At the beginning of the fifth century CE, a small group – their names and biographies now lost in the anonymity of the archaeological record – gathered at Henchir (Hr.) el-Hami, a site on the edge of the fertile plains of the Thusca region in the Tunisian Tell, to enact a ritualized performance.Footnote ¹ Texts composed in the urban centers of the day portray this era as a period of uncertainty and upheaval, marked by the growing power of competing Christian factions and official orders ending pagan rites. But in a premodern world, state authorities had little direct coercive presence, and imperial hegemony was always enacted and experienced in rather more indirect ways.

The location this group chose had been a place of worship for more than twenty generations. In the first century BCE, as this part of Africa felt the growing authority of Rome in the wake of Punic Carthage’s defeat, worshippers had first come together at Hr. el-Hami to engage in a distinctive offering rite, probably referred to as a molk, directed toward a god invoked in Punic inscriptions as Baal Hammon. Such molks unfolded in a series of discrete steps, a unique ritual chaîne opératoire: perinatal infants and/or ovicaprines were burned on a pyre, remains from the pyre were collected in an urn, that urn was buried in the sanctuary, and (often) a carved-stone stele was erected alongside the deposition. These ritualized actions took place in a particular type of open-air sanctuary marked by rows of deposits and stelae; modern convention, colored by a dash of Orientalism, has dubbed these sanctuaries “tophets” after a biblical site near Jerusalem where “children were passed through the flame.” Similar rites had been performed for more than a millennium, especially at central-Mediterranean sites settled by Phoenician-speaking migrants from the Levantine coast – most notably, at Carthage itself, where thousands of these molk offerings were made for more than 600 years. But by the time the first worshippers gathered at Hr. el-Hami, such practices were almost entirely confined to North Africa. In this rural sanctuary, generations made similar burnt offerings, although the god invoked came to be shown on stelae with anthropomorphized iconographies related to Saturn: a senior, bearded, male deity, his head veiled, holding a billhook (falx) as an identifying attribute.

Although this type of burnt offering seems largely to have ceased in the sanctuary by the late second century CE, those who came to the sanctuary in late antiquity made their own molk. It looked rather different than the earlier offerings on the site (Figure 1.1). In an open-form cooking pot, rather than a closed-form urn, the worshippers placed the burned remains of a lamb or goat, and buried this. Around the cooking pot, they arranged four miniature kraters, each standing around thirteen centimeters tall, capable of holding less than half a liter of material, and sealed with a clay plug; four lamps; and two coins. Coins and lamps were certainly deposited alongside earlier molk offerings – but seemingly not in such a carefully curated manner, to create such a neat tableau. The kraters were custom-made for the occasion at a local workshop; they are otherwise unparalleled in the ceramic repertoires of the region or the period. Similar kraters are, however, depicted on stelae from the region erected centuries earlier. Such stelae may well have been the models that inspired these unique ceramic pieces: a testament to the power of images to shape material and practices.Footnote ²

Figure 1.1 Early fifth-century CE molk-style deposit in the sanctuary dedicated to Baal Hammon at Hr. el-Hami.

Photo: courtesy A. Ferjaoui.

Some of the objects deposited at Hr. el-Hami were already quite old at the time they were buried; the coins had been circulating for two generations. Although the lamps were new – some were never even lit – two of them were made in shapes and styles that reprised much older forms.Footnote ³ Like the rites themselves, the objects were antiques or were modeled on much earlier prototypes to evoke a sense of the distant past, but used in a distinctive though recognizable way. The whole assemblage was not an exact reproduction of any molk that had occurred in the same place centuries before. Instead, it was a creative attempt to reimagine such ancient rites and enact them in the worshippers’ present.

Molk-style rites, and the carved monuments often erected as part of these rites, are one of the most distinctive features of the religious landscape of North Africa under the Roman Empire. Nearly 3,000 votive stelae, directed almost exclusively to Baal Hammon or Saturn, have been excavated or – more usually – found during surface survey at more than 100 sites, from Volubilis in the far west of Morocco to Oea along the central coast of Libya. Molk-style rites are unattested outside of this part of the Maghreb in the imperial period, and both Baal Hammon and Saturn seem to play comparatively minor roles in the lives of worshippers across much of the Mediterranean; the only known temple to Saturn outside of Africa was the one on the Forum at Rome, and around 95 percent of the epigraphic dedications to Saturn from across the Roman Empire come from Africa.Footnote ⁴ The stelae, and the rites of which they were a part, create a distinctive “objectscape” across the Maghreb: something truly regional in geographic terms, something that seems utterly alien to Rome, Italy, and the rest of the Mediterranean world.Footnote ⁵

Because of their archaeological visibility through stelae and seemingly pan-African distinctiveness, molk-style rites and their archaeological correlates have forever been entangled with larger metanarratives of the Maghreb’s place in the wider world: those ideological scaffolds into which fragmentary archaeological evidence from the past is fitted in order to give it significance. Often, the practice of molks over 1,000 years – from the eighth century BCE at Carthage through the fifth-century CE gathering at Hr. el-Hami – continues to be held up as proof of cultural “continuity,” Phoenician religious “survival,” or African “permanence” in the region at least from the first millennium BCE down through the Arab conquest. The rites, the god(s) to whom they are directed, and the people who enact that worship are effectively removed from history, frozen in time. A change in imagery on the stelae, the appearance of Saturn, is taken as a kind of meaningless veneer atop a fundamentally unchanging god and his similarly unchanging worshippers. Other aspects of life and experience in Africa under Roman rule – for example, political and military control, the development of new urban forms, and the economic boom that undergirded these – are all seen as dynamic and changing; religion, used synecdochally for culture, by contrast, is made static. But the deposit at Hr. el-Hami immediately shows this idea of permanence to be nothing more than a fairy tale: every ritualized event was a creative reimagination and remaking of tradition, rather than a mechanical reproduction of past acts.

