THE CONTINENTAL FLORA OF SOUTH SWEDEN 355 



fact that he would explain by the scanty supply of water. However, if the 

 species not suffering from drought has formed fruit, the formation of seeds is 

 not good; he says: »Meist sind aber ihre Samen schr dimn und scheinen mir 

 nicht keimfahig» (pp. 87 ff.). 



Another circumstance which should be pointed out is that the seeds, as a 

 rule, seem to ripen so late in the autumn that they may be injured by frost. 



These investigations into the formation of fruit by Cynanchum thus seem to 

 show that any continuance of the inland dispersal of the species from its occur- 

 rences in rocky cliffs is impossible, or at least extremely insignificant. 



But in that case the question immediately presents itself: how has Cynanchum 

 been able to attain its present occurrences furthest in the interior of the country? 

 This brings us to a very remarkable peculiarity in the distribution in the south- 

 east of Sweden, which is common to several species. 



With regard to certain species mentioned above, emphasis has been laid on 

 the abundant distribution in the skerries and in certain undulating districts on 

 the mainland lying immediately within those skerries. These mainland areas all 

 fall considerably below the highest marine limit, and thus have not become part 

 of the mainland until a comparatively late stage in the post-glacial upheaval of 

 the Swedish land. Their broken terrain shows that during certain phases in the 

 upheaval of the land they formed skerries of about the same kind as the present 

 ones. Consequently it is very natural to assume that the species in question 

 were abundantly distributed in those skerries also. When the skerries became a 

 continuous mainland, over which woods spread their shady covering, the species 

 lost a large number of the occurrences they had on the shore slopes, but they 

 were able to maintain their position on escarpments and cliffs and on the shores 

 of inland lakes in those regions. 



As regards Cynanchum, of course, it might be conceived that there might be 

 a spreading outwards^ from low shore-slopes fairly well protected from the wind 

 to similar spots, as they from time to time emerged from the sea; while a 

 spreading inland to suitable rock)^ escarpments could in any case take place on 

 a smaller scale, and that a continued spreading in this last- named direction, from 

 escarpment to escarpment, would be impossible in case the complicated pollina- 

 tion biology of the species rendered impossible the formation of fruit in such 

 localities. 



As regards Cynanchum vincetoxicum, the occurrences that lie furthest inland 

 and highest above sea-level would in that case be the oldest habitats of the 

 species in these regions. If this is really the case, the internal distribution-limit 

 ought to coincide approximately with some former coast-line. As is shown by 

 Plate 9, this is the case in Ostergotland and Smaland. Apart from some minor 

 divergences, which mark valleys penetrating furthest into the country, the occur- 



