In a recent number of these PROCEEDINGS, Emerson' suggests that certain breeding results are incompatible with my view of the genetic structure of complex in Oenothera.2 He points out, first, that crossing-over between genes is as great in the multiple rings as in the simple ring pairs and, secondly, that this crossing-over is not accompanied by interchange of non-homologous segments and consequent change of configuration in the progeny.* My view, however, is that, while certain genes, which are sometimes associated with complexes, cross over freely without interchange, the essential and permanent differences between complexes must depend upon genes which are normally incapable of crossing-over, owing to their being differently arranged in the chromosomes following structural change, and owing to their being situated next to the spindle attachment. Thus Emerson believes that the differences between complexes can be resolved into differences between independently segregating genes while I believe that they are inherently incapable of such resolution because they are associated with structural changes such as translocation, deficiency and reduplication. Since I put forward this suggestion new'evidence bearing on the question has appeared, particularly from observation of interchange heterozygotes in maize4 and Pisum.5 In these it is found that interchange has occurred in the arms of the chromosomes at some distance from the spindle attachment. It follows in such cases, as E. R. Sansome has pointed out, that a ring of six can (and indeed must) arise from two interchanges in such a way as to give interstitial homologous X-segments in two chromosomes both of whose ends are non-homologous. This statement may be generalized as follows: Every successive interchange by which a ring of four is increased to six, eight and so on, will give one such new relatively translocated segment in opposite complexes unless two interchanges coincide. Thus, consider a ring of four and one pair of the following segmental constitution: