¹ For the various functions of the Roman emperor, especially in the civil sphere, Millar, F., The Emperor in the Roman World (1977/1992Google Scholar) is fundamental; see also Campbell, J. B., The Emperor and the Roman Army, 31 B.C.–A.D. 235 (1984)Google Scholar, for the emperor's military roles. On the importance of the ‘symbolics of power’ and the ‘power of images’ in the overall configuration of imperial power, authority, and legitimacy, see especially Alföldi, A., Die monarchische Repräsentation im römischen Kaiserreiche (1970) = RömMitt 49 (1934), 3–118Google Scholar and RömMitt 50 (1935), 3–158Google Scholar (citations from the 1970 edition); P. Veyne, Le pain et le cirque (1976), ch. 4; S. Price, Rituals and Power: The Roman Imperial Cult in Asia Minor (1984); P. Zanker, The Power of Images in the Age of Augustus, transl. A. Shapiro (1988; German orig. 1987). Two recent studies that explore different aspects of the symbolics of imperial power are J. E. P. Davies, Death and the Emperor: Roman Imperial Funerary Monuments from Augustus to Marcus Aurelius (2000), and C. Ando, Imperial Ideology and Provincial Loyalty in the Roman Empire (2000).

² Clupeus virtutis: RG 34.2. For imperial virtues in the late antique panegyrics, see Seager, R., ‘Some imperial virtues in the Latin prose panegyrics: the demands of propaganda and the dynamics of literary composition’, Papers of the Liverpool Latin Seminar 4, 1983 (1984), 129–65;Google Scholar this topic is also addressed in many of the essays in M. Whitby (ed.), The Propaganda of Power: The Role of Panegyric in Late Antiquity (1998). On imperial virtues in general, see Charlesworth, M. P., ‘The virtues of a Roman emperor: propaganda and the creation of belief’, PBA 23 (1937), 105–33;Google ScholarWallace-Hadrill, A., ‘The emperor and his virtues’, Historia 30 (1981), 298–323;Google ScholarFears, J. R., ‘The cult of virtues and Roman imperial ideology’, ANRW II.17.2 (1981), 827–948;Google ScholarClassen, C. J., ‘Virtutes Imperatoriae’, Arctos 25 (1991), 17–39Google Scholar.

³ The authority to mint coins has traditionally been recognized as a prerogative of the state. See discussion in P. Grierson, Numismatics (1975), 95–7; for the ancient world, see briefly M. Crawford, ‘Roman imperial coin types and the formation of public opinion’, in Brooks, C.et al. (eds)Google Scholar, Studies … Grierson (1983), 51. Whether the types were chosen by the emperor himself or by a low-level bureaucrat has no bearing on the official character of imperial coins.

⁴ See, e.g., Howgego, C., Ancient History from Coins (1995), 62 ffCrossRefGoogle Scholar.

⁵ For the application of statistical methods to the study of ancient coins, see Carcassone, C. and Hackens, T. (eds), Statistics and Numismatics. Statistique et Numismatique. Paris, 17–19 sept. 1979 (1981)Google Scholar. For economic studies based on large samples of imperial coins, see Carradice, I., Coinage and Finances in the Reign of Domitian, A.D. 81–96, BAR Int. Ser. 178 (1983)Google Scholar; R. Duncan-Jones, Money and Government in the Roman Empire (1994), to be read with Metcalf's, W. review in RSN 74 (1995), 145–59;Google ScholarHobley, A. S., An Examination of Roman Bronze Coin Distribution in the Western Empire A.D. 81–192, BAR Int. Ser. 688 (1998)Google Scholar, on coin circulation.

⁶ The quotation is from Wallace-Hadrill, A., ‘Image and authority in the coinage of Augustus’, JRS 76 (1986), 66–87,Google Scholar at 69. Similarly, on the importance of the imperial virtues on the coinage of Vespasian, Fears, op. cit. (n. 2), 900, writes that ‘in a quantitative sense, at least, the Virtues assumed a prominence far beyond their role in Augustan imperial imagery’. A statement of this sort simply cannot be sustained without the support of numerical evidence. On the need to quantify these sorts of claims in context, see discussion in Laslett, P., ‘The wrong way through the telescope: a note on literary evidence in sociology and in historical sociology’, British Journal of Sociology 27 (1976), 319–42CrossRefGoogle Scholar.

⁷ See Potter, D. S., Prophets and Emperors (1994), 110–30CrossRefGoogle Scholar, on the different forms of communication available to emperors.

⁸ RIC I ² (1984), xxi–xxiiGoogle Scholar. Sutherland provides a list of twenty-nine ‘major collections’ (xvii–xviii), but does not indicate which of these collections were consulted in order to produce the frequency-estimates.

⁹ On the unreliability of the major collections as indicators of frequency, see Volk, T., ‘Mint output and coin hoards’, in G. Depeyrot et al. (eds), Rhythmes de la production monétaire, de I'antiquité à nos jours (1987), 141Google Scholar; I. Carradice, ‘Towards a new introduction to the Flavian coinage’, in M. Austin et al. (eds), Modus Operandi: Essays … Rickman (1998), 97. Duncan-Jones, op. cit. (n. 5), 134, n. 28, writes that the scarcity ratings in RIC ‘depart wildly from what the hoards imply’.

