344 Method of estimating Bacterial Density 



no significant difference in the proportion of exceptions between the 

 several groups of observations. 



Similarly the distribution of the exceptions in time, in which we have 

 shown the different groups to agree, may be best shown by taking the 

 totals, irrespective of the number of plates in each set. If this is done 

 we have a 2 x 10 contingency table, of which the value of ^^ proves to 

 be 57-826. 



Since n' = 10, the chance of such a distribution occurring under 

 conditions of random occurrence in time is about 4 x 10~^. It is indeed 

 obvious from inspection of Fig. 2 that the exceptional values occur in 

 groups together, although perfectly normal values continue to occur 

 throughout the worst of these epidemics. During the first outbreak 

 seven exceptions occurred with 14 normal values among them; the second 

 epidemic period was more prolonged and included 27 exceptions and 

 46 normal values. In the second half year of the experiment only six 

 exceptions occurred, of these two occurred on the same day (355) during 

 the last fortnight, when duphcates were taken, and two others, 338 and 

 340, were but two days apart. 



Bearing these points in mind, we have no hesitation in concluding, 

 on purely statistical evidence, that the exceptionally variable sets of 

 platings were due to two causes: — {a) a predisposing cause which is at 

 work throughout the epidemic period, and (6) some additional circum- 

 stance, in the absence of which the coimts obtained will still be normal. 



8. Special Organisms which affect the Number of 

 Colonies developing 



In the daily counts above considered, a uniform technique was 

 followed throughout, and fresh batches of medium were made up at 

 frequent intervals. It is conceivable that occasional differences in plating 

 technique, in the medium, or in counting the plates may by chance have 

 occurred on certain days. It is however most unlikely that any such 

 differences can have extended over the long periods covered by the 

 epidemics of high variance, without the fact being noticed. In seeking 

 a predisposing cause of variance, covering these periods, therefore, one's 

 attention is naturally drawn to possible changes in the soil itself or in 

 its population. 



It is known that certain micro-organisms, when growing on the 

 medium, exert an inhibitory action on the development of colonies by 

 other forms. The appearance of such an organism in the soil population, 

 during certain periods, might therefore give rise to periods of higher 



