180 DEVELOPMENT OF THE WING-VENATION OF ODONATA, 



the posterior border or "sole" of the now stocking-shaped or 

 Italian loop. The loop is still only two cells wide for the most 

 part, but tends to widen at both ends, so that it is not unusual 

 to find three or more cells in a row, both at the base and towards 

 the tip. Finally, by the convergence of the distal ends of A, and 

 Cuob, an exceedingly elongated and pointed "foot" may be formed, 

 such as is to be seen in ]}feurothemis (Needlvdm, loc. cit., fig. 18, 

 p.724). 



We have now to consider the formation of the midrib or 

 symmetrical axis of the Italian loop. By reference to Fig. 11, it 

 will at once be seen that this is not formed about a trachea at 

 all. The midrib is, in fact, nothing more nor less than a cubital 

 supplementary sec^o?' (Cuspl), formed on exactly the same lines 

 (and even carrying, in Heniicordnlia, the same mass of pigmenta- 

 tion) as Mspl under M^ and Pvspl under Rs, though attaining a 

 far greater measure of usefulness owing to its fortunate position. 

 In cases where the "toe" is not fully formed, as in Ilemicordulia, 

 this Cuspl runs straight to within one cell's breadth of the "sole." 

 But where there is a well-developed toe (e.g., in Diplacodes, Plate 

 xi., fig. 9), it is cut off by yet another branch of Cuo (CuoJ, which 

 descends from Cug about midway between its point of bifurcation 

 from Cuj and its point of branching to give oflP Cugb, and runs 

 almost straight to the tip of the toe. Hence, in this extreme 

 case, the midrib is formed chiefly by Cuspl; but its apical end, 

 penetrating into the toe, is formed abom the new branch from 

 Cu,. 



Having now traced the development of the anal trachea 

 through all its stages, we are in a position to give the true homo- 

 logies of the much-discussed Italian loop. First of all, it is not 

 an anal loop in the sense that the ^Eschnine loop is, for it is 

 formed as much from Cu., and its derivatives as it is from A. Of 

 the branches of the latter, only Ao enters into its composition. 

 It is, strictly speaking, a cubito-aiial loop^ and should be desig- 

 nated as such, with the abbreviation Cual. The basal side of 

 Cual is in all cases Ao. Nearly always, it descends directly from 

 Ac. But, in some genera, it may lie a little distad from or 

 proximad to Ac. In all such cases, where larvae are obtainable, 



