Академический Документы
Профессиональный Документы
Культура Документы
OF RITUAL CHANGE
The Case of Hadrumetum
Matthew M. McCarty
INTRODUCTION1
Ritual practices throughout the Mediterranean and beyond underwent a series of
important changes over the period of Romes hegemony. Yet while these changes
are often recognized, they are rarely discussed in depth, particularly for regions
outside of Italy; when they are treated, assumptions are made about how and what
ritual can mean or do, and frequently such assumptions differ based on the
region under discussion. By focusing on a particular case study, a sacrificial rite
that occurred in a relatively similar form over seven centuries in the North African
settlement of Hadrumetum (modern Sousse in the Tunisian Sahel; figure 1), I
want to reflect on how ritual and its transformations can be interpreted. This particular rite, usually involving sacrifice, often of an infant, the burial of the sacrificial remains in an urn, and the erection of a commemorative stele over this deposit, occurred in many Western Phoenician/Punic settlements throughout the
central Mediterranean from the Archaic period.2 In the Classical and Hellenistic
I am grateful to Simon Price for discussing the material with me, and to Angelos Chaniotis,
Teresa Morgan, Jo Quinn, and Charlotte Potts for their comments on portions of the text. I am
also deeply grateful for the assistance and hospitality offered to me by the Institut National du
Patrimoine in Tunisia, and in particular, M. Habib Ben Younes and Mme. Lamia Fersi, and to
Mme. Wafa Messaoudi for discussing her current dissertation research on the chronology of
the material from Hadrumetum. This research was conducted with a grant from the Tweedie
Exploration Fund.
On the rite in general: Le Glay 1966b, 297-358; Mosca 1975; Bnichou-Safar 1988 and 2004;
Ferjaoui 2007. The definition of Punic is, of course, problematic: Prag 2006. Whether or not
child sacrifice was actually practiced, or whether the urns represent the symbolic dedication
of pre-deceased children to a deity, remains hotly contested: cf. Moscati and Ribichini 1991,
with earlier bibliography; although more recent studies on the topic have appeared, these rarely break new ground. While the ancient texts that describe child sacrifice may be engaging in
some form of derogatory othering similar charges were levelled against both Christians
and Jews at present, the nature of the osteological evidence precludes a firm conclusion
about the nature of this sacrifice. The inscriptions on stelae, however, may well point to child
sacrifice.
198
Matthew M. McCarty
periods, however, the rite ceased in most areas.3 By contrast, from the second century BCE, the practice spread to a range of sites along the North African coast and
in inland Tunisia and Algeria; over the Imperial period, it came into even wider
use, spreading further inland as a range of military and commercial networks developed and encouraged the movement of people, some of whom brought such
practices with them.4 My argument is three-fold: in general, that there is a relationship between ritual forms and discursive religious concepts or meanings, visible when the archetypal form of a liturgy is altered; on the methodological front,
that understanding ritual change is best done by examining the frameworks used
to describe and represent ritual acts; and on the historical front, that incorporation
within the Roman Empire deeply altered the ways in which ritual communication
with the gods was conceived, at least in the region under examination.
Ritual lies at the heart of most recent accounts of the religion of the Urbs; indeed, for Scheid and others, Roman religion is ritual.5 Beard, North and Price
provide an account of Roman religion focused largely upon practice that emphasizes the protean nature of any meanings fixed upon a rite by ancient participants
or modern scholars.6 While drawing on ritual to paint a picture of Roman religion,
a particular relationship or lack thereof between gesture and significance is
supposed, often supported by contemporary anthropology; the recognition that
meaning can lie in reception, for example, has resulted in the awareness that the
same rituals can have a range of potentially divergent meanings for different participants.7 Such possibilities have created a distrust in pinning significance beyond
social function on any set of formal actions, and have led to an effort to strip many of the conceptualized layers of ritual away, resulting in a presumed divorce
between ritual form and meaning beyond Scheids epistemological realism, or at
least a gap largely unbridgeable by the modern scholar.8 Yet, as examination of
4
5
6
7
8
Outside of Africa, the rite is attested in Sicily at Motya from the late eighth to the early third
century BCE, although stelae seem to cease being produced around the time of the Greek
conquest in 397 (Moscati 1992, 19-21). In Sardinia, definite evidence for the rite appears at
Nora from the sixth century BCE to the Hellenistic period (Moscati and Uberti 1970, 43-45);
at Sulcis from the eighth century to the third century BCE (Moscati 1986, 81-84); at Monte
Sirai from the fourth century to the second century BCE (Moscati 1992, 27); and at Tharros
from the seventh to the third century BCE, although no stelae are attested in the very last phases of sanctuary use (Moscati and Uberti 1985). There are also more fragmentary remains
probably tied to this the rite in Classical Cagliari (Moscati 1981).
Ben Abid Sadallah 2003, 103-109; McCarty forthcoming.
E.g., Scheid 2005; Scheid 2007. There are notable exceptions, including Bendlin 2001; Ando
2008 for Roman religion as knowledge (without ample consideration of what religious
knowledge might be and how it is constructed), but accepting many of Scheids premises.
On ritualism as modern area of enquiry and methodology that predetermines certain conclusions, Bremmer 1998.
Beard, North, and Price 1998, 48, 125-134.
Cf. Beard, North, and Price 1998, 48; Rpke 2007, 110; Scheid 2007.
Cf. Chaniotis 2005, 144; Scheid 2005, for a minimalist view that ritual establishes a power
hierarchy. Rpke 2007, 97-106, tries to bridge the gap via structuralist categories and an idea
of ritual as communicative itself a problematic premise (cf. Humphrey and Laidlaw 1994,
199
Hadrumetum shall show, such a divorce need not be absolute; instead, there are
clearly-observable relationships between ways of thinking and liturgical acts, and
ones which bespeak a discourse on ritual.
Ritual is far more rarely discussed in accounts of religion in the Roman Imperial period beyond Italy, even in studies that are interested in looking at diachronic
cultural and religious shifts; instead, the deities and social frameworks of cult are
often given priority.9 When ritual change in the Roman provinces is discussed,
liturgical forms are often taken to have cultural weight, with changed rituals reflecting Romanization (or any number of roughly equivalent terms used to describe the process of cultural change) and continued rites from the pre-Roman into
the Imperial period demonstrating cultural stasis or resistance.10 Frequently, the
changes singled out are new actions added to and occasionally replacing the
types of rites that existed before; in the northwest provinces, for example, practices such as the dedication of arms in bodies of water seem to give way to rituals
such as the erection of votive altars, while Derks argues that the Roman votum
was adopted as a new practice.11 In North Africa, the dedication of votive statuary
in stone seems to represent a new ritual act in the Imperial period.12 Such new
features are often suggested to be evidence of Romanization in themselves. 13
While such transformations are important, and may suggest much about the potential for additive extension in such a polytheistic system beyond the pantheon and
into the realm of cult practices, less frequently discussed are changes within regionally-specific rites that continue in their basic form into the Imperial period.14
The sacrificial rite at Hadrumetum falls into this latter category. The supposed
continuity of this ritual has been used as evidence either for the permanence de
la psychologie religieuse des Africains or for cultural resistance.15 Underlying
10
11
12
13
14
15
72-76). For ancient Athens, Parker 2005, 155-159, 369-379, provides a nuanced view of what
ritual could accomplish in a social context.
As much as it might be explicitly disavowed, the basic thesis of Toutain 1907, which divided
gods into the categories based on origin, has been largely maintained in studies of Roman
provincial religion; deities are taken as most indicative of different religious and cultural
forms of expression. This deity-based approach most recently for North Africa: Seba 2005;
Cadotte 2007.
E.g., Derks 1998, with the votum as a Roman cultural form in the northwest, followed by
van Andringa 2002; Millett 1995, 99, suggesting ritual changes can demonstrate cultural
synthesis, and thus that individual forms are culturally marked in isolation; Fulford 2001,
looking at special deposits in Britain as a cultural practice with an assumed bond between action and mentality.
Haynes 1997; Derks 1998, 217-231.
Schrner 2009, 257f.; it is worth noting though that small terracotta figurines have been
found in pre-imperial sanctuaries in Africa, the appearance of larger figures of three-quarters
lifesize to over-lifesize scale primarily in the first century CE may be tied to the more widespread adoption of a statue habit which included the dedication other forms of statuary,
including honorifics.
E.g., Haynes 1993; Derks 1998, 231; Schrner 2009, 257.
Cf. Bendlin 1997.
E.g., Le Glay 1966b, 492; Bnabou 1976, 370-375; Ben Abid Sadallah 2003. Exceptions:
Rives 1995, 142-151, suggesting a process of change and Romanization driven by local eli-
200
Matthew M. McCarty
both the accounts that see new rites as markers of Romanization and those that
see continued rites as evidence of resistance or localness is the premise that
ritual form is indicative of culture, loosely defined as a set of values, worldviews,
and their corollary actions. While few archaeologists would accept the addition of
Italian terra sigillata to a local repertoire as indicative of a deep cultural change,
somehow liturgical form has been seen as a fundamentally different form, capable of conveying mentalities or culture in itself.
The scholarship around religion at Rome and in the provinces thus moves in
two very different directions, based on the assumed meanings and significances
(or lack thereof) of ritual form. This is not entirely surprising given that anthropologists, often cited in support of one view or another when interpreting ancient
material, are themselves divided over the potential of ritual and changes therein to
signify anything, ranging from Staals complete denial of meaning, to Humphrey
and Laidlaws view (particularly popular among Classicists) that ritualized action
is a semiotic blank upon which meanings are projected, to Rappaports more optimistic view that ritual is fundamentally a set of socially-encoded meanings.16
Without automatically denying that liturgical actions may lack any intrinsic
meaning, it is thus necessary to see how much can be said about the relationship
of ritual forms to ideas, either individual or social, to establish inductively the
possibility (or lack) of connection between forms and thought patterns or culture
in the Roman world. Looking at periods of change offers the best arena to do so,
for it is in dynamic shifts that such relational systems, in this case between form
and conception, become clearest. As a close examination of the material from
Hadrumetum shall show, ritual changes, when they affect the basic archetypal
structure of a liturgy, can be associated with wider changes in thought and culture.
Demonstrating such a bond, even if loose, can pave the way for wider conclusions
about the historical significance of ritual forms and their transformations.
In order to examine this relationship, it is first necessary to survey the archaeological evidence from Hadrumetum. Then, it will be possible to make a firstorder interpretation of this material to reconstruct at least those ritual forms which
have left archaeological remains and to isolate diachronic changes and variations
in the rites. Finally, arguing that representation provides at least some evidence
for social knowledge and conceptualization, it will be possible to see the relationships between ritual form and patterns of thought.
16
te; Schrner 2007a, looking at the changed contexts, such as timing, of the sacrificial rite in
the Imperial period.
Staal 1989; Humphrey and Laidlaw 1994; Rappaport 1999. It is worth noting, however, that
Humphrey and Laidlaw eagerly generalize about the universal aspects of ritual theory from
Jainism, whose fundamental supposition between the emptiness of external, personal ritual
gestures and the transformation of internal spiritual states is not at all commensurate with general ideas about the nature of Roman religion put forth by those Romanists who cite Humphrey and Laidlaw.
201
Foucher 1964.
I use the term tophet, attested in the Hebrew Bible to denote a sanctuary outside of Jerusalem
where children were sacrificed, only as a conventional descriptor for this type of sanctuary in
North Africa where burnt offerings were buried in urns.
202
Matthew M. McCarty
change in ritual forms, or at least in those forms that entailed the manipulation of
the physical environment and objects in a lasting manner. 19
The earliest excavated level in the tophet, Level 1, dated to the seventh/sixth
century-fifth century BCE, contained a series of urns buried in small shelters built
from stones.20 Inside the urns, the burned remains of infants, still with their nursing teeth, were discovered. In addition, amulets and jewellery, many with Egyptianizing motifs such as the Horus falcon, were placed in the vessels. The urns
were frequently covered with sherds of pottery before deposition.
The second level in the tophet, Level 2, was separated from Level 1 by a level
of stone paving slabs, laid probably in the late fifth century BCE, and layer of
black earth in which urns were deposited; this level continued until the third century BCE. The urns, which still contained infants but also began to include lambs
either with the children or alone, contained less jewellery. The vessels were fre19
20
The description of the site will follow the data presented in a summary fashion in Cintas
1948; although Cintas claims to have deposited his original fieldbooks in the Archives du
Service des Antiquits et Arts de Tunisie (3 note 11), archival research has failed to locate
them. I am grateful to my colleagues at the INP for helping me try to find them.
The chronology of the tophet at Hadrumetum is currently undergoing revision thanks to new
research into the ceramics assemblages by Wafa Messaoudi, which will hopefully offer more
precise dates for the transitions at the site, particularly in the earlier phases. Until her thesis is
complete, Cintas dating must remain in place, but I am grateful to Mme. Messaoudi for her
preliminary thoughts. Fortunately, for the arguments put forth here, the relative chronology
matters more than the absolute chronology, and this remains unchanged from Cintas report.
203
quently topped with ceramic incense burners, rather than the broken sherds of
Level 1. In addition, a series of carved stelae were discovered; at least two were
found in situ, their bases planted over the deposits. These stelae will be discussed
in detail later, but it is worth signalling some of their decoration in broad terms:
one stele shows a worshipper approaching an enthroned deity; two others show
figures before large incense burners.
At some point in the third century BCE, the ground level in the sanctuary was
again raised and levelled, this time with a layer of yellow sand and then a black,
sandy fill, creating Level 3, which appears to have been in use until the early second century BCE. Urns continued to be deposited in this layer, still topped with
incense burners, although jewellery was no longer offered, and animal remains
became as common as human.21 The stelae erected in this level were much larger,
and for the first time, a series was inscribed with dedications in Punic; most followed a well-known formula in use also at Carthage and a tophet in Cirta, beginning with an invocation to the deity, continuing with the formula ndr ([that]
which vowed) and the name of the dedicant with patronymics, and often ending
with some variation of the phrase km ql brk (because he heard [his] voice, he
blessed [him]).22 It is worth noting that these stelae were found at a slightly higher elevation than the numerous anepigraphic stelae of the level, and it may thus
be possible to speak of two distinct phases within the same stratum, 3a and 3b.
Those of level 3b were also carved from a different stone than the earlier stelae.
At the base of one of the stelae were discovered a small lead spoon and candelabrum atop which a lamp or incense burner could be placed.
Then, in the early second century BCE, the ground level was raised substantially as a very thick layer of sand, topped with ash, probably the remains from
ritual pyres, was spread through the sanctuary, sealing Level 3 and creating a new
set of deposits in Level 4. Urns containing burned bones continued to be buried in
this level, although the proportion of small mammals came to nearly equal that of
infants. Stelae, generally of smaller dimensions than in Level 3, were placed
above the deposits; Cintas suggested that some showed signs of having been reworked from Level 3 stelae, especially to make them exactly the same height as
the increasingly standard-sized stelae of Level 4.23 The stelae in this level were
frequently inscribed, and decorated with a limited range of motifs, including bottles, baetyls, and the so-called sign of Tanit, in very low relief or painted on
stucco. In addition to the now tightly-packed rows of relatively homogenous-sized
stelae, a 4x4m area of burned earth was discovered, devoid of stelae. The charring
of the ground suggests that a pyre (or series of pyres) had been constructed there.
21
22
23
It is worth noting that Stagers excavations in the tophet at Carthage suggested the exact opposite trend, with more infants later: Stager 1982. While this could point to regional variation,
it may also suggest that any report on the contents of the Hadrumetum urns, never published
in full, was incomplete or not entirely accurate.
On the epigraphic stelae, Fantar 1995.
Cintas 1948, 35.
204
Matthew M. McCarty
The penultimate level, Level 5, sat atop Level 4. A coin of Germanicus, struck
as part of a restitution series under Titus, was discovered among finds for this level, offering a terminus post quem for the closing of the stratum in the late first
century CE.24 Urns continued to be buried in this phase of the tophets use, but
contained only animal remains; infant bones disappear at this point. The deposits
were also packed more densely, with each stele sitting over several urns. The stelae themselves in this level tended to be smaller and thinner, and most showed
evidence of fire damage. Although the basic iconographic repertoire included motifs used in the earlier levels, six stelae from this layer represent an entirely new
subject: sacrifice at an altar. Numerous bronze coins of the Demeter/horse type,
attested in use at other sites through the first century CE, along with a series of
round, coin-sized lead cut-outs were found all through this level, seemingly placed around the stelae.25
The fire damage on the Level 5 stelae may be indicative of a disaster which
led to the creation of the final ancient level found at the site, Level 6. Over Level
5, a series of large paving stones were laid, sealing the level, at some point after
the late first century CE. The paving stones also included broken and damaged
stelae from an earlier level, perhaps part of a taboo that demanded such offerings
be kept within the sanctuary, a phenomenon visible elsewhere across the region. 26
In areas, Level 5 was preserved with stelae and offerings in situ under the pavement; elsewhere, any remains of Level 5 were cleared away first, leaving Level 6
right above Level 4. Stelae were planted on this pavement, around which were
deposited numerous small unguentaria; no urns were buried at this level, nor were
bones found. The stelae, squatter in form than those of Level 5, some of which
were recarved from earlier levels, frequently depict single animals, especially
bulls and rams.
