168 ME. R. J. TILL YARD ON THE RECTAL BREATHING-APPARATUS 



Lamellate Type into sub-types, we might propose the term Arclnlamellate 

 for the sub-type exhibited by the Synthemini, as contrasted with the Neo- 

 lamellate sub-type found in the other genera. Within the Neolamellate sub- 

 type we might then recognize two forms of lamellse, viz., the Corduline form^ 

 tbe more asymmetrical, and the Lihelluline form, the more symmetrical. 



Number of Capillaries in the Gill-basket. — It will be sufficient to indicate 

 the approximate number in the five genera above studied. In Synthemis the 

 capillaries are only of moderate fineness, and may be reckoned at only 100 to 

 each lamella. Hence we get a total of 12 X 12 x 100, or only 14,400 capil- 

 lary loops in this gill-basket. In Austrocordulia, there are probably 150 or 

 more capillaries in each lamella. Hence the total for this genus will be 

 28 X 12 X 150, or 50,400, allowing 28 as the average number of lamellae in a 

 hemibranch. In Hemicordulia the greater fineness of the capillaries may be 

 represented by reckoning 200 in each lamella. Thus the total for the genus 

 is 25 X 12 X 200, or 60,000. In Cordulephya the narrowness of the lamellse 

 allows us only to reckon 150 capillaries in each, in spite of their fineness. 

 Counfcino-, then, 23 lamellae to the hemibranch on the average, we have a total 

 of 23 X 12 X 150, or 41,400. In Orthetrmn, again, we can reckon on at 

 least 200 capillaries in each lamella, with an average of 25 lamellae to the 

 hemibranch. This gives us 25 X 12 X 200, or 60,000, the same as in Hemi- 

 cordidia. The greatest total is given by Diplacodes with an average of 

 28 lamella? to the hemibranch, yielding a total of 28 x 12 x 200, or 67,200 

 capillary loops. This number, though it falls short of that calculated for the 

 Foliate Type, belongs, it must be remembered, to a much smaller larva, 

 besides which, the arrangement of the gills in the Lamellate Type is such as 

 to give unrivalled advantage in the extraction of oxygen. We need not be 

 surprised therefore at the fact that the larvae of Diplacodes, Cordulephya, and 

 other Libellulid genera develop very rapidly, and pass into the imaginal 

 state well within a year from the time of hatching. What we may well 

 wonder at, however, is the fact that the larva of Austrocordulia, in spite of 

 its highly developed gill-basket, takes four or five years to mature. This 

 points, indeed, to some special disability not connected with the oxygen- 

 supply. That disability is, I think, the sluggish habit of the larva, which 

 hides away under rocks or debris, and feeds but seldom. 



Comparison of the Efficiency of the Gill-basket in the various Types. 

 The numbers given as the totals for the capillary loops are in all cases 

 only approximate, and are intended to give an idea of the comparative 

 numerical abundance of capillaries rather than their actual numbers. If we 

 look upon each capillary as a unit for the extraction of oxygen, they enable 

 us to compare the relative values of the diff'erent types of gill-basket as 

 oxygen-extractors. It must not be forgotten, however, that the value of 

 the unit also depends upon the length of it exposed to the direct action 



