Pottery and P-Values: 'Seafaring Merchants of Ur?' Re-Examined
Roaf, Michael, Galbraith, Jane, Antiquity
A 1977 ANTIQUITY paper included a study of Neutron Activation Analyses of Ubaid pottery from Mesopotamia and the Arabian Gulf (Oates et al. 1977). A re-examination finds that the interpretation of the statistical analysis was mistaken and shows that the data do not prove the existence of 'seafaring merchants of Ur' in the Ubaid period.
In 1977 an important article entitled 'Seafaring Merchants of Ur?' [hereafter SMU] was published in ANTIQUITY (Oates et al. 1977), its title recalling Oppenheim's 1954 seminal discussion of trade in the Gulf in the late 3rd and early 2nd millennia BC. It drew attention to an astonishing archaeological discovery of the previous decade: pottery, which looked exactly like 5th-millennium BC Ubaid pottery from sites in southern Mesopotamia, had been found in Saudi Arabia, Bahrain, and Qatar more than 450 km from the southernmost Ubaid site in Mesopotamia. Since the publication of SMU, similar pottery has also been reported from Kuwait, the Bushire peninsula in Iran, and most recently the United Arab Emirates. Although others interpreted these finds as showing the Ubaid culture originated in the Gulf or as proving a widespread trade in ceramics, the authors of SMU more modestly suggested the pottery arrived in the Gulf from southern Mesopotamia in the equipment of Ubaid traders or fishermen. This interpretation was partly based on Neutron Activation Analysis (NAA) and petrographic examination (thin section analysis occasionally backed up by electron microprobe study of individual mineral grains) of pottery sherds from ten sites: four in southern Mesopotamia (Ur, Tell al Ubaid, Eridu and Uruk (Warka)), three in Saudi Arabia (Ain Qannas, Dosariyah and Abu Khamis), one in Bahrain (Al Markh), and two in Qatar (al Da'asa and Ras Abaruk). 115 sherds were subjected to Neutron Activation Analysis and 74 sherds (none from Uruk or Ras Abaruk) and 3 soil samples (from Ur and Qatar) were petrographically examined. Some sherds were subjected to both types of analysis. Those in the Neutron Activation Analysis were classified (Oates et al. 1977: 226-7) as 'painted Ubaid', 'plain Ubaid', 'plain red' or 'coarse red'. The sherds in these 18 site/type groups are listed in TABLE 1.
The analyses led the authors of SMU to four principal conclusions:
Proposition 1 Many of the Ubaid sherds from Arabian sites could have been manufactured in southern Mesopotamia.
Proposition 2 The coarse red sherds found on sites in Saudi Arabia Were different from the Ubaid sherds and may have been made locally.
Proposition 3 The compositions of the sherds from Ur were recognizably distinct from those of the sherds from the other three southern Mesopotamian sites. (Oates et al. 1977:225-7 asserted that the compositions of the 21 sherds from Ur were distinctly different from the compositions of all but one of the sherds from the other southern Mesopotamian sites: they concluded that the remaining sherd although found at Al Ubaid was made at Ur.)
Proposition 4 The compositions of 22 of the sherds from the Arabian sites were so similar to those of sherds from Ur that they 'must certainly . . . have originated in pottery workshops at Ur' (Oates et al. 1977: 228).
Proposition 1 is not controversial: the Arabian Ubaid pottery has its closest stylistic parallels with pottery of the later Ubaid period which had its origins in the earlier Ubaid period in Mesopotamia (Oates 1986: 84-5). But SMU also suggested that some of the Arabian Ubaid sherds (in particular the two sherds from Ain Qannas, Oates et al. 1977: 232) were not similar in composition to those from southern Mesopotamia, and Courtois & Velde (1984: 90), using a different analytical technique, concluded that two painted Ubaid sherds from Khor in Qatar did not originate at any well-known Ubaid site in southern Mesopotamia. Similarly Proposition 2 seems plausible: the coarse red ware found on Arabian sites has not been found on Ubaid sites in Iraq, while kiln wasters of this ware have been found at Dosariyah (Oates et al. …