Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

ISSN 2239-5393

Continuities and Contexts


The Tophets of Roman Imperial-Period Africa
Matthew M. McCarty
(Princeton University, Princeton, U.S.A.)

Abstract
The stele-sanctuaries of Roman imperial-period North Africa have long been taken as prime examples of
both cultural and cultic continuity from the first millennium BC through late antiquity. Rather than seeing
continuity as a neutral redescriptive category in narrativizing the past, it may be more useful to
examine the means by which individual sanctuary communities created connections to their pasts through
their practices and spaces, and of what these pasts may have consisted. In order to do so, I examine two
of the type-cases for the types of transformations seen in the religious life of Roman Africa: the tophets
and later temples at Hr. el-Hami and Thugga, where the material remains draw into question traditional
accounts of clear continuity.
Keywords
Tophets, North Africa, Saturn, Roman, continuity, ritual.

Du Ve sicle av. jusqu la fin du IIe sicle ap. J.-C., cest donc le mme
dieu quont vnr les Africains, bel exemple de continuit que na
nullement entame limplantation romaine. (LE GLAY 1966b: 13)

The central Maghreb occupies a distinct space in both the raw geographic and the
cultural imaginings of the ancient world an island between the Mediterranean and
Sahara whose geography ensured a balance of connectivity with the wider Classical
world and buffering from it1. In other words, a land where the changes that marked the
Roman imperial period might look very different from, or lag behind, other parts of
empire: a phenomenon generally cast as an inability to Romanize, as resistance to
new ideas and models of comportment, as survival, or as persistence. And the imperialperiod tophets the conventionalized term for sanctuaries where infants and/or lambs
were burned, their remains collected in an urn, buried, and often marked with a carved
stele as well as the deity worshipped, Baal Hammon/Saturn, often stand as the
foremost indices of this historical, cultural, and cultic continuity.
Certainly, in contrast to Sardinia and Sicily, where such sanctuaries seem to have
been abandoned in the generations following the islands incorporation into the sphere
of Roman hegemony, North Africa presents a very different trajectory: not only did
visitors still make sacrifices in older tophets, including the sanctuary at Hadrumetum,
well into the imperial period, but from the late second century BC onwards,
communities established large numbers of new tophet-style sanctuaries (fig. 1). The
landscape of stele-sanctuaries in North Africa was as much a product of Roman
1

SHAW 2003.

Studi Epigrafici e Linguistici 29-30, 2012-2013: xx-xx

M. M. McCarty, Continuities and Contexts

hegemony in the generations following the destruction of Carthage as of Punic


precedent2. Why tophets in these two regions followed such different trajectories in the
second-first centuries BC has yet to be explained convincingly; here, though, I want to
focus specifically on the nexus of problems surrounding religious continuity in the
imperial period.
Given that cults often make claims of tradition and antiquity, that even the most
radical religious innovations are often cast as revivals, it is not surprising that religion
whether in tophets, other sanctuaries, or necropoleis often forms the basis of
arguments for relative cultural and religious stasis in North Africa through the imperial
period3. Not only are tophets and sanctuaries of Saturn treated as guarantors of the
relative permanence of the religious, social, and political landscapes of North Africa
simply by virtue of their presence, but by their long periods of use.
Such a picture of stasis has been compounded by the nature of the discoveries from
these sites, which stand in contrast with the preserved and carefully excavated stratigraphy of tophets in Sardinia and Sicily, through which diachronic changes in practice can
be charted in detail. Often, North African tophets are attested only by the discovery of
their stelae on the surface, or in poorly-documented excavations4. Stripped from their
association to related finds, made by local workshops who rarely participate in the types
of technical or stylistic changes seen in cosmopolitan marble-carving workshops, such
stelae are nearly impossible to order chronologically independent of a presupposed
narrative of Romanization, where deeper relief, more Classical forms, and Latin
script/onomastics are evidence of second- and third-century dates5. If this scheme works
in some cases, the meta-narrative of stylistic progress fails in others; it leads, for
example, to the Boglio stele having long been dated nearly 150 years too late6.
The problems of charting material chronologies as opposed to meta-narrative
chronologies are equally present in other North African tophets where nineteenth- and
early twentieth-century amateur archaeologists documented at least some architecture,
pottery, and small finds. A few sanctuaries, like Thuburnica, never had plans published,
while others, like Thinissut and El-Knissia, only plotted horizontal relationships
between features and finds. Such plans themselves can be implicated in creating an

2
3

4
5

MCCARTY 2010a.
BELL 1997: 212-242, on rituals claims to traditionalism. In Africa: LE GLAY 1966; BNABOU
1976; FONTANA 2001 on Punic continuity at Leptis Magna through religion/burials; BRIANDPONSART HUGONIOT 2005: 141-273.
LE GLAY 1961; 1966a.
LE GLAY 1966b: 14-57; MCCARTY 2010a: 20-25. For the style of the works, the availability of
skilled and dedicated stone-carvers may matter more than chronology: cf. JOHNS 2003. As for the
problems with Latin script/onomastics as evidence for dating (this model articulated most clearly in
MCHAREK 1982), particular contexts might call for different perhaps deliberately archaizing
choices: for example, the son of Marcus Avianus a perfectly Latin name when making a
dedication to Baal chooses not only the Punic language, but a markedly Punic name for himself:
Baalshillek (JONGELING 2008: 66-67).
The Boglio stele, long dated to the late third/fourth century, actually finds its closest parallel in a
second-century AD stele from Hr. el-Hami, suggesting not only the need to down-date the Boglio
stele, but the problems inherent in such a predetermined, evolutionary chronological schema.

SEL 29-30, 2012-2013: xx-xx

illusion of stasis, collapsing several centuries of material to a single, visible horizon7.


More recent attempts to understand the phasing of these sanctuaries are often based
either on general ideas about architectural evolutions, or on the findspots of individual
pieces of mobilier, as though such objects could not be moved and re-used8. As with the
finds of groups of stelae, such analyses presuppose certain narrative schemes: of either
unidirectional change, with no room for archaisms or quotations of the past, or absolute
stasis, with objects never moving or being re-deployed in new contexts of use.
Even those sanctuaries that have been scientifically excavated more recently (Hr. elHami, Hr. Ghayadha) and offer some evidence of different phases and diachronic
change lack clear stratigraphy. Still, transformations of the types of activities practiced
at the sites can be mapped: above all, the end of child sacrifice and its replacement with
other types of rites9. These shifts raise a series of basic questions: how do we
understand the changes that took place in tophets in the Roman imperial period? Under
what circumstances does an individual or community take up new ritual practices, or
abandon old ones? Are such changes, in fact, ruptures with the past?
When recognized, transformations in tophet-cult usually get described via one of two
metaphors: as new features grafted onto old, or as cosmetic veneer. Both images
claim that the rites performed, and the range of significances they might create (or be
freighted with), were static and unchanged, just as the deity to whom the offerings were
directed was simply an old, Punic god Baal Hammon updated with a new, Latin
name Saturn10. Individual signs might be replaced by substitutes, but the larger
semiotic field remained constant. The model embraced by Alfred Merlin in his 1910
account of Thinissut still underlies many accounts of tophets and other cult sites in the
imperial period: Enfin, nous constatons la vitalit persistante des anciennes
religions au temps de la suprmatie de Rome ce sont les mmes dvotions qui
subsistent, avec les mmes symboles et les mmes pratiques; enrichies dapports plus
ou moins rcents 11.
Yet both notions continuity and change (alongside its variants, including all of the
synonyms for Romanization) are but modern, redescriptive schemata used to
narrativize the past, to link our momentary historical and archaeological snapshots, and
to set those linked chains within the political frameworks of the present. A number of
recent studies have examined how these narratives have been mobilized and
instrumentalized in modern North Africa12; my goal here is not to revisit the uses to
7

8
9

10

11
12

While CARTON (1908) was acutely aware of diachronic change in architecture and rites at ElKnissia, for MERLIN (1910: 51), Thinissut proved long-term continuity from the Punic to the
imperial period.
LZINE 1959; DRIDI SEBA 2008.
E.g., CINTAS 1948; ELHAM: 110; SHAW 2013. Perhaps the most important account of the supposed
change from human to animal victims, RICHARD 1961, remains a deeply flawed study, presupposing
a trajectory inherited from Pierre Cintas (MCCARTY forthcoming). MCCARTY 2010a focuses on the
social dimensions of these changes.
E.g., LE GLAY (1966: 485) on limpossibilit de la romanisation des mes; more recently,
CADOTTE 2007. On Hr. el-Hami: La culture punique apparat lpoque romaine non pas comme
une persistance mais comme une permanence. Quant aux influences dpoque romaine, elles viennent
se greffer sur lhritage carthaginois (ElHami: 123).
MERLIN 1910: 51.
DONDIN-PAYRE 1994; GUTRON 2010.

