The mortars from rock-cut hydraulic structures of as-sila (Sela) in Southern Jordan: Mineralogical characterization and radiocarbon dating

www.maajournal.com Mediterranean Archaeology and Archaeometry Vol. 21, No 2, (2021), pp. 37-67 Open Access. Online & Print. Copyright: © 2021. This is an open-access article distributed under the terms of the Creative Commons Attribution License. (https://creativecommons.org/licenses/by/4.0/). DOI: 10.5281/zenodo.4643739 THE MORTARS FROM ROCK-CUT HYDRAULIC STRUCTURES OF AS-SILA (SELA) IN SOUTHERN JORDAN: MINERALOGICAL CHARACTERIZATION AND RADIOCARBON DATING Rocío Da Riva1*, Francisco Javier Santos Arévalo2, and Marisol Madrid i Fernández1 1Departament d'Història i Arqueologia, Universitat de Barcelona, GRACPE research team, Spain 2Centro Nacional de Aceleradores (Universidad de Sevilla, CSIC, Junta de Andalucía), Spain Received: 26/12/2020 Accepted: 27/03/2021 *Corresponding author: Rocío Da Riva ([email protected]) ABSTRACT One of the aims of the 2016 campaign in as-Sila was to conduct a survey in order to identify cisterns, channels and structures related to water use at the top of the settlement, using the 2015 survey map of the site as a base. The investigations have revealed a complex and sophisticated hydrological network with a great diversity of water structures (carved in the sandstone and designed for the collection, storage, transport and distribution of rainwater). All the structures have been identified and described using a total station, their location associated with the general topography of the site, and the contexts photographed. This study presents the results of recent analyses of lime-based mortars from rock-cut hydraulic structures collected during the 2016 archaeological campaign at the site of as-Sila/Sela, in the governorate of Tafilah in southern Jordan. Mineralogical and petrographic analyses were performed on 16 samples of mortars by means of x ray diffraction (XRD) and thinsection petrography (OM), and 12 AMS radiocarbon dates were taken from them. In spite of the difficulties in dating lime-based mortars and the problems inherent in the interpretation of the data, here we present the most complete analysis currently available of mortars from an archaeological site in southern Transjordan. KEYWORDS: mortars, hydraulic structures, Transjordan, hilltop settlements, mineralogical analyses, thinsection petrography, AMS radiocarbon analyses 38 R. DA RIVA et al. Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 1. INTRODUCTION AND OBJECTIVES The site of as-Sila is located in the northern Edomite Plateau, west of the Wadi Arabah, very close to modern-day Busayra in southern Jordan (Fig. 1). The site stands on a rocky promontory 200 m above the surrounding wadis. Sela was an important centre in the Late Iron Age (1000-539 BCE) 1 , as the presence of remains from this period testifies – among them, the impressive Neo-Babylonian monument of King Nabonidus (556–539 BCE) with a relief representing the king and the remains of a cuneiform inscription (MacDonald, 2004, Site 134; Da Riva, 2019; Da Riva, 2020) 2 . Modern-day as-Sila is one of the sites potentially identified with Sela of the Hebrew Bible, but this identification is uncertain. The relative chronology of the surface material (pottery) confirms occupation during at least the Late Iron Age, the Nabataean/Roman, Ayyubid, Mamluk and Ottoman periods. The cuneiform inscription of Nabonidus provides an independent date in the Late Iron Age, and the results of the 14C analysis of timber recovered from a building excavated in area F of the promontory (House 1, see Da Riva et. al., forthcoming) provide absolute chronologies for occupation in the Mamluk and Ottoman periods. The research work carried out at Sela since 2015 has revealed a large-scale site of approximately 43 ha, with a considerable number of architectural structures of different periods, sizes, morphologies and, presumably, functions: rock-cut houses, remains of large buildings, towers, and so on. The most notable feature of the site, however, is the presence of water structures: channels, tanks, rock-cut water cisterns and other structures used for rainwater harvesting and its storage, transport and management (Shqiarat Mansour 2019). More than a hundred of these structures have been documented in the course of the surveys, making Sela an ideal and unique site to study water management in the highlands of southern Jordan 3 . 1 Here we follow Bienkowski’s chronology for Edom (MacDonald, 2015, 24). 