366 



NA TURE 



[August 20, i! 



theory. Nor do I ; on the contrary, I regard it as an important 

 aid to the general application of the Darwinian theory. 



Apart from the objection, which I have above expressed, to 

 the treatment of this question by IVlr. Dyer as one of orthodoxy, 

 I note with regret that he has (unintentionally no doubt) mis- 

 represented what I have actually said in my letter. Whereas I 

 have expressly cited Mr. Darwin's principle of " correlation of 

 variations," Mr. Dyer writes of the " extended correlation 

 principle of Prof. Lankester." This form of pleasantry could 

 be it.self extended, but more of it would be unworthy of your 

 pages. As a matter of fact, what I said as to correlation was 

 very little more than a citation of cases and their theoretical 

 ■explanation given by Mr. Darwin. To represent this as in any 

 way parallel to the independent and anti-Darwinian theories of 

 Mr. Romanes and Mr. Bateson, as is done by Mr. Dyer, is 

 misrepresentation. The theory of Wells as to the black races 

 of the tropics was used by me in order to illustrate my suggestion 

 as to correlation leading to a development of useless specific 

 •characters. I might, equally well for my purpose, have used 

 any of the other cases collected by Mr. Darwin. Whilst I 

 spoke of Wells' case as "striking and suggestive," I at the 

 same time expressly referred to it as "a more or less hypo- 

 thetical case." Mr. Dyer has no justification, so far as I can 

 see, for stating that I use this case as a foundation for wide 

 generalisation. I made no wide generalisation, but adduced the 

 wide generalisation at which Mr. Darwin, after collecting and 

 •considering a large variety of cases, arrived, viz. that correlation 

 of variation does occur largely and generally in the organic 

 world. This wide generalisation is, I say, not mine ; it is Mr. 

 Darwin's. If this generalisation be accepted — and we may 

 reasonably hope that the apostolic and orthodox do accept it — 

 then it seems to me in the highest degree probable that an 

 obscure specific difference of structure — highly effective as life- 

 preserving or progeny-en.surmg — will more or less frequently 

 •carry with it as a correlated variation a more obvious and 

 measurable character in some remote part of the body, not 

 effective, that is to say, not useful. Hence I conclude that 

 there may be specific characters (not by any means all or always) 

 which are not themselves useful, though readily observed. That 

 is the whole of my contribution to the present di.scussion. It 

 ■does not seem to me to involve anything rash or surprising, 

 though such is its character according to Mr. Dyer. 



I may remind the readers of Nature that some years ago, in 

 ■these columns, I adduced this same principle, viz. Mr. Darwin's 

 principle of correlation of variations, as one sufiicient to remove 

 some of the difficulties in the way of the doctrine of natural 

 selection brought forward by the Duke of Argyll. What the 

 Duke called " prophetic germs" might, it seemed to me (and 

 still seems to me), when not explicable as lapsed rudimentary 

 structures, be accounted for as variations or new structures cor- 

 related with other useful and therefore selected variations, 

 although not yet themselves useful. A useless variation corre- 

 ilated with a useful one must (it seems to me) be supposed to pass 

 through initial stages in which it is too small, or otherwise 

 insignificant, to be useful (its utility or harmfulness being 

 swainped in the utility of the correlated useful character), and 

 only after attaining considerable development becomes either 

 ■useful or harmful, and therefore subject to selection, possibly 

 ■under some slight change of environment. 



I regret all the more my differences with my friend Mr. Dyer, 

 because the explanations they have involved leave me so little 

 space in which to refer to Prof. Weldon's courteous and interest- 

 ing letter. I have only one point to correct in his statement of 

 my position in relation to that taken by him. The attempt to 

 reconcile the dicta of Hume or Kant, or even of Mill, with the 

 experience and approved practice of those who make it their 

 ■business to investigate natural phenomena, would be an interest- 

 ling undertaking, possibly one beyond the powers of living man. 

 It is certainly not one upon which I shall here and now 

 embark. 



The point whereon Prof. Weldon has misunderstood my con- 

 tention is this. After describing a phenomenon (death-rate) 

 preceded invariably by two or more phenomena of structure or 

 function, he says : " Under these circumstances. Prof. Lankester 

 thinks it legitimate to pick out one of these antecedent 

 phenomena and to speak of it as the only effective cause of 

 change in death-rate, the other phenomena, although equally 

 imiversal, being merely unimportant concomitants of this one 

 ■essential change." Prof. Weldon is mistaken in stating that I 

 think it legitimate " to pick out" without qualification, or at 



NO. 1 399, VOL. 54] 



haphazard, any one of these antecedent phenomena, and to 

 speak of it as the only effective cause. I .should object to such 

 a proceeding on much the same ground as that on which I object 

 to his calling " any and all " of those antecedent phenomena 

 effective causes. 



