² ‘Der Truppensold der Kaiserzeit’, Neue Heidelberger Jahrbücher,x, 1900, pp. 218–241, supplemented for the pay of immunes, principales and officers in ‘Die Rangordnung des römischen Heeres’, Bonner Jahrbücher, cxvii, 1908Google Scholar. My obligations to Domaszewski's work are profound and will be evident. I do not deal with the donatives, although they might substantially supplement a soldier's income, since they were irregular in amount and our authorities afford us no means of making a complete statement of them.

³ Passerini, A., Le Coorti Pretorie, Rome, 1939, pp. 100–114Google Scholar; cf. Athenaeum, xxiv., 1946, pp. 145–59Google Scholar.

⁴ e.g. Diod. xiv. 16, 5, .

⁵ de vita populi Romani' ap. Non. p. 853L, ‘stipendium appellabatur quod aes militi semenstre aut annuum dabatur’.

⁶ vi. 39, 12,

⁷ E. Nischer in Kromayer-Veith, Heerwesen und Kriegführung der Griechen und Römer, pp. 328–30.

⁸ Festus, s.v. ‘Praetoria cohors’ p. 249L, ‘Scipio enim Africanus primus fortissimum quemque delegit, qui ab eo in bello non discederent et cetero munere militiae vacarent et sesquiplex stipendium acciperent.’

⁹ N.H. xxxiii. 44–5, especially 45, ‘in militari tamen stipendio semper denarius pro X assibus datus est.’ On the date of the currency change cf. Mattingly, Num. Chron. 1934, pp. 81 sqq.

¹⁰ JRS. xxvii, pp. 101 ff. I am indebted to Dr. J. G. Milne for a helpful discussion of this problem.

¹¹ Polyb. ii. 15, 6.

¹² Suet. Div. Iul., 26, 3, ‘legionibus stipendium in perpetuum duplicavit.’

¹³ Ann. i. 17,6, ‘denis in diem assibus animam et corpus aestimari.’

¹⁴ Cf. Caes., BC iii. 76Google Scholar; after marching a ‘iustum iter’ Caesar ‘duplicato eius itinere VIII milia passuum ex eo loco procedit’. The ‘iustum iter’ was probably 20 miles, cf. Veg. i. 9; Kubitschek, P-W. ix. 2309 n. 2.

¹⁵ Div. Iul. 26, 3, ‘frumentum quotiens copia esset etiam sine modo mensuraque praebuit ac singula interdum mancipia et praedia viritim dedit.’

¹⁶ Ib. 68, 1, ‘(optulerunt) milites gratuitam et sine frumento stipendioque operam.’

¹⁷ BC iii. 53, 5, ‘cohortemque postea duplid stipendio, frumento veste, cibariis militaribusque donis amplissime donavit.’

¹⁸ Bell. Afr. 47, 4, ‘In Africa autem non modo sibi quicquam non adquisierant aut paraverant sed etiam propter annonae caritatem ante parta consumpserant.’

¹⁹ App, . BC iv. 117, 493Google Scholar, .

²⁰ Tac., Ann. xv. 72, 1Google Scholar; Suet. Nero, 10.

²¹ It may be suggested that administrative convenience would account for the different treatment on this hypothesis accorded to praetorians and legionaries. In the provinces the Government had in any case to organize the supply of food for the troops; at Rome they could draw on the supply brought in for the capital. It might have appeared simpler in the latter case to let them buy in the market and pay them proportionately more, while in the former it was easier not to charge for the rations issued, instead of making deductions from an increased rate of pay.

²² P. Gen. Lat. I.

²³ Doubted by Johnson, Economic Survey of Ancient Rome, 2, p. 670 but, as pointed out by Premerstein, Klio, iii. p. 2 the soldiers concerned bear the three nomina and the same papyrus contains other documents undoubtedly referring to a legion.

²⁴ Ann. i. 17, 6.

²⁵ Veg. ii. 14.

²⁶ Dom. 7, 3, ‘addidit et quartum stipendium militi, aureos ternos.’

