Fig. |. Bronze head of Augustus from Meroe. Before 25 B.C. British Fig. 2. Augustus, head of statue from Prima Porta. After 20 B.C. Vatican, ‘Museum. Cast, Oxford. Braccio Nuovo. Cast, Oxford.31 Typology and diversity in the portraits of Augustus R.R.R. Smith including discussion of D. BOSCHUNG, DIE BILDNISSE DES AUGUSTUS. DAS ROMISCHE HERRSCHERBILD 1.2 (Gebr. Mann, Berlin 1993). xv + 237 p., 239 pl., 9 Beil. ISBN 3-7861-1695-4, In the history of the Roman imperial image Augustus’ portraits occupy a special position in terms both of self-representation and of their reproduction. Augustus was the first Roman leader whose image was widely disseminated and replicated, and we have far more of his sculptured portraits than of any other emperor. They have been much studied both in older and more recent studies, from which a near-consensus has emerged on the level of both facts and interpretation. There were three main Augustus portraits which presented him first in the manner of a Hellenistic-style leader, later in the classicising manner of Polykleitan art. Boschung’s excellent new book in the revised Das rémische Herrscherbild series! now pro- vides the complete evidence, collecting and illustrating all the surviving versions of Augustus’ portraits in one place — some 210 sculptured heads, busts, and statues. The picture of Augustus’ image that he presents is inevitably more complicated than that based on partial collections of the material, but, while there is much to adjust in detail and some new ideas to propose, by and large Boschung’s volume provides the underpinning for the consensus view, both for the isolation of the three portraits and their meaning Boschung is primarily concerned with the appearance and details of the main portrait conceptions or types, and his book is a state-of-the-art application of the typological method in the study of imperial portraits. Since it stands very close to the material, it may be useful to provide some preliminary sketch of this method and why it is important. It is easy to misunderstand detailed copy comparison in the study of imperial portrait sculpture (and ideal sculpture) as some kind of misguided, positivist exercise, out of touch with ancient realities and complexities. Some English-speaking scholars seem, in general terms, sceptical of how imperial portrait types worked and behaved, what they were for, and why they matter. Detailed copy comparison that involves the counting and verification of the components of repeated hairstyle systems can be off-putting, but some engagement with the phenomenon and its modern scholarship is necessary before it can be dismissed. ‘Type and replication: identity and recognition Augustus’ portrait image stands at the beginning of what became regular imperial portrait practice, that is, the creation of successive, centrally defined, three-dimensional images (heads or busts) which were each then made available around the empire and assiduously replicated in local contexts as part of a whole physical apparatus of loyalty-demonstration (figs. 1-2). Other image-types, for example, classical statues of divinities and athletes, such as the Doryphoros and Pheidian Athenas, or portraits of culture heros, such as Aristotle and Demosthenes, had already in the late Republic been the object of organised replication for the Taalian villa market, so that some centres would already have been familiar with and have had practice in the reception and replication of a given sculptured image. Augustus, however, remains the first living subject whose image went through this system. The creation and appearance of the centrally defined portrait image (the original model, the portrait type), its 1 The first volume of the revised series to appear was D. Boschung, Die Bildnisse des Caligula, Das riomische Herrscherbild 1, 4 (Berlin 1989). The new direction of the series was discussed by the present writer in JRS 82 (1992) 270-73.32 R.R.R. Smith dissemination in three-dimensional models (plaster, bronze, marble), and its reception, replication, and adjustment to its local context are the subject of modern typological studies of imperial portraits. No texts speak about the vital first two stages, of creating and distributing the portrait models, and the whole phenomenon is deduced entirely from archaeology, from the random selection of local replicas, versions, and variants that survives to us. Good recent work has put the whole subject on a much surer scientific footing, with rules of method and evidence that are quantifiable and verifiable? While it is important to distinguish the practice of the modern method from the ancient processes of the phenomenon (some critics see only the positivist modern practice), they do share a basic common purpose, that is, recognition and identification. The modern practice of the typological method consists of a detailed comparison of all the surviving replicas and variations of a given image. The aims are firstly to establish the appearance, design details, and style of the original model from the closest replicas; secondly, to assess the range of adjustment and variation of the other versions; and thirdly, to exclude those portraits that fall outside that range, that did not include (or deliberately avoided) any of the tell-tale elements that would connect them with the type. The result is a list or ‘recension’ of copies all certainly after one model, and so (obviously) all representing one subject. That subject may then be named from an inscribed example, usually on coins. The identification that the method provides is not the name — that is something separate — but the membership of a particular piece among the versions of the type. The standard to which each piece is being held is higher than a mere portrait resemblance in the modern sense. All elements of posture, physiognomy, and hair are compared, but usually the easiest elements of the design to count and compare precisely and objectively are the various stylised patterns that represent hairstyles. These natural and casual-looking hair patterns were clearly defined with great care and precision in the central models, and the local copies take as much from them as they want, starting with the ‘individually’ distinguished hair schemes over the forehead which often were made to act as identifying markers of early imperial portrait types. This method of copy comparison and counting of lock formations establishes that large groups of seemingly dissimilar images do in fact all refer at some distance, and with varying levels of local intrusion, to a common model or type. The method shows that there is not a continuous spectrum of variation within the various versions of one emperor's portraits, but rather a variation around a set of repeated design elements. Despite the accidental semblance of ‘real’ variety and change, the point of reference is not life, the real changing appearance of the subject, but another image. This is primarily, then, a method to isolate and identify the portrait heads of a given emperor and to exclude those private portraits (or portraits of other imperial figures) that have merely a portrait resemblance in the modern sense. It has been developed as a corrective to the old optimist tendency, especially in the study of individual pieces, that wanted to detect portraits of emperors rather than unnamed private portraits. What is required of a version after a given type is repetition of a sufficient number of key features, especially verifiable and countable details. A general resemblance is not enough — ot even necessary. It is one of the interesting and important paradoxes of imperial portraits, demonstrated by the typological method, that two portraits with a very similar appearance can be shown to represent different subjects, while two that look very different can be shown from their repetition of certain key details to have been intended as versions of the same model 2 Among important studies are: K. Fittschen, “Zum angeblichen Bildnis des Lucius Verus im Thermen- Museum,” Jaf 86 (1971) 214-52; M. Bergmann, Marc Aurel (1978); and K. Fittschen, P. Zanker, Katalog der romischen Portrits in den Capitolinischen Museen und den anderen kommunalen Samimlungen der Stadt Rom I. Kaiser und Prinzenbildnisse (Mainz 1985).Typology and diversity in the portraits of Augustus 33 (and so obviously to represent the same person). So it can be shown, for example, that the head of ‘Lucius Verus’ in the Terme Museum or the attractive bust of ‘Hadrian’ from the Olympieion in Athens represent neither L. Verus nor Hadrian but were contemporary private portraits.’On the other hand the unusual heads of Marcus Aurelius from Gerasa and of Hadrian from Lepcis, although they do not much resemble those emperor's portraits, are certainly versions based on one of their respective main types. The