o50 Method of estimating Bacterial Densitu 



variation cannot be detected with certainty in a small set of plates, we 

 recommend that occasional sets of 10 or 20 plates should be prepared 

 from time to time, and that if necessary every batch of medium prepared 

 should be tested in this way, the colonies being counted after seven days. 



10. The x^ Index of Variability applied to other 

 Bacterial Count Data 



It has been shown by the use of the ■)^ index of variability, that the 

 great bulk of Cutler's data on soil bacteria appears to be true samples 

 from the Poisson Series, and that therefore the accuracy of these results 

 is known with precision; also that, by the same method, a small 

 proportion of exceptions may be detected in which some definite dis- 

 turbing cause has interfered with the accuracy of the results. It is 

 therefore desirable to apply the same test to other sufficiently extensive 

 bodies of material, in order to ascertain if, by other methods, a similar 

 degree of accuracy can be obtained, and faihng that, if further light can 

 be thrown on the problems of the dilution method. Data from four 

 sources have been examined in this way. 



(A) Buddin's counts of soil bacteria at Rothamsted, using a gelatine 

 medium. 



(B) Counts of soil bacteria pubhshed by Engberding (1909(12)). 



(C) Breed and Stocking's tests of the accuracy of counting B. coli 

 in milk (1920(13)). 



(D) W. Owen's bacterial counts in sugar refinery products (1914 (14)). 

 In the aggregate we have tested over 1000 sets of parallel plates; 



owing to the bulk of the total examined it is possible that a small 

 proportion of arithmetical errors has been included, although the 

 appHcation of the method is much more expeditious than that of the 

 preliminary investigation of Cutler's data. Only the obvious and 

 unquestionable features of each body of data will be dealt with. 



(A) Buddins data 



A very large number of bacterial counts were made at Bothamsted 

 by W. Buddin, to whom we are indebted for permission to make use of 

 these data. The actual plate counts, though not published, formed the 

 basis of bacterial number estimations used in Buddin's work on the 

 effect of antiseptics on soil(i5). 



The platings in this work were made on a nutrient gelatine having 

 the following composition: — ^Wittes peptone 40 grams, Lemco 20 

 grams, NaCl 20 grams, gelatine 480 grams, distilled water 4000 c.c. 



