The Supposed Roman Manpower Shortage of the Later Second
Transcription
The Supposed Roman Manpower Shortage of the Later Second
The Supposed Roman Manpower Shortage of the Later Second Century B.C. Author(s): J. W. Rich Source: Historia: Zeitschrift für Alte Geschichte, Vol. 32, No. 3 (3rd Qtr., 1983), pp. 287-331 Published by: Franz Steiner Verlag Stable URL: http://www.jstor.org/stable/4435854 Accessed: 06/09/2010 15:49 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=fsv. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. Franz Steiner Verlag is collaborating with JSTOR to digitize, preserve and extend access to Historia: Zeitschrift für Alte Geschichte. http://www.jstor.org THE SUPPOSED ROMAN MANPOWER SHORTAGE OF THE LATER SECOND CENTURY B.C. I. Introduction In early Rome, as in other archaic states, only those who could afford to equip themselves could serve in the army. In due course, pay for military service was introduced and arms came to be supplied by the state.' But Roman conservatism ensured that the link between property ownership and military service was maintained, even when Rome acquired extensive military commitments overseas. Until the end of the second century service in the legions was, except in grave emergencies, restricted to freeborn citizens possessing a certain property qualification, who were known as assidui. Marius was the first commander to enrol men into the legions without regard for the property qualification. "I am very grateful to Professor P. A. Brunt, Mr. M. H. Crawford and the members of Professor H. B. Mattingly's seminar at Leeds for their comments on earlier drafts of this paper. All dates are B. C. unless otherwise stated. I refer to the following modern works in an abbreviated fashion: A. E. Astin, Scipio Aemilianus (1967), Cato the Censor (1978); A. H. Bernstein, Tiberius SempronisusGracchus, Tradition and Apostasy (1978); P. A. Brunt, Italian Manpower 225 B.C. A.D. 14 (1971), Social Conflicts in the Roman Republic (1971); M. H. Crawford, Roman Republican Coinage (1974) (= RRC), The Roman Republic (1978) (= RR); Les "Divaluations"a Rome: epoque republicaine et imperiale (Rome, 13-15 novembre 1975) (1978); D. C. Earl, Tiberius Gracchus, A Study in Politics (1963); E. Gabba, Esercito e societd nella tarda repubblica romana (1973) (English translation: Republican Rome, the Army and the Allies (1976)); W. V. Harris, War and Imperialism in Republican Rome 327-70 B.C. (1979); M. K. Hopkins, Conquerorsand Slaves (1978); P. Marchetti, Histoire economique et monitaire de la deuxieme guerre punique (1978); T. Mommsen, Romisches Staatsrec/b&(1887-8); C. Nicolet, Le mitier du citoyen dans la Rome ripublicaine (1976) (English translation: The Worldof the Citizen in Republican Rome (1980)); J.C. Richard, Les Origines de la Plebe Romaine (1978); H.-C. Schneider, Das Problem der Veteranenversorgungin der spateren romischen Republik (1977); Y. Shochat, Recruitment and the Programme of Tiberius Gracchus (1980); R.E. Smith, Service in the Post-Marian Roman Army (1958); D. Stockton, The Gracchi(1979); R. Thomsen, King Servius Tullius(I980); A. J. Toynbee, Hannibal's Legacy (1965). l There may be substance in the traditional association of the introduction of pay with the start of the war with Veii (Livy IV.59.11; Diod. XIV.16.5): see C. Gatti, Acme 23 (1970), 131-5; Crawford in L'Italie priromaine et la Rome ripublicaine (Melanges offerts a Jacques Heurgon) (1976), I. 197-207, especially 204-5. Polybius tells us (VI. 39.15) that deductions from pay were made x6v TLVOg 0tXoij 7tQo(6E-qft6)OL, i. e. for replacement arms. I infer from Plb. VI.21.6-7 that in his day soldiers were supplied with arms free of charge when they enrolled: so Gabba, ES23-4; Brunt, IM 405; Nicolet, Mitier 156. For the view that soldiers still had to provide their own equipment see Toynbee 11.514-5; Marchetti 244, 247-8. The members of the first class presumably had to pay for the loricae which they alone wore (Plb. VI.23.15). 288 J. W. RICH It is commonly supposed that by the later second century the Romans' insistenceon retainingthe propertyqualificationcreateda gravemanpower problem.2This view may be summarizedas follows. In the course of the century many peasantproprietorsgave up their holdings; most of this land was taken over by large landholdersand worked by slave labour. The dispossessedpeasantsno longersatisfiedthe propertyrequirement,and so the numberof assiduidropped.The drop was so steep that the remainingassidui could not meet the demandsof the levy without considerablehardship,and men came to fearfor Rome'ssecurity.Variousattemptswere madeto resolve the problem: the property qualification,already reduced during the HannibalicWar,was at some point reducedagain,and at leastpartof the purpose of the agrarianschemesof Laeliusand the Gracchiwas to increasethe number of assidui.But these expedientswere of no avail. The only solution was to abandonthe propertyqualification,but for long no one was preparedto takea step so repugnantto Roman sentiments.At last Mariusmadethe breakand later commandersfollowed his example. This doctrinerequiresus to supposethatthe Romanswere so unyieldingin their devotion to the principlethat only men of propertyshould servein the armythat they allowedan entirelyartificialmanpowercrisisto develop.If the Romansdid behavein this way, they displayeda rigiditywhich was not only foolish but uncharacteristic:althoughthey were a conservativepeople, their conservatismwas generallytemperedby pragmatism.It would be wrong to asserta priorithatthe Romanscannothaveactedin this way, but we oughtnot to acceptthat they did so without good evidence.In this paperI shallexamine the various argumentson which the doctrineoutlinedabove rests. I hope to show that the casefor it is veryweakandthatour evidencecanbe satisfactorily accountedfor without supposingthat a shortageof assiduidevelopedin the later second century.3 II. The Burdenof the Levy and the Number of Assidui This section is devoted to two topics on which some preliminarywords of clarificationare necessary:the natureof the burdenwhichthe levy imposedon the assidui,and the size of the class of assidui. 2 See especially Gabba, Athenaeum 27 (1949), 173-309 and 29 (1951), 171-8 = ES 1-56 = RRAA 1-24, 171-282; Smith 1-10; Earl 30-40; Astin, Scipio 161-174, 196-7; Brunt, IM 75-7, 402-8 and SCRR 11-19; Schneider 17-28, 65-101; Crawford, RR 99-108; Stockton 8-9, 31-5. The rejection of the doctrine that there was a shortage of assidui is one of the main contentions of Shochat, but in other respects his views differ widely from those which I put forward in this paper. Scepticism about the supposed shortage of assidui has also been expressed by J. F. Lazenby, Hannibal'sWar(1978),245. The SupposedRomanManpowerShortageof the LaterSecondCenturyB.C. 289 We are very poorly informed about how the levy was conducted.4 Polybius tells us that when a levy was held all eligible citizens were required to present themselves at Rome for selection, but there are grave difficulties in supposing that the procedure he describes was in force in his own day.5 Although there is no evidence for their involvement before the first century, it may be that already by the time of the Hannibalic War local authorities were responsible for the conduct of the levy over much of the ager Romanus.6 Conquisitores may sometimes have been despatched by the central government to seek out recruits, although before the first century this practice is attested only at two critical moments in the Hannibalic War.' However, it seems to me likely that the principle still held good that, when a levy was held, those eligible should present themselves for selection, even if many did not have to go to Rome. When a sudden crisis occurred, the usual procedures of the levy could be dispensed with and a tumultus declared. Men normally exempt might then be enlisted, and the commander would enlist men as expeditiously as possible, both at Rome and on the way to the trouble spot.8 Unless they enjoyed exemption (vacatio), assidui were liable for military service during the period in which they were iuniores, that is, between their seventeenth and forty-sixth birthdays; in exceptional circumstances boys under 17 and men of 46 and over might be enlisted.9 Infantrymen might be called on to serve up to sixteen campaigns (stipendia), or twenty in emergencies; equites were liable only for ten.10 In early days legions were normally discharged after a single campaigning season, so that a man might be enlisted many times but each individual period of service did not last more than a few months. From the time of the Punic Wars periods of continuous service lasting several years became common. In the second century legionaries enlisted for wars fought overseas in an area where Rome did not regularly maintain an army normally served for the duration, and legionaries serving in a province which was permanently garrisoned (as were the two Spains and after 4 In general on the workings of the levy see J. Marquardt, Romische Staatsverwaltung2 II.380-9; W. Liebenam, RE V.591-615; Brunt, IM625-634; Nicolet, Metier 128-140. 5 Plb. VI. 19-2 1, with the criticisms of Brunt, IM 625-7. 6 Brunt, loc cit. holds that only those living near the city still had go to Rome; he is followed by H. Galsterer, Herrschaft und Verwaltung im republikanischen Italien (1976), 105-6. Polybius is defended by E. Rawson, RBSR 39 (1971), 15 and F. Gschnitzer, Hermes 109 (1981), 69-71: they hold that all except those who lived in colon*ae and municipia still had to go to Rome. 7 Livy XXIII. 32.19; XXV 5.5-9. 8 Cf. RE V. 603-4; Brunt, IM 629-630. The lower age limit: Tubero ap. Gell. X.28.1 (= fr. 4P.); Livy XXII. 57.9, XXV.5.8, XXVII.11.15; Plut. CG 5. The upper age limit: Plb. VI.19.2; Cic. Sen. 60; Varro ap. Censorin. XIV.3; Livy XLIII.14.6; Dion. Hal. IV.16.3; cf. Livy XL.26.7, XLII.31.4, 33.4. On vacationes see Mommsen, Str. III.241-4. 10 Plb. VI.19.2. The figure of sixteen campaigns depends on an emendation: see Walbank ad loc. 19 290 J. W. RICH 146 Macedonia) were discharged only when the senate and their commander thought fit. There is some evidence that a man was deemed to be entitled to discharge after six years' continuous service."1Men who had undergone a long period of continuous service were not exempt from being enlisted again. We have no means of telling how much service assidui were on averagecalled upon to perform throughout their lives.'2 Those who fell below the property qualification were known as proletarii or capite censi.3 Like freedmen, they were normally liable only for service in the fleet and were enrolled in the legions only in emergencies.14We do not know how often such enrolments took place. We are told that opificum quoque vulgus et sellularii, minime militiae idoneum genus were enlisted in 329 and freedmen in 296, but these notices may not be historical.'5 There is unimpeachable evidence that proletarii were enrolled as troops at the beginning of the Pyrrhic War."6In the First Punic War the proleta ni were no doubt fully employed in manning the fleets. In the immediate aftermath of Cannae an army had to be amassed as quickly as possible from the manpower available in the immediate environs of Rome; the dictator M. lunius Pera enlisted as volunteers both slaves (who came to be l Livy XL. 36. 10; App. Iber. 78. Not all were so fortunate, but I doubt whether we can safely infer from Lucilius 509-10 M (of whose context we know nothing) that there were cases of men serving in Spain for continuous periods three times as long. 12 On length of service see Smith 6-7; Toynbee II. 72-80; Astin, Scipio 169-170; Brunt, IM 399-401; Harris 44-5. 13 The capite censi were those whose property was so low that the censors did not record it but registered them by their caput alone. On the etymology of assiduus and proletarius see G. Wesenberg, RE XXIII. 631-2; Richard 367-371. (The attempt of A. Pagliaro, Helikon 7 (1967), 395-401, to deny the derivation of proletartus from proles is unconvincing). I have followed the usual view that the capite censi were not a sub-class of the proletarii (as stated by Gell. XVI.10.I 1-14) but identical with them (so Festus 253L; Cic. Rep. II.40 seems to me ambiguous on this point). There is no ancient authority for the statement of Nicolet, Metier 109 that the capite censi were those who declared only their own person and the proletarii those who had children to declare. 14 Plb. VI.19.3; Gell. XVI.10.1 1-14. For the novelty of Marius' enrolment see the sources cited below n. 176. On military service by the proletarii before Marius see Gabba, ESI I ; Brunt, IM 395, 402; Shochat 27-31. 15 Livy VIII.20.4, X.21.4. 16 Oros. IV.1.3; Aug. Civ. Dei I11.17. Orosius implies that the enrolment took place at the time of the first campaign against Tarentum in 281, but it cannot have been necessary then. It is usually supposed that the proletarii were enrolled as a garrison for the city of Rome when Pyrrhus landed in spring 280. Two obscure fragments are commonly associated with this incident: Ennius Ann. 183-5V; Cassius Hemina 21P tunc Marcus praeco primum proletanios armavit. On this interpretation of the corrupt Hemina fragment Marcus is emended to MarCiusand praeco to either praetor or procos.; the Marcius is identified as Q. Marcius Philippus (cos. 281) (see MRR I. 191-2). If this interpretation is correct, we should accept Hemina's authoriy that this was the first (but not the last) emergency enlistment of proletarii, rejecting the passages cited in n. 15 as unhistorical. The Supposed Roman Manpower Shortage of the Later Second Century B.C. 291 known as volones)and men held in prison for debt."7As the HannibalicWar draggedon, Rome's reservesof legionarymanpowercameunderstrain.Livy recordsa numberof measurestakenin responseto this shortage,18 but we hear of only one instance of the enlistment of non-assidui: the recall to the standardsin 207 of the voloneswho had disbandedthemselveson the deathof theircommanderin 212.'9Theremay havebeen a few enrolmentsof freedmen andproletariinot mentionedby our sources,but we should, I think,conclude that, apartfrom the men recruiteddirectlyafterCannae,the legionaryforces fielded in this war were almostentirelycomposedof assidui.20 Why more use was not made of non-assiduimust remaina matterfor conjecture.2' We never hear of non-assiduibeing enlisted in the legions between the HannibalicWar and Marius' break with tradition.22Some may have been enrolled when a tumultuary levy was held. Livy records declarationsof tumultusin 198, 193, 181 and 178; the first was occasionedby a slave-risingat Setia,the other three by setbacksin northernItaly.23We cannot tell whether any tumultuarylevies were held in the period from 167 for which we lack Livy, but there cannot have been much occasionfor them now that northern Italy (exceptfor the Alpine valleys)had been pacified.24It may be thatlevying 17 Livy XXII.57.11, XXIII.14.3-4; App. Hann. 27; Zon. IX.2; Val. Max. V1I.6.1; Oros. IV.16.8-9; Fest. 511L; Macr. 1.11.30-1. The volones are discussed by N. Rouland, Les esclaves ronains en temps de guerre (1977), 45-58. Boys under seventeen were also enlisted on this occasion (Livy XXII.57.9). 18 In 214 the censors are said to have penalized those who had not served: XXIV. 18.7-9. In 212 the senate decided that boys under 17 should be enlisted and two boards of trumviriwere sent around the country districts to seek out those who had roborissatis adferenda arma: XXV. 5. 5-9. In 207 the exemption normally granted to the inhabitants of coloniae maritimae was set aside: XXVII.38.3-5. "9 Livy XXV.20.4, 22.3-4; XXVII.38.8,10; XXVIII.10.11,46.13. 20 The freedmen enlisted in 217 (Livy XXII.11.8) may have served only in the fleet (so Brunt, IM 395 n.6). On naval service by proletarii in the Second Punic War see Brunt, IM65-6, 421-2, 668-9. 21 It is also curious that our sources make no explicit mention of ingenuous proletarii; the tiumviri of 212 were ordered to inspect omnem copiam ingenuorum. In my view the solution to both puzzles may be that at this period most ingenui and almost all the ingenui living outside the city of Rome ranked as assidui. It is usually thought that the property qualification was reduced at some time during the war; on this see below section V. 22 The words expedito pauperem plebeium atqueproletarium have survived from Cato's speech De tribunis militum (ORF3 fr. 152). It is uncertain whether this should be taken as an exhortation to enlist proletarii; if so, it may not have been meant seriously. See now Astin, Cato 118 n. 46. XXXII.26.10-12; XXXIV.56. 11-13; XL.26. 6-7, 28.10; XLI.5.4. The tumultuary levy of 201 (XXXI. 2.5-6) seems to have been conducted in the Cisalpina itself (though see Briscoe ad 23 loc.). 24 Such a levy may have been held by Q. Metellus Macedonicus and Cn. Servilius Caepio when as privati cum imperio they crushed a slave revolt at Minturnae and Sinuessa in 134 or 133 (Oros. V.9.4.; Obseq. 27, 27b; see E. Badian, ANRWI.1.684-5). 