SHA-FISHERIES LABORATORY. 143 
is the real significant difference between 1,000 and 1,010? 
This is the legal quibble of “the heap,” and it is to be 
avoided in two ways. (1) By regarding bacteriological analyses 
as results to be supplemented always by epidemiological 
results. This is the line taken by the Local Government Board. 
Whether or not mussels with rather more or less than 
1,000 B. cola per individual are to be condemned, depends 
on collateral evidence. 
(2) By using the results statistically. Quantitative 
biological results are useless for comparisons unless they are 
__ treated and interpreted statistically. The crude data of observation 
are less reliable than statistically corrected data. When we 
find from experiment that five mussels contain, on the average, 
1,000 B. cols each, we must assume that the figure 1,000 is 
subject to an error, the size of which depends on the methods, &c. 
It may be that any result from 900 to 1,100 is to be regarded 
as equally probable, given a certain standard of probability 
adopted at the outset. 
If, then, there is collateral evidence against the shell-fish, 
we may regard (say) a result of 900 as the same thing as 1,100, 
and so condemn the mussels. If, on the contrary, there is no 
prejudicial collateral evidence, we may regard a result of 
1,100 as not enough to condemn the mussels. 
A Legal Definition of B. colt. 
It is desirable, and will become more so as further Enquiries 
are held, that there should be a legal definition of B. colt. 
Bacteriologists, at least research bacteriologists, are already 
probably im agreement as to the characters diagnostic of 
this micro-organism, but a uniform treatment of material 
submitted for analysis is very desirable in public health 
routine work. The methods of analysis may, of course, vary 
according to the laboratory, but they should be so directed 
that they will give the mean numbers of B. coli contained 
in each of (say) 5 or 10 mussels, cockles or oysters ; and they 
