1918] DEVRIES—MASS MUTATIONS 385 
grains of grandiflora with lorea egg cells. It points to a high 
amount of mutated sexual cells, but my cultures were too small 
and too few to justify a further discussion of this interesting 
point. 
I have also crossed the two mutants with one another. The 
results were as follows in June 1916: 
grandiflora ochracea lorea Total 
CO; ochraces Mlorea. 2 22 8 ) 30 
©. lorea Xochracea. .. 6252. 31 23 I 55 
Phe i VE Ou pas 53 31 I 85 
FerceGthee ek ee 62 37 I 
The results of the reciprocal crosses may be assumed to mean 
the same hereditary conditions, even as in the crosses of the pale 
mutant with the species. The specimen of /orea seems to be due 
to a corresponding mutation in the ochracea, showing that this 
mutability is not as wholly absent here as the results of self- 
fertilization seemed to indicate. 
In all these crosses the Jorea marks must be assumed to be reces- 
sive to the grandiflora character. I have not made any second 
generations to decide this question, but the results of my crosses 
with allied species will fill up this gap and show that in crosses with 
lorea this type is split off, as a rule, in the second generation in pro- 
portions which correspond to the law of Mendel. 
O. grandiflora mut. gigas (fig. 3) occurred in one specimen among 
the 1180 plants of my cultures of 1915, pointing to a coefficient of 
mutation of o.1 per cent. This mutant attracted my attention in 
May and was planted separately with some other seemingly aber- 
rant specimens. It opened its first flowers in the middle of August. 
They were strikingly larger, with broad, thick petals, a thicker tube 
of the calyx, thick filaments, anthers, and lobes of the stigma, and 
a rich supply of pollen. The flower buds were almost conical and 
the pollen was rich in quadrilateral grains, one of the characters of 
the gigas mutants of allied species. The nuclei of the young buds 
were investigated by my assistant Mr. C. VAN OVEREEM, who also 
counted the chromosomes in the young roots of the seedlings of 
the following year. The number was invariably 28, showing the 
