ic Feb., 1910.] Tests 7^it// Cultures of Rofl't-t/ibcrclc Bachrla. loi 



the same number of plants. \o replanting was necessary to fill up gaps, 

 since Jx)th the peas and clover gave 100 ]:)er cent, germination. The 

 experiments were begun e.n 13th May, and the seedUngs were well above 

 ground by ist June. The plots were all surrounded bv wire netting, and 

 they were harvested about the middle of October, cut close to the ground, 

 immediatelv placed in sacks and weighed on a sjjecial balance capable 

 of weighing se\-eral hundredweight to a fraction of an ounce. 



Although in large experiments done in the field the unavoidable error 

 due to uncontrollable conditions mav amount to as murli as to or ::o 



ROOT SYSTEMS OF PEAS FROM WATER CULTURES. 

 1. Inoculated. -J. rniiiocuhueil. 3. Inoculated. No Nitrates. 



per cent., even in good experiments, and in bad ones may be much greater, 

 in small plots like the above where the conditions can be so thortaighh 

 controlled, the limit of error mav not be more than 2 to 5 per cent. 

 Bearing this fact in mind, the plots show conclusively that inoculation 

 did not benefit either the clover or peas in the slightest degree, but rather 

 the reverse. In all thr plots, tubercle-^ were present on the roots, and 

 they were as abundant in the uninorulated j)lot as in the inoculated o.ie. 

 The tubercles were possibly not (]uite so abundant in the plot to which 

 nitrates were added, but, nevertheless, the vield in this plot was greater 



