GENETICS: W. E. CASTLE 
501 
Such values have not been observed and are logically impossible. 
(4) The hypothesis of non-linear arrangement is simpler because it 
eliminates secondary hypotheses needed on the linear hypothesis to 
harmonize greater with lesser cross-over values, particularly the 
hypothesis of double crossing-over. 
Point (1) is dismissed by Morgan and his associates with the remark 
that it is "probably not intended to be taken seriously.'^ I, therefore, 
renew the suggestion and invite "serious' ' consideration of it. They 
argue that chromosomes are thread-like and that accordingly linkage- 
systems must be thread-like. This seems to me unconvincing, because 
we do not know how much of the visible chromosomes is composed 
of genes and how much is something else. Further it may be that 
the genes are not in the chromosomes at all but merely attached to 
them, especially if the attachments are molecular rather than mechani- 
cal. All the observational facts concerning linked genes would be 
equally well satisfied by such a view. It should be borne in mind 
that we have no evidence whatever that the chromosomes are genes 
but only that the genes are in some way connected with them. Ac- 
cordingly the spatial relations of genes can not be legitimately inferred 
from the shape of the chromosomes. 
Against point (2) the chief attack of Morgan and his associates is 
directed. No objection is made to the method employed of recon- 
struction in three dimensions, but the reliability of the data used is 
repeatedly assailed. Now the data consisted of Table 65 of Morgan 
and Bridges on which the authors based their Diagram I showing the 
arrangement of the genes as linear. Surely this same data could 
legitimately be used in testing the hypothesis that the arrangement 
is non-linear. Moreover the Table 65 is, so far as I know, the only 
table in which the authors have ever given a comprehensive summary 
of their data. At the head of the table, they say "In Table 65 all 
data so far secured upon the sex-linked characters are summarized." 
Now my model, based on actual measurements of distances indicated 
by Table 65, showed conclusively that the arrangement of the genes 
can not possibly be linear. No alternative is therefore left to Morgan 
and his associates except either to repudiate their Diagram I or to 
repudiate the data on which that diagram was ostensibly based. 
They choose the latter alternative. In discussing the relation to 
each other of any three genes, they now reject all data except such 
as are based on simultaneous observation of all three loci. This is 
done for the seemingly good reason that in this way all disturbing 
