//^

JOURNAL OF GENETICS

CAMBRIDGE UNIVERSITY PRESS

C. F. CLAY, Manager

LONDON : Fetter Lane, E.G. 4

LONDON : H. K. LEWIS AND CO., Ltd., 136 Gower Street, W.C. i

LONDON : WILLIAM WESLEY AND SON, 28 Essex Street, Strand, W.C. 2

CHICAGO : THE UNIVERSITY OF CHICAGO PRESS

BOMBAY, CALCUTTA, MADRAS : MACMILLAN AND CO., Ltd.

TORONTO : J. M. DENT AND SONS, Ltd.

TOKYO : THE MARUZEN-KABUSHIKI-KAISHA

All rights reserved

w

JOURNAL OF GENETICS

\\\

KDIIKI) HY

VV. BATIsSON, M.A., K.R.S.

OIRECrOR OK 1HK JOHN INNKs HORTK Ul.rURAl. I NM I 1 1' t l«)'<
A N D

R. C. PUNNKTT, M.A., I .R.S.

ARTHUR BALFOUR PROKKSSOR OK (.KNKTICS IN THK UNIVhRSlTV OK CAMHRIIm.k

Volume VIII. 1918—1919

Cambridirc :

at the Univcrsitv Press
I 9 I 9

m

tei

t

CONTImNIS

No. 1 (Dei-cnilHT, I '.Ms)

H. C. Pu^'N^nT and thr Into \\ (I. IUii.kv. (J.nrtic Stiulirs in
liahhits. I. On t\\v Inhoritniicr nf \\'«ii;lii. (W itii twrlxr ti\i
figurt»s) I

R. C. PuNNKTT. NtjU'oii tlu' Oriniii of .i Mutation in tlir Swc*-! IN-jl

(Witli ont» toxt-Hgurt")) ........ JT

S. Ikrno. On Hyhridisation of sonu* Sjxvios of Sa/u\ (Witli

PlaU» 1 ixiul ono ti'xt-tigure) . . ^ . . . . ;'».',

B. MiYAZAWA. Studies of Inheritance in tlic .lapant'sc (\niro/rnf>is.

(With Phit« II and one t43xt-figure) ...... .'»'.♦

No. 2 (April, 11)19)

EL S. Salmon. On Forms of the Hop {//nmn/us Lnpu/us L. ) resistant

to Mildew (Sph(ierotfieca IIuninH (DC) l>un.); II. . s.J

W. Bateson. Studies in Variegation. I. (Witli PlaU's 111 and I \'

and one text-figure) ......... '.♦.'{

S. C. Harland. Inheritance of certain eharact^'r.s in the ( 'owjM'a

{Vigna sinensis). (With one text-figure) . . . . lUl

(). WiNGE. On the Relation between nunil>er of ('lironiosoMMs and

number of Types, in /,«^%r«.s' e.specially. (With I'lat*- \' ) I.J.'J

No. 3 (June, 1919)

E. J. Collins. Sex Segregation in the P»ryopliyta. (With lMai«- \' I

and five text-figures) .1 .'.!♦

J 0H8. Schmidt. Racial Studies in Fishes. II. H\|MMimrnt.il Invcstj

gations with Lehistfs rt'ticahitus (J^t<'rs) Kr^aii. (With <.ri.

graph) . . , . . 117

J. E. DuEKDEN. Crossiiii; tin' North Afiitau and SuiuJi .Vfijc.ui

Ostrich. (With Platr VII anil two t.xtJiLiur.s) I ."..">

W. Batkson and Ida Sutton. Double Flowfis :in<i S\ l.inka;;*- in

Beyonia. (With Plate V ill; 1 •.»•♦

PAGE

vi Cmitents

No. 4 (September, 1919)

H. Onslow. The Inheritance of Wing Colour in Lepidoptera.
I. Abraxas grossulariata var. lutea (Cockerell). (With Plates IX
and X, table, and twenty-five text-figures) . . . . . 209

J. W. Heslop Harrison. Studies in the Hybrid Bistoninae. III. The

Stimulus of Heterozygosis. (With two text-figures) . . . 259

Edward Hindle. Sex Inheritance in Pediculus humanus var cor-
poris. (With one chart) . . . . . . . .267

C. C. Little. Colour Inheritance in Cats, with Special Reference to

the Colours Black, Yellow and Tortoise-Shell . . . .279

J. B. S. H ALDAN E. The Probable Errors of Calculated Linkage
Values, and the most Accurate Method of determining Gametic
from certain Zygotic Series . . . . . '. .291

J. B. S. Haldane. The Combination of Linkage Values, and the
Calculation of Distances between the Loci of Linked Factors.
(With one text-figure) 299

COKRIGENDA IN VOL. VII.

P. 201, line 13, for K^J^0 ^o + iLj^^) («o-4rofo)
D 5

read /f, = (i±/-«» ,„ + '! :_^'l) (,„._4r„(„).
P. 201, last line, for homozygotes read heterozygotes.

CORRIGENDA IN VOL. VIII.

P. 74, line 27, for equally 3 : 1 read equally 1 : 1.

P. 79, last line but one, for (17 - 1) read (17 - 13).

PI. II, fig. 5. The stem shews some brown pigment in the original drawing which has
been lost in the reproduction.

Volume VIII DECEMBER, huh No. l

GENETIC STUDIES IN UAHBITS.

I. ON THE INHERITANCE OF WEIGHT.

By R. C. PUNNETT. F.R8.,
AND THE LATE Major P. G. BAILEY. R.F.A.

(With Twelve text-figures.)

CONTENTS.

PA«E

Introduction 1

General account of the experiments 2

A. Flemish-mixed cross 3

B. Flemish-Polish cross 5

The Grovk-th Curve 7

Sexual diflferences 15

A. Mature weight 15

B. Rate of growth 17

Inbreeding 19

Literature referred to 21

Tables of Data 21

Introduction.

Several investigators have published data relating to the inherit-
ance of weight in rabbits. The earliest experiments appear to be those of
Huth ('87) which were undertaken with the idea of tvscertaining whether
continuous inbreeding through brother and sister mating led to any
diminution of size. The results are not published in a form which allows
of analysis, since the individual weights of the different animals, whether
parents or progeny, are not given. Huth however came to the con-
clusion that inbreeding over six generations did not lejul to any
deterioration in size. More recently Ciistle ('09) took up the (piestion
of size inheritance in this species. He admits that his statistics are
unsatisfactory as he wiis unable to keep many of his animals until the
adult state was reached. Most of his compiirisons were consequently

Joom. of Gen. tiii 1

2 Genetic Studies in Babbits

made at an age of 18 weeks, and this, as will appear later, is too early
for trustworthy conclusions. Nor did he realize the marked sexual
differences which may occur, and which must, as we shall explain later,
be taken into account in work of this nature. He came to the con-
clusion that weight inheritance is blending in character, and that neither
dominance nor segregation in the Mendelian sense is recognisable.

Castle's work was done before Nilsson-Ehle had put forward the well
known hypothesis of multiple factors. More recently MacDowell ('14),
in continuing Castle's experiments, has come to the conclusion that his
data are all in accordance with this hypothesis. MacDowell's work is
mainly concerned with skeletal measurements, and though periodical
weighings of his animals were made he does not attach much signifi-
cance to the results, pointing out various factors which tend to militate
against accuracy.

Quite recently a few observations on the growth of rabbits have been
published by Da vies ('17) who states that when a small breed is crossed
with a medium sized one the young approximate to the size of the
smaller parent. But he admits that the cross was only made in one
way using the animal of the smaller breed as the mother. He is inclined
to consider that the reciprocal cross may give a different result, as he
holds that larger size depends more upon the mother than upon the
father.

We cannot yet pretend to know very much about the inheritance of
weight in the rabbit. It is clear however that investigations of this
nature in the mammal do not yield such clean-cut results as in birds
(cf. Phillips ('12), Punnett and Bailey ('14)), 'and it is evident that a
great deal of laborious work must yet be done before we can understand
a process of which the theoretical, as well as the economic, importance is
so considerable. We do not therefore propose to discuss the matter
from a general point of view until we are in possession of fuller and
more complete data than those of which, we may now proceed to give
some account.

General Account. Our experiments were started in 1912. The
breeds originally chosen were the Flemish \ one of the largest of rabbits,
and a strain of mixed Himalayan-Dutch-Havana origin which had been
formed in the course of some earlier experiments on coat colour. We
shall refer to this as the Flemish-mixed cross. The three breeds which
entered into this strain are all on the small side and on the whole differ

1 We are indebted to the kindness of Dr B. N. Salaman for procuring these animals
for us.

R C. PUNNKTT AND TIIK LATK P. ill BaILKY H

littlo from une aiiuther in |N>int of ninv*. The* Htmin wiim chom^ii i»ii
account of iu peculiar p<it.torn. niucv it wuh h<»|KMl, hy croHHJn^ it with n
self-coloun'd nice, U» obtain data on \\w inhrritanre of whit<' marking
on the coat Those ex|K)rin)ent« an* ntill in pro)(n'w and the rrHiiltM will
be presentini in nonir futiin* |Mi|M'r. In WH't n furtln r set of rx|Hri-
tnents was Ht-art'od. UHtn^ lus pan*nt nwrs thr KlmiiMh and thr l*olihh^
the latt«»r Kmu^ thf HUjallest bn'ed of nibbitw in donu'stiaition.

The work, so far Jis it has ^one, is ncetvHsarily of a preliminary natun*.
We did not begin by ci*ossin^ strains of unifonn sizr.a |)oinl of thr first
imporUmce in work of this kind. Our reason for not doin^ so wjih of
course the impt>ssibility of finiling them. The " pun; " breeds of rabbit>j
of which we have had expi^ricnce shew Huetuati(His of size, oft m con-
siderable, which c^innt>tbc put down t<> ill health, alteration of conditions,
and si> forth : and we have little doubt that this is true for all reco^nnscd
bree<ls. The student of genetics, like the chemist, oftt'ii has to jiurify
his raw material as a preliminary to critical research. In the j)resent
case this is a matter of some years. But while the standardisiition of
our niJiterial was pnxieeding we carried on the crossing exi^'riments
referred to above with a view to obtaining useful experience for more
extensive and criticiil work in the future. That the results hitherto
obtained are indecisive is no disappointment, having reganl to the nature
of the material and to the resources at our command. But in spit<' of
the limitiitions of our data we believe that they will prove of service in
directing the attention of other workers along these lines to jK)ints
hitherto unconsidered in connection with the inheritance of weight. On
some of these it is ho[>ed that further light will be thrown when the
material now being worked up becomes available for crossing purpo.ses.

A. Flemish-mixed cross. The original Flemish (/ {N 169) was mated
with two does of the Himalayan-Dutch-Havana strain, ^V17 and X \9,
which weighed respectively 6 lbs. 2 oz. and 0 lbs. 0 oz. These two does
were closely related to one another as is shewn by the accompanying
pedigree (Fig. 1). To what extent the members of this strain of mixed
origin varied in weight we are unable to say, since records bearing upon

' Davies ('17) speaks of modern Polisb, Himalayan, Dutch, and Tan varieties approxi-
mating closely in size to their wild prototype, and he also mentions 3 lbs. as the wii^ht of
a Himalayan doe. This is very much less than the wei^'ht of such a doe 10— 1'j years
back. Apparently the trend of the fancy has been towards smaller si/.e for these breeds in
recent years.

' For these animals aad for sundry information about them our thiinks arc due li»
Professor J. Stanley Gardiner.

Genetic Studies in Rabbits

this point were not kept while it was being formed \ The experiments
involving $ iV 19 are less extensive and may be considered first. Six
i^i animals were reared (cf. Table II, p. 22) and, as shewn graphically in
Fig. 2"^ they were intermediate in size with the exception of one indi-

9 X crHaiva.Ti8L X 9

Fig. 1.

Pedigree of the two does, ^ Nil and ? N 19,
used in the Flemish-mixed cross.

9Nt9

1

9O223

M

(j*Nl69

f

Fig. 2. Graphic representation of weight distribution in Fj and F2 generations from the
Flemish-mixed cross, ? iSTlQ x <? iV 169. The numbers above each of the columns
separated by broken lines denote lbs. Each black circle represents an individual and
is placed according to the weight of the individual in lbs. and oz. The record number
of those rabbits which have been mated together is given, cf. Tables I and II,
pp. 21—22.

1 Considerable numbers of animals involving these three breeds were reared by me
between 1907 and 1912, I do not recall any marked differences in size. The great
majority, if not all of them, were probably between 5 and 7^ lbs. E. C. P.

^ Unless any statement to the contrary is made in Tables I— V each weight given is
the maximum attained during the first twelve months (cf. p. 7).

R C. PUNNRTT AND TIIK LATK P. (1. lUlLKV 5

vidual (<^ 0 224) which wiw of thi» wiiuf weight jim it** inothor. An
#*! generation (cf. Tiiblo 111, p. 28) wjw nuinKl fnun two iiu'iuIxth of the
^i family which wi»rt» nrurly of thf munv sizf. It conHJHtt**! of 18 indi-
vidualHatui oxhibit<Ml the consi<K'nil)lr nm^o nf variation shrwii in Fi^. 11.
Points of int<M\»«t an», (n) that the avcrap' weight is rlom* t«» that of the
orii^inal small |»ari>nt (AMJ)) ami considtTahly Irns thiiii that <»r thi*
Fi gi»nen\tit)n, {h) that only one animal atUiinr«l th«' mean iwircnUil
weight, and (c) that in several c^wos animals wore produce<l which win*
much smaller than the ori^^inal small jwiicnt (.V 19). As compared with
Fi the /*, generation shews a marked shitting towanis smaller size. The
numbers are Uh^ small to draw any deduction of value jxh U) there Iwing
an increase of variability in F^ jus comjKired with F^.

The ex[x»riment.s with ^ N \7 are more extensive and have Ix-en
ciirrie<l thmugh several genemti»)ns. The 15 F^ animals reared shew a
wide mnge of variation, some of them being nearly as small as their
nu»ther while othei-s are c«>nsideral)ly heavier than tln'ir father (ef
Table 11, p. 22, and Fig. 8). F«)ur /', animals involving time different
matings were subse<juently used t<» give an F, generation. These four
animals wore all within 1 lb. of one another. The F.^ generation t^iken
together exhibits great variability (cf Fig. 3) extending some way
beyond the ninge of the original parents on either side. Though the
mean of the /T, generation jis compared with that of the F^ generation is
shifted towards the smaller size this is not so marked a phenomenon as in
the preceding Ciise. In this connection we shall have more to say later.

It was our original intention to breed an F^ generation from a pair
of the heaviest F2 animals as well as from a pair of the lightest. Con-
sideration of space however compelled us to restrict ourselves to the latter
part of the programme. An F^ generation of 23 was reared from two of
the smaller F^ animals (JO 187 and J* 0 192). As compared with their
parentis the mean of this generation again shews some shifting to the left
in Fig. 3. At this point it wtis decided to try the effects of continuous
inbreeding from the smallest of the ^^3 generation in order to ascertain
whether a fairly true-breeding strain of small size could eventually be
extracted. It is hoped to expand this part of the experiment in the near
future and to test the effects of close inbreeding with considerably
larger numbers.

B. Flemish-Polish cross. A small Polish buck weighing 3 lbs. was
successfully mated with two Flemish does. In one c;use the doe (O 138)
was more than three times as heavy ;vs the P«»lish buck. A single pair
of Fi animals w;vs reared and proved to be nearly interimdiate in size

Genetic Studies in Rabbits

o

r%

•

_

—

—

•

O)

t

cx)2

■z.

-

CO

o

Of

• —

•Si

CO
- CM

z
01-

Y

-•

•
•

h-

nP

-'

#

•

r*

»

ses

•

-

•~

•

r-%

—

CD

J

•

/8100

•
•

•

>•

r

•
•

•

%-^

h-

^^'op

•

i

z

lO

•

-•

•

-•

68ldi

£/

Av

Lt|_/

1^

CO _Q

s «

.2 H

Ol OQ

cx> -*^

I ^

.a ^

.2 -2

.2 f«4

« CO

o8 <«

CL

IL C. PUNNRTT AND THK LATK P. (i. BaILKV 7

between the pnrentM (of. Tuhh* V, p. 25, and Fig. 4). Two Iitu»rH were
bnxi fn)in thiH pjiir g^^'^^K * ^ ^'a Hninmls in nil. Th<* aviTiigf si/.f nf
tht»8f wiu* (iJHtinctly U'hh than that «»f thr |Nin'n(«, hnt, although th**
variability wiiMConHitlorahh', nothing was pnMluc<*<l apiiroiu'hing th<* Nniall
Hizv of the grandfather. Possihly this niay ho dm- nu'icly to th«- small
ninnl)er of F^ animals roariMl.

From tho othor Flemish doe (0 208) the numhors an- grcator. Th<*
ft>ur /*, animals wore closely intornjodiatv in size hctAvccn the jwinuts
(cf. Table V, j). 25, and Fig. 5). From these two /', pairs :U) /aotfspring
were n»Ared to maturity. The results are clo.sely comfwirable with those
nbUiininl from the otht^r Flemish -Polish mating. There is considerable
variability in Fo though very few individuals reach<'d the F^ size and
only one exceeded it, and that but slightly. Xo individual shews any
approach to the large size of the Flemish. 8f>m<' shewed a fairly close
approjich to the small size of the I'olish grandfather, and it seems not
unlikely that with larger nutnbers animals of this extreme small size
might have reappeared in F,,.

Taking all of the exi>eriments togi-ther the chief j)oint of interest
about them is, un(juesti«)nably, the failure of the larger form to reapjx-ar
in Fi in certain of the crosses. It is marked in the Flemi.sh-Polish
crosses and evident in the Fo family from iYl9. On the other hand it
does not seem to occur in the other Fo generation from the Flemish-
mixed cross, viz. that from iVl7. The non-appearance of an expected
class in F^ has occ^isionally been recorded for sundry characters, but the
only instance which we can recall in connection with size is tliat of
Ejist s Nicotiana cro.s.ses ('17), where, fiom the cross between N. Lanffs-
dorffii and N. alaUi, the small corolla length was recovered in F., with-
out difficulty while nothing appioaching the long coiolla of N. (data
reappeared. With this brief account of the crossing experiments we
may now proceed to discuss various points to which they have given rise.

The G-rowtli Curve. We have alieady stated that the weight reconled
in our tables is the maximum weight attained by the animal during the
first twelve months of life^ Generally speaking, a rabbit grows rapidly
during the first half year of its existence, after which the monthly
increment graduall}^ becomes smaller. Somewhere between 7 and
10 months in the smaller breeds a maximum weight is reached, after
which there is usually some decline, connected probably with the ful-
filment of sexual maturity. Latei* <tn an incicase in w«'ight is again

* With some ftw exceptions where ^rowtli was slow and maturity delayed. Tliese are
noted in Tables I — V, pp. 21—25.

8

Genetic Studies in Rabbits

•

CO

05 6
0+

00

r^

O

-•-

o
o>

O

(O

^

1

•

g

o

-s

a

o

pw

,a

,

OD

ca

<»

Pi

f^

>

H

3

<a

00

H

s

O

^

en

<u

03

n

^

hr

bO

^^

^

ra

tt

g

(S3

ni

«

(D

t*H

o

a

00

""*

oR

o

.2

03

^

OQ

a>

-o

tS

ci

<D

o

^

^

o

§

-s

o

^

flH

q>

•^

(U

X

^

00

ft

o

O

s

o+

P4

o

R C. PUNNRTT AND TllK LATB P. G. BaILBY

9

3
y'JPoUsh

7
9O203f

t?

•• ••rt pn

• •:« I

nr* t

• •

Fig. 5. Graphic ropresentAtion of weight distribution in Fi and F^ ^enerationfl from
Flemish -Polish cross, ? () 203 x j Polish. For furtht-r details as to exact weiKhts, st-e
Table V, p. 25.

noticeable in most cases, and frequently the weight eventually attained
is greater than the maximum registered before the age of 12 months.
This later increase is more marked in does than in bucks, and is probably
connected with the deposition of fat. The general features of the
growth curve are illustrated in Fig. 6, which shows graphically the rate

18 10 20

Fig. 6. Growth curves of 3 Fj animals from Flemish-Polish cross. The curve for d" P 117
is almost identical with that for <f 7^116 and has been omitted for clearness. The
black dots on the two female curves indicate the ages at which litters were produced.

10

Genetic Studies i7i Rahhits

of growth of three F^^ animals from the Flemish-Polish cross (cf. Table V,
p. 25). In the large breed to which our experience has been limited^
the Flemish, growth is relatively slower and the curve is consequently-
more drawn out. In the case of the four animals whose growth is shewn
on Fig. 7 the maximum weight, prior to cessation of growth, was not
attained until an age of 14-15 months was reached, and it was also
noticed that these animals were very slow in arriving at sexual maturity.
Nevertheless the usual decline in weight subsequently set in, and in
this respect the heavier breed conforms to the type of the curve exhibited
by the lighter animals of Fig. 6. It would be of interest to learn
whether a rise in weight occurs later in life, but owing to lack of space
we have been unable to keep the animals long enough to test this point.
It will be noticed from Figs. 6 and 7 that at seven months the F^ animals

Weight
in lbs.

7-

6-

Agein • T"

months! 8

1 — r

9 10

~i r

11 12

n 1 1 1 1 1

13 14 15 16 17 18

Fig. 7. Growth curves for 4 Flemish rab.bits. (Cf. Table I, p. 21.)

were markedly heavier than the Flemish. Their early growth was
much more rapid and their sexual maturity much earlier. The Flemish
animals which provided the data for Fig. 7 had been inbred for several
generations (cf Fig. 12, p. 20), and it is not impossible that inbreeding
may be connected with slow growth and delayed sexual maturity. Early
maturity is not necessarily connected with small size as is shewn

R. C. Ptonrtt and thk latk P. G. Baii.ky 11

graphically by the record of thrcn* Polish nibbilH iti Fig. 8. T\\v J (QH^l)
was over 10 tnonthM old lH»lon' any doclino of weight wt in, while neither
of the ffff attAiniMl their full weight nntil U and 12 nionthH r»'Hj)eo-
tivvly. Here again sexual maturity was late. Though the ihs* was run
with the buck sineral times earlier it was imt until she was 1 1 months
old that she priKiuccd a litter. It is worthy of note that these aninwils
are the pnxluct of certainly two genenitions of brother x sister mating.
Whether the inbreeding extended further bark we are unable t<i say.

Wright
mlbi

3-

1

— T 1 1 1 \ 1 \

7 8 9 10 11 12 13

Fig. 8. Growth curves for 3 Polish rahbits.

A further point of interest concerns the growth of the F., animals from
one of the Flemish-Polish crosses, viz. that from the Flemish $ 0 2().S.
Growth curves were made for all of these animals and it was found that
the period in which maximum growth was reached before the temiK)rary
decline set in varied from 8-13 month.s. The distribution of these
animals over the different periods is shewn in Fig. 9. In no cjuse wjus
maturity reached as early as in the F^ parents, while in several ca.ses it
was delayed until the age of 13 months (cf. Fig. 10). As might have
been expected the smaller F., animals mature on the whole earlier than
the larger one.s. The average weight for those maturing at 8 months
is 4 lbs. 2^ oz. and this increa.ses fairly regularly, having regaixl to the
paucity of numbers, up to the average weight of 5 Ib.s. 1^ (►z. for those
maturing at 13 months. Neverthele.ss a consideration of average weights
does not in this c;ise tell the whole story. Fairly large F.^ animals may

12

Genetic Shtdies m Babbits

mature early, e.g. Q 75 which reached 4 lbs. 8 oz. in 8 months ; while on
the other hand a small rabbit may mature late, e.g. QIS which took
12 months to reach the weight of 3 lbs. 12 oz.

That early maturity is to some extent independent of size is an
inference borne out by the results of the Flemish-mixed cross. The
distribution of the ages of maturity in relation to weight for three

montnt

Fig. 9.

1 1 \ 1 1 1 1 1 1 r

3 4 5 6 7 8 9 10 n 12 13

Growth curves for 3 F^ animals from Flemish-Polish cross. (Cf. Table V, p. 25.

1
14

. I llii

8 9 10 n 12 13
4..2^2 4..4 4..4 4..11 4..6 5.-1^

Fig. 10. Shewing distribution of weight in relation
to age of maturity in Fo, animals from Flemish-
Polish cross. (Cf. Table V, p. 25.)

R C. Pdnnktt and tiik latk P. G. Bailky

13

g«nenitioii8 is shitwn in Fig. 11. No n-conlH wen? kept «»f the nuiturity
age of the mixe<1 Htniin which wa« uh<'<I. X 17 ami X I!) hml their finit
littt^re at 11 and 10 months reH|H»otively, while other tUns in the saine
strain first littered at 9-11 uionthn. The avenige maturity age of the
strain was prolwibly 2-iJ moiithH earlier than that of th«' Flrmish. Th«*
Fi luiimals are notably later in maturing than those from the Kli'inish-
Polish cnws, though three uf them wen* distinctly in iwlvancc; of the n-st.

9
7.6

10

11 12 13
7.15 6.14

14

I { I I

8
7.5

9 10 11 12 13 14
6.3 7.2 6.15 6.8 712 6.10

9 10 11 12
5.81^ 5.0 5.8 5.14

13

14

Fig. 11. Shewing distribution of weight in relati«»n to age
of maturity in Fi, F-,, and F3aniinalg of Flemish-mixed
cross. (Cf. Tables II— IV, pp. 22—24.)

The Fi generation, which w;is raised from among the later maturing (»t
the Fi animals, shews considerable scatter, maturity in some c<ises being
attained as early ;vs 8 months, while in others it is fully jus lat<' as in
the Flemish (cf. Fig. 11). The curve appears to be a bimodal one, but
the small numbers and the necessarily rough method of classiHwition

14 Genetic Studies in Rabbits

hardly warrant us in attaching much significance to this feature.
Greater interest attaches to the fact that the maturity age here is less
dependent upon weight than in the case of the Flemish-Polish cross.
It is true that the group of lowest average weight (6 lbs. 3 oz.) matured
early and that the heaviest group (7 lbs. 12 oz.) matured late, but apart
from these groups there is no tendency in earlier maturity being asso-
ciated with lower weight. An association of increased weight with later
maturity is rather more marked in the case of the F.^ generation, but
the fewer numbers together with the smaller range of variation both
detract from its significance.

The possibility that weight might be affected by such features as
size of litter, age of doe, and season of birth naturally occurred to us,
but examination of our records from this point of view has not led to
any positive result. In this respect we agree with MacDowell ('14,
pp. 27-28) and we do not think it necessary to go more fully into an
analysis of our data in this connection. Castle ('09, p. 39) considers that
variation in size is to some extent dependent upon variation in the
quality of food. Davies ('17) on the other hand holds that, provided the
food supply is adequate, size is a matter of heredity. Our general
impression is that the last-named author is right, but we recognise that
the point can only be definitely settled by definite experiment. We have
endeavoured to keep our animals as far as possible under uniform con-
ditions throughout the year. The hutches are all in a large shed free
from draughts and warmed in very cold weather. Until quite recently
the animals have always had a liberal allowance of oats together with
such green food, roots, etc. as the season affords. They are bedded on
sawdust and hay and we have hitherto been very free from sickness — a
feature which is at once brought out by the growth curves (c£ McDowell,
'14, p. 43).

On the whole our experience suggests the following tentative con-
clusions with regard to rate of growth and maturity.

(1) The growth curve normally shews a steady rise, gradually
flattening with age. About the period of sexual maturity it usually
receives a definite check. As a rule this is temporary and the animal
subsequently becomes somewhat heavier than it was at maturity^

(2) Though animals belonging to large breeds may mature more
slowly than those belonging to small breeds it does not follow that age

1 In the rat growth is apparently continuous, and there would appear to be no definite
check at sexual maturity (cf. Donaldson, '15, pp. 66 and 67).

R C. PUNNBTT AND TIIK LATK P. il. HaII.KV 15

uf maturity is cU>Ht»ly com»laUHl with hv/.v, Thr vt-ry Hinall PoHmH mhhit
inattm'8 slowly — |in»hably a ginnl t\vi\\ mnn- hIowIv than a lar^rr fonn
such as the Dutch.

(3) Sixe and early maturity an ' to sunu* rxti'iit traiif*ii»itt<<l iinl«'-
pi'mleiitly. Karly maturing lar^t* aniuials i\h wrlj ]im latr iiiiitiiriti)^ Miiall
*»iie8 may occur U>^ethiT in the same F., family after a <-rosM.

(4) CiXKSsing (iislincl biveds n»ay in some vast's ha*! to \» ly rarly
maturity in F^. In other ouses this (Uh's not appear to In- tiJu-.

The data taken t^j^ether su^j^est that early maturity may <lf|H'nd
u|>on st)mi» factor, or factors, indrjK'ndcnt <>f size, thnu;(h probably a
larger nibbit wouM mature later than a smaller one where both wen-
similarly eonstitut<?d with regaixl to th»' sjK'ciHc genetic taetoi-s u|Hin
which onsi»t of maturity di'jK'iuled. The [)resent <lat^i are far to<» slender
for [Mjsitive infercnce, and we do not pretend that they oti'er more than
an indicjition of a pn>Ht{ible line of encpiiry for the future.

Sexual Differences.

A. Mature Weig/U. It is well known that in many animals the
male is normally the heavier of the two sexes when mature. This is
probjibly true of the majority of mammals, and there is cxjx'riment.d
evidence that it is also true for certain birds'. It is not, however, tru**
for nibbits. Our data shew that in some ca.ses the average weight of
bucks and does from a given jmir of rabbits is approximately e(jual : in
other cjises the average weight of the doe is markedly greater than that
of the buck. In no Ciise which has come under our notice does the buck
certainly exhibit the marked preponderance which may occur in the doe.
The position may be indicated by the following abstract (Table A) of
the datii contained in Tables I-V.

The figures shew that on the whole there is no marked difference
between the sexes in the Flemish. Indeed they shew an actual e<piality
for the whole of the animals reconled. This equality is however to be
discounteil by the fact that a larger proportion of males than of females
belongs to the earlier generation when the weights for both sexes were
considerably higher (cf. Table I, p. 21). For the later g<'nerations c<»ii-
sidered apart the average wi'ight of the females is rather higher than
that of the males. We are inclined to the view that our original
Flemish were of mixed constitution with regard to weight factoi-s (w hat-
ever these may be) and that the preponderance of weight among the

» Cf. Darwin ('91) ii. p. 281 ; Donaldson (15) ; Phillips ('12) ; I'unnett and Bailey ('li).

16

Genetic Studies in Rabbits

TABLE A.

Males

Number
12

Females

Flemish

Number
13

Average weight
8 lbs. 2 oz.

Average weight
8 lbs. 2 oz.

„ (later generation)

5

7 lbs. 5 oz.

8

7 lbs.

9^oz.

ExiV17x2V169

11

7 lbs. 3 oz.

4 .

, 8 lbs.

13 oz.

ExiV19xiV^169

5

7 lbs. 7 oz.

1

7 lbs.

14 oz.

Above Fi generations together...

16

7 lbs. 4 oz.

5

8 lbs.

10 oz.

Ex 039x041

4

7 lbs. 5 oz.

3

7 lbs.

5 oz.

Ex 2V 230 X iV 233

8

6 lbs. 60Z.

12

7 lbs.

13 oz.

ExiV230xO41

9

6 lbs. 7 oz.

5

6 lbs.

1 oz.

Ex 0 223x0 222

6

6 lbs. 7oz.

12

6 lbs.

7 oz.

Ex 0187x0 192

14

5 lbs. 5 oz.

9

6 lbs.

Ooz.

F2 Flemish-Polish (ex 0 203) ...

17

4 lbs. 6 oz.

19

4 lbs.

8ioz.

(ex 0138)...

5

5 lbs. 4 oz.

6

5 lbs.

14 oz.

males in the earlier generation is accidental. In later generations,
though the numbers are small, the weights generally shew less fluctua-
tion and the females are consistently heavier^

When we come to the Flemish-mixed Fx generation the figures are
striking. The does average nearly 1| lbs. more than the bucks. In
view of the small number of the former it is not unlikely that this
difference would have been diminished had larger numbers been bred.
Nevertheless it is too substantial not to be regarded as having some
significance. The F^ generation is of much interest. In two out of the
four matings, viz. 0 39 x 0 41 and 0 223 x 0 222, the sexes average the
same weight, in one {N 230 x 0 41) the bucks average rather more than
the does, w^hile in the remaining one {N 230 x N 233) the average weight
of the does is very markedly greater than that of the bucks. The only
F^ mating was from two animals, % 0187 and c/ 0 192, both derived
from the F^ generation ex iV230 x iV233. 23 F4 animals were i^ised,
and here again the average weight of the does was markedly greater
than that of the bucks.

In the case of the Flemish-Polish crosses the same phenomenon of
greater average weight among the does was noticed in the F^ generation
ex $ 0 138 X c/' Polish, while in that from J 0 203 x Polish the average
weight of the two sexes was nearly equal. It is possibly of significance
that in the former cross the Flemish doe (0 138) was a large one, and in
the latter a small one (0 203).

Evidently we must recognise that in the rabbit a factor, or factors,
are to be found which bring about an increase of size in the doe as

^ This is supported by the results from a further generation of three Flemish now being
reared. At the age of 10 pionths the c? weighed 6 lbs. 0 oz. , while the two does were
respectively 6 lbs. 8 oz. and 6 lbs. 4 oz.

R C. PUNNBTT ANDTHK LATK P. O. BaILKV 17

compared with the buck. That thin factor (or factont) may be tranH-
iiiitt<?<i by the buck appiMirH pn)l)abK* fnmi th<* rfHullH obUiint><l fn»u»
J A^230. With two brnthcn* of alinoMt exactly the kiuih' wei^'ht, N 2'M\
and 041, she gave in one cjwe a family in which the pniiiindcnitHM; of
weight wjus rather on the side of the bucks and in the <)th<'reji«<' u family
in which the doe» wen* markedly heavier. That the fiu'tor (or factorM)
may also be transmittiMi by the <loe is indiaited by the rcHultM of the
Flemish- Polish cn^sses. With the same Polish (/* the heavier FleinJMh
doe (0 138) gave rise to an F^ generation when; the inarke<l di.Hcrejwinry
in weight between the sexes apjx'ari'd, while with a lighter Flemish J
(0203) the Ft generation shewe<I little dift'erence in the average? weight
of the bucks and does. So far jus they go, the dat;i suggest the o|>era-
tion of a factor which leads to an incre^ise in the weight of the doe that
contains it, though not at all, or only to a slight extent, in that of the
buck. Much more work however must be done before we am arrive at
certainty on this point, but there seems little doubt but that it is a
complication which must be borne in mind in work dealing with the
inheritance of weight in this species, and possibly of others also.

B. Rate of Growth. A marked difference in the rate of growth
between the two sexes has long been recognised in man where stiitistics
bearing upon the point are more plentiful than in other animals. Up to
about 13 years of age the female grows rather more rapidly than the male.
After this point the male overtakes the female and eventually becomes
markedly heavier^ This differential growth is probably connected with
the earlier onset of puberty in the female sex. Our records allow of
some examination of this point in the rabbit, and we have prepared the
following brief abstract (Table B) from the data given in Tables II-V

TABLE B.

Males Females

Number

Average weight '
4 months 12 months

Proix)rtion of

later growth

to 4 months

growth

Number

Average
4 months

1 weight
12 montlis

I'ro|K)rti<»n of

later growth

to 4 months

growth

FiexNnxS169

11

4-2^

7-3

•73

4

4')

813

1-04

FiexN230xN2SS

8

3-9*

6-6

•77

12

3 12

713

1-08

Fj ex .V 230x0 41

7

4-3

6-7i

•54

5

3-8i

6-U

•72

F, ex 0223x0 222

6

4-0

6-7

•49

12

3 15 J

6-rt

•r)3

F3 ex 0 187 X 0 192

8

31U

5-4A

•42

4

40\

r,-2

•53

Fi Polish-Flemish

14

2-3i"

4-6

•99

20

2-3

4-8

11)6

^ According to Darwin (Descent of Man, 1891, p. 283) a fiimilar differential sex K'ruwth
is found in the Scotch deer-hound.

' Under this heading the figure in front of the dtciiual point denotes pounds while
the figure or figures after it denote ounces — e.g. 312 is to be read as 3 lbs. 12 oz.
Joum. of Gen. viii 2

18 Genetic Studies In Rabbits

(pp. 22 — 25). The average weights at 4 months and at 12 months^ are
given separately for the two sexes from matings in which 12 or more
offspring were reared. It is evident that up to the age of 4 months
there is little difference between the sexes. Where the sexes eventually
attain approximately the same weight (e.g. ex 0 223 x 0 222), or where
the male is slightly heavier (e.g. ex iV230 x 041), the weight of the
female at 4 months is below that of the male. Where, as in most of the
other cases, the female eventually reaches a markedly higher weight she
is rather heavier at 4 months. But in all cases the growth subsequent
to 4 months is greater in the feaiale than in the male. This is brought
out in the last column which shows the proportion of the weight at
4 months that is added between 4 and 12 months. In every mating
this proportion is higher in the case of the does.

A point of interest brought out in Table B is that the greater the
weight to which the animals eventually attain the greater appears to be
the proportional increment added after 4 months, for there is not much
difference at 4 months between the larger and the smaller animals in
the Flemish-mixed series of experiments. The data, so far as they go,
suggest that it might be more economical to breed a larger number of
smaller rabbits for killing at about 4 months than to. rear a smaller
number of larger ones. The latter would have to be kept more than
twice as long to gain the advantage of their greater size, and even then
the total weight would be less than double that of an equal number of
smaller ones killed at 4 months.

The results of the Flemish-Polish cross offer an interesting contrast
to those of the Flemish-mixed cross. Though the F^ animals are all on
the small side they are on the whole late in maturing (cf. p. 11). There
is practically no difference between the sexes either at 4 months or at
maturity which occurs at various ages between 8 and 12 months
This applies only to the F^ animals from $ 0 203. Full records of the
smaller F^ generation ex % 0 138 are not available.

One further point of difference in growth between the sexes is
brought out by an examination of the F^ records. It concerns the
extent to which the heavier and lighter animals in such a mixture can
be distinguished at a relatively early age ; or, to put it in another way,
how far the animals heavier at maturity give evidence of this at the age

1 The animals were generally weighed at intervals of a month. Since the litters were
produced irregularly the weight at 4 months sometimes refers to animals a little over and
sometimes a little under this age. Members of the same litter were of course always
weighed at the same age. Weight at 12 months means maximum attained by 12 months.

R C. PUNNKTT AND TIIK LATK l\ ii. HaILKV

19

of 4 months. For testing this |)oint the iivAilahli* n^conis consiHt of
^^<f<f rtiul :iO ?? belonj(in^r to i\\v F., ^M-nonituui of tin- Flnuish-
mixtnl cross, together with U ^^f and 20 ^ ? of thf h\ g«n«THtion
from the Flemish- Polish cross. Both ^f ^f and J J in each wisr havr

Ar«rM«irei«ht

of mMH ttnn«r

< lb*. Sot.

TABLE (

A vermes wnlitht

of malm of

6 Ibi. S OS. and over

AvcnMTc wriKht A veracv ««lcht

of frttuuM under of frmiUM over

7lbii.

7 lb*

At 4 At 13
month* month!

t\ Kenenuiou from
Flemish -mixed

35

61

Averace weicht

of malet under

4 Iba 8 OS.

At 4 At 13

months nionthn

Fi generation from
Flem
cross

^neration from )
Flemish-Polish \ 115^ 41

At 4 • At 12
moothn month*

4-5

7-2

Avenu(o weight

of malee of

4 Iba. 8 OS. and oTcr

At 4

month*

2-9

At 12
month*

4- 12

At 4 At 13

month* months

313^ 0-2

At 4 At 12

month* tmmtlui

313

T-l.-ii

Avvnwce weiKht Avera^rv weljcht

of feniale* under of female* nf

4 Dm. 10 ox 4 Ih*. 10 ox and over

At 4 At 12

month* month*

2-4

4«3

At 4 At 12

montliH month*

2-2 4-14

been divided into two roughly eijual groups, one containing the light<T
and the other the heavier animals. For each group the average weights
at 4 and 12 months respectively are shown in Table C. The result
shews that (^ff which are lighter at maturity are also markedly lighter
at 4 months, but that all the $ ?, heavier as well as lighter, are indis-
tinguishable in size at the earlier age. This result is strengthened by
the fact that it is equally true for the Flemish-mixed and the Flemish-
Polish animals. For the same chiss of material the bucks which are
destined to become heavier start to outstrip the lighter ones at an earlier
age than is the case for the does.

Inbreeding. In so far as our limited experience goes close inbreeding
appears to lead to diminution in weight', and perhaps also to some delay
in the age of maturity. Decrease in weight is most marked in the axsc

* This is contrary to the opinion of Huth ('87) who inbred rabbits for seven generations
to test this point. Though the average weight of his animals at the end of his experiment
was as heavy as at the beginning he does not state the numbers upon which ihese averages
were based. Nor does he give the weights of the parents in any case. His tigures do not
preclade the possibility that he selected the heaviest parents in each generation to breed
from, so that any efifect due to inbreeding might have been masked by unconscious selection
of genetic factors for increased size. That close inbreeding may be accompanied by an
increase in size, provided that the largest and most vigorous animals are selected for
breeding from, is brought out by the recently published experiments of Miss H. D. King
('18) with rats.

2—2

20

Genetic Studies in Babbits

of the F
materia

6

einish as is shewn graphically in Fig. 12. No doubt our origina
was heterogeneous with regard to the genetic factors tha

Cf Nl69

(5*02011
9O203'

• $P« •(5^P,225
228

Fig. 12. Graphic representation of weight distribution in Flemish animals.
(Cf. Table I, p. 21.)

govern size, but even so the rapid diminution can hardly be accounted for
by mere sifting out of genetic factors. Our experience with the Polish,
though even more limited, agrees with that of the Flemish. Experi-
ments are now in progress to ascertain the effects of continued inbreeding
on weight both in these two breeds and in an extracted strain of small
size from the Flemish-mixed cross (cf. Table IV, p. 24). It is hoped
also in this way to obtain homogeneous material which, by a system of
reciprocal crosses, can be used to elucidate the mode of transmission of
size.

The experiments of which an account is given above form part of a
series of investigations on heredity in rabbits for which the means have
been provided out of the Fund controlled by the Development
Commission,

H. C. PUNNRTT AND THK LATK l\ (i. lUlLKV

L>!

LITKRATUHK RKFKKHKI) TO.

Castle, W. E. "Studies of InhcriUnw in KaliMU," Publication N... lU, Com^fi*
/HttitHtiOH of WiutAingfoM^ IIKMI.

Darwin, C. Tkr Ik'^rrnt of Man. in*! Kailinn, isin.

DaVIKB, V. J. '* TIlO (inm-th of HaM.it«. " Tin' Ihiuuir, Krr/„iu;fr and Mnrl, Juin- H,

1917.
D0NALI>8ciX. H. If. n^ flat. Philmlclphin, 11)1.").
East, E. M. " InheritAnoo in C'rosMcs In'twocn .Vtrotinna hntfjtuiorjii hikI Xtrotiana

oUitaS' (JeHftic*y 1917.
HCTH, A. H. Thf Marriagr of X«i,r hi,*. 2\u\ VaUWou, l>»n«U.n, 1887.
Kino, H. D. StiulioH <»n Inbroetlinj;. I. Tho oflr»>ctM in inl>nMMling on the gr«»wtli

and variAlnlity in the Ixxly weight of th<» .ilhino rat. Jtmrn. A>/>. Xo<J. 1!)1h.
MacDoweli^ K. C "Size Inhorit;incc in IlfihhitH," Puhlication No. 1!K>, Carwgif

InMitufioH of WoA^iiufto/i, 1914.
Phillips, J. C. '• Size Inheritatice in Diuks," Jitunt. Evp. Zool. 191 1'.
PUXNKTT, R. C. and Bailky, \\ (I. " On Inhorit^nco <.f Weiglit in Poultry," .hwru.

of Genetic*^ 1914.

TABLE T.

Throaghoat the Tables the figure before the decimal point denotes pounds and that
after the decimal point denotes ounces — e.g. 8*2 represent.s ft lbs. 2 oz.

Ffejnish.

Weight

Weight at
4 months

Weight

Weight at
4 months

<?X169

8- 8

—

? 0 105

9 10

'

,^0134

810

5- 5

? 0 136

915

5-7

<f0135

8-12

5- 5

? 0 138

9- 9(')

6-3

,^0137
c?0139

8- 3

9- 9

5- 8
5 11

? 0 203

713

5-5

V ex 0 105 X \ 169

<f O201

713

4 10

—

—

—

^0202

9- 6

415

—

—

—

<f O204

8- 9

5- 4

—

—

— ,

c? P 61

7- 3

—

? P 56

7-15

— \

<! P 62

7- 0

?P 57
?P 58
?P .59

7- 3
7-U
714

—

^ ex O203x O201

</P 76
<? P 78

8- 1
711

—

9 P 77

8-11

' ex O 13H -: n 139

^ ' 1

S P225

6-10 (-'»

? P226

6-13 (-')

- »

? P227

7- 1 ('■*»

- CXP77 /T.l

? P228

7- 5<--')

- 1

(1) WvvM at l:i

iionilis.

(2) Wf'i^'ht at 14 iiinntli>.

22

Genetic Studies in Rabbits

TABLE II.

F^ generation from Flemish Cross.

Weight

Maturity
age in
months

Weight
4 months

Maturity
age in
Weight months

Weight

at
4 months

SN221

6-

8

11

4- 5

?iV230

8-3 11

4-14

dN228

7-

4

12

4- 8

?0 37

9-4 11

4- 0

SN229

7-

0

12

4- 4

?0 39

8-6 9

4- 6

SN2S1

6-

7

12

4- 0

?0 40

9-7 11

4- 0

c? N 232

6-

6

12

4- 3

criV233

7-

8

11

4- 0

> exiV17(°)xi^l69

cTO 34

7-

6

9

4- 4

[

JO 35

7-

7-

11

3-10

c?0 36

7-

9

11

4- 5

c?0 38

8-

3

—

4- 3

jO 41

7-

9

—

4- 2

<?iV235

8-

3

11

4-13

?0223

7-14 (^) —

5-4 >

criV237

7-

6

12

4-10

c?i^238

7-10

1]

4- 6

- ex N 19(6) xA^ 169

c?0 222

7-10(3) —

4- 8

•

J 0 224

6-

6

9

410

(3) Weight at 21 months.

(4) Weight at 21 months.

(5) Weight = 6-2.

(6) Weight = 6-6.

Weight at 9 months was = 7*3.
Weight at 6 months = 6-14.

H C. PUNNRTT AND TIIK LATK \\ (I. HaILKV

L>:{

TABLE III.

F^ (f^iuf rat iou from Fl^mixh mmn.

Wrltht

Maturitr

IIHMtlhl

WrJglil

at
1 luoiilh*

WrIglH

Maturity
acr III

lliotltlia

Wi.l«ht
at

i motiili*

^0195

7 18

10

4- 0

?P 75

Gil

'.1

\

<f0197

C- 3

10

3 la

f <t 19G

<;-i5

10

4- G 1

,rO 198

711

10

4- 1

9 0 200

8- C

10

3 13

ex O 39 .

<rO 199

7- 9r» 10

4- 0

)

,fO107

5- 7

13

2 12

9 O KNl

7- G

13

3- 7

^0109

G- 2

M

3- 13

9 O 108

8- 2

13

4- 3

(TO 111

0- 4

H

313

? (> 110

8- li

13

3 14

^Olli

5 12

13

3 7

9 (> 113

9- 0

13

4 2

<^Ol91

6- 9

10

3- 4

9 0 114

8- 2

10

4- 1

rf0192
,{0193

G- 0

7 15

12
13

3- 3

4- 1

9 <> 187
? O 189

G- G

8- 2

12
13

3 1 4
4- 3

^ rx \ 2:40

iP 51

G12

10

4- G

?0 190
9P 52
?P 53
9P 54

9 /' 55

8 13
714
7- G
7- G
7- 5

10
12
10
10
12

4 4

3- 4
3- G

3- :i
.{• 2 '

<fP 97

7- 6

10

4 15

9 P 99

5-10

12

3- 7

SP 98

GIO

12

4- 3

V P 130

5- G

12

3- G

<fP100

6- 3

12

3 14

9 P 131

7- 9

12

3- 11

<fP127

6- 4

10

4- 3

9 P 132

G- 8

10

3 10

<fP128

6- 4

10

4- 0

9 P 134

5- 7

10

3- 8

■ ex \ 230

,?P129

6- 2

10

315

cfP133

6- 8

12

4- 4

<fP204

515

9

—

dP2U

6- 9

11

-

iP 82

G14

12

314

9P 81

G- C

12

3- 8 .

c?P103

510

10

314

9P 83

G- 5

12

313

^P104

6- 7

10

4- 1

9P 85

G -5

12

3- G

<fP105

614

10

4- 8

9P106

6- 8

11

4- 0

,rP107

515

12

4- C

9 P 108

7 12

11

4- 7

cfPlGO

614

9

510

? P 159
9P198
? P 199

G12
G- 9
5- 3

10

12

9

5- 1
3 14
3- G

> ex 0 223

9 r 200

6- 9(') U

4 1

9 P 201

G-12

U

4- 5

9 P 202

G- 3

10

312

9 P 203

G- 4

12

4 1

041

\ 233

0 41

(7) Weight at 10^ months; cf. brother, <) 198, wlioso \v»iglit ihci not clmn^e b«t\v«<n
lOi— 12 months.

(8) Weight at 14 months.

24

Genetic Studies in Rabbits

TABLE

IV.

Fg geiieration from Flemish cross

Maturity
s>g& in

Weight
at

Maturity Weight

Weight

months

4 months

Weight

age in
months

at
4 months

c?P 90

5- 9

12

—

?P 88

5- 9

12

_ >^

<fP 94

6- 1

11

—

?P 89

5-13

12

<?P121

6- 0

11

3-13

?P 91

5- 9

11

c?P122

6- 4

11

3- 7

?P 92

6-13

12

JP123

5- 5

11

—

?P 93

514

12

S P 124

5-15

11

—

?P120

6-15

12

3-4

<?P125

4-14

11

—

?P189

4-15

11

4-4

crPi26

413

12

—

?P190

6- 7

9

4-4

- ex 0 187 X 0 192

cfP192

5- 3

9

3-15

?P191

6- 2

9

4-6

c?P193

5- 3

10

311

(^P194

412

9

3- 7

J P 195

5- 3

9

4- 3

•

crP196

4 12

11

3- 6

cfP197

4-13

10

3-13

,?P233

Weight
5-2

/^4 generation from Flemish cross.

Weight at
4 months

3-0

?P234

Weight at
Weight 4 months

5-6 3-1 } exP189xP196

cTQ 79
jg80

Weight

F^ generation from Flemish cross.

Weight at
4 months

2- 8

214

Weight

?Q78
?<?81

Weight at
4 months

214
2

■■1\

ex P 234 X P 233

R. C. PUNNRTT AND THK LATK P. G. BaILKY

2r)

TABf.K

V.

Flemish- Pot i*h crot§.

Wi.lfl,t

W«l«ht kt
4 niontht

WnlKht

W.l«ht»l
4 monUia

dPlU

4*13

4*2

?/M15

5-9

*"*^ I - /».1A« . fl

iPin

«• 2

4 1

?riiH

6-8

3.j^^exO/03x^/'

4Q%U

6- 7W

—

VO190

G-14(»»

~ \ ex <) 138 y J I

h\ (jenerafion

Welfht

MaturitT

•ire In
niontha

Weight

»t

4 month!

Weight

III

Weight

•t

4 nionthN

iPt2S

5- 4

13

3- 6

9Q 10

410

11

1- 6 \

i P 224

4 12

13

212

?<^11

4-13

11

Ml

iQ 9

311

<>

1- 3

?gi2

4 13

11

Ml

iQ U

4- 2

10

1- 7

?g 13

4- 0

10

MO

iQ 25

4- 2

11

—

? Q 20

4- 3

12

' vx V 115/ /'

iQ 27

4- 5

10

—

?(^37

3 12

10

114

,fQ 3a

4- 8

9

114

? V3H

4- H

10

2- 3

<fQ 76

315

8

2- 3

?g75
? V77

4- 8
4- 5

H
10

2- 5

M2 ^

JP217

4- 2

12

2- 8

?g 3

410

10

2- 1 >

<fP218

4 12

13

2- 9

9^4

4-10

10

2- 0

^Q 1

4- 5

9

2- 1

? (^ 5

5- 5

12

2- 7

^Q 2

4- 8

9

111

?(? G

4- 1

8

210

«^<? 15

3- 12

12

2- 6

?<^ 7

4- 2

10

2- 1

dQ IS
<tQ 19

4- 8
4- 7

12
10

2- 5
2- 1

?(^ 8
v<.> IC

4- 5
4- 5

12
10

2- 3
214

- ex P 118x7^

iQ 20

415

11

212

?(^17
? (^210«)
?Q22
?<;>55
?(?57

315
5-10
415
4-11
4- 5

10
13
11
11
9

210
2 13
2 10
2- 4
2 13

ill 36

5- 1

? //39

G-10

_

_

dH 37

0- 7

—

—

?//40

G- 2

—

iH 38
^H41

414
415

—

—

? // 42
?H44

6- 4
6- 0

—

—

w ex G 190 X G

,;H 43

5- 1

—

—

?H4o
? 7/4(3

5- 9
4-12

—

—

(9) Weight at 10.^ months.
(10) Weight at 13 months = 514.

14 ,, =511.

15 ,, =510.

NOTE ON THE ORIGIN OF A MUTATION IN
THE SWEET PEA

By R. C. PUNNETT. F.U.S.

(With One Tcxt-fi^nre.)

Many instances of the sudden appeanince of new forms in plants
and animals have been recorded in recent years, and spt'culation has
been rife Jis to the moment at which they may be regarded jus havin^^
originated. Perhaps the view most favoured is that the new form takes
its origin from some abnormal division during the f(»nnati()n of the
gametes. Nevertheless there are biologists who havi' placed on rcconl
their opinion that it may occur at some other stage in the life-history
of the form that exhibits the new character'. The princij^il ditticulty
in coming to any decision on this point is that in almost all cases on
record the new character has not been first observed in accurately pedi-
greed stock. After observation it has frequently been made the subject
of careful experiments in order to test its genetic nature, but this of
course does not help us with the problem of its origin. Even in Droso-
phila, with its century of mutants, there does not appear to hv a case
where the new^ form can be traced backwards through definite indi-
viduals for several generations. For this reason I have thought it worth
placing on record the following facts in connection with the apjx'ai-ance
of a new form of sweet pea in pedigree cultures. The form in question
is the so-called " cretin," already described by Mr Bateson and myself
in an earlier number of this Journal'-. It is a monstrous form of which
the chief characteristic is the straight stigma protruding through tiie

' See more especially Johannsen, /r« C'oH/e;rnf« InteruatiouaU de Geni^tique, Paris,
1911, aud Emerson, The American Saturalixt, June 1913. A Kood general discussion on
the subject is to be found in Baur's EittJUhruug in die erperimentelle Ver>rhungnUhre,
2 Auf. 1914, pp. 288 seq.

- Journnl cj GenetirK, Vol. i. 1911.

28 The Origin of a Mutation i7i the Siveet Pea

cleft keel (cf. Fig. 1). The standard and wings are generally smaller
than in the normal flower and fail to expand fully, but in these respects
a good deal of variation is to be found. The cretin is however always
characterised by one other feature ; it is invariably sterile on the female
side. The fact that this peculiar form appeared as a single individual
in a pedigree culture has already been recorded. Data have also been
given to shew that it behaves as a simple recessive to the normal form^
There arises the question whether the evidence is consonant with the
view that the original mutation occurred in the maturation divisions of
the germ cells, or at some other stage. To attempt to answer it involves
a consideration of all the details connected with the coming of the

Fig. 1. Two flowers of the original cretin plant, No. 14618/1907.
Flowers of other cretins are figured on PI. XL, Journ. Gen.
Vol. I. 1911.

cretin, in so far as they are known, together with those of its subsequent
behaviour.

In 1903 a cross was made between two white sweet peas, Blanche
Burpee (long pollen) and Emily Henderson (round pollen)^ From 3
purple F^ plants three large F^ families were raised in 1905. From one
of these F^ families, No. 309 containing 187 plants, the seed of 29 in-
dividuals was saved to give an F.;. generation. These 29 families were
raised in 1906 and resulted in 2083 individuals all of which were normal

1 Journal of Genetics, Vol. i, 1911 ; ibid. vol. iii. 1913.

^ Cf. Bep. Evol. Gomm. Roy. Soc. iv. 1908, pp. 9 seq. where details will be found of the
various characters entering into the cross.

R (\ PUXXKTT IMI

(cf. Tiihli' I). Krwiii oiu' of the F^ fninilii's, N<». .'M)4 conuiiiiiii^r IM planth.
uochI Will* mivckI fn>iii W iiulividiialM to ^ivf an /*, ^^ciicratioii. From
thctk* 14> plants then* ivsuIIimI an b\ ^rnrration ol Ills individiiaU ol
which all wen.' normal siive ont* (rf. Tahle I). Thf txcrption wiw th«-
original cn'tin which appvanMl in a family of .j'i plants niintMl from th«

TAHLK 1.

Roxml NmloT No irf Kvoml N«m of N.i. of Ko^inl i:«-ooii| No iif

No» erf ilcrlwd r, |»Uatii No« of lU-rivwl A, pUnU N.m of Niw. of pUnta

/*, |4atiU familial in >,, i>lMit« fAiiillitw In AmiUiiIa <lciiTei| A\ In

Vm 1906 faniily VMt, I<«u7 fmnlly I'J07 faniillca fftmllr

309 » 801 159 - _ ._ _

— « .Wi 121 — — — _ „

— « ;^ 60 — — - _ _

— • H02 (11 - - —

304 181 304^ 140 62*

.T

— ^ aG:i :r» • H7 ir.

— I' ao.'i 12(; — ^ us H7
— »• :W6 (»i — «" i4'» Ha
— «« 307 ♦;7 " IV) '

— " 360 6« — '- lol ' 12H

— " 308 5'.» — »* 152 W.)

— >« 368 2»; — >■ 153 242

— •' m\) 82 — '^ i:)4 (H

— "* 30*) 25 — •*• 155 30

— •» 370 21 — a« 150 23

— » 380 24 —^ 157 90

— « 310 54 —-J-' 158 -20

— *s 381 20 —^ 159 48

— » 371 62

40 «

111

1

— •'•

145

6

_I0

110

26

_. II

147

01

__I2

TH

r.'i

— It

ll'.l

•>

_-H

150

2

— I'.

151

1

17

152

27

Total 195

— »* 311 20

— =" 372 190

— » 373 25

— *» 374 193

— » 375 57

— » 313 75

— ^1 382 39

— » 376 49

— » 377 57

— » 380 57

Total 1118

Total 2083

• Family in wLich the cretin appeaieil. The record number of the cretin was 116"*.
Forther details as to the nature of the i'2, F^, and t\ families will be found in Report IV
to the Ecolution Committee of the lioijal Soiieti/, pp. 14, 15.

F:, plant 304V1906. This family of 52 \va.s iiumlKred U(j in HK)7. The
appearance of the cretin 1»'(| to the sjiving of sn'd i'vom normal indi-
viduals of this family, but since many <if them had been pulled uj) before

30 TJte Origin of a Mutation in the Sweet Pea

the cretin was discovered the numbers saved were fewer than could
have been wished. For various reasons the sowing of these seeds was
postponed until 1912. They germinated poorly and the F^ families from
9 plants consisted of but 195 individuals (cf Table I). No cretin how-
ever was found among them. Seed was collected from the four plants
Nos. 145-148 and sown in 1913. The 135 plants which resulted were
all normal. The series of experiments was not continued beyond the F^
generation.

The cretin then had its origin in a single seed of the F^ plant
No. 30471 906 and was the only case of its kind in a family of 52 plants.
None of the 13 sister plants of 304^ produced a cretin among a progeny
of over 1000, nor did such a plant appear in the large F2, generation
of 2083 individuals of which 304*^ was a member. Though the F^ gene-
ration raised from the sister plants of the cretin was not large, yet four
of the families were certainly of sufficient size to have produced cretins
had they been heterozygous for this simple recessive character. The
evidence taken together renders it unlikely that the origin of the cretin
was due to the meeting of two germ cells which had each lost the
normal factor. Were the mutation of germinal origin we should be
inclined to place its occurrence in the parent plant of 304**, viz. in the
F^ plant 309^ and we should have expected cretins to form about 25 ^\^
of the family in which they first appeared. Again we should have
looked for their further appearance in some of the F^ families grown
from sister plants of the cretin itself We are led therefore to suppose
that the appearance of this peculiar form is due to a change in the in-
dividual at some stage after fertilisation whereby the factor for the
normal flower was either dropped out or altered during the somatic
divisions.

It has been assumed that the cretin always behaves as a simple
recessive, and some evidence has already been published in support of
this assumption. More extended experience during the past few years
has served to confirm this view. Crosses between cretins and normals
of various families have been carried to the F^ and F^ generations and
in no case is there any reason for supposing that the cretin behaves
otherwise than as a simple recessive. A brief summary of the results
is given in Table II. Over a period of seven years 80 families have been
bred in which cretins occurred. Out of 5520 plants recorded in these
families 4198 were normal and 1322 were cretins — a proportion not far
removed from the expected ratio 3:1. There is therefore no reason for
supposing that the difference between the cretin and the normal is other

RC.PrNNKTT :n

than thiit of a single facUtr, nm) thin, taken in runjiinction with the cir-
cnniHliuict'M uniler which it u\iu\v it-n u|>|H'iinuir«>. Hii|)|H>rts ihc view thiit
the origintil plant anm.\ not through the uni«>n nftwo ^^Tni rcIlM which

TAHLK II.

Y«»r wlicn No «»f N«». of No. of No of

(niwn |»Unt* nomuiU rrBtin* famlltv*

1910 > G40 4Hn IM 'J

1913 « Hi:> .-iiK) 'i-i.-V IJ

1913' lOae 7.W '270 2()

1914 711 502 149 '2

l9lo 14.V) 1124 331 2^

1910 H73 080 193 11

ToUIr ... 6520 419H 1322 HO

KxpeetdtioH nt^f i:iHU

hafi K>st the normal fjictor, but through sonir nwlical alt<r;iti»Mi in th<*
zygote aft^'r union between two nornial gametes had already taken
place.

' Recorded in Journal of Geiwtict, Vol. i. 1911, p. 29.'>.

' Recorded in Journal of Genetics, Vol. in. 1913, pp. 102, 103.

ON HYBRIDISATION OF 80MK SPWJIKS
OF SALLW

By S. IKKNO. F.M.L.S.

(With Plate I and One Tcxt-H^nin'.)

CONTKNTS.

PA<»K

Introduction 33

I. Methods of Investigations 34

II. Breeding Ex(>eriment8 3r»

A. Results of the HybridiaatioHH done in 1910 M

{a) Habit of Stem 37

(6) Hairiness of Leaves 3h

(c) Stipules 40

(rf) Colour of Stigma 41

(e) Character of Catkins 41

B. llesults of the Hybridisations done in 1911 ... 51

III. Conclusions o6

Explanation of Plate I 57

Table showing Successes and Failures of Hybridisations of Various

Species of Salix 58

Introduction.

The well-known work of Max Wichura on the hybridisiition of
Salix- \fill ever remain the master-piece of investigations of such kind,
but as its publication dates back to more than fifty years ago, and as
since then no extensive researches on the hybrids of this genus have
ever been undertaken by any one, except some experiments made by
Kemer von Marilaun for comparing the date of first Howering of parents

* Some of the results contained in this paper were published in /><-/. M<i<j. Tokyo,
Vol. XXX. 1916, pp. 316—320.

2 Die Bastardbe/ruchtung im Pjianzenreich erViutert an den iJasturdcn der U'eiden.
Breslau, 1865.

Joum. of Gen. viii 3

34 On Hyh'idisation of some Species of Salix

and hybrids^, it would not be without interest to study them from the
standpoint of modern genetical science. Thus very often the view
has been expressed that hybrids between various species of this genus
always breed true in later generations. The objects of my experiments,
which were begun in 1910 and continued till now, were to ascertain,
first whether segregation of characters takes place, and secondly whether
m this process, if it occurs, the Mendelian ratio can be detected.
As stated below, since willows come to flowering only in their third
year or even still later, the completion of the work will naturally
require a very large number of years, and if I continue it, I must
necessarily leave a very large space of my experiment-garden occupied by
hundreds of plants for a long duration of time. Thus, some inconclusive
results contained in this paper could not be brought to a definite end
without the cultivation of a considerably larger, number of plants,
perhaps ten times as many as those already raised. This is quite
impossible for the present author, who has only a small piece of land at
his disposal and has yet to perform there much other work. So it was
decided to discontinue the experiments on Salix, except as regards
certain points, and to publish the results so far obtained. Although,
as above noticed, the breeding experiments dealt with in this paper
were commenced in 1910 they contain many imperfections, and are
very far from being complete, so that this paper may be perhaps
regarded as a sort of preliminary communication.

I. Methods of Investigations.

Flowers of various species of Salix open in Tokyo generally at the
end of February and at the beginning of March, though in some, such
as >S^. Caprea, S. triandra var. nipponica, they open only at the end of
April.

For the experiments of hybridisation male branches were cut off
and brought to a warm room, a few days before the opening of flowers,
and placed with their bases in a bottle full of water. When some
flowers began to open and to shed pollen I collected the latter in a Petri
dish by rubbing inflorescences with a hair pencil. As several days are
wanted for the opening of all flowers in one branch I repeated the
process every day till all flowers opened and began to shrivel. The
Petri dish with pollen was preserved in a cold, dark place. Pollen-

1 Plunzeiilcben, Bd ii. 2 Ausg. 1905, p. 510.

8. Ikkno 35

grains will remain eflri«ctivo for a very lon^ tiinf, hh Wichura hiu*
already indic4it<Ml».

Branches* with fiMnalo iiirtnrt»Mcencf.M \v<'n* i'IK'Ios<k1 in a |Ni|><r haj(
two or thriH* days bi»fore their flowers Ix'^an t<» o|h'Ii. Whrii thin Untk
place the inflorescomvs wrre nihhcd with a hair-|H'nril rovm'd with
pollen, and thisprow'^s was rejK'at^'d fors<'Vi'nil dayH till all <)|H'nril Howitm
had been dusted with it. Tho pajKT Imi^ w^ih rnnovrd when the
pnHruding stiginius of all flowt'rs had shrivelled and there wjis n<» more
danger of con Uunination by undesired |x»IIen. If hybridisjition HticccedH
the stigintis shrivel aft^T two or throe days, and the ovaries Ix-gin to
swell gradually; but if not, the stignuus remain jx»rfectly fresh for many
days, and the Ciitkins finally fall off. Fruits are generally rifH' in May.
Seeds were sown in a pt)t immediately aft<'r their collection, because, as
Wichura hjis already shown', they very soon lose their germinating
power. Seeds begin to germinate after two or three days, and seedlings
grow fairly rapidly. In March of the next year they are transferre<l
from the pot to the earth. Many of them come to flower in their thin I
year but sometimes much later.

II. Breeding Experiments.

Though the hybridisations between several species of Salix cultivated
in our Botanical Garden in Komaba near Tokyo were performed with
success*, I shall deal here chiefly with that between Salix purpurea
var. multinervis (below designated simply as S. multinervis, Japanese
name = Inukoriyanagi) and S. gracilistyla (Japanese name = Xeko-
yanagi)*.

These two species grow wqld in the vicinity of Tokyo, and in 1910
I found one female plant of «S. multinervis and one male of S. gracilistyla
cultivated in our Botanical Garden which had some time previously
been transplanted from their respective natural localities. In that
year I effected the hybridisation between these two plants, a process
which seems to take place with great difficulty, because though I have
dusted many female catkins with protruding stigmas repeatedly with an

* Wichura, /. c. p. 5.
« L. c. p. fi.

' The table placed at the end of this paper will indicate all hybridisations done by me,
with their respective successes and failures.

* As to the scientific names of these two .S'a/ix- species the opinions of our systt-matists
do not accord with each other, and I do not know whether or not the nanus employed in
this paper are really the correct ones.

3—2

36 On Hyhridisation of some Species of Salix

abundant quantity of pollen, the large majority of ovaries did not come
to maturity, and I got only a few good seeds, which have given rise to
fourteen jPi-individuals. In 1911 the same hybridisation was repeated
on the same female tree used in 1910, and I got almost fifty seedlings.
The result of the latter hybridisation was entirely different from that
in 1910, as described later in this paper (p. 51 ff.). The reciprocal
hybridisation was not done in 1910, for the female plant of 8. gracilistyla
was not then available in our Botanical Garden. That year however
I brought some branches of its female plant from a wild growing
locality, some twenty miles away from Tokyo ; they were cultivated as
cuttings, and came to flowering in 1912. One of these female individuals
was hybridised in 1912 with the male plant of >S^. multiyiervis. This
hybridisation failed, and nq ripe seeds were obtained. In 1 918 the two
reciprocal hybridisations were repeated, and as I was able to obtain a
certain number of seedlings from one of them, S. multinervis % x S. graci-
listyla </ , my experiments wdll, as far as possible, be continued on these
plants. The other, S. gi'acilistyla ^ x S multinervis c/" , failed again ; it
seems to be well-nigh impossible \

A. Results of the Hybridisations done in 1910.

The fourteen individuals obtained as the result of the hybridisation
performed in 1910 began to flower in 1912, and were found to consist of
ten females and four males which show clearly the hybrid nature in
their vegetative organs as well as in their catkins.

The corresponding characters of the two parents and the F^ plants
dealt with in this paper are as follows :

S. mnltine7'vis. Stem and branches erect. Leaves glabrous on both
surfaces^ without stipules (cf Text-fig. 1 B). Catkin, either male or
female, sparingly hairy (Plate I, fig. 2 J^ — $ ). Stigma bright scarlet-
coloured.

8. gracilistyla. Stem and branches spreading. Leaves hairy on
the lower surface, especially along mid- and side-veins, stipulate (cf.

^ In both 1912 and 1918 a number of female catkins of *S'. gracilistyla were repeatedly
dusted with plenty of pollen of S. multinervis, yet only very few ovaries in each catkin (only
1 — 10 out of some 400 ones in one catkin) have swollen to a certain extent, and after a
few days all of these catkins have fallen off. In 1918 however only one catkin remained
on the tree, and some 10 small fruits on it came to open, but were found to contain a few
tiny seed-rudiments with seed-hairs normally developed.

2 Except few, very young leaves yet inrolled in the bud, which are very sparingly
hairy.

8. Ikkno 37

Text-fig. 1 A). Cntkin, fithrr nmie <»r ftnmlo. (i<'nH4|y oovmKi with long
gray hairs (Plat** I, fig. 1 </••—? ). Stigma ^wvu.

Fi plant. St^in ami hranrhfs spn^nding. I><uvtM glahnmH dti Imth
8iirfact»8, cither with or withnut stiptiirs. Catkin, mostly dmHily hairy
as in the one p»»ront (Plato i. fig. .*{ ^f J ). or s|Minngly ho jls in tho
other (Plate I, fig. 4 c/" — J ). acronling to individuals. Stigma Hcarlot-
colourcd.

Below I will compare with each other the characters above enumer-
ated in the P-, F^- and i^^-generations respectively.

(a) Habit of Stem.

S. multinervis has an erect stem with its branches directed upwanls,
as is usual in many trees, while in S. gracili.sti/hi tin- stem ramifies noar
to the base into many slender branches, which are directed horizontally,
and oflen lie ujxjn the gi-ound. This fact causes a great dit!"erence in
the habit of the two species: the former is erect and high, whih* the
latter is low ; but as in the latter the branches spread horizontally in
all directions it comes to occupy a uiuch larger spact- of ground than

38 On Hyhridisatioii of some Species o/Salix

the former. All F^ plants are at first exactly similar to 8. gracilistyla
in this respect, so that the spreading habit may then be considered to
be dominant to the erect, but as the tree becomes older many branches
which are directed upwards are produced as in S, multinervis, thus
these older plants have become intermediate between the two parents.
From these F^ plants I obtained 442 in the F^ generation, of which 224
are spreading and 218 erect. The fact that the erect habit is a pure
recessive character against the spreading has been also proven in the
following way : I fertilised an extracted erect F2 'plant by the original
erect parent (= S. multinervis), and found the offspring, numbering 89 in
all, to be erect without exception. The segregation of the characters
under consideration in F2 generation has thus been clearly shewn, but
the ratio of the two kinds of plants is very different from what we
might have expected had this segregation taken place in the usual
Mendelian fashion.

(b) Hairiness of Leaves.

In >S^. multinervis the leaves are quite glabrous on both surfaces,
whilst in 8. gracilistyla they are at first hairy on both surfaces, and
throughout their life on the lower, being more or less densely covered
with long gray hairs, especially along the veins. Leaves of F^ plants
resemble the former entirely in this respect, because they are perfectly
glabrous on both surfaces, so that the non-hairiness may be considered
to be dominant to the hairiness ^ In the F^, generation I got 425 plants
in all, of which 35 1 have perfectly glabrous leaves like the one parent,
and 74 leaves more or less hairy underneath like the other. It may
here be remarked that among the hairy leaves the degree of hairiness
is very different in different individuals, and that I was unable to find
even a single plant which is so densely hairy as in the one parent,
S. gracilistyla. This is perhaps due to the fact that we have here a
large number of factors concerned in the hair-production, and it seems
not unlikely that the cultivation of a much larger number of F^ plants
may give rise to a certain number of such leaves which are as hairy as
in >S'. gracilistyla.

In respect to the character "hairiness of leaves" I will mention
here the hybrid between S. m.ultinervis and S. viminalis^ The former
has quite glabrous leaves as above stated, while the latter, which is
found here as well as in Europe, has, as is well known, leaves very

1 Some of the inner leaves in buds, and sometimes a few very small ones crowded
together at the base of young branches, are more or less hairy.

8. Ikkno 39

densely covertxl iin<lemeat.h with hmg Imiix «»f HJIvory lustro. The >',
hybrids made in either of the two reciprocal wiiyH have leavcH which are
always hairy undenieath but much Icms densely so than in S. viminaliH.
In the F^ j(eneniti«>n fi-oin N. multinervvi x S. viminnJis I wjih able t-o ^vi
6nly 7G planU in all. Of these 'M have leaves <|uite glabrous as in the
one parent, whilst the ivmaining 45 were hairy in various degrees, and
of the latter one plant was found to |)os8i«ss l«»aves which were jw densely
hair)' as in N. viminalis.

The segn'giition of the character "hairiness" is acconlingly (piite
evident in both cases above described, but the projxirtion of dominants
»uid recessives in F,j is very ditferent from the usual Mendeli.in ratio, and
reminds us of the occurrence of a complex segregation.

Heiv u»ay we be allowed to make a little digression. As stated
above, the leaves of F, hybrids between »S. niultiiiervis and f/radiistt/ld
are wholly glabrous, whilst in hybrids between the former and S. viitiinalis
they are hairy underneath though le.ss densely .so than in the latter, so
that the.se hybrids may be regarded in this respect as intermediate
between the two j^iirents. The hairy condition is apparently recessive
in the former case, and dominant (or strictly s[)eaki!ig intermediate) in
the latter. How such different conditions may occur in spite of the
fjvct that we have u.sed the same species S. multinervis — one and the
same tree in both cases — must of course remain a matter of conjecture
80 long as no extensive culture of several later generations of such
hybrids has been made, but the following may be perhaps one of the
probable explanations based on the presence-and-absence hypothesis.
Let H represent the ftictor (or the factor-complex) for the hairy condition
in S. gracilistyla and let I represent the inhibitory factor contained in
S. multinervis, then we have S. gracilistyla = HHW and S. midtinervis
= hh 1 1, therefore Fi=Hhli, and since the factor I is able to suppress
wholly the hair- producing action of H we have in the hybrid Hhli
leaves which are entirely glabrous. Since in the hybrids between
S. multinervis and viminalis leaves are hairy in contrast to tho.se
between the former and S. gracilistyla we are led to think that the
factor (or the factor-complex) for the hairy condition in S. vimiiialis is
different from that in S. gracilistyla. If we represent that factor by
H\ then we have S. viminalis = WH^W, S. multinervis =h^h^\\, and
Fi = H'hHi. The si\me inhibitory factor I which was responsible for the
entire suppression of the hair-producing action of the factor H in
S. gracilistyla may be regarded as being less i)otent against H' than
against H and able to prevent the action of H' (►nly partially, so that in

40 On Hybridisation of some Species of Salix

the hybrid H^hHi leaves are hairy, though less densely so than in the
parent S. viminalis.

(c) Stipules.

Leaves of S. wultinervis are exstipulate (cf. Text-fig. 1 B, p. 37),
whereas those of S. gracilistyla are stipulate (cf. Text-fig. 1 A). The F^
plants may be said to be a mosaic of the two parents respecting the
behaviour of stipules, for each individual is always provided with
both kinds of leaves, stipulate as well as exstipulate, and even in one
branch those with stipules may alternate with those without them.
The degree of their development is also very variable in different leaves,
because they are sometimes very conspicuous (cf Text-fig. 1 C, p. 37),
sometimes very insignificant being represented by mere tiny scales ; not
rarely we have one unpaired stipule on one side of the leaf In F^
plants we see generally the same behaviour of stipules as in F^, for
then they are provided both with stipulate and exstipulate leaves.
Besides such plants we have some F^ individuals where we could yet
find no stipulate leaves, i.e. where all leaves are exstipulate exclusively ^
Thus of 232 plants examined we have 170 with both kinds of leaves
and 62 with exstipulate leaves only. The latter plants are already six
to seven years old, and are pretty advanced in their growth, for many
of them are more than 1 metre, and some .even IJ metre high, and are
provided with a large number of branches. One might therefore be
led to conclude that they really lack stipules, and are the segregates
in a Mendelian sense, but I think that such a definite conclusion
may yet be considered as too hasty, because I have many times
experienced the fact that plants which were at first provided with
exstipulate leaves only, were later found to produce some leaves which
are clearly stipulate. Thus it is not unlikely that the 62 plants referred
to above, in which exstipulate leaves were exclusively found so far,
may in future bear some stipulate ones. And if the latter alternative
really holds good, then perhaps w^e have here a case of the so-called
" blending inheritance " or " constant intermediate inheritance " where
the behaviour of stipules in F^ , which is intermediate between that of
the two parents, always repeats itself throughout later generations.
But if, on the contrary, it be proved beyond all doubt that in F^ we
have a certain number of plants with exstipulate leaves exclusively, the
segregation of the characters " stipulate " and " exstipulate " may be

^ All F2 plants were examined for stipules every year, and several times each year in
different stages of the development of their branches.

