Heterosis

HETEROSIS

A record of researches directed toward explaining
and utilizing the vigor of hybrids

Edited by

JOHN W. GOWEN

Professor of Genetics
Iowa Stafe College

I tc l»

IOWA STATE COLLEGE PRESS

AMES • IOWA

Copyrlghi 1952 by The Iowa Sfate College Press

All rights reserved. Composed and printed in fhe

United States of America

Preface

Heterosis grew out of a desire on the part of Iowa State College to gather to-
gether research workers from marginal fields of science, each with something
to contribute to a discussion of a central problem of major national interest.
The problem of heterosis, as synonymous in large part with that of hybrid
vigor, formed a natural theme for discussion. As the reader will note, many
fields of science have contributed or stand to make significant contributions
to the subject. Major steps in the advance have led to divergent views which
may be rectified only through joint discussions followed by further research.
The conference of students of this problem was held June 15 to July 20, 1950.

In furnishing the opportunity for these discussions by active research
workers in the field, Iowa State College hoped: to facilitate summarization
and clarification of the accumulated data on the subject, to encourage formu-
lation and interpretation of the observations in the light of present day bio-
logical information, to stimulate further advances in the controlled success-
ful utilization and understanding of the biological processes behind the phe-
nomenon of heterosis, and to increase the service rendered by this discovery
in expanding world food supply.

Iowa has a direct, vested interest in heterosis. Today the agricultural
economy of the state is based upon hybrid corn. The scene portraying a hy-
bridization block of corn, shown here, is familiar to all who travel within the
state as well as to those in surrounding regions, for this method of corn
breeding has been shown to be surprisingly adaptable and useful in producing
more food per acre over wide areas of the world's agricultural lands.

Iowa's indebtedness to heterosis, generated through crossing selected and
repeatedly tested inbred strains, is well known. Few outside the workers in the
field realize the full magnitude of this debt.

With the progressive introduction of hybrid corn in 1936 there came a
steady increase in corn yields over both the former yields and over the
yields of other agricultural crops, as that of tame hay, which were not sub-
ject to this genetic method of yield improvement. It seems likely that in no
other period of like years has there been such an increase in food produced
over so many acres of land. The return from hybrid corn has been phenome-
nal, but it is now evidently approaching an asymptotic value. It behooves us
to find out as much as possible about the techniques and methods which

VI

PREFACE

made these advances possible. Even more we should determine what is going
on within the breeding and physiological systems through which heterosis
finds expression, if further increases in yields are to be obtained or better
systems of breeding are to be developed.

Toward this end the conference topics were arranged under four major

j^'u'X ''■*Je.Ai

mi

'.»*..

Controlled heterosis in the making through pollinations and fertilizations of selectively
purified genetic strains of corn (maize). (From G. F. Sprague.)

lA

z a:

E z
o

O V> -J

Z -10-

1 1,

4 O <t

K - Z

u >- e

> * .

100

/

'^'

^

90-

/

/

PERCENTAGE OF CORN ACREAGE
PLANTED WITH HYBRID SEED

80

/

/

70

/

'

1

60

1

i

1

/

\

-.

s.

/

\

/

\

50

f

-7]

/

ACRE YIELD
OF CORN

r\

/

\

/

\

^

40
30
20

/'

\

1

\

/

w

/

N.

\

1

\

/

I

\

/
/
/

--'

■ --

^-.

_.\

/

/

.-'

\

\
\

>

/

/
/

\

1
1
1

\

— "" "

"'^ ^.

1

ACRE YIELD OF
TAME HAY /

/^-

y'

\

1

10

/

/

/

0

_^ —

— -^

1930

1935

1940
YEARS

1945

1950

Trends in acre production of maize before and after heterosis was in use and tame hay
over which there has been no such breeding control 1930-1950. (From G. F. Sprague.)

PREFACE vii

headings. The early history and development of the heterosis concepts and
the cy to logical aspects of the problem occupied the first week. The contribu-
tions of physiology, evolution, and specific gene or cytoplasmic effects to the
vigor observed in hybrids were dealt with the second week. The third week's
meetings covered postulated gene interactions, as dominance, recombination,
and other possible gene effects. During the fourth week breeding systems and
methods of utilizing and evaluating heterosis effects were considered. In the
final week the students considered the problems that lie ahead and recent
methods of meeting them.

