The Maltese islands, famed for their Neolithic temples, are not well known for their Roman antiquities. After pioneering work by the polymath Sir Themistocles Zammit in the early decades of the 20th c.,Footnote ¹ the baton was picked up again by Anthony Bonanno, who singlehandedly has greatly advanced knowledge of Roman Malta over the past 50 years through copious influential papers in journals and conference proceedings, as well as in two books.Footnote ² Monographs by others have been dedicated to such subjects as the Roman economy of the Maltese islands, and Roman pottery from Malta and Gozo, both local products and imports.Footnote ³ Until now, however, there has been no single volume devoted exclusively to the Roman structures that have been excavated or recorded (and in some cases survive), nor to the disiecta membra of many others, which have in the past been largely ignored. A warm welcome, therefore, to this handsome volume, which is magnificently illustrated in color throughout, apart of course from historical images that have been taken from the archives. The color photographs are by Daniel Cilia, who has made major contributions to Maltese archaeology in other publications over the years.Footnote ⁴ His stunning aerial views in this book of the villa at Ta’ Kaċċatura (46) and of the temples at Ras ir-Raħeb (48) and Tas-Silġ (202–3) are a particular feast for the eye.

The intended readership is nowhere stated, but the book is clearly aimed at two quite distinct groups of readers. The first is the non-specialist one: introductory sections on Roman architecture, the domus, and the classical orders are clearly intended for that market. But the book will be welcomed also by scholars for its wealth of detail and the comprehensiveness of its contents, together with footnotes throughout and copious references. After introductory matter, the core of the book is contained in two chapters, “Building activity in Roman Malta” (47–85), with a largely chronological framework, and “Buildings and their decoration in Roman Malta” (87–221), which presents the relevant material in detail. A short chapter on “Materials and techniques” (223–43) and a brief conclusion (245–53) round off the book.

So what are the distinctive characteristics of Maltese architecture in the Roman period, when the islands administratively formed part of the province of Sicily? There are three main types of structure: urban buildings, rural villas, and temples; but the dossier of information on each (through no fault of the author) is not as complete as one would like, nor is it particularly plentiful. The full inventory, including very fragmentary remains and structures found long ago and no longer visible, totals just 59 sites, including five that are listed twice and two record inscriptions belonging to temples that have never been found (79–81).Footnote ⁵ Curiously the sanctuary site of Ras il-Wardija on a clifftop on the western tip of Gozo (see below) is not listed. The book is full of justifiable complaints about the inadequacy of the records of the early excavators and the non-publication of major research projects in the later 20th c., mainly by foreign teams.

Little is known of the topography of the two principal towns, Rabat on Malta (the ancient Melita) and Rabat (Victoria) on Gozo (Gaulus). The hint, on admittedly fragmentary evidence, that both seem to have had defensive walls is of interest, but, like Palermo's and those at S. Teresa on Pantelleria, these seem to me to be more likely of Punic date than Roman.Footnote ⁶ Pride of place goes to the domus in Rabat on Malta, with its fine Late Hellenistic mosaics (ca. 100 BCE?), including three figured emblemata (one of them in imitation of Sosos of Pergamon's famous composition of doves drinking: Plin. HN 36.184).Footnote ⁷ Borders include a pattern of stepped swastika meander and squares in perspective, and rich floral and fruit designs incorporating theater masks, of the type seen at Palermo and Pompeii as well as on Delos. Cardona (hereafter C.) does not speculate about who is likely to have laid them, but the three-dimensional variety of Greek wave pattern on a pavement with the satyr and nymph emblema is unknown in the Greek East and finds its only parallel at Tyndaris in Sicily.Footnote ⁸ This might suggest the presence in Malta of craftsmen from Sicily,Footnote ⁹ hardly surprising since the Maltese islands were part of the same province. The entablature of the peristyle at the Rabat domus is typical of Late Hellenistic Doric; this and the Sicilian version of Hellenistic Corinthian were the preferred choices in this period. How long the Rabat domus was occupied is unknown, but it was long enough for one of the pavements to be extensively (and crudely) patched. A new project by Heritage Malta and the University of South Florida on the houses adjacent, poorly excavated a century ago, might provide some welcome chronological exactitude, although some or all of these may be later if the conclusion that stone from the rich house was robbed for building them is correct.Footnote ¹⁰ A set of half a dozen marble statues found in the house, including one of Claudius and an exquisite one of a young woman here identified as his daughter Claudia Antonia,Footnote ¹¹ is taken as a striking sign of loyalty to the imperial house by a high-ranking resident; or do these statue fragments belong to an abandonment phase, transferred here from elsewhere, already in part broken up and awaiting the lime kiln?

