1Located in the highland valley of Dasht-i Morghab (1,900 m above mean sea level; fig. 1 and fig. 2), the site of Pasargadae was founded sometime after 547 BC as the first residence of Cyrus the Great (559-530 BC). The site marks the royal seat of the vastly expansive range of the early Achaemenid Empire and presents a curious park-like appearance with its orderly designed garden in which sparsely located palaces and buildings are to be found (Stronach, 1994; Boucharlat, 2009). Although under the reign of Darius, Persepolis was chosen as a new, ex-nihilo foundation in the wider, more open valley of Marv Dasht, some 40 km downstream from Pasargadae; the site of Pasargadae probably retained a certain role of importance, both as a continued centre of political (ritual) importance (Briant, 1996; Boucharlat, 2006) as well as a main stronghold and treasury location (Stronach, 1978). The site maintained this role throughout the Achaemenid period, up until the arrival of Alexander, who took possession of the treasury and whose historiographers were struck by the vivid image of a still well maintained park or ‘Paradeisos’ surrounding the royal tomb of Cyrus (see esp. Arrian).
2In the absence of any larger archaeological sites or major tells in the area, little information is available on more mundane settlements or the countryside supporting Pasargadae, let alone its role as an urban centre in the small highland valley. In recent years however, rescue excavations in an area to the south of Pasargadae, in the smaller valley named Tang-i Bolaghi, have provided ample information from at least three smaller sites of probable rural character (Askari Chaverdi and Callieri, 2006; Kaim et al., 2007; Asadi and Kaim, 2009; Helwing and Seyedin, 2009) and certainly one site of a more residential and ‘elite’ nature (Atai and Boucharlat, 2009). Together with rock-cut irrigation canals found alongside these sites (Kleiss, 1991; Atai and Boucharlat, 2009), these recent findings provide promising information of a more diverse as well as diachronic framework, steadily extending our appreciation of the Pasargadae countryside. Furthermore, along the northern and eastern fringes of the Pasargadae plain, a set of five monumental dam remains, constructed within thesame Achaemenid period, hint at the scale of investments in regional infrastructure. While remains of similar endeavours in the later periods have not yet been encountered, these dams (with their date now confirmed by the recent fieldwork) present an important footprint of human impact in reshaping the region. First reported on by German archaeologist and architect W. Kleiss (1982, 1987, 1988, 1991, 1992, 2000), the dams are predominantly found in dry riverbeds (fig. 2 and tab. 1) and must have acted principally as check dams, preventing violent flash floods that occur in winter and spring time. In the framework of the Joint Iranian-French Archaeological Project at Pasargadae, these dam sites were studied in detail and rescue excavations or surface documentation was carried out on those sites carrying architectural remains, Sad-i Shahidabad and Sad-i Didegan.
3Within the scope of this article, we will focus on the latter line of archaeological evidence, describing the dam site of Sad-i Didegan, located 22 km to the north of Pasargadae. Within this case study, the geomorphological pre-dam environment is reconstructed, the dam’s function as well as the construction rationale and logistics are studied and the hydrological and sedimentary impact of the dam is reviewed. This approach combines archaeology and geomorphology to better understand the choice of site location for Sad-i Didegan, the construction methods used, the functioning, hydrological and sedimentological impact and the wider level of investment it represents in the region.
Fig. 1 – General map of Iran.
Fig. 1 – Carte générale de l’Iran.
1: elevation <500 m; 2: elevation between 500 m and 1000 m; 3: elevation >1000 m; 4: Kur River basin; 5: state borders; 6: cities; 7: mentioned lake sites.
1 : altitude < 500 m ; 2 : altitude comprise entre 500 m et 1000 m ; 3 : altitude > 1 000 m ; 4 : bassin-versant du Kur ; 5 : frontières ; 6 : villes ; 7 : lacs mentionnés dans le texte.
Fig. 2 – General geography of the wider Pasargadae area.
Fig. 2 – Carte générale de la région de Pasargades.
1: dry riverbeds; 2: perennial river; 3: anticlines; 4: Achaemenid dam sites; 5: Sad-i Didegan catchment; 6: extent of fig. 2.
1 : écoulements temporaires ; 2 : écoulements pérennes ; 3 : anticlinaux ; 4 : barrages achéménides ; 5 : bassin-versant du Sad-i Didegan ; 6 : extension de la fig. 2.
4The study area is situated within the context of the Kur River Basin (or Neyriz drainage), one of the largest endorheic basins in the Zagros fold-thrust belt (fig. 1). Within the basin, the smaller Rud-i Polvar is the principal tributary to the Kur River, draining a catchment of ca. 9,000 km2, with peaks rising as high as 3,853 m and the lowest point at 1,592 m above mean sea level (at the confluence with the Rud-i Kur). Adding to the structural northwest to southeast oriented drainage of the Zagros mountains and the Rud-i Kur in particular, the Rud-i Polvar presents a transverse drainage pattern (fig. 2). Along its course, the river cuts across two dominant anticlines of Jurassic, Cretaceous up to Miocene age (Tsuneki and Zeidee, 2008). Like many other canyons in the Zagros (Oberlander, 1968), this drainage system originates in a complex interplay of antecedence and superposition (Rigot, 2010). The resulting canyons (labelled 'Tang' in Persian) are narrow and deeply incised, at times only housing the riverbed.
