² During excavation by Pre-Construct Archaeology for Wimpey Homes, directed by Alistair Douglas; for the site see Britannia 34 (2003), 347. Jenny Hall made it available at the Museum of London.

³ Only the top-right serif survives of the first letter in line 2, but it does not belong to S, and the sequence FT can be excluded.

⁴ During machine clearance of an area damaged by metal detectorists, before excavation by Surrey County Archaeological Unit. Kathryn Ayres, the Finds/Archives Officer, made it available.

⁵ The script, an Old Roman Cursive (and thus earlier than the fourth century), is too crude to be dated. II for E is unusual, and suggestive of an earlier date, but 4,000 denarii would be a large sum of money before the late third century.

⁶ Uncertain letters have been dotted, and undeciphered traces of text represented as …. The bands of damage are represented as ║. In (i) the E of creo is capital-letter, but elsewhere E is written as II. The scribe wrote Senis and corrected it to Senilis without deleting S. In (ii) the denariis sign consists of only two intersecting lines and possibly part of a third, but this reading is confirmed by the numeral which follows.

⁷ The text is too fragmentary for interpretation, but it may be a ‘curse tablet’ addressed to the (unnamed) local god, with reference to 4,000 denarii. There is no evidence of a connected text, only the two personal names, Senilis Senni, and Aurelius Se[…], and possible trace of others.

⁸ It is now in the Brewhouse Yard Museum, Nottingham Castle (inv. no. 1974.337), where Anne Insker made it available to RSOT, who will publish it with full commentary as ‘A Roman inscribed tablet from Red Hill, Ratcliffe-on-Soar (Notts.)’, in Antiquaries Journal 84 (2004), forthcoming. It was first examined in 1964 by (Sir) Eric Turner, who published the first ‘curse tablet’ from the site in JRS 53 (1963), 122–4. Another tablet is published in Britannia 24 (1993), 310–13, No. 2. For the site see Ebbins, S. and Palfreyman, A., ‘Redhill Iron Age and Romano-British site, Nottinghamshire: a new assessment’, Transactions of the Thornton Society, Nottinghamshire 107 (2004), 17–40.Google Scholar

⁹ The lettering cannot be dated, but the tablet was found just below the surface with twenty-two coins, one third-century, the others all of the period A.D. 367–92. This is not an absolute date, since the tablet may have been residual, but the use of the preposition de with the accusative also suggests a fourth-century date.

¹⁰ For a full commentary see the paper cited above (n. 8). This is the first instance of the verb annotare (better adnotare) in a British ‘curse tablet’, which is used here with the preposition de (exceptionally with the accusative case) to mark the objects of theft. These are duas ocrias, as in the Caistor St Edmund tablet (Britannia 13 (1982), 408, No. 9), which were leather gaiters worn by farm workers and huntsmen to protect their legs (Palladius 1.42); an axe and knife, tools appropriate to woodland management or land clearance; and MACIA for ma(n)ica(m), leather glove(s), as in tablets from Bath (Tab. Sulis 5) and Uley (Britannia 27 (1996), 439, No. 1), which were likewise worn by farm workers and huntsmen (Palladius 1.42 again). In writing ASCIA, SCALPRV and MACIA, the scribe omitted the final −M, as often in sub-standard texts influenced by the spoken language, since there was a marked tendency not to pronounce the final −;m of the accusative case, even in educated speech. The formula ‘whether man or woman’, in all its variants, is frequent in British ‘curse tablets’. It is similarly abbreviated, as S B S M, s(i) b(aro) s(i) m(ulier), in an unpublished Uley tablet. Another well-attested formula is the promise to the god of a proportion of the stolen goods.

¹¹ The letters NN in line 1 are ligatured. There is a space after C in line 5. The two dotted letters are uncertain.

¹² Words have now been separated, and those which can be interpreted are transcribed in lower-case letters. Letters omitted by abbreviation or oversight are supplied in round brackets. The letter for which another was written by mistake is restored in square brackets. The letters C and I in line 3 have been transposed.

