¹ Pollio is mentioned at Eclogues 3. 84, and he also receives the Fourth Eclogue. I accept the argument of G. W. Bowersock, HSCP 75 (1971), 73–80, that the dedicatee of the Eighth Eclogue is not Pollio but Octavian.

² Williams, Gordon has expressed misgivings, however, in Tradition and Originality in Roman Poetry (1968), 45Google Scholar. I must admit to careless use of the words myself, in JRS 64 (1974), 40–61 and HSCP 79 (1975) 265–300.

³ Here let me remark that the literary relationships which I will describe in this paper have many striking resemblances to those treated by Beljame, A., Men of Letters and the English Public in the Eighteenth Century, 1660–1744, tr. Lorimer, E. O. (1948)Google Scholar.

⁴ Compare the more vivid but similar catalogue of dependants given by Marquardt, J., Privatleben², 205–6Google Scholar.

⁵ I say ‘dependants’ here rather than ‘clients’ in the hope that the former will be accepted as a looser term, yet even ‘dependants’ may promote a false impression. We tend to think and write as though Roman society were neatly divided into a client population and an upper class of patrons. But the reality was more confused. In theory and to some extent even in practice, courtesies and officia were performed as between one friend and another, and they were mutual. The ‘client’ offered gifts to the ‘patron’, but the ‘patron’ also made presents to the ‘client’; dinners were sometimes exchanged; and ‘client’ and ‘patron’ recited poetry and applauded it by turns. The activities mentioned in these pages, therefore, should not be seen as a burden which fell exclusively on the lesser partner in a friendship. In the second place, although clientes certainly existed in the entourage of a great man, the word was not applied to most of those whom we are in the habit of calling ‘clients’. Clientes are the almost faceless numbers in the outer reaches of a man's acquaintance; persons known on more intimate terms are amici or sodales. Moreover, clientes and sodales do not constitute separate groups in the sense that one consists of dependants, the other of equals. In the setting of a great man's house, all, or almost all, the visitants will be in some degree beholden to the great man's favour. What determines whether one is another man's dependant or his better (there is hardly any middle ground) is not class simply, but the precise weight of his dignitas (age, family back-ground, wealth, honour) in comparison with one's own.

⁶ See Schanz-Hosius, II⁴ (1935), 489–91. The dating rests entirely on the assumption that the Calpurnius Piso celebrated in the poem is the conspirator of the year 65.

⁷ See Schanz-Hosius, II⁴, 188–9. The date of the poem is fixed approximately by a reference in line 12 to Messalla's consulship, which he held in 31 B.C., and by the absence of any reference to his later military feats, including particularly his triumph in 27.

⁸ Lines 35–6 (‘convenientque tuas cupidi componere laudes/undique quique canent vincto pede quique soluto’) indicate that the author expects other writers to be trying the same approach.

⁹ The suffix reveals the word as a French coinage: -age is the reflex of -aticus, a suffix which served in Latin to form adjectives, from which in turn a small number of substantive forms developed, like viaticum. But in early French, on the analogy of viaticum/voyage, the suffix -age had an enormous development in the forming of substantives. See Meyer-Lübke, W., Grammatik der romanischen Sprachen 11 (1894), 521–3Google Scholar. The word patronagium, cited once from a late (1448) text in the Du Cange lexicon, was evidently re-imported into Latin from French; it is unattested during the early Middle Ages. For this information I am indebted to the kind offices of Dr. Theresia Payr, who assures me that the word has not turned up in the materials for the Mittellateinisches Wörterbuch, and of Mme. Anne-Marie Bautier, who has similarly verified its absence from the Novum Glossarium.

¹⁰ The conception of a ‘patron’ saint is not entirely without analogy in Roman paganism: compare Plautus, , Rudens 261–2Google Scholar, ‘bonam atque opsequentem deam atque hau gravatam/patronam exsequontur benignamque multum', and Catullus’ invocation to his Muse (1. 9), ‘o patrona virgo’.

¹¹ I am grateful to Professor Carl Hammer of Carnegie-Mellon University for timely direction over this unfamiliar ground, and am indebted for instruction to Feine, H. E. A., Kirchliche Rechtsgeschichte: Die katholische Kirche⁶ (1972), 261–2, 397–8Google Scholar; The New Schaff-Herzog Encyclopedia of Religious Knowledge VIII (1910), 389–92, s.v. ‘patronage’; Godfrey, J. A., The Right of Patronage according to the Code of Canon Law (1924), 11–18 and 38–47Google Scholar; and Wood, S., English Monasteries and their Patrons in the XIIIth Century (1955)Google Scholar.

