• 1 Butler 1913 (architecture); Littmann, Magie Jr. and Stuart 1913 (Greek and Latin epigraphy).
• 2 De Vries 1998, with Villeneuve 2001 and Kennedy 2004, p. 86-91.
• 3 Bader 2009 (I. Jordanie 5/1).

2The material studied here is part of the larger batch of Greek and Latin inscriptions that have been recently discovered in the Jordanian Ḥawrān, thanks to a close partnership between the Department of Antiquities of Jordan and its office in al-Mafraq, the Umm al-Jimāl Project, and the French-Jordanian team of the ‘Inscriptions de la Jordanie’ (‘Greek and Latin Inscriptions in Jordan’). Of special interest are a Greek dedication to Zeus Keraunios by a beneficiarius in the service of the provincial governor under the Severi and a Latin inscription dealing with the building of a tower under the authority of Flavius Maximinus, dux of Arabia, in the fourth century ce. The purpose of this article is to present both documents for the first time and to contrast them with the archaeological and epigraphic material relating to the presence of the Roman army at Umm al-Jimāl from the creation of Provincia Arabia (106 ce) to the reign of Justinian (527-565 ce). In this respect, our study should complement the work of Howard Crosby Butler and the Princeton University archaeological expeditions from 1904-1905 and 1909,1 the results of the Umm al-Jimāl Project initiated by the late Bert de Vries in 1972,2 and the corpus of Greek and Latin inscriptions from northeast Jordan published by Nabil Bader in 2009.3

• 4 IGLS 13/1, 9010, 9011, 9015, 9107.
• 5 Hollard 2004.

5The name of Zeus Keraunios (i.e., “thunder-wielding”) could fit many deities of the Roman Near East, including Baalshamin, the Lord of Heavens. Given the military context of the Umm al-Jimāl dedication, Zeus was here perhaps none other than Jupiter Hammon, the tutelary god of the third legion Cyrenaica, whose main camp was at Bostra and whose command was entrusted to the governors of Arabia since the reign of Hadrian. Jupiter Hammon had his own temple in the capital city of the province.4 His expected status as a lightning god was emphasized in the coinage minted at Bostra in the name of the third legion under Antoninus Pius, since one of the known types shows the legionary eagle (aquila) holding the thunderbolt (fig. 4).5

• 6 PIR² F 227; Sartre 1982, p. 88, no. 25, with IGLS 13/1, 9072-9074; Thomasson 2009, p. 141, no. 35: (...)
• 7 Christol and Drew-Bear 2014, p. 301-306; Eck 2022.
• 8 I. Jordanie 5/1, 124, from the Commodus Gate; cf. IGLS 14, 412, for the date, maybe 177-178 ce. On (...)

6The dedication can be dated approximately thanks to the mention of Q. Flavius Balbus, whose governorship in Arabia is thought to have taken place between ca. 213 and 221 ce,6 and perhaps more precisely in 216-217 ce, under Caracalla.7 This document adds nothing to the discussion of the exact period of Balbus’ legatio. On the other hand, it furthers our knowledge of the role of Umm al-Jimāl as a place of settlement for the military. Until now, it was already known that the Roman army had been permanently established on the site at the latest since the time when the governor of Arabia had supervised the construction of a uallum, i.e., a mere palisade combined with an earth bank or a more resistant masonry rampart, in 177-180 ce.8 As will be seen later, it is also questionable whether the Latin inscription of 212/213 ce which deals with the construction of a new fort (castellum nouum) and which was reused at Qaṣr al-Ḥallabāt also referred to the fortifications of Umm al-Jimāl. This will be discussed in connection with the later castellum.

• 9 On beneficiarii: Schallmayer, Eibl and Ott 1990; Ott 1995; Nelis-Clément 2000.
• 10 France and Nelis-Clément 2014, on the concept and material realities of the stationes in the Roman (...)
• 11 E.g., Gatier 1998, p. 393, no. 84 (AE 1998, 1458), for a Latin dedication of an altar from Riḥāb b (...)
• 12 I. Jordanie 5/1, 145.
• 13 I. Jordanie 5/1, 100, found in a late antique house east of the so-called Barracks.

