Full text of "Journal of genetics"

See other formats

JOURNAL OF GENETICS

CAMBRIDGE UNIVERSITY PRESS

iLonion: FETTEE LANE, E.G.
C. F. CLAY, Manager

also

H. K. Lewis, Gower Street

and

William Wesley and Son, 28, Essex Street, W.C.

CfUinliurBl) : 100, PRINCES STREET

Berlin: A. ASHER AND CO.

Hetpjia: F. A. BROCKHAUS

i^tto Sorfe : G. P. PUTNAM'S SONS

Bombaa anJ dalculta: MACMILLAN AND CO., Ltd.

JOURNAL OF GENETICS

EDITED BY

W. BATESON, MA., F.R.S.

DIRECTOR OF THE JOHN INNES HORTICULTURAL INSTITUTION
AND

R. C. PUNNETT, MA.

PROFESSOR OF BIOLOGY IN THE UNIVERSITY OF CAMBRIDGE

Volume I. 1910— 1911

Cambridge :

at the University Press

191 1

Cambrtirge :

PRINTED BY JOHN CLAY, M.A.
AT THE UNIVERSITY PRESS

QH
A-51
MJGA-

)lo

CONTENTS.

No. 1 (November, 1910)

PAOE

Frederick Keeble and Miss C. Pellew. White Flowered Varieties

of Primula siiiensis ......... 1

Redcliffe N. Salaman. The Inheritance of Colour and other
Characters in the Potato. (Plates I — XXIX, one coloured,
and 2 Text-Figures) 7

Fkederick Keeble and Miss C. Pellew. The Mode of Inheritance

of Stature and of Time of Flowering in Peas {Pisum sativum) . 47

E. R. Saunders. Studies in the Inheritance of Doubleness in

Flowers. I. Pehinia. (Seven Figures) ..... 57

L. Doncaster and F. H. A. Marshall. The Etiects of one-sided

Ovariotomy on the Sex of the OflFspring ..... 70

No. 2 (March, 1911)

R. P. Gregory. Experiments with Primula sinensis. (Plates

XXX— XXXII and 2 Text^Figures) 73

M. Wheldale. On the Formation of Anthocyanin . . .133

Florence M. Durham. Further Experiments on the Inheritance

of Coat Colour in Mice 159

vi Contents

No. 3 (August, 1911)

PAGE

L. DoNCASTBR. Some Stages in the Spermatogenesis of Abraxas

Grossulariata and its Variety Lacticolor. (Plate XXXIII) . 179

W. Bateson and R. C, Punnett, The Inheritance of the Peculiar

Pigmentation of the Silky Fowl. (4 Text-Figures) . . .185

H. M. Leake. Studies in Indian Cotton. (Plates XXXIV, XXXV,

4 Text-Figures and 2 Diagrams) ...... 205

Redcliffe N. Salaman. Heredity and the Jew, (Plates XXXVI —

XXXIX, and 6 Text-Figures) 273

No. 4 (November, 1911)

W. Bateson and R. C. Punnett. On Gametic Series involving

Reduplication of Certain Terms. (Plate XL and 1 Text-Figure) 293

Edith R. Saunders. Further Experiments on the Inheritance of

" Doubleness " and other Characters in Stocks. {2 Text-Figures) 303

L. DoNCASTER. Note on the Inheritance of Characters in which

Dominance appears to be Influenced by Sex .... 377

Correction. On Plate XXXIV, to face p. 208, /or " Monopodial " read " Sympodial,"
arid for " Sympodial " read "Monopodial."

Volume I NOVEMBER. 1910 No. 1

WHITE FLOWERED VARIETIES OF PRIMULA

SINENSIS.

By FREDERICK KEEBLE,
Professor of Botany, University College, Reading ;

AND Miss C. PELLEW,
Research Student, Botanical Lahoraiory, University College, Reading.

[It was intended that this paper should be published simultaneously
with an extensive memoir by Mr R. P. Gregory on inheritance in
Primula sinensis. Mr Gregory's paper is already in type ; but
owing to its length and to delay incidental to preparation of the
coloured Plates illustrating it, we have been obliged to hold it over
for the next number of the Journal. — Edd.]

White Flowered Varieties. White flowered varieties of Primula
sinensis are of two kinds, one with red or reddish stems (coloured
stems) and the other with green stems. Coloured stemmed whites,
when crossed with a variety with coloured flowers, yield an Fi with
white or tinged white flowers. Green stemmed whites, when similarly
crossed, yield an F^ with coloured flowers. Since the white or tinged
white ^1 plants give rise, on selfing, to white and coloured flowered
plants in the proportion of three white to one coloured, it is inferred
that the coloured stemmed whites carry the factors for colour, but that
pigment formation is inhibited by the presence of a dominant white
factor. Since, also, green stemmed whites give rise, when crossed with
a colour variety, to a coloured -Pi, it is inferred that they lack the
dominant white factor as well as one or more of the colour-factors.

Jonm. of Qen. i 1

2 White Primula sinensis

Thus, of white varieties of Primula sinensis hitherto investigated,
those with coloured stems are "dominant whites," and those with green
stems " recessive whites."

One exception to this rule is already known : the green stemmed,
white variety Pearl having been shown to be a dominant white.

The purpose of this note is to record the existence of what appear
to be exceptions to the rule of dominant white among coloured stemmed,
white varieties.

The evidence is based on the gametic behaviour of Snow King, a
variety which has white flowers and dark red stems.

Plants of Snow King, raised in 1908 from seed obtained from
Messrs Barr, proved true to type, except for an occasional magenta
flaking of the petals of a few plants. The variety again bred true to type
in 1909.

In 1 908, three plants of Snow King were used for crossing with the
following coloured varieties :

Reading Pink (pale pink flowers, green stem).

Crimson King (dark red flowers, reddish stem).

Pink Stellata (pale magenta flowers, reddish stem).

A green stemmed variety with pink flowers a shade deeper than in
Reading Pink, numbered 2 A.

It should be remarked that, in green stemmed, coloured flowered
varieties of P. sinensis, the deeper flower colours of the self-coloured
types are not fully developed. Such plants however carry the factors
for the deep colours ; for, when they are crossed with coloured stemmed
varieties with pale coloured flowers, the deeper shades are fully
developed in the coloured stemmed ofifspring.

The Fi generations, obtained from the crosses between Snow King
and the several plants enumerated above, were as follows : —

Expt. No. Cross Description of f , plants

20-2 Snow King X Crimson King 1 10 tinged white : 9 magenta

52 Beading Pink X Snow King 1 5 ,, ,, : 3 ,,

200 Pink Stellata X Snow King 12 ,, ,, (nearly pure white)

2 A (Green stem x Snow King, flowers pink) 8 pale magenta

A uniform F^ family of whites or tinged whites occurs in only one
of these crosses. In No. 2 a, the F^ consists of coloured flowered plants
and, in Nos. 20*2 and 52, it is composed of tinged whites and coloured
in about equal proportions.

1 The same plant of Snow King was used in crosses 20-2 and 52.

F. Kkeblk and C. Pellew 3

In order to investigate the meaning of these results which are in
disaccord with those obtained hitherto with coloured stemmed whites,
coloured and white tinged plants of the Fi generation were selfed,
and the F, generation examined. The results were as follows: —

F^ from coloured flowered F^ plants.

Experiment No. 20*2, a magenta plant selfed.

^2. Observed 20 coloured : 8 white and flaked white.
Calculated 21 „ 7

^ 1

» <^ >? ••• >j jj n

Experiment No. 52, two magenta plants selfed.

F^. Observed 54 coloured : 19 white and flaked white.

Calculated 55 „ 18 „ „ „

3 1

i> ■•■ » » »

Experiment No. 2 Al, two magenta plants selfed.

F^. Observed 77 coloured : 22 white and flaked white.
Calculated 74 „ 25

F, from tinged white F^ plants.
Experiment No. 20*2, a tinged white selfed.

F^. Observed 29 white and tinged white : 12 coloured.
Calculated 33 „ „ „ 8

» ■*•" >» i> i> " »

Experiment No. 52, two tinged whites selfed.

F^. Observed 63 white and tinged white : 15 coloured.
Calculated 63 „ „ „ 15 „

»» -'•" » »> » " »>

Experiment No. 200'1, a white plant selfed.

Fj. Observed 13 white : 9 coloured.
Calculated 18 „ 4

»» 1" »> " i>

In the F, from coloured plants, we obtain approximately 3 coloured :
1 white, and in the F^ from tinged white sister plants we have approxi-

1-8

4 White Primula sinensis

raately 13 white (and tinged) : 3 coloured. A departure from the 13 : 3
ratio should be noted in Experiment 200*1 . This must be attributed
to the fewness of the F^ plants grown, until more evidence can be
obtained.

It was noticeable that some of the white plants of F2 from white
and coloured Fi, showed a considerable increase of flaking as compared
with that observed in certain plants of Snow King. Among those
flaked, white plants from coloured Fi plants, there occurred one or two
plants bearing flowers with a very faintly tinged ground.

Further investigations will, it is hoped, demonstrate the significance
of these facts.

On the basis of the numbers obtained in F^, we arrive at the
following conclusions : — The plant of Snow King used in Experiment
No. 200, which gives a tinged ^1, is homozygous (TTTT) for the domi-
nant white factor.

That used in Experiment No. 2 A, which gives a magenta ^1, is
homozygous (ww) for the absence of the dominant white factor. Since
the flowers of this plant are white, it lacks a colour factor. That is, its
gametic constitution is cw. Since the stem is red, the loss of colour
factor has regard only to the flower and not the stem.

Writing Snow King cw and plant 2a, Cw,

Fi = Ccw = coloured.

The plant of Snow King used in Experiments Nos. 20*2 and 52
which give both coloured and tinged white in F^, is heterozygous (Ww)
for the dominant white factor. Since the variety as a whole breeds
true to whiteness, the heterozygous (Ww) plants must lack colour
factors. Their gametic constitution is cWw.

Snow King (cWw) x Crimson King or Reading Pink (Cw).

Fi = Cc Ww, white or tinged white and Gcww, coloured.

i^, = 9 CW, 3 cF, 3 C«;, 1 cw. F^ =lGw,2 Ccw, 1 cw.

= 9 white + 3 white + 3 coloured + 1 white 3 coloured : 1 white.

= 13 white : 3 coloured.

In order to investigate further the nature of the factors necessary
for the production of colour in Primula sinensis, plants of recessive
white Snow King were crossed with the recessive white, green stemmed
varieties of Ivy leaf (for a plant of which we are indebted to Mr Bate-
son) and Snow-drift.

F. Keeble and C. Pellew 5

From Ivy leaf x Snow King an /*, was obtained consisting of
4 flaked white on dark red stems, and 1 flaked white on reddish stem.
Snow-drift by Snow King yielded an F^ consisting of 24 magenta
flowered plants with reddish stems. Thus a fully coloured Fy is
obtained as the result of a cross between two white flowered varieties.
The F^ generation from these crosses has not yet been obtained.

Table of Flotoer and Stem colour in F^

stem Flower coloor

White

"■^

Ko. of

Expt
>o.

Dark

and

Pale

plants not

Reddish

red

Magenta

Pink

tinged

pink

flowered CrooH

Fi family from

(20-2-1

—

— \

Snow King

magenta Fj plants

1 -

—

Fi family from ting-

(20-2 -2

—

Crimson King

ed white Fi plant

( —

—

(52-3

—

Fi families

—

from magenta

52 -5

Fi plants

—

i -

—

^ Reading Pink

52-4

—

Fj families

—

8
12

Snow King

from magenta.
F\ plants

[52-6

—

14
2

—

— ^

— —

I -

—

1 J

/2a1

—

— \

Fg families
from 2 magenta -\
Fi plants

2a5

1
9

2^x
[ Snow King

—

— ,

Fs family from f 200-2

—

— I

Pink Stellatax

white Fi plant

I —

—

— j

Snow King

THE INHERITANCE OF COLOUR AND OTHER
CHARACTERS IN THE POTATO.

By REDCLIFFE N. SALAMAN, M.D.

Introduction.

The experiments here described were begun in the spring of 1906
and are still being continued ; the work has been carried on in my
garden at Barley in Hertfordshire. Although the subject material of this
research was my own choice, at the time it was determined on I was
quite ignorant of the very special advantages as well as disadvantages
which the Potato offers for the Mendelian student. To Professor
Bateson and Professor Punnett I owe a debt of gratitude for the
encouragement they have always given me and the time they have so
kindly devoted to examining and criticising my work.

The potato plant as grown domestically in England is a perennial,
that is to say, it is raised from tubers vegetatively year by year. Most
of our varieties bear flowers, but only a very small proportion set seed ;
this peculiarity will be considered more fully later, and has already
been dealt with in detail (9) ^

The potato flower bears five anthers (sometimes six or seven)
arranged in a cone through whose apex projects the stigma. The
anthers dehisce at their distal extremities, the pollen, when there is
any, falling on to the knob-shaped stigma which projects but a short
distance beyond the cone's apex.

When cross fertilizations are made, the flower which is to act as
the female parent is emasculated before the bud is open while both
anthers and stigma are still unripe.

The flowers are borne as a cyme, on axial stalks, each bloom having
a short stem about an inch long, and at a distance of half an inch

^ The nambera in brackets refer to the Bibliography.

8 Colour and other Characters in the Potato

below the base of the flower there occurs a ring of cork. In all
potatoes the flowers have a great tendency to separate at this point
from their stems : the tendency is more marked in those flowers
where the anthers are sterile. If such a flower is used as the female
parent the chances of a successful cross fertilization are somewhat less
good than if the fertilization is made on one with fertile anthers owing
to this habit of separation. In all potato plants, however, when grown
out in the open, successful fertilization, be it " selfing" or "crossing," is
a hazardous undertaking, and I personally do not succeed in getting
more than about "5 % of the individual flowers I handle to set seed.

Once the ovary begins to swell there is little fear of separation
taking place at the cork ring, indeed the stem gradually thickens and
carries the berry late into the autumn.

All my work has been carried on without placing the flowers in
bags. The reasons for not adopting special precautions were that
when bagged the flower invariably drops, that bees and the like never
approach a potato flower though a small fly often lives in the bottom
of the corolla, that the flower is constructed for self-fertilization, and
that the quantity of pollen is so scanty as to render fertilization by the
wind in the highest degree improbable. Each year I have sterilized a
number of flowers and purposely left them unpollinated, in no instance
has any fertilization taken place. In two instances out of some
hundreds so treated the ovaries swelled till they attained a diameter
of 3/16 in., but they contained no seed and dropped.

Although the potato, owing to its scanty pollen, its frequent
sterility, and its delicate flower, is not an ideal subject for Mendelian
research, it does still off"er to the experimentalist one redeeming char-
acter. An individual plant can always be "carried on " by means of its
tubers into the next season's work, and whether it be for the sake of
comparison or for the purposes of further fertilization this property is
of the utmost service.

The Scope of the Observations. Attention has been concentrated
mainly on the heredity of characters of the tubers, for the haulm or
foliage of the potato plant, though variable in habit of growth, size,
shape, texture and colour, does not lend itself readily to this type of
work. The foliage more especially is so variable in different parts of
the same plant, whilst the differences between one type of foliage and
another, however apparent, are so difiicult to define that except in one
instance, which will be considered later in detail, I have not made out
anything sufficiently definite.

R N. Salaman 9

The colour of the stem is always correlated in some degree with
that of the tuber, but whereas one meets with innumerable white-
tubered plants, yet, as far as my experience goes, in all of these some
colour may be found, if not in the stem, then in the shoot which
emerges from the tuber in spring.

Very definite Mendelian segregation of colour in the stem occurs
when the black or deep purple pigment, such as is seen in " CJongo," is
introduced, but in the case of the red- and white-tubered plants the
quality of the pigment being constant, it is the quantity that varies
and that is not readily to be measured. In one family of 100 seedlings
I ascribed values to the colour as seen in the stem. The parent was a
plant with a medium quantity of pigment in the stem. The degrees of
pigmentation in the stems of the seedlings were divided into " strong,"
" medium," and " weak," and the numbers in each class bore to each
other as nearly as possible the relation of 1 strong : 2 medium : 1 weak.

The absence of distinct and definable gradations between the various
degrees of colour, as well as the possible personal bias in the classifica-
tion, is my reason for not publishing the results of the observations on
colour in stem and foliage which were made in every individual plant
during the four years' work covered by this paper.

Observations on the colour of the flowers have been made, but only
in the case of seedlings of the potato known as Lindsay's etuberosum
has anything of interest been observed : a description of the phenomena
in the flowers is given in the section dealing with this peculiar variety.

Observations on the pollen have disclosed some interesting facts in
connection with heredity of sterility and have confirmed East's (4)
observation of the relation between amount and viability of pollens.

The incidence of disease {Phytopthara infestans) has been closely
watched, but only in the case of the Lindsay etuber, q.v., has anything
definite been observed.

The fact that there has been till now no really immune variety to
work with has prevented any headway being made in this direction.

The Material used. All the observations, excepting those dealing
with the peculiar variety already described by Sutton (s), and known
as Lindsay's etuberosum, have been made with ordinary domestic
varieties. The most useful of all the potatoes employed has been
Sutton's " Flourball," which indeed gives the key to the understanding
of them all. The black pigment was introduced by the potato known as
the " Congo," a potato which is of a deep blue-black both within and
without and which is used domestically for salads. One variety which

Colour and other Characters in the Potato

proved of value was a white kidney potato known as " Record." It
was brought out by Messrs King, of Coggeshall, but it has entirely
gone out of cultivation as far as could be ascertained, not only in
England generally, but in my garden also, and my notes of its characters
are unfortunately not very full.

I give here a list of the domestic varieties I have used.

In self and cross fertilization.
A. Flourball (Sutton).
Record (King).
Congo.

Reading Russet.
Red Fir Apple.
Queen of the Valley.
Bohemian Pearl.
Sole's Kidney.
Early Regent.
Prof. Maerker.
S. etuberostim.

For observations on pollen.

Varieties in list A.
Ringleader.
Supreme.
Dutch Cornwall.
Peckover.
The Dean.
Purple Eyes.
Up-to-Date.
Duke of York.

S. commersonii ^
S. tuberosum
S. verrucosum
S. maglia

- species.

Several other varieties were used in class A without success.

Sterility of Anthers. Con tabescence.

Darwin (3), in considering the origin of sterility, describes a con-
dition not uncommonly found amongst plants of various families in
which the anthers are more or less twisted up or aborted and contain
no pollen. Darwin called this condition "contabescence," and described
how it might be propagated by layers, cuttings, etc., and even by seed.

Gaertner first observed the condition and described a similar change
affecting the female organ (6),

Bateson described in the Sweet Pea a similar phenomenon and found
it recessive to fertile anthers (i).

The potato " Record," which possesses no pollen in its anthers, was
crossed by Sutton's " Flourball," which possesses abundant pollen : 20^
of the 32 F^ plants which bore flowers not one of which contained any

1 In 1910 26 of the i^^ plants flowered and they were all sterile.

R N. Salaman 11

pollen. Two individuals of the F^ family were fertilized by a derivative
of " Flourball A," very rich in pollen, and gave rise to 39 plants, 19 of
which bore pollen and 20 bore none : the expectation on the assump-
tion that sterility is dominant being here equality.

In the "Congo" potato the anthers are entirely devoid of pollen,
though they are not usually aborted or crippled. A plant of this
variety was crossed by a " Flourball " seedling, and out of 18 ^^ plants
which flowered, 8 had abundant pollen and 10 had none : here again
the expectation was equality, " Congo " being heterozygous in sterility.
Two F^ plants possessing abundant pollen were selfed, and of 44
plants examined, 41 possessed pollen and 3 possessed but a few grains
of immature pollen. Why these plants should not have borne a fair
quantity of pollen seeing that the F^ parents must have been recessives
and should have bred true, it is not possible to say. All three
examples came out of one family.

A second cross with " Congo," viz. by " Reading Russet," gave only
a small F^ family, three plants bearing flowers, two containing pollen,
and one none.

Similar results were obtained in the cross " Red Fir Apple " and
" Reading Russet," F^ being part pollen producers, part sterile, whilst
jP*, from the pollen bearing F^, gave 9' plants all pollen producers.

The flower of the " Red Fir Apple " is heliotrope in colour and the
anthers are aborted.

" Queen of the Valley " has heliotrope flowers with sterile anthers.
Crossed by " Flourball " one plant gave a series of F^ plants of which
some bore pollen and others none, although exact notes as to their
characters in this family were not taken. One of the F^ plants was
crossed by a " Bohemian Pearl " seedling, and gave rise to a long line
of pollen producers.

The heredity of male sterility in the potato is obviously the converse
of that described by Bateson in the Sweet Pea, for the condition here
is distinctly dominant. Bateson found it partially coupled mth green
axils in certain families. In the case of the potato, the only evidence
of sterility being coupled with any other character was of a negative
sort. Working with a large number of established varieties as well as
with those plants which arose in the course of this work, I never found
a plant possessing pale heliotrope flowers that had other than sterile
and contabescent anthers, whilst those that were further tested proved

^ In 1910 22 more F- plants flowered and all possessed pollen in the anthers.

12 Colour and other Characters in the Potato

to be heterozygous as regards sterility of anthers. No connection was
observed between the condition of the male and female organs.

The presence of pollen in the anther being as we have seen a
recessive character, it is of some interest to note how it behaves in
selfed families. Unfortunately these pollen observations were not
begun till 1909, although the breeding experiments began in 1906.
Still a good deal of information may be extracted from the early notes.

Thus, in 1906, a red-tubered seedling derived from a "Flourball"
plant in 1904, was "selfed," and gave rise to a large number of
seedlings. One white-tubered plant {D) was reserved. From this a
further generation was bred, and from this again another, so that in
this case the family has been handed through five generations, and in
all the anthers have had abundant pollen though the quality of the
pollen was bad.

Two other lines, A and 0, derived from " Flourball," have been
bred through three and four generations respectively, and the recessive
character, viz. presence of pollen in the anther, has remained true.

The occurrence of spontaneous sterility, due to absence of pollen,
has already been mentioned as having taken place in the F"^ generation
of the family " Congo " x " Flourball " ; it has also been observed in
some other families where it was unexpected, but in all these cases it
has occurred in normal and not deformed or strictly " contabescent "
anthers. It is possible that " contabescence " is not a simple character
but that absence of pollen and deformity of anther are due to separate
factors between which exists an intimate linking.

The relations between quality and quantity of pollen and the shape
of pollen in varieties and species of Solanum are discussed elsewhere (9).

Heeedity of Characters in the Haulm.

The difficulties in relation to haulm characters have already been
adverted to ; although to experts constantly reviewing crops of well-
grown varieties it becomes comparatively easy to diagnose a variety by
the general appearance of the foliage, and by inspection to designate
at once such and such a potato as an " Up-to-Date " variety, or a
" Ringleader " type, and so forth, yet if one closely compares any two
foliages, taking corresponding specimens from various parts of the plant,
it will be found very difficult to describe any constant differentiating
character between any two varieties; there are differences no doubt,

R. N. Salaman 13

but they do not admit of such definition as to fit them for Mendelian
analysis.

The cross of " Red Fir Apple " and " Reading Russet " was made in
1906 for the purpose of tuber colour observations, and in 1909 a large
family of some 120 individuals of F^ plants were raised.

The " Red Fir Apple " has a somewhat distinctive foliage, the
leaves are relatively small, ovate with sharp apices, peculiarly soft and
silky to the touch, and, in addition, have a character which entirely
distinguishes them from " Reading Russet " and most other varieties.
The leaf has a peculiar twist in its axis, this twist being seen in all the
upper leaves and often down to the lowest when the plant is 18 inches
high or more.

The condition of leaf twist here in question must be clearly distin-
guished from that which occurs as a pathological condition in many
varieties ; in such cases the plants are dwarfed, the stems shrunken,
the axes of the branches very shortened, and the leaves on them
crowded together. The individual leaves also are much twisted, crenate
and small.

In the "Red Fir Apple" the twist is less violent, it is not associated
with crenation, and the plants are thoroughly healthy, vigorous and of
good size.

" Reading Russet " possesses a much coarser foliage, the leaves are
big, broad, blunt, flat, smooth, hard and coarse ; the green colour is of
a deeper shade than in " Red Fir Apple."

The four F^ plants which were examined were intermediate as regards
shape and texture of foliage, but resembled " Red Fir Apple " shape
rather than " Reading Russet " ; no twist in the leaf axis was observed.

In F- an analysis was made of the plant's foliage characters as seen
in the table below.

The characters taken are all leaf ones.

" Reading Russet " shape. Broad and blunt leaf.

„ „ texture. Few stiff hairs, glazed surface to leaf.

" Red Fir Apple " shape. Ovate, sharp apex to leaf

„ „ texture. Soft and silky.

Twist. Twist in the axis of the leaf.

Intermediate shape. Leaf shape neither " Reading Russet "

nor " Fir Apple " in type, but re-
sembling more closely the latter.
„ texture. Softer than " Reading Russet " and

harder thau " Fir Apple."

14 Colour and other Characters in the Potato

Foliage of F"^ Generation.

" Reading Russet " texture. " Reading Russet " shape 10

„ „ „ Intermediate shape 1

Intermediate texture. " Reading Russet " shape 4

„ „ Intermediate shape 40

" Fir Apple " shape 12

" Fir Apple " texture. Intermediate shape 9

„ " Fir Apple " shape 42

Total number of J^2 plants .... 118
Twist in leaf 27

In considering these figures it must be remembered that it is a
matter not only of considerable difficulty to classify the living plants
according to the shape and texture of their leaves, but that the
personal element is paramount in such a classification. More particu-
larly do such remarks apply to the consideration of texture and to the
intermediate forms. Certain features, however, are readily and unmis-
takably recognized ; these are the twist in the axis of the leaf and to a
lesser degree " Reading Russet " shape.

The intermediate form of leaf is much more like the " Fir Apple "
leaf than the " Reading Russet," and the former may therefore be con-
sidered dominant, whilst the twist in its leaf is recessive.

If the " Reading Russet " shape and texture are recessive, then it
should occur combined in the F^ family in the ratio of 1 : 15 and here
it is 1 : 12.

The twist in the leaf occurred 27 times out of 118, that is practically
in the ratio of 1 : 3, and it was associated 23 times with the " Red Fir
Apple" shape, the remaining four having intermediate shapes and none
showing " Reading Russet " shape.

Allowing again for the difficulty in distinguishing the intermediate
form from " Fir Apple " shape and texture, it would seem to be a fact
that this peculiar twist in the leaf is definitely linked up with the
" Fir Apple " characters of shape and texture. None of the eleven
plants possessing " Reading Russet " shape showed the slightest sign of
a twist. The same consideration leads one to believe that " Reading
Russet" texture is coupled up with "Reading Russet" shape; ten out
of eleven times it is recorded as being so linked whilst the eleventh

R. N. Salaman 15

time " Reading Russet " texture was united to intermediate shape,
which might possibly be an error of observation.

These observations demonstrate at least that such fleeting and
difficult characters as leaf shape and texture in the potato segregate
in the sexual generation.

This year^ a fresh F^ family of this cross is being raised, and close
attention will be paid to their foliage character.

The Shape of the Tubers.

No character seemed at first sight more elusive and less likely of
solution in respect to its heredity than that of shape. Whenever I
spoke to experts I was told that from the best " kidney " types you
could pick out "rounds," and that exhibitors had won prizes both for
" rounds " and for " kidneys " from one and the same potato.

East (5) notes four cases where originally " long " tubered varieties
produced as bud sports rounded tubers; in two cases these "round"
tubers reproduced themselves vegetatively true to " roundness," while
the other two relapsed in the following season.

The oval varieties he notes as producing on single plants entire
crops of very elongated tubers, which however did not grow true in
subsequent years.

My observations would lead me to think that these bud sports in
" kidney " and oval potatoes are quite common and are to be explained
by their heterozygous composition as regards " roundness."

A frequent cause of trouble in dealing with the shapes of tubers is
the nomenclature. The terms used to describe the diflferent shapes are
sufficient for the purpose of the gardener, but they connote no scientific
accuracy.

Where the cylindrical potato ends and the kidney begins, where
the latter ceases and the "pebble" starts, and where both merge
into the round is a problem which it would be hopeless to attempt to
solve by the mere classification of tubers.

It is only by the isolation of a type and its fixation as pure when
bred sexually that the problem can be solved.

In describing the shape of a potato, two points can be regarded as

1 In 1910 out of 71 F2 seedlings on Ang. 3rd 6 showed the "Fir Apple" twist, on
Aug. 23rd 14 had developed it.

16 Colour and other Characters in the Potato

fixed, viz. the point from which the tubers grow out from the stolon,
and the most distal point from that, which in 19 out of 20 cases coin-
cides with the central of the crown of eyes at the distal end. It is
from this eye that the earliest and strongest shoot grows out. The
line between these two points is the long axis, the breadth and depth
are respectively the greatest measurements in each direction measured
at right angles to the long axis and to each other. Adopting the
conventional terms for potato shapes, the names long, kidney, pebble,
and round appear to have the following meanings : —

A long potato is one in which the long axis is between \^ and 2|
times the greatest breadth, and the depth is equal to the breadth.
The ends are either blunt, as in the " Congo," giving the tuber a
cylindrical appearance, or they are pointed as in B, Plate XXIV.

A kidney potato is one in which the length is usually between
1^ times and twice the breadth, and the depth is considerably less than
the breadth, giving the tubers a flattened appearance which is charac-
teristic. The measurements of three specimens, unselected, of well-
known " kidneys " are : —

"Myatt's Ashleaf " :

Length.
Inches

Breadth.
Inches

Depth.
Incnes

Ratio

(1)
(2)
(3)

2,
3
2,

12/16
4/16

1,
1,
1,

9/16
7/16
7/16

1, 3/16
1, 3/16
1, 2/16

=44 : 25 :
= 48:23:
= 36:23:

;19

;18

"Sutton's Ideal":

(1)

(2)
(3)

2,
2,

7/16
5/16
4/16

1,
1,

8/16

10/16

7/16

1, 4/16
1, 4/16
1, 4/16

= 39 : 24 ;
= 37 : 26 :
= 36:23

:20
;20
:20

"Table Talk":

(1)
(2)
(3)

3,
3
3,

1/16
1/16

1,
2
1,

14/16
15/16

1, 6/16
1, 9/16
1, 8/16

= 49 : 30 :
= 48 : 32 :
= 49 : 31 :

;22
:25

:24

"Sir John Llewellyn'

> .

(1)
(2)
(3)

3
2,
2,

13/16
11/16

1,
1,
1,

10/16
10/16
13/16

1, 2/16
1, 4/16
1, 7/16

= 48 : 26
= 45 : 26
= 43 : 29

: 18
:20
:23

The Lapstone Potato is a bluntly elliptical or oval potato which is
much broader than it is deep.

The Pebble Shape. This term includes a vast number of rather
irregularly shaped tubers — tubers for the most part obtusely elliptical
and almost as broad as they are long.

R. N. Salaman 17

Below are some typical specimens : —
"Beading Basset," see Plate XXI.

Length

Breadth

Depth

RAtio

(1)

2, 6/16

1, 15/16

1, 7/16

= 38 :31

(2)

1, 15/16

1, 12/16

1, 3/16

= 31 : 28

(3)

1, 15/16

1, 13/16

1, 8/16

= 31 :29

"Flourball,"

see Plate I.

(1)

1, 15/16

2, 1/16

1, 8/16

= 31 : 33

.24

(2)

2, 3/16

2, 9/16

1, 13/16

= 35 : 41

Round Potatoes. The tubers are practically globular, as in " Wind-
sor Castle."

An examination of these different descriptions is enough, almost in
itself, to convince one of their artificiality, but when one comes to close
quarters with them by breeding various pure lines and by crossing, one
is soon convinced of the fact.

If Plate I, seedlings of " Flourball," be now examined, it will be
seen that it is easy to pick out^

Longs Nos. 14, 48, 135.
Kidneys „ 21, 87, 88, 123.
Pebbles „ 74, 90, 91, 154, 179;

but a close inspection shows a number of tubers which might be
described as round, but which are not globular. They are short, and
as deep as they are wide, such as Nos. 40, 89, 92, 112, 132, 138, 155,
156, 162, 185—10 individuals out of a total of 43.

If now we turn to Plates II, III, IV, V we shall find a family of
100 individuals all bred from one of these peculiarly shaped tubers (A).
The whole family present a striking uniformity of appearance and
similarity to the parent. Exceptions, however, there are, and they
are figured in full in Plates IV and V.

Turning to these plates we see photographed all the available
tubers from each of these individual plants, and it will be at once seen
that each individual plant in Plate IV contains striking examples of
this " round " type amongst its tubers.

1 It should be said that the representatives of the individual plants here shown are
when there are ovals and others more resembhng "rounds" present on the same root,
always the oval. The bias in favour of the "longs" as against the "rounds" has been
purposely made in the composition of all the plates, in order that the recessive "round,"
when present, shall be free from the suggestion that it is only a variant form of the
dominant "long." If therefore the effect to the eye be less convincing the deductions
that are drawn rest on a firmer basis.

Jonm. of Gen. i 8

18 Colour and other Characters in the Potato

On Plate V, Nos. 67, 87, 91, 94, only further illustrate the fact that
though certain tubers of a plant in this family may be more or less
oval, yet other tubers on the same plant will be found to be of this
peculiar " round " type.

One exception, however, stands out, and this is No. 100, which is
definitely unlike the parent type and all its 100 other sister plants.

It is possible that it arose from a stray tuber and does not belong
to this series at all — a view that has some plausibility, seeing that two
years before " Flourball " seedlings were grown on this ground. Efforts
are being made this year (1910) to obtain selfed seed from this plant.

On Plate VI a further illustration {Q family) of this "round" type
of potato is seen; it arose from a "Flourball" plant, but not the same
one as the line A.

Seed from four of these plants has been saved and a batch of seed-
lings of G* were planted in October 1909 and hurried forward; on
April 26, 1910, they were examined and all the seedlings bore tubers,
varying from |^ to f in. diameter, true " rounds " in shape. Those
of the Q^ seedlings which have formed tubers have also developed
typically "round" ones\

It thus appears that there is a certain definite type of "round"
potato that can be extracted from Sutton's " Flourball," and which can
be bred sexually pure through at least two generations after having
been isolated.

Before following further the evidence as regards the heredity of this
type and its behaviour when crossed with other types, it will be best to
discuss more fully its shape and variations.

The tuber shape, which is under consideration and which for the
purposes of my work I have called "round," is to be found white, or
coloured as red or black.

No relation has in the course of this research been shown to exist
between shape of any kind and the pigmentation either of haulm or
tubers.

The "round" tubers may be furnished either with "deep" or "fleet"
eyes. It will be shown later that depth of the eye is itself a character
inherited on Mendelian lines, and my experiments fail to show any
relationship between depth of eye and shape of tuber. The size of the
tuber is of course variable, but I have not found, however one may have

1 Aug. 29, 1910. Although the G family has not been completely harvested there is
evidence that the G^ family consists of three "longs" to one "round," and that the G^
and G* families are pure to "roundness."

R. N. Salaman 10

bred it, this type of " round " potato assuming large proportions ; few
examples with a diameter over 2 inches occur, although oval and
kidney from the same original parent stocks may be of large size and
weight.

A typical specimen of this " round " type is represented by the first
tuber of G*, Plate VI. The tuber is apple-shaped, its upper or proximal
end as well as its distal or crown end is depressed, and the height is
less than either its width or its depth. The actual dimensions are : —

Length Breadth Depth Ratio

1. 5/16 2, 2/16 1, 1/16 =21 : 34 : 17

One of the tubers of the parent A has the following measurements: —

Length Breadth Depth Batlo

1, 5/16 2, 2/16 1, 1/16 =21 : 34 : 17

The most characteristic feature is the stumpiness of the tuber in
relation to its breadth.

Potatoes are raised commercially by the vegetative method, thus a
crop of " Magnum Bonums " raised to-day should be regarded as merely
an offshoot — a cutting so to speak — of a seedling raised some time
before the year 1876. In other words the tens of thousands of tons
which in the past 34 years have been grown of this stock are for
scientific purposes merely replicas of a particular tuber of a particular
individual, and hence the continuity through the intervening years of
the variety's characters. Tubers that are grown by this vegetative
means, within limits, reproduce themselves in their original shape more
or less exactly, though I think, and hope to prove, that the degree to
which a potato reproduces its shape vegetatively depends in large
measure on its gametic constitution.

It may therefore be confidently expected that whilst a crop raised
from a typical "round" such as .4 by vegetative means will remain
perfectly true to type (and this indeed has been proved in the case of
A itself, by growing it in 1908 and 1909), a crop raised say from the
fifth tuber of No. 67, Plate V, might produce tubers more or less
uniform and unlike the type A. A family raised by seed from any
of the individuals, however aberrant in shape, will probably produce
a set of seedlings at least as uniform as the family A itself.

The variation of this "round" type, if grown vegetatively, so far as
my experience goes, is very slight or indeed none at all. The variations
of the type as raised sexually by seed are slight but definite, being

8—2

20 Colour mul other Characters in the Potato

towards greater length and approaching the pebble shape. In diagram
the type and extreme variation may be represented as below : —

Fig, 1, These drawings are tracings of sagittal sections of potatoes — the long and trans-
verse axes are shown — the depth cannot be shown.

Height and breadth are here represented, the depth being relatively
great.

The " round " type is not a potato that recommends itself for its
beauty or its economic qualities as regards shape ; its merit is derived
from the fact that there is very good reason to regard it as a gameti-
cally pure type, and that " roundness " in the sense in which it has
been used here is a simple Mendelian character. The further evidence
in support of this thesis will appear as we proceed to discuss other
shapes.

A seedling of "Flourball" was selfed in 1906, and in 1907 a large
number of seedlings were raised from it, one only of which was again
selfed in 1907. The plant was carried forward by tubers to 1908, 1909
and 1910. In both 1907 and 1908 it produced seed, but in these two
years only four plants came to maturity, and they produced the tubers
numbered in Plate VII, D\ D\ 1908, D^ and D\ 1909. The seedlings
from 1909 seed have not yet formed their tubers.

The tubers of plant D are quite unlike the " rounds " of the A
family, they are oval and more or less kidney-shaped. The offspring
of these, only four in number (excluding the seedlings now growing),
comprise distinct types.

R N. Salaman 21

D*, 1908, a long pyriform tuber.

D*, 1909, cylindrical tubers tending to kidney shape.

/>, 1908, oval or blunt kidney with a sister tuber nearer circular.

jj , 1 juy ), ,, ,, »

The numbers in this case are all too small to draw precise deduc-
tions; all that can be said is that D does not represent a fixed type,
that, on selfing, it gives both longs and ovals.

In 1908 this same D was crossed by A, and on Plate VIII the family
is shown, or rather two families, because two D plants (D' and D^) both
grown from tubers of the original D of 1907 were fertilized by pollen
of A.

A glance at the plate is enough to show that one has here two
types of tubers, the " round " that we have already discussed on the one
hand, and a series of ovals and kidneys on [the other. The "rounds"
are:

Nos. 3, 4, 5, 8, 13, 14, 15, 16, 18, 19.

3, 6, 7, 8, 10, 12, 14, 18, 19, 20, 21, 22, 28.

That is, 10 out of 19 in the first family, and 13 out of 30 in the
second family. Total, 23 out of 49.

One has, in other words, "rounds" and not "rounds" in practically
equal numbers; and it must be remembered that one counts here only
those as " rounds " which come well up to the standard already given
for a typical " round " such as either A, G^ or G'.

The result of this cross admits of a direct Mendelian interpretation,
for inasmuch as A is pure to " roundness," D must be heterozygous in
that character — a fact which was already strongly indicated before.
And the " non-rounds " must be all heterozygous in shape. If now one
examines more closely the " non-rounds," one sees that they are made
up of good kidneys such as Nos. 1 {D^ x il), and 1, 4, 11 and 26 of
(D^xA); of cylindricals, such as 5 and 23 (D* x J.), while the
remainder are ovals and pebbles difficult to place, but which include
among themselves abundant examples of the same shape as the
parent D.

The experiment therefore as portrayed in Plate VIII is capable of
being interpreted as meaning, not only that an oval " pebble " such as
shape D is heterozygous as to " roundness," but that a true kidney and a
true cylindrical may also be heterozygous in the same degree. Further,
if "roundness" (i.e. shortness of axis) is the one allelomorph here in
action, then " non-roundness " or length is the other. Later evidence

22 Colour and other Characters in the Potato

will be given proving that there is a tuber shape true to length, but
before bringing this evidence forward it will be necessary to discuss a
little further the nature of the kidney and the shapes which are
heterozygous.

Plate X shows a family derived from the cross of H^, a kidney
whose origin will be described later, and the typical "round" A. The
" rounds " can be picked out most readily.

The typical " rounds " are :

Nos. 4, 6, 7, 16, 17, 19, 22, 25, 26, 27, 29, 30, 34, 35, 36,
38, 39, 40, 42, 45, 49,

i.e. 21 out of 44, practically half.

A kidney potato of so typical a shape as H^ is therefore heterozygous
in shape, and length, and must clearly be dominant to " roundness."
Excellent specimens of kidneys occur in the family, and they must also
be heterozygous.

It is interesting to note that No. 46 is more or less cylindrical,
and that it is heterozygous and probably a merely variant form of
kidney.

The hybrid nature, in regard to shape, of the kidney may be regarded
as settled, that of the pebble follows as a necessity, but we have in
support two sets of crosses.

A pebble-tubered plant iT" was crossed by the same " round " A
as has been used before (see Plate XI). H^'^ is a typical pebble tuber
and another of the same root-crop can be seen on Plate IX. The family,
consisting of 47 individuals, is seen at once to break up into two types,
the " round " and the ovals of different degrees.

The " rounds " :

Nos. 1, 2, 3, 4, 10, 11, 13, 13a, 15, 17, 18, 19, 26a, 29,
31, 32, 33, 34, 40, 46, 48, 49.

22 out of 47 are all typical.

Emerging from this union of pebble and "round" occur really
good kidney tubers such as 26, 38 and 41, as good or better than
those produced in the family H^ x A, where the parent was a typical
kidney.

The next cross, and perhaps the most convincing, is represented
in Plate IX. It was made between a kidney potato, " Record " on the
one hand, and the pebble-shaped " Flourball " on the other. The
ofTspring number 32, of which Nos. 12, 13, 18, 21, 24, 25, 26, 30 are
all typical "rounds"; i.e. 8 out of 32, or 1 : 4, the expected proportion

R. N. Salaman 23

if both the kidney and tlie pebble-shaped parent are heterozygous
as regards shape, i.e. " length," and amongst the dominants some are
excellent kidneys, others pebbles. No. 3 is interesting because it
shows on one and the same root a cylindrical potato and a pebble, a
form which has just been shown to be heterozygous.

The arguments and the evidence in support of them, as to the
heredity of the tuber shapes have, so far, all turned on the fact that
there exists a variety of " round " potato which is recessive and breeds
true; at the same time all examples that have been so far brought
forward contain directly "Flourball" blood. It might therefore be
supposed that the whole structure of my contentions rest on this
keystone — this " Flourball " derivative — and that if this latter be
removed the ai-gument and deductions would fall to the ground.
It becomes necessary, therefore, at this stage to describe an experi-
ment entirely free from such an objection, at least as far as I am
aware. A cross was made in 1906 between "Red Fir Apple" and
" Reading Russet." " Reading Russet " is a pebble-shaped potato
and " Red Fir Apple " a long cylindrical. F^ was not examined
critically for shape; the note as to the 117 young seedlings raised
in 1907 is that about one-quarter bore " round " tubers, of these only
nine survived, and only five of them were reared in 1909. Four indi-
viduals are shown in Plate XXI, and the fifth one, which was omitted,
was a long-shaped tuber. On the whole the evidence is rather in
favour of F^ being a mixture of " longs " and " rounds " in the propor-
tion of 3 : 1, but of the F^ "rounds" we have no examples. The F^
generation, however, is represented by 120 individuals contained in
the two families Z^<^' and D^*\ both derived from the selfing of a
kidney-shaped F^ plant.

The first family, D^*\ consists of 60 individuals; of these 52 are
represented in Plate XXII, and of the eight missing, five were long and
three " round." When the plate is examined, and still more the actual
individuals, the " rounds," such as we have already become accustomed
to, are to be found at once, and the following typical examples are
seen, Nos. 1, 2, 22, 35, 37, 46, 47, 49, 61, 63 and 64, which in addi-
tion to the three not figured, makes the total of 14 out of 60 or
nearly 1 : 3.

The second family, X'<*>, Plate XXIII, affords some very striking
examples of typical "rounds" such as Nos. 6, 47, 52. The family
contains 59 tuber-bearing individuals, and of these Nos. 6, 10, 17,
19. 22, 24, 29, 30, 33, 40, 47, 52, 54, 61 are typical "rounds," i.e.
14 out of 59 or 1 : 3.

24 Colour and other Characters in the Potato

In the two families containing 119 tuber-bearing individuals, 29
are "round," that is 1 in 3, as would be expected in an F^ family
from a heterozygous parent in which "roundness" was recessive.

It remains now to consider the evidence bearing on the existence
and nature of the dominant shape in its pure form. So far, it has
been shown that length of tuber is dominant and that the degree
of dominance is variable, i.e. the hybrid form is not constant, the
heterozygous tubers varying from a long kidney to an ovoid. On
Plates XXII and XXIII, amongst the long tubers are undoubtedly
pure dominants, but which exactly they are, and how to distinguish
them from the impure dominants with certainty nothing but breeding
experiments could determine.

It is, however, significant that by selecting those individuals whose
tubers were the most uniformly long, it was found that out of the 119
members of the L family already described there were 34, or a little
more than one-quarter, that could be picked out as being probably
pure in respect to length.

Fortunately better evidence is to hand in respect to individuals
homozygous in the character of length.

A potato, called "Sole's Kidney," yielded abundant seed in 1906,
in 1907 several hundred seedlings were planted^ and they all came
true to type, viz. a long attenuated kidney, see Plate XXVI. One of
these seeded and 50 seedlings were raised in 1909, and every one
of these were long kidney form, see Plate XXVI. It would seem,
therefore, that this potato G, " Sole's Kidney," is a pure dominant as
regards length.

Another kidney, "Bohemian Pearl," was sown in 1907 and a very
large number of seedlings (family B) raised ; these were not examined
very critically in respect to size and shape, but were noted as being
uniformly long and pyriform : one selfed naturally, and of the five
seedlings raised three bore long tubers, and two bore oval tubers,
Plate XXV. These ovals are distinctly flattened and are not " rounds."
They have been grown in 1909 and have retained their shape. Had
there been any appreciable number of oval or "round" tubers in the
first batch of 300 seedlings raised in 1907 it would undoubtedly have
been noted ; on the contrary, my own and my gardener's impression is
that nothing but "longs" occurred. There is in my mind but very
little doubt that the stock B is pure to length. Efforts are being made
to self the oval tubered plants this season.

1 I was presented with several hundred of the seed of both these stocks by the Manager
of the Cambridge University Farm.

R. N. Salaman 25

In 1908 a cross was effected between a pebble-shaped tuber {Af,
Plate XXIV) and a seedling of the family B carried on by tuber from
1907 ^ The issue of this union forms a striking example of the effect
of crossing a heterozygous by a dominant long. The whole family of
39 individuals is without exception long or oval, and includes the most
elegant kidney and one or two cylindricals, see Plate XXIV.

In three experiments cylindrical potatoes were employed as the
female parent. In the first " Red Fir Apple," a cylindrical, was crossed
by " Reading Russet." There is good reason to believe that the F^
family really consisted of three " longs " and one " round," though the
small number of survivors, viz. 11 in the first season, does not assist
one to any definite conclusion. Those of the F^ family which survived
1909 are shown on Plate XXI. " Red Fir Apple," though long and
cylindrical, is therefore in all probability heterozygous as regards
length. It is of interest that, since it has been cultivated in my
garden, it has become shorter and broader and less cylindrical; on
the other hand "Congo," which was used in the second and third
experiment, maintains its truly cylindrical shape. Plates XII and
XXV.

In the second experiment "Congo" was crossed by a "Flourball"
seedling of 1906. The "Congo" tubers are typically cylindrical,
the seedling " Flourball " was not especially described^ but the F^
series, see Plate XXIX, consisting of 29 individuals, all of which
bore kidney-shaped tubers, is evidence that the " Flourball " seedling's
parent must have been "round" and that "Congo" must be a pure
dominant ; for if neither of these suppositions are true, then we should
have expected pure "rounds," which are conspicuously absent, or if
the " Flourball " seedlings were pebble or heterozygous in shape, then
half of the K seedling family should be pure " longs," which they are
not. F^ families were raised from K^ and if*, both elongated and
more or less kidney-shaped. The following proportion of " rounds "
and " longs " occurred

Bounds

Longs

Family K«

210

Family K^

78 279

* The B line planted in 1908 from the pollen of which this cross was made, was
grown from long tubers arising both from the plant which gave the seed ball in 1908
and from its sister plants, sown indiscriminately.

* The absence of a description of shape implies that it was "round" or "pebble"
shaped and not markedly distinct from the parent " Flourball."

26 Colour and other Characters in the Potato

i.e. 1 : 3'6. The families are illustrated in Plates XIII, XIV, XV,
XVI, XVII, XVIII, XIX, XX.

In the third experiment "Congo" was crossed by " Reading Russet."
Only four F^ plants survived, and the tubers of these, Plate XII, are
elongated, but here again the numbers are not large enough to draw
conclusions from.

The dominant character of length in the tubers has been isolated
or identified in the potato G, and is represented by a very elongated
kidney ; in B, where it is more pyriform ; and in " Congo," where the
ends of the tubers are blunted and the tuber has a cylindrical
appearance.

It is not improbable, as was suggested earlier, that the allelomorphic
pair to the character manifested in the " round " potato is length of
axis, and that the kidney and cylindrical shapes, though inseparable
with respect to length, are dependent on other factors governing shape
besides that governing the length of the main axis.

The dominance of the long potato tuber over the short is analogous
to the dominance of the giant over the dwarf plant, as Mendel showed
in the Pea Family. This dominance probably rests on the same ana-
tomical basis, viz. the respective length and number of internodes
involved. Tubers are borne on underground stems, called stolons, and
the eyes may be regarded as buds or nodes, so that the number of eyes
present may represent the number of internodes condensed into the
length of a tuber. A study of the tubers from this point of view is
not yet complete, but it is quite clear that as a general rule the
" round," i.e. short axis potatoes, have less eyes than the long axis ones,
i.e. they represent fewer internodal lengths.

It has already been shown that the dominance of length is not
equal in degree : sometimes the heterozygote is of the most attenuated
form, but more often an intermediate shape is assumed varying from
kidney to pebble and oval. The ordinary kidney of fair breadth is
probably always an heterozygote.

The Variations in the Shape of Tubers. The amount of variation
has already been indicated in the case of the " round " potato ; in the
"long" it is rather less. If "(7" and "Congo" be taken as pure "longs,"
then, accepting the typical well-grown tuber of each sort, it is apparent
that they are as to their proportion between length and breadth much
the same, and the form is fairly uniform.

By far the greatest variation in shape, both amongst the indi-
vidual members of the same family and the several tubers of the

R. N. Salaman

same iDdividual, is met with in the case of the heterozygous
variety.

The examples of heterozygous potatoes which have been tested,
viz. " Flourball " D\ H\ H'\ K*, K* and L\ varying as they do from
kidney to pebble, testify to this.

The degree of variation in the shape of tubers of some given sort
is in itself very variable, but I think it would be acknowledged that
the kidney types vary most. A striking example of this is shown on
Plate XXVIII, reproduced by permission of Messrs Sutton, where a
kidney potato, " Superlative," is photographed in the clamp, and whilst
the majority of the tubers are kidneys, a large percentage are best
described as pebbles.

The variety H^, Plate X, so clearly demonstrated to be hetero-
zygous, is a remarkably uniform kidney shape, but out of less than
half-a-bushel it is possible to pick out potatoes varying from a very
long to an obtuse ellipse. Fig. 2.

Fig. 2. These drawings are tracings of sagittal sections of potatoes of the individoal H'.
The long and transverse axes are shown. The depth is less than the transverse
diameter.

The Depth of the Eye.

The potato tuber has scattered on its surface buds from which grow
the shoots ; the buds are known as " eyes."

The potato eye consists essentially of two parts, a central spot or
shoot, and an overhanging ridge or brow which is curved, and whose
concavity always points downwards or distally.

The eye is recognized to occur in two forms and is known as either

28 Colour and other Characters in the Potato

" shallow " or " deep," The " shallow " eye is a superficial eye, i.e. the
central growing point is not depressed but is level with the general
surface of the tuber and the brow is but very slightly marked.

Typically " deep " eyes are those of " Congo " and most of the
family K (" Congo " x " Flourball " seedling) and A^^, whilst typically
" shallow " eyes are seen in A''' ; H' x A, Nos. 5, 37, 41. The " shallow"
eye is a distinctive and an easily recognized feature. Briefly the
"deep" eye is dominant to the "shallow," and the heterozygous "deep"
eye is never quite so " deep " as the typically " deep " one. In " Flour-
ball " the eye is " deep " but not remarkably so ; of its seedlings 14
out of 43 were definitely "shallow." In the family A, of 98 seedlings
21 were " shallow," and A the parent may be regarded as having the
standard impure " deep " eye.

The D^ X A families contain 16 "shallow"- and 33 "deep "-eyed
individuals.

The H^ X A families contain 22 "shallow"- and 71 "deep "-eyed.
K^ is a further example of an impure dominant " deep "-eyed potato.
Of the 73 seedlings of this family 23 are " shallow " and 51 " deep."

Two F^ families were raised from the cross of " Red Fir Apple " x
" Reading Russet." These two families differ a little in respect to
the eyes. Both were raised respectively from sister tubers of the indi-
vidual F^ plant (L^). Both parent plants grown from these tubers had
"shallow" eyes, one family, Z^<^', consists of 54 individuals, all of which
carry " shallow "-eyed tubers. In the other family, Z^'^', Plate XXIII,
out of 55 individuals 5 (Nos. 4, 15, 51, 52, 59) must be described as
medium, i.e. the eye is distinctly depressed and the brow is evident,
though not heavily developed. The only other " shallow "-eyed potato
that was selfed was " Bohemian Pearl," all the individual plants which
have arisen from it that have come under my notice are " shallow "-
eyed. Of the first generation there were some hundreds, of the second
only five.

If all the families arising out of matings of impure dominant eyes
be put together, we obtain the following :

ShaUow

Deep

"Flourbair

seedling selfed

2)1 X 4

H^ X A

Hio X A

... ... ...

£9 ...

... ... ...

Total ... 91 to 265

This is almost exactly 1 : 3.

R N. Salaman 29

K* is an example of a pure "deep "-eyed potato; all the 284
seedlings of which are " deep "-eyed.

This family, K*, further illustrates a curious phenomenon. Certain
individuals, such as iT"*', Nos. 28, 84 and 95, appear at first sight to be
" shallow "-eyed. When, however, they are examined with their sister
tubers from the same plant, it will be seen that the " shallowness " is
only present at those points where an outgrowth or protuberation is
taking place : elsewhere in the same tuber or on its sisters, the eyes are
"deep" iT'*^. No. 28 is apparently "shallow," but here also outgrowths
are just beginning. A true " shallow "-eyed potato is "shallow" in
every tuber of the plant and a true " deep " is equally " deep " in every
tuber. The heterozygote is more variable and, though " deepness " is
dominant, the eye is often shallower than in the tubers of a pure
dominant "deep" eye.

The potato "eye" is therefore, like shape, a distinct character
inherited on Mendel ian lines.

The Coloub of Tubers.

The colour is due to the presence of pigmented cell sap in the
cells of the superficial layers. The white skinned or, more correctly,
yellow skinned tuber, owes its colour on the one hand to the presence
of the cork in the upper layer of the corky tissue, and on the other
to the absence of any red or purple pigment. The red potato contains
a vermilion pigment in solution and the black potato, which is in
reality an intense purple, derives its colour from a deep blue purple
sap pigment which, seen under the microscope in contrast with the
red, is quite distinct.

It was pointed out in the Introduction that potatoes of all colours,
including the whitest — with white flowers — showed more or less purple
pigment in the shoots, arising from the tubers in spring, if not in the
haulm also. Vilmorin (lo), in his catalogue of all the known varieties,
makes three classes in which the tubers possess white shoots; it is
probable that small deposits of pigment were overlooked. Out of the
1200 separate and distinct varieties he describes some 45 as having
white shoots. Often the pigment occurs in punctate deposits which
need a lens to distinguish them clearly, but the pigment is unmis-
takably present. From this fact it would seem clear that all tubers,
coloured or not, possess the chromogen base, i.e. using the notation

30 Colour and other Characters in the Potato

employed in the Mendelian analysis by Bateson, Miss Saunders and
others, all potatoes possess the factor C. Miss Wheldale, who has
very kindly examined many of ray tubers from this point of view of
pigment analysis, confirms this view. If, then, colour can be present
in the haulm and even in the shoot and still not be developed in
the tuber, it would seem that there must be some factor which acts
as a "developer" of pigment, and in its absence the tuber is white
(yellow). The supposition that this factor might be an inhibitor of
colour is negatived by the fact that white are recessive to coloured
tubers.

It is necessary now to observe how the potato plant behaves in
actual breeding experiments.

The white potato breeds true.

Several hundred, about 600 in all, of seedlings of " Bohemian
Pearl " and " Sole's Kidney," both white potatoes, were raised, and
all the plants that bore tubers at all carried white ones only.

A " Bohemian Pearl " seedling was selfed and gave a half-dozen
white-tubered seedlings.

A "Sole's Kidney" gave 300 white-tubered seedlings, and one of
these selfed and produced fifty seedlings, all of which were white-
tubered.

A w^hite-tubered variety (D) extracted from " Flourball " has been
bred now through three generations and gives rise to nothing but
white-tubered plants.

The variety "Early Regent" sown this season has produced 125
white-tubered plants and none carrying coloured tubers.

The Colour Gomposition of the Red Potato. If seedlings of
"Flourball" be grown and these, after harvesting, divided up in
respect to colour, it will be found that red-tubered plants are to
white as 9:7,

The numbers in my experiments were : —

1907 271 Red plants 217 White

June 1909 71 „ 60 „

Oct. 1909 24 „ 19 „

Aug. 13, 1910' 54 „ 44 „

Total 420 „ : 840 ,

Ratio 9 „ : 709,

There are still about 100 plants to be harvested.

R. N. Salaman 31

The ratio 9 : 7 is one very well-known in Mendelian analysis and is
evidence of the interaction of complementary factors belonging to
separate pairs of allelomorphs.

Now if R be considered the factor which in presence of the
developer D converts the chromogen into a red pigment, then the
zygotic composition of " Flourball " should be written RrDd, which
will on selfing give plants with the following composition : —

9 RD=' Reds
S Rd = Whites
S Dr = Whites
I dr = White

Further, it will be seen that there are five kinds of white and four of
red plants, viz. — whites of the composition : —

Rrdd, ddrr, RRdd, rrDD, rrdD,

and reds of the composition,

RRDD, RrDd, RrDD, RRDd.

Of the red it is at present only possible to distinguish three
kinds, viz.,

RRDD, RrDD, or RRDd and RrDd. Of these RrDd we know as
the parent or type, the pigmentation of which is weak.

RrDD or RRDd has been raised twice out of " Flourball " seedlings,
and each case has given red and white tubered seedlings in the propor-
tion 3:1. Thus,

Family A 70 red 27 white
„ G* 12 „ 5 „

The colour of the tuber RrDD is distinctly stronger than the colour
of the ordinary " Flourball." There is good reason to hope that the
type RRDD will be isolated this season : such a potato will breed true
to red. "Reading Russet," a pale red, selfed in 1909 and planted out
this year, already gives evidence of a 9:7 ratio. Amongst the whites
no certain distinction has yet been made between the possible kinds,
nor have two whites been yet successfully mated ; an experiment which
when the two whites contain, one the R factor and the other the D
respectively, will probably give rise to a coloured potato*.

* This year, 1910, a laige number of crosses between Tarioos whites have been
effected*

32 Colour and other Characters in the Potato

" Flourball " has therefore yielded three types of potato which have
been identified by reason of their gametic qualities, namely, two reds,
one giving reds to whites in the ratio 9 : 7, another red to white in the
ratio 3 : 1, and a white variety.

In order to elucidate further the colour factors the white variety D
was crossed by the 3 : 1 red variety A and the result was

27 Red to 22 White.

This ratio is presumably to be taken as approaching equality, as
9 : 7 ratio would be here impossible.

If the formula of A be RrDD then this particular white potato
must be rrDD ; similarly if A be RRDd then the white variety must
be RRdd. It is here assumed that A = RrDD, and the family D
therefore will be represented by rrDD, it could of course be equally
well rrDd.

A cross of peculiar interest was made between " Flourball " and a
potato called " Record " which, although of attractive appearance, was
of such frail constitution that it has entirely died out everywhere.
The result of the cross was a family H. Of the 30 individuals which
lived through the following years 19 were white and 11 red. The
numbers are small, but enough at least to show that the whites are
in a very distinct majority. If the notes of the H family be examined
from its first origin, one finds that there were 28 whites to 12 reds and
two with no tubers, and that the mortality has taken place amongst the
white and tuberless.

The formula for " Flourball " was shown to be RrDd, and there are
two possible formulas for a white potato which would, in union with
" Flourball," give rise to a family having a majority of whites. They
are rrdd and rrDd respectively ; — the first would give a family of
three whites to one red ; the second would give a family of five whites
to three reds. The numbers in the H family are not large enough to
decide with certainty which formula for " Record " is the more correct.
We have seen that the mortality affected those plants which were
either white tuber bearers or tuberless, and that the approximation of
the final result of two whites and one red is due to this mortality
amongst the whites. Whether it is possible that plants pure to the
absence of pigment factors are more weakly than others cannot, on the
present evidence, be asserted, but the facts suggest such a possibility.

Two white-tubered members of the H family were crossed by the
red potato A, whose gametic composition we may assume to be RrDD,

R. N. Salaman 33

seeing that on selBng it gives three red and one white. The results
were different in each case —

W xA gave 29 red 19 white

lP»x A „ 18 „ 27 „

Total 47 „ 46 „

In either case it is possible that larger numbers would have shown a
nearer approach to equality.

It must however be noted that the family H^" x A, had far less
pigment in its stem than H^ x A, and that the possible results of
mating whites with reds of A's composition are equality, if the white is
rrBD or rrdd, or three red to one white if Rrdd.

One other cross was made between a pale red and a white-tubered
plant.

" Queen of the Valley " was crossed by a red seedling of " Flourball "
and the F^ generation consisted of seven red to three white. One of
these a pale red, M^, was crossed by a white seedling of the white
" Bohemian Pearl " B. Forty-one seedlings grew and 38 survived
to form tubers. Of these

19 had red and 19 had white tubers.

This result of equality suggests that the composition of the two
parents may have been — (M^) RrDd x {B) rrDD. if' is probably
RrDd and not RRDD, RrDD, etc., because it is a particularly feeble
red and might therefore be assumed to have the least possible factors
that would give a red.

Two reds, one very deep red, viz. " Red Fir Apple," and the other
a weak one, " Reading Russet," were crossed. " Reading Russet " has
now been selfed, and this year we shall learn its composition, but its
colour is weak like that of " Flourball," and it has probably the same
gametic composition, viz. RrDd^. " Red Fir Apple " is of a very deep
colour and might be RRDd. The F^ raised were 117 seedlings, but only
11 of them came to maturity, viz. eight red, and three white, indicating,
as would be expected from the union, a 3 : 1 ratio.

RRDd X RrDd = 3 red : 1 white.

Two plants arising both from tubers of the same individual of the
F^ family, viz. L^ and L*, were selfed and produced in the F^ generation
large families in which the ratio of red and white was 3 : 1.

1 The 1910 seedlings of "Reading Russet," so far as yet harvested, are divided into
14 red-tubered plants and 10 white-tnbered.

Joam. of Qen. t 3

34 Colour and other Characters in the Potato

The numbers in the latter are not conclusive in themselves, because
only selections of these families were actually planted out ; but amongst
the young seedlings, before planting out, there were 23 red to 8 white
and the appearance of the harvested selections fully bear out the sug-
gestion of a 3:1 ratio.

Purple Coloured Tubers. — The "Congo" potato is a cylindrical
potato of almost a black colour, the pigment extending within the
tuber somewhat irregularly. The " Congo " flower, which is white with
a purple tinge at the base of the petals, is completely sterile in the
male organs, and it was therefore only used as a mother plant.

Two crosses were made —

1. Congo X Reading Russet. There were eight seedlings and only
four survived until the late autumn of 1906, of these

Two were black like " Congo,"
Two bright red.

But on July 25, 1907, there was a fifth plant with white tubers which
died out subsequently.

The numbers are too small to make any deduction as to ratios, but
there is one factor of great importance which stands out, viz. — that out
of a union of a deep purple and weak red, there have segregated out
deep purple (black), bright red and white.

The next cross was —

Congo X Flourball Seedling. This cross was effected in 1906. The
" Flourball " seedling was a stray plant growing in one of the experi-
ment lines containing " Ringleader " and was used as pollen parent.
" Ringleader " itself did not flower that year. Except that it was
a red tubered variety nothing further can be told about it, as it was
unfortunately not preserved. Its pollen was used in the cross with
"Queen of the Valley" and, as has been mentioned before, it is probable,
for the reasons already given, that it was a red of the formula RrDD or
RRDd.

The F^ generation contained 29 plants and these were
13 Black tubers.
12 Red tubers.
4 White tubers.

Here again the important features are the complete segregation and the
appearance of the white tubers.

Before discussing the possible constitution of " Congo," it will be
best to consider the F^ generation.

R. N. Salaman 35

In 1908 two of the F^ plants, viz. K* and K^ both selfed and large
families were planted ; those of K' did well, the K^ family fared badly
in the wet summer of 1909.

K* Family. K*, Plate XXIX, is a black (le. deep purple) potato.
Several seedballs were collected from the plants, and one coming
from a plant .ff^*"* was planted in its entirety. Originally 301, there
were harvested but 160 seedlings. The tubers of the jP* family
separate at once into blacks, reds and whites in the proportion of 77
black, 29 red, 54 white; the reds are either quite pale and similar to
" Flourball " or " Reading Russet," or they have more purple colour
and resemble " Red Fir Apple."

Of the whites about one-sixth (9 in 54) are quite pure, i.e. no tinge
of colour can be seen in the tubers or eye before sprouting, whilst the
remainder may have a trace of colouring usually purple, in the eye or
the skin and more especially in any scars following a wound by fungous
disease or other lesion. Such pigment is minute in quantity and often
needs a lens to demonstrate its presence. The reds are roughly of two
kinds, a deep strong group, and a pale. The proportion between these is
23 deep red, and 6 pale red, and they can be classed fairly readily into
these main groups. The blacks are all alike, viz. deep purple. In con-
sidering the factors which underlie the phenomena of colour in the red-
and white-tubered potatoes we assumed the presence of the two factors
R and D. The purple potato is obviously bringing a fresh factor besides
these into the field and this new or " purpling " factor can be called P.

If ^* has the gametic formula Pp, Rr, Dd, then on selfing we
should get plants or biotypes with the following gametic constitutions :

27 plants of the composition PRD = purple.
" » >»

9 I, M

3 „ „

■'• » »

The numbers for the K^ family are :-

= white (tinged).

= red.

= white.

prd

= white.

Purple

Bad

White

Calculated nambers

Actual Numbers

9—2

36 Colour and other Characters in the Potato

The results^ are sufficiently close to give one some confidence that the
phenomena are correctly represented by the assumption of the factors
PR and D that have been supposed to be at work.

The sister family K^ adds additional evidence of a strong nature.
Several lots of seed of K^ plants were sown and in all some 300 seedlings
raised. The majority were however planted in selections and therefore
are of no use for quantitative purposes. All the groups, however,
coincided in one feature — none produced a single red tuber; and the
evidence from the selected groups strongly favour the view that purples
to whites were as 9:7, whilst the groups that were planted in full
give 26 : 14. The parent plant of such a family must be homozygous
in the purpling factor and heterozygous in its two other colour factors.
To K^, therefore, should be given the zygotic formula PP, Mr, Dd.

Having considered K^ aod K^, we can now turn back to the original
cross and the F^ family. The F^ family consisted of 13 purple, 12 red,
4 white. It is obvious that as regards P, " Congo " must be heter-
ozygous, further we knew the " Flourball " seedling was red and
therefore contained RD. If we represent the cross

" Congo " PpRrDD x " Flourball " seedling RrDD

we get 12 purple, 12 red, 8 white.

The result of these experiments on colour inheritance would seem to be
(1) that whilst colour may be present in the stem to any degree, a
special developer D is necessary to bring it out in the tuber, (2) that
redness is dependent on a separate factor R, (3) that purple is
dependent on a further one P, and (4) that the purple colour cannot
be developed except in the presence of all three factors PRD.

In all the experiments there has been much to suggest that the
degree of the " redness " is due to the homozygous condition or other-
wise of the plant as regards both R and D, but the evidence has not
been given in full because the classification into shades of " redness "
would be too empirical and dependent on personal judgment. In one
group the distinction was clearly made out, viz. in the family A where
the formula was shown to be RrDD (or RRDd) the deep reds were to
the remaining reds as 24 to 48, whilst in the K^ group the reds were
23 deep red to 6 pale red. Amongst the blacks (purple) no distinction
could be made.

1 If the disproportionate mortality of the whites be remembered, the actual numbers
will be seen to be not so far removed from the calculated ones. Thus the number of
whites, had the mortality in all classes been equal, would be 66 instead of 54.

R N. Salaman 37

SOLANUM ETUBBROSUM.

The plant with which I have worked is identical with that used by
Mr Sutton (8) and described and figured so fully by him. I obtained
my tubers from Kew, whence it was sent to me with the name of
"Maglia," though the misnomer was realized later. Mr Sutton has
been good enough to see my plants growing, and has no hesitation in
confirming that they are the same as his own obtained from Mr Lindsay
of Edinburgh Botanical Gardens and which he has described under the
name of " etuberosum." The Rev. Aikmau Paton's supply of etuberosum
was derived from mine, and his results, as far as they are published,
confirm mine in many particulars.

It is not necessary to decide as to whether this plant is the one
originally described by Lindley in 1834 as etvberosum ; the general
feeling is that it is not the same, but that it is a plant of the greatest
interest is none the less true though its name be a borrowed one.

The contention of Sutton (s) that S. etuberosum is the parent plant
of our domestic varieties has been considered by me in an earlier
paper(9). Wittmack(i2) has also discussed this question, and though
I do not share his opinion that etuberosum is an ordinary S. tuberosum,
variety I, nevertheless, agree with hi m that there is no reason to regard
it as the parent type of our domestic varieties.

The etuberosum plant is a low growing one with very light green
leaves which are of a different tone to any other I have had growing in
my garden. It rather suggests the dusty appearance of the olive. The
haulm spreads at its lower end, sending out lateral branches parallel to
the ground.

The average size of the leaf is 2 J inches by 1 inch ; the surface is
soft and rather woolly ; the veins are marked, but the leaf not curled or
rugose. Compared with most domestic varieties the nodes of the stem
would be considered short, but they are, in proportion to the rather
dwarf-like habits of the variety, about normal in length.

Pigment in the stem is red, patchy, extending feebly into the
petioles, and visible in the axils. The flowers occur in close clusters,
and are of an extremely beautiful lilac, which, viewed from above, has a
peculiarly soft appearance. This is due to the fact that the pigment is
on the under surface of the petal, that is outside when the flower is
closed. This lilac colour differs considerably from the heliotrope seen
commonly in domestic varieties. The anthers are delicate and form

38 Colour and other Characters in the Potato

a close cone similar to that seen in the various true wild species,
and through the apex projects a short style ending in a simple knob.
The anther contains abundant pollen.

The corolla is very definitely wheel-shaped, the tips of the petals
recurve ; they are rather sharp and hairy, and the calyx is hairy and its
five processes are long.

The tubers are borne on rather long stolons. They are white and
round, but the shape (Plate XXVII) is not typical of " round " as
we have met it before in this paper. The tubers are irregular, neither
oval nor long, but are often depressed at various points, so that
although the general shape is round, the actual circumference is not
necessarily circular.

The size is variable. When the tubers were first cultivated here
they were not more than 1^ inches in diameter; in 1909 I had some
up to 3 inches in diameter.

The taste is bitter.

In 1906 Mr Sutton informed me that he had for over 20 years tried
to self and cross this variety and had failed. In that year, however, a
plant bore one berry. I, also, after repeated trials, in 1906 succeeded
in making a cross. In 1907 Mr Sutton again obtained selfed berries,
and some tubers I had sent to the North of Scotland set seed naturally
and crosses were made. Hence, after over 20 years of observed sterility,
this variety suddenly flowers out into fertility in Reading, Scotland
and North Herts, which, as we shall see, has cost it dear. The tubers
in both 1906 and 1907 showed no variation, except a slightly enlarged
size. In 1908 when the plant first set seed naturally in Barley, it was
noticed that the tubers of one plant had a slight violet tinge in the
skin in places ; this plant set seed in addition to one other, and 30 of
the seedlings came from this plant. There is no evidence that the
seedlings are, as a whole, different from those which did not show this
vegetative variation.

The fertilization of the plants took place naturally, but at a date
when all the other potato plants in ray garden had ceased flowering
and when some F^ " Congo " crosses, which were close by, had already
formed good-sized berries.

Immunity to Disease. (Phytophthora infestans.) During the culture
of this variety in Reading it was noted for its immunity to disease.
In my garden it was in

1906. Perfectly immune from disease in haulm and tubers.
Three hybrid seeds only obtained.

R N. Salaman 89

1907. Very slight touch of disease on haulm, none in tuber.

No seed.

1908. Slight disease in haulm, none in tuber. Set seed freely.

1909. No disease in haulm on September 3, but some later,

considerable disease in tubers. No Seed.

1910. Some disease in haulm in August. Selfed and crossed

seed.

The incidence of disease amongst the seedlings was remarkable,
those attacked by disease were in some cases consumed away and all of
them, excepting one which was but very slightly touched in the haulm
and quite free in the tuber, were most seriously damaged. Out of 40
seedlings 34 were diseased and six were untouched, to these might be
added the one only just touched by disease on a leaf or two, making
seven. The ratio of 33 : 7 is of course suggestive of a 3 : 1 ratio.
Resistance to disease being, as Biffen(2) found in the case of wheat, a
recessive. Further careful observation will be needed before anything
more definite can be asserted. Id is a most striking fact that although
the parent etuberosum plant was for 20 years and upwards noted for its
immunity to disease, yet directly its sexual life begins that immunity
goes. The chain of events, the fact that the F^ family contains a
number of immune plants, suggests that with the onset of sexual
activity some disturbance in the mechanism by which the plant had
hitherto security its immunity to Phytophthora had occurred — and that
the dominantly susceptible state of the plant apparently heterozygous
in this respect, has as it were been uncovered and its true nature laid
bare.

The immune seedlings in 1910 demonstrated afresh their resistance
to Phytophthora. The etuberosum seedlings were so planted that on
either side of an immune plant was a susceptible one, whilst immedi-
ately behind was a row of ordinary domestic potatoes. The susceptible
seedlings and the ordinary potatoes were devastated by disease. Before
the end of July the haulms of both these latter were destroyed- Up
till the beginning of September the immune plants were unscathed.
Signs were not wanting that the immune plants had been attacked but
had successfully withstood the enemy. Pale spots were seen on some
of the green leaves during the height of the disease, whilst on these
spots on a few fading leaves colonies of Cladosporium epiphyllum were
found. The presence of the bright green healthy immune plants

40 Colour arid other Characters in the Potato

standing out in the naidst of the blackened and diseased debris which
marked the site of their destroyed neighbours formed a very striking
picture. Successful crosses have been made this year between the
immune seedlings and domestic varieties.

The Flower. It has been already noted that the flower of this potato
is of a very delicate lilac and that the pigment is on the under surface.
The petal is entirely self-coloured ; there is neither an intensification or
a weakening of the general tone in the central region of the petal, as
one so commonly finds in potato flowers.

The flowers of the seedlings offer considerable variations. Of the
40 plants 20 flowered, and of these —

Nine plants were exactly like the parent, i.e. uniform colouring on
under surface ;

Two plants were similar to parent but double the intensity of
colour ;

Three plants had the same general colouring as the parent, but
with a deep-coloured tongue in the middle of the petal, and in one it
was noted (probably true for all) that the colour in the tongue was both
in the upper and in the lower coats of the petal ;

Three plants had white flowers with purple tongues in the centre
of the petal, the colour in the tongue being on the upper surface ;

Three plants were pure white.

The sequence of the diverse flowers can be readily explained on the
following hypothesis — that we have two pairs of characters at work —

A. Colour. a. Colour absence.

B. Uniform distribution of colour h. Distribution of colour in a

on under surface. pattern on upper surface.

We then get —

6 : Bh. Aa. = Parent type.

2 : Bb. AA. = „ „ with deeper-coloured tongue.
1 : AA. BB. = „ „ but deeper colour,

S : A. b. = White with coloured tongue.

3 : a. B. = White.
1 : ab. = White.

The numbers are too small to lay much stress on an explanation
such as the one given, but the phenomena fall readily into line.

R N. Salaman 41

Shape of Tuber. The tubers of etuberosum are, as already mentioned,
" round " — the seedlings comprise both " rounds " and " longs," and
amongst the latter are kidneys. The numbers are 18 round, 14 long.
It is evident that the " roundness " of etuberosum is of a quite different
order and with a different hereditary value to that of the domestic
varieties, and moreover, it is obvious that the " round " here is dominant
to the " long," whereas in the domestic types it was recessive.

The Eyes. The eye of the parent tuber is " shallow " and very
insigniHcaut. The seedlings can, as regards the tuber eye, be at once
divided into " deep " and " shallow."

These are 26 " shallow " to 8 " deep."

" Shallow " eye is therefore clearly dominant : in the domestic
variety it is as clearly recessive.

The Colour of the Tuber. It will be remembered that, although the
etuberosum tuber is white, yet in 1908 certain tubers were noted to
have shown a slight purplish tinge. It is not therefore surprising to
find that the seedlings are varied in colour and that the parental white
is a dominant.

The colours of the seedling tubers are white and deep purple. The
latter are identical in colour to those purple tubers dealt with in the
earlier part of this paper.

The numbers of the different colourings are —

White 13

White tinged 12

Deep purple (black) 13.

25.

The numbers suggest that purple is a recessive character and that
white is a simple dominant. In the domestic varieties the reverse is
true. No reds were formed.

Crosses with Domestic Varieties. In 1906 I succeeded in effecting a
cross with " Queen of the Valley." Three seedlings only grew, and they
all died out. Mr Paton(7) crossed etuberosum by the white kidney
" Duchess of Cornwall," and he obtained 13 seedlings, the colour of
12 of which he describes, viz.

9 white, 2 purple, 1 red,

showing the dominance of white. It is of further interest to note that
he describes the shape of ten of them. Eight are " round " and two
are "long" (kidney and oval), again showing the dominance of the
etuberosum type of " roundness."

42 Colour and other Characters in the Potato

Crosses with S. etuherosum and maglia.

Sol. etuberosum x Sol. maglia (deep purple)

One seedling white tuber.

Sol. maglia x Sol. etuberosum.

One seedling white tuber.

Here again the " white " of etvherosum is dominant to the purple of
the recognized species maglia.

The relation of S. etuberosvm to other potatoes. Although the name
" etuberosum " has been used in this paper, it has been done rather for
convenience than with any idea of establishing its identity with the
species described by Lindley.

Whether S. etuberosum is to be classed with the domestic varieties
or as a native species is a question that may have an increasing import-
ance. It has been shown in this paper that in respect to such important
characters as shape, eye and colour of tuber it behaves in a diametrically
opposite way to the domestic varieties, and it is, therefore, likely that
it is distinct from them. On the other hand, its white is dominant to
the muglia purple, and its own purple is also recessive; so that in
respect to this character it certainly more closely resembles maglia.

The flower of etuberosum is much smaller and more compact than
that of the domestic potato, and is much more like the wild S. etuberosum
and S. maglia, and its scheme of colour as described here has no parallel
amongst the domestic varieties.

There would seem, therefore, to be no adequate reason at all for
classing S. etuberosum amongst domestic varieties ; on the other hand,
it has certain characters akin to those of recognized specific types, such
as S. maglia.

It has been suggested that the diversity of the S. etuberosum
seedlings shows it clearly to be a hybrid. That may be, but we can
feel at least equally sure that its parents are not domestic varieties.

Conclusions.

Very briefly the following conclusions have been reached in this
paper.

Domestic Varieties.

1. The twist of leaf, as seen in " Red Fir Apple," is a recessive
character.

R. N. Salaman 43

2. Length of tuber is dominant to " roundness."

3. Depth of " eye " is dominant to " shallowness."

4. Purple is dominant to red in the tubers.

5. Red is dominant to white, but is dependent on the presence of
two factors in addition to a chroraogen.

6. S. etuberosum is not subject to the same laws of dominance as
the domestic varieties of potatoes.

7. That amongst the seedlings of S. etuherosum occur some which
are at present immune to the attacks of Phytophthora in/estans.

8. That immunity to the attacks of Phytophthora xnfestans is in
S. etuherosum a recessive character.

9. S. etuherosum may be a hybrid and, if so, its parents are possibly
native species.

I take this opportunity of tendering my thanks to my head gardener,
Mr E. Jones, for the assistance he has rendered, and the great care he
has shown in the raising of the seedlings.

DESCRIPTION OF PLATES.

PLATE I.

Tubers of seedlings of Sutton's "Floorball" selfed. "Bonnds" are — Nos. 40, 89, 92,
118, 132, 138, 155, 156, 162, 185.

PUVTE 11.

Family of seedlings of parent A selfed. The majority of the tubers are normal "rounds";
the least typical "round" has been chosen to represent each individual root. On
Plates IV. and Y. can be seen the sister tubers of the more abnormally shaped

"round" tubers.

A family continued.

PLATE iU.

PLATE IV.

All the available tubers of each root crop are shown of those individuals who vary from
the typical " round." In all cases one or more typical " rounds " occur in eaeh
root crop.

PLATE V.

Same as Plate lY. No. 100 is probably a stray plant and not a member of this family.

44 Colour and other Characters in the Potato

PLATE VI.

The G family, consisting of six individuals with their root crops are shown. G*, Q^
and G® are more or less typically "round."

PLATE VII.

The D family — Top row — Three tubers of parent plant. D^ and Ifi, 1908, are the seedlings
raised in 1908 from D (1907) selfed, D^ and D^, 1909, are seedlings raised in 1909
from D (1907) selfed.

PLATE VIII.

Seedlings of the family raised from cross D y. A. The family consists of half "rounds"
and half "non-rounds." The "rounds" are Nos. 3, 4, 5, 8, 13, 14, 15, 16, 18, 19, and
8, 6, 7, 10, 12, 14, 18, 19, 20, 21, 22, 28.

PLATE IX.

Seedlings of the family raised from the cross "Eecord" x "Flourball." "Eecord" is
a kidney, "Flourball " a pebble-shaped potato (neither parents shown). One quarter
of the seedUngs are "rounds," viz., Nos. 12, 13, 18, 21, 24, 25, 26, 30.

PLATE X.

Seedlings of the family raised from the cross E} {F^ of family B., Plate IX) x A. Half the
seedhngs are "round," viz. : Nos. 4, 6, 7, 16, 17, 19, 22, 25, 26, 27, 29, 30, 34, 35, 36,
38, 39, 40, 42, 45, 49.

PLATE XI.

Seedlings of the family raised from the cross E>^ (i^i of family H, Plate IX) x A. Half
the family are "rounds," viz. : Nos. 1, 2, 3, 4, 10, 11, 13, 13a, 15, 17, 18, 19, 26a, 29,
31, 32, 33, 34, 40, 46, 48, 49.

PLATE XII.

Family J raised from the cross "Congo" x "Beading Eusset." The fifth seedling, a
long white-tubered one, died out and is not shown here.

PLATES XIII— XVIII.

The family raised from the individual K^ (i^' of "Congo" x "Flourball" seedling,
see Plate XXIX). This family for convenience has been divided into sub-families
K'c?, K^^, etc., according to the particular seedball from which the seedUngs were
grown. " Rounds" are to "longs" as 1 : 3 in this series, and the eyes are all deep
with the exceptions noted in the text.

PLATES XiX, XX.

The family raised from selfing K^ (F^ of "Congo" x "Flourball" seedling, see Plate
XXIX) the "rounds" are rather deficient, viz. : 13 to 60; the eyes are deep to
shallow, 3 : 1.

R N. Salaman 46

PLATE XXI.

The family L, raised from the cross of "Red Fir Apple" x "Reading Rasset." In
the f, No. U, a kidney has been omitted.

PLATES XXM, XXIII.

F*, family raised from L\ selfed. The rounds are 1 in 4, viz.: Nos. 1, 2, 22, 35, 37,
46, 47, 49, 61, 63, 64 (Plate XXII). Five long- and three roand-tubered individuals
have been omitted. In Plate XXIII the " rounds" are Nos. 6, 10, 17, 19, 22, 24, 29,
80, 33, 40, 47, 52, 54, 61.

PLATE XXIV.

The family raised by crossing SP (F^ of " Queen of the Valley " x " Flourball "' seedling)
x" Bohemian Pearl" long-tubered seedling. Nos. 2 and 20 which in the plate look
"round" are in reality much flattened and are clearly not rounds. Two other typical
long members of this family have been omitted.

PLATE XXV.

Examples of tubers, not from individual roots, of B,
" Bohemian Pearl " seedlings long and oval.

"Congo." The long tubers are much more common than the stunted.
"Red Fir Apple." The tubers in 1909 were all more or less stunted as shown in
the Plate.

PLATE XXVI.

C, 1907, one of the seedlings of "Sole's Kidney."
C, 1909, representatives of 4 seedlings of C, 1907.

PLATE XXVII.

Family raised from selfing Lindsay's etuberosum. The long-tubered seedlings are here in
the minority. The ravages of the disease are clearly seen.

PLATE XXVIII.

(Reproduced by kind permission of Messrs Sutton of Reading.) The kidney potato "Super-
lative" in clamp. The variability of shape amongst the kidney and pebble-shaped
tubers is very marked.

PLATE XXIX.

The F^ family raised by crossing "Congo" x "Flourball." The segregation of the
colours Purple, Red and White are well shown. The shapes are all "long" and the
eyes all " deep," demonstrating the dominance of these characters.

46 Colour and other Characters in the Potato

BIBLIOGRAPHY.

1. Bateson, Saunders and Punnett. Rep. Evol. Comm. Roy. Soc. 1904,

Vol. II. p. 91.

2. BiFFEN. Journ. Agric. Sc. 1907, Vol. ii. p. 109.

3. Darwin. Animals and Plants, 1890, Vol. ii. p. 149.

4. East. Rep. Connecticut Agric. Exper. St. 1907—8, p. 429.

5. . " Transmission of Variations in Asexual Reproduction." Rep. Con-
necticut Agric. Exper. St. 1909—10, p. 120.

6. Gaertner. Versiiche und Beohachtungen iiher Befruchtung-organe, Stuttgart,

1844, 849, S. 117.

7. Paton. J. R. Hort. Soc. Vol. xxxv. p. 33.

8. Sutton. Linn. Soc. J. Bot. Vol. xxxviii.

9. Salaman. Linn. Soc. J. Bot. 1910, Vol. xxxix. p. 301.

10. ViLMORiN. Catalogue M^thodique et Synonymique de Pommes de Terre,

Paris, 1902.

11. WiTTMACK. Bericht. d. Deutscht. Bot. Ges. 1909, Bd. xxvii. S. 28.

12. . Zdt.f. wiss. Landunrt. 1909, Bd. xxxviii. erganz. Bd. v.

JOURNAL OF GENETICS, VOL I. NO. 1

PLATE I

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE li

JOURNAL OF GENETICS, VOL I. NO. 1

PLATE

^V -Deed t ir\9S . G,„f-

73 7S 7^ 77 7S Tf 80 81

8 2 83 54 Z5 S7 SS S«? 1o qi

^2^ **3 'n 'JS <|4 q7 q^ qq

■ ■/ -| •-• -J-J

^|00 «OI

JOURNAL OF GENETICS, VOL I. NO. 1

PLATE IV

•

JB "•d'305

40|

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE V

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE VI

^ ^

G 3

1^' .

;*^$

JOURNAL OF GENETICS, VOL I. NO. 1

PLATE VII

l^oj.

X)' •^'^S

jy"^ l^aS

1^ ■*

J^' •i*'1

I>" ..,^^.

JOURNAL OF GENETICS, VOL I. NO. 1

PLATE VIII

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE IX

Ur^

^^0^

JOURNAL OF GENETICS, VOL I. NO. 1

PLATE X

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE XI

m^mm^

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE XII

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE XIII

- •

: •

•^

JOURNAL OF GENETICS, VOL I. NO. 1

PLATE XIY

•4

wB ' 9

■s.

f^^

Jf^

"^ '

^^v,^

^.•^

JOURNAL OF GENETICS, VOL I. NO. 1

PLATE XV

««f

^^^

o
u

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE XYI

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE XVII

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE XYIII

JOURNAL OF GENETICS, VOL \. NO. 1

PUTE XIX

1^8 ^

rar«

^f 9

U /3

IS IS l<\ 23 31 33 35 48 J5 ^^'/ 71

•

K^'

ji^ 8q qo ^1 3 4 «1 iz i3

^ <j IS »•*' ^^

23 33 ^y i ^ 7

S IX '7 /s z/ 22 2^ 24 30 3| 3Z 34 35

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE XX

JOURNAL OF GENETICS, VOL I. NO. 1

PLATE XXI

<=5^-o'?-uC L^

^ 'L'~i'<:A..>y\^MJiji^

'^Cc<a-cCiAn_ok^ 'l^-cr^/aA^C

OU\xmJly;i

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE XXII

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE XXlii

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE XXIV

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE XXV

0...I

1 > %

(o A/ C^ O

••i„§^

I ^ }^

JOURNAL OF GENETICS, VOL I. NO. 1

PUTE XXVI

(i»

I til ^

If»

JOURNAL OF GENETICS, VOL I. NO. 1

PLATE XXVII

JOURNAL OF GENETICS, VOL I. NO. 1

PLATE XXVIII

[Reproduced by kind pertu.^.^.^,,, .,/ Messrs Sutton of Reading

X
X

z
cr

THE MODE OF INHERITANCE OF STATURE AND
OF TIME OF FLOWERING IN PEAS {PISUM
SATIVUM).

