² During excavation by Worcestershire Archaeology directed by Jamie Wilkins. Derek Hurst sent details and photographs, including an RTI scan.

³ The altar would have to be turned upside-down (as in fig. 1) to read the text, meaning that a blank, uninscribed altar has been re-used for a different purpose: see the next note.

⁴ Line 1: the first letter looks like a square ‘C’, but there is trace of a downward extension for F, making it taller than the succeeding letters. The names IVCVNDO or SECVNDO cannot be read, nor is there any hint of a religious dedication. Line 2: the cross-stroke of T extends shallowly to the right, and is cut by a vertical stroke. This can be read as TT, but the form is unusual and complicated by a possible diagonal below, as if for ‘L’. But the sequence ɅTLII would be difficult, whereas names in Att– such as Attius are common. Also possible is ɅTIII for Atei, but the name Ateius is less common. Lines 2 and 3: despite the two incomplete letters, which look most like V (or O) and R, Mo|riduno is an attractive reading since it is a well-attested place-name. Rivet and Smith identify Moridunum with Carmarthen and perhaps Sidford (Devon), but the sense of ‘sea-fort’ would also suit Pilning, since it adjoins the Severn Estuary at its crossing-place. This altar-like stone may then be a boundary marker, identifying the ‘farm’ or ‘estate’ of Attius/Ateius (for fundus in this sense, compare Tab. Lond. Bloomberg 50.6, with note), but a good parallel for the wording has not been found.

⁵ In the garden of the Yorkshire Museum, from where Emily North sent details and photographs. It is now YORYM : 2017.312, but when and where it was found originally is not known.

⁶ Line 3 is much the best preserved, its letters except for H having lost only their very bottom. It can be restored as the funerary formula heredes faciendum curaverunt which, since it would have completed the text, not only establishes the width of the original but shows that it consisted of only three lines, even though the lower moulded border has been lost entirely (and was presumably part of an adjoining slab). Line 1 would have carried the name of the deceased, but the traces here are faint and ambiguous. Line 2 is less badly worn, and its traces accord with [.]ATRONOC, with just enough room for a narrow P at the beginning. This word [p]atrono is followed by quite a good C, but the next word cannot be restored. Most likely is c[arissimo], for a dedication to their ‘dearest patron’ by his freedmen and heirs: at York compare RIB 680, … here[edes et lib(erti)] patrono [bene mer]enti fecerunt. But he might have been ‘patron’ of the colonia Eboracensis or an ‘association’ (collegium) within it: a collegium of beneficiarii is attested at York by RIB III, 3193.

There is an incomplete lewis hole in the top of the slab, in the right-hand edge above the end of the surviving text, which would suggest there were originally two such sockets, used to lower the complete slab into place. It may thus have been a sarcophagus lid, but it is not shaped accordingly, and a sarcophagus would have been inscribed, not on its lid (except for the initial D M), but on its front. So this slab, which is certainly not a free-standing tombstone, was perhaps once part of a funerary structure, a mausoleum.

⁷ During excavation by the Vindolanda Trust directed by Andrew Birley, who made it available. It is fully published in this volume of Britannia by Alexander Meyer, Alex Mullen and Joonas Vanhala.

⁸ Meyer, Mullen and Vanhala (see previous note) collect parallels for the phallus and its inscription, and examine cacor: not a mis-cut of the noun cacator (‘shitter’), but a new form of the verb caco (‘I shit’), its first-person singular present passive cacor (‘I am shitted upon’). They discuss what was meant by this obscenity, and whether the phallus is addressing Secundinus or is even identified with him.

⁹ With Nos. 6–11 below during excavation by Newcastle University School of History, Classics and Archaeology, and the Historic England Archaeological Projects Team under the direction of Tony Wilmott and Ian Haynes, by whom and Diana Blumenberg they were made available for study at Vindolanda. For the excavation, see Britannia 53 (2022), 410–12, where Fig. 6 shows the altar in situ. Also found was a miniature sandstone altar, only 62 by 150 mm, 45 mm deep, but it was not inscribed.