19* J. W. 292 RICH officersenrolledsome proletariiwithout officialsanctioneitheras volunteers or as conscripts,25but in view of the outcry raisedby Marius'actionthis is unlikely to have been common. So in this period as before the legions were composed almost exclusivelyof assidui. The demands made on the assiduiduring the Hannibalic War were enormous. In 218-216 citizen losses are unlikely to have been much below 50,000.26 If Livy's legion lists areauthentic,20 or more legionswere in service every year from 214 to 206, with a peak of perhaps 25 in 212. Although I am sceptical about a good deal of Livy's information on this subject, I think it unlikely that the totals he implies should be much reduced.27Calculations of the number of men serving can only be conjectural, for many of the legions must have been seriously under strength. Brunt estimates that there were never less than 60,000 legionaries in service between 215 and 207, with a peak of 80,000 in 212.28 These figures are likely to be of roughly the right order of magnitude, and may even err on the side of caution. The number of legions and legionaries in service in the years 200-168 have been calculated by Afzelius and Brunt on the basis of Livy's evidence.29 Although some points of doubt subsist, the margin of error is not likely to be great. On Brunt's figures the average number of legions in the field in these years was 8.7. If we assume that the legions were kept up to a nominal strength of 5,500, the average number of legionaries was 47,850. In thirteen years ten or more legions were in service. During the war with Antiochus Rome fielded 12-13 legions: her effort then was of the same scale as in the Hannibalic War. From 167 we lack Livy, but Brunt has calculated the number of legions in service on the basis of our fragmentary evidence and reasoned conjecture.30 Although there were no doubt some years in which more legions were fielded than appear in his figures, these do in my view show that the demands of the legionary levy, although still high, were not as great in this period as in Cf. Crawford, RR 107; Hopkins 30-1 n.40. Brunt, IM 419-420. Brunt assumes that only four legions fought at Cannae, but Polybius' figure of eight legions could be right; see now Lazenby, op. cit. n.3, 75-6. 27 For reconstructions of the number and distribution of legions during the Second Punic War based on Livy's information see G. De Sanctis, Storia deiRomaniIII.2 (1916), 631-7; Toynbee II. 647-651. A new analysis is offered by Marchetti 13-94. The criticisms of Livy's information about legions made by M. Gelzer, Kleine Schr2ften (1964), III. 220-255 are in my judgment not satisfactorily answered by Brunt, IM 644660. However, the attempt of F. Gschnitzer, Hermes 109 (1981), 59-85 to revive Kahrstedt's view that Rome's legionary forces were much smaller than Livy implies does not convince me. 25 26 28 1M417-420. A. Afzelius, Die romische Kriegsmacht wahrend der Auseinandersetzung mit den hellenistischen Grossmachten (1944), 34-61; Brunt, IM 422-6 (reckoning only two legions in Spain from 178 instead of Afzelius' four). 30 IM 404, 426-434 (with table on pp. 432-3). Earl 30-1 and A.N. Sherwin-White, JRS 67 (1977), 73 n.75a exaggerate Rome's legionary commitments. 29 The SupposedRomanManpowerShortageof the LaterSecondCenturyB.C. 293 200-168, let alone the Hannibalic War. It is unlikely that as many as ten legions were fielded except for two brief periods, 148-6 and 105-1, and possibly the year 135. The average number of legions in the intervening period according to Brunt's table is 6.4 - at full strength, 35,200 men. A particular problem is presented by Cisalpine Gaul. We have positive evidence for the presence of armies there in only nine of the years between 166 and 105, but no doubt armies served there in other years as well. Brunt arbitrarily allows for two legions there in every year down to 135 and for none thereafter except when consular governors are recorded, but he considers it not unlikely that two legions served there in most years after 135. This is possible, but, since the area had been completely pacified (except for the Alpine valleys), it seems to me much more likely that from well before 135 military activity there was only occasional. Thus in some years before 135 Brunt's total of legions may actually be too high. The fact (if such it is) that the demands made on the assidui were in general rather lower after 168 than in the earlier years of the century has implications of some importance for the present enquiry. Some decline in the number of assidui may have taken place in the course of the second century without the burden of legionary service becoming greater than it had been in the early years of the century; the burden would only be greater if the decline in the number of assidui was steeper than the decline in the number serving in the legions. Unfortunately our sources give us no indication of how many assidui there were. Some scholars hold that the census figures represent assidui.31If they are right, there cannot have been a shortage of assidui in the later second century: although the figures recorded for that period show a decline from the peak reached in 164/3, they are still higher than any earlier figures.32However, I cannot accept this view of the census figures. The following argument seems to me decisive. Livy and his epitomator have a standard formula for presenting the figures: censa sunt (Romanorum) capita tot. It is natural to conclude that this formula was used by the censors themselves when they published the figure at the close of the lustrum. There is no doubt that non-assiduiwere censi, registered by the censors, as the term capite censi attests. It is perverse to suppose that the words censa sunt were used in the formula in some special 3' This view was originally propounded by E. Herzog, Commentationes Philologae in Honorem T. Mommseni(1877), 124-142. Modern writers who have adopted it include Gabba, ES 17-21, 521-535; Earl 35-7; G. Pieri, L'Histoire du cens jusqu'a la fin de la republique romaine (1968), 177-182; J. Molthagen, Historia 22 (1973), 439-445; Schneider 79-80 n.96, 89-90, 150-6; Shochat 9-45. 32 Cf. P. Fraccaro, Opuscula 11 (1957), 54 n.1; Brunt, IM 23. Gabba fails to appreciate the contradiction between his view of the census figures and his belief in a second-century shortage of assidui: cf. ES 22-3. The census figures are conveniently tabulated by Toynbee 1.438-440 and Brunt, IM 13-14. 294 J. W. RICH sense under which such personswere excluded."Most probably,the figures representadult male citizens.'4 Two authors writing about ServiusTullius'centuriateorganizationmake statementsabout the proportion of proletarii:Cicero says that there were more men in the single centuryof proletariithan in almostthe whole of the first class, Dionysius thattheproletariioutnumberedall five classes."5 Whether there was any historicalperiod for which eitherof these statementswas valid we can only guess. Bruntholds that in 214 there were about 100,000citizensqualifiedby age, birthandpropertyfor militaryservice,andthatthe numberwas thenincreased by a reductionin the propertyqualification.'6This view is based on Livy's statement(XXIV.18.7-9) that in 214 the censorsfound that only some 2,000 iuniores(excludingthose who could pleadillnessor vacatioiusta militiae)had not servedquadriennio,takenwith Brunt'sown calculationthat by the end of 215 the number who had "ever served" in the legions was about 108,000.37 However, it is difficultto believethat therewas a totalmobilizationof assidui, let alone that the governmentallowedmattersto go so far beforeloweringthe We must, I think, supposethatthe 2,000 men singled propertyqualification.'8 out for punishmentin 214 were by no meansthe only qualifiedmen who had not servedsince218. Perhapsthe 2,000 werenot thosewho hadnot served,but those who had not presentedthemselvesat the levy. There may be deeper distortionin Livy: the 2,000 are sent to Sicilyad Cannensisexeritus reliquias and, as Brunthimself admits,Livy's accountof that force cannotbe accepted in fu1139; there seems to be some overlap between the censorial punishments recordedby Livy at XXIV.18.2-9 and XXVII.11.12-15. Two very differentsuggestionshave been made about the numbersof the assiduiin the second centuryto illustratethe supposedshortage.Astin takesa cautiousview. He thinksthat the numberof proletariiandfreedmenmayhave stood at about 80,000 in 164 and eitherremainedstationaryor showed some Shochat 12-17 has not succeeded in rebutting this argument. " So e. g. K. J. Beloch, Die Bevolkerung der griechisch-romischen Welt (1886), 312-9; Toynbee 1. 438-479; Brunt, IM 15-25. Another possibility which in my view deserves more attention than it has so far received is that the figures represent only those adult male citizens who were sui iuris (cf. F. C. Bourne, CW45 (1952), 129-134, 180-2). 35 Cic. Rep. II. 40; D.H. IV. 18.2, VII.59.6. 36 IM 64-6, 77, 405. Cf. Harris 44-5. 37 This figure rests on the unexplained assumption that the censors' investigation preceded the levy of 214 (the figure of 105,000 given at IM 64 is a misprint). However, even if those enrolled in 214 are included in the figure of those who had "ever served," Brunt is surely right that the gap between this figure and the number of iuniores implied by the census figures cannot be bridged (contra Shochat 23-5). 38 Cf. Hopkins, JRS 62 (1972), 192-3. 33 39 IM 652, 654-5. The Supposed Roman Manpower Shortage of the Later Second Century B.C. 295 increasethereafter;the decline in the census figuresregisteredafterthat date thus fell entirelyon the assidui,whose numbersfell by 8 percent or more.40He seemsto envisagethatthe totalof assiduiremainedwell over200,000.It is hard to supposethat therecan have been a seriousshortageof qualifiedmenfor the levy if the numberof assiduiwas as high as this in the later second century. Indeed, the burdenof the levy would probablyhave been lighterthan in the earlierpartof the century,when the citizen body was smallerand the number of men servinghigher. At the other extreme,Bruntsupposesthatthe numberof assiduiof military age may havefallenas low as 75,000 by the latersecond century.41It is worth bringingout the implicationsof this suggestion.On Brunt'sown reckoningthe numberof proletariimustin thatcasehavebeenat leastthreetimesas large.42As we haveseen, the averagenumberof men servingin the legionswas then about 35,000. On the simplifyingassumptionsthatthe assiduiconstituteda stationary populationwith anexpectationof lifeatbirthof 25 andthatallassiduienteredthe legionsatthe ageof seventeenandservedcontinuously,a body of 75,000assidui would have had to serve about 11.5 years each to maintainan army of this size.43 In practicethe averagelifetimeservicewould have been much higher: not all will haveenrolledat seventeen,and few servedcontinuouslyfor such a long period; death, disease and the accidentsof recruitingwill have ensured that many assiduiservedless thanthe average.It follows thatlargenumbersof assiduiwould have servedfor the full sixteenyearsfor which they were liable or longer. If the numberof assiduidid drop to somethinglike 75,000, the Romangovernment'sconductwas very remarkable.To maintainthe principle that militaryserviceshould be confinedto men of property,they imposed a crushingburdenof militaryserviceon whatwas at most a quarterof the citizen body - and that the most well-to-do - while the rest went scotfree.In the rest of this paper I consider whether there is good reason to suppose that an artificialmanpowershortagedid develop on anythinglike this scale. 40 Scipio 337. IM 77, 405; cf. Harris 44 n.5. 42 On the size of the citizen body in the second century see Brunt, IM 68-83. The ratio of iuniores to seniores is estimated at 3:1 by A. Afzelius, Die romische Eroberung Italiens (340-264 v. Chr.) (1942), 100; cf. Brunt, IM 53 n.2. On the U.N. Model Life Table for expectation of life at birth of 25, adult males over 45 constitute 26.5 per cent of all adult males. 43 This calculation is modelled on those of Hopkins 34-5, using the model life tables published in United Nations, Population Studies No. 25, Methods for Population Projection by Sex and Age (1956). On Roman expectation of life see also Hopkins, Population Studies 20 (1966), 245-264; D. Engels, CP 75 (1980), 116-8. 41 296 J. W. RICH III. AgrarianChange Agrarian change provides the starting point for the theory that a serious shortage of assiduideveloped in the later second century: large numbers of peasants, it is held, lost their holdings and with them their status as assid'i. That the second century saw important changes in the pattern of Italian agriculture can hardly be doubted. In outline the characterof these changes has long been familiar. Rome's foreign wars led to an influx of wealth for which agriculture was the most attractive investment and made available large numbers of slaves. As a result there took place a great expansion of agricultural enterprises whose permanent work force was composed of slaves. These were of two main types, according to region: medium-sized plantations, of the kind for which Cato wrote his handbook, and transhumantpasturage. At first there was probably plenty of land available for those seeking to invest as a result of the disruptions of the Hannibalic War and the confiscations of territory from ex-rebels, but as the century progressed the peasantry (citizen and allied) must have come under increasing pressure from rich men eager to expand their holdings. Long terms of military service overseas helped to undermine the ability of peasant families to resist these pressures, and many succumbed. Recent research has done much to refine this picture, and several writers have properly stressed the danger of exaggerating the deracination of the peasantry44. The sweeping generalizations of Appian and Plutarch and the preoccupations of the agronomists sometimes led older writers to give less than due weight to both general considerations and scattered literary indications which show that, although the peasantry may have been in decline, small farmers (both freehold and tenant) never ceased to play a significant part in Italian agriculture. Field surveys have now provided further evidence of the survival of the peasantry. The surveys conducted by the British School of 44 Modern discussions of agrarianchange in second century Italy include Toynbee II.286-312; Astin, Scipio 161-5 and Cato 240-266; M. W. Frederiksen, Dial Arch. 4/5 (1970/1), 330-357; Brunt, IM 269-284, 345-375; V. A. Sirago, L'agricultra italiana nel II sec. a.l. (1971); E. Gabba, Ktema 2 (1977), 269-284 (reprinted with modifications in E. Gabba and M. Pasquinucci, Strutture agrarie e allevamento transumante nell' Italia romana (III-I sec. a.c.) (1979), 13-54); Bernstein 77-97; Hopkins 1-25, 48-64; F. de Martino, Storia economica di Roma antica (1979) 1.59-123; D. W. Rathbone, JRS 71 (1981), 10-23; A. Giardina and A. Schiavone (ed.), Societa romana e produzione schiavistica. I. L'Italia: insediamenti e forne economiche (1981). On the survival of smallholders see especially Frederiksen and Rathbone, and P. D. A. Garnsey, cited below nn. 53-4. Earl 23-30 goes too far in denying that there was in any sense an "agrariancrisis". Rathbone argues interestingly that the expansion of large estates may have taken place without much expropriation of smallholders, but his view that casualties in war brought about a continual decline in the number of smallholders in the third and second centuries is based on an exaggeratedestimate of the military demands made on the Latin colony Cosa (pp. 18-9) and takes no account of the Roman census figures. In general on the peasantry in ancient Italy see J. M. Frayn, Subsistence Farming in Roman Italy (1979). The Supposed Roman Manpower Shortage of the Later Second Century B.C. 297 Rome in Southern Etruria have revealed a high density of rural settlement in both the Republican and early imperial periods with small farms forming a high proportion of all farm sites.45 A similar pattern of settlement has been detected by surveys of the territory of Cosa and of the Biferno valley on the borders of Samnium, although in parts of these territories larger 'villas' came to predominate.46 The uprooting of the Italian peasantry was a long drawn out process which continued long after the second century and was never completed. In the Gracchan period there will have been few, if any, districts of Italy from which subsistence farmers had altogether disappeared and many in which they continued to flourish. However, it would be excessively sceptical to doubt that by the time of the Gracchi a good many peasants had given up their holdings. Any drop in the numbers of smallholders which had taken place in the first three decades of the second century was probably more than compensated for by the large-scale colonial and viritane assignations made in that period, but thereafter there were few, if any, land assignations.47 Thus by the later second 4 The survey reports were published in PBSR 26 (1958), 63-134; 30 (1962), 116-207; 31 (1963), 100-158; 33 (1965), 70-112; 36 (1968). T. W. Potter, The Changing Landscape of South Etruria (1979) provides a synthesis of the results; he discusses rural settlement in the classical period at pp. 120-137. See also D. B. Nagle, Athenaeum 57 (1979), 411-441. The lack of a firm chronology for black glaze ware prevents the analysis of changes within the Republican period. Much of the black glaze from Southern Etruria is of types normally dated to the third century; Potter, op. cit. 95-6 explains this in terms of regional conservatism, but M. H. Crawford, Athenaeum 58 (1980), 497 thinks that it may be evidence of depopulation in the second and first centuries. For some other possible evidence for the decline of the peasantry near Rome see Crawford, JRS 71 (1981), 158. 46 The Wesleyan University survey of the ager Cosanus is reported by S. L. Dyson, Journal of Field Archaeology 5 (1978), 251-268, concluding that settlement patterns were very similar to those observed in South Etruria. A new survey is under way under the direction of Professor A. Carandini; his conclusions for the Valle d'Oro district are set out in A. Carandini and S. Settis, Schiavi e Padroni nell' Etruria Romana. La Villa di Settefinestre dallo Scavo alla Mostra (1979), 32-41 and pannelli 5-7. Carandini holds that by the first century smallholdings were completely ousted from the Valle d'Oro by rich villas, but for criticisms of this view see Rathbone, art. cit. n.44, 20-1. Some of the villas, like that excavated at Settefinestre, date to the first century, but others date from the mid-second century if not earlier (Rathbone, loc. cit.). The Biferno survey is reported by G. Barker, J. Lloyd and D. Webley, PBSR 33 (1978), 33-51. They show that smallholdings survived in the vicinity of Larino, but in the lower valley had been replaced by larger units by the later first century. For preliminary reports of surveys in the Albegna and Liri valleys, where only small farmsteads have been detected, see G. Barker and R. Hodges (ed.), Archaeology and Italian Society (B.A.R. Int. 102, 1981), 269-275 (S. L. Dyson) and 275-287 (E. M. Wightman). 