Hostname: page-component-5d59c44645-zlj4b Total loading time: 0 Render date: 2024-02-24T17:34:08.686Z Has data issue: false hasContentIssue false

The Poets Of Bu Njem: Language, Culture and the Centurionate*

Published online by Cambridge University Press:  14 March 2012

J. N. Adams
All Souls College, Oxford


A good deal has been written about the origin and recruitment of centurions, but their language use has not been analysed as an index to their literacy, culture, educational levels, and (in some cases) bilingualism. This paper will be about the sociolinguistics of the centurionate. I take as my starting point two poems from Bu Njem in the African desert, one of which has only recently been published for the first time.

The military outpost of Bu Njem (Golas, Gholaia) lay 200 km south of Cape Misurata in the desert of Tripolitania. The fort, the construction of which began some time after 24 January 201, was garrisoned by a uexillatio of the legio III Augusta, and later by a numerus collatus. From here there survive ostraca of various kinds dated mainly to the 250s, but in this paper I am going to deal with the unlikely topic of poetic activity within the camp. Curiously, we have poems set up on stone in the name of two centurions, both of them acrostichs which spell out the centurions' names. The first, in iambic senarii and dated 202–3, has the name of Q. Avidius Quintianus, the second, dated to early 222, that of M Porcius Iasucthan; Iasucthan is of Libyan origin. This poem, on a charitable view, is composed in hexameters. The poems provide remarkable evidence for the cultural and linguistic level of centurions, if one can make the assumption that the texts were the responsibility of the two centurions themselves.

Copyright © J. N. Adams 1999. 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.)


1 See, for example, Le Bohec, Y., La troisième légion Auguste (1989), 441, n. 443.Google Scholar

2 See Marichal, R., Les ostraca de Bu Njem, Suppléments de ‘Libya Antiqua’ VII (1992)Google Scholar; also Adams, J. N., ‘Latin and Punic in contact? The case of the Bu Njem ostraca’, JRS 84 (1994), 87112Google Scholar.

3 This poem has been comprehensively discussed by Rebuffat, R., ‘Le poème de Q. Avidius Quintianusa à la déesse Salus’, Karthago 21 (1987), 93105.Google Scholar Note too the earlier discussions by Lavagnini, B., ‘Epimetron. Il centurione di Bu Ngem (Q. Avidius Quintianus)’, RFIC NS 6 (1928), 416–22Google Scholar, and particularly Kroll, W., ‘Metrische lateinische Inschrift aus Tripolitania’, Glotta 19 (1931), 151–2Google Scholar. See also now Courtney, E., Musa Lapidaria. A Selection of Latin Verse Inscriptions (1995), no. 40.Google Scholar

4 For the editio princeps of the poem, see Rebuffat, R., ‘Le centurion M. Porcius Iasucthan à Bu Njem (Notes et documents XI)’, Libya Antiqua NS 1 (1995). 79123Google Scholar.

5 For the present participle in building inscriptions, see (e.g.) RIB 1060, 1091, 1202, 1235. I owe these references to I. Mednikarova.

6 For an example at the Colossus of Memnon, see A. and É. Bernand, Les inscriptions grecques et latines du Colosse de Memnon (1960), no. 13.

7 Agens prae cunctos is difficult (see Rebuffat, op. cit. (n. 3), 95–6). Lavagnini and Kroll thought that prae cunctos was a mistake for praecinctos, and this view has its attractions, particularly since it is now clear that the verb could mean ‘enclose, protect’: TLL X.2.438.7ff. (‘-untur ea, quae munitionibus instruuntur, firmamentis teguntur sim.’; l. 42 ‘multitudo animantium’ (sc. praecingitur)).

8 Courtney, op. cit. (n. 3), 264, construes ll. 1 and 3–4 as follows: ‘quaesii quod uotum commune memoriae traderem proque reditu redderem’.