¹⁰ See Buttrey, T. V., ‘Calculating ancient coin production: facts and fantasies’, NC 153 (1993), 335–51,Google Scholar at 338–45, and idem, ‘Calculating ancient coin production Il: why it cannot be done’, NC 154 (1994), 342–52, on the ‘intractable problem’ of determining number of coins from number of dies. Buttrey exposes the absence of hard evidence behind the figure of 30,000 coins per die employed by M. Crawford, Roman Republic Coinage (1974), 694 (and by many others since) to measure the annual production of Republican denarii, emphasizing that the rate of coin production per die is not a constant but a variable (which rules out extrapolation). Howgego, C., ‘The supply and use of money in the Roman world 200 B.C. to A.D. 300’, JRS 82 (1992), 3,Google Scholar notes that individual dies in England between 1281 and 1327 produced anywhere from 5,000 to 74,000 coins. This variability, from a comparable pre-industrial context for which mint records do survive, makes meaningful calculations based on die counts from the Roman world very difficult. This is the basic problem that De Callataÿ, F., ‘Calculating ancient coin production: seeking a balance’, NC 155 (1995), 289–311,Google Scholar cannot overcome in his response to Buttrey. For a recent attempt to measure production levels in relative terms, see Lockyear, K., ‘Hoard structure and coin production in antiquity — an empirical investigation’, NC 159 (1999), 215–43Google Scholar.

¹¹ This is the method that Carradice, in a preliminary study for a new edition of RIC II (op. cit. (n. 9)), has used to correct the first edition's classifications of rarity and commonness for Flavian coin types.

¹² Reverse types which happened to be on coins hoarded for economic value will naturally be better represented in the hoards, but the ‘over-representation’ of these types will be minimized both by the large number of reverse types in circulation at any one time and by the size and chronological spread of the sample. The division of hoards into ‘circulation’ hoards and ‘savings’ hoards is also irrelevant for this sample, and is in any case no longer considered tenable for the imperial period. See Kent, J. P. C., ‘Interpreting coin-finds’, in Casey, J. and Reece, R. (eds), Coins and the Archaeologist, BAR Int. Ser. 4 (1974), 185;Google Scholar M. Crawford, ‘Numismatics’, in idem (ed.), Sources for Ancient History (1983), 199 ff.; Volk, op. cit. (n. 9), 159, n. 3; Duncan-Jones, op. cit. (n. 5), 67 ff. On hoarding and coin finds in general, see Grierson, op. cit. (n. 3), 124–39, and for the ancient world, Crawford, op. cit., 187–207; see also Buttrey, T. V., ‘The content and meaning of coin hoards’, JRA 12 (1999), 526–32,Google Scholar a review of D. Backendorf, Römische Münzschätze des zweiten und ersten Jahrhunderts v.Chr. vom italienischen Festland (1998).

¹³ Suet., Aug. 75: ‘modo munera dividebat … modo nummos omnis notae, etiam veteres regios ac peregrinos’, ‘sometimes he distributed favours … other times he gave out coins of every type, even those of the old kings and foreign coins’.

¹⁴ See discussion in Carradice, op. cit. (n. 5), 57–60. For criticism of samples based on groups of hoards (sometimes referred to as ‘aggregate’ or ‘composite’ hoards), see R. Duncan-Jones, Structure and Scale in the Roman Economy (1990), 39; Buttrey, op. cit. (n. 10, 1993), 335 ff.; cf. Howgego, op. cit. (n. 10), 2–4, on limits to quantification in general. It cannot be stressed too much that these criticisms address the problems with using hoard-generated samples for economic and monetary, not iconographic analysis.

¹⁵ For the sources, see Appendix.

¹⁶ For a preliminary report on the Trier hoard, see Gilles, K.-J., ‘Der grosse römischen Goldmünzenfund aus Trier’, KJ 34 (1994), 9–24Google Scholar. Duncan-Jones, op. cit. (n. 5), App. 10, 266 ff., provides a list of published gold hoards.

¹⁷ Hobley, op. cit. (n. 5), for example, created a sample of c. 25,000 imperial aes coins largely on the basis of evidence from local museum collections, which are not necessarily representative of circulation patterns in antiquity; see my review, AJN 11 (1999), 160–4Google Scholar.

¹⁸ For convincing arguments associating the move with Nero's reform of the imperial coinage in A.D. 64, see Metcalf, W., ‘Rome and Lugdunum again’, AJN 1 (1989), 51–70,Google Scholar with earlier bibliography. It should be noted that mints outside of Rome were in operation during the period A.D. 69–235 and that some of the coins assigned in the catalogues to the Rome mint may in fact have been produced elsewhere. But since these coins were more or less indistinguishable from those of the Rome mint, they still in effect represented the ‘official’ image of the emperor and should not distort the present analysis.

¹⁹ See Buttrey, T. V., ‘Vespasian as moneyer’, NC ⁷ 12 (1972), 89–109Google Scholar.

