Dans le département des Hautes-Pyrénées, le site de Saint-Lézer a livré plusieurs vestiges d'une vaste enceinte longue d'env. 950 m, construite à la fin de l'époque romaine. Certains de ces fragments permettent d'entrevoir quelques aspects les plus originaux de l'architecture des parties hautes du monument et de tenter de restituer, au-dessus du chemin de ronde, un parapet et ses traverses. Ce dispositif largement attesté par ailleurs dans le monde antique gréco-romain et qui a été récemment reconnu à Saint-Bertrand-de-Commiges semble donc avoir été également adopté par les bâtisseurs de l'antiquité tardive du piémont pyrénéen.

Le site de Saint-Lézer se trouve à 18 km au nord de Tarbes, dans le département des Hautes-Pyrénées.

Le site est établi au sommet d'une colline argileuse qui porte le nom évocateur de Castelbieilh (ou Castelvieilh: le vieux château). En forme d'éperon d'axe Nord-Sud, barré au Sud, il est occupé dès la fin de l'Age du Bronze et durant tout l'Âge du Fer. À l'époque romaine, la colline est dotée d'une vaste enceinte maçonnée longue d'environ 950 m dont il ne subsiste aujourd'hui que des fragments épars jonchant ses versants. Par la suite, approximativement un quart du site est réaménagé au Moyen-Âge: l'on y implante alors une motte castrale désignée sous le nom de Tuco.

L'hippodrome de Tyr est l'un des mieux conservès du monde antique. Les malheurs du temps et la mort de l'èmir Maurice Chèhab, qui en avait dirigè la fouille, n'en ont pas encore permis la publication scientifique. Des inscriptions y ont ètè trouvèes, dont l'èmir Chèhab m'avait confiè le relevè et la publication. Les èvènements du Liban ont interrompu ces travaux. Les inscriptions de l'hippodrome sont encore inèdites; des rèvisions et quelques complèments d'information s'imposent, peut-être difficiles à effectuer ou à obtenir, car il est à craindre que plus d'un document n'ait ètè perdu ou dètèriorè durant ces annèes dramatiques. J'ai toutefois souvent eu l'occasion de faire mention des plus importantes de ces inscriptions. La bienveillante autorisation de la Direction Gènèrale des Antiquitès du Liban me permet d'en donner ici les textes eux-mêmes. Cette première prèsentation du dossier anticipe une èdition plus critique et un commentaire plus complet, qui devraient prendre place dans le deuxième volume, en prèparation, consacrè aux inscriptions dècouvertes dans les fouilles de Tyr.

Ces inscriptions, peu nombreuses, forment trois groupes. L'un comprend trois mosaïques de pavement, conservèes en place, qui ont quelque peu souffert des èvènements. Un deuxième groupe est celui d'inscriptions sur fragments de colonnes, dont il n'est pas possible de dèterminer avec toute la prècision ou la certitude souhaitable les lieux de dècouverts, et dont je ne sais ce qu'elles sont devenues; ces inscriptions indiquaient les emplacements rèservès, sur les gradins, aux deux grandes factions, à d'autres groupes ou à certains personnages. Un troisième groupe est celui d'inscriptions peintes qui, près des accès et sous les galeries de l'hippodrome, marquaient les emplacements rèservès aux commerçants; dèjà difficiles à dèchiffrer avant 1975, ces inscriptions sont aujourd'hui apparemment effacèes; je n'ai pu en retrouver aucune lors de brèves visites en 1997 et 1998.

Following the rescue excavation of part of the Roman cemetery and a Roman mausoleum (Site 200) on Dhahret Slama, on the SE edge of the ancient town of Leptiminus on the east coast of Tunisia, the continued pressure of modern building led to the discovery of a remarkably well preserved underground burial-complex, as well a considerable extension of the above-ground cemetery which we had previously sampled. The underground burial-complex includes a number of tunnels and chambers, undoubtedly part of a much larger area of catacombs. These underground structures are cut into the sandstone and limestone ridge known as the Formation Rejiche which runs along the coast.3 This geological formation was suitable for the cutting of Punic-type chambers, of which a number have been located on this side of Leptiminus, and for catacombs, of which other examples have been identified in the same region, at Sousse, Sullecthum, and probably Thapsus.

The existence of underground structures became clear following enquiries made of local residents, who spoke of substructures and vaults beneath their houses. Residents noted that after heavy rains water vanished rapidly into the ground; we now realize that it disappears into the underground voids of built or carved structures.

My article on Roman wall-painting and social significance (JRA 14, 33-56) provoked two immediate reactions. I would like to show that both scholars misinterpreted my argument and intentions.

