Chapter III

The 2016 Excavation Campaign

by Thomas Lepaon


The excavation campaign of 2016 lasted from April 11th to May 25th. It was legalized by a permit issued on behalf of the Department of Antiquities of Jordan. Jordanian archaeologists of the University of Jordan and French specialists of the Mission Archéologique Française de Jerash conducted the active fieldwork under the supervision of two inspectors of the Department of Antiquities and employed local workmen. The archaeological interventions focused on two separate spaces: “area M” between the “northern hall” and the frigidarium of the eastern baths, and the semicircular annex (S) to the hall (Fig. III. 1 and Pl. III. 1, 1; Pl. III. 1, 2). The entire compound of archaeological investigation considered in the research program was divided into 5x5 m squares numbered continuously and oriented according to the axis east-west and north-south of the monument (Pl. III. 1, 1; Pl. III. 1, 2). For clarity, the archaeological description of the discoveries will be made by area and it will follow the reverse chronological order.

Fig. III. 1 The areas M and S excavated in 2016 between the frigidarium of the baths and the semicircular exedra of the northern hall.
III. 1. The Transitional Basin Room (Area M)

Area M (Fig. III. 2) occupies an area of approximately 285 m² as an transitional architectonic space between the frigidarium and the northern hall. Its masonry and openings are arranged according to the symmetry axis of the building in north-southern alignment. This Room M communicated with the frigidarium through three openings (OUV. 55001; 55002 and 55003) pierced thru the southern wall (MUR.55001) (Fig. III. 2 and Pl. III. 3, 1; Pl. III. 3, 2). At its northern side, Room M provides access to the northern hall along its on its entire length via a portico of eight columns placed on high pedestals between two pilasters (POR.55001).

The level of topsoil before the archaeological operations were determined between 561.50 m east and 561.00 NGJ m west at the Jordanian General Level (NGJ) (Fig. III. 3). The terrain has thus a slight inclination towards the wadi (5.5%). Seven squares have been opened as a part of the archaeological work in 2016 (Pl. III. 1, 1; Pl. III. 4, 1; Pl. III. 4, 2; Pl. III. 6).

Fig. III. 4 Southern profile of the filling of pit F.55001.
III. 1.1. Modern occupation of area M (Agr. 55001)

Immediately below the topsoil (US.55000), several modern occupation layers were identified (US.55006; 55007 and 55030). These layers contained fragments of recent plastic objects. Some Hashemite coins attest their rather recent deposit.

The excavation of square 79 quickly showed the presence of a very large pit (F.55001) in front of the central passage (Pl. III. 1, 1).This pit extended beyond the sector east and north. It has been dug into the Umayyad and Byzantine levels,  and the bottom has been reached (US.55062 - Fig. III. 4). Its filling (US. 55004) consisted of modern garbage (glass bottles, plastic bags, etc.), and the spot had already been excavated by the Department of antiquities in 2004. The torsi and limbs of the marble statue of the dancing Satyr of the type “with fruit scarf and panther” holding the Dionysus child in his left arm (see here cat.-no. 6) were discovered at that time in this pit. The resumed excavation in 2016 yielded knowledge of the precise extension of this dump pit and allowed the removal of all modern intrusive material.

Fig. III. 5 Levels of sterile colluviations discovered in square 77.
III. 1.2. Abandonment of area M (Agr.55002)

Under the modern levels followed a succession of sterile silt layers. They are characterized by a significant thicknesses and common soil characteristics such as dark brown color and homogenous texture. They contained no archaeological remains (Fig. III. 5).

III. 1. 3. Re-occupation of the ruins of area M (Agr. 55003)

After the wall’s destruction, area M underwent new occupation. To the northeast, the wall (MUR. 55004 - UC. 55029) was built in the level of abandonment that succeeded the destruction of the bath building (US. 55044). Wide of about 0.50 m, this masonry was recognized over 3.10 m length (Fig. III. 6; Fig. III. 7; Pl. III. 2, 1).