The six strata from the site clearly reveal a series of changes in ritual; yet how
much can be said about the liturgical forms from this data?
24
25
26
205
27
28
29
30
31
Bnichou-Safar 1988.
Ferjaoui 2007.
Rappaport 1999.
Archetypal is here understood as an external, canonical form of a particular set of gestures,
akin to how it is defined in Humphrey and Laidlaw 1994, 158, as opposed to a more universalized sense of archetype.
Chaniotis 2005, 150. A similar hiatus may also be observable in a North African sanctuary
which is often pointed to as evidence for cultic continuity from the pre-imperial period into
the High Empire, Thugga; analysis of the ceramics recorded suggests that there is no evidence
206
Matthew M. McCarty
Ritual sacrifice and deposition must have been the most important event if
not the only one to take place in the tophet at Hadrumetum. The arrangement of
the stelae in tightly-packed rows from at least Level 4, and especially in Levels 5
and 6, limited the amount of open space available for circulation or gathering in
the sanctuary. While Cintas excavations uncovered only a portion of the tophet,
and excavations in Carthage have not revealed the full extent of the Salammb
sanctuary, the amount of free space seems to have been limited by the proliferation of stelae and deposits; as a point of comparison, Henchir el-Hami seems to
have consisted solely of space for the stelae and urns to be deposited, and perhaps
a dump for sacrificial waste including excess bones and ash further away. 32
Furthermore, these spatial limitations suggest that each individual sacrificial event
was hardly a grand spectator occasion, but involved only a small number of participants in the ritual. Up until at least the second century CE, when many open-air
stele fields had temples built over them, these sites were hardly multi-purpose
sanctuaries, but were created and maintained for the purpose of one particular
ritual process: sacrifice, deposition, and stele erection.33
Little can be said about the preparations made for the sacrifices that took place. The archaeological documentation begins at the pyres upon which the sacrificial victims were incinerated; whether any processions were involved (as seems
unlikely given the spatial restraints), how the victims may have been prepared or
decorated, the prayers and music which may have accompanied the acts cannot be
known. The inscribed stelae offer some hint of a preliminary action by referring to
a ndr a vow. But what exactly taking such a vow involved is unclear. Such a ndr
may well have been undertaken to achieve a specific goal, as Kleitarchos suggests
in his third century BCE comments on Carthaginian child sacrifice, and certainly
involved the promise of a victim.34 Presumably, once this goal had been met, the
32
33
34
of cult at the site in the first through the late second centuries CE: Krandel-Ben Younes 2002,
171.
The only real open spaces detected at Salammb were a series of paths designed to allow the
circulation, in a relatively limited manner, of people through the sanctuary, rather than for
large processions; the largest of these was only 1.5 m wide: Icard 1922, 205; Lapeyre 1935,
83. At Henchir el-Hami, although the space was not delimited by a temenos wall, the rows of
stelae left little room in the sanctuary space for circulation, nor were there discernable traces
of contemporary activity beyond the confines of the stele field: Ferjaoui 2007, 20.
Over the course of the first and second century CE, a series of built temples was constructed
that gradually replaced the tophets. At Thinissut, a built sanctuary can be dated to the first
century CE thanks to the presence of terracotta vaulting tubes, which came into use only in
the Augustan period: Merlin 1910, 14f., 27; I am grateful to Lynn Lancaster, pers. comm., for
the dating. At Henchir el-Hami, the tophet went out of use and rites moved to an altar and
building in the late second century: Ferjaoui 2007. At Thugga, a temple was built in CE
194/5: Khanoussi and Maurin 2000, no. 38 = CIL VIII 26498; a similar temple was constructed at Ammaedara between 198 and 208 CE: LAnne pigraphique 1912, 209.
The Phoenicians, and especially the Carthaginians, honouring Kronos, whenever they seek to
obtain some great favour, vow one of their children if they are especially eager to gain success, dedicating it (kathagien) to the deity. Kleitarchos, Die Fragmente der griechischen Historiker 137 F 9 ed. Jacoby.
207
votary had to keep his (or, more rarely attested, her) end of the deal, and offer the
sacrifice.
Due to the fascination of modern scholarship with child sacrifice, the infant
remains from Hadrumetum have been studied far more closely than the remains of
the animals, and so more can be said about the treatment of the former.35 Specific
sites seem to have been left clear for the building of temporary pyres, which received no monumental or lasting structure; the 4x4 m patch of charred earth found
in Level 4 at Hadrumetum would have been one such site.36 The even burning of
the bones suggests that the children were already dead at the time of their cremation, and that they were placed on the pyres on their backs.37 As the burning went
on, portions of lambs or goats, usually new-born, could be thrown onto the pyre;
their bones, slightly less calcified than the infants, were found associated with
approximately 15-20% of the sacrificial remains that were examined from Hadrumetum and Carthage.38 In addition, small animals could also be cast onto the
pyre.39 The remains of the pyre were then cooled with water, resulting in longitudinal fractures on the bones from the sudden change in temperature.40 The bones
were collected and placed in an urn; even small bones were collected, with very
little ash in most cases, suggesting that this process also involved sieving the pyre
remains.
After sacrifice came deposition. In the earliest phases, Levels 1 and 2, jewellery and beads could be offered along with the victim, set in the urn. The vessel
was then sealed; the incense burners used for such a purpose in Levels 2-4 suggest
that at some point prior to this phase, perhaps during the burning, sieving or in
some intermediate step, they were used to provide pleasing odours. The urn was
then buried in the sanctuary; the organization of stelae bespeaks some form of
sanctuary authority responsible for overseeing placement.
Other rites followed the deposition. A stele had to be carved and erected. It is
unclear whether the stelae were set up immediately following sacrifice and urn
deposition or at a later date; if the former, the worked stone monument would have had to have been commissioned well in advance of the sacrificial rite.41 Further
offerings were made and rites conducted around the stelae, above the deposits.
The discovery of numerous lamps at the base of the stelae has long suggested that
35
36
37
38
39
40
41
Bnichou-Safar 1988; the new study from Henchir el-Hami looks far more closely at the
animal remains: Bdoui and Oueslati 2007.
A similar space was discovered at Salammb: Bnichou-Safar 2004, 56.
Bnichou-Safar 1988, 60.
Bnichou-Safar 2004, 119.
Bnichou-Safar 2004, 52.
Bnichou-Safar 1988, 65.
The personalization of many of the stelae strongly suggest that they were not bought off-therack, at least at Carthage. Some of the stelae, for example, represent aspects of the dedicants
profession, such as the hammer and tongs, presumably for a metalworker, on Corpus Inscriptionum Semiticarum I, 735. That being said, several display features which may suggest the
possibility that they were purchased pre-made, with only the names filled in by the stonecutter; for example, Mendleson 2003, 21 no. Pu6, ends with the name of the dedicant, Amotmelqart, daughter of, without any evidence that the fathers name had ever been added.
208
Matthew M. McCarty
these rituals took place at night, although a more symbolic use of such paraphernalia should not be excluded.42 Likewise, the bronze stand and incense (?) spoon
discovered in Level 3 suggests the use of incense again at this stage, and the scattering of either coins or substitutes from Level 5 onwards. Given the limitations of
archaeological chronologies, it is only possible to say that such events took place
after the erection of the stelae, not how long after. Such post-stele offerings may
represent a continuation of a single set of actions begun with the building of a
pyre, or the repeated use of the deposition site as a locus of ritual observance on
further future occasions.
Many of the details offered by the archaeological record are hazy, but still allow a general picture of the ritual practice at the site to be developed. This description of what happened at the Hadrumetum tophet offers a kind of ideal, archetypal ritual form. Assuming that there was such an ideal, as there must have been
to create such observably similar archaeological deposits resulting from the repetition of particular acts, the variations to this basic form require comment.
Such variations might affect the course of the ritual, and the type of traces it
left, without changing its fundamental form or significance of the ritual. For example, in his discussion of Ndembu ritual, Turner describes how certain colours
were key to the ceremony; new objects or features of the rite could be added or
substituted as long as they fit into the basic structural scheme of colour-coding. 43
Similarly, Humphrey and Laidlaw describe how within a fixed liturgical form of
Mongolian sacrifice, the requisite sheep was unavailable so a goat was used instead, without altering the nature or perceived significance of the ritual.44 Variations might mean nothing at all.