This is a book about such molk-style rites – or rather, about the traditions that were woven around and through such practices by stelae, about how stelae did the work of creating such traditions and helping imagined communities group together, about how images and spaces configured the shapes and boundaries of those communities through their acts of signification, and about how these objects made North Africa and its inhabitants part of the material hegemony of the Roman Empire. Molk-style rites were not some static, unchanging, and essential feature of North African peoples. Instead, these practices were largely semiotic blanks, signs whose ability to represent and create meaning was fluid and negotiable, constrained mainly by the visual and spatial signifiers woven around them by carved monuments. This shared signifying package – a meshwork of actors, acts, and images – worked to make sense of and to reproduce changing configurations of authority, prestige, and power, as the social forms of empire were lived, experienced, and produced in miniature within provincial towns and villages. That, after all, is the power of tradition: It can be the instantiation of social power and the medium of social change without the overt appearance or even the intentionality of doing either.

In recognizing the mutability of these traditions, in historicizing the objects and actors that have been seen as part of a monolithic and ahistorical North African culture, this book is also an attempt to take a step toward decolonizing narratives of culture and religion in Africa. Decolonization is, of course, a deconstructive practice – or perhaps even a destructive one. It demands dispelling not only the essentializing culture histories that continue to structure how we think and speak of (ancient) Africa, but also the methods, practices, and intellectual frameworks by which those narratives have been made and perpetuated – in finding new ways of knowing and writing counternarratives. Decolonization requires recognizing the perpetual entanglements between pasts and presents, between the material and the imaginary. This can be done only when we move beyond the realm of ideals and abstractions and place our histories and archaeologies on more explicitly material footings. And so this book will challenge the essentializing and ideal culture histories that continue to dominate the study of North Africa and its religions in antiquity, drawing on pragmatic semiotics to build instead a material, social history of signs, practices, and the making of imperial hegemony.

The Past Colonial: African Metanarratives and Marcel Le Glay (or, Why We Still Need to Decolonize)

Colonialism – as both a fact of material domination and as those more insidious ways of shaping knowledge production – has always defined archaeological and historical studies in North Africa. Despite more than fifty years of calls to decolonize studies of the North African past,Footnote ⁶ and of the Roman provinces and the field of archaeology more generally,Footnote ⁷ the practices and frameworks of colonialism remain deeply embedded in the field. Missions of archaeological exploration were tied to building knowledge of and thus control over the landscape.Footnote ⁸ The selection, excavation, recording, and curation of particular sites and objects – especially those tied to the Roman army and agricultural infrastructure – were mobilized in support of state-and-settler military and economic endeavors.Footnote ⁹ The institutions that defined cultural heritage and the personnel who organized and disseminated archaeological research may have had diverse agendas, but most explicitly or implicitly worked to legitimize European control of African territory and its inhabitants.Footnote ¹⁰ The privileging of Latin epigraphy as the primary source of knowledge about the past not only throws focus on urban, elite, Latinate monument-makers, but deliberately disenfranchises the modern peoples living alongside but unable to read these texts.Footnote ¹¹ But perhaps even more subtly, the wider metanarratives and conceptual categories into which the material past was put stem from particular colonial ways of knowing and connecting material objects to some wider significance.

Nowhere is this clearer than in the study of molk-style rites, religion, and the worship of Saturn. The experiences, concerns, and stories told about French colonial rule in Tunisia and Algeria are so deeply entangled with imaginations of the ancient past as to be their fundamental structuring principles. At the heart of most studies of religion and culture in the region lies a project of essentializing culture-history, built from the kind of metaphysical idealism that continues to shape the data, practices, methodologies, questions, and conclusions that govern the study of ancient North Africa. Challenging this, and proposing a material alternative when looking at stele-cults, offers a path forward not only for North African studies, but for Roman provincial archaeology more broadly.

The entanglements of colonial practices, ways of knowing, and the interpretation of ancient cult in Africa emerge most clearly in the work of Marcel Le Glay, a French historian and archaeologist who took up a post in French-controlled Algiers in 1949 and rose to become adjunct director of the Algerian Antiquities Service in 1955 until the violence of Algeria’s war for independence led him to return to France in 1961. Le Glay’s work serves as the basis for nearly all modern accounts of religion in Roman Africa, and is regularly held up as a methodological and interpretive model in North African studies.Footnote ¹²

It is hard to overestimate the impact of Le Glay’s Saturne africain (Reference Le Glay1956–62) on subsequent scholarship dealing with the Maghreb. Although technically focused on material related to one deity, over the course of his two-volume catalogue of thousands of stelae, inscriptions, statues, and anything else that could be related to Saturn, and his third-volume historical synthesis, Le Glay painted a much wider picture of the Maghreb’s cultural history and place within the world. In offering the sort of grand narrative that has since fallen out of favor in postmodernity (but upon which imaginations of the past still depend), Le Glay provided a template for writing the cultural history of the Maghreb; most accounts of the region since have implicitly or explicitly accepted the central tenets of Saturne africain. Le Glay’s picture of the cult of Saturn has become a kind of synecdoche, a model in miniature for the cultural history of Roman Africa.Footnote ¹³ Le Glay was, of course, not the first to try to focalize the culture of ancient North Africa through the cult of Saturn; he largely echoes earlier accounts that established many of the boundaries of investigation.Footnote ¹⁴ Still, it is Le Glay’s oeuvre that is most often cited today, largely on the authority of his seemingly empiricist catalogues of material. In the generations since, almost every work that tackles religion in Roman Africa builds directly upon Le Glay’s foundations – sometimes slotting new data into the narratives he built, sometimes simply applying his model to other deities, sometimes extending his conclusions with nuance – but rarely challenging the central conclusions or the approach.Footnote ¹⁵ Yet decolonizing the study of ancient Africa requires recognizing the intellectual and methodological underpinnings of such work, disrupting them, and building alternative narratives grounded in other ways of knowing.

