Bryn Mawr Classical Review

Bryn Mawr Classical Review 1999.03.20

J. T. Smith, Roman Villas: A Study in Social Structure.   London and New York:  Routledge, 1997.  Pp. xxxiii, 378.  ISBN 0-415-16719-1.  $100.00.  

Reviewed by Marc Kleijwegt, University of South Africa, Pretoria (
Word count: 4024 words

Excavation reports of Roman villas are published every year in considerable quantity, most of which are not perused, let alone studied in detail, by social historians. Monographs which aim to present a comprehensive synthesis of the material are much rarer and when they appear social historians should take note, especially, as is the case here, when the focal point of discussion is how architectural arrangements reflect social relationships. The theme is important and one which would benefit immensely from closer cooperation between archaeologists and historians. Smith's topic intersects with current debates on the social structuring of Roman houses and the nature of Romanisation. While much of the discussion in these areas, at least among Anglo-Saxon scholars, focuses on Italy (mainly Pompeii and Ostia) and Britain, Smith has included in his analysis villas from most European countries, east and west. Yet in other areas his focus is unfortunately more limited. He leaves out texts, pictorial representations, the architectural history (all of which are very prominent in John Percival's classic introduction to the subject) and the study of artefacts.1 While most students of Roman villas will agree that such material may hinder rather than enhance our understanding, the exclusion of artefacts is surprising in view of the recent attention this material has evoked. Its exclusion from Smith's examination is justified in a brief passage stating that 'few illuminate the use of rooms, and the difficulty of interpreting them is illustrated by the contrast between fine objects, especially those connected with feasting and dress, and the simple houses in which they are sometimes found' (p.6). A curious observation, especially in the light of Smith's persistent belief that social relationships can be read from house plans, for if there is a disparity between artefacts and the rooms where they are found how can we be sure that a social hierarchy can successfully be established on the architectural arrangements alone? For good measure one should compare the recent discussions on the value of the study of artefacts by Penelope Allison and Joanne Berry.2 It is clear that Smith believes that only one type of evidence can help us in explaining the social structuring of a villa, that is, the correct interpretation of house plans, 'irrespective of whether this goes counter to contemporary laws or descriptions' (p.16). Smith's main contention in this work will not be unfamiliar to those who have taken note of his previous publications. He believes very strongly that social reasons are the prime determinants of architectural form. Here there is plenty of room for Smith to express his disagreement with other students of the Roman villa. Polemics aside, Smith's ideas deserve serious consideration for the merits they may have for a better comprehension of the social occupation of the villa in particular, and Roman forms of housing in general.

The work is divided into three parts, although the central argument runs continuously through all of them. Part I discusses aims and methods (pp.3-20); Part II, the most substantial one, presents an analysis of villa types (pp. 23-216); Part III, 'The Villa System in Operation: Modes of Change', deals with the transformation from native houses to Roman villas and discusses how continuity or discontinuity in building structures may be detected and explained (pp. 219-301). The whole is well illustrated by numerous house plans drawn by A. T. Adams. Particularly helpful is a list of villas and other sites together with the original publications in which the plans may be found (pp. 340-59).

Part I presents an outline of the aims and scope of the book. There is sufficient clarity in these areas, but it is unfortunate that Smith has not taken more space to develop his thoughts. In little more than a page he mentions the uniqueness of the approach that he has adumbrated, the significance of house plans as evidence of social relations and the usefulness of comparative evidence from the study of late medieval and early modern houses. In the following section possible objections to Smith's methodology are briefly discussed. Some of the material outlined here is surprising. I know of no social historian who subscribes to the view that 'a rich man might choose to build a comparatively small house or a poor man beyond his means, so invalidating any close correlation between size or form of house and social class' (p. 4). And surely archaeologists will also acknowledge that the statistical probability of such a hypothesis is counter to predominant patterns of human behaviour. Moreover, it is counterproductive (and potentially disingenuous) to support with an example from seventeenth-century England the view that a Roman house was indeed a form of display or conspicuous consumption. That the same mentality dominated the social thinking of the upper classes in the Roman world is fairly evident in the historical sources. Knowledge of this would have made unnecessary the startling remark later in the book that 'it seems that the size and grandeur of private houses rarely caused political difficulties in Roman times' (p. 178).3 In general terms it is curious to note that a scholar who is so keenly interested in the social functioning of the Roman house is so uninformed of publications in this area.4 The history of villa classification (pp. 6-9) is a story of a few wise men and many fools. Next is a page on why there is so little attention in this book to the villa as a form of architecture, which allows Smith to criticize some of the more recent views in villa reconstruction, such as the one that maintains that villas in Britain had two or more storeys (p. 10). There follow brief discussions on what a villa is and how a site should be referred to. The final section in this chapter deals with the question of whether Smith's sample is representative. All this comprises a somewhat miscellaneous collection of thoughts, some of which could surely have been better incorporated in a general preface.

