NORTH AMERICAN ODONATA. 213 



eyes are united on the top of the head, and the abdomen develops 

 lateral longitudinal carina ; here are the most powerful in flight of 

 all Odonata. 



As a second line from the Petaluroid forms, come the Cordulegas- 

 terinse ;* here the median lobe of the labium remains cleft, the eyes, 

 although often meeting dorsally, do so only in a single point, but the 

 genital valves disappear. 



Thirdly are the bulk of the Gomphinse which ultimately have the 

 median labial lobe entire, lose the genital valves and the ovipositor, 

 have the meso- and metathoracic ganglia united, but preserve the 

 primitive characters of separated eyes and absence of lateral ab- 

 dominal carinpe. 



From some point along the Gomphine line, the Libellulida3 can be 

 conceived as having arisen. From Gomphine ancestors they in- 

 herited the absence of an ovipositor, and perhaps the union of the 

 last two thoracic ganglia ; the eyes meet upon the top of the head, 

 the labium loses all trace of the bifid (bilateral) character of the 

 median lobe, which is quite small as compared with the lateral lobes ; 

 a change takes place in the venation of the wings, so that the triangle 

 of the front wings is elongated in the direction of the short axis of 

 the wing, and the triangle of the hind wings (in the more specialized 

 genera) although retaining its direction of elongation, comes to lie 

 with its inner side in the prolongation of the arculus ; and lateral 

 carina appear on the abdomen. The relationship between the two 

 subfamilies of the Libellulidse is not yet apparent; the most im- 

 |)ortant systematic character separating them is a small tubercular 

 projection present on the hind margin of the eyes of the Cordulinse. 

 Perhaps we are to look upon such Corduline genera as Somatochlora 

 as the most specialized of all the Odonata. 



The preceding view of the relationships of the various groups to each other is 

 based entirely on morphological evidence. There is, apparently, only one im- 

 portant morphological fact which is not in favor of regarding the Calopteryginse 

 as the most primitive group, and that is that these dragonflies have an ovipositor, 

 while neither the Perlina nor the Ephemerina possess such. For this reason Mayer 

 regarded the Libellulidse as most approaching the hypothetical Protamphibiou, 

 one of whose characters (see above) was "no ovipositor;" he consequently 

 looked upon the Agrionidse on one hand and the Aeschnidse on the other, as 

 derived from a Lib;ellulid stock. If the Protamphibion did not have an ovipos- 

 itor, to look upon the Calopterygiufe as the writer does, means that the Gom- 



* Here and elsewhere throughout this paper the subfamily Cordulegasterinse 

 is equivalent to the legion Cordulegaster of Selys alone, and not of the extent 

 of the Division Cordulegasterina of Kirby (35). 



TEANS. AM. ENT. SOC. XX. AUGUST, 1893. 



