Commensalism and common sense: a new approach to archaeological ceramics

John Hawthorne


Archaeologists may have jumped the gun by moving directly from counts of ceramics to economic theory. This paper demonstrates that in the case of African Red Slip ware, ceramic abundance and scarcity can be more closely correlated with changes in the size of the vessels than macroeconomic trends. Periods when the pots are small tend to be well represented in the archaeological record, and conversely, periods when the pots are larger tend to be less well represented. It is argued that this is the result of changes between individual and communal eating: when eating individually is the norm, each person needs their own (small) dish, but communal eating requires fewer, larger vessels. The paper concludes with some speculation as to why such a simple idea has not been considered before.


This paper outlines a proposition which, although laughable in its simplicity and apparent obviousness, has received no attention in the archaeological literature, despite its very serious ramifications. In essence, it is argued that, in some cases at least, periods when pottery is rare correspond to periods of communal dining (commensalism, i.e., when the norm is for several people to eat from the same vessel); concomitantly, periods when pottery is abundant may correspond to periods of individual dining (that is to say, when the norm is that each diner has their own vessel). This is because the great size of the pots required for communal dining means that few are needed. The implication of this is that patterns which have up until now been interpreted in terms of supply and demand can now be seen to result from changes in eating habits. If this is a phenomenon which is found beyond the case studies presented here, which common sense suggests will be the case, then some very serious rethinking of the archaeology of all ceramic periods is needed.

This paper is in three sections. The first demonstrates the utility of this way of thinking about ceramics through the example of a particularly common later Roman fineware, African Red Slip ware. The second examines notions of communal and individual dining in greater detail, emphasising the considerable temporal and geographical distances which may be affected by this hypothesis. The third attempts to explain why this hypothesis, which appears to be the most basic common sense, has not been thought of before in archaeology.

African Red Slip Ware: a new interpretation

African Red Slip Ware (ARS) is one of the most common later Roman finewares. The standard works of reference are by Hayes (1972) and Carandini (1981). ARS was produced from the late first century AD until the mid seventh century in the area of modern Tunisia and exported around all of the Mediterranean, reaching even to Scotland in the north and Ethiopia in the south at the peak of its distribution (Carandini 1981:11). The form repertoire consisted of a series of bowls, dishes and plates, with some closed forms which were not commonly exported.

Figure 1. Total sherds per year as calculated from seven surveys. This graph uses summed absolute totals, but the same pattern is found if the average is calculated. The surveys are: Caesarea (Leveau 1984), Ager Tarraconensis (S. Keay pers. comm.), Monreale (P. Perkins pers. comm.), Mirabile (P. Perkins pers. comm.), South Etruria (P. Perkins pers. comm.), western Sardinia (P. Perkins pers. comm.) and Alicante (Reynolds 1993).

The importance of ARS in later Roman archaeology is that its frequent presence on sites has allowed its use as an indicator of fluctuations in trade and demography. This is especially true of field surveys. In recent years, attempts have been made to quantify the fluctuations in the quantity of ARS found on several surveys over time (Fentress and Perkins 1988; Cambi and Fentress 1989). The results of these analyses show that there is one particular pattern which was almost always found on the sites; this is the peak of ARS in the later second century and subsequent massive decline in the third, with sporadic recovery in the fourth (Figure 1). Until it was discovered that this pattern was common to all sites, its occurrence on individual sites was interpreted as the result of local economic and demographic crises (cf. Potter (1979) on the South Etruria Survey). More recently, in the wake of the work of Fentress and Perkins (1988) and Cambi and Fentress (1989), interpretations have focused on the possibility of an economic crisis in the production area, North Africa, which then had a knock-on effect on the consumption sites. There is a problem with this second approach, however. This is that the trough of ARS in the third century on west Mediterranean sites corresponds exactly to the expansion of its distribution to the eastern Mediterranean as well as the west. If there was enough ARS ware to cover both basins of the sea, it is difficult to see how there would be insufficient to continue supplying the western sites. Thus, another explanation is called for.

Figure 2. Mean vessel volume.

