¹ Inscriptions on STONE (‘Monumental’) have been arranged as in the order followed by R.G. Collingwood and R.P. Wright in The Roman Inscriptions of Britain Vol. i (Oxford, 1965) and (slightly modified) by R.S.O. Tomlin, R.P. Wright and M.W.C. Hassall, in The Roman Inscriptions of Britain Vol. iii (Oxford, 2009), which are henceforth cited respectively as RIB (1–2400) and RIB III (3001–3550). Citation is by item and not page number. Inscriptions on PERSONAL BELONGINGS and the like (instrumentum domesticum) have been arranged alphabetically by site under their counties. For each site they have been ordered as in RIB, pp. xiii–xiv. The items of instrumentum domesticum published in the eight fascicules of RIB II (Gloucester and Stroud, 1990–95), edited by S.S. Frere and R.S.O. Tomlin, are cited by fascicule, by the number of their category (RIB 2401–2505) and by their sub-number within it (e.g. RIB II.2, 2415.53). Non-literate graffiti and graffiti with fewer than three complete letters have generally been excluded. When measurements are quoted, the width precedes the height. Roman inscriptions found in Britain, including writing-tablets, can be found online at <romaninscriptionsofbritain.org>.

² In a medieval rubbish pit during excavation by Pre-Construct Archaeology, and published (sf 44) in the forthcoming final report.

³ Although now abraded, it runs parallel to the setting-out lines and at right-angles to the left edge, both edges being not quite at right-angles to the face but receding slightly. The reverse face is rough and untooled, but probably original, since there is no sign of its having split off a thicker slab; and other Roman inscribed marble slabs found in London (RIB 24, 26, 36) are quite as thin. The marble, probably Italian Carrara, was imported and would have been used sparingly. This is not a free-standing tombstone, but a plaque for insertion into a masonry structure, probably a niche for cremated ashes in a columbarium.

⁴ Notes line-by-line. 1. Only part of the first letter survives, the edge of a vertical stroke with serifs at top and bottom. It is indented, to mark this line as the heading. It might be D M with the name of the deceased carried over into line 2 ending with –ae, but neater than this clumsy division would be the formula DM et memoriae, as in RIB 23 (London). 2. The name would then begin with T, whether an abbreviated praenomen or the nomen. 3. LIA would then be towards the end of the cognomen, for example –lianus, –liaris, –liatus, whether masculine or feminine. 4. There is no letter after A, marking it as an abbreviation for ‘years’, whether accusative a(nnos) after q(ui) u(ixit) or similar, or genitive a(nnorum), in either case followed by the numeral.

⁵ In the grounds of a house in Grange Road, Fenham, Newcastle, c. 0.8 km west of the Roman fort and Vallum crossing. It was donated to the Great North Museum: Hancock, from where Jo Anderson sent details and photographs. It has since been examined by Nick Hodgson and Lindsay Allason-Jones, whose observations are included here.

⁶ The very top of C and the extreme lower right arm of A can still be seen, but to their right, where S would be expected, the surface is smooth and uninscribed. Below this line of letters, except for one or two marks which may only be casual, there is no sign of lettering. It seems therefore that the inscription was never completed. Since it is a thin stone slab like those inserted into turfwork, not a masonry facing-stone, it would seem to have been intended for the Vallum (compare RIB 1361–5, all from the Vallum at Benwell), but was left unfinished when it was realised that Vallum slabs should be inscribed in ‘centurial’ format, not as imperial building-stones or dedications beginning with the Emperor’s name and titles.

⁷ By the previous owner of the Bastle, High Rochester, where it still is. It came from the western wall of the paddock west of the house, a location quite near that of RIB III, 3492, another tombstone fragment, found in a field wall ‘immediately south-west of the fort’. Richard Carlton sent details including photographs.

⁸ The verb [u]ixit would have been followed by the age at death, expressed in years, months and days, the terms abbreviated to their initial letters, and each followed by a numeral. In the bottom line below, D for d(ies) is certain, but is preceded by N, not the expected numeral of m(enses). N does not suggest any formula, but perhaps it was a mistake by the stone-cutter for a numeral such as IV.

⁹ By a metal-detectorist, PAS ref. BERK-A1A5EF. Details from Philip Smither, county Finds Liaison Officer, who sent a photograph of the fragmentary bezel from which Fig. 4 is drawn.

¹⁰ The cutter distinguished Ʌ from M by making the first and third diagonals of M vertical. Visually the first M might be N incomplete, but there would not then be enough room for an[ima] m[ea] as in RIB II.3, 2423.4. There was apparently a space between the second M and the (lost) E, so as to balance AMA above. They are separated by a broad horizontal line. For the legend ama me on a gold ring, see RIB II.3, 2422.2, and compare 18 and 48, which are similar. It also occurs in moulded letters on a copper-alloy brooch (RIB II.3, 2421.1) and incised on a glass bottle (RIB II.2, 2419.73).

