440 CITY MILK SUPPLY 



the plate method, are found. By the other, each clump and each chain 

 of streptococci is counted as a unit. The latter method usually gives 

 much lower counts and corresponds more closely with the plate count; 

 it is the method adopted by the Bureau of Chemistry in milk work when 

 the microscopical method is used. The Slack method shows approxi- 

 mately as many bacteria as the plate method does. 



The standard methods adopted by the American Public Health 

 Association in 1916 make the plate method the standard one but the 

 usefulness of the Breed method in certain lines of milk work is recognized 

 and its use advised when direct counting is undertaken. 



In the stained incubated-slide method recently proposed by Frost, J^o 

 c.c. of milk is mixed with standard nutrient agar and spread over 

 a definite area of a sterile glass slide. When the agar is hard, this little 

 plate culture is put in the incubator for about 6 hr. under conditions that 

 prevent evaporation. It is then dried, given a preliminary treatment to 

 prevent the agar from firmly binding the stain, stained, decolorized and 

 cleared. When the dried and stained plate culture is viewed under the 

 microscope the little colonies are definitely stained and appear highly 

 colored on a colorless or slightly colored background. The colonies 

 can be readily counted and the number of bacteria per cubic centimeter 

 calculated. 



Accuracy of Bacteria Counts. It has long been known that dissimilar 

 bacteriological results are frequently obtained when samples of the same 

 milk are submitted to different laboratories for examination. It was 

 felt that these discrepancies were in large part due to slight differences 

 in the media used and also to variations in technique that obtained in 

 different laboratories; so that American Public Health Association pub- 

 lished Standard Methods of Milk Analysis which have been generally 

 accepted throughout the United States. In spite of these methods being 

 used, divergent results continued to be obtained and this, in 1914, led 

 four laboratories in New York City, viz., the New York City Public 

 Health, the Borden, the North and the Lederle laboratories to ask H. W. 

 Conn of Wesleyan University of Middletown, Conn., to act as referee 

 of a cooperative test that lasted 7 months and involved the making of 

 about 20,000 bacterial analyses by various methods with the object of 

 determining whether the discrepancies are due to inevitable difficulties 

 in bacteriological analysis, or whether such analysis has such inherent 

 obstacles that it cannot be made reliable, or whether the different methods 

 of technique employed in the several laboratories were responsible for 

 wide variations in the result. In case differences in technique proved to 

 be a factor of considerable moment, it was the intention to determine 

 how nearly identical the various methods might be made and what error 

 in bacteriological milk analysis would still remain. The general sum- 

 mary and conclusions arrived at from the test are these: 



