The Heredity of Senatorial Status in the Principate

Abstract Since Mommsen, it has been a tenet of Roman history that Augustus transformed the ‘senatorial order’ into a hereditary class, which encompassed senators, their children, grandchildren and great-grandchildren in the male line. This paper shows that the idea of a hereditary ordo senatorius is a myth without foundation in the evidence. Augustus and his successors conferred new rights and duties upon relatives of senators, but did not change their formal rank. Moreover, the new regulations applied not to three generations of descendants, but only to persons who stood under a senator's patria potestas during his lifetime. Emperors protected the honour and property of these filii familias of senators, in order to incentivise them to participate in politics and invest their wealth into munificence. The Supplementary Material available online gives all known early imperial holders of the title clarissimus vir in the province of Africa (Supplementary Appendix 1), all known early imperial clarissimi iuuenes (Supplementary Appendix 2) and all known early imperial clarissimi pueri (Supplementary Appendix 3).

History, 1 and has been endorsed by leading historians of the Senate such as Richard Talbert, Werner Eck and Géza Alföldy. 2 Modern scholars derive this interpretation of the 'senatorial order' from the work of Theodor Mommsen. In the third volume of Römisches Staatsrecht, Mommsen identied six rights and duties that separated members of the 'senatorial order' from other citizens: they had the right to wear the toga with the broad purple stripe (latus clauus); they bore the status designation clarissimus; they participated in meetings of the Senate; they were prohibited from entering marriages with descendants of slaves; they were banned from engaging in certain forms of commerce; and they were exempt from the duty to nance acts of municence in their cities of origin. But more important than the precise nature of these entitlements and obligations, was what they implied about the nature of the 'senatorial order' as a class. Mommsen explicitly likened the 'senatorial order' to a 'hereditary peerage'. On his reading, like the nobilities of the ancien régime, so the Roman senatorial elite was an 'aristocracy of birth', whose members enjoyed a set of rights and duties that separated them from other citizens. 3 By describing the 'senatorial order' as a hereditary aristocracy, Mommsen offered later generations of scholars a powerful model to understand the legal structure of the Roman elite and the terminology employed in ancient texts. The strength of his explanatory framework emerges most strikingly from the fact that it was adopted by scholars who disagreed with Mommsen in their interpretation of individual items of evidence. In two separate papers, published in the 1970s, André Chastagnol and Claude Nicolet noted that in some early imperial texts descendants of senators are described not as members of the ordo senatorius but as equestrians. 4 Yet this discovery did not lead them to abandon the theory of a hereditary 'order'. In their view, if there was some 'ambiguity' in the ways in which relatives of senators were described, this was simply due to the fact that the full legal consequences of the reform took some time to be worked out. According to Chastagnol and Nicolet, such terminological 'confusion' only disappears when under Caligula the right to run for senatorial magistracies was restricted to members of the 'senatorial order'. 5 But the two scholars did not doubt that Augustus transformed the ordo senatorius into a hereditary class. More specically, Chastagnol and Nicolet argued the birth of the 'senatorial order' occurred in 18 B.C.E. In this year, by increasing the minimum census for senators from 400,000 HS (the same as for equestrians) to 1,000,000 HS, and by prohibiting three generations of their descendants from marrying the offspring of slaves, Augustus brought into being a new notion of the ordo senatorius as a group which encompassed not only senators themselves but also their families. 6 When Mommsen developed his theory, and when Chastagnol and Nicolet adopted and rened it, it was not unreasonable to think of the Roman elite as a hereditary group. Until the late twentieth century, it was widely believed that almost all sons of senators became senators too. Under such conditions, it was plausible to conceptualise the Roman elite as a class whose membership was hereditary not only in practice, but also in law. But in 1985, Keith Hopkins and Graham Burton published a ground-breaking paper in which they demolished the long-standing belief that nearly all male descendants of ofceholders participated in politics. Through an analysis of a sample of all consuls appointed between 14 and 235 C.E., Hopkins and Burton demonstrated that only a minority of sons of senators followed their fathers' careers. Most led a life of otium, free from the expenses and dangers of political life. 7 Some rejected this radical conclusion out of hand. 8 However, only two scholars put forward an alternative interpretation of the evidence on which Hopkins and Burton's argument was based. By recalculating how many sons of Antonine consuls became consuls too, Johannes Hahn and Paul Leunissen sought to demonstrate that the high turnover within the Senate was due to a decline in birth rates, not a lack of political participation. 9 But as Burton pointed out in a reply to their paper, low fertility cannot explain why 'ordinary consuls', holders of the most prestigious magistracy, had more politically successful sons than the slightly less prominent 'suffect consuls'. Moreover, on any demographic model, if the great majority of sons of senators embarked on a political career, many more brothers would have held senior ofces than are actually attested. 10 Hopkins and Burton's demonstration that most sons of senators did not run for high ofce is convincing.
But this insight creates a difculty for the theory of a hereditary 'senatorial order'. If most children, grandchildren and great-grandchildren of a senator did not embark on a political career, but nevertheless inherited his rank, this would mean that in the rst two centuries C.E. there were several times more politically inactive members of the 'order' than actual senators. Hopkins and Burton explicitly endorsed this hypothesis. They argued that the existence of the 'order' was one of the major reasons why so many descendants of senior ofceholders withdrew from politics: 'In sum, the sons of senators inherited elements of their fathers' status, and that alone diminished their need to become full members of the senate in Rome.' 11 But there is a problem with this argument. No such non-ofceholding members of the 'senatorial order' are attested. Hopkins and Burton ascribed the lack of evidence to the bias of surviving sources: 'Those who did not achieve high positions were less likely to be honoured in a commemorative (as opposed to a tombstone) inscription.' 12 Of course, as a general proposition, it is true that politically inactive persons are less well documented than senators. But the early Empire is one of the best attested periods in Mediterranean history. In their local communities, politically inactive descendants of senators would have been exceptionally inuential gures. Surely some of them should be recordedif not in honoric inscriptions, then at least in literature, legislation or (as Hopkins and Burton suggest in the quotation above) epitaphs. The complete lack of evidence for this group of persons suggests that something is wrong with the theory of senatorial heredity.
In this paper, I challenge the conventional view that Augustus transformed the 'senatorial order' into a hereditary class. I argue that that the Romans did not recognise any difference between senatus and ordo senatorius. As in the late Republic, so also in the early Empire, membership of the Senate depended on fullling a minimum census and securing election to a republican magistracy. This is not to deny that Augustus conferred some new rights and duties on relatives of ofceholders; but the impact of these measures was much more limited than conventional interpretations suggest. Augustus neither created a hereditary 'order', which extended multiple generations after a senator's death, nor made rank hereditary. Rather, his reforms regulated the behaviour of a small group of young men and women who came under the patria potestas of a senator. 7 Hopkins and Burton 1985. 8 Alföldy 1986;Jacques 1987. 9 Hahn and Leunissen 1990. 10 Burton 1995. 11 Hopkins and Burton 1985, esp. 189-93, quoting 193. 12 Hopkins and Burton 1985: 191-2. I begin by re-examining the evidence on which the conventional view of senatorial heredity is based. I trace the ways in which descendants of senators are described in law (I), literature (II) and epigraphy (III), and explore the meaning of the new status designation clarissimus (IV). These different bodies of evidence show that senatorial rank did not become hereditary in the early Empire and that the Romans did not conceive of the 'senatorial order' as a group distinct from the 'Senate'. I then develop an alternative interpretation of early imperial regulations dealing with the relatives of senators. If these new laws are situated in the context of the Roman institution of patria potestas, it becomes clear that they affected not all children, grandchildren and great-grandchildren of an ofceholder, but only those of his descendants who stood under his patria potestas during his lifetime (V). The purpose of these regulations was to safeguard the honour of future members of the imperial elite and motivate them to embark on a senatorial career (VI). Augustus and his successors sought to increase political participation amongst ofceholding families both for ideological reasons and because they depended on the wealth of large landowners to nance public services in Rome (VII).

I THE LEGAL CONSTITUTION OF THE SENATE
1 The formation of a 'senatorial order' is conventionally associated with the marriage legislation of Augustus. 13 The following provision of the Lex Iulia de maritandis ordinibus, passed in 18 B.C.E., allegedly transformed the Senate into a hereditary class: The lex Julia provides that: 'A senator, his son or his grandson, or his great-grandson by his son shall not knowingly or fraudulently become betrothed to or marry a freedwoman, or a woman who is or has been an actress or whose father or mother are or have been actors. Nor shall the daughter of a senator, his granddaughter by his son, or great-granddaughter by his grandson become betrothed to or marry, knowingly or fraudulently become betrothed to or marry such a man. Nor shall any of these people knowingly or fraudulently become betrothed to or marry such a woman.' 14 This provision prohibited senators and their descendants for three generations in the male line from marrying men and women of servile descent. If challenged in court, children from such unions lost their rights of inheritance, and the fortunes of the deceased would be divided between the imperial sc and the accusers. 15 But does this mean that the law created a 'senatorial order' whose members enjoyed a clearly dened set of rights and duties? On the face of it, the excerpt does not warrant such a far-reaching interpretation. The Lex Julia does not contain the term ordo senatorius. Moreover, neither juristic commentaries nor historians offer any hint that the law changed the place of descendants of senators in ofcial orders of precedence. 16 The Lex Julia provides a surprisingly unstable foundation for the theory that Augustus transformed the Roman ruling class into a hereditary 'order'.
