1. Traces of blood serum albumin have been demonstrated in normal milk.

2. The content of blood serum albumin is increased in many of the milks which are poor in non-fatty solids and there is evidence of some correlation between low non-fatty solids and high blood serum albumin content of the milk.

3. The experimental data support a conclusion that deficiency of nonfatty solids in milk in some cases may be due to dilution of a normal secretion in the udder by a fluid of the nature of oedema fluid, i.e. to a pathological rather than a physiological condition.

4. Some evidence, which obviously needs confirmation, has been adduced indicating that the casein-nitrogen content of milk as determined by Moir's technique may be slightly too high. The accuracy of the globulin figures seems also to be open to doubt.

The nitrogen distributions in 62 samples of whole morning milk covering a range of solids-not-fat content from 6·5 to 9·0 per cent, are discussed.

Samples high in solids-not-fat showed greater constancy in nitrogen distribution over a consecutive range of samples than those low in solids-not-fat.

Rich samples were characterised by high total nitrogen, low non-protein nitrogen and high casein nitrogen. Poor samples were characterised by low total nitrogen, high and variable non-protein nitrogen, low casein and consequently high albumin and globulin nitrogen. In samples low in solids-not-fat the albumin and globulin distribution was variable.

The casein/albumin/globulin ratio was roughly 14:2·5:1 for samples high in solids-not-fat, but the casein dropped in its proportion to globulin as the solids-not-fat decreased and the albumin showed also a slight parallel decrease.

The samples low in solids-not-fat which were studied appeared to be abnormal secretions, since they showed increased albumin, globulin and chloride contents.

The thanks of the author are due to C. E. Lesser and A. WagstafE for the figures for fat and solids-not-fat.

1. Figures are presented which show that there is a very appreciable variation in the salt and moisture contents of small plug samples taken from different cheeses from the one vat.

2. The fact that a cheese plug has to be ground to a paste before good duplicate results for salt content can be obtained is a further illustration of the variation in salt content within a single cheese, and of the slowness of diffusion of salt in Cheddar cheese.

3. The extent of diffusion of salt into unsalted Cheddar cheese has been measured in two different experiments. The results show that diffusion is very much slower than in Limburger cheese, probably owing to differences in moisture content and texture

4. It is well known that salt has an inhibiting action on the growth of bacteria in cheese. With variation in salt content from place to place in a cheese it is therefore to be expected that local variation will be met in the bacterial content. This may explain in part at any rate, the difficulties encountered in obtaining concordant results for estimations of number of bacteria in cheese.