2i6 ANGIOSPERMAEDICOTYLEDONES 



is vertical, 14-15 ram. long, and closely apposed to the somewhat longer (16 mm.) 

 upper lip so as to close the throat of the corolla. It cannot be entirely depressed. 

 Visitors must obviously be large strong insects, such as humble-bees or hawk-moths, 

 which, much as in Antirrhinum and Linaria, are obliged to force their body or head 

 between the lips before they can touch the stigma. As shown in B, the stigma 

 springs forward so far when the lower lip is removed that contact with it must be 

 inevitable. That large insects effect crossing also seems to be indicated by the 

 distance of 2^ cm. between stigma and nectary. I have also several times seen 

 stamens torn out of the upper lip, evidently by the exercise of force on the part of 

 a large insect. One of the flowers figured (C) was protogynous, for the anthers 

 were still undehisced while the stigma {H) appeared to be mature. On dehiscence 

 subsequently taking place {D) a visitor would obviously strike the filaments and 

 sprinkle itself with pollen. It is noteworthy that the filaments, anthers, style, and 

 reflexed edges of the upper lip are all alike glabrous, there being no special 

 arrangement to afford resistance and thus increase the efficiency of the sprinkling- 

 apparatus. But there are hairs on the edges of the superior part of the upper lip, 

 which possibly prevent the pollen from falling on one side, as do the hairs on the 

 anthers of some other species. 



There are also variations in the breadth of the slit in the upper lip, and the 

 amount of protrusion of the stigma. The latter may {B) project (i-i^ mm.) or not 

 (as in C, or even less). The differences as .to the breadth of the opening in the 

 upper lip and the shape of this lip are illustrated by B and C, Fig. 312. The 

 explanation of these variations I must leave to those who have the opportunity of 

 studying fresh material. 



Self-pollination appears to be effected only with difficulty. As even so far north 

 as Alten (70 N. lat.) fruits are set abundantly, Warming is of opinion that crossing 

 must be effected by certain humble-bees.' 



2166. P. lanata Cham. (Warming, ' Bestovningsmaade,' p. 47, Bot. Tids., 

 Kjobenhavn, xvii, 1890, pp. 214-15.) Warming describes the flower mechanism of 

 this northern species as agreeing with that of P. hirsuta. The automatic self- 

 pollination, which inevitably takes place, is effective in Spitzbergen. And as there 

 are no humble-bees, which alone are strong and clever enough to suck nectar 

 legitimately, what has been said about P. hirsuta applies in a still higher degree to 

 this species. 



Andersson and Hesselman (' Bidrag till Kanned. om Spetsbergenso. Beeren Eil. 

 Karlvaxtflora,' p. 116) describe the corolla in the var. dasyantha Trautw. as being 

 rose-red in colour with a darker tube, and projecting about 12-13 mm. beyond the 

 calyx of 8-9 mm. Ekstam says that the flowers possess a pleasant odour resembling 

 that of almond blossoms, though this is sometimes absent (' Bliitenbiol. Beob. a. 

 Spitzbergen,' p. 7). The end of the style does not project from the corolla, but is 

 spirally coiled, while the anthers closely surround the stigma and dehisce on the side 

 turned towards it so that autogamy is inevitable. Flowering begins in Spitzbergen 

 towards the end of June and lasts till August, and fruits are quickly ripened at the 

 end of August or the beginning of September. 