The explanation forwarded here is that the drop in quantities of ARS in the third century corresponds to an increase in the average vessel volume (Hawthorne forthcoming a). A comparison of Figures 1 and 2 shows that the changes in vessel capacity are almost exactly inversely correlated with the changes in sherds. When pots are small there are many sherds, and vice versa. That this is not the result of differential breakage rates is clear; an analysis of over 500 sherds from Carthage by the writer and over 200 from Lepcis Magna by M. Attree has shown that the average breakage rate is constant across time, never deviating significantly from 5-10% (Hawthorne, forthcoming a). The possibility thus arises that the two phenomena are actually linked - that the reason for the drop in sherds is the rise in vessel capacity. Given that the average vessel diameter increases from approximately15cm to nearly half a metre from the second to third centuries, we are clearly dealing with a very significant change, a change of such magnitude that a link between the two variables is not implausible. The argument is lent further support when an additional calculation is attempted. If the volume of each form is multiplied by its sherd count (and calibrated by the breakage rate) we arrive at a figure for 'total overall volume', that is to say, a surrogate for the average total volume that we might expect to be represented by all the vessels on a site at any one point in time. This is shown on Figure 3. The graph is spiky, as we might expect given the nature of the calculation, but in comparison to Figure 1, it is clear that the level of the total overall volume remains generally constant through time.

Figure 3. Sherds * volume, calibrated by the average breakage rate for each form.

It is interesting to observe that similar changes occur in other wares at the same time. The above analysis has been undertaken on the tableware versions of ARS, but the ARS cooking wares, which were also exported to the west Mediterranean, show a parallel increase in size (Hayes 1972:47). The problem with including these in the analysis is that their dating is still uncertain; according to Hayes (1972) they lasted only until the early third century, whereas Carandini (1981) claims they survived until the late third. As the most substantial changes in tableware vessel capacity occur in the late third century, this is clearly an issue which needs to be resolved. Perhaps of more significance, though, is the fact that the change to larger pots is also seen in a ware which is produced outside Africa, Spanish sigillata. Although Spanish sigillata has an entirely different form series, with different shapes and decoration, the underlying pattern is the same: in the second century there was a predominance of small bowls which were replaced by large dishes in the third century (MezquÌriz 1981, 1983).

Thus, there is a strong case for suggesting that the changes in the quantities of ARS on sites in the western Mediterranean from the second to third centuries are closely related to changes in the size of the vessels. But what does this represent in terms of the use of the pots? The most likely explanation for the change is a move from individual dining to communal dining as Carandini (1981:15) has previously suggested. Given that the most common ARS pots of the third century are nearly half a metre in diameter, it seems unlikely that they were intended for use as individual pots, so the change to communal dining seems a reasonable explanation. It may further be argued that the change is related to the rise of early Christianity in Africa, with the adoption of communal dining seen as a reaction to the individual dining of pagan Rome (Hawthorne forthcoming b and c) but that is another story.

It remains to consider the potential impact of this way of looking at communal and individual dining on a broader scale, and also to examine the reasons why such a hypothesis has never been generated in archaeology before.

Individual and communal dining

Defining individual and communal dining is not as easy as might be imagined, as elements of one may be present in any given real-world example of the other. However, we might seek refuge in ideal types. Thus:

Communal dining. Following Bats (1988:23), communal dining can be defined as when a group receive their food from a vessel which is used by all. They may have individual settings to actually consume the food, or may simply eat directly from the bowl.

Individual dining. Again, following Bats (1988:23), individual dining can be defined as when not only the eating of food is done from individual vessels, but also the serving. There is no central vessel for serving.

We might then extend this to define 'periods' of communal or individual dining as times when it seems that one form of dining dominated to the virtual exclusion of the other.

To illustrate how such systems may have worked, we must turn to analogy. The individual ideal type is essentially that which is the norm in the West today, at least in informal domestic contexts. Each diner has their own place setting and communal vessels tend to be restricted to bowls for bread rolls or fruit. In Europe this system has been in evolution since the sixteenth century (Mennell 1985; Braudel 1973:124-39; Farb and Armelagos 1980:204-8; Hammond 1993:109-120). Thus, the majority of the tableware will be of a size suitable for individual servings, and in any given context the ratio of crockery pieces : diners will be at least 1:1 and probably considerably greater.

