4IO 



NATURE 



[August 25, 1904 



warrant a positive statement as to the total absence of 

 segregation, for in the few families raised from pairs of 

 artificial intermediates some dubious indications of segre- 

 gation have been seen. 



The rarity of true failure of segregation when pure strains 

 are crossed may be judged by the fact that since the revival 

 of interest in such work hardly any thoroughly satisfactory 

 cases have been witnessed. The largest body of evidence 

 on this subject is that provided by De Vries. These cases, 

 however, present so many comple.xities that it is impossible 

 to deal with them now. While so little is definitely known 

 regarding non-segregating characters, it appears to me 

 premature to attempt any generalisation as to what does 

 or does not segregate. 



Most of the cases of failure of segregation formerly 

 alleged are evidently spurious, depending on the appearance 

 of homozygotes in the second generation (F,). 



One very important group of cases exists, in which the 

 appearance of a partial failure of segregation after the 

 second generation (F,) is really due to another phenomenon. 

 The visible character of a zygote mav, for instance, depend 

 on the coexistence in it of two characters belonging to 

 distinct allelomorphic pairs, each capable of being in- 

 dependently segregated from its fellow, and forming in- 

 dependent combinations. For the demonstration of this 

 important fact we are especially indebted to Cu^not.' We 

 have indications of the existence of such a phenomenon in 

 a considerable range of instances {mice, rabbits (Hurst), 

 probably stocks and sweet peas). 



Nevertheless, there are other cases, not always easy to 

 distinguish from these, where some of the gametes of F, 

 certainly carry on heterozygous characters unsegregated. 

 As an example, which seems to me indisputable, I mav 

 mention the so-called " walnut " comb, normal to Malay 

 fowls. This can be made artificially by crossing rose-comb 

 with pea-comb, and the cross-bred then forms gametes, of 

 which one in four bears the compound unsegregated." We 

 may speak of this as a true synthesis. 



In another type of cases segregation occurs, but is not 

 sharp. The gametes may then represent a full series 

 ran.<?-ing from the one pure form to the other. Such cases 

 occur in regard to some colours of Primula sinensis, and 

 the leg-feathering of fowls (Hurst). In the second gener- 

 ation a nearly complete series of intermediate zygotes may 

 result, though the two pure extremes (if the case be one 

 of blending characters) may still be found to be pure. 



Resolution and Disintegration. — Besides these cases, the 

 features of which we now in great measure comprehend, 

 we encounter frequently a more complex segregation, 

 imperfectly understood, by which gametes of new tvpes, 

 sometimes very numerous, are produced by the crossbred. 

 Each of these new types has its own peculiarities. We 

 shall, I think, be compelled to regard these phenomena as 

 produced either by a resolution of compound characters 

 introduced by one or both parents, or by some process of 

 disintegration, effected by a breaking-u'p of the integral 

 characters followed by recombinations. It seems impossible 

 to imagine simple recombinations of pre-existing characters 

 as adequate to produce many of these phenomena. Such 

 a view would involve the supposition that the number of 

 characters pre-existing as units was practicallv infinite — a 

 difficulty that as yet we are not obliged to face. However 

 that may be, we have the fact that resolutions and dis- 

 integrations of this kind — or recombinations, if that con- 

 ception be preferred — are among the common phenomena 

 following crossing, and are the sources of most of the 

 breeder's novelties. As bearing on the theoretical question 

 to which I have alluded, we may notice that it is among 



1 When aic . . . X a^y . . . gives in Fj or Fo a character (not seen in 

 the original parents), which from Fn or later may breed true : not because 

 aa, i^, cy do not severally segregate, but through simultaneous homo- 

 zygosis of, say, aa and /3^, giving a zygote aapficy . . . which will breed 

 true to the character ap. 



- Owing to this behaviour, and to the simultaneous production of single- 

 comb (? by resolution), there are, even in pure Malays, five types of indi- 

 viduals, all with " walnut " combs— as yet indistinguishable— formed by 

 gametic unions /-x/, r>Xr/, r^Xr, rpy fi, r/,y.s. Of these kinds three can 

 at once be distinguished by crossmg with single ; but whether >-x/ can be 

 distinguished from r/X5 we do not yet know, [r, rose ;/, pea ; s, single; 

 rp walnut ] In this example four allelomorphs are simultaneously segre- 

 gated, one being compound. Neglecting sexual diiferentiation, there are 

 therefore ten gametically distinct types theoretically possible ; but of these 

 only^7/r are distmguishable by inspection. 



