CLASSIFICATION OF DIPTERA. 6! 



Without extending these remarks to an undue length 

 it may be well to set down here the chief lines of evolu- 

 tion in diptera. As they seem to me they are as follows: 



1. Reduction in the number of longitudinal veins from the primitive 



eight or nine; and of their branches. 



2. Shortening of the most anterior and posterior of these, and the re- 



duction of the basal cells. 



3. Weakening of the posterior veins; loss of the marginal vein behind; 



loss of the primitive cross-veins. Powerful flyers with strong 

 orienting powers have a supporting venation before the hind 

 margin, produced either by adventitious cross-veins or the 

 closure of cells. 



4. Loss of antennal joints from thirty or more to three or two, 



by reduction in number of homologous joints; by the progress- 

 ive fusion of the distal joints into the so-called 'complex'; or 

 into the style or arista; or by the atrophy of the basal joint. 

 The development in size of the simplified antenna; or the 

 production of structural peculiarities. 



5. Loss of palpal joints, and, as in the antennae, the development of 



the simplified palpus. 



6. The development of holopticism from a primitive clichopticism. I 



do not believe that the reverse is probable the Acalypterae 

 have not descended from the Calypterse, for instance, and these 

 latter are, in this respect at least, as in others, the more highly 

 .specialized insects, just as Bibio is more highly specialized than 

 Mycetophila. 



7. LOSS of ocelli; diminution and loss of the compound eyes, espe- 



cially characteristic of ectoparasites. 



8. Diminution in number of abdominal segments; the closer fusion of 



the thoracic segments. 



9. Loss of tarsal joints; loss of empodium. 



These of course are not all the lines of evolution in 

 diptera, but I believe that they are all irreversible, that 

 evolution has never recovered anything once function- 

 ally lost. Moreover all, or nearly all these lines of evo- 

 lution are polyphyletic, resulting in numerous cases of 

 parallel resemblances which must be taken into account 

 in an} 7 attempt at true classification. Heteropeza among 