¹¹ It has been analysed by portable X-ray fluorescence, showing it to be 70% lead, 20% tin, 10% other metals (information from Mike Fulford). This accords with the composition of the first two sealings ever analysed, said to be c. 70% lead, 30% tin (I.A. Richmond, ‘Roman lead sealings from Brough-under-Stainmore’, TCWAAS n.s. 36 (1936), 104–25, at 122–3). Richmond notes that this is close to that of modern solder, 67% lead, 23% tin, with the advantage of lowering the melting-point and making the alloy harder than pure lead.

¹² With the next two items during excavation by the University of Reading (Britannia 50 (2019), 453–4) directed by Mike Fulford, who made them available (sf 790, 545 and 683 respectively). They are published in his forthcoming final report with six other grey coarseware sherds having graffiti made after firing. Only two are literate, the single letter T (sf 198) and part of two letters (sf 989), perhaps […]NI, the owner’s name in the genitive. The others are non-literate: a rim-sherd (sf 239) with three notches, an elaborate ‘star’ (sf 97), and two plain ‘crosses’ (sf 531 and 990).

¹³ Only the second letter, O, is perfectly legible; but the first is similar in appearance, and can be seen as a closed C; and the last two, V and S, are distorted but still fairly legible. The other four are indistinct but separate from each other, so that something can be guessed from their outlines and the few surface traces. There is a gap at the apex, with no trace of a letter here, but perhaps the remains of a ‘swag’ or wreath. The letters CO identify the bust as that of a Constantinian emperor, but without the usual accompanying imperial titles of D N and/or (P F) AVG. The die-cutter also omitted T, but ended the name with VS, a regular name-ending. But Constans is an easier reading than Constantinus or Constantius, especially since Constans visited Britain in 342/3. One of his sealings is already known from London (RIB II.1, 2411.23), but it is quite different.

¹⁴ The first letter is a short vertical stroke met by two diagonal strokes joining each other at right-angles. Just enough remains of the original surface to show that the letter ended here, and that its vertical did not extend to the height of R. It is therefore F, not K. Capital F is sometimes written like this, as in a tile graffito also from Silchester (RIB II.5, 2491.1). A gap may have followed FR, as if the letters were an abbreviation (compare the tile stamp RIB II.5, 2489.42, TIFR), but it is easier to take them as the beginning of the manufacturer’s name, the rest of it now lost. The few possibilities include Fronto and Frontinus, but since the stamp is otherwise unknown, this cannot be certain. In the left part of the tile, much damaged, are the remains of the tile-maker’s ‘signature’, two concentric arcs made in the surface while it was still soft.

¹⁵ The name-ending […]ta indicates a feminine cognomen, not a nomen (which would typically end in –ia), so the second name is unlikely to be hers as well, an impression reinforced by the difference in script. The next letter is probably P with its loop unfinished, not an incomplete A (unlikely before O) nor a badly made F, since many names begin with Pol–, but Fol– is almost unknown. This name is probably that of another owner, but conceivably was a man’s name in the genitive, defining her in some way; whether her father, husband or even owner, if she were a slave. The last is unlikely, however, as would be [… liber]ta Pol[…] for a freedwoman, since liberta is usually abbreviated to L in formal nomenclature, and would be unlikely in this informal context.

¹⁶ With the next three items during the excavations published as Alfred H. Cocks, ‘A Romano-British Homestead, in the Hambleden Valley, Bucks’, Archaeologia 71 (1921), 141–98. These graffiti were read and drawn by Thomas May at pp. 184–5 with figs 18.1–4, but not included in RIB II. Only the next item (Setomina) is displayed in Discovering Bucks Museum, Aylesbury (acc. no. 1994.57.1753), where it was seen by Scott Vanderbilt. The others are in store, but photographs (Figs 8–10) were sent by Brett Thorn, who confirmed that this item was incised after firing. May says it was ‘scratched before baking’, but this is true only of the underlying zig-zag pattern. May also notes a samian base sherd (Drag. 27, stamped CVCALI[M]) which is ‘scratched’ underneath with R. The initial letter of the owner’s name.

¹⁷ This reading assumes that the graffito was made symmetrically, with three letters in each line. O is rather slight, just two small incisions not conjoining, and only the second diagonal of final Ʌ is well marked, but there is just trace of the first. May drew ‘T’ to the left of the second line, and the final Ʌ as ‘V’ followed by the tail of ‘S’, but the photograph does not support this. His reading was ANT | IOXVVS, which he suggested was a blundered ANTONIVS. The omission of I from ɅNTONIɅ seems more likely.

¹⁸ S is extended downwards to mark it as the initial letter, and II is for E. The reading is certain, but Setomina, by its termination a woman’s name, is not otherwise attested.

¹⁹ The bowl has been reconstructed from its remaining sherds, and the uninscribed surface to left and right of the inscribed sherd in Fig. 10 is modern infill. The first letter is broken by the edge and might be D, suggesting the name [Alexan]drina, but O offers many more possibilities. The name Victorina is common, although others are possible, such as Memorina. Epitaphs of women called Victorina often reduce the genitive/dative name-ending –ae to –e, a frequent ‘Vulgarism’.

