It has been argued (p. 5) from figure 1 that the group of small 

 fish taken In June, 1938, 'with a dominant mode near 70 ram. represented 

 fish spavmed in 1938. According to the scale studies, the scales of 

 these fish were consistently v:ithout annuli. It may therefore be sup- 

 posed that fish v^ith one and two rings on their scales at that time 

 of year were of year classes 1937 and 1936, respectively. In November, 

 one annulus became evident near the margin of some specimens identified 

 from figure 1 as belonging to year class 1938 • By March, all the mem- 

 bers of this class had one annulus, while those having tv;o and three 

 were now identified with year classes 1937 and 1936, respectively. 



If viTe choose an arbitrary boundary to separate "narrow" from "wide" 

 marginal increments, say the 32.5 mm. point in the data used for fig- 

 ure h, it is possible to define a time when the annulus became evident 

 during 1938-39. The propor.tion of fish having "narrow" marginal in- 

 crements, in percentage, by months, is given in figure $. In southern 

 California, according to figure 5, an annulus became evident, on the 

 average, in pilchards of year classes 1937 and 1938 early in January, 

 when 50 percent of the specimens examined showed a "narrow" marginal 

 increment, and 50 percent a "vride" one. In central California the an- 

 nulus did not appear until late February for year class 1938, early 

 March for year class 1937. 



This difference between central and southern California held con- 

 sistently during the months for which comparable data are available 

 (fig. 6). Since the scales vrere read without knowledge of the catch 

 localities , this is further evidence that the marks called annuli had 

 not been formed sporadically. 



The annulus, being a very narrow zone, practically a line, became 

 evident only when enough growth had occurred to expose it. Therefore 

 it must have formed some time previous to the late fall and winter, 

 probably in the late summer and fall, when, Judging from figure 5, 

 growth of the scales is slight. 



If the evidence be accepted that the marks counted as annuli are 

 really year marks, it remains to be shown whether the counts can be 

 made -virith reasonable consistency. To test this, a stratified random 

 sample of 2k2 scales that had already been read was drawn (by Mosher) 

 from among some 3,000 scales, so as to represent with approximate equality 

 the first four year classes. The labels of the slides were then masked, 

 and the scales read for the second time (by V/alford). The results, 

 given in Table I, indicate that one person can count what he defines 

 as year marks consistently. 



To test the similarity of readings by two persons working sepa- 

 rately, a sample of 367 scales from the fall fishery of 1939-UO that 

 had been read by VJalford was read independently by Mosher, with no 

 previous experience at scale reading, and recorded as described on 

 page 6. The results, given in table II, show a high percentage agree- 

 ment between the two sets of readings, and indicate that at least tv^o 



4l 



