Figure 1: The distribution of confirmedStone Age sites (left) and confirmed pre-Islamic sites (right),showing the apparent move inwards of sites through time.
With generous aid from the AmericanInstitute for Yemeni Studies, the second season of the MiddleHadramawt Archaeological Survey (MHAS) was undertaken in October1999. This survey, the fieldwork component of my doctoraldissertation at the University of Pennsylvania, aims to expand ourunderstanding of settlement patterns in the Wadi Hadramawt, as theychanged through time. In reaching this goal, the 1999 season soughtto create a new, comprehensive, map of archaeological sites in theregion between Qatn and Tarim. Assisting and supervising me in thisendeavor was Hussein al-'Aydarus of the General Organization forAntiquities, Manuscripts, and Museums.
Though a number of previous expeditions havevisited the Wadi Hadramawt, survey coverage in the (sub)regionbetween Qatn and Tarim&emdash;herein referred to as the "MiddleHadramawt"&emdash;has lacked consistency. For example, the Frenchmission that excavated Shabwa conducted surveys in the westernreaches of the Hadramawt system and also in the Wadi 'Idm to theeast, but only, to my knowledge, visited a few sites near Shibam (inthe middle of the "Middle Hadramawt"). Likewise, though theSoviet-Yemeni expedition that excavated Raybun conducted fairlyintensive mapping and survey, their coverage east of Qatn is patchy.The survey component of the NYU expedition to Jujah was even morenarrowly-focused, seeking primarily to add detail to the French andRussian sketch maps. Moreover, primary attention was paid by thesesurveys to Pre-Islamic Hadrami and Himyari sites, at the expense ofearlier and later sites. Indeed, the only prior survey thatspecifically sought to find Stone Age sites in the Middle Hadramawtwas conducted in the early 1960s by the American Foundation for theStudy of Man. It is hoped, then, that the survey data from the MHAScan be used to augment these prior surveys to create a new map that,with reasonable consistency and the accuracy afforded by modernmapmaking technologies, represents all archaeological timeperiods.
The 1999 season of the Middle HadramawtArchaeological Survey, however, is conceived of simply as an initialstep. Upon creation of the new regional map, another field seasonwill be undertaken to explore a more narrowly-defined archaeologicalquestion. That season, currently planned for 2001, will seek toexplain patterns in the location of Middle Hadramawt sites.Specifically, the upcoming season will focus on sites from the FirstMillennium BC through the early Islamic period, in hopes ofcorrelating site size, kind, and distribution with the changing traderoutes and political climate. Though this analysis is still to beundertaken (relying, as it does, on detailed maps and collectionsfrom the upcoming season), some preliminary trends can already benoted.
Broadly speaking, the more deeply onetravels into the longer tributary wadis of the Wadi Hadramawt, thegreater the density of stone-age sites. Wadi Sarr, Wadi Bin 'Ali, andWadi Dhahab, in particular, each have dozens of Stone Age sites ofvarious types. Conversely, Pre-Islamic and early Islamic sites (tothe degree that the latter can be identified archaeologically) arefound in greatest concentration within the Wadi Hadramawt and at themouths of its major tributaries. The zone extending westward fromShibam, as shown in Figure 1, is especially densely packed withPre-Islamic sites.
Figure 2: A stone ring and clearedsurface suggest that this was the site of a Stone Age hut orencampment.
Typically, Stone Age sites in the MiddleHadramawt are stone alignments, cairns, and rings on the scree slopesof the wadis and their rock ledges (Figure 2). Very rarely do thesesites have any easily datable material, though occasionally neolithicprojectile points and palaeolithic handaxes have beenfound&emdash;the latter in great quantity at two sites deep withinWadi Bin 'Ali (Figure 3). Nevertheless, the heavy "desert varnish"patination on these sites' stones indicates their great age. It ispresumed that most of the stone rings are the remains of ancient hutsor tents, and that the cairns and alignments are graves and mortuarymarkers. Better preserved specimens of these constructions have beennoted so that future excavation can test these hypotheses and,hopefully, provide accurate dating.