²⁰ May reads VIR…, which like his drawing implies that V is the first letter and that R, although damaged, is acceptable. There is no photograph. The name Virilis is very common, here nominative or genitive.

²¹ With the next thirteen items during excavation by Wardell Armstrong Archaeology directed by Frank Giecco, who sent details and photographs. For previous lead sealings from the site, see Britannia 54 (2023), 435, Nos. 13 and 14; 56 (2025), Nos. 12–17.

²² The suprascript bar of the numeral cuts the diagonals of V. On the reverse, despite the odd resemblance of ɅLR to ALP on Nos. 16 and 17 (sf 5736 and 5747 in Fig. 11), the A is ‘open’ (Ʌ) and the R has a tail; in any case, a sealing would hardly name both an ala and a legion. ɅLR is followed by an arc which can reasonably be understood as an incomplete centurial symbol.

²³ Obverse and reverse dies are the same as those of three other sealings in Fig. 11, sf 5666 (the next item, No. 14), sf 3324 and sf 433. The latter two have been published already as Britannia 56 (2025), 501, No. 13; and 521, Add. (g), correcting Britannia 54 (2023), 435, No. 13. There is a fifth instance of both dies, also from Stanwix: Britannia 56 (2025), 500, No. 12.

²⁴ The three initials of a decurion, not an abbreviated cognomen Anc(…). In this and the next item (No. 14) there would have been a crescent above and a ‘star’ below, as can be seen in sf 3324 and 433, both in Fig. 11, but they did not register or have worn away.

²⁵ The letters A, L and E are ligatured together just like those in the other sealings, but P and E have both registered badly. There is also little trace of the ‘star’ above and below.

²⁶ As already noted, the initials of the decurion.

²⁷ Above and below the obverse legend is a ‘branch’-like motif. This obverse die is not the same as that of the other ala Petriana sealings in Fig. 11. To expand the reverse, there are many possible cognomina in M–, but the decurion might well be Flavius Marcellinus, [F]lavius Marcellinus dec(urio), who dedicated an altar at Milecastle 49 (RIB 1870).

²⁸ Both dies are the same as those of the next item (No. 17, sf 5747), so the damaged reverse can be fully restored; it is shared by sf 817 (Britannia 54 (2023), 435, No. 14). On the obverse, the second A is separated from ALP by a space, and may be understood as the title A(ugusta). This is not included in the other sealings of the ala, but is attested by RIB 957 (Carlisle) which records its full titles: this slab is dedicated by one of its prefects, prae(fectus) alae Augustae Petrianae torq(uatae) m(illiariae) c(iuium) R(omanorum).

²⁹ Both dies are the same as those of the previous item (No. 16, sf 5736).

³⁰ A is apparently ‘open’ (Ʌ), and L has lost its horizontal stroke. The edge of the die can be seen after P, so it was not followed by E or A, unlike the previous items.

³¹ The obverse is damaged in the centre by the binding cord, but is apparently a rather wide N, its second (diagonal) stroke ligatured to the vertical stroke of E. This N is joined above by half a ‘star’, its horizontal perhaps serving as the suprascript bar of a numeral II below, as if it were combined with N. There are many different sealings of cohors II Nerviorum (RIB II.1, 2411.106–141), and this would seem to be a variant in which (like 2411.110) the numeral was not explicit.

³² Another sealing of the cohort (RIB II.1, 2411.116) also features a standing bird facing left, but from another die. They would have been impressed by the decurion with his signet.

³³ The middle letter is a composite formed by a vertical stroke joined below by a horizontal stroke and two short diagonals, perhaps an elaborate A extended downward to fill the space. The initials of the decurion.

³⁴ One of five found (Fig. 12), all neatly lathe-turned with one face dished or countersunk. One of them (sf 5596) is not inscribed, and another (sf 5582) has only a confusing medley of incisions left by successive owners, but the other three (sf 5581, 5580 and 5579), Nos. 21–23, all belonged to the same man, Coccus, as the first or second owner.

³⁵ This graffito occupies the prime position, suggesting that it names the first owner. The letter-forms are the same as those of COCCVS in the next item (No. 22), and evidently by the same hand.

³⁶ This reading assumes that the three letters were made after inverting COCCVS, which seems likely, since it would then make the previous owner’s name unreadable. The other way up, the letters might be IɅV or even IɅL, but ɅVI can be understood as the second owner’s name abbreviated: Avitus for example, or his three initials, A(ulus) V(…) I(…).

³⁷ This graffito occupies the prime position, suggesting that it names the first owner. His name is not certain since princeps, although frequent as a title or military rank, is comparatively uncommon as a personal name, so perhaps PRIM for Prim(us) was intended. There is a short diagonal stroke to the right of N, as if for M, but it seems to underlie it, and be related to another diagonal to its left, as if forming an underlying, unrelated letter.

