Hostname: page-component-586b7cd67f-tf8b9 Total loading time: 0 Render date: 2024-11-23T22:21:23.489Z Has data issue: false hasContentIssue false
  • English
  • Français

The Twentieth Legion at Wroxeter and Carlisle in the First Century: The Epigraphic Evidence*

Published online by Cambridge University Press:  09 November 2011

R.S.O. Tomlin
R.S.O. Tomlin
Affiliation:
Faculty of Modern History, Oxford University
Get access Rights & Permissions [Opens in a new window]

Extract

There is only one piece of explicit evidence that the Twentieth Legion was ever based at Wroxeter (Viroconium), a tombstone discovered in 1752, since published as RIB 293. That the evidence is explicit, however, has been questioned.

Type
Articles
Information
Britannia , Volume 23 , November 1992 , pp. 141 - 158
Copyright
Copyright © R.S.O. Tomlin 1992. Exclusive Licence to Publish: The Society for the Promotion of Roman Studies

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

1 E. Schallmayer, K. Eibl, J. Ott, G. Preuss, and E. Wittkopf, Der römische Weihebezirk von Osterburken I: Corpus der griechischen und lateinischen Beneficiarier-Inschriften des Römischen Reiches (1990), 808–13, Index 7. This great corpus is cited below simply as ‘Schallmayer’. See also A. von Domaszewski, Die Rangordnung des römischen Heeres (ed. B. Dobson, 1967), 313–3 (Register A 1, s.v. beneficiarius); and N.B. Rankov, The Beneficiarii Consularis in the Western Provinces of the Roman Empire (unpubl. Oxford D.Phil. thesis, 1986).

2 op. cit. (note 1), 811, s.v. beneficiarius praefecti legionis. None of the texts says explicitly whether the praefectus is castrorum or pro legato.

3 Beneficiarii ranked as principalis, but this expansion of PR (which no one has suggested) may be rejected as redundant here. It was not used by beneficiarii on their altars and tombstones.

4 Speidel, M.P., ‘Princeps as a title for ad hoc commanders’, Britannia xii (1981), 7–13.CrossRefGoogle Scholar

5 Schallmayer, op. cit. (note 1), No. 290 = E. Vorbeck, Militärinschriften aus Carnuntum (2nd edn, 1980), No. 300, reads be(neficiarius) leg(ati) pr(ovinciae), but this only reproduces the original reading of Bormann, in RLiÖ v (1904), column 133, who expanded the text by reference to CIL VII. 156 (now RIB 293). The two texts thus hang together: see further, note 24.Google Scholar

6 RIB 292, 294, and probably 296 (whose attribution to Legio VIIII by Alföldy, G. in BJ clxvi (1966), 641, is unnecessary). For the dating, see note 26.Google Scholar

7 Britannia: a History of Roman Britain (3rd edn, 1987), 67Google Scholar, following Webster, G., Trans. Birmingham Arch. Soc. lxxiii (1957), 107, note 2. It should be noted, however, that the unspoken premise that the tombstone of Legio XX is contemporary with those of Legio XIIII is unfounded. Unlike Secundus of Legio XX, none of the soldiers of Legio XIIII has a cognomen, an omission already rare by the mid-first century (compare Suetonius, Vitellius 2.2). (Enough survives of RIB 296 to be able to reconstruct the line-width and to deduce the lack of a cognomen.) Secundus' tombstone (RIB 293) thus ought to be later than the others, in itself a sign that Legio XX succeeded Legio XIIII at Wroxeter.Google Scholar

8 ibid. and p. 75. The conclusion is surely correct; I will argue that there is also good epigraphic evidence. For Graham Webster, who also locates Legio XX at Wroxeter, ‘this interpretation is entirely based on historical inference’: see his review of H.R. Hurst, Gloucester: the Roman and Later Defences (1986), in Trans. Bristol and Glos. Arch. Soc. cvi (1988), 229Google Scholar. In effect this is because Legio VIIII Hispana is accounted for, and Legio II Augusta (or most of it) is thought to have moved from Exeter to Gloucester to replace Legio XX. Archaeologically, Webster kindly informs me, the evidence of two periods in the legionary barracks at Wroxeter strongly suggests a change of garrison. Jarrett, M.G., ‘Legio XX Valeria Victrix in Britain’, Arch. Camb. cxvii (1968), 77–91Google Scholar, thinks that Legio XX probably moved to Wroxeter, but admits ‘there is no positive evidence for this’ since RIB 293 ‘must be discounted’ (ibid., 80, note 16). Hurst (op. cit., 127) argues that Legio XX retained Gloucester as its base, the only evidence for Wroxeter being RIB 293, which ‘carries little conviction’.

9 Since c. 57, in the governorship of Q. Veranius, the most likely date for the occupation of Wroxeter by Legio XIIII: see Webster in G. Webster (ed.), Fortress into City (1988), 123.

10 ILS, pp. 368–82. See also E. de Ruggiero, Dizionario epigrafico di antichità romane, IV, s.v. Legatus.

11 These details surely came from the legion's own records when Secundus was buried by his comrades. ANORV for annorum is not an oversight, but due to the tendency not to pronounce final -m and to confusion in writing geminated consonants: see Colin Smith, ‘Vulgar Latin in Roman Britain: epigraphic and other evidence’, ANRW 11.29, 893–948, at 918 and 925.