²⁰ Volk, op. cit. (n. 9), 159. Volk also quotes Thordeman's law on the composition of coin hoards: ‘The content of each coin-find stands in a certain ratio to the amount of the coinage during the period covered by the find, and that according to the law of high numbers this proportion reaches increasing agreement the larger the find is numerically’ (144). See also R. Reece, ‘The “normal” hoard’, in Carcassone and Hackens, op. cit. (n. 5), 335–41. For the suggestion that large hoards might be particularly anomalous, see Lockyear, op. cit. (n. 10), 220.

²¹ The imperial mint's total volume of production is much debated and, on the available evidence, probably impossible to quantify. For a very rough sense of the order of magnitude, Duncan-Jones, op. cit. (n. 5), 168, has estimated that over half a billion denarii were produced under Septimius Severus alone (table 11.2). These calculations were based on a projection of ‘7,982 coins per silver die’ (164), a figure which Duncan-Jones himself concedes is ‘obviously elastic’ (165).

²² For concise introductions to the main problems involved, see K. Hopkins, ‘Introduction’, in Garnsey, P.et al. (eds)Google Scholar, Trade in the Ancient Economy (1983), ix–xxv; W. Harris, ‘Between archaic and modern: some current problems in the history of the Roman economy’, in idem (ed.), The Inscribed Economy: Production and Distribution in the Roman Empire in Light of Instrumentum Domesticum, JRA Suppl. 6(1993), 11–29.

²³ Hopkins, K., ‘Taxes and trade in the Roman Empire (200 B.C.–A.D. 400)’, JRS 70 (1980), 101–25,Google Scholar at 112–16 (with fig. 4). Note also the explicit qualifications to this model which Hopkins makes (103–4), often ignored by those who attack it.

²⁴ Duncan-Jones, op. cit. (n. 14), 30–47 and 187 ff. on provincial taxation. Duncan-Jones has published a number of other studies arguing for regional as opposed to empire-wide coin circulation: ‘Mobility and immobility of coin in the Roman Empire’, AIIN 36 (1989), 121–37;Google Scholar op. cit. (n. 5), 172–9; ‘Empirewide patterns in coin-hoards’, in King, C. E. and D. Wigg (eds), Coin-finds and Coin Use in the Roman World (1996), 139–52Google Scholar; ‘The monetization of the Roman Empire: regional variations in the supply of coin types’, in Paul, G. M. and M. Ierardi (eds), Roman Coins and Public Life Under the Empire (1999), 61–82CrossRefGoogle Scholar. See below, n. 26, for some problems with Duncan-Jones' methods and conclusions.

²⁵ Howgego, C., ‘Coin circulation and the integration of the Roman economy’, JRA 7 (1994), 12–16,Google Scholar with references.

²⁶ The data that Duncan-Jones cites to show dissimilarities in regional coin populations, in fact, are often open to different interpretation. I cite two examples. The first is the representation of Trajan RIC 147 in six hoards, as a percentage of all Trajanic denarii in those hoards (Duncan-Jones, op. cit. (n. 14), 41). The figures are: Tell Kalak (Syria): 1.3%; Reka Devnia (Lower Moesia): 0.8%; La Magura (Dacia): 1.8%; Stockstadt (Rhineland): 3.3%; Bristol (Britain): 3.3%; Londonthorpe (Britain): 4.0%. For Duncan-Jones these figures represent a ‘strongly contrasting pattern’ in the hoards from East and West, but these differences are almost negligible for the purposes of this study, which emphasizes much broader patterns. The second example is drawn from denarii of Marciana and Matidia (Duncan-Jones, op. cit. (n. 14), 41, repeated in op. cit. (n. 5), 175). Duncan-Jones notes that denarii of Marciana and Matidia do not appear in a sample of over 600 Trajanic denarii from four hoards in North-West Europe, and appear only once in a sample of over 1,900 denarii from three hoards in Italy, but do appear once for every 150 Trajanic denarii in hoards from Syria and Egypt (the total number of coins from the Syrian and Egyptian hoards, an important factor for the significance of this observation, is not provided). He sees strong evidence here of ‘regional differences’, but such uncommon coins really cannot support this type of argument. More and better evidence is needed to prove that silver coins circulated only in regional pools.

²⁷ Howgego, op. cit. (n. 25), 14.

²⁸ Regional circulation of precious-metal coins in the third and fourth centuries: J.-P. Callu, La politique monétaire des empereurs romains de 238 à 311 (1969), 390 ff.; idem, ‘Structure des dépôts d'or au IVe siècle (312–392)’, in E. Frézouls (ed.), Crise et redressement dans les provinces européennes de l'empire (milieu du Ille–milieu du IVe siècle ap. J.-C.) (1983), 157–74; Kent, J. P. C., RIC VIII (1981), 74–7, 96Google Scholar ff.

²⁹ W. Eck, ‘Senatorial self-representation: developments in the Augustan period’, in F. Millar and E. Segal (eds), Caesar Augustus: Seven Aspects (1984), 129–67, discusses certain aspects of this process.

