Skip to main content

Full text of "Heredity in relation to evolution and animal breeding"

See other formats


Gift of H.W. Sheldon 










Printed in the United States of America 


THIS little book is based on a course of eight 
lectures delivered in November and December, 
1910, before the Lowell Institute, Boston, as well 
as on a course of five lectures delivered before 
the Graduate School of Agriculture held under 
the auspices of the Association of Agricultural 
Colleges and Experiment Stations at Ames, 
Iowa, in July, 1910. The hope is entertained 
that it may be of service to students and that it 
will also interest the general reader. 

The writer wishes to express his gratitude to 
the Carnegie Institution of Washington for per- 
mission, in its preparation, to draw freely upon 
published and unpublished material derived from 
investigations aided by the Institution. 

Acknowledgment is also due to the following 
persons, or to their publishers, for permission to 
use figures from their publications, as indicated 
in the text : Prof. E. B. Wilson and The Mac- 
millan Co., Prof. H. S. Jennings and The Ameri- 
can Naturalist, Dr. W. B. Kirkham and The 
American Book Co. 


JUNE, 1911 




















INDEX 183 



1. Egg and sperm of the sea-urchin, Toxopneustes 9 

2. Fertilization of the egg of .A^m's 12 

3. Egg of a mouse previous to maturation Facing 14 

4. Maturation and fertilization of the egg of 

a mouse Facing 14 

5. Diagrams showing the essential facts of 

chromosome reduction in the development 

of the sperm-cells 17 

6. An ordinary fern 21 

7. The prothallus of a fern 23 

8. Diagram showing the chromosome number in 

the spermatogenesis of ordinary animals 

and of the wasp 24 

9. Diagram showing the relation of the body to 

the germ-cells in heredity 29 

10. A young, black guinea-pig Facing 30 

11. An albino female guinea-pig Facing 30 

12. An albino male guinea-pig Facing 30 

13. Pictures of three living guinea-pigs and of the 

preserved skins of three others . . . Facing 32 

14. A black, female guinea-pig, and her young 

Facing 34 

15. An albino male guinea-pig Facing 34 

16. Two of the grown-up young of a black and of 

an albino guinea-pig Facing 34 




47. Diagrams to show the number and size of the 

classes of individuals to be expected from 
a cross involving Mendelian segregation 
without dominance 135 

48. Photographs to show variation in ear length 

of two varieties of maize, of their F x off- 
spring, and of their F 2 offspring . .Facing 138 

49. Diagram of sex-determination in partheno- 

genesis 162 

50. Diagram of sex determination when the female 

is homozygous, the male heterozygous .. . 167 

51. Diagram of sex-determination when the female 

is heterozygous, the male homozygous . . 170 

52. Diagram of sex-limited inheritance when the 

female is a heterozygote 173 

53. Diagram of sex-limited inheritance when the 

female is a homozygote, as in the red-eyed 
Drosophila 175 




THE theory of organic evolution has prob- 
ably influenced more fields of human 
activity and influenced them more pro- 
foundly than has any other philosophic deduc- 
tion of ancient or modern times. By this theory 
philosophy, religion, and science have been rev- 
olutionized, while in the practical arts of educa- 
tion and agriculture, twin foundation stones of 
the state, man has been forced to adopt new 
methods of procedure or to justify the old ones 
in the light of a new principle. 

The evolutionary idea has forced man to con- 
sider the probable future of his own race on 
earth and to take measures to control that fu- 
ture, a matter he had previously left largely to 
fate. With a realization of the fact that or- 


ganisms change from age to age and that he 
himself is one of these changing organisms man 
has attained not only a new ground for humility 
of spirit but also a new ground for optimism and 
for belief in his own supreme importance, since 
the forces which control his destiny have been 
placed largely in his own hands. 

The existence of civilized man rests ultimately 
on his ability to produce from the earth in suf- 
ficient abundance cultivated plants and domes- 
ticated animals. City populations are apt to 
forget this fundamental fact and to regard with 
indifference bordering at times on scorn agri- 
cultural districts and their workers. But let the 
steady stream of supplies coming from the land 
to any large city be interrupted for only a few 
days by war, floods, a railroad strike, or any 
similar occurrence, and this sentiment vanishes 
instantly. Man to live must have food, and 
food comes chiefly from the land. 

A knowledge of how to produce useful animals 
and plants is therefore of prime importance. 
Civilization had its beginning in the attainment 
of such knowledge and is limited by it at the 
present day. If, therefore, this knowledge can 



be increased, civilization may be advanced in a 
very direct and practical way. Before Darwin 
the practices of animal and plant breeders were 
largely empirical, based on unreasoned past ex- 
perience, just as was in antiquity the practice 
of metallurgy. Good plows and good swords 
were made long before a scientific knowledge 
of the metals was attained, but without that sci- 
entific knowledge the wonderful industrial de- 
velopment of this present age of steel would 
have been quite impossible. In a similar way, 
if not in like measure, we may reasonably hope 
for an advance in the productiveness of animal 
and plant breeding when the scientific principles 
which underlie these basic arts are better under- 
stood. Two practical problems present them- 
selves to the breeder : (1) how to make best use 
of existing breeds, and (2) how to create new 
and improved breeds better adapted to the con- 
ditions of present-day agriculture. We shall 
concern ourselves with the second of these only. 
The production of new and improved breeds 
of animals and plants is historically a matter 
about which we know scarcely more than about 

the production of new species in nature. Selec- 



tion has been undoubtedly the efficient cause o 
change in both cases, but how and why applie< 
and to what sort of material is as uncertaii 
in one case as in the other. The few grea 
men who have succeeded in producing by thei 
individual efforts a new and more useful typ 
of animal or plant have worked largely b; 
empirical methods. They have produced ; 
desired result but by methods which neithe 
they nor any one else fully understood o 
could adequately explain. So there exists a 
yet no true science of breeding but only ; 
highly developed art which was practiced a 
successfully by the ancient Egyptians, th 
Saracens, and the Eomans as by us. Th 
present, however, is an age of science; w 
are not satisfied with rule-of-thumb methods 
we want to know the why as well as the hoi 
of our practical operations. Only such knowl 
edge of the reasons for methods empirically 
successful can enable us to drop out of ou 
practice all superfluous steps and roundabou 
methods and to proceed straight to the marl 
in the most direct way. The industrial his 
tory of the last century is full of instances ii 



which a knowledge of causes in relation to 
processes, i. e. a scientific knowledge, has 
shortened and improved practice in quite un- 
expected ways. So we may not doubt the ulti- 
mate value in practice of a science of breeding, 
if such a science can be created. 

A beginning has been made during the last 
ten years, starting with the rediscovery of 
MendePs law of heredity in 1900. This book 
will be concerned largely with the operations 
of that law. 



A the outset we may with profit inquire 
what is meant by heredity. When a 
child resembles a parent or grand- 
parent in some striking particular, we say it 
inherits such-and-such a characteristic from 
the parent or grandparent in question. By 
heredity, then, we mean organic resemblance 
based on descent. 

Eesemblances due to heredity may exist even 
between individuals not related as ancestor and 
descendant, as for example between uncle and 
nephew. Here the resemblance rests on the 
fact that uncle and nephew are both descended 
from a common ancestor, and they resemble 
each other simply because they have both in- 
herited the same characteristic from that an- 
cestor. This form of inheritance is sometimes 
spoken of as collateral in distinction from direct 



inheritance. In all cases alike community of 
descent is the basis of resemblances which can 
be ascribed to heredity, whether direct or col- 
lateral. Mother and child, no less than uncle 
and nephew, resemble each other because they 
have received a common inheritance from a 
common ancestor. 

Three biological facts of fundamental im- 
portance to a right understanding of heredity 
were known imperfectly or not at all in the 
time of Darwin and Mendel. These are (1) 
the fertilization of the egg, (2) the maturation 
of the egg, which must precede its fertilization, 
and (3) the non-inheritance of " acquired ' 
characters. These we may consider in order. 

Every new organism is derived from a pre- 
existing organism, so far as our present ex- 
perience goes. It may not have been so al- 
ways. Indeed, on the evolution theory, we 
must suppose that living matter originally 
arose from lifeless, inorganic matter. But if 
it did, this may have occurred, and probably 
did occur, under physical conditions quite 
different from those now existing. At the 
present time the most exhaustive researches 



fail to reveal the occurrence of spontaneous 
generation, that is, the origin of living beings 
other than from pre-existing living beings. 

In asexual methods of reproduction a new 
individual arises out of a detached portion of 
the parent individual. Such methods of origin 
are varied and interesting, but do not concern 
us at present. In all the higher animals and 
plants a new individual arises, by what we call 
a sexual process, from the union of two minute 
bodies called the reproductive cells. They are 
an egg-cell furnished by the mother and a 
sperm-cell furnished by the father. 

There is a great difference in size between 
egg and sperm. The egg is many thousand 
times greater in bulk, as seen in Fig. 1, for 
example, yet the influence of each in heredity 
appears to be equal to that of the other. This 
fact shows unmistakably that the bulk of the 
reproductive cell is not significant in heredity. 
A large part of the relatively huge egg can 
have no part in heredity. It serves merely as 
food for the new organism, furnishing it with 
building material until such a time as it can 
begin to secure food for itself. The essential 



material, so far as heredity is concerned, is 
evidently found in egg and sperm alike. It 
is plainly small in amount and possibly con- 
sists merely in ferment-like bodies which, ini- 

FIG. 1. Egg and sperm (s) of the sea-urchin, Toxopneustes, 
both shown at the same enlargement. (After Wilson.) 

tiate certain metabolic processes in a suitable 
medium represented by the bulk of the egg. 
The amount of a ferment used in starting a 
chemical change bears no relation, as is well 
known, to the amount of the chemical change 
which it can bring about in a suitable medium. 



The equal share of egg and sperm in deter- 
mining the character of offspring is well shown 
in the following experiment. An albino guinea- 
pig is one which lacks in large measure the 
ability to form black pigment. Apparently it 
does not possess some ingredient or agency 
necessary for the production of pigment. Now, 
if an albino male guinea-pig, such as is shown 
in Fig. 15, be mated with a black female guinea- 
pig of pure race, such as is shown in Fig. 14, 
young are produced all of which are black, like 
the mother, none being albinos, like the father. 
Fig. 16 shows black offspring produced in this 
way. Exactly the same result is obtained from 
the reverse cross, that is, from mating an al- 
bino mother with a black sire. It makes no 
difference, then, whether the black parent be 
mother or father, its blackness regularly domi- 
nates over the whiteness of the albino parent, 
so that only black offspring result. This fact, 
which has been repeatedly confirmed, shows that 
the black character is transmitted as readily 
through the agency of the minute sperm-cell 
as through the enormously greater egg-cell. 

Let Us now consider what happens when egg 


and sperm unite, in what we call the fertiliza- 
tion of the egg. The egg is a rounded body 
incapable of motion, but the sperm is a minute 
thread-like body which moves like a tadpole 
by vibrations of its tail. In the case of most 
animals which live in the water, egg and sperm- 
cells are discharged into the water and there 
unite and develop into a new individual, but 
in the case of most land animals this union takes 
place within the body of the mother. We may 
consider an illustration of either sort. 

The fertilization of the egg of a marine 
worm, Nereis, is shown in Fig. 2. The thread- 
like sperm penetrates into the egg. Its en- 
larged head-end forms there a small nuclear 
body, which increases in size until it equals 
that of the egg-nucleus, with which it then fuses. 
The egg next begins to divide up to form the 
different parts of a new worm-embryo. To 
each of these parts the nuclear material of egg 
and sperm is distributed equally. Since this 
development takes place wholly outside the 
body of either parent it is necessary that the 
egg contain enough food to last until the young 
worm can feed itself. This food material is 



FIG. 2. Fertilization of the egg of Nereis. 
A. The sperm has entered the egg and is forming a minute 
nucleus at a*. The egg-nucleus is breaking up preparatory 
to the first maturation division. B. The egg-nucleus is 
undergoing the first maturation division. Notice the con- 
spicuous rod-like chromosomes separating into two groups. 
The sperm-nucleus ( $ ) is now larger and lies deeper in the 
egg. C. A small polar-cell has been formed above by the 
first maturation division of the egg. A second division is 
in progress at the same point. The sperm-nucleus is now 
deep in the egg and is preceded by a double radiation (am- 
phiaster). D. Two polar-cells are fully formed. The ma- 
tured egg-nucleus is now fusing with the sperm-nucleus. 
An amphiaster indicates that division of the egg will soon 
take place. (After Wilson.) 



represented in part by the conspicuous oil- 
drops seen in the egg (the heavy circles in 
Fig. 2). 

The egg of a mouse needs no such store of 
nourishment, since in common with the young 
of other mammals the mouse-embryo nourishes 
itself by osmosis from the body fluids of the 
mother. The mouse-egg is accordingly smaller. 
Stages in its fertilization are shown in Fig. 4. 
In A the sperm has already entered the egg. 
Remnants of its thread-like tail may still be 
seen there. Nearby is seen a nuclear body 
derived from the sperm-head. Opposite is 
seen the nuclear body furnished by the egg 
itself. The two nuclear bodies fuse and their 
united substance is then distributed to all 
parts of the embryo-mouse, just as happens in 
the development of the worm, Nereis. 

There are reasons for thinking that the 
nuclear material is especially important in re- 
lation to heredity and that the equal share of 
the two parents in contributing it to the em- 
bryo is not without significance, for inheritance, 
as we have seen, is from both parents in equal 
measure. In cases where the inheritance from 



each parent is different it can be shown 
that the offspring possess two inherited possi- 
bilities, though they may show but one. Thus 
in the case of a black guinea-pig, one of whose 
parents was white, the other black, it can be 
shown that the animal transmits both qualities 
(black and white) which it received from its 
respective parents, and transmits them in equal 
measure. For, if the cross-bred black animal 
be mated with a white one, half the offspring 
are black and half of them white. The cross- 
bred black animal inherited black from one 
parent, white from the other. It showed only 
the former, but on forming its reproductive 
cells it transmitted black to half of these, white 
to the other half. Hence the cross-bred black 
individual was a duality, containing two possi- 
bilities, black and white, but its reproductive 
cells were again single, containing either black 
or white, but not both. 

Now it has been shown in recent years that 
the nuclear material in the reproductive cells 
behaves exactly as do black and white in the 
cross just described. This nuclear material 
becomes doubled in amount at fertilization, 


FIG. 3. Egg of a 
mouse previous to 
maturation. (After 

FIG. 4. Maturation and fertilization of the egg of a mouse. 
A. The first maturation division in progress. B, The first 
polar-cell fully formed ; the second maturation division 
in progress. C. The second maturation division com- 
pleted ; the second polar-cell is the smaller one ; near it, 
in the egg, is the egg-nucleus, and at the left is the sperm- 
nucleus. D. A view similar to the last, but showing only 
one polar-cell, the second; note its twelve distinct 
chromosomes ; near the sperm-nucleus in the egg, at 
the left, is seen the thread-like remains of the sperm-tail. 
(After Kirkham.) 


equal contributions being made by egg and 
sperm. This double condition persists through- 
out the life of the new individual in all its 
parts and tissues. But if the individual forms 
eggs or sperm, these, before they can function 
in the production of a new individual, must 
undergo reduction to the single condition. 

This reduction process is called maturation; 
it is well illustrated in the case of the mouse- 
egg, whose fertilization has already been de- 
scribed. The large nucleus of the egg-cell, as 
it leaves the ovary, is either broken up or about 
to break up preparatory to a cell-division. The 
most conspicuous of the nuclear constituents 
are some dense, heavily staining bodies called 
chromosomes, about twenty-four in number. 
In Fig. 3 each of these is split in two, prepara- 
tory to the first maturation division. The egg 
now divides twice, both times very unequally 
(Fig. 4), forming thus two smaller cells called 
polar cells, or polar bodies. They take no part 
in the formation of the embryo. The chromo- 
somes left in the egg after these two divisions 
are only about half as numerous as before, or 
about twelve in number. These form the chro- 



matin contribution of the egg to the production 
of a new individual. It is possible that other 
cell constituents undergo a similar reduction 
by half during maturation, but of this we have 
no present knowledge. 

The known fact of chromosome reduction, 
of course, favors the current interpretation 
that the chromosomes are bearers of heredity, 
though it by no means proves the correctness 
of that interpretation. In the egg of Nereis, 
as well as in that of the mouse, two matura- 
tion divisions precede the fertilization of the 
egg. See Fig. 2. In B the first maturation 
division is in progress; in C the second is in 
progress; and in D both polar cells are fully 
formed, while egg and sperm nuclei are unit- 
ing. Similar processes occur in eggs gener- 
ally, prior to their fertilization. 

Like changes occur also in the development 
of the sperm-cells. In Fig. 5 the original or 
unreduced condition of the chromosomes in a 
cell of the male sexual gland is shown (at A) as 
one of four chromosomes to a cell. After a series 
of changes involving as in the maturation of the 
egg two cell-divisions, we find (at H) that the 



products of the original cell contain in each case 
two chromosomes, half the original number. 

FIG. 5. Diagrams showing the essential facts of chromosome 
reduction in the development of the sperm-cells. (After 

These chromosomes make up the bulk of the 
head of the sperm which forms from each of 


these cells, its tail being derived from other 
portions of the cell. 

It follows that not only eggs but also sperms, 
prior to their union in fertilization have passed 
into a reduced or single state as regards their 
chromatin constituents, whereas the fertilized 
egg, and the organism which develops from it, 
is in a double condition. It will be convenient 
to refer to the single condition as the N condi- 
tion, the double as the 2 N condition. 

From a wholly different source we have 
evidence strongly confirmatory of the conclu- 
sion that the fertilized egg contains a double 
dose of the essential nuclear material. By arti- 
ficial means it has been found possible to cause 
the development of an unfertilized egg. The 
means employed may be of several different 
sorts, such as stimulation with acids, alkalies, 
or solutions of altered density. In such ways 
the development has been brought about of the 
eggs of sea-urchins, star-fishes, worms, and mol- 
lusks, which normally require fertilization to 
make them develop. 

The sea-urchin egg has been made to develop 
more successfully than any other. This has 



occurred even after the egg had undergone 
maturation, being reduced to the N condition. 
From the development of such reduced but 
unfertilized eggs fully normal sea-urchins have 
been obtained which even contain developed 
sexual glands. On the other hand it has been 
found possible to break the egg into fragments 
by shaking it, or cutting it into bits with fine 
knives or scissors. It has also been found 
possible to bring about the development of 
an egg fragment so obtained, a fragment 
which contained no egg nucleus. This result 
has been attained by allowing a sperm to enter 
it and form there a nuclear body. No adult 
organism has yet been reared from such a 
fertilized egg-fragment, but so far as the de- 
velopment has been followed it progressed 

There can accordingly be no doubt that the 
nuclear material of a sperm-cell has all the 
capabilities of that of an egg-cell and can in- 
deed replace it in development. Accordingly, 
when, as in normal fertilization, both an egg 
nucleus and a sperm nucleus are present in 
the cell, a double dose of the necessary nuclear 



material is supplied. The second or extra dose 
is, however, not superfluous. It probably adds 
to the vigor of the organism produced, and in 
some cases at least, materially affects its form. 
For many animals and plants exist in two 
different conditions, in one of which the nu- 
clear components are simple, N, while in the 
other they are double, 2 N. Thus in bees, 
rotifers, and small Crustacea the egg may 
under certain conditions develop without being 
fertilized. If the egg develops before matura- 
tion is complete, that is in the 2 N condition, 
the animal produced is a female, like the 
mother which produced the egg. But if the 
egg undergoes reduction to the N condition 
before beginning its development, then it pro- 
duces a male individual, an organism, so far 
as reproduction is concerned, of lower meta- 
bolic activity. 

In many plants, too, individuals of N and 
of 2 N constitution occur, which differ markedly 
in appearance. Thus the ordinary fern-plant 
is a 2 N individual, but it never produces 2 N 
offspring. Fig. 6 shows an ordinary fern- 
plant, which produces spores on the under 



FIG. 6. An ordinary fern, which reproduces by asexual spores. 
The fern is shown reduced in size at 382; a portion of a 
frond seen from below and slightly enlarged, at 383; a 
cross-section of the same more highly magnified, at 384. 
Notice in 384 the sporangia, and in 385 one of these dis- 
charging spores. (After Wossidlo, from Coulter Barnes 
and Cowle's Textbook of Botany.) 




surface of its fronds. Each of those spores 
is a reproductive cell which, like the mature 
eggs and sperm of animals, is in a reduced 
nuclear condition (N). These spores germi- 
nate, however, without uniting in pairs and 
form a plant different from the parent, just 
as the mature egg of a bee, if unfertilized, 
develops into an individual different from the 
parent, in that case a male. The plant which 
develops from the spore of a fern is small and 
inconspicuous and is known as a prothallus. 
See Fig. 7. It produces sexual cells (eggs and 
sperm) which, uniting in pairs, form fern-plants, 
2 N individuals. Thus there is a constant alter- 
nation of generations, fern-plants (2 N), which 
produce prothalli (N), and then these produce 
again fern-plants (2 N). 

The fact is worthy of note that in an animal 
or plant which is in the single or N condition, 
there occurs no chromatin reduction at the 
formation of reproductive cells. Its cells are 
already in the single condition, and they 
probably cannot be further reduced without 
destroying the organism. The 2 N fern-plant 
forms reproductive cells, its spores, which are 



in the reduced condition, N, and these germi- 
nate into the prothallus, which accordingly is 

FIG. 7. The prothallus of a fern, which reproduces by sexual 
cells, eggs and sperm. The eggs are borne in the sac-like 
"archegonia," just below the notch hi the figure. They, 
like the sperm -forming "antheridia," lie on the under sur- 
face of the flattened prothallus which is here viewed from 
below. Notice the root-hairs or rhizoids by which the 
plant feeds. Highly magnified. (After Coulter, Barnes, 
and Cowles.) 