³⁸ This graffito was made with PRIN inverted. The letter-forms of COCCVS are the same as in the previous item, and evidently by the same hand. Coccus wrote his name as the second owner, initial C cutting the downstroke of N.

³⁹ It is not clear from the photograph whether these lines cut into COCCVS, as if to delete a name already there, or were made centrally by the first owner as a mark of identification. But their prime position, and the awkward placing of COCCVS, suggest they were made first.

⁴⁰ COCCVS again, so presumably by the same hand, but its letter-forms are somewhat different. O is much the same, two curving strokes not conjoining, but no space was left for S, which distorted it; and CC is also different.

⁴¹ Final S is only a short diagonal stroke. Like the next item, this is the owner’s name, whether nominative or genitive, ‘(the property) of …’.

⁴² N and S are both incomplete, but can be restored with certainty. The beginning of the diagonal second stroke of L is just visible in the broken edge.

⁴³ During excavation, but without note of the exact context. It is now in Colchester Museum (COLEM.2000.166.6348), from where Ben Paites sent a photograph by Douglas Atfield (Fig. 15). It is fully published by O. Humphreys and G.J.C. Davis as ‘An iron letter die from Colchester, Essex (UK)’, Instrumentum 62 (December 2025), 34–6, which was sent by Glynn Davis.

⁴⁴ There is no close parallel in RIB II.1, 2409 (dies), but Humphreys and Davis (see previous note) discuss many non-British iron letter dies including CIL xiii 10023, 26, which also reads ‘M’. They suggest that the Colchester die identified a local craftsman or manufacturer like Cintusmus, the coppersmith (aerarius) who inscribed RIB 194, a bronze plaque found nearby on the playing fields of the Royal Grammar School, Colchester. M would be the initial letter of his name.

⁴⁵ PAS ref. ESS-0E46ED.

⁴⁶ During the same excavation by Worcestershire Archaeology as the altar Britannia 54 (2023), 425, No. 1, but in the fill of a fourth-century ditch. Derek Hurst sent a photograph of the sherd identified by Jo Mills.

⁴⁷ Only T is complete and certain. There is just the tip of the previous letter, most like another T, which offers more possibilities than E, C or S. F offers none. T is probably followed by I, despite the diagonal stroke below, since TLV would be an impossible sequence. This stroke, if deliberate, was perhaps a first attempt at V. This V is oddly formed, consisting of two detached diagonals, spaced rather wide. There is too little to be certain, but the letters are probably part of the owner’s nomen, with many names possible.

⁴⁸ By a metal-detectorist, PAS ref. KENT-BCBF20. Details from Richard Hobbs.

⁴⁹ C is cut with bold vertical serifs at both ends, like the two I’s with their bold horizontal serifs at top and bottom. It is not O made like retrograde D, as in the reading SISOI adopted by PAS. The name Siscus is not directly attested, but might be a variant of Sisgus, the father’s name of Veiagenus Sisgi f(ilius), miles ex cohorte Raetorum (CIL xiii 6240, Worms); compare also the feminine cognomen Ulpia Siscia (CIL iii 3272).

⁵⁰ During excavation by Oxford Archaeology East. Details from Ian Rowlandson, who made it available.

⁵¹ The elaborate cross-stroke of T is cut by the next two letters, and a space was left between V and R. In the broken edge to the right of R is trace of a fifth letter, its angle suggesting V rather than I, for the nominative Taur[us] rather than genitive Taur[i] or a derivative such as Taur[inus]. This name Taurus (‘bull’) is quite a common cognomen, already found in Britain in RIB II.5, 2491.124 (a tile with the drawing of a bull) and Tab. Lond. Bloomberg 29. The incisions are filled with a hard black deposit, presumably applied by a subsequent owner to ‘delete’ the name of Taurus.

⁵² By a metal-detectorist, PAS ref. DECD8E, and acquired as Treasure by the British Museum (2006,1007.1), from whose photograph Fig. 20 is drawn.

⁵³ L has an exaggerated top-serif, perhaps to separate it from the line above. This reading differs from that of the British Museum and PAS: ‘… which appears to read IV / II’ perhaps ‘[V]IV(AS) followed by an abbreviated personal name’. There is space to the left of the four letters, occupied by two or three incised points, but no trace of any lettering. VIVAS rings are well attested (see n. 55 below), but this is more easily read as a simple ownership-inscription.

⁵⁴ By a metal-detectorist and declared Treasure, PAS ref. LIN-AF94FC.

⁵⁵ opto with subjunctive ualeas (‘I pray … that you are well’) is frequent in the closing greetings of letters at Vindolanda and elsewhere, so this ring is not explicitly Christian. Compare other rings with similar legends, but with uiuas or uitam: RIB II.3, 2422.25, Iul(ius) Bellator uiuas; 77, uita(m) uolo; Britannia 56 (2025), 499, No. 8, uir bone uiuas. But this ring dates from the fourth or early fifth century (Henig 2006, Type III), by which date the legend is likely to be Christian; compare the Christian uiuas in deo found on rings (RIB II.3, 2422.14, 15 and 70; Britannia 45 (2014), 446, No. 28 and perhaps 513, Britannia 56 (2025), 499, No. 43. It is the only instance of imperative uiue instead of subjunctive uiuas.

