁴ Compare the pediment of the Wolf and Twins at Corbridge which measures 1·8 m wide and 0·6 m high. Arch. Ael., 4th ser. xxi (1943), 173, 81 × E.Google Scholar

⁵ The use of a right-angle would mean that all three subsidiary pieces of pediment could have been cut from a single slab 0·45 m square.

⁶ For the form of the K see, e.g. RIB 882 and 883. The E, S and T in 1. 2 appear to be additions. They could have been balanced by cutting a leaf stop at the beginning of the line. Five letters are missing at the beginning of 1.2.

⁷ Ihm translates Domesticae as ‘of the homeland’ (RE v 1.1296 article ‘Domesticae’) comparing the description of the Matres as suae (as RIB 654, 2055) and patriae (as RIB 1318). Wright translates ‘of the household’ in RIB 652, 2025 and 2050. The restoration of Matribus here is virtually certain, the only other possibility being LARI]BVS DOMEST(ICIS); but such a dedication is apparently only attested once CIL iii 4160 = ILS 3607 (Savaria).

⁸ CIL xiii 8021 appears to identify the two groups: Matribus sive Matronis Aufaniabus domesticis, AE 31 No. 15 on the other hand distinguishes between them: Aufanis et Matribus Domesticis. For the evidence from Bonn see H. Lehner, ‘Römische Steindenkmäler von der Bonner Münsterkirche’, Bonner Jahrbücher 135 (1930), 1–48 and H. Lehner and W. Bader, ‘Baugeschichtliche Untersuchungen am Bonner Münster’ in ibid. 136/7 (1932), 3–216.

⁹ See Lehner op. cit. pis. VIII, X, XI, XV, XVI, XIX.

¹⁰ See ILS index Vol. iii 2, p. 699, municipal arcarii; ibid., 721, curatores arcae of collegia; and ibid. 726, arcarii attached to great households. For the presence of a collegium fabrorum and of Cogidubnus at Chichester see RIB 91. For an arka from Caerleon see RIB 385 and G. C. Boon Isca (1972), 138 n. 364, where the restoration ex arc[a col(legii)] rather than ex arc[a publ(ica) (RIB) is proposed.

¹¹ Excavations directed by Ann Ellison for the Committee for Rescue Archaeology in Avon, Gloucester and Somerset. For the site see Britannia ix (1978), 457Google Scholar and for an earlier find of a fragment of a defixio, Britannia iv (1973), 324, No. 2Google Scholar. This, and the following two items are three of the best preserved of 168 examples, many fragmentary or completely illegible, found in 1977. They are at present undergoing conservation in the research laboratory of the British Museum by courtesy of whom moulds have been supplied of a preliminary selection so that electrotype copies can be made at the Institute of Archaeology, London. The two editors have worked independently from these but have combined to produce an agreed, but provisional, text. They would like to express their thanks to Ann Ellison for every assistance and to the Trustees of the British Museum for a grant towards the cost of making the electrotypes and to John Olive for preparing them, as well as to Ian Marriott for making available his word and letter-group Index of Curses. This shows that the closest parallels to the language occur almost exclusively on other curses from Britain, specifically from Lydney, RIB 306, Kelvedon, JRS xlviii (1958), 150Google Scholar and Ratcliffe-on-Soar, JRS liii (1963), 123Google Scholar. This conforms with the observation made by A. Audollent Defixionum Tabellae (1904) p. xix, that the language of curses shows strong regional variations.

¹² (a) 5, eraptum recte ereptum.

(a) 8 ff., compare the instruction to the god in No. 3 below, lines 4 ff. and the Lydney curse lines 7 ff. nollis petmittas sanitatem donec perfera(t) usque templum Nodentis.

(b) 2–3, repraesentaverint from repraesentare, a technical term meaning ‘to pay on the nail’ (compare (a) 6 erogat, (a) 8–9 sanitatem (b) 6 devotionem (b) 7 and 8 expostulaverit, all comparatively rare and presumably used as technical terms.

(b) 4–5, rapuerunt, indicative where a subjunctive would have been expected for the verb of a subordinate clause in an indirect command, compare No. 3 (a) 5, circumvenit.

¹³ (a) 1. Commonitorium, late Latin, compare the technical language in No. 2 above (note 12 on (b) 2–3).

(a) 5. circumvenit. For the indicative see the note (n. 12) on rapuerunt (No. 2 (b) 4–5).