12 ILS, pp. 368–82. The emperor's name was sometimes used instead.

13 It should not be read on the altar from Carnuntum: see note 5. In theory this form would have occurred in proconsular Africa before the reign of Gaius, if the proconsul ever seconded a legionary to the staff of one of his legates.

14 The Romans in Britain (1st edn, 1932), 21 = (2 edn, 1969), 19. Burn qualifies pr(aetorii) as ‘possible’ and does not discuss it. I overlooked his prior suggestion when I wrote this paper, and it was brought to my notice by Mark Hassall.

15 ILS III, pp. 395–6 and 443–4 (of the Praetorian Cohorts, misleadingly indexed as praetorius by R. Cagnat, Cours d' Épigraphie Latine (1914), 454).

16 RIB 505; Britannia xix (1988), 490, No. 4Google Scholar. A third beneficiarius legati legionis of Legio XX is on record: AE 1951, 194 (Schallmayer, op. cit. (note 1), No. 867, Castelmadama, near Rome), the tombstone of Mommius Cattianus, mil(es) leg(ionis) XXV(aleriae) V(ictricis), benefic(iarius) leg(ati), cornicularius leg(ati), optio coh(ortis) I. The legate is obviously in command of the legion: compare ILS 2666b. Cattianus was then promoted twice within the legion, before his death a long way from Chester, apparently on detached service (see M.P. Speidel in ANRW 11.10, 854–5 = Mavors 1 (1984), 69–70). So it is likely that when he was a beneficiarius he was still at Chester. A fourth example depends on a baseless restoration of CIL VIII 27966 (Lambaesis), and is rightly not admitted by Schallmayer (No. 745).

17 CIL III. 3412 (ILS 2409). Schallmayer, op. cit. (note 1), Index at p. 810, collects twenty-four provenanced inscriptions attesting a beneficiarius legati legionis. Sixteen of these occur at the legion's headquarters, none of them at the headquarters of another legion, and none of them at the provincial capital unless the legion was also based there. Six of the other eight provenances are easily explained, and the remaining two were presumably the dedicator's native place, these inscriptions being dedications prompted by personal religiosity.

18 ILS 990, Cn. Domitius Lucanus (cos. c. 79), ‘praetorio legato provinciae Afr[i]c[ae] imp(eratoris) Caesar(is) Aug(usti)’.

19 ILS 991. Cn. Domitius Tullus, ‘qui cum esset candidatus Caesar(is) pr(aetor) desig(natus) missus est ab imp(eratore) Vespasiano Aug(usto) legatus pro praetore ad exercitum qui est in Africa at apsens inter praetorios relatus.’

20 CIL VI. 1388 (ILS 1090), the tombstone of Q. Cornelius Senecio Proculus (PIR2 C 1451), who was deputy-governor to his father, proconsul of Asia in c. 162. He was thus a legatus pro praetore provinciae Asiae.

21 HA Hadr 3. 9, ‘legatus praetorius in Pannoniam inferiorem missus; compare Pliny, NH XXVI.4, Manilium Cornutum e praetoriis legatum Aquitanicae provinciae’. The anonymous author of the ‘Augustan History’ probably found the term in his source: by his time, although consularis survived (in a new sense), the terms legatus praetorius and legatus consularis had long been obsolete. He knew, however, that legates were graded by the equivalent or the qualifying magistracy in Rome, since he parodies the terminology in Trig. Tyr. 33. 1 (the career of ‘Censorinus’), ‘quarto pro consule, tertio consularis, legatus praetorius secundo, quarto aedilicius, tertio quaestorius’.

22 Agricola 7.3. The term legatus praetorius is not used of Larcius Priscus in a dedication to be dated shortly before his consulship in 110, which recalls that he deputised as legionary legate for the consular legate of Syria: ‘legatus Augusti leg(ionis) III Scythicae pro legato consulare provinc(iae) Syriae’ (AE 1908, 237; repeated by ILS 1055). The reason for the omission is that he was not even an ex-praetor at the time, having been transferred to Syria in emergency from his previous post as quaestor of Asia: see R. Syme, Tacitus (1958), 16.

23 See previous note and ILS 1015, AE 1912, 5, and (in Greek) E. Ruggiero, op. cit. (note 10). 536–7; Suetonius, Vesp. 4.1. Tacitus (Agric. 8.2; 14) also uses the informal term consularis, in accord with contemporary usage: see Speidel, M.P., ‘The captor of Decebalus: a new inscription from Philippi’, JRS lx (1970), 142–53Google Scholar, esp. 151–3; A.K. Bowman and J.D. Thomas, Vindolanda: the Latin Writing-Tablets (1983), Nos 21.9, 30.5, 37.15 (c. AD. 103) with note. By now this usage was probably of long standing, to judge by P. Mich. 466 (dated 26 March 107), a letter written by a legionary in the new province of Arabia to his father in Egypt, which not only refers to the (praetorian) governor as τōν ὑπατιχōν λεγepsilon;ωνōς (i.e. the consularis) but refers to him in his capacity as legate of the province's one legion as τōν ὑπατιχōν της λεγεωνōς (i.e. the consularis legionis). I have not been able to consult Vidman in Studi in onore di Cesare Sanfilippo 1 (1982), 657–66, cited as AE 1982, 19, who apparently reaches a similar if less extreme conclusion by arguing that the terms consularis and legatus consularis are early-Trajanic if not Flavian in origin.