¹² Marquardt had perceived this important restriction on usage: ‘Am Ende der Republik hat der eigentliche Stand der Clienten insoweit aufgehört, als er nur noch durch die Freigelassenen repräsentiert und der Name patronus, der sich ursprünglich auf die Clientel bezog, ausschliesslich von dem Freilasser gebraucht wird’, Privatleben, 204. The only instance I know of in which the words patronus and cliens occur together outside the narrowly restricted contexts I have named is found in Horace, Epist. 1. 7. There Volteius Mena is termed the cliens of Philippus (l. 25), and is made to address Philippus with the words o patrone (l. 92). On the uniqueness of the terminology in this situation, see Neuhauser, W., Patronus und Orator: Eine Geschichte der Begriffe von ihren Anfängen bis in die augusteische Zeit, Commentationes Aenipontanae 14 (1958), 112–15Google Scholar. It is not unimportant that Mena is in fact a freedman (cf. l. 54), though not Philippus' freedman: he may be slipping into the deferential mode of address he practised towards his former master. Finally, I know of two cases in which cliens and patronus are juxtaposed in situations not clear enough for one to be able to establish a context, ILS 6577 and Pliny, NH 34. 17—though my guess is that both are concerned with the patronage of towns.

¹⁸ For the use of clientela and cliens during the Republic, see Hellegouarc'h, J., Le vocabulaire latin des relations et des partis politiques sous la République (1968), 54–6Google Scholar. Even during the Republic, use of the word clientela appears to be avoided in social (as opposed to political) contexts.

¹⁴ Although Hellegouarc'h (op. cit., n. 13) recognizes that Latin usage frequently glides over distinctions of status, this recognition does not deter him from founding his discussion of amicitia upon a distinction between the two parties (p. 41). Nevertheless, since his treatment deals mainly with the period of the late Republic, it makes a useful companion study to this paper, which concentrates on the following period. Certain differences in our respective purposes should be noted, however. Hellegouarc'h takes great pains to distinguish between terms used for friends who are equals and terms used for friends who are not. Whereas I do not think that this distinction is very well reflected in the vocabulary itself. Secondly, Hellegouarc'h tries to elicit nuances which differentiate the various words which make up the vocabulary of amicitia, seeking, for example, what distinguishes amare from diligere, or necessitudo from familiaritas, and he tries to present a more or less complete catalogue of the different words employed. From my point of view, the terminology is made up of several clusters of indifferent synonyms, and I am concerned to describe only what is common and ordinary language, not to provide a comprehensive lexicon.

¹⁵ Information about these persons is most handily obtained from the ‘General List of Contemporary Persons’, 738–62 of Sherwin-White, A. N., The Letters of Pliny: A Historical and Social Commentary (1966)Google Scholar.

¹⁶ The bilateral value of the notion expressed as amicitia is also evident in remarks like ‘amatur a me plurimum nec tamen vincitur’ (Pliny, Epist. 2. 13. 8) and ‘hunc hominem adpetentissimum tui… sic ama tamquam gratiam referas. neque enirn obligandus sed remunerandus est in amoris officio, qui prior coepit’ (Epist. 7. 31. 7).

¹⁷ cf. Hellegouarc'h, op. cit. (n. 13), 63, ‘le mot amicitia lui-même est celui qui désigne d e la façon la plus générale les rapports favorables établis entre deux hommes ou deux groupes politiques’; and 142, ‘amare est évidemment [le verbe] qui exprime l'amicitia sous sa forme la plus générale’.

¹⁸ cf. Hellegouarc'h, op. cit., 147–9 and 207.

¹⁹ cf. Hellegouarc'h, op. cit., 68–71.

²⁰ Comes and convictor/convictus are often found in the company of these other words, but they are not precisely synonyms. Comes usually denotes, not a relationship, but an activity or function, referring to someone who happens actually to be accompanying the rich man on his way. Cf. Hellegouarc'h, op. cit., 59–60: ‘il désigne la “cour’ d'un personnage de haut rang, particulierement lorsqu' ils lui font une escorte d'honneur au cours de ses déplacements dans la ville.’ Convictor and convictus, although they can refer to the actus convivendi, tend in ordinary parlance to evoke the narrower context of the convivium, and to describe dinner parties and persons who happen to be present at them.

