188 PROCEEDINGS OF THE AMERICAN ACADEMY 



deed, and long-clawed petals ; but, with its almost terete siliques, 

 having an obscure stipe, and oblong seeds, it best accords with lodan- 

 thus, Torr. & Gray, — which genus, again, is not well, if at all, to be 

 distinguished from Pachypodium, Nutt. {Thehjpodium, Endl.), and this 

 name, though badly chosen, may have to be adopted. It would be 

 better if we could settle upon the name of Thelypodium. 



No. 684 (or 689 ?) of Coulter's Mexican collection, from Ziraapan, 

 appears to be a strict congener of lodantlms pinnatijidus. 



A Revision and Arrange^nent (mainly hy the fruif) of the 

 North American Species of Astragalus and Oxytropis. 

 By Asa Gray. 



In view of the species known to him by the fruit, it was natural 

 enough that Linnreus should distinguish from Astragalus his genus 

 Phaca. Being established, and augmented with a considerable number 

 of species, it is not surprising that the two genera should still have 

 been maintained long after the neat carpological character which alone 

 distinguished them was found in some cases to fail. Perhaps there are 

 not very many large genera in botany which do not at some point grad- 

 uate into some other. But in the present case, — not to refer to the 

 eminent unsafeness of all leguminous genera founded on the legume 

 alone, and to the inutility of genera which are not recognizable by 

 habit or floral structure, — it has now become evident that the distinc- 

 tion between Phaca and Astragalus breaks down so completely and 

 so variously, that the two genera are, as I suppose, no longer tenable. 

 Indeed, from analogous instances we might expect that the intrusion 

 into the cell of one or the other suture, even when, as here, quite con- 

 stant in species, would not of itself be of generic consequence. An 

 obvious alternative to combining these two Linnfean genera is to pro- 

 ceed further in division, by taking the form and texture of the legume 

 into generic account. The manifold diversity which the fruit exhibits, 

 and also the vastness of the group, would invite to this course. But a 

 study even of the North American species only — especially as here 

 grouped mainly in reference to the fruit — demonstrates its impracti- 

 cability. In place of two genera with outlines here and there blended, 

 we should have twenty, most of them still less definable. A partial 

 attempt of this kind, probably one of the best that could be made, was 

 that of Nuttall, in proposing his two genera, Homalohus and Kentro- 