On the other hand, variations might well have communicated something.
Rappaport argues discursively that while the invariant part of a ritual encoded
canonical social knowledge through the very structure it provided, the variations
in ritual could present self-referential messages about the actors.45 Before considering the potential meaning (or lack thereof) of any such variation in forms, it is
worth first trying to isolate the different variables that did change.
There are a series of permutations in the rites that occur within the same levels. Up until the beginning of Level 6, the basic liturgical structure of vowsacrifice-deposition-stele erection-further offerings remained relatively constant.
The two clearest variations across the entire span of the sites use are in the number and type of victims offered. Many of the stelae, particularly from Levels 3-5,
sat atop multiple urns, each containing a burned victim. The number of urns varied, suggesting multiple victims offered within a short time span before the erection of the stele; the same is true both at Carthage and at Henchir el-Hami.46 The
42
43
44
45
46
Gielly 1927, 14, remarking that many of the stelae had smoke-blackened bottoms, suggesting
lit lamps at their feet; Cintas 1948; Le Glay 1966b; Ferjaoui 2007.
Turner 1967, 59-92.
Humphrey and Laidlaw 2007, 265.
Rappaport 1999, 50-56.
Bnichou-Safar 2004; Ferjaoui 2007, 50.
209
rhetoric of the inscriptions on the stelae over the deposits makes clear that these
were offerings made by individuals; none of the published inscriptions from Hadrumetum marks a joint dedication, and these are extremely rare at Carthage and
other tophets as well.47 The variable marked by the greater numbers of offerings
was not necessarily the participation of more offrands, but instead some aspect of
the status of the individual sacrificant involved or the situation that occasioned the
act. Such a difference in number of victims may well have carried with it some
social or religious significance, but this almost certainly had to do with the specific circumstances of the offering; for example greater desperation in seeking the
gods favour or greater wealth or standing may well have encouraged larger sets
of offerings. What is clear from the multiple deposits is that, at least in those
aspects of the liturgy which left physical traces, the same rite occurred as with
single offerings, but simply expanded to include more victims. Such variation,
then, represents not a change in the structure of the ritual, but something specific
about the circumstances in which a particular offering was made.
Another aspect of the ritual seems to have been variable at first, but then to
represent a change in liturgical form: the victims offered. While there appears to
have been a mixture of animals and infants in the offerings deposited in Levels 24, this ratio gradually changed until Level 5, when only animals were discovered
in the urns. This change certainly did not mark the kind of natural evolution away
from barbarism that Le Glay casts it as, nor can this be seen directly as intervention from Rome, the result of the oft-discussed Tiberian edict banning human sacrifice; a similar practice of infant deposition continued well into the second century
CE at Henchir el-Hami.48 In the early phases, when the two types of offerings
were made side-by-side, the lambs may have been the kind of substitution sacrifice they are often cast as, akin to the dedication of coin-shaped lead sheets instead of real coins in Level 5.49 It may be impossible to say what specific factors
led to the diachronic change in victims. Yet perhaps the most notable aspect of
this change is not the cessation of infant sacrifice itself, but the movement from
what was originally a liturgical variable to a canonical sacrifice of animal victims
alone.
In addition, at several points over the course of the sites history, features of
the liturgical practice changed, and phases of the action either done away with or
added. In Levels 2-3, the deposition of jewellery in the urns stopped, something
cast by some as a change in eschatological views but which could equally be representative of changes in social patterning and treatment of valuables.50 In Level
47
48
49
50
Only one joint dedication from Carthage is attested in CIS: I 384. None are found at Cirta. A
few more are attested in the Imperial period on votive stelae without find contexts from a limited number of sites, including Cuicul and Nicivibus, often made by husbands and wives.
Le Glay 1966b, 314; Ferjaoui 2007, 98-102. Rives 1994 convincingly argues that evidence
for such an enforced ban on the practice is lacking, and the references in Tertullian are rhetorically motivated.
Le Glay 1966b, 332-340.
Eschatological view: Bnichou-Safar 2004, 164.
210
Matthew M. McCarty
5, coins were incorporated in rites around the stelae, a feature also found at the
contemporary tophet at Henchir el-Hami.51
The most important changes appear to have occurred in Level 6, however,
when the archetype of the liturgy underwent a major transformation. The act of
gathering ashes from a pyre in an urn and burying that urn ceased; no such deposits were found in Level 6. This has frequently been used to argue that sacrifice as
a whole ceased at the site, a view which will be challenged below; the only change directly observable in the archaeological record is this lack of deposition.52
Stelae continued to be set up, however, and offerings of unguentaria, presumably
at least some of which held perfumed oil, were made around the stelae. This represents perhaps the largest change in ritual form seen at the site.
Did such shifts matter, changing the significance of the ritual, or can they be
given cultural labels, tied to broader systems of ideas, outlooks, and values? Are
such dynamics in the rituals simply evidence that even within a fixed and emphatically traditional structure, variations could occur, perhaps at most altering any
nuance of meaning that could be attached to the rites by participants? Answering
such questions, and understanding the import of changes in ritual form, requires a
few words about the interpretation of ritual.
211
ual.55 Most notably, Rappaport suggests that form and significance cannot be fully
divorced in ritual.56 The canonical, invariant form of a rite might create meaningful content via, at a minimum, its ordering of actions, while variables could demonstrate something about the particular circumstances, especially the social contexts, of the rite enacted.57 Such accounts at least raise the possibility that changes
in ritual can suggest wider changes in thinking. The problem, of course, is still
moving from actions to interpretation without presupposing some model of how
behaviours might encode any meaning. Since access to the full system of conventions and signs which may have been shared by the worship community at Hadrumetum is impossible, firmly defining any ideational content of the rituals also
remains elusive. Instead, then, of looking at the potential content of ritual meaning, it may perhaps be better to look at the ways in which the act itself was described, for in such descriptions, a relationship between thought and ritual action
might become clear.
The best way, then, to understand the significance of ritual change at Hadrumetum is to look not only at the shifts in liturgical forms, but in the range of ways
they were explained. Unfortunately, we lack texts from or about these rites, but do
have the range of sculpted stelae erected as part of the rites, many of which display images tied to the ritual practice. Such representations are not necessarily
documentary, as will be shown, but instead represent a selective and ideal description of the rites.58 And descriptions, far from objective, can reveal different
sets of conceptual priorities. Furthermore, such stelae were publicly displayed in
the sanctuary, visible to all who participated in rites there. Through their descriptions and emphases, the stelae not only represent the perspectives of individual
carvers and/or patrons on the ritual, but may also have played an active role in
shaping the perceptions of all viewers about such rites. Indeed, the representations
on the stelae were part of social knowledge, and can offer a perspective on the
importance of the different types of changes in the ritual form marked above.59
Humphrey and Laidlaw 1994, 2007; Scheid 2005, 280; Scheid 2007.
Rappaport 1999.
Rappaport 1999, 69-106.
On the meaning of ritual representations, cf. Gordon 1989.
On social knowledge and religion: Durkheim 1994, 98.
Unfortunately, the collection of stelae from the site has not been published in its entirety;
Cintas only singled out a few representative examples, and the rest, most of which were deposited in the Muse de Sousse, can no longer be firmly associated with specific archaeological
contexts. I am grateful to Wafa Messaoudi, currently in the process of cataloguing the pieces,
for discussing the 400 known stelae with me.