2 Water management in Sela presents parallels with other settlements in the area, such as Ba'ja III, Umm al-Biyara, Jabal alQseir, etc. (Lindner, 1992). A study of the water structures in Sela is the object of an ongoing thesis at the University of Barcelona by Figure 1. Map of Jordan showing the location of Sela (after Porter, 2004, 375) During our first campaign (2015) we undertook a survey of the site and produced a cartographic base for future studies (Fig. 2). The work was carried out using case studies in the area, mostly in the vicinity of Busayra, and archaeological data and charts stored in the JADIS (Jordan Antiquities Database and Information System) – MEGA (Middle Eastern Geodatabase for Antiquities) of the Department of Antiquities and Tourism in Jordan. The archaeological field survey of 2015 was non-intrusive and extensive (Da Riva et. al., 2017). Given the size of the site (42,0089 ha), in order to facilitate the surveys and the fieldwork, in 2015 we decided to divide the surface in a series of areas, indicated by the letters A-L on the sketch map (Fig. 3). R. Marsal under the supervision of R. Da Riva and J. C. MorenoGarcía. 3 All the information gathered by the project has been placed at the disposal of the Department of Antiquities of Jordan (DoAJ) of the Ministry of Tourism and Antiquities in the form of reports and site cards to be included in the Middle Eastern Geodatabase for Antiquities (MEGA) Jordan database: http://megajordan.org. THE MORTARS FROM ROCK-CUT HYDRAULIC STRUCTURES OF AS-SILA (SELA) IN SOUTHERN JORDAN 39 Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 Figure 2. Cartographic plan of Sela with the structures documented in the course of 2015 and 2016 surveys (by E. Jariri, D. Gaspar and J. Muñiz,) 40 R. DA RIVA et al. Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 One of the main features of the 2016 campaign was the survey conducted to identify cisterns, channels and structures related to water use at the top of the mountain of as-Sila, using the 2015 survey map of asSila as a base. All the structures were identified and described in the course of the 2015 and 2016 archaeological surveys with a total station, they were individually labeled, contextualized in the different areas of the site, and linked with the architectural arrangements. We also associated the location of the water structures with the topography and geology of the site so as to identify the general hydrological network. Figure 3. Sketch map of Sela (by R. Marsal after Corona Atlas & Referencing System, http://corona.cast.uark.edu/) In the course of the last campaign in Sela in the autumn of 2018, the Sela team studied the cuneiform inscription of Nabonidus with the assistance of professional climbers. The aim of this campaign was to elucidate the archaeological context of the monument and, later, to create a 3D model of it. Some of the water structures were studied at the same time, and archaeologists and pottery specialists undertook a detailed survey of the whole site, both on top of the promontory as well as in the eastern side of the wadi, collecting sherds for analysis (Da Riva, 2020). During the 2015, 2016 and 2018 campaigns, the water structures of Sela were identified and their contexts photographed, but as they were not excavated, no typology could be established. However, a thorough investigation of the mortars of some cisterns was carried out in order to complement the information obtained from the direct field study carried out in the structures of Sela. The main objectives of this investigation were: a) to observe whether the diversity of the cisterns was in line with the variety in the production of the mortars that protect and waterproof the hydraulic structures; and b) to date them in order to contextualize the use of the cisterns within a certain period of time. With these objectives in mind, during the archaeological campaign carried out at the site in the spring of 2016, a sample of 16 mortars associated with hydraulic structures and residential areas were selected for archaeometric study (Da Riva et al., forthcoming). Mineralogical characterization by means of XRD and petrographic analysis were performed, and finally some of the mortars were selected for radiocarbon dating. In order to avoid bias, the mortar samples were taken randomly; that is, there was no prior selection of the cisterns from which they were taken. 