• 1 Their flurry of activity also responded to the competing interests of those who were pillaging anc (...)
• 2 On the di Manes, see now King 2020.
• 3 Carton 1909, p. 43. For a comparatively recent synthesis of the employment of cupula grave monumen (...)

1Over a century ago, fascination with the discoveries at the Cemetery of the Officiales at Carthage in the 1880s and 1890s inspired amateur and professional archaeologists alike to fan out across Tunisia in search of ancient necropoleis1. Their work soon established that cremation had been the predominant mode of burial in the province of Africa Proconsularis between the mid- to late first century and first part of the third century C.E., and that the graves generally attested similar sequences of practices to those at Rome, including the dedication of the deceased to the di Manes on epitaphs2. By 1909, L. Carton was able to distill these pagan cemeteries to three core traits: (I) the prevalence of block-like funerary monuments (“caissons”), the most widespread and “African” of which was the cupula monument, which took the form of a horizontally-laid half-cylinder; (II) the use of tiles or transport amphoras as sarcophagi; and (III) typical grave assemblages of cinerary urns, lamps, and coins3. He recognized some regional differences, such as the preference for cippi-altars over cupulae at the Cemetery of the Officiales at Carthage and, among grave goods, a greater propensity for terracotta figurines in the Sahel than in the High Tell. He also correlated some patterns to cultural influences, for example linking the equipping of tombs with libation tubes to the larger, more “Romanized” urban centres.

• 4 For example, at Sala (Boube 1999), Tipasa, Cherchell, Sétif (for the latter three sites, see bibli (...)
• 5 Plin., nat., 5, 3, 24 (Teubner) : Libyphoenices uocantur qui Byzacium incolunt.
• 6 Moore 2007, p. 77 ; Stirling 2007, p. 116.
• 7 Detailed publication of recent fieldwork at Dougga (Dougga 2020) includes a fine-grained analysis (...)

3While there have been important studies of Roman-period cemeteries in several parts of North Africa4, a demonstration of the diversity of local practices is best achieved by examining a geographically-restricted area within which the necropoleis of several towns have been sufficiently documented to enable comparison. A prime zone for this purpose is along the east coast of Africa Proconsularis, defined by a narrow rectangle of approximately 250 km ­north-south, stretching from Carthage to Thaenae (just south of present-day Sfax), and 90 km east-west, between Raqqada (near Kairouan) and the coast (fig. 1). This selection limits the number of major pre-Roman cultural influences that might underlie certain behaviours, to the local indigenous, the Punic Carthaginians, and the Libyphoenicians who inhabited the stretch between Hadrumetum and Thaenae5, which was known at various times in antiquity as Byzacium or Byzacena, and today roughly corresponds to the Tunisian Sahel. Another reason to choose this zone is the contiguity of building materials in cemeteries: while monolithic stone grave monuments typified the High Tell and the western parts of the province, in the selected area the prevalent material was masonry-and-rubble6, which allowed more options for the integration of ritual stages and provisions with (or encased within) the monument, as will be discussed. Sites of the High Tell (including Thugga, Bulla Regia, and Althiburos), constitute a separate grouping by their different cultural influences (particularly Numidian) and their geology, which enabled solid stone monuments7.

• 8 Mahjoubi 1970, p. 10 (Raqqada) ; Ben Abed, Griesheimer 2001, p. 587 (Pupput) ; Larminat 2012, p. 2 (...)
• 9 The funerary rituals accompanying both cremation and inhumation accomplished the needed ritual goa (...)

4Even though inhumation continued to be practiced while cremation was at its height, it is worth considering cremation burials separately. The latter involved not only an additional step (the ritualized burning of the body), but also different attitudes to certain ritual acts, indicated by the number of objects in the grave and the provision of structures for post-funerary offerings – cremation graves generally seem to have had more of both8. In addition, from one site to the next within this study, the diversity of choices that were made for the structure and provisioning of cremation graves seems far greater than those for inhumations. Clearly the cremated dead were perceived differently from the inhumed dead.9

• 10 For considerations of the relationships between landscape and the mortuary practices of ancient No (...)

5While scholars tend to approach ancient polytheistic religion from the perspective of its roles in maintaining human-divine relations and reinforcing social norms within the community, the cemeteries to be examined here add another component for consideration, namely how funerary practices were partly shaped by local circumstances – not simply the region or general environment10, but aspects that were specific to a town or communit y, such as their demographics, the availability of land, and local resources and infrastructure. Such factors seem to have led each community – and, sometimes, each sub-community that was served by a separate cemetery – to develop its own distinct practices of negotiating the process of transitioning and interacting with the dead. It is worth noting that, while there were patterns distinctive to each site, there was always room for individual agency, from choosing which items to use at any ritual stage to the distinctive architectural and ornamental designs of monuments. However, the focus of this paper is on the impact of the locally-conditioned environment upon collective or group behaviour at each cemetery. Understanding the local milieu of social and especially funerary behaviours is the first step in contextualizing how individuals positioned themselves in the performance of those religious rites.

• 11 For example, at Hadrumetum and Thaenae, there were separate monumental and humble cemeteries. The (...)

6Logically, some variation in mortuary practices is to be expected due to the vastly differing nature of the towns, in terms of their location, size, and the population that they served. Further complicating the situation is the fact that many towns had more than one cemetery. While many burial zones served the general populace, certain others may have been chosen for their proximity or because they catered to a specific subset of the population based on profession, socio-economic status, or another categorization11. Within each cemetery, evidence for some form of organization, the level of crowding, and the presence or absence of funerary monuments is noteworthy, because they relate to the availability and administration of land for burial purposes, as well as the ability (and the degree of social pressure) to erect a visible memorial. Related to the latter point, the concern for protecting a grave is also relevant, since poorly-marked graves were likely to be disrupted by subsequent burials.

• 12 Signalled by Lavigerie (1881, p. 29-41) ; the main excavation reports, as opposed to epigraphy-foc (...)
• 13 Delattre 1888, p. 152. Lassère (1973, p. 54) presented evidence that both sectors were operational (...)
• 14 More than 600 of the epitaphs are published in CIL VIII, suppl. 1, pp. 1301-1338 (including discus (...)
• 15 A.-L. Delattre, cited by Lavigerie (1881, p. 34).
• 16 Lavigerie 1881, p. 37-39 ; Audollent 1901, p. 432. As Carlsen (2020b, p. 23) noted, only two inscr (...)

7Among the cemeteries of Carthage with cremation burials, the most extensively explored was that of the Officiales, located just outside the walls of the metropolis of Carthage, near the amphitheatre and straddling the route that connected to the decumanus maximus12. It received cremations between the first and early third centuries. The full extent of the sector on the east side of the road, known as Bir es Zitoun, is not recorded; it contained graves dating back to potentially the early years of the Roman colony. Its counterpart across the road, Bir el Jebbana, had enclosure walls defining a space of c. 1000 m² and contained graves of Imperial date13 (fig. 2). The abundance of surviving epigraphy – circa 1000 epitaphs – indicates that both sectors mainly served the same, specialized group of adults and children: Imperial slaves and freedmen who were attached to the Imperial administration at Carthage, and, to a lesser degree, soldiers and veterans, especially those who had served in the urban cohort14. The plan for each sector was generally open, although the graves were apparently grouped by family and by profession15. This combination of status and organization may signal the existence of funerary collegia that ensured the proper cremation and burial of their members16, even if the formulaic funerary epitaphs normally credited family members for doing so. Grave markers ranged from modest stelae to mausolea with stucco relief, although the cippus-altar was the preeminent monument. There was no regularity to the alignment or orientation of the graves. Over time, both sectors received so many monuments that many abutted one another, thereby creating obstacles for visitors and, presumably, for the performance of rites at certain graves.

• 17 Norman, Haeckl 1993, p. 244 ; Clerkin 2013, p. 14 n. 35.
• 18 Norman 2002, p. 305. The discovery and description of the sculptures were presented by Annabi (198 (...)
• 19 Annabi 1983, p. 6 ; Clerkin 2013, p. 14.

