QUANTITATIVE BACTERIAL ANALYSIS 243 



a given milk sample, and if successive decimal fractions of 1 c.c. 

 are taken for plating, the plates will contain as follows : 



Plate No. 1234 5 6 



1 c. c. 1/10 1/100 1/1,000 1/10,000 1/100,000 

 The number of bacteria thus placed in each plate will be 

 2,000,000 200,000 20,000 2,000 200 20 



But the colonies produced by these will be (approximately) 



Knowing that plates containing over 200 do not develop a colony 

 for each growable bacterium placed in it, but only for the strongest 

 ones, reject all but plate No. 5. This contains 200 colonies; they 

 developed from 1/10,000 c.c. of the sample; hence 1 c.c. of the 

 sample must contain 200 X 10,000 = 2,000,000 colonies. 



On the other hand, counts below 40 colonies per plate are also 

 unreliable, from the mechanical difficulties attendant on removing 

 so small a number accurately from a mixture. If a dilution is 

 made so low as to secure only 10 — 30 colonies per plate (as in 

 No. 6 plate above), the chances of any one plate showing a number 

 of colonies truly corresponding with the dilution made is rather 

 small. Plate No. 6 is so diluted that the proper proportion is 

 only 20 colonies per plate, but the chances are that 15 or 30 or 

 some other number rather than 20 (but not far from it) will actu- 

 ally grow. It will be seen that the error introduced by considering 

 such low dilution accurate would be in the case given reporting 

 of a count of 1,500,000, 2,500,000 or 3,000,000, whereas the true 

 count is 2,000,000. 



It is quite important to thoroughly understand this principle, 

 as great errors in reporting counts may obviously be made if they 

 are neglected. Thus, if seven different laboratories report on a 