9 Inter priores et futuros is obscure. For the possibilities see Rebuffat, op. cit. (n. 3), 96–7.

10 The antecedent of quem is either nomen et numen or deae; the agreement (of gender) is wrong on either interpretation.

11 I take solis flammas feruidas as object of the (hitherto) intransitive verb nuto (see below). An alternative possibility, adopted by Courtney, op. cit. (n. 3), 264, no. 11, following a suggestion of Kroll, would be to make corpora (13) subject of delenirent, in which case flammas feruidas would become object of delenirent (= ‘so that their bodies might ease the fiery flames of the sun by swimming’). This interpretation I find unsatisfactory for various reasons: (i) the object of delenio is characteristically a person or part of a person or a condition or feeling suffered by a person, rather than an external force (TLL V.1.432.74ff., 433.25ff.); corpora is therefore a much more appropriate object of delenirent (with the soldiers as subject) than is feruidas flammas; (ii) the instrumental nando, juxtaposed with delenirent, would harmonize far more satisfactorily with a personal subject of the verb than with corpora, if that were subject (a point made by Rebuffat, op. cit. (n. 3), 99); (iii) if corpora is subject of delenirent and feruidas flammas object, a comma must be placed after austri, and ut therefore postponed to fifth position in its clause; the strong break in the middle of l. 12 would be unparalleled in the piece. If, on the other hand, a comma is placed after feruidas, ut neatly occupies second place in its line and clause, and l. 12 forms an uninterrupted colon.

The meaning of nutantis is far from clear (for some speculations, see Rebuffat, op. cit. (n. 3), 99), the problem no doubt partly created by the needs of the acrostich. Courtney translates ‘in these ever sandy dunes of the south’ ( = nutantis austri), claiming that ‘the verb can mean “inclining towards the south”’ and citing Manil. 2.906. But the parallel is far from exact, and it seems to me, as it has seemed to others, that a reference to the Scirocco would be in order here, which would entail taking auster in reference to the wind rather than merely to the south.

12 I take this line (with Kroll, op. cit. (n. 3), 152) as explanatory Of facti in l. 14.

13 On the absence of an antecedent (ei, eius) here, see below, p. 125.

14 Deuotissimi is probably nominative plural, despite the fact that the nearest nominative is singular (Flauius Sossianus): therefore a constructio ad sensum, its number determined by the association of the uexillatio with Sossianus (see Rebuffat, op. cit. (n. 4), 87).

15 A cliché which the writer would have seen in other building inscriptions: cf., e.g. CIL III.4726, 4796, 5797 (all with the spelling conlabs-); for a literary example, see Suet., Cal. 21, and on inscriptional examples, see TLL III. 1573. 45ff.

16 Tertia Augustani is an odd phrase on which Rebuffat's note (op. cit. (n. 4), 88) is somewhat astray. He states: ‘La forme tertia augustani est assurée par le Code Théodosien, IV, 12, 3 en 321: tertiis augustaniis [sic]. En revanche, la Notitia Dignitatum, Occ. 5, 254 et 7, 251 donne tertio augustani, mais peut-être est-ce une erreur de la tradition’. But the example from the Theodosian Code is not a parallel for tertia Augustani, and I do not believe that tertio Augustani is an error. If tertia (sc. legio) were personalized into a (collective) plural signifying the members of the Third Legion, the form of the (substantivized) adjective would become tertiani, and similarly Augusta would become Augustani on the same principle. Logically tertia augusta would produce tertiani augustani, and such names are attested in the Notitia Dignitatum (e.g. Occ. 7.235 secundani Italiciani for secunda Italica; cf. Occ. 7.84 with Seeck ad loc). tertii Augustani is an illogicality, tertio Augustani on the other hand could be explained as a compound of later type with linking morpheme -o, as for example in mulo-medicus (for which class of compounds, see Leumann, M., Lateinische Laut- und Formenlehre (7th edn, 1977), 390)Google Scholar. In the elliptical expression tertia Augusta (with legio understood) tertia would have been felt as adjectival, with Augusta substantival, and that is why a compound may have emerged, tertia Augustani on the other hand seems to be a conflation of tertia augusta and tertiani Augustani.