³⁰ For the significance of the honorific cognomina of Metellus and Sulla, see discussion in Fears, op. cit. (n. 2), 877–82. Fears also notes the earlier and informal associations of specific virtues and qualities with certain gentes, such as virtus with the Cornelii Scipiones, libertas with the Junii, etc.

³¹ S. Weinstock, Divus Julius (1971), 80 ff., 233 ff.

³² For the notion of virtus, clementia, iustitia, and pietas as a canon of imperial virtues derived from Greek thought, see Charlesworth, op. cit. (n. 2); Wickert, L., ‘Princeps (civitatis)’, RE 22.2 (1954), 2231;Google Scholar Weinstock, op. cit. (n. 31), 228–30. This view is demonstrably incorrect, as Wallace-Hadrill, op. cit. (n. 2), and Classen, op. cit. (n. 2), have shown, among other reasons because the Roman term pietas has no Greek antecedent among the cardinal virtues of the king. Perhaps, as Wallace-Hadrill has suggested, it was simply the number of four which was meant to suggest a philosophical canon (303). On the clupeus virtutis, see K. Galinsky, Augustan Culture (1996), 80–90, esp. 84–8 on the four virtues.

³³ Potter, D. S., ‘Political theory in the senatus consultum Pisonianum’, AJP 120 (1999), 65–88,Google Scholar at 71.

³⁴ See Wallace-Hadrill, op. cit. (n. 6), on the importance of reading the obverse and the reverse of a coin as different parts of a single, unified message.

³⁵ On the synthesis of image and text in Roman visual culture, see e.g. Hölscher, T., ‘Die Geschichtsauffassung in der römischen Repräsentationskunst’, JDAI 95 (1980), 279–81;Google Scholar Eck, op. cit. (n. 29), 132–3; Woolf, G., ‘Monumental writing and the expansion of Roman society in the early Empire’, JRS 86 (1996), 27–9Google Scholar.

³⁶ Hes., Theog. 384 (Nike), 902 (Dike and Eirene); cf. 135 (Themis), 902 (Eunomia), 934 (Phobos). For the Greek background to the personification of abstract ideas, see H. A. Shapiro, Personifications in Greek Art: The Representation of Abstract Concepts 600–400 B.C. (1993). For the translation of this mode of representation into Roman political culture, see Weinstock, op. cit. (n. 31), 228–30; Fears, op. cit. (n. 2), 828 ff., 875–7.

³⁷ See Mattingly, H., ‘The Roman “Virtues’”, HTR 30 (1937), 103–17,CrossRefGoogle Scholar and Fears, op. cit. (n. 2), on the religious dimension of the imperial virtues, with emphasis on the virtues as recipients of cult. For Wallace-Hadrill, op. cit. (n. 2), this ‘religious metaphor’ serves to ‘identif [y] the view of the emperor as a “charismatic” one’ (315).

³⁸ Mattingly, op. cit. (n. 37), 107.

³⁹ This definition of ‘personification’ is borrowed and slightly adapted from that offered by J. C. M. Toynbee, ‘Picture-language in Roman art and coinage’, in R. A. G. Carson and C. H. V. Sutherland (eds), Essays … Mattingly (1956), 216.

⁴⁰ Wallace-Hadrill, op. cit. (n. 2), 308 ff.

⁴¹ These divisions are not conceptual and are only intended to isolate different iconographic categories; some of the ‘inanimate objects’, for example, may be read as symbols for one of the other categories (e.g. a club to represent Hercules). A few comments on the placement of certain types. Within the category of ‘gods, goddesses, and minor deities’ I have placed Genius types and those of the hero Hercules. The ‘objects/miscellaneous’ category includes inanimate objects (altars, buildings, military equipment, religious implements etc.), animals, events (e.g. adventus), imperial titulature and scenes not involving the emperor or members of the imperial family (e.g. a lictor burning debts). Depictions of the reigning emperor's predecessor have been included in the category of ‘the emperor and the imperial family’. Finally, the ‘provinces’ category includes Hadrian's Adventus, Exercitus and Restitutor series.

⁴² On personification in Roman imagery see e.g. Hamberg, P., Studies in Roman Imperial Art with Special Reference to the State Reliefs of the Second Century (1945), 15–45Google Scholar; Toynbee, op. cit. (n. 39); Hölscher, op. cit. (n. 35), 273–9. Kuttner, A., Dynasty and Empire in the Age of Augustus (1995)Google Scholar, 69 ff., discusses the importance of studying personifications not individually but in groups.

⁴³ Katzenellenbogen, A., Allegories of the Virtues and Vices in Medieval Art (1939)Google Scholar; see also the references collected in E. Kantorowicz, The King's Two Bodies (1957), 114, n. 80.