²⁷ P. Gen. Lat. 1, cf. p. 59 below.

²⁸ Dio, lxvii. 3, 5 = Zon. xi. 19,

²⁹ P. Gen. Lat. 4, probably referring to legionary pay after the increase, shows three equal instalments each of 297 silver drachmae; on the amount cf. p. 59 below.

³⁰ Dio, lv. 23–25.

³¹ , Dio, lv. 24, 9.

³² Tac., Ann. i. 17Google Scholar. The demand for an increase to one denarius a day was not pressed.

³³ Domaszewski adduced a further argument from the praemia fixed by Augustus in A.D. 5. Holding that they were fixed when the legionary's rate was 150 denarii and the praetorian's 500 he explains the praemia as respectively 20 × 150 and 10 × 500. But there seems no sufficient reason why the government should have chosen to give the praetorian a gratuity fixed at only ten times his annual rate of pay when they fixed that of the legionary at twenty times his rate. The relation is roughly that which had originally existed between legionary and praetorian rates of pay; probably the same proportion had been observed in granting parcels of land or money in lieu thereof, and this proportion had not been disturbed in respect of such grants even after the increase of praetorian pay.

³⁴ Dio, liii. 11, 5, .

³⁵ Ann, i. 17.

³⁶ Ib. i. 8: Suet. Div. Aug. 101.

³⁷ Suet. Div. Aug. 25, 2, supported by Dio, lv. 31, 1; Vell. ii. 111, 1 for the first occasion (A.D. 7.), Dio, lvi. 23, 3 for the second (A.D. 9.). Cf. Macr. Sat. i, 11, 32, ‘Caesar Augustus in Germania et Illyrico cohortes libertinorum complures legit, quas voluntarias appellavit.’

³⁸ Suet. Div. Claud. 10, 4. (Jos., . A.J. xix. 247Google Scholar, gives 5,000). For Nero, Tac, Ann. xii, 69Google Scholar; Dio, lxi. 3, 1. After the increase of pay by Domitian, Marcus Aurelius and Verus gave only 3,000, according to Dio, lxxiii. 8, 4 (though SHA. Ant. Philos. 7, 9 gives 5,000); Pertinax, however, promised only 3,000 (Dio, lxxiv. 1, 8, 4); thus throughout the first two centuries the donatives do not appear to have stood in any permanently fixed relation to pay. Cf. what is said of Severus below. Passerini, Athenaeum, xxiv, p. 155, points out that Macrinus promised 5,000 denarii to his troops and that this sum is not a multiple of the annual pay of 750 denarii proposed by Domaszewski (and accepted below), Dio, lxxviii. 34, 1.

³⁹ Domaszewski's calculation of the distribution of the 100,000 denarii which Hadrian gave to the urban plebs and the Army on the occasion of the adoption of L. Aelius Caesar is too highly conjectural to be of any value in supporting these conclusions.

⁴⁰ Dio, lxxiv. ii, 5, .

⁴¹ Dio, lxxi. 3, 3.

⁴² Cf. Heichelheim, , Klio, xxxvi. p. 105Google Scholar.

⁴³ Passerini, loc. cit.; Hdn., iii. 8, 4, . I agree with Passerini that must refer to stipendium and not annona, as again iv. 4, 6. An increase in pay under Severus is confirmed by ILS. 2438, 2445, 9099, 9100 (dedications to Severus and his house ‘ex largissimis stipendis et liberalitatibus quae in eos conferunt’).

⁴⁴ 5, 2.

⁴⁵ Hdn. l.c.; SHA. Sev. 12, 2, ‘denique militibus tantum stipendiorum quantum nemo principum dedit’. Passerini, Coorti, p. 109 n. 2, Athenaeum p. 153 n. 3, is wrong in referring this to donatives and not taking ‘stipendia’ in its natural sense. Although stipendium can mean donative, cf. SHA. Macr. 5,7, this sense is less natural, nor is there good evidence that Severus gave unusually large donatives, cf. n. 54.

⁴⁶ P-W. s.v. ‘anularium’.

⁴⁷ ILS. 2354.

⁴⁸ ILS. 2438, cf. 9097 n. 9.

⁴⁹ ILS. 9097.

⁵⁰ P. Lat. Gen. 1 (Klio, iii. p. 23).

⁵¹ ILS. 9100.

⁵² Dio, xlvi. 46; SHA. Sev. 7, 6.

⁵³ Truppensold pp. 232 ff.

⁵⁴ It is in fact an unexplained paradox that Severus' recorded donatives throughout this reign were less genetous than those of his predecessors.

⁵⁵ Dio, lxxviii. 36, 3.

⁵⁶ Hdn., iv. 4,8, says that after Geta's death, . For the meaning of σιτηρέσιον cf. the parallel passage in SHA. Carac. 2, 6, ‘addidit denique his quasi fidelioribus erga se stipendium’.

⁵⁷ See n. 56.

⁵⁶ Dio, lxxviii. 19. As Passerini, , Athenaeum, xxiv, p. 154Google Scholar, remarks, means that this was to be a second donative of 750 drachmae, not a donative equal to the annual stipendium.

⁵⁹ Dio, lxxvii. 24 = Exc. Val. 394, For ἐς χιλίας Domaszewski conjectured ἐς χιλίας, but this is quite uncertain.

⁶⁰ lxxviii. 36, 3.

⁶¹ Domaszewski held that centurions received no increment from Caracalla; for rejection of this view see below.

⁶² Veg., ii. 6. Assuming 5,000 legionaries on an average in the 33 legions, an increase of 250 denarii a head gives 36,250,000 denarii; the balance is easily accounted for by increases to officers, principles and men in other units.

⁶³ ILS. 9097.

⁶⁴ Frank, T., Ec. Survey, v. p. 92Google Scholar and literature there cited. The doubling and trebling of prices in Egypt may have been partly due to local conditions and is more than can be accounted for by the debasement of the coinage.

⁶⁵ Passerini, op. cit., pp. 151–6, in criticising Domaszewski fails to record all his arguments and misses their cumulative force.

⁶⁶ P. Lat. Gen. 1, see von Premerstein, , Klio, iii. pp. 1 ffGoogle Scholar. Lesquier, , L'armée romaine d'Egypte d'Auguste à Dioclétien, (Cairo, 1918) pp. 248 sqqGoogle Scholar; Johnson, op. cit. pp. 670 ff.

⁶⁷ P. Lat. Gen. 4; Johnson loc. cit.

⁶⁸ Op. cit. p. 670.

⁶⁹ See note 6. C. Gracchus provided that the State should pay for the soldier's clothes (Plut. C. Gracc. 5), but this was evidently of short-lived effect.

⁷⁰ Frank, T., Ec. Survey, i. pp. 188–9Google Scholar, on the basis of Cato, De r.r. 56 ff. reckons the annual cost of food for a slave at 69 denarii a year (if wheat is taken at 3 sesterces, cf. p. 192, this should be increased to 75). A soldier received the same ration of wheat that Cato allowed for a slave (cf. Polyb., vi. 39. 13), and can hardly have consumed less of other foodstuffs. His boots and clothing must have cost more than those of a slave, to whom Cato allowed a tunic, blanket and pair of wooden shoes every other year, at a cost guessed by Tenney Frank to be 9 denarii annually. An Egyptian legionary under Domitian spent as much as 60 denarii on clothing, but he might have been accounted ‘nitidus’ in Republican days. The cost of arms and probably tents has still to be added.

⁷¹ Tac., Ann., i. 17Google Scholar.

⁷² Ib., xv. 72, 2, ‘Addidit sine pretio frumentum, quo ante ex modo annonae utebantur’; Suet. Nero 10.

⁷³ Polyb., vi. 39, 13.

