148 



R. H. RICHENS 



habit, but it is probably only a single clone from among many other small-leaved elms 

 that make up a much wider entity, widespread in north Northamptonshire and elsewhere 

 in the northeast Midlands, and which may eventually be shown to intergrade continuously 

 with t\pical U. carpinifolia. 



The preceding discussion has been somewhat critical of the status of the microspecies 

 and varieties of Ulmus already described. Yet it is clear, in the case of the vegetatively- 

 reproducing elms at least, that taxonomic discrimination is possible between very minute 

 categories, and between individual clones. It is true that many of the samples can be 

 connected into a web in which numerous pairs of samples cannot be significantly separated, 

 but this must not be allowed to obscure the fact that most of the samples do differ signi- 

 ficantly fromi each other. The final picture that emerges is that of an aggregate of clones, 

 perhaps several hundred in East Anglia, whose interrelations can be expressed in an 

 n-dimensional network such as that in Fig. 3. 



Since the present treatment is perhaps unsympathetic to classical taxonomic pro- 

 cedures, it is pertinent to enquire what is to replace it. It is possible that further work 

 will provide a complete series of intermediates connecting even the major species U. 

 glabra, U. carpinifolia diT>d U. procera. Should then only a single species be recognised ? 

 It is not necessary to go so far. Should morphological discontinuities be absent, it is 

 still possible to utilise discontinuities of another sort. 



Thus, it can be seen in Fig. 3 that Ic (samples 87, 97 and 98) are only connected 

 with the other members of I by a single bond to sample 44. That is, connectivity between 

 Ic and the rest of I is lower compared with connectivity within Ic and v/ithin the rest 

 of I. Such a zone of low connectivity can be regarded as a type of taxonomic discontinuity 

 and used as a basis for discriminating between the two contiguous groups. 



Should even this type of discontinuity fail, it would still be possible to fall back 

 on relatively low frequency of individual types in the field as a basis for discrimination. 

 Thus, even w^ere multifarious connecting links discovered between U. carpinifolia and 

 17. procera, the greater frequency of elms representing the typical species as compared 

 with the intermediates, would probably, of itself, serve to justify the discrimination 

 between these species. 



The present evidence suggests that the three species U. glabra, U. carpinifolia and 

 U. procera can be legitimately distinguished on the basis of one or other of the types 

 of discontinuity discussed above. With regard to the subdivision of U. carpinifolia, it 

 seems best to eschew altogether the classical taxonomic categories. As an alternative, 

 subordinate systematic groups can be conveniently defined in terms of the range exhibited 

 in a number of selected characters. In the present paper, five quantitative characters 

 have been principally studied. If each of these is regarded as a dimension in a 5 -dimen- 

 sional anisotropic taxonomic space, then, by citing the lower and upper limit for each 

 characteristic, we define a 5 -dimensional parallelepiped, which can then be used as a 

 taxonomic category. This, in fact, is what has been done in Table 2. The highly dis- 

 tinctive assemblage of elms referred to as Ic can be denoted by Ulmus ' 44853, 45973 ' or 

 even by U. carpinifolia ' 44853, 45973.' This suffices to exclude all the other elms collected 

 in the area and seems to offer a better and more objective method of characterizing the 

 East Anglian elm population than the taxonomic methods used hitherto. 



Geographical Analysis 



The first point requiring mention is the distribution of the zones in which elms 

 were not collected, These include the fens, but exclude the " islands " such as Ely 

 where elms are abundant. This explains why no samples were obtained in minor squares 

 92 to 97. The chalk scarp, which runs diagonally across the area, is also devoid of elm, 



