SEA-FISHERIES LABORATORY. 137 
The Experimental Error. 
The mussels are unequal in volume, so that 5 taken now 
- or here do not necessarily have the same quantity of liquid 
in the shell, or the same mass of soft parts as 5 taken afterwards 
or elsewhere. The suspension of ground-up tissue in water is 
not uniform, for the hard muscle fibres do not disintegrate 
easily, and these fall to the bottom, while the lighter and 
softer tissues make a more stable suspension. There are errors 
of calibration im the pipettes and measuring flasks. Much 
more important than the above sources of error is that due 
to the variable composition and reaction of the agar medium. 
The materials, as bought, vary from time to time, and the 
preparation is usually the work of laboratory assistants who 
cannot always be overseen. 
If the number of micro-organisms inoculated is large— 
some grow so slowly as not to be recognised, the excessive 
number of organisms may diminish the nutritive capacity of 
the medium. Thus, a series of inoculations from “ decimal 
dilutions’ do not necessarily give multiples of 10 organisms, 
for the lower dilutions usually give fewer colonies than they 
ought to give. 
On the whole we may neglect these experimental errors 
as likely to be insignificant in comparison with those attribut- 
able to random sampling. 
The Original Error of Random Sampling. 
The mussels sampled are variable with respect to the 
quantity of micro-organisms taken up and retained, just as 
they are variable with respect to any morphological or physio- 
logical character that can be measured. Therefore, any sample 
of m mussels will differ from any other sample of » mussels 
collected from the same small area. Even if that sub-area 
is quite small it will still contain a very great number of mussels. 