Chapter 2, on methods of investigation and assumptions, is more pertinent, although equally fragmented in its presentation. The section dealing with origins of plan analysis, for example (pp. 13-4), would have been better placed with the history of villa classification (pp. 6-9). The next section on the use of evidence from vernacular architecture is a more detailed outline of the point mentioned in the first chapter, but here illustrated with examples. Smith emphasises the importance of the position of the hearth relative to the entrance as an index of social relations and stresses that historical families needed only a certain number of rooms for everyday use. Houses with more rooms than seem absolutely necessary for a conjugal family thus elicit the question for what purpose these may have been used.

These points, and others not mentioned here, play a key part in the main body of the work, but their presentation in the introductory sections is lacking in organisation and coherence. The whole can hardly be called a proper methodological underpinning of what is surely (especially in the main part) a meticulous piece of scholarship. It is worth noting that in these introductory pages there is no reference to the kin-group, arguably the most important sociological concept used by Smith to explicate the social functioning of the house. In sum, one gets the distinct impression that these chapters were written after the main part of the book had already been completed, and kept to a bare minimum so as not to create an overlap with the more detailed observations submitted there. The result is most unsatisfactory. What is mostly lacking is a full introduction to the theories about the social composition of Roman provincial houses. It is only by collecting the various remarks scattered throughout the work that the reader becomes aware that some archaeologists believe that villas were inhabited by Romans (and not by Romanised provincials), by conjugal families with their servants, or by the nebulous bailiff and his family, viewpoints which Smith consistently combats as unexamined assumptions and as being in conflict with the archaeological evidence. In a book that is dealing with the social functioning of the provincial house and principally aimed at historians and social archaeologists, a direct dialogue with these views in an introductory section would have been desirable. Smith has missed an important opportunity here to tie his work to recent research on the ideological fallacies of the study of Romanisation.

Part II forms a lengthy and mainly technical discussion of various types of houses in order to elucidate their architectural and social functioning. Classifying houses according to typologies has the obvious advantage that evidence which is available in one house can be used to enhance our understanding of a similar house elsewhere. As Smith notes, no two houses of any size are really alike. The most that can be expected is that they show resemblances, that is, if the core elements are more or less the same (p. 87). Where these conditions apply Smith is arguing that the differences may be attributed to the various stages of social dynamics involving two or more households. There are several elements which Smith considers to be of significance in deciding upon social equality or inequality between the various households occupying a house: the position of the hearth in the hall; the availability of cellars beneath a living-room and access restricted through it, suggesting that the household living there controlled the supplies; the positioning of doorways; the availability of hypocausts and mosaics illustrating superior status; the use of lobbies to shield private rooms from visitors and servants. Which room was the superior one is usually indicated by better appointment and finish, not necessarily by size. If a work hall was part of the house and not relegated to a subsidiary building, it was usually the biggest room (pp. 132-3). A more subtle means of detecting social variance within a house is by way of looking at symmetrical arrangements. Smith argues that perfect symmetry, along the classical canons of Roman architecture, was seldomly achieved in a house for the reason that the social requirements of the various households were never strictly egalitarian (cf. p. 92 on Newell).

