The use of mortality models in archaeozoological recognition of stock husbandry goals, particularly milking, has been criticised on three grounds: (1) primitive livestock do not "let down" milk in the absence of their offspring, so the infant slaughter characteristic of the "milk model" actually suppresses lactation; (2) pre-modern herders do not optimise for a single product, but also raise for meat the offspring from milk-animals; (3) mortality approximating to the "milk model" might reflect biased sampling of infant disposal areas or heavy natural infant mortality. These problems of uniformitarianism, optimality and equifinality are reviewed in turn. (1) uniformitarianism: it is argued that cattle, sheep and goats are "pre-adapted" to milking, and that problems of let down are related to husbandry conditions (nutrition, housing) and can be overcome in various ways. (2) optimality: the culling decisions of recent Greek herders, specialising in dairy production for market, conform to optimising expectations. (3) equifinality: assemblage formation can be explored through comparative analysis at intra- and inter-site level. It is concluded that, although mortality models measure potential rather than actual exploitation and although milking is consistent with a range of mortality patterns, only mortality approximating to the "milk model" can be cited as evidence in favour of intensive dairying.
Archaeozoology, dairying, milking, mortality models.