8 MR. G. BENTHAM ON THE 



composed of five sepals, each with its midrib, the lateral dehis- 

 cence would naturally take place in the thinner spaces between 

 those nerves or midribs, as is clearly shown in Campanula, where 

 the thin part of the pericarp between the dissepiments coincides 

 with the thin part of the calyx-tube between the nerves or ribs. 

 In Downingia, however, the ovary, two-celled at an early stage, 

 becomes one-celled by the drying up of the exceedingly thin dis- 

 sepiment, and the two placentae remain attached to the inner 

 walls of the cavity, each one opposite the junction of two sepals. 

 "When the capsule at maturity has to split open between the se- 

 pals, four of them are held together in pairs by the adnate pla- 

 centae, and between these pairs there is on one side one slit, and 

 on the other side the intervening odd sepal induces two slits, 

 one on each side of it ; and we have as the natural result 

 three valves, two of them placentiferous and the third naked, 

 as described. This is the only instance among Lobeliese proper 

 of the lateral dehiscence, which is also exemplified, but apparently 

 in a single slit, in the Chilian Cypliocarpus among Cyphiese, a 

 genus otherwise allied in many respects to Dow?iingia, and in a few 

 genera of Campanuleae, although there rather in short valves than 

 in long slits, excepting, indeed, in Githopsis, an outlying member of 

 Campanulese, as Downingia is of Lobeliese, and connected with 

 the latter in some measure in this respect as in native country. 



In geographical distribution the Campanulaceae may be reckoned, 

 for the most part, amongst the herbaceous races of extratropical 

 origin, and in this respect analogous to a large portion of the 

 Compositae. There is nothing also to oppose the idea of their 

 early development having been contemporaneous with that of the 

 Compositae, although their subsequent progress has been so much 

 more limited. But I confess myself quite unable to see any 

 grounds for supposing, with Delpino, that Lobeliese are the 

 parents of Compositae. Moreover, if the two orders had a com- 

 mon parent, it must have been a very remote one with a long- 

 passed extinction of all the races which had formed the inter- 

 mediate stages. The intervening gap which now separates them 

 is too wide and deep. In the important point of the pistillary 

 structure no genus or species of the tribe Lobeliese or of the 

 whole order of Campanulaceae shows any approach to that which 

 is so uniform in Compositae ; nor indeed does any of the whole 

 group of Campanulaceous orders, unless it be some slight indica- 

 tion in one or two species of Goodenovieae, the furthest removed 



