130 W. R. B. ROBERTSON 
as one would expect in instances where a vital factor had been 
dropped. He has mutations which, with one or possibly two 
exceptions, are of a retrogressive nature; i.e., lacking something 
necessary which was present in the parent species. The gigas 
variety was due evidently to a doubling of the number of chromo- 
somes (Gates ’09). It seemed to lack nothing, but the other 
mutants seemed all to lack traits, some more useful, some less 
useful, which were present in the parent species. These phenom- 
ena, it seems to me, point to something for their basis like the 
abnormal variations in reduction divisions, such as I have de- 
scribed in Tettigidea parvipennis. 
It is interesting to compare the number of mutations De 
Vries obtained from lamarckiana with the number of chromo- 
somes. His number of chromosomes was 14, seven pairs. He 
obtained seven mutations from his plants. One of these, gigas, 
was evidently due to a doubling of the number of chromosomes. 
It, however, was not a defective mutant, and so may be left 
out of account here. The other mutants seemed to have some- 
thing lacking, and there were six of them. Of these scintillans 
seems to have been heterozygous, producing on inbreeding, 
lamarckiana and scintillans. It also produced, in a small per- 
centage, some of the mutants most frequently produced by 
lamarckiana. Possibly scintillans had one over-deficient chro- 
mosome, such as I have postulated for yellow mice or the golden 
snapdragon. Many sterile pollen grains were found. Possibly 
the cause of sterility lies here. On this hypothesis, the fact that 
scintillans can produce oblonga, lata, and nannella is not sur- 
prising. The other chromosome pairs are just as liable to acci- 
dents in germ cells of scintillans as of lamarckiana. Lamarck- 
lianas may be produced by scintillans; why may not the mutants of 
lamarckiana be produced also? Each of the other five mutations 
might be based respectively upon deficiencies 1n one of the re- 
maining five pairs of chromosomes. Since these remaining 
mutants breed true in each case, it would be supposable that, in 
order to show, they must be in the homozygous condition. Thus 
we may possibly account for the five remaining mutants. This 
hypothesis, it seems to me, is worthy of consideration here. I 
