540 



SCIENCE 



[N. S. Vol. XLIV. No. 1137 



nearer view. Pending this, judgment must be 

 suspended. We are told that the breeding 

 numbers prove the factors to be in four linked 

 groups. We would like to take each one sepa- 

 rately and follow the proof regarding its link- 

 ages. As yet there is no means of doing this. 

 Of the evidence the book avowedly gives illus- 

 trative specimens merely, and even the long 

 array of Drosophila papers leaves great gaps 

 unfilled. Take the first or sex-linked series. 

 The book tells us that more than 40 factors 

 have been located in it and arranged in order. 

 Respecting the great majority we have no de- 

 tails at all and as to most of the remainder 

 very few. There are, however, six that we can 

 examine in the light of the data summarized 

 by Sturtevant in Zeits. f. Vererbungsl., 1914, 

 the last considerable body of evidence to hand. 

 The factors concerned, called T, W, V, M, 

 R, Br, are represented as arranged along the 

 chromosome in such a way that two, T and W, 

 are at the zero end, two more, V and M, near 

 together at 33.5 and 36.5, and the remaining 

 two, R and Br, also near together at 53.3 and 

 57.7. The numbers indicate that the members 

 of each set of two are closely linked, for with 

 fair consistency the breeding ratios are those 

 characteristic of close " coupling," namely, 

 riAB: lAb : IaB : nab, and of "repulsion" in 

 the form IAB : nAb : naB : lab, the value of n 

 being much greater than 1. The relations of 

 T and W to V and M are also of this kind, the 

 coupling being of course less close. But taking 

 Y with R, W with R, V with R, or M with R, 

 we meet numbers of a very different order, 

 and it is not clear by what system they have 

 been interpreted. For instance, we find the 

 following extraordinary series given, 



for T with -K 



as repulsion 342 58 466 19 



as coupling 235 50 194 56 



for W with It 



as repulsion 567 143 697 91 



uncoupling 294 61 175 108 



for V with R 



as repulsion 147 147 520 36 



for M with B 



as repulsion 430 795 1,716 189 



as coupling 4,189 93 850 1,033 



The numbers in which the new combinations 

 come are then added in each case and set out 



as percentages of the totals, these percentages 

 being taken as indications of the linear dis- 

 tances between the loci in which the factors 

 are presumed to be. To those accustomed to 

 series of this class, these numbers are so aber- 

 rant as scarcely to suggest prima facie that 

 they represent Mendelian series at all, and it 

 seems at least improbable that they can be 

 used to calculate percentages comparable with 

 those obtained from the various comparatively 

 normal series by which for instance Y and M, 

 V and Bz, 5 Y and W, or W and M are inter- 

 related. Throughout the experiments indica- 

 tions of differential viability recur, largely 

 masking the true proportions of the classes, 

 but as has been remarked by the authors in 

 reference to certain special cases, the incidence 

 of this differentiation is so irregular that 

 allowance can not be made for it in any con- 

 sistent fashion. Meanwhile the data look so 

 intractable that a doubt has sometimes arisen 

 whether the account here given may not be a 

 consequence of some radical misunderstanding 

 of the author's meaning. 



One is tempted further to ask whether all 

 parts of the several proofs are really independ- 

 ent of each other. In the present state of the 

 evidence only the authors themselves can posi- 

 tively answer this question. They declare that 

 all the factors are proved to be disposed in 

 four separate systems of linkage, but the argu- 

 ment that they are thus arranged contemplates 

 a very great variety of possibilities not obvi- 

 ously included in this scheme. For example, 

 the fact that two pairs of gens or factors give 

 50 per cent, of cross-overs might in the 

 authors' view be a consequence of the location 

 of the two pairs in distinct chromosomes. It 

 may equally be a consequence of the two being 

 in the same chromosome but at the terminal 

 and central positions respectively. It may also 



s The numbers given for V and Br by Sturte- 

 vant are misprinted, 260 standing for 2,660. Thus 

 emended they are fairly normal. The worst ex- 

 amples all involve J5, and it might be suspected 

 that this was a source of special difficulty, but 

 analogous numerical abnormalities occur also in 

 the ' ' second chromosome ' ' series, nor can any 

 hypothesis of differential viability be readily ap- 

 plied to such figures as those quoted above. 