¹⁰ The original text was probably more than three lines long, but there is no good trace of lettering below; nor any trace on the capital above. In line 1, only the top survives of the first letter: to judge by the C which follows, it might be C or G. The letter after CɅ might be a cramped N or M. The line is completed by quite good trace of Ʌ, T and O, which would suggest a dative (or ablative) termination in –ato. This might be a dedication ‘to’ a god, but there is no D for deo, and the initial CC / GC is inexplicable. If the first letter were G, it might be for g(enio), but dedication to the ‘genius’ – of a century perhaps, centuriae – would have been followed by a genitive. Lines 2 and 3 are too fragmentary to suggest anything, and S apart from a bold diagonal is very faint.

¹¹ In unknown circumstances, but with the location and date deduced from its exact similarity to the eight plaques in the Water Newton Treasure now RIB II.3, 2431.4–11. The property of a deceased estate, it was auctioned by Mallams (Abingdon) in August 2022, where it was noticed by David Ganz. By courtesy of Mallams it was then examined by Anni Byard, Martin Henig and myself, and adjudged to be genuine (BM Treasure No. 2022 T789).

¹² With the next five items in the same excavation as the altar (No. 4 above, with note). This sealing is <31057>, the next is <31986>.

¹³ The letters Ʌ, L and E are ligatured together for al(a)e, as in sealings of the ala II Asturum and the ala Sebosiana (RIB II.1, 2411.83 and 87 respectively). D for d(ecurio) in the reverse confirms the reference to an ala, but which ala is uncertain. The expansion offered here follows that of sealings of the ala Tungrorum found at Papcastle (Britannia 36 (2005), 487–8, Nos. 23–26) which read $ \widehat {{{\rm \Lambda LE}}}$ TV|NG. But in this Birdoswald sealing, V is incomplete and rather shallow, and even if TV can be read, it is not continued by NG below. Another difficulty is that this obverse resembles the reverse of a sealing of ala I Asturum (RIB II.1, 2411.82) which reads $\widehat{{{\rm \Lambda E}}}$TVD, likewise above a palm branch. This is the only instance of TVD on the sealing of an ala, and it does not identify the ala; its meaning is uncertain (perhaps t(ut)ud(it) for ‘struck’ or ‘stamped’), but it is frequent on the sealings of cohors VII Thracum (RIB II.1, 2411.213, with note). But the Birdoswald sealing has no D after TV, so Tu(ngrorum) is still quite possible.

¹⁴ The format is like the reverse of the ala II Asturum and the ala Tungrorum sealings cited in the previous note, the decurion's three initials above D for d(ecurio). But the two letters to the right of C are incomplete and confused by surface irregularities. They look more like T Ʌ than I T, even though C I T is an attractive reading, given the frequency of C(aius) I(ulius) in Roman nomenclature. For another instance of C T Ʌ, compare RIB II.7, 2501.150 (York), where it is an ownership-graffito on samian.

¹⁵ The obverse is less well-defined than the reverse. The vertical stroke of ‘I’ is continued by a second vertical stroke below, with nothing to its right. It may have been intended as a second digit, unless this was lost in the hole; a space here, between CI and L, seems unlikely. The cohort might thus be numbered I or II, but I[I] seems more likely: both cohorts were in Britain, but only the Second is already attested by a sealing (RIB II.1, 2411.108).

¹⁶ The hole makes it uncertain whether the centurion abbreviated his cognomen to one letter or two, but one would have been sufficient, since he used three letters to make his nomen explicit, ANT for Ant(onius).

¹⁷ <92001> The numeral could be read the other way up as ‘12’, but was probably meant to be read down the handle. It would have complemented another numeral, now lost, at the top of the handle or perhaps on the lip or other handle. This would have given the capacity of the amphora in modii, most likely ‘6’ or ‘7’. There were sixteen sextarii to the modius, which was equivalent to 8.754 litres. Such two-part notes of capacity, in modii and sextarii, are frequent on Dressel 20: see RIB II.6, pp. 33–4; and for examples, 2494.3–88.