The wall is composed of quadrangular soft yellowish limestone blocks and an upper drum of column. The blocks are not regularly chiseled and no mortar was used in the construction (Pl. III. 2, 2). The northern face of this wall received a cover of mortar about 0.02 m  thick (UC. 55145). It is tapered at the base of the construction on the level of a riffle of stones of medium sizes (UC. 55040). The nature of the structure suggests that it could be the remnants of a pool. Southeast of this excavation area (squares 89, 90 and 99), the Roman masonry has preserved traces of late occupation. Thus, initially covered with marble veneer, the façades of the walls show secondary recess holes (F.55012 - Fig. III. 8), horizontal holes that might have supported a horizontal wooden board serving as a shelf (F.55013) or a small recess (F.55003) which in its interior still preserves a whitish coating (US.55013) (Fig. III. 9). In front of these holes, the remains of whitish plaster coatings were discovered (UC.55085, 55090 et 55091) (Fig. III. 10; Fig. III. 11).

Fig. III. 6 MUR.55004 seen from the East.
III. 1.4. Demolition of area M (Agr.55004)

Several indications for the destruction of the Roman building have been discovered. South of room M (squares 89 and 99), many collapsed architectural  blocks such as fluted chest column, architraves, ledges, etc. were found in situ (Fig. III. 12 and Pl. III. 3, 1). There was no architectural element which testifies the presence of a roof. To the north of room M (square 77), the destruction of the building was also marked by the collapse of column drums and capitals (US. 55054 and 55056) (Fig. III. 13). These architectural elements are identical to the ones still in place around the northern hall. However, unlike the blocks collapsed to the south, no element of an entablature was discovered.

In this destruction horizon an important statuary fragment belonging to a colossal statue of Aphrodite out of white marble came to light. Preserved to a maximum height of 1.645 m, the torso consists of a plinth supporting the lower body of the deity and an Eros riding a dolphin (see here cat.-no. 4 Fig. IV. 25). This fragment corresponds to the statuary type “Aphrodite of Troas” well known in Rome during the 1st and 2nd centuries A.D. On the pedestal, two Greek inscriptions of several lines inform that the statue has been offered by the priest Demetrios, stepson of Asklepiodoros in the year 216 of the Gerasene era (=154 A.D.) under the reign of Antoninus Pius.

III.1.5. Recovery of decorative elements (Agr. 55005)

Before its demolition, room M experienced the re-covery of its rich materials. This plunder is particularly visible by elements of veneers. The majority of the decorative colored marble revetment slabs was still found with their iron dowels. Sometimes, workers did not hesitate to dig the mortar located at the base of walls to recover all of the metal nails (F.55008 and 55009) (Fig. III. 14 and Pl. III. 3, 2). Similarly, the hydraulic design belonging to the initial phase of the monumental bath has likely been recovered during this phase (F.55007).

The vast pool occupying the entire space of room M has been dismantled and uncovered (F.55014 - cf. III. 1. 5. 3). Only the lower parts of this installation are still in place in the northern (UC.55012) and southern sectors (UC. 55103 and 55106). Few traces showing the ancient construction process were also found. Thus, to the north (square 77), a working layer related to the deposit of plunder has been excavated (US.55116). The southeastern sector (square 99) preserved the re-mains of a small fireplace (F. 55015) (Fig. III. 15). It was found surrounded by marble veneers which were broken from the Roman building and laid directly on the flat ground (US.55055). These settings can be interpreted as a temporary installation during the recovery work.

III. 1. 6. Remains of Roman occupation (Agr. 55006 and 55007)
III. 1.6.1. Masonry

Only the masonry bordering room M to the south and to the north have been uncovered during the  2016 campaign (Pl. III. 3, 2). To the south, the wall Mur. 55001 consists of an elevation (UC.55100) resting on a strong foundation (UC.55101) (Fig. III. 16). This wall is constructed out of regular large stones (about 0.65 m high) dressed in soft limestone. The blocks of the northern façade of the elevation of this masonry (UC. 55100) are marked with a peripheral chasing. The masonry was covered by a colored marble revetment as indicated by fragmentary slabs in situ and many holes for metal doweling. This adorned masonry is limited to those walls forming the frigidarium which adjaces along the southern flank of room M (Agr. 55007). To the west, an approximately 0.35 m wide opening pierces through this wall (OUV.004) (Fig. III. 17). Oriented north-east/south-west, this opening originates from the time of the construction of the wall. Subsequently, this opening was closed by blocks in a later period and covered by plaster coating (UC.55141 - BOU. 55001). To the north, a vertical opening marks the front of this masonry (F.55016) (Fig. III. 18). Of quadrangular plan (0.20 x 0.20 m), this opening does not seem to be contemporary with the construction of the wall. Indeed, at the location of the opening, the faces of the blocks of masonry were originally the peripheral carvings which have been removed during the excavation.