Little is known in detail of the public buildings of either town, represented by broken pieces of architrave, or capitals, or column shafts, all lovingly recorded by the author, or else referred to in inscriptions, like a temple of Apollo at Rabat on Malta or another of Proserpina on the Mtarfa hill nearby.Footnote ¹² It is not known exactly when marble construction first reached the Maltese islands, but datable capitals and architraves in that material are not earlier than the 2nd c. CE, and the preponderance of Proconnesian among white marbles, estimated at 78 percent of the total, might point to the same conclusion.Footnote ¹³ The standard range of polychrome marbles from Tunisia, Greece, and Turkey have, however, (rather surprisingly) not so far been recognized in Maltese archaeology apart from a very few fragments (228). A few of the better-preserved limestone entablatures, including some known since the 17th c.,Footnote ¹⁴ are not readily paralleled anywhere in their particular style and combination of elements, and suggest the work of an imaginative local workshop or workshops. C. has also identified a genuinely original Maltese contribution in a variety of the Corinthian capital in marble, in which the central helices at the top are replaced by a long-stemmed fleuron with leaves on either side (144, 250).

Of rural sites (161–89), the most intelligible is the villa at Ramla Bay on the north coast of Gozo, in a gorgeous seaside location. It was first excavated over a century ago when there was no attempt to understand its phasing or overall chronology.Footnote ¹⁵ Of the residential part, only a half dozen rooms in the south and west wings were explored. The layout of the annexed bath-suite is more understandable: a changing room, square frigidarium with marble-slab floor, two tepidaria with a sudatorium opening off the first, and finally a small caldarium, are all clearly identifiable.Footnote ¹⁶ There are two small pools, one opening off the frigidarium and the other accessible from the caldarium. The villa's remans were backfilled to ensure its preservation, but the site is in increasing danger from marine erosion: an additional structure of ashlar blocks has been exposed in more recent years on the beach to the north of the known villa, and lidar surveying suggests as much as a third of this complex still lies buried.Footnote ¹⁷

Another villa, Ta’ Kaċċatura, also uncovered in the early 20th c. (1910–15), is still visible, although it is badly in need of a mise en valeur.Footnote ¹⁸ This is a small villa of a dozen or so rooms, but clear marks of a staircase on the wall of one room shows that the dwelling had, at least in part, an upper floor. There is a small central court with four columns per side. C. includes an excavation photo of three small blocks with cavetto and astragal, and one piece of gray marble has three letters of a text on it. Otherwise, there was nothing much in the way of flamboyant decoration (there are no mosaics, either here or at Ramla Bay); a very small set of baths has, however, recently been recognized at Ta’ Kaċċatura, too late for inclusion in C.'s book. No clear date for the villa is given by him, although he suggests there were “a number of building phases” (163). Ashby was convinced that some of the walls belonged to an earlier villa on the site and indicated that he thought these might be pre-Roman.Footnote ¹⁹ He did at least present some of the pottery that he found, and it included 4th/5th-c. CE African red slip ware, even though that was not recognized as such at the time.Footnote ²⁰ It may be (if this is not a post-abandonment garbage deposit) that the villa was still occupied then, even though the visible structure may well have been built much earlier in the Imperial period, with a possible predecessor before that.Footnote ²¹

