Hostname: page-component-cd9895bd7-8ctnn Total loading time: 0 Render date: 2024-12-24T18:52:17.408Z Has data issue: false hasContentIssue false
  • English
  • Français

Roman Army Pay Scales*

Published online by Cambridge University Press:  24 September 2012

M. Alexander Speidel
M. Alexander Speidel
Affiliation:
Basle
Get access Rights & Permissions [Opens in a new window]

Extract

How much did Rome pay the soldiers serving in the legions and the auxilia, who expanded and defended her empire? The answer is of some significance not only to the history of the Roman army but to the political, social, and economic history of the Roman Empire in general. Many a learned article has therefore been devoted to this matter and steady progress has been made. Yet problems remain, the evidence being scanty and often not readily intelligible. Work on the 600 and more writing-tablets from the legionary fortress of Vindonissa (Switzerland), currently in progress, has turned up a missing link in the chain of evidence. The new text, a pay receipt of an auxiliary soldier, reveals a new sum and thus allows the reconstruction of the Roman army's pay scales through the first three centuries A.D. The overall pay model given below reconciles all the hitherto known evidence.

Type
Articles
Information
The Journal of Roman Studies , Volume 82 , November 1992 , pp. 87 - 106
Copyright
Copyright ©M. Alexander Speidel 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 RMR, pp. 241f. and 255; cf. also Wierschowski, L., Heer und Wirtschaft. Das römische Heer der Prinzipatszeit als Wirtschaftsfaktor (1984), 131Google Scholar. and 228 (n. 58). Stipendia continued to be paid in the fourth century: P. Panop. 2 passim (A.D. 299/300), P. Oxy. 1047 (early fourth century), Paneg. Lat. III (XI).1.4 (mid-fourth century). Domitian added a quartum stipendium (Suet., Dom. 7.3) after his victory over the German tribe of the Chatti in A.D. 83. A sestertius of A.D. 84 with the legend Stip Aug Domitian (cf. Kraay, C. M., ‘Two New Sestertii of Domitian’, American Numismatic Society Museum Notes 9 (1960), 109–16Google Scholar) reveals the date and confirms Suetonius' statement of a fourth pay-day. Later, most probably after Domitian's death in A.D. 96, the stipendium Domitiani was abolished. By the late second century at the latest we find the old system of three pay-days reintroduced (cf. RMR 71 and Fink's comments ibid, p. 253). In Dio's time only the pay-rise was remembered (Dio LXVII.3.5).

2 1 January: RMR 72.7; 73 fr. h; ChLA 466; 473; 495; P. Panop. 2.37; 58; 201; 292. 1 May: RMR 66fr. b I 30, 71 fr. a 1; 10 fr. b 5. 1 September: RMR 66 fr. b II 3; ChLA 495; P. Oxy. 1047; P. Oxy. 2561.

3 Jahn (1984), 66ff.

4 For easy comparison all figures will be given here and below in sestertii. Sestertii, four of which make a denarius, seem to have been the basis on which the soldiers' pay was originally calculated (cf. Jahn (1984), 65) although the stipendia were paid mainly in denarii (cf. Doppler, H. W., ‘Die Münzen’, in Ch. Meyer-Freuler, , Das Praetorium und die Basilica von Vindonissa (1989), 107–19, esp. 110f.Google Scholar, and the documentary evidence of the papyri (1) and the new Vindonissa pay receipt (11)).

5 Suet., Iul. 26.3 may imply that Caesar had already fixed this sum by doubling the legions' previous pay: ‘legionibus stipendium in perpetuum duplicavit.’ No legionary pay-rise is recorded for the reign of Augustus.

6 Tac., Ann. I.17.4: 10 asses per day in A.D. 14. This equals 912½ sestertii a year, which shows that Tacitus (i.e. the rebellious soldier speaking) gives no more than an approximation (if he was not implying a ‘military year’ of 360 days). The intention was clearly to dramatize the soldiers' situation, which is why their pay was broken down to the day. Dio LXVII.3.5 reports that the pay per pay-day before A.D. 84 was 300 sestertii.

7 Domitian's quartum stipendium consisted of three aurei (= 300 sestertii) (Suet., Dom. 7.3; cf. also n. 1). After abolishing the stipendium Domitiani the old system of three pay-days was reintroduced, but now every soldier received 400 sestertii (Dio LXVII.3.5).

8 All we learn from Severus' Vita (HA, Sev. 12.2) and Herodian (III.8.5) is that the increase was greater than all previous ones. Jahn (1984) has shown this increase to have been 100 per cent. Jahn's convincing arguments can now be confirmed (cf. below, VI, and n. 89). There seems to have been no pay-rise during the reign of Commodus; cf. Passerini, A., ‘Gli aumenti del soldo militare da Commodo a Massimino’, Athenaeum 24 (1946), 145–59Google Scholar.

9 Caracalla increased the soldier's normal pay by a half to win over the soldiers after he killed his brother: Herodian IV.4.7; cf. also Dio LXXVIII.36.3 who states that Caracalla's increase cost Rome 70 million denarii yearly around A.D. 218. At this time, it seems, Caracalla's pay-rise was at least partially taken back by Macrinus (Dio LXXVIII.12.7; 28.2; cf. Pekáry, Th., ‘Studien zur römischen Währungs- und Finanzgeschichte von 161–235 n.Chr.’, Historia 8 (1959), 443–89Google Scholar, esp. 484). Cf. also Dio LXXVIII.28.3 and 36.1 for Macrinus paying the soldiers recruited during his reign the rates Septimius Severus had established. As this, according to Dio, was one of the reasons for Macrinus' overthrow, Elagabalus almost certainly restored the previous pay scale (cf. Jahn (1984), 66 n. 49).