8. Ikrno 41

considered to take place, and then we have to deal with "altemaU*"
instead of ** blending inheritance." It will hi* socn howc'Vrr from what
was statc<l above that we are unabK* ju* yrt t<» prove the <)ccurrence
of either the one or the other kind of inheriUince. We may alm» here
remark that so long as any one, wishing to prove the Hegregation of the
allelomorphic characters under considenition, takes plants with exstipu-
latc leaves for the standanl, or to boiTow the wonl of Nilsson-Khle, for
the "analyser*," he would never be able t^o arrive at a definit<^; con-
clusion, owing to the j)ossibility that plants which w(»re at first provided
with exstipidat-e leaves only might later devel(>p some stipulate ones.
If, on the contrary, we could find in t\ even one plant with all its leaves
provided with stipules like the one parent S. gracUistyla it would be
possible t^ reach sjifer conclusions ;is to the occurrence of segregation
of the allelomorphs under consideration. But not even one single such
plant has been obtAined till now. Probably the problem will not be
definitely sohtnl without breeding exiK'riments conducted on a far larger
scale than wjvs possible for the present author. In short, my experiments
.have not been able toprovi* the segregation of the characters "stipulate"
and " exstipulate."

{d) Colour of Stigma.

In S. multinervis the stigma is bright scarlet, while in S. gracilistijla
it is green. In F, plants it is scarlet as in the former. In F^ the
segregation of the two opposite characters is quite evident. Thus we
have 115 and 16 plants with scarlet and green stigmas respectively,
while 7 plants have greenish-red stigmas. If we add those with scarlet
and greenish-red stigmas together, we have 122 red and 16 gi-een, i.e.
almost 8 red : 1 gi-een. The segregation of allelomorphic charactei-s
* red" and "green" is thus clear, but as in other characters hitherto
enumerated the usual Mendelian ratio cannot be detected.

(e) Character of Catkins.

In S. gracilistyla the catkin, either male or female, is long and
broadly cylindrical, and very densely covered with long gray hairs
(Plate I, fig. 1, J* — J), whereas in S. multinervis it is much shorter
and narrower, and very sparingly hairy (Plate I, fig. 2, J" — J). The
chief difference between the catkins of these two species lies thus in
the degree of hairiness : in the one they are densely hairy, while in the
other they are sparingly so. The distinction between the two in this

. ^ Lund* Universitets Arsskri/t, N.F., Afd. 2, Bd vii. 1911, p. 18.

42 Oil Hyhridisatlon of some Species o/Salix

respect is sharp, and there is never found any transitional form between
the two.

In ^1 I have obtained two sorts of individuals : one of them has its
catkins densely covered with long gray hairs as in the one parent
S. gracilistyla (below designated as plants of 6^-type from the word
gracilistyla) (Plate I, fig. 3, ^ — $ ), whereas the other has its catkins
resembling those of 8. multinervis, i.e. sparingly hairy (designated below
as plants of iW-type from the word " multinervis") (Plate I, fig. 4,
^ — J). The distinction between the two types of catkins is generally
as sharp as between those of the two original parents, though in respect
to the male catkins the distinction between the two types is sometimes
difficult to be made out.

Let us first describe the results obtained in the F^ and i^a genera-
tions. As above stated (p. 36) I had only 14 i^j individuals, which may
be classed as follows :

1. <9-type 2. M-type

11 3

4<r 7? 0<f 3?

Not one single male plant of i/-type was obtained in i^i, but a few
plants belonging to this category appeared in F^. as stated below.

The following crosses among the F^ plants were made in 1912 and
1913, viz.:

1. 6^-type % X G^-type c/".

2. i/-type $ X G-type c/".

The fertilisation if-type $ x if-type (/ was not possible then, because,
as just stated, no male plant of if- type appeared in F^. In 1914,
however, I got one male F2 plant of this type resulting from the hybrid
i¥-type J X (r-type (^ above mentioned (Plate I, fig. 4 </), and I have
done the fertilisation between this male and the female ilf-type Fi
plant used in 1912 and 1913, thus :

3. il/-type F,^x if- type F^J',
The results of these three fertilisations are as follows :
1. G-type 9 X G-type $.

G-type M-type Total

78 c? 109 ? 2 c? 30 ?

Totals 187 32 219

( = 85 -4°/,) ( = 14-6 7o)

S. Ikkno 4:)

l»<{ 49 f 7i H?

77 67 l/> IM

( = 48-4 «0 ( = 4'i-17o) ( = 9-»*'/c,)

If we mid togi»thor plants of Af-iy\Hi an<l those of N«'W ty|H', l)<»tli
being very similar to each other (cf below, p. 44) we have

(7-t]r|ie Af-ty|»e + New tyiw ToUl

159

3. M-fyp^ F, 9 X Af-typp F., ^ '•

tf type A/ type New type Total

3<J 0? 2i 8? 2cr 3?

83 ,; 44 ?

26 i 57

Totals

77

(=48-4°/,)

H2
( = 51-6°/,)

Totals 3 10 5 18

(=16-7°/o) ( = 55-5%) ( = 27-8 7,)

If we add together, as before, plants of J/-type and those of the
New type, we have

G-type 3f-type+New tyiie Total

3<f 0? 4<f 11?

Totals 3 15 18

( = 16-7°/o) ( = 83-3°/o)

If we add together all offspring derived from the three fertilisations,
we have

G-type A/-type Newty|)e Totals

1. G-type X G-type 187 82 0 219

2. If. type X G-type ... 77 67 ll 159

3. J/type X .l/.type ... 3 10 5 18

Totals 267 109 20 396

( = 67-4°/3) ( = 27-57o) ( = 5-l.°/o)

' For description, Bee below, p. 44.

- This year (1918) almost all the offspring derived from this fertilisation were found to
possess some buds containing young catkins, but many of the latter ceased to grow in their
very young stages, having been killed by the severe frosts of last winter, so that I was able
to ascertain in 18 tree.s only the type of catkins borne by each of them.

44 On Hybridisation of some Species of Salix

or, if we add plants of i/-type and those of New type together, we have

G-type M-type+New type Totals

1. G-type X G-type ... 187 32 219

2. ill- type X G-type ... 77 82 159

3. 3I-type X M-type ... 3 15 18

Totals ^7~ 267 129 396

(=67-4%) ( = 32-6°/,)

As will be seen from the above tables the fertilisation between the
(r-type $ and the 6^- type (/ , which we may perhaps consider as
corresponding to the self-fertilisation of a certain F^ plant which bears
hermaphrodite flowers, gives rise to many (r-type and few ilf-type
F2 plants (cf. No. 1 in the above tables). The diametrically opposite
behaviour will be seen in the fertilisation between Tlf-type F^ % and
ilf-type F2 c/ plants, for then comparatively many ilf-type and com-
paratively few G-type plants are produced (No. 3 in the above tables).
In contrast to the two above cases the fertilisation between % and </
plants belonging to the two different types gives rise to the offspring
of both types in almost, though not quite, equal numbers (No. 2 in the
above tables). From these experiments we see that each of the F^
plants, whether 6^-type or il/-type, is heterozygous, and gives rise
by a fertilisation corresponding to self-fertilisation in hermaphrodites,
not only to offspring of the type similar to itself, but also to a
small proportion of those belonging to the other\ The occurrence of
segregation of the catkin character under consideration in F^ is thus
quite evident ; it must however be remarked that the ratio of the
numbers of plants of both types is then very different from the usual
Mendelian one.

Before proceeding further, I must make some remarks about plants
marked as " New type " in the above tables. These plants which have
arisen in F.^ from either if-type x (r-type (in the ratio of almost 10 7o)
cf the above tables) or ilf-type x if-type (in that of almost 28 °/^), but
never from (r-type x (r-type, are very similar to those of if-type, and
differ from the latter simply by the entire absence of hairs in catkins
(Plate I, fig. 5, (/ — ?)^ The catkins of this form appear to be much
more intensely black than in plants of (r-type or ilf-type, but this is
only apparent, because though the bracts of the two latter are in

1 "With some reserve in the case of IT-type plants, because I was not able to get male
M-typed plants in F^, and have depended upon the fertilisation M-type Fi x ilf-type F2.

" At least to the naked eye, because examined under the microscope, the bracts of
catkins of this form are found to possess a few short hairs.

8. Ikkno 4o

reality just as bhick ixs thow o( this ih'nv fnnn, tluir coltnir is |Nirtly
coiicchIihI by haii-s covi'riiij; tlu'iii. In (lii« lu'w fonii there are twn
kindji of fciimle plant-H, the one |H>.sHeHHi!ig red Hti^nnax and the other
green onen. In 191 (>, when they appeared for th** firnt time, a few
femiile catkins (riowers with green stignuus), which <lid not look «juit«'
healthy, wen* fertilis*Hl with jHiIlm from the mah* catkins of iju- sjuim'
type, and a few see<is wiTe obtained, w hich »iid not );«'rminat;4*. This
year(l!)lS) I re|H'ati'd the same fertilisation (flowi-rs with rc(| stigm;i,s),
and got many si'cds, which were able to germinate and have given lise
U» a certain number of seedlings. Whether or not the latter breed
true t-o the ty|>es of their parents is of cours<' yet unknown ; it is, nev<r-
thele^, not unlikely that this new type is a muUint — possibly a
loss-mutant priKJuced on account of the loss of the factor for the*
hair-formation — and arising after hybridisiition. Mutation after hyl^ri-
disation hsis, jus is well known, been sometimes discovered in (Jenot/ieni
by de Vries', (Jates-, etc., and also in HKbu^s by Lidfoi-ss'.

What I have described above about the formation of the two tyjKJS
of plants afti'r hybridisiition between S. gvacilistyla and S. inultinervis
is merely the description of the results actually gained, and I must now
go into their interpretation. The question is, should the appearance of
both (i-type and J/-typc otispring in F^ be regarded ;is the result of
segregation, or may this fact be explained in any other way ? The
exact answer to such a question cannot be given without further
breeding experiments, and I am, for the present, only able to make
certain hypotheses about it.

According to the first hypothesis one of the two original parents,
either S. gracilistyla or aS*. nudtinervis, should be regarded as being
heterozygous, at least in respect to the catkin character under con-
sideration. Thus if, for instance, we denote the (r-type character by
G and its absence, i.e. the Jl/-type one, by g, and if we further give to
S. gracilistyla and S. multiuervis the formulae Gg and gg, respectively,
we have in the fertilisation of the two ASWi^-species a back-cross
Gg X gg = Gg -f gg, thus explaining the production of the two types of
plants in F^. It will be of course the same, if we consider S. multinervis
to be heterozygous and S. gracilistyla to be homozygcjus reganling this
character. Is then either of the two *S'(//ia;-species under consideration

* See, for instance. Die Mutationstheorie, Bd i. pp. 211, 212 and Bil ii. pp. 425, 426;
also GrupptniceiKe Artbihhtng, p. 302 flf.

* The Mutation Factor in Evolution, p. 2HG ff .

' Zeits.f. ind. Abstaminunys. und Vererbumjulchn', Bd xii. l'U4, pp. 1 — 13,

46 Oil Hfjhrldisation of some Species o/Salix

really heterozygous in this respect ? For the elucidation of this question
I have made the fertilisation between male and female trees of each of
the two species — the very same trees which were used in my hybridisa-
tion experiments ; this fertilisation might correspond to selfing in
hermaphrodites. Seeds obtained by this process were sown, and plants
developed from them — 70 in S. gracilistyla and more than 100 in
S. multinervis — and were found to be exactly similar to their respective
parents in all respects. There is therefore no reason for considering
either of the two parents to be heterozygous regarding the catkin
character under consideration, and the first hypothesis should be dis-
carded.

According to the second hypothesis one sex, either male or female,
of one of the two species, is regarded as heterozygous and the other
homozygous in respect to the catkin character. Thus suppose, for
instance, the male plant of >S^. gracilisti/la to have the formula Gg, and
the female GG, then the fertilisation between them which is a back-
cross gives Gg ^ and GG J , and as G is dominant over g, S. gracilistyla
breeds always true despite the heterozygous nature of its male plant.
The hybridisation of gg $ (= S. multinervis) by Gg ^ (= S. gracilistyla)
should give in F^ the zygotes represented by Gg and gg, but as the
female G-type plant should be always homozygous (= GG) according
to our presupposition, we should have in F^

thus not one female G^-typed plant should then appear, which is
contrary to the fact actually seen, because I obtained seven female
(r-type plants (p. 42).

What will be the case, if we suppose the presence of the inhibitory
factor of hairs I ? Thus, for example, suppose the male plant of
S. multinervis = W, its female plant = li, >S'. gracilistyla (both male and
female) = ii, then the hybridisation li J x li ^ gives in F^ 11 and ii, and
since, according to our presupposition, the male if-type plant should
be always homozygous (= II) we should have in F^

li ¥ +il J+iic/',

thus not one male i/-typed plant should appear, which indeed accords
with the fact, for I got no such plant in F^ (p. 42). The F.^ offspring
arising from the fertilisation between male and female (?-type F^
plants (i.e. ii?xii(/=ii?+ii^) should however undergo in F^ no
segregation, and give G-type plants exclusively, which is contrary to

S. Ikkno 47

fact, bociiiiso I gt»t thmugh this ffrtiliMation O'-tyiH* ivi well ns M-ly\n'
plants in this giMioration (p. 42).

Thus the hypothosis which rr^unls cuic ol tin- two s«'x<*,s as iK-ing
hfti»n»zygou8 <l«>es not acronl with the \m:\^ actually ohsrrvrd and is
untenable.

The thin! hyjMjthesis is t'oun<le(l on what is variously called " inijMr-
fection of dominance/' "reversal of d«»nnnanc('," m- " fluctiiatiun in
dominanee," because our c^usc sii'ins. nl lea-st, ap|»ai«iit,|y very much
ivh\te<l U} that plienonienon. We have many examphrs of the latter
in pi>ultry acconiing to Hateson and Puiniett', tis well as l)av<'nport'.
To cit4.» only one example from the latter author, extra-toed individuals
of poultry mated with normal ^ive extra toe only in 7.*^ °/ of th«'
ot^spring, the remaining 27 "j having the normal number (jf toes^ yet
that both kinds of the offspring are heterozygotes was proven ])y the
fact that each of them mated inter se ha>! exhibited segregation in FJ.
Acconling t^> the author just named extra toe is dominant to normal,
but in 27 ° '. of the offspring this dominant character wjis not i)owerful
enough to exhibit itself; we have here to deal with the phenomenon
which is called " imperfection of dominance," etc.

The appearance of 11 G-type and 3 3/- type plants in /', of our
SdliX'Cross would, according to this hypothesis, be due to the latter
phenomenon, and the (7-type character which is generally dominant
to the other should be regarded as having failed, in the present case, to

Q ^ 100
be so in — ^= 21 % ^^ the offspring. The fact that the Fi plants,

1 1 -|- o

whether G^-type or J/-type, are heterozygotes, and undergo segrega-
tion in F<,, has also been proven. Thus, ' according to the present
hypothesis, the production of the two types of plants in F^ is not to be
regarded as a process of segregation.

Let us now examine whether this hypothesis explains the facts
actually observed. First of all, it must be marked that what some
authors regarded as the "reversal of dominance" or a phenomenon
similar to it was found sometimes on further inquiry to be explained
in quite another way. Thus, for instance, Coutagne ' and Kellogg*^

> Reports to the Evolution Committee of the Jioijul Society, Report II, 11)05, pp. 1 14—1 16.

■« Carncifie Inst. Washington Ptibl. No. 121, 11)00; Amer. Xat. Vol. xliv. I'JIO,
pp. 129—135 ; Amer. Breeders' Assoc. Vol. vi. 1911, pp. 29—32, etc.

•'» Carnetjie Inst. Washington Ptibl. No. 121, p. 19, Table 10.

* L.r. pp. 20, 21, Tables 11 aud 12.

^ Recherches e.rperimentahs sitr V Ilerf:dile ehez is I'ers ii Soie. These. Faculte d.
Sciences, Lille, 1902. '• Leland Stanford Junior i'nic. VubL, University Series 1, 190h.

48 071 Hybridisation of some Species of Salix

discovered that in crosses of some silk-worms which spin yellow and
white cocoons, respectively, the dominance is variable, because in some
yellow is dominant to white, while in others the reverse takes place ;
this is due, as the latter author thinks, to strain or individual idiosyn-
crasies, but Toyama^ has proved experimentally that this phenomenon
may be better explained as the effect of a mixed breed, containing
recessive as well as dominant whites, than as that of individual idiosyn-
crasies. Almost a similar explanation is applicable to what Correns
and Lock have observed in hybrids of Maize. In Maize, alha x cyanea,
where blue is dominant to white Correns found in F^ 94 °/^ blue
individuals and 6 °/„ white ones^ ; also in Maize, Moore's Concord
(white) X Black Mexican (black) Lock found that black was dominant
to white, but that sometimes the reverse takes placed As first pointed
out by East^ and afterwards by Lock himself^ this was due to the fact
that "a supposed pure white strain" used in the hybridisation was
composed in reality of a number of genotypically different individuals
which, though pure for white when selfed, differ among themselves in
carrying some invisible factors which react differently in the production
of colour^ Thus we have here to deal, not with the "reversal of
dominance," but with a " mixed breed," almost in the same way as in
the case of the silk-worms above enunciated. It may perhaps be
reasonably doubted, whether also in the so-called "reversal of dominance"
in poultry we have not to deal with similar circumstances as in Maize
just mentioned.

Our case in Salix is however somewhat different from that of Silk-
worms or Maize, inasmuch as the fourteen F^ hybrids are derived from
one and the same female plant fertilised by pollen taken also from one
and the same male plant, so that if our case were really explicable on
the basis of the hypothesis founded on the reversal of dominance it
must necessarily follow that the G^-typed catkin is sometimes dominant,
sometimes recessive to the if-typed one in the same individuals, which
does not seem very probable. It appears to me much more reasonable
to consider that though either one of the two types of catkins, for
instance the (r-typed one, is in reality always dominant to the other,

1 Zeits.f. ind. Abstammungs- u. Vererbungslehre, Bd vii. 1912, pp. 252 — 288.

2 Bibliotheca Botanica, Heft liii. 1901, pp. 53 — 55.

3 Ann. B. Bot. Garden Peradeniya, Vol. iii. 1906, pp. 117—129.

4 The Connecticut Agric. Exp. Station, Bull. 167, 1912, pp. 57—100.

5 Ann. B. Bot. Garden Peradeniya, Vol. v. 1912, pp. 257—264.

6 Lock, I. c. p. 257.

H. Ikrno 49

it« apparent r^oeanffeneaa is ciuitkHl in buiiu' cii»«'h by the inHui?nce of
other factors contniiUHi in tluMn, c'H|K'cially Moinr inviHihh* facU)rH, an
explanation similar U) that Hrst j)n)|M>.stM| by Ktwi about the Maizr-
cross just cite<i. Let u\v thru describe bc*low my views re^anljn^ this
question. Sup|)08e that either one of the two jNirents urnlrr eon-
sidemtion, iS. wultincrvis, for instance, carries some such fjictors in .i
heten)zyg<.»us condition'. In gametic fonnation thf latter will undergo
segregation, so as to give rise to gameti*8 containing ditVerent combirui-
tions of invisible fjict^irs. In the fertilisation betw('<«n mah- and frmair
plants of this species ganietes ditTering in respect to invisible factors
may come to copulation, yet the (►flfspring will always breed tru<' to
their pjirent type, at least all of them will agree in their catkin
chanicter, bi^nuse since the factors for the latter character an- in th<«
same homozygous condition in all of them, there will be no nason
why the catkin belonging to any other than the J/-type will come to
development, so that in this case the difference of invisibh; factors in
different offspring will be perfectly indifferent towards the development
of this character.

Quite different results may however be expected in the hybrid is^i-
tion S. muUinervis x S. gracilistyla. All F^ hybrids will agree now in
carrying the same factors concerning the catkin character in the sfime
heterozygous condition, for instance Gg'-, while they will differ am<jng
themselves in containing invisible factors differently combined, just as
in the former aise. It is then reasonable to consider that these
invisible factors, owing to the difference in the mode of their combina-
tions in the various offspring, will co-operate with Gg, so as in the one
case to let G dominate over g, and in the other to induce just the
contrary effect, thus producing, respectively, the G- or the J/-type in
different individuals. In short, whether the one or the other type of
catkins will make its appearance, may be regarded as being due to the
influence of invisible factors accompanying the catkin factors. Thus if
our view be true, the phenomenon seen in F^ is to be regarded as
being due to the segregation of invisible factors, but not to that of the

' Factors may be present, which do not by themselves aluue produce any visible effect, or
at least can produce some effects which are so inpignificnnt as to escap<^ our eyes. Such
factors are "invisible ones"' which are able to produce visible effects only in co-operati<fn
with other ones, either visible or invisible. Further, it is here supi>08ed that only one of
the parents carries invisible factors, but it will make no difference whatever in our loj^'ic. if
we consider them to be carried by both parents.

2 G =: G-type, g = absence of G - J/-type ; the factorial composition may really be much
more complex, but it is here so represented for the sake of simplicity.

Joarn. of Gen. viii 4

50 On Hybridisation of some Species of Salix

catkin factors themselves, because the latter are retained in the same
heterozygous condition in all F^ offspring. What we have seen in F.^,
(pp. 42 — 43) is however the segregation occurring on account of the
heterozygosity of the catkin factors carried by F^ plants.

Let us now go to F^. As already described (pp. 42 — 43) the fertilisa-
tion 6^-type X (r-type gives rise to many (r-type and few i/-type
plants, and the ilZ-type x if -type gives rise to many if-type and few
(r-type ones, whilst in the if-type x (r-type plants of both types are
produced in almost, though not quite, equal numbers. The explanation
of this peculiar mode of i^2-segregation will, as I think, naturally follow
from our hypothesis adopted about F^ plants. We have supposed
(p. 49) that each of the F^ plants, whether (r-type or i/-type, possesses a
similar factorial constitution in respect to the catkin character, which
we have represented by Gg ; in F^ we should have then on account of
the segregation

GG-f wGg + gg

in all cases, n being any positive integer equal to or greater than 2.

As already - noticed (p. 49) each of the (r-type F^ plants carries
besides the factors Gg a certain combination of invisible factors which
we may for instance call X, and which acts together with the latter, so
as to give rise to (r-type catkins exclusively; accordingly all the F2
offspring derived from the fertilisation (r-type x G^-type will contain X ;
and so of the F^ plants GG + z^Gg + gg, GG and ?iGg (the latter
under the influence of X) should be (r-typed, whilst only gg should be
ilf-typed, thus explaining the fact that the F2 offspring consist largely
of (r-type plants.

On the contrary, as each of the if-type F^ plants carries besides Gg
a combination of invisible factors which we may call Y, and which acts
together with the latter, so as to give rise to if-type plants exclusively,
we may, by similar reasoning as above, come to the conclusion that of
the offspring GG + ?i . Gg + gg derived from the fertilisation ilf-type x
iV/-type, nGg (under the influence of Y) and gg should be il/-type, and
only GG, (r-type, thus explaining the fact that the F2, offspring are
then largely il/-type.

In the fertilisation TIf-type x (r-type we have to deal with the two
kinds of combinations of invisible factors, X and Y, which are provided
with diametrically opposite characters. It is clear that in F^ some
offspring will receive X, whilst some others Y, and that then Gg accom-
panied by X will be (r-types, whilst other Gg accompanied by Y, will be

8. Ikkno 51

i/-type& In uther words, of the h\ offHprin^ 00 und gg will l>clung to
G-typca and ^/-type«, rt»8|K»ctivi'ly, whilst Og will belong jwirtly to the
one and jjartly U> the other. Thus the ntiiiilHT t»f plants of hnth tyj>eH
Mhould be theoretically tM|ual to each other, which acconiM, a« wt? have
alremly 8een. with the fact n-ally obsrrvrd.

This explanation of the behaviour of our cross in /', and F.^ '\h
naturally mere hypothesis which needs to be subjecte<l to ex|>eriinental
verification. The latter would b** however extremely difticult, if not
absolut'cly im{x>ssible, but I intend to continue my work in this
direction, as far as I can.

To summarise, the formation of the two ty|)es of aitkins in F, is not
to be looked uixm jus the ri'sult of segregation of the aitkin chanicter ;
the occurrence of the latter process in F., luis however been clearly
proven, though the ratio of the two tyjx's pro<hiccd in each of the
three kinds of fertilisiition is (juite different from what we might have
expected in usual Mendelian cases.

B. Results of the Hybridisations done in 1911.

As already stated (p. 36) I have repeated in 1911 the same hybridisa-
tion done in 1910. The female plant wiis the very s<ame tree used in
1910; whether or not the male plant was just the same as that used in
1910 is now unknown, but it belongs, at least, to the same vegetative
line (in the sense of Fruwirth) or the same clone (Webber) as the latter,
because in 1911 all male plants of S. gnicilistijla in our Botanical
Garden were exclusively derived from the cuttings of the same plant
used in 1910. This hybridisation succeeded pretty well, and I got
nearly fifty seedlings. They were, however, contrary to the result of
the hybridisation in 1910, not hybrids at all, at least externally. They
were nothing but S. niultinei^vis, and when they came to flowering, all
of them have proven, to my great astonishment, to be female individuals
without exceptions, or in other words, the offspring were of purely
ttuUenial type, so-called /a^se (Millardet) or unilateral hybrids (de Vries).

The production of either purely paternal or maternal j)lants as the
results of hybridisation — what Bateson calls " Monolepsis"' — h;is been
sometimes met with by various authors. Thus Gartner' obtained from
the hybridisation Melandriuni rubruui % x Sileiie noctiflora ^ only

* Report to the Kvolution Committee of the Royal Society, lleport I, 1902, p. loo.

* Versuche und Beohachtungen Hber die liaxtdrderzeuyuuy im Pjlanzenreich, Stuttgart,
1849, p. 37.

4—2

52 071 Hybridisation of some Species of Salix

two hybrids and many M. ruhrum, i.e. plants of purely maternal type.
In some species of Fragaria Millardet^ got plants of both paternal
and maternal types by hybridisation — the well-known "hybridation
sans croisement" or "fausse hybridation 2." This was repeated on
Fragaria virginiana ^ x F. elatior ,} by Solms-Laubach, who got hybrids
of purely paternal type, thus confirming the results of the French
author ^ The same kind of hybrids as the latter were found recently by
Collins and Kempton in the Gramineae Trifsacii'm dactyloides $ x
Euchlaend meadcanaj*, which breeds true in later generations — what
these authors call " Patrogenesis^"

Hybrids of purely maternal type were obtained by Hurst from the
Orchid Zygopetalum Mackayi fertilised by some species of Odonto-
glossum, Oncidium and Lycasta^.

A very detailed study was made by Lidforss on a great number of
species of Rubus^. According to him the hybridisations, R. acuminatus,
R. divergens, R. dissimulans, R. plicatus as females by R. caesius as
male, and R. polyanthemus as female by R. Bellardi or radula as males,
for instance, have given genuine and false hybrids in nearly equal
numbers, whilst the hybridisations, R. polyanthemus, R. insularis,
R. Lindenhergii as females by R. caesius as male have given rise ex-
clusively to false hybrids. The false hybrids of Lidforss were always of
purely maternal type, and were found to breed true in later generations.

To cite another example, de Vries obtained by the hybridisation
Oenothera Lamarckiana $ x 0. biennis ^ false hybrids of paternal type,
which were found to breed true in later generations''. The same
hybridisation carried on in the New York Botanical Garden has given
quite different results, because it has given rise to four distinct types
of hybrids (MacDougal, Vail, Shull and Small)^, and the latter four
authors came to the conclusion that the difference of their results from
those of de Vries might be due to the influence of some factors, such
as individual qualities as well as external conditions. It might however

1 Mimoire de la Soc. des Sciences physiques et naturelles de Bordeaux, 4 S6rie, torn. iv.
• 1894, pp. 347—372.

2 Besides Fragaria false hybrids were observed by Millardet in Rubus (I.e.) and Vitis
(Revue de Viticulture, torn. xvi. 1901, the original paper not seen, reviewed in Winkler,
Prog. Rei Bot. Bd 11. 1908, p. 344.

3 Bot. Zeit. I. Abt. Jahrg. 65, 1907, p. 53.

4 Journ. of Heredity, Vol. vii. 1916, pp. 106—118.

5 Journ. R. Hort. Soc. Vol. xxix. 1900, pp. 104—106.

6 Zeits.f. ind. Ahstammungs- und Vererbungslehre, Bd xii. 1914, pp. 1 — 13.

' Die Mutationstheorie, Bd 11. p. 31, and Gruppemveise Arthildung, pp. 156 — 159.
8 Carnegie Inst. Washington Publ. No. 24, 1905, p. 17 ff.

S. Ikkno r)3

be asked, whether the Oenoiheni -hik^ch^h U8e<i by the»t» iiuthorH hiul
genotypic constitutioiiH exactly similar to thns(» wnvil by the Dutch
botaninL If thi» be n^ally true, then this hybri<liwiti<ni stuMiiM t^» have
mm\e remMiibhince to what I have seen in .S'a/i.r-crnHH unHer ennMi«h»-
ration.

My rc»8ult«(»n the hybridisjition Salir inuld'ncrvis x Sulix gnu^lUUfUi
ap-ee with those reconled by Lidt'orss fur liiibns, inasmuch us 1 havi*
also obUiiniMl lv>th genuine hybrids jis well us plants uf purely mat<*rnal
ty|K'. The (iiflferenee between the tw(> hybridisiitions li«*s oidy in this:
while Lidfoi^vs obtained l>oth real and false hybrids in one and the same
year. I was aWe to obtain them in diti'erent years (1910 and 1911).
That the dilTerence of results in thwso two yeai-s is due to the ^'rnolypic
differences of male and female plants used by me may be absolutely
denietl, for. jvs above stated, the female plant used in both civses w.us
one and the Siime tree, and the male plant in 191 1 was either the same
with, or at least derived in a vegetative way from, that used in 1910.
As I obtained ditlerent results in ditferent years, one might be dis-
j)osed to think that whether the ot!*spring will be real hybrids or of
purely maternal tyi)e is dejK'udent on external conditions, which indeed
may be true. But I think it ecpially likely that our plant contains
two kinds of eggs, giving genuine and false hybrids, respectively,
just as we have in Thalictrum purpurascens^ and some species of
Hieracium^ two distinct kinds of eggs, i.e. those .which can develop
only after having been fertilised, and those which are able to develop
parthenogeneticiilly, though the final decision of this (piestion will be
of course impossible without performing further breeding experiments.

Now some remarks about the sex of false hybrids. Rubus is always
hermaphrodite, so that both real and false hybrids are naturally of this
nature. The sex of those of Melandrium obtained by Gartner is
unknown, for he has stated nothing about it^ Those of Fragaria
virginiana % x F. elatior J^ were either male <a- female*. Those of Salix
got by me, almost fifty in number, were female without a single excep-
tion, so that they may be well siiid to be of maternal type in the most
strict serine of the luord, for even the sex has been inherited. Further-
more, that our false hybrids will breed true in later generations like

1 Overton, Hot. (iaz. Vol. xxxiii. l'.K)2, pp. 3G3— 374 ; lU-r. <L heutxch. Hot. Ge*.
Bd xxu. 1904, pp. 274—283.

* Ostenfeld, IloUiuhk Ti(hfkrift, Bd xxvii. l'.*(»6, pp. 221 -J.'iO ; Xeitx. f. itnl. Ahst,im-
mungit' u. Vererhnunslehre, Bd in. 1910, pp. 241—28^).

» Lx. * SoliiiH-Laubacli, / c. p. o3.

54 On Hybridisation of some Species of Salix

those of Ruhus, is highly probable, though not yet actually proven.
This problem will be one of the objects of my future study.

That in the case of Salix the formation of false hybrids is due
neither to parthenogenetic development of the oosphere nor to the
vegetative production of embryos from nucellar cells is quite evident in
view of the fact that female inflorescences covered with paper bags
were never able to bear even single fruits. Many authors think ^ that
in the formation of false hybrids pollen has nothing to do with fertilisa-
tion, but acts merely by irritating egg-cells in some way and enables
them to develop into embryos without being fertilised ; such process is
called pseudogamy, a word first proposed by Focke^ Giard-^ thinks
that false hybrids of purely paternal type are derived from maternal
cytoplasm with the male nucleus alone, the female nucleus degenerating
(Merogony !), and that those of purely maternal type are derived by
pseudogamy, some stimulus to development being given by the pollen-
tube without entrance of the sperm-nucleus into the egg. All these
are however mere hypotheses which are simply more or less probable,
and which ought to be proven cytologically. The only cytological
investigation on false hybrids of plants is that of Strasburger on
Fragaria*, which did not confirm the hypothesis of Giard above stated.
Thus in the hybrid F. virginiana % x F. elatior ^ the former author
could observe no degeneration of the egg-nucleus, while, on the contrary,
not only was he able to see clearly the fusion of the sperm- and the
egg-nuclei, but he was able to count in the mitosis of the" fusion-nucleus
the diploid number of chromosomes.

False hybrids have also been observed in animals, and there is a
series of papers concerning hybrids of purely maternal type, though
they never reached the adult stage. In these hybridisations, or hetero-
geneous fertilisation, as it is often called, eggs of the Echinoids
(as Sphaer echinus J Strongylocentrotus, Echinus, Arbacia, etc.) were
fertilised by sperms of the Echinoids, the Crinoids (as Antedon), the
Mollusks (as Mytilus), the Vermes (as Chaetopterus). Since in these
hybridisations the systematic affinity of the two parents is always very
remote from each other, only the larvae, in more or less advanced stages
of their development, were obtained, and these have always proven to be

1 For instance, Hurst, I.e. p. 106 ; Winkler, I.e. p. 333.

2 Die Pflanzenmisehlinge, Berlin 1881, p. 525.

3 Gomptes-rendus de la Soe. de Biologic, einquantenaire de la Soc. 1899, Vol. jubilaire,
p. 665 ; Gomptes-rendus hebdomad, des Seances de la Soc. de Biologie de Paris, Vol. lv.
1903, p. 779 (original not seen ; cited according to Solms-Laubach, I.e. p. 53).

4 Histol. Beitrdge, Heft vii. 1909, pp. 43—46.

a. Ikkno 55

nf pureli/ maienuil tt/fte (VoriKm, IltThst, Ku|N'IwicH<*r. (t(Kll(»\vHki jun.,
Biiltzor, Morgiin. etc. vie). CyUAof^\ci\\ invrstij^iitidnH of thenf falHc*
hybrids won* mmlo iils<> by niftny aii thorn. The fiict hjw Ixm'ii n'VfnlcKl
that in the hybriciisatitm of th«' Iv-hinoids thr s|MTiiiat<ozoH jilwayH mU'r
thf ogg-ryUiplitsin. 'V\\v behaviour of th«' .s|H'nn-nMclruM wjis ho\vrvj«r
foiiml always nt>t t^» be tht» winie. For iiistaiiee, in Homr nxm'H (as
Kchiims J x Mi/tihis^, Stroiuft/locentrotns J :•; }fi/tilns ^) th«* r^^- aii<i
the sjK»rin-inieK'i do not come to real fusion, and the latter undrr^o
^nulual de)(i'neration (Kupelwieser') ; in many other <";ts«s l>oth nuelri
ftise U) each other, and in later stages the paternal ehroniatin is
ebniinat^Hl in s<)nie way (Haltzer'. Kupelwicscr^ (i<Kll«'wski jun.*).
Acconlin); t«» Baltzer'. in the hybrid Sphaerec/nnus J x StrotHfi/htren-
tnitus (fy which is intt'rnu'diate between the two pan*nts, n<» ehroiuatin
elimination takes place, whilst in S(ron(/iflocentn>tus ? x Sp}i(U'rechiitiiH(^ ,
which is of purely maternal typ<', he w;us able to ob.serve the elimination
of pit-ernal chromatin during the fiixt cleavage.

On the contnirv, in the hybrid Echinus J x Antedon </, which is of
purely maternal ty|K', an<l where also the sperm- and egg-iniclei fuse
with eivch other, no eliminati«ui of paternal chromatin w;vs obsi*rved
(Goillewski jun.'') ; the latter authoi- has also observed that the chromo-
somes deriveil both from the egg- and the sperm -nuclei participate
in the first mitosis of the Qgg (fii*st cleavage), though he could not
there distinguish paternal and maternal chromosomes from each other.
Biiltzer", who confirms in general the statement of Godlewski jun,, was
able even to distinguish both kinds of chromosomes. The results of
the cj'tological studies of Sti-asburger on Fracjaria above cited are
thus in exact accordance with this discovery of Godlew.ski jun. and
Baltzer on Echinus x Antedon, because in both cases no elimination of
p{\ternal chromatin succeeding the fusion of the two nuclei does take
place. What cytological phenomena will occur in the formation of false
hybrids of Saluc should be one of the objects of our future study.

* Archil' f. Enticicklungsviechanik der Organismeu, Bd xxvii. 190l>, pp. 134—462.
» Archivf. Zelljorschung, Bd v. 1910, pp. 497—621.

' Ibid. Bd VIII. 1912, pp. 352—395.

* Archivf. KntwicklungKinechanik, Bd xxxiii. 1912, pp. 196—254.
» L.C.

* Archil- f. Kntirickhnigsmechanik. Bd xx. 1906, pp. 574—643.
' L.c.

56 On Hybridisation of some Species o/Salix

III. Conclusions.

After the appearance of the work of Wichura the view prevailed
that hybrids between various species of Salioc breed true in later
generations. My hybridisation experiments conducted on a few Salioc-
species have shown that this is not true, at least in respect to certain
characters. According to these experiments the erect habit of stem is
dominant to the spreading, the hairy character of leaves is dominant
to the non-hairy in one case, and recessive in another, red stigma is
dominant to green, and all these characters were found to exhibit
segregation in F2 generation. Hybrids between plants with stipulate
and those with exstipulate leaves exhibit a mosaic character, for some
leaves have stipules and others none ; the occurrence of segregation of
this character in F2 is not yet proven.

In the hybridisation >S'. multinervis x S. gracilistyla the so-called
0-type and the il/-type offspring, differing in catkin character, appear
in Fy. This phenomenon has not yet been explained beyond all doubt,
and various hypotheses have been proposed for it. Of the latter the most
probable is that which supposes that either one of the parents (or both)
is heterozygous in some invisible factors; the offspring derived from the
hybridisation under consideration will then carry them in different
combinations, and this genotypic difference will influence the factors
concerning the catkin character, so as to give rise in some cases to the
G^-type, and in others to ilf-type. Thus the appearance of the two
types of catkins in F^ is not due to the segregation of the catkin factors
themselves, for all F^ plants will agree in carrying the latter in the
same heterozygous condition. Their real segregation was found to take
place first in F^) the peculiar mode of this latter process has been
explained on the basis of the hypothesis adopted in the case of F^
plants.

The segregation of many allelomorphic characters has thus been
conclusively proven, but in every case the proportion of individuals
bearing each antagonistic character is very different from 3:1, 15:1,
63 : 1, etc. etc., usually seen in Mendelian hybrids. It would not how-
ever be surprising that I was unable to demonstrate the usual Mendelian
ratios in >SfaZiic-hybrids, because neither Lotsy' nor Wichler^ was able to

^ Zeits. f. ind. Ahstammungs- und Vererbungslehi-e, Bd viii. 1912, pp. 325 — 333;
IV ^ Conference Internationale de Genetique, 1913, pp. 416 — 428.

^ Zeits. f. ind. Ahstammungs- und Vererbuvgslehre, Bd x. 1913, pp. 175 — 232.

Fig. 1. Gracilistyla c? — ?

Fig. 2. Multinervis i — ?

KiK. I. M iy\H^ ♦ 7

PLATE I

Fig. 3. G-tyi^e cT — ? .

Fig. .3. Mui;inn:'j cT — ? .

8. Ikkno 57

diaoem them in species- hybrids of Autirrhinutn iind Dianthxis, which
were studiiHi in detail fnun the Htiind|H)iiit of iiuMlem >(i*nftic Mciciicv,
though there were some ran* exceptions in the former ^enus. It is
ver)* likely thiit in siirh ruses u ^n/iit iiumlMT nf fju't^)rs are nuKUTiHKl
in the development of each character, and conH«Mnn«ntly a coui))lex
Si'gre^tion takes place in F«, though it is e<)ually undeniahle that
Ihis Si»Krej^tion may Ih' suhject to some law other than Mrndrlian,
hitherto unknown t^) ns.

The plants arising; lus the result.s nf the hybrid isiition d«»rje in IJMO
between ^'. m»///i*/*errKs' % xS. uvacxlistyla ^ were real hybrids, but those
pnxluciHl fmm the hybridisation dune in l!)ll Ix'tween the s.ime male
and female tn»es were the so-GdK'd "falsi' hybrids" of purely maternal
type because they were nothing but N. innltinervis. They have not yet
been prove<i to breed true, though this is probable. All false hybri»ls
thus obtaineil were of one sex, namely female. Cytological investigations
are necessarj- heiv.

In conclusion, 1 wish to thank Mr S. Xohara, Mr Y. Tanihara and
Mr M. Ando, who were ver\' helpful to me during the present investi-
gation.

EXPLANATION OF PLATE I.

Some of tlie figures are photographs by Dr Koininami, to whom my thanks are due.
All slightly reduced from natural size.

Fig. 1. Salix (jraciliMijla. Above d" , below ? .

Fig. 2. Sail J iiiiiltiuervis. Above (^ , below ? .

Fig. 3. Multinervig x gracilistyla i*', , G-type. Left s , riglit ? .

Fig. 4. Multivervig \ pracilistyla /•'] , J/-type. Left d" , right 9 .

Fig. 5. Mutant (?) arising from the fertili.sation .y-tyjx^ x r;.type. Above d , Inflow ? .

58 On Hybridisation of some species q/*Salix

Table showing Successes and Failures of Hybridisations of VaHaus
. Species 0/ ^a\ix*

Successes
Purpurea ?

X Purpurea multinenus ^

X Oracilistyhi s

X Viminalis i
Purpurea multineiris ?
• X Purpurea sei'icea s

X Qracilistyla cf

X Opaca s

X Viminalh j
Qracilistyla ?

X Purpurea s
Viminalis ?

X Pm-purea i

X Purpurea multinervis d

Fallm-es
Purpurea ?

X Pui'purea sericea i

X Opaca s

X Triandra uipponica ,f
Purpurea sericea ?

X Purpurea multinervis i

X Gracilistyla <s

X Triandra nipponica i

X Viminalis S
Oracilistyla ?

X Puiyurea multinervis cf
Opaca ?

X Purpurea multinert^is i
. X Gracilistyla s

X I'liandra nipponica s

X Viminalis i '
Caprea ?

X Purpurea multinervis i

X Opaca <s

X Triandra nipponica i

X Viminalis s
Triandra nipponica ?

X Caprea <?

' Seeds from this cross were obtained by Mr Nohara.

STUDIES OF INHEIUTANCK IN THK JAPAN KSP:
CONVOLVULUS,

By B. MIYAZAWA.

Agriculinml Expeinmeut Station of the Prefecture uf Kaimganui,

near Yokohama.

(With PlaU^ II and One Text-figuiv.)

[Xote. In the pivsont stiiU* of the postal service it has not been
possible U) submit this piper to the author for revision. Ei>l).]

CONTENTS.

rA(»E

Introduction <iO

Experiments 61

(a) Fi generation 61

(/>) F^ , 62

1. Leaf colour 63

2. "Hukurin" 63

3. Flower colour 65

(r) ^3 generation 66

1. Leaf colour 6(J

2. "Hukurin" 67

3. Flower colour 67

{(i) F^ generation "*•

1. Leaf colour 70

2. "Hukurin" 71

3. Flower colour 71

(<•) Back-croHsing and F> .... 7*

DiscnsHion of results 7.>

Summary ^^

Explanation of Plate II ^'-^

60 Studies of Inlieritance in the Japanese Convolvulus

Introduction.

The Japanese Convolvulus, closely related to the Morning Glory of
the Americans and known under the popular name " Asagao\" is very
extensively cultivated here since immemorial time as an ornamental
plant, and contains an abundant number of races which are characterised
by remarkable variation in the form and colour of leaves as well as
flowers. As I have been studying the hereditary behaviour of several
characters in this species for some years, and have reached definite con-
clusions in some respects, I am going to publish here the results of these
investigations. All experiments contained in this paper were conducted
in my garden in Yokohama.

The inheritance of the Japanese Convolvulus has already been studied
by three authors, Tanaka^, Toyama^ and Takezakil Of these I will
speak below only about the investigations of Takezaki, some of whose

Corolla with "hukurin," seen from above.

1 This plant has been variously called by our systematists Ipomoea hederaeea, Pharhitis
hederacea, P. Nil, etc., and I am not able to decide myself which name is really the
right one.

2 Idengaku Kyokivasyo (A text-book of Genetics in Japanese), Tokyo, 1915, pp. 32 ff.
and96ff.

3 Nippon Ikusyugakukicai Kicailio (Journal of the Japanese Breeders' Association),
I. 1, 1916, pp. 8, 9.

4 Ditto, pp. 12, 13 with many tables.

B. MiYAZAWA HI

results Are in tign'iMuent with iiiiiu\ A(*c(inliii^ t<i him ihv ^rctn coioiir
of leaves behaves as dominant UiwanlH the* yellow (rhlttrina), and in
Ft the ratio of grtn^n and yell«»w plants in ',\: 1. Thr ^'enrtie lN>havioiir
of flower colour is very complex, hut if we elaHnify pinntM simply into
thitse with e<»loured tind those with whitr flowi'i-x, white in r<'e<'HMi\r.
and in f\ the ratio of the two kin<ls of plants is .*{:!. In sninr **(
these ct»loured flowers tht' eorolla is whitr at its margin, sn as to torm
a nng-sha}KMi white patch (sre the tcxl-Hi(.), what .lajwmese ganhnirs
call the "hukurin'." Take/jiki studied the inhiTitJUJc** of whit«-
niargineil flowers, and foun<l that the " hukurin "' is pnKluc<M| hy a
special fjict4>r acting as a white dominant at the margin of the corolla
so that the hybrid bt^ween a nice with white-margined flowers and
another with fully -colounni ones wjis foun<l t^> pnKluce the former kind
of flowers in F^ and to segregate in F.. int«» th*- ratio '} white-margined : 1
fully-coloured. Mori'over, he rej)orted that in certain cases there is even
a factor which inhibit>s the action of that prmlucing the "hukurin"
part.

Experiments.

The plants originally used in my experiments are characterised as
follows :

A. Leaf is yellow^ {chlorina) (PI. II, fig. 6), and flower white, though
its throat is tinged with extremely light magenta (PI. II, fig. 2).

B. Leaf is green (PI. II, fig. 5), and flower dark-red' (PI. II, ^y^. I ).
These two j^arents were cultivated for two years before my exjxTi-

ments had begun, and since then this cultivation has been continued
during five years. Both of them were found during cultivation to
breed true entirely to their resjK^ctive types.

In 1913 I performed the hybridisation between these two plants in
both reciprocal ways, and in 1914 three individuals from each were
grown for the purpose of further experiments.

((/) F^ Generation.

Leaf was green : that is, green is dominant to yellow. Flower-
colour was entirely different from that of either parent, and was light

' I shall sometimes use this word to indicate such a white patch.

* The word "yellow'' is used always in this paper for brevity's sake, but naturally it
means yellowish green.

' This colour corresponds nearly to No. 42 (Kouge) of the "Code des Couleurs" by
Klincksieck and Valette, Paris, 1908.

62 Studies of Inheritance in the Japanese Convolvulus

magenta^ The corolla is not however fully coloured, and it is white at
its margin not wholly, but only near each of the five notches of its limb.
Such white patch is also called " hukurin," and the words " hukurin"
and " white -margined " used below refer always to flowers which are
edged with white partially in such way. Both reciprocal hybrids were
entirely similar to each other (PI. II, fig. 3).

(6) F2 Generation.

The mode of segregation of flower-colour in F^ is rather complex.
Not only are there found flowers of white, dark-red, and magenta
colour exactly similar to that of the two original parents and the F^
plant, respectively, but we have also those of scarlet colour (PI. II, fig. 4),
and in each of these colours — dark-red, magenta, and scarlet — there are
three gradations of their intensity, sharply distinguishable from each
other. The detailed study of the segregation of flower-colour is now
under way, and will be dealt with in a future paper. For the present
time, for simplicity's sake, I will call magenta and scarlet simply by
the collective name red, and make no distinction of the intensities of
colour just noticed.

The details of the segregation of leaf- and flower-colour in F^ are
shewn in Table I.

TABLE I.

F,

generation

•

White-margined
Leaf- Flower- or
colour colour fully-coloured

A

XB

BxA

Grand
totals

«a ''

h

c

Totals

d'^

e

"^ /~~

"t^Is

' j white-margined
'" \ fully-coloured

20

7

46

73

9

4

48

61

134

6

4

16

26

5

1

20

26

52

green -^

, , , (white-margined
dark-red-^, „ , ^
1 fully-coloured

9
4

2

0

29
15

40
19

5
2

5

5

22
10

32
17

72
36

^ white

15

4

23

42

11

5

36

52

94

/ , ( white-margined
red ... w „ 1 •,
( fully-coloured

5

5

27

37

3

8

30

41

7&

1

2

14

17

2

2

14

18

35

yellow -

dark.redJ;*;''-°,"8'°;'^
( fully-coloured

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

white

4

2

13

19

1

1

11

13

32

Totals

64

26

183

273

38

31

191

260

533

We will consider now

leaf-colour, "

hukurin "

and flower-colour

separa

Ltely.

1 This lies between No. 566 and 571 (Violet rouge) of the " Code des Couleurs."

2 The letters a—f in the Tables indicate the different individuals of the 6 Fi plants
which were bred from.

B. MiYAZAWA 63

1. Lmtf -colour,

Tho ri'tiulU of my iiiV(>Mti^ition am* in |H'rf<'ft jicronl with th«)H«' of
lakoxnki (p. tU). aiul it will hv rradily mth fn»in Taihir II that hciv
the Mi*gregiiti<m ciccurs in the Hiniplost Mcn<lilian liwhion.

TABLE II.

UomilU KxiH»cU«1

F\ plant* lirecii Yellow ToUU

.lx/?(« + fe + r) 200 73 273

Hx. 4 (<! + «+/) 188 72 2(»0

ToUIb ... 388 146 533 399-76

2. '* Ifuhirin."

As bi'tbre stjitod, in spito ot" thi' fact that neither the one nor the
other of the parents shews externally any sign of the " hnkurin," this
chanietvr appears in the F^ plants, and moreover, it will be seen from
Table III that in Fo the ratio of plants with white-margined and
those with fully-coloured Howers is 8:1. As of coui*se we cann(^t
distinguish between the white-margined and the non-white-margined
condition in perfectly white flowers, plants with the latter kind of
flowers are not included in this Table.

TABLE IIL

Kesults

Exi>ected

/*, pUnU

White-
margined

FuUy-
culoured

Total!

White-
5 margined

Fully-
coloured

a

i

AxB (a)

34

11

45

33-75

11-25

:l: 0-75

±9186

.. W

14

6

20

1500

5-00

± 1-00

i 1-937

ic)

102

45

147

110-25

36-75

± 8-25

±5-250

BxA id)

17

9

26

19-50

6-50

± 2-50

±2-208

(e)

17

8

25

18-75

6-25

± 1-75

±4 165

M - (/)

100

44

144

108-00

3600

± 8-00

±5196

Totals 284 123 407 305-25 10175 ±21-25 ±8-736

From the above table we see that the number of plants with white-
margined flowers really obtained is always smaller than might be
theoretically expected, except in ^ x B(a). We have however to make
here the two following remarks. In the first place, the area of the
" hukurin " piirt was very variable according to individuals, notwith-
standing the fact that all plants were grown under exactly similar
conditions. Thus not rarely the " hukurin " was represented by very

' Deviation from the theoretical number.
2 Standard error.

64 Studies of Inheritance in the Japanese Convolvulus

insignificant white spots in the five notches of the corolla; moreover,
even in one and the same individual, which has very slightly white-'
margined flowers, I was able to discern the " hukurin " sometimes clearly
but sometimes not at all, according to different stages of their develop-
ment, so that it would not be improbable that some plants with such
very slightly white-margined flowers were erroneously entered as being
without them. Secondly, I have learned by experience that the mode
of cultivation has great influence over the production of the " hukurin."
Plants were generally grown in a field, but some of them were cultivated
in pots, for example a certain number of (c) and (/) in Table III. The
difference of the results due to the method of cultivation will be
explained by reference to Table IV.

TABLE IV.

Cultivated in field

Cultivated in pot

White-

Fully-

White-

Fully-

Fi plants

margined

coloured

Totals

margined

coloured

Totals

AxB

{c)

38

13

51

64

32

96

BxA

if)

53

19

72

48

24

72

Totals

91

32

123

112

56

168

Percentage

73-98

26-20

66-66

33-33

As will be seen from the above Table, while in the field culture
plants with white-margined flowers and those with fully-coloured ones
are 74 and 26, respectively, i.e. are almost exactly in the ratio 3 : 1, in
the pot culture there are 67 and 33, respectively, i.e. the number of
plants with white-margined flowers is relatively much smaller in the latter
case than in the former. Plants in (a), (6), (d), and (e) were all cultivated
in the field, and we see that here the ratio of the two kinds of plants is
nearly equal to 3 : 1 in each case, and the very small deficiency of plants
with white-margined flowers from the theoretical expectation in these
cases may be probably due to the first of the two causes above mentioned.
That in the case of pot culture we see always a definite deficiency,
may be perhaps due to the fact that pots are generally too dry in
summer without special precautions. As is well known through the
investigations of several botanists, the formation of anthocyanin in leaves
is very much accelerated when leaves live under very dry conditions.
Thus, according to Wheldale^ we see the development of anthocyanin
mPelargmiium which was insufficiently watered ; also Miyoshi^ observed

^ The Anthocyanin Pigments of Plants. Cambridge, 1916, p. 24.

2 Journ. Coll. Science, Tokyo Imp. University, Vol. xxvii. 1909, pp. 1 — 5.

B. MiYAZAWA 65

that leaves of triH»8 in the Kiwt Indios, Ctylon and .lava n(i(hn <luriu>(
the dry period in the same way as autuiniial leaves do in the tein|x'nite
regions. Again Pellew' n.»|X)rU that the amount of pigment in |M'tals
of Inith whiU* and blue plaiitw of Camfxttnila carpatkit varies aceonling
to the moisture condition of the soil, tlowi-rs hecoming much darker
after rain.

In our Ciise it would not therefore be unlikely that owing to the
summer drought some anthocyanin would devel(»p in the " hukurin "
part and make whito-margine<l tiowei-s look like fully-colouriHl ones,
especially in plants grown in pots.

3. Flower-colour,

As will be seen from Table I there occur no dark-red Howers in
yellow-leaved plants. As flowers of this colour are found exclusively on
green plants, it might perhaps be concluded that some coupling or
repulsion took place between flower- and leaf-colour, l^ut such is not
really the case, jis will be e;\sily seen from Table V.

TABLE

V.

Leaf-
colour

Results

Totals

Expected

a

F,pUnt8

No. of
coloured
flowering

plants

No. of

white

flowering

plants

No. of
coloured
flowering

plants

No. of

white

flowering

plants

£

AxB(avh + c)

Green

158

42

200

15000

50 00

iS-OO

±6124

BxA{d + e+/)

»t

136

52

188

14100

47-00

±5 00

±5-937

Totals

294

94

388

291-00

97-00

±3-00

±8-592

AyB{a-i-b + c)

Yellow

54

19

73

54-75

18-25

±0-75

±3-700

BxA{d + e+f)

>>

59

13

72

54-00

18 00

±500

±3-674

Totals

113

32

145

108-75

36-25

±4-25

±5-241

Gross totals

407

126

533

399-75

133-25

±7-25

±9-997

In this Table plants are classified into two groups according to their
flower-colour, i.e. those with white and those with coloured flowei-s.
From this we see that the ratio of individuals of these two classes, both
in green as well as yellow plants, is 3 : 1, despite the fact above noticed
that in the latter there were found no plants with dark-red flowers. It
will be readily understood from these considerations that we have here
to deal with neither coupling nor repulsion.

In green plants the number of those with red, dark-red and white
flowers respectively is in the ratio 2 : 1 : 1, as shewn in Table VI.

J Jounud of Genetics, Vol. vi. pp. 317—339.
Joum. of Gren. viii

66 Studies of Inheritance in the Japanese Convolvulus

Fi plants
J xB(a + &4-c)
BxA{d + e+f)

Results

TABLE VI.

Expected

Red
99

87

Dark-
red White Totals

59
49

42
52

200

188

Dark-
red White

Red

100

94

50
47

50

47

±8-00
±5-00

R + W:D «

± 9-00 ±6-124
± 2-00 ±5-937

Totals

186 108 94 388 194 97 97 ±3-00 ±11-00 ±8-529

(c) Fs Generation.

Seeds were obtained from 31 J^2 plants, with which to study the F^
generation. The following are the results of these studies.

1. Leaf -colour.

It will be seen from Table VII that in respect to leaf-colour we
have obtained exactly the same results as in F^ (compare Table II).

TABLE

VII.

Pedigree

Results

Expected

No. of

~N

f 2 plants

Green

YeUow

Totals

Green

Yellow

a

«

7

8

6

14

10-50

3-50

±2:50

± 1-620

9

36

7

43

32-25

10-75

±3-75

±2-889

22

76

26

102

76-50

25-50

±0-50

±4-373

31

80

19

99

74-25

24-75

±5-75

±4-265

44

39

11

50

37-50

12-50

±1-50

±2-810

45

10

3

13

9-75

3-25

±0-25

±1-561

11(«)

12

6

18

13-50

4-50

±1-50

± 1-836

14(a)

6

2

8

6-00

2-00

±0-00

±1-225

15(a)

11

5

16

12-00

4-00

±1-00

±1-732

15(6)

12

1

13

9-75

3-25

±2-25

±1-561

39(6)

7

1

8

6-00

2-00

±1-00

±1-225

Totals 297

87

384

288-00

96-00

±9-00

±8-485

From their behaviour in F^ it was apparent that 20 of the 31 jPa
plants were homozygous for leaf-colour. Of these 10 were green and
10 were yellow. Table VIII gives the total number of F^ plants
obtained from these 20 homozygous F^ individuals.

TABLE VIII.

Leaf- Total number

colour of of families

F<i plants of F-^ plants

Leaf-colour
of f 3 plants

Green
Yellow

10
10

Green

241

0

Yellow

0
564

B. MiYAZAWA 67

We have got exiictly the same iwsuIIh iih in h\, ixh nht'svu in
Table IX (compare Table III).

TAHLK

IX.

R«>ulu

ToUU

WhiU.
nuMKinecl

Fully
coloured

A

/>,pUnU

WhIU
nuuflowl

Fulljr
coloured

«

14

43

21

64

4H()0

1600

±500

±2 289

16

37

11

48

3600

1200

±1000

±3-00<)

19

So

14

49

36-75

12-25

±1-75

± 3030

81

76

23

99

74-25

24-75

±1-75

± 4-30H

33

30

7

27

20-25

6-75

±0-25

± 2-250

38

18

6

23

17-25

5-75

±0-75

± 2-077

39

32

11

43

32-25

10-75

±0-25

±2-839

55

24

13

37

27-75

9-25

±3-75

±2-684

11(a)

11

3

14

10-50

3-50

±0-50

±1-620

15(6)

10

2

12

900

300

±1-00

±1-500

ToUIs 306 110 416 31200 10400 ±600 ±8832

From the above Table we see that the ratio of plants with white-
margined and with fully -coloured flowers is 3:1, and in this case, when
we compare the ratios of the number of these two kinds of plants in the
field- as well as in the pot-cultures to each other we see also in the
latter case a certain deficiency of plants of white- margined flowers.

We have got 4 families of plants which contain the " hukurin "
factor in homozygous condition, and 9 families where it is entirely
absent, as shewn in Table X.

TABLE X.

Total number of White- Fully-

families in F^ margrined coloure<l

White-margined ... 4 109 0

Folly -coloured ... 9 0 301

3. Flower-colour.

It would be a priori easily seen from the results in F.. that all F..
plants with white flowers will produce in i^j again those with white ones.
Though I could not obtain many seeds from plants of these families the
results shewn in Table XI will fully confirm this expectation.

We could find no families of plants which breed true constantly to
dark-red flowers.

5—2

68 Studies of Inheritance in the Japanese Convolvulus

TABLE

XL

Pedigree

Leaf-
colour of
F3 plants

D^

Flower-colour of F, plants

No. of
t\ plants

rk-red

Red

White

{ ^

green

0

0

23

a \ 10

>»

0

0

10

Ue

»>

0

0

12

)8 56

yellow

0

0

56

r

7
7

45

J green
1 yellow

0
0

0
0

8
6

1 green
yellow

0
0

0
0

10
3

Totals

128

We have obtained the two families of plants which breed true to
yellow leaves and red flowers, as indicated in Table XII.

Pedigree

No. of

Fi plants

16

49(6)

Leaf-
colour of
J3 plants

yellow

TABLE XII.

Flower-colour of F-^ plants

Dark-red
0
0

Eed

48
18

White
0
0

Totals

The families which segregate into plants with red and those with
white flowers are found only among yellow F^ plants ; in F.^ the ratio of
red and white is 3 : 1, as shewn in Table XIII.

TABLE XIII.

Pedigree Leaf- --'"

No. of colour of Dark-
JF2 plants Fz plants red

17
18
19
38
39
59
60

yellow

Results

Expected

Red White Totals

64 20 84

25

13

6

7

20

77
49
23
43
35

102
62
29
50
55

Dark-
red

0
0
0
0
0
0

0 46

22

68

Red
63-00
76-50
46-50
21-75
37-50
41-25
0 51-00

White
21 00
25-50
15-50
7-25
12-50
13-75
17-00

±1-00
±0-50
±2-50
±1-25
±5-50
±6-25
±5-00

±3-969
±4-373
±3-410
±2-332
±3-062
±3-211
±3-571

Totals

337 113 450

0 337-50 112-50

±0-50

±9-186

The families of plants which segregate into those with dark-red
and those with white flowers are found only among those which remain
constantly green in F.^, and the ratio of dark-red and white is 3: 1, as
shewn in Table XIV.

B. MiYAZAWA

00

TABLE

XIV.

RmmiXlM

Kx|«ci«.l

PMllflW

oolottrof

iUit

Kml

WhIU)

•

r.^li

lUtrk

WhlU

> TouU

«

14

green

0

65

14

79

0

ft'i-i.'i

i«-7r,

L 5-75

i 8H4«.I

Si

0

•27

H

H.'S

0

'HV2rt

8-7r.

i o-7r>

J '2'rsi

55

0

87

11

4H

0

:moo

12-00

tioo

laooo

11(6)

0

7

8

10

0

7-50

2r,o

! or>o

xl 870

«(•)

0

S

•2

."»

0

3-7-,

1-26

L o^r,

i 0 (HJ7

M(6)

0

8

1

<J

0

(r7r>

2-25

Ll^r,

i 1-29<J

t4(6)

0

«

2

11

0

8-26

2-75

10 75

Ll'130

Totaln

0 16f. 41 1U7

1477/

49-25

8- 25

fc « 078

The fkiniiies of plants which segregate int^) those with re<l and those
with dark-nni flowers are found only among those which segregate into
green and yellow plants in Z'.,. The results are in<licated in Ta})I«' X\'.

TABLE

XV.

Fe.lMrm.

Le*/-
colour of
Fj plants

KesulU

No. of
Fi i»Unts

Ke<l

l>ark re<l NS hit*

T(>Ul8

Oraiyl toUlu

9

green
yeUow

23

7

13
0

0
0

30 i

7

43

31

green
yellow

48
19

32
0

0
0

80 }
19 i

1»9

44

i green
} yellow

22
11

17
0

0
0

39 j

50

14 (a)

S green
) yellow

4
2

2
0

0
0

2]

8

15(a)

green
yellow

G
5

0

0
0

5(

10

15(6)

J green
) yellow

2

1

9
0

0
0

'li

12

39(6)

(green
1 yellow

3

1

3
0

0
0

;i

7

In the above Table the total number of individuals with red and of
those with dark-re<l flowers in green plants is 108 and 81, respectively,
and although this ratio seems to be somewhat different from the exjx*cte<l
2 : 1, yet the deviation lies within the range of thrice the standanl ermr,
because the former is ±11 and the latter is ± 6-481,

We had the two families of plants which segregate<l in the same way
as in F^j as shewn in Table XVI.

We have no yellow plants with white flowers in No. 11 (d), but this
is no doubt due to the small number «»f experiments. In other families

'0 Studies of Inheritance in the Japanese Convolvulus

TABLE 'XVI.

Results

Expected

Pedigree

No. of

f 2 plants

22

11 (a)

Leaf-
colour

green

No. of
coloured
flowering

plants

No. of

white

flowering

plants

Totals

54

21

75

8

4

12

No. of
coloured
flowering

plants

56-25
9-00

Totals

22 yellow

11 (a)

62

18

25

8
0

87

26
6

65-25

19-50
4-50

No. of

white

flowering

plants

18-75

3 00

21-75

6-50
4-50

±2-25
±1-00

±3-25

±1-50
±1-50

±3-750
±1-500

±4-039

±2-208
±1-060

Totals 24

Gross totals 86

8
33

32
119

24-00
89-25

24-00
89-25

± 0 ±2-450

±3-25 ±4-724

we see that the ratio of individuals with coloured and white flowers
is 3 : 1.

In the above Table, if we classify flower-colour of green plants into
red and dark-red, we see that their ratio is 2 : 1, as was the case in F.^,
(see Table VI). This is shewn in Table XVII

Pedigree

No. of

Fi plants

22

11(a)

Red

35

6

Results

Dark-red Totals
19 54

2 8

TABLE XVII.

Expected

Red

36-00

5-33

±1-00
±0-67

±3-697
±1-334

Totals

41

21

62

41-33

20-67

±0-33

±3-712

{d) F4, Generation.

Seeds were obtained from 43 F^ plants. The results in F4, are shewn
below.

1. Leaf -colour.

Only the families of plants which segregated into green and yellow
plants are shewn in the following Table.

TABLE XVIIL

Results

Expected

Pedigree

^^

a

± 2-75

No. of
Fi plants

9—23

Green

74

Yellow
21

Totals
95

Green
71-25

Yellow
23-75

a

±4-221

22— 1

78

30

108

81-00

27 00

± 3-00

±4-500

22— 4

51

16

67

50-25

16-75

± 0-75

±3-544

31— 1

43

15

58

43-50

14-50

± 0-50

±3-298

31—10
44— 2
44— 6

39
79
34

9
15
10

48
94
44

36-00
70-50
33-00

12 00
23-50
11-00

± 3 00
± 8-50
± 1-00

±3-000
±4-198
±2-693

Totals

398

116

514

385-50

128-50

±12-50

±9-817

B. MiYAZAWA 71

In the alxwe Tablo tlu» doviatinn nf thr t<a4il nuiulMT (=» t 12«r)()) in
somewhat Iargt»r than the Htniulanl eiT«»r (- ± J) 817), hnt the ciifliTfnce
betwtH?n them in not very hirg«'. FurthtTiiiore, if w«» examine each
family sejianit^yly we siH» that only Id N<>. 44—2 in th<» drviatinn larger
thaii the Htandanl ern»r hut evi-n hon« not lar^or than twice the laM^T.
so that the i\*8ult^ in this ease an» similar to thnsi* gained in F.^ and /*,
(see Tablet! II and VII).

2. ** Iful'uriii.'*

SecMJs were «)btaine<i from 17 F^ plants with white-margined flowers
and we had in /\ H fan\ilies of plants which hri'ed true to the " hukurin "
condition. The results from 9 families of plants which exhibited the
segregation are shewn in Table XIX.

TABLE XIX

•

R««ulu

KxiKjctecl

NoTof
F.pUnU

Whlt«-

FuUr
coloured

loUls

White-
1 margine<l

FuUy-
coIoure<l

a.

i

14— 2

55

26

81

1)0-75

20-25

±

5-75

± 3-897

14- 9 .

74

22

96

72 00

24 00

±

2-00

± 4-243

81-10

:«

13

48

3600

12 00

±

100

x 3-(M)0

32— 7

29

6

35

26-25

8-75

u.

2-75

- 2-562

38— 2

21

9

30

22-50

7-50

±

i-r,o

:i. 2-535

39- 1

69

29

98

73 50

24 50

.u

4-50

± 3-824

39— 3

26

11

37

27-75

9-25

±

1-75

± 2-633

39- 7

So

33

118

88-50

29 50

-i-

3-50

± 4-704

55— 4

34

14

48

36-00

12-00

±

2-00

i 3-000

Totals

428

163

591

443-25

147-75

±15-25

±10-527

From these results we see that they are in perfect agreement with
those obtained in F^ (see Table IX). As the number of plants with
white-margined flowers was relatively smaller in jx>t- than in field-
culture, we find always some deficiency of plants with white-margined
flowers under the ex{)ected number.

3. Flower-colour.

The results are shewn in Table XX, («)—(/<).

The segregation shewn in {a) is similar to that occurring in F. ; that
shewn in (6) and in (c) is, respectively, similar to that shewn in Tables
XV and XIV. The segregation shewn in {d) w;us not observed in F^.
The segregation shewn in (e), (/), {(j) and {h) is similar to that in
Table XI (a), Table XIII, Table XII, and Table XI (yS), respectively.

72 Studies of Inheritance in the Japanese Convolvulus

TABLE XX.

(a)

(6)

(c)

[d)

{e)

Pedigree No.

of F3 plants Leaf -colour

22— 4

9—23

22— 1

31— 1

31—10

44— 2

green
yellow

green

yellow

\ green

\ yellow

j green

\ yellow

( green

I yellow

green

yellow

Red

23

13

51
21
51
30
29
15
22
9
54
15

Dark-red White
11 17

0 3

23

0
27

0
14

0
17

0
25

0

Totals
44— 6

14— 2
22—11
32— 7
55— 4

J green 207

\ yellow 90

( green 0

\ yellow 0

green

106

0

34

10

81
18
35

48

Totals

/ 9-4
14—19

31— 9

32— 6
44— 3

green

182

35
96
18
26
35

Totals

/ 10— 1
10— 5
23— 1
32— 5
45— 1

green

210

0
0
0
0
0

Totals

,22— 3

38— 1

39— 1
39— 2

(/) \ 39— 3
59—12
60— 1

yellow

57
24
98
39
37
24
63

0
0

0
0

25
6

17
9

57

0
0
0
0
0

5

5

11

51

10

82

20
8

40
16
13
6
24

Totals Grand totals
51
16

74
21
78 \
30 (
43
15
39
9
79
15

313
90
34
10

106
24
52
57

239

35
96
18
26
35

210

5

5

11

51

10

82

77
32
138
55
50
30
87

67

95
108
58
48
94

403

44

Totals

342

127

469

B. MiYAZAWA

73

TABI.K XX (routinfir,/).

(9)

(M

f9ditrv9 No

l.i«f colour

K«l

iHirk rwl

Whilr

TouU

16 - 4

vol low

8<i

0

0

H<1

16- 9

»1

0

0

31

17 18

31

0

0

31

17- M

84

0

0

34

17— lii

88

0

0

33

' 81- S

80

0

0

30

>-«

30

0

0

3<»

88— H

27

0

0

27

89— 7

IIH

0

0

IIH

44- 1

2H

0

0

2H

44-10

86

0

0

H6

69-11

29

0

0

2<l

V ToUU ...

563

0

0

563

(17-7
1 56- 1

yellow

0

0

:^o

30

,,

0

0

4(

4i

1 59-10

,.

0

0

25

25

ToUls

«.♦'.♦

We may now consider some {K)ints in connection with Talile XX

(a) In green plants we have 84 phints with coloured and 17 with
white flowers, and the ratio may be taken as 3:1, the deviation and the
standard error being + 4-250 and ± 8092, respectively. Of the above
34 plants with coloured flowers 23 individuals have red and 11 dark-
red ones, thus their ratio is 2:1, the deviation and the standard error
being + 0-333 and ± 2-450, respectively.

In yellow plants we have 13 plants with red and 3 with white
flowers, and the ratio may be taken as 3:1, the deviation and the
standard error being + 1-000 and ± 1-732, respectively.

(b) In green plants, except No. 44 — 6, we have 207 plants with
red and 10(5 with dark-red flowers, thus their ratio is 2:1, the deviation
and the standard error being ± 1-667 and ± 8-340, respectively. In
No. 44 — 6 I could find, in spite of careful observations, no green plants
with red flowers, but here some yellow plants with dark-red flowers
made their apjx^arance, which were never found otherwise. This
peculiar case will be the subject of my future paper.

(c) We have 182 plants with dark-red and 57 with white fl(»wei*s,
thus their ratio is 3:1, the deviation and the standard error being
± 2-750 and ± 6 694, respectively.

(d) All breed true to dark-red flowers.

(e) All breed true to white flowers.

74 Studies of Inheritance in the Japanese Convolvulus

(/) W.® l^^ve 342 plants with red and 127 with white flowers, thus
their ratio is 3:1, the deviation and the standard error being + 9-750
-and ± 9-377, respectively.

(g) All breed true to red flowers.

Qi) All breed true to white flowers.

As will be seen from above, there are some cases where the deviation
is larger than the standard error, but these differences are not very large,
and it may be safely concluded that the results of all these experiments
^re in accordance with expectation. Furthermore, from the results of
jPg and F^ we may deduce the following facts :

1. No homozygous green plants with red flowers were found.

2. In the offspring derived from green plants with red flowers leaf-
■colour always segregates into green and yellow, while the segregation
of flower-colour is either exactly similar to that in F^, or different from
it, in so far as no white flowers are produced.

{e) Back-crossing and F^.

In 1916 the back-crossing of one .^i plant {— A x B) by both of
the two parents was done.

The results of Fj x A are indicated in Table XXI.

r

rABLl

: XXI.

Leaf- colour

Ked

Dark-red

White

Totals

Green

33

0

48

81

Yellow

45

0

38

83

Totals ... 78 0 86 164

In this case the flower was either white or magenta as in Fj , and all
coloured flowers were white-margined.

From the results in F2 it may be a priori expected that the ratios of
green and yellow plants, and that of red and white flowers, are 1 : 1
respectively. Indeed we have obtained 81 green and 83 yellow plants,
thus our expectation was so perfectly fulfilled that no further com-
ment is necessary. The ratio of plants with red and white flowers is
■equally 3:1, the deviation and the standard error being ± 4-000 and
X 6-043, respectively.

The results of back-cross F^x B are shewn in Table XXII.

In this case we should expect from the results of F2 that leaf-colour
would remain constantly green, that the ratio of plants with white-
margined and fully-coloured flowers would be 1:1, and finally that the

B. MiYAZAWA 75

TAHLK XXII.

, 4 Whit«niarKinc<l 'Ih /

I "*^ f Kullyoolouml n i *'^

Orwin ,^ . , < \Vhil4'iniirRined 10 i

f Fiillycoloured 26 )

' While 0

Yellow .. 0

ratio of planus with ivd and dark tcmI Howcrs would Ix- also 1:1. I>a
118 now oxaniino Table XXII to soo whether or not our exp'ctation is
fulfilliHl. Firstly, all plants an' green. Secondly, there are -Mi plants
with white-margined and 40 fully-coloun'd flowers, thus their ratio is
1:1, the deviation and the standard t'rror being ± 5-500 and ± 4 oOO,
respt»ctivi'ly. Again, there are 4() plants with red and .S5 plants with
dark-ixnl rtowt'i><, the deviation and the sUuidard error being espial to
those of the latter axse, respectively. Thus we see that in every ca.«M'
the deviation is larger than the standanl error, but the diti'erences
betw^een them are not large, so that it would not be unrejusonable t(»
consider that we see in both c;ises segregation in the ratio 1:1.

The F^ plant used in the back crosses just above mentioned was
self-fertilised ; and the results of the examination of the F. generation
thus obtained, consisting of Gol individuals in all, have fully confirmed
those shewn in Table I.

Di.scus.sioN OF Results.

It will be readily seen from all the experiments above mentioned
that the hereditary behaviour of leaf-colour is in exact accordance with
that obtained by Takezaki (p. (jl).

The results on the "hukurin" are also the same, at least in some
cases, as those reported by him, and in such crises the presence of a
factor for producing the " hukurin " part has been duly proven.

If we simply classify plants into those which can produce anthocyanin
on the corolla, at least partially, and those which cannot, their ratio in
Ft, F3, ete., is 3:1. Now since, for the formation of anthocyanin, at
least two factors are necessary, we may denote them by C and R, re-
spectively. Then the dark-red colour is to be repre.sent<'d by CCRR.
The white colour, a« we may infer from the results of experiments,
should have one (»f these factors; supjwse the latter to be C, then

76 Studies of Inheritance in the Japanese Convolvulus

the two parents and the F^ hybrid are to be represented as follows,
respectively :

parent A = CCrr,

5 = CCRR,
F^ = CCRr.

From these considerations it will be quite evident that the ratio of
plants with coloured and white flowers is 3 : 1.

I will go now to the consideration of the interrelation existing
between the hereditary behaviour of leaf-colour and dark-red flower-
colour. Flowers of the latter colour never appear in yellow plants but
exclusively in green ones. It was stated before that this constitutes no
case of coupling or repulsion (p. 65), and the results of experiments which
are now to be described led me to the conclusion that in the presence
of a certain factor D, tlte flower is either dark-red or of some other colour
according as the green factor G is in either homo- or heterozygous con-
dition (or altogether absent).