7The new inscription shows that, at least under the Severi, warrant officers (optiones) favored by the provincial governors as beneficiarii (holders of a beneficium) were temporarily detached from the third legion Cyrenaica and assigned to a military statio at Umm al-Jimāl, in addition to the soldiers of the local garrison.9 Here as elsewhere, the physical framework of the station probably consisted of a modest building designed as a guardhouse associated with a small sanctuary for the use of its occupants and perhaps playing the role of a relay of the imperial post (cursus publicus).10 More precisely, the text is part of the constantly growing series of dedications addressed by beneficiarii to the gods of their statio to thank them for their assignments or to celebrate the end of their mission.11 Two other Greek inscriptions from Umm al-Jimāl could be linked to the same practice. The first one recalls that a beneficiarius contributed to the decoration of a monument.12 The second one looks even more like the dedication of T. Flavius Artemidoros, because it was engraved on the shaft of a basalt altar and addressed to Zeus (fig. 5):13

• 14 IGLS 13/1, 9010: “Jupiter Optimus Maximus holy Genius Hammon” (Iuppiter Optimus Maximus Genius san (...)

9At the end of line 3 and in the following lines, the epithet κ̣[εραυνίῳ] may be restored, although other possibilities are not excluded (e.g., κ̣[υρίῳ] or the personal name of the donor beginning with a kappa). Either way, the name and titles of Zeus would once again suit the god of the third legion Cyrenaica, who was also described as “holy” at Bostra by means of a Latin epiclesis (sanctus) synonymous with the Greek epithet ἅγιος.14

• 15 Cf. IGLS 14, 8, 9 and 14, for the possible three-year mission of Flavianus in the city of Adraa be (...)

10The brevity of the new dedication makes it impossible to know more about T. Flavius Artemidoros. The onomastic formula that designated this Roman citizen is quite ordinary, so that it is not surprising to find homonyms in the Empire. Similarly, the duration of his mission remains unknown. At the beginning of the third century ce, the posting of beneficiarii to a given station routinely lasted six months, from the ides of January to the ides of July, then from the ides of July to the ides of January, and so on. However, longer missions are also documented, including in the limes Arabicus.15 The main point is that provincial authorities relied on such officers to carry out military police and law enforcement operations at Umm al-Jimāl, in direct contact with the nearby village community of al-Ḥirrīʾ and independently of the garrison hosted in the local castellum.

• 16 Brown 1998; de Vries 1998, p. 229.

11As with the castellum, the question of the location of the statio and its shrine now has to be asked. At first sight, there is little hope of finding any trace of these facilities, as the pre-fourth century buildings of the whole area appear to have been largely disrupted and dismantled due to the specific development of the late antique settlement of Umm al-Jimāl. We can nevertheless venture a proposal. In its initial state, the so-called Praetorium was a small public building, whose careful construction may date back to the late second or third century ce (fig. 6).16 It is tempting to view it as the station of the beneficiarii.

• 17 I. Jordanie 5/1, 121-123.
• 18 PLRE 1, p. 577, “Maximinus 6.”
• 19 I. Jordanie 5/1, 125 (367-375 ce). Lines 3-4: [sa|l]utisferumq(ue) for salutiferumque; cuntis for (...)

12Epigraphic sources provide only little information on the presence of Roman soldiers at Umm al-Jimāl for about a century and a half after the time of the Severi. Inscriptions mention road repairs and unidentified activity of the provincial governor during the Tetrarchy.17 At the end of this period, large-scale military work was undertaken during the joint reign of Valentinian, Valens, and Gratian (367-375 ce). This is what we learn from a small series of three Latin inscriptions. According to the first two texts, two towers were probably built in the same year 368 ce under the authority of Flavius Maximinus, comes and dux of Arabia,18 through the efforts of the Nonodalmatae, and to the care of Agathodaimon, tribune of the soldiers in this unit. The first document, found near Church 22 (North East Church), is not fully preserved at the end:19

• 20 I. Jordanie 5/1, 126 (368 ce).

13The second inscription, reused in House XX, provides a complete version of the text:20

• 21 Blouin 2016, p. 79-82.
• 22 Notitia Dignitatum, Or. 5.37, where the equites nono Dalmatae are mentioned among the uexillatione (...)
• 23 I. Jordanie 5/1, 149, funerary inscription for Aurelia Quirilla, wife of Veranus, sex(quiplicarius (...)