24 The Carnuntum inscription cited above (note 5) is the other instance of a beneficiarius legati praetorii. There is no obvious reason for Schallmayer's ‘third-century’ date. If this is inferred from the second-century division of Pannonia into a ‘consular’ (Upper) and a ‘praetorian’ (Lower) province (compare the dating of No. 342, below), then why is the ‘praetorian’ legate found at the capital of the consular province? The lettering of the altar, though irregular, contains no visible ligatures; and the name of the dedicator, M. Caelius Mansuetus, with its tria nomina and Italian nomen, also suggests an earlier date. By contrast, the b(eneficiarius) leg(ati) leg(ionis) from Chester is probably third-century (RIB 505, with note). Schallmayer, op. cit. (note 1), Index p. 810, collects five instances of an officer called beneficiarius legati consularis, but three are based on false inferences; there are in fact only two explicit instances of the term, Nos 342 (ILIug. 1. 329) and 895 (CIL V. 5451). No. 342 he dates to the end of the second-century/first third of the third century, but this is evidently inferred from its reference to a ‘consular’ governor, who is taken to be that of Pannonia Superior as distinct from the ‘praetorian’ governor of Pannonia Inferior; this inference is unnecessary, and the handsome lettering suggests a much earlier date. No. 895 he dates to the second half of the second century/the third century, but evidently on the assumption that L. Cominius Pollio miles leg(ionis) XIII Gem(inae) beneficiarius legati consularis was serving in Dacia, where it was necessary to distinguish the ‘consular’ governor. Again, this inference seems unnecessary: the man's nomen is Italian and, like the use of the praenomen, suggests an earlier date. The absence of abbreviation in his title is remarkable, and suggests that it was unfamiliar: the ubiquitous BF COS had not yet become standard.

25 From RIB 292 and 294 at least; RIB 296 is too fragmentary to tell.

26 PW XII, s.v. Legio, at 1731, citing CIL XI. 395 (ILS 2648). For the date of Boudica's defeat, see R. Syme, Tacitus (1958), 765. The form ‘Boudica’ is supported by Cassius Dio LXII.2.1 (etc.), and was established on philological grounds by Jackson, in Britannia x (1979), 255. Whether the title Martia Victrix was granted as early as 60 is unknown; the year 61 is assumed here, together with other consequences of the victory. For the legion's prestige see Tacitus, Hist. 11.11, ‘praecipui fama quartadecumani, rebellione Britanniae compressa. addiderat gloriam Nero eligendo ut potissimos’. V.16, ‘domitores Britanniae quartadecimanos.’CrossRefGoogle Scholar

27 McPake, R., ‘A note on the cognomina of Legio XX’, Britannia xii (1981), 293–5, citing RIB 200 and (by implication) CIL V. 533 (ILS 2702) (compare Tacitus, Ann. XIII.30). To these can be added AE 1924, 78 and 1966, 124.CrossRefGoogle Scholar

28 ibid. His purpose is rather to establish the latest date at which Legio XX did not bear the cognomina, and to prove that Valeria was not anterior to Victrix, than to establish the earliest instance of the cognomina. This indeed would still be difficult (see below, p. 154), even if McPake had not excluded evidence irrelevant to his main purpose, to judge by his comment (ibid., 293) that ‘it is only in Dio's review of the legions of the Empire that these titles are written out in full’. He is followed in this by Lawrence Keppie in The Making of the Roman Army (1984), 138–9, with 239–40, note 14, which cautiously cites only AE 1980, 445. True, the cognomina Valeria Victrix are always abbreviated in inscriptions from Britain, but there is ample epigraphic evidence elsewhere for them written out in full: see CIL 11. 3853; III. 186 (ILS 2657, in Greek); VIII. 1322; VIII. 14854 (ILS 2764); XIV. 4059; AE 1950, 251 (fragment only, in Greek).

29 Ancient Society 13/14 (1982/3), 275–6 = Mavors IV (1988), 282–3, citing CIL III. 11233, XIV. 3955 (ILS 2740) and JRS lii (1963), 190, No. 1. See also Frere, op. cit. (note 7), 63 and 79, note 24.

30 Interim reports of the excavations will be found in the annual Roman Britain surveys in Britannia, and the final report of the Annetwell Street excavations should soon appear. I will publish the writing-tablets there with photographs and commentary. For the interesting ‘notice board’, now blank, see Caruana, I., ‘A wooden ansate panel from Carlisle’, Britannia xviii (1987), 274–7. For stilus tablets from other sites, including Castle Street, see next note.CrossRefGoogle Scholar

31 ‘Addresses’ scratched on the outside tend to be all that is legible: see Britannia xix (1988), 496–8, Nos 30–35; xxii (1991), 299–300, Nos 24–5.Google Scholar

32 This is not a euphemism for rediscovery, it is literally true: a cramped excavation under the floor in advance of installing a new lift (see pp. 45–109, above). This tablet was found with the one addressed to a seplasiarius (see note 50) and other fragments, in silt filling a hollow in cobbles laid over a military ditch, the filling of which contained samian deposited in the early 80s.

