l62 GENETICS: MORGAN, STURTEVANT, BRIDGES Proc. N. A. S. 
leave only gas as interstitial material. This is, however, not the only 
possibility. It is not difficult to conceive of a distribution of forces in the 
heterogeneous surroundings of an igneous mass such that locally the liquid 
might be sucked out of a crystalline mesh, which would acquire a miarolitic 
texture, temporarily, at least. The consequences of this action would 
normally differ in no essential particular from those produced by simple 
squeezing out of the liquid. It is probable, however, that this sucking 
action could be operative at a very late stage of crystallization when 
squeezing out of liquid may be impossible or at least very unlikely. The 
kinds of liquid that form some of the alkaline rocks may perhaps be re- 
moved in this manner from quartz-mica rocks at a very late stage of their 
crystallization and possibly only such action can effect the separation of 
these liquids. In certain regions of the earth's crust where tangential 
extension is the dominant expression of the forces acting (Atlantic struc- 
tures) the development of alkaline rocks might be a prominent feature 
though the conditions requisite to their formation would undoubtedly 
occur locally elsewhere. 
^ The Later Stages of the Evolution of the Igneous Rocks, J. Geol., 23, 1915, Suppl., 
pp. 1-91. 
- Howie, Robt., Summary Rept. Geol. Survey Can., 1916, Coleraine Map Sheet msert, 
p. 228; also Knox, G. K.,Ibid., p. 229, et. seq. Reference is made to granitic "batho- 
liths" in the anticlines, p. 232. 
THE EVIDENCE FOR THE LINEAR ORDER OF THE GENES 
By T. H. Morgan, A. H. Sturtevant and C. B. Bridge 
Department of Zoology, Columbia University 
Communicated February 25, 1920 
Despite Castle's dictum that we "have failed in two different attempts 
to establish the linear theory in the case of the three genes yellow, white 
and bifi-d," we are bold enough to maintain that the data furnished, and 
still furnish, the proof called for. We wish to call attention to the fact 
that in his last paper Castle ignores our proof of the linear order that is 
furnished by building up the whole chromosome (or even large sections of 
it) by "distances" so short that no double cross-over classes appear. 
Castle asserts that we have rejected "nearly 99 per cent" of our data in 
the construction of the yellow, white, bifid section of the map. As a matter 
of fact no data have been omitted. In this case, as always, the order of the 
loci was determined by experiments that involved all of these loci at once. 
The order having been established the next step was to determine the rela- 
tive distance between the loci by the use of all the available data. We 
have emphasized in our reply to Castle that there are several sources of 
variability in linkage values such as age, temperature, genetic factors. 
The variability due to these causes far outweighs that due to random 
sampling. It is, therefore, inadmissible to compare data from different 
