Bacteria Developing at Different Temperatures 247 



ratios for such samples as do contain bacteria of this tyj)c arc alxive 5. 

 The figures are shown in Table 13. 



Normal bacteria-B. coli ratios. — In determining the normal ratio 

 between the 20° bacteria and the B. coli, our available data cover 

 a wider variety of sources and a larger number of samples than is 

 the case of the ratios previously discussed. With raw sewages 

 57 per cent of the samples examined have ratios between i and 5, 

 and 26 per cent of the samples have ratios between 5 and 10, the 

 other ratios being distributed on both sides of these limits. The 

 normal ratio for raw sewages, then, is probably just below 5. 

 The treatment of sewage by septic action or by straining tends to 

 eliminate abnormal ratios, as is shown by the fact that 75 per cent 

 of the septic sewage samples and 83 per cent of the strained sewage 

 samples have ratios between the i and 5. A somewhat smaller 

 percentage of the ratios for the effluents from sewage filters is found 

 between i and 5, although the normal ratios for these classes of 

 samples still remain within those limits. A distinction between 

 the effluents of the sand and trickling filters, whose action is entirely 

 aerobic, and the contact filters, whose action is partly anaerobic, 

 is manifest by the considerable percentage of samples of the former 

 classes which have ratios less than i, as compared with the small 

 percentage of the latter class. The distribution of the ratios 

 over a broader scale is also notable, indicating the fluctuating 

 character of the types of bacteria contained in the effluents from all 

 types of sewage filters. The normal ratios for the Merrimack River 

 water and Applied 216 are also between i and 5, with a large major- 

 ity of the ratios outside these limits, falling below i. The effluent 

 from Filter No. 216 has a normal ratio between i and 5, with 20 

 per cent of the samples having a ratio of o, while the effluents from 

 the slow water filters are characterized by the fact that 50 per cent 

 of the samples have a ratio of o, the other ratios being fairly w.-ll 

 distributed, with a normal ratio somewhat above 5. The variable 

 character of the sea-water samples is evident from the distribution 

 of the ratios. With this class of samples the normal ratio again falls 

 between i and 5, 39 per cent of the samples having ratios Ix^wecn 

 those limits, with 21 per cent lying between 5 and 10, 14 jkt cent 

 falling below i, and 19 per cent with a ratio of o. The ditTcrcncc 