2 Another reform that supposedly contributed to the formation of a new ordo senatorius is reported by Suetonius in his biography of Augustus.
In order to accustom descendants of senators more quickly to politics, he allowed them immediately after the toga uirilis to assume the broad purple stripe and to attend meetings of the Senate. And when they began military service, he gave them not merely a tribunate in a legion, but the command of an ala of cavalry as well; and to furnish all of them with military experience, he usually appointed two of these laticlauii to command each ala. 17 According to Suetonius, upon reaching maturity, descendants of senators (liberis senatorum) received the status symbol of senators, the toga with the broad purple stripe (latus clauus). Henceforth, they served as junior ofcers in the imperial army (tribuni laticlauii), fullled various administrative functions in Rome (the so-called uigintiuiri) and were allowed to attend meetings of the Senate as spectators. By creating this new preparatory stage for a senatorial career, Augustus tried to motivate them to participate in politics. 18 But again, this does not mean that the reform made senatorial rank hereditary. Other texts show that sons of equestrians or city councillors also frequently served as laticlauii. 19 More importantly, laticlauii did not enjoy senatorial rank. This clearly emerges from several passages in Dio, which Nicolet has carefully examined. 20 For example, when in 9 B.C.E. Augustus' stepson Drusus was buried, his body was carried by sons of senators and of equestrians: 'his body was carried to the Campus Martius by the equestrians, both those who belonged to the equestrian order proper, and those who were of senatorial descent'. 21 Dio here clearly states that sons of senators were equestrians. This is conrmed in the discussion of cash subsidies given in 4 C.E. to prospective ofceholders: Augustus supplied them with the senatorial minimum census, 'because many young men, both those who descended from senatorial families and the other equestrians, were impoverished through no fault of their own'. 22 The same usage is found in Dio's description of Augustus' funerary cortège: 'Then the equestrians, both those who descended from the order and the others, ran around the pyre.' 23 The 'others' mentioned by Dio can only be laticlauii: they descended from senators, but were members of the equestrian order. As mentioned in the introduction above, Nicolet and Chastagnol attempt to reconcile this evidence with the theory of a hereditary 'senatorial order' by suggesting that the status of descendants of senators remained 'ambiguous' in Augustus' and Tiberius' reigns. On this reading, only after Caligula would they be consistently classied as members of the 'order'. 24 But this interpretation is unconvincing. Cassius Dio is not an Augustan, but a Severan author. The precise terminology employed in these passages, and the fact that he felt no need to explain it to his readers, clearly show that still in the early third century it was taken for granted that laticlauii were equestrians.
Upon closer inspection, then, the latus clauus creates further problems for the view that senatorial rank became hereditary in the early Empire. To be sure, the fact that descendants of senators automatically received the toga with the broad purple stripe, whereas others had to apply for it, made it easier for them to start a political career. In this way, as Augustus hoped, the creation of the laticlauii may indeed have increased political participation amongst ofceholding families. But the reform did not change the formal rank of descendants of senators or make them members of a 'senatorial order'.
3 According to Cassius Dio (54.2.5), Augustus prohibited senators, their children and their descendants (τοῖς ἐγγόνοις), 'at least in so as far as they belonged to the equestrian order' (τοῖς γε ἐν τῇ ἱππάδι δῆλον ὅτι ἐξεταζομένοις), from performing in gladiatorial shows and other public entertainments. In other words, only those relatives of senators who fullled the equestrian minimum census of 400,000 HS were subjected to the regulation. Under his successor Tiberius, this prohibition was extended to include all equestrians and three generations of their cognatic descendants and ancestors. 25 The fact that Dio again classies descendants of senators as equestrians further undermines the theory of a hereditary 'order'.
4 Two nal regulations on relatives of senators are only attested in excerpts of the Pauli Sententiae, a legal handbook compiled sometime in the third century. 26 According to a fragment that has survived in a manuscript in Leiden, 'senators and progenitors under whose potestas they stand' were prohibited from engaging in tax-farming, merchant shipping or commercial horse breeding. 27 This was a restatement of an ancient rule. According to Livy, the Lex Claudia, passed in 218 B.C.E., prohibited 'senators and whoever had a senator as father' from owning merchant ships. 28 There are two differences between Livy and the Pauli Sententiae. First, according to the latter text, the law outlawed investments not only in trade, but also in tax-farming and horse-breeding. It seems that sometime in the early Empire the scope of the Lex Claudia was expanded. Perhaps Hadrian had modied the law? Dio relates that this emperor prohibited senators from investing in tax-farming, either directly or through intermediaries. 29 Second, the statement in Pauli Sententiae affects a different group of persons than the earlier law. While the original Lex Claudia regulated the economic activity of senators and the persons under their patria potestas, the later text discusses senators and their own patres familias: 'the progenitors ( parentes) under whose potestas they stand'. But this discrepancy does not mean that the law had changed. It is important to keep in mind that the Pauli Sententiae was not a lawcode, but a collection of juristic opinions. By suggesting that prohibitions on the fortunes of senators should also apply to their pater familias, the author expressed his personal view on how the Lex Claudia should be interpreted. His interpretation makes sense, since the pater familias was the legal owner of all assets managed by persons who stood under his patria potestas. 30 Perhaps the discussion was inspired by a specic case, in which a senator's father had invested in one of the prohibited enterprises. In any case, the Lex Claudia did not change the formal rank of senators or imply that their relatives had become members of a hereditary 'order'. 25 See EAOR III 2 = Stelluti 1997: no. 100a with the commentaries and discussions by Levick 1983;Stelluti 1997: 1.173-6;Baltrusch 1989: 195-206. 26 On this text, see Levy 1930;Liebs 1989;Ruggiero 2017. 27 Paulus, Sent., Frag. Leid. 3, 'Senatores parentesue eorum, in quorum potestate sunt, uectigalia publica conducere, nauem in quaestum habere, equosue curules praebendos suscipere prohibentur' (for the text, see Archi et al. 1956 5 Another fragment of the same handbook, preserved in the Digest, states that not only senators themselves, but also their children, grandchildren and great-grandchildren in the male line could not be forced to pay for compulsory public functions in their hometowns: Senators and their sons and daughters, whenever born, and equally their grandsons <and granddaughters>, great-grandsons and great-granddaughters by the son are removed from their scal domicile (origo), even if they retain municipal dignity. 31 The author of the Pauli Sententiae asserts that even those descendants of senators who were city councillors were not subject to local liturgies: they 'are removed from their scal domicile, even if they retain municipal dignity', as the text puts it. Again, the opinion was probably inspired by a specic case; strictly speaking, it was superuous to add that the immunity affected children of senators 'whenever born' ('quoquo tempore nati nataeue') and 'even if they retain municipal dignity' ('licet municipalem retineant dignitatem'). Presumably, a senator wished to know whether he was liable to pay for the public functions of a son who had been elected into a local city council and who had been born before his own promotion to the imperial elite. The jurist afrmed that the immunity from municence in cities outside Rome also applied in this borderline situation. But irrespective of why the statement was included in the Pauli Sententiae, what matters for our purposes is that non-ofceholding relatives of senators are classied as persons 'of municipal dignity', not as members of the 'senatorial order'.
To sum up, an initial review of the legal evidence suggests that the theory of a hereditary 'senatorial order' rests on shakier ground than is conventionally realised. While it is true that Augustus regulated the lives of descendants of senators in new ways, there is no sign that their formal rank changed. On the contrary, all evidence considered so far suggests that they were classied not as members of the 'senatorial order', but as equestrians or persons of 'municipal dignity'.