10 Maximinus Thrax doubled the soldiers' pay: Herodian VI.8.8. After Maximinus Thrax there seems to have been no further increase of the stipendia (cf. Jahn (1984), 66, 68), only the two other forms of soldiers' income, annona and donativa, were increased (cf. D. van Berchem, ‘L'annone militaire dans l'empire romain au 3e siècle’, Mem. Soc. nat. des Ant. de France 80 (1936), 136f.; Jahn (1984), 53ff.).

11 The conversions are based on the following rates: 1 sestertius = 1 drachma = 7 obols, or 1 denarius = 28 obols.

12 For this new pay document see below, IV.

13 On the dates, cf. Jahn (1983), 222f., who compares lay-out and script of the papyri to RMR 70.

14 The figure is given here as convincingly restored by Jahn (1983), 221. The difference of ½ obol from the sum in ChLA 446 cannot be expressed in asses and may have to do with fluctuating currency exchange rates (cf. Jahn (1983), 223). The reading of the exact amount of obols may also be doubted.

15 cf. P.Panop. 2.197f., where the pay of a praepositus equitum promotorum legionis II Traianae is listed separately.

16 RMR 68 was formerly presumed to mention legionary soldiers because of the tria nomina of the recipients. M. P. Speidel, ‘The pay of the auxilia’, RAS I, 83–9, esp. 86 and nn. 8–10, and more recently Mócsy, A., ‘Die Namen der Diplomempfänger’, in Eck, W. and Wolff, H. (eds), Heer und Integrationspolitik (1986), 437–66Google Scholar, have shown that as early as the first century A.D. the tria nomina are no proof for either Roman citizenship or type of unit.

17 M. P. Speidel, op. cit. (n. 16), 86. Readers may note that the present author, M. A. Speidel (the ‘Younger’), Basel, is the nephew of M. P. Speidel (the ‘Elder’), Honolulu.

18 Earlier suggestions of the 5:6 pay model (cf. Johnson, A. Ch., ‘Roman Egypt to the reign of Diocletian’, in Frank, T., An Economic Survey of Ancient Rome II (1936), 670ff.Google Scholar, Passerini, A., Le coorti pretorie (1939), 101Google Scholar n. 2, and Forni, G., Il redutamento delle legioni da Augus to a Diocleziano (1952), 32ff.Google Scholar) were lacking an explanation of the figures given in the papyri.

19 These can be described as the 1:3, 3:5, and 2:3 theses. For a short summary and the literature see Jahn (1984), 58ff., esp. nn. 17 and 18.

20 M. P. Speidel, op. cit. (n. 16), 145 quoted the career of the ‘Captor of Decebalus’ (AE 1969/70, 583; cf. Speidel, M. P., ‘The captor of Decebalus, a new inscription from Philippi’, JRS 60 (1970), 142–53Google Scholar = RAS I, 173–87, esp. 179f.), who was promoted from the rank of a vexilliarius equitum of the legio VII Claudia to a duplicarius alae. According to the 1:3 thesis this would have meant a severe pay-cut. For more evidence see legio VII Claudia, 180 and n. 43; cf. also Wierschowski, op. cit. (n. 1), 7ff., esp. for the high-ranking of the alae.

21 M. P. Speidel, op. cit. (n. 16), 87; Kaimio, J., ‘Notes on the pay of Roman soldiers’, Arctos 9 (1975), 39–46Google Scholar, esp. 41: ‘Unfortunately, all my attempts to find a mathematical solution to the problem of 84 denarii 15¾ obols have failed’.

22 ChLA x, 7ff.

23 Jahn (1984), 64f. and idem (1983), 224ff.

24 Jahn (1984), 64f. and idem (1983), 225ff.

25 For the full publication of this tablet and all other Vindonissa writing-tablets, see M. A. Speidel, Die römischen Schreibtafeln aus Vindonissa (forthcoming); for the tablets already published idem, ‘Neue Inschriften auf Schreibtäfelchen aus dem Schutthügel des Legions-lagers Vindonissa’, Jahresbericht der Gesellschaft Pro Vindonissa 1986 (1987), 49–64Google Scholar, esp. 49 with the literature. Cf. also idem, ‘Entlassungsurkunden des römischen Heeres. Eine hölzerne Entlassungsurkunde aus dem Schutthügel des Legionslagers Vindonissa’, Jahresbericht der Gesellschaft Pro Vindonissa 1990 (1991), 591–64; idem, ‘Römische Schreibtäfelchen aus Vindonissa’, Specima nova, pars prima (forthcoming).

26 16 × [7.3] cm; the lower half is missing. The remaining upper half shows on its inside four lines of cursive script. The blank space after the last line reveals that no further text is missing, apart from, perhaps, the closing-formula Actum Vindonissae on the now missing lower half. The outside of the tablet is blank. The tablet was probably found in the rubbish dump (‘Schutthügel’) of the fortress. The reading has been established with the help of enlarged photographs and a microscope.

27 cf. CIL v.4698 (Brixia). On this inscription Clua was the name of the father of a certain Esdrila. Assuming a similar dissemination of both names, Clua may have originated from the northern Italian Alpine region, the alpes Raeticae, perhaps from one of the valleys north of Brescia (cf. Untermann, J., ‘Namenlandschaften im alten Oberitalien’, Beiträge zur Namenforschung 10 (1959), 126ff.Google Scholar). Here, Raetian tribes are known to have lived (Strabo IV.6.8) and the indigenous names, according to J. Untermann (151ff.), seem to be of Raetian origin. On Clua's origin and the recruiting area of his unit, cf. Hartmann, M. and Speidel, M. A., ‘Die Hilfstruppen des Windischer Heeresverbandes’, Jahresbericht der Gesellschaft Pro Vindonissa 1991 (1992)Google Scholar. Cf. also Holder, A., Altceltischer Sprachschatz (1896–1904), III, 1238Google Scholar: Cloa (Scarponne), III, 1240: Clu (Langres).

