Vol. XXviii] ENTOMOLOGICAL NEWS. II 



In this paper I have used Megalagrion for these Hawaiian 

 forms. The genus Megalagrion was established by McLach- 

 lan* for blackburni and oceanicum, with blackburni the type of 

 the genus. Perkins later described ketcrogamias and con- 

 sidered it a close relative of these. He also pointed out that 

 certain individuals of species in Group 6, especially of kaiiai- 

 cnse, showed the venational characters of Megalagrion, thus 

 preventing the use of this generic name to set off these three 

 with the richly veined wings from the other Hawaiian forms. 

 Perkins placed the entire group in the old genus Agrion, now 

 Coenagrion. Studies of the penes in these forms show that 

 the genus Coenagrion can probably be broken up and that 

 these Hawaiian species are a compact group quite distinct 

 from the other groups. As Megalagrion has been used for 

 some of these Hawaiian species, it will then become applicable 

 to the entire series of Hawaiian Agrionines as I have used it. 



I had hoped that the penes more than the other characters 

 might give some clue to the relationship and probable origin of 

 this group. The penes do show that these Hawaiian Agrion- 

 ines in spite of the great range in their appearance and struc- 

 ture are a compact group and undoubtedly have been derived 

 from some single ancient immigrant that had strayed into the 

 islands. It lines the Odonata up with what is already known 

 about the birds (Drepanidae), the land snails (Achatinellidae) 

 and those orders of insects in which there are large endemic 

 genera with apparently diverse but really closely related spe- 

 cies. These strange groups have probably in each case been 

 derived from some single ancestor which has strayed into the 

 islands in the remote past. 



More difficult is the origin and probable relationship of this 

 ancestral Agrionine. A study of the penes in the species listed 

 by Kirby as Coenagrion shows that the extra-Hawaiian forms 

 fall into at least two groups, the group of which puella is the 

 type (see figs. 39, 46-49) and the group of which llndenii is 

 the type (see figs. 40-45). As lindcnii has been given generic 

 rank by Navasf as Cercion lindcnii, probably Cercion can be 



*Ann. and Mag. Nat. Hist. (5), Vol. 12, p. 237. 

 fBroteria 6, p. 55, 1907. 