8Another burial zone at Carthage was located immediately southwest of the city, by the circus, in what is now the Yasmina district. Excavations have uncovered a small section that was in operation from the late first century C.E. The few recovered inscriptions commemorated non-servile individuals17. The tombs themselves, ranging from humble burials to large monuments adorned with stucco relief, suggest a diversity of socio-economic representation, including the pair, a presumed husband and wife, who were commemorated by portrait statues of a charioteer and a woman18. As at the Cemetery of the Officiales, cippi were the most popular choice for monuments19. Although some monuments abutted others, there seems to have been an overriding orthogonal layout to the structures that kept them orderly and ensured that were corridors for pedestrians.

• 20 This summary is based on Ben Abed, Griesheimer 2001 ; 2004b.
• 21 Ben Abed, Griesheimer 2001, p. 555 no. 10, with response by S. Lancel on p. 591.

9At the southern base of the Cap Bon peninsula, just south of present-day Hammamet, Pupput was located at the crossing of two major roads, the north-south route linking Carthage to Hadrumetum and the east-west route connecting the peninsula to the interior, via Thuburbo Maius20. It was elevated to colonial status under Commodus. Its main necropolis, which was located north of the town, dated back to the late first or early second century C.E. It grew to cover at least seven hectares, eventually becoming so densely packed with grave monuments that as little as 1.68 m² of space per tomb existed in the open southern zone. There were also some burials that were either unmarked or signalled only by a small sand tumulus; many of these subsequently seem to have been disturbed or cut into by later graves. The cemetery contained dozens of walled compounds, varying in size from 17 to 156 m². No inscriptions survived21, but the size and density of the cemetery justify interpreting it as having served a cross-section of Pupput’s population. The growth of the burial zone seems to have been organic, with one compound accreting onto another and no regularity of tomb placement in open areas. As space progressively filled up with cupulae, cippi, and walled enclosures, it would have become increasingly challenging to navigate one’s way through the cemetery. The key organizational feature of the cemetery as a whole seems to have been the maintenance of two east-west alleys to provide essential passage across the width of the necropolis, although by the end of the 2nd c., even these were being encroached upon by funerary constructions.

• 22 See Alexandropoulos 2007b, p. 278-308 for discussion of individual mints, which in the early Empir (...)
• 23 According to Mahjoubi (1970, p. 8), the accidental discovery of the necropolis in 1960 was the fir (...)
• 24 As noted at Thaenae by Espinasse-Langeac (1898, p. 142), and Jeddi (1995, p. 150 ; at Hadrumetum b (...)
• 25 For treatments of select epitaphs from Henchir Methkal, see Aounallah, Ben Abdallah, Hurlet 2007, (...)
• 26 Another assessment factor could be the age distribution and health of the deceased. To date, exten (...)

10The remaining sites for consideration in this section were all located in east-central Proconsularis, what would become Byzacena in late Roman times. Of the sites from this region to be studied, Hadrumetum (Sousse), Leptiminus (Lamta), and Thaenae (near Sfax) were active ports that had been of sufficient significance to be granted the right to mint coins under Augustus22, from which time their fortunes grew to the point that they achieved colonial status in the early to mid-2nd c. In contrast, Raqqada was located in the interior (a few kilometres south-southwest of present-day Kairouan) and, as far as surviving evidence indicates, was a humble rural locale23. Few funerary inscriptions, most of them fragmentary, survive from the excavated cemeteries of these towns, making demographic assessment difficult24. Among the preserved inscriptions, although there was sporadic mention of an Imperial slave, administrator, or veteran25, most epitaphs did not identify the deceased beyond name and age at death. The presence of generic epitaphs, in combination with a vast and densely-occupied burial zone, could be taken to signify that a cemetery served the general population, notwithstanding the possibility that certain sectors may have served specific families or groups, as at Carthage26.

• 27 At the turn of the 20th century, it was sometimes known as the “Camp Sabattier” (also spelled Saba (...)
• 28 Lacomble, Hannezo, 1889, 117 ; Gauckler in Ordioni, Maillet 1904, 432.

11Of the necropoleis ringing Hadrumetum, the most extensive in Imperial times was located west of the town, beyond the Punic-period cemetery27. Closely packed with graves, it extended for a distance of c. 4km from Hadrumetum, along the Roman road towards Theueste (Tébessa). It included both communal burial areas and walled enclosures of varying sizes that may have served families, professional colleges, or other groups28. Although some sectors had closely-set graves, there was reportedly no intercutting. Most burials were covered by masonry monuments (predominantly cupulae), although some were unmarked. There were also cremation urns in mausolea and hypogea, although the latter more typically contained inhumations.

• 29 For a synopsis of archaeological evidence for Punic through late antique cemeteries at Leptiminus, (...)
• 30 Gauckler 1897b, 467-468 (who saw no grave monuments) ; Smet 1913, 329.
• 31 Hannezo, Molins, Montagnon 1897, p. 298-299.
• 32 Smet 1913, pl. XXIX. Since the walls were not dated, it is possible that they were constructed aft (...)
• 33 Foucher 1967, p. 140-141, plans I and II, 5-6. Foucher (1967, p. 140) briefly noted cremation buri (...)
• 34 Foucher 1967, p. 141.

12At Leptiminus, two main burial zones have been excavated29. The Henchir Methkal necropolis was at the southwest end of town, in the vicinity of present-day Bou Hajar. The earliest noted burials, which were inhumations, were in Punic-style shaft-tombs; no humbler graves of that period were reported. By sometime in the first century C.E., those underground tombs were covered at ground level by new, more modest burials, predominantly cremations. Two early investigators described the cremation burials as being in a largely undifferentiated zone of superposed layers of ash and bone, with some grave goods mixed in30. However, the cemetery’s earliest explorers, in 1895, thought that they could discern three distinct strata of burials above the Punic tombs. Their Level 3 was the earliest (“Neopunic”, or first century), while Level 1 was the closest to the surface and contained some Christian elements31. According to their description, there were no surviving ground-level grave markers except in Level 2, where certain graves were covered by either masonry markers or flat tiles. Smet’s 1913 plan shows some vestiges of enclosure-like walls and a handful of masonry tombs, the latter seeming to delimit the northeastern extent of the area that he investigated32. In 1955, Foucher identified a walled compound that included a mausoleum and masonry graves, which had been robbed out but seemed to have been for inhumations33. It may be, then, that the infrequent adoption of grave superstructures at Henchir Methkal coincided with an increase in inhumations, perhaps during the second half of the second century, when, on the opposite side of town, the East Cemetery was coming into use34.

• 35 Hannezo, Molins and Montagnon (1897, p. 300-301) also investigated some inhumations in what they c (...)
• 36 Stone et alii 2011, p. 178, 182.
• 37 For full publication of the East Cemetery, see Leptiminus 4 2021 ; for previous reports on these s (...)
• 38 There were also mausolea and hypogea in the East Cemetery, but to date only inhumations have been (...)

13The East Cemetery extended along the Dhahret Slama and Jebel Lahmar, a ridge and hill that today serve as the overlapping boundary between the eastern suburbs of ancient Leptiminus and the growing western spread of present-day Lamta35. Although Punic shaft-tombs dotted this eastern zone in pre-Roman times, the land was little exploited until the middle Roman empire, when it was put to both funerary and industrial use36. As a result, the East Cemetery faced the least competition for space for cremation graves among the cemeteries in this study. Within these burial grounds, neighbouring sites 302 and 304 have yielded the most information about cremation and ensuing funerary practices37 (fig. 3). Their cremation graves date to between the late second and mid-third centuries, corresponding to the last decades of cremations in our study area. Every identified cremation grave, including those for individual infants, received a masonry-and-rubble superstructure (a cupula, stepped tomb, or cippus), mainly set within an open communal area, but occasionally within one of the walled compounds38.

• 39 Salomonson 1970, p. 65 n. 143.
• 40 Mahjoubi 1970, p. 10.

14To the interior of the region, the necropolis at Raqqada probably served most or all of the population of that rural town. Its full size has not been established, but Salomonson estimated that a combination of initial salvage work and more methodical excavations in the early 1960s may have exposed over 450 cremation and inhumation graves, dating between the first and fourth centuries39. While there were no masonry superstructures over the graves, three scraps of inscriptions indicate that stelae once signalled some burials40. As was the case during the earlier phase of cremations at Henchir Methkal, there was little attention to demarcating individual burials, a situation that inevitably led to some violations of existing graves as new ones were being dug.