²¹ For example, Petronius 30. 11; Laus Pisonis 119, 134; Martial 10. 10. 11, 10. 74. 2; Juvenal 5. 64. 9. 72.

²² Cultor is the equivalent of cliens at Ovid, Ars Am. 1. 722; Seneca, De brevitate vitae 2. 4; Laus Pisonis 109, 133; Martial 9. 84. 4; Juvenal 9. 49. Ernout-Meillet, Dictionnaire étymologique de la langue latine ⁴, discuss only the possibility of a derivation of cliens from cluere. But an etymology current in late antiquity connected it with the verb colere: cf. Servius on Aen. 6. 609, ‘si enim clientes quasi colentes sunt, patroni quasi patres…’, and Isidore, Orig. 10. 53, ‘clientes prius colientes dicebantur a colendis patronis.’ That writers of the classical period already assumed the same etymology is suggested by Seneca's collocation of words at Epist. 47. 18, ‘colant tamquam clientes’, and by Horace's construction of the noun with an adverb at Epist. 1. 7. 75, ‘mane cliens’.

²³ A few examples out of many: Cicero, Amic. 69; Seneca, Epist. 47. 18; Laus Pisonis 113; Martial 6. 50. 1, 12. 68. 2; Statius, , Silvae 1. 4. 36;Google Scholar Pliny, Epist. 2. 9. 6, 7. 31. 5; Tacitus, Dial. 9. 5; Juvenal 7. 37.

²⁴ Cicero, Att. 5. 4. 2, Fam. 9. 5. 2; Petronius 5. 4; Martial 6. 50. 3, 10. 9. 3; Pliny, Epist. 1. 5. 15; Juvenal 7. 91.

²⁵ Seneca, Epist. 3. 1; Martial 5. 57. For the use of the word generally, see M. Bang's appendix in Wissowa, L. Friedlaender-G., Darstellungen aus der Sittengeschichte Roms, IV ^9–10 (1911), 82–8Google Scholar.

²⁶ The origin of the term is discussed by Fraenkel, E., Elementi Plautini in Plauto, tr. Munari, F. (1960), 182–3Google Scholar. To the examples from Plautus, Terence, Horace and Juvenal cited by Fraenkel, one might add Columella 1 praef. 9; Seneca, De firmitate sapientis 15. 1; Statius, Silvae3. 2. 92; Martial 2. 18. 5 and 8, 2. 68, 3. 7. 5, 4. 19. 13; Anth. Lat. 407. 1 and 408. 2 (? Seneca), 252 (? Florus).

²⁷ It is remarkable to see that these words are used no matter how debased the relationship. Cf. Martial 9. 2: ‘Pauper amicitiae cum sis, Lupe, non es amicae / et queritur de te mentula sola nihil. / ilia siligeneis pinguescit adultera cunnis, / convivam pascit nigra farina tuum. / incensura nives dominae Setina liquantur, / nos bibimus Corsi pulla venena cadi; / empta tibi nox est fundis non tota paternis, / non sua desertus rura sodalis arat; / splendet Erythraeis perlucida moecha lapillis, / ducitur addictus, te futuente, cliens; / octo Syris sufrulta datur lectica puellae, / nudum sandapilae pondus amicus erit …’

²⁸ Nevertheless, it might be worth citing texts where this language is used of two parties clearly identified as poet and rich friend: for Terence and Scipio Africanus, Porcius Licinus frag. 3 (p. 45 Morel); for Saleius Bassus and Curiatius Maternus, Tacitus, Dial. 5. 2; for Martial and Pliny, Pliny, Epist. 3. 21. Credit for insisting that during the early Empire the Romans ordinarily spoke of amici and amicitia where we speak of ‘patrons,’ ‘clients,’ and ‘patronage’ should go to Allen, W., ‘On the Friend-ship of Lucretius with Memmius’, CP 33 (1938), 167–81,Google Scholar and to Allen, W. and DeLacey, P. H., ‘The Patrons of Philodemus’, CP 34 (1939), 59–65Google Scholar.

²⁹ Tacitus 1 (1958), 93.

³⁰ The Letters of the Younger Pliny, tr. Radice, B. (1963), 219Google Scholar.

³¹ cf. also Epist. 5. 17. 1, ‘scio quanta opere bonis artibus faveas’ and 6, ‘faveo enim saeculo ne sit sterile et effetum’, in a letter describing a recitation given by a young noble; and i. 14. 5, ‘neque enim est fere quisquam qui studia, ut non simul et nos amet,’ again in the context of faithful attendance at readings of poetry.