At the northeastern corner of the wall MUR 55001, the excavation of this probable gutter displayed the wall decoration made of fine marble slabs (UC.55076) resting on a layer of soil (UC.55075) (Fig. III. 19 and Pl. III. 3, 2). The morphology of these slabs (less than 1 cm thick) and the use of the sediment for their fixing upon the wall appears quite different from the features observed in the other sector of excavation where plates are are usually between 3 to 5 cm thick and an application of sediment does not exist. This plating seems to be a later addition but it is in the present phase of study  yet impossible to specify the chronology.

To the east, the wall Mur. 55002 forms an elevation (UC.55011) (Fig. III. 20). Although similar to UC. 55100, this masonry belongs to a secondary phase of the building (Agr.55006). On the one hand, it set against the masonry MUR. 55001 closing the frigidarium to the north. On the other hand, the construction of MUR. 55002 has required the partial carve of the original north façade of the wall MUR. 55001. This observation seems to attest that wall MUR. 55002 belongs to a different phase of the construction of wall MUR.55001 and it does definitively not belong to the initial construction phase.

To the north, the wall UC.55115 served as the foundation of the portico (POR.55001) opening into the northern hall (Fig. III. 21 and Pl. III. 4, 1). This construction consists of large soft limestone blocks in a regular layout (about 0.65 m high). No indication is available to attribute this construction to one of the two isolated phases previously discussed.

III. 16.2. Floor and Piping

The remains of a paved floor discovered between the masonry UC.55100 and UC.55011 (SOL.001) can be defined as the oldest known in 2016 in this sector (Agr. 55007) (Fig. III. 22).

This floor consists of slabs of hard limestone which survived only in fragments (UC.55140) resting on a bedding of gray mortar (UC.55069). Despite the incomplete state of the pavement, the impressions of the missing slabs in the mortar are still visible and permit to restore the layout of the paving (Fig. III. 22 and Pl. III. 3, 2).

A channel for the receipt of a pipe (CAN.55001) pierces the ground level in east-western direction, about 0.50 m wide and 0.25 m deep, following a slight inclination of the floor to the west. The bottom of this channel was coated by multi-layered hydraulic mortar (UC.55093 and UC.55094). Between its single layers sometimes a thin package of homogeneous, compact clay has been inserted (US. 55092). This mortar is probably related to the construction of this channel. The interior of the channel groove was filled with a layer of sandy soil (US.55071). From a chronological point of view, this pipe must be attributed to the second phase spotted in the sector (Agr.55006). Indeed, this hydraulic device dug into the floor pavement required a rather coarse piercing thru the masonry of the western wall (UC. 55100) (Fig. III. 23). In contrast to this, the throroughfare was carefully executed thru the eastern wall (UC. 55011) (Fig. III. 24). In addition, this pipe is connected with a vertical opening observed previously (F. 55016) attributed to the second phase spotted in the sector (Agr. 55006).

North of the channel, further traces of architectural installations were found in the space between the masonry and the foundation wall UC.55101 (Fig. III. 25 and Pl. III. 3, 2).

Built against the eastern wall decoration (UC. 55076), a thick layer of mortar (UC. 55077 and UC. 55095) covered the upper part of the foundation wall UC. 55101. Several terracotta pipes, quadrangular and circular in section, sometimes broken and not systemically flat, were drowned in this layer of mortar. Within this layer, different prints related to an ancient device, impressed during the application of this mortar, are still visible. To the east, the first application (F. 55017) is perfectly square plan (0.15 m) and pierces the mortar throughout its full thickness. Symmetrically opposite, the second (F.55018), less well preserved, has the same characteristics. Finally, an impression in the mortar joins the wall MUR. 55001 and MUR. 55002. The northern sector of this impression is perfectly straight and its end is marked by two notches. The morphology of these prints suggests that it may be the marks left by the timberwork frame of a door.

1.6.3. The pool (BAS.001)

The traces of a vast basin occupying the full width (9.50 m) of area M have been discovered within the squares 77, 89 and 99, respectively, in front of the access central vault (OUV.001) and the foundation wall of the portico to its north (POR.55001) (Pl. III. 2, 1; Pl. III. 2, 2; Pl. III. 3, 1; Pl. III. 3, 2). The bottom of this basin (SOL.002) consisted of quadrangular slabs of large format (between 0.50 and 0.70 m), out of hard rose limestone (UC. 55061, 55104, 55107) (Fig. III. 26; Fig. III. 27; Fig. III. 28; Fig. III. 29). Within this coherent pavement system, the northern sector of it (UC.55113) has slabs of modest size (0.30 x 0.40 m). The altitude of this pavement lies between 558.18 and 558.25 NGJ m indicating that the bottom of the pool is in level. The total extension of this basin still unknown because only the north and south boundaries have been sounded.