II SENATE AND SENATORIAL ORDER IN LITERATURE
Legal texts give no indication that the status of relatives of senators changed in the early Empire. Does early imperial literature show traces of a hereditary senatorial class? In the Republic, the term ordo senatorius referred to the same social group as the word senatus. 32 The only difference in usage was that the ordo senatorius highlighted the 'rank' (ordo) of a senator, 33 while the word senatus evoked more immediately the institution of which he was a member. 34 If Augustus transformed the Roman elite into a hereditary class, we would expect that the semantics of the word changed accordingly. The expression 'senatorial order' (ordo senatorius) should refer no longer only to senators themselves, but also to their children, grandchildren and great-grandchildren in the male line. This is what is normally assumed. According to Chastagnol and Nicolet, though traces of the old usage survived until the reign of Caligula, the term began to be used in 'un nouveau sens' from the reign of Augustus. 35 Hopkins and Burton concur: 31 Dig. 50.1.22.5, with Mommsen's emendation, 'Senatores et eorum lii liaeque quoquo tempore nati nataeue, itemque nepotes <et neptes>, pronepotes et proneptes ex lio origini eximuntur, licet municipalem retineant dignitatem.' 32 As shown by Nicolet 1976: 33-4 andaccepted by Chastagnol 1973: 584-5;1992: 34-5. 33 OLD s.v. ordo 4 andTLL 9.2.961. Cohen 1975;Nicolet 1984a;Rilinger 1991 explore Roman conceptions of ordo. 34 OLD s.v. senatus 1. 35 Chastagnol 1973: 585-93, quoting 585;1992: 37-40, quoting 37;cf. Nicolet 1976: 35-8, esp. 35 n. 37: 'Sous l'Empire, ordo senatorius désigne clairement les sénateurs et leurs familles (Suet., Calig. 17.3), et donc la qualité 'under the emperors, the term came to include senators' immediate family and descendants'. 36 However, is it actually true that in the Principate the expression 'senatorial order' referred not only to senatorial ofceholders, but also to their relatives? Surprisingly, this assertion has never been systematically tested. I have collected all instances in which the expression ordo senatorius is employed in the Latin literature of the Principate. From the time of Augustus until the third century, it appears thirty-four times. The occurrences may be divided broadly into three groups.
Firstly, in eight cases, the expression is employed to describe promotions into the 'senatorial order'. In all these examples, it can unambiguously be determined that members of the 'order' are synonymous with the around 600 ofceholders who were members of the 'Senate'. For example, when Valerius Maximus records that a man has been 'adlected into the senatorial order', this can only mean that he has become a senator in the full sense of the wordadlectio is the legal term for the formal process of admission into the Senate. 37 Throughout the period from the time of Augustus until the third century, there is no change from republican usagethe term ordo senatorius refers to the same group of persons as the words senatus or senatores. 38 Secondly, in another six instances, the 'senatorial order' is mentioned in discussions of a loss of status. Men were 'expelled from the senatorial order', 'lost' or 'gave up their rank'. Again, this means not merely that they were demoted from a diffuse 'senatorial class', but concretely that they lost their seat in the Senate. For in all cases, it can be shown that the disgraced persons were current or former ofceholders. Again, no shift in the semantics of the word is discernible. The meaning of the word ordo also does not change over time, as Nicolet and Chastagnol assumethe evidence mainly consists of texts written in the late rst and second centuries, long after the alleged reform of the 'senatorial order' carried out by Caligula. 39 des ls de sénateurs, même lorsqu'ils n'ont pas encore exercé les honneurs.' ('Under the Empire, ordo senatorius clearly denotes senators and their families (Suet., Calig. 17.3) and thus the condition of sons of senators, even if they have not yet held ofce.') On the 'ex senatorio ordine aurigantibus' in Suet., Calig. 18.3, see n. 42 below. 36 Hopkins and Burton 1985: 193. 37 Val. Max. 2.2.1, 'in ordinem senatorium allectum'. Chastagnol 1992: 97-120 and Millar 1992: 293-7 trace the development of adlectio in the early Empire. 38 When Tacitus in Hist. 3.52.3 refers to 'Plotius Grypus, who had recently been adlected into senatorial order ('nuper a Vespasiano in senatorium ordinem adscitum') and put in charge of a legion', he must mean an adlectio inter quaestorios. A legionary command was the rst ofce senators customarily accepted after their quaestorship. Suetonius' discussion of the social origins of Vitellius' family in Vit. 1.3 -'who returned to the city and into the senatorial order' ('repetisse urbem atque ordinem senatorium')concludes a section on their political achievements. Similarly, in Suetonius' biographies of grammarians and in the commentary of Pomponius Porphyrio on Horace, membership in the ordo is explicitly linked to a political career: Suet., Gram. et rhet. 25.7, 'in ordinem senatorium atque ad summos honores processerint' ('they advanced to senatorial rank and to the highest honours') and Commentum in Horatium ad Sermones 1.6. 34-5 (ed. A. Holder, Pfaffenhofen, 1894, 263.13-14), 'qui se inserit senatorio ordini et rem publicam administrare cupit' ('who inserts himself into the senatorial order and desires to govern the state'). Finally, Tac., Hist. 2.86.1 'senatorium ordinem reciperauerat' ('he had recovered senatorial rank') refers to Antonius Primus' restoration to full senatorial membership and Tac., Hist. 2.82.2 'plerosque senatorii ordinis honore percoluit' ('he adorned many with the honour of senatorial rank') alludes to Vespasian's expansion of the Senate. 39 Suet., Iul. 43.1, 'repetundarum conuictos etiam ordine senatorio mouit' ('he also removed those who had been convicted of extortion from the Senate'), discussing former governors. Tac., Hist. 4.44.2, 'Octauium Sagittam et Antistium Sosianum senatorii ordinis egressos exilium in easdem insulas redegit' ('after they had left the senatorial order, he exiled Octavius Sagitta and Antistius Sosianus to those same islands'): Sagitta appears in Ann. 13.40.1 as tribune of the plebs and Sosianus in Ann. 14.48-9 as praetor. Tac., Ann. 6.48.4, 'Carsidius Sacerdos praetorius ut in insulam deportaretur, Pontius Fregellanus amitteret ordinem senatorium' ('to deport the ex-praetor Carsidius Sacerdos to an island and to strip Pontius Fregellanus of senatorial rank'): the persons in question are a former praetor and a consul suffectus, see PIR 2 C 451 and P 800. Tac., Ann. 12.4.3, 'at Silanus, insidiarum nescius ac forte eo anno praetor, repente per edictum Vitellii ordine senatorio mouetur quamquam lecto pridem senatu lustroque condito' ('But Silanus, unaware of the plot andas it turned outpraetor in that year, was suddenly expelled from the Senate through an edict of Vitellius, despite the fact that the review of the Senate Thirdly, there are eighteen examples in which the word was used to distinguish the rank of senators from other status groups. These passages may be divided into two sub-groups. In six cases, texts refer to the status of an individual member of the 'senatorial order'. In four of these six, other evidence enables us securely to identify the person as a senatorial ofceholder. 40 In the remaining two examples, the uiri ordinis senatorii in question are otherwise unattested. However, they were clearly highly informed political operators: Seneca mentions a Rufus, uir ordinis senatorii, who told a joke about the emperor Caligula's sacricial practice, and Tacitus calls two participants of the Pisonian conspiracy uiri ordinis senatorii. The fact that they appear in the context of high politics strongly suggests that they were full members of the Senate. 41 In the other twelve cases, the word ordo senatorius is employed to differentiate groups of men of 'senatorial rank' from other social strata. 42 Also in these examples, it can almost always be shown that the authors had in mind full members of the Senate, and not their relatives. For example, in his discussion of the aftermath of the trial against Libo, Tacitus writes that 'the estates of this prominent senator were divided amongst his accusers' and that those amongst them 'who were of senatorial rank' ('qui senatorii ordinis erant') were in addition honoured with praetorships. The reference is to the group of senior had already been carried out and the census concluded'). Tac., Ann. 12.52.3, 'laudati dehinc oratione principis, qui ob angustias familiares ordine senatorio sponte cederent, motique qui remanendo impudentiam paupertati adicerent' ('those who voluntarily resigned senatorial rank were praised in the emperor's speech, those who by staying added infamy to poverty were expelled'): about a speech in the Senate. See also Val. Max. 2.9.3, '… perinde ac malo exemplo luxuriosum in ordine senatorio retentum non esse' ('because his extravagance set a bad example he was not kept in the senatorial order'): an exemplum from the Republic, in which the supposed ordo did not yet exist. 40 Known ofceholders: Tac., Agr. 4.1, 'pater illi Iulius Graecinus senatorii ordinis' (tribune of the plebs and praetor: see AE 1946. 94 = CIL 6.41069); Val. Max. 5.8.5, 'A. Fuluius uir senatorii ordinis' (the same person is called a senator in Sall., Cat. 39.5); Vell. Pat. 2.69.1, 'quo tempore Capito, patruus meus, uir ordinis senatorii, Agrippae subscripsit in C. Cassium' ('at that time my uncle Capito, a man of senatorial rank, signed Agrippa's indictment of Gaius Cassius'): the prosecution of Caesar's murderers took place many years before the alleged formation of the senatorius ordo. Tac., Hist. 4.68.2 reports that Arrecinus Clemens, 'quamquam senatorii ordinis' ('despite being of senatorial rank'), was appointed praetorian prefect by Vespasian. AE 1947.40 = Suppl It 1.4 shows that he had previously been praetor and later became consul twice. On his career, see further Eck 1999. 41 Sen., Ben. 3.27.1, 'Rufus, uir ordinis senatorii' ('Rufus, a man of senatorial rank') and Tac., Ann. 15.49.4, 'Flauius Scaeuinus et Afranius Quintianus, uterque senatorii ordinis' ('Flavius Scaevinus and Aranius Quintianus, both of senatorial rank'). 42 Livy 35.6.5, 'ut duos legatos ex ordine senatorio mitteret ad L. Cornelium consulem' ('that he would send two envoys of senatorial rank to the consul Lucius Cornelius') and 43.2.3, 'ut in singulos a quibus Hispani pecunias repeterent quinos recuperatores ex ordine senatorio daret' ('that for each man from whom the Hispani demanded compensation, he would appoint ve judges of senatorial rank') both clearly refer to senatorial ofceholders; Suet., Aug. 35.1, 'decem ualentissimis senatorii ordinis amicis sellam suam circumstantibus' ('ten of his strongest friends of senatorial rank surrounding his chair'): the reference is to the period before the alleged formation of the 'senatorial order'; Val. Max. 5.2.6, 'Q. Terentius Culleo praetoria familia natus et inter paucos senatorii ordinis splendidus' ('Quintus Terentius Culleo from a praetorian family and as eminent as few other members of the senatorial order'): on any interpretation, 'inter paucos senatorii ordinis splendidus' must refer to a senior ofceholder; Suet., Tib. 23.1, 'inlatum deinde Augusti testamentum, non admissis signatoribus nisi senatorii ordinis, ceteris extra curiam signa agnoscentibus, recitauit per libertum' ('After he had Augustus' testament brought in, and only men of senatorial rank were let in to sign, the others acknowledging their seals outside the senate-house, he had it read out by a freedman'): admission to the senate-house would seem to exclude minors or women. The 'cuneos equestris ac senatorii ordinis' upon which Domitian according to Suet., Dom. 4.5 showered gifts clearly are the same groups as the 'senatui equitique' mentioned earlier in the same sentence. According to Suet., Calig. 18.3, Caligula organised a series of races in which 'all charioteers were of senatorial rank' ('nisi ex senatorio ordine aurigantibus'); pace Nicolet (n. 35 above), there is no evidence they included women or children. Pseudo-Cicero, Declamatio in L. Sergium Catilinam 111, 'interitum crudelissimum senatorii ordinis' ('the most cruel extinction of the senatorial order') refers to Catiline's supposed plan to exterminate all senators. Pseudo-Cicero, Declamatio in L. Sergium Catilinam 116, 'pro defensa consulum uita seruataque senatorii ordinis dignitate' ('for having defended the lives of the consuls and preserved the dignity of the senatorial order') denotes the Senate as an institution. ofceholders who brought Libo to trial, several of whom had earlier been mentioned by the historian. If Tacitus describes them as qui senatorii ordinis erant and not as senatores, this is not because they were only members of a senatorial class but not of the Senate. Rather, ordo is the better word because what the historian wishes to emphasise here is their 'senatorial rank', not their participation in the institution of the 'Senate': Tacitus contrasts them with other accusers who were of equestrian statusunlike senators, those men were rewarded for their services solely with nancial benets but did not receive public ofces. 43 Two passages feature a subtle change in the ways the word is used. Both come from Tacitus. In Book 13 of the Annals, the historian tells the story of one of Nero's drinking tours through Rome, during which he picked a st-ght with 'Iulius Montanus, of senatorial rank, but who had not yet held ofce'. 44 The terminology is a departure from previous practice. In the late Republic, membership of the Senate was obtained by holding the quaestorship. There could be no senator 'who had not yet held ofce'. The fact that Montanus could be counted a member of the ordo before he had obtained the quaestorship was the product of a reform introduced by Augustus. As noted in Section I.2 above, the rst emperor gave adult sons of senators who planned to embark on a political career the right to wear the traditional dress of senators, the toga with the broad purple stripe (latus clauus). Montanus was such a laticlauius, who was allowed to attend meetings of the Senate and served as military tribune in the imperial army, but had not yet held ofce and thus was not yet a full senator. 45 Signicantly, when Suetonius discusses the same incident, he calls Montanus quidam laticlauius. 46 Tacitus observes the same usage in a passage from the Histories in which he relates the story of Vespasian's general Cornelius Fuscus. Fuscus was born into a senatorial family but then abandoned an ofceholding career. 47 Tacitus describes his withdrawal from politics in a sartorial metaphor: 'out of a desire for a leisured life, Fuscus in his early youth took off (the symbols of) senatorial rank'. 48 It is possible that by the time he gave up his political ambitions, Fuscus had already secured election to the quaestorship. More likely, he was a laticlauius, who was expected to undertake a political career, but not yet a formal member of the Senate. This interpretation is supported by Tacitus' language: by his 'taking off' (exuerat) the symbols of senatorial status, the historian refers to his rejection of the toga with the broad purple stripe. As in the case of Montanus, so also in the passage about Fuscus, Tacitus seems to count laticlauii as men of 'senatorial rank'. For Chastagnol, the discussion of Montanus' rank in the Annals is a key piece of evidence for his view of the consolidation of a hereditary 'senatorial order' in the reign of Caligula. 49 But it is doubtful whether Tacitus in this episode (and in the parallel passage on Fuscus) intended the term ordo senatorius to be understood in a technical sense. As detailed above, in other texts laticlauii are classied as members of the equestrian order. More likely, this unparalleled usage was a deliberate stylistic choice. Tacitus usually avoids technical language, or employs it in idiosyncratic ways; in the words of Tony Woodman, 'the use of apparently technical language in a non-technical way is typically Tacitean'. 50  which Tacitus deliberately shunned in his historical works. By paradoxically depicting laticlauii as senators who had not held ofce, he avoided technical terminology inappropriate for the genre of historiography. 51 The two passages in the Annals in which laticlauii are depicted as men of senatorial rank are the only evidence that the Latin terminology of status changed in the early Empire. There is no instance in which the word ordo senatorius refers to more distant descendants of senators, or to their wives. This is remarkable. While the effects of Augustus' measure to permit a select group of elite males to participate in senatorial meetings can be traced, what should have been the most important aspect of his reformsthe formation of a hereditary senatorial classdoes not seem to have left any traces in extant literature. A second foundation of the theory of senatorial heredity turns out to be hollow.

III INSCRIBING SENATORIAL RANK
Early imperial law and literature thus show no traces of a hereditary 'senatorial order'. Are inscriptions more forthcoming? The time from Augustus until the Severan period is the 'golden age' of Latin epigraphy. Thousands of inscriptions discovered across the Mediterranean and western Europe have yielded detailed information about the social, economic, cultural and religious history of the Roman Empire. While the gaps in our evidence remain considerable, we are better informed about the early imperial elite than about most other pre-modern governing classes. 52 It is true that stone inscriptions were usually put up to honour successful ofceholdersthose who refused to engage in a political career are thus usually not epigraphically attested. As a result, extant texts rarely deal with those relatives of senators who withdrew from public life. 53 Still, some scattered evidence enables us to identify the formal status of politically inactive descendants of senators. Strikingly, whenever such persons are epigraphically attested, they are described not as members of the 'senatorial order', but as equestrians.
In the senatus consultum de Pisone patre, Gnaeus Piso, an adult son of the accused, who had accompanied his father to Syria, but who had not yet held a senatorial magistracy, is explicitly described as a person 'who was not yet member of the senatorial order' ('qui ordinis senatori nondum esset'). 54 A series of inscriptions from Sparta, whose prosopography has been carefully analysed by Anthony Spawforth, gives the names and achievements of the two sons of Tiberius Claudius Brasidas, one of only two Laconians who ever became senators in Rome. The texts date from the Severan period. They attest to the priesthoods of the imperial cult held by the descendants of the senator and to their intimate relations with Roman governors and high ofcials. Signicantly, however, in all these inscriptions, Brasidas' sons never carry the status designations of senators but indeterminate epithets such as 'most worthy' (ἀξιολογώτατος) or 'the all-rst' (παντάπρωτος). In line with the traditional interpretation, Spawforth nevertheless assumes that they were members of the ordo senatorius: 'both, as the sons of a senator, would have inherited senatorial rank'. But the consistent lack of senatorial titulatures in all inscriptions makes it hard to believe that this was in fact the case. 55 From the province of Asia, another three high priests of the imperial cult are known who were 51 sons of senators. They never carry senatorial status designations either. 56 It is also notable that as leaders of the provincial imperial cult, all these men must have invested substantial sums in municence. Yet they did not take advantage of the immunity from local liturgies allegedly enjoyed by descendants of senators (discussed in Section I.5 above).
