SEA-FISHERIES LABORATORY. 915 
distributions, the heterogeneity remains, and makes a 
bad-fitting theoretical curve.* 
There is no doubt that a catch of plaice, even one 
made by an experienced fisherman in such a way as to 
ensure regular action of the trawl net, contain hetero- 
geneous material; for we have only to determine the 
ages of the individual fishes to see that the catch 
contains fish of at least two age-groups. In the majority 
of the’ catches tabulated in this Report, age-groups I, 
II and III are present, that is, groups differing by one 
complete year of age. Now the mean difference in 
length of plaice of successive years of age, within the 
range represented by these records, may be taken as 
about five centimetres; and this difference is sufficient, 
when two groups are present in approximately the same 
’ proportions, to prevent the close fit of a frequency curve 
~ to the rough data. For the meaning of a theoretical 
frequency curve is that, in a series of measurements of 
one variable character in a great number of individuals 
of the same species, the deviations on either side of the 
mean follow one of a few probability laws, each of 
which, as experience has shown, can be represented by 
an algebraic or transcendental formula. But if the 
series of measurements include those of the same 
character in two species mixed together in the material 
examined, the distribution obviously cannot be expressed 
by the same probability formula. (By “‘species’’ is 
meant a distinct kind of thing differing in some respect 
from another.) 
If the number of measurements is relatively small 
(say 500), and if the sample contains fish which belong 
mostly to one age-group, a fairly close fit might be 
obtained. But when there are approximately equal 
*See Palin Elderton, Op. cit., p. 143. 