By FREDERICK KEEBLE,

Professor of Botany, University College, Reading ;

AND Miss C. PELLEW,
Research Student, Botanical Laboratory, University College, Reading.

The experiments recorded in the present paper, though incomplete,
throw some light on the nature of the Mendelian factors which deter-
mine stature in peas (Pisum sativum) and on the mode of inheritance
of earliness and lateness of flowering in this species. The experiments
were designed originally to investigate the latter problem — left un-
decided by Mendel's classical experiments on the inheritance of " time
of flowering," As our work proceeded, it became evident that this
function of the plant is not unconnected with certain, definite, morpho-
logical characters. Hence it becomes necessary to follow the course of
inheritance of these characters, e.g. length of intemode and thickness
of stem. In doing this, we have been led to conclusions concerning,
not only the relation of these morphological characters with flowering
period, but, also, the part which these characters play in determining
the stature of peas. We deal first with the latter point.

Stature. Certain garden races of Pisum sativum grow tall and give
rise to tall-growing offspring ; other races are dwarf and breed true to
this character. Adopting Bateson's classification (1909, a) we call
"tall," those varieties which grow 5 — 6 or more feet high, "dwarf,"
those which range from 9 inches to 3 feet. Between dwarf and tall are
various " half-dwarf" races which reach a height of about 4 feet The
actual height attained by the various races is determined in any given

48 Inheritaiice in Pisum

year, partly by gametic constitution and partly by the external condi-
tions to which the plants are subjected during their growing period.
Thus the two half-dwarf varieties Autocrat and Bountiful, used in our
experiments, are so constant with respect to stature as to be described
by Messrs Sutton and Sons, to whom we are indebted for seeds, as
being, the former variety 8 — 4 feet, the latter '?>\ — 4 feet in height.
Nevertheless, during the constantly wet summer of 1909 both Autocrat
and Bountiful reached in the College Gardens at Reading an average
height of 5 — 6 feet. We refer to these well-known seasonal fluctuations*
in height in order to point out that particular care is required in the
interpretation of the results obtained in any one year and in the
comparison of the statures of plants grown during different years.

The cross Autocrat and Bountiful and its reciprocal. This cross,
made in 1907, and repeated in 1908, yielded an F^ generation, the
plants comprising which were considerably taller than either parent
grown under like conditions. The average height of F^ plants was
7 — 8 feet, that of the parent plants 5 — 6 feet.

jPi selfed, yielded offspring {F^ which ranged in height from 8 feet
down to 1^ feet. In all, 192 F^ plants were recorded. Of this number,
61 plants were the progeny of a single F^ plant of Autocrat x Bountiful
grown in 1908. The remaining 131 plants of the F^ generation were
descended from four F^ plants of the cross Bountiful x Autocrat.
The seeds from these four plants were, owing to a mistake, harvested
together. There is, however, no recognisable difference between the
descendants of the single family from Autocrat x Bountiful and those
derived from the four F^ plants of Bountiful x Autocrat. We will
therefore consider the 192 F^ plants as a whole.

The F2 plants, showing such marked differences among themselves
with respect to height, fall into four groups which, for the moment,
may be defined as follows: — F^ type, Autocrat type, Bountiful type,
and Dwarf type. Moreover, when classified in this way, the numbers
of plants in the four groups show a close approximation to those
expected in the F^ generation derived from a dihybrid cross; that is
one in which two pairs of characters are involved.
Thus : ¥2=

f, type

Autocrat

Bountiful

Dwarf

Observed

114

Calculated

108

: 3

: 3 :

Inspection of the parent plants. Autocrat and Bountiful, reveals the
fact that, besides other, apparently minor, differentiating characters,

F. Keeble and C. Pbllew 49

these two lialf-dwarf varieties are distinguished from one another by
two well-marked characters, namely, length of internode and thickness
of stem. Thus Autocrat, whose normal height is 3 — 4 feet, has thick
stems (with large fleshy foliage of a bluish green colour) and short
internodes of about 3 inches in length. Bountiful, whose normal height
is 3^ — 4 feet, has thin stems (with foliage smaller than that of Autocrat
and of a yellowish green colour) and long internodes (5 — 7 inches). It
may be noted incidentally that the rates of growth of these two varieties
are very different ; the growth in length of the axis of Autocrat being
markedly slower than that of Bountiful. For example, when Autocrat
and Bountiful are planted at the same time. Autocrat grows one foot
whilst Bountiful grows two. The slowness of growth in length is asso-
ciated with short internodes. The two varieties differ also with respect
to mode of branching. Autocrat forms three to five branches at or
near the ground-level. These branches develop at the same rate till
they and the main axis are about 2 feet in height and then one axis
takes the lead. Bountiful shows a less fixed mode of branching.
Among the 1909 plants, some branched at the ground-level (2 — 4
branches), others formed their first branches a foot or so above the
ground-level. Generally speaking, thick stem appears to be associated
with branching, and thin stem with single stem, at the ground-level
We are engaged in endeavouring to work out the anatomical bases for
thick as opposed to thin stem, and for long as opposed to short inter-
nodes, and the bearing of these factors on growth.

We will now consider the factors, thickness of stem and length of
internode, in relation with stature. That these factors maybe taken
as valid representatives of those which determine height is evident
from the following considerations : —

Fi plants, 7 — 8 feet high, have all thick stems with long internodes
(6 — 9 inches). If the factor for thick stems is represented by T, and
its allelomorph (thin stem) by t, and if the factor for long internodes is
represented by L, and its allelomorph (short internode) by I : then the
gametic constitution of Autocrat is Tl, that of Bountiful is tL, and
hence the gametic constitution of F^ = TtLl. We ascribe the great
height of F^ plants to the presence of the factors T and L and to their
dominance over t and I. The suggestion may be hazarded that the
greater height and vigour which the F^ generation of hybrids commonly
exhibit may be due to the meeting in the zygote of dominant growth-
factors of more than one allelomorphic pair, one (or more) provided by
the gametes of one parent, the other (or others) by the gametes of the

Joam. of Gen. i 4

60 Inheritance in Pisum

other parents. This provisional interpretation of increased vigour of
^1 plants, has at all events the merit of being less obscure than the
hypotheses which are current in the literature of plant physiology
(Jost, 1907).

We return now to the F^ of the crosses between Autocrat and
Bountiful. Since the constitution of F^ = TtLl, its gametes have con-
stitutions :—TL :Tl:tL: tl, and hence, when F^ plants are self-fertilized,
we expect the usual 9:3:3:1 ratio ; i.e. in 16 plants, 9 with both,
dominants {T and Z); 3 with one dominant; 3 with the other and 1
with the two recessives {t and I). That the expectation is realised is
seen from the following table in which the results already given are
recorded in terms of Tt and LI.

2^8 =

F, type

Autocrat

Bountiful

Dwarf

Observed

114

Calculated

108

: 3

: 3 :

Of the 13 dwarf (^Z) plants, all but one were below 3 feet in height,
the three dwarfest being respectively 1^, 1|, and 2 feet. Table II
gives the records of height, of thinness or thickness of stem, and of
length of internode of the plants which we regard as true dwarfs. The
classification is of course open to the objection that thick and thin are
but qualitative terms, and that, in difficult cases, the criteria are purely
subjective. An answer to this objection is that the records in Table II
were made before we were aware that the characters "thin" or "thick"
were of any considerable importance. We include in Table II the
records of the characters oi F^ plants of a stature up to 4 feet. A com-
parison of the descriptions of the stems of the shorter plants (above the
horizontal line in Table II) with those of the stems of the less short
plants (below the line) confirms, as we think, the view which we have
expressed above, that the dwarf plants have thin stems and also short
internodes. With regard to the range of variation in height, both
among the dwarfs, and among the plants of the constitution Tl, it is
probable that the character of number of nodes, and also that of the
position of the first flowers, are also of importance. It is hoped that
further investigation of these characters among the F^ families will
determine this point.

We conclude from the above experiments that tallness in peas
(P. sativum) depends on the presence of two factors, long internode, and

F. Keeblk and C. Pellew 61

thick stem: that these factors are Mendelian in their inheritance;
being dominant respectively to short intemode and thin stem factors.
Half-dwarf peas are of two kinds. One kind, represented by Autocrat,
owes its semi-dwarfness to lack of the long intemode factor. In the
absence of this factor, the thick-stem factor cannot effect more than a
sturdy, medium growth. The other kind of semi-dwarf lacks the thick-
stem factor, and, in the absence of this factor, the long intemode factor
cannot build the stem-segments of a sufficient length to produce tallness
in the plant. It may be urged that this, after all, is but a common-sense
view of the way in which growth in length is effected : that only plants
with long internodes among annuals are likely to be tall ; and that only
when stems are sturdy may internodes reach their full length. This
may well be and it is certainly not a reproach to Mendelism that it
may lead to the discovery of the obvious which, without the method,
remains obscure.

The conclusions which we have reached as to the gametic constitu-
tion of tall, semi-dwarf and dwarf peas may be summarised thus : —

Tall = TL.

Semi-dwarf = tL or Tl.
Dwarf = tl.

In a cross described by Lock (1905) we have what seems to us an
interesting confirmation of this view of the chief factors involved in
stature of Pisum. Lock's comment on the case is as follows {op. cU.
p. 414): — "This cross seems to afford an example of remarkable inten-
sification of both the allelomorphic characters of the same pair, viz.
tallness and dwarfness — the former in F^ and both in F. and later
generations." The cross in question was one between Satisfaction — a
variety which at Peradeniya grew to an average height of 46 feet —
described as of robust growth (which we may take to mean thick stem),
and with internodes of an average length of 1'74 inches, and a Native
Pea of less than 3 feet in height, but varying much in different years,
with thin stem, and internodes of an average length of 1^ inches. In
the four plants of the F^ generation, the internodes were of an average
length of 2*4 inches — longer than in either parent — the height of the
^1 plants was about 6 feet, and the number of internodes was the same
as in Satisfaction. It would seem that in the increased length of
intemode of the Fi plants (an average of 24 inches as compared with
the 174 inches of Satisfection) is seen the influence of thick stem on
an intemode which, when combined with thin stem, is of an average

4—2

52 Inheritance in Pisum

length of 1^ inches. In i^j. the preponderating type resembled the i^i
plants, and the appearance of dwarfs, shorter than either parent, with
internodes of 1"0 — 1'2 inches in length (the proportion of long to short
being 19 : 6), confirms our belief that the characters thick and thin
stem, long and short internodes were the chief stature-factors involved
in this cross. Probably the difference in the number of^niodes intro-
duces a complication, but the small numbers grown in F^ and the lack
of further records, prevent a full analysis. We should mention that
this cross was made primarily by Mr Lock with the object of inves-
tigating the characters of the testa of the seeds of Pisum.

In conclusion, with respect to the question of tallness and dwarfness,
it is evident that a closer investigation will reveal facts of great import-
ance to an understanding of the physiology of growth.

Time of flowering : earliness and lateness. Certain varieties of peas
are well known and prized for their stability with respect to time of
flowering, and therefore it is to be supposed that the character is
hereditary. Mendel many years ago commenced experiments with a
view to determine the mode of inheritance, but few records of these
experiments are left to us. In Mendel's memoir on the hybridizing of
peas (1909b) we find the following: — "As regards the flowering time
of the hybrids the experiments are not yet concluded. It can, however,
already be stated that the time stands almost exactly between those of
the seed and pollen parents, and that the constitution of the hybrids
with respect to this character probably follows the rule ascertained in
the case of the other characters."

By the use of the varieties Autocrat and Bountiful for such an
experiment, the advantage is gained of a long space of time between
the flowering periods of the two varieties : the former variety flowers,
in normal seasons, about 30 days after the latter. Thus, in 1909, from
sowings made in April, 23 out of 28 plants of Bountiful were in flower
on June 2nd, whereas Autocrat, sown at the same time, was only
just coming into flower on June 30th (see Table I). In spite, how-
ever, of the favourable nature of our material with respect to the
character under consideration, we cannot claim to have arrived at a
complete understanding of the mode of inheritance of earliness or
lateness of flowering. Nevertheless, we publish our records, and our
attempts to analyse them in Mendelian terms, since they appear to
show definitely not only that the problem is capable of solution, but
also the nature of the difficulties which have to be met before the
solution is obtained. It will be seen from the records of the time of

F. Keeble and C Pellew 53

flowering (Table I) that the Fi generation is intermediate with respect
to time of flowering between the parents Autocrat and Bountiful. In
1909, whereas 23 out of 28 plants of Bountiful blossomed by June 2nd,
and whereas Autocrat was beginning to flower by June 30th, 10 of the
plants of ^1 (of a total of 12) were in flower by June 21st; and the
remaining plants were in flower by June 30th. The ^i plants of
Autocrat crossed Bountiful and those of the reciprocal cross, grown in
1908, confirm this result. From the appearance of such an intermediate
form in i^i, it may be supposed, either that there is incomplete dominance
of lateness over earliness, or that there are two (or more) factors con-
nected with the time of flowering ; the meeting of the two, or more,
dominant and antagonistic factors, from either parent in the Fi plant,
giving an intermediate time of flowering. In the former case, the F^
plants, obtained by selfing F^, may be expected to give the 1:2:1
ratio ; but, though segregation of early and late occurs in F^, it is not
of this simple type. If we tabulate the observations on time of flower-
ing, not of the F^ generation as a whole, but of the several categories
of that generation, viz. thick long {TL), thick short (Tl), thin long (tL),
and thin short (tl), we obtain the results shown in the accompanying
Table.

TABLE I.

The Accelerating Injluence of Long Intemodes on Time of Flowering

of F^ Plants.

(Times of Flowering of Boantiful, Autocrat and Fj are given for purposes of comparison.)

Numbers of Plants iu Flower

Thick
Long

Thick
Short

Thin
Long

Thin
Short

Date of
Flowering

Bountiful

Autocrat

June 2

—

June 10

—

June 17

—

June 21

—

June 30

—

July 6

(a few
beginning
to flower)

These results show that plants with long intemodes, of both thick-
and thin-stemmed types, flower, on the whole, earlier than the short
internode types. Since long internode is dominant to short internode,
^1 plants may be subject to the same accelerating influence with respect
to time of flowering as those of the groups, long thick and long thin.

54 Inheritance in Pisum

We suggest therefore that lateness is dominaut to earliness, and that
the reason why the plants of F^ flower before those of the late parent
Autocrat, is that they possess the factor for long internodes, their
gametic constitution being TtLl. The fact that the position of
the flowers on the stem, in F^ plants, is about the same as in
Autocrat, lends some support to the view that late is dominant to
early. There was however a considerable range of variation in this
character, both in Autocrat and in the F^ plants. Moreover, owing to
the branched habit of Autocrat and of the ^i plants, and to the fact
that this character was not considered until rather late in the season,
it was impossible in some cases to recognise the main stem, i.e. the
stem which flowered first, and this may have spoilt the records to
some extent. This character of the position of the first flowers on the
stem has been supposed to indicate time of flowering (1905b). Our
records show that Autocrat bears its first flowers, on the average, at the
thirteenth node. Bountiful at the seventh node, and the ^i plants at
the twelfth. Many records of this character were made among the F^
plants. The average result of these records points to the conclusion
that low-flowering indicates earliness, high-flowering lateness, but there
were many exceptional cases among individuals. Further investiga-
tions among F^ families, homozygous in respect of the many other charac-
ters involved, should provide a solution to this question.

Proceeding then on the basis that lateness is dominant to earliness,
we observe, in the F^ generation, that the flowering period spreads over
more than a month, from June 2nd — July 6th, that whereas many (36)
plants of F2 flower as early as the early parent, few flower so late as
Autocrat (Table I). That time of flowering is influenced by seasonal
conditions is undoubted ; but the marked differences in flowering-time
between the various plants of F^ show that the mode of influence of a
given season is determined to a surprising degree by internal factors.
A more detailed examination of the distribution of earliness and late-
ness of flowering among the F^ plants, brings out several facts which
lend support to the conclusions that time of flowering, though inherited,
is modified in its expression in the zygote by morphological characters
such as thickness of stem. As we have shown, the F^ plants group
themselves into four classes: — thick long {TL\ thick short {Tl), thin
long {tL), and thin short {tl). If we chose arbitrarily the date of
flowering of Bountiful (June 2) as early and regard for our immediate
purpose all plants flowering after this date as late, we find, on scruti-
nizing the distribution of lateness and earliness among these classes,

F. Keeblb and C. Pellew 65

that most of the thick-stemmed plants with short or long intemodes,
are late (141 late, 6 early); that, of the thin, short-noded plants, 6 are
early and 7 late, and that of the thin, long internoded plants, 24 are
early and 8 are late. It is therefore apparent that there is a relation
between morphological, vegetative characters and period of flowering.
Thick-stemmed plants contain a very high proportion of late plants,
thin-stemmed plants contain an almost equally large excess of earlies.
As has been shown by Bateson, departures from normal, Mendelian
expectation which manifest themselves by discriminate distribution of
a character among the members of a generation, may be interpreted by
the aid of the hypothesis of gametic coupling. Applying this hypothesis,
and assuming that the coupling between thick stem and late factor is
of the 7:1:1:7 order (Bateson, 1909, <yp. cit. p. 159), we arrive at
the following results: —

TLB TLe TIB TU tLB tU

Calcnlated on trihybrid scheme

81 27 27 9 27 9

Observed

109 5 32 1 8 24

Calculated on 7 : 1 coupling ...

996 8-4 83-2 28 9'2 276

2-8

8-4

r= thick stem, L =

=long intemode, E = late flowering.

Though the numbers are not large enough to demonstrate the exist-
ence of 7 : 1 coupling between thick stem and late flowering factor, yet
their general run and fairly close approximation to those expected on
the basis of such coupling make it probable, in our opinion, that these
two factors are connected with one another in this manner.

The nature of the influence exerted by the long-stem factor in
inducing precocity of flowering we are not yet prepared to discuss, nor
can we deal with the general physiological problems suggested by these
observations ; but will content oureelves with pointing out that, before
a full analysis of physiological properties such as those of time of
flowering can be made, a not inconsiderable amount of breeding work
must be done with the preliminary object of obtaining suitable material,
i.e. material which consists of groups alike in all respects save in the
one which it is proposed to investigate. We learn from the foregoing
preliminary experiments that it is not enough to cross any late with
any early pea, for, as is indicated by these experiments, lateness and
earliness are connected, in a manner not to be suspected on a priori
grounds, with definite, morphological, vegetative characters.

The authors have pleasure in acknowledging that in carrying out
their experiments they have been aided by a grant from the Royal
Society.

Inheritance in Pisum

TABLE II.

Autocrat x Bountiful and Reciprocal Cross.
Description of F^ plants of Heights between 1^ and 4 feet.

Label

Date

Flowering

Height

Stem

Length
Internode

FoUage

I>= Plants
reckoned
as Dwarf

5 X 3/5/10

June 2nd

li feet

Thin

3 inches

Very small

3 X 5/2/60

„ 9th

If „

Thin

2| „

Bountiful type

3 X 5/2/43

„ 2nd

2 „

Thin

3 ,,

„ „

5 X 3/6/7

July 6th

2i „

Thin

n „

Very small

5 X 3/6/3

June 2nd

2i „

Thin

2 ,,

Small ( = Bount.)

6 X 3/4/34

„ 2nd

2i „

Thin

2 „

? Intermed.

5 X 3/2/4

„ 2nd

2i „

Thin

2 „

Bountiful

5 X 3/4/19

„ 16th

2i „

Thin

3 „

Small

5 X 3/5/34

„ 30th

2i „

Thick

s „

? Intermed.

5 X 3/6/6

„ 10th

2| „

?Thin

2 „

3 X 5/2/20

,, 2nd

2^—3 feet

Thin

3 „

Bountiful

3 X 5/2/34

„ 21st

24-3 „

? Thin

^ .,

? Autocrat

5 X 3/2/14

„ 16th

2^—3 ,,

? Bountiful

6 X 3/5/12

„ 16th

?Thiu

3 „

Intermed.

5 X 3/5/41

„ 2lBt

?Thin

i „

Diseased

5 X 3/1/5

„ 21st

3— 3^ „

? Thick

3 „

5 X 3/3/5

„ 30th

3-3i „

? Thick

3 — 3^ inche

3 Small

5 X 3/1/10

„ 16th

3i „

Thick

Small ( = Bount.)

3 X 5/2/8

„ 16th

3i— 4 „

Thick

? Autocrat

3 X 5/2/52

„ 2nd

3i— 4 „

? Thick

Autocrat

5 X 3/4/29

„ 21st

3i— 4 „

? Thick

? Autocrat

REFERENCES CITED IN TEXT.

1905. Reports to the Evolution Committee, ii. p. 68.
1905. R. H. Lock. Studies in Plant Breeding in the Tropics, p. 403.
1907. JosT. Plant Physiology. Translation by R. J. Harvey Gibson, p. 375.
1909a. Bateson. Menders Principles of Heredity, p. 19.

1909b. Bateson. Mendel's Principles of Heredity. Translation of " Experiments
in Plant Hybridization," Gregor Mendel, p. 337.

STUDIES IN THE INHERITANCE OF
DOUBLENESS IN FLOWERS.

I. PETUNIA.

By E. R. SAUNDERS,
Lecturer and late Fellow, Neumham College, Cambridge.

The tradition that the production of double flowers is largely a
matter of external conditions has already been shown in the case of
Matthiola to be at variance with the results of breeding experiments
carried on for several years ^ The evidence, on the contrary, clearly
shows that in this case doubleness, like the other characters investigated,
is inherited according to definite laws, and in accordance with the
Mendelian principle of segregation*. With a view to making a com-
parative study of the inheritance of doubleness in plants a series of
experiments has now been undertaken in various other genera. In the
case of Petunia the results have already reached a point at which
a definite statement can be made, and it is with these results that the
following account is concerned.

1 Of the many beliefs still held regarding the occarrenoe of donbles in Stocks, the
only one which I have so far been able to confirm is that seed which has been kept
produces a higher proportion of doubles than that more recently harvested. This appears
to be true to the extent that the seeds destined to give rise to donbles retain their vitality
rather longer than those which give rise to singles. The higher proportion observed
is not therefore due to any effect of age on the constitation of the seed, bat to an original
difference in viability.

* A general statement of these results has already appeared, and a more detailed
account is now in preparation. (See Reports to the Evolution Committee, Royal Society,
n. p. 29, 1905 ; m. p. 44, 1906 ; it. p. 36, 1908.)

68 Double Petunias

The material used in these experiments included the following
forms : —

(1) P. violacea {phcenicia). Flowers deep magenta with very
dark throat. Pollen blue.

(2) P. nyctaginiflora. Flowers white with yellow flush in the
throat. Pollen yellow. Of stouter habit than the preceding species
and with larger flowers.

(3) P. hyhrida grandiflora. Garden hybrids, (a) Flowers magenta
or magenta and white, variously striped or blotched. Corolla plain
edged. Pollen blue, (h) Var. fimhriata. Flowers nearly pure white.
Corolla fringed. Sepals broad and slightly curled. (Lady of the
Lake.)

(4) Countess of Ellesmere. A garden variety. Flowers rose-
coloured with throat nearly white. Pollen white.

The plants were raised from seed. The two species P. violacea and
P. nyctaginiflora and the garden form Countess of Ellesmere are all
single-flowered. The seed from which the grandiflora plants were
raised was stated to yield a proportion of doubles, and a mixture of
singles and doubles was duly obtained. In growers' catalogues it is
generally stated that the seed which is guaranteed to produce doubles
has been obtained from flowers (i.e. singles) artificially fertilised with
the pollen of doubles. This, as will appear presently, seems to be the
only method of producing double-flowered plants from seed (see p. 60).
The proportion of doubles obtainable is variously quoted as 20 — 40 per
cent. The object of the present experiments was to discover under
what circumstances doubles may be expected to occur, and also, if
possible, to determine whether the proportion of doubles obtainable
was constant.

A. Description of the double flower.

The plants which will bear double flowers may be recognised before
the flower expands by the shape of the bud which is short, thick and
blunt, whereas that of the single is long, slender and pointed. In the
single flower we have a simple funnel-shaped corolla, five epipetalous
stamens, and an ovary with a slender style terminating in the expanded
disc of the stigma (see fig. 1). In many cases the connective is pro-
longed above the anthers in the form of a petaloid structure varying in
size from a short process so small as to be easily overlooked after the

E. R Saunders 59

anthers have dehisced to flat expansions of considerable size (see fig. 7).
But in these cases the stamens, always five in number, are otherwise
normal. The gynoecium is also normal, and the corolla forms one
petaloid funnel-shaped structure. The flower is obviously single. In
the doubles the flower tube is filled with a number of additional
petaloid structures and stamens (see figs. 2 and 3), or in rare cases
mostly with additional stamens (see figs. 4 and 5). These extra
petaloid structures are often variously folded, generally flat but oc-
casionally funnel-shaped, more or less adherent below and free above.
When folded the more deeply coloured, morphologically upper surfaces
are generally opposed, the less deeply coloured, often hairy under
surfaces being outside; but in the open flower the expanded upper
portions of these structures come to lie for the most part with the
upper surface exposed to view, thus giving a uniform colour eflfect.
They vary considerably in size and number even in the different
flowers on one individual. Many bear anther-like structures con-
taining pollen, and some have occasionally been found with a structure
resembling a stigma. The number of stamens proper is also variable,
being usually more numerous in flowers with few petaloid structures and
vice versa. The several members of the corolla and androecium may
fuse to form an outer, single, conspicuous, and somewhat massive
envelope, within which are concealed much smaller petal-like structures
and stamens forming a central mass, which may arise at a distinctly
higher level than the outer envelope owing to the development of an
intemode. Or they may form three or four well-developed envelopes
composed of petal-like structures and adherent stamens which can be
successively peeled off. A further important characteristic of the
double flower is the malformation of the gyneecium. The whole
structure is often completely deformed, but when this is not the case
and the style and stigma appear to be normal, the ovary is seen to be
larger than in the single, and when opened is found to contain perianth
parts, stamens with well-formed pollen, and in some cases also ovules
below or among these other structures. All attempts to use the doubles
as seed-parents however proved unsuccessful. Fertilisation produced no
result. Hence the double character could only be introduced into the
pedigree on the male side.

The flowers on any individual are of one type, either all single or
all double as the case may be. Among a large number of flowers from
double-flowered plants only one was found in which both corolla and
androecium appeared to be single, and in this case the flower was

60 Double Petunias

malformed, the corolla being split and the segments curled ; the ovary
was not opened. The remaining flowers on the plant showed the usual
degree of doubleness. Among the flowers of single plants only two were
observed in which there was any approach to doubling, and in each case
the remaining flowers on the individual were normal singles. In one of
these flowers a single large petaloid structure had developed in the
corolla tube ; in the other a similar structure arose near each of the five
stamens, the line of adhesion to the corolla coinciding with that of
a stamen and forming a common decurrent ridge. It was noticed that
in single plants kept through the winter under unfavourable conditions
the first flowers produced in the following spring were often deformed,
the corolla being split and infolded but without showing any tendency
towards doubleness.

B. Results of breeding experiments.

The general results of the experiments carried on during the last
five years may be briefly stated as follows: —

1. When a single is crossed with a double, doubles as well as
singles occur in the first {F^ generation.

2. When such F^ singles are self- fertilised^ or fertilised inter se the
resulting offspring are all single. Doubles in fact are only obtained
when the pollen of doubles has been used to fertilise the seed-parent,
so that this operation must be repeated in each generation.

3. The proportion of singles in a mixed family is probably always
in excess of the doubles.

Details of the experiments are given in the accompanying Tables.
The results recorded in Tables I and II show that singles, whether
belonging to one of the type forms or derived from a previous cross,
when fertilised with pollen from a double yield a mixture of singles and
doubles in the first generation. Out of a total of 41 families thus bred,
40 included some doubles. As regards the remaining case in which no
doubles were recorded there is little doubt that their absence is due
solely to the small size of the family (4), and that a larger sowing
would have given the usual mixture.

^ If protected under muslin or glass and left undisturbed violaeea and hybrida rarely
set seed ; even when artificially self-fertilised many pollinations give no result. On the
other hand nyctaginijlora, under the same conditions will often set seed, and does so
readily when artificially fertilised with its own poUen. Further experiments concerning
the sterility of these forms are now in progress.

E. R Saunders

It also seems clear that in such mixed families the singles pre-
ponderate. This was the case in 33 out of 38 families, and although
in the remaining five the doubles were equal in number to the singles
or slightly in excess, it is very improbable that the deficiency of
singles in these cases is real. In families 9, 10, 31, and 33 the numbers
recorded are too small to be conclusive, and in family 35 the result
(9 single, 11 double) is within the range of deviation which might
be expected to occur, if, as appears to be the case in several families,
the true ratio represents but a slight excess the other way. At present
the data available are hardly suflScient to determine with certainty the
real proportion of singles and doubles occurring in these families. Until
the general occurrence of doubles in unions of this kind had been
established the number rather than the size of the families was of first
considei-ation. If for the moment, however, we consider only those
families with more than 10 members we find that they fall naturally
into two principal groups, in one of which the numbers suggest the
possible ratio 3 s. : 1 d., while in the other they approximate closely
to the ratio 9 s. : 7 d. Grouping these families in this way we get the
result shown below : —

Beferenoe
number of

Number
of offspring

Reference

number of

family

Number
of offspring

Single Double

Reference

number of

family

Number
of offspring

famUy

Single

Doable

Single Doable

22 11

34 17

9
17

6
13

28
29

18
13
13

13
11

32
35
36
38
40

24
9
53
16
14

21
11
35
14

Total

136

397

307

Where a ratio
ofSs.: Id. cal-
culated to the
nearest whole
number would

Where a ratio
of 9 s.: 7 d. cal-
culated to the
nearest whole
number would

give

125

give

308

62 Double Petunias

9 families giving a total of 136 single, 31 double where a ratio of
3 s. : 1 d. would give 125 single, 42 double.

16 families giving a total of 397 single, 307 double where a ratio of
9 s. : 7 d, would give 396 single, 308 double.

2 families not included in either of the above groups giving a pro-
portion of 2 single : 1 double.

As yet it is not clear whether the occun-ence of these different
ratios indicates that more than one factor is concerned in determining
singleness and doubleness, or whether it results from the fact that
the proportion of germ cells carrying singleness and doubleness varies
in different individuals. In view of the results obtained with Stocks,
the former explanation seems the more likely.

The results given in Tables III and IV show that singles belonging
to the various type forms, whether self- fertilised or crossed with another
type yield only singles (see Table III); and further, that cross-bred
singles having one parent single and one double are equally unable
to produce doubles when self-fertilised or fertilised inter se (see
Table IV), although the same individuals yield both singles and doubles
when crossed with pollen from a double.

It would therefore appear that the pollen of all the singles tested
(23) was homogeneous as regards the presence of some factor x which
is essential to the manifestation of singleness, and which is absent
from some at least of the ovules. Whether the female germs are
homogeneous in this respect, and are all thus deficient ; or whether they
are heterogeneous, some lacking the necessary factor and some not is at
present uncertain. Precisely the same may be stated in regard to the
pollen of the doubles. In some of the grains some necessary factor is
evidently wanting, but whether this is the case in all the male germs
is not yet clear. It may however be safely asserted that whichever
alternative represents the true condition as regards the ovules in the
single, the converse will be found to hold good for the pollen of the
doubles. For the results obtained would equally follow whether it
were the ovules of the single which were homogeneous and the pollen
of the doubles that was heterogeneous, or whether the reverse were the
case. Analogy with Stocks^ would suggest that the first-mentioned

^ In the account of the results obtained with Stocks {Evolution Reports, loc. cit. ) it is
stated that the homogeneous pollen of the heterozygous (ever-sporting) singles carries
doubleness (i.e. absence of singleness), but that among the ovules some carry doubleness
and some singleness. This mode of expressing the difference in constitution between the
male and female germs is permissible if we suppose that the occurrence of singleness or

E. R Saunders 63

alternative may be likely to prove correct (viz. ovules of singles hetero-
geneous, pollen of doubles homogeneous as regards absence of the factor
x) ; but the fact that if this were so we should expect a certain pro-
portion of Petunia singles to be homozygous as to singleness, and
therefore incapable of yielding doubles when crossed with the pollen of
a double, whereas, as a matter of fact, no such singles were met with,
lends considerable support to the opposite view (viz. ovules of singles
homogeneous, pollen of doubles heterogeneous in regard to absence
of a;).

Thus we find in Petunia the same peculiar type of gametogenesis
which has already been shown to occur in Matthiola. In both cases
segregation proceeds in such a way that certain factors are distributed
differently among the ovules and the pollen grains. It may also be noted
that in both instances doubleness behaves as the recessive character,
singleness as the dominant, but in other respects the two cases present
an interesting contrast. In the double Stock, as is well known, the
flower is completely sterile, whereas in Petunia the male organs are
functional in the double though the female are not. Further it appears
that although both in the single Stock which constantly throws doubles,
and in the single Petunia which yield doubles when fertilised by a
double, the pollen is homogeneous in respect of some factor needed to
produce singleness, the homogeneity is brought about by the absence of
this factor in the Stock, by its presence in Petunia. Consequently
doubles are obtained in the Stock when heterozygous individuals are
self-fertilised, or fertilised inter se, but not in Petunia. Lastly, in the
Stock a heterozygous single fertilised with double-carrying pollen yields
an excess of doubles ; in Petunia on the other hand singles crossed with
pollen from a double yield a majority of singles.

Summary.

1. Single Petunias belonging to the following forms : P. molacea,
P. nyctaginiflora, P. hybrida grandifhra, and Countess of Ellesmere,
whether self-fertilised or crossed with each other, yield only singles.

doableness is determined by the presence or absence respectively of a single factor. Now
however that the accumulated evidence points to the probability that more than one factor
is involved this difference between the male and female germs is more correctly expressed
in terms of some factor the. presence of which is essential to singleness (as above in Petunia)
than in terms of the character singleness itself.

64 Double Petunias

2. Cross-bred singles derived from one single and one double parent
also produce only singles when self-fertilised or fertilised inter se.

3. Singles crossed with pollen from a double yield doubles in the
first generation.

4. In families containing a mixture of singles and doubles, the
singles are in excess of the doubles. There is some evidence to show
that in some cases the ratio approximates to 9 s. : 7 d. and in others to
3 s. : 1 d. The occurrence of the ratio 9 s. : 7 d. in many of the cross-
bred families strongly suggests that more than one factor is concerned
in determining the occurrence of singles and doubles, and this view is
in harmony witii the conclusions formed in the case of Stocks.

5. The male organs are functional in doubles, but the gynoecium is
more or less deformed, and when fertilised yields no seed, hence the
double character can only be introduced on the male side.

6. Doubleness behaves as the recessive, singleness as the dominant
character.

7. Gametogenesis is of the peculiar type which has already been
shown to occur in Matthiola, the factors for singleness and doubleness
being distributed differently among the ovules and the pollen grains.

8. The pollen of the singles is homogeneous as regards the presence
of some factor essential to the manifestation of singleness.

9. With regard to the constitution of the ovules of the singles and
the pollen of the doubles it may be said that the results obtained on
crossing are such as would occur, if either the ovules were homogeneous
and the pollen heterogeneous as regards the absence of some factor
needed to produce singleness ; or if conversely the ovules were hetero-
geneous and the pollen homogeneous in respect of this factor. The
fact that all the singles appeared capable of yielding doubles when
crossed with the pollen of a double points strongly to the first alternative,
but the impossibility of making reciprocal crosses renders direct proof
difficult.

The expenses incurred in connection with these experiments have
been in part defrayed by a grant from the British Association for the
Advancement of Science.

E. R. Saunders

TABLE I.

Showing the mixture of singles and doubles obtained in /*,, in the ease
of the type forms^ from the cross single 9 x double $ .

Form of onion

Reference
number
of family

f rur»»^»»

Single

seed-parent

Double

pollen parent

x^^mnbcr Oi v^ud|/iuik

Single

Doable

»»

„

»j

»>

»»

„

—

„

»»

♦»

»f

C£

H (var. ^mbn'ata)

»»

»>

»»

»>

V=xnolaeea. N = nyctaginiflora. C£=Goante88 of Ellesmere. H=kybrida gran-
diflora.

Jonrn. of Oen. i

Double Petunias

TABLE 11.

Showing a similar mixture of singles and doubles resulting from the union
single $ x double S , where one or both of the individuals employed was
descended from a previous cross.

Form oi

' union

Reference
number
of family

^fnmH*»r '^' C\fFoT\t^net

Single

seed-parent

Double

pollen-parent

J.1 uiiiUcr

Single

Double

(single

.HxN)x double H

agle HxN)x double H

double H

»»

„

(single HxN)x Self

double H

(^x doubled)

{Nxdonhle H)

(single HxN)

double H

5 9

The total number of individuals belonging to the type forms used as seed-parents in
experiments 1 — 41 was as follows :

6 plants of Violacea

7 „ ,, Nyctaginiflora

5 ,, ,, Countess of Ellesmere
5 ,, „ Hyhrida grandiflora

Total 23

E. R Saunders 67

TABLE III.

Showing that doubles do not occur when singles belonging to the
variotis type forms are self-fertilised or intercrossed.

Ftvm (tf onion

Single

pollen-parent

Reference

number of

famUy

Nomber of Offs;
Single

'ring

Single

•eed-parent

Doabl

self

—

»»

„

—

»t

i«

—

„

45 -

. —

self

—

self

48 .

—

self

50 .

—

»♦

—

i»

—

„

»»

—

»»

—

„

—

••

—

„

—

many (total not recorded)

—

143

—

„

»i

—

»>

—

»t

—

»»

many (total not recorded)

—

i> II >i

—

{NxV}

(NxV)

—

{NxV)

self

—

„

—

„

•t

—

„

—

„

—

73
74

2
2

[single flxxV)

self

—

••

—

Double Petunias

TABLE IV.

Showing that doubles do not occur when the singles derived from a cross with
a double are either self-fertilised, or crossed with other singles similarly
derived.

Form of union

Single

seed-parent

(single H x double H)

Single

pollen-parent

self

(^■x double H)
»>

>»
(single HxN)x double H

>>
(single H x double H)

(single H xN) X double H
(single H xN)x double H

self

(single H x double H)

(single HxN)x double H
N X double (NxH=si, double)

Reference
numl>er

Number of Otfspring

of family

Single

Double

—

* A double which occurred in this batch was evidently a rogue as the flower had some
of the characters of nyctaginifiora.

EXPLANATION OF FIGURES.

I am indebted for the accompanying figures to Miss D. F. M. Pertz, to whom I here
tender my best thanks.

Fig. 1. Single flower seen split longitudinally.
Fig. 2. Usual type of double flower showing extreme petalody, seen from above. The

functional stamens are concealed by petaloid structures. (See next figure.)
Fig. 3. Similar flower seen in longitudinal section.
Pig. 4. Less common type of double flower. Stamens numerous, but supernumerary

petaloid structures few and small. The corolla tube is curiously folded so as to

form a kind of cup round the stamens. (See next figure.)
Fig. 5. Same flower in longitudinal section. Between the lower region of the corolla tube

which rises vertically, and the upper part which lies horizontally is seen the curious

double bend which forms the cup-like structure surrounding the stamens. The ovary

is aborted.
Fig. 6. Group of stamens and a small supernumerary petaloid structure belonging to

the same flower showing fusion for some distance above the point at which they

become free from the corolla tube.
Fig. 7. Two stamens showing prolongation of the connective.

E. R. Saunders

Fig. 6.

Fig. 3-

Fig. 4-

Fig. 5-

THE EFFECTS OF ONE-SIDED OVAEIOTOMY ON
THE SEX OF THE OFFSPEING.

By L. DONCASTER,

Fellow of King's College, Cambridge,

AND F. H. A. MARSHALL,

Fellow of Christ's College, Cambridge.

{From the Physiology Laboratory, Cambridge.)

It is now widely believed that sex is determined not by conditions
acting upon the organism after fertilisation, but by determinants or
" factors " existing in the gametes themselves. Since this view came
into prominence several hypotheses have been put forward, suggesting
that gametes bearing the factor for one or the other sex are produced in
separate gonads. Some have believed that in vertebrates one testis
yields male-producing spermatozoa, the other female-producing, but this
has been disproved in rats by Copeman^ It is also known to stock
breeders that bulls from which one testicle has been removed, give calves
of both sexes. Meanwhile evidence has been accumulated that in
several groups of animals it is the egg rather than the spermatozoon
which plays the more important part in sex-determination, and in
accordance with this, the opinion has been held that one ovary produces
female eggs, the other male eggs. That this is not a general rule is
proved by the case of birds, which have only one ovary, and in
Amphibia by the experiments of H. D. King^, but in a recent book^
Dr Rumley Dawson has maintained that this hypothesis is valid at least
for man, and probably for other mammals. Direct evidence of a con-

1 Experiments described at the Physiological Society, May 1908.

2 Biol. Bulletin, xvi. p. 27, 1909.

^ The Causation of Sex, London, 1909.

L. DONCASTER AND F. H. A. MARSHALL 71

elusive kind is difficult to obtain in man, since even if children of both
sexes are bom after single ovariotomy, it is rarely possible to prove that
the ovary has been completely removed. It therefore seemed worth
while to test the matter critically in some other mammal, and with that
object the experiments described below were made on rats.

Two female albino rats were taken, and in May 1910 the right ovary
with the greater part of the fallopian tube was removed from one of
them, and the same parts from the left side of the other. Both animals
rapidly recovered from the operation and on being put with a buck,
shortly became pregnant. The female from which the right ovary was
removed gave birth to seven young on July 8. The young all died soon
after birth, and one of them was almost entirely eaten by the mother.
The rest were preserved for examination, and it was found on dissection
that there were four females, one male, and one was too much decom-
posed before being preserved for its sex to be determined with certainty;
it appeared to be a female.

The rat from which the left ovary had been removed gave birth to
five young on July 28 ; one young died shortly after birth ; it was dis-
sected when quite fresh and proved to be a male. The remainder lived
until August 22 when they were killed and dissected ; there were three
females and one male, giving three females and two males in all. On
the same day the two rats which had been operated on were killed and
dissected. In neither could any trace of ovary or ovarian tissue be
found on the side from which the ovary had been removed. In that
from which the left ovary was taken out there was about ^ inch of
fallopian tube, ending apparently blindly; in the other the right
fallopian tube had been cut ofif at its junction with the uterus. In each
case the uteri were normal. They were congested on both sides in the
rat lacking the right ovary, which was probably on heat at the time of
killing. In the female (left ovary removed) which had suckled its
young up to the time of killing all the mammae on both sides were
normal and functional. In both rats the remaining ovary was ex-
ceedingly large, and had doubtless undergone compensatory hypertrophy
in consequence of the removal of the ovary of the other side^ The
relatively large size of the litters (7 and 5) produced from one ovary
may be thus accounted for. That the litters were produced from one
ovary in each case is further shown by the fact that on microscopic
examination it was found that in the rat from which the right ovary was

^ Cf. Carmichael and Marshall, Journal of Phytiology, voL xxxvi. p. 431.

72 Ovariotomy and Sex

removed the remaining (left) ovary contained at least seven corpora
lutea, and the remaining (right) ovary of the second rat contained at
least eight. These corpora lutea were all of similar age in each animal,
and clearly distinguishable from the older luteal tissue present in the
ovaries.

These facts seem to us to indicate without any doubt that in the rat
it is not true that ova determining one sex are produced from one
ovary, and those determining the opposite sex from the other, for each
rat, with one ovary completely removed, produced young of both sexes.
This does not of course prove that the " right and left ovary hypothesis"
is not true for man, but its definite disproof for another mammal detracts
from its probability. It should be pointed out however that the
evidence for alternate male and female ovulations in man, collected by
Dr Rum ley Dawson and others, is not in any way affected. In our
opinion the weakest part of his evidence is that dealing with the pro-
duction of ova determining different sexes by the two ovaries, and it is
not impossible that this hypothesis may be false, and yet that in
general alternate ovulations may be of different sex, so making sex-
prediction possible. It is very desirable that those who have extensive
opportunities of testing this hypothesis — which involves knowing not
only the date of birth and whether the child is " full time " in each
case, but also whether the menstrual periods are normal and regular —
should have the matter in mind and keep records whenever possible.

[Note. The operations described were performed by F. H. A.
Marshall ; the dissections by L. Doncaster.]

Volume I MARCH, 1911 No. 2

EXPERIMENTS WITH PRIMULA SINENSIS.

By R. p. GREGORY, M.A.,
Fellow of St John's College, Cambridge ; University Lecturer in Botany.

CONTENTS.

Page

istboduction 74

Hbtebostylism 78

Abnormal cases 84

Leaf-Shape 86

Palmate and Fern-leaf 87

Ivy-leaf 87

Habit 88

Double Flowers 89

Inheritance of ordinary double 91

Chaeactebs of the "Eye" of the Floweb .... 91

Large yellow eye x small eye 92

White eye x small yeUow eye 94

White eye x large yellow eye 94

GoLOUB 94

A. Stkh-Coloubs 95

Inheritance of Stem-colonrs 96

Partial Suppression of Colour 100

B. Floweb-Coloubs 101

Inheritance of Flower-colours 103

Partial Suppression of Colour 105

Inhibition 106

Experimental results : (1) Pale colours . . 108

(2) Full colours ... 109

(3) Inhibition . . .115

(4) Flakes . . . .121

Gametic Cocpldig and Repulsiox 124

Descbiption of Plates 130

Jonm. of Gen. i

74 J^xperiments with Primula sinensis

Introduction.

The experiments, of which the present paper is the outcome, were
begun in 1903 by Mr Bateson and the present writer jointly, and in
1905 we published an account of our observations up to that time upon
the inheritance of heterostylism^ Although I am alone responsible
for the views put forward in the present paper, and for any errors
which it may contain, the work with which it deals has been done in
association with Mr Bateson, to whom much of such progress as has
been made is due. Mr Bateson has given me the most generous help,
not only in the elucidation of the results, but also in the practical
business of carrying on the experiments. I am further indebted to
him for giving house room to a large number of plants each year.

The plates illustrating the various coloured forms which have been
met with in the course of the experiments are reproduced from the
beautiful and accurate water-colour drawings of Miss M. Wheldale, of
Newnham College, Cambridge.

I wish to take this opportunity of acknowledging again my in-
debtedness to Messrs Sutton and Sons, who have most kindly given
assistance in many ways during the course of this enquiry.

My thanks are due also to the Botanic Garden Syndicate of Cam-
bridge University, and to Mr R. I. Lynch, Curator of the Botanic
Gardens, for the provision of housing, materials and labour.

The principal objects of our investigations in Primula sinensis have
been the inheritance of heterostylism and of colour. At the same time
records have been kept of certain other characters, the inheritance of
which has been found to be, for the most part, of a simple type and
does not require any special comment here^

Heterostylism^. The dearth of short-styled plants occurring in the
families raised from the self-fertilized heterozygote, which was noticed
in our earlier experiments, is still maintained even in the larger
numbers now obtained. On the other hand the same plants, crossed
by the long-styled, give an excess of short-styled offspring. Our results
do not as yet give a decisive answer as to whether these divergences,

1 Bateson and Gregory, Roy. Soc. Proc. B, Vol. 76, 1905, pp. 581—586.

'■^ Some of these results have already been mentioned; see Bateson : "The progress of
Genetics since the rediscovery of Mendel's papers," Prog. Rei. Bot., Vol. i. 1907, pp. 373,
383 ; Mendel's Principles of Heredity, Camb. Univ. Press, 1909. Gregory: "The inherit-
ance of certain characters in Primula sinensis," Brit. Assoc. Rep., Leicester, 1907,
pp. 691—693.

3 Bateson and Gregory, I.e.

R. P. Gregory 76

in opposite directions in the two cases, are to be regarded as merely
accidental, or whether they may have some significance, either in
connexion with observed differences in the fertility of the various
unions between plants of different form, or in other ways (p. 83).

Colour. The colour of the stems and flowers in the coloured races
is due to the presence of coloured sap. The colour may be absent
from the flowers, which are then white, or from the stems, which are
then green. Colour, both in flower and stem, is presumably produced,
as in other cases, by the interaction of two or more complementary
fectors. I have had no decisive case of the production of an ^i with
coloured flowers from the mating of two albinos, but Keeble and
Pellew^ record a coloured Fi from the mating of the red-stemmed
"Snow King" with the green-stemmed "Snowdrift." Similarly as
regards the stem-colours, I have no example of the production of a
coloured ^i from the mating of two green-stemmed plants, but in two
cases (p. 97) heterozygous plants with coloured stems have given
unmistakably the ratio 9 coloured : 7 green stem.

There exist several distinct types of coloration, both of the stem
and of the flowers. Thus, the stem may be fully and evenly coloured
(Plate XXX, figs. 1, 2), or it may possess only a faint colour, which is most
easily recognized in the young leaves and leaf-stalks (Plate XXX, fig. 5).
The faint colour is, in some cases, an elusive character, and the plants
bearing it are only with difficulty to be distinguished from those
devoid of colour in the stem. The inheritance of these two kinds of
pigmentation of the stem may be explained most simply if we assume
the existence of two separate and independent chromogen factors, each
of which reacts with the common activator to produce, one the full
colour, the other the faint colour (p. 96).

The colours of the flowers and stems are inter-related in such a way
that the more deeply coloured flowers never occur in conjunction with
stems wholly green. Flower-colours may then be divided into two
classes, namely, full colours, which are found only on plants having
fiilly coloured stems ; and pale colours, which occur on plants having
green or faintly coloured stems. White flowers may be associated with
stems of any kind.

When the albino " Snowdrift " (Plate XXX, fig. 7) was crossed with
types having fully coloured flowers and stems, the ^2 contained only
one real albino to every fifteen pigmented forms. These coloured forms
were of three kinds, (1) full colours on red stems, (2) a type known in

> Journ. of GeneticM, Vol. i. 1910, p. 4.

6—2

76 Experiments with Primula sinensis

horticulture as " Sirdar" (Plate XXX, fig. 4 ; Plate XXXI, figs. 44, 45),
(3) pale colours on faintly coloured or green stems. The " Sirdars " have
a peculiar distribution of the colour. The pigment of the petals is one
of the full colours, but it occurs in separate minute dots and the edges
of the petals are white. Associated with flowers of this kind, the stems
have pigment at the bases of the petioles and pedicels, the rest of the
stem and leaves being green. The inheritance of the " Sirdar "
character may be described conveniently if the " Sirdars " be looked
upon as belonging to the fully coloured series, while they lack a factor,
the presence of which is required to bring about the even distribution
of the colour which is found in the full colours. The full colours and
" Sirdars " together constitute three-fourths of the total F^ population.
The remaining one-fourth consists of pale colours and whites in the
ratio 3 : 1. The significance of the ratio 15 pigmented forms : 1 albino,
and the relation of the pale colours to the full colours, is discussed in
the text (pp. 103, 104).

The full colours are divisible into three classes, namely, (1) shades
of magenta, (2) shades of red or crimson, (3) shades of blue.

The pale flower-colour is always a shade of pink, never magenta or
red. This colour, in its deepest shade, is that of Sutton's " Reading
Pink " (Plate XXX, fig. 13).

Full colours are dominant to pale colour; magentas are dominant to
reds, and blue is recessive to all magentas and reds.

Whites may be dominant or recessive to colours.

Suppression of colour, partial or complete, by dominant factors is a
common phenomenon in Primula sinensis. Some of these factors affect
the colour of the flowers only, and one, at least, affects the colour of
both flowers and stems.

When plants, which otherwise would have coloured flowers, are
homozygous in the factors which suppress flower-colour, the flowers
are quite white (dominant whites) ; when they are heterozygous in the
inhibiting factors, the flowers are sometimes white, but are more often
tinged with colour, the depth of the tinge varying with the races used
and with the temperature of the house.

As regards the suppression of flower-colour, the evidence reveals a
curious complication in that the operation of two inhibiting factors,
affecting distinct areas, can be separately traced. Of these factors, one
suppresses colour in the peripheral parts of the corolla, the other affects
the gynoecium and central part of the corolla. In consequence it follows
that in F^ from fully coloured plants with coloured stigmas x dominant

R P. Gregory 77

whites, there appears the peculiar type knowD as "Duchess" (Plate XXXI,
figs. 27, 28), in which the flower is white peripherally and has a coloured
centre. The mating of " Duchess " with plants having coloured flowers
and green stigmas, gives a tinged white F^, exactly like that produced
by the mating of coloured, red stigma x dominant white.

Various light shades of flower-colour behave as dominants to the
deep shades ; this dominance is due to the presence of factors which
efifect the partial suppression of the colour. These factors are quite
distinct, as regards their inheritance, from those described in the pre-
ceding paragraph.

Similarly, the light shades of stem-colour are dominant to the deep
shades. The suppression of stem-colour is only partial, even in plants
homozygous for the suppressing factors, and no dominant green stems
are known.

Flowers of a light shade may be borne by plants having deeply
coloured stems, but the deep flower-colours never occur on stems not
deeply coloured. It is clear, therefore, that the factor which effects
the partial suppression of stem-colour exerts its action also upon the
flower-colour.

Besides the varieties constituted by combinations of the factors
already enumerated, there occur various types having flakes or patches
of colour (Plate XXXI, figs. 56 — 59) ^ As in other oases where such
flaking has been encountered, the genetics of these varieties is not
altogether clear, but in the case of Primula sinensis, as will be seen
on reference to the text (p. 122), it is possible to frame a hypothesis
which would give results consistent with those observed.

Gametic Coupling and Repulsion (p. 124). Complete repulsion
between the factor for the structural character of short-style and the
magenta colour-factor was observed in a series of experiments in which
short-styled salmon-pinks were mated with various long-styled plants
carrying the magenta factor.

The cases of partial gametic coupling which have been met with
are interesting in that, in many of them, the two middle terms of the
F^ series are much larger, relatively to the end terms, than they are in
the majority of cases previously recorded. In the case of the coupling
between magenta colour and green stigma, the results of several
experiments approximate closely to the expectation based on the
hypothesis that a coupling of the form 7:1:1:7 is present in the

^ Flaked forms mnst be carefully distingaished from " Sirdars."

78 Experime7its with Primula sinensis

gametes of one sex only, those of the opposite sex consisting of equal
numbers of the four kinds (p. 128). Other cases however are apparently
not susceptible of complete explanation on these lines, and it seems
possible that they may indicate the existence of lower forms of coupling
than any given by the gametic series

n-\ : 1 : 1 : w-l.»

Further experiment however is needed before any definite opinion
can be expressed upon this point.

The history of P. sinensis, since its introduction into England in
1820, has been given by Mr A. W. Sutton^ and further notices by other
writers have appeared from time to time^ It is interesting to notice
that the earliest illustrations^ of the species represent short-styled
plants of the ordinary habit (not stellata) with palmate leaves, light red
stems, and light magenta flowers — all dominant characters.

Heterostylism.

In an earlier report Mr Bateson and the writer showed that the
inheritance of the characters of long and short style is of a simple
Mendelian type, the short style being dominant®.

All the short-styled plants originally obtained for the purpose of
these experiments proved to be heterozygous, but from their progeny
pure short-styled plants have now been obtained. Nine such plants

^ Bateson, Saunders and Punnett, Rep. Evol. Comm. Roy. Soc. iv. 1908, p. 3.
Lower series would be given by the general expression

n-x : X : X : n-x
where x is any odd number less than - . The expression may be made a general one,

including all forms of partial repulsion as well as coupling, if x be taken as any number
less than n. The F2 series would then be given by the expression

371^ - X (2n -x) : X (2n -x) : x (2n -x) : (n - x)^.

^ Journ. Roy. Hort. Soc. Mar. 1891, xiii. p. 99.

3 Gard. Chron. 1889, p. 115 ; Ibid. 1890, p. 564 ; Ibid. 1892, p. 12; Ibid. 1902, p. 269.

* Bot. Reg. 539, May 1, 1821, under the name P. praenitens, and Lindley's Collectanea
Botaniea, Tab. VII, 1821. The plants figured in the two works are clearly of very similar,
if not identical, types. In Lindley's plate the drawings of the dissected flower apparently
represent the short-styled form ; the flowers shown on the plant have rather the appearance
of long-styled flowers.

5 Bateson and Gregory, Roy. Soc. Proc. B, Vol. 76, 1905, pp. 581—586.

R. P. Gregory 79

have been used as parents : their offspring are shown in the following
table:

Number of

short-styled ^lort- Long-

Croas pUntsnsed styled styled

Pare short-styled X Self 8 252 0

Pure short-styled ? x Long-styled <f ... 5 290 0

Long-styled ? X Pure short-styled (J ... 4 247 0

Totals — 789 0

Several thousand plants have been raised from crosses of (long-
styled X long-styled), all the offspring being long-styled.

Heterozygous short-styled plants.

The results of crosses in which heterozygous short-styled plants
were used are shown in the accompanying table (p. 80).

Although the results are in general harmony with simple expec-
tation, yet the observed numbers diverge rather widely from the
calculated ones*. The divergences are, moreover, in opposite directions
according as the cross is of the type (DR x DR) or of the type
{DR X R) and the reciprocal form. The heterozygous short-styled
plants, self-fertilized, show a deficiency of short-styled oflfspring as
compared with the expected ratio of 3 short : 1 long ; the same plants,
crossed either way with long-styled plants, give an excess of short-styled
oflFspring. In the former case the divergence from the ratio 3 : 1 is
more than twice as great as the probable error of a random sampling
taken from a population mixed in that ratio ; and an equally great
divergence occurs in the results of the crosses {DR x R) and (R x DR)
taken together-.

* The results of two experiments, each of which would have the effect of slightly
increasing the divergence, have been excluded from the totals given in the Table (p. 80).
In each case a long- styled plant ? was crossed by a short-styled cT ; the <? parents
were known to be heterozygous, but the offspring, 9 in the one case, 5 in the other,
consisted of short-styled plants only. The cases are excluded owing to the possibility
that the two short-styled plants were behaving in a manner similar to that of the
abnormal case described on p. 84. One other very aberrant family has also been
excluded, owing to the possibility of error. This family was produced from a mating
(Fj ? X long-styled <f ) and consisted of 27 short-styled and 9 long-styled plants (expected
equality).

' The probable error for random sampling of N individuals of two kinds mixed

in the proportion p : q 'w given by the formnla p.e. = •6745 - — ^ . The errors given in

the table show the probable departures from the exact ratios 3 : 1 and 1 : 1 respectively
for the numbers concerned. I am indebted to Mr A. B. Bruce, of the Cambridge
University Department of Agriculture, for this formnla.

Experiments with Primula sinensis

I-

>~1

•w

■e

•o

«b

«s

-H

0-.

f^l

X -

»o

i-H

i-(

l&g

i-H

«o

S3 o s

i-H

^ 1

"-I

f-i

■^

f-K

•o

a
o

■H

l-H

I-l

o»

X -

ot-

I-H

»o

T»l

£»

1-H

■g

»«i

<j«

lis

H X «

5®S

i-H

^ ^"S

r t

'"i

"-i

»«(

e>3

»,

■»

-H

«5

«o

*?,

■*i -<

■e

«»

iffl

i-(

I:-

l-l<JI«

III

^ iZ°^

1 :

^ 2

■^

00 o

GD a

-*J> o

CQ g)

g*^

•

"S J3

-5 -2

a
'53

■^■^

■1^ a

1
§

>> 02

R. P. Gregory

TMe shotving individual families raised from the cross (DR xDR).

Yeu

Number

in family

Short

Long

Katio
(*:l)

d=(2^1-x

cl<

9»

Fi-$horU X self

1905

2-80

53-21

-11

•1210

2-299

0-91

: 1

19-11

2 00

4^0000

82-000

214

: 1

47-08

-77

•5929

13 044

3-33

86-66

-42

•1764

4-586

2-54

116-90

-37

•1369

6-298

1-82

56-43

1-09

1-1881

36-831

3-60

82-79

-69

-4761

10^950

3-64

185-70

•73

-5329

27-178

4-50

99-00

1-59

2-5281

55-618

3-17

79-25

-26

•0676

1-690

3-39

267-80

-48

-2304

18-202

2-86

77-23

•05

-0025

-068

. 106

2-53

268-18

-38

-1444

15-306

1-25

: 1

33-75

1-66

2-7556

74-401

1906

2-50

210-00

-41

-1661

13-952

2-95

244-80

•04

-0016

-133

4-67

79-38

1-76

3-0976

52-659

2 14

47-08

-77

•5929

13 044

114

2-45

279-40

-46

-2116

24-122

1908

3-36

204-90

•45

•2025

12-353

3-00

7200

•09

•0081

•194

4-57

178-20

1-66

2-7556

107-468

3-44

137-60

•53

•2809

11-236

Other heterozygous shorts x self

1904

4 00

80-00

109

1^1881

23-762

10
10

6
9

4
1

300

ia5 00

•09

•0081

•162

1905

2-40

40-80

•51

•2601

4-422

1906

2-00

96-00

•91

•8-281

39-749

1907

3-00

20400

•09

-0081

-551

1910

2-67

117-50

•24

-0576

2-534

Totals

1226

897

329

84-83

3569-75

—

654-816

Weighted
Probable

mean
error o

ratio = -
f weieh

ted I

3569-75
1226

nean ratio =

1 = 291 :
•6745 A /

Zp<P

= 09.

Batio=2-91=i=-09:l.

Experiments with Primula sinensis

Table

showing

hidividual families raised from

the crosses

{DR

xR)

and

{D X DR).

Number

in family

Ratio

Year

Short

Long

(x:

px d-

= (1-23-

■x) d»

pdi

Fi Short

X Long.

Short ¥

1905

•41

: 1

9^84

-82

•6724

16'1376

1-14

: 1

34-20

•09

•0081

•2430

2-62

: 1

123-14

1^39

1-9321

90-9087

1-25 ;

: 1

22^50

•02

•0004

•0072

•83

: 1

43^01

•40

-1600

5-2800

•67!

; 1

13^40

•56

-3136

6-2720

1-62 :

34-02

•39

•1521

3-1941

1-25 ;

; 1

45-00

•02

•0004

-0144

•65:

18-20

•58

•3364

9-4192

100;

: 1

34-00

•23

•0529

1-7986

-Fi Short

X Long.

Short i

1905

•82 :

: 1

25-42

•41

•1681

5-2111

1-37 ;

; 1

26^03

•14

•0196

•3724

Other Heterozygous Shorts.

Shorts

1904

113 :

1921

•10

•0100

•1700

10
14

7
7

1^40 :

33^60

•17

-0289

•6936

1907

113 :

36^16

•10

•0100

•3200

1^33 :

65^17

•10

-0100

•4900

1910

•87 :

24-36

•36

-1296

3-6288

110 :

23-10

•13

-0169

•3549

Other Heterozygous Shorts.

Short <

1904

17
1

15
0'

113 :

36-16

•10

-0100

•3200

1-38 :

26-22

•15

•0225

•4275

1905

•82 :

16-40

•41

•1681

3^3620

1-00:

24-00

•23

•0529

1^2696

6
15

5
6

1^10:

23-10

-13

•0169

•3549

1907

1^94 :

97-00

•71

•5041

25-2050

1910

•84 :

38-64

•39

•1521

6-9966

Totals

724

382 ;

342 28-80 : :

891-88

—

182-4512

Weighted mean ratio = ^ : 1 =
Zp

Probable error of weighted mean

891-88
" 724 ■ ~

ratio = -6745

123 :

2pd2
:p{n-l)

= -10.

Ratio = 1-23 ±-10 : 1.

R. P. Gregory 83

This would seem to imply the presence of some disturbing cause
affecting the regular Meudelian distribution, but it is important to
ascertain what reliance may be placed on the ratio determined from
the sum of all the families taken together. If the total results be
tested by the discordance of the results in the individual families which
make up the total \ it is found that, in the case of the (DR x DR)
crosses the approach to the normal 3 : 1 ratio is close, the observed
result being 29 1 ± "09 : 1 (Table, p. 81). In the (DR x R) and
(R X DR) crosses the observed result is 1-23 + lO : 1 (Table, p. 82)
the theoretical ratio for 724 plants being 10 + 01 : 1.

Examined in this way, the results obtained at present perhaps
scarcely afford a clear indication as to whether the above noted diver-
gences are to be regarded as merely accidental, or whether they may
have some significance in regard to the observed differences in the
relative fertilities of the various kinds of legitimate and illegitimate
unions.

Any significance, which the foregoing results may have in this connexion, lies in the
possibility that the observed differences in the fertility of the legitimate and illegitimate
onions^ may be, in part, due to differences in the fertility of the various kinds of gametic
anions, or rather (since the results of the matings (DR x R) and (R x DR) are in sub-
stantial agreement) to differences in the mortality of the three kinds of zygotes arising
from these unions.

All the experiments on relative fertility are in agreement in showing that the union
(short-styled plant x short-styled plant) is distinctly the least fertile, while the legitimate
unions are the most fertile. Assuming that all forms of gametic union are equaUy
fertile, the cross (DR x DR) would give offspring in the proportion 1 DD : 2 DR : 1 RR
while the cross (DR x jR) would give 1 DR : 1 RR. But if there are differences in the
fertility of the various kinds of gametic union, the observed deficiency of short-styled
offspring in the cross (DR x DR) might be due to the small number of pure short -styled
plants which are produced, while the excess of short-styled offspring in the cross (DR x R)
might be due to greater fertility of the union (D x R) as compared with that of the union
(R X R).

1 I am greatly indebted to Mr F. J. M. Stratton, of Gonville and Cains College,

Cambridge, for this method of examining the results.

^ See Darwin, Forms of Flowers, pp. 38 — 43, 246. Darwin found that the ratio of the
fertility of the two legitimate unions taken together to that of the two illegitimate unions
was 100 : 53. With this ratio that given by my experiments agrees very closely, but the
fertility of the long-styled form, whether fertilized by its own or by the other form of pollen,
is greater in the case of my plants than that observed by Darwin. The figures are

Long X Short Short x Long Long x Long Short x Short
Average number of seeds per capsule 33 25 21 11

It is to be presumed that the short-styled plants used by Darwin and Hildebrand
included, like mine, heterozygoos as well aa pore individuals.

84 Experiments with Primula sinensis

If we assume for the moment that the observed divergences from the simple Mendelian
ratios are due to differences of this kind, then, if di, 02, 63 represent respectively the
fertilities of the gametic unions Short x Long, Long x Long, and Short x Short, the results
described above would give

$1 :02 : 03=100: 89 : 44.

These figures are quantitatively in general agreement with the relative fertilities, as
determined by the average number of seeds per capsule, of the various kinds of union
between plants of different form, the corresponding figures being 132 or 100 : 84 : 44
(see p. 83, footnote). The comparison must not be pressed too far, since the actual
fertilities of the various unions, observed in any set of experiments, would depend in part
upon racial characters. The agreement is however rather suggestive and, taken in con-
junction with the results of our examination of the observed numbers by other methods, is
suflScient to justify further investigation.

The point can be tested experimentally by determining the constitution of all the
short-styled plants in a number of large F2 families; we should then find whether there is
any significant divergence from the theoretical proportion of 1 pure : 2 heterozygous
short-styled offspring.

Abnormal Cases.

A case was described in the previous report^ in which the entire
series of crosses made with a certain short-styled plant (No. 6/3) showed
a definite and consistent departure from the normal expectation. The
evidence already given showed that No. 6/3 behaved as an ordinary
heterozygous short-styled plant when used as the female parent in
crosses with long-styled plants, while its male gametes almost exclu-
sively bore the dominant character^ The case promised to be of some
interest, but unfortunately all the plants used as parents for succeeding
generations proved to be normal pure short-styled plants, giving short-
styled offspring only, when selfed and crossed either way with long-
styled plants. No further elucidation of the case is therefore possible.
The F^'s from crosses of this race with long-styled plants showed normal
distribution of shorts and longs in the offspring, and are included in the
F^ table given on p. 80. The results of all the crosses in which this
particular race was used are recorded in the tables given on pp. 85, 86.

1 Bateson and Gregory, loc. cit. p. 584.

2 By an unfortunate error the statement made in the first paragraph on p. 585 of the
previous report is inverted. The context makes it clear that the statement should have
read: "the ovules of No. 6 gave a mixture of longs and shorts, and consequently were
of two kinds, while all the plants raised from it as male were shorts."

R. P. Gregory

Table showing the results of crosses made with No. 6/3 and its progeny

in direct descent.

Short-

xSelf

Short-Style

d? xLon

g-styled i

Long-styled

? X Shor

t styled (f

styled

_j_

parent

Number

Short-

Long-

Number

Short-

Long-

Number

Short-

Long-

of family

styled

of family

styled

of famUy

styled

r 37/4

39/4

24/4

40/4

35/4

6/3 J

43/4

72/4

45/4

74/4

/ 126/5

127/5

80/5

128/5

87/5
104/5
143/5

0
0
1

37/4^

149/5

177/5

183/5

205/5

■'■

213/5

149

4(?3)

[ 21/6

[126/5

was not used

[126/5

was used as i

126/5 \

for crosses of

parent

m one

cross.

this

type

which [

^ve no

seeds]

[ 24/7

27/7

2/7

211/6 ^

28/7

25/7

[21V6

was not nsed

3/7

212/6

■

for crosses of this type]

58/7

' 26/7

29/7

4/7

31/7

57/7

21»/6 ^

32/7

122

• Beoorded as "doubtful."

Experiments with Primula sinensis

Table showing the constitution of the F^s raised from crosses in which 6/3
and its progeny were used.

f 1 Short-styled >
{DR X DR)

cSelf

Fi Short-styled ?
styled <? (DR

X Long-
xR)

Long-styled t x
styled <J(iix

F, Short
DR)

*F, Short-
styled plant

Number
of J-',
family

Short-
styled

Long-
styled

Number
of Fi
family

Short-
styled

Long-
styled

Number

of Fa

family

Short-
styled

Long-
styled

241/4

98/5

351/4

124/6

125/5

402/4

182/5

431/4

133/5

(134/5
I135/5

f 96/5

] 142/5

1 146/5

451/4

136/5

(137/5
(188/5

452/4

139/5

652/4*

174/5

721/4

201/5

(202/5
(203/5

182/5

1281/5

22/6

1431/5

24/6t

41/7

7/8

271/7

20/8

281/7

21/8

581/7

28/8

Totals

420

150

* The origin of the Fi plants is shown in the preceding table. 65'V4 was a short-
styled plant raised from the double pollination (see Bateson and Gregory, I.e. p. 585) of
a long- styled ? x self and 6/3.

t Two plants from this family were bred from and gave respectively :

xSelf

241/6 ? X Long-styled (f Long-styled ? x 24i/6<?

241/6
242/6

Short
61
76

Long

0
21

Short
38

Long
0

Short
58

Long
0

Leaf-Shape.

There is a considerable range of variation in the form of the leaf in
Primula sinensis. Besides the common palmate and fern-leaf varieties,
Messrs Sutton have raised a strain in which the peculiar lobing of the
leaf is repeated in the petals, which also somewhat resemble the leaf in
form\ Of other types, the Ivy-leaf is described below ; while I have

1 Roy. Hort. Soc. Journ. Vol. xxxv. Pt. i. 1909, p. xxxvi. The leaves of this variety
are described as approaching those of Ivy ; it may be well, therefore, to point out that the
character is a different one from that of the strain to which I have applied the name of
" Ivy-leaf " in this paper.

R. P. Gregory 87

this year obtained a plant which possesses very deeply palmatiBd
leaves. In addition to these variations, which affect the general aspect
of the leaf, there also occur less noticeable ones ; as an illustration the
case may be cited of a plant, which occurred in an F^ family this year,
the leaves of which had serrate, instead of the usual crenate, margins.

Palmate and Fern-leaP.

The palmate character is dominant, though a slight difference can
sometimes be recognized between the pure and heterozygous palmate
types. The shape of the leaf has been recorded in 27 F^ families raised
from crosses between palm- and fern-leaf, the numbers obtained being
1370 palmate, 457 fern-leaf {expectation : 1370-25 : 4o6-75).

Ivy-leaf.

In 1907 Mr A. W. Hill kindly gave me a monstrous plant (Plate
XXX, fig. 5) which occurred among a batch of seedlings raised by him
from seed obtained from a nurseryman. The leaves are palmate, but the
margins are not crenate, as they are in the ordinary form of leaf This
peculiarity of the leaves is always accompanied by abnormal develop-
ment of the flowers, which are very much reduced. The abnormality
is much more marked in the early flowei's than in the later ones, and if
the plants be grown as biennials or perennials it is generally possible
to obtain good seed from such as survive. A seedling raised from the
original plant is shown in Plate XXXII, fig. 60. It will be seen that
the early leaves (the lower ones in the photograph) have more divided
edges than the later ones, and bear a closer resemblance to the leaves
of the ordinary palmate form.

The absence of crenation of the leaf margin behaves as a recessive
character. The ^i from the cross with the ordinary palmate form is a
normal palmate plant. The F^s raised from {F^ x self) have given 703
palmate, 241 Ivy {expectation : 7080 : 236-0). Crossed with the ordi-
nary fern-leaf, the Ivy-leaf gives again a normal palmate plant (Plate
XXXII, fig 61). This F,, selfed, gives an F^ (Plate XXXII, fig. 61, the
bottom row of plants) consisting of normal palmates, normal fern-leaves,

' Bateson: "The progress of Genetics since the rediscovery of Meadel's papers,"
Prog. Rei. Bot., Bd. 1, 1907, p. 373 ; MendeVt PnncipUt of Heredity, Camb. UniY.
Press, 1909, p. 24.

88 Experiments with Primula sinensis

palmate ivy-leaves and fern ivy-leaves, the numbers obtained at present
being

Palm.

Fern.

Palm-Ivy.

Fern-Ivy.

173

1631

54-4

181

Expectation :

It is clear therefore that we are dealing with two independent
characters, namely, (1) the shape of the leaf and (2) the crenation
of the margin; and it is the absence of the latter character which is
accompanied by the abnormality of the flower structures which is
characteristic of the Ivy-leaved variety.

Considering the character of crenation only, the crosses have given
922 crenate, 312 non-crenate {expectation: 925'5 : 308'5).

Habit.

The hybrid between the typical P. sinensis and the "stellata"
variety is the well-known "pyramidalis" form^

The principal characters in which the parent types differ from one another are :
Sinensis. Stellata.

(1) Inflorescence compact. Early elongation of the main axis above

the primary umbel, with production of
secondary and tertiary umbels.

(2) Shorter pedicels. Long pedicels,

(3) Calyx cylindrical, with numerous Calyx tube narrowing at the top and
teeth ; more or less enclosing the unfolded shorter, so that the corolla protrudes before
corolla. beginning to unfold; calyx teeth = the

number of the petals (5).

(4) Corolla lobes imbricate, crenate. Corolla lobes scarcely, if at all, over-

lapping ; heart-shaped.

The hybrid is intermediate between the two parents; in respect of the characters
of the inflorescence it approaches more nearly to the stellata form ; the calyx has 10 — 15
teeth ; the degree of crenation of the margins of the petals is somewhat variable, but
generally well marked.

From the study of a plant (No. 54/9, see Plate XXXII, fig. 64) which Messrs Sutton kindly
gave me last year, it is clear that a plant, although capable of producing offspring nearly
resembling the sinensis type, may itself approach somewhat nearly to the stellata form.

A series of flowers taken from the plant in question is shown in Plate XXXII, fig. 64,
There is some range of variation in the corollas of individual flowers, some of which are
scarcely crenate at all ; the plant also resembled the stellata form in its elongated axis
and long pedicels.

The plant, when selfed, gave 21 offspring, of which 2 were true stellata, 12 were
clearly intermediate, 7 approached sinensis, but of these seven 3 showed a strong

1 Bateson, MendeVs Principles of Heredity, Camb. Univ. Press, 1909, pp. 26 and 68.

R. P. Gregory

tendency to the development of high spires of flowers, and the corollas protmded from
the calyx in the young bud.

A detailed study of the various characters of these offspring suggests that we are not
yet justified in regarding the differences between the tinensit and ttellata types as
depending upon one factor. If it should prove that the characters of the axis, of the
oalyx and of the corolla may be inherited independently, the character designated here
as Mtellata must be taken to refer to the form of the corolla.

The stellata form used in the great majority of my experiments
was a strain known as "Primrose Queen" (Plate XXX, fig. 12, and
Plate XXXII, figs. 62 and 63, No. 37/9). The F^, resulting from the
cross of this with a plant of the typical sinensis habit, consists of
sinensis, pyramidalis and stellata forms. The original " Ivy-leaf" plant
also proved to be a stellata form. When this plant is used. Ivy-leaves
of course appear in the F^, in addition to the forms already mentioned.
The F, Ivy-leaves are presumably of different forms, corresponding with
the forms met with in the normally developed plants, but, owing to
the poor development of the flowers and inflorescence, it is impossible
to say more than that, in some, the petals were more or less crenate.

It is not easy to draw a sharp line of distinction between the
pyramidalis forms and the true sinensis type ; in the following table
they are therefore grouped together.

The numbers obtained are

/■.xSelf

xStMata.

Number of
/j famUies

Sinensix
and inter-
mediate

Stellata

Ivy

Xumber of
Ft families

Sinensis
and inter-
mediate

Stellata

1030

342

—

Expectation

—

10290

343

—

37-5

Crosses in which
Ivy-leaf was used

t 2
1

151

—

Total

1181

409

—

Expectation

—

1192-5

397-5

—

Double Flowers.

Two types of doubling of the flowers in Primula sinensis are known
to me, in both cases in long-styled plants, though experiments are in
progress which, it is hoped, will give short-styled doubles'.

' Short-styled doubles of the type shown in Text-fig. A, have now (Feb. 1911) been
obtained, in the F^ from short-style, single x long-style, double. So far as the morphology
of the corolla is concerned, the short and long-styled doubles very closely resemble one
another ; they differ of course in the size of the pollen grains and in the length of the
style.

Joom. of Gen. i 7

Experiments with Primula sinensis

In the more common double (Text-fig. A), the supernumerary
segments are inserted at the throat of the tube, one segment occurring
opposite each petal. The anthers are somewhat exsert, and are attached
just at the base of the supernumerary segment ; the position of the
anthers might easily lead one to suppose that the flower was " thrum-
eyed," were it not for the long style and the size of the pollen. The
supernumerary segments are reversed ; that is to say, the external side
resembles the upper (internal) surface of the normal petal, while the
internal side is like the back of the latter^

A B

In the old-fashioned double (Text-fig. B) the doubling is more
complete than in the more usual form, and a number of supernumerary
segments occupy the centre of the flowerl The supernumerary segments
are of different orders; the primary segments are inserted, one opposite
each petal, on the corolla tube at the constriction which, in the normal
type, would mark the position of the stamens. These primary segments
are not reversed, but they bear secondary supernumerary segments
which show the reversals The latter are attached to the primary
segments at, or rather below, the region corresponding with the throat.
Our plants of this type are of a pale pink, so that the reversal of the
colouring is not so conspicuous as in the full coloured races of the
ordinary double, but it shows clearly in that the yellow "eye" at their
base is on the external side, while the internal side resembles the outside
of the primary segments and of the ordinary petal. The plants bear
no stamens at all and the female organs are generally represented by

1 Cf. Masters, Vegetable Teratology, 1869, p. 449.
- Cf. Masters, loc. cit. p. 315.

•* In both kinds of doubles the morphology of the reversed segments is obscure, and it
is not clear that these structures are of the same nature in the two cases.

R. P. Gregory 91

a group of foliar carpels', surrounding an axis on which are borne naked
ovules. Proliferation of the axis is frequent. Hitherto I have not been
able to raise any seed from these plants, but some cuttings, taken late
in the season and only coming into flower in May last, have developed
what appear to be normal ovaries, and it is hoped that experiments
will be possible in the future.

Inheritance of ordinary double.

The ordinary form of doubleness is a recessive character*. When
crossed with singles, it gives a single ^i, which on self-fertilization
gives singles and doubles in the proportion of 3 : 1. The actual
numbers obtained in 15 families are 762 singles, 284 doubles
{expectation: 78^5 : 261'5y.

The donble race ased in all the foregoing experiments had its origin in a white single
obtained from a nurseryman in 1903. The plant proved to be heterozygous, throwing
singles and doubles. Every degree of doubleness was exhibited among the various
individuals of this race, and the phenomenon was repeated in some of our F^'s. On
the other hand, certain plants, derived from the same strain, produced nothing but
full doubles, and in the F^i from their crosses with singles, the distinction between
the singles and the doubles was quite sharp, all the latter being fully donble.

Characters of the "Eye" of the Flower.

In the majority of horticultural strains the yellow or yellowish-green
" eye " of the flower occupies a small and well-defined area round the
mouth of the corolla tube. Besides this type of eye ihere exist two
other kinds; in the first, the eye occupies a much larger area, the
yellow colour extending well over the bases of the corolla lobes
("Primrose Queen," Plate XXX, fig. 12 and Plate XXXII, figs. 62
and 63, No. 37/9) ; the second type is represented by the white-flowered
race " Queen Alexandra," in which the eye is not distinguished from
the rest of the corolla, the whole flower being uniformly white (Plate
XXX, fig. 11 and Plate XXXII, fig. 62, No. 34/9).

Eye-characters are inherited quite independently of any of the
other characters which I have studied, but they affect certain other
characters with which they may occur in combination in the same

> Cf. Masters, loc. cit. pp. 262, 297.

- Bateeon, MendeV$ Principle* of Heredity, Camb. Univ. Press, 1910, p. 199.

' The discrepancy is almost entirely due to one F^ family which consisted of 66 singles
and 45 doubles. Five other ^2*8 from the same parents however gave 188 singles,
61 doubles.

7—8

92 Experiments with Primula sinensis

individual. The effect of the large yellow eye in giving rise, in the
absence of the factor for short-style, to the " homostyled " form has
been fully described on previous occasions'. Both the large yellow eye
and the white eye have effects when combined with certain colour
characters of the flower. Certain coloured forms possess a blotch of
deep colour, which in flowers with the ordinary eye occupies a well-
defined area at the base of the corolla lobes (Plate XXXI, figs. 50, 51).
If this character be combined with the large yellow eye, the deep colour
is, so to speak, pushed further outwards, and forms a rather ill-defined
band round the periphery of the area occupied by the pigment of the
eye^ But, so far as my observations go, when "Queen Alexandra" is
crossed with the same coloured race, the blotch of deep colour is
not developed in the F^ plants which have the white eye, though the
corresponding forms with the ordinary eye are blotched.

(1) Large yellow eye x small eye.

The accompanying table (p. 93) shows the results, inclusive of
those previously published', which have been obtained from crosses of
the " homostyled " plants with both short- and long-styled plants having
the ordinary eye. The crosses in which the F^ plant was selfed show
a considerable deficiency of large-eyed offspring, and in those cases in
which the small-eyed parent was short-styled, the deficiency is almost
confined to the short-styled offspring. The crosses of the form
{DR X R) have given results which, in the aggregate, do not differ
appreciably from expectation, though again, in those cases where we
are also concerned with short and long st)'le, the distribution of the
offspring among the four types is not very smooth*, and is particularly
irregular in one aberrant family (given separately in the table) where
the excess of short-styled offspring with the small eye is very marked.

It is only in the early years, however, that any great discrepancy
manifests itself Very few crosses have been made with short-styled
parents since 1906 ; but experiments with long-styled plants have been

^ Bateson and Gregory, loc. cit. pp. 582 — 584.

* For illustration of flowers of this kind see Bateson, MendeV» Principles of Heredity,
Camb. Univ. Press, 1910, Plate VI. figs. 19, 21.

^ Bateson and Gregory, he. cit. p. 584.

•* In this connexion it must be borne in mind that in the crosses between short and
long style there is throughout a deficiency of short-styled offspring when the f j is selfed,
and an excess when the Fj is crossed with the long-styled. This would, of course,
have a disturbing efifect in cases such as that under notice.

R. P. Gregory

g
^

tc £- 00

^ 2 -

2 111

I i

rH ;o 00 1-1

ag" s

■^ 1-1 «o

S. S ^

5i»

Isl

tCa -"K ^

Bo's

.9ae '*

A S S S3

6 iH I

r 8?

'So

r>.

a»

C4 00

n to

1 «
^ 1-1

v»

94 Experiments with Primula sinensis

continued, and the totals for the last three years are 972 small-eyed,
326 large-eyed {expectation : 973-5 : 324-'6). It is therefore impossible
to attach any great importance to the discrepancy in the early years,
though at the same time it remains unexplained.

(2) White eye x small yellow eye.

The white-eyed race ("Queen Alexandra") is a recent addition
to my collection and only a few F^ families have been raised from
crosses in which it takes part.

The heterozygote resulting from the cross with a small-eyed race
can be distinguished, on close examination, from the pure "Queen
Alexandra " by a faint appearance of yellow or yellowish-green, which
is most pronounced on the rays corresponding with the median line of
each petal (Plate XXXII, fig. 62, No. 35/9) i. Three F^ families have
been raised from the gelf-fertilized hybrid, and have given 182 white
eye and heterozygous, 67 small yellow eye {expectation : 186-75 : 62'25).

(3) White eye x large yellow eye.

The heterozygote resulting from this cross is not distinguishable to
the eye from that of the preceding case (Plate XXXII, fig. 62,
No. 36/9). The one F^ family raised from the hybrid by self-
fertilization has given 52 white eye and heterozygous, 12 large yellow
eye {expectation : 4^ : 16). An attempt to separate the pure from the
heterozygous white-eyed offspring gave 19 with no trace of colour in
the eye, 33 with faint yellow rays.

Colour.

The various forms of red stem, and the colours of the flowers, are
due to the presence of coloured sap. Both in the stem and in the
flower the simple colour may be modified by the action of numerous
factors which affect its distribution, intensity and tint. There is a close
relation between the colour of the flower and that of the stem, in that
fully coloured flowers are only produced by plants having fully coloured
stems. The deepest colour in the flowers of a green-stemmed plant
is that exhibited by the pale pink strain known as "Reading Pink"
(Plate XXX, fig. 13), while the white-edged type exemplified in "Sirdar "
is characteristic of plants in which the stem-colour is restricted to the

1 The contrast between the yellow rays and the white ground is somewhat intensified
in photographic reproduction.

R P. Gregory 96

collar and bases of the petioles (Plate XXX, tig. 4 ; Plate XXXI,
figs. 44, 45). The degree to which colour is developed in the stem
may therefore be taken as an index of the limits within which the
colour of the flowers will be confined. All the red-stemmed whites
which I have examined were found to be white in virtue of factors
which inhibit the development of colour in the flower, though their
range of action does not extend to the stem'.

A. Stem-Colours.

Various types of coloured stems are illustrated in Plate XXX. The
plants shown in figs. 1, 2, 4 and 5 all possess, in varying degrees, the
common purplish-red sap. Sap of this colour is present in the stems
of all the races which have the usual magenta or red flowers, and
though there are, no doubt, minor dififerences in the tint in diflferent
races, it is scarcely possible in practice to make any distinction between
forms which differ in so slight a degree. There are, however, two kinds
of flower-colour which are associated with distinctive stem-colours;
in the races which have blue flowers the stem has a corresponding
colour, as compared with that of the commoner purplish-red types;
while the clean red stem, shown in fig. 3, is, so far as my observations
go, limited, in the fully coloured form, to the strain known as " Orange
King" (fig. 8).

In coloured stems the red sap may be distributed over the whole of
the stems and petioles (Plate XXX, figs. 1 and 2), or it may be developed
only in certain regions, the other parts being green. Fig. 4 shows a
form in which the colour occurs only in the collar and bases of the
leaf and flower stalks ; in plants with coloured flowers this type of stem
is always associated with a peculiar distribution of the flower-colour
which is characteristic of the strain known as "Sirdar" (Plate XXXI,
figs. 44, 45). In fig. 5 there is represented a lower type of stem-colour,
in which the colour is most pronounced in the young petioles. It is
often only faint, and is sometimes scarcely discernible in the older leaf
stalks, so that the character is somewhat elusive. It is dominant to
the complete absence of sap-colour exhibited by " Snowdrift " (fig. 7),
but the discrimination between the various types in F^ is ditficult,
the more so since " Snowdrift " brings in a factor which reduces the
apparent colour to a minimum.

' Keeble and Pellew record the existence of a recessiTe white on red stem {Joum.
Genetic*, Vol. i. 1910, p. 1).

96 Experiments with Primula sinensis

In the foregoing types the colour extends into the root-stock and
roots, and in the faintly coloured forms its presence is much more easily
detected there than in the stem, where the colour of the sap is masked
by the green colour of the chlorophyll.

The plant represented in fig. 6 is the " Ivy-leaf." In this form the
colour can be recognized most readily in the young petioles, and it also
appears, though more faintly, in the pedicels. In older leaves the
colour may bf noticed at the base and sometimes along the edges of
the leaf-stalk. It does not appear to extend to the root-stock and roots.

Outline of the inheritance of stem-colour.

In its general outlines, the inheritance of stem-colour is simple.
Thus, the red-stem crossed with a green-stem gives an F^ in which
the red-stemmed offspring are either approximately 9 in 16, or 3 in 4,
according to the constitution of the green-stemmed parent. The full
colour crossed with the faint colour (fig. 5) gives, in F^, 3 of the former
to 1 of the latter, and similarly the faint colour behaves as a simple
dominant to the complete absence of colour.