33 Since Cassius Secundus' note of hand is incomplete, we cannot be entirely sure the debt was due to the loan of 100 denarii rather than the purchase-price of something, but a loan is almost certain, both from the natural meaning of debere (‘to owe’) and by comparison with the other texts. These are all papyri, with the exception of No. 2 (below). For instances of soldiers borrowing or lending money, see:

(1) Harrauer, H. and Seider, R., ‘Ein neuer lateinischer Schuldschein’, ZPE xxxvi (1979), 109–17Google Scholar, with further commentary by Shelton, in ZPE xxxviii (1980), 202Google Scholar, and by Gilliam, in ZPE xli (1981), 277–80. (Dated 28 August 27; an eques of an ala borrows 200 drachmas [50 denarii].)Google Scholar

(2) Unpublished stilus tablet fragment from Vindonissa, kindly communicated by Michael A. Speidel. (Dated 25 January 90; an eques of an ala borrows an unknown sum, probably from a legionary.)

(3) P. Mich. VII 438 (CPL 188), improved text in ChLA V, 50, No. 303. (Dated 8 August 140; an eques of a cohort borrows 79 denarii.)

(4) P. Fouad Ier 45 (CPL 189). (Dated July 153; an eques of an ala borrows 50 denarii.)

(5) P. Mich, VII 440 (CPL 190). (Fragment only, dated July/August 162.)

(6) Benoit, P., Milik, J.T., and Vaux, R. de, Discoveries in the Judaean Desert 11 (1961), 240–3, No. 114. (Fragmentary Greek text, probably dated 171. A legionary borrows the equivalent of 50(?) drachmas [12½ denarii], probably from a civilian.)Google Scholar

(7) P. Mich, VII 445 (CPL 194), improved text in ChLA V, 18, No. 284. (Dated 188; one legionary borrows from another.)

(8) P. Mich, VII 161 (CPL 128), improved text in ChLA V, 34, No. 294. (Fragment only, second-century; a marine(?) borrows from a colleague(?).)

(9) Priest, N., ‘A loan of money with some notes on the ala Mauretana’, ZPE li (1983), 65–70. (Greek text dated June/July 205; a priest borrows 116 drachmas [29 denarii] from a signifer of an ala.)Google Scholar

(10) ChLA XI, 4, No. 467. (Petition by a legionary veteran presumably to an officer, to compel a legionary to repay a loan of 100 denarii contracted when they were both serving soldiers.)

34 None of these fragments (from Tullie House and Annetwell Street) carries any visible trace of an inked text on the fore edge identifying the contents, as on the Vindonissa tablet (see note 33, No. 2); this practice is well attested at Pompeii.

35 It was found in a later (1989) excavation than those mentioned in note 30, and is unpublished and not yet fully deciphered. Both words are found in the Vindolanda tablets. There is apparently a reference to ‘ten military cloaks’ in one of the wooden leaf tablets from Annetwell Street (No. 31 in my report).

36 For fidepromissio see Gaius, Inst. III. 116–20. But compare Pippidi, D.M. and Russu, I.I., Inscriptile Daciei Romane 1 (1975), 229, No. 39 (CIL III, p. 945), ‘t(antam) p(ecuniam)… fide promisit Veturius Valens’ (the vendor's promise of reimbursement); ibid., 201, No. 33 (CIL III, p. 931), ‘s(upra) s(cripla) dari fide sua promisit lulius Alexander’ (a promise to repay 140 denarii borrowed, with interest).Google Scholar

37 Lines have been scored across the Vindonissa tablet (see note 33, No. 2) and Michael A. Speidel compares Hunt, A.S. and Edgar, C.C., Select Papyri 1 (1970), Nos 69, 74, and 75, etc.Google Scholar

38 See ILS Index v, Consules aliaeque anni delerminationes, especially Vol. III, pp. 326 ff.

39 H.M. Cotton and J. Geiger, Masada II: The Latin and Greek Documents (1989), 35ff., No. 722. R.O. Fink, Roman Military Records on Papyrus (1971), No. 37 (P. Gen. Lat. 1).

40 CIL XII.2602 (ILS 2118). The dates quoted are ‘imperial’ only in form (neither Domitian nor Nerva was yet emperor), but the intervening dates are genuinely so: Vespas(iano) X cos., Domit(iano) VIIII cos., Domit(iano) XIIII cos.

41 See note 33, No. 8.

42 op. cit. (note 11), 923.

43 See note 33, Nos 1, 3 (restored), 4, 5, and (in Greek ) 9.

44 ibid., No. 3.

45 ibid., Nos 1 and 3.

46 Tacitus, Agric. 38, exacta iam aestate… lento itinere. For the chronology see A.R. Birley, The Fasti of Roman Britain (1981), 77–9. The news of Mons Graupius soon followed Domitian's German triumph (Agric.39.1) and his assumption of the title Germanicus; the latter can now be dated between 9 June and 28 August 83: see T.V. Buttrey, Documentary Evidence for the Chronology of the Flavian Titulature (1980), 52–6.