17 On creto, see below, p. 121.

18 For an alternative interpretation of this line, see below, p. 123.

19 There is a parallel to this use of compendium in a similar context at Apul., Mund. 27 ‘an non eiusmodi conpendio machinatores fabricarum astutia unius conuersionis multa et uaria pariter administrant?’ (‘is it not thanks to an economy of effort of this kind that inventors of machines employing the ingenuity of a single rotation operate at the same time many diverse things?’): cited at OLD s.v. 5 with the meaning ‘compendious device’. In both passages the sense hovers between abstract and concrete, with a type of machine signified in the concrete meaning.

20 The compound delonge is attested particularly in the Old Latin Bible translations, but occasionally also in technical texts (e.g. Vegetius, the Latin Dioscorides, Anthimus): see TLL V.1.469.39ff. It was excluded from high literature.

21 On this word see Rebuffat, op. cit. (n. 4), 91. It may be added that the Romance evidence (OFr. arche, Opr. area) shows that arcus developed a feminine by form *arca (see von Wartburg, W., Französisches etymologisches Wörterbuch I (1928), 130Google Scholar), of which arcata seems to be a derivative; the derivative in-atus of arcus is arcuatus (see TLL, s.v.). arcata too is reflected in Romance (OPr. arcada ‘arche d'un pont; arcade’ (v. Wartburg)). Though this last is feminine, the substantivized adjective in Latin (Iasucthan) is likely to be neuter (plural). Similarly arcuatus was used as a noun in the neuter (TLL 11.374.3ff.), = arcus. *arcatum thus corresponds to arcuatum, but with a root of different gender. The term in Iasucthan could thus be an early anticipation of two Romance outcomes.

22 animaduerto is complemented by a quod-claus instead of the acc. + inf., in the typical late manner: cf. Sidon., , Epist. 1.2.4Google Scholar.

23 titulis ornare may well have been a familiar collocation (cf. CLE 511), but with titulis bearing a concrete sense (‘inscription’: OLD s.v. 2). Here the word seems to be more abstract (cf. perhaps OLD s.v. 7b ‘(w. gen. of cause or defining gen.) the distinction, honour (arising from, consisting in)’, uirtutis deuotionis seems to form an asyndeton bimembre.

24 See below, p. 119.

25 See below, p. 122.

26 A text with translation and commentary can now conveniently be found in Courtney, op. cit. (n. 3), no. 34.

27 See in general Courtney, E., ‘Greek and Latin acrostichs’, Philologus 134 (1990), e.g. 7, 1113CrossRefGoogle Scholar.

28 For some interpretations of f.c.mac., see Rebuffat, op. cit. (n. 4), 98.f. is taken as = fecit, c. = cum, cura or curante, and mac. = magistro or magistri. It seems perverse though not to take f.c. as standing for faciendum curauit, especially as Avidius wrote centurio leg…faciendum curauit to accompany his verses, mac. may well not be part of the acrostich. It is not an obvious abbreviation of magister, though it could be justified. It is incidentally likely that the first five lines, with their accumulation of official material and titles, come from a different source from the ‘poem’ proper, possibly from Sossianus himself or from an official examplar.

29 Rebuffat, op. cit. (n. 4), 114, scans fieri as - -.

30 The poems which Courtney, op. cit. (n. 3), 22; cf. 257, singles out from his selection (nos 34, 117; for the first, see above, and for the second, from Venafrum, see also CLE 1319 = ILS 5150) as so defective in metre that an analysis of their faults is pointless are more correct than the poem of Iasucthan.

31 Nicolau, M., ‘Les deux sources de la versification latine accentuelle’, ALMA 9 (1934), 65.Google Scholar

32 See Koller, H., ‘Die Silbenquantitäten in Augustinus’ Büchern De Musica’, Mus. Helv. 38 (1981), 262–7Google Scholar.

33 See e.g. W. S.Allen, Accent and Rhythm. Prosodic Features of Latin and Greek: a Study in Theory and Reconstruction (1973), 80–2; for the phenomenon in the history of Latin, see M. Leumann, op. cit. (n. 16), 55, 252–3.

34 For the date of Commodian, see Krestan, L., ‘Commodianus’, RAC 3 (1957), 248Google Scholar; Der Kleine Pauly 1 (1964), 1260–1; Herzog, R. and Schmidt, P. L., Handbuch der lateinischen Literatur der Antike IV (1997), 629.Google Scholar

35 See further Nicolau, op. cit. (n. 31), 59, Leumann, op. cit. (n. 16), 252. On Commodian's versification, see also Perret, J., ‘Prosodie et métrique chez Commodien’, Pallas 5 (1957), 2742CrossRefGoogle Scholar.

