Mutations and Evolution. 83 
may be supplemented and confirmed by studies of the germ-cells. 
For, while her counts are probably correct, yet the phenomena of 
fragmentation of chromosomes in somatic cells make it highly 
desirable that they should all he checked by observations on the 
chromosomes during meiosis. She has determined the chromosome 
number in 234 plants, including sixteen different mutant types, 11 
of which types had 15 chromosomes, three had 16, and two were 
triploid. That (E. mut. albida has 15 chromosomes has been 
verified by counting the number in thirteen more plants. As will 
be seen from Table I, she finds an extra chromosome also in a new 
mutant called bipartita , in subovata , elliptica (?) and “ type 2806,” all 
from lamarckiana. In addition, a second dwarf type, unlike 
nanella lata , is found, lata as well as lata X lamarckiana give “ type 
4499” with 15 chromosomes, while lata self-pollinated produces 
exilis, exundans and “type 5365.” None of these latter forms are 
described, so it is impossible to judge of their claims to separate 
status, but it seems clear that the three forms appearing in the 
offspring of lata selfed are secondary in origin, i.e. t bear some 
special relation to lata. All that can be said about these forms 
at present is that there appear to be at least seven distinct types 
which might arise through duplication of different chromosomes, 
and that in addition a number of these types appear to be more 
closely related to lata. This is certainly the case with semilata 
Gates, and probably to a different degree or in a different way 
with several others. This may be connected with other differential 
factors, but it is useless to speculate about the matter until the 
facts themselves are clearer. Mut. oblonga apparently has some 
forms with 15 chromosomes and some with 14. The same is true 
of rubrinervis. In this case the true rubrinervis of de Vries, a 
rather weak form, has 14 chromosomes plus a small one. The form 
which other investigators have called rubrinervis , derived from seeds 
of de Vries, is taller and more robust and certainly has 14 chromo¬ 
somes. It is the form of my cytological studies (Gates 1908), the 
form from which rubricalyx originated (Gates 1911), and may be 
called rubrinervis Gates for identification. It is more like the 
subrobusta of de Vries, which is derived from rubrinervis x 
lamarckiana , and may perhaps be identical with that form or with 
the closely related erythrina (de Vries, 1919) to be discussed later. 
Miss Lutz concludes that a very large percentage of the mutant 
offspring of (E. lamarckiana have 15 chromosomes, there being both 
a larger number of distinct types and a higher percentage of mutant 
individuals with the extra chromosome. One would feel safer 
