1898-1902. No. 2.] VASCULAR PLANTS OF ELLESMERELAND. 91 



''pili furcato-stellati" also covers the short pedicels, here more inter- 

 spersed with simple hairs. The inflorescence is, during the flowering 

 season, very condensed, forming almost a head, hence the specific name. 

 Later on the scape is somewhat stretched, but often the pods also sit 

 densely clustered. There are specimens, however, especially from Spits- 

 bergen, which have the pods further apart, and sometimes the lowest is 

 removed from the others. The flowers are generally 2 — 5, rarely more, 

 in the raceme. 



The flower is very small with narrow, linear sepals, that are more 

 or less hairy, with long, forked, or generally simple, hairs. The petals 

 are pure white, spathulate, rounded, or more or less emarginate. They 

 are so narrow as not to touch each other with their margins. The 

 flower is so different from that of all other species, as to make D. sub- 

 capitata immediately distinguishable when living and in a flowering 

 state. Also when the pod has begun to develop, the sepals and petals 

 will still remain for a time. 



The pod is at first broad-lanceolate, with sparse short hairs, but 

 later on it becomes more ovate or almost circular, rather thick, and 

 quite shiny glabrous, purplish-brown. 



In habit, D. subcapitata resembles small stunted forms of D. alpina 

 as well as of D. fladnizensls and D. hirta, and it may sometimes be 

 difficult enough to distinguish herbarium specimens. This is why I can- 

 not always assert that the specimens from other countries, I have seen, 

 really belong to it notwithstanding a great resemblance. Especially small, 

 glabrous, D. alpina specimens in fruit, are very difficult to separate 

 from it. 



D. subcapitata was rather common in clay or gravel fields with 

 sparse vegetation along the Southwestern fjords, flowering about the 

 end of June, and developing its pods in a very short time. 



Occurrence. South Coast: Harbour Fjord, at the Western en- 

 trance (2437); Muskox Fjord, inner part (2118, 2140); Goose Fjord, Gull 

 Cove (2896, 3821), Falcon Cliff (2872), Castle Point (3960), Yellow Hill 

 (3591, 4212), East of 3rd quarters (3187, 3431, 3482), Ptarmigan Gorge, 

 Gallows Point (2991), valley at the bottom. West coast: Lands End 

 (2850), Braskerud Plain (709, leg. Isachsen). The latter is doubtful, as 

 are also some fragments from the Hayes Sound region that may belong 

 to it: Cape Rutherford (322), Bedford Pim Island (4191). I did not ob- 

 serve it the first summer, but the above fragmentary fruiting specimens 

 seem to represent it. Further it is not improbable, that the D. rupestris 

 var. parviflora, Oliver, mentioned by Hart, Bot. Br. Pol. Exp., p. 25, 