Le Glay’s central argument is deceptively simple: The wide spread of the sacrificial rite involving votive stelae and especially the god toward whom it was directed, a near-supreme indigenous African god who borrowed a number of traits from the Phoenicio-Punic Baal Hammon, and who “became” Saturn in the imperial period, provided clear evidence of a unity of belief across North Africa: A common Africanness ran from Morocco to Libya, which later Roman-style trappings could only faintly paper over. The essence of the god – the core being that defined Baal Hammon/Saturn – was unchanged as a result of Roman domination. The observable material changes in the cult – new names and images for the deity, adjustments in sacrificial practice – could be discounted as mere veneers.

The full scope and import of Le Glay’s arguments become clear in the rhetorical flourish of his conclusion: “The religion of Saturn … stopped deep Romanization … At the head of the African pantheon, Saturn thus bore some part – and not a negligible one – of the responsibility for Rome’s failure in Africa and the return of the region to the Orient.”Footnote ¹⁶ For Le Glay, the cult of Saturn both illustrated and explained the longue durée patterns of cultural history in the Maghreb. The “religion” of Saturn, “dieu nationale” of the Africans, bore all of the characteristics that Franz Cumont had, a generation earlier, seen as characteristic of “Oriental cults”: It was nearly monotheistic in its exaltation of a supreme deity; it involved a kind of mystical belief with strong astral components; it had ethical expectations of its worshippers; and it was preoccupied with personal salvation, often achieved through mystery rites.Footnote ¹⁷ These Oriental traits diffused into the empty space of a primitive Berber belief in an abstract, natural spirit (of which, it should be noted, there is absolutely no positive evidence!), providing personality and organization that were lacking in an earlier evolutionary stage.Footnote ¹⁸ The supposed Phoenician origins of Baal Hammon let Le Glay add Baal Hammon/Saturn to this basket. In this case, Le Glay’s insistence on the Orientalness of Saturn and the god’s worship in Africa led to a number of tortured readings of evidence (many of which will be challenged here), but it also gave him an explanatory mechanism for later patterns of religious history in the region. The cult expressed, made visible, and structured an African form of spirituality that made North Africans less resistant to the novelty of Oriental, monotheistic Christianity in late antiquity, but it was ultimately this “permanence of the religious psychology of the Africans” and the “impossible Romanisation of souls” that made North Africa “return” to the Islamic Orient rather than stay part of Roman Europe.Footnote ¹⁹ Variations on these conclusions are common in works on North Africa, its culture, and its religion.

Le Glay’s work was, of course, itself situated in a particular moment and circumstances – ones that remade a version of ancient pasts via the experience of his colonial present. Its main narrative direction seems to have been shaped directly by the end of French domination in Africa. Largely researched and written in the midst of the violence of Algeria’s war of independence from France (1954–62), and published after Le Glay had joined the stream of émigrés returning to France, the whole conclusion reads as a kind of apology for France’s failure to Europeanize the Maghreb.Footnote ²⁰ Not even the Romans could change African mentalities and bridge a fundamental binary in thought; how could the French be expected to succeed? The dialectic between the historiography of Rome’s colonization of the region and France’s own colonial projects in the present has always been a deeply ingrained feature of scholarship on North Africa, shaping both the practice of archaeology and assuring its conclusions.Footnote ²¹ That Saturne africain partook in this effort to colonize antiquity with the dynamics of its own present is not surprising; what may be more striking is how different the conclusions about the unchanging indigène sound than those Le Glay himself had reached a decade earlier, when he explicitly recognized the “plasticity” and adaptability of religion in ancient Africa.Footnote ²² The changing political circumstances of French empire played a direct role in reshaping narratives about Rome’s empire in Africa. The question ultimately guiding and fixing the telos of Saturne africain – why did the Roman and French empires fail to keep hold of Africa? – was very much a product of France’s “failure” to maintain its overseas territory. And the answer to this question lay in something abstract and intangible, but made visible in the material record of stelae, statues, and inscriptions: mentalities.

Le Glay’s tripartite division of African Saturn’s personality – an indigenous god fusing with the Semitic Baal Hammon and then papered over by a Roman veneer of name and image – not only serves as a metaphoric model for North Africa’s cultural history in the longue durée, but simultaneously draws on and perpetuates a narrative born of nineteenth- and early twentieth-century colonial ethnography. As ethnographers, historians, and administrators generated schemata to understand lands and peoples brought under French control – with the violent conquest of the former Ottoman regency of al-Jaza’ir from the 1830s onward, and then the official absorption of Tunisia as a protectorate in 1881 – they divided the inhabitants of the land into three main cultural groups: indigenous “Kabyles” or “Berbers” (collated into a “people” by ibn-Khaldoun’s medieval history and its 1858 French translation in a way that directly collapsed millennia of history), Arabs from the Orient, and European colonists.Footnote ²³ Le Glay was not the first to retroject these three monolithic cultural blocks into the distant past, as permanent and unchanging cultural mentalities, nor was he the last. Nearly every introduction to North Africa under the Roman Empire begins by describing the cultural and religious history of the region as a series of distinctive layers, a clear stratigraphy atop indigenous bedrock, born from successive waves of colonial migration and foreign domination in a predictable historical pattern: Oriental Phoenician, European Roman, Oriental Arab, European French. Through all of this, a nation of indigenous Libyans (as the Greeks called them), Africans (as the Romans and Europeans called them), or Berbers (as the Arabs called them) persisted; that these terms are so often used interchangeably in modern scholarship, in ways that elide the divergent perspectives of outsiders over millennia of history, simply demonstrates the ways that groups of people are denied history. And even when not elided directly, works like Le Glay’s account of ancient “Gauls in Africa” do this same work even less subtly.Footnote ²⁴ Imaginations of the past and of colonial presents are made to resemble one another, eliding chronological gaps to focus on permanent, intangible essences of peoples, manifested in patterns of thought and ultimately in material culture.

