180 
CHARLES L. PARMENTER 
Complexes of the second class. In fourteen cells of this class 
there is one point in one chromosome and in four cells there is 
one point in each of two chromosomes which, to persons hyper- 
critically inclined, might possibly appear uncertain. To one 
acquainted with the material, each of these points is entirely 
clear, and even when accepted as subject to interpretation it is 
very plain how the interpretation should be made—so plain that 
I am certain that the count of twenty-eight chromosomes is 
accurate and dependable. But for the sake of unquestionable 
fairness I have placed these cells in a separate group. As to 
the exact nature of the interpretations in these eighteen com¬ 
plexes, four of them have some small portion of only one chro¬ 
mosome so covered by others that it cannot be traced over its 
entire length without losing sight of it as stated above (p. 178). 
Two other cells had two chromosomes of this nature. Five 
complexes have a single chromosome lying in such a relation to 
another chromosome that it might possibly be interpreted as a 
part of the other chromosome (e.g., fig. 23, chropiosome i), and 
in three more cells there were two such chromosomes. In the 
remaining five complexes a single chromosome was so situated 
or otherwise involved, that it might be interpreted that there 
were two chromosomes present (e.g., fig. 21, i). 
In considering all the interpretation possible in each of these 
eighteen cells the minimum number in any one of them would 
be twenty-seven and the maximum number thirty. Even grant¬ 
ing this much variation, it is far removed from that expected in 
a series of chance variants as Della Valle claims them to be. 
The points in question were sketched as described above 
before the chromosomes were counted, so that the determination 
of the number of chromosomes was not influenced, either con¬ 
sciously or unconsciously, by a knowledge of how many chromo¬ 
somes were present or by how they should be sketched in order 
to produce the expected number. This procedure and the fact 
that the number counted always agreed with the number present 
in the forty-five cells of class I make it practically certain that 
the enumeration is correct. It should be emphasized again that 
these cases are only subject to question when hypercritically 
