ro 
These figures have been used for the graphical representation (percentage curve) 
in Fig. 13, the fully drawn lines. In the 1907 sample, the sizes fall between 19 and 26 
cm., by far the greatest percentage being for the sizes 20, 21, and 22 cm. The average 
size for the whole sample is 21.6cm. The sample from 1908 shows greatest percentages 
at 22—24 cm., the average for the whole being 23.2 cm., while in that from 1909, the 
mass of individuals will be found between 24—26 cm. with an average for the whole 
of 25.2 cm. 
A comparison of these three samples shows a distinet increase in size during the 
period from 1907—1909. Theoretically, this may be explained in either of the two fol- 
lowing ways: the fish may have grown differently im the three years, or the three 
samples may have consisted of fish of different ages. We will therefore consider the 
composition of the samples with regard to age, as seen in the following table. 
2 years 3 years 4 years 5 years 6 years 7 years 
1907.. 38.4 61.5 ee: 0.2 
1908.. 34.1 8.3 57.6 ene sea Baa 
1909 < « 4.3 34.8 12.5 46.2 2.1 0.2 
From this table, and from the graph based on the figures (Fig. 14) it will be seen, 
that the 1907 sample consisted of two and three year old fish, the 1908 sample of two 
and four year olds, and the 1909 sample of three and five year old fish. The composi- 
tion with regard to age exhibits thus a so extraordinary variation as to warrant our con- 
sidering it as the only reason for the difference in average size exhibited by the samples. 
This is even more evident if we consider the composition of the samples, not from the 
point of view of the ages represented, but from that of the percentages in which the dif- 
ferent years (year classes) appear. Presented in tabular form, this is as follows: 
Year class 1907 1906 1905 1904 1903 1902 1901 
TONER RO ke 0.2 
08 ar I RE Ba ee | ees 
ID... 0 15 843 16. 42 Bil 0.2 
Year class 1904 among fat herring. 
The table above exhibits the peculiarity, that the year class 1904 has in all three 
samples furnished an exceedingly large number of individuals. It is thus evident, that 
this rich year class, which in the autumn of 1907 was 2?/, years old, in autumn 1908 
3?/,, and in autumn 1909 42/,, is the principal cause of the great difference in the com- 
position of the three samples. This is also clearly indicated by the dotted curves (Fig. 
13), showing the distribution, in point of size, of the individuals belonging to the year 
class 1904 in the samples for the three years. It is natural that these (dotted) curves 
should agree so well with the curves of percentage of all individuals in the samples (the 
full line curves) since the year class 1904 furnished so great a percentage of the samples 
in all three years. | 
Examination of these samples thus leads us to two conclusions: 
1) The difference in the composition of the samples in point of size is due to the 
fact that they were of different composition with regard Lo age. 