¹⁸ <91031> The potter's signature, whether written by himself or a literate supervisor. The first letter is apparently I, but the slight curve at top and bottom would allow L. The two strokes of V do not meet, so they might be LI or even II. P is ligatured to I, which extends downward. The incomplete final letter, marked as such by its exaggerated descender, is certainly S; but because of the break above, it is unclear whether this S was preceded by a cramped V or CI, or even by CIV, if the (lost) V were formed by a shallow loop linked to the top of the long S. The name might thus have ended in –PIVS or –PICIS (genitive), or even –PICIVS. Despite the uncertainty of L and V, a possible reading is Laupi[ciu]s. But this name is attested only once, as the cognomen (perhaps Celtic) of Valerius Laupicius, who commemorated his brother just north of Rome (CIL xi 4085), his brother being a legionary in Legio II Italica. It is not an obvious name for a south-Spanish potter, but is worth considering until a better instance of his signature is found.

¹⁹ <32031> The first I is broken by the edge, but cannot be part of N: Anulius is found as a (rare) nomen, but would have been followed by a cognomen. The second V was made with an initial vertical stroke, which was corrected to a diagonal. Iulius is very frequent as a nomen, but is often used by itself as a cognomen. On samian, for example, see RIB II.7, 2501.265, 266 and 267.

²⁰ <34608> The graffito is much abraded. Its first two letters descend from an apparent cross-stroke, but this is probably casual, since the downstroke of T would have been made first. The third letter is now only a patch of abrasion, but the horizontal lower edge and the spacing suggest L. Apart from this L, the graffito resembles RIB II.7, 2501.254 (York), IVIɅCI, ‘presumably Iviaci’. But the name Iviacus is not otherwise attested, nor is Iulacus. Around the circumference below, but still more coarsely incised and very much abraded, is another clockwise series of letters which also ends in CI. Its final I is much larger than the initial I of IVLɅCI to its right, and is separated from it by a space. The letters before CI are too abraded to be legible, but in view of the repeated CI they may well be another attempt at the same name.

²¹ By metal-detector, PAS ref. LANCUM-A1CA02. It is being kept at the Museum of Lancashire until a local museum can buy it from the finder. Cleaning and conservation may then make it possible to improve this reading from a photograph (see fig. 12), to which John Pearce has contributed.

²² The restoration of IOVI is certain, and the text would have begun with Jupiter, so the width lost to the left can be estimated. To the right is much less certain. Jupiter would have been followed by Juno and Minerva, to make the Capitoline triad, and Victory might have been followed by a formula referring to ‘the other gods and goddesses’ as in RIB III, 3460 (Carlisle), but the formula of dedication and the dedicator(s) are unclear. In line 3, Herculan[us] is surely the name of a dedicator, not a reference to Hercules. It is preceded by what might be ]VIT, a verbal ending, but the third letter (‘T’) looks more like a damaged P or R. To the left of this V may be a distorted S, making the restoration still more difficult. In line 4, COH is preceded by DI, which would suggest a genitive termination, perhaps that of the century to which the dedicator(s) belonged. But the preceding letter looks like a damaged E, and a (genitive) name in –edi would be un-Roman, especially since the next letter to the left may be an incomplete M, raising the possibility that the dedicator was medi(cus), the ‘medical officer’. The numeral of the cohort is more like VII than VI, to judge by the length of the suprascript bar and the possible trace of a digit in the broken edge, which would suggest that the cohort was the Seventh Thracians. They probably garrisoned Brough, since many of their lead sealings have been found there.

²³ With the next item in a demolition layer during excavation by Wardell Armstrong Archaeology directed by Frank Giecco, who sent details and photographs. For the site and excavation see Britannia 53 (2022), 412–15, where this sealing is Fig. 9.

²⁴ The letters A, L and E are ligatured together as in the Birdoswald sealing (No. 6 above, with note), but one horizontal stroke is now incomplete. The ala Petriana was the only milliary ala in Britain, but has produced just two sealings, the other being RIB II.1, 2411.84 (Stanwix), both dies of which are different.

²⁵ The reverse should identify the decurion responsible (compare the next item and No. 6 above, with note), but its interpretation is uncertain. The central X may be a large ‘cross’ or ‘star’ separating the initials of his nomen and cognomen, A(…) and C(…). The ‘crescent’ below (which looks like C tilted forward ninety degrees) may represent the usual D for d(ecurio). Such a symbol is not attested, but in the other sealing of this ala (see previous note) there is a crescent and two stars below H R D, the decurion's two initials followed by d(ecurio).