Communal eating in its purest form, on the other hand, is best illustrated by the example of medieval Europe. Although use would sometimes be made of individual bread trenchers, the emphasis was on a number of diners eating from a communal bowl, usually only using their fingers (Hammond 1993:111; Farb and Armelagos 1980:207-8). The number of diners per bowl varied. In England there existed the mess system whereby a number of people in a dining room ate together in groups. John Russel in 1430 laid down that there could be up to 4 people per mess although the number would drop with the increasing status of the participants (Hammond 1993:118). In courtly circles at least, there were also rules on the rationing of food to messes. In 1478, Edward IV ruled that every two men in a mess should share a loaf of bread, every four men a gallon of ale and every three men a dish of meat or fish (Hammond 1993:118). Therefore the size of the tableware is likely to be such that it may contain sufficient food for several people, and the ratio of crockery pieces: diners will be 1: many.

Thus it is likely that there will be significant differences between periods of individual and communal dining in terms of the number of vessels represented, but given that the same volume of food will be consumed, we might expect that the total volume of the vessels in each period will be approximately the same, as is the case for ARS when the volume of each form is multiplied by the number of sherds (Figure 3).

The potential areas for application of this hypothesis are legion. Changes between periods of communal and individual dining are common throughout history. This may be observed in the first millennium BC in the Mediterranean as a whole (Bats 1988) and indeed for several millennia before this (Latham 1982). As noted above, Mediterranean Europe and Britain are well known examples (Mennel 1985; Braudel 1973; Farb and Armelagos 1980; Hammond 1993), but it is also seen in contexts as diverse as colonial south Carolina (Ferguson 1991) and contemporary China (Croll 1983). Given the frequency with which these changes occur, and given the seeming obviousness of the basic hypothesis presented here, it must be asked, why has this not been thought of before?

Common sense and ceramics: a critique of ceramic quantification

It is probably fair to say that the quantification of archaeological ceramics has progressed beyond the stage of being merely a methodology and is now an art form in its own right. As well as the more familiar wrangles over the relative merits of sherds vs. weights vs. estimated vessel equivalents (EVEs), there are now debates over the merits of some of the more exotic methods such as the Pottery Information Equivalents (PIEs). Orton et al. (1993) provide an intelligible guide to the latest research in layman's terms. However, with this proliferation of ever more complex techniques there is a danger that we may lose sight of one of the fundamental aims of quantification, and indeed forget why it is a fundamental aim. According to Orton et al. (1993: 166) the most basic question is, "How much pottery is there?" We can go beyond this, however, and ask, "Why do we care how much pottery there is?" This is by no means a facile question, as the answers we might think of are illuminating. The standard answers must be that we quantify ceramics to study such things as trade, demography, site hierarchies and so on - in other words, large scale phenomena. Even studies which use quantified data to look at social issues (see Yentsch 1991) only consider the light that different classes of pot may throw on status, power relations and so on. Quantified studies of pots for the sake of examining eating habits alone are rare (Bats 1988 is an exemplary exception), but - and this is the crucial point - there is never any conception that eating habits might affect the overall quantified data. The only variables that we allow to affect our quantified data are respectable macro-variables, such as trade and demography. Alternatively, changes in the materials used for food containers are sought as explanations of quantitative change. These substitutes have the virtue that they are comparatively untraceable archaeologically; metal and glass are recycled, and wood does not survive well. Thus there is a further disincentive to look at vessel size, as there must always be uncertainty as to the amount of quantitative variation which may already be explained by substitution.

There is perhaps also a more insidious reason lurking at a more technical level. Historically, ceramic quantification in archaeology has been dominated by excavations. Excavations have the problem that change over time cannot be looked at in an absolute manner, as the amount of pottery recovered from any period of time is heavily dependent upon the amount of earth excavated from that stratum. Thus there is no incentive to look at variables which may produce absolute variation between periods - but this is exactly the sort of analysis which is needed to see the patterns, as the examination of ARS demonstrates. This is beginning to change now, with the proliferation of field surveys, which are not so bound by the problem of unequal stratigraphy.


