METHODS 93 



Instead of venom or toxin, let us inject into an animal 

 quite harmless substances like milk, calf's blood or beef 

 liver. To all appearances we shall have no means of know- 

 ing if the animal has been modified by the injection. If not 

 disturbed at all by the first injection, it cannot be less 

 disturbed by the second. And yet it is hard to say that a 

 colloid substance can be quite harmless ; injected in suffi- 

 cient doses, it will clearly kill the patient by indigestion. 

 But the symptoms of uneasiness after the injection of a 

 reasonable dose are so slight that it is impossible to observe 

 a diminution of them in the second injection. 



Luckily, at least in certain cases, a very important pecu- 

 liarity allows us to verify indirectly a variation of the organism 

 even with colloids which seem quite harmless. This remark- 

 able peculiarity has to be studied in the next chapter ; it 

 is the starting point of a fruitful method known for some 

 years back under the name of serumtherapy. 



First, there is an observation useful to make. 



Habit has limits, as we explained. Once an animal is 

 habituated to the maximum under given conditions, it is able 

 to undergo these conditions without experiencing hence- 

 forward any variation. When, therefore, a living being 

 has been long subjected to circumstances which have not 

 varied, the law of functional assimilation gives way to the 

 rigorously exact law of assimilation, which we verified as 

 an approximate law in the third part of this book. This 

 is how we were able to verify assimilation without variation 

 in an anthrax bacteridium tvatched by a skilful experimenter 

 in its bouillon. We might say the same thing of a bacteri- 

 dium which has already killed several sheep and conse- 

 quently acquired the maximum virulence for sheep, should 

 it multiply without variation in a new sheep. The law of 

 functional assimilation is relative to the periods of change 