47 G. Tibiletti, Athenaeum 28 (1950), 232-9 argued that assignations may have continued in the period 167-134 on much the same scale as before, but, although our sources are certainly defective, foundations on a significant scale would probably have left some record and it is hard to see why Ti. Gracchus sought to obtain land for allotment by such controversial means, if land assignations had been continuing steadily in the previous decades. Cf. Toynbee II. 208-210, 658-9; E. T. 298 J. W. RICH century there will have been a significantdrop in the number of peasant smallholders.How greatthe drop was we cannottell, althoughwe know that enough peasantshad been ousted to arousethe concernof contemporaries. It has been generallysupposedthat all those citizenpeasantswho lost land ceasedto rankas assidui.In my view, theremayhavebeenmanywho retained that status. As Brunthas argued,the valueof the propertyqualificationwas verylow.48 Our only figure for the price of land is Columella's 1,000 sestertiifor a iugerum.49 At that rate the figures for the property qualificationgiven by Polybius and Cicero would be equivalent to 1.6 or 0.6 iugera respectively.50 Now Columella'sfigure may well have been somewhatexaggeratedfor his own day,51and land valueshigherthen than in the second centuryB.C.; but there is no reasonto suppose that land values rose sharplyafter the second centuryB.C., and in any case not justlandbut almostall a man'spropertywas takeninto accountin fixinghis censusrating.Bruntis surelyrightto conclude that assiduiwere requiredto own no more than a house and gardenat the most. Manyformerpeasantproprietorsremainedin the country:it was thesemen Someof them may who constitutedthe bulk of Ti. Gracchus'ruralsupport.52 have been able to get the tenancy of a smallholding,but most must have subsistedin the main by casuallabour. Considerableuse was madeof hired free labour on slave-runestates.53But such work was largelyseasonal.Few families can have got all their living from this source alone. Most probably eked out their earningswith the produce of a small patch of land. Some of these plots will have been leased, but some may have been owned, either retainedfrom the formerpropertyor boughtwith money obtainedby the sale of the holding.A good manyformersmallholdersmayhaveowned a house,in Salmon, Roman Colonization under the Republic (l969), 112-5. Veil. I.15.3 assigns the colony of Auximum to 157, but for the possibility that the correct date may be 128 see Salmon, loc. cit. and Athenaeum 41 (1963), 5-13. 48 Brunt, JRS 52 (1962), 74; IM 405-6; SCRR 14. Cf. Schneider 13-16. The calculations of Shochat 20, 67-8 are based on the erroneous assumption that income was taken into account in the census assessment. 49 Colum. RR III.3.8. 50 Reckoned at the pre-tariffing rate of 10 asses to the denarius. On the figures for the property qualification see below Section V. 51 Cf. R. P. Duncan-Jones, The Economy of the Roman Empire (1974), 49-51. 52 Brunt, JRS 1962, 69-72; Astin, Scipio 345-6; Bernstein 86-97; Stockton 18-21. 53 Cato Agr. 1.3.; 4; 5.4; 144ff; Varro RR 1.17.2-3. See Brunt, JRS, 1962, 72; K. D. White, Roman Farming (1970), 347-350; P. D. A. Garnsey (ed.), Non-Slave Labourin the Greco-Roman World(1980), 41-3 (Garnsey), 65-72 (J. Skydsgaard); Rathbone, art. cit. n.44. Rathbone holds that except near Rome most of the seasonal agricultural labour was provided by peasant smallholders; in his view, few peasants had been expropriated and 'the villa systems and peasant smallholdings were complementary modes of agricultural production' (p. 15). The SupposedRomanManpowerShortageof the LaterSecondCenturyB.C. 299 some cases acquired by purchase, but perhaps more often the ancestral home. Peasants who lived in the towns and went out to their fields by day, as many may have done, could have lost their land without losing their homes.' Thus many of those dispossessed peasants who remained in the country may have owned enough property to satisfy the modest requirement for an assiduus. We cannot be sure that those who were no longer qualified by property invariably ceased to rank as assidui. Censors may have been reluctant to accept the claim of former assidui that their property had fallen below the minimum, especially if they were still resident in the country. Probably only a few of those who drifted to Rome had enough property to qualify as assidui, but it is possible that others retained that status along with their membership of the rural tribes.55 I do not doubt that a decline took place in the number of assidui as a result of the agrarian changes. However, our evidence about these changes does not help us to determine the extent of the decline: we do not know how many peasants lost their lands or how many of those who did ceased to rank as assidui. What we know about the agrarianchanges is compatible with the view that the number of assidui fell a long way below the level obtaining in the early second century, so that, although the demand for legionaries was now less great, the burden of the levy on the assidui was greater. It is, however, no less compatible with much less pessimistic estimates of the fortunes of the assidui. It is likely that there was some increase in the number of assiduiin the first part of the second century as the citizen body as a whole expanded, and it may be that, although the number later declined, it remained above the level of the early years of the century. Even if it dropped below that level, it may not have fallen so far that the reduced demands of the levy constituted a significantly greater burden on the remaining assidui. IV. Contemporary Anxieties about Manpower Those who hold that there came to be a serious shortage of assidui suppose that this gravely concerned many contemporaries and that C. Laelius and Ti. Gracchus hoped to relieve the shortage by agrarianlegislation. It was, on this For the view that many peasants lived in the towns see e. g. White, op. cit. 345; Brunt, IM 345. The first serious investigation of where peasants lived is by P. D. A. Garnsey, PCPS 205 (1979), 1-25, arguing that most peasants lived in the countryside. Garnsey may be right: he argues convincingly that the origins of the 'agro-towns' of present day southern Italy are mediaeval; the field surveys mentioned above provide evidence for dispersed settlement; and there are a number of colonies where it is clear that there was space for only a minority of the settlers within the walls. However, even Garnsey admits (pp. 10, 15) that some peasants whose holdings were near a town are likely to have lived there. 5' For such persons remaining in the rural tribes see H. M. Last, AJP58 (1937), 468-472; E. S. Staveley, Greek and Roman Voting and Elections (1972), 142, 201, 258 n. 374. 300 J. W. RICH view, one of the principal aims of Gracchus' law to restore the status of assidui to men who had been reduced to proletaii by granting them allotments. There are a number of difficulties in this view. It may be doubted whether men would have been eager to receive allotments which would make them liable to what had become, on this view, a crushing burden of conscription.56 We do not know to what extent, if at all, the law converted proletarii into assidui: it is not certain that the allotments counted towards a man's census rating,57and, if the argument advanced in the previous section is correct, many of the beneficiaries may never have ceased to rank as assidui. The most serious difficulty, to my mind, is that the hypothesis is in conflict with our evidence. Appian and Plutarch tell us a good deal about contemporary anxieties about manpower, but they say nothing of a decline in the number of assiduicaused by men ceasing to satisfy the property qualification. They speak rather of a decline in the numbers of the free population caused by a failure to rear children. In this section I examine this discrepancy. It is a notorious difficulty of Appian's account of the agrariancrisis that he sometimes implies that it was only Italian allies who lost their lands and aroused Gracchus' concern.58To that extent Appian was (for whatever reason) certainly mistaken: it is a matter of dispute whether the solicitude of Gracchus and his contemporaries extended to the allies as well as the citizens, but it cannot have been confined to the allies.59Apart from this problem, there are no serious inconsistencies in the accounts which Appian and Plutarch give of the manpower problem. The peasantry, they tell us, were impoverished by the loss of their lands, and so were unable to rear children. The resulting decline in the free population was all the more alarming because the number of slaves on the land was increasing; these were, Appian says, at best militarily useless and might actually be a threat. Gracchus undertook to set this state of affairs to 56 Cf. H. H. Scullard, JRS 54 (1964), 199, and J. E. A. Crake, Phoenix 20 (1964), 174, not adequately answered by Badian, ANRWI.1.718. 5' The allotments remained public land; they were made ager privatus by the agrarianlaw of 111, and line 8 of that law is taken by some writers to show that it was only then that they were counted for their holders' census assessment. For this view see P. Fraccaro, Opuscula 1I (1957) 87-8; F. C. Bourne, CW45 (1952) 181; Shochat 42, 87. Arguments against it are presented by Brunt, IM 78; Bernstein 128-131. '8 There can be no doubt that by 'ITakXL6TCELAppian was referring to Italian allies, not to citizen country-dwellers: cf. P. J. Cuff, Historia 16 (1967), 177-188. 59 Shochat 20-2, 77-85 attempts to save Appian's credit: Gracchus in his view planned to shift the burden of recruitment from the citizens to the allies, and so 'whereas in his speeches Tiberius suggested distributing land to Roman citizens in the name of justice, with the object of improving their economic situation, he justified the division of land to the allies by claiming that this would increase their share of military service" (p. 84). Both Shochat's interpretation of Appian and his account of Gracchus' motives seem to me quite untenable. We need not consider here the vexed question whether the allies were entitled to receive allotments under the law: for ingenious recent contributions to the debate see Bernstein 137-157 and J. S. Richardson, JRS 70 (1980), 1-11. The Supposed Roman Manpower Shortage of the Later Second Century B.C. 301 rights; according to Plutarch, Laelius had attempted to do the same earlier but had desisted when opposed by the powerful. Appian tells us that the purpose of Gracchus' measure was to procure eiicivb6Qac.Gracchus' plan on his view was evidently to restore prosperity to those who had been impoverished by giving them land and so to enable them to rear children once again.60 Some writers have claimed to discern references to the property qualification and the problems to which it is supposed to have given rise in Appian's words. These claims are in my view unconvincing. Brunt correctly points out that the words EvcavQLtaand ouavb?ctaused by Appian are ambiguous: they may refer to the quality as well as the quantity of manpower.61 It may well be that Appian meant to convey not only that there was a shortage of qualified men, but that those who were left were of poor physique. It does not seem to me that in the context in which Appian uses them these terms could reasonably be taken to refer to a shortage of citizens possessing the property qualification. A reference to the distinction between assidui and proletarii has been detected in the following passage from Appian's account of the speech which Gracchus made at the assembly at which the vote was to be taken on his law: EL .. .tXCLOV 8L-'TL. Tta xoLvC( XOLV 6tav4Etu1atL XaiL CiL yvrOLWTEQog &E'L XQaL xCiLToL; '0F,QovTo;g 6 3tOkiLTTg )(Q@OLWTEQO 6 0tQQliaTYL(T1atoXEC[LOU But the opening words of the next 6TriIooLoLLE?oVVOUcEQ0o O XOLVwVO6.62 sentence (ovUX 0g noXi 8E Trv OUYXQLOLV g 66botov tnitvEyxdv) suggest that the contrast in all these comparative phrases is the same, namely that between citizens and slaves.63 &nokE4ou recalls Appian's earlier insistence on the uselessness of slaves for military purposes.64Any citizen could be said to be a XOLVwVo5in the respublica. Both Appian and Plutarch in fact imply that those who had lost their land were still liable for military service.65 On the view for which I have argued in the previous section, there may have been many of whom this was true. 60 App. BC I.7.30; 9,35; 10.40; 11.43. 45-6, 27.122, 124; Plut. TG8.3. The decline of the free population: App. 7.30 Toi,; 6' 'ITctkL6TC(; UL,yftrl; xCti voCvb6ta xEaxeXLaVtTave TQuxo0vou; JtEVL TE xCdi o(poQ(aL; xai TLTEcL?;a, 9.35 rQlxxog... oERv0o? oViOoE tEQi TOV) 'ITC(XXOu yEvovU . .(.f1ELQOkEtVOV ... xCT'oXyoV EL &ITOQiaVxai 6kLyaCtV6QLCV, 10.40 oi 6' au 7ErVv1rET &-VTW16VQ0VT0 7I,itoiQoq?tiv, (t F-VZOQt0(Lg TIeQLqE(eoaL, ?5 7MViaV tUXa<T'V Plut. oi 7TEmvTE- ... xCai 63T' avm'g -yovCCLV, ovU 5UV&V6EVo tg lLEXoUV ITE7[Ci6oUv &VaTQOcPll GxrrE TCXii T1V 'IakCiav &tnaGav6IyaV6QiaV tXekEU&wV CidO&G1hL. Gracchus' aim of ectv8Qv'a: App. 11.43, 45. On Laelius' proposal see Astin, Scipio 307-310. 61 IM 77, 189; cf. Earl 39, L. Gallo ASNP 3.10 (1980), 1240-3, 1259. 62 App. 11.44. Badian, ANRW 1. 1.686 n. 50; Nicolet, Rome et la conquete du monde mediterraneen 264-27 avant J-C. 1. Les structures de l'Italie romaine (1977), 127; Stockton 32. 63 So Gabba ad loc. ("il confronto e per Iiberio fra il cittadino e lo schiavo"); Bernstein 146. 64 App. 7,29; 9.36. 65 App. 7.30 (cited above n. 60); Plut. Oc Cit.oL 7tEtvTljEg outTcZTaLQ eaTeiLCTL TE7QO&Vp.0vg If Gracchus really said that those who fought for Italy were homeless vagrants naXQELXOV ciCUToi3g. 302 J. W. RICH Suchotherevidenceas we havefor contemporaryanxietiesaboutmanpower also impliesthatit was aboutthe numbersof the freepopulationthatmenwere concerned. ScipioAemilianus,censorin 142/1, protestedagainstthe use of adoptionto MetellusMacedonicus,censor in 131/0, retainthe privilegesof parenthood.66 urgedthatall citizensshouldmarryliberorumcreandorumcausa.67Althoughit was, of course, not novel for the guardiansof Romanmoresto show concern for the propagationof the race,68these incidentsmay plausiblybe taken as evidencethat at the time men were concernedabout citizen numbers. Appian tells us that, when Scipio Aemilianus was appointed to the commandagainstNumantiain 134, he did not hold a levy, becauseRomewas fightingmany warsat the time and therewere plentyof menin Spain;instead, he took out a force of 4,000 volunteersraisedfromforeignkingsandcitiesand his Roman clients and friends.69Accordingto Plutarch,the senateprohibited the many men who wished to accompanyScipio from going on the grounds that Italy would be deserted (6g FiQ>oU tij 'ITaXLag;cooivrI5).7? If Plutarch'sstory is historical(which is by no meansbeyond doubt), we must supposethat the senaterefusedto allow Scipioto hold a levy.7"The numberof men involved may have been large: Scipio may have proposed that most, perhapsall, of the troops in NearerSpainshouldbe replaced,in whichcasethe number of Roman and allied troops to be levied might be of the order of 20,000 or more.72 Even so, the senate's decision was probably the product (Plut. 9.5), this must be either rhetorical exaggeration or another instance of the common complaint that soldiers returned home to find their farms expropriated, on which see Brunt, IM 642-3. 66 Gell. V. 19,15-16. See Astin, Scipio 322-4; T. P. Wiseman, JRS 59 (1969), 61. 67 Livy Per. LIX; Suet. Aug. 89; Gell. 1.6; cf. Lucilius 678-686M. The attempt of Shochat 50-2 to show that Metellus was moved simply by moral considerations and was not concerned for citizen numbers is unconvincing. On p. 52 he interprets Metellus' words (cited by Gellius) saluti perpetuae potius quam brevi voluptati consulendum est to mean that "the individual should act with moderation, taking thought for his long-term good." This overlooks the reason Metellus gives: quoniam ita natura tradidit, ut nec cum illis satis commode nec sine illis ullo modo vivipossit. 68 Cf. D. Daube, Proc. Class. Ass. 74 (1977), 10-25. 69 App. Iber. 84. 70 Plut. Mor. 201A = Apophth. Sc. Min. 15. 71 Plutarch's account is defective in at least one respect, namely the implication that Scipio was not allowed to take volunteers. There is a close parallel to this episode in the story that Scipio's grandfather, the first Africanus, was only permitted a volunteer force in 205 (Livy XXVIII.45.13-46.1 ; App. Lib. 7; Plut. Fab. 26.2; Zon.IX.11). That story may be a fiction based on the episode of 134 (so e. g. Brunt, IM 655-6). However, it may be historical (so e. g. Gelzer, Kleine Schniften(1964), 111.245-7); if so, it is possible that Scipio elected to take only volunteers in 134 in conscious emulation of his grandfather (for such emulation in his career see Astin, Scipio 20-2, 46). Scipio might well have preferred to avoid the odium of holding yet another levy for the hated Spanish wars (see below section VI). 