36 On the name, see Rebuffat, op. cit. (n. 4), 97.

37 Shipley, F. W., ‘Carmina Epigraphica and some problems of the Latin hexameter’, PAPA 58 (1927) xxx–xxxiGoogle Scholar, similarly notes that the uneducated would only obtain an impression of ‘the more obvious and salient features’ of the hexameter as they heard it read. His classification of defective hexameters (some work at the beginning as well as the end, but others only at the end, with the middle more or less unknown territory) is useful, and brings out well the feature (the rhythm of the line ending) which was the most salient to an uneducated reader/listener of the period when phonemic distinctions of vowel quantity had been lost.

38 See Rebuffat's useful collection (with errors marked) of imperfect African carmina, op. cit. (n. 4), 116–19.

39 From Rome, note for example the epitaph of Allia Potestas(cf. Horsfall, N., ‘CIL VI 37965 = CLE 1988 (Epitaph of Allia Potestas): a commentary’, ZPE 61 (1985), 251–72Google Scholar), which has a few shortenings in final syllables (ll. 14, 40); for another Roman example, see CLE 1150.3 (spirito); cf. Galletier, É., Étude sur la poésie funéraire romaine d'après les inscriptions (1922), 303Google Scholar. For Gaul, , see CLE 465.3Google Scholar (uixi), 765.2 (praecepit), 769.9 (cum uerecundia); for Spain, , CLE 541.8Google Scholar (paruolo; c. third-century according to Buecheler); and for Moesia (Nicopolis), CLE 492.2 (inclusae), 19 (nobili), 20 (carmini). See further Galletier, op. cit., 302–4, Courtney, op. cit. (n. 3), 25. It should, however, be noted, first, that there had been a long history of vowel weakening (including shortening) in final syllables in Latin (see below), and the occasional shortening of long vowels in this position is of a different order from the lengthening of short vowels under the accent (on the difference between shortening in final syllables, and lengthening under the accent, see L. Mueller, De re metrica (2nd edn, 1894), 442). Secondly, many poems which allow such licences from time to time otherwise have substantial sections which scan correctly. The degree of error in Iasucthan cannot be exaggerated.

40 On apocope of final -e, see, e.g. Drexler, H., Einführung in die römische Metrik (1967), 58–9.Google Scholar

41 See Leumann, op. cit. (n. 16), 107–10.

42 See Leumann, op. cit. (n. 16), 110; see also Fitch, J. G., ‘Sense-pauses and relative dating in Seneca, Sophocles and Shakespeare’, AJP 102 (1981), 303–5Google Scholar (on the increase of -ǒ for -ō in the later plays of Seneca); Horsfall, N., ‘Criteria for the dating of Calpurnius Siculus’, RFIC 125 (1997), 176–7Google Scholar.

43 See Kaster, R. A., Guardians of Language: the Grammarian and Society in Late Antiquity (1988), 352–3Google Scholar.

44 See Leumann, op. cit. (n. 16), 56, 252–3.

45 CLE 44 = CIL IV.5092 (Vēnerem, ûbi); see Leumann, op. cit. (n. 16), 56, Väänänen, V., Le latin vulgaire des inscriptions pompéiennes (1966), 19Google Scholar.

46 Herman, J., ‘Un vieux dossier réouvert: les transformations du système latin des quantités vocaliques’, BSL 77 (1982), 285302Google Scholar, = idem, Du latin aux langues romanes. Études de linguistique historique (1990), 217–31.

47 Reservations must also be expressed about the unrefined statistical comparison of ‘Roman’ with ‘African’ carmina, without attention to the possible authorship of each individual poem. A poem found in Rome may in fact have been composed by an outsider. We will discuss below (p. 129) a ‘Roman’ carmen which in fact bears the name of a Thracian.

48 For his few abnormalities of scansion, see Kroll, op. cit. (n. 3), 152. harenacis (11) has to be scanned with a long first a.