¹³ During excavations for Barratt, Chester, directed by Vanessa Clarke of Earthworks Archaeological Services. Will Walker of Earthworks Archaeology and Dan Robinson, Keeper of Archaeology at the Grosvenor Museum, Chester, provided details including a photograph, and discussed the reading with us. For preliminary notices, see Vanessa Clarke and Dan Robinson in The Past Uncovered (Chester City Council, October 2002), 3.

¹⁴ The difference in letter-spacing is probably because the first line contains a formula abbreviated to single letters, the second line a single word. They share the same style of lettering.

¹⁵ In the second line, the initial space shows that this is the beginning of a word. The first and fourth letters, of which only the upper loop survives, are either B, P or R; probably the beginning of the name Petronius (or perhaps Petronianus). The formula VSL(L)M, which usually concludes a dedication, is occasionally followed by a consular date; however, this does not seem to be so here. Perhaps PETR[…] is the name of a dedicant or official person.

¹⁶ During excavation by North Pennines Archaeology funded by Leader Construction, directed by Frank Giecco, who sent a drawing and photographs.

¹⁷ There may be medial points after [I]VLIA and before and after ET and ET, but this is not clear from the photographs; nor do they confirm whether the tail of E survives before RM, or the top of F/E at the end of the line below […]A ET. Germanus is the most likely name (it occurs at Old Penrith, RIB 934), but Firminus remains possible, lulia is found as a cognomen in RIB 1483, a usage which is well attested for Iulius. After she is identified as fconijux, the conjunction el introduces another female name, that of a daughter presumably. A second et introduces a third name, most likely that of a second daughter.

¹⁸ With the next eight items during excavation by Carlisle Archaeology Ltd for Carlisle City Council's Gateway City Millennium Project (Britannia 32 (2001), 337); for inscribed coarseware from this excavation see Britannia 33 (2002), 360–3, Nos 7–11. The vessels were identified by Margaret Ward for Oxford Archaeology (North), which made them available. The graffiti were all made after firing, and will be published by RSOT in the final report with some minor graffiti omitted here, including ‘marks of identification’ and two graffiti deliberately erased.

¹⁹ The ‘cross-bar’ is vertical. To judge by its size, the graffito consisted of only this letter, the initial letter of the owner's name.

²⁰ The first two letters of the owner's name. Possibilities include Aper, Apollinaris, Aprilis.

²¹ There is just enough space before M to make this the first letter. The third letter resembles S, but was made differently: first the angular ‘C’ and then a downstroke. In the fourth letter, the beginning of a diagonal second stroke is just visible three-quarters down the vertical stroke. The Celtic name-element *maglos (‘chieftain’, ‘prince’) is found in many personal names in Gaul and Britain, most of the examples being sub-Roman (e.g. Brigomaglos at Vindolanda, RIB 1722), but for the Roman period in Britain, compare Brigomalla at Bath (Tab. Sulis 30, 1) and Maglorius at Leintwardine (RIB II.8, 2504.20, mis-read as Maslorius). In Gaul compare the derived nomen Maglius (CIL xiii.1701).

²² There is just enough space to the left of the first letter, and a hint of the loop, to make it probably initial P.

²³ The graffito may be complete. Perhaps the initial letter of the owner's name rather than a numeral (‘5’).

²⁴ Compare Britannia 28 (1997), 461, No. 20 (also Carlisle), another samian dish inscribed with X and V.

²⁵ The graffito is complete, but to judge by the previous item (No. 13), V may have been accompanied by X.

²⁶ By a builder, ‘in a pit’. Given by the finder to Mr H.J. Mclusky, who subsequently presented it to Welwyn Hatfield Museum Service, Mill Green, Hatfield. Details from Mr A.G. Rook, who made the sherds available.