Although the character of the stem in " Sirdar " is, in its lighter
shades, not very different in appearance from that of other faintly
coloured types, the inheritance of stem-colour can be most simply
explained if the " Sirdars," which appear in certain F^s, are regarded
as forming a part of the fully coloured population, lacking, however, in
the factor {F) which effects the even distribution of the colour in the
stems and flowers. We have then a factor {R) for colour, and epistatic
to R, and without effect in its absence, a distributing factor F^. In
order to provide for the existence of the forms with some faint
colour in the petioles we require to assume the existence of another
factor {Q) determining this character, which is independent of R and F
and is unaffected by them, except in so far as the faint colour is not
discernible when R and F are present.

In crosses between plants with the lower grade of stem-colour and
those without colour, the last factor (Q) only comes into play, and the
3 : 1 ratio is obtained in F^ (Table, p. 98, II.). Since " Sirdars " have
only occurred in my experiments in cases in which "Snowdrift" was

1 The use of a so-called distributing factor is intended merely as providing a simple
means of formulating the observed results. The relation which subsists between the
" Sirdar" types and the self -colours is probably different from that which obtains between
flakes and self-colours (p. 122).

R P. Gregory 97

used, we are justified in assuming that all the other races which have
been used possess the factor F ; consequently, crosses between the full-
coloured stem and the faint colour merely exhibit the segregation of
the factor R, the effect of Q, which is present in all the offspring,
being masked when R is also present (Table, p. 98, III.). The same
applies to the crosses between the full colour and the green stem,
but in this case one-third of the offspring have clean green stems
(Table, p. 99. IV.).

In the F^& from crosses between " Snowdrift " and races with fully
coloured stems " Sirdars " occur ; and if the factors inhibiting flower-
colour be absent, the F^ is found to contain approximately, in every 16
plants, 9 with fully coloured stems and flowers, and 3 " Sirdars " ; while
of the remaining 4, 3 may have faint colour in the petioles, or they may
all be devoid of colour in the stem, according to the presence or absence of
the factor for faint colour in the coloured parent. The total numbers
obtained in these crosses (Table, p. 99, V. C) show some divergence
from the expectation set forth above, in giving an excess of " Sirdars."
The divergence is however almost entirely due to the results obtained
from the first two families raised, which gave 142 full, 75 "Sirdar"
and 67 faint and green. In the later experiments a close approximation
to the theoretical proportions has been maintained, the numbers
obtained being 329 full, 110 " Sirdar," 134 faint and green. In crosses
between " Snowdrift" and red-stemmed dominant whites, the "Sirdar"
character cannot be determined with any accuracy in those offspring
which have white flowers. In these F^s the observed numbers of full-
coloured stems and light stems (including " Sirdars ") approximates very
closely to the expected ratio of 9 : 7 (Table, p. 99, V. B).

There remain however cases in which the 9 : 7 ratio is clearly
indicated in F^s from which " Sirdars " are absent (Table, j). 98, I.).
Only two such cases have been met with, but the result strongly
suggests that, in Primula, as elsewhere, at least two complementary'
factors are necessary for the production of colour. In one of the cases,
the character of faint stem- colour was not recorded separately, and we
only know that the family consisted of 51 fully coloured and 33 light
or green stems. In the other case, the coloured parent was a dominant
white, and the offspring consisted of 49 fully coloured, 13 with colour
in the petioles, and 25 devoid of colour, or 49 fully coloured, 38 light
stems (.9 : 7 = J^8•9^ : 38-06). In so far as reliance can be placed upon
the distinction between plants with faint colour in the petioles and
those devoid of colour, this result further suggests that one comple-

98 Experiments with Primula sinensis

mentary factor (C) is common both to the factor for full colour {R) and
to that for faint colour {Q), so that the combination CR gives full
colour^, and the combination GQ gives faint colour in the collar. The
constitution of the hybrid would then be Gc Rr Qq, and the F^ would
consist of the three types in the proportion of 36 full : 9 faint : 19 green,
or, in a total of 87 plants, 48-94 : 12-23 : 25-83.

The only matings of the " Ivy-leaf" from which F^s have as yet been
obtained, are its crosses with " Snowdrift " and with full-coloured forms.
The ^2 from the cross with "Snowdrift" is chiefly interesting in con-
nexion with the partial suppression of stem -colours, and is considered
more fully under that head (p. 101). Unlike the majority of the
experiments on stem-colour, in which the observed results agree witli
the expectation very fairly closely, there is a great dearth of light-
stemmed offspring in the F^'s from the crosses between "Ivy-leaf" and
plants with fully coloured stems. The deficiency is most marked in
the class in which the light colour is combined with the "Ivy-leaf"
habit, but is also apparent, though in less degree, in the light-stemmed
plants of the normal kind. There do not however appear to be
sufficient grounds for supposing that any novel phenomenon occurs
in these cases.

Table shovnng the results of experiments in regard to stem-colour.

I. Red stem (C EF Q) x Green stem {c rF q)

Fi selfed 1 family Red 51 Light 33

Colour in petioles 13 Green 25

Red 49

38
Totals ... "lOO 71

Expectation (9 : 7) 96-2 74-8

11. Faint colour in petioles (C Q) x Green stem (C q)

Fi selfed 9 families Faint colour 366 No colour seen 130

Expectation (3 : 1) 372-0 124-0

III. Red stem (G RF Q) x Faint colour (C rF Q)

Fi selfed 11 families Red stem 384 Faint colour 120

Expectation (3:1) 378-0 126-0

» Strictly speaking this combination gives the parti-coloured type " Sirdar," but as no
"Sirdars" appear in this F-y we are not here concerned with the distribution of the
full colour.

R P. Gregory 99

IV. Red ttrm (C RF) x Green item, no colour teen (C rF)

\ selfed 11 families

Red stem 39a

No oolonr seen 128

Expectation (3.1)

S92-25

13075

\ X green stem 6 families

99.

104

Expectation (1 : 1)

101-5

V. CroBiea giving " Sirdars." Red stem (C RF Q) x " Snowdrift " (C rf q)

In 1905 many plants were discarded as seedlings ; as there is no record of the flower-characters of theae
plants, the " Sirdars " cannot be distinguished from the other light-stemmed types. In the Fj's from (" Snow-
drift "x dominant white) the flower-characters of a proportion of the family are masked by the presence
of the dominant white character, and in such cases the " Sirdars " cannot be certainly distinguished from
other light stems.

A. 1905 crosses.

F, selfed 4famiUe8 Red stem 76 Light stem (including >^

' ' Sirdar ) )

Expectation (9 : 7) 69-2 53-8

B. Dominant white x " Snowdrift."

Fi selfed 8 families Red stem 440 „ 337

Expectation (9 : 7) 4371 3399

C. Ck)lonred, red stem x " Snowdrift."

2?, selfed 11 families Red stem 471 "Sirdar" 185 Fa^^* colour) ^^^

and green )

Expectation (9:3:4) 482-1 160-69 214-25

VI. Red stem x " Ivy-leaf."

Bed stem

Light stem

F] selfed 7 families
Expectation

Palmate
626

587-25

Ivy
197
195-75

Palmate Ivy
177 44

195-75 65-

823
783-0

221
2610

The red stem of " Orange King " (Plate XXX, fig. 8).

"Orange King" originated in horticulture a few years ago, and was
obtained by Messrs Sutton as a sport from " Crimson King." The
"Orange King" character of flower and stem is recessive to that of
" Crimson King," and in the F^ the two forms reappear in numbers
approximating to the 3 : 1 ratio. This result would indicate that
a single factor suffices to restore those characters which diflferentiate
" Orange King " from " Crimson King." The only other matings of
" Orange King" of which I have experience are those with " Snowdrift."
The hybrid resulting from this cross is indistinguishable to the eye
from the hybrid between "Crimson King" and "Snowdrift." The

100 Experiments with Primula sinensis

" Orange King " characters of stem and flower are however so intimately
associated that the fuller consideration of this case may be deferred
until the section dealing with flower-colour (p. 114),

Partial Suppression of Colour.

The light shades of the colour in the stem are dominant to the
intense shades. This fact is well illustrated in the F^s from (" Ivy-
leaf" X deep red stem), where the red-stemmed plants fall into two
sharply separated categories. The numbers obtained are :

Light Intense

3 families ... 157 49

Expectation ... 154'5 51'5

Similar sharply divided categories are found in families raised from
the cross of a deep red stem with the jPj of (" Snowdrift " x deep red
stem). The numbers obtained in these crosses are :

Light Intense

9 families ... 198 202

+ 3 doubtful (occurred in one family)
Expectation ... equality.

It is clear from these cases that the light class, taken as a whole,
may be explained as being due to the presence of a single factor,
epistatic to the factors for colour, which diminishes the intensity of
the pigmentation (pallifying factor). In the F^'s produced by the self-
fertilization of the Fi from the cross ("Snowdrift" x deep red stem) there
are forms intermediate between the light and the very dark red stems,
and the separation between the classes is by no means sharp. No doubt
many of these intermediate forms are the result merely of heterozygosis
in the factors for colour and for its partial suppression. In different
pure races, however, and in the hybrids produced by their matings,
colour is developed to very different degrees, and in order to account
for the detailed phenomena it would probably be necessary further to
elaborate the simple scheme put forward here, which is intended only
to apply to the general outlines of the phenomena of the partial
suppression of stem-colour.

The partial suppression of flower-colour follows, in general, very
similar lines to that of stem-colour, but is independent of the latter,
at least to the extent that light flowers may occur on deeply coloured
stems.

In the lower grades of stem-colour the same relation subsists between the light and
intense states as in the fully coloured types, but the separation of the categories is

R. P. Gregory 101

of course a matter of much greater practical difficulty. The point has been studied with
some care in the cross ("Snowdrift" x "Ivy-leaf" ^). The Fi has a faint trace of colour
in the young petioles; the F^ consists of (1) plants with full colour in the young petioles,
which grade through rather lighter forms to (2) those in which faint colour in the
petioles can be recognized with certainty; and these again grade, through doubtful
forms, to (3) those in which no colour can be detected. Precise numerical results cannot
be given, but so far as can be judged the constitution of the families can be fully explained
without the assumption of any other factors than those for colour and for its partial
suppression.

B. Flower-Colours.

The various colours exhibited by Primula sinensis may be classified
as (1) full colours, which jjaay exist either in the self or in flaked
patterns (Plate XXXI, figs. 56 — 59), and are always associated with
fully coloured stems; (2) "Sirdars" (Plate XXXI, figs. 44, 45), in
which the characteristic distribution of the full colour is associated with
a definite type of stem-colouring; and (3) pale colours (Plate XXXI,
fig. 46) which occur only on green or faintly coloured stems.

White flowers may occur in association with stems of any kind, and
may be dominant or recessive to colours. The dominant whites owe
their character to the possession of factors which inhibit the development
of colour in the flower (see under "Inhibition," p. 105).

The full colours and " Sirdars " may be sub-divided into blues,
magentas and reds; in the pale class, however, no distinction of this
kind can be drawn, for the pale forms which correspond with the
magenta full colours (and give magenta offspring when crossed with
a red) are quite indistinguishable to the eye from those which
correspond with the red class (and give only red offspring when
crossed with reds).

Colours belonging to all these classes appear in the offspring of
certain hybrids; the sharpness of the separation between the various
classes of full colours varies, however, in different cases, and though in
the majority the classes are fairly readily distinguished, in others
intermediate forms occur. Whether these intermediate forms are
always heterozygous cannot yet be said ; in the few experiments in
which they have been tested they have proved to be so^

' The "Ivy-leaf" used in the experiments on stem-colour was heterozygous for the
pallifying factor. Hence the appearance of "Ivy-leaf" here as the parent lacking the
pallifyiag factor, and previously as the parent bringing in that factor.

' A plant with red stigmas, which probably belonged to the red class but had flowers
of a colour somewhat intermediate between the magenta and red classes, has since proved
to be homozygous for its type of colour.

102 Experime7its with Primula sinensis

I have not yet undertaken any systematic experiments with the
blue-flowered strains of Primula sinensis. Bhies occurred among the
offspring of a certain magenta plant obtained in 1903, in such propor-
tions as to corroborate the more extended results obtained by Messrs
Sutton, which show that the blue colour is an ultimate recessive \

For the purpose of these experiments it has been found convenient
to work mainly with well-known horticultural strains, which provide a
series of fixed standards of colour. The colours of the races of which
principal use has been made are illustrated in Plate XXXI. For con-
venience of reference descriptions of the various types are given below.

Description of strains used in experiments on colour.

Recessive White.

" Snowdrift." (Plate XXX, figs. 7, 10.) Fern-leaf, green stem, white, green stigma.
Pale colours.

" Beading Pink." (Plate XXX, fig. 13. ) Palmate, green stem, pale-pink, green stigma.
Full colours.

Salmon Pink. Palmate, purplish-red stem (light), salmon-pink, green stigma, short
style.

Rosy Magenta. (Plate XXX, figs. 19, 20.) Palmate, purplish-red stem (light),
magenta (rosier than Fj type, light), green stigma.

"Crimson King." (Plate XXX, fig. 9.) Palmate, purplish-red stem (deep), deep
crimson, red stigma.

"Orange King." (Plate XXX, figs. 3, 8.) Palmate, red (not purplish-red) stem,
pink flowers, red stigma.
Dominant Whites.

Double White. Palmate, green stem with colour in leaf bases (Plate XXX, fig. 6),
double white flowers, green stigma.

" Primrose Queen." (Plate XXX, fig. 12.) Palmate, purplish-red stem (light), white
flowers, green stigma, large yellow eye.

"Queen Alexandra." (Plate XXX, fig. 11.) Palmate, purplish-red stem, white,
green stigma, white eye.
Colour uncertain (see p. 122).

"Ivy-Leaf." (Plate XXX, fig. 5, Plate XXXII, fig. 60.) Palmate, non-crenate, stellata,
green stems with colour in leaf bases'^, flowers? very pale colour flaked, green
stigma. The "Ivy-leaf" is a very monstrous type, the non-crenate character of
the leaves being always accompanied by partial abortion of the floral organs.
Stamens are often absent and the corolla may be reduced to a tube surrounding
the style, without petal-lobes. Petal-lobes, when developed, may be only small
strap-shaped structures. Owing to the poor development of the corolla the colour
of the plant used in the experiments cannot be determined with certainty. Such
plants as survive usually become fertile in the second year, producing however

1 Bateson's MendeVs Principles of Heredity, Camb. Univ. Press, 1909, p. 135.

2 The colour is insufiiciently shown against the dark background in the plate.

R. P. Gregory 103

only small quantities of pollen. A very common character of the " Ivy-leaf" is
that the axis of the inflorescence forms a more or less conical elongation above
the whorl of pedicels, at the apex of which carpellary structures may be
developed, or ovules may be borne on an exposed disc, which is sometimes
surrounded by small lobed expansions (probably carpels) each terminating in a
knob resembling a stigma ^ In extracted Fg-forms with green stigmas these
expansions are green, in those with red stigmas they are coloured.

Outline of the inheritance of flower-colour.

When a plant with fully coloured stems and flowers is crossed with
the albino " Snowdrift," the F^ consists of

Full-colours, " Sirdars," Pale colours and Whites

in the ratio of 9 full : 3 "Sirdar" : 4 pale colour and white. Although
the number of whites recorded in these F^Q is somewhat less than
1 in 16, there can be no doubt, I think, that this represents the
proportion in which they really occur*.

In a family of this kind, the plants having fully coloured stems
always have fully coloured flowers ; that is to say, the full colour, when
present, is distributed throughout the whole plant. Consequently, it
is not necessary in this case to draw a distinction between stem-colour
and flower-colour, since the colour of both behaves as a single unit'.
The inheritance of the full colour, then, follows the scheme outlined in
the case of stem-colour (p. 96), in which the relation of the " Sirdars "
to the full colours is also explained.

The place of the pale colours in the scheme must be left undecided
until further data are available. It may prove that they constitute an
independent series of colours, comparable with the faint stem-colours
in their relation to the full colours ; or they may perhaps result from
the resolution of the combination of factors to which the full colour

1 The structures described by Mr L. Crawshay in a malformed Primula (Journ. Roy.
Hort. Soc. XXXVI. 1910, p. xxix) are apparently of the same nature.

' The discrepancy is almost certainly due to the difficulties attending the separation
of the pale colours from the whites. The fact that we have sometimes detected a trace of
colour in pure " Snowdrift," when the plants have been kept cool, points in the same
direction.

' Keeble and Pellew's experiments (Journ. Genetics, Vol. i. 1910, p. 1) indicate that in
certain pigmented forms one, at least, of the factors which determine the production of
colour may be absent from the flowers, which are then white, though it is present in the
stem, which is therefore coloured. This evidence that, in certain cases, the factors for
colour are not distributed throughout the whole plant, is indirectly supported by the
results of my experiments with the red-stemmed dominant white "Primrose Queen"
(pp. 116, 123).

104 Exjjeriments with Primula sinensis

is due. If the former suggestion should prove to be correct, the fact
that all our fully coloured races, when crossed with " Snowdrift," have
given pale colours in t\ ; and the further fact that two heterozygous
"Sirdars" have thrown only "Sirdars" and whites, would be merely
fortuitous results depending on the particular races which have been
used. It may be noted that, if the pale colours are an independent
series, certain matings between F^ "Sirdars" and pale pinks should
give full colours, while others should not do so; the alternative case
would seem to imply that all these matings should give full colours.

The primary colour of the fully coloured flower is red\ The
numerous shades of red are due to the presence or absence of factors
which reduce the intensity of the pigmentation, and other factors which
produce slight changes of tint. In the simplest cases the magenta
class may be regarded as due to the action of a factor epistatic to the
factors which give rise to the red colour; in other cases, however,
the proportions of the magenta and the more rosy class indicate the
9 : 7 ratio (see under " Rosy Magenta," p. 110) ; and in yet another case
an intermediate, mated with a clean red, gave typical magentas among
its offspring. There exist corresponding shades of magenta for many,
if not all, the numerous shades of red.

The flaked or splashed forms of coloured flowers show a considerable
range of variation in the degree to which the flaking is developed, and
in the size and form of the coloured areas. The distinction between
the red and magenta colours in flakes is often attended with some
difficulty. In self-coloured red flowers it will often be noticed that
a bluer tint is developed at the edges of the petals, and in forms in
which the colour is weakly developed just round the eye a similar
bluish tint will be noticed in this region. In the same way, there
seems to be a tendency for the red colour to pass into a bluer tint at
the edges of the coloured stripes and splashes, and in flowers showing
fine as well as coarse splashes, it is often to be noticed that the coarse
splashes are red, while the minute dots of colour, viewed with the naked
eye, would certainly be put down as magenta^ My experience of flaked
flowers is limited to the F^a of crosses in which the "Ivy-leaf" took

' The relations of blue to the other colonrs have not been worked out. The fact that
blues appeared in small numbers in a cross in which the rest of the coloured offspring
were red suggests that blue is either hypostatic to red, or, if it forms an independent
series, is masked by red.

2 A somewhat similar difficulty occurs in the "Sirdar" type, owing to the optical
effect of the intermingled coloured and colourless dots. In this case, however, the distinction
between magentas and reds can be made readily with the help of a microscope.

R P. Gregory 105

part; the results are such as to indicate that the flaked condition
behaves aa a recessive to the self-colour (see, however, p. 122).

The pale-coloured flowers on green stems are scarcely affected in
appearance by the presence or absence of the numerous factors which
produce such marked changes in the fully coloured types of flower.
It is often by no means easy to recognize the pale colour when it occurs
in the flaked condition ; this is no doubt an optical difficulty, for the
lower forms of this colour in any case need careful examination in
order to distinguish them from white. Among the pale-pinks there
occur forms in which the colour is more pronounced peripherally, others
in which it is central, others again in which it forms peculiar bands.
But the difficulty of observation is such that no attempt has yet been
made to study the inheritance of these variations.

Partial Suppression of Colour.

As in the case of stem-colours, the intense colours of the flowers are
produced only in the absence of a factor which diminishes the intensity
of the pigmentation, and so gives rise to the dominant light shades.
The partial suppression of flower-colour may be brought about by
either of two factors, of which one affects the flower only, the other
the whole plant. Hence light flowers may occur in association with
dark stems, but deeply coloured flowers are limited to plants with
deeply coloured stems.

In many F^s there occur classes intermediate between the lightest
and the very deep types, but, though the existence of such classes may
be clear enough, it is difficult, if not impossible, to draw any sharp line
between them, and, as in the case of stem-colours, it must remain
undecided whether one pallifying factor, in its various pure and
heterozygous combinations, is sufficient to account for all the shades,
or whether a series of such factors is involved.

The factors which effect the partial suppression of colour seem to
diffier in degree rather than in kind from the factors which, in pure
races, completely inhibit the development of colour in the flower.

Inhibition.

In the red-stemmed " Dominant WhitesS" the whiteness of the
flower is due to the presence of a substance which inhibits the

1 Gregory, Rep. Brit. Auoc., Leicester, 1907, p. 692.
Joam. of Gen. i 8

106 Experiments with Primula sinensis

development of colour in the flower^ It has recently become clear
that this inhibition is due to the action of two separate components,
each of which has its own localized effect. The one component is
present in the majority of the races which have coloured flowers, in the
form of a factor which prevents the development of coloured sap in the
ovary, style and stigma, and gives the green stigma. The second factor,
on the other hand, affects only the peripheral parts of the corolla, and
in the absence of its fellow, gives rise, in fully coloured forms, to the
characteristic "Duchess" type of flower (Plate XXXI, figs. 27, 28), in
which coloured sap occurs only in the gynoecium and in the flushed eye
of the corolla^ In the pale colours the stigma is only faintly coloured,
and the presence of coloured sap can be most easily detected in the
placenta and ovules. The recessive green stigma (which corresponds
with the recessive white flower, and is green through the absence of
colour and not from its inhibition) has been recognized experimentally
in F^ plants fi-om the crosses of "Snowdrift" with "Crimson King."
The factors for inhibition may of course be present in plants which are
devoid of the factors for colour ; thus the green stigma of " Snowdrift "
is of the dominant kind, and other green-stemmed whites have been
met with, which possess both the factors for inhibition.

Plants which contain the factors for colour and are heterozygous for
the inhibiting factors have tinged white flowers with green stigmas, the
depth of the tinge varying with the intensity of the underlying colour
(Plate XXXI, figs. 21, 24—26, 32). A heterozygous form of '•' Duchess "
is represented in "Sir Red vers Buller" (Plate XXXI, fig. 29), and various
other forms, depending on the presence or absence of the magenta and
other factors epistatic to colour, exist (figs. 30, 31). In all of them the
peripheral part of the corolla is tinged to a greater or less degree, and
the full colour is only developed immediately around the eye.

One other character of flower-colour should be mentioned here. In
certain varieties there occur spots of deep colour on the petals just
external to the eye (Plate XXXI, figs. 50, 51). The inheritance of this
character is, in itself, simple ; but the full development of the spots is
limited by the operation of other factors^. Thus, the deep spots are

1 In certain races belonging to this class an occasional splash or stripe of colour may
often be observed, sometimes in only one or two, sometimes in many of the flowers.

2 The flush rouud the eye is often only faint, especially in flowers of the stellata
variety. The flush is an independent character limited to plants with red stigmas (see
p. 120).

'^ Bateson, Mendel's Principles of Heredity, Camb. Univ. Press, 1909, p. 138.

R P. Gregory 107

not fully developed unless the stigma is coloured ; nor, even if the stignia
be coloured, are they developed in plants which have the white eye of
the "Queen Alexandra" type (Plate XXX, fig. 11)'. Again, the spots
are deeply coloured only in deeply coloured flowers, their appearance in
flowers of a light shade somewhat resembling that which they assume
in plants with green stigmas. The limitation imposed in these cases
results from the dominance of an inhibiting character. There are also
limitations due rather to the lack of a coloured base ; the spot is not
visible in pale-coloured flowei-s, nor again in the flaked patterns of
full colour, unless it should happen that the colour is distributed in
any of the petals in a wide stripe covering the area occupied by the
spot. Such petals exhibit the spot, which may not be visible in other
petals of the same flower.

Plants in which the development of the spot of deep colour is inhibited by the factor
for green stigma have flowers of a definite type, characterized by the presence of a
well-defined brownish spot. The character is a different one from the diffuse brownish
band which appears in some plants as the flowers fade (Plate XXXI, figs. 54, 55), and is
very clearly marked in the young flowers (Plate XXXI, fig. 50), becoming less conspicuous
as they grow older (fig. 51). This "ghost" of the spot is well seen in the Fi from
('•Crimson King " x •' Rosy Magenta"), and in the F^ all the plants with red stignus
have the spot of deep colour. The inheritance of the character is further illustrated
in the subjoined experiments in which a series of F^ pale pinks were crossed with
" Orange King."

Green stignma Red stigma

No spot

Beference
Number

Ghost of
spot

No spot

Spot

36/10

—

37/10

—

38/10

—

39/10

—

40/10

No plants

41/10

—

42/10

—

43/10
44/10

No plants

45/10

—

46/10

—

47/10

—

* These three had light stems, and a brownish marking in the region of the spot
somewhat resembling the marking which represents the spot in plants with green
stigmas.

• In plants with the large yellow eye the spot is pushed outwards, so that it occupies
the same position relative to the eye pigment as it does in the usual type (see Bateson,
loc. eit. Plate VI, figs. 19, 21).

a-2

108 Experiments with Primula sinensis

Experimental results. (1) Pale colours.

Pale-pink (Plate XXX, fig. 13 ; Plate XXXI, fig. 46). Pale-pinks
have occurred in the F^s of all ray crosses between full colours and
" Snowdrift," as well as among the progeny of certain heterozygous
full colours obtained from various sources. It is also the characteristic
coloured form thrown by heterozygous dominant whites having green
or only slightly coloured stems. If the pale-pink be crossed with
" Snowdrift " the resulting F^ shows some dilution of the colour.

Heterozygous pale-pinks can throw nothing but pale-pinks and
whites, and this they do in the proportion of 3 pinks : 1 white, the
numbers obtained being 51 pink, 16 white. One of these plants
crossed with "Snowdrift" gave 23 pink, 17 white.

My experiments throw no definite light on the question of the
dependence of the colour on two complementary factors, a chromogen
and a ferment, but in this connexion the cross between "Ivy-leaf" and
" Snowdrift " should be mentioned. Both parents appear white, while
the hybrid has definite though faint colour in the flowers. In F.2
plants with definitely coloured flowers form approximately 9 in every
16 plants, the observed number being 144 coloured in a total of 273
plants. Subsequent experiments with the " Ivy-leaf," however, suggest
the possibility that, instead of its being a white, as I had supposed, it
may have the very pale pink colour in the flaked condition (see p. 122).

The pale-pinks may or may not carry the magenta factor. Of
10 pale-pinks tested by crossing with reds, 6 were pure for the magenta
factor and gave 65 offspring, all magenta ; 2 were heterozygous and
gave 30 offspring, 14 magenta, 16 red ; and 2 were without the magenta
factor and gave 16 offspring, all red. One other, mated with a magenta
throwing magentas and reds, gave 5 magenta, 5 red, and was therefore
without the magenta factor.

The same set of experiments served to reveal other characters
carried by the pale-pink. Nine F^ pale-pinks from the cross (" Crimson
King" X "Snowdrift") gave offspring when crossed with "Orange
King." In the resulting families there occurred intense and light
colours, in one case rosy-magentas as well as the usual kind, in another
case deep crimson-magentas together with reds very like "Crimson
King," while in some cases the spot of deep colour was present in all
the offspring having coloured stigmas, in others in only a proportion of
them^ One pale-pink without colour in the stem was found to have

1 See Table, p. 107.

R. P. Gregory 109

the recessive kind of green stigma, all the offspring resulting from its
mating with "Orange King" having coloured stigmas^ As was to be
expected from the origin of the pale- pinks, none of the offspring showed
the colour characters of " Orange King," the stem-colour being always
purplish-red, and the colours of the flowers those of types found in
"Crimson King " F,'s (Plate XXXI, figs. 33, 36, 39, 41, 43). The pale-
pink strain " Reading Pink," crossed with " Orange King," gives a red
(Plate XXX, figs. 15, 16) rather towards the magenta side of the class
and having purplish-red stems.

(2) Full colours.

Salmon-pink. The race of this colour which has been used for
experiment was derived from a heterozygous crimson, or crimson-
magenta, which threw forms like itself, together with salmon-pinks and
blues. The crosses in which this race has been tested give very simple
results, since the race was pure for the light colour, and was without
factors producing, the minor variations of tint. Heterozygous salmon-
pinks may throw pale-pinks only, or whites may appear in addition ; in
either case the proportion of full colours in the offspring follows the
stem character. Crosses between such heterozygous salmon-pinks and
either " Snowdrift " or the pale-pink carrying magenta show the simple
operation of the magenta factor; crosses of this kind have given
44 magenta, 52 pale colours.

Salmon-pink x " Snowdrift." The F^ from this cross is a magenta
with light red stems. In the F^ there were obtained, in 3 families :

Foil coloan " Sirdsra " No ooloar in stems

Ma^nta Salmon Magenta Salmon Pale-pink White

57 J6 16 6 19 10

52-3 17-4 17-4 5'8 23'3 7'8

The expectation, given in italics, is based on the scheme already set
forth, namely, that the full colours represent the "Sirdars" -h a factor
which effects the even distribution of the colour.

The salmon-pink is one of the few short-styled races with which as
yet detailed experiments upon the inheritance of colour have been
made', and a most interesting relation between the structural character

' See Experiment 41/10 in the Table, p. 107. The pale-pinks used in Experiments
37/10 and 38/10 bad faintly coloured stigmas.

' The obvions advantages of working with pure horticultural strains entail the
disadvantage of working exclasively with long-styled plants, since the short-styled form
is eschewed by florists.

110

Experiments with Primula sinensis

of short-style and the magenta colour has been revealed. In the F^s,
bred from plants heterozygous for both characters, the salmon-pinks
are invariably short-styled. The results clearly indicate complete
repulsion in gametogenesis between the two dominant factors, short-
style and magenta. The case is dealt with fully on p. 125.

Rosy-Magenta. For the strain of this colour with which experi-
ments have been made I am indebted to Messrs Sutton. Very similar
types appear, as part of the magenta class, in the F^^a of certain crosses
between reds and either "Snowdrift" or pale-pinks carrying magenta.
The colour of the root-stock in this race bears the same relation to the
colour which appears in the ordinary magentas as does the flower-
colour in the two cases. The cross with " Snowdrift " gives an F^ of the
ordinary magenta type. In the F^ the rosy-magentas take the place
of the reds, but the distinction between the two classes is of course less
obvious than that between magentas and reds. Like the salmon-pink,
the rosy-magenta does not carry the factor for faint colour in the stem,
and in the light class the stems and roots are devoid of coloured sap, so
far as can be seen. The F^ obtained in one experiment of this kind
suggests a ratio of 9 magentas : 7 rosy-magentas, the numbers obtained
being :

Full colours

' Sirdars '

No colour in stem

Reference
Number

9/9

Magenta
37

Rosy-
magenta

Magenta
14

Rosy-
magenta

Pale-
pink

White
4

In the next two, however, the usual 3 : 1 ratio obtains :

Full colours "Sirdars" No colour in stems

Reference
Number

Magenta

Rosy-
magenta

Magenta

Rosy-
magenta

Pale-
pink

White

23/9

, 4

17/10

Totals

One can scarcely believe that the result shown in Experiment
No, 9/9 is only a fortuitous departure from the 3 : 1 ratio, nor does it
seem likely that it is due to experimental error in the separation of the
classes, for both No. 9/9 and No. 23/9 were recorded within a day
or two of one another, and in each case the separation of the classes
was confirmed by another observer. The same rosy-magenta parent
was used in Experiments 9/9 and 23/9, and one of its offspring in
Experiment 17/10. The different results are not necessarily con-
tradictory, for if the difference between magenta and rosy-magenta

R. P. Gregory 111

does, in reality, depend upon the combination of two factors (of which
"Snowdrift" must be assumed to have both) the rosy-magenta used
in the 1909 experiments may have been heterozygous for one of them,
without giving us any clue other than that which is suggested by these
experiments. The mating between a sister plant of the rosy-magenta
used in experiment No. 17/10 and a dominant white gave magentas
and rosy-magentas in F^. The separation between the two classes was
somewhat doubtful, but they apparently consisted of 20 and 19 plants
respectively. So far as this observation carries weight, it tends to
support the view that the difference between the two classes depends
on the combination of two factors.

" Crimson King." In all its crosses " Crimson King " gives a great
variety of coloured forms in F.^, and it is clear, both from the number
of these forms, and from the comparative rarity with which the
" Crimson King " t}^e itself reappears, that its visible characters
result from the interaction of several factors which are partially or
wholly independent of one another in segregation.

A series of F^ forms from the cross with the dominant white
" Queen Alexandra " is shown in Plate XXXI, figs. 22 — 43. The types
possessing some form of inhibition will be dealt with under that head
(p. 115). Among the coloured forms (figs. 33 — 43) various types of light
and dark magentas and reds occur, with or without the coloured stigma.
This last character is recessive to the factor inhibiting the development
of colour in stigma, and the observed numbers of green (colourless)
stigmas and red stigmas approximate very closely to the ratio 3 : 1.
But in the great majority of my experiments the two kinds of stigma
are not evenly distributed among the magentas and reds, and there is
clear indication of the existence of partial gametic coupling between
the two factors magenta and green stigma (p. 127). "Crimson King"
has the factor determining the spot of dark colour on the petals and
accordingly this character appears in deeply-coloured flowers which
have the coloured stigma and the ordinary or large yellow eye.

" Crirtison King " x " Snowdrift." The F^ from this cross is an
ordinary (light) magenta. The F^ contains fully coloured forms corre-
sponding with those just described^ and in addition to these there
occur magenta and red " Sirdars " (figs. 44, 45) in light and deep
forms, pale-pinks (fig. 46) and whites, the last two classes having green

1 The white eye is a character derived from "Queen Alexandra" and does not appear
in the experiments with " Snowdrift."

112

Experiments with Primula sinensis

or only faintly coloured stems. The magenta and red classes form
parallel series of light and intense shades ; the two classes as a whole
are readily distinguished, though there usually occurs a small number
of individuals whose proper position may be a matter of some doubt.
In this connexion it may be remarked that the presence of the red
stigma seems to have the effect of giving the flower in general a redder
appearance than that of the corresponding type with green stigma.

Two ^2 families raised from this cross in 1907 show some departure
from the normal in the ratio of full colours and " Sirdars " ; the
numbers obtained were :

Full colours

"Sirdars"

Pale class

Magenta

Red

Magenta

Bed

Pale-pink*

White*

Stigma Stigma
green red

33 15

12 5

14 9

6 2

49 16

8 4

24 8

10 2

Totals 82 31

20 9

38 17

16 4

113 29 55 20

* The distinction between these two classes is not sharp.

The case does not perhaps merit any great consideration in view of
the return to the normal ratio when the experiment was repeated in
the succeeding years, and the lack of any other indications of a depar-
ture from the normal distribution of self-colours and " Sirdars."

Three families raised subsequentl}' gave :

Full colours

Magenta

Red

Stigma Stigma Stigma Stigma

green red green red

16 7 7 5

14 3 6 1

13 4 3 5

" Sirdan

Pale class

Magenta

Red

Pale-pink* White

Stigma Stigma
green red

9 2

1 2

12 4

5 3

2 0

11 1

1 0

12 1

Totals 43

35t

* Distinction not sharply drawn.

t Of 6 of these which had some colour in the stem, 4 had coloured stigmas, 2 green.

The five families taken together give 245 magentas, 82 reds ; 234
green stigma, 93 red stigma; the calculated numbers in each case being
245*25 of the larger class, 81 '75 of the smaller. In the first two

R. P. Gregory 113

experiments the distribution of the two kinds of stigma among the two
classes of colours follows the normal 9:3:3:1 ratio, being :

MagenU MagenU Red B«d

green stigma red stigma green stigma red stigma

Fall colours
Sirdars

120

122-0

40-7

136

Totals
Expectation

In the later experiments there is considerable departure from this
distribution, the first class being small and the last large. But it is to
be noticed that in these two cases there is considerable departure from
the normal ratio of 3 : 1 in each of two pairs of characters under con-
sideration, the numbers observed being 77 magenta, 33 red ; and 78
green stigma, 32 red stigma. There seem to be no grounds for regard-
ing this discrepancy as other than a chance departure from the normal,
but it of course has a very material effect on the numbers observed in
the four groups when the two pairs of characters are considered in
conjunction with one another. If the theoretical ratio of 9 : 3 : 3 : 1
be weighted so as to allow for the two discrepancies a fairly close
approximation to the observed numbers is obtained :

Magenta
green stigma

Magenta
red stigma

Bed
green stigma

Bed
red stigma

Observed numbers

Expectation from weighted ratio

54-6

22-4

23-4

9-6

There is therefore no clear indication that partial gametic coupling
between the factors for magenta and green stigma occurred during the
gametogenesis in the ^j plants used in these experiments ; the point is
of some interest because partial coupling of these two factors is clearly
indicated in many of the experiments in which " Crimson King "
was used.

" Crimson King " x Rosy-Magenta. The F^ from this cross is a
magenta of a rather deeper kind than that of the Fi from (" Crimson
King " X " Snowdrift "). In the F^ there occurs, in addition to the
ordinary magentas and reds, a curious parti- coloured type in which
irregular masses of full colour are distributed over a lighter ground.
These " Strawberries " (Plate XXXI, fig. 49) apparently belong to the
red class and only occur in small numbers, probably as one in 64 of
the total offspring.

The magentas and reds may be subdivided into classes differing
from one another in a minor degree. Thus, in the red class there

114 Experiments with Primula sinensis

are dark reds, of which a few approximate to "Crimson King," terra-
cottas of two shades, one bluer (Plate XXXI, fig. 47), the other a clean
red (fig. 48) and light reds corresponding with both the shades of
terra-cotta ; in the magenta class a similar series of forms occurs. The
grading between the sub-classes is close and I am not able to give any
precise numerical results as to the proportions of the various types.
The distribution of the green and red stigma among the magentas and
reds clearly indicates the existence of partial gametic coupling between
the factors for magenta and green stigma (see p. 127).

^3 families have been raised from certain of the F^ forms in the
hope of elucidating their relations to one another and to the " Straw-
berries." The bluer terra-cotta appears to be differentiated from the
red kind by the addition of a single factor, but for the most part the
results are complex and further data are required for their detailed
analysis. One result, however, is of interest in connexion with the
relation between the magenta and red colour. An F^ plant with
peculiar deep rosy flowers and red stigma, when selfed, gave forms
like itself and strawberries ; a light red with green stigma, self-
fertilized, gave light reds, terra-cottas of both shades, and strawberries,
all with green stigma. The two plants were crossed together recipro-
cally, and the two families thus obtained consisted of typical magentas,
reds (including light reds and terra-cottas) and strawberries, all with
green stigma.

"Orange King." (Plate XXX, fig. 8.) "Orange King" originated
with Messrs Sutton as a sport from a strain of " Crimson King " ; it bred
true from its first appearance. The F^ from the cross with " Crimson
King" bears an exceedingly close resemblance to the latter; the mature
flowers of the hybrid are probably not to be distinguished from those
of the pure race, but in the young flowers there is a slightly more
magenta tint than in the pure strain of " Crimson King " with which I
have worked. In the F^ from this cross there were obtained 55 plants
like the F-^, and 14 "Orange King"; some very slight differences in
the depth of the colour were noticeable among the latter. The ex-
tracted "Orange King" had the true red stem-colour, as compared
with the purplish-red colour of the forms resembling " Crimson King."

" Orange King " x " Snowdrift." The F^ of this cross is indistin-
guishable to the eye from that of the crosses of either the Rosy-
magenta or "Crimson King" with "Snowdrift." The constitution of
the F^ follows the general lines of the F^ from ("Crimson King"x
"Snowdrift") but is of course rather more complex, since the ^i is

R. P. Grbgory 116

heterozygous for the factor determining the purplish-red stem and deep
colour of " Crimson King," which is present both in that race and in
*' Snowdrift." In addition therefore to the types found in the "Crimson
King" F^ there appear extracted "Orange Kings," and a new class
consisting of plants with pink or pale-pink flowers and stem-colours
ranging from red collar to reddish stem. These plants are no doubt
derivatives of "Orange King," whose appearance they rather recall;
but further experiment is required upon this point, as well as upon the
further point as to whether the " Sirdar " character is recognizable as
such, if, and when, it occurs in the " Orange King " series of pigments.
The numbers obtained in two F, families were :

Pink, PiJepink,

red collar to faint tin^e or .Mliite,

Full colour " Sirdar " '' Orange King " reddish stem no colour m stem green stem

111 33 5 29 52 12

178 64

The numbers given in the last three classes can only be regarded
as approximately representing their relative sizes, since one can hardly
avoid some experimental error in a separation guided by external
appearance only. It will be seen that, if the pink class prove to be
derivatives of " Orange King," the numbers obtained agree with the ex-
pectation based on the hypothesis suggested by the result of the cross
(" Crimson King" x " Orange King"), namely, that the subtraction of a
single factor will suffice to explain the behaviour of the "Orange King"
type of pigment.

The existence of some form of partial gametic coupling between the
magenta and green stigma is clearly indicated (see p. 127).

(3) Inhibition of Colour in the Flower.

All the red-stemmed whites with which I have worked have been
found to possess the factors which inhibit the development of colour
in the flower; when crossed with the albino "Snowdrift," they have
given colours in F^. Since fully coloured flowers only occur in con-
junction with fully coloured stems, the stem-colour of the dominant
white is a guide to the flower-colours which may appear in the F^;
those with full red stems will give full colours, while those with no
more than a tinge of colour in the stem can only give pale-pinks.
The precise ratio in which the coloured forms appear in Fj is still in
doubt. In the F^'s consisting of whites and pale-pinks only the former
are in excess of the expected ratio of 13 : 3. Owing to the difficulty
of distinguishing these faint colours, no great weight could be attached

116 Experiments with Primula sinensis

to this discrepancy, were it not that in some F^%, which contain plants
with fully coloured stems, there is again a considerable excess of whites
in the red-stemmed class, where the distinction between white and
coloured forms can be made with certainty. The numbers which have
been obtained are' :

stems not fully
coloured (including
Red stem those resembling " Sirdar " *)

Dominant White ''■ ~ ' ^

Parent White Magenta "Sirdar" Pale pink White

Giant White '

(18 5 0 2 4

(33 8 3 6 33

"Primrose Queen" 66 13 5 7 36

* Without the character of the flower-colour as a guide it is scarcely possible accurately
to distinguish the " Sirdar" type of coloured stem from other low grades of stem-colora-
tion.

Before passing to a detailed consideration of these results, it is well
to recall the fact that in the F.^ from crosses between plants having
coloured flowers and stems x the albino " Snowdrift," oil the red-
stemmed oflfspring have coloured flowers, whites being found only in
the green-stemmed class. These results, together with the fact that
all my red-stemmed whites proved to be dominant whites, suggested
that the factors for full colour are common to the whole plant, and
that, in general, red-stemmed whites are white in virtue of the
suppression of the colour in the flower by inhibiting factors 2.

Turning now to the results of the crosses between " Giant White" x
" Snowdrift," it will be seen that the red-stemmed class consists of
whites and colours, in proportions which do not diverge so greatly from
the expected ratio (3 : 1) as to exclude the possibility of accounting for
all the whites on red stems as resulting from the suppression of colour
in the flower.

In the red-stemmed class of the ^2 from "Primrose Queen" x
" Snowdrift," however, the whites are much more than three times
as numerous as the plants with coloured flowers. The observed ratio
of colours to whites agrees closely with the expectation based on the
hypothesis that the production of colour in the flower, even in the
red-stemmed offspring of this cross, depends upon two complementary
factors, for both of which the ^1 was heterozygous. An F-^ heterozygous
for these factors and for inhibition, would give an F^ consisting of
9 coloured : 55 white ; the numbers obtained are 13 coloured, 66 white
(easpectation : ll'll : 67'89).

1 The earlier experiments only give qualitative results, as many plants were discarded
before the characters of the flower could be accurately determined.

2 Gregory, Rep. Brit. Assoc., Leicester, 1907, p. 692.

R P. Gregory 117

Other experinietits made with " Primrose Queen " definitely support
the view as to its constitution which is entailed by this hypothesis.
The results of Keeble and Pellew's experiments with the red-stemmed
"Snow King"* indicate that in certain cases the factors for colour may
be absent from the flower, though present in the stem, and consequently
that certain red-stemmed plants may have white flowers in the absence
of inhibition. On the other hand, the mode of inheritance of the full
colour in my crosses between coloured, red stem x " Snowdrift " suggests
that in certain other cases the factors for colour are common to the
whole plant, both stems and flowers.

Dominant white x Coloured, green stigma. The simplest cases illus-
trative of the operation of the factors which inhibit the development of
colour in the flower are those in which a dominant white is crossed
with a coloured form having green stigmas. The ^i in these cases is
white or tinged-white, the depth of the tinge depending, under uniform
conditions^ upon the intensity of the colour of the coloured parent, and
to some extent upon the particular race of dominant white used. The
^2 from this cross consists of whites, tinged whites and colours, all with
green stigmas. The numbers obtained are :

fi X coloured,
F^ X Self green stigma

/'ttemiliea

White and
Tinged white

Coloored

782

271

Expectation

789-75

263;i5

Number of White and

Ft familiee Tinged white Coloured

3 59 58

Equality

The experiment has been repeated in a slightly different form by
crossing coloured plants with the ^i of (Dominant white x Recessive
white). The numbers obtained from these crosses are :

Reference Xomber

of 1*1 plant

White

Coloured

28/4

4/6

36/6

30/6

61/9

Totals

247

~ 274

Expectation

260-5

> Joum. Genetic*, Vol. i, 1910, p. 1.

* The depth of the tinge is dependent upon the conditions under which the Fi is
grown, and its maximam development is only obtained by keeping the house as cold as
is possible without injury to the plants. At higher temperatures very Uttle tinge is
developed, and the F^ from the cross of such an intense colour as " Crimson King" with
a dominant white is scarcely tinged.

118 Experwients with Primula sinensis

The "dominant white" parent of Nos. 26/6 and 30/6 was one
which gives a very fully tinged ^i when crossed with " Crimson King "
— the coloured race with which 26/6 and 30/6 were mated ; the excess
of coloured offspring shown in their crosses may therefore be in part
due to experimental error, through the inclusion of some deeply tinged
forms with the light colours, and in the absence of any other indica-
tions of departure from normal segregation one does not feel inclined
to attach any great weight to the discrepancy shown here.

Dominant white x Coloured, red stigma. The ^i from this cross is
again a tinged white with green stigma (Plate XXX, fig. 18; Plate XXXI,
fig. 21). The F2 from one of these crosses — that between "Queen
Alexandra" and "Crimson King" — is illustrated in Plate XXXI,
figs. 22 — 43. As concerns the factors for inhibition, the F2 consists of
four classes, namely, (1) whites and tinged-whites, with green stigma
(Plate XXXI, figs. 22 — 26); (2) plants in which the peripheral part of
the corolla is white or tinged, the central part flushed, with red stigma
("Duchess" and "Buller" types; figs. 27 — 31); (3) coloured, green
stigma (figs. 33, 34, 38 — 41); (4) coloured, red stigma (figs. 35 — 37,
42, 43). The four classes are in the proportions of 9 : 3 : 3 : 1, the
observed numbers being :

White and tinged-
svhite, green stfgma

" Duchess " and " Buller "
forms ; red stigma

Coloured,
green stigma

Coloured,
red stigma

193

Expectation 191-25 63-75 63'75 21-25

" Duchess." The " Duchess " types which appear in these -Fg's are
shown by experiment to be homozygous for the peripheral inhibiting
factor. Crossed with a coloured, red stigma, they give " Sir Redvers
Buller," which in turn gives "Duchess," "Buller," and fully coloured, all
with red stigma. The F^ types resembling "Buller" are therefore those
which are heterozygous for the peripheral inhibiting factor.

" Duchess " X green stigma. " Duchess," crossed with plants with
green stigma, gives a white or tinged-white -Fj. The result is the same
whether the parent having the dominant green stigma be a coloured
form or a recessive white (" Snowdrift "), except that in the former case
the Fi has a rather deeper tinge.

In certain cases the flowers of the ^1 have a distinct tinge of colour
in the corolla-tube, just below the region of the insertion of the anthers,
although no tinge at all may be discernible in the petals^ The charac-

1 A similar character lias been observed in one other experiment where the Fi from
(Dominant white x Crimson, green stigma) was crossed with a dominant white. In this
case the character was coupled with that of short-style.

Croas

Namber of
famUies

' Daebess '

' x" Snowdrift" ...

'Duchess'

' X '* Sirdar "

' Duchess "^

'x "Ivy-leaf"

R. P. Grkcm)ry 119

ters of the F,'s from the various crosses which have been made are
shown below :

VntnhAr ni

Deaeriptton
21 plants. White, petals tinged, no

tinge in tube.
28 plants. White, no tinge seen.
42 plants. White, with faint tinge
in petals ; no tinge in tube.

" Dnchess " X Dominant White 1 8 plants. White, with distinct tinge

in tube.
" Dnchess "xfj (Dominant White X" Snowdrift") 2 White, no tinge seen, 15 plants;

White, tinged in tube, 14 plants.

" Duchess " X Rosy- Magenta 1 12 plants. White, rather fully

tinged in petals.

The Fi fix)m the cross between " Duchess " and " Snowdrift " con-
tains a long series of types, for to the various inhibited and coloured
forms corresponding with those obtained in the F.2 from (Dominant
white X CJoloured, red stigma) there are added the " Sirdars," pale-
pinks and whites on green stems which are characteristic of the F^'s
from crosses between "Snowdrift" and plants possessing the factors for
colour. And since the " Duchess " used in these experiments was of
the red class, red as well as magenta forms of each coloured type are
present. The numbers obtained in three families were :

Green stems, red collar
Bed stems (" Sirdar " tn>e8) Pale-pinks WMtes

No colour No colour
Bed seen in seen in
Green stigma Bed stigma Green stigma Bed stigma oc^ar stem stem

Tinged
White and white and Tinged

Tinged white Coloured Coloured White Sirdar White Sirdar

117 47 37 47 14 18 7 5 19 56

The tinged whites with green stigma are of two kinds, namely,
(1) those which resemble the Fi in having a more or less evenly
distributed tinge, which becomes more pronounced as the flowers fade,
and (2) those with a definite central tinge surrounding the eye and
most conspicuous in the young flower. All the tinged whites with red
stigma have the colour disposed in the centre after the " Duchess "
style.

At the time when these families were recorded the distinctive
character of the forms resembling "Buller" had not been recognized, and
some of them were included with the class " Coloured, red stigma " ;
in the table the two classes of red-stemmed plants with red stigma are
therefore taken together ; it will be noticed that there is a deficiency

120 Experiments with Primula sinensis

in these two classes as compared with the corresponding classes with
green stigma ; on the other hand in the " Sirdar " classes the propor-
tions of green stigma and red stigma are slightly less than 3 : 1, but
further experiment is required before any suggestions can be made
as to any possible significance of these departures from normal distri-
bution.

A further generation w^as raised by selfing one of the offspring of
the cross [" Duchess " x ^i (Dominant white x " Snowdrift ")]. The
most interesting point brought to light by this experiment is the fact
that there occur whites (? with no tinge) having red stigmas, but
without the central flush of deep colour which is characteristic of the
" Duchess" strain \ The numbers obtained were :

Ked stem Green stem, red collar No colour in stems

White,

green stigma

red stigma

34*

13t

Duchess, White, White, White,

red stigma green stigma red stigma green stigma

4 16 5 19

• Two with definite central tinge.

t Three with definite central tinge.

Tinged-whites with red stigma and without the central flush
occurred also among the offspring of a cross between " Duchess " and
the i^i of ("Ivy-leaf" X " Crimson King"). From this cross 31 plants
were obtained, 16 with green stigma, 15 red stigma. Those with green
stigmas were white or slightly tinged (like the F^ of "Dominant
white " X coloured, red stigma) ; those with the red stigma were deeply
tinged, but whereas some were of typical "Duchess" or "Buller" types,
others were without the deep central flush.

The deep central flush of " Duchess" and "Buller" is therefore not a
necessary consequence of the absence of the factor inhibiting colour in
the stigma; it would appear rather that the character is an independent
one, but, like the deep spot of colour just external to the eye (p. 106),
is dependent for its full development on the presence of colour in
the stigma. We may surmise that the definite central tinge found in
some whites with green stigmas represents this character in combina-
tion with the green stigma.

Green stigma in coloured flowers. The results showing the behaviour
of the green and red stigma in crosses between colours are :

Number of families Green stigma Bed stigma

4 315 116

Expectation 323-25 107-75

1 In Stellata flowers the "Duchess" flush is often only poorly developed, but the
phenomenon is of a different kind from that referred to here.

R. P. Gregory 121

(4) Flakes.
The Fj's from crosses between the "Ivy-leaf" and coloured races
contain flakes (Plate XXXI, figs. 56—59) in addition to the self-colours.

"Ivy-leaf" x "Crimson King" The Fi of this cross, and of that
between "Ivy-leaf" and "Orange King," is indistinguishable to the eye
from the Fi of the crosses between " Snowdrift " and the same coloured
races.

In the F,, the self-coloured flowers on red stems constitute a series
of types similar to those of the F^ from ("Crimson King" x "Snowdrift");
the same series is probably repeated in the flaked patterns, though
the distinction between the shades of red and magenta is much less
easily made in the flakes. The flaking may be coarse, the coloured
areas taking the form of wider or narrower radial stripes (Plate XXXI,
fig. 56), or very fine flakes may be present in addition to the coarser
ones (fig. 58). The flaking appears generally to be strongly marked in
plants with red stigmas (figs. 57, 59).

As in many other cases, the distribution of green and red stigmas
among the magentas and reds clearly indicates the existence of partial
gametic coupling (see pp. 127, 128).

All the ofispring with light stems have some amount of colour in
the bases of the leaves, as does the "Ivy-leaf" itself. The light-
stemmed class consists of pale-pinks, and whites flaked with pale
colour. The flaking in this class may be very sparse, and in that case
is inconspicuous as the colour is so faint, but it was observed in all the
plants except four. The F^ contains no " Sirdars."

The numbers obtained were :

Red stems Light stems

Beference Self Pale WTiite, flaked White, no

Number ooloor Flake pink pale pink flake seen

61/10 1 „ , , , I 97 33 30 6 2

Palmate leaves

62/lof '^*"^'^'^^^' 1173 42 35 25 2

Total, palmate 270 75 65 31 4

61/10 \ I 24 10 2 0 0

\ Ivy leaves s + 3 undetermined*

62/10) I 40 16 12 3 0

+ 11 undetermined* +5 undetermined*

Total, Ivy leaves 64 26 14 3

+ 14 undetermined* + 5 undetermined*

Grand total 334 101 79 34

* Owing to the poor development of the flowers.

Joum. of Gen. i

122 Experiments with Primula sinensis

Taking the red- and light-stemmed classes together, the self-colours
are 413, the flakes 135, numbers which suggest that the flaked
condition is a simple recessive, the expectation in such a case being
411 colours : 137 flakes. The distribution of the self-colours and flakes
among the red and light stems is however irregular, especially in the
palmate plants of 62/10.

"Ivy-leaf" x Dominant white. Up to the present time F^b have
been raised from only one cross of this kind, that of ("Ivy-leaf" x "Prim-
rose Queen "). The F^ is noteworthy for two reasons : (1) no self-colours
are obtained, all the coloured offspring being flaked ; and (2) no pale-
pinks occur. The numbers are :

Red stems Light steins

White Magenta flakes White Flakes

Palmate leaves 93 26 36 —

Ivy leaves ... 43 4 8 —

The flakes grade from fully-flaked to small and sparse flakes of
colour. It may be noticed that the young flower-buds of the flaked
forms are quite strongly tinged, even though the flaking may prove to
be sparse. The great excess of whites, as compared with flakes, among
the Ivy-leaved offspring is probably of no great significance, as it
may well be due to the reduced corollas of the Ivy-leaves. The
plants with light stems were carefully examined in view of the results
obtained from the cross of ("Ivy-leaf " x "Crimson King"), and no trace
of flaking was observed in any. It may be remarked however that
both "Primrose Queen" and "Ivy-leaf" carry the factor which partially
suppresses flower-colour, and even the full colours are very light.

" Ivy-leaf" x " Snowdrift." The F^ from this cross has definite,
though faint, colour in the flowers. In the F^, plants with definitely
coloured flowers form approximately 9 in every 16 plants, the observed
numbers being 144 definitely coloured in a total of 273 plants
(t6 *^f ^'^^ — 1^3'5). The plants recorded as definitely coloured were,
so far as could be judged, self-colours ; in one at least of the remainder
the "ghost" of a coloured flake was recognized. In this cross, again, both
parents bring in the factor which partially suppresses flower-colour.

Discussion of the " Ivy -leaf " crosses. The appearance of the "Ivy-
leaf" plants and the characters of the Fi obtained from their crosses
with colours, led me to look upon " Ivy-leaf" as a recessive white ; but
the result of the crosses with " Crimson King " suggests that this view
will need revision, and that the plant may really possess the pale-pink

K. P. Gregory 123

colour in the flaked condition. A re-examination of the parent " Ivy-
leaf" in the light of this suggestion failed to reveal any definite
coloration, but the pale colour is at best sometimes hard to discern
and in the flaked condition might escape even close inspection,
especially in such poorly developed flowers as are characteristic of the
" Ivy-leaf." The suggestion is moreover supported by the fact that
Keeble and Pellew^ obtained a flaked Fi from the cross of an "Ivy-leaf "
of this strain with "Snow King." This view of the constitution of
the " Ivy-leaf" would agree well enough with the results of the cross
with " Snowdrift," for the latter possesses the factor for self, as
against flaked, colour, and we should therefore expect a ratio of
9 self-coloured : 7 flaked and white.

In the same way the absence of pale-pinks in the F^ of the cross
with " Primrose Queen " may perhaps be put down to the difficulty of
recognizing the colour in its most dilute and flaked condition-. The
complete absence of self-colours from this ^3 is interesting in view of
the results of the cross between " Primrose Queen " and " Snowdrift,"
and suggests some considerations as to the relation between flakes and
self-colours. For if the self-colours result from the addition of a
" distributing " factor epistatic to the factors for colour, it is clear that
"Primrose Queen" must be without this factor; but in that case
one-third of the coloured ofi^spring obtained in the F., from (" Primrose
Queen " x " Snowdrift ") sh<mld be flaked, and no flakes have been
obtained in this cross.

If, then, the conception of distributing factors is to be retained, it
would be necessary to construct an elaborate scheme of factors, for the
existence of which there is at present no evidence. In the absence of
such evidence, it is more simple to suppose that one, at least, of colour
factors may exist either in the flaked or in the distributed condition.
The ^1 from (self-colour x flake) then appears self-coloured because the
flaked character is masked when the flower as a whole is coloured';
and the segregation which takes place in the hybrid consists in the

* Joum. Genetics, Vol. 1. 1910, p. 4.

* Flaked pale-pinks have now (Feb. 1911) been definitely recognized in F, from this
cross. A red-stemmed magenta flake, self-fertilized, gave two kinds of ofifspring, namely
(1) plants with red stems and flowers flaked magenta, (2) plants almost devoid of colour
in the stem, in the flowers of which the flakes of pale pink were rec(^:nized with certainty.
Temperature of the house, 5-5° F.

* Whether any pale self-colours, crossed with flakes, would give an Fj of a visibly
flaked character depends upon the relation between full and pale colours, which is not yet
fully understood.

9—2

124 Experiments with Primula sinensis

separation of the flaked and distributed forms of the same factor, and
not in a segregation of the factors for flaked and for self-colour from
their respective " absences."

On this hypothesis the results of the crosses with "Primrose Queen"
may be explained as in the subjoined scheme, where

X, Y, the colour factors in the distributed condition ;
X\ Y', the colour factors in the flaked condition ;
R, inhibition.

Assuming the constitution of the parents to be " Snowdrift" = xYr;
" Primrose Queen " = XyR ; " Ivy-leaf" = XTr; then

"Snowdrift" x "Primrose Queen," F^ = XxYyRr,
F2 should give 9 se(/-coloured : 55 white
Observed 13 „ : 66^ „ "

Calculated ll'l „ : 67'9 „

"Ivy-leaf" x "Primrose Queen," F, = XXY'yRr,
F2 should give 3 flakes : 13 whites
Observed 26 „ : 93^

Calculated '22'3 „ : 96-7 „

"Ivy-leaf" x "Snowdrift," F, = XxYrrr,

F^ should give 9 self-coloured : 7 flaked and white
Observed 144 „ : 129

Calculated 153-5 „ '.119-^

The conception of the relation between the flaked and self-colour
characters set forth above does not pretend to do more than provide a
means by which the results of the experiments may be described. It
brings us no nearer the solution of the problem as to how the flaked
distribution is brought about, nor is it intended as implying that the
colour-factors themselves may not be the same in the flakes as in the
self-colours, the mode of distribution of the colour being determined
independently.

Gametic Coupling and Repulsion.

Evidence has been obtained of the existence of (1) a repulsion
between the factor for the structural character of short-style and certain
factors affecting the colour of the flower, and (2) of a partial coupling
between two colour factors.

1 The numbers given are those obtained in the red-stemmed class only, since the
distinction between coloured and white green-stemmed plants is not critical.

R P. Gregory 125

(1) Repulsion between short-style and colour characters.

Short-style and Magenta-colour. The results of my crosses in which
a short-styled coloured race (Salmon-pink, p. 109) was mated with
various long-styled plants carrying the magenta factor, show that in
the gametogenesis of the hybrids so produced, a complete repulsion
between the factor for short-style and the magenta factor takes place.
The numbers obtained in these experiments are given below, together
with the expectation based on the assumption of complete repulsion
between the two factors under consideration.

Magenta short-styled Fj, giving niagentas and saJmon-pinks.

Short-«tyle Long-style

Magenta Salmon

Observed Numbers ... 54 18

Expected „ ... 48 24

Tinged-white short-styled F^, giving tinged-white, magenta, and salnumrpink.

Shortrgtjle Long-style

Tinged-white Magenta Salmon Tinged-white Magenta Salmon

Observed nnmbers 157 46 19 66 20 0

Expected „ 173-25 38-5 1925 57-75 1925 0

Tinged-white short-styled F^, giving tinged-white, magenta, saimon-pink,

and hlrie.

Short-style Long-style

Tinged- Tinged-

white Magenta Salmon Blue white Magenta Salmon Blue

Observed nnmbers 15 4 3 2 10 6 0 0

Expectation {omitting^ ^^.^ ^.^ ^.^ _ j,.^ ^.^ ^

iluet) I

The deficiency in the tinged-white short-styles is referred to below.

Short-style and inhibition. Certain families raised from one
heterozygous short-styled tinged-white (No. 51/9) have shown an
interesting departure from the normal distribution of the four kinds
of offspring. The results of the experiments made with this plant are :

Offspring

Cron

51/9xSelf

51/9 $ X Long-stjle, coloured i
51/9 ? X Long-style, coloured <f
Long-style, coloured $ x 51/9 i

She

Tinged-
white

5
7

irt-style

Long-style

Beferenoe
Komber

66/10

67/10

68/10

79/10

Colooied

Tinged-
white Coloured

9 3
11 4

5 5
17 8

126 Experiments with Primula sinensis

In Experiments 66/10 and 68/10 the distribution is normal, but in
Nos. 67/10 and 79/10, where we expect equality of all four classes, the
two middle classes are much larger than the end terms, and though the
total numbers are small, the divergence is such that it can scarcely be
dismissed as fortuitous.

The relative sizes of the four classes in the families Nos. 67/10
and 79/10, indicate that any repulsion which may take place must be
of a low order. The family of 67/10 was raised from seeds obtained
from two capsules; but in Experiment 79/10, pollen was taken from
only one flower, so that the low type of repulsion indicated by the
constitution of that family cannot be regarded as due to a mixture
of families of more than one kind.

For the solution of the problems presented by the results of these
experiments further data are required. The constitution of the F^^
obtained from our other plants heterozygous for short style and for
inhibition, throws little light on the case. In these families the
distribution of the four kinds of offspring does not depart very greatly
from the normal. The numbers obtained are :

Short-style Long-style

White and
Tinged-white

Coloured

White and
Tinged-white

Coloured

Observed ..

327

126

128

Expected . .

348-75

116-25

38-75

There are small departures from the 3 : 1 ratio in the cases both of
the short- and long-style and of inhibition and colour ; the deficiency of
dominants of both kinds has of course a marked effect upon the size
of the first category.

The excess in the two middle classes and the deficiency in the first
class appear to be more or less constant throughout the families, which,
combined together, furnish the totals given above.

Short-style and tinge in corolla-tube. In one experiment clear
indications were obtained of a coupling between the short-style and
a tinge in the corolla-tube just below the level of the anthers. So
far as the character of the petals was concerned, the family consisted
of tinged- whites and whites ; of the whites, however, some were tinged
in the tube, others were colourless. Of the short-styled whites, 20 were
tinged in the tube, 1 was colourless ; of the long-styled whites 5 had
the tinge and 11 were colourless. The ratio of short- to long-style in
this family was very much less than the expected 3 : 1, the numbers
being 33 short, 21 long. The asymmetry of the four classes is no doubt

R. P. Gregory

127

partly a result of this, and the numbers obtained suggest that a coupling
of a fairly high order was present.

(2) Coupling between colour characters.

With the exception of the mating between "Crimson King" and
" Snowdrift," all the experiments in crossing either " Crimson King " or
" Orange King " (red, red stigma) with plants carrying the factors for
magenta colour and green stigma, have given results which point clearly
to the existence of a partial coupling between these factors in gameto-
genesis. The results in general show some deficiency of the two
dominant characters, magenta and green stigma, as compared with
the expected ratio, in each case, of 3 D : 1 R ; the classes consisting
of plants having one or both recessive characters are therefore unduly
large, and it is necessary to make allowance for this in attempting to
compare the observed numbers with the expectation calculated upon
various systems of partial coupling. The distribution of the four
characters in the oflFspring possessing fiilly-coloured flowers and red
stems is set out below :

Constitution of F^

Cross
" Orange King " x "Snowdrift '

" Crimson King " x Rosy Magenta

"Crimson King"x "Queen Alexandra'

" Crimson King " x •• Ivy-leaf "

Magenta

Red

Beference

Green

Red

Green

Red

Number

stigma

10/9

11/9

33/9

Total

101

17/9

18/9

24/9

32/10

33/10

Total

61/10

62/10

137

Total

206

In these results, that obtained in No. 18/9 stands alone in that the
fourth term of the series is distinctly larger than either of the middle

128 Experiments with Primula sinensis

terms. On the basis of the 3:1:1:3 system of partial coupling, the
122 plants of which the family consists should be distributed in the
proportions of

78-1 magenta, green stigma : 133 magenta, red stigma :
13*3 red, green stigma : 17*2 red, red stigma.

The distribution thus calculated would, however, give magenta : red =
green stigma : red stigma = 3:1, while the observed numbers are
magenta 85, red 37 ; green stigma 88, red stigma 34. That is to say,
there is a deficiency of plants bearing the dominant characters, and,
consequently, the first term will be smaller and the fourth term will
be larger than the calculated numbers. Apart from this divergence,
there is a close approximation between the observed and calculated
numbers, and there can be little doubt that the partial coupling was
of the type 3:1:1:3.

The crosses between " Orange King " and " Snowdrift " have given
results which may perhaps allow of the same explanation, but in the
remaining experiments the fourth term is definitely smaller than the
middle ones. Each family was raised from seed obtained from several
capsules borne on one plant ; and, until the completion of experiments
which are now in progress, it is not possible fully to analyse the
results. For the time being it may be pointed out that a very close
approximation to the observed numbers is given by the assumption
that a coupling of the form 7:1:1:7 is present in gametes of one
sex only, gametes of the opposite sex being produced in equal numbers
of all four kinds. The numbers are

Magenta,
reen stigma

Magenta,
red stigma

Red.
green stigma

Red.
red stigma

411

416'8

96-2

74-8

Observed

Calculated for 7:1:1:7 coupling
in gametes of one sex only

As was stated above\ the distribution of the four kinds of offspring
in the F^^ from the cross (" Crimson King " x " Snowdrift ") gives no
clear indication of the existence of any form of partial coupling during
gametogenesis of the ^i. In two F^. families raised in 1907, the results
differ very little from the simple 9:3:3:1 ratio. In the later experi-
ments there is some departure from this ratio, principally due to the
dearth of plants carrying the positive characters, magenta and green
stigma.

1 p. 113.

R. P. Gregory 129

In conclusion, it may be pointed out that here, as elsewhere, families
raised from sister plants do not necessarily follow the same system of
distribution. Thus the parent of the family 18/10, which apparently
conforms to the 3:1:1:3 system, was the sister plant to the parent of
the family 17/10, in which the fourth term is smaller than the middle
ones.

Note added Febraary 17, 1911. Since the foregoing was written some interesting
results have been obtained in connexion with the phenomena of coupling and repulsion.
These results are briefly described below. The constitution of a certain type of coloured
flower, which was previously uncertain, has also been ascertained.

Coupling and repulsion. (1) Magenta and short-ttyle. On p. 12.5 a series of experiments
is described in which a complete repulsion between the factors for short-style and magenta
colour is shown. It will be noticed that in this series of experiments one of the dominant
characters was possessed by the one parent race, the other by the other parent. In a new
series of experiments a race possessing both domumnt characters (i.e. magenta, short-
style) was mated with races which had the recessive characters only. The results show
that when the cross is made in this way, partial coupling takes place between the factors
for the two dominant characters.

In the experiments in which the recessive parent was a long-styled red with doable
flowers, the coupling shown is almost certainly of the form 7:1:1:7; in these experi-
ments there is no indication that coupling occurs between either the factor for short-style
or that for magenta and any third factor.

In a second set of experiments, in which the recessive parent was the long-styled
"Crimson King," the form of the coupling between magenta and short- style is as yet
uncertain, the numbers obtained being almost exactly intermediate between the expectation
based on the series 7:1:1:7 and that based on the series 15 : 1 : 1 : 15. In these
experiments there is clear evidence that the factor for magenta is coupled, not only with
the factor for short-style, but also with the factor for green stigma. The coupling between
magenta and green stigma is of a mnch lower order than that between magenta and short*
style.

(2) Light red ttem and green stigma. A new instance of complete repulsion between
two factors has been obtained. The factors are (1) the pallifying factor for stem colour
(p. 100), and (2) the factor for green stigma. This repulsion was observed in the progeny
of a cross between "Crimson King " and "Ivy Leaf." Certain individuals of the F^ from
this cross were tested by self-fertilization. Three of these plants, all having light red
stems and green stigma, were found to be heterozygous in these characters, giving offspring
having either light or dark red stems, and either green or red stigmas, but none of the
dark-stemmed offspring had red stigmas.

Flower-colour. The deeply tinged type of flower shown in Plate XXXI, fig. 32, the eonsti-
tntion of which was formerly in doubt, has proved itself to be heterozygous for the factors
which inhibit flower-colour. Self-fertilized, it has thrown tinged whites with green stigma,
" Duchess " and " Boiler " types with red stigma, coloured with green stigma and coloured
with red stigma. The flush shown in the "Duchess "and "BuUer" types is of a very
deep kind, and the coloured types have flowers of a very deep crimson, at least as deep as
that of " Crimson King."

130 Experiments ivith Primula sinensis

DESCRIPTION OF PLATES.

The figures illustrating the colours of the flowers are from water-colour drawings
by Miss M. Wheldale.

PLATE XXX.

Figs. 1 — 7 illustrate various types of coloration of the stem.

Fig. 1. Dominant white with deep purplish-red stem.

Fig. 2. Dominant white with light purplish-red stem.

Fig. 3. " Orange King," red stem (cf. the more usual purplish-red colour of the stem).

Fig. 4. " Sirdar." The characteristic distribution of the flower-colour, which ia

associated with this type of stem, is shown in Plate XXXI, figs. 44, 45.
Fig. 5. Green stem, with slight purplish-red colour in young petioles.
Fig. 6. "Ivy-leaf." Non-crenate leaves, and monstrous flowers. Stems green, with

some purplish-red colour in young petioles.
Fig. 7. " Snowdrift," stem devoid of sap-colour.

Figs. 8 — 20 illustrate the colour of the flower of various pure races and Fi's.
Fig. 8. "Orange King" (cf. fig. 3).
Fig. 9. " Crimson King."
Fig. 10. "Snowdrift" (cf. fig. 7).

Fig. 11. "Queen Alexandra" (dominant white, white eye).
Fig. 12. "Primrose Queen" (dominant white, large yellow eye).
Fig. 13. "Beading Pink." The deepest colour found in association with green stems

devoid of sap-colour.
Fig. 14. Fx (" Reading Pink " x " Snowdrift ").
Figs. 15 and 16. Young and mature flowers of the Fi from ("Reading Pink " x " Orange

King").
Fig. 17. Fi ("Crimson King" x " Snowdrift").
Fig. 18. Fi ("Crimson King " x Dominant white, ordinary eye).
Figs. 19 and 20. Rosy-magenta, young and old flowers.

PLATE XXXI.

Figs. 21 — 43 illustrate the cross between "Crimson King" and "Queen Alexandra."

Fig. 21 is the Fi; figs. 22 — 43 illustrate the series of F^ forms.

Figs. 22 — 26. Inhibited whites, green stigma.

Fig. 22. White, white eye.

Fig. 23. White, ordinary eye.

Fig. 24. Tinged- white, white eye (Fj type).

Fig. 25. Tinged-white, ordinary eye.

Fig. 26. Fuller tinged-white, tinge central

Fig.

28.

Fig.

29.

Fig.

30.

Fig.

31.

Fig.

32.

Figs

. 33

figs.

Fig.

33.

Fig.

34.

Fig.

35.

Fig.

36.

Fig.

37.

R P. Gregory 131

Figs. 27—31. Plants with the factor for inhibition of colour in periphery of petals, bat

with red stigma.
Fig. 27. Bed " Duchess," f i tyi)e of eye.

Magenta "Duchess," Fi type of eye.
"Sir Redvers Buller" (red), ordinary eye.
Red "Buller" of rather bluer type, white eye.
Magenta " Buller," ordinary eye.
Light red, ? inhibited form, green stigma i.
—37 represent types belonging to the Bed class. Figs. 33, 34, green stigma;
3o — 37, red stigma.
Red, ordinary eye.
Bed (bluer type), white eye.
Bed, white eye.

Bed, ordinary eye, band of deep colour round the eye.
Deep red, white eye.

Figs. 38 — 43 represent types belonging to the Magenta class. Figs. 38 — 41, green stigma ;
figs. 42, 43, red stigma. Figs. 41 and 43 are corresponding forms with green and
red stigma respectively.
Fig. 38. Magenta, Fi type of eye.
Fig. 39. Magenta, ordinary eye.
Fig. 40. Fuller magenta, white eye.
Fig. 41. Bosy-magenta, ordinary eye.
Fig. 42. Magenta, Fi type of eye.

Fig. 43. Bosy-magenta, ordinary eye, spots of deep colour round the eye.
Figs. 44 — 45 represent additional coloured types which occur in the F^ from (" Snow-
drift" x " Crimson King").
Fig. 44. Bed "Sirdar."
Fig. 45. Magenta " Sirdar."

Fig. 46. Pale-pink on green stem (cf. figs. 13, 14).
Figs. 47 — 49. Other types belonging to the red class from the fg from (" Crimson

King " X Bosy-magenta).
Fig. 47. Terra-cotta (bluer type), green stigma.
Fig. 48. Terra-cotta, green stigma.
Fig. 49. Strawberry.

Figs. 50 — 55 represent various types of special coloration just external to the eye.
Figs. 50, 51. Magenta, red stigma, with the spots of deep colour which are only fully

developed in flowers possessing the red stigma. Young and mature flowers.
Figs. 52, 53. Magenta, green stigma, corresponding with the foregoing types. Young

and mature flowers.
Figs. 54, 55. Young and mature flowers of a magenta with green stigma, showing the
rather diffuse band of brownish colour, which only becomes conspicuous in the older
flower (fig. 55).
Figs. 56 — 59 represent Flaked types.

Fig. 56. Flakes medium to coarse ; no fine flakes, green stigma. Stellata.
Fig. 57. Some coarse flakes ; finer flakes rather peripheral in distribution ; red stigma.

Sullata.
Fig. 58. Fully flaked, coarse and fine flakes, green stigma.
Fig. 59. Fully flaked, coarse and fine flakes, red stigma.

1 See note added February 17, 1911, p. 129.

132 Experiments with Primula sinensis

PLATE XXXII.

Fig. 60. Seedling plant of "Ivy-leaf," showing the non-crenate leaves.
Fig. 61. Illustrating the cross between " Ivy-leaf" and " Snowdrift " (Fern-leaf, crenate).
Top row: "Ivy-leaf" (left), "Snowdrift" (right).
Middle: Fi (palmate, crenate).

Bottom row: the four F2 types (1) palmate, crenate, (2) palmate, non-crenate,
(3) femleaf, crenate, (4) fern-leaf, non-crenate. The four types occur in the
proportions 9:3:3:1.
Fig. 62. Showing the Fi's from crosses of the white-eyed race "Queen Alexandra " with
the large-eyed " Primrose Queen," and with " Snowdrift " (ordinary eye).

Top row: "Primrose Queen," No. 37/9; "Queen Alexandra," No. 34/9;
" Snowdrift," No. 1/9.

Second row: 36/9 Fi from ("Primrose Queen " x " Queen Alexandra").
35/9 Fi from ("Queen Alexandra " x " Snowdrift ").
Fig. 63. Showing the Fi& from crosses of the large yellow eye, stellata, with the ordinary
eye in the stellata and typical sinensis forms.

Top row: Stellata, white, ordinary eye. No. 55/9; " Primrose Queen," No. 37/9 ;
"Snowdrift" 1/9.

Second row: 48/9 Fi from {Stellata, ordinary eye x " Primrose Queen").
38/9 Fi from ("Primrose Queen " x «{ne7Mi«, ordinary eye).
Fig. 64. Showing the variation in the form of the corolla in a plant No. 54/9 and its
offspring.

Top row : flowers from the same plant, No. 54/9.

Second and third rows : Flowers from four plants illustrating the various types
produced by the self-fertilization of 54/9. The flowers from each plant are arranged
in pairs, one above the other. The first three represent the nearest approach to
sinensis, the intermediate and the stellata types in the Giant form. The last pair of
flowers are those of a stellata plant which did not possess the Giant character of its
parent.

JOURNAL OF GENETICS, VOL I. NO. 2

PLATE XXX

JOURNAL OF GENETICS, VOL I. NO. 2

PUTE XXXI

JOURNAL OF GENETICS, VOL I. NO. 2.

Fig. 60

Fig. 61

PUTE XXXII

Fig. 62

Fig. 63

E'^ ■

F#~

^r.*^H

r#j

^f # J

^H '^^^1

B ^£ARes^^^^^|

^H //vre «.ME o/Are

^^^1 SreTZArA^^I

^H Smaz.!. 6rELL*^^^H

Fig. 64

ON THE FORMATION OF ANTHOCYANIN.

By M. WHELDALE,

Fellow of Newnham College, Cambridge.

Nature of Problem ajs'd Preliminary Statement of
Conclusions.

The problem to which I have attempted to give a solution in the
following pages may be briefly stated as follows : — what are the
chemical processes which underlie the formation of anthocyanin ?

In my attempt to arrive at some solution I have used as data the
results of observations upon the general distribution of pigment, its
formation in relation to other metabolic processes and to the chemical
constituents of the tissue : also the conditions, both natural and
artificial, which lead to its appearance, and lastly, the detection of
enzymes which may be involved in its production.

As an outcome of this general investigation, I venture to bring
forward an hypothesis which may afford in many respects an explana-
tion, in terms of chemical reactions, of the mechanism underlying the
phenomenon of soluble pigment formation. At the same time I may
say that I look upon my suggestions as tentative and as having value
possibly only in so far as they combine together into a general scheme
a number of more or less isolated facts. I fully realise that the under-
lying causes of such a phenomenon are of a complex nature and may in
reality demand a very different explanation from that which I am able
to offer.

The ultimate object of the enquiry is the identification of the
Mendelian factors for colour. There is little doubt that the formation
of anthocyanin does involve a series of progressive reactions each of
which is controlled by a certain enzyme. In variation, whatever that
may be, the loss of these enzymes gives rise to different colour varie-
ties. Hence the greater the complexity of the pigment-forming process

134 On the Formation of Anthocyanin

in any species, the greater the number of derivative varieties we may
expect to appear under cultivation. Only an exact knowledge of the
chemical reactions involved in the formation of pigment will enable us
to explain the mechanism of colour inheritance and the cause of differ-
ences between varieties.

The main conclusions arrived at in the present paper may be
summarised as follows : —

(1) The soluble pigments of flowering plants, collectively termed
anthocyanin, are oxidation products of colourless chromogens of an
aromatic nature which are present in the living tissues in combination
with sugar as glucosides.

(2) The process of formation of the glucoside from chromogen and
sugar is of the nature of a reversible enzyme action :

Chromogen + sugar '*~~^ glucoside + water.

(3) The chromogen can only be oxidised to anthocyanin after
liberation from the glucoside and the process of oxidation is carried out
by one or more oxidising enzymes :

Chromogen + oxygen = anthocyanin.

(4) From (2) and (3) we may deduce that the amount of free
chromogen, and hence the quantity of pigment formed at any time in
a tissue, is inversely proportional to the concentration of sugar and
directly proportional to the concentration of glucoside in that tissue.

(5) The local formation of anthocyanin which is characteristic of
the normal plant is due to local variation in concentration of either
the free sugars or the glucosides in the tissues in which the pigment
appears. The abnormal formation of pigment under altered conditions
is due to differences in the concentration of these same substances due
to changes in metabolism brought about by these conditions.

(6) On the above hypothesis the formation of anthocyanin is
brought into line with that of other pigments produced after the
death of the plant, as, for example, indigo, the respiration pigments of
Palladin, etc.

Results obtained by Previous Investigators.

Although the soluble pigments of plants have afforded material for
a considerable amount of investigation, the sum total of results gives
us very little knowledge either of the composition of these substances
or of the processes which underlie their formation.

M. Whbldale 135

With regard to their chemical nature, I have previou8ly(19) given
reasons for the statement that the red, purple and blue pigments,
collectively termed anthocyanin, are oxidation products of chromogens
of an aromatic nature'. That, moreover, these chromogens, in the form
of glucosides, are present in solution in the cell-sap throughout the
living tissues and in the unoxidised state cause no colouration, but
under certain conditions through the agency of an oxidase the chromo-
gens may be oxidised to pigments, i.e. anthocyanin. This point of view
is in agieement with that held by other investigators: Wigand(21),
Pick(16), Mirande(9), Laborde(7), Overton(13) and Palladin(14), who
have considered the soluble pigments either to be themselves aromatic
compounds or to be intimately connected with tannins and allied
substances.

That oxygen is necessary for pigment formation and that the
oxidation is probably brought about through the agency of an oxidase
has been suggested by Mirande(9), Palladin(14), and by Buscalioni and
Pollacci(l).

Katic(5) and Molliard(ll) have also shown experimentally that
oxygen plays an important part in the appearance of pigment in
certain organs.

So far, however, no hypothesis has been formulated as to the more
exact mechanism of pigment formation, the reasons for its appearance
only under certain conditions and for its localisation in definite organs
and parts of the plant.

Wigand(21), it is true, has pointed out that the occurrence of antho-
cyanin in autumnal leaves, evergreen leaves in winter, injured or dying
leaves, flowers and ripening fruits, indicates a connection between
lessened assimilative activity and the production of pigment, though
the nature of the connection remains unexplained.

Overton (13), on the other hand, basing his conclusions on results
obtained from feeding leaves and plants with sugar solution, maintains
that increase of sugar in the cell gives rise to formation of anthocyanin.
He considers the pigment itself to be a glucoside of which the non-
sugar part is a tannin-like compound.

Again no indication is given as to the exact nature of the con-
nection between the excess of sugar and the appearance of pigment.

Kati6(5), Molliard(ll), Mirande(9) and Palladin(14) also support
the statement that an accumulation of sugar increases the formation of

1 In many cases, probably, members of the flavone and xanthone classes of com-
pounds.

136 On the Formation of Anthocyanin

pigment. In addition Kati6 has shown experimentally that the pro-
duction of pigment, though dependent upon the presence of oxygen, is
independent of the presence of carbon dioxide.

Some important additions to our knowledge concerning the forma-
tion of anthocyanin have recently been published by Combes(3). This
author has shown that the reddening of leaves is accompanied by an
accumulation of oxygen in the tissues and that the disappearance of
pigment on the other hand is accompanied by a loss of oxygen. In
addition he has shown that red leaves contain proportionally greater
amounts of glucosides and sugars than green leaves of the same plant.
Combes considers the cause of oxidation to be this accumulation of
glucosides and sugars which may arise from various external causes.
These substances accelerate the processes of oxidation and hence the
gaseous exchange may be fundamentally modified.

Statement of Hypothesis.

From evidence which will be given in detail later I have been led
to conclude that the formation of anthocyanin from a chromogen
depends upon two processes, in which two different enzymes at least
are involved.

The first reaction is reversible and may be expressed as follows : —
Glucoside + water ■^*i chromogen + sugar.

The same enzyme may be supposed to accelerate both the synthetic
and hydrolytic reactions.

The second reaction is one of oxidation and is carried out by an
oxidising enzyme or oxidase : —

Chromogen + oxygen = anthocyanin.

It must be further assumed that the chromogen can only be
oxidised after liberation from the glucoside.

On the basis of this hypothesis, it follows that the amount of antho-
cyanin in any tissue depends upon the amount of free chromogen, and
the latter, in accordance with the reveraibility of the first reaction, is
directly proportional to the concentration of glucoside and inversely
proportional to the concentration of sugar in the tissue.

The frequent appearance of pigment, under abnormal conditions, in
tissues which are normally unpigmented, justifies the supposition that
every part of an anthocyanic plant is provided with this mechanism for
the formation of pigment.

M. Wheldale 137

The local appearance of pigment in various plant tissues thus
resolves itself into a matter of local variations in the concentration of
sugars and glucosides in the tissues.

In order to test the validity of the hypothesis as stated above, it
must be ascertained whether the conditions which give rise to forma-
tion of pigment are such as would influence the amount of glucosides
and sugars present, and in this way it should be possible to demonstrate
a connection, if it exists, between the two phenomena.

I have classified my evidence from various sources under the
following headings : —

(1) Analogous reactions.

(2) Distribution of anthocyanin.

(3) Concentration of sugars and glucosides in various tissues.

(4) Existence of enzymes.

(5) Sugar-feeding.

Evidence from Analogous Reactions^

Many of the reactions involved in plant metabolism are known to
be of a reversible or balanced nature. Excess of sugar, for instance,
may be converted into starch and thereby stored up in an insoluble
form which is again hydrolysed into sugar when required. Similarly
oils may be hydrolysed into fatty acids and glycerine, and these products
again synthesised into oils. Synthesis and hydrolysis are also con-
tinually taking place between the disaccharides and the monosaccha-
rides. Cane-sugar is synthesised from glucose and fructose and
hydrolysed into the same products: dextrose is synthesised into maltose
which is hydrolysed into dextrose and so forth.

As a typical reversible reaction we may quote the hydrolysis of
ethyl acetate. When ethyl acetate is treated with water, hydrolysis
into acetic acid and ethyl alcohol commences at once, but as soon as
any products of hydrolysis are formed, the reverse action is also set up,
namely the synthesis of ethyl acetate from acetic acid and ethyl
alcohol. Hence in any mixture of the four substances, ethyl acetate,
acetic acid, ethyl alcohol and water, two opposite reactions will proceed
at diflferent rates : —

Ethyl acetate + water -^ ethyl alcohol + acetic acid.

1 In connection with enzyme action I have freely qnoted from Bayliss, The Natwre
of Enzyme Action, 1908.

Jonm. of Gen. i 10

138 On the Formatio7i of Arithocyanin

After a fime a certain relative concentration of the four constituents
results and at this stage the velocities of the two reactions are equal
and equilibrium is established.

If to a system of this kind, a catalyst, such as hydrochloric acid, is
added, the equilibrium position has been shown to remain unaltered.
From this it may be inferred that both the hydrolytic and synthetic
reactions are equally accelerated by the catalyst.

In plants the greater number of reversible reactions are of a hydro-
lytic nature and are controlled by special catalysts, known as enzymes,
produced by the living organistn.

It is now known that a considerable number of these enzymes, as,
for example, invertase, maltase, lipase, diastase and emulsin, can be
extracted from the living tissues and their activities can be demon-
strated under artificial conditions outside the plant. It is then found
that in many cases the velocity of the hydrolytic reaction is so much
greater than that of the synthetic that the equilibrium position is very
near complete hydrolysis. When such is the case, we may infer that
there is some, though very little, reversibility of the reaction. Hence
if enzymes behave in the same way as inorganic catalysts, it should be
possible to show that they are able to again synthesise the products
they produce in hydrolysis if the right conditions can be found.

Croft Hill was the first observer to give experimental proof that
enzymes accelerate synthetic processes, though in the special case
investigated by him the synthesised product was an isomeric form of the
compound hydrolysed. From a concentrated solution of glucose he
obtained, through the action of maltase, small quantities of isomaltose
which was again hydrolysed in dilute solution.

Since then many other cases have been discovered, such as the
synthesis of the ester, ethyl butyrate, by lipase from a mixture of ethyl
alcohol and butyric acid, of the glucoside, salicin, from saligenin and
glucose, and of cane-sugar by invertase from glucose and fructose.