⁷⁴ Tac., Ann., xv. 39. 2Google Scholar, ‘Pretium frumenti minutum usque ad ternos nummos’—evidently a minimum. According to Pliny, , N.H., xviii. 90Google Scholar, ten sesterces was an average price for a modius of meal.

⁷⁵ Wilcken, Gr. Ostraka, 1128–46; receipts for corn etc. given ἅχρι τοῦ ὀψωνίου (1129), ἅχρι λόγου συνάρσεως (1135), ἐν προχρ[εία], cf. 1131, 1145. Van Berchem, , L'Annone militaire dans l'Empire romain au troisième siècle (Paris 1937) p. 136Google Scholar, cites these for his view that the stipendium was supplemented by payments in kind as early as Severus; they seem to me to make against it. It is also quite uncertain whether there is any reference to the military annona in Dio, lxxviii. 34, 3, describing Macrinus' restoration of Caracalla's concessions to the Army: . The parallel passage in 28, 2, refers only to the reduction of pay () and deprivation of privileges and immunities () which Caracalla had given them.

⁷⁶ Fayum Towns, CV, col. 2, 18.

⁷⁷ Ib. Col. II for withdrawals by soldiers of sums deposited no doubt with the signifer, to pay debts to the unit.

⁷⁸ Suet., Dom. 7, 3Google Scholar. Veg., ii. 20, describes a practice (which he calls ancient) by which the soldier was required to deposit half of his donatives with the signiferi; this rule was in force at a time when the troops were paid in kind and literate soldiers exceptional; i.e. in the third century, if not later. Veg. does not mention Domitian's restriction which may have been shortlived.

⁷⁹ Veg. l.c. Heichelheim, , Ec. Survey, iv. p. 180Google Scholar, puts the annual cost of living in Syria during the second century at between ioo and 140 denarii. Yet in Syria prices were apparently higher than in Egypt. This gives an idea of the soldier's opportunity for saving.

⁸⁰ De vir. ill. 73, 1.

⁸¹ RG. 3, 3. ‘Millia civium Romanorum sub sacramento meo fuerunt circiter [quingen]ta. Ex quibus dedu[xi in colon]ias aut remisi in municipia sua stipen[dis emeri]tis millia aliquanto [plura qu]am trecenta et iis omnibus agros a[dsignavi] aut pecuniam pro p[raemis mil]itiae dedi’.

⁸² RG. 17, 2.

⁸³ Hardy, Contra, CQ. xiv. pp. 187 ff.Google Scholar, whose view is, in my judgement, untenable even with his assumption that during most of his reign Augustus had only 22 legions.

⁸⁴ RG. 15, 3.

⁸⁵ Dio, li. 3, 1, 4, 5, .

⁸⁶ Antony had 30 legions, Grueber, , Coins of the Roman Republic in the British Museum, ii. 526–30Google Scholar, but probably two-thirds of the legionaries were Orientals, Kromayer, , Hermes, 33, 68Google Scholar, n. 1, cf. Jos., AJ. XIV. 15, 10Google Scholar; Cuntz, , Jahreshefie, XXV. pp. 70–81Google Scholar; and Augustus is concerned in RG. 3, 3 only with Roman citizens, of whom Antony had perhaps only 50,000; possibly Augustus includes these.

⁸⁷ Inferred from BGU, IV. 1104, 1108.

⁸⁸ The lowest date assumes 16 years service.

⁸⁹ RG. 16, 1. Hardy, op. cit. p. 190 rejects Hyginus statement (p. 77 Lachmann) that in 29 B.C., as well as in 14, some veterans were settled outside Italy, and suggests that this is a confusion arising from the settlement in colonies overseas of expropriated Italians; Dio, li. 4, puts the colonies in Italy only and RG. 16,1 is compatible with this. In that case each veteran received land worth 5,000. sesterces. But this seems uncertain.

⁹⁰ RG. 15, 3.

⁹¹ RG. 3, 3, ‘iis omnibus’ is against a supposition that soldiers discharged in these years received praemia from the aerarium, not from Augustus.