M. M. McCarty, Continuities and Contexts

which these pasts are put post-antiquity, but to reflect on how we interrogate epigraphic
and material evidence from tophets to arrive at notions of continuity. Too often, the
concept of continuity in accounts of individual stele-sanctuaries is used loosely and
uncritically, without articulating exactly what constitutes continuity: a problem
shared in the study of cult sites across the ancient world13.
Is the best index of cultic (and cultural) continuity, as Le Glay, Cadotte, and others
have argued, the worship of a particular deity14? If so, and if we acknowledge that any
identity for a god is negotiated through the gods name, image, and the way that worshippers deal with the deity, then surely the choice of a new name (Saturn), a new epithet (Augustus), a new image-type (enthroned, himation-clad elder male with pruningknife), and new types of interaction (including the end of child sacrifice) problematizes
a claim of continuity based on the god15. Likewise, giving priority to the person (or
personality) of the deity often stems from Christianizing assumptions about gods
rather than either symbol-systems or practices sitting at the heart of ancient religion16.
Is continuity best examined as the persistence of a particular symbol-system, as
Merlin implies? As ritualized activity of various forms in the same place a definition
favoured by those working on the transition between the Bronze and Iron Ages17? A
repeated set of ritualized activities, which as with all ritual, especially in the absence
of a master-script might undergo variations and innovations, but somehow still relate
to an ideal and abstract archetype?
In what follows, I want to make two central claims. First, that rather than presupposing redescriptive grounds for navigating between the poles of continuity and
change, it may be more fruitful to look at how cult communities themselves marked
aspects of their religious life as connected to and contiguous with or disjoined from
particular pasts18. Moving beyond a general notion of invented tradition, I will
explore the specific strategies and grounds upon which two different communities, at
Hr. el-Hami and Thugga (fig. 2), created and imagined links to particular pasts and
presents. Since the significances created by ritual participants depend on how their
actions are set within relational contexts, especially between sets of gestures and
between gestures and their environment, I will focus here instead on cult practices, and
how communities set their ritual acts within wider frameworks via both the objects and
gestures used and via their placement in space19. My second claim, on a more positivist
level, is that the evidence for continuity in ritual practice regular use of a site in a
13

14
15

16

17
18

19

SOURVINOU-INWOOD 1989; HUSSLER KING 2007, where the concepts remain central structuring principles even if clear explanations are hardly given.
LE GLAY 1961, 1966a-b; CADOTTE 2007; BRIAND-PONSART HUGONIOT 2005: 153.
Cf. LIPKA 2009 for constructing gods based on names, images, practices, times, and institutions. On
naming: MCCARTY 2010a: 44-59; images: MCCARTY 2011b. For rites, see below. For placing the
god in time, see the changes outlined in SCHRNER 2008.
RPKE forthcoming; I am grateful to Nic Terrenato for drawing my attention to Rpkes arguments
here.
Recently: FELSCH 1996-2007.
This is not, of course, to deny that even the concept of religion in the ancient world is problematic,
and itself a redescriptive (though not unuseful) category: NONGBRI 2013.
MCCARTY 2011a for the connection between ritual and representations at Hadrumetum; more
generally, BELL 1997: 173-177.

SEL 29-30, 2012-2013: xx-xx

similar manner over a given period at these two sites is much less secure than often
acknowledged. In the absence of material evidence, such regular usage cannot simply
be assumed, but must be proved as much as any suggestion of disjunction.

1. Henchir el-Hami: Recreating Rites


Since its excavation between 1992 and 1994, the sanctuary at Hr. el-Hami has been
taken as the type-case for the transformations of pre-imperial sanctuaries in the Roman
imperial period: from Baal Hammon to Saturn, from human to animal and vegetal
sacrifice, from tophet to temple, and from Punic/Numidian to Roman in the forms (but
not substance) of art and rites20. The excavator, Ahmed Ferjaoui, describes the significance of the site as proof of the permanence of Punic culture in the Roman imperial
period albeit with a few Roman influences grafted on, particularly in the
substitution of different types of offerings21. As with the general story told of tophets in
North Africa, it is the worship of a particular god Baal Hammon/Saturn that
guarantees continuity in the face of some ritual changes. Yet a fifth-century AD deposit
from the sanctuary suggests that this narrative is hardly so neat, revealing how
individual aspects of a rite could be made deliberately archaizing in order to create links
to a sanctuarys past.
Although the shallow depth of sanctuary below the modern surface precluded
detailed understanding of change based on stratigraphy, the pottery assemblage,
alongside Punic, Numidian, and Republican coins minted at Utica, suggests that the
sanctuary was founded at the end of the second century BC, or more probably, in the
early first century BC22. Up until the late second century AD, infants and lambs were
burned, buried in locally-made two-handled amphorettes in neat rows, and some of the
deposits marked with stelae in an open-air sanctuary (fig. 3, A). Towards the middleend of the second century AD, the cultic activity on the site moved north of the stelefield: an altar, small multi-roomed structure which used some older stelae in its foundations, and enclosure wall were built (fig. 3, B). Hundreds of unguentaria discovered
around the altar suggest a shift towards offerings of perfumed oil, while cookwares
suggest animal sacrifice and dining, rather than holocaust. The god to whom such rites
were directed is assumed to have been Saturn, the Romanized Baal-Hammon,
although there is no direct evidence. Such a move, from open-air stele field to temple
and altar parallels contemporary changes suggested to have taken place at other sites,
especially Thugga, in the late second century/early third century AD23.
If the building of the temple marked a clear disjunction in practices, links with the
past were still maintained via the sanctuarys location adjacent to the stele field. The
new temple and altar seem to avoid disturbing earlier offerings, although the precise relationship between the two parts of the sanctuary is unclear. Some of the earlier stelae
20
21
22
23

FERJAOUI 2002; ElHami.


ElHami: 123.
ElHami: 62-63.
None of these cases, save Hr. el-Hami, is wholly certain. Thugga: see below; Ammaedara: LE GLAY
1961: 323-324; Thubursicu Numidarum: LE GLAY 1961: 366-367, proposes that a sanctuary in the
forum area was moved c. 202 to a hill south of the city; Mactar: PICARD 1984, though the evidence
that the temple under the museum was the site of the stele-sanctuary is inconclusive at best.

M. M. McCarty, Continuities and Contexts

were removed and built into the walls of the temple; others seem to have fallen in situ
in the stele field, though whether this happened at time of the temples construction, or
post antiquity, is not certain. Some of the stelae could have remained on view while the
sanctuary community conducted their new rites just to the north, visible reminders of
past practice, even if they were not curated for re-display the way stelae at Thinissut
and Ammaedara were24. Likewise, the transition from one rite to another may have
been gradual rather than abrupt, given the crossover of unguentarium types between the
two areas25.
One of the most remarkable features of the site is a late-antique deposit, cited as evidence for continuity at the site in an era when urban pagan temples were going out
of use and often being repurposed and when sacred violence was raging in some
areas between newly-empowered Christian groups and non-Christians26. In late antiquity, visitors made what may have been the final offering at the site, and the last attested
tophet-style rite in all of North Africa: a cooking pot, filled with ash and burnt bone
from a lamb or goat, surrounded by four miniature chalices, each containing gray
sediment and topped with a clay plug; four lamps; and two coins (fig. 3, C; fig. 4)27.
The dating of the deposit is contested. Although Ferjaoui focuses on the coin of
Constantius II to suggest a date in the fourth century for the offering, the ceramic
assemblage suggests a slightly later date, in the fifth-sixth century. While there are two
parallels for the cooking-pot, unusual for its lack of everted rim, in the sanctuary, other
examples come from dated contexts at Carthage in the fifth-sixth century28. Two of the
lamps are Deneauve VIII, and probably datable to the mid-late third century; the other
two are Deneauve XII, and find their closest parallels in the fourth-fifth century AD29.
If pottery chronologies in this period are slightly fuzzier than the more rigid terminus
post quem provided by a coin, the ceramic assemblage seems to point at the earliest to
the fifth century AD. In addition, given the problems in the supply of base-metal
coinage in the late fourth and fifth centuries (especially acute in North Africa
immediately after the Vandal invasion), and that fourth-century coins seem regularly to
have stayed in circulation through the fifth century, the coins in the deposit offer only a

24
25
26

27

28

29

MCCARTY 2011a.
ElHami: 25.
LEONE 2007 may paint too uniform a picture of this transition following imperial legislation; Hr. elHami provides an important exception and alternative, albeit in a less urban context. Unfortunately,
stratigraphic excavation of Roman-late antique sanctuaries in Africa is rare, and I am unaware of
evidence for similar types of offerings, save for the late antique statue-caches: a very different type of
phenomenon. Sacred violence: SHAW 2011, ch. 5.
ElHami: 78. The latest datable stele dedicated to Saturn was set up in AD 323: BESCHAOUCH 1968.
New material from Hr. Ghayadha, however, shows activity that may have taken place there to the
seventh century, and given the sites remoteness, such activity was most likely cultic (MCHAREK et
al. 2008). Otherwise, evidence for late-antique visitation of sanctuaries dedicated to Saturn comes
primarily from coins recovered; whether these represent the cultic use of the sites, their destruction, or
other activities remains debated, as at Bou Kournein.
Parallels at Hr. el-Hami: DEL VAIS in ElHami: 340, nos. 54-55. Carthage: FULFORD 1984: 163, nos.
16-17, with dating suggested between 425-500, with earliest example in a deposit of around 450.
DEL VAIS in ElHami: 357-358. Lamp no. 3 is Bonifays Deneauve VIII.2 (225-250) Lamp no. 4 is
Bonifays Deneauve VIII.4 (250-300).