5Squeezed in between these main folds of the Zagros mountains, discrete alluvial valleys present the main areas for agriculture and settlement as early as the proto-Neolithic and Neolithic (Goff, 1964; Bernbeck et al., 2005). The plain of Pasargadae, bordered to the north by the Tang-i Hana and to the south by the Tang-i Bolaghi, represents one of the larger valleys, measuring 25 km in length by 12 km across. As recent studies by M. Kehl and A. Skowronek (2002), M. Kehl et al. (2005, 2009) and J.-B. Rigot (2010) have shown, the 16 m high main alluvial terrace (T1) within these plains was well established by the Holocene Climatic Optimum. As further discussed by J.-B. Rigot for the Tang-i Bolaghi and confirmed by M. Kehl’s observations along the Rud-i Kur, a second and last major period of terrace building (T2) is probably to be dated to the second half of the first millennium AD, corresponding to the Sasanian or Early Islamic period, with the T2 terrace level some 10 m below the main terrace. Within the higher reaches of the river, to the north of the Tang-i Hana anticline, Holocene alluvial deposits are only marginal and occupy a narrow strip along the riverbed, not wider than 2 km. In contrast to the Pasargadae plain, these highlands present a barren landscape of largely Early Pleistocene glacis material, gently sloping down from the high Zagros anticlines (Gray, 1949). Within these upland areas (>2,000 m), vegetation cover is extremely limited and presents anArtemisia steppe vegetation, characteristic for the central Iranian plateau (Zohary, 1963). While a less favourable location for major agricultural deployment, the upland area corresponds to the main drainage area for the Rud-i Polvar and the higher mountains retain important snow cover in wintertime.
6The general climate of southern Iran, and that of the northern Fars area in particular, is semi-arid characterised by hot dry summers and rather cold winters. As for any other semi-arid region, precipitation generally shows a high inter-annual variability with a mean annual precipitation and temperature of 350 to 500 mm and ca. 18°C respectively. In the absence of any nearby available meteorological station, these data derive from the Zarghan and Dorudzan stations, located 60 km to the west of Pasargadae (http://www.irimo.ir), though Pasargadae must show similar patterns (Ehlers, 1980). The rainfall season is largely confined to the December to May period and shows episodic rains (and snow at higher elevations). Exceptional storms, originating from the Mediterranean and travelling southeast through the Zagros, can furthermore produce erratic floods, with resulting river discharge up to 32 times the average flow. An example being that of the Rud-i Kur, with 845 m3/s on December 7th 1954 (Justin and Courtney, Irrigation Corporation of Iran, 1965). Although hydrological figures are only available for the Rud-i Kur, the smaller Polvar river catchment must present a similar hydrological pattern, with a parallel importance of spring rain and melt-water runoff from the higher elevations in the catchment. Firstly, the latter restock the groundwater reserves, sustaining perennial river base flow throughout the summer months. Secondly, within the geomorphological framework described above, these discharges also provoke significant spring floods and overbank deposits extending from the highly incised perennial river-channels. Such floods are still vivid in living memory, and further traces can be found in historical sources (Houtum-Schindler, 1891); they were moreover observed at archaeological sites in the Tang-i Bolaghi small alluvial basin (Helwing et al., 2010). Finally, while climate proxy for the Fars area has only recently become available, evidence from long established cores from the wider Zagros mountains such as at Lakes Zeribar and Mirabad (Van Zeist and Bottema, 1991; Stevens et al., 2001) indicate the present day climate to have been well established by 4.5 ka BP, much earlier than the Achaemenid period. Although further published proxy results are awaited, preliminary pollen analysis from core sequences at Lake Maharlou near Shiraz (Djamali et al., 2009) might additionally indicate that the first millennium BC was a period of increased aridity. Though of course very susceptible to minor changes, local variation or human impact on vegetation cover, the limited records to date would thus corroborate more regional analyses (Roberts et al., 2011). For the time being we can presume that the Achaemenid landscape would have been quite similar to the present day situation, with comparable climatic conditions and hydrological regimes.
7Within these, or even drier conditions, it is not surprising that several important check dams were installed, both on perennial rivers (in the case of Sad-i Shahidabad) as well as in dry tributary beds. The location of several dams upstream of the Tang-i Hana anticline, with no clear exploitable or irrigable arable land nearby, additionally shows that these dams are to be viewed as passive, protective measures, in operation for the more economically valuable lands downstream, in the vicinity of Pasargadae.
8Situated in one of the smaller dry river beds to the north of the Tang-i Hana anticline (fig. 3), the Sad-i Didegan site location is that of a relatively small catchment of ca. 46 km2 and of modest proportions when compared with catchments of other known Achaemenid dams (tab. 1). The higher grounds of the catchment are dominated by the previously described gently sloping Early Pleistocene glacis, with limited vegetation cover. Within the basin, only a handful of present day hamlets are to be found and agricultural surfaces are low in number. Few archaeological sites are presently known, though no extensive surveys have so far been undertaken.
Fig. 3 – Geomorphological map of the Sad-i Didegan zone and its main units.
Fig. 3 – Carte des unités géomorphologiques de la zone de Sad-i Didegan.
1: limestone; 2: old pediment; 3: young pediment; 4: old terraces; 5: young terraces; 6: Holocene colluvial deposits; 7: dam remains; 8: entrenched track; 9: present-day tracks.
1 : calcaires ; 2 : glacis ancien ; 3 : glacis récent ; 4 : terrasses anciennes ; 5 : terrasses récentes ; 6 : dépôts colluviaux d’âge holocène ; 7 : vestiges du barrage ; 8 : chemin creux ; 9 : pistes actuelles.
Tab. 1 – Characteristics of the Achaemenid dams in the Pasargadae area.
Tab. 1 – Principales caractéristiques des barrages de l’époque achéménide dans la région de Pasargades.