28 Some irregularities may cause surprise. When copying the date, Clua omitted the cognomen Quintilianus of the second consul. Dating by suffect consuls outside Italy was very uncommon (cf. Eck, W., ‘Consules ordinarii und consules suffecti als eponyme Amtsträger’, Actes du colloque en memoire de Attilio Degrassi Rome 27–28 Mai 1988 (1991), 15–44, esp. 30ff.Google Scholar) and may shed some light on military administration customs of the early empire. The use of both forms of the letter ‘e’: E (in ‘Raetor(um)) and II (being the normal form on stylus tablets) in the same text or word, was, admittedly, unusual. Yet examples can be found with ease: cf. e.g. Bakker, L. and Gallsterer-Kröll, B., Graffiti auf römischer Keramik im Rheinischen Landesmuseum Bonn (1975) no. 349Google Scholar; Tomlin, R. S. O., Tabellae Sulis (1988) no. 53Google Scholar; CIL XIII. 10009, 6, 119a; 10010, 188, 228d2, 2281, 251e passim. For the colloquial expression eques Raetorum cf. below n. 32.

29 Vegetius, Ep. rei mil. II. 20 reports, that the signiferi, who had to be litterati homines, were in charge of the troops’ money and responsible singulis reddere rationem.

30 The legionary horsemen were assigned, instead, to the centuriae (cf. Speidel, M. P., ‘Ein Silberring aus Baden für die Reiter der 21. Legion’, Helvetia Archeologica 70 (1987), 56–8Google Scholar).

31 Otherwise unknown. He may have been a member of the legion (perhaps an eques legionis) ad tradendam disciplinam immixtus (Tac, Agric. 28), as this was often done during the early Empire. Cf. also AS 1969/70, 661; CIL III.8438 and esp. M. P. Speidel, ‘A Spanish cavalry decurion in the time of Caesar and Augustus’, RAS I, 111–13.

32 For such colloquial expressions, cf. Tab. Vindolanda 1988/944 (unpub.): equites Vardulli for equites cohortis I fidae Vardullorum milliariae civium Romanorum equitatae and Tab. Vindolanda 1985/183 (unpub.): Vocontii for equites alae (Augustae) Vocontiorum civium Romanorum (for these tablets, cf. e.g. Birley, R., The Roman Documents from Vindolanda (1990), 29 and 9)Google Scholar. The speech the emperor Hadrian gave on his inspection of the troops at Lambaesis, recorded in ILS 2487, 9133–9135 (A.D. 128), uses both terms equally: campus Commagenorum and in the next line: eq(uites) coh(ortis) VI Commagenorum (ILS 9134). Cf. also M. P. Speidel, ‘Ala Maurorum? Colloquial names for Roman army units’, RAS I, 109–10.

33 No alae Raetorum are known (the late Roman ala I Flavia Raetorum: Not. Dign. Occ. xxxv. 23 and ala V Raetorum: Not. Dign. Or. XXVIII. 30 were upgraded cohortes (equitatae?), cf. E. Birley, ‘Raetien, Britannien und das römische Heer’, RA, 259–71, esp. 266 n. 33; M. P. Speidel, ‘The Roman army in Arabia’, RAS I, 229–72, esp. 248f.). Of the many cohortes Raetorum the following are known to have had cavalry detachments: I Raetorum eq.; I Raetorum eq. c. R.; III Raetorum eq.; V Raetorum (eq.?); VII Raetorum eq.; VIII Raetorum eq. During the first half of the first century A.D. there is no evidence of where any of these troops may have been stationed. Of the cohors VII Raetorum stamped tiles have been found in Vindonissa dating around the mid-first century A.D. (CIL XIII.12457, 12458; cf. Hartmann and Speidel, op. cit. (n. 27)). For the Vindonissa alae, cf. M. A. Speidel, ‘Römische Reitertruppen in Augst’, ZPE 91 (1992), 165–75.

34 This is how ‘accepi stipendi proximi LXXV’ is to be understood. The expression written out fully is also found in RMR 70 passim: ‘accepit stipendi …’ and in P. Yadin 722, 4 and 11: ‘accepi stipendi …’, RMR 68, ChLA 446 and 495 show only ‘accepit stip.’, RMR 71 and 72: ‘accep. stip.’ R. Marichal has shown (ChLA x, p. 14), that stipendi was a genitive (‘génetif de relation’), linked to the verb accipere (‘adverbialer Genetiv’), and specifying the nature of what was received (‘Genetiv der Rubrik’), rather than the amount (‘il n'est en rien question de quantité, mais de nature’). Hence his translation ‘reçu comme solde’, which is followed above: ‘I … have received … as next pay.’

35 cf. e.g.RMR 70 (A.D. 192); 73 (A.D. 120–50); ChLA 473 (second/third century A.D.).

36 RMR 73 fr. a i 24.

37 cf. S. Daris, ‘Le truppe ausiliarie romane in Egitto’, ANRW 11.10.1 (1988), 743–66, esp. 752f. For advanced payments of grain, cf. RMR 78, 2 and 9 (second/third century A.D.).

38 cf. Speidel, M. P., ‘Outpost duty in the desert. Building the fort at Gholaia (Bu Njem, Libya), Antiquités africaines 24 (1988), 99–102CrossRefGoogle Scholar. Marichal, R., ‘L'occupation romaine de la Basse Egypte: le statut des auxilia (1945), 54f.Google Scholar, explained the missing stipendia of several soldiers in RMR 70 by their absence from the camp at the time the money was paid or the record made respectively.