These essentializing culture-histories that actively dehistoricize the region and its inhabitants are so deeply baked into the study of North Africa that even works that explicitly aim to take postcolonial or anticolonial perspectives often draw on the exact same conceptual frameworks.Footnote ²⁵ Writing one of the few studies of the ancient Maghreb that has resonated in postcolonial studies more broadly, Marcel Bénabou sought to return agency to native North Africans by outlining the variety of ways that they resisted Roman domination – with armed rebellion, but also through the Saturn cult. Instead of the failure to “Romanize” the cult, Bénabou switches the emphasis and value and uses Le Glay’s conclusions to demonstrate the “Africanization” of Roman Saturn.Footnote ²⁶ The worship of Saturn through votive stelae was not a mark of cultural survival or passive permanence – a fossil impeding evolutionary development – but rather a vital means of rejecting Roman models and authority. And it is not Saturn who keeps his worshippers forever Berber, but rather his Punico-Berber worshippers who maintain their cultural uniqueness. While the shift toward human agency rather than passive psychology is a welcome turn, ultimately Bénabou’s narrative is – as Yvon Thébert pointed out immediately in reviewing the work – simply an “inverted history.”Footnote ²⁷ In a similar vein, more recent work has used the cult of Saturn as evidence of Punic survival and a lack of cultural Romanization, valorizing the maintenance of pre-Roman cultures,Footnote ²⁸ or casting stele-cults as evidence of the hybrid blending of Near Eastern and African culture in ways that echo contemporary political attempts to balance pan-Arabism and African nationalism in Tunisia.Footnote ²⁹ Regardless of the value ascribed to it, the narrative of a relatively unchanging, national, indigenous cult runs throughout North African studies.

The repetition of common conclusions and the maintenance of colonial culture-historical narratives even in avowedly anticolonial scholarship hint at the way North African studies work within a kind of hegemonic framework. Particular categories, ways of knowing, and metanarratives provide enduring structures into which material is fitted. But nearly every aspect of this framework arises from deeply problematic conceptual foundations.

Problematic Ideals: Ethnicity, Religion, and Romanization

Were these categories and metanarratives not so deeply embedded in the ways stories are told of the ancient Maghreb, they might be easy to dismiss in the wake of recent scholarship not only in the humanities and social sciences, but even in accounts of nearly every other part of the Roman world. The ideas about ethnicity, religion, and culture that still have wide currency in North African studies are themselves testaments to how deeply colonial frameworks have become engrained in the field, how they have become axiomatic. To paraphrase Bourdieu, these categories go without saying because they come without saying.Footnote ³⁰

The notions of culture and identity underlying most accounts of ancient North Africa, including Le Glay’s, stem from a problematic notion of ethnicity. These posit that groups of people (Libyan, Phoenician, Roman) are predetermined by shared origins, ancestry, and homelands and possess a shared national way of thinking and conceptualizing the world. Ethnicity, narrowly defined, determines how people think and act individually and collectively. This focus on origins and immaterial mentalities has repeatedly been shown to be a product of particular nineteenth-century European forms of nationalism forged in a laboratory of colonial encounters, a myth that crumbles when confronted with the host of other ways individuals might group together, act together, or draw boundaries among themselves.Footnote ³¹ Identities can be (and were) multiple, fluid, and contextual. Situating oneself in the world was a process of triangulation, of imagining, experiencing, and being freighted with linkages and disconnections. Particular axes of identification can be foregrounded or ignored as situations (or actors) demand; it is not clear that “being African” was ever of primary concern to the people who inhabited the Maghreb, much less that they had some collective consciousness waiting to be discovered.Footnote ³² Indeed, David Mattingly has recently argued that identities, concerns, and lifeways in ancient Africa might better be conceived along the lines of urban, military, and rural communities: a positive step away from ethnicity, but one that still imposes fixed (and sometimes ethnically colored) categories atop the ancient peoples of Africa.Footnote ³³ Moving beyond ethnicity as the defining feature of cult in North Africa demands looking again at how worship could be part of an active strategy of “grouping together,” of actively defining, experiencing, categorizing, and organizing communities.Footnote ³⁴

And this brings us to the second fundamental problem in how worship is studied in North Africa: as religion, a cultural correlate of national mentalities and centered upon a particular national god. Both religion and culture are often bracketed from political, economic, and social histories in North Africa, whether shunted to separate chapters of synthetic works or simply juxtaposed as something rather different. The very title of Saturne africain centers a god as the primary object of inquiry and establishes him as a national, African deity, but this has long been a tradition of accounts of religion in North Africa. After all, the notion of national gods as the defining feature of religion in the Roman provinces was most strongly articulated by a North Africanist, and almost every account of cult in the region looks at the worship of individual deities in isolation.Footnote ³⁵