The outstanding feature of Ta’ Kaċċatura is its almost perfectly preserved, partly rock-cut cistern situated alongside, just above the villa, with a flat roof of large ashlar blocks supported by huge freestanding stone pillars (163–66). It is not clear if it is contemporary with the villa or belongs to some earlier period, such as the Punic or later Hellenistic (4th/3rd c. BCE?). There are other, smaller examples of the same type of cistern elsewhere in Malta; one at Ta’ Ġawhar is illustrated by C. (212–13).Footnote ²² This stands beside a round tower, 15 m in diameter and with walls 3 m thick, one of a group of six such towers in the Maltese landscape (but they are not found on Gozo). David Trump, who excavated at Ta’ Ġawhar in 1960 and found evidence for destruction levels of the 3rd c. CE, was convinced that they were Roman,Footnote ²³ and C. follows him in that chronology (210). If that is right, it is reasonable to assume that the adjacent cistern is of Roman date too, and presumably the one at Ta’ Kaċċatura as well. The military function of the towers is now supported by the very recent discovery of two pieces of ring-mail body-armor among the 1960 finds.Footnote ²⁴ But there is no military reason why such powerful towers need have been erected on the island during the long centuries of the pax Romana, and ascribing them to the troubled times of the First Punic War perhaps makes more sense, especially since Ta’ Ġawhar has also produced pottery of Punic date, and the use of ring-mail is not incompatible with such a chronology.Footnote ²⁵ If it is of that date, the type of cistern with flat-slab roof may also belong to the 3rd c. BCE rather than (much) later, suggesting that it supported a farm at Ta' Kaċċatura of the same period, underneath the visible villa.Footnote ²⁶ This is another conundrum of Maltese archaeology that needs unpicking by future research.

A further old excavation is of rural baths with geometric mosaics at Għan Tuffieħa, uncovered in 1929–30 but sadly backfilled in 2021 to prevent further degradation. C. tentatively suggests (186) that “the idea of this bathing complex serving a private residence may not be entirely ruled out,” but stand-alone rural baths do not make sense without a villa nearby to serve.Footnote ²⁷ In fact, the long corridor with a series of small mosaic-paved rooms and a large (ornamental?) pool adjacent to the south (away from the baths) may themselves be part of the villa to which the known bath-suite was attached. The suggestion that these rooms formed (a problematic) part of the baths, and were “used as changing rooms, sleeping quarters for visitors, or for the gratification of carnal lusts” (187) is unnecessary if they formed part of the villa proper. The date proposed by Zammit, the original excavator, of “the late first to the early second century AD,” is noted without comment,Footnote ²⁸ but the geometric polychrome mosaics look to me to be later (ca. 150/225?).Footnote ²⁹ Recent work, so far unpublished, may tell us more; certainly, this is a site that would repay extensive modern excavation.

The multi-phase plan of another villa at San Pawl Milqi, uncovered by an Italian mission, is more or less incomprehensible (177), but C. does his best to guide us through the excavation results, so far as they are known, with full discussion of the architectural fragments that the site has yielded, mostly in all likelihood Late Hellenistic.Footnote ³⁰ That seems to have been the phase when the site prospered most; in the later Imperial period, three olive presses were installed in a large courtyard, while what had once been residential rooms with frescoes of the 2nd c. BCE show no sign of later redecoration.Footnote ³¹ Żejtun is yet another villa where much more work is needed to yield a coherent plan and more precise dating evidence.Footnote ³² C. presents some interesting pieces of entablature from there with overhanging rounded cavetto cornices in the Punic tradition,Footnote ³³ perhaps demonstrating their continued use into the Roman period (late 2nd/1st c. BCE?), when the villa appears to have been built. The recent suggestion that pottery evidence points to the likely existence of a Punic sanctuary underlying the villa, however, leaves open the possibility that the cornices come from an earlier building.Footnote ³⁴ Another major point of interest at Żejtun is a further Roman olive-oil processing plant, of the kind for which there is considerable evidence all over the Maltese islands.Footnote ³⁵

