Since the processing of crops has a direct effect on the composition of archaeobotanical samples, variation due to crop processing must be filtered out before using samples for the study of crop economy and husbandry. As the archaeological context of plant material does not often provide a reliable basis for the identification of crop processing products and byproducts, this paper presents a method based on the statistical comparison of archaeological samples with ethnographically collected material and on an internal statistical analysis of the archaeological samples themselves. These analyses make use of differences in the categories of weed seeds extracted at each stage of the crop processing sequence and are thus more reliable and more widely applicable than simple calculations of the relative quantities of grain, chaff and weed. Ethnographic and archaeological material from Greece is used to illustrate the method and some of the wider implications of the results are discussed. -- AATA