1902.] NATURAL SCIEXCES OF PHILADELPHIA. 215 



from an early ancestral organ of diffuse arrangement one part 

 change? into one set of organs, another part into another set. Thus 

 from the gastrovascular system of a simpler Cnidarian -with, e. cj. , 

 a structure like Hydra, are found in higher forms developing intes- 

 tine and gonads (e. (j., Turbellaria). In all such cases where the 

 substitution of parts is gradual, it is obviously difficult to determine 

 where one organ ends and another begins, and hence to determine 

 how far back into the pliylogeuy an organ extends. And the sub' 

 stitution is generally gradual, so that the new organ either arises 

 from a part of the first, or the first becomes gradually moulded over 

 to form the new; much more rarely is the first completely lost 

 (ontogenetic development of the Diptera). It is not probable that 

 a part in any case is completely lost without first aiding in the forma- 

 tion of another part ; and this is the more true Avhen we consider 

 that accidental mutilations are probably not inherited. And further, 

 it must be borne in mind that a character which is essentially con- 

 servative throughout one group may be variable in another : thus 

 the external cuticle is comparatively uniform in structure throughout 

 the Nematoda, but very variable in the Gordiacea (in some species 

 of Avhich it differs even in the two sexes). 



A word may be said just here as to the value for classification of 

 the statistical study of individual variation. The examination of 

 the relative amount of variation in different parts of the same 

 species shows of course which are most and least variable. It shows 

 also that in some species there is less general variation than 

 in others. The modern school of statistical variation has not 

 brought to light any new points of view, but it has the advan- 

 tage of deriving its statistics from a large number of individ- 

 uals, and of reproducing these statistics in concise mathematical 

 forms. Before its oncoming observers contented themselves with 

 such loose expressions as ' ' very variable " " slightly variable, ' ' 

 etc. The accurate recording of individual variations is the scien- 

 tific method of showing which parts are most conservative, and a 

 character (e. g., number of joints in an antenna) which has been 

 shown to be veiy variable cannot be regarded as of much value ; 

 that is to say, such a character should be considered of little value 

 in the diagnosis of the species. Yet in closely related species such 

 a character may be eminently stable, so that in general no deduc- 

 tions can be drawn from these variation statistics to species other 