This assessment that senatorial status was not hereditary is conrmed by the usage of the word 'senatorial order' in Latin inscriptions. All men epigraphically attested as members of the 'order' were active in politics, not mere relatives of senators. In most texts, promotion into the 'order' is explicitly associated with formal adlectio to the Senate, at the rank of a magistracy dened by the emperor. 57 However, there was some ambiguity surrounding the status of laticlauii. In four inscriptions, put up in the reigns of Trajan and Hadrian, the conferral by the emperor of the toga with the broad purple stripe is depicted as admission into the 'order'; only later did the persons in question achieve the entry magistracy for the Senate, the quaestorship. 58 This usage is reminiscent of Tacitus' description of the status of Montanus and Fuscus. 59 The fact that laticlauii were sometimes depicted as men of 'senatorial rank' reects the problems of classication posed by this in-between group, whose members wore senatorial dress and participated as spectators in senatorial meetings, but had no voting rights in the assembly. It may be no coincidence that both Tacitus' historical works and these inscriptions were produced in the early second century C.E. As will be explored in the next section, there are other signs that in this period the status of descendants of senators became an urgent question, in ways in which it had not been before. But for present purposes, another fact is more noteworthy. In all of imperial epigraphy, there is no single text in which a politically inactive descendant of a senator or his wife is described as member of the 'order'. Strikingly, the supposed extension of the senatorial class left no trace in inscriptions.
A third pillar of the theory of senatorial heredity has collapsed. The combined evidence from imperial law, literature and epigraphy makes it hard to sustain the conventional view that senators transmitted their rank to their children, grandchildren and great-grandchildren in the male line. But before we explore the implications of this nding, it is necessary to look at a fourth and nal body of evidence that is usually thought to support the traditional theory. clarissimus, their wives clarissima femina, their adult sons clarissimus iuuenis, infant sons clarissimus puer and infant daughters clarissima puella. 60 The Greek translations λαμπρότατος and λαμπροτάτη, employed without accompanying nouns, appear around the same time. 61 The spread of status designations that were employed by both senators and their relatives might seem proof that senatorial rank had now nally become hereditary. But yet again, a closer look at the evidence problematises this interpretation.

Clarissimus Vir
Let us begin with the status designation clarissimus uir. This title began to appear in the reign of Hadrian. Chastagnol suggests that it always denotes actual senators: 'Ce titre de clarissimus uir s'applique seulement aux sénateurs effectifs.' 62 By contrast, Hopkins and Burton argue (with some hesitation, but in keeping with their theory that senators transmitted their rank to politically inactive descendants) that some non-ofceholders also laid claim to this title: 'some of those recorded in tombstone inscriptions as having senatorial status … may not have been full members of the senate but rather politically inactive members of the senatorial order.' 63 Several thousand inscriptions containing the title uir clarissimus are currently known. A complete analysis of this vast corpus of texts is impossible. As a case study, I have collected all clarissimi uiri attested in Africa Proconsularis from the Antonine period until the accession of Diocletian. Sixty-four of seventy-two holders of this title can be shown to have exercised senatorial ofces. Another six served as curator, an imperially appointed municipal administrator, or as patronus, patron of a city. 64 As Graham Burton and François Jacques have shown, until the Tetrarchy these functions were almost always fullled by senators or senior equestrian ofceholders. 65 For the remaining two clarissimi uiri, we have no information on their public lives. 66 Of course, it is theoretically possible that they were politically inactive descendants of senators. But this is unlikely. If senators automatically transmitted their rank to their children, grandchildren and great-grandchildren, non-ofceholding clarissimi uiri would have far outnumbered actual senators. Yet no single such person can securely be identied. Moreover, as will be recalled from the previous section, when politically inactive descendants of senators are attested in other regions of the Empire, they never carry senatorial titles. 67 Chastagnol's view that the title uir clarissimus denoted full members of the Senate is surely correct.

Clarissimus Iuuenis
The same scholar maintains that the status designation clarissimus iuuenis referred to laticlauii, young men who fullled some of the preparatory functions for a senatorial career, either as uigintiuiri in Rome or tribuni laticlauii in the provinces. 68 Burton 1979 andJacques 1984. 66 Supplementary Appendix 1, nos 13 and 68 (son of a consul and father of a praefectus urbi). 67 See nn. 54-6 above. 68 Chastagnol 1992: 172. of all clarissimi iuuenes attested from the Antonine period until the accession of Diocletian conrms this interpretation. From all regions of the Roman Empire, thirty-six holders of this title are known. The number is interesting in itself. If every son, grandson and great-grandson of a senator had the right to call himself clarissimus iuuenis, it would be difcult to explain why only so few holders of this title are attested. But it ts well with the hypothesis that this status designation normally denoted persons who exercised one of the administrative posts or military functions customarily exercised by elite males before they formally joined the Senate. A closer look at the identity of known clarissimi iuuenes provides further corroboration. No fewer than fourteen of them are explicitly recorded as holders of one of the civilian or military ofces reserved for laticlauii. 69 Another ten exercised senatorial magistracies or priesthoods. 70 For the remaining ten clarissimi iuuenes, no ofces are explicitly attested. 71 But at least four of them served (alongside their fathers) as patrons of their local communities. 72 Another two were sons of equestrians; clearly, they had received the toga with the broad purple stripe because they planned to join the Senate. 73 All of this supports Chastagnol's view that the term clarissimus iuuenis denoted men in their high teens or early twenties in a preparatory stage for an ofceholding career.

Clarissimus Puer
The status designation clarissimus puer is more interesting. Some holders of this title were infant children; the youngest attested clarissimus puer was six months old. 74 Here at last, it might be thought, there is unambiguous proof that senatorial rank became hereditary in the early Empirehow else could it be explained that babies now held senatorial titles? But again, an in-depth examination of the evidence gives a more complex picture. Of fty-six known clarissimi pueri, only twenty-eight (50 per cent) were descendants of a senator, 75 fourteen (25 per cent) came from equestrian or municipal families, 76 and the family origins of the remaining fourteen (25 per cent) clarissimi pueri cannot be identied. 77 The fact that this title was only used so rarely, and often for boys who were not related to senators, makes it difcult to believe that it was understood as a hereditary status designation. More likely, it should be interpreted analogously to the status designation clarissimus iuuenis. It seems that clarissimi pueri were boys below the age of maturity, who had the right (once they reached the requisite age) to obtain the latus clauus. If this is correct, the title implied an expectation that the person in question would run for senatorial magistracies. Conversely, clarissimi pueri who were unable or unwilling to embark on a political career would need to abandon their title. That this is precisely what happened is suggested by the fact that (as mentioned above) no non-ofceholding clarissimi uiri are known.

Clarissimae feminae and puellae
The status designations for female relatives of senators emerge later than those for their male counterparts. The title clarissima femina appears for the rst time in two texts produced in the reign of Antoninus Pius. However, the stonecutters employed two unique abbreviations for it: CL F and CLAR FEM. 78 Only from the reigns of Marcus Aurelius and Commodus onwards is the status designation clarissima femina commonly used, and now uniformly abbreviated with the conventional CF. As Marie-Thérèse Raepsaet-Charlier observes, this suggests that it was only in this period that the title received some form of ofcial sanction. 79 Also around the same time, the Greek translation λαμπροτάτη is rst attested. 80 In some ways, the adoption by the wives of senators of the new status designations was a surprising development. After all, the Senate dened itself as an ofceholding elite, and in Rome, women (outside the eld of religion) were normally excluded from exercising any form of public power. Moreover, in the early Empire, married women normally remained under the patria potestas of their fathers; they did not become part of the agnatic family of their husbands. The fact that the adoption of senatorial titles by women had nothing natural or inevitable about it is shown by the terminology employed by rst-century and early second-century authors in talking about female relatives of senators. In this period, there was no word through which the status of the group of persons whom we call 'senatorial women' could be expressed; strikingly, collocations such as senatoria femina or senatoria puella are not attested in surviving texts. If they alluded to the rank of female relatives of senators, Latin authors employed circumlocutions such as mulier natalibus clara 'a woman of distinguished birth', splendida femina 'eminent woman' or inlustres feminae 'illustrious women'. 81 The lack of language through which the rank of female relatives of senators could be designated suggests that the Romans had no clear conception that the wives of senators were a group which was different from, or superior to, other female members of the propertied elites of the Empire.