To the north, a low wall (UC 55111) is installed in a distance of 0.40 m from the foundation structure of the portico POR. 55001 (Fig. III. 30). This construction of yellowish soft limestone (0.10 to 0.13 m width) is preserved on 0.50 m above the bottom of the pool (UC. 55113) (Fig. III. 31). The southern part was covered with slabs (0.05 m thick) of hard white/pink hard limestone (UC. 55112). These slabs are similar to those used for the bottom of the pool. To the north, the space between the masonries UC. 55111 and UC. 55115 was filled with a layer of clay containing no artefact (US. 55117). The latter, the pavement of the basin and east-western masonry UC. 55111 rest upon a compact layer of small stones and brown sediment (US 55119) which is set against the masonry UC.55115. To the south of it, a veneer wall, similar to the one excavated in the northern one, obviously marked the limit of the basin (UC. 55103 and UC. 55106) (Fig. III. 32; Fig. III. 33). Their implementation, however, is different. The veneer was set against brick masonry (UC.55102) built against the southern foundation wall of the area M (UC. 55101).

A terracotta structure, 0.72 m wide, consists of several bases of square bricks of a similar size (0.35 x 0.35 x 0.03 m). They are arranged in two lines running parallel to the foundation wall. The last base of brick is a more modest feature allowing to leave the central part empty. This device, which is present on the entire length of the masonry UC.55100, terminates at the location of the niche at the center of room M (recorded as UC.55105 in square 89) (Fig. III. 34). In this niche of segmental plan, the revetment (UC.55106) is directly placed against the wall and fixed by iron dowels (Fig. III. 35). Finally, a thick layer of mortar warrents the water-proof tightness of the basin between the ground and the plated elevation (UC.55114, 55143 and 55144).

III. 2. The Exedra (Area S)

Area S  has a semi-circular plan and measures about 22,40 m in diameter. Theoretically restored it must have covered an area of about 200 m² (Fig. III. 36 and Pl. III. 4, 2). It opens directly southward into the northern hall over its entire length through a portico of six columns between two pilasters (POR.55002). The topsoil level prior to the archaeological intervention had a considerable inclination westward (562,75 m NGJ east and west 561,55 m NGJ). Eight squares (53, 54, 63, 64, 73, 74) have been excavated in the 2016 season down to the level of the Roman floor.

Fig. III. 36 The exedra (area S) during excavation.
III. 2.1. Modern occupation of area S  (Agr. 55008)

Area S had already been partly excavated in 1984. Following this former archaeological operation, the area has been used as a dumping place for the surrounding modern domestic dwellings. The release of levels corresponding to these modern occupation has been registered in 2016 with the identification label “US. 55000.” A highly grown therebinth tree and the northeastern bulk limitation threatened the security of public life  and the planned excavation of the area (Fig. III. 37). With an official approval by the municipality authorities as well as the neighboring landowners, this tree has cut down to its roots. To the west of this tree and south of the public passageway leading from the promenade piazza of Meidan al-Hashemiye to the municipal market area a recent pavement (Fig. III. 38) covered the planned excavation terrain. Also this was removed by the municipality with a bulldozer.

III. 2.2. Abandonment of the area (Agr.55009)

The excavation of the areas not explored in 1984 has found a succession of unexplored sterile sedimentary layers in all respects similar to those observed in space M (Fig. III. 40).

III. 2.3. Demolition of the Exedra (area S) (Agr.55010)

Several indications pointing to the destruction of the ancient monument were discovered (Fig. III. 41 and Pl. III. 5, 1). These are several architectural blocks such as architraves, ledges bases, etc.) came to light in a position which was caused by collapse (US.55141). Again, no element was identified which testifies the presence of a ceiling.

Beside architectural blocks, partly delicately decorated, two previously unknown statue bases appeared among the fallen blocks (Fig. III. 42; Fig. III. 43; Fig. III. 44; Pl. III. 6; Pl. II. 8). The first one (Fig. III. 42; Fig. III. 43) survived only in its upper part. A fragmentary Greek inscription preserving only the first line is engraved on it. According to Pierre-Louis Gatier it reads “to the good fortune”. This is the accustomed introduction formula for an ex voto, in this case certainly an honorary statue dedicted by the municipial community or individual donators. The upper surface of the block has two series of holes and a chiseled imprint of a foot sole indicating the standing pose of the statue missing today. This type of mounting devices by dowelling is characteristic for sculptures cast in metal. The second base (PiS. 55032) displays as well traces of the missing statue’s mounting, but it lacks any evidence of an inscription.

