CONFIGURATIONS OF BIVALENTS OF HYACINTHUS. 481 



But the four large bivalents of Hyacinthus show, like the 

 large bivalent of Uvularia, many connections not at the ends 

 (Fig. i). It has been pointed out in regard to Uvularia (Belling, 

 1926) that the simplest hypothesis is that these connections 

 (nodes) not at the ends represent places where two of the four 

 chromatids have undergone segmental interchange by fracture 

 and recombination. In Hyacinthus it can apparently sometimes 

 be seen with the microscope that two of the four chromatids are 

 bent back at a node, so as to continue along the same sides of the 

 bivalent (Fig. 3). It has also been shown that the homologous 

 chromosomes of the rings and V's formed by the large bivalent of 

 Uvularia, acted when separating as if they were not merely 

 twisted across one another, but had undergone a process which 

 produced some interlacing of chromatids at the nodes. This 

 would prevent the simple untwisting of the homologues at the 

 anaphase, and such untwisting has been shown not to occur in 

 Uvularia (Belling, 1926). This would also lead (as has been 

 abundantly shown by Janssens, 1924, and others, in animals) 

 to the separation of whole upper and lower halves of vertical 

 rings and V's, and to one chromatid passing up and one down 

 from both sides of horizontal rings or V's, which may get smaller 

 as the process advances, without opening up. This is what takes 

 place in Uvularia, and apparently also in Hyacinthus. 



In Hyacinthus, as already stated, the homologues are not always 

 connected at one or both of the extreme ends, but are connected 

 at other places (nodes). This is especially the case with the 

 four long chromosomes. Hence a study of these may show, by 

 the nature of their configurations and their mode of separation 

 at the reduction metaphase, whether the nodes correspond to 

 what would be expected if they were due to segmental inter- 

 change between chromosomes (crossing-over of genes). 



If the nodes in the long bivalents of the hyacinth are due to 

 segmental interchange, the following phenomena should be 

 observable: (i) the nodes should occur at different points in 

 the bivalents in different cases; (2) these nodes should be at 

 equal distances from the ends of both homologues; (3) the nodes 

 should be visible at the late prophase (diakinesis stage or earlier) 

 as well as at the metaphase; (4) the horizontal rings or V's 



