lent, 



rrow 

 The 



532 ANGIOSPERMAEDICOTYLEDONES 



so noiseless, that I found it impossible to discriminate which flowers were first 

 visited by them. 



The flowers in the second stage, with curved corolla-tubes, assume a darker 

 tint in the course of the next few days, ultimately becoming a dirty orange-brown 

 colour. The tips of the petals roll up more and more, and the fragrance exhaled 

 in the evening gets less and less, but the position of the stamens and style remains 

 the same. A certain amount of nectar is still secreted, and the stigma continue 

 receptive for a time. Although the visits of hawk-moths become less frequent, 

 the possibility of cross-pollination remains for some days. 



Warnstorf states that inside the basal half of the corolla-tube, in line with the narro 

 under-lip, there is a longitudinal yellow ridge, indicated externally by a groove 

 surface of this ridge is beset with small sessile glands secreting abundant nectar, the little 

 drops of which collect in the base of the tube. An insect must possess a proboscis 

 at least 15 mm. long in order to reach the beginning of this store. The style with 

 its capitate stigma usually projects about 28 mm. from the corolla. Self-pollination 

 is rendered very difficult, if not quite impossible, in such flowers. There are, 

 however, others in which the stigma only projects about 1 mm. beyond the anthers, 

 and here it is obvious that autogamy is greatly facilitated should insect-visits 

 fail. The anthers dehisce in 30-40 minutes after the flower has opened. The 

 pollen-grains are white in colour, sticky and coherent, tetrahedral, rendered opaque 

 by numerous short spinose warts, and 88-100 fi in diameter. 



Kerner and Warnstorf give the same hour for the opening of the flowers as 

 myself; the former also states that the fragrance is strongest from 6 p.m. till 

 midnight. He further describes the subsequent curving of the corolla-tube, adding 

 that in this way direct contact between stigma and anthers, and therefore automatic 

 self-pollination, may be brought about. I have not myself seen this ; in the flowers in 

 the island of Fohr it is almost impossible, the stigma projects so far beyond the 

 anthers. My own observations in Helgoland prove that the mechanism is not 

 everywhere the same. Here the buds are quite horizontal; except that in free 

 inflorescences not crowded by others they are at first vertical, subsequently becoming 

 inclined. Anthers and stigmas mature simultaneously, and three of the stamens are 

 as long as the style, so that their pollen-covered anthers must touch the stigma, 

 automatic self-pollination being therefore inevitable. The two other stamens are 

 shorter by the length of an anther, and consequently serve for cross-pollination only. 

 The secretion of nectar is so copious that the corolla-tube is often half-filled, giving 

 access to Lepidoptera with a short proboscis. Even long-tongued humble-bees can 

 obtain pari, of the spoil, at the same time effecting cross-pollination. 



Hermann Miiller observed a reduction of the corolla-tube from 22-5 mm. in 

 length to 6 mm. in plants growing under ' unnatural conditions of life,' being 

 subjected to the drip from a roof, which apparently killed them. 



Visitors. In Fdhr, besides the above-named legitimate pollinators (Sphinx 

 convolvuli and S. ligustri), I observed other Sphingids (Macroglossa stellatarum Z., 

 Deilephila elpenor Z., Smerinthus ocellatus Z.), a Noctuid (Plusia), and po-dvg. hover- 

 flies (Syrphus, Eristalis, Rhingia, Syritta). I have also seen the humble-bee Bombus 

 hortorum Z. 5, skg. legitimately and effecting pollination. This insect cannot, 

 indeed, get all the nectar, but obtains a considerable part of it. On the island of 