At each daily conference the speaker of the day presented a formal morn-
ing lecture covering his subject. In the afternoon, he led a conference session
on the subject of the morning lecture. At this time, all present had an oppor-
tunity to participate.

Accompanying, and as a supplement to the Heterosis Conference, a
Methods Workshop was held from July 3 to July 13. The Workshop was de-
voted to recent techniques for evaluating the kinds of data which occur fre-
quently in animal breeding experiments. Workshop meetings were organized
by Professor R. E. Comstock of North Carolina State College and Professor
Jay L. Lush of Iowa State College.

The meetings were led by men from several institutions besides Iowa
State College. Professors Oscar Kempthorne, Jay L. Lush, C. R. Henderson,
G. E. Dickerson, L. N. Hazel, F. H. Hull, A. E. Bell, A. M. Button, J. Bruce
Griffing, C. C. Cockerham, F. H. W. Morley, R. M. Koch, and A. L. Rae
contributed much to this phase of the program. It is with regret that it is
impossible to present the meat of the methods presented and developed in
the Workshop and the afternoon discussions. To many, this material con-
tributed much to the merit of the conference and the use to which the results
were put later.

In the field of worth-while living, as well as to see heterosis in operation,
conferees were guests, on various weekends, of three nearby companies putting
heterosis to the practical test of commercial seed stock production in crops
and live stock — the Ames Incross Company, the Farmers Hybrid Corn
Company, and the Pioneer Hi-Bred Corn Company.

Finally, the organization of the conference was the product of the joint
effort of the genetic group of Iowa State College. This group transcends all
departmental lines having as the common interest what goes on in inher-
itance. They were Jay L. Lush, G. F. Sprague, Oscar Kempthorne, S. S.
Chase, Janice Stadler, L. N. Hazel, A. W. Nordskog, Iver Johnson, W. A.
Craft, J. Bruce Griffing, and John W. Gowen.

In last analysis it was the interest of the audience and their participations
in the discussions that made the Conference worth while. The papers cover-
ing material presented by the leaders of these discussions follow.

Table of Contents

1. Early Ideas on Inbreeding and Crossbreeding .... conway zirkle 1

2. Beginnings of the Heterosis Concept . . . George Harrison shull 14

3. Development of the Heterosis Concept H. k. hayes 49

4. Preferential Segregation in Maize M. M. rhoades 66

5. Inbreeding and Crossbreeding in Seed Development . . . r. a. brink 81

6. Physiology of Gene Action in Hybrids w. Gordon whaley 98

7. Hybrid Nutritional Requirements william j. robbins 114

8. Origin of Corn Belt Maize and Its Genetic Significance

EDGAR ANDERSON and WILLIAM L. BROWN 124

9. Heterosis in Population Genetics . Adrian o a. buzzati-tra verso 149

10. Fi.xing Transgressive Vigor in Nicotiana Rustica . . harold h. smith 161

11. Hybridization in the Evolution of Maize . . . paul c. mangelsdorf 175

12. Biochemical Models of Heterosis in Neurospora . . sterling emerson 199

13. Nature and Origin of Heterosis th. dobzhansky 218

14. Plasmagenes and Chromogenes in Heterosis . . . donald f. jones 224

15. Specificity of Gene Effects m. r. irwin 236

16. Genetics and Cytology of Saccharomyces .... carl c. lindegren 256

17. Genetic Implications of Mutations in S. Typhiniurium . . h. h. plough 267

18. Dominance and Overdominance james f. crow 282

19. Gene Recombination and Heterosis leroy powers 298

20. Gene Interaction in Heterosis A. j. mangelsdorf 320

21. Inbred Lines for Heterosis Tests? Gordon e. dickerson 3vS0

22. Specific and General Combining Ability .... c. R. Henderson 352

23. Rotational Crossbreeding and Heterosis L. m. winters 371

24. Gamete Selection for Specific Combining Ability

e. l. pinnell, e. h. rinke, and h. k. hayes 378

25. Monoploids in Maize sherret s. chase 389

26. Early Testing and Recurrent Selection g. F. sprague 400

27. Heterosis in a New Population E. j. wellhausen 418

28. Recurrent Selection and Overdominance fred H. hull 451

29. Hybrid Vigor in Drosophila John w. gowen 474

30. Estimation of Average Dominance of Genes

R. e. comstock and h. f. robinson 494

Bibliography 517

Index 537

IX

67867

CONWAY ZIRKLE

University of Pennsylvania

OGIC^^

Chapter 1

Early Ideas on Inbreeding
and Crossbreeding

In tracing the historical background of a great scientific advance or dis-
covery, the historian nearly always has the opportunity of showing that the
scientists who receive the credit for the work are really late-comers to the
field, and that all the basic principles and facts were known much earlier.
Finding these earlier records is always something of a pleasure; comparable,
perhaps, to the pleasure a systematist experiences in extending the range
of some well known species.

The historian may be tempted, in consequence, to emphasize these earlier
contributions a little too strongly and to re-assign the credits for the scientific
advances which have been made. In the present state of the history of sci-
ence, it requires only a little searching of the records to discover contributions
which have been overlooked and which are very pertinent to the advance
in question. This wealth of data, which accumulates almost automatically,
seems to deserve emphasis. But great steps forward generally are made
not by the discovery of new facts, important as they are, or by new ideas,
brilliant as they may be, but by the organization of existing data in such
a way that hitherto unperceived relationships are revealed, and by incor-
porating the pertinent data into the general body of knowledge so that new,
basic principles emerge.

For example, even so monumental a work as Darwin's Origin of Species
contains few facts, observations or even ideas which had not been known
for a long time. The work of many pre-Darwinians now appears important,
especially after Darwin's synthesis had shown its significance. Of course,
this does not belittle Darwin in the slightest. It only illustrates the way
science grows.

The emergence of the scientific basis of heterosis or hybrid vigor is no

1

2 CONWAY ZIRKLE

exception. Practically all of its factual background was reported before
Mendel's great contribution was discovered. Even workable methods for
utilizing hybrid vigor in crop production were known, but it was not until
the classic post-Mendelian investigations of Shull, East, and Jones were
completed, that heterosis took its proper place in genetics. The following
discussion of the importance of heterosis will be confined to its pre-Men-
delian background.

Heterosis can be described as a special instance of the general principles
involved in inbreeding and outbreeding. To fit it into its proper niche, we will
trace first the evolution of our ideas on the effects of these two contrasting
types of mating. Since our earliest breeding records seem limited to those
of human beings and primitive deities, we will start with the breeding
records of these two forms.

Hybrid vigor has been recognized in a great many plants during the
last two hundred years. We will therefore describe briefly what was known
of its influence on these plants. Because heterosis has reached its greatest
development in Zea mays, we will trace briefly the pre-Mendelian genetics
of this plant, and show how the facts were discovered which have been of
such great scientific and economic importance.

The ill effects of too-close inbreeding have been known for a long time.
Indeed, Charles Darwin (1868) believed that natural selection had pro-
duced in us an instinct against incest, and was effective in developing this
instinct because of the greater survival value of the more vigorous offspring
of exogamous matings. One of his contemporaries, Tylor (1865), noted that
many savage tribes had tabooed the marriage of near relatives, and he
assumed that they had done so because they had noticed the ill effects of
inbreeding. The Greeks looked upon certain marriages between near rela-
tives as crimes. This has been known almost universally ever since Freud
popularized the tragedy of King Oedipus. At present, we outlaw close in-
breeding in man, and our custom is scientifically sound.

We are apt to be mistaken, however, if we read into the standards of our
distant preceptors the factual knowledge which we have today. The in-
tellectual ancestors of European civilization approved of inbreeding and
actually practiced it on supposedly eugenic grounds. The fact that their
genetics was unsound and their eugenic notions impractical is irrelevant.
They had their ideals, they were conscientious and they did their duties.
The Pharaohs married their own sisters when possible so that their god-
like blood would not be diluted. Marriage between half brother and sister
was common in other royal families of the period. Actually, as we shall see,
the two great pillars of European thought, Hebrew morality and Greek
philosophy, endorsed inbreeding as a matter-of-course.