(a) 7–8, (b), 1, 4. For the abbreviation of supradictus and suprascriptus compare Ratcliffe-on-Soar side (b) 5 deo ssto and CIL vi index.

(a) 8–9, (b) 7–8. For this clause compare Ratcliffe-on-Soar (a) 8 si mascel sifemina and Kelvedon, 2, si mulrer si mascel.

(b) 1–2, tertiam partem, compare the decima pars offered to the god at Ratcliffe-on-Soar, (b) 5–6, and the demediam partem at Lydney, 4.

(b) 3, 7, exsigat. The same verb is used at Ratcliffe-on-Soar (a) 3 in the sense of ‘hounding’ a thief.

¹⁴ 1 The dedication to Mars–Mercury like the original dedication to Mars Silvanus in No. 3 (a) 2 is unusual. 4 Compare Revue de Philologie xliv (1920), 12Google Scholar, No. 27 (Trier) where a woman is cursed q(u)ia mihi fraude(m) fe(cit). Here, however, the writer may not be thinking of the specific fraud that has been done to him but is using the gnomic perfect.

¹⁵ During excavations for the Lincoln Archaeological Trust, directed by Christina Colyer, M. J. Jones and B. Gilmour. Information from M. J. Jones, who sent a rubbing by B. Gilmour. For the site see above p. 294.

¹⁶ In 1. 1 a single serif only survives; 1. 2: the first letter is probably not an o but could be a B, while of the last letter only a serif survives; 1. 3: the final letter could be an F or R less probably a B; 1. 4: only a serif to the left of the top of the A survives and the letter could also be an F; 1. 6: there are just possibly traces of an up-right stroke along the vertical break at the beginning of the line; 1. 7: the final letter could be a c or G. Of the three fragments which do not join, two are uninscribed while the third carried part of a slightly curved vertical stroke followed by an erasure.

¹⁷ Since the subject of the inscription was an imperial freedman, the office (honos) that he held must have been as a member of a college of seviri augustales, the only appointment in municipalities open to freedmen. On election he would have had to pay a fee (summa honoraria), but could also have provided additional funds for a project as here; see in general R. Duthoy ‘Les * Augustales’, Aufstieg und Niedergang der römischen Welt ii (Principal) 16.2, 1254–1309 and especially 1281, on the significance of the ob honorem formula. This phrase also occurs at the end of RIB 67 (Silchester) where it was presumably followed by the municipal office held. For work on a public building being financed by a magistrate (aedile) either as his election fee or in addition to it, see RIB 707 Brough-on-Humber. For other imperial freedmen in the province see RIB 179 and 643 and Britannia vii (1976), 378, No. 1. Seviri augustales have previously been attested at York, RIB 678, York and Lincoln, JRS xi (1921) 101Google Scholar = AE 1922 No. 116 (set up at Bordeaux by a freedman who was a sevir at both coloniae) and possibly Cirencester, Britannia ii (1971), 289Google Scholar, No. 3.

¹⁸ The numeral and the following letters VIR are divided as here between lines in ILS 6253 and the Bordeaux inscription cited in note 17 above.

¹⁹ It is possible that two or more men were mentioned on the stone since there are a number of cases where building projects of this kind were undertaken jointly.

²⁰ During excavation for the York Archaeological Trust directed by Mr R. A. Hall. Mr A. G. MacGregor sent a squeeze and other details. It will go to the Yorkshire Museum.

²¹ The Castle Yard cemetery lies just to the south-east, RI seems to be the end of a word, perhaps path or matri; RIM would then be part of the dedicator's name (e.g. Primus and its cognates). There are of course other possibilities.

²² Mr P. S. Austen provided a squeeze and photographs and full details. The altar will be published in Arch. Ael. 5th ser., vii (1979), forthcoming, by him and Dr D. J. Breeze, whom we thank for allowing us to read and discuss their paper in advance.Google Scholar

²³ It is already attested on late coins of Hadrian. See Birley, E., ‘The Religion of the Roman Army: 1895–1977’, in Temporini, H. and Haase, W. (edd.), Aufstieg und Niedergang der römischen Welt, iiGoogle Scholar (Principal), 16.2 (1978), 1506–41, at 1513–15. Such dedications were ‘official’: two of the British examples were found beneath the aedes principiorum (RIB 990, 1127), and a third in the principia well (2092).

²⁴ RIB 893–5, 897, 905, cf. 903. Dr Breeze's arguments are conclusive. That Chesters was originally a cavalry fort was conjectured by Birley, E., Research on Hadrian's Wall (1961), 172.Google Scholar

²⁵ By Mr R. E. Birley, who made this and the next item available. They are both in the Vindolanda Museum.