This paper has argued that it may be profitable for archaeologists to give more attention to the size of ceramic vessels, with an eye to possible changes between communal and individual dining, as it may be that in many cases this will reveal otherwise unnoticed patterns in the data. This has been demonstrated in the case study of African Red Slip ware. The importance of the hypothesis for ceramic studies is that it raises the possibility of adding to the standard economic and social explanations by considering pots for what they originally were - containers for eating from.


I am grateful to Simon Keay for encouraging me to pursue this idea and to Phil Perkins for supplying much of the data. I would also like to thank Peter Durham for his constructive criticism.


Bats, M. 1988. Vaiselle et alimentation ý Olbia de Provence (v.350 - v.50 av. J.C.). ModËles culturels et categories cÈramiques. Revue Archeologique de Narbonnaise, Supplement 18. Paris: Centre National de la Recherche Scientifique.

Braudel, F. 1973. Capitalism and material life, 1400-1800. London: Wiedenfeld and Nicholson.

Cambi, F. and E. Fentress. 1989. Villas to castles. First millennium AD demography in the Albegna Valley. In K. Randsborg (ed.), The birth of Europe: Archaeology and social development in the first millennium AD. Analecta Romana Instituti Danici, Supplementum XVI. Rome: L'Erma di Bretschneider. pp 74-86.

Carandini, A. (ed.) 1981. Atlante della Forme Ceramiche. I. Ceramica fine romana nel bacino mediterraneo (medio e tardo impero). Rome: Instituto della Enciclipodia Italiana.

Croll, E. 1983. The family rice bowl. Food and the domestic economy in China. London.United Nations.

Farb, P and G. Armelagos. 1980. Consuming Passions:. The anthropology of eating. Boston : Houghton - Mifflin.

Fentress, E. and P. Perkins. 1988. Counting African Red Slip Ware. AntiquitÈs Africaines 24, 205-214.

Ferguson, L. 1991. Struggling with pots in Colonial South Carolina. In R. McGuire and R. Paynter (eds), 28-39.

Hammond, P. W. 1993. Food and feast in Medieval England. London: Alan Sutton.

Hawthorne, J.J. forthcoming a. Vessel volume as a factor in ceramic quantification: the case of African Red Slip Ware. In Computer Applications and Quantitative Methods in Archaeology 1996. Proceedings of the 24th Annual conference held in Iasi, Romania.

Hawthorne, J.J. forthcoming b. Post-processual economics: The role of African Red Slip Ware vessel volume in Mediterranean demography. In TRAC 96: Proceedings of the 6th annual meeting of the Theoretical Roman Archaeology Conference.

Hawthorne, J.J. forthcoming c. African Red Slip Ware and the economy of Severan Proconsularis. In Acta di I Convegno Internazionale di Studi Severiani, Albano Laziale.

Hayes, J. 1972. Late Roman pottery. London: British School at Rome.

Latham, J.E. 1982. The religious symbolism of salt. ThÈologie historique 64. Paris: Beauchesne

Leveau, P. 1984. Caesarea de MaurÈtanie et ses Campagnes. Rome: Šcole FranÁaise.

McGuire, R. and R. Paynter (eds) 1991. The archaeology of inequality. Oxford: Blackwell.

Mennell, S. 1985. All manners of food: eating and taste in England and France from the Middle Ages to the present. Oxford: Blackwell.

MezquÌriz De Catalan, M.A. 1981. Terra Sigillata Hisp·nica.In Carandini ed. (1981), 247-323.

MezquÌriz De Catalan, M.A. 1983. Cer·mica Sigillata Hisp·nica. Historia y criterios tipolÛgicos. Separata del Bolletino del Museo Arquelogica Nacional, Tomo 1.2. Madrid,

©John Hawthorne 1996


E-mail assemblage

©assemblage 1996