[Notes de géographie historique], reprinted from Bulletin de l’Institut Français d’Archéologie Orientale, Vol. 15, pp. 69 ff.) locates the frontier between Syria and the Ḥeǧâz somewhere below the oasis of al-ʻEla’. This essay is brilliant as regards its equipment of learning, but on a more detailed investigation of the sources quoted it is obvious that the author has arrived at results which are scientifically inaccurate. He does not distinguish the physiographical from the administrative frontiers, pays no attention to classical reports, and does not interpret the Arabic authors faithfully.
From the oldest times the southern slope of the aš-Šera’ range formed the frontier between the settlers and the nomads. This is clear both from the Bible and from the Assyrian records. The classical authors took over these native frontiers and gave the separate regions new names; the Arabic authors changed nothing in this natural limitation of frontiers except the names. The southern frontier of Syria coincided with the southern frontier of Arabia Petraea (later of Palestina Tertia, or Palestina Salutaris) which led from the Red Sea across the mountains of Iram (Ramm) to the southern slope of the aš-Šera’ range.
On the border of Arabia the Roman imperium had a twofold limes, an internal and an external one. The internal limes followed the edge of the settled and cultivated territory, while the external one led through the frontiers of the territory of the nomads, to whom the Romans paid annual salaria. The internal limes was permanent and therefore strongly fortified; the external limes, on the other hand, was not fixed; it contained no permanent Roman garrisons and therefore no fortified camps. North of Iram (Ramm) and in the aš-Šera’ range there is an abundance of remains of Roman strongholds and fortified camps which would seem to confirm the information given in the Notitia dignitatum. South of the aš-Šera’ range, however, I did not find a single Roman remain; nevertheless the settlements of Madiama (al-Bedʻ), Onne (ʻAjnûna), Bada’, and especially Hegra (al-Ḥeǧr) were known to the classical writers, who would certainly have mentioned it had there been Roman garrisons in them, or if the remains of Roman encampmen had been preserved there. But we search in vain for such references in the classical writers and for Roman camps in the northern Ḥeǧâz. From this it is clear that both Madiama and the other oases mentioned above were situated, as Eusebius correctly states, trans Arabiam, and that they did not belong to the political administration of the province of Palestina Tertia, or Palestina Salutaris. It does not follow from this that they were not situated in an area enclosed by the external limes. This is obvious from the inscription at Ṛwâfa, where the tribe of the Thamudenoi built a temple in honor of the Emperors Marcus Aurelius Antoninus and Lucius Aurelius Verus (see above, p. 185). The Thamudenoi bordered on what was then Arabia Petraea and later Palestina Salutaris, whence they received salaria for acknowledging the supremacy of the Roman and Byzantine emperors into whose service they let themselves be hired. Indeed, they were even appointed as Roman or Byzantine officials. This is a usage for which there is evidence in the Assyrian records, and it has continued until now, but we cannot infer from it that the territory of the Thamudenoi, or of the Ǧuḏâm tribe after them, formed a permanent part of the Roman Empire and belonged to Syria. As soon as a chief, who was at the same time a Roman official