Spinning or Sailing?: The Boat Models from Eridu

Article excerpt

A sceptical view of received wisdom is much to be encouraged, and in that sense we welcome Thomas Strasser's reinterpretation, in the December ANTIQUITY, of the so-called boat models from Eridu. There are, however, strong arguments for rejecting his spinning bowl hypothesis. The known spinning bowls from Egypt and Palestine are massive and heavy, in order to provide tension against which to pull and ply or twist the fibres being prepared. (Technically these bowls are not for spinning per se: Barber 1992: 72.) They must be sufficiently stable to stay in place and not overturn. Barber suggests that they are also used for 'wetting' and that these bowls were associated specifically with the production of linen (Barber 1992: 72). The eastern Mediterranean examples have heavy handle-like loops within the base which show thread wear on the undersides. The Eridu vessels are far too fragile for such usage and the Eridu 'thwart' is not sufficiently heavy to have survived the necessary tension: nor would the boat-shape itself have been stable (see illustration in Safar et al. 1981: 227). Strasser also suggests - on the model of Aztec spinning bowls - that the socket, preserved in one of the boats and previously assumed to have held a mast, in fact held a rotating spindle from which the thread was drawn. However, the surviving socket is off-centre, which would have produced an unreliable wobble, as would the shape of the vessel itself.

The Eridu 'boats' date from the early 5th millennium BC, a time when no spinning bowls are attested anywhere. They are found, moreover, within a culture in which yarn and thread are traditionally produced by the use of handheld spindles (widespread by the time of the Neolithic villages of the 7th millennium BC and more suitable for wool). Unlike Egypt, no spinning bowls are illustrated even in later periods, nor have they been found, whereas the use of hand-held spindles continues (for example, the 3rd-millennium BC spinning ladies of Mari: Parrot 1962: plate 11). The only Mesopotamian bowls with 'inner handles' are a small number, largely from the Early Dynastic period, with interior clay 'flaps', originating at the rim and usually isolating a central area in the bowl. Their purpose is not clear; the three- and four-flapped varieties have been interpreted as pot-stands, possibly for collecting seepage from the supported vessel (cf., inter alia, Abu Salabikh: Moon 1987: 43; the Nuzi 'goose neck potstand': Starr 1937: plate 95B; Tepe Gawra Level IV: Speiser 1935: plate 29b; and the massive single-strap example from Habuba Kabira Sud: Strommenger 1970: figure 24). These show no sign of having served as spinning bowls, nor would their design have been efficient for such a purpose (see now Allen in press). In the eastern Mediterranean the earliest spinning bowl would appear to be that from Myrtos, illustrated by Strasser (Early Bronze Age): such vessels are not attested in Egypt before the Middle Bronze Age (12th Dynasty, with a single model dating to the 11th Dynasty), and appear later still in Palestine. Strasser suggests that the boat interpretation is 'precocious' and 'incongruously early'; the same can be said of the spinning bowl hypothesis.

Nor has Strasser read carefully the Roaf & Galbraith article (1994) which is said to 'cast doubt on' the neutron activation analyses of 'Ubaid pottery from sites along the Arabian Gulf with which the Eridu boats have been coupled. Roaf & Galbraith suggest more sophisticated statistical techniques and point out several regrettable errors, the most serious of which involves the attribution of the same data to two sites. The original computer data are now lost, but the first print-out, at which time only the Arabian sherds had been analysed, shows these data to be correctly attributed to the Arabian site of Dosariyah, so it is the Eridu results that should be ignored (Oates et al. 1977:226-7 and figure 2; Roaf & Galbraith 1994: 773). This unfortunate error in no way invalidates the overall results which are further supported by Kamilli's thin-section data. …