CTJLTURAL REQUIREMEXTS OF BACTERIA 333 



This particular experiment has been selected from a number 

 of a similar nature since it shows a fairly clean separation into 

 two fractions. Not infrequently one fraction or the other will 

 by itself reactivate the infusion to some extent and produce 

 slight growth, but almost invariably both together are better. 



Two points are illustrated in the above experiment. In the 

 first place, an actual separation has taken place. It is not a 

 distribution of the activating material through both fractions 

 with a corresponding dilution, which might reduce the concen- 

 tration below that required. In the second place, it is essential 

 to have some idea about the quantity of material which is being 

 added, best in terms of its equivalent of the original casein. 

 Ob\'iously, if, as is really the case, a quantity of casein hydroly- 

 sate, or of the first HgS04 precipitate, equivalent to 0.25 gm. 

 casein will reactivate 50 cc. of a mixture of decolorized infusion 

 and glucose salt solution, one must not test a later preparation 

 or fraction with a concentrated solution, using perhaps an equiva- 

 lent of 5 grams of casein or more. If such were done, and the 

 preparation found active, one might still have lost 95 per cent 

 or more of the active material and not recognized it, for bej^ond a 

 certain optimum, growth is not increased by multipljing the 

 quantity of reactivating material considerably. 



It has not been found possible so far to define the conditions 

 of this precipitation so exactly as always to obtain a complete 

 separation, and while it was useful in leading to the discovery 

 that there were two substances involved, it has now been given 

 up in place of a more certain method. 