⁵⁶ In a deposit dated c. a.d. 62–65/70 during the same excavation by MoLA as the stylus writing-tablets published in R.S.O. Tomlin, Roman London’s First Voices (2016), but only available after this had gone to press. In 2019 it was in the ‘Last Supper at Pompeii’ exhibition at the Ashmolean with an online description by MoLA based on the full account written for Sadie Watson, Archaeology at Bloomberg (2017), but not published there. A summary account, ‘Roman London’s First Voices: a stylish postscript’, appeared in Newsletter 21 (spring 2018) of the Oxford Centre for the Study of Ancient Documents. Both were used by Denis Keyer for his full publication, ‘Inscription on a Roman Stylus from London’, Hyperboreus: Studia Classica 25.2 (2019), 340–50, which improves both reading and interpretation. It is also discussed by Anna Willi, ‘Inscriptions for inscribers: texts on Roman writing equipment’, Journal of Epigraphic Studies 5 (2022), 63–104, which compares it with other inscribed styli, of which it is the most elaborate. Its best parallel now is the bronze stylus from Austria just published as G.E. Thüry, ‘Ein Stilus mit Inschrift aus Carnuntum (Niederösterreich)’, Instrumentum 63 (June 2026), 17–22, which is a mu[n]us (‘gift’) inscribed in similar dot-punched letters with a metrical dedication to the donor’s lady-friend (domina). In the Bloomberg excavation c. 200 iron styli were found, but this is the only one with an inscription.

⁵⁷ Making it uncertain sometimes whether a letter has been lost at the beginning or end of a word, and if so, whether it was omitted or (more likely) lost to corrosion. Missing letters have been restored between square brackets.

⁵⁸ The text is metrical, a couplet of iambic senarii ending in rogo followed by an elegiac couplet, the hexameter beginning with si fortuna dar[e]t, the pentameter with longa. Keyer notes CIL vi 36202 = Carmina Epigraphica Latina 1545 as an example of this format. Line-by-line notes. 1, ab urbe: the ‘city’ is not named, so it would have been Rome, the city. adfero (‘I bring’): E is indistinguishable from F, and R is incomplete. 2, acul[eat]um (‘pointed’): there is only just enough space for the missing –EAT–, the two vowels being scanned as a single syllable by synizesis. 3, rogo (‘I ask’): metrically this belongs to the iambics, and either governs the preceding clause or is added parenthetically if ut habe[a]s is dependent on adfero. si fortuna dar[e]t (‘If only Fortune …’) introduces an exclamatory wish. 4, longa uia (‘long way’): the metre requires this to be ablative, uia being scanned as a single syllable by synizesis. Cen[i]: the third letter is certainly N, not V (as first suggested, for ceu). The lost letter must be I, with even a hint of the three dots required. cen[a] has also been suggested, but there is not enough room for the missing A and the space required before SACCVLVS; and an allusion to ‘dinner’ (cena, ablative) would be quite obscure. Caenius/Cenius is attested as a nomen, and may have also been used as a cognomen (genitive Ceni); otherwise understand Cenus, from which it would be derived, and compare the Celtic name-element –cenus (‘son of’). For the genitive Ceni compare the potter’s stamp m(anu) Ceni (CIL vii 1336, 287, Colchester). Cenus with uacuus (‘empty’) can then be seen as a learned pun: in Greek the adjective κενός means ‘empty’.

⁵⁹ With the next three items during the same excavation by Pre-Construct Archaeology as No. 1 above, and published (sf 44) in the forthcoming final report. This includes four non-literate graffiti, such as ‘X’ (a plain ‘cross’).

⁶⁰ Below them is a third graffito, perhaps unrelated. Part of two diagonal strokes, and probably part of a third coinciding with a crack in the surface, but too damaged to tell if they were made before firing, let alone by the same stylus as (a) and (b). Perhaps a numeral, but too little remains for a reading.

⁶¹ Compare the long S in the ‘signature’ of Victor, Victoris, found twice on Dressel 20 in Britain (RIB II.6, 2493.77 with Britannia 55 (2024), 424, fig. 18).

⁶² Final R is certain. The preceding letter looks like M, but this would be an impossible sequence, so it is probably A cutting the end of a (now) incomplete M.

⁶³ sf 141. S with medial line is the symbol for s(extarii), one sextarius being equivalent to 0.547 litres, with 16 to the modius (equivalent to 8.754 litres). For similar notes of capacity on the rim of a Dressel 20, seven modii followed by the number of sextarii, see RIB II.6, 2494.44 (South Shields), m(odii) vii s(emis), and 46 (London), m(odii) vii s(extarii) xii[…]. Even closer, but without M and S, is 38 (London), (modii) vii (sextarii) iiii.