What I think is the reassonable course in such a case — sup- 

 posing that a man wishes to ascertain, as fully as may be, the 

 relation of these phenomena to one another, is that he .should 

 frame in his mind a hypothesis as to how any one or more of the 

 phenomena, invariably associated with a given death-rate, can 

 operate so as to effect an increase or decrease in that death-rate. 

 This, no doubt, will require a large knowledge of the surround- 

 ing conditions not usually to be acquired uith ease, and an 

 analysis of the antecedent phenomena in question, often of a 

 prolonged and laborious character. Sometimes, however, .such 

 a hypothesis will present itself very readily and with much 

 antecedent probabiHty. However attained, the hypothesis will 

 remain merely a guess until it is tested. It can be tested either 

 by experiment or by observation of appropriate natural instances. 

 By repeated testing, involving often great ingenuity and pro- 

 longed labour, the hypothesis is either confirmed or discarded ; 

 po.ssiblya new hypothesis is adopted, so to speak, en route, and 

 established as in all probability true. When— and not until — 

 this process has been gone through, the naturalist will be, more 

 or less according to the extent of^ his work, in a position to 

 filaie the phenomena in their true relation to one another and to 

 the ultimate phenomenon proposed for investigation, viz. death- 

 rate. If the study of the antecedent or associated phenomena 

 by means of hypothesis and test-experiment has not been and 

 cannot be carried out, the naturalist can not (it seems to me) 

 reasonably either " pick out " one of them and assert that it is 

 the cause of increased or decreased death-rate, nor (still less) 

 declare that all the antecedent or associated phenomena are the 

 causes and produce the effect of increased or decreased death- 

 rate. If he does do so, he appears to me to be evading the task 

 before him, which is to "explain," that is, to place in their 

 true order and relation a complex group of related phenomena. 



The appeal to analogy no doubt frequently leads naturalists 

 to rapid conclusions as to the causal relations of the phenomena 

 of organisms Often (but by no means always) such conclusions 

 are erroneous : but it is not on that account desirable to reject 

 the argument from analogy in reasoning about such matters. It 

 must be used with knowledge and with caution. It seems to 

 me, that in considering a complex case in which actual experi- 

 ment is as yet wanting, it is often more useful to formulate 

 provisional conclusions as to what is cause and what only con- 

 comitant effect by the aid of argument from analogy, than to 

 deliberately reject all attempts at analysis, and to " lump" all 

 the constantly associated phenomena as "causes." Surely in 

 the case of Prof. Weldon's crabs, most naturalists would take 

 the view that the frontal measurements may puisihly be opera- 

 tive in saving the life of the crab, or may be only a correlative 

 of some other life-preserving structure ; that its quality in this 

 respect should be inquired into by means of hypothesis and 

 experiment : and that, until this is done, it '\i, premature to speak 

 of a particular frontal proportion as having for its effect the 

 survival of those crabs distinguished by its possession. 



The chief task of the student of living things seems to me to 

 lie in the .search for such explanations, even though the task is 

 in some cases to all appearance at first sight hopeless, and even 

 though too hopeful and imaginative spirits may be led, not un- 

 frequently, to propound explanations which are insufficiently 

 supported by observed facts, or are demolished by the observa- 

 tion of other facts. Voucannot (it seems to me) reduce natural 

 history, as Prof. Weldon proposes, to an unim-.iginative statistical 

 form, without either ignoring or abandoning its most interesting 

 problems, and at the same time refusing to employ the universal 

 method by which mankind has gained new knowledge of the 

 phenomena of nature— that, namely, of imaginative hypothesis 

 and consequent experiment. I think that most naturalists will 

 agree with [ohannes Miiller that " Die Phantasie i.st ein unent- 

 behrliches Gut." E. Kav Lankester. 



Dinard, Bretagne, August lo. 



Habits and Distribution of Galeodes. 



WiT.i, you be good enough to publish in your widely-read 

 paper the following notes on the geographical distribution of 

 So/fu^^a (Galeodes) araneoides, as the publislied accounts of that 

 arachnid are incomplete in that respect. 