SEL 29-30, 2012-2013: xx-xx

very loose terminus post quem30. Altogether, this special deposit seems more probably
to date from the early-mid fifth century AD, rather than the fourth31.
Identifying the offrands, even in general terms, remains the other main challenge in
interpreting the deposit. Whether an individual, a family, a temple-community or other
voluntary association, or a civic group is impossible to determine on the grounds of the
deposit itself. Tophets/temples of Saturn and the rites practiced within them seem to
have structured very different types of communities at different times and places
some focused on small, family units, others on larger, hierarchical communities with
prestigious sacerdotes32. Changes in the society of participants at Hr. el-Hami seem to
have followed a trajectory common at other sites: in the tophet, individual/family
offerings that responded to occasional pressures; in the temple, more formalized,
regular community-based offerings around an altar that included large numbers of
unguentaria. Yet there is no guarantee that this unusual fifth-century deposit was made
by a similar sanctuary community33.
Rather than evidence of a clear continuity, an unbroken progression of rites from the
first century BC, this fifth century offering instead indicates an imaginative re-creation
of much earlier practices. The central feature of this structured deposit was the burnt
offering: the other items in the deposit orbit around the cooking-pot, with the lampnozzles all aimed inwards towards it to highlight its importance. Although it is unclear
whether the pot contained the full holocaust of an ovicaprine, or the selection of some
parts of an animal following butchery and consumption, curating the burnt remains of
an offering and burying them had not been done at Hr. el-Hami for roughly 250 years
when this offering was made. Not only had the rite of holocaust and deposition been
abandoned at Hr. el-Hami much earlier, but no buried holocaust deposits later than the
end of the second century AD have been discovered anywhere in North Africa34. Even
if some stelae date from later periods, these did not necessarily accompany burned and
buried remains; at Hadrumetum, for example, carved stelae continued to be erected in
the final stratum (second century AD), even if burned and buried offerings largely
ceased35.
The particular rite of holocaust and deposition, central to tophets, was reprised
after seemingly being abandoned everywhere for two hundred years. Of course, the
open-form cooking pot used for this deposit is a very different shape than the closedform amphorettes preferred earlier: this was not a direct replication of earlier rites and
their trappings. As such, the deposit raises the questions of how the dedicants composed

30

31
32
33
34

35

REECE 1984; GUEST 2012. Cf. SAUER 2011 for coin-circulation and offerings in the late-antique
northwest.
DEL VAIS in ElHami: 358 offers a similar date.
E.g., BENSEDDIK 2006 for family being central to the cult around Sitifis; MCCARTY 2010a: 91-151.
MCCARTY, forthcoming.
Second century deposits: Lambafundi (LE GLAY 1966a: 114-115); Thamugadi (LE GLAY 1966a:
125-129). If Tertullian, Apol. 9 has been used to suggest the continuity of child sacrifice after around
200, not only is his claim unverified archaeologically, but the lack of urn-deposits after this date also
suggests that substitution sacrifices (as practiced earlier) were also abandoned around this time:
SHAW 2013.
MCCARTY 2011a.

M. M. McCarty, Continuities and Contexts

their ritual act, and how memories of such practices were preserved and maintained
over several generations in the absence of regular ritual performance.
Other aspects of the Hr. el-Hami deposit similarly look back to much earlier offering
practices. The two coins placed in this deposit recall the coins dedicated alongside burnt
remains in the first phase of the sanctuary. If there were around 200 coins found in the
urn-field up to the reign of Domitian, these coin offerings seem almost entirely to
disappear from the site in later periods, and the few later coins found might well have
been lost rather than intentionally deposited36. Coin-offering, like the deposition of
burnt sacrifice, was very much a practice of the distant past at Hr. el-Hami when this
special deposit was made.
Nor do coin offerings seem to have been common in other North African sanctuaries
in the imperial/late-antique period, judging from the paucity of finds. At both
Hadrumetum and Hr. Ghayadha, for example, coins, common in the first century BC,
largely disappear in the first century AD37. The same is true at El-Knissia, where just
as at Hr. el-Hami the Roman imperial coins disappear after the reign of Domitian,
despite later cultic activity on the site38. While arguing from absence is always
dangerous, especially when it involves valuable and re-usable metal objects, the dearth
of coin-offerings attested in second-fifth century North Africa stands in sharp contrast
to the situation in the northern provinces, where low-denomination coins were among
the most common dedications39. This does not, of course, preclude the possibility that
any such offerings in North Africa were simply given to the sanctuaries differently,
perhaps collected in a coin-box rather than deposited individually. Nevertheless, the
dedicants of this special fifth-century deposit at Hr. el-Hami were not simply borrowing
from contemporary practices in other sanctuaries to piece together their own offering,
but again drawing on a ritual practice that seems to have fallen out of regular use for
several hundred years.
The four lamps arranged around the central burnt offering suggest that the fifthcentury offrands were not simply looking to previous rites that took place at Hr. elHami. At Hr. el-Hami, lamps do not seem to have played a large role in most of the
rites from the first century BC to the third century AD. In the stele-field, fragments
from only five lamps were recovered: one late Hellenistic and three Deneauve VII,
datable to the second century AD. The only nozzle fragment shows signs of burning.
This is hardly enough evidence to suggest that the rites were regularly held at night,
both because the lighting of lamps as a votive act need not be confined to the provision
of light and because of the paucity of material40. In the north part of the sanctuary at Hr.
36
37

38
39

40

ALEXANDROPOULOS in ElHami: 448.


At Hadrumetum, coin offerings seem to decline rapidly in the first century AD, similar to the pattern
seen at Hr. el-Hami. While the majority of coins found at Hr. Ghayadha come from the sanctuary,
most are Punic-Numidian (8 as opposed to 3 Roman imperial coins), suggesting that coin offering
was practiced early.
CARTON 1908: 19 Numidian/Republican coins, 12 Roman imperial coins to Domitian.
SAUER 2011. Admittedly, the northwest provinces have been more extensively surveyed and
excavated.
Nighttime: ElHami: 110. The notion that such sacrifices took place under the cover of night, really
supported only at Nicivibus by the epigraphic formula sacrum magnum nocturnum, has more to do
with the trope of bad (or illicit) religion being done secretly in darkness than actual evidence for

SEL 29-30, 2012-2013: xx-xx

el-Hami, the area of the altar and temple, a larger though still relatively small
number of lamps was recovered (18); the majority of these show clear signs of having
been lit.
The relative lack of lamps in the sanctuary stands in contrast to their regular
appearance at other North African stele-sanctuaries of the late Hellenistic/early imperial
period, hinting at the local variations in tophet rites. At Salammb in the third-second
century BC, lit lamps were a regular feature of the rites surrounding urn-deposits, with
their smoke blackening the bases of some stelae; at El-Knissia, over three thousand
lamps were discovered in favissae with the stelae, almost all (except for votive
miniatures) showing signs of having been lit; at Thuburnica, over a hundred lamps were
placed in front of the niches of the temple and found in a deposit of mixed ash and
pottery adjacent to the stele-field; at Bou Kournein, sixty lamps were discovered around
the altar41. In the temple at Thinissut, which itself seems to have replaced an urn-field in
the first century BC/AD, lamps were found in situ, placed around the niches for statuedisplay; not only would they have illuminated the statues, but lighting them would have
served as a devotional act in its own right42. Lamp-lighting continued to be an important
ritual act, especially at many tophets, through late antiquity: two late deposits,
consisting primarily of lamps and plates, were buried in the sanctuary at Hr.
Ghayadha43. If illuminating lamps was a regular practice at many tophets and other
sanctuaries, a rite that continued regularly through late antiquity, this had not been the
case in the earlier phases at Hr. el-Hami. This suggests two things: first, that despite
commonalities among offerings across tophets, there was not a fixed script for the
denouement of rites shared among different sanctuaries, but a family of loosely related
practices that orbited around the central triad of holocaust, burial, and stele-erection.
Second, when the fifth century deposit was made at Hr. el-Hami, the dedicants were not
necessarily looking back to how exactly earlier rites had been conducted there, but were
instead including practices and ritual forms drawn from a wider pool of options.
The fact that the lamps in the fifth-century Hr. el-Hami deposit were buried as part
of the offering also suggests a different use of the objects44. Only one is recorded as
showing signs of burning on its nozzle, and it is one of the lamps datable to the third
century; the traces of burning may be from another use context during the roughly 150
years between its creation and its burial. The other three lamps may have been unused:
they were not apparently lit as part of the rites themselves, but simply removed from
circulation45. The lamps found in all of the sanctuaries cited above were either
discovered in situ, having been lit at the base of a stele or statue as part of a devotional
act, or found collected in favissae, having been culled and buried with earlier offerings.
Unlike the assemblage from Hr. el-Hami, these lamps were not part of individual

