450 TRANSACTIONS LIVERPOOL BIOLOGICAL SOCIETY. 



and 448 as that it will be greater than 210 and less than 

 252 or greater than 448 and less than 730. It is 

 conventional to regard it as probable that it will lie 

 within the range, mean + the probable error. That is 

 to say, the mean of our estimation is any number greater 

 than 252 and less than 448, and in making a comparison 

 of this analysis with a " standard " we must bear this 

 in mind, for any number within the range is equally 

 probable. 



One might possibly use such an estimate for all analyses 

 made by the same methods. Obviously, in comparing 

 the results of two analyses, the probable error of the 

 difference of the averages must be calculated and taken 

 account of. I can find no indications of these statistical 

 precautions in the literature relating to shell-fish 

 bacteriology. One must insist that they are not fanciful 

 precautions, but the application of plain common sense. 



These considerations would apply if we were quite 

 certain that we could isolate an organism, call it Bacillus 

 coli, and be quite sure that it was an indication of the 

 transmission of organisms, only temporarily* altered in 

 character, from the human intestine to the body of a 

 shell-fish. But there is no convincing evidence that we 

 are able to do this. We do not know how many distinct 

 strains of B. coli inhabit the human intestine; nor 

 whether or not there are strains differing so little as to 

 be indistinguishable by the methods in use from those 

 of the human intestine, but with an entirely different 



* Consider the recent work on ' ' pure lines " in heredity. A Bacillus 

 we may assume is subject to variability. Let this variability be represented, 

 occasionally anyhow, by small mutations. Then even in the limited time 

 that a bacillus might take to reach a shell-fish from a water-closet, there 

 would probably be time enough for the formation of " pure lines," that is, 

 permanently altered strains of organisms. Let us suppose a bacillus will 

 divide every four hours, in three days there will be 18 generations and 

 2 18 individuals. The environment during these three days will have been 

 represented by many different series of conditions. 