method can thus sort out the emperor's portraits from the mass of similar portraits influenced by his image and from which his image emerged. It is possible that there were also ‘non-type’ portraits of the emperor, that is, portraits made without any reference to a centrally defined model that were identified by inscription or context. There is, however, no way of identifying them archaeologically, of separating on any but subjective criteria such ‘non-type’ imperial portraits from contemporary private portraits. And since no such portraits survive with other means of identification, it remains an entirely putative category. The ancient purpose of making, distributing, and replicating such centrally defined types was simply to ensure recognizability, to create a recognizable identity legible to the ancient viewer. Roman emperors had no distinguishing dress, attributes, or statue types to set them apart from others. Quite the contrary, the imperial image worked hard to express the idea that the emperor was not in principle a different kind of person from other citizens. In theory, a viewer was meant to know that he was looking at the emperor's image simply because he recognized it as that person whom he knew also happened at that time to be emperor. In practice, of course, other aspects, such as scale and context, might also tell him, but in theory he should recognise the emperor as he would any other citizen's imago. The requirements of this elaborate visual facade, so pronounced in the portraits of the ‘bourgeois'-style rulers of the 2nd c., were what made the ‘system’ of typological reproduction of the Roman imperial image so important. Without it, the only part indicating ‘emperor’ would be elided, and the emperor's portrait would sink into the background of contemporary self-representation. Carefully defined and individualised types presented a sharply distinguished identity for each ruler, and if properly replicated would secure his recognizability. This portrait ‘system’ was at its loosest under Augustus precisely because he went the furthest in creating a distinctive style recognisable as ‘imperial’ without further help. ‘The same basic purpose lay behind the phenomenon of replication in the marbles used to decorate Italian villas: easy recognition. These classical-looking marble figures replicated, in a different medium and context, earlier images — such as the Doryphoros or the Demosthenes — chosen from an agreed cultural canon, whose best known, ‘classic’ images came to be readily recognised through frequent repetition of a restricted repertoire. These images were exemplary specimens of an earlier, foreign high culture, and derived their meaning for villa owners as approved and recognised representatives of that culture. Ancient villa-owners might have been surprised, even offended, at the Roman-ness now so often identified in modern scholarship as one of the most important characteristics of these marbles. That these expensive cultural ‘ormaments should turn out to speak more loudly of a new and different aesthetic will than of their owners’ Hellenic culture, they might perhaps have found irksome. We can sometimes detect adjustment and contemporary input in these marbles, but it may be doubted if it was ever an intended part of the patron’s interests. ‘An important theoretical difference from imperial portrait copies is that replicas of the Doryphoros or the Demosthenes referred to a known and fixed image from a bygone era. Augus- tus’ portraits, on the other hand, although in terms of practical manufacture they referred to 3 Terme ‘Lucius Verus’: Fittschen, Jdl 86 (1971) 214-52. Olympieion ‘Hadrian’: P. Zanker, Provinzielle Kaiserportrits: 2ur Rezeption der Selbstdarstellung des Princeps (Munich 1983) 14, pl. 4.3-4 4 Zanker (oupra n. 3) 18-20, pl. 9 (Hadrian), 39, pl. 23.1 (M. Aurelius).34 R.R. R. Smith other images, to pre-defined models, in theory they referred to the appearance of the living emperor. This means that, provided identity and recognizability were preserved, replication might be much freer, especially since in practice other elements, such as context, scale, and inscriptions, would also usually help with identity. Marble versions of the Doryphoros and the Demosthenes, on the other hand, would, beyond a fairly narrow range of variation, cease to be satisfactorily recognisable. This is in fact clear in the surviving marble record where the manner of replication of Augustus’ portraits, for example, is much more varied than that of the Doryphoros or the Demosthenes. It is all the more striking because these old statues were reproduced over a much longer period. In the 2nd c,, imperial portrait replication became ‘tighter’, especially in Italy. But in the early empire, when Augustan and Julio-Claudian style were more distinctively ‘imperial’, an appearance of difference in the emperor's portraits from the central model may have been felt to evoke usefully the illusion that it referred to the real living emperor. Although they had a theoretically different relationship to their model, the purpose of replication in both the imperial image and villa marbles was in each case the same — easy recognition of the subjects. Local reception, variation, and adaptation The surviving examples of imperial portraits are small parts of the once massive approba- tion of the emperor organised from below by communities and individuals, eager to demonstrate their loyal adherence to the imperial family. Imperial models were made available to local centres, which employed them to make statues and busts. There was a wide range of factors that could change the appearance of a local version after a given, centrally defined imperial portrait type: scale, material, setting, posture of statue or bust, copying technique (measured, free-hand, or entirely from memory), local workshop preferences, local stylistic preferences — all of which could result in adjustment and re-styling. The purpose was never any kind of searching after an originality for its own sake, but rather appropriate adaptation to local circumstances — the setting of the image, techniques of the executing workshops, and conceptions of the local community of viewers for whom the image was intended. There was here a clear motive for changes and ‘improvements’ to the model, to adapt it to different local ideas of the emperor's rdle. Such a motive did not exist in the reproduction of old images such as the Doryphoros and the Demosthenes which were made for a much narrower range of contexts and audiences and with an unchanging conception of what the model stood for. Obvious, structural ‘improvements’ would be to little purpose. While the Demosthenes always remained a symbol of classic rhetoric, Augustus’ portraits had to represent the emperor in some contexts as civilis princeps, in others as ‘saviour of the oikoumene.’ Given that there were centrally provided portraits of the emperor (and no other model of explanation will account for the observable phenomena) and that they could tolerate a wide range of variation in their various local deployments, it becomes a question of scholarly prefer- ence and interest whether the appearance of the original central model or its local variants are more interesting or important. It depends what questions one is asking. In terms of understanding imperial ideology at the centre, the former is more important. In terms of understanding individual pieces, the second perspective is the more important — though of course the second perspective depends on the first. As with the villa marbles that reproduce classical statues and portraits, local and contemporary adjustment can be detected and assessed only against a more or less precise mental image of the original. The distinction in this area is between, on the one hand, the character of the earlier statues and what the closest replicas can tell us about them, and, on the other hand, the adaptations and alterations that copyists might make — in material, scale, style, and expression — to make them more appealing for the villa market. It is Boschung’s interest and concem to collect all the versions of Augustus’ portraits and to sort and analyse them in such a way that the clearest picture of the central types on which they all more or less depend emerges. Other aspects of local reception are not ignored, but theyTypology and diversity in the portraits of Augustus 35 are not the primary focus. They are used to establish as far as it is possible the original highly detailed, ideologically-charged portrait icon — something like a visual text, against which the rest of the ragged ‘manuscript tradition’ may be assessed. This is the concern of the Herr- scherbild series. A pioneering study such as P. Zanker’s on provincial imperial portraits puts the other perspective in the foreground: local