The two fundamental typologies of provincial houses are the hall house and the row-type house. Hall houses are found over a wide area of Europe and originate within an indigenous tradition. All are good-sized buildings and some are very large indeed; all are capable of accommodating more than a nuclear family, or more than one nuclear family, not necessarily the same thing. Smith is very dismissive of alternatives which may be presented to his own view that hall houses were occupied by more than one household. The assumption that the hall was used by a family living on the one end and working on the other is considered 'an inconceivably wasteful use of so large a space' (p. 45). An owning family and its servants living on opposite ends of the hall is regarded as an impossibility for 'there is little sign of the spatial differentiation that might be expected in a situation where there is great disparity of wealth' (p. 45). All these are valid arguments, but their being cast in the form of negative findings does not necessarily provide eloquent proof that hall houses were always occupied by more than one household. Smith's ideas can be more properly tested in the case of row-type houses.

Row-type houses take the form of a row of two to five (sometimes even more) rooms which are interspersed with smaller squarish or corridor-like rooms. There is usually a main room which is somewhat bigger than the other ones. This could be either a living-room or a representational room. The conclusion is that such houses were occupied by more than one household. The best evidence for such an assumption is the physical evidence for the existence of two front entrances (p. 59), but such evidence is exceptional. For the majority of houses Smith has to rely on the positioning of rooms and doorways and the use of reverse symmetry to accentuate the relative status of individual rooms and their occupants. In practice the theoretical bottom-line of Smith's arguments is the assumption, taken over from studies of early modern households (p. 57), that the average household had need of only one sizeable room and access to other facilities. No attempt is made to ascertain whether such an assumption might be valid for the Roman provinces. As can be imagined, the general difficulty lies in identifying how many smaller and bigger rooms formed a distinct unit occupied by one household and therefore in reconstructing how many households lived in a particular house. The middle room, as can be demonstrated for instance in the case of the villa at Newton on the Isle of Wight, is usually representational and designed for the shared celebration of religious ceremonies. Newton is unusual in retaining evidence of doorway positions, and thus it can be demonstrated that the middle room did not communicate with the rooms on either side. Newton is instructive for another reason as well, for the two flanking rooms are not of equal size. It was most probably the smaller room to the west which was the more important one since it had mosaics and the east end shows no signs of there having been any mosaic floors (p. 48). Smith does discuss some alternatives to his interpretation. For example, he supposes that the apartments at each end may have been occupied by husband and wife and their respective servants, but immediately dismisses the whole idea as impossible. There is no discussion, however, of separate living-rooms for the children. Smith appears automatically to assume that they would live and sleep together with their parents. Another factor which is completely overlooked by Smith is that villas of some size would have needed guest-rooms to accommodate acquaintances and friends. The value of these inferences is unknown for the moment, but it is evident that if proven to be correct this would reduce the number of rooms available for other households of the same kin-group.

A significant component that was added to most houses is the porticus-with-pavilions. Three reasons can be mentioned for introducing it to an already existing structure: aesthetic, cultural and social. It is unclear from Smith's discussion whether one was dominant over the others, nor is it evident how the two main ones (the desire for enhanced social differentiation and the wish to advertise Romanitas) influenced each other. The main social reason, Smith argues, must be sought in the desire to have additional high-status rooms. In terms of the social development between the various households Smith sees the creation and function of pavilions as an extension, as it were, of the dual occupation of the hall. The latter was still used as living-space but left to kinsfolk who were dependants, while the pavilions were taken over by the senior households (p. 119). Symmetry was an important factor in the construction of porticusa-with-pavilions, yet house plans, if indeed accurately reproduced, suggest that symmetry was hardly ever achieved. It is one of Smith's main achievements in this work that he has identified this as a significant element in the construction of houses by Romanised provincials. Two things need to be mentioned here. First, Smith convincingly demonstrates that in view of the skills of Roman provincial builders the lack of symmetry evident in some parts of the house, small though it may be in some cases, must be deliberate and significant of some social differentiation (pp. 121-2). Second, Smith's conclusion that lack of symmetry indicates social desires goes counter to many attempts, Mylius' one in the case of Mayen being an infamous example, to reconstruct villa plans along the lines of the classic canons of symmetry.

