crossed by a fertile plant with a light
axil, the F_1 plants are all fertile with dark axils. But such plants in
F_2 give fertiles with light axils, fertiles with dark axils, and steriles
with dark axils in the ratio 1 : 2 : 1. No light axilled steriles appear
from such a cross owing to the repulsion between the factor for dark axil
(D) and that for the fertile anther (F).

These four cases have already been found in the sweet pea, and similar
phenomena have been met with by Gregory in primulas. To certain seemingly
analogous cases in animals where sex is concerned we shall refer later.

Now all of these four cases present a common feature which probably has not
escaped the attention of the reader. In all of them _the original cross was
such as to introduce one of the repelling factors with each of the two
parents_. If we denote our two factors by A and B, the crosses have always
been of the nature AAbb x aaBB. Let us now consider what happens when both
of the {94} factors, which in these cases repel one another, are introduced
by one of the parents, and neither by the other parent. And in particular
we will take the case in which we are concerned with purple and red flower
colour, and with long and round pollen, _i.e._ with the factors B and L.
When a purple long (BBLL) is crossed with a red round (bbll) the F_1 (BbLl)
is a purple with long pollen, identical in appearance with that produced by
crossing the long pollened red with the round pollened purple. But the
nature of the F_2 generation is in some respects very different. The ratio
of purples to reds and of longs to rounds is in each case 3 : 1, as before.
But instead of an association between the red and the long pollen
characters the reverse is the case. The long pollen character is now
associated with purple and the round pollen with red. The association,
however, is not quite complete, and the examination of a large quantity of
similarly bred material shows that the purple longs are about twelve times
as numerous as the purple rounds, while the red rounds are rather more than
three times as many as the red longs. Now this peculiar result could be
brought about if the gametic series produced by the F_1 plant consisted of
7 BL + 1 Bl + 1 bL + 7 bl out of every 16 gametes. Fertilization between
two such similar series of 16 gametes would result in 256 plants, of which
177 would be purple longs, 15 purple rounds, 15 red longs, and 49 red
rounds--a proportion of the