F
i
g
u 212
r
e conical
Matthew M. McCarty
hat seated upon a winged sphinx-throne. In his left hand, he holds a tall
sceptre; his right is outstretched, palm open. Before him, on a smaller scale,
4 stands a second figure with long hair, left hand at side, right extended outward,
: palm open. The image has little to do explicitly with sacrifice and rites at the tophet, but it does present the relationship between a deity and a worshipper, and an
S ideal about how that relationship should be perceived. The gesture, which is fret quently emphasized on contemporary stelae from Carthage by displaying a frontal
e open palm, is one of verbal prayer and benediction, part both of a set of ritual gesl tures shared across the Mediterranean and especially common in the Phoenicioe Punic world.61 The votary approaches the larger god, seated on a throne in a way
that emphasizes the unequal power relationship between the two, in prayer, and
w the god, represented as present, responds. 62 Placed over an urn offered to fulfil a
i vow, presumably after receiving the gods favourable intervention, the
t relationship between votary and deity is thus made explicit on the stele. The
h aspect of the rite emphasized is the verbal one, of prayer/vow and response,
alongside the presence of the deity. The act of sacrifice is not highlighted at all.
g
The two other stelae from Level 2 present scenes very similar to each other;
o only one is illustrated here (figure 4). In each case, a tightly-wrapped figure is
d seated at right, extending a round object forward; in the example shown, the figd ure raises its right hand in the prayer/blessing gesture. In front of the figure stands
e a tall, narrow object, topped with a ball, above which appears a crescent and disk,
s two motifs again common on most Punic stelae for their heavenly evocations. The
s seated posture suggests the divine nature of these two figures; in Phoenicio-Punic
. art, only gods (or god-like kings) are shown seated.63 The object at left has been
suggested to be a thymiaterion (incense burner), probably a correct identification
L based on comparison with surviving pieces in bronze from within the Phoenicioa Punic world as well as a relief from Adloun (in Phoenicia) showing a similar scet ne with smoke clearly rising from the burner.64 The iconography focuses again on
e the deity, and, in the case of the relief illustrated, the deitys response of benediction. It is worth noting that this was the phase in which incense began to figure
5 more prominently in the rite, as urns began to be topped with incense burners.
t
h
- 61
e
a
r 62
l
y
3
r
d
c
e
n
t
u
r
y
B
63
64
There is, unusually in Carthaginian iconography, basic consensus about this motif as one of
prayer and blessing: Hours-Miedans 1951, 32; Picard 1976, 116; Brown 1991, 134-136. See
also the article of M. Lopez in this volume (p. 47).
The representation is unlikely to be a cult statue; there is very little evidence, save exaggerated literary accounts from Greek and Latin authors, for large-scale cult statues in the Punic
world prior to the first century BCE; indeed, two of these accounts (Diodoros 20.14.6; Pliny,
Naturalis historia 36.39) focus on the bronze statue from which infants were dropped onto
the pyre in rites of sacrifice, a possibility ruled out by the regular calcification pattern of the
burned bones: Bnichou-Safar 1988, 60f. No statue bases are attested, and the norm seems to
have been, at largest, 3/4 life-size statuary, and mostly much smaller, if the naiskos (?) bases
from the sanctuary at Kerkouane are any guide: Fantar 1986, 154.
E.g., Gubel 1997, 60; Gubel 2002, 74f. no. 75.
For similar thymiateria from sites with a Phoenicio-Punic presence, Fontan et al. 2007, 164f.;
Gubel 2002, 122f. no. 119, shows a similar thymiaterion with wisps of smoke rising from it.
213
Figure 3 (left). Baal of Sousse, limestone stele, late 5th-early 3rd century BCE (Cintas 1948, fig.
49).
Figure 4 (right). Limestone stele with goddess, late 5th-early 3rd century BCE.
The stelae from Levels 3-4 display a range of abstract motifs, many of which
particularly in Level 3 can also be found in contemporary Carthage; not, perhaps, surprising, given that Hadrumetum seems to have been closely allied with
Carthage during this period and served in 203 BCE as Hannibals winter quarters.65 These include the so-called sign of Tanit, the open palm, series of baetyls,
the caduceus, and the so-called sign of the bottle.66 What each of these might
represent is debated; given that they appear to be arbitrary symbols, their precise
associations may be impossible to reconstruct. Nevertheless, all seem to bear
some relation to the gods the sign of Tanit appears almost exclusively in sanctuary contexts or at thresholds, perhaps as a kind of apotropaic symbol, suggesting
its divine power.67 The baetyls have long been recognized as serving as aniconic
testaments to the presence of a deity.68 While a range of interpretations have been
put forward for the caduceus, the most convincing is that these objects represent
some kind of cult paraphernalia, probably a sort of standard that would again
65
66
67
68
214
Matthew M. McCarty
highlight the divine nature of the god or the objects associated with the caducei,
frequently baetyls.69 After all, these caducei frequently appear inserted into triangular bases. Finally, the so-called bottles have been suggested to represent the
urns carrying sacrificial remains given to the deity, but are more probably other
types of aniconic indicators of the god, for they are frequently shown placed on
pedestals and sometimes also, like the baetyls, flanked by caducei. 70 Thus, as with
the stelae from Level 2, the primary focus of the iconography is indicating the
deity in some way, even if the nature of the representation has changed. The epigraphy, more frequently incorporated as part of the stele-monument in this phase,
likewise frequently focuses on the gods, beginning with an invocation to them,
stressing the role of the presumably verbal ndr, and often ending with the formula
of prayer and blessing, again drawing attention to the gods agency and the verbal
nature of the relationship.
Figure 5 (left). Limestone stele with sacrificial scene, first century CE.
Figure 6 (right). Limestone stele with sacrificial scene, first century CE.
69
70
Brown 1991, 131-134, with earlier bibliography, suggesting that the caduceus is the GrecoRoman symbol appropriated for Carthaginian art; Lipinski 1995 makes a more convincing case that it is a cultic standard, based on the bases it appears with and the ribbons often shown
streaming from it.
Cintas 1948, 60-62; Brown 1991, 138-141.
215
Yet in Level 5, an important group of stelae demonstrate a very different iconography, and one otherwise unattested at the site or at other tophets prior to
the Roman Imperial period.71 These all show narrative scenes of sacrifice. On one
stele (figure 5), a group of three tunicate men appears at right, carrying a lamb
between them. At left, the legs of another figure are visible below a break: this is
probably a scene of bringing the animal to sacrifice. And it is notable that the victim is emphasized in this manner, given that no reference is made to the animal
victims in the previous levels. Moreover, unlike the earlier inscriptions, the act is
cast as communal: three people are involved, not just the single dedicant named in
the Punic epigraphs. Two other stelae show similar scenes of groups of people
moving towards altars.72
Three additional stelae represent other moments of a sacrifice underway. The
first (figure 6), heavily weathered and partially illegible, shows a square altar at
centre, with a large figure at left reaching over it to make a preparatory offering;
to the right stand two smaller figures, reaching upward and holding other offerings. The second (figure 7), fragmentary, shows an altar at right, draped with a
garland. To the left, a figure stands in profile, wearing a short mantle that falls in
thick folds down his side; he extends his right hand over the altar, perhaps to drop
incense onto it. Behind him, positioned frontally, stands a smaller figure in a
short, belted tunic. In his left hand, he supports a tray; in his right, he holds a
bucket. The third stele (figure 8), presents a similar scene. Atop a raised platform
stands a flaming altar; to the left, a striding man, clad in toga pulled over his head
as a veil, drops incense onto the altar with his right hand while carrying a small
incense box in his left. On the other side of the altar, a smaller figure approaches,
tray held over left shoulder, bucket in right hand.
Figure 7. Limestone stele with sacrificial scene, first century CE. Sousse (Cintas 1948, fig. 136).
71
72
216
Matthew M. McCarty
Figure 8. Limestone stele with sacrificial scene, first century CE (Cintas 1948, fig. 134).
Finally, the stelae from Level 6 demonstrate a final, major shift in iconography: smaller, rougher, they show only animals, frequently set in an architectural
frame, as in figure 9. Here, a ram is represented between two columns that support
a pediment.
217
The stelae from the first century BCE/first century CE Level 5 mark a radical
departure from those of the previous levels. Indeed, the visual focus is shifted
from aniconic displays of the deitys presence and presumably through his presence, his efficacy at accomplishing the favour sought in the vow to the ritual act
of sacrifice. If marking and prioritizing any ritual act as metonymous for the set of
rites practiced, the earlier stelae focused on verbal communication with the deity
rather than sacrifice. In inscribed stelae, both the ndr, always mentioned, and the
frequent use of the formula he heard his voice, he blessed him on stelae from
Hadrumetum and Carthage draw attention to this verbal rapport between votary
and god.73 Although much earlier, the Level 2 Baal stele demonstrates a very
different relationship between god and votary than the stelae of Level 5, one that
is both epiphanic and still primarily verbal, as both figures raise their right hands
in gestures of prayer and speech. Similarly, three mid-second century BCE stelae
from the Salammb tophet (out of thousands of contemporary examples), although they show scenes of figures at an altar, represent the sacrifice as a fait accompli, the sacrificial bulls head burning on the altar, the sacrificant raising his
hand in such a gesture of verbal prayer.74 The stelae of Level 5 thus demonstrate a
new emphasis on the act of sacrifice, rather than the god or prayer. This change in
priority suggests a shift in how the rites were viewed and which elements came to
be prioritized as metonymous for the whole range of activities conducted, even if,
at this point, the basic archetype of the rites themselves remained relatively unchanged.