adaptation to provincial ideas and techniques. Augustus’ portraits: explaining ersity Any doubts that we can say or know anything useful about the original models of surviving imperial portraits or villa statues should be dispelled by recent detailed research in this area, such as that of Boschung. The replica record of Augustus’ portraits is amongst the most varied of any emperor and more varied than any replica series of classical ideal sculpture, but, as Boschung demonstrates, while we can argue about particulars and details, the method clearly works. Any theoretical counter-hypothesis would need to propose a comprehensive model that will account for the beguiling combination of diversity and recognizable typology within this body of well studied material. This is an issue with wider implications than the portraits of Augustus. Type, replication, and local adaptation were central phenomena in a wide range of ancient products. The diversity of the surviving versions of Augustus’ main portrait types perhaps requires special or additional explanation. In the early days of the imperial portrait ‘system’, one might have expected such flexibility and tolerance. Reproduction certainly became more exact and systematic in the surviving imperial portrait copies of the 2nd c. A.D. This is, however, not a full explanation because precise, scale replication had already been achieved as a matter of regular production for some sections of the late Republican villa market. Other factors were probably more important. Firstly, Augustus was the longest reigning princeps and had an impressive, empire-wide impact. His portraits cover the widest range of place and time of any emperor. Secondly, most versions of his image employed for more than 40 years a single unchanging portrait type. And thirdly, that one portrait type incorporated an unusually elevated personal style that could retain an Augustan identity through quite extensive physiognomical and stylistic adjustments. In all these elements — length of use, unchanged type, and personal style — Augustus’ portraits were unique. These factors created the variety that causes formidable problems of typological classification. Although it does not cast doubt on the ‘system’, the variety requires sensitive and patient handling, such as Boschung gives it. Augustus’ portraits: research history Modern study of Augustus’ portraits began with J. J. Bernoulli's Rémische Ikonographie IL. 1 (1886) which collected a lot of material, identified a range of sculptured portraits, and exclud- ed others. Bernoulli recognized some of the importance of the arrangement of signature locks on the forehead, but the portraits were still grouped by different ages and phases of the emperor's life to which the images could not really be made to fit. The forehead lock formations played a prominent réle in O. Brendel’s dissertation of 1928 which distinguished 5 types and their replicas, using stricter method and the terminology of original, copy, and version, borrowed explicitly from G. Lippold’s excellent study of copies after earlier Greek originals. Without illustration, Brendel’s publication did not have the impact it deserved. Its results were subsequently disputed and contradicted by various scholars, doubtful of the methodology. Two of Brendel’s types are now generally recognized to represent Augustan princes. 5 Zanker (supran. 3). 6 —. Brendel, Ikonographie des Kaisers Augustus (Diss. 1928, published Niirnberg 1931); G. Lippold, Kopien und Umbildungen griechischer Statuen (Munich 1923).36 R. R. R. Smith More recently, a third full corpus planned by P. Zanker resulted in important preliminary studies; Der Actium-Typus (1973) and the wide-ranging exhibition catalogue prepared with K. Vierniesel, Die Bildnisse des Augustus (1979),” which outlined the now-agreed typological sequence and set the images of Augustus in a wider context of public display and of the portraits of earlier and later imperatores. The latter remains the best, most accessible treatment of many aspects of the subject not covered by the Herrscherbild series. Discussion, refinement, and some disagreement followed in detailed studies by F. Johansen, U. Hausmann, A.K. Massner, B. Schmaltz, G. Grimm, K. Fittschen and again by P. Zanker in the early entries of the first volume of the Capitoline portrait catalogue.’ Meanwhile, in 1986, Zanker had transferred to Boschung the wide-ranging collection of material for the planned new study, and the present Herrscherbild volume is the result of Boschung’s working-up of this data and photographic collection (xii-xiv, 3-4). It is on its complete documentation and consistent illustration of every known and published piece that the value of this volume depends. Only with such a dossier of photographs taken from consistent angles can the sophisticated analysis of type, replica, and adaptation be verified. Boschung’s typology Boschung’s work is a typological study aimed at the reconstruction of Augustus’ centrally designed portrait images. It is not about Augustus’ portraits in their local context, except in as far as they help understand the types. The goal is deliberately limited. It is not concerned except incidentally with the wider deployment of Augustan imagery in different statue types, groups, narratives, cameos, reliefs, or coins. It is concerned simply to establish the visual texts of the main successive changes in Augustus’ public image. Boschung finds the same three types, as before but gives them a much firmer grounding and shows on what they are really based. Many additions, corrections, improvements, and original points are made along the way. The most important novelties are the suggested isolation of two new early portrait types attested in a few versions to be dated in the late 40s B.C., before the period of the first of the three agreed main types. Chapter 1 sets the scene of previous research and describes assumptions and methods. Chapts. 2-3 are the heart of the book: the establishing of the types and their replicas — the three agreed types and two new ones (chapt. 2), and their relative chronological sequence (chapt. 3). Chapters 4 and 5 discuss more summarily the chronological and regional dimensions, that is, the dating and the distribution of the individual replicas. Chapter 6 concludes with good discussion of the sources on Augustus’ appearance and some listing of Augustus statues known from literature and inscriptions (a forthcoming study on Julio-Claudian portraits in their contexts is promised). What follows here attempts to describe the important results of chapts. 1-3, and then to raise some general issues in a discussion of the topics of chapts. 4-6. 7 P. Zanker, Studien zu den Augustus-Portrits I. Der Actium-Typus (Gottingen 1973); K. Vierneisel, P. Zanker, Die Bildnisse des Augustus (Exhib. Cat. Miinchen 1979). 8 F Johansen, “Le portrait d’ Auguste de Prima Porte et sa datation,” Studia romana in honorem P. Krarup sepluagenarit (1976) 49-87; U. Hausmann, “Zur Typologie und Ideologie des Augustusportrats,” ANRW T1122 (1981) 551-98; A-K. Massner, Bildnisangleichung. Untersuchungen 2ur Entstehungs- und Wirkungsgeschichte des Augustusportrits (43 v.Chr, 68 nChr.}. Das romische Herrscherbild 1V (1982); F. Johansen, “Augustusstatuen fra Primaporta og to portraetter af Primaportatypen i Glyptoteket,” ‘MeddeiGlypt 40 (1984) 97-118; P. Zanker, in Fittschen-Zanker I (supra n. 2), nos. 1-9; B. Schmaltz, “Zum Augustus-Bildnis Typus Primaporta,” RémMitt 93 (1986) 211-243; G. Grimm, “Die Portraits der Triumvimn C. Octavius, M. Antonius und M. Aemilius Lepidus. Uberlegungen zur Entstehung und Abfolge der Bildnistypen des Kaiser Augustus,” RomMitt 9 (1989) 347-64; K. Fittschen, “Die Bildnisse des Augustus,” in G. Binder (ed), Seeculum augustim III (Darmstadt 1991) 149-86,Typology and diversity in the portraits of Augustus a7 Boschung defines his terms in chapt. 