²⁶ D is now incomplete, but its position mid-way confirms the reading of d(ecurio), the decurion whose initials are in the line above. Here, to the right of the second F, trace of two horizontal strokes can be read as E. It was probably ligatured to L. The decurion is otherwise unknown, and (unusually) there is no obverse to identify his ala, but in view of where the sealing was found, it is likely to have been the ala Petriana again.

²⁷ With the next item during excavation by Carlisle Archaeological Unit, and summarised in Britannia 23 (1992), 148; partly published in J.J. Wilkes (ed.), Documenting the Roman Army: Essays in Honour of Margaret Roxan (2003), 183. Both items are now in Tullie House, Carlisle, with a fuller description of their content.

²⁸ The other face was used more than once and the extensive traces are now illegible, but four vertical lines were drawn across its middle portion, and there are long sinuous lines elsewhere, which suggest that a loan-note was being ‘cancelled’.

²⁹ Notes line-by-line:

1. 1, ex auctione. This phrase occurs in Pompeian stylus tablets to refer to ‘my’ auction or that ‘of’ the person named, meaning the sale of property belonging to them; for example, TPSulp. 81, ex auctione P(ubli) Servili Narcissi.

2. 2, Senticae. This feminine name is not common, but is well attested in north-west Britain by the ornate tombstone of Aelia Sentica (RIB III, 3217) at Low Borrow Bridge.

3. 3–4. These two entries have been written so close together that their letters overlap. At the end of each line is a note of the price, the denarius sign followed by a numeral, confirming that they are two entries, not one. But none of the other entries overlap in this way, and the likely explanation is that one entry was written to correct the other: the first would have been erased in the wax, and the second written in its place. One item is a neuter noun ending in –ium, which cost ‘5 denarii’. The other, which cost ‘7½ denarii’ (with s(emis) for ‘half’), looks like horreum (‘granary’), but this is inherently unlikely since the other entries relate to clothes.

4. 5, sintosem. There is a just a hint of the cross-bar of T, but the next letter is apparently a cramped O, not the E which might have been expected, if the word is a variant spelling of synthesim. This word is Greek, and i for y and t for th are frequent in transliteration. It is likely that the accusative of synthesis was intended, since the other entries relate to clothes and synthesis means an ‘outfit’ or set of clothing: compare Tab. Vindol. II, 196.4, where its meaning is discussed by the editors. The numeral after the denarius sign might end in II for ‘22’, but the two incomplete strokes look more like the ‘V’ in numerals below.

5. 6, auctione. The reading is guaranteed by its repetition in 1 and 9, each time followed by a personal name. The letters after AVC have lost their upper part, but their traces are consistent with TIONE. The name is apparently Alcimi (genitive), although surface corrugation has removed the tip of A, and most of L and C. This is the first instance of Alcimus in Britain, but although Greek by origin, it is often found in the Latin West.

6. 7, tossiam. The cross-stroke of T has been lost in the wood-grain. The word is rare, but is well attested in the list of clothing (etc.) sent from Britain by its governor, Claudius Paulinus (CIL xiii 3162), which includes a ‘British rug’, tossiam Brit(annicam). Three such rugs, in the variant spelling tosseas iii, are listed in Tab. Vindol. II, 192.6, followed by a ‘cloak’ (sagum) in 192.8; and the word probably recurs in 439.10.

7. 8, sagum. ‘Ten military cloaks’ are mentioned in an ink tablet from Carlisle (Tab. Luguval. 24), decem saga m[i]litaria. At Vindolanda, as well as the tablet just quoted, there are ‘three cloaks’ in Tab. Vindol. 207.2; ‘cloaks’ in a letter listing clothing (uestem) (255.9); and sagum (singular) in a long list of items with their price (App. 184.back 6).

8. 8, (denarii). The numeral, as in line 5, ends in two incomplete strokes which look more like V (‘5’) in the line above than II (‘2’).

9. 9, auct[ione]. A is complete, and T retains its cross-bar. After A, the scribe wrote C by mistake, but corrected it by writing V over it. There is only faint trace of N and E, but the restoration is certain in view of lines 1 and 6, where again auctione is followed by a personal name in the genitive.