NO. J817, VOL. 70] 



examples of this complex breaking-up that a great pi-oportion 

 of the cases of partial sterility have been seen. 



No quite satisfactory proof as to the actual moment of 

 segregation yet e.xists, nor have we any evidence that all 

 characters are segregated at the same cell-division. Correns 

 has shown that in maize the segregation of the starch 

 character from the sugar character must happen before the 

 division forming the two generative nuclei, for both bear 

 the same character. The reduction-division has naturally 

 been suggested as the critical moment. The most serious 

 difficulty in accepting this view, as it seems to me, is the 

 fact that somatic divisions appear sometimes to segregate 

 allelomorphs, as in the case of Datura fruits, and some 

 colour-cases. 



In concluding this brief notice of the complexities of 

 segregation I may direct attention to the fact that we are 

 here engaged in no idle speculation. For it is now possible 

 by experimental means to distinguish almost always with 

 which phenomenon we are dealing, and each kind of compli- 

 cation may be separately dealt with by a determination of 

 the properties of the extracted forms. Illustrations of a 

 practical kind will be placed before j'ou at a subsequent 

 meeting. 



The consequence of segregation is that in cases where it 

 occurs we are rid of the interminable difficulties which beset 

 all previous attempts to unravel heredity. On the older 

 view, the individuals of any group were supposed to belong 

 to an indefinite number of classes, according to the various 

 numerical proportions in which various types had entered 

 into their pedigree. We now recognise that when segre- 

 gation is allelomorphic, as it constantly is, the individuals 

 are of three classes only in respect of each allelomorphic 

 pair — two homozygous and one heterozygous. In all such 

 cases, therefore, fixity of type, instead of increasing 

 gradually generation by generation, comes suddenly, and 

 is a phenomenon of individuals. Only by the separate 

 analysis of individuals can this fact be proved. The sup- 

 position that progress towards fixity of type was gradual 

 arose from the study of masses of individuals, and the 

 gradual purification witnessed was due in the main to the 

 gradual elimination of impure individuals, whose individual 

 properties were wrongly regarded as distributed throughout 

 the mass. 



We have at last the means of demonstrating the presence 

 of integral characters. In affirming the integrity of segre- 

 gable characters we do not declare that the size of the 

 integer is fixed eternally, as we suppose the size of a 

 chemical unit to be. The integrity of our characters 

 depends on the fact that they can be habitually treated as 

 units by gametogenesis. But even where such unity is 

 manifested in its most definite form, we inay, by sufficient 

 searching, generally find a case where the integrity of the 

 character has evidently been impaired in gametogenesis, and 

 where one such individual is found the disintegration can 

 generally be propagated. That the size of the unit may be 

 changed by unknown causes, though a fact of the highest 

 significance in the attempt to determine the physical nature 

 of heredity, does not in the least diminish the value of the 

 recognition of such units, or lessen their part in govern- 

 ing the course of Evolution. 



The existence of unit-characters had, indeed, long been 

 scarcely doubtful to those practically familiar with the facts 

 of variation (cp. De Vries, " Intracellulare Pangenesis," 

 1889), but it is to the genius of Mendel that we owe the 

 proof. We knew that characters could behave as units, but 

 we did not know that this unity was a phenomenon of 

 gainetogenesis. He has revealed to us the underworld of 

 gametes. Henceforth, whenever we see a preparation of 

 germ-cells we shall remember that, though all may look 

 alike, they may in reality be of many and definite kinds, 

 differentiated from each other according to regular systems. 



Numerical Relations of Gametes and their Significance. 



In addition to the fact of segregation, Mendel's experi- 

 ments proved another fact nearly as significant ; namely, 

 that when characters are allelomorphic, the gametes bear- 

 ing each member of a pair generally are formed in equal 

 numbers by the heterozygote, if an average of cases be 

 taken. This fact can only be regarded as a consequence 

 of some numerical symmetry in the cell-divisions of gameto- 

 genesis We already know cases where individual families 



