854

Inv.no. T02.38 (one fragment of six). BM Registration no. 2004,0501.71. Dimensions: 91 × 21 mm. Period: II/III. Location: V02.19, below Rooms I and II of Period IV structure. A.R. Birley 2003, 110, fragment (d); 2009, no. 6 (d). Scappaticcio 2009, esp. 61–3.

FIG. 1. 854. 91 × 21 mm. Virgil, Georgics 1.125.

Three joining fragments of a tablet, with three edges broken; only the left edge is original. 855 is on the other side. Inscribed along the grain in the ‘literary’ hand described above.

The marks between the upper edge and the text itself are not ink. The tips of letters survive in the bottom edge, but they are so slight as to be illegible; we cannot even say whether they are by the same hand or not. They should certainly not be read (with Scappaticcio) as s[e]g(niter), a tutor’s comment (cf. 118.2).

As Birley has recognised, this is Virgil, Georgics 1.125. But the three letters after coloni, although by the same hand in darker ink, as if the pen had just been dipped, do not continue the quotation (since 126 would begin ne signare quidem). The first two letters are certainly iu, after which there is only the angle of a downstroke curving to the right. Birley suggests it is p, for Iup[piter], which is possible in the sense that we do not know how p was made in this hand. (In ORC, but not in Capital, it would have had a curving foot.) But Scappaticcio, to enlarge her untenable theory about iusto (see 855 introduction), reads s for ius[to], which is impossible, since the letter curves below the line to the right, and not to the left for s. In this respect it closely resembles the l of coloni, which raises the possibility that the writer was writing another line of Virgil, the only candidates being Georgics 2.163, which begins with Iulia, or Aeneid 1.288, which begins with Iulius.

Lines of Virgil are used for writing practice in Latin papyri from Egypt, no doubt in a military milieu; instances are noted in Tab. Vindol. II, p. 66. Noteworthy is P.Tebt. II 686 = ChLA V 304, in which Georgics 4.1–2 has been written six times in Capital over a discarded account in ORC. 854 is now the first evidence of the Georgics in Roman Britain.

855

Inv.no. T02.38 (one fragment of six). BM Registration no. 2004,0501.71. Dimensions: 91 × 21 mm. Period: II/III. Location: V02.19, below Rooms I and II of Period IV structure. A.R. Birley 2003, 110, fragment (d); 2009, no. 6 (d) ‘back’. Scappaticcio 2009, esp. 63–5.

FIG. 2. 855. 91 × 21 mm. Literary composition (?).

Three joining fragments of a tablet, with three edges broken; only the right edge is original. 854 is the other side. Inscribed along the grain in two different hands (reading downwards): (a) the ‘literary’ hand described above; (b) consisting of large, elongated letters, i.e. address script. Since (b) was inscribed first, we begin with it.

The hand is typical of the ‘address script’ used on the back of a letter to name the recipient (Tab.Vindol. II, pp. 43, 52–3) and, as already noted, identifies the tablet used for 854 and 855 as a discarded letter addressed to Iustus, like the other fragment found with them (868). Two fine lines have been drawn across his name to erase it. Between them, and more or less above the o, is a short diagonal stroke which Scappaticcio takes to be an apex, but is more likely to be owing to the return of the writer’s pen before he drew the second line. There are two more such lines on the joining fragments above, where the tips of letters survive in the broken edge. The letter-tip to the right is probably the long I of iusto, but the others are illegible, and we see no merit in Scappaticcio’s suggestion (63), although it is accepted by A.R. Birley (2009, 277) that they read s[egnite]r (cf. 118.2). Possibly they are the end of a place-name, erased.

We should also mention, if only as a curiosity, that Scappaticcio understands iusto throughout as an adverb, as if a comment meaning ‘justly’ has been added to writing-exercises by a tutor. She admits that this ablative with adverbial force would be unparalleled (69, n. 54), and that she has not convinced Birley who even told her about ‘address script’ (69, n. 51); he has since justly observed (loc. cit.) that such a comment would be inappropriate to ‘the wildly inaccurate rendering of the Virgilian line’ in her reconstruction of 855. But let us turn to this hand (a), the ‘literary’ hand described above.

The tops of most letters have been lost in the broken edge. The tails of f (twice) and r descend below the line, but the other marks there, except for the two lines of erasure already mentioned, are not ink. The text presents difficulties of reading and interpretation, owing to damage and because it is ‘literary’ but unmetrical.

Scappaticcio interprets the text as a blundered attempt at Virgil, Aeneid 10.860–1, ‘adloquitur maerentem et talibus infit: | “Rhaebe” …’ (Mezentius’ address to his weeping horse Rhaebus). She reads it as ]iṭ res cuịtụr merentem [[t]]ṛhebae, which is accepted by A.R. Birley 2009, 277, and has the merit of connecting merentem with Thebae by positing a complex sequence of mistakes and corrections, so it deserves careful consideration. We conclude, however, that it depends on readings which are impossible or perverse. After the first two letters, she accepts Birley’s res with misgivings; justified, since he reads two identical letters as r and s, although neither of them resembles the r in merentem. Fortunately, res plays no part in her reconstruction, since she regards it as corrected in tiny letters below to all[o], which is her reading of the marks there which are not ink. After res, she reads the lower half of e as a diminutive c, double l as u, i and t correctly, and finally ur, despite there being one diagonal too many (part of the bold m) and no downstroke at all for r, but only two specks of damage. This *all[o]cuitur is taken to be Vindolanda’s attempt at Virgil’s adloquitur.

Next merentem is certain, and easily understood as maerentem written by ear. The omission of et talibus infit then requires explanation, and Scappaticcio notes (64, n. 31) that the phrase is also omitted here by two Virgilian MSS, e and c, but does not add that they read ac talia fatur instead. The last word in the line is also difficult. The antepenultimate letter is essentially a downstroke curving left, with something above it. It is perhaps a defective b, too defective to be determinate whether ORC or Capital. The word The[b]ae, ‘Thebes’ (whether in Egypt or Boeotia), must be literary, but Scappaticcio sees it as a mistake corrected by a second hand, the tutor’s, to Rhebae, the vocative Rhaebe written by ear. Whatever the merits of a ‘correction’ which left two Vulgarisms uncorrected (e for ae, and ae for e), it must be emphasised that there is no evidence of an r written over the t. There is neither downstroke, nor the diagonal, only a speck of ink high up between t and h which might belong to either letter.

We reject Scappaticcio’s interpretation, therefore, not for its tortuous series of hypotheses, but for having no foundation in palaeography. Instead we have wondered whether this text may not be something more interesting than a blundered line of Virgil, at least as regards intellectual life at Vindolanda: a piece of free literary composition. There is a good instance of this in the Latin papyrus already cited (P.Tebt. II 686 = ChLA V 304). Not only were Georgics 4.1–2 and some non-literary tags used for writing practice in Capital, but from the same hand apparently, there are some lines in ORC about the mythological labours of Hercules.

We have certainly sought in vain for 855 in the prose canon. The initial ut is followed, not by the subjunctive, but by the perfect indicative; we should therefore understand ita, or restore [sic]ut. The sense would be the same in either case. If the subject of fefellit has not been lost, it must be The[b]ae, an easy mis-spelling of Thebe. This lady was the estranged wife of Alexander of Pherae who procured his murder in 358 b.c., ‘… just as Thebe deceived her innocent (husband)’. It will be objected that the tyrant Alexander, even if innocent of the adultery she suspected, was hardly ‘innocent’ in the general sense; and no echo can be heard of Cicero’s treatment of the murder in De Off. 2.7.25 and Inv. 2.49.144. But tyrannicide was a favourite topic for rhetorical exercises on a law-court theme (controversiae), and perhaps this is a fragment from the prosecution of Thebe, or a woman like her, for murder. Or even a sarcastic remark by the defence.

