ARCHEOLOGY IN SOUTH ARABIA — YAN BEEK, COLE, AND JAMME 539 



the present wadi floor. Yet, at one time, the wadi floor must have 

 been level with the present surface of this block of silt, and perhaps 

 both were even higher. It seems probable that this block of silt has 

 been preserved by chance, while the remainder of the surrounding 

 valley floor has eroded 2 metei^ to its present level. If this is the case — 

 and there appears to be supporting evidence) elsewhere — any sites that 

 may have stood on the silt have been washed away without leaving 

 a trace. At the same time, the height of the silt floor seems to have 

 increased at the eastern end of the wadi, near its confluence with Wadi 

 'Idim; when wells are dug in this area, pre-Islamic artifacts are 

 often fomid a meter or so below the present surface. In other words, 

 a gradual grading of the western two-thirds of the valley floor may 

 have destroyed all evidence of pre-Islamic towns in that area, while 

 deposition of this graded material together with silt from Wadi 'Idim 

 has buried pre-Islamic tow^ns below the surface in the eastern part of 

 the main wadi. Although this explanation may not be correct in all 

 details, it seems likely that some such phenomenon is responsible for 

 the disappearance of what must have been many townsites in the main 

 wadi. 



Some pre-Islamic towns in Wadies 'Idim and Du'an go back to at 

 least the 9th or 10th century B.C. and possibly a century or so earlier, 

 judging from similarities between the pottery from these sites and 

 that of Ha jar Bin Humeid in Wadi Beihan. Many also come down in 

 time to the early post-Christian centuries, and possibly to the Islamic 

 period. Altogether it is probable that the entire pre-Islamic period 

 is represented in the collections made at townsites durmg the 

 reconnaissance. 



One extremely important site of this period is a large field of ruins 

 of a pre-Islamic irrigation system in Wadi Du'an. Within an area 

 of about 2 square kilometers, there are several hundred — and perhaps 

 as many as a thousand — partially exposed mounds revealing sections 

 of canals and sluices, some of which are in an excellent state of pres- 

 ervation (pi. 7:2). All were constructed of stone and mortar. That 

 these ruins belong to the pre-Islamic period is certain from an exam- 

 ination of the installations themselves and associated artifacts. This 

 type of construction was not used in the Islamic period anywhere in 

 the Hadhramaut drainage ; as noted above, Islamic period installations 

 are entirely of earthen construction. Further, all the sherds found on 

 the silt surface surrounding these canals and sluices are weathered, 

 heavily patinated pre-Islamic sherds ; not one Islamic sherd was found 

 in this area. By contrast, the field plots of the Islamic irrigation sys- 

 tems have only Islamic sherds, except for those instances described 

 above where pre-Islamic sites were ploughed up and incorporated in 

 the field system. 



