CONTROL OF THE PUBLIC MILK SUPPLY 441 



"1. The standard methods of Milk Analysis, published by the American 

 Public Health Association, are in emphatic need of revision. These standard 

 methods lay great emphasis on some of the least important points, while they 

 neglect to lay any emphasis on some of the most important ones. The revision 

 of these methods is now in the hands of at least three committees, one appointed 

 by the American Public Health Association, one by the Society of American Bac- 

 teriologists, and one by the Association of Dairy Instructors. (Revision of 1916 

 has since been adopted.) 



"2. Individual analyses under the best conditions are subject to considerable 

 variation, so that no single individual count can be properly relied upon. This 

 emphasizes the necessity of demanding an average of two or more plates in deter- 

 mining the bacterial content of any sample of milk. 



"3. The question of the exact composition of the media to be used is of far 

 less significance than that of the methods used in the manipulation. Wide varia- 

 tions in the composition of the media do not make any noticeable difference in 

 the bacterial count. 1 



"4. Greater care is needed to unify laboratory methods than has hitherto 

 been given. When the work of these four laboratories was compared at the out- 

 set it was found that there were very wide differences in the analyses of duplicate 

 samples, due chiefly to differences in laboratory technique. 



"5. These variations in the analyses of duplicate samples of the same lot of 

 milk have been found to be due to several causes: 



"(a) Laboratory errors. These occasionally appear, due doubtless to con- 

 fusion which is sure to arise when large numbers of samples of milk are analyzed 

 at the same time. 



" (6) Irregularities in methods of laboratory technique. These are several in 

 number, the more important seeming to be the following: 



"I. Shaking of the samples. Wide variations were found in the vigor and 

 the extent of the shaking to which the samples of milk and the dilutions are 

 subjected by the different laboratories. While this factor does not make a very 

 great difference in results, it is one of the irregularities that should be eliminated. 



"II. Amount of dilution. The counts from highly seeded plates are uni- 

 formly lower than the counts of the same milk from low-seeded plates. The 

 best results are obtained only when the plates contain somewhere between 40 

 and 200 colonies. Hence the number of dilutions made in any analysis will 

 materially affect the results. 



"III. Methods of counting. This has seemed to be the cause of the widest 

 amount of irregularity. The greatest difference was associated with the use or 

 non-use of a counting lens, or to differences in magnifying power of the lens used. 2 



"Even when this difference is eliminated the further results show that the 

 personal equation in counting is still a factor of very large importance. When 



1 For this reason and in order to produce uniformity, there has already been 

 adopted by the American Public Health Association a change in the standard methods 

 by which beef extract can be substituted for beef infusion and the acidity of the stand- 

 ard agar fixed at 1 per cent., thus making it uniform with that of the standard used 

 in water analyses. 



2 The recognition of this fact has already led to a modifying of the standard meth- 

 ods, which now require a lens of 3)^ diameters to be used in counting all plates. 