Three temples have been uncovered in Malta and Gozo. Two of them (191) are dramatically situated on cliff-tops, Ras ir-Raħeb on Malta and Ras il-Wardija on Gozo.Footnote ³⁶ Both were poorly excavated, but they appear to be of the open-plan Punic type built on the flat, common in Roman north Africa, without a hint of raised podia or other Roman influence.Footnote ³⁷ The chronology, however, is uncertain: they probably belong to the Republican period when Punic influence in Malta (as in western Sicily too) was still strong. Pottery from the Italian mission at Ras il-Wardija, which stretches from the 3rd/2nd c. BCE to the 1st/2nd c. CE, may support that conclusion. At Ras ir-Raħeb, a coin of Constantius II (337–61 CE) documents a Late Roman visitor, but whether the temple was in ruins by then is unclear. A case can and has been made for it to be identified as the site of the temple of Hercules mentioned by Ptolemy, although available evidence is slight.Footnote ³⁸ Both sites would repay modern scientific excavation.

The third temple, Tas-Silġ on Malta (191–207), is a remarkable site, situated on a hilltop with a magnificent sweeping panorama over the eastern part of the island.Footnote ³⁹ One of two adjacent megalithic temples of late Neolithic date (ca. 3000/2500 BCE) was reused anew as a shrine in Phoenician times (8th/7th c. BCE). Either then or later, fresh walls were provided in front of it, together with a new entrance. Then in Late Hellenistic times, when politically Malta was under Roman control, the Phoenician/Punic forecourt was demolished, and in its place was built an open paved piazza on a grand scale, with spacious porticoes surrounding it on three sides. The latter, paved in opus signinum with white tessera inlay, were probably built in the late 2nd or early 1st c. BCE.Footnote ⁴⁰ The columns, all of limestone, were mostly in the Doric order, but Sicilian Corinthian was used as well, apparently on the east side near the temple entrance.Footnote ⁴¹ The paving of the court respected some largely demolished Phoenician walling, being laid up to but not over it.Footnote ⁴² This is truly a place of memory, with later builders aware of, and respectful of, preceding shrines of Phoenician and Neolithic date. Numerous inscribed votive potsherds indicate worship of Astarte here. Cicero mentions the sanctuary in the same breath as Hera's on Samos for their antiquity and sanctity,Footnote ⁴³ Astarte having now been transformed into Juno Regina through interpretatio Romana.Footnote ⁴⁴ It is the veiled head of Astarte/Juno that features on Melita's coinage from the end of the 3rd c. to the 1st c. BCE, almost certainly a reference to this shrine.Footnote ⁴⁵ Ongoing excavation at Tas-Silġ as well as study of unpublished earlier records continues to yield fresh information.Footnote ⁴⁶