10. 10, Accauni. Like Alcimi (6), this name too is damaged by the contraction of the surface across the middle of the letters, but the traces suit this reading. The name is rare, but is attested at Bordeaux as Acaunus (CIL xiii 685).

11. 11–12. The entry for Accaunus’ auction is too damaged to be legible. In line 11, dolabram was suggested in Wilkes 2003, 183 (see n. 27 above): the word certainly seems to have ended in –RAM, with quite good traces of O and L, but an ‘entrenching tool’ is unlikely in what seems to be a list of clothing. Line 12 may end with a numeral and s(emis), as in 4, but even the denarius sign has been lost in the damage.

³⁰ As with the previous item, the cursive has been transcribed in capitals and words separated where possible. TT of Attio is oddly formed, probably because the stylus was not lifted between strokes, and the final O of Suburano has been lost in a medley of strokes. These include a very large S which spans all three lines; it apparently represents the expected COS for co(n)s(ule), since there is no trace of C and O nor even space for them. Attius Suburanus was suffect consul in a.d. 101, but his second consulship in 104 was ‘ordinary’ and used to date the whole year. His colleague is not named, probably to save space, just as in Tab. Vindol. III, 581.20, where their consulship is used as a date-heading but only Suburanus is actually named. In the line above, there would have been a day/month date, but this is now illegible. Immediately after the consular date is BENO or BVNO, or perhaps QVINO, but this is unexplained: perhaps the place of execution, but no such place-name is known, and it would surely have been preceded by actum (‘executed’).

³¹ During the same series of excavations as the two lead sealings (Nos. 13 and 14 above, with note). The tiles are noted in Britannia 53 (2022), 414, with Fig. 8.

³² This stamp has been found at other sites in Carlisle: see RIB II.5, 2483 (vi–li), where its expansion and interpretation are discussed.

³³ By metal-detector, PAS ref. PUBLIC-A895F7. Sally Worrell sent details of this item and other inscribed Portable Antiquities.

³⁴ What this numeral refers to is unknown. According to the standard of 327.45 g assumed for the Roman libra (RIB II.2, 2412, p. 1), eight ounces (unciae) would be 218.30 g. But this lump does not appear to have been cut down from a larger piece of about that weight. Nor can the V be read as the ‘ounces’ symbol followed by III (compare RIB II.2, 2412.70); and in any case, three ounces would be only 81.86 g.

³⁵ By metal-detector, PAS ref. SF-7B3489.

³⁶ The third diagonal of M is cut by a short cross-stroke which might be casual, but probably indicates the letter A ligatured within it. The god would then be Mars or the Matres, not Mercury or Minerva.

³⁷ In circumstances unrecorded, but now in Colchester Museum (COLEM: 1936.654.1), on whose website it was seen by Scott Vanderbilt.

³⁸ An abbreviated personal name, perhaps Cat(us), since the feminine form Cata occurs on samian at Colchester (RIB II.7, 2501.131). Otherwise a cognate name or Catullus (etc.). The Museum cannot be sure ‘without more context’ that the graffito is Roman like the sherd, and not an excavator's joking reference to Colchester Archaeological Trust, but it does not look modern and the letter-forms are acceptably Roman.

³⁹ By Mechanical Removals and Installations, Gravesend (M.R.I.), from where Steve Trim sent details and a photograph. It was found eroding from the foreshore with pottery, salt-making briquetage and animal bones.

⁴⁰ The name is uncommon, but must derive from Similis. This is the first instance of his stamp.

⁴¹ By metal-detector, PAS ref. LIN-1958BA.

⁴² Reading and interpretation are uncertain, but this is perhaps a personal name abbreviated to its three initials, T(itus) A(…) A(…). PAS reads it the other way up as M, perhaps M(ercurio), but the cross-bar of T is evident, being made with two short strokes; and the next letter is Ʌ rather than (if inverted) V, since its second diagonal cuts into the first. There is little trace of the fourth diagonal of ɅɅ, and these two letters might be a crude M, but TM would be more difficult to interpret.

