T etracyclidce. 2 21 
the Geraniales-Sapindales plexus, and by no other. Examples of 
isomery of the androecium and alternation of stamens with petals 
are to be found in all the larger natural orders of Geraniales and 
Sapindales—in the latter, the more advanced cohort, more frequently 
than in the former. A bicarpellary ovary is not of such common 
occurrence ; but examples are forthcoming in most of the larger 
orders, while in the D section of Geraniales, including the Poly- 
galaceae, the pistil is nearly always composed of two carpels. 
Unfortunately, however, the Contortse have left no traces of 
their progress from polypetaly to sympetaly in the shape of penta- 
cyclic forms ; neither a second staminal whorl nor any hint of it 
ever occurs. Nor can they be regarded as the direct descendants in 
any sense of a group so highly evolved as the Ericales, although 
these originated most probably from the same stock—the geranial. 
The Pentacyclidae, we may premise, seem to have left no descendants 
among the Tetracyclidae, although the latter have almost un¬ 
doubtedly been derived from the same archichlamydeous stocks. 
The immediate antecedents of the Contortse may have been poly- 
petalous forms in which the economy principle had reached its 
limiting expression in an isomerous alternating androecium and a 
bicarpellary gynsecium ; or they may have been pentacyclic and 
sympetalous, the tendency to economy being active, in association 
with a corolla-tube, until the typical Contortse were evolved. In 
either case the intermediate forms have been extinguished. 
We regard the Contortse, then, as derived from the geranial 
stock, like the Ericales, on some such lines as indicated in the 
accompanying diagram; and we may suggest that while the latter 
have employed their evolutionary powers, so to express it, in the 
direction of specialization of habit and details of floral structure, 
the Contortse have reserved their efforts for the realization of the 
economy tendency. This refers, of course, to the passage from 
polypetaly to sympetaly ; we have now to consider the question of 
evolution and affinities within the group itself. 
We need first to search for those members of the cohort which 
are nearest to the archichlamydeous stock—in other words, we 
must endeavour to find those members which are relatively primitive; 
and it is at this stage that we feel with particular keenness the 
disadvantage of the evolutionary break to which we have drawn 
attention above. A pleiomerous androecium we cannot find; a 
gynaecium with more than two carpels occurs so rarely that very 
little significance must be attached to it, unless this can be associated 