856

Inv.no. T02.38 (one fragment of six). BM Registration no. 2004,0501.71. Dimensions: 43 × 11 mm. Period: II/III. Location: V02.19, below Rooms I and II of Period IV structure. A.R. Birley 2003, 110, fragment (b) ‘front’; 2009, no. 6 (b) ‘front’. Scappaticcio 2009, esp. 65–7.

FIG. 3. 856. 43 × 11 mm. Appendix Vergiliana, Copa 28 (?).

Fragment of a tablet, three edges broken. The top edge is original, since the other side is part of the heading of a letter to Verecundus (867). Inscribed along the grain in the ‘literary’ hand described above.

Birley has identified the text as Virgil, Aeneid 7.373, ‘[his ubi nequiquam dictis ex]perta Lati[num]’, but this attractive idea is excluded by the two (fragmentary) letters we have dotted. All that remains of the first is an upward-curving line of ink above the break, which cannot be part of p. In ORC the top stroke of p almost invariably slopes downwards, but is occasionally horizontal; it never curves upward. In Capital, which is what we would expect in this hand, p would much resemble modern capital P and so have a loop in the top half, part of which ought to be visible on the tablet. This letter can only be c, g or s. The other is now a vertical downstroke turning rightward at the foot, marked as e rather than i by the beginning of a medial stroke. Its top has disappeared in a loss of surface towards the edge.

The text, by ascending gently towards the top edge and then descending, seems to respect the traces below, which are written in a different, cursive, hand. This would imply that they belong to the discarded letter on the other side, perhaps as part of the place-name in the ‘address’, an implication likely in itself, and strengthened by what may be another two lines of erasure (cf. 855). We certainly share A.R. Birley’s doubts (2009, 277) about reading the adverbial iusto with Scappaticcio (see 855 introduction).

In trying to identify the text, we have considered the possibility of free composition again (cf. 855), supposing Virgil’s [spes in]certa (Aeneid 8.580) were completed by latet, but as it happens, there is a known line which would fit exactly: Copa, 28, ‘nunc uaria in gelida sede lacerta latet’ (‘the spotted lizard now lurks in its chilly home’). Despite Servius’ attribution of this short poem to Virgil, modern scholars prefer an unknown follower who knew Propertius but not yet Ovid, writing very soon after Virgil’s death in 19 b.c. (cf. for example Iodice 2002, 182–3 for a terminus post quem of c. 16 b.c.). It would thus have been available to students at Vindolanda a century later and, like 854, would imply an interest in Virgil which extended beyond the Aeneid.

857

Inv.no. T01.15. BM Registration no. 2004,0501.14. Dimensions: 39 × 105 mm. Period: I. Location: the outermost ditch. A.R. Birley 2003, 90; 2009, no. 1.

FIG. 4. 857. 39 × 105 mm. Strength report.

Three joining fragments of a leaf, complete at the top and part of the bottom, although this was presumably only the top half of a diptych with a text running over into the lower half. There are remains of two v-shaped notches and two tie-holes at the top. The surviving left edge coincides with the left-hand notch and tie-hole, which suggests that about one-third of the width has been lost entirely at the left; and with it, almost all of the entries below line 5, except for some of their numerals. The text is written across the grain and parallel to the short edge. The back is blank. It is a fragment of a strength report very similar to 154 and probably very close in date, which fits the archaeological evidence. What survives of the name of the unit and of the prefect in lines 2 and 3 shows that this is also a report of the cohors I Tungrorum and it could well be the work of the same hand (see note to line 6). 154 suggests the readings offered in lines 2–5 and these are mutually supportive in indicating a loss of some nine or ten letters at the left in 857. Since the dates of the two reports are different, there is no reason to assume that the entries of the various dispositions of groups and individuals, though broadly similar in character, are identical. We take the opportunity in the notes below to offer some further comments and observations on some of the readings in 154.

1. Ịạnuạṛịạṣ. The reading is uncertain, but we must surely begin with a date and since nu looks good, January seems inevitable. Before it, A.R. Birley (2009) reads K(alendas) but we cannot see k, though we do not find as at all easy. We cannot suppose that the first line was indented but it is difficult to envisage a date clause long enough to fill the gap. Perhaps Pridie No]ṇạṣ, though pridie is normally abbreviated to pr. What follows the date is badly smudged and could be compatible with almost anything. The parallel in 154 suggests n(umerus) p(urus), what is identifiable here amounting to little more than the two bold downstrokes of the putative n.

3. For the prefect Iulius Verecundus, see 154.2–3 and 210–12. The latter group of tablets is archaeologically assigned to Period II and we must reckon with the possibility that the First Tungrian cohort was at Vindolanda in Period I and part of Period II (see R. Birley 2009, 183).

4. The numeral for total strength (in 154, probably 752 men) has presumably been lost at the beginning of the line, which concludes with the same total of centurions: in is (centuriones) vi. As in 154, the heading of ‘those absent’ follows in the next line.

5. [ex eis absen]ṭes. This is apparently a heading for the entries which follow, as implied by the gap below it. Judging by 154, it was not followed by a numeral at the end of the line. This sub-total (summa eorum) would naturally be added after all the entries had been totted up, towards the foot of the diptych; but this would have been on the second, now lost, leaf (see below, note to 15).

6. The only surviving letter — two bold horizontals — could be several things, but is compatible with g by analogy with 154.1, 5, 22; if correct, this form is distinctive and would strengthen our impression that 154 and 857 were written by the same hand. The diagonal above it is probably not ink. Therefore restore [singulares le] (ati); and it is apparently followed by the same numeral as in 154 (46), which implies that this report is reasonably close in date to 154 (but with a minimum gap of about five months between January and June). There are traces of letters after the numeral, as A.R. Birley notes, and these suggest a reference to one centurion. The scan of 154 suggests that there are faint traces there too after the numeral xlvi and they would fit in is (centurio) i , as in 857.6. If that is correct, we need to ‘lose’ one of the other centurions in 154, most probably in line 8, where we should read (centurio) i rather than iị. There are also possible traces of a numeral at the right in 154.6, suggesting that this entry (officio Ferocis) should be understood as separate from what precedes and follows.

7. There is an indeterminate trace before a probable o. By analogy with 154, we might expect this to be the Londinio entry (one centurion absent), but we are unlikely to have had [Londin]io here since the restoration is too short (we expect about ten letters lost). From this point on, A.R. Birley’s transcript has an extra line (9), we think wrongly; so hereafter our line numbering does not match his.

8. Perhaps a numeral under a patch of surface dirt.

10. Traces of two letters, perhaps the end of a word; perhaps we could even read ]is, then two digits looking most like ii.

11. The scan of 154.

12. suggests that in Gallia is correct there, though we are now more doubtful about prof]ecti. Here we seem to have viii with a broad v, and just possibly a long i thereafter, which would strengthen the idea that this is the same entry as 154.12 where the number is viiii. A centurion would certainly be expected to accompany them, but it is hard to read [in is] (centurio) i in line 12, which seems to end with a v.

13. The line is lost, but there is a possible trace of a digit towards the right.

14. The trace of the first digit would suit x.

15. The digits are fairly well preserved; the third one apparently trails leftward below the line, but we think this is because it was the last digit; it is too slight and too close to the others to be a centurial sign (which in any case one would expect to have a line to itself and be preceded by in is). The numeral iii is much too small to be a [summa eo]rum sub-total (so A.R. Birley, his line 16, which we therefore reject), and we understand ]um as the end of a gerund, or more likely, a supine like ṣṭipendiatum in 154.14 (Tab.Vindol. III, App. 154).