Part III deals with the transformation of the native house and the emergence of the villa. Smith offers a good discussion of the evidence for pre-conquest houses, but it is clear that the evidence is not as detailed as it is for the final phase of the villa. Chapters 15 and 16 undoubtedly are the most important chapters in this part, and perhaps in the whole book. In these chapters Smith discusses the social development of individual villas rather than the social implications of the typology to which they belong. It is at this point that some weaknesses in Smith's arguments may be discerned. I found it particularly hard to test the implications of his assertion (p. 271) that '[w]here the principal house of a villa was enlarged over a long period of time it was due to an increase in population or an improvement in the standard of living or a bit of both'. The two principal factors are not in doubt, but how can they be neatly separated, especially if one postulates that an improvement in the standard of living may have generated a desire for more and better appointed rooms, without a concomitant rise in population being necessary? Moreover, up to a point, an increase in population may in the short run have prompted no alterations, and in the long run may have resulted in a decline in living standards. Any explanation offered here contains too many unknown variables and is therefore ultimately untestable.

One of the houses which Smith discusses in more detail in these pages is Blankenheim. In Phase I it is a typical hall house with only a few separate, more or less equal, rooms and much space in common. Smith suggests a kin-group of fairly large dimensions (seven or eight households, cf. p. 265) and near-equal social relations. Phase II saw some alterations, for instance one room received a hypocaust and was entered through a lobby which was also the entry to another room. This suggests that one household started to stand out from the kin-group. However, the middle hall and a shrine room were left unaltered, 'no doubt because the social relations which were expressed from time to time in ceremonies and feasts continued substantially unaltered' (p. 267). In Phase III the middle part was completely reordered. The hall was divided up into smaller rooms, creating the impression of a row-type house, and the religious function was divided between the baths and a newly constructed temple outside the house. At this stage there arose three markedly stratified apartments and a fourth block of rooms for people who 'may have been kin but were virtually dependants' (p. 268). Thus in Smith's opinion the house went through the stages of social near-equality in I, predominance of one household in II and predominance of three households in III, with a concomitant moving down the social scale for the remainder of the population. Leaving aside the validity of the implications of the architectural changes, what sort of development might be responsible for creating this sort of social zig-zagging, especially between Phase II and III? The development from I to II is readily understandable, but how do the three superior households emerging in III relate to previous developments? Are we still dealing with the same kin-group that inhabited the house in I?

What is problematic in Smith's argumentation is that there were only minimal changes to some houses, with the core staying the same over a considerable length of time, which implies that the general composition of the house, Smith's duality, did not change significantly either. Especially in houses that remained virtually unaltered over many decades, this is somewhat contrary to expectations based on demographic evidence. In some cases, he suggests, overpopulation was siphoned off to nearby cities (cf. e.g. p. 293), but it is not clear how this process can be proven, nor is it likely that evidence in support of this theory will ever become available. Another point which needs to be raised is that the majority of houses reflect slight instances of social differentiation, with one household gradually standing out from the clan. While it may be the case that houses were constructed or adapted with such a differentiation in mind, it is not at all obvious how this process came about in social terms. Is such a development to be envisaged already within Celtic customs of partible inheritance or is it the gradual outcome of increased exposure to a Romanised framework? The discussion on pp. 278-283 fails to enlighten me. For instance, Smith (p. 281) contrasts the architectural features of a society based on partible inheritance with a Romanised landscape. The periodic division of land accompanied buildings of temporary earthfast construction which were replaced comparatively frequently. A house built of stone, however, was less prone to decay and the gradual emergence of a leading family of the clan may have been a factor in starting the fossilization of social differentiation. My question would be the following. If Romanisation had the effect of making inroads into the egalitarian structure of the clan, what reason is there to believe that partible inheritance survived the first steps of Romanisation? In other words, if Romanisation eroded the informal structure of the clan, what reason was there to continue to express social relations in the duality so typical of the Roman provincial house? And why is that social differentiation expressed in such a minimal degree of architectural visibility?