⁴³ With the next three items during excavation (Britannia 43 (2012), 336) by Pre-Construct Archaeology, from where Eniko Hudak and Jon Butler made them available with other graffiti, still more fragmentary, which will all be published in the final excavation report being prepared by A. Douglas for Transactions of the London and Middlesex Archaeology Society.

⁴⁴ Since they both end in X, graffito (iii) may repeat (ii). Its first numeral, three incomplete diagonals, is either ‘3’ if complete (III), or if incomplete, either ‘4’ ([I]III) or ‘8’ ([V]III). One modius (sixteen sextarii) was equivalent to 8.754 litres, making these possibilities 31.73 litres, or 40.47 litres, or 70.75 litres. This sherd was found with two others, but they do not conjoin, and the exact amphora-type has not been identified. But according to the Southampton University (2014) website, Roman Amphorae: a digital resource, the average capacity of the Africana 2A Grande was 60–65 litres, of the Africana 1 Piccolo 39–42 litres, which would suggest [I]III X (40.47 litres) as the only match.

⁴⁵ Marianus is quite a common name, already found twice in Britain: on a tombstone from Old Carlisle (Britannia 45 (2014), 432, No. 2) with the cognate name Marinus, and on a samian platter from Vindolanda (Britannia 36 (2005), 493, No. 42).

⁴⁶ S is scratched over a small ‘star’ (✻), perhaps a previous ownership-mark. Like the cross-stroke of A, the third stroke of N was made twice. S is certainly the first letter, but to the right, it is uncertain whether the name continued after N, or was abbreviated to SAN. Probably it was written in full, whether Sanctus or its derivative Sanctinus. Both names are quite common, and they occur together in the London Bridge curse tablet (Britannia 18 (1987), 360, No. 1), where they are written as Santinus and Santus with omission of the velar [k], which may indeed have been the case here.

⁴⁷ A personal name, whether or not abbreviated. Although such a short fragment, it is included here since two coarseware sherds were found at Tobacco Dock in 2002, inscribed VI and […]VIC[…] (Britannia 42 (2011), 449, Nos. 13 and 14). Names such as Victor and Vitalis are very common, but perhaps it is the same owner.

⁴⁸ During excavation by the University of Manchester Archaeological Unit (Britannia 36 (2005), 419), for which see R.A. Gregory, Roman Manchester: The University of Manchester excavations within the Vicus 2001–05 (2007). Richard Gregory made it available.

⁴⁹ The first and last surviving letters are now incomplete. The first might be a large K, but looks more like the right-hand half of a six-pointed ‘star’ (✻), whether this was a previous ownership-mark or (as it would appear) deliberately incorporated into the owner's name. The last letter is either I or V, but the sequence would suggest I. The graffito is too incomplete to be certain, but seems to be an elaboration of Proximus, the owner's name, with s reinforcing the [ks] of x, a frequent ‘Vulgarism’. Compare the dedicator of RIB III, 3180, Q(uintus) Nerat(ius) Proxsimus.

⁵⁰ During excavation by the Vindolanda Trust directed by Robin Birley. It is now in the British Museum, where it was photographed by Alex Meyer, Alex Mullen and myself, with thirteen other tablets, nine of which are noticed in E. Birley, R. Birley and A. Birley, Vindolanda Research Reports II, The Early Wooden Forts (1993), where this tablet is illustrated on p. 26 as fig. 6.1 (Inv. 532), but not transcribed. We hope to publish them all in due course.

⁵¹ The downstrokes begin with a short upstroke like a serif. C is a long gentle curve, not a semi-circle. The cross-stroke of T is an upward diagonal. S is crossed with a medial line, the usual symbol for sextarii (compare RIB II.6, 2494.10). This is an improvised pittacium, the term used by Petronius (Satyricon 34) for the label attached to an amphora identifying its wine. The Vindolanda label also notes the quantity, for which compare a dipinto from Newstead (RIB II.6, 2492.7) and a graffito from Chester (RIB II.6, 2494.10). picatum (‘tasting of pitch’) was a premium wine from the Rhone valley, attested by amphora dipinti such as AE 1988, 864 (j), pic(atum) exc(ellens), and 865, pic(atum) uet(us). It is also noted by Pliny (Nat. Hist. xiv.3.18, 4.26, 6.57; xxiii.24.47) and Martial (xiii.107), and a ‘motto’ beaker ironically compares it with Aminean, which Pliny (xiv.3.21) regards as the best wine of all: parce picatum da amineum (CIL xiii 10018, 135), ‘go easy on the picatum, serve amineum’.

