Summary One fact emerges from a consideration of the literature upon the methods at present in use for determining the bacteriological content of milk, viz. no one method is capable of giving exact information as to the numbers of living bacteria present. The plate count has been shown to be subject to a number of limitations to its accuracy, but it is possible that these are much more the result of failure to appreciate the importance and effect of observing the protocols of a standard technique, than of insurmountable causes of error inherent in the method. Any one laboratory should be able to return results which are comparable from sample to sample, but as has been shown, it is unlikely that results will agree between laboratory and laboratory unless the most rigid precautions to secure uniformity are enforced. This latter fact does not seem to be of very great importance, since the most that can be expected is that results shall reflect the condition of the milk at the time of examination. These will be governed by the previous history of the milk. Figures have been given which show (p. 120) that, if milk is properly produced and handled, the bacterial content will be, even 24 hours or more after milking, within the range where the plate count as a method of computation is more reliable than any other known method. From the point of view of keeping quality and value for some manufacturing purposes, milk is either good or bad, and the obvious inaccuracy of the plate or any other method is a matter of small importance when the bacterial content has once exceeded 500,000 per 1 c.c. Nevertheless there is no sort of justification for attempting to separate milk into classes, which are not broad enough to include the variations which have been repeatedly shown, although not always on sound lines, to occur in the figures. It may be possible so to separate milk of very low bacterial content, but it is not possible to do so when the bacterial content exceeds, say, 200,000 per 1 c.c. It must always be remembered that the colonies appearing on agar plates arise from groups as well as from individual bacteria. In view of the fact that keeping quality, the basis on which the consumer judges milk, depends upon the kinds as well as upon the numbers of bacteria, it appears wise to supplement bacterial examinations with determinations of keeping quality, and, in view of their special significance, of organisms of the coliform group. It seems that, although the methods of Breed and Frost are excellent for rapidly classifying milk into broad grades, their accuracy is necessarily diminished by their very nature, in that the results obtained are dependent upon computation to a greater extent than the plate method and, although it is possible to secure good results, the time consumed in getting these is as great, if not indeed greater, than by the plate method. The reductase test, although it has a large sphere of usefulness as a factory method, will not give accurate information as to the numbers of bacteria in milk.