Smith is a persuasive observer of house plans with a pronounced aversion to anachronistic interpretations. For example, he dismisses the notion of bedrooms separate from living-rooms and the idea that Romanised provincials needed as much light in a room as people in industrialised countries. These may be small points but Smith demonstrates how they have led to serious misconceptions about the occupation and use of space in Roman villas. He consistently argues that parallels, whether implicitly or explicitly drawn, with Victorian England are confusing and meaningless (cf. the disparaging remark on the bourgeois household in togas at p. 92) and prefers comparisons with post-Roman Welsh and early modern English society. His deconstruction of earlier attempts to make sense of Roman provincial houses is acute and acerbic. Not many people in the field will be pleased by his dismissive tone, but this does not detract from the importance of his statements. His main argument, the continuation of partible inheritance as the predominant socio-cultural pattern of the western provinces which is then reflected in the use of space, has met with a variable response so far. This may well be, in the present state of our knowledge, an untestable hypothesis, but it is merely the possible social underpinning of the observations he has made of various types of houses.

With this book Smith has produced a radically different perspective on the Roman villa in the western provinces of the Roman Empire and simultaneously has called into question many of the accepted notions about Romanisation. It is the most eloquent support of John Percival's perceptive statement, made more than twenty years ago: 'Could it be, in fact, that the social structure in these areas is not a Roman one at all, but a native one dressed up in Roman trappings?'.5 Even if one disagrees on some points with Smith's arguments, or perhaps with all of them, one must surely recognise the value of his work. He has created a fundamental awareness of the social functioning of provincial houses and pointed out many similarities between houses in different areas of the western provinces of the Roman Empire. These similarities, in Smith's opinion, are to be attributed to similar responses to the same social circumstances. This view is necessarily controversial and requires further investigation, as is the contention that the Roman authorities were somehow involved in making available to provincial architects information on how to formalise social differences into more or less uniform architectural constructions (cf. pp. 284; 288). What Smith has done here in a thought-provoking way is to put in doubt the traditional explanation of the social structuring of the Roman provincial house. In its place arises a new framework, not yet perfected, which views the house as a more natural product of indigenous and Roman influences. Smith's arguments assume a predominantly native ordering in the occupation of these houses, a theme which is of great interest to the current revaluation of Romanisation.


1.   John Percival, The Roman Villa: An Historical Introduction, London 1976.
2.   Penelope M. Allison, 'Roman Households: an archaeological perspective', in Helen M. Parkins (ed.), Roman Urbanism: Beyond the Consumer City, London and New York 1997, 112- 47; idem, 'Artefact Distribution and Spatial Function in Pompeian Houses', in Beryl Rawson and Paul Weaver (eds.), The Family in Roman Italy: Status, Sentiment, Space, Oxford 1997, 321-55; Joanne Berry, 'Household artefacts: towards a re-interpretation of Roman domestic space', in Ray Laurence and Andrew Wallace-Hadrill (eds.), Domestic Space in the Roman World: Pompeii and Beyond, Portsmouth, RI 1997 (Journal of Roman Archaeology Supplementary Series Number 22), 183-96.
3.   Cf. for instance the house of Valerius Publicola which was by public opinion considered to be too imposing and tied up with notions of tyranny (Plut. Publ. 10; cf. Liv. 2.7.12 and Dion. 5.19.1), cf. Monica Affortunati and Barbara Scardigli, 'Aspects of Plutarch's Life of Publicola', in Philip A. Stadter (ed.), Plutarch and the Historical Tradition, London and New York 1992, 109-32, esp. 114. For a general survey of the symbolic importance of the aristocratic house and the ambiguous evaluation of luxurious houses in particular see Catharine Edwards, The Politics of Immorality in Ancient Rome, Cambridge 1993, 137-73. Evidence for houses being razed to the ground, mainly for political reasons, can be found in Cicero, De domo sua, 101 and Val. Max. 6.3.1.
4.   Richard P. Saller, 'Familia, domus and the Roman concept of family', Phoenix 38 (1984), 336-55; Andrew Wallace-Hadrill, Houses and Society in Pompeii and Herculaneum, Princeton 1994.
5.   Percival (cf. n.1 above) 135.

Read Latest
Index for 1999
Change Greek Display
Books Available for Review

HTML generated at 13:26:17, Friday, 03 April 2009