













. 



























. 









REVISION OF THE BRITISH WILLOWS. 451 



flowers, some of them would with difficulty be discriminated from 

 cinerea otherwise than as slight modifications. 



As belonging to Salix sordida I put all plants which, however 

 like cinerea they may be, have the filaments of the stamens more 

 or less united to each other. Connate filaments may, according 

 to Wimmer, be distinguished from cladostemmic ones by the 

 branches forming an acute and not an obtuse angle; and one of 

 the best characters of hybridization with 8. purpurea is the pre- 

 sence of connate stamens. 



In these Woody Island plants an extreme degree of connation 

 is rare. Frequently the filaments are united for a little way above 

 the base only ; but in the same catkin free, slightly united, and 

 more distinctly connate stamens may all be found— a phenomenon 

 which may be seen in S. rubra and other hybrids of 8. purpurea. 



Other points in which these cinerea-\ike plants are variable 

 are the shape, colour, and pubescence of the leaves, stoutness and 

 pubescence of the twigs*, shape of the catkins, and colour of the 

 anthers. In all these particulars they approach, or recede from, 

 8. cinerea by almost imperceptible gradations. Probably, as 

 Wimmer suggests, the cinerascens modifications of 8. sordida 

 have a somewhat different origin from those which are nearer 

 8. purpurea, having been possibly produced by purpurea ? X 

 cinerea tf, or, what seems more likely, by the crossing of sordida 

 with cinerea. 



The $ of the cinerea-like forms is a much more difficult plant 

 to distinguish than the tf , since of course we have not the assist- 

 ance afforded by the connate filaments. I have indeed found 

 some plants whose catkins are so like those of cinerea, that it is 

 on ty by the resemblance of their leaves and habit to some 6 

 Plants of sordida that they can be referred to that species. 



In addition to the Woody Island plants, I have seen specimens 

 from a rf bush found at Dalmarnock, on the Tay above Dunkeld, 



M 



*he Tay below Perth. 



Woody 



un der 8. sordida (as var. rubella), though I am inclined to sus- 

 pect that it may be a cross of S. rubra (with which it grows) 

 an <l 8. cinerea (i. e. 8. purpurea X S. viminalis X S. cinerea). In 

 Jj» leaves and habit is is near S. purpurea, but has free stamens. 

 1116 very beautiful catkins are small or moderate in size ; the 

 ^opened anthers orange-red, but when burst yellow from the 









• . 