Name
|
Dimensions (length x width x height)
|
Catchment surface
|
Catchment elevation
|
Dating Evidence
|
Canal traces
|
Sad-i Shahidabad
|
700 m x 50 m x 15 m
|
4,875 km2 (perennial river)
|
2,060-3,853 m
|
(Early-) Achaemenid (excavated, 14C)
|
Dam crown
|
Sad-i Didegan
|
150 m x 105 m x 21 m
|
46 km2
(dry riverbed)
|
2,040-2,442 m
|
Achaemenid (architecture, 14C)
|
Dam base
|
Sad-i Alafi 1
|
215 m x 45 m x 6 m
|
241 km2
(dry riverbed)
|
1,950-3,228 m
|
Achaemenid (14C, bone)
|
Disappeared
|
Sad-i Alafi 2
|
180 m x 35 m x 4 m
|
186 km2
(dry riverbed)
|
1,950-3,318 m
|
First millennium BC (relative)
|
No traces
|
Sad-i Tang-i Saadatashahr
|
400 m x 40 m x 8 m
|
423 km2
(dry riverbed)
|
1,856-2,569 m
|
First millennium BC (relative)
|
No traces
|
9Located on the northern flank of the Kuh-i Tang-i Hana mountain range (with highest peaks at 2,350 m), the southern end of the Sad-i Didegan valley is visually dominated by the exposed dark greyish, bare limestone hill slopes, rising 210 m above the valley bottom (fig. 3). The limestone is closely bedded and of Middle Cretaceous date, as evident from the included Orbitolinas fossils (Geological Survey & Mineral Exploration of Iran, s.d.). For the most part folded and uplifted in the Mio-Pliocene period (Harrison, 1968; Berberian and King, 1981), the bedrock is highly fractured with beds dipping in a south-western direction, at an angle up to 30°. As evidence for high tectonic activity, faults cut across the beds in an orthogonal way, leaving evident fault planes with fault mirrors and related deposits of fault breccia noted throughout the limestone exposure. A second area of main tectonic activity is the actual narrow canyon, which cuts across the limestone bedrock and shows a major strike slip fault, with a notable shift causing the western limestone to advance northwards. The resulting prominent outcrop of the limestone bedrock is responsible for the asymmetric structural topography of the valley and the sheltered location of the canyon entrance. This has played an important role in the construction of the Achaemenid dam, the western end of which is anchored onto the limestone bedrock (fig. 4).
Fig. 4 – Selected pictures.
Fig. 4 – Photos sélectionnées.
A: Western face of the dam, anchoring on limestone bedrock, dam indicated by white dashed line. B: Eastern side of the dam, superimposing the old glacis material, dam indicated by white dashed line. The arrows point out the canal remains. C: Old pediment material. D: Preserved canal remains. E: Exposed section of the Holocene deposits. F: Displaced canal remains. G: Inter-valley canal.
A : Flanc ouest du barrage, sur le substratum calcaire, barrage indiqué par une ligne discontinue. Les flèches indiquent les traces du canal d’évacuation. B : Flanc est du barrage, construit sur le glacis ancien. C : Dépôts de glacis ancien. D : Vestiges du canal en place. E : Coupe dans les colluvions holocènes ; F : Restes remaniés du canal. E : Témoignage du canal reliant les bassins-versants des barrages de Sad-i Didegan et Sad-i Shahidabad.
10Dominating the surface of the slopes of the mountain, is a large deposit of gently sloping scree material, characterised by the presence of considerable calcrete banks of clast supported angular to subangular limestone talus debris, cemented by calcite (fig. 3 and fig. 4). A similar cover can be found on the higher grounds of the catchment and the wider slopes of the highland plain. Within the Sad-i Didegan valley bottom itself, several remains of this deposit can still be recognised and were mapped as old terraces (fig. 3). Just upstream of the canyon entrance, a residual hill is to be found on the left bank of the valley, rising 21 m above the valley floor (top surface at 2,065 m). It is built of erosion material derived from the Cretaceous limestone and shows a prolonged period of exposure, cementation and consolidation, thus hinting at an Early Pleistocene date. Superimposing the old glacis material is a similar angular to subangular limestone talus debris of a younger unconsolidated scree phase (fig. 3). The later deposit covers the lower elevations of the surrounding hillslopes and encircles the above-mentioned residual hill on both its eastern and southern side, isolating the older consolidated material from the limestone massif.
11Of presumed Late Pleistocene date is the fourth major unit encountered in the geomorphological survey, indicated as 5 on fig. 3. It consists of a 12-m thick deposit found as lower-lying infill on the valley floor, thus superimposing the infrequently encountered Tertiary mudstone in the valley bottom or building upon the older scree margins, described above. The unit is distinctly different from any preceding deposits and can be characterised as unstratified homogenous slightly fine sandy silt, with very few dispersed small angular to subangular gravel. It stands out as the most evident unit since its remnants are clearly marked by terraces, with a preserved surface elevation of ca. 2,055 m. The deposit resembles to a high degree the loess and loess-like sediments described by M. Kehl (2009) for the Iranian plateau in general and the Persepolis area in particular (Kehl et al., 2005). As also discussed by J.-B. Rigot (2010), this unstratified, unweathered deposit is to be correlated with a cooler and drier period or periglacial context in which the loess, derived from local or more distant mountains, was banked up on the sparsely covered slopes and valley floors of the Zagros mountains (Thomas et al., 1997). Subsequently, the fine silt was reworked, eroded and transported, and in so doing the hillslopes were fully deprived of this original fine cover. Such conditions are well known from the Zeribar and Mirabad pollen sequences and throughout Iran such loess deposits have now been securely dated in multiple studied sections (Karimi et al., 2009). Without any evident palaeosoils, this loess-derived alluvial deposition must have witnessed a high sedimentation rate and is attributed to the period of the Last Glacial Maximum (LGM). At that time the secondary or weathered loess was redeposited in lower lying areas, such as the Sad-i Didegan valley for which we can anticipate an ice-plug at the narrow canyon entrance to have caused the sedimentation of this fine silt, largely originating from areas upstream of the dam site.