1: type, replica, and replica group. A type is a central- ly defined model; replica is used for any copy or version after a type; and a replica group is secondary series of portraits that follow a common model that is itself a variant of one of the main types — such variants too can be shown to have been widely disseminated. Speculative argument addresses the nature of the centrally defined model: perhaps a clay or wax model, from which imperial workshops might make ‘original copies’, that is, direct copies after the model (in bronze or marble, I presume), which might then be exported and made available to other workshops in Rome, Italy, and provincial centres. Boschung argues forcefully against the idea of one famous statue of Augustus, a well-known dedication in Rome, serving as a model for replication. Only the head of early imperial portraits (often in the 2nd c. the bust too) was replicated, and it seems unlikely that the model was ever ‘published’ as an original public monument, Unlike copies of ideal statues, there was no known ‘original’ to which the copies referred. It might be added that this ‘concealed original’ suited better the fictional premise that all the portraits referred to the living Emperor himself, and that the model had to be a head capable of reduction, magnification, and adaptation to busts, statues, relief, coins and ‘cameos. Boschung only here (6-8, with notes) provides a brief account of the variety of statue formats that are found with Augustus’ portraits. The first portrait type discussed is what Boschung calls the ‘Alcudia’ type (formerly called the ‘Actium’ or ‘Octavian’ type), agreed to be Octavian’s portrait type of the 30s B.C. and found mainly in the west (11-22). There is a relatively coherent core group of c.10 replicas, the most detailed, best-looking, and most Hellenistic of which is the head in Alcudia (cat. 6). Like Zanker, Boschung argues for seeing this type as the portrait of Octavian also used on the Divi Filius sestertii, as well as on the later ‘Octavian’ series of the 30s. This would place the creation of the type as early as c.40 B.C. The type continued in use for a long time and was mixed with, and ‘calmed’ under the influence of, the classicizing Prima Porta type. Of such later versions, Boschung identifies another c.18 examples — the best, for example, being the new equestrian bronze portrait in Athens (cat. 7). Genuine typological problems occur where local sculptors reworked the model, perhaps from memory, combining key elements of different types. The well known bronze portrait from Herculaneum carrying a thunderbolt (cat. 15) and the uncharacteristically vigorous bust from Fondi (cat.16) might as well be classified under reworkings of the Prima Porta type as under reworkings of the Alcudia type. It will be seen that Boschung dispenses with the inherited nicknames of Augustus’ main types, and calls each of them after what he regards as the best surviving replica. This practice has an old pedigree in the Herrscherbild series, but is cumbersome and results in important types receiving such names as ‘The Type of Louvre MA 1280’. The practice also has the unfor- tunate effect of promoting unduly one particular head as the best copy against which others are measured. In fact they are all more or less adjusted versions, adapted in some way or another. ‘The lead copy is more a good point of departure than a substitute for the model against which to assess other versions — it should be under as much ‘reception’ scrutiny as the rest. ‘Two new types are next isolated, each in only two or three replicas, which, Boschung later argues (61-62), belong in a very early period of experimentation with Octavian’s image, between Caesar's death and the creation of the Alcudia type, which then drives the two earliest types from the field. Boschung is first at pains, however, to establish the two new types on the typological analysis of the replicas (23-26), quite independently of such historical reconstructions. More on this below. The type often called the Forbes type (after a head in Boston, cat. 34), is re-named the type of Louvre MA 1280, a head which, according to Boschung, is its best exemplar (cat. 44). This is 9 Zanker, in Fittschen-Zanker I (supra n. 2) no. 1, there revising some views expressed in Der Actium- Typus (supra n. 7).38 R. R. R. Smith clearly a separate type with distinctive short forehead hair, known in some 15 copies and as many variants, which Boschung divides into two later and separate editions or strands that are replicated independent of the first centrally defined type. He calls these two strands the ‘Stuttgart replica series’ (11 copies) after a bust in Stuttgart (cat. 58), and the ‘Copenhagen 611 series’ (3 copies) after a fine head from Sardis now in Copenhagen (cat. 60). There are formidable problems here due to the concurrence of the type in the workshops with the Prima Porta type. To many the type has seemed, though clearly a distinct type, merely a hairstyle variant on the Prima Porta type, to be dated sometime after it but before 13-9 B.C. when the new type was used for the Augustus on the south frieze of the Ara Pacis (cat. 56). Boschung argues, however, that the few best copies show a distinct, more Hellenistic style, on the basis of which he would place it between the Alcudia and Prima Porta types, that is, say 30-27 B.C. Most other copies show influence of the Prima Porta type, on this theory, because the type was not in use for long before the Prima Porta type was issued. The Stuttgart and Copenhagen series would then be later editions, of, say, the 10s B.C. Much here is very speculative. ‘Of Augustus’ main type, known after its best copy, the head of the Prima Porta statue in the Vatican (fig. 2), Boschung lists some 147 copies and versions. It is the most easily recognized of Augustus’ portraits, and the simplified signature pattern of claw-locks over the centre of the forehead and the calmer, classical-looking physiognomy are its hallmarks. It is now generally agreed to have been created for the settlement of 27 B.C. and to embody and express visually the title ‘Augustus’. This date, compared to the dates of the other types, is relatively secure. The fine bronze head of Prima Porta type from Meroe now in the British Museum (cat. 122: here fig. 1) is dated by its context before 25 B.C.,19 and Boschung argues persuasively (with the aid of helpful drawings of the hair schemes) that the cistophori of ¢.27-25 B.C. have Prima Porta type hair. Normally the earliest dated coins employing the portrait type in question give the date, or rather terminus ante quem, for its introduction. But the changes in Augustus’ portraits were generally too slight and located too much over the front of the forehead for the die- cutters to pick them up. Out of the 147 listed versions of the main type, Boschung finds only a dozen close replicas (38-47). First there is a core group of 5 heads (the high quality heads in Chiusi, cat. 86; Merida, cat. 130; Syracuse, cat. 195; and Vienna, cat. 206; and the head of the Vatican statue from Prima Porta, cat. 171), which are the closest and most detailed in the reproduction of the hair system (though even these best copies are quite variable in physiognomy). Close to these stand a further 8 copies that are similar in all respects, only less detailed in the reproduction of all the particulars and secondary elements of the hair scheme. Well-known pieces here include the Copenhagen bust from the Fayum (cat, 112) and the head of the togate statue from the Via Labicana (cat. 165: below fig. 3). Only a few (some 14) of the many remaining versions are categorized into sub-groups that represent identifiable workshop series or replica strands (these, unlike the replica series inside the type of Louvre MA 1280, are not considered ‘strong’ or independent enough to be taken up as separate divisions within the catalogue). This leaves a very large number of versions (120) that cannot be usefully classified more precisely in relation to the main type and represent a very large range of chronological, regional, and contextual receptions and adaptations. This proportion of close replicas to adaptations and adjusted versions is strikingly low. The relative sequence of the three main portrait types is the subject of chapt. 3. It discusses what may be called the consensus sequence of Fittschen and Zanker, that is, (1) Actium-Aleudia (2) Prima Porta, and (3) Forbes-Louvre MA 1280, refutes other and more recent heterodox 10 On the context and date: D. E. L. Haynes, “The date of the bronze head of Augustus from Meroe,” Alessandria ¢ il mondo ellenistico-romano: studi in onore di A. Adriani (Rome 1983) 177-81.Typology and diversity in the portraits of Augustus 39 opinions (of U. Hausmann and B. Schmaltz),”" and argues for a new sequence of (1) Actium- Alcudia, (2) Forbes-Louvre MA 1280, (3) Prima Porta. The hard evidence (coins, the Meroe head in London, and the Ara Pacis) will really only allow secure dates for the Actium-Alcudia type (in the 30s) and the Prima Porta type (27 BC. or immediately after). The Forbes-Louvre MA 1280 type remains open to dispute as to its date (sometime before 13 B.C,). This chapter closes with a synthetic account of the origins of the three types and their historical interpretation, seeing a move from the image of an energetic ruler-dynast in the 30s to an elevated ‘art figure’ (Kunstfigur) modelled in classicising, especially Polykleitan, formal language (62-65). This is set out as an agreed and un-problematic position. It may be helpful to summarize Boschung’s main typological findings (the old names of the types are given in parentheses): 44-40 B.C. (1) Spoleto-Béziers (new) type: 2 copies (2) Lucus Feroniae (new) type: 3 copies 40-30 B.C. @) Alcudia (Actium, Octavian) type: 28 copies ‘The copies and versions continue into the Claudian period mixed with and influ- ‘enced by the Prima Porta type. Perhaps 10 of the copies/versions are pre-27 B.C. 30-27 B.C. (4) Lowore MA 1280 (Forbes) type 30 copies ‘Very few show no influence of the Prima Porta type, that is, it was soon replaced. by the Prima Porta type. The versions divide typologically into three ‘replica series’, as follows: Original ype (30-27 B.C 15 copies Stuttgart series (17 B.C.) 11 copies (one, Ara Pacis) ‘Copenhagen 611 series: 4 copies 27 BC. (6) Prima Porta (main) type: 147 copies The clear, succinct ordering of this material and its full documentation are a considerable scholarly achievement. The following remarks raise points of discussion and interpretation. New Octavian types? ‘The two new types, represented together by a total of only 5 heads, are perhaps given greater prominence than they deserve. They have their own sections in the catalogue and their own imagined historical scenario, The new types require careful testing. The new ‘Spoleto-Béziers’ type has only these two copies (from Spole- to and Beziers, cat. 1-2). They have only a two-lock difference in the forehead hair scheme from the Alcudia type, and they might (in my opinion) as easily be classified as a small sub-division within the Alcudia type. The second new type, the ‘Lucus Feroniae' type, comprises three heads: a head in London bought in Rome in 1888, a head found at Lucus Feroniae in 1959, and a veiled head in Venice (cat. 3-5). The Venice head still seems to me a fairly clear and recognisable version of the Alcudia type." OF the other two pieces, the London head is at best wholly over-worked, at worst an outright fake. Boschung defends the London head on the grounds that it shares a common and hitherto unattested model with the Lucus Feroniae head which was found only much later. The Lucus Feroniae head is, however, a rather summary, middle-grade piece and does not necessarily entail the same model as the London head. Both the hair at the back of the head and the forehead hair are sufficiently different between the two heads to allow them to be disconnected. One could be left with only the Lucus Feroniae head, which on its own would be much better classified simply as a weak version combining reminiscences of the Alcudia type and the Prima Porta type. The physiognonmy clearly draws on the former, while the weak and hesitant forehead hair dimly remembers both types. It was a version pethaps worked from memory. ‘The versions of both Boschung’s two new types show much less overall distance or difference from the Alcudia type than is routinely permitted between the varied versions of the main Prima Porta type. One should also ask why there are so few copies of both the new types. Ifthe early Alcudia type continued in use 11 Hausmann, Schmaltz, cited above n. 8. 12 So Zanker, Der Actium-Typus (supra n. 7) 17-18.40 R.R. R. Smith well into Augustus’ reign, why not also these two types when they are so little different in their presentation of Octavian as dynast. With more than 200 portraits surviving, it would not be unreasonable to expect more versions of the new types if they had ever been separate types current in local workshops. Dating the copies ‘The levels of precise and particular knowledge to which we can aspire in this field are often placed too high. Many things that rest on quite different levels of evidence and probability are presented as equally certain, and some good lines of argument are pressed too far to yield firm results where caution and flexibility would be more appropriate. One such area is the dating of individual copies and versions, the subject of chapt. 4. While opposed to received models of style development that simply will not work for this material or this period (p.67), Boschung still feels able to date a very large number of Augustus portraits to phases of ‘Augustus’ reign and the reigns of his successors, both with surprising precision and in a surprising number of cases. Naturally, where one can see the features of a later imperial portrait ‘infecting’ or intruding into an ‘Augustus head, one can say it belongs probably in that reign — as, for example, in the colossal Claudian- looking head from Cerveteri cat. 174) — but such cases are in the minority. Beside the rigours of typological identification, the simple listing of copies from different phases of Augustus’ reign and from later reigns, though succinct, looks subjective. Some of the results are intrinsically surprising — for example, the greater ‘number of Augustus heads said to be from Caius’ short reign than from the reign of Claudius. ‘The number of externally dated Augustus heads is not enough to allow such precision, especially given the wide range of their provenances, contexts, and ‘levels’. Only the following 7 portraits are roughly dated by external evidence: (1) the Meroe head, 30-25 8.. (cat. 122: above fig.); (2) the Prima Porta statue, after 20 BC. (cat. 171: above fig, 2); (3) Ara Pacis, 13-9 8. (cat. 56); (4) the Veli head, Tiberian (cat. 175); () the colossal hhead from Lepcis, soon after A.D. 23 (cat. 200: below fig. 5); (6) the seated figure from Lepcis, Claudian (cat. 30); and (7) the Herculaneum bronze statue, Claudian (cat. 15). The widely divergent opinions expressed about the dating of individual Augustus portraits by other scholars reveals the high degree of subjectivity involved (one need only compare, for example, the datings given by Boschung with those of Zanker in 1973 and 1985).! The precise dating of copies and versions after classical statues isa similarly subjective exercise easily exposed by its divergent results. A few can be dated, many more cannot, Such dating becomes more than a matter of a correct label for a particular piece when it supplies arguments about the dates of the central types. For example, the proposed date of the introduction of the type of Louvre MA 1280 (Forbes), that is, 30-27 ®, depends on the dating of the few copies that are closest in detail and most representative of the type. The eponymous head in Paris (cat. 4) especially has a softer, more ‘Hiellenistic’-looking handling, and so is said to be pre-27 8. (on another set of assumptions), others are more ‘classical’-looking, so later. But one may ask on what criteria (other than by assumption of the point at issue) cone can tell if the more Hellenistic-looking ones really are earlier. By ‘more Hellenistic’ is meant simply a more nuanced, more expressive physiognomical treatment, which, as many of the copies and versions of the other types show, could be preferred in some contexts at any date (for example, the portraits from Chiragan, cat. 29; Fondi, cat. 16; and Herculaneum, cat. 15) In other words, there is no guaranteed connection between accuracy of detail (that is, hair details) and chronological proximity to the model, or between accuracy of hairstyle and general stylistic fidelity. The Alcudia head (cat. 6) is the finest, most detailed exemplar of the Alcudia-Actium type. This version also happens to have a more Hellenistic posture and physiognomical handling, with carefully modelled accents, strong chin, and an expression of energy. But merely because in its hair the copy is among the most detailed, it does not mean that this copy is also necessarily the earliest. In this case the more Hellenistic character of the type is guaranteed by the related coinage (of the 30s B.C.) which presents an Octavian with dynamic hair and youthful energetic posture Provincial reception and typological classification ‘More flexibility might also help in the area of typological classification. More graded levels of certainty, ‘more explicit recognition of the great number of unknowable variables that make a local version different from its model would suit analysis of the material better. Typology is primarily a sophisticated tool of correct {identification and of exclusion of alien pieces from an emperor's portrait lis. Other connections, processes, and relationships of particular versions to their models at the level of the copying workshops are more hypothetical. 12 See Zanker, in Fittschen-Zanker (supra n. 2) p.2, nLTypology and diversity in the portraits of Augustus 41 ‘The distribution of the versions of the different Augustus types in Italy and the Empire, East and West, and the manner oftheir local reception is the subject of chapt. 