VARIATIONS IN BAcTERIA COUNTS 223 
reasons why the plate method gives widely discrepant results. Among the 
most important causes are: (1) the failure of certain species to produce 
visible colonies on the medium and in the incubation temperature used; 
(2) the tendency of many species to exist In groups of two or more individ- 
uals, which groups are broken apart with varying degrees of completeness 
during the plating operation; (3) too few or too many colonies to the plate; 
(4) the inhibiting or beneficial effect of diffused by-products from the 
growth of certain species on other species within the radius of diffusion; 
(5) the personal element involved in carrying out the method. Widely 
varying, results from the same sample of milk under the same conditions 
of incubation temperature and medium would still be caused by the 
clumping tendency, by the number of colonies on each plate, and by the 
personal element entering into the manipulations. 
Hill and Ellms (1897) early called attention to the unreliable results 
obtained from over-crowded plates used in water analysis. The Standard 
Methods stipulate that there shall be not less than 30, and not more than 
200, colonies to the plate, altho Breed and Dotterrer (1916) conclude 
that limits of 30 and 400 are nearly as satisfactory. 
Altho the Standard Methods call for plain agar incubated at 37° C. 
for forty-eight hours, comparative counts published from time to time 
have shown that a carbohydrate medium and a longer incubation period 
at a lower temperature have many advantages. Heinemann and Glenn 
(1908), from their work on the effect of incubation temperatures and media, 
reached the following conclusions: 
1. Since pathogenic bacteria are always difficult, and in most cases 
impossible, to find in milk, a high temperature of incubation has no advan- 
tage over room temperature from this viewpoint. 
2. Incubation at 20° C. is superior to incubation at 37° C. because 
both a higher count and a better differential count are obtained. 
3. Dextrose is preferable to lactose as an addition to the medium. 
4. Miik is usually consumed before the results of bacterial examinations 
are available. Accordingly bacteriological «nd chemical examinations 
should have as their principal objects the improvement and control of 
the general supply; and accuracy being of greater importance than quick 
results, the loss of a day in its interest is irrelevant. 