72 The morale of the troops in the province was low (references in Astin, Scipio 136 n. 3). No reinforcements are recorded after those of 140 (App. Iber. 78), so that, on the principle that a man The SupposedRomanManpowerShortageof the LaterSecondCenturyB.C. 303 chiefly of political jealousy and the claim that Italy would be left deserted hardly more than a pretext.'3Such a pretextwould, however, probablynot have been used if it was not felt to have some plausibility.It has sometimes been adduced as evidence that declining numbers of assiduihad led to a shortageof recruits.74But I doubt whether it can mean that, althoughthere would still be plenty of free men, Italy would be empty of assidui.The point must ratherbe that Italy would be empty of free men.5 Grossly exaggerated though this was, the claim reflectedand exploitedcontemporaryfearsabout the numbersof the free population. The impoverishmentof a substantialproportionof the peasantrycould have led to a decline in the numbersof the free population.There may have been some reductionin the frequencyof marriage,and it is likely that manypeople Whethersucha declinedid in facttakeplacewe sought to limit theirfamilies.76 do not know. The census figuresgive us evidencefor adultmalecitizens,but pose grave problems of interpretation.77They show a gradual decline, sometimeschecked,from a peakof 337,022in 164/3 to a low point of 317,933 in 136/5. In 131/0 the numberregisteredrose slightly to 318,823. Only two more figures are recorded for the second century. They are suspiciously similar:394,736 in 125/4 and 394,336 in 115/4. The decreasebetween 164/3 and 136/5 may be the resulteitherof a declinein the citizenpopulationor of an increasein the numberof those who werefailingto register.If we accepteither of the figurespreservedfor 125/4 and 115/4 as authentic,the explanationof the earlierdecreasein termsof non-registrationis almostcertainlycorrect.Evenif both figures are corrupt and the true figures showed a continuingdecline, increasingnon-registrationmight still be at least part of the cause. If a decline did take place in the numbersof the free population,it cannot have been on the scale which the dramaticstatementsof our sourcessuggest. The lowest recordedcensusfigurefor this period,thatof 136/5, was only 6 per cent lower thanthat for 164/3 andwas higherthanany recordedbeforethat.It has been inferredfrom this that our sources must have misrepresentedthe was entitled to discharge after six years' continuous service (above n.1 1), most, if not all, may have served their term. 73 Cf. Astin, Scipio 184. According to Plutarch the senate begrudged Scipio money as well as men. The excuse for this must have been the poverty of the treasury; this too was an argument of doubtful validity (cf. Crawford, RRC635-6). One may compare Cicero's allegations against Piso: cum exhauriebas aerarium, cum orbabas Italiam iuventute (Pis. 57). 74 E. g. Astin, Scipio 171; Brunt, IM407 n.2; Badian, ANRW 1.1. 685 n.48. 75 This was certainly what later writers meant when they complained of solitudo Italiae: cf. Brunt, IM 128, 345ff. 76 On the relationship between poverty and family limitation in Roman Italy see Brunt, IM 76-7, 136-154; J.K. Evans, CQ 31 (1981), 435-8; W.V. Harris, CQ 32 (1982), 114-6. 77 Lucidly discussed by Brunt, IM77-83. Against the view that the figures represent assidui see above section 11. 304 J. W. RICH natureof contemporaryfearsaboutmanpower.The decline(realor supposed) in the nativestock may havebeenthe subjectof some concern,but it cannot,it is argued, have given rise to the acute anxiety attested by the sources or prompted Laelius and Ti. Gracchusto bring forwardcontroversialagrarian proposals;the conclusionis drawnthat for the chief causeof alarmwe must look elsewhere- to the assidui.78In my view this rejectionof our evidenceis unwarranted. The declinein the censusfiguresfrom 164/3 and in particularthe low figure recordedin the last censusbeforeGracchus'tribunate,thatof 136/5(a dropof 9,509 from the previouscensus),may well haveseemeddisquietingenoughto contemporaries,and they may have thought that the true picturewas worse still. Manumissionswere probably frequentin this period; men may have concludedthat the numberof ingenuiwas decliningfasterthanthe figuresfor all adult male citizens recordedby the censors.'9A decline in child-rearing could have only a delayed impact on figures for all adult males, and other figures which may have been available- for example,the total of iuniores may have shown a sharperdecline.Suchinformationas could be obtainedfor the Italianallies may have suggestedthat their numberswere decliningfaster thanthose of the citizens." Concernfor the nativestock may, in any case,not have startedfrom the census returns.These figuresmay have done no more than lend colour to alreadyexisting fears, based on subjectiveimpressions about the countryside,the complaintsof the dispossessedand the conviction that peasantswho had lost their lands could not but be reluctantto bear children. It would have been naturalfor those who felt such fears to give expressionto them in highly colouredlanguage:thus Gracchusin his speeches may have painted a picture of the depopulationof Italy not unlike that of Appianand Plutarch. The Romangoverningclasswas only too proneto exaggeratedfearsfor the state'ssecurity:thus the war againstCarthage,in which Gracchushad begun his militaryservice,had been undertakenin the groundlessbeliefthat, as long as Carthagestood on its existingsite, Rome could not be safe.81However, it 78 For this line of argument see especially Brunt, IM 75-7. 7 Scipio's famous outburst in 131 (references in Astin, Scipio 265-6) attests contemporary awareness of the increase in freedman numbers. An increase in non-registration, if it did take place, is less likely to have been apparent at the time. 80 Brunt, IM 84-90, 97-9, concludes from an analysis of the census figure for 70/69 that the rate of increase from 225 was much lower for those who had been allies than for those who had been citizens before the Social War. 81 Harris 163-254 demonstrates that fears for their own security played a smaller part in bringing about the Romans' wars of conquest than has often been supposed, but he overstates his case, notably on the origins of the Third Punic War (pp. 234-240). On that question Astin, Scipio 270-281 remains the best discussion (though he represents the Romans' fears as more reasonable than they were). The arguments of Cic. Leg. Agr. II.87-95 presumably did not seem as absurd to his audience as they do to us. The Supposed Roman Manpower Shortage of the Later Second Century B.C. 305 would perhapsnot have been so unreasonablefor men to fearthat, if nothing was done to check the declinein the nativestock andthe risein slavenumbers, Rome's security would be imperilled.No one knew how many slavesthere were in Italy. The SlaveWarin Sicily vividly illustratedthe danger.It was a slave revolton a scalewithout precedentin the Greco-Romanworld, andwas crushedwith considerabledifficulty; Sicily was declareda consularprovince each year from 134 to 132. Minoroutbreaksoccurredin Italy too. Men might well trembleat whatwould happenif revolton the Sicilianscalewereto spread to Italy.82 I conclude that we may accept the evidence of our sources. Men were gravelyconcernedabout the decline of the nativestock, citizen and perhaps also allied, and the increasein the numberof slaves.LaeliusandTi. Gracchus brought forward their agrarianproposals in the hope of checking these developments. Under Gracchus' law and probably also Laelius' proposal peasantswould be settledon landnow workedby slaves;slavenumberswould thus be reduced, and the settlers, protected under Gracchus'law from the encroachmentsof the rich by the stipulationthat their lots could not be alienated,would, it was hoped,prosperandonce againbe ableto rearchildren. It might still be maintainedthat, in additionto theirconcernfor the native stock, Ti. Gracchus,Laelius and their contemporarieswere also concerned about an artificialmanpowershortagebroughton by the maintenanceof the propertyqualification.But the sourcesprovideno supportfor sucha view, and those who hold it must accountfor their silence. V. The PropertyQualificationsof the CensusClasses It is now usually supposed that the property qualificationfor military servicewas reducedtwice, once in the HannibalicWar and once duringthe second century.These reductions,if they were made,must attesta shortageof qualified men.83 However, no ancient writer refers to a change in the qualification.The supposed reductionsare merely a modern hypothesis to account for some of the differentfigures for the qualificationgiven by our 82 & T Cf. Badian, ANR WI.1.684-5. The point was made by Gracchus himself: App. BCI.9.36 br' ( &O aTQQTtUO XCaiOtZOTE eg bvtJXEtEvag 6EoT6nO;g J (p, T6 EvayXog 4 6oiotxI tn1VVEyXV tV XLXECX(La &83EOTCV7tcfoS Vito f?EfQWtOVTWOV yEVO6LEVOV,bTJJAvOv X)&XELVwvnZo3 yEw?yiag, xaLt TOV tn' ai'VToi; 'PwRa(wv noiXegoovo1 6XL8ov oii& ItQaXuv,dXX' 'Eg TE XQ6VOl XCei T0oiat; XLV-6V(V noItxeag XO hNTQuanTa. 83 Shochat 49-50 holds that the qualification was reduced at some time between 133 and 123, although there was no shortage of assidui, because men were reluctant to be recruited. But, if there was no shortage, it is hard to see why reducing the qualification should have been seen as a way to resolve the problem of reluctance to serve. There is no reason to suppose that the attitude to the levy of those made liable for it by this reduction would be significantly different from that of those already liable. 20 306 J. W. RICH sources.The sourcesalso reportdifferentfiguresfor the qualificationfor the first propertyclass; these too are commonlytaken to representthe ratingsin force at differentperiods. In this section I assess these explanationsfor the discrepanciesin our sources.84 The accounts which Livy and Dionysius give of the centuriatesystem ascribedto ServiusTullius agreein all except a few particulars.5Dionysius' sourcespresumablygavethe propertyratingsin asses,like Livy: Dionysiushas converted them into minae, making the equation of the denariuswith the drachmawhich was standardby his day and reckoningat the pre-retariffing rateof ten assesto the denarius.86 His figuresagreewith those givenby Livy for the first four propertyclasses,namely 100,000,75,000, 50,000and25,000 asses.Livy gives the ratingof the fifth class as 11,000asses,whereasDionysius gives it as 121/2minae,i. e. 12,500 asses. We have no other evidencefor the ratingsof the second, third and fourth classes. Polybius in his accountof the Romanarmytells us thatonly thosewho had propertyto the valueof 400 drachmaeor morewereliablefor militaryservice, and that only those ratedover 10,000drachmaewore a lonrca.87 Sincethe lonrca was restrictedto the first classin the Serviansystemas describedby Livy and Dionysius, the secondfigureis generallytakento be the firstclasscensus.It is usually supposed that Polybius was reckoningwith the Attic-Alexandrian drachmaand so equatedthe drachmawith the denarius,and that his figures relate to the period before the retariffingof the denarius.In that case, the figurescorrespondto 4,000 and 100,000assesrespectively,and Polybiusis in agreementwith Livy and Dionysius on the first class, althoughnot on the fifth88.It is a difficulty of this view that it requiresus to treat Polybius' apparentlypreciseequationat 11.15.6of 1/2 as with 1/4 obol as an approximation. Some writers, starting from this equation, suppose that Polybius' 8' Discussions of the problems of the property qualifications include: Mommsen, Str. III. 230, 237-8, 248-251; Kubler, RE III. 1521-3; G.W. Botsford, The Roman Assemblies (1909), 84-91; A. Rosenberg, Untersuchungen zur romischen Zenturienverfassung (1911), 33, 40-3; H. Mattingly, JRS 27 (1937), 99-107; Gabba, ES 3-30; F. W. Walbank, A Historical Commentary on Polybius I (1957), 176, 698, 706; Toynbee I. 512-5; Brunt, IM 402-5; R. Thomsen in Classica et Mediaevalia FranciscoBlatt Dedicata (1973), 194-208 (hereafter CMB) and Actes du 8eme congres internationalde numismatique, 1973 (1976), 377-381 (hereafter Actes); Crawford, RRC621-631; Nicolet in Devaluations 248-269 (with discussion at 270-2); Marchetti 154-6, 168-198, 262-272. 8 Livy 1.43; D.H. IV.16-18. 86 For the equation of the denarius with the drachma see F. Hultsch, RE V. 1628-9. Presumably Dionysius reckoned at ten asses to the denarius because he knew that this was the original ratio. Dionysius' use of this ratio does not constitute evidence that these ratings were in fact introduced when the as was on the sextantal standard, as is asserted by e. g. Gabba, ES 3 n.2; Thomsen, CMB 205-6, Actes 377, KST 151; Nicolet, Devaluations 244, 247. 87 VI.19.3, 23.15. 88 For this view see especially Walbank and Thomsen, cited above n.84. The Supposed Roman Manpower Shortage of the Later Second Century B.C. 307 drachmawas the equivalent of 12 or 121/2 asses,giving figures for the census classes of 4,800 or 5,000 and 120,000 or 125,000 asses respectively.89 But it seems unlikely that Polybius could have expected his readers to understand without any explicit indication that he was using a weight standard other than the Attic-Alexandrian.90 Yet another figure for the fifth class appears in Cicero, Gellius and Nonius, according to whom the proletarii were those whose property was worth no more than 1,500 asses.91Cicero's statement occurs in a passage of the De Republica which purports to describe the centuriate system instituted by Servius Tullius, but, whereas Livy and Dionysius assign 80 centuries to the first class, he gives it only 70; in this respect his account reflects the centuriate organization as it was in 129 (the dramatic date of the dialogue) and in his own day.92Gellius goes on to distinguish the capite censifrom the proletariias those whose property was worth no more than 375 asses. It is commonly thought that a mistake has somehow arisen here, and that the capite censi and proletarni were in fact identical.93 A variant tradition on the first class census is preserved in three sources. According to Pliny, it was 120,000 asses in Servius Tullius' day. Gellius tells us that the rating was 125,000 asses and that those who possessed it were called classici and the rest infra classem. Festus says that the term infra classem denoted those who had less than 120,000 asses.94 As is generally agreed, the 89 Mattingly, art.cit. n.84, held that Polybius, using the Aeginetan standard, reckoned the drachma as equivalent to 11/4 denarii. Marchetti 168-198 and in Devaluations 196-200, 211-6 holds that Polybius, using the Rhodian standard, reckoned the drachma at 3/4 denarius and calculated at the post-retariffing rate of 16 asses to the denarius. 90 Yet another solution is proposed by A. Giovannini, Mus. Helv. 35 (1978), 258-263, who supposes that Polybius made the following equations: I drachma = I denarius; I obol = 2 (retariffed) asses; 8 obols = 1 denarius. But Polybius cannot have reckoned (or expected his readers to reckon) with a drachma of other than six obols, as Giovannini's theory requires him to do (notably at VI.39.12). 91 Cic. Rep. II.40; Gell. XVI. 10.10; Nonius 228 L. 92 In the "reformed" centuriate organization there were two centuries in the first class for each of the 35 tribes. This arrangement must be subsequent to the creation of the last two tribes in 241; it is implied by various passages in Livy's account of the Second Punic War (XXIV.7.12; XXVI.22.2-5; XXVII.6.3), but some writers reject this evidence and date the "reform" to 179 on the basis of Livy XL.51.9 (notably Nicolet, RHDFE4 39(1961), 341-358). For bibliography on the problems of the "reform" see R. Develin, Athenaeum 56 (1978), 346-377 (whose solution does not convince me). 93 See above n.13. 9' Pliny NH XXXIII.43 maximus census CXX assium fuit illo rege, et ideo haec prima classis (there is no warrant for deriving this statement from Timaeus, contra Crawford, Devaluations 271-2). Festus IOOL 'infra classem' significantur, qui minore summa quam centum et viginti milium aeris censi sunt. Gell. VI. 13 'classic' dicebantur non omnes, qui in quinque classibus erant, sedprimae tantum classis homines, qui centum et viginti quinque milia aeris ampliusve censi erant. 'infra classem' autem appellabantur secundae classis ceterarumque omnium classium, qui minore 20- 308 J. W. RICH discrepancybetween Gellius' figure and that of Pliny and Festus is almost certainlythe productof a scribalslip. Probablyit is Gelliuswhose figureis in error. In the samepassageGelliusgoes on to statethatthe termsclassicusandinfra classemwere used in Cato's speech in favour of the lex Voconia.This law, passed in 169, preventedmen owning property above a certainvalue from institutingwomen as their heirs. We may reasonablyinferfrom Cato as cited by Gellius that the limit fixed by the law was identicalwith the first class census.95Gaius'figurefor the Voconianlimit is 100,000asses,the sameas that given by Livy, Dionysius and Polybius for the first class census.96However, Dio, reportingexemptionsto the Voconianlaw grantedby Augustusin A.D. 9, statesthe limit as 25,000 drachmae,whichmustmean100,000sestertii, anda figureof 100,000sestertiifor the limit is also impliedby Pseudo-Asconius.9' Thus we have to reckonwith three conflictingtraditionson the first class census and the limit underthe Voconianlaw: 100,000asses(Livy, Dionysius, Polybius, Gaius), 120,000 (or 125,000) asses (Pliny, Festus, Gellius), and 100,000 sestertii (Dio, Pseudo-Asconius). For the fifth class our sources preserve four figures: 1,500 asses (Cicero, Gellius, Nonius), 4,000 asses (Polybius), 11,000 asses (Livy) and 12,500 asses (Dionysius). We do not know when the Roman citizen body was first divided into property classes with assessments in asses. Probably this development took place after the regal period but well before the first Roman coinage, the as being merely a pound of bronze.98 As most writers have recognized, the first census ratings must have been much lower than the ratings attributed to Servius Tullius by Livy and Dionysius. The evidence of Polybius and Gaius shows that there was a period when the figure given by Livy and Dionysius for the first class was in force. It is usually supposed that the same is true for their figures for the intermediate classes and for one of their figures for the fifth class; on this view, whatever be thought of the rest of the account of the Servian system reproduced by Livy and Dionysius, the assessments were anachronisms introduced from a much later period. It has traditionally been thought that these ratings were introduced when the as was on the sextantal summa aens, quod supra dixi, censebantur. hoc eo strictim notavi, quoniam in M. Catonis oratione, qua Voconiam legem suasit, quaeri solet, quid sit 'classicus',quid 'infra classem'. On the original significance of the distinction between classici and infra classem see most recently Richard 360ff and Thomsen, KST 167ff. 95 Contra Nicolet, Devaluations 262 n.25. Gellius' words do not, however, warrant the conclusion that he found the figure he gives for the census rating in Cato: cf. Thomsen, CMB 207, Actes 378-9. ' Gaius II.274. 