Yet religion in a deeply interconnected Mediterranean world imagined as full of gods was not confined to the realm of internal mental processes. Nor, in a polytheistic world, was it particularly theocentric. And it was not bracketed from a host of other human activities. Seeing religion as such depends again on frameworks developed in nineteenth-century, usually Protestant, European encounters with Others.Footnote ³⁶ The multiplicity of alternative approaches to religion in the Roman world is dizzying: seeing religion as ritualized practices that structure communities of gods and humans, where “to do is to believe (faire, c’est croire)”;Footnote ³⁷ as a form of symbolic, communicative discourse; as “lived ancient religion,” the dynamic set of experiences and actions shaped and reshaped by individual religious agents;Footnote ³⁸ as the discrete relational nodes where particular flows of materials, persons, and competencies might intersect, resulting in a particular act or object.Footnote ³⁹ While these approaches might differ in many ways, they all agree that religion was situated and mediated, embedded in a host of other social practices and relationships.Footnote ⁴⁰ The molk-style rites and stelae of the Maghreb did not exist as a form of religion isolated from other human activities, and so they demand to be resituated rather than ascribed to unchanging ethnic modes of thought.

And there is a third problem, of presupposed teleologies that determine the direction of narratives about the past, especially in terms of “Romanization.” We have already seen how Le Glay’s story of Saturn was predicated on the experience of Algerian independence, the outcome of “failed Romanization” dictated by the abandonment of the French département. The stories told of Africa under Rome, like those of many former parts of the empire, are almost always told by measuring the success or failure of reproducing models from central Italy, or of becoming part of a homogenously conceived Europe.Footnote ⁴¹ Not only does describing individuals, groups, or the styles and iconographies of objects as “Romanized” (or not) remain commonplace, but “Romanization” itself is often cited uncritically as an explanatory mechanism for observed changes through time, just as a lack of observable change is explained as “resistance.”

Rehashing the myriad critiques of Romanization models that have become commonplace in Roman provincial studies is hardly necessary here.Footnote ⁴² The term is used in so many different ways that it has become an empty signifier; even within North African studies, the term began as a descriptor of official state policy around the juridical promotion of towns – a far cry from the ways Francis Haverfield and most accounts of the northwest provinces focus on Romanization as a cultural process.Footnote ⁴³ The model often presupposes neat binaries (Roman versus non-Roman),Footnote ⁴⁴ when “Roman” was itself a discursive category of identity that may have only occasionally mattered to ancient people in certain contexts.Footnote ⁴⁵ And Romanization has no more causative value than older notions of acculturation; the concept offers neither a useful descriptive yardstick nor a causal explanation. A new structure for telling the story of Africa under the Roman Empire is necessary.

All of these problems are not only entangled with notions of nationalism that were defined alongside European colonialism, but can be diagnosed as the interrelated symptoms of a fundamental aspect of how North Africa continues to be studied: through the lens of epistemological and metaphysical idealism. This kind of idealism sees the material world as a reflection or representation of something deeper; material and media are secondary predicates to the immaterial ideals they represent, which allows them to be dismissed as unimportant veneers. The ultimate goal of inquiry is to excavate through the material world and arrive at things that exist beyond these physical reflections: ethnicity, mentality, religion, a god. But these abstractions are, as we have seen, chimeras constructed as part of and mobilized for colonial control. What is needed are new models that are not predicated on idealism, but are instead grounded and situated in the material past.

Decolonizing: Two Paths Forward on Theoretical Ground

Two closely related paths forward offer the opportunity to reevaluate, challenge, and shift the narratives and conceptual frameworks in ways that move beyond the colonial hegemonies that shape the study of the ancient Maghreb. The first is to move from the colonialist idealism that continues to sit at the heart of North African studies and toward a more materially oriented epistemology. Doing so can build upon wider movements in Roman provincial archaeology and the material turn in the humanities more generally. This involves replacing the kind of essentialism that haunts studies of North African culture with a focus on relationality, re-historicizing peoples, practices, and forms of worship by situating them in time, space, and a host of other social entanglements. But it also means recognizing the ability of objects like stelae to act within and shape the world. While the move to materialize Roman provincial studies and religion has gained ground, this has often been along the lines of New Materialism, in ways that downplay both the asymmetrical relationships that structure empire and the signifying power of objects. Instead, I suggest that the second path forward involves drawing on pragmatic social archaeology, which allows for a better relational modeling of material and social power in colonial contexts.

First, rather than seeing the material objects generated through archaeology as a kind of epiphenomenon that papers over “real” mentalities, we can begin by acknowledging the central role material plays in worlds, past and present; there are no mentalities without material, for humans think with and through things.Footnote ⁴⁶ It has become commonplace to call for materializing studies of the past, especially in the Roman Empire, and to a lesser extent, in the study of both ancient and modern religion.Footnote ⁴⁷ But what would materializing religion in ancient Africa entail?

Materially oriented approaches almost all share a commitment to moving beyond essentialism and the kind of categorizing “container thinking” that persist in North African studies and within other strains of Roman archaeology more broadly, and that are often (as we saw) closely linked to colonial ways-of-knowing.Footnote ⁴⁸ Instead of fixed categories, they emphasize relationality: everything exists and acts within a constellation of relationships. These relationships are never fixed, but fluid and mutable; this is as true of discursive constructs like identity or even personhood, which could be constantly and contextually triangulated, as it as of the flows of objects that connected places and individuals in a globalizing Roman Mediterranean.Footnote ⁴⁹ These relations are not predetermined or teleologically fixed, but emergent in ways that demand historicization, and always malleable according to actions of particular agents.