⁶⁴ sf 125. The lettering is laboured, with some strokes being made twice, and O made awkwardly with an angular first stroke; perhaps because the writer was cramped for space. This may also be why he did not complete the name by adding R.

⁶⁵ sf 129. The graffito is complete, and must have identified the vessel in some way. Straight lines or diagonals are often incised across the width of a rim or foot-ring, their purpose unstated, but probably a numeral or mark of identification. But unlike them, these lines do not extend across the whole width and look more like letters. One numeral followed by another (as if ‘4’ and ‘5’) is unlikely. Probably an abbreviated personal name, Iuv(…), one of those derived from iuvenis (‘youth’) such as Iuventius or Iuvenalis.

⁶⁶ Now in the Museum of London, and published by F. Grew and N. Griffiths in ‘The pre-Flavian military belt: the evidence from Britain’, Archaeologia 109 (1991), 47–84, at pp. 80–1 with a drawing (No. 198) = AE 1996, 944.

⁶⁷ This is the only belt-fitting with a maker’s mark in Grew and Griffiths (see previous note), and the maker Montanus is otherwise unattested.

⁶⁸ The provenance is not recorded, but it was bought by an American airman stationed at Mildenhall, Suffolk, in 1987–90, who also bought a copper-alloy statuette of a warrior found at Shuckleburgh, three miles from Diss, during roadworks in 1990. Perhaps the ring was found then as well. It was inherited by his daughter in Texas, who consulted Martin Henig about both items, sending him photographs. His dating of the ring and stone, and other comments, are incorporated here.

⁶⁹ The letters are rectograde but must be read anti-clockwise, implying that the gem-cutter intended them to be read clockwise in impression, like a coin legend, but omitted to cut them retrograde. The nomen Mannaius, not Mannaus (which is unknown), is almost unique; it is attested once, with the same praenomen, as C(aius) Mannaius C(ai) l(ibertus) [S]phaerus in a Latin list of names found at Ephesus, probably those of Italian businessmen active there. This stone is now in Oxford and published by Alison Cooley in ‘Two Latin inscriptions from Ephesos in the Ashmolean Museum’, ch. 23 of From Document to History (Brill Studies in Greek and Roman Epigraphy, Vol. 12, 2019), who dates it tentatively to 35 b.c. This would suit the likely date of the gemstone, increasing the possibility that it was originally incised for the Gaius Mannaius who owned and then manumitted a slave called Sphaerus.

⁷⁰ PAS ref. NMS-3035FB. Information from the county Finds Liaison Officer, Helen Geake, in whose keeping it now is.

⁷¹ MER is separated from MARTI by the horizontal line which divides the panel in half. A is ligatured within M. Other rings dedicated to Mercury are known from Britain: RIB II.3, 2422.29, 30, 51 with Britannia 48 (2017), 486, Add. (c); Britannia 25 (1994), 306, No. 39; 27 (1996), 451, No. 29; 48 (2017), 477, No. 42. But none to Mars, or to Mars and Mercury; but two or three of the Uley tablets are addressed to them both (Tab. Uley 3, 8 and perhaps 19).

⁷² NI, not M, since the two letters are unlike M on the bezel with its four diagonals. They are separated from ACI by an uninscribed gap. The meaning of NI ACI is unclear, whether it is two words abbreviated or a single word strangely spaced out. Since –aci suggests a genitive ending, perhaps the five letters represent a personal name, the ring being the property ‘of’ this owner. *Niacus is not attested as a name, but perhaps Nicomachus was intended, reduced for lack of space and with –ci for –chi.

⁷³ During excavation by Cotswold Archaeology, from where Jo Mills identified the sherd and sent it for examination.

⁷⁴ The long tail of R is cut by the end of a curving diagonal stroke curtailed by the broken edge, an incomplete letter which must be a vowel; its angle marks it as A, ‘open’ A (Ʌ), since the speck below is damage, not a medial point. This is the second incomplete TRA[…] samian graffito from Britain, the other being RIB II.7, 2501.548 (Catterick). The only common name in Tra– is Tranquillus in its various forms, which is likely here. In Britain compare Tab. Vindol. 196.12, Tranqu[illo]; RIB 897, Tranquilae; 1539, Tranquila. The foot-ring is very worn, but has traces of diagonal incisions across its width; not lettering and hardly numerals, but probably marks of identification.

⁷⁵ By a metal-detectorist, PAS ref. NLM-7F59DA. Information and photographs from the county Finds Liaison Officer, Martin Foreman.

⁷⁶ DTO is followed by ‘the very edge of another letter, probably T’. More than seventy TOT rings, mostly of silver, have now been found, but it is very rare for the god’s abbreviated name to be preceded by D or DEO, indicating that the ring was dedicated to him, rather than used as a signet. Compare the Scredington ring (Britannia 56 (2025), 510, No. 38 with note), incised: DEO TOT.