97 Dio LVI.10.2 (omitting the word drachmae in accordance with his usual practice); Ps.-Asc. 247 St. 98 Cf. G. Pieri, op cit. n.31, 48-50; Crawford, art. cit n.l.; Richard 386ff. The Supposed Roman Manpower Shortage of the Later Second Century B.C. 309 standard.9Of the two figures for the fifth class, most writersaccept Livy's 11,000 asses as historical and dismiss Dionysius' 12,500 asses as a fiction, obtainedby halvingthe figure for the fourth class.'1' It is commonlysupposedthat the first classcensuswas at some point raised from 100,000to 120,000(or 125,000)asses;Crawfordhas recentlyarguedthat 120,000asseswas the earlierrating.The view thatthe figureof 100,000sestertii representsyet anotherchange in the census goes back to Boeckh and is still current;other explanationshave also been canvassed.101 Older writerscommonlyheld that the qualificationfor militaryservicewas at some point reduced from I 1,000 to 4,000 asses, but attempted to account for the figure of 1,500 assesby denying that the property qualificationwas the Solutions same for the fifth classand for liabilityfor the levy and tributum.'02 along these lines are in clearconflictwith the sourcesand must be rejected.'03 In an influentialarticle,first publishedin 1949, Gabbaarguedthat the fifth class ratingwas reducedtwice, with the resultthat the legions were gradually "proletarianized."Almost all subsequentwritershave acceptedthis view.'04 Gabba ascribedthe first reduction,from 11,000 to the Polybianfigure of 4,000 assesto the HannibalicWar,and this remainsthe dominantview. Gabba argued for the date 214; this has also been advocatedby Brunt, but for 9 This view is based partly on an invalid inference from Dionysius' use of the ratio of ten asses to the denarius (see above n. 86), partly on the (in my judgment correct) belief that the ratings would be implausibly high if expressed in libral asses. '00 Some writers (e. g. Nicolet) leave open the possibility that it is Livy's figures rather than Dionysius' which should be rejected. Toynbee I.512-4 regarded both as historical; in his view the qualification was reduced from 12,500 to 11,000 asses in the later fifth century. Nicolet supposes that the common source from which Livy's and Dionysius' accounts of the Servian system derive was a work of the early second century (see especially art. cit. n. 92; Quad. Accad. Naz. Linc. 221 (1976), 111-148); Gabba thinks that it may belong to the Sullan period (Athenaeum 39 (1961), 98-121; Quad Accad. Naz. Linc. 1976, 149-150). Thomsen, KST 153-6, holds that the common source may be Piso and that he derived all of his account, except for the attribution of the arrangements to Servius Tullius and the information about military equipment, from a document in the censorial archives which set out the arrangements in force in the period between the introduction of the sextantal as (212/11) and the reform of the centuriate organization, which he dates (following Nicolet) to 179 (cf. above n.92). 0O1For the view that the census was twice raised see A. Boeckh, Metrologische Untersuchungen ,uber Gewichte, Muinzfuisseund Masse des Altertums (1838), 427ff., and among recent writers Thomsen, CMB 206-8, Actes 3 78-81, Devaluations 2 70; Nicolet, art. cit. n. 84; Marchetti 262-272. For Crawford's views see RRC 621-5, 631; Devaluations 271-2. Some older writers (cited by Steinwenter, RE XII. 2419-20) attempted to explain the discrepancy on the limit fixed by the lex Voconia in terms of the supposed equation between the libral as and the sestertius; against this equation see Crawford, RRC 622. 102 E. g. Mommsen, Botsford, Rosenberg, as cited in n.84. 103 The sources give no hint of any such distinctions, and Gell. XVI.10.1 0-l1 is explicit that the proletarn, who owned not more than 1,500 asses, were not normally liable for military service. See Fraccaro, Opuscula 11 (1957), 152 n.1 1; Gabba, ES 6-8. 104 Scepticism has been expressed by Lazenby, loc. ct. n.3. 310 J.W.RIcH differentreasons.'05Some writersdate this reductionin the earlierpartof the second century.'6 The terminusante quemfor the secondreduction,from4,000to 1,500asses, is taken to be 129, the dramatic date of the De Republica. Gabba held that it was introduced in 129 or shortly before, since he took the increase in the census figures between 131/0 and 125/4 to reflect the reduction in the qualification; this argument is invalid if, as I have argued above, the census figures represent not assidui but all adult male citizens.'07 It is generally supposed that the reduction must have taken place after Polybius wrote Book VI, probably about 150.108 Brunt thinks that Polybius' figure may have been out of date at the time he wrote, one of a number of anachronisms which he suspects in Polybius' account of the Roman army; in his view, the reduction may have taken place as early as 171.1'9 Two recent advances in our understanding of the numismatic history of the Roman Republic have important implications for these questions: the establishment of a firm chronology for the devaluation of the bronze coinage, and Crawford's demonstration that the sestertiusreplaced the as as the unit of reckoning about 141. It is now clear that the earlier part of the Second Punic War saw a progressive and drastic devaluation of the bronze coinage. Crawford's account is as follows." 0 At the outbreak of the war the as was still being coined on the 'libral' standard, in fact of ten ounces."' The standard was reduced to six ounces (semilibral) in 217, but continued to fall in the following years. Stability was simultaneously restored to both the silver and the bronze coinages in 211 105 Gabba's arguments for this date (ES 12-13) lack cogency. The shortage of rowers which made it necessary to call on slaves in 214 (Livy XXIV.11. 7-9) could have been the result of factors other than the enrolling in the legions of men who formerly served in the fleet. There is a good deal of evidence for light-armed troops in the legions before 211 (see Walbank on Plb. VI.21.7); Livy's claim (XXVI. 4.4-10) that velites were introduced in 211 may be based merely on a story (of doubtful value) that in skirmishes at Capua cavalrymen carried velites into battle. On Brunt's argument see above section II. 106 Nicolet, Devaluations 257, 259 (either during the Second Punic War or in 179); Marchetti 154-6, 262-272 (suggested date 169). See also Crawford, RR 101. 107 See above section II. Gabba's dating for the reduction is accepted by those more recent writers who take the census figures to represent assidui, cited above n.31. 108 Cf. C.O. Brink and F.W. Walbank, CQ 4 (1954), 97-122; C. Nicolet in Polybe (Fondation Hardt, Entretiens XX, 1974), 209-265. '09 IM 604, 625. Rejected by Crawford, RRC 625; Harris 49. For their view that the reduction took place between ca. 150 and ca. 141 see below. 110 RRC28-35, 43, 595-6, 615-6, 625-8. Crawford marginally revises the account offered by Thomsen, Early Roman Coinage (1957-1961). For Thomsen's present views (in general agreement with Crawford) see Devaluations 9-22. " This weight standard had been in force since the fourth substantive bronze issue (RRC no. 21, dated by Crawford to 269-266). The Supposed Roman Manpower Shortage of the Later Second Century B.C. 311 or possibly 212: the denarius was introduced (tariffed at ten asses) and the as was issued on a weight standard of two ounces (sextantal). It seems unlikely that this account will be significantly modified by further research."2 Thus by 211 the purchasing power of the as had fallen to about one-fifth of what it had been at the start of the war.'13 Crawford and Nicolet have recently argued that the qualifications for the classes remained unchanged in these years in spite of the progressive devaluation of the as; Nicolet holds that it was the "Servian" ratings of Livy and Dionysius which were in force."4 This is difficult to accept. Ratings ranging from 11,000 (or 12,500) to 100,000 "libral" asses are much too high to be plausible; a reduction in the real value of the census rates to one-fifth of the earlier level would have had profound social and political consequences which would surely have been unacceptable to the ruling oligarchy."5 We can certainly exclude the possibility that the fifth class census stood at the "Servian" figure (11,000 or 12,500 asses) before the Second Punic War and was reduced to the Polybian figure of 4,000 asses during the war: this would have meant a reduction in real value to about seven per cent of the pre-war level! It seems much more likely that higher ratings were introduced to take account of the devaluation. Possibly there were two upward revisions of the ratings, for the censuses of 214 and 209 respectively. The rating of the fifth class (and perhaps others too) may have been fixed rather below the equivalent in real terms of the pre-war figure in view of the losses in the war and the continuing high demand for legionary manpower. Those who adhere to the traditional view that the "Servian" ratings were introduced after the as had become sextantal must now accept 211 as the earliest possible date for their introduction. On this view too it seems most 112 Marchetti believes that the denarius and the sextantal as were introduced in 214 and that the as was reduced to the uncial standard and retariffed at sixteen to the denarius in 211: RBN 117 (1971), 81-114; Numismatique antique. Problemes et methodes (1975) 75ff.; Devaluations 23-29, 195-209; Hist. icon. et monet. 174-197, 281-352. I am not competent to assess his arguments; numismatists seem unconvinced (cf. Thomsen, cited n.110). His attempt to reconcile his date for the uncial standard with his view that the "Servian" ratings were sextantal and remained in force until 169 is very strained, and his claim that the denarii with the symbol XVI need not be connected with the retariffing is implausible. 113 Nicolet, Devaluations 258 n.16, 271, is surely wrong to suggest that the devaluation may not have been fully reflected in prices. 114 Crawford, RRC 627; Nicolet, Devaluations 257-9. 115So Thomsen, Devaluations 270; Nicolet himself expresses misgivings (ib. 271). Both Crawford and Nicolet seek support in the consular edict of 214 fixing contributions of slaves according to property (Livy XXIV.11.7-9). But there is no reason to think that the scale prescribed bore any relation to the ratings of the property classes. The property assessments made at the census of 220 were used because none had been made since. The phrase aut cuipostea tanta res esset facta may mean that account was taken of the subsequent devaluation (cf. Marchetti 299-301). 312 J. W. RICH unlikely that the fifth class census was reducedto the Polybianfigureduring the Second Punic War. The reductionwould come when the worst of the manpowerdifficultieswereover, althougha suitablecontextmightbe foundin 207.116 Much more serious, the life of the "Servian"figurefor the first class census would have been very short. Why should any ancient author have attributedto ServiusTulliusa rating which was in fact in force for so brief a period ?117 Thus in the light of what we now know aboutthe numismatichistoryof the HannibalicWar,it can, in my view, no longerbe maintainedthatthe fifthclass censuswas reducedfrom the "Servian"to the Polybianfigureduringthe war. It remainspossibleto supposethatthe "Servian"ratingswereintroducedafter the as becamesextantaland the fifth class ratingwas reducedto the Polybian figure at some time in the first half of the secondcentury.However,one may wonder whether Polybius would in that case have recordedthe new figure without mentioningthe change.He claimedto be describingthe Romanarmy as it was in the HannibalicWar,and in some respectshis accountmay reflect the practiceof even earlierperiods.'18 Crawfordhas shown conclusivelythat the sestertiuswas adoptedby the Roman state as the official unit of reckoningin place of the as in or shortly before 141. His suggestion that the retariffingof the denariusat 16 asses occurredat the same time and as part of the same reformis attractive.119 What happenedto state assessments,which had been previouslyexpressed in asses,when the sestertiusbecamethe official unit of reckoning?Crawford holds that for administrativeand politicalconveniencethese assessmentswere converted into the same number of sestertiiso that their real value was quadrupled.He also thinksthatto createmoreassiduithe fifth classcensuswas reduced from 4,000 to 1,500 asses between about 150 and 141. These two propositions are surely incompatible.If, as Crawfordsupposes, the rating became 1,500 sestertiiin 141, it was thereby raised substantiallyabove the Polybian figure. If the Roman governmentwas preparedto quadruplethe value of the qualification for military service in 141, no concern can have been felt in that period about the numberof assidui. 116 For the emergency measures relating to manpower taken then in response to Hasdrubal's approach see above nn. 18-19. 117 Gabba's original account was vulnerable to this objection: on his view the denarius system was introduced ca. 218, the "Servian" ratings were introduced after that and the fifth class census was reduced to the Polybian figure in 214. 118 Cf. Brunt, IM 627. Polybius remarks on a change of practice at VI.20.9. 119 RRC 621-5. Cf. Devaluations 21-2n.63 (Thomsen), 149 n.Il (Crawford). H. Zehnacker in Les devaluations a Rome: epoque romaine et imperiale 2 (Gdansk, 19-21 octobre 1978) (1980), 31-49 argues convincingly from references in Cato Agr. (dismissed as interpolations by Crawford, RRC 621 n.2) that private use of the sestertius as a unit of reckoning was established before its adoption by the state. The SupposedRomanManpowerShortageof the LaterSecondCenturyB.C. 313 Crawford'stheory about stateassessmentsshouldin my view be rejected.120 So drastican increasein the census valuesseems in principleunlikely.It may be that the first class census and the limit underthe lex Voconialaterstood at 100,000sesterti4but this could havebeen broughtabouttot by conversionin 141 but by a laterincrease.In any case,this is not the only possibleexplanation of the evidence.As Nicolet has pointed out, the passageof Pseudo-Asconius which mentions the figure of 100,000 sestertii is confused, and the writer may well have converted100,000assesinto sestertiiin error."2'It is not impossible that a similarmistakelies behind Dio's statementon the Voconianlimit. The qualificationfor locupletioresliberti laid down in the lex Papia need not be identical with the first class census.'22The other evidence assembled by Crawfordfor assessmentsbeing expressedin the same numberof assesand sestertiican carry no weight.'23It seems to me most likely that, when the sestertiuswas adoptedas the unit of reckoning,assessmentsformerlyexpressed in asses were converted into a quarter that number of sestertii so that there should be no changein their realvalue.'24 Although Crawford'stheory about state assessmentsis unacceptable,his demonstrationthat the sestertiuswas adoptedas the unit of reckoningabout 141 holds good and must be takeninto accountby those who wish to explain the traditionsof a fifth class census of 1,500 assesand a first class census of 120,000assesby supposingthat the ratingswere changedat some point in the second century. Those who hold that the fifth classwas reducedfromthe Polybianfigureto 1,500 asses must suppose that it stayed at that ratingfor only a short period: unlessPolybius'figurewas obsoletewhen he wrote, the reductioncannothave takenplace beforeabout 150; about141,the ratingwould havebeenconverted into perhaps 375 sestertii. This hypothesis seems to me unattractive: it is hard to see why a figure which was in force for perhapsless than a decadein the 120 So Nicolet, Devaluations260-4 and Thomsen, ib. 270; not in my view effectively rebutted by Crawford, ib. 271-2, 121 Nicolet, Devaluations 263, showing that the writer has misunderstood Cicero's neque census erat. The source from which he took his information on the lex Voconiamay have stated the limit simply as centum milia without specifying the unit of reckoning (as Pseudo-Asconius himself did the second time he mentioned the figure). 122 Gaius III.42; Just. Inst. 111.7.2. 123 There was no need in 141 to convert the penalty of 25 asses for iniuria specified in XII Tab. 8.4, since it was no longer effective law; 25 sestertii in Paulus, Coll. Mos. Rom. II.5.5 can only be a jurist's error. I do not understand the relevance of Crawford's speculations on the "basic daily wage" since on his view it was only state assessments, not state payments, which were converted to the same number of sestertii. It is natural that nominal assessments should be fixed at I x the unit of currency. 