386 Decomposition of Organic Matter in Soil 



of determining whether the numbers of bacteria on the several plots 

 at a given date corresponded with the order of productiveness. As a 

 rule no correspondence has been observed. Kellermann and Allen's 

 results^ are as follows : 



Verj" pro- Productive Poor 



ductive 



Plot No — 40 190 290 10 30 180 



Bacterial numbers, rail- 

 lions per gram ... 0-02 0-21 0-003 0-27 0-44 0-16 0-06 



Percy E. Brown^ also failed to obtain any close connection : 



Plot No 607 604 602 901 601 609 



Yield of maize, bushels per 



aero 52-7 50-7 46-0 43-2 33-5 32-5 



Bactei-ial numbers, millions 



per gram 2-8 3-3 2-6 2-5 2-1 2-7 



It could hardly be expected, however, that any very close con- 

 nection would be traced. The productiveness of a plot of land is 

 determined by a number of factors, of which bacterial activity is only 

 one, and, at any given moment, not necessarily a particularly im- 

 portant one. 



A better method for studying the connection between bacterial 

 numbers and the extent of the decomposition is to follow the changes 

 on a single plot systematically during a long period, and see how closely 

 the rate of decomposition is related to the changes in bacterial numbers. 



This we have done for three seasons, and we have brought the 

 results of our investigations together in the present paper. 



We have followed in detail the changes in amount of nitrate in the 

 soil, of CO2 in the soil air, and of bacterial numbers in the soil. The 

 determinations have been made at frc(|ucnt intervals, — usually some 

 10 to 14 days, and still more frequently at critical periods, — ^and the 

 results have then been plotted. 



The analytical results do not tell us precisely how much CO2 and 

 nitrate are formed in the field because CO2 diffuses out and nitrate 

 washes away to an unknown extent. Any attempt to enclose the soil 

 involves separating it from the main body and putting it under more 

 or less artificial conditions, which would defeat the whole object of the 

 work. For our purpose we mainly want the fluctuations rather than 

 the absolute quantities, and these are shown to a considerable extent 

 by the curves. When the curves are rising production is obviously 



1 U.S. Depl. of Agric. Burran nf Plant Induslry, Bull. No. 211, 1911. 

 3 Journ. Research Bull. 2, 1911; Gentr. Balct. Par. 1912, Abt. 11, 35, 234-272; and 

 Journ. Agric. Research, 19X6, 5, 855-869. 