The Hebrews, who derived mankind from a single pair, were compelled
to assume that the first men born had to marry their sisters — as there were

EARLY IDEAS ON INBREEDING AND CROSSBREEDING 3

then no other women on the earth. Indeed Adam and Eve themselves were not
entirely unrelated. The marrying of a sister was obviously respectable, and
it seems to have occurred routinely among the Hebrews and their ancestors
for several thousand years. Abraham's wife, Sarah, was also his sister. At
times even closer inbreeding took place. Abraham's nephew, Lot, impreg-
nated his own two daughters. The latter instances occurred, however, under
exceptional circumstances — and Lot was drunk. But as late as the time of
King David, brother-sister marriages took place. The imbroglio between
David's children, Tamar, Ammon, and Absalom, shows that a legal mar-
riage between half-brother and sister would then have been a routine oc-
currence.

The Greeks also could hardly have had scruples against inbreeding, as
evinced by the pedigrees they invented for their gods. Their theogony shows
many instances of the closest inbreeding possible for either animals or gods
in which the sexes are separate. Zeus, the great father of the gods, married
his sister, Hera. Their parents, Kronos and Rhea, also were brother and
sister, and were in turn descended from Ouranos and Gaea, again brother
and sister. Thus the legitimate offspring of Zeus — Hebe, Ares, and Hephaes-
tus— were the products of three generations of brother-sister mating.
Moreover, the pedigrees of the Greek heroes show an amount of inbreeding
comparable to that in our modern stud books for race horses. They were
all related in one way or another and related to the gods in many ways. A
single example will be cited. Zeus was the father of Herakles and also his
great-great-grandfather on his mother's side. Herakles' great-great-grand-
mother, Danae, who had found such favor in the eyes of Zeus, was herself
descended from Zeus through two different lines. With immortals, back-
crossing offered no real problems.

East and Jones (1919) have pointed out that close inbreeding was com-
mon among the Athenians even at the height of their civilization. These
scientists were of the opinion that most of the freemen in Attica were
rather closely related to each other. Marriage between half brother and
sister was permitted, and marriage between uncle and niece fairly common.
A Grecian heiress was nearly always taken as a wife by one of her kinsmen
so that her property would not be lost to the family. Common as inbreeding
was during the flowering of Greek culture, it was as nothing compared with
the inbreeding which occurred in the period after the Trojan War and before
the true historical period. In this intervening time, Greece was divided into
innumerable independent political units, many of them minute. One island
six miles long and two miles wide contained three separate kingdoms.
Political boundaries as well as bays, mountains, and seas were functional,
isolating mechanisms; and the Greeks were separated into many small
breeding units for fifteen to twenty generations. Isolation was never com-
plete, however, and there were enough wandering heroes to supply some

4 CONWAY ZIRKLE

genie migration. There were also some mass migrations and amalgamations
of different tribes. The general situation was startlingly close to the condi-
tions which Sewall Wright (1931) describes as the optimum for rapid
evolution.

We may be tempted to explain as cause and effect what may be only an
accidental relationship in time; and, while recognizing that it is far fetched,
to ascribe the sudden appearance of what Galton called the ablest race in
history to the ideal conditions for evolution which their ancestors had. We
would also like to consider, as the necessary preliminary to the hybrid vigor,
that period of inbreeding which preceded the flowering of Grecian culture.
This hybrid vigor we would like to recognize as an important factor in the
production of the great geniuses who flourished in the later, larger city
states of Greece.

So much for the classical attitude toward endogamy. It slowly changed,
and exogamy which had always existed became the exclusive custom. At
the time of Sophocles, all forms of inbreeding were not considered ethical
and pleasing to the gods. The sin of Oedipus lay in his having made a for-
bidden backcross rather than in mere inbreeding which was lawful. We do
not find any records of degeneracy appearing in his children — indeed his
daughter Antigone was a model of feminine virtue. It seems that close
human inbreeding came to an end without its ill effects ever having been
recognized.