But once wives and daughters of senators began to employ clarissimate titulatures, their distribution follows similar rules than titles for men. The key passage is an excerpt from Ulpian's treatise De deicommissis, included in the Digest: Women who marry clarissimi are included in the expression clarissimae personae. The term does not encompass the daughters of senators, except if they marry uiri clarissimi. 82 In Ulpian's view, daughters of senators were only allowed to call themselves clarissima femina if their husbands were senators too; if they married below their rank, they would lose their title. As Raepsaet-Charlier has shown, usage in surviving texts generally follows this principle. Most clarissimae feminae whose relatives can be identied were wives of senators. 83 Conversely, daughters of senators who married below their rank were classied as equestrians. For example, during the Secular Games of 204, two daughters of senators who had married equestrians participated in the spectacles amongst the matronae equestres. 84 There are exceptions. In several inscriptions, mostly put up on private estates of large landowners and dated to the reign of Commodus and to the second quarter of the third century, wives of senior equestrian ofceholders bear 78 The rst text is an unpublished inscription reported by Raepsaet-Charlier 1987: 104 in which Aquilia Blaesilla is honoured for the construction of a building in Lepcis Magna, the second, CIL 6.17543, a funerary inscription set up by some freedmen of Sallustia Frontina. 79 Raepsaet-Charlier 1981: 194-8. 80 Arjava 1991: 26-9 collects the evidence on the term λαμπροτάτη; 29-31 on the rarer συγκλητική (which appears only in Asia Minor). 81 Plin., Ep. 8.18.8-10; Suet., Otho 1; Tac., Ann. 3.49 with Chastagnol 1992: 173. 82 Dig. 1.9.8. 83 Raepsaet-Charlier 1981: 200-12. 84 Iulia Soemia (PIR 2 I 704) and Aelia Gemellina (PIR 2 A 1058) in EAOR I 43 with Raepsaet-Charlier 1983: 185-8. See also Vedia Phaedrina (PIR 2 V 327) and Claudia Crateia Veriana (Raepsaet-Charlier 1987: no. 230) in IEph 980 with Raepsaet-Charlier 1983: 189-91.

T H E H E R E D I T Y O F S E N A T O R I A L S T A T U S I N T H E P R I N C I P A T E
the title clarissima femina. 85 Chastagnol explains this pattern by arguing that the rule reported by Ulpian had only been introduced by Caracalla and was later abolished; on his reading, both before and afterwards, daughters of senators who married equestrians continued to enjoy clarissimate status. 86 By contrast, Raepsaet-Charlier suggests that clarissimae feminae with non-senatorial husbands received special dispensations from the emperor. 87 A third possibility is that Ulpian's rule was not consistently enforced. This would explain why most inscriptions that attest clarissimae feminae without equestrian husbands come from private contexts.
In any case, even though female designations appear slightly later than their male counterparts, their distribution yet again highlights that clarissimate status was inextricably intertwined with ofceholding. Like sons of senators who did not hold high ofce, so also daughters whose husbands did not engage in politics normally lost the title of their father.

Summary
The emergence of clarissimate titulatures was a signicant development. The fact that there was now a set of titles which described not only senators themselves, but also their female relatives and underage males who had received the right to run for senatorial magistracies, marks a departure from previous practice. This change may be linked to other contemporaneous transformations in epigraphic protocol. As seen in the previous two sections, from the early second century C.E. onwards, commissioners of inscriptions and some elite authors began to experiment with new ways of describing the status of underage males who had received the right to run for high ofce, but who were not yet members of the Senate. 88 Why did the line of demarcation between laticlauii and other members of local elites become an urgent issue from the Antonine period onwards? One explanation may be that the incorporation of growing numbers of provincial landowners into the Roman ruling class increased competition for the latus clauus. As laticlauii now frequently grew up in provincial cities and lived in the same milieu as other wealthy local landowners, it became more important to differentiate between persons who held ofce and fullled liturgies in their own municipalities and those who spent their political careers and invested their wealth in the imperial capital. If the new titles were responses to competitive tensions at the borderline between imperial and local elites, this would explain why this this new ofcial terminology of status only began to be widely employed in the Antonine period and why the earliest texts containing clarissimate titulatures were commissioned not in Rome, but in the provinces or Italian cities. 89 But irrespective of how the rise of clarissimate titulatures is best explained, what matters for current purposes is that these new titles were not hereditary. All known clarissimi uiri Arminius Donatus, amen perpetuus (from a rst marriage). Her son Arminius Donatus is recorded as CP. 86 Chastagnol 1979;: 173-99. 87 Raepsaet-Charlier 19811983: 191 with n. 39. 88 See nn. 44-51 and 57 above. 89 See also nn. 127-9 below. are senators, and only persons who were preparing for a political career were described as clarissimi iuuenes or clarissimi pueri. Similarly, woman normally were only called clarissimae feminae or clarissimae puellae if their fathers or husbands served as ofceholders. There is no evidence for the large number of non-ofceholding persons of clarissimate status that would have existed if membership in the 'order' extended not only to ofceholders themselves, but also to three generations of agnatic descendants. Like the legal, literary and epigraphic evidence, so also the distribution of clarissimate status designations does not t the traditional view of senatorial heredity. The theory of a hereditary ordo senatorius, which encompassed three generations of agnatic descendants of senators, can thus safely be laid to rest. It is now time to explore the broader signicance of this nding.
V PATRIA POTESTAS AND THE LEGAL CONSTITUTION OF THE SENATE From a certain perspective, it may seem surprising that the regulations reviewed in Section I of this paper had no more visible effects. After all, Mommsen was right that Augustus and his successors conferred special rights and duties on descendants of ofceholders. On the face of it, it seems plausible to expect that these measures transformed the children, grandchildren and great-grandchildren of senators into a new legal class. But as the previous sections have shown, there are hardly any traces of this class in early imperial literature and epigraphy. It is not merely that the term ordo senatorius is never employed to describe relatives of senators. With the exception of Cassius Dio, Tacitus and some inscriptions that record the status of sons of ofceholders who served as laticlauii and participated in various rituals in Rome, the new legal class supposedly created by Augustus' reforms does not appear in surviving texts at all. In particular, we hear nothing about the many politically inactive children, grandchildren and great-grandchildren of senators who lived in Italian or provincial cities and who now presumably were subject to a different set of laws than other local landowners. For example, the Album of Canusium, produced in 223 C.E., proudly lists the names of the senators and senior equestrian ofceholders who served as patrons of this Apulian town. It also carefully catalogues all members of the local city council and praetextati (future city councillors, the municipal equivalent to laticlauii). But locally resident relatives of senators are not mentioned in the document. 90 There is an apparent tension between the legal evidence, which suggests that the status of descendants of senators changed in the early Empire, and other sources, which are almost completely silent on the impact of the new laws.
One regulation poses the problem with particular urgency. It will be remembered that, according to the Pauli Sententiae, senators and three generations of their descendants were immune from fullling nancial obligations in their hometowns. 91 On the face of it, it would seem that this rule removed thousands of wealthy landowners from the reach of city councils. But there is precious little evidence of the impact of this regulation. In early imperial legal literature, there is no other examination of the scal status of descendants of senators, despite the fact that three long chapters in the Codex Iustinianus, and an entire book of the Digest, discuss in detail the conditions under which different groups were exempted from local municence. 92 Literary texts are no more forthcoming. For example, Pliny's native region of northern Italy was home to a 90 CIL 9.338 = Dessau, ILS 6121 = ERCanosa 1.35 (with useful commentary and prosopography). 91 Section I.5 above. 92 See esp. Dig. 50 and Cod. Iust. 10.32, 10.39 and 10.42. disproportionate number of senators. 93 So was the province of Bithynia and Pontus which he governed towards the end of his life. 94 But in none of his letters does Pliny touch on conicts arising from attempts by municipalities to force descendants of senators to pay for local obligations. The same silence in epigraphy. Not a single inscription survives in which a senatorial relative advertised immunity from local liturgies, or that he or she had generously outed such an immunity out of civic patriotism. All of this presents a riddle. As Fergus Millar noted long ago, the silence in legal, literary and epigraphic texts makes it hard to believe that children, grandchildren and great-grandchildren of senators were exempt from local nancial obligations. 95 But how then to make sense of the excerpt of the Pauli Sententiae, which unambiguously conrms this immunity?