Adopting a materialist approach also means recognizing the ways that “things” – in this case, carved stelae, ceramic vessels like those deposited at Hr. el-Hami, the built structures where ritualized acts took place – could act as such agents, could have an ability to shape their worlds rather than only passively represent something beyond themselves. Of course, the precise nature and contours of this object agency, and how it relates to human agency, are deeply contested within archaeology and beyond.Footnote ⁵⁰ In Roman archaeology, recent material approaches have built on the New Materialism of figures like Bruno Latour, John Law, and Michel Callon. Many of the approaches that embrace actor-network theory (ANT) or assume that relationships between people and agentitive things are largely symmetrical: that nonhuman actants have the same degree and kind of agency as human actors.Footnote ⁵¹

The first danger with this sort of approach is that, perhaps primarily in a fit of polemic radicalism, it can overemphasize flat, horizontal, and equal relationships.Footnote ⁵² In such works, people and things are ontologically the same and can act on each other in the same manner as equals. Indeed, these ANT approaches often explicitly downplay or villainize hierarchical and asymmetrical relationships, especially among persons. Taken to extremes, this can lead to (what one hopes are hyperbolic) calls to “unite in a defence of things … those subaltern members of the collective that have been silenced and ‘othered’ by the imperialist social and humanist discourses.”Footnote ⁵³ Those who have suffered violence and inequality through imperialism are not humans, but nonhuman things; the asymmetries in human:thing relations deserve greater attention than those among human actors. Miguel John Versluys has even identified the main problem hindering development in Roman archaeology precisely as the dominance of postcolonial theory and its own privileging of unequal power dynamics.Footnote ⁵⁴

But the Roman world was a world of deep inequalities, of lived hierarchies, of systemic violence, and of discrepant experiences.Footnote ⁵⁵ Postcolonial work has certainly shed new light on those facets of antiquity, moved us away from sanitized and valorized imaginations of imperial splendor, and drawn attention to the ways that representations of the past in historical and archaeological accounts are implicated in perpetuating other, contemporary inequalities. Postcolonialism did not manufacture those hierarchies or asymmetrical relations; it simply calls on us to recognize and account for their ubiquity historically, historiographically, and in our own practices.

The great insight of postcolonialism writ large is that “culture” is as much a part of making imperial power as military domination, economic exploitation, and administrative control.Footnote ⁵⁶ Controlling individuals and groups – whether in ancient Latium, Africa, or Britain, or in modernity – involved developing particular constellations of social power in multiple dimensions that rarely acquiesce to being categorized neatly.

It may be better to think about imperial power in terms of hegemony: control over myriad forms of production, including symbolic production, in ways that can shape and reshape habits.Footnote ⁵⁷ For Edward Said, hegemony is that “system of pressures and constraints by which the whole cultural corpus retains its essentially imperial identity and its direction.”Footnote ⁵⁸ The totality of culture is touched by and flows toward imperial control. Here, Jean Comaroff and John Comaroff’s anthropological understanding of hegemony might offer more room for countercurrents, resistance, and alternatives in their juxtaposition of ideology and hegemony. Building from their reading of Antonio Gramsci, they posit that ideology might be considered a dominant view, but one still contestable; hegemony, by contrast, is comprised of those things made so natural as to be beyond contestation, “that part of a dominant worldview which has been naturalized and, having hidden itself in orthodoxy, no more appears as ideology at all.”Footnote ⁵⁹ Or, put another way, those “signs and practices, relations and distinctions, images and epistemologies drawn from a historically situated cultural field that come to be taken-for-granted as the natural and received shape of the world and everything that inhabits it.”Footnote ⁶⁰ Still, it is telling that the Comaroffs, despite their material inclination, separate signs from practices and cast ideologies and hegemonies as operating through the material and embodied world. Similar uneasy pairings of “images and ideologies” – loosely related but somehow separable – can be found throughout current studies of ancient material and visual culture.Footnote ⁶¹ Indeed, the challenge for archaeology is to move beyond immaterial (un)consciousness and ground the operation of hegemonic power more firmly in material and relationships.

Understanding the Roman Empire thus demands grappling with how material culture was implicated in ordering the world via the creation of unequal forms of social power, rather than collapsing those relationships into a host of symmetrical interactions and object-flows. Any path forward in Roman archaeology, especially when concerned with people and things on its margins, must offer to do that.

The other danger with these ANT-related approaches is that by making representationality the archnemesis of archaeological studies, they minimize and ignore a major dimension of how material things work. As Stratos Nanoglou suggests, “representations are as material as anything else.”Footnote ⁶² This may be obvious in the case of something like a carved votive stele, set up as a billboard to communicate and signify, to stand in for something other than itself – although what and how it signified might change radically as similar-looking objects were situated in new fields and relations. But even beyond figured objects, the late antique deposit at Hr. el-Hami only makes sense as a representation and recreation of earlier rites at the site. The practice of any tradition always represents and signifies earlier acts. Representationality is a fundamental quality that emerges in the relationship between humans and objects, presents and pasts.

Rather than indicting representationality as a whole, it may be better instead to rethink how objects – and figured objects in particular – might represent. The problem with the approach that Le Glay and others take is not in seeing objects as reaching for something beyond themselves, but rather that they imagine a constant, 1:1 relationship between signifier and meaning. A “Classical” style on a stele signifies “Romanization,” rather than, say, access to dedicated stone carvers trained in particular conventions. Worship of Saturn always signifies belonging to an imagined African nation, rather than, say, belonging to a particular family or group of laborers. A falx always stands for a particular god of viti- or oleiculture, rather than the labor that generated a worshipper’s offering. It is this narrow conceptualization of representation, where meanings are passive, singular, and fixed, that is the problem – not representationality itself. What is needed is a more robust account of how representational signification itself might work.