The value to the plant of even a slight reversible action has been
pointed out by Bayliss(lA), for if the synthesised product is removed
from the sphere of action as rapidly as it is formed, either owing to its
insolubility or by translocation, a considerable amount of synthesis may
eventually take place.

There is also in many enzyme actions a special retarding influence
exerted by the respective products of action in addition to that due to
reversibility of the reaction. Usually the retarding effect exerted by
one of the products of action is greater than that exerted by the other ;

M. Wheldale 139

or again one may have a retarding influence and the other none, as for
instance in the case of invertase, which is retarded by fructose, whereas
glucose has no eflfect.

There is little doubt that the retarding influence is due to the fact
that the enzyme enters into some form of compound with the sugar
and is thereby removed from the sphere of activity, with the resultant
slowing down of the hydrolytic process.

There is a similar retardation in many cases in the synthetic process
due to the combination of enzyme with the substrate. A full account
of these retardation processes is given by Bayliss in The Nature of
Enzyme Action.

The enzymes with which we are chiefly concerned in the present
paper comprise the glucoside-splitting class. The term glucoside is
applied to a large number of substances occurring in plants, all of
which have the property of being hydrolysed by enzymes or by acids
into glucose and one or more other products such as alcohols, aldehydes,
phenols, etc.^

In some cases a glucoside, as, for example, xanthorhamnin, is only
hydrolysed, as far as we know, by one particular enzyme, rhamnase,
though sometimes one enzyme, such as emulsin, will hydrolyse a con-
siderable number of diflferent glucosides, i.e. arbutin, salicin, coniferin,
syringin, helicin, amygdalin, aesculin, daphnin, and others.

An interesting point in connection with the glucoside-splitting class
of enzymes is the rapidity with which the hydrolytic reactions take
place when the plant is submitted to the action of chloroform vapour
or other anaesthetics. Injury to the tissues will also bring about the
same result. This reaction is readily detected if the products of
hydrolysis have a characteristic odour, as for instance in the case of
cyanogenetic glucosides, of which amygdalin is the best known example.
Amygdalin occurs in bitter almonds and in the kernels of peaches,
apricots, plums and other fruits of the Rosaceae. By emulsin it is
hydrolysed according to the equation: —

C«H„0„N -f- 2H,0 = aH«0 + HCN + 2C5H„0.

amygdalin benzal- hydrogen glacose

dehyde cyanide

and the progress of the reaction can be detected by means of the
characteristic odour of the products.

^ An sccoont of these substances is given by £. F. Armstrong in The Simple Carbc-
hydrates and Glucotides.

10—2

140 On the Formation of Anthocyanin

The mustard oil glucosides, sinigrin and sinalbin, occurring in
mustard and other Cruciferae, also give products with a characteristic
odour of mustard oil on hydrolysis : —

CioHieO^NS^K + H,0 = CsH.CNS + C«H,A + KHSO4

sinigrin allyl sulpho-cyanide

The hydrolysis of glucosides on autolysis in chloroform or through
injury, can also be detected when the non-sugar component of the
glucoside molecule is an aromatic substance, which when free from
glucose is subsequently oxidised to a coloured compound through the
agency of an oxidising enzyme (oxidase) ; in this case the development
of colour indicates the progress of the reaction.

In some genera the pigments produced in this way after death or
injury to the plant are red, purple or blue, and hence attention has
been drawn to the phenomenon, and the products so formed have been
used commercially for dyeing purposes. The best known examples are
the "indigo plants," Indigofera spp., Isatis tinctoria, Polygonum,
tinctorium, etc.

The processes taking place in the formation of indigo have been
very fully investigated and can be expressed as follows : —

C^H^OeN + H,0 = CeHj^Oe + CsH.ON

indican glucose indoxyl

2C8H,ON + 03 = m.,0 + QsHioO.N,

indigotin

The first reaction is brought about by a glucoside-splitting enzyme,
indimulsin, which hydrolyses the glucoside, indican ; the second by an
oxidase which oxidises the colourless indoxyl to the pigment indigotin
or indigo.

Another striking example is the rapid formation of a bright
red pigment in the flowers and leaves of Schenkia blumenaviana on
autolysis in chloroform as described by Molisch(lO).

Though the formation of a brightly coloured pigment after death
is a comparatively rare phenomenon, yet many plants rapidly turn
brown or reddish-brown when placed in chloroform vapour. The
same effect is produced by mechanical injury and sometimes by
immersion in absolute alcohol (Mirabilis Jalapa). Extracts from such
plants give a blue colour with guaiacum tincture and there is little
doubt that the production of pigment is due to the oxidation of an
aromatic substance through the activity of the oxidase.

M. Wheldale 141

This appearance of pigment on autolysis is especially frequent
among genera of the Labiatae, Boraginaceae, Scrophulariaceae and
Umbelliferae, though it is also generally characteristic of some of the
genera in most Natural Orders. Often, as in the Ranunculaceae, this
phenomenon is peculiar to all species of certain genera {Anemone,
Helleborus, Clematis, Trollius, CdLtha and Aconitum), which rapidly
yield brown pigment on autolysis in chloroform, whereas all species of
other genera {Ranunculus, Paeonia, Aquilegia and Thalictrum) give no
colour in the same time of exposure.

It is probable that the processes involved in the formation of post-
mortem pigments are in all cases analogous to those known to take
place in the production of indigo. The aromatic chromogen, from
which the pigment is produced, is combined with sugar in the form of
a glucoside in the living cell. In such a form the chromogen cannot be
attacked by the oxidase. When the cell is subjected to chloroform
vapour, the velocity of the hydrolytic reaction is accelerated and the
whole of the glucoside is split into chromogen and sugar. The free
chromogen is then fully oxidised by the oxidase.

According to the view held by Palladin(14), these aromatic glucosides,
together with the glucoside-splitting and oxidising enzymes, form an
important system in the plant for the purpose of oxidising respirable
materials, and the post-mortem pigments have been termed by him
" respiration pigments." In the living unpigmented cell, the processes
of oxidation, reduction and the glucoside splitting reaction are so
balanced that no free pigment appears. To quote Palladin(lo): —
" Einer sparsamen Hausfrau vergleichbar, halt die Zelle die Chro-
mogene verschlossen und verausgiebt sie in geringen Mengen fiir
Oxydationsprozesse. Die Ausgabe wird durch ein die Prochromogene
spaltendes Enzyme besorgt."

To the chromogen in combination with sugar as a glucoside,
Palladin has applied the term " prochromogen." He also includes
anthocyanin among the respiration pigments, but does not ofifer any
very definite explanation of the appearance of anthocyanin in various
plant organs.

I am inclined to believe that anthocyanin itself has no direct
respiratory function in that it is absent from albino varieties, which do
not appear to suflfer from the loss of pigment, since they grow and
reproduce quite as vigorously as the pigmented types.

From the description of enzyme actions given above it will be seen
that a series of reactions such as I have supposed to take place in the

142 On the Formation of Anthocyanin

formation of anthocyanin is known to occur in connection with the
oxidation of aromatic compounds in the plant. The essential difference
between such reactions as lead to the formation of indigo, and those
which have been suggested for anthocyanin, lies in the nature of the
oxidase. In the former case the oxidase continues its function after
the death of the cell, but so far there is no evidence of anthocyanin
being formed in extracts from the plant', and it seems highly probable
that it is a process which is difficult or perhaps impossible to induce
under artificial conditions. The nature of the oxidases concerned will
be discussed later.

Distribution of Pigment.

The various organs of the plant in which anthocyanin may appear
can be enumerated as follows :

Under normal conditions.

(1) Veins, midribs and petioles of many leaves. Herbaceous stems
and the young stems of shrubs and trees.

(2) Leaves of red-leaved species (Ainaranthtis, Goleus, etc.).

(3) Leaves of red-leaved varieties of green-leaved types {Fagus,
Corylus, Beta, Atriplex, etc.).

(4) Young developing leaves (Quercus, Rosa, Crataegus, etc.).

(5) The older leaves of many plants {Fragaria, Aquilegia, etc.), and
sometimes the whole plant (many Umbelliferae, Galium aparine, etc.)
towards the end of the vegetative season.

(6) Variegated leaves in which the chlorophyll is absent from
certain areas.

(7) Flowers and ripe fruits.

Under abnormal conditions.

(1) Leaves which have been injured either mechanically or through
the attacks of insects and fungi.

(2) Autumnal leaves.

(3) Leaves exposed to low temperatures, such as evergreen leaves
in winter (Hedera Helix, Ligustrum vulgare, Mahonia sp., etc.).

(4) Leaves exposed to drought.

^ Except in so far as I have been able to induce a formation of colour in an extract from
Primula flowers by means of Horseradish peroxidase in presence of hydrogen peroxide.
Proc. Gamb. Phil. Soc. Vol. xv. 1909.

M. Wheldalk 143

Leaves. The majority of leaves during the active vegetative period
are entirely without soluble pigment so far as the eye can detect.
Nevertheless it is possible that the leaves of anthocyanic plants may
contain a small amount of pigment though it is not apparent as such.
The leaves of albinos, for instance, are frequently of a brighter and
lighter shade of green than leaves of the pigmented type even when
the latter are without obvious pigment, and this difiference can often be
detected before the plant flowers. The deeper colour may, however, be
due to some other cause.

When pigment is present in the veins and midrib, as is normally
the case in many leaves, it is usually confined to the epidermal (gene-
rally upper) and sub-epidermal layers.

In leaves which are more or less permanently red (Amaranthus spp.),
the pigment is commonly present in the epidermis only, both upper and
lower, all over the leaf, but in the midrib and veins it may appear in
the sub-epidermal layers also.

In red -leaved varieties {Atriplex hortensis v. rubra. Beta vulgaris
V. rubra, etc.) arising from a green-leaved type, the pigment is again
usually only present in the epidermis, both upper and under, of which
the cells are intensely coloured.

It is an interesting fact that the guard-cells of the stomata in the
epidermis of permanently red-leaved plants and red-leaved varieties
are colourless when all the surrounding epidermal cells are intensely
coloured.

The cases of abnormal formation of pigment in leaves may now
be considered. If a leaf is subjected to any kind of injury, this is
accompanied in many plants by a more or less intense colouration of the
tissues. The injury may be a mechanical one, such as tearing of the
lamina, partial breaking of the midrib, petiole or stem, or the removal
of a portion of the midrib. In each case the leaf becomes pigmented
in the part distal to the point of injury. Sometimes the whole leaf
when severed from the plant and lying in a fairly moist situation will
turn red or purple. Injury may also be brought about by attacks of
insects, leaf-boring larvae, aphides and fungi. In all such cases pig-
mentation results.

Sections of leaves which have been injured show a different distri-
bution of pigment from those which are normally coloured. Antho-
cyanin is most frequently present in the palisade parenchyma, often in
the spongy parenchyma, and more rarely in the epidermis and veins,
unless these were originally coloured in the normal leaf.

144 On the Formation of Anthocyanin

Hence we may state that in injured leaves the formation of pigment
commences in tissues which in the healthy plant are usually unpig-
mented.

The same internal distribution of pigment is found in leaves red-
dened by low temperature, i.e. autumnal leaves and evergreen leaves in
winter, and also in the older dying leaves of plants at the end of their
vegetative season or after exposure to drought.

It is an interesting coincidence that the phenomenon of increased
pigmentation accompanying age is also characteristic of young develop-
ing leaves. In these again the pigment is formed in the assimilating
tissue, chiefly palisade parenchyma, though it may also appear in the
epidermis.

Anthocyanin is very frequent in variegated leaves and it is then
often limited to the stripes or patches free from chlorophyll (variegated
Zea Mais). In other cases (Codiaeum sp., Acalypha sp., Tradescantia
sp.), the whole leaf may be pigmented.

Stems and Petioles. The distribution of pigment in petioles, herba-
ceous stems and the young stems of trees and shrubs is very much the
same as in the midribs of leaves. Anthocyanin is usually confined
either to the epidermis alone or to one or more sub-epidermal layers in
addition, of which the cells are frequently collenchymatous in structure.

Flowers and Fruit. In the corolla, anthocyanin is located in the
epidermis, usually both upper and under, sometimes only upper. The
upper pigmented epidermal cells are almost always more or less pro-
longed into papillae but this prolongation is less characteristic of the
under epidermal cells.

In fruits the colouring matter may be limited to the epidermis and
sub-epidermal layers or may extend into the inner tissues.

Concentration of Sugars and Glucosides in various Tissues.

To ascertain the relative concentrations of sugars and glucosides in
the different tissues of a leaf is a difficult problem.

The presence or absence of glucose can be detected micro-chemically
by means of Fehling's solution (22), and to some extent glucose, fructose,
maltose and cane-sugar can be differentiated micro-chemically by a
modification, employed by Grafe(4), of Senft's(17) phenyl hydrazine
method. For detection of differences in amount I have not found these
methods reliable.

M. Wheldalb 145

Since no absolute reliance can be placed on the above tests, it is
only possible to draw deductions indirectly from such evidence as we
possess from other sources.

Broadly speaking the concentration of sugars in a leaf depends
upon :

(1) The assimilative activity.

(2) The starch-forming activity.

(3) The rate of translocation of sugars.

Since these three factors are more or less interdependent and form
in co-operation a self-regulating mechanism, the concentration of sugar
as the outcome of their combined activities may under normal con-
ditions remain fairly constant. But if a tissue has assimilative without
starch -forming power or vice versa, we have perhaps some basis for
conjecture as to the concentration of its sugar-contents compared to
that of other tissues possessing both activities. There is a like possi-
bility if the different activities are affected in varying degrees by
changed conditions, and this question will be considered again in
connection with abnormal reddening in leaves.

The question of the concentration of aromatic glucosides in a leaf
is even more problematic. Kraus(6) has given experimental evidence
for regarding the assimilating leaf as the seat of metabolism of aromatic
substances. This author found, as a result of analysis, that aromatic
compounds^ accumulate in a cut leaf exposed to illumination but
decrease in a leaf kept in darkness. He moreover states the amount
of aromatic substances formed to be proportional to the assimilative
activity of the leaf.

Palladin(14) also holds the view that the aromatic materials of a
plant are manufactured from the carbohydrate series. In corroboration
of his view, he quotes the results of Waage(18), who obtained an increased
amount of phloroglucin in leaves fed on glucose solution ; also those of
Biisgen(2), who found that the tannin contents of plants increase in
glucose cultures.

On such evidence as we have, we may say that the concentration of
aromatic substances in a leaf depends upon : —

(1) The amount of sugars present in the leaf.

(2) The rate of translocation of aromatic substances.

^ In this case, tannins, bat the precise natore of the products is immaterial provided
they belong to the aromatic series.

146 On the Formation of Anthocyanin

In the formation of anthocyauin the following reactions must be
taken into consideration : —

Aromatic glucoside + water "^"^ aromatic chromogen + sugar.

Sugar — »> aromatic chromogen.

Aromatic chromogen + oxygen = anthocyanin.

The following possibilities may therefore arise. The amount of
pigment is directly proportional to the amount of free chromogen.
Increase of sugar would naturally lead to decrease of free chromogen,
but if at the same time additional chromogen is formed from the sugar,
the ultimate concentration of the glucoside, if it is not removed by trans-
location, will be increased to such a degree that the final result is an
increase of free chromogen accompanied by formation of pigment.

A decrease of sugar, on the other hand, will increase the free
chromogen, but at the same time it will lead to a decrease in the
concentration of the glucoside, so that the final result is a decreased
amount of free chromogen and less possibility of pigment formation.

Or to state the case rather differently : so long as the concentration
of glucoside remains low either as a result of translocation or of decreased
formation, the amount of free chromogen is negligible, but if the
concentration of glucoside is raised beyond a certain point as a result
of diminished translocation or continual formation, the synthesis of free
chromogen and sugar can no longer take place and the former becomes
oxidised to anthocyanin.

In the normal green leaf the absence of pigment from the mesophyll
is in all probability due to the rapid translocation of aromatic gluco-
sides away from the leaf. It is difficult to ascertain the precise reason
for the presence of pigment when it appears in the epidermis of the
lamina and in the epidermis and sub-epidermal layers of the veins and
petiole. It may be caused either by low concentration of sugar or by
increased concentration of glucosides due indirectly to excess of sugar.
These tissues are without chlorophyll and the power to assimilate, but
at the same time they are also apparently devoid of starch-forming
capacity, since starch does not as a rule appear in them, so that the
sugar concentration may or may not be greater than in the mesophyll
of the leaf.

In general the chlorophyll-containing tissues are most free from
pigment, the non-chlorophyllous more frequently pigmented. Hence
the appearance of pigment is undoubtedly connected with the concen-
tration of sugar, but I am at present unable to give the exact sequence
of events which affects the reversibility of the reaction.

M. Wheldale 147

That a relationship exists between pigmentation and assimilation is
further borne out by the appearance of anthocyanin in old leaves,
variegated leaves (with parts free from chlorophyll), autumnal leaves,
leaves exposed to drought or low temperature and in flowers and
ripening fruits. In all these cases the same difficulty arises as to the
real cause, since the starch-forming power may be diminished as well
as the assimilative. Starch does not as a rule appear in petals ; and in
fruits the colouring matter is often limited to the epidermis and sub-
epidermal layers which are free from starch though the flesh of the
fruit may be full of starch. In variegated leaves the chlorotic portions,
in which pigment often appears, are unable to form starch. I have
made a number of observations upon the starch contents of green leaves
and of leaves, from the same plant, reddened as a result of cold,
drought, etc., and I have found the red leaves almost invariably to
contain less starch than the green.

It must also be borne in mind that the translocation of both sugar
and glucosides may be hindered by low temperature, drought, age, etc.
I am inclined to believe, in the absence of more direct evidence, that
the reddening under these conditions is due to diminished translocation
of glucosides combined with increased formation of these substances
due to the presence simultaneously of excess of siigar.

Results lately published by Combes (3) corroborate this view to some
extent. Combes has made comparative estimations of the glucosides
and sugars in both red and green leaves of Ampelopsis hederacea in
which reddening was due to light intensity, and in Rosa canina,
Mahonia aquifolium and Sorbus latifolia showing autumnal colouration.

His results may be expressed as follows : —

Sugars

Dextrins

Ulucoeides

Inaolnble
CarbohTdntes

Ampelopsis hederacea

green

•74

2-78

2-43

2-42

red

•98

1^88

2^79

502

Bo$a canina

green

2-42

1-30

8^22

972

red

264

1-23

8-24

533

Sorbm latifolia

green

•71

1^15

220

1199

red

•80

1-07

2^52

120

Mahonia aquifolium

green

•57

•80

341

2-38

red

1-30

•60

430

8^78

From these numbers we see that the concentration of glucosides
and sugars in red leaves is greater than in green, that of dextrins
greater in green than red, whereas the amount of insoluble carbo-

148 071 the Formation of Anthocyanin

hydrates varies, being sometimes greater in one, sometimes in the
other.

Since the concentration of both glucosides and sugars is greater in
the red leaves, it is reasonable to interpret the pigment formation as
being due to accumulation of glucosides, in which case the reaction

Glucoside + vs^ater — *- sugar + chromogen

would give rise to more free sugar (apart from other causes) in the red
than in the green leaf from which the glucosides are continually
removed, so that the concentration of glucoside is lowered

Sugar + chromogen — *- glucoside + water.

Kraus(6) has also shown that red autumnal leaves contain more
aromatic substances than normal red leaves.

Results, however, which are more convincing than these just stated,
are those connected with the phenomenon of reddening produced by
injury. Instances have already been quoted of pigment formation due
to injury to the cortical tissues of the midrib and petiole or to the
removal of a portion of the midrib or main veins of a leaf. According
to Kraus(6) the path taken by aromatic substances in translocation is
the vascular system of the leaf, but whether by the phloem or the
surrounding parenchyma is not stated. In any case injury to the
vascular system of the leaf or the living tissues of the petiole or stem
would involve an accumulation of glucosides in the parts distal to the
point of injury. It has been recently suggested by Mangham(8) that
the sugars travel from the leaf by the phloem. If such is the case, the
injury may also lead to accumulation of sugars and hence indirectly to
more glucoside.

Combes(3) has shown that decortication in spp. of Spiraea induces
reddening of the leaves above the point of operation. A similar obser-
vation has been made by Kraus for Cornus alba, and I have myself
observed a similar result following upon decortication in Ribes Grossu-
laria and a species of Salix.

Combes (3) has shown by analysis that there is a large increase of
both glucosides and sugars in the leaves of Spiraea which had reddened
as a result of decortication.

The following are the numbers given : —

Insoluble
Sugars Dextrins Glucosides Carbohydrates

Spiraea paniculata green 2-21 1*01 1-64 10-75

red 4-26 '92 6-15 26-58

M. Wheldalb 149

Kraus(6) also found that some cut leaves redden when placed in
water in bright sunshine, and on analysis gave greater quantities of
aromatic substances than control leaves kept in the dark.

Evidence for the Presence of Enzymes.

If the formation of anthocyanin is dependent upon enzyme action,
it should be possible to obtain evidence of the existence of both
glucoside-splitting enzymes and oxidases in the tissues of anthocyanin
plants.

Glucoside-splitting enzymes. For the detection of glucoside-splitting
enzymes I have employed the following method. The tissue to be
examined is well ground and thoroughly washed with 75°/^ alcohol: it
is then dried in air and extracted with distilled water. These processes
are carried out as far as possible under sterilised conditions. The
water extract is then added to a definite quantity of salicin solution
and kept, together with a control flask, at a temperature of from
36° — 40° C. for 24 hours. The following reaction then takes place : —
Salicin + water = saligenin + glucose.

The saligenin is extracted from the liquid by shaking with ether
and after evaporation of the ether its presence can be detected in the
residue by means of ferric chloride with which it gives a violet
colouration.

By this method I have demonstrated the presence of a glucoside-
splitting enzyme in the following : — leaves of Gorylus Avellana, Rumex
crispus, Taraxacum officinale and Primula sinensis, flowers of Cytistis
scoparius, Aquilegia vulgaris, Viola tricolor, Antirrhinum majus.
Primula sinensis, Narcissus pseudonarcissvs, Cheiranthus cheiri, Fritil-
laria imperialis, Polyanthus sp., Helleborus orientalis, Pyrus japonica,
Prunus avium, Galanthus nivalis. Narcissus Tazetta, Pelargonium,
zonule, and tubers of Solanum tuberosum.

These results show that glucoside-splitting enzymes are widely
distributed. In other species a negative result was obtained but this
is to be expected, since all such enzymes may not be able to hydrolyse
salicin.

If glucose solution is added to the salicin solution plus the enzyme
the hydrolysis of the salicin is greatly retarded.

Also if the preliminary treatment with alcohol as described above is
omitted and a water extract is made from the fresh plant tissues and
added to salicin, very little or no hydrolysis of the latter takes place.

160 On the Formation of Anthocyanin

This retardation is doubtless due to the fact that the water extract
contains, in addition to the enzyme, the glucosides and sugars present
in the plant. Thus the products of hydrolysis of the glucosides derived
from the plant would retard or entirely prevent hydrolysis of the salicin
added. By treatment with alcohol, all glucosides and some part of the
sugars are removed previous to extraction with water.

Oxidases. It has been previously mentioned that Palladin(14)
considers anthocyanin to be a respiration pigment. That oxygen is
necessary for its production has been shown experimentally both by
Molliard(ll) and Katic(5).

The dependence of pigment formation on the presence of oxygen
can be readily demonstrated in a very simple way. If cut leaves of
Taraxacum officinale are placed in sugar solution so that the lamina is
partially immersed, reddening only takes place in the portion exposed
to air and not in the submerged part. The oxygen may also be
excluded by greasing part of the leaf with vaseline. The greased
portion remains green while the ungreased portion develops a con-
siderable quantity of pigment. Similar results have been obtained
with leaves of Heracleum sphondylium, Sambucus nigra and Hiera-
cium sp.

Apparently no reverse process of deoxidation takes place when a
coloured leaf is greased so as to prevent all gaseous exchange. If
anthocyanin constitutes a medium for the transference of oxygen, we
should expect the colour to disappear as a result of reduction when
coloured leaves are deprived of oxygen, especially since respiration is
one of the last " vital processes " to disappear. The strongest argument
against Palladin's hypothesis is the existence of well-developed albino
varieties of an almost innumerable number of species.

The question of the oxidising enzyme presents some difficulty. In
all plants forming post-mortem pigments, oxidases can be detected by
means of guaiacum tincture, with which the extracts give a strong and
rapid direct action. Yet blueing of guaiacum is not limited to these
cases, for a less rapid direct action is also given by other plants
(Lathyrus, Matthiola), which do not form pigments on autolysis. All
the guaiacum-blueing species I have examined have been anthocyanic,
and the possibility suggests itself that the oxidases may form antho-
cyanin in the living plant but a post-mortem pigment after death.
There is some evidence in favour of this supposition : first, when a
plant forms anthocyanin and also a post-mortem pigment, the former is
converted into the latter on autolysis and the organs which contain

M. WUELDALE 161

most anthocyanin produce the greatest quantity of brown pigment.
Secondly, when fully pigmented flowers of the type in any species
(cultivated spp. of Chrysanthemum, Petunia and Pyrethrum) give a
strong oxidase reaction, tinged or less intensely coloured varieties
frequently give a less strong reaction, which may indicate that some
part of the oxidising mechanism has been lost from the varieties, as
I have previously suggested (19) for Lathynis and Matthiola.

On the other hand, very many anthocyanic plants give no direct
action with guaiacum, although nearly all living tissues give an indirect
action (i.e. after addition of hydrogen peroxide). It is possible that the
direct action is inhibited in these cases by some strong reducing sub-
stance in the plant. It is also more than probable that anthocyanin
oxidases are of a nature totally different from those connected with
respiration pigments and may, many of them, not react with guaiacum.
For the present no other explanation appears available.

Some work on oxidising enzymes has been recently published by
Moore and Whitley (12). These authors do not support the hypothesis
of Bach and Chodat, i.e. th^t an oxidase consists of a mixture of two
enzymes, an oxygenase which acts upon certain substances in the plant
forming peroxides and a peroxidase which transfers the additional
oxygen atom from the peroxide to other bodies.

When both enzymes are present, the plant extracts have a direct
blueing action on guaiacum, but when the peroxidase exists alone,
hydrogen peroxide must be added before blueing results (indirect
action).

Moore and Whitley suggest that only one enzyme — peroxidase — is
involved in the process and that the blueing, to a greater or less degree,
of guaiacum by plant extracts, is due to the existence of more or less
organic peroxide in the tissues and that no special enzyme involved in
the formation of peroxide can be detected.

This point of view greatly simplifies the conception of oxidation
processes. I am nevertheless of the opinion that peroxide formation in
the plant may be controlled by an enzyme, though it may not be pos-
sible to extract this substance and to obtain an expression of its activities
under artificial conditions.

Since, moreover, the presence of organic peroxides in plants is
directly connected with the appearance of post-mortem pigments, it
must follow that the metabolism of this class of plants differs in some
fundamental respect from that of others; and in my opinion such a
constitutional difference may quite well involve the existence of at
least one special enzyme.

152 On the Formation of Anthocyanin

Sugar-Feeding.
It is obvious that in the consideration of such an hypothesis as that
which I have formulated, any evidence of a connection between in-
creased pigmentation and increased concentration of sugars brought
about by artificial feeding of plants or parts of plants with various
sugars would be of considerable value.

Such a method of research has been adopted by Overton(13). This
author maintains that the conversion of sugar into starch is lessened
by a lowering of the temperature. Hence the pigment of autumnal
leaves and evergreen leaves in winter is due to excess of sugar induced
by low temperature.

In order to test his hypothesis, Overton made a number of sugar-
feeding experiments with both water and land plants. The water
plants were grown either submerged or floating in solutions of cane-
sugar, glucose, fructose, etc. In the case of land plants, the cut ends of
leafy stems or the petioles of leaves were placed in the solutions.

Many of the species used {Hydrocharis morsus-ranae, Utricularia
spp., Lilium Martagon, Ilex aquifolium, Hedera Helix, Ligustrum
vulgare, Saxifraga spp., Crassula spp., Aquilegia vulgaris, Tm^axacum
vulgare, Eupatorium cannahinum and Epilohium parviflorum), showed
increased formation of pigment, but in other cases (Potamogeton spp.,
Lemna spp., Fritillaria imperialis, Mahonia aquifolium, Anthriscus
sylvestris, Rubus spp., white flowers of Pelargonium zonale, and Anemone
japonica) there was a negative result.

Increased colour sometimes appeared in control plants kept in
distilled water under good illumination.

Corroborative results have also been obtained by Kati6(5) with
plants of Hydrilla, Elodea canadensis, Hydrocharis morsus-ranae, leaves
of Sagittaria natans, Canna indica, Veronica chamaedrys, Rosa sp.,
Saxifraga cordifolia, Pittosporum undulatum and Bellis perennis.

Overton has proved his results to be due to the chemical nature of
the dissolved substance and not to any osmotic action by the use of
control solutions of neutral salts, i.e. sodium chloride, sodium sulphate,
potassium sulphate of osmotic strengths equivalent to those of the sugar
solutions employed. In no case where a neutral salt was used, was there
any increase in pigmentation. In Lilium Martagon, an increase of
pigment resulted from the use of ethyl and methyl alcohol solutions.

In view of Overton's suggestion that increased sugar concentration
may under some conditions be brought about by a decreased starch-
forming capacity, I thought it possible that some connection might
exist between the reddening of leaves and starch formation in sugar-

M. Wheldale

153

cultures. I therefore made a number of sugar-feeding experiments
with various species and at the same time I examined the starch
contents of the leaves after this treatment.

The leaves employed were first kept in the dark until starch-free,
and a piece of each leaf was placed, after this treatment, in methylated
spirit as a control. Portions of the same leaf were then floated, upper
surface downwards, in two dishes, one containing 3°/^ cane-sugar solu-
tion, the other distilled water. Both dishes were placed under a bell-jar
together with a dish containing strong caustic potash solution, air being
allowed to enter the bell-jar only by means of a tube containing soda-
lime. Control dishes of sugar solution and distilled water containing
further portions of the same leaf were placed under a bell-jar without
potash solution. After an interval of from 7-10 days, any development
of pigment was noted, and the leaf portions were then placed in methy-
lated spirit until colourless and sections, after treatment with strong
chloral hydrate solution and iodine, were examined for starch contents.

The results are tabulated as follows : —

Species

3 per cent

cane sugar

—carbon

dioxide

Distilled

water

— carbon

dioxide

3 per cent.

cane stigar
+carbon
dioxide

DistiUed

water
+carbon
dioxide

Development of

Pigment either
with or without
carbon dioxide

Viola tricolor

no starch

no Starch

no starch

CapseUa bursa
pastoris

abandant
starch

no starch

abandant
starch

Lactuca sativa

no starch

Reseda lutea

abandant
starch

no starch

abandant
starch

Matricaria sp.

abandant
starch

no starch

abandant
starch

Cheiranthws
cheiri

abundant
starch

no starch

abandant
starch

considerable

amount of

starch

—

Nieotiana
tabacum

abandant
starch

no starch

abandant
starch

Aquilegia vul-
garis

some starch
in places

no starch

Epilobium sp.

abandant
starch

no starch

abandant
starch

some starch

Ilex aqui/olium

very little
starch

no starch

very little
starch

no starch

Ligustrum
vulgare

no starch

very little
starch

no starch

Mahonia
aquifolium

very little
starch

no starch

crammed
starch

Rumex crispus

very little
starch

no starch

very little
starch

no starch

Rubus fnUicostu

crammed
starch

no starch

crammed
starch

no starch

Joam. of Gen

154 On the Formation of Anthocyanin

These results show that there is some connection between pro-
duction of pigment and the capacity for forming starch from the sugar
provided. As a rule, the leaves which turn red are those which form
least starch from the sugar solution and several, in fact, form very
little or no starch even under normal conditions.

Hence experiments on sugar-feeding further strengthen the view
that reddening is due to an increase in the concentration of sugar
which ultimately leads to an increase in concentration of glucosides ;
the latter, being formed from sugar, naturally accumulate in excess
since there can be no translocation from the severed leaf

Palladin(14) also maintains that the amount of aromatic chromogen
is increased by sugar-feeding. In his experiments equal portions of
leaves of Rumex patentia were placed in water and 20°/^ cane-sugar
respectively. After four days the pieces in cane-sugar had developed
anthocyanin, those in water none. The sugar-fed and the control portions
were then heated with water and equal amounts of the extracts treated
with horse-radish peroxidase and hydrogen peroxide. The extract from
the sugar-fed portions produced considerably more pigment than that
from the control portions. This view is quite in accordance with my
suggestion that sugar-feeding leads to increase of free aromatic chro-
mogen.

With Vicia Faba leaves in sugar-cultures Palladin obtained a
different result. In this case the extracts gave less not more pigment
with peroxidase and hydrogen peroxide, whereas cultivation in water
only increased the amount of free chromogen. As an explanation
Palladin suggests that the free chromogen combines with sugar to form
a glucoside — prochromogen — and as such cannot be oxidised by the
oxidase.

I should suppose the explanation to be as follows : — The chromogen
in Vicia Faba is of a different nature from that in most plants in that
it is oxidised by tyrosinase, and we may therefore suppose it to be a
tyrosin-like compound and not capable of being synthesised from sugar
alone. The increased concentration of sugar would only form a gluco-
side from the existing chromogen and thereby decrease the amount of
free chromogen and would not increase the total amount of glucoside.
Cultivation in water would tend if anything to decrease the amount of
sugar and hence the amount of free chromogen would increase. It may
be also added that sugar-culture does not produce colour in Vida
Faba leaves.

The question as to whether sugar-feeding does or does not directly

M. Wheldale 165

increase the concentration of aromatic glucosides is one which can only
be solved by quantitative estimation. I am at present engaged in
experiments in connection with this point.

Application to Mendelian Factors.

A question which now arises is how this hypothesis I have for-
mulated fits in with our knowledge of the relationship between colour-
varieties and the type from which they are derived.

In the first place I shall deal with a variation which, though com-
paratively rare, may be most closely connected with the reactions
controlled by the glucoside-splitting enzymes.

There are anthocyanic species which have given rise to varieties
having some organ or part fully pigmented with anthocyanin, whereas,
in the type, the same organ or part is unpigmented or only slightly so.
The following are examples : —

Flower. The type in Bellis pei'ennis, Cyclamen persicum, Primula
acaulis, P. elatior, Cheiranthus cheiri, Crataegus oxyacantha,
Achillea millefolium, is either without, or is only slightly
tinged with, anthocyanin, but fully coloured varieties are
known.

Leaf. Fagus sylvatica, Coryllus Avellana, Beta vulgaris, Atriplex
hortensis, Perilla nankinensis. Carina indica, Plantago major,
Brassica sp., Lactuca saliva, produce red-leaved varieties.

Fruit. The orange and banana have varieties in which the flesh
and pericarp respectively are pigmented with anthocyanin.

We may assume that the coloured varieties arise through the loss
of some factor from the type, and in some cases it has been shown that
the coloured variety is recessive to the type. If the petals and leaves
of the coloured varieties are examined microscopically, it is found that
the pigment is invariably limited to the epidermal cells, and it is
reasonable to suppose that the loss of the factor is also limited to the
epidermis. Previously (19) I have termed this unknown factor a reductase
or inhibitor, but if the views I have expounded in the present paper
are correct, the appearance of pigment in the epidermis might be
explained on the supposition that the enzyme controlling hydrolysis
and synthesis of the glucoside is absent from this tissue. Hence the
chromogen is free from sugar and can be oxidised. In the type the

11—2

156 On the Formation of Anthocj/anin

equilibrium position is such that very little or no free chromogen is
present in the tissues subject to the variation : in the variety the
equilibrium position is possibly one of complete hydrolysis and the
tissue as a result becomes considerably pigmented.

In the second place, there are anthocyanic species in which the
type has coloured flowers, and has given rise to a large number of
derivative varieties. Many of these have been fully described in Men-
delian literature, and several classes of varieties can be recognized
which are applicable to a number of different species.

The main classes can be distinguished as : —

I. The blue or purple anthocyanic class.

II. The red anthocyanic class.

III. The albino or non-anthocyanic class.

Both I. and II. may in many cases exist in sub-classes common to
both; i.e.

(a) The tinged class.

(6) The pale class.

There is no further evidence in the present paper beyond that
which I have previously given(20) as to the nature of the factors, the
absence of which causes loss of blueing power and albinism respectively.
They are in all probability oxidising enzymes, though I am by no
means unwilling to admit that blueing may in some cases, considering
the great number of possibilities in plant-metabolism, be due to
alkalinity of the cell-sap brought about by some definite enzyme
action. I am uncertain as to the nature of the factor, the absence of
which causes the tingeing.

I. (b) constitutes the type in many species and deeper varieties of
both red and blue classes, i.e. I. (c) and II. (c) are known. They are
recessive to the type and are due to the loss of some factor. It now
seems probable that this factor is not a partial inhibitor or limiting
factor of a reductase nature such as I have suggested, but a controlling
enzyme, i.e. one which synthesises and hydrolyses the glucoside. Whereas
loss of this enzyme may give rise to coloured varieties when the type is
merely tinged and quite unpigmented, when the type is already
coloured the loss deepens the colour by increasing the amount of
pigment formed.

Sometimes the loss is limited to the flower only — Lathyrus, Matthiola,
Althaea, Cheiranthus; in other cases, the intense pigmentation of the

M. Wheldale 157

flower is accompanied by increased pigmentation of the epidermis of
the leaves which in the type are un pigmented : example — deep-flowered
varieties of Antirrhinum majus, Dianthus barbatus.

It is diflficult to devise a method for demonstrating the absence of
an enzyme when the latter may be confined to the epidermis alone. It
is possible that some micro-chemical method may be found.

REFERENCES.

lA. Bayliss, The Nature of Enzyme Acti(yn, 1908.

1. BuscALioyi, L., and Pollacci, G. Le antocianine ed U loro sigmficato bto-

logico nelle piante, 1903.

2. BusGEX. Chem. Centralb., 1890 and 1894.

3. Combes, R. Du role de Poxyg^ne dans la formation et la destruction des pig-

ments rouges anthocyaniques chez les vegetaui. C. K Acad. d. Sciences,
mai, 1910. Sur le degagement simultan^ d'oxyg^ne et d'anhydride carbo-
nique au cours de la disparition des pigments anthocyaniques chez les
v^etaux. C. R. Acad. d. Sciences, juin, 1910. Les Changes gazeux des
feuilles. Rev. gen. de Bot. torn, xxn., 1910. Production d'anthocyane sous
I'influence de la decortication annulaire. Bv.U. Soc. bot. France, torn, n.,
1909. Recherches biochimiques sur le d^veloppement de I'anthocyane
chez les v^etaux. C. R. Acad. d. Sciences, 1909. Rapports entre les
composes hydrocarbon^ et la formation de I'anthocyane. Ann. d. Sciences
nat. ^ s^rie, 1909.

4. Grate, V. Studien iiber den mikrochemischen Nachweis verschiedener

Zuckerarten in den Pflanzen-geweben mittels der Phenylhydraziiunethode.
Sitzungsher. d. k. Akad. d. Wiss. Wien, Math. nat. Klasse, 1905.

5. Katic, D. L. J. Beitrag zur Kenntnis der Bildung des roten Farbstoffs in

vegetativen Organen der Phanerogamen,

6. Kracs, G. Grundlinien zu einer Physiologic des Oerbstoffs. Leipzig, 1889.

7. Laborde, J. Sxu" le mecanisme physiologique de la coloration des raisins

rouges et de la coloration automnale des feuilles. C. R. Acad. Sci., 1908.

8. !Mangham, S. The Translocation of Carbohydrates in Plants. Part I.

Science Progress, October, 1910.

9. MiRANDE, M. Sur I'origine de I'anthocyanine d^uite de I'observation de

quelques Insectes parasites des feuilles. C. R. Acad. ScL tom. cxlv., 1907.

10. MoLiscH. tTber ein neues, einen karmin-roten Farbstoff erzeugendes Chromogen

bei Schenkia blumenaviana. Ber. d. d. bot. Gesell., 1901.

11. MoLLiARD, M. Action morphogenique de quelques substances organiques sur

les vegdtaux sup^rieurs. Rev. g^. de Bot. torn, xix., 1907 ; also, Production
experimentale de tuberculcs blancs et de tubercules noirs k partir de graines
de Radis rose. C. R. Acad. Sci., 1909.

158 On the Formation of Anthocyanin

12. Moore, B., and Whitley, E, The Properties and Classification of the
oxidising Enzymes and Analogies between Enzymic Activity and the
EflFects of Immune Bodies and Complements. Biochemical Journal,
Vol. IV., 1909.

1 3 Overton, E. Beobachtungen und Versucbe fiber das Auftreten von rothem
Zellsaft bei Pflanzen. Prings. Jahr. f. wiss. Bat. Bd. xxxiii., 1899.

14. Palladin, W. tJber das "Wesen der Pflanzenatmung. Biochem. Zeitsckr., 1909.

15. t?ber Prochromogene der pflanzlichen Atmungschromogene. Ber. d.

d. Bot. Gesdlsck, 1909.

16. Pick, H. Ueber die Bedeutung des rothen Farbstoffes bei den Phanerogamen

und die Beziehungen desselben zur Starkewanderung. Bot. Centralh.
Bd. XVI., 1883.

17. Senft, E. tJber den microchemischen Zuckernachweis durch essigsaures

Phenylhydrazin. Sitzungsber, d. k. Akad. d. Wiss. Wien, Math. nat. Klasse,
1904.

18. Waage, T. Ber. d. Deutsch. botan. Ges. 8.

19. Wheldale, M. The Colours and Pigments of Flowers with special Reference

to Genetics, P. R. Soc. B. Vol. lxxxi., 1909 ; also On the Nature of
Anthocyanin. Proc. Cam. Phil. Soc. Vol. xv., 1909.

20. Note on the physiological Interpretation of the Mendelian factors for

Colour. Rep. Evol. Com. Roy. Soc, Report v., 1909.

21. Wig AND, A. Die rothe und blaue Farbung von Laub und Frucht. Bot. Hefte.

Forschungen a. d. hot. Garten zu Marburg, 1887.

22. ZiMMBRMANN. Botanical Microtechnique.

FURTHER EXPERIMENTS ON THE INHERITANCE
OF COAT COLOUR IN MICE.

By FLORENCE M. DURHAM.

In Report IV of the Evolution Committee of the Royal Society, I
published a preliminary account of the results of my breeding experi-
ments to determine the inheritance of coat colour in mice. I now
propose to complete that account by giving the results of my investiga-
tions into the genetic behaviour of pink-eyed mice with coloured coats
and also of yellow mice.

I propose to begin with an account of the pink-eyed mice with
coloured coats, but at the same time to leave the question of the
behaviour of pink-eyed mice with yellow coats until I deal with dark-
eyed yellow mice, and to confine myself at first to pink-eyed mice of
any coat colour except yellow.

The albinos have been dealt with in Report IV. The pink-eyed
mice with coloured coats as stated in Report IV have only apparently
unpigmented eyes. Examination of sections of the eyes microscopically
reveals the presence of pigment both in the retina and iris. The
amount of pigment present is however so little, that it is extremely
difficult to say of what colour it is.

There is a correlated absence of pigment in the hairs of these mice,
so that they are much paler in colour than any of the corresponding
varieties of dark-eyed mice. But this absence of pigment in the eyes
and hair of the pink-eyed mice has a genetic significance different
from that of the dilution of coat colour in the dark-eyed mice. For
in the case of the dark-eyed mice, the absence of a factor which effects
the den.se deposition of pigment in the hairs gives rise to what are
known as the dilute forms, and for each coloured type there is a dilute
variety. The pale colours of the pink-eyed mice are not due to the
same cause, and cannot be explained in the same way. For pink-
eyed mice behave genetically like the concentrated and diluted varieties

160 Inheritance of Coat Colour in Mice

of dark-eyed mice and carry the conditions of concentration and dilu-
tion just as they do, and in their colourings the effects of these
are shown. The paleness of colour therefore which accompanies the
pink eye must be due to some other cause. This statement however
applies only to those mice in which yellow pigment is absent. For
it is possible to produce pink-eyed yellow mice with hair as deeply pig-
mented as that of dark-eyed yellow mice. These will be dealt with later
on. Also in the case of the pink-eyed agouti mice, while the black and
chocolate pigments are there in very much diminished quantities the
yellow banding may be as deeply coloured as in the hair of the ordinary
agouti. It is possible to arrange the pink-eyed mice in classes cor-
responding to those which have been distinguished among the dark-
eyed mice.

Pink-eyed mice which behave genetically like black mice are of a
pale greyish colour and were named lilacs by Mr Darbishire(6) who
was the first to breed them and kindly gave me two living specimens.

In order to distinguish them from other lilac mice, on account of
their colour, I have called them " blue lilacs." They breed perfectly
true mated inter se. Mated with chocolate mice, they never throw any
other colour but black in Fj.

In the F2 generation from this mating two new varieties appear
which I have named "chocolate-lilac" and "champagne" i^'cafe au lait"
of Cuenot) respectively.

The chocolate-lilacs vary very much in appearance in depth of
colouring, but the colour is always browner than that of the blue lilac
more resembling that of the silver fawn. For this reason I called them
chocolate-lilacs, and I thought at first they were a chocolate variety of
the pink-eyed mouse. But when mated with chocolate mice they throw
a mixture of blacks and chocolates.

Chocolate-lilacs mated together throw blue lilacs, chocolate-lilacs
and champagnes.

The champagne mice, mated with chocolates, throw only chocolates
and are I believe the pink-eyed variety of chocolate. Mated inUr se,
they breed perfectly true, I therefore regard the blue lilacs as the
homozygous pink-eyed variety of the dark-eyed black mouse, the
chocolate -lilac mouse as the heterozygous variety of dark-eyed black
(throwing chocolate) and the champagne as the homozygous chocolate
pink-eyed form.

When the various forms are mated with the dilute forms of dark-
eyed mice, blues or silver fawns, then in the F^ generation pink-eyed

F. M. Durham 161

mice without the factor for concentration are produced. These when
mated with blues or silver fawns throw only the dilute varieties, whereas
pink-eyed mice descended from unions between pink-eyed mice and
dark-eyed mice of the concentrated form only throw concentrated forms
when mated with the dilute varieties. There is a great deal of varia-
tion in the depth of colour of the pink-eyed mice and I think that the
presence or absence of the factor for concentration accounts for this.
Unfortunately I did not recognize this fact early enough in my experi-
ments to be able to give numbers in support of this view. In the case
of the champagne mice, however, a different variety which I called
" silver champagne," arose and always appeared in the F^ generation
from a mating between champagne and silver fawn. These silver
champagnes mated with dilute forms always gave dilute forms.

When the chocolate lilac mouse is mated with the ordinary wild
colour or golden agouti mouse, the ^i is always golden agouti. All
possible forms should appear in F^. Black, chocolate, golden agouti,
cinnamon agouti, blue lilac, chocolate lilac, champagne, pink-eyed golden
agouti, pink-eyed cinnamon agouti.

The pink-eyed agoutis, golden and cinnamon, are very much alike in
appearance. In fact at first and for some time I took the pink-eyed
cinnamon agouti to be a pale or dilute form of the pink-eyed golden
agouti, and owing to the small amount of pigment present, I thought
that the pink-eyed golden agouti must be the cinnamon variety.
However, the genetic behaviour of the two forms when mated with
chocolate showed their differences.

The pink-eyed golden agouti gives only golden agouti when mated
with chocolate and the pink-eyed cinnamon agouti gives only cinnamon
agoutis as a result of mating with chocolate.

The small amount of pigment present makes the microscopical
determination very difficult.

Pink-eyed coloured mice are recessive to dark-eyed mice and when
mated inter se never throw the dark-eyed form.

Taking all the results irrespective of colour and classifying only
according to eye-colour, then as a result of mating pink-eyed mice with
dark -eyed mice in F^ I obtained

875 dark eye, DE, 303 pink eye, PE, observed
883-5 „ 294-5 „ calculated.

From matings between heterozygous DE with PE
105 DE, 113 PE, observed
109 109 calculated.

162 Inheritance of Coat Colour in Mice

In the case of the first mating I made between blue mice and blue
lilac the numbers yielded in the F^ generation are peculiar.

Instead of a ratio of 9 : 3 : 4 as I expected, I got 27 blacks, 17 blues
and 18 blue lilacs.

The Fi mice were black and therefore the blue lilacs were carrying
the determiner for concentration.

The formula for the blue lilacs may be represented as eDB, where
e is the absence of dark eye, D the factor for concentration, B the factor
for blackness.

The blue mouse can be represented as EdB, where E is the presence
of dark eye, d is the absence of concentration.

The figures given above may possibly be an indication of spurious
allelomorphism between the factor for dark eye and the concentration
factor.

The F^ mating would then be EdeD x EdeD. The results would
then be a ratio of 2 black to ] blue to 1 blue lilac, giving calculated
results of 31 black to 15'5 blue to 15*5 blue lilac.

I was unable to repeat the combination owing to either the blues
used being heterozygous in chocolate or the blue lilacs heterozygous in
concentration.

If this interpretation be correct, then all the blacks should be hetero-
zygous and all the blues homozygous. Unfortunately I only mated a
few of the offspring. 6 blues only were mated and 3 of these died
without young, the remaining 3 were homozygous ; 5 blacks were mated,
3 died without young, one had only 4 young and these were all black,
and the fifth was heterozygous.

The results of mating chocolate-lilac mice with dark-eyed varieties
may give rise to various heterozygous forms.

Thus the F^ generation of a mating between chocolate-lilac and
blue (giving black Fi) was

19 black, 2 blue, 5 blue lilac, 6 chocolate-lilac.

If the mating was EeDdBb x EeDdBB, the calculated numbers
would be 18 black, 6 blue, 4 blue lilac and 4 chocolate-lilac.

From a mating of chocolate-lilac and black heterozygous in blue
giving black ^i, I got

10 black, 4 blue, 6 blue lilac and 2 chocolate-lilac.

If the mating were as above between EeDdBb x EeDdBB, there
should be 12"3 black, 4*3 blue, 27 blue lilac and 2*7 chocolate-lilac.

F. M. Durham 163

Blue lilac x chocolate, eeDDBB x EEDDhb gives black F,.

Fi. Observed
Calculated

Here no blue lilacs were obtained but an excess of champagnes.
Bltie lilac x chocolate, eeDdBB x EEDdhh gives blues and blacks.

BUck

CbocoUte

Blue
lilac

Chocolste-
lilac

Champagne

20-7

6-9

2-3

4-6

2-3

Black Blue

Chocolate

SUver
fawn

Blue Chocolate-
lilac lilac Champagne

Fj. Observed 4 3

11 0

Chocolate-lilac x chocolate, eBeb x

EbEb

gives blacks and choco

Fr.

Fi from blacks „, ^
' Black

Chocolate

Blue
mac

Chocolate-
lilac Champagne

Observed ... 16

3 0

Calculated ... 14 4

4-8

1-6

3-2 1-6

From black and chocolate

Observed ... 8

0 5

Calculated ... 10-8

10-8

3-6 3-6

From chocolate x chocolate

Observed ... —

—

— 16

Calculated ... —

40-5

—

— 13-5

Bltie lilac x silver fawn, eeddBB x EEddbb giving blue F^.

Bine

SUver
fawn

Blue
mac

Chocolate-
lilac Champagne

Fi. Observed 43

10 3

Calculated ... 43*9

14-6

4-9

9-7 4-9

Chocolate-lilac x silver /aim, eeddBh x EEddbb giving blue Fi.

Blue

SUver
fawn

Blue
Ulac

Chocolate-
lilac Champagne

Fj. Observed ... 16

0 2

Calculated ... 20-7

6-9

2-3

4-6 2-3

The champagnes in these last two cases were silver champagnes.

Silver fawn x champagne giving chocolate F^,
EeDdbb x EeDdbb.

Chocolate

SUver
fawn

Champagne

Observed

Calculated .

6-75

2-25

These champagnes should have been of two sorts, champagne and
silver champagne.

164 Inheritance of Coat Colour in Mice

Silver fawn x champagne giving silver fawn F^,
Eeddhh x Eeddhh.

Silver
fawn

Silver
champagne

F2>

Observed

Calculated ...

9-75

3-25

Chocolate heterozygous in pink-eye x chocolate-lilac,

Eebh X eeBh.

Black

1
Chocolate

unocoiate-
Hlac

Champagne

Observed

Calculated ...

2-5

Blue X champagne giving black F^,

EeDdBh x EeDdBh.

Black

Blue

Chocolate

Silver
fawn

Blue
lilac

Chocolate-
lilac

Champagne

Observed

Calculated

3-5

1-17

•4

•5

Blue X champagne giving blue and black F^,
EeddBb x EeDdBh.

Black

Blue

Chocolate

Silver
fawn

Blue
lilac

Chocolate-
lilac

Champagne

Fi-

Observed

Calculated

Blue carrying chocolate x champagne giving blue and chocolate,
EeddBh x EeDdhh.

Black

Blue

Chocolate

Silver
fawn

Chocolate-
lilac

Champagne

Observed

Calculated

2-1

1-3

Blue heterozygous in pink-eye and chocolate x champagne,
EeddBh x eeDdhh.

Silver Chocolate-
Black Blue Chocolate fawn lilac Champagne

Observed ... 3 0 5 3 1 1

Calculated ... 1-6 1-6 1-6 1-6 3-2 3-2

Blues carrying pink-eye mated together,
EeddBB x EeddBB.

Blue

Blue lilac

Observed

Calculated

18-75

6-25

F. M. Durham 165

Golden agouti x chocolate-lilac gives golden agouti F,,
GgBbEe x GgBhEe.

Pink -eyed
Golden Cinnamon Pink-eyed cinnamon Blue Chocolate-

agouti agouti Black Chocolate agouti agouti lilac lilac Champagne

Ft. Observed 83 8 31 2 26 7 0 11
Calculated 72-9 24 3 24 3 8 1 24 3 8 1 27 5 4

4
27

Golden agouti y. pink-eyed agouti gives golden agouti F^.

Golden agouti Pink-eyed agouti
Fa- Observed ... 32 17
Calculated ... 3675 12-25

F,.

Cinnamon agouti mated with chocolate-lilac giving cinnamon agouti

Cinnamon Pink-eyed

agouti Chocolate Cinnamon agouti Champagne

Fa. Observed ... 14 4 3 1

Calculated ... 12-3 4 1 4-1 1*4

Agouti heterozygous in pink-eyed agouti x pink-eyed agouti.

Agouti Pink -eyed agouti
Observed ... 11 8

Calculated ... 9*5 9-5

Agouti heterozygous in pink-eye and chocolate x pink-eyed agotUi
heterozygous in chocolate.

Pink-eyed
CSnnamon Pink-eyed cinnamon Blue Chocolate-

Agouti agouti Black Chocolate agouti agouti lilac lilac Champagne

Observed

Calculated

5-85

1-95

-65

5-85

1-95

•65

1-3

•65

Agouti heterozygous in pink-eye and chocolate x black ditto.

Pink -eyed
Cinnamon Pink-eyed cinnamon Blue Chocolate-

Agouti agouti Black Chocolate agouti agouti lilac lilac Champagne

Observed 61 10 3 0 6024

Calculated 93 93 3 1121

Pink-eyed agouti x pink-eyed agouti. From this mating I obtained

Pink-eyed Pink-eyed Chocolate-

agouti cinnamon agouti lilac

37 4 8

There were no blue lilacs and no champagnes. The explanation of
this may be that the pink-eyed agoutis were not all carrying the same
characters.

Another case I cannot explain is the following :

An albino heterozygous in E was mated with a yellow carrying
agouti. From the agoutis Fi of this union I obtained

17 agouti, 5 black, 1 chocolate-lilac, 1 champagne and 8 albinos.

166 Inheritance of Coat Colour in Mice

There were no chocolates, no cinnamon agoutis, no pink-eyed agoutis
of either sort, and no blue lilacs.

I have tried other raatings of various sorts but the numbers yielded
are too small to be worth quoting.

Yellow Mice.

The genetic behaviour of yellow mice differs in various particulars
from that of other mice ; and there is at present no very satisfactory
explanation possible to account for this.

Hagedoorn(l) is the only one among many breeders of yellow mice
whose experiences are not in accordance with my own. But from his
account of his experiments, it is clear that he was using a different
type of yellow mouse from that employed by the rest of us.

The type, which I and other breeders have used, must be regarded
as a heterozygous dominant. For it never breeds true, no homo-
zygous form has yet been obtained; and when mated with mice of
other colours than yellow, some of the offspring are always yellow.
Hagedoorn's mouse was a recessive and did breed true. His experi-
ments are of interest as showing that another type of yellow mouse
exists, but his results need not be considered further here.

I made 185- matings in all between yellows bred in every kind of
way, but every one of these yellows proved to be heterozygous.

As a result of 127 matings between yellows I obtained 448 yellows
and 232 other colours. I purposely excluded from the list all matings
from which sables and albinos were obtained, so as to count only the
pure yellow forms. Albinos can carry the yellow determiner, and the
sable mouse, which is perhaps only a variant of the yellow, presents so
many peculiarities as I shall show later on that for the present purpose
I preferred to exclude it.

As a result of 104 matings between yellows and other colours I
have obtained 297 yellows and 336 other colours.

The problem created by the absence of pure yellows has been dis-
cussed by Cu^not(2), Castle (5), Wilson, Morgan and others. There are
two possibilities: (1) that in fertilization the zygotes, yellow x yellow, are
never formed; (2) that these zygotes are formed but perish. If they
are not formed we should expect the ratio of yellow to non-yellow to
approximate in .Pg to 3 : 1, because the number of spermatozoa is
indefinitely large; if on the other hand such zygotes are formed and
perish, the F.^. ratio should be 2 : 1.

F. M. Durham 167

The F^ numbers obtained are as follows :

Yellow Non-yelloir

Cu^not(2) 263 100

Castle (5) 800 435

My own 448 232

1511 767

Expectation at 2 : 1 1518-6 759 3

Expectation at 3 : 1 1708-5 5695

From these figures there can I think be no longer any serious doubt
that the pure yellow zygotes are actually formed in fertilization, but that
for some unknown cause they are unable to develop. The case becomes
therefore exactly comparable with that observed by Baur(7) for the
varietcUes aureae, which form albino embryos incapable of existence.

It has been argued that if this representation is correct the average
numbers per litter should be less for the mating yellow x yellow than
for yellow mated with some other colour, and Cuenot and Castle record
a difference of this kind, giving the following averages:

YeUow X Yellow

Yellow X Non-reOow

Cuenot

3-38

3-74

Castle

4-71

6-57

From my experience I incline to doubt whether much importance
can be attached to differences of this order.

The following averages have been compiled from an ample series,
75 litters being the lowest included.

yellow X yellow

yellow X other colour

black X black

black X other colour (not yellow)

chocolate x chocolate

chocolate x other colour (not yellow)

agouti X agouti

agouti X other colour (not yellow)

albinos x other colour (not yellow)

I have not mated albinos together often enough to make it worth
while to compare the results of mating albino x albino with the other
figures.

Only mice which lived long enough to have their colours determined
are included in these averages, but Castle's figures evidently are based
on the numbers actually born. It is clear nevertheless that large
differences exist where no special disturbance, analogous to that we are

3-90

young

3-97

4-60

3-99

»»

3-96

«i

3-93

3-47

3-32

>»

4-27

168 Inheritance of Coat Colour in Mice

considering, is to be suspected, and I doubt whether the observations
can be used either for or against the conclusion that the ratio of yellow
to non-yellow in F»/\%1 -. 1.

The non-viability of pure yellows raises an important physiological
question, but we have no indication as to what may be its cause. It
should be remembered that the mortality may, for aught we yet know,
occur at any age between fertilization and maturity.

In the report to the Evolution Committee (3), I have already stated,
that the pigments of the eye of the yellow mouse may be black or
chocolate but never yellow. If the yellow mouse throws chocolate
young but never black the eye will be found to be pigmented with
chocolate, often chocolate pigment will also be found in the hairs of this
animal.

A yellow mouse which throws black young will have black pigment
in the eyes and some black pigment will always be found in the hair. I
have never found black pigment in the hair of a mouse with chocolate
only in the eyes.

I have examined several hundred yellow mice and never found an
exception to this statement.

The hair and the eyes are a key to the genetic behaviour, or one
may equally well say the genetic behaviour is the key to the pigments
of the hair and eyes of the yellow mouse. Both black and chocolate
pigments will be found in the eyes of the yellow mouse with agouti
determiner.

Yellow mice are subject to an abnormal development of fat in their
tissues. All the fat depots become loaded to an extraordinary degree.
This development of fat renders them unable to breed. It is a well-
known fact to the breeders of Fancy mice.

The question of dilution is also a difficulty in yellow mice. Yellow
mice vary very much in their colouring. Some are very deep yellow,
some much paler, some are deeply coloured dorsally and very light
underneath, pale almost to whiteness. I do not mean piebald, but the
colour fades off gradually to a very pale cream. The result is that it is
very difficult and often impossible to decide whether a mouse belongs
to the dilute variety or not. Of course many mice are so pale all
over, one would not hesitate to class them as dilute yellows, that is
creams. But there is a very large section whose classification can only
be determined by their genetic behaviour. To illustrate the difficulty
I will mention the case of two mice which I bred together and classed

F. M. Durham 169

as creams and they threw chocolates. If they had been real creams
they should have thrown silver fawns. Another cream mouse which I
had grew a chocolate streak, late in life, down its back, a reversal of the
ordinary procedure.

When yellows are bred with pink-eyed mice, pink-eyed yellows will
appear in Fo as deeply coloured as the original yellow mouse which was
grand-parent. As stated before the yellow bar of the pink-eyed agouti
mouse is so deeply coloured and so bright that the inexperienced observer
would put them in the yellow class. I believe that the so-called pink-
eyed yellow mice of Plate's (4) classification must be really pink-eyed
agoutis, either golden or cinnamon.

The pink-eyed yellow mice when produced behave exactly like the
dark-eyed yellows. I have never succeeded in obtaining a homozygous
pink-eyed yellow and when mated together they do not throw 3 yellows
to 1 other colour; mated with any other colour they always throw some
yellows. The dark eye is dominant to the pink eye, but the yellow
colour behaves independently of the eye colour when pink-eyed yellow
is mated with dark-eyed any other colour.

Pink-eyed yellows mated together throw pink-eyed yellows, blue
lilacs, chocolate-lilacs and champagnes according to their genetic con-
stitution.

From the matings of pink-eyed yellows I have obtained the follow-
ing results,

17 matings PE7 x PEY gave 57 PEY, 45 PE other colour.

19 matings PEY x PE other colour gave 32 PEY, 33 PE other
colour.

Before proceeding to give the tables of the results of the various
matings I have carried out, I must now refer to two cases in which
I obtained yellow mice by breeding together other varieties than yellow.
In each case the mice had pink- eyed ancestry.

Case I. This mouse was not strictly speaking a yellow mouse. I
could not class it as an agouti simply or as a sable. It was very yellow
in colour, with the agouti barring on the dorsal surface and a yellow
belly. It resembled a very yellow agouti with a yellow belly.

Its ancestry is shown by the following scheme :

Pink-cyed cfaooolate-lilac Yellow throwing agoutis

1188 X 1450

, i ,

1667 <f X 1555 J
^^^___ I

3 black 3 agoutis 1 blue 1 pink-eyed 1 yellow

champagne agouti
1825
Joaxn. of Gen. i 12

170 Inheritance of Coat Colour in Mice

Both 1667 (/"and 1555 % were agoutis and not to be distinguished
in any way externally from any ordinary agouti.

I mated the yellow agouti mouse (1825 ^) with 6 does, but unfor-
tunately the matings were not all successful. With a chocolate %
there were 20 young (not one of which was agouti), 1 yellow, 4 blacks,
7 sables, 1 chocolate yellow belly, 3 chocolates, 3 albinos, 1 chocolate-
lilac.

Mated with a yellow mouse carrying chocolate he gave 2 yellows,
2 sables, 3 blacks and 1 chocolate.

I tried him four times with agouti mice but in each case there was
no result. I had hoped by such matings to obtain agoutis which would
throw yellows or sables.

None of the offspring mated together produced any agoutis.

Case II. A champagne % was mated with an agouti ^. In the
first generation there were

1 agouti, 4 cinnamon agouti, 1 chocolate.

The agouti which was a (/ was mated with the only % a cinnamon
agouti, and there resulted

1 sooty yellow, 2 silver cinnamon agouti and 1 black.

Unfortunately death carried off the yellow before she could be
mated. Subsequent litters of the parents did not contain any yellows.

In the subjoined tables the calculations are made on a 2 to 1 basis
instead of the ordinary 3 to 1, adopting the conclusion indicated above.

TABLE OF RESULTS.

Dark-eyed Yellows.

Yellows carrying chocolate mated together :

Yellow Chocolate

136 68 observed

136 68 calculated

Yellows carrying chocolate x chocolate :

Yellow Chocolate

66 46 observed

56 56 calculated

Yellows carrying black and chocolate mated together :
Yellow Black Chocolate

65 35 9 observed

72 27 9 calculated

observed

10-7

calculated

Black

•

observed

calcalated

F. M. Durham 171

Yellows carrying black and chocolate x chocolate :
Yellow Black Chocolate

23 11

21-4 10-7

Yellows carrying black x chocolate :
YeUow
6
12

Yellows carrying black and chocolate x black heterozygous in chocolate :
Yellow Black Chocolate

25 17 18 observed

30 15 15 calculated

Yellows carrying black x black :

Yellow Black

29 24 observed

26-5 26-5 calculated

Yellows heterozygous in black, chocolate and albino mated together :
Yellow Black ChocoUte Albino

59 27 5 30 observed

45-3 34 11-3 30-4 calculated

Albinos heterozygous in yellow and chocolate x chocolate heterozygous in albino :
Yellow Chocolate Albino

12 5 7 observed
8 4 12 calculated

In the following tables the yellows are not separated into yellows
and creams on account of the diflSculty stated above of distinguishing
between them.

Yellows heterozygous in chocolate and silver fawn mated together :
Yellow and Cream Chocolate Silver fawn

5 7 5 observed

11-2 4-2 1-4 calculated

Yellows heterozygous in chocolate and silver fawn x silver fawn :
Yellow and Cream Chocolate SUver fawn

33 11 13 observed

29-0 14-5 14-5 calculated

Yellows heterozygous in chocolate and silver fawn x chocolate heterozygous in silver
fawn:

Yellow Chocolate SUver fawn

13 10 9 observed
16 12 4 calculated

Yellow heterozygous in black and albino x albino heterozygous in yellow and black :
Ydknr Black Albino

5 2 7 observed

4-6 9*8 6-9 ^culated

172

Inheritance of Coat Colour in Mice

Yellow X Agouti gives Yellow and Agouti,
Fi yellow x Fj yellow :

Yellow

Agouti

observed

61*2

30-6

calculated

Fi yellow x Fi agouti :

Yellow

Agouti

observed

■

calculated

Yellow heterozygous in

agouti X chocolate :

Yellow

Agouti

observed

calculated

Yellow heterozygous in agouti x black :

Yellow Agouti

11 18 observed

14*5 14*5 calculated

Yellow X Chocolate-Lilac gives Yellow and Black.
Fi yellow x Fi yellow :

Dark-eyed
yellow

Black

Chocolate

Pink-eyed
yellow

Blue
lilac

Chocolate-
lilac

Champagne

observed

calculated

Here there was an excess of champagnes, no blue lilacs and only one
chocolate-lilac.

Fi yellow x Fi black :

Dark-eyed
yellow

Black

Chocolate

Pink-eyed
yellow

Blue
lilac

Chocolate-
lilac

Champagne

3 observed

28-8

21-6

7-2

9-6

2-4

4-8

2-4 calculated

Here there was an excess of chocolate-lilacs, no blue lilacs.
Dark-eyed yellow heterozygous in pink-eye x pink-eyed yellow :

Dark-eyed
yellow

Black

Chocolate

Pink-eyed
yellow

Blue
lilac

Chocolate-
lilac

Champagne

7 observed

26-4

9-9

3-3

26-4

3-3

6-6

3*3 calculated

FixFii

Yellow X Pink-eyed Agouti gives Yellows.

Dark-eyed Pink-eyed Pink-eyed

yellow Agouti yellow agouti

25 17 11 5 observed

28-8 14-4 9-6 4-8 calculated

F. M. Durham

173

Dark-eyed yellow heterozygoas in chocolate and pink-eye :

Dark-eyed
yellow

Chocolate

Pink-eyed
yellow

Champagne

observed

9-G

4-8

2-4

calculated

Dark-eyed yellow heterozygoas in chocolate and pink-eye x chocolate heterozygoas in
pink-eye :

Dark-eyed _ Pink^eyed

yeUow
8
8-4

Chocolate

7
8-4

yellow
1
2-8

Yellow heterozygous in pink-eye x blue lilac :

Dark-eyed
yellow

Black
0

Chocolate
2

Pink-eyed
yellow

Blue
lilac

Champagne
6
2-8

Chocolate-
lilac

2 0 2 2 1

Yellow heterozygoas in pink-eye x chocolate-lilac :

observed
calculated

Champagne

0 observed

Dark-eyed
yeUow

Black
2

Chocolate
4

Pink-eyed
yellow

Blue
lilac

Chocolate-
lilac Champagne

observed

Yellow heterozygous in pink-eye and chocolate x champagne :

Champagne

Dark-eyed
yellow

Chocolate
4

Pink-eyed
yellow

observed

14 3 9

Here there is an excess of pink-eye.
Pink-eyed yellow x pink-eyed yellow heterozygous in black and chocolate :

Pink-eyed
yellow

Blue
lilac

Chocolate-
lilac

Champagne

1 observed

28 8

3-6

7-2

3-6 calculated

Here there is an excess of chocolate-lilac possibly due to the fact
that some of the yellows were heterozygous in blue lilac only and
others in chocolate only, and those mated together would give chocolate
lilacs only, no blues and no champagnes.

Pink-eyed yellow x pink-eyed yellow heterozygous in chocolate only :
Pink-eyed yeUow Champagne

18 12 observed

20 10 calculated

Pink-eyed yellow x pink-eyed agouti :

Pink-eyed yellow Pink-eyed agouti

15 7 observed

11 11 calculated

Pink-eyed yellow heterozygoas in pink-eyed agouti :
Pink-eyed yellow Pink-eyed agouti

12 8 observed
18-2 6-6 calculated

12-3

174 Iiiheritance of Coat Colour in Mice

Pink-eyed yellow x chocolate-lilac ;

Pink-eyed Blue Chocolate-
yellow lilac lilac Champagne

8 2 8 8 observed

10-4 2-6 5-2 2-6 calculated

Pink-eyed yellow x champagne :

Pink-eyed Chocolate-

yellow lilac Champagne

7 4 7 observed

9-0 4-5 4-5 calculated

Sable Mice.

Among the yellow mice I used for my experiments were some
individuals, which produced sables when mated with blacks or choco-
lates. As these appeared very early in my experiments, I at first
concluded that sables would always result from such matings. Subse-
quent investigation however showed that the power to produce sables
was limited only to certain mice and that it was a hereditary quality.
At present I am unable to offer a scheme which correctly represents the
relation of sables to the other colours.

Sable mice are well known to the Fancy. They differ from yellow
mice in having a dark black or brown streak down the middle dorsal
region while the rest of the mouse is yellow. The streak may be very
narrow, when the mouse is said to be a light sable, or very broad when
the mouse is a dark sable. As a general rule, the hairs in this dark
streak show an agouti pattern, being black or chocolate barred with
yellow. But this does not mean that the mouse is carrying agouti
determiner. But it is possible to produce sables in which the barring
of the dorsal hairs is absent, and at various times I have had black,
blue, chocolate and silver fawn mice which differ only from the ordinary
forms by having yellow bellies, and which from their genetic behaviour
must be classed with the sables. They always moulted subsequently
into ordinary sables.

The appearance of the sable mouse varies very much according to
age. During the first few months the marking is very definite, but as
age comes on the sable appearance is lost, so that a mouse, which was
a very good specimen at three months may be hardly distinguishable
from an ordinary yellow mouse at 18 months old. The amount of
yellow in its colouring increases with the successive moults.

Sables are not to be confused with sooty yellow mice, which result
from mating ordinary yellows with blacks or chocolates. The sooty
yellow is a dirty colour all over and never shows a definite pattern.

F. M. Durham 175

I have never bred a homozygous sable mouse. Bred together,
sables may throw sables, yellows, blacks, chocolates, and also agouti, if
they are carrying the agouti determiner.

Yellows carrying the sable determiner mated together will throw
sables, and sables mated together may throw yellows. By mating
together yellows carrying sable I have obtained

111 yellows, 38 sables, and 69 other coloured mice.

By mating yellows carrying sable with other coloured mice, not
yellows, I have obtained

78 yellows, 55 sables, and 80 other coloured mice.
Mating together sables, I have obtained

161 sables, 43 yellows, and 142 other coloured mice.
Mating sables with other colours, not yellow, I have obtained

93 sables, 90 yellows, and 174 other coloured mice.

Examination of the records suggests, that there is more than one
sort of sable mouse, and that it is possible to produce sables which
never throw yellows at all.

Thus I had as a result of 5 matings between blue sables, 29 blue
sables, and 23 blue mice, and no yellows at all.

4 matings between blue sables and dark sables gave
16 sables and 8 other colours (no yellows).

7 matings between blue sables and blue gave
20 blue sables and 19 blues.

On examination of the results produced by mating sables together,
I find that the matings in which yellows were produced, the oSspring
consist of 62 sables, 43 yellows and 64 other colours, while the offspring
of the matings in which no yellows were produced, consist of 99 sables
and 78 other colours, suggesting a 9 to 7 ratio.

The matings of sable x other colour show that the families in
which yellow appeared consisted of

48 sables, 90 yellows, and 107 other colours,

in the remaining families there were 45 sables and 67 other colours.

Matings between sables and yellows .w^ithout the sable determiner
give

23 yellows and 18 other coloured mice, no sables.

Matings between yellows carrying sables with sables give
14 yellows, 28 sables and 17 other coloured mice.

176 Inheritance of Coat Colour in Mice

These results suggest that sable is recessive to yellow. But at the
same time it seems extraordinary that some of the sables should be
able in their turn to throw yellows, and at present no adequate
explanation is forthcoming. The fact that there is probably more than
one kind of sable may supply the basis for explanation, but the question
ought to be more fully worked out.

Sables, like yellow mice, show a tendency to become abnormally fat.

Besides obtaining ordinary sable mice, another form appeared un-
expectedly in my experiments. These I called reversed sables, because
in them the agouti pattern was reversed. The base of the hairs was
yellow and the barring was black or chocolate.

This marking was only apparent in the young mouse; after about
6 months these mice moulted completely yellow, but it was a very con-
spicuous feature in early life.

I twice obtained mice which were agouti all over with the pattern
reversed. They behaved exactly like sables and never threw any
agouti young.

Matings between sables in which the reversed sables appeared gave

49 sables, 21 reversed sables, 19 yellows, 41 other colours.

Matings between sables and other colours, not yellow, in which
reversed sables appeared,

12 sables, 12 reversed sables, 9 yellows, and 29 other colours.

The reversed sables are recessive to the other sable, mated together
they produced reversed sables and other colours but never ordinary
sables, and mated with ordinary sables did not produce reversed sables.

Sables which could throw reversed sables when mated with reversed
sables, gave

10 sables, 8 reversed sables, 5 yellows, and 18 other coloured mice.

Owing to the great variation which obtains amongst sables, it has
not been possible to classify them very satisfactorily.

Matings between dark sables (broad dorsal streak) x light sables
(narrow dorsal streak) gave

8 dark sables, 3 light sables, 10 other colour.

Dark sable by yellow gave

10 dark sable, 5 light sable, 8 yellow, 8 other colour.

Sable X agouti gave

7 sable, 7 yellow, 17 agouti.

F. M. Durham 177

Sable X heterozygous agouti gave

29 sable, 26 yellow, 15 agouti, 23 black, 4 chocolate.
Fi sables from sable x agouti gave

17 sable, 8 yellow, 5 agouti, 4 black, 2 chocolate.
Yellow carrying sable x yellow carrying agouti
6 sable, 14 yellow, 7 agouti.

Agouti Mice.

I have made some further matings between agoutis on account of
the suggestion made by Cuenot (2) that a chocolate mouse was to be
regarded as the dilute form of black. He made this suggestion in order
to account for the results of breeding agouti by chocolate.
I therefore mated agouti with blue (dilute black).
Fi was all agoutL
i\ gave 30 ag., 13 dil. ag., 10 black, 4 blue,

32 10-7 10-7 3-6 calculated.

These dilute agoutis are well known to the Fancy as Silver Brown,
though a better name would be silver agouti. According to Cuenot
the cinnamon agouti would be the dilute form, and not the silver
browns.

If cinnamon agouti is mated with silver fawn (dilute chocolate) the
^1 is cinnamon agouti.

F.2 39 c. ag., 10 dil. c. ag., 9 choco., 4 silver fawn,
34-8, 11-6, 11-6, 3-9 calculated,

so that in both cases of golden agouti and cinnamon agouti it is possible
to produce a diluted form.

Other matings are : agouti x silver brown gives agouti F^:

Agouti Silver brown

Fi 45 16 observed

45-75 15-25 calculated

Silver cinnamon x silver fawn gives silver cinnamon :

Silver cinnamon agouti Silver fawn
Ft 25 10 observed

26-25 8-75 calculated

Agouti X cinnamon agouti gives agouti F^ :

Agoati Cinnamon agouti

F% 9 4 observed

9-75 3-25 calculated

178 Inheritance of Coat Colour in Mice

This evidence is perfectly consistent with the scheme which I
previously published (3) expressing the relations of black, chocolate,
blue, silver fawn, to each other, and shows conclusively that Cu^not's
representation of chocolate as a dilution form of black is incorrect.

Finally I wish to record the result of mating agoutis together
heterozygous in black. The results should yield a ratio of 3 to 1, but
my numbers are not in accordance with this ratio.

I obtained 76 agouti, 37 blacks.

The agoutis which result from mating agouti with black are much
darker than the ordinary agouti and very often there is a markedly
dark streak down the middle of the back.

The expenses of these experiments were partially defrayed by a
grant from the Royal Society.

REFERENCES.

(1) Hagedoorn, Archiv Entwickelungsmeckanik, Vol. xxviii. 1909.

(2) Cu^NOT, Arch, de Zool. Exp. et Gm. Vols. i. ll. lii. vi.

(3) Durham, Report IV. Evolution Committee Royal Society.

(4) Plate, Festschrift zum 60 Gehurtstage Richard Hertwig, Bd. ii. 191.

(5) Castle, Science, N. S., Vol. xxxii. 1910.

(6) Darbishire, Biometrika, Vol. ii. 1903.

(7) Baur, Ber. Deutsch. Bot. Ges. xxv. 1907.

Volume I AUGUST. 1911 No. 3

SOME STAGES IN THE SPERMATOGENESIS
OF ABRAXAS GROSSULARIATA AND ITS
VARIETY LACTICOLOR.

By L. DONCASTER, M.A.,
Fellow of King's College., Cambridge.

It has been shown by various cytologists, especially in the United
States, that in certain insects and other animals, the sex of the
individual is related to the presence or absence of a particular
chromosome in the nucleus of one of the gametes from which that
individual was produced. It has also been frequently pointed out
that the behaviour of the chromosomes in the maturation of the
gametes is exactly adapted to bring about Mendelian segregation, if
the members of an allelomorphic pair of characters are determined
by a pair of chromosomes which separate in gametogenesis. In the
Currant Moth {Abraxas grossulariata) I have shown ^ that a pair of
very definite Mendelian characters is intimately associated with sex,
in such a way that one of them is never borne (before fertilisation)
by eggs which will produce females. The two forms have the typical
grossulariata character, and the lacticolor character respectively, and
breeding experiments show that the grossulariata determinant is never
borne by female-determining eggs. It therefore seemed that a study
of the gametogenesis of this species oflfered exceptional hope of showing
the relation between a Mendelian character and a chromosome, if such
relation exists. The investigation cannot be regarded as completed,
but in the account which follows of the results obtained, it will be seen
that the hope of identifying a chromosome as the bearer of a hereditary
character has not been fulfilled, although other phenomena of consider-
able interest have been observed.

' Evolution Committu Roy. Soc. Report, it, 1908, p. 53.
Joarn. of Gen. i 18

180 The Spermatogenesis of Abraxas

Since it is in the egg that the incompatibility between the sex
determinant and a body character is found, logically it is in the
maturation of the egg that the chromosome which determines this
character, if it exists, should be sought. But the practical difficulties
in studying the maturation of an egg provided with a thick shell are
so great that the spermatogenesis was investigated first, in order to
find out whether any differences exist between the chromosome groups
in the pure forms of the two varieties, and in the heterozygote produced
by crossing them.

The spermatogenesis lasts for a considerable period in the late larval
and early pupal stages ; during the first week or so of pupal life all
stages may be found from spermatogonia with division-figures, through
growth stages to spermatocytes with first and second divisions, and the
conversion of the spermatids into spermatozoa. The testes are divided
into several compartments, each containing numerous follicles, and
within these compartments at one end or side spermatogonia are
found, and from these the later stages may be traced in fairly regular
order, all the cells in one follicle being at nearly the same stage. The
male and female pupae can be distinguished by the pits representing
the genital openings of the moth, and the material was prepared by
opening male pupae in Ringer's fluid (NaCl 0'7 gr., CaClj 0*23 gr,,
KCl 0"3 gr.. Water 1000 c.c), removing the orange-coloured testes
which are united in the middle dorsal line, and fixing them for about
half to one hour in Flemming's strong fluid. After being well washed
in water, and taken gradually through successive alcohols into cedar
oil, they were embedded in paraffin and cut into sections 6/i in
thickness. They were stained on the slide with Heidenhain's Iron
Haematoxylin. Testes of larvae were also examined in the same way.

The spermatogonia are rather small, and not so clearly divided into
follicles as in the later stages. Some follicles or groups of cells show
mitoses, in which the chromosomes form a flat equatorial plate and
very regular anaphase figures, but the chromosomes are so small and
numerous that I have not been able to count them with complete
accuracy. The number however is clearly at least 50 (Fig. 1), and
a study of the later divisions makes it probable that it is in reality 5G.

After the last spermatogonial division the nuclei undergo a growth
stage, during part of which the chromatin takes the form of a great
number of fine granules almost filling the nucleus. Some groups of
cells among the spermatogonia occasionally degenerate, and produce
masses of deeply staining material. As growth proceeds, the normal

L. DONC ASTER 181

cells become very definitely arranged in follicles, in which a space
begins to appear in the centre. From this stage onwards threads
may be seen running out from the cells into the cavity of the follicle,
ending in small vesicles or transparent masses of cytoplasm. Frequently
it can be made out that these threads are attached to the centrosome
of the cell, as described by Meves (Anat. Am. xiv. p. 1), and in some
cases it appears that there are accessory threads like those figured by
Meves in the Bee {Arch. Mikr. Anat. Bd. Lxx. p. 414).

From this stage onwards the fate of the different follicles varies, for
as Meves ^ has described in Pygaera and Voinov' in other Lepidoptera,
there are two distinct types of spermatogenesis. The first and more
normal type is found chiefly in larvae before pupation and in very
young pupae (1 — 3 days), but both types are found concurrently except
in advanced pupae, in which only the second is usually found. In the
first type the primary spermatocytes enlarge considerably, and before
the maturation divisions the nucleus contains about 28 chromosomes
scattered round the nuclear membrane (Fig. 2). These then form
a flat equatorial plate, in which 28 chromosomes may easily be counted';
they are not all equal in size, but none can be identified as differing
conspicuously from its fellows (Figs. 3 — 5). Seen in side view, the
chromosomes are elongated or dumbbell-shaped, and after division
they travel to the poles with great regularity and form a vesicular
nucleus (Figs. 7, 8, 9). From the rarity of its occurrence this stage
probably persists but a short time, and it breaks down to give a second
division spindle. Here again the chromosomes are regularly arranged,
and 28 may usually be counted without difficulty, about half the size
of those of the fii-st division (Fig. 10). The cells divide (Figs. 11, 12),
and give rise to spermatids which develop into spermatozoa in a manner
closely similar to that described by Meves in the "eupyrene" spermatozoa
of Pygaera.

In the second type of spermatogenesis the growth-phase of the
primary spermatocytes appears to be shorter, and they prepare for
division when their diameter is conspicuously smaller. The nuclei
contain a number of chromatin granules, which in the prophase of

1 Arch. Mikr. Anat. Bd. lvi. (1900), p. 56.5, and Bd. uu. (1903), p. 62.

- Arch. Zool. Exp. et Gin. 1903, Notes et Revues, p. xlix.

' In one testis of a heterozygote between the grossulariata and lactieolor varieties some
equatorial plates appear to contain 29 chromosomes. In some figures this appears to be
due to the fact that a few of the chromosomes have already divided, for some are markedly
smaller than the rest, but in one case 29 can be counted in which there is no direct evidence
for division (Fig. 6).

1»— 2

182 The Spermatogenesis of Abraxas

the first division are mostly arranged round the membrane, and
approach 50 in number (Fig. 13). As the division approaches there
are indications that these small chromatin masses become associated
in pairs, and before the spindle is formed between 20 and 30 of these
pairs may be counted (Fig. 14). This doubleness is much less pro-
nounced in spermatocytes of the first type, and in them the preliminary
diploid condition is not found. The first division of these smaller cells
differs greatly from that described above, for the equatorial plate is so
irregular that an exact count has never been possible ; the number of
chromosomes, however, approaches 28 (Figs. 15, 16). The anaphase
figures are very remarkable ; at the beginning the chromosomes usually
show a dumbbell-like shape, but they do not all divide simultaneously,
with the result that while those in the centre of the equatorial plate
still retain their position, those near the edge have divided and the
halves may have nearly reached their respective poles. The spindle
thus may appear almost covered with scattered chromosomes, which
finally all reach the poles ; a vesicular nucleus is then formed and the
cell divides (Fig. 17).

The second spermatocytes appear to begin their division almost
immediately; it is essentially like the first except that the cells,
spindles and chromosomes are distinctly smaller and the arrangement
of the latter even more irregular (Figs. 18, 19). At the close of the
division the cells become spermatids, with a vesicular nucleus and
conspicuous "Mitochondrion Korper," differing from those produced
by the large regular spermatocytes only in their smaller size (Fig. 20, a, h).

It will be seen that the process described is closely similar to that
found by Moves in Pygaera, except that a single spermatid nucleus
is formed instead of each chromosome forming a small separate vesicle.
Abraxas also differs from his description in the fact that the " normal "
spermatogenesis occurs chiefly in the late larval and early pupal stages,
and the small irregular spermatocytes are most conspicuous in the
rather later pupae. In my first sections of pupae several days old
I found no large regular spermatocyte divisions, and even in pupae
3 — 5 days old they are often quite scarce; while in larvae not yet
spun up some search may be required before the small irregular type
is founds According to Moves both types occur in about equal
numbers in the pupae of Pygaera.

^ I owe the suggestion that the regular type might occur more frequently in larvae, and
that the irregular type may be abnormal, to Prof. E. B. Wilson, who has kindly examined
some of my preparations.

L. DONC ASTER 183

The later development of the small spermatids appears to be
identical, as far as I have been able to follow it, with that of the
" apyrene " type in Pygaera as described by Meves. When the
developing spermatozoa become aggregated in bundles, it may generally
be seen that these are of two sizes ; in the larger the nuclei are at one
end, being transformed into the heads of the spermatozoa, but in the
smaller the nuclei are scattered along the tails of the spermatozoa and
appear to be degenerating. A few such degenerating nuclei may some-
times be seen in the large bundles (Fig. 21, a, 6). In teased prepara-
tions of the testes of the imago, the small spermatozoa in which the
nuclei have degenerated appear to remain in bundles, which suggests
that only the large nucleate ones are functional in fertilisation. This
is confirmed by sections of testes of the imago in which nearly all the
spermatozoa are of the small degenerate type, while in the vas deferens
nearly if not quite all are of the normal kind. Probably therefore the
degenerating spermatozoa do not leave the testis. In the imago the
degenerate spermatozoa are in the same stage as in the pupa ; I have
not seen the final stages leading to complete loss of the nuclei described
by Meves in Pygaera.

In conclusion, it should be said that the observations here described
apply equally to pure (wild) grossulariata, to the variety lucticolor,
and to the heterozygote between the two varieties. The formation
of " apyrene " spermatozoa is thus not connected with the lacticolor
variety, and my observations do not give any indication of the cause
of the phenomenon. The suggestion of Meves that " apyrene " sperma-
tozoa are capable of fertilising an egg but not of transmitting the
paternal hereditary characters is not borne out by breeding experiments,
nor do these confirm the suggestion that the two types of spermatozoa
determine diflFerent sexes in the fertilised egg.

I have only made a cursory examination of the mitoses in the
germ-cells of the female, but those in the egg-tubes (oogonia) of a
full-fed larva do not differ recognisably from the spermatogonia!
mitoses described above.

184 The Spermatogenesis of Abraxas

PLATE XXXIII.

EXPLANATION OF FIGURES.

All the figures except Fig. 21 were drawn with a Zeiss immersion apochromat. 3 mm.
N. A. 1-40, and Compens, Oc. 12.

Fig. 21, a, 6 was drawn with a 12 mm. objective.

Fig. 1. Spermatogonial mitoses.

a. Equatorial plate in face; about 54 chromosomes visible.
h. Metaphase in side view.

Figs. 2 — 9. First spermatocyte divisions, large "normal " type.

Fig. 2. Early prophase: chromosomes on nuclear membrane. The nucleus extends
through more than one section : only those in one section are represented.

Figs. 3 — 6. Equatorial plates in face. Fig. 3, wild grossulariata; Fig. 4, lacticolor;
Fig. 5, heterozygote. These three show 28 chromosomes. Fig. 6, heterozygote,
showing 29 chromosomes, probably owing to premature division of one.

Fig. 7. Equatorial plate, side view.

