August 25, 1904] 



NA TURE 



411 



show such departure from normal expectation that either 

 the numbers produced must have been unequal, or sub- 

 sequent disturbance must have occurred. But so far no 

 case is known /or certain where the average of families does 

 not point to equality. 



The fact that equality is so usual has a direct bearing on 

 conceptions of the physical nature of heredity. I have 

 compared our segregation with chemical separation, but the 

 phenomenon of numerically symmetrical disjunction as a 

 feature of so many and such different characters seems 

 scarcelv favourable to any close analogy with chemical 

 processes. If each special character owed its appearance to 

 the handing on of some comple.x molecule as a part of one 

 chemical system, we should expect, among such a diversity 

 of characters and forms of life, to encounter some pheno- 

 menon of valency, manifested as numerical inequality 

 between members of allelomorphic pairs. So tar, equi- 

 valence is certainly the rule, and where the characters are 

 simply paired and no resolution has taken place, this rule 

 appears to be universal as regards averages. On the other 

 hand, there are features in the distribution of characters 

 after resolution, when the second generation (F„) is poly- 

 morphic in a high degree, which are not readily accounted 

 fur on any hypothesis of simple equivalence ; but none of 

 ihese cases are as yet satisfactorily investigated. 



It is doubtful whether segregation is rightly represented 

 as the separation of iit'o characters, and whether we may 

 not more simply imagine that the distinction between the 

 allelomorphic gametes is one of presence or absence of some 

 <listinguishing element. De \'ries has devoted much atten- 

 tion to this question in its bearings on his theory of Pan- 

 genesis, holding that cases of both kinds occur, and attempt- 

 ing to distinguish them. Indications may certainly be 

 enumerated pointing in either direction, but for the present 

 I incline to defer a definite opinion. 



If we may profitably seek in the physical world for some 

 parallel to our gametic segregations, we shall, I think, 

 find it more close in mechanical separations, such as those 

 which may be effected between fluids which do not freely 

 mix, than in any strictly chemical phenomenon. In this 

 ^vay we might roughly imitate both the ordinary segre- 

 gation, which is sensibly perfect, and the curious' impurity 

 occasionally perceptible even in the most pronounced dis- 

 continuities, such as those which divide male from female, 

 petal from sepal, albino from coloured, horn from hair, 

 imd so on. 



Gitmctic L'nions aud their Consequctiees. 



Characters being then distributable among gametes 

 according to regular systems, the next question concerns 

 the properties and features presented by the zygotes formed 

 by the union of gametes bearing different characters. 



.\s to this no rule can as yet be formulated. .Such a 

 helerozygote may exhibit one of the allelomorphic characters 

 in its full intensity (even exceeding it in special cases, 

 perha[)s in connection with increased vigour), or it may be 

 intermediate between the two, or it may present some 

 character not recognisable in either parent. In the latter 

 case it is often, though not always, reversionary. When one 

 character appears in such intensity as to conceal or exclude 

 the other it is called dominant, the other being recessive. 

 It may be remarked that frequently, but certainly not 

 universally (as has been stated), the phylogenetically older 

 character is dominant. A curious instance to the contrary 

 is that of the peculiar arrangement of colours seen in a 

 breed of game fowls called Brown-breasted, which in com- 

 bination with the purple face, though certainly a modern 

 variation, dominates (most markedly in females) over the 

 Black-breasted type of Gallus bankiva. 



In a few cases irregularity of dominance has been observed 

 as .in e.xception. The clearest illustration I can offer is 

 th.-it of the extra toe in fowls. Generally this is a dominant 

 character, but sometimes, as an exceptional phenomenon, 

 it may be recessive, making subsequent analysis very 

 difficult. The nature of this irregularity is unknown. A 

 remarkable instance is that of the blue colour in maize seeds 

 (Correns ; R. H. Lock). Here the dominance of blue is 

 frequently imperfect, or absent, and the figures suggest that 

 some regularity in the phenomenon may be discovered. 