5. This might in some ways have been the most important part of the study. Boschung is committed to the term ‘replica’ (really the wrong term) forall extant Augustus heads, and it leads him to place all the many and varied versions into perhaps too tight a framework. Copies which from this perspective perform badly are squeezed and argued into a more direct relation to the model than they perhaps warrant. The surviving versions seem also to be treated as somehow a good and typical cross-section of what once existed, rather than as the very small and random selection that they surely are. A larger, looser framework of reception and variation is perhaps needed to allow space for the surviving examples to breathe, to allow space for all kinds of variant reception (now missing) around the survivors and that would make more sense of those survivors. It is very difficult otherwise to trace a coherent picture of local reception in different provinces. In each area there were too many variables of context, statue type, setting, and local preference to allow the survivors to be truly representative. If Augustus could melt down 80 silver images of himself in the city of Rome very early in his reign (RG 24), how many other images were there in bronze and marble in the cities of the empire by AD. 60? We have just 210 — ‘many in one perspective, very few in another. It is one of the most important and striking results to emerge from Boschung's complete documentation of ‘Augustus’ main type that out of 147 heads less than one-fifth are copies which follow the type closely enough to be useful for its reconstruction. The other four-fifths are variants and adaptations, each quoting a few identifying signature-elements of hair and face, but otherwise more concerned with theit own local environment. This is striking because later, in the 2nd c, tie proportions would be roughly reversed. That is, in the portraits of, say, Hadrian or L. Verus, the majority of the survivors can comfortably be called replicas, while a minority embody interesting local adjustment and preference. Generally the most accurate imperial portrait replicas tend to be from Rome and central Italy, while the more adjusted and adapted versions tend to be from provincial centres. There are, however, no firm rules. ‘There are close copies in the provinces, where some are likely to be imports from Italy, but certainly not all. This very varied pattern of reception is of great interest in itself, and the only way to bring it out is to catalogue or list an emperor's portraits not by how well they perform typologically but by region. Individual pieces always have their most meaning in their local contexts. Listing by provenance and by province would preserve and represent this interesting pattern more faithfully. Boschung has brief discussion of the replicas by region (Rome, Italy, Western Provinces, Eastern Provinces, Egypt), and some of his plates bring out the regional dimension well, but it is lost entirely in the austere arrangement of the catalogue in which the pieces are entered under their type alphabetically by the city in which the pieces are now located. Asking only or chiefly about the relationship to the type allows an inventory of secure images of the emperor to be drawn up bbut it does less than justice to the character of much of the material. ‘The Augustus-Doryphoros theory Although it is not a major part of Boschung’s concern, other perspectives on the historical interpretation of Augustus’ main type might be discussed. All are agreed that the more dynamic, energetic style seen in the early portrait type of the 30s B.C. was replaced in the main type in 27 BC. by a calmer, plainer, more aloof style, created with a classicizing formal vocabulary. This interpretation can have a strong and a weak formulation, and it is the strong formulation that currently dominates. Boschung seems to give the strong version, but in some respects it is not quite full-strength: ‘The new design translates the physiognomy of the princeps into a classicistic formal language, derived especially from Polykleitan works. The ideal forms of the new type show the emperor ageless and removed from the accidents of mortality; they set him in line with the figures of myth. No longer is an individual representation ofthe emperor's face sought, rather it lends his physiognomy with the exemplary forms of Greek art. The personality of the emperor retreats behind an art figure (Kumsifigur), which through its aesthetic qualities must have seemed unassailable. (Boschung p.64) 14 For the replication of the main types of Hadrian and L.Verus: Fittschen in Fittschen-Zanker (supra n. 2) nos. 46-54 and 73, with Beil. 22-38, 50-55.42 R.R.R. Smith In the strongest formulation, Augustus’ main type is said to have been based not merely on Polykeitan works in general but in visually explicit terms on the Doryphoros of Polykleitos from which the ancient viewer would carry over specific meaning about the emperor's nature. ‘Through frequent repetition in scholarly books and articles, this has now become a thorough- going orthodoxy. The idea is seductive and gives a precise reference, a visual inter-textuality to Augustus’ image that its interpretation would otherwise lack. It also contains some observable truth — there are classical-looking elements in Augustus’ main portrait type. It has proved so appealing, so true to modern views of Augustus’ subtle and sophisticated appropria~ tion of earlier visual models that a collective scholarly vision has concentrated on those versions of his main type and those aspects of them which most support such a theory. But, we may ask, did Augustus really style himself after the Doryphoros? If true it would be, from many points of view, astonishing. ‘The Augustus-Doryphoros interpretation claims both philological and visual evidence. Both require careful handling. The philological argument employs a long chain of connections and assumptions. In answer to the question raised by the visual evidence — why should Augustus model himself after the Doryphoros? — a passage of Quintilian has been frequently quoted and taken to show that the Doryphoros was thought to embody the idea of the vir gravis et sanctus, a concept close to that of augustus (sanctus and augustus could, depending on the context, carry a similar semantic resonance). However precise this last connection (between augustus, ‘venerable’, and sanctus, ‘blessedi’) may have been in the minds of Roman viewers, the Quintilian passage will simply not bear the first connection, between the Doryphoros and vir sanctus, that the theory requires. In the passage in question, Quintilian is lambasting in highly coloured rhetorical terms the degradation of contemporary public-speaking exercises (declamationes) by unreal content and styles of soft, decadent, effeminate charm. He compares contemporary declaimers to slave- dealers who castrate boys to make them pretty. They smooth out strength and manliness in speech for a pleasing effect. ‘This debased eloquentia may please modern audiences by its effeminate and voluptuous charms, but I absolutely refuse to regard it as eloquence at all; for it retains in itself not even the slightest trace of a virile or uncorrupted man, not to mention of a serious and moral one (quae ne minimum quidem in se indicium masculi et incorrupti, ne dicam gravis et sancti viri astentet). When the most famous painters and statue makers wanted to carve or paint the finest possible bodies (corpora quam speciossima), they never fell into the error of taking as models some (sc. eunuchs like) Bagoas or Megabyzus, but righily selected as truly beautiful the well known Doryphoros, equally suited to the campaign and the palaestra, and the bodies of other war-like and athletic youths. Shall we then, who are endeavouring to mould the ideal orator equip eloquence not with weapons but with timbrels? (Quint. 5.12.21-22, translation adapted from Loeb Classical Library)! ‘The passage raises a lot of questions and is interesting from a number of points of view. We need only note the following points internal to the logic and argument of the text. (A) It is eloquentia itself, the main subject of the passage, that should show the mark of the vir gravis et sanctus, not the Doryphoros. (2) The Doryphoros is introduced merely as part of an analogy concerning the proper procedure in selecting models: for the representation of fine bodies one picks recognized images of athletic strength, not some eunuch. 