DEGENERACY OF FLOWERS. 279 



conditions, either many carpels having one or few orales 

 in each, or a few carpels with many seeds, as in the 

 RanunculaceoB. This primitive condition rapidly vanishes in 

 passing to CalyciflorcB and Ganwpetaloe. 



(2) Having reduced the nuraher of carpels to a definite 

 quantity, five, i.e. one cycle of the prevailing f type, this 

 number remains tolerably persistent, but does not show a 

 large percentage. 



(3) The combination of five carpels with a reduced number 

 of ovules, i.e. one or two in each cell, or 5-10 ovules in all, 

 is pretty uniform for the first two divisions, but almost 

 disappears under Gamopetalm, the orders Sapntaceae, Nolanece, 

 and one or two Euhiacece, (e.g. Erithalis) representing this 

 condition. 



(4) and (5). Here we see a steady increase in the 

 percentages in passing from Thalamifloree to Gamopetalce, in 

 which the number of carpels is still further reduced ; but 

 the number of ovules runs in two directions, being either 

 numerous or few. 



Two questions arise at this point. If one result of insect 

 agency is to bring about increased specialization in flowers 

 (yet, in proportion as they become specialized, so, inversely, 

 is the number and variety of insect visitors diminished), 

 how is it that some (e.g. Foxglove and Orchids) produce 

 an enormous number of seeds ; while others (e.g. Labiata, 

 Coinpositce, etc.) produce few or only one in each flower ? 

 The second question is whether a plant is better ofE for 

 having so many more seeds than another. Recognizing 

 reproduction as the sole end of plant life, so that a plant 

 should bear as many good seeds as possible, it is noticeable 

 that the two largest orders, Compositoe and Gramineas have 

 never more than one seed to each flower. Again, comparing 

 Labiaiae with Scrophularinece, according to the Genera 