⁹² For praetorians after A.D. 5 see following note. ILS. 2338 for an aquilifer and curator veteranorum who received ‘praemia duplicia’ in 29 B.C.

⁹³ Dio, lv. 23, . See also note 33.

⁹⁴ See note 59. Nischer's statement (op. cit. p. 527) that the praemia were increased in proportion to the increase in rates of pay is unjustified.

⁹⁵ iii. 3, 8.

⁹⁶ Ec. Survey, V. p. 170

⁹⁷ Meagre evidence for Syria only in Ec. Survey, iv. pp. 149–52Google Scholar.

⁹⁸ Dio, li. 4.

⁹⁹ 6%, Col., iii. 3, 8; 5% in the alimentary inscriptions.

¹⁰⁰ Tac., Hist., iv. 19Google Scholar; the Batavians demand ‘augeri equitum numerum’ as part of a programme for bettering themselves. In Hadrian's allocutio to the troops in Africa (ILS. 2487) he says, ‘Difficile est cohortales equites etiam per se placere, difficilius post alarem exercitationem non displicere.’ If the equites alares were expected to be more efficient, they were probably better paid.

¹⁰¹ Cheesman, Auxilia of the Roman Imperial Army, p. 35.

^101a L'Occupation romaine de la basse Egypte (Paris 1945), pp. 14–6Google Scholar. I am indebted to Mr. Eric Birley for knowledge of this work. Lesquier had already used the papyrus, but, owing to a misinterpretation, dated it to 180–3.

^101b Tac., Hist., iv. 19Google Scholar.

^101c As Marichal accepts Domaszewski's view that Domitian added a fourth, and Commodus a fifth stipendium to legionary pay, each of 75 denarii, he ought to have argued that it stipendium in the papyrus does not mean the total annual pay, it must be not a third but a fifth thereof. This would certainly be impossible; but the inference to be drawn is not that stipendium can here only mean a whole year's pay, but as argued above, that increases of pay did not take the form of the addition of another equal instalment to the 3 which the legionary received before Domitian. Incidentally Marichal's own theory requires him to suppose that Domitian increased auxiliary pay by one third, although it continued to be paid in a single instalment each year. This is a further illustration of the illogicality into which Domaszewski's theory of the nature of stipendium leads its adherents.

^101d In that case the Batavian demand for ‘duplex stipendium’ was strictly a demand for an increase from 135 to 225 denarii; but an objection cannot be founded on this, since ‘duplex’ need not be taken in a mathematically exact sense, cf. n. 14.

¹⁰² See Johnson, op. cit. pp. 304 sqq. for these and further examples. Johnson reckons the expenses of the labourer in 79, including taxes but not shelter, at 220 drachmae, but allows that this may be a little too high.

¹⁰³ It was an auxiliary cavalryman who paid 103 denarii for arms, cf. n. 76. It is true that P. Berlin 6866 shows only one deduction from the auxiliary's wage, a collatio for an unspecified purpose of 4 denarii 22½ obols. (I am not convinced by Marichal's argument that this should be identified with the very much smaller contribution ‘ad signa’ made by the more highly paid legionaries.) But it must be remembered that the men in question were being employed away from their unit, and were therefore not drawing on army rations; instead they must have had to buy food for themselves. As for clothes, they were not necessarily a charge on every instalment of pay; and this is more obviously true of arms, which would only have required periodic renewal. It is, however, conceivable that in Severus' reign soldiers in all units were less subject to stoppages from pay.

¹⁰⁴ P. Hamb. 39.

¹⁰⁵ Fayum Towns, CV, col. 2, lines 2 and 18, records an auxiliary of an ala as having saved 1562 denarii; admittedly cavalrymen must have been paid more than auxiliary footsoldiers (supra), but hardly so much more as to make it possible that one of them should have saved the equivalent of nearly 20 years' wages of an infantryman, if Marichal were right. For auxiliaries with freedmen, land and other property see e.g. ILS. 2500, 2531, 2567; Wilcken, Chrest. 376; BGU II 591, 455; I, 4. As for auxiliaries having families, a high proportion of the soldiers of P. Berlin 6866 were born ‘castris’.