858

Inv.no. T01.19. BM Registration no. 2004,0501.18. Dimensions: 26 × 53 mm. Period: I. Location: the outermost ditch. A.R. Birley 2003, 92.

A fragment of a tablet with writing across the grain, presumably one half (the lower?) of a diptych. It appears to be complete at the left and perhaps also at the bottom. If it is complete at the left, the narrowness of the surviving fragment ensures that there must have been a sizeable portion lost at the right, although that does not necessarily mean that we have lost a lot of writing.

FIG. 5. 858. 26 × 53 mm. Fragment of a strength report (?).

The lettering is worn and incomplete, and there are dark discolorations which are not ink. The back is blank. In format it is a list of places and special duties to which men and (possibly) centurions have apparently been detached. Compare RMR 63 (‘Hunt’s pridianum’), especially 31ff. (detachments ‘ad’ this and that). It is not clear from what survives whether each line represents a separate entry, or whether some entries (for example 5–6 and 7–9, and perhaps 3–4) consist of a place-name followed by identification of the detachment; we could, of course, have a mixture of both kinds of entry (lines 1 and 2 and perhaps 7 being single-line entries). In several respects the text resembles a strength report like 154 and 857, which raises the question of whether 858 is a missing fragment of 857. But examination of the originals established that they do not match, and here there are no particular palaeographical features which suggest the same hand. There must have been many of these documents, and they would have been dumped en bloc when obsolete, so the shared provenance and similar appearance are far from decisive. We have not attempted to translate this lacunose text.

1. There may be faint traces at the edge above, but nothing that is certainly ink. The first letter might be l rather than b, the second could be e; the word certainly ends with ia. This suggests Belgia vii[ . An unknown place-name, apparently, but possibly a slip for Belgica; the pre-Conquest name is Belgium, but the only authority for (Gallia) Belgia is a variant reading or slip of the pen in the Ravenna Cosmographer (see TLL II 1803.50 = Itin.Rom. II, p. 59).

2. Londịṇio: the first four letters are reasonably clear, but what follows is complicated by surface dirt. There may be a centurial symbol at the right edge, with a digit (i) to follow; this would match the spacing in 154.9.

3. There may be further traces after m, perhaps one or two letters lost before a trace of i, thus .aṃ[..]ị, which could yield a locative place-name, though we cannot suggest anything suitable. There may be a trace at the right edge which could be a broken digit (v?), but not if lines 3 and 4 are a single entry (see above).

4. ..estra[: there are possible traces following, but nothing clear. If this were a single-line entry we would expect a locative place-name. If a two-line entry, perhaps a word analogous to that in line 9; it would be possible to imagine, for example, ..estra[ri(i)], but we have no suggestion for anything suitable.

5. This looks like an -iconio place-ending, preceded by the faint trace of r, even perhaps of ir. The first letter could well be u/v with the right-hand side lost. This would give us Viriconio, an attested variant of ‘Viroconium’ in the Antonine Itinerary, which Rivet and Smith (PNRB, 505) even suggest ‘might more properly be Viriconium’. If this strength report is to be dated c. a.d. 85–95 (Period I), as the archaeology suggests, it might just fall in the period when there was a legion at Wroxeter: the fortress is thought to have been dismantled in c. a.d. 90, having been redundant since the Twentieth Legion was relocated at the end of Agricola’s campaigns (White and Barker 1998, 50; Hassall 2000, 51–67). But the place was still an important centre of communications.

6. This must be an ad entry (purpose or function for which detached); compare line 8. It is not clear whether these entries are explanatory of the previous (place-name) entry, or independent when, as here, there are no clear traces of numerals at the right edge. What follows ad is very uncertain.

7. Brigae. For this place-name, see 190.c.38; 292.c.2; 581.33 (note). There is no firm evidence for its identity, cf. A.R. Birley 2002, 135–8 (Kirkbride?). Just possibly traces of numerals at the right edge, in which case this will be a single-line entry.

8. Certainly an ad entry. The first letter after ad is a good a, unless the first element of m. Slight traces thereafter, and a bold horizontal in the area of discoloration, most resembling t, although n is also possible. The word might end –tos or –tus. At the right iuss[u (see 862.5 and note), followed by a name in the genitive, is possible if the amount lost would accommodate this. Otherwise, we would have to understand this as a number (lvii[) and assume it is a separate entry from what follows in lines 9 and 10.

9. The sequence calcar- is a good reading, and the last two letters are perhaps to be read as is rather than ii, though the latter is defensible. If we read calcarii, it presumably means ‘lime-burners’; if calcaris, it could be the locative of the place-name Calcaria (Tadcaster, see PNRB, 288–9) and can be taken with what follows in line 10 as a two-line entry. This would better suit the pattern of other entries. There is no sign of any writing at the right-hand edge in line 9. Lime was available close to Vindolanda, see 156.4, 314.2 and cf. R. Birley 2009, 61; there is an early-modern lime-kiln a mile away.

859

Inv.no. T02.41. BM Registration no. 2004,0501.72. Dimensions: 6 × 70 mm. Period: II or III. Location: V02.19A, remains of buildings below Rooms 1 and 2 of Period IV structure.

FIG. 6. 859. 6 × 70 mm. Fragment of a strength report (?).

A small fragment of a tablet with writing across the grain. The exiguous content which survives, particularly the large numbers in lines 9 and 10, suggests that this is quite likely to be part of a strength report.

1. Presumably [intra (or extra) p]rovin[ciam, cf. 154.10.

5. Perhaps the end of an entry of the form ad … ]am.

9–10. Very large sub-totals, 360 and 370+, are appropriate to the end of a strength report, i.e. the praesentes and absentes. The grand total of 730+ compares well with the total strength of 752 recorded for the cohors I Tungrorum in 154.3.

860

Inv.no. T03.16(2). BM Registration no. 2004,0501.57. Dimensions: 13 × 21 mm. Period: IV. Location: V38 (drain below stable-block or workshop). A.R. Birley 2005, 92, fig. 136.

FIG. 7. 860, Side B. 13 × 21 mm. Fragment of a strength report (?).

A small fragment with writing, probably by the same hand, on both sides (it is not known which is the front and which is the back). The writing on Side A is across the grain and on Side B along the grain. It is possible that the text on Side A is part of an account (see A.R. Birley 2005, 92) but in view of the magnitude of the numbers (see further the note to line 3), we wonder whether it might refer to the size of a military unit and thus be part of a strength report or similar document. The letters have been jotted hastily, without always completing them, to judge by the m and o on Side B (if that is what they are): this haste and imprecision would be consistent with the scribbled identification of a familiar type of document, in which only the date would really be important. Complementary to this, though equally tentative and uncertain, is the possibility that the text on Side B might be read as a reference to the cohors III Batauorum, already attested in the tablets (see above and note 14, but those derive from Period III whereas 860 is attributed to Period IV (and the cohort was in Raetia by a.d. 107)).

Side A.3. The d appears to be written as a correction over something else, or vice versa. It might have been a denarius symbol, or the first traces at the left, which only partially survive, might be the remains of a denarius symbol. If that is the case, this will be part of an account rather than a strength report, but the figure of 764 denarii, if that is what it is, is abnormally large for a Vindolanda account. On the other hand, 764 is appropriate to the strength of a milliary cohort (see 154.3).