47 op. cit. (note 9), 132; compare 20, fig. 1.2. See also P. Salway, Roman Britain (1981), 144. Strictly speaking, the document is proof only of the two legionaries’ presence in Carlisle, but since they presumably took part in the Mons Graupius campaign, it may be inferred that they had recently arrived here. That they had returned there, that Carlisle had been some sort of legionary base for years, is also a reasonable inference but needs to be confirmed archaeologically. See below, p. 152, and note 59.

48 op. cit. (note 7), 90.

49 In what follows I am excluding much of the epigraphic evidence for the Twentieth Legion at Carlisle, stone and tile mostly, since it is undated and may well be later than the first century: see Tomlin, R.S.O. and Annis, R.G., ‘A Roman altar from Carlisle Castle’, Cumb. Westm. lxxxix (1989), 77–91Google Scholar. To the evidence cited ibid., 87, add Britannia xvii (1986), 440–1Google Scholar, Nos 29–31 (stamped tiles); No. 26 (branded wooden bung, where LEG XX should probably be read); and RIB 2411.80 (lead sealing). Detachments of the Twentieth and the Second Augusta were brigaded together at Carlisle in the third century (art. cit., with Britannia xx (1989), 331, No. 4)Google Scholar. Second Augusta is also attested by a tombstone(?) (Britannia xix (1988), 492–3, No. 8Google Scholar), stamped tiles (Britannia xvii (1986), 441, Nos 32–3, etc.Google Scholar) and lead sealings (Britannia xxii (1991), 298, Nos 10–11). Lead sealings of Sixth Victrix have been found (ibid., 12–13), and stamped tiles of Ninth Hispana. RIB 944, 962, and 963 may also be legionary.Google Scholar

50 Britannia xxii (1991), 299, No. 24, where the meaning of seplasiarius is discussed. Compare CIL XIII. 6778 (Mainz), ‘seplasiar(ius) in leg(ione) I Ad(iutrice)’. But others were civilians, e.g. ILS 7606 (Cologne), ‘negotiator seplasiarius’.Google Scholar

51 Britannia xix (1988), 496, No. 31. It was originally the first leaf of a stilus-tablet letter with this address scratched on the outside, but later it was split into two halves, which were reused by being bound together as a ‘diptych’.Google Scholar

52 Bowman, A.K. and Thomas, J.D., ‘A military strength report from Vindolanda’, JRS lxxxi (1991), 62—73. This very important new document locates only 296 men of Coh. I Tungrorum at Vindolanda, less than half the unit, compared with 337 at Corbridge; and this fragmentation is beginning to look ‘relatively normal’ on the northern frontier in c. A.D. 90 (ibid., 68). There is also some evidence of legionaries at Vindolanda, perhaps belonging to Legio II Augusta (ibid.). The early tombstone of a legionary of the Twentieth at Carvoran (RIB 1826) is also interesting, but unfortunately it cannot be precisely dated; it must be one of the earliest instances of the cognomina Valeria Victrix. What these legionaries were doing is still unclear; whether they were building parties, or actual garrisons (in the sense of holding detachments which could be concentrated again as required).Google Scholar

53 Richmond, I.A., ‘Excavations at the Roman fort of Newstead, 1947’, PSAS lxxxiv (1949—1950), 1–38. The ‘Agricolan’ fort at c. 10.5 acres is unusually large, and it was enlarged to more than 14 acres after the withdrawal from Scotland in 86/7, but the relevant garrisons are not known. The altars of C. Arrius Domitianus, centurion of Legio XX, are not dated, but one (RIB 2123) came out of the well of the Antonine principia together with two boars in stone (emblem of Legio XX), and is thus presumably Antonine. The other centurion of Legio XX attested at Newstead, L. Maximius Gaetulicus, is certainly Antonine (compare AE 1985, 735). Richmond found a possible stable in the Flavian I fort and a possible pair of legionary barrack-blocks in the Flavian II fort, both of them buildings notoriously difficult to identify. Curie found gladii and an iron legionary helmet (Newstead, 164, and pi. xxvi) in ‘Flavian’ pits, as well as cavalry helmets and other cavalry equipment. However, the helmet is identified as ‘cavalry’ by Russell Robinson (The Armour of Imperial Rome (1975), 94–5), even if it is not clear why, and gladii were not exclusive to legionaries. [See further below, addendum at the foot of p. 158]Google Scholar

54 Compare Britannia xix (1988), 498, No. 33Google Scholar, addressed to Iulius Macrinus. There are some twenty-eight personal names attested in the wooden leaf writing-tablet fragments from Annetwell Street (sec note 30), as many as eight of them Iulii, including Iulius P[…] (No. 13), Iulius Apollofnius] (No. 36), and the recipient (No. 27) of the greeting quoted in the next note: ‘opto bene valeas, lulf’. For the lead tag of Iulius Suritus of the centuria Gabiana, presumably a legionary, see Britannia xx (1989), 334, No. 14.Google Scholar

55 No. 20 in my report (see note 30), a letter endorsed a Fulvio Natale collega; and No. 27, a letter concluding with the greeting ‘[saluta or similar] ∣ Pra[ese]ntem verbis meis et collegas ∣ om[ne]s. opto bene valeas, Iuli.’ Fulvius Natalis and Iulius were Roman citizens, and thus (if soldiers) are more likely to have been legionaries than auxiliaries; but note that there is only one other instance in Britain of the name Fulvius Natalis, also Flavian, which was borne by the joint-heir to a trooper in the ala Indiana who may have belonged to the unit himself (RIB 108). (The name is not distinctive enough to identify the two, but the possibility is there, especially since the ala Indiana was probably withdrawn from Cirencester in the 70s for service in the North.) There is a third reference to ‘colleagues’ in a fragment (No. 34), which reads: ‘[…] ∣ in modis quatt(u)or […] / iuss(us) cum colleg[is…],’; where orders are being given for concerted action. The term collega implies that both parties are of the same official or military rank: compare Livy vm.6.15 and other texts cited by me in my commentary to No. 20.