The Nordics also were unaware of any degeneracy inherent in inbreeding.
Their great god Wotan included a bit of inbreeding in his plan for creating
a fearless hero who could save even the gods themselves from their im-
pending fate. Wotan started the chain reaction by begetting Siegmund and
Sieglinde, twin brother and sister. The twins were separated in infancy.
They met again as adults and, recognizing their relationship, had an il-
legitimate affair — begetting the hero Siegfried. Although Siegfried was not
exactly an intellectual type, he was certainly not a degenerate — represent-
ing rather the ideal male of a somewhat primitive culture.

As the centuries passed, incest was extended to cover brother-sister
mating, even when the parties involved were unaware of their relationship.
There is no need to cite here the many examples of the later tragedies based
upon this plot. It soon became an almost universally accepted standard in
literature, from epics to novels. The luckless Finnish hero, Kullervo {The
Kalevala, Rune XXXV), thus brought disaster to his family by seducing his
sister unknowingly. Defoe's long suffering heroine Moll Flanders (1722)
had to abandon an apparently successful marriage when she discovered that
her husband was her brother. On the other hand, as late as 1819, Lord
Byron defended brother-sister marriage passionately in his drama Cain —
but this was a scandalous exceptioii to the rule. The marriage of kin nearer

EARLY IDEAS ON INBREEDING AND CROSSBREEDING 5

than first cousins had become legally and morally taboo. Perhaps we may
follow Westermarck in assuming that endogamy became passe, not because
its biological ill effects were recognized, but because men knew their kins-
women too well to marry one of them if they could possibly get a wife
elsewhere.

It is possible that we have thus far paid too much attention to inbreeding
and outbreeding in man. Our excuse is that there are almost no other records
of inbreeding from classical times. There are no plant records, of course, for
sex in plants was not understood in spite of the general practices of caprifica-
tion and hand pollination of the date palm. Records of inbreeding and out-
crossing in domestic animals are almost completely lacking even in the
copious agricultural literature of the Romans. Aristotle's History of Animals.
576al5 (Thompson 1910) does state that horses will cover both their mothers
and their daughters ". . . and, indeed, a troup of horses is only considered
perfect when such promiscuity of intercourse occurs" — but he seems to
be almost alone in referring to the subject. Later on in the same book
(630b30) he cited a happening which we quote.

The male camel declines intercourse with its mother; if his keeper tries compulsion, he
evinces disinclination. On one occasion, when intercourse was being declined by the
young male, the keeper covered over the mother and put the young male to her; but, when
after the intercourse the wrapping had been removed, though the operation was completed
and could not be revoked, still by and by he bit his keeper to death. A story goes that the
king of Scythia had a highly-bred mare, and that all her foals were splendid; that wishing
to mate the best of the young males with the mother, he had him brought to the stall for
the purpose; that the young horse declined; that, after the mother's head had been con-
cealed in a wrapper he, in ignorance, had intercourse; and that, when immediately after-
wards the wrapper was removed and the head of the mare was rendered visible, the young
horse ran away and hurled himself down a precipice.

This behavior of the stallion was considered so remarkable that it was
described by Aelian, Antigonus, Heirocles, Oppian, Pliny, and Varro.
Varro confused the tradition and made the horse bite his keeper to death.

It is fairly safe for us to assume that in both classical and medieval times
the flocks and herds were greatly inbred. Transportation difficulties would
have insured inbreeding unless its evil effects were realized, and we have at
least negative evidence that they were not. Varro, who gave many detailed
directions for the breeding of all domestic animals, does not even mention
the question of kinship between sire and dam. We do have an interesting
literary allusion by Ovid, however, to the routine inbreeding of domestic
animals in his account of the incest of Myrrha in the tenth book of the Meta-
morphoses. The affair between Myrrha and her father Cinyras was like that
of Oedipus and his mother Jocasta. The fates had decreed that Myrrha
should become the mistress of her father. Torn by her unholy desires she
debates the matter with her conscience. Her better nature argues (From
the metrical translation of Brookes More, 1922):

6 CONWAY ZIRKLE

But what more could be asked for, by the most
Depraved? Think of the many sacred ties
And loved names, you are dragging to the mire;
The rival of your mother, will you be
The mistress of your father, and be named
The sister of your son, and make yourself
The mother of your brother?