Figs. 8, 9. Early and late anaphases.

Figs. 10 — 12. Second spermatocyte division, large "normal" type.

Fig. 10. Equatorial plate, a. heterozygote. b. lacticolor.

Fig. 11. Equatorial plate, side view.

Fig. 12. Telophase.

Figs. 13 — 19. Spermatocyte divisions, small irregular type.

Fig. 13. Early prophase of first division ; the nucleus extends through about three
sections, one of which is represented. There were altogether over 50 chromosomes.

Fig. 14. Later prophase, one section of nucleus. Altogether 18 double and 10 or more
single chromosomes were counted.

Fig. 15. Equatorial plate of first division. 31 chromosomes appear to be present, pro-
bably owing to premature division.

Fig. 16. Same stage, side view.

Fig. 17. Anaphase of first division, showing scattering of chromosomes.

Figs. 18, 19. Second spermatocyte division, early and late anaphases.

Fig. 20. Spermatids in process of conversion into spermatozoa. Beconstruotions com-
bined from more than one section.

a. (Above) large "normal " type ("Eupyrene," Meves).

b. (Below) small abnormal type ("Apyrene," Meves).

Fig. 21. (Drawn with J in. objective.) Bundles of nearly mature spermatozoa.

a. (Above) normal " Eupyrene."

b. (Below) abnormal "Apyrene."

JOURNAL OF GENETICS, VOL I. NO. S

PLATE XXXIII

/W'

• ' V.K; \

*•=/>.

<tf*^

rfii.

'^^ -- —

<i^

•f* *

< >

o . o

THE INHERITANCE OF THE PECULIAR
PIGMENTATION OF THE SILKY FOWL.

By W. BATESON, M.A., F.R.S.
AND R. C. PUNNETT, M.A.

CONTENTS.

PAOB

Introduction 186

General statement of resnlts 186

Detailed results of the various crosses 191

(1) The Fi generation 191

(2) The F2 generation 192

(3) Fi X Brown Leghorn 194

(4) Fi X fully pigmented birds 195

(5) The fully pigmented F^ birds 197

(6) The ppii strain : a test of the hypothesis . . 197
The constitution of the Brown Leghorn hen .... 199

Exceptions 200

The grades of pigmentation 200

Silky crosses other than with Brown Leghorns . . . 202

Introduction.

During the past six years we have been engaged upon a series of
experiments connected with the inheritance of the peculiar pigmenta-
tion of the Silky Fowl. A brief account of the general features of this
interesting case has already been published by one of us\ but as our
experiments are now concluded we are able to give in greater detail
the evidence upon which our views are based. As a description of the
Silky Fowl may be found in any of the standard works on poultry it
is unnecessary for us to dwell upon the characters of the breed except
in so far as they enter into this particular series of experiments. One
of its most remarkable features is the extraordinary abundance of black
pigment which is generally distributed among most of the mesodermal
» W. Bateson, Mendel's Principles 0/ Heredity, 1909, p. 181.

186 Peculiar Pigmentation of the Silky Fowl

tissues of the body. Seen through the thin epidermis this pigment
gives the bird the appearance of a fowl with a black skin, deep purple
comb and wattles, and dark slaty shanks. The iris is heavily pigmented,
and the usually white earlobe takes on a more or less intense turquoise
tint which is especially noticeable in the hens. The somatic peritoiieum,
the periosteum and pia mater are inky black from the pigment with
which they are crowded. The splanchnopleure is much less pigmented,
and the liver seen through this looks its normal colour. The muscles
also have a blackish appearance, but we have not made any histological
examination to determine the exact distribution of the pigment here.
It is apparently confined to membranes of mesodermal origin, and is
absent from the lungs, liver and other viscera, while at the same time
the plumage is white. There is never any trace of it in the allantois,
or other foetal membranes.

Our experiments with this breed were begun with the idea of
investigating the nature of the form of comb by which it is characterised,
but we had not proceeded far before it became evident that the in-
heritance of the peculiar pigmentation promised more interesting and
novel results. As the case presents features unlike any hitherto met
with elsewhere it will be convenient if we give a brief outline of the
main results and of the interpretation before passing to a detailed
examination of the experimental data.

General statement of results.

The breed with which the Silky was originally crossed was a strain
of Brown Leghorns which had been in our possession since 1899. The
first indication of something unusual was the definite difference found
in the reciprocal crosses between these two breeds. While the mating
of Silky % X Brown Leghorn j/* resulted in chicks of both sexes with
little pigmentation, the mating of Brown Leghorn $ x Silky f^ gave
a markedly different result. From this mating the (/• chicks had only
a little pigment and were indistinguishable from those resulting from
the previous cross, but the % chicks were all deeply pigmented, differing
but little in this respect from a pure Silky ^ On breeding the F^ birds
together there resulted in either case an F2 generation consisting of

1 The Fi chicks all had coloured plumage and subsequent breeding showed that the
white of the Silky behaved as a simple recessive (cf. Rep. Evol. Comm. Roy. Soc. iv. 1908).
Our experiments have led us to infer that the pigmentation is quite independent of the
colour of the plumage.

W. Bateson and R. C. Punnett

187

chicks of various grades of pigmentation ranging from the deepest
pigment to none at all. The F^ generation however presented certain
distinctions according as a pigmented, or a non-pigmented ^i hen was
used (see p. 193).

Brown Leghorn Silky

Br.LO X

cTbp.l.

1 I r

1 1

Fig. I.

Silky Brown Leghorn
#" X ^

Br. L.C

(5f

CfBr.L

I — I \ — I — f I — I — \ — I — I — I I — r

Fig. 2.
In these figures
(J O represent un pigmented birds.

deeply pigmented birds.

birds with some grade of pigmentation other than the
deeply pigmented type.

The F^ birds were also crossed with the pure Brown Leghorn and
the results were strikingly diflferent according as the sex of the ^i was
male or female. When the F^ $ was crossed with the Brown Leghorn (^
none of the offspring were deeply pigmented, and this was true for the

188 Peculiar Pigmentation of the Silky Fowl

pigmented F^ ? as well as for the unpigmented. But when the ^i j/"
was crossed with a Brown Leghorn $ about one in eight of the offspring
were deeply pigmented and these were alivays females. To assist the
reader in following this somewhat complicated case we append a rough
scheme. It will be understood of course that the scheme gives no
indication of the proportions in which the various classes are produced,
neither for the moment do we attempt to differentiate between the
various grades of pigmentation other than the fully pigmented state.

We may now state briefly the interpretation to which our various
experiments have led us. We consider that three factors are involved
of which two are directly concerned with the degree of pigmentation.
These are (a) a pigmentation factor (P), and (/S) an inhibition factor (/)
which can prevent the full development of the pigmentation ^ The
various grades of pigmentation met with depend upon the various
compositions of the zygotes in regard to these two factors ; e.g. a bird
of the constitution PPii will be fully pigmented, a bird of the con-
stitution Ppli will be slightly pigmented, while birds of the constitu-
tion ppll, ppli, or ppii will be unpigmented (see also p. 200).

The third element with which we are concerned in these experi-
ments is sex. Here we have made certain assumptions. We regard
the female as differing from the male in possessing a special element,
F, of which the hereditary behaviour is like that of any other
Mendelian factor. Moreover we consider that the female is always
heterozygous for this factor so that the zygotic constitution of a female
is Ff while that of a male is ff. Further we suppose that in such
zygotes as are heterozygous for both F and / there occurs a repulsion
between these two in gametogenesis so that F and / do not pass into
the same gamete. We may allude to the cases of the inheritance of
the lacticolor variety of Abraxas grossulariata^ and of the red eye of
cinnamon canaries' in which similar phenomena can be shown to follow
the same system of descent.

It must be expressly stated that the suggestion that females are
heterozygous for femaleness is offered without prejudice as to the
possibility that males may also be heterozygous in maleness. The
systems followed by the descent of colour-blindness* in Man and by

^ The condition of the gamete from which either or both of these factors are absent we
shall denote in the conventional way by the use of the corresponding small letters p and i.
^ Doncaster, L., Reports to the Evolution Committee of the Royal Society, iv. 1908.
3 Durham, F. M., Reports to the Evolution Committee of the Royal Society, iv. 1908.
* Mendel's Principles, 2nd imp., 1909, p. 195, note.

W. Bateson and R. C. Punnktt

189

that of the white eye recorded by Morgan in Drosophila^ clearly point
to the existence in those cases of a repulsion between a factor for
maleness (M), and factors respectively for colour-blindness and for the
red eye. The operation of the system of sex-limitation is similar in all
these examples, the only difference being that in the one group the
repulsion is from the factor F, in the other from the factor M.

iiPPf

gives
gametes

iPf]

[9]

lippFf

gives
gametes

\ipF

[9]Iippff

gives
gametes

^Pl\

ipF]

lpfi\ fpf^

fpf6

ipF

Jpf9

Ipf

iPfi

ipF
iPfi

ipF

^Pfi

I I

U^iiPpS iiPpFfii^

gives
gametes

[ IPf\
\ipf J

gives
gametes

IP/

iPF^

IPf

iPfi

IPf IPf

ip'F'% 1 ipf i

iPF^

Ipf
ipF^

^pfS

iPf
iPF i

\pfi

iPf iPf
ipFf ipfi

f/z-f

ipf \ ipf
ipf 9 \ ipf 6

lippffU\

gives
gametes

Ipf

iPF%

Ipf

ipf S

Ipf
ipF^

Ipf
ipf 6

Fig. 3-

Recognition of the existence of factors both for femaleness and for
maleness of course involves the assumption that ova bearing F can only
be effectively fertilised by sperms not bearing M, and vice versa. For
that supposition no independent evidence yet exists, and we note that

1 Morgan, Science, 1910, N. S. xxxii. p. 120.

190 Peculiar Pigmentation of the Silky Fowl

Morgan^ has made observations on Cumingia (Mollusca) distinctly
unfavourable to it. At present however we think it is the most
acceptable account of the facts ascertained both as to the heredity
and the variability of sexual characters.

Ilppff

gives
gametes

[t]
iiPPFf

gives
gametes

iPF

iPf

Bp. Legh.

[^]IippFf

gives
gametes

lpf\

ipFj

U]IiPpff liPfFfi^^

gives
gametes

i IPf]
Ipf

iPf
ipf

gives
gametes

Br. Legh.

Ilppff 16}

gives
gametes

ilpf

Ipf

--V

?//.

ipF
IPf ^

IPf

Ipf $

IPf

Ipf$

IPf
iPF^

IPf

ipF ^

'^fS

IpF

Ipf9

Ipf
IPf$

Ipf
iPF^

Ipf
ipF^

Ipf

iPf ^

ipF
iPff

iPf
IPf S

iPf

Ipf $

iPf
iPFf

iPf

ipF i

ipf 6

ipF
ipf 9

fpfi

%f6

ipf
iPFf

ipf
ipF ?

Ipf
ipF^

Fig. 4.

The Silky Fowl normally breeds true to the fully pigmented con-
dition and we consequently represent the cocks and hens of this breed
as ffPPii and FfPPii respectively. The Brown Leghorn on the other
hand never produces pigmented birds and we therefore regard it as
being entirely without the factor P. But it possesses the inhibitor
factor 7; and for reasons which will appear later the cock must be
1 Morgan, Payne, and Browne, Biol. Bull. 1910, xviii. p. 76.

W. Bateson and R. C. Punnett 191

normally homozygous and the hen heterozygous for this factor. Con-
stitutionally therefore we look upon the cocks and hens of this breed
as being respectively ffllpp and Ffllpp. To illustrate what we
imagine to happen in the several generations produced by mating
a Silky ff with a Brown Leghorn % as well as in the reciprocal cross
we have drawn up the appended schemes (Figs. 3 and 4) for comparison
with Figs. 1 and 2. These schemes also indicate the composition on
our hypothesis of the generations shown ar)d we may now proceed to
test their validity by the facts witnessed in these and other forms of
mating.

Detailed results of the various crosses.

1. The F^ generation.

(a) From the Silky %.

[Nature of mating FfPPiixffppII.]

We have bred from Silky $ x Brown Leghorn j/" on two occasions
and ill neither case had any of the chicks more than a slight amount
of pigment (cf. Fig. 1). Many of these were reared and in the adult
state were almost indistinguishable in general appearance from pure
unpigmented birds. Careful examination however revealed traces of
pigment as patches either on the wattle, skin, or shanks. In most
cases the presence of some pigment was most readily detected beneath
the skin in the periosteum of the femoro-tibial or of the tarso-metatarsal
joints. Dissection showed that some pigment was nearly always present
in the ribs and in the occipital region of the skull. There was frequently
a little peritoneal pigment more especially in the region of the ribs
and some in the occipital pia mater. The amount of pigment varies
somewhat and may be very slight. In some cases the chicks are
recorded in our notes as being without pigment, but most of these

TABLE I.

Reference

Males

K»tare of mating Full

Some None

FuU

Some None

1905

Pen 16, 349

SUky 9 X Br. L. <f —

—

1907

„ 7, 495

—

1909

„ 7, 150

^2? X „ —

—

192 Peculiar Pigmentation of the Silky Fowl

records refer to birds which were not killed and critically examined.
The extent to which the pigment development is inhibited exhibits
individual variation, and it is likely that in some birds the inhibition
is so complete that they are indistinguishable from birds which lack
the pigmentation factor. Here we may mention also that we bred
from a fully pigmented F^ hen ($ 150, see p. 197) with results similar
to those which we obtained from the Silky hens.

(y3) From the Silky </.

[Nature of mating Ffpp li x ffPPii.']

Our original Silky f^ was mated at different times with two pure
Brown Leghorn hens. The F^ males from this mating were precisely
similar to those produced from the reciprocal cross. The F^ hens
however were nearly as deeply pigmented as the Silky (cf Figs. 2 and 4).
To the one exception, a slightly pigmented $, we shall return later
(p. 200).

TABLE II.

Males Females

Reference Nature of mating Full Some Hone Full Some None

1905 Pen 51, 404 \

1906 „ 18, 404/

1908 „ 18, 2811

1909 „ 18, 281 J

Br. L. ? X Silky i — 8 8

,. .. — 29 31

2. The ^2 generation.

(a) From the Silky ^.

[Nature of mating FfPpiixffPpIi.']

In Figure 3 we have already indicated the results which are to be
expected from this form of mating. One quarter of the total offspring
should be completely unpigmented while the remainder should be
equally divided between the fully pigmented and the partially pig-
mented classes, the expected ratio being three fully pigmented, three
partially pigmented, and two unpigmented out of every eight birds.
Moreover the ratio should be the same for each sex. In Table III the
results of six such matings between F^ birds are given. There is a
slight excess of fully pigmented $ % due to the unusually high pro-
portion of such birds in one of the matings (1909, Pen 4, 329), but on
the whole the facts are in close accord with expectation.

W. Bateson and R C. Punnett

193

TABLE

III.

Reference

Nature of m&ting

Males

Females

FuU

Some

None

FuU

Some

None

1907

Pen

15,

283

Fi ? (full pig.) xFiS

23,

114

1907)
1908(

22,

121

1908

467

—

1909

329

22,

148

1909)
1910)

20,

374

. 15

Total

Expectation

52 ■ 3

52-5

62-5

(y8) From the Silky ?.

[Nature of mating, FfPpIixffPpIi.^

As indicated in Figure 4 expectation is here different from that in
the preceding case where the F^ $ was from the cross Browu Leghorn %
X Silky f^. The slightly pigmented F^ $ is here heterozygous for the
inhibitor factor, /, and there comes ioto play the repulsion between
/ and F so that all the male gametes produced by such birds contain /,
while this factor is carried by none of the female gametes. From this
mating therefore we should not expect any fully pigmented males since
every bird of this sex must contain /. Nevertheless, as the data in
Table IV show, many of the males resulting from this mating were recorded
as being heavily pigmented. By far the greater number of chicks in
this generation were killed and recorded on hatching, and a peculiarity
of the f^ ^ booked as fully pigmented lay in the fact that the toes of
these birds were always light in colour. We regard these birds as of

TABLE IV

Reference

Nature of mating

Males

Females

FuU

Some

None

FuU

Some

None

1906

Pen 9,

467

Fi ? (unpig.)

xFi<f

—

1907

„ 11,

459

„ 20,

461

1908

„ 19,

459

„

—

,. 19,

393

., 22,

467

—

Total

Expec

tatum

18-75

37-5

1875

194 Peculiar Pigme7itation of the Silky Fowl

the constitution ffPPII or ffPPIi and suppose that in the presence
of a double dose of the pigmentation factor the effects of the inhibitor
are in considerable measure overcome in the younger stages. In cor-
roboration of this view we may state that we reared several of these
deeply pigmented ^^ and that they all became far less heavily pig-
mented in appearance as they approached maturity. In external
appearance indeed they showed little more pigmentation than the F^
cocks. This explanation is the natural one if we regard the constitu-
tion of the slightly pigmented F^ % as FfPpIi, and further evidence
(p. 198) in favour of this view will be adduced from an entirely
different set of experiments.

3. fj X Brown Leghorn.

(a) Brown Leghorn $ x F^ (f.

[Nature of mating, FfppIi x ^Ppli.]

On our hypothesis this form of mating should give a specific result,
for while the ^^^^ should all be either without, or with comparatively
little, pigment, one quarter of the $ $ should be fully pigmented
(cf. Figs. 3 and 4). We have bred a considerable number of birds
(nearly 700) in this way, and the figures given in Table V show that
this expectation is closely realised. None of the 336 (^(^ produced

TABLE V.

Males Females

Reference Nature of mating FuU Some None Full Some None

1906 Pen 9, 207 Brown ? x i^^i <? — 28 8 29

,, 33, 248

„ 33, 159

1907 „ 11, 203

„ 12, 264

,, 12, 159

8 11

13 5 18

34 8 24

13 3 8

19 3 18

„ 15, 347 Br.L. ?xFic? — 34 11 37

„ 16 ? ?Br.L. Brown? xFiS — 18 2 14

„ 20, 129 „ ' — 22 7 19

„ 22, 101 „ _ 42 8 28

„ 23, 98 „ — 24 7 18

1908 ,, 5, 101 „ _ 17 3 20
„ 16, 345 Br. L. SxjPjc? — 20 7 16
,, 22, 129 Brown ?xFi<f — 40 8 31

1909 „ 20, 347 Br.L.^xFiJ— 4 1 4

Total — 336 82 280

Expectation — 336 90-5 271-5

W. Batkson and R. C. Punnett 195

were deeply pigmented, while of the 362 $ $ 82 were deeply pig-
mented, a proportion approximating fairly closely to the expected
quarter. We should add that owing to a deficiency of pure Brown
Leghorns some of the hens used were light-shanked brown birds of
Brown Leghorn extraction. With regard to the transmission of pig-
mentation these 4>ehaved similarly to the pure race.

(^) ^1 ? (unpigmented) x Brovm Leghorn ^.

[Nature of mating, FfPpIi xffppll.]

Two -F, $ $ of this nature were crossed with a Brown Leghorn (^
and gave 26 j/'j/' and 18 $ $ of which none were deeply pigmented.
This again fits in with our hypothesis (cf Fig. 4), for no deeply
pigmented birds are to be looked for from this mating.

4. fj X Fully pigmented (PPii) birds,
(a) F,^xPPii%.

[Nature of mating, FfPPii x ffPpIi.]

The expected result from this form of mating is equal numbers, in
both sexes, of chicks with deep pigmentation and of chicks with some
pigmentation. We have made this mating twice with the following
results :

TABLE VL

* Hales Females

Full pig- Some pig- Full pig- Some pig-
Reference Nature of matiiig mentation mentation mentation mentation

1906 Pen 33, 349 Silky ?xFi<r 5 2 3 5

1909 „ 17, 114 i- 2 fall pig- ? X F, <r 20 18 19 22

Total 25 20 22 27

Expectation 235 235 23-5 23'5

These results are obviously in close accord with expectation, but it
must be mentioned that $ 114 also gave one $ chick recorded as
without pigmentation.

()8) jPi % {slightly pigmented) x Silky ^.

[Nature of mating, %^PpIi x ^^PPii.]

Since on the hypothesis the gametes produced by the Fi% are %Pi,
%pi, (^PI, and ^pl it follows that all the female chicks will contain
P without /, while of the male chicks all will be heterozygous for /
while half will be homozygous for P. La discussing the nature of the

Joum. of Gen. i 14

196 Peculiar Pigmentation of the Silky Fowl

F^ generation from this type of ^i $ we have already seen reason for
supposing that the male chicks of the constitution PPIi are practically
fully pigmented on hatching, but that the pigment becomes much
reduced with advancing age. Hence the expectation for the present
type of mating is that all the % $ will be deeply pigmented, and that
the (/"(/ will hatch either as deeply pigmented chicks, or as chicks
with some pigment only — these two classes being produced in equal
numbers. Table VII gives the results of the two cases in which we have
made this mating. While the general result is in accordance with
expectation the Table shows that there are two $ % from each mating
which are not fully pigmented. To these exceptions we shall recur
later (p. 200).

TABLE VII.

Males Females

Reference Nature of mating Full Some None Full Some None

1907)

^g^g Penis, 459 Fj ? (slight pig.) x Silky <? 12 13 -

28 2 —

1907)
1908}
1909 „ 16, 467

1908f " ^' ^^^l „ ., 12 18 — 28 2 —

Total 24 31 — 56 4 —

Expectation 27-5 27-5 — 60 — —

(7) Fi % {fully pigmented) x Silky ^.

[Nature of mating, % (^Ppiix ^ ^PPii.]

Since the gametes of neither parent carry the inhibitor factor and
since those of one parent all contain the pigmentation factor, the
expected result of this mating is fully pigmented chicks only, of both
sexes. The mating has been made on three occasions and as Table VIII
shows the results are in accordance with expectation.

TABLE VIIL

Males

Females

Keference

Nature of mating

FuU pig-
mentation

Full pig-
mentation

1907 Pen 18,

121

Fi ? full pigmentation x Silky <?

1907 )

1908) " ^®'

114

»» »»

1907 )

1908 / "

283

»» f>

We have already alluded to the deeply pigmented hens which
resulted from crossing the F^ ^ with the Brown Leghorn % . On our

W. Bateson and R. C. Punnktt 197

hypothesis these birds are in constitution FfPpii and consequently
should give the same result as the deeply pigmented ^i % when crossed
with a pure Silky ff. We have made this cross on two occasions and
in accordance with expectation all the chicks were deeply pigmented
(cf. Table IX).

TABLE IX.

Males

Females

Keference

Nature of mating

Fun pigmentation

Full pigmentation

1907 Pen 18,

344

?Ppux Silky ,f

1907 „ 9,

376

l» u

5. Crosses with deeply pigmented F^ birds.

In the course of our experiments we have made crosses with two
deeply pigmented F^ birds, viz. f^ 40 (from Pen 15, 283 of 1907) and
% 150 (from Pen 23, 114 of 1907). Each of these birds was as deeply
pigmented in the adult stage as the pure Silky, and when bred together
they gave only fully pigmented offspring (12 ^f^ and 1\ % %). <^ 40
was also mated with a pure Brown Leghorn $ and gave 2\ ^^^ with
some pigment together with 33 deeply pigmented % $. But he is
recorded as giving also one deeply pigmented ^ and 2 $ $ which
were not deeply pigmented. To these exceptions we shall return
and will merely state here that we regard them as due to a peculiarity
in the behaviour of the Brown Leghorn hen. $ 150 behaved like a
pure Silky when crossed with a Br. L. {/• (p. 192), and we look upon
both these F^ birds as of the constitution PPii.

6. The ppii strain.

In this account we have so far been concerned with the results of
various crosses between the Silky and the Brown Leghorn breeds. By
a happy accident we are able to adduce independent and cogent
evidence in favour of the interpretation which we have put forward.
In 1907 we bought a Silky ^^ which proved to be heterozygous for P
(i.e. Ppii). Mated with an Eg}'ptian hen, a brown bird with light
coloured shanks, he gave unpig merited as well as deeply pigmented hens.
Two of these unpigmented birds were mated back to the heterozygous
Silky f^ in 1908 and as was expected gave deeply pigmented and
unpigmented birds of both sexes, viz. 18 j/'j^ deeply pigmented, 15 <^ff
unpigmented, 21 %% deeply pigmented, 17 %% unpigmented. In
this way we were able to establish a strain of birds containing neither

14-8

198 Peculiar Pigmentation of the Silky Fowl

the pigmentation nor the inhibiting factor in either sex. These birds,
on our system of notation, must be represented as ppii, and the
possession of such a strain provided us with the means of testing
the constitution of the ^i (Silky x Brown Leghorn) birds in the
simplest and most direct way.

The ^1 f^ on the hypothesis produces four kinds of gametes in
equal numbers, viz. fPI, fPi, fpl and fpi. Crossed with Ffppii such
a bird should give in both sexes equal numbers of birds with and
without pigment. Again among the pigmented birds there should
be equal numbers of deeply pigmented birds, and of birds with a small
amount of pigment only. Table X shows that these expectations were
closely realised in fact.

TABLE X.

Males Females

Reference

Nature of mating

Full

Some

None

Full

Some

None

1909

Pen 4,

408

ippiixF^^

—

„ 17,

274

„ 22,

261

„

Total

Expectation

We have also made a similar set of experiments to test the gametic
output of the slightly pigmented Fi $ (ex Silky $ x Br. L. (^). The
constitution of such birds on the hypothesis is FfPpIi and owing to
repulsion between F and / the gametes produced are of four kinds
only, viz. FPi, Fpi, fPI, fpl (cf p. 188). Mated with c/c/ of the
constitution ppii such birds should give equal numbers of pigmented
and unpigmented chicks in both sexes. And since the female gametes
which contain P all lack the inhibiting factor, all the $ $ pigmented

TABLE

Nature of mating

Males

Females

Reference

Full

Some

None

Full

Some I^one

1909 Pen 3,

467

Fj (slight pig.) ? X sppii

—

1 4

^^^^l 13
1910 i " ^^'

459

» >>

—

2 12

1910 1 " ^^'

393

„ >,

—

— 21

1910 „ 22,

4? ?

^ Ppii X ippii

1 56

Total

102

101

4 93

Expectation

— 99

W. Bateson and R. C. Punnett 199

at all should be deeply pigmented. On the other hand all the male
gametes of the F^ % which contain P contain also /, and consequently
none of the pigmented ^/'j/' produced should be deeply pigmented.
Table XI which gives the details of four such experiments shows how
closely this expectation is realised, and offers strong corroborative
evidence of the view here taken of the nature of the gametes produced
by this type of jP, %. The five exceptions recorded we shall refer to
again (p. 200).

The Constitution of the Brown Leghorn Hen.

While the Brown Leghorn </ is homozygous for the inhibiting factor,
the % is on our hypothesis always heterozygous for this factor. And
since we assume repulsion to take place during gametogenesis between
the factors F and / it follows that she produces two kinds of gamete,
viz. Fpi and fpl. The possession of the ppii strain enabled us to
devise a pretty experiment to test this view. By mating a Brown
Leghorn $ with a cock of the constitution ^pj9u we obtained a number
of unpigmented chickens of both sexes. On our hypothesis only the
f^ (^ should receive the inhibiting factor, being in constitution ffppli,
while the % % should be F/ppii, and consequently lacking the inhibiting
factor. This ditference between the sexes with regard to the inhibiting
factor should be brought out by a cross with fully pigmented homozygous
birds (PPii), for while the females should give only fully pigmented
chicks, the males may be expected to produce fully pigmented and
partially pigmented chicks of both sexes in equal numbers. During
the present year a cockerel (ex Br. Leg. ? x ppii ^) was mated with
a pure Silky hen, and four sister pullets were put with an F^ fully
pigmented ^ (No. 40, ex Pen 15, 283 of 1907) already shown to be
PPii in constitution. The results are shown in Table XII and are in
accordance with expectation.

TABLE XII.

Males TemmlM

=3 I § 5 I §

Reference Kstore of mating h dS S; ^ oS ;:;

1910 Pen 8, 150 Silky 9 x <? (ex Br. L. ? x jjpti <r ) 14— 33 —

,, 24, 4? 9 ? ? (ex Br. L. ? X ppii <r)x <r PPii 18 — — 26 — —

200 Peculiar Pigmentation of the Silky Fowl

Exceptions.

In our account we have mentioned certain exceptions which occurred
in several of the various matings. These are :

Table II, p. 192 . ex Br. L. ? x Silky i 1 ? partially pigmented

p. 197 . ex Br. L. ? x FPU , , F, \^' ' r^ -'"^ "^T""'"^

( 1 cf fully pigmented

Table VII, p. 196 . ex F^ {Pplij ? x Silky s 4 ? ? partially pigmented

Table XI, p. 198 . ex F, (Ppli) ? xppii ^ j ^ ' / T'"*"^ Pigmented

( 1 (f fully pigmented

In all these cases the $ $ should have been fully pigmented and
the fff^ should have been partially pigmented on our hypothesis. It
will be noticed that wherever these exceptions occurred the mother
was a bird heterozygous for both F and /. These cases raise the
question whether the normal repulsion between F and / in such birds
may not occasionally break down, and whether in addition to Fi srndfl
gametes they may not produce FI and the complementary fi gametes.
This appears the more likely as in two out of the four cases a fully
pigmented (^ also appeared as an excej>tion ; and in Table VII even
if such birds appeared they would not be noticed, since fully pigmented
f^f^ are one of the classes normally produced from the mating of
slightly pigmented F^ $ and the Silky </. We incline therefore to
think that upon occasion the repulsion between factors may be im-
perfect, though whether this imperfection is sporadic, or whether it
can be conceived as part of some orderly scheme we do not yet know
enough to say.

The Grades of Pigmentation.

The dependence of pigmentation upon the presence or absence of
two factors (P and /), as well as upon the heterozygous or homozygous
condition of the individual with regard to either or both of them,
would naturally lead the observer to look for a considerable range of
variation in the pigmented condition. For in the full zygotic series
are the nine possible combinations, FPU, Ppii, PPII, FPU, PpII,
Ppli, ppll, ppli, ppii. The great majority of the chicks with which
we dealt in these experiments were killed and recorded on hatching,
and our practice was to refer them in so far as pigmentation was
concerned to one of the following grades, viz. none, faint, slight, some,
moderate, much, full, very full. Though not corresponding accurately
to the various zygotic constitutions, these empirical grades nevertheless

W. Bateson and R. C. Punnett 201

aflford some indication of them. Where P is not present the bird is
always unpigmented, though with regard to / it may be either //, It,
or n. Where / is absent the bird is nearly always fully pigmented
whether homozygous or heterozygous for P, though it is probable that
chicks recorded as with much pigment may sometimes have been in
constitution Ppii. The birds classed as " very fully " pigmented were
probably in most cases PPii though sometimes they may have been
exceptionally deeply pigmented birds of the constitution Ppii. Where
both P and / are present some pigment would appear to be always
present though the amount is subject to fluctuation. Thus F^ birds
of both sexes (ex Silky $ x Brown Leghorn </), and the ^ birds
(ex Brown Leghorn % x Silky j/") are of the constitution PpIi, but
in respect of the intensity of their pigmentation they might belong to
either of our three classes "slight," "some," or "moderate," and our
experience has been that these classes grade very much into one
another. Birds with " much " pigmentation are in general either
PPII or PPIi, though an occasional bird of the Ppii class might
be referred to this group. The class PpII is doubtless to be found
among the birds with "faint" or "slight" pigmentation.

The grade of pigmentation would also appear to diflfer somewhat in
the two sexes, for among birds similarly constituted for these two
factors P and / the females are generally a little more pigmented
than the males\

This case of the Silky pigmentation is interesting in connection
with the production of intermediate forms. In an F^ family bred from
Silky $ X Brown Leghorn </ all the nine possible zygotic combinations
of P and / occur in one or other sex. It would be possible to choose
birds of such breeding and to arrange them in a series exhibiting
continuous gradation from full pigmentation to none at all. Yet we
now know that such a series is due to the interaction of three definite
factors (inclusive of the sex factor), and that the continuity in variation
manifested is in reality founded upon a discontinuous basis. Moreover
we may point out that the mating of partially pigmented males of the
constitution PPIIff with partially pigmented females of the constitu-
tion PPIiFf would result in the establishing of a race breeding true
to an intermediate condition of pigmentation in spite of the underlying
discontinuity involved.

' This fact is interesting in connection with the common experience of fanciers that
in black-feathered breeds which have yellow skins, it is easy to obtain males with clear
yellow shanks, bat the females almost always have some black pigment in the shanks.

202 Peculiar Pigmentation of the Silky Fowl

Silky crosses other than with the Brown Leghorn.

During the course of our experiments we have crossed the Silky
with other fowls beside the Brown Leghorn, but as the crosses with
the last named promised the most definite results our attention and
resources were mainly devoted to these. We may however mention
a few points of interest which have arisen in connection with some of
the other crosses.

Our original Silky f^ was mated in 1906 to a white Rosecomb
bantam. All the chicks (5 (/cT and 7 $ $) were deeply pigmented
on hatching though as they reached maturity the pigment became less
marked in the cockerels. A few cases are already on record in which
a Silky was crossed with another breed and all the resulting offspring
of both sexes were deeply pigmented'. It is worthy of note that in
such cases the breed with which the Silky was crossed possessed dark
shanks. This was certainly so for the Spanish used by Tegetmeier and
Darwin as well as for the Rosecombs used by ourselves ; and we infer,
though this is not explicitly stated, that it was also true for the frizzled
fowls used by Davenport.

We may mention two cases from our experiments which are con-
sistent with this view. When a Silky ^ was mated with a dark-
shanked mongrel ? {F.^ ex W^hite x Brown Leghorn) 2 out of the 13
male chicks produced were fully pigmented. The remaining 11 male
chicks exhibited a varying amount of pigment, while all the 11 female
chicks showed the full pigmentation (1906, Pen 18, 150). In the other
case an F^ ^y ex Silky $ x Brown Leghorn (/, was also crossed with
a dark-shanked mongrel Leghorn hen bred similarly to the last (1906,
Pen 9, 604). Out of the 19 male chicks from this mating two were
deeply pigmented, while with light-shanked hens the cock gave the
usual result (cf. p. 194). We must suppose therefore that the factor
or factors upon which shank pigmentation depends can influence the
factors concerned with the development of the pigment found in the
Silky fowl, but at present we do not know sufiicient about the nature
of these factors to make any more definite statement.

Though our experiments have led us to infer that the strain of
Brown Leghorns with which we worked was homogeneous in respect
of the factor modifying pigmentation we nevertheless have evidence

^ Cf. Tegetmeier, The Poultry Book, 1873, p. 268 ; Darwin, Animals and Plants, 2nd
edit., 1899, p. 253 ; Davenport, Inheritance in Poultry, 1906, p. 60.

W. Bateson and R. C. Punnett 203

suggesting that this is not necessarily the case for all light-shanked
birds. An example may serve to illustrate our meaning. During
1008 and 1909 the fully pigmented F^ ^ mentioned on p. 197 was
crossed with a Brown Leghorn % and gave a typical result, viz. slightly
pigmented ^^ and fully pigmented % %. During both of these seasons
he was also run with a light-shanked % belonging to our recessive
white strain ^ With her he gave 19 male chicks varying from slight
to moderate pigmentation, but of the 18 female chicks 8 were fully
pigmented and 10 showed only a slight to moderate amount of pigment
(1908-9, Pen 24, 53). From this and other similar experiments it
seems natural to infer that some light-shanked hens may carry other
factors capable of modifying the Silky pigmentation besides that which
we have been able to demonstrate in the Brown Leghorn.

Lastly we may refer to a cross which we made between our original
Silky cock and a hen which was homozygous for the dominant white
factor (1907, Pen 18, 397). All the offspring (18 ^^ and 22 ? ?)
showed some pigment, sometimes a good deal, and this as a rule was
distributed in small irregular patches, but we were unable to notice
any diflference between the two sexes. We think it not unlikely that
the hen used was potentially a dark-shanked bird, and that the
offspring of both sexes would have exhibited full pigmentation had not
its development been in some way checked by the dominant white
factor. The results however were complex and lack of opportunity
prevented us from following up the cross, but we have thought it worth
placing these cases on record since they indicate that radical differences
in constitution may exist among light-shanked birds, and that the
behaviour of our strain of Brown Leghorns with regard to the Silky
pigmentation is not necessarily typical of birds with unpigmented
shanks.

1 An aocoant of the origin of this strain will be found in Reports to tht Evolution
Committee of the Royal Society, in. p. 19, rv. p. 28.

STUDIES IN INDIAN COTTON.

By H. M. LEAKE.
CONTENTS.

PAOK

Introdaction 205

The genns Gossjpinm and the types used in the investigation . . 208
The experiments :

(a) The colour of the corolla 212

(ft) The red colouring matter of the sap 214

(d) The type of branching and the length of the vegetative period 230
(«) The leaf glands 238

Correlation • . . . 241

liiterature 243

Tables I— XXIX 244

Introduction.

Cotton forms one of the main crops of large tracts throughout
India and is consequently of considerable agricultural importance.
The fibre of the majority of the forms found under cultivation
is, however, very poor and in a few cases only of sufficient quality
to find a market in England. The consumption is chiefly local and
an important industry has arisen with numerous mills the bulk of
whose out-turn is coarse yarn and cloth for which a considerable
demand exists. The problem of improvement in the quality of the
raw product is one which has exercised the minds of numerous in-
vestigators throughout India for nearly a century and was referred to
the author as the problem of most pressing importance when he entered
Government service in IGO-i. The experiments were commenced at
Saharanpore in 1905 when a series of the Indian forms were first grown
and observed and they have been pursued without interruption first at
that place and later as part of the work carried on by the Research
Section of the United Provinces' Agricultural Department at the

206 Studies in Indian Cotton

Cawnpore Station. Though the practical conclusions have of necessity
throughout received detailed attention the wider aspect has not been
neglected. The broadest interpretation, in fact, has been placed upon
the subject in the belief that by such means only can the breadth of
view be obtained which is essential to that comprehensive understand-
ing of the group of types under experiment which alone will lead to
success in practice. With this aim in view the range of the indigenous
forms has, as far as possible, been determined, the various types
isolated and grown in pure culture, and crosses made between them.

The fact that the ultimate goal of the experiments is the improve-
ment of the forms generally cultivated has nevertheless imposed certain
restrictions which it is necessary to review here. The object is ex-
clusively an improvement of the forms grown in the United Provinces.
These Provinces are characterised by a comparatively severe winter, of
a severity sufficient not only to check all growth in the cotton plant
but to render all previously-formed branches incapable of flower produc-
tion. Before this can occur, a considerable amount of fresh growth
must take place and, by the time flowers commence to form, the brief
temperate period has given place to a summer so intensely hot and dry
that little or no fruit is set. Forms, therefore, such as are commonly
cultivated in the milder districts of Southern India have been grown
only with considerable difficulty. It has been found practically impos-
sible to isolate pure types of these and in many cases the only record is
one of complete failure to pass from one generation to the next. A full
investigation of such forms can only be accomplished in a climate more
suited for their cultivation.

During the past few years there has been frequent reference in
India to " plant-to-plant " selection as a means of improving the quality
of the staple. This term " plant-to- plant selection" is one which has
received extended application in India and is there used to denote that
selective process by which the crop is grown from the seed of definite
selected plants. Fertilisation is allowed to take place naturally and
the effects of possible cross-fertilisation are disregarded. It is a method
which has been advocated on the assumption that cross-fertilisation
does not occur in nature — a view that has been maintained by Gammie
(8 and 9). On the other hand observations to the contrary have been
made by Balls (1) working in Egypt on a different series of types and in
India by Burkill (4), Fyson (7) and the author (11). Also throughout
their work both Middleton (13) and Watt (19 and 20) constantly indi-
cate their belief, not only that natural crossing takes place, but that

H. M. Leake 207

certain of the races recognised by them have directly arisen by the
intercrossing of other extant races. It is not proposed to enter into
this question here, and it will suffice to say that abundant proof that
crossing is of frequent occurrence has been forthcoming since the
author's first note (11) on this subjects Under the circumstances,
however, a word as to the procedure adopted in the present experiments
is necessary.

The seed received from all sources has invariably given a crop con-
taining numerous, and frequently most diverse, forms. The seed of
those appearing recognisably distinct is collected and sown separately'.
From the similarity or dissimilarity of the offspring the purity or the
reverse of the parent can be determined. If the parent appears to be
pure the most typical individuals among the oflfspring are selected and
the flowers of these are protected. The form is, in future generations,
raised from the seed of flowers thus self- fertilised only. The unit of the
parental series on which these experiments are based is, therefore, a
series of individuals derived by repeated self-fertilisation through a
greater or less number of generations, from a single individual. Such
a unit may form the sole representative of a tjrpe but, in the majority
of cases, the type, as defined, includes several such units usually dis-
tinguishable by some small difference in one of the characters. The
unit is, therefore, comparable to the " pure line " of Johannsen.

In all cases when it is proposed to make a cross, the flower of the
seed parent is emasculated in bud after removal of the petals and, after
emasculation, both before^ and after fertilisation, protected by a paper
bag for two days by which time the stigma has usually dropped. The
flower from which the pollen is to be obtained is also protected in like
manner before the bud opens. The manipulation is simple and among
several hundreds of plants in the Fi generation not a single case of
accidental selfing has occurred. The parental types have been grown in
each successive season from the seed of protected flowers. In the
majority of instances the bud is simply covered with a bag which is
removed after two days. This method has been found, however, to lead
to a considerable degree of sterility and in some cases it has been
necessary to adopt the method of intercrossing different plants of the

^ Vide also Balls (2) which has appeared since the above was written.

* Since several characters are not recognisable nntil ripe fruit is developed, it is
usnally impossible to select these plants antil it is too late in the season to obtain seed
from them by self- fertilisation.

* The flower is never fertilised at the time of emasculation. Compare Hartley (23).

208 Studies in Indian Cotton

same generation. Such crossing is confined to individual plants of the
same pure line and its success is of some interest in view of the facts
noted by Darwin (5)\ Where the produce by self-fertilisation is required,
as in the F^ and subsequent generations, the flower is invariably pro-
tected in spite of the considerable labour of handling some 9000 flowers
in the course of four to six weeks. In no case has the produce of an
unprotected flower been included in the results given below.

The genus Gossypium and the types used in the
investigations.

For reasons which will appear in the course of this paper the author
does not consider it advisable at the present time to put forward any
scheme of classification to which reference can be made for identifica-
tion of the types handled by him. Nor is he able to accept in full any
of the classifications hitherto advanced. Only those types to which
reference is made are therefore briefly described and referred to their
place in the schemes in current use. The oldest of these are the
classical studies of the genus Gossypium by Todaro and Parlatore (15
and 14) where comprehensive schemes for the classification of the genus
are to be found. At a more recent date The Indian Cottons have been
dealt with by Gammie (9) and lastly Watt has reviewed the whole genus
in full detail in his Wild and Cultivated Cottons of the World (20).

The Indian cottons fall into two marked groups distinguished from
each other by the type of secondary branching. Arising from the main
axis, which is invariably a monopodium, the secondary branches may
either be monopodia or sympodia. The type in which all the secondary
branches are sympodia has not been observed though it frequently
happens that individual plants of certain types exhibit sympodial
secondary branching only (cf. PI. XXXIV, facing p. 208).

Nevertheless, in pure races, the number of monopodia produced at
the base of a sympodial type is invariably limited and the two groups
stand in obvious contrast on this point.

Monopodial branches are in most cases, though not invariably,
ascending while the sympodial branches are usually spreading and the
two groups lie in marked contrast to the eye. This difl'erence appears
to be fundamental and not limited to the appearance. The flowers are
invariably borne on sympodia which take the form of leafy cymes or,
more strictly, monochasia. In the sympodial group, therefore, the
1 See also Goebel, K. in Darwinism and Modem Science, p. 401.

JOURNAL OF GENETICS, VOL. \. NO. 3

PLATE XXXIV

Monopodial type.

Sympodial type.

H. M. Leake 209

flowering period commences with the development of the secondary
branching, while in the monopodial group this period is delayed until
the tertiary branches arise. In India the cultivated monopodial types
are in the minority and occupy distinct tracts. Cross-fertilisation be-
tween the different types is consequently of rare occurrence, arising
under exceptional conditions only, and intermediate types are few.
The sympodial types on the other hand are widely cultivated through-
out continuous areas and consequently, in the absence of any control
over the seed supply, have become inextricably mixed through natural
cross-fertilisation. The occurrence of crossing between the monopodial
and sympodial types is, as in the case of the monopodial types, suffi-
ciently infrequent even where these types are grown in close proximity.
The monopodial commence flowering about five months after the sowing
period while the sympodial are in full flower in three months and are
producing only stray flowers when the monopodial types are in flower.
While, therefore, natural crosses are fairly frequently observable among
the offspring of monopodial types, such crosses have not occurred within
the author's experience among the offspring of s3'mpodial types grown
in like proximity.

The types which have been employed in the experiments detailed
below are characterised in the subjoined list. In this list no attempt
has been made to arrive at a full and accurate botanical description of
each type. The references to the current schemes of classification will
sufficiently indicate the broad outlines of the type in question while
below are given only such characters as it is desirable to emphasise
owing to the position they occupy in the course of the experiments
about to be detailed.

Monopodial types.

Type 1. Perennial ; secondary branches ascending sharply at an
acute angle. Leaf factor' is less than entire 2 ; plant almost glabrous.
Bracteoles small, triangular, margin entire or dentate. Corolla yellow.

This plant is the G. obtusifolium Roxburgh Flora Indica of Gammie
(9) and Watt (20). The various forms to which the specific name obtusi-
folium has been given at different times have been dealt with by
Burkill (21).

Type 2. Perennial ; with secondary branches spreading. Leaf with
a factor less than 2. Stem and leaves densely covered with short hairs.

1 A detailed account of the leaf factor is given below (p. 221). It is the valae obtained
for the ratio a-b:e, vide Fig. 1, p. 220.

210 Studies iii Indian Cotton

Bracteoles deeply auriculate or reniform, deeply serrate, spreading in
fruit. Corolla yellow, petals small. Stigma heavily glandular. Capsule
inflated and nearly spherical with a sharp mucronate apex.

This plant is the G. herhaceum Linn, of Todaro (16) and Gammie
and the G. obtusifolium var. Wightiana of Watt (20).

Type 3. Perennial " tree cotton " ; secondary branches ascending
sharply at an acute angle. The entire plant of a deep red, or purple
colour. Leaf with a factor greater than 3 ; frequently with an extra
tooth on one or both sides of the central lobe. Bracteoles small, tri-
angular ; margins entire or with the tip dentate. Corolla deep-red.
Stigma eglandular. Capsule usually 3 celled, ovate.

This plant is the Gossypium arboreum of Linn. Sp. PI.; Parlatore(14);
Todaro (16); and the G. arboreum type of Gammie (9) and Watt (20).

Sympodial types.

Annuals with a few only, or none, of the lowest secondary branches
monopodia, the remainder sympodia ; the monopodial branches ascending
and the sympodial spreading.

Type 4. A tall plant, in later stages drooping under the weight of
fruit. Leaf large, with factor less than 2 ; lobes commonly 3 or with
two small accessory basal lobes. Young stem and leaves sparsely hairy.
Bracteoles small, entire or with few small apical teeth, closely enveloping
bud and fruit. Corolla yellow with deep-red " eye." Petals large, semi-
transparent. Stigma eglandular or with few glands only. Capsule
commonly 3 celled, ovate.

This plant is the Gossypium indicum Lamk. of Gammie (9) and
G. Nankin var. bani of Watt (20).

Type 5. An erect plant, in later stages drooping under the weight
of fruit. Leaf factor less than 2 ; lobes 5 — 7. Young stem and leaves
hairy. Bracteoles large, entire or with few small apical teeth loosely
enveloping bud and in fruit sometimes reflexed. Corolla yellow with
deep-red " eye " ; petals opaque. Stigma eglandular or with few glands
only. Capsule commonly 3 — 4 celled, ovate.

Type 6. An erect plant differing from type 4 in the greater rigidity
the main stem and the angle at which the secondary monopodia arise,
in this case about 4.5°, and in the corolla which is white. The petals
are small, scarcely projecting beyond the bracteoles.

Type 7. Plant erect with secondary monopodial branching, when
developed, sharply ascending. Leaf factor less than 2 ; flower white.

H. M. Leake 211

This type differs from the last in two respects. The secondary mono-
podia! branches, if developed, are sharply ascending. Frequently, how-
ever, they are absent, and even when present reduced in number in
plants where the growth of the main axis has not received a check, to
one, or at most, two with vigorous growth. The plant is consequently
strongly asymmetrical. For the same reason the length of the vegeta-
tive period is very brief and the first flowers develope while the plant
is still quite small. Growth continues throughout the season, the plant
maintaining a marvellous fertility.

Type 8. A tall plant, in later stages drooping under the weight of
fruit. Leaf factor greater than 3 ; lobes 5 — 7 with an extra tooth,
on one, or both sides of the central lobe, frequently developed. Young
stem and leaves hairy. Bracteoles entire or with few apical teeth.
Corolla yellow with deep-red " eye." Stigma eglandular or with few
glands only. Capsule 3 — 4 celled, ovate.

Type 9. A plant differing from (6) in the colour of corolla only
which is white and scarcely protrudes beyond the bracteoles.

Types (4) — (9) fall into the G. neglectum and G. roseum of Todaro (16),
the G. neglectum Tod. of Gamraie (9) and the G. arhoreum vars. neglecta
and rosea of Watt (20).

Type 10. A tall plant with the main stem weak and early drooping.
Leaf factor greater than 3; lobes 5 — 7. Bracteoles entire or with
few apical teeth, large and continuing to grow with the developing
boll. Corolla pale-yellow with deep-red " eye." Stigma eglandular.
Capsule ovate very large with numerous seeds.

This plant is the G. cemuum of Todaro and Gammie and the
G. arhoreum var. assamica of Watt (20).

Type 11. A tall plant with leaf factor greater than 3; leaf lobes
5 — 7 ; stem and leaves of a deep-red or purple colour ; bracteoles entire
or with few apical teeth. Corolla with deep-red " eye," petals white,
tinged with pink along margin and the portions exposed in the bud.

This plant is the G. sanguineum Hassk. var. minor of Gammie (9).

In the above description no reference has been made to the glands
which occur on the under-surface of the main ribs of the leaf The
presence, absence and number of these glands was at one time con-
sidered a point of some systematic importance. For the present it may
be noted that most of the types above described can be divided into
three groups ; that in which the leaf has no glands, that in which the
leaf commonly has 3 gland.s, and an intermediate group, in which
the majority of the leaves possess only one gland situated on the

Joam. of Gen. i 15

212 Studies in hidian Cotton

under-surface of the mid-rib. A detailed discussion of the leaf glands
is reserved for subsequent treatment.

The experiments.

In the crosses that have been made between pairs of the above
types the characters that have been observed in greatest detail include
the following :

(a) The colour of the corolla.

(b) The red colouring matter of the sap.

{d) The type of branching and length of vegetative period,
(e) The leaf glands.

3 (a). The colour of the corolla.

In all the types of Indian cottons now under consideration, there
occurs at the base of each petal a deep-red or purple spot or " eye."
The remainder of the petal is uniform, either yellow in colour or white,
while in the two types, 3 and 11, the colour is red (cf. Plate XXXV).
The red colouration of the petals in these two types is simply a mani-
festation of the red sap colour which is present throughout the plant.
It is not characteristic of the petals alone and cannot be dealt with
exclusively as such.

According as the factor producing the yellow colour is present or
absent, there arises a simple pair of allelomorphic characters peculiar
to the corolla, of which the presence of the colour producing factor
is dominant. This is shown in the cross between a type with yellow
petals and one with white petals.

The results of such a cross between type 4, with a yellow, and
type 6, with a white, flower are set out in Table I.

In the F^ generation of this cross the plants are all yellow, the
colour being indistinguishable from the yellow of the parent type 4.
The yellow is completely dominant.

In the F2 there occurs a separation into two groups, yellow-flowered
and white-flowered. The numbers handled are not large and small
weight can be attached to the ratio of 2'1 yellow to 1 white obtained,
in which the proportion of yellow is considerably below expectation.
It is noteworthy, however, in this connection that throughout the
provinces where these types are cultivated, although the crop presents

H. M. Leake 213

a remarkable range of intermediate types, the white-flowered types are
recognisably the hardier. On the correlation that apparently exists
between hardiness and white flower there is, at present, no definite
information available but such correlation would tend to produce a
preponderance of whites.

The figures for the F, generation are unfortunately meagre owing
to the large loss (some 90 °/J of young seedlings caused by the early
continuous nature of the rains in 1909. The figures are too few to
aflford any numerical guide as to the proportion of pure dominants
among the yellow F^ plants. They indicate, however, that in the F,
generation there occur plants of the type DD giving only yellow
offspring and others of the type DR which give both yellow and white
offspring. The F^ plants with a white corolla on the other hand give
whites only. These results are in entire agreement with those recorded
by Fyson (7) and a comparison of the two series of results gives strong
evidence as to the part played by vicinism in the field.

Type 10 is characterised by a flower in which the full yellow of the
petals of the type previously considered is replaced by a pale but
distinct yellow. This pale yellow behaves as a simple recessive to the
full yellow. From a cross between this type and type 8 which is
characterised by a full yellow petal the Fy generation is obtained in
which the petal is indistinguishable from that of the parent type 8,
the full yellow thus being completely dominant. In the F2 out of
140 plants, 41 possessed the pale yellow corolla giving a ratio of
2'4 plants with full, to 1 with pale, yellow petals. Similarly the full
yellow in type 2 appears to be completely dominant over the pale
yellow of type 10 giving, in the ^1, plants of which the petal colour
is indistinguishable from that of the parent type 2, and, in the F^,
91 plants with full yellow, and 66 plants with pale yellow, petals. In
both these cases and in the latter especially there occurs a large excess
in the actual, over the expected, number of recessives which it is
necessary to indicate though without further comment since, through
inability to cope with the entire series of experiments as at first
planned, a number of crosses, including those involving type 10, had
to be abandoned after the F, generation had been recorded and the
presence of this excess has not formed the subject of further experi-
ment. In no case has the pale yellow flowered been crossed with a
white flowered type.

It may be here noted that these are the only two cases in which
complete dominance has been observed in the cottons under consideration.

15-2

214 Studies in Indian Cotton

In all other cases, as will subsequently appear, dominance is incomplete.
The fact may be contrasted with the observation of Balls (1) in another
group of cottons in which the impure form is recognisable as of a pale
lemon yellow distinct from the full yellow of the parent.

3 (6). The red colouring matter of the sap.

In types 3 and 11a red anthocyanic colouring matter is present in
the sap and communicates an intense red to the entire plant — stem,
leaves and flowers. This colour is especially marked in the young
chlorophyll bearing tissues, but it is also distinct in such organs as the
stigma, anther and fruit. The young leaves are of an intense purple
which fades, however, as the leaf developes until in the mature leaf,
only the ribs and veins show the colouration distinctly, the lamina
retaining merely a slight, and, in cases, barely distinguishable, colour.
From the remaining types this colouring matter is absent and their
foliage is green, and the petals either white or yellow.

The Fi generation of a cross between either of the types 3 and
11, and, in fact, any type in which the red colour occurs, and the
types in which the colour is absent, bears the red colour which may
be said to be dominant. The intensity of the red colour is, however,
sufficiently diminished to render the cross readily distinguishable from
the parent.

The F2 generation is readily separable into two sections according
to the presence or absence of the red colour. The proportion of
coloured individuals to colourless which has been obtained in the
experiments under review are given in Table II,

Among the coloured individuals, however, there is a considerable
range of intensity in the red sap. In the foliage this is apparent in
the extent to which the colour suffuses the leaves. In the least
intense form the mid-rib and two main lateral ribs of a young, but
fully expanded, leaf are suffused, and the minor veins and lamina green.
Further intensification occurs when the larger veins appear as red lines
set in the green lamina, and finally the whole lamina may be suffused
as in the red parental type 3. The records of this character in its
relation to foliage character show the colours as limited to (a) the ribs,
(6) the veins, or (c) the lamina diffused throughout. While the limits
between these three are not very distinct, the division offers a fair
guide to the purity or reverse of the plant under consideration in
r3gard to this character.

Table III (a) shows that in 5 cases out of 66 an error was made in

H. M. Leake 215

over-estimating the intensity of the colour, while in 2 eases only out
of 138 the error was made in the direction of under-estimation giving
a combined error of approximately 1 in 25. In Table III (6) a larger
error occurs, the intensity of the colour having been over-estimated in
4 cases out of 63 and under-estimated in 22 cases out of 212, making
a combined error of 1 in 10. As a generalisation it may be said that
in the intensity of the colour in the leaf there exists a character by
means of which it is possible to separate with a fair degree of certainty
the pure dominants of the form DD from the impure dominants of the
form DR.

The presence of the red factor does not, however, merely find its
expression in the leaf. As has been stated it is universally present
and is readily identified in organs devoid of chlorophyll such as the
petals. Under these circumstances it would seem probable that the
intensity of the colour would be most readily determined in an organ
like the petal where the colour is not masked. A reference to Table IV
will show, however, that this is not so. Gradation in the case of the
petals does not occur as one of intensity but one of area. The petal is
either entirely red or red with areas, greater or less in extent, situated
round the eye in which the red colour is absent. Such plants, in the
case of a cross between a red, and a yellow, flowered type have
been recorded as having the petals red on yelloiu. As Table IV (a)
indicates, among 201 plants, 30 were recorded as having the petals
red and the remaining 171 plants as having the petals red on yellow.
Actual experiment has, however, shown that in reality 63 of the plants
used as parents were of the form DD. The petal colour, therefore,
failed to distinguish between the pure, and impure, forms in 33 cases
out of 201. This error of approximately 1 in 6 compares very un-
favourably with the error of 1 in 25, which was obtained when the
colour of the leaf was considered.

Table IV (6) is derived from the cross between a type with red, and
one with white petals, and in it only that section of the F2 genera-
tion in which the dominant yellow occurs has been considered. In
this example an error occurs in 49 out of 196 instances making an
error of 1 in 4. In both cases the error is considerably enhanced when
the determination is made on the flower instead of the leaf. In spite
of the masking effect of the chlorophyll, therefore, the intensity of the
sap colour is most readily identified in the young leaf and, determined
in this manner, affords the most accurate guide to the purity, or
impurity, of the plant with respect to this character.

216 Studies in Indian Cotton

The results detailed in Table II may now be expanded to include
greater detail and the group possessing sap colour divided according
to the intensity of that colour. This expansion is effected in Table V.
Correcting the totals in this table as far as subsequent experiment
renders possible the numbers become, DD 373, DR 810, RR 384,
giving a ratio of I'OO : 2-17 : 1-03. The widest variation from the
expected result occurs in the cross between type 2 and type 3. This
is a cross between two monopodial types, and, for reasons already
explained, it has been found almost impossible to handle this cross,
which has not, in consequence, been carried beyond the F2 generation.
It is impossible, therefore, to say how far the lack of the pure dominant
form is real. It is noticeable that this lack is associated with a large
proportion of the impure vein form and that, consequently, the deter-
mination from the leaf may not be as accurate as in the cases more
fully investigated.

In the above considerations on the behaviour of the red colouring
matter in the sap no distinction has been drawn between the various
types used in which this colouring matter is absent. It is necessary
briefly to consider these types under two groups, namely those in which
the petals are yellow and those in which the petals are white. It is
the identical cross made by Major Trevor Clarke and, as described by
Watt (20) (page 336), one on which he founded great hopes. These
experiments of Trevor Clarke are the subject of frequent note in the
Journal of the Agri-Horticultural Society of India of that date (1867-
1870), but no full details have been traceable. A similar cross is referred
to by Fletcher (22). Here again full details are not given but, in as far
as both red arid yellow flowered plants appeared in the Fi generation,
it would appear that the red parent plant was a heterozygous form.

(i) Type 3 x type 4. Type 3 possesses a full red colour both in the
foliage and flower which in type 4 is absent, the foliage of this type
being consequently green and the petals yellow. The cross, which has
been carried as far as the Fs generation, may be taken as an example of
the case in which the first of these groups is employed. The results
obtained from this cross are set out in Table VI. The numbers
obtained in this series bear a ratio in close accordance with Mendelian
expectation and it is evident that in this cross an example of the
simplest case occurs, namely that in which a single pair of allelomorphic
characters is concerned. This pair is composed of the two factors —
presence or absence of the red colouring matter — the present condition
possessing partial dominance over the absent. This being so, it follows

H. M. Leake 217

that the red parent must possess the yellow factor in addition to the
red. That this is so will be seen from an examination both of the base
of the petal, which usually exhibits a slight yellow colouration on the
external surface, and of diseased flowers, in which the petal almost
approximates to the red on yellow condition of the impure cross.

(ii) Type 3 x type 9. As far as the present discussion is concerned
this cross differs from the last only in the fact that one of the parents,
type 9, has a white, instead of a yellow, petal. The cross is of consider-
able interest because on it have been based the greatest hopes of
obtaining an improved cotton suitable to the needs of the United
Provinces and the results have consequently been investigated in some
detail. The present interest, however, does not arise from this aspect
but concerns the flower colour. In the F^ generation the corolla is
indistinguishable from that of the ^i of the cross previously described
and is of the class which has been above denoted as red on yellow, the
red petal having round the eye a border of greater or less extent of
yellow. (PI. XXXV.)

From the self-fertilisation of the jP, generation plants are obtained
which can by corolla character be divided into four groups :

(1) Corolla red or red on yellow.

(2) „ red on white.

(3) „ yellow.

(4) „ white.

In addition to the two original, two additional types of corolla have
made their appearance. If now, as in the previous cross, the colour of
the foliage is taken into consideration six groups become recognisable.
These are :

Flower

Foliage

(1)

corolla red

colour extending to lamina.

XXXV. 4

(2)

„ ,, on yellow

,, „

,, veins.

XXXV. 5

(3)

„ ,, on white

»> 5»

,, lamina.

XXXV. 6

(4)

,, „ „

J» »l

,, veins.

XXXV. 7

(5)

„ yellow

colourless.

f»

XXXV. 8

(6)

„ white

«>

XXXV. 9

This behaviour is readily explained on the assumption that two
pairs of allelomorphic characters are here being dealt with :

(a) Presence of the red factor which has been shown to be dominant
to absence of the same.

(b) Presence of the yellow factor which has been shown (p. 213)
to be dominant to absence of the same.

218 Studies in Indian Cotton

The red type 3 possesses the two dominant, and the white type 9
the two recessive, factors. Denoting these two pairs by the letters Rr
and Yy, the two parental types will bear the constitution i2F and ry,
and the six groups which have been recognised the constitution given
below with the numerical proportion between the individuals which is
assigned to each group :

(1) hryy 1)

RRYy 2

(2) RrYY 2)

RrYy 4^^

(3) RRyy 1 \^

(4) Rryy 2 J

(5) rrYY 1 , ,

rrYy 2

(6) rryy 1 I

The plants of the first group can be separated into two subsidiary
groups, the members forming the one being pure with regard to both
characters, while those forming the other will be pure with regard to
the red, and impure with regard to the yellow, character. Groups (2)
and (5) can be similarly divided and in all cases this division will be
recognisable in the offspring. How far these assumptions are borne
out in experiment will be seen from Table VII where the results of
this cross are set out in detail. In all cases the expected groups have
been formed and the actual numbers are in close accordance with those
expected on the above scheme.

The facts concerning the petal colour and the red anthocyanic
colouring matter of the sap are, therefore, fully explained on the
assumption that two pairs of allelomorphic characters enter into con-
sideration, these two pairs being composed of the two factors producing
the red colour and the yellow colour respectively, the presence of the
colour producing factor being in both cases the dominant, and its
absence, the recessive, condition.

Starting with the red and the white flowered type, it has been
found possible not only to produce, but to produce in a state of purity,
two other types, one having a yellow (PI. XXXV), and the other a red
on white (PI. XXXV), flower. Apart from complications introduced by
the consideration that one of the parents is a monopodial, late flowering
type, which may be put aside for the moment, the yellow flowered
form is recognisable as type 8, and similarly the red on white flowered

H. M. Leake 219

form is comparable to type 11, a type which is found cultivated in the
Punjab.

The conclusions drawn from the results obtained from the series
derived from the direct crosses as described above, receive confirmation
from a second series obtained from crosses between the Fi generation
and the parental type. Owing to illness and consequent limitation of
the working period, it became impossible to complete the records of
this season and a part of this series had to be abandoned. The some-
what meagre records which were obtained are tabulated in Table VIII.
The number is too small to admit of any numerical comparison, the
character of the offspring can alone be considered. In all cases
involving one pair of characters only, the cross with the dominant
parent has given only dominant and intermediate forms and that with
the recessive parent only recessive and intermediate forms. In the
single instance in which two pairs are concerned the cross between
the intermediate form (RrYy) and the parent possessing both dominant
characters (RRVY) has given offspring similar to the pure dominant
(RRYY and RRYy) or to the F, intermediate (RrYY and RrYy),
while that with the parent possessing both recessive characters has
given, in addition to the form with both recessive characters, three of
the four recognisable intermediate forms, that with a red (or red on
yellow) flower and colour extemling to the veins {RrYy), that with red
on white flower and colour extending to the veins (Rryy) and that
with a yellow flower (rrYy). These forms are, in all cases, such as
would be expected. In the one case where the recessive only has been
obtained, the number of individuals (2) is too small to make the
absence of the intermediate form a matter of any moment.

Before concluding this section the cross between type 3 and
type 10 may be briefly referred to. It has been already shown (p. 213)
that the pale yellow of type 10 is recessive to the full yellow of types 2
and 4, and from the experiments last quoted it is apparent not only
that a yellow underlies the red in the petal of type 3, but that this
yellow is identical with the full yellow of type 4. It would, from this,
appear probable that the cross between type 3 and type 10 would be
comparable with the cross between the two types 3 and 9 just dis-
cussed. This expectation is borne out in experiment. The plants of
the ^1 generation of this cross are in all their petal characters similar
to those of the cross between type 3 and type 9, that is of the form
which has been denoted by the term red on yellow. In the F^ genera-
tion four types of plants as distinguished by their petal colour appear :

220

Studies in Indian Cotton

(a) Corolla red or red on yellow.

(b) „ red on pale yellow.

(d) „ pale yellow.

The number of individuals occurring in each group has been found
as follows :

(a) 263, (b) 88, (c) 83, (d) 17. Except for the paucity of the
individuals in group (d) these numbers agree fairly with the Mendelian
ratio of 9 : 3 : : 3 : 1.

Further the two groups (a) and (b) are capable of subdivision in
accordance with the degree to which the red colouring matter suffuses
the leaf Owing, however, to the crosses from type 10 being discarded,
no full records of this appearance are available and it can only be noted
that, to the extent of these incomplete records, the two crosses between
types 3 and 9 and between types 3 and 10 are strictly comparable.

3 (c). The leaf factor.

The term leaf factor has been described by the author in his first
introductory note to the cotton work undertaken by him (11).

Fig. 1.

It is the numerical value obtained by dividing the difference
between the two measurements a and b in the accompanying diagram

H. M. Leake 221

(Fig. 1) by the measurement e. It is not proposed to enter into a
detailed discussion as to the significance of the constancy of this factor
for the various types of Gossypia. It may be noted, however, that its
identification was purely empirical and it is not to be taken as an
absolute figure for each leaf of a plant ; there is a fair range of fluctua-
tion as would be expected in the measurements of any series of
multiple organs. In spite of these fluctuations it is a matter of little
difficulty to recognise what may be termed a " typical " leaf and there
is a very marked agreement between the leaf factor, as determined
on such " typical " leaves, of individuals of the same type.

The degree to which the leaf is incised forms a striking feature of
the plant and has been adopted freely as a means of classification.
Todaro (16) divides the Indian group (subsectio Indica) of Gossypia into
two sections :

A. Lobi breves, ratione longitudinis latiusculi.

B. Folia palmato-partita, lobis angustis, oblongis, vel elongato-
lanceolatis.

Gammie (9), though he does not accord this character of the leaf a
primary position in his scheme of classification, throughout refers to
two gioups with the leaf lobes either broad or narrow.

Watt (20) uses the leaf character to subdivide the section of
" Fuzzy seeded cotton with united bracteoles." He distinguishes three
groups :

Leaves two-thirds palmately (sometimes almost pedately) 3 — 7 lobed.

Leaves half-cut into 3 — 5 (mostly 3) lobes.

Leaves less than half-cut into 5 (more rarely 3 or 7) lobes.

It will be noticed that while these three schemes deal generally
with the same character there is some difference in detail in the exact
points involved. Watt simply deals with the degree of incision which

is, perhaps, most closely given by the ratio r •

Todaro's group B, as fully defined, is distinguished by not only the
factor J- but by the breadth of the lobe, thus including the measure-
ment e ; while for his group A he makes use of an expression which is,
perhaps, the best form of definition that could be found for the author's

" leaf factor." Gammie refers simply to the ratio , — %-r of the lobe,

'^ •' breadth

which is identical with the leaf factor.

In a preliminary series, among other measurement determinations.

222

Studies in Indian Cotton

Fig. 2.

The top left figure is that of a broad lobed leaf, with leaf factor less than 2 ; the bottom
figure is that of a narrow lobed leaf with leaf factor greater than 3 ; the top right figure
shows an intermediate leaf with leaf factor 2-5.

H. M. Leake 223

the ratio y was determined for a large series of plants but was found

to be quite inconstant and useless as a means of identifying types
which were readily distinguishable by eye. On the other hand in the
leaf factor an expression was found not only for such differences as are
of sufficient magnitude to be recognised by the eye but also for such
as, though definite and constant, are elusive to the eye and incapable of
adequate verbal definition. While there is thus found in the leaf factor
a means of defining and expressing to a degree of minuteness hitherto
impossible, what appears to be a unit character of the cotton leaf, it is
necessary to beware of pressing it too far. It is physically impossible
to measure every fully developed leaf and obtain fi:om such measure-
ments an average. " Typical " leaves must be selected and in such
selection the door is opened for the introduction of a considerable
personal element. In the experiments recorded determinations have
been made on at least two such " typical " leaves from each plant and
the average between the two values so obtained is taken as the leaf
factor of the individual.

Before dealing with this character in detail therefore both the
magnitude of the error met with in these determinations and the
exact meaning to be ascribed to the term " typical " require brief
consideration.

It is clear that a larger experimental error is to be expected in the
leaf factor of types with narrow lobed, than those with broad lobed,
leaves. In the latter case the three measurements employed in the
calculation are all large and errors of measurement proportionately
small. In the former case, on the other hand, the divisor e is small
and the errors proportionately large. The experimental error, con-
sequently, increases as the value of the leaf factor rises. When this
value falls below 2 the error, which is accepted, is normally less than
015 from the mean (giving a total range of 0'3) and, when this value
lies above 3, this error may reach 0*3 (with a total i-ange of 06).
These figures indicate the extreme variation met with. Where the
error exceeds this amount duplicate determinations have been made.

The recognition of this leaf factor was, as has been stated, in the
first place purely empirical and resulted from an attempt to find some
method of denoting by symbols the differences between the various
characteristic shapes of the cotton leaf. In the selection of leaves
used in the determinations certain precautions were found to be
necessary and were consequently adopted. That such precautions were

224 Studies in Indian Cotton

necessary receives recognition in the use of the word " typical," These
precautions require examination since, in a purely arbitrary deter-
mination of this nature, some control is required to ensure that the
restrictions imposed by their use are not of a nature to render valueless
the figures so obtained. Such a check has been found in the measure-
ment of the leaves of one individual of each of the several pure
types isolated, only the earliest leaves of the main stem and the
diminutive leaves at the base of each branch being excluded. These
measurements were made at intervals of about a week throughout
the season, each leaf being thus measured as it became fully expanded.
The results of one such determination in the case of a plant of type 5
are set out in Table IX. For the purpose of their understanding the
leaves may be grouped into four sets :

(1) Leaves borne on the main stem.

(2) „ „ monopodial secondary branches.

(3) „ „ tertiary branches.

(4) „ „ sympodial secondary branches.

It will be noticed that the monopodial secondary branches alone
bear tertiary branches which are almost invariably sympodial. The
values obtained for the average leaf factor of these four groups are
respectively :

(1) 1-82, (2) 1-84, (3) 1-73, (4) 1-72.

It will be noticed that the leaf factor of the leaves borne on the
monopodia is definitely larger than that of the leaves borne on the
sympodia whether these be secondary or tertiary branches. The value
of the leaf factor as determined for the leaves arising from the mono-
podia, differs by between 0*06 and 0*04 from the value obtained by
the empirical method of selection of " typical " leaves. This error lies
well wdthin the limits of the experimental error as defined above. The
" typical " leaf, therefore, may be defined as that leaf which possesses a
factor having a value equal to the average of the factors of all leaves
arising from the monopodial branches. It is not, as was anticipated
when the author's earliest note (11) was published, the average of the
factors of all the fully developed leaves. This result is in perfect accord
with the main precaution which on empirical grounds it has been found
advisable to take, namely, to select leaves from the monopodia. It is
these leaves that the eye naturally selects as being typical of the plant.
It is perhaps unnecessary to detail more than one further precaution
which it has been found advisable to adopt. This is to avoid the

H. M. Leake 225

determination of the value e where an accessory notch (vwie Fig. 1,
p. 220) occurs in the re-entrant angle at the base of the main lobe.
Such precautions are obviously necessary and cannot affect the value
of the leaf factor as a definite character.

The determination of the leaf factor for many thousands of plants
has brought one remarkable feature into prominence. While every
value has been obtained for the leaf factor from 1 ("broad" lobed)
to 5 (" narrow " lobed) no case has been observed in which a plant
with intermediate value (between 2 and 3) for the leaf factor breeds
true to this character. All pure plants, and consequently all types, are
divisible into two distinct groups :

(1) With a leaf factor less than 2.

(2) „ „ greater than 3.

Within the limits 1 to 2 occur all the " broad " lobed types, while
within the limits 3 to 5 occur all the " narrow " lobed types*.

The accuracy of the expression — that is, the measure of agreement
between different individuals of one type — is such that it is possible
to recognise within, and isolate from, a type, otherwise pure, races
separable only by the leaf factor. It seems probable that the existence
of such "pure lines," t-o use Johannsen's term (10), is a phenomenon of
general occurrence throughout this series of Gossypia and in some of
the types such forms have been isolated. Thus within type 4 occur
three "pure lines" with leaf factors of 1*37, 1*46 and 164 which have
been isolated, and from type 9 "pure lines" with leaf factors of 3*34
and 359 which have similarly been isolated. Opportunity has not been
forthcoming for treating this question in the detail it deserves and it
seems probable that with a more detailed examination the number
might be considerably increased. Indications of the existence of such
" pure lines " are apparent in Table X.

The behaviour of the leaf factor when crossing occurs.

When a plant with leaf factor less than 2 is crossed with a plant
with the leaf factor greater than 3 the leaf factor of the plants of the
Fi generation is found to approximate to the mean of the two parental
leaf factors. Table XI illustrates this point. At the time the crosses
were made the character had not been identified and the figures given

^ In the fields plants are frequently found with a leaf factor less than 3 and greater
than 2, and on this fact among others the author has based his views on the occurrence
of cross-fertilisation under natural conditions (11).

226 Studies in Indian Cotton

for the parental leaf factors are not those of the actual plant but
the average of the type as given by the offspring (produced by self-
fertilisation) in the two subsequent generations. In two cases only is
the variation from the parental mean at all marked and in both these
this difference is not shown by the reciprocal.

In the F^ generation a continuous series of forms is produced in
which every value of leaf factor between the parental limits is obtained.
Diagram 1 illustrates one such case and is derived from the series given
in Table XIII (6). It is here noticeable that, while the series appears
continuous, in that every value of leaf factor (within the limits imposed
by the experimental error) occur, the number of individuals is by no
means regularly distributed throughout the series — in other words, the
frequency of each class exhibits marked variation. The curve is, in
fact, multimodal (Davenport (6)) and possesses three modes. The posi-
tion and value of these modes are instructive. While the values of
the outer modes differ but slightly from the values of the two parental
leaf factors, the value of the intermediate mode shows a fair degree of
approximation to the value of the mean between the leaf factors of the
two parental types. The proportion between the number of individuals
grouping themselves about these three points is 1 : 22 : 1. The
curve retains its trimodal nature, if for the actual values obtained by
direct measurement of the leaves of individual plants — the values
here given — the mean value of the leaf factor of the ^3 offspring be
substituted.

A similar curve has with one exception been obtained in every case
submitted to a critical examination. In this instance, the cross between
type 2 and type 3, there is no trace of a multimodal curve and the
ratio between the number of individuals in each group (Table XIII («))
diverges markedly from that obtained in the instance given above.
Lack of opportunity and the difficulty of handling a cross between two
monopodia! types have rendered it impossible to continue investigation
into the behaviour of this cross and for the present it must remain
undecided whether, on further examination, this too will fall into line
with the example more fully investigated or whether a different series
of phenomena is here instanced.

So far the results have been described in outline only, and as
a close examination of the tables will show, are only approximate.
Complete agreement is, perhaps, hardly to be expected in dealing with
a character which, as has been already shown, cannot be measured
with absolute accuracy. It will be observed that the modal values of

H. M. Leake

227

the leaf factor in all cases exceed the corresponding parental or mean
parental value, the excess being practically identical (031, 035 and
0"38) in the three cases. This excess, though small, appears definite
but has so fiir received no explanation.

1 i i...

^ i 1 T "T^

! i t i

jT:::i:::;::"::""it"-i ^ ■ ::: . ^

_ii * ...t±±i..i.^ I

I t

_i r

:: :::: ::^^ i i 1 1

II _ _ ±1.... I ::i±_

Parents

11 16 21 26 31 36 41

Type 4 x Type 8
Diagram 1.

In one case only has each plant of the F^ generation been self-
fertilised and the ^3 generation raised from the seed so obtained. The
results are set out in Table XIV. In this table the extreme and

Joom. of Gen. i 16

228 Studies in Indian Cotton

intermediate groups are given in a condensed form so that the offspring
of all plants, the average leaf factor of whose offspring differs by O'lO or
less, are grouped together. Full details of individual plants are only
given at the two points where the change from the pure to the impure
form takes place. It will be seen from this table that a marked
difference exists in the behaviour of the individuals belonging to the
three groups into which the F^ parents fell. The offspring of those
Fn individuals of which the leaf factor was less than 2 have, with few
individual exceptions, a leaf factor which is less than 2. In the same
manner the offspring of plants with the leaf factor greater than 3'2
have a leaf factor which is greater than 3. As will be seen from the
table the exceptions are relatively few and it may be said in general
terms that the individuals of the two groups, having the leaf factor less
than 2 and greater than 32 respectively, are pure with regard to this
character. The dotted vertical lines in the Table drawn between the
columns representing the values 2"1 and 2*2, and between those repre-
senting the values 2*8 and 2*9, indicate the limits of experimental error
recognised in the two groups. It will be noticed that in 5 only out
of 1283 cases the limiting value of 2*1 is exceeded and in 7 out of
1274 cases the limiting value of 2*9 is not reached. These exceptions
will form the subject of further investigation. It is, of course, possible
that these plants have been introduced by accident. Nothing, however,
in the further examination of these individuals lends support to this
view.

The third group, which is characterised by the intermediate value
of the leaf factor, is not, like the previous groups, pure in this respect.
Such plants have invariably given offspring which, as a group, exhibit
the entire range of values obtained for the leaf factor. It will be noted
that, though this variability exists, the average value of the leaf factor
of the Fs generation from this intermediate group differs but slightly
from the mean of the two parental values and further that the number
of individuals comprising the three groups are in almost complete
accordance with Mendelian expectation (1 : 2*04 : 1), while the mean
values of leaf factors for the three groups taken severally show but
slight variation from the values obtained for the corresponding groups
of the jPa generation.

It is impossible to avoid being impressed by the similarity which
exists between these results and the more typical examples of
Mendelian phenomena.

It has frequently been pointed out {vide Bateson (3), p. 53) that

H. M. Leake 229

dominance, which formed so striking a characteristic of the earlier
experiments on these lines, holds no position of fundamental importance
in Mendel's own statement of his law. In the present instance there is
a complete absence of dominance and the direct offspring of a cross are
as markedly distinct from one, as they are from the other, parent.
It is possible, however, to discern more than this. The two factors
appear capable of blending in any proportion, and there thus appear
a continuous series of forms showing all stages from the typical broad
lobed individual, with a leaf factor less than 2, to the typical narrow
lobed individual with the leaf factor greater than 3. Owing, however,
to some influence, of which, as yet, nothing is understood, these various
degrees of blending do not occur with equal frequency. This is greatest
at the point represented by a blending of equal proportions of the two
factors and becomes less and less as this proportion becomes unequal,
but increases again when the proportion of one or other of the factors
is reduced to a negligible quantity or is entirely absent.

This capacity of blending in unequal proportions is further shown
by a comparison between the value of the leaf factor of the F^ parent
with the mean value of that of the ^3 offspring. This comparison is
given in the three last columns of Table XIV. The difference between
these two values for the whole series is 0*07, a figure well within the
limit of experimental error, which is, however, in a few individual cases
exceeded. It may be generally stated, therefore, that the value of the
parental leaf factor is the mean of the values for the offspring. Con-
sequently, when unequal blending occurs in any plant, the number of
offspring falling within the group whose leaf factor enters in greatest
proportion into the blending will exceed the number of offspring which
fall within the other group. In other words the ratio of the offspring
having a leaf factor less than 2 to offspring having a leaf factor greater
than 3 will increase as the parental leaf factor diminishes from the
mean value of 26 and will conversely diminish as the parental leaf
factor increases from this mean. That this is the case the detail
columns of Table XIV clearly show.

It is now necessary to glance for a moment at the lower limit of the
" narrow " lobed group. It has been stated that this limit is 30, a
figure which has, with one exception, been adopted in Table XL
Reference to Table XIV, however, will show that the lower limit for
the pure forms with narrow lobed leaves is 32 — a figure which exceeds
the value of the corresponding parental leaf factor. In this connection
it is noteworthy that a value of 3*5 is throughout obtained for the

16—2

230 Studies in Indian Cotton

mean leaf factor of this group. It is possible that this figure, 3*52,
more accurately represents the true value of the narrow lobed parent
than that actually obtained by experiments (313). This latter figure
is based on six determinations only and it is a matter for regret that
more determinations were not possible. Not only, as has already been
remarked, is type 3 difficult to handle, owing to its monopodial habit,
but it has been found to be in a marked degree self-sterile. In the first
generation only six plants were obtained by self- fertilisation, while in
the second, numerous attempts were all unsuccessful. While, therefore,
the value 3*13 has been adopted in these calculations it must be noted
that this value is extremely low for the type 3 as determined on a set
of pure, but unrelated, plants of this type. Acceptation of the figure
3'52 as more nearly representing the true parental value, while
accounting in full for the difference of "38 found between the value
of the narrow lobed parent and that of the corresponding F^ group,
accounts only partially for the difference of 0*35 between that of the
parental mean and of the intermediate group, and fails entirely in the
case of the difference of 0'31 between the broad lobed parent and its
corresponding F^ group. These differences must for the present remain
without explanation.

The few cases in which the F^ generation has been crossed with the
parent types are given in Diagrams 1 and 2. In all cases the F^
intermediate, when crossed by the broad lobed parent, has given only
intermediate and broad lobed offspring and, when crossed by the narrow
lobed parent, only intermediate and narrow lobed offspring. The
number of intermediates is far too small for any value to be attached to
comparison of their relative numbers and of the mean value of leaf
factor. It is impossible, therefore, to draw any further conclusion than
that, within the limits imposed by their paucity, these figures are in
complete accord with the expectation based on the conclusions derived
from the direct series.

3 {d). The type of branching and the length of the
vegetative period.

The differences which exist in the form of the secondary branches
and in the length of the vegetative period between the various types
under consideration have been briefly noted above (p. 209). The
intimate connection which has been found to exist between these two
characters in the Indian cottons has already been pointed out by the
author in Part 2 of his introductory note (12). In a still earlier

H. M. Leake 231

publication Balls (1) foreshadows a similar interrelation between the
type of secondary branching and the length of the vegetative period
in the Egyptian and American upland*.

Since the publication of the note referred to, a most interesting
communication from J. V. Thompson to the Agri-Horticultural Society
of India has been met with in the Journal of that Society for the year
1841, in which the intimate relation between the type of branching
and date of flowering is clearly indicated. In this communication he
states :

" The cultivated varieties of cotton I find may be divided into two
classes, viz. early and late kinds ; this precocity or tardiness being
inherent in the particular variety, and derived from a peculiarity
hitherto unnoticed, and which it will not be difficult to explain. It
may be observed that all the varieties have a natural tendency to
produce a central main stem furnished with a leaf at intervals of a few
inches ; in the axillae of each leaf-stalk resides a pair of germs or buds,
placed in the same plane or side by side ; one of these germs is destined
to produce flowers only, the other only branches. In the early kinds
the former or flowering branches alone are developed, while the late
kinds expend their force exclusively in the production of multiplying
branches. This peculiarity must for ever unfit these late kinds for a
cold climate, such as Northern India." For the full communication,
which is of some length, the reader is referred to the original source (18).
Sufficient has been quoted, however, to show how fully the importance
of the connection between the branching habit and the length of the
vegetative period had thus early been recognised. The importance of
two axillary buds, which is also indicated, has previously been dealt
with by the author (12) in a preliminary note but has no concern with
the experiments now under treatment

It has already been noted, when defining the types which have
been employed in these experiments, that the Indian cottons fall into
two well-defined groups, those in which the secondary branches are
always, or nearly always, monopodia, and those in which the secondary
branches are always, or nearly always, sympodia. As long as observation

* Since the above was written Balls. 147 and 155 Bureau of Plant Industry, U.S.
Department of Agriculture have been received. In these the authors draw attention
to this same point. According to them, however, this character is induced to vary in the
types investigated by them as a consequence of change in environment. This and other
differences in the method of branching between the observations of these investigators
and those of the author are not concerned with the subject matter of this paper and
most be left for consideration at a subsequent period.

232 Studies in Indian Cotton

is limited to pure types these two groups are readily distinguished.
When, however, the progeny of" crosses between types belonging to
these two groups come to be considered, every gradation between the
two extreme forms is found and it becomes a matter of extreme
difficulty in individual cases to define the degree of approximation to
one or the other extreme type. In such intermediate plants the
passage from one type of secondary branching to the other is usually
abrupt, the earlier branches being monopodia and the later sympodia.
It is, therefore, possible to divide the main stem into two portions,
a lower portion in which the branches are monopodia and an upper
portion in which the branches are sympodia. The character can then
be conveniently expressed as the percentage of the entire stem bearing
monopodial branches. Expressed in these terms a pure monopodial
type is indicated by the number 100 and a pure sympodial type by
the number 0. It has already been stated that no pure type has been
isolated which invariably produces sympodial secondary branches only.
A few monopodial branches may in all cases occur at the base of the
primary stem. It is convenient, therefore, to denote these also by the
symbol 0 which indicates all such sympodial types as have been found
to breed true. In like manner the symbol 100 may be used to denote
cases in which a few of the most apical branches are sympodial. In
the earlier experimental stages it was considered sufficient to recognise
four divisions only :

(1) The full monopodial type indicated by 100.

(2) Approximately three-quarters of the secondary branches mono-
podial, indicated by the symbol 75.

(3) Approximately one-half of the secondary branches monopodial*
indicated by the symbol 50.

(4) The sympodial type indicated by the symbol 0. Recently the
separate forms have been recorded in greater detail by which the
fraction, recorded in tenths, of the main stem bearing monopodial
secondary branches is used as a basis for division. By this method
10 groups are formed, the relation of which to the four groups given
above is shown below.

100 75 50 0

100 90 80 70 60 60 40 30 20 10 0

In this notation the figures 100 and 0 apply respectively only to
individuals in which sympodial and monopodial secondary branches
are entirely absent.

H. M. Leakb 233

It is clear that this system of record, though the best that has been
devised, is subject to considerable disadvantage. It is, at the best,
approximate and moreover the determination is only possible when the
main stem has received no check to growth. In practice this continued
growth of the main stem is rendered a fact of comparatively infrequent
occurrence from the climatic conditions prevalent at the early stages of
growth. These conditions favour insect life of all kinds and the larval
stage of Earias sp. is commonly met with on the cotton plants. This
pest penetrates the young stem at the leaf axil and from this point
bores its way downwards. The stem so attacked withers and growth is
continued by an enhanced development of the secondary branches. In
such cases it becomes difficult and frequently impossible to determine
this character even approximately.

The length of the vegetative period is most readily expressed in the
number of days from the date of sowing to the appearance of the first
flower. Unlike the previous character this lends itself to accurate
record. The fields are visited daily and the plants in flower for the
first time noted. Yet numerous subsidiary influences are here found to
aflfect the date of production of the first flower and render the figure,
though accurate in itself, only approximately accurate as an indication
of a definite individual character. The more important of these
influences may be here referred to.

In the first place, there has been found a considerable seasonal
variation ; that is, a considerable difi"erence in the length of the
vegetative period of a pure type from one year to another. Hence
the figures obtained for one year only are strictly comparable and it
is possible to compare the results of two or more years by introducing
a seasonal factor by the addition (or subtraction) of which the results
of any two years are rendered comparable. This is illustrated in the
column of Table XV for the years 1907 and 1908.

In the second place the length of the vegetative period is materially
influenced by the method of cultivation. Two methods have been
employed in the course of these experiments. In the first the seed is
sown in pots and the young plants, when a month to six weeks old,
planted out. In the second the seed is sown in the ground about
a month after the sowings in pots have been eflfected. Here only
indirect comparison is possible and the efifect of such variation in the
method of cultivation is shown by a comparison between the third and
the first two columns of Table XV.