3) The Doryphoros is not a unique touchstone: other figures of war-like and athletic youths 15 The clause of most interest in this context — quae ne minimum quidem in se indicium masculi et incorrupti, ne dicam gravis et sancti viri ostentet_ — is rather weakly translated there as follows: “it retains not the slightest trace of purity and virility in itself, not to say of those qualities in the speaker” This translation is, however, correct in the general sense of the passage in collapsing gravis et sancti into masculi et incorrupti.Typology and diversity in the portraits of Augustus 43 were also used as models by sculptors and painters. (4) While proper eloquence may be required to show signs of a vir gravis et sanctus, it is clear that in the corresponding part of the analogy what is required by sculptors and painters is not any moral characteristic but a fine (naked) body (corpora quam speciossima). (8) While sanctus often means ‘blessed, holy’ and so can intersect with augustus, ‘venerable, majestic’, it is clear that it is used in this passage in its moral sense (‘upright, blameless, virtuous, pure’: OLD s.v. sanctus 4). In the construction of the relevant sentence, sancti parallels incorrupti, as gravis parallels masculi. This passage cannot really do the work that the Augustus-Doryphoros theory assigns to it. In general, the Augustus-Doryphoros theory requires a very optimistic view of the general level of art-historical knowledge and sympathy on the part of the Roman viewer. We may legitimately ask the following questions, to which the theory requires unequivocal affirma~ tive answers. Would ancient viewers have recognized a head of the Doryphoros on its own? Would ancient viewers have recognized Polykleitan athletic heads as distinct from others of the mid-th c. — for example, Pheidian or Kresilan ones? And above all: Would the overt asso- ciation of the emperor with Greek athletics and art have been a welcome one to most Roman viewers? A negative answer to all these questions, especially the last, seems at least possible. Few probably would press the claims of the full figure of the Prima Porta statue (as opposed to its head) to have been explicitly or recognizably modelled on the Doryphoros statue. It ‘would in any case be of limited significance, since it is but one version not repeated in any other full figure of the emperor. The Prima Porta statue shares with the Doryphoros only the walking pose of the feet. The raised right arm and the armoured, clothed body would close off for the average viewer any ready association. (The idea of ‘average viewers’ may seem to beg, many questions, but to grasp the most widespread, middle-brow level of understanding at which public statuary in antiquity was aimed, it is a necessary if blunt tool.) It is the visual evidence rather of the head types that is the core of the Augustus- Doryphoros theory. The head of the Prima Porta Augustus and the head of the Doryphoros are frequently illustrated side by side, in three-quarter view, to show similarities. Similarities are clearest in the brows, posture, and hair length: Augustus’ main type has smooth sharp brows, it tus its head slightly to its right, and it has approximately the same length of hair. But many heads share such general features without evoking the Doryphoros in particular. Obvious differences can be as easily enumerated in both hair and face — especially if one juxtaposes the head of the Doryphoros with one of Augustus’ more ‘physiognomical’ portraits in straight frontal views (figs. 3-4). Augustus’ hair design is not axial-symmetrical, rather it has an asymmetrical design of large and small locks designed to look (for alll the ‘iconic’ effect it gathers through repetition) natural, casual, irregular, and portrait-like. The Doryphoros, on the other hand, has an axial hair system of flat, even, overlaid locks of a manner that came to form part of the vocabulary of the ‘ideal’, that is, of gods and heroes, as opposed to that of portraits, Augustus’ main type also had a marked ‘portrait’ physiognomy. The cheekbones are modelled and asymmetrical; the nose is long and curved, modulated in width, and the straight classical profile broken at the root of the nose. In construction the face is broad at the brow, narrowing to a shorter triangle below — quite unlike the rather formless oval of the Doryphoros’ face, with its deep rounded chin. Most strikingly ‘portrait’ and ‘Roman’ in ‘Augustus’ image are the ears protruding from the sides of the head. The ears of Greek portraits seem never to protrude in this manner. Such ears were part of the vocabulary of the late Republican self-image. They are a constant feature of the copies and variants of Augustus’ portrait and were clearly part of the model. ‘One could choose to emphasise these ‘portrait’ elements that put distance between Augustus’ image and images such as the Doryphoros, instead of the rather general similarities. And one might argue that, whatever effect of impassive calm, serenity, or majesty that Augustus’ mainFig. 3. Augustus, head of togate statue from Vii Fig. 4. Doryphoros, head of bronze herm from Fig. 5. Augustus, head from a colossal acrolithic sta Labicana, Rome. Musco Nazionale Romano. Photo: Herculaneum. Naples, Museo Archeologico Nazio- _tue, from Lepcis. Tiberian, soon after A.D. 23. Tri- DAI Rome Neg. 57.1180. nale. Photo: DAI Rome Neg. 64.1805. poli, Archaeological Museum. After Boschung pl. 118Typology and diversity in the portraits of Augustus 45 type was designed to carry, use of these elements of recognized portrait vocabulary was meant to separate that majestic effect from its ultimate source in 5th-c. art. Indeed, it might be argued that neither the Doryphoros nor any specifically classical Greek resonance was meant to be explicitly evoked for the ancient viewer. The ‘strong’ Augustus-Doryphoros theory depends on a few versions, such as the head of the Prima Porta statue (cat. 171: above fig. 2) and the ‘Chiusi head (cat. 86), which have rather hard, metallic forms, especially in the brows. But in most of the finest ‘core’ replicas of Augustus’ main type, neither the Doryphoros, Polykleitos, nor any particularly Greek 5th-c. art-style is explicitly legible. Some of the replicas employ a more ideal structure but with a soft, non 5th-c. handling (for example, the heads in Syracuse, cat. 195, and Vienna, cat. 206). More employ neither overtly high classical forms nor an overtly ideal structure. Rather they have subtly-designed portrait features that are to be taken as fine and handsome rather than ideal or divine. Such ‘physiognomical’ replicas among the core group that seems to stand close to the type are the head from Merida (cat. 130) and the head of the togate statue from the Via Labicana (cat. 165: above fig. 3). These versions steer a clear and careful course between portrait signs and a styled elevation If the Doryphoros, Polykleitos, and high classical athletes in general are not immediately legible in most of the good copies of Augustus’ main type, they were probably also not in the original model. An obvious tendency of the local reception of Augustus’ image was to bring out that latent elevation, idealism, and divinity which his subjects wanted to see and which was appropriate to veneration of the emperor initiated from below but not to his image as projected from above. Here in local contexts one does catch Polykleitan, or at least high classical, formulations of the type — for example, in portraits such as a head in the Palazzo Colonna (cat. 169) or, most obviously, the colossal head from Lepcis (cat. 20: fig. 5). Since the very best replicas vary widely in their balance between a representation of ‘ideal’ and ‘portrait’, we can never be sure what the balance was in the original model. It can at least be argued, however, that the model carefully concealed its strong ideal components beneath a refined layer of handsome portrait physiognomy, a layer easily swept away in the local reception of the type by subjects uninterested in Augustus’ subtle visual positioning of his public image. Between portrait identity and permanent youth The colossal Augustus head from Lepeis (ig. 5) is of great interest from this point of view for the semantic clarity of its constituent elements. The face is purely ideal in structure, composi- tion, and formal handling: it describes the emperor's godlike nature through the recognized vocabulary of high classical forms. The hairstyle is a careful if simplified version of the main type and provides ‘portrait’ identity. The two parts thus spell out ‘Divus’ and ‘Augustus’. The protruding ears also provide an important portrait-marker, stressing that the head represents a different kind of divinity from those of the older Capitoline-Olympian pantheon. The Lepcis head is an extreme example and shows up what an Augustus head that was explicitly modelled on Sth-c. forms would actually look like. It also points up what was the intended significance of the ideal elements. Neither Augustus nor his subjects had much interest in evoking Greek art or Greek athletic associations in the emperor's image. Instead the classical elements may be seen to point in another direction. Since the Hellenistic period, when the art of the 5th c. could be looked back on as ‘ideal’ and above reality (supra verum), the pure classical style had become the language of images of gods and of youths. Divinity is expressed openly in such far-reaching classical reformulations as the Lepcis head, but most replicas steer away from such overt divine associations. Permanent youthfulness on the other hand, is one of