1 62 BIOLOGICAL LECTURES. 



some other mode must operate in producing these structures, 

 a mode, which, it would seem, depends rather on independently 

 inherited and typically formative forces in individual organs. 

 This latter formative mode appears to us more difficult than the 

 former for the very reason that it requires a whole series of 

 independent, typically localized individnal formations for the 

 building up of a single structure. But that this formative mode 

 must in reality be carried out with very great ease is shown by 

 the difference in the rich and beautiful pattern of birds' plumage 

 in closely allied species, although in every such plumage every 

 feather, characterized as it is by its position on the body and 

 its relation to the other feathers, must have its own typical pat- 

 tern, differing from that of neighboring feathers in a typical 

 manner. 



" Certainty " in causal deduction cari only come from experi- 

 ment, either from " artificial' ' or from ''natures'' experiment, 

 such as variation, monstrosity, ox ot\\e.r pathological phenomena; 

 this certainty, however, is only to be obtained by adhering to 

 various precautions which are often difficult to follow. 



In an experiment performed under the most favorable condi- 

 tions, only one of the components known to us is or will be 

 changed, and through the results of this change we appre- 

 hend those phenomena which are connected with this compo- 

 nent. 



In practice, however, matters are not so simple ; for in 

 organic objects even after artificial, analytical experiment we 

 often experience the greatest difficulty in tracing back the 

 effects to their true causes ; in the first place we are obliged 

 to repeat the experiment often in order to obtain constant 

 results and then it must be modified in various ways in order 

 that we may be able to determine the true causes. This is 

 because the conditions are so complicated that we do not know 

 the primarily altered components even by means of artificial 

 interference, since, when we suppose we have succeeded in 

 changing only a single component, accidental external or inter- 

 nal conditions or unintentional collateral effects of our own 

 interference have already affected several components. Only 

 when we are perfectly sure that in reality no other than the 