Of course, the one major archetypal change attested in Level 5, the shift from
victim as variable to victim as canonical, may receive a certain amount of recognition on the stelae. Certainly in the procession to sacrifice stele, the victim is
highlighted in a new way. Yet this is only one of the many stelae from the level; it
was only in Level 6 that animal victims came to prominence in the representations
on the site.
It has been suggested that such visual changes mark no more than the appropriation of images from Rome or some other centre like Carthage a new, and
potentially meaningless, visual veneer for what were old and unchanging rituals. 75
The basic theme preliminary rites of libation or incense-offering at an altar is
familiar from numerous reliefs and altars of the Imperial period in Italy, especially
the elaborate state monuments.76 Yet the stelae from Hadrumetum demonstrate no
mere copying of a model, but do, in fact, suggest active choice, a re-conceptualization and new description of the series of rites performed by the participants in
these rituals at Hadrumetum.
73
74
75
76
218
Matthew M. McCarty
Direct iconographic borrowings for religious monuments are, of course, attested in North Africa and elsewhere. Both the relief of Tellus, now in the Louvre, and the relief showing Mars, Venus and Divus Iulius in Algiers, both found in
Carthage, reproduce images from the Urbs in the former, the famous figure from
the Ara Pacis, in the latter, most probably the cult statues from the Temple of
Mars Ultor.77 Likewise, the famous carved imperial cult altar from Carthage, probably dedicated by P. Hedulus in the Julio-Claudian period, repeats a series of
images from Rome and reproduced in other provincial capitals like Augusta Emerita: the Lupercal, Aeneas fleeing Troy, Roma seated atop arms, and a scene of
sacrifice.78 The sacrificial scene on the altar is particularly close to Italian examples (figure 10): it shows a togate sacrificant, head veiled, holding an incense
box in his left hand and dropping incense onto a flaming portable altar. To his left
stands an assistant (victimarius), clad in a limus (an apron-like garment), mallet
over his shoulder, holding a sacrificial bull. Two camilli (temple attendants) stand
on the other side of the altar with sacrificial paraphernalia, while a tibicen (flutist)
plays the double-flute behind the altar.
The connection between such an image and the sacrificial scenes from imperial monuments in Rome and central Italy is striking, particularly when set alongside a scene such as that from the imperial cult shrine on the forum in Pompeii
77
78
The bibliography on both monuments is vast; on the Carthage Tellus, Simon 1967, 323f.;
Zanker 1988, 314. On the Algiers relief: Doublet 1890, 84f.; Zanker 1970, 20; Krause 1979.
Altar and sanctuary: Cagnat 1913; Poinssot 1929; Seba 2005, 271-274. Augusta Emerita
Aeneas group and references to other attestations: de la Barrera and Trillmich 1996. On the
spread of these images: Boschung 2003.
219
(figure 11): the focus on the togate sacrificant making preparatory offerings, the
reference to the coming blood sacrifice through the presence of bull and victimarius, the associated attendants holding a range of paraphernalia, and the presence of a musician to make sure that no untoward sounds disturbed the rite. The
most notable difference is the lack of architecture background on the Carthage
altar, possibly attributable to its slightly earlier date. Such a close appropriation of
Italian models, however, is not surprising perhaps in the provincial capital, carved
from marble imported from Italy and commissioned to adorn a temple to the imperial cult set up by a man with important business connections in Italy.79 Yet
such close and clear appropriations of sacrificial scenes are otherwise rare in the
region; the only other such scenes attested are a fragmentary sacrificial procession
involving a pig from an elite tomb at Utica, and another fragmentary procession
akin to that on the Ara Pacis found in Carthage.80 Indeed, such close examples
simply serve to throw into contrast the difference between these scenes and the
stelae from Hadrumetum, which do not demonstrate such a close appropriation of
Italian or cosmopolitan models while still representing a similar set of concerns to
such pieces.
Figure 11. Imperial cult altar from Pompeii. First century CE.
79
80
The only other references to Publius Hedulus are a series of brick stamps found in and around
Carthage with his name: CIL VIII 22632.72. On two bricks, his stamp appears alongside that
of C. Iulius Antimachus (CIL VIII 22632.65), whose stamp also appears on bricks distributed
in Italy (CIL XV 1202). The greater spread of Antimachus stamps suggests he was engaged
more heavily with distribution, and thus the officinator, but also indicates that Hedulus had
business connections in Italy. Most recently on double-stamped bricks: Steinby 1993, 139.
Gauckler et al. 1910, 72, cat. no. 1173; Lund 1998; Schrner 2009.
220
Matthew M. McCarty
The image of the three men carrying a ram to sacrifice (figure 5), for example,
is unprecedented in Roman art. The moment chosen for representation is distinctly
different: these are not preparatory rites for the blood sacrifice at an altar, nor a
formal procession, but instead a group shown presenting their offering. The emphases are, however, commensurate with those on the Carthage altar or Italian
monuments: the victim is highlighted, the communal side of the rite is accentuated
in contrast to the earlier stelae from Carthage and the Baal of Sousse stele. The
two well-preserved stelae showing sacrifice at an altar similarly demonstrate a
similar focus on certain elements of the rite that goes beyond the mere appropriation of visual forms. The moment chosen for representation is the same as that
selected for both state monuments and a range of personal votive altars in Italy
and elsewhere, the preparatory offerings either before (or instead of) blood sacrifice.81 The way that ritual was described and represented at Hadrumetum thus
changed in a manner that used its own, local terms to highlight moments and aspects of the ritual favoured in representations in many parts of the empire.
Finally, Level 6 demonstrates another change in visual priorities and ways of
describing the rituals that took place at Hadrumetum. The lack of urns deposited
in this level has often been taken to mean that sacrifice ceased at the siteusually
itself taken as evidence of the impoverishment of the worship community and its
decline.82 The latter certainly is not true; not only was Hadrumetum experiencing
the same second-century economic boom as the rest of Africa, attested by the
building of major public amenities and the rich decor of the few excavated residences there, but the repaving of the sanctuary at the beginning of Level 6 bespeaks a major, costly project and the vitality of the community.83 Lack of urns is
not evidence for lack of sacrifice, but as discussed above, only for evidence of a
change in the ritual archetype: collection of remains and deposition was no longer
practiced. Comparison with other tophet sites still active in the first/second century CE suggests an alternative explanation: that the rites had shifted from private
sacrifice in fulfilment of a vow to communal sacrifice around an altar, possibly
followed by a sacrificial meal.
Such a shift is visible at at least two other sites, and probably more, if not as
explicitly. At Henchir el-Hami, for example, the tophet-style deposition ceased in
the late second century CE, in favour of a more public rite celebrated around a
monumental built altar which involved the use of perfumed oils; the number of
81
82
83
221
unguentaria discovered at its base probably bespeaks the wide participation in this
rite, rather than the more individual modality of offering represented by the urn
deposits and the offerings made around individual stelae.84 Similarly, a range of
cookware and fineware fragments, almost entirely lacking in the tophet phase of
the site, were discovered, suggesting communal dining.85 Likewise, at Thinissut,
the building of a temple in the first century CE limited access to the hillside upon
which urns had previously been deposited.86 Although urn deposition may have
continued a bit longer, and stelae were set up (perhaps curated rather than newlycarved) in a side room of the sanctuary, a monumental altar surrounded by a porticoed courtyard for participants demonstrates the new focus of the sanctuary. 87
Likewise, at a slightly later date, a dump room was created at the rear of the
sanctuary, accessible through a small window, in which bones, ash, and dining
wares were thrown: sacrificial remains were not being deposited, but dined upon,
by a community.88 Related changes may likewise be observable at other sites,
where altar and temple complexes replaced or were added to stele-fields.89 While
the limited excavation at the Hadrumetum tophet does not allow the confirmation
of this hypothesis, it seems probable that rites there underwent a similar change;
deposition ceased, but animal sacrifice, emphasized by the subject matter of the
stelae, did not. Instead, this aspect of the ritual act the offering of an animal victim was highlighted in a new way on the Level 6 stelae.
The representations on the stelae, themselves indicative of how the ritual
might be thought about and whose public presence was capable of shaping social
knowledge about ritual practice, thus demonstrate a series of connections with the
changes in ritual form observed at the site. The images used on the stelae were
actively chosen by the carvers and/or patrons, and can demonstrate how, within a
social setting, such actions may have been perceived.