49 See above, n. 38.

50 The language of the Vindolanda writing-tablets: an interim report’, JRS 85 (1995), 87–8Google Scholar.

51 See Adams, op. cit. (n. 2), 103.

52 See Adams, op. cit. (n. 2), 96, 103.

53 lab- for lap- also represents an attempted recomposition which, even if it is a mistake, at least superficially reveals the writer's ability to associate the participial form with the present stem of the verb. But since the spelling is common in this verb in inscriptions recording building repairs (see n. 15), it could in this case merely have been copied from other inscriptions which the writer had seen.

54 See, e.g., W. D. Elcock, The Romance Languages (1960), 105, 145.

55 It reflects the confusion of consonant- and i-stem nouns. For some evidence, see Adams, op. cit. (n. 50), 99–100.

56 Harrison, S. J., Vergil Aeneid 10 (1991), 178Google Scholar ad loc.

57 See Mattingly, H., Sydenham, E. A. and Sutherland, C. H. V., The Roman Imperial Coinage IV.2 (1938), 159, 161Google Scholar. I am grateful to P. M. Brennan for suggesting this line of approach to the problem.

58 See Mattingly, and Sydenham, , The Roman Imperial Coinage III (1930), 393Google Scholaruirtus aeterna Aug. (A.D.192).

59 See Bowman, A. K. and Thomas, J. D., The Vindolanda Writing-Tablets (Tabulae Vindolandenses II) (1994), 66Google Scholar, for a collection of such lines; to these should be added the expression conticuere omnes, found at Silchester.

60 See TLL V.1.879.19ff. (a relatively late usage, quoted first from Tertullian).

61 facio had long shown a tendency to replace more precise verbs in colloquial or substandard writings (see Hofmann, J. B. and Szantyr, A., Lateinische Syntax und Stilistik (1965), 755Google Scholar). Examples abound, for instance, in the Peregrinatio Aetheriae. In one chapter (20) chosen at random note facto ibi triduano, facta est oratio, uirtutes faciant multas, fecimus …biduum, mirabilia fecerint, quae hodie faciant.

62 Rebuffat, op. cit. (n. 4), 95, states with reason that dictatores ‘est évidemment ici une expression qui désigne les chefs, le centurion et le magister’, and goes on to remark that attestations of the word which do not designate the Republican magistrate are extremely rare. In my opinion it is a mistake to attempt to identify here any sort of technical use of the word belonging to the Empire. The usage embodies soldiers' humour, and is a nickname given by them to the overseers of the work. Bits and pieces of soldiers' wit are preserved (see Heraeus, W., ‘Die römische Soldatensprache’, ALL 12 (1900), e.g. 265Google Scholar), including a few nicknames (Heraeus, 278). One of these last, recorded by Tacitus at Ann. 1.23.3, was applied by soldiers to a sadistic centurion (cedo alterant). One of the non-technical examples of dictator cited at TLL V.1.1003.61 (Fronto p. 226.23N.) is described there as used ‘per iocum’.

63 See, e.g., Löfstedt, E., Philologischer Kommentar zur Peregrinatio Aetheriae (1911), 88Google Scholar.

64 See Fischer, B., Novae Concordantiae Bibliorum Sacrorum iuxta Vulgatam Versionem critice editam (1977)Google Scholar, citing more than fifty examples.

65 See Adams, op. cit. (n. 2), 96–103.

66 See Bernand, É., Inscriptions métriques de l'Égypte gréco-romaine. Recherches sur la poésie épigrammatique des grecs en Égypte (1969), 598CrossRefGoogle Scholar. The text is printed as no. 168 in Bernand.

67 See ll. 24–5, and on their interpretation, Bernand, op. cit. (n. 66), 606–7.

68 See Wölfflin, E., Ausgewählte Schriften (1933), 268.Google Scholar

69 See TLL V.1.1652.31ff.

70 See Hofmann-Szantyr, op. cit. (n. 61), 539; Norberg, D., ‘Zura Infinitiv in lat. Frage- und Relativsätzen’, Glotta 27 (1939), 261–70Google Scholar, esp. 268.