²⁷ The first (incomplete) letter is a short vertical stroke, conceivably part of G. But if the bowl was inverted when the graffito was made, it would be possible to read: […]IΛIII[…], possibly […]iani(a)e, ‘(property) of […Jiania’; or (but less likely): […]IΛII[…].

²⁸ Full details, photograph and rubbings from Robert Hopkins, who will publish it in the Journal of Roman Pottery Studies 12 (Papers presented to Kay Hartley, ed. Pamela Irving), forthcoming. Brenda Dickinson has identified the potter as Severus iii of La Graufesenque (Die 7c, A.D. 75–100). The dish was purchased by General Pitt Rivers between 1880 and 1900. The label, of which half survives, records the provenance, and states that it was ‘b(ough)t Webster’. This was probably the Bicester antiquities dealer of that name, according to Mr Hopkins, who notes that the purchase was not entered into the General's private catalogue, which is now in Cambridge University.

²⁹ This name is not attested, but it is a possible nomen derived from a cognomen *Rabirinus, which would itself derive from the nomen Rabirius.

³⁰ During excavations by the Museum of London Archaeological Service directed by John Sygrave for the Moorgate Investment Partnership. Information from Jackie Keily and Dr Angela Wardle of the Museum of London Specialist Services.

³¹ Other possibilities are the genitive of Victorius, or the abbreviated genitive Victori(ni), or a dative, ‘for Victor’.

³² During excavation by Pre-Construct Archaeology for Wimpey Homes, directed by Alistair Douglas; for the site see Britannia 34 (2003), 347. PCA provided a drawing and photograph, and Jenny Hall made the tile available at the Museum of London.

³³ There is no visible word-separation. In line 1, letter 3 is C or T (but compare line 2, letter 1); letters 5 and 8 are either B or D, but they are evidently different and, although the sequence DEMBRA reads more easily, letter 5 looks more like B in line 2; letters 9 and 10 might also be AR, and are evidently repeated at the end of line 3.

³⁴ During excavations for M. Gilbert by Archaeological Project Services directed by James Albone, who provided a drawing and full details. For the site, see Britannia 34 (2003), 333.

³⁵ This is acceptable as a note of capacity for a Dressel 20: see RIB II.6, p. 33.

³⁶ By Mr J.C. Green, and reported to the British Museum. Its current location is uncertain, but it was examined by Paul Holder in April 2000, who sent full details. He will publish it in Roman Military Diplomas V.

³⁷ Surviving letters only, without specifying how many have been lost in each line. Letter-heights are 3 mm (outer face) and 2.5–3 mm (inner face).

³⁸ CIL xvi.69 (now in the British Museum), which is the only other document to locate the ala Agrippiana miniata in Britain. The Aldwincle fragment (inner face) writes tribunicia in full, which only happens in the early part of Hadrian's reign; by A.D. 124 it is abbreviated. The outer face, with its list of units, replicates that of CIL xvi.69 almost exactly. Dr Holder thinks, with the late Dr Margaret Roxan concurring, that there would not have been a similar diploma issued in either A.D. 121 or 123.

³⁹ Like the next item (No. 22), with a metal detector. Lindsay Allason-Jones, Museum of Antiquities, Newcastle upon Tyne, sent photographs of them both. No. 21 (accession number 2002.14) has been donated to the Museum, and No. 22 (accession number 2003.5) is likely to be.

⁴⁰ The reading is uncertain except for the enlarged initial Q, which is apparently not followed by V, so it must be an abbreviated praenomen or nomen. The next letter looks like L, perhaps followed by a very small O, then by LL narrower than the first L. So it is tempting to recognize the governor Q. Lollius Urbicus, who recommissioned High Rochester in c. A.D. 139 (compare RIB 1276), but lead sealings usually name centurions or decurions.

⁴¹ This is the unit's first sealing to be found; it was based at High Rochester in the early Antonine period (RIB 1276), and at Lanchester from at least A.D. 218/22 (Britannia 19 (1988), 492, No. 10). The reverse is apparently not lettered, but is perhaps a comb-like device within an oval border.