There are many instances in which the intensity of flower-colour
varies according to the homo- or heterozygous condition of the factor
concerned in pigmentation. Thus the flower-colour was found to be
lighter in heterozygous than in homozygous individuals, for example in
Atropa Belladonna^, Datura Tatula x D. Stramonium'^, Linum usitatis-
simum'^ and Antirrhinum majus^. Although our case has not to deal
with the intensity of flower-colour, I think that it has to be ranked
among the same class of phenomena as those above cited. Similar
examples are also found in respect to the pigmentation of other plant
organs, as in Corchorus capsularis^, Egyptian cotton^ Indian cotton^, and
Phaseolus vulgaris^. Saunders reported an interesting case of the con-
nection between the factors for hoariness of leaves and flower-colour in
Stocks'^ Colour is due here to the presence of two factors C and R in
the zygote. In certain strains of Stocks, the hoariness of the leaves has
been found to depend also on the presence of two factoids H and K.
Between these two pairs of factors there is a certain relationship, viz.

^ Bateson and Saunders, Rept Evol. Com. Roy. Soc. 1901, pp. 1 — 160.

2 L.c.

^ T. Tammes, Rec. Trav. Bot. Neerl. Vol. viii. 3, 1911, pp. 201—288.

^ E. S. Finlow and I. H. Burkill, Mem. Depart. Agric. India. Bot. Vol. iv. 4, pp. 73—92.

5 w. L. Balls, Journ. Agric. Sci. Vol. ii. 1908, pp. 346—379.

« H. de Vries, Ber. Deut. Bot. Ges. Vol. xviii. 1900, pp. 83—90.

7 H. M. Leake, Journal of Genetics, Vol. i. 1911, pp. 205—272.

8 G. H. Shull, Amer. Nat. Vol. xvir. 1908, pp. 433—451.

^ E. R. Saunders, Proc. Roy. Soc. Vol. lxxxv. B, 1912, pp. 540—545.

B. MiYAZAWA

77

that the hoarinoss due to H and K is only manifested when C and R
are both present Hence an albino (as regardH anthocyanin) nmy contain
both H and K, and nmy y^'t Ih» glahroiiH Ix^oaust* it ainnnt contain at
the saine time both C and R. An anthocyanin form, on the other hand,
which is glabrous, carries of course C and R, but C4in only conUiin either
H or K, and not Ixith ; when it carries C and R, <ih well an H an<l K, it
is hoary and colon reil.

The relationship existing between the factor for leaf-colour and that
for flower-colour in our Cotivnlvulus is very similar to the last nH*ntion(*<l
case in Stocks. The fact that the dark-red colour apj)ears excluHively
in flowers of green plants will hi} explained in like manner as in the casi*
of Stocks. If we denote for example the fivctor for green leaf-colour by
G and that for dark-red flower-colour by D, then the jwirents would Iw

A =ggdd,

^ = GGDD.

The Fi hybrid is thus GgDd, so that it is heterozygous for the
factor G. We will suppose that D can produce red colour but not dark-
red, when G is either heterozygous or absent in the zygote.

We should have in F., the following plants :

Colour

No. of

Plants

of leaf

of flower

planU

GGDD

green

dark-red

1

GGDd

>>

»>

2

GGdd

»»

white

1

QgDD

red

2

GgDd

,,

,,

4

Ggdd

,,

white

2

ggDD

yellow

red

1

ggDd

»>

,,

2

ggdd

>»

white

1

Totals

16

3tes may

be arranged

as follows

:

( dark-red flower 3

Green leaf -1 red

6

i white

„ 3

/ dark -red flower 0

Yellow leaf \ red

3

i white

„ 1

That the theoretical expectation just mentioned is well fulfilled, may
be seen from Table II (p. 63), Table VI (p. 66), and Table V (p. 65).

78 Studies of Inheritance in the Japanese Convolvulus

The offspring derived from these F^ plants were studied in order to
ascertain, whether the production of the Fo. plants with the above
mentioned genotypic constitutions has been realised.

In the families containing plants which always produce white
flowers, Table XI shews that a corresponds to the formula GGdd,
y8 to ggdd and 7 to Ggdd.

We could get no family corresponding to the formula GGDD in F^,
though we had some (cf. Table XIV) corresponding to the formula
GGDd. It will be noticed here that notwithstanding the fact that
there should be theoretically one GGDD and two GGDd in F^ we had
seven GGDd and none of GGDD, but this may perhaps be merely a
matter of chance and without special meaning.

The results in respect to the plants of other genotypic constitutions
are as follows :

Table XVI corresponds to GgDd.
,, XV „ „ GgDD.

,, XII „ „ ggDD.

„ XIII „ „ ggDd.

Thus all results secured in ^3 progenies are fairly well in accordance
with the theroetical expectation, except GGDD.

Furthermore, let us examine the results in F^ to see whether or not
our expectation is fulfilled. First of all, we have the families corre-
sponding to GGDD in Table XX (c?), and other families are similar
to those in F^, It will be noticed also here that we have had no single
constant family containing green plants with red flowers till we have
attained the F^ generation, and moreover, according to our theoretical
expectation it should appear neither in F^^ nor F^. This fact alone
suffices perhaps to confirm our hypothesis above mentioned that in the
presence of the factor D,G will produce dark-red colour in its homozygous
and red colour in its heterozygous condition.

Next I will pass on to the results of back -crossing. According to
our theory the ratio of plants with dark-red and red flowers in F^y. B
should be 1:1, and this was really the case, as will be seen in
Table XXII. In i^i x J. there should be no plant with dark -red flowerg,
and this is really the fact, as will be seen in Table XXI. Thus
again the results of back-crosses are in perfect accordance with our
expectation.

Further, I have made various crosses between some of the F^ indi-
viduals to each other, and also between them and either one of the

B. MlVAZAWA

79

two original |)arentR. The resulta of thone ex|K»riinent8 are shewn in

the Table XXIII.

TABLE XXIII.

KlowcTHSolottr

I atUtnpUd
(10— l)x(ie— 9)

(17— 7) X (17-18)

(W- 8) X (M- 4)

(«— 3) X (22-11)

(81— l)x(31— 5)

(31—10) X (31— 9)

(44— 2) X (44—10)

(55— 15) X (38— 1)

(56— l)x(38— 1)

(59—10) x (16— 9)

^ X ( 9— 4)

A X (16— 9)

Jx(38— 1)

(32— 5) X n

J grwn
) yellow

( Rreen
J yellow

i green
) yellow

j green
) yellow

( green
f yellow

green
yellow

( green
i yellow

j green
} yellow

( green
\ yellow
( green
I yellow
( green
( yellow

{ green
) yellow

j green
/ yellow
j preen
} yellow

Red

:»o

0

0

04

18

27

7

0

29

22

24

0

26

44

4

0

0

27

0

39

10

0

0

9

0

8

0
0

liarkrwl WhiU

0

0

u

0

0

0

0

0

0

3

0

3

0

2

0

0

0

0

0

0

25

0

0

0

0

0

0

0

0

7

0

0

0

0

0

21

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

6

11

0

0

0

TuUJs
60i

o(

0^
64 (

21
30

9

0

29

22
49^

Oi
26 i

44 (

"I

0
39
10

0

:i

0

14

11

0

(innd toUla

50

64

61

51

49

70

11

48

39

10

14

11

From what was above described about the results of self-fertilisation
of /^, plants it is clear that the genotypic constitutions of F3 plants and
the relative number of various kinds of the offspring as the results of
their crosses are as follows :

Let us see whether the results of these experiments and <Mir expec-
tation are in agreement with each other.

(10 — 1) X (16 — 9): All the hybrids have the genotypic constitution
GgDd and should be green-red.

(17 — 7) X (17 — 1): All the hybrids have the genotvpic constitution
ggDd and should be yellow-red.

80 Studies of Inheritaiwe in the Japanese Convolvulus

TABLE XXIV.

CroBses <

Uenotypic constitutions

Genotypic constitutions
of the offspring

attempted

of the parents

Ratio

(10-

- 1) X (16- 9)

GGdd xggDD

GgDd

(17-

- 7) X (17-13)

ggdd XggDD

ggDd

(22-

- 3) X (22— 4) .

ggDd xGgDd

GgDD, GgDd, Ggdd, ggDD,)
ggDd, ggdd j

1:2:1:1:2

(22-

- 3) X (22—11)

ggDd xGGDd

GgDD, GgDd, Ggdd

1:2:1

(31-

- l)x(31— 5)

GgDD XggDD

GgDD, ggDD

(31-

-10) X (31— 9)

GgDD xGGDD

GGDD, GgDD

(44-

- 2) X (44—10)

GgDD XggDD

GgDD, ggDD

(55-

-15) X (38— 1)

GGdd xggDd

GgDd, Ggdd

(56-

- 1) X (38— 1)

ggdd XggDd

ggDd, ggdd

(59-

-10)x(16— 9)

ggdd XggDD

ggDd

^ X ( 9— 4)

ggdd xGGDD

GgDd

A X (16— 9)

ggdd XggDD

ggDd

^x(38- 1)

ggdd XggDd

ggDd, ggdd

1:1

(32-

- I)x5

GGdd xGGDD

GGDd

(22 — 3) X (22 — 4) : The genotypic constitution of these hybrids is
very various, as will be seen in the above table. If we classify them
according to leaf- and flower-colour there should be 3 green-red, 1 green-
white, 3 yellow-red, and 1 yellow-white. Now according to this expec-
tation the deviation and the standard error are ± 2-250 and ± 1-985,
respectively, in green plants, and they are ± 4-500 and f 2-371, respec-
tively, in yellow ones. Thus the results agree almost entirely with our
expectation.

(22 — 3) X (22 — 11): The genotypic constitution is here also various,
as seen in the above Table. The ratio of red and white should be 3 : 1 ,
and the experimental result 7 : 2 shews that our expectation is almost
perfectly fulfilled.

(31 — 1) X (31 — 5) : As the hybrids have the constitution GgDD and
ggDD, the ratio of green-red and yellow-red should be 1:1, and the
plants really obtained are 27 and 22 : thus here also our expectation has
been fairly well fulfilled.

(31— 10) X (31— 9) : As the hybrids have the constitution GGDD
and GgDD, all jDlants should be green and the ratio of red and dark-red
1 : 1, and we have had 25 red and 24 dark-red, the results being thus
fairly well in accordance with our expectation.

(44—2) X (44—10) : The hybrids have the constitution GgDD and
ggDD, so that green-red and yellow-red should be in the ratio 1 : 1. As,
however, the plants of these two kinds are 26 and 44, respectively, their
ratio seems to fit somewhat badly with the expectation, but the deviation

H. MiYAZAWA 81

and thr sUiidarH frmr Ikmii^ i !MMH) and ± 4ls;}, ri!»i>ectivt'ly, the
rv^ults may bo said U) Ih' in acconlancc with the t'XiK'ctation.

(55— 15) X (38 — 1): Ah thi« hybridM havo thc^ conMlitutioii GgDd
and Qgdd, then' should he I gn*cii-nMl ami 1 ^nM-ii-whitc, and wr havt*
4grt*en-re<i and 7 groon-whito in n^ality. Thtmgh the empirical innnbtTs
are very small, the deviation and the standanl ermr are ± 1 ^(H) and
± 1-658, re«jK»ctively, an<l our (>x|MH-tation is fulHlh'd.

(56—1) X (38—1): As tho hybrids have thr constitution ggDd and
ggdd all i)lanUs should hv yellow, and the ratio of red and whit<' 1:1,
in fiict we have 27 yellow-ri'd and 21 ydlow-white.

(59_10) X (16—9): The hybrids have always the constitution ggDd,
and we have 89 yellow-red.

A X (16 — 4) : The hybrids have the constitution GgDd, and we have
10 green-red.

A X (16 — 9): The hybrids have the constitution ggDd, and wv have
9 yellow-red.

A X (88 — 1): As the hybrids have the constitution ggDd and ggdd,
all plants should be yellow and the ratio of red and white 1:1, indeed
we have 6 yellow-red and 8 yellow-white.

(32 — 1) X B: All hybrids have the constitution GGDd, and we
have obtained 11 green-dark-red.

When we examine all the above results we find some cases which
seem to fit badly with our theoretical expectation, though all these lie
within the range allowed by the theory, and such cases are no doubt due
to the small number of individuals included in each family. It may be
noticed moreover that, on the one hand, all kinds of plants which are
theoretically expected to occur in any family were found there to appear;
and on the other, in no single family plants of such kinds which should
not occur there accoi-ding to our theoretical expectation were ever found
to appear.

Summary.

1. The green colour of leaves is dominant to yellow, and the segre-
gation in F2 takes place accoixling to the 3 : 1 ratio.

2. The factor producing the " hukurin " is present in the parent
with white flowers. This condition is dominant to full colour and in F.,
the segregation occui-s according to the 3 : 1 ratio.

3. The results mentioned in 1 and 2 agree with those obtained by
Takezaki.

82 Studies oj Inheritance in the Japanese Convolvulus

4. If we denote the one parent by GGDD and the other by ggdd,
there exists the interrelation between the factors G and D, inasmuch as
in the presence of D the production of the dark-red flower-colour takes
place when G is present in homozygous condition, and that of the red
(magenta or scarlet ) colour, when G is present in heterozygous condition
or altogether absent. The hybrids F^ (= GgDd) will thus bear always
flowers of red (= magenta) colour.

EXPLANATION OF PLATE IL

Fig. 1. A dark-red flower from the one parent.

Fig. 2. A white flower from the other.

Fig. 3. A magenta flower with " hukurin " from the Fi plant.

Fig. 4. A scarlet flower from a F^ plant.

Fig. 5. Leaf and a portion of stem from a green plant.

Fig. 6. Leaf and a portion of stem from a yellow plant.

All figures are from water-colour drawings by Mr N. Midusima.

JOURNAL OF GENETICS. VOL. VIII. NO. 1

PLATE II

I

V

Miduaiiua del

Volume VIII APRIL, VJVJ No. 2

ON FORMS OF THE HOP (IIUMULUS LUPULUS L.)
RKSISTANT TO MU.DKW {SPIIAEROTIIECA
HUMUrj (DC.) BURR.); H'.

By E. S. salmon,
Mycologist, S.E. Agricultural College, Wye, Kent, EngUiud.

In a previous article* attention wjus called to the fact that certain
forms of the Hop (Huutulus Lupulus L.) are resist<ant t-o the attjicks <>{'
the }\oi^-i\\\V\q\s {Spluierotheca Huinuli (DC.) Burr.). These '* innnune "
plants fall into two groups, ((/) certjiin individual seedlings of the wild
hop raised from .seed obUiined from Vittorio, Italy ; (6) the female
variety with yellow leaves known as the "golden hop." The present
article describes further exj)eriments carried out during 1917 with these
and other plants.

Group (a). Of this gi*oup 2 seedlings were discovered in 1914, and
7 seedlings in 1916. As already mentioned, the two 1914 seedlings were
planted out during the winter 1914 — 15 in the Experimental Hop-
garden at Wye College; the next season one plant (Ref. No. OR^H)
proved to be female, the other (Ref. No. 0RS9) male. These two
plants were sufficiently established by the winter of 1916 to enable
"cuts" to be taken from them; 5 were potted up from OR^H and 2
from Oi^ 39. These potted plants were the ones used in the following
experiments (Expers. 1 to 5) :

Exper. 1. A ix)tted "cut" of 0RS9 and a similar potted "cut" of
another seedling (Ref. No. Z 54) of the wild hop from Italy were treated
as follows: a fully-expanded leaf at the 3ixl node from the base of the shoot
(which had 7pjiirsof leaves) was sprayed with water,using an "atomiser,"
until numerous small drops had collected on the leafs upper surface ;
conidia were then placed on these drops at three similar places on each leaf
The conidia were taken from various " powdery " patches of the mildew
occurring on different susceptible hop-plants stiinding in the greenhouse

^ See Journ. Agric. Sci. Vol. viii. p. 455 (1917).
Joum. of Gen. viii 6

84 ReHiHtance to Mildew in Hops

where the experiment was carried out. In tho case of this and of all
the experiments described below in order to make the inoculation
material hh unifoim .is possible (or nil the plants in each experiment,
the conidia lakcn lioin ('.irli " iiovvdcry" patch wore distributed equally
as far as posHil)l(! on all Mic |)lniil;8 in the process of inoculation. The
drops of water w(jro found to have cvapoinlcd in a f(^w hours' time.

By the Slst day after inoculaUon the leal" of Z^A was infected at
the three inoculated places, wlicro clusters of conidiophores occurred.
,Nine days later these p.itdK s h.id become "powdery" with conidia, and
patches of mildew wciv .ilso pivsent on four other leaves and at four
places on the stem. No Uww of any infootion resulted on 0-R39. The
two plants stood side by side, and thus exposed to the same chances of
inoculjition from Hurroundim^ inildcucoxcicd pl.mls. Tii this cxpcrimont
the inuiHually prolonged " inciiU.ithui jM'iiod of the mildew \v;is doubt-
less due to the very abnormal weather conditions of the period (March,
11)1 7) when rapid elianj^^'s of teniperatiii-e oeeiiiiiMl. Under thiise
conditions the shoots of both plants made scarcely any growth, and the
l(>aves showed a slight injury round th<"ir maiL;iiis which turned brown.
riuler Mi( se a(lveis(^ couditions of gi-owth the immunity of OJ? 39
remained uiiclianged.

Exper, 2. One leaf of a potted "cut" of 0/^ 38, and one leaf of a
similar potted " cut " of another seedling (Uef No. Z39) of the wild hop
from Italy were inoculated as in Eayper. 1. By the 31 at day the leaf of
Z89 bore small patches of mildew with weak clusters of conidiophores
at the three inoculated places ; tho leaf of OR 38 bore similar patches
of mildew at two of the three inoculated places. Nine days later the
inoculated loaf of Z 39 still bore weak sub-powdery patches at the three
places, but the leaf was now beginning to die ; on the leaf of OR 38 the
weak clusters of conidiophores — scarcely more than " sub-infection " —
were still visible at the two places. At the end of the Experiment —
58 days from the inoeidation — the inoculated leaf of Z39 had withered
up, and nil tlu leax. s ot th(^ original shoot (which had scarcely elongated
under the ahnormal w ( i i h . i conditions noted in Ewper. 1) were brown
at I heir ed^cs; a new basal shoot bore loaves with numerous patches of
niildtw (.11 them. With regard to Oli*AH, the original shoot, which had
rcMnaiiied chi>cki>(l in growth for some time, had now elongated and was
two feet long; the!(> was no trace of any mildew on the inoculated leaf,
nor elsewhere^ on l\\r numerous h>avi>s, although all the leaves were now
exposed to IVeijnent inoeulations from adjacent mildew-COvered plants.

K. 8. Salmon h.")

It is cloar thiit tho HUNcoptibility hIiowu Uvrv h\ (>/i',\H wiw Htrictly
Im'al or U'in|x»nirv. Viirious hy|xtthoHrH may l>r julvmiccil to jwcmmt
for it: (I) that the i-niHlitions of growth t»Mii|H»rarily IikIiummI moiiu-
iimoiint of Hiiwr|»til>ility ; (2)that inm'uIatiouM when* a ^nn-at niiiiilMT of
conidia art' uhimI may cause at the |)la<'<> of iiionilatioii a ntrictly l«K>al
infcctit)!!. — JVM in the oaHos reronli»(l ot "Muh-iufectiori'"; (M) that the
plant OA'.'iH is ^rjulually Ihmh|; clian^^Ml in itn "conMtitution" an the
n*8ult «»f cultivation in manun'd ^^romul ; (4) that the conidia used in
this Kx|HTiment ha<l exceptional |H»werM of infection (which weakened
iw the mildew j;rew on the ))lant, so that the conidia pnxluced there
wi'rt» iinahh* to infect the plant further). The Huhject in diwuHwd
further at |Mi^e 8(5.

AVyxT. 3'. Two .shoots (of e(jual hn^th and vi^'our) of two |M>tt<d
"cut.s" o\' ()R'\H and Z',V.) were chosen, and nn each sh«>ot 1 leaf (at the
3rd n<Mle from tht^ ajM'x) wjis irxK-ulatcd at three places^ The leaf in
each vixso was just exj)anded. Hy the IDth day the iiKK-ulated leaf of
Z89 bore small, di'iisely |K>wdery patches at all the inoculated places,
and 4 other leaves on the shoot bore each a powdery patch. No infec-
tion resulted on any leaf of ()Rl]H.

Exper. 4. Two shoots (of eipial length and vigour) of a |)otted
"cut" of 0/^38 and of a seedling plant of the wild hop from Italy -of
unknown susceptibility — were chosen, and on each shoot two leav<'s (at
the 3rd node from the apex) were inoindat^ed at three places. On the
12th day the seedling plant wjis fully infcjcted at all the six places;
there were also, by this date, numerous powdei^ patches on seven othei-
leaves. No infection whatever had occurreci on OH'.IH. The same
results were recorded on the 21st day.

hJjcj)€r. 5. The plants used were potted "(Mits" of 071*38, the var.
neo-mexicanus (Kef No. A A 9) and a ccM-tain seedling (Kef. No. ()C 32) ;
one leaf (at the 3nl node from the ap«!x) of a vigorous shoot, <nie foot
long, of ejich plant was inoculated at three jilaces. Kavh leaf was
copiously in<K:ulate(l with conidia taken from the sam<; sources. On the
7th day the plants A A !) and 0^*32 were infect<(l at the three inocu-
lated places on th<'ir leaves, where inimerous, densely powdtiy patches
(xxjurred ; other leaves on both the; plants also bore |)atclies of mildew.
No trace of infection lijul resulted on 0/j* 38, either on the in<»culate(l

' Sec The Neic Pln/tohyiMt, Vol. in. p. 110 (I'.JOI).

'•* In thiH Experiment and in all the followin;^ oneH, the conditioiiK of growth were
normal.

^ In each of thcHe ExperimontH u difTeicnt potted *' cut" wuh UHtil.

i\ 2

86 Resistance to Mildew m Ho2)s

leaf or elsewhere. On the 16th dayS however, a leaf (at the 5th node
from the apex) of the shoot of OR 38 bore on. its upper surface one tiny,
powdery patch, of mildew ; the inoculated leaf as well as all the other
leaves (at the 11 nodes of the stem) of the plant were at this date and
subsequently quite free from mildew. There were no signs in this
plant of any weakness or abnormal growth, and a close examination of
the one leaf bearing the patch of mildew failed to show any difference
in its colour or any injury to the epidermal cells which might account
for this strictly local susceptibility. The patch of mildew measured
2 mm. X 1mm., and was powdery with conidia. These conidia were
removed with a sterilised scalpel, and placed in a film of water on a leaf
(at the 2nd node from the apex) of the same shoot of Oi^38. The
other leaf at the same node was similarly inoculated with conidia taken
from patches of mildew on another normally-susceptible hop-plant. In
neither case did any infection result 2. The little patch on the leaf of
OR 38 noted above did not increase in size, although it produced a fresh
crop of conidiophores, so that 14 days later it was again powdery. The
patch then soon began to die away, and 6 days later was dead. After
the disappearance of the mycelium, a minute patch of brown, dead
epidermal cells became visible, such as is often found at. the centre of a
flourishing patch of mildew. This plant of OR 38 grew vigorously
throughout the remainder of the season, but although continuously
exposed to inoculation by conidia from upwiards of a hundred mildew-
covered plants surrounding it, no trace of any further infection resulted.

Here, again, as ill Eayper. 2 (noted above) there was certainly no
general breaking down of the immunity of the plant, but only a strictly
local susceptibility. Of the hypotheses advanced above (p. 85) to
account for this phenomenon, (2) would appear to be ruled out.

If we summarise the results of the above experiments, we find that
the 1 inoculated leaf of OR 39 resisted infection ; of the 5 inoculated
leaves of OR 38, 4 resisted infection and 1 became feebly infected, while
a very small, strictly localised patch of mildew appeared from some
unknown source on one uninoculated leaf Of the 5 leaves inoculated
of three other seedlings of the wild hop from Italy all became fully
infected, and in the case of each of these seedlings most of the leaves

1 The date was May 22, 1917.

2 A young leaf on this shoot of OR 38 was pricked with a pin at two places, causing
respectively one and three holes, and then inoculated over the wounds with conidia taken
from patches of mildew on various hop-plants. No infection resulted. (In other cases
susceptibility to the attack of the "wrong " biologic form of a mildew has been induced by
this method: see Annals of Botany, Vol. xix. p. 125 (1905).)

E. 8. Salmon 87

not inoculiiUKl Ix'camc Riibstvjufntly infecUnl. The haiiic wjih iiltw) the
case with one plant of the var. fieo-tfiexinniuM, and with oiw " hybrid "
seedling? (Rof. No. ()Cl\2). It is clear that (hiring' the exiKTiinrntH there
wfts no ^Mieral breaking down of the " innnunity " of the two 8<»e<llin^
OHllH and ()R:V,K but in the wi.se of O/i 8S. ..nly a .strictly h^cal or
UMnporary sustvptibility due to unknown causes. This is confirmed by
the general bi»haviour HubsiMjuently <»f these phint^ in the ^reenhruise
chirin^ the season. Besides th<' one |)Iant of 0/^3!) and th<* four phints
of OH 3H use<i in the above-mentioned Kxjx'riments, an additional
potted "cut" of each seedling was present in the greenhouse. These
seven phmt.s ma<le during the season a strong, "healthy" grnwth, e^ich
producing seveml stems, 4 to 6 feet high, with large leaves of a dark-
green colour. Although the conditions were ideal for the growth and
dissemination of the mildew — jis was evidenced by the ftict that it was
only ni'cessiiry to stand a healthy (susceptible) seedling hop-plant in the
greenhouse am<»ng the mildewed plants to find it in a few weeks' time
more or less smothered with mildew — these seven plants placed under
the siime conditions for several months showed no trace of infection
beyond that temporarily induced in Eocpers. 2 and 5.

The behaviour of the original plants OR 88 and OR 89 may now be
noted. These two seedlings proved immune in the greenhouse during
the season 1914 ; during the winter 1914 — 15 they were planted out in
the Experimental Hop-garden at Wye College. In 1916 they flowered,
one {OR 38) proving to be female, and the other, male. During the
summer of 1916 both plants kept free from mildew \ although it occun-ed
on all adjacent plants ; by October 3rd, however, mildew was observed
on both plants. On this date OR SS showed patches of mildew on
several of the leaves and on one hop of a late shoot; while OR 39 showed
mildew on one leaf each of two late lateral shoots. In 1917 both plants
remained free from mildew* until the autumn. In October, however,
ORSH showed a fair amount of mildew (with perithecia) on its hops,
chiefly on the peduncles but also on the bracts and bracteoles ; OR 39
showed one small patch of mildew on the under-surface of one leaf of a
lately-developed lateral shoot.

It is clear, therefore, that both OR 38 and OR 39 when grown in a
manured hop-garden produce late in the growing sea.son leaves which
are more or less susceptible to mildew, and that OR 38 under these
conditions produces "hops" (strobiles) which are decidedly susceptible

^ No direct inoculations were niade on these plants in the hop-garden.

88 Resistance to Mildew in Hops

to mildew. Whether the resistance to mildew as shown in 1914 by the
original seedling plants of OJ?38 and Oi2 39, and in 1917 by the "cuts"
taken from them, will be found to disappear after the plants have been
grown for some time in manured ground remains to be seen. It is
intended to carry out inoculation experiments with "cuts" taken in
successive years.

We will consider now the behaviour in 1917 of the seven seedlings
of the wild hop from Italy which showed immunity in 1916.

Exper. 6. One of the above seedlings (plant a) and another seedling
of unknown susceptibility (plant b) of the same parentage and age were
inoculated. Each plant was inoculated at three places on 2 leaves (at
the 2nd and 3rd node . from the apex), the shoot of each plant being
of the same length and apparent vigour. By the 18th day plant h
was infected on one leaf at the three places, where there were large
patches of densely clustered conidiophores. By the 27th day the other
leaf of plant h showed patches of mildew at two of the three places of
inoculation. The plant a showed no trace of infection.

For the remainder of the growing season this immune seedling a
together with the six other immune seedlings of 1916 stood in the
greenhouse among some hundreds of virulently infected hop-plants,
under conditions which ensured a continual inoculation of their leaves
and stems with conidia. None of these plants showed a trace of mildew.
For two consecutive years, then, these seven seedlings of the wild hop
from Italy have proved persistently immune under conditions in which
other seedlings of the same parentage and age have proyed very sus-
ceptible. Three of these seven immune seedlings have now been
planted out in the hop-garden.

Group (b). Complete immunity to mildew was shown in 1916 by a
form of H. Lupulus with yellow leaves obtained under the name of
"golden hop" from Messrs Bide and Sons, Nurserymen, Farnham. This
plant was further tested in 1917 in the following experiments.

Exper. 7. Two leaves (at the 4th node from the apex) on a plant
of the "golden hop" obtained in 1916 from Messrs Bide, and on a
two-year-old seedling hop of unknown susceptibility, were inoculated at
three places on each leaf By the 10th day the latter plant bore large
densely powdery patches of mildew at the six places of inoculation. No
trace of infection was visible at this date, or subsequently, on the
" golden hop." This plant, together with three other potted " cuts " of
the " golden hop " from the same source, stood in the greenhouse con-

E. 8. Salmon 89

tinuously exposed to infection throughout the seiison ; all the pliintH
remained jx^rsistently immune, just jus in lOKi.

Trial was now made ot* a yellow-leavi'd fi«malc variety purchaned
from Messrs Hunyaixl, Maidstcme, in 1912. The plants used were "cuts"
taken from established plants in the hop-garden and put into |)ots in
the winter of 1916— 17.

Exper. H. The two plants used were both potted "cuts," one (Ref.
No. ^89) was a seedling of the wild hop from Italy, and the other
(Ref. No. 341) was the yellow-leaved variety obtained from Messi-s
Bunyard. On both plants 1 leaf (at the 2nd node from the apex) on
shoots of equal length was inoculated at two places with conidia from
the same source. Owing to the abnormal weather conditions (see above,
p. 84) it was not until the 24th day that any infection was visible on
ZS^\ and then only a few weak conidiophores occurred at one of the
places of inoculation ; by the 33rd day weak, clustered conidiophores
were visible at the two places on the leaf (which was now brown at the
edges), and small, vigorous powdery patches occurred on three other
leaves and at one place on the stem. No trace of infection resulted on
341, although this plant stood by the side of Z39 throughout the
growing season. It was clear that the abnormal weather conditions
had no effect upon the immunity of this yellow-leaved plant.

Exper. 9. Two "cuts" in pots of -^39 and 341, with shoots of six
nodes and of equal length, were used in this experiment. The tw^o
shoots were first " atomised " with water and then inoculated by shaking
over them a virulently infected hop-plant, from which the conidia fell
in clouds, — with the result that most of the leaves on both shoots
became visibly whitened with the mass of conidia. By the 11th day
6 leaves of Z 39 w^ere heavily infected, many of the patches of mildew
already bearing clustered conidiophores. By the 15th day 7 leaves
and parts of the stem were smothered over with powdery patches. No
trace of infection occurred on 341, which stood by the side of ^39
throughout the season.

Exper. 10. Shoots of ZS9 and of 341 of equal length and vigour
were selected, and 2 leaves on each shoot — one partly unfolded leaf at
the 2nd node fnjm the apex and one just expanded leaf at the 3rd node
— were inoculated at three places on each leaf. By the 13th day the
2 leaves of ZS9 were infected at all six places; by the 20th day the
2 inoculated leaves bore densely powdery patches, and 3 other leaves
were also mildewed. By the 30th day 14 leaves of Z39 were mildewed.

90 Resistance to Mildew in Hops

No trace of infection occurred on 341, which stood by the side of Z39
throughout the season.

Six plants of this " golden hop " (341) from Messrs Bunyard stood
exposed to infection in the greenhouse throughout the growing season ;
all remained persistently immune".

The " golden " leaved plants (mentioned above) obtained from Messrs
Bide and Bunyard are both female, and agree in the characters of the
shape of the leaf and its coloration. A comparison of mature plants
to establish the identity of the two has not yet been possible.

In 1917, as in all previous seasons, the 9 plants ("hills") of the
" Golden Hop " planted in the hop-garden showed no trace of mildew
on leaves or hops throughout the growing season.

A male variety possessing "golden" leaves also exists. In 1910
some plants were sent to me by Messrs Bunyard under the impression
that they were the female plant. These were used in the following
experiments.

Exjper. 11. A young leaf (half expanded) on a shoot. 2\ feet long of
a " cut " in a pot was inoculated with conidia from the same source, the
three following plants being used : (1) a seedling (Ref. No. 0032) of the
cultivated variety " Bramling " ; (2) the Russian variety " Shpaltski " ;
(3) the (/ variety with yellow leaves. By the 7th day the leaf on all
the plants was equally and virulently infected; the leaf (now fully
expanded) being covered over almost continuously with densely clustered
conidiophores.

Exper. 12. Young leaves (partly expanded) on shoots of equal
length of the Russian variety " Zemshevi " and the (/" variety with yellow
leaves were inoculated. By the 7th day the yellow-leaved plant showed
infection at the three places of inoculation ; the '' Zemshevi " variety
was also similarly infected.

Three other " cuts " in pots of the ^ variety with yellow leaves were
placed in the greenhouse and exposed to inoculation by placing viru-
lently infected plants around them. They all became infected. It is
clear that this ^f plant with yellow leaves is susceptible to a normal
degree to /S. Hiiirndi.

1 An attempt was made to induce susceptibility by injury to the leaf. A young leaf of
341, attached to the stem, was pricked with a pin, 30 holes being made in the half of the
lamina on one side of the mid-rib. The injury inflicted did not kill the leaf cells except
those immediately surrounding tlie hole. The whole leaf was inoculated, but no infection
resulted.

E. 8. Salmon 91

The ftiots tis regards the origin and correct name of these yellow-
leavetl or "golden" fonuH are hard to discover, owing jNirtly to the
impossibility of corre«|K»ndence with (Jerniany. A certain amount of
information is being collecte<l, which it is hoped to publish later.

Summary.

1. Individual seedlings of the wild hop {JIitmulu,s Lupuluji L.) when
grown in a gn'enht>use may he innnune as regards leaf and stem to the
attacks of the unldew Sphacnttheca Iluniuli (I)( \) Burr. This immunity
has been shown by the s;ime seedlings throughout the growing season
for two consecutive yeai-s.

2. Such immune seedlings when planterl out in the hop-garden
may show susceptibility late in the growing se;ison iis regards the leaf
and " hop " (strobile).

3. An immune plant in the greenhouf^e may show strictly Icxial
susceptibility without the general immunity being lost. (Expers 2 and 5.)

4. A yellow-leaved female variety of H. Lnpulas is immune to
S. Humuli.

5. A yellow-leaved male variety of H. Lupidus is susceptible to
S. Hamuli.

STUDIES IN VARlPXiATlON. I.

U\ W. HATKSON. M.A., KH.S.

(With Platos III and IV aixl One Tcxt-fi^Miic.)

Thk phononuMia of variegation dnc to absence or deficiency ot
chlorophyll have for some time been a special object of study at the
John Innes Horticultural Institution. The inter<'st of the subject lies
in the circumstjinee that in variegated plants an opjx)rtunity is given (»f
witnessing somatic distribution «»f a character, deficiency of chlorophyll.
alrciidy known to be in many plants a Mendelian recessive. It is true
that up to the present time no direct experimental evidenc** exists sufti-
cient to prove that the chanicters, presence and absence of chlorophyll,
heterozygously combined together in fertilisation, am actually lead to
the production of a variegated zygote; but from the general coui-se of the
phenomena of mosiiicism, presenting not very rarely. two allelomorphic
differences in juxtiiposition in the same plant, we may assume without
much reservation that this interpretation is admissible. Baur', indeed,
speaking of a blue Veronica bearing a white-Howered branch, observed
by de Vries, is disjx>sed to refer such ca.ses to original mutation by
loss, r;\ther than to somatic segregation of charactei-s in heterozygous
combination. Evidently there is at present no means of positively
distinguishing the two possibilities, but I incline to regard .somatic or
vegetative segregation as on the whole the more acceptable account.
If this hypothesis be the true one, we have in variegation a visible
model or plan of segregation by which the properties of the germ-cells
are certainly determined in many instances, and we may at least enter-
tain the possibility that in plants segregation in properties not thus
producing visible .somatic effects may also be similarly determined.
The series of examples which will be descril)ed in the present and
succeeding papers illusti-at** mi.scellaneous features in this special kind
of segregation. Apart from any question of wider application \\\v
phenomena are, 1 think, of obvious genetic im[>ortance.

' y. /»//■/ /»MH.7, Ac. p. 218, Note.

94 Studies in Variegation. I,

Part I. Reversal in Periclinal Chimaeras.

Variegated plants having a white subepidermal layer extending
over a green core, however fertilised, give exclusively white or albino
offspring, which of course die after a short existence. Conversely those
having a green skin over a white core give green offspring only. The
significance of this observation was first emphasized by Baur. We have
seen numerous examples of such behaviour in the course of our work,
of which a list will eventually be given.

The general appearance of these chimaeras, as Winkler and Baur
have called them, is familiar. It is noticeable that in some of them the
thickness of the " skin," whether white or green, may remain with great
constancy the same over very large areas of leaf-surfaces. In white-
skinned forms which are thus regular (e.g. Holly and Box) the deficiency
of chlorophyll affects chiefly the subepidermal layer. In other plants
(e.g. Nicotiana colossea var. variegata) there is continual irregularity,
some leaves having only the subepidermal layer white, while in others
the underlying layers are similarly affected to varying depths, in most
white-skinned plants the edges of the leaves are solid white throughout
their whole thickness, so that each leaf has the white marginal band
characteristic of '' varietates albo-viarginatae" as they are styled in
horticulture. The width of these white edges is sometimes fairly
constant, but generally varies considerably.

The condition in which the core is white and the skin green is far
less common, and hitherto we have seen none in which the green layer
is uniformly one cell thick. Generally the edges are for a considerable
breadth solid green, the thickness of the green layer diminishing
towards the centre of the leaves where the white core shows through,
being sometimes entirely exposed (as in Coprosma). Irregular bands
of solid green are often prolonged from the margin into the middle of
such leaves. Several of the green-skinned chimaeras have the peculiarity
that their stems are destitute of chlorophyll or nearly so. For example,
in the green-skinned form of Coprosma, of Vinca major, and of one of
the Pelargoniums the stems are almost white. In the green-skinned
ivy-leaved Pelargonium also chlorophyll is almost entirely absent from
the stem, but, owing to a great development of red anthocyanin in the
cortex, the stems are a full pink, whereas in the white-skinned form of
the same plant the anthocyanin is confined to a thin layer of the cortex
in the stem. In connexion with this type of variegation mention
should be made of another, somewhat analogous, in which the absence

W. Batk8(»n 95

of chlorophyll is cnrriiHi a Htn^e fiirthtT. In this, not only in the nUMn
whitv, but the pt^tioloH ami thr ccntrrs of th»' havi's and HtipulfH an*
also throughout thfir thirkness dostitutr of rhlomphyll. Of thJH
condition I know only two |K»rfi'ct examples, a ////(/m/u/ra aini \\u'
PelRrgt>niuni "Freak of Xatun*" niis<M| l>y Mrssi^ (aiinrll. DrtaiU <»!
the»e planU will l>o j^iven in a sul)S('<|uent |Ni|K'r.

The oocum»nce to which I wish now to cjili attention is a Humaiic
change such that a sjwrt arises in which the relative positions of the
green ami whit^' jwrt.s are reversed. This phenomenon of (•(unjilete
reversjvl has now occurri'd in five? distinct plants, Knouymnit japoniruM
Uttifoltus, Coprosma Baueri, and three Pelargoniums, viz. an ivy-leaved
variety, and two of the zonal class, Mmc Solleroi and ('aroline Schmidt.
In none of the examples is any evidence forthcoming jus to the cause of
the change, nor can any suggestion be offered as to the natun; of the
disturbance provoking it.

Euouymus japouicus latifolius van variegata.

On the occasion of a visit to Messrs May's nurseries the reversed
specimen was noticed among a large batch of well-grown plants of thi,s
horticultunil variety. The shoot of the green-skinned form' was a
strong branch arising in a sharply marked area of the stem, well above
the level at which the cutting had been divided from the original plant.
The growing point of the main stem must, at the i)oint from which
the sport arose, have formed simultaneously a white-skinned and a
green-skinned segment, and in this latter area a bud had arisen which
developed into the green-skinned branch. Neither among the many
plants seen at Messrs May's nor among numerous specimens of the
variety since examined in various gardens, including several very large
plants many feet high, has any similar piece been met with. Hut
wholly white and wholly green areas are formed not uncommonly on
the white-skinned variety. If in such an area a bud is included, it
gives rise of course to a branch wholly white or wholly green as the
case may be. In Fig. 3 a leaf having such a wholly green area is
shown. The green-skinned form, on the contrary, though a considerable
quantity of it has now been grown, has not produced any substantial
variation. In it, as in the white-skinned form, the number of cell-
layers forming the "skin " is sometimes greater and sometimes less, but
no white areas or white-skinned parts have appeared. The sU'm in
this case is green.

' Mr Bintner tells rae that this variety is grown in continental nurseries under the
name of Due dWrxjou.

96

Studies in Variegation. I.

The white-skinned form alone has flowered under observation. It
failed to set with its own pollen.

The text-figures show in section through the leaves the distribution
of chlorophyll in the leaves of the two forms.

Green skin. White core.

White skin. Green core.

Euonymus japonicus latifolius var. vai'iegata.

Coprosma Baiieri.

The white-skinned var. variegata of this New Zealand plant is well
known. In 1877 J. Barbier figured in Rev. Hort. Belg. ill. p. 32 the
reversed or green-skinned form which had been lately brought out by
Messrs B. S. Williams. To it he gave the name G. Stocki. It is also
sometimes called var. picturata. A few years ago Sir William Lawrence
presented to this Institution a cutting of this identical variety which
had arisen at Burford as a sport from the ordinary variegata. The two
forms are shown in Figs. 6 and 7. The distribution of the green in the
green-skinned form is approximately the converse of the distribution
of the white in variegata. Its stems however are white. The green-
skinned plant has, with us, produced some wholly green shoots.

W. llATKSOX U7

Pelargoniums,

In PolargDnium Miiu» S«>Ilen»i (Fig. IO)thi' n^vrrwil hits «K*(Mirrfcl on our
planto 8overal timoM. Oiu'o ii whoir hruiu'h of gr«'fn-MkiniM'<l Icjivi-h was
fonixxi, for the inoHt |Nirt i\» in Fig. 12, liut amongst th<MM a leaf a|)|H'an(l
having the whole of c»no side grcon iim shown in Fig. 14. On anoth<T plant
of Mnie Sollen>i h 8h<M>t a|)|H»artMl Ix'uring many Icavrs which wrn- wholly
white, but the leaf suinding lowest on the shoot, viz. the first Icat from
the stem, hml the structun* shown in Fig. 13, half being white and th<*
other half griH'n over white. In individual leaves jMitches of reversal
have been formed jvs in Fig. 11. Such green-skinned |)atches include,
I believe, always some jwirt of the leaf-margin, and on their internal
boundary they are delimited from the whitc-over-green parts by a white
bi\nd indicating that in the area in which the two kinds of arrangement
abut on each other, the deficiency of chlorophyll extends below the sub-
epidermal layer (compjire Fig. 16).

It is a peculiarity of Mme Solleroi that, so far iis I have observed, no
flowers are formed on the white-skinned parts, but the green-skinned
branch produced a truss of pink flowers. These flowers however were
ill-formed' and destitute of pollen. The pistils were, I believe, also
defonned, but by inadvertence no note of their condition was made.

On a large pink-flowered ivy-leaved Pelargonium reversal ha.s also
occurred sporadically. Most often the reversiil is confined to a part of
a leaf, usually the whole of one side (as in Fig. 9), but more than one
whole branch of the reversed kind has independently appeared. Flowers
on the white-skinned parts are fertile, producing (as such plants habit-
ually do) long white carpels in the fertilised fruits, but the green-skinned
form has not yet flowered.

The white-skinned Pelargonium Caroline Schmidt very often pro-
duces wholly green sports. We have here had also several individual
leaves on this variety as shown in Fig. 16, composed of a mosaic of the
typical and reversed kinds, but hitherto no reversed shoot.

The phenomenon of reversal is evidently rare and exceptional. No
example other than those enumerated hjis yet been seen among the
many white-skinned plants grown here or examined elsewhere. We

* Note. The variegated Pelargonium "Freak of Nature'' mentioned above (stem and
centres of foliar organs white ; edges of foliar organs green) bears deformed flowers having
both male and female organs aborted. But sports occur some wholly green, others wholly
white, and the flowers on both these are perfect, ripening seed on self- fertilisation, and
producing seedlings respectively wholly green or wholly white. This plant has had one
small green-skinned branch which has not yet flowered.

98 Studies in Variegation. I.

have, for instance, several hundred yards of Euonymus radicans, var.
variegata used as an edging-plant. Wholly green shoots are common
on this plant, and wholly white shoots not rare, but no reversal has yet
been seen. Among many hedges of white-skinned holly also no reversal
was found.

Cases superficially mistakable for reversals are not uncommon in
various plants. For example, in the white-skinned Pelargonium used
by Baur in his observations (of which he kindly gave me a cutting some
years ago) leaves like that shown in Fig. 15 occasionally appear. At
first sight the condition recalls that of Figs. 11 and 16, but on closer
examination this is seen to be due in reality to the formation of small
solid green areas associated with irregularity in the number of layers
devoid of chlorophyll. In this variety, as in many white-skinned forms
the appearance of wholly green areas is not very rare.

Obviously the occurrence of reversal, and of areas wholly green or
wholly white, are consequences of some instability arising in the growing
point, but there is nothing to indicate the cause of such instabilities.
The formation of wholly green areas in white-skinned plants may no
doubt be described as a bursting out of the green core, and might be
attributed to some greater vigour of the green parts, but these expres-
sions are merely descriptive. Injury may be suggested as a probable
cause. White shoots do indeed arise with special frequency round old
scars on the boles of white-skinned hollies, but green shoots, which might
be expected to burst through are extremely rare, if they occur at all in
such places. The suggestion of injury is plainly inapplicable to such
cases as the Pelargoniums described in this paper.

It would be interesting to ascertain whether the green-skinned forms
ever change back again, and the absence of any example of this trans-
formation may be worth noting.

As mentioned in introducing the subject, the consequence of somatic
reversal is that the genetic properties of the plant are completely
changed. Naturally this fact leads to a surmise similar to that sug-
gested by the behaviour of root-cuttings \ In the variegated chimaeras
we can visually distinguish the properties of the cortex, but is it not
probable that similar genetic distinctions may exist which are not thus
visible ? May not the phenomenon of reversal exist in regard to them
also, bringing into the cortex, and so into the germ-cycle, properties
previously contained only in the deeper layers ?

1 See "Root-Cuttings, Chimaeras, and Sports," Jour. Gen. Vol. vi. 1916, p. 75.

JOURNAL OF GENETICS, VOL VIM. NO. 2

PLATE III

JOURNAL OF GENETICS, VOL VIM. NO. 2

Fig. 8.

Fig. y.

PLATE IV

Fig. 16.

llL'. Hi

W. IUtkson 09

EXPLANATION OF PLATES III AND IV.

Thi» drAWtntri> li^><» which Uiono nUlvnwrn- prrixirctl wrrr lutuh- by Mr ( . H. 0"l« r»l<»ck,
Tho various lonm of fjTt'Oii (lr|HMul on tlie iuhuIm-i (tf Iav«TM whirli air ^•r^l•n or \vhlt«

Fig. 1. KuomifmMt jiip(uncu* Ititifolius, wliitoHkinncd variety, liaviiij^ n unn-u coir.
Fig. 2. IWvorsotl form of tlio kaiuo, having a k>»'<'" ^l^iii over h wliiir con-.

Fig. S. Ii»»af from Rtiothcr hush of the KRmr having' om- Hid«' wh<»lly ^'rom hikI ilir «itli«r

i«i(lt> whito skinned. (Slijjhtly too hluo in torn : Fi^'. I roin-ctly ^,'iv«K tin cMlr.m (•(

Uio dorsal nurfacojt. )
Fign. 4 and *>. Hack and front of a leaf of tlw sjinic liavinp an nrca wliolly i^Mcni ;h>|k ;iii»H!

next thr luidrib on one si(io.
Fig. t». i'i'pntitnui liitucri, var. vuricijtitit ; tlio wliito-skiruM'd fomi.
Fig. 7. Tho j«anu' ; j^n^-n skinned form.

Fip. 8. Ivy leavini iVlarponium sliowint^ tlio mixture of tlio two kiuils of rliimnrra.
Fig. 9. Leaf of the same : one side wliite-skinned the other side for tho incmt piirt

green -skinned.
Fig. 10. Madame Solleroi, zonal Pelarj,'onium: the white-skinned form (Tlu- veins in

this figure are too wide. They are correctly represented in tlie other figures.)
Fig. 11. The same with reversed areas at apex of leaf.
Fig. 12. The same: green-skinnetl leaf.

Fig. 13. The same : leaf all white on one side, Rreen-skinned over most of tlio other sidr.
Fig. 14. The same: leaf half green and half green-skinned.
Fig. 15. Baur's white-skinned zonal : leaf showin^r irregularities in the uninbir of lay«rs

de6cient in chlorophyll.
Fig. 16. Zonal Caroline Schmidt : leaf with two separate areas of reversal.

Joura. of GtMi. VIII

INHRHITANCK OF CKRTAIN ( 'If AKA( TKliS IN
THK COWPKA (VniNA S/Xh\S/S).

h\ S. ('. HAKLAXI). n.S( (LoMu.
ImfHTuiI DefHirtineiit of A(fn('nftftrr Jar t/ir Wfst Imlies.

(With ()n«' 'r«'xt-tii^Min'.).
("ONTKNTS.

11.

Introduction

I. Flowor colour .......

(1) Park X pale ........

(2) Dark x wliite

(3) I»alex\vhit«

The Pattern of the Seed Coat

The Experimental Results ......

(1) Black, small eye x Rounceval, solid colour

1*2) Red solid x Black, small ey<' .....

(3) Red solid x Brown, small eye ....

Summary .........

The Colour of the Pattern of the Seed Coat .

Relation of Flower Colour to Pattern of the Seed (^o.it

Summary-

Reference.-j

Tahles I— XVII

III.

101

nn

U)'l
\u.\
104

lot
10.-.
10.-,
\m
111

112

11.^

117
lis
111>

I\TR<)l)r(TI().N.

The cowjKJJi is ;i very suitable plant as a subject for t^eiittic iinrsti-
gjition in the Tropics. It is oa.sy of culture, «tccupies little space, ami
usually pnnluces enough seed per })lant foi- a progeny row of ;i hundred
or more plants. It comes to maturity in from two to three months, and
as it can be s<^)wn almost at any time of year it is by n.. means difficult
to study three gem-rations of a cro.ss ammally. The investigations
described in this paper deal with the iidieiitance ..f taeiois r(.ncenied in
the production of the following charact* is :
I. The colour <'>f the flower.

II. The pattern of the .seed coat.

III. The C(»lour of the pntt«-rn of the see(| coat.

7-2

102 Inheritance in the Cowpea

I. Flower Colour.

There are three principal types of flower colour in the cowpea.
These may be described briefly as follows :

Dark. Possessing much anthocyanin coloration, producing a flower
the prevailing colour of which is a more or less deep reddish violet.
Colour is most developed in the region of the throat and on the wings.
The keel is usually without colour, but a slight amount of violet streaking
may be present.

Pale. Distinguished from the Dark form chiefly in the lesser de-
velopment of anthocyanin coloration. The standard is almost white
but the wings are faintly streaked with violet. The keel is devoid of
colour.

White. Anthocyanin colour entirely absent, the flower being pure
white except for a faint primrose tinge on the standard in the neighbour-
hood of the throat.

(1) Dark by Pale.

Several reciprocal crosses were made. The flower colour of the first
hybrid generation was in all cases indistinguishable from that of the
Dark parent. Dark is thus completely dominant over Pale.

The F^.

In the F2 three groups of families were grown. The numerical
results are presented in Table I. A survey of them shows that segre-
gation occurs into Dark and Pale. The ratio of Dark to Pale is the
simple Mendelian one of 37) to IP.

TheFs.

In the Fs a large number of families were grown. The results are to
be found in Tables II, III, and IV. From these results it will be seen
that in F3 some of the Darks bred true, while others segregated into
Dark and Pale in the 3 — 1 ratio. The Pales, with certain exceptions
explained below, bred true.

Gametic contamination of the F. results through Natural Grossing.

The ^3 families of cross 1 from F^ parents with Pale flowers all bred
true. With the exception of one family, this was also true of the
families of cross 2. In cross 5, 71 families from F^ Pales were gro\vn.
14 of these families contained Darks, but in small numbers. At

S. (\ IIahland lO:}

thr tiiiii" Nvluii ihr K. «»r rn»sHi'.s I an<i 2 weir Imiii^ slinlitd a airrlnl
sonix'h l"aili'<l l(» r^Vful tin* pivsi'iu!*- nt .uiy insect likely tn (muihc cnms-
ftTtilisiilion. During th*- jM-riiMl whm tlw F. ai mtss 5 was in flower a
lari^' hbu'k rarjH'ntiT Ium- t X i/l(K'(tp(i sp.) mad*' ils apjieanince. an<l often
visiitMl tlu* Mowoix. liringhravy iMxIitMl this insect cansecl the extrusion
of the stigniJis of all the flowers which it visited, and it seems clear that
it is ea|K»ble of cross-pollinat inij th*- jjlaiits. The presence of a few
Darks in faniilii's which should have hied tiiie to l*ale, nnist. he .uscrilied
Ui ius activities.

Genetii' HeUtiionaln p of Darh ((ml rule.

( )n tht' evidt'uce so far jncsented il is |)enMissd)le to conchide that
the two typos of cowpea Hower, Dark and Pale, const it,ute an allelo-
inorphic jKiir.

In order to state definitely that the two tyjies differ by a single
fivctor it must be shown that :

1. A ratio of .S dominants to 1 recessive exists in F.,.

2. The ratio of i)ure to heterozygous dominants is 1:2.

3. The recessives all breed true.

The results c<jnform to these conditions, for the ratio of pure to
heterozygous dominants as shown by their behaviour in F:i cultures is
88:35, i.e. 2*4 : 1*0, a fairly close approximation to expectation. Some
of the families were carried to ^4 and Fr,. The results are of no par-
ticular significance, as they merely serve to confirm those obtained in
previous examinations.

(2) Dark by White.

The F^ of crosses between Dark and White flowered types was Dark.
The results of the F.^ and F,, are presented in Tables V and VI. A
summary of the results is as follows :

In F. the parental types reappeared and the ratio of Dark to White
w;vs 3 Dark to 1 White. In F,, three families only were grown. Two
of these were from F., Darks, and showed segregation into Dark and
White in the 3 — 1 ratio. The remaining family was from a White
flowered F., plant. It bred true to White.

While the evidence does not permit of complete certainty attaching
tt) the conclusion that Dark and White also form an allelomorphic pair,
there can be little doubt that such an interpretation of the results is
justifiable.

104 Inheritance in the Cowpea

(3) Pale hy White.

The White used in this experiment was the same as that used in
the series of crosses just described. Owing to a great increase in the
number of certain insect pests at the time when the cultures of this
cross were being grown, it has only been possible to record the F^ results.
In two crosses the F^ was of the Dark type.

Interpretation of the Experimental Results.

It has been established most clearly that the two colour types Dark
and Pale form an allelomorphic pair. From a less complete series of
results it has been concluded that Dark and White also form an allelo-
morphic pair. The Fy of Pale by the same White proved to be Dark.
These results can be explained by the assumption that flower colour in
the cowpea is due to the interaction of two factors :

L. A factor for Pale.

D. A factor which increases the amount of anthocyanin colour in
the flower but has no visible effect except in presence of L.

According to this hypothesis the number of possible homozygous
types resulting from a combination of these two factors is four, viz. :

LLDD Dark
LLdd Pale
HDD White
lldd White.

Dark (LLDD) and Pale (LLdd) differ in one factor D. Hence they
are allelomorphic to each other. Dark (LLDD) and White (HDD) are
similarly related. Pale (LLdd) by WhitQ (HDD) will give an F^ of
composition LIDd which is Dark and will give in F^ a ratio of
9i) : 3P : 4Tr. One of the Whites will be the double recessive Hdd.

II. The Pattern of the Seed Coat.
Dr W. J. Spillman (1911) has described several types of pattern
characteristic of the seed coat of various races of cowpea, and in his
paper gives a diagram of those that he worked with. A representation
of his diagram is given in Fig. 1.

w^m

f

Fig. 1. Forms of the " eye" or pigment area, in the seeds of
the cowpea. (After Spillman.)

S. r IIaki.am) 105

He ri'ffrs U* Forms a — c iih onliiiary vyo or .smiill ryr. Koriiis e
1111(1 /* art' tn'rtU'd iw onr mulcr the imiiu' ll«»l.st«'in, ihun tin* (roloiir
(iiilUTn of 11 viirioty hiivini; that namr. Korni r/ is cjillrd bir^a- eye, and
is eoiiMidoriHi i\s xhv lictfrozyi^nt*' Ix'twrin H<»lst-4Mii ami Small ^y(^
Form ff n'pnrsontss a k^ciutically ditTt'iciit ty|H' of rye In it thr |»i^-
mcnUHJ aiiNi surnmrnls l\w hilum, Imt thr micrnpylar rnd nf ihr Hrua
has the marj^in vt-ry indistinct ; fine <l<»ts of |Mt^mrnt, extend i»v(i- tin-
micropylar cnti of th*- scrd. Spillnian cmIIs this latt«i- t.y|M' tin- Watson
oyi", from a variety rhanic'tcrisrd hy this ty|M' of |».ill«rn.

Ho investigat.i*d the results of 21 crosses iMtweeii Small eye and
Solirl colour. In all crust's the Ft was Solid coloui-. In the F.. he
oht^iined the following types: Solid, Watson, Holstein, Lirge eye, und
Small eye. The ratio between them approximaU'd to !>:*}: I : 2 : I.

He wivs able t^ fonnulate four hypotheses to exj)lain his ratios. A
summary may hv given of the hypothesis which he consi<lers as best
adaptrcd to explain his results.

The behaviour of the cross Small eye by Solid colour indicates that
these two ty|>es ditTer in two factors which are transmitted inde|x;ndently
of each other. Acconlingly the Watson pattern may be regarded as due
to a fiict^r W, and the Holstein pittern iis due to a factor H. The
formulae* of the tyj)es appearing in F. will be:

1. Solid WWHH

2. Watson WWhh
8. Holstein wwHH
4. Small eye wwhh.

The present writer has investigated in detail the results of three
crosses between Small eye and Solid colour. In inj cjise have such simple
results as those of Spillman been obtained, but the data secured have
provided an indirect confirmation of the latter's work. An account of
the results obtained in the present work will now be given.

The Experimental Results.

Cross 1. Black eye) , jRouncevaP.
Small eyej (Solid colour.

Several crosses were made, and the F^ was in all cases Solid colour.

The F,

The results are j>resented in Table VII. The following lyi)es ap-
peiired : Soli<l, Watson, Holstein types e and /, Large eye spotted, and

' For description .see p. 114.

106 Inheritance in the Cowpea

Small eye. These are the same types as were obtained by Spillman,
with the exception that the type Large eye spotted replaces the Large
eye. Large eye spotted is identical with the Large eye of Spillman
except that spots of pigment are usually found on the body of the seed.
It must be admitted that certain types would have been classified as
Large eye if a very close examination had not been made. Indeed it
happened that several pods had to be examined in some plants before
the pigment spots were seen. The type Large eye did not appear in
any of the crosses studied by the present writer. Large eye spotted
was classed as a Holstein. Actually it showed similar genetic behaviour
to Spillman's Large eye, and may be regarded as the heterozygote be-
tween Small eye and Holstein types e or f.

The present writer does not wish to cast doubt upon the correctness
of Spillman's classification, for it is quite probable that the types with
which he worked were of different genetic constitution.

A survey of the F2 results will render apparent the fact that the
Solids and Holsteins are in excess of the expected numbers while there
is a deficiency in the Watson and Small eye classes. Is the deviation
from expectation of genetic significance ? Assuming that the ratio
9:8:3:1 is the correct one, then all the Solids will contain both
Watson (W) and Holstein (H), and it is possible to determine which of
these factors is responsible for the deviation above alluded to. The
proportions in which W and H occur in the F2 families will be found
in Table VIII, from which it is clear that the deviation from the
9:3:3:1 ratio of Table VII is due to the fact that a greater number
of types with the Holstein factor are produced than are expected. The
ratio of W to w is close to 3 — I but the ratio of H to h is 4-9 to 1*0.
Thus from the F2 results it seems that in this cross the two factor hypo-
thesis of Spillman does not apply.

TheF,.

The Fs results are set forth in Table IX. A summary of them is as
follows :

1. Solid may either (a) breed true, (6) segregate into Solid and Watson
(H : h = 4-8 : 1'O), (c) segregate into Solid and Holstein (W : w = S'l : I'O),
(d) segregate into Solid, Watson, Holstein and Small eye (H : h = 7*2 : 1*0
and W:w = 2-6:l-0).

2. Watson segregates into Watson and Small eye {W : w = 2'6 : 1'O).

3. Holstein segregates into Holstein and Small eye (H : h = 4*7 : 1'O)
or breeds true.

4. Small eye breeds true.

S. V. Haklani)

lo:

It riMiiaiii.s ii(»\v t«i i*oiiMi(l«T thr n>Hiilts in tin- li>;l)t <>! Spiliimurh
hy|Hithosis that in a rrtww U'twtM'n Solid and Small r\f two laetoix
W and H an* roncrrnttl. Kmni tin* F,j rrsnlUs «»f ('msN I it was himmi
that iUv nitii* of H U» h was 4M: 1*0 instrad «>( :{ : I, and that this
(i(*viatiun hsi to a rorn'.s|H»ndin^ dtviatinn in thr nitin iMtNsrrn thrfnur
types. Fnun thr alxivr sunnnarv nf thr f", n-snlts it is rvi<lrni that
thi» nititi of H to h is a^ain very ahnonnal while tin- ratio of W In w is
of thi' nonnal :i — I tyiH'.

Art'onlinij to Spillinan s hypothesis the L^enotyjHs appeaiin^^ in F.
would Ih' as follows :

MiiniiijtyicoUK
WWHH Solid
WWhh WatKoii
wwHH Holstein
wwhh Siimll ovo

HoternzyKdTiK
wwhH Holstein throwing' MI: ISK
wWhh Wiitrtoii tljrowink' HH' : 1.S7-:
wWhH Solid throwiriK 9S : AW : MI : l.S/-;
wWH H Solid throwing '^^ : 1 "'
WWhH Solid throwing •^S : 1//.

Thus <»f' evi'iy }> Solids. I shimld breed true, 4 should be dihybrid,
2 should be inonohybrid {S:H), and the remaining two should be
monohybrid {S: W ). C«»mpare the actual facts. The number of /'',
tamilies from F.. Solids wjvs 44. Of these 6 bred true, 18 were dihybrid,
5 weix^ monohybrid (.S: W ), and 15 were monohybrid (S:H). The
coiTes{X)nding exjX'cUition in each case is 49, 19*6, 9*8, and 98. The
proportion of homozygous and dihybrid families is seen to be well in
accordance with expectation, but there are three times as many families
segregjiting into Solid and Holstein as into Solid and Watson. Further
if this cross were a simple c<ise of the inheritance of two factors it would
be found that the ratio of pure to heterozygous Holsteins would be 1 : 2.
This is not so. Out of 22 families grown from F. Holsteins, 12 segre-
gate and 10 are pure — a result which indicates that another hypothesis
must be sought to explain these facts.

Imagine that the Solid parent instead of being of the constitution
WWHH wjis WWH,H,H,H,, i.e. that it contained two factors each of
which could produce the Holstein pattern. And suppose also that each
Holstein factor could produce the Solid pattern when in assijciation
with th<' Watson fact«>r. The types appearing in the F., of the cross
WWH,H,H,H, by wwh.h.hA w(nd(l then be: 27WH,Ho(Solid), 9WH,h,
(Solid), 9Wh,H, (Solid). 9wH,H, (Holstein), riWh.h, (Watson). :3wH,h.,
(Holstein), 8wh,H., (Holstein) and lwh,h., (Siiiail eye).

The ratio between the four types would be 4-5S : lUV : \^)JI : 1^7i.
Of the 45 Solids 7 would breed true, 14 would throw N and //, 8 would
thn^w S and W. and the remaining Hi would throw all four types.

108 Inheritance in the Coivpea

Examining the F-^ families in the light of this hypothesis the matter
can be set forth as follows :

{a) The Solids.

Number of families

Expectation with IH factor
Expectation with 2H factors ..

S only
4-9
6-8

S:W:H'.SE

19-6
15-6

S:W
9-8
7-8

9-8
13-7

Obtained ... 6 18 5
{h) The Holstewis.

Number of families

H only H and SE
Expectation with IH factor ... 14-7 7-3
Expectation with 2 H factors ... 11-7 10-3

15

Obtained ... 12 10

From these results it is evident that so far as the F.^ families are
concerned the hypothesis is well confirmed that two Holstein factors
are involved in this cross.

The two H factor hypothesis leads to further important consequences.
These are :

(a) 4 Solids out of 45 should give in i^g a ratio of lbS:lW.

(6) 4 Holsteins out of 15 should give in F.^ a ratio of \bH : \SE.

A study of the F.^ families which show segregation into 8 and W
does not lead to the belief that the 15 : 1 ratio occurs. The number of
fomilies is small. Some of the segregating F^ Holsteins may possibly
show the 15 : 1 ratio, e.g. 1-10—7 (31 : 3), and 1—10—16 (16 : 1) but
again the number of plants grown is too few to determine this point
with certainty.

(c) Out of the 18 families which segregate into all four types in
F-i, 50 per cent, are expected to show the 9:3:3:1, and 50 per cent,
the 45 : 3 : 15 : 1 ratio. With such small numbers it is scarcely possible
to get more than an approximate idea of whether both types of ratio
occur, but taking the families as they stand we have :

Family No. 1—10—26.

8 W H SE

Obtained 40 1 8 1

Calculated ... 35-1 2-3 11*7 0-8

Family No. 1—6^—9.

S W H SE

Obtained 66 5 19 1

Calculated ... 63-9 4-3 18*3 1-4

on 45 : 3 : 15 : 1 basis.

S. v. IIahlam) \{)\)

Thum* iX'.Hiill^ slnui^I}' MU^^i'Ht that thr 4/) : M : I *> : I niti(» <l«Mh ikhmit.

In ri'^inl to th«' Ix^havioiir nf tin* \Vatji«»nH it is clmr that wh<*th<T
t»iio or two Holsti'in factors arr cnnciriu*!. l\v<» WjitMnn.s nut of thn***
wouKi Ix- iuonohy)tri() himI the thii^l \voul«l hn^Ml trin'. 10 liiinilicH wm*
^»\vii fn»in F, Watsons. All prnvoH t«» Ix* iimnuhyhrnl. The al)H4ii<M-
of the |Hin' ty|H' sr^ins to lUM'd liirtlui inv«'sti^ali<in. Small .m- ImIi.ivcs
in /', a(x*onliiik( to I'xjKrtation ami l>rr«Mis trur.

A lurthrr |H»iut imw iummIs «li.s(Missi(>n. ( )m the \\\<* H tarim Ii\|k.-
thosis thr nitio nf H Lo h in F. slmuM hr !,*) : I, .ind ihr rat in iH-twcrn
N, W, If. and .S'/s sh.Mild hr 45 : .S : 1.') : I. This is not totind U* \h- ih*-
wuH'. Thr nitio n\ 49 : l() in F, of H to h is far fnun <'X|KCtatinn, an<l
tiMusion has already hwu niadr t»» the ahiKunial lalius foinxl in the
het<.M\)zvi,^»iis fannlios of th<' /',. If the t\v(» H factor hyi>oth«sis is
corn.'ct how can tho diviation from the 15: I ratio he accounted fnr !*
Two possible explanations sugi^est themselves.

{(t) The Solid parent may have been heterozygous tor on*- <»i ihc
two H fact<»rs, and in that wise some of th«' F.^. familit-s would show the
9 : .S : 8 : 1 ratio and others the 45 : *3 : 15 : 1 ratio. The numbers ol)-
tiuneii in the F.2 families are too small for any discussion of this p(»int
to have much value. As they stand, however, they do not lend any
support to this theory.

(6) The ratio of 4*9// to III in F.^ may be accounted for by assuming
linkage between the two H factors. The /^., results could not be ;ie-
counted for by such an assumption. No hypothesis can at present be
suggested which will explain in all the results. The (piestion is evi-
<lently a conipliciited one, and the whole experiment is being lepeated
in the ho^x* of elucidating the various points that have arisen.

The unexpected results which were obtained in Cross 1 in respect
of the inheritance of pattern on the seed coat led to the study of two
other crosses between Solid and Small eye, viz. Cross 2, Red Solid ])y
Black eye, and Cross 5, Red Solid by Brown eye. The results of these
two crosses will now be set forth and discussed.

(•ros.s2. Red Solid by l^''^'^'^ ^y"-
*^ (Small eye.

The F, was Solid. In the F. segregation occurred into Solid,
Watson, Holslein Saddle (types e and /' of Spillman). Holsttin Large
eye s}K)tt<'ci, and Small eyt*. All the Holsteins weic classified t(»gether.
From the numeriwil n-sults, which are presented in Table X, it will be
seen that the ratio of ir to ir is close to :i — 1 while the ratio of // to //

110 Inheritance in the Cowpea

is 312 : 14, i.e. 22-8 : 1*0. Further, the numbers obtained of each type
correspond closely with expectation on a ratio of 45 : 3 : 15 : 1. It seems
that the hypothesis that two Holstein factors might be involved in a
cross between Solid and Small eye holds good in this case without the
peculiar complications found in Cross 1.

TheF,.

The F.^ results are to be found in Table XI. The chief points worthy
of notice are as follows :

1. Solid either bred true, segregated into S, W, H, and SE, segre-
gated into S and W, or segregated into S and H.

The number of families of each of the above types was well in accord-
ance with the two factor hypothesis.

Number of families

Type of behaviour

-Sonly

S'.W:H:SE

Obtained
6
9

Expected
4-4

8-9

S:H

8

7-8

S:W

2

3-9

It would seem that in the families which segregate into all four
types, occur the two types of ratio 45 : 3 : 15 : 1 and 9:3:3:1. As an
example of the latter ratio, may be given family No. 2 — 4 — 40, and of
the former ratio, family No. 2 — 11 — 20.

s

w

H

SE

Obtained

57

18

32

8

Expected

63-6

21-5

21-5

7-2 (on 9 : 3 : 3 : 1 basis)

Obtained

53

3

22

1

Expected

55-5

3-7

18

1-2 (on 45: 3 : 15 : 1 basis)

2—4—40
2—11—20

The two families which segregate into Solid and Watson are
apparently both of the 15:1 type.

The Solids which segregated into Solid and Holstein did so in the
3 — 1 ratio.

2. Watson should have exhibited two kinds of behaviour, (a) bred
true, (6) segregated into Watson and Small eye. Actually all the 5
Watsons segregated into Watson and Small eye in the 3 — 1 ratio. It
is somewhat remarkable that no pure Watsons were isolated. The
same point was noticed in regard to the Watsons of Cross 1.

3. Holstein bred true or segregated into Holstein and Small eye.
Had a representative group of F2 Holsteins been taken for the F.^
cultures the ratio of pure to heterozygous Holsteins should have been
7 to 8. Ten Holsteins were purposely selected showing the typical

S. C. Harland 1 1 1

HolsU'in iMittom — tin- S*mIHI<« form (ty|M's e ami /' of Spillinun), utui
5 of tht' I^irj^' vyv spi»t.t4Hl ty|M». It will Ix' nrtj that, th*- SjuMU* lonn
Imni triu' while the I^ir^» vyv s|x)ttr<l pmvcMl to Ix* hrtcinzv^nHis. Id
m»j»i\»gatin^ fiiinilieH the ratio of // to A apiM'und to hr of tin- \'> : 1
ty|K\

4. Ni» fainiliej* wimh' ohtaiurd showin^^ tin- lu-havioiu nl the Small
eye fbnn in A*,.

To sum up: on tJn' wholr. thr cvidriicr from this cross is in favour
of the view that it is of the nature W H, H. I>y w h, h...

..... I . (Brown eye;.
Cross :\. Hed Solid by -1 , ,, ^
•^ (Small eye.

Thi* h\ of this er(»ss was Solid. In the F. scgretrat ioii oecurred into
Solid. Watsson, Holstein (Sad<IIe and Lir^^c rye s|M»ti<'d) and Small ey«'.
The results are presented in Table XII.

The ratio of II' to ir w;vs a<(ain approxiujatdy .S — I. while the latio
of// t-o /* w;vs lOG to I(). From a pr<liminary surxcy of these figures
it seenus probable that the civkss is of the same type as Cross 2, i.e.
W H, H._. by w h, h... The ratio between the four types w;is near to
•io : 8 : 15 : 1. In the F^ a large number of familii's were grown from F,
Holstvins in order to obtain more exact figures as to the mimber of
types which were pure and heterozygous respectively. Broadl}- speaking
the Holsteins were classified into two groups, (a) the Large eye spotted
form, and {b) the Saddle form {e and/of Spillman).

Of 106 Holsteins examined in F. ^)2 were; of the former and 54 were
of the latti'r ty|K'.

The F,

The /'.; residts ai-e pi-e.sented in Table XIII. Th<' main [joints are
as follows :

1. The residts obtaim-d conform generally to tho.se of Cross 2.

2. The families from F., Solids wei-e 2^^ in nundier. The number
of families which bred true, segregated into *S', If, //, and .S'A'. into
«S' an<l IF and into *S' and //, was again in accordance with the two
H faet^»r hviM>th<'sis, as will hv seen from the summarized lesulis IkIow.

Nuiiil>er (

iif fain

ilifs

_,_

Tyi>e of Ix-liavidur

ohtaimnj

K

X|K,-<t.-.l

Pure S

•J

•^i)

S: U:I! : SK

7

;^-2

S and If

I

11

•S and //

10

7 1

112 Inheritance in the Cowpea

3. Holstein Large eye spotted proved to be invariably heterozygous.
Some of the Holsteins were obviously of the XhH : \^E type, but
families of the SH.lSE type— e.g. No. 5—5—20 (52 : 18) were also
present. In many families the number of plants is too small to decide
which type of ratio prevails. /Pwo of the families apparently segregated
into Large eye spotted and Small eye. In these families classification
was difficult and some of the plants were referred first to the Large eye
spotted form and then to the Saddle.

4. Holstein Saddle proved to be homozygous. There was great
variation in the extent to which the pattern was developed on the seed
coat, but it was not possible to effect an accurate classification of the
differences. It may be possible to distinguish by external appearance,
the different Holstein combinations, but this would need a more careful
examination than the present writer has been able to give.

It will be noticed that a few of the families threw an occasional
Solid. This is probably due to accidental crossing by the carpcntei* bee,
Xylocopa sp., and the set of results can scarcely be discarded on this
account, though due care should be exercised in weighing the whole
mass of evidence.

Since Holstein Large eye spotted is always heterozygous and Holstein
Saddle homozygous, it is interesting to go back to the F2 results and
consider them anew. In the Fo of this cross there were altogether
116 Holsteins. Of these 10 were not classified further. Taking the
rest, 52 were Large eye spotted, and 54 were of the Saddle form. On
the two H factor theory the ratio of pure to heterozygous Holsteins
is 7:8, and the ratio actually obtained 7*6 : 7*4 (54 : 52). Thus the
evidence from the behaviour of F.^ Holsteins in F. is strongly in favour
of the two H factor hypothesis.

Summary of Results of Investigation on Pattern of the Seed Goat.

In two crosses involving Red Solid and Small eye it has been shown
that three independently inherited factors are concerned. These are :

W — The factor for the Watson pattern.

Hi — A factor for the Holstein pattern.

Hg — A second factor for the Holstein pattern, having the same
effect as Hj. A combination of W and either of the Holstein factors
produces the Solid type of pattern, while absence of all three factors
produces the Small eye pattern.

S. ('. Hahlani) Wi]

In aiioth(*r cmHs of S<»li<l iukI Small ryv |M'<Miliar nitioM wfw
obUiiruil. though tho behaviour of (vrtnin n| th«' Fj h«'tvn»/y^<)UM
fainilit*»4 in(iinit4M| that thror t'actoi'H wt»n* aluo coikmtihmI.

Thr rxist^'HCr nf two Hi>lst'<Mt) fartors prcMhirin^ ihr sjuiH viMihli-
ort'tH't. alono or in rontbination, is analu^Mus t^» th** cljissic (mm<' oI tin-
colour of wheat i^niins Mtu<li<'<l by Nils.s4»n-Khh' (IIM)H) .ind bv tin-
Howarris (1912). Thry foun«l that thr R«m| rojonr charactir in c.rtani
typi's wjus ihw t4> thrco sojianitc nnits f«»r red. NiJHson-EliJc also
ob8orvo<l that a bhvck konn'llo(| nat variety may |k>ss(ss nion* than nn«-
unit for black, t'arh unit alnm- U'int; able to pnMlucr the tvjMcal blark
colonr.

Ono |M»int nnist htic In* cniphiusiztMl most strongly. The prrsmt
writ4'r, fvt'n if lu' h;Ls I'stablishiMJ little ronclnsivdy, hjis at l<*;ist drmoFj-
stnit^nl the inadvi.sjibility of fitting a series of results to the nearest
pntbible nitio. and then ;iscril)ini( exce.ss of defieieney of certain t.v|Ms
to chanci" or to tlu' small innnlx'r of plants t^i-own.

The results t)btaine(l by I )r S[)illman are now <'.usily explained.
The Solid forms used by him must have be«n in constitution W H, h_,
or W h, H,. Kither of these would give an F, ratio of 9.S' : 8 M' : .S// : I S/^J,
when cr«>.s.sed with Small eye.

III. The CoLorR of the Pattern of the Seed ('o.vt.

(a) The /actor H.

One of the pjirents of CnKss 1 , the Black eye variety, is characteri.sed
by the apjK'amnce of a <lark red or purple coloration on the tip of the
young jxxl. The colour first a|)pears when the pod is two oi- threr davs
old, and lasts until the [x>d is nearly ripe, when it is obscured by the
drying up of the \hm\ which then takes place.

The red ti|) of the immature pod is a pt^rfectly distinct chamcter,
and is found to occur only in plants which pos.se.ss black colonition of
the t^^'St^i {);\tt4*rn. It is always accompanied by more «>r less n*d colour
in the calyx and jx-duncle.

It has been |)ointed out by Morgan et al. (1915) thai a so-called
unit character is only the most obvious ov most significant product of
the j>Kstulat«d factor, and that a single factor may affect a plant or
animal in many different ways. An example of a single genetic difien-nce
which affects th.- <ntirc organization of a plant is seen in the jx-culiar
types known as (Vinkled I)warf Rogues which occur occiisionallv in
fields «»f Sea Island Cotton. Tli*- present wiitci- (19h))lias shown that

114 Inheritance in the Cowpea

the normal Sea Island type behaves as a simple dominant to the
Crinkled Dwarf Rogue. The latter type differs from normal Sea Island
in respect of almost every morphological character and also in certain
physiological characters. The whole plant is much smaller ; the leaves
have a crinkled and mosaic appearance and have ragged edges. Both
buds and bolls are shed much more easily than in Sea Island and there
is a general reduction in the size of all parts of the plant.

In the case of the association of the red tip of the young pod with
the presence of black in the testa, and of coloration in calyx and
peduncle, it will be shown that this combination of characters as a
whole is allelomorphic to the absence of such a combination, i.e. to pod-
tip devoid of colour, seed devoid of black (in this case the colour of the
seed is brown), calyx and peduncle without red colour. This may be
explained by assuming that several completely coupled factors are
concerned, but it seems more reasonable to suppose that all these
effects are manifestations of a single factor, B.

The Experimental Results.

Cross 1. Black eye by Rounceval.

Black eye. Black present in seed coat ; young pod with red tip, red
colour present in calyx and peduncle (B).

Rounceval. Seed coat uniformly brown, immature pod, calyx, and
peduncle green {h).

The F^ showed complete dominance of B.

The F^ and F^ results.

In Tables XIV and XV will be found the results of the F^ and F^.
Reference to these two Tables* will show that the ratio of B to h in
F2 is 3 — 1, that in F.^ all the families from F2 recessives bred true and
that of the families from F^ dominants 18 bred true, while 40 segregated
into B and 6, again in the 3 — 1 ratio. The ratio of pure to hetero-
zygous dominants is very close to the expected 1:2.

To sum up : the two sets of characters comprised under the symbols
B and h, are allelomorphic to each other.

(h) The Brown and Red colour of the Pattern of the Seed, Goat.

In the cross Brown eye by Red, a study was made of the inheritance
of the colour of the seed coat pattern. The extent of the colour varies
with the extent to which pattern is developed. Thus in dealing with
Small eye types it is often impossible to separate different shades,
especially those which are even difficult to classify when the coloui' is

S. (\ IIaui.ani) IIT)

viMibIc owr thr whole tvMtn, ju* in tin* Solid ty|K's. In this jMirticiihir
cr<»«s, f«>r rxiiiiiplo. it i** otTtJiiu thai at Inixt iIuim- ditVonnf slmdrs of
bi-own ti))|H>nr in Kj, but owin^ to ihr (lifKnilty o| chiKsifyin^' iircnrahly,
the browns hav»» all Ikhui placed in oin- vIohh.

Brown-eye ran Ih* deserihcd as a brown with a lainl piirpir cast.
The cohair of R^mI is nither \u\\rr than that of the vaii^ty \{t'i\ Kipp« r.
which is cIjussihI by ( '. V. ri|x'r (lfM2) jis maroon. Pi|M'r stat<'s that
undiT th<' kjronp of pink seeded varieties is a nin^j* of colours fioin
viniuxHuia to brick re<l. When the |H'as an- a^ed thcne colours darken
so that they aix' v(Mt difficult to distiui^niish from maroon. In this
IMijX'r the c«>lour of the variety Rod will he described as red. It must
b<' underst^MHl that the results which an' h<'re described apply (»ni\ to
this imrticular cross. It is certain that th«' relations of the test.a
coloui's in cow|H';us are far from simple, and th(^ workint( out ot all (jI
them would Iv a task of no mean magnitude.

The Kvpentnental Results.

The plantsS of the F^ showed complete dominance of brown, but it
was n<n possible to tell whether the particular shade of brown of the F^
wjia the .s;ime ;us that of the brown eye parent.

The F,^

The results of the F., are presented in Table XVI. Segregation
iKx^urred into (a) browns of various shades, (b) dark maroon, difficult to
distinguish from black when quite ripe, (c) a maroon less intense in
colour, (d) red of the same shade as the red parent.

The plants were classified into three groups, brown, maroon, and
red. The ratio between brown, maroon and red was near to 1 2 : .'-J : I.
The ratio of brown and maroon to red was 18'2 : 10.

It may be concluded that two facto i-s are concerned, M pioducing
the maroon colour, and N producing the brown colour. N is (Inminaiit
over M an<l the tyjies appearing in Fo will be as follows:

f) N M Brown.
.S N m Brown.
.S n M Maroon.
1 nm Red.

The various shades ot brown and maroon may perhaps be accounted
for by iissuming that the het«'rozygot«'s aic dit!'<'rent in appearance from
the pure lyiKs.

In onler to contirm this hyjKttlusis a laim- nunibe?- (•! families were
grown in F,.

.Tourn, of (ien. viii S

116 Inheritance in the Cowpea

TheF,.

The results of the F.^ will be found in Table XVII. It will be seen
that :

1. Brown either breeds true (18 families), segregates into brown
and maroon (15 families), segregates into brown and red (7 families), or
segregates into brown, maroon and red (15 families).

2. Maroon either breeds true (6 families), or segregates into maroon
and red (11 families).

3. Red breeds true (12 families).

So far the results appear to confirm the hypothesis which has been
put forward. It remains now to examine the proportions in which the
various types occur in the segregating families, and also to see if the
number of families following the above types of behaviour is in con-
formity with the hypothesis.

The families which segregate into brown and maroon should do so
in the 3—1 ratio. The ratio obtained 348 : 107 (3*3 : 1*0) is fairly close
to expectation. The 3 — 1 ratio should also be obtained in families
segregating into brown and red. The actual numbers recorded were
18 red and 65 brown, giving a ratio of 2*9 : 1'O, which is again close to
expectation. The proportion of brown, maroon and red in families
splitting into all three types should be in the ratio 12:3:1. The sum-
marized results are placed below.

Brown Maroon Red

Obtained 474 llfi 63

Expected 489 '7 122-4 40-8

Here the results again appear to confirm the hypothesis though the
number of reds is somewhat above expectation.

The 3 — 1 ratio is expected in families which segregate into maroon
and red. The numbers obtained were 281 maroon and 119 red, a ratio
of 24 to 10. This deviation from the 3 — 1 ratio is perhaps due to
fluctuation.

(a) The Brown.9.

Number of families

Type of behaviour

Observed

Expected

Brown only

18

18-3

Brown and Maroon

lo

9-2

Brown and Ked

7

9-2

Brown, Maroon and Red

1.5

18-3

{h) The Mm

'oons.

Maroon only

11

llv3

Maroon and Red

6

r>-7

S. (\ Haklani) 117

It will now be conHidered whothor tlu* iminbur •>! liiiiiilit-h f«>ll«)win^'
the different ty|K»« of bt^haviour Ih in Jiceoniance with theory. A com-
parison of the observefi and ex|)«'ct«'<l resultH Ih given just above.

The nmin featun' «»f the above n»HuItjH '\h thr <iifference in the
numb(*r of families .s«»jrit»^itin^ int^» bmwn and numMin and brown and
ixmI n'8|)e<'tivt'ly. Thesr .should hv «M|ual in inimbcr wht're;us those
which follow the ft>nner ty|H' of behaviour an* nion* than double tho.M<'
which are of the latti^r tyiK\ Sinee. however, the h\ reaultH as a whole
confirm the two factor hyjH)the8is, it is rea.sonable to Huppf»8e that tho
deviation is due to tluct nation.

At this [K>int it may Ix' observed that the meaning of the various
i^hntles of colour found in F., Ix'came a littli' clearei- by observation of
the F3 families. The observations may be sinnmari/ed us follows:

1. The brown which seiriegated into brown an<l mariMin. oi- into
bi-own man>on and red, wjus slightly tinged with purple.

2. The bn»wn which si^*gregatt<l into brown and red was very jjiile
and devoid of purple. Then- w.us apparently no difi'orence between
Nm Nm. and Nm nm.

.*!. The darkest shade of maroon was always homozygous. The
intenneiliate maroon segregated into daik maroon, intermediate maroon
and red in the 1:2:1 ratio.

Suminari/ of Investigations on the Inlieritance of Pattern Colom:

1. The black iwittern colour of the vaiiety Black ey<' is due to a
single factor B, which is dominant to its absence.

2. The factoi*s involved in a cross between biown and red are two
in number. These are N — a t'actoi- which causes the production of
brown, an<l M — a factor which produces a maroon colour. N is <lomi-
nant over M. In the absence of both factors the seed coat is re<l. N
and M are indejxmdent of each other in inheritance.

lieldtion of Flower Colour to Pattern of the Seed Coat.

In the cro.sses between S(>lid and Small eye, it was noticed that a
genetic correlation existed betwien pattern and flower colour. Thus
plants c<mtaining the Watson factor were always dark flowered. Plants
with no Watson factor, i.e. the Holstein and Small eye types, weic p.ile
tlowered. It may be concluded that Dark flowei- and Watson pattein
are both manifesUitions of a single fact<a'.

If the flower is white there is no pattern on th(? seed coat. In a
cross between Black eye and a white .seeded type (Para), th«' /-', was
Black Solid and in F.. segregation occun-e(l into Black Solid, lilack

118 Inheritance in the Cowpea

Watson, Black Holstein, Black eye, Brown Solid, Brown Watson, Brown
Holstein, Brown eye, and White. The ratios between these types have
not been fully worked out, but it is clear that the factor for Dark
flower and Watson pattern, for Holstein, and also for Black, can be
carried by an albino type (white flowered and white seeded).

The subject of colour correlation in the cowpea has been discussed
by Dr Spillman (1913). His note, however, is short, and does not con-
tain the experimental results on which his conclusions are based. Some
of his statements may, however, be discussed. According to him all
varieties having white or cream coloured seeds have white flowers and
are devoid of anthocyanin in stems and leaves. This is not always true.
In the cross Red Ripper by Para, a certain number of types in F^ have
white seeds and are devoid of anthocyanin in stems and leaves but also
have a distinct violet tinge on the wings of the flowers.

He further states that the flower colour and the anthocyanin in
stems and leaves are dependent on two Mendelian colour factors :

{a) The general factor for colour in the seed coat.

(6) The special factor for black which when added to a variety
having coffee coloured seeds converts the seed colour to black.

It has been shown in this paper that the factor for black affects the
seed coat, the pod tip, the calyx and the peduncle. It does not appear
to have anything to do with the flower colour or the factor which causes
anthocyanin to appear in stems and leaves, except that it is without
visible effect except in presence of such a factor. It is, of course, possible
that Spillman was working with a different factor for black, and all that
the present writer wishes to point out is that Spillman's conclusions are
not of universal application.

Summary.

1. In the preceding pages an account has been given of the mode
of inheritance of factors affecting the flower colour, pattern of the seed
coat, and colour of the seed coat pattern of Vigna sinensis. These
factors are :

L. A factor which produces the type of flower colour known as
Pale. The factor has visible effect only in types with the Holstein and
Small eye patterns. It is very possibly the factor mentioned b}^
Spillman (1913) as the one responsible for the production of anthocyanin
in stems and leaves.

D. A factor which has no visible effect except in presence of L,
when it converts the Small eye pattern to the Watson pattern and the

S. (\ Haki.ani) 119

Ht>Uu»in p}itt4*rn to iStilid colour, at ihv sjimr time chHU^iii^ thr flnw«T
colour t4) Dark.

H,. The fiicUir which c«>iivcrtss the Sumll i>yr |Mitt«Tn tn thr H«ilHt,cin
pHlU>n).

H^ The crtwt o( this fiictor is similar to t hat ai' H,.

B. The faot-or Tor l>l.ick in the seed coal, which also manifests itwelt
by the pnxiuction ol a red lipped \mh\, and a calyx and pednncU* with
more or less anthtn-yanin pigmentation.

N. Thi* factor tor a hrown or biitV colour of the s«'ed (!oat pattern.

M. The tiu'tor tor a liark mar<H>n t(»8tii pattt-rn.

2. In one cross between Small eye and Solid (colour the behaviour
of certain (»!" thi* h\ families indicated that both the Holstein fjictors
wore invi>lve<l. The rati«» of Holstein to \o Holstein in F.. was, how-
ever, widely nnnived from thi* «'xpi'cte(i 15:1, being ^fh TO. It is
point'Od out that linkage between the two Hnlstein factors woidd a('(M)unt
for the Fj results but not for thnse of /^ No hypothesis can at present
be suggeijted which will account for both F.^ and F:i results.

3. The following pjiirs of factors apjx'ar to be inherited inde|X'n-
dently of each other: L and D. D and H,, D and H,, H, and H,. M
and N.

4. A short statement is put forward in criticism of certain results
obtained by'Spillman (191*^) in regard to colour-correlation in the
cowpea.

REFERENCES.

Harlaxd, S. ('. 11)10. -'Oil the Genetics of ( Viiikled Dwarf Rogues in S&v Isknd

Cotton." iV*^it Indian Bulletin, Vol. .\vi. pp. 82-84 and 353-355.
Howard, A. and (i. L. C. 1912. "On the Inheritance of some Charactei-s in

Whciit, I.' Mem. Dept. Afjr. Indi'f, Bot. Ser. 5, Xo. 1, pp. 47.
MoROAN, T. H., Sturtev.vnt, a. H., Mullkr, H. J., and Bridges, ( '. B. 11)15.

"The Mech.uiisni of Mendelian Heredity.'
XiLi>.soX-EHLK, H. 1908. "Eiuige Ergel>nis.sc von Kreuzungen l)ei Hafcr und

Weizen." Botaniabi Notiser.
Piper, (\ V. 1912. " AgricMiltural Varieties of the Co\vi)c;i and Inmiediateiy

lielateii S|>ecie.s." C.S. Dept. of Af/r. Bur. PL Indm. Bid., 229.
Spillman, W. .1. 1911. '' Inheritmce of the *Kye' in Vigna." Ainericatt

yatnraliM, 45, No. 537, i)p. 513-523.
. 1913. "Color Con-cliition in Co\v[>e;is." Science, X. Ser., 38, Xo. 974, p. :W2.

120

Inheritance in the Cowpea

TABLE I.

The F^ results of crosses between Dark and Pale.

Cross

Crosses involving

(1) Black eye (Pale) x Kounceval (Dark)

Family
1
2
3
4
5
6
7
8
9

—10
—G

Dark
26
22
11
30
51
24
108
38
90
20
54

Pale

12

13

5

7

9

8
42

9
25

8
19

Totals

Total all crosses

Calculated...

423
1185
1183-5

145

394-5

Ratio
Dark to Pale

Totals

474

157

3-0:1-0

(2)

Black eye (Pale) x Bed (Dark)

2— 3

75

19

2— 4

69

21

2— 5

13

10

2— 6

19

9

2—10

32

10

2—11

37

12

2—12

43

10

Totals

288

91

3-2:1-0

(5)

Brown eye (Pale) x Red (Dark)

b—G

77

41

5— 1

75

25

5— 3

16

8

5— 5

46

13

5— 8

15

3

5— 9

105

29

5—11

26

9

*

5—12

41

13

5—13

22

4

2-9:1-0
3-0:1-0

S. ('. IIaki.and

ILM

I'AHLK 11

RftuU'H ttt the t\ ifnirrnhoH. /'\itni/trK from /'*, l)iirln< Hrt/irtffifnu/
iiiftt Dark atuf J\t/r.

Kwntly

lUik

Till.-

K»inll.»

hark

l'»t.<

CruMl. 1

r> A

10

1

CrosH 2.

2 3 59

2H

7

Black ovo I

5 1

M

n

Black cvr

2 3 -64 »

42

13

by 1

7 i

10

6

hy I{4><1

2 3 66

11

13

liuunnval ]

- 7 M

12

1

2 3 -83

fV.i

35

^ 7-11

27

12

2 4 40

76

40

7 12

14

2

2 1 43

16

5

7 17

66

27

2 I 48

35

H

7 1'.)

16

2

2 - 4 51

41

17

7 21

IH

4 1

2 4—78

37

11

7---2,'5

11

3

2 4 83

71

36

-^ 7-29

16

5

2 11 19

41

16

— 7—39

5o

26

2-11—20

56

24

— 7—40

15

M

2 -11-36

71

17

- 7-45

61

23

2 11—37

49

16

— H- 1

28

10

2—12-17

24

16

— H— 3

30

12

2—12-18

32

12

H— 9

30

9

2-12—27

47

11

— 8—12

21

8

2-12-31

32

11

I - 9— 1

26

6

2-12-37

14

1

I- 9— 3

16

7

2—12- -47

44

16

I— 9— 9

13

3 '

2-12—49

4

3

I - 9—10
— 9—12

16
16

12
1

Totals ...

890

331

l_ 9_15

16

3

Calculated

915-8

305-2

— 9—19

6

7

iiatio of Dark to Pale .

2-7

10

— 9-21

12

1

L— 9—23

11

2

L— 9 — 33

25

15

Cross 5.

r,_ 1-37

17

5

I— 10— 1

24

8

Brown eye

5— 1—50

17

7

l_10_ 2
I— 10— 3

45
15

17
4

by Red

5— 1-66
5— 6-17

15
11

4

4

l_10— O

52

25

5— 8—20

20

2

I— 10— 9

31

10

5—11— 2

8

10

1—10—12

•22

()

5—11—46

15

4

1-10-17

22

12

5—11—48
5—G - 1
5-G - 2
5—G - 7

15
13
17

6

7
3

4

[ -10— I'.t
[_10— 20

9
11

2
2

1-10—23

9

2

9

I _ 10— 26

12

8

5—G —10

12

4

1_G— 6

71

19

5—G —13

12

7

I— G —23

28

14

5—G —Un

20

15

I— 9— 7
1_ 9_28

21
25

10

ry—G —lib
5—G —23

6
23

3
2

1_ 9_96

65

19

Totals ,.

-,__G —26
5- G -55

18
22

270

7
9

Totals ...

11-22

400

103

Calculated

1141-5

380 -5

Calculated

276-8

93-3

Ratio of Dark

to Pale ..

2-8

: 10

Ratio of Dark to Pale . .

2-6

10

122

Tnheritance in the Cowpea

TABLE III.

Results oj

F^ generation.

Families J

roni, F.2 Darks breeding true to Dark.

Family

Dark

Pale

Family

Dark

Pale

Cross 1.

L_ 7_12

15

0

Cross 2.

2_ 3—15

17

0

Black eye

I— 7—29

10

0

Black eye

2_ 3—17

60

0

by

1_ 7_46

41

0

by Bed

2— 3—82

28

0

Kounceval

1- 7-75

41

0

2— 4—24

16

0

1—8—2

15

0

2 4—41

59

0

1_ 9_ 2

6

0

2_ 4—44

43

0

1— 9— 8

20

0

2- 4—54

43

0

1— 9— 9

6

0

2—12- 3

35

0

1_ 9_io

28

0

2—12—14

19

0

1_ 9—11

10

0

2—12—26

52

0

1_ 9_23
1_ 9_26

13

6

0
0

Total

...

372

0

1_ 9_96

84

0

l_10_ 2

41

0

Family

Dark

Pale

1—10—19

8

0

Cross 5.

5— 8— 55

19

0

1—10—23

9

0

Brown eye

5— 8—106

23

0

I_10_25

32

0

by Bed

5— G— 5

23

0

1—10—26

49

0

5— G— 21

45

0

l_G —24

81

0

5— G— 54

35

0

Total

515

• 0

5— G— 56

14

0

Total

159

0

TABL

E IV.

1

^iesults of F.^

generation. Pales breeding true to Pales.

Family

Dark

Pale

Family

Dark

Pale

Cross 1.

— 5—3

0

8

Cross 2.

2— 3— 80

0

48

Black eye 1

— 7—7

0

115

Black eye

2— 3— 81

0

16

by 1

[— 7—11

0

39

by Bed

2— 3— 87

0

36

Kounceval

L_ 7_22

0

4

2— 3—103

0

28

L— 7—39

0

80

2— 4— 42

0

12

L— 7—47

0

11

2— 4— 53

0

49

L_ 7—73 ,

0

96

2— 4— 62a

0

6

L— 8— 1

0

39

2— 4— 62&

1

60

L— 8—10

0

41

2— 4— 65

0

49

L— 9— 1

0

31

2— 4— 75

0

19

L— 9— 3

0

23

2—10— 1

0

14

L_ 9—12

0

17

2-10- 21

0

4

i— 9—17

0

16

2—10— 27

0

25

L— 9—20

0

5

2—11— 13

0

26

L— 9—21

0

13

2—11— 21

0

6

L— 9—22

0

14

2—11— 42

0

47

L— 9—24

0

21

- 9—32

0

42

Totals .

11

445

L— 10— 12

0

31

—10—16

0

17

Family

Dark

Pale

—10—21

0

86

Cross 5. ■

5— 1— 9

0

19

—G— 6

0

90

Brown eye

5— 1— 10

0

34

— G— 23

0

41

by Bed

5— 1— 17

0

11

— G— 65

0

27

5— 1— 19
5— 1— 28
5— l_ 55

0
0

0

26

18

8

Total ...

0

907

1 Probable

vicinist.

S. ('. IIaklani)

12:]

TA BLK I V — rnutinurJ.

KmhIIv

l>iirk

rail-

CroM5.

.*»

1

7U

0

Jl

Brown ore
byR«d

&-

•J

n

(1

:.9

y-

» -

15

1)

41

:»

»~

IB

0

:iO

.*» -

:»-

I

5

(VI

&- B~

12

0

18

.*» —

A-

lA

0

24

A-

A—

M

1

70

5-

A-

3H

I)

.Vi

ft

5--

51

(1

82

*-

A-

45

(1

35

5-

6—

5a

(1

28

5—

5

54

0

7

h-

8—

4

0

HH

5

8-

18

2

37

?"~

8-

3&1

0

24

5 —

8—

m

1

:^H

5—

H~

31

1

ni

5—

8—

3.-1

u

6<;

"> —

8—

.-H)

I

82

."> —

8-

59

2

35

5-

8~

67

0

41

5 —

8—

77

II

78

5-

8-

79

0

65

,->—

8-

81

0

81

5—

8— 85

0

35

5—

8

94

3

47

5—

8—101

0

49

/> —

8—109

0

36

r>—

8—111

0

33

5 —

8—113

0

12

5-

8—115

0

44

5 —

8-

117

0

92

5—

8—

135

3

40

Katnlly

l>«rk

l'»ir

5 9

3

(1

30

:> 9

(i

n

93

:. 9

8

0

59

:. 11

26

0

22

5 11

36

<l

17

5 -11

.w

1

57

5 12-

11

0

44

6 G

10

3

35

5 -n

12

(»

20

5 G

16

U

22

5 G

19

(1

tl

5 G

20

0

19.

5 -0' -

22

0

22

5 G

24

0

22

.•i -0-

25

0

K

0 -fi

27

1

M

r. G-

31.i

0

»«

r> -G

31/;

2

28

5 -f? -

41

{)

54

:> G -

45

1

23

5 «

49

0

24

5— G -

.",0

0

13

ry—(J -

51

0

15

rt—G -

58

0

16

H—G —

77

0

20

5-G —

86

0

23

5— G-

88

0

6

5— (;

98

2

•25

5— G -

100

0

70

5— G ^

104

II

«;

5-G -

A

0

30

Totals

30'

262r,

Probably vicinists.

Table v. Th*i F., ref<nlts of crosses hrhvenn, Dark and Wh'iln.

Family
8—1
8—2
A—\
A— 2
J— 3

Totals

Calculated

Katio

Dark
25
24
12
25
25

111^

113-3

2-8

White

8
12

5

s
7

40

37 -H
10

TABLE VL 71t'- F, r«'sidts 0/ crosses belu:etii l>arkoini \Mi\t>
1. Families from f'^ Darks segregating into Dark and White.

Family
6—1 — 1
6—1—2

Totals
Calculated
Ratio ...

Dark
21
32

53
57
2-3

White

8

15

23

Family G— 1 — 3 from a White h'l threw 45 whites in ¥

124

Inher'itance in the Cowpea

TABLE VII.

TheK,

generation of the cross Small eye

(type c)

by Solid colour.

Family

Solid Watson

Holstein

Small eye

Cross 1.

1_

- 1

19 5

9

3

Black

eye

1-

- 2

13 2

12

0

by Brown

1-

- 3

7 3

2

2

1-

- 4

19 6

6

0

1_

- 5

36 5

6

2

1_

- 6

18 2

5

1

1_

- 7

88 13

28

9

1_

- 8

31 7

6

3

*

1_

- 9

80 9

20

5

*

1_

-10

16 4

7

1

1

1-

-G

38 12

18

1

Potals ...

365 68

111)

27

:■■■

3261 109

109

36

1 Highest

expectation on 9 : 3

: 3 : 1 basis.

TABLE VIII.

Showing proportions in

which W and H occur i'n

the F.j^ of Cross 1.

. Family

W

w

Jiatio

H

h

Eatio

1— 1

26

12

28

8

1— 2

22

13

25

2

1— 3

11

5

9

5

1—4

30

7

25

6

1— 5

51

9

43

11

1— 6

24

8

23

3

1— 7

108

42

116

23

1— 8

38

9

37

10

1— 9

90

26

100

14

1—10

20

8

23

5

1— G

54

19

56

13

Totals ...

474

157

3-0: 1-0

485

100

4-9:1-0

Calculated

473-3

157-8

438-8

146-3

TABLJ

E IX.

Analysis of pattern in F^. Gross 1. Black eye by Rounceval.

For convenience use will be made of the following symbols :

fif = Solid, fl'=Holstein, jr=Wat8on, £ = Smalleye.

Family

s

w

H

E

(1) The Solids ]

L- 7— 9

24

1

9

1

L- 7—11

25

2

9

3

L_ 7_12

21

5

7

1

L— 7—29

5

2

1

2

L— 7-25

17

5

4

1

L — 7—45

51

10

21

2

L— 8— 1 •

28

1

8

2

L— 8— 3

24

5

10

1

L_ 8—17

17

4

7

0

L— 9— 3

15

1

7

0

L_ 9—19

3

2

5

0

I- 9-23

8

3

2

0

L_ 9—27

11

4

21

1

L_10— 19

4

2

2

0

S. ('. IIaui.am>

ii>:>

TAHLK l\ rotUiiiur,/.

Katnllr

>

II

K

1 The Sill uU 1 10 20

7

n

0

1

1 10-26

10

K

I

1 G- «

A<1

19

1

1 — « 28

27

11

3

1 9 11

H

_,

1 9-22

10

1 9 -2.5

H4

11

1 10 27

HH

13

1 7 - 2

9

__

6

1 - 7 l'.»

ir,

2

1 7 21

IK

—

4

1 7—40

15

-

7

1- 9 9

3

3

1_ s~V2

1(>

--

1

1 - 9 15

15

3

1 - 9—21

12

1

1 9 9r.

()5

19

1 10- 1

24

—

8

1 10- 2

31

—

10

--

1_-10- 3

15

4

1-10-^ 5

52

25

-

1_10_17

20

3

1_ 10-23

7

—

2

1_ 7_45

41 "^

l-_ 8—10

41

—

—

—

1— 9— 8

20

1— 9—17

16

_

1_ 9—24

21

—

—

1_ 10—25

32

—

—

—

> w

H

E

Ratio

Summary : 393 56

151

20

W : w
H : h

= 449: 171 = 2-6: 1-0
= 544:76 =72: 1-0

140 29

H:h

= 4-8: 10

317 —

103

—

W : ic

= 3-1: 1-0

Family

WH

Wh

>''H

>i-i.

Tli« Wateontf 1—5—3

6

—

2

1— 5— 4

27

9

1- 7-17

—

66

26

1_ 7_39

—

54

26

1— 7—47

—

10

• —

1

1— 9— 1

25

6

1_ 9—10

16

12

1_ 9_28

—

19

—

4

Totals

—

223

—

86

Calculated

—

231 -8

—

77-2

Ratio

—

2-6

—

10

.The HoUteinH 1— 7—18

_

_

U

2

1_ 7_27

17

4

U 7—73

—

79

17

1- 8-14

25

10

1— 9— 2

_

5

1

1- 9—11

—

—

9

1

1_ 9_20

3

2

1-10- 7

—

—

31

3

126 Inheritance m the Cow pea

TABLE IX — continued.

Family \VH Wh wH wh

3. TheHolBteins 1—10—16 — _ 16 1

1—10—18 _ _ 14 1

1—10—21 — _ 69 17

1_G— 24 — — 66 15

Totals — — 348 74

Calculated ... — — 316-5 105-5

Ratio — — 4-7 : I'O

1_ 7— 6 — — 16 —

1_ 7_22 — _ 4 —

1_ 7__30 _ _ 8 —

1_ 7_35 _ _ 18 —

1_ 8— 2 _ _ 16

1— 9_32 _ _ 42 _

1—10— 8 — — 14 —

1—10—13 _ _ 46 —

l._G— 53 _ -_ 87 —

1— G— 65 — — 27 —

4. Small eye 1—7—7 — — — 117

1— 7—46 _ _ — 41

Total — — 227

1—7—7 — — —

1— 7—46 _ _ —

1—10— 4 — — — 37

"tS^I 7.. 7. — ^^^ — 195"

TABLE X.

TJbt #2 "fesults of the cross Red Solid by Black eye.

Family Solid Watson Holstein Small eye

2— 3

64

5

18

2— 4

63

3

19

2— 5

3

4

—

2— 6

12

6

2—10

31

1

9

1

2—11

36

—

11

1

2—12

37

3

9

—

Totals

246

12

76

2

236-31

15-8

78-8

6-3

1

Expectation on

a 45

:3:15

: 1 basis.

TABLE

XI.

The F^ results of the cross Red Solid by Stnall eye.
,S=Solid, fF=Watson, if=Holstein, E = Smalleye.

Family

s

w

H

E

The Solids 2— 4—40

57

18

32

8

2- 4-48a

14

2

5

2_ 4— 48&

28

6

4

2— 4—83

65

1

35

2-11—19

11

0

5

2—11—20

53

3

22

2—11—37

47

1

15

2—12—17

18

3

12

3

2 12—47

39

5

15

1

S. ('. Haklam)

1-j;

TAHLK :

\\

rovt

iiinfd.

KmuIIt

,v

W

II

t:

1. ThoSolidi. H 4-41

lu

2

^ 4 Hi

42

1

—

2- 8 Att

2r»

7

•i- 8- (M)

42

13

•J - :i h:i

83

35

•J 1 Ai

10

17

2 1 1 H(J

70

10

s //

2 12 18

HO

12

Uatin-T350: 114

2 12 27

40

11

3 1 : 10

2 12 37

13

3

—

2 3 i:.

21

2 a 17

17

2 4-24

15

2 12 H

35

-_

2 -12-11

13

—

2 12 2r,

11

—

—

2. The Watsons 2 3 fir.

34

11

2 3 72

1

2

2 4 13

6

—

4

2 12—31

—

23

--

9

2-12—49

-

.5

—

3

Totals

r,9

—

2**

Cjilcnlatotl

73-5

—

24 5

3. The HoUteins 2— 3—80

44

4

2— 3— 81a

-^

34

1

2— 4-53

—

42

6

2— 4-62

—

—

50

3

2— 3— Slh

"_

ii;

.^

2- 3-103

._

—

28

—

2- 4- 42

—

—

12

—

2- 4- r,5

_

39

2— 4- 75

—

17

2—10- 1

--

—

13

.2_10— 27

—

—

25

—

2—11 13

.

20

2—11— 21

—

—

0

_

2—1 1_ 42

—

—

47

—

TA BLK

XIl.

Th^ F, rexnlts <,,

fth

' vroas h

.Vr/ S

of iff hy

Jiro^rtt t'tjf.

Patiiily ?

^lid

Watson

Holstein

Small eye

■> G

73

3

34

3

5— 1

70

3

22

3

5— 3

13

2

7

1

5— 5

43

3

7

0

5— 6

12

3

1

2

5— a

LOO

5

22

0

5 — H

23

3

8

1

5—11

30

4

11

1

5^12

20

1

4

(I

Tntftiw ;

<90

27

iir,

23

390 •<

P

25-

II

130-2

8-7

' Expectation f»n 45 : 3 : 15 : 1 basis.

128

Inheritance in the Coivpea

TABLE XTII.

The F,^ results of Cross 5. Red Solid by Brown eye.
/S = Solid, JF= Watson, ir = Holstein, £ = Small eye.

Family S W

The Solids 5—1—50 10 7

5— 6— 17 10 1

5—11— 48 12 3

5— G— 1 12 1

5_(j_ 23 21 2

h—G— 26 17 1

^—G— 55 21 1

5— G— 5
5_G— 56

51
14

5_ 8— 55

17

2

_

_

5— 8—106

16

7

5_G_ 21

42

31

—

—

5— G — 54

32

22

—

—

5_ i_ 37

17

5

5- 1— 6&

15

—

4

—

5_ 8—120

20

—

2

—

5—1 1- 2

8

10

5—11— 46

15

—

■ 4

—

5_(j_ 2

17

3

s u

5-G— 7

21

11

Katio 179 : 71

5— G— 10

29

—

13

2-5 : 1-0

5_G— 13

31

—

16

—

5— G— 14

6

—

3

—

2. The Watsons. No Watsons grown from this cross.

3. The Holsteins. (1) Saddle Holsteins.

Family

W

5_ i_ 10

—

5— 1— 17

—

5— 1— 72

5— 1— 79

5— 2— 6

—

5_ 3_ 16

—

5_ 5_ 15

—

5_ 5_ 16

—

5_ 5— 24

5— 8— 10

1

5— 8— 23

—

5_ 8— 31

1

5— 8— 35

5— 8— 67

—

5— 8— 77

5_. 8— 85

—

5— 8-113

5— 8—115

5— 8—135

—

iJ3

34
11

5
21
59
30
41
24

7
35
57
65
66
41
78
38
12
44
40

^ The two forms bred true (15 and 13), in F^.

' The" two Watsons bred true in F^ (16 and 6 respectively).

3 All Saddle. . .

S. C\ Harlanu

129

TABLK Xni-ron/ifMiW.

t>

U G

A—

» - H

f.-ll - iii

5—11 !MJ

5-12 11

5-(; 12

5-0 - 111

&_o _ io

6—0

f- u

•i (

25

6-(.

•il

5- (

41

5— G

45

:> (i

- 50

.v-<.

r,.H

5 r.

58

5 f.

- 77

r, (,

8(>

r, -(.

MS

5 (.

100

98

'i'2
17

IM
20
11
IV

G
47

r>4

44

82
2G
16
20

2a

0
70

(2) Tho HolHt4>inR of typo Lai^r «\v(' with spots. (Typo 2.|

Family
5 1 y
5— 1— 19
-,_ i_ 28
5— 1— 55
5— 5— 1

.-,„ ;>_ 12

5— 5— 20
5_ -,— 21
o — .> — i't

r> .-»— r,i

.-,_ 8— 4
.-,— 8— 18
-,— 8— 23
5— 8— 59
5— 8— 79
.-,_ 8— 81
r>— 8 - 94
r,_ 8-101
5^ 8—109
5_ S-117
;-,_ 9_ 3
"> -11— 50

r, a — k;
5 - r; — 22
.-, _ a __ 27
5_r; HI
r>—(i— 49
5-r; 50
o- a -,1

5 — ^»' - 57
5_r; -- 98
5 r; 104
5 f,- - 1

// (2)

//(.-.)

H

11

7

1

12

12

2

9

5

4

:^

5

0

32

31

1

U

7

0

35

17

18

21

2t

1

14

20

I

48

31

H

13

17

4

IG

19

2

14

17

2

IG

IM

1

25

33

7

45

37

5

17

2'.>

1

22

23

4

14

19

2

30

41

5

10

10

10

42

45

9

32

i;

12

10

8

1)

1

14

9

5

10

12

2

2G

13

4

H

r.

1

2

2

1

«♦

15

1

2

4

_

1 1

IG

1

All Sii.ldlr

130

Inheritance in the Cowpea

TABLE

XIV.

meration of the cross Black eye

{E) by Roun

FamUy

B

h

Ratio

1— 1

36

4

1— 2

26

8

1— 3

9

7

1— 4

29

8

1— 5

45

15

1— 6

25

7

1— 7

117

33

1— 8

39

8

1— 9

83

32

1—10

21

7 ■

\—G

53

19

—

Totals ..

483

148

3-3: 10

TABLE

XV.

Tlie F ^results of the cross Black eye (B) by Rounceval (b).

(1) Families which segregate into B and h.

Family

B

b

1— 5— 4

28

8

1— 7- 2

14

1

1— 7- 6

12

4

1— 7— 9

12

4

1— 7—17

69

28

1— 7—18

12

4

1— 7—19

10

7

1_ 7_2l

17

5

1_ 7_23

20

5

1— 7_25

12

2

1— 7—27

18

3

1_ 7_30

5

3

1— 7—35

14

4

1— 7—40

15

7

1— 7—45

69

15

1— 8— 3

31

9

1_ 8—14

23

12

1— 8—17

22

6

1_ 9_i3

13

3

1_ 9__i4

7 ■■

2

1— 9—15

17

1

1— 9-19

7

3

1 - 9—25

72

23

1_ 9_27

16

6

1— 9—28

17

6

1_10_ 1

12

10

1—10— 2

50

11

1—10— 3

18

1

1_10_ 4

24

13

1—10— 5

58

19

1—10— 6

12

3

1—10— 7

26

8

1_10_ 8

12

2

1—10—13

36

10

1_10_14

7 ■

4

1_10_17

22

6

1_10_18

13

2

1_10_20

8

3

l_10_27

36

15

I— a _53

29

8

Totals... \..

915

281

Calculated

897

299

Batio

3-2

1-0

j (2) Families which breed true to li.

Family B b

- 7—12 15 -

- 7—29 10 -

- 7—46 41 -

- 7—55 41 -
-8—2 15 -

9— 8

— 9— 9

— 9—11

— 9—23

— 9—26

— 9—96
—10— 2
—10—19
—10—23
—10—25
—10—26
— G— 24

20

16

10

13

6

84

41

8

9

32

49

81

Total

497

(3) Families which breed true to h.

Family

— 5—3

— 7—7

— 7—11

— 7—22

— 7—39

— 7—47

— 7—73

— 8—1

— 8—10

— 9—1

— 9—3

— 9—12

— 9—17

— 9—20

— 9—21

— 9—22

— 9—24

— 9—32
—10—12
—10—16
—10—21
— G— 6
— G— 23
— G— 65

Total

8

115

39

4
80
11
96
39
41
31
28
17
16

5

13
14
21
42
31
17
86
90
41
27

907

8. C. Hakland

i:n

TABLE XVI.
Thtt t\ results of Cn}9» 5. /irown ryr {lirown) hy Hed (Ited).

Paniilj

Itrown

Murtxm

K«d

ft— <;

7ft

80

U

ft— 1

ftH

1ft

ft

ft— a

12

6

8

o - 8

17

6

3- 4

—

ft— ft

40

1ft

3

ft - 6

11

1

3

ft— 8

9ft

25

10

ft- 9

26

ft

3

ft- 11

40

10

2

5-12

17

6

1

taU

391

122

39

441-6'

110'4

36-8

Expectation on 12 : 3 : 1 basiH.

TABLE XVIL

Tlf F^ results (ifCro.ss 5. Jirotcn eye (Jh'ofvn) by Ibid {W;d).

1. The Browns,

Family

Brown

Maroon

5—1— 28

18

5— 3— 16

30

—

5— 5— 1

64

5—5- 20

70

5— 5— 21

46

—

5_ 5— 56

14

5_ 8— 23

57

—

5— 8- 50

82

5— 8— 51

82

5— 8— 59

35

. —

5— 8— 67

41

5_ 8— 79

65

5— 8- 85

35

5— 8—101

49

_

5— 8-115

44

—

5—11— 36

17

5_r;_ 16

38

o-G — 45

44

—

5— G— 5

22^

6

5_G — 31

25

3

5— r; — 49

18

6

5—G— 51

12

3

5— a — 58

13

3

5—G— 86

19

4

5— 1— 19

19

7

5_. i_ 79

16

5

5— 2— 6

42

17

5— 5— 15

30

11

5— 8— 18

28

9

5— 8— 3.5

51

15

5— 8—109

25

11

5— 8—117

13

1

Red

1. The Browns.

Family

Brown

Maroon

Ked

5—11— 48

15

6

—

5— 5— 36

37

_.

15

5— 8— 94

38

—

9

5— 8-135

28

—

12

5— 9— 3

24

—

6

5—G— 19

31

10

5_G — 77

14

—

6

5—G — 13

22

3

3

5—G — 23

17

1

7

5 -a _ 25

4

1

1

5—G — 31

35

8

5

5—G — 41

34

12

8

5_G - 54

23

0

5

5_G - 98

15

4

6

5— 1— 9

15

2

2

5-- 1— 55

4

3

1

5— 8— 31

47

12

6

5— 8— 77

60

14

4

5_ 8— 81

63

13

5

5— 8—111

22

7

4

5— 9— 6

69

u;

8

5— 9— 8

43

12

4

0-11— 26

18

3

1

2. The Maroons.

5— 5- 12

13

4

5_ 8— 23

23

10

5—11— 2

12

6

5—11— 50

--_

56

38

Jonm. of Gen. \ui

132

Inheritance in the Cowpea

TABLE XVII— cow^iwwed

2. The Maroons.

Family Brown Maroon Red

5— G— 12
5_G— 21
5— G — 22
5_(j_ 26
5_G— 53
5— G —100

17
33
14
20
23
49

3
12
8
5
3
21

5_ 5— 39

_

4

5_ 5— G

14

5_ 5_ 45

35

5— 8—113

12

h—G— 2

20

5— G— 7

—

21

—

3. The Beds.

Family

5— 1— 17
5_ i._ 37
5— 1_ 50
5_ i_ 66
5— 5— 16
5— 6— 17
5_ 8— 10
5_ 8— 55
5_ 8—106
5_ 8—120
5—11— 46
5_12_ 11
5— G— 10

Brown Maroon

11

Red

11
16
24
19
24
15
35
19
23
22
19
48
422

1 Probably due to vicinism.
- And 1 black.

ON TIIK UKLATION liF/rWKKN NUMHKIl OF
(MIKOMOSOMKS AND NUMBKIl OK TYPKS,
IX LATIIYIWS ESPKOIAI.LY.

Hy (). WIX(iK.
(With Plato V.)

The iiuiiiIkt of simultaiu'ously and indepondently segregating pairs
of factoi-s' in the species investigated by genetic experts has, as we know,
never yet been found so high Jis to exceed the haploid chroniosonie rniniber
of the s|X'cies. Conseipiently, there is still nothing to subvert the theory
that the genes have their morphological equivalent in the chromosomes
and that these latter are — or can be — individually dissimilar in a given
biotyjx; as regards the genes included.

It is a (juestion of very gi'eat theoretic importance, whether the
simultiineously segregating pairs of factors, not mutually connected, can
ever exceed the number of chromosomes in the species concerned, this
being, so to speak, a decisive point as regards the value of the entire
section of the stud\' of chromosomes related to the science of genetics.
If a biotype could be found to exhibit segregation of but a single paii-
of fact'Ors in excess of the number indicated by the haploid chiomosome
value, then, properly speaking, the theory as to the value of chromosomes
as bearers of the genetic, segregating units would collapse at once. The
nice agi'eement between reduction division and segregation would thus
be irrevocably destroyed.

At a first glance, it might seem likely that we should be able, in
highly varying species, to segregate without great difficulty a gieater
number of ty[>es than the chromosome number found for the species
permits. We can, however, obtain a surprisingly large number of com-
binations even from a (juite small number of chromosomes, and it must
also be Ixjrne in mind that the theory of agreement between leduction
division and Mendelian segregation would by no means be destroyed
even if we did succeed in finding a greater number of indei)endently
mendling piiii>> of genes (or pairs of gene-complexes) within a Linnaean

* By this expression is meant pairs of factors, or groups of pairs, between which the
phenomenon of linkjifje is not found.

184 Number of Chromosomes in Latliyriis

species, than corresponds to the haploid chromosome number. Even a
species with but a single chromosome in the haplophase, and two in the
diplophase, might be allowed to contain an unlimited number of inde-
pendently segregating genes, as the two chromosomes in the diplophase
might very well be genotypically different between individuals within
the species. The point is, that according to the theory, there must not
be more independently mendlihg pairs of factors in a given biotype
(individual or clone) than the chromosome number of the biotype indi-
cates. As soon as two individuals (not to speak of more) not belonging
to the same biotype are introduced into the experiment, the possibility
of new combinations is considerably increased, unless the two original
individuals are homozygotic.

An organism with only one chromosome in the haplophase (^=1)
will naturally only be able to have two different chromosomes at the
outside in the diplophase. Let us call these A and a. Two different
types of gamete can then arise, viz. A and a, and these can form three
different types of zygote: A A, Aa, and aa, of which two are homo-
zygotic.

With ^= 2, a given biotype can be doubly heterozygotic ; i.e. AaBb,
and four kinds of gametes can be formed, viz. AB, Ah, dB and ab, of
which in F^ it will be possible to obtain nine diploid combinations,
AABB, AABb, AAbb, AaBB, AaBb, Aabb, aaBB, aaBb, and aabb, of
w^hich four will be homozygotic in both characters.

Where a; = 3, 8 different gametes can be formed, giving 27 different
diploid combinations, of which 8 are homozygotic.

In a word : With a given haploid chromosome number, x, we can by
self-fertilisation and segregation of a single individual obtain, theoretically
speaking, at the outside 2^ different gamete types and 3* different diploid
biotypes, of which 2* will be homozygotic in all characters.

If we are to entertain any hope of controverting the theory of identity
between reduction division and segregation, then naturally it will be
necessary to work with organisms having a low chromosome number,
and capable of self-fertilisation. A plant with only eight chromosomes
(a? = 8) will on self-fertilisation be capable of forming 256 different
gametes, and will in F^ segregate 6561 different types.

If the segregation experiments be commenced with more than one
biotype, which of course will as a rule be necessary when working with
species not capable of self-fertilisation, the question becomes more
complicated.

Two individuals with x=l can in the diplophase differ in both

(). WlN(iK 1 .*];■)

chn>inam>inos, tho ono (liplnbimit having i\\r fnnimla A A,, the nthcr
CI (I,. Oil loriuation of tho ^aiiu»tt*s, we can thou i»])tjuii four ditVrn'Mt
typi»s, A — ii, — tt — <i, ; two fnun each. In /',, we have 2x2 = 4- ^cno-
typically tliHt'ivnt <iiploi<i combinations, viz. Aa Aa^ /I, rt -/!//,, and
those will, on nHiuction (livisi<»n, aj^ain throw af^ thr four gametes
A — Ax — (I and (/,. In F,^ an<l the foliowini; oftspring gencrationH, we
Clin by five combination of these find 10 i^enotypically ditic'rent coMd»i-
niitions, viz. A A — -1 A , — Aa — Aa^ — A ^A , — A ,a — A ,rr, — (ut — rm, — fr,r/, ,
of which four are homozygotic ; one for each gamete type.

On considering in the same way an organism with two chromosomes
in the haplophjuse (.c = 2), and presuming the segregation ex|K'riment
to commence with the crossing of two individuals having different
genotypic value for all the chromosomes of the diplopha.se ((^ight in all
in the two individuals) then each individual will be able to form four
gametes, or eight diti'erent gametes from both. On crossing these forms,
we obuin U) ditferent biotypes in I\: and in F.,, where the gametes (16
different) can combine altogether freely, there can arise 100 genotypically
distinct diploid types, of which 16 will be homozygotic in all charactei-s,
i.e. one for each gamete type formed by F^ .

Where .r = 3, we can by crossing two different individuals easily
obtain, theoretically, eight gametes from each, i.e. 16 diffc'rent in all.
In Fi, 64 types will have arisen, able to form in all an ecpial number of
gametes, and in Fo, 1000, of which 64 are homozygotic.

Brieffy then, if a segregation experiment he commenced by crossing
two individuals of the species to be investigated, then tve can in F^ obtain
4* and in F^ and the folloiuing generations 10^ genotypically distinct
fornus, of which 4^ will be homozygotic in all characters, whei^e x indicates
the haploid cJiromosome number of tJie species.

If, for instiince, we commence by crossing two entirely diff*erent
individuals, heterozygotic throughout, of a species with eight chromo.somes
in the haploid phase, then in F^, there can arise 65586, and in F., a
milliard different types.

On extending the analysis so as to include a further number of
individuals, as to whose genotypic constitution nothing is previously
known, then the number of possible combinations will of course be far
higher even than this, as also when "crossing over" takes place. In this
last case we cannot reckon the possibilities beforehand.

The advantage of working with self-fertilising organisms is thus
entirely evident. It would be even better if we could make our genetic
experiments with organisms where the haplophase was a richly developed

136 Number of Chromosomes in Lathyrus

and independently living individual, as for instance in the liverworts.
This I have already pointed out in a previous work : '* The Chromosomes.
Their numbers and general importance" (Comptes rendus des travaux
du Lahoratoire de CoHsberg, Vol. xiii, Copenhagen, 1917). A species
such as Marchantia polymorpha, with eight chromosomes, would thus
only form, at the outside, 256 types on crossing between two individuals;
and by taking an individual of a monoecious species, self-fertilised,
then eo ipso we should obtain but one type — a pure line — if we may use
this term also for cultures of haplobionts.

Lathyrus odoratus is one of the plants with which we may hope
sooner or later to elucidate the important question as to agreement of
reduction division with Mendelian segregation, as students of genetics
have for a long time back been occupied with the genotypic features of
this species.

Acting upon a suggestion from Prof R. C. Punnett, of Cambridge,
concerning the number of chromosomes in Lathyrus odoratus, I proceeded,
in the summer of 1918, to fix and examine material of the species in
question. In a written communication to me, Prof Punnett stated that
Mr R. P. Gregory had informed him that there were doubtless seven or
eight chromosomes in the haplophase. Otherwise, as far as I am aware,
nothing is stated in extant literature as to this ; no mention is made as
to any species of Lathyrus either by Tischler in his excellent " Chromo-
somenzahl, -Form und -Individualitat im Pflanzenreiche " {Progr. Rei
Bot., Vol. V, 1915, p. 164) or by Ishikawa in his likewise very compre-
hensive work, "A list of the number of chromosomes" (The Bot. Magazine,
Tokyo, Vol. XXX, 1916, p. 404).

On two occasions, in the early spring and early summer of 1918,
Prof Punnett sent me, from his cultures at Cambridge, freshly fixed
material of young flowers of L. odoratus. On investigation, however,
the former consignment was found to be at a too advanced stage, as the
tetrad division had throughout been completed. The second batch of
material, again, was too young, no stage beyond synapsis being discernible.
In order not to delay the investigation further by correspondence under
the present abnormal conditions prevailing in the postal service, I
therefore fixed material myself, obtaining it from a garden at Marselis-
borg, near Aarhuus (Jutland) in the month of July. The material was
fixed in Carney's liquid, and stained with Delafield's haematoxylin.
Besides the above mentioned species, I also fixed, in September, by the
same means, some young flowers of Lathyrus latifolius, cultivated in the
experimental nursery of the Carlsberg Laboratory, Copenhagen*

(). WiMJK i:j7

A cyU)lo^ic«l iiivi'Mi-ijfiition, carried nut fsm-ntially with a \u\\ U)
aaccrtaining the chnmiosomc munlHT. j^avi* thr followiu^r rcMult.

hifln/rus odttratuji L.

The rhn»iiu»8«>inett art* largr aixi soim-whal. <'l<n>^Mt<'<l in lh»- hrtin*-
typic uietJiphiuH*. Thry '^re pifsviit. tn thr iiumlu'r nl' 7 (=./) (Figs. 1
ami 2). Thort' is no ditVrrcncT, <»n the whoir, in the si/c of thr dinVnnt
chn)nu»Si»nu»s. Splitting and transition to the ;m.i|)h}i.m' prorrcd fairly
regularly. Onv of the chromosome |>airs am, however, at limes l»e
sepanite<i slightly earlier than the remainder (Fig. .S). The heterotypic
tolophjise may often give an impression that the chromosome number
isgToati»r than seven, owing t<> the fact that the chromosomes at this stage
are bent to an angle, and also exhibit splitting in the plane through
which the subsequent homoeotypic division takes place. The chromo-
somes being bent to an angle (the angle pointing toward the pole of the
nuclear spindle) will in a certain position each appear {is two separate
chromosomes (schematically shown in Fig. 4) and as the chromosome is
further split in the plane of the angle, we find, extremely often, apjmrent
groups of four chromosomes, to the number of seven, i.e. 28 altogether
(Fig. 5). Fig. 6 shows the chromosomes in the metaphase of the homoeo-
typic division; here also we Ciin with great certainty count seven chromo-
somes in eiich of the two nuclear plates of the spore mother cell. In
the anaphase of the homoeotypic division also, we may find what would
seem like more than seven chromosomes, the chromosomes here being
likewise bent to an angle, though not split.

An investigation of the chromosomes in somatic cells showed entire
agreement with the figure found in the haphjphase, i.e. 14 (Fig. 7). As
is usually the case, the chromosomes here were of slenderer form than
during reduction division.

Lathyrus latifolius L.

This j>erennial species appeared in every respect as L. odoratu.s, and
the cytological picture for the two species is so uniform that prepara-
tions of the one might be taken for those of the other. Not only is the
chr«»mosome number likewise seven, <us the heterotyi)ic metaphase in
Figs. 8-9 shows, but the chromosomes themselves are also entirely alike
in size and shape. Here, as in L. odoratua, twice the number could be
counted with pertect certjiinty in somatic cells (Fig. 10).

These two species, then — the oidy species of Ldthijrns hitherto
investigated — present an instiince of the ordered icgularity with which
in larger or smaller systematic groups in the animal and veget<ible

138 Number of Chromosomes in Lathyrus

kingdoms, the chromosome members of closely related forms are found
to be themselves related ; i.e. either entirely alike, or multiples of a
cardinal number characteristic of the group — as I have explained in my
work above quoted.

The chromosome number seven is a comparatively low value to find
among phanerogams, and should, a priojH, count in favour of the employ-
ment of Lathyrus in genetic experiments. The theory as to agreement
between reduction division and Mendelian segregation permits — com-
mencing either with self-fertilisation of one heterozygotic individual or
by crossing of two pure lines — that there be found at the outside seven
independently segregating pairs of factors (or pairs of factor groups). A
segregation experiment commenced in one or other of these two ways
should not lead to the production of more than 128 different biotypes,
all homozygotic throughout.

As to how far it may prove possible to find more than seven inde-
pendently segregating pairs of factors or pairs of factor groups in a
single biotype of the species, is a question which future investigations
must decide. Personally I feel confident that they will not be found.

I have observed nothing in my material which might lead to the
interpretation that chiasmatypy, in the sense of Janssens, is met with in
Lathyrus. In my opinion, even in organisms in which breeding experi-
ments have resulted in the view that parts of the chromosomes are
exchanged during the reduction division, this process is not usually
going on at so late a stage as shown by Janssens. More probably it
occurs in synapsis where it cannot readily be demonstrated by micro-
scopical methods.

Carlsberg Laboratorium, Copenhagen,
15 October 1918.

EXPLANATION OF PLATE V.

All figures are drawn with aid of Abbe's camera lucida, using Zeiss' homog. immers.
2 mm. and comp. oc. 18.
Figs. 1 — 7, Lathyrus odoratus. Figs. 8 — 10, L. latifolius.
Figs. 1 — 2. Pollen mother cell. Hetero typical metaphase (polar view).
Fig. 3. Pollen mother cell. Heterotypical meta-anaphase (side view).
Fig. 4. Two chromosomes from heterotypical anaphase (schematically).
Fig. 5, Pollen mother cell, Heterotypical anaphase (polar view). Apparently 7 groups

of 4 chromosomes present.
Fig. 6. Pollen mother cell. Homotypical metaphase.
Fig. 7. Somatic cell with 14 chromosomes.

Fig. 8. Pollen mother cell. Heterotypical metaphase (polar view).
Fig. 9. Pollen mother cell. Heterotypical metaphase (side view).
Fig. 10. Somatic cell with 14 chromosomes.

JOURNAL OF GENETICS, VOL VIM. NO. 2

X

/■

*/
iv

*^'

\

1 \

■r

0

/ / ■

\

^1

/

X

PLATE V

f^:^

'r' -iJ

/

•v

/

VoMMi: VIII JUNE, 101 !» No. 3

SEX SEGREGATION IN THE HRYOPHYTA.

By E. J. COLLINS. B.A. ( amb., B.Sc. L«.iul.
Botanist to the John Innes Horticultural Institution.

(With Pluto VI and Five Text-figures.)

Whilst the majority of plants show an alternation of generatiorm
in the complete life cycle, it is in the Bryophyta alone that the gamcto-
phytic {n) phjvse shows an independent growth combined with a high
degree of morj)hological differentiation. The study of sex in this group
does not involve, as it does in the highest forms of plant life, the con-
sideration of sporophy tic tissues differentiated as "sex organs" in respcjnse
to the parasitic nature of the gametophyte generation. It may well be
that very considerable light would be thrown on the nature of sex and
much information gained jis to its mode of inheritance by further studies
among the lower plant groups.

In recent years no more interesting and suggestive pai)ei-s concerning
sex in Mosses have been published than those of El. and Em. MarchaP,
and a very brief outline of these papers will serve to show the scoj>e of
their researches.

By compounding a suitable culture fluid, these investigators have
grown what might be regarded Jis pure line cultures of the various
mosses experimented with. Starting with the spores from capsules of
dioicous forms, these were shown to be uni-sexual, inasmuch as the
protonemata developed from them produced leafy axes which were all
of one sex, either male or female. The male and female axes so pro-
duced were approximately equal in number, and the protonemata ob-
tained vegetatively from them gave leafy axes of the s«iine sex as the
parent axes. Protonemata were then induced to develop aix)sixjr()usly
from sporophytic (2n) tissue, and these produced a prop<jrtion of herma-
phrodite axes. The projx)rtion of hermaphrodite axes however wjvs very
low; e.g. in Bryum caespiticium, of 1738 axes 1579 were* male, 154

* J/«fifi. de Vacad. roy. de lielgique, HK)6 ; liitU. de Vacad. roij. de liehjique, Clause dc»
Sciences, 1907, 1909, 1911.

Joam. of Gen. viii 10

140

Sex Segregatio7i in the Bryophyta

hermaphrodite and 5 female, a result which shows a striking preponder-
ance of male axes. No sporogonia were borne upon any of the aposporous
products as the sex organs were sterile. Abnormal organs of mixed sex
also were found in B. caespiticium and Mnium ho7'num. The investigators
concluded that there was cause to think that the uni-sexuality of dioicous
mosses in the haploid phase was absolute, and due to the presence — to
the exclusion of the other — of only one sex determinant, and that the

2n Sporogonium ^ (Sporoph/te) ^

(Antherldium d Antherozoid) n i

y spore

"•■ Protonema (/») ?

Diagrammatic Life Cycle of Dioicous Type.

^2n Sporogonium $ (Sporophyte)

i:'spores{n)

'^ Protonema (/?) ^

Diagrammatic Life Cycle of Monoicous Type.

maturation process was the occasion of the segregation to the state of
purity of sexual characters in the spores. In other words a clean cut
sex segregation occurred at the reduction division of sporogenesis. Un-
doubtedly the low proportion of hermaphrodite axes resulting from
aposporously developed protonemata • is a serious hindrance to the ac-
ceptance of the theory, for on the hypothesis we should have expected
that plants arising aposporously from sporogonial tissues would all be

E. J. Collins 141

heriimphnidito. Tho ditticulty is recogniHeti by Kl. and Km. Miiix:hiil,
but in alloviiition of this objection th» y mh^'^^'mI that jiinon^ thi- male
plants may have b<»en many in which the develnpmrnt of J or^anM may
have lH*en suppre88e<l. Th»' a|H)s|H>rous pHwhirt^ of th** dioirouM forms
B, caespiticium, li. avgenteum, H. rapillarc, Ji. fall(Lr and M. htnumn
were without exception sterile. A|)oH]M)rou.s |)r<Kiiiets of moiioicotis Innns
were fertile, pnxlucing diploid gametes. 'I'hcHc united in frrtiliHiition,
and by further a|HW|)orou8 development from the s|)oroj(onium tetraploid
gametophyt<»s wei*e forme<l.

In supjx>rt t>f the theory of sex segn'gation at HjX)rogeneHiH thr
reconled sex behavi»>ur of Sphaerocarpm terrestins is often (pioted. Ah
is well known, the s|H)res of this HejMitic are shed in the orif^inal tetrad
groups and aeconling to V. Douin, two ^ and two J plants result from
the development of the spores of each tetrad. More precisely, of 81 s{x>re
groups examined 64 showed two J" and two J plants, 13 germinated
inn>erfectly and 4 showed results not in agreement with ex|Xictation.

Recently Allen' hjus rejwrted a chromosome ditierence correlated with
sex diffei-ences in Spha^rocarptis Donelli, and his limited observations
indicate' that of the spores of each tetrad, two give J* and two give
J plants as in X terrestins. At the reduction division two of the four
spores each receive a large chromosome and these develop female plants,
whilst the other two receive a small chromosome each and give rise to
male plants.

It will be seen that the theory emphasises the dual nature of the
sporophyte generation, in that it carries both sex determinants. The
determinants however biing about no expression of sex in the sporo-
gonium, but assuming that dioicous sex plants were foreshadowed by
two types of spores, a differentiation of sporophytic tissues in conformity
with the change might ensue. Thus changes would be initiated which
Would give the s{X)rophyte the appearance of "sex monacism," and
inhibition of the differentiation of one or other of the " sex organs '
would letwl to the dioicous spor<>phyte. Should the relative importance
of the two generations in the life cycle become reversed, the parasitism
of the -sexed plants would bring about a still greater differentiation ot
ti.ssues of the sporophyte in accordance with the physiological need.
Needless to .siiy the changes would lead to the impression of "six
charactere " ujx)n the sporophyte.

Scheme I is suggested as an expression of the theory for dioicous
fonns, whilst Scheme II would apply to monoicous types. In the latter
» Science, N.S. Vol. xlvi. 1917. p. 4G0.

10—2

142

Sex Segregation in the Bryophyta

scheme an alternative reading is given {A and B). If the purity of the
gamete for sex characters be maintained, a sex segregation in the somatic
tissues of the gametophyte is assumed to occur (B). This point will be
considered a little later.

Quite naturally the question arises, how far can the facts of sex in-
heritance gained from the study of animals, be applied to these plant
forms, if indeed they can be applied at all ? At the outset it must be
borne in mind that in general, meiosis accompanies gametogenesis in
animals, and sporogenesis in plants. Exceptions occur in plants, as for
example in the Fuci, where there is no alternation of generations and
reduction occurs at gametogenesis.

Assuming that in a monoicous (^) form of moss, the gametophyte
produced J^ gametes bearing the male factor, and $ gametes bearing

SCHEME I.

Sporophyie

Apcsporoas Itiie.

<r

( ( (

@

1579. /S4. 5
Oierile axes.

0., g)< Spores

(

)

Leafi^

> Gajrteiropkyte.

(ny (J)^ Gameles j

__3poropAyi€

Dioicous ij/pe.

K. J. (V)LLINS

143

sen KM K II

Sporopkyte.

>J5<JCvI

^ A ® ® B ®<^''-'" t

^CdmeU^A^e.

S€x-s*^r*^aftJj

SborcpAyi^.

^n ^ameiophyie

MonoLcous ti/pe.

either the cT <>t' % fjvctor, then two types of sporogonia would be pro-
duced, ((/) those bearing exclusively male elements whose spores {(J)
would result in J" gametophytes only and (6) those carrying both male
and female facUjrs whose spores (^) would give rise to monoicous (^)
gametophytes. Such a possibility is not excluded \ although it is
generally assumed, without definite knowledge to the contrary, that the

* In liryum mamillatum the monoicous condition is accompanied by distinct male
plants. It is not possible to say whether such plants have been derived from spores pro-
duced by sporogonia exclusively male in character, or whether of the spores in any one
capsale some bear j and others A factors.

144 ' Sex Segregation in the Bnjophyta

spores of any moss capsule of a monoicous species produce monoicous
gametophytes. The further assumption, that maleness is recessive,
would mean the extinction of the monoicous habit.

Again in dioicous species it is generally agreed that the spores of
any single capsule develop to give rise to either J^ or ? axes, but the
theory would lead us to assume, as with monoicous forms, the existence
of certain capsules, all the spores of which produce male axes exclusively,
and others whose spores produce monoicous axes in which maleness is
latent.

Although the experiments upon which El. and Em. Marchal based
the theory of sex segregation have not been repeated, the theory has
been subjected to considerable criticism. Recently M. Wilson^ found
organs of mixed sex together with normal sex organs upon naturally
growing plants of Mnium hornum. A cytological examination was made
of this material and a normal antheridium which happened to be in
spermatogenesis showed the haploid number of chromosomes. From
this it was evident that meiosis had occurred, and it was not probable
that the plant had been produced aposporously as the theory would
demand. Also it was suggested that a low proportion of hermaphrodite
axes, such as was produced in aposporous cultures, might very well be
found in nature following a normal process of sporogenesis.

My own experience relates to the well-known form Funaria hygro-
metrica which has been described by many competent observers as
monoicous, whilst others, equally competent, have described it as a
dioicous species. Under these circumstances Boodle ^ some years ago
undertook an inquiry into the question of the distribution of the sex
organs, and after examining material from many localities came to the
conclusion that the plant was monoicous, and dioicism, if it occurred, was
comparatively rare. It was found that the shoot which bore archegonia
terminally, arose as a lateral branch of the axis bearing the terminal
male "flower," and no instance was observed in which a female axis
produced a male branch. Boodle concluded that the dioicous habit was
attributed to the plant because young plants bearing the discoid male
" flowers " only were found, and that if female branches were gathered
they were generally torn from the male axis to which they were attached,
the presence of a basal tuft of rhizoids on the female branches giving
them the appearance of distinct plants.

During the course of some experiments conducted for the purpose
of determining the best cultural conditions for the production of proto-

1 Annals of Botany, 1915. ^ Annals of Botany, 1906.

K. .1. (\)LUNS 145

nenmt-ii fnun various innss ori^jms and spoivs, ihrt'o cultiin's of Fuiutntt
hi/(;n)metnc*i wtTr nmdf in Marchals' nutrii-nt li(|ui(i. Thrsr rulturcH
wrre ohtainrd (ci) tVoni antlu'ridia takm from ]i Hiiigh* male " How(;r,"
(6) fnun |)*Ti^onial loavos, ami (r) from the M|K>n'H shi'd from one ri|M'n«(l
ca|)suK\ P2ach culturt' whm sufticiontly ^rowii was |H)un'(| out u|)<ni
soil ct>ntainiMl in a small carthciiwari' pot, owr which hoilin^' wat«'r had
boiMi pi>ure<i smnv littK' time previously. The jhiIs were then eov<'re(i
and 8t<KHl in a eold frame. When the |)rot()nemal felt was observed
gniwin^ strongly, the coverings were removed and no further attenti«»n
was given to the pots. Later however it was oh.served that whilst the
sward of plants obt^iiiu'd from the culture from spores had produced a
dense crop of sporogonia, a close swani of plants with large discoid luale
"flowers" had appeared in the two cultures resulting from antheridia
and jvrigonial leaves. A second crop of discoid male " floweix " was
pnxluced by these same cultures but no s|K)rogonia were produced at
any time'. Photographs of two of these cultures, A from sj)ores and
li from antheridia, are repnxluced on Plate VI.

It apjx'ars i)ossible that vegetative development from structures
borne on male and female axes respectively may, if a sex segi'egation
has actually occurred somatically, lead to the production of distinct
male and female plants. If such is the case the jmrity of the (jamete in
monoicoits forms is secured by a somatic segregation in haploid tissue.
It is interesting to recall in this connection that El. and Em. Marchal
have stated for FunaHa, that J" axes and % axes may arise from the
same protonema and that this has led to a confusion concerning
monoicous and dioicous forms. Thus the point at which segregation
occurs is not necessarily fixed, but may be shifted backward in the life
cycle until it occurs at sporogenesis. Thus the dioicous habit of the
gametophyte may, as we can imagine, have been established. In this
way the dioicous condition might co-exist with the monoicous (autoicous),
a state of affairs known U) occur in some mosses, e.g. Dicranella crispn,
or the autoicous condition might be accompanied by distinct male plants
e.g. Bryum mamillatum.

Other forms might be mentioned which show varying sex conditions,
whilst vegetative propagation from sex- segregated axes would also learl
to the prcKluction of the various sex-forms of <jne species.

The generalized scheme figured on p. 146 deals with sex segregation
and vegetative reproduction in both monoicous and dioicous types.

^ Because of this interesting result, perichaetial leaves associated '.vith the archegonia
in Funaria are now being made the subject of a similar experiment.

146

Sex Segregation in the Bryopliyta

GENERALIZED SCHEME.

@e Sporopkyh

Zeafydxas

<

® ® ® ^

Ve^el&lively reproduced axes.

Vegclalively reproduced axc5.

* SS. Point at which it is assumed that a segregation of sex occurs.

PLATE VI.

For explanation see p. 145.

JOURNAL OF GENETICS. VOL. VIII NO. 3

PLATE VI

A. riiints <:iown from spores, Monoicon.'j plants boarin^' spoio^'onin.

B. Plants grown from antheridia. Males only,

RACIAL STUDIES IN FISUKS

II. EXPERIMENTAL INVESTIGATIONS WITH LPJlilSTES
RETICULATUS (PETERS) REGAN

By JOHS. SCHMIDT; D.Sr.

Director of the Carlsberg Physiological Laboratory,
Copenhagen, Denmark.

(With OnoGi-aph.)

I. Introduction,

The purpose of the experiments about to be discussed was to
contribute information on the rather obscure question whether, or to
what extent, quantitative racial characters are hereditary.

The tropical -Americiin Cyprinodont Lehistes reticulatus (Peters)
Regan' wjis employed in the experiments. I have previously used this
little aquarium-fish in experimental investigations, namely for the
purpose of demonstrating the importance of environment on the
numbers of organs (dorsi\l rays).

Lebistes reticulatus is, like so many of its relatives, viviparous, and
under favourable conditions the female brings into the world, at intervals
of about 4 weeks, a considerable number of young. The young jx)ssess
at birth the full number of vertebrae, dorsal rays, etc, which is therefore
recognisable immediately after birth.

The ex}:)eriments fall into two groups, of which the first helps to
elucidate the importance of eodernal factors (temperature) upon the
number of dor&il rays. The second is concerned with the <juestion
whether hereditary differences, i.e. differences dependent upon internal
factors, may be proved to exist in different individuals. Before I proceed
to discuss the experiments, I may draw attention tu the fact that the

* C.Tate Began, "A revision of the Cyprinodont Fishes of the Subfamily Poeciliinae,"
Proc. Zool. Soc. London 1913, Vol. ii. pp. 977—1018, 1913.

148 Racial Studies in Fishes

number of dorsal fin rays in Lebistes reticulatus varies from 5 to 8. By
far the most usual number is 7.

A more detailed account appears in Vol. XIV, Nos. 1 and 5, of the
Comptes-Rendus des Travaux du Lahoratoire de Carlsberg, Copenhagen.

II. Importance of External Factors.

The principle was to vary the temperature for the same pair of
parents from one period of pregnancy to the other, and then determine
the number of dorsal rays in the various broods of offspring. In the
beginning I had no means of maintaining a constant temperature in
the aquaria, and was therefore compelled to limit myself to stating that
the animals in the experiments were kept at a " low," " medium " and
" high " temperature, in wHich " medium " temperature was ca. 6° above
" low " and ca. 3° below " high." " Low " temperature was generally
equivalent to ca. 19° Centigrade, varying between. ca. 17° and ca. 23°.

In the experiments 5 different pairs of Lebistes reticulatus were
used. The results of these investigations can be seen from Tables I — V,
each of which shows the number of rays in several broods from the

TABLES I — V. Number of dorsal rays in offspring of the same pairs
of parents at different temperatures.

TABLE I.

c^ 7 X ? 7.

No. of
rays

High

temperature

Bom 12 March

Medium Low
temperature temperature
Born 13 April Born 1 June

Low
temperature
Born 25 July

High
temperature
Bom 25 Sept

8
7
6

9

6

20

25
13

1
33

4

4
16

n
a

P.E. A.
Fl.

15
7-600

±0-532

, ±0-093

±0-465

20
7-000

38
6-658

±0-493
±0-054
±0-270

38
6-921

±0-390
±0-043
±0-215

20
7-200

±0-414
±0-062
±0-310

TABLE II.

c? 8 X 9 8.

No. of rays

Medium temperature Low temperature
Born 12 April Born 29 May

8
7
6

15
29

27
5

»

n
a

<r
P.E.

Fl.

44
7-341

±0-483

A. ±0-049

±0-245

32
6-844

±0-399
±0-048
±0-240

Jons. Schmidt 149

TAHLK III. t^H X 9 7.

Medium t«nt|trrmturr l^iw t4'iit|irr«(urp
Nn. of m.Tii l^im 3.1 Mhy lUirii H .Inly

8 8 —

7 r.» »i

G - (I

n

A?

37

a

7 140

6 SSS

<T

10 -aril

} 0-3W

IM<

:. A.

±0-031

.i OOU

Fl.

±0155

± 0-220

TABLE IV. c^6 X 9G

.MeiUuni

Mf^lium

IX)W

t«in|>enturc

tem|>eraturc

temiHTatun-

No. of niVH

Horn 1 April

Itorii 27 .May

llorn 13.1uly

8

1

1

—

7

11

31

15

6

3

(>

21

n

15

38

36

a

6 867

6-868

6-417

<r

i 0-565

±0 4:M)

±0">05

P.E.

A.

: 0-098

±0048

± 0057

Fl.

±0-490
TABLE V.

±0-240

c^ 6 X 9 G.

±0-285

Medium

Low

Low-

temperature

temperature

temperature

No. of rays

JJorn 7 May

Born 19 .June

Born 14—15 Au^uflt

8
7

11

17

35

6

11

18

5

—

2

—

n

11

30

53

a

7-000

6 500

6-660

<r

±0-636

±0-487

P.E.

A.

±0-078

±0-045

Fl.

—

±0-390

±0-255

same pair of parents. The date of birth of the young is noted in each
ca.se, iis also whether developed at low, medium, or high temperature.

It is distinctly evident from the tables that the different broods do
exhibit a difference in the number of dorsal fin rays, and it is further
seen that the average number of rays was greater where the young had
been develojyed at a high temperature than luhere their development took
place at a low temperature.

In all my later experiments this result has been confirmed. I will
content myself with di.scu.ssing a single one of the later exp<3riment.s,
which, technically speaking, h;ul the advantage over the preceding ones
in that the tempemture in the aquaria could be kept constant, there

150 Racial Studies m Fishes

being a fluctuation of only one-tenth of a degree (± 0*1). The two
parents had respectively 7 and 5 rays in the dorsal fin. The experi-
ment took place partly at 25°, partly at 18°. The three first broods
were produced at 25°, following which the parents were maintained at
18°, from the day the third brood was born until the birth of the fourth
brood. After this the temperature was raised again to 25°, at which
degree the development of the last broods of young took place. The
result of the experiment is given in Table VI.

TABLE YI.

Number of dorsal rays in offspring of the same pair of parents at
different temperatures.

No. of Specimens

25° 25° 25° 18° 25° 25°

^^:Pi Brood 1 Brood 2 Brood 3 Brood 4 Brood 5, 6, 7 Brood 1, 2,

3, 5, 6, 7

rays in

offspring Born 15/5 1918 Born 10/6 1918 Born 6/7 1918 Born 21/9 1918 Born 25/10, 21/11,
V ^ ' 24/12 1918

7 8 13 29 6 51 101

6 1 2 3 13, 4 10

5 _ _ _ 1 — —

n 9 15 32 20 55 . Ill

a 6-889 6-867 6*906 6-250 6*927 6910

a ±0-333 ±0*352 ±0-296 ±0-550 ±0*262 ±0-283

P.E.A. ±0-075 ±0*061 ±0-035 ±0-083 ±0024 ±0018

Fl. ±0*375 ±0*306 ±0*177 ±0-415 ±0*129 ±0«092

Thus we see, that whilst the broods developed at 25° had an average
of 6*91 rays, the average number of rays fell to 6*25 at 18°, The
difference between the averages was thus 0*660 and the probable error
of this difference ± O'OSo.

As all experiments in this connection have given a similar result, it
may be taken as proved that the number of rays in the dorsal fin of the
offspring is affected to a considerable degree by the temperature to which
the mother is subjected whilst in a state of pregnancy. Remarkable
besides is the great difference in the duration of pregnancy at the
different temperatures; which at 25° lasted ca. 1 month, at 18° more
than 3 months.

III. Importance of Internal Factors.

The object of the experiments about to be discussed was to investi-
gate whether hereditary differences in the number of dorsal fin rays
could be proved to exist. The principle of the experiments was to
maintain different pairs of parents in the same environment, and see
whether the offspring were different as regards the number of rays.

JOIIS. SOHMIDT 151

Thf 8pecinu«n» oiiiploytMl in the fX|MTiiiH*nt.s were H('I<-ct4'(l from two
racfj* with w-hich sincf 1915 srh»ctinn rx|MTiiiirnU \\iu\ hvm umlrrUikcn,
pjirtly U»wiinls a hi^h. jNirtly towanlH a low duiiiIkt of doi-Hal rays.
The jwrentsH had. roHiHTtivoly. both H and hoth (J rays in tho dornjil Hn.
In i»ach case the spi'ciniens were kept at a constant t<'ni|H'mtun', viz. 25 .
In addition, in «»rder U) srcurc uniform environment, the H|H-eimen«
whoiw offspring should In* comj)ared were placed in the same a<|uarium,
sejmnittHl only by a trelliswork of thin ^lass tubes. In other wohJh
they livinl in ijuite the same body of w.iter, maintained at a constant
temperature. The aipiarium contjiined no plants, l)ut a continuous
stream of atmospheric air bubbled through the water.