N throughout. But when the prothallus forms 
reproductive cells, no reduction occurs. Its 
egg-cells and its sperm-cells in common with 



all other cells of the prothallus are already 
in the reduced condition without any matura- 
tion divisions. The result of their union in 
pairs, at fertilization, is the formation of 2 N 
combinations that germinate into fern-plants. 
Similarly in the case of a male animal which 

FIG. 8. Diagram showing the chromosome number in the 
spermatogenesis of ordinary animals (upper line) and of the 
wasp (lower line). 

has developed from a reduced but unfertilized 
egg, no reduction occurs at the formation of 
its sperm-cells. In an ordinary male animal, 
one which is in the double or 2 N state, the 
development of the sperms is attended by re- 
duction to the N condition. In this process 
there occur two cell-divisions producing from 
each initial cell four sperms. See Fig. 5, and 


Fig. 8, upper line. But in the male wasp, 
whose cells are in the N condition at the be- 
ginning, one of these divisions is so far sup- 
pressed that the resulting cell products are of 
very unequal size, and the smaller one contains 
no nuclear material. The other then gives rise 
to two sperm-cells, each possessing the origi- 
nal N nuclear condition, while the small non- 
nucleated cell degenerates. See Fig. 8, lower 

In conclusion, I wish to introduce two tech- 
nical terms, which it will be convenient for us 
to use in subsequent discussions. These are 
gamete and zygote. A reproductive cell (either 
egg or sperm) which is in the reduced condi- 
tion (N) ready for union in fertilization is 
called a gamete. The result of fertilization is 
a zygote, a joining together of two cells each 
in the N condition. The result is a new or- 
ganism, at first a single cell, in the 2 N 




1903. "The Heredity of Sex." Bull. Mus. Comp. Zool- 
ogy, 40, pp. 189-218. 

1898. "Embryons sans noyau maternel." Compte rendu, 
Academic des sciences, Paris, 127, pp. 528-531. 

1909. "Le sexe chez les Oursins issus de parthe- 
nogenese experimentale." Compte rendus, Academic des 
sciences, Paris, 148, pp. 453-455. 


1907. Maturation of the Egg of the White Mouse." 
Trans. Conn. Acad. of Arts and Sciences, 13, pp. 65-87. 


1899. "On the Nature of the Process of Fertilization and 
the Artificial Production of Normal Larvae (Plutei) from 
the Unfertilized Eggs of the Sea-urchin." Amer. Journ. 
of Physiol, 3, pp. 135-138. 


1905. "Die X-Generation und die 2 X-Generation." 

Biologisches Centralblatt, 25, pp. 97-117. 

1908. "Die Spermatozytenteilungen bei der Hornisse 
(Vespa crabo L.)." Arch. f. mik. Anat. u. Entwick., 71, 
pp. 571-587. 


1896. "The Cell in Development and Inheritance," 370 
pp., illustrated. The Macmillan Co., New York. 



IN the last chapter we discussed two bio- 
logical principles which, if clearly grasped, 
greatly simplify an understanding of the 
process of heredity. These are as follows: 

(1) A sexually produced individual arises 
from the union of two reproductive cells (or 
gametes), each of which contains, so far as 
heredity is concerned, a full material equip- 
ment for the production of a new individual. 
Accordingly, the newly produced individual is 
two-fold or duplex as concerns the material 
basis of heredity. 

(2) If the new individual becomes adult and 
forms gametes, the production of these will be 
attended by a reduction to the simplex or 
single condition as regards the material basis 
of heredity. 



To these two principles we may now add 
a third, viz.: (3) The individual consists 
of two distinct parts: first, its body destined 
to die and disintegrate after a certain length 
of time; and, secondly, the germ-cells con- 
tained within that body, capable of indefinite 
existence in a suitable medium. 

The fertilized egg or zygote begins its in- 
dependent existence by dividing into a number 
of cells. These become specialized to form 
the various parts and tissues of the body, 
muscle, bone, nerve, etc., and by becoming thus 
specialized they lose the power to produce any- 
thing but their own particular kind of special- 
ized tissue; they cannot reproduce the whole. 
This function is retained only by certain un- 
differentiated cells found in the reproductive 
glands and known as germ-cells. They are 
direct lineal descendants of the fertilized egg 
itself. If they are destroyed the individual 
loses the power of reproduction altogether. 

External influences which act upon the body 
may of course modify it profoundly, but such 
modifications are not transmitted through the 
gametes, because the gametes are not derived 



from body-cells, but from germ-cells. This 
relationship first pointed out by Weismann 
may be expressed in a diagram, as in Fig. 9. 
Only such environmental influences as directly 
alter the character of the germ-cells will in 
any way influence the character of subsequent 
generations of individuals derived from those 

FIG. 9. Diagram showing the relation of the body (S) to the 
germ-cells (G) hi heredity. (After Wilson.) 

germ-cells. Body (or somatic) influences are 
not inherited. This knowledge we owe largely 
to Weismann, who showed experimentally that 
mutilations are not inherited. The tails of 
mice were cut off for twenty generations in 
succession, but without effect upon the char- 
acter of the race. Weismann also pointed out 
the total lack of evidence for the then current 
belief that characters acquired by the body 

are inherited. The correctness of his view that 

body and germ-cells are physiologically distinct 



is indicated by the results obtained when germ- 
cells are transplanted from one individual to 

Heape showed some twenty years ago that 
if the fertilized egg of a rabbit of one variety 
(for example an angora, i. e. a long-haired, 
white animal) be removed from the oviduct of 
its mother previous to its attachment to the 
uterine wall, and be then transferred to the 
oviduct of a rabbit of a different variety (for 
example a Belgian hare, which is short-haired 
and gray), the egg will develop normally in 
the strange body and will produce an individual 
with all the characteristics of the real (an- 
gora) mother unmodified by those of the foster 
mother (the Belgian hare). Young thus ob- 
tained by Heape were both long-haired and 
albinos, like the angora mother. To this ex- 
periment the objection might be offered that 
the transplanted egg was already full-grown 
and fertilized when the transfer was made, and 
that therefore no modification need be expected, 
but if the egg were transferred at an earlier 
stage the result might have been different. In 
answer to such a possible objection the follow- 


FIG. 10. A young, black guinea-pig, about three 
weeks old. Ovaries taken from an animal like 
this were transplanted into the albino shown 

FIG. 11. An albino female guinea-pig. Its 
ovaries were removed, and in their place were 
introduced ovaries from a young, black guinea- 
pig, like that one shown in Fig. 10. 

FIG. 12. An albino male guinea-pig, with which 
was mated the albino shown in Fig. 11. 


ing experiment performed by Dr. John C. Phil- 
lips and myself may be cited. 

A female albino guinea-pig (Fig. 11) just at- 
taining sexual maturity was by an operation 
deprived of its ovaries, and instead of the re- 
moved ovaries there were introduced into her 
body the ovaries of a young black female 
guinea-pig (Fig. 10), not yet sexually mature, 
aged about three weeks. The grafted animal 
was now mated with a male albino guinea-pig 
(Fig. 12). From numerous experiments with 
albino guinea-pigs it may be stated emphati- 
cally that normal albinos mated together, with- 
out exception, produce only albino young, and 
the presumption is strong, therefore, that had 
this female not been operated upon she would 
have done the same. She produced, however, 
by the albino male three litters of young, 
which together consisted of six individuals, all 
black. (See Fig. 13.) The first litter of young 
was produced about six months after the oper- 
ation, the last one about a year. The trans- 
planted ovarian tissue must have remained in 
its new environment therefore from four to 
ten months before the eggs attained full growth 



and were discharged, ample time, it would seem, 
for the influence of a foreign body upon the 
inheritance to show itself were such influence 

In the light of the three principles now 
stated, viz. (1) the duplex condition of the 
zygote, (2) the simplex condition of the 
gametes, and (3) the distinctness of body and 
germ-cells, we may proceed to discuss the 
greatest single discovery ever made in the 
field of heredity, Mendel's law. 



1911. "On Germinal Transplantation in Vertebrates." 
Carnegie Institution of Washington, Publication No. 
144, 26 pp., 2 pi. 

1890. "Preliminary Note on the Transplantation and 
Growth of Mammalian Ova within a Uterine Foster- 
mother.' 7 Proc. Roy. Soc., 48, pp. 457-458. 
1897. "Further Note," etc. Id. 62, pp. 178-183. 

1893. "The Germ-Plasm." Translation by Parker and 
Romfeldt. Chas. Scribner's Sons, New York. 

FIG. 13. Pictures of three living guinea-pigs (A, B, (T), 
and of the preserved skins of three others (D, E, F); 
all of which were produced by the pair of albinos shown 
in Figs. 11 and 12. 



teacher of the physical and natural 
sciences in a monastic school at Briinn, 
Austria, in the second half of the last cen- 
tury. He was, therefore, a contemporary of 
Darwin, but unknown to him as to nearly 
all the great naturalists of the period. Al- 
though not famous in his lifetime, it is clear 
to us that he possessed an analytical mind 
of the first order, which enabled him to plan 
and carry through successfully the most origi- 
nal and instructive series of studies in hered- 
ity ever executed. The material which he used 
was simple. It consisted of garden-peas, which 
he raised in the garden of the monastery. 
The conclusions which he reached were like- 
wise simple. He summed them up, the results 
of eight years of arduous work, in a brief 
paper published in the proceedings of the local 



scientific society. There they remained un- 
heeded for thirty-four years, until their author 
had long been dead. Meantime biological sci- 
ence had made steady progress. It reached 
the position Mendel had attained in advance 
of his time, and Menders law was rediscov- 
ered simultaneously in 1900 by De Vries in 
Holland, by Correns in Germany, and by 
Tschermark in Austria. It gratifies our sense 
of poetic justice that to-day the rediscovered 
law bears the name, not of any one or of all of 
its brilliant rediscoverers, but of the all-but- 
forgotten Mendel. 

The essential features of this law can best 
be explained in connection with some illustra- 
tions, which I choose for convenience from my 
own experiments. If a black guinea-pig of 
pure race (Fig. 14) be mated with a white one 
(Fig. 15), the offspring will, as explained on 
page 10, all be black; none will be white. 
To use Mendel's terminology, the black char- 
acter dominates in the cross, while white 
recedes from view. The black character is, 
therefore, called the dominant character; 
white, the recessive character. 



g d CO 


la I 

11 I 


.SP bB P 

ft -a ^ 


|-3 & 



I | IS I 

n5 o cd^ N^ 


But, if now two of the cross-bred black in- 
dividuals (Fig. 16) be mated with each other, 
the recessive white character reappears on the 
average in one in four of the offspring (Fig. 

FIG. 18. Diagram to explain the result shown in Fig. 17. 

17). Its reappearance in that particular pro- 
portion of the offspring may be explained as 
follows (see Fig. 18) : The gametes which 
united in the original cross were, one black, 
the other white in character. Both characters 



were then asociated together in the offspring; 
but black from its nature dominated, because 
white in this case is due merely to the lack of 
some constituent supplied by the black gamete. 
But when the cross-bred black individuals on 
becoming adult form gametes, the black and 
the white characters separate from each other 
and pass into different cells, since, as we have 
seen, gametes are simplex. Accordingly, the 
eggs formed by a female cross-bred black are 
half of them black, half of them white in char- 
acter, and the same is true of the sperms 
formed by a male cross-bred black. The com- 
binations of egg and sperm which would natu- 
rally be produced in fertilization are accord- 
ingly 1BB:2BW:1WW, or three combina- 
tions containing black to one containing only 
white, which is the ratio of black to white off- 
spring observed in the experiment. 

Now the white individual may be expected 
to transmit only the white character, never the 
black, because it does not contain that char- 
acter. Experiment shows this to be true. 
White guinea-pigs mated with each other pro- 
duce only white offspring. But the black in- 




dividuals of this generation are of two sorts, 
- B B and B W in character. The B B indi- 
vidual is pure, so far as its breeding capacity 
is concerned. It can form only black (B) 
gametes. But the B W individuals may be 
expected to breed exactly like the cross-bred 
blacks of the previous generation, forming 
gametes, half of which will carry B, half W. 
Experiment justifies both these expectations. 
The test may readily be made by mating the 
black animals one by one with white ones. 
The pure (or B B) black individual will pro- 
duce only black offspring, whereas those not 
pure, but B W in character, will produce off- 
spring half of which on the average will be 
black, the other half white. These two kinds 
of dominant individuals obtained in the second 
generation from a cross we may for conven- 
ience call homozygous and heterozygous, fol- 
lowing the convenient terminology of Bateson. 
A homozygous individual is one in which like 
characters are joined together, as B with B; 
a heterozygous individual is one in which unlike 
characters are joined together, as B with W. 
It goes without saying that recessive individ- 
4 37 


uals are always homozygous, as WW for ex- 
ample. For they do not contain the dominant 
character, otherwise they would show it. 

It will be observed that in the cross of black 
with white guinea-pigs black and white behave 
as units distinct and indestructible, which may 
meet in fertilization but separate again at the 
formation of gametes. Mendel's law as illus- 
trated in this cross includes three principles: 
(1) The existence of unit-characters, (2) domi- 
nance, in cases where the parents differ in a 
unit-character, and (3) segregation of the units 
contributed by the respective parents, this seg- 
regation being found among the gametes formed 
by the offspring. 

The principles of dominance and segregation 
apply to the inheritance of many characteristics 
in animals and plants. Thus in guinea-pigs a 
rough or resetted coat (Figs. 23 and 24) is domi- 
nant over the ordinary smooth coat. If a pure 
rough individual is crossed with a smooth one, 
all the offspring are rough ; but in the next gen- 
eration smooth coat reappears in one fourth of 
the offspring, as a rule. Again, in guinea-pigs 
and rabbits a long or angora condition of the 


FIG. 20. Radiograph of a hand similar to those shown in 
Fig. 19. Notice the short, two-jointed fingers. (After 


fur is recessive in crosses with normal short 
hair. All the immediate offspring of such a 
cross are short-haired, but in the next genera- 
tion long hair reappears in approximately one 
fourth of the offspring. 

In cattle, the polled or hornless condition is 
dominant over the normal horned condition; 
in man, two- jointed fingers and toes (Figs. 19 
and 20) are dominant over normal three- jointed 
ones. This is clear from an interesting pedi- 
gree given by Farabee of the inheritance of 
the abnormality in a Pennsylvania family (see 
Fig. 21). In no case was an abnormal mem- 
ber of the family known to have married any 
but an unrelated normal individual. It will 
be seen that approximately half the offspring 
throughout the four generations of offspring 
shown in the table were of the abnormal sort, 
- short-bodied and with short fingers and toes. 

In each of the cases thus far considered a 
single unit-character is concerned. Crosses in 
such cases involve no necessary change in the 
race, but only the continuance within it of two 
sharply alternative conditions. But the result 
is quite different when parents are crossed 





-i' I"' 

IQ-+ l^ 


* I 





o* g 

| | 

bC -2 

> ^ 





FIG. 22. A smooth, dark guinea-pig. 

FIG. 23. A rough, white guinea-pig. 

FIG. 24. A dark, rough guinea-pig. The new 
combination of characters obtained when 
animals are mated like those shown in 
Figs. 22 and 23. 


which differ simultaneously in two or more 
independent unit-characters. Crossing- thembe- 
comes an active agency for the production of 
new varieties. 

In discussing the crosses now to be described 
it will be convenient to refer to the various 
generations in more precise terms, as Bateson 
has done. The generation of the animals orig- 
inally crossed will be called the parental gen- 
eration (P) ; the subsequent generations will 
be called filial generations, viz. the first filial 
generation (F ), second filial (F 2 ), and so on. 

When guinea-pigs are crossed of pure races 
which differ simultaneously in two unit-charac- 
ters, the F offspring are all alike, but the F 2 
offspring are of four sorts. Thus, when a 
smooth dark animal (Fig. 22) is crossed with 
a rough white one (Fig. 23) the F t offspring 
are all rough and dark (Fig. 24), manifesting 
the two dominant unit-characters, dark coat 
derived from one parent, rough coat derived 
from the other. But the F 2 offspring are of 
four sorts, viz. (1) smooth and dark, like one 
grandparent, (2) rough and white, like the 
other grandparent, (3) rough and dark, like 



the F generation, and (4) smooth and white, 
a new variety (Fig. 25). It will be seen that 
the pigmentation of the coat has no relation 
to its smoothness. The dark animals are either 
rough or smooth, and so are the white ones. 
Pigmentation of the coat is evidently a unit- 
character independent of hair-direction, and as 
new combinations of these two units the cross 
has produced two new varieties, the rough 
dark and the smooth white. 

Again, hair-length is a unit-character inde- 
pendent of hair-color. For if a short-haired 
dark animal (either self or spotted, Fig. 26) 
be crossed with a long-haired albino (Fig. 27), 
the F offspring are all short-haired and dark 
(Fig. 28) ; but the F 2 offspring are of four 
sorts, viz. (1) dark and short-haired, like one 
grandparent, (2) white and long-haired, like 
the other, (3) dark and long-haired, a new com- 
bination (Fig. 29), and (4) white and short- 
haired, a second new combination (compare 
Fig. 25). 

Now the four sorts of individuals obtained 
from such a cross as this will not be equally 
numerous. As we noticed in connection with 


FIG. 25 

FIG. 2 9. 

FIG. 25. A smooth, white guinea-pig. A second new combination of 
characters, but obtained first among the grandchildren of such an- 
imals as are shown in Figs. 22 and 23. 

FIG. 26. A short-haired, pigmented guinea-pig. ("Dutch-marked" 
with white.) 

FIG. 27. A long-haired, albino guinea-pig. 

FIG. 28. Offspring produced by animals of the sorts shown hi Figs. 26 
and 27. One shows the " Dutch-marked " pattern as a belt of 
pale yellow ; the other does not. Both are short-haired and 
pigmented (not albinos). 

FIG. 29. A long-haired, pigmented guinea-pig, " Dutch-marked " 
with white. Its parents were like the animals shown in Fig. 28 ; 

sVi/vnrn in TTir 

9A anrl 97 


the black-white cross, dominant individuals are 
to the corresponding recessives as three to one. 
Therefore, we shall expect the short-haired in- 



FIG. 30. Diagram to explain the result of a cross between the 
sorts of guinea-pigs shown in Figs. 26 and 27. L stands 
for long hair, S for short hair, D for dark hair, and W for 
white hair. Dominant characters are indicated by heavy 

dividuals in F 2 to be three times as numerous 
as the long-haired ones, and dark ones to be 
three times as numerous as white ones. Fur- 
ther, individuals which are both short-haired 



and dark should be 3 X 3 or 9 times as nu- 
merous as those which are not. The ex- 
pected proportions of the four classes of F 2 
offspring are accordingly 9:3:3:1, a propor- 
tion which is closely approximated in actual 
experience. The Mendelian theory of inde- 
pendent unit-characters accounts for this re- 
sult fully. No other hypothesis has as yet been 
suggested which can account for it. 

Suppose that each unit has a different mate- 
rial basis in the gamete. Let us represent the 
material basis of hair-length by a circle, that 
of hair-color by a square, then combinations 
and recombinations arise as shown in Fig. 30. 
The composition of the gametes furnished by 
the parents is shown in the first line of the 
figure; that of an F individual (or zygote), 
in the second line; that of the gametes 
formed by the F individual in the third line. 
L meets S and W meets D in fertilization to 
form an F x individual double and also hetero- 
zygous as regards hair length and hair color, 
but these units segregate again as the gametes 
of the Fj individuals are formed, and it is a mat- 
ter of chance whether or not they are associated 



as originally, L with W and S with D, or in a 
new relationship, L with D and S with W. 
Hence we expect the F individuals to form 
four kinds of gametes all equally numerous, - 
L W, S D, L D, and S W. By chance unions of 
these in pairs nine kinds of combinations be- 
come possible, and their chance frequencies will 
be as shown in Fig. 31. Four of these com- 
binations, including nine individuals, will show 
the two dominant characters, short and dark; 
two classes, including three individuals, will 
show one dominant and one recessive charac- 
ter, viz. dark and long; two more classes, in- 
cluding three individuals, will show the other 
dominant and the other recessive character, 
viz. short and white; and lastly, one class, in- 
cluding a single individual, will show the two 
recessive characters, long and white. The four 
apparent classes, or, as Johannsen calls them, 
phenotypes, will accordingly be as 9:3:3:1. 
This is called the normal Mendelian ratio for 
a dihybrid cross, that is, a cross involving 
two unit-character differences. 

One individual in each of these four classes 
will, if mated with an individual like itself, 



breed true, for it is homozygous, containing 
only like units. The double recessive class, 
long white, of course contains only homozygous 
individuals, but in each class which shows a 
dominant unit, heterozygous individuals out- 
number homozygous ones, as 2 : 1 or 8 : 1. Now 
the breeder who by means of crosses has pro- 
short Dark. Long Dark. Short White. Long White. 

1 S D. S D i L D. L D i S w - S w i L w. L w 
s s D. L D s L D. L w s s w - L w 

2 S D. S w 
_S D. L w 

933 1 

FIG. 31. Diagram showing the kinds and relative frequencies 
of the young to be expected in F 2 from the crossing of 
animals shown in Figs. 26 and 27. 

duced a new type of animal wishes, of course, 
to " fix " it, that is, to obtain it in a condi- 
tion which will breed true. He must, therefore, 
obtain homozygous individuals. If he is deal- 
ing with a combination which contains only 
recessive characters, this will be easy enough, 
for such combinations are invariably homozy- 
gous. His task will become increasingly diffi- 
cult the more dominant characters there are 
included in the combination which he desires 
to fix. 