⁵² With the next six items during excavation by the Vindolanda Trust directed by Andrew Birley, who made them available at Vindolanda.

⁵³ SF 23138 is complete, but the surface is corroded. Like many weights, it is under-weight according to the standard of 327.45 g assumed for the Roman libra (RIB II.2, 2412, p. 1), but it coincides with three other two-pound weights found in Britain: RIB II.2, 2412.42 at 638 g, 105 at 632 g and 106 at 639 g.

⁵⁴ SF 23237. The graffiti are inverted as to each other. The lettering of (i) is larger and better-formed: C is almost circular, R stylishly made with an exaggerated bottom-serif. But despite some difference in style and appearance, the two graffiti record the same name. The first two letters of (ii) are certainly CR, the fourth letter T; and what remains of the others is quite compatible with those of (i). The only difficulty in them both is the fifth letter (see below).

⁵⁵ SF 23086. The first letter is incomplete, only a short diagonal stroke, but was evidently an angular C since the cognomen Cornutus is quite common. But this is its first occurrence in Britain, and the century is otherwise unattested.

⁵⁶ SF 23443. This sequence does not suit a cognomen on its own, even by taking IC as an incomplete R: Nucrinus is very rare. More likely is a nomen and cognomen in the genitive case, written without a break. Minucius is quite a common nomen.

⁵⁷ SF 23389. The second letter is incomplete, but enough remains of both horizontal strokes to make D almost certain. Although less than three letters, the graffito is included here because it probably consisted of two letters only: compare the almost complete Dressel 20 from Vindolanda incised after firing with the same two letters, AD (Britannia 36 (2005), 495, No. 69); also the handmill from Vindolanda (RIB II.4, 2449.2) which reads AD, (centuria) Ad(…). They abbreviate a personal name to its first two letters, the most likely being Adiutor, since it is a frequent cognomen already found in the Vindolanda tablets (Tab. Vindol. 199 and 588, and compare 214).

⁵⁸ SF 23312. The graffito is complete, and unlikely to be an abbreviated personal name, since a third letter would surely have been added. For numerals on samian, compare RIB II.7, 2501.837–870, but their purpose is not stated.

⁵⁹ SF 23411. The letters are incomplete, and the first might be R, but its incomplete lower loop suits B. The third letter might be T, but this would surely have been tilted further to the right, to align its vertical with the diameter of the base; it is much more like Ʌ, ending a word. The graffito thus seems to be a woman's name ending in –bia, but with only a few possibilities such as the nomina Fabia and Vibia treated as cognomina. To read […]ria would suggest more possibilities such as Maria or Valeria, but R (as already said) is difficult.

⁶⁰ During excavation by the Central Marches Archaeological Research Group directed by John Allen, published by him in Transactions of the Shropshire Archaeological and Historical Society 88 (2013). For the spoon and its inscription, see pp. 71–2, where they are illustrated with reading and full commentary by Catherine Johns. Its discovery was noted in Britannia 22 (1991), 250 and 27 (1996), 451 n. 46, but the reading was quoted as utere felix vivas. Now in Shrewsbury Museum (SHRMS: 2019.00205), from where Emma-Kate Lanyon sent photographs.

⁶¹ The reduction of [ks] to [s] is a trivial ‘Vulgarism’, but this seems to be the only instance of felis for felix in the stock phrase utere felix. It is followed by only three letters, VIV, since there was not enough room to complete vivas.

⁶² By metal-detector, PAS ref. SF-F57791 with information from David Brear. Sold by TimeLine Auctions, 21 February 2023 (Lot 136). It is the third example known, the others being RIB II.3, 2421.44 and Britannia 32 (2001), 396, No. 39.

⁶³ The brooch was made in the ‘region’ of Lagitium (Castleford): see Britannia 32 (2001), 396, n. 53. The local museum, Wakefield Museum, tried to buy it but was outbid.

