588 REPORT—1904. 
reply that the Owl’s egg existed before the Owl; and if we hesitate about the 
Owl, we may be sure about the Bantam. The parent zygote, whose offspring 
display variation, is giving off new gametes, and in its gametogenesis a segre- 
gation of their new character, more or less pure, is taking place. The significance 
and origin of the discontinuity of variation is therefore in great measure evident, 
So far as pre-existing elements are concerned, it is an expression of the power 
of cell-division to distribute character-units among gametes. The initial purity 
of so many nascent mutations is thus no longer surprising, and, indeed, that such 
initial purity has not been more generally observed we may safely ascribe to im- - 
perfections of method. 
It is evident that the resemblance between the parent originating a variety 
and a heterozygote is close, and the cases need the utmost care in discrimination. 
Tf, for instance, we knew nothing more of the Andalusian fowl than that it 
throws blacks, blues, and whites, how should we decide whether the case was one 
of heterozygosis or of nascent mutation? The second (F,) generation from 
Brown Leghorn x White Leghorn contains an occasional Silver-Grey or Duckwing 
female. Is this a mutation induced by crossing, or is it simply due to a recom- 
bination of pre-existing characters? We cannot yet point to a criterion which 
will certainly separate the one from the other; but perhaps the statistical 
irregularity usually accompanying mutation, contrasted with the numerical sym- 
metry of the gametes after normal heterozygosis, may give indications in simple 
cases—though scarcely reliable even there. These difficulties reach their maximum 
in the case of types which are continually giving off a second form with greater or 
less frequency as a concomitant of their ordinary existence. This extraordinarily 
interesting phenomenon, pointed out first by De Vries, and described by him 
under the head of ‘ Halb-’ and ‘ Mittel-Rassen,’ is too imperfectly understood for 
me to do more than refer to it, but in the attempt to discover what is actually 
taking place in variation it must play a considerable part. 
Just as that normal truth to type which we call heredity is in its simplest 
elements only an expression of that qualitative symmetry characteristic of all non- 
differentiating cell-divisions, so is genetic variation the expression of a qualitative 
asymmetry beginning in gametogenesis. Variation is a novel cell-division.1 So 
soon as this fact is grasped we shall hear no more of heredity and variation as 
opposing ‘ factors’ or ‘ forces ’—a metaphor which has too long plagued us. 
We cease, then, to wonder at the suddenness with which striking variations 
arise. Those familiar with the older literature relating to domesticated animals 
and plants will recall abundant instances of the great varieties appearing early in 
the history of a race, while the finer shades had long to be waited for. In the 
sweet pea the old purple, the red bicolor, and the white have existed for genera- 
tions, appearing soon after the cultivation of the species; but the finer splitting 
which gave us the blues, pinks, &c., is a much rarer event, and for the most part 
only came when crossing was systematically undertaken. If any of these had 
been seen before by horticulturists, we can feel no doubt whatever they would 
have been saved. An observer contemplating a full collection of modern sweet 
peas, and ignorant of their history, might suppose that the extreme types had 
resulted from selective and more or less continuous intensification of these inter- 
mediates, exactly inverting the truth. 
We shall recognise among the character-groups lines of cleavage, along which 
they easily divide, and other finer subdivisions harder to effect. Rightly con- 
sidered, the sudden appearance of a total albino or a bicolor should surprise us 
less than the fact that the finer shades can appear at all. 
At this point comes the inevitable question, What makes the character-group 
split? Crossing, we know, may do this; but if there be no crossing, what is the 
cause of variation? With this question we come sharply on the edge of human 
1 The parallel between the differentiating divisions by which the parts of the 
normal body are segregated from each other, and the segregating processes of 
gametogenesis, must be very close. Occasionally we evén see the segregation of 
Mendelian characters among zygotic cells, 