⁴⁴ On the importance of the obverse portrait of the emperor, see C. E. King, ‘Roman portraiture: images of power?’, in Paul and Ierardi, op. cit. (n. 24), 123–36. After the coinage, imperial portraiture was surely the most important medium for spreading the emperor's image. For the order of magnitude of imperial portrait production, see Pfanner, M., ‘Über das Herstellen von Porträts: Ein Beitrag zu Rationalisierungsmaβnahmen und Produktionsmechanismen von Massenware im späten Hellenismus und in der römischen Kaiserzeit’, JDAI 104 (1989), 157–257,Google Scholar who estimates (178–9) that between 25,000 and 50,000 portraits of Augustus were produced between 30 B.C. and A.D. 14, which corresponds roughly to an average production of between 500 and 1,000 portraits per year. Following G. Alföldy's figure of 1,000 cities in the Empire, Pfanner figures that each city would have received a new portrait of Augustus every one to two years (178). On the dissemination of the imperial image in general, see H. Kruse, Studien zur offiziellen Geltung des Kaiserbildes im römischen Reiche (1934); Zanker, P., ‘Prinzipat und Herrscherbild’, Gymnasium 86 (1979), 353–68;Google ScholarPekàry, T., Das römische Kaiserbildnis in Staat, Kult und Gesellschaft (Das römische Herrscherbild III.5) (1985)Google Scholar; C. B. Rose, Dynastic Commemoration and Imperial Portraiture in the Julio-Claudian Period (1997). For the role of statues and portraits in establishing an imperial ‘presence’ throughout the Empire, see also Price, op. cit. (n. 1), ch. 7; cf. K. Hopkins, Conquerors and Slaves (1978), 221 ff.

⁴⁵ This group of eleven virtues is the same as that provided by Wallace-Hadrill, op. cit. (n. 2), 310, n. 56, with the exception of Constantia, which was not minted between A.D. 69 and 235. For a tabular presentation of all the personifications minted by all emperors from Augustus to Constantine (without treatment of relative frequency), see Gnecchi, F., ‘Le personificazioni allegoriche sulle monete imperiali’, RIN 18 (1905), 354–9Google Scholar. Fears, op. cit. (n. 2), 841–5, n. 67, provides a useful bibliography (through 1981) for most of these virtues.

⁴⁶ In addition to the eleven imperial virtues, the personifications which appeared on imperial denarii between A.D. 69 and 235 were: Abundantia, Aeternitas Annona, Bonus Eventus, Concordia, Fecunditas, Felicitas, Fides, Fortuna, Hilaritas, Honos, Iuventas, Laetitia, Libertas, Moneta, Nobilitas, Pax, Perpetuitas, Salus, Securitas, Spes, Tranquillitas, and Victoria.

⁴⁷ All coins minted at Rome under the reigning emperor are included in that emperor's totals (e.g. coins of Titus and Domitian minted under Vespasian are tabulated under Vespasian). The coinage of the imperial women has also been included under the totals for the reigning emperor. In those cases in which there were two (or more) emperors (e.g. Septimius Severus, Caracalla, and Geta), I have also tabulated all coins under the senior emperor. Finally, I have not included in Fig. 1 (or in the figures that follow) the tabulations for emperors whose reigns were too brief to make a significant impact on the Empire's coin population, or for whom there was not sufficient evidence.

⁴⁸ Whether this sudden drop is to be explained by the brevity of Macrinus' reign or by a new, ‘non-senatorial’ and ‘non-traditional’ representation of the Principate is unclear.

⁴⁹ Wallace-Hadrill, op. cit. (n. 2), 311–14, argues that the virtues were first given prominence under Hadrian as an official reaction to the emphasis on human virtues in Pliny's Panegyricus: ‘In offering a gallery of imperial virtues, the mint [sc. under Hadrian] responds to the mood of the times’ (313).

⁵⁰ When Wallace-Hadrill, op. cit. (n. 2), writes that the virtues were ‘at all times a secondary phenomenon on the coinage’ (313), he is, sensu stricto, correct, but the quantitative evidence presented in Fig. 1 shows how much this sort of blanket statement can and should be refined.

⁵¹ Shaw, B. D., ‘Body / power / identity: the passion of the martyrs’, JECS 4 (1996), 269–312,Google Scholar examines the development of patientia as an important philosophical and above all Christian virtue in the first three centuries A.D. (esp. 291 ff).

⁵² It has been noted that Iustitia types were uncommon (by e.g. Charlesworth, op. cit. (n. 2), 113; Lichocka, B., Justitia sur les monnaies impériales romaines (1974)Google Scholar), but when scholars have referred to the ‘infrequency’ of the type they rely, it seems (their evidence for these claims is never explicitly cited), on the number of types found in the catalogues — not necessarily a reliable indicator of the actual frequency with which a given type was minted (see above). The quantitative evidence in this case serves to document what until now has only been inferred.

⁵³ Millar, op. cit. (n. 1), 3–6, 228, 240, 252, 271, 469, 507, 516, 527–9, 535, 549, 636 ff. and passim. For the emperor's role as judge and the ideology of imperial justice, see also J. Béranger, Recherches sur l'aspect idéologique du principal (1953), 270 ff.; Wickert, op. cit. (n. 32), 2248–53.

