This is a high-quality definitive publication of two of the slightly misnamed ‘villas romaines’ or rich town houses situated within the parc archéologique on the Odeon Hill in the NE sector of Roman Carthage – the Maison de la Rotonde and the Maison du Cryptoporticus. The exploration of houses in this area was initiated at the start of the twentieth century by Paul Gauckler and continued through the 1920s and 1930s, during which time the present houses were exposed but not studied or published. The present project was launched within a Franco-Tunisian programme of research on African mosaics.
La Rotonde was excavated extensively by the present team and they organise the results in five phases (états) spanning the first to seventh centuries a.d. For the house of the Cryptoporticus, more limited sondages were carried out, together with architectural analysis, so the results are less substantial and mainly relate to the peristyle house which was constructed probably in the second century and in use until the seventh century. For both houses there are extended studies of the floor decorations, principally mosaics, with opus sectile in the House of the Cryptoporticus, and of the painted wall plaster. The first volume concludes with synthèses and an appendix on the mise en valeur of the two houses. The second volume presents finds reports for the two sites organised according to the stratigraphic framework or, in the second case, by sondages. Synthetic studies follow on the coins, pottery and glass.
The organisation of the information and its clarity deserve warm appreciation. The first volume in particular contains a mass of sometimes complex information: throughout, the arrangement is logical, the text is clear and concise, the illustrations are beside the relevant parts of the text and of high quality. The fold-out plans at the start of Volume 1, allowing the detailed text to be read with constant reference to a plan, are an effective part of the presentation and the whole-figure scales (1:200, 1:250, etc.) are particularly appreciated. The difficulty may, nevertheless, be mentioned, of using multi-phase plans with the phase under consideration outlined in colour. All the evidence should certainly be shown at some point, but if, in the interpretative phase plans, only evidence for the phase under consideration is given, not only is communication clearer, but they clarify any ambiguities in the phasing.
The specialist studies are thorough and well-informed; they provide convincing support for the wider picture of the sites and are valuable additions to these specialist areas. I shall focus principally on the contributions made to our general understanding of Roman, Vandal and Byzantine Carthage. A non-exhaustive list would include: chronological development of two insulae in the NE part of the Roman city; background stories and sequences of two grand town houses; their architectural character and decoration; new information on Roman and later urbanism at Carthage.
A first question is when this area first became urbanised using the gridded street system of Roman Carthage. The earliest remains on the site of the Maison de la Rotonde were eight burials of Punic date (later fourth to the second half of the second century b.c.). Several burials seem to have been disturbed in Late Antiquity or later, and one appears to have been partly reinterred in an urn perhaps on the occasion of the early fifth-century construction of the Rotonde; another was a stone cist excavated previously. These burials were aligned E–W, on the same alignment as the Roman buildings. Tertullian's reference to burials being disturbed by the construction of the Odeon c. a.d. 200 (De Res. mort. 42.8) is cited, together with Bénichou-Safar's study of Punic burials. Phase I Roman remains were confined to the E half of the insula between cardines IX and X, those on higher ground further W being cut away by later terracing. There appeared to be two approximately equal-sized properties extending for c. 18 m on the street frontage, ranging in date from pre-Flavian to mid-second century. The building plans were fragmentary, but construction was in handsome-looking opus Africanum masonry, with floors of opus signinum. Each property had a round bottle-shaped cistern and towards the rear of that to the S were two cement-lined tanks and a discharge basin. These were interpreted as a garum-making installation, but a less sea-related and smelly use is surely more plausible on this hill 0.5 km or more from the coast within what looks like residential/commercial property. The excavators emphasised the lack of pisé constructions such as were found at a similar date in the Circular Harbour North Side excavations. In the next insula to the S, where the Cryptoporticus was located, sondages also suggested the presence of structures aligned with the grid from the mid-first century a.d.