Unfortunately no records are available by which the direct influence

234 Studies in Indian Cotton

of the method of cultivation may be calculated, for in no case has the
same type been grown by both methods in a single season. In 1907
and 1908 all the pure types were grown in pots, while in 1909 they
were sown in the field. To obtain a comparison between the two
methods of cultivation it is necessary to resort to an indirect method
based on the crosses. In 1908 the entire ^2 generation obtained from
the crosses was raised in pots while of the seed of these plants only
that of which a small amount was available was, in 1909, sown in pots,
the remainder being sown in the field. In Table XVI is given the
result of the comparison between the length of the vegetative period of
the offspring of plants having a similar period when these offspring are
grown under the two conditions. The difference due to the method of
cultivation varies from a minimum of 21 days to a maximum of 31 days
and, generally speaking, the greater the length of the vegetative period
the greater will be this difference.

A similar result is reached from a comparison of Tables XVIII —
XXI. Tables XVIII and XX are based on the pot series and involve
only the seasonal difference between the two years 1908 and 1909,
which is foHind to be five and three days respectively. In Tables XIX
and XXI, based on the field series, in addition to this seasonal differ-
ence there also occurs the difference due to the method of cultivation,
and the combined differences are in the two cases 31 and 28. By
subtraction the average difference due to method of cultivation alone is
found to be in the one case 26, and in the other 25 days.

From the above it is noticeable that the difference in length of the
vegetative period due to the method of cultivation is fairly constant for
all types, increasing only slightly with the increase of what may be
termed the standard vegetative period of the plant. The seasonal
difference, on the other hand, depends in considerable measure on the
type, being less for early flowering than for late flowering types. While,
therefore, it is possible to reduce two series, differing only in the
method of cultivation, to one standard, this is not possible when a
seasonal difference enters into consideration.

In addition to these two main causes, which, it will be noticed,
affect the entire series, the length of the vegetative period of individual
plants may be influenced through several minor causes and the actual
figures, though accurate in themselves, are thus rendered only approxi-
mate as a record of the standard length. Thus in a few cases the
young flower buds have been observed to fall without opening {vide
note to Table XVII) and an abnormally long vegetative period has been

H. M. Leakb 235

the consequence. Again, dwarfing arises through numerous causes and
leads to delay in the production of the first flowers. In one case plants
of a monopodial type, with a normal vegetative period of over 200 days,
commenced flowering within 100 days from the date of sowing and
before they had been planted out. All cases where any such abnor-
mality is apparent have been omitted from the following records.

The interrelation between the type of branching and the len,gih of
the vegetative period.

The two characters just dealt with are mutually dependent. A
plant of the sympodial type will commence flowering shortly after the
secondary branches have developed, while a plant of the monopodial
type will not flower until the tertiary branches develope. This lengthen-
ing of the vegetative period is shown in Table XV, in which the length
of the vegetative period of some of the more important types are
recorded. The interdependence becomes still more marked when a
continuous series, such as is obtained in the F^ and subsequent genera-
tions of a cross, is considered. For this purpose the plants may be
associated into groups in which the length of their vegetative periods is
similar, each group being formed by the plants which flower during a
ten-day interval. This method has been adopted for the series derived
from the F^ generation of the crosses between types 3 and 4 and
between types 3 and 9, and the results are recorded in Table XVII
(cf. author's note). The figure given opposite each ten-day interval
indicates the average type of branching occurring in plants falling
within that interval and is obtained by adding the numbers indicative
of the type of branching of each plant (100, 75, 50 or 0) and dividing
by the total number of plants.

Tables XXII — XXV show the same interrelation in the ^3 series only
in a slightly diSerent and more detailed manner, the ten stages latterly
recognised in the type of branching as above described, and two- and
five-day intervals being respectively substituted for the four stages and
the ten-day intervals. The closeness of the interrelation is given by
the coefficient of correlation (Davenport (6)). This has been worked
out for the series given in Table XXIV and found to be '6819.

This interrelation, or correlation, is, therefore, a definite fact depen-
dent on the limitation of the flower-producing habit to the sympodial
branches. What appear to be two characters are merely two outward
expressions of the same structural peculiarity. In other words a

236 St'udies in Indian Cotton

definite reason exists for the correlation between these two measurable
and apparently distinct characters, and it is permissible to select the
one that appears to be more reliable for the purpose of recording the
habit of the plants under consideration.

While in neither case has an accurate method of record been
obtainable, the date of appearance of the first flower is at once more
readily determined and obtainable in a larger number of instances.
The measure of the length of the vegetative period, therefore, probably
affords a means of indicating the habit of the plant which is more
accurate than the direct record of the percentage of monopodial second-
ary branches, and has been adopted to record the behaviour of this
character when plants of the two groups are intercrossed.

The habit of the offspring from a cross between monopodial and

sympodial types.

In the ^1 generation derived from a cross between a plant belonging
to a monopodial and one belonging to a sympodial type, the length of
the vegetative period is intermediate between those of the two parental
types. This is shown by Table XXVI in which the relative lengths of
the vegetative periods of the F^ generation and of the two parental
types are detailed. This table further shows that while the Fy^ genera-
tion is intermediate in this respect, it does not hold a position
corresponding to the mean of the two parental values but in all cases
approaches the sympodial type. In this table the seasonal variation is
eliminated by comparison of the F^ generation with the offspring of the
parent plants.

In the F^ generation the plants form a continuous series in which
every stage from early flowering to late flowering forms occurs. It is
noticeable, however, that while those individuals of the F^ series which
have the shortest vegetative period are in flower as soon as, or even
before, the plants of the parental type, in no case does the vegetative
period equal in length that of the monopodial parental type. In other
words, while the full sympodial type appears comparatively frequently
the full monopodial type only rarely does so. The divergence from the
mean length of the parental vegetative periods noticed in the F^ gene-
ration is here even more marked.

Diagram 2 illustrates these results for a single instance of a cross
between a monopodial and a sympodial type. Owing to the seasonal
variation above noted it is impossible to compare the periods for

H. M. Leake

237

successive generations directly, and each must be compared with the
values for the parental series grown in the corresponding season. It is
impossible here to distinguish more than one mode ; there is no trace

F2I909

Parents
1909

Fi 19u-

Parents

n^Ll^,--fet;:^^ I1IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

fe:"^' ::l:::-f— -j-zn: -^

!.::':j.J::' jEili — ^rtrr: ^

! . . ! : . . — -— — ~H"'

;■■;;■; ; 'i ' ■ ^ "T*"

'■■^' ■■ ■-- •---• : - ^-

::;:;■,, ""' : '* ] ' ' ' ' ■ . * " " ' ; : :

-r^-t;-! --|--H-I.I| 1 i ' 1 1 ilijil

U -- ^ ■- -- -^^^^--- ----- ~ - ----^^^

::::.-T"

..... ... . . i

i '■■'■■'' 1 ■■'■ ■ - — — ' ' i ' ~H — -■■■■■'■■■- — 1 j

m ::,'|llli|li|ir :4'!i':' 1 ' \I\±M

■ : ; ■ t ■ ■ — • — — -m (X —^ ■ ;■:■■, -H-

1 1 ' ' ■ 1 ! ' ' 1 ' : ' • 1 ' ! 'itl'li 1' ■ ' ' ■ - • ■ ■ ■ ■ ■ ■■.■■■ ,.,. )., ■ ■ 'I'l — U-|

1908 70 ^° ^ ^00 1^° ^20 130 140 150 160 170 180 190 200 210 220

Type 3 x Type 10
Diagram 2.

of a curve of frequency with three modes such as was found in the
case of the leaf factor, nor has any instance of such a curve been
obtained for the character under consideration.

238 Studies in Indian Cotton

In the present instance there appears to exist an example of partial
dominance combined with incomplete resolution of the component factors
in the subsequent generations. It must, however, be admitted that the
experimental error is undetermined and, from a consideration of Tables
XVIII — XXI\ this would appear to be considerable in comparison with
the magnitudes under measurement, and sufficiently large to render
the character ill adapted to such analysis as has been attempted. The
impossibility of determining this error was in itself sufficient to render
the advisability of attacking this question as a purely theoretical
problem exceedingly doubtful. The behaviour of this character is,
however, a matter of vital practical importance. As has been stated, it
is essential that a plant should be of the sympodial type if its cultiva-
tion in the United Provinces is to be a commercial success. At the
same time the majority of Indian cottons with a really valuable staple
belong to the group of monopodial types. The chief hope of improve-
ment of the cotton crop in the United Provinces, therefore, has been
based on the isolation of pure sympodial forms with the staple of the
monopodial type.

3 (e). The leaf glands.

There frequently occur on the under surface of the leaf one to
three (and rarely four) glands. When there is a single gland it is
situated on the midrib a short distance from the point where this
leaves the petiole. In addition to this gland, two more glands may
occur similarly situated but on the two main lateral veins — giving
three glands in all. The stage in which only one of these laterally
situated glands developes is commonly met with. In one or two cases
only have four glands been observed and in all such cases the addi-
tional gland is situated on the midrib. It is not a condition which
enters into the course of these experiments.

The number of glands is definite and as a leaf character lends itself
to ready determination. But the leaf is a multiple organ of the plant
and it becomes possible for a plant to possess leaves differing in the
number of their glands. Difficulty arises in this case similar to that
met with in the leaf factor, and such as of necessity arises when the

^ Of these tables No. XVIII only is given in extenso. The subsequent Tables
XIX — XXI are abbreviated and give the combined details for those plants the average
dates of flowering of whose offspring fall into successive five-day periods. These three
tables, in their expanded form, agree in all particulars with Table XVIII.

H. M. Leakb 239

character of a multiple organ is employed as a plant character, owing
to the character, definite for the organ, being indefinite for the plant.

It is possible, however, to recognise two distinct forms in which the
leaves are either all eglandular or all glandular. It is true that an
intermediate condition has rarely been observed in which a few of the
leaves may bear a minute and rudimentary gland. The condition is,
however, extremely rare and though the plant would on direct observa-
tion usually be entered as eglandular its true character will be identified
through the occurrence of glandular offspring on selfing.

If plants belonging to a single type and bearing glandular leaves
be arranged in series according to the proportion of leaves bearing 1, 2
or 3 glands, the series will be practically continuous. Nevertheless it
has been found possible to recognise three fairly distinct stages which
have received the following notation :

(1) Glands 1 ; in which all or nearly all the leaves bear a single
gland.

(2) Glands 1 — 3 ; in which the majority of the leaves bear one
gland but those of the main stem and possibly one or two of those of
the monopodial branches bear three glands.

(3) Glands 3 — 1 ; in which the majority of the leaves, including
all those of the main stem and monopodial branches, bear three glands.
A few leaves of the sympodial branches may also bear three glands.

It has been found possible to isolate and grow in a state of purity
forms in which the leaves are eglandular and forms which fall within
the third stage as given above. Plants with the leaf glands 1 — 3 on
the other hand have invariably given mixed offspring^

There remains for consideration the second stage in which the
leaves have a single gland only. This too may occur as an intermediate
condition between the eglandular form and that with glands 3 — 1, and
in such cases does not breed true. It appears probable, however, that
it may also occur as a pure form. Within the author's experience
plants of type 2 have invariably leaves with one gland, but, for reasons
already given, this type has not been very fully investigated and,
perhaps, to an extent hardly sufficient to justify the statement that one
leaf gland is characteristic of the type though there can be little doubt
that a pure form of type 2 so characterised does exist. This pure
form with a single leaf gland does not enter further into the present

' The two cases noted in Table XVIII form apparent exceptions to this statement bat
most, in the absence of farther evidence, be considered as extreme instances of divergence
from the expected ratio.

240 Studies in Indian Cotton

experiments in which the 1-gland stage will be grouped with the 1 — 3
stage to form one intermediate group.

Excluding types 1 — 3 and 11, in all the remaining types two forms
have been isolated which are characterised respectively by the absence
of leaf glands and by the presence of these glands in the 3 — 1 stage,
and both these forms have been pure bred. In type 1 the three forms
have all been observed but their purity or the reverse has not been
tested by experiment ; type 2 has already been dealt with ; in type 3
the 3 — 1 gland form has alone been met with, while of type 11, though
the 3 — 1 form has similarly been the sole one observed, it is impos-
sible to speak with much certainty since the plants on which the
observations have been made are all derived from a single source.

In the cross between type 3 and type 4 to which reference has
already been made an eglandular form of type 4 was used as parent.
This cross, therefore, illustrates the behaviour of this gland character
under the influence of cross-fertilisation and the results are set out in
Tables XXVII and XXVIII. In the ^i generation the plants are
uniformly of the intermediate form (glands 1 — 3) while in the F^
generation the two parental forms reappear. It will be observed from
Table XXVII that while the ratio between the eglandular and glandular
forms agrees closely with the expectation there occurs among the
glandular forms a large excess of that with the glands 3 — 1 and the
same is found to hold among the F^ offspring of the impure F^ parents
(Table XXVIII, last two columns).

That this excess is due to the classification of certain intermediate
forms as pure 3 — 1 forms is proved by the fact that 52 individuals
which had been so characterised were found to be in reality impure.
The F^ plants must in fact be considered as forming continuous series
from the pure eglandular form to the pure glandular 3 — 1 form though,
from the very nature of the case, the former is more readily identified
than the latter. It has been seen that plants with a single leaf gland
occur; and, if the 1 — 3 stage be considered as the full intermediate,
this stage must be considered as an approach to the eglandular con-
dition. In the same manner there appears to occur a stage which
approaches the fully glandular condition sufficiently closely to be with
diflSculty separated from it. By examining the plants at the end of the
season it is possible to distinguish two conditions which may be termed
the 3 — (1) stage, in which even the latest leaves of the monopodia bear
three glands, and the 3 — 1 stage, in which these bear only one or at
most two glands. It is not yet certain, however, that this distinction

H. M. Leake

241

will afford a means of separating the impure forms, nor is it a method
which becomes available till after the work of fertilisation is long over.
At present no method of discriminating with certainty between the
pure and impure forms during the major portion of the life of the plant
has been discovered.

Correlation.

One instance of correlation has already been dealt with in section .3 (d)
on the type of branching and the length of the vegetative period. In
this instance the correlation was seen to depend on a recognisable
feature — the flowers are only formed as a development of the apical
buds of the sympodia the growth of which is carried on by the main
lateral bud. In the present section reference will be made to two
other instances of correlation, but in them the feature on which the
interrelation between the two characters depends is thus not recog-
nisable.

Fig. 3.

There appears to be complete correlation between the size of the
petal and the colour of the flower. If the petals be white in colour
they will be small and hardly project beyond the bracteoles ; on the
other hand, if the colour be yellow, they will be large in length, about
twice that of the bracteoles (vide Figs. 3 and 4)*.

> The difference is well shown by a comparison between Plates 14 a and 16 Watt (20).

242 Studies in Lidian Cotton

The petals may be of one of two sizes, either small, when they lie
within the bracteoles whose length they do not exceed, or large, when
they project beyond and are about double the length of the bracteoles
{vide Figs. 3 and 4). The exact size of the larger petal varies somewhat
with the particular type but in no case approaches that of the smaller,
and the two stand in marked contrast without intermediate form.
There appears to be complete correlation between the size of the petal
and the colour. The smaller petal is invariably white and the larger
petal invariably yellow. Among the plants under experiment, which
now amount to over a hundred thousand, and among cottons under
cultivation in the field no single exception has been observed. The
correlation holds with the simple yellow and white types and also with
those types in which a red colour is superimposed. It follows from
this that all plants with a red on yellow flower, such as type 3, have
large petals, while plants with a red on white flower, such as type 11,
have small petals. The cross between types 3 and 9 illustrates this
point well ; in all cases both plants with red on yellow, and those with
yellow flowers, whether pure or impure, have large petals, while in the
plants with red on white flowers, whether pure or impure, and in those
with white flowers, the petals are small.

A further instance of correlation, and one which is of considerable
importance both practically and on account of its bearing on the argu-
ment of section 3 {d), has been found to exist between the presence of
the red colouring matter and an increase in the length of the vegetative
period. There is a distinct retardation of the commencement of the
flowering period when the red sap colour is present. This is shown in
Table XXIX. In this table the unit is a plant of the F^ generation
and the figure is, for the pure forms, taken as the average of the F^
offspring and, for the impure forms, as the average calculated from
only those ^3 offspring which are, judging by the depth (to lamina), or
absence, of the red colour, pure in this character.

In the light of this correlation it is necessary to reconsider the
results detailed in section 3 {d). In that section attention was drawn
to the monomodal curve as indicating incomplete resolution. No
distinction was, however, made between plants with, and plants without,
the red colouring matter. It would appear possible that a separation
of the plants into two groups dependent on the presence or absence of
the red colouring matter might disclose two trimodal curves, whose
presence is rendered obscure through superposition. Table XXIX,
however, in which such a separation is effected, shows no such trimodal

H. M. Leake 243

curves and it has not been possible to obtain from the records available
any clear indication of their existence. For the present, therefore, it
is impossible to do more than recognise that in this correlation between
the flower colour and the length of the vegetative period may lie the
explanation for the failure of the early and late flowering characters to
fall into line with other Mendelian phenomena.

LITERATURE.

1. Balls, W. L. Joum. of Agricultural Science, Vol. ii. No. 2.

2. Tear Book of Khedivial Agricultural Society, 1909.

3. Bateson, W. MendeTs Principles of Heredity.

4. BuRKiLL, I. H. Joum. and Proc. Asiatic Society of Bengal (New Series),

Vol. m. No. 7, p. 517.

5. Darwin, C. Effects of Gross and Self-fertilization in the Vegetable Kingdom.

6. Davenport, C. B. Statistical Methods.

7. Ftson, p. F. Memoirs of the Department of Agriculture in India (Botanical

Series), VoL ii. No. 6.

8. Gam](IE, G. a. The Indian Cottons.

9. Memoirs of the Department of Agriculture in India (Botanical Series),

Vol. II. No. 2.

10. JoHANNSEN, W. UebcT ErUichkeit in Populationen und in reinen Linien.

Jena, 1903.

11. Leake, H. M. Joum. and Proc. Asiatic Society of Bengal (New Series), VoL rv.

No. 1, p. 13.

12. Joum. and Proc. Asiatic Society of Bengal (New Series), VoL v. Na 1,

p. 23.

13. Middleton, T. H. The Agricultural Ledger, 1895, No. 8.

14. Parlatore, F, Le Specie dei Cotoni-Firenze, 1866.

15. ToDARO, A. Osserv. Sui Specie dei Cotoni coltivati in Palermo, 1863>.

16. Relazione Sulla Cultura dei Cotoni, 1877-78.

17. Prodromus Monographic Generis Gossypii.

18. Thompson, J. V. Proc. Agricultural and Horticultural Society of India, 1841,

Dec., p. 15.

19. Watt, Sir G. Dictionary of the Economic Products of India. Article on

Gossypium.

20. The Wild and Cultivated Cotton Plants of the World.

21. Burkill, I. H. Metnoirs of the Department of Agriculture in India (RotamcsX

Series), VoL L No. 4.

22. Fletcher, F. Joum. of Agrictdtural Science, VoL ii. p. 281.

23. Hartley, C. P. U.S. Department of Agriculture, Bureau of Plant Industry,

BulL No. 22. .

Joam. of Gen. i 17

244 Studies in Indian Cotton

TABLE l\

Flower Colour. Type 4 {yellow coloured) x Type 6 {white flowered).
Fi 68 plants all yellow flowered

\ ratio

109

plants yellow flowered
21

52 plants white flowered
1

F2 plants used as parents

102 6

J yellow 65 35 34 0

^ I white 0 0 11 100

^ No difference has been observed between the direct cross and its reciprocal. The
two have, therefore, been grouped together in this and subsequent tables.
^ Number of offspring too small to be a reliable guide to purity of parent.

TABLE II.

The occurrence of the red colouring matter in vegetative organs.

Types

Coloured
RR and Rr

Colourless
rr

Total

3x 2

106

135

3x 4

224

293

8x 5

299

102

401

3x 8

180

244

3x 9

374

120

494

3x101

351

100

461

Total

1534

484

2018

Batio

3-17

417

1 Determined on flower colour only.

H. M. Leake

245

TABLE III.

The intensity of the. red colouring nuUter in the leaf as an indication

of purity.

Leaf of Fi parent
recordea as

Constitution, as
by Ft offspring,

determined
of the form

ToU

(o) Type 3 x Type 4

Lamina

Veins

Ribs

116

Total

141

204

Ratio

2-2

(6) Type 3 x Type 9

Lamina

Veins

Ribs

188

197

Total

194

275

Ratio

2-4

TABLE IV.

The intensity of the red colouring matter in the petal as an
indication of purity.

Flower of /".parent
recordea as

Constitution, as determined
by Fi offspring, of the form

Total

(a)

Type 3 X Type 4
Red
Red on yellow

28
35

136

30
171

Total
Ratio

63
1

138
2-2

201

(6)

Type 3 x Type 9
Red
Red on yellow

11
46

136

14
182

Total
Ratio

139
8*4

196

17—2

246

Studies in Indian Cotton

TABLE V.

7'Ae F^ generation of crosses between type 3, in which the red colouring
matter is present, and types in which it is absent.

Coloured Colourless

Lamina

Veina

Eibs

Total

(1)

3x2
Ratio

96
9-6

29
2-9

135

(2)

3x4

114

228

4x3

—

Total

771

242

123

147

293

Ratio

(3)

3x5

146

5x3

114

117

255

Total

115

176

174

102

401

Ratio

1-7

(4)

3x8

134

8x3

■08

110

Total

110

121

244

Ratio

2-0

1-1

(5)

3x9

132

137

252

9x3

124

139

242

Total

973

256

276*

121

494

Ratio

2-8

1-2

Grand total

358

739

825

384

1567

Ratio

2-3

1-1

^ 5 of these shown by experiment to be impure.
22 „ „ „ pure.

34 „ ,, ,, impure.

* 22 „ „ „ pure.

Flower colour.

TABLE VI.

Type 3 {red flowered) x Type 4 {yellow flowered).

Fi 38 plants with flowers red on yellow and the red colouring matter extending

to veins.

RR
Fj Foliage (lamina)
77
Ratio 1-1
Used as J 61 lamina
parents 1 2 veins

RR+Rr

(Ribs or Veins)

(Total coloured)

(Colourless)

147

224

2 1

5 lamina

204

136 veins

RR Rr RR+Rr rr

(Lamina) (Ribs or veins) (Total coloured) (Colourless)

Fs 1328 832 1692 2524 773 12451

Ratio 107 2-18 1-90

1 And 4 red plants. A consideration of other characters indicates that 2 of these are

without doubt either volunteer plants or have arisen through an accidental mixing
of seed.

H. M. Leake

247

'^i.

•2

-8.

S 00 OS

O -^ i-H

X X

eo OS

^.^

M !

=^6

f 1

<^ '

,,^

^1^

^ fl

^-.

"-^

„-«

t s

«Doeo

fcfc 1

©»

00
0*

tH "U

2«5

^^_^

_,_^

^ 1

I-l

c— 1

§1

00 '^e*

o«

»o

e«

^ >.

■^0.

«o

•*

c«

^ '*

a-~

1 -

s 1

on
11

r-l

§5
11

" r1

l-l

§5
11

.-( 1-1

S<1

^_^

g a

5
1

r-i

I-l

l:fc &

9
>>

--^

B^^

CO
04

^ s

§
1

oooe
Si

§ 1

i-<

C<1

0»

-HO

I-H

f-l

5
>>

•«*' so

!^ 3

t I

IQO

i-l

a; g

» a

§5

osooee

^ 1

1-1

®«

^11

a
o

eS M

1
3

'«ao

^ 1

i-H

M a

1 flg-

o o

.2 O.C8 *e

?;»

u, ^

o o

o -9

248

Sttidies in Indian Cotton

TABLE VIII.

Flower colour. F^ plants x parents.

(Type 2 x Type 3) x Type 3

{red (lamina) x yellow} x red (lamina)
(Type 2 x Type 8) x Type 2

{red (lamina) x yellow} x yellow
(Type 3 x Type 4) x Type 3

{red (lamina) x yellow} x red (lamina)
(Type 3 x Type 4) x Type 4

{red (lamina) x yellow} x yellow
(Type 3 x Type 9) x Type 3

{red (lamina) x white} x red (lamina)
(Type 3 x Type 9) x Type 9

{red (lamina) x white} x white
(Type 4 x Type 6) x Type 4

{yellow X white} x yellow
(Type 4 x Type 6) x Type 6

{yellow x white} x white

Red on Red on
Red Red white white
(lamina) (veins) (lamina) (veins) yellow

white

— 2

1 —

15 — _ —

2 —

— — 8 —

TABLE IX.

Detail of individual of type having leaf factor 1'88.

Number

of
leaves

Leaf
factor

Main stem

1-82

Secondary branched

(a) Monopodial

arising from leaf 7

1-81

Number

Tertiary

Leaf

branches

leaves

factor

arising from leaf 8 21 1-86

arising from leaf 9 17 1 '90

Sympodial

arising from leaf

1-64

„

1-66

arising from leaf 10

1-72

» >>

1-74

>> >>

1-70

»> >>

1-79

Monopodial

arising from leaf

1-74

j> >>

1-86

H. M. Leake

249

TABLE IX (continued).

arising from leaf 10

arisiiig &om leaf 11

arising from leaf 12

arising from leaf 13

arising from leaf 14

Number

of
leavea

Leaf
factor

Tertiary
branches

Sympodial

Nnmber

of
leaves

factor

arising

from leaf 8

1-78

1-72

1-81

t»

1-66

1-74

»»

1-83

1-76

1-64

1-86

arising

from leaf 8

1-75

>»

1-68

1-74

»>

1-87

1-75

1-91

arising

from leaf 5

180

„

1-84

1-74

1-73

1-69

1-76

1-68

1-71

»>

1-78

„

1-72

1-71

1-84

arising

from leaf 5

1-61

f »

1-65

1-78

„

1-73

1-88

arising

from leaf 5

1-71

»»

1-73

»»

1-76

1-72

»»

1-78

1-76

1-69

1-74

1-77

arising

from leaf 6

1-52

»»

>•

1-57

»>

10
11

1-75
1-79

250

Stiidies in Indian Cotton

TABLE IX

{continiied).

Number

of Leaf
leaves factor

Tertiary
branches

Number

leaves

Leaf
factor

15 1-76

arising from leaf 5

1-73

1-85

1-76

„ 10

1-68

arising from leaf 16 8 1-83

arising from leaf 17 16 1"71

18 13 1-97

arising from leaf 19 16 1-81

arising from leaf 24 14 1*78

Average of leaves on
Monopodia

— 1-84

(6) Sympodial

arising from leaf 20

1-76

1-75

J J

1-66

2-07

1-81

» »

1-73

1-78

1-80

1-79

1-63

1-69

>»

1-66

1-65

1-67

1-71

>»

1-70

1-68

„

J J

1-60

„

1-69

, J

1-85

Average of

leaves

Sympodia

—

1-72

arising from leaf 5 2 1*73

arising from leaf 1 1 1*72

„ 5 2 1-55

„ 7 2 1-79

11 2 1-73

arising from leaf 1 6 1*87

„ 10 1 1-73

arising from leaf 1 5 171

„ 2 1 1-79

„ 6 2 1-57

„ 7 1 1-59

Average of leaves on

tertiary branches — 1'73

H. M. Leake

251

TABLE X.

Variation of the leaf factor toithin the type.

Extremes of
leaf factor

Leaf factor

(average
of offspring)

Nnmber of offspring

used in

determination

Max.

Mtn.

1907

1908

1907

1908

Type 4

1-57

117

1-37

—

1-71

1-27

1-46

1-49

1-92

1-42

1-65

1-73

Type 5

1-80

1-56

1-68

1-73

1-94

1-73

1-78

1-80

1-98

1G9

1-88

1-84

Type 6

1-98

1-69

1-81

1-88

Type 8

3-83

2-96

3-35

3-26

3-64

3-55

3-59

3-71

Type 9

418

318

3-64

407

4-34

3-80

416

—

TABLE XI.

The relation between the leaf factor of the F-^ generation of a cross
and those of the parents.

ParenU

rUen

Leaf factor
of parents

Mean of
parental
leaf factor

Leaf factor
of offspring

Difference

Number
of fi

eed Pc

Seed

PoUen

indiTidoals

Type 3 Type 2

313

1-45

2-29

2-26

-003

, 3

, 4

313

1-45

224

2-21

-003

, 3

, 4

313

1-64

2-38

2-49

+ 011

, 4

, 3

1-46

313

2-29

2-42

+ 013

, 4

, 3

1-64

313

2-38

2-45

+ 0-07

, 3

, 5

313

1-78

2-45

2-70

+ 0-25

, 5

, 3

1-78

313

2-45

—

, 2

. 8

1-46

3-59

2-52

2-18

-0 34

, 2

, 8

1-46

3-34

2-40

2-37

-003

. 8

, 2

3-59

1-46

2-52

2-49

-003

, 8

, 4

3-59

1-64

2-61

2-53

-0 08

. 8

, 4

3-34

1-64

2-49

2-54

+ 0-05

, 4

. 8

1-46

3-59

2-52

2-36

-016

. 4

. 8

1-64

3-34

2-49

2-42

-007

252

Studies in Indian Cotton

TABLE XII.

The leaf factor. Type 3 (1. f. > 3) x Type 4 (1. f. < 2).

Leaf factor 313 1-401.

Mean 2-27.

F\ 15 plants mean leaf factor 2'21.

Leaf factor

>2 and <i3

Number of individuals

Eatio

Mean leaf factor

1-71

143

2-3

2-57

1-3

3-54

Leaf factor

2 and <3-2

>3-2

Individuals used as parents

Ratio

Mean leaf factor ^

1-72

143

2-3

2-62

64
1

3-51

Leaf factor

<2 >2

> 2 and < 3

<3 >3

Number of individuals
Ratio

1222 5

784
. 1

1602
204

791

6 1273

Mean leaf factor

1-71

1-62

2-62

3-42

3-51

1 The mean value between 1'35 and 1*46, the values of the pure lines used in this
experiment. At the time when the original cross was made these two forms had not
been isolated.

^ The value of each individual is here taken as the mean of the values obtained from
its offspring.

H. M. Leake 253

TABLE XIII.

The leaf factor. Re-appearance of parental values in the F^ offspring.

(a) Type 2 x Type 3.

Leaf factor 1-40 3 13.

Mean 2-26.

Fi 21 plants mean leaf factor 2-33

Leaf factor <2 =>2and<3 >3

Number of individnals

40 82

Ratio

3-6 7-4

Mean of leaf factor ...

1-63 2-43

3-41

(6) Type

4 X Type 8.

Leaf factor 1-52 3-47.

Mean 2-49.

28 plants mean leaf factor 2-39

Leaf factor

<2 >2 and <3

Number of individnals

47 102

Ratio

1 2-2

Mean of leaf factor ...

1-66 2-59

3-42

254

Stvdies in Indian Cotton

TABLE XIV. The leaf factor. F^ generation

'2 ^

<l-49
1-50— 1-59
1-60— 1-69
1-70— 1-79
1-80— 1-89
1-91
1-98
2-05
2-06
210— 2-19
2-20— 2-29
2-30— 2-39
2-40— 2-49
2-50— 2-59
2-60— 2-69
2-70— 2-79
2-80— 2-89
2-90— 2-99
3-00— 309
3-10— 315
3-17
3-18
3-19
8-20— 3-29
3-30— 3-39
3.40_3-49
3-50— 3-59
3-60— 3-69
3-70— 3-79
3.80—3-89
3-90— 3-99
4-24

— 12 13 8 1
11 23 36 33 25

6 20 44 70 79

3 13 27 88 148

— 1 5 19 44

2 1

— 4

— 1
1 1

— 4

— 1

— 24
111
111
9 14 14

13 10 14
4 11 13

13 25

14 26
3 7
8 11
3 5
2 2

— 2

— 1

4
2

10 — 1 —

43 13 9 —

131 67 35 8

54 62 35 14

3 4 11

8 10 13 5

32 16 13 10

33 34 22 12
16 13 4 8

27 15
14 12

_ 1 _ 1 _ 2 1

1—1111 —

9 7 15 6 5 7 3

14 24 14 n 10 10 7

8 16 14 17 12 4 2

11 24 29 39 39 31 23

19 30 45 47 50 42 36

6 14 23 23 24 21 20

5 7 24 28 34 35 32

2 8 5 17 28 24 33

1 2 4 10 21 15 15

_ _ _ 3 7 5 11

- 1 1 — 5 8 2

2 1

5 3

4 6

19 16

28 37
19 23

29 28
24 16
16 13

8 14

10 8

2 3

2 5

2 7

3 4
16 13
16 18

4 10

16 11

17 18
14 11

11 8

12 17
2 5

10 9

9 15

17 31

8 12

2 5

3 —
— 1

H. M. Leake 255

from the cross. Type 3 x Type 4.

O- Oa 0°F on

> S S So £ S ^ c('C

0-22 #2-2=»' «a< "

« Til lo «) r- 00 OS o .-I c« eo rH « to «>aD^2ea5o8S^ ^.g £

eb eb as « « « « rh tp -♦ -* ti< Tf >* •*-*c^*>^^o >o --

a S5 •< Z < a

— — — — — — — — — — — — — — — — 2 1-68 84 1-48 +0^

— — — — — — — — — — — — — — — — 8 1-66 149 1-54 +012

— — — — — — — — — — — — — — — — 17 1-77 284 1-64 +0-13

— — — — — — — — — — — — — — — — 24 1-75 522 1*74 +001

— — — — — — — — — — — — — — — — 12 1-81 237 1-84 -003

— — — — — — — — — — — — — — — — 1 1-91 11 1-93 -002

— — — — — — — — — — — — — — — — 1 1-79 46 1-98 -019

— — — — — — — — — — — — — — — — 1 2-26 21 2-05 +0-21

— — —— — — — — — — — — — — — — 1 2-67 9 206 +0-61

5 3 — — 111— 1— — — — — — — 5 2-31 141 218 +013

2 3— 1— 3 1 — ________ 10 2-47 187 225 +0-22

85113— 1 — — 1 — — — — — — 8 2-38 179 235 +0K)3

15 117 4322 11 l______20 2-56 450 246 +010

16 19 13 10 6 8 2 5 2 !______ 24 268 634 2-54 +0-14

7 118 126 44 12 l_ — ____15 2-74 314 264 +010

20 11 17 8 10 10 6 76 8 51 — 1__ 20 2-73 468 2-74 -0K)1

10 14 12775932 23 — — 11— 16 296 315 2 84 +012

6766 10 6484 3 12— — — 1 11 292 219 295 -003

252531443 2—112—— 4 322 126 3-06 +0-16

14 64 10 55243 1_____1 7 302 131 312 -0-10

2 12 11__— __ — — __ — — 1 3-51 19 317 +034

1 1 1 2 — — — — — — — — — — ~— 1 291 8 318 -0-27

_2_-____ — — — — — — — — 1 3-22 4 319 +0-08

954 5 2 3 — — — — — — — — — — 3 3 47 61 323 +0-24

30 23 23 12 9 3 2 — — — — — — — — — 7344 145 3-33 +011

36 47 27 29 18 17 12 6 2 3 1 _ _ _ _ _ 14 3-54 274 3-44 +011

7 23 33 25 21 18 10 8 9 2 12 1 — _ — 10 362 192 3-55 +007

23 24 44 40 33 25 18 18 8 10 5 — 1 — 1 — 13 372 274 365 +007

5 10 12 12 14 11 15 7 9 9 4 2 — — — — 6 387 119 375 +0-12

149469366 86112— — 4 3-95 70 3-85 +0-02

166656 17 89 11 74-2 — 2 3 399 91 3-93 +006

— — — —1122112123 — 1 1 3 92 17 4-24 -0-32

256

Studies in Indian Cotton

TABLE XV.

The length of the vegetative period as affected by the season.

Sown in

Pots

Field

Season

1907

1908

Difference
= seasonal
variation

1909

Type 3

146

210

+ 64

146

» 4

111

+ 28

» 5

110

+ 30

„ 6

114

+ 24

» 7

—

106

+ 28

,. 9

117

+ 24

,, 10

115

+ 19

Seasonal variation for monopodial types 64.

„ ,, ,, sympodial types 25.

TABLE XVI.

The length of the vegetative period as affected hy the method of

cultivation.

Type3xType4

Type3xType9

Pots

Field

Number

of
plants Days

Ditfer-
ence

Pots

Field

Number

of
plants Days

Period in
Days

Number

of
plants

Days

Number
of

plants

Days

DifTer-
ence

Below 101

105

— —

—

127

— —

—

101—110

108

30 85

109

8 88

111—120

115

52 89

118

30 93

121—130

122

36 96

122

41 98

131—140

126

31 95

130

20 102

141—150

127

19 98

132

9 105

151—160

132

3 104

132

—

161 and over 2

123

__ —

—

135

_ _

—

H. M. Leake

257

TABLE XVII.

The length of the vegetative period. The F^ generation from the cross.
Type 3 x Type 4 and Type 3 x Type 9.

Type3xType4

Type

3xTypc9

Interrals

Number of days from sowing

to flowering

Number

of
plants

Number
indicative
of type of
branching

Number
of

plants

Number
indiottlTe
of type of
branching

Above 170

—

166—170

161—165

156—160

151 — 155

146—150

141—145

136—140

131—135

126—130

121—125

116—120

111—115

—

106—110

—

101—105

—

below 101

—

In this table, which is based on the F-i generation, only the four degrees, indicated by
the numbers 100, 75, 50, 0, of secondary branching have been recognised.

' Of these five plants two are of the sympodial type. One of these was dwarfed, and
the date of appearance of first flower consequently very late. The second produced
flowers at the extremities of the sympodial branches only. On account of these two
plants, which appear somewhat normal, the figure for this interval is abnormally low.

258

Studies in Indian Cotton

TABLE XVIII. The length of the vegetative period. The F.^ generation

_, 01 'S >
.2 2 <=iS. ^

QbQ0a6ciC>oiO5Ci

— — 2

— — 1

— 1 —

106
128
115
281
282
110
300
329
104
118
280
150
241
320
151
131
327
142
284
114

65
219
170
119
139
163
127
138

87
166
162
324
156
314
112
242
195
153
158
274
181
140 — — — — — —

210 — — — — — —

225 — — — — — —

134 — — ___ —
264 _ — _ — __
264 — — — ___
214 ______

221 ______

1 — —

3 —

3
2
1
2
3
4

1 _ 1 _____
____1 i__

3_2 l___ —
__3_1 3 l_

__2 l__l_
2122121_

14 5 1 —
11111

1 — 2 —

— 212
11— —

— — 1 —
1 _ _ —

1 — 1 _

1 _

— 1

2 —

3 1

— 1

1 —

2 1
2 1

— 1

111111

— 1 1 2 — 6

— _ 1 _ 2 1
1 _ _ 1 — 1

— _ _ 2 _ 1
1 _ 1 — _ 1

— 1 2 1 _ 1

— — 1 4 _ 1
1 1 _ _ 1 1

— — 1 — 1 1

— — — — 1 1

— _ 1 1 _ 2
1 _ _ _ 1 _

_ 2 — 1 1 1

— 1 1 1 — —

— — 1
2 1 —

— 12

— — 1
2 — 1

— 1 —
113
11-

— — 1

1 — —
— — 1

- — 1

- — 1
4 — 1 -

— — 2 1 -

— 11 — 1

1 —

- 3

— 1
1 —

4 2 2

— 2 —
111
1 — —

— 42

— — 1
2

1 — 1

2 —
— 1

2 —

1 —
1 1

1 —
1

— 2

— 1
1 1

— 3
4 —

1 —

— 2
1 —
1 1

— 1

— 3
3 3

1 —

2 5

H. M. Lkake 259

from ihA cross. Type 3 x Tt/pe 4. Pot Series.

o V ^ o flo o^g-25 2

1"^ 5 g

_________________ _ _ _ 10 107 -10 97

__________ — — — — — — — — — _ 7 112 -14 98

_1 _______ — -___ — — — — — — 10 101 +2 103

_________ _ __________ 18 100 +3 103

__________ — _ — — _ — _ — — — 22 99 +5 104

_1 _______ — __ — — ___ — — — 10 110 -5 105

__________ — — — — — — _ — _— 13 107 -2 105

__________ — __ — — — — — — — 24 113 -7 106

___ 1 ________________ 10 109 -2 107

________________ — — — — 21 102 +5 107

__2— — — — — — — — — — — — — — — — — 22 98+9 107

1 ___________ — — _ — _ — — — 5 113 -5 108

_ ___________________ 6 132 -22 110

________________ _ __ 19 118 -8 110

__ 1 _ 1 ______________ 22 115 -5 110

1_ — __1 ____ — — — _ — ___— 22 115 -5 110

___________________ 12 114 -4 110

____________________ 8 111-1 110

__ — — _ — — _________ — __ 14 109+1 110

___________________ 21 109+1 110

_ _ _________________ 16 108 +2 110

__1— ___ — — — ____ — — — __ 11 1-24 -13 111

1 _________ 1 ________ 21 114 -3 111

_1 1 ___________ _____ 12 119 -7 112

1 1 2 1 l____l_ — _______ 25 117 -5 112

___________________ 8 113-1 112

___________________ 22 112 — 112

1 — 1 ________________ 5 117 -4 113

_1_____ — _ — — — — __ — — — — 17 131 -16 115

— 1 _ — ___ — ___________ 9 115 — 115

2 1 1___1 _!__________ 31 113 +2 115

_1 !________________ 13 119 -3 116

1 1— — — __ — _l_ — _ — _ — ___ 19 119 -3 116

__ — ___—_____ — — _____ 7 116 — 116

1 ___ — _ — —— 1 !_____ — — _ 11 108 +8 116

2— —- — _ — __ — _— — ______ 11 132 -15 117

1 2 _ I ____ 1 __________ 19 121 -7 117

2 1___— !____________ 16 118-1 117

2 — 1_1— — — — — _____ — _ — _ 14 115 +2 117

2_1 1 1— _ — _ — — __ — — ___— 16 155 -37 118

1 — 1 1 1 — 1- l__________ '20 127 -9 118

4 3 2 l_l _____________ 43 120 -2 118

2 _ _ — 1 1 ___ — 1 _____ — — — 16 130 -11 119

— 2 1__ — — — —— — — _______ 24 128 -9 119

1— _1 1 — !____________ 11 119 _ 119

2 — 1 — 1 1 _____________ 13 148 -28 120

1_2__1____— 1 _ — ___ — — 18 141 -21 120

4 6 2 _— 1— — _1 ____ — __ — _ 36 130 -10 120

1— — — 1— — __ — _ — _______ 9 126 -6 190

Joam. of G^en. i 18

260 Studies in Indian Cotton

TABLE XVIII

2^ c^

"^££.S'^°°00<30<»<'*^05(3505OOOOOf-lr-l^.-(i-l5<IIM

les _______! 2 — 1— — — 1 l____l_4

100 — ________________3 1 1 1 1 1

273 __________________ 1 1 3 1_

335 ______________ 1 _____! 1 1

71 _________________l_l_l_

187 ___________l 2 3__1_1 — 1 3 2

182 _ _ ____________ 1 — 2 l___l 1

132 ______________l__2 2__3 2

147 _________-____l 2__— 1 2_1 1

90 _____________ 1 ____ 2 _ 3 _ 1

179 ___________ 1 _______ 4 _— 1

310 — — — — — — — — — — — — — 2 2__1__1 6 4

97 ________________ 1 _____ 3

263 ______________2_ — _2 2 1___

224 _____________ 1 _______ 2 —

223 ___________i___i 3___l 1 3

328 ______________ 1 __ 1 ___ 1 _

155 _ — _ — ___________2 — 2— — 2 1 1

275 _____________ l______2 2 2

259 __ — _ — _____l_l____l 1 1 1 — 3

246 _ — ____ — ______l____2_2 3 1

75 _ _ _______________ 1 1 i_l 1

193 _________________i_4— — 1

194 _ _ ______________ 1 _ 1 __ 2 2

266 —-_ — ________— _____l_2 3 2

249 _ _ _____________ 1 __!____

185 ______________ 1 2 1 l__l_l

271 ____________________ 1 2 4

235 __________________ _ __ 1 1

227 ________ _ ______ 1 ___ 1 _ 1 —

77 ________________ i___ 2 1 1

245 ____________ 1 _________ —

176 ________________ 1 ______

215 _______________________

212 __________1____1_1_1 2 1 1

175 _________________ i_____

272 _______________ _ ____1_ —

256 _______________ ___1— — — 1

345 ____________________ l_l

229 ____________ — __ — — — — __1 1

251 ________________ l___l 1 3

89 —____ — -___ — __ — _-_ — _ — — — —

255 ___________—_ — _____ — — — 1

269 _______________ — __ — — — — —

93 _________________ _ _ _ __1

220 __________ — _ — — — — _— 1 ___ 1

245 _______________ — _ — — — — — —

244 ________________ — — — — —— —

240 _________ — — — __ — _ — — — — — — —

H. M. Leakb 261

(continued).

ra ° s \
a t, '^ * ti

,.Hf->l<Hl-lf-tl-ll-l<-ll-l<Hl-l'Hf-ll-ll-l>-Hl-l>-)>-I^Og^-g C

4 2 1 — 1 i__2— — — — — — — — — — — 22 117 +3 120

1_— 1 — 1- — — — — — — — — — — — — — 11 117 +3 120

2 4__ — — — — — — — — — — — — — — — — 12 156 -35 121

— __ 2 — — — — — — — — — — — — — ——— 6 124 -3 121

3 1 ______ — — — — — — — — — — — — 7 123 -2 121

3 — 1 2 — 3 1— — — 1— — — — — — — — — 25 122-1 121

2 1 1— _l 1— — __ — — — — — _ — — _ 12 121 — 121

_2 2 l___ — — — — — — — — — — — — — 15 120+1 121

2 3 1 1 1— — — — — — — — — — — — — — — 16 119 +2 121

__2 1 1— — 1— — — — — — — — — — — — 12 141 -19 122

_1 2— — 1— — 1— — — — — — — — — — — 11 129 -7 1-22

7 4_2 1— — 1— — — _ — — _ — _ — — — 31 112 +10 122

1 i__— 1— — — — — — — — — — — — _— 7 166 -43 123

— 3 2 1 2 1— — — — — 1— __ — — — — _ 17 141 -18 123

— 2 1 — — 1 — — — — — — — — — — — — — — 7 128 -5 123

3 1 1 1 — 4— — — 1— — — — — — — — — — 22 128 -5 123

1 1 _________! ___ — __ — _ 6 118 +5 123

1 3 — 1 — _ — — 1 — - 1 — — — — — — — — 15 118 +5 123

1 l___i____i _ — _ — — — — — — 11 163 -39 124

1 2 2 — 2 — 1 1 — — 1 _ _ 1 _ — — — — — 20 143 -19 124

— 5 5 __ 2 — — — — — I — — — — — — — — 22 133 - 8 125

_3 3 — 1— — — 1— — — — — — — — — — — 13 133 -8 125

1 — 2 — — — — — — — — 2 — — — — — — — — 11 130 - 5 125

2 1 — — — 1 — — — — — 2 — — — — — — — — 12 129 - 4 125

— 1 — — 3 — 1 — — 1 — — — — — — — — — — 14 141 -15 126

1 5 2— — 1— — — — 1 — 1— — — — — — — 13 132 -6 126

2 4 6 3 — — — — — — 1 2— — — — — — — — 25 126 — 126

1 5_ — — _ — _2 — 1— — — — — — — — — 16 153 -26 127

1 1 1 1 i_ — — _ — — — — — __ — — — — 7 133 -6 127

_1 — 1— __ — _ — — 1— — — — — — — — 6 131 -4 127

1 — 4 1 2 2 1 1— — — — — — — — — — — — 17 134 -6 128

2— — 11— — 11— —— — — — — — — — — 7 133 -5 128

1 11— — 1 — I— — — — — — — — — — — — 6 129-1 128

4 — 1 — — 1 ——— — — — — — — — — — — — 6 126 +2 128

4 2 4 5 4 4 1 1 1 — 1— — — 1 _— _ — — 36 124 +4 128

— — 1 1 1 1 — 1— — _______ — — — 6 129 — 129

1 —__ — ______ 1 _ _ ______ 3 155 -25 130

1 1 — 1— — 1 — 1 — i_ — _______ 8 148 -18 130

— 2— — — — — — — — 2— — — — — — — — — 6 135 -5 130

4 5 4 1 1 1 — 4— — — 1— — 1— — — — — 24 134 -3 131

1 1 2 3 2 — 1— — —— — — 3 1 3— — — — 23 150 -17 133

2 2 — 4 1 2 1 1 — 1— — 1— — — — — — 1 16 143 -9 134

•2 2 1 — 1 1— — — 1 i_i_i___ — — 12 140 -6 134

1 — _— 1 _ 1 ____ 1 _____ — — — 4 155 -20 135

— 1 2 1 2 1 1 1 — 1 4_____ — — — — 15 153 -18 135

— 2 1 3 — 2— — — 1 1 2— — — 1 1— — — 16 1-29 +6 135

— 2 — 2 1 1— — 1— — — 1 — 2— — — — — 10 138 — 138

— 2 1 1 1 1 — 1 1 2— — 1 — 1 — 1— — — 13 139 — 139
_____1 1 1 ______ 1____— 4 133 +7 140

18—2

262

Studies in Indian Cotton

TABLE XIX. The hngth of the vegetative period. The F, generation

«0 CO CO O

05 0> 05

Below 80 — — — — — — — 2 3 4 3 4 1 l_— 1 l___ —

80— 84 — — — 1 2 2 13 8 32 43 74 72 66 50 45 18 19 14 13 12 10 6

85— 89 — — — 1 — 1 6 8 15 44 68 83 78 59 57 40 31 33 45 30 33 28

90— 94 1 _ _ _ _ 1 _ 2 12 19 30 61 63 65 62 52 66 58 63 80 76 70

95— 99 2 1 — _ — _ _ 3 i 5 13 18 26 31 39 44 32 35 68 75 9t 97

100—104 — — — — — — — — 1— 1 2 3 3 1 6 5 7 16 20 30 27

105—109 — — — — — — — — — — — 1___ 1 1 1 2 3 10 12

110—114 — — — — ___________ !_____ 1

TABLE XX. The length of the vegetative period. The F^ generation

O rH 1— I r-4

Below 104— ___3 21 1— 1— 1 1— 1 1___ — — —

105—109 — 1235 775 57 13 76 17 12 16 13 651— 2

110—114 — — 1 6 3 10 7 8 16 17 21 20 21 31 31 34 46 35 23 26 30 9

115—119 — — _ 2 7 8 5 9 8 21 20 22 18 34 46 58 79 76 57 64 60 36

120—124 ———14 125 7 7 9 10 11 11 16 40 36 53 44 63 70 46

125—129 — — — — 1 12 2 4 2 3 3 4 5 10 26 32 42 41 63 86 66

130—134 — _________ 2 1 2 1 2 9 14 16 17 30 44 42

135—139 — — — ___________ — — — 1 3 3 9 11

140—144 — _____________ 1— 2— 1 4 8 5

145—149 — — — — — — — — — — — — — — — — — — — — — —

150—154 — —__________ — — — — — — — — — —

TABLE XXI. The length of the vegetative period. The F^ generation

Below 80 — — — 1 — 1 1 2 3 l_ — — — — ____ — — —

80— 84 — — 1317 852434517113—1 — —

85— 89—1 1 1 2 6 11 17 24 18 19 16 18 15 13 22 13 14 8 8 3 2

90— 94 1 — — 1 3 8 14 20 27 18 24 24 16 29 40 42 35 48 27 22 21 11

95_ 99 _ _ _ _ 1 _ 7 6 13 4 6 14 9 17 19 35 24 33 26 26 18 14

100—104 — — 1 — 1 1 3 2 5 4 9 5 6 3 11 14 25 30 17 22 19 21

105—109 1— — — — — — — — 2 1 4 1 6 5 4 14 11 15 17 13 11

110—114 — — — — — — — — — — — — — — 2— 1 2 2 3 4 6

115—119 — — — — — — — — — — — — — — — — — — — 1 4 3

H. M. Leakk 263

from the cross. Type 3 x Type 9. Pot Series.

O-4-^^-H-HlM?JCq5<ir»50 33 35 3523S*9

— — — — — — — — — — — — — — — — — — 1 106 20 79-27

6 2 2 — — — — — — — — — — — — — — — 33 112 512 83-29

12 13 3 3 — — — — — — — — — — — — — — 37 116 691 87 -29

56 26 30 16 12 5 5 2 3 1 ____ _ ___ 40 123 942 92-31

63 37 46 33 16 12 5 3 4 4 2 l _ _ 2 — — — 43 125 815 97 -23

13 9 20 12 11 8 6 3 1 3 — 2 1 1 _ _ — _ 14 136 212 102 -34
4 17 21 7 14 5 3 1 — 2 — 2 — 1 — 1 — — 6 137 109 106 -31

— 1 1 — — 11 — 2 1 1 _ _ _ _ 1 _ _ 2 127 11 111 -16

/rom the cross. Type 3 x Type 9. Pot Series.

->^ ^*- ^'^ ^ i-
o *» ""J" as » o ■?» -* 5s X o CO •* to X o~>-22dCi-^^
35 33 35 33 35 -^ ■* ~>< •* ■>*" lo >o lo o «o «rio3*S;«*>5

Z -sj Z^

C v. ^2

z^ ^^ S

— — — — — — — — — — — — — — — — — — 1 117 12 101 -16

— 2 — — — — — — — — — — — — — — — — 10 119 142 108 -11

10 5 2 1 2 — — — — 3 — — — — — — — — 2S 120 418 113 - 7

25 24 13 10 5 3 4 2 — 4 4 — — — — — — — 47 122 724 117 - 5

42 39 24 15 9 15 5 6 8 6 2 i __ _ i _ _ l 41 126 610 122 - 4

78 47 37 31 20 25 19 16 19 18 20 2 1 2 1 — — — 47 130 729 127 - 3

40 48 24 28 21 18 19 32 25 9 11 12 8 4 2 — — 2 37 132 483 133 + 1

10 11 8 5 5 9 7 9 8 7 7 3 — 5 2 1 12 9 139 127 136 - 3

5 9 12 14 6 7 9 10 10 12 11 10 6 11 4 2 5 7 12 142 171 141 - 1

_ 211 — — 2741232 2 — — 2— 4 137 31 147 +10

— — — — — — — — — 12 — — 1— 11— 1154 6 154 —

from the cross. Type 3 x Type 9. Field Series.

- z"" ^-^ z= ^ =

X O 0» 'I* «0 X O CO -* so X O 9« ■* O X o*?-2£e32-5a«39<

— — — — — — — — __________ 1

— — — — — — — — __________ 3

1 1 — — — — — — — — ________ 13

8 7 1 3— — 1— — — — — — — — ___ 33

4 10 1 6 3 2 3 2 1— _l !_____ 20

12 13 9731831 151— — — — — 1 20

11 786745625323- — — — 4 14

2 3 2 7 4 4 2 4 — 2 2 3 1 1 _ _ _ _ 5 138 57 112 -26
1 11 — 3 1 3 — 2 1 — 3 — 1— — — — 2 140 25 115 -26

111

-34

116

-33

115

234

-28

124

451

-32

124

306

-27

129

264

102

-27

130

178

107

-23

264

Studies in Indian Cotton

TABLE XXII.

The interrelation between the length of the vegetative period and the type

of branching. Type 3 x Type 4. Pot Series.

Branching

ofTays'^ 10 20 30 40 50 60 70 80 90 ^'^^

Below 90 3 6 2 — — — — — — 19 19

92 2 4 4 1— l___27 ~

94— 8 4 !_____ 24 26

95 2 1 10 2 — — — — — 32

96 — — 8 1______

98— 3 4 3 — — — — — 30 31

100 2 4 4 4 3 i___33

102 — 4 2 1 2 1 _ _ _ 34

104 2 3 4 5 2 — — — — 31 34

105 1 3 5 2 4 2 — — — 36

106— 3 4 2 1 2 — — — —

108 2 5 12 5 4 — — — — 31 33

110 — 3 4 3 1 1 _ _ — 34

112 1 4 7 11 3 — 1 _ _ 86

114 1 1 5 3 4 3 1 _ — 42 35

115 — 4 5 1 1 1 _ _ _ 35

116 — _ 5 — — 1 1___

118 — 1 8 11 5 3 4 1 — 45 44

120 — — 6 12 3 2 2 — — 43

122 — 4 4 7 7 4 4 1 — 46

124 — 2 4 4 3 5 1 i — 46 47

125 — — 4 2 7 6 1 _ — 50

126— 15 64272 — —

128 — — 2 2 3 6 7 4 — 61 56

130 1 — 2 4 4 4 6 2 1 58

132 — — — 2 9 4 5 1 1 59

134 — — _ _ 2 2 7 2 — 67 62

135 — — — 1 2 1 2 — 1 64

136 — — _ — — 2 2 1 — —

138 — — 1 1 — 1 1 1 — 56 64

140 — — — — — 2 1 — 1 70

142 — — — — 1 1 1 3 1 73

144 — — — — 1 3 — — 1 64 65

145 — 1 _ _ _ 1 1 4 — 65

146 — — — — — 4 1 — — —

148 — — — — — 1 1 3 — 74 68

150 — — — — — 2 1 1 — 67

152 — — — — — 1 2 — — 67

154 — — — — 1 — 2 1 — 67 66

155 — — ~ — — 1 _ _ _ 75

156 — — — — — — — — 1 — 80

158 — — — — — — 1 _ — 70

^rioF 1 ^^^ ^^^ ^^^ ^^^ ^^^ ^^^ ^^^ ^^^ '^^'^

H. M. Leake 265

TABLE XXIII.

The interrelation between the length of the vegetative period and the type
of branching. Type 3 x Type 4. Field Series.

Brandling
Namber iQ 20 30 40 50 60 70 80 90 ^^ ^

aidt^B

Below 65— i___ — — — — 15 20

66 1_ — — — — — — — —

68— 1 1_ — — — — — 25 18

70 3 2 i__ — — — — 16

72 4 6 3__ — — — — 19

74 7 11 2 — 1 __ — — 19 19

75 7 15 1 . 2 — — — — — 20

76 10 17 8 2 — — — — — —

78 27 40 29 7 1 — — — — 21 21

80 46 65 32 12 1 1 i _ _ 21

82 30 76 58 16 7 6 — — — 25

84 24 59 44 16 9 6 2 — — 27 26

85 6 25 25 13 2 i _ _ — 32

86 3 19 18 15 4 6 5 2 — —

88 17 33 35 26 13 11 4 2 1 33 36

90 7 17 19 22 12 11 5 2 1 39

92 1 5 20 15 16 13 7 13 i _ 40

94 5 16 20 15 10 10 6 1 — 38 39

95 3 6 11 8 4 7 4 — — 41

96 3 7 16 12 13 8 3 2 1 —

98 6 12 24 31 18 18 10 5 — 43 44

100 4 14 21 30 23 31 16 4 2 47

102 8 8 18 28 32 20 20 3 2 47

104 5 9 10 34 25 25 9 1 3 46 49

105 — 2 5 5 8 10 4 1 2 50

106 42379 11 51 — —

108 — 4 9 12 10 19 12 3 — 52 52

110 — — 2 12 9 18 9 3 1 56

112 — — 1 8 13 10 8 1 1 55

114 1 1 — 4 5 12 7 2 — 56 55

115 — — — 4 1 3 1 _ _ 52

116 — — — — 2— 1___

118 — — 1 1 2 6 5 1 — 6058

120 — — — 2 4 2 3 — — 55

122 — — 1 — 2 — 4 1 — 61

124 — — — 1 1 4 2 1 — 61 61

125 — — — 1 _ _ 2 — — 55

126 — — — — — — — 1__

128 — — 1__ i___45 59

ISO — — —■_ — — _ — —

182 — — — — 1 1__ 16666

134 — — — — l__ i_65

^^^Jl® I 83 84 88 94 98 100 101 104 104

266

Studies in Indian Cotton

TABLE XXIV.

The interrelation between the length of the vegetative period and the iyj>e
of branching. Type 3 x Type 9. Pot iSeries.

Branching

Number
of days

2 day
period

5 day
period

Below 85

—

26 -

—

25 -

—

100

—

33 -

102

—

104

—

105

—

31 -

106

—

108

—

110

—

33 -

112

—

114

—

115

—

37 -

116

—

118

—

120

—

40 -

122

—

124

—

125

—

50 -

126

—

128

—

130

—

55 -

132

—

134

—

135

—

57 -

136

—

138

—

140

—

61 -

142

—

144

—

145

—

64 -

146

—

148

—

150

—

65 -

152

—

154

—

155

—

64 -

156

—

158

—

160

—

70 -

1G2

—

164

—

165

—

74 -

Average [
period j

116

110

113

118

123

129

137

144

—

H. M. Leake

267

TABLE XXV.

The interrelation b'-tioeen tlie leiu/lh of the vegetative period and the type
of branching. Type 3 x Type 9. Field Series.

^rsDching

Noinber
of <U]r8

2 day
period

5d«r
period

Below 05

—

29 -

—

27 -

—

27 -

—

32 -

—

31 -

—

39 -

—

100

—

43 -

102

—

104

«>

—

105

—

44 -

106

—

108

—

110

—

50 -

112

—

114

—

C .

—

115

—

58 -

116

—

118

—

120

—

47 -

122

—

124

—

135

126

128

—

1
1

1
2

—

65 -

130

—

66 -

132

—

134

—

Over 135

—

136

—

138

—

140

—

Average )
period j

104

113

121

—

268

Studies in Indian Cotton

TABLE XXVI.

The relatioii between the length of the vegetative period of the F-^
generation of a cross and those of the parents.

Monopodial Parent

Sympodial Parent

i^i

generation

Mean of
Parents

^A^

Type Days

Type

Days

Maximum

Minimum

Mean

Diffen

Type 3 146

Type 4

114

108

„ 3 146

,, 5

113

111

„ 3 146

» 8

112

118

„ 3 146

„ 9

120

123

105

„ 3 146

„ 7

104

TABLE XXVII.

The leaf glands. F^ and F^ generations of the cross. Type 3 (leaf glands
3 — 1) X Type 4 (leaf glands 0).

Leaf glands 3 — 1 Leaf gland 0

Fi 15 plants Leaf gland 1 — 3

F<i Leaf glands

1—3

3—1

Total glandular

Number of individual . . .

113

100

213

Katio

1-7

1-6

Used as parents

107

197

Pure Impure
60 4

Pure Impure
2 105

Pure Impure
38 52

Corrected distribution ... 62

Expectation 65

161
130

38
65

201
195

H. M. Leake

269

TABLE XXVIII.
The leaf glands. The F^ generation of the cross. Type 3 x Type 4.

Leaf glanda

ToUl Total F3

Character of Fj No. of «rlan- indivi- ExpecU- Character of F^

parent plants 0 1—3 3—1 dfular duals tlon offspring

60 1381 — —

— 1381

Leaf glands 0

1 8
1 4
1 32
1 18

4
2
0
1

0
1

1
0

—

818

872

Leaf glands 0

4 62

Leaf glands 1 — 3

1 —

—

1080

1744

Leaf glands 1 — 3

—

Expectation ...

105

557
567

822
1134

899
567

1712
1701

—

Leaf glands 3 — 1
Expectation ...

199

278

222

556

690

278

912
834

1591

872

Leaf glands 3 — 1

38 — — 679

679 — Leaf glands 3—1

270

Stiulies in Indian Cotton

TABLE XXrX. Correlation between the presence oj the red

Jjeai iMTit^aoo<N-*^aooiM-<*i50Qoo

Flower colour colour ^^'^'''SSSSSSSIhS^S

Type 3 x Type 4
Pure forms

Red on Yellow lamina — — — — — — — — — 1 — 3 4 6 4

Yellow green ______4 4 12 20 6 5 5— 1

Impure forms

(Red on yellow lamina — — __ — _ 1— — 3 1 2 3 10 12

JYellow green 1 1 1 _ 4 3 — 5 11 9 15 11 18 11 4

Type 3 x Type 9
Pure forms

Red on yellow lamina — — — — — — — — — — — — — 2 1

Red on white lamina — — — — — — - — — — — — 1 1 1 —

Yellow green _________ 1 2 1 3 3 1

White green _____i_l_ 2 2— 4 3 1

(a) In one character only.
Impure forms

(Red on yellow lamina — — — — — — — — — — — — 1 5 —

/Red on white lamina — — — — — — — — 1 — — 1 — 2 —

(Red on yellow lamina — — — — — — — — — — — — 2 — —

(Yellow green _____1__ 1 2— 1 3 1 —

(R«d on white lamina — — — — — — — — 1 — 1 1 2 — 2

(White green _____1 — 1 i_3 2 2 — —

(fc) In both characters.

Red on yellow lamina — — — — — — — — — — 1 3 2 2 2

Red on white lamina — — — — — — — — — 2 2 1 1 3 3

Yellow green __ — — — 221 8265251

White green __ — _3 2 — 2 5 1 3 4 1— 1

H. M. Leake 271

colouring matter of lite sap and a lengthened vegetative period.

6 77 5 43 4 4 1 1__________ 124

3 2 — — 1_______________ 111

1111 7 8 15 8 7 4 2 1 1— ________ 125

4 3—2 21 1_____________ 113

1 12__1 3 1__1 1________ 131

__1 2 14 l_l___l___l___ 129

_ 13_ 2_ !_____ — _____ _ — 119

2 21_— 1— — — — — _1— __ — _ — — 117

1_31 4423__5 1 — 1— __ — _ 131

— 14 4 22 4 1 3_3 2 !_______ 132

1 1_2 22 2__1 1 1 1 l______ 133

1 3 1 3 __ — __— — ___ — — ____ 120

3 12 1 12 1 l_________i__ 124

4 2 2— — — 1 1 ________ ____ 119

6 44 5 22 6— — 1 2__ — — _ — ___ 125

1 31 2 32_— 1 — 1 l_l______ 125

1 3 2 2_2_ — _ — — _ — — — _____ 114

3—3 1— — _ — — 1 __________ 113

272 Studies in Indian Cotton

PLATE XXXV.

EXPLANATION OF FIGURES.

Petal Colours— Cross Type 3 x Type 9.

1. Parent-Type 3.

2. Parent-Type 9.

3. Fi generation of cross.
4—9. F2 generation of cross.

4. Full red form — pure or giving 4 and 6.

5. Impure red form — giving 4, 5 and 8 or 4, 5, 6, 7, 8 and 9.

6. Pure red on white form — giving 6 only.

7. Impure red on white form — giving 6, 7 and 9.

8. Yellow form — pure or giving 8 and 9.

9. Pure white form — giving 9 only.

JOURNAL OF GENETICS, VOL I. NO. 3

PLATE XXXV

1»~

HEREDITY AND THE JEW.

By REDCLIFFE N. SALAMAN, M.D.

The object of this paper is to lay before Anthropologists some
results in the domain of Ethnology which, though arrived at by methods
as yet foreign to anthropological research, promise a rich harvest in
every direction. Mendelian methods, by which is meant the analytical
observation of specific characters in the individuals and their occurrence
in the immediate offspring, have for the last decade been the all-powerful
weapons of the modem student of heredity. To the Botanist and
Zoologist who can plan his experiments as he will, the results have been
immediate and surpassingly important. To the student of mankind,
whether he be the anthropologist or the medical man, the application
of the method is of necessity limited. It is impossible to frame his
experiments according to design and it remains with the enquirer to
search out from the confused mass of facts those which conform most
nearly to the requirements of an experiment

Painstaking collections of family histories and pedigrees have
already shown that in man several abnormal conditions behave as unit
characters. A classic example is that of Brachydactylism(5) in which
the deformed hand condition is dominant to the normal. The principles
which underlie Mendelian research are well known and need no
repetition here. So far this type of research has hardly been applied
to man except with respect to diseased or abnormal conditions of one
sort or another. Some opponents of Mendelism have directe<l their
criticism to the fact that when black mates with white in man, the
offspring is a blend, and in future generations complete segregation
does not occur. From this observation some have gone further and
implied that to so complex and withal hybrid an animal as man, such
crude principles as those of Mendelism could not be expected to hold
good. The work of G. C. Davenport and C. B. Davenport(4) on the
mating of negroes and whites shows that the problem is by no means

274 Heredity and the Jew

hopeless, and that the apparent absence of segregation on Mendelian
lines is due to the fact that the difference between black and
white is a matter not of one factor but probably of a series of
distinct colour factors. Thus if it be assumed that the colour of the
negro differed by four such positive characters from that of the white
and that each of these in an ascending order were dominant to the one
below, then the children belonging to the F.2 generation, that is, the
grandchildren of the original cross, would only show one white-skinned
individual in every 256 ; whilst if the number of intervening factors
instead of being four were six, then a purely white individual would be
expected to occur once only out of 4096 grandchildren. One feature
examined by Hurst(7) gave consistent and valuable results, viz. eye
colour. He clearly showed that eye colour might be of two kinds, that
in which both surfaces of the iris were pigmented, the other in which
only one surface was. The latter condition is recessive and gives
rise to the true blue eye of the anthropologist. Where the dominant
character is present the eyes are of various shades of brown or of
green. The independence of this character in segregation is of the
greatest importance when one considers how some anthropologists have
talked of a blue-eyed, fair-haired, long-headed race as if it were an
impregnable complex and not a synthetic accident. The further such
researches as Hurst's can be carried into the heredity of individual
features, the clearer will become our notions of racial types. The facts
I am about to describe in relation to the Jews will, I think, bring this
point out in a clearer light.

The racial position of the Jew has engaged the attention of all
modern ethnologists. The problem is extremely difficult because, on
the one hand, we have the oft asserted and by no means easily disproved
statement of the Jews themselves that they are pure Semites, whilst
observers such as Ripley, von Luschan and others, point out that the
Jew of to-day has no uniform cranial characters, that on the whole he
is decidedly brachycephalic, whilst the typical Semite such as the
Bedawyn is essentially dolichocephalic.

Judt(lO), whilst regarding the Jews as belonging to one type although
with variations, is assured that they cannot be designated Semites on
account of the prevalent brachycephaly.

Renan concluded that religion was the one bond of the Jews, and
that there was no single but several Jewish types. Most authorities
are at least agreed that the Jews up to the time of the destruction
had freely intermarried with the surrounding people, and the Biblical

R. N. Salaman 275

evidence supports this view. The words of Ezekiel — " Thy father was
an Amorite and thy mother a Hittite " — were not said merely in scorn.
There seems no reason to doubt that the original band of Abrahamites,
themselves of the same Semitic stock as the Assyrians, had mixed
during all the Biblical period with at least three different racial groups,
with the native Semitic Canaanitish tribes similar to themselves, with
the Hittites, and with the Amorites. The Hittites, of whom several
conventionalised representations in Assyrian and Egyptian sculpture
exist, are now considered to be practically identical with the present-
day Armenians, the highly brachycephalic people who possess the so-
called Jewish nose. Of the Amorites very little is known, but it is
generally stated that they were a long-headed blonde race. It is quite
possible that they were blonde and it is not at all improbable that the
Amorites, like the Philistines, were non-Semitic and related to the
Central European people. It is to the Amorites that the constant
occurrence of blondeness amongst Jews is by most authors ascribed and
the Pan-Germanic school go so far as to identify the Amorites with the
Nordic race. This latter theory, fanciful at the best, is, as I hope to
show, entirely repudiated by the observations I shall soon detail.

After the destruction of Jerusalem the Jews gradually spread
throughout Europe and the north coast of Africa. In Egypt, Jewish
colonies had existed for hundreds of years prior to this, and small
outlying groups were doubtless settled elsewhere, but the penetration
of Europe by Jews in any quantity began from the second century. It
is not necessary here to follow in any detail the paths of their
migrations. It should however be noted that from a very early date
the division amongst the Jews into those of the African and Mediter-
ranean Littoral, including the Iberian Peninsula on the one hand, and
those of Central Europe on the other, was established. The former
group are known as Sephardim, the latter as Ashkenazim.

The general type of face amongst the Sephardim is somewhat
different to that commonly met amongst the Ashkenazim. The colouring
is more uniformly dark, the nose less frequently characteristic. They
resemble more closely the Southern European peoples. Notwithstanding
this, the great majority may be always recognised as Jews by their
appearance, whilst one frequently meets amongst purely Sephardic
families individuals who are in no way different from their Ashkenazic
brethren.

The Sephardim are often described as being the aristocrats of the
race and of a finer and more delicate type and purer blood. Whether

Joum. of Gen. i 19

276 Heredity arid the Jew

the Sephardic community represents the aristocracy or not, depends
upon what one means by the terra. If by aristocracy is meant a
dominating class of the same stock, or a conquering invading people,
then the Sephardim hold, in respect to the Ashkenazim, no position as
aristocrats. If, by aristocrats, a class is meant which has, so to speak,
precipitated itself from out of the body of the general people by reason
of superior mental or physical attainment, then again the Sephardim
fail to establish their claim as aristocrats because, since the dispersion,
the two sects have never lived in that close communion in which such
precipitation could occur. On the contrary, the two classes have held
themselves rigidly apart up to the last fifty years or so. The original
distinction between the two groups would seem to have been essentially
geographical. During the Middle Ages the Sephardic Jews lived under
far better conditions than their Ashkenazic brethren in Europe, and in
that way they were brought into much more intimate contact with
general culture than the Ashkenazim who were thrown on their own
resources. In this sense, therefore, the Sephardim may be considered
aristocrats.

In point of view of the purity, that is to say the absence of mixture
with outside blood, during the last 1800 years, there is no doubt that
the Ashkenazim can show a far cleaner bill than the Sephardim who
are known to have absorbed in no small quantity both Moorish and
Iberian blood, so that the boast of blue-bloodedness comes to have
a meaning other than that generally assumed.

The composite nature of the Jew as he left Palestine has already
been stated and the question at once arises, was this complexity
increased by intermarriage with European races during his wanderings ?
Many authorities, and recently more especially Fischberg(6), have
argued that the Jew has absorbed, during the last two thousand years,
blood from all the European stocks. Ripley is assured of it. Whilst it
is obviously impossible to prove that there has been no intermixture
during the last eighteen hundred years, yet it is, I think, more than
probable that that intermixture has been absolutely minimal. The
historic evidence is naturally incomplete on either side. Those who
think the intermixture was important in quantity point out the well-
known fact of the conversion to Judaism of considerable numbers in
Rome, but they forget that it was these very Judaised Romans who
were the early Christians. Then one is reminded that in the eighth
century the kingdom of the Kozars in South Russia was converted to
Judaism. This is true, but as Zollschan points out, all we know is.

R. N. Salaman 277

that the King and his immediate court were converted ; and according
to Joseph Jacobs after the destruction of the Kozar Empire it was
the Jews of that district who formed the Karaite sect, and this sect
has remained absolutely distinct from the rest of the European Jews,
A further wholesale conversion is that of the Falashas, an Abyssinian
negroid people, of whom we shall have a little to say later. They do
not, however, in any way aflfect the question of the purity of the
present-day European Jews as there is no communion whatever
between them.

When one considers the melancholy condition of the Jews in
Central Europe throughout the entire Middle Ages, how they were
despised and despoiled in every land, is it likely that any Gentile, much
less any number, would willingly seek admission into their flock,
especially when one remembers that the entry of the male Gentile
necessitates the Abrahamic covenant I It might be thought that with
the Renaissance and the spread of culture, the opportunity had arisen
for a greater intimacy between the Jews and their Gentile brethren,
but so far was this from being the case that it was now that the greatest
paradox in history took place. To the Gentile, the period of the
Renaissance brought culture and freedom of thought, to the Jew it
brought the Ghetto and the bondage of the Rabbi. The Ghetto walls
acted as an impenetrable barrier between Jew and Gentile up to the time
of Napoleon, who was the first in Western Europe to break them down.
In Galicia and Russia, where still the majority of Jews live^ the Ghetto
life — none the less real though the walls are gone — still exists. During
the last two or three generations intermarriage has taken place and
become increasingly common in Western Europe, but it has little
bearing on our problem. The ofifepring of the intermarried in the
great majority of cases, passes over to the Gentile population, whilst
those that retain their connection with the Jewish community are
cognisant of their origin. It would be possible to follow this question
in far greater detail but I do not think that it would serve any useful
end. All the historic evidence would seem to bear out the contention
that from the second century till at least the beginning of the nineteenth,
the Jewish people (Ashkenazim) in Europe absorbed into their own
midst practically no blood from the races with whom they came in
contact At the same time it is known that a leakage, varying in
degree, of Jewish blood to the outside was always taking place, and
this loss occurred then as now, at such points on the periphery where the
community came into the most intimate contact with the outside world.

19—2

278 Heredity and the Jew

Ethnologists may be said to agree that the Jew is not racially pure,
but on the other hand they have to admit that the Jews constitute
a definite people in something more than a political sense, and that they
possess though not a uniform, still a distinguishing type.

Nothing is more confusing than the varied accounts of the shapes
of head, nose, eyes, and colour of the hair of Jews in different countries,
and if one's only acquaintance with Jews were through the literature
of anthropology one would be inclined to think that the " chosen
people " had no existence apart from books, and the imagination of the
anti-Semites. It is with no small degree of comfort therefore, that one
finds Ripley(12)^ making the following statement. "Who has not, on
the other hand, acquired a distinct concept of a Jewish face and of
a distinctly Jewish type ? Could such a patent fact escape observation
for a minute ? " Again Weissenberg(14) says " The Jew in an anthropo-
logical sense forms no specific type, but the facial expression is
absolutely characteristic." Fischberg is not so whole hearted as to the
general occurrence of this characteristic facial expression but he does
recognise it and considers it not strictly a physical trait but rather
an expression of the soul. Others will tell us that this Jewish expression,
so impossible to define, is merely an emblem of the ceaseless wanderings
and the countless agonies of the Jew — of the tausend-jdhrigen Schmerz, as
Heine calls it. Others again tell us it exists because the Jew is landless,
and if only he were once more back in his native land the facial type
would vanish. All, however, practically agree that whether blonde or
dark, tall or short, long headed or round headed, the Jew is a Jew
because he looks like one. The peculiar facial expression is at least
not the outcome of recent times. We have evidence of the greatest
antiquity. In the Assyrian sculptures, 800 B.C., are depicted Jewish
prisoners who are thoroughly Jewish (PI. XXXVI. and PI. XXXVII. fig. 1)
and Petrie(ll) has brought home from Memphis terra-cotta heads dating
500 B.C. of Jews at once recognisable by their Jewishness. On a forest
roll of the pre-expulsion times in England, is a pen and ink sketch, or
one might rather say a caricature of a certain Aaron, "Son of the Devil,"
dated 1277 which, crude though it is, hits off a distinctly Jewish type
(PI. XXXVII. fig. 2). The great master Rembrandt has given us numerous
drawings of Jews. He was mamly attracted by the Sephardic Jews,
but whatever the shape of their face may be, the curious expression
that we recognise as Jewish, never escaped the artist. More interesting
than the examples given of the persistence of this facial expression

1 Loc. cit. p. 399.

R. N. Salaman 279

is the fact that the Samaritans of to-day who live in the land of their
fore-fathers, have an unmistakable Jewish expression, and this though
their heads are dolichocephalic and those of the majority of Jews
brachycephalic.

At this point one might with advantage consider the relation which
the existence of the Kohanim has to the question of Jewish type.
The Kohanim are the traditional descendants of the tribe of Aaron.
There is, of course, no written record of such descent, but the hall-mark,
as a rule, is shown by the name of Cohen or some modification of it.
It is not at all unusual, however, to find people not possessed of the
name of Cohen, who are still Kohanim. It is most improbable that
anyone could, and much less would, assume the title of Kohen without
having a right by birth because it conveys neither social distinction nor
advantage, whilst on the other hand, it brings in its train some un-
doubted disabilities, the chief of which directly concerns us and is, that
no Kohen, according to Jewish law, can marry a stranger, a proselyte
or the daughter of a proselyte, or a divorcee : so that we have
a sect whose descent may be regarded as strictly Jewish. If now we
review the physiognomies of the various Kohanim, it will be found that
they exhibit no type in any way distinct from that of other Jews. Every
phase of Jewish bodily form will find its representative amongst the
Kohanim, so that one is inclined very much to the view that whatever
value may be ascribed, and I personally think a very high one may be,
to the purity of descent of the Kohanim during the last 2000 years,
practically the same value may be ascribed to their brethren amongst
whom they live.

What the elements are which go to make up the expression of
a face that is at once so elusive of description and yet so characteristic,
it is difficult to say. The nose is often peculiar, not because of its
length or even its convexity which may be often outdone in non-Jews,
but by the heavy development of the nostrils. Jacobs has described
this " nostrility " and has most aptly compared the Jewish nose to the
figure six with a long tail. Remove the tail, he says, and the Jewishness
will disappear. The eyes are generally elongated, and a fairly character-
istic feature is the length of the upper eyelid. The face which
exhibits the expression of Jewishness is never of the angular type with
square jaw, a type which is indeed extremely rare amongst Jews. Far
more usual is it to find rounded features, long sloping jaw, fairly
developed chin which is round and not square, a good-sized forehead
devoid of that angularity in the region of the temples which is not

280 Heredity and the Jew

uncommon amongst Teutonic people. However it may be brought
about, there is no doubt that the character of Jewishness is a real one.
Weissenberg(15) relates that he put several hundred photographs of
Russians and Russian-Jews without peculiar dress or other distinguishing
feature before two scientific friends, one a Jew, the other a native
Russian. His Jewish friend picked out 70 °/^ of the Jewish subjects
correctly and the Russian 50 %. If so high a percentage of Jews
could be identified by their looks alone in a photograph it is not
surprising that the opinion is current that the Jew may be recognised
wherever he goes. Notwithstanding the fact that the great majority
of Jews look Jewish, it cannot be denied that one meets, not rarely,
individuals, perhaps more often men than women, who do not exhibit
this type and who are either indistinguishable or at least practically
indistinguishable from North Europeans. It is relying on these
apparently non-Jewish faces, that Fischberg and others have rashly
assumed that they are the direct results of mixture with the surrounding
people. I think I shall be able to offer some evidence which will show
that this view is untenable.

Impressed with the great frequency and the distinctiveness of the
Jewish type of face, it occurred to me that this character might form
excellent material for research on Mendelian lines. Intermarriage
to-day with the English is very common in Anglo-Jewry, and one had
only to follow out such cases of mixed marriage to obtain results
comparable to those the genetic student has been obtaining in plants
and animals. My method has been to collect personally, as far as
possible, all cases of mixed marriage and to obtain the assistance of
those on whom I could rely, and whose duty it was merely to state
whether they considered the children of the mixed marriages of their
acquaintance as Jewish or Gentile in appeai-ance. Most of my
observers were quite ignorant of the purpose of my examination and
of the results I expected, whilst none were conversant with Mendelian
or other theories of heredity. All who have assisted me have been
themselves Jews and I have noted a distinct tendency on their part to
claim, wherever possible, a Jewish type of face for the children they
have examined, and although, as I shall show, the results are entirely in
the opposite direction, yet what error there is, is distinctly towards
increasing the number of supposed Jewish faces in the offspring of
mixed marriage. Wherever possible, I have seen the children myself
or have obtained photographs, but in at least half of them, I have had
to rely on others. In doing so I have been rather encouraged than

R. N. Salaman 281

otherwise by finding that the bias of my assistants has been always

against the results which they, to their own surprise, have found. In

all cases the Jew is of the Ashkenazic section and the Gentile is either

a native of England or Northern Europe.

Briefly, the results of the intermarriage of Jew and Gentile may be

stated thus (Table I).

TABLE I.

Father

Gentile

Jew

First Genera

Mother
Jewess
Gentile

tion.

Children

Number of Families
50
86

GentUe

240

Jew
15 "
11

Intermediate
4
4

Total 136

328

In 50 families where the father was Gentile and the mother a Jewess,
there were 88 Gentile-looking children, 15 Jewish, and 4 intermediate in
type. In 86 families where the father was Jewish and the mother Gentile,
there were 240 Gentile-looking children, 11 Jewish, and 4 intermediate.
In both cases the intermediates are practically Gentile-lookiug. Adding
the two classes together we find that there are 336 Gentile children to
26 Jewish, i.e. 13 Gentile to 1 Jewish. The result is a surprise to both
the anthropologist and to the Mendelian. To the former who looks for
blending, we have the fact that so far from blending, we have no less
than 93 °j^ of the mixed bred offspring resembling one parent only.
To the Mendelian some surprise must occur, that the dominance is not
absolute, but this is, to a slight extent, due to the Jewish bias in the
observations, and to a much greater extent, to a Jewish permeation of
the English people in certain localised districts which is much more
prevalent than is generally suspected. I have, whilst making these
observations, come across certain cases where I was assured that in a
certain family the father was a Jew, the mother a Gentile. In one such
I examined the children carefully and found that two were without doubt
Gentile in appearance whilst one was equally without doubt Jewish.
I then discussed the family history with the parents and I was able to
obtain the pedigree shown in fig. 1 which at once explains the occurrence
of the Jewish child. In another case I found a very similar state of
affairs but I was unable to trace it further as the non-Jewish parent
objected to elucidate the Jewish blood in her grandparent which she,
however, admitted. In a third and fourth case where complete
dominance was expected but not obtained, I have reason to believe
that it will be discovered that the Gentile parent has Jewish ancestors.

282

Heredity and the Jew

In determining the nature of so complex a character as the facial
expression, the personal equation of the observer must play an important
part. I have in some cases found that observers not specially acquainted

Family D.

-P

)xCf

9 f ?

Fig. 1.

0 = Jewish appearance.

0 = Gentile appearance.

^ = Gentile appearance and birth.

with the subject, although agreeing that a given individual of the first
generation is of Gentile appearance have yet felt that there was
somewhere lurking in the face an expression which suggested " Jewish-
ness " and there is very little doubt that such opinion may often be
well founded. I have myself come across a few cases where without
doubt the recessive Jewish facial expression has come to the surface as
the individual grew older. One case was particularly apparent. The
parents were characteristically Jewish and non-Jewish respectively,
there was a large family of which I saw one personally and the
remainder in photographs. Most of them were, to my mind, not Jewish
at all, but the one whom I was interviewing, though not in any way
strikingly Jewish, would probably have been recognised by many people
as such. His age was about 45 and he assured me, and his assurance
was confirmed by his wife, that when he was a young man he was never
by any chance recognised as a Jew in public. This same individual
has married a Gentile and has three children who are, I think, without
doubt totally non-Jewish in appearance. It is not without surprise
that one finds that very many of the leading families of this country as
given in Burke, contain Jewish blood and I know of at least one case
where two parents, neither Jewish in appearance, have a daughter who

R N. Salaman 283

is typically Jewish. A reference to Burke showed that in the family
tree of both parents was Jewish blood.

To obtain portraits of families for the purpose of exhibition has
been a most diflScult matter, but I am able to show in Plates XXXVIII.
and XXXIX. a few examples.

To the student of heredity, the phenomenon of dominance is, after
all, a matter of secondary importance. The vital question that he has
to deal with is, whether the character in question is one which
segregates or not, i.e. when in an individual the character and its
opposite are both present, are these two opposite characters represented
together in the sex cells or gametes, or does one go to one gamete and
the other to another ? Two methods are open to us in testing this
question, one to observe the matings of the hybrid individual with
those possessing recessive character only, the other to observe the
matings of such hybrid individuals with each other. Of the matings
of hybrid with hybrid I have not found a single example. This is
hardly surprising when one considers the vastly greater choice the
hybrid has of finding his mate either in the Jewish community or
in the outside world. Of matings between hybrid and Jew I have
9 families where the Jew is the father and the hybrid the mother,
giving rise to 25 children, 13 of whom are undoubtedly Gentile and 12
are unequivocally Jewish. 4 families where the father is hybrid and
the mother Jewish, contain 7 children of which 2 are Gentile and
5 are Jewish. Taking the families together their offspring consist of
1.5 Gentile and 17 Jewish children, the Mendelian expectation being
equality. Besides these matings, I have been able to collect a certain

TABLE II.
Hybrid and Jew.

Children

Nomber of Families Father Mother ,, *- >,

Jew Hybrid Gentile Jew

9 „ „ 13 12

4 Hybrid Jew 2 5

Total 13 — — 16 17

number of families where a hybrid has married a Gentile. In 4 the
father is hybrid, the mother Gentile, with 8 offspring all Gentile in
appearance. In one the mother is hybrid and father Gentile with 3
Gentile offspring (cf. Table III). I have indirect knowledge of several
other families comprising a large number of children all of whom are

19—5

284

Heredity and the Jew

said to be Gentile in appearance, but I have not included them as the
observations were not sufficiently reliable.

TABLE III.
Second Generation.

Children

Number of Families

Father
Gentile

Hybrid

Mother
Hybrid

Gentile

Gentile
3

Jew

Total 5

In figs. 2, S, 4, and 5 are given further pedigrees showing the
results of the matings of hybrid individuals with Jews and Gentiles
respectively.

Family A.

pxCf cT-x-^

cf-

Fig. 2.

Family B.
pxCT ^x«r

CT ^

Fig. 3.

©©

R. N. Salaman
Family C.

285

— #

0x ®

"1
© e

Fig. 4.

Family E.

9 ^ ^

PI XXXVII fig4* IP! XXXVII fig 3

CT
figl

PI XXXIX

fig 2
PI XXXIX

fig2
PI XXXVIII

figl

PI XXXVIII

fig 3

rig4

PI XXXIX

? 9 9 .

fig 4 fig 5 fig 6 fig3

V ^ '

PI XXXVIII

Fig. 5.

0 = Jewish appearance.
0 = Gentile appearance.
^ = Gentile appearance and birth.

The conclusion to which these results inevitably lead is that
the Jewish facial type, whether it be considered to rest on a gross
anatomical basis, or whether it be regarded as the reflection in the
facial musculature of a peculiar psychical state, is a character which
is subject to the Mendel ian law of Heredity.