41

42
43

44
45

practice. Of course, given the amount of time necessary for a fire to burn a complete holocaust, when
such sacrifices were made, it is possible that they, like funeral pyres, were left to burn overnight.
Salammb: BNICHOU-SAFAR 2004; El-Knissia: CARTON 1908: 96-105; Thuburnica: CARTON
1907; Bou Kournein: TOUTAIN 1892.
MERLIN 1910.
MCHAREK et al. 2008: 143-144, dating the deposits to the fourth-fifth century AD, contemporary
with the one at Hr. el-Hami. For the ceramics, BOURGEOIS 2008: 257-258.
DEL VAIS in ElHami: 357-359.
Cf. STEWART 2003: 195-207 on lamps in Roman cult.

10

M. M. McCarty, Continuities and Contexts

structured deposits. Not only are the lamps a new addition to the rite at Hr. el-Hami,
and not reflective of earlier practices there, but employed in the ritual in a
fundamentally different manner.
The most unusual feature of the Hr. el-Hami deposit is the group of four doublehandled chalices with clay plugs. Two were opened by the excavation team, and their
contents examined; the other two were subjected to non-invasive x-ray analysis. Under
the covers, only a grey sediment was found, with very few inclusions: tiny fragments of
bone and small land-snails46. What is clear is that the chalices did not hold burned
bones collected from a holocaust. In the absence of further analysis, little can be said
about these offerings, except that whatever the cups contained originally, it was
important enough to merit the clay seals. If the forms of the vessels are themselves
unusual, with parallels more easily attested in the iconographic repertoire of mosaics
rather than in other ceramic assemblages, at least one other, similar chalice was found
on the site47.
Rather than a custom order of ceramics and offerings for this particular rite, the
objects instead seem to have been drawn from a variety of sources. Two of the lamps
used had potentially been in circulation for over a century when paired with the two
more recent ones. While very unusual, and without obvious published parallels, the
chalices were all hand-made using clays that vary in fabric/inclusions. If the general
profiles are commensurate, the base of the handles, mouldings on the body, and
decoration of the foot differ considerably. The chalices were thus not objects produced
together as a group for this specific occasion. The opportunistic collection of offerings
to be used for the fifth-century rite (alongside the uniqueness of the deposit), hints that,
rather than being an occasion planned-for and provisioned well in advance, the rite was
enacted to respond to a specific set of unforeseen pressures at a given moment48.
This fifth-century deposit thus represents another disjunction in practice, a fifth
century break from the communal sacrifices and meals that may have continued around
the altar49. Yet in breaking with contemporary norms of practice and technologies for
dealing with the gods, the offrands drew connections with a cultic past.
The special deposit at Hr. el-Hami was the product of a hybrid set of rites, and
reveals much about the way ritual practices might be re-imagined, re-interpreted, and
altered. Certain features of the deposit and the acts that resulted in its formation the
burning and burial of a sacrificial victim, the inclusion of coins seem deliberately archaizing, meant to create resonances with much earlier practices that had not been performed at Hr. el-Hami, or anywhere else, for over 200 years. To speak of this deposit as
46
47

48

49

BDOUI OUESLATI in ElHami: 452-453.


Parallels: DEL VAIS in ElHami: 358, no. 13. Other chalice at Hr. el-Hami: DEL VAIS in ElHami: 339,
no. 45. The lack of other parallels in published ceramic assemblages from sanctuaries, settlements,
and necropoleis including early excavations where such unusual forms stood a better chance of
being recorded, may attest the rarity of this form in ceramic. The incised contour lines and tongue
pattern on the foot of no. 10 may hint at metal prototypes.
While a number of vessels used for the burnt remains of sacrifice in tophets are opportunistically reused, this too may be a product of the lack of institutionalization of the rites and their response to
specific, unplanned stressors: MCCARTY forthcoming.
The chalice and cooking-pots from the temple area that parallel those of the special deposit, alongside
lamps and African D ware in forms datable to the fourth-fifth century (CAMPISI in ElHami: 366; DEL
VAIS in ElHami: 375-376), suggest that some form of occupation continued on the site.

SEL 29-30, 2012-2013: xx-xx

11

a continuity from earlier cult rites is not wholly accurate; on the other hand, for the
offrands who engaged in this rite in the fifth century, these archaisms created a link to
earlier sacrificants at the site and their technologies for dealing with the gods. The
offrands, though, did not exactly reproduce such earlier offerings. The fifth-century votaries may not have been directly familiar either the liturgical order (and its variations)
practiced at Hr. el-Hami or other sites in earlier generations, for the rite is
elaborated substantially and includes features not regularly present at Hr. el-Hami
lamps and chalices and used in a manner distinct from earlier practices at other sites
the lamps.
Explaining this single, late deposit is more difficult in the absence of clear parallels
or other structured ritual deposits datable to this period. Shifting the date of the deposit
to the fifth century moves it into a very different historical context than the mid-late
fourth century. Even if the ritual itself was meant to respond to an unforeseen stressor
in the life of the community the proximate cause for the deposit being created the
unusual form chosen may relate to wider pressures.
The early fifth century was, after all, an era of increasing repression of practices seen
as incompatible with orthodox Christianity. In 399, Honorius sent agents and legislation
to Africa to help stamp out non-Christian practices; two years later, church councils at
Carthage issued canons specifically targeting rural cult-centers50. It was also an era
when a number of temples in North African cities had possibly been abandoned, and
when others were being retrofitted for new civic or Christian cult use; the temple of
Saturn at Thugga, for example, was quarried for materials to build the fourth-early fifth
century church just below, while at Thuburbo Maius, a temple was converted to a
church in the mid-late fourth century51.
It may be overly facile to call this strange, archaizing deposit a form of resistance to
the Christian conflicts and pressures of the early fifth century, and yet it seems difficult
to withdraw the offering entirely from this frame. Perhaps, as with the original
establishment of many tophets along the African littoral and down into the Medjerda
valley/Siliana plain in the first century BC, the late-antique offering at Hr. el-Hami
represents an attempt to reclaim and reconnect with a past under pressure from a new
hegemonic power. If nothing else, the attempt to reanimate a long-dead form of ritual
practice suggests that the cult community held a very different view of the
reproducibility of the past, of that pasts potentially continuous presence, than the view
being loudly championed by a contemporary North African bishop52.
To speak of the deposit at Hr. el-Hami in terms either of continuity or change, in
terms of persistence or Romanization, misses the central tension at play in the offering:
between deliberately and recognizably archaizing features of this offering, its new and
strange dimensions, and the more typical sacrificial rites taking place in the temple. The
dedicants at Hr. el-Hami rejected contemporary ritual forms and technologies those
attested around the altar for dealing with the gods, in favour of revived, remembered,
and adapted tophet-style rites. The offrands reprised what they thought had taken
place in that place: re-establishing old ritual acts was as important as where in space
50
51
52

CCL 149: 196-197, canons 58, 60. Cf. RIGGS 2001.


Most recently, SEARS 2011, demonstrating the problems in dating these changes archaeologically.
For Augustines notions of progress: MOMMSEN 1951.

M. M. McCarty, Continuities and Contexts

12

those rites were performed. Neither place alone, nor the deity (to whomever the offering
was directed), was a full guarantor of ties to antiquity: the forms of offering were
necessary to set this act in line with past tradition. This is a very different strategy, and
way of establishing contexts and frames for tophet rites, than those chosen by a
sanctuary-community at Thugga.