³² Compare these metaphors with the language used by Statius, Silvae 3 praef., 5–11 (line numbers in this and subsequent references to the prose prefaces of the Silvae correspond to A. Klotz's Teubner text, second edition, 1911): ‘tibi certe, Pollio dulcissime… non habeo diu probandam libellorum istorum temeritatem, cum scias multos ex illis in sinu tuo subito natos… quotiens in illius facundiae tuae penettali seductus altius litteras intro et in omnis a te studiorum sinus ducor.’

³³ cf. also 7. 68, 7. 80, and 10. 93.

³⁴ cf. 7. 12, 8. 12. 1, and Martial 7. 52.

³⁵ See 4. 86. 6–7, 7. 26. 9–10, 7. 72, 10. 33; cf. Statius, Silvae 4 praef. 43–5.

³⁶ Let me emphasize again that I am not talking about relationships between poets and the emperors, who did on several occasions bestow large sums on poets.

³⁷ cf. also the gifts itemized in 7. 92 and 12. 36.

³⁸ Some of the evidence for these holidays is gathered in Balsdon, J. P. V. D., Life and Leisure in Ancient Rome (1969), 121–6Google Scholar.

³⁹ Appian's letter may be found on pp. 227–8 of van den Hout's edition of Fronto, or in the Loeb edition by C. R. Haines (1962) 1, 264–9.

⁴⁰ This passage of Fronto's letter is found on pp. 230–2 of van den Hout, or in vol. I, pp. 275–7 of the Loeb; translation is by Haines. At an earlier period, Cicero is less explicit than Fronto about gift-giving, but his viewpoint is similar. Note first that the subject barely appears among the issues requiring the casuist's attention in the De amicitia (see sections 70 and 79–80): surely not because the exchange of gifts was unheard of or unimportant in Cicero's time, but because the scale on which it was practised was too small to create problems. In the De officiis, Cicero divides beneficia into material gifts and services, and argues that the latter are far more befitting a gentleman (2. 52 f.). Note especially section 64, where he raises the same consideration as Fronto: ‘habenda autem ratio est rei familiaris, quam quidem dilabi sinere flagitiosum est, sed ita, ut inliberalitas avaritiaeque absit suspicio. posse enim liberalitate uti non spoliantem se patrimonio nimirum est pecuniae fructus maximus.’ Finally, compare the hierarchy of beneficia set up by Seneca at De beneficiis 1. 11, where the kinds of gifts ordinarily exchanged among friends occupy the lowliest place.

⁴¹ L'ordre équestre à l'époque républicaine 1 (1966), 441–56.

⁴² ‘Republican and Augustan Writers Enrolled in the Equestrian Centuries’, TAPA 99 (1968), 469–86.

⁴³ ‘Natus in Etruria Volaterris, eques Romanus, sanguine et affinitate primi ordinis viris coniunctus’, according to the ancient vita (ll. 4–5, p. 37 of W. V. Clausen's edition of Persius, 1956).

⁴⁴ See Epigrams 5. 13. 1–2 : ‘sum, fateor, sem perque fui, Callistrate, pauper/sed non obscurus nec male notus eques’; cf. 3. 95. 9–10, which attests Martial's tribunate.

⁴⁵ Silvae 5. 3. 116–20, alluding to the golden bulla of the knights.

⁴⁶ CIL x. 5382 = ILS 2926. The text of this inscription has recently been discussed by S. Monti in RAAN 40 (1965), 79–110; pace Reynolds, J. in JRS 61 (1971), 146,Google Scholar Monti does not seem to be arguing against the identification of the cohort captain and the satirist.

⁴⁷ See the vita in Reifferscheid's edition of the fragments of Suetonius, p. 76.

⁴⁸ See Pliny, Epist. 3. 7.

⁴⁹ Argonautica 1. 5–7 identify the poet as quindecimvir sacris faciundis.

⁵⁰ These figures are taken from Duncan-Jones, R., The Economy of the Roman Empire: Quantitative Studies (1974), 4, n. 2Google Scholar, who gives the evidence for them.

⁵¹ Carcopino, J., Daily Life in Ancient Rome, tr. Lorimer, E. O. (1940), 66Google Scholar, had suggested that this was the significance of the equestrian census, and he documented his suggestion with the first two of the passages I quote here.