15The tribune in charge of the works had a Greek name which betrayed his Egyptian origin: Ἀγαθοδαίμων.21 His unit was composed of horsemen who had obviously been placed in the service of the dux of Arabia, head of the border troops stationed throughout the province and subordinate of the magister militum per Orientem. Later in the fourth century ce, the equites Nonodalmatae joined the palatine armies of the eastern Empire.22 A handful of Latin dedications of monumental tombs from the fourth century ce might relate to the officers of the same uexillatio and their families in Umm al-Jimāl.23

• 24 SEG 7, 1164 = IGLS 16/5, 1211.
• 25 Unpublished Greek dedication of a castellum kept at al-Mafraq in the Al-Bayt University Museum, in (...)
• 26 CIL 3, 14381, 14382; Littmann, Magie Jr. and Stuart 1910, p. 127-128, no. 229; Kennedy 2004, p. 76 (...)
• 27 On lines 7-8, [i]n | [--- s]ede in earlier editions. On line 10, previous editors read praef(ecto)(...)

16Other building operations were carried out in Arabia under the authority of the dux Maximinus at Ḏībīn in 365 ce,24 at Ṣubḥiyya in 369/370 ce,25 and at Dayr al-Kahf between 367 and 375 ce.26 The Dayr al-Kahf inscription deserves to be quoted in full as it is very similar to the texts from Umm al-Jimāl:27

• 28 I. Jordanie 5/1, 127, on a lintel reused in the so-called Cathedral or Church 14; cf. PLRE 1, p. 4 (...)

18A third, more recent text from Umm al-Jimāl mentions the construction of a whole burgus by the equites Nonodalmatae under the supervision of the superior of all the troops of the Diocese of the East, Iulius, magister militum (per Orientem), who resided in Antioch, in 371 ce (fig. 7):28

• 29 PLRE 1, p. 929, “Vahalus.”
• 30 Vahal, Vahalis, etc., of Germanic or Celtic origin according to Scheungraber and Grünzweig 2014, p (...)
• 31 Pace Sartre 1985, p. 225.
• 32 Littmann, Magie Jr. and Stuart 1913, p. 134: “The name Vahalus may be the Latin transcription of a (...)

20In 371 ce, Vahalus had replaced Agathodaimon as head of the equites Nonodalmatae.29 The name of the tribune is reminiscent of the ancient name of a river in the Batavian country, the present Waal in the Netherlands.30 However, it is not known as a personal name among Germans or Celts, so it is better to consider it as a transcription of a Semitic anthroponym. Given the presence of the Latin H, which could be used to render the ʿayn, it was not related to the Arabic wʾl and the Nabataean wʾlw,31 but to the Arabic name of the Nubian ibex (Capra nubiana), wʿl, which was extensively used as a personal name in ancient North Arabian languages (especially Hismaic and Safaitic) and from which the less common wʿlw was also derived in Nabataean Aramaic.32 In any case, the idea of local or regional recruitment should be favored for the tribune.

• 33 Isaac 1992, p. 178-186; Scheungraber and Grünzweig 2014, p. 112-114, s.v. “Burg 1.” For other ment (...)
• 34 Epitoma rei militaris 4.10: castellum paruolum quem burgum uocant.

23If we look at all the available documents together, not only two, but at least three towers (turres, sing. turris) were probably built at Umm al-Jimāl in 368 ce before a whole burgus was inaugurated in 371 ce. Although the situation seems quite clear, the interpretation of the Latin term burgus is not self-evident. This word of Germanic origin does not necessarily refer to a tower, contrary to what an abusive comparison with the Greek πύργος may have suggested.33 The three dedications of real towers belonging to the fourth-century defenses of Umm al-Jimāl prove that something else was at stake. The burgus was in fact a fortlet, a castellum paruolum in the words of Vegetius.34

• 35 Butler 1913, p. 165-166.
• 36 Brown 1998.
• 37 Littmann, Magie Jr. and Stuart 1913, p. 134.
• 38 De Vries 1986, p. 231-232, 238, fig. 10.2; de Vries 1993, p. 435-436, 445-446, 447, 452, fig. 4; K (...)