Side B.2. m is broken by the edge and has lost its first stroke; as often in stilus texts, the third of the four conventional diagonal strokes was omitted; o is also rather like a stilus-tablet letter, being a hooked stroke followed by a straighter one, which do not actually join. Before that, c looks good, but the first stroke of h descends in a way not seen elsewhere in the tablets; following that, iii with a bar over it looks good. If this is correct we presumably expand/restore it as […u]m coh(ortis) iii [Batauorum]. We do not know what they called their ‘strength reports’ at Vindolanda: see the commentary to 154 (Tab.Vindol. II, p. 91) for the unlikelihood of the document being a pridianum. This fragment implies a neuter in –um (e.g. diurnum?), which would not be surprising (cf. renuntium, 574–9 etc.).

861

Inv.no. T01.39. BM Registration no. 2004, 0501.35. Dimensions: 210 × 80 mm. Period: IV. Location: V42, Room 4 of a building adjacent to the Period IV praetorium. A.R. Birley 2003, 95–9; 2009, no. 7.

FIG. 8. 861. 210 × 80 mm. Account: (i) Left-hand side; (ii) Right-hand side (the small fragment placed at the top of the lower leaf in A.R. Birley 2009, plate on p. 292, does not fit there and is omitted from our plate).

Double leaf (diptych) now fragmented, with a few pieces missing. The edges survive almost entire, with double tie-holes and notches to left and right. The account consists of 37 lines of text arranged in three columns, a format only paralleled by 184, except that Cols i and ii are written on the first leaf, and Col. iii on the left half only of the second leaf; the right half is blank. Col. i is badly worn, and largely illegible after i.2. Like 184, the account is headed by the name of a century and records names, each followed in the next line by a commodity with its cost, but the latter are not indented. Six names and one commodity are annotated with d, meaning apparently that they were still ‘owing’, an annotation comparable with the check marks found in 184. The heading ‘century of Ianuarius’ indicates that the men are infantry soldiers, despite the problem of Tagomas the uexellarius (see below, 25 note), and raises the same question as 184 of whether this account is ‘official’; they both probably represent transactions between soldiers and a civilian entrepreneur.

1. There is slight but sufficient trace of the centurial sign, which would resemble a modern ‘7’ as in 862.3. The men were thus infantry soldiers; there is no sign of another heading. Their purchase of ‘spears’ (lanceae) does not contradict this, despite the contemporary letter from Carlisle (Tab. Luguval. 16 = Britannia 29 (1998), 55–63) which details the two types of lancea ‘lost’ by members of a cavalry ala. Infantry also used lanceae, notably Catiline’s improvised legionaries in 62 b.c. (Sallust, Cat. 56.3) and the Third Legion combating the mounted Rhoxolani in a.d. 69 (Tacitus, Hist. 1.79; cf. Josephus, BJ 3.95). But the presence of Tagomas the uexellarius remains a problem, since he was a cavalryman (line 25, see further below).

2. scuta is complete (the diagonal stroke belongs to the line below), and must be scutarius (‘shield-maker’) abbreviated. It cannot be a commodity (‘shields’), since Crescens’ purchase is entered in the next line with its cost, the format used throughout. scutarii are not included in the Digest’s list of immunes (50.6.7), but are well attested at Vindolanda (160.4, [sc]utarius, and 184.21, Lucius scutarius), and at Vindonissa (Tab.Vindon. 168, no. 35, Valerio scutario). The term scuta(rius) was probably added for identification, like uexellarius in 25 and uenator in 27: the names Crescens and Victor are common, and already attested at Vindolanda; Tagomas is much less common, but there was another man of this name here, the faber in 862.11.

3. The traces are compatible with the word-ending -aria (a neuter plural commodity), and there is space for three or four letters before it; to judge by the spacing of sudariu[m] in 36, [sud]aria is a possible restoration.

4. The termination of this name is unusual, apparently -ato or even -aio, if the seeming cross-stroke of t is only discoloration along the fibre.

5–11. These lines are so badly worn as to be illegible except for the end of the name in 8 (perhaps -as) and traces of cost at the ends of 5, 7 and 11. Names evidently alternated with costed commodities, as in the rest of the account, but it is not possible to see whether any of these names was preceded by the annotation d (see note to 21).

11. The denarius-symbol is followed by two strokes which are slightly different from each other, but the traces are compatible with the same reading as 17. Since four asses would be only one-quarter of a denarius, and the total (37) ends in half a denarius, another four asses are missing; the end of a lost entry, presumably, or of two which both ended in ‘2 asses’.

12. The name is otherwise unattested.

13. Another neuter plural commodity in -ria costing 1½ denarii like line 3, but there does not seem to be enough space to restore [s]u[da]ria. We agree with Birley that the first letter might be c, but not that cu[cum]a[m] (‘kettle’) is a possible restoration.

14. An illegible name.

15. The trace of the first letter is compatible with d, but the second letter looks like a, not o. Birley’s restoration of [do]labra[m] (‘axe’) is attractive, but would require a difficult mis-spelling, *dalabram.

16. Another difficult name. Birley reads the well-attested Atrectus, but his r is not possible: it is either a or (as the second diagonal suggests) m. The marks between this line and the upper edge are casual discoloration, not ink.

17. The digit after the denarius-symbol is followed by two short horizontal strokes for asses, of which there were 16 to the denarius. Examples are also found in Tab.Vindol. III, 596, 764, App. 182, 184; see further 596 introduction.

18. An unattested name, presumably German.

19. The only feasible restoration seems to be men[sam] or men[sas], ‘table’ or ‘tables’. Furniture is out of place among these commodities of personal use or wear, so this may be an unknown technical term. Vitruvius (10.11.6) applies mensa to the ‘bed’ of a catapult, which is also out of place here, but the joking reference to slices of bread as mensas in Virgil, Aeneid 7.116, suggests that it might be applied to any flat, ‘tray’-like item.

21. The first surviving annotation of d to the left of a name, as also in 23, 25, 27, 29 and 33. In 36 it is placed against a commodity, sudariu[m], but this is presumably a slip at the very end of the list; Birley notes as ‘a curious coincidence’ that its only occurrence in 184 (37) is also before sudar(ium) rather than a name.

Here in 861, Birley understands d as ‘a rather drastic abbreviation’ for duplicarius, but recognises that it is almost unparalleled. He cites CIL III, 1189 and might have added AE 1955, 67, but in both it is conjectural. Better would have been M.P. Speidel 1994, nos 524 and 556, which are two probable instances among the inscriptions of the imperial mounted guards (equites singulares Augusti) at Rome; but even this collection offers many instances of the abbreviation dup(licarius) or dupl(icarius), which is standard not only in stone inscriptions, but also in military documents (RMR, pp. 540 and 557) including Vindolanda (312.back 3). We cannot see the need to note that some of the soldiers charged were on double pay; if identification were needed, as for Crescens, Tagomas and Victor, it would have been added after their name. But there was obvious need to note any sums outstanding, so we suggest that d is for debet (‘he owes’). debent (‘they owe’) is explicit in 181.10, and the abbreviated deb(et) is part of a marginal annotation to 586. d is found as a one-letter abbreviation in an account (808) where the context is too slight to establish its meaning, but it certainly forms part of longer epigraphic formulas such as d(are) d(ebebit). Its use here should be compared with the practice in 184, where some commodities have check marks before them, corresponding to the d already noted before sudar(ium) (184.37). This entry, like three others in 184, was then crossed out by a horizontal line indicating that it had subsequently been paid. Similar crossings-out are also found in 182.

The names Albanus and Albinus both occur in Vindolanda ink-texts, and are popular in Gaul, but would not fill the space available; 3–4 letters have been lost. This must be a derived name such as Alb[ian]us.

23. Birley tentatively reads Liber[iu]s, but there is a clear space after Liber, and no further writing on this line, so the name is complete.