39 For possible outposts, cf. Hartmann and Speidel, op. cit. (n. 27).

40 M. P. Speidel, op. cit. (n. 15), 87; Jahn (1983), 66.

41 CIL XII. 2602 = ILS 2118.

42 cf. J. Gilliam, ‘Dura rosters and the Constitutio Antoniniana’ RAP, 289–307, esp. 292ff.; idem, ‘An Egyptian cohort in A.D. 117’, RAP, 309–15, esp. 309 and n.3.

43 Confirmation of this will be found in the later data (see below p. 96, and n.93). The ratio between the income of a miles and an eques in the legion, in theory, may also have been calculated on the same basis as in the cohorts (750 sestertii:900 sestertii before A.D. 84), i.e. 5:6. This would lead to 1,080 sestertii per year, a sum easily divisible by 3, suggesting a stipendium equestre of 360 sestertii. But this sum does not reconcile with the figures of P.Panop. (cf. below pp. 99–100).

44 CIL VIII.2532, 18042 = ILS 2487, 9133–9135: Difficile est cohortales equites etiam per se placere, difficilius post alarem exercitationem non displicere: … equorum forma armorum cultus pro stipendi modo.’

45 AE 1969/70, 583; M.P. Speidel, op. cit. (n. 20), 146f.; cf. also idem, op. cit. (n. 16), 87 and n. 18.

46 AE 1969/70, 661 from A.D. 55/8–71/4.

47 CIL VIII.2354 add. = ILS 305. He was transferred from the same legion (I Augusta) to the same ala (Pannoniorum) as the above M. Licinius Fidelis several years earlier. One may also note the late second-century career of M. Aurelius Paetus, who was promoted from eques alae to sesquiplicarius legionis (AE 1977, 720; cf. Y. LeBohec, La Troisième légion auguste (1989), 205 and n. 214), which, according to the above pay scales, also entailed a pay-rise.

48 For confirmation see below 96f. and 99f. The demand of the Batavian Cohorts in A.D. 69 for ‘donativum, duplex, stipendium, augeri equitum numerum’ (Tacitus, Hist.IV.19) — a passage quoted with great regularity whenever the pay of the auxilia is being discussed — is of no value in helping to determine the basic pay of the auxilia. For a detailed discussion of the passage, cf. Wierschowski, op. cit. (n. 1), 9ff.; see also M. P. Speidel, op. cit. (n. 16), 87 n. 19.

49 Tacitus, Ann. I.17 reports deductions for: vestis, arma, tentoria.

50 M. P. Speidel, op. cit. (n. 16), 86, who first recognized the I per cent deduction, suggested an exchange fee for conversion of denarii to drachmae. Yet, as I per cent of the stipendia in RMR 70, ChLA 446 and 495 appears to have been deducted although they were paid in denarii (and obols), this deduction is perhaps not to be explained as a conversion fee. Hence Jahn (1984), 63 n. 36, surmised its use for an institution or purpose, benefiting all soldiers of the unit. Watson, G. R., ‘Documentation in the Roman army’, ANRW, II, 1 (1974), 493–507Google Scholar, esp. 499, suspected a service-charge for book-keeping.

51 cf. above, II, with our suggestion that the horseman Clua received his money in advance because he was about to leave the camp. If this is correct, none of the above explanations (n. 50) would fully apply, which might explain the absence of the I per cent deduction. It is, of course, equally possible that the I per cent deduction was not yet in force at the time the Vindonissa tablet was issued.

52 We follow Marichal's convincing proposition (ChLA I, p. 25), that the 128 drachmae in RMR 69, 1. 5, deducted in victum (?), included 28 drachmae for the saturnalicium, leaving the standard 100 drachmae in victum as found with the second and third stipendium of this document.

53 These standard stoppages and 4 drachmae, ad signa were the only deductions made during the second stipendium of Q. Iulius Proculus and C. Valerius Germanusin A.D. 81 (RMR 68), equalling c. 42 per cent of the stipendium (cf. also n. 54).

54 This may also have been true for the legionary account in RMR 69 though we cannot be certain since the greater part of the last entry for the quartum stipendium with the deductions is missing. Of the first stipendium c. 75 per cent was kept back, the following two show deductions of c. 50 per cent. The items for which these deductions were made are lost.

55 cf. e.g. RMR 73.

56 Masada II, 45. Cf. ibid., 41ff. for a detailed discussion of the differences between the Masada document and the pay records in RMR.

57 The text and the reconstructions given here are the editors’ (Masada II, 46f.); cf. also their commentary 47ff. The expansion of the date in l.1 is uncertain and could also be understood as IMP VES]PAS [A]V[G VI TIT]O IIII CO[S, i.e. the year 75 (cf. op. cit., 47f.). The reconstruction of l. 2 seems open to doubts, since it lacks a grammatical link between the two words (cf. op. cit., 48f.).

58 The editors of the document understood the expression accepi stipendi, by suggesting the genitive to relate to the sum of 50 denarii, which was obviously not the full stipendium. Hence their translation ‘I received of/from my pay’ (pp. 44,47). On the other hand, they quoted RMR 68, 69, and 70, where they believed the same expression to denote the full sum, despite the fact that these documents only show the full sum minus I per cent. As the new Vindonissa tablet proves, accepi stipendi could indeed be followed by the full stipendium. It must, therefore, be translated ‘I have received as pay’, the genitive denoting the quality of the money rather than the amount (cf. above n. 34).

