Identifying Tophets
Despite the word’s appearance in the Hebrew Bible, “tophet” is a strictly conventional term used to group a range of archaeological sites that share certain recurring traits. We do not know what these spaces were called in antiquity—they were probably simply referred to as a bt (house) or qdš (shrine) in Phoenician (Amadasi Guzzo and Zamora López 2013)—and so bracketing these sites from other places of worship already represents an interpretive step.
Tophets are among the most easily recognizable sanctuaries in the Phoenician world because the rite conducted within them has left two distinctive signatures in the archaeological record. The first is a field of buried urns containing burnt bone and/or vegetal remains. The second is the presence of carved stone monuments that take various forms, from cippi to thrones to stelae, erected alongside the buried urn; these often survive as surface finds, reused in later buildings, or culled during ancient cleanings of the sanctuary and buried together in a deposit referred to by the Latin term favissa. When found in large quantities in such secondary contexts, stelae can suggest the existence of an unexcavated tophet-like sanctuary.
The other main commonality through time and space is the god honored by these offerings. Baal Hammon is almost the exclusive recipient named on stelae in Punic, although from the sixth century bce, he is accompanied by his consort Tanit at Carthage (Garbati 2013). In the Roman imperial period, stelae related to tophet-like rites are mostly dedicated to Saturn as worshippers reimagined Baal Hammon within new systems of associations (Le Glay 1966; McCarty 2016).
These signatures have allowed the identification of over 100 tophets stretching from Morocco to Malta (map 21.1). The spread of the rite across this wide area came in at least four distinct phases. The first wave of tophets was established in the central Mediterranean as Phoenician colonists founded settlements in North Africa, Sicily, and Sardinia (D’Andrea and Giardino 2011). In the fifth–third centuries bce, a few new tophets were founded as settlers from these early colonies moved to new homes and brought their rites with them. For example, some residents of Sulcis reestablished the inland Sardinian site of Monte Sirai in the fourth century bce, founding a tophet in the process (Bondì 1995), while the tophet in Numidian Cirta was founded and used largely by the resident Punic population (Benseddik 2012: 32–33). As Punic territories fell under Roman control in the third–second centuries bce, tophets in Sardinia were slowly abandoned. In North Africa, the opposite happened: following the destruction of Carthage, a host of new tophets sprang up, especially in central Tunisia and along the Atlantic coast (D’Andrea and Giardino 2013; McCarty 2018a). Finally, in the late first and second centuries ce, as the institutions of the Roman Empire—especially army recruitment and deployment patterns—drove migration from the coasts into central Algeria, these migrants also founded new tophets where they settled (McCarty 2010). The spread of the rite through time and space was neither automatic nor through simple diffusion but, rather, the result of concrete and diverse factors; historicizing the appearance of these sanctuaries already hints at the ways these sanctuaries might not be part of a monolithic phenomenon.
Map 21.1 Tophets and tophet-like sanctuaries in the Phoenician world.
Source: Map by M. McCarty, with base-map under license from Esri and data provided by Esri, USGS, NOAA.
Reconstructing the Rites
The rite that resulted in the site formations we call tophets is explicitly named on some of the monuments from these sanctuaries: in Phoenician it was called a mlk (usually vocalized as molk), a term that also appears in the Hebrew Bible to describe burned sacrifice, usually of children (Eissfeldt 1935; Stavrakopoulou 2004).
The urns, their contents, and their contexts allow a partial reconstruction of the mlk through interdisciplinary examination (Bénichou-Safar 1988; Docter et al. 2003; Stager 2014). While detailed studies of urn contents are a positive step forward, bringing that same level of scrutiny to the wider excavation of tophet sites is essential to understand the full ritual process; recent excavations of first-century bce tophets at Althiburos and Zita, with focus on the micro-archaeology of cult, may add substantially to our knowledge of the mlk.