2. SELA: SETTLEMENT AND CHRONOLOGY Although surface finds and excavations at the summit indicate that as-Sila/Sela was occupied during several periods (at least the Iron Age and the Nabataean, Mamluk and Ottoman periods), the site’s most extensive occupation and use were during I millennium BCE, in the Iron Age II period, the time of the Biblical Edomites and the Neo-Assyrian and NeoBabylonian empires. In fact, Sela is one of the socalled "hilltop settlements", a type of mountain site that flourished in the region in the Late Iron Age (Ben David, 2015), and there are indications of pottery and THE MORTARS FROM ROCK-CUT HYDRAULIC STRUCTURES OF AS-SILA (SELA) IN SOUTHERN JORDAN 41 Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 architectural structures comparable to those found at other nearby Iron Age II sites, such as Umm al-Biyara, Busayra, Hesban, Ba’ja and Umm el-Ala (Lindner and Farajat, 1987; Lindner, 1992). Indeed, the study of the surface pottery carried out by archaeologists Najjar and Herles in 2018 revealed a variety of periods and a broad chronological framework, which bears witness to the long-term occupation of the settlement. Interestingly, no Bronze Age or Iron Age I pottery was found in the course of the survey, although of course an ephemeral occupation of the site in those periods, or even earlier, cannot be ruled out. The extent and continuity of these occupations is difficult to estimate at this stage in the research, but the many structures related to rainwater harvesting and storage documented during the surveys suggest that the occupation was intense, at least for certain periods. All this evidence suggests that the beginning of the permanent settlement of Sela dates from Iron Age II, when the techniques of water management (catchment, distribution and storage) were mastered. Accordingly, Sela was only habitable on a stable basis if water was properly managed, and this was only possible by means of the construction of water channels, reservoirs and cisterns. The available evidence suggests that water management in Sela began during the Late Iron Age and continued until the Ottoman period. 3. THE HYDRAULIC STRUCTURES OF SELA A notable feature of Sela is the presence of approximately a hundred cisterns, water reservoirs perforated or carved into the sandstone, presumably filled with rainwater through surface channels incised in the rock. These cisterns are exceptional for their variety; some of them are carved into the sandstone in the ground, and others are cut into the standing sandstone boulders. The shapes and sizes vary, as does the level of sophistication: some are technologically complex, with settling or sedimentation basins for stopping sediment and for separating mud and sand before the water was allowed into the main cisterns (Fig. 4). Figure 4. House with canals, sedimentation basin and cistern D15 Some cisterns are more or less oval in shape; others are rounded, and others still are rectangular or square, with corners forming right angles. Some have plaster or water-proof mortar on their sides and/or a 42 R. DA RIVA et al. Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 slab of stone to close them; these additions would have helped to improve the quality of the water, lengthen the potential storage time, and reduce the loss of water through evaporation. It is plausible to think that the structures were reused over time, and that mortars were added during different periods of use; so a cistern may have been manufactured in one particular period, but the mortar currently present in it may be more recent. The variability in size and morphology of the cisterns may respond to chronological differences. We assume that different artisans in different periods used different technologies and elements to manufacture the cisterns and their mortars, which implies that the system was used in more than one period. However, we did not attempt to establish an accurate and definitive typology of the structures because of the difficulty in obtaining a 14C-dating and the lack of independent archaeological evidence. Most of these cisterns are now filled with debris and sediment, so there is no way of dating or analysing them without undertaking an archaeological excavation. In some of them a preliminary inspection was performed using a Kong Cevedale tripod (Fig. 5). In addition to the cisterns, a series of rock-cut channels and pipelines were detected in different areas of the site. Some of the channels are more than 20 m in length, and they seem to have been used to funnel the water into the cisterns. All these elements suggest the existence of an extremely sophisticated and well-developed system for harvesting and storing rainwater using water channels, pipes, pools, and underground cisterns. Figure 5. Studying cistern D18 in September 2018 Water structures have been documented in the promontory of as-Sila as well as in the wadi area below (areas A and B in Fig. 3 above). The samples from the analysis published here were obtained in the course of the 2016 campaign in areas