12The final and most recent phase of morphogenetic dynamics, most probably of Holocene date, is presented by a phase of incision and colluvial reworking of the fine-silty Late Pleistocene terrace material (fig. 3). Covering a small surfaced incision and infill of the valley beds in between the terrace remnants, this deposit is 30 m in maximum width. The prevailing dynamics are still today of colluvial nature, as shown by the gently sloping terrace foothills and the absence of a clear established alluvial level. A studied section located in the incised wadi bed demonstrates the build-up of this deposit (fig. 4 and fig. 5; see fig. 6 for the section’s location). Field descriptions allow a tentative reconstruction of the main Holocene dynamics within the catchment. The section furthermore provided an opportunity to check for any dam related sediments or changes in sedimentological dynamics induced by the dam. The exposed section was 1.7 m high with the top of the section at an elevation of 2,049 m. In the upper part, the valley infill is built of slightly fine sandy silt with few dispersed gravels (<1 cm) and it shows three short interbedding deposits of imbricated, normal graded fine to medium subrounded to subangular gravel (<5 cm) set within a slightly fine sandy silt matrix (fig. 5). The lower part of the section is separated by a clear upper limit and it presents a homogeneous, slightly more clayish silt with only very few dispersed granules of limestone. At the actual base, again a more gravel-rich layer was encountered which in all likelihood corresponds to the base gravel of the valley floor. While the lower deposits highly reflect the Late Pleistocene material described above, the preliminary interpretation of the upper segment of the section shows a prevalence of colluvial material with only episodic and limited phases of observable major wadi flow. Finally, as a last phase, the reincision into this colluvial deposit must have occurred at a time of increased erosion within the basin itself or within the wider Rud-i Polvar and Rud-i Kur catchment. For the wider area, such an incision phase has been dated to the middle Holocene period or secondly, to the Sasanian to early Islamic era (Rigot, 2010).
13For the study area however, we have no evidence to firmly assign any date. Nor do the results unequivocally allow us to suggest the final reincision can be linked to a period of dam failure and renewed incision. For the time being we can merely observe the prevalence of colluvial material within the most recent deposits with only limited intervening periods of major wadi flow. In contrast to dam sites with century long sedimentation (Calvet and Geyer, 1992; Coque-Delhuille and Gentelle, 1997; Francaviglia, 2000) it is also interesting to note that no clear dam-related sediments are present. Most likely, the dam did not function as a sediment trap for a prolonged period of time. As a working hypothesis, we are to assume it only functioned seasonally and with a very limited storage volume.
Fig. 5 – Section in the Holocene colluvial valley bottom.
Fig. 5 – Coupe dans le fond de la vallée colluvial holocène.
1: colluvial deposits; 2: beds of gravel deposits; 3: homogeneous slight fine sandy clayish silt.
1 : dépôts colluviaux ; 2 : lits de graviers ; 3 : dépôts argilo-limoneux homogènes.
14During a short season of fieldwork in the spring of 2009, the Sad-i Didegan’s architectural remains and surrounding environment were carefully documented. No excavation proper was carried out, but the detailed mapping of the in-situ preserved blocks as well as of ca. 100 blocks, found displaced in the riverbed, allows for a solid reconstruction and study of the architectural remains and water-controlling infrastructure. Furthermore, the wider region around the site was plotted, surveyed and studied. Primary cartographic sources used for this were the available geological map (Geological Survey & Mineral Exploration of Iran, s.d.) and aerial photographs from the Iranian National Cartographic Centre (NCC, Summer 1998). Topographic maps of the NCC (1:25,000) were also used for the production of a local digital elevation model (DEM) and the estimation of the maximum reservoir extent and volume. The maps were supplemented by tiles from the global SRTM mission (GLCF, 2010) to provide a general regional evaluation of the catchment and the characteristics of the different dams and their catchments. The dataset was processed using the Soil and Water Assessment Tool (Neitsch et al., 2005) so that the individual catchments could be compared. For the study, the Sad-i Didegan remains were viewed in their natural context, with both the pre-construction environmental history and landscape dynamics studied, together with any possible dam-related deposits. The major geomorphological units of the southern zone of the Sad-i Didegan catchment were mapped and identified in the field, both from surface observations and from available exposures. Samples for laboratory analysis and optical stimulated luminescence dating (OSL) were collected from all main sedimentary units and construction-related deposits, and a similar approach was adopted for other dams in the region where at times, no definite archaeological elements for dating were available. Though fruitful results were to be expected (Shaw et al., 2007), the OSL samples did not reach the Ghent University laboratory. Unfortunately, they were confiscated at Tehran airport, despite the official written authorisation delivered by the Iranian Centre for Archaeological Research. Since 2009, no joint Iranian-French fieldwork or renewed sampling could be undertaken, hence no OSL dating results can be presented.
15Within the described natural framework of the small valley, the construction of the Sad-i Didegan dam can be contextualised and some crucial elements of site location and construction methods are now more easily understood. At present, remains of the original dam are found on the left and right bank of the valley, while those in the valley floor itself were swept away. Furthermore, in between the earliest visits by W. Kleiss in the 1980’s and the joint mission’s first reconnaissance in 2004, recent infrastructural works and machine excavators are responsible for a partial destruction of the site (fig. 4, fig. 6 and fig. 7). Also within the riverbed downstream of the dam a number of the (already dislocated) architectural remains were pushed from their original location in the riverbed, up the southern valley flank (fig. 6, zone C). On the right bank, traces of the dam are minimal, and the clearest remains of the original dam correspond to two piled concentrations of angular boulders, superimposing the natural limestone outcrops. In between these two boulder concentrations, a low mound of fine silty material rises up to 2,065 m (fig. 4). On the opposite side of the riverbed, on the contrary, the dam body is largely preserved (fig. 4). Anchoring onto the well-consolidated residual hill of old glacis material, a clear non-natural extension stretches 85 m to the northwest, with its axis clearly in line with those remains mapped on the western side rock outcrops. Similar to the western side, two packings of angular to subangular boulders can be observed at both the upstream and downstream side of the dam. These boulders are generally of 20 cm to 30 cm in diameter and make up the packing of the dam body. In between, the core of the dam is composed of homogeneous and compacted slightly fine sandy silt.
