PROGRESS IN BIOLOGICAL IN-QUIRIES, 1927 235 



hot water to a similar consistency. In addition to the liver and 

 heart mixtures with beans and clam meal a mixture of equal parts 

 of beef liver and beef heart was fed to both the brook and the 

 rainbow trout. The results of the experiments with the brook trout 

 were very consistent, but in the case of the rainbows the data are 

 confusing in some respects. However, it is believed that the results 

 obtained with both species justify the following conclusions: 



Possibly the most striking result of the experiments was the evi- 

 dent superiority of the mixture of beef heart and beef liver to either 

 heart or liver alone. With respect to growth, the superiority of the 

 mixture over beef liver is not pronounced, but in the tests with both 

 brook and rainbow trout there was a striking decrease in mortality 

 among the fish fed the liver and heart mixture. When fed alone, 

 beef liver brought about much better growth than beef heart, but the 

 mortality of the fish on the latter diet was a great deal lower than 

 among those fed liver. As regards mortality, beef heart was on a 

 par Avith the heart and liver mixture. 



Sheep liver gave by far the poorest results of any of the meat 

 foods used and resulted in relatively poor growth and high mor- 

 tality among both brooks and rainbows. The experiments of pre- 

 vious years also indicate the inferiority of sheep liver, and it is 

 believed that this food should not be fed to fingerling trout. 



The relative value of the substitute foods used is not as clear as 

 in the case of the strictly meat diets, but it is evident that of the three 

 clam meal is by far the most promising. With brook trout mix- 

 tures of clam meal and beef liver gave results comparable to those 

 obtained from feeding the beef liver and heart mixture. However, 

 this mixture did not make quite as good a showing with rainbow 

 trout. 



Turning to the vegetable products, it was found that mixture of 

 beef products with soy-bean meal and Mexican beans gave results 

 inferior to those obtained with clam-meal mixtures. With brook 

 trout the soy-bean meal mixtures were better than those containing 

 Mexican beans, while with the rainbow trout the reverse was true. 

 On the whole, the results of the feeding experiments at Holden indi- 

 cate that it is inadvisable to feed cereals or beans to fingerling trout. 



Feeding experiments with yearling trout were less satisfactory, 

 owing largely to inadequate facilities and the small number of brook 

 trout of this age available for experimental purposes. It was nec- 

 essary to conduct the experiments in raceways where the fish were 

 exposed to various enemies, which resulted in considerable loss. 

 These losses, combined with those resulting from cannibalism and 

 gill disease, make it difficult to interpret the results correctly. Sheep 

 liver was selected for the diet of the controls, as frequently (owing 

 to its lower price) it is fed to yearlings and brood fish. The other 

 experimental lots were fed sheep liver in combination with soy-bean 

 meal, Mexican beans, wheat middlings, shrimp bran, or " clam 

 heads." The wheat middlings, beans, and soy-bean meal were cooked 

 before being mixed with the liver. 



On the whole, a ration of 50 per cent sheep liver and 50 per cent 

 '' clam heads " gave the best results. The fish grew rapidly and 

 were exceptionally active and vigorous, with good color. " Clam 

 heads " is the dried refuse from clam canneries and is composed 