⁴² During excavation by the Institute of Archaeology, Oxford University, and the School of Archaeology and Ancient History, Leicester University, directed by E. Sauer. For the site see Britannia 34 (2003), 330–33, and South Midlands Archaeology 33 (2003), 92–105, where the roundel is published at 102–3 with fig. 25.

⁴³ By metal-detector; see British Archaeology 73 (November 2003), 22–7, with excellent photographs; also Current Archaeology 188 (October 2003), 324–5. It is described independently by Sally Worrell in Section II above, p. 326. That the pan is ‘too good to be true’ is argued by Ian Smith in Current Archaeology 190 (February 2004), 190, but without having seen the original; there is a rejoinder by Guy de la Bédoyère (ibid.). Jane Stewart, Finds Liaison Officer, Birmingham Museum and Art Gallery, made it available to RSOT, who has also spoken to the finder, and discussed it with MWCH and Guy de la Bédoyère.

⁴⁴ The letters are not much serifed, but R in DRACONIS and the second (but not the first) R in RIGORE carry a bottom-serif. I in RIGORE has a wide upper serif, and almost resembles ‘T’. A is ‘open’ (without cross-bar); the first two strokes of N (which has a marked forward slope) and of M both resemble this A.

⁴⁵ Logically, perhaps, and by analogy with the Rudge Cup and Amiens Skillet (see below, n. 47), the text should begin with MAIS, but there is a space before MAIS, whereas CAMMOGIANNA crowds against RIGORE, which suggests that DRACONIS was already there when MAIS was cut, but that RIGORE preceded CAMMOGIANNA. CAMMOG is quite generously drawn, whereas IANNA is noticeably lower in height, as if to save space before RIGORE, which it almost touches. There is apparently no separation between RIGORE and VALI, except that VALI is lower in height, something which was forced on the graver by the encroaching band of decoration below; nor did he leave a gap before AELI, or between it and DRACONIS; after that, however, there is the space already mentioned before MAIS, and minimal separation of the place-names which follow, at its slightest between VXELODVNVM and CAMMOGIANNA. The generous spacing given to L in VXELODVNVM, which resembles a separation, is only due to miscalculation.

⁴⁶ By separating all the words (see n. 45), and ‘correcting’ the spelling (see n. 47).

⁴⁷ The pan joins three groups of related vessels which, to judge by their distribution, were made in Britain: see C.N. Moore, ‘An enamelled skillet-handle from Brough-on-Fosse and the distribution of similar vessels’, Britannia 9 (1978), 319–27. Its style of decoration sets it apart from the others, including the Rudge Cup and Amiens Skillet with their schematic rendering of Hadrian's Wall (RIB II.2, 2415.53; A.L.F. Rivet and C. Smith, The Place-Names of Roman Britain (1979), 232–3), but all three vessels, as ‘souvenirs’ apparently of the western end of Hadrian's Wall, carry similar but not identical texts.

The Rudge Cup reads: a Mais Ahallava Uxelod(un)um Camboglan(na)<s> Banna. The Amiens Skillet: Mais Aballava Uxelodunum Cambogla[ni]s Banna (A)esica. Also relevant are the corresponding sequences of Wall forts in the Ravenna Cosmography (107, 28–9; 107, 11) and the Notitia Dignitatum (Occ. 40. 44–9, but reversed and as emended by Hassall in Goodburn, R. and Bartholomew, P. (eds), Aspects of the Notitia Dignitatum (1976), 112–14).Google Scholar Ravenna reads: Maia Avalana Uxelludamo Gabaglanda. The Notitia: [Mais] Congavata Aballaba [Axeloduno] Amboglanna.