59 Masada II, 51; cf. also 44f.

60 See also the abbreviations accep. and stip. as in e.g. RMR 68, 71, 72, ChLA 446, 495.

61 In his comment on RMR 68 (p. 248), R. O. Fink reached the same conclusion.

62 It could also be argued that C. Messius had no money at all in his deposit, which would also explain the absence of the entries concerning the depositum.

63 In any case, it seems, the debts were not automatically deducted from the soldiers' savings: the amounts in deposito and in viatico remained untouched: e.g. RMR 70 fr. a i, 28ff.; ii, 25ff.; fr. b i, 9ff.; 22ff. The new Vindonissa tablet may show how such debts could originate.

64 Because of the fragmentary state of the papyrus, the possibility that Messius ran up further debts cannot be totally excluded.

65 cf. the editors’ comments, Masada II. 39 and 51ff.

66 P.Vindob. L. 135; cf. H. Harrauer and R. Seider, ‘Ein neuer lateinischer Schuldschein: P.Vindob. L. 135’, ZPE 36 (1979), 109–20, Taf. IV. For further comments on this text see J. F. Gilliam, ‘Notes on a new Latin text’, RAP, 429–32; M. P. Speidel, ‘Auxiliary units named after their commanders: four new cases from Egypt’, RAS I, 101–8. J. Shelton, ‘A note on P.Vindob. L135’, ZPE 38 (1980), 202.

67 By analogy to the equal sums deducted in victuml sumptuarium from both the legionaries’ (P. Yadin, 722) and the auxiliaries’ (RMR 68) stipendium and the equal pay for cavalrymen in the legions and in the alae (for confirmation, cf. also the third-century data below), we assume that the stoppages for the horsemen's barley-money were also equal in both types of unit. Differences in stoppages, it seems, were mainly due to different equipment and clothes, hence Hadrian's remark: ‘equorum forma armorum cultus pro stipendi modo’, ILS 2487 (cf. above n. 44).

68 If boots and socks were deducted with each stipendium rather than at the beginning of the year, he would have had 193.2 sestertii left: the deduction then would have been 7 denarii or 28 sestertii:3 = 2.3 denarii or 9.3 sestertii. There is no apparent reason, however, why the equites alae should have paid for their boots and socks more often than the equites legionis. For the equal treating of equites alae and legionis see above n. 20 and pp. 92–3 and below pp. 99–100ff.

69 All other stoppages, mainly for clothes (in vestimentis) did not occur regularly, and Secundus will have avoided them, if he could. As the pay accounts of the second stipendium of Q. Iulius Proculus and C. Valerius Germanus show (RMR 68), it was possible to keep deductions at a minimum (cf. above n.49). Cf. also Tacitus, Ann. XIII.35: ‘fuisse in eo exercitu veteranos … sine galeis, sine loricis, nitidi et quaestuosi, militia per oppida expleta.’

70 Even if Secundus invested the borrowed money so that he could not dispose of it for a longer period of time, it could be argued that he probably needed no extra money for daily living expenses, since these were covered by the deductions from his pay.

71 For a possible reconstruction of the supply services, cf. Rodriguez, J. Remesal, ‘Die Procuratores Augusti und die Versorgung des römischen Heeres’, in Vetters, H. and Kandler, M. (eds), Der röische Limes in Österreich 36/1; Akten des 14. Internationalen Limeskongresses 1986 in Carnuntum (1990), 55–64Google Scholar. In La ‘annona militaris'y la exportatión de aceite Betico a Germania (1986), 91ff.Google Scholar, esp. 94, the same author expressed the view that because of the many deductions from the soldiers' pay, hardly any money actually changed hands. This can have been no more than a general tendency during the first century A.D. as is shown by the accounts of the second stipendium of Q. Iulius Proculus and C. Valerius Germanus in RMR 68 (A.D. 81). Well over 50 per cent of these stipendi was actually paid out (cf. above nn. 53 and 54). Cf. also the soldiers' loan on the above P.Vindob. L 135 (A.D. 27), promising the repayment of 200 drachmae with the next stipendium. For the second-century developments, cf. below and especially the soldiers' loans CPL 128, 188, 189, 194.

72 cf. also A. R. Birley's suggestion that some of the soldiers described by Tacitus, Ann. XIII.35 (cf. n. 69) as ‘nitidi et quaestuosi’ ‘had been making money from selling “duty-free goods”’ from the army's supplies to civilians (‘The economic effects of Roman frontier policy’, in King, A. and Henig, M. (eds), The Roman West in the Third Century. Contributions from Archaeology and History, BAR Int. Ser. 109 (1981), 39–53, esp. 46)Google Scholar.

73 Occasional deductions survive for the repair of armour and helmet (‘re[f̣(ectio) loric(ae) et casid(is) I s(emis)’: ChLA 446, cf. Jahn (1983), 220 and n. 13) and for servants' food (‘ṣolvit tess(eras) baronum LX’: ChLA 495, cf. M. P. Speidel, ‘The soldiers' servants’, Anc. Soc. 20 (1989), 242 and n. 19).

74 cf. above n. 71.

75 Another way of decreasing stoppages was to keep the sums deducted at a fixed rate over periods of great inflation. This can be observed e.g. with the deposits which the equites cohortis XX Palmyrenorum had to pay for their horses: 125 denarii in both A.D. 208 and 251 (RMR 99 = ChLA 311 and RMR 83 = ChLA 352; cf. Davies, R. W., ‘The supply of animals to the Roman army and the remount system’, Service in the Roman Army (1989), 153–73)Google Scholar. Already in A.D. 139 equites alae did not pay significantly more, as a receipt for the return of such a deposit (RMR 75 = ChLA 397) reveals the ‘pretium equi’ to have been ‘[d]enạṛịọṣ c̣ẹṇṭụm[-’ (ll. 3 and 5; cf. esp. Marichal's comments in ChLA IX, p. 103).

