84 J. H. ORTON. 
decided to use some other fixed point of more constant position. The 
point chosen 1s in all probability the one shown in the figure of the 
Herring supplied by the Board. It is the point from which the perpendicu- 
lar 7 arises, and marks the origin of the median caudal rays from the 
muscular part of the tail. The muscular part of the tail is covered by an 
epidermis of metallic appearance, and is in nearly all cases sharply marked 
off from what may be regarded as the tail fin proper where this kind of 
epidermis is absent. The caudal fin-rays are slightly embedded in the 
fleshy part of the tail. Thus the point chosen for measurement may be 
stated shortly to be the origin of the mid-caudal rays from the fleshy part of 
the tail. 
The origin of these rays is, however, a concave line, as indeed is shown 
in the Board’s figure, and the point actually measured is the line at right 
angles to the long axis of the fish which forms a tangent to the posterior 
border of the fleshy part of the tail. This line is apparently the same as 
the perpendicular No. 7 shown in the Board’s figure. 
There were only a few fish in which this point was at all difficult to 
determine and these were among Sample IV. It is of course well known 
that the posterior scales extend over the mid-caudal fin-rays. 
ACCURACY OF MEASUREMENTS. 
Before the large samples were examined a batch of 33 fish was examined 
twice, in order to obtain some determination of the error in measuring 
under the conditions in which the samples would be examined. The fish 
measured were not in good condition, so that it is probable that the errors 
observed in this case would be the maximum error, especially as more 
practice in measuring was obtained afterwards. In this experiment the 
average difference in the two sets of readings was less than 1 mm. in all 
measurements except 4 and 6, in which the average difference was 1-3 
mm. and 1:1 mm. respectively. These 33 fish were examined in 47 and 
45 minutes respectively, i.e. about the rate of 44 per hour, about the 
maximum rate for the large samples. After this experiment assistance 
was obtained in making measurements for Characters 3, 4, 5, and 6, and 
there can be no doubt that the accuracy of the measurements was thereby 
increased. In all measurements therefore it may be confidently stated 
that they are correct on the average to one millimetre, and in the case of 
1 the average error is probably not more than -5 mm. It is believed that 
only isolated errors of measurement occur of as much as 3 mm., but 
errors would increase in frequency towards zero. 
It is, however, possible that occasional errors of observation may occur 
of as much as 5 mm. where the -5 cm. line on the scale has been read as a 