124 Zehnacker, art. cit. n.1 19, holds that for reasons of conservatism the census ratings continued to be expressed in asses. This seems unlikely in view of the use of the sestertius for the equestrian census, on which see the remarks of Nicolet, Devaluations 264-7. 314 J. W. RICH middle of the second centuryand was alreadyobsolete by 129, the dramatic date of the De Republica,should have been ascribedby Cicero to Servius Tulliusor should have been the ratingcited by Gellius and Nonius. Gaius' statementon the limit fixed by the lex Voconia,takenwith Gellius' citationof Cato'sspeechon the law, permitsus, in my view, to concludewith reasonableconfidencethat, when the law was passed in 169, the first class census was 100,000asses.It seems to me perverseto maintainthat, in spite of the Gelliuspassage,the Voconianlimit may not havebeenthe sameas the first class census,125or that Gaius' figure may not be the limit specified in the originallaw.'26Those who supposethat the firstclasscensuswas 120,000asses at some time in the second century must, I think, choose between two hypotheses. They may suppose that the census was raisedfrom 100,000to 120,000assesat some point afterthe passingof the lex Voconiain 169,127and convertedinto perhaps30,000 sestertiiabout 141. Holders of this view who cannotacceptthat Polybius'figurewas no longerin forcewhen he wrote must date the introduction of the 120,000 asses rating after about 150; this hypothesisis vulnerableto the objectionthat the life of the ratingwould have been very short.The alternativeis to supposethatthe censuswas reducedfrom 120,000to 100,000assesbefore 169 and convertedinto perhaps25,000sestertii about 141. On this view, Polybius failed to remarkthat the ratehad recently been changed. The main way in which recent writershave attemptedto accountfor the variousfiguresgiven by our sourcesfor the census ratesof the first andfifth classes is to suppose that the figures representrates which were in force at differentperiods. I have arguedthat the fifth class census cannot have been reducedfrom the figureof Livy or Dionysius to that of Polybius duringthe HannibalicWarand that some of the hypothesesaccordingto which the rates of these two classeswere changedduringthe secondcenturyareless attractive than has been supposed.It remainspossible that some of the discrepanciesin our sources may have arisen in this way. Another hypothesis of this type which might perhapsbe worth consideringis that the fifth class censuswas 1,500 asses at the time of the "libral"as and was increasedto the Polybian figureduringthe HannibalicWarto take accountof the devaluationof the as. There are other ways in which the discrepanciesmay have come about. Error,eitherby ancientauthorsor by scribes,may accountfor some of them. 125 See above n.95. So Marchetti 268 n.50 (absurd) and Crawford, RRC631. Crawford supposes that both the census and the Voconian limit were reduced from 120,000 to 100,000 asses between 169-8 and ca. 150, but it seems unlikely that Gaius would have cited a figure which was not that of the original law and was in force for so brief a period. The 30,000 sestertii of Livy XLV.15.2 need bear no relation to the first class census. 127 So Thomsen, cited n.101. 126 The SupposedRomanManpowerShortageof the LaterSecondCenturyB.C. 315 It has long been recognizedthat the disagreementbetweenthe 120,000assesof Festus and Pliny and the 125,000assesof Gellius is probablythe productof scribal error. As I have suggested above, misunderstandingby Dio and Pseudo-Asconius may have been responsible for their figure of 100,000 sestertii.It may be that the figure which the common source of Livy and Dionysius gave for the fifth classwas 12,500assesand that Livy's 11,000asses is a scribalerror.Misunderstandings or scribalerrorsmay lie behindsome of the other figures- possibly even Polybius' figurefor the fifth class. The antiquarianor annalistfrom whom the accountsof the Serviansystem in Livy and Dionysius deriveassignedto the first class a ratingwhich was in force in a later historicalperiod (which may or may not have been the time when he himselfwrote). We cannottake it for grantedthathe did the samefor his figuresfor the otherclasses:it is worth notingthat,if his figurefor the fifth classwas 12,500asses,all four figuresaresuspiciouslyneatfractionsof the first class rating. Our three sourceswhich speakof a first classcensusof 120,000or 125,000 assesare all concernedwith early Rome: Pliny says that this was the ratingin the time of ServiusTullius,while Gelliusand Festusmentionit in conncetion with the archaic distinction between classici and infra classem. They may well deriveultimatelyfrom a commonsource,perhapsan antiquarianwho gavean account of the Serviansystem significantlydifferentfrom that of Livy and Dionysius. We have no warrantfor assumingthat the figurefor the first class census given by this tradition was, like that of Livy and Dionysius, a retrojectionof a ratingactuallyin force in a later period. That is a possible explanation,but others are no less likely: the figure could have been the productof antiquarianspeculationor sheer invention. Cicero attributedthe figureof 1,500assesfor the fifth classcensusto Servius Tullius,while the other sourceswho mentionthis rating,Gelliusand Nonius, give no indicationof when it was in force. It may well be that this figuretoo derivesfrom an antiquarianwriterand was neveractuallyin force.'28 On the first class census it seems to me not unreasonableto hazard a hypothesis.The most economicalsolutionis to supposethatthe variantfigures are the productof erroror invention,and that the censuswas fixed at 100,000 assesin or soon after211, when the as becamesextantal,and remainedat that rate until 141, when it was convertedinto probably25,000 sestertii.On the remainingclasses my conclusion is negative.For the intermediateclasseswe haveonly the doubtfulevidenceof Livy andDionysius,while for the fifth class the discrepanciesof our evidenceseem to me intractable.In my view, the only 128 It is true that the figure given for the number of centuries in the first class in Cicero's account is one which was in force in later times, but it does not follow that this was the case also for his figure for the fifth class census. 316 J. W. RICH prudentcourse is to acceptthat speculationabout the historyof these census ratingsis fruitlessand to admitour ignorance. The view that the propertyqualificationfor militaryservicewas progressively reducedderivesmuchof its plausibilityfromthe factthatit fits well with receiveddoctrineon Romanmanpower:manpowerresourceswere evidently understrainin the HannibalicWarandon the usualview a shortageof assidui developedduringthe secondcentury.It would thus smackof circularityto use the supposed second century reduction in the property qualificationas evidencefor the shortageof assidui.I have arguedin this sectionthatthereare no good groundsfor advancingeitherthis or any other hypothesisaboutthe level of the property qualification.If that is correct, the evidence of the propertyqualificationcan provideno supportfor the doctrinethattherewas a shortageof assidulin the later second century. VI. The Levy beforeMarius If the Romangovernmentwas facedby a seriousshortageof recruitsin the later second century, one might expect that it would have avoided all unnecessarymilitarycommitments.But the Romansundertooka numberof wars in this period in areaswhere armieswere not normallystationed,and some at least of these wars could readilyhave been avoided.129 A varietyof evidencerelatingto the levy has been taken as supportingthe view that a shortageof assiduidevelopedin this period.In this sectionI seekto show that these claimsare without substance. (i) Popularattitudesto the levy. It is often saidthat men becamelesswilling to serve in the legions in the courseof the secondcenturyand thatone of the reasons for this was that, as the numberof assiduideclined,the burdenon those who remainedbecameheavier.In fact, it is by no meansclearthat men becamegenerallyless willing to serve.130 Livy's silence shows that there was no seriousmanifestationof unwillingness to be conscriptedfrom men normallyliable to serve in the first thirty years of the century.Disputes over the levy took placein 193 and 191, but in both cases they concernedparticulargroupswith a claimfor exemption."3' 129 130 Cf. Harris 245-9. On popular attitudes to the levy in the second century see especially Smith 5-8; L. R. Taylor, JRS 52 (1962), 20-2, 26; Toynbee II.92-100; Astin, Scpio 167-172; Brunt, IM 396-402, 404-5; Harris 46-50; Shochat 55-60. Harris effectively questions the assumption of earlier writers that throughout the century most conscripts served reluctantly; he supposes that in the earlierpart of the century citizens were generally willing to serve, but that by the middle of the century their willingness was becoming much more selective. Shochat supposes that citizens did become generally unwilling to serve, although there was no shortage of qualified men; he exaggerates the weakness of the senate's response. I discuss the special problem of the recall of veterans below (sub-section (iii)). 131 Livy XXXIV. 56-9, XXXVI. 3.4-6; see Harris 48 n.2. The Supposed Roman Manpower Shortage of the Later Second Century B.C. 317 Service in the war against Perseus seemed attractive when it began in 171: men remembered the gains that had been made by those who fought against Philip and Antiochus.'32 By 169 no progress had been made in the war. The consuls reported that they were having difficulty in completing the levy because men were not answering to their names, but two praetors claimed that the consuls were seeking to avoid unpopularity by not enlisting those who were reluctant to serve; the levy was transferred to the praetors who completed it successfully after the promulgation of a censorial edict.'33 In 149 there was general eagerness to enlist for the war against Carthage in the expectation of easy victory and rich booty.'34 By contrast, the protracted, hard and unrewarding wars fought in Spain from 154 to 133 evoked much discontent.'35 The first instance occurred in 151. Polybius tells us that, because of unattractive reports of the recent fighting, "a sort of extraordinary panic overtook the young men, such as the older men said had never happened before," and they sought to evade service both as officers and in the ranks. The tribunes disputed with the consuls over exemptions and took the unprecedented step of throwing the consuls into prison. At least one concession was made: the ballot was used for the first time in the levy, evidently to decide which of the recruits should serve in the Spanish legions and which should enjoy brief and safe service in Italy.'36Two obscure episodes in 140 may reflect resentment at the Spanish levy: Ap. Claudius Pulcher obtained a decree of the senate that there should not be more than one levy in the year and a tribune sought unsuccessfully to impede the departure of the consul Q. Servilius Caepio for Further Spain.'37 Further incidents occurred at the levy of 138, when troops were presumably being raised for the consul D. lunius Brutus to take to Further Spain. The new recruits were taught an exemplary lesson: one or more deserters from the Spanish army were flogged and sold into slavery before their eyes.'38 Once again the tribunes pressed claims for exemption and threw the consuls into prison.139It is true that some Romans went to Spain 132 133 134 Livy XLII. 32.6. Livy XLIII. 14.2-6,15.1. App. Lib. 75. 135 On the character of the fighting see especially Plb. XXXV.1, 4.2. Lack of booty: App. Iber. 54; Pliny NH. XXXIII. 141; but see Brunt, IM 396 n.2. 136 Plb. XXXV.4; App. Iber. 49; Livy Per. XLVIII; Oros. IV.21.1; Val. Max. 111.2.6.Polybius' claim that it was Scipio Aemilianus' offer to serve which brought men to their senses evidently exaggerates the effect of his gesture. Livy seems to have combined Polybius' account with that of an annalist. The episode is well discussed by Astin, Sczpio42-6. 137 Livy Oxy. Per. LIV. This account of Claudius' action depends on the restoration of the word delectus; its significance is discussed by Astin, Scipio 144. The suggestion of Harris 49 n.5 that the levy Claudius prevented may have been for Sicily, not Spain, requires too early a date for the outbreak of the Sicilian slave revolt. 138 Livy Per. LV; Ps.-Frontin. Strat. IV.1.20. 139 Cic. Leg. III.20; Livy Per. and Oxy. Per. LV. 318 J. W. RICH with Scipio as volunteersin 134, but these were all clients or friendsof the commander.140 We hear of no resistanceto the levy in the remainingyears of the second centuryafterthe wars in Spainended. It is clearthat in the secondcenturythe attitudeof the assiduitowardsa levy varied considerablyaccordingto the characterof the war for which it was being held. It may be that the mid-century wars in Spain provoked unprecedentedresistanceto the levy simply becausenone of Rome's earlier wars overseashad seemedso unattractive. If it were true that the assiduibecamegenerallyless willing to serve in the legions in the course of the century,it would not necessarilyfollow that the reason was a decline in their number.Many differentfactorscould help to bring about such a change. Plutarchis the only ancientauthorto assertthattherewas a declinein men's willingness to serve. He links the developmentto the dispossessionof the peasantry,but says nothing of a drop in the numberof assidui;accordingto him it was the dispossessedpeasantrywho no longerreadilycameforwardfor service."4'If the argumentadvancedin section III aboveis correct,Plutarch's explanationis not to be dismissedout of hand:many of the dispossessedmay Still have ranked as assidui.It would not be surprisingif those whom conscriptionhad helped to ruin lost their relishfor fightingRome'swars. It could also be that familieswho were mainly dependenton wage labourfor their subsistencecould less readilysparea manfor prolongedmilitaryservice than those who lived off the produceof their own smallholding.'42 (ii) C. Gracchus'lexmilitaris.One of C. Gracchus'lawswas concernedwith Plutarchtellsus of two of amelioratingthe conditionsof servicein the army.143 its provisions:clothingwas to be suppliedfreeof charge,andno one was to be conscriptedbelow the age of seventeen.Others may be conjectured:some other deductions from pay may have been stopped, and the number of stipendiamen were liable to servewas probablyreduced."' 140 Appian, cited above n.69. According to Plutarch, many men wished to go, but he is wrong to say that the senate prohibited men from going as volunteers (above, nn. 70-1). 141 TG 8.3, cited above n.65. 142 On the underutilization of manpower on peasant holdings see Hopkins 24-5; Crawford, RR 102-3; Rathbone, JRS 1981, 15, 19, 22. On the comparative attractions of work as a mercennarius and military service see the judicious remarks of Brunt, IM 411-412. 141 Plut. CG 5.1; Diod. XXXIV/XXXV. 25.1. Plutarch's clear implication that Gracchus passed just one law on military matters is to be preferred to Diodorus' vague reference to v6LO L (contra G. Bloch and J. Carcopino, Histoire romaine II (1935), 245; Gabba, ES 15-16). 144 Gracchus may have retained the charge for replacement weapons and armour (for which see above n.1) to ensure proper care of arms. If he did remove it, it was later restored, as were deductions for clothing: Tac. Ann. I.17.6; G. R. Watson, The RornanSoldier(1969), 103-4. On the legitima stipendia see below sub-section (iii). The Supposed Roman Manpower Shortage of the Later Second Century B.C. 319 The prohibition on the conscription of those under seventeen merely reassertedthe existing state of the law.145It is commonly said that Gracchus must have included this provision because there had been recent cases of personsbelow the minimumage being conscripted,andthatthis is evidenceof a shortageof qualifiedrecruits,since, as Astin puts it, "armiesdo not enlist This is not plausible.Personsnot normally boys when they can find men".146 liable to serve could be conscriptedby a decree of the senate, but it seems unlikelythat the senatewould have sought to alleviatea shortageof assiduiby authorizingthe conscriptionof assiduiunderthe age of seventeen.Sucha step would simply have made the plight of the unfortunateassidui yet more wretched;the naturalway to achieve a marginalalleviationof a shortageof assiduiwould have been to reduce the propertyqualification.Commanders who conscriptedmen not liableto servewithout the senate'sauthorityranthe risk of an appealto the tribunes;147 they would in any case be more likely to choose adultproletarii,who on this hypothesiswere numerous,than assidui underseventeen. It would be in keepingwith Romanlegislativepracticeif Gracchus'military law did not merely prescribechanges from existing procedureand reassert rulesby which Gracchusset specialstore, but gavea comprehensivestatement of whatwas henceforthto be the law on the conditionsof militaryservice.148In that case many of its provisionswould simply be tralaticianreenactmentsof the existing state of the law (which had presumablyrested up till now on custom rather than statute). I suggest that, while the other known or conjecturedprovisions of Gracchus'law were changes, the prohibitionon conscriptionunderseventeenwas tralatician.'49 (iii) The laws reducingthe 'legitimastipendia.'Plutarchtells us that Ti. Gracchusannounceda numberof legislativeproposalswhen he was seeking reelection,one of which was for the reductionof the periodof militaryservice, 145 146 Above, n.9. Astin, Scipio 172; cf. Smith 8 and Brunt, IM405. Shochat 54 holds that youths had been recruited because there was a shortage of qualified men who were willing to serve; this seems most unlikely. 