444 TRANSACTIONS LIVERPOOL BIOLOGICAL SOCIETY. 



obtain an 100 per cent., a 96 per cent., a 90 per cent., 

 and a 77 per cent. " coli-like " microbe! But is there 

 really anything sound in the assumption? Clearly not, 

 for if we regard a bacillus as an organism defined by 

 certain characters, we expect that it will exhibit all 

 those characters. It is true that a species of animal or 

 plant may apparently lose a character (as in Mendelian 

 inheritance), also a character may be variable in 

 magnitude : thus the number of fin rays in the dorsal 

 fin of a plaice may vary, yet the fish is undoubtedly a 

 plaice. But the inability of a bacillus to ferment and 

 clot milk, for instance, is an absolute disappearance of a 

 character, and not a decrease in its magnitude. Besides, 

 the laws of variation and heredity in higher organisms 

 are sufficiently well known to allow us to appraise the 

 modifications of character, but this is certainly not the 

 case with bacteria. Variability in these organisms is 

 only just beginning to be studied, and the results so far 

 attained have not, in the very least, affected public 

 health practice. 



Let us remember what is the object of these analyses. 

 It is to detect, in shell-fish, &c, microbes (B. coli) which 

 have inhabited the human intestine and have then found 

 their way into a new habitat. We detect these microbes 

 by finding that they produce changes in various chemical 

 substances, and that is all we know about them that is 

 of value in recognising them. We find bacilli, which 

 produce all or some of the chemical changes induced by 

 the human B. coli, in many other situations. Now let 

 us assume that a colon bacillus may undergo " loss of 

 attribute" and still remain the same biological species. 

 Why should we say that it is only, say, 77 per cent, a 

 Bacillus coli? 



The "Quintuple Preferential" method, then, works 



