THE PRESIDENT'S ADDRESS. 75 



nature of the material (opal) of which the microsclere is com- 

 posed ; (2) the nature of the medium in which it is deposited, 

 viz. the colloidal cytoplasm of the cell ; and (3) the presence of 

 the cell-membrane, by which the growth of the spicule is to some 

 extent restrained and guided. All three are, however, doubtless 

 dependent upon the hereditary constitution of the mother-cell 

 (including, of course, its nucleus), for while the mother-cells in 

 siliceous sponges secrete hydrated silica, those of the Calcarea 

 secrete carbonate of lime, and so on. 



We have next to inquire how it is that, if the specific forms 

 of sponge microscleres are of no importance to the sponge, such 

 very remarkable forms should ever have arisen in the course 

 of evolution. We have to remember in this connection that we 

 are dealing not merely with a few isolated and unrelated forms, 

 but with progressive evolutionary series along lines as definite 

 as any other lines of evolution with which we are acquainted, 

 and which certainly seem to require some directive force to 

 explain them. If we were dealing with adaptive characters we 

 should at once say that the result was due, as in the case of the 

 megascleres, to the natural selection of small, fortuitous, favour- 

 able variations ; but the fact that the characters in question are, 

 for the most part at any rate, not adaptive, seems, at first sight 

 at any rate, to rule natural selection out altogether. 



It might be suggested, however, that the solution of the 

 difficulty is to be found in the well-known principle of correlation. 

 In accordance with this idea certain characters of an organism 

 are inseparably linked together with other characters in such 

 a way that any variation in the one must be accompanied by 

 a corresponding variation in the other, though the reason why 

 such characters should be so linked together is often by no means 

 obvious. To upholders of such a view as this the analogy of 

 by-products, upon which I laid so much stress at the beginning 

 of my address, may, I think, prove useful. Although I doubt 

 whether the hypothesis of correlation is adequate to meet the 

 present case completely, it certainly seems worth while to 

 examine it a little more closely. 



I may illustrate my meaning by reference to the action of a 

 few drops of acid upon an alkaline solution of litmus. Two per- 

 fectly distinct results will be produced. The solution will become 

 acid and it will change from blue to red. You may desire for 