The exjKTiment falls inti> two series, A and B. In scries A there
were employed partly (^ 2G9 and J 270 (ejich with (i rays), partly (/ 2()7
and J 2G8 (each with 8 mys). For series B there were employed jwirtly
</" 27-t and ? 27.S (each with (5 rays), partly ^ 27G and ? 275 (each
with 8 rays). All the exjx*rimental fish in series A were kept in the
same aquarium which stood at the side of the one in which all the fishes
belonging to series B were placed.

From the appended Table VII and from the graph on p. 152 one
remarks that there wtis in both series a very great difference in the

TABLE VII.

Nunther of dorsal rays in offsprimj of four differeyit pairs of parents
at a constant temperature of 25° C.

Series

A

Series B

<?289x ?270
both 6 rays

c^267x ?268
both 8 rays

cf274x ?273
both 6 rays

cf276x ?275
both 8 rays

No. of
r»y8 in

>(T8priDg

Brtxxl 1, 2. 3,
Born 23/9, 17/10,
4/12 1918

4

911,

Brood 1, 2, 3. 4

Born 27/9. 23/10, 19/11,

17/12 1918

Brood 1.2,3,
Born 29/9, 24/10, 2
18/12 1918

4
11/11,

Brood 1. 2. 3

Bom 29 9. 26/10,

26 11 1918

8
7
6

62
25

54
10

51
23

40
3

n

a

P. E. A.
Fl.

87
6718

±0-455
±0-033
-0165

64
7 844

±0-366
±0031
±0-154

74
6 680

±0-466
±0-037
±0183

43
7-980

±0-258
±00-27
±0-133

average number of rays in the offspring of fishes with () anfl with 8 rays,
namely in the first series 1"131 (Probable error of ditierencr = 004.'))
and in the second 1-241 (RE. Diff. = 0045).

This difference cannot be due to difference in environment because
the fishes swam in the same aquarium, indeed in the very s^ime water

152

Racial Studies in Fishes

00
CH-
X
00

QO i>» «0

o

X

QO r»

PQ

00 r* «©

1

1

O)

i

2

a

00

o

p=3 d

le

bb

oc

•■^ J

O

fl

^

^ -5

1.

2

1

o

o

g

53 '^

ft .2

,J3

-^

__ "S

p^

0)

fl -^

s

^

S -w

c^

O

E-1

i g

"S

^

^ B

^

_^j

2 cJ

,

-^

1

j2

f^ o

I
a.

<

^

-3

ft

oc

M

:3

»«5
1

^6

o

■4^

^5

go

LU

I

03

.9

-s

o S
^1

u.

o

QQ

1-2
9 S

O

P4

1

&

-3^

z

'o

1

o

o

o

(U to

1-
<

-1

•1-1 Oj

z

o

l-H

^

>^ o

<

-J

Q.
X

UJ

.3

S3
1

in

H

00

!

o

1

be «!

§

• rH

<1

^ ^

•5

'5c

s

.2

1 i

i -2

.2

sn

2

ft -u>

0)

^

o

^ ee

g

■*^

-t^

o ^

i

■§

1

§3

-4J

;h

5b

CO

2

li

o

?

QQ

C» (Q

(I)

^

?a ^

t

ft
2

°l

^

cC

s

°o

0)

ClD

1 -^

cs

r]

J3

^

^

.9

H

ft

2 t— 1
ft

P

;h

Jons. Schmidt !;'):{

at rt consUiiit U'mjH?nittire and with regular vcntilati<»t). The* conditionH
with R'gjinl U> tmifonuity won* in my opjuinn (h«* inoHt rjivonnihlr
ptkssiblo and I cannot but conclude from the priHcnt ex|)crinient, thai
the difference pt-oved to erutt in the offspring of parents with resj)€ctivel y
6 and 8 my* is of hereditary (genoty/ncal) nature.

IV. Concluding Remarks.

The investigations here treated fall into two groups: (1) KxjM'ri-
ments in which the same mother was exposed to different environments
in dirterent jn-riods of pregnancy and (2) Experiments in which different
mothers weiv exposed to the stntie environment, have thus succeeded in
ehicidating thesi' n\tht»r compliaited (piestions.

It has been shown that tht' number of organs may be very susceptible
to environment, but that this fact cannot — under suitable experimental
conditions — disguise the fact, which we specially wanted to demonstrate,
viz. that there are or may bi' differences of hereditary nature between
the various individuals.

This proof is of considerable interest for our view upon the nature
of " races " in fishes, and supports in a high degree the opinion expressed
by me at a previous occasion': "My view then, with regard to the
nature of ' races ' in fishes, as characterised by our population analyses,
is briefly this : A fish ' race ' is largely a statistical conception. It
implies a mixing of different genotypes, and the average values
characterising the ' race ' are primarily dependent upon the (luantita-
tive proportion between these ; only secondarily on the environment."

' Jobs. Schmidt, "Racial Studies in Fishes. I. Statistical Investigations with Zoarce«
viviparugy L.," Journal of Genetics, Vol. vii. p. 117, 1918.

CROSSING THE NORTH AFRICAN AND SOUTH
AFRICAN OSTRICH.

l^v J. E. DUERDEN, M.Sc, Ph.D.,

Professor of Zoology, Rhodes Universitij College, Grahamstown ; Officer-
iti-Charge, Ostrich Investigations, Groot/ontein School of Agriculture,
Middelburg, C. P., South Africa.

(Wilh Plate Vn, ami Two Text-figures.)
CONTENTS.

PAOE

Introduction 156

North African Ostrich 157

South African Ostrich 157

Nature of the Material . . 159

Dimensions 162

Colour 163

Northern Ostrich 165

Southern Ostrich 166

Cross-bred Ostriches 166

Bald Head.Patch 168

The Egg 170

Egg of North African Ostrich 170

Egg of South African Ostrich 171

Eggs from Cross-matings 172

Eggs from Cross-bred Hens 173

The Wing Quills 174

North African Ostriches ....... 175

South African Ostriches 176

Cross-bred Ostriches 177

Survival of 42-plumed Ostriches 178

Scutellation of Middle Toe 181

Claw on Fourth Toe 185

Discussion 186

Factorial Constitution , 18«3

Adaptive Value of Changes 190

Establishment of Characters 193

Specific distinctness of Northern and Southern Ostrich . 196
Joarn. of Gen, vrii 11

156 Crossing the North and South African Ostrich

Introduction.

The continent of Africa, with the adjoining parts of Arabia,
Palestine and Asia Minor, is the natural home of the ostrich genus
Striithio. Beyond the confines of Africa however the wild bird is now
extremely rare, if it exists at all ; while in Africa it is slowly passing
away as the continent becomes occupied by the white settler. The
domesticated bird on the other hand has greatly increased in number
during the fifty years of ostrich farming, amounting to near one million
in 1913, though since considerably reduced owing to the less demand
for plumage as a result of the prolonged war.

Zoologists recognise four species of the two-toed ostrich : the North
African ostrich, Struthio camelus Linn., the South African ostrich
S. australis Gurney, the East African ostrich, S. massaicus Naumann
and the Somali ostrich, S. molyhdophanes Reichenow. The two last
mentioned are not however well-established species, appearing to repre-
sent intermediate types of the other two. On the other hand the
northern and the southern birds have well-defined characteristics
separating them, connected with size, colour, nature of the Qgg and
other minor features. Observing them side by side no one would
hesitate in assigning them specific distinction.

Recently a unique opportunity has presented itself for studying
numbers of the northern and southern ostrich under similar conditions,
and also the behaviour of their characters when the two are crossed.
In 1912 the Government of the Union of South Africa imported 132
specimens of the North African ostrich from Nigeria \ with the object
of possibly improving the domesticated strains built up from the
original South African wild bird. The imported birds were stationed
at the Grootfontein School of Agriculture, and the breeding experiments
to be conducted with them were placed in charge of the writer.

The main object of the investigations is the practical one of deter-
mining to what degree the plumage of the southern bird can be improved
by crossing with the northern, but in the course of the work many other
questions have arisen which have an interest to students of genetics
generally. The experiments have been in progress for over four years,
and during that period about a hundred cross-bred chicks (i^i) have
been hatched as well as a score or so of pure North African chicks ;
at the present time some of the first crosses have reached the age
at which they are beginning to breed, but only two chicks belonging
1 Report on the North African Ostriches imported into South Africa in 1912. Union
of South Africa, Department of Agriculture, Pretoria, No. 2, 1916.

J. E. Dl'KKDKN li)7

to th«» m-Kxmfl hyhri<l genemtion (f\) Imvr yvl Immh narrd. Tlw
earlier mating were carriiMl out with whiohrvcr of thr northrrn bircin
hup|)em'<i U> attain sexual maturity, irrrs|M'rtivo of their phiiiiagf
qualiticM or other chanieti'rs; but with thf abundant niat<Tial now
avaihible cn>sse8 iixv made with a definiiv pur|)<>s<! in view. Th<! long
periml Ix'tween the niatiirity of one gcniTation and thr n<'Xt, usually
between throe and four years, necessarily renders j)rogress slow'.

The distinguishing characters of the northi-rn and the southern
ostrich an* jvs follows :

A. The North Afric^m Ostrich, »SV/M^/a'o cr(m<'///.«< Liiuj. The sp<*cies
is hirger and strong(»r than the South African bird, tht; h<'ad reaching
about 8J feet fmm the gnaind ; the length of the body from the tip of
the bi'ak t^) the end of the tail is about 8 feet, and the total weight
about 275 lbs. The neck is also longer, about 3A feet in length, and the
body feathers extend upwanls for about IJ feet. The legs are longer,
thicker and uu)re robust, the knee joint being at le«ist 4 feet from the
ground, and the feet are larger; a claw is sometimes present on the
small toe and the seniles over the large toe may show one or, rarely, two
breaks.

The number of wing plumes, or remiges, is about 8() to each wing,
but varies from 33 to 39.

The colour of the skin in immature birds of both sexes, <is well ;is of
mature hens, is a creamy yellow, while the mature cock is bright red or
scarlet on the legs, heiid and neck, and^ red or pink over the body
generally.

The crown of the head has a bald patch, either single or partly
divided (Text-tig. 1, p. 158).

The egg is smooth, as if jx)lished, practically free from deep pittings
or pores, and larger and more rounded than that of the southern bird.
The average long diameter is 6'15 inches and short diameter 535 inches
and weight 3 lbs. 11 oz. (PI. VII, fig. 3).

Found in Northern and Western Africa, and in times past ranged
eastwards to Abyssinia, Arabia and South Palestine.

B. The South Afriwm Ostrich, Struthio australis Guvnvy. Smaller
than the North African, the less size being due to the slender and
shorter legs and neck rather than to any difference in the size of the

' The wild ostrich breeds when about four years old but the domesticated bird, largely
as a result of hiph nutrition as a chick and young bird, along with a certain amount of
anconscious selection, now usually breeds when between two and three years old, though
chicks are sometimes hatched from birds under two years.

11 2

158 Crossing the North and South African Ostrich

body. The head extends about 7| feet from the ground, the length from
the tip of the beak to the end of the tail is 7^ feet, and the total weight
about 240 lbs. The neck is about 3 feet long, and the body feathers
pass upwards for about 1 foot ; the knee joint is about 3 J- feet from the
ground ; a claw is rarely present on the small toe, and the scales on the
tarsus and third toe are usually continuous.

Text-fig. 1. Head of North African ostrich showing baldness. The oval area towards
the hinder border represents the pineal spot.

The wing plumes average about 36 to each wing, but vary from 33
to 42.

The skin of the neck, body and legs is pale yellow in chicks, dark
grey in the mature hen, and dark blue in the cock. In the sexually
ripe cock only the beak, front part of the head, naked skin round the
eyes and tarsal scales are a bright scarlet, and the rest of the tarsus and
toes pink, but the redness does not extend above the ankle.

The crown of the head is without any bald patch, and often bears a
tuft of longer hair-like feathers in the middle.

The Qgg is deeply pitted all over, smaller and more oval than in the
northern bird. The average long diameter is 6 inches and the short
diameter 5 inches ; the weight about 3| lbs. (PI. VII, fig. 3).

J. E. DUKHDKN 159

InhabiU pnicticnily tho wholo of tho sub-continent of South
Africa.

Ob«H»rve<i alonpiido one another, im can now Im^ <lone at (JrcMitfrintein,
no one could fail to tiislin^uish \\\r nnrth<'rn from th«* Houthcrn ostrirh.
The greater height of the former, the stronger liinhH and th<' light
yellow of the skin an* obvious features, while the sexually nuiture cock
is still nion* distinctive on account of the brilliant scarlet of the head,
ni*ck and legs, and the red colour of the b(Mly generally. The naked
patch on the head and the smooth, larger eggs are just ;us consUmt dis-
tinguishing features. Many small differences of feather structure occur,
and are of much imj)ortance to the practical ostrich farmer. Hitherto
all the distinguishing characters of the imjK)rted northern ostrich have
been retained under the new conditions of South Africa and re-appear
in the progeny, showing that they are hereditary and inde{)endent of
climatic and other environmental influences.

Whether the northern and the southern ostrich are to be regardei^i
as distinct species depends largely upon one's conception of the term
species and will be discussed later. It may be noted however that the
two are found to interbreed freely and reciprocally, and the crosses or
hybrids have also been proved to be fertile, both inter se and with either
of the parent forms ; at the same time the two races have many dis-
tinctive characters which are germinal in their origin.

Nature of the Material.

From a genetical point of view the material on which the investiga-
tions are being conducted is heterogeneous. The imported Nigerian
birds are such as were procured by the Arabs from wild nests, and then
reared in kraals for the sale of their plumage. Methodical ostrich
farming, as conducted in South Africa, is scarcely known in North
Africii, and chicks are not bred in captivity ; hence the birds are un-
influenced by any artificial selection. They exhibit much variation in
the details of feather structure, and many distinct commercial types of
plumes are represented, though all the birds come from one area. On
importation the greater number were chicks about six months old.
The majority of the older birds failed to become adapted to their new^
environmental conditions, and many died from lack of iminunity to the
various parasitic diseiises affecting the southern bird. A selection has
been made of those producing the most desirable plumage and the
others discarded.

160 Crossing the North and South African Ostrich

The South African birds employed are such as have been produced
by gradual selection during the fifty years or so since ostrich farming
was first established at the Cape. The foundation stocks naturally
consisted of wild birds, and the best of these and their progeny have
been employed in building up the superior strains of to-day. The
ultimate object in ostrich breeding is simple and well defined. The farmer
selects only for feather production; no other character of the bird is
taken into account, as any weakness in constitution or breeding power
is scarcely observable. Also, practically only one feather ideal exists,
namely, the largest feather combining all the many desirable features
at their maximum. The technical " points " relate to details concerning
length, width, strength, shapeliness, density and lustre, in all of which
ostrich plumes vary greatly. None of the original South African strains
possessed the highest expression of these in combination, and the object
throughout has been to gather into one all the best features available.
As yet no breeder has succeeded in doing this, though many are nearing
the desired end. The ostrich farmer clearly appreciates the distinct-
ness of the various characters of the plume, though in his selection for
mating he proceeds mainly on the assumption of a blending inheritance,
and in practice the method is succeeding, even though progress is slow
and much variety is encountered in the progeny. When the ideal type
of plume has been built up it is understood that it must be " fixed " by
a measure of in-breeding, and at present widely divergent crosses are
rarely made.

It is generally conceded that, notwithstanding all the selection
which has taken place, no advance has been made on the best of the
feather points originally scattered among the foundation stocks, except
such as can be ascribed to improved nutrition and other conditions
dependent upon domestication ^ All that the farmer has done is to
combine in the one plume the best of the features originally distributed

^ As long projecting outgrowths of the epidermis, with a core or medulla of highly
vascular nourishing dermis, ostrich plumes during their six months' growth are remarkably
sensitive to the nutritive and other physiological conditions of the bird; even the varia-
tions of blood-pressure between day and night often leave their impress upon the growing
feather in the form of "bars," while climatic conditions may make or mar the success of
the feather crop. The greatest difference in value obtains between a well-grown and an
imperfectly grown crop of feathers, and this accounts for the great care bestowed upon the
management of the birds, and the highly stimulating food supplied. Probably the high
grade ostrich is the best cared-for and most pampered of all our domestic animals. No
better subject for studying the influence of a varying nutrition and blood-pressure upon a
growing structure could be chosen than its long plumes, growing at the rate of a quarter
of an inch a day.

J. K. DUERDRN \(^\

Among the many wild HtminH; but it hiis hwu found iiniHiHsiblc to
choiigv liny «»f tluM'hiinioUTistics hrymul wluitniitiin* providrd. Tiiking
the fentun'M 8o|Ninitidy th** ostrich pliiinc atVonls a noteworthy inHUmcc
of the ini{M>ssihiIity of iniproveimnt. heyoinl thr maxima originally
pri'MMit, by meuiui of continued s«'l<'et.ivr bn-eding. All thjit tlu* pnK-rKs
haH achiev(sl is t-o s<'gn'gate the ehanicU^rs niost desired; moreovir, in
connei'tion with the |>oints of the feather no hint of any sj)ort or muta-
tion ever <H*eurs.

When t»om|)aring closely the many connnercial varietit's of ostrich
phnnes, ench repivsenting a separat** tyjK? to the specialist and having
a distinct value, the (juestion arises as to how far th(» many differences
in size, density, shajH' and lustre should be considered as Huctuating
variations, or how far they are eJementjiry chanicters. Farming ex[K'ri-
ence has fully proved that selective breeding within a typo will not
change any of the minutiae of the type. If a farmer requires any
particular featuiv juJded to his strain he must procure it from birds
whose plumage displays it: no <legree of breeding will otherwise produce
it. It is thus shown that the variations distinguishing the types are
germinal, not environmental, and should therefore be reganted as repre-
senting definite elementary characters. Yet how many of these elementary
characters must be represented in even a single ostrich plume! No
doubt the same multiplicity of small characters appeals to the specialist
in an intensive study of any domesticated stock. Experimental investi-
gations are usually undertaken on one or on only a few of the pronounced
characters, but when all the many details have to be considert^d which
are of the greatest importance to the practical breeder, there appears
no limit to w'hich analysis can be carried'.

The present paper is confined to the behaviour in crosses of certiiin
of the more prominent characters which distinguish the northern and
southern ostrich, such jis the dimensions, colour, the bald head patch
and details connected with the egg, and others which the two have in
common, such as the number of plumes on the wing, the scales on the
middle toe and the claw on the fourth toe. An attempt is made to
arrive at the germinal constitution of the ostrich so far as concerns these
characters, as well as their adaptive value and manner of establishment
in the race.

' Two preliminary attempts have already been made to analyse the various plume
characters of the parents and progeny in cross-matings, Breedimi ExpcrimcntK ivith North
African and South African Ostriches, I. "The Plumes of Parents and Chicks " ; II, "The
Flames of the Second and Third Clippings," Local Series, Nos. 13 and 17, Department of
Agriculture, Union of South Africa, 1917.

162 Crossing the North and South African Ostrich

Dimensions.

On account of its restless nervous nature and the difficulty of fixing
upon constant determinable points the live ostrich is not a creature
which lends itself to accurate bodily measurements. In any troop it
will be found that the members differ much among themselves, and the
same individual varies at different ages and according to its nutritive
condition. Hence the northern and southern birds and the crosses from
them can be compared only in general terms, as when seen side by side.
The average North African ostrich is a much taller bird than the South
African, being longer in the legs and neck. The head reaches to a
height of from eight to nine feet from the ground whereas in the Cape
bird it extends only seven to eight feet. The feet, legs and neck of the
Nigerian bird are also more robust. The general dimensions of the
body do not differ much in the two, the greater size of the northern
being mainly a result of the longer legs and neck. As a chick and
young bird the body of the northern ostrich however tends to narrow
behind more than the southern, but later this becomes a feature largely
dependent upon the nutritive state. The relative sizes admit of the
two being easily picked out in a mixed troop, the heads of the northern
birds towering a foot or so above those of the southern.

The cross-bred birds at maturity stand higher than pure Cape birds,
but are not so high as the Nigerian. As chicks the body tends to
narrow behind more than in Cape chicks, so that, with the slightly
longer legs and tapering body, they appear decidedly more slender than
Cape chicks of the same age. On the whole it can be said that as
regards size, especially length of limbs and neck, the hybrids follow
neither one parent nor the other but are intermediate between the two,
though the statement is not one which can be supported by actual
measurements.

The two chicks of the second cross-bred generation already reared
are now a year old and as regards their general size are strikingly like
the South African grandparents as contrasted with the North African,
including the shorter, less robust legs and neck. When mingling with
first generation cross-breds of the same age the difference is most
marked, and no one would hesitate in regarding them as pure Cape
birds. Such a result is at least suggestive that the distinctive sizes of
the northern and southern ostrich will undergo segregation in the F2
generation, but further chicks will be necessary before the real nature
of the segregation can be determined.

J. E. DUKRDKN IfK]

C0IX)lTR.

The 8kin or body colours of the oHtrich, jus <liHtinct from those of the
plunmgi^, vary from the chick to the julult sta^e, an- (iirt'ereiit in the hen
ami the cock, am! change in the latti-r with the breeding' stjite. They
also vary with the physiolo^icnl condition of the bird, m-cordin^' as the
surfiwe of the skin is clean or covered with scMirf. When low in con-
<iition the skin bec<uues dry and scaly, thereby masking the true coiotii-s ;
but lis a higher physiological state is reached the scurf" peels off or is
preeniMi oft', and the true fresh colour is n^vealed. This is particularly
the Cixse at the beginning of the breeding season when the skin coloui-s
are at their brightest. The colour is readily .seen on the naked legs
and under the wings, around the eyes and beak, and elsewhere by
turning aside the overlapping feathers. Chicks of both .sexes are
practiadly alike and even young birds show little distinction. The hen
remains throughout of the same colour us young birds, but the cock
undergoes a change beyond, and in places assumes a brilliant scarlet as
the nuptial state is attjiined. From the chick onwards the colour dis-
tinctions between the North African and South African ostriches are
strongly marked.

The red and sciirlet colouration of the cocks of both races, as well as
the rich dark blue of the Cape bird, are found to be dependent upon the
presence of the testes, while the black plumage is dependent upon the
absence of the ovaries. South African cocks which have been castrated
while young never assume the red and scarlet skin colours, but retain
the light or dark grey of all young birds and mature hens. On the
other hand the plumage of castrated cocks attains the normal blackness
of the sex as contrasted with the greyness of the hens, from which it-
may be inferred that the formation of the black pigment of the feathers
is not subject to any influence from the male gonads. Spayed hens
retain their ordinary body colour but the normally grey feathers are
found to assume the blackness of the cock, showing that ordinarily the
secretions from the ovaries exercise an inhibitory influence on the
formation of black pigment in the feathers of the hen, though having
no action on the .skin colour ^

> Prof. T. H. Morgan, Amer. Nat. \o\. li. Sept. 1917, in the case of tlie cock bird of
the Sebright bantam which is "heu-feathered," has proved experimentally that when
castrated a complete change in the plumage occurs, normal cock feathers appearing. He
considers that, as in the hen, some internal secretion, acting through the gonad, must
inhibit the development of the secondary sexual characters in the hen-feathered cock.
Morgan also refers to certain experiments by Goodale who has found that when the hen of

164 Crossing the North and South African Ostrich

The secondary sexual colours of the skin and plumage of the ostrich
are thus determined by altogether different influences ; the full attain-
ment of the one is dependent upon the presence of the testes and of the
other upon the absence of the ovaries. Two North African birds at
Grootfontein, although about six years old, have shown no signs of
sexual maturity ; they retain the cream yellow of all northern young
birds and mature hens, but have the black plumage of cocks. Evidently
some abnormality is connected with the internal gonads, but from the
external appearance of the birds it is impossible to say whether they
are cocks or hens. It may be noted that the removal of the ovaries or
testes, especially after a bird has attained maturity, has little or no
effect on certain of the sexual instincts. Thus a castrated hen will go
through the characteristic snapping of the beak and fluttering of the
wings as if broody, and will even crouch to receive the cock ; while the
castrated cock will perform his ordinary "rolling" display and even
mount a crouching hen.

In determining the sexual colours of the male ostrich the testes
clearly give rise to some secretion, presumably of an enzyme nature.
This must be produced at first in small quantities, and the colour changes
come slowly ; but as the testes ripen and become functional more of the
enzyme must be forthcoming, for the colour intensity increases and
remains brilliant throughout the mating period. With the beginning
of the six weeks' period of incubation the testes become less active,
pairing ceases, less enzyme is produced and the colour fades. The
differences between the sexual colours of the northern and southern bird
are well defined, and must be germinal in the first instance; but the
factors must act through the gonads, and presumably these exert their
influence by means of specific enzymes. Even if we regard the germinal
factors as themselves enzymes, as Troland ^ and others would have us do,
those concerned with the sexual colours must express themselves through
the gonads.

Though the scarlet colour of the cock is a secondary sexual character
it may well be doubted whether it has any influence on the mating of
the birds, or any preferential value in the eyes of the hen, as is so often

ordinary breeds of fowls is spayed she develops the full male plumage, as is also proved
above for the hen ostrich. Seeing that the plumage of the cock ostrich is more valuable
than that of the hen the results from spaying the latter have an economic bearing and the
practice is followed by some farmers.

1 Troland, L. T., "Biological Enigmas and the Theory of Enzyme Action," Amer.
Nat. Vol. LI. June, 1917.

J. K. DUKRDKN Hif)

auppcKHiHl to Ih» the Ci\»c with tho hright iniptiiil colourH of bin Is. For
northorn cocks arc a bright scarh't <»vcr all their ex|M)Hc<l |)arU at the
time of si'xual ri|KMies.s while s^uithcrn cm'ks are scjirlet only over the*
head and the tJirsus and are tar less striking in tlu'ir g<'n<'nd ap|)«*arancr,
yet ft northern htMi will crouch just as freely for the latter as for the
fonner. Occasionally ostriclns exhibit a dim su^^^estion of |)referential
nmtin^, l>ut in pnictice it is tound that any hen will jwiirwith Jiny vah-U,
and in " cainpini^ otV" as bre<Mlei>i the fanner never takes into accoinit
any |)ossible preferenci's on the part of the birds themselves. In a st;itc
of natun*. t»n the o|>en vi>ld. a c«K'k gathers round him one or more hens
jvs the breeding scjuion apj)i()aches, and v<'rv definite spatial limitations
become established am«»ng the diti'erent breeding sets, and woe betide
any cock which may wander (»n the area appropriated by another. In
all this however thi' hens are purely j)iussive and indifferent, and are
prone to lay in the s,ime nest, jus many as (JO or 70 eggs being sometimes
found in the one shallow depression. Further, Jis in most other birds,
the plumage is at its highest stiite of development at the beginning of
the mating se;uson, as if still further adding to the attractiveness of the
cock. Yet farmers .is often as not clip the plumes before mating birds,
and so preserve them from wear and tear, without however any influence
on the readiness with which pairing takes place.

The Northern Ostrich. In North African chicks the skin is a bright
deep yellow, almost orange, over the legs and head, and a slightly paler
yellow over the body and neck. As maturity is reached the hen becomes
a light yellow, the tarsal scales assuming a light or dark horn\' brown.
Some northern hens are slightly pink over parts of the body, and the
colour may show through the white downy covering of the neck.

The North African cock undergoes remarkable colour changes as
sexual maturity is attained, which are a sure guide to the farmer as to
the breeding condition of the bird. The deep yellow of the chick is
gradually replaced by a light yellow, this by pink, and then by red,
reaching a bright scjirlet over the legs, body, neck and head as the
actual inating period is reached. The bright scarlet colour contrasts
strongly with the jet black body-feathers, white waving wings, erect
light brown tail feathere and fleecy white down of the neck, and makes
of him a glorious creature as he prances about in his breeding camp in
all the pri(Je and pugnacity of his sex. The nuptial colours pde greatlv
when nesting begins, and also when the breeding season is over, the
body being reduced to a pale pink or brick C(jluur. At its height so

166 Crossing the North and South African Ostrich

sensitive is the colour to the physiological state of the bird that close
observation often reveals variations in the intensity within the same
day, as well as from day to day.

The Southern Ostrich. The skin of South African chicks is at first
pale yellow in colour and afterwards dark grey. Highly fed Cape
chicks may show a rich deep yellow round the eyes and beak, though
this does not continue for more than a few months. Mature southern
hens are a dark grey over the legs, body, neck and tarsal scales.

The Cape cock is at first a dark grey or steel colour, much like the
hen, but as sexual maturity is gained he assumes a fresh, bright blue
over the greater part of the body, while the tarsal scales, beak and naked
parts round the eyes become a bright scarlet ; the small scales over the
sides and hind part of the tarsus may also be red or pink, but ordinarily
none of the red colouration extends beyond the tarsus, nor over the body
and neck.

Thus northern chicks are a deeper yellow than southern chicks.
They pass to a pale yellow and the hen remains at this stage, but the
cock passes beyond to a pink and then a scarlet stage. The pale yellow
of southern chicks is early replaced by a dark grey which persists in the
hen, but is followed in the cock by a blue or blue grey as sexual maturity
is reached ; moreover, only the tarsal scales, beak, and skin round the
eyes assume the bright scarlet which characterises practically the entire
body of the northern bird.

In southern cocks the red colour of the northern would appear to be
latent, or perhaps wholly obscured by the dark blue; for on recovering
from an injury to the neck or body it is often found that the scar of the
new skin shows a reddish tinge.

Cross-bred Ostriches. The colour of the skin of cross-bred chicks is
intermediate between that of northern and southern chicks. The legs,
body and large scales are a pale yellow, which is lighter than that of
Cape chicks but never so deep as that of Nigerian chicks. The adult
cross-bred hen retains the light yellow body colour, though usually it
becomes a little darker compared with the chick. The colour remains
darker than that of the pure northern hen, but is invariably lighter than
the pure Cape.

The cross-bred cock retains the uniform light yellow of the hen
until sexual maturity approaches. He then assumes a pink tinge in
places and later the bright scarlet. As noted, however, it is in the

J. E. DlTKRDKN !«7

extent of the nnl colounition, not in itH intenMity, thiit tin; norilu'rn iind
southern cocks ciiffcr ho conspicuously ; in the former it is difTtiseci
practically all over, while in the latter it is limited to the head and legs
below the ankle joint. The sexually mature cross-bred cock is decidedly
intenuediate biM ween these two as reganls the area <>f the body Jissuming
the red colour. The head and tarsus are scarlet as in both parents, but
only a slight pink colour appeai-s on the upjxjr part of the leg and also
over the neck, and may even tinge the other parts of thr body, though
without approaching the bright red of the northern parent. The various
cross-brtnl cocks naturally differ as regards the degree and extent of the
colouration, but they never wholly follow one parent or the other. In
extreme cases the body colour may be a grey blue almost like the
southern cock or a grey yellow nearly like the northern, but all kinds
of intenntnliate tints are to be met with, even in biixls from the siime
nest.

In all the cross-bred ccx^ks thi' red of the neck is displayed to a
greater or less degree through the white downy covering. Sometimes
It is only apparent when the small hairy feathers are turned Jiside and
the loose skin put on the stretch. It then appears as red showing
through a bluish ground, the two producing a purple. From this it
w^ould seem iis if both the blue of the southern bird and the red of the
northern were represented, the resultant purple being the product of the
two. It is evident that the degree of redness of the body and neck is
partly dependent upon the other body colours. If the latter is dark
blue it naturally tends to obscure the red, while if the body colour is
only a pale yellow the red becomes more obvious. The nuptial colours,
dependent upon the presence of the testes, are superimposed upon the
true body colours.

As regards the two Fo chicks the colour of the body, legs and neck
is quite as dark as that of any Cape hen, showing no influence from the
lighter colour of the northern grandparent and the intermediate light
colour of the cross-bred parents. Both being hens however the colour
is not so distinctive as it would be in the case of cocks. Taken as a
preliminary result it certainly suggests that the colours of the northern
and southern birds have a separate factorial basis, and that segregation
will take place in the second cross-bred generation. The rearing of
further F.. chicks will be awaited with interest as likely to solve the
problem.

168 Crossing the North and South African Ostrich

Bald Head Patch.

The crown of the head of the South African ostrich is covered with
short, hair-like feathers, which often form a tuft of longer hairs in the
middle. A bare pineal spot^ present in all ostriches at the back of the
head, is so small in the adult as to be only noticeable when the feathers
are turned aside. The North African ostrich on the other hand is
distinguished by having the top of the head for the most part naked, a
bald patch beginning at the back and extending forwards in a shield-
like fashion between the eyes (Text-fig. 1, p. 158). The area is roughly
pear-shaped, but may be partly divided down the middle. In diagnostic
descriptions the baldness is considered to be a character of some
importance in separating the northern species from the southern and
is even mentioned in the writings of Pliny 2.

The extent and shape of the naked space vary a little in different
ostriches, but all the North African birds at Grootfontein display it to
a greater or less degree. It is quite independent of the pineal spot, and
its posterior border may either include this (Text-fig. 1) or pass in front
of it. In some birds, instead of forming a continuous patch, it is divided
more or less down the middle, having then a decided bilaterality. Often
a tuft of long, hair-like feathers remains towards the middle of the
hinder border, corresponding with the tuft in the southern bird, and
gradually disappears forwards. The area is covered with a horny, scurfy
layer, which peels off at times, exposing a fresh, clean surface of the
skin with the hard bony skull immediately below.

The baldness is not apparent in the North African ostrich chick
when first hatched. At that time the head is covered with short bristly
down as in the South African, and the character becomes established in
the course of the first six months or so of growth. It is gradually
formed by the dropping out of the hairy feathers from about two months
onwards, and in a batch of chicks of the same age practically all stages
in the loss can be observed, the feathers to remain longest being those
of the middle tuft. No sharp line of separation occurs between the
naked and the covered part of the head ; a few stunted feathers represent

1 An extra-cranial pineal body has lately been discovered in the ostrich. At a certain
stage of development it shows as a black pigmented area or vesicle which later disappears,
and only a dark, oval area, devoid of feathers remains in the newly hatched chick and per-
sists throughout life. Apparently the ostrich is the only bird with such a well-defined
pineal body, recalling that of the reptilia and persisting as a pineal spot.

'^ Hist. Mund. lib. xi. cap. xxxvii.

J., E. DrKKDKN H)9

the gnuiuiil IninHition, while the nuMlium tiitt nuiy or nuiy imt ]N>rHiHt.
No ei)rn\M|M)mIiiig falling out of thf fcjithcTH cvrr (nkrs place in the
Capt» chick.

Nfttunilly JH^me interest ha« Ix'rn attii(h«<l U* thr Ixhaviour of the
Iwild |)iitch in th<» cn^sscs of (he northern and southern ostrich. Of the
humlriHl or so cn)ss-hnMl chicks which have hwu hatched none iit first
showed any signs of Iwildness, hnt in every case the f<'athers iM'gan to
fall out when the chick was two «»r three months old, and at six months
the [Mitch was established as completely jis in adult N«>rth African
i»striches. Thus the baldness of the northern bird is shown to be
dominant over its absence in the southern bird.

The two second i^eneration chicks already reared are now w<'ll nvei-
the age at which the bald patch becomes established, and in one of thr-iii
the head remains coveri'd with hairy feathers jis in southern binls while
in the other the baldness h;is been formed ixs distinctly us in any
northern bird. The F.^ chicks thus atibrd evidence that factorial segre-
gation tjikes place in the second hybrid generation, and there can be
little question that when sufficient chicks of this generation have been
obtained it will be found that baldness behaves as ;i homozygous dominant
in strictly Mendel ian proportions.

The bald head piitch is therefore a distinctive Mendelian unit-
charactvr sepjvrating the northern and the southern ostrich. The
differences jissociated with the dimensions and colours of the birds, and
also those of the agg, are differences of features common to b(jth, but in
the Cape bird there is nothing suggestive of the baldness of the Nigerian.
It is an entirely new character which has appeared in the latter race of
ostriches, but not in the former. It may be regarded as a mutation,
and was presumably fully developed from the beginning, for though it
varies somewhat in its extent and form the differences are no more than
can be regarded as fluctuating variations. That it is germinal in its
origin is manifest since it appears in all chicks, both pure and cross-l)red,
while its dominance in all the latter proves that the parents are du|)lex
or homozygous with regard to it.

It Ciin hardly be suppo.sed that the baldness h;is arisen in response to
any external influi-nce, for it is unlikely that anything environmental
could affect the top of the head of the northern bird which w(»u]d not
have a corresponding action upon its southern relative, even if it were
possible that any influencing of the kind could bring al)out a corre-
s}x»nding change in the germ phusm. Nor can it be deemed to have any
adaptive value. It lends strong support to the view maintained by

170 Crossing the North and South African Ostrich

Bateson, and also by Morgan, that new characters make their appearance
as a result of changes in the germ plasm, without any reference to external
influences, or any utilitarian value or need of the individual. Since the
baldness is now present as a duplex dominant in all the imported birds
it must have originated long ago in the history of the northern ostrich,
sufficiently long for the change to have affected all the individuals. For,
as will be shown later, there is good reason to suppose that in the ostrich
a new character appears at first in only a few members, but gradually
extends to more and more, by the continued change de novo in the germ
plasm of the nulliplex members of the race.

The Egg.

As in all other birds the eggs from the same ostrich and also from
different ostriches vary within certain limits, as regards size, shape and
surface characters. Beyond these fluctuating variations however certain
well-defined differences distinguish the egg of the North African from
that of the South Afi-ican bird (PL VII, fig. 3).

Egg of North African Ostrich. The egg of the northern bird is
practically always larger than that of the southern, the shell is almost
free from obvious pores or pit tings, and presents an ivory-like smooth
surface. Usually also the northern Qgg is rounder in shape or less oval.
Measurements have been taken at the nest of the long and short dia-
meters of four series of eggs and are as follows, in inches :

TABLE I.

Measurements of Eggs of North African Ostrich.
Series A Series B

Long

Short

Long

Short

Diameter

Diameter

Diameter

Diameter

1

6-00

5-19

1

6-12

5-50

2

5-94

519

2

6-50

5-50

3

600

5-25

3

612

5-38

4

612

5-31

4

6-38

5-25

5

6-19

5-31

5

6-12

5-25

6

6-19

5-31

6

6-25

5-38

7

6-12

5-31

7

6-38

5-25

8

6-00

5-25

8

6-25

5-38

9

600

6-25

9

6 00

5-50

10

6-00

519

10

6-25

5-50

11

5-94

5-19

11

6-25

5-38

Average

6-05

5-25

6-24

5-43

J. K. DrKHDKN 171

TABLE I — rontitiMt/.
HcriM C S4TioH 1)

I^onf

HiKirt

t>UiiMt«r

I>t*m««(«r

1

625

ri»8

»

«a:.

581

a

6i»4

5 12

4

IVOO

r)i«»

5

». rj

5-44

fi

0 25

ryM

7

0-26

r>38

8

r.()c

525

9

(;i2

5:iH

10

fi'i.".

5-44

11

r»0()

61'.»

12

r,i2

5»8

IS

r.-06

5-31

14

6-25

5-50

Ixin*

H|M»rt

iMjunrt^r

l>latnpl«r

1

«0<'.

5-38

2

r,-3i

5 JW

:j

♦1-26

5-38

4

ncxj

5-44

5

0 12

5-31

r»

(i'25

5 31

7

6-25

5-88

Average fil9 5 37

Averat^e ri-14 5-34

Thus the jivmigo long diameter of 4M northern eggs is (Mo inches
and the short diameter 5'35 inches, representing an average difference
between the two axes of ()"8 inch.

Egg of South African Ostrich. The egg of the southern bird is
deeply pitted all over the surface, and pits often larger and more plentiful
at the air-chamber end, hence the shell does not present the ivory
smoothness of the northern egg. A Cape hen will sometimes prcnluce
a neiirly smooth, round vgg, but never to so marked a degree ;is th«'
typical Nigerian hen. Also the latter may occasionally lay eggs with
deeper pittings than usual, especially in the first one of the season.
Among a number of eggs from northern and southern birds mixed to-
gether no mistake can however be made in separating the one type from
the other.

The pitting which gives such a marked character to the southern egg
is jissociated with the respiratory pores of the shell. In the northern
shell the pores are so small and open so close to the surface as to be
sciircely visible to the naked eye, and are mostly scattered singly with
but little grouping. Hence the surface appears almost uniformly smooth.
though fine pores can be detected with a lens. In the southern egg the
shell pores are larger, sunken below the general surface and mostly in
small groups, varying from about six to twelve in a group. It is the
close grouping of the sunken pores which gives rise to the pitted surface.
In eggs which have been in the nest for some time dirt tends to accumu-
late within the pits and thus accentuates their presence, whereas in the
northern egg the \xn*'s art* ton small and shallow. Jn both ty|X'S the

Joum. of Cren. vui 12

172 Crossing the North and South African Ostrich

outer enamel layer shows differences in thickness and with it the polished
character of the surface. All the eggs are a cream or yellow colour
when freshly laid but fade considerably on exposure.
Measurements have been taken of 20 eggs as follows :

TABLE II.

Measurements of Eggs of South African Ostrich.
, Series A Series B

Long

Short

Long

Short

Diameter

Diameter

Diameter

Diameter

1

6-00

4-62

1

6-25

5 12

2

5-69

4-81

2

5-81

4-88

3

6-00

5-12

3

5-75

4-94

4

5-62

4-88

4

5-94

5-06

5

6-25

4-81

5

6-00

5-00

6

d'69

4-50

6

5-81

5-00

7

6-12

4-81

7

6-00

5-00

8

5-81

500

8

5-88

5-00

9

6-06

5-00

9

6-00

5-00

10

600

5-00

10

5-94

5 06

Average

5-92

4-85

5-93

5-00

Thus the average long diameter is 5*92 inches and the short dia-
meter 4*92 giving a difference of 1 inch. With such variable structures
as eggs a larger series of measurements is desirable in order to secure
a more reliable comparison. They serve to show however that on
the average the northern egg is about a quarter of an inch longer
(6"15 — 5*92 = 0*23 inch) and two-fifths of an inch broader than the
southern egg (5"35 - 4*92 = 0'43). The mean difference in the two
diameters is 0*8 inch for the northern and 1 inch for the southern,
indicating that the former are rounder or less oval than the latter. .

Eggs from Cross-matings. In breeding for cross-bred chicks the eggs
follow the characteristics of the hen whatever the cock may be, that is,
the eggs laid by a northern hen mated with a southern cock are large,
round and unpitted, while those from a southern hen mated with a
northern cock are smaller, oval and pitted. Thus as regards size, shape
and surface features, the egg as laid is uninfluenced by the male bird
and partakes wholly of the nature of the hen. This is what would
ordinarily be expected, seeing that the germ from the cock unites only
with the germ of the hen, and scarcely any further change takes place
before the egg is laid. As the albumen, shell-membranes and shell are
formed in the oviduct of the hen after fertilisation it is difficult to see
how the coverings of the egg could be influenced. Instances are adduced
however where in crosses of other birds giving differently coloured eggs

.1. K. DlKUDKN ITo

the cxxjk HoiMUH u* exrriMHo moiuo inHucncc, t lu« phtMioinonon Immii^' H|M»k«'n
of a« Xonia {Jouni. Ifermiitt/, Vol. vi. No. T)). Thr divrrsr rlmra<t«ix of
the eggs of the northmi nml southern ostrich atVonl a ffo<M| test ca.s«-
of the }K>ssihility of Xenia occurring, hut from nom- of tho croHs-imitings
has atiy indimtion of the phfiioinenoii horn forthconiiii^.

KgpsJ'rom Cross-bred Hens. In eross-hnd hens are naturally coiu-
biiUKl the jxxssihilities of both the northern and the southern parents, and
the characters of the e^^ lai<l hy them are just as mueh a part of the
make-up of the hini as are the nutre obvious body features. The (pies-
tion therefore arises whether the e^gs lai<l by cross-bn'd hens will follow
those of one |)Jirent or the other, or be somethini^' intermediate betwe«'n
the two. Scores of eggs laid by cross-bred hens have been examined at
the nest and in the incubator and in all cases have been found t<» b<-

TABLE ill.
Mtotfuremtint^t of £</(/.<< oj" CronsJtrff/ Ostriclfx.

Scries B

Seri<

i» A

IxjnK

Short

PiameWr

1 >ianieUT

1

Ull)

r,l2

rj

0-25

5 12

3

012

ol'.j

4

612

5-OH

5

012

.")12

G

♦rl2

5-06

7

6-06

5 12

8

()-00

:yi)0

11

5-88

oOU

IxjtiK

Short

I)iainet«r

Diameter

1

6-25

.512

2

-,•81

4-94

3

5-94

4-94

4

0-00

5-00

5

6 19

512

0

5-88

4-94

7

r,-2o

5-12

8

(J-19

r,-()0

9

6 19

oUH

10 5-94 o-OO 10 5-88 5 00

Average 6-08 5-08 (rOO 5 02

Series C

Long IMameter

Short Diameter

1

5-94

5-12

2

5-81

512 •

3

5-88

5- 12

4

5-75

5 06

5

600

5-12

0

6-06

506

7

5-62

512

H

5-91

5 12

9

5-88

5(»6

10

5-8H

5 12

11

5-81

5- 12

12

5-8H

5<I0

13

gim;

5 12

14

5-94

5 12

Average 'yH\) 5 09

12—2

174 Crossing the North and South African Ostrich

intermediate as regards size, shape and the nature of the shell between
typical northern and southern eggs. Obvious pittings occur over the
shell, often more numerous towards the air-chamber end, but are never
so plentiful nor so deep as in the eggs from the southern bird. The egg
has neither the full size nor the roundness of the northern ostrich, but
is larger than the southern, and its general surface is more enamel-like.
Naturally variations occur in the eggs laid by different hens, and some-
times they approach those of the one parent more nearly and sometimes
those of the other. The degree of pitting and smoothness of the surface
of the shell do not admit of more than a general statement, but the
dimensions of 34 cross-bred eggs are available for comparison with those
of the northern and the southern eggs.

The average long diameter of 34 cross-bred eggs is therefore 6*01 inches
and short diameter 5*06 inches, the difference between the two diameters
being 0'95.

The measurements of the three series may be compared as follows :

43 North African Eggs

34 Cross-bred Eggs

20 South African Eggs

Though not elaborate enough for many purposes the results suffice to
indicate that as regards size and shape the cross-bred eggs are inter-
mediate between those of the northern and the southern bird. They
apparently approach somewhat nearer to the southern than the northern,
but with such variable objects as eggs an indication of this kind maybe
deemed of little value when only small numbers are available.

The intermediate nature of the cross-bred eggs, as regards size,
shape and the nature of the surface, may be taken to suggest that the
different characteristics of the eggs of the two races of ostrich are de-
pendent upon separate factorial representation in the germ plasm, as in
the case of the dimensions and colours of the birds. Also the factors
are not alternatives, for in the hybrid Qgg no one character of the parents
is dominant or recessive to the other, but each strives, as it were, for
expression, the result being something midway between the two.

The Wing Quills.

In farming practice the number of plumes to the wing of the ostrich
is an important matter, though selective breeding has hitherto been
concerned with the quality of the plumes rather than with their quantity,
it having been assumed that not much variation in number occurs. The

Long

Short

>iameter

Diameter

Difference

6-15

5-35

0-80

6-01

5-06

0-95

5-92

4-92

1-00

J. K. DrKunKX

17:>

first n»w ol* pluiiies, llu' win^ t|uillN nr rrmi^tM, iialmlrM the iainiliiir
whiu? plmiici* of the ostrich which arc by far l\\v most vahuihli* ns coin-
pnrLMJ with the first iiml «ccon<l rows of upjMT-cnvcrts which an* also
clipptHl, and an* hhick in thr cock and ^rcv in thr hen. Ah the coverts
altiTnati' with the wing <|uills the innnbrrs in all the rows an* definit<'ly
c*>rrelate<l, .h«» that for pur|)oses of coniparisnn aiunn^^ ditV^nnt birds
attention can be confined to the first n»w. thr nini^^'s (PI. VII, fi^^s. I
and 2)

North African Ostriches. The first-row fcathei-s on each win^^ havr
been connted on 25 of the original imported North African birds and
the n^sults are given below. It will be noted that a ditterence of one or
two phinies is often found between one wing anri the other, })ut the
cocks and hens show fto distinction. The number on the wing varies
from 38 to 39, the arithmetical mean of the series bi'ing 3()-54 ; repre-
sentinl gniphically they approximate to a normal fre(juency curve with
the mode at 3(). Manifestly the birds represent a mixed i>opulation, a
result of indiscriminate breeding in a race in which the numbers difier
by small amounts; but indications are not wanting that a pure line win
be built up of each number. We may regard each bird as heterozygous
with regjird to number of plumes, and a mixture of the kind given below
is what would be expected seeing that the birds come from a single area
in North Africa where no farming selection is practised.

TABLE IV.
First-rou^ Plumes on Wings of Imported North African OstricJies.

Riglit
Winfc

Left

RiRht

I^ft

Wing

Wing

Wing

1

Hen, No. 11

39

38

14

Hen, No. 87

36

37

•2

Hen, No. 20

37

37

15

Cock, No. 92

37

36

3

Hen, No. 40

36

37

16

Hen, No. 105

36

37

4

Hen, No. 41

34

35

17

Hen, No. 108

37

36

5

Hen, No. 45

36

36

18

Cock, No. 115

38

38

()

Cock, No. 50

36

37

19

Hen, No. 116

33

34

7

Hen, No. 63

37

38

20

Hen, No. 130

35

36

8

Hen, No. 69

38

39

21

Cock, No. 141

35

34

9

Hen, No. 71

30

35

22

Cock, No. 252

3 s

39

10

Hen, No. 75

37

36

23

Hen, No. 277

35

36

11

Cock, No. 78

37

36

24

Cock, No. 284

3()

37

12

Cock, No. 84

35

36

25

Hen, No. 287

38

39

13

Cock, No. 85

38

38

The number of plumes on the wings of 15 pure North African chicks
reared at Grootfontein from the importation, are also represented and
give approximately the same arithmetical mean as the above, namely
36*7, though without the low numbers 38 and 34. The chicks are from

176 Crossing the North and South African Ostrich

three separate breeding sets and may represent some slight selective
influence compared with the mixed importation.

TABLE Y.

First- row Plumes on North African Ostrich Chicks reared at Grootfontein.

Eight

Left

Wing

Wing

1

No. 333

36

36

2

Nq. 334

36

37

3

No. 335

37

37

4

No.l

37

37

5

No. 2

38

38

6

No. 3

36

36

7

No. 4

36

36

8

No. 5

36

36

Ri^ht
wing

Left

Wing

9

No. 6

36

37

10

No. 7

39

38

11

No. 8

38

37

12

No. 9

35

36

13

No. 10

38

37

14

No. 11

36

38

15

No. 12

36

35

South African Ostriches. In order to ascertain the number of plumes
on the South African ostrich for comparison with the North African
recourse has been had to the troops on various ostrich farms. Included
among them are representatives from the best and most widely known
ostrich strains of South Africa. It is only necessary to give the detailed
countings of one series of 48 birds as an example, the average for the
others agreeing closely. As much interchange of birds and chicks has
taken place since ostrich farming commenced it is most unlikely that
any additions to the series will vary from the averages here given.

Farmers^ Series. No. 4.

TABLE VI.

er of Plumes

on South African Ostrici

Right

Left

Wing

Wing

1 Hen

39

39

2 Cock

35

35

4 Cock

37

36

5 Cock

36

37

6 Hen

36

36

7 Cock

37

36

8 Hen

37

36

9 Cock

37

38

10 Cock

36

36

11 Cock

36

35

12 Cock

37

37

13 Hen

39

38

14 Cock

36

36

15 Cock

38

38

16 Hen

36

36

17 Cock

38

37

18 Cock

36

37

19 Cock

37

36

20 Cock

38

37

21 Cock

39

38

22 Hen

34

34

23 Hen

38

38

24 Hen

37

36

25 Cock

39

40

Right
Wfng

Left
Winf

26

Cock

39

39

27

Cock

36

35

29

Cock

39

39

30

Cock

38

38

31

Hen

37

34

32

Hen

37

38

33

Hen

36

35

34

Hen

36

37

35

Cock

36

36

36

Cock

36

35

37

Cock

35

36

38

Hen

35

35

39

Hen

37

38

40

Cock

36

35

41

Cock

39

88

42

Cock

39

38

43

Cock

40

39

44

Cock

35

34

45

Hen

37

38

46

Cock

36

36

47

Hen

37

36

48

Hen

37

36

49

Cock

37

37

50

Cock

39

38

J. K. DrKRDKN 17

; <>l hvo

rAnuiTH :

^ericK an* a.s loll

lowH :

No. 1

'iri hiuU

.'i(VHM

No. '2

'24 birdii

:u;:.M

No. 8

n» binU

8r.«»7

No. 4

4H liinlH

;ifi-M7

No. r,

10 binU

.■j«»«:t

ToUl MoAii

:m-7H

The avonij(o number of plunu's on tin- South AlVican ostrirh is
therefore the same a.s that on the North African, an important con-
clusion which could not have been arrived at without the opjxjrtunity
of counting hirge numbers of each.

As the northern ostriches now at (irootfontein were all procund
originally by the Arabs of Nigeria as chicks from wild nests, and are
uninfluenced by any artificial breeding, we may presume that their
plumes repR\sent the avemge for the North African wild bird, and we
have therefore go<xl reason for concluding that the ostriches over tht;
whole continent of Africa pnMluco the .same average number. From
this it follows that during the fifty years of ostrich farming in South
Africa no advance ha^ been made on the number of plumes originally
present on the wild bird. On the average the domesticated birds of
to-day produce the same quantity of plumes as the original birds with
which the first ostrich farmers commenced in the sixties.

Though somewhat remarkable at first sight this result is scarcely to
be wondered at if we bear in mind the principles underlying ostrich
breeding : Farmers have bred for quality ; quantity has never been taken
into account. Great advances have been made in the so-called quality
characters of the individual plume, but in doing this no attention has
been given to the number of feathers which one bird produces as com-
pared with another, and therefore no numerical change hiis taken place.
It is a good instance of the principle that no progress is ever made as
a result of indiscriminate breeding, unless a character has some selection
value, or mutations are taking place.

Cross-bred Ostriclies. Seeing that the northern and southern birds
have the Siime average number of plumes and are a mixture of hetero-
zygotes, no change is to be expected in the number of plumes on cross-
bred chicks compared with what would be procured by mating two
northern <jr two southern birds. The table given below is an example
of the results which have been obtained. The arithmetical mean of the
parents is 36*24 and of the chicks 3028, but for a larger series the
average is 36'31 which agrees more closely with that of the two race.s.

178 Crossing the North and South African Ostrich '

A hint- at factorial purity is indicated seeing that the extremes 33, 34
and 39 are not represented.

TABLE YII.

dumber of first-row Plumes on Cross-bred Chicks from mating a
North African cock with a South African ken.

Parents :

North African cock, No. 78
South African hen, No. 225

Chicks

1

No. 300

2

No. 301

3

No. 315

4

No. 316

5

No. 318

6

No. 320

7

No. 321

8

No. 322

9

No. 323

Right Wing

Left Wing

37

36

36

36

36

36

35

36

38

37

36

37

37

36

35

36

36

36

36

35

38

37

Survival of ^2-plumed Ostriches. Among the Cape birds in the
Grootfontein flock two^ have been discovered with 42 plumes to the
first row, though the rest have the usual average of about 36. At first
it seemed as if two distinct strains of ostriches were represented in
South Africa, as compared with the single strain in North Africa, one
with approximately 36 plumes as the average and another with 42.
The countings on farmers' birds have however given no support for this
view ; they have disclosed no individual bird exceeding 40 plumes, nor
any influence from a 42-plumed strain. Hence it is concluded that the
occurrence of ostriches with 42 plumes is altogether exceptional among
Cape birds, and has had no recent influence on the general average.
Likewise none of the Nigerian birds has more than 39 plumes, nor any
of the chicks reared from them ; so the influence of a 42-plumed strain
is non-existent in North Africa.

As regards their origin it is manifest that the 42-plumed birds re-
present a distinct departure from the ordinary 86-plumed birds. Were
no other evidence available the high number might be looked upon as
a meristic mutation and, as will be proved later, the birds give progeny
with such high numbers as to show that the extra plumes are not merely
the extreme limit of a fluctuating series but have a factorial value. They

' Of the two original birds one has since met with an accident and died. The birds
were procured several years ago from two farmers widely apart, without any suspicion of
their number of plumes. It is noteworthy that though search had since been made among
the same flock yet in neither case has another 42-plumed bird been found.

J. K. DrKRDKN 171>

havi' als<i Ihh'Ii (Muisidonsi in tin* li^ht t»f n'ViTHioiis to an jurlicr jinc<«Htrv,
but fulliT con»i»li'niti»»n Kwl.s uh to arcount. r>r thrm otlHTwJHr. l{<'tM'nt
«>hm;rvation8. to Ix* fully <l«'»tTil>e<i in a later j)a|Mr, hav«' shown that th«*
•xttrirh pn^s«'ntss us with nunurouH stji^os iinlicatin^ thr roui-sv of llw
dojjoiuTation which the win^ and Io^h have undcr^on*' up to thr jurm-nt.
118 well as tho course likely to hr followed in thr futun'. Survivals of
many antH»stnd chanictcristics an* to In- found ainon^^ th«* plmtifid
nmt<Tial now availahh* for study. Thus, while ordinarily only on<' in-
complete^ row of undcr-coverts is present, a fanner's strain exists in which
a second row of under-covertHS is almost complete, ami several memlx-rs
of a thin! row also <x*cur. and single m<'mhers of both rows are (K-c;ision-
ally met with on other birds. All slaves are t<» Ix* found from a complete-
niw of under-coverts to the usual (»ne where H to 10 are wanting at one
endof therow(Pl. Vll.fig. 2); and conditions of a like nature are to be met
with in the second row of upper-coverts. Vestigial down is to l>e found
on most ostriches over the wings and tail, though it is usually stated to
be absent. Further, while usually burie<l in the Hesh of the wing, the
thin! digit sometimes beai-s a second phalanx and projects freely from
the surface, and even bears its own plumes, a primitive conditi<^n sug-
gestive of the fossil bird Archneupteryx. The claw is usually absent
from the small, fourth toe of the foot but still survives in a few ; while
the sciiles on the middle toe show the beginnings of loss by one or two
breaks in their continuity (Text-fig. 2, p. 182). Experiments hitherto
carried out all indicate that the individual losses have proceeded as
retrogressive mutations, (»n definite factorial lines and in well-defined,
determinate directions.

In view of all these survivals of many of the earlier characteristics
of the ostrich the 42-plumed bird may with good reason be regarded as
a survival of a stage when the average number of plumes to the wing
was larger than at present. On this interpretati<m the 36-plumed birds
of to-day are to be considered as degenerate in the number of wing quills,
as they are in many other respects. The practical endeavour is now-
being made to build up a pure strain of ostriches bearing 42 plumes, for
with the increase of the other njws of plumes in correlation with the
wing (juills it bec«jmes possible to provide the farmer with an ostrich
giving about 25 more plumes than he receives from his birds at
present, the "quality j)oints " being also of the highest. The demon-
strati«>n below that the high number is not merely the extreme of a
fluctuating series, but is factorial in its nature, renders this possible.
Whether bv continued s«'lecti(»n thi' number 42 will ever be exceeded

180 Crossing the North and South African Ostrich

is doubtful, seeing that the factors for any higher number have probably
been altogether lost to the race, even if they were ever present in the
ancestral ostrich.

Table VIII shows that when the 42-plumed southern cock is mated
with various North African hens of the 36 strain the average number
of plumes in the progeny is practically intermediate, namely 39*56, the
lowest number being 37 and the highest 42 ; they do not regress to the
general average. The numbers form an approximately normal curve
with the mode at 40. None of the birds hitherto employed as breeders
can be deemed " pure " as regards the number of plumes ; and the 42-
plumed bird is probably heterozygous like the rest ; hence the fluctuating
series represented below.

TABLE VIII.

Number of first-row Plumes on Cross-bred Chicks from mating a
4r'2-plumed Southern cock and 3Q-plumed Northern hens.

Right Wing

Left Wing

Right Wing

Left Wing

1

No. 226

42

41

13

No. 311

40

40

2

No. 228

39

40

14

No. 230

41

40

3

No. 229

41

42

15

No. 232

39

41

4

No. 242

40

40

16

No. 233

41

42

5

No. 243

39

40

17

No. 234

42

40

6

No. 302

37

38

18

No. 237

40

39

7

No. 303

38

38

19

No. 238

39

38

8

No. 304

40

40

20

No. 1

39

38

9

No. 307

38

38

21

No. 2

39

39

10

No. 308

40

39

22

No. 3

38

39

11

No. 309

37

39

23

No. 4

40

41

12

No. 310

38

39

24

No. 5

41

40

Experiments are being undertaken to determine how far it is possible
to extract numerically pure lines, especially as regards the two extremes
33 and 42, but progress is necessarily slow. Until this has been done
full proof will be lacking that each plume has its own factorial repre-
sentation, though all the evidence points in this direction.

The 33-plumed birds represent the extreme of the loss of wing quills
which has taken place in the ostrich of to-day compared with the maxi-
mum of 42 plumes. If, as we seem bound to suppose, some intrinsic
influence is at work within the germ plasm inducing slow retrogressive
changes, it appears not unlikely that by in-breeding pure 33-plumed birds
it will be possible to increase the action of the degenerative force and
produce a still further loss of plumes. By selection it should be possible
to control the further evolution of the ostrich with regard to the number
of its plumes.

J. K. DrKUDKN 1«1

As iiutinl I'lirlirr. tanning pracliri' him rMtnhli.shcd ns clciirly as cmld
he fxpt'cUni that thr diffrrt'iit '* |H»ints " of tin* plumes an* factorial in
their U^haviour. and an they vary in thr varioiin Htniins a wparaU*
>(i*nninal n'pn^srntation for vach may hv juHsumrd. Kven Mtich a simpl*-
structuml part ax the central shaft, of the feather shows many ditVerent,
typos which ap|H»ar either distinct or as inU-rmediates in crosses. The
length of the Imrhules an<l tht'ir closeness on the harhs are also inatt<M-s
i>f much economic im|>ortanc<' in determining the "density" of tin- j)lunies.
and the tarmer never «|uestions their distinctness in breeding. At the
present stage of the ostrich industry, where crossing is pnvctised t<» such
a high degri'e. the factorial analysis (jf the individual j)hnue would he
a prolongiMi undertaking, hut will become feasible as the farmer finds
himself in a j)osition to build up pure strains.

SCUTELLATION oF MiDDLK ToK.

Along the tVont of the tiirsus extends a series of large, nearly rect-
angular sciiles, which in most c.ises continues uninterruptedly to the end
of the big middle toe. Over the leg the contiguous edges of th<* scales
simply meet, but they are imbricated where the tarsus joins the toe and
also over the latter, thus allowing for the bending movements of the
toe during walking and running. Along the tarsus the scales retain
approximately the same size, but at the junction with the toe they
usually become somewhat smaller, enlarging again distally. Occasionally
a distinct break in the continuity (jccurs at the joint, several of the large
scales disappearing and being replaced by insignificant ones like those
which cover the surface of the limb generally ; and in rare cases a second
break in the continuity takes place over the joint about the middle of
the toe, thus giving a proximal and distal series of digital scales (Text-
fig. 2, p. 182).

The number of scales fluctuates in different individuals, and also on
one limb as compired with the other; and occasionally irregularities are
introduced owing to single scales being partly subdivided. At the
breaks the large scales tend to pass insensibly into the small scales of
the limb, hence any enumeration is only approximate. A few countings
are given in Table VIII, p. 188.

The break in the continuity occui*s rather rarely, especially in
southern ostriches. Out of 20 Cape birds of mixed breeding only one
showed an interruption, while in 20 mixed Nigerian birds a single break
occurred in three c;ises and a double break in two. As the figures in

182 Crossing the North and South African Ostrich

Text-fig. 2. Tarsi and feet of Northern Ostrich. The scutellation shows a strong break
between the tarsus and the large, inner third toe and the beginning of a second break
over the middle joint of the toe. The small, outer, fourth toe bears only a few scales
and the claw shown is unusually well developed.

J. K. Di'KKDKN 18:]

TABLE VIII.
Xutnbf'r of Scale's in Tarmjmlal ScnlfUation.
A. ContinuoM* :

KiRht Tiuniiii

fth.I T.*

l^ftTamuiiaiul Ttw

1

a

4

«7

55
53
50
67

B.

If 'if A one bifnk :

TamuH

TiK!

Tamui

I

To«

1
2

27
•2H

16

17

27
30

17
19

C.

H'ith two brfakf

Taniuii

Toe

TainuM

Toe

1
•2
3

29
30
32

5 H
() 9

29
31
31

5 8

6 9

7 11

Table VIII show, the breaks represent a deHnite loss of scales. Taken
along with the other facts of degeneration in the foot, the losses are
without doubt to be regarded as the first evidence of degeneration in the
middle toe of the ostrich, the first, second and fifth having already disap-
peared and the small fourth being well on the way. The breaks evidently
represent independent unit characters, retrogressive mutations, in course
of introduction within the whole race, the process having gone a little
further in the northern ostrich than in the southern.

In all ostriches the tarsal scutellation is now distinct from that over
the small toe, only 8 to 10 scales occurring distally (Text-fig. 2). Com-
parison with other birds would, however, lead one to expect that the
two series were originally continuous', as they are still in the great
majority of ostriches with regard to the middle toe.

Breeding experiments prove that the breaks between the tarsus and
middle toe are germinal in their nature. Where no break occurs in
either of the parents the progeny also show no loss of scales. Thus in
13 cross-bred chicks from a southern cock and a northern hen, both
with a continuous scutellation, no loss of scales occurred. When how-
ever one of the parents bears a break and not the other, then, as indicated
below, approximately one-half of the chicks displays the loss, proving
that the factor for the break is dominant but that the germ plasm is
simplex or heterozygous with reference to it.

' At extensive series of birds' feet is shown on p. 425 of Sedfiwicks, Students' Test-book of
Zoology, Vol. ii. 1905, where however the Kcutellation of Stnithio camelim is erroneously
represented, i\\o scales of tlip small to«' boinp depicted as continuous with those of the
tarsus.

1 84 Grossing the North and South African Ostrich

TABLE

IX.

Scutellation in Parents and C)

Parents :

No Break

North African cock, No. 78 x

South African hen, No. 225 —

Cross-bred Chicks :

No. 314

X

No. 315

X

No. 316

No. 318

—

No. 319

No. 320

No. 321

X

No. 322

No. 323

X

No. 300

X

No. 301

—

Break

The heterozygous condition with regard to the break is what would
be expected, assuming that the character is one which is in process of
introduction within the race, and that it behaves in Mendelian fashion.
At present the mutation is found in comparatively few individuals, and
in a state of nature there is little chance that a bird showing the break
would mate with another in a like condition, but rather with one having
the scales continuous. If the change first took place in a homozygous
duplex manner in a few individuals there is small likelihood that these
would mate with others in like condition, but with nulliplex individuals.
The first crosses would be dominant and simplex, and these mating with
other nulliplex birds would give half simplex dominants and half nulli-
plex, which is what the experiments indicate. As shown below, certain
significant results have been obtained on in-breeding some of the cross-
bred birds.

TABLE X.

Scutellation in F^ chicks comjiared with Parents and Grandparents.

Parents :

North African cock, No. 9
South African hen, No. 225

Fi Crosses :

Cross-bred cock. No. 182
Cross-bred hen, No. 179

F2 Crosses :

No. 1

No. 2 ...

No. 3

No. 4

No Break

X

Break

J. K. DlKUDKN iHf)

In this cast' thf original |Mux?nU wen* a North AtVinin cck-U without
any Icmw of nailos and a South African h<»n with a Hin^h* hrcak. Of the
four ortHpring n'ariMl thn**' arr without the break while it «K!eurH in
No. 17i), the hen used in the »«xperinient. From th<* uiatiug of brother
and sisttT four chicka wert* hatched, two of which had two bnaks in th«'
scutellation. one had only a singh' break and one ha<i no bn*ak. The
result may Iv i\»garded jus highly suggestive that the iidi<'rcnt t<iidrncy
towanls the loss of scales can Ik? accentuated by in-breeding, and de-
generation thus accelerate<l along definite prescribed lines; for after the
single break the next step in the loss is a second break over the middle
of the t(H». In the course of the investigations it hjus become evident
that i\n inherent tendency exists in the ostrich towards the loss of various
parts of the fore and hind limbs in a continuous determinate direction,
as well as of it** plumage, and it is not unlikely that by in-breeding the
degenenitive tendency can be accentuated. The accumulation of fuller
data must howevtr be awaited before the suggestion as regards the loss
of scales can be regarded as more than tentative.

Claw o\ Fourth Toe.

The claw on the small, fourth toe has for the most part disiippearefi
from the ostrich, though it is occasionally present, more often on North
African than on South African birds (Text-fig. 2). In 25 mixed northern
birds it occurred on six specimens and was wanting on the othei*s, while
in 20 mixed southern birds it was found on only one individual. Every-
where it is feebly developed, especially by comparison with the big claw
on the middle toe, and is always non-functional, never reaching the
ground. Where best formed it projects for about half an inch from its
socket, while at other times it is almost hidden in the integument, and
can scarcely be felt with the finger ; but all intermediate sizes can be
obtained!. Usually it is equally developed on both the right and left
foot, though sometimes a difference is observed.

In crosses where both pirents are without the claw the progeny are
also devoid of it, though in a nest of IG chicks from clawless parents
a distinct claw appeared in one (Cross-bred No. 304). Where however
one parent is clawed and not the other it appears in about half the
progeny, showing that the clawed birds are dominant heterozygotes.
Out of a totid of 3(1 chicks hatched from breeding paii^s, where one
parent was clawed and not the other, the numbers were actually eipial,
18 chicks clawed and 18 unclawed.

186 Crossing the North arid South African Ostrich

Thus, as in the case of the loss of scales over the big toe, the evidence
is conclusive that the presence of the claw follows the Mendelian pro-
portions for heterozygotes when breeding takes place. Where the
parents are nulliplex as regards the factor no claw usually appears in
the progeny, but where a clawed individual mates with a clawless the
structure appears in practically half the progeny.

With the small proportion of clawed to clawless individuals among
both northern and southern birds it is to be expected that most of the
clawed ostriches will be heterozygous as regards the factor for the claw
for in a mixed assemblage the chances of a clawed bird mating with
another clawed one are very remote. A clawed bird will almost certainly
be the progeny of one clawed and one clawless parent, and hence will be
a simplex dominant, and when in turn mated with a non-clawed bird
will give progeny half of which are again simplex clawed and half
clawless.

Discussion.

The northern and the southern ostrich illustrate in a clear manner
how distinct species of animals may arise on the basis of germinal or
factorial changes. It may be assumed that both are the descendants of
an original stock in which the characters were all alike, and that in the
course of time alterations have taken place in the germ plasm which
give the marked differences now separating the two. Whether the
changes have any adaptive value or not will be discussed later. Con-
sidered as a whole the broad genetical conditions are fairly simple, as
no other representative of the Ratitae exists in Africa with which in-
dividual ostriches could have hybridised. As we know the two-toed
ostrich to-day, it may be assumed to have evolved entirely within the
continent, though in Tertiary times extending into Eurasia as far as
Southern India, fossil remains having been found in the Siwalik deposits.
That the two species had a common origin, and are not yet far apart,
may be inferred from the fact that they interbreed and the offspring are
fertile, also that similar degenerative changes are going on in the germ
plasm of each (parallel mutations).

Factorial Constitution. The results from the crossing of the two
species, though admittedly very incomplete, afford certain evidence as
to the present germinal constitution of the ostrich, and indicate directions
along which changes are taking place. Everything points to the dis-
tinctive characteristics of the two species as having separate factorial

J. K. DrKHDKN 1H7

representation in the ^riu phutni. Thr \mU\ {rntch Ih a tin it chanicter
which htis appeannl in the northern without any c'orreH|x)n(lin^ chan^i*
in the mmthem binl. In all the individualH nmKi for erosHin^ it iK'haveh
in a honn»zygi>us numner, showing that it haw iH'conie fully cMtabliMhe*!
throughout the northern nice. It Hupj)ortH the pri*Menc<! and absmci-
ooaoeption of genetic factors ; the factor for baldness is present in duplex
form in the genu cells of the northern ostrich but is absent from thosr
of the southern. In cross-breds of the finjt generation it is simplex
dominant, and in cross-breds of the second generation it is found to
segregate.

The dimensions and colonics of the ostrich, a,s well jis the varic^us
features of the egg, have manifestly a more complex fact<jrial representa-
tion than the bald p<itch. In the ancestral ostrich the factors for each
of the characters were doubtless common to all the individuals, but
germinal divergence has since taken place. The cross-breds serve to
establish that each species now has its own separate factor or factors
controlling its size, for those of the one are neither dominant nor re-
cessive to those of the other. They do not constitute an alternative
pair, a presence and an absence, as in the wise of the bald piitch. Hence
in the zygote resulting from the cross-mating the simplex factors for
both races are represented, the factor for the small size of the southern
bird and that for the large size of the northern. In the resulting soma
the two sets of forces are, as it were, each independently striving for
expression, and naturally the individual cannot be both sizes at the
same time, but appears a.s an intermediate, a resultant of the interaction
of two distinct tendencies, one from the northern parent and one from
the southern.

Similarly with the body colours. Each species may be deemed to
have its own distinctive factors controlling its colours, for in crosses they
do not behave as alternative pairs. In the cross-bred bird each factor
endeavours to exert its pure specific influence, the result being a com-
bination of the two, a cohjur which is intermediate and partakes of the
nature of both ; sometimes one factor may gain a slight ascendency and
sometimes another. The same conclusions as to factorial distinctness
have been arrived at with regard to the various characters of the eggs
which appear as intenuediates in the F, crosses. The exact gametic
constitution, as regards the dimensions, coloui-s, and eggs, can however
be determined only after a number of chicks of the second generation
have been reared.

The factorial representiition fnr the plumes must b«- complex to a

Journ. of Gen. vui 18

188 Grossing the North and South Africa}} Ostrich

still further degree. As already indicated, the numerous " points " in
the feather to which the farmer attaches importance behave in breeding
practice as if each were under separate control. On the other hand we
have abundant evidence that in ordinary breeding, and also in degenera-
tion, each plume usually acts as an independent whole, as if some factor
or group of factors controlled it in its entirety, irrespective of its
numerous component factors. A complexity is however introduced by
the presence of vestigial, imperfectly formed feathers, a few of which
are sometimes found beyond the last fully developed plume in a row,
and also as vestigial down on the wings and tail. Where vestiges occur
during degeneration we can only surmise that the factors which control
the whole plume may drop out piecemeal. They show that meristic
structures do not always appear or disappear in their entirety. We
may have a part of a feather as well as a whole.

Among the mixed assemblage of ostriches of the present day the
genetic factors controlling the number of wing plumes in each bird may
be regarded as heterozygous, for from any pairing we get a fluctuating
series around some mode, less wide in some cases than in others ; but if,
as seems likely, it is found possible to extract pure lines for each of the
numbers from 42 to 33 these will then breed true and may be expected
to be homozygous.

The loss of scales from the middle toe evidently represents a germinal
change which is actually in progress in both northern and southern birds
at the present time. It proceeds along parallel lines in both races, its
first manifestation being a loss of scales at the joint between the tarsus
and toe and then another over the middle of the toe. The breaks re-
present a loss of structural parts of the foot, though they are dominant
over continuity. As yet- the germinal change involved in the first break
has affected only a small proportion of birds and the second break a still
smaller proportion. The other facts of degeneration in connection with
the foot, the claw and scales over the small toe, and the loss of three toes,
are taken to justify the assumption that the breaks represent still further
degeneration which is in progress for the ostrich race as a whole. If,
when first introduced, the change is a homozygous one, there is small
likelihood that, in mating, the homozygote will meet with another homo-
zygote. Until the mutation is introduced among a considerable number
of birds the chances are that pairing will take place with a nulliplex
individual, in which case all the progeny will be heterozygotes ; these
in turn are more likely to pair with a nulliplex, and as regards the break
the offspring will be simplex dominant and nulliplex in equal numbers.

J. E. DUKRDKN 1H<»

As alretuly pn»vwi both tlu* iiorthtTii and th«* nouthorn hinln showing
the mutation Iwhuve i\» heti^rozygoU's in crosses.

Like the loss of the scales from the middle toe the loss of the claw
on the fourth Uh» is a degenerative change in pr^gn'ss within th«* ostrich
race as a whole; likewise it is germinal in its nature, and in crossi's
follows Mendelian lines. Also it can readily Ix; mlmitted iis jwirt of an
establishfHl degenerative scheme which has Ix'cn going on in the fiH)!
for a viu<t jK»riod si^eing that the first, second and fifth toes have already
di8apjK»are<l. The loss is however an individual une, not oim* affecting
the race as a whole simultaneously ; but compariMl with that ot the sailes
the absence of the claw has reached a stage where by far the majority
of the race an^ aftected. The loss <»f the claw factor may be deemed to
take place de notv> in individual biixls, and in course of time has affectiMl
largiT and larger numbei*s. In crosses the presence of the claw is
dominant over it« absence, but, from the small pro|)<>rtion of individuals
now possi^ssing one, the chances of a clawed bin! mating with another
• clawed bird must be very remoU* ; hence, as the investigations prove,
the few showing the claw are heterozygotes.

The fact that a germinal change, such as is involved in the loss oi
the claw or the scales on the toe, occui-s in individual birds anywhere
throughout the continent, strongly suggests that, whatevt'r the degeneni-
tive influence may be, it acts on the germ phism of the ostrich as a whole,
wholly irrespective of environmental conditions. It gains expression de
novo at different times in different individuals, but in the end affects all
the members of the genus, as had happened in the case of the losses
already sustained. The gradual loss of the scales, as well as that of the
claws and feathers, indicate that retrogressive evolution is taking place
piecemeal on strictly factorial lines, but in a continuous determinate
manner as regards the race. If, as seems likely, we are to regard the
baldness of the northern bird as the first step in the loss of the head
plumage then we have an instance of the retrogressive changes affecting
only one of the species (divergent mutations), while the loss of the scales
and claw is in progress for the entire race (parallel mutations).