F, G, H, K and L. Areas F, G and H are located to the north and east of the site, near the monumental entrance to the summit, while areas K and L lie to the south and southwest respectively. Structures D13 and D115 are located in area F. The entrance area (F) includes a monumental gate, a hollow tower (perhaps a cistern, or maybe a silo, identified as D10, see Fig. 28 below), and some “houses” with stone-walls (Lindner et al., 2001, 252258). The gate is rock-cut, with a reinforcement of stone ashlars to build the walls. D01 (Fig. 6), D03 (Fig. 7), D05, (Fig. 8), D09 (Fig. 9), D16 (Fig. 10) and D63 are THE MORTARS FROM ROCK-CUT HYDRAULIC STRUCTURES OF AS-SILA (SELA) IN SOUTHERN JORDAN 43 Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 from area G. In area G we found the rock-cut layout of a large structure, some houses and cisterns. D22 (Fig. 11) and D107 are from area H. This area presents two large rock-cut structures, one of which at least seems to have functioned as a sanctuary: there is an altar and a sort of sacrificial pile or baytilos (Lindner et al., 2001, 263). D38 (Fig. 12) is from area K. The central area K is the largest at the site; here, we found numerous cisterns, rock-cut houses and also a canal. Surface finds (pottery, stone and metal objects) coincide with those of other areas of the site. D57 and D59 are from area L. In this western part, we found a tower and some rock-cut structures, including houses and cisterns. We also detected a limestone outcrop, which was probably the quarry for the plaster used for the walls and floors of the cisterns and the houses. Near the limestone outcrop, we found iron slags and some fragments of basalt objects. Figure 6. Cistern D01 44 R. DA RIVA et al. Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 Figure 7. Cistern D03 Figure 8. Cistern D05 The observations below regarding the morphology of the cisterns are preliminary; a detailed study could not be carried out, as most of the cisterns were filled with sand, debris and/or vegetation, and none of the structures were excavated. D01, D05, D22 and D63 lack a regular shape; the cisterns are completely filled so they could not be studied or measured. D03 and D09 are both oval in shape; D03 is not silted, and THE MORTARS FROM ROCK-CUT HYDRAULIC STRUCTURES OF AS-SILA (SELA) IN SOUTHERN JORDAN 45 Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 they were both inspected in September 2018. D13 is filled with some vegetation and debris; it has a circular opening and four canals are connected with it. These canals are covered with rectangular sandstone slabs. The canals and the cistern were inspected using a tripod in September 2018 (Fig. 13). D16 is partially filled; it is nearly oval in shape, and it was inspected in 2018. Cisterns D38, D57 and D59 (filled) are also oval-shaped. D107 and D115 are indeterminate, but the presence of lime-based mortars points to their use as hydraulic structures, and so samples were taken from these two structures. Figure 9. Cistern D09 52 R. DA RIVA et al. Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 Figure 15. Optical micrograph of the thin section of lime mortar type 1. The image in plane polarized light shows fragments of detrital quartz as aggregate and the lime binder matrix (Soto, 2017, p. 13, fig. 2.2.) Figure 16. XRD patterns for the categories of association of crystalline phases as detected by XRD of mortar type 1 (Soto, 2017, p. 14, fig. 3) THE MORTARS FROM ROCK-CUT HYDRAULIC STRUCTURES OF AS-SILA (SELA) IN SOUTHERN JORDAN 53 Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 For type 2 (Fig. 17), the binder is defined as micrite (1–4 µm) or microsparitic calcite, or a combination of both, in proportions of 60–70% with respect to the total sample. Aggregates include sub-rounded or subangular detrital quartz and fragments of sparitic limestone, both in proportions of 2–5%. Smaller amounts of red sandstone, crushed ceramics, plants and hematite (1–5%) are present. Some of the samples of this group show vuggy porosities (1mm) in proportions of 10–20%. The porosity of the mortars decreased in time due to the process of carbonation, as a secondary calcite gradually filled the pores. Interestingly, in experiments carried out with lime mortar prepared in the laboratory, the strongest specimens were the ones with the highest porosity because this characteristic allows a faster and more complete carbonation (Vyšvařil, et al., 2017). The XRD indicates the presence of quartz and calcite as the main mineralogical components in the preparation of this mortar (Fig. 18). Figure 17. Optical micrograph of the thin section of lime mortar type 2. The image in plane polarized light shows vuggy porosities in the lime binder matrix (Soto, 2017, p. 25, fig. 20.2.) Figure 18. XRD patterns for the categories of association of crystalline phases as detected by XRD of mortar type 2 (Soto, 2017, p. 25, fig. 21) 54 R. DA RIVA et al. Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 The mortar binder in type 3 (Fig. 19) is mostly micrite, in amounts of 70–80%. The main aggregates are subangular detrital quartz (5-10%) and microsparitic limestone fragments (2–5%), and, to a lesser extent, rock with biomicritic texture (composed of bioclasts and micritic matrix), plant remains, ashes, massive hematite, and fragments of crushed ceramics. Some samples show fissures, probably related to drying processes. In this case, the binder/aggregate ratio is 3/1. The XRD of this mortar confirms quartz and calcite as the dominant minerals in the preparation (Fig. 20). Figure 19. Optical micrograph of the thin section of the lime mortar type 2. The image in plane polarized light shows intraclasts in the lime binder matrix (Soto, 2017, p. 37, fig. 40.2) THE MORTARS FROM ROCK-CUT HYDRAULIC STRUCTURES OF AS-SILA (SELA) IN SOUTHERN JORDAN 55 Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 Figure 20. XRD patterns for the categories of association of crystalline phases as detected by XRD of mortar type 3 (Soto, 2017, p. 38, fig. 41) Finally, for types 4 (Fig. 21) and 5 (Fig. 22) the binder consists of microsparitic calcite in proportions of 5060% with respect to the total sample, and the main aggregates are fragments of microsparitic limestone and sub-rounded detrital quartz (10-30%). The main differences are in the additional aggregates, plant remains (2–5%), fragments of crushed ceramics, carbon and ashes in proportions that vary between 2% and 5% in the only sample of type 4; and fragments of gastropods, gypsum and ashes in variable portions of 1– 5% are visible for type 5. The proportion of binder in these two types is lower than in the total sample of the mortars studied and seems to be a third different recipe used for the preparation of the mortars analysed in this study. Diffractograms of these two types show quartz and calcite as the dominant mineralogical components of mortar type 4 (Fig. 23) and also dolomite in the case of type 5 (Fig. 24). 56 R. DA RIVA et al. Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 Figure 21. Optical micrograph of the thin section of the lime mortar type 4. The image in plane polarized light shows quartz and limestone as aggregates in the lime binder matrix (Soto, 2017, p. 39, fig. 43.1.) Figure 22. Optical micrograph of the thin section of lime mortar type 5. The image in plane polarized light shows quartz and limestone as aggregates in the lime binder matrix (Soto, 2017, p. 39, fig. 45.3) THE MORTARS FROM ROCK-CUT HYDRAULIC STRUCTURES OF AS-SILA (SELA) IN SOUTHERN JORDAN 57 Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 Figure 23. XRD patterns for the categories of association of crystalline phases as detected by XRD of mortar type 4 (Soto, 2017, p. 40, fig. 44) Figure 24. XRD patterns for the categories of association of crystalline phases as detected by XRD of mortar type 5 (Soto, 2017, p. 41, fig. 46) 58 R. DA RIVA et al. Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 5.2. Mortar Radiocarbon Dating We are well aware of the difficulties presented by the methods currently used to date mortars (Hayen et al., 2017; Hajdas et al., 2017; Urbanova et al., 2020). Despite all the problems posed by the dates obtained, and the conviction that they do not provide us with a reliable absolute chronology of the mortars, we believe it is necessary to offer the most complete analysis possible. In this article we present new chronological data in the hope that future techniques will make it possible to obtain a wholly reliable absolute dating. The idea behind the radiocarbon dating of mortars is clear: immediately after the preparation of mortar a chemical reaction begins in which the Ca(OH)2 of the mortar reacts with CO2 from the atmosphere to produce calcium carbonate (CaCO3), creating a hard and durable binder. During this hardening process, the mortar fixes carbon from the atmosphere in its structure, and a radiocarbon signal is thus automatically created which is related to the time of building. Unfortunately, despite the efforts of researchers, mortar is still among the most difficult materials for radiocarbon dating to assess, due to its complex composition and certain other phenomena that can affect its radiocarbon content. The first problem appears when the hardening process takes years or even decades to finish, a situation which disrupts