147 Cf. the incidents reported in Livy XXVIII.38.3-5; XXXIV.56.9-11; XXXVI.3.4-6. For the senate authorizing the conscription of men not normally liable for the levy see Livy XXV.5.6-8; XL.26.7; XLII.31.4, 33.4.Brunt, IM405-6, is right that such men could be conscripted without legislation (thus in 212 the senate authorized the conscription of boys under seventeen, but it was deemed necessary to pass a law ". . . uis perinde stipendia procederent, ac si septem decem annorum aut maiores milites facti essent": Livy XXV.5.6-8), but he overlooks the role of the senate. 148 The repetundae laws provide a good illustration of this legislative method. See especially Cic. Rab. Post. 9. 149 The two provisions which Plutarch records may have been selected more or less at random from a more comprehensive account by Plutarch or an intermediary. 320 J. W. RICH but whether Gracchusactually made these proposals is much disputed."50 Ciceroin the Pro Corneliomentioneda senatusconsultum of 109callingfor the abrogationof quae legesremmilitaremimpedirent;Asconius,commentingon the passage,says that the consul M. Iunius Silanuscarriedthe abrogationof severalrecent laws quibusmilitiaestipendia It is thusclearthat minuebantur."5' between 133 and 109 two or more laws were passed,presumablyby popularis tribunes,reducingthe numberof yearsfor which citizenswere liableto serve; one of these may have been C. Gracchus'lex militaris.The occasionfor the repeal of these laws was probably, as Asconius seems to imply, the reappearanceof the Cimbri,againstwhom Silanuswas sent; the new effortin the war againstJugurthamay have been anotherfactor.'52 It is usuallysaid that the laws reducingthe legitimastipendiawerepassedin responseto the growing distastefor militaryserviceand repealedbecauseit would otherwisehave been difficultto find enough qualifiedrecruitsfor the force to be raised in 109, which probably numberedabout 10,000.153 This interpretationis in my view mistaken. What was in questionwas the total periodof servicewhich a manmightbe calledupon to performduringhis lifetime;beforethe reductions,this hadbeen sixteen years.'54 Very few men in the second century can have served continuouslyfor periodsas long as this.The menwho would benefitfromthis legislation would be veterans,who would either be protected from recall altogetheror, if recalled,entitledto an earlierdischarge. The recallof veteranshad been a causeof contentioneversincelong periods of service became common.'55Commandersnaturally tended to prefer experiencedsoldiersto tiros. No doubt some veteranswerehappyto returnto the standards,and, as we haveseen,thereweresomewarsin whichservicewas especiallyattractive.156 However, recallcould be bitterlyresented.In 200 the senateruledthat only volunteerscould be takenfrom Scipio'sveteransfor the war againstPhilip,but 2,000 men mutiniedin 199,claimingthatthey hadbeen '5 Plut. TG. 16.1; cf. Dio fr. 83.7. For a recent defence of the authenticity of these proposals see Bernstein, TSG 215-219; see also Shochat 85-6. 151 Asc. 68 C. 152 Metellus' preparations against Jugurtha: Sall. BJ 43.3-4. We do not know whether deductions for clothing were restored now or later (see above n.144). 153 E. g. Gabba, ES 33; Smith 9; Brunt, IM407; Harris 46, 50. Shochat 65 holds that the laws were repealed because of "the reluctance of the populace to join the army and the desire to raise the standard of the soldiers." '54 Above n.10. 55 On this question see especially Smith 5-7; Astin, Scipio 169-170. " Those who protested in 171 (Livy XLII.32.6ff) were complaining not at being recalled but at the loss of their rank. Some enthusiastic soldiers may have achieved records almost as impressive as that with which Livy credits Sp. Ligustinus (XLII.34). I cannot share the confidence in the details of Ligustinus' story evinced by G. Perotti, CISA II (1974), 83-96. The Supposed Roman Manpower Shortage of the Later Second Century B.C. 321 As might be expected, enrolledagainsttheir will despitethe senate'sruling.157 reluctanceto reenlistwas particularlyevidentat the time of the Spanishwarsof the mid-century,servicein which was so much disliked.Q. FabiusMaximus Aemilianustook out an armyof rawrecruitsto FurtherSpainin orderto spare the veteransof recentwars, and the armywhich was sent to Q. Pompeiusin Nearer Spain in 140 was similarly composed.158Astin has plausibly suggested that the exemptions which were being claimed by tribunes in 151 and 138 were for veterans who had already served long periods.'59 The laws reducing the legitima stipendia were passed, in my view, because in the Gracchan age tribunes were ready to address themselves to the longstanding problem of the recall of veterans, to which the recent Spanish wars had given a new prominence. The laws would have had the effect of making it more difficult for commanders to stiffen new levies with seasoned troops. It was surely for this reason that the senate recommended their abrogation.t60 (iv) The ratio of allied to Roman troops. Brunt has argued that towards the end of the second century the Roman government increased the proportion of troops supplied by the Italian allies to compensate for the decline in the number of citizen assidUi.161 Polybius tells us that the allies supplied the same number of foot as the citizens, but three times as many cavalry; he implies that these ratios were in force both in the Hannibalic War and in his own day.'62This generalization is not borne out by the detailed evidence supplied by Livy. The ratio varied from period to period, and within the same period between different armies and different years. The allied cavalry was normally much less numerous than 157 Livy XXXI.8.6, 14.2; XXXII.3.2-7. 3,000 Scipionic veterans were enrolled as volunteers by Flamininus in 198 (Livy XXXII.9.1; Plut. Fl. 3.1) and 5,000 (including allies) by L. Scipio in 190 (Livy XXXVII.4.3; Brunt, IM 395 n.1, is wrong to link this passage with XXXVII.2.3). 158 App. Iber. 65, 78. "9 Astin, Scipio 170. Our optimate sources present them in a less favourable light: Plb. XXXV.4.6 with Walbank's note; Livy Per. XLVIII (pro amicis suis). 160 We may compare the senate's decree of 171 empowering the consul who obtained the province Macedonia to enrol veterans (Livy XLII.31.4, 32.6, 34.4). How the assembly's agreement was secured in 109 we can only guess. Perhaps men were persuaded that the national interest required the abrogation; possibly veterans were poorly represented at the meeting (dominated by city-dwellers ?). 161 IM404, 684-6; cf. Crawford, RR 101, 129-130. Gabba had already suggested (ES 33, 70-1) that there was a tendency in the later second century to spare citizen manpower at the expense of the allies, but, as Brunt remarks (IM 686 n.1), "the texts he cites show only that allied contingents remained an indispensable part of the Roman army." Shochat 89-93 accepts that the proportion of allies was increased; in his view this was a response to the reluctance of Roman citizens to serve. It cannot be shown that (at any rate after the end of the Spanish Wars in 133) there was such marked and general unwillingness to serve among the citizens as he supposes, and it seems to me unlikely that the senate would have condoned such reluctance in this or the other ways which he suggests. 162 Plb. I11.107.12, VI.26.7., 30.2. 21 322 J. W. RICH Polybius says. At the start of the Hannibalic War it was probably normal for the allies to furnish significantly more infantry than the citizens, but, when many allies revolted in the course of the war, the proportion of allied foot must have declined, perhaps to something approaching parity.'63From 200 to c. 179 the ratio of allied to citizen foot ranged between 3:2 and 2 :1; there is some evidence to suggest that in the 190's especially high demands were made on allies who had been disloyal. In the 170's the proportion of allied foot dropped nearly to parity.'" After 168, when Livy fails us, we do not have enough evidence to tell whether Rome kept her demands as moderate as in the 170's.165 Velleius says that the allies were justified in taking up arms in 91 to get the citizenship, since in every year and in every war the allies furnished twice as many foot and horse as the citizens.166It would have been natural for the Italians to exaggerate their services as passions became inflamed, but Brunt argues that such evidence as there is for the last years of the second century suggests that by then Velleius' claim may not have been wide of the mark.'67 Orosius (doubtless following Livy) tells us that in 110 A. Postumius was in command of an army of 40,000 in Numidia.168 Postumius is unlikely to have had more than two legions, i. e. 11,000 foot and horse. Brunt supposes that the regular allied contingents were twice as numerous and that the balance was supplied by allies not serving ex formula togatorum.'69 There are other possibilities: the number of allies not in the formula might be greater, or Orosius' figure erroneous. Valerius Antias said that at Arausio in 105 the Romans lost 80,000 soldiers and 40,000 campfollowers and only ten men survived to tell the tale.'70 Diodorus gives the more moderate figure of 60,000 for the Roman losses.'7' There were probably four legions in the Roman force. Brunt takes Diodorus' figure to imply an allied contingent about twice as large as the citizen. But Brunt, IM 677-681; cf. V. Ilari, Gli Italici nelle strutture militari romane (1974), 154-9. Afzelius, op. cit. n.29, 62-79; Brunt, IM681-4; cf. lhari,op. cit. 158-166. I suspect (cf. n.27 above) that there may be more inaccuracies in Livy's information than these writers allow, but not so much as to cast doubt on the overall trends implied by his figures. 165 Brunt, IM684, infers from Paus. VII.16.1 and App. Ill. 10 that a ratio near parity obtained in the armies fielded in Achaea in 146 and in Illyricum in 135. '66 Vell. II.15.2. 167 Loc. cit. n.16. Brunt also suggests that Appian's statement (Hann. 8) that in the Hannibalic War the allies had to furnish twice as many men as the citizens is an anachronistic application of the later ratio to the Hannibalic War. Ilari, op. cit. 166-171, holds that the proportion of allies remained slightly above parity from 179 to 91. 168 Oros. V.15.6. 169 Postumius' force included Ligurians and Thracians: Sall. BJ38.6. 170 Livy Per. LXVII; Oros. V.16.3-4. 171 Diod. XXXVI.1, presumably following Posidonius. Diodorus' use of the words btiXExTot and Xoy&be&in this passage is puzzling; he certainly cannot intend a distinction between citizens and allies, as the Loeb editor, F. R. Walton, supposes. 163 '64 The SupposedRomanManpowerShortageof the LaterSecondCenturyB.C. 323 Diodorus may have included allies not in the formula and he too may have exaggerated the Roman losses.172 Diodorus says that the force which L. Lucullustook againstthe Sicilian This should presumably slaves in 103 included14,000Romansand Italians.'73 be reckonedas a legion of 5,500 and 8,500 allies, a ratio of 17:11. Sulla said that Catulushad 20,300 men and Marius32,000 at the battleof Vercellae.'74It is hard to resist the conclusion that the ratio of allies exformula togatorum to citizens in this force must have been well below 2:1. Brunt's suggestion that exclusively allied troops had been detached to garrison the towns of northern Italy looks like special pleading, and he overlooks the possibility that the figures may have included allies not in the formula. The case for the view that at the end of the second century the ratio of allied to Roman troops was normally 2:1 does not seem to be strong. Even if it were correct, it would not follow that the proportion of allies had been raised to compensate for a decline in the number of citizen assidui. As we have seen, the proportion had been not far short of 2 :1 in the first two decades of the century. We do not know why the proportion dropped nearly to parity in the 170's or how long it remained so low. VII. Mariusand the Levy It used to be widely held that Marius transformed the Roman legions from a force raised by conscription from men of property into one composed of volunteer professionals, most of whom were penniless. A number of more recent writers notably Gabba, Smith and Brunt, have pointed out various defects in this thesis: the property qualification was so low that many of those who served before Marius must have been very poor; there was no sudden change in the length of service; conscription continued to be extensively used, and assidui still served. However, although these writers seek to minimize the significance of Marius' enrolment of capite censi they still suppose that Marius brought about a lasting change. Marius' action, in their view, was a response to the shortage of assidui; later commanders followed his example and the artificial crisis of manpower which had been brought about by the maintenance of the property qualification was thus resolved.'75 In this section I assess this interpretation of Marius' enrolment of capite censi I shall argue, first, that Marius' action can be accounted for without supposing that there was a shortage of assidui, and, secondly, that it was an 172 173 174 On the unreliability of the casualty figures given even by good sources see Brunt, IM694-7. Diod. XXXVI.8.1. Plut. Mar.25.4. 175 Gabba, ES30-56; Smith 9-10; 27ff; Brunt, IM406-413. For a more traditional view of the significance of Marius' action and references to earlier literature see J. Harmand, L'armeeet le soldat a Rome de 107 a 50 avant notreere (1967), 11-20. 21* 324 J. W. RICH isolatedincidentand that it was laterand in quite differentcircumstancesthat the propertyqualificationfor militaryserviceceasedto be observed. Sallustputs Marius'enrolmentof capitecensiin his first consulshipin 107 before his departurefor the war against Jugurtha.Plutarch, Florus and Exuperantiusagree,but a varianttraditiondatingthe episodeto the Cimbric Scholars Waroccurs in Pseudo-Quintilianand Gellius notes both versions.176 have surely been rightto preferSallust'sdating;the CimbricWarversionwas perhapsthe invention of a pro-Marianwriter seeking to representMarius' action as a necessaryexpedientin a time of nationalcrisis.177 for the legions The senatehad empoweredMariusto raisea supplementum alreadyin Numidia.'78Disregardingthe customaryproceduresof the levy, Mariusassembleda forceentirelyfromvolunteers;he did not debarthosewho lackedthe propertyqualificationand the majorityof those he took in fact did not possess it. 79 The breachof conventioninvolved in enlistingunqualified men evoked much disapproval,but, since none of them had been enrolled againsttheirwill, no sanctionscould be taken.Mariustook morementhanthe senate had decreed,'80but even so the numbersinvolved were not large. At most, the numberdecreedmay have been about 3,000, the numberenlisted about 5,000.181 Sallustrecordstwo explanationsfor Marius'enrolmentof capitecensi.One is favourableto Marius:he acted in this way becauseof a "shortageof good men" (inopiabonorum).The other is unfavourable:recruitsof this type most suited the purpose of a man seeking potentia.'82 The second explanation appealedmore to Sallust'sown preconceptions,as the space he devotedto it 176 Sall. BJ86.2-4; Plut. Mar. 9.1; Val. Max. 11.3.1; Ps.-Quint. Decd. 111.5; Gell. XVI.10.14; Flor. I.36.13; Exup. 2; Lyd. de mag. 1.48. 177 Gabba, ES 31-2. 178 Sall. BJ 84.3, quoted below. 179 Sall. BJ 86.2 ipse interea milites seribere, non more maiorum neque ex classibus, sed uti cuiusque lubido erat, capite censos plerosque. Volunteer armies had been raised by Scipio Aemilianus in 134 and perhaps by Scipio Africanus in 205, but they had not been authorized to hold a dilectus (above section IV and n.71). For the enrolment of men lacking the property qualification as a breach of convention see also the other sources cited in n.176. Plutarch may be although right that it broke not just convention but the law (3aMx T6v v6Otov xai ThV aouvr?Lav): the property qualification had no doubt rested on custom rather than statute up to C. Gracchus' day, it was probably reasserted in his lex militans, if the view of that law for which I argued above (section VI (ii)) is correct. On Brunt's view that Marius was entitled to enlist unqualified men by virtue of his discretionary imperium see above n.147. I do not understand why R. Mariano, Labeo 26 (1980), 354-364, supposes that the senate authorized Marius' action. 180 Sall. BJ 86.4 cum aliquanto maiore numero quam decretum erat. 181 Cf. Brunt, IM 430. 182 Sall. BJ 86.3 idfactum alii inopia bonorum, alii per ambitonem consulis memorabant, quod ab eo genere celebratus auctusque erat, et homini potentiam quaerenti egentissumus quisque opportunissumus,cui neque sua cara, quippe quae nulla sunt, et omnia cum pretio honesta videntur. The SupposedRomanManpowerShortageof the LaterSecondCenturyB.C. 325 betrays, but it is certainly wrong: it was twenty years later and with Sulla's examplebeforehim that Mariussoughtto dominatethe stateby forceof arms. Some of the other sources offer another explanation:Mariusenrolled men from the humblestclasses becausehe himself was a "new man."'83It seems most likely that this was a laterhistorian'sfancy.184 Most modern historiansexplain Marius'action in terms of the supposed shortageof assidui.Althoughthe force which Mariusneededwas so small,he would, it is supposed,have had difficultyin findingenoughassidui,or at any rateenough assiduiwho were willing to serve.To avoid losing popularityby conscriptingmen againsttheir will, he took unqualifiedvolunteers.'85 At first sight Sallustprovidesimpressivesupportfor this view, since one of the two explanationswhich he cites is that therewas a shortageof good men, which must mean, in this context,qualifiedmen.'86But the credentialsof this explanationmay be no betterthanthose of the otherexplanationwhich Sallust cites: both show evidentbias,andmay well be laterinventions.'87 It is unlikely that Mariusor his sympathizersattributedhis action at the time to inopia bonorum,for Mariuswould hardlyhave concededthat his recruitswere not boni. The view that Mariusdisregardedthe property qualificationin order to avoid unpopularityimplies that there was a great differencein the attitudes towards service under Marius held by those who possessed the property qualificationand those who did not. A passageearlierin Sallust'snarrative suggests that this was not the case: (senatus)supplementumetiam laetus decreverat,quia nequeplebi militiavolentiputabaturet Mariusaut belli usum 183 So Valerius Maximus and Florus, cited above n.176, probably following Livy. M. Sordi, Athenaeum 50 (1972), 379-385, holds that the explanation derives from Marius himself. Against this suggestion see Gabba, Athenaeum 51 (1973), 135-6. 