76 P.Mich, VII.438 = CPL 188; cf. esp. Gilliam's comments and improved readings in RAP 53–9., esp. 54ff.

77 ‘labentem disciplinam retinuit ordinatis et officiis et inpendiis’ (HA, Hadr. x.3).

78 cf. Jahn (1983), 223 and especially the many examples of soldiers acquiring food, clothes, and even weapons mainly from or through their relatives, cited in Wierschowski, op. cit. (n. 1), 112ff. The earliest and the majority of these examples date to the early second century. Wierschowski therefore, too, comes to the conclusion, ‘dass sich seit dieser Zeit (the time of the Geneva papyri) das System der Soldatenversorgung seitens der Armee gewandelt haben muss’ (121).

79 van Berchem, op. cit. (n. 10); idem, ‘L'annone militaire est-elle un myth?’, Armées et fiscalité (1977), 331–40.

80 P.Hamb. 39 = RMR 76 (A.D. 179): 25 denarii per year for ϰϱάστις; (green fodder, esp. for horses: cf. Liddel, Scott and Jones, Greek-English Lexicon s.v.). As the deduction of 16 denarii for barley from each horseman's stipendium, made over one hundred years earlier, implies, this was not the full sum cavalrymen spent on horse fodder. By the fourth century at the latest, soldiers also received free rations for their servants’, cf. M. P. Speidel, op. cit. (n. 73), 242 and n. 20.

81 cf. only the many fractions of obols recorded in RMR 70 and ChLA 446. Jahn's argument (1983), 223ff., that payability of the stipendia was achieved by enforcing the 1 per cent deduction, does not seem convincing, for the sums credited (accepit stipendi) and those actually handed out (reliquos tulit) in RMR 70 show fractions of obols. If the military accountants had ever tried to achieve payability in round sums, it seems they should have been able to do better. Certainly the soldiers' yearly pay was calculated irrespective of its payability in thirds after a I per cent deduction.

82 cf. above p. 89 with n. 24.

83 Jahn (1983), 222ff.

84 Jahn (1983), 225, it seems, reached the same conclusion. His arguments, based on the assumed ratio of 5:6 between the legions and the auxilia (cf. Jahn (1984), 66ff.), can now be confirmed.

85 P.Panop. 2.292f.

86 This sum would allow for any number of duplicarii and any even number of sesquiplicarii. Some centurions’ pay may also have been included (cf. below p. 104f.). This understanding of the figures in P.Panop. does not allow for bonuses of the kind found in RMR 70 and ChLA 446 and 495. Perhaps, therefore, they were no longer included in the stipendia at this time.

87 Jahn (1984), 67.

88 Only if one of the pay-rises of Septimius Severus and Maximinus Thrax is totally denied, can this sum be reached.

89 P.Panop. 2.36ff.

90 Jahn (1984), 67 n.55.

91 The unlikely theoretical alternative being 294 stipendia of 250 denarii.

92 The surprisingly small number of soldiers in both the cohors XI Chamavorum (max. 131) and the ala I Hiberorum (max. 105) need not be the units’ full strengths (as Jahn (1984), 61 and nn. 28–30 seems to assume). It is perhaps more likely that the units, whose full strengths at this time are unknown, were split up into several detachments in different camps: cf. e.g. A. K. Bowman, ‘The military occupation of Upper Egypt in the reign of Diocletian’, BASP 15 (1978), 25–38, esp. 33. If correct, this might explain why the ala I Hiberorum, when the above pay arrived, was under the command of only a decurio (Besas; cf. P.Panop. 37).

93 In consequence, the figure suggested above of 1,050 sestertii before Domitian's pay-rise is also confirmed. The alternative presented above of 1,080 sestertii (cf. n. 43) can now, in all probability, be ruled out, for it cannot be run through the second- and third-century pay-rises to fit the sums of the Panopolis papyri.

94 He explained the figure by assuming the scribe of the papyrus had actually meant to write 343,200 denarii, which is divisible by the basic footsoldiers' stipendium (343,200/600 = 572). The mistake happened because the scribe, according to Jahn, misheard διαχοσίας for Τϱιαϰοσίας: cf. Jahn (1984), 68f.

95 For a list, see esp. Domaszewski, A. v. and Dobson, B., Die Rangordnung des römischen Heeres (2nd edn, 1967), XI–XIII and 29–37, esp. 29ffGoogle Scholar. Cf. also Jones, A. H. M., ‘The Roman civil service (clergical and sub-clergical grades)’, JRS 39 (1949), 38–55Google Scholar, esp. 44.

96 CIL XII.2602; cf. Domaszewski and Dobson, op. cit. (n.95), 31. Cf. also Breeze, D., ‘Pay grades and ranks below the centurionate’, JRS 61 (1971), 130–5Google Scholar, esp. 133, who suggested that the cornicularii were not actually mounted, but received equestria stipendia ‘simply as a means of increasing their pay’.

97 Divisible, of course, into e.g. 3 double and 1 basic (= 2 pay-and-a-half and 4 basic (or 2 double)) horsemen's pays and 282 double foot-soldiers' pays, etc.

98 For centurions attested in the officium of governors, cf. Jones, op. cit. (n. 95), 44 and n. 60.

99 cf. esp. Berchem, op. cit. (n. 79). If the supplies in kind did not suffice, the difference was paid in cash. Perhaps this is in part the explanation for the supernumerary 25 sestertii and 125 sestertii respectively in ChLA 446 and 495 (cf. above p. 89). The figures given in the Panopolis papyri have recently been discussed by R. Duncan-Jones, ‘Pay and numbers in Diocletians's army’, now in idem, Structure and Scale in the Roman Economy (1990), 105–17.

