NATURAL SELECTION 257 



Some of the most striking instances of resemblance between 

 insects belonging to widely separate groups are those be- 

 tween Hymenoptera and Diptera. If we consider the single 

 dipterous family, the Syrphidae, we find some species which 

 are indistinguishable (when flying) from wasps (Chrysotoxum 

 cautum, Paramixogaster spp., etc.) or from bees ( Volucella bombylans, 

 Pocota apiformis, etc.) . Often the resemblance is due to the modi- 

 fication of the body in different ways, as when a long twelve- 

 segmented antenna is imitated by one of three long segments, 

 or the folded wings of a wasp are imitated by a longitudinal 

 cloud along the costal margin of the wing of a fly (cf. also 

 Sturtevant, 1921 ; Nicholson, 1927). At first sight it seems 

 impossible to attribute such resemblances to parallel evolution, 

 even in part. But to judge the question properly it is necessary 

 to consider the whole range of colour-pattern found in the 

 Syrphidae. We then find that there is a complete series from 

 ' fly-like ' forms to bee- or wasp-like forms. It is difficult to 

 imagine that the little-modified members of such a series are 

 really mistaken for Hymenoptera by their enemies : it would 

 appear rather that there is a definite tendency in the Syrphidae 

 to produce bee- and wasp-like types ; possibly, when a certain 

 degree of resemblance has been hit off, selection may contribute 

 to completing the resemblance. In other words, such mimicry 

 is not the product of selection alone, and it is impossible in 

 any particular case to say what part selection has actually 

 played. Sturtevant {I.e. p. 202) has criticised the view that 

 occurrence of parallel mutations plays much part in mimicry. 

 He objects to drawing a distinction between mimicry and the 

 protective resemblance of insects, etc., to other objects (as 

 stick insects, leaf insects, etc.). But, as a matter of fact, resem- 

 blances to the inanimate background are already known to 

 be due to more than one cause — viz. either hereditary consti- 

 tution or power of changing colour during the life-history 

 (see discussion of specific differences in colour, p. 279). Again, 

 Sturtevant points out that parallel evolution cannot make the 

 leg of a fly resemble the antennas of a wasp. Generally 

 speaking this is true, but in the Syrphidae and many other 

 dipterous families, long, three-segmented antennas, super- 

 ficially resembling those of wasps, are well known to occur in 

 forms not resembling wasps in colour. Lastly, it is impossible 

 to show without elaborate genetic analysis that two mutations 