²⁶ For the last ‘Veteres’ altar from Chesterholm, see Britannia viii (1977), 432, No. 22.Google Scholar

²⁷ During consolidation by the Department of the Environment.

²⁸ The first letter is far from certain. The final letter(s) resembles a D with extensions of the vertical downwards and the upper curve leftwards to suggest a ligatured T.

²⁹ During demolition of a field wall by Mr R. E. Birley, who made it available. It is now in the garden of the Vindolanda Museum, Chesterholm.

³⁰ Excavations for the Cambridge Antiquarian Society Archaeological Research Group directed by Joyce Pullinger. Mrs Pullinger provided full information on this and the following five items and submitted the sherds for inspection.

³¹ For the site which is bounded by Castle Street, Mount Pleasant and Shelley Row, see Current Archaeology 61 (= vi No. 2, published April, 1978), 57–60.

³² The word dolium, abbreviated to DOL occurs in painted inscriptions on amphorae (CIL iv 2796, 5573, 5519, 5572 and 5577). However, where the content is clear it specifically refers to the ‘vat number’ from which the contents of the amphorae concerned were decanted, and the use of the word here, by the potter seems inappropriate.

³³ The first letter appears originally to have been cut as a c and then changes to an s. The following two verticals could represent IC, A or R.

³⁴ For the name see CIL xii 1929 (Vienne). coc(c)vs also occurs as a samian stamp.

³⁵ During excavations by Exeter University directed by J. Collis and C. G. Henderson. Information from P. Bidwell who submitted the sherd for inspection.

³⁶ Of the last letter only the bottom of the left-hand stroke survives and it is very tempting to take this as an accidental mark and restore as L(ucii) lul(i)i (H)ipponi[ci, ‘(property of) Lucius Julius Hipponicus’.

³⁷ Excavations by Exeter Museums Archaeological Field Unit directed by P. T. Bidwell and J. Pamment. Mr Bidwell supplied information and submitted the sherd for inspection.

³⁸ Presumably an indication of ownership. If this reading is correct, the personal name AVITA appears to be the only one possible.

³⁹ During field-walking. It is now in the Bowes Museum, where it and the next two items were made available by Mr D. Coggins.

⁴⁰ Perhaps part of the owner's name. For quern-stones from German military sites inscribed with the name of a c(enturia) or con(tubernium), see Jacobi, H., ‘Römische Getreidemühlen’, Saalburg-Jahrbuch iii (1912), 75–95Google Scholar, at p. 85 (with p. 21).

⁴¹ Like the next item, during excavation for the Bowes Museum and Durham County Council directed by Mr I. M. Ferris and Mr R. F. J. Jones. Both items are now in the Bowes Museum. For the site, see p. 284.

⁴² The forms of the a, b, and n may be noted. We are grateful to Dr J. D. Thomas for pointing this out to us, and to Professor E. Birley for his comments on the subject-matter.

⁴³ Iuventianus and Karinius are the only exceptions, but of course Virilis and Vitalis are ambiguous.

⁴⁴ Including variant spellings and cognate forms. Karinius, strictly speaking, is a nomen formed from a cognomen, but this distinction was irrelevant by the fourth century. There are only two Celtic names, Cunovendus (already attested as Cunovindus [Britannia ii (1971), 292Google Scholar, No. 14, now in Newcastle]) and Catugnavus (apparently unattested, but a plausible formation akin to Catugnatus). The high proportion of ‘Roman’ names in this random sample from late Roman Britain, presumably in fact from a frontier unit which was recruited locally, deserves notice.

⁴⁵ The headings are probably not pure numerals, since ‘104’ (etc.) are implausible as totals, c is more likely to be an abbreviation followed by ‘4’ (etc.), perhaps C(ohors), since legionary cohorts were numbered; but the names which follow cannot be those of centurions (centuria understood), since there are too many of them in Column 1 at least. It would make more sense if we supposed that a c was omitted before the v of Column 2, or that a horizontal stroke (which would have completed CI) was omitted, but there is no surface indication that either was ever present. Whatever the expansion of c, the genitive case of almost every name makes it difficult to see this document as some kind of membership-roll, say of soldiers detached for brick-making. (One would expect legionaries, at least in earlier documents, to be identified by cohort and century; the usual brick-maker's tally identifies only one or two workmen and itemizes the work done, like the well-known ‘Primus fecit X’ from Leicester [EE vii 1143]). A mass of names in the genitive case may recall the lists of witnesses, usually seven in number, appended to a Roman legal document like a will or military diploma. Is this brick a memorandum of signatures copied, and subsequently checked, from an original document or documents, for example the receipts there must have been for rations issued to individual soldiers?