• 41 Feuille 1938-1940, p. 641-642. Excavation at the necropolis of Thaenae has recently resumed : http (...)
• 42 Espinasse-Langeac 1892, p. 140 ; Barrier, Benson 1908, p. 25 ; Feuille 1938-1940, p. 641.
• 43 Fortier, Malahar 1910, p. 84 and pl. XIX.
• 44 For typologies of grave monuments at Thaenae, see Barrier, Benson 1908 and Feuille 1938-1940.

15Finally, back on the coast at Thaenae, three burial zones spread in a fan stretching at least 800 m north of the town walls41. The northeast zone potentially came into use first, in the first century C.E.; it was situated along the road leading north to nearby Taparura (Sfax). The other two zones seem to have spread along the southern and western routes42. Between those burial zones and the town walls was a large, enclosed ustrinum, a communal cremation area measuring some 23 by 33 metres43. Graves were typically marked by one of a variety of monument types in ­masonry-and-rubble (most commonly cupulae) or contained within a mausoleum44 (fig. 4). Compounds appear to have been rare. Although the graves were set fairly close together, there was enough space between them to allow for small gatherings and pedestrian traffic.

• 45 Delattre 1898a, p. 84.
• 46 For the chronological indicators, see Lassère 1973, p. 25-57.

16Although direct evidence for cemetery administration is generally lacking, indirect evidence may indicate the choices that were being made. At Carthage, for instance, there must have been some form of centralized organization to deal with issues resulting from the high usage of the Cemetery of the Officiales. Although there was no evidence of an enclosure wall at Bir es Zitoun, its surface area was clearly somehow contained, for as it became built up, the old graves were covered over and a new stratum of burials began above. Repeated ground-raising resulted in superimposed layers that eventually rose a total of 7 m45. In the first century C.E., graves were marked by stelae. Masonry cippi-altars began to appear as early as the Flavian era and had become the characteristic grave marker by the Antonine era46; the implications of this change will be considered in the section on commemorative rites, below. Although the nearby, walled Bir el Jebbana sector was also densely filled with cippi-altars, no successive strata of graves were reported there.

• 47 Delattre 1888, p. 152 ; 1898b, p. 217-218.

17As another indication that land was at a premium at the Cemetery of the Officiales, the two sectors were so crowded with tomb monuments that making one’s way across the necropolis would have been difficult. There was limited space for conducting rites at a grave. At this site, the preference for upright cippi-altars, rather than long cupulae, may reflect these spacing constraints: these markers generally had a limited footprint of between 0.5 and 1 m per side, but rose tall, at c. 1.5 m on average47. Some of the more elaborate cippus monuments, those surmounted by an arched niche, stood much taller. In contrast, cupulae at several other towns in the study area were usually at least long enough to cover the entire area of the pyre (with the body laid out on it), but were shorter, averaging about 1m tall.

• 48 As cited by Lavigerie 1881, p. 33 ; Delattre 1888, p. 154 ; 1898b, p. 217.
• 49 Gauckler 1896, p. 152 ; 1897a, p. 89 no. 3 ; Delattre 1898b, p. 218.

18Within the Cemetery of the Officiales, the density of occupancy extended beyond the visible monuments, for many of them were designed to hold two or more occupants. Multi-urn tombs were designed for future additions of family members, in that empty urns were encased within the monument, each attached to its own libation tube, through which the new cremated remains (‘cremains’) could be dropped to reach their final resting place in the urn below; Delattre reported finding some cremated bones in the tubes48. As each new person was added, another epitaph was affixed to the monument; if there was no room on the front, the plaque went on the side. In other cases, a tomb that had been designed for single occupancy, with one urn, received additional cremains. One cippus (fig. 5), for example, contained a single urn and originally had an epitaph that was centred on the main façade between the garland and vegetal motifs. However, a second epitaph, which obscured part of that décor, attests the later, unplanned addition of another person’s cremains. Thus, the number of individuals who were buried in this cemetery significantly exceeded the number of monuments. Occasionally, the exterior access to the tube of a grave was partially closed off by a marble or lead grate49. While this insert might have restricted the type of commemorative offering to liquids or might have been intended to repel vermin, another possibility is that it prohibited additional human remains from being added to the tomb. The overall impression at this cemetery is one of a tension between the desire to commemorate the dead with a substantial monument and the reality of intense pressure for space.

• 50 See, for example, Lacomble and Hannezo (1889, p. 126) (Hadrumetum) ; Carton (1909, p. 37-38) (Gurz (...)
• 51 Bailet 2012, p. 544-547.
• 52 Stirling, Moore 2021, p. 99.

19In the other cemeteries, in contrast, monuments encasing two (or more) urns were exceptional50. Such cases are distinct from situations in which an unfortunate coincidence of deaths led to two or more individuals being buried together. At Pupput a few graves contained a neonate and a woman who may have been the infant’s mother, while others combined an adult (sometimes male) with one or more subadults51. One grave in the East Cemetery at Leptiminus contained the cremains of a young woman and a child52.

• 53 Noy’s study (2000) included a review of the ancient terminology and practical considerations for b (...)
• 54 We follow McKinley (2000, p. 40-42 ; 2013, p. 150) in using the terms “pyre goods” and “grave good (...)

20Cremations could take place either on the spot where the grave would be located (“primary” or bustum cremations) or at a pyre in a communal area, such as the aforementioned ustrinum at Thaenae53. In the latter case, the cremated remains, potentially including accompanying objects, would then be transported to the gravesite (“secondary” cremation burials). The finds in cremation graves could therefore come from ritual placements either on the pyre at the time of cremation (“pyre goods”) or in the grave at the funeral (“grave goods”), collectively through “rituals of deposition”54. In listing grave assemblages, most investigators have not distinguished which objects were pyre or grave goods. Items that they described as entirely burnt or melted presumably had been exposed to the cremation fires. In addition, any noted organic remains most likely survived by virtue of having been carbonized on the pyre, unless contextual evidence signals their addition at a later stage.

• 55 Bodel 2000, esp. p. 136-139 ; 2004, esp. p. 147-162.

21At those cemeteries that experienced frequent traffic, one might expect to find evidence that the pyre and grave assemblages were somewhat stereotyped, in terms of having core “sets” of goods. These “sets” may have become habituated due to people’s common exposure to the rites. Alternatively, the very regularity of burials may have encouraged the presence of an entire industry of providers of goods and services, which in turn led to a greater standardization of mortuary materials, including fuel and pyre materials; urns; items of ceramic, glass, and other materials; scented products such as perfumes and incense; and food and drink. There is no known epigraphic evidence in North Africa for the funerary professionals who were attested in Italy as of the late Republic55, but busy cemeteries at large urban centres would have been prime locations for such specialists to guide the funerary processes. In contrast, those who made use of quieter cemeteries may have had to be more self-sufficient and gather supplies on an ad hoc basis, leading to greater individualization of certain elements.

• 56 Hannezo, Molins, Montagnon 1897, p. 299. Gauckler (1897b, p. 467) specified that the metal urns we (...)
• 57 The full dimensions of the burial zone at the east side of the city is not known. The overall fune (...)

22Leptiminus’ two cemeteries provide a striking contrast in this regard. At the busy Henchir Methkal necropolis, Hannezo et alii reported that, across the various phases, each grave contained two lamps, a lidded ceramic or metal vase (presumably the urn), ceramic dishes (including fine-wares in Level 2), two or more terracotta or glass unguentaria, and one or two bronze coins, in addition to which further items might be added56. Although the ubiquity of such fulsome assemblages is likely exaggerated, the explorers must have been encountering these kinds of items frequently to have formed that opinion. On the other side of town, at the East Cemetery, where we estimate that an average of only one burial per year was taking place in the excavated sector at its height (counting both cremations and inhumations)57, not only were there no patterns to the items in the grave, but there were also fewer items than at Henchir Methkal and almost no fine-wares. One might not have predicted that difference from the elaborate masonry monuments that each person who was buried in the East Cemetery received, especially in contrast to the minimally-marked graves at Henchir Methkal.

• 58 We propose these two phases of ritual assemblages based on our study of the grave catalogue of the (...)
• 59 Salomonson 1970, p. 64.