the assumed temporal association with the time the structure was built (Hale et al., 2003). A second problem appears due to the interaction of the structure with water; this may lead to the presence of deposits of carbonate transported by the water and very likely unrelated to the carbonate originally present in the structure. As a consequence, this may alter the age of the mortar. Another issue arises from the possible calcination remains coming from the original limestone, which has no radiocarbon at all. Finally, it is very common to find more geological carbonate in the aggregates used in the production, thus adding a new problem to the system (Marzaioli et al., 2013; Lubritto et al., 2015). For all these reasons, extreme care must be taken when using mortar as a radiocarbon sample. A thorough mineralogical and petrographic characterization of the samples is essential prior to the specific treatment for radiocarbon dating. At present there are two main strategies for treating and dating mortar: sequential acid dissolution and forced suspensionbased methods, including the Cryo2Sonic method (Hayen et al., 2017). Both try to separate the anthropogenic carbonate generated during the hardening of the lime mortar from other sources of carbon which are in fact contaminants: residual carbonate from incomplete combustion of the primitive lime, and carbonates added as aggregates. The first option assumes that the anthropogenic carbonate reacts faster to acid than the rock carbonates (Van Strydonck et al., 1986). Several fractions of carbon dioxide are analysed from the same mortar sample and the results are used in combination in order to obtain the age of the mortar (Hajdas et al., 2017). The second option treats the sample by repeatedly freezing and thawing, and then grinding, and sieving to separate the anthropogenic fraction and then perform a single radiocarbon analysis (Nawrocka et al., 2005). The mortars in this study were prepared and measured at CNA (Centro Nacional de Aceleradores) in two different batches. The first batch comprised samples CNA4189–4193 and was prepared before the Cryo2Sonic method was first introduced at the CNA lab. These mortars were prepared using a standard carbonate sample preparation procedure as follows: after visual inspection, a small aliquot was taken and a soft leaching with HCl was applied to eliminate the outer part of the sample, which is more likely to be affected by external carbonates or recrystallization processes. Then, about 15 mg of carbonate material was used for graphitization with a CHS-AGE system (Wacker et al.(2), 2010; Wacker et al., 2013). The second batch comprised samples CNA4391–4397 and was prepared by the suspension-based method. Cryo2Sonic is the best known version of the suspension-based method (Marzaioli et al., 2013; Addis et al., 2019). Briefly, the mortar is sequentially introduced in liquid nitrogen for five minutes and on a stove at 80ºC, also for five minutes. The cycle is repeated three times to make the mortar brittle and easy to break. The sample is then wet sieved to select fine grain under 500 m and the material is allowed to settle until complete sedimentation. Then, the sample is ultrasonicated for 30 minutes and the resuspended material is collected by siphoning and discarded. After the addition of more water, the rest of the material is again allowed to settle, and the material resuspended in a second ultrasonication is selected, centrifuged, and dried. The carbonate fraction selected is then graphitized in a CHS-AGE system. In both batches, about 15 mg of the carbonate is dissolved in phosphoric acid and the carbon dioxide produced is transferred to the reactors of the AGE system, where it is transformed into graphite, which is then pressed in a sample holder and is ready for AMS measurement in the Micadas system at CNA (Synal et al., 2007). AMS targets are analysed in a batch containing unknown samples, standard samples for normalization, and blank samples for background correction. Both standard and blank samples are used to correct the AMS measurement of unknown samples, since AMS does not directly give the nominal value of the samples. In the AMS measurement the concentrations of THE MORTARS FROM ROCK-CUT HYDRAULIC STRUCTURES OF AS-SILA (SELA) IN SOUTHERN JORDAN 59 Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 the three carbon isotopes, 12C and 13C, which are stable, and 14C, which is radioactive, are determined, since all of them are needed to obtain the age of the sample. The final analysis of the data is performed using the BATS tool (Wacker et al., 2010(1)) to obtain the final Radiocarbon Ages and 13C values, which are also measured in the AMS system. The 13C parameter indicates the relative concentration of the stable isotopes and is used to correct the fact that different coetaneous materials contain slightly different radiocarbon concentrations. 