BY R. J. TILLYARD. 325 



ment of any Libellulid connectioa — yet possesses this remarkable 

 nymphal lal)ium of distinct Libellulid form, and distinctly non- 

 uEschnid in every particular. We now find that Syuthemis, 

 Aviiich in its imaginal development is absolutely Cordnline, with- 

 out a single J^schnid characteristic, possesses a larva exceedingly 

 similar to that of C or duleg aster. The gap, therefore, between the 

 two large families is in this way completely bridged. If we place 

 the genus Syntheniis at the end of the Libelhdidce, in the sub- 

 family Cordtdiince, and begin the ^Eschnidce with C or dtdeg aster, 

 we have a very good linear connection between the two. Further* 

 if we see in Synthemis and Cordulegaster two of the most archaic 

 survivals in the whole of the Anisoptera, it is to be expected that 

 such forms would shew small divergence compared with the 

 larger and more specialised groups. 



The Larval Gizzard. — Dr. Ris has pointed out to me the great 

 importance of the form of the larval gizzard as a character for 

 the systematisation of large groups. In dragonfly nymphs the 

 gizzard is a well-developed organ, lying within the fii'st or second 

 segment of the abdomen, and furnished on its inside v/ith a set 

 of four or more chitinoiis plates or *' folds," carrying au armature 

 of strong hard teeth for the reduction of the food-masses. In the 

 Anisopterid families two main lines of development are found, 

 viz., (1) the Libellulid form, which possesses four plates arranged 

 in bilateral symmetry, i.e., one pair of one form, and one pair of 

 another; and (2) the ^schtiid form, in which the four plates are 

 radially symmetrical, i.e., all alike in form. Dr. Ris found an 

 almost exactly similar pattern in all the Libelhdid nymphs he 

 examined, both of the Libellidince and Cordidiince. But in the 

 uLschnidce, where a fair number of archaic forms have been 

 placed, the radially symmetrical development differs considerably 

 in the various subfamilies. 



I have carefully examined the gizzard of Synthemis eustalacta, 

 of which a diagram is given below, in Text-fig. 2. The diagram was 

 prepared by cutting the gizzard open longitudinally, spreading it 

 out carefully on a glass slide, and placing another lightly above 

 it; so that the teeth are not ab.solutely flattened, but are viewed. 