2. Thugga: New Contexts and Traditions


In many ways, the narrative told of the sanctuary of Saturn at Thugga parallels that
at Hr. el-Hami. A late second/first century BC stele-field was replaced by a built temple
towards the end of second century AD; indeed, the shift observed at Thugga has offered
the main context for historical interpretation of the Hr. el-Hami sanctuary and others
(Mactar, Thubursicu Numidarum, Hr. Ghayadha), creating the tophet-to-temple
narrative. Yet two features in the archaeology of the site caution against this neat
picture: first, the evidence for continuous ritual use of the site is problematic, and
second, the new rites adopted, and the architectural context into which they are put,
frame those rites very differently: not as practices tied only to a past, but as deeply
entwined with the contemporary religious life of the city.
Under and alongside a temple dedicated in AD 195 by L. Octavius Victor Roscianus,
set along a sharp slope to the northeast of the city centre, numerous remains of earlier
tophet-style rites were discovered, including approximately 500 stelae and urns with
burnt offerings, some in situ, tucked into crevices and along ledges across the slope, the
rest collected in favissae (fig. 5)53. If the few published ceramics from the site indicate
activity from the second century BC onwards, the stelae suggest a similar chronology54.
The inscribed dedications include communal offerings in neo-Punic of a type common
in the area between Thugga and Mactar in the first centuries BC/AD, as well as Latin
dedications by offrands whose names (Sittius, C. Iulius) likewise suggest a terminus
post quem in the mid-late first century BC55.
In AD 36/7, a patron of the Thuggan pagus, Lucius Postumius Chius, built a temple
to Saturn whose inscription survives built into the Byzantine wall. At the same time,
Chius also paid for a number of other monuments: given that these benefactions seem
to cluster in the forum area, it is most likely that the new temple to Saturn was also
located there56. Two further inscriptions, found at a distance from the sanctuary, record
the building and the subsequent restoration of a temple to Saturn, probably in the first
and mid-second century AD respectively57. The first was found 6km away from the site,
at Hr. Ben Mansoura. The second, recording the restoration of the temple vetustate
consumptum, was found in the forum area, built into a modern house; this led Louis
Poinssot to speculate that the original temple had been the same temple dedicated in
53
54
55

56
57

CARTON 1897; LANTIER POINSSOT 1942; POINSSOT 1958: 66.


KRANDEL-BEN YOUNS 2002: 163-171.
Communal dedications: JONGELING 2008: 78-79, nos. 1-3; 81, no. 1; 95, no. 11; 106, no. 39; 137,
no. 105; 139, nos. 110-111; 140, no. 116; 149, no. 13; 152, no. 21; cf. MCCARTY forthcoming. DFH
250-252 for the Latin stelae.
AE 1914.172=DFH no. 23.
CIL 8.10619=27417; ILAfr 551 = DFH no. 126. The letter-forms on the restoration inscription
suggest a date in the early second century AD: DFH 252.

SEL 29-30, 2012-2013: xx-xx

13

36/7, and that the temple stood somewhere nearby and may even have been one of the
rooms opening onto the west edge of the forum58. Of course, many of the inscriptions
found around the forum and reused in post-antique buildings themselves come from a
secondary context: the Byzantine wall59. As a result, most recent commentators have
suggested that the restoration inscription refers to a sanctuary over the tophet site, to the
northwest of the town60. The history of the sanctuary would thus run as follows:
established as a tophet in the late second/first century BC, with a small temple-building
added in the first century AD (possibly Chius temple or the one referred to in the
inscription from Hr. ben Mansoura); this temple was restored in the second century, and
then replaced by Roscianus grander temple in AD 195.
I want to suggest instead a very different narrative of the sanctuarys use. From the
foundation of the sanctuary in the second-first century BC, both individuals and the bl
(citizens?) of the town offered cult in the open-air sanctuary, including the burial of
burnt sacrificial remains and commemoration with stelae61. These rites can be documented continuously through the early/mid first century AD by the onomastics on the
stelae. Then, a period of either abandonment or of greatly reduced activity on the
hillside: the cult may have been displaced to the city-centre with the building of either
Chius or the civitas temple, adopting new rites in the new location, or these new temples might have been cast as newly-founded sanctuaries without a link to hilltop tophet.
Rather than originating from the hilltop sanctuary, it seems more likely that this set
of building/repair inscriptions instead comes from multiple sanctuaries. If the findspot
of the restoration inscription does not guarantee that the temple stood directly on the
forum, it is also worth noting that none of the blocks built into the Byzantine wall can
be associated with the temple of Saturn on the hill. At the time the Byzantine wall was
built, the Saturn sanctuary was already out of use, and a potential quarry for materials:
the late fourth/early fifth-century chapel near the temple cannibalized its stones62.
Likewise, the dedication to Saturn for the health of Commodus, reused in a wall
running through the Temple of Concordia near the city centre, does not necessarily
come from the sanctuary on the hill; such dedications are common in public spaces and
in sanctuaries devoted primarily to other deities63. It thus seems probable, given that
material from the Saturn-temple does not seem to have made its way down to the
forum-area for re-use, that the temple-restoration inscription similarly did not originate
from the hillside sanctuary. This would mean that there were at least two sanctuaries to
Saturn at Thugga, and perhaps a third at Hr. ben Mansoura, where the civitas temple
inscription was found: such duplication of sanctuaries is not uncommon within a town,
and for stelae-sanctuaries, finds a parallel at second-century Lambaesis64.
In addition, no material recorded from the hillside sanctuary can be dated between
the mid first century AD and the construction of Roscianus temple. If there was an
58
59
60
61
62
63
64

POINSSOT 1916: 31, followed by LE GLAY 1961: 208.


DFH 253.
DFH 251-2; SAINT-AMANS 2004: 348-357.
Communal dedications: JONGELING 2008: 76, no. 2; 78, no. 5.
POINSSOT LANTIER 1925: 231.
ILTun 1399.
MCCARTY 2010b: 38. RIVES 1995: 147 also favours multiple sanctuaries.

14

M. M. McCarty, Continuities and Contexts

earlier building on the site, the later temple obliterated its architectural presence
entirely; nothing is visible on-site, nor did excavators record any structures. Carton does
record a few objects datable to the imperial period excavated at the northwest corner of
the sanctuary under the portico: glass unguentaria, a Roman-style lamp with tondo, and
a piece of terra sigillata with a stamp that Carton read as CNAMF in planta pedis65.
The piece is almost certainly Italian sigillata, given general import and consumption
patterns at Carthage and documented by the Thugga survey; this, however, would mean
that Carton misread the stamp, as no comparable stamps are attested among Italian
producers66. The most likely reading of the stamp is CNATM, with elision of the A
and T and Cartons F being the toes of the foot; this is the reading given in CIL and by
Poinssot, who includes this stamp, rather than a second piece of terra sigillata from the
sanctuary, in his publication of inscriptions from Thugga67. This reading is further
supported by the presence of Cn. Ateius Ma()s wares at Carthage, Leptiminus, and
sites in Tripolitania, showing that the workshop exported to North African markets68.
The potter provides a date for the piece in the early-mid first century AD69.
Nevertheless, the materials mentioned by Carton were seemingly part of a deposit
formed when Roscianus temple was built, and may be residual; they thus do not offer a
firm chronological indicator for the activity in the sanctuary itself. That said, even if the
materials do come from cult activity on the site, they do not rule out a hiatus in ritual in
the sanctuary between the mid-late first century AD and 195.
The number of stelae found at the sanctuary just under 500 also does not argue
against a shortened period of use. There are no numbers provided for the sacrificial urns
at Thugga, so the stelae serve as only surviving proxy for the frequency of cult activity.
Stele-dedication was not a necessary part of tophet rites, and different communities at
different periods dedicated stelae alongside their burnt sacrifices at different rates. If 52
stele-fragments are attested at the relatively rural sanctuary of Hr. el-Hami over the
roughly 250 years when tophet rites were practiced (0.25/year), at the more urban Cirta,
more than 850 stelae are attested over approximately 350 years of use (2.4/year). If
active from c. 150 BC-AD 50, the sanctuary community in the equally urban Thugga
replete with a temple to Massinissa, a princely tower-tomb, and a substantial number of
imported goods attesting the citys wealth and cosmopolitan character would have
been dedicating stelae at a similar rate (2.5/year). Although these numbers are highly
artificial, they at least demonstrate that the sheer number of offerings at Thugga does
not rule out an abbreviated period of stele-dedication and activity in the tophet.
Nor should the choice of this site for Roscianus temple indicate an unbroken chain
of regular ritual practices until that moment. Such re-activation, and appropriation of
the perceived antiquity, of a cult site is hardly a unique phenomenon in the Roman (or
65
66

67

68
69

CARTON 1898: 383. Both CARTON and LE GLAY read this as C. Namf(amo).
FULFORD PEACOCK 1995; RICE 2011; CIOTOLA 2000: 59, who draws attention to the lack of
Gallic wares, in contrast to sites in Algeria and Libya. LE GLAYs assertion (1961: 210) that it must
be an ARS imitation of Italian ware is unlikely.
CIL 8.22645, 44d; POINSSOT 1911: 67. LE GLAY (1961: 210) assumes that POINSSOT is referring to
a second stamp, when it is clear from CIL and POINSSOT that they are re-publishing the piece found
by CARTON.
OCK no. 298, vessels no. 22.582-3.
Ateius pieces were also found, in great quantity, at Pompeii.