¹⁰⁶ Cheesman, op. cit., p. 46. Macrinus' plan to pay new recruits to the army at the rates fixed before Caracalla while continuing those rates for the benefit of soldiers already in the service (Dio, lxxviii. 28, 3) may have been based on experience in cohortes c. R.

¹⁰⁷ ILS. 2574.

¹⁰⁸ Rangordnung, pp. 70–2.

¹⁰⁹ This view is based (i) on the statement of Veg., ii. 7, ‘duplares duas, sesquiplicares unam semis consequebantur annonam’; (the existence of a rank paid at double the private's rate is shown also by ILS. 470, mentioning duplarii under Elagabalus; 9098, dedication under Severus by ‘[milites cons]ecuti dupli stipendii beneficium’;) (ii) on the fact that the inscriptions of the scholae principalium (ILS. 2354, 2438, 9097, 9100) show an anularium paid to the cornicen (an immunis) of 500 denarii, to librarii and exacti of 800, to the cornicularius praefecti of 1,000, to the optiones ad span ordinis of 1,500. Domaszewski assumed with great probability that these payments were the equivalent of one year's pay.

¹¹⁰ See note 6.

¹¹¹ App., BC. iv. 100Google Scholar (speech of Cassius), . (The last words, in particular, show that the differing amounts of the donatives promised to different ranks rested on a basis of correspondence; and it is hard to see to what they could have corresponded except the differences between the scales of pay applicable to those ranks.) Cf. 120 (speech of Antony), . This makes it probable that in 42 B.C. tribunes were paid only twice as much as ordinary centurions; and primi ordines and even primi pili presumably received less than tribunes. Later all these officers seem to have secured substantial increases in pay, differentiating their rank more sharply from ordinary centurions and other ranks, see below.

¹¹² The evidence for such an increase is that ILS. 2452 (Marcus) mentions 7 centurions in the first cohort, 8 in the sixth, 7 in the eighth whereas CIL. xiii. 6801 (Severus) mentions 11 in the first cohort. Domaszewski identified the 4 extra centurions with those employed on the staff, viz princeps praetorii, centurio strator, commander of the equites singulares, and exercitator of the equites singulares and held that Severus made them all centurions of the first cohort, i.e. in his view primi ordines (Rangordnung, p. 98).

¹¹³ Domaszewski, Rangordnung, p. 111, cites ILS. 7178, ‘T. Aurelio T. fil. Papir(ia) Flavino primipilari honorat[o] a divo Magno Antonino Aug(usto) (sestertium) (quinquaginta) milia n(ummum) et [sestertium] viginti quinque [milia nummum] et gradum promotionis [ob] alacritatem virtu[tis adv]ersus hostes Ca[rpos] et res prospere et va[lide ges]tas’. He infers (i) that the two donatives were given to him on reaching centurionate and primi ordines respectively; as the scale of donatives always corresponds to that of the pay the doubling implies double pay for primi ordines; (this seems probable;) (ii) the donative of 6,250 denarii is explicable as 5 units of 1,250 denarii, but the legionaries' pay was reckoned in units of 75 denarii, which stand in no clear relation to units of 1,250; on the other hand the praetorians' pay was reckoned in units of 250 denarii, consequently the unit used for reckoning the centurion's donatives, and therefore his pay was 5 times that of the praetorian. The reconstruction is conjectural and leads to the difficulties stated above. It is quite easy to suppose that Flavinus received half a year's pay as donative.

¹¹⁴ See previous note.

¹¹⁵ Domaszewski held that they received grants on discharge of 150,000 denarii, cf. Suet. Gaius, 44: ‘in exercitu recensendo plerisque centurionum maturis iam et nonnullis ante paudssimos quam consummaturi essent dies, primos pilos ademit, causatus senium cuiusque et imbecillitatem; ceterorum (? other primi pili or centurions other than primi pili) increpita cupiditate commoda emeritae militiae ad† sescentorum milium summam recidit’. If ‘sescentorum’ is right, the passage must refer to primi pili and they then received a grant on discharge, after Gaius reduced it, 50 times greater dian the legionary's. This is not impossible; their pay may have been 66 times greater; but it is suspiciously high. Emendations of the text are arbitrary and uncertain. A theory cannot be built on this passage.

