THE ORIGIN OF VARIATION 35 



lost when the animals are returned to the control environment. 

 It is very difficult to decide what degree of permanence in the 

 modification must be established to prove the possibility of 

 the process. It is at least necessary that the modification 

 should be partially maintained for at any rate one generation 

 after the return to control conditions. Actually, in quite a 

 number of experiments no return to the control environment 

 was ever attempted. 



4. The Value of Negative Evidence. — No amount of un- 

 successful experiments can prove that modifications do not 

 gradually become hereditarily stable. Under natural condi- 

 tions it might require many thousands of years for the 

 modification to become permanent. 



On the other hand, the experiments should not entail 

 subjecting the animal to conditions very unlikely to be met 

 with in nature. If many thousands of years are required to 

 produce a stable modification, it is probable that only a few 

 simple agencies, such as low or high temperature or changed 

 salinity in the sea, can be effective. Few other environmental 

 factors are likely to operate steadily for long periods. 



(b) Experimental Evidence, (i) Experiments on Protozoa. — 

 This work has been summarised critically by various authors 

 (see references in Robson, 1928, p. 168 ; and Hammerling, 

 1929). The bulk of the work (Jennings, Jollos and others) 

 concerns such forms as Paramoecium and Arcella and consists 

 in their habituation to altered temperature-conditions or to 

 doses of arsenic or calcium salts. Reversible modifications 

 (' Dauermodifikationen ') are frequently found. Some (e.g. 

 ' calcium-dauermodifikationen ') are in all probability deter- 

 mined by changes in the cytoplasm and reversion follows on 

 the return to normal asexual reproduction after conjugation 

 (e.g. in Paramoecium). In Bacteria also the changes are still 

 manifested after transplantation to a new medium. 



(ii) Experiments on Metazoa. — There is substantial evidence 

 that lesions are not inherited. We need mention only such 

 practices as circumcision, modification of shape of head or feet, 

 docking of tails, etc., which produce no heritable effect after 

 hundreds of generations (cf. also Agar, 1931). 



There is a large number of experiments which may be 

 set aside or regarded as so questionable as to be practically 

 worthless as evidence. These are dealt with very briefly. 