⁴⁶ Of the initial letter nothing survives except a horizontal stroke appropriate to c, G, S or T. The complete word was presumably the genitive case of one of the rare personal names in -utto.

⁴⁷ Excavations for the Mucking Excavation Committee directed by Margaret Jones who provided information and a drawing of the vessel.

⁴⁸ Information and rubbings from S. Weller who carried out a watching brief for the Billericay Archaeological and Historical Society. The cemetery was probably associated with the site of the farmstead excavation at Beauchamps Farm (TQ 764936) for which see JRS lviii (1968), 197Google Scholar and lix (1969), 223.

⁴⁹ Excavations for the Committee for Rescue Archaeology in Avon, Gloucestershire and Somerset, directed by Ann Ellison who provided details and a drawing. For the site see Britannia ix (1978).Google Scholar For three curse tablets from Uley see Nos. 2–4 above.

⁵⁰ The full motto, as found on a complete set of fittings, will have read optime maxime, con(serva) numerum omnium militantium, ‘(Jupiter) Best (and) Greatest protect (us) a troop of fighting men all’. No complete set has been found in Britain, but individual attachments or parts of them, have been found at various sites including Silchester, High Rochester, and Aldborough, For a recent discussion with full bibliography see G. C. Boon, Silchester: The Roman Town of Calleva (1974), 66–8, 309, n. 7.

⁵¹ Excavations for the East Hertfordshire Archaeological Unit directed by C. Partridge who provided details and submitted the sherd for inspection.

⁵² For the name CEN(N)ATVS of which CENATA would be the feminine form and from which CENATINVS would be derived see F. Oswald Index of Potters Stamps on Terra Sigillata (1931), 72. The pre-Roman context of this find makes it of considerable interest.

⁵³ Excavations for Welwyn Archaeological Society conducted by A. Rook who provided a photograph of this and the following item.

⁵⁴ Following the c the writer first mistakenly cut an E which he then altered to H.

⁵⁵ If a personal name almost certainly the common nomen AVIDIVS in the dative since the only other possibilities are excessively rare.

⁶⁶ Excavation by the East Hertfordshire Archaeological Unit directed by C. Partridge who provided details and submitted this and the following item for inspection. Allen and Hanbury's factory lies east of the A10 on the western side of the town, some 100 m north of the river Lea.

⁵⁷ For the nomen CENNIVS/A, see CIL ix 5209 (Asculum Picenum). It is classed by Holder, Alt-keltischer Sprachschatz, as Celtic.

⁵⁸ Presumably part of a personal name with the stem -ent in an oblique case (as Clementis), or one derived from such a name (as Clementinus).

⁵⁹ Excavations by the East Hertfordshire Excavation Group by C. Partridge who provided information and supplied rubbings. For the site see Britannia iv (1973), 299Google Scholar, under Puckeridge (ii) a, init.

⁶⁰ Presumably an indication of capacity in which case M could either be for modii (1 modius = 8·754 litres), or, alternatively, for mensura ‘capacity’, the actual unit of measurement, possibly congii, being understood (1 congius = 3·283 litres). For the latter interpretation see Britannia viii (1977), 436Google Scholar, No. 44.

⁶¹ Excavations for Canterbury Archaeological Trust directed by K. and Marion Blockley. Information from the director of the Trust, T. Tatton-Brown who provided details and a copy of a drawing by G. Hulse. For a full discussion of the object citing parallels, see the note by H. Chapman in Antiqu. Journ., forthcoming.

⁶² The cognomen Valenus appears to be unmatched.

⁶³ By Mr J. Bradshaw who provided full details and submitted the objects for inspection. We would like to thank John Kent, Richard Reece and George Duncan for discussing the seals with us.

⁶⁴ Haverfield, F., Arch. Journ. xlvii (1890), 233Google Scholar (= EE vii, 1149). Illustrated by C. Roach-Smith, Collectanea Antiqua vi, 120. For the possible significance of imperial seals (exemption from, or prior payment of import dues) see S. J. de Laet, Portorium (1949), 165. For a recent find of numerous imperial seals of broadly similar type see note 68 below.