The parallelism of the changes going on in both species of ostrich is
readily understood if we regard the two as having a common origin from
the same germ plasm, with all its inherent tendencies. Many <>f the
parallelisms of evolutionary changes found in other forms of life may
also be deemed to be indications of a distant common origin for part at
least of their germ plasm. Darwin in his ()ri(/in, \). ITH, remarks: "As
all the species of the same genus are supposed, on my theory, t«» have

18—2

190 Crossing the North and South African Ostrich

descended from a common parent, it might be expected that they would
occasionally vary in an analogous manner." All the parts of the germ
plasm may be held to be the same for any " pure " species ; some parts
are changed and give us the distinctions between species ; other parts
diflfer in producing generic separations, and still more fundamental parts
in giving ordinal, class and phylar distinctions; but corresponding
changes may take place in the germ plasm which remains common to
phylum, class, order or genus and so give rise to parallel mutations, the
Analogous Evolution of Prof. H. F. Osborn^, while divergent changes in
the common germ plasm would produce Polyphyletic Evolution. In the
ostrich specific changes have taken place, and others embracing the
genus are in progress.

Adaptive Value of Changes. We may briefly consider whether the
changes set up in the germ plasm, and expressing themselves in the soma,
have any adaptive significance in the life of the ostrich. Following
largely the teachings of Bateson and Morgan, few writers now are pre-
pared to admit that germinal changes are a response to external environ-
mental influences, or have appeared in response to some need of the
organism, or have necessarily some selection value; rather it is held
that they are dependent upon some intrinsic cause which may vary in
different cases. Though we may not know much as to the conditions
under which the changes are brought about till oncie effected and
manifested in the body, we may discuss the question of their utility or
otherwise to the individual and their influence on the evolution of the
race.

The bald head patch on the northern ostrich is probably as neutral
in its effect on the bird as it is possible for any character to be ; it is
impossible to think of it as exerting any beneficial or harmful influence
compared with the feathered condition in the southern bird. Both races
are equally successful. It should probably be regarded as the first step
in the loss of the head covering, thus introducing plumage degeneration
to this region of the bird, following upon losses which have already been
effected over the legs, wings and body and which are presumably still
in progress. In this case the baldness raises the whole question of the
adaptive significance of the loss of plumage going on in the ostrich, only
the bare facts of which can now be noticed. In chicks the outer surface
of the leg, from the knee joint to the ankle, is well covered with feathers
which practically all drop out before maturity is reached. The under

1 H. F. Osborn, The Age of Mammals, Macmillan and Co., 1910, pp. 29—34.

J. K. OlIKKDKN 191

surface of the win^ is now pnicticully ujikrcJ. Only <»ni' row of under-
ooverte pt»raist8. ami it in hanlly ever complete, while rare HurvivaU of
members of the »eci»n(l and third rows indicnte that the under Hurface
was at one time well covertMl. The win^ (Hiills and up|M'r-covertH are
many le,H8 in number in some stniins than in oth<»rH, and th<* under
covering of down for the IhkIv generally hiis all but ilisjipjx'anMl, vestigial
plumules apiKMiring only around the base of th** large feathers of the
wing and tail. Pn>bably no bird is at present so naked lus the ostrich.

It is cpiestionable however whether the loss of plumage hitherto has
any serious influenci^ on the well being of the ostrich. It is not incom-
patible with its present existence. But should the losses continue to
a much further degree the absence of a protective covering may begin
to be felt; while should the number of wing (juills become still further
lessened their inability to cover the usual quantity of eggs (12 to IG)
during incubation may aflfect the number of chicks reared.

It is likewise impossible to Jiscribe any selection or utilitarian value
to the colour differences between the northern and the southern ostrich.
The light colour of the chicks, young birds and hens of the former com-
pared with the dark colour of the latter can hardly be reganled as either
advantageous or disadvantageous. The intense scarlet nuptial coloura-
tion of the northern cock as contrasted with the restricted scarlet of its
southern rival has already been shown to give it no preference in the
eyes of its sombre mate. On the natural veld it might i>ossibly add to
its conspicuousness, supposing any greater display value on the African
plains could be thought of than that of an ordinary cock with his intense
black body plumage, white wings, white neck and light legs'.

In like manner the marked differences associated with the egg of
the two races can scarcely be deemed to have any adaptive value. The
bigger northern bird may be expected to produce a larger egg than
its smaller southern relative, but from a selection point of view nothing
in reason is to be said in favour of its rounder shape, smooth surface and
absence of pitting, in comparison with the more oval shape and pitted
surface of the other. Both are equally successful in artificial ;is well as
natural incubation.

On mechanical grounds some advantage may accrue from the loss of
the fourth toe, the claw of which has almost disappeared from the race,

* Ex-President Roosevelt in African Game Atiimals, has given much consideration
to the question of protective colouration and considers (p. 181) that " Cock ostriches
alwajs show jet black, and are visible at a greater distance than any of the common game;
the neutral tints of the hens making them far less conspicuous."

192 Crossing the North and South African Ostrich

just as the loss of the first, second and fifth toes has for long conferred
a mechanical advantage by transferring practically the whole of the leg
movements directly to the big middle toe. If however the degenerative
forces are so relentless as they appear to be and should next begin to
attack the big toe there could then be no question of the ultimate
influence of the genetic changes upon the well-being of the bird, for
with the loss of all its toes it is inconceivable that the extinction of the
ostrich would not be imminent. While the losses of the scales in the
case of the single break on the middle toe and the still rarer double
break are deemed to be the first steps in this direction, it is conceivable
that they are for the time being advantageous in the flexions and ex-
tensions of the toe.

On the whole then no evidence is forthcoming that the differences
between the northern and the southern ostrich have arisen because of
any direct utilitarian consideration; and the same can be said of the
retrogressive changes common to both. Having appeared, they may
come to have an adaptive value ; but even for this there is no support
except perhaps as regards the loss of the toes. On the other hand there
is much to indicate that, if the degenerative losses continue in the various
directions already initiated, we may look forward in the dim future to
the sad spectacle of a wingless, legless and featherless ostrich, if ex-
tinction does not supervene.

As already remarked it is impossible to resist the conviction that we
have in the ostrich some intrinsic influence, slow but continuous in its
action, which is bringing about the gradual loss in piecemeal fashion of
the various parts of the wing and the legs as well as of the plumage,
wholly irrespective of external influences or adaptive considerations.
The losses are separate mutative changes so far as the individual is con-
cerned, yet the result for the race is continuous, determinate degeneration
along several directions. If, as seems to be the case, the losses hitherto
have no adaptive value, then natural selection is in no ways concerned
with them, though it will become operative when degeneration has
proceeded so far as to interfere with the ordinary activities of the bird.
Some adaptive value may be ascribed to the loss of the three toes from
the foot, and also to that of the fourth which is in progress, yet it could
hardly be conceded when, the same degenerative tendency continuing,
the only remaining toe is attacked. It is manifest that an evolutionary
change may be advantageous up to a certain point but disastrous when
continued beyond.

The Nagelian idea that evolutionary changes have taken place as

J. K. DlTKHDKS \m

a rwult of mmu» inU^nml vitnlJHtio foixv. acting alto^rth«T in(l«'|M'mlently
of ext<?riml inHuonces. and jmK^otMlinjr alon^ drfiniU* linoH, irr«'H|K*ctiv<'
of mlaptivo (*<>n8i(lt>nitions, sonns t<> he gaining ground at the prcwiit
time among l)ioIogi8t.8'. The degeneration phenomena pn^senU'd hy the
ostrich apjH»ar t4» conatitnte as clear an example in support < if it ;us <•( mid
be addiieiH), while the gent'tical result.s se<'m to atVonI what lm,s hitherto
been lacking, namely, the cliroet application of mutation and Mendelian
principles U* continuous determinate changes, such as confront th<' com-
panitive anatomist and the palaeontologist. The main evolutionary
conception associated with mutation is fortuitous discontinuity, but in
the ostrich we |)erceive how discontinuous changes in the individual
may proceiHi along definite lines and result in determinate continuous
evolution for the race as a whole. The loss of sailes or single feathers
in individual birds may seem to be nothing more than haphaziinl chance
occurrences, but when considered for the race they indicate an onlerly
progress towards definite end results.

Establishment of Characters. If none of the changes which have
tAken place between the northern and southern ostrich have any selec-
tion value we may well enquire how the differences have actually become
established. Undoubtedly geographical isolation ;us regards North and
South Africa has played some part. Whatever intermediate forms may
be found in the intervening areas, the ostriches in the more extreme
parts of the continent must have evolved independently on one another
for long ages, though not to such a degree as to bring about infertility
between them. Some changes, such as the bald patch, and those con-
nected with the size and colour of the body and the nature of the egg,
are now distinctive between the two races, while others, such as the loss
of plumage, the loss of the claw on the small toe, and of scales on the
large toe, are common to the ostrich race as a whole.

Assuming the characteristics for the race to have been the same
originally, and that the distinguishing features of to-day have no selec-
tion value, we may first enquire how, for example, such a unit character
as the bald head patch has come to be dominant and duplex for the
northern species, while altogether absent from the southern. • On the
factorial theorj' of variation we a.ssume that some definite, hereditable
change took place in the germ plasm of the northern ostrich, a,s a result
of which the feathers fall out from the top of the head at a cerUiin age.
If we further admit that the number of ostriches for the area was con-

* C. B. Davenport, •* The Form of Evolutionary Theory that modern Genetical Research
seems to favour." Amer. Sat. Vol. l. Aug. 191G.

•194 Crossmg the North and South African Ostrich

stant, that all were equally fertile and that breeding at random took
place then a single pair of mature birds would on the average give rise
to only a single mature pair. At the beginning we may allow that the
mutation occurred in the germ plasm of, say, a single bird, and repre-
sented a double or duplex dose and was dominant. Such a bird mated
with one in which the change had not taken place would give offspring
all of which would be simplex dominants for baldness. Only two of the
progeny would reach maturity and mating with two nulliplex individuals
would give four mature birds of which two would have the factor simplex
and two would be nulliplex. Thus on the conditions postulated the
number of birds showing baldness would never increase beyond two and
both would be simplex, that is, the new character would retain the same
proportion throughout the history of the race. There would be no
swamping of the character and no increase of individuals showing it.
In the same manner if the factorial change took place in the germ
plasm of a number, x, of birds simultaneously, their influence on its in-
troduction would be to the extent of 2x, if all the progeny were simplex.
If matings took place between simplex pairs instead of between simplex
and nulliplex then the result at maturity would be one duplex dominant,
two simplex dominants and one nulliplex on the average for each two
pairs ; in other words, the mating would result for the time being in a
loss of one-fourth of the number bearing the character, but the original
number would be restored if the duplex dominant paired with a nulliplex,
for two simplex individuals would result.

Thus under the conditions stipulated — a new character of no selection
value, a stable population, free intermingling and equal fertility — con-
ditions which it must be admitted are closely approximated in the
natural life of the ostrich, the complete introduction of a new unit
character in duplex form would occur only by the germinal change
taking place as many times as there are individuals making up the
race. On the other hand the character could be introduced in a simplex
form by the change being effected in half the number of individuals.
It follows that unless a new character has some selection value it can-
not be bred into a race ; it must be introduced de novo for each homo-
zygous increase and half the number of times for the heterozygous
increases.

As regards the bald patch therefore the germinal change must have
been effected as many times as there are individuals making up the
northern race, for the experiments have proved they are all homozygotes-
We can scarcely conceive that the alteration would be carried out simul-

.1. K. 1)! KKDKN Ift.*)

taiieouHly in nil the imli\ idualH. |)i*OHumubly it atVccli'd u W'w at Hrhl umi
otheni gnuiually. For a lon^ time in the history of th«« race noine of
the binis wouM be duplex for Imldness, some would be Miniplex and the
r(>8t would Ih' nulliplex. Mr (J. W. Manly (Punnttt's MendvUsm, IfUl,
\\ 136) haw shown that under eon<litions such lus are stipulate*! for the
tiatrich the jnipulation would rapidly fall into a stablr condition with
reganl to the pro|)ortion of the three forms, whatever may be tht* pn»-
portion to start with. If the |)opulation consists of p h(nnozy^ot<'s of
one kind, v homozy^tes of the othrr kind and Sry heterozygotes then
he points out that such a population would be in eipiilibriuni for a
particular fact-or so long as the condition (f = pr is fulfilled. The pro-
portions which satisfy the e(|uation are exceedingly numerous and in
case of any disturbance of the equilibrium, a.s by the app(;aranc<' (fe novo
of the charact<?r, it will be restored after a single generation.

From the foregoing we gjither how little etTect undi'r natunil con-
ditions the importation of the hundred and thirty-two northern birds
would have upon the southern race if the population of the latter were
stationary. If all the northern birds reached breeding age and mated
with the southern they would in the end give rise to only double the
number of simplex, bald-headed ostriches, and the number would neither
increase nor decrease. How far the other characters — dimensions, colour
and nature of the egg — would influence the southern race cannot b<!
determined until their actual factorial values have been worked out.

The conditions represented by the loss of scales from the middle toe
and the claw on the fourth toe further help in an understanding of how
mutative changes are introduced into a large assemblage. It is seen
that the loss of scales has occurred in individuals of both the northern
and southern birds, though more frequently in the former. The change
however must have appeared independently north and south, for the
distinctness of the other characters, especially that of baldness, proves
that no intermingling of the two species has taken place towards the
extremities of the continent. Appearing at first in comparatively few
individuals, and presumably in duplex dominant form, it is most unlikely
that a bird having the break would mate with another showing the same
loss, but rather with one in which the scutellation was continuous. The
duplex bird mating with a nulliplex would give F^ progeny in all of which
the break would appear, while germinally they would be simplex. The
simplex condition would tend to be retained until such time as the char-
acter became prevalent and opportunity occurred for a simplex to meet
with a simplex, when a duplex condition would arise. It is significant

196 Crossing the North and South Africmi Ostrich

however that in all cases where an ostrich showing the break has
been paired with one in which it is wanting, the bird has proved itself
to be heterozygous, giving progeny of which approximately half display
the break and half the continuous scutellation. Thus while baldness
is a mutation fully established for the northern race and germinally
duplex, the loss of scales from the third toe is a mutation only partly
established and germinally simplex. A character in course of intro-
duction within a race will for a long time be mainly in a simplex or
heterozygous form ; later, as the mutation appears de novo in more
and more individuals, the population will tend to consist of duplex,
simplex and nulliplex birds, until in the end all will be duplex or
homozygous.

As regards the race as a whole the claw on the fourth toe reveals
conditions somewhat similar to those of the loss of scales over the middle
toe, but is a character which has almost disappeared. Experiments
have shown that the presence of the claw is dominant over its absence,
and matings with a nulliplex individual give progeny half simplex
dominant and half nulliplex, proving that the clawed individuals are
heterozj^gotes. This again is what would be expected considering the
small proportion of clawed birds, and the remote likelihood that a hetero-
zygote would mate with a heterozygote. If only heterozygous individuals
are to be found then any further loss of the factor will presumably take
place as a simplex, and to be completely lost to the race the change
must take place as many times as there are heterozygous birds. As
for the introduction of a new character so for the loss of an existing
character, it cannot be bred out under the conditions postulated, but
must drop out germinally.

Specific distinctness of Northern and Southern Ostrich. Whether the
northern and the southern ostrich are to be regarded as separate species,
or only as sub-species or varieties of a single species, raises the ever-
recurring, but undefinable question as to what constitutes a species. In
the foregoing we have available all the data which the systematist could
possibly desire to enable him to reach a decision. A germinal character,
baldness, occurs in one, but is wanting in the other, while the dimen-
sions and colours of the body as well as certain features of the egg are
also distinctive and germinal. The characters are retained when the
members of one race are subjected to the same environmental conditions
as the other, showing they are not dependent upon external circum-
stances. They can all be regarded as distinct elementary characters in

.1. K. DrRKDKN 197

the IK' Vrieaian m»nm», iiml the combiniitionH ini^ht well Im* hM Ut
warrant us in regnnling the two w* HfK'cifiailly cliHtinct.

On tho other hand the birds are provi'd to interbrecKl freely, and the
offspring an» fertile, both inter se and with the |)arenUil foniiH. Thj*
fact that similar degenenUive processes — loss of pluiiuige, scales and
claws — an* pnK^ee<iing in both also jx)ints to a close grmiinal relation-
ship. In the opinion of many systcmatists the jihysiological fact of
fertility alone wotdd Iw deemed to justify specific unity.

The ostrich nmges over all the habitable j)arts of Africa and there is
everj' likelihood that in intermediate areas between north and south a
mingling of the two races goes on, producing a mixed po[)ulation, com-
posed of all possible combinations of the two sets of chanicters. Thus
in the East African Ostrich, S. massaicus, as the writer has found in
visiting the ostrich farms in British East Africa, the colour of the hen
and immature cock is a cream yellow while the mature cock hjus the
head, neck and legs scarlet, and the birds are sijmewhat larger than the
southern. The bald patch is present and the eggs are pitted. The
Somali Ostrich, <S^. molyhdophanes, is described jvs a smaller, darker bird
than the southern, but the bald j^atch is wanting, and the colouration is
like that of the southern and the eggs are pitted.

If the entire ostrich population of Africa were gathered together we
are probably justified in thinking that all intermediate forms would be
forthcoming between typical northern and southern birds. An excep-
tion would occur however in the case of the bald patch, for however much
inter-crossing had taken place the character would never be intermediate,
but would be wholly present or absent; and though the dimensions,
colours and eg^ characters appeared in varying intermediate degrees as
a result of crossing we should still have the knowledge that their distinc-
tive nature could be extracted by selective breeding. When discussing
intermediates the possibility of segregation should always be borne in
mind. In the present instance, where all the facts are known, inter-
mediate forms grading from one species to the other have no direct
bearing on the question of specific distinctness. Among the African
fauna especially, experimental breeding would probably establish that
many so-called species and sub-species, often founded upon one or a few
specimens, are in reality intermediates or hybrids of other s2>ecies.

So long as we have the facts before us it is of small moment whether
we regard the northern and southern ostriches as distinct species or
not. It becomes a matter of individual predilection whether greater
importance should be given to somatic differences or to physiological

198 Ci'ossing the North mid South African Ostrich

similarity. Without being biassed in either direction it appears to the
writer to make for convenience to regard them as distinct under the
names bestowed by Linnaeus and Gurney.

DESCRIPTION OF PLATE VII.

Fig. 1. Outer surface of fore-wing of ostrich, with plumes clipped off, to show arrange-
ment of wing quills on upper (post-axial) border and rows of upper-coverts.
The claw on the bastard wing is not visible.

Fig. 2. Under surface of fore-wing of ostrich which is naked except for the single, incom-
plete row of under- coverts.

Fig. 3. Egg of North African ostrich (to the left) and South African ostrich (to the
right).

JOURNAL OF GENETICS. VOL VIM. NO. 3

PLATE VII

DOUBLE FLOWERS AND SEX-LINKAOE
IN BEGONIA.

By W. BATKSON, M.A., F.R.S., and IDA SUrrOX,
Student in the John Innes Horticultural Instxtution.

(With Plate VIll.)

Begonias are monoecious plants, having thrir flowers armngcd in
axillary cymes. In normal plants the flower which terminates each
dichasium is a male ; and, in the simplest arrangement, njwn either side
of this stands a female. For one or both of these females may b<'
substituted a continuation of the inflorescence, which again at each
dichasium ends in a male, this system being indefinitely repe^ited.

Since doubleness affects only those flowers which stiind terminally,
being that is to siiy in normal plants males, an investigation of the
inheritance of this condition offered attractions, as being not unlikely
to throw light on the genetics of sex. In passing it may be remarked
that since a female flower can be replaced by an inflorescence, wherejus
a male flower is not thus replaceable, from these morphological relation-
ships we are led to infer that the female flower contains something that
the male hjis lost. The male flower may be thus compared to a reces-
sive, dropped out of the inflorescence which can be produced further in
the heterozygous state.

The investigation wjis begun in 1908 by fertilising the normal female
flowers of double Begonias with pollen from singles of unknown origin.
Subsequently further crosses were made between doubles and a horti-
cultural strain of singles which was declared to have bred true for some
generations. The results have been full of complications such that,
after many years work, it has become evident that no simple factorial
scheme is f<jllowed, and that segregation in regard to single and double
flowers must in these plants be a process liable to considerabhj irregu-
larity. In general the single is a dominant as Bond also found. T\\v
recessive doubleness reappears in F^y but the numerical proportion of
F^ doubles is low and fluctuates widely. There are many transitional
forms, which render accurate classification and enumeration impossible,

200 Double Flotvers and Sex-Linkage m Begonia

and not very rarely several of them may appear on the same plant.
Some of the more interesting of these forms will be spoken of later.

An average of many i'^a families gives about 1 double in 32, but in
several large families no doubles at all appeared, and this average has
certainly no general significance. From F^ plants crossed back reci-
procally with various doubles, similar irregular numbers were obtained
and no approximation to analysis could be made. To render the com-
position of these families intelligible lengthy descriptions would be
required and little purpose would be served by the publication of such
details. Their interest lies chiefly in their value as an indication that
in regard to a character which in so many plants is distributed geneti-
cally according to strict allelomorphic rules, great irregularity may
elsewhere prevail. Whether this irregularity is in any way connected
with the monoecious structure of Begonias cannot of course be declared.
Such a conclusion is by no means improbable.

The purpose of the present paper is to make known a curious dis-
covery which resulted when Begonia Davisii was brought into the series
of experiments. The plant is one originally found in Peru by Mr Davis,
collector for Messrs Veitch, and first flowered by them in the year 1876 \
Inasmuch as this is a real species, breeding perfectly true on self-
fertilisation, it seemed suitable for use as a reliable single for crossing
with doubles. When however these crosses were made it was found
that any double fertilised by pollen of B. Davisii gives onli/ double-
flowered offspring — 405 plants have been thus raised, and of these only
18 are recorded as having less than complete doubling. The male
side of Davisii is therefore exclusively double-bearing. Since the
same plant fertilised with its own pollen gives only singles, the
female side must be inferred to be exclusively single. Tested however
with the pollen of a double, it gave a result which we cannot satis-
factorily interpret. Fertilisation with pollen of doubles cannot always
be accomplished, since thoroughly petalodic flowers do not produce
pollen. A good many doubles nevertheless when starved or poorly
grown do produce anthers and pollen, as for example the well-known
double called in horticulture Begonia Lloydii. B. Davisii $ fertilised
by Lloydii j' gave 72 thorough singles and 42 with traces of petalody,
a condition we have not yet seen in Davisii itself The genetic nature
of these slightly petalodic plants is not clear. If they can be formed
when the pollen of Lloydii is used, we should expect them to appear
when Davisii is fertilised with its own pollen, for this pollen used on
^ See Wynne, The Tuberous Begonia, 1888, p. 10 and Bot. Mag., t. 6252.

W. Batkhon and Ida Sutton 201

doubles gives scarcely anything but extreuu* doublcB. Slightly ]>c*tal(xlic
plants came also oocanionally, among largt> nunilM^ni of HJnglcH, in fauiili<«
raised from DavUii female fcTtilimnl with |>ollc'n of heterozygouH plantn
{Ft from double x ningle (f). Whon thi» jMillon of Daviini in uwhI on
such Fi plants the proportion' of recesMive, doublf-bi'uring, ova, of coiirmj
appears ; and since })erfectly reliable {wllen of doubles is ditticult to
obtain, the {)ollen of Davisii may be substituti'd for it.

AIUt discovering the |x»culiar genetic constitution of l)av\mi we
naturally expt»cted that the results of reciprocal crosses mjule between
doubles and F, plants (from double x single) would at least 8*>metimes
show linkage of doubleness with either the male or the female side.
For this investigation a considemble amount of material is now available
and we are satisfied that in general heterozygotes do not show any
regular phenomenon of this kind. In contrast however to the usual
absence* of consistent sex-linkage, one plant raised from the female side
of Davisii fertilised by Lloydii was provcKl U) possess such sex -linkage,
though less complete than that of Davisii itself. The plant, self- ferti-
lised, set badly and only 2 plants (singles) were thus raised. As regards
its female side we have the evidence that with Davisii jwUen it gave
11 singles, and with Lloydii pollen 5 singles and 1 slightly petahxiic,
from which it may be inferred that the ovules were at all events
predominantly single-bearing. The male side tested on Lloydii gave
27 doubles, 14 half doubles and 5 slightly petalodic (see Nos. 23 — 26 in
Table on p. 20(5).

As to the presence of sex-linkage in other heterozygous individuals
the evidence is as yet conflicting. Some plants show it, whereas others
do not, and we cannot as yet perceive any circumstance either in the
way in which the plants were made up or in any other resjiect which
accounts for these differences. We give in the Table (Nos. 31 to 45
on p. 207) specimens of these various behaviours.

The case naturally recalls other examples in which sex-linkage has
been observed in plants. In three of these the male side has been
specially distinguished as being associated with the recessives, though
whether this is an accidental circumstance due to the way in which the
plants were originally bred cannot yet be declared, but in Petunia, as
shown by Miss Saunders singleness, the dominant, was carried by all the
poll en -grains, and by some only of the ovules of the single-tlowered plants.
In Matthiola the pollen was all double and for the most jxirt wirrying
cream plastid-cr>lour (Saunders); and in a plant of Campanida carpatica
' This pro|>ortion, as Table II exemplifies, is appan'Utlv quite irregular.

202 Double Flowers and Sex-Linkage in Begonia

and its descendants, the pollen bears white flower colour and femaleness,
the factors for blue and for the hermaphrodite condition being carried
by the ovules (Pellew). In this case as in Begonia the sex-linkage was
not general but special to a particular plant and its descendants. Of
these examples the plastid-colour is the only one in which the converse
combination has yet been built up, though perhaps the others may
hereafter be obtained ^

The condition in Oenothera " velutina " described by de Vries must
be very similar, the recessive dwarf character being carried by the
pollen. In the corresponding case of Oenothera " laeta" the evidence
also points to the pollen being all dwarf, and to the existence of a mix-
ture of tails and dwarfs among the ovules, in spite of which the plants
do not throw dwarfs on self- fertilisation. This absence of dwarfs on
selfing constitutes a puzzle exactly like that of the presence of slightly
petalodics in Davisii x double </ ^

When in hermaphrodite flowers the male and female sides are
genetically distinct we feel fairly sure that the segregation of these
allelomorphs occurs not later than the formation of the anther-rudiments,
but in B. Davisii it presumably happens even earlier and not later than
the formation of the male flowers. Those who incline to regard the
reduction division as the stage at which alone segregation can be effected
may no doubt be tempted to suggest that in B. Davisii, for instance,
pollen grains bearing the dominant factor are in reality formed but in
some unexplained way fail to take part in fertilisation. As a mere
suggestion of a possibility that theory cannot as yet be absolutely
excluded, but in this special example it is more than usually difficult to
accept, since the pollen of B. Davisii is to the eye exceedingly uniform
and regular. There are none of the shrivelled grains which are gene-
rally looked upon as the bearers of missing elements. Though less
significant, the absence of seeds partially defective is also noticeable.

In applying the term sex-linkage to such cases as this I am following

1 Since in the original form the ovules were mixed and the pollen was all recessive,
the " converse " might appear in one of two forms. Either (1) the ovules might be all
dominant and the pollen mixed ; or (2) the ovules might be mixed and the pollen all
dominant. As Miss Saunders's plants were tested by self-fertilisation and not by crosses
with recessives it cannot yet be declared which of the above possible constitutions they
possessed, but she considers there is a presumption that they were really arranged on the
second of the two plans. (See Saunders, Jour. Gen. iv. pp. 332 and 359 and compare
Pellew, Jour. Gen. vi. p. 320, &c.)

2 For a discussion of these Oenothera cases see W. Bateson, Problems of Genetics,
1913, p. 113.

W. Hatkson and Ida Sutton '2i):\

the suj(j^»8tion injuio l)y Miss PclKw in her disc^usHion of " TyjX'H of
Sogn^^itioii'." Tho |m>|)not.y nl' the coiniKirison botwccn the jihs* ►elation
of ft chanict'er with ono of tho sexes in the aise of a herinnphriKlit^' plant
luui the phenomenon in bisexual animals commonly calle(| sex-linkage
may In* (jiiestiontMl. hut until we know more preeis<*ly how s<'X in animals
is rolatini t^» the phiMiomena in the flowering plant, no unjustiHabh;
Assumption is made an<l no si'rious confusion can hn aiused by their
use. If, following om* methrnl of interpretation, we regard poll(;n-mother
colls, IkMug the latest diploid stage, as the equivalent of male animals,
we can rcivsonabiy speak of thi' character — here doubleness — carried by
the j>ollen-gniins, jus linked with maleness, and singleness as linkc^i
witli femaleness. The comparison, though not certainly valid is at
present defensible. The relation of the hermaphrodite to the dioecious
condition, whether in animals or in plants h;is not yet been represented
by any factorial scheme which is thoroughly satisfactory. On a survey
of the various sexual arrangements followed among plants we meet a
difficulty in attempting to choose any fixed moment common to all the
cycles, which can serve as a starting point for the institution of homo-
logies. The difficulty is intensified when we proceed to the case of
animals. One obvious suggestion is that the reduction-division pro-
vides such a common fixed point. Though I am not disposed to look
upon that event iis the only occasion on which Mendelian segregation is
effi^cted, I readily agree that many segregations presumably do happen
then, especially that by which sex is usually determined among animals.
Such observations however as those of the Marchals and the new evidence
discovered by Collins- show almost beyond question that even within
the group of Mosses sex-segregation may occur at different moments in
the different cycles.

With equal propriety we may regard the actual gametes as the fixed
point common to all and therefore hgniologous in all the cycles, but we
have still to face the difficulty that such a critical segregation as that
which determines sex (and probably others) may be sometimes effected
at the reduction-division, sometimes before it, as at least in monoecious
flowering plants, and sometimes after it as in Collins's Funaria.

The facts practically drive us to the conception that the ordinal
|X>sition of the reduction-division can be shifted in the cycle, and that
segregations which in some cycles precede reduction are in other cycles

' Jour. Gen. 1917, vi. p. 310.

- In the present number of Jour. Gen.

Journ. of Gen. viii 14

204 Double Flowers and Sex-Linkage in Begonia

postponed until reduction has been already undergone. The problem
is not unlike that so often raised by the differentiation of parts com-
posing a meristic series. In one Lizard the nth vertebra carries the
pelvis and undergoes special modification. In another Lizard the ver-
tebra thus differentiated is the n 4- mth in ordinal series. Morphologists
have long discussed whether in allotting homologies among vertebrae
we should be guided by the differentiations, or by the ordinal positions.
When once the true nature of segregation and differentiation is under-
stood the question is seen to lose all significance \ and having no precise
meaning is incapable of being answered. For the individuality of the
segments is not respected or maintained in variation, nor are differ-
entiation and nuQierical change necessarily interdependent. We may
easily satisfy ourselves that the numbers may vary and that within con-
siderable though unascertained limits the functions and differentiations
of the segments may be redistributed. I can scarcely doubt that we
must similarly interpret the series of divisions and differentiations of
which the life-cycles consist.

In the Tables we represent the plants as of five classes. Singles are
those in which the male flowers have not been seen to have more than
the four normal petals. The slightly petalodic class have generally one
or two, though occasionally rather more extra petals or petalodic anthers.
These two classes cannot be quite strictly instituted, and plants having
flowers of both kinds are common. The half-double class ranges from
the slightly petalodic to the really double, but nevertheless it is a fairly
uniform class. Doubles and full doubles are not essentially distinct, but
the termfidl was applied only to flowers in which the petals were very
numerous and close.

As was stated above, peculiar and transitional forms are common.
In particular some difficulty is caused by structures consisting of female
and male flowers imperfectly resolved from each other^. Such flowers
can generally be recognized by examination of the bracts, but when
this condition of imperfect resolution is combined with some degree
of petalody the degree of doubling cannot be determined with much
confidence.

Since double flowers stand terminally, that is to say in the male
position, we supposed at the beginning of these experiments that double
flowers were necessarily petalodic males. Happening however to examine

^ See Problems of Genetics, p. 66.

2 Noticed by Bond, Jour. Gen. iv. 1915, p. 341.

W. Batkson and Ida Sutton 205

the varii'ty vhIUhI (imf Zoppoliii, wo wrrr Htruck by tin* fact that, the
double rtowers, though UTiiiinal, an* in rrahty moiWiinl /ctiKiles. Th»Te
is no inferior ovary, but at the bases of the jxitalH are inJUHses of exjK)He<i
ovules'. This arningeinent is normal for the variety and gives it a
iiuwt chanietvristic apjH'arance. Further search among double Ht-gonijis
8howo<i that many aiv in essentially the same condition, though the
amount of ovules develojK'd varies greatly. Probably most of i\w fine
exhibition blooms are nunlified fi'male flowers, though in them the;
ovular tisstie may be riMlucetl to a mere trace at the bjise of occasional
peUils.

Whether any of these plants are altogether inculpable of prfnlucing
anthers, however much they may be starved, we do not know. Our
experience inclines us to think that some plants cannot produce anthers,
though we have cert^iinly seen thoroughly double Howers of the ovule-
containing kind on plants which had borne double males containing
anthei*s. But apart from this cjuestion we CJin easily recognize a chuss
of doubles, of which Lloydii is a good instance, in which the double
flower is essentially male; and though they may be fairly perfect doubles
when well grown, this kind of double can readily be starved into pro-
ducing pollen. The view that plants, e.g. Graf Zeppelin, in which the
terminal flowers are female, instead of male as normally, may be homo-
zygous females is rather attractive, but we see no means of testing it ;
nor if such an idea could be entertained, would it at all account for the
fact that in a full double which must certainly be accepted as a reces-
sive, homozygous in doubleness, the normal female flowers standing in
the latei-al positions are single. Beyond this point we see as yet no
means of pursuing the analysis.

Since B. Davisii is a genuine wild species and bears exclusively
single flowers, the conclusion to which our observations have led us,
namely that its male side is genetically all double, seems not a little
remarkable.

' Flowers having this structure were referred to by Wynne, I.e., p. 13, and parts of
them are figured by Bond, Jour. Gen. iv. PI. XVI. Their morphology is obscure, but
it seems natural to regard the carpellary walls as represented by a mass of petals. We
have never seen a normal female standing in the male position.

14—2

206 Double Flowers and Sex-Linkage in Begonia

Details of Experiments relating to Begonia Da visil

Eeg. No.

4

49/13

5

20/14

6

30/15

7

28/13

8

4/14

9

30/18

10

22/14

11

25/14

12

27/15

13

85/18

14

10/17

15

4/17

16

5/17

17

15/15

18

41/16

19

3/17

20

42/16

21

2/17

22

36/16

Davisii selfed, 4 families, 200—300 raised all true to type
Lloydii selfed, 45 true

Various doubles J ertilised hy Davisii $ .

Single

Lloydii X Davisii <} ... ... —

Graf Zeppelin X do. ... ... —

49/13 (as above) double x do. —

20^/14 (as above) double x do. —

A double x do. —

28/13 (as above) double x do. —

A double x do. ... ... —

Argus xdo —

Hollyhock X do — —

Louis Boucher X do. ... — —

Fleur de Chrysantheme x do. — —

Davisii ? fertilised hy double.

Davisii 2 x Lloydii s ... 72 42

Reciprocal crosses with a half double.

Davisii x 3V14 half double ... 39 15

Slightly

Half

Fully

petalodic

double

Double

Double

■

32

28

6

3

16

3

14

139

—

9

8

78
13
17

z

3*'/14 half double x Davisii
The same half double 3^/14
selfed

1?

.3

28

2

49

17

Reciprocal crosses with a heterozygous single.
105 1 —

68 14 5

42 4 —

Davisii ? x 22i/14 hetero- )
zygous single ]

221/14 heterozygous single

X Davisii s
The same heterozygous )
single 2^1/14 selfed j

22oyi4 heterozygous single

X Davisii s
The same 2^0/14 selfed

67
45

9
10

17

34
1

109

8

23

24
1

The following are tests of two plants bred in Experiment No. .14,
Davisii $ x Lloydii ,}. In the first group 10717, a single, was used :
in the second group 10717, a slightly petalodic, was used. In both, the
male side proved to be predominantly double.

23
24
25
26

Reg. No.
12/18
13/18
14/18
15/18

20/18
22/18
23/18
24/18

102/17 X DamsM c? ...

Davisii^ x 102/17

lO'^/n X Lloydii

Lloydii xW^in

Single

11

6

5

Slightly
petalodic

1
5

Half
double

14

1
5

1

Double
14

Fully
double

13

27

28
29
30

lO^inx Davisii s

Do. X Lloydii d

Lloydii xlOy 17

Graf Zeppelin (fully double) |
X 10^/17 )

14
2
1

8
2
1

11

JOURNAL OF GENETICS, VOL VIM. NO. 3

PLATE VIII

The flowers and leaves of liequuid t>ui isi

W. IUtkson and Ida Sitttdn

20:

Kxpenmeuts illujitrattng behamour of vanoua heteriKygoxui giuf/les
(.33717,17717, 16717,84717).

In 33717 tho sin^Ir factor wnit in frnin the iiioIIht's si<l«', ami tlnT«*
is a dour iiHlicalion that the jhiIIi'M was pndoininantly douhh'; but in
17717 and IGVn, similarly bnnl, tht' pollon w;is prrcloniinantly Hin^^l**.
In .'UM7, the «»xact nripHKNil of 33*/17, tho sin^'lc factor wmt in from
tho father's side, and the imnd)ei*s though insuthcicnt, do not .suggest
8OX -linkage.

31 71/18 33»/17 Relfoil

82 7*2/18 Do. X 7'/ 17 double ^

83 73/18 3.')8/17 double x 33V17 <r

SliKlitlv Mftif Fiilh

SiiiKlo iK^UhxIic lioulilc Doiihli! (tutihl

45

43
31

7
25

0
27

Plate VIII shows the flowers and leaves of Begonia Davisii.

43

84

53/18

17'/17 xLW/'i<f

18

85

54/18

Lloudiif x'l7V17 ...

4

2

—

—

—

86

56/18

.S,5«/17 double X do

4

—

1

1

3

37

67/18
34/18

35"/17 double X do

41

8

2

2

7

38

16-Pelfed

27

3«)

35/18

Do. X Hoydii

55

17

—

11

17

40

36/18

Lloydiixl6'^in

88

24

—

3

3

41

37/18

Gmf Zeppelin X do

.: 50

7

1

—

4

42

38/18
74/18

35'*V 17 double X do

55

23

15

14

43

34V17 selfed

73

4

1

44

76/18

Do. X 353/17 double ...

47

16

11

16

45

75/18

35«/17 double x 34V17

0

2

3

—

1

VoLUMK VITl SKITKMHFIJ. 1019 N(». 4

THE INHKHITANCK OF WlNi; COLOUR
IN LKlMDOPTKIiA.

I. AFWAXAS GROSSULARIATA VAR. LUTKA ((V)ckkrell).

By II. ONSLOW.

(With PlaU-s IX and X. TahK' aiul twciity-Hvc Tc.xt-Hgurcs.)

CONTENTS.

PAOK

I. Introduction 200

II. The "TinU.mctor" 211

The colour scale 212

III. Method of presenting results 214

{a) Curves showing the distributions of the colour-values 214

(6) The frequency distributions ...... 215

(f) Experimental and statistical errors 216

(rf) Colour of hind wings, and effects of desaturation and sex . 218

{r) Table of families, and pedigree 219

IV. Description of certain varieties . . . . * . . 220

(a) The melanic series 221

(6) The distribution of the black pigment in suffused varieties 223

V. The yellow pigment 224

The distribution of the pigment in scales of insects of varying

shades of yellow 226

VI. The original material 227

VII. Breeding results 228

(a) Var. lutea x A. grossulariata ...... 228

(6) Var. lutea x var. lutea 2.30

(r) Hybrid (/u/fa x^ro««.) X var. /M/«a 231

(d) Yiyhri^ {lutea X gross.) >^hy\}x'\di [lutea y. gross.) . . 235

(e) Hyhrid {lutea X gross.) X A. grossulariata .... 239
VIII. Summary 239

I. Introduction.

Among the Heteroccra varioties which exhibit changes both of

colour intensity and colour (quality are not unconiiiion. The former

may be broadly classed as melanic varieties. The latter, which are less

Joorn. of Gen. viii 15

210 The Inheritance of Wing Colour in Lepidoptera

frequent, often show a change from red to yellow, or from yellow to
white, the ground colour alone being affected. I am at present investi-
gating several examples of both these varieties.

The present communication deals with the yellow variety of Abraxas
grossulariata, in which the markings are normal, and bright orange or
yellow replaces the usual white ground. Little or nothing is known of
the pigments involved in such changes, or indeed of any insect pigments,
with the single exception of the white and yellow colours of the Pieridae,
which Hopkins^ has shown to be due respectively to uric acid, and to
some unknown reduction product of the same acid. It is possible that
a similar relationship exists between the yellow and white pigments of
A. grossulariata ; and if this be so, an enzyme may be found to control
the changes in both cases. I hope to be able to undertake a chemical
study of the pigments at an early date.

The experiments were undertaken in order to investigate the genetic
relationship of the white and yellow ground colours in crosses between
A. grossulariata and its variety lutea. The type insect is usually
papery white (see Plate IX, Nos. 25 and 26), or sometimes a very pale
shade of cream ; and the depth of the yellow ground in lutea varies
considerably. Preliminary crosses showed that the white colour of the
type insect was only completely dominant if the yellow of the variety
was pale; and it became evident very soon after the experiments
were commenced^ that this white colour did not behave as a simple
Mendelian character. It is in fact an example of a character which
varies more or less continuously; and such cases frequently present
problems for which it is difficult to find a satisfactory explanation. As
a rule, the F^ heterozygote from lutea x grossulariata is not white, but
can be distinguished from the type by a tinge of yellow which
occasionally reaches an appreciable depth. The total range of variation
in all crosses is considerable, and extends from the papery white of the
type insect, through the palest shades of lemon, to a bright reddish
orange (see Plate IX, Nos. 1 — 26).

Since the material could not at once be divided into discontinuous
classes, several attempts were made to grade the insects and to place
them by inspection in four arbitrary groups. This method had to be
abandoned as no reliability whatever could be placed in the judgments,
even when these were made under similar conditions of lighting, etc.
An instrument of some kind for determining the colour appeared
essential, and several were examined for this purpose. The colour-
. 1 F. G. Hopkins, Phil. Trans. Vol. 186, Part ii. p. 661, 1895.

H. ()N8L0W 211

wheel, notwithstJiiKiing its inrchjinirnl disjulvanUi^en. h«.H imich in itj<
favour, hut was Htially alMitxItiiifd tor a connnrn'ial instruiiHiit railed
the " Tint-oinetvr." TUv n-iuson ftir this srK'ctinu was that with th«'
colour-wheel tht»n' is ua i\\vihiH\ n( di'tiiiin^ nr rfconlin^ thr (olours «if
the (lises employed: it is thoreforr iui|H»ssihlr for any fulun- nhs.rvrr
to n»produe«» these discs, and eonsefpiently the colours of any readings
taken with them. The " Tintonu'trr " however supplies a unit, whieh
though arhitnu'v. is recoverahle and sjitisties the other essentials of a
standanl. Also, it is placed on the market at a UKKleratc^ price. The
scale consistnS of a series of coIouhmI gbtsses, carefully <lyed and stan-
danlistnl by comparison with the ^busses of other scales, so that sev<'ral
opjHirt unities for ern)r are introducted. For this rcjuson it must be
mlmittinl that the eolniir-wlK-el would he prefenvhle, since with it. there
is hut one judi^ment to introduce I'rror. Unfoit unately. !i(» series of
sU'imlard colour discs can he pnK'ured. They should be based on some
physical constant such jus wave length, so that they could be che-cked
eiisily in avse (»f fading.

It may be added that sometimes the colour-wheel can be used with
advantage cond)ined with some simi)le optical arrangement such as is
provideti by the "Tintometer," to secure conditions of eipuil illumination.

11. The "Tintometer'."

The "Tintometer" (Fig. 1) consists essentially of a rectangular tube
B, slightly tapered and about 10 inches long. At the narrow end there
is an eye-piece A : at the other end there are two apertures which
admit light. The tube is mounted on a base to which it is inclined at
an angle of about 45''. Just above the apertures there are two rows of
grooved slots G, which receive the graded standard slips of coloured
glass F, for intercepting the beams of light before they reach the eye.

The apparatus is used tis follows: a piece of mirror is put imme-
diately under the two apertures and the instrument placed in diffused
daylight, preferably from a north window. The instrument is now
moved until both fields of view are equally illuminated ; all objects such
as window-sjishes, trees, etc., being ayoided. The unknown coloured
object is then phiced under one aperture, and the specially preparefl
white background C, made of firmly compressed plaster of Paris, under

^ Further details for the use of this instrument for other purposes are to be found in
Mecuiirement of Light and Colour Setu<ntiom (George Gill and Sons) and Light and Colour
Theories (E. & F. N. Spon, Ltd.) by J. W. Lovibond. the inventor.

15—2

212 The Inheritance of Wing Colour in Lepidoptera

the other. The colour slips F are then placed in the grooves until an
equivalence of colour has been obtained in both fields of view. The
coloured object may of course be placed indifferently on either side
without affecting the result. In making the measurements here re-
corded a low power lens B, magnifying about three diameters, was

A

Fig. 1. The "Tintometer " arranged for reading the colour of an insect,

A. Eye-piece. B. Tube. C. Compressed plaster of Paris background. D. Lens.
E. Cardboard diaphragms. F. Standard coloured glasses. G. Slots to hold glasses.

placed between the insect and one aperture. A variety of different
sized diaphragms were also cut in black card. These were placed in the
metal slots E in order to mask the black markings of the insect, which
might otherwise have interfered with the colour determinations by
introducing a contrast effect. A second diaphragm of the same size
was placed on the opposite side from the insect in order to equalise the
two beams of light.

The colour scale.

The colours of the glass slips used are yellow, red and blue. The
scale consists of twenty units, each divided into ten parts, every one of
these parts being again divided into a further ten fractions. It is
however only with the paler colours that an increment of 0*01-0'05
becomes appreciable. These three colours can be combined to form any
other colour required. Thus, one unit of red + one unit of yellow = one

H. Onslow 213

unit of onuiffi" ; and a yolli>w-unin^t' colour may ha uuuiv by ct)iiibiiiing
one unit of nil with twt) or more uniU* of yt'llow. Further, one unit of
nnl + one unit of yellow + »)IK' unit of bhu* = one unit of neutral tint, or
black.

It is affiruKil by tho nuikers that the thn'r c«)|our-uiiits whi<!h ^o U>
make up the neutral tint are e<|ual, that is Ut nny, when time ditVrrent
eolourtnl units an* conibiniHl then* is n<» residual colour. All units and
fractions ai*e siiid to be cheeked by this test., and, further, all the glasses
of one set are interchanged with those of another and the units verified
by CHKss checking. To enable obsi'rvers to verify their scales the makers
publish colour readings, obUiined by dissolving a known weight of a
pure substance such jus |M)tjissium-ferricyanide in a giv«'n voIuiih* of
water, a known depth of which can bi' examined in cells of ditt'erent
thicknessi'S. The only objection to this procedure is that even s(j|utions
of the most stable substances such jus picric acid are com|j{iratively
inconstant, and may show either a fading, or an increiuse of colour on
standing.

In tuxler to obtain the colour measurement of a given in.sect it
should be pinned upon a strip ot cork and su placed that the desire(J
portion of the wing, magnified by the lens, comes immediately under
one aperture. A suiUible diaphragm is then chosen to cut out most of
the black markings, while leaving exposed the central yellow portion of
one wing. A red glass slip of a certain value is next chosen, and this
is combined with a yellow slip, representing about twice as many units.
Such a combination will give a bright orange which may be either too
intense or too dilute. By the selection of suitable units or fractions
both the red and the yellow are alternately increased and decreased,
till a colour exactly matching the wing is obtained. Care should be
taken to limit the length of each observation to five seconds, in order to
avoid the disturbing effects due to fatigue. It is often found impossible
to obtain a perfect match with the red and yellow colours only, owing
to the fact that the colour of the wing is desjiturated by the addition of
black. In this event the exact colour e(jui valence may be obtained by
adding some fraction of a blue unit, e.g. when the units on the slips
of each colour employed in obtaining a {x^rfect match are added together,
the toU\l may be found to come to :

Red Yellow Blue

ry6 : 9-8 : 0*4

These colours will not however be the same as those appreciated by
the eye. When converting the former into the latter, as must always

214 The Inheritance of Wing Colour in Lepidoptera

be done, it should be remembered that one unit of bhie, and one unit of
red, and one unit of yellow when combined give one unit of black or
neutral tint : and, further, that one unit of red and one unit of yellow
together give one unit of orange. Now it is clear that in the above
example the eye will only receive 0*4 units of black, because not more
than 0*4 units can traverse all three colours. Since 0*4 units of red have
now been required to produce the neutral tint, only 5*2 units of red
remain. These, combined with an equal number of units of yellow,
give 5*2 units of orange. Now 0'4 + 5*2 units of yellow have been used
to produce the neutral tint and the orange, which when subtracted from
the total 9'8 units of yellow, leave 4*2 yellow units remaining. Thus
to obtain the visual colours from the glass units: treat the blue as
black; subtract this from the red to obtain the orange; and subtract
the red from the yellow to obtain the yellow. The two expressions thus
found may be considered virtually as the two halves of an equation.
As an example of the readings usually obtained the following colour
measurements of two insects are taken from the protocols :

Wing

Left fore
„ hind

Eight fore
„ hind

Sta

Red
5-3

2-8

4-5

2-8

ndard glasses used

=

Visual colour

Insect
17 0^9

5)

17C?44

>5

YeUow Blue

12-9 0-9
5-4 0-7

9-0 0-8
5 0 0-3

Black
0-9
0-7

0-8.
0-3

Orange

4-4
2-1

■ 3-7
2-5

Yellow
7-6
2-6

4-5
2-2

Throughout the paper all figures in square brackets refer to colour
measurements. The first number is the orange value, the second

Or. Yel.
number the yellow value, thus : [2*5 : 2*2]. As a rule, the black value,
if any, was so small that it has been omitted.

III. Method of presenting Results.
(a) Curves showing the distributions of the colour-values (Figs. 15 — 25).

It was found that the colours of the insects did not fade appreciably
when they were kept in the dark. The colour of every insect was taken
separately in the manner already described (see p. 213), the conditions
of lighting, etc., being kept as constant as possible. With the readings
obtained from a series of insects, curves to show the distribution of the
colour- values can be constructed by plotting these values as ordinates,
the abscissae being determined by the numerical positions of the indi-
viduals in the series, after the colour- values of all the insects have been

H. ()n8F/)W iMf)

armngiyl in onlor of nmf^nitudt*'. Ktwh absciHsji then n^pn'srnlH an
individual insivt.

Thi» diHtributions which rt'wult lor t'arh family, or j^roup of fumilicK,
have lK»i»n ^ivt'n in fidl al tin* rnd of thr jmjM r. lis it wjim cuUMidin'd a
mort* conris«* and acouniU' uh^iIukI (»f puhlishinL^ ih** <lata than any form
of UbK'.

In each figure the heavy black line di'note.s t he orange values. ( )range
is the donunating colour, but owing to the particular dyes used in making
the coIouixhI glaitsi's. m»>re yell<>w units are always re(juired than Vi'<\.
This exct^ss of yellow is much less iniportant than the orange, since a
considerable inci-ejvse or decrease in it alt<*rs the colour torje much less
than quit<} a small change in the red. It is nevertheless significant,
so the values have been shnwn in each c.-usc by means of a small circle
on the same perptMidictdar as the corn's|)onding orange value. It will
be seen at once that these yellow values do not follow the sami* order
as the orange, and an oscillating curve results. This merely means
that of two given orangt's, the paler, i.e. the one containing the least
red, may contain more yellow than the darker, or rice versa. For this
reiison it ha.s been thought well to arrange the yellow values, also, in
their order of magnitude. The n-sulting curve is shown by a line of
crosses, but it must be remembered that any given cross does not
necessiirily refer to the same insect Jis the orange value on the same
perpendicular. The yellow curve arranged in the sann' order as the
orange has been called " yellow a," the yellow curve rearranged in its
own order of magnitude "yellow 6." Generally speaking it will be seen
that the yellow values are in most cases approximately equal to the
orange, so that the curve " yellow h " runs roughly parallel to the
orange curve.

In the case of single families, the colour-values of the two parents
have been printed in the margin in s<piare brackets, beside the arrow
which indicates, at the appropriate point along the colour scale, the
orange value of each parent. Where more than one family is included
the mean value of the parents h;is been given.

(6) 71) e frequency distributions (Figs. 2 — 14).

Although the curves showing the distribution of the colour-values
have been given chi«*Hy to serve the purpose of a record, yet for the sake
of convenience and simplicity, and in onler to bring out the mostsiUient

' ' This method has been HUggested for anthropological studies by Eug. Dubois (Man,
Vol. VIII. June, 19(>8).

216 The Inheritance of Wing Colour in Lepidoptera

features, the colour-values of each family or group of families have been
expressed as a percentage frequency distribution. Among other advan-
tages this method tends to remove differences caused by the size of the
group dealt with. The following method was adopted when converting
one curve into the other.

The colour scale was divided into intervals of 0*4 units eaph (e.g.
from 0-0 to 0*4, and 0*4 to 0*8, etc.). The number of individuals occurring
in each colour interval was found, and the result divided by four, so
that the percentages might all refer to the frequency in J^ of a colour-
unit. These average numbers of insects of any given colour-value were
then plotted as ordinates, and the colour- values as abscissae. On account
of certain errors mentioned below, an interval of 0*4 units was chosen,
since this gave a reasonably smooth curve ; though occasionally small
oscillations appear, which probably have no significance.

The orange values are shown by the heavy line. Though of less
importance the yellow values have been treated by the same method and
are represented by a fine broken line. Roughly speaking it is seen to
run parallel to the orange curve, as might be expected. As in the case
of the distribution curves the colours of both parents have been shown
in the margin below the arrow which represents their orange values.

(c) Experimental and statistical errors.

Whenever a frequency distribution rises to two maxima they corre-
spond to those portions of the figures showing the distributions of the
colour-values which are flattest, and which consequently contain the
greatest number of insects of one colour^ The points at which the fre-
quency distributions reach a minimum (about 2*0 — 2 5) correspond to
those portions of the curve showing the distribution of the colour-
values where the fall is most rapid, and therefore to those points at
which segregation appears to occur.

The question that must then be considered is whether the fall is
real or whether it can be more easily attributed to the various sources
of error. Errors due to technique, such as labelling, are, I think, insig-
nificant. There remain those due to taking the colour measurements.
The accuracy with which the readings can be taken varies considerably.
A difference may be just discernible between two colours, one measuring
1-00 and the other I'Ol units, but no change will be appreciated between
two other colours, one measuring 1000 and the other lO'Ol units. To
be discernible the increment in the last case should be about 01

HIack

< >raiiKe

VeU..w

3-2

3G

—

2-8

2-7

—

2-7

2U

—

2-G

2 7

2-3

2 0

0-2

5 0

6 0

OH

4-5

5-7

100

1-4

1-4

H.Onslow 217

(though in pnictico it apjMMirs to b** Ickk), In'cnuHf the Htinmlu.s nM|uin*(l
U) cuujk' a chan^f of Honsation is always a (Idiinti' fraction of tin* original
stimuluH. Thr jmliT coIohfh win thrn^fon* he nuMiHnriMl with ^rrat^T
*iccuracv than thi* darker. The chief error, however, is eauHe<| mther by
the ditfen'nees in tht* ei>Kmr t>f the wiii^ an-a selectee I for meiuiureiiH-nt,
than by the inability of the eye t.o <iiscrinniiale limr shades. As a
rule the yi'llow cohmr bec«)nies slightly paiei- towards the jHiiphery «»f
the wing. Therefore since exactly the same area cannot always again
be found, the experimental error is incn^Jised by rei)eating the mejisure-
nients. The following figures give some i<lea of \,\\v, ditterences in the
c<^li>ur of an average specimen :

Area rouml (liscoidal spot

., ht'tween discoiilal s\mA and central fascia

„ near hinder angle ...

,, near costa ...

,, near apex ...
Central orange fascia
C)ran>je Khoulder knot ...
Black diHcoidal spot

On account of this variability a i)ractice was always niade of selecting
the large yellow area surrounding the discoidal spot, the more perfect
wing of the two always being chosen. When this area was torn, or too
badly rubbed, the nearest portion, and therefore the one most approaching
the same colour, was selected. To avoid prejudice, all the readings were
taken before the curves were constructed, and in most cases the readings
were repeated twice. It was found from experience that at about the
range in question there was a maximum experimental error of 01 colour-
units. With regard to the (juestion whether the data prove that the
families and groups of families dealt wdth are really heterogeneous, and
not single samples, I have shown the evidence to Mr Udny Yule. He
has most kindly considered the matter and reports as follows : "As regards
figures obtained in the DR x RR cross (Figs. 4 to 9) I do not think there
can be any doubt that the data can be held to prove segregation. The
gap is wide and well marked and occurs in the same position in several
families or groups of families. In the case of the DR x DR cross
(Figs. 10 to 13) so well marked a separation cannot be expected, as the
second maxinnim does n(it rise to a sufficient height, and the distribu-
tion of the 72's only causes a comparatively slight hump on the tail of
the distribution. A check on the result can however be obtained by
building up the distribution that should be expi!cted on the theoretical
basis, showing that fair agreement is obtained." This check has been
applied with fairl}' satisfactory results. In the ca.se of the DR x DR

218 The Inheritance of Wing Colour in Lepidoptera

cross the fit is especially good, the second maximum being no higher
on the theoretical basis than on the experimental. In the case of the
DR X RR cross, the maxima, especially the second, though no higher,
are shifted somewhat to the right. This is no doubt due to the fact that
the average colour-values of the parents in the former case are con-
siderably higher than the average value of the yellow grandparents of
the DR X RR cross, since the yellowest insects were selected for breed-
ing. It seems to be fairly clear therefore, from the foregoing considera-
tions, that a considerable degree of segregation actually takes place.

{d) Colour of the hind wings, and effects of desaturation and sex.

In every case the hind wings are paler than the fore wings, and in
a large number of insects separate colour readings were taken of them.
In many of the medium and paler specimens the hind wings are quite
white, about [I'O: 0-7]^ In the case of one family, '17C (see Figs. 5
and 18) the orange values of the hind wings have been shown as a curve,
but as no special importance was attached to these figures, elsewhere
they have been omitted. On the curve (Fig. 18) it is seen how the paler
insects uniformly have white hind wings, irrespective of small differences
of colour in the fore wings.

Mention has already been made of the fact that many of the colour
measurements required the inclusion of a small quantity of black, before
exact equivalence could be obtained. This amount is very variable, and
occurs most irregularly in any series of insects. The highest values are
always among the deepest yellows, but even among these there are
some devoid of all black pigment. Usually the pale yellow forms have
no black, but white insects generally require a small quantity. The
deepest yellows no doubt actually contain a little melanic pigment,
which under the microscope appears as an almost imperceptible mottling
upon the scales. In the white insects the greyish colour of the surface
is caused rather by displacement or removal of the scales, than by the
presence of lyielanic pigment. These injuries are more apparent in
white insects than in yellow ones. Owing to the misleading effects of
injury; and to this lack of uniformity in the black values, they were not
considered of sufficient importance to record in every case. In family
'17 C, however (see Fig. 18), these values have been indicated by means

^ N.B. All figures in square brackets refer to colour measurements : the first number
denotes the orange value, the second number the yellow value. The black values have
been omitted.

Pedig

centralipuncta

iochatca flavipalliata ''.. ' "^ centralipuncta

T Y mgrosparsata) .,

WiW grossulariata
V 06

f flavipalliata
^ (Pale lemon)

(^ lacticolor

14 B.

_j flavipalliata

Wild gross. Wild gross.

(Kent) (Kent)

Wild gross.
(Cambs.}l

;i 4,^10,_p,l-4^1 0^5,2 6^,1 6^ 10 ^25 ^n28^02 ^25 ,-^0-5 Wi30^0'7 ^2e/-^07^M

Wild Wild

gross. gross.

(Kent) (Kent)

albipalliata ''""'>'

?l ^ .

0 chryso
I

Wild
gross,

(Kent)

"I —■ 1 1 1 \ — I 1 r

^10^08<5o6SPo8*2oCJi4^02 "l5 •26Vo-5 " IS ^06 • 3-5V o 4 " 1 7 ^0 sQPo-? ^ 06 •^ 1 5 VOB ^ 1 2

'I5D. 'I5R;i15G.'I50.

Wild gross. Wild
(Kent) iii mgrosparsata
fata nigricostata (Kent)

I5C.

I5Q, 15 F. '15 K. '15

S. '15

M. '15

Lost,

(Kent)

16 Z '16

nigri-
[ftensrs - cosiata Identity lost

\cta I — ^ ^^ (—! r-

^0 8^ 0-8 ^>3
^03X0-5 • $-7

impuncti-
V fasciata +

X. 'i6R.'i6\/.'l6G. '1^. 'I6N.'I6M. 'I6KF 'l6C. 'I6E

Wild
gross.
'Kent)

A-2 ^4-2 Y7i0-3\^l-8 ^I'l ^'10 ^06^1 2^39^3 5 fDO-6^40 ^35 ^,4-3^35^2-5 ^42^
3-8 "3-5 Vo-4 >f 11 ^1-6 ^0 5*^04*^ l8^3-4* 3 5 Vo-7 T3-7 T3-3 • 42 " 37" 1 7 • 36 •

I— I ' . ^ I J "T-^ I LJ

1 r 1 ■ I 1 1 ■ I

2 5 wi4-2^^3

2

1

6 ^2 9 ^41 ^
8 T21 • 48 ■

=! I I

^i7W.l7K. 'l78;l7S/i7B;i7Q[l7UU. 'I7a. ^l7^;i7D;i7fl;i7?;l7F?'l7t

N.B. The colour- ralues are denoted by t^e figures to the right
of each individual. The upper figures- Vefer to the orange values, the
lower figures to the yellow values.

0 = Homozygous for yellow, i i.

© = The same combined with lacticolor.

0 = 1-
0-1

IT.

Wild groti
HampctUatc (Ltyton)

n • • • ^

.Upunci^ "•«>^'^-

-1 • ' 1 '

A 3* (Uwpatliaia (Tf grotsul

J ariata

^ 17

■T—

14a[

• ai

^ 1 I

▼ ae

^ la

1905
1906
1907

FAMILIES BRED BY DR DONCASTER.

'06-21

of!

No yellow Ancettry ' A "7 9 Q

■0716. ^^^^

I 0 ^40

No yellomt Aneeti

-. "071.

r<.ji^,9\. *;; t;: ♦'- e?

"08-3. r

'08b

1908 <?;: <K\i <*•« <*""'"•

I

'0817

c,08?0.

1909
1910

'096.

'0919

or?; *!

-J L-

'09-I8.

(f tinged

1012.

1015.

'I5L

Wild Gross (Kent J
migricostata t

Wild gross.

(Kent I

( nighcostata *

{nigrocinda t

•|5R

Wild gross.
^lO ^\J I ^,3 0 ^3 7^13 ^,3 8 ^OQ (Kent)
^ la X08 • 26 "ae^o? " 34 voe
I — ^_J

* The colour of these insects could not be measured, but
record of their emergence makes it probable they were homozyj
yellow.

nigricostata t

,37^34 ^.8 8 ^3 7 wO ^3 7^14 3 w- 3 7 wi 3 7 ^0 8 r^ 10 ^0 9 ^15 ^10 rrf" ^'3 /^O 7 ^4 3^4 5 /*
» 39 ~34 • 44 ~33*3^4^ 44* 31 • 38 V04^10 Sf 04 ^ \A ^0 8 ^ 11 f 05*49T45^

J= "

Mum.

'I7R. l7e:i7Y 'l7E.'l7P.'l7e.'i7C.

m.

» the inhibition of yellow (white) I L
9tned with lacticolor.

3 - Heterozygous for the inhibition of yellow. 1 i

(Very pale yellow, or white )
3 "The same combined with lacticolor.

II. Onslow i>19

of diainond-shaiKMl (lots iimlrr tin* onin^f value of thr upproprijit^'
insect

In prt*|}firin^ the rurvcH Ui nhow the diHtribution of tin- <*o|nur
remlin^. the scxos won* iit first kept s«'|Mirat4'. In family 'IGA (srf
Fi^. 21) the females have Imvii shown al»»ne. just iwlnw the eurvr which
oomprisi\M b«»th wxes. Possibly there is a slii^ht t«'nileney for the fi'inai* .s
to be jMiler than the males, ami in the family in «jm'stion it can be sc«ii
that nither a larger pn)jM)rtion than usual of the pah'r forms are female.
On the wht)le, however, the sex apiM'ai-s to have little etVeet. Acconlingly,
elsewhen.' individuals of b<ith sexes have been treatiMJ together. Further,
in the cnxsses, the ^iex «)f the yellow parent appeaix t-o have no etfeet on
the offspring.

(e) Table of families, and pedigree.

In many of the families there occur some of the varieties which arc
describe<l on pp. 220 — 223. Their names have been given below th«'
curves showing the distribution of the colour-values at the end of the
paper, and to the right of the name is shown the sex and the number
of the insect in the curve. These numbei-s refei* of coui-se to the figures
along the base line of the curve, so that any individual ciin be looked out,
and the colour-value rea(J off from the curves. For instjince in family
'16G (see Fig. 24) the variety violacea occui*s. It is marked cT 24.
From the colour- values of the curve it may be seen that the 24th
insect is [r0:0"4]. The colour- value of any other variation may be
looked up in the same way. When, however, several families are com-
bined in a single curve, a difficulty arises. There is nothing to show in
a compound curve which in.sects belong to which family, and theivfore
it cannot be found from inspection which individuals in a given family
do not .show the pjirticular variation under consideration.

Reference to Table I will give this information, as the sex and
number of every individual in each ftimily has been there tabulated.
The gametic con.stitution of every insect used for breeding, ;is indicated
by the offspring, is shown by the symbols employed in the pedigree
(see Table II). The .sex, colour-value and any other details known are
also given, and the colour-values of all the offspring may be determine<l
by reference to Table I and the appropriate figure. All the details
subsequent to 1914 have been comjxjsed from my own data. F(»r reconls
previous to 1914 I am indebted to the Rev. (i. H. Raynor, wh<» also
gave me my original material. A few of the insects bred before 1914

220 The Inheritance of Wing Colour in Lepidoptera

were sent to me, and the colour was measured in the usual way. In
the other cases, although the colour could not be determined, it is clear
from the description whether the insect was a homozygous yellow, but
it is not always so clear whether it was a hybrid or a type. In these
circumstances the probable constitution of the insect has been indicated
by the symbol, and a note of interrogation has been added to show that
the question is doubtful. The pedigree of six families of set insects,
kindly given to me by Dr Doncaster, has been inset on Table II.

IV. Description of certain Varieties.

" As is well known, A. grossulariata is an extraordinarily variable
species, and Mr Raynor has named and described a number of its
aberrations^ His nomenclature therefore has been used whenever
possible, but in two cases I have been reluctantly forced to adopt
new names to denote slight variations for which no published name
could be found. The following descriptions include all the variations
which have been met with in the course of the experiments, but it is
by no means a complete list.

Var. chrysostrota.

(See Plate IX, Nos. 31 to 42.) This name is given to the variety
oi lacticolor, which is entirely suffused with yellow.

Na.r.flavi- or albipalliata.

(See Plate IX, Nos. 1, 7 and 19.) The black markings are absent
from a broad area between the black basal blotch and the discoidal
spot, which gives a white (or yellow) mantle about that area of the
insect. This variety only occurred among the parents of my strain,
and not much information could be derived from Mr Raynor's records
as to the nature of the inheritance of this character. In four families
both parents of which were flavi- or albipalliata less than 7 per cent, of
the offspring showed the variation.

Var. fulvapicata.

(See Plate IX, Nos. 3 and 28.) The orange colour of the central
band is continued to the apex of the wing, and the black marking at the
apex is obsolete. The incomplete development of the scales at the
apex, occasionally the result of inbreeding, may cause a superficial
resemblance to this variety.

^ ''Notes on Abraxas Grossulariata and how to rear it," Entomologist's Record, Vol. xiv.
p. 321, 1902 and Vol. xv. p. 8, 1903.

H. 0N8L0W 221

Var. impu ucti/'t^uiUi .

(Soi* PluU' IX, No. 24.) Thin viirioty hiw tmt h<Tn pn*viimftly
de»cribt»d. Tho blark ImmJ of H|H)ts on ihr ouUt widr of th*- onm^<*
faRcin nuiv Vh' nunph't^'ly or |Mirtiiilly ol)sol«'tr<'. It wtiiih thiit. this
variety shouM have a closo coniK^ctioii with th<* pr('(MHiifi^ oip*, hut
there in no ovi<ii»nce fn)ni which to docido. Kithcr thr |H)rti«>ii nf th<'
band near the inner margin or the ecntnil portion nuiy Im- ohso|rt<\
without the ajKW of the wing beinj; aflftJcUMl. In onlrr to HistingMi.sh
between the HifteriMit amounts of black present in thr hand of thes«*
insects, in Table II and elsewhere they have been (jualiHed in the
following way :

impnnctifasciaUi ^ signifies that about half the band is obsolete.
„ // signifies that three quarters of the })and is

obsolete.
„ . /// signifies that the band is completely oKsolete,

or only traces of it present.

(a) The Melanic Series.

A certain number of varieties may be collected under this heading.
They consist of variations in the size and position of the black markings
on the wings and body, and of a suffusion of dark pigment over the
whole wing surface. For convenience they will be considered together.
In several cases, although different names have been given to the
variations, one oflen only denotes a more advanced stage of melanism
than the other. The furthest stage of melanism is found in var. varleyaki
(Porritt)'. I am at present investigating the relationship of this variety
to the type as well as to lacticolor.

Var. radiiita.

(See Plate IX, No. 5<S.) Marginal spots radiated, generally on the
fore wings only. This variety may of* course occur in grossalariata as
well as in lacticolor.

Var. nigricostata.

(See Plate IX, Nos. 16, 22 and 28. No. 44 also shows traces of a
black stripe.) The bhvck stripe may extend from the shoulder knot either
to the discoidal spot or throughout the whole length of the costa. Some-
times the black marking below the discoidal spot is confluent, so that a

^ An excellent coloured illustration of var. varleijata (Porritt) is given by W. Bowater,
Plate XXVII, No. 45, Journal of Geneticg, Vol. in. 1914, p. 299.

222 The Inheritance of Wing Colour in Lepidoptera

second black stripe runs from the middle of the shoulder knot towards
the third marginal spot, which it may or may not reach, leaving one or
two small wedge-shaped areas of yellow or white in the centre of the
wing. (See Plate IX, No. 16.) The different lengths of this black
stripe have been indicated in the following way :

nigricostata signifies costa black as far as discoidal spot.
„ / signifies costa black as far as orange fascia.
„ // signifies costa black as far as the apex of the wing.
„ /// signifies second black stripe running more or less
completely across the wing.
Var. hazeleighensis.

(See Plate IX, No. 30.) Fore wings almost filled with black except
for small white specks in the middle of the costal margin. The hind
wings are sometimes banded, but are often unaffected. This variety
frequently occurs in the cross varleyata x grossulariata.

Var. nigrocincta.

(See Plate IX, Nos. 16 and 30.) This variety has not been pre-
viously described. It is caused by the dorsal spots being extended
round the body so as to form black rings. In some specimens the body
may be almost entirely black. The variety seems sometimes to be
correlated with the amount of black pigment on the wings, but this is
not always the case, because the dorsal spots in many specimens of
varleyata are quite normaP.

The degree of pigmentation is indicated in the following way :

nigrocincta + signifies black rings incomplete.
„ /f signifies black rings complete.

„ -h+i signifies dorsal area entirely black.

In some insects the black rings are only found on certain abdominal
segments. In these cases the correct segments are indicated in brackets,
just below the variation on the distribution figures.

Var. violacea and var. semiviolacea.

(See Plate IX, Nos. 11 and 27.) Either all four wings or only one
pair may be sufi'used with a purplish brown bloom that gives the insect
a scorched appearance. Under the microscope the pigment is seen to
be purple, and evenly distributed over the surface of the scales, which
have a faintly mottled appearance. Otherwise it strongly resembles
the suffusion of iochalca {vide infra).

1 Compare the insect illustrated by Bowater (see note p. 221).

H. Onslow 223

Var. nigntsfKintdta.

(»StM» I*lrtt4» IX. No. 2i).) All tho \\'\i\^ Jirr firirly M|MH-kl«M| with
black, which fjivt*8 tht' iiimn't a H<M»(y a|)|M>aniii<'(', timt. «'xt<inls «iv«t tht*
oraiigi^ fiiMciii, It is siiid that this variation is n<»t inherited, and
Mr IWritt has UtU\ uw that he t*ail«'(i to hrrcH a ^rratcr prrcrnta^r of
ajK-ximcns from uiffntsiHtrsttUi ? x niifvosparitaUt J than ho \v;ls ahl*-
to obUiiii fn»ni wiM larvao colloctoti in the siuno n('i^'hlMMirli<M.<l.
Mr li^iynor has also had a vory similar cxjx'rirnc*'. I (»ncr oht,;iin«Mi a
sin^lo s|)«»oinion of inffntspdrsata J^ from larvae cojlrctcd in Kent, whif^h
was painni to a nither<iark female from the s;ime Im^ality : out of 12 off-
spring only one % showed a very faint trace of melanic sut!*u.sion.

Var. iochalca.

(See Plate IX. No. ()().) 'Phis variety hears a certain relati(»nship
to the prectnling one and to vwlacea. in that they all have a melanir
suffusion which extends over the surface of the wing. Th<' ditV«'renc«*
in apfH»ar.ince is however most striking, as may he seen by reference to
Plat4» IX. In iochdlca the wings are suffused with a pecidiar metallic
hue duo to the addition of black pigment to the scales of the y«'llow
variety of kicticolor (i.e. chrysostrota), which caus(»s a desaturation of
the yellow colour. In nigrospavsaUt on the other hand, the suffusion
is more intense and often has a mottled appearance, especially when
observed through a lens. It is true that the specimen illustrated
(No. 29) has a white and not a yellow ground. Nevertheless, when
nigrosparsata is combined with lutea, the difference is if anything more
striking.

(h) Distribution of black pigment in sufficsed viwieties.

A yellow insect showing this suffusion was bred too late to be included
in the plate. A seiwrate illustration, however (see Plate X), shows the
difference between the dejx)sition of the melanic pigment in the yellow
variety of nigrosjxirsaki, and in a specimen (^f iochalca as they appear
under the micro.scope. The j>ortion of the f(n*c wing of the nigrusjxtmatif
illustrated is much closer to the costa than that of the iochalca, and
consequently the scales are smaller and narrower. A comparison of the
two figures shows that the scales of iochalca (see Plate X, Fig. I) are
a uniform buff, the black pigment being rather more abundant near the
base of the scales than at the tip. For this reason th<' pigmentation is
somewhat obscured, bec«'iuse the darkest portion is always covered by the
pale overlapping jxjints of the scales in the row preceding it. The wing

224 The Inheritance of Wing Colour in Lepidoptera

of nigrosparsata (see Plate X, Fig. 2) on the other hand is seen to have
the black pigment concentrated in certain scales, the majority of yellow
scales being practically devoid of dark pigment. In contrast to the
case of iochalca however, the dark pigment appears to be concentrated
in the tip of the scale rather than at the base, which accentuates the
mottled effect already caused by the localisation of the black pigment in
a few scales. The few grey scales of iochalca scarcely affect the colour.
Mr Raynor has kindly given me data concerning several pairings
with iochalca. It appears to be recessive to A. grossulariata and to
breed true when mated together. A pairing between iochalca </ and
iochalca ? gave 13 females and 6 males, all iochalca. Another family
of iochalca J^ x albipalliata J gave 5 females and 1 male iochalca, and
12 males and 15 females of other varieties, 14 of these females being
lacticolor. It must of course be remembered that the inheritance of
t/his variety may be complicated by the fact that it probably never occurs
except in combination with lacticolor. Experiments are in progress
with a strain of this variety kindly given to me by Mr Raynor.

Var. cupreofasciata.

(See Plate IX, No. 59.) This variety has the melanic suffusion of
iochalca, restricted to the orange fascia and shoulder knot. It can be
seen that the fascia has exactly the same appearance as that in No. 60,
Plate IX, and may be contrasted with the bright yellow fascia of the
ordinary lacticolor insects (Nos. 55 and 56). Mr Raynor tells me that
insects of the variety cupreofasciata are invariably descended from
iochalca ancestry.

V. The Yellow Pigment.

The cause of the difiference in the intensity of the yellow pigment
in the pale and the deep varieties is a problem of considerable importance,
since it may possibly give a clue as to the nature of the factors which
determine the varying shades of colour. There are several possible
conditions which might account for these colour variations.

(1) The pigment itself may vary in constitution, and consequently
in cplour.

(2) The concentration of the pigment, which may be either diffused
or in the form of granules, may be increased in the deeper varieties.

If the composition of the pigment varies the change need not be
very fundamental, but might merely consist in an alteration in the

H. Onslow -j-j")

position or nuinlH»r of rrrUiin chciiiiail ^'rnu|>H'. If thr clian^rr m
quantiUitive ami tho pi^tiRMit Ih (iiri'uHo<i, th(> iimount " in H<»liiti«>n" umy
be incn»ai»e<l, or the pigment may IxroiiH' a^^rr^at^Ml in .small graimlcH
or inassoH. The orange-red pigments of the l*ieri<lae an- all y«H<»w in
mpitM)Us solution, the orange colour being probably due t^) inen;i,s«<l
oonoentnUion". The jwile yellow butterfly (\diaH hyale^ is said l«»
conUvin mostly diffused pigment, wherejus the more <leeply coloun'<l
C7. edusa contains gnundar nuwses of pigment between the walls of each
scale.

(3) A similar deeixMiing «)f colour might be j)rodueed by an increase
in the number of scales, i.e. several overlapping layers of sc;iles woidd
give a deejKT shade than a single layer.

(4) Finally, the position of the pigment within the scale itself
might affect the colour, which would depend im the proximity of the
pigment to the upper surface, e.g. it might be chieHy in the upj)er or
lower membrane of the scale, or it might be between the two*.