III. 2.4. Recovery of the decorative elements  (Agr. 55011)

Comparable to area M, the floor and parietal decoration has been uncovered in 2016 (F.55010). Marble slabs have almost entirely been looted leaving only the prints in the mortar base. Parietal veneers revetting the lateral wall MUR.55005 (UC.55128) as well as the semicircular and quadrangular niches. These structures were completely covered by revetment slabs as proved by the holes for metal clamps (Fig. III. 45).

The excavation of 2016 highlighted an extensive re-use of area S over allsurfaces uncovered so far (US. 55050; 55109 and 55110). The excavation of these layers has delivered abundant material of marble as well as rich ceramic finds. The preliminary analysis of the pottery allows to establish a range of production between the 6th and the 7th centuries A.D. Unexpectedly in this horizon, otherwise thoroughly dismantled of its marble elements for the purpose of lime burning, a second statuary marble fragment came to light (Fig. III. 46). It is a female portrait head (see here cat.-no. 14), fairly well preserved, which at first glance resemples the physiognomical features of Julia Domna, wife of Severus and mother of Caracalla. Such an identification, however, cannot be verified at the present state of research, and we must therefore strongly assume that the represented matrona was either an unknown female member of the imperial court or, even more likely, a lady of the local Gerasene nobility in the fashion attitude of the empress.

III. 2.5. Remains of Roman Occupation (Agr. 55011, 55012 and 55013)

The semicircular exedra area S displays at least three phases of occupation in connection with the bathing use of the Roman thermes.

The earliest phase (Agr. 55013) includes all structural walls forming the large semicircular exedra and the southward adjacent portico opening into the northern hall. The 2016 campaign supplemented the Jordanian excavation of 1984 during which only the southwest corner, three pedestals of columns and the base of the pilaster of angle based on a foundation wall had been cleared (MUR. 55005 - UC. 55142). In 2016 the curved wall of the exedra (MUR. 55006 - UC. 55128) was unearthed starting from the west along its entire depth of 2.05 m (Fig. III. 47; Fig. III. 48). The alterating sequence of semicircular (NCH. 55001 and 55003; r ± 1.06 m) and square niches (NCH. 55002; 1.35 m x 0.50 m) recessing in the interior curved wall was confirmed as it has been shown already on the plan of 1984 excavation (cf. also above Fig. II. 1).

The curvature of the excavated segment allows to restore the radius of the exedra at approximately 11,50 m. The semicircular patio thus reached an asumed surface of ca 200 m². This space was originally paved as a coherent marble floor (UC. 55125, 55126 and 55127). The slaps rested upon a bed of greyish mortar (UC. 55146, SOL.55003). Later occupation largely looted the marble tiles, but their impression left in the mortar during their installation, permits to restore the original layout.

The second phase (Agr. 55012) manifests in the construction of an outer wall (MUR.55006) reinforcing the older curved exedra screen (UC. 55129). This supplementary wall, which has been excavated this year only along the upper erased layers of masonry, is built with large limestone blocks. For the construction of this new wall, ancient constructors destroyed a part of the masonry of the older curved screen (UC. 55128). In the same chronologial phase, small recesses were hewn into the profiles of pedestals and column bases of the portico (POR.  55002). This may indicate that access to the semicircular room has been controlled during this period.

Finally, area S underwent a third phase (Agr. 55013) of architectural transformation marked by construction of walls MUR.55008 (UC.55134) and MUR. 55009 (UC.55135) inserted between the columns of portico (Fig. III. 49; Fig. III. 50). These screening walls achieve a thickness from 0.80 to 0.90 m, and are made out of re-used blocks. One of these spoils belonged presumably to a pilaster of the northern hall. If this oberservation is correct, the screens between the columns of the exedra have been installed when the adjacent hall was not intact anymore. The boulders of the secondary screening walls are not built with regularity and no mortar was applied to consolidate the joints. At the same time, the function of the niches as places for the presentation of statuary has been abandoned: The quadrangular niche (NCH.55002), for instance, has been filled by large unit blocks (UC. 55133).

Plates