In phase II on the Rotonde site, dated late-second to early-fifth century, uniform constructions extended over much of the width of the insula. The notional width of an insula is 35.3 m: the Phase 1 structures extended for c. 30.6 m, becoming c. 33.8 m with a Phase 2 rebuild towards the W. These were of solid masonry construction, with many mosaic floors, including polychrome floral and geometric designs in a Phase II (probably fourth-century) rebuilding in the W part of the southerly property. The broad interpretation is comfortable domestic, but it is frustrating that with so much of the plans and with details of floors, cisterns, drains, etc. it was not possible to go further than a division into two broad ‘lots’. Were these blocks in multiple occupancy as mentioned in Chapter 13, perhaps of more than one storey? There seems to be no clear-cut indication of this, such as staircases, but the southern property is progressively cut westwards into the higher ground towards cardo IX, suggesting basements. A long corridor in the S property could indicate adjacent apartments and the N property seems particularly well endowed with cisterns: one beside cardo X had its base at the same level as the floors of ground-floor rooms, which might indicate upper floor levels. An extended wrestle with the well-recorded structural details could yield more on building organisation and functions.
Phase III, first quarter of the fifth century, was the point at which the area was rebuilt unifying the two former properties and creating a building complex extending E–W for c. 38.5 m, so notionally encroaching on the streets bounding the insula to W and E. The new building included a range at least c. 40 m long by 8 m along the W side of cardo X, backed (to the W) by a large terraced court, apsidal-ended room and rotunda of the Rotonde. Although in outline the plan is suggestive of a grand residence, there were difficulties of interpretation. Within the court and rooms to the W, no floors survive because of a later cutting down of the level. An apparent lack of finish throughout and the discovery of an amphora base containing 301 coins (up to the early fifth century), set within a wall of a room at the S end of the courtyard, led to the suggestion that this was an unfinished building project, perhaps suspended on the occasion of the Vandal capture of Carthage (a.d. 439). This interpretation acquires additional rhetorical emphasis in the synthèse, but is unconvincing. The room whose wall contained the coins was secondary to the main phase III layout (‘murs postérieurs’), so casting doubt on the ‘unfinished project’ interpretation of that layout. As Brenot says in the coin study in Volume 2, the latest possible date of a coin actually found is early-fifth century, notionally extending to 423. By manipulation of percentage incidences in comparison with other broadly contemporary coin groups and an argument that the pot was disturbed, with its latest pieces being removed, he suggests a deposition date of 425–435; from this a small jump can be made to 439. He also assesses the value of these coins as equating to 100 days' food for a prisoner or a dozen loaves of bread: as he says, not a fortune. Is it really necessary, then, to introduce a big political event to explain the existence of these coins, set in a pot in a wall, designed for easy retrieval, so presumably a store of small change? On the overall functioning of the complex at this stage, the presumed two-storey range beside cardo X is shown in a section drawing to have had floors below the level of the courtyard. Access from the street to the house may be represented by Room 3, with a substantial entrance 2.65 m wide flanked by piers, but it would still be necessary to climb to the courtyard level on the W side of this (where no evidence appears to survive): a functional/commercial use for most of this streetside range seems likely.
In phase IV there was an adjustment of the building to a lower courtyard level, with a fully rebuilt peristyle of 24 columns, making it one of the grandest houses in Carthage. The layout from phase III seems to have remained, although the S side of the peristyle was built up with rooms containing mosaic floors, and a N limit was set beside an alley leading W from cardo X, situated above the N part of the phase III building. Dating evidence for this phase seems not earlier than the end of the fifth /earliest sixth century. The excavators believe that phase III was a ‘chantier … inachevé’ (p. 136) and seem inclined to push the dating as early as they can, to the earliest sixth century. In the summary this becomes ‘une vaste demeure réalisée à l'apogée du royaume de l’époque vandale’ (p. 326). This seems problematic: if we accept the Vandal-period dating (pre-533), it does not explain how the phase III project managed to remain dormant for a century before being completed. More plausible, surely, on general grounds and in view of the secondary structure to phase III with the coins, is that this was the rebuilding, after a century of use, of a house completed in the early-fifth century. Should the rebuilding date then be before or after 533? I would see it as paralleled by rebuildings from other parts of the city and typical of the period after the Belisarian recapture of Carthage, but I can see that this is as historically subjective a view as placing it in the apogée du royaume vandale. Surely now, with all the archaeological study of late-antique Carthage, we can step up a level: can the dating evidence for a latest fifth/early sixth-century terminus post quem not just be presented on its own merits, with the appropriate qualifications? This would acknowledge that the relationship with 533 is ambiguous, but also that there is a cumulative process, in which we progressively learn more about the construction of archaeological deposits around this time and more about how Carthage might have looked either side of 533: we do not have to be historically dogmatic until the evidence is clear. Later in the Byzantine period (phase V) the property looks to have been divided perhaps into three (rooms S of the peristyle, rooms relating to S and N parts of the peristyle, with an added room to the NW). The latest coin in a disused deposit in cistern C4 was of Heraclius (612–41). Two inhumations in stone cists typical of the final phase of activity were found in the apsidal-ended hall and another in the rotunda. Pottery was suggested to stretch to the end of the seventh/early-eighth century.