The archaeologically attested portions of the rite began with the selection of a main offering: in tophets down to the first century bce, this was almost exclusively one or two infant(s) and/or ovicaprine(s) (i.e., sheep or goats, whose bones are nearly indistinguishable). The ratios of infants to ovicaprines varied over time and from site to site. For example, of the urns from the ASOR excavations at Carthage (ca. 750–146 bce), 54 percent contained only infant remains, 19 percent contained infant and lamb remains, and 12 percent contained only lamb remains (Stager 2014: fig. 26). Over time, the proportion of infants increased (Stager and Wolf 1984). By contrast, at Motya (ca. 725–350 bce), the ratios are almost reversed: 61 percent of the urns contained only animal remains, 8 percent contained humans and animals, and 24 percent contained only humans (Ciasca et al. 1996). While the lamb sacrifices are generally treated as “substitution sacrifices,” with a lamb given in place of an infant, based on third-century ce Latin inscriptions from Nicivibus, the Phoenician inscriptions suggest that the two offerings are distinct practices: the mlk is qualified either as a mlk b’l/mlk ‘dm (mlk of a citizen/person) or mlk ‘mr (mlk of a lamb). Such variations in the object burned suggest that a multiplicity of rites might take place in these sanctuaries, even if the ritual process was formally quite similar.
This offering was placed on an open-air pyre, causing uneven temperatures and burning of the bone. Secondary offerings might be placed on the pyre alongside the main offering: cow, pig, fish, bird, and occasionally wild animal (e.g., deer) bones are also found in the urns. These are always small portions of the animal, though, unlike the infants and ovicaprines on their own. The whole process of burning the offering was lengthy, complex, and almost certainly required professional skill—either a priest or an attendant to build and tend the pyre over several hours.
Once cooled, the bones were collected; the presence of different anatomical parts suggests that this process was somewhat random (Melchiorri 2013: 238). The collection process may also account for parts of multiple individuals in a single urn; although some urns definitely contained two infants, those that contained more than two only have a few fragments of more individuals, hinting that the bone collector might miss pieces that would then be collected from a subsequent pyre at the same location. The remains were then placed in an urn, and occasionally other unburned objects were added, such as jewelry or protective amulets.
The urn was then buried in the tophet space, sometimes set into crevices in bedrock or in small “boxes” constructed of stone slabs, often in organized rows that hint at a sanctuary authority overseeing the practice. A few worshippers marked their deposits with a carved monument. That the number of urns dwarfs the number of monuments has been used to suggest that each represented a distinct rite and kind of offering; however, the hole dug for the urn and into which the monument’s base was set appears to be a single cut in photos of trench sections at Carthage, while at the Hellenistic tophet of Althiburos, the monuments were integral to the stone boxes containing the urns, demonstrating that the stele was erected at the time of the deposit’s burial (Xella 2013). Still, the presence or absence of a monument also reflects variegated and individualized practices, even within a single site: the possible extension of a ritual core that comprised the act of burning and burial.
The Nature of the Rites: Human Sacrifice?
The question that has dominated studies of tophets—often at the expense of other questions—is whether or not the archaeological remains provide evidence for the Punic human sacrifices described by Classical authors (for the Phoenicians in Classical texts, see chapter 44, this volume). This debate has raged since the discovery at Salammbô, as scholars have tried to explain the finds either as live child sacrifice or—especially given the high rates of infant mortality in the ancient world—as a spe
cial burial ground for children where they might posthumously be dedicated to particular gods (e.g., Bénichou-Safar 2010; Moscati 1987; Ribichini 1987).
Untangling the nature of the rites practiced in tophets depends upon different types of evidence: literary testimonia, inscriptions, archaeology, and the bones themselves. Of course, the debate is not strictly empirical, but it has been shaped by wider historiographic concerns, from French colonialism to the emergence of an independent Tunisia, to postcolonialism (for the Phoenicians in Tunisia, see chapter 48, this volume). While no piece of evidence is conclusive on its own, together these sources point to the probability of live child sacrifice (Xella et al. 2013), conducted by individuals seeking specific favors from the gods.