FIG. 32. A long-haired, rough albino guinea-pig ; 
male, 2002. 


The most direct method for him to follow is 
to test by suitable matings the unit-character 
constitution of each individual which shows the 
desired combination of characters, and to reject 
all which are not homozygous. In this way a 
pure race may be built up from individuals 
proved to be pure. Such a method, however, 
though sure, is slow in cases where the desired 
combination includes two or more dominant 
unit-characters, for it involves the application 
of a breeding test to many dominant individ- 
uals, most of which must then be rejected. It 
is therefore often better in practice to breed 
from all individuals which show the desired 
combination, and eliminate from their offspring 
merely such individuals as do not show that 
combination. The race will thus be only grad- 
ually purified, but a large stock of it can be 
built up much more quickly. 

We may next discuss a cross in which three 
unit-character differences exist between the par- 
ents, instead of two. If guinea-pigs are crossed 
which differ simultaneously in three unit-char- 
acters, color, length, and direction of the hair, 
a still larger number of phenotypes is obtained 



in F 2 , namely, eight. A cross between a short- 
haired, dark, smooth guinea-pig (compare Fig. 
22) and one which was long-haired, white, and 
rough (Fig. 32) produced offspring in F which 
were short-haired, dark, and rough (compare 
Fig. 24), these being the three dominant char- 
acters, two derived from one parent, one from 
the other. The F 2 offspring were of eight dis- 
tinct types, two like the respective grandpar- 
ents, one like the F individuals (parents), and 
the other five new, shown in Fig. 33. They are 
short white rough, short white smooth, long 
white smooth, long dark rough, and long dark 
smooth. The largest of the eight apparent 
classes (phenotypes) was the one which mani- 
fested the three dominant characters, short, 
dark, and rough, which had been the exclusive 
F type; the smallest class was the one which 
manifested the three recessive characters, long, 
white, and smooth. Theoretically these two 
classes should be to each other as 27 : 1. Of 
the twenty-seven triple-dominants, twenty-six 
should be heterozygous. 

A comparison of this case with the one just 
previously described shows what an increas- 


FIG. 33. Five new combinations of unit-characters obtained in gen- 
eration F 2 , by crossing the animal shown in Fig. 32 with animals 
like that shown in Fig. 22. 


ingly difficult thing it is to fix types obtained by 
crossing, if the number of dominant characters 




FIG. 34. Diagram to show the gametic combinations and 
segregations involved in a cross between guinea-pigs differ- 
ing in three unit-characters. L stands for long hair, S for 
short hair, W for white hair, and D for dark hair; R for 
rough, and Sm for smooth coat. Compare Figs. 22 and 32. 

in the selected type increases. On the theory of 
unit-characters the gametic combinations and 
segregations involved in this cross are as shown 



in Fig. 34. The nature of the gametes formed 
by the parents crossed is shown in the first row ; 
the composition of the F x individuals, immedi- 
ately below. In the two lower rows are shown 
four different sorts of gametic splittings which 
may occur in the F x individuals, producing thus 
eight different kinds of gametes. If, in re- 
ality, the F! individuals form eight kinds of 
gametes, all equally numerous, and chance 
unions in pairs occur among them, there should 
be produced eight corresponding sorts of indi- 
viduals numerically as 27:9:9:9:3:3:3:1. 
In a total of 64 individuals there should be on 
the average one pure individual in each of the 
eight different classes. The class numerically 27 
in 64 manifests three dominant characters ; those 
which are numerically 9 in 64 manifest two domi- 
nant characters ; those which are numerically 3 
in 64 manifest one dominant character. Among 
each of these there will be on the average one 
pure individual, but the class which contains 1 
individual in 64 is a pure recessive, for it contains 
no dominant character. This combination, then, 
requires no fixation. It will breed true from the 





1903. " Mendel's Law of Heredity." Proc. Am. Acad. 
Arts and Sci., 38, pp. 535-548; also, Science, N. S., 18, 
pp. 396-406. 

1905. " Heredity of Coat-characters in Guinea-pigs and 
Rabbits." Carnegie Institution of Washington, Publica- 
tion No. 23, 78 pp., 6 pi. 

1905. "Inheritance of Digital Malformations in Man." 
Papers of the Peabody Museum, Harvard University, vol. 
3, No. 3, pp. 69-77, 5 pi. 



WE have noticed that when a black 
guinea-pig of pure race is mated 
with a white one, only black off- 
spring are produced; and that when rough- 
coated guinea-pigs are mated with smooth- 
coated ones, only rough-coated young are pro- 
duced; and that when short-haired guinea- 
pigs are mated with long-haired ones, only 
short-haired young are produced. The char- 
acter which in each case is seen in the young 
we call dominant, that which is unseen we 
call recessive. Thus black is dominant over 
white, rough coat over smooth coat, and short 
coat over long coat. 

A question which has given much concern to 
students of heredity is this, upon what does 



dominance depend? Why should black domi- 
nate white rather than the reverse f 

In poultry, indeed, the relations are often 
reversed, white dominating black. Why is 
this? Several attempted explanations have 
been made, but none of them is thoroughly 
satisfactory. The one which has found wid- 
est acceptance is this: In the dominant in- 
dividual something is present which is want- 
ing in the recessive. Thus, in the black guinea- 
pig there is present some ferment-like body 
or some ingredient of black which is wanting 
in the albino. Accordingly, the black guinea- 
pig forms pigment, a thing which the albino 
can do only feebly or not at all. The distinc- 
tive something of the black parent therefore 
dominates a corresponding nothing of the 
white parent. White fowls, on the other hand, 
are not albinos. They have pigmented eyes. 
Accordingly they do not lack the power to 
form pigment, owing to the absence of -some 
necessary ferment or pigment ingredient. 

White guinea-pigs occur which are in a way 
comparable with white fowls. They look ex- 
actly like albinos, except that their eyes are 
5 53 


black, whereas the eyes of the albino are pink. 
If such a black-eyed white guinea-pig is crossed 
with an albino of the sort shown in Fig. 15, 
the young produced will be black all over. 
Now this result shows that the black-eyed 
white animal possesses what is lacking in the 
albino as compared with the all-black animal. 
It would seem, therefore, that it lacks some- 
thing different from what the albino lacks, and 
that a cross of the two supplies both lacks, the 
albino supplying what is wanting in the black- 
eyed white, and vice versa. Accordingly, wholly 
black offspring result from the crossing of the 
two white races. 

But the case of white poultry is different 
from this, since white poultry lack nothing 
that is necessary to produce the complete 
black plumage. For when white fowls crossed 
with black ones produce white offspring, if 
these offspring are then bred with each other, 
they produce both white offspring and black 
ones in the ratio 3 to 1. White fowls, there- 
fore, are able to produce the black condition. 
This ability is in the white individual held in 
abeyance, it is not exercised. Why, we do 



not know. Some suppose it to be held in 
check by an additional unit-character, an in- 
hibiting factor, but we have no direct evidence 
that such a factor exists. All that we are 
warranted in saying at the present time is 
that black and white in poultry represent dif- 
ferent conditions of pigmentation, alternative 
to each other in heredity. In crosses of the 
two, white is ordinarily dominant over black, 
but in crosses between certain strains of white 
and black poultry this relationship is reversed, 
as Bateson has shown. 

In still other cases, a cross of white with 
black fowls produces offspring which resemble 
neither parent closely, but which are in reality 
intermediate. They are known as blue or 
Andalusian fowls. They manifest a dilute 
condition of black, such as one might obtain 
by mixing lampblack with flour; they are in 
reality a fine mosaic of black with white. 
Such a condition has thus far been obtained 
only from a cross of black fowls with a pecu- 
liar strain of impure sooty whites. This strain 
undoubtedly contains the mosaic pattern but 
without sufficient black pigment to make it 



plainly visible. A cross with a black race 
makes it visible. No one, however, has suc- 
ceeded in " fixing '' a blue race, that is, 
in obtaining a strain which would breed 

When two blue individuals are bred together 
they produce black, blue, and white offspring 
in the ratio 1:2:1. The blacks are homo- 
zygous, B B ; the whites also are homozygous, 
WW, but the blues are invariably heterozy- 
gous, B W. Blue accordingly in this case is 
called a heterozygous character, one which is 
due to the presence in one zygote of two unlike 
unit-characters, which invariably segregate from 
each other at the genesis of gametes, but which 
jointly produce a different appearance from 
what either produces by itself. If a strain of 
Andalusian fowls should ever be secured which 
would breed true, it would have to come about 
by the association of black with white in a 
non-segregating relationship, so that both 
would be transmitted in the same gamete. 
That is, one would have to secure in the same 
gamete with white enough black pigment to 
bring out the latent mosaic pattern, and fur- 



ther, one would have to secure a homozygous 
race of fowls which formed such gametes. 

Success would be most likely to attend the 
experiment if one selected always the sootiest 
whites obtained from blue parents, for blue 
results, as we have seen, from the association 
of more black with the white and in the pat- 
tern borne already by the white race. 

A much-debated case of inheritance which 
involves this principle of unfixable heterozy- 
gous characters occurs among fancy mice, in 
the variety known as yellow. A wonderful 
series of color varieties exists among mice 
kept as pets, equalling or perhaps surpassing 
that known in the case of any other mammal. 
All these varieties appear to be derivatives 
of the common house-mouse, with which they 
cross readily. All are capable of explanation 
as unit-character variations from the condition 
of the house-mouse. Among all these varie- 
ties yellow is most peculiar in its behavior. 
In crosses it is dominant over all others, yet 
is itself absolutely unfixable. 

If certain strains of yellow mice are crossed 
with black ones, the offspring produced are of 



two sorts equally numerous, yellow and black. 
From this result alone it is impossible to say 
which is the dominant character, but breeding 
tests of the offspring show that yellow is the 
dominant character. For the black offspring 
bred together produce only black offspring, 
but the yellows bred together produce both 
yellow offspring and black ones. The curious 
feature of the case is that when yellows are 
bred with each other no pure yellows, that is, 
homozygous ones, are obtained. Hundreds of 
yellow individuals have been tested, but the 
invariable result has been that they are found 
to be heterozygous ; that is, they transmit yel- 
low in half their gametes, but some other color 
in the remaining gametes it may be black, 
or it may be brown, or it may be gray. The 
black, brown, or gray animals obtained by 
mating yellow with yellow mice never produce 
yellow offspring if mated with each other. 
This shows that they are genuine recessives 
and do not contain the yellow character, which 
is dominant. 

Now ordinary heterozygous dominants, when 
mated with each other, produce three domi- 



nant individuals to one recessive. Accordingly 
we should expect yellow mice, if, as stated, 
they are invariably heterozygous, to produce 
three yellow offspring to one of a different 
color, but curiously enough they do not. They 
produce two yellows (instead of the expected 
three) to every one of a different color. About 
the ratio there can be no reasonable doubt. 
It has been determined with great accuracy 
by my pupil, Mr. C. C. Little, who finds that 
in a total of over twelve hundred young pro- 
duced by yellow parents almost exactly two- 
thirds are yellow. Instead of the regular 
Mendelian ratio 3 : 1, we have then in this case 
the peculiar ratio 2 : 1, and this requires ex- 
planation. The explanation of this ratio is to 
be found in the same circumstance as is the 
total absence of pure yellows. Pure yellow 
zygotes are indeed formed, but they perish for 
some unaccountable reason. For a yellow in- 
dividual forms gametes of two sorts with equal 
frequency, viz. yellow and non-yellow (let us 
say black). For, if yellow individuals are 
mated with black ones, half the offspring are 
black, half yellow, as already stated. 



If now yellow individuals are mated with 
each other we expect three sorts of young to 
be produced numerically, as 1:2:1, viz. 1 Y Y, 
2 Y B, and IBB. But since observation shows 
that only two combinations are formed which 
contain yellow to one not containing yellow, 
and since further all yellows which survive 
are found to be heterozygous (YB), it must 
be that the expected YY individual either is 
not produced or straightway perishes. As to 
which of these two contingencies happens we 
also have experimental evidence. Mr. Little 
finds that yellow mice when mated to black 
ones produce larger litters of young than when 
they are mated to yellow ones. The average- 
sized litter contains something like 5.5 young 
when the mate is a black animal, but only 4.7 
when it is a yellow animal. It is evident, then, 
that about one young one out of a litter per- 
ishes when both parents are yellow, and this 
undoubtedly is the missing yellow-yellow zy- 
gote. The yellows which are left are hetero- 
zygous yellow-black zygotes, and they are to 
those that perish as 2:1. They are also to 
the non-yellow zygotes as 2:1, the ratio ob- 



served also among the surviving young of 
yellow by yellow parents. 

This interpretation of the 2 : 1 ratio observed 
in this case is strongly supported by a similar 
case among plants, in which the evidence is 
even more complete. A so-called " golden ' 
variety of snapdragon, one in which the foli- 
age was yellow variegated with green, was 
found by the German botanist, Baur, to be 
unfixable, producing when self-pollinated fully 
green plants as well as golden ones, in the 
ratio 2 golden : 1 green. The green plants were 
found to breed true, that is, to be recessives, 
while the golden ones were invariably found 
to be heterozygous. Baur found, however, by 
germinating seeds of golden plants very care- 
fully, that there were produced in addition to 
green plants and golden ones a few feeble 
seedlings entirely yellow, not variegated with 
green, as the golden plants are. These, for 
lack of assimilating organs (green chlorophyl), 
straightway perished. Clearly they were the 
missing pure yellow zygotes. 

Some Mendelian characters, while not them- 
selves heterozygous and so unfixable, are never- 



theless produced only when two independently 
inherited factors are present together. A 
character of this sort does not itself conform 
with the simple Mendelian laws of inheritance, 
but its factors do. Herein lies the explanation 
of atavism or reversion, and the process by 
which reversionary characters may be fixed. 

Atavism or reversion to an ancestral con- 
dition is a phenomenon to which Darwin re- 
peatedly called attention. He realized that it 
is a phenomenon which general theories of 
heredity must account for. He supposed that 
the environment was chiefly responsible for the 
reappearance in a species of a lost ancestral 
condition, but that in certain cases the mere 
act of crossing may reawaken slumbering an- 
cestral traits. Thus he noticed that when 
rabbits of various sorts are turned loose in a 
warren together, they tend to revert to the 
gray-coated condition of wild rabbits. And 
when pigeons are crossed in captivity they 
frequently revert to the plumage condition of 
the wild rock pigeon, Columba livia. In plants, 
too, Darwin recognized that crossing is a fre- 
quent cause of reversion. The explanation 



which he gave was the best that the knowl- 
edge of his time afforded, but it leaves much 
to be desired. This lack, however, has been 
completely supplied by the Mendelian princi- 
ples. An illustration or two may now be 

When pure-bred black guinea-pigs are mated 
with red ones, only black offspring are as a 
rule obtained. The hairs of the offspring do 
indeed contain some red pigment, but the black 
pigment is so much darker that it largely 
obscures the red. In other words, black be- 
haves as an ordinary Mendelian dominant. In 
the next generation black and red segregate 
in ordinary Mendelian fashion, and the young 
produced are in the usual proportions, three 
black to one red, or 1 : 1 in back-crosses of the 
heterozygous black with red. All black races 
behave alike in crosses with the same red in- 
dividual, but among red animals individual 
differences exist. Some, instead of behaving 
like Mendelian recessives, produce in crosses 
with a black race a third apparently new con- 
dition, but in reality a very old one, the agouti 
type of coat found in all wild guinea-pigs, as 



well as in wild rats, mice, squirrels, and other 
rodents. In this type of coat reddish yellow 
pigment alone is found in a conspicuous band 
near the tip of each hair, while the rest of the 
hair bears black pigment. The result is a 
brownish or grayish ticked or grizzled coat, 
inconspicuous, and hence protective in many 
natural situations. 

Some red individuals produce the reversion 
in half of their young by black mates, some 
in all, and others, as we have seen, in none, 
this last condition being the commonest of the 
three. It is evident that the reversion is due 
to the introduction of a third factor, additional 
to simple red and simple black. It is evident 
further that this new third factor, which we 
will call A (agouti), has been introduced 
through the red parent, and that as regards 
this factor, A, some individuals are homozy- 
gous (AA) in character, others are heterozy- 
gous (transmit it in half their gametes only), 
while others lack it altogether. Further ob- 
servations show that it is independent in its 
inheritance of both black and red; it is in 
fact an independent Mendelian character, which 



can become visible only in the presence of both 
black and red, because it is a mosaic of those 
two pigments. If the F x agouti individuals 
are bred together they produce in the next 
generation (F 2 ) three sorts of young, viz. 
agouti, black, and red, which are numerically 
as 9:3:4. This evidently is a modification of 
the dihybrid Mendelian ratio 9:3:3:1, result- 
ing from the fact that the last two classes are 
superficially alike. They are red animals with 
and without the agouti factor respectively; but 
this agouti factor is invisible in the absence 
of black, so that both sorts of reds look alike. 
Together they number four in sixteen of the 
Fo offspring. 

Fig. 35 is intended to show how the inde- 
pendent factors behave in heredity. The black 
parent contributes the factor B, the red par- 
ent, R and A, so that the zygote, or nw indi- 
vidual, contains the three factors necessary to 
the production of agouti. When the new in- 
dividual forms gametes (sex-cells), these will 
l)c of four different kinds, for A is independ- 
ent of B and of R and may pass out with 
either one in the reduction division which sepa- 



Agouti Agouti Agouti Agouti (fixed) 
Black Fed Black 


FIG. 35. Diagram to show the gametic combination and re- 
combinations which occur in the production and fixation of 
an atavistic coat-character in guinea-pigs. 

Row 1 shows the character of the gametes formed by the 
parents crossed; row 2 shows the character of the FI agouti 
individuals resulting from the cross; row 3 shows the two 
different sorts of gametic splittings which may occur in the 
production of gametes by the FI agoutis, and how four differ- 
ent kinds of gametes result; row 4 shows how among such 
gametes four different kinds of unions may occur that will 
produce agouti young. The BA-BA combination, it will 
be understood, could result only from the union of a BA 
gamete with another gamete of like constitution. Below 
each of the four combinations is indicated the kinds of young 
which an animal of that sort would produce if mated with 
an animal like itself. The numerals show the expected 
relative frequencies of the four sorts of combinations. 



rates B from R. That division accordingly 
may occur either so as to form gametes B and 
R A respectively, or what is equally probable, 
so as to produce gametes B A and R. Obser- 
vation confirms this interpretation, for it is 
found that the reversionary agoutis do not 
breed true, but produce young of the three 
sorts, agouti, black, and red, as expected. We 
expect black individuals from unions of B with 
B, or of B with R; we expect red individuals 
from unions of R with R or with R A, and 
from unions of R A with R A ; we expect agoutis 
to be produced by any gametic union which 
brings together the three factors B, R, and A. 
There are six chances in sixteen for the oc- 
currence of such a union, when the rever- 
sionary agoutis are bred together. In fact, 
however, agoutis are produced much oftener. 
Approximately nine out of sixteen of the 
young have been found to be agoutis. The 
unexpected excess of agoutis in our experi- 
ments was fully explained when these second- 
generation agoutis were tested individually. 
It was then found that they are of four sorts 
as regards breeding capacity. The first sort 



produces the three kinds of young, agouti, black, 
and red; the second sort produces only agouti 
young and red young; the third sort produces 
only agouti young and black young. The fourth 
sort produces only agouti young, i. e. repre- 
sents the fully fixed agouti type, the completely 
recovered wild type. 

In the chart (Fig. 35) are indicated cer- 
tain gametic unions which would lead to the 
production of these four classes of agoutis. 
The probable frequencies of their occurrence 
on the basis of chance are 4:2:2:1. 

Experiment made it clear that R as an inde- 
pendent gametic factor is not necessary to the 
production of the agouti character, as was at 
first thought to be the case, but that any 
gametic union which includes both B and A 
will produce an agouti individual whether 
E is or is not present. Yet a microscopic 
examination of the agouti hair shows that 
red pigment is present in a distinct band 
near the hair-tip. As a matter of fact all 
black individuals, even when they breed true, 
probably form some red pigment along with 
the black, but its presence is overlooked when 



the more opaque black is distributed through- 
out the whole length of the hair. When, how- 
ever, black is excluded from the hair-tip, the 
red then becomes visible as the agouti mark- 
ing; elsewhere the hair appears black. Eed, 
then, we may assume, is always present with 
black in sufficient quantities to produce the 
agouti marking if the factor A is present (ab- 
sence of black from the hair-tip). This ex- 
plains why blacks never give the reversion in 
any sort of cross, but it is always brought 
about through the agency of the red parent. 
If a black individual contained the factor A, 
it would no longer be a black individual, but 
an agouti one. 

The existence of a third factor, A, in cases 
of reversion in coat-character among rodents 
was long overlooked merely because it does 
not represent a distinct pigment or set of pig- 
ments, but consists in a particular kind of 
pigment distribution on the individual hairs. 
The agouti hair is due to a definite cycle of 
activity of the hair follicle in forming its pig- 
ments, first black, then red, then black; the 
wholly black hair is due to a continuous process 
6 69 


of pigment formation without alternation in the 
character of the pigments produced. 

In rabbits as well as in guinea-pigs rever- 
sion to the original wild type, in this case 
gray, may be obtained by crossing a black ani- 
mal with a yellow one. In guinea-pigs the 
yellow (or red) animal which will yield this 
result cannot be distinguished in appearance 
from one which will not; but in rabbits the 
yellow animal which will give reversion has 
a white belly and tail, while the one which 
will not give reversion is not so distinguished. 

