AN ACCOUNT OF THE GENUS ARGKMONE. 133 



the varieties of P. orientale, which is the type of this latter section ; 

 occasionally it occurs in cultivated, more rarely in wild, examples 

 of several other species. Then the pccuharity of horned sepals is 

 characteristic of Fcqmver pavoninuvi, which is otherwise very nearly 

 related to P. Anjemone and P. hyhridwn. 



But the arrangement is moreover as unnatural as it is in- 

 convenient, for the position in question is based on a misdescription 

 of the stigmatic lobes. In place of having concrete stigmas, as in 

 the Papavers and the majority at all events of the Meconopses, the 

 lobes remain discrete, as they do in Chelidonium § Stylo/ihorum. 

 The lobes are moreover erect, and alternate with the placentas ; 

 the structures so often described as radiating stigmatic lobes 

 opposite the placentas are nothing more than horizontal pro- 

 longations of the sinuses between the lobes. But even this 

 distinction, which has been greatly used in most systematic 

 arrangements of the Papaveracea;, is of little real importance, for 

 we now find that it is necessary to include in JSleamopsis forms that 

 exhibit this very peculiarity. Taken as a whole, however, its 

 generic characters ally Argemone most closely with the genus 

 Romneya. This genus, supposed, when its name was employed by 

 Mr. Bentham to designate the tribe IlomneijctB, to be characterised 

 by having discrete ripe carpels, is now found to have a fruit that is 

 not distinguishable from the fruit of a Meconopsis or an Argemmie, 

 and thus clearly connects Platystujma — which, as Mr. Greene has 

 shown, does not deserve to be recognised as generically distinct 

 from Platysteiiwn — with Argemone. Arctomecon too, now that its 

 structure is accurately known, cannot be generically separated from 

 Romneya, in spite of the intruded placentas of the latter,* for the 

 differences between Romneya Coulteri and Arctomecon californicum\ 

 are not greater than those between Papaver somnifenim and P. midi- 

 caule, not nearly so great as those between Meconojms rohusta and M. 

 Henrici. And indeed the generic distinction between Romneya and 

 Argemone is, on close examination, found to be of the slightest; 

 there is only the partial separation of the tips of the styles in the 

 first, and the presence of horns on the sepals of the last, left to 

 differentiate them. We thus pass among the genera of trimerous 

 type from PLatystemon, where the ripe carpels are discrete, or nearly 

 so, through Romneya, with its united ripe carpels, but partially 



* This difference, at first sight considerable, is rendered insignificant by 

 the occurrence of two Meconopses in China, one {M. chelidonifolia) with an 

 ovate fruit and deeply intruded Pajmver-like placentas, the other {M. OUveriana) 

 with a narrow cylindric fruit and nerviform ones ; these two species are in every 

 other respect so alike as to be indistinguishable. 



t A careful examination and analysis of A. californicum and A. humile leads 

 me to fear that Mr. Coville's separation of these as species, though certainly 

 convenient from the local point of view, cannot be sustained when the order is 

 examined from the monographer's standpoint. This applies with even greater 

 force to A. Merriavii, which, however, I only know from ]Mr. Coville's drawing 

 and his excellent description. The polymorphism in this alpine species is by 

 no means so excessive as that displayed in the corresponding Papaver (P nudi- 

 caule) ; even the tendency to 2-mery shown in A. hwidle is paralleled by the 

 tendency to 3-mery shown in P. nudicaule proper. 



