jSTew York Agricultural Experiment Station. 115 



a definite ratio whereby the results obtained by the one method 

 can be interpreted on the basis of the other method. This makes 

 it impossible to determine with a reasonable degree of certainty 

 just what dilution to use in making plates, if it is desired to sup- 

 plement the microscopic work with a cultural study. 



A much wider relative difference exists between the two counts where 

 the bacteria are few in number than where they are are numerous, 

 with a rapidly decreasing difference as the numbers of bacteria 

 increase. This is not only true in a comparative series of counts 

 but it holds true for many individual samples. When the plate 

 count averages less than 10,000 per cubic centimeter the total num- 

 ber of individual bacteria as seen by the microscope is about 44 

 times as great, but when the plate count approximates 1,000,000 

 per cubic centimeter the total number of individual bacteria is only 

 about 5 times as great. The difference, however, between the plate 

 count and the microscopic count where each isolated bacterium and 

 clump is regarded as an individual object is much less, being only 

 about 17 times as great as the plate count when this averages less 

 than 10,000 per cubic centimeter. The two counts are practically 

 the same when the plate count approximates the 1,000,000 mark. 

 This clearly indicates that the bacteria found in unpasteurized market 

 milk containing 1,000,000 or more bacteria per cubic centimeter 

 practically all grow on nutrient agar media if incubated at ordinary 

 temperatures. 



These facts demonstrate that the bacterial count obtained in milk 

 by the direct microscopic method is equally as good if not a better 

 criterion of its bacterial content than the count obtained by the 

 plate method. Whatever may have been the cause of such vari- 

 ations as those cited on page 100, it is unquestionably true that the 

 microscope reveals the actual germ content of such samples as the 

 one taken on February 19 (Table II) more accurately than does the 

 plate count. Since the number of bacteria present in a given sample 

 of milk is indicative of the care and of the contaminating influences 

 to which the milk has been subjected, the plate method failed in 

 this particular case to reveal what it is generally supposed to reveal. 

 In other words the plate count gives the idea that this particular 

 sample of milk was of a much better quality than those taken on 

 February 18, 20 and 24 from the same farm, while the microscope 

 shows that this was not true. 