⁷⁷ During excavation by Wessex Archaeology (Britannia 27 (1996), 435), for which see Vaughan Birbeck, ‘Excavations on Iron Age and Romano-British settlements at Cannards Grave, Shepton Mallet’, Somerset Archaeology and Natural History 144 (2000), 41–116. The graffito is illustrated on p. 88, as fig. 19.2. Kevin Trott made the sherd available.

⁷⁸ The first letter is apparently an elaborate G made with at least five strokes. The second name begins with S, its tail very short; then ‘open’ A (Ʌ), its first diagonal sinuous and extending well above the second, like the Ʌ of Gammi, which would suggest it was written by the same hand. The two names are thus not those of successive owners, but a patronymic followed by the owner’s name. The third letter looks like a narrow O, but the sequence –ao–would be impossible; however, cursive B is suggested by the faint trace of a curving line above and a space to the left. The name Gammus/Gamus is very rare, but the same sequence occurs a century earlier, in a military diploma from Raetia dated a.d. 64 (CIL xvi 5): the recipient’s wife is (in the dative) Sabinae Gammi filiae; Sabina, daughter of Gammus, described as ‘Helvetian’ (Helvetiae).

⁷⁹ PAS ref. SF-204CA6, and fully described in Britannia 56 (2025), 476–7. It has now been given to West Stow Museum.

⁸⁰ The curving outline of C can just be seen, followed by the first diagonal of A. There is only faint trace of the final F, which is required by the sense. This reading is supported by a trulla found in Hungary (CIL xiii 12301, 11) which is stamped in just the same way: Fig. 33(b), above, shows it drawn in Archaeologiai értesítö 5 (1885), 26, where fig. 1 d reads the final letter as V (followed by CIL), but fig. 1 c shows it could well be a damaged F. Both pans are thus by the same maker, Caratus, a third example probably being the fragment published as CIL xiii 12301, 12. These parallels are noted by C.N. Moore in ‘Roman bronze skillets from the river Witham’, Lincolnshire History and Archaeology 10 (1975), 59–60, including the long-lost stamped trulla read as C.ARAT. This is RIB II.2, 2415.12, for which now see Add. (b), below.

⁸¹ In a pit dated c. a.d. 120–50 during excavation by Wessex Archaeology, in whose keeping it is. Rachael Seager Smith sent photographs and other details.

⁸² There is just enough space to the left of the first letter to identify it as the first, its vertical stroke extended downward as if to mark this. It is cut by a curving diagonal which meets the broken edge, and would have extended in a leftward curve for R, but this curve has been lost. Visually ɅɅ or M would be possible, but not before L. The next letter is a perfunctory O, two curved strokes not conjoining, which cannot be II (for E) or V, which in this hand would be two diagonals meeting in a ‘v’. The final A includes a downward medial stroke, as if to mark it as the final letter. These letters are quite well preserved, but the reading Ralonata cannot be certain, since this name, that of the owner, a woman, is otherwise unattested; its etymology is unknown, but perhaps cognate with the unique nomen Ralonius (CIL vi 741).

⁸³ PAS ref. SUSS-52B6EF.

⁸⁴ The three initials of a Roman citizen’s tria nomina, much the most common nomen in I– being I(ulius). The abbreviated praenomen might be an angular S for S(extus), but looks more like G for G(aius). For such a combination, compare the brick-stamp RIB II.5, 2490.1, GIS, probably G(aius) I(ulius) S(…).

⁸⁵ With the next five items during excavation by Ecus Archaeology in Grange Road, Fenham, immediately north of the fort. They are published in the final report with fourteen other graffiti made after firing, almost all of them incomplete, and either non-literate (including an elaborate ‘star’) or only one or two letters.

⁸⁶ For the same graffito on a mortarium rim, see RIB II.6, 2497.34 (Manchester), and compare the coarseware rim sherds below (Nos. 49 and 50). Such ‘numerals’, often only one, two or three parallel lines across the width of the rim, are often found on coarseware (RIB II.8, 2503.34ff.) but their purpose is never stated: probably they identified the vessel within a set or place of storage, but they may have only been a mark of identification.

⁸⁷ Only the lower tip of the first letter survives in the broken edge, but its angle suits E, not A. The cognomen Marcellinus is very common, and at a Roman fort is more likely than the much less common Celtic name Bellinus.

⁸⁸ For the same graffito on coarseware rim sherds, see RIB II.8, 2503.36 and 37, and compare the next item (No. 50).

⁸⁹ For the same graffito on coarseware rim sherds, see RIB II.8, 2503.42, 43, 45, 46, 47, 49, 50, 51.

⁹⁰ The first ‘letter’ might be a badly made X, as if for the numeral ‘15’. But these three figures are not necessarily letters or numerals; perhaps only a series of diagonal notations like the next item (No. 52), or a mark of identification.

⁹¹ Only the upper-left tip remains of the third pair. For similar graffiti on coarseware rim sherds, see RIB ii.8, 2503.89, 90, 91. It is uncertain whether they are numerals, or only marks of identification.

