OP ARTS AND SCIENCES. 131 



habit. The genus Eriogijnia is separated from Spircea by its cespitose 

 habit, peculiar foliage, racemose inflorescence, and loose seed-coats. 

 It is described also as having united filaments, but this does not ap- 

 pear to be the case. The stamiuodia attributed to it are merely the 

 crenate lobings of the margin of the disk. It is therefore only in its 

 foliage and seed-coats that it differs from Petrophytum. The dehis- 

 cence of the coriaceous carpels by both sutures is the same in both. 

 No more satisfactory disposition of our present abnormal species 

 occurs to me than to transfer the section Petrophytum (excluding the 

 additional species referred to it by Maximowicz) to Eriogynia, and to 

 add E. unijlora as a section Kelseya. This leaves Spircea somewhat 

 more homogeneous, and brings together species that are similar in their 

 dwarf habit and not essentially unlike in other respects. The deter- 

 minate inflorescence of Kelseya occurs also in aS'. Ulmaria, etc., and in 

 S. {Chamcehatiarid) Millefolium. 



The genus Spircea has been recognized generally, and by botanists 

 of the highest authority, as a composite one, which it was better to 

 retain as made up of a number of well marked sections than to divide. 

 The one notable exception is Maximowicz, who has made a very 

 careful study of the whole group and whose conclusions are not to be 

 rejected unadvisedly. In addition to Physocarpus, as distinct from 

 Neillia * (to which it was referred by Bentham & Hooker), and 

 Eriogynia, which genera seem to me well founded, he has separated 

 several other genera, placing Aruncus and the Asiatic Sibircea in 

 his group Sjnrceece, — Chamcehatiaria {S. Millefolium), with the Old 

 World Sorbaria'f and Spirceanthus, among his Gilleniece (as stipu- 

 late and having the carpels opposite the sepals instead of alternate 

 with them), — Holodiscus (S. discolor and the very similar S. Ameri- 

 can aS'. argentea) among the Potentillece, — and Filipendula ( Ulmaria) 

 among the Sanguisorbece. The two latter genera are removed from 



* The Neillia capitata, Greene, can be in no way separated from the ordi- 

 nary P. opidi/oliufi. His N. malvacea also, judging from the characters, appears 

 to be a common form of P. Torreyi, though there are perhaps characters otlier 

 than those given by him upon which that species can be divided. 



t As respects the names adopted by Maximowicz, Sorbaria as the sectional 

 name of Seringe he considers as having precedence by right of priority over 

 Lindley's later generic name Schizonotiis. Botli names, however, are antedated 

 by Rafinesque's Basil ima (1815), and this makes it unnecessary in any case to 

 disturb the Schizonotus of Dr. Gray. Filipendula is rightly preferred to Ulmaria, 

 inasmuch as it was adopted by Linnaeus himself in several of the early editions 

 of the Genera, and the genus was definitely characterized by him as the equiv- 

 alent of both the Tournefortian genera. 