SEL 29-30, 2012-2013: xx-xx

15

post-Roman) world: examples abound, including the sanctuary of Demeter and Kore on
the slope above Corinth70.
Arguing from archaeological absence, especially given how material was recorded
by Carton and Poinssot, is dangerous. Verification of a gap in ritual activity on the site
would require a new campaign of excavation. Yet what is clear is that the case for
regular use of the tophet area from the first century AD to AD 195 is no more certain
than a potential hiatus. Continuity cannot be assumed any more than rupture in the
absence of evidence.
At the very least, then, we are confronted at Thugga with a gap in the evidence of
regular ritual use of the hillside sanctuary from the mid first century AD, when the site
may have been replaced by a temple in another location and with the new temple, a
potentially very different style of ritual that did not involve burned offerings and their
burial. This potential change has important ramifications for the larger narrative told of
tophets in the imperial period, since the dating of the transition at Thugga has been one
of the anchors of accounts that recognize a sea-change in the religious life of the
province in the late second century AD. The tophet to temple transition at Thugga
may look rather different in both chronology and continuity than the shift seen at Hr.
el-Hami, when a temple was built in the late second century adjacent to the older stelaefield. The rates and forms of ritual change in North Africa were hardly homogenous,
but varied widely among different communities; there was no set trajectory to the
Romanization of cults, even if similar patterns, similar families of the modalities of
change, can be observed at different sites71.
Seeing a sharp reduction or disappearance of cult activity in the hillside tophet from
the mid first century AD also lends new importance and interest to the building of
Roscianus temple. If the offrands of the fifth-century deposit at Hr. el-Hami sought to
recall a cultic past by reactivating and reimagining past rites in an archaizing pastiche,
Roscianus and the community of temple-users at Thugga employed a very different
strategy for creating a link to past cult.
The ritual practiced within Roscianus temple a courtyard surrounded by a vaulted
portico, with three vaulted cellae on the west side seems to have been markedly
different than that of the earlier tophet. Instead of incineration of a victim on a pyre, the
burial of those remains, and commemoration via a carved monument in the midst of
densely-packed rows of stelae, the open courtyard created a stage for a more public
sacrifice, likely with a sacrificial altar at center, although the paving-stones of the court
and anything above them were robbed out post-antiquity72. A marble statue of Saturn
enthroned, almost certainly carved by a workshop from Carthage, could watch the
denouement of these rites, as could a large group assembled in the shade of the
portico73. The other main architectural evidence of what happened in the new temple
comes from the cisterns common in most North African buildings with a large surface
area to aid in the catchment of an important resource and from the wear on the
stylobates supporting a small propylon at the center of the east portico (fig. 5, A),
70

71
72
73

BOOKIDIS STROUD 1997. Cf. HORDEN PURCELL 2000: 406-411, arguing for the importance of
landscape itself, rather than its sacralization by human agency, as the cause of such survivals.
MCCARTY, forthcoming.
Cf. EINGARTNER 2003 for a similar phenomenon at Thamugadi.
MCCARTY 2011b.

16

M. M. McCarty, Continuities and Contexts

implying frequent foot-traffic around the portico. In terms of ritual forms, there was a
clear disjunction with past practices, whether they continued at Thugga up until the
building of Roscianus temple, or whether they had petered out earlier with the building
of the first century AD temple(s) in the town.
Instead, the placement of the new temple was what ensured the sanctuary and its
cults ties to antiquity. Rather than simply replacing an active stele-field with changed
forms of sacrifice, placing the temple over an older, if defunct, sanctuary created a
resonance with these past practices and cultic life, in spite of the disjunction in practice.
The plan of Roscianus temple may also draw connections to the past in a slightly
different way. Laid out as a so-called temple cour, with three cellae set on one side of
a porticoed courtyard, the sanctuary follows a temple design popular across North
Africa74. Patrizio Pensabene has demonstrated that such plans owe their origin to
Phoenicio-Punic sanctuaries, even if the basic layout could be adapted in various ways
to suit the needs of different communities and/or deities75. The choice of such a plan at
Thugga to replace an earlier open-air stele-field was not to reproduce the cult as it had
existed before in that place, or was imagined to have existed. Up until the terrace of the
temple was built, burying sections of the tophet under up to 2m of earth to create a level
platform for the temple, earlier stelae and deposits seem to have remained on view,
potential reminders of the type of cult that had been practiced there (fig. 6). The plan of
Roscianus temple then is not an example of conservatism, of preserving things as they
were, or of continuity with a Punic or Libyan past, but of inserting the sanctuary within
an imagined sense of what might constitute traditional forms. Perhaps more
significantly, this new layout also set the sanctuary within a genre of temples being
built and/or reconstructed at Thugga in the second-third centuries, including Temple B
and the Temple of Tellus76.
If the placement of the temple created a resonance with past cult of a different kind
on the site, and the plan of the temple partook in imaginations of a particular tradition, a
renovation of the temple shortly after its dedication suggests how the cult could also be
set into other, contemporary contexts as well, and placed within a wider web of civic
and cultic life. Roscianus temple was originally entered from the east, via two steps
leading to a door that opened onto the portico (fig. 7, A). There are no foundations for
monumental exterior embellishments: no columns, no porch, just the stairs and the
outer wall of the precinct. Soon after the temple was built, though, a new porch and
faade was added to the sanctuary77. This new porch sits on a separate foundation than
74
75

76

77

TILMANT 1989 offers a useful overview of the temples cour and scholarship on the topic.
PENSABENE 1990; TILMANT 1989 removes the temples from a Levantine sphere and sets them
wholly in the west.
TILMANT 1995. The identification of Temple B remains problematic: SAINT-AMANS 2004: 287293. Temple B, as it stands now as five rooms opening onto a large court, was built in the middle-end
of the second century based on ceramics; the Temple of Tellus dates by its inscription to AD 261
(SAINT AMANS 2004: 362-363), but reuses a much earlier lintel dedicated to Tellus in its entryway:
another means by which connections to the past were made. Given the archaeological focus on urban
facies of the High Empire, earlier levels and antecedent temples have rarely been excavated at
Thugga.
The column bases on the porch reproduce, in slightly larger form and with slightly different
proportions, the moulding sequence of the interior colonnade, suggesting an attempt to link the two
spaces architecturally and potentially a close chronology.

SEL 29-30, 2012-2013: xx-xx

17

the terrace upon which the temple was built, guaranteeing a later date for the porchs
construction. Four Corinthian columns stood atop a stylobate set along the front side of
this porch (fig. 7, B). The fact that the stylobate sits at the same level as the entrance to
the sanctuary suggests that the floor-level of the porch would have been at the same
level as the entryway, obsoleting the two steps from the earlier phase. In addition,
probably as part of a third phase of reconstruction, a raised platform, not architecturally
bonded into the porch, was built extending outward.
One of the problems created by this new faade was of access to the temple: there
are no traces of a staircase or any other means of negotiating the sharp drop between the
level of the porch and the sloping ground below. Instead, entrance to the precinct seems
to have been provided via a side door and set of rooms constructed on the south side of
the portico (fig. 5, D). The new columnar faade, erected at some point in the third
century AD, did not serve as an entrance to the temple; it was built as a monumental
showpiece atop a tall podium that heightened the visibility of the already monumental
monolithic columns set at front to the building. If much has been made of the plan of
the temple of Saturn as a temple cour, leading to the suggestion of the cult as native
and bracketed off from civic and Roman cults in the city, the elevation of the
temple the sanctuary as presented to those looking at it from the outside firmly set
the sanctuary in relation to the range of other temples with podia and columnar facades
that had been (and were being) built in the town from the mid second century onwards,
from the Capitolium to the Temple of Pietas. As much as the temples cult might have
been framed by the fuzzy context of tradition in its plan, it was also set firmly within
the wider sphere of the contemporary religious life of the town.

