INTRODUCTION TO THE METHOD 

 migration into their present habitats in terms of the Quaternary Ice Age (Manton, 



1934, 1937)- 



The case of watercress {Nasturtium) is sHghtly more intricate because the chromo- 

 somes are more numerous and smaller, and there are also a greater number of types of 

 plant to consider. A selection only of the relevant photographs is given in Fig. 3 a-/, 

 the magnification being the same as for Biscutella, but, to facilitate the interpretation in 

 view of their smaller size, some explanatory diagrams are added (Fig. /\.a-d). The 

 monoploid chromosome number (which is also the 'haploid' or gametic number of 

 the lowest form) is here 16, and the polyploid series again consists of diploids, triploids 

 and tetraploids which, in this case, possess chromosome numbers of 32, 48 and 64 

 respectively. Sample views of the somatic chromosomes showing the diploid and tetra- 

 ploid chromosome numbers in the unpaired state are contained in Fig. 3 a and b, 

 and the only further point of importance to add about the origin of the series is that 

 in this case all three members of it are wild plants widespread in Europe, although the 

 triploid, as in Biscutella, is the hybrid between diploid and tetraploid which, in the 

 watercress, has occurred spontaneously. 



Chromosome pairing at meiosis in wild plants of diploid, triploid and tetraploid 

 watercress is shown in Fig. ^c-e, with an artificially produced autotetraploid (Fig. 3/), 

 obtained from the diploid by treatment with colchicine, added for comparison. In the 

 flw^otetraploid (Fig. 3/), chromosome pairing closely resembles that in tetraploid Biscu- 

 tella, allowance being made for the smaller size and greater total number of the chromo- 

 somes; numerous quadrivalents are formed. In the wild watercress polyploids, on the 

 other hand, multivalent groups are completely absent. The tetraploid (Figs. 3^, 4c) 

 forms 32 pairs and the triploid (Figs. 3^, ^.b) invariably develops 16 pairs and 16 uni- 

 valents, whether the plant studied be the wild triploid or an artificially synthesized 

 hybrid between the wild tetraploid and the diploid. This means that polyploidy in the 

 wild watercress cannot be simply due to the multiplication of identical sets of chromo- 

 somes. There must be two different sorts of monoploid sets contained in the polyploids, 

 one of which is identical with that of the low-numbered Nasturtium officinale R.Br., 

 and which can pair readily with the chromosomes of that species when hybrids are 

 formed (as in the triploid) but the other of which is not homologous and which com- 

 pletely fails to pair in the triploid and does not form quadrivalents in the wild tetraploid. 

 The origin of this second set of 16 chromosomes is still unknown, though, from the 

 morphology of the fruits in the wild tetraploid, it is suspected to be a species oiCardamine. 



The watercress series in the wild state is, therefore, not an autopolyploid series as in 

 Biscutella but an a//opolyploid one, and the wild tetraploid must be recognized as 

 in origin an interspecific, or in this case probably an intergeneric, hybrid since the 

 genus Nasturtium contains no other known species. This hybrid, at some former period, 

 doubled its chromosomes and became fertile and stable, as in the case of Spartina 

 Townsendii. It is therefore desirable in the case of watercress to separate the wild 

 tetraploid taxonomically from the diploid and to relegate it to a separate species to 

 which the name Nasturtium uniseriatum has been given as a descriptive title to record 

 the most distinctive morphological difference by which the tetraploid can be recognized 

 in the field without a chromosome count, namely, the arrangement of seeds in the 