24Once again, it is difficult to recognize the fourth-century fortifications on the ground, as all inscriptions that show their existence were discovered in secondary locations. With regard to the burgus, the Princeton expedition scholars considered either the so-called Praetorium,35 which eventually did not show any fourth-century construction phase,36 or by pure guesswork “an outpost of the main castra” built in the late second century ce.37 In 1981, the excavations conducted as part of the Umm al-Jimāl Project started to uncover a large Roman castellum to the east of the site.38 It was an irregular quadrilateral of about 1 ha, with walls originally just under 2 m wide, four main gates flanked by towers, projecting towers at all four corners, interior rooms built against the walls and in the courtyard area, and a building identified as principia in its northern half (fig. 9). In its southwest corner, the fort was paired with a large reservoir fed by an aqueduct from a dam to the northeast. The very broad periodization put forward has so far not been supported by any detailed publication of the stratigraphy: while it is sure that the fort was dismantled in late antiquity, the alleged idea of a Tetrarchic foundation (ca. 300 ce) may be relevant, but it still seems to lack solid evidence. We are left to speculate on the possible building or reconstruction of this major fortification and its towers ca. 368-371 ce, under the reign of Valentinian, Valens, and Gratian.

• 39 I. Jordanie 5/1, 128, “found in the courtyard of a house in the south-western corner of the city, (...)

25A Greek inscription dates the construction of the later castellum to 412 or 413 ce (fig. 10):39

• 40 I. Jordanie 5/1, 92, dated June 17, 411 ce; cf. PLRE 2, p. 106, “Fl. Pelagius Antipater 3.”
• 41 Sartre 2007, p. 204-208; Yon 2018, p. 66-69.
• 42 Notitia Dignitatum, Or. 37.4 (Tricomia); 37.15 (equites promoti Illyriciani, Tricomia); George of (...)

27The same dux of Arabia is known from the Greek dedication of another castellum in al-Bāʿij, very close to the site.40 At Umm al-Jimāl, the construction of the fort was supervised by a junior officer whose rank was just below that of tribune. The name of Bassos could be the Greek transcription of the Latin Bassus or the transcription of an Arabic anthroponym taken from the name of the cat, bs.41 It is too common to draw any conclusions about the origin of the primicerius, but there is reason to wonder whether this soldier came from Tricomia ([Τρικω?]μίας) and whether this garrison site was located in Umm al-Jimāl.42 If such an assumption were to prove correct, then it would have to be concluded that the unit stationed at Umm al-Jimāl at the beginning of the fifth century ce was that of the equites promoti Illyriciani.

• 43 Butler 1913, p. 171.
• 44 Parker 1998; cf. Villeneuve 2001, p. 213-214, and Kennedy 2004, p. 90-91, with methodological cave (...)
• 45 I. Jordanie 5/1, 130, “reposant sur le sol à l’extérieur, près de la maison n^o 51.”

28Howard Crosby Butler linked the inscription to the first state of the Barracks (fig. 11).43 Since the Princeton expedition, limited excavations under the outer wall of the building have produced ceramics from the fourth and fifth centuries ce, which did not invalidate this hypothesis.44 The study of the Barracks also revealed major alterations between the early sixth and the early seventh century ce (repair of the southeast tower which was then adorned with Christian inscriptions, addition of a tower on the west side, etc.). During the same period, the surrounding wall of the whole settlement was the subject of repairs. A recently published Greek inscription may relate to this set of operations. It shows that the work done in the castellum (κάστελλος) was the responsibility of Kornelios, an officialis from the staff of the military governor of Arabia.45

• 46 Littmann, Magie Jr. and Stuart 1910, p. 22-23, no. 18; Meimaris 1992, p. 231, no. 262.
• 47 H. × W. × D.: 36 × 86 × 14 cm (a); 36 × 45 × 14 cm (b). Cross, S-shaped abbreviation mark. Tabula (...)
• 48 Line 2: ἀπά for ἀπό. Lines 3-4: δούξ for δουκός; τὰ | κάστραν for τὸ κάστρον; ἔτι for ἔτει.