•Mendel is often represented as having enunciated domin- 

 ance as a general proposition. That this statement should 



NO. 1817, VOL. 70] 



still be repeated, even by those who realise the importance 

 of his discoveries, is an extraordinary illustration of the 

 oblivion that has overwhelmed the work of the experimental 

 breeders. Mendel makes the specific statement in regard to 

 certain characters in peas which do behave thus, but his 

 proposition is not general. To convict him of such a 

 delusion it would be necessary to prove that he was ex- 

 ceptionally ignorant of breeding, though on the face of the 

 evidence he seems sufficiently expert. 



A generalisation respecting the consequences of hetero- 

 zygosis possessing greater value is this. When a pair of 

 gametes unites in fertilisation the characters of the zygote 

 depend directly on the constitution of these gametes, and 

 not on that of the parents from which they came. To this 

 generalisation we know as yet only two clear exceptions. 

 I'hese vciy curious cases are exactly alike in that, though 

 segregation obviously occurs in a seed-character, the seeds 

 borne by the hybrid (F,) all exhibit the hybrid character, 

 and the consequences of segregation in the particular seed- 

 character are not evident until the seeds (Fj) of the second 

 (F,) generation are determinable. Of these the first is the 

 case of indent peas investigated especially by Tschermak. 

 Crossed with wrinkled peas I have found the phenomena 

 normal, but when the cross is made with a round type the 

 exceptional phenomenon occurs. The second case is that 

 discovered by Biffen in the cross between the long-grained 

 wheat called Polish and short-grained Rivett wheat, demon- 

 strations of which will be laid before you. No satisfactory 

 account of these peculiarities has been yet suggested, but 

 it is evident that in some unexplained way the maternal 

 plant-characters control the seed-characters for each gener- 

 ation. It is, of course, likely that other comparable cases 

 will be found. 



Appearances have been seen in at least four cases (rats, 

 mice, stocks, sweet peas) suggesting at first sight that a 

 heterozygosis between two gametes, both extracted, may 

 give, e.g., dominance; while if one, or both, were pure, 

 they would give a reversionary heterozygote. If this occur- 

 rence is authenticated on a sufficient scale, we shall of course 

 recognise that the fact proves the presence in these cases 

 of some pervading and non-segregating quality, distributed 

 among the extracted gametes formed by the parent hetero- 

 zygote. As yet, however, I do not think the evidence 

 enough to warrant the conclusion that such a pervading 

 quality is really present, and I incline to attribute the appear- 

 ances to redistribution of characters belonging to indepen- 

 dent pairs in the manner elucidated by Cu^not. The point 

 will be easily determined, and meanwhile we must note 

 the two possibilities. 



Following, therefore, our first proposition, that the 

 gametes belong to definite classes, comes the second pro- 

 position, that the unions of members of the various classes 

 have specific consequences. Nor is this proposition simply 

 the truistical statement that different causes have different 

 effects ; for by its aid we are led at once to the place where 

 the different cause is to be sought — Gametogenesis. While 

 formerly we hoped to determine the offspring by examining 

 the ancestry of the parents, we now proceed by investigating 

 the gametic composition of the parents. Individuals may 

 have identical ancestry (and sometimes, to all appearances, 

 identical characters), but yet be quite different in gametic 

 composition ; and, conversely, individuals may be identical 

 in gametic composition and have very different ancestry. 

 Nevertheless, those that are identical in gametic composi- 

 tion are the same, whatever their ancestry. Therefore, 

 where such cases are concerned, in any considerations of 

 the physiology of heredity, ancestry is misleading and passes 

 out of account. To take the crudest illustration, if a hybrid 

 is made between two races. A, B, and another hybrid 

 between two other races, C, D, it might be thought that 

 when the two hybrids .\B and CD are bred together, four 

 races. A, B, C, and D, will be united in their offspring. 

 This expectation may be entirely falsified, for the cell- 

 divisions of gametogenesis may have split A from B and 

 C from D, so that the final product may contain characters 

 of only two races after all, being either AC, BC, AD, or 

 BD. In practice, however, we are generally dealing with 

 groups of characters, and the union of all the A group, for 

 instance, with all the C group will be a rare coincidence. 



It is the object of Mendelian analysis to state each case 

 of heredity in terms of gametic composition, and thence to 