There are also three or four references to ‘brothers’ (fratres), like ‘colleague’ a term appropriate between members of the same unit. One (No. 37) may belong to a letter of commendation. Another (No. 23) contains greetings from Primus and Anoncletus to their ‘brothers’ Euphemius, Secundus and Rusticus: names of Greek etymology are also found in the Vindolanda tablets, but certainly do not suggest legionaries.

56 No. 3 in my report (see note 30), written across the grain (transversa charta) unlike the letters, and thus a military document: see Bowman and Thomas, op. cit. (note 23), 40. For a detailed discussion of the readings see my commentary, but domo B[…] in the first fragment, written […] .]domob[[…], looks like a note of a soldier's place of origin, and so perhaps do […]gerasa[…] and (next line) […]gera[ in the other fragment. The Misenum fleet sided with Vespasian in 69 (Tacitus, Hist. III.57ft.) and its survivors may have been rewarded by the chance of promotion to legionary service, like those of the Ravenna fleet (ibid., 50); if this occurred during the levy of recruits directed by the young Agricola (Tacitus, Agric. 73), they may have been used to reinforce the British legions (see Birley, op. cit. (note 46), 76–7) rather than incorporated in the two Adiutrix legions already organised. A veteran of the Misenum fleet discharged in 71 was a Syrian from Gerasa (CIL xvi .15 = ILS 1990), and presumably he had younger compatriots who opted for legionary service.

57 The full reading of this text, which may in fact describe him as a beneficiarius of Agricola, I intend to discuss elsewhere, together with the wooden leaf-tablet fragment apparently addressed to an eques singularis of Agricola.

58 Britannia xxi (1990), 320Google Scholar. E. Birley, Research on Hadrian's Wall (1961), 137, citing Bushe-Fox in Archaeologia lxiv (1913), 299 ffGoogle Scholar. Th e size of the fort is a verbal estimate by Ian Caruana, but see also Charlesworth in Hanson, W.S. and Keppie, L.J.F. (eds), oman Frontier Studies 1979 (1980), 210.Google Scholar

59 Ian Caruan a kindly sent me a copy of his draft report on the Tullie House excavation, in which this suggestion is made, but whether legionaries occupied distinctive barracks is still uncertain.

60 No. 19 in my report (see note 30); for the reading see note 57. The leaf-tablets contain two further references to ‘cavalry’, but neither of them necessarily refers to Carlisle: No. 21. a fragment of a letter endorsed ‘a Clementi[no] ∖ equit[e]’ and No. 33, a fragment reading ‘[…]vae turma S[…].’ Names of Celtic etymology also occur, and may have been borne by auxiliary cavalrymen: No. 24 was received by Verconus and Ollocus; No. 28 is endorsed ‘a Sufeva(?)’; and No. 38 mentions a Vocontius.

61 P.A. Holder, The Roman Army in Britain (1982), 109, but reading C1L XIII. 11709 and 6236; Tacitus, Hist. III.6.

62 ibid., 108; Tacitus, Hist. 1.70, IV.49. The earliest evidence is RIB 1172 (Corbridge), a late-Flavian tombstone of a signifer: see Birley, E., ‘Dalswinton and the ala Petriana’, Dumfriesshire and Galloway Trans, xxxv (1958), 9–13 = Mavors IV (198), 248–50; and note also the ‘early’ formulas Dis Manibus (unabbreviated) and h(ic) s(ilus) e(st), which hardly ever occur in inscriptions from northern Britain.Google Scholar

63 The provenance of RIB 1172 is beyond reasonable doubt, since it was found at Hexha m Abbey with other building stone from Corbridge, but it is possible the regiment was only in transit. However, the natural interpretation is that it was stationed there. There is no archaeological evidence of what the garrison of the primary (late-Flavian) fort was, but the size of the fort, though unknown, seems to have been much too large for a quingenary ala on its own. See M.C. Bishop and J.N. Dore, Corbridge: Excavations of the Roman Fort and Town, 1947-80 (1989), 129. A composite garrison is probable, in view of the evidence that 337 men of Coh. I Tungrorum were based there in c. A.D. 90: see Bowman and Thomas, op. cit. (note 52).

64 No. 32 in my report (see note 30), where the whole fragment will be published with photograph and commentary. This is much the earliest instance of cohortalis as a substantive.

