VARIATIONS IN BACTERIA COUNTS 243 
The results given in tables 3 and 4 show some striking facts concerning 
the effect upon the counts of these two different methods of exposure 
of the plates to the uniform conditions of temperature and ventilation 
maintained in the incubator. In table 4 there is a marked difference 
in the results from all top samples and samples in the corner piles, as 
compared with those from the bottom of the piles and particularly from 
those at the bottom of the piles on the inside of the block. In order to 
show these variations in counts, the standard deviation, the coefficient 
of variability, and the probable error were calculated for the entire number 
of samples in each set and for certain groups of samples which were packed 
in the solid block. These mathematical expressions are shown in table 5. 
TABLE 5. VariaATIONS IN Counts Dur To Meruops or PILING PLATES AS SHOWN BY 
COEFFICIENT OF VARIABILITY, STANDARD DEVIATION, AND PROBABLE ERROR 
Nae Coefficient 
Samples used - WOnigE Average Standard Probable of 
for calculations 2 1 count deviation error variability 
Semp es (per cent) 
Air space between all plates 
(iD 8). 8eseos ae 100 | 237,350*| +49,014*} 33,059* 20.6 
Plates in solid block, all samples 
(iio 4) .ses Cope er 90 | 307,000 | +105,000 70,822 34.2 
Samples in corner piles only 
ONostrsG IS 18)i seis ect 20 | 368,000 +45, 485 30,679 12.3 
Samples in side piles only (Nos. 
Peon ondn ta, 14, 15,16, 17) 50 | 330,000 +64, 000 43,168 19.4 
Samples in center piles only (Nos. 
Sy Dy TU) ID) eS eran 20 | 195,000 | +138,016 91,742 69.7 
Samples from top plates of each 
lle. oo BN ho Rane 18 | 365,000 +39, 200 26,440 10.7 
Samples from bottom plates of 
Barc hepllerraentartsc a ors oes 18 | 201,000 | +123,200 83 , 098 61.3 
*Not comparable with corresponding figures from samples shown in the remainder of the table 
because of difference in bacterial content of the milk. 
The significant figures in table 5 are those representing the coefficient of 
variability, altho the other figures contribute toward a more comprehensive 
understanding of the variations found in the different groups of samples. 
The coefficient of variability obtained from samples when a free circulation 
of air was allowed between the plates was 20.6 per cent; whereas this 