13C is included in the calculation of the Radiocarbon Age, which is performed as defined by Stuiver and Polach (Stuiver and Polach, 1977), and is the experimental result of the dating process. It is expressed in years BP (Before Present) and is defined under several assumptions which are known not to be true; thus, the Radiocarbon Age does not express a real calendar age. The main assumption is that the atmospheric radiocarbon concentration has remained constant at a specific value through history. In order to transform the Radiocarbon Age into a calendar age it is necessary to carry out a calibration procedure. This consists essentially of a comparison of the Radiocarbon Age obtained for the sample with the values of the Radiocarbon Age of a set of samples with a known calendar age (e.g., tree rings independently dated by dendrochronology), which form the calibration curve. The calibration procedure can be performed using several free access software programs; here we used CALIB software (Stuiver and Reimer, 1993) with the IntCal13 calibration curve (Reimer et al., 2013). Calibration is performed at the standard level of 2 (95.4% confidence level). The calibration result is a set of age ranges and the probability that the real calendar age of the sample belongs to each specific range. Calibrated ranges should always be used to follow the archaeological analysis. The results are shown in table 2. For each sample a unique lab code (CNA#) is assigned identifying the sample through the process in the laboratory. Radiocarbon age and 13C are presented as the values obtained from the AMS measurement and the following data analysis. Calibrated results are presented as the whole set of age ranges with their probability given by the calibration software (Millard, 2014). The data are presented in chronological order to simplify the archaeological analysis. A grey background is used for the first batch of samples which were prepared following a simple carbonate pretreatment. Fig. 25 presents the probability distributions of the calibrated ages for the twelve samples graphically to provide a schematic view of the historical periods obtained in the results. Table 2. Results of Mortar Radiocarbon Dating in chronological order. Samples with a grey background belong to the first batch of samples in which a standard carbonate pretreatment was applied. The Cryo2Sonic procedure was applied to the rest of the samples.  13C values in the table were obtained from the AMS measurement. CNA# User Code Age BP 13C Calibrated ranges 4192.1.1 SL16.D16.2 3 15030 -11.5 1498–1382 BC (90.0%) 1340–1310 BC (10.0%) 4191.1.1 SL16.D38.9 2 98030 -8.7 1371–1359 BC (1.2%) 1297–1113 BC (98.8%) 4393.1.1 SL16.D03.5 2 89030 -4.6 1192–1170 BC (3.3%) 1165–1144 BC (3.3%) 1131–977 BC (93.4%) 4193.1.1 SL16.D107.13 2 66030 -12.2 895–868 BC (8.7%) 857–854 BC (0.6%) 850–794 BC (90.7%) 4394.1.1 SL16.D57.10 2 65030 -9.6 894–870 BC (6.3%) 849–792 BC (93.7%) 4392.1.1 SL16.D22.8 2 50030 -13.0 787–699 BC (27.9%) 696–540 BC (72.1%) 4396.1.1 SL16.D09.6 2 30030 -6.5 404–356 BC (82.1%) 286–235 BC (17.9%) 4395.1.1 SL16.D59.11 1 26025 -8.1 670–778 AD (92.9%) 791–805 AD (2.1%) 812–826 AD (1.7%) 840–862 AD (3.3%) 4391.1.1 SL16.H1.UM3.2 (D115) 1 00030 -18.1 983–1049 AD (82.2%) 1086–1124 AD (14.5%) 1137–1150 AD (3.3%) 4189.1.1 SL16.H1.UM3.1 (D115) 98030 -21.1 993–1055 AD (51.0%) 1077–1153 AD (49.0%) 4190.1.1 SL16.H1.6.169 97030 -21.7 1018–1059 AD (36.3%) 60 R. DA RIVA et al. Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 1065–1154 AD (63.7%) 4397.1.1 SL16.D13.1 91025 -12.7 1035–1189 AD (99.1%) 1199–1202 AD (0.9%) Figure 25. Calibrated age ranges at the 2  confidence level. Names of the samples have been shortened for convenience, eliminating the common SL16. part of the original user codes. Table 3. Combined results of the archaeological, petrographic, mineralogical and radiocarbon analyses Sample CNA# Hydraulic structure Type of mortar Type of structure Chronology Area of Sela 4391.1.1 H1.UM3.2 (D115) 1 Indeterminate (tank?) X-XII centuries AD F 4189.1.1 H1.UM3.1 (D115; reform) 1 Indeterminate (tank?) X-XII centuries AD F 4190.1.1 H1.6.169 4 Indeterminate (floor?) XI-XII centuries AD F 4397.1.1 D13 3 Cistern with circular opening X-XIII centuries AD F 4192.1.1 D16 2 Cistern with oval shape XV-XIV centuries BC G THE MORTARS FROM ROCK-CUT HYDRAULIC STRUCTURES OF AS-SILA (SELA) IN SOUTHERN JORDAN 61 Mediterranean Archaeology and Archaeometry, Vol. 