616 PROCEEDINGS OF THE NATIONAL MUSEUM. 



but as a rule taxonomists search onl}^ for characters indicating a 

 difference in kind. 



Specialization may take place in two very different wajs^-'^Jirst, bj^ 

 an addition or complication of parts, Hpecializatlon hy addition; second, 

 by a reduction in the number or in the complexity of parts, upeciaUza- 

 tion hy redact !o)i.'' The specializations to be considered later are all 

 of this lattei' type. 



It should also be borne in mind that when an organ disappears in 

 any phyllum or line of ascent it can not reappear in the descendants 

 of this phyllum, though they might develop a substitute for it. t^ven 

 if such a substitute should be developed, it is not probable that the 

 substitute would resemble the organ so closely as to be mistaken for it. 



In determining the phylogeny of any group, those characters indi- 

 cating a difference in degree of specialization of the same kind are the 

 most useful in allotting the rank of the different groups. Every large 

 group has numerous characters indicating a difference in degree of 

 specialization of the same kind. Certain of these characters show the 

 ascent of the group as a whole, while others show only small lateral 

 lines of ascent or a sidewise development. Characters indicating a 

 sidewise development frequently arise independently several times, 

 and do not indicate anx'thing as to the line of ascent of the group as a 

 whole. This is illustrated by the presence or absence of the radial 

 cross-vein in the families Xiphydriidfe (tig. 85) and Tenthredinidte. 

 The presence or absence of this cross-vein is of value in indicating the 

 line of ascent of the genera of each of these families, but is worthless 

 so far as indicating any rank between the families themselves. There- 

 fore care nmst be taken to differentiate between those characters that 

 show the ascent of the group as a whole and those characters that 

 show only a sidewise development. 



The front wing of the original progenitor of the Hymenoptera, and 

 therefore of the Tenthredinoidea, was imdoubtedly very similar to the 

 one already described as the typical hymenopterous Aving (fig. 8). This 

 wing contains not only all those parts that are generall}' wanting in 

 the Hymenoptera, but the various parts are arranged in the most 

 primitive condition known to us, as can readil}^ be seen by comparing 

 this Aving with those of the Xyelida> (hgs. 31-35) and Lydidsi? (tigs. 

 36-43). No hymenopterous wing contains all the veins shown in the 

 tj^pical wing, but by combining the wings of the families just named 

 the wanting parts can be readily supplied. 



The characters that have been found the most useful in determining 

 the ascent of the Tenthredinoidea are the position of the radial cross- 

 vein, the position of the medio-cubital cross-vein, and the reduction of 

 the anal cells of the front wings. 



Hitherto the special modifications of the wing veins of the Tenthre- 

 dinoidea have been considered in detail, particularly with respect to 



