MB. A. W. BENNETT ON PAKNASSIA PALUSTKIS. 25 



fcipecies of that geuus as passing by every grade into a true attach- 

 ment between the stamens and the calyx, and the same variability 

 occurs also among Violacese. 



On the difference in the structure of the seeds I do not Jay so 

 much stress, as, if their exalbuminous character is to be taken as 

 an essential point, Parnassia must either be referred back to 



Hypericacese, with which it has no other affinities, or be hopelessly 

 consigned to the solitary confinement of a separate order. • The 

 extrorse stamens are, however, connected with an important phy- 

 siological function presently to be described. In his ^ Genera of 

 'iXorth-American Plants,' Prof. Asa Gray describes the anthers of 

 Farnassia as introrse, and gives a drawing of P. Caroliniana as an 

 illustration. I do not, however, find any other observer to agree 

 with Prof. Gray's observation in this respect, except two Ame- 

 rican botanists, Dr. Torrey and Mr. Chapman, who have pro- 

 bably borrowed their descriptions from him ; nor do any speci- 

 mens which I have been able to examine of this species confirm 

 any departure in this resj^ect from the ordinary type of the 

 genus,' 



Before pointing out what seem to me the affinities between 

 Farnassta and some tropical genera with which it has not been 

 generally associated, a few remarks may not be out of place on 



b ■ 



the physiological structure of our British species. The true 

 morphological value of the remarkable glandular petaloid scales 

 of Parnassia has been a subject of much discussion. The advo- 

 cates of its affinity with Hypericum of course consider these scales 

 to be modified polyadelphous stamens united together at the 

 base. The fact, however, that notwithstanding the countless 

 number of specimens examined by some German botanists, I can 

 find no record of a single flower having ever been gathered in 

 which the glands have reverted into pollen-bearing anthers, seems 

 to me a strong argument against this hypothesis. In certain 

 Himalayan species these scales seem entirely to lose their stami- 

 noid appearance, and to be simply bifid or trifid at the apex, or 

 even almost entire. I am rather disposed, on the other hand, to 

 consider them to be a modified inner row of petals, the glands 

 having an unmistakeable function, as we shall presently see, 

 connected with the disti*ibution of the pollen. Dr. Buchenau 

 (Botanische Zeitung, vol. xx. p. 307} goes so far as to view the 

 glands as metamorphosed carpels !, having found a specimen in 

 which they are rolled up in a carpellary fashion. The 4 stigmata 