23At the Raqqada necropolis, a different phenomenon highlights the strength of local practices within a community’s rituals of deposition. Our review of its grave assemblages indicates that there were two discernible phases, distinguishable by the characteristic “set” of items58. During the first phase, to which between 11 and 13 cremation graves can be assigned, all but four had a glass unguentarium as a pyre good and every grave except one had one or two lamps (Loeschke forms 1 and 4), datable to between the late first and early second centuries C.E. There were no cook-wares. The sole example of a table-ware was a thin-walled cup in one grave. Then, in the 2nd c. (with 21 datable cremation graves), glass unguentaria disappeared from the assemblage and there were no identifiable pyre goods. Lamps (now predominantly Deneauve form VII) remained the most popular grave good, in 15 graves, now joined by a casserole (a variant of Hayes form 183/184), in 14 graves. Half of the casseroles were lidded and the same proportion had an African-red slipped (ARS) dish or plate that was set, inverted, over the lidded or unlidded casserole. Although no contents were observed within the casseroles, their bases were consistently burnt59. The casseroles and ARS dishes may attest a funerary meal that was cooked and served at the grave during this second phase. The surviving evidence therefore suggests some significant, shared modifications to the mortuary rites at this cemetery.

• 60 Carton 1909, p. 43. To be clear, the graves could also contain other items ; Carton’s focus here w (...)
• 61 The small bottles are labelled in the reports by various terms including unguentarium, fiole, bals (...)

24Another means of highlighting the relationship between rituals of deposition and the locally-conditioned environment is to test Carton’s statement, cited at the beginning of this paper, that the standard items in the African cremation graves were cinerary urns, lamps, and coins60. To those, we can add another frequently-mentioned item, an ­unguentarium, a small bottle presumed to have contained perfume or unguents61. Examining the patterns for those four categories of objects on a site-by-site basis reveals the tremendous variability that underlies Carton’s blanket assertion.

• 62 For illustrations, see Delattre 1888, p. 8 figs. 2, 4, and 5 ; Gauckler 1897a, figs. IV and VI.
• 63 Gauckler 1907, p. 479-481, p. 506 no. 467, p. 510 no. 480.

26At the Cemetery of the Officiales at Carthage, the most common urn shape was a neckless vase with its maximum diameter at the mid-belly or shoulder, sometimes with handles at the upper body62. This morphology was quite similar to, and therefore probably influenced by, that of terracotta urns at Rome, as might be appropriate for the burial grounds of Imperial servants. Carthage is the only known site in the study area to have adopted that form as its primary choice. In contrast, less than 15 km southwest of Carthage, the rural settlement at Djebel Djelloud was burying its cremated dead in small, trough-like stone ossuaries, sealed with a lid63.

• 64 Pupput : Pupput 2004, p. 117 ; Bonifay 2004, p. 48-50. Hadrumetum : Choppard, Hannezo 1893, p. 194 (...)
• 65 Carton 1909, p. 37-38 (Gurza) ; Foucher 1964, pl. XXXIII.d (Hadrumetum).
• 66 Stirling, Moore 2021, 103-110.

27At Pupput, Hadrumetum, Leptiminus’ East Cemetery, and Thaenae, local domestic wares were adapted for use as cinerary urns. Two-handled common-ware jars have been attested at all four sites64 (fig. 6). In the region of Hadrumetum and the East Cemetery, casseroles, of either form Hayes 183 or a variant of Hayes form 198, could instead contain the cremated remains. The casserole-urns were covered by an inverted or upright Hayes form 185 lid, the centre of which had been pierced to allow offerings to pass from the libation tube into the urn65 (fig. 7). In the case of the East Cemetery, these jars, casseroles, and lids had almost certainly come from the nearby potting complexes, which produced those types in the same fabric. Residue analysis indicates that they had likely not been used for storing or cooking food prior to arriving at the cemetery. Moreover, they had firing flaws that, although minor, may have rendered them unfit for domestic use, yet acceptable for burial purposes66.

• 67 E.g., Choppard, Hannezo 1893, p. 194-196 ; Carton 1902, p. 230 ; Goestchy 1903, p. 164 ; Icard 190 (...)
• 68 Hadrumetum : Choppard, Hannezo 1893, p. 194 ; Foucher 1964, pl. XXXIIId. Carton (1909, p. 36) note (...)
• 69 Delattre 1888, p. 8.

28Another type of urn seems to have been specific to the area of Hadrumetum: a narrow-mouthed, globular terracotta urn with pre-firing holes in its lower body (fig. 8), the latter trait frequently causing its discoverers to compare the jars to present-day couscous pots67. While this form was distinctive, the perforations reflect a broader phenomenon both there and in the East Cemetery of Leptiminus, where holes were added at some point after firing to the bases of the jar- and casserole-urns68. These holes would have allowed drainage into the soil below of any foods and liquids that reached those urns, via libation tubes that connected to the tomb exterior for receiving post-funerary offerings. A similar phenomenon, but of unknown frequency, also took place at the Cemetery of the Officiales at Carthage, in certain cippi whose masonry encased two superimposed voids. The upper void contained the urn and a few goods, while the lower one included further items and pyre debris69. Commemorative offerings deposited in the libation tube drained through the urn in the upper cavity, down into the lower one, and from there into the ground.

• 70 Pupput : Ben Abed, Griesheimer 2001, p. 567 ; the cremains and ashes were typically covered with a (...)
• 71 E.g., Ordioni, Maillet 1903, p. 547-548.
• 72 Stirling, Moore 2021, p. 104 ; Jeddi 1995, p. 144, p. 149.

29However, in some primary burials, the dead did not receive a cinerary urn at all. Their bones and ashes were simply left in situ or swept into a heap for the funeral, a practice that seems to have been predominant for busta at Pupput and Thysdrus (El Jem), and pervasive among the cremation graves at Raqqada70. An unknown proportion of busta graves at Hadrumetum lacked an urn71. One note of caution that must be applied here is that sometimes cinerary receptacles were made of perishable materials. For instance, the only traces of wooden containers may be nails, appliqués, or organic stains, as seen at some tombs at the East Cemetery (Leptiminus), while in the humble cemetery at Thaenae some cremains had been wrapped in fabric72.

• 73 Delattre 1898a, p. 98 ; Gauckler 1897a, p. 102 ; Larminat 2012, p. 301 ; Moore, Stirling 2021, p.  (...)
• 74 Lamps in hypogea and mausolea, on the other hand, provided illumination for visitors : e.g., Barri (...)
• 75 DAGR., s.u. « lucerna, lychnus » (J. Toutain).

30In the first century C.E., the lamps that accompanied the dead might be imported, especially from Italy, or be ­locally-sourced. During the second century C.E., African-made, coarse-ware lamps, especially of Deneauve types VII and VIII, came to dominate the market. Most investigators have not recorded whether lamps in graves were entirely burnt from inclusion on the pyre, had carbon traces only on their nozzle from having been lit, or were pristine and presumably unused. However, in the cases where the lamps’ condition has been noted or where clear photographs are available, it seems that most examples had no discoloration, even if they bore the metal wick-holder that made them ready for use73. The general practice, then, was to provide a new lamp to serve the deceased in the afterlife74. Whether or not this situation continued pre-Roman customs is unclear, since the condition of pre-Imperial lamps in tombs from the study area has also mostly gone unrecorded. However, it has long been recognized that the broader Roman world favoured unused lamps in the grave75.

• 76 Bonifay 2004, p. 30-31.
• 77 Pupput 2004.
• 78 Inverted lamps : Pupput 2004, figs. 56, 65, and possibly also fig. 50.

31Lamps were the artifact that was most consistently found in graves at Pupput. Some 500 examples have been noted76, although they have not yet been correlated in print to the mode of burial or other contextual considerations, except for the portion of the southern sector that was presented in the preliminary report77. Within that sector, 12 cremated individuals, from teenagers to adults, were buried between the mid-2nd c. and the Severan period. All but three graves contained one or two lamps. In those primary burials where it was possible to identify the orientation of the body because the human remains had been left in place after cremation, the lamp was most often placed by the feet. Photographs indicate that the lamps might be placed either upright or upside-down and that they were not discoloured by fire, so they must have been added at the funerary stage78.

