556 Gates .— The Trisomic Mutations of Oenothera. 
included three plants with 16 chromosomes, one of which resembled cana. 
Also from semigigas and its crosses with Lamarckiana and gigas the 
following 15-chromosome forms were obtained : lata , cana , pallescens , and 
liquida. Similarly from Oe. biennis semigigas x biennis was obtained albi- 
nervis with 15 chromosomes and three rosettes with small blue-green leaves, 
one of which was shown to have 16 chromosomes. 
Such forms may of course arise from 8 + 8 chromosomes, or particularly 
in cases where the mother plant is triploid, from 9 2 + 70 7 '. That the con¬ 
dition is still an unbalanced one is shown by the male sterility of these 
forms. 
We may now consider briefly the significance of these complicated facts, 
and it is well to point out in the first place a fact which has as yet been 
recognized by very few geneticists, namely, that the great majority of the 
known mutations in the Oenotheras involve altered chromosome numbers, 
chiefly simple trisomic. Indeed almost the only exceptions are (1) brevistylis, 
rubricalyx, and gigas narietta, which involve simple Mendelian differences, 
and (2) rubrinervis , nan ell a, perhaps laevifolia , and a few others which may 
possibly have arisen through crossing-over. Irregular chromosome distri¬ 
butions are then concerned in the great majority of Oenothera mutations. 
It was first pointed out many years ago (Gates, 1908 ) that owing to 
the usually weak attraction between the chromosomes in Oenothera during 
meiosis, irregular chromosome distributions (now spoken of as non-disjunc¬ 
tions) may and do occur, giving rise to pollen grains (and probably also mega¬ 
spores) with eight chromosomes. It was also pointed out that simultaneous 
non-disjunction of two pairs of chromosomes in opposite directions would lead 
to the formation of pollen grains all having seven chromosomes, but two of 
them lacking both members of one pair and two both members of another 
pair. Since then an attempt has often been made, by giving letters to the 
chromosomes, to work out the behaviour of the various mutations on 
a chromosome basis. Since no consistent scheme could be arrived at, the 
results were not published. Miss Lutz ( 1917 ) has made a similar attempt, 
but confesses failure. It is perhaps worth while to view the matter afresh 
in the light of our present knowledge. Let us apply this view first to 
mutations like rubrinervis and nanella. The striking facts in their genetic 
behaviour are (1) that they arise sporadically from Lamar ckiana, (2) that 
they breed true except for occasional aberrant forms, (3) that in crosses 
with Lamar ckiana they give the two parent types in various proportions, 
(4) that when crossed together they give (de Vries, 1913 , p. 214) in F l 
Lamar ckiana and subrobusta (which we have seen is closely related to rubri¬ 
nervis). The Lamar ckiana breeds true, while subrobusta in T 2 splits out 
dwarfs ( rubrinervis , nanella) as a Mendelian recessive, and rubrinervis , both 
of which breed true. Subrobusta continues to split in the same manner in 
later generations. This is shown in diagram form on p. 557. 