¹¹⁶ Cf. Suet., Tib., 48, 2Google Scholar, ‘ex senio mortem, ex morte compendium captans’.

¹¹⁷ He took their grants on discharge (as to which see note 115) to be 10 times their annual pay.

¹¹⁸ Domaszewski cited CIL. xiii. 3162, the inscription honouring Sennius Sollemnis, who received a salary of 6,250 denarii for 6 months' service with Claudius Paullinus, legate in Britain, in what capacity is obscure—‘ad legionem sextam adsedit’. This apparently comes from Elagabalus' reign, but on the hypothesis here presented ordinary centurions already received 12,500 denarii at this time. If Domaszewski's reconstruction of tribunician salaries etc. from this inscription were right, it would apply to the early third century, and the rates of pay proposed for centurions and consequently for lower ranks would be too high. (Dessau, , Gesch. d. röm. Kaiserzeit, i. p. 248Google Scholar, is surely wrong in holding that the tribunus semestris was paid for a full year; he cites no evidence.)

¹¹⁹ Dio, lxxviii. 22, 5.

¹²⁰ Hdn., vi. 8, 8; immediately after his usurpation . Domaszewski, , Rh. Mus., lviii. p. 383Google Scholar, n. 1, gratuitously supposed that Alexander Severus had reduced pay to the level fixed by Septimius, and that Maximin therefore increased it again only to 1,000 denarii; as Passerini, points out, Athenaeum, xxiv, pp. 157–8Google Scholar, this would be contrary to what else is known of Alexander's subservience to his soldiers. Passerini's own hypothesis that Maximin raised the pay only of those serving with him is, as he recognises, unnecessary, and seems improbable; if Herodian's words are to be limited at all in their reference, they must be limited to the corps of recruits of whom Maximin was in command and who had raised him to the purple, but it would have been too paradoxical to pay recruits more than veterans.

¹²¹ Veg., ii. 8. Rostovtzeff, Storia Ec. e Sociale, p. 556 n. 26 (English edition, p. 622) reads in the Aragua, inscription (CIL. iii. 14191Google Scholar) ‘mil(item) cen(tenarium) frum(entarium)’; if this were certain, it would give further support to the hypothesis that as early as Philip's reign a centurion was receiving 100,000 sesterces—on the assumption that the frumentarius was paid as highly as a centurion.

¹²² Grosse, , Römische Militärgeschichte von Gallienus bis zur Beginn der byzantinischen Themenve-fassung (Berlin, 1920), pp. 115–7Google Scholar.

¹²³ Loc. cit.

¹²⁴ ILS. 569.

¹²⁵ E.g. praefecti legionis, ILS. 545, CIL. iii. 3529, tribunes of the praetorian guard, 3126, ILS. 1332.

¹²⁶ ILS. 2777.

¹²⁷ ILS. 2778.

¹²⁸ ILS. 5695.

¹²⁹ ILS. 9479.

¹³⁰ The supplements proposed by Domaszewski (Rangordnung, pp. 185 ff.) for the latter part of the inscription which associate the title ‘protector’ with the higher ranks that he subsequently held of tribune and legionary prefect are quite uncertain and afford no foundation for Domaszewski's theory of three grades of protectores. (Equally, as is well known, Domaszewski was wrong in holding that the protectorate was conferred only on praetorian centurions, cf. ILS. 2777, 9479, CIL. xiii. 8273, 8291.)

¹³¹ Ges. Schr., viii. p. 439 (Eph. Ep., v. pp. 136–7).

¹³² See also ILS. 9478, δουκη(ναρίῳ) ἑκ π[ρο]τηκτόρων.