16The resulting overall image is that of an earth built gravity dam of ca. 90 m wide, 21 m high and with a crest length of about 150 m. The total original volume of the dam amounts to some 75,000 m3. Much like earlier dam examples such as the 3rd millennium BC dam of Sad-el Kafara in Egypt (Smith, 1971), or those examples from the flourishing 9th to 7th c. BC Urartian kingdom (Garbrecht, 2004), its construction is of a somewhat generic type, with two ‘walls’ or boulder masses backing up and protecting the dam’s core. In contrast to later Iranian examples of Mediaeval (Safavid) dams built within the actual canyon (Hartung and Kuros, 1987), the Sad-i Didegan dam is constructed 150 m north of the canyon. Its actual location is clearly not haphazard, reflecting both structural and practical considerations. First of all, the dam body anchors onto solid natural units with the limestone bedrock underlying it, and to the west of the dam. On the southeastern side, the old glacis calcrete provides a solid base. Furthermore, the chosen location allows for easy (maintenance) access on both the upstream and downstream side while finally, the necessary construction materials are found at the site, or within a few hundred metres. The outer walls amount to a total volume of 42,000 m3. The prime sourcing area probably corresponds to deposits of younger glacis materials, with a major exposure just downstream of the chosen site location (fig. 3). The earth core represents a total reconstituted volume of 33,000 m3 with sediments clearly related to the fine pseudo-loess deposits encountered in the Late Pleistocene terraces just upstream. Although no clear quarry fronts or pits could be observed, these deposits must again originate from close by. As a working hypothesis, we assume the eastern bank of the valley to originally have housed an extension of the Late Pleistocene terrace material. This terrace was then further built upon and quarried with a possible vague limit to the north, indicated on fig. 6. Additionally, a clear 400-m long entrenched track originates from this zone and encircles the old glacis hill along its northern slope, up onto its highest point (fig. 3 and fig. 6). With an average slope of 3.6%, this path would well have allowed the transport of these sediments to the highest point of the dam, from where it could be poured. As a remark in this regard, it is to be noted that the available absolute dating material derives from one of the sampled sections of the central infill. Here, charcoal from the well compacted and homogeneous silty construction produced a radiocarbon date of 2470±25 BP (sample KIA-40535), providing a first terminus post-quem for the construction of the site of 762-416 cal. BC [2σ, calibrated using OxCal 4.1 (Bronk Ramsey, 2009; Reimer et al., 2009)]. Like for all other raw materials, we can assume the limestone for the canal at the base of the dam, to have been quarried locally. Indeed the highly fractured nature of the local limestone bedrock, and the less detailed stone working of the canal blocks point to a local source, though no quarry traces or any zones of a more extensive haphazard quarrying were found. If not to be found elsewhere, the needed volume of limestone (ca. 250 m3) could originate from the bedrock outcrops now covered by the dam remains.
17With a total volume of some 70,000 m3, the labour requirements on site must have been considerable. Though higher manual labour productivity figures are estimated by C.J. Erasmus (1965), rather moderate estimates of 3 m3/d are applied here (Coche and Muiir, 1992; Daniels-Dywer, 2000). For the excavation of the silty dam core material alone, already some 10,000 labour days would have been needed. Likewise, the sourcing of boulder packing from its original young glacis context would require a similar labour force. Though a more elaborate calculation could provide figures for the added transport of the materials and the final construction and fine masonry, further data is however needed for such an approach. Although the brief calculations and elements brought forward somewhat over-simplify the construction process, and merely represent a first stage of the work, it can nevertheless be clear that for the full construction period a multiple number of labour days were required. Historically, the assigned work and travel of large groups of kurtaš (nonspecific ‘workmen’) is well documented from the centralised Achaemenid economic records of the Persepolis Fortification Texts archive (PFT), with numbers of individual work groups exceptionally large with as many as 1,500 individuals (Briant, 1996). Again according to the archive, such high numbers were not available for Pasargadae, although groups of workers amounting to 280 individuals, who received rations over several months or years, are reported (Arfaee, 2008). Although the PFT archive is surely fragmentary, still only partly published (Henkelman, 2008) and does not allow thorough comparison, such figures provide some perspective for the magnitude of infrastructural works such as at Sad-i Didegan. We can only speculate that the construction of the dam would have involved numerous, perhaps hundreds, of workmen employed over a prolonged period of months.
Fig. 6 – Topographic map of the Sad-i Didegan dam remains.
Fig. 6 – Carte topographique de l’état actuel du barrage de Sad-i Didegan.
1: riverbed and present-day reservoir; 2: limestone bedrock; 3: areas of recent bulldozer disturbance; 4: earthen dam core; 5: limits of the dam’s boulder packings; 6: exposed boulder packings; 7: area of architectural remains; 8: 1 m contour lines.
1 : fond du chenal et réservoir actuel ; 2 : lit rocheux calcaire ; 3 : zones de destruction récente par bulldozer ; 4 : cœur de terre du barrage ; 5 : limites des contreforts du barrage ; 6 : portion visible des contreforts écroulés ; 7 : emplacements des blocs ; 8 : isohypses équidistantes d’1 m.
Fig. 7 – Restitution of the canal remains at the base of the dam.
Fig. 7 – Reconstitution du canal d’évacuation à la base du barrage.
1: reconstructed water flow; 2: restituted blocks of the covering course; 3: reconstructed walls of the covering course; 4: in-situ preserved blocks of the bedding course; 5: bedding course structural elements; 6: restituted dam core; 7: cross-section.
1 : reconstitution du courant d’eau ; 2 : emplacement restitué des blocs de la deuxième assise de pierres ; 3 : limites restituées des murs de la deuxième assise de pierres ; 4 : emplacements originaires des blocs de la première assise de pierres ; 5 : éléments structurels de la première assise de pierres ; 6 : cœur du barrage restitué ; 7 : profil.