• 79 Hannezo, Molins, Montagnon 1897, p. 299.
• 80 Moore, Stirling 2021, p. 158-159 (graves G-105, G-020, G-071). A lamp was also a pyre good at a bu (...)

32At Henchir Methkal (Leptiminus), Hannezo et alii claimed that each grave had held one or two lamps79. In contrast, at the East Cemetery, lamps were found in only three of ten intact or fairly intact cremation tombs, each with disparate treatment. The earliest grave, dating to the late 2nd c. and preserving no skeletal traces due to robbing, had one lamp as a pyre good and two as grave goods. The next grave, slightly later and belonging to an infant, had one lamp as a pyre good. Finally, the third grave, datable to the first half of the third century and belonging to an adult (possibly male), had one lamp as a grave good80. At neither cemetery was there any noted pattern of placement within the grave.

• 81 Delattre 1898a, p. 98 ; Gauckler 1897a, p. 102.
• 82 Barrier, Benson 1908, p. 53.
• 83 Jeddi 1995, p. 143, p. 150.

33At the Cemetery of the Officiales at Carthage, one or more lamp(s) formed an intrinsic part of the grave assemblage. Often the lamp received habituated placement with other artifacts; for instance, it might be placed on a patera over the cinerary urn, or it could be set next to the urn and covered by a dish or cup81. That level of stereotyped behaviour contrasts with the situation at Thaenae, where it appears that discrete graves did not generally contain lamps; instead, mausolea were the usual source of the lamps that Barrier and Benson found82. In the humbler cemetery at Thaenae that Jeddi excavated, lamps were found only with inhumations, even though cremations were taking place there contemporaneously83.

• 84 As noted by Cagnat and Chapot (1916, p. 336), caution must be used when relying on coins as dating (...)
• 85 Cagnat 1909, p. 202 ; Desnier 2004, p. 16-17 ; Alexandropoulos 2007a, p. 439-440 ; Visonà 2014, p. (...)
• 86 Slim 1992-93, p. 370-372, p. 384 ; Ben Jerbania 2008 ; 2013.
• 87 Cagnat 1909, p. 202 ; Desnier 2004, p. 16-19.
• 88 On the symbolic value of coins in funerary contexts, see Stevens 1991.

34At both cemeteries and sanctuaries across Roman North Africa, the proffered coins were generally ­low-denomination bronzes and commonly either centuries old (Punic or Numidian issues) or so worn that they were illegible84. Cagnat and others have argued that the poor penetration of the African provinces by early Imperial coinage had compelled inhabitants to keep old coins in circulation through at least the early Imperial period85. However, the situation for east-central Africa Proconsularis differed, in that its ports had long been established stops within the overseas trading networks. Furthermore, as previously mentioned, several towns had been granted minting rights under Augustus and Tiberius. Even in the Byzacene interior, Thysdrus and its region had enjoyed centuries of economic integration with the western Mediterranean before its cremated dead received coinage and imported fine-wares in the early Imperial period86. It could be argued that, by the Imperial period, old coins had little current monetary value and were therefore easily donated to the dead. A counterpoint to that argument is the evidence from both votive and funerary settings of perforated and otherwise altered old coins, possibly converting them to amulets or pendants, but at any rate indicating that with age they took on new purposes beyond commercial exchanges between the living87. Even unmodified coins may have held symbolic power due to their antiquity, their metal, or perhaps even their motifs88.

• 89 Gauckler 1897a, p. 102.
• 90 Hannezo, Molins, Montagnon 1897, p. 299.
• 91 Lacomble, Hannezo 1889, p. 111 ; Hannezo 1890, p. 448.
• 92 Ordioni, Maillet 1904, p. 451 ; Gauckler in Ordioni, Maillet 1904, p. 435.
• 93 Desnier 2004, p. 15 ; Larminat 2012, p. 301.
• 94 Calculation based on the grave catalogue of Mahjoubi 1970, p. 10-22.
• 95 Espinasse-Langeac 1892, p. 142 ; Gauckler 1904, p. CLVI ; Barrier, Benson 1908, p. 48, p. 54 ; Jed (...)

35Antique coins received ritualized placement within graves at the Cemetery of the Officiales at Carthage, where the coin was carefully set on the discus of the lamp, which itself might be covered by a patera or dish89. At Henchir Methkal (Leptiminus), coins were set on either a lamp or a dish90. The coin-lamp pairing was also an occasional find in the necropoleis of Hadrumetum and Sullecthum (Salakta)91, although in at least one Hadrumetine sector no coins were found in graves92. In fact, at many cemeteries, coins were not a standard component of the grave assemblage. Specific results from Pupput are forthcoming but seem to be in the order of only slightly more than one out of every ten graves93. A similar proportion was evident at Raqqada (4 out of 43 cremation graves)94. At the East Cemetery of Leptiminus, in contrast to the situation across town at Henchir Methkal, only two of 13 cremation burials had coins, neither one set on other items. Coins were almost absent from Thaenae95. Carthage and Henchir Methkal therefore stand out as exceptions in terms of the regularity of coins in cremation burials. While residents of small, rural towns like Raqqada and Djebel Djelloud may have lacked the means to make such donations, the infrequency of coins at the monument-rich cemeteries of economically-healthy ports such as Hadrumetum, the East Cemetery (Leptiminus), Sullecthum, and Thaenae indicates that coins were simply not desired elements in cremations or the subsequent burials in their respective parts of Africa Proconsularis.

• 96 Outside of our study area, recent work on votive unguentaria includes the investigations at Henchi (...)
• 97 In addition to reviewing recent scholarly interest in sensory experiences of past humans, Price (2 (...)
• 98 Carton 1907, p. 82.
• 99 Robinson 1959, p. 15.
• 100 Two large houses and a paved street were subsequently built over part of the cemetery : Slim 1992- (...)
• 101 Slim 1992-1993, p. 366-369.

36Small bottles for perfumes and unguents are attested from both votive and funerary contexts in ancient North Africa96. In most cases, they seem to fit within a broader Mediterranean belief that both the gods and the dead derivedsome pleasure from aromas, while the wafting aromas would have formed evocative sensory associations for the mortals who participated in rites97. Beyond their value as offerings, the application of scents to the body of the deceased while it was being prepared and to the pyre also served to mask (at least partly) the disturbing smells of incipient decomposition and then of burning flesh, much as placing pine-cones, pine branches, or incense on the pyre did. Scented offerings could also be given at the funerary and commemorative stages. In both votive and funerary settings, the on-site deposition of unguentaria indicates that the containers themselves were considered part of the gift, perhaps as lasting physical evidence of the ritual donation of ephemeral scents. In fact, at the Neo-Punic sanctuary at Ciuitas Popthensis (El Kenissia, near Hadrumetum), votive offerings included both real and “dummy” unguentaria – objects in the shape of unguentaria, but lacking a hollow cavity in which contents could be held98. Evidently at that sanctuary, offering a ­simulacrum could substitute for the act of making a scented offering, another indication of the perceived value of tangible materials as embodying intangible acts. In the last centuries B.C.E. around the Mediterranean, including in North Africa, unguentaria were normally made of terracotta. Over the course of the first century C.E., glass unguentaria increased in popularity to the extent that, by the end of that century, they had largely replaced the ceramic versions99. This shift had not yet taken place among the cremation burials at a cemetery at Thysdrus when it went out of use, sometime between the Flavian period and the early 2nd c. C.E.100. Ceramic unguentaria seem to have been common as both pyre and grave goods there, but only one glass version was reported101.

• 102 Delattre 1898a, p. 99 ; Gauckler 1897a, p. 88 (grave no. 1). Although Foucher (1964, p. 197) claim (...)
• 103 Foy 2003, p. 67 ; 2004, p. 71 ; Pupput 2004, p. 132-134.

37The period of the abandonment of the Thysdrus cemetery coincides with that of the rise of cippi-tombs at the Cemetery of the Officiales at Carthage, where a ceramic or glass unguentarium was reportedly encountered in most graves. Glass unguentaria were often pyre goods, presumably placed on the funerary bier after their scented liquids had been poured out. The melted and distorted remnants of these vessels were then buried with the ashes in the grave, sometimes in the cinerary urn itself102. As already noted, the same phenomenon happened during the first phase at the necropolis at Raqqada, but that practice ceased there at some point during the second century. At Pupput, in contrast, glass unguentaria seem to have been relatively late choices, dating to the late second or early third century, and were apparently introduced during the funeral rather than at the cremation103.

