140 EXPERIMENTS UPON LEGUMINOUS CROPS 



the original sample from arable land. For these reasons it is 

 necessary to reduce the estimate made of the annual gain of 

 nitrogen, but however large an allowance be made on this 

 score, it is still difficult to account for the magnitude of the 

 accumulation, especially for Geescroft, where, as the botanical 

 analysis showed, there were no leguminous plants. 



The other sources of nitrogen which may be invoked, such 

 as the rain, dust, and absorption from the atmosphere, would 

 equally affect the arable land, yet, as the various unmanured 

 plots on the wheat, barley, and rotation fields show, there is 

 no evidence of a corresponding gain of nitrogen on the arable 

 land. 



The only explanation that seems at all probable depends 

 on the intervention of the bacterium Azotobacter chroococcum 

 (Beyerinck), which possesses the power of fixing atmospheric 

 nitrogen without any host plant, and which has been found in 

 all the Rothamsted soils. On the arable soils this bacterium 

 would not be able to effect much fixation because of the lack 

 of recent organic matter, by the combustion of which the 

 necessary energy for bringing the nitrogen into combination 

 could be obtained. On the wild grass-land, however, there is 

 every year an accumulation of vegetable matter which would 

 supply the bacterium with its needed carbohydrate, and in 

 consequence considerable fixation is possible. The greater 

 gain on the Broadbalk land may be due to the presence of 

 leguminous plants in the herbage, or to the comparative rich- 

 ness of the soil in calcium carbonate, since the Azotobacter 

 has been found to be active only in soils containing calcium 

 carbonate. 



It would appear from the chemical analyses that the Gees- 

 croft field has never been subjected to the chalking operations 

 previously described (p. 28), since the surface soil in 1904 

 contained only 0*16 per cent, of calcium carbonate, about the 

 same quantity as was found in the soil of the adjoining un- 

 cultivated common land, whereas the soil of Broadbalk wilder- 

 ness contained as much as 3*3 per cent. 