Figure 3: One of dozens of handaxes foundon the surface of two sites within Wadi Bin'Ali.
Despite the paucity of stone alignments andcairns in the main wadi, their location on the walls of the wadimakes it unlikely that they were all obliterated by the wadi's laterinhabitants. It is much more likely, then, that these types of siteswere never present in the main wadi in the first place. This, inturn, suggests that their placement represents either the actions ofjol-dwelling populations which would descend into the wadis to erectthese monuments, or wadi bottom dwelling populations that wouldtravel into the outer wadis to erect these monuments. I suspect thatthe former hypothesis is true&emdash;but either case stronglysuggests that these cairns and alignments served ritual (and probablymortuary) functions.
If there were Stone Age populations livingon the floor of the main wadi, their traces are now surely coveredunder the accumulated alluvium from thousands of years of saylirrigation. This is a point that has been repeatedly invoked indiscussions of Yemeni archaeology. While certainly a factor affectingthe preservation and discovery of numerous sites, its effect shouldnot be overstated. Two types of silt were clearly visible in some ofthe larger wadis. The first type, well known, is light andfine-grained and deposited by sayl irrigation. The second type,however, is darker, coarser, less clearly stratified, and oftenoverlain by gravel beds. Occasionally, these gravel beds also havecairns and stone rings built upon them, proving that the alluviumbelow them (the latter type of silt herein described) cannot beanthropogenic. This observation supports Gardiner's discovery ofLevallois flakes in the silt in Wadi 'Amd in the 1930s and argues forcaution in using alluviation as a marker for humanactivity.
In contrast to the Stone Age sites in theMiddle Hadramawt, a whole class of which may be hidden beneath lateralluviation, the range and distribution of Pre-Islamic sites isprobably reasonably intact (even if, as is likely, some individualsites have been lost). The clustering of Pre-Islamic sites on the newmap, then, seems purposeful&emdash;and I expect that a great numberof the sites are located where they are, in part, because of theincense trade route. Sites found in the deepest reaches of thetributary wadis are almost exclusively graffiti stones. Sometimescovered with palimpsests of personal names written in the Musnadscript, these stones probably served as markers to travelers enteringthe wadis from the jol, and records of their passing. The clusteringof settlement sites around Shibam, however, indicates that this was ahub of activity. This strategic location at the confluence of WadiBin 'Ali and Wadi Hadramawt saw traffic coming northwards throughWadi Bin 'Ali as well as westwards from the major sites of Sunnah andMashgah in Wadi 'Idm&emdash;and these sites' inhabitants no doubtserviced and profited from the incense trade.
Indeed, the general pattern of Pre-Islamicsettlement in the wadi&emdash;the even spacing of the wadi's majorsites and the relatively blank spaces in between&emdash;suggests a"string of pearls" along the caravan route. A closer inspection willtest the contemporaneity of these sites, as well as the travel timesbetween them. It will also explore the role of lookouts and possiblecaravansaries such as al-Ghuraf at the mouth of the Wadi 'Idm. In theWadi Hadramawt, practically every outcrop has a fortress (see Figure4). But while the identified fortresses are almost exclusively fromthe Islamic period, closer inspection should reveal the location ofPre-Islamic fortresses which would be expected to have watched overportions of the trade route.
Figure 4: The mud brick core of aPre-Islamic or early Islamic fortress overlooking the confluence ofthe Wadi 'Idm and the Wadi Hadramawt.
Preliminary inspection of the 1999 MHAS dataposes more questions than answers. However, one question strikes meas particularly interesting: why does the area around Shibam have somany more Pre-Islamic sites than other parts of the wadi? Somepossible explanations are posed here, to be tested in the upcomingfield season. Next year's work, then, will focus most closely on thiszone, with careful mapping and collections from every knownPre-Islamic and early Islamic site there, in hopes of establishingthe chronology of those sites as well as any functionaldifferentiation between them. This analysis should suffice for mydoctoral dissertation, but other questions, such as the nature ofthis region's earliest settlement, beg inquiry and, with luck, willhave me returning to the Wadi Hadramawt for many years tocome.
