It is well known that the fat content of cow's milk increases as the cow is being milked. CrowtherO) showed that this effect is reduced as the period between what is now known to be let-down, and actual milking is increased. Johansson (2) has described the effect in detail and has measured the influence of a number of factors on this change in composition. He has shown that the rise in fat percentage is greatest in cows giving a large yield of milk, or milked after comparatively long intervals. Incomplete milking, or 'super milking' with the aid of Pitocin, were shown to have little effect on the average fat percentage measured over a period of several days. These experiments suggest that abnormally high or abnormally low intramammary pressures do not preferentially decrease fat secretion, any pressure effect apparently acting on both fat and serum in the same way. The writer has found (3) that the distribution of fat globule sizes does not usually change greatly throughout the milking of the cow.In contrast with the cow, the sow shows no trend in the fat content of her milk throughout milking, nor is there any definite change in the fat globule distribution (4). Microscopic examination shows that the globules in sow's milk are somewhat smaller than those of cow's milk, and do not appear to cluster. Sow's colostrum is different in this respect, in that the globules cluster and, unlike sow's normal milk, the colostrum creams readily.Because of the differences between sow's and cow's milk, the writer has undertaken an examination of current theories explaining the trends in composition of cow's milk, and has modified one of the theories in an attempt to explain all of the facts. The following section is a brief review of theories at present in circulation.The creaming theory. This theory (5,6) suggests that the richness of cow's strippings is due to the cream being drawn last. If creaming in the udder could occur, it would certainly explain the effects observed, just as the non-creaming properties of sow's milk would explain the lack of a trend in the fat content of the latter. Two basic facts, however, make the theory untenable:(1) Most of the cow's milk is held in the alveoli prior to let-down. The diameter of the filled alveoli is of the order of 40-80 units (7, 8) so that any creaming which occurs, does so within a small volume. When the alveoli contract at let-down, the contents of the many alveoli will be mixed so that creaming effects will be largely eliminated, except perhaps for the small number of alveoli which empty via a ductule leaving the bottom of the lumen.(2) By milking a goat on its back after massaging tbe udder, Games (9) showed that the fat trend is still maintained, while Crowther(i) showed that the tendency for the fat