1 8 CONNECTICUT GEOL. AND NAT. HIST. SURVEY [Bull. 



The descriptions of nearly all species listed in subsequent pages 

 have been made directly from the specimens themselves and were 

 compared with descriptions by other workers, including the orig- 

 inal in many cases. Descriptions of the rarer species have in 

 some instances been compiled from the work of other authors. 

 Every effort has been made to bring the systematic part up to 

 date and to make it accurate and reliable. It is to be kept in mind, 

 however, that variation among the dragonflies is considerable, and 

 it is often impossible to state exactly the limits of these variations 

 in terms that the reader will easily grasp. Generic descriptions 

 have been made rather full, and care taken within large families to 

 include the same characters in each for the sake of comparison. 

 They form a summary of the matter contained in the keys rather 

 than a reference to the latter. 



Attention is called to a character of some taxonomic value in 

 the separation of the Cordulinae and Libellulinae. This is the 

 amount of curvature in the mesopleural suture. In all specimens 

 examined, the corduline type shows this suture to be practically 

 straight from the wing base to the katepisternum (infraepister- 

 num) whereas in all Libellulinae examined, there is a decided 

 upward bend just before reaching the katepisternum (Fig. 40). 



The Catalogue of Muttkowski has been mainly followed as to 

 nomenclature of species and families but is modified in a few 

 instances on account of work done since the publication of that 

 list. As regards systematic tangles, the Odonata are compara- 

 tively free, which is for the most part due to the excellent work 

 of Calvert, Walker and Williamson, and to the earlier work of the 

 Belgian, Baron de Selys. It is to be hoped that future workers 

 follow the lead of these scientists. For the sake of convenience, 

 the Anisoptera have been divided into five families instead of two 

 as commonly used. This arrangement may be criticised but has 

 some precedent in the publication of Hart and Needham (1901) 

 and in the statement of Tillyard (1917, P- 258). This also seems 

 to be a more natural grouping, since the families show nearly equal 

 differentiation in structure. In point of numbers, on the other 

 hand, the families are very unequal, but this is held to be no serious 

 obstacle to their adoption as families. 



The External Anatomy of the Dragonfly 



The parts of the skeleton of any insect are similar, even in 

 widely separated groups ; but through the process of evolution 

 many parts have been lost or so modified that they are only to be 

 recognized by means of "landmarks" in the structure of the body 

 wall. Moreover, the features used in classification are often so 

 different in different orders that specialists do not agree as to the 