• 104 Hannezo, Molins, Montagnon 1897, p. 299. Gauckler (1897b, p. 467-468) also described the ceramic f (...)
• 105 Certain graves contained a few shards or flakes of unmelted glass that were too small and sparse t (...)
• 106 Gauckler 1907, p. 467 ; the location was described as above the ossuary, so the unguentaria may ha (...)

38At the Henchir Methkal necropolis of Leptiminus, Hannezo et alii reported that the typical grave assemblage across all phases included at least two unguentaria, either spindle-shaped or piriform, and in either terracotta or glass104. While it may be tempting to nuance their report by assuming that the ceramic examples were in earlier graves and the glass in later ones, finds at the same town’s East Cemetery give reason to exercise caution about dating graves by the material of these petite bottles, at least at Leptiminus. In the East Cemetery, no glass unguentaria were identifiable in pyre or grave assemblages105. Instead, solitary examples of piriform ceramic unguentaria were found within a handful of cremation and inhumation burials. Since the earliest of those graves dated from the late second century C.E., these small vessels were not contemporary; they had probably been salvaged from earlier tombs in the vicinity. At Djebel Djelloud, near Carthage, even older terracotta ­unguentaria, this time fusiform in shape, were occasionally found around the libation tubes of cremation graves, while glass ­unguentaria (whose condition was not noted) were sometimes found in the ossuaries of the same graves, below them106. At both cemeteries, the antiquity of the small ceramic vessels must have been valued and perceived to enhance the ritual acts. In these examples of re-use, it is not known whether the containers were refilled with aromatic offerings for the rites or whether they constituted the offerings themselves, as with the dummy versions at Ciuitas Popthensis.

• 107 Moore, Stirling 2021, p. 151-153. Burnt pine shells were also found in primary cremation graves at (...)
• 108 Stirling 2004, p. 440. Olive wood was found to be the predominant wood type (86.47 % of the carbon (...)
• 109 Chevy 1904, p. 89.
• 110 Ov., fast. 5, 436-440 ; Plin., nat., 18.118 ; Matterne, Derreumaux 2008, p. 110.
• 111 Bailet 2012, p. 544 ; Larminat 2012, p. 304.

40One aspect of the study of rituals of deposition that needs more attention is the evidence for perishable goods, such as food, drink, and scented offerings, which can be brought to light through ceramic, faunal, palaeobotanical, and residue studies. At the East Cemetery of Leptiminus, organic pyre offerings included stone pine-cones (presumably burned for their scent) and pine-nuts, pomegranates, grapes, figs, dates, almonds and walnuts, and bread107. Carbonized olive pits were also present, although it was unclear whether they were intentional offerings or intrusive by virtue of their ubiquity in an oleoculture-intense region where olive trimmings and pits were a common fuel source108. Chevy reported finding burnt pinecone in a cremation grave at Souassi el-Ksiba, c. 10 km south-southwest of Hadrumetum109. No beans or lentils have been observed at any site in the study area, despite their use as funerary foods in Italy, as mentioned in Roman literary sources and confirmed by finds at Pompeii110. As for faunal remains from the cremation stage, only one example has been reported, a piece of chicken in a cremation grave at Pupput111. However, if boneless meat had been placed on the pyre, it would not have left visible traces.

• 112 Raqqada : Mahjoubi 1970, p. 18 (tomb B 25) ; Pupput : Pupput 2004, p. 112 (tomb 137) ; Leptiminus  (...)
• 113 Pupput 2004, p. 132 (tomb 134 : a Hayes form 8A bowl) ; Moore, Stirling 2021, p. 145-146, fig. 4.1 (...)

41Infrequent finds of entirely-burnt ceramics provide indirect evidence for offerings of food or drink at the cremation stage: a jug at Raqqada, a Hayes 182 lid at Pupput, and two Hayes 184 casseroles in a grave at the Leptiminus East Cemetery112. The latter two sites also each provide an example of fine-ware that must have been used for pouring offerings out onto the pyre, then was ritually broken and its pieces separated, since part of the vessel was found, entirely blackened, in the grave, while the rest was recovered, unblemished, either in or outside of the grave113. Since pottery that was used in cremation rites may have been excluded from the burning pyre, and since vessels from either busta or ustrina were not necessarily included in the grave, the ceramic evidence for this stage of rites is presumably incomplete.

• 114 Delattre 1888, p. 154 ; 1898a, p. 98-101 ; Gauckler 1897a, p. 88-89, 91, 95 ; Hannezo, Molins, Mon (...)

42As for offerings at the funerary stage, in the cremation graves of the Cemetery of the Officiales and Henchir Methkal, investigators consistently found ceramics that either might be remnants of a funerary meal or may have held food or drink to provide sustenance after burial114. Unfortunately, most early excavators at other sites merely noted they had found pottery, without regard for context, frequency, or type. The old reports from Thaenae give the impression that ceramics other than urns were uncommon in graves.

• 115 Based on Pupput 2004, p. 121 (tomb 603), p. 125 (tomb 604), p. 127 (tomb 1165), and p. 135 (tomb 1 (...)
• 116 Moore, Stirling 2021, p. 141 (grave G-071) ; Mahjoubi 1970, p. 18 (grave B 33). At Pupput, S. de L (...)
• 117 Calculations based on the grave catalogue of Mahjoubi (1970).
• 118 Pupput 2004, p. 112 (tomb 137) ; Moore, Stirling 2021, p. 146.

43Evidence for funerary ceramics from more recent excavations adds to the larger picture of variability between, and sometimes within, sites. As previously mentioned, sometime in the second century, most mourners at Raqqada adopted the custom of placing cook-wares and dishes in graves at the funeral. Their counterparts in the mid-to-late second century and later at Pupput and the East Cemetery of Leptiminus did not equally embrace that ritual act. Only three of the 12 published graves from the southern sector of the Pupput necropolis contained a casserole, while a fourth grave had a fine-ware bowl115. At the East Cemetery, no cook-wares or dishes in graves derived from a funerary meal, despite the standard inclusion of such vessels in burials across town at the Henchir Methkal necropolis. As at the cremation stage, it is plausible that ritual ceramics were disposed of outside the grave instead. Sometimes food was placed in the grave without a pottery receptacle, as signalled by eggshell at the East Cemetery and Raqqada116. As for liquids, at Raqqada, nine out of 21 cremation graves datable to the 2nd or early 3rd c. contained a jug, in five cases incomplete and therefore perhaps ritually broken117. Jugs were uncommon offerings at the funerary stage in the cremation graves at Pupput (one published example) and the Leptiminus East Cemetery (two)118. Across the depositional stages, detection of traces of food, potable liquids, and scented offerings awaits specialized residue testing in future investigations.

• 119 Lacomble, Hannezo 1889, p. 124 and pl. III, no. 3.
• 120 Taillade 1904, p. 368.
• 121 Ordioni, Maillet 1903, p. 550.
• 122 Ordioni, Maillet 1903, p. 550.

45Examples from the necropolis at Hadrumetum include terracotta figurines, in one case of Bes119, and in another of a female (possibly Dea Nutrix) carrying a baby120. The latter figurine was found in the masonry of an infant’s grave. Six lamps and two ceramic cups were built into another tomb121. Finally, in a semicircular wall-monument with embedded urns, the masonry encased five lead leaves that were similar in appearance to curse tablets, yet had no visible writing122.

• 123 Moore, Stirling 2021, p. 165 (tomb G-020).
• 124 Moore, Stirling 2021, p. 139-141 (tomb G-071).
• 125 Moore, Stirling 2021, p. 148 (tomb G-102/103).
• 126 Moore, Stirling 2021, p. 159 (tomb G-004).
• 127 Moore, Stirling 2021, p. 148 (tomb G-019). Since only rim sherds were fully analyzed, it is unknow (...)

