153 



table matter, Tims Ave see that all tlu^ ex])eriineiits ao:roo with each 

 other. 



Ill the case of peas the results aic entirely different, for we see, 

 as in Ville's former experiments, that hy the side of a plant weiohinc; 

 less than a gram there will be another i)lant weii>hin<>- 10 or I.") or 20 

 grams, and even more, without its being })ossible to attribute the 

 difference to any apparent influence coming from the outside. There 

 is a regime of absolute irregularity, and an examination of the roots 

 shows that the irreguhyity is proportional to the presence or absence 

 of tubercles on the roots, whence arises the connection above men- 

 tioned. 



It now only remains for us to distinguish between cause and effect. 

 Is this appearance of these nodosities in itself merely a consecpience of 

 the greater vigor of the plants, or ought we, on the contrary, to see 

 in these very tubercles the origin and cause of that greater vigor ? The 

 following exiDeriment will show us which of these two hypotheses is 

 correct : 



When to the same soil of sterile sand which served for the preced- 

 ing experiments only 5 grams of good arable soil dissolved in 25 

 cubic centimeters of water was added, the peas grew in a natural 

 manner and produced, on the average, from 15 to 20 grains of dried 

 crop. Each stalk contained, on an average, 150 milligrams of nitro- 

 gen, although there were scarcely 10 contained in the soil. In every 

 case there was a fixation of nitrogen in the gaseous state amounting to 

 nearly half a gram. 



Under the same conditions a seed of lupin produced. a crop of from 

 42 to 45 grams, containing more than 1 gram of nitrogen. 



French grass (sainfoin) produced the same results, and in all cases 

 we see that the roots of these different plants are abundantly pro- 

 vided with tubercles; but if the artificial soils and the solutions of 

 earth employed in these experiments have been sterilized by the action 

 of heat the plants remain invariably i)uny and produce less than 5 

 grams of dried material per stalk. In this case the tubercles are 

 always wanting. 



Under cover, in pure air to which a little carbonic-acid gas has 

 been added, the results are a little less favorable than in the open air, 

 but they still show an important fixation of nitrogen in the case of 

 Leguminosffi infected with bacteria. 



These principles, then, represent the determining cause of the 

 phenomenon, and the systematic addition to the soil of a])propriatc 

 germs will enable us hereafter to reproduce at will the experiment of 

 Ville, which was formerly attended with sudi uncertain results. 



In the Museum of Natural History, Breal has obtained results sim- 

 ilar to those of Ilellriegel and Wilfarth. In one of his experiments 

 a pea containing 9 milligrams of nitrogen, in a soil of poor gravel, 

 but into which bacteria had l)een sown, produced a plant weighing 

 103 grams in a green state, 82.8 grams when dried, and containing 

 358 milligrams of nitrogen — that is to say, 40 times as much as the 

 seed. The pea vine, which was 1.40 meters long, produced 14 ripe 

 pods; the gain in nitrogen thus realized corresi^onds to about 255 

 kilograms per hectare. 



In another experiment, a small plant of lucerne grass provided 

 with tubercles and weighing -10 grams, and likewise in a soil of 

 sterile sand, gave a crop weighing 332 grams when green, 85.5 when 