⁹² During excavation directed by Paul Buckland (Britannia 7 (1976), 317, where its discovery is noted), in the Hadrianic destruction levels of a Flavian building and thus from the first fort. Also recorded by Ian Betts, ‘A scientific investigation of the brick and tile industry of York to the mid-eighteenth century (University of Bradford thesis, 1985), 131, but not by R.P. Wright in his ‘Tile-stamps of the Ninth Legion found in Britain’, Britannia 9 (1978), 379–82, where it would have been his type 7. It is thus missing from RIB II.4, 2462.9B, as seen by Paul Buckland, who sent a photograph (Fig. 42) and other details.

⁹³ At the eastern margin of the northern extra-mural settlement by the metal-detectorist Tom Shurlock, who has given it to English Heritage at Helmsley. Information and photographs from Martin Millett.

⁹⁴ A reference to Danum (Doncaster), a fort 40 miles south of Aldborough, is much more likely than the genitive of Danus, a personal name almost unknown. But face (ii) is more difficult. Its only surviving letter might be read as either V or Ʌ (‘open’ A). Its placing suggests that another letter has been lost beside it, but there is no trace of this in the broken edge. If indeed V, it might be a numeral or the sender’s name abbreviated. Two letters would be the initials of his nomen and cognomen. More rewarding is Dani, since it is the first epigraphic instance of the place-name Danum, in the locative Dani, which is otherwise attested only by the Antonine Itinerary and Notitia Dignitatum, in the ablative Dano.

⁹⁵ During excavation directed by Martin Millett, and made available by Jerry Evans. Also found were three small samian sherds, each with an incomplete letter surviving from a graffito.

⁹⁶ S was made with a sinuous downstroke having a short upward return at the foot. These four letters are the end of the owner’s name, followed by the numeral ‘7’ recording the amphora’s capacity in modii. ‘6’ or ‘7’ is usual for Dressel 20, and it was probably followed by a second numeral for sextarii (see note to No. 35 above).

⁹⁷ When it was bought by Irene Manton from Charles Ede, antiquities dealer in London, and read by David Roberts of the British Museum as ‘VITEL or MTIFL’ for Vitellius or Ma(…) Ti(…) Fl(…). Information from Adam Robinson of the Peter Scott Gallery, Lancaster University, where it now is. He sent a better image (Fig. 45) of its illustration in ‘Between science and art: Irene Manton’s collection of antiquities’ by Anna Reeve in Journal of the History of Collections 36.2 (2024), 339–52, fig. 2, which was sent by Raphael Isserlin. The stamp is otherwise unknown.

⁹⁸ The first letter is probably M with a diminutive first diagonal to save space, rather than N or V. F looks more like E, but the lowest horizontal stroke is shorter than the others and extends left, as if an exaggerated bottom serif. The nomen Matius is not common, unlike the cognomen Flavianus (or perhaps Flavus/Flavinus) which like Flavius is sometimes abbreviated to FL.

⁹⁹ Anunciación Carrera de la Red, ‘The year of the discovery of the Stannington diploma’, Notes and Queries 72.2 (2025), 157–9, shows that ‘1760’ was wrongly inferred by CIL from the report to the Society of Antiquaries in December 1761 that it was found in ‘the preceding spring’ or ‘last spring’.

¹⁰⁰ See No. 44 above (with note).

¹⁰¹ Hilary Cool, ‘A circus cup from Annetwell Street, Carlisle’, Glass News 55 (summer 2025), 12–15, with a reconstruction drawing (fig. 2). She notes its well-dated archaeological context (a.d. 83/4–93/4) in a room of the fort’s south gate, the implication being that it was used and broken there by soldiers on duty.

¹⁰² Musclosus the winner (his arm raised), and three also-ran. His name should be corrected from ‘Muclosus’ in the translation of the Caistor St Edmund fragment (RIB II.2, 2419.5). VA is easily understood as an abbreviated ua(le), contrasted with AE ligatured for a(u)e, the intervening V being represented by the second diagonal of A. The vocative termination –e was omitted from Musclose, perhaps by confusion with a(u)e.

¹⁰³ The second line of the obverse was tentatively read as VCO, but is clearly VC followed by a reversed C enclosing a ‘star’. This separates Ↄ, the centurial symbol, from VC, which are the initials of the centurion’s nomen and cognomen, unless I has been lost to the left for an abbreviated cognomen Iuc(undus). The ‘almost illegible’ reverse can now be seen as the numeral VI to the right of two verticals, which presumably abbreviate legionis in some way. Many of the legionary sealings in RIB II.1, 2411 are made from a similar, small rectangular die. This would be the first sealing of a legionary vexillation, but many stone inscriptions record detachments of the Sixth Legion in north-west Britain.

¹⁰⁴ Information from Daniel Holdsworth communicated to the RIB website. These routes are listed by the Yorkshire Mountaineering Club, Yorkshire Gritstone Vol. I (2012) at <ukclimbing.com/logbooks/crags/hetchell-1873>, and include some with Roman names such as ‘Caligula’, ‘Tiberius’, ‘Tarquin’s Terrace’, and ‘Up Pompeii’.