164 
GENETICS: C. W. METZ 
Proc. N. a. S. 
cism misses the mark. It need only be stated that Castle was wrong in 
supposing that the cross-overs between yellow and white should be sub- 
tracted from the total before the white bifid cross-over value is calculated. 
If, as Castle states, he cannot conceive of a mechanism "which would 
tie two genes together in such a way that they will subsequently separate 
from each other oftener than they will remain together, yet this is what 
the idea of cross-overs in excess of 50 per cent amounts to," it would seem 
to follow that he has not really understood the mechanism that we have 
described, and which he has attacked ; for, whether such a mechanism really 
exists or not, it is nevertheless a conceivable mechanical device that could 
do just this thing. 
We have left to Dr. Metz the opportunity to answer Castle's criticism 
relating to D. virilis. 
To sum up: M^e believe that we have met all the pertinent criticisms 
that Castle has brought forward of our methods and conclusions, and that 
he has failed to meet our criticism of his three dimensional model. 
THE ARRANGEMENT OF GENES IN DROSOPHILA VIRILIS 
By Charlks W. Metz 
Station for Experimental Evolution, Carnegie Institution of Washington 
Communicated by T. H. Morgan, March 6, 1920 
In connection with his general attack on the theory of the linear ar- 
rangement of genes Castle^ has questioned my use of this theory in ex- 
plaining the genetical results obtained in Drosophila virilis. I did not 
consider it necessary to reply to this criticism because Castle's general 
position was shown by Sturtevant, Bridges and Morgan^ to be untenable 
as regards Drosophila melanogaster, with which D. virilis agrees in its 
mode of inheritance. Subsequently, however. Castle has reaffirmed 
his belief in the superiority of his hypothesis over that of linear arrange- 
ment^ and has apparently misconstrued the silence on my part with re- 
spect to Drosophila virilis. This would seem to call for a brief reply. 
From my paper on eight sex-linked characters in D. virilis^ Castle 
concluded that the genes dealt with could not be arranged in a linear series, 
and by applying his three dimensional hypothesis to the case he made 
certain predictions (6, p. 36) which he now wishes used as a test of his 
hypothesis. His general arguments in this case are the same as those 
he used pre\^iously in reference to Drosophila melanogaster, and since 
they are being treated by those toward whom they were first directed, 
I will avoid repetition by confining myself to his predictions. These apply 
to four undetermined cross-over values in D. virilis, namely, magenta- 
hairy, glazed-rugose, frayed-forked and frayed-glazed. 
It should be pointed out first that in making these predictions Castle 
apparently overlooked the statement in my paper (pp. 113 and 125) that 
the frayed stock had been lost shortly after it was obtained. The absence 