Clearly a correct decision of these cpiestions is of the utmost im-
portance, for if the pigment varies (piantitatively and not <jualit;itively,
the factor or f;ictors which control the amount must be capable (»f acting
in a quantitative manner upon the pigment-producing mechanism.
Now this mechanism, in many if not in all ciises, consists of an enzyme
(usually oxidising) which acts upon a colourless chromogen. It is
difficult to see how the addition or subtraction of a factor — a change

* In the yellow flower pigments for example the pale yellow flavone Apigeniu

O
HO

HO CO
becomes the much more deeply coloured pigment Luteolin,

_ O^ ^ .OH

HO '"

HO CO
i.e. an extra hydroxyl group appears in the side ring.

« F. G. Hopkins, Phil. Trans. Vol. 186, B, p. 678, 1895.
» W. Geest, Zs. tciss. Imektenbiol. Berlin, Vol. iv. p. 162, 190r».

* A case of dilution of colour owing to localisation of pigment occurs in the hairs of
black and blue mice and rabbits. H. Onslow, Proc. Roy. Soc. B, Vol. 89, p. oG, 1915.
Joom. of Gen. viii 16

226 The Inheritance of Wing Colour in Lepidoptera

which cannot but be called qualitative — can cause a quantitative
difference in the amount of pigment deposited by an enzyme-substrate
system. The only factors that are capable of influencing such a system
quantitatively would appear to be those such as time, temperature,
mass action, hydrogen ion concentration, etc. Increased pigmenta-
tion may of course be produced by the removal or suppression of
inhibitors, but if there are a great variety of shades to be accounted for,
as in the case of lutea, this may lead to the employment of an absurd
number of inhibitors before the phenomena can be accounted for.

Distribution of the pigment in scales of insects of varying shades of
yellow.

A preliminary microscopical examination of a number of insects
showed that the colour was by no means evenly distributed throughout
the wing, but that patches of pale and more deeply coloured scales
were often intermingled. Even in a single scale the pigment was by no
means uniformly deposited. On the whole, however, it can be said that
in the orange varieties the scales are a much deeper yellow than in
paler specimens. This observation was confirmed by examining the
scales by transmitted light. To investigate the condition of the pig-
ment more carefully a number of sections were made of wings from
individuals of various shades. As is well known, a scale is a flattened
sac and appears in section as shown in Plate X, Figs. 3, 4 and 5.
With considerable difficulty thin uniform sections with an average
thickness of about 2 to 3 yu, were obtained. An inspection of these showed
at once that the pigment was present " in solution " within the chitin,
and in no case could any granular pigment be found even in the deepest
yellow insects. A scale from a pale insect is shown (Fig. 5) having a
value of about [1'8 : 2*0]. Very pale and white specimens have so little
pigment that thin sections are practically invisible when mounted in
balsam. A scale from a moderately yellow insect (Fig. 4) with a colour-
value of about [2'8 : 3*0] is also shown. This section is cut near the base
of the scale, not far from the root, as is indicated by the considerable
increase in the distance between the two walls at the centre. The
deepest orange section (Fig. 3) came from an insect with a colour- value
of about [4'0 : 4'0]. Clearly the pigment is diffused throughout the chitin,
and therefore any increase in colour must be caused, either by a quali-
tative difference in the pigment, or by an increase in its concentration.

H. Onslow 227

VI. The Orkjinal Material.

I am indebu»d to tho kindn('.H.s of the lit'v. (J. H. lijiynor for the
maU'rial with which the brooding ox|H'riinonts wcr(3 airried out. This
strain of lutea origiuatcd in 1J)04 with two pair insectH bred from wild
larvae collocU»d in the neighbourhood of Warrington. These in.s(?(!ts
were shown to nie : the (f had a value of about [17 : lo], and the J a
value of about [1*0 : 18], bein^f only just tinged with yellow.

Dr Donciwter most generously gave me six families of set insects
bred by him between 190()-li)10, and with his permission, 1 have
included the data from the^ families with my own. Eiich family bred
b^ Dr Donc4ist<?r has been marked with an Jisterisk wherever it (x:curs.

My original material consisted of two batches of ova sent by Mr Riiynor
in 1914. The female piirent of one family ('H B), was a wild insect
previously fertilised by a yellow male. The wild $ was aiptured in
Milltown Park, Dublin, the locality in which the ''Q variety," described
by the Rev. J. M. Woodl(X)k\ w;i8 found. All the F^ generation was
therefore heterozygous for yellow. The parents of the other family
('14 A) were described as a " piir of yellows." One of these yellows
which I was able to examine had a low colour- value, i.e. [17 : 17]. As
a rule, homozygous yellows, as can be seen from Fig. 16, have a much
higher value, at least above [2*4 : 24]. Moreover, it will be seen from
the pedigree that the male in question wjis bred from a jlampalliata %
X grossulariata (f . Consequently it is much more probable that the
insect was a yellowish heterozygote, than a homozygous yellow, and
therefore the family '14 A is BR x RR, and not RR x RR as was first
thought. This view is supported by the fact that of the total offspring
(10 c/'c/' ^nd 12 ? J) 11 had an orange value greater than 2*6, and 11
an orange value less than 16. The material has therefore been treated
in accordance with this supposition.

The larvae were reared with the usual precautions, great care being
taken with the labelling, and every endeavour made to avoid intro-
ducing either eggs or very young larvae from one box to another. It
seems almost impossible, however, to prevent a larva that has escaped
from reappearing occasionally amongst the food when another brood is
being dealt with. No doubt a small percentage of errors may be
attributed to this cause, as for instance the appearance of some of the
palest individuals in the lutea x lutea matings (see Fig. 16). In order
to avoid the ill-effects of in-breeding, type insects were obtained for

» J. M. Woodlock, Journal of Genetics, Vol. v. p. 183, 1916.

16—2

228 The Inheritance of Wing Colour in Lepidoptera

the purpose of crossing, by means of larvae or pupae, from Cambridge,
Kent and Yorkshire. Notwithstanding this precaution and the fact
that the species resists in-breeding very well, all ill-effects were not
avoided. In 1916-1917 many of the females became sterile, and the
families of others were very small. Dysentery was also severe in 1916
and 1917, as may be seen from the small lutea x lutea families in the
latter year.

VII. Beeeding Results.

The frequency distributions (Figs. 2 — 14) have been constructed from
the curves showing the distribution of the colour-values (Figs. 15 — 25)
in the manner described on p. 215. The orange colour- values of the
parents have been marked with arrows on the scale of colour-units at
the bottom of the figures, and the full colour-value has been printed in
the margin. When several families are combined in one curve, the
arrows indicate the mean orange value of the parents. When both
parents are of the same gametic constitution, one arrow refers to the
male and one to the female parent. In the case of a cross containing
several families, the arrows refer to the varieties of the parents regardless
of their sex. The names of all the families included in any curve are
printed on the figure, together with the year in which the eggs were
laid, thus :— 16 M, '17 E, etc.

(a) Var. lutea x A. grossulariata.

For the sake of clearness it is well to divide the colour scale into
five arbitrary classes, corresponding to the five columns of insects in
Plate IX. This division of the colour scale in both types of distribu-
tion curves helps to bring the phenomenon of segregation into evidence.
The white or lowest class includes colour-values, from 0*0-0'8, of the
type insects (Plate IX, Nos. 25 and 26). The second class, from 0*8-1 '6,
are the slightly tinged individuals heterozygous for yellow (Plate IX,
Nos. 19 to 24). The third class rather deeper in colour, from 1*6-2*4,
is as a rule also heterozygous for yellow (Plate IX, Nos. 13 to 18).
It does not contain more than a very few individuals, because it embraces
just that part of the scale where segregation takes place, and where the
frequency is therefore lowest. The fourth and fifth classes comprise
the homozygous yellows, from 2*4-3*2, and from 3*2-4*0, or above
(Plate IX, Nos. 1 to 12).

The i^i heterozygotes lutea x gross, cross can usually be distinguished
from the type by slight differences of form or colour. Over a thousand

H. Onslow

t>L>9

type siKKiiinens, coIIccIimI fnun varinns localities, Imvc
iiml in no insect <li<l the c(jlonr-viilue exceed |ll :0
great niajority wen* |0 7 : O'd] or less.

l)een e

:il. will

xiuiiined,

•rejL« the

17 a "16 P 'ISC '14 a
'17 K. 16 T '15 P. '08-5.''
'17 a '15 K. 08-20'
'17 6. ^5M.

'I5Q.

4 Colour-units

[0-G:0-5]
Mean colour-value of
grossulariata parents

[3-3: 2-8]

Mean colour-value of

lutea parents

Fig. 2.

(Cf. Fig. 15.) Curve showing frequency distribution of the orange and yellow
colour- values of the offspring from 14 pairings of hitea x grossulariata.

All the families resulting from pairings of lutea x grossulariata are
small and have been combined in a single curve (see Fig. 15). The mean
orange value of the yellow parents was 3'3 or nearly six times as yellow
as the mean value of the type parents, which was 0*6. The frequency
distribution (Fig. 2) reaches a maximum at about 1*0, showing that the
factor for yellowness is not completely recessive, because the majority
of the offspring are more deeply coloured than their type parents.
However the orange colour represented by TO unit of orange is a very
pale cream, rather paler than insects Nos. 20 and 21, Plate IX. It
was found to make no difference in the offspring whether the male
parent was a grossulariata or a lutea, so the reciprocal matings have
been combined in one curve. They can be easily separated, however,
by reference to the pedigree and Table I. The curve though rather
narrow is not unlike an ordinary frequency distribution. The range of

230 The Inheritance of Wing Colour in Lepidoptera

variation is small, though the slight broadening of the curve, just before
a colour-value of 2*0, shows that a small percentage of the offspring are
very perceptibly tinged with yellow. The result of mating one of these
rather yellow hybrids with a type insect will be seen in section (e), p. 239.
The broken line denotes the curve formed by the excess of yellow in
the colour measurements, and it runs roughly parallel to the orange

curve.

J? 8

E

a
t 6

'17 D.

^ ,

- '17 F.

A

1

_ '17 H

/\

.

V S.'

1 \

- 17 T 16 B.

/ \

-

17/3. 16 D.

/ '*— " \

" 17 5.

^ ! M

- V t A

#/ \

-.X f

'17 ^. ^/ \

\

I \

v.""

1

^

i 2

3

4

5

2-

Colour-units

[3-6: 3-4] [3-9: 3-7]

Mean colour- values of

lutea parents

Fig. 3. (Cf. Fig. 16.) Curve showing frequency distribution of the orange
and yellow colour- values from 13 pairings of lutea x lutea.

(h) Var. lutea x var. lutea.

This curve (Fig. 3) is again composed of a number of small families,
and though asymmetrical, it is more like a normal frequency curve
than Fig. 2. It reaches a maximum immediately above the arrows
indicating the mean orange value of the respective parents. The base
of the curve is fairly broad showing that the range of variation is
considerable. Reference to Fig. 16 shows that about 10 insects have
an orange value below 2*5. Although some of the parents had a low
value, as may be seen from the pedigree, these offspring are unusually
pale, and there can be little doubt that some of them are due to experi-
.niental errors, i.e. of labelling or while feeding. The general appearance
of the black pattern on one or two of them is quite dissimilar to that of
the other members of the family, and these no doubt have strayed from
elsewhere. This explanation cannot account either for all the pale
individuals in this cross, or for the deepest yellows in the Fi generation.

H. Onslow

•2:n

The bnikon line indicating tho oxochs yi-Ilow in Kij^. .'{ H netj. niiiH iih
usual cItHk'ly |M»nillfl to the onm^t' rurvr. I'xcfpt for a Hhiirp |MMik in
the pale n»jjion. whi<'h is no doiiht tuily an irnt,nilarity raiiM<'(l hy an
insutticiency t>f nunih»'i-s.

(r> /////>/•/(/' {lutea x (fntss.) x var. Intra.

A grinip of small families, none of t hem lar^e enough to he treated
8ejwn»tely. has Wen Hi-st dealt with (Fig. 4). Am would be ex|x'ct«Ml
in this tyjK' of mating (Dli x HH) the ofVspiing segregate into tw(j
8epj»rat<» classes. Tlu' hybrid insects have a value of ()H-2"4, and the
yellow insects one of 24 or above. The curve reaches a minimum at 2*2,
and the two maxima occur at just about the mean orange value of the
yellow and hybrid parents respectively. The areius of the two curves
should be tH]ual, and should repn*sent the number of individuals in each
class; actual measurement shows that the percentage «»f the dominant
form is 54, and that of the recessive form 40. The yellow values show
a rather definite decline for the luted homozygotes.

>.8
o

c
o

3

V 6

17 0.

*I6H

'15 e;

14 A

17 R.

'15 N.

17 Y.

'15 0.
*I5R
'15R

5 Colour-units

[1-3 :M]
Mean colour-
value of
hybrid parents

[31: 2-6]
Mean colour-
value of
lutea parents

Fig. 4. (Cf. Fig. 17.) Curve showing frequency distribution of the
orange and yellow colour-values of the oflfspring from 10 pairings of
hybrid {lutea x gross.) x lutea.

* Although the word "hybrid" strictly applies to species-crosses, it has been retained
instead of the preferable term "cross-bred," since it was too late to make alterations in the
figures at the end of the paper, where the first term had unfortunately been printed.

232 The Inheritance of Wing Colour in Lepidoptera

Figs. 5, 6, 7 and 8 show the curves given by four separate families in
: which the offspring were numerous enough to be treated by 4ihemselves.
The essential features are much the same as in the last case. In family
'17 C (Fig. 5) the BR individuals are indeed slightly darker than the
% parent (hybrid) whereas the RR insects are nearly all paler than the
</ parent (lutea) which was very deep [43 : 4-9]. Two separate curves
have been added to this figure to indicate the orange values of the
hind wings, and the black values of the fore wings (see p. 218).

1 2 3 4| 6 Colour-units

[0-7: 0-5] [4-3: 4-9]

Colour-value of Colour- value of

hybrid parent , lutea parent

Fig. 5. (Cf. Fig. 18.) Curve showing the frequency distribution of the
orange and yellow colour- values of family '17 C, hitea x hybrid
{lutea X gross.).

The orange colour of the hind wings is also shown by means of a dot-and-
dash line. The small quantities of black which occur so irregularly
in the fore wings are shown by means of a thin continuous line.

Family '16 E (see Fig. 6) shows a somewhat similar curve; the
yellow (f parent [2*9 : 21] was not quite so dark as in the last figure,
and the J parent unfortunately was lost. However a record exists
saying that it was rather deeply coloured for a hybrid: the orange
colour-value being probably not less than 1*7. The result seems to be
a certain increase in the number of yellows. The double maximum

II. Onslow

233

seen at a value of about *M) is probably only dui* t-o tho small iiunilM.*r
of individuaU available.

5 Colour-units

Hybrid
parent loRt

[2-9: 2-1]

Colour-value of

Intra parent

Fig. 6. (Cf. Fig. ![).) Curve showing freijuency distribution of the orange
and yellow colour-values of family 'IGE, hybrid {Ititea x (jross.) x lutea.
The ? parent (hybrid) was unfortunately lost.

Family '1015* (Fig. 7) shows almost exactly the same princijxil
features as the preceding curve. Segregation appears to occur at the
same point, and the homozygous yellows are again somewhat in excess,

5 Colour-units

Fig. 7. (Cf. Fig. 20.) Curve showing frequency distribution of the orange and yellow
colour- values of family "10-15 *, hybrid {lutea x gross.) x lutea. Parents missing.

Bred by Dr Doncaster.

234 The Inheritance of Wmg Colour in Lepidoptera

but unfortunately the parental values cannot be shown as the insects
were unobtainable.

Family '16 A (Fig. 8) shows yet another cross of the same nature.
The yellow parent is again very dark, and the homozygous yellows are
in excess of the hybrids. In this family the females have been kept
divided from the males, and the orange colour- values of their fore wings

>»

o

c
o

3

0) 6

' h '

■

a/ i
•'a

' / \ '•

'16 A. "

^' / \ i

Jf

il : \\9

# ,1

c^

1 , \l^

1

2

3 4 5

[1-3: 0-7]

Colour- value of

hybrid parent

5 Colour-units

[4-9 : 4-6]

Colour-value of

lutea parent

Fig. 8. (Cf. Fig. 21.) Curve showing frequency distribution of the orange
and yellow colour-values of '16 A, hybrid {lutea x gross.) x lutea.

The orange values of the females alone are shown by means of a dot-and-
dash curve.

have been shown by a line of dots and dashes, which differs in no essential
from the same curve for both sexes (see p. 219). Reference to Fig. 21
shows that the excess of yellows over hybrids is greater than in the three
preceding cases, there being at least 64 yellows to 20 hybrids, or a ratio
of 76: 24 (per cent.).

Now supposing the yellow colour develops in the variety lutea owing
to the loss of an inhibitor /, from the white type //, then this excess of
yellows might of course be accounted for on the hypothesis of another
factor X, a deepener, which has no visible effect upon yellow ii, but which
turns the pale hybrids li into deep yellows. If for instance the yellow
parent of '16 A had been Xx, then half the hybrid offspring would have
been liXx, and half lixx. Those carrying X and I would be changed
into deep yellows, and the result would be 75 per cent, deep yellows, and
25 per cent, pale yellows or whites. The ratio observed in this family
('16 A) is very close to the result expected on this hypothesis.

H. Onslow

235

In the three prece<ling faniiliea tho yellowH have iilwiiyH hcrii in
excesH. Tho result of eoinbining them* thnu' faiiiili»*H is to maki' the
ratio of the receanive f«)nu to the dominant (JS : :\2 (jmt ct-nt.), a sonn*-
what K»wer |)n>j>*>rtion t»f yellows than in 1(1 A. In Fi^^ 4 thr pro|Hnti<»n
of yellows to hybrids wa.s about •V():54(|)er cent.), but the evidence dti ivid
from the four large families of alxmt XO inserts rarh, is undoublrdly mon*
valuabK> than that derived tVom a nuinlHT of small on«'s. Furtli«T the
offspring fnmi all pairings of the ty|>e PR x RH have been combined in
one curve. Fig. 9, which resembles the fivi* preceding curves in most
essential pi^ints : and results in a nitio of 41 : 50 (per cent.). But jls
a matter of fact there are not sut!icient data from which to draw a definite
conclusion, especially as no confirmatory evidence is forthcoming from
the other crosses.

6

1 1

1 1

>»

u

g^

c

3

/ *. 1 *J

DRxRR. -

o-

/ \ \^

g 4

/ ' \tt

Av.

v>

^i \\^

/ ^\

0)

/ \

t«

^l ^\

/ \ ~

I 2

f , \"<

( ^->^

o

1 1

I /

\ ,•'' "^ ^ \

X

\ /

k

~ Vx

Q.

" / / S

\ ^

i;

N. "^

- / ^

V^

5

^v*^-.

• o

^

x^ ^2

0

1

2

3

4 £

Colour-units

[1-1: 0-8]

[3-6: 3-5]

Mean colour-value of

Mean colour- value of

hybrid

parents

lutea \

jarents

Fig. 9. Curve showing frequency distribution of the orange and yellow colour-
values of the offspring from all the hybrid {lutca x gross.) x lutea families
shown in Figs. 4, 5, 6, 7, and 8.

(d) Hybrid {lutea x gross.) x hybrid {lutea x gross.).

The frequency distribution shown in Fig. 10 is composed of a number
of small families of this cross, none of which were large enough to be
treated by themselves.

If the heterozygous yellows, which are often distinguishable from the
white type by their pale yellow colour, were really to form a distinct class
to themselves there should be three maxima in Fig. 10. Clearly there
are only two, showing that the pale yellow heterozygous class passes
by insensible changes into the pure white DD class with an orange value
of less than 08. It should however be carefully noted in Fig. 22 that
there are 48 insects or 24 per cent, with a colour value of 08 and

236 The Inheritance of Wing Colour in Lepidoptera

under, whereas in Fig. 17 there are about 1 per cent. But since the
maximum frequency for heterozygous j^ellows (about 1*0) lies so close to
that of the white type insects (about 0-6 or 0*7), the ability to distinguish
colours is not sensitive enough to permit of the two maxima being

2 3 4 5 Colour-units

[1-0: 0-9] [1-1: 0-9]

Mean col our- values of

hybrid parents

Fig. 10. (Cf. Fig. 22.) Curve showing frequency distribution of the orange
and yellow colour-values of the offspring from 16 pairings of hybrid
{lutea X gross.) x hybrid {luteax gross.).

separated. Segregation between the heterozygous and pure yellows is
distinct, though the frequency of the yellows does not reach a very
pronounced maximum. The expected ratio, 25 : 75, is in this case
exactly the observed ratio. It must not be forgotten that the length
of the colour scale occupied by the yellow class is as great as that
covered by both the type and the heterozygous classes, but it is the
area of the curve not the height which corresponds to the number of
individuals it contains. Some of the extracted yellows are very deep
in colour, almost as dark as a pure-bred yellow, but the average orange
colour-value (about 3'0) is rather less than what it is for pure yellows
(about 3-6).

Figs. 11 and 12 are made from two fairly large families of the same
type of cross. Fig. 11 is in most details similar to Fig. 10 except that
both the yellow and orange curves rise to rather a sharper second
maximum. Fig. 12 is not so characteristic because the family is only a

H. ()nsu)W

237

very sniall ono. A« bt^fort\ Fign. 10, 11 and 12 have \kvi\ ccunbined and
the result is shown in Fi>;. \'X

Colour-units

•3:1-1] [1-5: 1-4]
Colour-values of
hybrid parents

Fig. 11. (Cf. Fig. 23.) Curve showing frequency distribution of the orange and yellow
ooloar-valaes of family '17 E, hybrid {luUa x gross.) x hybrid (liitea x gross.).

>.'6
o

c

3

© 12

A

■|6 G.

/ \'^
/ \^
/ \^
/ x*^

if \ \

-

<N

r

2 3 4

4-*

0

[0-9: 0-4] [1-0: 0-6]

Colour-values of

hybrid parents

4 Coloup-units

Fig. 12. (Cf. Fig. 24.) Curve showing frequency distribution of the orange
and yellow colour-values of family '16 G, hybrid {littea x gross.) x hybrid
{lutea X gross.).

The essential features are the same as in Fig. 10, but the curve,
especially in the yellow region, tends to oscillate more. This is because
the points on the curve have been taken rather closer together than

238 The Inheritance of Wing Colour in Lepidoptera

usual. The ratio of the area of the yellow class to the other two is about >
22: 78 (per cent.).

>>

o

c

0)

3

o-

■DC

d

c

<D
O

£.
<0
Q.

10

1

— 1 1 1

" fl \

DRxDR.

1 rf

1 ''

-

i J

/ cJcf

X

^^\v. !

1

i 3 4 5

Colour-units

[1-1: 0-9] [1-2: 0-9]

Mean colour- values of

hybrid parents

Fig. 13. Curve showing frequency distribution of the orange and yellow
colour- values of the offspring of all the hybrid {lutea x gross.) x hybrid
(lutea X gross.) families shown in Figs. 10, 11, and 12.

16

o 8
o

s.

■ k '\7 W.

'15 G.

■ .

. /\ '16 0.

'09-6?-

1 / \^

1 \^

_

X / \^

>-/-K W

_ ' / \ \

1 1 "^ \

,1 \ .\

/ A
1 / ^ \

-

/ ^ \

I ^ *

' / ''

\

w

~

1 / «,

0

\

1 1 ^

t).\

\

/ t

c"

. \

1 / , a

a

^ \

, / u

>.

\ \

i '^

t ^

v.^ .

1

1

2

3

4

Fig,

Colour-units

[0-9 : 0-6J [1-4 : 1-0]

Mean colour-value Mean colour-value
of gross, parents of hybrid parents

14. (Cf. Fig. 25.) Curve showing frequency distribution of the
orange and yellow colour-values of the offspring from four families of
hybrid {lutea x gross.) x A. grossulariata.

H. (Xnsu)W 239

(f) Hybrid {lutea x gross.) x A. grossularintd.

A few |)iiinn^ only of this ty|H« of mating won* iimdr, and the fonr
faniilio8 rt^sulting an* shown in Fi^. 14. With the exception of family
*17 W the hyhrici jwrents wen* all a vrry |Mih' yellow, and the result, a.s
might l>e exjXH^tAni, is a typical frecpieney distrihution. Th«' hybrid
parent of *17 \V hacl a high colonr-value [24 : 22]. The oflfspring are
much nion» nnmerons than is represented in P^'ig. 25. There were indeed
82 insivts, but 64. which had a colour-value between [07 : 0*.')] and
[0't):0'4] and which showed no other variations of interest, were set free
as stxm as they emerged. The insects of family '17 W which were
entered in the curve (Fig. 25) may be found by reference to Table I.
The mther deep colour of the J parent does not seem to have hml any
visible effect on the colour of the oflfspring.

VIII. Summary.

The yellow ground colour of Abraxas grossulariata var. lutea
(Cockerell) is incompletely dominant over the white ground of the type.
It was found impossible to divide the F^ and F.y generations into distinct
classes, because the colours of the insects form a continuous series
varying from white, through the palest yellows, to a deep orange.

A commercial instrument for the measurement of colours on an
arbitrary standard scale of colour-units, called the " Tintometer," is
described. By its means the colours of the insects have been given
numerical values, thus enabling the complete data of the breeding ex-
periments to be expressed in the form of curves showing the distribution
of the colour- values. When these curves are converted into percentage
frequency distributions, the F^ generation resembles an ordinary chanc(^
di.stribution. But on the other hand the Fo generation, etc. are at once
seen to give curves having more than one maximum caused by the
tendency of the colour factors to segregate according to the ordinary
Mendelian laws.

Only some of the DR individuals are distinguishable from the 1)D,
on account of a faint tinge of yellow in their wings. The difference
between the two classes, when expressed in colour-units, is not sutiicient
to enable the maxima representing them to be separated on the curves.

The eflfects of sex and of variations of colour in different parts of the
same insect upon the measurements are considered, and the various
sources of experimental and statistical error are discussed. The gap on

240 The Inheritance of Wing Colour in Lepidoptera

the curves between the two classes is so clearly defined, and occurs so
often in the same position in various families, that there can be no doubt
it is significant and proves that even in the DR x RR families segre-
gation really takes place.

A certain number of variations, which appeared during the course of
the experiments, are described and figured. Very little is at present
known of their genetics, but a table is given, enabling all details of their
occurrence in the curves showing the distribution of the colour-values
to be easily found. Further experiments are being conducted with
var. varleyata as well as certain other varieties.

Special attention has been paid to the manner in which the pigments
are deposited within the scale, and to the exact modifications which are
the cause of certain varieties. Variations in the position and concen-
tration of the melanic pigment are, for instance, the only factors which
govern the difference between the two suffused varieties iochalca and
nigrosparsata.

A microscopical examination of the scales, both in situ and when
sectioned, shows that in the deepest colours the yellow pigment is
diffused throughout the chitinous walls of the scales, without the
formation of any granules. The intensity of the colour must therefore
be determined, either by variations in the concentration, or by a change
in the nature and consequently in the colour of the pigment itself. If
the yellow variations are simply due to differences in the concentration
of the colouring matter, suitable factors capable of controlling such
quantitative changes must be formulated.

The chemical nature of the white and yellow pigments has been left
for the subject of a further investigation.

\Note added on July 19th, 1919.] Since the above was written, the
1918 broods have emerged. On the whole the additional evidence from
these insects (over 700) confirms the previous conclusions, but one fresh
point of interest has been observed. In all the families previous to
1918 (except Dr Doncaster's) no lacticolor was used for pairing, lest
this factor should complicate the case. In 1918 however the following
pairings were made :

18 V lacticolor $ [0*8 : 0*8] x lutea ^ [3-2 : 2-5]

18 III lutea % [3-7 : 4-4] x lacticolor ^ [0*9 : 0-8]

18 V lutea $ [3-2 : 3'5] x lacticolor ^ [0-8 : 0-8]

In families 18 III and 18 V, the females were all lacticolor, since the

cf parent was of this variety, and it was noticed that all these $ % were

H. Onslow 241

poculiarly pale, wheriMva inoHt of tho ^f if hiul an appreciable <|uaiitity
of yellow. In family *1H V, for inHtance, then* wen* Wl J % iuxl 20 ^f J ,
their ru8pectivo coloun* being iih followH:

Average cx)lour of % J wjw [11: 1 Oj

Avenige colour of ^^^^^ was [1*7 : \'h\
5 % % wen^ <leej)cr than [1-2:1 :i]

lOc/c/" L2<>.-J<iJ.

Similarly, in the small family '18 III, there were ^ %% aFid « c/</» <h«'
respective colours being jus follows :

Average colour of J ? wjus [11 : li]
c/'c/' was [21:1 •!)].

It was at first thought that there was some j)eculiarity in lacticolor
which inhibits the development of the j)ale yellow colour cli.ir;u't<'risin;4
the heU^rozygous form oi grossulanata. This seemed the innrc pn>l)al)l<!
becjiuse one of Dr Doncaster's brood, already recorded, namely, family

'08-5 lactkolvr ? [10: 10] x Uicticolor f [40:30] (see Fig. 1.')),

contained, in spite of the deep colour of the cT parent, 11 c/c/* '^"''
14- J J only two of which were deejxT than [12 : V\\ The remainder
had an average colour-value of [I'O : 10], there being practically no
diflference in the colour of the ^ ^ and the % % .

In family *18 V, the % parent was lacticolor and consequently both
ffff and J J were heterozygous grossulariata. A comparison between
these % ? and the lacticolor J ? of the reciprocal cross shows that it is
femaleness, or some factor associated with it, rather than lacticolor that
prevents the development of the pigment in the pale yellow^ /",$?.
This can be seen from the fact that the grossulariata J J were even
paler than the lacticolor J J from the reciprocal pairing, whereas the </(/
in the two families were approximately the same depth of yellow.

In family '18 V there were 23 $ $ and 23 (f^ the colours of which
were as follows :

Average colour of $ $ was [10: 10]
Average colour of (/J* was [1*4 : 16].

It was previously suggested on p. 219 that there was "a slight tendency
for the J J to be paler than the cTcT' ' ^^^^ this difference was not iiion-
than 8 — 9 ^/, in the heterozygous J J either of the I\ <>r F. generation.
Although this point has not been tested (except in faniily 'Ob')) by
Journ. of Gen. viii 17

242 The Inheritance of Wing Colour in Lepidoptera

pairing a white to a yellow lacticolor, the above explanation seems
probable. The question is however doubtful and the experiments are
being continued with a view to elucidating the problem.

I am indebted to Mr Raynor and Dr Doncaster not only for the
material they have given me but for much kind help throughout the
experiments and in preparing the paper. To Prof. Punnett I am also
grateful for encouragement and advice throughout the experiments.
Mr Udny Yule has been kind enough to give me the benefit of his
advice and opinion on the statistical points of the paper. I am especially
indebted to Miss Moodie for her constant care of the larvae, etc. during
several years, for without her cooperation the experiments would not
have been possible.

H. Onslow

243

TAHLK I.

Table to m%ahl« iwiivuimiU of any otufjamifi/ to br found in ritrrrn
comfHh^d of morf than mvn famihf.

NMMof YKl.U>WyTYPK

fMnllj {mm FVt l*^)

'17 B o* oH. 1<. »9. 7H. 7»

V V17. i'i. 47. 4«. 71. HI. 82

•17 K o' o 1*. J»*. •*•'*. •'*••. •'»>. 72. K\

^ V 1^. ••^4, 34, 57. 127

•17a o'o >.'-'. 5,7.25

'^ V^O. '-*«. 31, 121
•17c o*J«

V^ Vf>''>. 67. 111. 11*^ 114
'16P o* o4.fi. 10. 21 , 48. 52, 117. 12'>

V i 3. 8(). 131, 130

'16T S J. 35, 14. Ill), 125

9 V -^6, 59.84, 107, 122, 12fi, 128

•15 C o* J 32, fi2, 146, 148, 14'J. 150

V? V^. 115, 1-23, 140

•15F S 56

•15 K ^^.^40, 113, 133, 141, 144

V ^5:^, 145

'15 M o*o 13.20,37,38,41,42,43,50,

54, 68, 69, 73, 77, 108, 109,
120, 124
^ o 132, 1,37

15 Q ^ o^a5, 135

o V 16, 75, 110, 130, 138

14 B cJ J9, 23, 46, 76, 86, 104, 106,

116, 118
$ $ 30, 45, 105

'08-5» ^ ^21, 63, 64, 66, 87, 88, 89,
90, 91, 92, 93
$ 0 28, 58, 60, 61, 94, 95, 96,
97, 98, 99, 100, 101, 102,
103

'08-20* (^^74,143,147,151
$ 0 70, 134, 139, 142

YELLOW X YELLOW
(see Fig. 16)

'17D c^ (^29, 30, 37, 59, 60, 67, 76,

77, 78, 79, 90, 94
o c 24, 26, 88, 97

'17 F

o"o45, 58
o % 75, 87

'17 H

o"o46, 47
$ 0 20, 32, 36

17 S

o^43

17 T

^Jc?71, 81,83, 84, 89

futiiljr

'Mft
•I7«

•i7r

•17^

•17^?

'17 m
16 B

16 D

17 0

17 R

17 Y

16 H
'15 E

15 N
15 0
'15 P

15 R
'14 A

9 $3, 5, 8, 11, 12, 31,34, 35,
55, 56, 57, 62, 68, 70, 72

• Bred by Dr Doncaster.

VKLI/>W . VKM-n\V cmUnuwl

(«•«. KiK. HI)

?.>22.23

,' S 1, 6, 7,9. H». 13. II, 17, 1H

? ,;3», 52, 53

V . 1J>. 41. 6«

.;2

.;5i

V , 25, 93
.;oM9,54

V ,^4, 15, 18. 88

;27

J ;16. 21. 2H, 33, 40, 44, 50,
63, r>5, 66, (>9, 73, KO, «2. 92,
95, 98

9 V42, 61, 74. 85. 86, 91, 9'.»,
100

? 96

MYRRH). VKIJ.OW

(Hcc KIr. 17)

(^ (5^21, 33, 66, 76, 79, 80, 87,

88, 135
? $8, 28, 29, 119

(J ^16, 18, 22, 27, .30,41, 71,

72, 81, 112, 113, 116, 132,

139, 140
$ $ 47, 59, 73, 82, 94, 97, 100,

101, 102, 106, 118, 121, 128,

129, 130, 131

^ o'14, 24, 49, 77, 99, 124, 138
9 9 11, 12, 19, 20, 103, 110,
114, 136

S (?3, 7, 45, 83, 90, 111
992, 34, 65, 68, 120, 122, 142

.J ^10, 31, 32, 38, 40, 43, 44,

50, 68, 74
9 91, 6, 13, 15, 37, 78, 107

S c^46, 53,60

9 9 25, 35, 51, 52, 89, 92, 123

S 0^5, 23, 95
9 9 75, 91

^ ^4, 48, 57, 69, 84, 85, 86.

93, 104, 108, 125
999, 64, 141

9 9 96, 133

^ ,^\1, 26, 39, 42. 54, 63. 70,

105, 115, 134
9 936, 55, 56, 61, 62, 67, 98,

109, 117, 126, 127, 137

17—2

244 The Inlieritance of Wing Colour in Lepidoptera

TABLE I — continued.

Name of HYBRID x HYBRID

family (see Fig. 22)

'17 P ^(J20,22, 52, 67

$128

'17 Q (J (^8, 23, 24, 74, 75, 80, 83,

112, 175, 176
? $ 18, 73, 114

'17 UU cJ c^76, 148, 149, 164, 167, 177
, 2 $ 6, 108, 147, 165, 166

'16 C ^(^36, 41, 79, 84, 116, 189,

190
$ $ 188, 197, 199, 200
16 S (^(^183,184,187,198

'16 KF (J 100

'16 M (^ (J30, 60, 81, 168, 169, 186

$ $ 115, 129, 144, 145, 153

'16 N c? <?25, 37, 39, 46, 47, 68, 106,

126, 138, 142, 154, 162, 171,
178, 194

$ $ 70, 71, 72, 82, 107, 125,

127, 134, 139, 140, 141, 157,
158, 159, 160, 170, 172, 179,
180, 181, 193

'16 R (^c? 33, 45, 50, 56, 161

$ $ 113, 143, 155, 173, 185

'16 V c?^49, 137, 156, 174

$ $ 26, 27, 42, 99, 133

'15 D (^ (J 109, 130, 151

$150

'15 L ^ ,-?13, 16, 17, 38, 64, 110

? $ 21, 93

Name of
family

'15 S

'07-23*

'09-19^

'10-12*

'17 W

'16 0

'15 G

09-6*

HYBRID X HYBRID-continued
(see Fig. 22)

^ cJ28, 86, 117, 118

$ $ 15, 19, 29, 43, 85, 124

(^(^2, 7, 10, 11, 14,51,59,65,

94,95,96,101,103,119, 120,

182
$ $1, 3, 8, 31,32, 40, 48, 53,

55, 58, 61, 63, 77, 87, 88,

89, 97, 104
(J ^35, 69, 78, 131, 152, 191,

192, 195, 196
? ?4, 34

(J (J 12, 54, 57, 66, 98, 102,
105, 111, 121, 122, 123, 135,
136, 146, 163

$ $ 5, 44, 62, 90, 91, 92, 132

HYBRID X TYPE
(see Fig. 25)
^ c^28, 38, 39, 40, 41, 42, 43,

57, 58
$ $16, 23, 24, 25, 33, 34, 35,
36, 44

c?37

? ? 21, 32
f^(?22, 26
$31

(J(^l, 2, 3,4, 5, 6, 9, 17, 27,
29, 45, 46, 47, 48, 49, 50, 51,
52, 53, 54, 55, 56

$ $7, 8, 10, 11, 12,13,14,15,
18, 19, 20, 30

N.B. All the above numbers denote positions of insects along the base lines of the
following distribution figures.

Bred by Dr Doncaster.

H. Onslow

245

^un-Jno|00

246 The Inheritance of Wing Colour in Lepidoptera

5: (0 5-0

It

O 3

O

Mean colour-
values of lutea
parents
[3-9: 3-7]

[3-6 : 3-4]

10

20

40

60

80 K

No. of inse

Fig. 16. (Cf. Fig, 3.) Curve showing the distribution of the orange and yellow colour- values of the offspring

from 13 pairings of lutea x lutea.
These families included no variations.

H. Onslow

° I

I. B

.' ^

3 a

( ,*

^, a

\

X "P*

•3 O

2

= 1

_2
s

3

a.

-':^

E

11

f^

a

o —

o >

•^

^r "^

^ o^

ll

?.

;n

S"3 5-: c4

fti 2

O J2

5 '^

as =

•n <D U^

j:4

o ^

"S to

> c

i ^

S OS

^ fl

o o

o •-

-»<

11

- Of

TS X~

•<

§ ^

kH

■^

Oj x"

t^

=

« r

OS OJ

s .s

'5

2

<V £3

-^ -s

;a4

u-l )-•

O ^

o -

ti i-<

-^ o

_,

n ^

1^

•ji

' "\

4:

'^ .—

-V

^ <j

^

-J cS

^

^

^ d

t^

c

) .= >

o

r ^ tf

.^

'~

248 The Inheritance of Wing Colour in Lepidoptera

W'l

1111111111

iViiumiz

80

ibc

No, of insects

Colour-va
of lutea

[4

parent
•3 : 4-9

Colour-v£
of hybr
parent

[0-7 : 0-5

Fig. 18. (Cf. Fig. 5.) Curve showing the distribution of the orange and yellow colour- values of family '17 C

hybrid (lutea x gross.) x lutea.

Two separate curves are added showing the black values of the fore wings, and the orange values of the hind wings.

The following variations occur in certain insects, whose position along the base line of the curve is indicated

by the accompanying numbers.

nigricostata-h ^ ^^36, 37, 38, 45, 49, 51, 52, 53 (radiated), 58,
61 (radiated), 62, 70
„ ?? 3, 14, 30, 63 (radiated), 74 (mVrociwcta /)

///^c?50, 69
nigrocincta // 2 35 (slightly radiated)

H. Onsuow

249

i'|^!ijjliij!j!|||!!f!i;»!{;isi|

I

Colour-1

of lutf

paren

[2-9:2

100
No. of insects

Fig. 19, (Cf. Fig. 6.) Curve showing the distribution of the orange and yellow colour-values of family

'10 E, hybrid (lutea x gross.) x lutea.
The following variations occur in certain insects, whose position along the base line of the curve is

indicated by the accompanying numbers.

Very broad orange fascia $ 35.
nigricottatd ^ ^32, o 966, 70 nigrociiicta $ ? 2, 23 (segments 1, 2 and 3)

Hybi
parent

/i^^Gl

fi 9 29

250 The Inheritance of Wing Colour in Lepidoptera

5-0

4-0

3-0

2-0

1-0

;:

tA

m

m

ii>

m

:6:

M

Til

n'-

<H^

<m

m

m

af

^

ml

¥t

Bm

*^i

10

20

40

60

80

TOO

s ■'.-.' >,■ No. of insects

Fig. 20. (Cf, Fig. 7.) Curve showing the distribution of the orange and yellow colour-values of family
'10'15^ , hybrid {luteax gross.) xlutea. Parents missing.

The following variations occur in certain insects, whose position along the base line of the curve is

indicated by the accompanying numbers.

r (^ (^2, 4, 5, 14, 20, 23, 24, 25, 27, 28, 30, nigricostata ^ ^ 19, 20, 23, 34, 36

lacticolor i 31, 33, 34, 35, 36, 38, 30

I $ $ 3, 6, 18, 26, 29, 42, 50, 51, 52, 53
( c? (?9. 11, 19> 20, 23, 27, 28, 39, 46, 48, 49
fulvapicata {551,6,10,14,17,18,3

/ 051, 10, 15, 47
// ^c? 13, 37,41

-hH-^^S, $45'
semiviolacea^Q
A. gross, (type) ^12, $ $8, 22
'^ Bred by Dr Doncaster.

36, 43, 45, 51

H. Onsu)W

2;") I

3olour-value
of hybrid
parent
[1S:0-71-

100
No. of insec

K., ., (Cf Fi.. H, Curve showin. the distribution of the orange and yellow colour-values of
^"^- -'• ^ fa.nily 'IG A, hybrid {lutea x i,ros,.) x lutea.

The following variations occur in three insects

whose position along the base line of the curve
isTndicated by the accompanying numbers.
nujricosiata Hi I ^3 „,^,,,,„,„ . 7O (segments 2 and 3). Broad orange fascia o O 09, 72

nigrocini:

ta I ^

(0

i-H

04-

-Of

^

f*

s

1— i

z ^

«

.

TjT

<M

X

1

S

a
to

a

1^'

«o

rH

>5

111
1.11

s r* s

•X-

Ci
I— 1

Of

V.

1

of

o

r— i

OS

OS

II

1

5

1

•X-

c3

(M

^Of

a

o

SP

t>»

.o

*>

crt

• i-c

»-i

<1)

b

«

s

^

»o

&

!^

CO

CO

^

r^

fi

-^

^g

^

2

CX)
S3

03

s

05

«

00

CO

^

^^

p

r>

s

H-

^

o

g. ,

>

«5

5
5

00

o

,£3

1 '

P^

.o

CO

CO

00

-3

O

a

a

'a

05

00

W5 C

!>

1— 1

»c J:

^

P

®

00

CO :

O

ffl

o

iO -p

o

^

rH

oo" 7

(U

-* c

^

-f^

>-,

lO

^•"^ ?,

o

u

rH

■^ Zj c

o

a

"looOf-^

<8

O

^

.

G^TiH ^'xi

o

o

a
o

d*
t^

2

00
-<*

rH
Of

e

CO

a

1 II

-2 s ^

•^

CO

1— 1

"

tH

o

ci

S

"^

K

"^

»

J3

.1-1

>%

S

►>

V

a

^

:;3

c8

d

fe

a

V

^

o

fe

fJH

<X>

o

a

fl

^

OS

t>D

iH

I-H

a

as

I-H
00
I-H

•'H

13

^

fH

^

1-t

^

o

Jl

t- i-H

OS

a

giof

(>» t-

>

Of

00 rH

'T"

bD

OfCH

.

O
i—l

1

t-

a

^^

rH

t- t-rH

o

3

Of

i

CO Of

CQ

r-i

^,^

i

:: ::

a

e3

t

^

1

§ g s

N

\^

un-jno|Oo

H. ()Nh5L0W

2"):]

>lour>value8

of hjbrid

parents

[1-3: M]

tl-5:l-4]

80 TOO

No. of insects

Fig. 23. (Cf. Fig. 11.) Curve showing the distribution of the orange antl yellow colour-values of family

*17 E, hybrid {lutea x gross.) x hybrid {lutea x (jross.).
The following variations occur in certain insects, whose position alon^ the ba.>^e line of the curve is

indicated by the accompanying numbers.
niprocincta (segments 3, 4 and 5) o60 niffrocincta a \ ^ 4, %^, 1»7

254 The Inheritance of Wing Colour in Lepidoptera

Colour-vali

of hybric

parents

[10 : 0-6

[0-9 : 0-4;

100
No. of insects

Fig. 24. (Cf. Fig. 12.) Curve showing the distribution of the orange and yellow colour- values of family
'16 G, hybrid (iwfea x gross.) x hybrid (lutea x gross.).

The following variations occur in certain insects, whose position along the base line of the curve is

indicated by the accompanying numbers.

nigricostata -f-^ ^ IS, 15, 4:5, Gl violacea^2'i semiviolacea )

„ „ (right wing only) ,^62 semiviolacea $ $ 7, 8 (fore wings only) nigricostata + S

„ (left wing only) (^47

$27

H. Onslow

or,

>;>

ftOfjTT:

i:::

ti:::::::::

••■■lasaali

::::::::»!

•■••■■■■■•a

:::::::::::

IhI;III!II

iuh

imiHliiiiiiiiji'l

TTTTjTinrT^mTT

Mean colour

value of

hybrid

parents

- [1-4: 101

Mean colour-
value of groxg

parents
• [0-9: 0-6]

100

No. of insects
. 25. (Cf. Fig. 14.) Curve showing the distribution of the orange and yellow colour-values of the offspring from

four pairings of hybrid {lutea x gross.) x A. (jrossulariata.
fcddition to the 8 o 2 , 9 ^ r^ included above, family *17 W contained 64 insects whose colour-value was less than [07 : 0-5].
i following variations occur in certain insects, whose position along the base line of the curve is indicated by the

accompanying numbers.
nily '17 W. hazeleighemis ^ ^38, 3<J, 40, 41, 43

mpunctifasciata -f- ^\\, JIG ,, o 0 33,36

iS^c>28, o G23, 24, 25, 44 Family '096 ^ •

uZrapirafa 9 23 nigricottata i ^ ^'6,^, 17, 27, 45, 46, 48, .50

nigricottata fff ^ <^28, 44, .58 lacticolor ^ ^ 1, 2, 3, 4, 5, 6, 9, 17, 27, 29, 45, 46, 47, 48, 49

„ 9 0 34, 35 50, 51, 52, 53, 54, 55, 56

$ 97, 8, 10, 11, 12, 13, 14. 15, 18, 19, 20, 30
%■ Bred by Dr Doncaster.

256 The Inheritance of Wing Colour in Lepidoptera

DESCRIPTION OF PLATES.

PLATE IX.
Abraxas grossulariata.

Beduced two-thirds natural size.
1—12. Var. lutea.

1—6. Orange colour-values from 3-2 — 4*0 and over.

1. Var. Jlavipalliata $ .

2. Var. lutea ^.

3. Var. fulvapicata (J .

4. Var. lutea $ .

5. Var. lutea $ .

6. Var. lutea <^.

7 — 12. Orange colour-values from 3*2 to 2-4.

7. Var. Jlavipalliata ^ .

8. Var. lutea $ .

9. Var. Jlavipalliata ^ .

10. Var. lutea ^.

11. Var. semiviolacea ^ .

12. Var. lutea ^.

13—24. Pale yellow hybrids {lutea x A. grossulariata).
13 — 18. Orange colour-values from 2-4 to 1*6.

13. Pale hybrid^.

19—24.

*14.

Pale hybrid $ .

*15.

Pale hybrid ^ .

*16.

Var S^iorocincta ++ ■
) niyricostata H+

*17.

Pale hybrid (^.

*18.

Pale hybrid $ .

Orange

colour-values from 1*6 to 0-8.

19.

Var. Jlavipalliata $ .

20.

Pale hybrid,^.

21.

Var. ^>^*'^i'^^**« c.
I impunctifasciata H- '

22.

Var. S^iSricostata^
I semiviolacea

23.

Pale hybrid $ .

24.

Var. impunctifasciata H $ •

* [Owing to an error in the reproduction which could not be corrected, insects
Nos. 14, 15, 16, 17, and 18, especially the last, appear very much too dark in colour. They
should be somewhat lighter than Nos. 12 and 13 and slightly darker than the insects in
the next column, Nos. 19 and 20.]

IT. Onslow 257

35 and 36. A. ffro^ttiiiirinlrt. OmnKroolour-TiUuM from 0*0 to 0-A.
35. Type ,\ (fWxIcy, Knit.)
26. T>|»o ;hn'<l J»y W. Nrwiiian.
37. Wr. frmiriohtcrit \\

t fuU'iipicata
'lA. Var. I niftricnatnta y .

\ armivioUtcrtt

'i[h Xnr. niijrofpitriiata ^. (Boxicy, Kont.)

( httirUiqhen$i»
MS. >«r. ■ :.

{ nujuH'inctn /

31 —60. Var. Uicticolor.
81 — 43. Var. cUr\j*o»troUt.
81 — 3(>. Orant^e coIoiir-Taliio** from .'^2 to 40 and over.

31. \m. chrytontrotii \; .

32. Var. c Ary«()*/r<»t/i '^ .

33. \m. chry$ontrota ^.

34. Var. chrytoftrota "J .
S.^i. V&T. chry$o*trot(i ^ .
3»>. Var. chrynottrota 9 .

37 — 4*2. Orange colour-values from 3 2 to 2 4.
37. \fix. chrysostrota ^.

as. Var. (/'-!'«'•""'"" 5.

/ fiitrnpicnta

\ chrysoHtrota
\ fitlvapicata '^ '

40. \rtr. ' • Q.

( ftilvapicata

41. Y&r. chryitoiftrota^ .

42. \&T. chryaostrota :^.

43 — 54. Pale yellow hybrids {chryKostrota x lacticolor).
43—48. Orange colour-values from 2-4 to 1-6,

43. Var. \<-f"WO»^^ota

ifulvapicata
/ chrysostrota

44. Var. -| fulvapicata 9 .

( nigricostata

, , „ ( chrysostrota ^
4o. Var. { ^ / . (^.

{fulvapicata

46. Pale hybrid.

47. Var. S'^^V^O'^'rota

f fulvapicata

48. Pale hybrid.

49 — 54. Orange colour-values from 1*6 to 08.

49. Pale hybrid 0 .

50. Pale hybrid 9 .

51. Pale hybrid J.

52. Pale hybrid $ .

53. Pale hybrid $ .

54. Pale hybrid $ .

Joum. of Gen. viii 18

39. \s

258 The Inheritance of Wing Colour in Lepidoptera

55 and 56. Orange colour-values from 0-0 to 0-8.

55.

Var.

lacticolor $ .

56.

Var.

lacticolor $ .

57.

Var.

lacticolor $ .

58.

Var.

radiata $ .

59.

Var.

cupreofasciata 5 .

60.

Var.

iochalca^.

Nos. 1, 7, 59 and 60 were very kindly given to me by the Kev. G. H. Eaynor, all the
other specimens were bred during the course of the experiments.

PLATE X.

1 and 2 were sketched with a Zeiss AA objective and ocular 5 giving a magnification
of about 200 diameters.

3, 4 and 5 were drawn with the camera and a Leitz 1/20 inch oil immersion objective
and ocular 9, giving a magnification of about 1360 diameters.

1. Portion of fore wing of var. iochalca (see Plate IX, No. 60) showing the way in which

the melanic pigment is diffused throughout the scales. The base of the scales contains
most pigment but the effect is masked by the pale overlapping tips of those in the
preceding row. There are a few grey scales, but not sufficient to modify the colour
appreciably.

2. Portion of the fore wing (near the costa) of var. nigrosparsata with yellow, not white,

ground (as in Plate IX, No. 29) showing the way the melanic pigment is concentrated
in certain scales. The tips of the melanic scales usually contain more pigment than
the base, and further the yellow scales are almost devoid of black, which gives the
insect a faintly mottled appearance.

3. 4, and 5. Unstained sections of the scales, about 2 to 3/a thick, of var. lutea from

specimens of varying depths of colour, showing the way in which the yellow pigment
is diffused throughout the chitin of the walls of the scale. Even in the deepest
yellow insect (3) there is no granular pigment.

JOURNAL OF GENETICS, VOL VIII. NO. 4

«^.'"V

S?"

'>•

^•■>

Sjc^N ^ r^ ?fe^> «&-?' ^ ^••>>

g^^ Ji' p #> p ^^

^;- ~5

.?)

#^-= 0^

^•H) %) %.-% ^f-^ ?fc,^ ^<:-.

*»^.- >i- M" 34:'- -M^ ^=

!*i^ ais>0

^S

&0 CO

I

O

PLATE IX

'' ^-■'

- QS^y-'.

^•v:

4.

1^":^

-li^^^

'Tu^'

JOURNAL OF GENETICS. VOL VIII. N0.4

PLATE X

"K^^

/\

^-r^s.

.VI n\l cfir-jn: liXJ.

C»-aorida'' "Jni v«ru\y rT

STUDIES IN THK HYHIUD HISTONINAK.
TIL THE STIMULUS OK HETKROZYOOSLS.

By J. W. MESLOP HARRISON, D.Sc.

(With Two Text-figures.)

Pahtly owing to the inriuence of Darwin and those who succeeded
him, and jwirily because it harmonises with the observed facts, the
dictum that cross fertilis^ition is a source of strength or of stimulus to
metabolic activity hjis become almost axiomatic ; very few workers, how-
ever, seem to have sjx'culated Jis to its method of action.

Whether the phnise be restricted to the actual cross fertilisation
of plants or extended naturally, as ever3'one understands, to include
the act of avoiding inbreeding in animals, some explanation of the
phenomenon must exist, and it is now proposed to consider what light
hybridisation experiments in the Bistoninae throws on the matter.

Almost immediately the experiments were initiated it was discovered
that the hybrid larvae were not only emphatically more robust than
those of the weaker parent, but they also surpassed in strength and
vitality those of the stronger form. For instance, attempts repeated
yeiir by year to rear Ni/.ssia gniecaria have uniformly failed ; yet when
this species was mated with Lycia Inrtaria the resulting larvae wt^-re so
sturdy and strong that, as larvae, their mortality rate was negligible
and, moreover, they fed up in the amazingly short period (»f six wcrks,
thereby anticipiting the pupation of the more vigorous parent by no
less than four weeks. Nor must this be deemed an isolated or exc('i)-
tional case; to a greater or less degree it illustrates the condition of all
the crosses.

Since there were ditierences in these partieulars, as soon as the s»t of
experiments was completed, an endeavour was made to correlate the
degree of sturdincss and acceleration of development with other facts,
and this had striking success, 'leaking the ring <>! hybrids in which
Lycia figures as the central genus, not only because that circle is
practically complete, but also because the* crosses concerned in it had

18-2

260

Studies in the Hi/brid Bistoninae. Ill

been reared with elaborate care, during the same season, and under
conditions likely to eliminate phenotypic variation in all its phases, for
another purpose, it was perceived that, as the phylogenetic divergence
of the second species in any particular cross from hirtaria increased,
there was a concomitant and proportional increase in the physiological
robustness of the hybrid organism which, in its iturn, entailed :

(1) A size increased beyond the theoretical expectation.

(2) An acceleration in the speed of feeding up. of the larvae.

(3) Great disease resisting powers.

(4) An enormous reduction of the time of lying over indulged in
by some of the species and which, presumably, should have affected any
hybrid in which that species took part.

Nyssia graecaria

Nyssia

PoecMopsis rachelae

Poecilopsls Isabel lae

Poecflopsis

PoecUopsis
Lycia ursaria Lycia hirtaria

zonaria

lapponaria

pomonaria

J. W. II. Harrison

201

Thf,s*» (H)rr%*lationH ran Ih' HinipliHctl and rrndfriMl innn» vivi<l hy the
npptMult'd Uiblo aiul K'^'M''** ^*'*^' bl'fnrr K'^'"K ^^rnr, it will pmvj* vi-ry
ludpful Ui j(ivt» a phylo^t'iu'ticnl tnM' of the H|H»rirH invdIviMl.

Hofoiv pnK't'iMiin^ t«» t'lnphasisc the most siilimt it( tin* n-latiniiM
gniphirally. it will bo wril t«) rxpami ami «Iiici(lat<* tlu* points nftTHMi
to in {Mimgraph 4 above and plaod in the l.ust cohnnn of the appended
t-abular Htjitenient..

of

cmmiiMtndinii

male

of

hlrtaria

malo

M<«n of
la- 1 two

Winir pt|MinM

of the iirnh'

h)t»riin>ftwwn

ijtHl two

InrrMwr of

li)hrlil

wiiiK i'X)Minit**

oviT itieaii

l'erc*-liUM(e

incrrawj
over mean

Orclrr of
pu|«tlon

KfTertuf

livbrldltyon

lylntf ijvrr

power* of

pupa*

pOMONOrtd 1

3*2 mm.

4*2 ri mm.

37"*2."» mm.

38 mm.

•7.'»mm.

2

LaHt

Nil

Poeciiopii* 1

31*>min.

■VIT} mm.

37 mm

3M'4 mm.

1-4 mm.

3 «)

Fourtli

Sli«ht

Poecih/uit |
lappomaria \

3-2 mm.

l"2-r» mm.

37*2.'>mm.

31J-.') mm.

2*2r> mm.

«

Third

(ireal

icmarin \

29*7 mm.

1*2 '5 mm.

:{(')■ 1 mm.

'.i'J mm.

'2'.» mm.

H

Sj-cond

-

35 -io mm.

•l'2-.'i mm.

:\HH mm.

401 mm.

1-3 mm.

3 •3'

Firnt

(treat

* Not roannl in the wimo season imd under the same envintnmcnt conditions an tlie
others.

All of the Bistonine species, but more particularly the Non-Hoannioid
section, which is of more interest to us at the present junctun^, instead
of emerging from the pupae together in the spring following pu|>iition,
agree, inasmuch jvs jjercentages, varying with the species, persist in
lying over as pupae from two to eight winters. Of the species here
considerefl, Xi/ssid <jraecaria is the most notorious offender as it rarely,
if ever, emerges in the second year, but commences by providing small
detachments in the third yeai*, followed by similarly small instalments
each se;v.*^on up to the eighth. L(ippo}iaria is the next W()rst culprit;
a limited number appear in the second year, the remainder appearing
at intervals in the ensuing three years. Ni/ssia zonaria is extremely
erratic in the actual numbeis lying over, but all come out before the
thinJ spring; Poecilopsis isahellae is nnich the siime, whilst in Lycia
hirt(iH(0 and Pitecilopsis pomonaria nearl}' all are bred when expected,
a very few remaining unchanged until the third year.

And the effect of hybridity is that the fui'ther the habit, in this
respi'ct, of the .second si)ecies in the cross from Li/cia, the greater the
acceleration in enu'rgence.and the nean-r its habit approaclns L. hirtaria
the less the state of ati'airs is altere(l, e.g. cross Lijcin hirtarKt and Xt/ssia
graeairid «tr L. /nrtarid and PttecUopsis Uipponarid and all emerge the
' Except extreme northern races whicli were not employed.

262

Studies ill the Hybrid Bistoninae. Ill

first year; however, mate L. hirtaria and P. pomonaria and the con-
ditions obtaining in the two parent forms are unchanged, the same
percentage exactly lying over.

With this explanation, we can now proceed to plot our graph showing
the relationship between the phylogenetical divergence of the female
parent from L. hirtaria and the increased vigour of the hybrid organism
as expressed in its size.

Y

3

-u

C/3

C
CO

a

If

•- E

(fi >M
cd o

V o
.E ^
<]>

bo

cd

(5!

Hirtaria xzpnaria

Hirtaria x lapponaria

Hirtaria xisabellae

X

•X Hirtaria xjrraecana*

Hirtaria x pomonaria

O Phylogenetic order

Isabellae is on a side chain

* This brood was not reared with the elaborate precautions taken in the other cases
to avoid phenotypic variation as, for obvious reasons, it is impossible to get new material
from Carniola.

The correlation between the two conditions is so exact as to give as
the curve of relationship a straight line.

In other words, the stimulus given to the organism by hybridity is
inversely proportional to the physiological affinity of the participating
species or, otherwise, it is directly proportional to the physiological
divergence, and the same holds true of all the other manifestations of
inciting force. Had mere " cross fertilisation " (to adapt the word to
include the fusing of gametes the affinities of which are more or less
remote) in itself been the sole inducing cause, it is exceedingly difficult

J. W. H. IIahhison -203

t^ HtH' ln»w th<»n» should Imvi* iiriMiii this pnntveji!niH> stiintilation IxTnum*.
when all is sjiid and cinni', all tin* s|Mrit'.s ••mployrd an* <li\ IiIimI hy pn*-
cistdy X\\v sani»» nunilHT of j^onrnitinns fnmi ih«' main l.tjcin st<K'k. It
musts tluMvfoiv. Ik» flin^'tly <lf|Hnd«'nl «ui tin- runmlativ.- ditVrn'ncrs
U»twtH*n xhv raot4»rs building up lln' various ^mutyprs (or iMntyjH-.s, it
preforrtsl !).

But. how ha\M' tht's«* s|M»cit's ;iiis«'n .-nKl ln>\v {\n tlnv ditVrr ' Wlutln-r
thoy have Ihm'U t'volvrd by Ljfnirraphical isolatinii (.r by mutation th«'ir
dit^'eivnco tVom tlu'ir innn<'diat<> phyl<>i^('n<'t ic ancestor ap|H'an'd jus a
change in thr valuo of th«> t^i-nrs ritlwr by loss, addition, dujilication or
othonviso ; wh«'ntH* it is i'vi<l»'nt, that.. ;is wr pass from the Lycia stem to
the furthest rt'nu»ve<l tbnn (physiologically and |)hylo^eii(ti(alIy ), wr
have. ;ki/*i* passu, accompanying ilirterences iu what wen- originally
homolog«)us genes plus the apjM'anince of genes not re|)n'sented in the
original stock. Therefore, if we |),iir A. Inrtavin with any <»f its (h-rived
fonns, we i\w geneniting zygotes' the <legree of h<'terozygosity of which
depends entirely- ujx>n the extent of the <liv<'rgence of th«' second form
from hirtaria. And this corollary must follow then-from, that the in-
cre;u<i'd and progressive met^ibolic action visible in the hybrid series is
directly and positively correlated with the advance in the heterozygosity
of the individual hybrid membei*s.

Nix?ess;irily, of couinc, a pr>int must be reached when, in the end,
total or pjirtial incompatibility^ of the two setsof chromosomes involved
will step in c^irrying with it, in some, physiological interference with
the normal development of the zygote and, in others, the total failure
of fertilisation ; thus limits are set to the heterozygotic acceleration of
the activities of an organism by means of hybridity.

As to how the influence of the heterozygosity works there are
several possibilities open. Firstly, one must take cognisiince of the
fact that any intruding spermatozoon consists of little but the nuclear
structures which, coming into intimate contact with the cytophusm
of the ovum, are so placed as to be able to act, react, and interact
with that cytophusm which presents it with entirely novel conditions of
environment ; thus the stimulus may be brought about. Secondly, the
heterogeneous nature of the zygote, the conseijuent conflicting tendencies
of its genes, the extra work induced by the interaction, all combine t<»
secure the laying down of great stocks of cytoplasm and stimuiat** to
an extraordinary degree cell-division, not only in the actual number of
cells, but in the sj)eed with which they are elaborated. On this view,
* As when Lycia hirtaria j and PorcilopKix nicfwliie , are mated.

264 Studies m the Hybrid Bistoninae. Ill

the greater the number of diverse determiners (so long as they are able
to work without upsetting the stability of the cell thereby causing it to
perish) the greater the stimulating effect.

Again, the stimulus may result mainly from the presence in any
given cell of a greater number of units than it was designed to receive,
causing a development of increased capacity to allow for these extra
units, and thus increased and more rapid growth.

In connection with this, it is well to note that, in reality, the size
of the cells in the only hybrid I have critically examined was actually
greater than that of either hirtaria or zonaria, the two parents.

Lastly, it must be remarked that the Mendelian factors are not the
only reacting features ; there are assuredly many nuclear elements not
of that type, although such, very likely, predominate, and the activities
of this element and the resulting impulses must have some effect.

It will be clear that, up to this stage, only heterozygotic stimulus,
as it appears in the F^ generation of hybrids, has been considered ; the
reasons for this are plain. Firstly, the F^ generations in the hybrid
Bistoninae are, for the most part, completely sterile, and, secondly, each
and every zygote generated in any particular cross possesses the same
intensity of heterozygosis, not only in the actual number of the opposing
factors, but in their quality. Every gamete thus is stimulated to exactly
the same extent, and, if we make due allowance for that variation which
would be exhibited even by a pure line, will reach approximately the
same size in the same time.

But when, as in the pomonaria-isabellae hybrids, the F^ generations
are fertile, a new set of circumstances arises on pairing these inter se\
circumstances, let it be emphasised, not those appearing in F^ insects
of a mono-Mendelian hybrid origin or of hybrids differing by a
relatively small number of opposing genes, but those developing in
an i^2 generation in which the interacting and conflicting genes reach
enormous figures. How soon, under these conditions, the number of
heterozygotes reaches appalling dimensions, and how quickly the homo-
zygotes descend to negligible figures may be gathered from the following
table :

Number of

Percentage of
homozygotes

50
25
12-5
6-25
3 12
100
2"

Number of

opposed genes

Number of

Number of

Number of

in the parents

individuals

homozygotes

heterozygotes

1

4

2

50

2

16

4

12

3

64

8

56

4

256

16

240

5

1024

32

992

4» 2» 2»^{2"-l)

J. W. H. Harrison 265

Nor Ih it t<» ho HUp|)«imMl llmt all <»f* thr hi*tvrnzy>;<»t<*H iin* h«»t/<'rnzy^<»iiH
Ut the wiiiio (U'^HM', for that in far fnmi Iwin^ h«». Thry pnKMMMJ fmiii
inunohybridH. throu^rh every HUi^e of inatnK'linouH ami |MilnK'liiinuH coni-
plexe8, right up to the full heU«ro/.ygouH coinphiiu'iit of thr /', frat«Tnity.

Hence the stimulating iinpuUe of het«Tozyg«».sity in f<'lt with uni<|ual
foroo in the diffep'nt zygoti»8, luul they attain vastly dirtrn-nt h'\/a'h.
Thus, as in characteristics so in size, the F^ generation shows cnnrrnous
variation as coinjMiretl with the conijNiratively even size and ap|M*anince
of the Fi lots. This state of atVaii*s wjis displayed vividly in the F^ and
Ft generations of the pomoiiarid-isabellac crosses.

Manifestly therefore, by the men* stimulus of tln'ir heterozygosis,
independently of any action of multiple genes for size and weight, th«*n'
will lx» ap))aR>nt size segregation in the F^ insects. Any att«'mpted
genetic analyses for size purjxjses which fail U> allow for heterozygotic
impulses are vitiated and useless.

To illustnite by examples, size and weight exjK*rimeiits have been
undertaken with guinea-pigs, poultry, rabbits, etc. in several of thes<\
dwarf strains of sjx'cial races have been mated with full-sized examples
of other breeds, thereby bringing into play unknown numbei-s of genes
resulting in the production of an equally uncertain number of hetero-
zygotic composites in the F.^ animals carrying with them undetermined
and indeterminable degrees of size acceleration. How can it be p<j.ssible
with this unknown factor interfering to conduct any analysis of size and
weight determiners ? — hence my opinion, expressed above, that many
of these analyses are useless. They are not only valueless because
heterozygosis will produce size groups of its own, but also because its
effects, added to those of the size and weight determiners themselves,
serve to throw accelerated specimens into groups to which on the merits
of possible weight and size factors they do not belong.

SEX INHKHITANC^K IN rHDICUIJS HUMAXrs

VAH. coiiroinsK

Hy KDWARl) IIINDLK, M.A.. Ph.D.,

Charles Kiii«fsl('j/ fA'cturer ami Ihio-Felhuv of Ma(fdideue

College, Cii mhridye.

{From the Quick Lahoi'atory, Cambridge.)
(With One Chart.)

The observations on the inheritance of sex in Pediculus hutuanus
\'iir,corj)oris, which form the subject of the following pages, were originally
commenced in the winter of 1912, when the writer made some (*x|)eri-
ments on rearing these panosites in connection with the transmission of
certain infections. In the course of these experiments it was observed
that, under the same conditions, very unequal numbers of males and
females were obtaine<l, and, therefore, individual pairs were isolated and
their offspring reared separately. It was then discovered that three
kinds of families might be obtained, namely, male, female, or male and
female.

Many series of experiments were begun, but only two of them were
carried through more than one generation, as the conditions of rearing
were not understood. Eventually one series was carried thn^igh to the
second generation, and another one to the fifth, the latter having to be
discontinuefi at the outbreak of war. The results obtiiined, therefore,
are incomplete, and much work remains to be done on the subject, but
as the writer is unable to pursue this line of investigation at present,
the records of the exjx'riments are being publishe<l in the hojK' that they
may be of assistance to students of the inheritance of sex.

In my absence Profes.sor Nuttall kindly published a short account of
some of the.se experiments in Parasitology (Hiiidle, 1917), but unfor-

* The expenses of this Investigation were defniyed, in part, by a Grant from the lioyal
Society.

268 Sex Inheritance in Pediculus hiimanus var. corporis

tunately certain errors have occurred in that article, and it is very difficult
to trace the histories of the various families. The present account con-
tains a more complete record of the experiments, and also corrections of
the former paper.

The cytological examination of fertilization in Pediculus has been
undertaken by Dr Doncaster, who is continuing the study of the sex
problem presented by this parasite.

Material.

In all cases the lice used were collected from verminous clothing
sent from London by Dr Hamer. The clothes-lice for any particular
series of experiments came off the same piece of cloth, and presumably
from the same host. In all cases they belong to the race Pediculus
humanus var. corporis, and there was no evidence of the presence of
var. capitis, although the possibility of the occurrence of hybrids between
these two races should not be ignored, especially in view of the results
obtained by Bacot (1917), and Keilin and Nuttall (1919) on hybridisa-
tion in lice.

On the arrival of a batch of lice, the larvae were removed from the
adults and placed on a piece of black cloth in a glass tube. This was
kept in an incubator at a temperature of about 30° C. and the lice were
fed twice daily, morning and evening, on the arm\

As soon as an adult emerged it was removed to a separate tube, in
order to avoid the possibility of any unknown crosses with other emerging
adults, and in all cases where a female was found in the same tube as
a male, it was not used for any breeding experiments.

A selected male and female were then placed together in the same
glass tube and provided with a small piece of black cloth, on which to
hold and to lay eggs, and, except in the case of a few experiments, were
kept in the same tube and fed together until they died.

When the lice were placed on the arm they usually began to feed
at once and gorged themselves in a few minutes. The female com-
menced to lay eggs about 24 hours after emerging from the larval
skin, and generally continued to lay 2 to 4 eggs daily for about four
weeks. It should be emphasised, however, that the number of eggs laid
by a female fed only twice daily is considerably less than the number

I. '
1 This method has now been improved upon by Nuttall (1917) who obtains much
better success with his " pill-box method," whereby the parasites may be kept continually
on the body.

E. UlNDI.K 200

(KXMirring wht'ii individualH an* krpt cniiiimuilly on tlu* UmIv, iindcT
which iMiulitioiiH iws many Jts !>'7 tlaily havt- Ihmii nronltMl.

Ill my ox|K'rimont..s {hv (vi\\n\o nirrly laid iin»r«» than SO i*^^ all told,
an«l in many vxim^h mnsidtTahly fowrr. so that the cunditiofiM imiMt Im«
n'ganli'<l jis having bt'i'n somewhat unfavourahlr.

Tho larvao usually hat^'hrd iVom thf c^^s aftrr 10 nr II <liiys, and
niuuIUHl thnv tinu's Ih'Auh' hriMimin^ ;ulult. Thr ti?st nmult ^mi'mlly
tKXMirnMl aft*T (» days, an<l the two t'olloNvin^^ rach afl^T 5 days, s<» that tin-
U»Uil drvclopmont from tiixt sta^c larva to adult usually t<M)k ahout
16 days. Thoiv wa,s a ciTtNain amount <»l" mortality amongst the larval
stJi^i's, JUS simwn in the rrcM»nls on pj). 270-271, wluro ihr nundur of
larvae emerging is shown in each case.

When feiniing such a lari^e mnuher **( lice it- wsis sometimes necessary
t<» plact* tw»» or three hnMnls on the arm at the sjime time, hut when-
ever jH>ssible thi'v wen' fe<l successively, in order to avoi<l the |K»ssil)ility
of their y^ettin^ mixi'd. In one or two instances, h(»wever, it is pmhable
that indiviiluals have wandered frtun nne hnnxl to another, as shown
by the counts of the larvae before an<l afti-r fe('^lin;^^ In all these causes
the expi'riment is marked with a «pier\'.

ImuRHliately an adult appeared it was isolated, ami it us<'d for
furthiT ex|K'riments w;is crossed with one of the (►pposite sex as soon
as possible. For practical reasons the adults could not be kept alive
until all the members of the brood had completed their development,
and in consequence some of the crosses had to be made with material
that Wiis available rather than that which one would have selected.
In view of the onerous task of feeding, all adults not intended for
further experiments were killed an(l preserved, in order to keep down
the numbers to a minimum.

The reconis of these experiments are summarised in the fV>llowing
table, in which is shown ( 1 ) the number of the family, e.r/. /^8 or Z 15 ;
(2) the nature of the female parent, e.g. ex ^l 8 signifies that the female
came from the brood of family ^4 3 ; (8) the nature of the male parent
similarly expressed ; (4) the total number of larvae that hatched ; and
(5) the number of each kind of sex among the adults which reached
maturity.

In some instances the same male, or the Siime female, was used for
more than one exi)eriment, and these are indicated by footnotes.

270 Sex Inheritance in Pediculus humanus var. corporis

RECORDS OF EXPERIMENTS.

First Series.

V umber of

Female

Male

Number of

Number and sex

family

parent

parent

larvae hatching

of adults reared

1^1

Ancestry unknown

Ancestry unknown

18

9c? <?

^Ala

a a

9

1 ? and 2 c? c?

A2

'> tf

17

12c? c?

^3

20

5 ? ? and 10 c? c?

^4

yt }i

35

26<? <?

2^5

5

4? ?

AQ

48

32? ?

Al

') >i

29

8 ? ? and 17 c? c?

Bl

exAB

ex ^3

36

3 ? ? and 14 c? c?

B2

ex AS

ex A 2

53

34? ?

BS

ex Ad

ex A 2

64

48? ?

BA

ex A la

ex Ala

—

1^

An accident to the incubator prevented any further experiments with this series.

^ The same individual male was used in both these experiments.
2 In this experiment two females and two males were kept together.

Second Series.

Vumber of

Female

Male

Number of

Number and sex

family

parent

parent

larvae hatching

of adults reared

L 1 Ancestry unknown

Ancestry unknown

45

27 c? <?

L 2

> »»

f >

»>

58

45 c? c?

L 3

» »>

J,

38

31 ? ? and 1 c? (?)

L 4

) >>

>»

»»

16

9? ?

L 8

> >j

>)

>»

42

18 ? ? and 13 c? c?

L 9

> »>

>»

»>

9

2? ?

Lll

» »>

>»

'J

39

28 c? <? and 2 ? ?

L12

> »»

>»

»>

19

9 ? ? and 2 c? (?

LIS

» >>

,,

>>

46

38 c? c?

1L15

' >' .

,, .

,,

9

6? ?

1 L 15 a

> j»

)>

J,

50

24? ?

L16

» >>

>»

>>

66

64? ?

L17

, ,,

,,

,,

42

27 ? ? and 2 c? c?

Ivl8

> >»

»>

>>

25

8 c? <? and 2 ? ?

L19

» >>

»»

,,

39

26? ?

M 1

exL19

ex

L 18

33

2 c? c?

M 2

ex L 8

ex

L 8

.... 12 ...

4 ? ? and 1 c?

31 3

exL19

ex

L 8 ....

.... Nil

Sterile

M 4

ex L 15 a

ex

L13

. .. 45

36? ?
Sterile

M 5 '.

exL 8

ex

LIS ..

Nil

M 6

ex L 3

. . . . , ex

L 13

20

10 C? c?

^ M 8

exL 8

ex

L18

.... 12

3 ? ? and 4 c? c?

lif 8a

exL 8

ex

L18

.... Nil ...

iV/10

exL 8

ex

L13

.... 16 ...

3 ? ? and 5 <? c?

Mil

exL19

ex

L13

.... 27

5 ? ? and 4 <? c?

2? ?

1?

M12

exL19

ex

L13

.... 19 ..

M13

exl.19

ex

L18

.... 19 ...

The same male was crossed with two successive females.

EL HiNDLK 271

Secofui SerUs — (continnni).

Nvnbwol
tmmlXj

S % ....

Pamato
ex M i

Mule
liarvnt

ex M 6

NutiihKrof
UrvM hatch! n«

7H ...

89 ...

19 ...

56 ...

81 ...

88 ...

Numhrr *n«1 ••t
ciT aduiu roarvd

r>4 / ^

S 8 ....

.. eil/ 4

ex M 6 .

28^,^

S 4 ....

X 5 ....

.. •xMlO

ex Al 4

ox 3/ G .

ex M 6

6 ? ? And 6 ^ ^
42 ? ?

N 6 ....
iV 7

.. exAf 8

ex A/10

0x3/ 6 .

ex A/ 6 .

8? ?»nd 12^ 4
7 ? ? and 15 ^ -<

S 8 ....
N 9

.. ex.V 2

ex M A

ex 3/ 6

ex A/ 8

80 ...

48 ...

Nil

7 ? ? and 10 ^ /

32 ? ?

*V 10

ex M A ....

ox 3/ H

Sterile

S 13

ex M 4

ex 3/ 10 .

47 ...

37 ? ?

X14 ....

ex 3/ 4

ex 3/ 10 .

61 ...