21, No 2, (2021), pp. 37-67 4393.1.1 D03 2 Cistern with oval shape XII-X centuries BC G 4396.1.1 D09 2 Cistern with oval shape V-III centuries BC G 4193.1.1 D107 5 Indeterminate (pool?) IX-VIII centuries BC H 4392.1.1 D22 3 Cistern with irregular shape VIII-VI centuries BC H 4191.1.1 D38 3 Cistern with oval shape XIV-XII centuries BC K 4394.1.1 D57 2 Cistern with oval shape IX-VIII centuries BC L 4395.1.1 D59 3 Cistern with oval shape VII-IX centuries AD L 6. DISCUSSION As can be seen in Tables 2 and 3, some results coincide chronologically with various periods that are represented by the surface material, and so they can be considered plausible. Types 1 and 4 are only documented in mortars dated between the X and the XII centuries AD, type 2 in mortars dated XV-III centuries BC, type 3 in XIV-VI centuries BC and VII-XIII centuries AD, and type 5 in IX-VIII centuries BC. These are dates in which Sela was occupied, judging by the type of surface pottery found: 4189-4190, 4193, 4391-92, 4394-4397. Some dates cluster around the XI-VI centuries BC (4192-4394, 4393, all of them corresponding to mortar type 2); there is another slightly more recent group in V-III centuries BC (4396, corresponding also to type 2, which is not found in mortars from other periods), and a third group can be dated between the VII and the XIII centuries AD (4189-90, 4391, 4395 and 4397, corresponding to mortar types 1, 4, 1, 3 and 3 respectively). Note that samples 4189 and 4391 coincide chronologically as well as in terms of the type of mortar (type 1), and both belong to the indeterminate hydraulic structure D115, which was excavated at House 1 during the 2016 campaign. Interestingly, mortar sample 4391, slightly more recent than 4189, belongs to a reform that was carried out in structure D115. Sample 4189 was obtained using the standard carbonate sample preparation procedure, and 4391 by Cryo2Sonic. From the same house sample 4190 was obtained, corresponding to mortar type 4, , and this was also analysed using the simple method 5 . Three results (4191-92, 4393, corresponding to mortar types 2 and 3) range between the XV and the XII centuries BC, so they are older than the oldest pottery documented on the surface of the site. In our opinion it is unlikely that Sela was occupied permanently in the Late Bronze age; if 5 Note that these dates are noticeably older than the 14C dates of the wood samples taken from House 1, for which the following absolute chronologies were obtained: sample CNA4194.1.1. from Layer 5 (Cal AD 1643-1682); sample CNA4195.1.1. from Layer 6 it had been occupied one would expect to find surface materials such as ceramics, but there are none. As for the methods used, the simple method (the standard carbonate sample preparation procedure) which is shaded in grey in Table 1, gives two unlikely and three likely results, whereas all the results obtained by Cryo2Sonic are more or less likely. This is in agreement with our current knowledge and sample preparation techniques in mortar dating, in which very specific preparation methods are used. Concerning the types, the differences observed especially in the binder/aggregate ratio suggest the use of at least three different recipes for the production of the lime-based mortars: one in which the binder represents 70–80% of the total sample (type 1 and 3); a second one in which it represents 60–70% (type 2); and a third one in which it accounts for approximately half (types 4 and 5). Moreover, intragroup variations are also observed. All the mortars show aggregates that would increase consistency and strength, such as crushed ceramics, ash and plants. These differences may be due to the techniques used by the various artisans or to the chronological differences indicated by radiocarbon dating. To sum up, even if the dates obtained are not totally unreasonable, what is in doubt here is the reliability of the currently available methods for dating mortars, which determine the results. An additional problem is that we lack independent chronological indications for dating the mortars or cisterns in Sela, such as morphology, type of manufacture, etc., able to confirm or refute the results obtained. For example, sample CN4392 from cistern D22 is dated between the VIII-VI centuries BC, which is in full agreement with, for example, the evidence from the Nabonidus inscription. But we lack an independent date for D22. In our opinion, all these data need to be confirmed in the (Cal AD 1486-1604); and sample CNA4196.1.1. from Layer 10 (Cal AD 1451-1529), see Da Riva et al., forthcoming. But we stress that wood samples can also be problematic, as this material is often intrusive.