By the term “udder count” in this series of papers is understood the colony count obtained when samples of early mid-milk (taken after about ½ pint has been removed from each quarter) are plated in milk agar and incubated for 2 days at 37°C. Dairy bacteriologists have long recognized the fact that cows with udder disease give high-count milk. In fact, some of those engaged in control work have made a practice of advising producers to search for mastitis when, in the absence of other causes, unusually high counts have been obtained in a series of samples. Therefore in low-count milk, at least, the milk agar plate gives some idea of infection in herds. The early observations, however, were not as a rule confirmed by satisfactory evidence of mastitis, i.e. by plating on blood agar or by cultural or serological typing of colonies picked off plates. This criticism also applies to many papers and statements in articles on mastitis as it affects other aspects of the dairy industry, e.g. off-taints, “slow starter” in cheese-making, and butter-making generally. Unless satisfactory evidence of mastitis is produced such work is valueless. Conclusions based on the results of the indirect tests are unreliable.