87 ? ?

X 17 ....

ex A/ 4

0x3/11 .

36 ...

12? Jand 1 ^ (?)
Sterile

O 1

ex \ 5

ex .V 2 .

Nil ...

5 ...

37 ...

30

U) i ....

ex S 8

ex iV 2 .,

1 rj

0 4 ...
»0 6 ....

.. exN 6

exN13

ex.V 2

ex iV 3 .,

13 ? 9 and 3 ^ <f

«0 7 .. .

ex X 5

ex .V 2 ..

16 ...

3? ?

(> 8 ....

ex S 5

ex N 2 .

36 ...

29 ...

IH ...

24 ...

10 ...

11 <^ -r

«0 14 ....

ex N 7

ex X 2 ..

9 <^ <f and 1 ? » (?)

3? ?

0 16 ...

ex N 7

ex N 2 .,

0 17

exN 5

ex .V 2 ..

2 rf (f

> 0 18

ex .V13

ex N 2 .

3 ? ? and 1 <i

'0 19 ...

(0 6)
exN 7

(0 2)
exiV 2 ..

11 ...

1 ? and 4 (^ <f

P 2

(014)
, ex O 7

(0 7)
ex 0 19 ..

4 ...

1 ?

P 3

ex 0 7

ex 0 19 ..

15 ...

3 ? ? and 1 ^

P13

ex 0 18

ex 0 6 ..

8 ...

4 ? ?

' The male ex N2 of 02 was also crossed with female ex N 13 of 06 (Family () IH).
* The male ex A' 2 of O 7 was also crossed with female ex N 7 of O 14 (Family O 19).
^ This ? may have strayed from O 16, as a larva from O 16 was accidentally included
with O 14.

A glance at the above records will show that there are at least three
distinct types of families, male, female, or mixed. Therefore, four kinds
of crosses are }X)ssible, namely,

A. ? ex female family x d ex male family.

B. ? ,, ,, X (? ex mixed family.

C. ? ex mixed family x ,j ex male family.

D. ? ,, ,, X (^ ex mixed family.

These four types of crosses willbeconsidered .separat<^lv(see Tables I — IV").

272 Sex Inheritance in Pediculus humanus var. corporis

TABLE I.

Female from a female family x male from, a male family.

umber of

Number of

Number of

family Female parent

Male parent

females

males

M 4 ex L lott

X

exL13

36

—

i¥ 6 exL 3

X

exL 13

10

M 11 ex L 19

X

exL 13

5

4

M12 exL19

X

ex L 13

2

—

N 2. ex 31 4:

X

exM 6

54

I^ S exM 4:

X

exM 6

—

28

N 5 ex M 4

X

exM 6

42

—

0 4 exiV^ 5

X

exjY 2

10

3

0 6 ex 2^ 13

X

exN 3

—

14

0 7 exiV 5

X

exiV^ 2

3

0 8 exN 5

X

exN 2

—

11

on exN 5

X

exN 2

2

0 18 exN 13

X

exN 2

3

1

0 1 exN 5

X

exN 2

Sterile

A. All three kinds of families have been obtained. Excluding
from consideration those broods in which the number of adults raised
is less than four, it will be noticed that there are two female, five
male, and three mixed broods, but in the latter there is a decided
preponderance of females over males, viz. 18 to 8. Another note-
worthy result is that obtained in N2, NS, and iV 5, where crosses
between adults from the same families in two cases produced male
broods, and in the other, a female one. It is evident, therefore, that
there must be two kinds of males, or two kinds of females, or two kinds
each of both sexes.

TABLE IL

Female from, a feraale faw^ily x male from, a mixed family.

Number of
family

M 1
M 3
M\^

N 9
^10
i^l3
NU
Nil

Female parent

ex L 19
exL 19
ex L 19

ex M
ex M
exM
ex M
ex M

ex O
ex 0

Male parent

ex L 18
exL 8
exL 18

ex M 8
ex ilf 8
ex ilf 10
exMlO
ex Mil

ex 0 19
ex 0 19

Number of
females

Number of
males

Sterile

1
32

37
37
12

Sterile

1(?)

B. It is unfortunate that the one certainly mixed brood (P3)
contains but four individuals. The result, such as it is, resembles
that obtained in the preceding series, as regards the preponderance of
females over males in the mixed broods. With regard to the other

brcMicis. it will bt* iiottrtHt that then* is a wry «liH'i(l«H| pn'iMnulcniiKU' of
fiMimlos, tho ninj^lr iiialr family (-'/I) Iniii^ (!niu|n»sr<| ni only twu iii-
tliviiliials, which an* Unt few to be c<»nHi(irn'<l.

TAHLK 111.

Ffniaii^ /nym a mi,rrd family v tmil^/nim n uuilr f'umil

NutnSrrof Niimbrrof Nuntti«>riff

raiitlljr Female iNurrnt Male |i«rrnt fpiuAln) malM

li '2 ex A 3 X vx A 2 «4

li 3 ex A 3 X ex A 2 |H

M 5 ox /. 8 X ex /. 13 SUrih-

-V 10 ex /. H X ox /. 13 3 6

-V 4 ex M 10 X ox .V (> fi fi

X 6 ox M 8 X ox M fi 8 12

X 7 ox 3/10 X ox .1/ 6 7 I")

.V 8 ox M 2 X ox 3/ fi 7 10

(> 2 ex \ 8 X ex .V 2 — 1

()14 ex.V 7 X ex .V 2 1 «

(> 16 ox \ 7 X ox .V 2 3 —

O 19 ex JV 7 X ex \ 2 1 4

/' 13 ox O 18 X ox O 6 4 —

C. Excluding from discussion the two families contiiining less than
four individuals, all the families are mixed with the exception of three
female broods. There is, however, a distinct preponderance of males
over females in the mixed broods, the respective totals of the two sexes
being 61 male and 33 female.

TABLE IV.

Female from a mixed family x mnie froin a mi j:ed family.

Number of Nunil)€r of Number of

family Female parent Male parent females males

Bl ex A 3 X ex A S 3 14

M2 ex L 8 X ex L 8 4 1

MS ex L 8 X ex L IS 3 4

/? 4 ex .-1 la X ex .4 la — 1

D. The results of this series are too few for drawing any definite
conclusions, but it is suggestive that the three families are all " mixed "
broods.

Order in which the adults hatched.

In none of the experiments was there any suggestion that the fenialf
first laid eggs of one sex, and towanis the end of her lift', when exhaust<'d,
laid eggs of the other sux. With reganl to the mixed broods, the males

Journ. of Gen. vin ^'-^

274 Sex Inheritance in Pediculus humanus var. corporis

and females alternated fairly regularly in the order of their emergence
from the third stage larvae. Records of the order of hatching were kept
in each case and the following example is typical of the manner in which
the sexes alternated :

Family N 6. %^^ ?</</d'</ Wc/cT./ ¥ ? ? ?^</ ¥•

Sterility.

Although it was not very uncommon to meet with cases in which
pairs had no offspring, it is probable that most of them could be explained
by the assumption that one or other of the parents was unhealthy and
thereby prevented from performing any sexual functions. However,
one pair {N 10) seems to constitute an example of mutual sterility, for
unlike most of these cases, the female laid as many eggs as an ordinary
fertile individual, and none of them showed any signs of development.
When an unfertilized female is isolated it usually lays a few shrivelled
eggs, but the numbers are very much less than in the case of a fertilized
individual. It is possible, of course, that in the case mentioned, the
mere mechanical effect of the act of copulation may have stimulated
the process of Qgg laying, but there was such a well-marked difference
in the numbers and appearance of the eggs laid, that it seems more
reasonable to interpret it as an example of sterility.

The same male crossed with two different females.

Three experiments were made in order to determine whether the
same male produced the same type of family when crossed with different
females. Families A 1 and A\a constitute the first example, from which
it appears that when crossed with one female only male offspring were
produced, whilst when crossed with another, a mixed family was obtained.
Families 0 2 and 0 18 form another example, but unfortunately only
one adult, a male, was reared from the first cross, whilst the second cross
resulted in one male and three females. The results, therefore, of this
experiment are inconclusive. In a third experiment, families 0 7 and
019, the first cross resulted in three females, whilst the second one
produced four males and one female, but in the two latter cases the
results are complicated by the fact that the female had also been crossed
with another male, as mentioned below. Nevertheless, the result of the
first example is supported by the others, and strongly suggests the
existence of two types of females in the body louse.

K. HiNDLK 275

The name /etna le crossed witli two nut i as.

In tho auk» o( fiunily O (J ihv foinaI<» wjih k«'|>l. with u iii.il*' r«»r 1 .'i <lavh,
(luring which it hiid a cfrlain nuiiihrr ulfg^rH, fmin which a niah* IiphkI
of 14 uialo^ wiw ri'urrd. Th»» friiiah* was (Inn i^nlatrjl for 2 ilayH and
Rft4?rwani8 cn»«si'(l with a ilitVcmit. inah-, thr nsult h«iii^' .shnwii in
fimiily O IS. Tho first a<hjlt t*\' this hioud to hatch was a mah* and lh«ii
foHowtMl tho thriH' fiMiiah's.

In a second oxporinu'nt, U 14 and U ID, a teniah- pnMiuccd the hhuu'
tyjK' of family when crossed with two ditVcrent nwdes.

These results are c«)nipIioat4'd by the fact that a fenude may lay
fertile eg^ for many days after being isolated from any male, and this
should be remembered when considering the results of the first ex|M'ri-
ment. In this example there is a strong suggestion of the exisU'Uce of
two ty[K»s of males, for the first individual was only male-prrxlucing,
whilst the second prenluced females. With reganl to the single mah'
that was the first to hatch in 018, it is jwssible that it may have
grown from an v^g fertilized by the first male parent, but in any cjuse,
there is a tjualiUitivc dift'erence between the two broods.

Hemiaphroditisni.

In view of the possible existence of hybrids of P. hunKinua var.
corjwris with the variety capitis, the male families in the above series
of exjK'riments have been carefully examined to see if any individuals
showed any signs of hermaphroditism, a feature which is especially
characteristic of hybrid races, especially in the males. In family Xli
one specimen exhibited a slight anal process which Dr Keilin informs
me is an indication of this condition, but apart from this doubtful
exception, no other individuals have been found which are not normal
in this respect. The results of these experiments cannot be explained,
therefore, by the assumption that parents from ditierent races were
employed, for, as previously mentioned, the original parents were all
typical corporis, and in Series 1 and 2 came from the same })iece of
clothing, where they must have lived side by side.

19 2

276 Sex Inheritance in Pediculus humanus var. corporis

M4 M6 M4 Me Mio Me M4 Me

M4 Mio Mia Me M4 Mio M4 Me

Ni3

Chart showing the ancestry of families P3 and P13. The nature of the family of
the parent is indicated in each case, ? ? signifying a female, c? «? a male, and
<f ? a mixed family. The experimental number of the family is also given.

Of each pair, the female parent is on the left and the male on the right.

E. HiNDLK 277

Summary.

1. Tho n\Hults of th«»Ho ox|)«Tiiiu«nt..s hHow that in /*. Iiumatniff viir.
corporis thnv typeH of fuinilioH may occur:

(ci) FiMimle families,
(6) Male families,

(c) Mixinl families, in which there is an indicAtion of some nitio
between the numbers of the two sexes.

2. There is a strong suggestion of the existence of two kinds of
females, as well as two kinds of males.

3. The attached chart gives the pedigree of two families carrie<i
through four generations.

REFERENX'ES.

Bacot, a. (1917.) " A coutribution to the bionomics of Pediculus humanu* {vetti-
menti) and Pedirulu* rtipiti's.'' Parasitology, Vol. IX. pp. 228 — 258.

HiNDi.E, E. (1917.) " Notes on the Biology of Pedicidxu humanusy ParatUologyy
Vol. IX. pp. 259—265,

Keilin, D. and Nuttall, G. H. F. (1919.) "Hermaphrfxliti.sm and other abnor-
malities in Pediculus hunianus." Parasitology ^ Vol. xi.

Nuttall, G. H. F. (1917.) '' 'V\\e Biology oi Pediculus humanusy Parasitology^
Vol IX. pp. 80—185.

COLOUR INHERITANCE IN CATS. WITH SPECIAL
REFERENCE TO THE COLOURS BLACK. YEL
LOW AND TORTOISE-SHELL.

hY i\ r. LITTLE.

Research Associate. Carneijie SUition for K.rpei'imeutal Evolution
(CoM Spring HarlM)r, Now Y«)rk).

I. Introductory.

This paper has two <>hject«<: (1) the critical examination of exjKTi-
mentAl data on, and of current hyjK>theses concrrnin^ the inhi-ritaneo of
black, yellow, and t-ortoise-shell coat colours in cats; (2) the suggestion
of possible explanations for the occurrence of (a) unexi>ect<*d colour
classes in oniinary crosses between blacks, yellows, and t<^rtoise-shell8,
and of (6) both sterile and fertile tortoise-shell males which appt-ar
extremely rarely.

The fact that the work of all investigators of this subject ha.s left
the two points above mentioned not Siitisfactorily accounted for justifies
an attempt to explain the observed experimental results, even though at
this time no additional breeding data are offered fijr consideration.

IL The facts requiring explanation.

The critical and apparently contradictory facts which have been
brought out by breeding experiments with cats, and which must bo
satisfactorily accounted for and explained, are brieHy jis folli)Ws:

(1) In crosses between yellow imtles and black fe nudes, whon* the
expectiition on the b;isis of complete scx-linkagc is black males and
tortoise-.shell females, black females are sometimes produced. { I )oncaster,
1913.)

(2) In cj'osses between yelloin males and tortoise-shell /enuiles, whi*re
yellow males, black mah's, yollow f.niah's and tort«»is('-sh»'ll fomalos are
the only chv-sses expoctod on tho b;i.sis of coiiiphto sox-linkai^i'. black-
females are sometimes produced. ( I )oncastoi\ 1 IM )i. )

280 Colour Inheritance in Cats

(3) In crosses between two yellow cats, although only yellow young
are expected, two aberrant results have been noted. *

(a) A mating of this type has produced tortoise-shell females
besides yellows of both sexes. (Doncaster, 1913.)

(b) A mating of this type has produced tortoise-shell females
and black males besides yellows of both sexes. (Whiting, 1918.)

(4) There is no record of two black cats crossed together having
given yellow or tortoise-shell young.

(5) Tortoise-shell males are produced much more rarely than any
of the aberrant classes recorded under headings 1, 2, and 3 above. (Don-
caster, 1913; Wright, 1918.)

(6) Such tortoise-shell males are usually sterile. (Cutler and Don-
caster, 1915.)

(7) If they are not sterile they apparently do not give tortoise-shell
sons, but breed as yellows. (Doncaster, 1913.)

In considering these facts, investigators have usually tried to explain
all of them by a single hypothesis. (Doncaster, 1913; Whiting, 1918.)
This has proved to be difficult and unsatisfactory. (Ibsen, 1916 ; Wright
1918.)

It is believed that the experimental evidence favours the existence
of two genetically independent agents at work in the production of these
aberrances, for

(a) The appearance of the unexpected individuals noted under
headings 1, 2, and 3 above, is relatively frequent, and produces regular
results involving neither sterility nor the formation of new colour types.
{b) On the other hand, the occurrence of tortoise-shell males is
very infrequent, not regular, and is in a majority of cases intimately
connected with sterility.

Such being the case, an effort will be made to explain the appearance
of the unexpected individuals noted under headings 1, 2, and 3 by one
hypothesis and the occurrence of tortoise-shell males by a different one.

III. The relation betv^een yellov^ and black.

One of the first points to be established is the nature of the genetic
relation between yellow^ coat colour and black coat colour.

In this connection Ibsen, 1916, and Wright, 1918, believe black or
extension of black pigment to the coat, to be epistatic to yellow or the
restriction of black pigment from the coat. Doncaster, 1913, and

C. C. LiTTLK 1>81

Whiting, 1918, oonnidor lUv two roat rolourn allrlomorphic, thr h<*t4»n>-
/.ygwte IxMiig roininonly tortoiMo-Hlu'll.

The t*Tiiuno|o^y umsl hy th«Mn Ih juh folloWH :

/fejern, 1!U(J: Hhick li in (loininant to oninp* h which is iMirtu' in th«*
A*^ chnimtisoino. UiuhT onlinary r(»inlitiniis the factor lor nnm^«* h in
clcw«»ly linkiHi to T, a fact<>r fi»r torU>iw-Hhcll which acts only in the
pn*Si'nco o( Ii-hh\rk. The fiMualc is A' A', the luah' A'^ in fnrnjula.

Wng/U, \9\H : Black is <hie to the action o| a factor A, while t-4»rtois4'-
sholl is jmniuccMl by hctorozy^<isis of an " exU'nsion " fartor p]. TorU»iw-
shell foniales arc thus AV, yellow males e-, and black males A*-, in
formula. The factor K is b(>rne in the A' chromosome. The female is
A' A', the male A'^ in ft>rmula.

Doncaster, li)18 considei-s that yellow ami bhvck are allelomorphic,
and expri'sses yellow by V, and black by li. Where both an- present, a
YB or tortoise-shell animal is pHnluced. The female is A' A', and the
male A'^ in formula.

W/nthifj, 1018 also nmsiders yellow Y to be allelomorphic with black
y, and supposes y to be borne in the X chromosome. The female is
homozygous, the male heterozygous for A"^.

In 1912 I employed much the siime terminology {is that of I )onc<ister,
but in view of the production of blacks and tortoise-shells by two yellows
and the failure of blacks when crossed inter se to produce anything
except blacks, it is probable that the relationship between these two
colours may be more accurately expressed in somewhat the following
manner : B a factor for the production of black pigment which is found
in all X gametes. Y a factor for the restriction of black pigment from
the coat allelomorphic to y, a factor for the extension of black pigment
to the coat. One " dose " of Y is normally completely epistatic to one
"dose" of B, thus producing yellow individuals; but two "doses" of B
to one of Y produces a tortoise-shell. The factor Y and its allelomorph
y are also borne in the A"^ chromosome. Thus :

YBX

YBX

Yellow female

YliX

0

Yellow male

yliX

yiiX

Black female

yliX

d

Black male

YliX

yliX

Tortoise-Hhell female

This type of relationship will become clear us the crosses are taken up
in detail, and is further made use of in explaining the occurrence of
tortoise-shell males.

282 Colour Inheritance in Cats

IV. An attempt to explain the appearance of unexpected indi-
viduals OF NORMAL COLOUR TYPES. (Headings 1, 2, and 3,
Section ii, above.)

It is tacitly assumed by all investigators that at some time or times
in the past, there must have been a genetic change, ridding certain
gametes of the epistatic colour factor, whether it be the Y of Whiting,
the E of Wright, or the T of Ibsen. Had this not been the case neither
the hypostatic form nor the tortoise-shell heterozygote could have
appeared.

We may, then, for the sake of argument accept the set of symbols
given above, and assume that tbe change from Y lo y must have occurred.
There is no experimental evidence to show how recently or how fre-
quently this change may have taken place, but if we assume that it is
still taking place in a portion of the gametes of certain individuals —
which seems entirely probable — all the results obtained under headings
1, 2, and 3, may be accounted for. Such a change from an epistatic to
a hypostatic condition would be directly comparable to the appearance
of the recessive pink-eyed mutation in a stock of dilute brown mice
recorded by the writer in 1916.

Animals in whose gametes this mutative process was occurring de
novo would show no trace of it in their own somatic characteristics, but
would, upon breeding, give results in agreement with the actual aberrant
classes obtained.

We should thus expect that an occasional yellow female would
form gametes yBX in addition to those containing YBX which she
normally produces. Similarly, certain yellow males would be found which
showed by their progeny that they were forming among their X gametes
some which were of the constitution yBX instead of the normal YBX
type.

Yellow males of this unusual kind would, when crossed with black
females, give among their progeny a certain number of black females, in
number depending upon the frequency with which the unusual yBX
sperm was formed. This fact would explain the aberrances listed above
under Section ii. Heading 1.

Similarly, such unusual yellow males would, when mated to normal
tortoise-shell females, give rise to a certain number of black females in
addition to the other classes normally expected. This would cover
category two of exceptions mentioned above (Section ii).

Finally, a yellow forming yBX gametes, when crossed with a normal
yellow or with one of its own type, would give rise to unexpected black

C. C. LiTTI.K 283

or tortoist^-shell young, th«* pn»|K>rtion (lojK»iHling ii|)<)n whothcT the ydlow
nmlo or the foinalo t>r both wen* concemc<| in the fonnation of th<* yliX
ganieti'H.

Thu» if the male was alone concerned, tortoim'-Hhell fenmleH. Iml lut
Hack males wouhl Ik* likely to apiKMir among the progmy. This appran*
to be the caiH» in the mating reeonled by Doncjuster ( HH.S) in wliirh two
yellows gave among their progeny three blue tenmlen with a cn*am
ooloureii patch (tortoise-shellH). If, (Ui the other hand, the fenmle {larent
was the unusual mut^itive individual, black males would f)ccur in aildition
to tortoise-shell females and yellows of both si'xes. This contlition was
realised in the c^use of female dilute yellow #23 (formerly owned by m<*)
whosi* breeding reconi is rejK)rted by Whiting, 1918. An explanation
of this sort would account for the aberrances note<l under Section II,
Heading 3, above.

From the number of tortoise-shell and black young obttiined in the
twH> cases referred U^, and from the numerical relation of the black
females under headings 1 and 2 (Section II) to the expected colour chusses
(Doncaster, 1913), it seems probable that yellow animals forming yBX
gametes do so in approximately 50 °/^ of the gametes they form, as
would a normal heterozygote.

In addition to yellow animals, certain tortoise-shell females might
theoretically be expected to show the same phenomenon. Such animals
would fonn an excess of, or jx)ssibly exclusively, yBX gametes, and,
in so far as they did so, would breed as blacks. Such an occurrence
would, however, give rise to no unexpected chisses of young in crosses,
but might result in the absence of some of those normally expected from
certain matings. Quite naturally this fact might, in a small number of
progeny, escape notice.

There is no evidence to show that the appearance of any of the classes
above referred to is in any way connected with a break in se.r-linkage or
with the occurrence of tortoise-shell males, and lue may therefore, until
such evidence is presented, fairly consider them as independently pro-
duced.

V. Criticism of existing hvpotheses to explain the appearance

OF UNEXPECTED INDIVIDUAI^S OF NORMAL COLOUR TYPES. (Head-
ings 1, 2, and 3, Section II, above.)

Attempts to explain the apiK*arance of the aberrant colour classes
referred to, have involved either {a) the breaking i»f .sex-linkage with
" crossing over " in the male, or (6) the (►ccurrence of a series of nKKlifying

284 Colour Inheritance in Cats

factors determining the relative degree of black and yellow pigmentation.
They may be separately considered as follows :

(a) Doncaster's hypothesis of a break in sex-linkage : this hypothesis,
which in a modified form is a basis for Ibsen's later explanation of the
appearance of unusual colour types, involves, if it is to explain the ex-
ceptional black females, the existence of "crossing over" in the male
between the X and the 6 chromosomes. Such crossing over has not, in
so far as I am aware, been demonstrated in any forms in which the male
is XO in formula as in cats. It further would suppose that, as tortoise-
shell males were formed by the same process, they would be expected to
occur with as great frequency as the exceptional black females. It
further leaves entirely unexplained the appearance of blacks or tortoise-
shells from a cross between yellow animals. These objections seem to
be of sufficient weight to throw the chances against Doncaster's or Ibsen's
hypotheses.

(6) Whiting's hypothesis of modifying factors which at one end of
the series would serve to make tortoise-shell animals yellow, and at the
other end of the series make them black, remains as a possibility though
seriously invalidated by certain points as follows :

(1) There should be records of black females (geno typically
tortoise-shell) which if crossed with other blacks should give yellow
males and tortoise-shell females, or if crossed with yellow males should
give unexpected yellow females. Neither of these results has been
recorded.

(2) Doncaster, 1913, reports that the three tortoise-shell females
produced from a single cross between two yellows were " blue with a
cream patch " thus showing that they were near the black end of Whiting's
modifier series. Inasmuch as under his hypothesis one of their parents
must have been at the opposite or yellow end of the series, it is difficult
to explain ho w and why many of its progeny should show the condition
characterising nearly the other end of a graded series.

(B) The occurrence of these young in a single mating makes it
seem likely that the particular animal was forming ordinary yBX gametes
in a considerable number.

(4) The tortoise-shell young produced by dilute yellow female
^ 23 already referred to, before she was sent to Dr Whiting, were normal
tortoise-shell in colour ; if anything, more nearly on the black end of the
graded series, than on the yellow. This case serves to support that

C. C. LiTTLK 285

reported by DonaisUT, ami tomiH to hHow that the yellow animal tnius-
mitt^i to its pifM/e/iy no peculiar set of mfHiijiers.

VI. An ATTF.MIT to explain the CXrUKHKNrK OK (a)STKHILK, ANI>
{h) KEHTILE TORTOISK-SIIKLL MAI.P>i.

(rt) The prtHiuction of sterile tortvise-shell males.

It is agnHMl by all th(>s(» who have n'|K)rte(l on brccciin^ oxpi^riinontsS
with cnt.s that the female apjK'ars to he homozygous, tin- male hetero-
zyginis, for sex. The former is therefore -Y-Y, the latter XO in formula.
This places aits in the siime category with Drosophila, and this in turn
means that one may rightfully turn, and in fact should turn, to the mag-
nificent work of Morgjvn and his associates in any attempt at explaining
a peculiar result which shows exceptional conditions of sex-linkage.

If one considers the phenomena of non-disjunction of the X chromo-
some in Drosophila, reported by Bridges in 1918, and later (IDUI a and
6) further established by him after an extensive series of breeding ex-
periments, one cannot fail to be impressed by the similarity between the
results of that process in Drosophila, and the observed experimental
facts in cats.

For example, non-disjunction is neither frequent in its occurrence
nor is it clear enough in its hereditary behaviour to give striking
numerical results in as slow breeding an animal as a cat, unless it were
watched for deliberately. In Drosophila it gives rise to two ver}- sig-
nificant exceptions to the normal sex-linked inheritance. First, it
produces animals apparently males, which are sterile, and second, mosaic
form^ are apt to arise in non-disjunctional stocks. If one considers that
the majority of tortoise-shell cats luhich are apparently males are stetnle,
and second that they are also a mosaic form in a sex where commonly
none is found, the comparison becomes interesting and extremely sug-
gestive.

We may now consider what the probable results of non-disjunction
would be, did this phenomenon exist in cats.

The chanicteristic of primary non-disjunction is that in oogenesis the
two A'' chromosomes go together into a single egg, leaving another egg
without even the normal single X. This may be shown Jis follows :

Non-disjunctional female XX
forms gametes A'A' and -

286 Colour Inheritance in Cats

If now the eggs of such a female are fertilized by sperm of a normal
male we have four possible types of zygotes.

Eggs Sperm Zygote

XX X XXX Dies

X X- " Near male " always sterile

XX 6 XXd Female with peculiar gametic condition

d d- Dies

Bridges has demonstrated that the XXX and 6- forms die, a7id that the
X- form although appearing like a male is always sterile. If now we
imagine a cross to be made between a tortoise-shell female cat showing
non-disjunction and a normal yellow male, we should have the following
condition :

Non-disjunctional
Tortoise-shell female YBXijBX Normal Yellow male YBXe

Forming gametes YBXyBX and — Forming gametes YBX and 6

Zygotes (a) YBXyBXYBX Dies

(6) YBXyBXd Tortoise-shell with peculiar gametic conditions

(c) YBX- "Near male" always sterile

[d] - d Dies

If now one assumes that absence of the 6 chromosome allows the " near
male " class (c) to develop into a tortoise-shell, disturbing the normal
relation of yellow to black to produce a somatic mosaic, we could account
for the appearance at rare intervals of tortoise-shell " near males " which
were not fertile. It seems not unlikely that the absence of the 6
chromosome might well upset the somatic relationships of certain of
the characters whose factors are carried by the X chromosome. This
would account for the appearance of a tortoise-shell "male" from a
mating of yellow male x tortoise-shell female. (Doncaster, 1913.)

Another mating which, according to Doncaster, has produced a
tortoise-shell male is that of yellow male by black female. Here, if the
black female showed non-disjunction, the following condition would be
found :

Black non-disjunctional female yBXyBX Normal yellow male YBXd

Forming gametes yBXyBX and - " Forming gametes YBX and d

Zygotes (a) yBX yBX YBX Dies

(ft) yBX yBX 6 Black female with peculiar gametic conditions

(c) — YBX Tortoise-shell? "near male" always sterile (as in

previous mating)
{d) — d Dies

The third type of mating reported by Doncaster as having. produced
a tortoise-shell male is that of black male with tortoise-shell female.
Here everyone is in difficulty. If, as Doncaster suggests, the occasional
crossing over of F, the factor for yellow, to a ^ gamete is responsible for

C. C. I.ITTLK 287

the pniduction of ii t4)rt4)iHi»-Hhell iiml(\ iinthing that omld hti|)|icn in
either the gnnu*U>8 of i\\v bhick timlr or of t hi* t/ort/oiw-Hhrll rrnmlc wotiM
pnxluoo )i Uirtoiw-shrll iiwilr. On Whiliii^'s hy|H»lh«*HiM wr Mhniild hiivr
to8UppoH(> that thi* tottoiso-Hhrll tciiiair, altJimi^h slw htrHilf .shnwiMl no
innrkiHi luotiiHorH (t»r »\\v would have bmi hiack), tranMinilM uiiUMually
hwivy iiHHiiliors to hiT «ous. Thosr ^jinu-U's wcuiM in turn havr to Im-
mot by an iMjually hoavysi'toi* uKMlificr.s from tin- black inaN*, or a yrllow
would result.

Further than this, by Whiting's hy|)oth('His thf yellow male is YXt) in
constitution, and this makes th«' soiin-e of the black that he must pnMliiee
fKunatieally under the mfluenee of mcMlifiji-s in onler to ln'come a tortoise-
shell, uncertain. This condition is, of coui-se, not im|)o8sible but is highly
impmlmble. Finally, the phenomenon of non-disjiuiction meets with
distinct dirticulties. Unless the black male forms gametes with neither
the A' nor 6 chromosomes present it would be hard to sec how the
tort-oise-shell male could be prmluced by this mating. Fijrmation of
sperm without A' or 6 would not be likely. Yet the {wssibility exists
and may therefore be considered. What seems to me altogether more
likely is that the breeder's records on which Doncjister bjised his obser-
vation were in this aise uncertain or incorrect, a circumstance quite
possible in cats even with the best possible intentions.

(6) The production of fertile tortoise-shell viales.

We have seen that peculiar tortoise-shell females of formula

YBXyBXB may possibly be produced by primary non-disjunction. If

now one of these females is crossed with a black or a yellow male peculiar

yellow males of the constitution ¥3X66 would be formed iis follows:

Non -disjunctional
Tortoise-shell female crossed with Yellow male

YliXyliXd YBXe

Forming gametes YBXyliX, YliXO, Forming gametes YliX and 0
YliX, 0, yliXd and yliX

Zygotes (a) YliXyBXYBX Dies

(6) YBXdYBX Peculiar yellow female

(c) YBXYBX Yellow female

(d) 6 YBX Yellow male

(«) YBXyBXd Peculiar tortoise-shell female

(/) YBX ee Peculiar yellow male

(g) YBX d Yellow male

(h) ee Dies

(i) yliXeYBX Peculiar tortoise-shell female

(j) yBXee Peculiar black male

(Jk) yBXYBX Tortoise-shell fumale

{I) yBX e Black male

288 Colour Inheritance in Cats

If now such a peculiar yellow, YBX96, is mated with any female showing
primary non-disjunction — an animal which might well prove to be a
fertile tortoise-shell male would be produced. Thus :

Non-disjunctional

Non-disjunctional

Black female

crossed with Yellow male

yBXyBX

YBxee

Forming gametes yBXyBX and -

Forming gametes YBXd, YBX, 0 and 69

(a)

yBXyBX

YBXd Dies

(&)

yBXyBX

YBX Dies

(0)

yBXyBX

d Peculiar black female

(d)

yBXyBX

ee Peculiar black female?

{e)

- YBXd

' ' Tortoise-shell male ''fertile ?

(/)

- YBX

" Tortoise-shell ' near male ' sterile" ?

(9)

- 6

Dies

{h)

- ee

Dies

Here the assumption is made that an animal formed from the combination
of gametes, YBX 6 and -, may he somatically a tortoise-shell, and that the
6 chromosome which is brought into the zygote by an X-bearing gamete
does not in all cases exert its full influence until gametogenesis. The
FBX^-male would then be supposed to develop somatically just as does
the FSX-animal, but upon gametogenesis the 6 chromosome of the YBX6-
male is able to prevent the sterility which exists in its absence. This
seems quite possible, for it appears that in Drosophila the 6 chromosome
is not needed for the development of the normal male somatic characters,
but that it is necessary, however, for successful gametogenesis in the
male.

A fertile tortoise-shell male would, when he formed gametes, behave
exactly like a normal yellow male. That is to say, although he was
himself the product of a combination of Xd and - gametes, he would in
gametogenesis form only X and 6 gametes, just as would a normal male.
This has been the breeding behaviour of the one recorded certainly fertile
tortoise-shell male (see Doncaster, 1913) which acted in crosses with
tortoise-shell females apparently exactly as a yellow male would have
done.

It wdll be seen that the above hypothesis, although somewhat com-
plicated, is nevertheless in accordance with experimental facts and
accounts for sterile and fertile types of tortoise-shell males ; it explains
their infrequency of appearance, and possibly their failure to transmit
their own colour pattern to their descendants ; it is supported by the luork
of Bridges with Drosophila — the most completely investigated form showing
a similar type of sex-linkage ; it is further capable of experimental tests.

i\(\ LiTTLK 2H«

VII. Criticisms ok kxistino iiYiM»TiiF><>>i to kxim.ain tiik

iK^crUHKNCE OK TOKTolSK-SIIKM. MAI.KS.

(1 ) l\)!ic4uttor'.s hy|x»thfHi.s, jus jiln-july |M>intr(| nut, nM|uinM "cniswiii^
ovor" ill tluMualr lK*twi*i'n A' jiikI ^ chnnnn.mnncs jic<»M«litinn imi .shuwii
to oxiat in giiinot<)j;i»m\siH of any A'^ nmlr rnriii. It. fMitlnr fails to
acctMint for {a) tlu^ comparativr infrtMjucncy tti' tintoi.sc-slnll nuilis an
COlM[>jinHl with alh'rrant Mack fnnalrs, {b) tlu' strrjljty of tin- majority
of torUnse-shi'll inah's, ami (r) their peculiar hchavi(Mir iti hnMciin^r

(2) Ibsen's hy|K)thesis dot's away with the nee<l of crossing over in
the nuUo, hut fails, as does Doncjvster's hypotliesis, to meet or explain
point* (a), (/>), or (c) statA'd above.

(3) Whitinj^'s hypothesis of nioditiers would not be aljle to ^'ive a
tortoise-shell male which acconlin^^ to his formula wouM be A' }'^ with-
out iwldinj; a fiwtor for black t^) the formula ^nven by him to be carried
in the A' ^imete. It further would supjH»se that by selection (which
iindoubttMJly ha,s t>ccurred) it would be possible t(» transmit the necess;iry
motiitiers t<.) a considerable number of his male progeny, thus fornnng
torU)ise-shell males — and this, though great efforts have been made, hjus
provLHl imjx)ssible. Whiting's hypothesis, like those of Doncaster and
of Ibsen, takes no account of the sterility of the majority of tortois<'-shell
males.

(4) Wright's hypothesis is that t(jrtoise-shell males are really A' A'
individuals in which the abnormality lies not in the cohmr but in the
sex. He likens them to certain sex intergrades alre;uly described in
some forms by other investigators. This hypothesis meets trouble when
a fertile tortoise-shell male is encountered. It also is contrary to th<'
evidence obtained by Bridges who shows that in Drosophila A' A' fonus
are females, even though they contain other abnormalities of chromosome
distribution.

VIII. Summary and conclusions.

(1 ) The genetic constitution of the normal colour varieties of cats ;is
regards yellow and black pigmentation appears to be as follows: Ji = a
factor producing black pigmentation, }' = a factor which restricts black
from the coat, y = a fiictor allelomorphic to Y^ and hy[)ostatic to it,
allowing black pigment to extend to the coat.

17/A' J7;A' Yellow female

17^V e Yellow male

yliX ylLX lilack fern ah-

yHX 0 Black male

YliX yftX Tortoise-slj.ll fem.iU-

Joum. of Gen. vui 20

290 Colour Inheritance in Cats

(2) The unexpected but normally pigmented individuals appearing
in certain matings (Headings 1, 2, and 3, Section ii) can be accounted
for by supposing that F becomes 3/ in a certain proportion of the gametes
of exceptional individuals.

(3) Sterile tortoise-shell males may possibly be " near males " formed
as a result of non-disjunction of the X chromosome and therefore YBX—
in constitution.

(4) Fertile tortoise-shell males may also be the product of non-
disjunction (secondary) and would be zygotes formed from the fusion of
gametes YBX6 and -. These males in gametogenesis would behave as
ordinary yellows.

LITERATURE CITED.

Bridges, C. B. 1913. " Non-Disjunction of the Sex Chromosomes of Drosophila."
Journ. Exp. Zool. Vol. xv. p. 587.

1916. (a) " Non-Disjunction as Proof of the Chromosome Theory of Here-
dity." Genetics, Vol. i. p. 1.

1916. {h) " Non-Disjunction as Proof of the Chromosome Theory of Here-
dity." Genetics, Vol. i. p. 107.

Cutler, W. D., and Doncaster, L. 1915. " On the Sterility of the Tortoise-Shell

Tom Cat." Journ. of Genetics, Vol. v. p. 65.
Doncaster, L. 1904. " On the Inheritance of Tortoise-shell dnd Related Colours

in Cats." Proc. Camb. Phil. Soc. Vol. xiii. p. 35.

1912. "Sex- Limited Inheritance in Cats." Sci N. S. Vol. xxxvi. p. 144.

1913. " On Sex-Limited Inheritance in Cats, and its Bearing on the Sex-
Limited Transmission of Certain Human Abnormalities." Journ. of Genetics^
Vol. ni. p. 11.

1914, " A Possible Connexion between Abnormal Sex- Limited Transmission

and Sterility." Proc. Camb. Phil. Soc. Vol. xvii. p. 307.

Hyde, R. R. 1916. "Mosaics in Drosophila Ampelophila.'' Genetics, Vol. i.

p. 581.
Ibsen, H. L. 1916. " Tricolor Inheritance. III. Tortoise-shell Cats."' Genetics,

Vol. I. p. 377.
Little, C. C. 1912. " A PreUminary Note on the Occurrence of a Sex- Limited

Character in Cats." Sd. N. S. Vol. xxxv. p. 784.

1916. "The Occurrence of Three Recognized Color Mutations in Mice."

Amer. Nat. Vol. l. p. 335.

Whiting, P. W. 1915. " The Tortoise-Shell Cat." Amer. Nat. Vol. xlix. p. 518.
1918. " Inheritance of Coat Colour in Cats." Journ. Exp. Zool. Vol. xxv.

p. 539.
Wright, S. 1918. "Colour Inheritance in Mammals. X. The Cat." Journ.

Heredity, Vol. ix. p. 139.

THE PROBABLE ERllOKS OF CALCULATKD LINK-
AGE VALUES. AND THE MOST ACOl'KATK
METHOD OF DETEUMININC; (JAMETIC FItoM
CERTAIN ZYOOTK^ SERIES.

Bv J. B. S. HALDANK, M.A..

Fellow of New College, Oxford.

In view of tho controversy lus to the cause of coupling and repulsion
it is desinible to know the probable err(»r of any j^iven deh'rniination
of a gametic ratio, and also t<i obtain from an observed zygotic serit^s
the most accurate possible estimate of the corresponding gametic series.
By the probable ern)r of an observation is meant of course a number
such that the difference between the true and observed values is ecpially
likely to exceed it, or to fall short of it.

We shall consider the case of a heterozygote AaBb which pnxluces

gametes in the proportions ^AB: — ~- Ab: ^ aB :^ <ib. In the

£t It JL Z

case of repulsion p is the cross-over value, in that of coupling 1 - p. If

we write the gametic series in the case of coupling as

xABAAh'.\aB:xab,
in the case of repulsion as

\AB : yAh : yaB : lab,

then X = -^ — , V = ^ .

The values of />, x, and y may be obtained directly from the cross
AaBb X aahb. Here if n is the total number of zygotes obtained, P the
observed value of p, and (AB), (ab) the observed numbers of zygotes of

compositions ABab and abab respectively, 7^= - . If n were

ft

infinite P and p would of course be equal, actually the probable error

of Pis

P{l-P)

v/'";" <■'■

•6745

V n

•JO -2

292 Probable Errors of Calmlated Linkage Values

A direct proof of this result from Bayes' theorem is given by
Todhunter(l) ; the proof generally given is for the probable error when
p, the true value, is known beforehand.

More accurately the probable error in excess is

AP(I-P) , -3033 (1-2P)

■6745^--A±__^^+— v- -^ (2),

that in defect

•674i

/P(l-P)_g033(l-2P)

\ n n ^ ^

This correction however is never important, and may be neglected
for all ordinary purposes if both nP and n (1 — P) are sufficiently large
. (say over 100).

Let x = X + ^, p = P + TT.

provided ir is small compared with P,

(1-P)

= oc + {X + ly IT.

Hence the probable error of X is

or

•6745(X + 1)^^ (4).

Similarly that of F is

•6745(F + l)yj (5).

It may be remarked that when p, x, or y are determined by this
method we automatically eliminate the effects of differential mortality
due to one factor only, or to both if they affect the mortality indepen-
dently. If however both factors affect the viability it will generally be
safer to employ Morgan and Bridges'(2) "balanced inviability" method.

To take a concrete example of the above calculation, Altenburg(3),
working with the factors M and S (Magenta and Green stigma) in
Primula sinensis, obtained from the cross MS . ms x mmss, from a count
of 3684 plants, a value of '884 for P. The probable error of this value

/•884X-116
3684

is therefore -6745 a/ ....... , or 00358.

J. H. K. ilALPANK 293

Henct* tho on»M8-<>v«T vuluo in Wili'M] /. A' - 7r>f» and i\H

....... . "•* ■-*^- 'l'l>'i^ rnith«r 7 imr S

AUHV

an* iinpritlvihlf vuIih's for .r, thr prohaMlit y «»t an « rmi us ^nat us nr

gn*aU»r than 511 lK»in^ only about 27 / , us may !»<• s».ii fVom a tal.I.- oi

values of "'■■ I f'^df.

\V«» shall now considn- the zyL^otic scrirs ol)tain«<l in F. \t\ mating'
iuter se thv h\ zyijotrs i-onsid^nd alK»v«'. Tln' rx|M'('t«M| |no|>..i tions an-

- + P' t n ^ ~ P' 1 / ' ~ P' If P' I
/ An : / Ah: / (i/i :'. ah,

4 4+4

when' AH.viv. rrpn'st'nt t.ln' visible characters of the zygotes, ('misider
a gn)Uj» of » such zyi;<)ti's. Put t = j)"^, and h't /„ Iw an aj)j)ro\iiiiaie
value of t, while T=f„ + S is the value which is most probable from the
given oKservation. Also let ^,, f.,, t^, f^ be the four values of / (al(ulat«'<l
from the four classes observed.

Then ti= - 2,

^=l-

4(^16)

f.= I

^{(tli)

t,=

4(a6)

.'. t, + t, = L + t,.
Hence for any value of 8 the probability of the observed proportions

2 + f , 1 - ^ 1 - f, U ;

— T— AB : — T— Ab : —7 aJi : , (ib

4 4 4 4

being obtained in a group of « zygotes is proportional to

Exp.

j-'1 f"!"'-M'"''

2 + ^„ + S

4

ia + S

4

or

" [(to - ^ + ^)- (^a - t; + 3 )■- 4 (/„ - f. 4 3 )-' ( fa - /. + 3 )•

1 - ta fa I

since h may be neglected in comi)arison with /„ or 1 -/„. th«»ugh not
with ta — ti, etc.

294 Prohable Errors of Calculated Linkage Values

When T is the most probable value this probability is a maximum,
and hence the expression in brackets is a minimum. The condition for
this is

(l-«^a+2(2-f^a)^a + (2+g(l-W

Hence, putting ^2 + ^3 = ^i + ^4,

^-^« + ^- 2VUg ^^)'

while P=^/T.

As our value of ta we may take t^ in the case of repulsion, ^ (^1 + ^4)
in the case of coupling. If greater accuracy is desired the value of T
thus obtained should be substituted for ta in equation (6), and a more
accurate value thus obtained. This proceeding is however rarely worth
while.

The same value for Tis reached more directly by Bridges' method (5),
where T is calculated from the coefficient of association

( AB) x(ah)-(Ab)x(aB) _ 4>T-1 , .
(AB)x (ah) + {Ah) X {aB) ~ 21^+1 '

as may readily be seen on substituting the value ^ (2 + fj) for {AB), and

so on. It is doubtful however if this method is any shorter than that
given above, unless a four figure table of values of P in terms of the
coefficient of association has been calculated in advance.

To take a concrete example, Punnett(4) working with the coupled
factors B and L in sweet peas obtained the F^ zygotic series

48315i:, 3905^, 3936X, 13386^.

Here n = 6952,

4xl338_
^'~ 6952 ~ ^^^^'
^^ = i(^^ + ^,) = -77475.

^ 3 X -77475 X -7796 + 2'77475 x -7699 _ ^^.^
•'• ^ == 2 + 4 X -77475 -'^^^^'

If we substitute "7743 for ta in equation (6) we obtain no change
in the first four decimal places.

J. H. K. IIaldank 205

Henoo /* « ^T - '8800, and the cniHs-ovcr valur in 12 00 7^,

and the ca)culate<] cx|x»ct4\tioii is

4821'7^A., 392-:}/?/. f\^2:\hL, VM-ylU,
against obeen'iKl niitnb<>rs

4881 BL, 390 m, :\m\ bL, VXIH hi.

We have now t-<i calculate the probable errors of the values ol
T, Py X, and Y^ obtained above.

The probability of any value T -^ a of t varies as

^ r.a

But the coefficient of a in the exponent vanishes, and a may b
neglected in compjirison with T or 1 — T unless n is small, hence th^
probability varies as

Hence the probable error of T is

/iT(2+T){l-T)

That of P is x-p of this, or

^^V (r+2P=)» <^>-

The probable error of X is

or, when X is large, approximately

•6745(X+1)a/^ "^-^ (JO.

That of r is

■-('•^../"■■^m?,:/-^'

or, when 1' is large, approximately

(i745<''+'»'^ ,n..

296 Prohahle Errors of Calculated Lhikage Values

Comparing these values with the probable errors, gi ven by formulae (1),
(2) and (3), of the values obtained by crossing F^ with the double re-
cessive we see that the latter are always smaller. When P is nearly 1
the accuracies are nearly equal, but when P is small the direct method

is — -p times as accurate as the indirect, i.e. ^ times as many zygotes

obtained by the indirect method must be counted, in order that it
should give a result as accurate as the direct method. When P = J,
the ratio of the probable errors is only 3 : 2.

Hence the ratios calculated from Pg are nearly as reliable as those
obtained from Pi x the double recessive in the case of coupling or weak
repulsion, but with strong repulsion they are somewhat unreliable.
Hence Bridges' stricture(5) on the unreliability of values derived from
Pa results is only justified in the case of strong repulsion.

Moreover, since its numbers vary as the square of the cross-over
value, the double recessive class in F^ is a more sensitive indicator of
repulsion than any class derived from the cross Pj x double recessive.
Thus the enumeration of Pg is somewhat more sensitive as a test for
linkage, and about equally accurate as a measure of its degree in the
case of coupling, though not of repulsion.

Differential mortality is eliminated by the above method under the
same conditions as by the direct method. This may be seen at once
from the fact that the coefficient of association is unaltered when {aB)
and {ah) are diminished in the same ratio.

It may be remarked that where there is incomplete coupling or
repulsion in unequal degrees (in other words finite but unequal cross-
over values) in both sexes, the above method of evaluating t gives t =pqy
where p and q are the cross-over values in the two sexes.

To return to Punnett's sweet pea experiment quoted above : from
formula (7) the probable error of P is

/2^
V ¥i

,„„ , _7743 X -2257 .__,

•"^^-/ 2-5486 X 6952 '"'••""^«"-

Hence the cross-over value is 12*00 + '28 7o-

From formula (9) the approximate probable error of X is

/7'3^
Hence a; = 7-333 + 197.

6745 X 8-33 a / ' %l^^^^ ^ or -197.

.1. H. 8. Haldank 297

A U\h]o of vnliioft of , I e~'*(it hHowm that 7 \h <|uit«' ii pmhalilr

valut* of jr, whilst tht» rhanro^ ii^iiinst an rrrnr a-s lar^'«- as ihaf iii\o|vi<l
by x«8 an* alMuit 50 Ui I.

It wotiKl \ic UKist (i(>sinihK> to cxaiiiiiic all rxtniil )iiikaL(<- data nii
th«»!U» lines. aii<l ih't^'nninr whrthrr int-4'^nil valiirs of .r ami y. if |H»H.Hil»l«'
of the fonn 2"— 1, lay within t.h«' zonr of |>rol»ahh' error in alxMit half
the wuHes. Wen* this found t<» he the ease it wouM he a stron^^ ar^ni
inent in favour of the reduplieation theory, if not it WiiuM tend to
disprove that the«>ry, at least in its present, form.

My thanks aix' due to Professor Kdgeworth, F.IIA , for valiiahl*-
adviee and criticism.

Summary.

1. F'ormulae are given for the proba])le errors of linkage and re-
duplication values calculated from the offspring of crosses of tvjMs
A Ii . ab X abab, Ab .oB y. ahab, A B . ah x A B . ah, and A h .aBx A l> . n H

2. A meth(Kl is given by which the best possi])le linkage values
may be calculated from the of^s])ring of the latter two crosses, i.e. from /',.

3. If this meth<Ml is em])lo3'ed, F^ is almost as accurate a means «»f
measuring linkage as arc the (►ffspring from F^ x douhle recessive; it is
also slightly more sensitive ;is a means for the detection of linkage.

REFERENX^ES.

1. ToDHLXTER. Hutonf of the Theory of ProhdhiUt 11^ p. r).')^.

2. MoiuiAN ami Bridcjfis. " Sex-linked inheritance in Drosopliil.i.' Curnrifo' /„sti

tution of Wcuthington^ 1910,

3. Altexburg. Oenetic*, Vol. i. p. 3r)4, 1910.

4. PuxNKTT. Journal of Genetics, Vol. vi. No. 3, 1917.

5. liRllXiE-s. American Xaturalixt, Vol. XLvni. p. 521, 1914.

THE COMBINATION OF LINKAOK VALUES, AND
THE CALCULATION OF DISTANCES BETWEEN
THE LOCI OF LINKED FACTORS.

By J. B. S. HALDANE, M.A.,
Fellow of New College, Oxford.

(With One Text-figure.)

On the theory that the degree of linkage between two factors
depends on the distance apart of their loci in a chromosome, Morgan
and his fellow- workers have taken the distance between two loci .is
proportional to the cross-over value ^ of the factors located in them.
This theory gives consistent results when the cross-over values are small,
but, as recognised by Sturtevant, and by Morgan and Bridg('s(l), is not
accurate for larger values. On the reduplication theory Trow(2) h;us given
a formula for the combination of linkage values which is shown below
to be inaccurate when the linkage is not close. In the present fKiper
a more ;iccurate theory of the relations i7iter se of the cross-over values,
and of their connexion with the distances apart of the loci of factors in
a chromosome, is developed. Some such theory is especially necessary
when dealing with a group of factors conUiining few membt'i*s not very
closely linked.

Suppose A, B, C to be three factors whose loci lie in that order in
the same chromosome. Let m be the cross-over value for .1 and B,
n that for A and C. If the chromosomes were perfectly flexible, so
that the fact of their having crossed between A and B did not diminish
the probiibility of their crossing again between B and (\ wr sh«uild
expect a triply heterozygous organism to j)rodure gamet«*s in th»- f«»l-

> If zvgot€h of composition AH . ah and Ah . ali ^ive Kametic sfri«-s

{\-m)Ali.mAh:malt:(\- m) uh and in AH : (1 in) A1> -.{l in) ali .in ah
respectively, then m is said to be the croBS-over value for the factor^ A and H.

300 Combination of Linkage Values

lowing proportions if it were of composition ABG.ahc, and similarly
for other compositions :

No cross-over ...

(ABC and abc).

(l-m)(l-n).

Cross-over between loci of A and B only ..

{aBG and Abc).

m(l-w).

5 and C only

{ABc and abC).

(l-m)n.

A and B and of B and C

(AbC and aBc).

mn.

Actually the last class has been shown to be in defect in many
cases. This has been thought to be due to the loops formed by the
chromosomes during synapsis having a modal length (3). If this were
so, we should expect to find an excessive number of double cross-overs
when the distance between the loci of A and G was equal to twice the
modal distance between points of crossing over. This phenomenon
has however not been recorded. The shortage of double cross-overs
can equally well be explained by the mere rigidity of the chromo-
somes, which makes sharp bending difficult. In the sex chromosome
of Drosophila the ratio ^ of observed to calculated numbers of double
cross-overs is '58 : 1 for eosin (white), vermilion, and sable (4) (where
m + n = -406), and -21:1 for vermilion, sable, and bar (3) (where
m + n = -239).

If the calculated number mn of double cross-overs occurred, the
cross-over value for A and C would be equal to the total number of
single cross-overs, i.e. to m (1 — n) -]- (1 — m) n, or m 4- n - 2mn.

If double cross-overs were impossible, but the full numbers of single
cross-overs occurred, as would happen if the chromosomes were straight
rigid rods, the cross-over value for A and G would obviously he m-\-n
(Morgan and Bridges' formula).

Finally if double cross-overs were impossible, and in every case
where one should have occurred according to the calculation above,
a single cross-over took its place, the cross-over value for A and G
would be m -I- n — 2mn + mn, or m +7i — mn. This case might be
approximately realised if the chromosome could not form loops shorter
than some definite length.

Hence the cross-ov«r values for A and G should be approximately
mi-n when m and n are small, m + n— %nn when their sum is large,
and m-\-n — mn for intermediate values.

Table I contains the observed values (5) for all triads of factors in
the sex chromosome of Drosophila for which each of the cross-over
values exceeds '1 (10 7o)- The first column gives the three factors
con6erned in each case ; the second and third columns give the cross-

^ Called by Muller the "coincidence."

J. H. S. IIaldank :u)\

over valiica for tho firnl lUid m^coml. luul H,.rMiwl atul thin! f/irtorn
runjKKJlivrly. iV m mm\ n. Thr f*nirth. liflli. umi nixlh roliiiniiM jfiv«-
the n\sultHS o( ihv thriH? pnivisioiDil .siiininiitinii (nnmjliu- «»)itniii<(l
Hb*>vo; tho st'Vciith ^ivo.s thi' «»hHrrvr<l ( ro.M.s-nvrr Viilin- (<»r ili«- lirMi

TAHLK I.

1

2

.1

4

.'.

r.

7

*

PWHoni

m

tt

m It n

iM It n mn i

m t N Tmn

OttM^rvnl

(laM

Whito iwiblo lolhiil jtr

•41-2

•230

648

•5.M

•15 I

HA)

V

Yollow wmiiliun nidimonUry .

•air,

•241

•586

•.■>03

•120

•429

V

bar

'Mr,

•239

•584

•502

•119

•479

y

White deiireMod bar

,, Millie forknl

•203

•3H()

•583

•.5(Ml

•129

•I3r,

> ■'

•412

•100

•572

•50<;

•410

•457

7

Yellow jMiblc riultnuMiiarv

•42U

•143

•572

•511

•4 19

r2*.»

i

bar

•429

•13«

•5r,7

•508

•449

•479

y

While \-crmilion fupcni

•aof)

•258

•5«i3

•4H4

■4(m;

133

y

Bifid vermilion forktti

•311

•245

•55r,

•480

•101

•425

V?

WhiU' sable rudimontAry

•412

•143

•555

496

•437

•421

A

Bifid vermilion riidimentHry ...

•311

•241

•552

•477

•4(»2

•427

7

White vermilion forked

•305

•245

•550

•475

•401

•457

7

, . iwble bar

■412

•138

•550

•49:{

•4:{6

•43<;

7A

Yellow miniatnro bar

•348

•205

•548

•47H

•407

•479

(i

White vermilion rudimentarv ...

•305

•241

•546

•472

•3'.>«l

•421

7

bar ... ' ...

•305

•239

•544

•471

•39H

•436

7

,, miniature bar

•332

•205

•537

•469

•401

•43»;

7

Yellow miniature nulinientary . . .

•343

•179

•522

•471

•419

•429

7

White miniature rudimentary .

•332

•179

•511

•452

•392

•424

7

„ reduplicatovl bar

•289

•206

•495

•435

•375

•436

(1

., furrowed forked

•303

•191

•494

•436

•37s

457

li'

Bifid miniature rudimentary ...

•30()

•179

•485

•430

•375

•4 '27

7^

White furrowed bar

•303

•179

•482

•4 '28

•374

•436

/^•'

Yellow vermilion sable

•345

•101

•446

•411

•37»;

•4-29

li

Facet venuilion sable

•32(i

•101

•427

•39 1

•361

130

a?

Depressed vermilion bar

•170

•239

•409

•3«;8

•328

•3.S0

fi-

White vermilion sable

•305

•101

•406

•375

34 1

112

a

Shifted vennilion bur...

•155

•239

•3«)4

•357

•3-20

•31 1

6!

White depressed vermilion

•203

•170

•373

•338

•304

•;io5

7 •

Yellow club vermilion

•177

•188

•365

•332

••29M

•315

rt

White lethal sb miniature

•156

•199

•355

•325

•295

■3.12

fi

Wliite club vermilion

•143

•188

•331

•301

•277

•305

;i

,, lemon vermilion

•145

•1*20

•2r,5

•248

•230

•305

a '

Vermilion sable forked

•101

•ir,o

•261

•245

•2*29

•215

'1y

„ ,, rudimentary...

•101

•143

•244

•230

■215

•241

,i

„ bar

•KH

•138

•239

•225

•211

•239

a.i

and third factors. In thu riglith coluinii thcsr nh.sci^vrd values an-
chissified us follows :

(Jreater than in & n . . a

Between lu + u and in -f- n — inn m

„ 111 + n — inn and m + /< — -"'" • 7

I>_*s.s than ni + n — 'linn . . . ^^

Those e.xactly e<jual to ni {- n are cla.ssiHcd a.s afS, and .s<. ..n. 'Ph«-

data are placed in the unler of the nia;^nnl ndr.s n\' m + //. W hrn- any

302 Combifiatiofi of Linkage Values

of the three observed vahies is based on a count of less than 500 in-
dividuals (in which case the probable error of the cross-over value
ma}^ exceed 1*5 °/^, as pointed out by the author (6) elsewhere) a query
is placed in the last column.

It will be seen that the observed values, when m-\-n exceeds *5,
lie almost wholly between m + n — mn and- m-^n — 2mn, as demanded
by the theory above. The three discordant values out of 19 are no
more than would be expected in view of the probable errors of the
observations due both to small numbers and differential mortality.
When m + n is less than *5 the results are somewhat more irregular,
as the calculated values from the three formulae are not very different,
but the majority of observations lie between m + n and m + n — mn, as
demanded by the theory.

This table also enables us to test the formulae given by Trow (2),
based on the reduplication theory. If reduplication takes place so that
A and B when coupled give the gametic series

qAB : lAb : laB : qab I cross-over value m = 1 ,

whilst B and G give the series

r BG :lBc:lbG:rbc{ cross-over value n = ) ,

V r + lj'

then A and G should give the series

(qr + l)BG:(q-\-r)Bc:(q + r)bG:(qr^l)bc

I cross-over value = ) .

\ qr + q + r -[- 1/

11 2

This latter value = H - ; — — = m + ^ — 2mn. Hence

q + l r+1 (g-hl)(r-f-l)

on this hypothesis the observed cross-over values for A and G should

cluster round m + n — 2mn, and approximately equal numbers should be

greater or less than it. In other words, as many values should fall in

class B as in classes a, /3, and 7 together.

The expectation is therefore 18 (S), 18 (a, 3, and 7); the actual
numbers are 3*5 (S), 32'5 (a, y8, and 7), reckoning the single value 7S
as half in each class. Hence the above form of Trow's theory is
untenable.

On a more complicated form of the same theory, which Sturtevant(7)
has shown to be impossible on other grounds, A and G when coupled

J. B. K. IIaldank :]0U

alone give a priiimry mticm sA(^ : \Ar . \<i(' snr and in zy^nivM nl rum-
position A Hi' . abc, A mTioti

(qrs + tt) A C : (7 + r) .4c : (7 + r) (iT : (r/r/* 4 n ) nr:

Crros.s-ovrr valin* =■ ' ).

f/rs \ 'I ^ r h HI

Ab thin vnluo is less than that of m f /* —2;/*//, it is .still iiiMir cliarly

im|H>HsibK\

Tho supporU*rs of tlir ri'duplimtion throry nmst tlunjnn- explain
the deficiency of the tKnibJe eross-over ehisses <»f )^ainet<* (whi<;h fmiu
a zygote? of conijMxsition ABC . ahc are Ah(^> and it lie), i )n the ehromo-
80me theor)' this is due to the rigidity of the chruinosonus, and until
an equally plausible explanation on the redtiplic^ition theory is givi-n,
the chn>mosome theory must be considered the more probable of the
two, so far Jis the class of evidence dealt with in this jm|H'r is concerne<l.

It has bt^en shown above that if .1, li and (' are three factors whose
loci lie in that onler in the same chromosome, and if //* and n are the
cross-over values for A^ B, and B, C n»spectively, then the value for A
and C is wH- n —pmn, where p is a number between 0 and 2, increasing
on the whole with m-^n, and having the value 1 when m + 71 = alxjut o.
The distJinces between loci may now be c^Uculated ;us follows:

Let X be the distance between the loci of two factoi*s, y their cross-
over value, and let the unit of distance be chosen so that when y is
sufficiently small .r becomes equal to y. This jissumption is legitimate
if we suppose that crossing over is as likely to occur (other things
being equal) at one point in the chromosome as another, i.e. that the
chromosome is equally flexible and breakable at all points. The unit
of distance is thus 100 times Morgan's unit.

If now we write y—f{x\ the form of this function being indeter-
minate,
.-. f (x -\- h) — f {x) -\-f(h) —pf{x)/(h), where h is any increment of x.

. f{x-^h)-f(x)_f(h)-pf(x)f{h)
h h

Now as h is decreased towards 0, - 7— tends to the limit 1.

h

dy ^ ^^/{x-hh)-f(x)
dx fc-M) h

= Lt-^^''^"-^^^^^-^^''^

A-M) 't

= 1 — j)f{r), where y> h;is the value assumed when //* ^ //, n ^ <>,
= 1 - P!/'

304 Combination oj Linkage Values

Therefore

X — I z , since x and y vanish together, and py <1.

J Q 1— pt

Hence if p were constant we should have

— 1 1 — e~^^
«=---iog«(i-i>yXor2/ = — -— (1).

Since however p varies between 0 and 2, the values of x 'must lie

— 1 1 — e~^^

between ^.and -Q-loge(l — 2y), those of y between x and — - — ; the

equation

3/ = ^ (2)

being nearly accurate for small values of x and y, the equation

X _ g-2a; _ 1

y= —2 — >orx=^\oge(l-2y) (3)

for large values of x and y, as is obvious, since for large values of x,
y approaches the value '5 asymptotically. The equation (2) corresponds
to Morgan's summation formula m + n, the equation (3) to Trow's
formula m-{-n — 2mn.

The equation (3) may be deduced more directly as follows for a
perfectly flexible chromosome :

Let a length x of the chromosome be considered as divided into
a very large number iV of small equal portions. Then the chance of a

X

cross-over in each of these is approximately ^. Hence the chance
(5f a cross-over in t of these segments and no more is

When N becomes infinite the limiting value of this expression, i.e.
the probability of exactly t and no more cross-overs in a length x, is

"^--tT W-

Hence the value of y for a given value of x is the sum of the
probabilities of all odd numbers of cross-overs.

.-. 3/ = Ci-Hc3-f C5-i-C7-[-

^ fx x^ x^ x' \

= ^ (r! + 3-! + 5-!+7-!+ )

= e~^ sinh x

= —^ (3).

J. U. H. FIaldank :]or)

In pnictict». howowr. «iwiii^ U* thi? n>(i«lity of tho rhn»iii<>H4iiiic, tla*
vahif of c, thiiH cnU'uluUHl in tini Hinull. ami thonr of r„, r,, «•!<-. 14h» lar>(«'.
They art* howi-vor inon* JMXMimU* for ^wnl lrIl^al^s. whin- tin- n^uUly
of the clm>iu<woint>.M atT«TU ihc n'.suh.M t«» a Ivhh ixtmt.

It in suggtNsUHl that tht* unit (»i'(iiMt.;iiKM> in n chrniiioHdiiK' im difiiH d
above bi* lA'niuKl a " inorpm,' on thf analogy of th«« ohm. voli, ,'\r.
Mor^iiH unit of ilistjuico is thrn*fon; a ct-ntimor^an.

To obtain a morf acounitr rrlation Iwtwrrn .r and y >%•• may ph»t
the curvi\H n^prfstiitin^' (Mjuations (2) and {'\), ami th«n nbt-iin •in
piric^illy a curvo lyini^ iK'twcm thr two which Ht.s the ol)s«'rv«d pmuIIh
a« closfl y JUS |x»ssihl('. This ha,s hr«'n d«»m' in thr fii^urt', whrn- lim- («/)
ri»piv.s*'nLs r«niation C2). t-urvc (/>) i'<|uation {'A), and curv** (/•)

•'• = ■"//- .', >«»Kr(l-2//) (.'»).

K<{uation (">) is nu'ivly choson to ^ivc* Jis ^o«m1 a fit as )Hi.ssihh'
juid h.'is pn»bably n«» theoretical significance. Tlie |M»ints repnsmtin^
observations an- plotted ;us follows:

The values of y in columns 2 and '.) of Table I are t^ik»*n, and tin*
corn."s|H)nding dist-:inces in mort^ans (values of./) read otf from curvr (c)
or Table II. which is c^vlculaLed from ecpiation (5). These latter are
addtnl together, and a {x)int plotted with their sum .-is absciss^i and the
observed cross-over value from column' 7 of Tal)le I :us ordinate. For
example the first row of Table I gives the following result:

The cross-over values '+12 and "2'M) correspond, according t<» th<*
curve (c), or better, by interpolation from Table II, to distances of •.■>4*J
an<l '261 morgans respectively. The sum of these distances is •SK), and
the obsi'rved cro.ss-over value from column 7 is '400. The point farthest
t<i the right is accordingly plotted with abscis.sa <S1() and ordinate MA).
Curve (c) gives the value 479 f(»r //, and the error «»f // is aco'idinglv
•019, or 19 /,.

It will be seen that IS of the observati(»ns lie above tli<' curve (r),
18 below, and that in only 4 cases, 3 of which an- among the it>uli.s
<pierie<i in Table I, does the error of ?/ ixceed 04 or 4 , . The pioliable
error of the cross-over values, ;us calculated from the curve, is I .S . nr.
omitting the (puriejl result.s. !(> /, . This result is not large cniiMderJUL,'
the probiible err«u-s of the values of// f<»r the points plotted, which raiiL^t
from 81 °/ downwards.

The curve gives s;itisf;ict.<»ry H'sults f«>r smaller cr<»ss-nV(r Nahn-s. but
these are not ph»tted. ;is they do not allow of much di^criminalinu Im -

Joam. of Gen. viii -'1

306 Combination of Linkage Vahies

tween the three equations. If the points had been plotted from either
line (a) or curve (6), 3J would have lain on one side, 32-| on the other,
as may be seen from Table I.

Hence the curve (c) may be taken as a fairly accurate guide to the
combination of linkage values, and this remains equally true whether
the chromosome theory is adopted or not. For this reason a series of
values of 100^ and lOOy {i.e. distances in centimorgans and cross-over
values as percentages) calculated from equation (5) are given in Table II.
As more results accumulate it should be possible to correct these values,
which are rather uncertain for large values of x and y.

TABLE II.

100?/ (Cross-over value as percentage) 0-0 5*0 8-0 100 11-0 12-0 13-0

100 'a; (Distance in centimorgans) ... O'O 5-1 8-2 10-3 11-4 12-5 13-6

100^ ... 14-0 15-0 160 17-0 18-0 19-0 20*0 21-0 22-0 23-0 24-0
100 .X ... 14-7 15-9 17-0 18-1 19-8 20-5 21-7 22-9 24-1 25-3 26-6

100?/ ... 25-0 26-0 27-0 28-0 29-0 30-0 31-0 32-0 33-0 34-0 35-0
100 a; ... 27-9 29-2 30-5 31-9 33-3 34-7 36-2 37-7 39-3 40-9 42-6

100?/ ... 36-0 37-0 38-0 39-0 40-0 41-0 42-0 43-0 44-0 45*0 46-0
100 .T ... 44-3 46-1 48-0 50-0 52-2 54-4 56-9 59-6 62-6 66-0 70-1

100?/ ... 47-0 480 49-0 49-5 49-7 49*8 49*9 50*0
100 a: ... 75-1 81-9 93-0 99-2 109-4 117*7 128-1 oo

As an example of the use of this table the following problem may be
taken :

'' The factors A and B give a cross-over value of 38*5 °l^, the factors
5 and (7 a value of 22-7 7,. What is the value for A and C? "

Fram the table we find by interpolation that the distance AB m
49*0 centimorgans, the distance BG 249. Hence the distance

AC = AB±BC= 73-9 or 241 centimorgans.

The cross-over value is therefore 46*8 ^/^ or 22*0 °/^, according as C lies
outside AB or between A and B. Morgan's formula would have given
61-2 7^ (an impossible value), oi: 15*8 7^; Trow's formula 43*7 7,^, or
28-9 7, (by solving the equation m 4- '227 - 2m x '227 = -385). On the
reduplication theory the result from AB + BG corresponds to the view
that the reduplication between A and G is " secondary " to those between
A, B and B, G; the result from AB — BG to the view that the redupli-
cation between A and B is secondary to those between A^ G and G, B.

It should be remarked that the existence of a quantity x which has
the property demonstrated above is not a conclusive proof of the chromo-
some theory, and indeed such a quantity may occur in certain forms
{e.g. Trow's) of the reduplication theory. However the fact that the

J. B. 8. Haldank :m7

VH)ut*t( (»f T r<>m'M|miMl t<» t.h<»m« tl«>injiii(|(Hl for \hv (liHtniici* nii thr hy|»«»-
thoiiiH thai thr fm*t«>rH iiru IikniUhI in a Hrmi-ri^itl <'hniiiiMs..iiH i?* n Htmii^
|M)int ill favour of that hy|Nitht«8iN.

Wo have How the ihiUi for a fairly juuMinit^' rMiimat*' «»! lln- UtUi\
length of the known {xtrtion of a rhronioHonic. r.7. the twx r|)n>np»Miiiir
in I>n>sophila Taking soinu of tlu- Ik-mI authonlicaUMl niriLHim iihuu
wc have :

FV-ton.

lOity

(CrtMn over valoo

In |icr osnl.)

100 j^ (fnin
Table III

Yellow White

11

11

WhiU'- Vermilion ...

SOA

:t/i 4

Vermilion Bar

•28 'J

2«ir,

B»rlx!lhal .c

H-:<

H-.-i

ToUIm 6^H 71-.*.

This i;ivos a total length of 7\o crntjinnr^^aiis a^MJii.Hi Mtir^Mii ami
Hri<i^os' i*stiinatr(<S) of t)(>'2. The HiHcrrjMiiicy is <lnr to th«' fad thiit in
some coni|)JinitiveIy lon^ se^nient-s of the chroinoH<uiic (cj/. lMtw««fi tin
K>ci of SaMi' and Kudinicntary, a distance of alnnit 1.') c«-ntiiiior^ans) no
fjict«.>rs have been located, and such dist;uices tend to !><• un<lerestiiiiat<'«l.
It may also bo duo in part to the large probable ern)r involved in usin^'
a large number of small distances.

From eipiation (4) wc may calculate the proportion ot chromosomes
giving t cross-overs in the known region. These valius are incoirect,
owing to the rigidity of the chromosome, c, being too low, the remainder
too high. The theoreticid values arc :

No cross-over in c„ = e~'''\ or 4f)l / of the chromosonu's

One „ in c, = 'Tloe""'', or ,*U-4 /^

Two cross-overs in c._. = ,^ , <jrl2G"/,^ „

•7153g-"15

Three „ in r, = , or 3() /^

rour „ in 04= ^^ , or M)

and so on.

The value of c, is Uhi low, the others too high. The nal value of
c, -f-Ca4- C5+ ... is the cross-over value of 4<)".S /,,and Morgan(.S) givrs
C2 + C4 -I- c« -H ..., the numbi'r of double cross-ovei-s (inchnling "juad-
ruples, etc.), as al>out 10 / , .so that q should be about 43 \Vh« 11

the relation between x and y is ;iceurat<ly known it will Ix- |>os.sible to
c«ilculat<i the values of c, with accunicy by integration.

308 Combination of Linkage Values

It is believed that the above method of estimating distances will
prove of considerable value when applied to comparatively long chromo-
somes in which factors are sparsely located, such as the second and third
in Drosophila, since there is no reason to suppose that the relation
arrived at between distance and cross-over value is peculiar to the sex
chromosome in Drosophila. The results of investigations on these chro-
mosomes should go far to confirm or refute the theory.

Outside Drosophila the best series of results on which to test it are
those of Altenburg(9) with the three factors M, S, and G in Primula
sinensis, quoted by Punnett(lO) in a recent paper. Here the cross-over
value for M and 8 is 11-6 %, for M and G 34-0 %, for S and G 40'6 7o,
each result being based on 3684 individuals. By Table II the distance
SM is 12'1 centimorgans, MG 40*9, and hence SG is 55*0 centimorgans
(assuming the loci to lie in the order SMG). Hence the cross-over value
for S and G should be 404 °l^, the observed value being 40*6 °/^, a very
nearly perfect fit. The addition formula gives 45*6 °/q, Trow's formula
37"7 7o- The probable error of the calculated result is '64 705 of the
observed '55 7o- Hence the probable value of their difference is *84 7o>
and though the close agreement is accidental, both the alternative for-
mulae are impossible.

In the case of Punnett's(lO) results for sweet peas the agreement is
also good, but owing to the closeness of the linkage, the three formulae
give nearly equal values. There is, however, no reason to suppose that
Table II does not represent with fair accuracy the relation between dis-
tance and cross-over value in all organisms, though the absolute value
in fifi of the unit of distance, or morgan, is presumably different in
different cases.

Summary.

By a consideration of the observed gametic ratios of the sex-linked
factors in Drosophila, the following results, among others, are arrived
at:

1. li A, B, and C are three factors lying in a chromosome in that
order, and if m is the cross-over value for A and B, n that for B and C,
then the value for A and G lies between m + n and m + n — 2mn, being
nearer to the former when m + n is small, to the latter when it is large.

2. A relation is arrived at, on the hypothesis that the chromosomes
are partially rigid, between cross-over value and distance, which permits
of the calculation of one of the cross-over values for three factors from
the other two, with a probable error of less than 2 °/^.

.1. B. S. IIaldank

:K)0

*X TliJM nOalittii nmy i\\m Ih» imM to mlnilat*' ih. hiul Inijjih n( n
chn»iiuiH«tiin\ and ihv iiuiiiIht uI <loul»|r uixi iriplr rn»HhuvrrH (41 )h>
exiKHrUxl ill a larj(i« iliHtaiic*'.

4, Thr n*HuItj< fniiii l>n»H.n»lnla jin* inroin|Mitit)|r with Tn»w'f» (nriii
of ihv riMiupliciititm tlntiry, Imt jMrhajw rmt with nthrr |Hifwilih- jnrtiiN
of its

5. Tho thfory (loNvlnjH'<l alxivc fits all thr «iI»»m ivrti «|jita m phuitw.

8.

!).

10.

KKKKHKNCKS.

MoKtiAN Jiiui HuiiHiKs. "S<'x-liiik«Hl inln'iilJiim- m I)n»'M.|.liilu ' funtr^/ir

/njttitufion <tf' WaAhlntjtnft, UHH, p. 21.
Tkow. " IViiuary ami S<Moiulary Ki^hipliiJitioii.' Jourmtl <.f(i,nfii,», V«.I. 11

p. 22. 1!»13.
MoKiJAN aiui Hl<ll»Uh>. /xx. rit. |». 13.
/^H-. rit. j». 37.

L<H . rit. |i. H J.

IIam»a\k. " TIu' Pn»lwil.l«' Krroi-s of Ol.s»Tvo<l Lmk.iu'' Viln*'--. " .In.inuxl ..^

(t'citfdcs. Vol. VIII. p. iM)i. i;m;».

STrRTKVAXT. **Tho H<Hltjpli« .it ion hyiMitliosiMippIitij Ut |)n>>«'phila. ,l»/i' ri'V,,,

XaturaliAt, V(»l. XLMII. p. .*»3r>, liM4.
MoHcJAN and HRIIHiKS. Aor. r/>. p. 8.
Altkmhiuj. rieiu'tics, Vol. i. p. 3')4, MHO.
PiNNKnT. " Ke4iiiplicviti<»ii .series in Sweet I'oics." Joiinml nf (irnctu*^ Vi>l. vi.

N.». 3, 1!)17.

05

0 4
0-3
0-2

/'

o

o o

■

- —

—

-O-

'i

3

^'

-*- -. — i—

- - ♦ — ■ 1 •

:

01

y

^\

-

— ,

. _

__ __,J

/

1

L. _

1 i

i . i

0 1 0 2 0 3 0-4 0 5 0-6 0 7 0 8 0 9 lO 11

X - di.'-tAncc

CAMBRIDGE : PRINTED BY

J. B. PEACE, M.A.,
AT THE UNIVERSITY PRESS

QH Journal of genetics

A31

AU6A

V.8
cop .2

Biological
& NMical
SchaU

PLEASE DO NOT REMOVE
CARDS OR SUPS FROM THIS POCKET

UNIVERSITY OF TORONTO UBRARY