There are discrepancies between these five texts, both in sequence and in spelling. In sequence, the pan joins the Notitia in placing Congavata / Coggabala between Mais and Axeioduno / Uxeloduno ‘on the line of the Wall’, and thus reinforces the identification of it as Drumburgh. (For argument to the contrary, see Smith, I.G., ‘Some Roman place-names in Lancashire and Cumbria’, Britannia 28 (1997), 372–83.)CrossRefGoogle Scholar But unlike the Notitia, the pan omits Aballaba, and also diverges from the Rudge Cup and the Amiens Skillet, for which the only fort between Mais and Uxeloduno is Aballava. It also diverges from Ravenna in this respect, and there is no obvious reason for the omission.

Hitherto, the Notitia was the only authority for the place-name CONGAVATA, which Rivet and Smith emend to Concavata (Latin, ‘hollowed’). The manuscript confusion between G and ? would be trivial, but not the anomaly which they also note, ‘[t]hat the name is Latin – wholly exceptional among the forts which make up the defensive system in this region.’ The pan might suggest a different emendation: *Congabata for CONGAVATA, the B/V confusion being just as trivial in manuscript transmission. (The next item in the list, ABALLABA for Aballava, is an example.) *Congabata would be formed from the substantive gabata (‘dish’) with a connective prefix, ‘dish-like’, a description of the ‘bold knoll’ of Drumburgh (C. Daniels (ed.), Bruce, CollingwoodHandbook to the Roman Wall (1978), 250).Google Scholar The word gabata was current in Roman Britain (it occurs in an unpublished Uley tablet), and Holder (in Alt-celtischer Sprachschatz, s.v.) thought it might be Celtic, but there is no direct evidence of this; in Latin it is first used by Martial, but occurs earlier in Greek, which has seemed to be a more likely origin.

The other place-names are not problematic. Mais is apparently a locative or ablative form, and it may be that the pan, like the Rudge Cup, omits the initial preposition a (‘from Maia’) which is retained by the Amiens Skillet. CAMMOGIANNA is only a copying-error, with M repeated by mistake for B (the name-element *cambo- guarantees the spelling), and with I written for L. These fort-names are preceded (or followed) by the sequence rigore val[l]i Aeli Draconis, which must be considered next.

First rigore val(l)i, ‘on the line of the Wall’. Rigor is a technical term frequent in the Roman land-surveyors, who define it as ‘whatever is seen to stretch straight between two points’ (Balbi ad Celsum: expositio et ratio omnium formarum iii.4 (Lang, Corpus Agrimensorum II, p. 98), translated by Campbell, B., The Writings of the Roman Land Surveyors (2000), 208).Google Scholar It is confirmed by epigraphic evidence, for example C1L iii.586 (Macedonia), where a surveyor uses land-marks to define tribal boundaries: ri[g]orem servari usque ad fontem, etc. Compare CIL viii.2728 (Lambaesis), in which a military surveyor describes how two tunnels failed to meet: they had both deviated from the true line, duae ergo partes relicto rigore errabant. There is also a large group of inscriptions (CIL vi.31540–57) (Rome) which define the course of the Tiber recto rigore, variously abbreviated.

rigore val(l)i therefore corresponds to the less-technical term used by the Notitia (Occ. 40.32) for ‘the line of the Wall’, per lineam valli. (The land-surveyor Balbus, already quoted, actually uses the word linea to define rigor.) VALI for valli, with its confusion between L and LL, can easily be paralleled for Britain (see Smith in Temporini, H. and Haase, W. (eds), Aufstieg und Niedergang der römischen Welt 11.19, 918)Google Scholar , where it may also reflect confusion between the Celtic name-element *val- / *valio- (‘strong’), in Luguvalium for example, and the Latin word vallum (‘wall’). Thus in the Antonine Itinerary, ‘Carlisle’ is once transmitted as Luguvallo (467, 2), but twice correctly as Luguvalio (474, 1; 476, 6).