In Stating the other side of the case Myrrha describes the "natural" in-
breeding of animals.

A crime so great — If it indeed is crime.

I am not sure it is — I have not heard

That any God or written law condemns

The union of a parent and his child.

All animals will mate as they desire —

A heifer may endure her sire, and who

Condemns it? And the happy stud is not

Refused by his mare-daughters: the he-goat

Consorts unthought-of with the flock of which

He is the father; and the birds conceive

Of those from whom they were themselves begot.

Happy are they who have such privilege!

Malignant men have given spiteful laws;

And what is right to Nature is decreed

Unnatural, by jealous laws of men.

But it is said there are some tribes today.
In which the mother marries her own son;
The daughter takes her father; and by this,
The love kind nature gives them is increased
Into a double bond. — Ah wretched me!

The debate ends as we would expect, and in due course Myrrha is de-
livered of an infant boy who certainly showed none of the ill effects of the in-
breeding which produced him. He grew up to be quite an Adonis. In fact
he was Adonis.

We can profitably skip to the late eighteenth century before we pursue
further the matter of inbreeding. This was the period when Bakewell was
emphasizing the importance of breeding in improving farm animals, when
the various purebreds were beginning to emerge, and when the efficacy of
artificial selection was beginning to be understood.

By the beginning of the nineteenth century, practical attempts to im-
prove the different breeds of cattle led to intensive inbreeding. A prize bull
would be bred to his own daughters and granddaughters. At first, the breed-
ers seemed to believe that a selection of the very best individuals followed
by intensive inbreeding was the quickest method for improving the stock.
On theoretical grounds this seemed to be the case, and great advances
were actually made by this method — but sooner or later something always

EARLY IDEAS ON INBREEDING AND CROSSBREEDING 7

happened. The inbred stock seemed to grow sterile, but vigor could be re-
established by outcrossing. The actual cause of degeneracy in the inbreds
was not understood until Mendelian inheritance was discovered, but the
remedial procedures of the practical breeders could hardly have been im-
proved on. We owe to them the basis of our finest stocks. They inbred to
add up and concentrate desirable qualities and then crossbred to prevent
degeneration, then inbred again and crossed again, all the time selecting
their breeding stocks most carefully. Charles Darwin (1868) described this
process most accurately and listed the pertinent publications.

There was a striking divergence in this work between theory and prac-
tice, which is just as well, as the only theories available at the time were in-
adequate. Those breeders who held that inbreeding was the suninmm bonum
did not hesitate to crossbreed when the occasion demanded, and those who
emphasized the virtues of hybridization inbred whenever inbreeding gave
them the opportunity of adding up desirable qualities. Darwin, himself,
stated, "Although free crossing is a danger on the one side which everyone
can see, too close inbreeding is a hidden danger on the other." We await
the twentieth century for a real improvement in breeding methods.

The first plant hybrid was described as such in 1716, and during the next
forty-five years many descriptions of hybrid plants were published. Some
attempts were even made to produce new varieties, but in retrospect the
work seems somewhat dilettante.

From 1761 to 1766, Josef Gottlieb Koelreuter (1766) published the several
parts of his well-known classic, and plant hybridization was put upon a
different and more scientific basis. His investigation of hybridization was
intensive, systematic, and scientific. He described, among other things,
hybrid vigor in interspecific crosses in Nicotiana, Dianthus, Verbascum,
Mirabilis, Datura, and other genera (East and Jones, 1919). He also observed
floral mechanisms which insured cross pollination and assumed in conse-
quence that nature had designed plants to benefit from crossbreeding. It is
worth emphasizing that hybrid vigor in plants was first described by the
person who first investigated plant hybrids in detail. Koelreuter continued
to publish papers on plant hybrids until the early nineteenth century.

