It has been demonstrated by Hewitt and Burrows (19U8) that it is 

 practically impossible to secure an unbiased sample of a population 

 when but a portion of the population is selected by weight due to 

 the effect of stratification in the net during the necessary draining 

 period. To avoid the necessity for reductions in weight due to the 

 possible overcrowding of the fish, the troughs were stocked at re- 

 duced weights so that the initial population, less mortality, could 

 be retained throughout the experimental period. 



The equipment used in the experiments duplicated that in use in 

 production procedures as nearly as possible,, The troughs were the 

 standard, concrete troughs of the Grand Coulee Project, 16 feet long, 

 16-1/2 inches wide, and with an average water depth of 10 inches. 

 The food was presented to the fish by means of a specially designed 

 ricer of a more limited capacity than that described by Mitchell 

 (19U1). The size of the food particle fed was varied by differences 

 in the hole diameter of the perforations in the bottom plate of the 

 ricer. These diameters varied from l/l6 to 5/32 inch dependent on 

 the size of the fish. 



At the beginning of each series of trials, sample groups re- 

 presenting 10 per cent of the fish in the entire group of experiments 

 were counted to determine the average number of fish per trough. At 

 biweekly intervals throughout the experimental period, the fish in 

 each trough were weighed 1 on a solution balance sensitive to one gram. 

 A standard, l5-second draining interval was used before the fish 

 were introduced into the weighing container. The total weight of 

 each pair of troughs on a single diet and the total number of sur- 

 viving fish at the end of the period were used to determine the size 

 of the fish and the total weight of the lot. With these data and 

 the expected average water temperature for the next bi-weekly period, 

 the amount of food to be fed per day was established by reference to 

 feeding charts prepared for blueback salmon (Table l). The ration 

 for each diet was weighed but distribution to the individual troughs 

 was made by eye. To avoid the possibility of an accumulative error 

 in the food distribution between troughs, the order of feeding was 

 alternated so that first one then the other was fed first. On short 

 term experiments (12-lU weeks) the fish were fed twice a day, 6 days 

 per week. On long term experiments (2\\ weeks) the fish were fed 3 

 times per day, 7 days per week for the first 12 weeks and 2 times per 

 day, 6 days per week thereafter. 



During 19hh and 19hS, the diets were ground one day in quantity 

 sufficient for two rations, mixed by hand, and fed the following two 

 days. In 19^7 and 19U8, the diets were ground, mechanically mixed, 

 and fed within an eight-hour period. In each group of experiments, 

 the raw products were held in frozen storage until required, ground 

 while still frozen, and held under refrigeration until fed. The pre- 

 pared diets were held at temperatures which would not allow them to 