A number of Classical authors from the fifth century bce onward describe Carthaginian child sacrifice (Simonetti 1983). Although the earliest mentions are often simple statements to the effect that the Phoenicians or Carthaginians conduct such rites (Sophocles, Andromeda fr. 122; Ps.-Plato, Minos 315b–c), as tensions between Greeks and Phoenicians heightened and authors adapted more baroque writing styles, these descriptions gained macabre details that heightened the emotional impact and purported “barbarism” of the rites. The third-century bce Greek historian Kleitarchos, for example, writes:
The Carthaginians, honoring Kronos [i.e., Baal Hammon], whenever they seek to be successful in some great thing, accordingly vow one of their children…burning it as a sacrifice to the god. Among them is set a bronze statue of Kronos, its hands stretched out over a bronze brazier which burns the child. When the scorching flames fall upon the body, the limbs contract and the opened mouth appears almost to laugh, until the contracted body slides quietly into the brazier. (FGrH 137fr9)
Such “outsider” accounts of the rites, penned by authors inimical to Carthage and who invoked child sacrifice with specific rhetorical goals, have rightly been questioned. The consistent patterns of burning on the excavated bones preclude the use of a mechanical, baby-dropping statue like the one Kleitarchos describes (Bénichou Safar 1988). Still, even the most fantastical slanders rely on a germ of fact, and baroque embellishment does not preclude truthfulness. Likewise, however objectionable the Orientalist rhetoric of those who first trumpeted the “sacrificial hypothesis” to explain tophet archaeology, their conclusions should not be dismissed out of hand.
Despite the lack of surviving historical or literary accounts penned by Phoenicians, a written “insider’s” view of the rites does survive in the form of the thousands of inscriptions on the stone monuments from Salammbô and the handful of inscriptions from other tophets (Amadasi Guzzo and Zamora López 2013). Although there are minor differences among the inscriptions, CIS I.2.511 illustrates the key components of the dedication formula: “To Lady Tanit, face of Baal, and to Lord Baal Hammon: [that] which vowed Arisham, son of Bodashtart, son of Bodeshmun, because he (the god) heard his (the dedicant’s) voice, he blessed him.” The inscription begins with an invocation that clearly sets the monument—and presumably the associated offering—in a dedicatory, rather than a funerary, context.
Aside from the name of the dedicant—almost always one man (but a few women do appear; Amadasi Guzzo 1988)—the rest of the dedicatory formula makes clear that the rite was a planned act rather than a response to an unforeseen death. The ending formula and invocation of a vow (ndr) clarify the ritual process: the dedicant would have made a contract with the gods to receive some favor; when the gods fulfilled their end of the deal, the dedicant would have offered sacrifice. The planned cycle of vow-favor-fulfillment precludes response to a sudden death. Similarly, dedicants would not have described the death of an infant as a blessing in the closing clause. Yet sometimes the formulae differ, and they look forward to a future blessing: “may he [the god] hear his [the dedicant’s] voice,” implying a different point in the sequence of dealing with the gods. The common archaeological form of the stele (and perhaps even the associated burned deposit) conceals a range of circumstances and moments at which the rite might be conducted by an individual offrand.
The form of the rites, too, suggests that the mlk was distinct from funerary practices and that tophets were not simply cemeteries. While much of the Punic world adopted inhumation in the sixth century bce, the children in tophets continued to be burned (Gras et al. 1991: 136–38). The dearth of excavated infant burials in the Phoenician world should not be surprising—infant burials from all periods are archaeologically underrepresented—and there is some evidence for infant inhumation in family tombs or in shallowly buried amphorae. While there are archaeological and ethnographic parallels for the special treatment of deceased children—whether distinctive rites or their own necropoleis—these often involve less time and expense than resource-draining cremation and monumentalization. Within the Phoenician world, in the Ibiza necropolis, for example, children’s burials received less investment than adult graves (Gómez Bellard et al. 1995). Most of the seeming overlaps between funerary and tophet rites date to the Roman imperial period (including a unique children’s necropolis-sanctuary at Thysdrus), and even these have rightly been challenged (Schörner 2010). The main commonality between tophet and funerary practices seems to be that both involved disposal of a human body, though by different means, in different places (but both on the periphery of settlements), with different assemblages of objects: hardly evidence that the two were commensurate.
Perhaps the most compelling piece of evidence against the funerary hypothesis is the fact that some dedicants offered an ovicaprine rather than an infant, or in addition to a child. At Tharros, in urns with only ovicaprines, the whole animal is attested, implying that the lamb or kid was subjected to the same ritualized practice as infants (Fedele and Foster 1988: 33). In these cases, the rite is clearly sacrificial rather than funerary.