3. Conclusions
To speak of continuity or rupture in the cultic life of North Africa and its tophets is
to create a narrative relationship between a given site and some imagined past: a
narrative potentially constructed quite differently than it would have been by the
communities using these sanctuaries. All too often, the grounds for claims of
continuity are based, as in the cases of Hr. el-Hami and Thugga, upon problematic
assumptions, concealing a host of disjunctions: in the deity, in the types of rites
practiced, in the architectural arrangements of the sanctuaries, in the move from tophet
to temple. Assuming continuity and stasis in cult also leads to overlooking potential
gaps and hiatuses in ritual, as may be the case on the tophet site at Thugga.
Rather than tallying aspects that seem to remain constant versus those that seem to
change (the deity, the place, the rites, the symbol system), it may be more fruitful to
examine the strategies by which communities created their links to the past, how they
made their cults traditional and situated them in relation to some past. This did not
happen in any static form, via continuous reproduction of ritual scripts, but via an
imaginative recreation of cultic traditions. At Hr. el-Hami, moving the second-century
temple to a space adjacent to the earlier tophet, with some stelae potentially left on
view, maintained links to past cult in the face of a wholly changed ritual apparatus:
links that were seized upon when a group of fifth-century worshippers reprised the
long-defunct tophet-style rites. These offrands, rather than directly copying rituals
which had ceased 250 years earlier, recreated these rites by combining a set of
archaizing offerings (burnt remains in a pot, coins) with new and unusual contemporary

M. M. McCarty, Continuities and Contexts

18

types of offerings (chalices), and elements sometimes used in tophet-rites (lamps), even
if not at Hr. el-Hami, put here to new use. If one basis for ties to the past depended on
the place itself, reconstituting an abandoned ritual practice served as a second basis, at
least for one group of fifth-century worshippers. At Thugga in the second-third century
AD, the site of the earlier tophet was re-activated as a locus of cult, though for a vastly
changed set of practices. These practices, though, were framed via an architectural link
to traditional cult in the form of Roscianus temple cour. Here, the tie to the past is
not just reimagined as a reprise of what had come before in that place, as with the
reconstitution of a particular rite at Hr. el-Hami, but in a more generic way, via a
temple-layout accepted as part of a traditional repertoire. At the same time, the framing
of the rites and temple were soon updated, to set the cult not just within the sphere of
tradition, but within the realm of contemporary civic display and cult at the site: even if
this meant re-orienting the entrance to the entire sanctuary.
No group exists without a sense of its ties to some history, and religion especially
in ritual practice serves as one of the central discourses wherein the past is
remembered, maintained, and colonized78. Rather than seeing continuity as natural, or
as the default, any such sense of continuity is the product of particular means and
media, of strategies that might vary across time and communities. Indeed, the types of
resonance and evocation of both particular past places and practices and more general
imaginations of tradition seen at Hr. el-Hami and Thugga already seem to differ
markedly from the strategies used at other, earlier tophets. At Carthage and
Hadrumetum, for example, imagined pasts, lived presents (and perhaps even
aspirational futures) were woven together via a most unusual sacrificial rite, freighted
with imagined antiquity; via monumental forms that could evoke a Phoenician past; and
above all, via extended genealogies of individual offrands: a feature conspicuous by its
absence from most imperial-period stelae-sanctuaries79.

Plus cest la mme chose, plus a change

78
79

ASSMANN 2006.
BONNET 2011, pace DANDREA GIARDINO 2011.

SEL 29-30, 2012-2013: xx-xx

19

References
AE =
ASSMANN 2006 = J. ASSMANN, Religion and Cultural Memory: Ten Studies, Stanford 2006.
BELL 1997 = C. M. BELL, Ritual: Perspectives and Dimensions, Oxford 1997.
BNABOU 1976 = M. BNABOU, La rsistance africaine la romanisation, Paris 1976.
BNICHOU-SAFAR 2004 = H. BNICHOU-SAFAR, Le tophet de Salammb Carthage: Essai de
reconstitution, Rome 2004.
BENSEDDIK 2006 = N. BENSEDDIK, Saturne Africain et les couples: recherches iconographiques,
Revue Archologique, 2006: 208-213.
BESCHAOUCH 1968 = A. BESCHAOUCH, Une stle consacre Saturne le 8 novembre 323, Bulletin
Archologique du Comit des Travaux Historiques et Scientifiques 4, 1968: 253-268.
BONNET 2011 = C. BONNET, On Gods and Earth: The Tophet and the Construction of a New Identity
in Punic Carthage, in E. GRUEN (ed.), Cultural Identity in the Ancient Mediterranean, Malibu 2011:
373-387.
BOOKIDIS STROUD 1997 = N. BOOKIDIS R. STROUD, The Sanctuary of Demeter and Kore, Princeton 1997.
BOURGEOIS 2008 = A. BOURGEOIS, tude de la cramique antique recueillie Henchir Ghayadha,
Antiquits Africaines 44, 2008: 241-259.
BRIAND-PONSART HUGONIOT 2005 = C. BRIAND-PONSART C. HUGONIOT, LAfrique romaine:
de lAtlantique la Tripolitaine 146 av. J.-C. - 533 ap. J.-C, Paris 2005.
CADOTTE 2007 = A. CADOTTE, La romanisation des dieux: linterpretatio romana en Afrique du Nord
sous le Haut-Empire, Leiden 2007.
CARTON 1897 = L. CARTON, Le sanctuaire de Baal-Saturne Dougga, Nouvelles archives des
missions scientifiques et littraires 8, 1897: 367-474.
CARTON 1907 = L. CARTON, Note sur la dcouverte dun sanctuaire de Saturne dans la Colonia
Thuburnica , CRAI, 1907: 380-384.
CARTON 1908 = L. CARTON, Le sanctuaire de Tanit el-Knissia, Paris 1908.
CCL =
CIL =
CIOTOLA 2000 = A. CIOTOLA, Il materiale ceramico rinvenuto nella ricognizione attorno a Dougga,
in M. DE VOS (ed.), Rus Africum, Trento 2000: 58-66.
DANDREA GIARDINO 2011 = B. DANDREA S. GIARDINO, Il tofet dove e perch. Alle origini
dellidentit fenicia, Vicino e Medio Oriente 15, 2011: 133-157.
DFH = M. KHANOUSSI L. MAURIN (eds.), Dougga, fragments dhistoire, Bordeaux 2000.
DONDIN-PAYRE 1994 = M. DONDIN-PAYRE, La commission dexploration scientifique dAlgrie: une
hritire mconnue de la commission dgypte, Paris 1994.
DRIDI SEBA 2008 = S. DRIDI M. SEBA, De Tanesmat Thinissut. Nouvelles observations sur
lamnagement dun lieu culte africain, in Lieux de cultes: aires votives, temples, glises, mosques:
IXe Colloque international sur lhistoire et larchologie de lAfrique du Nord, Tripoli, 19-25 fvrier
2005, Paris 2008: 101-118.
EINGARTNER 2003 = J. EINGARTNER, berlegungen zum religisen Hintergrund nordafrikanischer
Reliefstelen des Saturn und der Dea Caelestis, in P. NOELKE et al. (eds.), Romanisation und Resistenz in Plastik, Architektur und Inschriften der Provinzen des Imperium Romanum, Mainz 2003:
601-608.
ElHami = A. FERJAOUI et al., Le sanctuaire de Henchir el-Hami: de Baal Hammon au Saturne africain,
1er s. av. J.-C. - IVe s. ap. J.-C., Tunis 2007.
FELSCH 1996-2007 = R. C. S. FELSCH, Kalapodi: Ergebnisse der Ausgrabungen im Heiligtum der
Artemis und des Apollon von Hyampolis in der antiken Phokis, Mainz 1996-2007.
FERJAOUI 2002 = A. FERJAOUI, De Baal Hammon Saturne: prsentation dun sanctuaire Henchir-