The Notitia shows that Hadrian's Wall was called (the) vallum, a term which is amply confirmed by the Antonine Itinerary (464, 1, a limile, id est a vallo; compare 466, 5, and 474, 1–2) and by two inscriptions from the Wall itself: RIB 2034, dedicated ob res trans vallum prospere gestas, and RIB 1445. Compare RIB 2200 and 2205, where building the Antonine Wall is said to be opus valli.

But should AELI be taken with VA(L)LI, or with DRACONIS? Is it an adjective of the Wall itself, or only the (imperial) nomen gentilicium of Draco? The Latin is ambiguous. The absence of a praenomen (whether P(ubli) or T(iti)) is far from decisive, but it lends support to the idea of taking AELI with VAL(L)I. It would then follow that Hadrian's Wall was literally so-called: vallum Aelium (not vallum Aeli). The obvious analogy is pons Aelius, Hadrian's Bridge at Newcastle; compare his refounding of Jerusalem as Aelia Capitolina. But while cities and forts might be named for emperors, frontier-names do seem to have been geographical. (For example the limes Tripolitanus in the Antonine Itinerary 73, 4 and ILS 8923, and the various African limites named in the Notitia.) However, Hadrian's Wall was almost the first – and certainly the most substantial – of the various frontier ‘walls’, and some linear structures took their names from emperors: roads above all, in a tradition dating from the Republic, for example the via Domitiana, the via Traiano, and a much later ‘frontier road’, the strata Diocleliana; but also aqueducts like the aqua Claudia, and the Rhine canal named after Claudius' father, the fossa Drusiana. It is conceivable, therefore, that Hadrian's Wall was initially called vallum Aelium, but also certain that the name did not maintain itself: witness the references already quoted to (the) vallum unqualified, and note also that the Augustan History, when it needs to distinguish the Antonine Wall from Hadrian's Wall (for which see Hadrian 11.2.), refers only to ‘another wall, made of turf’ (Antoninus Pius 5.4, alio muro caespiticio). If the pan preserves the original name of Hadrian's Wall, therefore, it must also be conjectured that it soon lapsed, presumably when it was abandoned for the Antonine Wall; this would date the pan to c. A.D. 125/40, after the forts were added to Hadrian's Wall, but before it was superseded by the Antonine Wall.

Finally the name, whether it was Draconis or Aeli Draconis. (There is an imperial freedman called Titus Aelius Draco, but the identity is unlikely, see CIL vi.20505.) The cognomen is Greek by etymology (‘snake’). The genitive case is ambiguous; either the name of the manufacturer / craftsman, or that of the client for whom the pan was made. (A bronze statuette, the Foss Dyke Mars (RIB 274), carries the names of both maker and clients.) If these vessels were made as ‘souvenirs’ of the Wall, perhaps they could also be ‘personalized’ with the name of the purchaser; the manufacture of military diplomas, certified copies of the law affecting a named veteran, offers an analogy. If so, the owner's name might have been Aelius Draco, with its implication that he was a recently-discharged auxiliary veteran who wanted a memento of his frontier-station.

⁴⁸ By Ms Sally Atkinson with a metal detector, in whose possession it remains. Information from Faye Minter, Finds Liaison Officer for the Portable Antiquities Scheme for Suffolk, who made it available through Sally Worrell.

⁴⁹ For other such tags or labels, see RIB II.1, 2410, and Britannia 20 (1989), 334, No. 14 (Carlisle); 342, No. 60 (Caerleon).

⁵⁰ (a) Apparently numerals erased by diagonal strokes, (b) Perhaps a commodity ending in -ipas followed by the numeral ‘36’.

⁵¹ Drawn from the original with the help of drawings by Donna Wreathall for Suffolk County Archaeological Service.

⁵² During excavations for Sussex Archaeological Society directed by John Manley and David Rudling who sent details together with a photo and rubbings. For the site see Britannia 34 (2003), 352–3.