⁵⁵ Or less probably RVE, RMF or RME.

⁶⁶ As R1C vii Trier, 520. A.D. 324–37.

⁶⁷ RIC vii Trier, 523, LRBC Pt. 1, No. 71 (pl. 11) A.D. 330–35.

⁶⁸ Identical seals have been found at Izvoarele (Sucidava in Moesia Secunda), see Culica, V. ‘Plumburi comerciale din Cetatea Romana Bizantinâ de la Izvoarele (Dobrogeâ)’, Pontica viii (1975), 215–62Google Scholar (Nos. 49–55 and No. 116). We are most grateful to George Duncan for drawing our attention to this article. Other seals from the site carry the name of Ephesus.

⁶⁹ During excavation for the Manchester University Department of Archaeology and the Greater Manchester Archaeological Group directed by Professor G. D. B. Jones. Mr M. Fitchett sent a photograph and details. For the site, see p. 291.

⁷⁰ Moeller, W. O., The Mithraic Origin and Meanings of the Rotas-Sator Square (1973)CrossRefGoogle Scholar, illustrates examples from Cirencester, Pompeii, Dura-Europos, and Aquincum (Budapest), as well as from the post-Roman period. His speculations are followed by a bibliography of more than two hundred items, to which Fishwick, D., ‘On the origin of the Rotas-Sator square’, Harvard Theological Review 57 (1964), 39–53CrossRefGoogle Scholar, is a valuable guide. Two more examples may be added. Beltz, W., ‘Noch zwei berliner Sator-Amulette’, Archiv für Papyrusforschung 24/25 (1976), 129–34Google Scholar, publishes (with two later parallels) a variant Greek version written on a late fourth-century potsherd from Egypt. It was also found in Portugal in 1970, inscribed on a brick before firing: Etienne, R. (and others), Fouilles de Conimbriga ii (Paris, 1976), 168–70Google Scholar, No. 372 (pl. XXVIII), with a useful discussion.

Ever since Grosser (in Archiv für Religionswissenschaft 24 (1926), 165–9) and other scholars demonstrated that the word-square could be a double anagram of ‘A – PATER NOSTER – o’ in a cross centred on the N, it has been widely accepted as a Christian cryptogram. Scepticism, however, may still be expressed. Many other (less attractive) anagrams are also possible. More important, the circumstances of use do not point to Christians. The word-square occurs twice at Pompeii, that is, in A.D. 79 at latest, when Christians in Italy were a small minority (there is no sure evidence of their presence at Pompeii); and it is doubtful whether ‘Pater Noster’ and ‘α et ω’ were yet in Christian use, indeed, whether they became exclusively Christian until much later. The Conimbriga brick is unstratified, but its lettering is dated to the first century by Etienne. The Aquincum example was inscribed before firing on a stamped tile from the legate's palace, the tile being dated to the early second century, that is, at least a century before Christianity seems to have reached the middle Danube (A. Mócsy, Pannonia and Upper Moesia (1974), 259); moreover, above the word-square itself are the words Roma tibi sub, which is the beginning of an ‘ancient’ palindrome quoted by Sidonius Apollinaris (ep. 9, 14, 4, ‘Roma tibi subito motibus ibit amor’). Thus the tile was surely inscribed by someone like Sidonius and his correspondent who was interested in palindromes, and nothing more. (See J. Szilàgyi, ‘Ein Ziegelstein mit Zauberformel aus dem Palast des Statthalters in Aquincum’, Acta Anliqua Academiae Scientiarum Hungaricae ii (1954), 305–10). The Cirencester example has been seen, perhaps significantly, as the only piece of archaeological evidence for Christianity in Roman Britain before the reign of Constantine (J. M. C. Toynbee, in JBAA 3rd ser. 16 (1953), 2, followed by S. Frere, Britannia: a History of Roman Britain (2nd ed., 1974), 372); its context was inconclusive. The four examples from Dura-Europos (from the first half of the third century) were all scratched or painted on the wall of a room in a pagan temple requisitioned by the military. Although the word-square has been seen as a secret Christian cryptogram, it is nonetheless strange that it never occurs in a Christian context (the Roman catacombs, for example) until Christianity has become the state religion: a Cross accompanies the word-square (in this variant form not a ‘Pater Noster’ anagram) on the late fourth-century ostracon already mentioned. Thereafter it occurs in Christian contexts, but this need only mean that Christians (by now in the majority) were interested in palindromes like everyone else. A further difficulty is the likelihood that ‘ROTAS-OPERA’ is the only five-letter word-square possible in Latin which makes syntactical sense of a sort, at the cost of coining the name (?) Arepo (see H. Polge, ‘La fausse énigme du carré magique’, Revue de l‘Histoire des Religions 175 (1969), 155–63, an idea anticipated by H. M. Last in JRS xliv (1954), 112–4). Thus, as the Aquincum tile would suggest, any magical or religious significance must be secondary. (Fishwick had argued that the ‘PATER NOSTER’ cross must be primary, but he had to admit a ‘disquieting doubt’ (op. cit., 52). On the whole, therefore, it seems possible that the word-square was adopted by Christians before the fourth century, but it is unlikely that they devised it or were its exclusive users. It should not be seen in isolation as evidence of Christianity in second-century Manchester.