We now know what is implied in the fixa- 
tion of a heterozygous character obtained by 
crossing. When A and B are crossed we ob- 
tain a third condition, C. C is due either to 
the simple coexistence of A with B, or to the 
coexistence with them of a third factor intro- 
duced with one or the other. In either case 
fixation will consist in getting into the gamete 
all the factors necessary to the production of 
C. In the first supposed case the zygote is 
A-B and the resultant is equivalent to C. Fix- 
ation will consist in getting a zygote of the 
formula AB AB. In the second supposed 



case the zygote produced is either A-CB or 
AC-B; fixation will consist in obtaining a zy- 
gote ACB-ACB; every gamete formed will 
then contain the three factors A, C, and B. 



1909. "Mendel's Principles of Heredity," 393 pp., illus- 
trated. University Press, Cambridge; also G. P. Put- 
nam's Sons, N. Y. [Contains translation of Mendel's 
original papers.] 


1907. " Untersuchungen iiber die Erblichkeitsverhaltnisse 
einer nur in Bastardform lebensfahigen Sippe von An- 
tirrhinum majus." Ber. d. Deuisch Bot. Gesellsch., 25, 
p. 442. 

1907. "On a Case of Reversion Induced by Cross-Breed- 
ing and its Fixation." Science, N. S., 25, pp. 151-153. 

1907. "The Production and Fixation of New Breeds." 
Proc. Amer. Breeders' Ass'n, 3, pp. 34-41. 

CASTLE, W. E. and LITTLE, C. C. 

1910. "On a Modified Mendelian Ratio Among Yellow 
Mice. " Science, N. S. } 32, pp. 868-870. 


1908. "Sur quelques anomalies apparentes des pro- 
portions Mendelie"nnes." Arch. Zool. Exper. (4), Notes 
et Revue, p. vii. 


1906. "Inheritance in Poultry." Carnegie Institution 
of Washington, Publication No. 52, 104 pp., 17 pi. 

1909. "Inheritance of Characteristics in Domestic Fowl." 
Carnegie Institution of Washington, Publication No. 
121, 100 pp., 12 pi. 




OUR knowledge of Mendelian phenom- 
ena is most complete in the case of 
color-inheritance. We find that the 
flower-colors of plants and the coat-colors of 
mammals are alike complex, and that what 
seem at first sight simple results may really 
depend on several independent factors acting 
conjointly. By analysis of such complex cases 
we are able to gain some idea of what the 
probable course of evolution has been in the 
production of the color varieties found among 
cultivated plants and domesticated animals. 

Thus among rodents (mice, rabbits, guinea- 
pigs) the coat is grayish, consisting of black, 
brown, and yellow pigments mingled together 
on the same individual hair in a pattern of 
greater or less complexity. 



The simplest variation from this ancestral 
type of coloration is albinism, a wholly unpig" 
mented condition in which the eyes are pink. 
This is due to the loss of the capacity to form 
pigment. Albinism is recessive in crosses. We 
explain it by assuming that something neces- 
sary for color production is wanting in the 
albino, and call that something the color-factor 
C, without necessarily making any assumption 
as to its nature. Another common variation 
is the loss of the pattern-factor of the indi- 
vidual hair, the agouti or A factor. An ac- 
count of the discovery of this factor was given 
in the last chapter. In consequence of the loss 
of this factor the pigments become mingled 
together without order, and the result is a 
uniform black, the denser pigment hiding the 

A third variation is the loss of the capacity 
to form black pigment (factor B), only brown 
and yellow pigments being left. Thus arise 
brown and cinnamon varieties. Through these 
three independent loss-variations there arise 
eight different color-varieties as follows: 



Gray (or agouti) = C B Br A; Cinnamon = C Br A 

Black = C B Br; Brown = C Br 

Albino (1) = B Br A; Albino (3) = Br A 

Albino (2) = B Br; Albino (4) = Br 

Proof of the correctness of this interpretation 
may be obtained from crosses. Suppose the 
four kinds of albinos described be crossed with 
the same colored variety, brown; albino 1 will 
produce gray offspring, albino 2 will produce 
black ones, albino 3 will produce cinnamon 
ones, and albino 4 will produce brown ones. 
The cross with albino 1 brings together all the 
four factors entering into the production of 
gray, viz. C, B, Br, and A, hence the young are 
gray. The cross with albino 2 brings together 
the factors C, B, and Br only. The result is 
black. The cross with albino 3 brings together 
the factors C, Br, and A; result, a cinnamon 
animal. The cross with albino 4 brings to- 
gether no factors except C and Br; result, a 
brown animal. 

Thus far we have considered merely varia- 
tions which arise by loss of one or more of 
the three unit-characters, A, B, and C. We 
may now consider variations which arise 



by modification without loss of these same 

Yellow varieties owe their origin to a re- 
duction in the amount of black or brown pig- 
ment in the fur, and to a corresponding in- 
crease in the amount of yellow. In some 
yellow animals, such as the sooty yellow rab- 
bit, black and brown pigments are not wholly 
lacking in the fur, but are only greatly re- 
duced in amount. They always persist in 
the eye. In other yellow animals, mice for 
example, the black or brown pigments are 
wholly absent from the fur, and they may also 
be greatly reduced in amount in the eye, as 
in the variety known as pink-eyed yellow, but 
in no yellow animal, so far as I am aware, is 
the production of black and of brown pigments 
wholly suppressed. 

In any mammal which possesses yellow varie- 
ties we can produce by suitable crosses as many 
different varieties of yellows as there are of 
gray, black, cinnamon, and brown varieties 
combined. For example, in mice, yellow indi- 
viduals of which, as was shown in the last 
chapter, are invariably heterozygous and pro- 



duce some other variety than yellow, even when 
mated with yellows, we can recognize the fol- 
lowing varieties distinct in breeding capacity, 
though all looking very similar. 

1. Yellows which produce yellow young and gray 

2. Yellows which produce yellow young' and black 

3. Yellows which produce yellow young and 

4. Yellows which produce yellow young and brown 

Albino varieties occur which correspond with 
each of these yellow varieties, viz. (1) albinos 
which if crossed with brown will produce yel- 
low young and gray ones; (2) albinos which 
crossed with brown produce yellow young and 
black ones ; (3) albinos which crossed with brown 
produce yellow young and cinnamon ones ; and 
(4) albinos which crossed with brown produce 
yellow young and brown ones. Such albinos, 
of course, differ from the corresponding yellow 
varieties merely by the general color factor C, 
which the albino lacks. If this is added by a 
cross, they produce the same visible result as 



the corresponding yellow variety in the same 

In addition to the modification which pro- 
duces yellow varieties, we can recognize sev- 
eral other modified conditions of the unit- 
characters A, B, C, and Br, which modifica- 
tions produce whole series of color varieties. 
For a modified condition of a single unit- 
character is capable of producing as many 
new varieties as there are possible combina- 
tions of the modified character with other unit 

One who attends a poultry-show cannot fail 
to be impressed with the great number of color 
varieties among poultry. Let him first observe 
these among fowls of common size, and if he 
then visits the bantam section he will find them 
all duplicated in miniature among the bantams. 
If a new color variety is brought out, it is only 
a short time until it finds its place among the 
bantams as well as among fowls of common 
size. The dwarf size of the bantam is clearly 
due to a modified condition of one or more unit- 
characters capable of combinations with as many 
different kinds of coloration as occur among 



poultry. The various combinations are of 
course brought about by crossing, and two gen- 
erations suffice theoretically for securing them. 
In mice, if one possessed only the albino 
variety last described, the one which corre- 
sponds with the brown-eyed yellow variety, - 
he could easily produce within six months every 
one of the various color varieties which have 
been mentioned. All he would have to do 
would be to catch some wild mice and cross 
these with his albinos. The immediate off- 
spring produced by the cross might seem un- 
promising; they would either be gray, exactly 
like wild mice, or else yellow. But if our 
breeder possessed the faith to breed a second 
generation from these animals, he would be 
rewarded by seeing all the color varieties which 
I have described put in an appearance, viz. 
yellows with black eyes, and yellows with 
brown eyes, blacks, browns, cinnamons, and 
grays, and albinos corresponding in character 
with each colored variety except for the lack 
of the color-factor C. 

It may be of interest to consider how some 
additional color varieties of mice have arisen, 



for of all mammals bred in captivity the mouse 
is probably richest in color varieties. In one 
series of these the capacity to form black or 
brown pigments is greatly weakened, so that 
the coat is less heavily pigmented and the eye 
is almost wholly unpigmented, and looks pink, 
due to the red color of the blood in the eye. 
This series we may call the pink-eyed series. 
All the common color varieties occur in a pink- 
eyed as well as in a dark-eyed series. Thus 
there are pink-eyed grays, pink-eyed blacks, 
pink-eyed cinnamons, pink-eyed browns, and 
pink-eyed yellows, as well as albinos which 
transmit the pink-eyed condition in crosses. 

Given a single pink-eyed individual in any 
one of these varieties, all the others may be 
produced from it by suitable crosses. Thus 
a pink-eyed gray crossed with brown produces 
in F reversion to the condition of the wild 
house-mouse, but in F 2 (that is, among the 
grandchildren) occur eight varieties, four 
dark-eyed and four pink-eyed. Gray, black, 
cinnamon, and brown occur, both in dark-eyed 
and in pink-eyed individuals, the latter being 
also far lighter in color than the dark-eyed 



varieties. The pink-eyed condition is there- 
fore in mice a unit-character modification of 
the pigmentation, independent of any of the 
pigment factors previously mentioned, since 
it can be transferred by crosses from asso- 
ciation with one of these to association with 
another. It may also be transmitted equally 
well through colored and through albino in- 
dividuals, though it produces a visible effect 
only in colored individuals. 

Another unit-character modification of the 
pigmentation seen in mice produces a series 
of dilute or pale pigmented varieties, but dif- 
ferent in character from the pink-eyed series, 
since their eyes may be dark, not pink. The 
pale modification of gray is known to fan- 
ciers as " blue-gray, " that of black is known 
as " blue," and that of brown is known as 
" silver fawn." The pale quality is inter- 
changeable between black, brown, and yellow 
pigmentation, so that if one has a pale gray 
variety he may by crosses obtain also pale 
black, pale cinnamon, pale brown, and pale 
yellow varieties. Or if one starts with pale 
yellow, he may by crosses with a perfectly 



wild mouse obtain also pale gray, pale black, 
pale cinnamon, and pale brown varieties, all 
within two generations from the cross. 

Now the pale modification is distinct from 
the pink-eyed modification, and independent of 
it in transmission. Accordingly, it is possible 
to have the two modifications combined in the 
same race. Thus arises a series of pale pink- 
eyed grays, blacks, cinnamons, browns, and 
yellows. Since paleness is in crosses recessive 
to intense pigmentation, and pink eyes are re- 
cessive to dark ones, it follows that a variety 
which is both pale and pink-eyed will breed 
true to those characteristics without fixation. 

The lightest colored of the pale pink-eyed 
varieties develop very little pigment indeed, 
yet the modifications to which they are due 
are wholly different in nature from the albino 
variation, as a very simple experiment will 
show. Cross together an albino of variety (1), 
page 74, which is a snow-white animal with 
pink eyes, and a pale pink-eyed, brown ani- 
mal, whose coat is pale straw color, and whose 
eyes, like those of the albino, are pink. Although 
both parents are pink-eyed, and one develops no 



pigment whatever in its fur, while the other 
develops very little, nevertheless the offspring 
are as dark as the darkest wild mice, eyes, 
fur, and all. They look just like common house- 
mice. This result shows that the albino varia- 
tion is something very different in nature from 
the modifications found in the pink-eyed brown 
parent, since each parent contains those con- 
stituents of the wild gray coat which the other 
parent lacks. 

I can think of no more instructive labora- 
tory experiment illustrative of Mendelian in- 
heritance than to follow through two genera- 
tions the cross just described, and to analyze 
critically the results obtained. One who does 
this can never be sceptical about the value of 
crossing as an agency in the production of new 
varieties. For in the second generation from 
the cross he will obtain (1) ordinary gray, 
black, cinnamon, and brown varieties; (2) pale 
gray, black, cinnamon, and brown varieties; 
(3) pink-eyed gray, black, cinnamon, and brown 
varieties; (4) pink-eyed and pale gray, black, 
cinnamon, and brown varieties; and lastly, 
albinos, which, if he has the patience to test 



them one by one, will prove to be of sixteen 
different homozygous kinds, to say nothing of 
the much more numerous heterozygous sorts. 

No mention has thus far been made of spotted 
races, in which a unit-character modification has 
occurred which results in a distribution of pig- 
ment to part of the coat only, the remainder 
being unpigmented. Although this modifica- 
tion apparently regulates the distribution of 
pigment over the body, it is independent of the 
general color factor C, since it is transmitted 
through albinos, which by hypothesis lack C. 

Spotting is also independent of all the other 
unit-character modifications which have been 
described. Consequently we have in mice four 
different series of spotted varieties, the in- 
tense spotted, the dilute spotted, the intense 
pink-eyed spotted, and the dilute pink-eyed 
spotted. In each of these series are gray, 
black, cinnamon, brown, and yellow individuals, 
making a total of twenty spotted sorts, all of 
which may be obtained from crossing a single 
pair of properly selected parents, such, for 
example, as an albino and a wild house-mouse 
of the kind every barn contains. 



The color variations of guinea-pigs are simi- 
lar to those of mice; the same series of unit- 
character changes has produced them with one 
exception. The pink-eyed modification is want- 
ing in guinea-pigs. We are therefore limited 
here to the intense series, the pale series, the 
intense spotted series, and the pale spotted 
series. In each of these occur gray (or agouti) 
individuals, black ones, cinnamon ones, and 
brown ones. 

The parallelism between the color variations 
in guinea-pigs and in mice received an inter- 
esting demonstration in a particular case. The 
brown pigmented series in mice has been 
known for some time, but in guinea-pigs the 
brown variety is of comparatively recent origin, 
and the cinnamon variety was wholly unknown 
until some three years ago. After an analysis 
had been made in terms of unit-characters of 
the color varieties of the mouse, it became clear 
that if the color variation of guinea-pigs fol- 
lowed a like course, a then unknown variety 
of guinea-pig, cinnamon, should be capable of 
production by crossing an agouti animal with 
a brown one. In 1907 a statement of the sci- 



entific expectation in the case was published, 
and a few months later I had the satisfaction 
of announcing its fulfillment in the second gen- 
eration (F 2 ) from the cross in question. 

The experiment progressed as follows : The 
parents were an agouti and a black, their F x 
offspring were agoutis in character ; but the F 2 
offspring were of four sorts, agouti, black, 
cinnamon, and brown. The cross thus produced 
two varieties new to the experiment, viz. black 
and cinnamon, the latter being a variety at 
that time new among guinea-pigs. 

The subsequent behavior, too, of the newly 
produced cinnamon variety is in harmony 
with expectation based on Mendelian prin- 
ciples. The cinnamon variety has not pro- 
duced agouti or black individuals, which from 
the formulae it will be seen it may not be ex- 
pected to produce, since it lacks the factor B. 
But it has in some cases produced brown in- 
dividuals, as it clearly could in case both par- 
ents to a mating were heterozygous (single) 
in factor A. 

On the whole the evidence seems very clear 
that the numerous color varieties of animals 
7 85 


kept in captivity arise chiefly from loss or 
modification of Mendelian unit-characters. 
Loss of a unit-character might easily come 
about by an irregular cell-division in which 
the material basis of a character failed to 
split, as normally. The consequence would be 
that the character in question would be trans- 
mitted by one only of the two cell-products 
produced. The cell lacking a character might 
be the starting-point of a race lacking the char- 
acter, as of a black race, derived from a gray 
one. On the other hand a modified condition 
of a unit-character might possibly result from 
unequal division of the material basis of a 
character, so that one of the cell-products 
would transmit the character in weakened in- 
tensity, the other in increased intensity. 



1907. "Color Varieties of the Rabbit and of Other Rod- 
ents: Their Origin and Inheritance." Science, N. S., 
26, pp. 287-291. 

1908. "A New Color Variety of the Guinea-pig." Science, 
N. S., 28, pp. 250-252. 

1909. "Studies of Inheritance in Rabbits." Carnegie 
Institution of Washington, Publication No. 114, 70 pp., 
4 pi. 




IN the last chapter we discussed the color 
variations of mammals, and we concluded 
that these result largely from the loss or 
modification of some half-dozen independent 
Mendelian unit-characters. As to the material 
basis of these unit-characters some interesting 
evidence has recently been collected by Riddle. 
Melanin pigment has been for some time known 
to be formed by oxidation. A variety of or- 
ganic compounds may undergo oxidation into 
melanin pigments ranging in intensity from 
light yellow to black; the greater the oxida- 
tion, the darker the product. But it is not 
certain, as assumed by Riddle, that the chemi- 
cal method of oxidation is the same in all cases 
or that the substance to be oxidized is the same. 
The results obtained from breeding experiments 



show that the capacity to form pigment of all 
sorts may be lost by a single variation, which 
we have called loss of the color factor, C. We 
do not know whether it consists in the loss of 
a substance capable of oxidation, or of the 
power to take some indispensable first step in 
the process of oxidation, perhaps due to loss 
of an enzyme; but we do know that when this 
particular variation has occurred, the power 
to produce other than albino individuals can- 
not be recovered by any known means except 
a cross with colored animals. We know also 
that the capacity to form specific kinds of pig- 
ment (yellow, brown, or black) is independent 
of the general color-factor, C, for albinos may 
transmit those specific powers without them- 
selves being able to form any kind of pigment 
at all, i. e. without possessing C. Any animal 
which forms pigment of one of the higher 
grades has the capacity apparently to form 
pigment also of the lower grades. Thus a 
black animal can form also brown and yellow 
pigment granules. Brown (chocolate) animals, 
however, lack the capacity to form black pig- 
ment. The oxidation, it would seem, can in 


this case be carried no further than the brown 
stage, because of the lack of some oxidizing 
agency necessary to the last stage in pigment 
production. The production of yellow is prob- 
ably a first or early step in the oxidation 
process preliminary to the production of brown 
or black, yet all yellow animals, so far as 
known, are able to take the further steps ; they 
retain the capacity to form either brown or 
black pigment to some extent, if only in the 

The variations thus far described are what 
De Vries has called retrogressive, i. e. due to 
loss or modification. A much rarer sort of 
variation has been called by De Vries progres- 
sive, i. e. due to gain, acquisition of some char- 
acter not before possessed by the race. I can 
call to mind very few cases which certainly 
fall in this category. One which it would seem 
must belong here is the rough or resetted con- 
dition of the hair in guinea-pigs, a variation 
similar in nature to the reversed plumage of 
birds, seen, for example, in the Jacobin pigeon. 
The rough coat of guinea-pigs is surely not 
an ancestral condition, yet it behaves as a 



dominant character in crosses. It can scarcely 
be explained by loss; the only alternative is 
to consider it an acquisition, unless we choose 
to consider it a modification of the normal 

Aside from the sorts of variations already 
discussed, which consisted either in the loss or 
modification of existing unit-characters or in 
the gain of new ones, we must also recognize, 
as a cause of permanent and heritable varia- 
tion, changes in the potency of unit-characters, 
i. e. their tendency to dominate in crosses. 

When a gamete containing a particular unit- 
character unites with a gamete not containing 
it, the zygote formed will ordinarily show the 
character in question fully developed. This re- 
sult following Mendel's terminology we call 
dominance. But dominance is frequently im- 
perfect and may even be reversed. The zygote 
in which a character is doubly represented fre- 
quently develops the character more fully than 
the zygote in which it is represented but once. 
If a black guinea-pig is crossed with a yellow 
one the offspring are black, but oftentimes of 
a slightly yellowish shade. Likewise if black 



is crossed with brown, the crossbreds are apt 
to develop in their coats more brown pigment 
granules than do homozygous or pure blacks. 
Nevertheless, we have no reason to question the 
entire purity of the gametes, both dominant 
arid recessive, formed by such cross-bred black 
animals. It is the dominance, not the segrega- 
tion, which is imperfect. 

In other cases still the dominance may be 
entirely reversed in character, owing to varia- 
tion in the potency of a unit-character. Thus 
in most rodents the gray or agouti pattern- 
factor of the hair, A, is dominant. A cross of 
black with homozygous gray, in rats, mice, or 
rabbits, produces only gray offspring, which 
in F 2 produce three grays to one black. But 
the so-called black rat, Mus rattus, a species 
distinct from the one which has given rise to 
the varieties kept in captivity, behaves in a 
different way, as shown by Morgan ('09). 
When crossed with its gray variety, the roof 
rat, Mus alexandrinus , it produces only black 
offspring, and in F 2 , three blacks to one gray. 
If we suppose the gray coat in this case to 
be due to the same factor as in other rodents, 



we must assign to it a different potency, or 
power of dominance, so that it produces a 
visible effect only when doubly represented in 
the zygote. 

In guinea-pigs, rabbits, and mice we have 
seen that the presence together in the same 
zygote of two factors, A and B, in any com- 
bination whatever, produces the gray or agouti 
coat. The two factors are A, the agouti or 
gray marking of the hair, and B, black pig- 
ment in the fur. If A is lacking, the coat is 
black; if B is lacking, it is brown, cinnamon, 
or yellow. If both are lacking, it is either 
brown or yellow. But if both are present, the 
wild or agouti type is produced. So far as 
the production of the agouti coat is concerned, 
it makes no difference whether either factor is 
singly or doubly represented in the zygote. 
Each factor has potency enough to produce the 
full effect either in a single or in a double 
dose. Accordingly, as we noticed in an earlier 
chapter, we can distinguish by their breeding 
capacity, though not by their looks, four types 
of agouti guinea-pigs or gray rabbits, viz.: 


1. A A B B, which breeds true, since it forms game- 

tes all AB; 

2. ABB, which produces agouti young and black 

ones in the ratio 3:1, since it forms 
gametes A B and B ; 

3. A A B, which produces agouti young and yellow 

ones in the ratio 3:1, since it forms 
gametes A B and A; 

4. A B, which produces agouti, black, and yel- 

low young in the ratio 9 : 3 : 4. For 
the gametes formed by this sort are of 
four kinds, A B, A, B, and neither 
A nor B. 