⁵³ By farm-workers digging peat, according to a contemporary letter preserved with the tablet, signed ‘M.Ll’. This may be Morgan Lloyd (1820–93) of Cefngellgwm, 2 km WSW of Bodyfuddau, who in 1839 helped survey the parish of Trawsfynydd. The farm dug its peat at SH 732 338–9, which is near the early but undated settlement of Ffridd Bod y Fuddau (SH 731 343), with three huts and, despite the altitude (350 m), ‘an extensive field system, defined by low banks and stone walls covering some 3 hectares, which is almost completely filled with narrow cultivation ridges’ (Peter Crew). The nearest major Roman site is the fort of Tomen-y-Mur (SH 706 386), 5 km to the north-west. The provenance is discussed by RSOT, ‘A Roman Will from North Wales’, Archaeologia Cambrensis 150 (2001, forthcoming).Google Scholar Soon after it was found, the tablet was taken to London for examination, where it was casually preserved for many years and ultimately came by descent to Stafford Ellerman, who has presented it to the National Museum of Wales. Mr Ellerman and Ralph Jackson of the Department of Prehistory and Early Europe at the British Museum made it available, and it has been photographed by Ian Cartwright and Jake Matchett. Caroline Auger, archivist of Shrewsbury School, Peter Crew of Plas Tan y Bwlch, Maentwrog (the Snowdonia National Park Study Centre), Steffan ab Owain of the Gwynedd Archives Service, leuan Thomas of Trawsfynydd, and Lesley Whitelaw, archivist of the Middle Temple, all helped with the question of provenance.

⁵⁴ This is not a letter-by-letter transcription, for which see RSOT (cited in n. 53). Letters whose existence can be deduced are supplied in square brackets, and letters omitted by the scribe in round brackets. Hooked brackets enclose letters he wrote by mistake. Letters read with some confidence, but which cannot be explained, are in capitals.

⁵⁵ There are three sentences, each indicated by extending the first word to the left. The testamentary formulas can be matched in Gaius' Institutes and surviving Roman wills, especially that of Antonius Silvanus (FIRA 111, No. 47). There is a full commentary on the reading and text by RSOT (cited in n. 53). Details of the testator and heir have been lost with the first five lines; it can only be deduced from line 15 (compare 19) that the heir was a woman. Since the testator was a Roman citizen, despite the remoteness of the find-spot, he may have been an auxiliary veteran of the unit at Tomen-y-Mur who settled after discharge at Ffridd Bod y Fuddau. Remarkably, considering how many wills were written, this seems to be the only tablet of a Roman will to have been found outside Egypt.

⁵⁶ By Michael de Bootman, the finder of the first fragment, who has presented it to Norwich Castle Museum with the landowner's consent. Adrian Marsden, Finds Liaison Officer for Norfolk Landscape Archaeology, sent details.

⁵⁷ Since neither face includes the names of witnesses, the fragments belong to Tabella I. It is further likely that face (b) comes from the outer face of Tabella I because it probably had a complete text. Face (a) would therefore come from the inner face of Tabella I. In face (a), line 6, Dr Marsden has suggested that the C of CAMP, which is quite clear on the fragment, may be an orthographic error for T, and that the unit was actually the ala [I PANNONIO]R(VM) TAMPI[ANA]. He points out that on the Malpas diploma (CIL xvi. 48 = RIB ILI, 2401.1, of A.D. 103), this unit is the second one listed and occupies a similar position on the tablet. It would then be the unit of the recipient (see face (b), line 5). The ala Pannoniorum Tampiana is well attested in the army of Britain, notably on the Brigetio diploma of A.D. 122 ( CIL xvi. 69), where the recipient was also a veteran of this unit. This restoration is perhaps supported by RMD 153 (A.D. 114, Pannonia Inferior), where ala I Hispanorum Campagonum occurs (probably) as the first unit, rather than the second as here and on the Malpas diploma. However, face (b), lines 7 and 8, are not readily restorable from the Malpas sequence of units.

⁵⁸ Where it was seen by MWCH in 2003. It is said to have been presented by Admiral Quaglia.