freeze solid but would still preserve the optimum feeding consistency. 

 Because of variations both in diet composition and the refrigeration 

 temperature, a two-day storage period was not found satisfactory for 

 the maintenance of optimism feeding consistencies. 



Every effort was expended to present the food in a form which 

 would be available to the fish. Salt, when combined with certain 

 raw products, causes a change in the physical consistency of the 

 diet. The addition of salt at the rate of 2 grams per 100 grams of 

 mixed food containing an adequate binding agent causes the diet to 

 assume a rubber-like consistency which does not leach out readily 

 when introduced into the water. When binding agents such as hog liver 

 and hog spleen were included in the diets, salt was added to produce 

 this bound quality. When no binding action was present in the diet 

 components and the product to be tested leached badly in the water 

 when fed, absorbent meals , sufficient to prevent leaching, were added. 



At the conclusion of each series of feeding trials, the data were 

 analyzed statistically by use of analysis of variance for paired ex- 

 periments as outlined by Snedecor (19^0, 19U6) . In this manner it 

 was possible to measure the differences which existed between the 

 troughs on each diet and isolate the effect of biological variation 

 and experimental error. By the establishment of the fiducial limits 

 of these errors,, it was possible to determine the significance of the 

 differences which existed between the mean weights of the groups of 

 fish fed different diets. 



SUMMARIZED DATA 



The results of the feeding trials for the individual years are 

 summarized in Tables 2, 3, h, and 5. Two control diets were included 

 in each group of experiments: the first consisted of 100 per cent of 

 flukey beef liver, and the second of the standard Leavenworth pro- 

 duction diet which contained 20 per cent each of beef liver, hog liver 

 and hog spleen, 30 per cent Oolumblsu River salmon viscera, and 10 per 

 cent flame-dried, salmon offal meal. In each table the diet number 

 representing a control diet is underlined. 



The mean lot weight represents the mean weight of the two groups 

 of fish fed each diet. No compensation has been made for the mortality 

 which occurred during the experimental period. The mean lot weight 

 serves to weight the gain of the fish against the mortality and, there- 

 fore, is believed to be a better measure of the diet than is the av- 

 erage weight of the surviving fish. 



The per cent mortality represents the percentage loss for the 

 entire experimental period. Where the data are broken into 12 and 