In order to solve this puzzle, it is useful to look more closely at the ways in which the group of persons affected by Augustus' reforms is described in the texts discussed in section I above: Only the Lex Julia and the immunity from liturgies explicitly affected three generations of agnatic descendants (though Dio's vague statement that the exclusion from spectacles affected senators, their children and other ἔγγονοι may imply the same). By contrast, the reform of the latus clauus was aimed at the 'children' of senators. Finally, the Lex Claudia targeted all persons who stood under their patria potestas. These differences are surprising. At rst sight, it might seem as if each of those regulations affected a different group of relatives. But this conclusion would be premature. In order to understand the terminology employed in these texts, they need to be situated in the context of the Roman institution of patria potestas. 96 As Yan Thomas has shown, until deep into the imperial period, Roman fathers normally exercised full property rights over the possessions of all their descendants in the male line. A pater familias could dispose of the belongings of these lii familias as he wished, even if they were adults, and even if they were senior ofceholders of the Roman state. Emancipation was exceptionally rare; it normally happened only in cases where the pater familias remarried, and his rst wife wished to prevent the children's share of her inheritance falling into his hands after her death. 97 If the legal structure of the Roman family is correctly understood, it becomes clear that all the regulations quoted above specically targeted the persons who stood under the patria potestas of senators. This is most obvious in the case of the Lex Claudia. Livy explicitly writes that this law applied to 'senators and whoever had a senator as father'. 98 The same is also true for the reform of the latus clauus. Suetonius calls the group of persons who had the right to receive the toga with the broad purple stripe liberi senatorum. 99 This expression is 93 Alföldy 1982. 94 Halfmann 1982: 637-9. 95 Millar 1983: 87-91, esp. 88-9. 96 Kaser 1975 surveys the functioning patria potestas in the High Empire. Saller 1994: 102-32 looks at the social and demographic pressures which mitigated the uitae necisque potestas of a pater familias in practice. Arjava 1998 traces the survival of patria potestas into Late Antiquity. 97 Thomas 1982 explores the effects on senior ofceholders of the Roman state of the subjection to the patriarchal authority of their fathers. 98 See n. 28 above. 99 See n. 17 above. usually translated as 'children of senators'. But as Ulpian points out, the term has a broader meaning: By the term liberi senatorum we mean not merely the sons of a senator, but all whom he is claimed to have begotten, or his children are claimed to have begotten, irrespective of whether they are the natural or adopted children of the senators from whom they are claimed to be born. But if somebody is born from a daughter of a senator, we need to look at the status of the father. 100 Liberi senatorum were not merely the 'children' of a senator, but all agnatic descendants who normally stood under his patria potestas. The same is true for the Lex Julia and the immunity from liturgies. When a Roman law affected the children, grandchildren and great-grandchildren of a person, it was another way of saying that it affected all his potential lii familias: a great-grandson was the most distant possible descendant who could theoretically fall under his patria potestas. 101 Signicantly, jurists call the group of persons affected by the marriage legislation liberi senatorum. 102 As with the other regulations, so also the Lex Julia and the immunity from liturgies only affected those relatives of a senators who fell under his patria potestas during his lifetime. 103 If Augustus' laws specically targeted the lii familias of senators, this solves the problem posed by the excerpt of the Sententiae Pauli. The immunity from liturgies only extended to those descendants of senators who stood under their patria potestas. As long as a senator was alive, assets controlled by all his descendants in the male line normally were his legal property. By forcing one of these lii familias to pay for public services, a municipality would have been able to extract wealth from a senator against his will. It is not surprising that in the view of the author of the Pauli Sententiae, this was illegal. But after a senator's death this immunity ceased to be valid. This explains why there is no other evidence for conicts around the scal status of descendants of senators. Unless they embarked on an ofceholding career in Rome, they would again be liable for liturgies in their hometowns.
If this interpretation is correct, this also explains why we hear so little about the group of persons affected by Augustus' reforms in early imperial texts. The new regulations affected far fewer people than is usually thought. Richard Saller's demographic simulations give a rough idea of who normally fell under the patria potestas of a Roman head of household. According to his 'senatorial' lifetable, only one-quarter of Roman elite males (27 per cent) had a paternal grandfather at birth; at age twenty-ve, this ratio drops to one in a hundred (1 per cent). Signicantly too, by this time, only around two-fths of males (39 per cent) still had a father. 104 These gures suggest that Augustus' regulations normally affected the children of senators, not their grandchildren or great-grandchildren. Moreover, most of the persons affected by the legislation were still underage. Interestingly, the distribution of clarissimate status designation is in line with Saller's estimates. Of the forty-six clarissimi iuuenes and clarissimi pueri who have known senatorial ancestors, 100 Dig. 1.9.10. 101 For the general principle, see Dig. 50.16.220.pr. For applications of it in practice, see, for example, Cod. Iust. 4.28.6.1;9.41.11;Dig. 1.6.4;1.6.5;1.7.2.pr;1.7.28;28.2.25.1;28.2.29;37.8.1 and 38.16.1.7. 102 See Celsus in Dig. 23.2.23, 'Lege Papia cauetur omnibus ingenuis praeter senatores eorumque liberos libertinam uxorem habere licere' ('The Lex Papia stipulates that all freeborn persons except senators and their children are allowed to have a freedwoman as wife'), and Ulpian cited in n. 100 above. 103 As usual in Roman law, this included unborn children in their mothers' wombs, but not persons born more ten months after a senator's death or loss of rank: see Thomas 2007 on the general principle and Dig. 1.9.7.1 on its application to descendants of senators. The status of emancipated children of senators was controversial. In the view of Paul in Dig. 1.9.6.pr., they were not affected by the legislation. By contrast, Ulpian in Dig. 1.9.7.1.pr. argues they should be treated 'as if' they were the lii familias of senators. 104 Saller 1994: 52. forty-three (93 per cent) were sons, three (7 per cent) grandsons and nobody the great-grandson of a senator. 105 In sum, the Roman institution of patria potestas claries the purpose of Augustus' reforms. The new regulations did not create a hereditary aristocracy, which extended multiple generations after a senator's death, but merely conferred some new rights and duties on those persons who stood under his patriarchal authority during his lifetime. This explains why we only hear about them in the context of imperial rituals conducted in Rome, or when they exercised one of the administrative or military positions reserved for laticlauii. In practice, most of these liberi senatorum were children or young adults who lived in the same household as their parents.

VI AMBITION AND WITHDRAWAL
If it is recognised that Augustus' regulations specically targeted the lii familias of senators, it then becomes clear that they had much earlier precedents. The original version of the Lex Claudia was passed in 218 B.C.E. 106 Similarly, when in the late second century B.C.E. the lex repetundarum created a new court that allowed provincials to bring suits against Roman magistrates, its jurisdiction extended not only to senators, but also to all persons 'whose father was a senator'. 107 The municipal law of Urso (modern Osuna near Seville) from 44 B.C.E. included a provision that 'senators of the Roman people' and 'sons of senators of the Roman people' had the right to sit on designated seats in spectacles. 108 At the same time, it is true that emperors intervened more aggressively into the lives of the lii familias of ofceholders than republican lawmakers. Not only had descendants of senators never been subjected to so many new regulations in such a short time; Augustus' reforms were also more ambitious in scope than earlier republican laws. This raises a new question. What motivated this increased interest of the imperial government in the descendants of senators?
On the one hand, the new regulations pursued ideological objectives. By allowing young elite males to wear the latus clauus and by prohibiting them from performing in spectacles or marrying into servile families, Augustus advertised that he was committed to protecting their freedom and moral integrity. Future ofceholders would be recruited from the same social class as in the Republic and their honour unsullied through contact with servile populations or occupations. These measures formed part of a broader effort by the Augustan regime to involve the children of ofceholders in public life. It has already been observed that laticlauii had a prominent role in imperial funerals. 109 Sons of senators also appear several dozen times in the records of the elite priesthood of the fratres aruales: 'freeborn boys, sons of senators whose fathers and mothers are still alive, taking the offering to the altar in libation dishes'. 110 Even more important was their participation in the transuectio equitum, the annual parade of the equestrian order on the Ides of July. 111 Each year, six laticlauii were handpicked to serve as seuiri equitum Romanorum, leaders of the faux military units (turmae) into which the ordo equester was divided. 112 Some funerary reliefs represent this ritual, showing the deceased participating in the procession on horseback, symbol of his equestrian rank, but wearing the shoes (calcei) of senators, representing the promise of a future promotion into the Senate. 113 In sum, by regulating the lives of future members of the imperial elite, and by involving them prominently in state ceremonial, Augustus staked out a claim that he wished to foster traditional forms of virtue and political commitment amongst them.
On the other hand, the new regulations had material effects. They contributed to keeping the wealth of senators within their families. The immunity from local liturgies prevented the alienation of assets from which ofceholders would nance their political careers. Similarly, as Andrew Wallace-Hadrill pointed out, by preventing descendants of slaves from becoming the heirs of senators, the Lex Julia encouraged the intergenerational transmission of wealth within elite families: 'Augustus aimed to stabilise the family in order to stabilise the transmission of property, and consequently of status, from generation to generation.' 114 It is also interesting that Hadrian expanded the scope of the Lex Claudia to include the capital intensive business of tax-farming. 115 By outlawing this form of investment, he reduced the risk that a sudden default of provincial debtors would prevent Roman magistrates from fullling their nancial obligations. Taken together, the diverse set of regulations introduced by Augustus and his successors may thus be said to have pursued a unied purpose. By safeguarding the honour and wealth of lii familias of senators, the imperial government made it easier for them to follow their fathers' careers.