Meanwhile other contributions had been made to our knowledge of the
effects of outcrossing and the mechanism for securing it. In 1793, Sprengel
depicted the structure of flowers in great and accurate detail, and showed
how self pollination was generally avoided. In 1799, Thomas Andrew Knight
described hybrid vigor as a normal consequence of crossing varieties and
developed from this his principle of anti-inbreeding. Other hybridizers
noted the exceptional vigor of many of their creations. Indeed, hybrid
vigor in plants was becoming a commonplace. Among the botanists who
recorded this vigor were: Mauz (1825), Sageret (1826), BerthoUet (1827),
Wiegmann (1828), Herbert (1837), and Lecoq (1845). Gartner (1849) was

8 CONWAY ZIRKLE

especially struck by the vegetative luxuriance, root development, height,
number of flowers and hardiness of many of his hybrids.

Naudin (1865) found hybrid vigor in twenty-four species crosses out of the
thirty-five which he made within eleven genera. In Datura his results were
spectacular. In reciprocal crosses between D. Stramonium and D. Tatula
the offspring were twice the height of the parents. Knowledge of plant
hybridization was increasing more rapidly at this time than the biologists
knew, for this was the year in which Mendel's (1865) paper Versuche iiber
Pflanzen-Hyhriden appeared. Mendel discovered hybrid vigor in his pea
hybrids and described it as follows:

The longer of the two parental stems is usually exceeded by the hybrid, a fact which is
possibly only attributable to the greater luxuriance which appears in all parts of the
plants when stems of very different lengths are crossed. Thus, for instance, in repeated
experiments, stems of 1 ft. and 6 ft. in length yielded without exception hybrids which
varied in length between 6 ft. and 1\ ft.

We shall cite but one more scientist who wrote on the general subject of
hybrid vigor in plants. This is Charles Darwin, whose Cross and Self Fertiliza-
tion in the Vegetable Kingdom appeared in 1876. This was a book of great
importance and influence, but no attempt will be made here to summarize
this work of nearly five hundred pages. At the beginning of his concluding
chapter, Darwin stated:

The first and most important conclusion which may be drawn from the observations
given in this volume, is that cross-fertilization is generally beneficial and self-fertihzation
injurious.

There is a special reason why this book of Darwin's is of such great
importance for any historical background to heterosis. Darwin worked
carefully and quantitatively with many genera, including Zea mays. He
measured accurately the amount of hybrid vigor he could induce, and he pub-
lished his data in full. His work stands in the direct ancestral line to the
twentieth century research on the subject, and the great advances made
from 1908 to 1919 are based solidly on this work. There are no great gaps
in the steady progress and no gaps in the literature.

Zea mays was brought to Europe in 1493 by Columbus on his home-
ward voyage. This was sometime before the great herbals were written,
so our first descriptions of the new grain are to be found in the books of the
travelers and explorers. Later, Indian corn appeared under various names
in the early herbals, and it was described in detail in the famous Krautehuch
of Tabernaemontanus, first published in 1588. The author obviously yielded
to his enthusiasm in devoting five and a half folio pages to corn and includ-
ing thirteen illustrations in his treatment. He was the first to describe the
results of xenia — the occurrence of difi'erent colored grains on the same ear —
but his explanation of the phenomenon has nothing to do with cross pollina-
tion. He ascribed it directly to God Almighty.

EARLY IDEAS ON INBREEDING AND CROSSBREEDING 9

And one sees an especially great and wonderful mystery in these spikes, Gott der Ilerr,
through the medium of nature which must serve everyone, disports himself and performs
wonders in his works and so notably in the case of this plant that we must rightly be
amazed and should learn to know the One True Eternal God even from his creatures alone.
For some of the spikes of this plant, together with their fruit, are quite white, brown and
blue intermixed. Thus, some rows are half white, a second series brown and the third blue;
and some grains, accordingly are mixed with each other and transposed. Again, sometimes
one, two, or three rows are white, the next rows blue, then again white and after that
chestnut-brown; that is, they are interchanged on one row and run straight through on
another. Some spikes and their grains are entirely yellow, others entirely brown, some are
white, brown, and blue, others violet, white, black, and brown: of these the white and
blue are prettily sprinkled with small dots, as if they had been artistically colored in this
way by a painter. Some are red, black, and brown, with sometimes one color next to the
other, while at other times two, three, even four colors, more or less, are found one next
to another in this way.