Not all changes in liturgical form can be linked to corresponding changes in
representations on the stelae; a number of variations, which may have been the
result of the specific circumstances in which the ritual was enacted, are not often
marked by the stelae. Similarly, the disappearance of jewellery and the addition of
coin-offerings, for example, do not change in any noticeable way the manner in
which the whole collection of gestures is visually described. This is not to deny
potential significances of these changes, but only to emphasize that these are not
reconstructable.
Far more marked is the way that changes to the archetypal form of the liturgy
(ndr-sacrifice-deposition-stele erection-further rites) are manifest in the stele representations. There is a basic re-orientation visible on the stelae in Level 5 from
an idea of communication with the gods that is based around a verbal exchange to
84
85
86
87
88
89
Ferjaoui 2007.
Campisi 2007a; Campisi 2007b.
Merlin 1910.
Merlin 1910, 29.
Merlin 1910, 14.
E.g., at Ammaedara and Thugga.
222
Matthew M. McCarty
one based around sacrifice at an altar, from individual rites to communal ceremonies. Such a change may be visible in Level 6, as the rites probably did move to a
communal altar sacrifice in a shift commensurate with that observable at other
related sites. Such a changed emphasis led to a change in how the rites were conducted; the deposition portion of the liturgy was dropped, perhaps in favour of
communal dining on animal victims. Likewise, as the variability of victims changed to become canonical, this shift was referenced on at least one stele from Level
5, and certainly in Level 6, where the victims become the sole focus of the stele
representations. The changes in conception and description of the rites were not
necessarily contemporary with such changes in archetypal form, but they do, at
least to an extent, map onto each other.
3 CONCLUSIONS
The archaeological and visual evidence from Hadrumetum offers a rare look at the
range of connections between ritual forms and ways of expressing those forms: a
chance to see how the act of describing ritual via visual representation changed
alongside the actual transformations in the rites, and to hint at the potential and at
least partially-recoverable bond between liturgical form and conceptualization.
Within a set of rites that involved child or animal sacrifice and culminated in
the erection of a carved stone stele, a range of changes in ritual form can be seen
at Hadrumetum, as can the ubiquitous variations that may have reflected the specific circumstances of an offering. Jewellery might no longer be offered, coins
might be added as a post-sacrifice offering, all the while maintaining a similar
ritual archetype at least for those actions which have left traces in the archaeological record. Yet not all of these changes carried the same significative weight;
those which left the basic ritual archetype unaltered may have added nuance or
confirmed the meanings of other actions, but are not easily relatable to the broad
changes in representation. As a result, the basic visual means of commenting on
and representing the ritual stayed largely the same from Level 2 to Level 4 on the
site (late fifth century BCE-first century BCE), emphasizing the presence of the
deity and the verbal prayers of the dedicant, despite some changes in the ritual
form. Level 5 (first century BCE-first century CE) saw little change in the archetypal rite, although coins were added as part of the set of offerings made in the
tophet. The one major shift was the change in victims offered: no longer was this
a variable feature of the rite, but became a canonical one. Such a new focus on the
victims appears in representations of the ritual.
This basic picture confirms that in antiquity, changes in ways of thinking
about ritual were not always immediately paralleled by changes in ritual form, a
relationship already well-documented for other cultures by Blochs work among
others: not an entirely surprising conclusion, nor a novel one. Yet it also shows
that conceptions and changes in ritual archetype were not completely divorced: in
Level 5, ways of perceiving the rites were reconfigured in a manner commensurate with the change in victims, advertising and encouraging the new use of exclu-
223
sively animal sacrifice. Likewise, although it took time, the new focus on sacrifice
as the key moment by which to metonymously represent both the ritual act and
perhaps more broadly the relationship between votary and god found in the stelae
of Level 5 eventually found expression in the reorganization of sacrificial rites in
Level 6 (after first century CE), which probably became communal and based
around an altar. Such connections at least raise the possibility that, if done carefully, analysis of changed ritual forms in the ancient world on the archetypal level
could reveal much about changed patterns of cognition and culture.
This is, in some ways, a minimalist conclusion in demonstrating only a weak
link, and a link not necessarily in any signified content of ritual action, but in the
ways of treating the ritual itself. But the material from Hadrumetum does nevertheless point to the existence of such a connection and to the shared, social element of the knowledge related to the sacrificial liturgy. It may thus be possible to
move beyond the realm of entirely personally-constructed and projected meanings
of liturgical acts and towards broader frameworks of social knowledge in ritual.
It must also be noted that the shifts observed at Hadrumetum point to a range
of social changes in the worship community at the site. Indeed, the movement
away from individually-offered deposits towards the communal sacrifice scenes
on the stelae of Level 5, and what may have been a more group-based ritual
around an altar by Level 6, bespeaks a re-structuring of the cult community.90
Whether this was a result which the ritual worked to create, or the ritual a manifestation of this external social re-ordering, is impossible to say. But such social
change is certainly one aspect of a broad range of cultural changes that occurred
in the region over the course of the first century BCE-first century CE in North
Africa revealed in the dynamics of ritual change.
Finally, a word about the cultural weight of such changes, since the cultural
weight of changes in ritual archetypes has been established. Even within a rite
whose final outcome erection of a stele was part of a long tradition specific to
parts of North Africa, the basic set of changes observed, both in terms of liturgical
forms and in terms of ways of representing them, demonstrate incorporation into
the wider sphere of the Roman Empire, and close ties to Italy in particular. Such
communal sacrifice around an altar that can be suggested by the remains in Level
6 was a ritual form not previously attested in this cult, yet by the second century
CE, this manner of relating to the gods had been adopted in several stele sanctuaries: liturgical acts known and used in other cults in Africa and across the Mediterranean were transferred for use in a new context. Such transfers even included
details of dress: on one of the Level 5 sacrifice scenes, the sacrificant is shown
with his head veiled in a manner typical of Roman rites, appropriating an idea
about the proper costume for such an act, probably from public, civic cults. Simi90
The argument in Schrner 2007b, that these rites were based around a family group, may well
apply to the earlier material and some later sites; there is little, however, to suggest that many
of the later tophets focused their expressions of communality around family units; indeed, the
sacrificial scenes in Level 5 at Hadrumetum suggest otherwise: they do not show the neat and
easily-recognizable family groups of Classical Greek votive reliefs, for example.
224
Matthew M. McCarty
larly, although the images used for representing sacrifice did not directly emulate
those of Italy, the visual means of describing and making sense of the rites was
similar to the representations from Italy. Indeed, it may even be possible to tentatively suggest the development of an empire-wide mode of discourse, although
with strongly local aspects, in the treatment of rites that emphasized many of the
same aspects and features of rites; Schrner suggests a similar local shift in the
representation of sacrifice in Asia Minor.91 Such changes may suggest that incorporation into the Roman Empire did affect the basic formulations used to relate to
the gods and think about ritual on a local level in Africa and elsewhere in the empire, even within such continued cults.
Nevertheless, it is also striking that stelae did continue to be erected in Level
6, even though their basic raison dtreto mark a sacrificial depositwas no
longer present. Despite the changes in the archetypal ritual practice at the site, the
lasting markers that guaranteed the accomplishment of the vow continued to be
erected. This may have occurred for any number of reasons the prestige value
associated with commissioning a stone monument (although this seems less likely
at Hadrumetum than elsewhere in the Imperial period, since the dedicants are
never named or figured), the importance of leaving a lasting testament to the fulfilment of a vow, the weight of tradition itself. At other sites, a similar phenomenon is attested although urn deposition seems to have ceased in the first or second century CE in the sanctuaries at Thinissut, Henchir el-Hami, and Ammaedara,
for example, the findspots of stelae seem to suggest that they were still curated
and displayed, if not commissioned, well into the third century.92 Even if, at all of
these sites, the archetypal rites fundamentally changed, and roughly similar shifts
to those at Hadrumetum in ways of representing ritual action can be observed, the
stelae still had a role to play in the cult, clothing the new archetypes with lasting
monuments that set these new rites in line as part of a recognizably old practice.
The discourse surrounding ritual, evident in representations, may have changed
dramatically over the course of the Imperial period and driven later changes in the
archetypal form of the liturgies, but maintaining an idea of tradition remained a
key aspect of both the acts and the descriptions of them.
Schrner 2006, 2009, although pushing for more direct appropriation of imperial images in
Africa.
The latest known newly-carved stele was erected in 323 CE, datable by the consular year in
the inscription: Beschaouch 1968.
225
Ben Abid Sadallah, L. (2003) Le culte de Baal Hammon dans la province dAfrique: entre tradition et romanisation. PhD dissertation, University of Tunis.
Bnabou, M. (1976) La rsistance africaine la romanisation, Paris.