⁶⁴ By metal-detector, PAS ref. SF-813507.

⁶⁵ Omega is inverted and to the left of Chi-Rho, Alpha to the right. But Rho is rectograde. If a retrograde inscription were intended, it was evidently bungled.

⁶⁶ By metal-detector, PAS ref. SUSS-687D22.

⁶⁷ The full inscription would have invoked long life (etc.) for the person named, using vivas for example, as in RIB II.3, 2422.14 and 70. The sequence –AGR– suggests his name was Agricola or Agrippa, or perhaps Syagrius.

⁶⁸ By metal-detector, PAS ref. WILT-A646EC.

⁶⁹ The letters are made with exaggerated serifs. Martin Henig comments that the hair-style is fourth-century, and that the ring is related to ‘Brancaster’ rings with square bezels, especially RIB II.3, 2422.15, a ‘betrothal ring’ with facing busts and the legend VIVAS IN DEO. He also compares 2422.25 and 42, with a single bust and IVL BELLATOR VIVAS and VIVAS respectively. This type of bezel with facing busts is widespread: see J. Spier, Late Antique and Early Christian Gems (2007), Nos. 25–66, and especially 35, inscribed SEPTIMI ELIA VIVATIS.

⁷⁰ In circumstances unstated, PAS ref. DYFED-12602A. It is ‘thought to have been scratched in while the clay was wet’, but from the photograph it looks as if the surface was already hard and brittle. After firing does seem more likely, especially since it would have been difficult to inscribe the receding inner face (see the profile in fig. 41) when the pot was unfired and entire. But it is hardly an assertion of ownership: ownership-names are usually scratched on the outer wall of a vessel or underneath it. The broken edge to left and right makes it uncertain that the graffito is complete, but since the word is neatly centred between them, it does look as if the writing-medium was already a sherd.

⁷¹ elephantus (usually masculine) means ‘elephant’ and (by extension) ‘ivory’, but is not attested as a personal name. The graffito does not look modern, but its reference and purpose are unknown.

⁷² Falkirk Museum acc. no. 2008-010-453, the photograph on whose website was seen by Scott Vanderbilt. Either reading is possible, but the gap between the successive diagonals favours NV rather than ɅN.

⁷³ And not IV Gall(orum), which was repeated by mistake. The correct restoration is noted there in n. 43, as seen by Scott Vanderbilt and made without comment in AE 2008, 800.

⁷⁴ Before its return to the site museum at Aldborough, where it now is. Published in Britannia since it was found a few months after the closing date for RIB II.1, where it is noticed in the note to 2411.36 (Binchester).

⁷⁵ The Aldborough sealing has lost the first letter, P, except for its very foot, in the hole made by the binding-cord; but it can be restored from the Felixstowe sealing. Both sealings were first read as P B I, p(rovinciae) B(ritanniae) I(nferioris), by comparing them with two others, RIB II.1, 2411.35 (York) and 36 (Binchester), which depict a bull facing right with the letters P B I in relief above. This was the badge of Lower Britain, as confirmed by 2411.37 (Coombe Down), which depicts Upper Britain (P B R S) with a stag. The commentary to 2411.34 (Felixstowe) notes that it ‘is now more damaged than when drawn by Roach Smith. What now appears as the lower half of the I is there illustrated as the right horn of the bull.’ But the Aldborough sealing (see fig. 42) confirms that the bull did indeed have two curving horns, its right horn just to the right of B. There is no sign of I for I(nferioris) here, so this letter was probably overlooked by the die-maker when he cut the horns where it should have been as well.

⁷⁶ Before its return to Tullie House, Carlisle, where it now is (CAR 89 ABB B (3) W <19>). Also noted in Wilkes 2003, 180–81 (see n. 27 above). The letters have become less legible than they were thirty years ago because the wood has contracted slightly, which has both accentuated the wood-grain with its horizontal ribbing and coarsened the incisions. There is no longer trace of quos towards the end of line 7, but otherwise the letters can still be distinguished from the extensive traces of a previous text which have been drawn in outline. This text also seems to have begun with a consulship-date of Domitian, and is likely to have been another such loan-note.