694 



SCIENCE 



[N. S. Vol. XXVIII. No. 724 



nymphal stage. Two species, Argia calida 

 Hagen and A. funebris Hagen, are known only 

 from the type specimens. 



Synoptic keys are given to the genera of the 

 six sub-families comprising more than one 

 genus each and to the species of forty-five of 

 the genera. Two genera {Hesperagrion and 

 Metaleptoiasis — ^both Agrioninae) and eighty- 

 one species have been described as new. 



Except in these eighty-one species and in 

 the genus Argia, the specific descriptions and 

 the figures on the plates are limited to features 

 unnoticed, or insufiieiently or incorrectly de- 

 scribed or figured in the previous literature, 

 which it is believed has been cited very fully. 

 The distribution of each species is given in 

 detail; the number and sex of the specimens 

 examined and the collector's name are stated 

 after each locality. To give the fullest in- 

 formation on such topics, the first of the two 

 tables in the introduction comprises an alpha- 

 betically-arranged list, by countries, of all the 

 localities from which Odonata are represented, 

 the state, altitude and temperature-zone of 

 each locality, the date of collection, the col- 

 lector's name, and often remarks on the phys- 

 ical character of the environment or the pre- 

 cise spot where the insects were gathered. 



The greater part of the nine years occupied 

 in the preparation of this work has been con- 

 sumed by the gathering and tabulating of 

 various characters — especially those of the 

 veining of the wings — which have been em- 

 ployed by previous writers to separate the 

 genera, or which seemed to lend themselves to 

 that purpose. These data (collected without 

 the aid of clerks or assistants), numbering 

 above one hundred and fifty thousand, were 

 reduced to percentages for each of the species 

 studied. Features which showed a variation 

 of ten per cent, or less were thereby assumed 

 to be of sufficient constancy to serve as generic 

 characters, and among these importance was 

 naturally assigned to those showing the least 

 degree of variability. Many of the specific, as 

 well as the generic, characters employed in the 

 work rest on a similar basis. At the same 

 time it must be remarked that the data are 

 not sufficiently numerous for any one species 



to furnish bases for mathematical formulse. 

 The limitations of time and strength and in 

 many cases also the available material forbade 

 the examination of more than twenty-five or 

 thirty individuals of a species, but not infre- 

 quently these were tabulated for twenty-five 

 different characters, which in the case of the 

 wind-details were noted for both sides of the 

 body in each specimen. Further statement of 

 this part of the work is not made here nor is 

 it more than hinted at in a large part of the 

 Biologia volume itseK, since it is hoped to 

 publish in another place tables of the percent- 

 ages obtained. 



Under each species, where possible, special 

 attention has been given to noting: (1) the 

 color changes through which the imago passes 

 from the time of transformation to the final 

 tints of old age — which has been done for 

 fifty-two species, the most extensive changes 

 being perhaps those of Hesperagrion hetero- 

 doxum (pp. 103, 377 and Plate VI., Figs. 

 1-6); (2) the geographical variations; (3) 

 the individual variations found in the same 

 locality. Owing to the conservative attitude 

 adopted towards species, many of these varia- 

 tions (2) and (3) will doubtless afford — indeed 

 have already afforded in one case — additional 

 " species " to later workers in this field, but in 

 these days one may perhaps reply to criticism 

 with de speciehus dividendis non disputandum. 

 In the matter of nomenclature some use has 

 been made of trinomials, in the sense of the 

 American Ornithologists' Union. 



The areas which have been most carefully 

 examined are portions of the Mexican states 

 or territories of Tepic, Jalisco, Guerrero, 

 Morelos, Distrito Federal, Tamaulipas, Vera 

 Cruz and Tabasco, the central belt of Guate- 

 mala from the Caribbean to the Pacific and a 

 few localities in Costa Rica. The odonatolog- 

 ically unknown area in Mexico and Central 

 America is, therefore much greater than that 

 which has been investigated. 



Of the physical data which have yet been 

 brought together, only those on temperature 

 are sufficiently complete to enable one to make 

 a natural division of Mexico and Central 

 America as a whole, with which the distribu- 