100 cf. Jahn (1984), 53ff. for comments and estimations especially on the figures given in the Panopolis papyri.

101 Jahn (1984), 53ff.

102 Breeze, op. cit. (n.96); cf. also J. F. Gilliam, ‘The Moesian “Pridianum”’, RAP 263–72,esp. 271f.and M. P. Speidel, op. cit. (n. 16), 88 and nn. 23–4.

103 AE 1976, 495 = 58 BerRGK (1977), no. 99 (Mainz-Weisenau); reign of Tiberius?

104 cf. Holder, P. A., The Auxilia from Augustus to Trajan (1980), 91Google Scholar, who finds confirmation for treble pay for the post of evocatus in the career of C. Iulius Macer, duplicarius alae Atectorigianae, before becoming evocatus in charge of 600 Raeti gesati during the first half of the first century A.D. (CIL XIII. 1041). This promotion entailed, according to Holder, a pay-rise.

105 cf. idem, 91. The evocati may later have been paid the otherwise highest pay rate below the centurionate, double horsemen's pay, i.e. the rate of a cornicularius (cf. n. 96 and n. 117). This assumption may find some support in the fact that legionary centurions were regularly appointed from those two ranks of the praetorian guard (cf. D. Breeze, ‘The organisation of the career structure of the immunes and principales of the Roman army’, BJ 174 (1974), 245–92, esp. 247ff.).

106 Forth the following see Jahn (1984), 69f.

107 Jahn (1984), 69 (cf. also ibid., 54). Hence, he concludes, they were not of equestrian, let alone senatorial rank.

108 Since we can now be more certain of the basic annual rate of 1,800 denarii for legionaries, the above ratios reached by Jahn (1984), gain further credibility. Their simplicity further suggests that the two sums of P.Panop. 2.197 and P.Oxy. 1047 were calculated on the basic pay of a legionary, which may be taken as an additional argument in favour of the two praepositi having been legionary centurions.

109 For the ranking of centurions, see Wegeleben, T., Die Rangordnung der römischen Centurionen (1913)Google Scholar. He surmised that all centurions in Cohorts II-X were equal in rank, differing only in seniority. Hence promotion was only involved upon transfer to the first cohort, then joining the senior grade of the primi ordines, of whom the primuspilus and the praefectus castrorum were the top ranks. This was accepted by E. Birley, ‘Promotions and transfers in the Roman army II: the centurionate’, RA, 206–20, esp. 206, and B. Dobson, ‘Legionary centurion or equestrian officer? A comparison of pay and prospects’, Anc.Soc. 3 (1972), 193–207, esp. 197 and n. 25, and 201f.

110 Later it may have been the cornicularius praefecti praetorio, receiving double horsemen's pay: cf. above n. 105.

111 cf. e.g. Durry, M., Les cohortes prétoriennes (1938), 264ff.Google Scholar; Watson, G. R., The Roman Soldier (1969), 98Google Scholar.

112 We are not informed how much a horseman in the praetorian guard received. If the difference in pay was the same as in the auxilia and in the legions, i.e. 150 sestertii per year before A.D. 84, we arrive at a yearly income of 6,300 sestertii for the cornicularius praefecti praetorio.

113 Breeze, op. cit. (n. 105), 247ff. Note also the many promotions from the rank of cornicularius. For the career prospects of the evocati Augusti, cf. also E. Birley, ‘Evocati Aug.: a review’, RA, 326–30.

114 Breeze, op. cit. (n. 105), 252.

115 cf. above n. 105.

116 On the commoda in general, cf. e.g. M. P. Speidel, ‘Cash from the emperor. A veteran's gravestone at Elecik in Galatia’, AJP 104 (1983), 282–6; Wolff, H., ‘Die Entwicklung der Veteranenprivilegien vom Beginn des I. Jahrhunderts V. Chr. bis auf Konstantin d. Gr.’, in Eck, W. and Wolff, H. (eds), Heer und Integrationspolitik. Die örmischen Militärdiplome als historische Quelle (1986), 44–115Google Scholar, esp. 50ff. The figure of 600,000 sestertii, although an emendation of a corrupt text, is generally accepted, cf. e.g. Dobson, op. cit. (n. 105), 193–207, esp. 198.

117 cf. CIL v.5832: P. Tutilius P.f. Ouf. veteranus, who died A.D. 29, formerly a signifer, aquilifer leg. and curator veteranorum, received praemia duplica ab Imperatore, and his pay grade must have been that of a duplicarius. L. Pellartius Celer Iulius Montanus, missus ex evocato et armidoctor. leg. XV Apol., boasted to have received 30,000 sestertii from Domitian, ‘quod ante ilium nemo alius accebit (!) ex hac militie (!)’, for he would normally only expect 24,000 sestertii (twice the amount of a normal soldier, i.e. 12,000 sestertii, cf. also n. 118) according to his pay grade as a duplicarius (AE 1952, 153: Aquileia). Cf. above n. 105 for the possible reduction of pay grades from triplicarius to duplicarius of the evocati.

118 Dio LV.23.1. Augustus had fixed this sum. It seemed to remain unaltered until Caracalla raised it to 20,000 sestertii (Dio LXXVII.24.1); cf. Wolff, op. cit. (n. 116), 52. It may be noted that all attempts to understand these sums as multiples of the stipendia seem to have failed, cf. Wolff, 52ff.

119 cf. Suet., Claud. I 1.3: ‘Gai… acta omnia rescidit.’