46In the East Cemetery at Leptiminus, a jug was found in the masonry of a stepped tomb for a baby (fig. 7)123. At another grave, this time for an adult (likely male; fig. 6 and 9), the construction layers of a cupula contained a jug and two cook-ware lids that had both been altered (post-firing) to have a central perforation, similarly to the lids for casserole-urns; in this case, they may have served as offering siphons for the jug’s liquid124. In another example, mortared into the façade of a stepped tomb that contained the co-burial of a young female adult and a 4-7 year old child was an antique terracotta unguentarium125. At other graves, items were sealed under the base of grave superstructures. Under the step of one cupula, there was a complete Deneauve VIII.2 lamp and a fragment of an Uzita 1 common-ware bowl, a vessel type that was found almost exclusively within graves (both cremations and inhumations) at this site126. The altar for another grave sealed a ceramic bead, an iron rod of indeterminate function, and at least the rims of a ceramic unguentarium and a jug127. These all seem to have been purposeful placements, rather than intrusive debris that entered the construction accidentally or as aggregate within the masonry.

• 128 E.g., Hilali 2009 ; Rebillard 2015.

48The final stage of ritualized interaction at Roman graves took place after the grave was closed, when, at both prescribed and individually-motivated times after the burial, survivors could return to the grave to make offerings. To date, scholarly consideration of this phenomenon for Roman North Africa has tended to focus on commemorative meals at the tomb, taking into account Roman literary sources, inscriptions, and evidence of offering tables, although most of the cited evidence for the latter two categories has come from Algeria128. Once again, the evidence from our study area demonstrates the significant local distinctions that such broad approaches can overlook.

• 129 Clerkin 2013, p. 14.

49A growing concern about paying rites to the dead is evident in the gradual change from stelae to cippi at Carthage’s Cemetery of the Officiales, beginning perhaps in the late 1st c. and complete by the mid-2nd c. From a practical perspective, the cippi provided greater protection for the remains. However, the transition also marked a change in attitude toward post-funerary visitors. The stelae at the earlier graves had served to identify the individual, without assumption about what the visitor would be doing at the grave site. Not only did the much larger and more ornamented cippi make a greater visual impact to draw attention to the deceased, but they also placed expectations upon the visitors. The monuments were specifically designed to resemble altars at which honours were due. Furthermore, their construction included libation tubes that would deliver the visitors’ offerings directly to the physical remains of the deceased in the urn below, a set-up that had not been possible with the graves marked by stelae. As such, the cippus graves demanded commemorative rites in ways that earlier graves had not. At Carthage, this situation was not particular to the specialized population of the Officiales burial zone, since cippi with libation tubes were also the preferred monument for individual burials at the Yasmina Cemetery129.

• 130 Apparently, there were occasional graves with tubes for inhumed individuals at Hadrumetum : Ordion (...)
• 131 Yasmina : Clerkin 2013, p. 97-99 ; Hadrumetum : Ordioni, Maillet 1903, p. 547 ; Leptiminus : Stirl (...)

50Cippi-altars were less common in the cemeteries of other towns, but the adoption of cremation across the study area typically brought with it a marked interest in providing a focal point for post-funerary offerings. Throughout most of the study zone, libation tubes (normally of terracotta, rarely lead) were commonly built into the masonry superstructures of cremation tombs, whereas they were unusual for inhumation tombs at any period130. Another option was a low offering table (mensa), that was built next to the grave and therefore lacked direct physical contact with the deceased. Unlike libation tubes, mensae served both cremated and inhumed individuals. Usually a grave had only one of these amenities, but at the Yasmina Cemetery (Carthage), the Hadrumetum necropolis, and the East Cemetery (Leptiminus), some graves had both, attesting an unusual degree of concern for sustain­ing the deceased131.

• 132 Barrier, Benson 1908, p. 48-49 ; Feuille 1938-1940, p. 646.
• 133 Ben Abed, Griesheimer 2004a, p. 312. Out of 65 examples enumerated by Ben Abed and Griesheimer (20 (...)
• 134 Icard 1908, p. 285 ; Icard’s report was incorrectly summarized in RT 1910, p. 254, as stating that (...)

51At other sites, there was a distinct preference for one type of amenity over the other. At Thaenae, libation tubes were the norm, but no mensae were reported132. In contrast, at Pupput, mensae were far more popular, with 156 examples recorded as of 2000, in contrast to only five libation tubes; however, those examples account for only c. 11% of the total number of observed tombs, so the vast majority of Pupput graves had no fixture for commemorative rites133. Libation tubes were also notably absent at Hammam Lif (located approximately 15 km south of Carthage), despite the standard use of cupulae and cippi there134. Thus, in spite of the popularity of the tubes at towns to both the north and the south, that option for making commemorative offerings did not appeal to the residents of Pupput and Hammam Lif.

• 135 Stirling, Moore 2021, p. 137 (citing an unpublished report by M. Malainey and T. Figol).
• 136 Intriguingly, one cremation urn at Dougga, far inland and outside our study area, contained a sing (...)
• 137 Capt. Blondont, as cited by Merlin (1908, p. ccvii).
• 138 Choppard, Hannezo 1893, p. 194 ; Delattre 1898b, p. 218 ; Gauckler 1897a, p. 92-93 ; Goetschy 1903 (...)
• 139 The cremains of a child were found in the tube of tomb 1323 at Pupput : Bailet 2012, p. 542.

52Commemorative offerings that were inserted into the libation tube of grave G-071 at the Leptiminus East Cemetery included eggs and an entire raw chicken, a reminder that the term “libation tube” is conventional and perhaps should be replaced by “offering tube”. The uncooked chicken had obviously not been part of a shared meal between the deceased and the living. At the same cemetery, in other graves with no visible faunal remains, residue analysis on the interiors of libation tubes and cinerary urns identified cadaverine135, a by-product of rotting flesh, so it is likely that boneless cuts of meat had been dropped down the tubes. In the coastal zone under study, fish might be expected as an offering, but its presence has not yet been confirmed136. Tubes in graves at Thaenae contained nuts, figs, and pinecones, which must have survived as a result of being burnt before being offered137. Sometimes ­non-ingestible items were deposited, such as the curse tablets (tabellae defixionum) that were found in libation tubes of tombs at Carthage and Hadrumetum138. Finally, in the case of pre-built monuments, the tubes could also deliver human cremains to awaiting urns, as most notably occurred at the Cemetery of the Officiales at Carthage139.

• 140 Ben Lazreg, Mattingly, Stirling 1992, p. 316 ; Ben Abed, Griesheimer 2001, p. 584.
• 141 E.g., Ben Abed, Griesheimer 2001, p. 585 ; 2004a, p. 317. Ben Abed, Filantropi, Griesheimer 2007, (...)
• 142 Ben Abed, Griesheimer 2001, p. 585 ; 2004a, p. 317. At the Yasmina Cemetery (Carthage), successive (...)
• 143 Ben Abed, Griesheimer 2001, p. 585 ; 2004a, p. 317 (Pupput) ; Moore, Stirling 2021, p. 147-148 (Le (...)

53Evidence for what was offered at mensae tends to be indirect, pending future residue studies on tomb structures and observation of clear stratigraphic relationships to specific graves. The tables that served cremation graves in the study area were not like the offering tables that sometimes accompanied votive stelae dedicated to Saturn, which could have carved hollows or representations of food and dishes in their upper surface. Instead, the top surface of the offering tables could be flat or concave140. Sometimes the surface bore carbon traces, indicating the cooking of food or burning of incense141. Cook-wares such as Hayes 181 pans and Hayes 183/4 casseroles have been found in the ash deposits around offering tables and altars; at Pupput at least, these vessels had been ritually broken after use142. Evidence for liquid offerings comes from paterae, which at both Pupput and Leptiminus’ East Cemetery were found around graves that did not have tubes, meaning that their liquids had been poured over an offering table or the grave itself. These vessels, too, had been ritually broken after use143.

• 144 These trough-receptacles must have been quite similar in appearance to the hollowed stone blocks t (...)
• 145 Based on personal observation of examples on display in the Musée de Sfax (May, 2006). Since the t (...)
• 146 Feuille 1938-1940, p. 648.
• 147 Barrier, Bensen 1908, p. 31, 51 ; Feuille 1938-1940, p. 648, 652 ; Fendri 1965, p. 40.