Bergmarm's views are compatible in all respects with mine expressed in my article. In two introductory sections (33-42) I drew attention to some major misunderstandings and misrepresentations of the chronological model of Roman wall-painting developed by H. G. Beyen by scholars who have recently focused on paintings in their spatial and social context. B. agrees “that we lose much by neglecting the contributions of 19th- and 20th-c. scholars”. Presumably she took offence at what she calls my “ad hominem recriminations” In order to state my argument clearly, I quoted from several recent studies, mentioning their authors and places of publication. I pointed out as straightforwardly as possible that some of their views expressed are demonstrably untenable or, in some cases, not as innovative as suggested. This was done not to incriminate the authors but to bring some clarity in a discussion which risks becoming embroiled. Unfortunately, Bergmann does not question the validity of any of my specific “recriminations”, limiting herself to some general observations.

She defines as my “main concern (…) that current scholars do not find Beyen's methods (and [my] own) worth emulating”, and prefer to employ other approaches “looking at paintings within larger ensembles and spatial networks” which “might shed some light on the rooms they embellished”. Nothing is less true. I agree with her that scholars understandably “find the search for and understanding of these relationships as appealing as attempting to validate Stufe C of Phase I of Beyen's Second Style” — I would say “far more appealing” That is the reason why I relegated that specific validation to an Appendix (53-56), in spite of the perspectives it opens up for social history (56), and devoted my main section (42-53; discussed neither by Bergmann or Hallett) to the study of wall-painting in its spatial and social dimensions (cf. also JRA 9 [1996] esp. 366-74). Moreover, I stressed that “the basic concepts underlying this [new] line of research are sound and stimulating” (34); that “any comprehensive account of Roman wall-painting should transcend the level of specialists' chronological ‘finger-exercises’ and encompass the synchronic dimension”; that “actually, this is a communis opinio by now, and rightly so” (36); and that the contributions of social historians able “to introduce more appropriate and refined concepts and vocabulary … can only be welcomed” (42). My real concern in sections 1 and 2 was to counter the current trend to ‘throw out the baby with the bath-water’, i.e., the diachronic with the synchronic dimension, since “a sound functional model of Roman wall-painting cannot dispense with careful attention to diachronic developments in both form and style” (36).

The comments on my article (JRA 13,104-32) by Blackman and Peleg, which are much appreciated, reflect prevailing discussions on Roman hydraulic engineering practices as well as on the relevant paragraphs of Vitruvius (8.6). Blackman focuses mainly on technical aspects, while Peleg aims at archaeological arguments.

Blackman starts with a discussion of the term ‘siphon’, which he regrets being used as it could lead to misunderstanding: ‘siphon’ would represent a “real” siphon, not an “inverted” siphon, suggesting that I must have meant that there was a real siphon at Aspendos. However, in archaeology “siphon” is generally accepted to represent an inverted siphon. There is no misunderstanding: at Aspendos we have an inverted siphon, and since real siphons are not known in classical aqueduct systems, there is no objection to using ‘siphon’ and ‘inverted siphon’ for one and the same notion. Subsequently he states: “The presence of air pockets and so on is almost irrelevant to operations [of siphons] if no point in the system lies above the Hydraulic Gradient Line; were it not so, no garden hose would work reliably” This is a misconception. It is only due to the fact that our garden hoses are connected to supplies with elevated pressures that we get water from them. If we would connect the hose to a low-pressure source (e.g., to a rainwater container standing at ground level), we have to straighten out the coils before water will emerge. If the hose is coiled up, for example, on a wall support but positioned below the free water surface in the container to which it is connected, while we are holding the free end somewhere near the ground, we have nothing but an inverted siphon with high points. If air pockets are irrelevant, water should come out, but it does not if air is in the hose. For problems associated with air pockets in gravity-driven closed conduit systems (which classical siphons must be considered to be), see G. Corcos, Air in water pipes (1989) and H. T. Falvey, Air-water flow in hydraulic structures (1980).

Since most coin hoards are found without any archaeological context, even when excavation after discovery is possible, prompt publication of a large hoard of denarii with its context is so unusual an event that a few extra comments and questions may be allowed.

I have doubts about the comparison with the site at Frocester (Glos.). The finds there suggested the remains of an iron-bound chest, and it might be that that was a strong box. If so, it was a permanent feature but was found empty. The Shapwick hoard was deposited in a scoop in the ground and was presumably a temporary hiding-place for a sum of money which might be roughly £150,000 to £250,000 in modern terms. The hiding of the hoard would, in a sense, be more ‘normal’, and certainly more safe, if it were done within the walls of a derelict building rather than an active farm. From the archaeological details this seems to be a possibility.