The binary between representation and material agency is not only a false one, but one that reinscribes the very categorical dichotomy between mind and body, thought and material that the material turn seeks to overcome. Moving “beyond representation” by denying or downplaying the ways that signification happens in the relationship between persons and things is simply creating a théorie inversée. In fact, it is precisely because they represent that objects can have power and agency, that they can shape relationships and make society.

And here, pragmatic semiotics can offer a means of critically assessing, rather than dismissing, the representational potential of objects, while also recognizing the potential impact that these objects can have in the world. Rarely used in studies of Mediterranean archaeology,Footnote ⁶³ pragmatic semiotics stems from the philosopher work of Charles Sanders Peirce, and has been explicitly and productively embraced for use in archaeological work by Robert Preucel, Zöe Crossland, and others.Footnote ⁶⁴

Like any semiotic theory, Peirce’s pragmatics is concerned primarily with the making of meaning from signs – an interest shared with the full range of interpretive archaeologies, but one that moves beyond reducing things to texts. And like the wider project of New Materialism, Peirce takes aim at the binary divide between mind and body, ideal and material that had dominated ways of understanding since Descartes.Footnote ⁶⁵ For Peirce, anything can be a sign – a word, an object, an event – as long as it points or refers to something beyond itself, the object (in Peirce’s terms) or its referent. This relationship between sign and signified might take any number of primary forms. It might be iconic, sharing specific, observable qualities and relating through resemblance (a photographic portrait that looks like its subject). It might be indexical, related by existential fact or causation (smoke caused by and thus signifying a fire). Or, it might be symbolic, related solely by observable convention – this is the kind of signification that New Materialism most often takes issue with.Footnote ⁶⁶ And for Peirce, a single sign could simultaneously reach out beyond itself via several of these relationships; the toga worn by a figure carved on a stele might iconically resemble a woolen garment known from lived experience, but might also have symbolic relationships, as a convention of juridically holding Roman citizenship.

But where Peirce’s work departs from most semiotic frameworks is in its rejection of a simple dyadic relationship between sign and significance – the kind of simple equivalency in representationality that underlies accounts like Le Glay’s and that Versluys and others rightly reject. Peirce adds a third element to this relationship, the interpretant: “an effect upon a person” wrought by that sign.Footnote ⁶⁷ The sign is allowed agency because of the relationship between signifier and referent: it exerts tangible effect on human interpreters through the interpretant. Although Peirce’s concept of interpretant changed through time, it is usually treated as another sign, opening a potentially endless chain of semiosis through myriad signifying relationships.Footnote ⁶⁸ Every signifier might be pregnant with endless possibilities, and every signifying event in this infinite chain of semiosis could reach for a new interpretant; what limits this potentiality is a community of interpreters who develop (or impose) particular privileged relationships between signs and interpretants. Signifying is a practice; like a Bourdieuvian habitus, particular interpretants can be learned as norms and socialized as expectations implicitly. Because of this, signs could be implicated not only in representing a community, but in making and defining its boundaries.

That material signs could have social ramifications is, of course, not a novel claim. Ian Hodder argued long ago that material symbol-systems were not merely representative of but constitutive of social relations, drawing on Anthony Giddens’s notion of structuration as the means of bridging situated agency and wider social structures.Footnote ⁶⁹ In anthropology, Webb Keane has argued that representations can be acts and things that have tangible impacts within the world, while in archaeology, Carl Knappett has argued that both the pragmatic and significative dimensions of objects need consideration.Footnote ⁷⁰ And Jeremy Tanner has explored how the style of Roman Republican portrait statues might be indexical and performative in ways that establish social relationships among the parties involved in their creation and consumption.Footnote ⁷¹

With the recognition that signs and the practice of signification could shape the relationships among people, asymmetries of power can once again reenter the picture. The ability of signs to build imagined connections, generate other signs, and effect people is a social process. As Robert Preucel writes, “The control of [signification] via strategic action permits the fixing of meanings, as sign combinations come to be interpreted together as semiotic ideologies.”Footnote ⁷² That is, the production of signifying habits – how an interpreter moves from signifier to interpretant and relates signs – is very much entwined with structures of authority. The habit of seeing Saturn-stelae, their images, and inscriptions as signs of African nationalism has been naturalized within a community of historians and archaeologists by its repetition.

North African studies, especially when dealing with the sign-heavy realms often dubbed “culture” and “religion,” have long been grounded on essentializing idealism bound up with colonial archaeologies, metanarratives, and the privileging of particular semiotic habits. Studies on the worship of Saturn are perhaps the most obvious example of this, but that also makes the stelae and sanctuaries of the region an ideal place to develop and test new approaches to Roman provincial archaeology.

A pragmatic semiotic approach to material culture in the Roman world offers a means of moving beyond the essentializing idealism bound up in the colonial archaeologies, metanarratives, and semiotic habits that still undergird much work on the Roman provinces, especially North Africa. Pragmatic semiotics allow for the agency of all objects – including figured ones – without surrendering a focus on human subjects discarded by New Materialism, or downplaying the asymmetries of power in which the material world was directly implicated in creating.

The Path Ahead

The asymmetries of power that emerged in the material world of the Roman Empire left no relationship – among persons, material signifiers, and interpretants – untouched. Even a form of practice like stele-erection, or gods and iconographies that seem regionally distinct and continuous from the Iron Age through the Late Empire, became entangled with, reproduced, and naturalized those forms of relationship that constituted imperial hegemony. Instead of a monolithic “religion of African Saturn,” stelae and associated worship practices were variegated, emergent practices, which made and remade a host of connections: between worshippers and gods, between both face-to-face and imagined human communities, and between humans and things. Instead of resistance to empire or Berber permanence, these practices not only participated in but also forged and naturalized a vertically structured, centralizing system of signs and social authority. This, ultimately, was the hegemonic force of being within the Roman Empire.

But these central arguments rest on a host of smaller claims and shifts in the categories of analysis used. Chapter 2 begins by re-historicizing stelae and how they came to be, as material objects in time and space. The adoption of molk-style rites and carved-stone monuments was not part of a monolithic and unchanging religion, but can be seen as part of four distinct historical moments. The limited adoption of these practices in the early Iron Age was entangled with the movement of peoples in migration and resettlement. But the production of stelae and molk-style rites in the first centuries BCE/CE, when groups across Africa adopted molk-style rites and the practice of dedicating stelae, was a distinctive phenomenon. This tophet boom was quite different than what had come before, and from what came after, as migration and movement tied to the Roman army drove a second boom in new tophet-like sites. In every period, though, it is possible to recognize how stelae were not simply part of some intangible religion or culture, but entangled in a host of material, social, and productive practices, from migration to stone-working industries.

Part II shifts from a diachronic account to thematic studies of North African stele-cults and their entanglement with imperial hegemony. Chapter 3 focuses on stelae erected in the long first century BCE and the indeterminacy of their signs, arguing that communities created a common set of signs across the region that resembled a colonial Third Space. These stretched far beyond stelae and molk-style rites, again pointing to the ways that cult was never an isolated phenomenon, but deeply situated. Material signifiers were as important – if not more important – than significances in creating those connections, and ultimately in creating a new cultural geography of Africa. Rather than veneers that can be dismissed as epiphenomena, these signifiers had the power to create imagined communities, and these communities had the agency to privilege different interpretants.

Chapter 4 examines the way that this indeterminate system became more and more determined in the second and third centuries CE, especially in the ways that a particular deity – Saturn – was constructed. While previous studies have emphasized the equivalency of Baal Hammon and Saturn as the guarantor of cultural continuity in ways that perpetuate colonial ideals, I argue that gods were constructed and made real through their signs, and that the changing signifiers embraced by stele-users foregrounded very different kinds of relationships and operated within different systems of semiosis. Stelae of the third–second centuries BCE made a god present indexically; stelae of the imperial period embraced iconicity in ways that were entangled with empire, including new divine epithets tied to imperial authority and new road systems in the province. And by the end of the second century CE, this iconic system could even work to perpetuate clear social hierarchies. In short, the social structures of empire were made and naturalized in ways that a god was signified.

Chapter 5 turns to the ways that epigraphic and visual signs on the stelae create new forms of grouping together for worshippers in the imperial period, focusing on the way a new subject became the center of stelae: human figures. The role of stelae shifted from marking divine presence to commemorating and individuating figures whose social positions were triangulated according to forms of prestige shared across the Roman Empire. The dedicants of stelae were not “African peasants”; rather, the stelae set these worshippers into a competitive and hierarchical social world through particular attributes, compositions, and titles like sacerdos. In demonstrating that stelae became arenas of social competition for worshippers, it is possible to move beyond the problematic narratives that see Saturn-worship as an Oriental mystery-cult of initiates.

Yet even if nearly every imperial-period stele embraced compositions that foregrounded and individuated persons, Chapter 6 argues they did not all signify the same kind of referent. Although the figures depicted are often labeled the “dedicant,” many adapt iconographic conventions widely shared across the empire to identify the figures shown as children and to individuate them. This is not only a marked shift from Iron Age stelae, but points to the ways notions of childhood and personhood were changing. And here, we may see the strongest evidence of how imperial hegemony was forged from local practices: the futures of children were exclusively imagined as wholly situated within signs and categories of the empire.

Chapter 7 examines changes in ritualized practice around the stelae, and how the images could play a role in naturalizing different configurations of worship practice and their social entailments. While past studies have focused either on supposed changes in sacrificial behavior – especially a teleological evolution from live child sacrifice to the offering of animals, which can no longer be supported – I instead argue that stelae from the only excavated tophet-like sanctuary used regularly from the mid first millennium BCE through the second century CE create a rather different shift. Down to the first century BCE, stelae set molk-rites as an individual transaction with the god, mediated primarily through voice; in the imperial period, stelae begin to reimagine rites as communal activities at an altar, often elevating a single sacrificer. This was not the product of copying Roman visual models, but instead a recasting of priorities and foci in an act. By the second century CE, the stelae had naturalized a different mode of worship, leading to the normalization of new practices in the same sanctuary space.

The ways that stelae and signs reshaped practice and sanctuary society can be seen even more clearly, and far more broadly, in Chapter 8. Here, a study of changes in sanctuary spaces points both to a diversity of forms that resists the cultural labels often freighted upon them, but also to the ways that elite benefaction might serve to create new practices and relations that ultimately set these same benefactors as centers of attention. At the end of the second century CE, a number of stele-sanctuaries were rebuilt in monumental forms that privileged central altars, the spectacle of animal offering, and dining. This shift in the spatial dimension of worship afforded new possibilities of practice and social ordering that closely resemble those of the wider imperial world.

A bundle of signifying practices within worship traditions that have long been seen as static or resistant to Rome changed in myriad ways over the course of the imperial period. These signifiers were part of a tradition that became uniquely African because of the dynamics of empire. And they were full participants in making Africa part of the Roman world.