18In the lower body of the dam, two metres higher than the limestone bedrock (2,044 m), remains of an ashlar-built canal were encountered. These were found in two zones, respectively on the upstream and downstream side (fig. 6, zones A and B). In between these zones of exposure, further unearthed parts of the canal must in all likelihood still be preserved. Though within the framework of the project, excavations were not included, the canal can nonetheless thoroughly be described on the basis of surface documentation and the individual mapping of dislocated blocks. At surface level, the two exposed zones were cleaned, closely mapped and studied. To these zones, two main areas of displaced blocks are to be added. The first is a bulldozer-pushed concentration of ashlars and smaller cobble material, to the south of the dam (fig. 6, zone C) while the other is located somewhat downstream (fig. 6, zone D). Within this area, large ashlars of up to 1.7 m in length were found clustered at the narrow entrance to the limestone canyon. The blocks correspond to the lower and covering course of the distal zone of the canal (zone B) and for a large number of these the original location could be retraced on the basis of their dimensions and architectural details. Since these blocks do not always show clear bulldozer-cut traces, they were probably deposited here, at the time of dam failure, when a cataclysmic mud or debris flow scoured the dam and carried these blocks up to the bottleneck location of the canyon entrance. At the upstream side of the dam, the actual entrance of the canal is no longer present, though a 6.7 m long exposed stretch of the canal highlights the original dimensions of the main channel through the dam mass (fig. 4). Constructed out of four courses of limestone ashlars, the canal has inner dimensions of 0.9 m in width by 1.25 m in height, easily accommodating regular cleaning and maintenance. The second area of exposed canal remains is located 40 m to the southeast. Only the canal bed is preserved in situ, over an area 15 m in length by 5 m in width. Although many of the original blocks have been washed away (or were removed deliberately), this canal bed has several traces indicating the position of its original covering courses (fig. 7). Notably wider than the canal remains mapped in area A, this second zone must have originally accommodated a larger canal infrastructure. This is further indicated by the presence of four rectangular depressions, found in alignment on the canal bed. Further dislocated bedding course blocks with corresponding recesses were found in zone D, indicating there would originally have been six of such recesses. Similarly to the site of Sad-i Shahidabad, these six recesses represent the lower end of vertical sluice shafts, for which the covering blocks were also encountered further downstream from the dam (fig. 7). These blocks had been recorded by W. Kleiss, and were now fully mapped. Originally, they measured up to 2.1 m in length and housed smaller channels of ca. 0.32 m by 0.32 m in size. For each of these smaller channels, vertical sluice shafts allowed the individual control of the resulting discharge. With no other canal blocks identified, no further covering blocks are to be expected downstream of these sluices and the flow through the channels would thus have aired and quickly rejoined the original wadi bed. The resulting reconstruction is that of a single main structure at the base of the dam, consisting of a large feeder canal and an accessible control infrastructure at the downstream flank of the dam. In contrast to the first proposed reconstruction by W. Kleiss (1988), who took the wide bed of area B as the basis of at least two canals, only a single 77 m long main channel was present, with an overall downward slope of 1.6% (fig. 9). In the six smaller outlets of the sluice infrastructure, the flow could be finely tuned by closing or opening one or more of these shutters, the downstream discharge could be regulated so that a targeted discharge could be maintained over a considerable period of time. Overall, the evacuation canals and the outlets operate much like a hydraulic culvert or gate structure, in which the smallest passage (being 0.32 m x 0.32 m) determines the hydraulic properties of the overall structure. By allowing for the possible friction or any hydraulic jump throughout the system, the canal’s general discharge can thus be calculated (Chanson, 2004). At the maximum hydraulic head (19 m), the discharge from the canal would be quite considerable and as high as 1.29 m3/s for each of the channels, resulting in a total discharge of 7.76 m3/s. At a more limited reservoir volume of 5 m depth (resulting head of 3 m), total discharge would still amount to 2.96 m3/s. Even with such small dimensions, these channels would provide a very large outlet capacity for the dam.
19From an archaeological perspective, the canal infrastructure also provides several elements of chronological importance. Generally, the stone working is that of classical, Ionian-inspired Achaemenid style with the use of pick and chisel and the diagnostic use of anathyrosis. At several locations, iron clamps were used to bind the blocks of the upright walls and of the bedding course. Especially at the location of the sluices, an area of important hydraulic pressure, clamp cuttings were frequently found. The shape of these clamp cuttings is highly variable and to some extent reflects the somewhat poor quality of the limestone. Nonetheless, quite clearly two distinct and known types can be recognised. Aside from the first type of clamp cuttings of classic widely splayed dovetail shape, rectangular hook clamps were also encountered at the site. As C. Nylander (1965, 1966a, 1970) has argued on the basis of clamp studies at the major Achaemenid Palatial sites, the latter type is generally attributed to the reigns after Cyrus the Great (559-530 BC). Additionally, two singular occurrences of tooth chisel marks on block surfaces at Sad-i Didegan would furthermore point to a post-Cyrus construction date (Nylander, 1966b, 1970; Stronach, 1978). At the same time however, the first and oldest type of clamps (i.e., dovetail-shape) was introduced earlier, at the onset of the construction of Pasargadae, and only continued be in use during the reigns of Cyrus, Cambyses and Darius (who died in 486 BC). The presence of both tools might well suggest the construction period of the canal to fall within a period of transition between the two technical types. Finally, corroborating such a date in the late 6th or early 5th c. BC is the clear evidence of copying whereby the six-sluice design presents an evident copy of a similar canal infrastructure at the nearby site of Sad-i Shahidabad (note that the Sad-i Shahidabad remains are in all likelihood to be dated to the 6th c. BC as the site shows only classic dovetail clamps and no toothed chisel marks, and thus is highly reminiscent of the architectural details of the earlier monuments at Pasargadae). As it will be shown below, supplementary traces of contemporaneity between both sites are available.
20Finally, the completed maps and data provide the basis of a rough estimate of the actual reservoir capacity. Despite the fact that the 1:25,000 topographic maps only provide a contour-interval of 20 m and the resulting DEM is therefore of a rough nature, the general magnitude of the reservoir can nonetheless be estimated. As tab. 2 shows, a completely filled basin would hold up to 24 Mm3 of collected run off. However, we can imagine such a situation to have occurred only in rare circumstances and only after years of continued filling. In this respect, present day values from the downstream modern dam of Sivand are of interest. Inaugurated in 2007, the dam is located on the Rud-i Polvar within the Tang-i Bolaghi valley (Fars Regional Water Authority, s.d.). Draining a catchment of 5,490 km2, the yearly submitted discharge is 100 Mm3. With the Didegan catchment only 46 km2 in size, proportionally it represents less than 1% of the Sivand catchment. Similarly, a calculation on the basis of the average annual run off at the Dorudzan gauge station (Justin and Courtney, Irrigation Corporation of Iran, 1965) would provide a total collected volume of ca. 6 Mm3. If we were to adopt these present day values, it would take some years to fill the basin completely. It is therefore highly unlikely for the Achaemenid dam to have functioned with a vast reservoir for a prolonged period of time and we can imagine that, if water was indeed stored temporarily, it would only amount to a few metres of storage.