At the House of the Cryptoporticus site, after the developments in the first century, the grand peristyle house was constructed not earlier than the turn of the first and second centuries. A major transformation, including the construction of the cryptoporticus, followed in the early-fifth century, giving the building a form which it basically retained until the end of antiquity. With reference to the street grid question, this building at its maximum extent, with the cryptoporticus on its W side, extended E–W for c. 44.25 m, about 8.75 m more than the notional width of an insula. Broise explains how this resulted from expansion on both the E and W sides and that provision was made to give cardo IX a width of 3.9 m by making it partly overlap the cryptoporticus.
Broise's essay on the evolution of the insulae examined within the urban context (Chapter 14) is persuasive on the reduction of street widths in the later Roman imperial period and perhaps in suggesting that 3.9 m was a legally accepted width. But, although it refers to the secondary addition of insulae in the NE part of Carthage, it seems curiously to miss the major point about when this part of the city was formally urbanised and how and when the street grid was applied to it: there seems to be an assumption that it was theoretically urbanised from the Augustan period according to the frameworks laid out by Saumagne (1924, 1928 and 1931) and Davin (1930). Thirty years ago I collected information suggesting that there were Roman-date burials within the urban area, possibly on the site of the Odeon (from a slightly unsatisfactory report by Gauckler, Revue archéologique [1902], 398) and more convincingly in the Dermech-Douimès area further E (Cahiers des Études Anciennes 17 [1985]). A short way NW of the present site, decumanus VIN at its junction with cardo VIE was shown in the Canadian excavations to overlay an obliquely-aligned Roman building of the second or third century, suggesting that the grid layout had only been applied after that date. This, with other indications, pointed to a less extensive, less theoretically-constructed, initial urbanisation than Saumagne in particular had suggested. So how does the present information bear on that? There are indications of the Roman alignment in this area already at the time of the Punic burials. Roman buildings on this alignment are convincingly shown from the middle of the first century a.d. A regular Saumagne-Davin insula of 35.3 m width is never evident for the site of the Rotonde, since as noted, the pre-fifth-century buildings are narrower than that and the later ones wider. (A small stretch of notionally axial N–S wall, M61, in phase 1 has evidently influenced Broise's assumption of its existence.) With a restored early wall on the E, the Cryptoportique can be made to fit across an insula in its pre-fifth-century state, but it was plainly wider thereafter. So, to be a devil's advocate, the urban alignment is demonstrated, but not the grid in any precise way, until its late Roman version with narrower streets. This could fit with the evidence from the Canadian excavations and with Broise's arguments for the ‘legality’ of narrower streets. Should we not then just accept what this suggests: a smaller Augustan grid, with expansion until an organised later imperial revision?
It will, I hope, be apparent that these volumes have high-quality contents, well presented so that discussion can be developed. I will, however, return to my one fundamental criticism, in deliberately tough terms. Why does such good archaeology as these volumes contain have to be dogged by the superficial historical analyses, particularly of the synthèses, but also present to some degree throughout? What we know archaeologically about fourth–seventh century Carthage is far richer than the limited set of inferences which can be made from textual references. Ultimately of course we make a synthesis, because this is a historical period, but there is a more interesting way of doing it than trying to impose political history events on every building sequence and set of finds. That said, these two volumes are an admirable addition to the literature on Carthage.
