Figure 2. --Typical scales of control (upper) and starved 

 (lower) sockeye salmon. Arrow indicates starvation 

 check. 



The results obtained by this method were 

 evaluated statistically. The tests employed 

 and their results are presented in the fol- 

 lowing section. 



RESULTS 



Before we could proceed with the analysis 

 of the marking success, it was necessary to 

 validate the use of the visual method for 

 designating scales as either marked or un- 

 marked. To do this, we hypothesized that the 

 number called marked is independent of 

 (1) reader and (2) of different readings by the 

 same reader. Of the 652 scales examined 

 from the 1960 experiment, reader A called 

 193 marked and 459 unmarked, whereas reader 

 B called 165 marked and 487 unmarked. The 

 total difference between readers was 4,29 

 percent. Comparison of these results by chi- 

 square revealed no significant difference at 

 the 5-percent level (x^z 3.00 with 1 d.f., 

 n. = 652). Comparison of two readings of a 

 cross section of the 1959 scales made over 

 1 year apart by reader A revealed that the 

 overall results were identical, 135 desig- 

 nated as marked and 77 unmarked. On the 

 basis of this evidence, the hypotheses were 

 not rejected, and with reasonable assurance 

 we accept the validity of the results obtained 

 by this method. 



Marking success as referred to hereafter 

 in this paper will be the results obtained by 

 the most experienced reader. 



Marking success 



The results of the 1959 tests (fig. 3) are 

 striking in two respects. First, excellent 

 marking success was obtained in the 2- and 

 4-week lots and little success achieved in the 

 6- and 8-week lots. Second, in no case did a 

 starvation mark appear on the scales in 

 significant numbers until the fish had been 

 returned to food for over 2 weeks. 



Strikingly similar results were obtained 

 in 1960 (fig. 4). Excellent marking success 

 was obtained in only certain test lots and 

 in no case did marks appear in significant 

 numbers until the starved fish had resumed 

 feeding for over 2 weeks. Marks were de- 

 tected in 67, 100, and 96 percent of the 4- 

 week test lot after 4, 6, and 8 weeks of 

 resumed feeding in that order. Seventy-one 

 percent of the 6-week test lot exhibited marks 

 after having been returned to food for 6 weeks. 