Tab. 2 – Calculated reservoir capacity of the Sad-i Didegan dam.
Tab. 2 – Capacité du réservoir du barrage de Sad-i Didegan en fonction de la hauteur d’eau.
Reservoir depth
|
Reservoir surface
|
Reservoir volume
|
20 m
|
364 ha
|
24,351,663 m3
|
15 m
|
234 ha
|
8,248,668 m3
|
10 m
|
73 ha
|
1,533,207 m3
|
5 m
|
4 ha
|
39,672 m3
|
21Ultimately, a more integrated reconstruction of the Sad-i Didegan dam can be proposed by combining both the elements of the geomorphological mapping and of these rough estimates of canal discharge and reservoir capacity. With its small incised valleys in the Late Pleistocene loess-material, the observed predominance of colluvial dynamics and the mere occasional occurrence of identifiable large flashes of wadi flow (fig. 5), we have good evidence that the dam had limited sedimentological impact. As the studied section at the valley bottom has shown, no clear hints of major alluvial phasing or dam-related sedimentation can be observed. Moreover, with the canal base only two metres above the bedrock base level below the dam, dead storage capacity is very limited. Any higher build-up of the sediment would have entailed sedimentation influx and the continued maintenance/clearing of the inner canal. Furthermore, this would have left notable banks of silty material (Calvet and Geyer, 1992). As an indication, the present day small lake as shown on fig. 6 presents the extent of this dead storage. In all likelihood, the dam was a genuine check dam and we must imagine it to have functioned seasonally and sporadically, with limited storage at times of winter and springtime wadi flow. In periods of agricultural water requirements, the system’s high volume sluice configuration would have allowed delayed and fine-tuned control of the discharge, into the perennial river of the Rud-i Polvar and on to the plain of Pasargadae. In such a hypothesis, the dam is clearly an oversized feature for which the resulting height and invested labour first of all reflects a considered design and effort to match the natural elevation of the dam’s solid base, with the old scree slope material and limestone bedrock on both the western and eastern side.
22With the Sad-i Didegan dam quite credibly only ever functioned below its full storage capacity, further evidence encountered in the region might hint at secondary use or even an intended different primary use of the dam location, and an explanation for the suggested limited use. Another trace of grand-scale hydrological investment, a large earthwork was found crossing the catchment divide between the Sad-i Didegan catchment and the adjacent valley to the west, home to the Sad-i Shahidabad dam (fig. 8). The earthwork presents a wide V-shaped trench of remarkable size, dug into the Pleistocene hillslope material. Especially in the central part of the depression, the flanking upcasts of fine material are considerable and mark the evident anthropogenic nature of the feature (fig. 8). This depression is up to 100-m wide and could be followed over a total length of 900 m. The maximum present-day depth is 7.5 m, and a rough minimal estimate of the total volume of the depression would provide a figure as large as 250,000 m3. The historic nature of the earthwork is evident with its trace clearly noticeable on historic satellite imagery (1970 CORONA KH-4B imagery). However, no further absolute dating material is present up to day. The sole relative dating evidence is composed of four cobble-built rectangular to slightly oval constructions of ca. 1.5 m to 1.8 m in length and up to 1 m in width. These were found clustered in the eastern bed of the depression and all but one is oriented in a north-northwest to south-southwest direction. Though only future sounding of these features can identify their function and possible date, they most likely represent graves of an unknown date or type. Though no extensive surveys or references are available for the Fars area (Boucharlat, 1989), and only limited attention has been devoted to present-day and recent practices of mobile pastoralist of the Qashqa’i confederation and other groups in the area, it can be proposed that these pre-date the Islamic era. First of all, their orientation is manifestly different from Islamic burials, which are oriented along a northwest to southeast axis with the head placed on the northwestern side (Schmidt, 1957). Since plain-earth burials in Fars in the pre-Islamic historic periods are hardly a widely documented and published subject (Boucharlat, 1991), only a provisional comparison can be made here with closest parallels to structures of predominantly Parthian, or possible Sasanian date. Graves of similar orientation were attested and dated at Tal-i Malyan (Balcer, 1978) as well as at the Spring Cemetery near Persepolis (Schmidt, 1957). However, the original surface structures identifying the grave, were not reported at these sites. Without further study or sounding, the date of the features discovered cannot be securely be ascertained, though in all likelihood they date back to the pre-Islamic period and - in the absence of any known Achaemenid commoner’s primary graves (for the most recent review; Basirov, 2010; Jacobs, 2010) - postdate the construction of the Sad-i Didegan dam.
23With no other associated remains surrounding this earthwork, the depression is to be interpreted as a massive ditch connecting the valley of Sad-i Didegan, to the east, with the larger, alluvial valley of SafaShahr to the west and the dam site of Sad-i Shahidabad. The latter represents yet another large Achaemenid dam. The dam, which raised water from the perennial river, the Rud-i Polvar, also shows a 600-m long depression, a presumed spillway channel, running along the left bank of the Sad-i Shahidabad valley and it connects to the western start of the large earthwork (fig. 8). With the evidence for its chronology, and taking into account the profoundly massive and daring character of this earthwork, we can only speculate that it was a planned hydrological feature, connecting both catchments and evacuating water from the large catchment of Sad-i Shahidabad into the smaller valley of Sad-i Didegan, for which we have already established its large reservoir capacity. However, as fig. 8 shows, the longitudinal profile of the depression is marked by a distinctly upward slope at the outset and then by a downward slope, resulting in a convex overall profile. Consequently, the inter-valley canal was never finished, and must have been abandoned at a rather early stage, for reasons of an economic nature, perhaps unsuccessful engineering, or, more likely, because of a change in regional hydrology (being the possible collapse of one of the two dam sites). In the absence of any alternative readings of the large depression, our working hypothesis is that it was a canal, particularly when compared with other examples of such interfluve (or inter-valley) canals of similar dimensions, known from 1st millennium BC northern Assyria (Ur, 2005). Even though the construction in this case clearly was left unfinished, the remains echo a major investment in available labour. It also demonstrates the contemporaneity of both dam sites and ultimately the more regional scale and elaborately planned character of the hydrological endeavours in the Pasargadae area.