Now in rats we have no evidence that the 
factor B has ever been lost, a matter to which 
we shall presently return ; but the agouti factor 
is apparently frequently wanting in ordinary 
rats, which are then black. For ordinary rats, 
then, the known combinations of A and B seem 
to be three, viz. : 

A A B B = the pure gray (wild type) ; 

ABB = heterozygous gray, which produces off- 
spring 3 gray : 1 black. This type is ob- 
tained by crossing black with wild gray; 

B B = pure black. 

Now in Mus rattus, as we have seen, the 
middle or heterozygous type is black, not gray 



in appearance, but it produces both the gray 
and the black types. So the same gametic 
formulae will account for both sets of facts, 
if we suppose merely that the potency of A is 
different in the two cases. In ordinary rats 
(Mus norvegicus) A produces the gray coat in 
a single dose; but in Mus rattus its potency 
is less, two doses are required to produce the 
gray coat. I am unable to frame any hypothe- 
sis other than this which will account for the 
reversal of dominance in one case as compared 
with the other. 

Yellow color in mammals affords another 
illustration of this same thing, reversal of 
dominance. Black and brown are in most 
mammals dominant over yellow in crosses, but 
in mice the reverse is true. The differential 
factor between black and yellow, if it is the 
same in mice as in other rodents, must be in 
one case potent enough to show itself if singly 
represented in the zygote, whereas in the other 
case it produces no visible effect unless doubly 
represented in the zygote. Yellow certainly 
seems to be a retrogressive variation from 
gray, black, or brown. The pigment granules 


remain in a lower oxidation stage in yellow than 
in black or brown. We suppose that in the 
yellow animal something is wanting which 
makes that further oxidation possible. This 
hypothesis would fully account for the observed 
recessive nature of yellow in the case of all 
mammals except mice. But here the capacity 
to form black or brown pigment is regularly 
present in the yellow individual but is held in 
check. We may suppose, therefore, that the 
differential factor, that which converts yellow 
into brown or black, must in this case be doubly 
represented in the zygote in order to produce 
brown or black fur, whereas in most mammals 
a single dose is effective. Accordingly, if the 
unmodified black or brown factor is represented 
only once in the zygote, and the yellow modi- 
fication is represented once, the latter will 
show, since the former is singly ineffective. 
The animal accordingly is a heterozygous yel- 
low, capable of producing also black or brown 
offspring. But mice are peculiar in that they 
cannot exist in the doubly deficient condition 
of a pure yellow zygote, consequently all 
yellow mice are heterozygous dominants, 



whereas other yellow mammals are homozy- 
gous recessives. 

In connection with this same case may pos- 
sibly be found the explanation of the complete 
absence of the yellow variation in rats. In 
nearly all mammals kept in captivity yellow as 
well as black varieties occur; this is true of 
horses, cattle, swine, dogs, cats, rabbits, guinea- 
pigs, and mice. In rats, however, a yellow 
variety is unknown. We know that rats are 
able to form yellow pigment, for all wild rats 
do form yellow pigment in their agouti fur, yet 
singularly enough no all-yellow rat has ever 
been observed, so far as we have any record, 
either wild or in captivity. A rat of this sort 
would command a high price at the hands of 
any fancier. Suppose the variation did occur 
in a single gamete. If, as in most mammals, 
it behaved as a recessive in crosses, it would 
not become visible, and might be carried along 
for untold generations without ever becoming 
visible unless two yellow gametes met. But if, 
as in mice, the yellow-yellow combination when 
formed quickly perished, then the character 
might never become visible. So the yellow 


variation may have occurred many times in 
rats, as it has in so many other mammals, but 
failed to become visible simply because it has 
the same potency as in most mammals, but is 
subject to the same physiological limitations 
as in mice, so that it cannot exist in a homo- 
zygous state. In that case the only evidence 
of its existence in a race would lie in a 
slightly diminished fecundity under inbreeding, 
as is found to be the case in yellow mice. 

Such sharply contrasted variations in the 
potency of characters as we have been discuss- 
ing are evidently of prime importance in evo- 
lution, making all the difference between a 
dominant and a recessive condition of a char- 
acter, or between the occurrence and the per- 
manent suppression of a particular variation. 
The character which is potent enough to show 
itself in a single dose will behave as a domi- 
nant character in crosses. We might call it 
ID/ i potent. That which must be present in a 
double dose to produce a visible result will 
behave as a recessive character in crosses. We 
might call it semi-potent. It is not impossible 
that the same character may as regards domi- 



nance behave in different ways under different 
circumstances, at one time dominating com- 
pletely, at another only feebly, and at other 
times not at all. 

Undoubtedly the chief condition affecting 
dominance is the nature of the gamete with 
which a union is made in fertilization. In 1905 
(Carnegie Inst. Publ. No. 23) I described a 
case in which a particular guinea-pig (male 
2002, shown in Fig. 32) having a rough or 
resetted coat gave a varying result in crosses. 
In crosses with most smooth animals his rough 
character dominated completely (see Fig. 24, 
which shows a son of the male 2002 by a smooth 
mother), but with one particular smooth ani- 
mal the dominance was very imperfect in all 
the young (Fig. 36), while with a second it 
was imperfect in half the young. The conclu- 
sion was drawn that gametes vary in potency, 
and that parents, too, differ as regards the 
potency of the gametes which they produce, 
some individuals producing gametes all of which 
are relatively potent, others producing gametes 
only half of which are potent, while still others 
produce gametes none of which are potent. 



Relative potency would, therefore, seem to be 
a character inherited in Mendelian fashion. 1 

Observations of Coutagne on silk-moths may 
be cited in support of this idea. Coutagne 
made crosses between races of silk-moths dif- 
fering in cocoon color, viz. between a race which 
spun yellow cocoons and another one which 
spun white cocoons. He found that some of 
the F! offspring spun yellow cocoons, others 
white ones. The F yellow cocooned animals 
when bred together produced F 2 progeny which 
spun some yellow, others white cocoons, the 
two sorts being as 3:1. In other words, yel- 
low in such cases behaved consistently as a 
dominant character. And the white-cocooned 
F 5 moths produced in F 2 cocoons of both colors, 
but in this case the white cocoons were to the 
yellow ones as 3:1. In other words, when 
yellow behaved as a dominant in F x it behaved 
as a dominant also in F 2 ; and the same was 
true of white. Each retained throughout the 
two generations the relative potency with which 

1 It is of course possible to interpret such a case as due to the 
separate inheritance of a factor which inhibits the development 
of the character, but it is doubtful whether this line of explana- 
tion can be successfully applied to cases presently to be described. 


it started. C. B. Davenport has also produced 
much evidence favoring the idea of varying 
potency of characters in recent papers based 
on his extensive studies on poultry. 

The case which I described in 1905 was one 
in which unusual potency seemed to inhere in 
the gametes of a recessive individual, one 
which apparently did not possess the character 
whose dominance was affected. But there occur 
also cases in which the varying gametic potency 
is associated directly with the character af- 
fected. One such I was able to describe in 
1906, that of an extra toe in guinea-pigs. It 
was found while building up a polydactylous 
race by selection and crossing it with other 
races that individuals varied in the potency 
which the character had in their gametes. In 
general the better developed the character was 
in an individual the more strongly was it trans- 
mitted, i. e. the larger was the proportion of 
polydactylous individuals produced in crosses. 
In no case, however, was this a recognizable 
Mendelian proportion, though both dominance 
and segregation seemed . to be taking place. 
Variation in potency was, however, unmistak- 


FIG. 36. An imperfectly rough guinea-pig. Produced by mating 
the guinea-pig, shown in Fig. 32, with a particular smooth animal; 
female, 2005. 

FIG. 37. A silvered guinea-pig. One in whose coat occur white 
hairs interspersed with pigmented ones. The amount of the silver- 
ing has been greatly increased by selection. 

FIG. 38. A. Front feet of an ordinary guinea-pig. B. Its hind feet. 
D. Hind feet of a race four-toed on all the feet. C. Ordinary con- 
dition of the hind feet of young obtained by crossing B with D. 

FIG. 39. Diagram showing variation in the color-pattern of hooded 


able and was transmitted from generation to 
generation. 1 See Fig. 36. 

It is an important question whether potency 
is a property of the unit-character or of the 
gamete, i. e. whether it affects all the charac- 
ters transmitted by a gamete or only a par- 
ticular one. Practical breeders as a rule favor 
the idea of garnetic rather than of unit-character 
potency, but this is probably due to a failure 
to discriminate between the two. They desig- 
nate as " prepotent " an individual supposed 
to impress all its characters upon the offspring, 
but it is very doubtful whether such individuals 
exist. It is easy to mistake for an animal 
potent in all respects one which is potent in 
one or two important respects only, especially 
if the observer is unaware, as every one has 
been until quite recently, that one character is 
independent of another in transmission. 

Conditions other than the character of the 
gametes themselves may determine the extent 

1 An alternative explanation is possible, viz. that the develop- 
ment of the fourth toe depends upon the inheritance of several 
independent factors, and that the more of these there are present, 
the better will the structure be developed. The correctness of 
such an interpretation must be tested by further investigations. 
8 101 


to which a character develops in the zygote, 
i. e. the completeness or incompleteness of its 
dominance in a particular case. For example, 
in salamanders, which apparently, like mam- 
mals, form skin-pigments of different sorts, 
such as yellow, brown, and black, Tornier has 
found that by feeding one may control the 
proportions in which chromatophores of the 
several sorts are formed in the skin. Abundant 
feeding causes preponderance of pigment of 
one sort, scanty feeding causes preponderance 
of pigment of another sort. Here external 
conditions determine the degree of development 
of characters. In other cases internal condi- 
tions may exercise a controlling influence. Thus 
in cattle the capacity to develop horns is a 
semi-potent unit-character, behaving as a re- 
cessive in crosses, heterozygotes developing 
only " scurs," that is, mere thickenings of the 
skin, or else no trace of horns at all. In sheep, 
moreover, horns are more strongly developed 
in males than in females, the presence of the 
male sex-gland in the body, or rather probably 
some substance given off into the blood from 
the sex-gland, favoring growth of the horns. 



In merino sheep the male has well-developed 
horns but the female is hornless ; yet if the male 
is castrated early in life no horns are formed. 

When a breed of sheep horned in both sexes, 
such as the Dorset, is crossed with one horn- 
less in both sexes, such as the Shropshire, horns 
are borne by the male but not by the female 
offspring. Both sexes, however, are heterozy- 
gous in horns, as is shown by their breeding 
capacity. For in F 2 occur both horned and 
hornless individuals in both sexes. The horn- 
less males and the horned females prove to 
be homozygous, but the horned males and the 
hornless females may be either heterozygous 
or homozygous. Accordingly the character, 
horns, behaves consistently as a dominant char- 
acter in one sex, but as a recessive in the 
other. Further, the presence of the male sex- 
gland in the heterozygote raises the potency of 
the character, horns, from semi-potent to uni- 
potent, as the result of castration shows. 

It is impossible to be certain that in a horn- 
less race the character horns has been wholly 
lost. It may merely have fallen so low in 
potency that under ordinary conditions it pro- 



duces no visible structures. The occasional 
occurrence of an imperfectly horned animal as 
a sport within a hornless race need not, then, 
occasion surprise. It would be a variation of 
the same sort as the extra toe in guinea-pigs 
(see Fig. 38), which, from a single sport, was 
built up by selection into a well-established race 
within a very few generations. This character, 
seemingly lost from the germ-plasm for an in- 
definite period, had perhaps merely fallen so low 
in potency that it no longer produced the fourth 
toe on the hind foot, though this was still pres- 
ent on the front foot. In the variant observed, 
the first polydactylus guinea-pig of my stock, 
the toe was imperfectly developed on one hind 
foot, doubtless as the result of an unusually 
potent condition of the character in one of the 
gametes which produced the individual. This 
manifestation of the character, though feeble, 
was sufficient to afford a guide for selection of 
those individuals which formed the most potent 
gametes, and so a polydactylous race was 
formed by selection and inbreeding. 

Great as has been the contribution of Men- 
delian principles to our knowledge of heredity, 



they do not reduce the whole art of breeding 
to the production of new combinations of unit 
characters through crossing. Selection is re- 
quired also, not merely among different combi- 
nations of unit-characters, but also among in- 
dividuals representing the same combinations 
selection is required of those possessing the 
desired characters in greatest potency. The 
further role of selection in evolution we shall 
need to consider in a subsequent chapter. 


CASTLE, W. E., and LITTLE, C. C. 

1909. "The Peculiar Inheritance of Pink Eyes Among 
Colored Mice." Science, N. S., 30, pp. 312-314. 


1902. "Recherches expe*rimentales sur 1'he're'dite' chez lea 

vers-a-soie." Bull Sci., 37, pp. 1-194, 9 pi. 

1909. "Breeding Experiments with Rats." American 

Naturalist, 43, pp. 182-185. 

1901-03. "Die Mutationstheorie." Leipzig, Veit and Co. 

See also the Bibliographies to Chapters III., IV., and V. 



IF, as suggested in the last chapter, the po- 
tency of a character in crosses may be 
modified by selection, why may not the 
character itself be modified by selection, or are 
not the two things perhaps identical, viz. modi- 
fication of the potency of a character and modi- 
fication of the character itself? Darwin firmly 
believed that the characters of organisms can 
be modified by selection, and he made this the 
foundation stone of his theory of evolution. 
De Vries and Johannsen, however, have taught 
us a different doctrine, maintaining that selec- 
tion is able to affect characters in superficial 
and transitory ways only, that the slight varia- 
tions in characters which we see everywhere 
among organisms have no evolutionary signifi- 
cance or permanent value ; that they come and 



go like the wavelets on the ocean beach, but 
have no more relation to evolution than the 
waves have to the tides. The brilliancy of the 
Mutation theory of De Vries, coupled with his 
great service to biology in rediscovering the 
Mendelian laws, has somewhat dazzled our eyes 
and led us, I think, to accept too readily his 
views concerning the efficacy of selection also. 
Ten years' continuous work in selection con- 
vinces me that much can be accomplished by 
this means quite apart from the process of mu- 
tation. The work of De Vries himself argues 
strongly in favor of this idea. To be sure, his 
interpretation of it is adverse to selection, and 
has seemed to most of us at times overwhelm- 
ingly convincing; but from his interpretation 
we may fairly appeal to the record of the work 
itself, and with this compare the record of our 
own work. 

One of the most extensive selection experi- 
ments conducted by De Vries was made on the 
common buttercup, Ranunculus bulbosus, which 
occurs as a weed in pastures and meadows in 
this country as well as in Europe. It has, as 
is known, regular 5-petaled flowers. An ex- 



animation of 717 flowers in the field made by 
De Vries in 1887 showed the rather frequent 
occurrence of 6 and 7 petaled flowers also, the 
average number of petals in the entire collec- 
tion being 5.13. De Vries set himself the task 
to see if the proportion of many petaled flowers 
could be increased or the number of petals to 
a flower be further increased. In both these 
respects he succeeded surprisingly well. As a 
result of five successive selections the average 
number of petals was raised from 5.6 to 8.6, 
the upper limit of variation from 8 to 31, and 
the mode (or commonest condition) from 5 to 9. 
Singularly enough De Vries concludes, in ac- 
cordance with general ideas which he had 
adopted, that selection had in this case done 
practically all that it could accomplish, that 
further selection, while it might advance the 
average somewhat farther, would have no per- 
manent effect in modifying the type. This be- 
lief seems to have rested on considerations 
such as these. De Vries had found, as had 
others, that variations which are heritable have 
their origin in the germ-cells only. He recog- 
nized that the tendency to produce double 



flowers in the buttercup is a heritable varia- 
tion and supposed it to be a unit-character, 
and so to conform with Mendel's law.. 

Now, if the tendency to produce double 
flowers were a simple Mendelian character it 
could exist in only three conditions, that of 
a recessive, that of a homozygous dominant, 
or that of a heterozygous dominant. But re- 
cessives and homozygous dominants are pure, 
that is, they form only one type of gamete, 
and selection therefore from among their pro- 
geny could produce no new type. As regards 
the heterozygous dominant type, this would 
itself be unfixable, and selection could accom- 
plish nothing permanent except by isolating a 
homozygous type. But such types should all 
be in evidence within two generations; there- 
fore, if a completely and permanently double 
type had not been discovered within the five 
generations covered by the experiment, such a 
type was not to be expected at all from the 
material in hand, unless either a wholly new 
unit-character were introduced or an existing 
one were profoundly modified. De Vries con- 
siders changes of both these sorts possible. 



He calls them mutations, and regards them as 
the sole means of evolutionary progress. But 
it is a peculiarity of his mutation theory that 
it regards only large changes in unit-characters 
as having any permanency, namely, such 
changes as mean a practical making over of 
the character. To borrow a figure from Bate- 
son, just as the gas carbon monoxide, C 0, 
may change into a very different gas, carbon 
dioxide, C 2 , by taking up a single atom 
of oxygen, but can make no less extensive 
change, since oxygen atoms do not split; so, 
according to De Vries, a unit-character may 
not change unless it changes profoundly. Vari- 
ous circumstances may modify the degree of 
its expression, but these are without perma- 
nent effect, since the character itself remains 

But there are both a priori and experimental 
grounds for questioning the correctness of 
De Vries ' conclusions. It is known that the 
chemical compounds within the germ-cells are 
not so simple in composition as C and C 2 . 
They are very complex substances, made up, 
it is thought, of very many atoms, often hun- 



dreds in a single molecule. If so, it is quite 
possible that an atom or two might be trans- 
posed in position within the molecule without 
wholly altering its chemical nature, and that 
thus slight changes in the germ-plasm might 
result, which, however, would be as permanent 
as more profound changes. 

The argument of De Vries against any per- 
manent effect of selection in modifying unit- 
characters has been greatly strengthened by 
the subsequent work of Johannsen and Jennings. 
Johannsen has found that if one selects from a 
handful of ordinary beans the largest seeds and 
the smallest seeds, and plants these separately, 
the former will produce beans of larger average 
size than the latter. Selection here has effect. 

But if the selection is made, not from a 
general field crop of beans, but from those 
beans borne on one and the same homozygous 
mother plant, then the progeny of the selected 
large seed will be no larger than that of the 
selected small seed. Selection here is without 

The different result in the two cases may 
be explained, according to Johannsen, on 



FIG. 40. Diagram showing the variations in size of eight different races of 
paramecium. Each horizontal row represents a race derived from a 
single parent individual. The individual showing the mean size in 
each race is indicated by a cross placed above it. The mean of the 
entire lot is shown at X X. The numbers show the measure- 
ments in microns. (After Jennings.) 



the principle of the " pure line." The pro- 
geny of a single self-fertilized homozygous 
bean plant constitute a pure line. They are 
all alike, so far as the hereditary transmission 
of size is concerned, for they are all derived 
from like gametes. The differences in size 
which occur among them are due to differences 
in nutrition, not to germinal differences, and 
they are not transmitted. But in a mixed 
population of beans, such as is represented by 
a field crop, differences of size occur which 
are due to heredity as well as those which are 
due to the environment. In the case of the 
former, selection naturally has effect; in the 
case of the latter, it does not. 

Jennings has obtained similar results in his 
studies of paramecium, a one-celled animal 
which multiplies asexually by dividing into two 
similar parts. It lives in stagnant water and 
may be reared in great numbers in a hay-infu- 
sion, for it multiplies with great rapidity, divid- 
ing two or three times within twenty-four hours. 
The variations in size which occur in parame- 
cium are shown in Fig. 40. 

When from an ordinary culture of parame- 


cium Jennings selects the largest and the small- 
est individuals respectively, he finds that the 
descendants of the one lot will be of larger size 
than the other. This looks like an effect of selec- 
tion upon racial size. But if selection is made not 
within a mixed population but among the de- 
scendants of a single individual, it is found 
that the descendants of large individuals are 
of no greater average size than those of small 

The explanation of this fact is to be 
found in the existence of what Johannsen 
has called pure lines. Jennings has been able 
to isolate eight distinct pure lines of parame- 
cium differing in average size, as shown in 
Fig. 40. The range of variation in size within 
one of these races is great, but if one selects 
extremely large or extremely small individuals 
within the same pure line, i. e. among the 
asexually produced descendants of the same 
animal, no change in the average size of the 
race is brought about. 

A very different result is obtained, however, 
if one mixes together several pure lines and then 
selects from the mixed race on the basis of size. 



