-81- 



weight. But sinoQ the average per stalk where two grew 

 in a hill was not far different from where three grew, and 

 since the nxunber where two grew were not a large per cent 

 of the whole, it seemed hardly worth the trouble to correct 

 for this error. Suppose ten per cent of the hills had two 

 stalks, weighing ten per cent more per stalk than where 

 three grew, the error would be but one per cent. 4his 

 would fall far below the range of experimental error. 

 Under same conditions doubtless correction should be made 

 where the number of hills with a smaller number of stalks 

 are any considerable per cent of the whole. The writer is 

 informed that at one station it is estimated that a hill 

 with two stalks is calculated to yield SOfo, and one with 

 one stalk 42 per cent as much as where three occur. 