⁵² See Duncan-Jones, op. cit. (n. 50), 4 and 21.

⁵³ On the semestris tribunatus, see the section s.v. ‘legio’ in de Ruggiero, Diz. Epigr. (by Passerini) IV, cols. 578–9.

⁵⁴ Perhaps another point should be made. I have listed only those poets who appear actually to have held the rank of eques Romanus. Yet one might also possess the equestrian census, or more than that, without meeting the other qualifications of an eques. It does not follow, therefore, that poets not known as equites, or even known not to be equites (like Phaedrus), were poor men.

⁵⁵ cf. Appian to Fronto, ‘friends too do not shrink from taking under wills. And why, pray, should a man take under a will, but take nothing from the living, when the latter gift is an even greater proof of affection?’ (van den Hout, 227; Loeb I, 267). See also Fronto's retort to this sophistry (van den Hout, 230; Loeb 1, 273).

⁵⁶ Martial 2. 91–2 and, for Suetonius, Pliny, Epist. 10. 94.

⁵⁷ See Kaser, M., Das römische Privatrecht² 1 (1971), 320–1Google Scholar.

⁵⁸ Duncan-Jones, op. cit. (n. 50), 21, cf. 25–7.

⁵⁹ Epigrams 1. 99. Cf. 2. 61. 11–12, which alludes to a legacy of 300,000 and two of 100,000.

⁶⁰ Epist. 2. 4. 2 and 6. 32. 2; cf. the dowry given by Aurelius Cotta to the daughter of his freedman Zosimus, ILS 1949.

⁶¹ Epist. 3. 21; cf. 6. 25. 3.

⁶² ILS 1949; Seneca, De benef. 3. 9. 2; Laus Pisonis 109–11; Martial 4. 67, 5. 19. 10, 5. 25, 12. 6. 9–11, 14. 122; Pliny, Epist. 1. 19.

⁶³ cf. also the vague allusions to largiri and pecunia in Seneca, De benef. 2. 14. 4, 3. 8. 2, 3. 19.1; Martial 4. 40, 5. 19. 9, 5. 42, 8. 19. 9, 12. 6. 9, 12. 13; Pliny Epist. 9. 30.

⁶⁴ Martial 2. 30, 3. 41, 4. 15, 6. 5, 6. 20, 8. 37, 9. 102, 10. 15, 10. 19. 2, 11. 76; Pliny, Epist. 3. 11.

⁶⁵ cf. Otho's friendly purchase of land for a speculator in Galba's Praetorian Guard, Tacitus, Hist. 1. 24. 2.

⁶⁶ cf. the provisions of Trimalchio's will, Petronius 71. 2, and the arrangements discussed in the Digest, 33. 7. 27 pr., 35. 1. 17, 39. 5. 9. 1.

⁶⁷ Siculus, Calpurnius, Eclogues 4. 152–5;Google Scholar Martial 1. 55, 8. 18. 9, 11. 18; Juvenal 9. 59–60; Lucian, , De mercede conductis 20. 22Google Scholar (line numbers as in Macleod's OCT).

⁶⁸ (Laelius speaking) ‘hospes et amicus meus M. Pacuvius’, Cic., Amic. 24.

⁶⁹ (Atticus speaking) ‘Antiochus, familiaris meus … quocum vixi’, Cic., Leg. 1. 54.

⁷⁰ ‘Est apud M. Pisonem adulescentem… Peripateticus Staseas’, Cic., De or. 1. 104.

⁷¹ ‘ad Cornelium Gallum se contulit vixitque una familiarissime’, Suet., Gram. 16. 1.

⁷² After having been forbidden access to the palace of Augustus, Timagenes ‘in contubernio Pollionis Asinii consenuit’, Seneca, , De ira 3. 23. 5–8;Google Scholar the whole passage is relevant for showing the closeness of Timagenes' association with Pollio.

⁷³ ‘receptus in Planci oratoris contubernium’, Suet., Gram. 30. 2.

⁷⁴ (Fronto speaking) ‘habitavimus una, studuimus una’ (van den Hout, 165).

⁷⁵ 9. 3. 5. 1. 39. 5. 7. 9.

⁷⁶ Epist. 4. 13. 10; cf. Epist. 3. 3. Lucian's De mercede conductis throughout speaks of domestic appointments for philosophers, grammatici, rhetors, and even musicians (see 4. 15–18 and 25–6, 25. 33, 36. 28–9).

⁷⁷ cf. Martial 6. 8.