Fig. 8 – Inter-valley earthwork in between Sad-i Shahidabad and Sad-i Didegan.
Fig. 8 – Canal (inachevé) entre les barrages de Sad-i Shahidabad et Sad-i Shahidabad.
A: Longitudinal section. B: Cross-section. C: Local map of the Sad-i Didegan and Sad-i Shahidabad remains. 1: roads; 2: perennial river; 3: dry riverbeds; 4: dam crest; 5: Sad-i Shahidabad spillway; 6: inter-valley earthwork; 7: cross-section of the inter-valley earthwork.
A : Profil longitudinal. B : Profil transversal (voir aussi fig. 2). C : Carte de la zone de Sad-i Didegan et Sad-i Shahidabad. 1 : route ; 2 : écoulements pérennes ; 3 : écoulements temporaires ; 4 : crête de barrage ; 5 : canal de trop-plein du barrage de Sad-i Shahidabad ; 6 : dépression allongée entre les bassins versants des barrages de Sad-i Didegan et Sad-i Shahidabad ; 7 : localisation du profil transversal de la dépression.
Fig. 9 – Cross-section of the Sad-i Didegan dam.
Fig. 9 – Profil transversal du barrage de Sad-i Didegan.
1: limestone bedrock; 2: Late Pleistocene terraces; 3: boulder packings; 4: evacuation canal.
1 : substratum calcaire ; 2 : terrasses du Pléistocène supérieur ; 3 : contreforts constitués de blocs ; 4 : canal d’évacuation.
24It is clear that the Sad-i Didegan dam was constructed in the Early Achaemenid period, and possibly at the outset, it was planned to function as a larger catchment control measure. With no clearly arable land nearby, the dam and water control system worked in accordance to the demands and control downstream and probably was planned to accommodate a higher volume than the valley itself would have yielded. The construction of the dam was (at least partly) envisioned within a larger and more elaborate consideration of regional hydrology. The dam represents an attempt to supply and reshape the landscape for the countryside supporting Pasargadae. The location of the site was carefully chosen for both hydrological and geomorphological reasons. First, the valley setting and bedrock provided both a solid ground and plentiful building material required to construct the dam and canal, directly on site. With its beneficial narrow canyon entrance, a high dam crown elevation was achievable and thus large parts of the relatively small catchment of Sad-i Didegan could be used as spacious reservoir, to be filled by an inter-valley canal, linking Sad-i Shahidabad and Sad-i Didegan. In addition, the archaeological evidence, hydrological calculations and geomorphological observations further support the argument. During its actual lifespan the Sad-i Didegan dam must have functioned within its own catchment, well beyond the canal’s technical capabilities. That the original plan was only partially completed is shown by the limited dam-related deposits on the valley floor and is furthermore confirmed by the absence of any major reincision that can be linked to a post-dam failure. Rather, the valley shows deposits and dynamics which can be attributed to the study zone’s natural (pre-dam) situation.
25Although we can clearly reconstruct the logic behind the function and construction of the dam, the last element of dam history, its lifespan and final failure, cannot be demonstrated as easily. We can hypothesise that the limited dead storage and large accessible stone canal would have been easily maintained and regularly cleaned and that, with such maintenance, the dam could have had a considerable lifespan. However, the limited sinter build-up on exposed canal blocks and the apparent absence of any major repairs to the architectural traces suggest it must have been at a period of neglect and possible abandonment that finally, the boulder cover partly gave way, ensuing erosion of the earth core (seepage and piping) and subsequently, the dam was breached. That this eventually happened in a dramatic fashion and resulted in an instantaneous dam failure is possibly shown by the apparent far-reaching deposition of several of the sluice blocks of the canal. Additionally, as the dam itself, or any of the other dam sites in the region, were not renovated or replaced by a more recent dam, either in the pre-Islamic or Islamic eras, must show the massive investment such infrastructure required. We can only await the results of continued palaeoclimate research within Fars and the Zagros as well as further excavations of the archaeological sites in the region which will provide a better understanding of the climate and hydrological regime at that time, and of the economic role played by these dams and the countryside of Pasargadae.
The Joint Iranian-French project and the operations within the Pasargadae area were funded by the Archaeological Comity of the French Ministry of Foreign and European Affairs, the Archéorient laboratory, CNRS-Lyon 2 University. Additional funding and travel grants were provided by the Research Foundation Flanders (FWO Vlaanderen) as well as by Ghent University’s BOF program. These operations would furthermore not have been possible without the support and kind authorisation of the Iranian Cultural Heritage Handicrafts and Tourism Organisation (ICHHTO), the Iranian Centre for Archaeological Research (ICAR) and the Parsa Pasargadae Research Foundation (PPRF) and their respective directors at that time: Dr. Mehdi Mousavi Kuhpar, Dr. Hasan Fazeli Nashli and Dr. Mohammad Hasan Talebian. We furthermore acknowledge the kind support of the Institut Français de Recherche en Iran (IFRI) who provided the total station for the general mapping of the site. Furthermore, the authors extent their gratitude to those colleagues who provided numerous comments and exchanges, and they especially would like to thank both Maurits Ertsen and Miss Bingjing Zhang for their input as to the hydrological properties of the canal infrastructure. We are also grateful for the text revisions and language remarks by both Eline Deweirdt and Bernadette McCall. Finally, we express our gratitude to the guest editors Matthieu Ghilardi and Yann Tristant, correctors and the four anonymous reviewers for their assistance and feedback. Any remaining mistakes however, remain our own.