the most striking constants of the surviving replica record, Tt refers to the comparatively young age at which Octavian-Augustus came to power (19 at the death of Caesar, 32 at the battle of Actium). This, more than divinity, was something that might legitimately be emphasised by the centrally-defined model46 R. R. R. Smith The Alcudia-Actium type of the 30s B.C. was a youthful portrait with thick hair and probably some expression of vigour and energy — more youthful than most, but within the contemporary tradition of late Republican reformulations of the Hellenistic portrait vocabulary, and with no specifically royal or ruler aspects. It was created, c.40 B.C., for Octavian as the 21-year-old son of Divus Julius. The main type of 27 B.C,, created for Augustus aged 36, should, under the prevailing norms of Roman portraiture, have been a more mature, more physiognomical portrait, playing up the princeps' resolution, experience, and vigour. Portrait art had the means to increase an appearance of mature strength in a younger subject as much as to rejuvenate an elder figure. Augustus’ image here marks a radical break with tradition. For us the Prima Porta type may have an effective air of inevitability, but what might have been expected in 27 B.C. was something more like the portrait of Agrippa who was precisely the same age as Augustus and is presented in a ‘mature’, Hellenistic-style commander portrait." Some local workshops clearly thought that this was what Augustus needed: good examples are the heavily-lined portrait on the Praeneste altar (cat. 63) and the dynamic head from Chiragan (cat. 29 — it draws as much on the main type as the Alcudia type), where deep- set eyes and expressive physiognomy give the modern viewer a greater sense of age The new type of 27 B.C. created a permanently youthful but identifiable portrait, using a combination of refined, ‘handsome’ physiognomy laid over or embedded in a (covert) ideal structure, that was not updated, revised, or changed in a further 40 years of rule (figs. 1-3). The age-level was very finely judged. The portrait is generally ‘youthful’ but a little ‘older’ than the earlier Octavian type and is clearly ‘older’ than the images of young Augustan princes and heirs, such as Caius and Lucius. In the best versions, the main type achieves a sort visual paradox that might be described as a mature, ageless, and authoritative youthfulness. The combination of portrait language and elevating elements seems also to have been very finely judged. To emphasise only the embedded (or non-explicit) classical components (the ‘Doryphoros’ elements) is to miss this balance and tension. The balanced combination of ‘ideal’ and ‘portrait’ makes the message and expression of Augustus’ portrait difficult for us to formulate correctly: it was probably a genuinely ambiguous and ambivalent image that opened the way to an unusually wide range of reception and variation in even the most careful, typologically precise versions. One may compare the smooth, ideal head in Vienna (cat. 206) with the more ‘physiognomical’ portrait in Merida (cat. 130), two versions singled out by Boschung as amongst the best replicas for visualizing the original model. Boschung discusses the ideal components of Augustus’ image, in one place giving a straight forward ‘Polykleitan’ view (chapt. 3), while in others (especially in chapt. 6) he brings out the physiognomical portrait aspects well, but he does not really produce a descriptive and interpretive formula to reconcile them. Here formulation is all Augustus’ image and later emperors A major difficulty in defining the effect and meaning of Augustus’ portrait — and part of its great success — is that it had so few immediate points of reference and itself quickly became a portrait icon that carried meaning simply by reference to the first princeps. Boschung interprets the sequence of changing portrait types within Augustus'reign but he does not stand back to look at Augustus’ portrait image in the wider context of the image-styles of later emperors. This longer perspective is needed to bring out the radical novelty of Augustus’ image and its unusual character. It requires some effort of visual imagination to recapture the startling daring of Augustus’ manipulation of his public image in 27 B.C. Augustus’ main portrait tumed its back on that of his Republican predecessors and on their 16 For Agrippa’s main portrait type and its replicas: F. Johansen, “Ritratti marmorei e bronzei di Marco Vipsanio Agrippa,” AnalRom 6 (1971) 17-48.Typology and diversity in the portraits of Augustus 47 ‘Roman’ reformulation of Hellenistic-commander styles; it created his own personal style. The truly personal and dynastic element of that style is best seen in the light of what comes after. Augustus’ portrait quickly defined an ‘imperial’ manner for his immediate successors. It moulded the basic form of the Julio-Claudian imperial image up to the middle years of Nero's reign. The portraits of Tiberius and Claudius, out of a kind of pious visual modestia towards the first princeps, display a more mature physiognomical age than Augustus’ portraits, but still have the stiff, plain, controlling Augustan vocabulary (near-frozen in some versions of Tiberius’ portraits). The pure, youthful Augustan style, accenting the ideal components, is deployed in endless variants for the young princes of the family and for the reigning youths, Caius and young Nero. (It becomes clear in the portraits of these young princes that ‘youth’ was the primary significance of classical form in this context.) Rejected by the Flavians and sidelined by the triumph of the mature ‘bourgeois’ imperial images of the 2nd c,, the Augustan style did not reappear in imperial circles until the 4th c. No mature Roman emperor after Augustus was again shown so youthful or ageless until Constantine — whose later images were explicitly modelled on those of the first princeps. Constantine's successful appropriation and reformulation of Augustus’ portrait vocabulary to create a new and more explicitly monarchical imperial style casts an interesting light on the interpretation and meaning of Augustus’ own portrait style. Constantine's diademed portrait type made explicit what had been only implicit in Augustus’ main type. Augustus was also the only emperor (until the 4th c.) not to have his portrait image updated at some point to become ‘nearer’ his own age.” The change at 27 B.C. was the only change of represented age that his image underwent — from vigorous youth to the majestic calm of a youthful agelessness. The radical and unusual nature of this change also perhaps explains why the type was not changed later. In 27 B.C. Augustus’ image moved back from the ‘normal’ next step, a change to a more mature, physiognomical portrait. Having stepped back into an area where normative elements were as, or more, important than traditional ‘portrait’ requirements, and having found in the main type a satisfactory icon to carry those ideological components, Augustus’ image had no further need of adjustment. Many aspects of Augustus’ portraits were unusual, and some were unique to him. He was probably the first intperator for whom centrally defined portrait types (‘concealed originals’) were designed and somehow made available locally. His portrait types were certainly the first to be replicated in large quantities. Compared with those of other emperors, the local versions of his central portraits have an unusually wide range of variation and chronological span. His main type employs a radically novel image-style, employed in its pure form by no later emperor until the 4th c. And although he ruled for more than 40 years, his portrait was not felt to require serious adjustment after the type designed for his accession to more legitimate power in 27 BC. — either at the centre, or by local workshops and their patrons in the provinces. Of all imperial portraits, Augustus’ main type was the most reproduced, the most adapted, the longest in service, and the most radical and elevated in style. The portrait style of the first and most successful princeps was somehow his own. It repre- sented something to which his successors might not legitimately aspire. Until Constantine, no imperial prince or emperor avoided having his image altered periodically in some way, to represent major age-changes in his personal appearance.'® Only the great Divus Augustus had been able to straddle such a wide gap between his person and his image. Lincoln College, Oxford 17 Caius really was young (24 at the start of a short reign), and the official portraits of young princes (Nero, M. Aurelius, L. Verus, Commodus, and Caracalla), whose portraits entered the ‘system’ while they were still boys, were subsequently updated several times in line broadly with their major age- phases. 18 Seen.i7.