120 For a description of this development, cf. e.g. Holder, op. cit. (n. 104), 72ff.

121 Dobson, op. cit. (n. 109). For the following, see esp. pp. 196ff. and 199ff.

122 HA, Pertinax 1.5–6. The increasing number of cornicularii praefecti praetorio and evocati Augusti of the praetorian guard promoted to praefecti cohortis and tribuni cohortis in the third century also shows that the pay of the legionary centurionate, to which they were normally promoted, and of the prima and secunda militia must have been similar at that time. Cf. Breeze, op. cit. (n. 105), 252.

123 Dobson, op. cit. (n. 109), 201. This has been accepted by H. Devijver, ‘La Prosopographia Militarium Equestrium. Contribution à l'histoire social et économique du principal, in The Equestrian Officers (1989) (= Mavors VI), 396–411, esp. 409.

124 CIL XIII.3162; cf. the commentary on this text by Pflaum, H.-G., Le Marbre de Thorigny (1948)Google Scholar. This is the only known sum to have been paid to an equestrian officer as a salary. Dobson, op. cit. (n. 109), 201 and Devijver, op. cit. (n. 123), 409 have taken the 25,000 sestertii to be half the annual pay of the ‘militia prima’.

125 It may be noted that the sum of 50,000 sestertii cannot be explained as a multiple of any of the above basic pay grades, which shows that the pay grades of the equestrian ‘militiae’, as a career of their own, were calculated on completely different grounds. An attempt to re-establish the remaining equestrian salaries without further evidence must therefore produce wholly conjectural figures. During the first century, it appears, all the equestrian officers were paid better than the legionary centurions (cf. e.g. ILS 9090, CIL IX.2564; XII.3177, 3178). This might suggest that the pay rates of the ‘militia equestris’ and of the lower procurators were kept level until Septimius Severus when they seem to have been raised (cf. also n. 127). At the beginning of the third century A.D. the salary of the militia secunda was, according to the career of Rufinus (RIB 1288 = ILS 1425), higher than the income of a sexagenarian procurator, who still earned 60,000 sestertii at the time (Dio LIII. 15.5).

126 This conclusion of Dobson's is based on the Trajanic career of T. Pontius Sabinus (ILS 2726). Cf. Dobson, op. cit. (n. 109), 252.

127 H.-G. Pflaum, RE XXIII, 1272f. Cf. also idem, Abrégé des procurateurs équestres (1974), 56ff. Commanding a milliary cavalry unit as the ‘militia quarta’ would also lead to a centenarian procuratorship. Under Septimius Severus and Caracalla some of the salaries of both equestrian officials and senators seem to have been raised; cf. e.g. Alföldy, G., ‘Die Stellung der Ritter in der Führungsschicht des Imperium Romanum’, Die römische Gesellschaft, HABES I (1986), 162–209Google Scholar, esp. 178, 180; cf. also P. A. Brunt, ‘Pay and superannuation in the Roman army’, PBSR 18 (1950), 50–71, esp. 69.

128 cf. J. F. Gilliam, ‘The appointment of auxiliary centurions’, RAP, 191–205, esp. 202 and n. 25; see also Domaszewski and Dobson, op. cit. (n. 95), 53 and 57.

129 contra Domaszewski and Dobson, op. cit. (n. 95), 56.

130 cf. e.g. ILS 305 (dec.alae–cent.leg.; Flavian-Trajan), ILS 2596 (dec.coh.–cent.leg.; mid/late first century); CIL v.522 (cent.coh.–cent leg.; mid first century). Cf. also Domaszewski and Dobson, op. cit. (n.95), 53f. and 56f. for further examples. During the first century A.D. (until Domitian's pay-rise?) it seems the equestrian officers were paid better than the legionary centurions: cf. above n. 127.

131 cf. e.g. Gilliam, op. cit. (n. 128); M. P. Speidel, op. cit. (n. 20), 183; Holder, op. cit. (n. 104), 86ff.

132 CIL v.522, mid-first century.

133 For the pay of the praetorian guard, see above VII and n. III.

134 Speidel, M. P., Die equites singulares Augusti (1965). 49Google Scholar.

135 cf. ibid., 50; Speidel, M. P., Guards of the Roman Armies (1978), 36 and n. 196Google Scholar.

136 2.5 is the maximum factor, if the praetorian guard is to remain the best paid Roman troop. (In this case an eq. sing. Aug. would have drawn 7,000 sestertii after Severus’ pay-rise, a praetorian 8,000.) Vet this is but a theoretical possibility, for if that factor is applied and if five times basic pay is accepted as the minimum salary of a decurion, there would have been too insignificant a difference between the legionary centurions’ pay and that of the decurio equitum singularium Augusti (e.g. 36,000 cent.deg.–35.000 dec. eq. sing. Aug. after Severus’ pay-rise). The same basic pay as the alares or their pay-and-a-half may in theory have been the basic pay of the emperor's horseguards. It may also be that their pay was not a multiple of the basic salary of the alares, but some independent (higher) amount below the pay of a duplicarius alae.

137 If the emperor's horseguards received 1.5 times the basic pay of the equites alae, a factor of 6, or 7 at the very most, could also be envisaged. Domaszewski, op. cit. (n.95), 7011ff., assumed that these ranks were paid three times basic legionary pay, which he believed to be 500 denarii per year during the reign of Septimius Severus. His assumption was based on the money presents given to members of military collegiae according to their rank. Yet these sums show no correspondence with the soldiers' income. Moreover, treble basic legionary pay as the income of auxiliary centurions and decurions would have brought a considerable pay-cut for the above mentioned praetorian L.Arnius Bassus upon his promotion to the auxiliary centurionate.

138 Alföldy, op. cit. (n. 127), 180.