Bendlin, A. (1997) Peripheral Centres Central Peripheries: Religious Communication in the
Roman Empire, in Cancik and Rpke (eds.) 1997, 35-68.
(2001) Rituals or Beliefs? Religion and the Religious Life of Rome, Studia Classica Israelica 20, 191-208.
Bnichou-Safar, H. (1988) Sur lincinration des enfants aux tophets de Carthage et de Sousse,
Revue de lhistorie des religions 205, 57-68.
(1989) Les stles dit de Sainte-Marie Carthage, in H. Devijver and E. Lipinski (eds.),
Punic Wars, Leuven, 353-364.
(2004) Le tophet de Salammb Carthage: essai de reconstitution, Rome.
Berthier, A. and R. Charlier (1955) Le sanctuaire punique del-Hofra Constantine, Paris.
Beschaouch, A. (1968) Une stle consacre Saturne le 8 novembre 323, Bulletin archologique
du Comit des travaux historiques et scientifiques 4, 253-268.
Bloch, M. (1986) From Blessing to Violence: History and Ideology in the Circumcision Ritual of
the Merina of Madagascar, Cambridge.
Boschung, D. (2003) Die stadtrmischen Monumente des Augustus und ihre Rezeption im Reich,
in P. Noelke et al. (eds.), Romanisation und Resistenz in Plastik, Architektur und Inschriften
der Provinzen des Imperium Romanum, Mainz, 1-12.
Bremmer, J. (1998) Religion, Ritual, and Opposition, in F. Graf (ed.), Ansichten griechischer
Rituale: Geburtstags-Symposium fr Walter Burkert, Stuttgart/Leipzig, 9-32.
Brown, S. (1991) Late Carthaginian Child Sacrifice and Sacrificial Monuments in their Mediterranean Context, Sheffield.
Cadotte, A. (2007) La romanisation des dieux: linterpretatio romana en Afrique du Nord sous le
Haut-Empire, Leiden.
Cagnat, R. (1913) Un temple de la Gens Augusta Carthage, Comptes rendus de lAcademie des
Inscriptions et Belles Lettres, 680-686.
Campisi, L. (2007a) Ceramica sigillata del santuario, in Ferjaoui 2007, 325-239.
(2007b) Ceramica sigillata dal tempio, in Ferjaoui 2007, 363-372.
Cancik, H. and J. Rpke (eds.) (1997) Rmische Reichsreligion und Provincialreligion, Tbingen.
Candellieri, V. (1989) Monetazione punica: il tipo testa di divinit/cavallino corrente od impennato, Turin.
Chabot, J.-B. (1916) Les inscriptions puniques de la collection Marchant, Comptes rendus de
lAcademie des Inscriptions et Belles Lettres, 17-34.
Chaniotis, A. (2005) Ritual Dynamics in the Eastern Mediterranean: Case Studies in Ancient
Greece and Asia Minor, in W. V. Harris (ed.), Rethinking the Mediterranean, Oxford, 141166.
CIL: Corpus Inscriptionum Latinarum, Berlin 1853.
Cintas, P. (1948) Le sanctuaire punique de Sousse, Algiers.
de la Barrera, J.L. and W. Trillmich (1996) Eine Wiederholung der Aeneas-Gruppe vom Forum
Augustum samt ihrer Inschrift in Mrida (Spanien), Rheinisches Museum 103, 119-138.
Derks, T. (1998) Gods, Temples, and Ritual Practices: the Transformation of Reigious Ideas and
Values in Roman Gaul, Amsterdam.
Doublet, G. (1890) Muse dAlger, Paris.
Durkheim, E. (1994) +++, in W. S. F. Pickering (ed.), Durkheim on Religion, Atlanta, ++.
Fantar, M. H. (1985) Kerkouane, cit punique du Cap Bon (Tunisie) II: architecture domestique,
Tunis.
(1986) Kerkouane, cit punique du Cap Bon (Tunisie) III: sanctuaires et cultes socit
conomie, Tunis.
(1993) Formules propitiatoires sur des stles puniques et nopuniques, in J. Quaegebeur (ed.),
Ritual and Sacrifice in the Ancient Near East, Leuven, 125-134.
226
Matthew M. McCarty
(1995) Stles pigraphes du Tophet de Sousse, Revue des tudes Phniciens-Puniques et des
Antiquits Libyques 9, 25-48.
Ferjaoui, A. (2007) Le sanctuaire de Henchir el-Hami: de Baal Hammon au Saturne africain, 1er
s. av. J.-C. - IVe s. ap. J.-C., Tunis.
Fontan, E. et al. (eds.) (2007) La Mditerrane des Phniciens: de Tyr Carthage, Paris.
Foucher, L. (1964) Hadrumetum, Paris.
Fulford, M. (2001) Links with the Past: Pervasive Ritual Behaviour in Roman Britain, Brittania
32, 200-218.
Gielly, P. (1927) Quelques lueurs nouvelles sur lemplacement de la Carthage punique et de ses
ports, Casablanca.
Gordon, R. (1989) The Moment of Death: Art and the Ritual of Greek Sacrifice, in I. Lavin et al.
(eds.), Acts of the XXVIth Congress for the History of Art, Philadelphia, 567-573.
Gubel, E. (1997) Cinq bulles indites des archives tyriennes de lpoque achmnide, Semitica 47,
53-64.
(ed.) (2002) Art phnicien: la sculpture de tradition phnicienne, Paris.
Haynes, I. P. (1993) The Romanization of Religion in the Auxilia of the Roman Imperial Army
from Augustus to Septimius Severus, Brittania 24, 141-157.
(1997) Religion in the Roman Army: Unifying Aspects and Regional Trends, in Cancik and
Rpke (eds.) 1997, 113-126.
Hours-Midan, M. (1951) Les reprsentations figures sur les stles de Carthage, Cahiers de Byrsa
1, 15-160.
Humphrey, C. and J. Laidlaw (1994) The Archetypal Actions of Ritual: A Theory of Ritual Illustrated by the Jain Rite of Worship, Oxford.
(2007) Sacrifice and Ritualization, in E. Kyriakidis (ed.), The Archaeology of Ritual, Los
Angeles, 255-276.
Icard, F. (1922) Dcouverte de larea de la sanctuaire de Tanit Carthage, Revue Tunisienne, 195205.
Khanoussi, M. and L. Maurin (eds.) (2000) Dougga, fragments dhistoire: choix dinscriptions
latines editees, traduites et commentees (1er-IVe siecles), Bordeaux/Tunis.
Krandel-Ben Younes, A. (2002) La prsence punique en pays numide, Tunis.
Krause, T. (1979) Zum Mars Ultor-Relief in Algier, in G. Kopcke and M. B. Moore (eds.), Studies
in Classical Art and Archaeology: A Tribute to Peter Heinrich von Blanckenhagen, Locust
Valley, 239-245.
Lapeyre, G. S. (1935) Fouilles rcentes Carthage, Comptes rendus de lAcademie des Inscriptions et Belles Lettres, 81-87.
Le Glay, M. (1961) Saturne africain: monuments I, Paris.
(1966a) Saturne africain: monuments II, Paris.
(1966b) Saturne africain: histoire, Paris.
Lipinski, E. (1995) Le caduce, in M. F. M. Ghaki (ed.), Actes du III congrs international des
tudes phniciennes et puniques, Tunis, 203-209.
McCarty, M. M. (forthcoming) Soldiers and Stelae: Votive Cult and the Roman Army in North
Africa, in M. della Riva (ed.), Incontri tra Culture nel Mediterraneo Antico, Rome.
Mendleson, C. (2003) Catalogue of Punic Stelae in the British Museum, London.
Merlin, A. (1910) Le sanctuaire de Baal et de Tanit prs de Siagu, Paris.
Mettinger, T. N. D. (1995) No Graven Image? Israelite Aniconism in its Ancient Near Eastern
Context, Stockholm.
Millett, M. (1995) Roman Britain, London.
Mosca, P. (1975) Child Sacrifice in Caananite and Israelite Religion: A Study in Mulk and MLK.
PhD dissertation, Harvard University.
Moscati, S. (1981) Documenti inediti sugli scavi di Nora, Rendiconti della Accademia dei Lincei
36, 157-161.
(1986) Le stele di Sulcis : caratteri e confronti, Rome.
227
228
Matthew M. McCarty
PHOTO CREDITS
Figure 2:
Figure 3:
Figure 4:
Figure 5:
Figure 6:
Figure 7:
Figure 8:
Figure 9:
Figure 10:
Figure 11: