R C. PUNKETT 93 



Hence in a series of three reduplications, two primary and one 

 secondary, that one is to be regarded as the secondary in which the value 

 of the reduplication is lowest. 



In this connection the two groups of families recorded in Table XI 

 are also of interest. In Group A from the mating DfN x dFn there is 

 repulsion between D and F, coupling between D antl N, and repulsion 

 between N and F. The first repulsion is not improbably on the 1 :15:15:1 

 system or something near it (cf j). 89), while the second repulsion and 

 the coupling are not far removed fi-om the 1:3:3:1 and the 3:1:1:3 

 systems respectively. The figvu-es are in general accordance with Trow's 

 hypothesis, but tlie numbers are not large enough to determine more 

 precisely the values of the reduplication systems, or tci decide which of 

 the two lower .series is the secondary one. 



In the families of Group B in Table XI, where the nature of the 

 mating is dfN x DFn, there is coupling between D and F but no 

 repulsion between N and D or between N and F. Here again the 

 experimental facts are in accordance with Trow's hypothesis, for where 

 one of the primary series shews no reduplication it follows that no 

 reduplication will be exhibited by the secondary series. 



* 



Conclusion. 



Finally attention may be drawn to some points in connection with 

 the value of the reduplication in the various cases discussed above. 

 Where only two factors are concerned we have regarded the redupli- 

 cation as of the form (?; — 1) : 1 where n is some power of 2, and we 

 suggested in a previous paper how such a series might be brought 

 about through alternating periclinal and anticlinal cell divisions ((5), 

 Fig. 4). The experimental data hitherto obtained from sweet peas fit 

 in with this view, but, as pointed out some years ago ((2), p. 9), they 

 are also in accordance with the form of the reduplication being 

 n : 1 : 1 : « where n again is some power of 2. It is only in cases 

 where n is very small that we can hope to distinguish between the 

 two without growing an impracticably large number of plants. At 

 present the NF x nf mating is the only one from which we can look 

 for a critical result on this point, and the available evidence suggests 

 that the reduplication here is 3NF : INf : InF : 3nf (cf p. 86), i.e. that 

 it is on the {n — 1) : 1 basis rather than on the n : 1 basis'. It may 



1 What is evidently a case of reJuplication on the 3 : 1 basis has been recently dis- 

 covered by Collins in maize ((6), p. 579), the cross in (juestion being one between an 

 Journ. of Gen. iii • 7 