Although bone analysis has been hailed as a silver bullet capable of putting the debate to rest on scientific grounds, the osteological evidence is itself hotly debated (Melchiorri 2013). In analyzing material from the ASOR excavation, Jeffrey Schwartz and his team (Schwartz et al. 2010, 2012, 2017) argued that the deposits at Salammbô could not have been the result of child sacrifice because 38 percent of 540 individuals examined were pre- or perinatal, reflecting the high rate of mortality around childbirth in antiquity. Their assessment was based on the size of the bones, tooth development, and the absence of neonatal lines on the teeth, which form approximately seven days after birth. However, Schwartz also examined the sex of seventy individuals based on pelvic ilia, and found that 37 percent were male and 54 percent female—an imbalance that may reflect active gender selection (historical infanticide tends to target females) rather than natural death. A second look at the same material (Smith et al. 2011, 2013) challenged these age findings, arguing that cremation caused shrinkage that not only reduced bone size but also occluded the neonatal lines, and so less than 1 percent of the infants were actually prenatal. Even with Schwartz’s aging techniques, the majority of the sample (62 percent) is aged zero to six months. The presence of older children—up to around six to nine years (Docter et al. 2003)—also suggests that the space was more than simply a response to infant mortality, and may even confirm Diodorus Siculus’s suggestion that some families bought and sacrificed slaves rather than their own children (Diod. Sic. 14.4). The osteological evidence does not disprove that the infants could be live sacrifices.
Given high rates of infant mortality, especially from conditions that can leave evidence on bone, more detailed analysis looking for pathologies could help solve this debate over child sacrifice. Unfortunately, the fragmentary condition of the cremated bones often precludes such analysis. Still, there are a few tantalizing hints: at least two individuals at Sulcis had evidence of cortical hyperostosis (Melchiorri 2013: 238). For now, though, the bone evidence remains incapable of solving the debate.
That many of the human offerings in tophets were probably the result of live child sacrifice raises questions of its own: What was the occasion fo
r such rites? A number of hypotheses have been put forth. From a functionalist perspective, some have suggested that tophet rites served as a form of population control, intended to reduce pressure on land and resources (e.g., Stager and Wolf 1984; Wagner and Ruiz Cabrero 2007). Such pressures may have existed in the early Iron Age Levant, but in a colonial setting like the Phoenician central Mediterranean—especially in the early days of a colony, when population growth was essential for a settlement’s survival—this was less of an issue.
There is also debate over whether the rites were periodic, repeated according to a calendar, or occasional. The ovicaprine remains in the urns provide approximate dates for the sacrifices, as lambing occurs at specific times of year: when age data is provided for the ovicaprines, they are very young, with most one to three months old, but some three to six months old (Fedele and Foster 1988; Stager 2014: 12–14). Coupled with Hebrew Bible evidence for springtime “first fruits” sacrificial rites in Israel/Canaan, the lamb remains have been used as strong evidence for mlk rites being periodic rather than occasional (Stager 2014; Garnand et al. 2013).
Yet this evidence is not wholly convincing. While the young lambs do provide a rough time of year for the deposits in which they occur, lambing happens twice yearly in North Africa (February–March and October–November); one- to three-month-old lambs could presumably be found at least six months in a given year. At the sites where the ages of the ovicaprines have been ascertained, they occur in only 31 percent (Carthage) to 47 percent (Tharros) of the urns; this is consistent with a pattern of including young lambs when they were available in rites that otherwise happened throughout the year.
Punic inscriptions from tophets give no evidence for specific occasions beyond the general formulae just described; in the Roman imperial period, while most Latin inscriptions are similarly vague, some specify that the rites were conducted for someone’s well-being. Still, the epigraphs stress the personal nature of each offering—made by individuals who had undertaken vows, asking a personal favor of the deity. The rites may thus have been occasional, stemming from personal crises and pressures, rather than systematic and calendrical.
The Oxford Handbook of the Phoenician and Punic Mediterranean Page 46