20

M. M. McCarty, Continuities and Contexts

el-Hami, in P. DONATI GIACOMINI M. L. UBERTI (eds.), Fra Cartagine e Roma, Firenze 2002:
59-77.
FONTANA 2001 = S. FONTANA, Leptis Magna. The Romanization of a Major African City Through
Burial Evidence, in S. KEAY N. TERRENATO (eds.), Italy and the West: Comparative Issues in
Romanization, Oxford 2001: 161-172.
FULFORD 1984 = M. FULFORD, The Coarse and Painted Wares, in M. FULFORD D. PEACOCK
(eds.), Excavations at Carthage: The British Mission I.2, Sheffield 1984: 155-231.
FULFORD PEACOCK 1995 = M. FULFORD D. PEACOCK (eds.), Excavations at Carthage: The
British Mission Volume II, Part 2: The Circular Harbour, Oxford 1995.
GUEST 2012 = P. GUEST, The Production, Supply and Use of Late Roman and Early Byzantine Copper
Coinage in the Eastern Empire, Numismatic Chronicle 172, 2012: 105-131.
GUTRON 2010 = C. GUTRON, Larchologie en Tunisie (XIXe-XXe sicles): jeux gnalogiques sur
lAntiquit, Paris 2010.
HUSSLER KING 2007 = R. HUSSLER A. KING (eds.), Continuity and Innovation in Religion in
the Roman West, Portsmouth, R.I. 2007.
HORDEN PURCELL 2000 = P. HORDEN N. PURCELL, The Corrupting Sea, Oxford 2000.
IlAfr =
ILTun =
JOHNS 2003 = C. JOHNS, Art, Romanisation, and Competence, in S. SCOTT J. WEBSTER (eds.),
Roman Imperialism and Provincial Art, Cambridge 2003: 9-23.
JONGELING 2008 = K. JONGELING, Handbook of Neo-Punic Inscriptions, Tbingen 2008.
KALINOWSKI 1993 = A. KALINOWSKI, Pottery from Unity 4000, in S. STEVENS (ed.), Bir el Knissia
at Carthage, Ann Arbor 1993: 155-177.
KRANDEL-BEN YOUNS 2002 = A. KRANDEL-BEN YOUNS, La prsence punique en pays numide,
Tunis 2002.
LANTIER POINSSOT 1942 = R. LANTIER L. POINSSOT, Les stles dcouvertes dans un favissa du
temple de Saturne Dougga, Bulletin Archologique du Comit des Travaux Historiques et
Scientifiques 1942: xiixxviii.
LE GLAY 1961 = M. LE GLAY, Saturne africain: monuments I, Paris 1961.
LE GLAY 1966a = M. LE GLAY, Saturne africain: monuments II, Paris 1966.
LE GLAY 1966b = M. LE GLAY, Saturne africain: histoire, Paris 1966.
LEONE 2007 = A. LEONE, Changing Townscapes in North Africa from Late Antiquity to the Arab
Conquest, Bari 2007.
LZINE 1959 = A. LZINE, Architecture punique, Paris 1959.
LIPKA 2009 = M. LIPKA, Roman Gods: A Conceptual Approach, Leiden 2009.
MCCARTY 2010a = M. M. MCCARTY, Votive Stelae, Religion, and Cultural Change in Africa
Proconsularis and Numidia, unpub. DPhil dissertation, University of Oxford, 2010.
MCCARTY 2010b = M. M. MCCARTY, Soldiers and Stelae, Bolletino di Archeologia Online 1, 2010:
(151.12.58.75/archeologia/bao_document/articoli/4_McCARTY.pdf).
MCCARTY 2011a = M. M. MCCARTY, Representations and the Meaning of Ritual Change: The Case
of Hadrumetum, in A. CHANIOTIS (ed.), Ritual Dynamics in the Ancient Mediterranean, Stuttgart
2011: 203-234.
MCCARTY 2011b = M. M. MCCARTY Y, Beyond Models and Diffusion, Centers and Periphery:
Religious Art in Roman Africa, in T. NOGALES I. ROD (eds.), Roma y las provincias, Roma
2011: 439-448.
MCCARTY, forthcoming = M. M. MCCARTY, Empire and Worship in Roman Africa, Cambridge.
MCHAREK 1982 = A. MCHAREK, Aspects de l'evolution dmographique et sociale Mactaris aux IIe
et IIIe sicles ap. J.C, Tunis 1982.
MCHAREK et al. 2008 = A. MCHAREK et al., Recherches archologiques Henchir Ghayadha/Bagat?
(Tunisie), Antiquits Africaines 44, 2008: 111-167.

SEL 29-30, 2012-2013: xx-xx

21

MERLIN 1910 = A. MERLIN, Le sanctuaire de Baal et de Tanit prs de Siagu, Paris 1910.
MOMMSEN 1951 = T. MOMMSEN, St. Augustine and the Christian Idea of Progress, Journal of the
History of Ideas 12, 1951: 346-374.
NONGBRI 2013 = B. NONGBRI, Before Religion, New Haven 2013.
OCK =
PENSABENE 1990 = P. PENSABENE, Il tempio di Saturno a Dougga e tradizioni architettoniche
dorigine punica, AfrRom 7, 1990: 251-293.
PICARD 1984 = C. PICARD, Le temple du muse Mactar, Revue Archologique 1984: 13-28.
POINSSOT 1911 = L. POINSSOT, Inscriptions de Thugga, Revue Tunisienne 18, 1911: 64-75.
POINSSOT 1958 = C. POINSSOT, Les ruines de Dougga, Tunis 1958.
POINSSOT LANTIER 1925 = L. POINSSOT R. LANTIER, Lglise de Thugga, Revue
Archologique 22, 1925: 228-247.
REECE 1984 = R. REECE, Coins, in H. HURST S. ROSKAMS (eds.), Excavations at Carthage: The
British Mission I.1, Sheffield 1984: 171-181.
RICE 2011 = C. RICE, Ceramic Assemblages and Ports, in D. ROBINSON A. I. WILSON (eds.),
Maritime Archaeology and Ancient Trade in the Mediterranean, Oxford 2011: 81-91.
RICHARD 1961 = J. RICHARD, tude medico-lgale des urnes sacrificielles puniques et de leur contenu,
unpub. PhD dissertation, Universit de Lille 1961.
RIGGS 2001 = D. RIGGS, The Continuity of Paganism between Cities and Countryside of Late Roman
Africa, in T. BURNS J. EADIE (eds.), Urban Centers and Rural Contexts in Late Antiquity, East
Lansing 2001: 285-300.
RIVES 1995 = J. B. RIVES, Religion and Authority in Roman Carthage, Oxford 1995.
RPKE, forthcoming = J. RPKE, Perspektiven fr eine Reichsreligion und die gesellschaftliche
Integrationskraft des Imperium Romanum, in M. JEHNE et al. (eds.), Religise Vielfalt und soziale
Integration, Mainz, forthcoming.
SAINT-AMANS 2004 = S. SAINT-AMANS, Topographie religieuse de Thugga (Dougga): ville romaine
dAfrique proconsulaire (Tunisie), Pessac 2004.
SAUER 2011 = E. SAUER, Religious Rituals at Springs in the Late Antique and Early Medieval World,
in L. LAVAN M. MULRYAN (eds.), The Archaeology of Late Antique "Paganism", Leiden 2011:
505-550.
SCHRNER 2008 = G. SCHRNER, New Pictures for Old Rituals, in B. CROXFORD et al. (eds.),
TRAC 2006, Oxford 2008: 92-102.
SEARS 2011 = G. SEARS, The Fate of the Temples in North Africa, in L. LAVAN M. MULRYAN
(eds.), The Archaeology of Late Antique "Paganism", Leiden 2011: 229-260.
SHAW 2003 = B. D. SHAW, A Peculiar Island: Maghrib and Mediterranean, MHR 18, 2003: 93-125.
SHAW 2011 = B. D. SHAW, Sacred Violence: African Christians and Sectarian Hatred in the Age of
Augustine, Cambridge 2011.
SHAW 2013 = B. D. SHAW, Lambs of God, unpub. ms. in progress.
SOURVINOU-INWOOD 1989 = C. SOURVINOU-INWOOD, Continuity and Change in Greek Religion,
CR 39, 1989: 51-58.
STEWART 2003 = P. STEWART, Statues in Roman Society: Representation and Response, Oxford 2003.
TILMANT 1989 = P. TILMANT, Introduction au phnomne des temples cour en Afrique romaine,
Revue des archologues et historiens d'art de Louvain 22, 1989: 9-16.
TILMANT 1995 = P. TILMANT, Dougga (Tunisie): tude du Temple de Tellus, Revue des
archologues et historiens dart de Louvain 22, 1989: 21-30.
TOUTAIN 1892 = J. TOUTAIN, Le sanctuaire de Saturnus Balcaranensis au Dj. bou Kournein,
Mlanges de lcole Franaise de Rome 12, 1892: 3-124.

M. M. McCarty, Continuities and Contexts

22

Figures

Fig. 1: Stele-sanctuaries established between 2nd century BC and 2nd century AD


(author; base-map data under license from Esri, USGS, NOAA)

Fig. 2: Sites discussed in the text (author; base-map data under license from Esri, USGS, NOAA)

SEL 29-30, 2012-2013: xx-xx

Fig. 3: Sanctuary at Henchir el-Hami, 1st century BC-5th century AD


(ElHami, plan, reproduced with kind permission from A. Ferjaoui)

23

24

M. M. McCarty, Continuities and Contexts

Fig. 4: Special deposit at Hr. el-Hami, 5th century AD


(ElHami fig. 85, reproduced with kind permission from A. Ferjaoui)

Fig. 5: Roscianus temple to Saturn, Thugga, AD 195 (after POINSSOT 1958, fig. 7)

SEL 29-30, 2012-2013: xx-xx

Fig. 6: Tophet deposits, with carved stelae over buried urns,


in situ below temple at Thugga, 2nd century BC-1st century AD (POINSSOT 1958, pl. 19)

25

26

M. M. McCarty, Continuities and Contexts

Fig. 7: Stairs of the original entrance to Roscianus temple,


Thugga, later obsoleted by building of porch, AD 195 (author)

You might also like