⁵⁴ TLL III, 1334–5 (IA 1). Cf. Cic, Part. or. 2: ‘aut saevitiam aut clementiam iudicis’. Clementia also had the wider sense of mercy toward those upon whom one takes vengeance. Seneca's definition captures both aspects, with emphasis on leniency in the exaction of poenae: ‘clementia est temperantia animi in potestate ulciscendi vel lenitas superioris adversus inferiorem in constituendis poenis, inclinatio animi ad lenitatem in poena exigenda’ (Clem. 2.3.1).

⁵⁵ TLL VII.2, 714–15 (esp. IA b). Cf. Ulp., Dig. 1.1.10: ‘Iustitia est constans et perpetua voluntas ius suum cuique tribuendi’. See discussion in T. Adam, Clementia Principis (1970), 31–9, on the relationship between clementia and iustitia in Seneca's De Clementia.

⁵⁶ For the dichotomy between iustitia and clementia, cf. Sen., Clem. 1.20.2: ‘hoc enim ad iustitiam, non ad clementiam pertinet’; see also Amm. Marc. 16.5.12 for contradictory demands on the emperor to show both clementia and iustitia. See Millar, op. cit. (n. 1), 516–17, on the potential conflict between these two virtues, and R. Saller, Personal Patronage under the Early Empire (1982), 56–7, on the related conflict between the emperor's obligation to give justice and to meet the demands of amicitia.

⁵⁷ The tentative conclusion of Wallace-Hadrill, A., ‘Galba's Aequitas’, NC 141 (1981), 37Google Scholar.

⁵⁸ Other explanations for the relative infrequency of Clementia and Iustitia types (based, as noted above, n. 52, on the potentially misleading number of types in the catalogues) tend to consider the types in isolation. Charlesworth, op. cit. (n. 2), 113, for example, argues that clementia was ‘too much a despotic quality’ for the Roman emperor. This view was accepted by Béranger, op. cit. (n. 53), 271, and Wickert, op. cit. (n. 32), 2243. The infrequency of Iustitia types is ascribed by Béranger to a shift in the meaning of iustitia from ‘l'abstraction’ of Justice to ‘l'attitude personelle’ of the princeps (270–1). Another possibility worth considering is that these ‘judicial’ virtues were not under-represented on the imperial coinage, but were instead over-represented in other sources (especially in the epigraphic record, reflecting the incentive to make a public and permanent record of favourable judgements).

⁵⁹ Strack, P., Untersuchungen zur römischen Reichsprägungen des 2. Jahrhunderts I–III (1931–1937), I, 154–64Google Scholar.

⁶⁰ See Wickert, op. cit. (n. 32), 2248–53, where aequitas and iustitia are treated as essentially interchangeable.

⁶¹ So Mattingly (BMCRE II, xlviii; III, xxxv ff.; IV, lvii). Cf. Wallace-Hadrill, op. cit. (n. 57), for a more recent statement of this view along with a summary of the debate and earlier bibliography.

⁶² Wallace-Hadrill, op. cit. (n. 57).

⁶³ Wallace-Hadrill concludes his analysis of the Aequitas type with the suggestion that the type ‘conveys a message close to the hearts of the issuing authorities: that the administration of public finances is honest and specifically that the coin they issue is value for money’ (op. cit. (n. 57), 37). This may be the case, but as Wallace-Hadrill himself notes in the same article, emperors were held responsible for what was done under them (21, n. 8). The public finances fell under the authority of the emperor, and the emphasis on aequitas, regardless of the mint officials' putative attachment to this virtue, clearly redounded to the credit of the emperor himself.

⁶⁴ Mattingly, BMCRE III, xcv. R. L. Gordon, ‘The veil of power: emperors, sacrificers, benefactors’, in M. Beard and J. North (eds), Pagan Priests: Religion and Power in the Ancient World (1990), 199–231, discusses sacrificial imagery and the iconographic emphasis on the emperor himself in the act of sacrifice.

⁶⁵ Strack, op. cit. (n. 59), I, 75 ff.; II, 51 ff., 169–71; III, 25 ff., 36 ff., 115–24. See also J. Beaujeu, La religion romaine à l'apogée de l'empire I: la politique religieuse des Antonins (96–192) (1955), 281 ff., esp. 285–6.

⁶⁶ The reciprocal nature of pietas is stressed by R. Saller, Patriarchy, Property and Death in the Roman Family (1994), ch. 5, esp. 105–14.

⁶⁷ TLL VII.1, 1246–50, esp. IB (‘goodwill’ or ‘kindness’), IIA (‘generosity’), and I ID (‘leniency’). For a general treatment, see J. Gaudemet, Indulgentia Principis (1962). See also Cotton, H., ‘The concept of indulgentia under Trajan’, Chiron 14 (1984), 245–66,Google Scholar for the term's establishment in the ideology of the Principate under Trajan, with particular attention to its ‘paternalistic’ aspect.

⁶⁸ TLL VIII, 1650–2, esp. IA Ia (on public benefaction) and IA1 β for the particular sense of munificentia as generosity in the giving of public games.

⁶⁹ Cicero, for example, ascribes the deliverance of the Republic from the ‘great dangers’ of the Catilinarian coniuratio to his own providentia: ‘quod virtute consiho providentia mea res pubhca maximis periculis sitliberata’ (Cat. 3.14).

