337 



At the time the experiments were planned the investigations of 

 Prazmowski had not been made public. The plants used were peas, 

 yellow lupine, beans {Phaseolm vulgaris), common locust [RoUnia 

 jyseudacacia), honey locust ( Ghditschia triac(()ithn.s),'Mu\ lahunium {Cyfisus 

 Laburnum), the object being (1) to test the tixation of nitrogen by 

 certain papilionaceous trees; (2) to study the eft'ects of using for the 

 inoculation of the plants of the different genera, (a) extracts of soils 

 in which each of the above-mentioned plants have been previously 

 grown, and (h) pure cultures of bacteria prepared from soil infusions 

 and from root tubercles; and (3) to determine whether the same form 

 of bacteria is capable of causing the growth of tubercles on leguminous 

 plants of diff'erent genera, or whether a special form exists for each 

 kind of plant, Avith which alone they are capable of living in symbiosis. 



The experiments were carried out in glass vessels of G.o liters capacity. 

 Each vessel had three openings in the sides, just above the bottom. 

 About a quart of pebbles Avas placed in the bottom of each jar, above 

 this a layer of sterilized cotton batting, and then the soil, which was 

 likewise covered with a hiyer of cotton. This artificial soil consisted of 

 pure sterilized quartz sand, with 5 per cent by weight of peat* (previ- 

 ously treated to extract nitrogen). To this was added half a per cent 

 of chemically pure calcium carbonate and a quantity of a nutritive 

 solution. The materials were all sterilized before filling in the vessels, 

 and each vessel and its contents were afterwards sterilized by heating to 

 95° C. on 3 separate days. Five plants were grown in each vessel. 



The soil infusions were prepared by thoroughly shaking GO grams of 

 soil with 300 grams of water and filtering, and were used without delay. 

 A bacteriological examination of the various infusions showed that 

 those of different origin differed not only with regard to the total num- 

 ber of bacteria they contained, but also with regard to the ])ro])ortioii 

 of colonies of the form designated by Beyerinckt as Bacillus radiri- 

 cola. The averages of numerous determinations of the number of bac- 

 teria per c. c. in the infusions of different soils were as follows : 



Total Bacilbts 

 miiiiber. radicicula. 



Pea soil infusion 1,980,000 78,000 



Lupine soil infusion 156, 000 None 



Robinia soil infusion 880, 000 78, 000 



Gleditschia soil infusion '. 310, 000 40, 000 



Laburnum soil infusion 1,300,000 143,000 



The lupine soil had been kept for several months before the infusions 

 were made, and had become very dry. No Bacillus radicicola was 

 found, and the bacteria present were all less active than those in the 



* The peat proved unsatisfactory, as it decomposed, and in all experiments except 

 the first series only half of the above amount was used, 

 t Bot. Ztg., 46 (1888). 



