216 The Mechajn^ism of Evolution in Leptinotabsa 



of single characters, and the neglect of these alone is sufficient to invalidate their 

 findings. The added difiBculty that these fluctuations were not capable of unlim- 

 ited accumulation through quantitative selection seemed an insuperable obstacle 

 to De Vries, and while there is no doubt of the general correctness of this limi- 

 tation and the frequent regressions to the mean, both limitations and regressions 

 are due to other agencies than those considered by De Vries. 



ANALYSIS OF HETEROGENEITY OF SOME SIMPLEST STRUCTURAL 



CHARACTERS. 



With structural characters it is necessary to apply the same criteria for theii 

 determination as in the color-characters, in that they must be the indivisible 

 resultant of the interaction of a single constellation of interacting agencies, the 

 absence of any one of which results in the non-production of the character, in 

 precisely the same manner that the absence of an agent in a complex physical 

 system results in the non-production of the proper product of the operation of the 

 mechanism. 



IN THE ADULT. 



Prime favorites with the statistician and the taxonomist are such " structural " 

 features of the organism as the dimensions of parts, length and breadth or 

 indices, form, i. e., elongate, robust, depressed, general aspect of the surface, 

 rugose, glabrous, punctate, and so on without limit. These characterizations of 

 the appearance and form of the object serve well in the operations of the taxono- 

 mist to distinguish the different species, but finer analysis is needed for the 

 investigation of the problems of the origin and meaning of diversity in organ- 

 isms, and especially in the effort to untangle some of the conflicting statements 

 regarding the problems of heterogeneity. 



In the case of structural characters, the same set of questions need to be asked 

 and investigated with care as in the case of color-characters, and the same exact 

 requirement needs to be followed so that homogeneous states shall be investigated. 

 Both must be the indivisible product of a group of interacting agencies, in which 

 the absence of any member is followed by the non-production of the character in 

 question. It is necessary to deal with simplest structural characters, as the basis 

 of the analysis. 



In the organism with which I have been working there is an unusual dearth of 

 spines, scales, or other ornamental characters of a " structural " type, only the 

 integumentary punctations being present. It has, therefore, been necessary to 

 use more important portions in the study of the characters of this class, although 

 it is by no means as easy to obtain simplest conditions as in color-characters. 



Body-weigM. — One of the first characters investigated was the variability in 

 the body-weight of the population, which, obviously, is the product of a large 

 number of contributing factors in the population and in the individual. A deal 

 of time was spent in 1899, 1900, 1901, and 1903 in the statistical study of body- 

 weight in L. decemlineata Say. Both the live weights and the dry weights were 

 obtained, but the outcome of the study was the demonstration that the method 

 and the material could give nothing of value. I was using materials from 

 nature, and these were obviously of unlike composition genetically, of dissimilar 