⁷⁰ Charlesworth, M. P., ‘Providentia and Aeternitas’, HTR 29 (1936), 107–32Google Scholar. See also Béranger, op. cit. (n. 53), 210–17, on the relationship between providentia and cura rei publicae. Because these applications of providentia depended on the foresight of the emperor himself, simple Providentia types must be distinguished from Providentia Deorum types, which have not been tabulated under this category.

⁷¹ cf. Sen., Ep. 94.26: ‘improbum esse qui ab uxore pudicitiam exigit, ipse alienarum corruptor uxorum’. For a general treatment of pudicitia with ample bibliography, see C. Micaelli's introduction to Tertullian's De Pudicitia in the Sources chrêtiennes series, vol. 394 (1993)Google Scholar.

⁷² Between A.D. 69 and 235 the type appears on coins with obverse portraits of Sabina (under Hadrian), Faustina II (under Antoninus Pius), Lucilla (under Marcus Aurelius), Crispina (under Commodus), Julia Domna (under Septimius Severus and under Caracalla), and Julia Maesa (under Elagabalus).

⁷³ So Strack, op. cit. (n. 59), II, 117–18, contra Mattingly (BMCRE III, cxxxi), who makes Pudicitia the emblem of religious sanctity.

⁷⁴ On virtus see J. Hellegouarc'h, Le vocabulaire latin des relations et des partis politiques sous la République (1963), 244 ff.; D. C. Earl, The Moral and Political Tradition of Rome (1967), 20 ff.; Galinsky, op. cit. (n. 32), 84.

⁷⁵ Two passages from Suetonius (Aug. 94.12; Ner. 25.2) and one from Eusebius (Vit. Const. 15.1) have been adduced to support imperial choice of types, but both authors may be assuming that what was done for emperors was done by them. The author of the De rebus bellicis includes suggestions for coin types in his advice to the emperor (3.4), but this only reveals one man's assumption that the emperor selected the types. Finally, a casual reference in Statius' Silvae suggests that under the Flavians the procurator a rationibus was responsible for the supply of bullion to the mint and the volume of coinage produced (3.104–5), but this is not the same thing as responsibility for the selection of types. None of this evidence is compelling, and we cannot even discount the possibility that it was still the tresviri monetales who chose the types (the college of tresviri monetales survived at least through the reign of Severus Alexander: CIL X.3850 = ILS 1181). See R. Wolters, Nummi Signati: Untersuchungen zur römischen Münzpragung und Geldwirtschaft (1999) 262–4, for a recent overview of the debate on the selection of types.

⁷⁶ As Wallace-Hadrill, op. cit. (n. 6), 86, notes, however imperial coin types were chosen, the understanding of the type, once the coin was in circulation, will have been exactly the same.

⁷⁷ Kloft, H., Liberalitas Principis, Herkunft und Bedeutung: Studien zur Prinzipatsideologie (1970)Google Scholar is basic. See also Veyne, op. cit. (n. 1), ch. 4; Millar, op. cit. (n. 1), 133–201.

⁷⁸ The Cyropaedia was allegedly read by Scipio Aemilianus (Cic, Tusc. 2.62, Q Fr. 1.1.23) and by Caesar (Suet., Iul. 87). For the influence of the Cyropaedia on Roman Republican political thought, see briefly Kloft, op. cit. (n. 77), 20–2; Rawson, E., ‘Scipio, Laelius, Furius and the ancestral religion’, JRS 63 (1973), 164–5Google Scholar (= Roman Culture and Society (1991), 86–7).

⁷⁹ Kloft, op. cit. (n. 77), ch. 3.

⁸⁰ Suet., Nero 10.1: ‘Atque ut certiorem adhuc indolem ostenderet, ex Augusti praescripto imperaturum se professus, neque liberalitatis neque clementiae, ne comitatis quidem exhibendae ullam occasionem amisit.’

⁸¹ See Gnecchi, op. cit. (n. 45), 354–59, for a list of the emperors on whose coinage the Liberalitas type appears.

⁸² So Kloft, op. cit. (n. 77), 79.

⁸³ Tac., Hist. 1.30.1: ‘Falluntur quibus luxuria specie liberalitatis imponit’, ‘Those for whom (Otho's) extravagance takes on the appearance of liberalitas are deceived’; 1.37.4: ‘falsis nominibus … parsimoniam pro avaritia … appellat’, ‘by mistaken names he (sc. Galba) calls it “thrift” instead of “greed”’; cf. 1.18.3 and 1.38.1 on Galba's stinginess.

⁸⁴ On the derivative nature of Vespasianic coin types, which borrow types and motifs even from ‘bad’ emperors such as Nero and Vitellius, see Buttrey, op. cit. (n. 19).

⁸⁵ Plin., Pan. 3.4, 25.3, 25.5, 27.3, 28.4, 33.2, 34.3, 38.2, 38.4, 43.4, 51.5, 86.5.