Such efforts to motivate descendants of ofceholders to participate in politics were particularly urgent in the 20s and early 10s B.C.E. when most of these measures were passed. In the civil wars, many senators been killed and hundreds of 'new men' had been admitted to the imperial elite. 116 But even after the return to peace, ofceholding families remained reluctant to join the Senate. Cassius Dio notes that several times in these years Augustus faced a shortage of qualied candidates for magistracies; as Amy Russell points out, the concentration of power at the imperial court and the decline in the power of ofceholders led to 'a crisis of senatorial recruitment'. 117 The policies pursued by Augustus and his advisers in this period were designed to reverse these trends. In 29, 18 and 13 B.C.E., three reviews of the Senate were carried out, during which 'unworthy' members were expelled and the overall size of the assembly reduced from more than 1,000 to around 600. 118 Also in 18 B.C.E. or soon afterwards, Augustus increased the minimum census for senators from 400,000 to 1,000,000 HS. As Claude Nicolet and Andrew Pettinger note, these reforms were not intended to weaken ancient ofceholding families; notably, for those who could not full the new wealth requirement, the emperor made up the difference from his own private fortune. On the contrary, by reducing the number of senators and raising the census, Augustus emphasised the greater dignity (dignitas) which elevated the ruling elite of the Empire T H E H E R E D I T Y O F S E N A T O R I A L S T A T U S I N T H E P R I N C I P A T E over other citizens. 119 The regulations on lii familias of senators need to be situated in the same context. By incentivising them to participate in politics and exhibiting them to the public in state rituals, Augustus advertised his determination to restore social distinctions of an imagined past and to moderate social mobility; as John Richardson puts it, 'the intention was clearly that membership in the senate would pass down the generations of a senatorial family'. 120 This goal did not lose urgency under Augustus' successors. As Hopkins and Burton have shown, social mobility in the early imperial Senate was higher than it had been in the late Republic. 121 Powerful structural forces worked against any attempts by Augustus and his successors to slow down the rate of renewal within the imperial elite. First, while the ction that the emperor was the rst magistrate of a restored Republic brought many symbolic and material benets to senators, it also exposed them to great risks. The persecutions and conspiracies which marked the history of early imperial Rome are indicative of the violence generated by a political system in which any successful senator was a potential emperor. The high incidence of violent death amongst early imperial senators disrupted continuity within the Senate. 122 Second, Roman political culture had always associated monarchy with slavery. It is true that early imperial ofceholders went out of their way to challenge this preconception. Shadi Bartsch and Matthew Roller have exposed the rhetorical techniques and ideologies developed by rstand early second-century authors to reconcile service to the emperor with the preservation of their inner freedom. 123 Rose MacLean has observed that senators now also discovered the old servile quality of obedience as a positive virtue. 124 But not all were convinced that the tensions between liberty and monarchy could be quite so easily resolved. Behind reassertions of the value of state service as the only appropriate arena in which elite males could prove their worth lurked widespread anxieties that the experience of civil war and autocracy had fatally undermined the freedom and moral integrity of the Senate. 125 In the view of many Roman men, a retreat from politics (otium) was preferable over membership in a slavish and morally corrupted ruling class. 126 Third, in the rst two centuries C.E., ever larger numbers of provincial property-owners obtained Roman citizenship and became eligible to join the Senate. 127 Already in the reign of Antoninus Pius, most consuls (the best attested group of senators) came from outside Italy. 128 The gradual incorporation of provincials into the imperial elite intensied competition. While the number of candidates eligible for membership multiplied, the overall size of the Senate remained steady. I have suggested above that the intensication of rivalry explains the quick spread of clarissimate titulatures; the new terminology provided a language to describe the boundary between those members of local elites who had received the right to run for senatorial magistracies and those who spent their 119 Nicolet 1984b: 90-6 (census) and Pettinger 2019, especially 60-2 (reduction in numbers). 120 Richardson 2012: 118. 121 Hopkins and Burton 1985. 122 Winterling 1999 and 2011 are the most sophisticated recent explorations of the paradoxical political system introduced by Augustus. Suggestive of the scale of violence in some periods is the remark in Suet., Claud. 29.2 that in Claudius' reign alone thirty-ve senators were killed. As Hahn and Leunissen 1990: 60-1 note, this is equivalent to more than one-tenth of the senators appointed during this period. 123 Bartsch 1994;Roller 2001. 124  political careers in their municipalities. 129 In any case, like the high incidence of violence and political withdrawal, so also the accelerated integration of provincial landowners into the imperial elite ensured that the Senate renewed itself more quickly in the early Empire than it had in the late Republic.
This increased turnover provides a context for the new interest shown by Augustus and his successors in the lives of the descendants of senators. In the face of widespread anxiety that the introduction of monarchy would lead many of the men who were best qualied for leadership positions in the Roman state to withdraw from politics, and enable unworthy candidates to reach high ofce, emperors sought to incentivise as many sons of senators as possible to follow their fathers' careers. In this sense, Mommsen and other modern scholars were entirely correct if they discerned in early imperial legislation a concern to facilitate the intergenerational transmission of status. The regulations were designed to compel sons of senators to embark on an ofce holding career; as Nicolet puts it, 'becoming a senator would in principle become an obligation for them.' 130 Where scholars such as Mommsen and Nicolet went wrong was in thinking that the legislation simply enshrined into law an underlying reality in which sons of senators almost always followed the careers of their fathers. The opposite was the case. Precisely because the Senate had become more unstable as a social group, emperors sought to slow down the pace of turnover within it.

VII IMPERIAL STATE AND SENATORIAL WEALTH
One nal question remains. Why were Augustus and his successors so keen to encourage heirs of leading families to participate in politics? After all, from a certain perspective, it was in the interest of monarchs to reduce the power of their highest-born subjects. In the view of Hopkins and Burton, Augustus and his successors deliberately sought to weaken the power of old senatorial families in order to shore up their own authority: 'aristocrats were those whose social power most threatened the emperors' supremacy and survival'. 131 Why did emperors nevertheless seek to convince the sons of ofceholders to follow their fathers' careers? Ideological reasons certainly played a role. In the view of Roman authors, the children of senatorial ofceholders, prominent equestrians and other wealthy citizens were more likely to exhibit the exceptional forms of commitment to the Roman state expected from holders of senior magistrates than men of low birth or servile descent. Cassius Dio expressed these preconceptions in his speech for Maecenas: 'for it is not possible either for a man of any real worth to be naturally lacking in political dedication (φρόνημα), or on the other hand for a man sprung from a servile sphere of life to acquire such dedication'. 132 But ideology was not the only reason why Augustus and his successors strove to stabilise the rate of turnover within the elite. There were also good economic motives to encourage the descendants of senators and other large landowners of free birth to participate in politics. In comparison to other ancient states, Rome had an extremely wealthy elite, 133 129 See Section IV.5 above. 130 Nicolet 1976: 35, 'devenir sénateur serait une obligation de principe pour eux'. See also 38, 'l'obligation, au moins morale, faite aux ls et descendants de sénateurs de briguer à leur tour les honneurs' ('the obligation, at least a moral one, imposed upon sons and descendants of senators to compete for ofces'). Also Mommsen 1887: 469 assumes that they 'normally' ('regelmässig') followed their fathers' careers. 131 Hopkins and Burton 1985: 121. 132 Cass. Dio 52.8.5. 133 Scheidel 2017: 62-85 notes that levels of wealth accumulation in Rome and China exceeded those of previous states and would not be reached again until the nineteenth century. Lewis 2007: 115 observes that even the estates of the great landowners of Han China were much smaller than those of their Roman counterparts. but a very weak state infrastructure. 134 As a result, many key public functions were nanced from the private fortunes of leading citizens. In particular, ambitious politicians invested vast sums in providing spectacles and other forms of largesse for the Roman people. 135 This reliance on private municence posed two problems for emperors. On the one hand, it was not in their interest that ofceholders would provide outsize investments in popular entertainments. If another senator showed too much generosity towards the populace, this might have been taken as a threat to the emperor's role as supreme patron of the populus Romanus. In order to reduce this risk, Augustus and his successors asserted a monopoly on the most impressive forms of public self-display, such as triumphs and public building, and introduced expenditure limits on large-scale spectacles. 136 On the other hand, if they wished to maintain the level of provision to which the Roman people had become accustomed in the Republic, monarchs could not completely forgo the wealth of the imperial elite. Yet, since the most spectacular prizes for political achievement were now reserved for the emperor and his relatives, senators no longer had the same incentives to invest their money in public functions as they had in the Republic. Augustus and his successors solved this problem by making it obligatory for ofceholders to invest some of their private money into select public functions. For example, quaestors had to make large cash payments into a fund for the construction of roads, praetors were asked to organise a series of expensive spectacles for the Roman populace and the curator aquarum was expected to invest some of his own money into the improvement of the water supply. 137 This system allowed emperors to harness the wealth of senators without risking serious challenges to their own position. But the vital role played by senators in the maintenance of civic infrastructure had an important corollary. It provided a powerful incentive for emperors to ensure that as many descendants of senators as possible would participate in politics. For if heirs of ofceholders withdrew from political life, the imperial state would no longer be able to draw on their wealth to nance municence in Rome. There was thus not only a powerful ideological reason, but also a strong economic motive, for emperors to encourage sons of senators to participate in politics.
In sum, by making it easier for the lii familias of ofceholders to attain the same status as their ancestors, Augustus and his successors not only sought to full the ideological expectation that a good ruler would share his power with the descendants of the most respectable families in the Empire, but also advanced their own material interests. They ensured that there would always be a sufcient number of wealthy landowners who were able and willing to serve as magistrates, and invest their private fortunes in municence in Rome.