AbstractThe water separator, a mechanical flotation device which is finding increasing usage among Mediterranean archaeologists, is compared with “tub” flotation which is widely practiced by North American archaeologists. The water separator was found to be two to three times more efficient than tub flotation. A redesigned separator is presented which should further improve this figure.The problem of sample size and the attendant aspects of representativeness are evaluated by a comparison of items of various classes recovered in a 100% separator sample compared to the much smaller tub flotation sample.Finally, the specific biasing effects of various recovery techniques with particular emphasis on fish remains is discussed.