⁸⁶ CIL VI. 1492 = ILS 6106: SECUNDUM LIBERALITATEM EIUS. Two other alimentary inscriptions commemorate imperial liberalitas: CIL XI.5956 (Pitinum Mergens, A.D. 138–61) and CIL XI.5395 = ILS 6620 (Asisium, undated). Note that a range of virtues could be ascribed to Trajan in the alimentary inscriptions, such as indulgentia (CIL IX. 1455 = ILS 6509, Ligures Baebiani), munificentia (CIL IX.5825, Auximum), and providentia (CIL VI. 1492 = ILS 6106, Rome). See discussion in Woolf, G., ‘Food, poverty and patronage: the significance of the epigraphy of the Roman alimentary schemes in early imperial Italy’, PBSR 58 (1990), 224–5Google Scholar.

⁸⁷ CIL VI.955 = ILS 286. On Trajan's building in the Circus Maximus, cf. Plin., Pan. 51.3–5; Dio 68.7.2; Paus. 5.12.6. Fell, M., Optimus Princeps? Anspruch und Wirklichkeit der imperialen Programmatik Kaiser Trajans (1992), 52–61Google Scholar, discusses the use of the title ‘optimus princeps’ before A.D. 114 (when it became an official cognomen).

⁸⁸ Veyne, op. cit. (n. 1), 621 ff. (‘Les bienfaits du prince’), esp. 638–42 on public building.

⁸⁹ CIL VI.8.2 40493: [… NERV]AE TRAIAN[ANO]… / […] LIBERALISSIMO […]. The two fragments of this inscription, both found in the Markets of Trajan, were originally published separately by R. Paribeni in NS 1933, 503, n. 221 and 518, n. 283. The fragments were put together and the inscription properly restored and published by Alföldy, G. in CIL VI.8.2 (1996)Google Scholar. This inscription now represents the earliest evidence for the use of the superlative epithet ‘liberalissimus’, next attested under Hadrian: AE 1969/70, 167 (Beneventum, A.D. 125–126); CIL VIII.2534 (Lambaesis, A.D. 117–138).

⁹⁰ On superlative epithets in honorific inscriptions to the emperor (down to Commodus), see Frei-Stolba, R., ‘Inoffizielle Kaisertitulaturen im 1. und 2. Jahrhundert n. Chr.’, MH 26 (1969), 18–39Google Scholar.

⁹¹ See A. Wallace-Hadrill, Suetonius: The Scholar and his Caesars (1983), 142–74, for a discussion of virtues and vices in Suetonius, esp. 166–71 on imperial expenditure; see also Kloft, op. cit. (n. 77), 155–6.

⁹² CIL VI.967 = ILS 309: HAC LIBERALITATE (Rome, A.D. 118); CIL VI.972: INDULGENTIA ET LIBERALITATE EIUS (Rome, A.D. 133); CIL XIV.95: INDULGENTIA ET LIBERALITATE EIUS (Ostia, A.D. 133); CIL XIV.4235 = ILS 318: [LIBE]RALITATES (Tibur, A.D. 136); CIL XIV.2460: [EX LI]BER[ALITATE] (Castrimoenium, A.D. 117–138); AE 1991, 694: LIBERALIT[ATE] (Caesarea, A.D. 117–138). For the use of the superlative epithet ‘liberalissimus’ in Hadrianic inscriptions, see above, n. 89. One inscription set up after A.D. 138 attributes liberalitas to Hadrian: CIL X.5963: PROFUSA LIBERALITA[TE] (Signia, post-A.D. 138).

⁹³ On the attachment of serial numbers to the Liberalitas type and the resulting ‘trivialization’ of personal generosity as a moral concept, see Kloft, op. cit. n. 77), 158–9.

⁹⁴ Bourdieu, P., Outline of a Theory of Practice, transl. R. Nice (1977), 164CrossRefGoogle Scholar.

⁹⁵ As Howgego, op. cit. (n. 4), 71, writes, the lack of good evidence for many of the standard questions ‘has rendered the debate somewhat sterile’. For a recent and extensive survey of these questions and the attendant debates, with ample bibliography, see Wolters, op. cit. (n. 75), 255–339.

⁹⁶ The Reka Devnia hoard alone constitutes over one half of all the silver coins in the sample. The overwhelming size of this one hoard might be suspected of having an inordinate effect on the overall hoard profile, but a few test cases show that the composition of Reka Devnia, although far larger than other hoards, is remarkably similar to that of smaller hoards. The variable annual percentages of Trajanic denarii and denarii of Severus Alexander in the Reka Devnia hoard, for example, closely mirror those of the via Braccianese hoard in Rome (6,409 denarii, unpublished), and the Elveden hoard in Britain, respectively (Duncan-Jones, op. cit. (n. 5), 114, fig. 8.2; 116, fig. 8.3). Caracalla's antoninianus, introduced in A.D. 215, represents 11.5 per cent of Caracalla's silver at Reka Devnia, and 12.2 per cent of his silver at Colchester (Duncan-Jones, op. cit. (n. 5), 138). At least three other hoards, then, are demonstrably similar in composition to Reka Devnia, and the combined evidence of these test cases suggests indeed that Reka Devnia ‘should provide something like a directory of central silver coin-output’ (Duncan-Jones, 133).