STUDIES ON ULMUS : I 



147 



seed is produced in the former species in very favourable seasons (cf. Henry, 1910). Such 

 vegetative propagation tends to increase and conserve variability (cf. MuUer, 1951), and 

 heterozygous types, if well-adapted to their environment, may well equal or even exceed 

 in abundance the relatively homozygous parent types. The frequency distinction, there- 

 fore,- that normally subsists between species and their hybrids is liable to disappear. 



Within Ulmus, no fundamental sterility barriers appear to exist between the English 

 species, and hybrids are readily produced (cf. Doorenbos, 1938). The chromosome 

 number of all the English species, is the same, as far as is known, although it is possible 

 that triploid or tetraploid clones may occur, as they do in U. turkestanica Reg. (cf. Krijthe, 

 1939). The subdivision of the genus into species must therefore rest entirely on the 

 discovery of discontinuities, and the data here reported support, on the whole, sub- 

 division of the English elms into the sexually-reproducing species U. glabra (VI), and the 

 two vegetatively reproducing species U. procera (II) and U. carpinifolia (I, III and IV). 

 There are, however, forms intermediate between these species and some of these may 

 well be hybrids, for example VII, intermediate between U. glabra and U. carpinifolia, but 

 it would be rash to presume too far on this point without further evidence as to the range 

 of variation of each species when not in proximity to the other. Thus, scabrosity in 

 U. carpinifolia, as in 64, suggests introgression from U. glabra. But it has to be remem- 

 bered that the juvenile leaves in U. carpinifolia are scabrous, and it would probably 

 require only a slight physiological change for this character to persist to the adult con- 

 dition. 



The distinction between U. procera and U. carpinifolia is also blurred. The smooth- 

 leaved elms in Ila may well be hybrids, and 24a in this subgroup occurs in the region 

 of overlap of these two species, but the full range of variation of U. carpinifolia is not 

 yet known, and it is possible that the smooth-leaved members of Ila, and even more 

 likely, lib, are only extreme forms of U. carpinifolia. 



Further subdivision of U. carpinifolia is obviously possible. Melville (1939, 1946, 

 1949) partitioned it into the three microspecies U. carpinifolia sens str., U. diversifolia and 

 U. coritana, excluding for the moment U. plotii and U. stricta. The data presented here, 

 however, suggest relationships which cut across and transcend Melville's groups. Thus 

 the large group I includes U. diversifolia, forms with a somewhat remote relationship to U. 

 carpinifolia sens. str. and U. coritana var. media, and forms receding considerably from 

 any of these. Typical U. carpinifolia sens. str. is found in III, but samples with the 

 characteristic leaf shape of this species appear also in I, IV and VII. U. coritana seems 

 a rather artificial aggregate. U. coritana var. angustifolia is nearest to VIII ; it is very 

 remote from var. media, in II, and var. rotundifolia in V. Thus the groups derived by 

 the present analysis include, in some cases, several of Melville's microspecies; in other 

 cases, one microspecies has to be partitioned into several distinct groups. A more con- 

 cordant return would be obtained, however, if Melville's three varieties of U. coritana 

 are treated as independent microspecies with an equivalent status to the other two. 



It is true that the type descriptions of Melville's microspecies relate to highly 

 characteristic and well-defined entities, but the analysis given here raises considerable 

 doubt as to the status of these types as foci of the populations into which the genus can 

 most naturally be divided. It is indeed probable that further work will reveal more inter- 

 mediate types and intergroup connections than those here described ; this would render the 

 status of the microspecies even more uncertain. The conclusion would seem to follow 

 that the application of Latin names to segregates of U. carpinifolia is of doubtful utility. 



The data given here are not adequate to discuss the status of U. plotii or U. stricta, 

 but the indications are that these species might turn out to be nothing but extreme 

 variants of U. carpinifolia. Typical U. plotii is a very distinct elm with a characteristic 



