12 



STATION BULLETIN 355 



of drops and yield does not give us a picture of the actual amount of 

 drop unless the magnitude or the actual change in weight of drops 

 with change in yield is also known. 



Fig. 3A shows the regression line for the relation between per 

 cent drop and yield of Northern Spy in the BFP block in 1935. The 

 regression line predicts the average decrease in per cent drops with 

 increased yield per tree. Thus, with a yield of 200 pounds per tree, 

 the drop averaged 8 per cent, while with a yield of 1000 pounds the 

 drop averaged 3 per cent; yet a tree yielding 1000 pounds of fruit 

 dropped twice as much fruit by weight as one yielding 200 pounds. 

 This is shown directly by calculating the regression ecjuation on the 

 basis of weight of dropped fruit (Fig. 3B). 



200 400 600 QOO 



YIELD (Pounds) 



fCOO 



200 



400 



^00 



80O 



1000 



YIELD (Pounds) 



Fig. 3. A. (Upper) Regression of per cent drops on yield; B. (Lower) 



Regression of weight of drops on yield. Northern Spy, 



1935, B. F. P. block, 60 trees. 



With Mcintosh in the BFP block, covering the period from 1932 

 to 1937, there was a correlation coefficient of - — .099 between per cent 

 drop and total yield, and .563 between weight of drops and yield. 

 From the regression equation, trees yielding 1000 pounds dropped 125 

 pounds of fruit and those yielding 3000 pounds dropped 44() pounds 

 of fruit during the 6-year period. Here there is no correlation be- 



