Jane Webster has made an eloquent and compelling case for writing slavery – and other forms of unfree labour – back into the historiography of the Roman world. Her argument weaves together questions of the nature of evidence, the principles of comparative study and the politics of disciplinary practice. My own concern has been with the archaeology of European colonialism and the postcolony in South Africa, several oceans and more than a millennium away from the Roman world (Hall 1992; 1999; 2000). Webster bridges this divide through a direct question: ‘it would be interesting to know whether Hall himself would feel that the “fact of slavery” . . . might generate recognizably similar material statements among other slave-owning peoples in other periods, including ancient ones’ (p. 112). This is to respond to this question, and to tease out some of the provocative implications in Webster's sense of the possibilities inherent in a wider archaeological frame of enquiry.

Webster's central tenet is that the possibilities inherent in comparative archaeology are underdeveloped. She sees two reasons for this, which can be separated into the profane and the epistemological.

The profane restraints on comparison rest in the organizational politics of universities, museums and other research organizations, in the overvalorization of ‘discipline’ as a means of legitimating access to employment opportunities, research funds and status. Thus ‘classicists’ need to differentiate themselves from other archaeologists, and from each other. Graduate students – necessarily aware of the realities of patronage – know which conferences and seminars to attend, and which to keep away from. Divisions of knowledge are perpetuated not because there is a logic to such differentiation in the questions asked or the evidence interrogated, but access to key resources is at stake. In putting this before us, Webster is overly kind to other fields of archaeological enquiry. ‘Historical archaeology’ was long reserved for the archaeology of colonial North America and the claim for a broader view is still comparatively new (Funari, Hall and Jones 1999; Hall and Silliman 2006). ‘Post-medieval archaeology’ is still largely distinct, despite the direct connections with colonial archaeology (Johnson 1996). ‘Underwater archaeology’ still organizes separately, despite the obvious point that shipwrecks were not spontaneously created beneath the sea.

Such issues of organizational disposition, while obviously important in the day-by-day politics of academic organizations and highly significant in the ways in which interpretations of the past are structured, have no inherent legitimacy. They rather signify the double meaning of ‘discipline’: as a way of ordering knowledge, and also as punishment for transgressions. Because profane restraints are intellectually arbitrary, breaking through their barriers offers exciting possibilities for new synthesis and interpretation of key aspects of the past. Webster opens up such a set of possibilities for a new understanding of slavery and unfree labour in the Roman world.

The epistemological question is more complex, and Webster comes back to it several times in her paper, clearly with some ambiguity, eventually asking the faint-hearted to ‘look away’ (p. 119) while she proposes a generalized principle for the archaeology of slave communities, wherever they are. Her dilemma is that, in proposing a comparative approach, she may be heading back to a modernist ‘laws-of-human-behaviour’ approach, whether as a set of Marxist propositions or as positivist generalizations in the tradition of Lewis Binford or Stanley South. I think that this concern can be addressed usefully in two ways.

Firstly, comparative study can set up provocative and instructive analogies that are valuable in pointing to new sources of evidence or lines of interpretation. To do this hardly implies anachronistic modernism: as Webster appropriately points out, ‘prehistory’ depends overwhelmingly on analogy for interpretation. Webster's own case study illustrates this well. In pointing out that graffiti may be potential evidence of slave lives in the Roman world, I do not read her as implying that, as a law of human behaviour, enslaved people write graffiti. I rather see her using evidence from slave societies in other times and places to suggest that romanists take a fresh look at a category of evidence they may previously have passed over too quickly. This is clearly useful when the Museum of London tells us that the only evidence for slavery in Roman London is four inscriptions from writing tablets and gravestones (www.museumoflondon.org.uk). By transgressing the disciplinary boundaries and offering analogy by comparison to another slave context, Webster is suggesting we look at what we may have for the Roman world in new ways.

Secondly, we can use material culture, texts and other kinds of evidence to look for recurrent patterns across contexts that help us see the underlying dynamics of power at work. This has been the approach that I have followed in a comparative archaeology of slavery in the Dutch East India Company's Cape of Good Hope and in broadly contemporary British Virginia (work to which Webster generously refers in her paper). Again, there is no need to evoke universal laws here. The point of comparison is rather to reach beneath the immediate and specific to look for the interests at play beneath the surface – the dynamics of labour relations, gender, the extraction of surplus benefit, status and claims to benefits. I owe this, of course, directly to Foucault's concept of power and discourse and the idea of the ‘statement’ as a manifestation of everyday power relations (Foucault 1972; 1979). In enabling an archaeology of power that cuts across the specificities of context without evoking a priori general laws of behaviour, Foucault has enabled the sort of comparative study that Webster signals as a key opportunity for better understanding the key role of unfree labour in the Roman world.

Such profane and epistemological restraints on a comparative archaeology of the Roman world come together in the peculiar privileging of textual sources. This is strange because, of course, historians have themselves long recognized the partiality of written sources and have shown repeatedly the immense value in other sources of information about the past – oral accounts, images, landscapes, architecture and the like. Leaving aside the obvious preferences of archaeologists, it is difficult to see the logic of seeking to ‘read’ the Roman world primarily through textual evidence, and for relegating other evidence to a secondary position. The huge attraction of any context in which material and textual sources are coterminous is that we can escape the constraints of interpretation through ethnographic analogy. To insist on the primacy of textual sources is to deny this set of possibilities – to put the profane interests of discipline ahead of the epistemological richness of a past understood through all its available dimensions.

In exploring these possibilities, Jane Webster's attention to graffiti will surely be productive. But I was taken by the possibilities in her observation that a ‘stigma of inferiority’ was attached both to slaves and to those who had been manumitted, to the point of emphasis in inscriptions on graves and other marks and claims to status. This points to ambivalence and to contradiction – to the identities of those who are enslaved but essential to the prosperity of their owners, or who are legally free and powerful but related to those still in enforced servility.

In looking at slavery in the colonial Cape, I have found that such ambivalence leads to contradictions, absences and overdeterminations that result in patterning in the array of textual and material evidence. For example, slaves at the Cape had a diet dominated by fish, a cheap and readily available food. Fish bones are well represented in archaeological assemblages, but fish is denied as part of the Cape's culinary opportunities in contemporary travellers’ accounts. Contemporary drawings and lithographs position fishermen and representations of fish markets ambiguously. Taken together, the textual, archaeological and graphic sources of evidence show absences, contradictions and overemphases that show how fish and fishing signify the ways in which slavery underpinned life in a colonial town in which more than two-thirds of the population were unfree (Hall, Halkett et al. 1990; Hall 1991; 1992, 1999; 2000).

I would predict that, freed from the constraints of disciplinary conventions, those with a close knowledge of the Roman world would find similar absences and ambiguities created by the stigma of inferiority, showing that slaves and slavery have been part of the historical and archaeological record all along.