ARISTOLOCHIACEAE 353 



downwards from the entrance is distinctly smoother than the other portion which 

 leads up from the lowest part of the tube to the expanded terminal chamber ; for, 

 if both are equally smooth the flies will have as much difficulty in climbing up into 

 the stigmatic chamber as in climbing back out of the flower. In Arum I have 

 repeatedly noticed that the small midges try to escape from their prison, not by 

 creeping, but by flying towards the hght, and get knocked down by the grating at 

 the mouth of the flower. So if in Arislolochia Sipho the inside of the tube is all so 

 smooth that flies can climb neither one way or the other from the lowest part, the 

 reason of their imprisonment must be sought only in the curvature of the two ends 

 of the tube, the one rising straight from the lowest part to the cage, while the other 

 part which rises up towards the entrance to the flower is so bent outwards at its 

 upper end that the insects flying towards the light knock against the end and fall 

 back again. They are set free by the shrivelling up of the corolla, which enables 

 them to creep out.' ('Fertilisation,' p. 518.) 



Hermann Miiller's opinion is rendered much more probable by observation 

 I made upon Arisarum vulgare in Capri. This plant is distinguished from species 

 of the genus Arum in the same way that Aristolochia Sipho differs from A. Clematitis, 

 i. e. by the absence of weel-hairs. In Arisarum I saw numerous tiny gnats and flies 

 which had crept into the flower trying to escape, and constantly striking against the 

 window-like, transparent stripes of the spathe, in their eff'orts to do so. Only a few 

 exhausted ones, which crept slowly to the top of the spadix, were able to get out. 

 This barrier, which is partly dependent for success on the stupidity of the flies, is so 

 good, that the plant may be carried about for some time without the escape of 

 a single insect ; when the spathe is cut open, however, th<;y fly away immediately. 

 This takes place similarly in Aristolochia Sipho. 



Correns declares these hypotheses regarding the dete .ition of flies in the flowers 

 to be insufficient, and thinks that it is scarcely possible to form a decision in Europe, 

 but only by observation of the plant in its North Ameri(;an home. He also says that 

 the narrow part of the perianth tube possesses no proper weel-hairs, but closely 

 packed, downwardly directed papillae, which are perhaps connected with the insects' 

 lengthened stay in the trap. The latter falls in two parts, the glabrous ' antechamber,' 

 and the trap proper, the lower two-thirds of which are of a black-purple colour, and 

 lined with white hairs. The ' trap-hairs ' are mixed with other hook-like ' twining- 

 hairs ' ; these easily break to pieces from the apex downwards as the flower grows 

 old, splitting through the sexta either into single cells or pairs of cells. Correns 

 definitely proved secretion of nectar by fixing the flower upside down for a few 

 hours. 



W. Burck is of opinion (Ann. Jard. bot., Buitenzorg, viii, 1890) that flowers 

 belonging to species of Aristolochia are adapted for self-pollination ; but Correns 

 considers Burck's entire arguments against cross- pollination as partly actually mis- 

 taken, and partly insufficiently substantiated. 1 urck overlooked the fact that the 

 species examined by him in Java were of Amei ican origin, and that plants which 

 in their native countries are adapted for cross-pollination, in other regions frequently 

 become autogamous, autocarpous, or even cleisfiogamous in consequence of the lack 

 of pollinating insects. E. Ule (Die Natur, Halla^, xlvii, 1898, pp. 207, 210) thoroughly 

 examined some Brazilian species of Aristolochia (A. macroura, A. brasiliensis, 



DAVIS. Ill A 2. 