29It also turns out that a neglected inscription allows us to place the renovation of the defenses of Umm al-Jimāl in the context of the emergence of a kastron in the early years of Justinian’s reign. This document was found in 1905 by the members of the Princeton expedition in the Umayyad residence of Qaṣr al-Ḥallabāt, about 20 km south of Umm al-Jimāl.46 Today it is preserved in the storerooms of the Jordanian Archaeological Museum in the ʿAmmān Citadel. The text was engraved inside a tabula ansata on a basalt plaque broken in two fragments (fig. 12).47 There is no problem reading it:48

• 50 Bisheh 1986, p. 161-162; Bisheh 1993, p. 49-50.
• 51 Littmann, Magie Jr. and Stuart 1910, p. 21-22, no. 17; Kennedy 1982, p. 39-40, no. 3; Kennedy 2004 (...)
• 52 Arce, Feissel and Weber-Karyotakis 2018; cf. Arce 2018, who inconsistently believes that the two b (...)

31As Ghazi Bisheh convincingly argued over thirty years ago, it is highly likely that this inscription was originally displayed at Umm al-Jimāl.50 The same can be said for all of the basalt blocks reused as building materials in the Umayyad residence of Qaṣr al-Ḥallabāt, including the slab which bears a Latin inscription relating to the construction of a castellum nouum with the help of detachments of four cohorts in 212/213 ce (fig. 13).51 It is now established that the numerous fragments of the famous constitution of Anastasius I (491-518) found at Ḥallabāt were brought from Umm al-Jimāl.52

• 53 Saliou 2020.
• 54 Gatier 2005, p. 109-110; Aliquot and al-Husan 2019-2020.

32The use of the Greek word κάστρον also makes a strong case for attributing the kastron inscription to Umm al-Jimāl. This polysemous term belonged to the late antique administrative language.53 Although it sometimes designated a fortress or a fortified residence, or even a monastery (in Sinai), it was most commonly used, in its Latin version castrum from the fourth century ce, and then in Greek from the sixth century onward, in relation to urban agglomerations of non-civic status, particularly in border areas, at a stage prior to the elevation to the status of city (ciuitas, πόλις). Therefore, the notion of kastron would be inappropriate to describe the modest fortlet of Qaṣr al-Ḥallabāt under Justinian. On the other hand, the concept was well suited to a series of large villages located along the steppe and exposed to the threat of nomadic raids in Provincia Arabia, including Mothana (Imtān), Riḥāb, Samrāʾ, and Umm al-Jimāl, if only the territory of Bostra is considered.54

• 55 It is enough here to gather epigraphic records on beneficiarii consularis serving in the province o (...)
• 56 Russell 1980; Guidoboni 1994, p. 264-267, no. 153.
• 57 E.g., AE 1948, 136 (Ḫirbat Umm al-Manāra, 334 ce), with Aliquot 2016.

33The present study has brought to light several hitherto unknown aspects of the history of Umm al-Jimāl in the long run. At the origins of the settlement, it now appears that a station of beneficiarii was established there no later than the reign of Caracalla (211-217). The warrant officers sent to the site were most probably detached from the headquarters of the third legion Cyrenaica in Bostra. With the capacity to act on behalf of the governor of Arabia, the highest authority in the province, they were responsible for police and intelligence operations around the adjacent village of al-Ḥirrīʾ. Their temporary missions were carried out independently from the local garrison stationed in the castellum nouum which was built or renovated in 212/213 ce and whose location is not known. We thus find here the soldiers of the Roman army in their traditional role of guardians of the territory entrusted to the provincial authorities on all the depth of the limes Arabicus. In this respect, the example of Umm al-Jimāl encourages further research into the network of beneficiary stationes scattered throughout the province between the second and early fourth century.55 The reasons for the decline of al-Ḥirrīʾ during the same period have yet to be determined. One might consider the effects of the earthquake that devastated Palestine and Arabia in the night of May 18-19, 363 ce,56 the recurring rise of territorial friction created by nomadic attacks, which epigraphic and literary sources occasionally mention,57 or a combination of all these factors. In any case, from the last third of the fourth century onward, the construction of important military buildings at Umm al-Jimāl under the authority of the dukes of Arabia led to the birth of a new fortified village which was to obtain the status of kastron in late antiquity.