DEVELOPMENT. 101 



petals and the stamens in alternating whorls of five ; probably this plant 

 should be worked over anew. In the mature flower the cyclic arrange- 

 ment is least evident in Castalia and but little more so in Lotos ; but, as 

 stated, in old flowers of Brachyceras, and open flowers of Hydrocallis it 

 seems to prevail in all of the parts (cf. Fig. 36). Probably cyclic arrange- 

 ment is always to be considered as derived from acyclic, and Nymphaea 

 seems to be in the transition stage. For such flowers the term hemicyclic 

 or spirocyclic (Engler, 1898) is best. It goes beyond the truth to place 

 Nymphaeaceae among acyclic orders as Goebel did (1887, p. 412). 



While still maintaining that the terms sepal and petal as commonly 

 used for waterlilies are physiologically correct, the morphological value of 

 the petals must be considered. We have shown that the anterior sepal 

 represents the bract, the lateral sepals the bracteoles, and the inner sepal 

 with the four outer petals the primitive sepals. For the last homology I 

 am indebted to a suggestion from my instructor, Dr. John W. Harsh- 

 berger. In development no distinction can be made between the rudi- 

 ments of the remaining petals and the stamens, while in N. alba the four 

 outer petals do have a distinct and simultaneous origin. Payer, therefore, 

 pointed out, and rightly so, that the inner petals are really modified 

 stamens, and the waterlilies are habitually double flowers in the same 

 sense as roses, almonds, etc., are doubled in cultivation. This at once 

 explains and is supported by the extremely variable number of petals and 

 the occurrence of transitional forms between petal and stamen. It seems 

 probable, however, that in last analysis all petals are modified stamens, 

 and in Lotos and Anecphya, where a naked space occurs between the 

 insertion of stamens and petals, and intermediate forms rarely if ever 

 occur, the distinction between the two classes of organs is being rapidly 

 drawn, and the homology is more distant than in Castalia. The fixation 

 of definitive petal and stamen characters upon the floral rudiments is 

 taking place before our eyes in the genus Nymphaea. It is striking, but 

 on second thought only natural, to find this kind of specialization going on 

 alongside of the change from spiral to cyclic symmetry. The two phe- 

 nomena are but manifestations of the same law of adaptation and differ- 

 entiation in form and function. 



After the stamen rudiment is established it grows considerably before 

 the archesporium appears. It has been claimed that the anther cells are 

 at first extrorse and are pushed around to the inner side by uneven 

 growth of the connective. In N. lotus, at least, this is not the case. Two 

 pairs of archesporial lines are differentiated in the substance of the anther, 