The larger animals then produce larger average 
offspring and vice versa. An examination of 
Fig. 40 will show why. Animals of the same ab- 
solute size are there placed in the same vertical 
row. If, now, one selects from the mixed popu- 
lation only the largest individuals, he will nat- 
urally secure representatives of only two or 
three pure lines, viz. of those lines which are 
characterized by the largest average size, and 
which, therefore, will produce large average 
offspring. If on the other hand he selects 
extremely small individuals, he will secure 
representatives of only the smallest races, 
which naturally will produce small offspring, 
so that selection seems to be effective in modi- 
fying racial size, but in reality it does this by 
sorting out the elementary constituents of the 

It is impossible to deny the soundness of 
the reasoning of Johannsen and Jennings. It 
is perfectly clear that the effects of selection 
should be more immediate and much greater 
in the case of a mixed race than in that of a 
pure line, but is it certain, as assumed by 
them, that selection is wholly without effect in 



the case of a pure line? We know the effects 
should be less, but are they nil? Concerning 
this matter we are perhaps justified in await- 
ing further evidence. For in the case of beans 
and of paramecium alike size is subject to very 
great variation through the influence of nutri- 
tion. Variations due to this cause are natu- 
rally not inherited, since the germ-cells are not 
affected by them, but only the body. But is 
it not possible that along with the striking size 
differences due to nutrition there may occur 
also slight size differences due to germinal 
variation within the pure line, that is owing 
to variations in the potency of the same unit- 
character or combination of unit-characters? 
To be sure, Johannsen and Jennings have not 
observed these, but this does not prove their 
non-existence. Others may yet be able to do 
so; indeed one case is already on record in 
which such observations have been made in the 
case of a small crustacean (or water-flea), 

Daphnia is a small transparent animal, about 
the size of a pin-head, which occurs in enor- 
mous numbers in fresh-water lakes and pools, 



forming a large part of the food supply of 
fresh-water fishes. It multiplies chiefly by the 
production of unfertilized eggs, those which 
undergo no reduction and which develop with- 
out fertilization into an individual like the 
parent. The germinal composition, therefore^ 
of all descendants produced in this way by the 
same mother should be identical, unless germi- 
nal composition can be modified in other ways 
than by reduction and recombination of unit- 
characters. Now the German zoblogist, Wol- 
tereck, has shown that, among the offspring 
developed from the unfertilized eggs of the 
same mother Daphnia, variations do occur 
which are heritable, so that if one selects ex- 
treme variants he obtains a modified race. 
Systematic zoologists recognize as a generic 
distinction between Daphnia and Hyalodaphnia 
absence from the latter of the rudimentary eye 
found in Daphnia. Woltereck observed that 
in a pure line of Hyalodaphnia the rudi- 
mentary eye, usually wanting, may occur in 
individual cases. He found further that it 
occurred in varying degrees of development, 
which ranged all the way from a group of 
9 117 


pigmented cells outside the brain, through 
stages in which cells were present without pig- 
ment, and others in which pigment was visible 
within the brain but no cells outside it were 
developed, and finally to those in which all 
traces of the eye had vanished, cells and pig- 
ment alike. By selection in three successive 
generations of the mother having the rudi- 
mentary eye best developed offspring were 
obtained, 90 % of which had the pigmented 
eye, and which would therefore pass for ani- 
mals of a wholly different genus. The degree 
of development of the organ in the last genera- 
tion was also greater than in the previous 
generations. Here within a pure line produced 
by parthenogenesis selection served to augment 
both the degree of development of an organ 
and the frequency of its occurrence within the 
race, a result precisely parallel to that which 
I obtained some years ago by selection in the 
case of a rudimentary fourth toe in the guinea- 
pig. The experiment with Daphnia is not open 
to the objection that may be offered to the 
guinea-pig experiment, that it is possibly a 
result of gametic segregation and recombina- 



tion, for in Daphnia the reproduction was ex- 
clusively by unreduced and unfertilized eggs. 

The rudimentary eye of Daphnia is an organ 
the development of which, so far as observed, 
is wholly independent of environmental influ- 
ence; but the case is different with another 
structure of Daphnia, upon which also Wol- 
tereck made observations, namely, a projection 
or spine borne on the head of the animal. 
This is not a constant structure, but is some- 
times present, sometimes wanting altogether, in 
the same pure line. In extreme cases it forms 
a great angular extension of the head forward. 
To a considerable extent its development is 
subject to control through the temperature of 
the surrounding water, but independently of 
such influence the degree of its development 
varies and is heritable. Although in general, 
just as in the experiments of Johannsen and 
Jennings, selection of animals with the best- 
developed spine did not increase the degree of 
development of the organ or the frequency of 
its occurrence, yet in individual cases such 
increase was observed, so that the structure 
occurred in over 50 % of the offspring. In 



such cases, then, it would seem that along with 
the cases due to environmental influence oc- 
curred others due to germinal variation. Al- 
though selection of the former would not in- 
fluence the race permanently, there is every 
reason to think that the latter would so influ- 
ence it, and did in the experiment. 

Accordingly the results of Johannsen and 
Jennings on the one hand, and of Woltereck 
on the other, are not necessarily in opposition 
to each other. Woltereck 's conclusions agree 
with those of Johannsen and Jennings so far 
as concerns the great bulk of the variations, 
those caused by external influences. All agree 
that they are not inherited. Woltereck, how- 
ever, observes also, what the others have failed 
to observe, that along with the non-inherited 
variations occur other similar but less numer- 
ous ones which are inherited. 

My own observations are entirely in har- 
mony with those of Woltereck. Like him, I 
find that selection may modify characters. In 
several cases I have observed characters at 
first feebly manifested gradually improve under 
selection until they became established racial 



traits. Thus the extra toe of polydactylous 
guinea-pigs made its appearance as a poorly 
developed fourth toe on the left foot only. 
Only 6 % of the offspring of this animal by 
normal unrelated mothers were polydactylous, 
but among his offspring were some with better 
developed fourth toes than the father pos- 
sessed. Such individuals were selected through- 
out five successive generations, at the end of 
wliich time a good four-toed race had been 
established. It was found in general that those 
animals which had best-developed fourth toes 
transmitted the character most strongly in 
crosses with unrelated normal animals. The 
percentage of polydactylous individuals pro- 
duced in such crosses varied all the way from 
to 100 %. By selection this percentage was 
increased, as was also the degree of develop- 
ment of the fourth toe in crosses. 

Another character which made its appear- 
ance among our guinea-pigs,, at first feebly 
expressed, was a silvering of the colored fur, 
due to interspersing of white hairs with the 
colored ones (see Fig. 37). The first indi- 
viduals observed to have this character bore 



white hairs on the under surface of the body 
only. By inbreeding, a homozygous strain of 
the silvered animals was soon obtained, one in 


+ 2 



FIG. 41. Chart showing effects of selection in eight successive 
generations upon the color-pattern of hooded rats. A, 
average condition of the selected parents in the plus series; 
B, average condition of their offspring. A 1 , average condi- 
tion of the selected parents in the minus series; B 1 , average 
condition of their offspring. 

which all the offspring were silvered to a 
greater or less extent. Selection was now 
directed toward two ends, (1) to secure ani- 
mals which were free from spots of red or 
white, a condition which was present in the 



original stock, and (2) to secure extensive and 
uniform silvering on a black background. In 
both these objects good progress has been 
made. We have animals which are silvered all 
over the body except on a part of the head, 
and the percentage of such well-silvered indi- 
viduals is relatively high. 

But the most extensive selection experiment 
which I have personally observed is one in 
which I have been assisted by Dr. John C. 
Phillips (see Figs. 39 and 41). Selection in 
this case has been directed toward a modifica- 
tion of the color pattern of hooded rats, a 
pattern which is known to behave as a reces- 
sive Mendelian character in crosses with either 
the self (totally pigmented) condition or the 
so-called Irish (white-bellied) condition found 
in some other rats. The extreme range of 
variation among our hooded rats at the outset 
of this experiment is indicated by the grades 
- 2 and + 3 of Fig. 39. Selection was now 
made of the extreme variates in either direc- 
tion and these were bred separately. Two 
series of animals were thus established, one 
of narrow striped animals, minus series; the 



other of wide striped, plus series. In each 
generation the most extreme individuals were 
selected as parents ; in the narrow series, those 
with narrowest stripe ; in the wide series, those 
with widest stripe. 


Results of Selection for Modification of the Color-pattern 
of Hooded Rats. 

Plus series. 

Minus series. 






































The result of the selection is shown graph- 
ically in Fig. 41 (compare Table I). The 
offspring in the narrow series became with 
each generation narrower; those in the wide 
series became with each generation wider, with 
a single exception. In generation two the wide 
stock was enlarged by the addition of a new 
strain of animals. This caused a temporary 
falling off in the average grade of the young, 
the two series overlapping for that generation. 
No new stock was at any other time intro- 
duced in either series, the two remaining dis- 
tinct at all times except in generation two. 
It will be observed that a change in the aver- 
age grade of the parents is attended by a 
corresponding change in the average grade 
of the offspring. The amount of variabil- 
ity of the offspring is not materially affected 
by the selection, but the average about which 
variation occurs is steadily changed, as are 
also the limits of the range of variation. 

The interesting feature of this experiment 
is the production, as a result of selection, of 
wholly new grades ; in the narrow series, of ani- 
mals having less pigment than any known type 



other than the albino; in the wide series, of 
animals so extensively pigmented that they 
would readily pass for the " Irish type," which 
has white on the belly only, but which is known 
to be in crosses a Mendelian alternative to the 
hooded type. By selection we have practically 
obliterated the gap which originally separated 
these types, though selected animals still give 
regression toward the respective types from 
which they came. But this regression grows 
less with each successive selection and ulti- 
mately should vanish, if the story told by these 
statistics is to be trusted. As yet there is no 
indication that a limit to the effects of selec- 
tion has been reached. 

From the evidence in hand we conclude that 
Darwin was right in assigning great importance 
to selection in evolution; that progress results 
not merely from sorting out particular com- 
binations of large and striking unit-characters, 
but also from the selection of slight differences 
in the potentiality of gametes representing the 
same unit-character combinations. It is pos- 
sible to ascribe such differences to little units 
additional to the recognized larger ones, but 



if such little units exist, they are indeed very 
little as well as numerous, and by adding to the 
effect of the larger ones they produce what 
amounts to modification of them. 



1906. "The Origin of a Polydactylous Race of Guinea- 
pigs." Carnegie Institution of Washington, Publication 
No. 49, pp. 17-29. 

1909. " Heredity and Variation in the Simplest Organ- 
isms." The American Naturalist, 43, pp. 321-337. 


1909. "Elemente der exakten Erblichkeitslehre." G. 

Fisher, Jena, 516 pp. 

(See Bibliography to Chapter VI.) 


1909. " Weitere experimentelle Untersuchungen iiber Art- 
veranderungen, speciell tiber das Wesen quantitativer 
Artunterschiede bei Daphniden." Verh. Deutsch. Zool. 
Gesellsch., pp. 110-172. 



WE shall now discuss a seemingly 
different type of inheritance from 
that discovered by Mendel, one 
in which the offspring are a true intermediate 
or blend between the parents, and in which the 
occurrence of segregation has not in all cases 
been certainly established. 

Differences in size between parents have been 
found to behave in this blending fashion. Rabbits 
are apparently favorable material in which to 
study size inheritance, for some races are fully 
twice as large as others. If a large rabbit is 
crossed with a small one the young are of inter- 
mediate size, and the F 2 offspring show no such 
segregation into large, small, and intermediate 
individuals as a simple Mendelian system would 
demand. For this reason size has been de- 


FIG. 42. Skulls of three rabbits. Father (1 and la), 
mother (3 and 3a), and son (2 and 2a). 


scribed as a non-Mendelian, non-segregating 
type of inheritance, but recent discoveries place 
this interpretation in doubt. Let us first con- 
sider what are the observed facts and after- 
ward the interpretation. 

Fig. 42 shows the skulls of three rabbits, - 
of the father at the left, of the mother at the 
right, and that of the son between. Notice the 
fully intermediate or blended character of the 
son's skull as regards both absolute dimensions 
and proportions. The intermediate character 
was possessed also by the next generation of 
offspring. Now this same cross, while pro- 
ducing a blend in size and ear-length, was 
yielding dominance and segregation of coat- 
characters. Fig. 43 shows a picture of the 
rabbit with the small skull in the cross just 
described. He was an albino and his fur was 
long. The mother, which had the large skull, 
was a sooty-yellow rabbit, with short fur and 
long ears (see Fig. 44). The son is shown 
in Fig. 45. His fur was black and short, the 
albinism and long fur of his father having 
become recessive in the cross in accordance 
with Mendel's law. The pigmentation is also 



intensified in the son, black having been re- 
ceived through the albino parent as a latent 
factor, which became fully active in the son. 
The excluded albinism, recessive in the son and 
his brothers and sisters, all seven of which 
were similar in character, reappeared among 
the grandchildren, as, for example, in the one 
shown in Fig. 46, which was short-haired. 
Other F 2 offspring were long-haired, some of 
them being albinos, others being pigmented. 
But the size and ear-length of the son were 
intermediate between the sizes and ear-lengths 
of his parents, and this intermediate character 
persisted without apparent segregation among 
the F 2 offspring. The animals in the pictures 
are unfortunately not all shown on the same 
scale, but the relative ear-lengths are sufficiently 

A Mendelian interpretation of blending in- 
heritance, illustrated in the inheritance of skull- 
size and ear-length among rabbits, has been sug- 
gested by my colleague Dr. East, and by others, 
an interpretation in which Mendelian dominance 
is indeed wanting but segregation nevertheless 
occurs, yet not of a simple kind, involving one 



or two segregating factors, but involving sev- 
eral such factors. Before entering into this ex- 
planation it will be necessary to discuss a fur- 
ther extension of Mendelian principles recently 

Some modified Mendelian ratios of particular 
interest have lately been obtained by the Swedish 
plant-breeder, Nilsson-Ehle (1909, Lunds Uni- 
versitets Arsskrift) in crossing varieties of 
wheat of different color. When a variety hav- 
ing brown chaff is crossed with one which has 
white chaff, the hybrid plants are regularly 
brown in F and 3 brown : 1 white in F 2 , but 
a particular variety of brown-chaffed wheat 
gave a different result. In 15 different crosses 
it gave uniformly a close approximation to 
the ratio 15 : 1 instead of 3 : 1. The totals are 
sufficiently large to leave no doubt of this. 
They are 1410 brown to 94 white, exactly 15 : 1. 
This is clearly a dihybrid Mendelian ratio, and 
Nilsson-Ehle interprets it to mean that there 
exist in this case two independent factors, each 
of which is able by itself to produce the brown 
coloration, though no qualitative difference can 
be detected between them. 



A still more remarkable case was observed 
in crosses between varieties of wheat of dif- 
ferent grain-color. Eed crossed with white 
gave ordinarily all red in F A and 3 red to 1 
white in F 2 , but a certain native Swedish sort 
gave only red (several hundred seeds) in F 2 . 
This result was so surprising that one cross 
which had yielded 78 grains of wheat in F 2 was 
followed into F 3 , with the following result: 

50 plants gave only red seed; expected 37 

5 " " approximately 63 R: 1 W; " 8 

15 " 15R:1W; " 12 

8 " " " 3R:1W; " 6 

" all white; 1 

The interpretation given by Nilsson-Ehle is 
this. The red variety used in this cross bears 
three independent factors, each of which by 
itself is able to produce the red character. 
Their joint action is not different in kind from 
their action separately, though possibly quanti- 
tatively greater. The F 2 generation should 
contain 1 white seed in 64. It happens that 
none were obtained in this generation. The 
next generation should contain in a total of 
64 individuals, the sorts actually observed as 


DQ 3 

* a.'-- 9 ! 


45 H 2 S 

C3 r 1-1 >-i 1> 
PH I I 3 "^ 
bC "g "g ct 

tttl f 

6 6 cj o 


well as a sort which would produce only white 
seed, the progeny namely of the expected 
white seed of F 2 , but as that was not obtained, 
the all-white plant of F 3 could not be obtained 
either. The expected proportions of the sev- 
eral classes in F 3 are given for comparison with 
those actually obtained. The agreement be- 
tween expected and observed is so good as to 
make it seem highly probable that Nilsson- 
Ehle's explanation is correct. Corroborative 
evidence in the case of maize has been obtained 
by Dr. E. M. East (Am. Naturalist, Feb., 1910). 
This work introduces us to a new principle 
which may have important theoretical conse- 
quences. If a character ordinarily represented 
by a single unit in the germ-plasm may become 
represented by two or more such units identi- 
cal in character, then we may expect it to domi- 
nate more persistently in crosses, fewer reces- 
sives being formed in F 2 and subsequent gen- 
erations. Further, if duplication of a unit tends 
to increase its intensity, as seems probable, 
then we have in this process a possible expla- 
nation of quantitative variation in characters 
which are non-Mendelian, or at any rate do not 
10 133 


conform with a simple Mendelian system. Con- 
sider, for example, the matter of size and 
skeletal proportions in rabbits. It is perfectly 
clear from the experiments described that in 
such cases no dominance occurs, and also that 
no segregation of a simple Mendelian character 
takes place, but it is not certain that the ob- 
served facts may not be explained by the com- 
bined action of several similar but independent 
factors, the new principle which Nilsson-Ehle 
has brought to our attention. Let us apply 
such a hypothesis to the case in hand. 

Suppose a cross be made involving ear- 
lengths of approximately 4 and 8 inches respec- 
tively, as in one of the crosses made. The F t 
young are found to have ears about 6 inches 
long, the mean of the parental conditions, and 
the F 2 young vary about the same mean con- 
dition. If a single Mendelian unit-character 
made the difference between a 4 inch and an 8 
inch ear, the F 2 young should be of three 
classes as follows: 

Classes 4 in. 6 in. 8 in. 

Frequencies 121 


(Compare Fig. 47, bottom left.) The grand- 
parental conditions should in this case reappear 






FIG. 47. Diagrams to show the number and size of the classes 
of individuals to be expected from a cross involving Mende- 
lian segregation without dominance. One Mendelian unit 
involved, bottom left; two units, middle left; three units, 
top left; four units, right. 

in half the young. This clearly does not oc- 
cur in the rabbit experiment. But if two unit- 



characters were involved, F A would be un- 
changed, all 6 inches, yet the F 2 classes would 
be more numerous, viz., 4, 5, 6, 7, and 8 inches, 
and their relative frequencies as shown by the 
height of the columns in Fig. 47, middle left, 1, 4, 
6, 4, 1. The grandparental states would now re- 
appear in % of the F 2 young, while % would be 
intermediate. It is certain, however, that in rab- 
bits the grandparental conditions, if they re- 
appear at all, do not reappear with any such 
frequency as this. 

If three independent size-factors were in- 
volved in the cross, the F x individuals should all 
fall in the same middle group, as before, viz. 
6 inches, but the F 2 classes should number 
seven? and their relative frequencies would 
be as shown in Fig. 47, top left. For 4 independ- 
ent size-factors, the F 2 classes would be more 
numerous still, viz., 9 (Fig. 47, right), and the 
extreme ear-size of either grandparent would 
be expected to reappear in only one out of 256 
offspring, while considerably more than half 
of them would fall within the closely inter- 
mediate classes included between 5% and 61/2 
inches, the three middle classes of the diagram. 



With six size-characters, the extreme size of a 
grandparent would reappear no oftener than 
once in 4000 times, while with a dozen such 
independent characters it would recur only 
once in some 17,000,000 times. It would be 
remarkable if under such conditions the ex- 
treme size were ever recovered from an ordi- 
nary cross. 

There is one means by which we can deter- 
mine with certainty whether in a particular 
case of seemingly blending inheritance segre- 
gation does or does not occur, namely, by com- 
paring the variability of the F x and the F 2 
generations. If segregation does not occur, F 2 
should be no more variable than F 1? whereas 
if segregation does occur, F 2 should be more 
variable. For, in a segregating system, the F 
individuals should all fall in a middle, inter- 
mediate group, but the F 2 individuals should 
be distributed also in classes more remote from 
a strictly intermediate position, that is, they 
should be more variable. But, in a non-segre- 
gating system, Fj and F 2 individuals alike 
should fall in the same intermediate group, 
that is, they should have the same variability. 



The matter should be easy of determination 
by observation of considerable numbers of F! 
and F 2 offspring. Investigations are now in 
progress to test this matter. 

My colleague, Dr. East, has found clear evi- 
dence that, in maize, size-characters, although 
they give a blending result in F , nevertheless 
give segregation in F 2 . The character to be con- 
sidered relates to length of ear in corn. A single 
illustration will suffice. The variation in two 
pure varieties is shown in the two upper rows of 
Fig. 48. The " Length " of each class is given 
in centimetres, its frequency just below at " No. 
Var., ' ' abbreviation for number of variates. The 
variation in the F offspring obtained by cross- 
ing the two pure varieties is shown in the third 
row, and that of the F 2 offspring in the lowest 
row. Note that the variability in the F gen- 
eration is not increased; its range is interme- 
diate between the range in the parental varie- 
ties. In the F 2 generation, however, the varia- 
bility is so increased that it includes almost 
the entire range of both parental varieties, to- 
gether with the intervening region. 

In the light of this evidence it is clear that 

FIG. 48. Photographs to show variation in ear length of two varie- 
ties of maize (upper row), of their F l offspring (second row), and of 
their F 2 offspring (third row). (After East.) 


in maize, seemingly blending is really segre- 
gating inheritance, but with entire absence of 
dominance, and it seems probable that the same 
will be found to be true among rabbits and 
other mammals; failure to observe it hitherto 
is probably due to the fact that the factors 
concerned are numerous. For the greater the 
number of factors concerned, the more nearly 
will the result obtained approximate a com- 
plete and permanent blend. As the number 
of factors approaches infinity, the result will 
become identical with a permanent blend. 

Theoretically it is important to know whether 
segregating units are involved in inheritance 
which we call blending; practically it does not 
matter much, since if these units are only as 
numerous as six or eight it will be practically 
impossible to undo the effects of a cross and to 
recover again the conditions obtaining previous 
to the cross. The great majority of the offspring 
both in the first and in subsequent generations 
following the cross will be strictly intermediate 
between the conditions crossed whether several 
units, an infinite number of units, or no units 
at all are involved. 



