1918} DEVRIES—MASS MUTATIONS 417 
the mutation theory. The names of mass mutation and secondary 
mutation, therefore, seem to be very appropriate, indicating, as 
they do, the true explanation of the phenomenon. 
G. TWIN HYBRIDS, CONSIDERED AS A RESULT OF MASS MUTATION 
In Gruppenweise Artbildung (5) I have devoted a large part to 
the study of the twin hybrids of O. Lamarckiana and its derivatives 
in their crosses with other species. I was convinced that some 
relation must exist between the cause of this curious phenomenon 
and the high degree of mutability of the species. I supposed this 
internal cause, whatever it might be, to be responsible in a large 
degree, not for the mutability itself, since this is not a special trait 
of the Lamarckiana, but for the exceptionally high degree of develop- 
ment of the quality in that species. 
Later investigations of different authors, and especially those of 
RENNER, have confirmed this conception, since they do not offer 
an explanation of the problem involved on the basis of other excep- 
tional qualities of my plant. The experiments described for O. 
grandiflora, however, prove that there is still another relation, since 
the twins may be considered as the result of the fecundation of 
sexual cells which are, for a large part, in the condition of mutated 
gametes. It is evident that in crosses these latter may give differ- 
ent hybrids from those of the normal gametes of the same parent. 
I shall now try to show that the results of my crosses confirm this 
view in almost all their details. 
We have to start from the assumption that the mass mutations 
take place in the same numerical proportions as those required by 
the formula of Mendel for monohybrids; in other words, that the 
two kinds of gametes are produced in equal numbers and among 
the pollen as well as among the egg cells. Fecundation with a 
different species must then produce two kinds of hybrids, each of 
them in about 50 per cent of the offspring. Our table for the pro- 
duction of laeta and velutina in such crosses gave on the average 
52 per cent for the first and 46 per cent for the latter, and thus fully 
confirms our conception. When the crosses are repeated with mut. 
ochracea instead of the type of the species itself, no twins must be 
the result, but only uniform hybrids of the type corresponding with 
