HISTORICAL SKETCH 15 



and much valuable information. In 1902 Giles published a second edition of 

 his work and in this the classification was adjusted as far as possible in accord- 

 ance with the Theobaldian views. In 1904 he published " A Revision of the 

 Anophelina" which follows closely the scheme laid down in Theobald's third 

 volume. 



In 1905 Blanchard published " Les Moustiques," which brings together in 

 most convenient form our knowledge of mosquitoes up to that time, both from 

 the systematic and the medical and economic sides. It contains a very complete 

 bibliography and on that account is a most invaluable work. The classification 

 followed is that of Theobald, but set forth much more clearly than by its author. 

 There are many emendations and changes in nomenclature in accordance with 

 the laws of priority. 



In conclusion, some remarks as to the merits of the different systems of classi- 

 fication seem called for. 



It is generally agreed that the best classification of any group of animals is a 

 natural one; that is, one that agrees with the phylogeny of the group. To 

 determine what characters are natural ones in this sense, considerable experi- 

 ence is required, not only with the group under study, but with characters of 

 insects in general. No single set of characters should be used, but the subject 

 approached from all possible points of view. There are certain general principles 

 that must not be violated. For example, generalized forms must not be derived 

 from specialized ones, and this will apply to special organs and to any stage 

 of the insect. Superficial characters, or those that are readily affected by 

 adaptation or specially useful to the insect, should not be made the basis of 

 fundamental divisions. Characters that appear in one sex only are un- 

 desirable to use. These represent, as Williston says, evolutional instability ; 

 for it often happens that a modification of an old character will appear in one 

 sex first, and be gradually transferred to the other sex. Therefore where 

 characters, not primarily related to reproduction, differ in the sexes, it may be 

 considered that such species have not reached equilibrium, and the use of their 

 characters for groups higher than specific is not warranted. 



Viewed from this point, it must appear that the classification of the mos- 

 quitoes has pursued a vicious course from the beginning. The first separation 

 of genera by Meigen was based on secondary sexual characters; the second 

 separation, that of Desvoidy, made use of superficial adaptational characters 

 of vestiture, thus violating from the beginning the principles above laid down. 

 Arribalzaga continued in the same course, adding to the undesirable list other 

 more detailed, and therefore more objectionable, characters of a secondary 

 sexual nature. Meigen's introduction of the relative length of the palpi in the 

 two sexes was particularly unfortunate. It was based upon a very few forms and 

 indicated in the most superficial way, and more than anything else hindered a 

 more exact study of the Culicidae. In fact no attempts to understand the palpi 

 of the different forms structurally were made until very recently by Neveu- 

 Lemaire (1902) and Felt (1905). 



