September 11, 1914] 



SCIENCE 



371 



parative study fully justifies tliis conclu- 

 sion. How, then, are we to regard this 

 insistent problem of parallelism and con- 

 vergence from the point of view of genetic 

 study? 



A belief in the "inheritance of acquired 

 characters," or, as it is sometimes ex- 

 pressed, "somatic inheritance," is at pres- 

 ent out of fashion in some quarters. But 

 though powerful voices may seem to have 

 forced it for the moment into the back- 

 ground, I would take leave to point out 

 that such inheritance has not been dis- 

 proved. All that has been done, so far as 

 I understand the position, is to show that 

 the evidence hitherto advanced in support 

 of it is insiifficient for a positive demon- 

 stration. That is a very different thing 

 from proving the negative. "We hear of 

 "fluctuating variations" as distinct from 

 "mutations"; and it is asserted that the 

 former are somatic, and are not inherited, 

 while the latter are inherited. This may 

 be held as a useful terminological distinc- 

 tion, in so far as it accentuates a difference 

 in the heritable quality. But it leaves the 

 question of the origin of these heritable 

 "mutations" quite open. At the present 

 moment I believe that actual knowledge on 

 this point is very like a complete blank. 

 Further, it leaves indefinite the relative 

 extent and proportion of the "mutations." 

 It is commonly held that mutations are 

 considerble deviations from type. I am 

 not aware that there is any sufficient 

 ground for such a view. It may probably 

 have originated from the fact that the 

 largest are most readily observed and 

 recognized as reappearing in the offspring. 

 But this is no justification for ignoring the 

 possibility of all grades of size or impor- 

 tance of heritable deviations from type. 



On the other hand, adaptation, with its 

 consequence of parallel or even convergent 

 development in distinct stocks, is an in- 



sistent problem. The real question is, 

 What causes are at work to produce such 

 results? They are usually set down to the 

 selection of favorable divergences from 

 type out of those produced at random. 

 But the prevalence of parallelism and con- 

 vergence suggests that those inheritable 

 variations, which are now styled "muta- 

 tions," are not produced at random. The 

 facts enforce the question whether or not 

 they are promoted and actually determined 

 in their direction, or their number, or 

 their quality, in some way, by the external 

 conditions. Parallelism and convergence 

 in phyletic lines which are certainly dis- 

 tinct impress the probability that they are. 

 Until the contrary is proved it would, in 

 my opinion, be wiser to entertain some such 

 view as a working hypothesis than posi- 

 tively to deny it. Such a working hypo- 

 thesis as this is not exactly the same as a 

 "mnemic theory," though it is closely akin 

 to it. It may perhaps be regarded as the 

 morphologist's presentation, while the 

 mnemic theory is rather that of the physi- 

 ologist. But the underlying idea is the 

 same, viz., that the impress of external cir- 

 cumstance can not properly be ruled out in 

 the genesis of inheritable characters, simply 

 because up to the present date no definite 

 case of inheritance of observable characters 

 acquired in the individual lifetime has 

 been demonstrated. Of course, I am 

 aware that to many this is flat heresy. At 

 this meeting of the association it amounts 

 almost to high treason. I plead guilty to 

 this heresy, which may by any sudden turn 

 of observation be transformed into the true 

 faith. I share it in whole or in part with 

 many botanists, with men who have lived 

 their lives in the atmosphere of experiment 

 and observation found in large botanical 

 gardens, and not least with a former presi- 

 dent of the British Association — ^viz., Sir 

 Francis Darwin. 