Of particular interest, but mentioned only in passing by C. (205), is an apparent restoration in Augustan times or later which deserved fuller treatment. A marble slab with Egyptianizing relief decoration, now in a private collection but taken from Tas-Silġ, shows side by side a snake (probably with a human head), a stylized atef-crown (in the center), and a large uraeus on the right, also stylized.Footnote ⁴⁷ The slab is broken off here, but may have been balanced by a further atef-crown and a further snake. Parallels elsewhere suggest that the snake with human head symbolizes Serapis as the Agathodaimon (“the good spirit”), whereas the atef-crown, originally worn by Osiris, came to be associated with Isis. The uraeus, originally linked with the cobra-goddess Wadjet, was used ubiquitously as the symbol of both royal and divine power in ancient Egypt. That the relief came from Tas-Silġ is demonstrated by the discovery of small fragments with identical motifs in an identical arrangement during the Italian excavations of 1963–70, at various points on the site, indicating that one of the buildings there was once decorated with an architectural marble frieze with a repeat pattern featuring these elements. We know that there was apparently some Augustan building work at Tas-Silġ, because a fragmentary Latin inscription, probably the oldest in that language in Malta, honors a certain ]EMPRO] / ]TRATIN[ / [A]VG IMP[. This must refer to L. Sempronius Atratinus, the suffect consul of 34 BCE and augur, whose award of the title imp(erator) for his victories in Africa in 21 BCE means that the Tas Silġ inscription can be ascribed to a date between then and Atratinus's death in 7 CE.Footnote ⁴⁸ Bonzano has suggested that the building work carried out by him included the erection of the Egyptianizing marble frieze, which she therefore sees as Augustan, and she suggests that it might have adorned at eye-level the walls of one or other of the two small square side rooms which flank the central entrance to the main temple.Footnote ⁴⁹ Even if the location suggested for the frieze is plausible, if unprovable, the Augustan date is more uncertain. If correct, it would be the earliest recorded instance (by over a century) of marble used in construction work anywhere in the Maltese islands.Footnote ⁵⁰ More significantly, the use of the so-called running drill, clearly visible on the relief, is more appropriate to work at the end of the 1st c. CE and in the first half of the 2nd c. (and later) rather than earlier.Footnote ⁵¹ So while the Atratinus inscription appears to indicate building activity of some sort at Tas-Silġ under Augustus, there might have been further work later, a little before or somewhat after ca. 100 CE, which included the Egyptianizing frieze.Footnote ⁵²

How long the temple remained in use is unclear, but African red slip pottery and lamps show continuing activity into the 4th and 5th c. Whether that activity was still religious in nature is hard to gauge: there seems to have been no new building work in the Late Roman period. Perhaps the sanctuary's very antiquity removed any desire to alter such a venerated place, running repairs apart. Later still, a small church was inserted, probably in the 6th c., with a baptismal font set in the middle of the Neolithic structure. Remarkably, even if with interruptions, Tas-Silġ remained a sacred site for millennia.

One striking aspect of C.'s book is the absence of any structures from the Late Roman period. There is pottery of this date from sites like Tas-Silġ and Ta’ Kaċċatura (as noted above) but no clear structural evidence of that period, and there appear to be no Late Roman villas. Yet the Maltese islands were not in decline then: there are a handful of Late Roman inscriptions, both public and private, and the rich catacomb evidence, not only in an urban setting (at Rabat) but also in scattered rural locations all over both islands (but especially on Malta), show a continuing, dense occupation in Late Antique times.Footnote ⁵³ Field survey, not fully developed as yet in the islands, may well change this picture in the future,Footnote ⁵⁴ but excavation of key Late Roman sites, both urban and rural, is an urgent desideratum. This is a challenge (along with the need to fill several other lacunae noted above) that, one hopes, present and future generations of Maltese archaeologists will be able to take on, funding permitting.Footnote ⁵⁵ The lack of a secure chronology for so many of the Hellenistic and Roman buildings in the Maltese islands is a long-standing and on-going problem.

At times, more contextualization of the material presented in C.'s book would have been useful; often the focus is too firmly fixed on the Maltese islands, without looking at the wider world.Footnote ⁵⁶ Ta’ Kaċċatura's cistern, for example, has close structural parallels in well-preserved examples of the same type near Scordia in southeast Sicily, as well as at Syracuse and elsewhere. All are likely to be pre-Imperial,Footnote ⁵⁷ strengthening the conclusion that the Maltese examples of the same type, with flat, large-slab roofs, all also belong to the 3rd/1st c. BCE rather than later. The satyr in telamon pose from Ramla Bay recalls one in a similarly private context in a Late Hellenistic house at Centuripe, also in eastern Sicily.Footnote ⁵⁸ The bibliography is full, but additional items in a few cases would have made it even more helpful.Footnote ⁵⁹ The great strength, however, of C.'s handsome book is that, in presenting a wealth of informative detail about Roman structures and their decoration in the Maltese archipelago, he has rendered an important service to scholarship by bringing this little known material to a wider audience. For that he warmly deserves both thanks and praise.