With the knowledge gained from these observations one can now
understand the somewhat conflicting reports that travellers and others
have given of those outlying Jewish communities which are found
on the Malabar coast where they are known as the Beni-Israel ; in
China where they are known to the Chinese as the " people who remove

286 Heredity and the Jew

the sinew of the leg " ; in Abyssinia where they are known as the
Falashas, and in Jamaica and the West Indies. The Beni-Israel of
India have been settled in India at any rate since 1400 of the present
era, but traditionally from pre-exilic times. They are essentially
a black people quite unlike the European Jew. They have always
been looked down on by their white brethren in India and they have
lived as the natives amongst whom they dwell, and with whom there is
little doubt they have freely mixed. In the description of them given
by Fischberg, he agrees that they are non-Jewish looking and dark
skinned ; he remarks, however, that every now and again a practically
ordinary white skinned individual with Jewish features occurs amongst
them. If, as is probable, the Jewish facial features are recessive to the
native, then it is only what one should expect to find that the great
majority of this isolated community are native-looking and that an
occasional recessive should crop out from the mating of two hybrids.

The Chinese Jews are an even more isolated group who probably
reached China through India, possibly a thousand years ago, but
traditionally at a far earlier date. I have only seen photographs of
these so-called Jews, in which they are typically Chinese in appearance,
even as regards the eye shaped The Chinese Jews have lost practically
all knowledge of the tenets of Judaism and there is but little doubt
that the Jewish facial type has been swamped by the Chinese.

The Falashas of Abyssinia are simply negroid. Some doubt whether
they had at any time any Jewish blood or whether they were not merely
converts by Jewish missionaries. Faitlovitch, who has spent many
years amongst the Falashas, whilst admitting that they are not Jewish
in appearance but on the other hand closely resemble the neighbouring
black peoples, assures me that it is his belief that originally and at a
very early period a considerable body of Palestinian Jews did settle in
Abyssinia. The Jewish settlers freely intermarried with the natives at
first, but during the last two hundred years they have become isolated.
If the origin of the Falashas is such, then the swamping of the Jewish
type of facial expression is only what would be expected when a
recessive character is introduced into a community of dominants.

In Jamaica and the West Indies Jews, from the 17th century and
onward, have played a very important part as traders and settlers.

1 Dr C. G. Seligmann informs me that in a number of crosses occurring in Australasia
and the east, whether between Chinese and White, Malay and Melanesian, or Malay and
White, the peculiar Mongolian eye with its epicanthus is always dominant. This fact is
amply borne out in the photographs of the hybrids which he has shown me.

R. N. Salaman

287

These Jewish settlers employed negro slaves, with a result that a number
of their illegitimate children have founded families bearing Biblical
names but negroid features.

It hiis already been noted that amongst the Ashkenazic Jews in
England and elsewhere, one does meet with individuals who have not

cjxcj (^ 9 §

99Gf

? P P P~~]

^' (3S & & (^(^^9

n '^

O0O0

Fig. 6.

(IS(^P

oooo

O000

0 = Exaggerated Jewish type with long nose.

0 = Jewish type — readily recognisable but not exaggerated.

9 = Gentile appearance and birth.

0 = Gentile appearance, mixed Jewish-Gentile birth.

0 = Gentile appearance, pure Jewish origin.

The original Parents were first cousins.

The Daughter (No. 6) whose features were of an extreme Jewish type married a Gentile
and their child is totally un-Jewish in appearance. The Son (No. 7) married a woman of
pure Jewish descent but with features entirely nn-Jewish. AU their children are of the
exaggeratedly Jewish type.

a peculiarly Jewish facial type and in some eases the keenest eyed
Jew would not recognise these men as his brethren. At other times it
is only the superficial observer who fails to recognise the type. I have
attempted to follow out the results of the mating of such non-Jewish-
looking Jews who may be said to have a " pseudo-Gentile " appearance
with Jews who have a pronouncedly Jewish cast of feature. The case
whose pedigree is shown in fig. 6 is an interesting example, the Jewish
features being of the most pronounced, whilst the pseudo-Gentile-
looking mate is equally pronouncedly un-Jewish. All the children are
as typically Jewish as the Jewish father. A sister of this same father,
whose features are indeed almost a caricature, married an English

288 Heredity and the Jew

Gentile husband, and she has a child who is without a trace of Jewish-
ness.

I have met with an abundance of cases which illustrate the same
phenomenon, but I have not classified them statistically nor do I show
the pedigrees, because it is rarely that one can describe individuals
without the smallest possible hesitation, as " characteristically Jewish,"
" Jewish," or " non-Jewish " in expression, as one is able to do in this
family. Nevertheless, I have not met an exception to the rule that the
pseudo-Gentile appearance is recessive to the fully Jewish, where the
Jewishness of the features are strongly pronounced. In those cases
where the Jewishness of the features is weak and more or less con-
jectural, then in raatings of such with the pseudo-Gentile type,
both Jewish and non-Jewish types may be found amongst the children.

The results, therefore, seem to show with very little doubt, that
this pseudo-Gentile face is an essentially different thing from its
Teutonic counterpart. Whereas the latter is dominant to the Jewish,
the former is as decidedly recessive. Such an apparent paradox as the
dominance of one type and the recessiveness of an apparently exactly
similar one is not unknown to the student of heredity. It has been
met with by Bateson and Punnett(l) in their research on the plumage
of fowls, and by Bateson in the colour of flowers (3). I have myself,
working on heredity in potatoes (13), come across one case where the
white potato is recessive to the purple and another where an apparently
similar white is dominant to the purple.

The facts that have been described above may, I think, throw some
light on the question of the purity or otherwise of the Jews. The
Jewish features have been shown to be recessive to the Northern
European (and I have cases indicating that they are recessive to the
native Italian), to the native Indian, to the Chinaman, and to the negro.
If then the Jew had freely intermixed with the European races as some
authors think is the case, it is obvious that, the characteristic facial
type being recessive, it would have been rapidly swamped. But the
very reverse is the case : it is the one thing which practically all
observers are agreed is common to the Jewish people. It has been
suggested by the Pan-Germanic school of Chamberlain and others, that
whatever good qualities the Jews possess are due to the admixture
in them of a fair-haired race, probably Amorites who were, according to
these writers, of Germanic origin. It is indeed more than probable
that the fair features found amongst Jews are derived from Amorites or
other people of non-Semitic blood in their early home, but it has

R. N. Salaman 289

already been shown that this non-Jewish type found amongst Jews
is recessive to the typically Jewish, whereas the German or Teutonic
type is undoubtedly dominant ; hence, if the non-Jewish type is
Amoritic, then it is quite certain that the Araorites were not Germanic.
Conversely if it is not derived from the Amorites, it is at least quite
certain that it cannot be Teutonic in origin.

In marriages between Sephardic Jews of a markedly southern
European or Spanish type and Ashkenazic Jews, the former's facial
characteristics seem always to be dominant. This fact, when one
remembers the infusion of Iberian blood in the Sephardim already
referred to, is not indeed surprising.

In a previous paragraph, it was stated that many people regarded
the Jewish expression as the result of age-long homelessness and
persecution. Whether it is meant that this expression is acquired in
the life of the individual or whether it is an example of the heredity
of an acquired character, is not decisively stated. My results would
seem to throw some light on this point. In the first instance, I have
frequently seen new-born babies with an unmistakably Jewish cast of
feature, and secondly, in those families arising from the mating of
hybrid and Jew where the children are brought up in a Jewish home
with Jewish surroundings, half the children are Jewish-looking, and
half are non-Jewish, a fact which the inheritance of an acquired
character fails to explain. Again, if the expression is the result of
landlessness and the tausend-jdhrigen Schmerz, is it not peculiar that
of two children born of the same parents and reared in the same home,
one should have it and the other not ? I think it is clear, therefore,
that this Jewish facial expression is a fundamental character, and it
is necessary to trace, if we can, its origin. All observers are agreed
that it cannot be described as Semitic. It is seen in, but is not the
peculiar property of the Armenians who certainly resemble the Jews
and who probably have in some degree a common ancestry. Is it perhaps
possible that this peculiar facial type has arisen from the fusion of
characters derived from two or more of the original races from which
the Jews sprang ?

The experiments of Bateson and others(2) with the sweet pea,
paralleled as they have been in the animal world, are not unsuggestive
in this respect. On mating together two apparently similar but really
distinct white sweet peas, they obtained the common purple pea. When
this latter was bred inter se, it gave rise to a series of purples, reds
and whites. Of each of these classes, some, when self-fertilised, bred

290 Heredity and the Jew

perfectly true, so that from the union of two apparently similar whites,
arose such distinct and dissimilar individuals as the red and the purple
pea. Could not this Jewish facial expression be due to the union of
characters in a manner similar to that which gave rise to the purple in
the pea ? In this way one would be able to explain on the one hand
the practically constant presence of the Jewish facial character, and on
the other, the wide divergence of head characters and the rest, which
is found throughout the Jewish communities of Europe.

It is necessary before leaving our subject, to enquire whether there
are no other characters common to the Jew which are as frequently
present as the facial expression, or which are in any way peculiar
to Jews. There would seem to be two instances of such peculiarity
which fulfil these conditions. The disease known as Amaurotic Family
Idiocy, the victims of which die in early childhood, is probably unknown
outside the Jewish people. Fischberg states that cases are met with
outside, but all the authorities I have been able to consult agree that
it is peculiarly Jewish. Another character which would seem to be
peculiar to the Jewish people as a whole, is the absence of alcoholism
in their midst. This is acknowledged by every authority. Indeed the
Jewish Board of Guardians finds it unnecessary to make any special
provision for alcoholic cases as distress arising from this cause does
not occur more often than once in a thousand cases, and my own
experience of over nine years at the Loudon Hospital fully bears out
the statement that drunkards are practically unknown. This absence
of the desire for drink cannot be ascribed to the result of religious
training. There seems to be a real lack of that desire for drink which
is so common amongst the North European races.

The deductions which might be drawn from these two sets of facts
can naturally have no very great weight, but they do, in conjunction
with what has gone before, strengthen the view that complex as the
origin of the Jew may be, close inbreeding for at least two thousand
years, has resulted in certain stable or homozygous combinations
of factors which react in accordance with the laws of Mendel and
which may explain the occurrence of the peculiar facial expression
recognised as Jewish.

R N. Salaman 291

DESCRIPTION OF PLATES.

PLATE XXXVI.

Jewish Prisoners bearing tribute from King Jehu to Shalmaneser II. 9th century, B.C.,
Brit. Mus.

PLATE XXXVII.

Fig. 1. Jewish Prisoners at lacbish. 9th century, B.C., Brit. Mus.

Fig. 2. Pen and ink sketch in margin of Essex Forest Roll, 1277 a.d.

Fig. 3. Jewish parent of Family E.

Fig. 4. Gentile parent of Family E. (see text, p. 285).

PLATE XXXVIH.

Fig. 1. Man of pure Jewish birth, brother to the man represented in Plate XXXIX. Fig. 1,
and husband to the woman shown in Plate XXXVIII. Fig. 2.

Fig. 2. Daughter of Jewish and Gentile parents represented in Plate XXXVIl. Figs. 3
and 4, and sister to woman shown in Plate XXXIX. Fig. 2, is non-Jewish in appear-
ance.

Figs. 3 and 4 are the non-Jewish looking children of parents shown in Plate XXXVIU.
Figs. 1 and 2.

Figs. 5 and 6 are the Jewish looking children of the same parents.

PLATE XXXIX.

Fig. 1. Man of pure Jewish birth brother to the man represented in Plate XXXVIIL

Fig. 1, and husband to woman shown in Plate XXXIX. Fig. 2.
Fig. 2. Daughter of Jewish and Gentile parents represented in Plate XXXVIl. Figs. 3

and 4. Herself non-Jewish in appearance.
Figs. 3 and 4. Non-Jewish and Jewish sons respectively of parents shown in Plate XXXIX.

Figs. 1 and 2.
Figs. 5 and 6. Two brothers thoroughly non-Jewish in appearance, the children of a

father of Jewish birth and appearance, and of a Welsh Grentile mother.

Note. I am greatly indebted to the ladies and gentlemen who have so kindly allowed
me to use their photographs to illustrate this paper. Far more striking examples could
have been shown, but permission to publish was in no case obtainable.

The description "Jewish" and "Non- Jewish" ascribed to the portraits is arrived at by
personal knowledge and by the emphatic assurances of nearest relatives. It may be noted
that photographs are not a really satisfactory means of demonstrating so peculiar a
character as that of Jewishness.

292 Heredity atid the Jew

LITERATURE.

1. Bateson and Punnett. Rep. Evol. Comm. Roy. Soc. Vol. iii. p. 18, 1906.

2. Bateson, Saunders, and Punnett. Ibid. Vol. ii. p. 84, 1905.

3. Bateson. MendeVs Prin. Hered. 1909, p. 105.

4. C. B. Davenport. A mer. Nat. Vol. xliv. No. 527, p. 641, 1910.

5. Farabbe. Papers of Peabody Mus. of Amer. Arch, and Ethn. 1905.

6. Fischberg. "The Jews. A Study of Race and Environment." Contemporary

Science Series, 191 1.

7. Hurst. Proc. Roy. Soc. Vol. lxxx. B, 1908, p. 85.

8. Huxley. Zeitschrift f. Demographic u. Statistik d. Jxid. Rasse, Heft ix. 1906.

9. Joseph Jacobs. "Appendix on Racial Characters of Modern Jews." Anthro-

pological Institute, Feb. 1885.

10. JuDT. Zeitschrift f. Demog. u. Stat. d. Juden, 1905, Ht. 5. Die Juden als

Rasse, 1901.

11. Petrie. "Palace of Apries Memphis," Vol. ii. Plate XXVIII. Brit. Sch.

Arch. Egypt, 1909.

12. Ripley. Races of Europe, 1900.

13. Salaman. Journ. Genetics, Vol. i. p. 41, 1910.

14. Weissenberg. "Die Siid. Russischen Juden." Arch. f. Anthrop. Vol. xxiii.

15. Globus, Vol. xcvii. 9. 6. 10, 1910.

16. ZoLLSCHAN. Das Rassenproblem. Wien, 1911.

JOURNAL OF GENETICS, VOL I. NO. 3

PUTE XXXVI

JOURNAL OF GENETICS, VOL I. NO. 3

PUTE XXXVII

-.-r^@^^

JOURNAL OF GENETICS, VOL I. NO. 3

PLATE XXXVIII

JOURNAL OF GENETICS, VOL I. NO. 3

PUTE XXXIX

Volume I NOVEMBER, 1911 Na 4

ON GAMETIC SERIES INVOLVING REDUPLICATION
OF CERTAIN TERMS\

By W. BATESON, M.A., F.R.S.
AKD R. C. PUNNETT, M.A.

In a paper recently published' we gave a brief account of some
peculiar phenomena relating to the coupling and repulsion of factors in
the garaetogenesis of the sweet pea and of several other plants. The
view there stated was that if A and B represent two factors between
which coupling or repulsion can exist then the nature of the F^
generation depends upon whether A and B were carried into the
F^ heterozygote by the same gamete or by different gametes. If the
heterozygote AaBb is formed by the gametes AB and ab partial
coupling between A and B occurs in ^2 according to a definite system,
and it must be supposed that the gametes formed by the heterozygote
belong to one or other of the series

SAB :Ab:aB: Sab,
7AB :Ab:aB: lab,
loAB : Ab : aB : 15ab, &c.

If on the other hand the heterozygote, AaBb, is formed by the gametes
Ab and aB repulsion occurs between A and B, so that only the two
classes of gametes Ab and aB are formed. In the account to which
we have alluded we supposed that such repulsion was complete, and
that the two classes of gamete AB and ab were not formed. Our work
on sweet peas during the present summer has led us to modify oar
conception of the nature of the gametes produced in cases where
repulsion occurs, and this modification will perhaps be made clearer if
we begin by giving an account of the experiments upon which it is
based.

1 This paper is also appearing in the 49th volnme of the Brnnn VerhantUungen which
is to be published as a Mendel FesUchrift.
» Proe. Roy. Soe. B, VoL 84, 1911, p. 1.
)9Vn. of Gen. I 20

294 Reduplication in Gametic Series

During the years 1906 and 1907 we were engaged upon an investi-
gation of the inheritance of the hooded character in the sweet pea, of
which an account appeared in Report IV to the Evolution Committee
of the Royal Society, 1908, pp. 7 — 15. Among several thousand plants
bred and recorded in this set of experiments there occurred a single
individual (in Exp. 35, R.E.C. IV, p. 15) exhibiting striking peculiarities
in the form of its flowers. These were small and much deformed
(cf. PI. XL, fig. 1). The standard failed to become elevated, the keel
was cleft distally so that the anthers were partially protruded, while the
stigma projected far beyond the petals, and was carried on in the line
of the carpels instead of being abruptly bent at right angles to them as
in the normal flower. At the time of its discovery, in reference to the
open " mouth," and the protruding " tongue " represented by the pro-
jecting style, the plant was dubbed " the cretin," by which term we shall
subsequently refer to this peculiar malformation. The fact that the
style protrudes is due to the malformation of the keel which is unable
to curve the growing style and cause it to assume its natural position.
Fuller experience of these cretins has shewn us that the petals may
sometimes be nearly as large as in normal flowers (cf. PI. XL, fig. 2),
and that the standard may sometimes become elevated in the normal
way (cf PI. XL, fig. 3). The size of the flowers may vary considerably
on the same plant, and hitherto where the larger form of flower has
occurred the plant has also borne others more nearly resembling the
original type. The degree to which the keel is cleft also shews some
variation, but in all cases these cretins have the peculiar and character-
istic straight stigma.

Our original cretin was found in 1907 and was used as the pollen
parent to fertilise various sterile^ sweet peas. The F^ plants, which
flowered in 1908, were all indistinguishable from normal sweet peas.
The normal form of flower (N) was completely dominant to the cretin {n),
and fertility {F) of the anthers was of course dominant to sterility (/),
We may draw attention to the fact that the crosses were in all cases of
the nature Nfx nF, one of the two factors entering with each gamete.
In the following year a single F^ family was raised and consisted of 51
normal fertile, 30 normal sterile, 33 cretin fertile, and 1 cretin sterile ^
The cretin character behaved as recessive to the normal flower, but the

* In this family and in one of those grown later both light and dark axilled plants
occurred. In each case the dark axil went in from the fertile cretin parent, and in Fg
there is some coupling between the dark axil and fertility. The numerical results however
are complex and must be left over for discussion until more material is available.

W. Bateson and R. O. Punnett

295

relative distribution of the different characters evidently pointed to
some form of repulsion between the normal flower and fertility. Had
it not been for the appearance of the single sterile cretin we could have
regarded this case as one of complete repulsion between the factors N
and F. The problem was to account for the sterile cretin, and at the
time we were inclined to regard it as due to an unaccountable failure of
repulsion between N and F. Lack of opportunity prevented us from
following up this case in 1910, but in the present year we sowed the
seed 'of the rest of the F^ plants harvested in 1908 and obtained details
of eight more families which are set out in the accompanying table
(Table I).

TABLE L

fieferenoe
Number

Normal
fertile

Normal
sterile

Cretin
fertile

Cretin
sterile

Number 5, 1909

72, 1911

73, „

74, „

—

75, „

, 76, „

, 77, „

, 78, „

—

79, „

Total

336

150

143

Expectation

330

150

These records shew that the appearance of a small proportion of
sterile cretins is a constant feature in these families and we suggest
that their presence may be accounted for as follows. The repulsion
between N and ^ is to be regarded as partial, and of such a nature
that the series of gametes produced by the ^j plant is NF, SNf, ZnF, nf.
Such a series of ovules fertilised by a similar series of pollen grains
would give rise to a generation consisting of 33 normal fertiles, 15
normal steriles, 15 cretin fertiles, and 1 cretin sterile. As the figures
given in Table I shew, this expectation is closely realised by the facts
of experiment, and we have little hesitation in regarding this explana-
tion as the correct one. Moreover we ar6 inclined to go further and to
extend the principle to all cases of repulsion in plants. We consider
then that where A and B are two factors between which repulsion
occurs in the gametogenesis of the heterozygote formed by union of

20—2

296

Reduplication in Gametic Series

the gametes Ah and aB, the gametes produced by the heterozygote so
derived form one or other term of the series

AB: ^Ab
AB'. 7Ab
AB : 15Ab

SaB : ab,
7aB : ab,
15aB : ab, &c.

And if we take 2n as the number of gametes in the series we may
generalise it under the expression AB : (n—l) Ab : {n — l)aB : ab.
As the repulsion increases in intensity it is obvious that the zygotes
of the form A ABB and aabb will become relatively scarcer, for there
will be only one of each of these two homozygous forms in the complete
series of zygotes. At the same time the ratio of the three zygotic
forms AB : Ab : aB approaches more and more nearly to the ratio
2:1:1 such as would occur if the repulsion were complete. This is
brought out in the upper part of Table II where we have set out some
of the gametic series in which partial repulsion is involved together
with the series of resulting zygotes. The latter, as the Table shews,
are covered by the general formula

(2n^+l)AB : {ri'-l)Ab : {n^--i)aB : ab*.

Partial repulsion '

from zygote

of form

AbxaB

Partial coupling
from zygote .
of form
ABxab

TABLE II.

Gametic

series

Number
of gametes
in series

Number of
zygotes
formed

Nature of zygotic series

AB^

(n-1)

2re

4na

2n2 + l

n«-l

n2-l

4096

2049

1023

1024

513

255

256

129

258

177

1024

737

225

4096

3009

961

128

16384

12161

127

3969

(n-1)

1 (

n-1

) 2n

4n2 3n2

-(2re-

2n-l

2n-l n

2-(2re

Hitherto the only repulsion series which we have been able to identify
with certainty is the one with which we have just dealt, i.e. 1:3:3:1
series for the factors N and F.

* The general formulae made use of here and in Table II are purely empirical, and
offer a convenient way of calculating the nature of the zygotic series from any series
of gametes.

W. Bateson and R. C. Punnett 297

It is probable, however, that the case of blue and long pollen' is one
in which the repulsion is of the 1 : 7 order. Up to the present time
we have had four families of the mating Bl x bL and the 419 plants
recorded in F^ were distributed in the four possible zygotic classes as
follows :

Beference Number

Blue long

Blue ronnd

Red long

Red round

Number 61, 1910

J^28, „

—

„ Fil, ,.

—

„ F32, „

—

Total 226 95 97 1

Though the evidence for partial repulsion rests here upon the single
red round plant which occurred in family 61, it is in reality very much
stronger than it appears at first sight, for the following reason. All the
plants in the above four families were hooded, i.e. lacking in the factor
for erect standard (E). As we have already pointed out^ the three
factors E, B, and L constitute a series such that if any two are brought
into a zygote by different gametes repulsion occurs between them.
Until the present round hooded red plant appeared we had never
encountered this combination in any of our experiments. It cannot
therefore be regarded as due to a stray seed from another family. And
it is evident that if the repulsion between any pair of these three
factors were complete such a plant could never arise. For in the
normal course the ehl gamete could never be formed. Only two
possibilities therefore are open. Either we must look upon it as an
unaccountable mutation, or we must consider that the repulsion between
B and L is partial. In the light of the evidence afforded by the cretin
sweet pea we prefer the latter hypothesis, and we are inclined to regard
the partial repulsion between B and Z as of the 1:7:7:1 type. On
this hypothesis we should expect one red round in every 256 plants
(cf. Table II) whereas experiment gave 1 in 419. At the same time
we recognise that the data are not yet sufficient to preclude the
1 : 15 : 15 : 1 system. It is worthy of note that the coupling between
B and L is usually on the 7:1:1:7 system, and it would be interest-
ing if in such cases as these the repulsion and coupling systems for a
given pair of factors were shewn to be of the same intensity. In most
cases this could not be tested in practice owing to the verj' large

1 Blue in the flower colour (B) is dominant to red (b), and long pollen (L) is dominant
to ronnd pollen ({)•

« Proe. Roy. Soe. 1911, p. 7.

298 Reduplication in Gametic Series

number of plants required. Thus the coupling between erect standard
and blue is on the 127 : 1 : 1 : 127 system, and if the repulsion were of
similar intensity we should expect only one hooded red in every 65,536
plants. We may, however, state that in this particular case we have
grown over 4000 plants without meeting with a hooded red, so that
the facts, so far as they go, point to a high intensity of repulsion for
factors exhibiting a high intensity of coupling. It is obvious that the
relation can only be worked out where the intensity of repulsion is low,
and it is_hoped that the case of the cretin may eventually throw light
upon this point when the system in which iV^ and F are coupled shall
have been determined.

The question now arises how these gametic systems are formed. In
each the characteristic phenomenon is that the heterozygote produces
a comparatively large number of gametes representing the parental
combinations of factors and comparatively few representing the other
combinations. In describing the original case of coupling, namely that
between the blue colour and long pollen in the sweet pea, we pointed
put that no simple system of dichotomies could bring about these
numbers, and also that it was scarcely possible that such a series could
be constituted in the process of gametogenesis of a plant, in whatever
manner the divisions took place. In saying this, regard was of course
had especially to the female side, and this deduction has become even
more clear in view of the fact that we now know a series consisting of
256 terms. It is practically certain that the ovules derived from one
flower of the sweet pea, even if all collateral cells be included, cannot
possibly be arranged in groups of this magnitude. A pod rarely contains
more than nine or ten good seeds at the most, so that if we even reckon
twelve potential seeds to the pod and eight potential gametic cells to
the ovule, the total is still only 96, which is much too few\ Nevertheless
our series of numbers is plainly a consequence of some geometrically
ordered series of divisions.

There is evidence also from other sources that segregation may
occur earlier than gametogenesis. Miss Saunders' observations on
Matthiola^ and on Petunia^ proved that in those plants the factors for
singleness are not similarly distributed to the male and female cells.

1 From the fact that in maize the endosperm characters are the same as those of the
seed itself we know moreover that segi-egation must have been completed before the
divisions at which the male and female cells which constitute the endosperm are set apart.

2 Kep. Evol. Committee R. S. IV, 1908, p. 36. . .J

3 Jour. Gen. i. 1911.

W. Bateson and R. C. Punnett 26k

The recent work of de Vries' on Oenothera biennis and muricata has
provided other instances of dissimilarity between the factors borne by
the male and female organs of the same flower. In all these examples
it is almost certain that segregation cannot take place later than the
formation of the rudiments of the carpels and of the stamens respec-
tively. The only visible alternative Is that in each sex the missing allelo-
morphs are represented by somatic parts of the sexual apparatus, which
for various reasons seems improbable. There is therefore much reason
for thinking that segregation can occur before gametogenesis begins,
but there is no indication as to which are the critical divisions.

Now that we may regard the formation of four cells of composition
AB, Ab, aB, ab, as the foundation both of the coupling- and of the
repulsion-series the problem is manifestly somewhat simplified. The
time, excluding gametogenesis, at which we can most readily imagine
four such definite quadrants to be formed is during the delimitation of
the embryonic tissues. It is then that the plant is most clearly a
single geometrical system. Moreover the excess of gametes of parental
composition characterising the coupling- and repulsion-series must
certainly mean that the position of the planes of division by which
the four quadrants are constituted is determined with regard to the
gametes taking part in fertilisation. Though the relative positions of
the constituents of the cells may perhaps be maintained throughout the
history of the tissues, it is easier to suppose that the original planes of
embryonic division are determined according to those positions than
that their influence can operate after complex somatic diflferentiation
has been brought about.

At some early stage in the embryonic development or perhaps in
later apical divisions we can suppose that the n — 1 cells of the parental
constitution are formed by successive periclinal and anticlinal divisions
of the original quadrants which occupy corresponding positions. The
accompanying diagram gives a schematic representation of the process
as we imagine it. Obviously it does not pretend to give more than
a logical or symbolic presentation of the phenomena. If such a sy.stem
of segregation is actually formed at the apex, it must be supposed that
the axes of the system revolve with th& generating spiral. Whatever
hypothesis be assumed the following points remain for consideration,

1. We are as yet unable to imagine any simple system by which
the four original quadrants can be formed by two similar divisions.
Evidently there must be two cell-divisions, and if in one of them we

» Biol. Centralbl. xxxi. 1911, p. 97.

800

Reduplication in Gametic Series

suppose AB to separate from ah, we caunot then represent the formation
of Ah and aB. Therefore we are almost compelled to suppose that
the original zygotic cell forms two similar halves, each AaBh, and that
the next division passes differently through each of these two halves,
in the one half separating AB from ah, and in the other half separating
Ah from aB. The formation of these four quadrants must take place

ABxab

Ab X aB
( Ab.aB I

n- 1

n -1

SAB

lAb

Iba

3ab

3Ab

IBa

lAB

Fig. 4.

3aB

in every case in which there is segregation in respect of two pairs
of factors, (For three pairs there must similarly be eight segments,
and so on.) The axes of this system may well be determined by the
position of the constituent parental gametes. Reduplication or pro-
liferation resulting in w - 1 gametes may then take place in either of
the opposite pairs of quadrants according to the parental composition.

W. Bateson and R C. Punnbtt 301

2. If in the gametes of any plant some factors are distributed
according to one of the reduplicated series and other factors according
to the normal Mendelian system — as we know they may be — the segrega-
tions by which such a system is brought about cannot have happened
simultaneously. Moreover if various reduplications can take place very
early iu some individuals and not in others, we cannot imagine how the
normal form of the plant remains unchanged, unless these reduplications
affect tissues originally set apart as germinal.

As possibly significant we note here the fact that in the embryonic
development of plants the order of the various divisions is known to
be subject to great variation and it is not inconceivable that such
disturbances of the order in which the planes of division occur may
indicate variations in the process of segregation \

3. We do not yet know whether independent reduplicated systems
can be formed in the same individual. In the sweet pea for instance
we have not yet seen the consequences of combining blue, erect standard,
and long pollen with the fertile-sterile, dark-light axil series, and much
may be discovered when such families come to be examined.

Animals.

The phenomena seen in animals may well be produced by the
segmentations in which the parts of the ovary or testis are determined.
Hitherto no case of coupling has been found in animals. Among the
phenomena of repulsion, however, of which many examples exist, certain
suspicious cases have been observed which may mean that in animals
reduplicated systems exist like those of the plants. Nevertheless at
present it seems not impossible that the two forms of life are really
distinguished from each other in these respects.

Terminology.

Lastly, in view of what we now know, it is obvious that the terms
" coupling " and " repulsion " are misnomers. " Coupling " was first
introduced to denote the association of special factors, while "repulsion"
was used to describe dissociation of special factors. Now that both
phenomena are seen to be caused not by any association or dissociation,
but by the development of certain cells in excess, those expressions

* See Coolter and Chamberlain, Morphology of Angiotperms, 1903, p. 187.

802 Reduplication in Gametic Series

must lapse. It is likely that terms indicative of differential multiplica-
tion or proliferation will be most appropriate. At the present stage of
the inquiry we hesitate to suggest such terms, but the various systems
may conveniently be referred to as examples of reduplication, by
whatever means the numerical composition of the gametic series may
be produced.

EXPLANATION OF PLATE XL.

Fig. 1. Photograph of the flowering stalks of two cretins. The flowers are here as fully

opened as they usually become in this variety, and they are represented slightly

smaller than natural size.
Fig. 2. Flower of cretin which has larger petals than usual. The standard however is not

elevated and the straight stigma protrudes beyond the rest of the flower.
Fig. 3. In the centre two flowers from a cretin in which, the standards are fully elevated.

On the right are two other mature flowers from the same plant shewing petals of the

usual cretin form. On the left are two old buds.

*"^

JOURNAL OF GENETICS, VOL I. NO. 4

PLATE XL

Fig. I.

Fig. 2.

Fig. 3-

FURTHER EXPERIMENTS ON THE INHERITANCE
OF "DOUBLENESS" AND OTHER CHARACTERS
IN STOCKS.

By EDITH R. SAUNDERS,
Lecturer and late Fellow of Newnham College, Ccembridge.

TABLE OF CONTENTS.

PAOK

Statement of conclusions arriTed at in the earlier experiments .... 303
Later experiments on the inheritance of " doableness " and plastid colour . . 306
I. Baces which were obtained only in the double-throwing form • . . . 306
II. Baces which occur both in the form of double-throwing and non-double-
throwing strains 311

III. Proportion of doubles obtained from the eversporting strains when self-
fertilised or inter-crossed 317

lY. Constitution of the zygote and segregation in the eversporting forms . 321

V. Segregation in Fi orossbreds derived from two eversporting forms and

statement of the results obtained in F2 324

YI. Constitution of the zygote and segregation in the pore-breeding (non-
double-throwing) strains 334

VJi. Segregation in Fj crossbreds derived from unions between eversporting
and non-double-throwing forms and statement of the results obtained

inFa 336

Yin. Summary 356

Appendix. Note 1. On the relative viability of seeds giving rise to singles and

doubles 361

Note 2. On the inheritance of the branched and the nnbranched

habit 368

Note 3. On certain sap-colours not dealt with in the earlier accounts,

and on the constitution of the sulphur-white race . 369

Statement of conclusions arriyed at in the earlier
experiments.

The experiments recorded in the present paper form a continuation
of those of which I have already given some account elsewhere*, and it
may be well, before considering these later records, to recall the main
conclusions given in the earlier accounts.

* Beports I — lY to the Evolution Committee of the Boyal Society. In regard to
"donbling" see II, 1905, p. 29; m, 1906, p. 44; lY, 1908, pp. 4, 36.

304

Doubleness in Stocks

Double stocks are completely sterile, forming neither pollen nor
ovules, and consequently they are always obtained from seed set by
singles.

Among the singles certain strains breed true to singleness, producing
only singles in successive generations, whether self-fertilised or inter-
bred ; these are referred to as no-d-strains. Other strains of singles,
indistinguishable to the eye from those of the previous class, yield
a mixed offspring of singles and doubles when self-fertilised or inter-
bred, the doubles being mostly (? invariably) in excess of the singles —
referred to as d-strains^.

The behaviour of these two types of singles may be graphically
contrasted thus :

no-d-single

d-single

singles

and so on indefinitely.

doubles (sterile)

singles

doubles (sterile)

singles doubles (sterile)

and so on indefinitely.

A strain composed entirely of c?-singles would thus be " ever-
sporting."

Further progress in the elucidation of this peculiar type of inheritance
was made when it was shown that the eversporting character results
from a difference in distribution of the factors concerned, among the
ovules and the pollen grains respectively. In a single belonging to an
eversporting strain the pollen grains all appear to behave alike and all
carry doubleness, whereas the ovules are evidently heterogeneous, rather
more than half carrying the double, and the remainder the single
character. These conclusions were arrived at through the different
results obtained in reciprocal unions between pure-breeding and ever-
sporting individuals. For while no-d-single $ x c^-single (^ gives Fi
plants all throwing doubles on self-fertilisation, the reciprocal cross
rf-single $ X no-d-single j/* gives Fi individuals of two kinds, viz.
those which, when self-fertilised, throw doubles, and those which
breed true to singleness. The composition of the resulting generations
in the two cases is compared below.

'■ Crosabreds are not here in question.

E. R. Saunders 305

fuhd-wag]e d-single no-d-mngle

pollen X ovules pollen x ovoles

Fi singlefl singles singles

• I II I . » . u,

Ff singles singles singles doubles singles singles doubles

F3 singles singles singles singles doubles singles singles singles doubles

As there is no reason to suppose that the ovules and pollen of the
no-d-singles are unlike in constitution, this difference in behaviour of
the ^1 crossbreds resulting from reciprocal unions must be due to
a difference in the composition of the ovules and pollen produced by
the dsingles; and the experimental data are in accordance with the
explanation already given, viz. that the d-singles produce two kinds of
ovules, but only one kind of pollen grain\

Moreover this interpretation is confirmed by the fact that doubles
are always produced in Fi from a cross between two rf-singles, whereas
doubles are never obtained in ^i when the mating is between a d- and
a wo-d-single. Doubleness in this respect behaves as a recessive.

So far the case is clear, and the explanation just given has been
amply borne out by subsequent experiments. But certain points in the
relations existing between singles and doubles still remained obscure.
Though it was now clearly established that the appearance of doubles
in Stocks is exhibited in an orderly and definite manner, and is entirely
independent of external conditions, it still remained to determine the
proportion of doubles thrown by the eversporting singles, and to
ascertain, if possible, whether this proportion is constant. Doubleness
behaves as a recessive to singleness; how then are we to account for
the production of doubles in excess ? Nor is doubleness the only
character which behaves in this remarkable way. In a certain race of
double-throwing singles, viz. sulphur-white, the plants are also ever-
sporting in regard to plastid colour ; every individual yields both whites

1 The conception of a difference in constitution between the ovules and pollen grains
of a plant was first put forward in 1908 in connection with the Stocks. It is interesting
to find that a difference in reciprocal crosses among certain forms of Oenothera has led
de Yries to the conclusion that differences between the ovule and pollen series of the
same plant may altso occur in this genus. (Cf. de Tries, BioU Centr. 1911.)

80^ Douhleness in Stocks

and creams. From independent experiments^ we know that white
plastid colour is dominant to cream, nevertheless among the offspring
of the sulphur-whites the dominant whites are not more numerous than
the recessive creams. Moreover the inheritance of plastid colour is
curiously bound up with the inheritance of singleness and douhleness ;
for whereas in the sulphur-white race the singles, so far as experiment
has yet gone, are all white, the doubles are for the most part cream,
though a few are white like the singles. It was with the aim of
elucidating these phenomena that the present experiments were under-
takfen, and in the following account I have attempted to show that by a
conception of coupling and repulsion^ among the factors, and a peculiar
but definite distribution of the factors among the reproductive cells de-
pending upon their sex, these hitherto unexplained facts can be related
to our previous knowledge, and brought together into a general scheme.

Later experiments on the inheritance of "doubleness"
and plastid colour. '

I. Races which were obtained only in the double-throwing forrn.

Two of the Ten-week wallflower-leaved varieties, viz. red (crimson)
and sulphur-white, appear to be obtainable only in the double-throwing
form. Direct proof of the eversporting character is obtained if doubles
are always found to occur when individuals of the race in question are
self- fertilised, while corroborative evidence is afforded by the indirect
method of crossing. For if the conclusion in regard to the character of
the pollen grains in eversporting races given above (p. 304) be correct,
it follows (1) that when an eversporting race is used as the pollen
parent in a cross with a true-breeding (no-d) race, doubles, though
absent in Fi, may be expected to occur in every family in ^2; (2) that

1 Eep. Evol. Committee, IV, 1908, p. 35.

2 The terms coupling and repulsion have been employed by Bateson and Punnett in
explanation of certain results obtained by them in the Sweet Pea, which seemed to
suggest that the inter-relation between certain factors was of the nature of attraction or
repulsion according as these factors were received separately from the parents or
associated together (see Proc. Roy. Soc. B, Vol. 84, 1911). In a later communication
which appeared after the present paper had been sent to press (see Verhandlungen des
naturforschenden Vereines in Brilnn, Bd. xlix. and also the present number of this Jownal
of Genetics), these authors suggest the substitution of the general expression "reduplica-
tion of terms " to cover both cases. Pending the acceptance of other terms which will
serve to distinguish results which would have been classed under the head of coupling
from those coming under the head of repulsion the original terms are here retained, as
convenieritly descriptive of the two types of results, not as connoting necessarily the real
cause of the phenomena. .

E. R Saunders 307

when two eversporting races are bred together, doubles will, on the
other hand, occur in each F^^ family as well as in each family in all later
generations, just as when either race is repeatedly self-fertilised.

The evidence at present available in each case may be summed up
as follows :

Red Race.

87 individuals were tested directly by self-fertilisation. The pedigree
of these plants is shown below.

Parent plant (A)

Fi Plants derived by self-fertilisation from the parent plant A

F2 „ M M >> 11 of thfi ^1 individuals

F3 „ „ ,. „ 3 „ F2

Fi „ „ „ „ 1 „ Fs „

Fs „ „ „ „ 1 „ Ft

Total 87

Doubles occurred in each of the 87 families (see Table III). Thus
every attempt to breed out the doubles proved unsuccessful, and the
evidence shows that this form, at least so far as the material used in
these experiments is concerned, is eversporting. Efforts to obtain from
other seed on the market a true-breeding (no-d) strain of this race
proved equally unsuccessful. Two or three large firms to whom
application was made were unable to supply such a strain ^

^ In the catalogues of the large Stock Growers the varions stock races are catalogued
in different colours, but not as a rule according as they do, or do not, produce doubles.
It has however been found that commercial seed, stated to give only singles, does in fact
breed true; and that from seed stated to yield doubles, doubles are obtained in such
abundance that for testing purposes small sample savings are suflBcient. It might
perhaps be supposed that, since the aim of the grower is to produce seed which will
yield as high a percentage of doubles as possible, a true-breeding strain, should it by
chance appear, would be at once discarded; and hence the fact that it had not been
found possible to obtain such a strain in the red race, might not necessarily indicate that
no true-breeding individuals occurred when the race was cultivated without selection.
But this assumption does not explain the fact, that in the case of the other sap-coloured
forms employed, true-breeding seed is on the market and easily obtainable. There is no
reason to suppose that modem taste demands a pure-breeding single in various other
shades but rejects it if coloured red. Nevertheless there is no doubt that a pure-breeding
red strain could at once be made by crossing an eversporting red with a no-d type.
If the resulting crossbreds are self-fertilised, F^ will contain a proportion of glabrous
red singles some of which will be found to breed true. We may therefore safely class the
red race with the other sap-coloured types as one which can exist both as a pure-breeding
and a double-throwing form. Whether a non-double-throwing sulphur-white race, i.e. to
say a white race composed entirely of individuals throwing a proportion of creams but
breeding true to singleness, can exist, or not, we cannot tell. At present no such race
is known, and we are unable to make it.

308 Doubleness in Stocks

Sulphur-white Race.

62 individuals were self-fertilised, but in this case they were not all

descended in one line. Their relationship is shown below.

1 Parent plant (-4)1

7 Fi Plants derived by self-fertilisation from the parent plant A

19 Fj ,, ,, „ ,, 3 of the Fi individuals
3 i^s >» 5> » >> 1 »» F2 „

1 Parent plant (J5)i

20 Fi Plants derived by self- fertilisation from the parent plant B

5 Fz „ „ ,, ,, 2 of the Fj individuals

5 Fz >> » >i >i 3 y, Fi „

1 Parent plant (C)

Total 62

Families were obtained from each of these 62 individuals and here
again doubles occurred in every case (see Table III). It is therefore
evident that this race also is wholly composed of eversporting individuals.
So much seems clear from the results of self-fertilisation, but it is only
on crossing that the real explanation of these results becomes apparent.

Reciprocal crosses between cZ-strains and ?20-c?- strains afford a con-
venient means of separately testing the ovules and the pollen of the
ci-strain, and it is through the different behaviour of such reciprocals
that we are enabled to understand the true cause of the eversporting
habit. At this point it will be convenient to consider transmission of
the double character by the pollen in these two strains ^

When the red or sulphur-white was used as the pollen parent in
a cross with a pure-breeding (no-d) strain all self-fertilised i^. plants,
with three exceptions, produced a mixture of singles and doubles in
^2 (see Table IV). In view of all the evidence it is unlikely that any
of these three cases really indicates a genuine exception ; each will be
fully discussed later (see pp. 309, 310).

The experiments with the red race were as follows : —

Pollen from 6 individuals of this race was used to fertilise 10 plants
belonging to 4 different pure-breeding strains. The number of seed-
parents in each case was as follows:

2^o-d-glabrou8 cream 4

„ „ white 4

„ ,, flesh 1

„ hoary white (Brompton) 1

Total ... 10

^ A and B were obtained from different growers.

3 Transmission by the ovules will be dealt with in a later section (see p. 323).

E. R. Saunders 309

91 of the resulting Fi crossbreds were self-fertilised to produce F,.

The number of these ^i plants derived from the 6 rf-parents used as f^,

representing in each case an equivalent number of pollen grains, were

respectively

57
19

Total "91

Doubles occurred in every F, family. Each of the 91 pollen grains
tested 7nust therefore have been carrying the double character.

In the sulphur- white race 7 individuals were employed as the
rf-pollen-parent in matings with 9 individuals belonging to 3 different
pure-breeding strains. The number of seed-parents used in each case
was as follows :

^o-d-glabrou8 cream 5

flesh 3

„ hoary white (Bromptou) 1

Total ... 9

93 of the crossbreds were tested as in the red race. The number of
these ^j plants derived from the 7 c?-parents were respectively

22
22
16
15

Total "93

Doubles were obtained in 90 out of the 93 families. It remains to
consider whether in the 3 families in which no doubles were recorded
their absence is probably real or not. It would seem that in two of the
three cases, at least, we may fairly regard the totals, viz. 8 and 17, as
too small to be conclusive, for we find among the mixed families a case
where the proportion of singles to doubles was as high as 20 : 1 (the
actual numbers were 40 s. 2 d.). This being so, it is clear that the two
cases in question fall within the range of what may be expecte«l from

Joum. of Gen. 1 21

310 Doubleness in Stocks

an F^ crossbred, bred as above, but from which nevertheless doubles
would be obtained if a further sowing was made. The remaining
exception was a family of 33 singles, but even this total constitutes no
very strong case for the genuineness of the exception, seeing that in
another case a result of 40 s. 2 d. (see above) was actually observed.
It represents, it is true, a greater excess of singles than was recorded in
any other family of the same parentage, but much stress cannot be laid
upon this point, since among the mixed F. families obtained when one
of the sap-coloured forms was used as the double-throwing parent in
similar matings, we find a case where the proportion of singles was as
high as 30 : 1 (the actual numbers were 60 s. 2 d.). An equally high
proportion might presumably be obtained with the sulphur-white ; so
that even in this last case it is quite possible that doubles would have
occurred in a larger sowing. Another possibility is worth noting in
this connection. The plant from which the F^, family of 33 singles was
derived was one of 46 obtained from pure-breeding creams which had
been fertilised with the pollen of sulphur-whites. The other 45 all
yielded a mixed offspring of singles and doubles. Now the strain of
sulphur-white used in this experiment evidently did not contain the
colour factor G found in the ordinary pure white glabrous race^ for the
mating with the cream produced offspring which were all cream, and,
as we should expect under these circumstances, all glabrous. Thus the
Fi plants obtained from crossing the cream with the sulphur-white are
indistinguishable in appearance from Fy^ plants derived from the same
cream parent by 5eZ/*-fertilisation. Where F^ shows reversion in colour
and surface character we know that we are dealing with a genuine
crossbred, but in this case we have no such proof It is in fact within
the bounds of possibility that the F^ plant which produced the 33
singles, although supposed to be a crossbred, may in reality have been
a pure-bred resulting from accidental self-fertilisation.

To sum up the evidence in regard to these two double-throwing
forms, red and sulphur-white :

Experiments carried through 6 generations showed that the 149
individuals tested were all throwing doubles. It therefore seems
beyond doubt that both forms are genuinely eversporting — that in

1 As stated in the Evolution Reports one of the two factors C and R which are essential
to the production of sap-colour is found in the pure white race, the other in the cream.
As white is there represented as containing C and cream as containing R, it will be
convenient to retain the same formulae here (see Report IV, p. 36). For a fuller account
of the constitution of the sulphur-white, see p. 370 of the present account.

E. R. Saunders 311

both cases every pollen grain is carrjang the double character. This
view receives strong confirmation from the results of cross-breeding.
184 pollen grains were tested by crossing with a pure-breeding form.
From the mixed character of the F^ families it was definitely ascertained
that 181 of these grains must have been carrying doubleness. The
absence of doubles in the 3 remaining families can scarcely be regarded
as other than accidental, since if genuine it would presumably imply
the production by the double-thro\ving forms of a certain number of
single-carrying pollen grains, a condition which is not borne out by the
results of self-fertilisation.

II. Races which occur both in the form of double-throwing
and non-double-throwing strains.

The question now arises as to the behaviour of those races which
can be obtained both in a pure-breeding and in a sporting form. Are
these d-strains also strictly eversporting ? In these cases is it also
impossible to breed out the doubles ? From the results which have
now been obtained it would seem that to these questions we may safely
return an affirmative answer. It will however be convenient to consider
the evidence from the sap-coloured and the non-sap-coloured forms
separately.

Commercial seed of both double-throwing and non-double-tbrowing
strains was obtained in the case of the two glabrous non-sap-coloured
forms white and cream, and of several sap-coloured forms, viz. very light
purple or azure (both hoary and glabrous), light purple, dark purple,
marine blue, flesh and copper (all glabrous)^ The seed supplied as
giving only singles was found, as previously stated, to answer to
description ; in no case were doubles obtained from such seed either
when the strains were self- fertilised, or bred together. The strains
stated to give doubles were tested both (1) by self-fertilisation which
affords the readiest means of detecting the sporting individual, though
it leaves undetermined the share in the results to be attributed to
pollen and ovules respectively ; (2) by crossing with pure-breeding
strains, a method which enables us to sample ovules and pollen inde-
pendently of each other. In the latter case the experiment has to be
carried to F, before a result is obtained.

^ Unless otherwise stated all races employed in these experiments were of the Ten-
week class.

21—2

312

Douhleness in Stocks

(a) Sap-coloured races.
i. Evidence from self-fertilisation.
The number of individuals tested in each case is shown below

Number of Individuals Tested.

Azure
hoary

Azure
glabrous

Light

purple

glabrous

Dark

purple

glabrous

Marine

blue
glabrous (

Flesh
?labrou8

Parent plants

Fi individuals

derived from
self-fertilisation

(all of
one

(belonging
to three

of parent plants

family)

families)

Fi individuals

derived from

self-fertilisation

(belong-
ing to two

9
(belong-
ing to six

(both from

one

of Fi plants

families)

family)

Fs individuals

derived from
self-fertilisation

(all of
one"

(belong-
ing to five

of F2 plants

family)

families)

Copper
glabrous

Totals
12

0 52

0 13

0 31

Totals

108

Twelve individuals belonging to different sap-coloured forms were
taken at random, and they and 96 of their descendants were self-
fertilised. Doubles were obtained from each of these 108 plants. (For
details see Table III.) Thus the evidence, so far as experiment has yet
gone, indicates that the double-throwing strains of these forms now on
the market are similar to the red and sulphur-white races in that they
are genuinel}' eversporting, and that it is in fact impossible to breed
out the doubles.

ii. Evidence from cross-breeding.

To obtain the further proof that the double character is being
carried by all the pollen in each of these sap-coloured strains necessi-
tates the raising of a large number of F^ plants which have been bred
by self- fertilisation from the mating no-d $ x c? </ where the ^ parent
belongs to the sap-coloured form which is to be tested.

Up to the present 36 ^1 plants representing as many pollen grains
contributed by 6 c?-parents have been tested in this way. The
parentage of these F^ plants and the number of pollen grains tested
in the case of each parent are shown below ; the composition of the F^
families will be discussed later (see p. 336 and Table IV); those marked
thus * have already been recorded (see Report II, p. 37).

£. R. Saunders

313

Number of F-i Mfttinga from which the /i

iriaaU tested pl^ta were derived

6 no-d-cream $ x d-light purple i (plant A )
14 „ „ X „ „ (plant B)

7 no-d-flesh ? x „ „ (plant £)
1 *RO-d-dark purple ? x d „ (plant C)
6 no-d-cream ? x d-azure <r

1 *no-d-flesh ? x d-dark purple i

1 *no-d-white % x d-copper*;

Total 36

Number of pollen

gT»iiu tested in the c

of each ^ parent

Total 36

All the 36 F^ cross-h'eds yielded dottbles in F,; hence all the pollen
tested must have been carrying the double character.
(6) Non-sap-coloured races.

The results recorded in the case of the glabrous white and cream
races are less consistent than those obtained with the sap-coloured
forms, but, if the conclusion which a review of all the evidence seems
to render most probable should prove correct, viz., that in the case of
the non-sap-coloured forms the seed obtained commercially was not
homogeneous but of mixed origin, some of it being pure-bred and some
cross-bred, such admixture would account for the discrepancies observed.
On this view the facts may be taken to indicate that, where pure-bred
material is used, the same results may be expected to follow whether a
sap-coloured or a non-sap-coloured form is employed; and that these
races, when pure-bred, are all in fact like the sulphur-white and the
red, strictly eversporting. The facts in full are given below.

i. Glabrous white race.

Seeds stated to yield doubles were procured from two different
firms. Sample sowings gave the expected mixture. Certain singles
occurring in this first and in later generations were tested as shown
in Table I.

Both lots of seed gave a different result fi-om that obtained with the
sap-coloured forms, for here the singles appeared to be mixed, some
giving doubles according to expectation, others not Thus in the one
lot, plant A, and in the next generation plant K were presumably
breeding true while plant B was not ; in the other lot plants H and /
and 21 of 7*8 descendants evidently belonged to the sporting class, while
plants C, D, E, F, G and / were in all probability breeding true. We
should not be surprised at a result of this kind if, either there had been
some mischance or want of care in the handling of the seed before it
was supplied, in which case we might regard the mixture of singles as

314 Doiibleness m Stocks

accidental and unimportant : or, if we had grounds for supposing that
we were dealing with a race in which some, but not all, of the pollen
grains were carrying doubleness. But neither supposition agrees well
with the facts. The evidence from cross-breeding, so far as it goes,
indicates that here, as in the sap-coloured forms, all the pollen grains
carry the double character; for, as shown in Table I, 20 Fi plants
derived from the mating no-d-creara $ x c^-white </• were tested, the
pollen grains from which they were derived having been furnished half
by plant B and half by plant / ; all yielded doubles when self-fertilised.
On the other hand the fact that results precisely similar to those
described above were obtained with the cream race renders explanation
on the accident theory very improbable.

ii. Glabrous cream race.

Seeds of the cream race were obtained from the same sources as
those of the white, and here too, in accordance with expectation, doubles
occurred in both sample sowings (see Table II).

In the case of lot 1 only three of the singles were tested, one (A) by
self-fertilisation, the other two (B and X) by cross-breeding. All three
yielded doubles, either in Fi (as a result of self-fertilisation or of crossing
with a d-strain) or in F^ (when the mating was with a no-d-strain), as
did also the two descendants of the self-fertilised plant A which were
tested (viz. plants iV^ and 0). In addition to these 5 plants, 12 singles
derived by cross-breeding from B were also tested ; each yielded doubles
in the next generation, a result which further confirms the eversporting
character of plant B.

In the case of lot 2, where more individuals were bred from, the
results indicate on the other hand that, as was found with the white
race, the singles were mixed, some yielding doubles and some not. Out
of 11 singles taken at random from this batch 10 were tested by self-,
1 by cross-fertilisation ; of these, 4; (plants C, D, E, F) appeared to be
breeding trueS the remaining 7 (plants G, H, /, J, K, L, M) produced
doubles either in the next generation, or in F2 if the mating was with
a no-c^-strain as was the case with plant M. In the case of plants
G and H the ofifspring were too few to give indication of the true

1 The number of offspring obtained from plant F by self-fertilisation was only 3 — far
too small a total to be taken as proof that the absence of doubles is real, but the evidence
from cross-breeding leaves no doubt that F was a pure-breeding single. Used as the
S parent in a mating with no-d-white it produced 10 plants in Fi, 9 of which were self-
fertilised, yielding altogether a total of 70 plants in F^ which were all single. Further,
when used as ? with two rf-strains the 23 individuals obtained in Fi were all single.

Table I showing in the case of the white glabrous race the number ar

with a no-d-strai7i, and

Seeds from Source 1

Sample sowing of Plant A x

commercial seed

Seeds from Source 2

Sample sowing of Plant C x
commercial seed |

200x

Plant D X

I
245 X

X (Plant Z)

I
134 X

Plant E X

I
37 X

17 X

I I I I I I I i I I

15 X —X 5x 7x 6x 9x 6x 7x 8x 56 x

5« 1« 6« 3« 2« 1« 1« 4« 3« 21«

Plant i?'x

I
many x
total not recorded

66 X

28 •

tota

I I I

— X —X 3:
6« Am 84

Seeds from Source 1

Table II showing the number and pedigree of the cream plants t

(d-glabrous red ? ) X !

Sample sowing of
commercial seed

Plant A X

1
Plant N X

33 X
62 •

Plant 0 X

52 X
70 •

Seeds from Source 2

i I I I I I

24 X 29 X 5x 36 X 9x 19 x

24« 18« 12« urn 179 25«

Sample sowing of Plant C x Plant D x Plant E x
commercial seed | | |
1 14 X 19 X

Plant P X

19 X

41 X

I I I

9 X 10 X 21 X

Ugree of the individuals tested hy self -fertilisation, or by crossing

\T€SllltS OOtCtmeCl. x =a single individoal. 9=a double indlTidaal.

(no-d-glabroas cream) X Plant Bx + 2 x + 4 <

xxxxxxxx

I I I I I I I I

121 X 56 x 116 X 49 X 107 x 124 x 160 x 8x

32« 20« 37« 25« 39« 57« 47« 6«

1 1

y X

recorded

Plant fix

I
— X
2*

(tu>-d-glabrons cream) X Plant I x Plant Jx +llx +7#

240x

T ! r ~i I I 1 1 1 1 1 — — I 1 1 1 1 1 1

I i I I I I I I I I 1 I 1 I i 1 \ i

xxxxxxxx XX xxxx XX 62 X 107 <

I I i I I I I I I I I I i I I I

X 5x oxlOx 6x 5x Ix 5x 8x 6x 2x 6x 5x Ix 4x 6x
!• 3« 10« 18« 59 im 4« 12« 12« 10« 4« 6« 6« 1« 4« 2«

h?j self-fertilisation, or hy crossing, and the results obtained.

(no-d-glabrous flesh ? ) X Plant Xx + 8 x + 17(

1
1

1
1
x

■ 1

1
22x
13 •

40x
26«

1
10 X
23«

1
20x

169

1
14 X

21*

glabrous white ? ) X Plant F x X ^^-g^abroas red <j and d-glabrous red j X Plant G x
^ 1 I d-glabrous white d | |

, I — ; i

1 r~rn — \ — \ I I

3x
14«

i I I

XXX

X X X X X

! I I I I

5x 4x Ix 8x

8x 23x

X 21 X

I
44x
16«

E. R. Saundkrs 315

proportion of single to double, but both plants were presumably
producing an excess of doubles as was also apparently K, probably J,
and certainly I ; L on the other hand yielded a proportion of about
3 s.: 1 d., i.e., the proportion we should expect from a cross-bred rather
than a pure-bred. In the next generation a single descendant from
each of the two plants / and / was selfed, and both like their parents
gave doubles in excess ; both in short behaved like eversporting indi-
viduals as we should naturally expect. In the case of K and L however
the results obtained in F^ are not so easily comprehended, for in neither
case did all the Fj singles yield doubles in F^. In fact the same
diversity of behaviour exhibited by the haphazard collection of singles
(plants C — if) is here found among the sister plants of a self-bred
family derived from one of these singles (K). 48 ^i descendants of K
were tested, 1 (plant U) by cross-fertilisation only, 47 by self-fertili.sation
either alone or in addition to cross-fertilisation. [Where self-fertilisation
shows that an individual was throwing doubles it is unnecessary for the
present purpose to complicate the pedigree further by introducing into
it the results of cross-fertilisation, and these results have therefore been
omitted where the evidence from self-fertilisation was sufficient.] The
former plant (U) and 40 of the latter again produced doubles in the
next generation, but the remaining 7 yielded only singles, the numbei-s
in these 7 families ranging from 8 to 68. How many among the
40 mixed families can be regarded as showing the true proportion of
singles and doubles is uncertain, since in many the totals are very
small ; moreover the seed was not sown until two years after it was
harvested, and in some cases germinated badly. (See later, p. 361, where
the probability that seeds giving rise to singles and doubles respectively
diflfer in viability is discussed.)

To sum up the foregoing results :

Plant iT as a matter of fact gave a very slight excess of doubles, but
among the Fi singles derived from K some were evidently giving doubles
in the proportion of only 1 d. : 3 s. while others were apparently breeding
true to singleness. Some of the F^ singles similarly yielded the pro-
portion 1 d. : 3 s. in F,. (See Table II.)

In the case of plant L, 31 Fi descendants were tested by self-
fertilisation ; 18 of the resulting jPj families included some doubles,
13 were composed entirely of singles, the numbers in the latter class of
families ranging from 7 to 34. If we review these 18 families we find
that in 13 the numbers agree well with the ratio 3 s. : 1 d., and that in
the remaining 5, none of which included more than 6 individuals, there

316 Douhleness in Stocks

is either equality or a slight excess of doubles. In the next generation
this diversity of behaviour was again apparent ; 3 F^ plants belonging
to 2 Fi families, both of which included some doubles, were tested
by cross-breeding. One was used as (/* in a mating with the d-red
strain and gave a total of 36 all single. Another was employed
as the </ parent with two individuals of the eversporting sulphur-
white race and gave a total of 92 (58 + 34) offspring again all single.
The third individual, which was sister to the last-mentioned plant, and
was similarly used as the f^ parent in a mating with the same two
sulphur-white individuals and also with another sulphur-white plant
gave, on the other hand, a mixture of singles and doubles, the singles
in each case being largely in excess. These results may be summarised
thus : L itself yielded singles and doubles in the proportion of 3 s. : 1 d.;
when self-fertilised the resulting ^i singles proved to be mixed, some
yielding again 3 s. : 1 d., others apparently breeding true to singleness ;
whether also some of the F^ singles were yielding a higher proportion
of doubles than 1 in 4 is not certain. In the F^ generation a similar
result was obtained, some of the F^ singles were evidently breeding true
to singleness, while others gave an excess of singles when mated with
an eversporting form.

There remain the plants M and X about which all that can be said
is that both were evidently able to throw doubles, but whether in excess,
or not, there is not sufficient evidence to determine.

Among the various matings of the cream race with other forms, only
two happened to be carried out in such a way as to enable the double-
throwing character of the pollen to be tested independently of the
ovules as is the case when a no-c^-strain is used as the $ parent in the
cross. 16 of the F^ plants resulting from these two unions were self-
fertilised, viz., 1 Fi from the mating wo-rf-flesh % x rf-cream (plant X)
^ and 15 from wo-c?-white % x d-cream (plant H) ^, and each yielded
doubles in the next generation. The total number of pollen grains
belonging to the two non-sap-coloured forms, which were tested in this
way, is then 20 from the white race (see p. 314) and 16 from the cream,
and all proved to be carrying the double character.

We find then that seed of the double-throwing strains of white and
cream, as supplied commercially, appears to differ from similar seed of
the sap-coloured races in that it is not homogeneous. Though the plants
raised are uniform and true to type in respect of other characters such
as flower colour and character of leaf surface, they differ in behaviour as
regards the double-throwing characters. They behave in fact as we

E. R Saunders 317

should expect a population to behave if it was composed of double-
throwing and non-double-throwing individuals of the same race breeding
indiscriminately together. Under these conditions we should expect
that some individuals would yield an excess of doubles, and that others
would breed true to singleness, and that in the latter case the succeed-
ing generations would be homogeneous and would behave like their
parents. Further, that other individuals though yielding a mixture
of singles and doubles would give an excess of singles. The plants
yielding an excess of singles would be cross-breds due to cross-breeding
between the eversporting and the true-breeding single, and each
succeeding generation of their descendants, if self-bred, would prove
heterogeneous, and give again true-breeding singles and singles giving
a minority of doubles.

From the facts detailed in the preceding pages it therefore seems
reasonable to suppose that in the case of the sulphur-white and the
various sap-coloured strains employed the samples of commercial seed
investigated were harvested from homogeneous populations composed
of eversporting individuals only. That in the case of the white and
cream races the populations from which the seed was collected included
pure-breds and cross-breds, some of the pure-breds being eversporting,
some true-breeding. A sample sowing in the case of these two strains
might therefore very well produce all three types of single, as indeed
appeared to be the case with the cream, where A, B, H, I and J ap-
peared to be eversporting, C, D, E and F true-breeding, L and probably
K either cross-breds or the oflFspring of cross-breds. The remaining
plants, viz. G, M and X, were also producing doubles, but the evidence
is insufficient to determine whether they were pure-breds or cross-breds.
This explanation indeed appears to be the only one tenable, for the
facts which have been given may be taken to put out of account any
question of accident in the handling of the seed before it was supplied,
or of the frequent occurrence among eversporting forms of pollen grains
carrying the single character.

III. Proportion of doubles obtained from the eversporting strains
when self-fertilised or inter-crossed.

i When self-fertilised. (For details see Tables III, VI, VII, and
VIII.)

As previously stated (Report III, p. 45) eversporting individuals,
when self- fertilised, usually give an excess of doubles. It may be noted

318 Douhleness in Stocks

in passing that the proportion of single and double plants among self-
bred offspring of eversporting forms presumably indicates the proportion
of single and double-carrying ovules in the parents, since the % gametes
are being tested, so far as appears, against a uniform standard — the
double- carrying pollen grain. A survey of the numbers recorded (see
Table III), especially where the totals are fairly large, whether obtained
as the result of a considerable sowing from one individual or by summing
the results of small sowings from many individuals, leads to the con-
clusion that the real ratio of single to double is either exactly 7 : 9, or
that it lies somewhere between this and equality. The proportion is
in fact such as we might expect from imperfect gametic coupling
where two pairs of allelomorphs are concerned. Coupling on a 7 : 1
basis for example gives the precise ratio 7:9; a 15 : 1 series gives
7*5 : 8'5 and the next higher term in the series a still nearer approach
to equality. On the whole the balance of evidence seems to point to
a 15 : 1 series, but very large numbers would be required to enable us
to decide this point with certainty, and until these are available we
may conveniently represent the ratio of the two forms by the general
expression

7 -H a; single : 9 — a; double,

where x has some value less than 1. We may suppose that the value
for X is probably the same in all the strains investigated and that the
considerable divergences occurring in many cases where the numbers
are small are not real but the outcome of a topographical scheme of
distribution of the different % gametes, in consequence of which the
flower unit may not afford an average sample. It is as though
the arrangement of the % gametes were regulated by some coarse
mechanism, so that in regard to such small regions as a single fruit
or part of a fruit, there may be great irregularity of grouping. With
a view to avoiding any effect of unconscious selection in the samples
sown, the practice was adopted of sowing all the seeds belonging to
some definite unit or area, as e.g. all from one fruit, or from one side
of the fruit, or if fewer still were required the necessary number were
taken in order from one end of the pod, and not selected at random
from the mixed seed of many pods. Several cases selected for further
sowings on account of the aberrant result obtained in the first instance
from small samples, later gave totals in accordance with expectation.
It is this irregularity of distribution which renders it difficult to
determine whether the higher or the lower of the ratios given above
should be accepted as correct.

E. R. Saunders 319

ii. When inter-crossed.

Table shotoing the totals obtained in F^from various matinga between two
everspor ting forms. (For details of the fcvmilies, see Table VI.)

Tjrpe of Union
d-glabrous cream ? x d-glabrous white ^

,, ,, 9 X d-glabrons red (f

d-glabroas red $ x d-glabrous cream cf

,, „ ? X d-glabrous sulphar-wbite i

d-glabrons sulphur- white ? x d-hoary white <?

„ ,, ,, ? X d-glabrons red (f

,, „ ,, ? X d-glabrous white (? ...

,, ,, ,, ? X d-glabrous azure (f ...

,, ,, ,, ? X d-gabrous light purple <f ...

d-hoary azure ? x d-glabrous sulphur-white <f

d-glabrous flesh $ x d-glabrous azure <;

d-glabrous light purple ? x d-glabrous red <f
d-glabrous red ? x d-glabrous light purple cf ...

d-glabrous azure ? x d-glabrous red <f

Totals exclusive of cases where suspected ' creams were used 50

d-glabrous cream ? x d-hoary white ^

,, .. ? X d-glabrous sulphur white <j

d-glabrous sulphur- white ? x d-glabrous cream <?

d-glabrous cream ? x d-glabrous flesh ,?

d-glabrous flesh $ x d-glabrous cream <?

d-hoary azure ? x d-glabrous cream <?

Totals including cases where suspected 1 creams were used 60 785 888

Table sJiowing the totals obtained when the F^ cross-breds from the above
matings between eversporting forms were crossed back with one of the
eversporting parent types. (For details of the families, see Table VIII.)

Type of Union

Glabrous red? ^ (glabrous red x glabrous

sulphur- white) <J

Glabrous red? ^ (glabrous sulphur-white

X glabrous red) <? ...

(Glabrous sulphur- white x ) , , ,

glabrous red) ? } x glabrous red cT

(Glabrous sulphur-white x | , , , , , ..

glabrous red) ? | x glabrous sulphur-white <?

(Glabrous red x glabrous ) , , , , , .^

sulphur-white)? J x glabrous sulphur- white <r

Glabrous cream x glabrous ) i . ,

sulphur-white)? [ x glabrous red <r

Totals 35 421 505

* Suspected, that is, of being of cross-bred origin and not truly eversporting.

Number

of
mAtlngs

Number

of
singles

Number

of
doubles

147

163

108

231

311

d 50

586

757

139

Number

of
matings

Number

of
singles

Number

of
doubles

305

377

320 Douhleness in Stocks

As shown in an earlier account* and in the accompanying Tables,
niatings between two double- throwing parents invariably gave doubles
in the next generation. We should expect, unless other complexities
are indicated, that the results in such matings will be the same as when
either c?-parent is self-fertilised, and that the proportion of doubles
obtained from a cross will, as in the case of self-fertilisation, furnish
an index of the proportion of double-carrying ovules produced by the
d-seed-parent.

Altogether 1673 plants were raised in the present experiments in
Fi, from 60 matings between various rf- strains, representing an
equivalent number of ovules from 45 c?-seed-parents ; or, if we exclude
all matings in which the cream plants K and L were employed on the
ground of suspicion as to their purity, we have 1343 F^ individuals
from 50 matings in which 37 cZ-seed-parents were used. If all these
50 families are summed together, we get a total of 586 singles and
757 doubles, or almost exactly 7 s. : 9 d.^ A certain number of these
c?-seed-parents were also self-fertilised, and we are therefore able to
compare the effects of own pollen versus foreign pollen on identical
individuals.

The results were as follows:

From crossing 15 individuals with pollen from

d-individuals of other strains 248 singles 298 doubles

where a ratio of 7 s. :9d. would give ... 239 ,, 307 „

or 7-5 8. : 8-5 d. would give ... 256 „ 290

From self-fertilisation of these same individuals 437 ,, 539 ,,

where a ratio of 7 s. : 9 d. would give ... 427 ,, 549 ,,

or 7-5s. :8-5d. would give 457| ,, 518^ „

Again, 35 matings were made, in which jPj from two eversporting
forms was crossed back with one of the eversporting parent types.
If the 35 F^ families are summed together, we get a total of

421 singles and 505 doubles,
where a ratio of 7 : 9 would give

405 singles and 521 doubles,
and a ratio of 7*5 : 8'5 would give

434 singles and 492 doubles.

1 Rep. Evol. Committee, IV, 1908, Table II. p. 40.

2 In certain cases the results obtained from one fruit were unexpectedly divergent
from those of another, and a considerable sowing had to be made before an indication of
the probable ratio was obtained. But where large numbers were recorded the results
appeared sufficiently uniform to justify the inclusion of the whole series in one total
as above.

E. R. Saunders 321

Numbers approximating either to the one ratio or the other were
similarly obtained when these F^ cross-breds were «e(/'-fertilised. These
results will be discussed later. (See Section V, and Table VII.)

From these facts it seems clear that among these eversporting forms
self-breeding and inter-breeding give similar results.

IV. Constitution of the zygote and segregation in the
eversporting forms.

In any attempt to construct a formula which shall represent the
behaviour of the eversporting forms, so far as it is at present known,
the following points must be taken into account:

(1) All the self-bred single descendants of an eversporting indi-
vidual appear to be also eversporting.

(2) It also seems certain that all individuals of the eversporting
types employed yield an excess of doubles. The evidence points to
a proportion of 7 -I- ir single to 9 — a? double, the value of x being less
than 1. We may take it that the occurrence of such a ratio precludes
the possibility of the appearance of singles and doubles being determined
by the presence or absence of one factor only.

(3) All the pollen grains of an eversporting individual apparently
carry doubleness, whereas the ovules are mixed, some carrying double-
ness and some singleness.

(4) The inheritance of singleness and doubleness appears to be
quite independent of surface character (whether hoary or smooth) and
of the character of the sap (whether coloured or colourless), but in
certain cases, if not in all, it seems to be in some way bound up with
the inheritance of plastid character (whether white or cream). In the
present account therefore we may disregard surface character and sap
colour, but it will be convenient to consider plastid character simul-
taneously with that of singleness and doubleness.

With regard to the conclusion given under (2) it may be noted that
where more than one factor is concerned in the manifestation of any
character it becomes theoretically possible for this character to appear
on crossing, even though both parents are breeding true in regard to
its absence. In a case, however, where the factors are not distributed
equally among pollen and ovules, this unequal distribution may in fact
prevent such a possibility, and this appears actually to be the case here.
As yet no mating between two true-breeding singles has ever produced
doubles. (For details of some of these cases see Report IV, Table III,
p. 40.)

322 Douhle^iess in Stocks

(a) General considerations in regard to singleness and doubleness
apart from plastid character.

The requirements 1, 2, 3, stated above, which concern the general
occurrence of singles and doubles apart from complications connected
with plastid colour, would be met if we suppose

(1) That singleness — the dominant character — results from the
presence of two factors {X and F), doubleness from the absence of
either or both.

(2) That in the eversporting forms these two factors are carried
only hy the ovules and exhibit partial gametic coupling^ The propor-
tion of singles and doubles obtained points to a coupling either on
a7:l:l:7 or on a 15:1:1:15 basis. Breeding experiments on
a much larger scale than it has yet been possible to attempt would be
required to determine with certainty which term in the series represents
the truth. But, as will shortly appear (see p. 324), certain results in
which plastid character has also to be taken into account are more
easily explained on the supposition of a 15 : 1 : 1 : 15 series, and we
may therefore adopt this value provisionally for the purpose of a working
hypothesis.

On this view we may represent the eversporting zygote (so far as
singleness and doubleness alone are concerned) thus

Xx Yy

and its gametes, where 2n are required to exhibit the whole series, in
general terms thus^

^ Coupling of the kind here indicated was first described by Bateson and Punnett in
the case of the Sweet Pea, where it was found that purple flower colour was partially
coupled with long shape of pollen. (See Rep. Evol. Committee, III, 1906, p. 9, and
IV, 1908, p. 3.) Several other instances of this kind are now known. For reference to
some of the more recently investigated cases, see Bateson and Punnett, Proc. Roy. Soc.
Series B, Vol. 84, 1911, p. 3.

^ It is realised that the fact that the pollen of eversporting races appears only
to carry doubleness merely proves the absence of X and Y in combination not of X alone
or Y alone. The absence of either factor alone is deduced from the fact that all the
singles of an eversporting race appear to behave similarly, which would not presumably
be the case if some of the pollen carried Z or 7 and some did not. Were the pollen
thus heterogeneous we should expect eversporting singles to be of two kinds, yielding
different proportions of singles and doubles. We have no knowledge as to the cause of
this inability on the part of the eversporting pollen to carry the factors X and Y. If we
suppose that the quality maleness {31) in this case repels the factors X and Y, the
assumption, though it accounts for the fact in question, carries us no further. It is
almost unnecessary perhaps to add that when it is stated that the pollen is unable to
carry a particular factor, the meaning which the statement is intended to convey is that
the pollen is not carrying that factor in a form in which its presence can be detected.

E. R Saunders 323

Ovules

Pollen

n-1 XY

all xy

1 Xy

1 x7

n — lxy

or if n is taken as 16, thus

Ovules

Pollen

15 XY

all xy

1 Xy

1 xY

15 xy

(b) Consideration of the relation of singleness and doubleness to
plastid character.

All forms having colourless plastids, and breeding true in this
respect, may be supposed to contain a factor W, and to be homozygous
in respect of this factor. In true-breeding cream forms W is absent.
In the sulphur-white race the case is more complex. The factor W,
though present in some of the ovules, is evidently absent from the
pollen, since matings with pure cream as % yield only plants with
cream plastids (see later p. 352, also Evol. Rep. IV. p. 39). Further
there appears to be some complex relation between this factor W and
the factors for singleness and doubleness, since, as previously stated, the
sulphur-white i-ace, even when self-bred, always gives a mixture of single
whites and double creams with a small percentage only of double whites
and apparently no single creams ^ These results can be accounted for
if we assume that in this form W is either coupled with, or repelled by,
one of the two factors X or F. The assumption that repulsion occurs
can only be made to fit the results observed if other assumptions are
also made involving further complexities ; it will therefore be simpler
to proceed on the hypothesis that the relation is one of coupling. We
will suppose the coupling to be between W and X, we shall then

^ More than 2000 plants have been recorded, and none of the singles among them had
cream plastids. The non-appearance of the single cream therefore seems hardly likely to
be due to accident. Should this however eventually prove to be the case it would be
unnecessary to assume the existence of any special relation between W and either X or Y
(as described above), since the scheme of coupling described for the two factors X and Y
extended to cover the case of three independent factors (X, Y and W) would give all four
forms in the required proportion, viz. many single white and double creams, a few double
whites and still fewer single creams.

324

Douhleness in Stocks

express the composition of this and other eversporting forms more
fully thus:

Sulphur-white race Pure cream forms

XxYyWw XxYyww

ovules pollen ovules pollen ovules pollen

15 XYW all xyW 15 XYW all xyw 15 XYw all xyw

Zygote
Gametes

Pure non -cream formal
XxYyWW

1 XyW 1 XyW

1 xYW 1 xYw

15 xyW 15 xyw

Expectation on self-fertilisation

480 single whites
544 double ,,

7*5 single whites
8-5 double ,,

480 single whites
512 double creams

32 double whites

or
7'5 single whites
80 double creams

•5 double whites

1 Xyw

IxYw

15 xyw

480 single creams
544 double „

7*5 single creams
8*5 double ,,

These formulae would give the uniform result of 7'5 singles : 8'5
doubles for all eversporting forms, and would explain the occurrence of
a small percentage of double whites in addition to single whites and
double creams, and the absence of single creams when sulphur- whites
are self-fertilised. The fact that the number of these double whites is
sometimes below, rarely above, the estimated proportion of "5 in 16 or
about 3 per cent, is an important point in favour of the adoption of 16
as the value for n rather than 8. Were a 7 : 1 : 1 : 7 series taken as
representing the gametic output, the expectation in regard to double
whites would be just twice the number given above.

We have now to see how far the assumptions made above in regard
to the relations existing between the factors for plastid colour and for
singleness and doubleness will satisfactorily explain the distribution of
forms with white and cream plastids among the singles and doubles
when the various eversporting forms are inter-crossed.

V. Segregation in Fi cross-breds derived from two eversporting forms,
and statement of results obtained in Fo.
As shown above (p. 320) matings between two eversporting forms
were found to give a slight excess of doubles as in the case of self-
fertilisation of either of the parents. This is in accordance with the

1 The term ' non-cream ' is used to indicate any form with uncoloured plastids
irrespective of the colour of the sap.

E. R. Saunders

325

scheme suggested above (p. 323), according to which the distribution
of the factors for singleness and doubleness is assumed to be the same
for all eversporting forms. We should naturally expect that a condi-
tion which obtains in both parents would also hold good in their
cross-bred offspring, and we may therefore conclude that, so far as the
factors for singleness and doubleness are concerned, the general scheme
of segregation in ^i cross-breds derived from two eversporting forms
will be the same as that put forward in the case of the parents (see
p. 322).

The distribution of plastid colour needs further consideration, since
in respect of this character reciprocal cross-breds from unions between
eversporting cream and non-cream forms give different results. Matings
of this type can be carried out in six different ways as shown below,
where the unions 2, 4 and 6 are the reciprocals of 1, 3 and 5.

Mating 1 d-snlphar -white ? x d-non-cream i

2 (2-non-cream ; x d-salphor-white g

3 d-non-cream 2 x d-cream ^

4 d-cream ? x d-non-cream ^

5 d-8ulphur- white ? x d-cream <f

6 d-cream ? x d- sulphur-white <f

The composition of the ovules and pollen uniting to produce the
single plants in F^ according to the scheme given above, together with
a general statement of the results obtained in F. for those cases which
have already been carried out, is given below.

Mating

2
8

Constitution of the gametes

anitiog to produce the

single plants in fi

Results obt&ined in F«

Omles

XYW
XTW
XYW

XYw

XTW
XYw

Pollen

xyW

xyw

xyW

xyw

Singles

with white

plastids

many

(a) many

(b) many

Singles

with cream

plastids

Doubles Doubles

with white with cream

plastids plastids

none many

(not yet carried out)
none none

few many

(not yet carried out)
(not yet carried out)

many

none

few(?)

Only three of these matings have as yet been carried to F^. It is
doubtful however whether the results of unions 2, 5 and 6, when
available, will throw any further light on the relation existing between
plastid colour and singleness and doubleness, since we may suppose

Joum. of Oen. i 22

326

Douhleness in Stocks

2 ^Zs-o
Wo2

saiqnoa

.saiStng

•-I 92 '-' «2 :2!

1/5 eft rH »-l

rH OS f-l «0 22
O OJ >-l OS "^

CC « «0 «£

seiqnoa QO «£ '-'

L BaiSnjg « S '"'

sppsBtd , , ,
saiqnoo;

SpnSBld _H (M W5

ajiqiimiM S O i-H

r-l w ^

gpi^sBid
aresJo qiiA

B8l3Utg

»o «S "^^

gpnseid
e^iqM q^m » ^
S8x3uis

eo "5 =2 t S
S ^ "^ S ^

gaiqnopmox O q|

— H OS OS

aonpojd o^

pasn sx^npiA ^ qo (M us 2
JO jaqintiK!

5*j

= ^
•-t! m

eS _d _e8 ^

"ao 6C SO 60

tg -rj "« 'tt

X XX X

•a

•3

B. R Saunders 327

that in case 2 the result will be the same as in case 3 which is already
known ; while in raatings 5 and 6 the result will presumably be the
same as if the seed parent had in each case been self- bred instead of
cross-bred. We have then to consider the results in matings 1, 3 and
4 in the light of the assumptions made above.

Mating 1. rf-sulphur-white ? x d-non-cream ^.

Five diflferent matings of this kind were made (see p. 326).

This type of mating gives only white plastids in JP,, a result fully in
accord with expectation. For the sulphur-white ovules which carry
creamness, by assumption, also carry doubleness ; hence when this race
is crossed with any non-cream form w^hose pollen carries doubleness,
cream will not presumably reappear in any succeeding generation. It
will have been bred out completely, though in the recessive condition,
in the F^^ non-cream doubles. Thus only those sulphur-white ovules
which carry the white plastid factor give rise to singles in F^. These
F^ singles are therefore all homozygous as regards the factor TF, since
the mating will have been between XYW ovules and xyW pollen;
segregation in their case may therefore be expected to proceed on the
same lines as in a pure-bred d-race with uncoloured plastids, and to
yield a similar proportion of singles and doubles. As regards the
question whether a ratio of 7*5 s. : 8"5 d. or 7 s. : 9 d. more nearly repre-
sents the facts, it happens that the numbers obtained in this case agree
better with the latter alternative (7 s. : 9 d.). It is just worth noting
however that the mating in which the largest record was obtained, both
absolutely and in proportion to the quantity of seed sown, and in which
therefore the result might be supposed to be the most reliable, viz. the
mating with d-glabrous red, leaves either alternative equally probable.
In three out of the five matings the experiment was carried to ^3,
where the complete breeding out of the cream was further confirmed,
but where again the evidence is not quite decisive as between the two
ratios. For, though the sum of the three totals gives 7 s. : 9 d., one of
the two larger sowings gives almost exactly 7'5 s. : 8'5 d. The results
are summarised below (see p. 328).

We may then conclude that in the unions of the form rf-glabrous
sulphur-white % x d-glabrous non-cream <^ segregation in single Fi
plants and in later self-bred generations is like that in any pure-bred
eversporting glabrous form with white plastids.

28—8

328

Douhleness in Stocks

.2 S^5_j
wop.

8^.2 "P."-
Wog, «^

saiqnoa

sax8nig co g

l-H

spiisw[d
saxqnoo:
spuseid
sgxqnod

g3X3u|S

spusBid
ga^auig

(M IN

saxSnis \v\03. \n 05

aonpojd «n
pasn si^npiA '^ ^

-ipni ''J
JO laqranil

« 3 45
^ -« 13

E. R. Saundehs

329

wiqnoa g

»I^8 sJ «

O
H

S3i8ins

saiqnoQ **

spii8«id
*»tiii qiim I I
saiqnoa

spi^ffBtd

8pt)8«id
«>niii q»ui 5 g

88iqnopi«;ox S S

aonpoido}
pasn ErprapiA cfl ih

-rpui i^ ' "H
JO laqinn j{

I I

oq S'

Ph Cm

►*;

a
1— 1

"3.

•3
.5

t-i

.2P

■3

«^

hi
0
0

X
X

S
IS

-*a

-4-1

■♦*

[5:

"3)

0
0

'>^

•-•

0
0

C3
(D

"S ^•

0 0

2
9

0 0

S 2

94 »4

&, »

O eS a *?

•s

330 Doiibleness in Stocks

Mating 3. c^-glabrous non-cream $ x c?-glabrous cream ^.

Only one mating of this kind was made, two cream plants being
employed as the pollen parents (see p. 329).

The totals from this mating were 270 singles and 281 doubles,
a result which agrees with the view provisionally adopted that
7 + a? : 9 — a; rather than 7 : 9 probably represents the true ratio of s. : d.
All but 8 plants were flowered and the 543 individuals recorded
included singles with white plastids and doubles with cream plastids, but
the reciprocal combinations of whiteness with doubleness and singleness
with creamness did not appear. Now in matings of this type all single
^1 plants will presumably be derived from the meeting of XFTT ovules
with xyw pollen, i.e. from unions in which all three dominant factors
are carried by the ovules and none by the pollen. In other words the
union is a union between the combination white plastid colour with
singleness brought in by the female parent and creamness with double-
ness brought in by the male. Since out of a total of 548 plants in F.^
all were either singles with white plastids or doubles with cream plastids,
it follows that redistribution of the factors in such a way as to lead to
the combination of singleness with creamness or of doubleness with
whiteness either does not occur when segregation takes place, or it
must occur very rarely. The above result seems to necessitate not
only that the F^ pollen should, as already inferred on other grounds
(see p. 321), all carry doubleness, but also that it should all carry cream-
ness. There is in fact a strong presumption that we have in these F^
plants a condition similar to that obtaining in the sulphur-white race.
In both cases the singles result from the union of the same combinations
of factors {XYW % x xyw (^). If none of the pollen of the sulphur-
white can carry the factors XYW although all three are present in the
sulphur-white zygote, it need not surprise us if the same should hold
good for an F^ cross-bred of the same composition. The absence in F^
of singles with cream plastids would thus be explained. The absence
of doubles with white plastids would seem to show that the factor W
must stand in some different relation to X and Y in the pure white
race to what it does in the sulphur-white. In the present case W,
which is introduced into the cross in combination with XY, appears
only to occur in combination with XY in the gametes of F^. If any
gametic combination is formed in which W is dissociated from X or F,
such as is presumed to occur in the sulphur-white, it must evidently be
rare, since no indication of such a gametic combination was apparent in
an F^ population numbering 543.

E. R. Saunders 331

If the above account is correct, then so far as can be seen a similar
result will ensue in the succeeding self-bred generations F,^ F^, &c.,
since in each case the singles produced by self-pollination would appear
to result from the meeting of XYW ovules and xyw pollen. The record
obtained in ^j, so far as it goes, is entirely confirmatory. 54 families
were raised and all included some doubles, a result which may be taken
to establish the double-carrying character of all the Fi pollen grains
from which the F^ parents were descended. The totals obtained in the
whole number of ^3 families were 354 singles and 372 doubles. Only
a few plants in each family were flowered ; they proved to be again all
singles with white plastids and doubles with cream plastids, thus
confirming the results obtained in F^.

We are thus led to conclude in regard to unions of the form
d-non-cream ? x d-cream j/": (1) That segregation in the male cells
of the cross-breds exhibits the same peculiarity as in a sulphur-white,
and that the pollen grains do not carry XYW, although all these three
factors are present in the F^ zygote. Thus the pollen of F^ is similar
to the pollen of the ^^ parent which was used to produce it. (2) That
the factor for whiteness (W) which is brought into the pedigree by the
? parent in combination with X and Y remains in association with
X and Fin the ? gametes of ^i- If exceptions occur in either case
they must be extremely rare.

Mating 4. d-glabrous cream ? x ti-glabrous and d-hoary non-
cream f^.

The matings were the converse of those just described, d-cream
being here used as the $ parent, d-red as the </.

Three different crosses of this kind were made, three cream plants
being used as male (see p. 332).

It will be convenient to consider the results under (a) and (6)
separately, since it may be that the appearance of an jPj single with
cream plastids in (6) — a combination not recorded in (a) — is due to the
impure nature of the cream plant used as the pollen parent in the
(6) mating.

In matings of this type all single Fj plants will presumably be
derived from the union of XYw ovules with xyW pollen, so that in
this case only X and Y are introduced by the female parent, W being
brought in by the pollen. We may therefore assume that here W will
be carried by at least some of the ^1 pollen, though whether by all, and
if not by all, by what proportion, we cannot on purely a priori grounds

332

Doiibleness in Stocks

predict. Similarly we may suppose that some of the F^ ovules which
carry XY will lack W, though whether all will prove to be thus
deficient we also cannot foretell.

MATING 4. Summary of results.

{«)^

Parental Tyi)es
d-glabrous cream (Plant fl) ? x d-glabrous red i

„ „ (Plant G) ? X d-glabrous red cT

„ „ (Plant (?) ? X d-glabrous white (?

St,

•si?

