MULTIPLE CHROMOSOMES 545 



stituent part and thus demonstrates the completeness of the 

 series. In class 3 there are ten separate chromosomes, of which 

 one is a multiple involving the accessory chromosome as in class 

 2, besides which there is a very large octad similar to the one 

 in classes 4, 5, and 6^ Class 6 differs from 3 only in respect to 

 the non-union of the accessory chromosome with a tetrad. The 

 one individual of class 5 a, of which I have only a smear prepa- 

 ration, presents the second instance of a supernumerary chro- 

 mosome which has appeared in my material up to the present. 

 Aside from this, the complex is like that of class 5. From all 

 these considerations I think it may be said that the numerical 

 variations within these specimens of H. viridis, instead of being an 

 indication of lack of stability in organization of the chromosomes, 

 are, in fact, very strong proof to the contrary. There is noth- 

 ing whatever to indicate that the integrity of any element is 

 lost, or, aside from the case of the supernumerary, that any 

 additional members are added to the complex. Variation is not 

 a question of loss or gain, but of relations. 



2. Chromosome numbers in general 



Variations in chromosome numbers, within the species and 

 larger groups certainly exist, as is clearly manifest in the work 

 of numerous investigators. The constancy of numbers is more 

 definitely established in some groups of animals than in others. 

 Thus the Acrididae have a common number for many of the 

 genera, but in the Hemiptera, as Wilson and Montgomery have 

 indicated, the families are much less uniform numerically. In 

 using the conditions of the family which I have most studied as 

 an indication of chromatin organization, I have recognized 

 that the nature of this organization might be different in other 

 groups, and I have not attempted to apply generally the imme- 

 diate conclusions to which I inclined. • Not only is this true but 

 the specific statement to that effect appears in an early paper 

 of mine ('08 a) in these words: 



I am quite prepared to admit also that in one species even, there 

 may be a variation in the integration of the chromatin material, result- 

 ing in some numerical variation of the chromosomes, without losing 