71 Hofmann-Szantyr, op. cit. (n. 61), 539.

72 Hofmann-Szantyr, op. cit. (n. 61), 440.

73 See Hofmann-Szantyr, op. cit. (n. 61), 555–6.

74 See in general Dahlén, E., Études syntaxiques sur les pronoms réfléchis pléonastiques en latin (1964)Google Scholar.

75 502 ‘qui et sibi quidem post unam horam sani fiunt’ See Ahlquist, H., Studien zur spätlateinischen Mulomedicina Chironis (1909), 34Google Scholar and Dahlén, op. cit. (n. 74), 120 (this last cited by Courtney, op. cit. (n. 3), 265).

76 See Hofmann-Szantyr, op. cit. (n. 61), 94, Lindsay, W. M., Syntax of Plautus (1907), 41Google Scholar.

77 See the detailed discussion of Le Bohec, op. cit. (n. 1), 147–84, with the general remarks at 148 on methods of recruitment to the centurionate.

78 See Bowman and Thomas, op. cit. (n. 59), 224, 227.

79 See Dobson, B., ‘The centurionate and social mobility during the Principate’, in Nicolet, C. (ed.), Recherches sur les structures sociales dans l'antiquité classique (1974), 99116, especially 101.Google Scholar

80 See Plb. 6.34.7–12; Best, E. E., ‘The literate Roman soldier’, Cj 62 (19661967), 123Google Scholar; Harris, W. V., Ancient Literacy (1989), 166–7Google Scholar.

81 On this inscription, see Pikhaus, D., ‘Le carmen de Cillium et l'épigraphique versifiée de l'Afrique romaine’, in Les Flavii de Cillium. Étude architecturale, épigraphique historique et littéraire du mausolée de Kasserine (CIL VIII, 211–216), Collection de l'École Française de Rome 169 (1993), 149–51Google Scholar.

82 See also Der Kleine Pauly 3 (1969), 626Google Scholar S. V. librarius.

83 For this connection drawn between the career of Petronius and the passage of Vegetius, see Birley, E., ‘Promotions and transfers in the Roman army II: the centurionate’, The Roman Army Papers 1929–1986 (1988), 208–9.Google Scholar

84 The post of trecenarius was of high status, but its exact nature is unclear. See especially Dobson, B. and Breeze, D. J., ‘The Roman cohorts and the legionary centurionate (Appendix II: trecenarius and princeps castrorum)’, Epigraphische Studien 8 (1969), 118–22Google Scholar.

85 See Donat., , GL IV.368.7Google Scholar.

86 See Kay, N. M., Martial Book XI: a Commentary (1985), 64 ad loc.Google Scholar

87 For centurions as falling short of the highest social class, see further Val. Max. 3.8.7 with Skidmore, C., Practical Ethics for Roman Gentlemen. The Work of Valerius Maximus (1996), 103CrossRefGoogle Scholar. I am grateful to T. P. Wiseman for these references.

88 The inscription is discussed from a different point of view by Le Bohec, Y., L'armée romaine sous le hautempire (2nd edn, 1990), 252Google Scholar; for bibliography, see also Pikhaus, op. cit. (n. 81), 136 n. 12, 138 n. 18.

89 Note, for example, the framing of each line by verbs, the hyberbaton in most lines, including the separation of primi from pili in a poetic manner (cf. Mart. 1.93.3), and the iteration of the compound residere with the simplex sedi.

90 See Balland, A., ‘Sur la nudité des nymphes’, Mélanges offerts à Jacques Heurgon (1976), 111Google Scholar.

91 Balland, op. cit. (n. 90), 2, speaks of ‘quasi-vers’.

92 See Pikhaus, op. cit. (n. 81), 150.

93 The texts are now re-edited and discussed in detail in the work cited at n. 81; see also Courtney, op. cit. (n. 3), no. 199.

94 See Pikhaus, op. cit. (n. 81), 151.

95 This is clear from the prose inscription which accompanies the verses; see also J.-M. Lassére in the work cited in n. 81, 220–1.

96 See A. and É. Bernand, op. cit. (n. 6).

97 To the centurions' inscriptions listed above can be added the military inscriptions nos 6, 9, 14, 20 (Greek), 38, 39 (Greek), 46, 56, 60, 74.