18 See especially Gabba, ES 33ff; Brunt, IM 407. 186 Both explanations presuppose the doctrine that only property owners could make trustworthy soldiers. The doctrine (which also appears in the accounts of Plutarch, Valerius Maximus, Gellius and Exuperantius) ignored the fact that the property qualification was very low. Pseudo-Quintilian represents Marius as rejecting the doctrine: cum scires non ex censu esse virtutem, praeterita facultatium contemplatione vires animosque tantum spectasti; cf. the use of bonus in Marius' speech, Sall. BJ 85.5, 48.9. On the usage of bonus in Sallust see U. Paananen, Sallust's Politico-Social Terminology (1972), 59-64. The rejection of this interpretation of inopia bonorum by Shochat 62-3 is perverse: Sallust clearly implies that the capite censi enrolled by Marius were not boni. 187 Sallust's memorabant can hardly be pressed as evidence that the explanations were current at the time: Gabba, ES 33-4, 41-3 and Brunt, IM406-7 hold that the view that Marius enrolled the capite censi as a means to potentia cannot be contemporary, but their arguments are not conclusive: the doctrine that the loyalty of men without property could not be depended on could have been current in second century Rome, and in 107 Marius' opponents may well have alleged and perhaps believed - that he aspired to excessive power. It is possible that the explanation in terms of inopia bonorum derives from the tradition in which Marius' action was attributed to the Cimbric War. 184 326 J. W. RICH aut studiavolgi amissurus.Sed ea resfrustrasperata;tanta lubidocumMario eundi plerosque invaserat. sese quisque praeda locupletem fore victorem domum rediturumalia huiuscemodianimis trahebant, et eos non paulrm orationesua Mariusarrexerat.'88 The senate,accordingto Sallust,thoughtthat there was generalreluctanceto serve, but in fact Mariushad arousedgeneral enthusiasm.Other considerationssupport what Sallustsays here about the popularattitude.189It is improbablethat therewere manyproletarii who were eager to become legionaries irrespective of the character of the service for which they were enlisting.'90 The belief that Marius would bring men easy victory and rich booty could have led to general eagerness to enlist such as is attested at the outset of the Third Macedonian and Third Punic Wars.19' If there was no shortage of assidui and widespread eagerness to serve under Marius, Marius could have held a traditional dilectus without loss of popularity, simply by avoiding enlisting anyone who was reluctant to serve. Why then did Marius break with convention? The answer, in my view, is that Marius acted as he did not to avoid loss of popularity but to make himself yet more popular than ever. All the men he took were not just willing but eager to serve. If he had respected tradition, many of them would have been disappointed. Marius broke with tradition to gratify their wish. In doing so he incurred the senate's disapproval, but that mattered little to the declared enemy of the nobility. Sallust's words imply that there were some assidui in Marius' force,'92 and there may have been more than he and the other sources were prepared to admit. But no doubt the majority did not have the property qualification. This is not surprising: the less a man had to lose, the more tempting was the prospect of easy enrichment. Many of them may have been drawn from the urban plebs.'93 Thus Marius' enrolment of capite censi may be satisfactorily accounted for without supposing that there was a shortage of assidui. I turn now to the question of what happened to the property qualification for military service after 107. 188 Sall. BJ84.3-4. 189 The miscalculation which Sallust attributes to the senate suggests that legionary service was generally unpopular, but Sallust is hardly a reliable witness on the senate. His language there and at BJ85.3 may reflect first century dislike of the levy (on which see the evidence collected by Brunt, IM635-8, especially Cic. Att. IX.19.1). '90 Cf. Brunt, IM410-3. 191 See above section VI (i). As Brunt observes (IM407 n.3), the hopes of booty were fulfilled. 192 According to BJ 86.2, the force comprised capite censosplerosque. 193 Contra Gabba, ES 38-9, 56-7; other views are reported by J. Harmand, op. cit. n.175, 16. The veterans of the Cimbric War were countryfolk according to App. BC I. 29.132 but this does not entitle us to infer the same of the men enrolled in 107. According to Sall. BJ 73.6, Marius' consular campaign aroused the enthusiasm of opifices agrestesque omnes, i. e. the urban and the rural poor. The Supposed Roman Manpower Shortage of the Later Second Century B.C. 327 None of our sources suggest that Marius' successors followed his example. Gellius and Exuperantius state that he was the first to enrol capite censi; this implies that others did so later, but they give no indication when this took place.194 It seems unlikely that many proletarii were enlisted as volunteers in the years after 107. Few commanders in that period would have been as ready as Marius to brave the senate's disapproval or could have offered attractions comparable to those which had made men flock to serve under him. Marius' acceptance of unqualified men as volunteers cannot have affected the exemption which such men traditionally enjoyed from conscription except in emergencies. It would be most implausible to suppose that the senate was led by Marius' action, of which it strongly disapproved, to take the view that the conscription of proletariiwould now be proper in normal times. As before, commanders who attempted to conscript proletarii without the senate's support would run the risk of an appeal to the tribunes.'95 Rome had more men in service in 105-101, the years of the Cimbric threat, than at anytime since the Third Punic War.'" Four legions were annihilated at Arausio. The emergency measures taken after that disaster included requiring all iuniores to take an oath to remain in Italy.197 Further demands were made on Rome's manpowerwhen the Sicilianslavesrose again in revolt.'98 It could well be that in these years the senate deemed that the situation was so grave as to require the recruitment of proletarii. However, the assidui had met even greater challenges in the past, and in my view may have been numerous enough to man the armies which Rome fielded now. We do not know whether Marius raised troops on a large scale in 104.'" If he did, we cannot be sure that he appealed once again to the proletaii. The situation was very different: he might indeed have promised that the victors would gain not just booty but land, but no one this time will have thought that victory would be easy. 194 Gell. XVI.10.14; Exup. 2. Gellius' statement is based on his distinction between proletarii and capite censi (on which see above n.13); he supposes that only proletanii had been included in the earlier emergency enlistments (ib. 13). Gellius himself is not altogether clear on the point (cf. ib. 11), and, even if the distinction were historical, it seems unlikely that officers raising troops in an emergency would have taken account of it. 195 Cf. above n.147. 196 Brunt, IM 430-3. 197 Gran. Lic. 14F. (The numeral XXXV should surely be emended to XXXXVI). 198 A Roman force was not sent out until 103 (Diod. XXXVI.8.1). In 104 the governor, P. Licinius Nerva, made do with such forces as could be raised locally (Diod. XXXVI.3-4; cf.1 &6tOQ()V O6vto)v E'L; ajtocToXiv O?QaTlwTWV koyacbwv). 199 Cf. Brunt, IM 430-1. Marius took over the two legions raised by P. Rutilius Rufus (Ps.Frontin. Strat. IV.2.2), and no doubt strengthened them with African veterans or new recruits. Brunt may be right, despite Plut. Mar. 24.2, that he raised two further legions for his army in Gaul. 328 J. W. RICH I conclude that no significant enrolments of proletarii, either as volunteers or conscripts, took place between Marius' enrolment of 107 and the Social War, except perhaps in 105-103 to face the Northern barbarians or the Sicilian slaves. Marius' action remained an isolated incident.200 The armies which were fielded by the two sides in the Social War constituted the greatest mobilization of Italian manpower which had yet occurred, exceeding even what had been achieved in the Hannibalic War.20'The Roman government drew on every available resource to meet the crisis. Freedmen were enlisted for coastal defence.202Freeborn proletarii must also have been called upon. We hear of one army composed mainly of men from the city of Rome :203 these must have been largely proletarii. The armies which fought in the civil war of 83-2 were hardly less numerous,204and were naturally raised by procedures very different from the traditional dilectus. The levying officers of this war will hardly have been concerned with the property qualification. Once again we hear of troops being drawn from the city of Rome.205 The manpower demands of the civil wars of 49 to 30 were on the same scale,206 and there can be no doubt that proletarii were called on again.207A levy was held in the city of Rome in 43.208 However, in the period between the two bouts of civil war far fewer men were in service. In the years 70-50 Italy supplied no more troops than in the quieter years of the second century.209 Our sources give no indication as to whether proletarii were recruited in these years, either as volunteers or as conscripts, but in my judgement it is likely that the property qualification was not now enforced. 200 I agree with Brunt (SCRR 99) that Saturninus' agrarianschemes were not a consequence of Marius' enrolment of proletarni. The majority of Marius' African veterans were assidui (if the supplementum decreed to Marius was 3,000, there were 8,000 citizens serving there in 107). In my view, most, if not all, of the veterans of the Cimbric Wars may have been assidui. In view of the low level of the property qualification there is no need for surprise that land allotments were attractive to assidui. 201 Brunt, IM 435-440, estimates that some 300,000 men were under arms in 90-89. 202 App. BC I.49.212; Livy Per. LXXIV; Macr. 1.11.32. 203 Dio fr. 100. 204 Brunt, I 441-5. 205 App. BC 1.82.373. 206 Brunt, IM 473-512. I cannot, however, accept Brunt's view (IM 409) that this is what Appian meant when he explained the disloyalty of the legions in 41 by the fact that ou1 Tot; 3TatTQOL; 'EHoLV tX xCatakrtyov ovv'yovTo (BC V. 17.68). Appian's words would be a very roundabout way of referring to the socio-economic composition of the troops. Brunt is right that Appian cannot mean that the legions were composed of volunteers, since compulsion had in fact been extensively applied. I take Appian's point to be that the troops had been raised by irregularprocedures, very different from those which would have been employed in normal times. 208 Cic. Phil. X.21; Fam. XI.8.2. The Pompeians held a levy circa urbem in 49: Caes. BC I. 14.4. 209 Brunt, IM 446-472, with summary at 446-7. 207 The Supposed Roman Manpower Shortage of the Later Second Century B.C. 329 In the late Republic levies were usually held only in particular districts; levies throughout Italy were required only in emergencies. Commanders raised their armies mainly through the agency of the authorities of the colonies and municzpia.210 An important role was also played by conquisitoressent out by commanders to seek out recruits.21'We cannot be sure how far the practice of this period differed from that of the second century, since there is so much uncertainty about how the levy was conducted in the second century.212 However, it seems to me likely that there had been a very substantial change in practice, that what had remained essentially a levying system in which those eligible were required to present themselves for selection had been transformed into one in which recruits were sought out, and that along with the other changes the property qualification had ceased to be observed. If this is correct, the Social War should be identified as the turning point. The levying procedures of the second century continued, in my view, to be observed down to the outbreak of that war; no departures from normal practice occurred in the years leading up to its outbreak except for the isolated incident of 107 and perhaps some emergency measures in 105-103. The levying system was transformed by the demands of the great wars of the years 90-82. Many of the new expedients adopted in those years persisted and became established practice in the post-Sullan period; the abandonment of the property qualification was one of these. That Sulla's successors should not have reverted to the traditional procedures of the dilectus need not surprise us. The seventies were troubled years. Those Italian communities which had hitherto sent allied contingents now contributed to the legions. A comprehensive list of iuniores can hardly have been available to the Roman authorities before the census of 70/69.213 The failure to complete the census thereafter would have made it impossible to keep the list up to date. The abandonment of the property qualification may not have greatly changed the social composition of the legions. I argued in section III that a high proportion of those impoverished peasants who stayed in the country in the second century may still have had enough property to qualify as assidui. In my view this may have been true in the first century too. The uprooting of the peasantry had continued, perhaps at an increasing rate. But the capacity of the countryside to support the dispossessed was no greater in this century than in the last. The drift to the city of Rome gathered momentum.214 Brunt has shown Brunt, JRS 1962, 74, 85-63, IM 631. Cic. Prov. Cons. 5; Mil. 67; Att. VIII.21.1. 212 See above section II 213 For the tabulae iuniorum, which must have played an important part in the traditional dilectus, see Livy XXIV.18.7. On the census of 86/5 see Brunt, IM 91-3. 214 See Brunt, IM 376ff; Hopkins 57-8, 96-8. 210 211 J. W. 330 RICH that the armiesof the late Republicwere raisedmainlyin the country;levies Although, as I were held in the city only rarely, at times of great need.2"5 believe,the levying officersof the late Republicregardedthemselvesas under no obligationto take accountof a man'sproperty,most of those whom they enlisted may still have possessedthe meagreamountof propertyrequiredto qualifyas an assiduus. VIII. Conclusion As I arguedat the beginningof this paper,it is in principleunlikelythatthe Roman governmentin the second century set such store by maintainingthe property qualificationfor military service that they allowed a grave, but entirely artificial, manpower crisis to develop. The doctrine that such a shortageof assiduidid occur rests on an accumulationof argumentsbasedon variouskindsof evidence.I hope to haveshown thatat everypoint the casefor the doctrine is weak and alternative,often preferable,interpretationsof the evidence are available.I conclude that we should abandonthe belief that a serious shortageof assiduidevelopedin the course of the second century. The number of assidui probably rose in the first decadesof the second century,as the citizen body as a whole expanded;thereafterthe changesin the pattern of agriculturemust have caused the numberto decline. We cannot, however,determinehow steep the declinewas. It is possiblethat the number did not fallbelow the levelof the earlyyearsof the century.If it did, it maynot havefallenso far that, althoughthe demandfor legionarieswas now less great, the burdenof the levy on the remainingassiduiwas significantlygreater. There is no good reason to reject the evidence of our sources that the concern about manpower which was one of the motives for the agrarian schemesof LaeliusandTi. Gracchuswas directednot to the numberof assidui, but to the decline (real or supposed)in the numbersof the free population (citizen and perhapsalso allied). The view that the propertyqualificationfor militaryservicewas progressively reduced in order to create more assiduiis only one of a numberof hypotheseswhich might be put forwardto accountfor the variantfiguresfor the qualificationin our sourcesandis open to a numberof objections.It is best to admit that we do not know how these discrepanciesshould be accounted for. None of what our sourcestell us about the levy in the latersecond century constitutes good evidence for the supposed shortage of qualified recruits. Marius, in my view, enrolled capite censi not becausehe would have lost popularityif he had confinedhimselfto assidui,but in orderto earnstill more 215 Above, n.210. The SupposedRomanManpowerShortageof the LaterSecondCenturyB.C. 331 popularityby enrollinga force composed entirelyof men who were eagerto serve, many of whom would have been disappointedif he had respected tradition. It is unlikely that succeeding commanders followed Marius' example;his action is best seen as an isolatedincidentwhich had no effect on subsequent levying practice. It was probably in the turbulentyears of the eighties that the property qualificationfor military service ceased to be observed. Universityof Nottingham J. W. Rich