A practical question of some importance is 
how to manipulate simultaneously blending (or 
seemingly blending) and Mendelian inheritance. 
This must be by a system of line-breeding in 
alternate generations, not in successive genera- 
tions. To test the practicability of this matter 
I several years ago set myself the task of com- 
bining in one race the large size of some lop- 
eared, yellow rabbits which I had, with the 
albino character of some small white rabbits 
of common race. A first cross produced gray 
rabbits of intermediate size, but no white ones. 
On inbreeding the gray animals, there were 
obtained in F 2 white young of intermediate 
size. These were now crossed again with the 
original yellow stock, and again colored young 
were obtained, but now with % of the desired 
increase in size. These bred inter se again 
produced albinos, this time of the % size. A 
third cross with the original large stock brought 
the size up to % of that desired, and combined 
it in F 2 with the desired albinism. Having 
satisfied myself of the correctness of the 
method, the experiment was now discontinued. 
By further crosses, especially with a fresh lop- 



eared stock, to avoid ill-effects of inbreeding, 
the size could have been still further increased, 
with judicious selection doubtless up to the ex- 
treme size of colored lop-eared rabbits. 

The general conclusion to be drawn is that 
in attempting to combine in one race by cross- 
breeding characters which exist separately in 
different races, one should first inquire very 
carefully how each character, in which the races 
differ, behaves in transmission, for on the an- 
swer to this question should depend the mode 
of procedure to be chosen. 

If simple Mendelian characters only are con- 
cerned, nothing is required but to cross the two 
races and select from the second generation 
offspring the desired combination. If blending 
characters only are concerned and F x yields 
the desired blend, this is secure without fur- 
ther procedure, except possibly selection to re- 
duce its variability; but if the desired blend 
is not yet secured, further back-crossing with 
one race or the other may be necessary. If, 
finally, both blending and Mendelian characters 
are simultaneously involved in a cross, then the 
method of combined line-breeding and selec- 



tion in alternate generations, already described, 
should be adopted. 



1909. (See Bibliography to Chapter V.) 
EAST, E. M. 

1910. "A Mendelian Interpretation of Variation that is 
Apparently Continuous." American Naturalist, 44, 
pp. 65-82. 


1910. "The Inheritance of Sizes and Shapes in Plants." 

American Naturalist, 44, pp. 739-746. 

1909. "Kreuzungsuntersuchungen an Hafer und Weizen." 
Lunds Universitets Arsskrift, 5, No. 2, 122 pp. 



WHAT is the probable source of the 
evil effects which have been fre- 
quently observed to follow in- 

By- inbreeding we mean the mating of closely 
related individuals. As there are different de- 
grees of relationship between individuals, so 
there are different degrees of inbreeding. The 
closest possible inbreeding occurs among plants 
in what we call self-pollination, in which the 
egg-cells of the plant are fertilized by pollen- 
cells produced by the same individual. A simi- 
lar phenomenon occurs among some of the 
lower animals, notably among parasites. But 
in all the higher animals, including the domes- 
ticated ones, such a thing is impossible because 
of the separateness of the sexes. For here no 
individual produces both eggs and sperm. The 



nearest possible approach to self-pollination is 
in such cases the mating of brother with sister, 
or of parent with child. But this is less close 
inbreeding than occurs in self-pollination, for 
the individuals mated are not in this case iden- 
tical zygotes, though they may be similar ones. 

It has long been known that in many plants 
self-pollination is habitual and is attended by 
no recognizable ill-effects. This fortunate cir- 
cumstance allowed Mendel to make his remark- 
able discovery by studies of garden-peas, in 
which the flower is regularly self -fertilized, and 
never opens at all unless made to do so by 
some outside agency. Self-pollination is also 
the rule in wheat, oats, and the majority of the 
other cereal crops, the most important econom- 
ically of cultivated plants. Crossing can in such 
plants be brought about only by a difficult 
technical process, so habitual is self-pollina- 
tion. And crossing, too, in such plants is of 
no particular benefit, unless by it one desires 
to secure new combinations of unit-characters. 

In maize, or Indian corn, however, among 
the cereals, the case is quite different. Here 
enforced self-pollination results in small un- 



productive plants, lacking in vigor. But racial 
vigor is fully restored by a cross between two 
depauperate unproductive individuals obtained 
by self-fertilization, as has recently been shown 
by Shull. This result is entirely in harmony 
with those obtained by Darwin, who showed 
by long-continued and elaborate experiments 
that while some plants do not habitually cross 
and are not even benefited by crossing, yet in 
many other plants crossing results in more 
vigorous and more productive offspring; that 
further, the advantage of crossing in such cases 
has resulted in the evolution in many plants 
of floral structures, which insure crossing 
through the agency of insects or of the wind. 

In animals the facts as regards close fer- 
tilization are similar to those- just described 
for plants. Some animals seem to be indiffer- 
ent to close breeding, others will not tolerate 
it. Some hermaphroditic animals (those which 
produce both eggs and sperm) are regularly 
self-fertilized. Such is the case, for example, 
with many parasitic flat-worms. In other cases 
self-fertilization is disadvantageous. One such 
case I was able to point out some fifteen years 



ago, in the case of a sea-squirt or tunicate, 
Ciona. The same individual of Ciona produces 
and discharges simultaneously both eggs and 
sperm, yet the eggs are rarely self-fertilized, 
for if self-fertilization is enforced by isolation 
of an individual, or if self-fertilization is 
brought about artificially by removing the eggs 
and sperm from the body of the parent and 
mixing them in sea-water, very few of the 
eggs develop, less than 10%. But if the 
eggs of one individual be mingled with the 
sperm of any other individual whatever, prac- 
tically all of the eggs are fertilized and 

In the great majority of animals, as in many 
plants, self-fertilization is rendered wholly im- 
possible by separation of the sexes. The same 
individual does not produce both eggs and 
sperm, but only one sort of sexual product. 
But among sexually separate animals the same 
degree of inbreeding varies in its effects. The 
closest degree, mating of brother with sister, 
has in some cases no observable ill-effects. 
Thus, in the case of a smal] fly, Drosophila, 
my pupils and I bred brother with sister for 



fifty-nine generations in succession without ob- 
taining a diminution in either the vigor or the 
fecundity of the race, which could with cer- 
tainty be attributed to that cause. A slight 
diminution was observed in some cases, but 
this was wholly obviated when parents were 
chosen from the more vigorous broods in each 
generation. Nevertheless crossing of two in- 
bred strains of Drosophila, both of which were 
doing well under inbreeding, produced off- 
spring superior in productiveness to either 
inbred strain. Even in this case, therefore, 
though inbreeding is tolerated, cross-breeding 
has advantages. 

In the case of many domesticated animals, 
it is the opinion of experienced breeders, sup- 
ported by such scientific observations as we 
possess, that decidedly bad effects follow con- 
tinuous inbreeding. Bos ( '94) practiced con- 
tinuous inbreeding with a family of rats for 
six years. No ill-effects were observed during 
the first half of the experiment, but after that 
a rapid decline occurred in the vigor and fer- 
tility of the race. The average-sized litter in 
the first half of the experiment was about 7.5, 



but in the last year of the experiment it had 
fallen to 3.2, and many pairs were found to 
be completely sterile. Diminution in size also 
attended the inbreeding, at the end amounting 
in the case of males to between 8 and 20 %. 

Experiments made by Weismann confirm 
those of Bos as regards the falling off in fer- 
tility due to inbreeding. For eight years Weis- 
mann bred a colony of mice started from nine 
individuals, six females and three males. 
The experiment covered 29 generations. In 
the first 10 generations the average number of 
young to a litter was 6.1; in the next 10 gen- 
erations, it was 5.6; and in the last 9 genera- 
tions, it had fallen to 4.2. But sweeping 
generalizations cannot be drawn from these 
cases. Each species of animal must probably 
be tested for itself before we shall know what 
the exact effects of inbreeding are in that case. 
In guinea-pigs, a polydactylous race built up 
by the closest inbreeding out of individuals all 
descended from one and the same individual 
has now been in existence for ten years. It 
consists of one of the largest and most vigor- 
ous strains of guinea-pigs that I have ever 



seen, and has shown no indications of dimin- 
ished fertility. 

In the production of pure breeds of sheep, 
cattle, hogs, and horses inbreeding 'has fre- 
quently been practiced extensively, and where 
in such cases selection has been made of the 
more vigorous offspring as parents, it is doubt- 
ful whether any diminution in size, vigor, or 
fertility has resulted. Nevertheless it very 
frequently happens that when two pure breeds 
are crossed, th6 offspring surpass either pure 
race in size and vigor. This is the reason for 
much cross-breeding in economic practice, the 
object of which is not the production of a new 
breed, but the production for the market of 
an animal maturing quickly or of superior size 
and vigor. The inbreeding practiced in form- 
ing a pure breed has not of necessity dimin- 
ished vigor, but a cross does temporarily (that 
is in the F x generation) increase vigor above 
the normal. Now why should inbreeding un- 
attended by selection decrease vigor, and cross- 
breeding increase it? We know that inbreed- 
ing tends to the production of homozygous 
conditions, whereas cross-breeding tends to 
11 149 


produce heterozygous conditions. Under self- 
pollination for 1 generation following a cross, 
half the offspring become homozygous; after 
2 generations, % of the offspring are homo- 
zygous ; after 3 generations % are homozygous, 
and so on. So if the closest inbreeding is 
practiced there is a speedy return to homo- 
zygous, pure racial conditions. We know fur- 
ther that in some cases at least heterozygotes 
are more vigorous than homozygotes. The 
heterozygous yellow mouse is a vigorous lively 
animal; the homozygous yellow mouse is so 
feeble that it perishes as soon as produced, 
never attaining maturity. Cross-breeding has, 
then, the same advantage over close-breeding 
that fertilization has over parthenogenesis. It 
brings together differentiated gametes, which, 
reacting on each other, produce greater meta- 
bolic activity. 

Inbreeding, also, by its tendency to secure 
homozygous combinations, tends to bring to 
the surface latent or hidden recessive charac- 
ters. If these are in nature defects or weak- 
nesses of the organism, such as albinism and 
feeble-mindedness in man, then inbreeding is 



distinctly bad. Existing legislation against 
the marriage of near-of-kin is, therefore, on 
the whole, biologically justified. On the other 
hand, continual crossing only tends to hide 
inherent defects, not to exterminate them; and 
inbreeding only tends to bring them to the 
surface, not to create them. We may not, 
therefore, lightly ascribe to inbreeding or in- 
termarriage the creation of bad racial traits, 
but only their manifestation. Further, any 
racial stock which maintains a high standard 
of excellence under inbreeding is certainly one 
of great vigor, and free from inherent defects. 
The animal breeder is therefore amply jus- 
tified in doing what human society at present 
is probably not warranted in doing, viz. in 
practicing close inbreeding in building up 
families of superior excellence and then keep- 
ing these pure, while using them in crosses 
with other stocks. For an animal of such a 
superior race should have only vigorous, strong 
offspring if mated with a healthy individual 
of any family whatever, within the same spe- 
cies. For this reason the production of 
" thoroughbred " animals and their use in 



crosses is both scientifically correct and com- 
mercially remunerative. 



1894. " Untersuchungen ueber die Folgen der Zucht in 
engster Blutverwandtschaft." Biol. CentrbL, 14, pp. 75- 

S. 0., and BARROWS, W. M. 

"The Effects of Inbreeding, Cross-breeding and Selec- 
tion upon the Fertility and Variability of Drosophila." 
Proc. Amer. Acad. Arts and Sci., 41, pp. 731-786. 

1898. "Versuche mit Kreuzungen von verschiedenen 
Rassen des Hausmaus." Ber. naturf. Gesellsch. zu 
Freiburg, 10, pp. 317-332. [Contains observations of 



THE value of a domesticated animal often 
depends in considerable measure on its 
sex. Therefore, if a means could be 
devised for controlling the sex of offspring, it 
would be of great economic value to the breeder. 
Endless attempts have been made to do this, 
and occasionally a claim of success has been 
made, but none of these claims has withstood 
the test of critical analysis or experiment. The 
hypotheses advanced to explain how sex may 
be controlled have been of the most varied 
character. In some the determination of sex 
has been supposed to inhere in the nature of 
the parents, in others it is referred to condi- 
tions of the gametes themselves. 

Relative age or vigor of the parents have 
been supposed to influence sex in various ways. 
The same idea has been advanced regarding 



the gametes themselves, it being supposed that 
early or late fertilization of the egg might 
influence its sex. Experimental evidence, how- 
ever, as to these several hypotheses is wholly 
negative, when one eliminates other possible 
factors from the experiment. Everything 
points to the conclusion that sex rests in the 
last analysis upon gametic differentiation, just 
as the color of a guinea-pig in a mixed race 
of blacks and whites depends upon whether the 
gametes which unite to produce it carry black 
or white. As the heterozygous black guinea- 
pig forms black-producing and white-producing 
gametes in equal numbers, so there is reason 
to think male-producing and female-producing 
gametes are formed in equal numbers by the 
parent, in many cases at least. But is it not 
possible that there may exist individuals which 
produce the two sorts of gametes in unequal 
numbers, and so would have a tendency to 
produce more offspring of one sex than of the 
other? Perhaps so, though we have no evi- 
dence that such a condition, if it does exist, is 
transmitted from one generation to another. 
On this point I made experimental observa- 



tions upon guinea-pigs extending over a series 
of years. Oftentimes I found an individual 
that produced more offspring of one sex than 
of the other, but this was probably due merely 
to chance deviations from equality. I could 
get no evidence that the condition was inher- 
ited, though the experiment was continued 
through as many as seven generations, includ- 
ing several hundred offspring. 

The essential difference between a female 
and a male individual is that one produces 
eggs, the other sperm. All other differences 
are secondary and dependent largely upon the 
differences mentioned. If in the higher ani- 
mals (birds and mammals) the sex-glands (i. e. 
the egg-producing and sperm-producing tis- 
sues) are removed from the body, the super- 
ficial differences between the sexes largely dis- 
appear. In insects, however, the secondary 
sex-characters seem to be for the most part 
uninfluenced by presence or absence of the 
sex-glands. Their differentiation occurs in- 
dependently though simultaneously with that 
of the sex-glands. 

The egg or larger gamete (the so-called 


macro-gamete) in all animals is non-motile and 
contains a relatively large amount of reserve- 
food material for the maintenance of the de- 
veloping embryo. This reserve-food material 
it is the function of the mother to supply. In 
the case of some animals, for example flat- 
worms and mollusks, the food-supply of the 
embryo is not stored in the egg-cell itself, but 
in other cells associated with it, and which 
break down and supply nourishment to the 
developing embryo derived from the fertilized 
egg. Again, as in the mammals, the embryo 
may derive its nourishment largely from the 
maternal tissues, the embryo remaining like a 
parasite within the maternal body during its 
growth, feeding by absorption. But in all cases 
alike the mother supplies the larger gamete 
and the food-material necessary to carry the 
zygote through its embryonic stages. The 
father, on the other hand, furnishes the bare 
hereditary equipment of a gamete, with the 
motor apparatus necessary to bring it into 
contact with the egg-cell, but without food for 
the developing embryo produced by fertiliza- 
tion. The gamete furnished by the father is 



therefore the smaller gamete, the so-called 

From the standpoint of metabolism, the 
female is the more advanced condition; the 
female performs the larger function, doing all 
that the male does in furnishing the material 
basis of heredity (a gamete), and in addition 
supplying food for the embryo. As regards 
the reproductive function, the female is the 
equivalent of the male organism, plus an ad- 
ditional function, that of supplying the em- 
bryo with food. When we come to consider the 
structural basis of sex, we find reasons for 
thinking that here, too, the female individual 
is the equivalent of the male plus an addi- 
tional element. The conclusion has very natu- 
rally been drawn that if a means could be 
devised for increasing the nourishment of the 
egg or embryo, its development into a female 
should be thereby insured, while the reverse 
treatment should lead to the production of a 
male. But in practice this a priori expecta- 
tion is not fulfilled. Better nourishment of the 
mother may lead to the production of more 
eggs, but not of more female offspring, as has 



been repeatedly demonstrated by experiment. 
Also poor nutrition of the mother may diminish 
the number of eggs which she liberates, but will 
not increase the proportion of males among the 
offspring produced. 

An excellent summary of evidence on this 
point was made by Cuenot in 1900. Attempts 
to influence the sex of an embryo or larva by 
altered nutrition of the embryo or larva itself 
have proved equally futile. Practically the 
only experimental evidence of value in favor 
of this idea has been derived from the study 
of insects, and this is capable of explanation 
on quite different grounds from those which 
first suggest themselves. It has sometimes been 
observed, as by Mary Treat for example, that 
a lot of insects poorly fed produce an excess 
of males. In such lots, however, the mortality 
is commonly high, and more females die than 
males, because the female is usually larger and 
requires more food to complete its development. 
The fallacy in concluding from such evidence 
that scanty nutrition causes individuals which 
would otherwise become females to develop 
into males was indicated years ago by Riley. 



Nevertheless an argument for the artificial con- 
trol of sex based on such evidence is from time 
to time brought forward, as, for example, a 
few years since by Schenk. The latest advocate 
of sex-control by artificial means is an Italian, 
Eusso (1909). He claims in the case of rabbits 
that by feeding the mother on lecithin or by 
injections of lecithin, the proportion of female 
births may be increased. His evidence in sup- 
port of this claim is, however, wholly inade- 
quate, and two independent repetitions of his 
experiments, made by Basile in Italy and by 
Punnett in England, have given entirely nega- 
tive results. 

An alternative hypothesis concerning the de- 
termination of sex has been steadily gaining 
ground during the last ten years, that sex has 
its beginning in gametic differentiation and is 
finally determined beyond recall in the ferti- 
lized egg by the nature of the uniting gametes. 
Instructive in this connection is a study of 
parthenogenesis, reproduction by unfertilized 
eggs. But before entering upon this, it may 
be well to review briefly the changes which 
regularly take place in the egg which is to be 



fertilized, and compare with this the changes 
which occur in eggs not to Be fertilized. 

In each cell of the ordinary animal there 
occurs a characteristic number of bodies called 
chromosomes. We do not know that they are 
any more important than other cell constitu- 
ents, but we know their history better. These 
are contained in the nucleus of the cell, and at 
the time of nuclear diyision they are found at 
the equator of the division spindle. For ex- 
ample, in the egg of the mouse (Fig. 4, A), the 
nucleus is seen to be in the spindle stage, and 
its chromosomes are gathered together at the 
equator of the spindle. There each of them 
regularly splits in two, and one derivative goes 
to either end of the spindle, and so into one of 
the daughter-nuclei. Thus each new nucleus 
has, as a rule, the same chromosome composi- 
tion as the nucleus from which it was derived. 

But the egg which is to be fertilized under- 
goes two nuclear divisions in succession, in only 
one of which do the chromosomes split (see 
Fig. 4, A-D). In the other division the chromo- 
somes separate into two groups without split- 
ting, and each group goes into a different cell 



product. Consequently, in each of these prod- 
ucts the number of chromosomes is reduced to 
half what it is in the cells of the parental body. 
Thus in the egg of the mouse, by maturation, 
the number of chromosomes becomes reduced 
from about twenty-four to about twelve. 

Similar changes occur in the developing 
sperm-cell (see Fig. 5). Starting with the 
double or 2 N chromosome number, there are 
formed by two nuclear divisions, with only one 
splitting of chromosomes, four cells, each with 
the reduced or simplex number of chromosomes, 
N. Consequently, when the sperm enters the 
egg at fertilization it brings in a group of N 
chromosomes (in the mouse apparently twelve), 
which, added to the egg-contribution of N 
chromosomes, brings the number in the new 
organism again up to 2 N (in the mouse twenty- 

Now, as regards the maturation of partheno- 
genetic eggs, those which are to develop with- 
out having been fertilized, three categories of 
cases deserve separate discussion. The simplest 
of these in many respects is found among the 
social hymenoptera (ants, bees, and wasps). 



See Fig. 49, left column. The eggs are, so far 
as we can discover, all of a single type. They 




\ ^~7^^ 

\y. \ . 

FIG. 49. Diagram of sex-determination in parthenogenesis. 
First row, nuclear condition of the parthenogenetic mother; 
second row, of her eggs when they develop without reduction, 
after forming a single polar cell; third row, condition of the 
eggs after complete maturation the fertilized egg in each 
case produces a male; fourth row, nuclear condition of the 
fertilized egg, always a female. 

undergo maturation in the manner already de- 
scribed, the chromosomes being reduced to the 



N or simplex number. The eggs of most ani- 
mals, after they have undergone reduction, are 
incapable of development unless fertilized, but 
those of the hymenoptera may develop either 
fertilized or unfertilized. In the former case 
a female is produced, in the latter a male. The 
simplex, or N condition is in this case the male, 
the duplex or 2 N condition is the female, natu- 
rally the one of higher metabolic activity, the 
one which forms the macro-gametes. 

In an earlier chapter I explained how the 
development of the sperm-cells in a male having 
the reduced or simplex number of chromosomes 
differs from that in the ordinary male. Eefer- 
ence to Fig. 8 may help to recall this. The 
cells of the male are in this case already in the 
reduced or simplex condition, N. In the pro- 
duction of the sperms the reducing division is 
omitted so far as nuclear components are con- 
cerned, so that each sperm formed contains the 
full simplex chromosome number, N. If it were 
less, the gamete formed would perhaps not be 
capable of transmitting all the hereditary char- 
acteristics of an individual. 

A second category of cases (Fig. 49, middle 


column) is represented by such simple aquatic 
organisms as rotifers and small Crustacea, like 
Daphnia. In these parthenogenesis occurs ex- 
clusively, when the food supply is very abun- 
dant and conditions otherwise favorable, whereas 
reproduction by fertilized eggs occurs only 
when external conditions, including food-supply, 
are not good. Under favorable conditions only 
female offspring are produced. The conclusion 
has naturally but erroneously been drawn that 
good nutrition in itself favors the production 
of females in animals generally, which is not 
true. The egg produced by Daphnia, or by a 
rotifer, under optimum conditions does not un- 
dergo reduction (see Fig. 49, second row). It 
remains in the 2 N condition, forming but a 
single polar cell. It is therefore unprepared 
for fertilization, and in fact it is not fertilized. 
Its sex is like that of the animal which formed 
it, female. Under unfavorable conditions, how- 
ever, the eggs of the rotifer and of Daphnia 
do not begin development until they have un- 
dergone maturation. They are also of two 
sizes (Fig. 49, third row), small eggs, which 
develop without fertilization and which form 



males, and large eggs, which require fer- 
tilization, and which form females. In this 
category of cases, as in that of the hymenop- 
tera, the egg which develops in the 2 N condi- 
tion, either from failure of reduction to occur 
in maturation or from fertilization following 
reduction, forms a female; whereas the egg 
which develops in the N condition forms a 

In a third category of cases there is a quan- 
titative difference in chromatin between male 
and female, just as in the foregoing cases, but 
this does not amount to a whole set of chromo- 
somes, N, but to only a partial set, one or two 
chromosomes (see Fig. 49, right column). This 
category of cases occurs in plant-lice (aphids 
and phylloxerans) ; evidence of its existence 
rests chiefly on recent observations made by 
von Baehr and Morgan. Females are formed 
by parthenogenesis without reduction, occurring 
under favorable conditions, just as in the case 
of rotifers. Females are also formed by fer- 
tilization following reduction under unfavor- 
able conditions, just as in rotifers. In both 
cases the female is 2 N. Males arise only by 
12 165 


parthenogenesis under unfavorable conditions, 
just as in rotifers, but the reduction which 
occurs before development begins is partial 
only. A whole set, N, of chromosomes is not 
eliminated in maturation, but only 1 or 2 chro- 
mosomes. Hence the male condition here is 
2 N 1 or 2. The condition of the gametes 
formed, however, is N in both sexes. In 
spermatogenesis, division of the germ-cells 
takes place into N and N - - 1 daughter cells, 
but the latter degenerate (like the non-nucleated 
cells of the bee and wasp), and only the former 
produce spermatozoa. Hence in fertilization 
only 2 N zygotes are produced, which are in- 
variably female. 

Summarizing the three categories described, 
we may say that in all known cases of par- 
thenogenesis, the female is in the duplex, 2 N 
condition, the male in the simplex (N) or par- 
tially duplex condition (2 N -- 1, or 2 N 2). 
The female in all cases has the greater chro- 
matin content. 

In a great many insects and other arthro- 
pods, which are not parthenogenetic, it is 
known that, although the male, like the female, 



develops only from a fertilized egg, neverthe- 
less the male possesses fewer chromosomes than 

FIG. 50. Diagram of sex-determination when the female is 
homozygous, the male heterozygous. 

the female. In such cases the female forms, 
as in cases of parthenogenesis, only N gametes, 
but the male forms gametes of two sorts, N and 



N 1 or N 2 (see Fig. 50). In consequence 
zygotes of two sorts result, those which are 
2 N, female, and those which are 2 N 1 or 
2 N 2, male. Thus in the squash-bug, Anasa- 
tristis, according to Wilson, the mature egg 
contains 11 chromosomes, the spermatozoa 
either 10 or 11 chromosomes, the two sorts 
being equally numerous. 

Egg 11 + sperm 11 produces a zygote 22 (2N), a female; 
Egg 11+ " 10 " " " 21(2N-l),amale. 

N in this species 11 ; 2 N == 22, the female ; 
2 N 1 = 21, the male. Males and females are 
therefore approximately equal in number, as in 
most animals where the two sexes are not sub- 
ject to unequal mortality. In the Mendelian 
sense the female is in such cases a homozygote, 
the male a heterozygote. The sex of an indi- 
vidual in such cases depends upon which sort 
of a sperm chances to enter the egg. 

But the experimental evidence indicates 
that both as regards sex and as regards 
heritable characters correlated with sex, these 
relations may in some cases be reversed, the 
female being heterozygous, the male homozy- 



gous. In such cases there is reason to think 
that structurally the male is 2 N but the female 
2N +. That is, the female is still the equiva- 
lent of the male plus some additional element 
and function. A structural basis in the chromo- 
somes for such a condition has been described 
by Baltzer in the case of the sea-urchin. He 
found the regular duplex number of chromo- 
somes in the male; but in the female, while 
the number was the same, one of the chromo- 
somes was larger than its mate, having an extra 
or odd element attached to it. In such a case 
the gametes formed by the male would all be 
N, but those formed by the female would be 
of two sorts equally numerous, viz. N and N + 
(see Fig. 51). Egg N fertilized by sperm N 
would produce a zygote 2 N, a male ; egg N + 
fertilized by sperm N would produce a zygote 
2 N +, a female. Hence here, as in other ani- 
mals, the sexes would be approximately equal, 
but the sex of a particular individual would 
depend upon which sort of egg gave rise to it. 
Upon the existence, as in the foregoing cases, 
of an unpaired or odd structural element in 
the egg, may perhaps depend the explanation 



of a curious sort of heredity known as sex- 
limited heredity. 

FIG. 51. Diagram of sex-determination when the female is 
heterozygous, the male homozygous. 

Every one who knows anything about poultry 
is acquainted with the popular American breed 
known as the barred Plymouth Eock. In this 



breed the feathers are marked with alternate 
bars of darker and lighter black. Pure barred 
Eocks breed true, but when crossed with other 
breeds, the male proves to be homozygous, the 
female, heterozygous in barring. For the male 
Eock crossed with a non-barred breed produces 
only barred offspring in both sexes, but the 
female Eock crossed with the same non-barred 
breed produces offspring approximately half 
of which are barred, the other half being non- 
barred. Further, the barred individuals in this 
cross are invariably males, the non-barred ones 
being females. Accordingly, the distribution of 
barring and non-barring in the cross is sex- 

The barred offspring produced by a cross 
between barred Plymouth Eocks and a non- 
barred breed, whether those offspring are males 
or females, prove to be heterozygous in barring, 
as we should expect, the barring factor having 
been received only from one parent, the barred 
one. Further, the non-barred offspring pro- 
duced by a barred Eock female crossed with 
a non-barred breed, do not transmit barring, 
hence they are pure recessives as regards bar- 



ring. Hence, also, we are forced to conclude, 
as already suggested, the female of the pure 
barred Rock breed is heterozygous as re- 
gards barring, and transmits the character only 
to her male offspring, her female offspring 
(if the father is non-barred) neither being 
themselves barred nor being able to transmit 

A pure Plymouth Eock race breeds true to 
barring merely because all its males are pure, 
for the females are not pure. This is shown 
by the following experiment. If a heterozygous 
barred male, produced by a cross between a 
Eock and a non-barred breed, is crossed with 
barred females, either those of a pure Eock 
race or those produced by a cross, the result 
is the same. The male offspring are all barred ; 
the females, half of them barred, half non- 
barred. This result shows that all barred 
females alike are heterozygous in barring. 

Sex-limited inheritance such as this finds at 
the present time its most probable explanation 
in the existence in the egg of an extra or plus 
element never found in the sperm, this element 
pairing with the sex-limited character in the 



reduction division. Thus, in the barred Rock, 
calling barring B, the male of pure race is 
plainly B B and every sperm is B. But the 
female clearly contains only one B and can- 



FIG. 52. Diagram of sex-limited inheritance when the female 
is a heterozygote, as in barred fowls. X, female sex deter- 
miner; B, barring. 

not be made to contain two. Perhaps a second 
B is kept out by some structural element, X, 
the distinctive structural element of the female 
individual. Then the eggs will be of two 
sorts: B and X. Since the sperms are all B, 
the first type of egg when fertilized will con- 
tain B B, a homozygous barred individual and 



a male, since it lacks X; the second type will 
contain B X, a bird heterozygous in barring, 
and a female, since it contains X. This agrees 
with the experimental result (see Fig. 52). 

A heterozygous barred male will form two 
kinds of sperm, only one of which will contain 
B, If such a male be mated with a barred 
female, four sorts of zygotes should result, as 
follows : 

Gametes of heterozygous barred male = B and , 
Gametes of barred female = B and X, 

Zygotes = BB (homozygous barred male); B 

(heterozygous barred male), B'X (barred female), and 

X (non-barred female). 

The observed result of this cross accords 
fully with the foregoing expectation. 

The sex-limited inheritance of barring in 
fowls may be explained, as we have just seen, 
on the assumption that the female is the hetero- 
zygous sex. The same is true of sex-limited 
inheritance in canary-birds and in the moth, 
Abraxas, according to Bateson and Doncaster. 
But these relations are exactly reversed in the 
pomace-fly, Drosophila ampelophila, according 
to Morgan. 



In Drosophila the female is apparently 

homozygous as regards some cell-structure, 

X, which in the male is never represented 

more than once. Accordingly the formula of 


FIG. 53. Diagram of sex-limited inheritance when the female 
is a homozygote, as in the red-eyed Drosophila. X, sex- 
determiner; R, red-eyes. 

the female is in such cases X X ; that of the 
male, X . Now the sex-limited characters in 
Drosophila seem to be bound up with the X 
structure, not repelled by it, as is barring in 
fowls. Accordingly, a sex-limited character 
may be represented twice in the female Droso- 
phila, but only once in the male; or in other 



words, the female may be homozygous as re- 
gards a sex-limited character, but the male can 
only be heterozygous (see Fig. 53). 

Drosophila normally has red eyes, but the 
redness of the eye is a distinct unit-character, 
sex-limited in heredity. Further males are 
regularly heterozygous in this character, while 
females are homozygous. For Morgan has ob- 
tained a race in which the eyes are white, 
owing to the loss of the red character; and 
reciprocal crosses of this race with ordinary 
red-eyed animals yield different results. The 
red-eyed female crossed with a white-eyed male 
produces only red-eyed offspring, but the red- 
eyed male crossed with a white-eyed female 
produces offspring only half of which are red- 
eyed, viz. the females, whereas the males are 

These different results in the two cases ap- 
parently come about as follows: 

First case. 

Gametes of red-eyed female = X-R and X-R, 
Gametes of white-eyed male = X and , 
Zygotes= XX-R (red-eyed female), and = -X-R 
(red-eyed male). 



Second case. 

Gametes of white-eyed female = X and X, 
Gametes of red-eyed male = X-R and , 
Zygotes = X-X-R (red-eyed female), and -'X 
(white-eyed male). 

A short condition of the wings in Drosophila, 
which renders the animal incapable of flight, 
is likewise sex-limited in heredity, as has been 
shown by Morgan. By crossing two races of 
Drosophila, each of which possessed a different 
sex-limited character, Morgan has been able to 
combine the two characters in a single race. 
Thus was obtained a race both white-eyed and 
short-winged. The synthesis cannot be made 
originally in a male individual, but only in a 
female. For only in the female can the two 
characters be brought together, each associated 
with a different X, since in the male only one 
X is present. Although each sex-limited char- 
acter seems to be attached to or bound up with 
an X structure, it evidently has a material basis 
distinct from X. Otherwise it would not be 
possible for the character to leave one X and 
attach itself to the other, as apparently takes 
place in the female when the combination of 



two sex-limited characters in the same gamete 
is secured through a cross. The combination 
is apparently secured in this way: 

Gametes uniting, X-R and X L, 

Zygote formed, X-R X-L, 

Its gametes, X-R and X-L, or X-R-L and X. 

One of the uniting gametes, X-E, is formed 
by the red-eyed, short-winged parent ; the other, 
X-L, is formed by the long-winged, white- 
eyed parent. The zygote resulting is a red- 
eyed individual, since it contains R ; it is long- 
winged, since it contains L ; it is a female, since 
it contains two Xs. Now, its gametes are of 
four sorts, as indicated. The first two sorts 
result from simple separation of the two Xs, 
each with its associated character, R in one 
case, L in the other. But the third sort could 
result only from the attachment of R and L 
to the same X, leaving the other X without 
either R or L as the fourth kind of gamete. 
This kind, which transmits neither red eyes nor 
long wings, would represent the new gametic 
combination, white-eyed and with short wings. 

The experimental evidence that gametes of 


these four sorts are formed by females of the 
origin described is as follows : When such a 
female is mated with a long- winged, white-eyed 
male, there are obtained female offspring, all 
of which are long- winged, but half of them are 
red-eyed, half white-eyed. The male offspring, 
however, are of four sorts, viz. red short, white 
long, red long, and white short. This result 
harmonizes with the hypothesis advanced. For 
if the gametes of the female are X-R, X-L, 
X-R-L, and X, and those of the male are X-L 
and , then the following combinations should 
result : 

X-L 1 X-R, red long female, 
X-L- X-L, white long female 
X-L- X-R-L, red long female, 
X-L- X , white long female, 

X-R, red short male, 

X-L, white long male, 

- X-R-L, red long male, 
X , white short male. 

This expected result accords with that ac- 
tually obtained by Morgan. 

Color-blindness in man is a sex-limited char- 
acter, the inheritance of which resembles that 



of white eyes or short wings in Drosophila, 
rather than of barring in poultry. 

Color-blindness is much commoner in men 
than in women. A color-blind man, however, 
does not transmit color-blindness to his sons, 
but only to his daughters, the daughters, how- 
ever, are themselves normal provided the 
mother was; yet they transmit color-blindness 
to half their sons. A color-blind daughter 
could be produced, apparently, only by the 
marriage of a color-blind man with a woman 
who transmitted color-blindness, since the 
daughter to be color-blind must have received 
the character from both parents, whereas the 
color-blind son receives the character only 
from his mother. 

Color-blindness is apparently due to a defect 
in the germ-cell, absence of something nor- 
mally associated there with an X-structure, 
which is represented twice in woman, once ii 
man. Color-blindness follows, therefore, 
transmission the scheme shown in Fig 53. 

If, as has been suggested, the determination 
of sex in general depends upon the inheritan< 
of a Mendelian factor differentiating the sexe* 



it is highly improbable that the breeder will 
ever be able to control sex. Male and female 
zygotes should forever continue to be produced 
in approximate equality, and consistent inequal- 
ity of male and female births could result only 
from greater mortality on the part of one sort 
of zygote than of the other. Only in partheno- 
genesis can man at will control sex, and until 
he can produce artificial parthenogenesis in the 
higher animals, he can scarcely hope to con- 
trol sex in such animals. 

Negative as are the results of our study of sex- 
control, they are perhaps not wholly without 
practical value. It is something to know our 
limitations. We may thus save time from 
useless attempts at controlling what is uncon- 
trollable and devote it to more profitable em- 



1909. (See Bibliography to Chapter IV.) 

1909. "A Mendelian View of Sex-heredity." Science, 

N. S., vol. 29, pp. 395-400. 

1900. "Sur la determination du sexe chez les animaux." 
Bull. Sci. de la France et de la Belgique, 

13 181 



1909. "A Biological and Cytological Study of Sex Deter- 
mination in Phylloxerans and Aphids." Journal of 
Experimental Zoology, 7, pp. 239-352. 

1910. "Sex-limited Inheritance in Drosophila." Science, 
N. S. t 32, pp. 120-122. 

1911. "The Application of the Conception of Pure Lines 
to Sex-limited Inheritance and to Sexual Dimorphism." 
The American Naturalist, 45, pp. 65-78. 

Russo, A. 

1909. "Studien iiber die Bestimmung des weiblichen 

Geschlectes." G. Fischer, Jena. 

1909. "Recent Researches on the Determination and 
Heredity of Sex." Science, N. S., 29, pp. 53-70. 

1910. "The Chromosomes in Relation to the Determina- 
tion of Sex." Science Progress, 5, pp. 570-592. 

For references to the earlier literature see CUENOT and 


Abraxas, sex determination in, 

Atavism, 62. 

von Baehr, 165. 
Basile, 159. 
Bateson, 37, 110, 174. 
Baur, 61. 
Bos, 147. 
Buttercup, 107. 

Cattle, polled, 39, 102. 

Ciona, 146. 

Color blindness, 179. 

Correns, 34. 

Coutagne, 99. 

Cross-breeding, effects of, on 

vigor, 149. 
Cuenot, 158. 

Daphnia, 116, 164. 
Darwin, 7, 62, 106. 
Davenport, 100. 
Donraster, 174. 
Drosophila, 146. 
sex limited inheritance in, 

East, 130, 138. 

Egg, fertilization of, 11. 
of Nereis, 12. 
of mouse, 13. 
of sea-urchin, 18. 

Factors, inhibiting, 55. 

multiple, 62, 131. 
Farabee, 39. 
Fern, 20. 

prothallus of, 22. 
spores of, 22. 
Fingers, inheritance of short, 


Fixation of a reversionary char- 
acter, 68. 

Fowls, Andalusian, 55. 
crosses of, 53. 

sex limited inheritance in, 

Gamete, definition of, 25. 
Guinea-pig, angora, 38, 42, 47. 

black crossed with white, 34. 

new variety of, 84. 

polydactylous, 100, 121. 

reversion in, 63. 

rough crossed with smooth, 

38, 41, 47, 98. 
Hare, Belgian, 30. 



Heape, 30. 

Heredity, collateral, 6. 

definition of, 6. 
Hyalodaphnia, 117. 

Jennings, 111. 
Johannsen, 45, 106. 

Little, 59. 

Maize, ear-length in, 138. 
Maturation of egg, 15, 20. 

of sperm-cells, 17. 
Mendel, 7, 34. 

Mendelian ratios, 35, 45, 50, 59. 
Mice, pale-colored, 81. 

pink-eyed, 79. 

spotted, 83. 

yellow, 57. 

Morgan, 91, 165, 174. 
Mus alexandrinus, 91. 
Mus norvegicus, 94. 
Mus rattus, 91, 94. 
Mutilations, inheritance of, 29. 

Nilsson-Ehle, 131. 

Parthenogenesis, artificial, 18. 

sex determination in, 162. 
Pigeons, reversion in, 62. 
Phenotype, 45. 
Phillips, 31. 
Prepotency, 101. 
Punnett, 159. 

Rats, 91. 

inbreeding in, 147. 

selection in, 123. 
Reduction, in fern, 22. 
Reversion, 62. 
Riddle, 87. 
Riley, 158. 

Rodents, coloration of, 72. 
Rotifers, sex determination in, 

Russo, 159. 

Schenck, 159. 
Self-fertilization, 145. 
Sheep, horns in, 102. 
Silk moths, 99. 
Snapdragon, golden variety of, 

Spermatogenesis, 17. 

of wasp, 24. 

Squash bug, sex determination 
in, 168. 

Tornier, 102. 
Transplantation of egg, 30. 

of ovary, 31. 
Treat, 158. 

Unit-characters, 38. 
De Vries, 34, 89, 106. 

Weismann, 29, 148. 
Wilson, 168. 
Woltereck, 117. 

Rabbits, size inheritance in, 

Zygote, definition of, 25. 




By DAVID STARR JORDAN, President of Leland Stan- 
ford Jr. University; VERNON LYMAN KELLOGG, Professor 
of Entomology; HAROLD HEATH, Assistant Professor of 
Invertebrate Zoology. 

Evolution and Animal Life. 

This is a popular discussion of the facts, processes, laws, and theories 
relating to the life and evolution of animals. The reader of it will have 
a very clear idea of the all-important theory of evolution as it has been 
developed and as it is held to-day by scientists. 8vo. Cloth, with about 
300 illustrations, $2.50 net ; postage 20 cents additional. 

Animal Studies. 

A compact but complete treatment of elementary zoology, especially 
prepared for institutions of learning that prefer to find in a single book 
an ecological as well as morphological survey of the animal world, i2mo. 
Cloth, $1.25 net. 

Animal Life. 

An elementary account of animal ecology that is, of the relations 
of animals to their surroundings. It treats of animals from the stand- 
point of the observer, and shows why the present conditions and habits 
of animal life are as we find them. I2mo. Cloth, $1.20 net. 

Animal Forms. 

This book deals in an elementary way with animal morphology. It 
describes the structure and life processes of animals, from the lowest 
creations to the highest and most complex. I2mo. Cloth, $1.10 net. 


This consists of " Animal Life " and " Animal Forms " bound in one 
volume. I2mo. Cloth, $1.80. 

Animal Structures. 

A laboratory guide in the teaching of elementary zoology. I2mo. 
Cloth, 50 cents net. 




Adolescence : Its Psychology and Its Relations 
to Physiology, Anthropology, Sociology, Sex, 
Crime, Religion, and Education. 

By G. STANLEY HALL, Ph.D., LL.D. Two vols., royal 
8vo, gilt top. Cloth, $7.50 net. 

This work is the result of many years of study and teaching. Tt is 
the first attempt in any language to bring together all the best that has 
been ascertained about the critical period of life which begins with 
puberty in the early teens and ends with maturity in the middle twenties, 
and it is made by the one man whose experience and ability pre-emi- 
nently qualify him for such a task. The work includes a summary of 
the author's conclusions after twenty-five years of teaching and study 
upon some of the most important themes in Philosophy, Psychology, 
Religion, and Education. 

The nature of the adolescent period is the best guide to education 
from the upper grades of the grammar school through the high school 
and college. Throughout, the statement of scientific facts is followed 
systematically by a consideration of their application to education, pe- 
nology, and other phases of life. 

Juvenile diseases and crime have each special chapters. The changes 
of each sense during this period are taken up. The study of normal 
psychic life is introduced by a chapter describing both typical and excep- 
tional adolescents, drawn from biography, literature, lives of the saints, 
and other sources. 

The practical applicauons of some of the conclusions of the scientific 
part are found in separate chapters on the education of girls, coeduca- 
tion and it> relations to marriage, fecundity, and family life, as seen by 
statistics in American colleges, with a sketch of an ideal education for 

Another chapter treats with some detail and criticism the various 
kinds and types of organization for adolescents from plays and games to 
the Y. M. C. A., Epworth League, and other associations devised for 
the young. 

The problem of the High School, its chief topics and methods, is 
considered from the standpoint of adolescence, and some very important 
modifications are urged. It closes with the general consideration of the 
relations of a higher to a lower civilization from this standpoint.