98 See A. and É. Bernand, op. cit. (n. 6), 133 n.1.

99 See Speidel, M. P., Die Denkmäler der Kaiserreiter. Equites Singulares Augusti (1994), no. 760 with 732.Google Scholar

100 See Speidel, op. cit. (n. 99), 424 on no. 760.

101 On the plural corpora ‘body parts’, see Adams, J. N., Pelagonius and Latin Veterinary Terminology in the Roman Empire (1995), 581.Google Scholar

102 With this substandard form of the perfect, cf. reguit in another centurion's inscription, ILS 2671.

103 So Speidel, op. cit. (n. 99), 424 on no. 760, though with two crucial misprints: ‘Das Gedicht ist aus derselben Feder wie 594 und 760’ (substitute 596 and 761); see also idem, 425 on no. 761.

104 It is worth recalling that the new fragment of the poet Gallus (see Anderson, R. D., Parsons, P. J. and Nisbet, R. G. M., ‘Elegiacs by Gallus from Qasr Ibrîm’, JRS 69 (1979), 125–55Google Scholar) was found in a military environment in Egypt, to which it was probably taken by an officer. At the pilgrimage site of Philae in Egypt one Junius Sabinus of the cohors III Ituraeorum has left four lines of Greek verse (IGP 159) which contain a rare verb (ἐγάνωσεν), lit. ‘brighten, polish’, used in a metaphorical sense ‘praise’ which may only be attested elsewhere in Pindar's Paeans (7.8). I owe this reference to I. C. Rutherford.

105 See E. Birley, ‘The origin of legionary centurions’, in idem, Roman Britain and the Roman Army: Collected Papers (1953), 104–24.

106 cf. B. Dobson, Die Primipilares. Entwicklung und Bedeutung, Laufbahnen und Persönlichkeiten eines römischen Offiziersranges (1978), 258–9.

107 See C. Sourvinou-Inwood, ‘ReadingGreek Death (1995), 180–216.

108 cf. Dobson, op. cit. (n. 106), 287.

109 cf. Dobson, op. cit. (n. 106), 320.

110 cf. Dobson, op. cit. (n. 106), 322.

111 For an up to date text, see F. Kayser, Recueil des inscriptions grecques et latines (non funéraires) d'Alexandrie impériale (1994), no. 19, pp. 69–73. I am grateful to Dr Benet Salway for drawing my attention to this inscription and for offering a convincing interpretation of its oddities.

112 See, e.g. CIL III.90, 402, 416, 7240, 7241, 7265, Speidel, op. cit. (n. 99), 688c (from Anazarbos in Cilicia).

113 See Corpus Inscriptionurn Semiticarum II.3962 = D. R. Hillers and E. Cussini, Palmyrene Aramaic Texts (1996), 0308 (Latin and Palmyrene), Starcky, J., Inventaire des inscriptions de Palmyre X (1949), 81, = Hillers and Cussini 1397 (Greek and Palmyrene). This second centurion (Julius Maximus) accompanied a caravan, and was honoured with a statue; he would almost certainly have been an Aramaic speaker.Google Scholar

114 See N. Lewis, The Documents from the Bar Kokhba Period in the Cave of Letters. Greek Papyri (1989), no. 11.

115 See Kolník, T., ‘Q. Atilius Primusinterprex centurio und negotiator: eine bedeutende Grabinschrift aus dem I. Jh. u. z. im Quadischen Limes-Vorland’, Acta Archaeologica Academiae Scientiarum Hungaricae 30 (1978), 62Google Scholar.

116 See Kolník, op. cit. (n. 115), 67.

117 See Kolník, op. cit. (n. 115), 66, 68.

118 See e.g. Dobson, B., ‘Legionary centurion or equestrian officer? A comparison of pay and prospects’, Ancient Society 3 (1972), 193Google Scholar; idem, ‘The significance of the centurion and “primipilaris” in the Roman army and administration’, ANRW II.1 (1974), 403–7.

119 See above, p. 118, and for additional details see Adams, op. cit. (n. 2).