65 Compare the new military strength report from Vindolanda (see note 52), which records a centurion(?) absent in London, and forty-six men serving as singulares leg(ati). Three lead sealings of the p(edites) s(ingulares) c(onsularis) have been found at Carlisle: RIB 2411. 92 and 93; Britannia xix (1988), 496, No. 28.Google Scholar

66 ‘cum adsuetus expeditionibus miles otio lasciviret’ (Tacitus, Agric. 16.3).

67 The same sura is borrowed by another legionary: see note 33, No. 10 (after the mid-second century). But except for ibid, Nos 6 and 7 (sura now lost), the other debtors are auxiliary soldiers, and the sums they borrow are not uniform.

68 Cassius Dio LXVII. 3.5, as preserved by Zonaras. One of them has confused ‘drachmas’ with ‘denarii’; presumably not Dio, since he had first-hand experience of legionaries and denarii. Fink, op. cit. (note 39), 241-5, and Nos 71 (1 May) and 72 (1 January).

69 Suetonius, Domitian 7.3, taken by Brunt in PBSR xviii (1950), 54Google Scholar, as a careless way of expressing ‘the equivalent of a fourth stipendium’. For a literal interpretation, see Watson, in Historia v (1956), 332–40. His best evidence is the fragmentary line 27 in Fink, op. cit. (note 39), No. 69, which implies a fourth annual payment of 297 drachmas [74¼ denarii]; but see Fink's commentary ad loc. on this difficult text.Google Scholar

70 Tacitus, Agric. 39.1. See note 46.

71 The unique sestertius of 84 published by Kraay, in American Numismatic Society Museum Notes ix (1960), 114–16, alludes to the pay-rise with its adlocutio reverse captioned STIP IMP AVG DOMITIAN, but does not resolve the question of whether there was an extra stipendium or simply enlarged stipendia. Nor does it date the pay-rise precisely. Since there was no aes coinage struck for 83, and the mint was ‘inactive’ after mid-83 (Buttrey, op. cit. (note 46), 55), the pay-rise could not be celebrated in that year. Perhaps the sestertius marked the first payment, not in August 83 (too early), but in January 84.Google Scholar

72 Fink, op. cit. (note 39), No. 117, i 8, with discussion on p. 266.

73 As published in RIB (for which see Roger Goodburn's exemplary Epigraphic Indexes (1983), 42), in JRS from 1956, and in Britannia from 1970. The doubtful cases are RIB 1779 and Britannia xix (1988), 492, No. 11Google Scholar. The cognomina were inscribed on the Gloucester tombstone, Britannia xv (1984), 333, No. 1: see xvii (1986), 454, corrigendum (e). The cognomina are also missing from RIB 1125 as published, but it is clear from the drawing (and Haverfield's reading) that the bottom half of the inset panel is badly worn; the letters VV should obviously be restored in the space available at the bottom right.Google Scholar

74 RIB 853. It is not a ‘centurial’ stone.

75 RIB 507 (Chester).

76 RIB 122 and 160 are incomplete, but otherwise it is certain these epitaphs all either began with Dis Multibus in full (RIB 158) or lacked it entirely, and concluded with h(ic) s(itus) e(st). The assumed movements of the legions are those summarised in Webster, op. cit. (note 9), 19–21. Here and elsewhere I am assuming that Legio XX abandoned Inchtuthil and Wroxeter, and replaced Legio II Adiutrix at Chester, immediately after 86: see Hobley, A.S., ‘Numismatic evidence for post-Agricolan Scotland’, Britannia xx (1989), 69–74. The two Bath tombstones, RIB 158 and 160, belong most readily to when Legio XX was stationed nearby at Gloucester; but see further, note 81.CrossRefGoogle Scholar

77 I have limited myself to inscriptions published in CIL or A E, but even this coverage is incomplete. Texts may be misunderstood (e.g. CIL XIII. 8286, which must be Legio XXX), the indexing is sometimes inadequate, and there are too many doubtful cases to be worth discussing here. But these inscriptions do amount to a large random sample.

78 See note 27.

79 (1) CIL XIII. 1900 (ILS 7025), dated by Devijver (Prosopographia Militiarum Equestrium s.v. V 1) to the end of the first century. No closer dating seems possible. (2)AE 1951, 194 (see note 16). The excellent lettering and the introductory D(is) M(anibus) s(acrum) coupled with h(ic) s(itus) e(st) suggest an early second-century date. (3) The other four inscriptions are cited in notes 92-5 and discussed below.

80 CIL in. 5577; V. 6632; XII. 678; XIII. 6780.

81 The only epigraphic reason for dating them to before 61 is the absence of the legion's cognomina. Historically such an early date is attractive, since the legion was then nearby at Gloucester. But they are no earlier in formulation than the other Bath tombstone of Legio XX, which includes the cognomina. This is RIB 156, which lacks any introductory formula and concludes with h(ic) s(itus) e(st). It shares these criteria with the legion's earliest tombstones from Chester, RIB 493, 498, 501, 502. But unlike them, it does not abbreviate stipendiorum, an ‘early’ feature found at Chester only in tombstones of the previous legion, Legio II Adiutrix, and elsewhere in Britain only in RIB 122 (Gloucester) and 2213 (Ardoch, but Flavian), a total of seven instances out of fifty references in RIB to ‘years of service’. For these reasons, RIB 156 is surely to be regarded as first-century, despite the problem of the man's origin and its formulation (natione Belga), and the much later date they suggest to the practised eyes of G. Forni (Il reclutamento delle legioni da Augusto a Diocleziano (1953), 190) and J.C. Mann (Legionary Recruitment and Veteran Settlement during the Principate (1983), 90).

