500 CONTKIBUTIONS TO OUR KNOWLEDGE OP SOIL-FERTILITY, ii. 



EXPKRIMENT ix. 



The results are much lower than those obtained in previous 

 experiments, and, as the method of isolation had presumably been 

 improved, it was thought that the shrinkage might have been 

 catised by the bacteria slowly dying out during the four months 

 that the soils had been stored in the laboratory. Storage, how- 

 ever, appears to affect them only to a slight degree; when new 

 soils wei'e tested, the i-esults did not show any pi'onounced loss. 



Experiment x. 



So far as the soils ai'e concerned, a certain amount of informa- 

 tion might be gained by collecting the various results, and making 

 an average. As the conditions were the same for each set, the 

 averaging will probably minimise tlie experimental error. The 

 averages of eleven counts were taken from Expts. i., ii., iii., iv., v., 