⁷¹ During excavation of the late Roman well described in Britannia ix (1978), 453–5Google Scholar. Miss L. Schaaf sent a photograph and details.

⁷² This abbreviation is attested in Gaul and Germany: CIL xiii 10024, 97a, 97b, 158, and esp. xii 5692, 7 (written as VT F). Utere felix is already attested on rings from Roman Britain: EE ix 1334 (= CIL vii 1327) and Britannia ix (1978), 580Google Scholar, No. 62. R.S.O.T.

⁷³ Excavation for the Southwark and Lambeth Archaeological Excavation Committee directed by C. Murray and Laura Schaaf. For the site see Britannia vi (1975), 271Google Scholar; vii (1976), 352. Information on this and the following five items from Joann a Bird who also provided rubbings.

⁷⁴ This tag is found on Rhenish motto beakers (CIL xiii 10018 26(a) and (b)).

⁷⁵ The first letter could be B or R. The only possibility, even if the name were abbreviated, suggested by I. Marriott's index of Personal Names is Sosistraten (nom.), CIL xiii 2591.

⁷⁶ Excavations by the Staines Archaeological Unit for the London and Middlesex Archaeological Society directed by K. Crouch. Information and drawing for this and the following item were supplied by Orpah Farrington. For the site see Britannia viii (1977), 409.Google Scholar

⁷⁷ Excavations by the Central Research Unit of the Department of the Environment directed by J. Hinchcliffe. Information from N. D. Balaam who made the tile available for inspection.

⁷⁸ See Britannia vi (1975), 288, No. 25.Google Scholar

⁷⁹ By R. A. Brown who submitted the sherd for inspection.

⁸⁰ For Celtic personal names with this common suffix see, for example, P. de Schaetzen, Index des terminaisons des marques de potiers gallo-romains sur terra sigillata (1956), 44–5, 27 names.

⁸¹ During excavation for the Department of the Environment directed by Mr J. S. Wacher. Miss J. M. Chisholm made it available. It will go to the Yorkshire Museum.

⁸² Perhaps the name Primitivus. There is a space after the T, but a second line is implied by the top of a diagonal stroke below the T.

⁸³ Excavations for the Department of the Environment and University College Cardiff directed by W. H. Manning. Information on this and the following item from Sabina Thompson, who made the objects available for inspection.

⁸⁴ Like the next item, during excavation for the Bowes Museum and Durham County Council directed by Mr I. M. Ferris and Mr R. F. Jones. Both items are now in the Bowes Museum.

⁸⁵ Information from Mr P. D. C. Brown.

⁸⁶ Information from Dr D. J. Smith.

⁸⁷ It was sent for study by Mr G. M. R. Davies of the Colchester Museum, but will probably go to the Ipswich Museum.

⁸⁸ Harrison, R. M., ‘An Early Byzantine Bronze Stamp’, Arch. Ael. 4th ser. xlviii (1970), 333–7Google Scholar. Christian legends are usual, but personal names in the genitive case also occur, like two examples in the British Museum (O. M. Dalton, Catalogue of Early Christian Antiquities and Objects from the Christian East in the Department of British and Medieval Antiquities and Ethnography of the British Museum (1901), 99, Nos. 487 [Leontiou] and 490 [Domitianou]). G. Galavaris, Bread and the Liturgy: the symbolism of Early Christian and Byzantine Bread Stamps (1970), 161–6, suggests that this type of stamp was used to mark loaves distributed after the funeral of the person named, or at commemorative meals. We are grateful to the British Museum for making its reserve collection available, and to Mr S. Bendall and Mr I. Roper for information about unpublished examples. R.S.O.T.