Apart from this question of cognomina, and the fact that Gloucester is closer to Bath, there would be no difficulty in dating RIB 158 and 160 to the Wroxeter period (66—86). RIB 158 was found with RIB 157, which must have been erected during 71—86, since it is a tombstone of Legio II Adiutrix (then at Lincoln or Chester). Legionaries from Wroxeter seem to have visited Bath: see Woodfield, P., ‘An intaglio from Wroxeter’, Britannia vii (1976), 284–5.CrossRefGoogle Scholar

82 See note 26.

83 Except at Mons Graupius, where the legions all played a passive role, it is impossible to penetrate the generalities of Tacitus’ Agricola to the actual fighting, so we are reduced to surmise. The Ninth Legion was unlucky as usual (ibid., 26), but we do not know whether the Twentieth was among its rescuers. Was southern Scotland won, like Mons Graupius, ‘with typical Roman efficiency; no more, no less’ (W.S. Hanson, Agricola and the Conquest of the North (1987), 139), or did one of the legions truly distinguish itself? More importantly, did Domitian believe it had, or wish it to be thought he did? Nero's favour to Legio XIIII Gemina had ensured its lasting loyalty to him (Tacitus, Hist. 11.11), and Domitian may have made the same calculation. Conversely he executed a governor of Britain for tampering with the army's loyalty, if that is the sense of the anecdote in Suetonius, Domit. 10.3: see Birley, op. cit. (note 46), 82–3. Perhaps the title Valeria Victrix was even a consolation prize to the legion for evacuating Inchtuthil.

84 A.L.F. Rivet and Colin Smith, The Place-Names of Roman Britain (1979), 499. Mark Hassall and others would rather see it as the site of a battle, perhaps Mons Graupius itself. That Victoria derives from Valeria Victrix is unlikely: one would have expected Victricensis.

85 CIL III. 11929, 11932, etc., as McPake points out (see note 27), but with different emphasis. The first-century station of the unit seems to be unknown.

86 RIB 493, 495, 498, 500, 508, 510.

87 For Bath (RIB 156) see note 81. For Carvoran (RIB 1826) see note 52.

88 Britannia xvii (1986), 429, No. 3; discussed by Hurst in Webster, op. cit. (note 9), 70.Google Scholar

89 A stone bath-house or similar structure has to be hypothesised, an isolated provenance much less likely than ‘a general period of stone buildings’, as Hurst observes, ibid, (note 88).

90 The only other ‘centurial’ stone from Gloucester (Britannia iii (1972), 353, No. 5), the legion's name not stated, almost certainly came from the defences.Google Scholar

91 The problem remains of what Legio XX (part of it, at least) was still doing at Gloucester, whether refurbishing an obsolete base or building the colonia. Hurst repeatedly insists on the primacy of the archaeological evidence: see Webster, op. cit. (note 9), 48, 54–6, 70–71.

92 CIL XIV. 2523 (ILS 2662). M. Durry, Les Cohortes Prétoriennes (1938), 198 with pl. IV. His decorations (torques, armillae, phalerae, but no crown) suggest the reign of Domitian, but this would have been during earlier service in another legion as a centurion: see V.A. Maxfield, The Military Decorations of the Roman Army (1981), 192 and pl. 12a.

93 CIL III. 12411 (ILS 2666b). Maxfield, op. cit. (note 92), 191—3. His epitaph, which begins D(is) M(anibus) and concludes h(ic) s(itus) e(st), looks early second-century. This would suit a late-Flavian date for his service in the Twentieth, since he was then transferred to the Ninth Legion and was honourably discharged before returning home to the Danube; although it had been a long career, beginning in the ranks with two posts before the centurionate, he presumably survived some years before his death aged 75 (an obvious estimate).

94 CIL VI. 3584 (ILS 2656). Maxfield, op. cit. (note 92), 191—2. The similarity of the two careers, Proclus and Vitalis, prompted Birley's suggestion (Carnuntum Jahrbuch viii (1963/64), 30 = Mavors IV (1988), 216—17) that they reflect a movement of troops from the Danube to Britain. This would help to date their service in the Twentieth (not before the 90s in any case), but raises other difficulties: see Maxfield, loc. cit., and Strobel, K., ‘Bemerkungen zur Laufbahn des Ti. Claudius Vitalis’, Tyche ii (1987), 203–9.Google Scholar

95 ILS 9200. Kennedy, D., ‘C. Velius Rufus’, Britannia xiv (1983), 183–96.CrossRefGoogle Scholar

96 In 84, not 83, to allow time for news of Mons Graupius to reach Rome, for Domitian to make the decisions it entailed (that the conquest of Britain was complete, that it should be consolidated, that Agricola should be recalled, that he and his army should be honoured), and for his decisions to be transmitted to Britain. After 7 November 83, at least. The epigraphic evidence would even accommodate a date as late as 86/7, if Valeria Victrix were a consolation prize to an angry legion (see above, note 83). But this is only a guess.