172 



CONTRIBUTIONS 10 OUR KNOWLKDGE OF SOIL-FERTILITY, xvi., 



EXPEBIMENT XX. 



The numbers in the three tests are not uniform, but they serve 

 to show the probable variation that the effect of dilution has 

 upon an acid or alkaline extract. For example, in the ten days' 

 test with Penic. cladosj). (xix.), the numbers with 80% w^ere 187, 

 and with '20%, 38. The reaction-curve passes through 86 for 

 + 0-1,91 for +0-08, and 99 for +0-02. A slight calculation 

 shows that the numbers at the 80% dilution are higher by 6%, 

 and at the 20% dilution higher by 15% than they should be on 

 account of the reduction of the acidity by the mere dilution. In 

 these weakly acid extracts, however, small differences such as 

 these are negligeable, as the counts themselves are liable to 

 greater fluctuations. 



A stage has been reached in the investigation at which it is 

 made clear that bacteria and moulds do not produce toxins, or, 

 if they do, the toxin is not capable of being demonstrated either 

 in nutrient solutions or in vegetable-humus. Any resemblance 

 to toxicity is probably caused by an alteration in the reaction of 

 the medium, and, to such alterations of reaction, the test-organ- 

 ism is very sensitive. 



The Growth of Amceb^e. 

 The attempt to obtain toxic substances among the byproducts 

 of certain bacteria and moulds had not been so successful as had 

 been wished ; in fact, it had been decidedly unsatisfactory. 

 Either the methods of producing the desired bodies were at fault, 

 or the micro-organisms were not such as would give the desired 

 results, although they had been selected as being the most likely 

 to do so. It is known that bacteria can give out bacteriolytic 