.23 o

same plant
after 3 years

of these after
3 years

2
after 2 years

Results obtained in Jg

^ .2

^as

«.-S--3

—

4
53

2
36

94 141 79 —

97 175 56

4
52

110
149

Totals in the case of the immediate sowings
,, „ ,, delayed sowings ^

49 66 48
193 320 137

65
263

,, . f ? d-glabrous cieam ? d-hoary white (? 2 9

^ ' \ (descendant of Plant K) ^ (intermediate strain)^ (after 3 years)

19 10

The experimental result was an F^ of 49 s. and 66 d. in the case of
the immediate sowings, and 193 s. and 320 d. in those that were delayed
until the second or third year after harvesting. 513 plants were
flowered and all had white plastids. Singles and doubles with cream
plastids either do not occur, or if they are formed they must be present
in very small numbers. We may therefore conclude that all (or all but
very few) of the F^ pollen grains must carry whiteness as well as double-
ness. Thus both here and in the reciprocal type of cross, so far as
experiment has yet gone, we find that the Fi pollen appears to carry
chiefly if not exclusively the particular combination of factors which is
present in the pollen of the father. Although the F^ zygote is hetero-
zygous in regard to these factors, the allelomorphs brought in by the
mother appear to be absent (or if not wanting altogether then very
rare) in the male cells of the cross-bred. But if all (or almost all) the
Fi pollen is carrying W it is evident that the distribution of W among

^ See Appendix, note 1, p. 361.

2 See note p. 326.

E. R. Saunders 333

the ovules can only be certainly determined by breeding to F„ since
the result in F^ will be the same (or almost the same) whatever be the
proportion of ovules containing W to those lacking it.

We may surmise however from the behaviour of the cross-breds in
the reciprocal union that most (if not all) of the ovules carrying single-
ness will lack W, and conversely that W will be present in most (if not
all) of those which carry doubleness. There seems in fact strong reason
to suppose that in matings between eversporting forms of unlike plastid
colour, the plastid colour allelomorphs are associated in F^, chiefly or
exclusively, the one with singleness the other with doubleness according
as each is associated with singleness or doubleness in the germ cells
which united to produce F^.

The cream plant used as the seed-parent in experiment (6) was
a descendant of plant K, an individual which yielded an excess of
doubles on self-fertilisation but which nevertheless is under suspicion
of not having been a genuine eversporting type, since some of its
ofifspring were found to breed true to singleness (see p. 315). This
being so, it is not improbable that this particular descendant of K was
also not genuinely eversporting but some form of cross-bred. It is
therefore at present uncertain whether we shall be right in ascribing
the appearance of the one single with cream plastids in the case of the
(6) mating to the spurious character of the cream plant used as the
% parent in this case, or whether we are to suppose that this form
would also have appeared in the (a) results if a larger sowing had been
made. For it is not certain that a total of 185 -f 328 = 513 plants,
recorded when the cream plants G and H were used, is sufficiently large
to exhibit the complete series in F^.

We may then state the conclusions in regard to matings between
eversporting forms thus :

1. Segregation in ^i cross-breds from two eversporting forms follows
the same course as in the eversporting parents, so far as the factors
X and Y are concerned. (See above, p. 320, where it is shown that
cross-breeding and inter-breeding between these forms give the same
proportion of singles and doubles.)

2. If the eversporting parents are of unlike plastid colour, all or
almost all the F^ pollen carries the same allelomorph for plastid colour
as the pollen which was used to produce F^; similarly, the other member
of the pair, which is borne only or chiefly by the ovules, is borne only
or chiefly by those ovules carrying the same combination of factors as
the ovule from which the F^ plant in question was derived, viz. the

334 Doubleness in Stocks

combination XY. Thus, where Fi is derived from ci-non-cream
$ X c?-cream ^f, all or almost all the F^ pollen carries lu, whereas W
is carried only or chiefly by the ovules carrying XY. When, on the
other hand, F^ is derived from cZ-cream (/• x c^-non-cream $ it is IT
which is borne by all or almost all the pollen, and w only, or chiefly by
the ovules carrying XY.

It remains to consider how far the scheme suggested will serve to
explain the results obtained when eversporting forms are bred with
true-breeding types.

VI. Constitution of the zygote and segregation in the
pure-breeding (non-double-throwing) strains.

The only information to be gained from the self- fertilisation of the
pure-breeding forms is that they breed true to singleness and to plastid
character. From this fact together with the evidence obtained from
the testing of the ovules and pollen independently by crossing, which
shows that the above result is not due to any complete coupling of the
factors concerned with either kind of germ cell, it follows that all these
forms are homozygous as regards X and F: those with white plastids
are also homozygous as regards the presence of W, those with cream
plastids as regards its absence (= w). A detailed consideration of the
results of crossing shows however that the simple formulae XYW and
XYw do not fully express the complex relations existing between these
pure-breeding strains and the eversporting forms when the two are
inter-crossed. It is evident from the results of these matings that
X and Fdo not occur under the same conditions in the pure-breeding
strains as in the eversporting forms ; otherwise the results as regards
the occurrence of doubles would presumably be the same when the
XY ovule of a pure-breeding individual meets the xy pollen of an
eversporting type, as when the XY ovule of an eversporting type is
either fertilised with its own pollen or with that of any other ever-
sporting form. But this is not the case, the percentage of doubles in
the one case (pure-breeding x eversporting) being always very much
fewer than in the other (eversporting x eversporting).

The results of matings between typical d- and wo-d-strains in fact
suggest that the resulting Fi cross-breds are only able to form those
gametic combinations of the XxYy allelomorphs from which they
themselves arose; if gametes are formed carrying different combinations
of these factors, they must be extremely rare. Thus when an F^

E. R. Saunders 335

cross-bred has been produced by the union of germ cells carrying XF and
xy respectively, it will again produce germs of XF and xy composition,
but the combinations Xy and xY apparently do not occur. X and F
behave in fact as though they were coupled, a condition which we may

represent thus XY. It will be well to point out that inability to throw
doubles need not necessarily imply that an individual is homozygous in
regard to the condition in which the factors X Fare present. For we may
suppose that if at any time an XF ovule of an eversporting form were
by chance crossed with pollen from a pure-breeding single of XF com-
position— a possibility which might easily come to pass now and again
when the diflferent strains are grown side by side in the open — we should
at once get a zygote of XYXY composition. This zygote and all its
posterity would behave as any tnie-breeding single so long as self-
fertilisation or inter-crossing among the offspring continued. The
heterozygous nature of a certain proportion of the individuals would
remain undetected and would be perpetuated indefinitely under these
conditions. It would only become apparent if any of these individuals
were crossed with an eversporting form ; then the different proportion
of doubles occurring in F^ families derived from sister F^ plants, on
crossing with an eversporting form, would disclose the peculiar hetero-
zygous nature of the nevertheless true-breeding parent.

Similarly it would seem that among true-breeding singles with
white plastids some may have the factor W coupled in some or all the
germ cells with the XF group — a condition which may be indicated
thus XYW — so that when such germ cells unite with an xyw ovule
or pollen grain of an eversporting form the resulting ^i cross-bred is
unable to form the combinations XYw or xyW. In other cases on the
other hand W appears not to be thus coupled, the results indicating a
redistribution of the factors in the ordinary way. Individuals breeding
true to singleness and to whiteness may therefore conceivably be of six
different kinds, according as the factors X, F, and W are wholly,
partially, or not at all coupled in the zygote, as shown below.

Possible types of true-breeding singles with white plastids.

Zygote Gametes

Homozygons

1 XYWXYW all XYW

2 XYWXYW all XYW

3 XYWXYW all XYW

336

Zygote
Heterozygous

4 XYWXYW

5 XYWXYW

6 XYWXYW

Douhleness in Stocks

Gametes

XYW and XYW
XYW and XYW
XYW and XYW

VII. Segregation in F^ cross-breds derived from unions between ever-
sporting and non-double-throwing forms, and statement of the results
obtained in F^.

The various possible unions coming under this head are the
following :

Eversporting form employed as seed parent

A.
Mating

d-cream ?
d-non-cream ?
d-cream ?
d-non-cream ?
d-sulphur-white ?
d-sulphur- white ?

X 7io-d-cream <?

X 7io-d-non- cream <?

X 7M)-d-non-cream <?

X 7io-d-cream s

X no-d-cream <r

X no-d-non-cream <?

B. EversiMrting form employed as pollen parent

Mating 7
8

10
11
12

7M)-d-cream ?
no-d-non-cream ?
no-d-non-cream ?
no-d-cream ?
no-d-cream ?

X d-cream <?

X d-non-cream <?

X d-cream i

X d-non-cream <f

X d-sulphur-white ^

no-d-non-cream ? x d-solphur- white i

In the case of matings 7 — 12 where the eversporting form is used
as the pollen parent we may expect every F^ family to yield doubles in
F^, whereas in the reciprocal unions (matings 1 — 6) some of the Fi
individuals may be expected to yield doubles and some to breed true
to singleness. Reference to Tables IV and V will show that such
was the case in each of the 8 types of union which have already been
carried out.

With regard to the proportion of doubles occurring in those F^
families which are mixed, it has been stated in the earlier accounts'
that they occur in the proportion of the simple Mendelian recessive,

^ loc. cit.

E. R. Saundbes 337

viz. 1 in 4. This appears undoubtedly to be the case in the great
majority of families, but there are a certain number of cases in
which a considerably lower percentage of doubles was obtained, while
occasionally the proportion was in excess of this amount. The cases in
which the proportion of doubles is very small are so marked and, in
some unions, of such frequent occurrence, that it seems clear that they
cannot be regarded as other than genuine — that they must in fact
represent a distinct ratio and not an extreme variation from the usual
3 s. : 1 d. On the other hand it seems highly probable that in the one
or two cases where the proportion of doubles recorded is distinctly
higher than 1 in 4 the excess is accidental.

It is the frequent occurrence of numbers approximating to the ratio
3 s. : 1 d. which suggests that, in the wo-d-strains, the factors for
singleness {X and Y) are so coupled that re-combination with x and y
in the gametogenesis of ^i cannot occur, in the manner described
above for matings between two eversporting forms, where ovules with
X and Y uncoupled meet xy pollen grains. This condition of single-
ness which is typical of non-double-throwing forms is, as stated above,
conveniently represented thus XY: and since as regards singleness
and doubleness, reciprocal heterozygotes of similar composition give
similar results, we may write XY for both the ovules and the pollen
of a typical no-d-ioTxn. But, as explained above (p. 335), a single
might breed true and yet not be pure-bred, a fact which should
not be overlooked in considering any unexpected result in F^. For in
any cross between an eversporting and a true-breeding type, made
in the form rf $ x no-d (^, a certain number of the F^ singles will have
the composition XY XY; they will breed true to singleness, and on
self-fertilisation will be indistinguishable from a pure-bred true-breeding
single having the composition XY XY. But the cross-bred true-breed-
ing single will presumably behave differently from a typical pure-bred
single, when crossed with an eversporting form; since in the one mating
the xy germ cells of the d-type will unite with certain germ cells in
the no-€?-type carrying an uncoupled XY group, and in the other, not.
We have already seen that of the double-throwing plants assumed at
first to be pure-bred eversporting, some were probably cross-bred ^ and
it may well be that now and again the same may be found to be the
case with a supposed type single. From such a true-breeding but

1 See pp. 313 — 317 where an account is given of the behavioor of commercial samples
of the white and cream strains.

338 Douhleness in Stocks

heterozygous single, when crossed with an eversporting form, we might
well expect a proportion of the F^ plants to yield a higher percentage
of doubles than 1 d. : 3 s. Of such cases however we have as yet no
certain example. It is otherwise however as regards those F^ families
which show a deficiency of doubles. Some of these cases, at least, seem
beyond question, but until we have a fuller understanding of the real
meaning of coupling it is difficult to obtain a clear insight into their
cause.

The results seem to show that the lower proportion of doubles
obtained from some ^i cross-breds in matings where sister F^ plants
gave 3 s. : 1 d. cannot be considered in all cases as an effect due to eiih&r
one of the parents apart from the other, but must be regarded as due
to a combination of factors brought together by their union. It was
found, for example, that the same true-breeding individual may give
only the usual proportion of doubles (3 s. : 1 d.) in F^ when crossed
with one eversporting strain, but will show a marked deficiency of
doubles in some families when crossed with another d-strain (see
Table IV, where 6 Fi derived from the union no-d-cream x d-azure
all gave the usual proportion of double plants, while in F2 from a
mating with c?-light purple, where the same cream individual had
been used as the seed parent, one or two families indicated a marked
deficiency of doubles). These facts suggest the possibility that the
conditions which result in the production of a single or a double may,
in some cases, be more complex than those represented in the simple
formula hitherto employed, according to which the occurrence of a
single is attributed to the presence of two factors (XY), the occurrence
of a double to the absence of either or both. This may remain true,
and yet it may also be that more than one such pair of factors exists,
and that the presence of the two members of one or other pair will lead
to the production of a single. The complementary distribution of the
two members of a second pair (X' Y') among- some of the d- and
no-d-strains respectively would lead to an increased production of
singles in F2 as the result of a union between a d- and a no-d-str-Am
which happened to contain the complementary factors, if these factors
are borne by ovules and pollen alike ; whereas unions between two d-,
or between two no-d-strains would be unaffected by the presence of one
only of the second pair of factors. It is in fact difficult to see how
otherwise results such as those obtained in mating 8 (see p. 341) are
to be explained, since it seems hardly possible to suppose that the
discrepancies can be due to a mere chance variation.

E. R. Saunders 339

Details of the several matings.

i. The parents are alike in plastid colour, and are homozygous as
regards the allelomorph concerned {W or w).

Mating 1 . d-cream $ x no-d-cream </*. Not yet carried out.

Mating 7 (reciprocal cross), no-rf-cream $ x ci-cream ^f. Not yet
carried out.

Mating 2. c?-non-cream $ x /u)-d-non-cream <^.

As stated above (p. 336) the expectation in such cases is that of the
Fj plants, some will breed true to singleness, and some will yield both
singles and doubles in F^. The results of several matings of this type
have already been recorded^ More recently another experiment of
this kind has been carried out on a considerably larger scale. The two
sets of results are summarised below :

Parental Types

'd-glabrous red $ x no-d-hoary white (incana) s

'd-glabrou8 flesh ? x ,, ,, >i <?

'^d-glabrous dark purple ? x Tw-d-glabroua white <r

'd-glabrous copper $ x no-d-glabrous flesh i

»d-glabrous dark purple ? x no-d-glabrous light purple i

d-glabroas red ? x no-d-glabrous white i

Totals 31 17 14

Altogether 31 ^i plants were tested; 14 yielded singles and doubles in
F2 while 17 gave only singles, where the expectation would he an almost
corresponding excess the other way, i.e. a proportion of 15 breeding
true : 17 which give doubles^. Experience has shown however in the case of
another character, viz. hoariness and smoothness, that where the expecta-
tion is as near equality as in this case a corresponding excess on the
wrong side is within the range of variation which we may expect.

With regard to the proportion of singles and doubles in the mixed
F2 families it seems probable that the plants used as parents were true
to type in constitution (viz. XYxy and XY XY respectively) and that
the real ratio in every mixed family in F^ was 3 s. : 1 d. If we add

^ Reports to the Evolution Committee.

2 Recorded in Report II, p. 37.

^ On the assumption, i.e. that the gametic series is on a 15 : 1 basis, see p. 322.

"SS-

351

II-
S)

340 Doubleness in Stocks

together the numbers obtained in the 9 mixed families in the last
experiment (see Table V), we get a total of

371 s. and 115 d.,

where a ratio of 3 s. : 1 d. would give

364-5 s. : 121-5 d.

Though in two of the families the proportion of doubles was
distinctly less than 1 in 4, it seems more likely, on the whole, that
these are cases of accidental variation in the direction of deficiency
from the ratio 3 s. : 1 d., than that they represent some considerably
higher ratio such as evidently occurs in certain other matings.
Assuming the constitution given above we may suppose that in the
gametogenesis of ^i an equal number among the pollen and the
ovules carry the combinations XTW and xyW (or XyW or xYW as
the case may be). The gametic types would be simply expressed
thus

Ovules Pollen

XYW XYW
asyW xyW

in the case of an F^ plant derived from an xy ovule, a corresponding
substitution being made in the case of an F^ plant derived from an
Xy or an xY ovule. This being so, we should expect the same result
in F3 from the F^ singles, as was obtained in F^ from the F^ plants ;
and this was the case. Of three F^ sister plants which were self-
fertilised, two gave a mixture in a proportion approximating to
3 s. : 1 d. (viz. 20 s., 9 d. and 5 s., 2 d. respectively) and one gave
a family of 21 all single.

The F^ families composed entirely of singles will have been pro-
duced by those ^1 plants which were derived from the union of XYW
ovules and XYW pollen. In all these families the double character
will have been bred out completely, so that not only will doubles be
absent in them, but they will be wanting in all later generations
derived from such F^ families. Confirmatory evidence of this fact was
obtained both in ^3 and ^4, after which the experiment was brought to
an end. The ^3 generation was raised from 41 self-fertilised F^ plants
belonging to 11 out of the 16 all-single families, and consisted of 898
plants all of which were single. In F4, seven families numbering all
together 77 individuals were again all single.

We may now consider the reciprocal cross.

E. R. Saunders 341

Mating 8. wo-c?-non-creara $ x rf-non-cream ff.

The expectation in this case is that all ^i plants will yield
a mixture of singles and doubles in F^.

A few experiments of this kind are given in the Evolution Reports,
and others have been carried out since (see Table IV). The results in
both cases are summarised below :

B • '^ * r*" *•■:: 3 S m

" § ■sc=~ = -¥:i

Parental Types o^ ^-> Z g •« ""•§

^ no-d-glabrous flesh ? x d-glabrons dark parple i \ 0 \

' no-d-glabroas white ? x d-glabrons copper i 10 1

* ,, I, ? X d-glabrou8 red <r 2 0 2

„ „ ? X „ cf 46 0 46

no-d-hoary white (Brompton) ? x ,, <j 2 0 2

no-d-glabrous flesh ? x d-hoary white (intermediate) jf ' 1 0 1

,, „ ? X d-glabrons light purple (f 7 0 7

Totals 60 0 60

Altogether 60 F^ plants were self-fertilised and all gave a mixture of
singles and doubles in F^. In the great majority of the families the ratio
approximated to 3 s. : 1 d., in fact, in all but the last mating it can
hardly be doubted that this represents the real ratio. If we take the
case where the largest number of F^ plants were bred from (as likely to
furnish the most reliable result), and sum all the families (46) together,
we get a total of 1640 s. and 505 d., where 1609 s. and 536 d. would
have been an exact result. In the case of the mating between no-rf-flesh
and d- light purple however there is a very large deficiency of doubles in
certain families, and it is evident that here some further complication
is present. The question that arises is, whether this result is due to
the particular constitution of one of the two parents, or whether it is
the result of a combination of factors, some of which are brought in by
the one parent, some by the other. This point is not easy to determine ;
on the whole, however, the evidence may be taken to point to the latter
alternative, other factors being concerned besides X and Y in the
manner suggested above (p. 338). The conclusions indicated by the
series of results bearing on the point may be put thus:

1. The no-c?-glabrous flesh individual used as seed parent in the
union with d-light purple did not happen to be employed in any other

1 Recorded in Beport 11, p. 37.
' See note, p. 326.
Joom. of Gen. i 28

342 Doiibleness in Stocks

mating, but several sister plants gave a deficiency of doubles in several
F^ families when used either as seed or pollen parent in matings with
various sulphur-whites — the only other c?-form besides light purple with
which flesh was crossed. Thus from one sister plant, used as the pollen
parent to fertilise a particular sulphur-white, five Fo families were
obtained, composed as follows :

Single

Double

Totals 153 24

Now this particular sulphur-white was used in many matings. It
gave a distinct excess of doubles, as we should expect, both when self-
fertilised (viz. 5 s,, 11 d.) and when crossed with pollen from five other
eversporting forms, the seven ^i families thus produced, comprising alto-
gether 57 s., 95 d. 29 F^ families raised from these latter matings gave
a total of 418 s. and 597 d. We have therefore every reason to suppose
that this particular sulphur- white was producing as large a proportion
of ovules carrying doubleness as the typical eversporting plant. It
therefore looks as though the deficiency observed in F2 from the mating
between flesh and this particular sulphur-white could not be attributed
wholly to the sulphur-white. But if the flesh parent is partly or wholly
responsible for the deficiency in this case, then we shall probably be
right in regarding the flesh parent as similarly responsible in other
matings where sister plants were employed.

2. On the other hand the c?-glabrous light purple individual used
as the pollen parent in the union with no-c?-flesh was also similarly
employed in matings with two no-c?-cream individuals. In both these
matings there was a distinct deficiency of doubles in ^2- Nine F^ cross-
breds in the one case gave 186 s. and 20 d., 5 F^ cross-breds in the other
gave 75 s. and 9 d. Now one of these same cream plants was also
similarly used as the $ parent in a mating with (i-glabrous azure, and
here none of the F2 families showed a deficiency, hence the cause of the
deficiency in the mating between cream and light purple cannot be
ascribed altogether to the cream, but must be due, wholly or in part, to
the light purple individual which was used as the pollen parent in the
mating with flesh as well as with cream.

E. R. Saunders 343

3. These results can be harmonised on the assumption that in addi-
tion to X and T a second pair of factors X'Y' are concerned in the
determination of singles and doubles, as suggested above (p. 338), and
that these factors have a complementary distribution in some of the
d- and no-rf-strains^ Thus the presence of Y' in the /io-<i-strains flesh
and cream, and of X' in d-sulphur-white and c?-light purple but not in
rf-azure (in the case of those individuals used in these experiments),
would mean a higher percentage of singles in F^ than 3 s. : 1 d. after
crossing the flesh or cream with the two d-forms, sulphur-white and
light purple, but not after crossing with azure.

We get confirmatory evidence of the genuine nature of this high
proportion of singles from the results in F^. We should expect from
analogy with simpler cases that Fi would behave diflferently from F^ in
that not all the singles would yield a mixed offspring but that about
one-third would prove to breed true to singleness. Those ^2 singles
which yielded a mixture would presumably give the same proportions
as the ^1 plants. This was found to be the case in the one kind of
mating in which the experiment was carried to F^. In the case of the
mating no-d-glabrous flesh % x c?-glabrous light purple ^T, 14 ^2 singles
were self-fertilised to produce ^3. Disregarding one family of 8 singles
as indecisive we find that among the remaining 13 families 4 were
composed entirely of singles, and 9 included a mixture of singles and
doubles; in two cases a proportion of about 3 s. : 1 d. was recorded, in
the other 7 the proportion of singles was considerably higher.

ii. The parents are of unlike plastid colour ; each is homozygous in
the allelomorph concerned ( TT or lu).

In these cases we have to consider not only the total number of
singles and doubles obtained, but also the proportion of each form
having white and cream plastids respectively.

^ A somewhat similar case in which the recessive form was found to occur in an
extremely small proportion in F*, owing to the presence of several factors in Fi, any one
of which alone sufficed to produce the dominant form, has already been investigated and
fully described by Nilsson-Ehle. This observer finds that if two wheats are crossed
together one having red grains and the other white, plants with white grains only occur in
Fn in the proportion of 1 in 64. This, he explains, is due to the existence in the red
wheat of three factors (i?i , R-i, R^, the presence of any one of which will suffice to render
the grain red. Hence only those F^ plants in which all three factors are absent will have
white grains, and these will only occur in the proportion of 1 in 64. (See Nilsson-Ehle,
Kreuzungtuntersuchungen an Hafer und Weizen, Lund, 1909.) The Stocks appear to offer
a parallel but more complex case, as in this instance pairs of factors instead of single
factors are concerned.

23—2

344 Douhleness in Stocks

Mating 3. d-cream % x wo-c?-non-cream ^.

Summary of results. For details see Table V.

Only one kind of mating of this type was made, viz. cZ-glabrous
cream $ x wo-c?-glabrous white (/•. The ci-cream plant il/ as $ was
crossed with the wo-d-white plant /as ^, and two jP, descendants of the
c?-cream plant iT as $ were crossed with an F^ descendant of plant
/ as j/* (see pp. 814 — 317 and Table II). Now plant K, it will be
recalled, was a sporting cream which gave excess of doubles, but
which was under suspicion of being a cross-bred cream since some of
its offspring appeared to breed true to singleness. At present therefore
we must accept the results obtained from this plant with some
reserve until they have been confirmed with material that is beyond
question.

In accordance with expectation some ^i families gave a mixture of
singles and doubles, others bred true to singleness. In the mixed F^
families the proportion of singles and doubles was evidently 3 s. : 1 d.
with one doubtful exception ; the bulk of the plants were singles with
white plastids and doubles with cream plastids, the former being in
excess, but a small proportion of singles with cream plastids, and
doubles with white plastids occurred in some cases. The F.2 families
containing only singles were composed almost exclusively of non-creams,
only 7 individuals in a total of 420 having cream plastids ; these 7
occurred in 2 F^ families derived from the suspected cream.

Altogether 26 F^ plants were self- fertilised to produce F2. 14 of
the F^ families were mixed, 12 were composed only of singles. Although
two or three of these latter families are too small to put their purity
beyond doubt, and may therefore possibly cause the percentage of
all-single families to appear slightly higher than it actually is, the
result as it stands cannot be far from the truth, and agrees well with
the expectation of 7 -f- « breeding true to 9 — a; giving both singles and
doubles.

In the case of the mixed F^ families, the ^1 parent was presumably
derived from the union of

xyw
or Xyw
or xYw ovules with XYW pollen

and a scheme of gametogenesis which would give the observed result,
where all four forms occurred in F^, might be imagined thus in the

E. R. Saunders 345

first case, the appropriate substitutions being made for the alternative
cases:

Ovules

PoUen

n-1 XYW

1 xyW

1 an/W

1 XYvf

1 XYw

n — 1 acyw

n — 1 xyw

wJiere the distribution of XY, xy, W and w is the same among ovules and
pollen, hut where partial coupling between the plastid colour factor and
the factors for singleness and doubleness occurs in such a way, that the
two most frequently occurring terms in the series represent the combina-
tions received from the parents, the two rarer terms the recombinations of
these factors.

The conception that heterozygotes containing the same components,
but having received these components combined in different ways, may
form different gametic series has recently been put forward by Bateson
and Punnett^ in explanation of certain facts observed in the course of
their experiments with the Sweet Pea. In certain cases where two
separate pairs of allelomorphs are concerned, and where particular
combinations occurred in the gametic series with greater frequency
than others, they found that the results obtained would be explained
if it is assumed that a heterozygote of composition AaBb, which has
been built up of the combinations AB and ab, again forms chiefly the
gametes AB and ah, only comparatively few in the series being Ab or
aB in composition. Whereas in an AaBb heterozygote, which has
received A from one parent and B from the other, Ab and aB are the
more frequent, AB and ab the less frequent terms in the series. In
Stocks a parallel case may be found in the type of union now under
consideration, viz. those in which the parents are of unlike plastid
colour, each being homozygous in the allelomorph concerned ( TT or w),
and in which the one is an eversporting, the other a true-breeding single.
Since two factors X, Y (or X'Y') are required to produce singleness, we
are here concerned altogether with three factors, viz. X, Y and W, but
X and Y being linked together in the true-breeding single, the two
behave as a single allelomorph. When an XYxyWw heterozygote has
received these components in the combinations XYW and xyw (as in

» Proc. Roy. Soc. Series B, Vol 84, p. 3, 1911.

346 Doiihleness in Stocks

the case ?io-c?-non-cream x d-cream), XYW and xyiv are the more
frequent, XYw and xyW the rarer terms in the gametic series. If,
on the other hand, the XYxyWw heterozygote has been built up
from XYw and xyW (as in the case no-c?-cream x d-non-cream),
then XYv) and xyW gametes are chiefly formed, those of XYW
and xyw composition being comparatively rare. The same scheme
of coupling, as already shown, holds in regard to the female germs
when the eversporting single is se^-fertilised, but here the symmetry
of the gametic series is disturbed by the fact that the male germs are
unable to carry either of the dominant factors X or Y. Every ever-
sporting single is an XxYy heterozygote and is built up from the
combinations XY and xy. In gametogenesis XY and xy ovules are
chiefly formed, only comparatively few, we may conclude, are Xy and
xY in composition, though direct proof in this case is not as yet possible.
For since all doubles are sterile we cannot apply the breeding test, and
at present therefore we are unable to demonstrate differences of com-
position between the doubles derived from Xy, xY and xy ovules
respectively.

We may now consider the results of the present experiment in
detail. If we accept the results as they stand, with the reserve
mentioned above, and compare the totals obtained from the 14 mixed
families with the results which would follow from gametogenesis on
the lines suggested above, we get :

Observed result

Singles

with white

plastids

457

Singles

with cream

plastids

Doubles

with white

plastids

Doubles

with cream

plastids

140

Calculated result if n = 16 .

442

18-5

135

„ if 71 = 32 .

451

9-5

144

This latter result agrees very closely with that obtained experi-
mentally. If we take the recorded result to be an average sample of F^,
it appears that in F^ n may have the value 32, whereas in the eversporting
parent, as previously stated, n probably = 16. In several of the smaller
families the two rarer forms were not recorded, but, on the whole, it
seems probable that their absence in these cases is accidental and is to
be accounted for by the small size of the F.i family. For the largest
family in which these two forms were absent numbered only 32, and on
the present supposition the expectation for both forms is less than 1 in
64, and even with the lower value for n would not be quite as high as
1 in 32. It may be worth while to note that the occurrence of even

E. H Saunders 347

larger F, families composed entirely of singles with white plastids and
doubles with cream plastids would present no serious diflBculty on the
present view. Segregation in these cases might be in accordance with
some higher term in the series. For with each successively higher
value for n the diflference in the proportion of the two more firequent
forms (singles with white plastids and doubles with cream plastids)
would be so slight, that it would not be practically appreciable in
experiments on the present scale ; whereas the proportion of the two
rarer terms (doubles with white plastids and singles with cream plastids)
would be successively reduced by about one-half, and hence in small
families it would be unlikely that they would be recorded. If the value
for n were suflficiently high the coupling would appear to be complete,
and these two forms would then appear to be altogether wanting. Or
again a like result would follow if the no-d parent were by chance a
form in which W was linked with XF in some of the germs (see above
p. 335). In this case we should expect the Fi plants derived from
the XYW germ cells to give F^ families composed of only the two
forms — singles with white plastids and doubles with cream plastids —
while those derived from XYW gametes would yield the F, families
which include all four types.

The all-single F.^ families will be produced by the F^ plants derived
from the XYw ovules. These cross-breds will naturally breed true to
singleness. In all these F.^ families we find either absence, or a marked
deficiency of individuals with cream plastids. This deficiency recalls
a similar result obtained in an earlier experiment where the parents,
both in this case eversporting, were of unlike plastid colour, c?-cream
being used as $, c?-non-cream as </. In this latter case XYw ovules
were also fertilised by pollen carrying W, and here a single with cream
plastids was obtained in F^ when a descendant of plant K was used,
but not when other cream individuals were employed. We may
suppose that the distribution of W among the ^i pollen grains will
be the same in both grosses (see p. 332, where this point is discussed).
In the present case, among 10 families which included a total of
279 individuals none had cream plastids; in the two remaining families
7 plants with cream plastids were recorded in a total of 141, but these
7 are not beyond suspicion, since some doubt exists as to the genuine
eversporting nature of the cream parent (a descendant of plant K).

To sum up ; we find that the union ci-cream % x no-rf-nou-cream ^
gave, in accordance with expectation, some all-single-femilies, and some

348 Doubleness m Stocks

mixed families with a proportion of 3 s. : 1 d. We may suppose (1) that
in ^2 families containing doubles all the four possible forms will occur
if in the non-double-throwing (^ parent W is not coupled with XY,
but that only the two parental forms will occur if W is linked with XY;
(2) that the small number of singles with cream plastids and doubles
with white plastids is due to partial coupling in the i^i gametes such
that W and w occur much more frequently in combination with the
factors with which they are associated in the parents than in other
combinations: (3) that in Fj cross-breds producing the all-single F2
families, all or most of the pollen must carry W, as has already been
seen to be the case when c?-non-cream is used as the </" parent instead
of wo-d-non-cream.

Mating 9. no-d-non-cream $ x cZ-cream (/".

This reciprocal cross was made with the same two strains.

Summary of results. For details see Table IV.

A true-breeding single glabrous white was crossed with pollen from
a double-throwing glabrous cream (plant H, see p. 314). Only four of the
Fi plants were tested ; each in accordance with expectation gave a mixture
of singles and doubles in F^. All four forms occurred, though all were
not recorded in each family. Singles with white plastids and doubles
with cream plastids were obtained from each of the ^1 cross-breds, the
former being in excess, but the other two forms were present in such
small numbers that larger sowings would be necessary to determine
whether their absence in the families in which they were not recorded
was real, or not. Quite possibly it is merely accidental. In two of the
families the total number of singles and doubles approximated to 3 s.: Id.;
in the other two the proportion of doubles was less, but the deficiency
is not so great (about 1 d. : 6 s.) as to render it beyond doubt that it is
to be regarded as genuine.

In this form of mating Fi is presumably produced by the union
oi XYW ovules with xyw pollen. If gametogenesis follows the same
course as in the reciprocal mating where the union is between xyw
ovules and XYW pollen (see above, p. 344), then, taking the two families
in the present case where experiment gave the expected proportion of
3s. : Id., we should expect a total of

Singles Singles Doubles Doubles

with white with cream with white with cream

plastids plastids plastids plastids

117 2-5 (nearly) 2-5 (nearly) 37

where 122 3 1 33

E. R Saunders 349

were actually observed. The agreement between the observed and
calculated results is so close that we may conclude that the same
gametic series is formed by the reciprocal cross-breds {xyw % x XYW ^)
of the present mating and {XYW % x xyw </) of mating 3. If n has
the same value in each of the F^ plants, and if in the present mating
the non-double-throwing parent is homozygous in XYW, we must
suppose that the absence of the two rarer forms in the two families
mentioned above is a chance variation.

Mating 4. c?-non-cream ? x no-d-cxeaxo. ^, Not yet carried out.

Mating lO. Reciprocal cross, no-d-cream % x d-non-cream ^.

Summary of results. For details see Table IV.

Five kinds of matings of this type were carried out, viz.,

7io-(i-glabrous cream % x d-hoary white (intermediate) (/".

„ „ ? X c?-glabrous white ^.

„ „ $ X d-glabrous red ^.

„ „ $ X c?-glabrous azure ^f.

„ „ $ X c?-glabrous light purple f^.

72 .F, families were raised, all of which included some doubles, the
proportion varying from 3 s. : 1 d. to a very much higher proportion of
singles. A point of special interest in this group of matings is that no
doubles with cream plastids were recorded in an F^ generation numbering
more than 3000. That is to say, in no case in which a non-double-
throwing cream has been employed in a mating with an eversporting
non-cream form has it yet been found possible to obtain the com-
bination of creamness with doubleness in F^, though a considerable
number of the single F^ plants have cream plastids. The total numbers
obtained were :

1666 singles with white plastids

773 doubles „ „ „

790 singles „ cream „

or about twice as many of the form with both dominant characters as
of either of those exhibiting one dominant and one recessive character.
In matings of this type F^ is presumably derived from the union
o{ XYW ovules with xyW pollen. Now if the two kinds of germ cells
which united to produce F^ were formed again by F^ without any
redistribution of the factors for plastid colour and for singleness and

^ See note, p. 326.

350 Douhleness in Stocks

doubleness, the result would be entirely in agreement qualitatively
with that actually observed, and not very different from it quantitatively.
For where

8W.

dw.

sc.

1615

807

1666

773

790

would have been an exact result,
were actually observed.

But it seems doubtful whether the case is in reality quite so simple
as this, for the proportion of the three forms occurring in the different
F^ families was not as uniform as on the above scheme we should expect
it to be, the deficiency of plants with cream plastids in some cases being
too great to be reasonably regarded as an accidental variation from the
ratio 3 white : 1 cream. On the supposition that the repulsion in F^
between XFand W is complete for both kinds of germ cells, we get

Ovules Pollen

XYw XYw

xyW ccyW

It follows that all the F^ singles with white plastids should be
heterozygous both as regards singleness and doubleness and also as
regards plastid colour. Out of 108 such F^ singles derived as follows :

70 from the mating between no-d-cveBxa % x c?-light purple ^
18 „ „ $ X c?-white (/*

20 „ „ $ X cZ-red ^

which were tested, 90 (viz. 61, 15 and 14 from the three matings
respectively) have already proved to be heterozygous as regards
singleness and doubleness. In the other 18 families doubles were
not recorded, but the numbers, none of which exceeded 11, are too
small to be regarded as decisively indicating that the F.2 parent was
unable to produce doubles. So far, then, the evidence presents no
difficulty in the way of the above supposition. But only 79 of these
families (viz. 55, 12 and 12 from the three matings respectively)
included plants with cream plastids ; and although the numbers in
the other 29 families in which they were lacking were mostly too small
to be conclusive, the fact that in one case as many as 40 singles were
recorded, all with white plastids, leaves it doubtful whether some of
the Fi plants may not be breeding true to whiteness, and consequently,
whether some F^ gametes may not be carrying the combination XYW.
If any such are formed they must evidently however constitute only

E. R Saunders 351

a small proportion of the whole number of gametes. On the supposi-
tion that the repulsion is only partial we might expect recombination
thus:

Ovulea Pollen

n — I X Yw n — lX Yw

1 XYW 1 XYW

1 xyw 1 xyw

n — I xyW n — 1 xyW

This would give a result in F2 almost precisely similar to that
produced by complete repulsion, except that there would be in addition
to the three forms given above a proportion of rather less than 1 in
1000 of doubles with cream plastids, if n = 16 ; or rather more than
1 in 4000 if n = 32. The difficulty of distinguishing between these
two possibilities will be apparent from an examination of the figures
given below, where the composition of the resulting F^ generations is
compared in detail in the two cases.

Coupling

partial

>i=16

Coupling
complete

Coupling
partial
n=32

Conpling
complete

Singles, plastids white

513

512

2049

2048

Singles, plastids cream

255

256

1023

1024

Doubles, plastids white

255

256

1023

1024

Doubles, plastids cream

Totals ... 1024 1024 4096 4096

As between complete repulsion on the one hand, or partial repulsion
on either a 15 : 1 or a 31 : 1 basis on the other, the evidence therefore
is not absolutely clear. If the former assumption (coupling complete)
should prove correct, then, as stated above, all F^ singles with white
plastids should prove heterozygous in singleness and doubleness and
also in plastid character; while all the singles with cream plastids
should be homozygous in both characters : further, the observed absence
of doubles with cream plastids will be absolute. If on the other hand
the repulsion is partial, then certain of the F2 singles with white plastids
will breed true both to singleness and to whiteness, others to singleness
though not to whiteness, others to whiteness though not to singleness,
while others again will be heterozygous as regards both characters:
similarly some of the singles with cream plastids will prove to be
breeding pure to singleness, others not. In this case it most be
supposed that with a larger count in F, an occasional double with
cream plastids would appear. The available evidence from the F,

352 Douhleness in Stocks

generation leaves the question still undecided. 38 F.^ singles with
cream plastids were tested, 33 from the mating with d-\\g\\t purple
and 5 from that with c^-red ; none yielded doubles in ^3. So far as
it goes this fact is against the view that the repulsion is only partial,
but again it is doubtful whether the experiment is on a sufficient scale
for the result to be regarded as conclusive.

iii. One parent is homozygous and the other heterozygous in
regard to plastid colour.

Mating 5. c?-sulphur- white % x no-d-cxeaxa ^ . Not yet carried
toF,.

Mating 1 1. Reciprocal cross, wo-c^-cream $ x c?-sulphur- white </•.

46 F^ families were raised, and doubles were obtained in all but one.
The probability that this all-single family was not derived from a cross-
bred has already been discussed (see p. 310).

As we should expect, the F^ generation all have cream plastids, for
jP, had cream plastids, being derived presumably from the union of
XYw ovules with xyw pollen. The proportion of singles and doubles
in F2. approximates in many cases to the ratio 3 s. : 1 d. In those
families in which the doubles amount to more than 1 in 4 it is doubtful
whether the excess observed is real, but among some at least of those
in which the proportion is less than 1 in 4 the deficiency is probably
genuine (see Table IV). An explanation of these cases has already
been suggested (see p. 338), the supposition being (as in the case of
Mating 10) that we are here dealing with the additional pair of factors
X'Y, X' occurring in the one parent and Y in the other, the union of
the two producing a higher percentage of singles than is the case where
X and Y alone are concerned. We shall therefore express the com-
position of the germ cells uniting to produce F^ more fully thus

XYX'w%xxyYw ^.

Mating 6. c^-sulphur-white $ x wo-ci-non-cream j/.
Four kinds of unions of this type were made, viz.

d-sulphur-white % x wo-rf-hoary white {incana) ^.
„ $ x no-d-hoa,ry red (Brompton) j/*..

„ $ X wo-c?-glabrous white (/*.

„ $ X no-c?-glabrous flesh </".

We may suppose that in this class of unions there will be at least
four different types of plants in Fj, and therefore that there will be

E. R. Saunders 353

diflferent types of families in F,. We may state the expectation in
general terms thus:

(a) Nearly half the F^ plants will be derived from the union

of XYW ovules with XYW pollen ; these will give only singles in F^,
all with white plastids.

(b) A small percentage of the F^ plants will be derived from the

union of XyW ovules with XYW pollen, and these should give a
mixture of singles and doubles both with white plastids.

(c) Half of the ^i plants will be derived from the union oi xYw or
xyw ovules with XYW pollen, and these may be expected to give all
four forms in ^2 (ie. singles and doubles with white plastids and singles
and doubles with cream plastids).

The results obtained may be summarised thus (for details see
Table V).

A total of 128 F^ families were raised, composed as follows :

(1) 65 families were composed entirely of singles with white
plastids. Of these no doubt some should be disregarded on account
of the small number of plants recorded. Leaving out of account all
families of less than 10 individuals there remain 46, representing
a total of 1303 individuals all single and all with white plastids (see
paragraph (a) above).

(2) 63 families included a mixture of singles and doubles.

(a) Two of these contained only plants with white plastids, but in
both the numbers were small and included only one double, so that
although they may represent the type of family given above under
(6) the evidence is insufficient for proof.

(y3) 11 families included all four forms, singles with white plastids
and doubles with cream plastids being largely in excess.

(7) Of the remaining 50 double-containing families there were
22 in which singles with cream plastids and 2 in which doubles with
white plastids were absent (the other three forms being present in each
case); and 26 in which singles with white plastids and doubles with
cream plastids only were present.

If we apply the same reasoning here as in the case of Mating 3,
where d-cream was used instead of d-sulphur-white (see p. 344), we
shall conclude that many (? all) of these 50 families would yield the
missing forms if a larger sowing were made. For here as in the earlier
case tbe great majority of the double -throwing cross-breds will be de-
rived from the mating of xyw ovules and X YW pollen ; recombination

354 Douhleness in Stocks

together with partial coupling of the factors for singleness and douhle-
ness and for plastid colour, giving, as previously stated (p. 346), both the
rarer forms in the proportion of only about 1 in 33 if w = 16, or 1 in 65
if w = 32, may well account for their apparent absence in a large number
of families in the present experiment. There seems in fact no reason
to doubt that, so far as the factors X, F, W aye concerned, the same
relation holds in both matings. But in the present case it seems
probable that in two of the unions another pair of factors come into
play, one member of this pair being present in the sulphur-white, the
other in ?io-rf-flesh, and also apparently in no-d-^hxte, but not probably
in either incana or the Brompton strain. The effect of the presence
of these additional factors in any Fi cross-bred will be to raise the pro-
portion of singles in the F2 family derived from this cross-bred as
described in Mating 11. Only in this way does it seem possible to
explain the frequent high percentage of singles in the F2 families
where the two ten-week strains were employed, when other forms
such as the two biennials gave the expected 3 s. : 1 d. The genuine-
ness of these results is confirmed by those obtained in the reciprocal
union.

Mating 12. Reciprocal cross. no-c?-non-cream $ x tZ-sulphur-
white c^. (For details see Table IV.)

Two kinds of mating of this form were carried out, wo-c^-hoary
Brompton white being used as the seed parent in the one case,
wo-cZ-glabrous flesh in the other. All four forms appeared in F^,
singles with white plastids and doubles with cream plastids greatly
preponderating; in many families in fact only these two forms were
recorded. The proportion of singles and doubles in the large F^ families
derived from the Brompton white was evidently 3 s. : 1 d. The families
derived from the no-d-flesh are mostly of small size. In some no doubt
the ratio is also 3 s. : 1 d. but in others there appears to be a distinct
excess of singles as in the reciprocal cross (see above). Altogether
48 F2 families were raised and doubles were recorded in 46 ; the larger
of the two remaining families consisted of 17 singles, the smaller of only
8 ; both were derived from the no-d-fiesh parent. As there is every
feason to suppose that every Fj derived from the union no-d % y. d (^
will produce doubles, and as the proportion of doubles in one or two
sister families is even less than 1 in 17, we may reasonably conclude
that a larger sowing would have given the expected mixture in these
cases also.

E. R Saunders 355

Summary of results of cross-breeding.

If we now put together the whole body of evidence obtained from
unions between true-breeding and eversporting forms we may summarise
the results as follows :

From those matings in which the eversporting form was used as
? 185 F^ families were raised ; 91 showed a mixture of singles and
doubles, 94 were composed of singles only. This latter total no doubt
appears larger than it is in reality through the fact that some families
are probably included in it, which, if a larger sowing had been made,
would have been found to contain some doubles. As however it is not
possible to tell exactly how many of these smaller all-single families
should be disregarded, the totals are given as they stand. But we may
take the results as fully establishing the fact that when the eversporting
form is used as $ in matings with a true-breeding strain, some F^
families will be mixed and some all single ; and that the proportion of
the all-single to the mixed will be the same as the proportion of single-
to double-carrying ovules in the % parent, viz. 1 -\-x single : 9 — a;
double where x is less than 1.

From the reciprocal form of mating 230 F^ families were obtained
and doubles were recorded in 227. In two at least of these exceptions
the evidence in regard to the seeming absence of doubles cannot be
regarded as conclusive, and it may be that in the remaining case the
same explanation also holds good ; or, it may be that this family did not
arise from a cross at all, but was the result of accidental self-fertilisation
which in this case would not betray itself in F^.

When one of the parents in these unions is homozygous in W
(plastids white) and the other in w (plastids cream) the proportion of
F^ singles and doubles having white and cream plastids respectively
indicates that in almost all the ^i gametes, whether pollen or ovules,
each of the allelomorphs W and w is associated chiefly with the particular
combination of factors for singleness and doubleness with which it was
combined in the ovule or pollen grain used to produce F^. Thus in
a mating between no-d-white and d-cream, W is borne for the most
part by those ^i gametes carrying XY,w by those carrying xy. Con-
versely when the mating is between c?-white and no-d-cxe&va it is IT
which is carried almost entirely by the XY gametes, w by those of
xy composition.

356 Douhleness in Stocks

VIII. Summary.

In the preceding pages an attempt has been made to work out
a scheme which will account for the behaviour, so far as we know it
'at present, of various races of Stocks in regard to the two characters,
colour of plastids and production of doubles. The results have already
shown that the relationship of the various factors concerned is by no
means simple, but it is not unlikely that as more facts come to light
still further complications will become apparent. It may be claimed
however that the scheme as it stands affords a useful working hypothesis
enabling us to grasp a complicated series of facts; moreover it is one
which can be tested in detail by further experiment on definite lines.
A solution which enables us to fit together so many pieces of the
puzzle must, one cannot but believe, prove to be substantially correct.

We may therefore venture to add to the conclusions already
formulated on pp. 321 — 324 the following general statements :

(1) All sap-coloured races of Ten Week Stocks so far investigated
(i.e. azure, light purple, dark purple, marine blue, flesh, copper, red) and
the two non-sap-coloured forms pure white and cream can occur under
two forms, a pure-breeding form and an eversporting form.

(2) The sulphur-white race — a race which is peculiar in being
eversporting in regard to plastid colour as well as in regard to douhle-
ness— is only known in the double-throwing form. It produces single
whites, double creams and a small percentage of double whites.

(3) Every individual in an eversporting strain yields doubles in
excess ; the proportion may be stated as 7 -I- a; single to 9 — a; double
where a; is less than 1.

(4) All the pollen grains of such strains appear to carry douhleness:
that is to say, in these strains the distribution of the factors for single-
ness (X and Y) is limited to the gametes of one sex.

(5) The ovules in every individual belonging to these strains are
mixed, the proportion of those carrying singleness and douhleness is
presumably the same as the proportion of singles and doubles among
the offspring, since the pollen is uniform.

(6) The proportion of 7 + a; single to 9 — a; double is most easily
explained on the supposition (1) that two factors at least are con-
cerned {X and F), (2) that the zygote is heterozygous in regard to
both, and (8) that in the case of the ovules these factors show partial
coupling of the kind with which we are already familiar in the Sweet

R R. Saunders 357

Pea*. Though, however, the scheme of coupling is based on the same
principle in the two cases, there is in the Stocks an additional com-
plexity owing to the limitation of the power of carrying these factors
to the gametes of the female sex.

(6 a) The scheme of coupling is such that the combinations of the
allelomorphs XxYy carried by almost all the ovules in an eversporting
individual are the combinations borne respectively by the male and
female germs which united to produce that individual ; the rarer terms
in the series are those which represent recombinations of these factors,
one factor in the recombination being derived from the male parent,
the other from the female. The number of gametes required to exhibit
the whole series being taken as 2n, we may represent the gametic series
in the eversporting forms in general terms thus :

Ovules Pollen

n-1 XY aU xy

1 Xy
1 xY
n — 1 xy

where XY represent factors, required for singleness, and where the
zygote has arisen from the union of an XY ovule with an xy pollen
grain. The value for n is probably 15 for the type forms though
in some cross-breds it may be 31 (or possibly some higher term in
the series).

(6 h) The above formula holds good for all the eversporting strains
investigated, hence when no further complications arise the proportion
of doubles remains the same whether these strains are self-fertilised
or inter-crossed. Thus we are able to understand how it is that these
eversporting strains produce a constant excess of the recessive (double)
form. The only other instance of the kind, at present known, that
seems in any way comparable, is that of one of de Vries' Oenothera
hybrids — a tall form which gave an excess of dwarfs ^

(7) Singleness in the pure-bred, non-double-throwing single is due
to the presence of the same two factors {X and F), but in these strains
these two factors are linked together {XY), so that when this type of
single is crossed with an eversporting form recombinations of the two
pairs of allelomorph do not occur.

(8) In addition to the pair of factors referred to under (6) and (7),
1 See note, p. 322. « Ber. der Deui. Bot. GeseU. Bd xxn. a, 1908, p. 667.

Joom. of Gen. i 24

358 Doubleness in Stocks

which are present in all the strains, there appears to be a second pair
of factors (X'Y'), the presence of which also renders the zygote single.
One member of this second pair appears to occur in some but not all
of the pure single strains, the other in some but not all of the double-
throwing strains.

(9) The effect of the coupling mentioned under (7) is that when
only the pair of factors common to all the strains is present {XY), the
mixed F^ families from a cross between the non-double-throwing
pure-bred and the eversporting single contain a proportion of about
3 s. : 1 d.

(10) When a similar cross is made between forms which contain in
addition one member of the second pair of factors mentioned under (8),
the distribution of the members of the second pair being complementary,
some of the mixed F^ families again show a proportion of about 3 s. : 1 d.,
but in others the proportion of singles is considerably higher.

(11) White plastids result from the presence of a factor {W),
cream plastids from absence of the same factor {w).

(12) Pure white or cream races are homozygous in W and w
respectively, but the sulphur-white race is heterozygous in regard to
this factor which is present in some only of the ovules and absent
altogether from the pollen ; moreover in this latter race W appears to
be coupled with one of the factors required for singleness. W^e may
represent the gametic series in this sulphur-white race thus:

Ovules Pollen

n — 1 XYW all xyw

1 ZyF

I xYw
n — 1 xyw

(13) The distribution of the allelomorphs W and w among the
gametes of F-^, where the parents are of unlike plastid colour, appears
to depend upon the conditions under which the plastid colour factor is
introduced into the cross, ie., whether by the male or the female germ,
and whether in combination with singleness or doubleness.

(a) When the union is between two eversporting forms, and when
W is introduced on the female side in combination with XY {XYW
ovules) and w on the male side with xy {xyw pollen) as in the cross
d-white $ X f^cream ^, all the F^ pollen appears to carry creamness
{w) as well as doubleness {xy) like the pollen of the ^ parent : and

E. R. Saunders 359

all the Fi ovules carrying singleness (XY) appear to carry whiteness
(W) like the ovules from which F^ is itself derived; while of those
F^ ovules which carry doubleness all (or almost all) lack W. If it
should be confirmed that only the two parental forms (singles with
white plastids and doubles with cream plastids) occur in F^, then
"all" will presumably be correct in each of the above cases; but
if, as analogy with other cases suggests, the factor W shows not
complete but partial coupling, of the same nature as that described
under (6 a) for the factors X and Y, then we may expect that of the
F^ ovules carrying doubleness almost but not quite all will lack W, and
that in a large sowing in F^ a few doubles with white plastids will
occur. In the event of this latter alternative proving true we should
be able to synthesise the sulphur-white form afresh from true-breeding
whites and creams. For the F^ single from d-non-cream x rf-cream is
formed from the union of the same combinations of factors as was ^i,
and will presumably therefore repeat the same gametic series. Hence
if the appropriate single white be selected in F, it may be expected to
behave like a pure-bred sulphur-white.

(6) In the reciprocal cross where single F^ plants are derived from
the union XYw ovules and xyW pollen a corresponding but reversed
distribution of the plastid colour factor explains the observed results.
Here W is introduced into the pedigree on the male side and is
evidently borne by all (or almost all) the germ cells of one sex — no
doubt the male — in Fi. Since the presence of the dominant allelo-
morph in all or nearly all the germ cells of one sex produces a constant
or almost constant result in F^ whatever the distribution of this factor
among the germ cells of the other sex, we are unable merely from the
F2 result to infer the distribution of W among the ovules. But we
may suppose from analogy that all (probably) of the ovules carrying
creamness will carry singleness, and that nearly all those carrying
doubleness will carry whiteness.

(c) When crossing occurs between two forms of unlike plastid
colour, one of which is an eversporting, the other a non-double-
throwing single, the distribution of the allelomorphs W and w appears
to be different in the Fi singles which are heterozygous in regard to
singleness from that in the singles which are homozygous in this
respect. In the heterozygous singles, which here have the constitution

XYxyWw, recombination of the four components XY, xy, W, and w
occurs in the same manner as described above under (6 a) for the four

24—2

360 Douhleness in Stocks

separate allelomorphs X, Y, x and y, but with this difference, that in
this case the same gametic series occurs among both male and female
gametes. That is to say the majority of both ovules and pollen in F^
exhibit the combinations occurring in the ovule and the pollen grain
which united to produce F^ ; the rarer terms in the series are repre-
sented by the recombinations of the two sets of factors. Thus when
the mating is between single white from a no-d-sXroiu of the form XYW
and double cream {xyw) we shall represent the gametic series in F^ thus :

Ovules

Pollen

n -

-1 XYW

n-1 XYW

1 XYw

1 xyW

11-

- 1 xyw

n—1 xyw

When on the other hand single cream and double white of the
form (XYw) and (xyW) unite to produce Fi the gametic series will

be as follows:

Ovules Pollen

n-1 XYw

n—1 XYw

1 XYW

1 xyw

n — 1 xyW

In the first case F^ is composed chiefly of singles with white and
doubles with cream plastids, the other two forms, doubles with white
and singles with cream plastids, being scarce.

In the second case, on the other hand, doubles with cream plastids,
if they occur, must be extremely rare (none have as yet been recorded,
though possibly they would occur in a larger sowing); doubles with
white plastids and singles with cream plastids are fairly numerous, and
singles with white plastids abundant. In the sister Fi singles which are
homozygovs in regard to singleness {XYXYWw) the distribution of
the plastid colour factor must be such that all or nearly all the gametes
of one sex carry W. Possibly the plastid colour factor W is here
associated with the gametes of one or other sex according as it is
brought in on the male or female side, as described above in the case
where both parents are eversporting in regard to douhleness (see 13 a
and 6).

E. R. Saunders 361

APPENDIX

Note 1. On the relative viability of seeds giving rise
TO singles and doubles

The belief that in Stocks a larger proportiou of doubles can be
obtained from old seed than from seed recently harvested is one of
long-standing, but it appears to be rather of the nature of a tradition
than of an opinion founded on a knowledge of definite facts. Discussing
this point in his paper entitled " Beitrage zur Kenntniss gefuUte
Bluthen" Goebel^ refers to a treatise by F. A. H. Thiele which shows
that the above view was current at the beginning of the last century.
Thiele-, as Goebel tells us, was seeking an answer to the questions how
can one obtain Stock seed which will produce a high proportion of
doubles and how can one recognise this seed ? a propos to the former
inquiry he mentions among other traditions current at that time the
view that the older the seed the more doubles does it yield. Com-
menting on this statement Goebel adds that though unsupported by
experiment it may very well prove to be the case, and might be
explained on the supposition that in course of time more and more
seeds which would have produced singles lose their power of
germinating.

Chat^, a French horticulturist, in a treatise on the cultivation of
Stocks, the culture of which had been carried on in his family for more
than 50 years, expresses himself on this point as follows : Experiment
has shown that seeds two years old give more doubles than seeds one
year old. In proportion as the seeds get old their power of doubling
increases, whilst the power of germinating diminishes*. Although
Chat^ here states that his view is based on experiment, it seems clear
that he did not recognise the possibility that the character of the flower
may already be pre-determined in the seed, and that his suggestion
that a seed which would have produced a double may, if kept,
eventually give rise to a single, is not a true explanation of the facts.

^ Pringtheinu Jahrbuch, Band xvu. p. 285, 1886.

* Prediger tu Pittenoitz bei Kyritz in Pommtm, Coslin, 1825.

* Culture pratique des GiroJUes. Paris, N.D.

362 Douhleness in Stocks

The present series of experiments has made it abundantly clear that
singleness or doubleness in the flower is a character which is dependent
solely upon the constitution of the germ cells from which it arose ; that
it is in fact already determined in the seed, and is entirely independent
of external conditions. Though the present results have disproved the
interpretation, they have confirmed the fact observed by Thiele, Chate
and others that the proportion of doubles obtained from old seed is
often higher than that given by seed more recently harvested. They
have also furnished incidentally a certain amount of evidence as to
the relative viability of the seeds giving rise to singles and doubles
respectively, and also as to the possibility of identifying those which
give doubles.

(a) Viability.

Although critical experiments on a very large scale, and specially
designed to this end, would be required to show the relative rates at
which the progressive loss in germinating power occurs among seeds
destined to give rise to singles and doubles respectively, the evidence
already available points strongly to the conclusion that in any lot of
seed which has been kept until the bulk of it is no longer capable of
germination the surviving remnant will be mostly if not exclusively
composed of those yielding doubles^ The effect of this greater viability

^ It is perhaps hardly necessary to state that the time during which the seeds retain
their vitality varies greatly with the quality of the seed. Cent, per cent, germination
was obtained in some cases after the lapse of three years ; even after seven years — the
longest period over which sample sowings from any one lot of seed were extended — a few
still retained their vitality. On the other hand badly ripened seed sometimes failed
altogether to germinate after three, or even two years.

In considering the results of repeated sowings made after a considerable lapse of time
the following point must be borne in mind. Where only a small number of seedlings
are obtained, it may be that all or most of the seeds in one or two better ripened pods
have retained their vitality while those belonging to all the other pods have died ; or on
the other hand it may be that only a seed here and there in each of several pods has
survived. In the former case the original ratio of single to double will presumably
be unchanged ; only in the latter case are we concerned with the question of a differential
death-rate. Unless the seed is unmistakeably of uniform quality, it is therefore desirable,
in experiments designed to test this point, that the seed of individual fruits should be
sown separately, though this method of procedure necessarily entails much waste of
space when very few of the seeds still survive. When the further fact is taken into
account that single fruits are often found not to afford average samples it will be seen
that any comparative experiment will be of little value unless carried out on a very
considerable scale. Furthermore only those cases should be taken into account in which
all or most of the remnant which germinated survived to flower. Where the number of
plants involved is in any case very small, the loss of several individuals before the
flowering stage may render the result quite untrustworthy.

E. R. Saunders 363

of seeds producing doubles is that in cases where sowings are made
from seed which has been kept for some seasons, a certain error is
likely to be introduced in the direction of making the proportion of
doubles appear greater than it actually is, the variations frf)m the
theoretical result in the case of old sowings being always in the same
direction. The fact that this increase in the proportion of doubles may
be obtained from a sample of quite good seed, after it has been kept,
shows that it cannot be attributed to a particular distribution of the
double-carrying ovules in different regions of the pod, such e.g. as that
those occurring in the distal region give rise to fewer doubles than those
occurring in the basal part, since in the case of Stocks all or nearly all
the ovules in each pod are naturally fertilised under favourable condi-
tions; hence a sample of loose seed is likely to represent all regions
equally. Moreover, direct experiment by means of halving the pods
transversely and sowing the seeds from the upper and lower halves
separately gave no indication of any such unequal distribution.

Subjoined are some of the more striking instances in which the
greater viability of the seeds which give rise to doubles is plainly
manifest.

i. In 1904 a glabrous dark purple plant gave a family of 11 single
and 5 double. This excess of singles was probably accidental since the
plant appears to have been a true eversporting individual, and no doubt
with a larger sowing would have given the usual preponderance of
doubles.

In 1908 about 400 seeds from 9 of the F^^ singles were sown : 44
germinated of which 35 lived to flower, 8 being single and 27 double.
The families were composed as follows :

'amily 1.

16 seeds

sown.

5 germinated

all 5 plants were donble

,. 2.

16 „

>»

all 3 plants „ „

,, 3.

. 19 „

2 were double, 1 died before
flowering

„ 4.

228 „

17 were double, 8 died
before flowering, and 8
were single

The seeds from the 5 other Fi plants gave no result.

In 1909, 30 seeds from one of the F^ singles were sown, only
1 germinated and this proved to be a double.

In 1910, > 130 seeds from another F^ single were sown ; 21 germi-
nated, all of which lived to flower ; 7 were single and 14 double.

ii In 1907 a glabrous white plant yielded an F^ of 83 singles and
100 doubles.

364 Dotibleness in Stocks

In 1908, 30 more seeds of this plant were sown ; only 2 germinated ;
both were double.

In 1909, 47 more seeds were sown ; only 7 germinated and again all
were double.

Of 5 seeds, harvested also in 1906, from a sister plant, but not sown
till 1909, only 2 germinated and both produced doubles.

In 1908 nearly 200 seeds harvested from 5 of the ^i singles were
sown ; 40 germinated of which 27 lived to flower, 5 being single and
22 double. The families were composed as follows:

Family 1. > 30 seeds so\m. 1 germinated and produced a double

,, 2. 34 ,, ,, 12 ,, 8 were double, 4 died before

flowering
„ 3. 33 ,, ,, 15 „ 8 were double, 4 died before

flowering and 3 were siogle
„ 4. 30 „ ,, 12 ,, 5 were double, 5 died before

flowering and 2 were single

All the seeds from the fifth ^i plant failed to germinate.

In 1910, 85 more seeds from 3 of these same ^i plants were sown,
but none germinated.

About 500 seeds from 20 others among the ^i singles gave a total
of 79 singles and 114 doubles. Here the proportion of seeds still
capable of germination, though less than 50 per cent., was considerably
greater than in the lot sown in 1908, and the result is not very different
from what we should expect had the seeds been sown in the season
following that in which they were harvested. From this and other
facts it is evident that the length of time during which the seeds retain
their power of germinating is not fixed but depends probably on the
quality of the seed in the first instance, and on the conditions under
which it is kept.

iii. In 1908, 69 seeds of a sulphur-white which had been harvested
in 1906 gave 23 singles and 32 doubles.

In 1910, 128 more seeds were sown; only 5 germinated of which
4 lived to flower : all were double.

A similar increase in the proportion of doubles was observed in
many cases where the seed was originally of bad quality, and where
only a small percentage germinated even when sown the following
season. This is well shown in the case of the two type forms from
which the largest sowings were made in 1910. Owing to a bad season
in 1909 a great deal of the seed harvested was of miserable quality
and a large proportion failed to germinate. Though no real line of

E. R Saunders 365

division exists, since all grades occur, some arbitrary classification must
be made for the purpose of comparison, and the line is therefore drawn
between those pods where at least half the number of seeds sown
germinated, and those in which less than half proved to be good. The
results may be summarised thus :

Nnmber of Namber of

seeds sown seeds sown

where less Namber of Number Namber where st Number of Namber Namber

th&D half seeds which of of least half seeds wbkh of of

Type germinated germinated singles doables germinated germinated singles doaUes

Marine blae

744

161

108

238

162

Light purple

790

237

125

1439

1040

447

494

In both cases the fruits containing the less good seed have given
a higher percentage of doubles.

As to the proportion of doubles actually obtainable from the various
types the numbers quoted in seed catalogues range from about 50 per
cent, to as much as 90 per cent. In the case of the Erfurt Ten Week
strains from 60 to 75 per cent, is given. This is a rather higher
proportion than was found to occur in the breedings here described,
where the average ranged between 53 and 57 per cent., though
a considerably higher proportion might now and again be obtained
in individual sowings. Chate^ believed his experiments to show that
a larger percentage of doubles could be obtained from the pods on the
main stem and from the lower ones on the primary laterals than from
those on the laterals of a higher order; and similarly from the seeds
from the lower region of a pod as compared with the upper : the
difference is given as 20 per cent, only of doubles from branches of a higher
order as compared with 65 per cent, from those of a lower order, and
30 to 35 per cent, from the upper region of the pods as compared with
75 to 80 per cent, from the lower region. These two latter numbers
would give an average of 55 per cent, for the fruit as a whole, which
agrees very closely with the observations contained in the present
paper, and with the avei*age which the theoretical considerations here
advanced would lead us to expect. No indication of the aggregation of
seeds giving rise to doubles in definite regions of the fruit was obtained,
although a number of observations were made with a view to testing
this point. 68 pods belonging to three different strains (red, marine
blue, and Princess May) were halved transversely, the seeds from the
upper and lower halves being sown separately. The same result was

' loc. eit. p. 79.

366 Douhleness in Stocks

obtained as in the case where the seeds were sorted according to shape
(see below). Sometimes a higher proportion of doubles would be
obtained from the lower half, sometimes from the upper, making it
evident that no constant difference exists in the two regions with
regard to the distribution of the two kinds of seed. It seems in fact
probable that the distribution observed by Chat^ was accidental, and
not the result of any general arrangement throughout the individual.

{h) On the possibility of distinguishing the seeds giving rise to
singles and doubles respectively.

In several papers by earlier writers, treating of Stocks, we find the
statement repeated that more doubles are obtained from the lumpy
irregular-shaped seeds than from the typical regular disc-shaped seeds.
No figures are quoted in support of this view, which is probably the
outcome of an association of ideas rather than of critical experiments,
which would need to be carried out on a considerable scale in order to
allow for any disturbing effect due to the frequent marked irregularity
of distribution which has already been noted. So far no indication
has been observed of any connection between the shape of the seed
and the character of the flower. The glabrous-red race being one in
which many lumpy or irregular seeds constantly occur, the seeds from
a certain number of pods belonging to this race were sorted according
to shape, the flat seeds being sown separately in one lot, the irregular-
shaped seeds in another. It was found that cases where more doubles
were obtained from the flat seeds were about as numerous as those
where the reverse was true, and that so evenly did the variations in
the one direction balance those in the opposite direction that the ratio
obtained from the totals in the two cases was almost identical. Thus
in 1 906 the seeds of 10 pods of the red race were sorted before sowing.
The results were :

From the flat seeds a total of 65 singles and 93 doubles or 1 : 1'43
lumpy „ 19 „ 28 „ 1 : 1-47

Similar sowings in other years gave similar results.

It seems much more probable that the irregular shape of the seeds
is connected with the way in which they are packed in the pod. In the
case of the cream race Princess May, and of a certain strain of sulphur-
whites, the pods are often some inches in length. The seeds are borne
at some distance from one another, and although a pod may contain
from 60 to 70 or even more, they do not overlap. They are so regular

E. R. Saunders 367

in shape that a lumpy seed can only be found now and again. In the
glabrous red, on the other hand, the pods are so short that though very
much fewer in number the ripe seeds are crowded together. Yet the
same proportion of doubles is obtained from each of the three strains.

We may therefore conclude that no system of selection based on
the shape of the seed will enable us to obtain a proportion of doubles
which is constantly above the average. In the case of certain
sulphur- whites however it is quite possible by sorting the seeds
according to colour to separate almost completely those giving rise
to singles from those producing doubles. The present experiments
have shown that there are at least two types of sulphur-white on the
market, one in which the seeds are small, brown, often irregular in
shape, and indistinguishable in appearance from those of a true-breeding
white ; the other in which the seeds are very regular, larger, of a lighter
yellowish colour, and similar to those of the cream race Princess May.
These two types have no doubt a different origin, and are different in
constitution (see later, p. 370). In the case of the type with the yellow
seeds it was found possible in well ripened pods to sort the very yellow
seeds which give rise to the creams which are all double from the less
yellow seeds which give rise to whites of which nearly all are single.
The following result will show the degree of accuracy which can be
reached by this method.

Of 81 seeds taken from one pod

48 were expected to give rise to creams 33 to whites

38 germinated 27 germinated

34 flowered 26 flowered

30 were cream doubles 24 were white and all single
4 were white and all single 2 were cream doubles

Of 72 seeds taken from another pod

44 were expected to give rise to creams 28 to whites

28 germinated 25 germinated

28 flowered 16 flowered

27 were cream doables 16 were white (15 single, 1 doable)
1 was white and single 0 were cream

Thus of the 60 doubles which were obtained 57 were recognised by
the seed-colour ; of the whites 5 were wrongly classed as probable
creams, but the remaining 40 were correctly identiBed, and 39 proved
to be single. A very slight error must however always remain in
sorting the singles from doubles, since the rare double white is not
distinguishable in seed-colour from a single white.

368 Douhleness in Stocks

Note 2. On the inheritance of the branched and
the unbranched habit.

Most races of Stocks branch freely, and in the case of biennial types
form large bushy plants. Of the various sorts used in the present
experiments the Ten Week strains all have the branched habit, as have
also among the biennials, incana and the Brompton strains raised by
Continental growers. The typical English Brompton is on the other
hand wwbranched, the single stout stem being prolonged above the
region of the leaves as a simple raceme. Both leaves and fruits in this
type are thick and somewhat fleshy. The unbranched habit is recessive
to the branched. "When a cross is made between an English type of
Brompton and a branched form the Fi cross-breds are bushy plants like
incana. In F^ the pure Brompton habit reappears in a proportion of
the plants. The sorting of the F^ plants is rendered difficult owing to
the fact that the formation of branches can no doubt be induced by a
variety of causes producing a check in growth. An injury to the
terminal bud or to the roots may cause a check in the growth of the
main axis and lead to the development of one or more axillary buds
which otherwise would have remained dormant. Injuries of this kind,
resulting in a check to growth, are very likely to occur when the young
plants are planted out, and hence in a strict count a certain number
of individuals are likely to be classed as normally producing branches
which in fact only do so owing to unfavourable conditions, or to accident ;
thus the proportion of individuals inheriting the unbranched habit is
likely to appear less than it actually is. In the one mating in which
an English Brompton stock was crossed with a branched form 394 plants
were raised in F^. Of these 66 were recorded as typical Brompton
plants and 31 others as being unbranched except for a single lateral.
These numbers suggest that the true proportion of plants inheriting
the unbranched habit in F^ is probably 1 in 4 as in the ordinary case
of a simple recessive.

The characteristic appearance of the unbranched as compared with
a branched type is shown in the accompanying figures showing two
of the Ffi plants derived from a cross between an English Brompton
and a branched Ten Week strain. (Fig. 1 shows the branched, fig. 2 the
unbranched habit.) The photographs were taken at the end of the
season when the plants were in fruit and the leaves had fallen. In the
Fi generation the Brompton plants presented a very curious appearance,
the single stem in many cases reaching a height of from 3 to 3| feet.

E. R. Saunders

369

Fig. 1.

Pig. 2.

Note 3. On certain sap-colours not dealt with in the earlier
accounts, and on the constitution of the sulphur-white

RACE.

Sap-colours.

Rose is obtained from unions where the colour factors C and R are
present together with a factor for paleness, provided the blue factor B
is absent from at least one of the parents. Hence it is obtained when
flesh or a certain type of sulphur-white (type 1 of p. 367) is crossed with
any form which gives a coloured jPj. If both parents lack B then it
appears in Fi , but if one or other contain B it does not occur till F,.
Thus when sulphur-white type 1 was crossed with red, flesh, cream, or
Brompton white, F^ was rose ; whereas when bred with azure or light
purple the rose colour only appeared in certain plants in F,. Owing

370 Doubleuess in Stocks

to the presence of the B factor, azure and light purple can never give
rose in the first generation, but in any mating with a 6-forni they will
presumably give it in F2.

Rose is epistatic both to the deeper colours carmine and crimson,
and to the purer red shades flesh and terra-cotta.

Lilac is a somewhat bluish pink form, the blue tinge becoming more
marked on fading. It occurs in ^2 from certain unions where flesh is
used, as, e.g. flesh x light purple or white incana. Its position in the
colour series has not yet been determined owing to the failure of the
crop in 1910.

Terra-cotta (? Rothbraun of German catalogues) is a full pure colour.
So far it has only been obtained in F^ from matings between flesh and
sulphur-white or cream. It is recessive to flesh, and possibly stands at
the hypostatic end of the scale of the pure reds as copper probably does
of the impure series.

Carmine and Crimson. These full red colours have hitherto been
spoken of collectively as " red." But carmine certainly includes three
distinct shades, and crimson probably more than one. The two colour
groups together form a very closely graduated series, and a full analysis
of these shades has not been attempted. When, as here, a considerable
deepening of the colour occurs between the unfolding and the fading
of the flower, the range of shades exhibited by individuals of a pale
grade may overlap those of an intermediate class, and so on up the
scale, thus increasing the difficulty of sorting.

The same difficulty is met with among some grades in. the blue
series, but the three main classes, dark purple, light purple and azure
or very light purple, are easily distinguished. Marine blue is a larger-
flowered form, in range of tint between unfolding and fading covering
almost those of azure and light purple together. The two paler forms
azure and marine blue, differ from the more deeply coloured purple
types in having brown and not green seeds.

Constitution of the sulphur -white race.

All sulphur-whites were found to behave alike when self-fertilised,
in giving a mixture of white singles and cream doubles ; all probably
also give a small percentage of white doubles. When bred with other
glabrous forms the results varied according to the type of sulphur-
white employed. Seed supplied by Messrs Haage and Schmidt proved
to belong to the second type described above (p. 367, seeds yellow, large,
regular). The plants crossed with glabrous cream gave F^ all glabrous,
and either all cream or mixed white and cream, according as the

E. R Saundkes 371

sulphur-white was used as <^ or $. When crossed with glabrous
white or glabrous sap-coloured strains F^ was hoary and sap-coloured.
If a full sap-colour as e.g. red was used, a full colour was obtained in F^.
This type of sulphur-white contains the hoary factor K\ and one of the
two factors G and R necessary for the production of sap-colour; the
one present must be the one which occurs in Princess May (= R). The
other colour factor (G) and the factor which turns red blue (B) are
both absent. We can therefore express the composition of this type
of sulphur-white thus — bcRK. The seed obtained frqp Herr Benary
showed the characteristics described under type 1 (p. 367, seeds small,
brown, irregular). This form evidently has the composition bCrK, and
has also a factor causing paleness, so that in a cross a full sap-colour
carried by the other parent becomes pale in Fi. This type when bred
with glabrous cream or a glabrous sap-coloured form gives F^ all
hoary sap-coloured; with glabrous white on the other hand it gives
Fi all glabrous white. Bred together these two sulphur-whites should
give a sap-coloured hoary Fi of a pale red colour (=rose). It was
hoped that plants from this mating would have been raised this year,
but unfortunately owing to the bad season in 1910 no good seed was
obtained. Indirect proof however is already forthcoming, for a mating
in the form

[sulphur-white (type 2) x glabrous red] x sulphur-white (type 1)
gave all rose hoary (217); whereas the mating

[sulphur-white (type 2) x glabrous red] x sulphur-white (type 2)

gave the expected result — half the offspring being red hoary and half
white smooth.

The expense incurred in the course of the present work has been in
part defrayed by a grant from the British Association for the Advance-
ment of Learning, and also during the present year by a grant from the
Gordon Wigan Fund. The experiments were carried out in one of the
allotment gardens of the Cambridge Botanic Garden, which for some
years, by the kindness of the Botanic Garden Syndicate, has been
permitted rent free.

I wish here to express my thanks to Miss Killby, who in the course
of the work has given me much valuable assistance in the garden, and
who kindly took the photographs here reproduced ; also to those friends
who were kind enough to raise and record a number of the plants.
» See Evolution Report IV. p. 36.

372

Douhleness in Stocks

•2 t-i-HC5O5iNU5W5C<5-^fHa0'*
gi-(COi-l OrH(Nt-i-iaO t-

•g « t- l.O IM

o ■>

a O CO 0» rH iH t- US 04 U5

S" T-t (N «5 iH
•S «3

S TttlNi-HOOOJ-^eOCfSOKN^OOOlCOi-l
O i-H COeO>-l-^rHiH

3 » * «

S Cqi-lTt(«50»0-»fTj<05C<li-|IN

o (N 1-1 eo i-M i-H

mcocoQooot-Qoco

rH iH eO rH

60 iH i-< CO C(5 lO

00 O iH «0 ?0 iH

"«

I-H
M

I-H

<«

•S

00,

•<!»

0
O

g SC~-^?0«5'<i<iNCO«5'<l<-*COi-l
?= ft

■a, « 05 MS «5 «D Tjt ,-( •»»( S» »0 CO eO rH CO 1-1 52, -g, O CO U5 lO (N i-l O) CO »0 iH I «o

B iH r-l rH (M ^ r-l CO 1-1 1-1 S<l CO c '

■§ l> 00 O OS 00 CO OS O <N OS •«»< CO U5 «c •*
O -^ lO»Oi-((MiH 1-H

J5 C05D'-IOOQOCOrHOOC<ICDO>'5
3i-iaO<Ni-ICO<M<N iH

ft ^

■3) e0«5<MO0S>OOi-IQ0»OTj((MC<10S

a -"H 1-1 -* -* 1-1 CO

o "^

3 «005<MC0050s>Oi-lOS<)i-IC^>niM«0
OrH CO rHf-llO ^O »0 •<*l

•2 OOOSCOTt<»C<Mi-lt~CO»«OCO
orH O^COOJCOi-l COWCOi-l

CO Cs O OJ -^

■S OS-*COf-l-*eO<MOOOrHCOCO'<1<»CO
§ (MrHdOIIN i-liH COCO(MtJ<i-I

3 oi-ifHCijcoOii-iosic-^oao
Si-i (N>o co-^eoooseOrH
o

-2 eot-coooostocooocoooioooosos

SPr-li-lrHi-l'H rH (MCMi-leO

"Soc^i-tOJcoust-THosoootoeo
a (M»o i-(C^iMi-(c-usi-<

E. R Saunders 373

"5©

3 f^

33 "

«2 — I. — "3

o _2

^ "B. C I >0 ^ 3J »3 35 U5 C<I £ je O 1-1 « 50 -Si O 05 IN -^ £2- JP X

I* § § ^ i s ^ 1

^ • i •§ -2 . i :i I

te"9-^'-l»-<C<l-HXXer5i-l 3"§OT(<t- °3«.-l(NTt< Se^ u'3<5=i °

•J3o ^"^ P 9 ^ O* m O "2 eSO-^ a>

O £

I ^ •§ ^ ^

as* £ ^ S 0» S •*

= "i 1 w 1 I |CJe-e<ii-i tc o lis o» "-^ «» •-< I I-* » i sc « g

Ocllll'H.-* s in o* B > ' £ I £ <-< -2

X X X 33 X g

•I a

O O ® C3 "* IB

a lo rj e« >o CO C4 ■« us cq >o 'sdr-ica soo-^om S >a >o 3 "t* 2 30

5" Sou,-* J ooo>t- g S •* ^ 2 'H

ff>Hc« 5'"' B>oxt-l J^*^' 5*" —

00 33 35 X CO S.

Joam. of Gen. i 85

374

Douhleness in Stocks

TABLE IV.

Showing the number of singles and doubles obtained in F^ families derived from
matings of the form no-d '^ x-d $ .

wo-d-Klabrous white ?

10-d-glabrous flesh ?
d-siilphur white S

no-d-glabrous

cream ?

d-glabrous red $

d-glabrous azure S

Single Double

Single

Double

Single Double

Single

Double

Single

Double

—

148

no-d-glabrous cream ?

16
13

9
6

6
16

1
1

1
7

1
3

d-sulphur white (f

Single

Double

Single

Double

?[33

111

no-d-elabrous

cream ?

d-hoary wLite (intermediate) i i

11
20

9
3

7
2

no-d-

-hoary white (Brompton) ?

d-glabrous

red-<

Single

Double

Single

Double

8
2

4
2

Single

Double

Single

Double

3
3

no-d-glabrous

cream 2

no-d-glabrous flesh ?

d-glabrous white <f
>■

Single Double

-;U

Single

Double

Single

Double

no-d-glabrous cream ?
d-glabrous red S

1
6

/■ —

mod-glabrous flesh 2

Single

Double

Single

Double
9

d-glabrous light purple <

Single

Double

Single

Double

208

no-d-glabrous

cream 2

116

d-glabfous light purple <?

26
60

1
2

Single

Double

Single

Double

242

»io-d-hoary white (Brompton) ?

118

9
16

d-sulphur

white S

107

167
6

85
7

32
12

Single

Double

Single

Double

139

125

no-d-glabrous white ?

d-glabrous

cream 3

Single

Double

Single

Double

1 See note, p. 326.

E. R. Saunders

375

TABLE V.

Showing the number oj singles and doitbles obtained in F^ families derived from mtUings
of the form d^ y. no-d S • In this case some of the families toill contain doubles and
some tvill be aU- single. [The all-single families in each mating are arranged on the
left, the mixed families on the right.^

d-glabroiu red $

rf-saluhur white ?
«io-d-glabroiu flesh S

d-«alphur white ?

no-d-giabroiu white i

•MHi-glabroua white i

Single

Double

Single

Doable

Single

Double

Single

Doable

Single

Double

Single

Doable

—

124

—

116

—

rf-sulphnr

white?

no-d-hoary white (tn<ami)<f

29
26

—

1
17

7
2

—

71
8

/^ ^

Single

DoaUe

Single

Doable

114

—

120

—

—r.

d-Bolphar

white?

NO-d-red hoary (Brompton) S

22
61

Single

DooUe

Single

Doable

137

d-glabrons 1

156
132

cream S

29
33

60
87
13
33
25

13
7
2
7
3

tu>-d-glabroaa white S

36
48
13
42

7
7
6
7

Mngle
36
79

Doable

Single
42
10

Doable
7
3

—

138

376

Doiibleness in Stocks

TABLE Vl.

Showing the number of singles
and doubles obtained in 50
Fi families derived from
matings between two ever-
sporting forms. (Seep. 319.)

Single Double

TABLE Vn.

11
5

5
1
2

14
7

14
8
52
2
31
78
13
17
16
13
8
18
23
7
1
1
12
13
16
6
12
86
4
2
13
9
18
10
29
11
25
17
7
12
8
21
4
19
12
11
8
S

24
7
7

Shovnng the number of

singles

and doubles obtained m 81

F^ families

when i

the Fi

cross

-breds f

rom matings

between two

eversporting

forms are

self-fertilised.

(See

p. 324.)

Single

Double

Single

Double

—

1*2

( 12

17)
141

1*94

*85

146

*12

TABLE VIII.

Shounng the number of singles
and doubles obtained in 35
F2 families when Fi cross-
bretds from matings between
two eversporting forms were
crossed back with one of the
eversporting parents. (See
p. 319.)

Single

Double

* Eecords marked with an asterisk were obtained from delayed sowings (see Appendix, Note 1).
Below are shown the matings from which the above families were derived.

Fam.
1-2 cream
3-4

5-7 red
8-11 ,,

28-55
56-57
58-62

X white

xred

X cream

X sulphur-
white

12-19 sulphur- white X white(hoary) 63-75 ,,
20-40 „ „ xred 76-79 cream

41 jj „ X white 80-81 ,,

42-44 „ ,, X azure

45 ,, ,, X light purple

46 azure(hoary) x sulphur- white

47 flesh X azure

48 azure x red

49 light purple x „

50 red x light purple

Fam.
1-13 red x cream

14-27 sulphur-white x red

X azure

x light purple

X white

X white(hoary)

xred

X white

Fam.
1-2 red x (red x sulph.-wh.)
3-5 ,, X (sulph.-wh. X red)
6 (sulph.-wh. X red) x red
7-31 ,, ,, X sulph.-wh.

32-33 (red x sulph.-wh.) x sulph.-
wh.

34-35 (cream x sulph.-wh.) x red

NOTE ON THE INHERITANCE OF CHARACTERS
IN WHICH DOMINANCE APPEARS TO BE
INFLUENCED BY SEX.

By L. DONG ASTER, M.A.

Fellow of King's College, Cambridge.

A NUMBER of cases have been described, in which it appears that a
character is dominant in one sex, recessive in the other. Such cases
fall into two categories, according to whether the character concerned
is inherited in the normal Mendelian manner, or is sex-limited in its
inheritance. Examples of the former type are the homed character in
sheep (horns dominant in the male^), and probably the white colour in
the butterfly Colias (white dominant in the female') ; of the sex-limited
type examples are colour-blindness, hereditary nystagmus and haemo-
philia in man, and probably the orange colour in cats'. In the latter
class it has frequently been stated that the character concerned is
dominant in the male, recessive in the female. Taking colour-blind-
ness as an example, we find the following facts. A colour-blind man
married to a normal woman has usually only normal offspring ; his sons
do not transmit the affection, but his daughters transmit it to some of
their male children, as in the following scheme:

^ X 9 i colovu*-blind man

(J 9 X (J (J normal man

i 6 9 9 9 normal woman.

A colour-blind man married to the normal daughter of a colour-blind
man may have colour-blind daughters as well as sons, thus :

I 1

i 6 f 9

* Wood, Journ. Agric. Science, in. 1909, p. 145.

* Geroald, Amer. Naturalist, 45. 1911, p. 257. In this case there is the complication
that homozygous white females have not been observed.

3 Doncaster, Proc, Camb. Phil. Soc. xni. 1905, p. 35. Since the pablication of that
paper I have obtained evidence, not yet conclusive, that the inheritance of the orange
colour is sex-limited. Experiments to test this more fully are being made.

378

Inheritance of Characters

The explanation commonly given of these facts has been that
colour-blindness is dominant in the male, recessive in the female, so
that the male heterozygote is colour-blind, the female heterozygote
normal ; a colour-blind woman can thus arise only when the affection
is inherited from both parents. It is also evident that the affected
male transmits the factor for the disease only to his daughters ; the
heterozygous female, however, transmits to some of the offspring of
both sexes. This sex-limitation of the transmission makes a different
explanation possible, which is also more in accord with other cases
of sex-limited inheritance.

Since the male transmits the factor for colour-blindness only to
his daughters, it must be assumed that the male in this case is
heterozygous for the sex-determiner. In former papers I have sug-
gested that if maleness is determined by a factor ^, femaleness by a
factor % epistatic to ^ when both are present, then a male individual
may be represented <^0, a. female jf $; i.e. that both sexes are hetero-
zygous for sex-determiners, with selective fertilisation between (/-bearing
eggs and 0-bearing spermatozoa, and between $ -bearing eggs and </*-
bearing spermatozoa ^ If we adopt this scheme as a working hypothesis,
and then represent normal sight by N, colour-blindness by absence or
modification of iV"(=n), and further suppose that N can only be borne
by gametes containing a sex-determiner ((/'or %, not 0), we obtain the
observed results.

Parents
gametes

gametes

m 0

(affected male)

(normal female)

Ns 0
(normal male)

Ns, 0

n (T iV ?
(normal female
heterozygous)

m , N i

N s , n ?

ns 0 Ns 0 iYc?7i? N^ N9

(affected male) (normal male) (normal female (normal female)

heterozygous)

ns 0
(affected male)

n<r, O

Ns ni

(heterozygous

female)

N s , n ?

I ' 1

m 0 Ns O Ji<?n¥ ns N 9

(affected male) (normal male) (affected female) (normal female

heterozygous)

1 Proc. Roy. Soc. B. 82, 1910, p. 88, B. 83. 1911, p. 476. My reasons for continuing to
prefer this scheme to that of the American writers, who represent the male as XO, the
female XX, will be given in a subsequent paper on the same subject. The argument in
the present case appUes equally to both schemes.

L. DONCASTER 379

This scheme is exactly comparable with T. H. Morgan's results on
the inheritance of the white eye in Drosophila\ in which no suggestion
of alternative dominance has ever been made ; if N represents the
factor for red eye and n its absence (white eye), the scheme does as
well for Drosophila as for human colour-blindness.

The scheme here outlined will apply to all cases of a character
apparently dominant in one sex only and also sex-limited in its
transmission by that sex (with the possible exception of the orange
colour in cats, the inheritance of which is not adequately known).
It will not apply to cases which show no sex-limitation in inherit-
ance (e.g. horns of sheep) ; in these it must probably be supposed
that a sex-limited modifying factor is present in one sex.

1 Morgan, Science, 32. 1910, p. 120; American Naturaliat, 45. 1911, p. 65.

CAMBBIDOE: FBINTED BT JOBN clay, M.A. at the UXIVEB8ITY PBE88

QH Journal of genetics

A31
AIJ64.
v.l
cop. 2

Biological
& Medical
Serials

PLEASE DO NOT REMOVE
CARDS OR SLIPS FROM THIS POCKET

UNIVERSITY OF TORONTO LIBRARY

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Books

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Journal of genetics"

See other formats