54Many graves at Thaenae had an unusual variation of libation amenity. Although a cupula-and-tube construction was the most popular type of monument, another common combination consisted of a grave stele that was set into the back of a small (c.0.20-0.35m per side), fairly rough-hewn stone trough that sat on the ground144. In the cavity of the trough, the floor was pierced with a hole that allowed libations to drain down, below the earth, into a two-handled ceramic jar that served as cinerary urn, with no intervening tube (fig. 10). While the cavity of the trough was large and deep, the drainage hole within it was quite small, limiting offerings to liquids and perhaps solids that were smaller than an egg145. A lamp found in one cavity indicates that the trough could receive a variety of commemorative offerings146, only some of which would have made their way to the cremains below. No signs of burning in the troughs were noted. Dozens of these grave markers have been recorded at Thaenae, concentrated especially (though not exclusively) in the northeastern sector of the necropolis147. They were found both immediately adjacent to mausolea, suggesting some form of subservience to the occupants of such structures, and as isolated graves.

• 148 Beyond the distinction between primary and secondary graves, Ben Abed and Griesheimer (2004a, p. 3 (...)

55Despite the differences from one site to the next, clearly the desire to have commemorative amenities at cremation graves was widespread. Graves that lacked such provisions therefore stand out. The absence of a tube or offering table often correlated to the lack of any monument at ground level. Although the relevant burials may originally have been signalled by a stele that no longer survives, in many cases they seem to have had either an ephemeral surface marker or nothing at all, to judge by the fact that such graves were often subsequently intercut by later burials, as though the grave-diggers had not been able to detect their presence. At Raqqada, out of 43 methodically-recorded cremation graves, only one had any kind of amenity, a libation tube, for a commemorative offering. At Henchir Methkal (Leptiminus), supposedly only a minority of burials had any kind of monument at the surface. Elsewhere, at some sites where monuments were common, such as Pupput and Hadrumetum, an unspecified proportion of busta graves lacked surface markers, tubes, or mensae148. In each of these cases, one might imagine that, if post-funerary offerings were being made, they could be poured over, or set onto, the ground in the general vicinity and directed at the deceased individuals as part of the collective of di Manes.

• 149 Gauckler 1907, p. 482-483 ; Lassère 1973, p. 49.
• 150 Carton 1909, p. 43. Wolski and Berciu (1973, esp. p. 379) argued that the employment of libation t (...)

56There is no single rationale that explains the varying preference for tubes, mensae, or no commemorative fixtures. The expense of a monument may have been too prohibitive for humble individuals and towns, but financial constraints cannot have limited almost the entirety of the population that buried their dead at Henchir Methkal, when, on the other side of Leptiminus at the more spacious East Cemetery, cremation graves typically had one or both of a tube and an offering table. The situation must have been more complex than Carton’s hypothesis that libation tubes were a phenomenon of larger, “Romanized” urban centres. Indeed, at rural Djebel Djelloud, where the names in the epitaphs indicate a relatively low level of Roman influence, libation tubes were a staple component of cremation graves149. At the other end of the spectrum, there were clearly towns that fit Carton’s definition, yet rejected the concept of delivering food and liquids directly to the buried remains150.

• 151 Regarding the pre-Imperial burials at Carthage’s Cemetery of the Officiales, Delattre (1898a, p. 8 (...)

57Some centres went further, spurning the fashion for adopting any kind of monument that demanded offerings. It would be useful to know how closely such towns were maintaining mortuary traditions that pre-dated cremations. Unfortunately, at the sites under study, the mundane burial practices of the period between the fall of Carthage and the early to mid-first century C.E. are poorly understood – that is, further study is required for scientifically-excavated burials that were more representative of the general population than those in shaft and chamber tombs151. Notably, many of the cemeteries under examination came into existence in the first century or later, coinciding with the arrival of cremation practices. In prior centuries, the inhumed dead must have been buried elsewhere. This geographic shift of burial grounds adds to the picture of changing perspectives about the dead that accompanied the adoption of cremation.

• 152 See Vismara 2015a for in-depth discussion of textual sources and the argument that the outcome was (...)
• 153 The focus of this article has been on individual graves, since they were subject to different expe (...)

58It remains a puzzle as to why widespread changes, such as the move to or away from cremation, received so little recognition in the ancient texts and epigraphy, when the archaeological evidence indicates significant transformations in how a population was thinking about preparing the dead for, and maintaining them in, the afterlife152. In the last few centuries prior to the first century C.E., most North Africans would have associated the destruction of the body by fire with the dedication of infants in sanctuaries to Baal Hammon and Tanit. Along the east coast of Africa Proconsularis, the adoption of more widespread cremation in some cases correlated to the creation of a new burial ground. Several cemeteries must have required centralized administration to oversee the provision of an ustrinum and/or the designation of private (compound) versus communal burial zones, where relevant. The fuel demands of both bustum- and ustrinum-type cremations would have significantly affected local wood consumption at busy cemeteries. The cremation added a stage of rites and offerings; indeed, the preparations and time required for fires to consume the body made it the longest of the ritual stages that took place at the cemetery. If the cremated remains were to be deposited in a container, either a specially-made urn or modification of a pre-existing vessel was required. The appearance of built grave monuments at ground level suggests that many tomb owners and their survivors felt that a more concrete show of honouring and protecting the dead was needed than a stele could provide153. These monuments in turn enabled those who so desired to add another ritual stage, the encasement of offerings within the mortar-and-rubble of the construction itself, and could give new, physical emphasis to the requirement that the living provide for the dead after the funeral, via offering tubes and mensae. In sum, between the first and mid-third centuries, the choice of the majority of North Africans to cremate, rather than inhume, their dead, had cosmological and material ramifications for both individuals and the communities to which they belonged.

• 154 As stated above, it was mainly at Hadrumetum and its vicinity that potters realized that there was (...)
• 155 ARS C fine-wares include those commonly known as “El Aouja” wares, named for the location where th (...)
• 156 Moore 2021b, p. 264-265.

63A cemetery’s relationship to industries such as pottery production also seems to have been relevant. Byzacene towns like Hadrumetum, Leptiminus, and Raqqada shared a common heritage of forms of cook-wares and common-wares. However, while residents of Raqqada apparently used domestic wares solely as food vessels for the grave, their counterparts at Hadrumetum and Leptiminus sometimes transformed equiv­alent forms into cinerary urns154. In the case of the East Cemetery at least, that practice was probably born of local circumstances, namely repurposing misfired pottery that inevitably arose from the significant output of the neighbouring kilns. Proximity to different pottery production centres also offers an explanation for a seeming discrepancy between the items in the cremation graves of Raqqada and Leptiminus in the second and third centuries, namely that African red-slipped table-wares were common in the graves of the rural town, but unusual in those of the port. Raqqada’s graves likely included table-wares because of its proximity to the inland production zone of ARS “C” wares and because it lay between that zone and Hadrumetum, the port from which those products were probably exported155. Meanwhile, archaeological investigations at Leptiminus have revealed that its extensive potting complexes fulfilled most of its residents’ needs for ceramics, but did not produce ARS wares; nor did the Leptitani import ARS in quantity before the late Roman period156. It seems that fine table-wares were simply not a significant component of their collective apparatus in the second and third centuries, when cremations were taking place. Meanwhile, the unusual use of trough-like stone receptacles at Djebel Djelloud and Thaenae presumably relates to broader lapidary activities at those sites. These and other kinds of site-specific conditions would inevitably have had an effect upon the typical physical signature of local graves and, by extension, the details of the mortuary acts being performed there.

• 157 The Moroccan, Algerian, and Tunisian (High Tell) sites that were mentioned in footnotes 4 and 7, a (...)

64Clearly, the practical manifestations of grief and com­memoration were intensely local even when communities used components from a widely-recognized system of symbols and practices. Beyond a town’s status and its cultural influences, it is worth thinking about how local circumstances affected group behaviours, resulting in the formation of distinctive ritualized acts. Applying a comparable approach to other parts of Roman North Africa would likely reveal further examples of how common rituals were adapted in response to location-specific factors157. Carton and his fellow pioneers of cemetery exploration laid the foundation for cultural sub-sets by observing differences at the broad regional level. It is now possible to push those sub-sets to an even smaller, ­community-level scale and to explore the stimuli that caused them, as we become ever more aware of the many different forms that being “North African” took in antiquity.