24. The sagum is the standard military cloak, explicitly ‘military’ in Tab. Luguval. 24 (Britannia 29 (1998), 66), and well attested at Vindolanda. Unfortunately the cost of the sagum in 192.8 is lost, since ‘2 denarii’ in 24 (the reading is certain) seems rather cheap: it is the same as a sagellum (32), and only twice the price of a sudarium (36); contrast 5 denarii, 3 asses, for a sagaciam (184.20) and 11½ denarii for a (sagum) infiblatorium (596.i.3).

25. This is the only line which continues on to the right leaf. Tagomas is amply annotated by Birley, who identifies him with the uexsillarius Tagamas in 181.14–15, where he is named immediately after the equites Vardulli, cavalry seconded from the part-mounted cohors I fida Vardullorum, of whom he was surely one. He is unlikely to have been their commander, unless the detachment was very small, since this would have been a post for a decurion. Without being aware of 862.11 and 864.5, which provide further evidence, Birley deduces from the inscribed amphora handle found in the same building (Britannia 34 (2003), 377, no. 37) that Tagomas was the preferred form of the name. He also reads the annotation to the left as a barred u for u(e)x(illarius), but this is not an attested symbol. The graffito to which he refers (now Britannia 36 (2005), 493, no. 48) does not really read TAGO; the excavators originally read TAPEO, and there is a more nuanced account in A.R. Birley 2005, 96 with fig. 138.4. After TA, there is C with a small scratch, perhaps casual; then a tiny hook, not necessarily a letter; and finally a triangle, one line extended, which if it were a letter would be O or D. Even admitting the identification with Tagomas, and the interpretation of the triangle as a barred u for u(e)x(illarius), we cannot see why it should be used here to annotate his name, since this was immediately followed by uexellarius in full. Instead, we read a slightly angular d for d(ebet) (compare 21, with note).

The term uexillarius derives from uexillum, the ‘flag’ borne by cavalry units as well as by detachments (uexillationes) of infantry or cavalry. It is used collectively of the members of a detachment (Saxer 1967, passim), but in the singular a uexillarius was specifically a ‘flag-bearer’. This would be an infantry post if the detachment were infantry, notably legionary veterans (CIL V, 4903, uexillarius ueteranorum legionis) and by extension the centuries of Vigiles at Rome (for example ILS 2169). For this reason, perhaps, inscriptions often specify that a uexillarius was ‘of cavalry’ (equitum). In general it was a cavalry post; the evidence is incomplete, but the modern consensus (after Cheesman 1914, 39–40; and see note to RIB II.7, 2501.14) is that there was only one uexillarius in an ala, but eight in a part-mounted cohort, one to each troop (turma). However, judging by the cohort at Dura, where there were four uexillarii in a single troop (RMR 1, xxxviii.23, xxxix.9, 13 and 17), the duty might also have been shared. More surprisingly, two troops there contained a uex(illarius) (centuriae) (RMR 2, xxxviii.8 and xl.9), perhaps in imitation of legions, where a cavalry uexillarius did indeed belong to a century (BGU II, 600, with M.P. Speidel 1986, 165–7), but it would be difficult to extend this anomaly to Vindolanda.

To conclude, therefore, although Tagomas might have been an infantryman, this would be unlikely, even if there were no reason to associate him with the equites Vardulli. Birley recognises the difficulty, and in his note to i.1 suggests that equites Vardulli were perhaps listed ‘from about line 9’. But there is no sign of this, and the presence of a cavalry NCO remains a problem.

27. Victor, like Crescens and Tagomas, is distinguished from homonyms. uenatores are included in the Digest’s list of immunes (50.6, 7), and CIL III, 7449 explicitly refers to im(m)unes uen(atores) of legio XI Claudia. At Birdoswald, the uenatores Banniess(es) formed a distinctive group (RIB I, 1905). At Vindolanda the prefect Cerialis refers to ‘his’ huntsmen (615, uenatores mei), and there are other allusions to hunting (594 introduction).

28. The numeral is rather cramped in contrast to the wide u elsewhere, but there is sufficient trace of the initial downstroke and diagonal before the bold final stroke. This is much higher than the other charges for lanceas, ‘1’ in 20, 22, 26 and 34; and ‘1¼’ in 17. (The numeral in 30 is lost.)

29. Birley reads the well-attested name Exomnius, but the first letter is nothing like e; it looks rather like u written twice; then x is possible, but looks more like e. There is no sign of o. Next m is possible, but would be unlike the well-preserved m in Tagomas (25); a better reading is rr. Verrinus is only attested once as a Latin cognomen (CIL II, 4242), but in the present context may rather be related to the Frisian name Verritus (Tacitus, Ann. 13.54).

30. The traces are rather slight, but compatible with another lanceas entry.

31. Birley’s Necalaṃẹṣ is attractive in view of RIB I, 1793 and 1794 (Carvoran), but requires some forced readings such as the stubby downstroke and long horizontal stroke as l, although they are nothing like the tall narrow l everywhere else in this tablet; me cannot be read at all.

32. The reading sagellu is straightforward, after which two short downstrokes, each accompanied by slight trace of a cross-stroke, provide the expected m. sagellum must be a diminutive of sagum alternative to the usual form sagulum, and (as sacellum) is also found in Tab. Sulis 33. Since it costs the same as a sagum (24), it was presumably finer in some way.

33. The name is not attested. Birley reads Atepo, a well attested Celtic name, but the initial n cannot be read as at: contrast the t of Tullio (35), and ag in sagum (24) and Tagomas (25), where the diagonal of a is completed before the next downstroke is made. But we endorse his previous (2003) reading of s, instead of p, in view of the letter’s upward second stroke which excludes p (see 856 note).

35. The name is well attested at Vindolanda (184, two instances, one a centurion; 185), but it was not thought necessary to identify him further.

36. sudarium is a frequent entry in 184, where it is noted that it was a towel or napkin, perhaps for use in the baths, which could be worn around the neck like a scarf (cf. Suetonius, Nero 51); it was worn thus to catch sweat (sudum), perhaps also to prevent armour from chafing. It also occurs in Tab.Vindol. III, App. 440.3.

37. The total is separated from 36 by space for four or five lines, and may have been added later, perhaps by another hand: the angular u is quite unlike the flat-bottomed u in previous lines. The total of ‘29½ denarii’ for eighteen commodities accords with the surviving figures. If we exclude the ‘5 denarii’ in line 28 as exceptional, the other thirteen figures total 17¾ denarii, leaving 6¾ denarii for the four lost figures in total. The average cost is much the same throughout.

862

Inv.no. T02.25. BM Registration no. 2004,0501.64. Dimensions: 93 × 113 mm. Period: II or III. Location: V02-19A, remains of buildings below Rooms 1 and 2 of Period IV structure. A.R. Birley 2003, 107–8; 2009, no. 5.

FIG. 9. 862. 93 × 113 mm. List of fabri.

A leaf with eleven lines of text written across the grain. Two tie-holes and notches at the top end. The top right corner is lost, and with it the end of line 1. A small part of the left edge is lost, and with it part of the first letter of line 4 and the first letter of line 6. Otherwise it is almost complete, but now fragmented into four pieces. Unusually the wood is oak, and special difficulty is caused by broken vertical ‘lines’ which look like ink, but must be grain, although sometimes it seems that ink has also run into the fibres. We do not agree with Birley that ‘most of the surviving lettering is particularly difficult to read’, since we have had the benefit of improved scans; but since his own reading is little more than ‘traces’ and disconnected letters except in lines 1–2 and faber in 7, we have not thought it necessary to indicate where it diverges from our own. The hand is idiosyncratic. Most striking is t, with its extended cross-stroke and sometimes exaggerated size, especially when it is the first letter of 8 and 11. The upper diagonal of a is often prolonged to the left, cutting the previous letter and extending still further. r has a long sinuous descender, and like t is sometimes exaggerated in size, notably as the second letter of 8. As in 160 (see below), there is occasional use of interpunct.

The text is a note of work done in the fabrica by the century of Firmus, dated 20 April. The heading may refer explicitly to fabri, and it is certain that three men are named and identified as faber, preceded by two items for the first faber and two for the second pair of fabri. They were hardly the actual workmen (since there must have been many more from a century), but rather the qualified craftsmen in charge of each project. The best parallel for the tablet as a whole is an Egyptian papyrus, P. Berlin 6765 = ChLA X, 409, which records work in a legionary fabrica over two successive days. Each of its two columns is headed by the date (18 and 19 April), and itemises objects made or rather processed (for example shields ‘flattened’); but personnel are not named, only entered collectively by military status (immunes, cohortales etc.). At Vindolanda the best parallel is 160, too fragmentary for full interpretation, which appears to record names in the nominative, followed by their trade and century; it too refers to faber (A12, B4, B9) and probably to [fe]rrum (‘iron’, A13). Tab.Vindol. III, App. 155 may also be noted; it too is dated (24 April), and records the numbers of men in officis, engaged in various duties relating to building-work.

1. After ma there is a vertical streak coinciding with the grain, better suited to i than r, perhaps followed by the edge of a; thus a date in April, not February.

After the date, between the second tie-hole and the broken edge, are two downstrokes, the second curving left, both cut by a diagonal; part of a loop follows. They cannot be Batau[orum?] as Birley suggests, but a truncated faḅ[…] resembling that in fabricae (2) is perfectly possible; the diagonal of a would be elongated, but not so acute, as usual. The word was hardly fab[rica], in view of line 2, but reference to fab[ri] here would anticipate faber nicely in 7, 10 and 12. After it there would be space for ad, as the preposition before opus, of which there is no sign in 2; the latter was evidently centred to complete the heading.

3. The centurial sign (resembling a modern ‘7’) is unmistakable, with its descender reaching the foot of the line below. The bold i which terminates Firmi overlaps with the descender of r from the line above, to the right of which are two vertical streaks in the fibre, but they are not in alignment with 3, and are probably not letters. A ‘century of Firmus’ is not attested, but in 180 one Firmus gives orders; he is evidently an officer, and the phrase iussu Firmi there (180.23, and perhaps also 12) is reminiscent of iussu Musụruni (centurionis) here in 5.

4. The first word presents problems of reading and interpretation. Only a curved stroke remains of the first letter, which is most likely u; then o, and c (like that in fabricae) rather than p; finally -ridem, the i being reduced to a streak in the fibre. Hence ụocridem, the accusative of *uocris, a noun not attested. In response to our enquiry, Dr Paul Russell (University of Cambridge) kindly supplied the following note which we reproduce verbatim:

The next word ends in actam or aptam (but the letter before the obvious t looks more like c in fabricae than p), and the upward stroke of f can just be distinguished from the cross-stroke of t and the downstroke of i at the end of Firmi (3). This reading is supported by factas (6) and facta (8) in the corresponding entries below. Although uecturam / uecturas means ‘transport (costs)’ in 615.A.2 and 649.ii.12 and 14, as Russell’s note suggests, the sense here is more likely to be concrete, ‘a vehicle (for transport)’, as in 600.ii.2, ad capsum ueturae, ‘for the body of the carriage’; the loss of c is a trivial Vulgarism (Adams 2003, 558–9).

5. For iussu with the name of the originating officer in the genitive, cf. 180.5, 12 and 23, 858.8 (?); RMR 63, ii 4 and 5. The centurion’s name is not attested, but the name Musuri[us?] occurs on a military diploma found at Lyon (CIL XVI, 147). The reading is certain except for the second vowel, which has been damaged in the break; but the traces and the spacing suit u better than i.

6. Like *uocris, this seems to be another technical term, perhaps *[c]ircolas for circulos, ‘circular’ objects apparently connected with transport. In view also of the fabri (7, 10, 12) and ferrum (9), we conjecture that they are iron hoops, ‘tyres’ made for wheels. According to Quintilian (1.5.8), cantus was the word for ‘the iron-binding of wheels’ (‘ferrum quo rotae uinciuntur’), but he regards it as not Latin; certainly it is rare, and there may have been a more homely synonym. Diocletian’s Prices Edict (15.31–40) quotes for wheeled vehicles ‘without iron’ (sine ferro); the Latin of this chapter is defective, but the Greek translation shows that iron fittings as well as tyres are meant: note especially 15.36, ‘u[ehicula cum canthis et ferr]aturis’. The adjective(?) after factas may refer to the fitting of these tyres, but we cannot decipher it.

The first digit of the numeral is cut by the descender of the centurial sign (5), and is certainly v, not x. A little after viii, and running at a different angle, there is apparently a fine diagonal stroke descending below the line; but we do not think it is part of the numeral, and are not even sure it is ink.

7. The name begins with Hue- and ends in -us, with five digits in between. There is some indication that they fall into two pairs each linked by a cross-stroke, followed by i, which makes Huennius an attractive reading, but not certain since the name is unattested (see, however, the nomen Vennia in Alföldy 1968, 52, no. 134). The aspirated u identifies it as German, and perhaps it should also be read in 184.27. Huennius(?), like Andauer (10) and Tagomas (11), is identified as a faber, which does not mean simply a workman in a fabrica (that would be fabricensis), but a specialist craftsman, an ‘engineer’ or ‘smith’, one of the many immunes listed by the Digest (50.6, 7). Evidently they were supervisors in charge of the various projects.

8. Initial t is exaggerated (as in 11), and the descender of r reaches the foot of 10 and the leaf-edge. The noun transuersaria (neuter plural) is used by Vitruvius (10.14.1) for the cross-beams under a testudo, a strongly-built siege engine on wheels. In this context of transport and iron-working, tra(ns)uersaria would be iron braces ‘across’ a vehicle to hold it together, one of the iron fittings to which the Prices Edict refers (see above, note to 6). trauersaria is an alternative form with reflexes in Romance (e.g. Fr. travers); the grammarian Velius Longus notes that trans was sometimes reduced to tra- in compounds (Keil, Grammatici Latini VII, p. 66).

There is a horizontal line above n(umero) viii, but this is not to mark the numeral; it extends from the t of facta. There is another just below, which belongs to the t of [p]roductum.

9. item is frequent in lists and accounts (at Vindolanda 180.10, 31, 34 and 37; 182 i.7) to introduce a new entry. ferrum (which in association with faber should probably be restored in 160.13) is qualified here by a compound of ductum, for which there is sufficient trace of the prepositional [p]ro. This technical sense is omitted from standard dictionaries, but is evident in Juvenal, Sat. 15.165–6, who uses produxisse (with the synonym extendere) in the sense of forging ‘deadly iron’ into weapons; and it is well illustrated by Cyprian, Ep. 69.8, ‘turibula quoque ipsa … conflata atque igne purgata in laminas ductiles producuntur’ (‘[pagan] censers are melted and purified by fire, before being drawn out into metal sheets’). But dictionaries do recognise the derived adjective productilis, frequent in the Vulgate in the sense ‘of beaten work, of highly wrought metal’ (to quote Lewis and Short).

10. The name is not attested. The prefix ande- and the element uero-s are frequent in Celtic personal names, but in the circumstances this is more likely to be a German name.

11–12. Initial t is greatly exaggerated so as to fill half a line, the rest of which is blank; the name was completed on the line below. Tagomas is not the uexillarius (861.25 note) since he is identified as faber.

863

Inv.no. T02.24. BM Registration no. 2004,0501.63. Dimensions: 14 × 107 mm. Period: II or III. Location: V02-19A, remains of buildings below Rooms 1 and 2 of Period IV structure. A.R. Birley 2003, 107.

FIG. 10. 863. Side B. 14 × 107 mm. Name-tag (?).

Fragment of a leaf preserving the whole width but now warped into a curve. The short edges are original. Both long edges are broken and notched towards one end, probably by the tie-holes, but perhaps for re-use as a label (see below). Inscribed both sides, A across the grain with part of an account(?), B along the grain with a personal name.

2. Perhaps a numeral ending in ‘35½’.

7. In an account, we would expect mandata or similar (cf. 610.19 note), but in view of B, this might be the name Manduorix. We cannot tell whether A is related to B (the latter identifying the purchaser) or not related (but re-used as a label).

The ‘traces’ are very smudged, and may be an erasure of an earlier attempt to write the same name, Manduorix. This is not attested, but both elements are frequent in Celtic names. Some names in –rix are feminine, for example Tancorix mulier at Old Carlisle (RIB 908). Dr Paul Russell kindly observes that mandu- (‘concerned with’) is the interpretation offered by Holder, Altceltische Sprachschatz (1896–1913), and noted secondarily in D. Ellis Evans, Gaulish Personal Names (1967), 223; more recent interpretations tend to prefer the sense of ‘pony’, see P. Russell and A. Mullen, Celtic Personal Names of Roman Britain at http://www.asnc.cam.ac.uk/personalnames/search.php?s_element=mandu-, X. Delamarre, Noms de personnes celtiques dans l’épigraphie classique (2007), 226. For the patronym, Vassinus might have been expected (cf. RIB 215), but there is no trace of a second s; instead the letter is c or more likely t. Vastus is found as a cognomen in Narbonensis (CIL XII 537; 2463), and it is the name of a potter at La Graufesenque (Marichal (1988), No. 77, cf. 203, No. 98). Russell notes that Vastinus would be a Latin derivative of a stem vasto-; this is the linguistically earlier form of vasso- ‘slave, servant’, a common name-element (cf. also Old Irish foss, Middle Welsh gwas ‘servant’, all of which < *wo-sto- (lit. ‘standing under)); see Russell and Mullen at http://www.asnc.cam.ac.uk/personalnames/search.php?s_element=uasso-; Delamarre, op. cit., 235; M. E. Raybould and P. Sims-Williams, Introduction and Supplement to the Corpus of Latin Inscriptions of the Roman Empire containing Celtic Personal Names (2009), 175, 238.

864

Inv.no. T02.16. BM Registration no. 2994, 0501.57. Dimensions: 67 × 20 mm. Period II or III. Location: V02-19A, remains of buildings below Rooms 1 and 2 of Period IV structure. A.R. Birley 2003, 104–6; 2009, no. 4.

FIG. 11. 864. 67 × 20 mm. Account.

Fragment with part of two columns of personal names. The right edge is original, with a tie-hole and notch. The top edge might be original. The others are broken. The complete name in line 3 locates the width of Col. i, and implies that no letter has been lost entirely in 1 and 2. The original leaf would have been wide enough for a third column to the left of Col. i. The remains of the denarius-symbol in 8 suggest that the text is some sort of account, and not just a list of names (cf. 161 introduction), but there is no annotation against individual names as in 609, so the format is unparalleled.

1. The name is not attested, but the reading is certain. In this hand, final -us is run together, with u reduced almost to a dash (as seen most clearly in 7), so the previous little stroke is part of g (compare that in 5).

2. The end of this name is difficult, and we conjecture that ts has been corrected to tus. Before it, acep is possible by understanding the defective first letter as the end of a. *Aceptus would be a mis-spelling of the well attested Acceptus due to failure to geminate the first consonant; this is rare, but there are two examples in Noricum (CIL III, 5131; Weber 1969, no. 355). 5. The initial t has lost its cross-stroke, but otherwise the reading is certain. There is no indication whether this is the uexillarius (861.25 note) or the faber (862.11–12).

7. Birley offers Attius Iustus, ‘but it must be conceded that the postulated -us of Attius is difficult’. We endorse this comment, and would be surprised to find nomen and cognomen in what appears to be a list of cognomina. But the reading of Mansuetus is straightforward. The name is already attested at Vindolanda (580.6 note), and like Proculus and Secundus is very common.

8. Two large diagonals intersect at the broken edge, and we take them to be the top of a denarius-symbol. To the right, trace of two strokes, an illegible numeral.

865

Inv.no. T02.45. BM Registration no. 2005,0501.75. Dimensions: 40 × 85 mm. Period: III. Location: V02-27A, dividing wall between Period III structures, below wall and drain to east of Period IV building. A.R. Birley 2003, 113–14.

Six joining fragments of an account written across the grain, the edges all broken except perhaps for the bottom. It was not possible to reassemble the fragments completely for photography, and there is a seventh, the sliver with trace of denarius-symbols not transcribed, which could only be placed somewhere to the right. The format is best seen at the bottom (12–14), but can be recognised in the lines above: a list of commodities, each numbered or quantified, with a note of its cost. Since the left margin is unknown, it is hard to guess what they were.

3. Perhaps a word ending in -cae.

5. cl is plain, so this is a numeral; l is extended right, as if to mark it as the final digit. This would suggest a round number, perhaps ‘350’ items. Compare 14 (below).

6–10. All that survives of the denarius-symbols in these lines is the horizontal stroke left, and part of one diagonal in 8, but they can be restored from 11–14.

8. Possibly [n]ucis, ‘nuts’, with the variety specified; the ‘traces’ which follow are compatible with m(odium) i, ‘one bushel’, but this is conjectural. Hazelnuts have been found at Vindolanda, and cf. 591.a.2, nucul(a)e. For nuts in the military diet, see Davies 1989, 198.

11. This entry has been crossed out by a horizontal line, perhaps meaning that it has been paid (see note to 861.21).

12. paria cannot be read (with Birley), not only because it is preceded by r, but because the first letter’s second stroke trends upward for c, not p. The word-ending -aria (for a neuter-plural commodity) is attractive, but would mean that a has an extra stroke to the right (cf. 855 introduction) which it does not have elsewhere in this text.

13. ides is preceded by a faint horizontal stroke, perhaps for s. There is no sign of a succeeding t, which excludes Birley’s reading of id est, which in any case would be redundant before the denarius-symbol and is not found before any of the others. He rejects the idea of [parop]sides (‘side-plates’, cf. 194.A.3), but this is a possibility.

14. This commodity was worth numbering exactly (‘263’), unlike the rounded figure in line 5; but at Vindolanda even coriander seeds were counted, to the number of ‘1,884’ (RIB II.8, 2503.1).

866

Inv.no. T02.57. BM Registration no. 2004,0501.86. Dimensions: 22 × 66 mm. Period: VIA (probably). Location: V02-40A, a ditch cutting through an Antonine causeway. A.R. Birley 2003, 115; 2009, no. 11 (one fragment only).

Three joining fragments of an account or list, inscribed across the grain. Part of the bottom edge survives, but the others are broken. Found with other small fragments, three of them inscribed, which it has not been possible to place.

FIG. 12. 866. 22 × 66 mm. Fragment of an account.

3. There is a superscript line over both letters, which are written in compressed form but acceptable as the abbreviation mo(dios).

5. After ur, perhaps u again; ia cannot be read (with Birley), since the diagonal is set too low, and there is no sign of the second stroke.

7. Birley reads brassicas, ‘cabbages’, but his diminutive c is clearly the second stroke of u, just as in 9 and 10. Before it, mass is certain, and neither b nor r can be read. Final m is difficult since it is incomplete, and just below it are two diagonal downstrokes of unknown purpose; they are not a barred numeral.

9. The first letter might be l.

10. Birley’s tentative [fa]ecis is not possible, since it makes ci out of u just like that in 7 (which he reads inconsistently as ic).

LETTERS (NOS 867–869)

867

Inv.no. T02.38 (one fragment of six). BM Registration no. 2004,0501.71. Dimensions: 43 × 11 mm. Period: II/III. Location: V02-19, below Rooms I and II of Period IV structure.

A.R. Birley 2003, 110, fragment (b); 2009, no. 6 (b). Scappaticcio 2009, esp. 65–6.

FIG. 13. 867. 43 × 11 mm. Fragment of a letter.

A fragment of a tablet which is incomplete at the left, right and foot. The text on the back is published as 856. The text on the front is the heading of a letter to Verecundus, as is guaranteed by the position of salu[tem in line 2. A.R. Birley (2009) erroneously sees this as the name of the writer or sender on the back of a letter and accepts the reading ]pṛạef[, which is incorrect. This recipient is not necessarily Iulius Verecundus, the prefect of the First Cohort of Tungrians who receives 210–12 and perhaps 745, see also 154 and 857. This cognomen is very common at Vindolanda (see Tab.Vindol. II and III, indexes svv.) and elsewhere in the North-West, see Lefebvre 2001. The hand is a fairly stylish cursive.

→

] Verecund[o suo

] vacat salu[tem

] traces [

… . .

‘… to his Verecundus, greetings …’

2. There may be a trace of t at the right. The way in which sal is so elongated as to foul the letters in the line above is remarkable.

3. There are traces of writing at the left which should belong to line 3, the first line of the letter proper.

868

Inv.no. T02.38 (one fragment of six). BM Registration no. 2004,0501.71. Dimensions: 50 × 15 mm. Period: II/III. Location: V02-19, below Rooms I and II of Period IV structure. A.R. Birley 2003, 110, fragment (a); 2009, no. 6 (a).

FIG. 14. 868. 50 × 15 mm. (i) Front: fragment of a letter. (ii) Back: part of an address.

A small fragment incomplete at the top and foot and probably at the left and right margins, containing part of a letter and an address on the back. Like 854, 855 and 856, which were found with it, this seems to be a letter saved for re-use as writing material, but the small fragment which survives shows no sign of re-use. At first sight this appears to be the start of the letter proper after the salutation, i.e. the left-hand column. It was not uncommon for the first line to be in ecthesis in respect of the lines following; and some letters do begin with misi tibi (frater), e.g. 259, 309, 697. But in that case, we would not expect an address on the back since it is normally on the back of the right-hand column. If this fragment is part of the right-hand column, it would be unusual, though not quite unique, that the first line of the right-hand column is in ecthesis (see 215, and the introduction thereto). We have considered the possibility that this is the end of the letter, assuming that Sanctum is not the object of missi and restoring line 2 as Sanctum [a me saluta or similar, but think this less attractive; the two lines are probably the work of the same hand but this is not decisive either way.

Front:

→

… …

missi tibi fra[ter

Sanctum [

… . .

Back:

→

…

] Iusto

…

‘… I have sent to you, brother, Sanctus … (Back) to … (?) Iustus …’

Front. 1. missi: for the gemination, see Adams 1995, 88–9.

2. The common cognomen Sanctus occurs at Vindolanda in 182 and 650.

Back. The hand is much like that of 855b, but ‘address’ scripts tend to be in very similar elongated letters and the hands of the two documents are probably different. Iustus must also be the recipient of 868, but should not be identified with the senders of 345 and 614A. It is quite a common cognomen. There may be space for a lost nomen before it.

869

Inv.no. T01.33. BM Registration no. 2004,0501.32. Dimensions: 90 × 45 mm. Period: II/III. Location: V42, below corridor between Rooms 3 and 4 of Period IV building. A.R. Birley 2003, 93; 2009, no. 2.

FIG. 15. 869. 90 × 45 mm. (i) Front: fragment of a letter. (ii) Back: part of an address.

A fragment consisting of part of the right-hand leaf of a diptych, probably incomplete at the top, with remains of the ends of lines of the left-hand column of a letter and four lines of the right-hand column, including the closing greeting, with an unusual wording. There are no certain traces above the visible text (pace Birley, the digital scan strongly suggests that remains of an apparent descender are not ink). The back has traces of the name of the addressee and possibly also of the place of destination above it and the name of the sender, Secundus, below. The content suggests we are dealing with a legal transaction of some kind.

Front:

→

i

… …

salut]ẹm

2 lines missing

].s

… …

ii

tum subscripseris usque

quo res exegerit. (m ²) bene

uale frater karissime tuo

Secundo

vacat

Back:

→

Traces

… io traces [

ṛạẹ ọḥ(ortis)

traces [

‘(Col. i) … greetings … (Col. ii) … you (will?) have subscribed to the extent which the business required. (Second hand) Farewell, brother, most dear to your Secundus. (Back) (Place-name). To …ius …, prefect of the cohort, [from Secundus.]’

Front ii.1. Birley hesitates between tum and cum but tum is certain and must be the end of a word which began at the foot of the preceding column (see introduction). It is followed by a mark which could be an interpunct, but since there are no other occurrences in this text it should probably be ignored. Birley understands the verb as perfect subjunctive (as also in line 2) but we could have future perfect in both cases, depending on what preceded.

2–3. The PHI disk shows that usque quo occurs in Hyginus, De Munit.Castr. 54 (‘usque quo linea exteriores comprehenderit’), Digest 5.2.29.2 (‘et ideo datae in eo libertates atque legata, usque quo Falcidia permittit’) and cf. 23.3.13 pr. , Pliny, HN 6.13; 17.175 and 208, and Servius, Aen. 12.147. There is no difference in meaning between usque quo and quousque; according to Szantyr 1965, p. 655, §357a, usque quo belongs to ‘late Latin’ and is ‘more vulgar’ than quousque (we are indebted to Dr J.N. Adams for this reference). Frontinus, De Arte Mensoria 18 has ‘transposito ferramento respicere priores oporteat, et perpenso coeptum quo usque res exegerit’. res exegerit is found three times in the Digest: 4.6.21.2; 10.2.5 pr.; 11.4.3 pr.

4–5. We should expect the closing greeting bene … Secundo to be in a second hand and this seems to be the case here, although the two hands are quite similar. This common cognomen had not hitherto appeared in the Vindolanda tablets (see also 864.ii.6) . There is no problem with reading tuo ¦ Secundo (despite Birley’s remarks ad loc.). He rightly observes that the form of greeting is unique in the tablets and rare elsewhere, the only examples apparently being those from Fronto which he quotes.

Back.1. Even if the top edge of the tablet is not fully preserved, as remarked above, we think this first line is part of a place-name in the upper part of the leaf. We could imagine, rather than read, uindol (so abbreviated in 338 and 343).

2. The traces are compatible with a short nomen, e.g. Iulio (there is not room for Flauio), but very faint traces thereafter, if that is what they are, cannot be read.

3. The reading is very tentative: r is reasonably good, and the other traces do not conflict with praef; more traces further to the right suggest coh.

4. Faint traces where the name of sender, which must be Secundus, would be expected. We could imagine ạ ….ṇḍ[ (i.e. ạ Ṣẹ ụṇḍ[o]) and there is a vertical mark further to the right which may or may not be ink. But it is also possible that we have a nomen with all or most of Secundo lost.