During the next century and a half, many other descriptions of the
occurrence of different colored grains on a single ear were published. I have
found about forty of them and there are doubtless many more. The earliest
correct interpretation of this phenomenon had to await the eighteenth cen-
tury and is contained in a letter written by Cotton Mather in 1716. Here
the different colored grains occurring together on an ear are ascribed to a
wind-born intermixture of varieties. This letter is the first record we have of
plant hybridization, and antedates Fairchild's description of a Dianthus
hybrid by one year. In 1724, Paul Dudley also described hybridization in
maize, and he was able to eliminate one of the hypotheses which had been
used to explain the mixture. As a broad ditch of water lay between the mix-
ing varieties, he could show that the mixed colors were not due to the root-
lets of different strains fusing underground, a view held at the time by
many New Englanders, both white and red.

Hybridization in maize was described again in 1745 by Benjamin Cooke,
in 1750 by the great Swedish traveler and naturalist, Pehr Kalm, and in
1751 by William Douglass. By the early nineteenth century, knowledge of
plant hybrids was widespread. Plant hybridization was becoming a routine
practice, and there is little doubt that different varieties of maize were
crossed many times by American farmers who did not record their breeding
experiments in writing.

Brown and Anderson (1947, 1948) have recently shown that the modern
races now grown in the corn belt are derived from both the northern flint
and the southern dent varieties. Hybridization in corn was easy to perform
and the results were easy to recognize. The intermixtures of colors were so
spectacular that they were frequently described, by Gallesio (1806), Burger
(1808), Sageret (1826), Gartner (1827), and others.

We detour briefly here into some of the technical aspects of xenia. Double
fertilization and the mixed nature of the endosperm were discovered by Na-
waschin in 1899. In 1881, Focke introduced the term xenia but he used it
to include what we now call melaxenia. Focke collected from the literature

10 CONWAY ZIRKLE

many supposed instances where the pollen influenced directly the color
and form of the flowers, the flavor and shape of the fruits, and the color
and content of the seeds. How many of these cases were really due to Men-
delian segregation we will probably never know, since the investigators did
not know enough to take proper precautions.

We can, however, divide the history of true xenia into three periods:
first, when its visible effect was considered a lusus naturae (1588); second,
when it was known to be caused by foreign pollen (1716); and third, when
the embryo and endosperm were recognized as two different structures and
when the influence of the pollen upon the latter was recorded specifically.
In the paragraph on Zea in the section on xenia, Focke cites the work of
Vilmorin (1867), Hildebrand (1868), and Kornicke (1876), who described
the effect of pollen on the endosperm.

We should note a brief comment on the subject which has been overlooked
and is earlier than the papers cited by Focke. In 1858, Asa Gray described
xenia in maize. He reported starchy grains in ears of sweet corn and many
different kinds and colors of grains on the same ear. He had two explana-
tions for this occurrence: (1) cross pollination of the previous year and (2)
direct action of the pollen on the ovules of the present year. It is obvious
that by ovules he did not mean embryos. This may be the earliest authentic
recognition of the real problem of xenia.

In reviewing the nineteenth century records of hybrid vigor in Zea mays,
we start with those of Charles Darwin (1876). Darwin planned his experi-
ments most carefully. He crossed and selfed plants from the same stock, and
raised fifteen plants from each of the two types of seed he had obtained.
He planted the seed from both the selfed and crossed plants in the same
pots, from six to ten plants per pot. When the plants were between one and
two feet in height, he measured them and found that the average height of
the plants from the selfed seed was 17.57 inches, while that from the crossed
seed was 20.19 inches or a ratio of 81 to 100. When mature, the two lots
averaged 61.59 inches and 66.51 inches, respectively, a ratio of 93 to 100.
In another experiment when the corn was planted in the ground, the ratio
of the selfed to the crossed was 80 to 100. Darwin called in his cousin, Francis
Galton, to check his results and Galton judged them to be very good after
he had studied the curves that he drew.

The direct connection between Darwin's work and our present hybrid
corn is shown by Darwin's influence on W^ J. Beal who was the real leader
in the American research designed to improve maize. Beal reviewed Dar-
win's book in 1878, and even wrote an article which was little more than a
paraphrase of what Darwin had published. Beal's own contributions ap-
peared a little later.

In 1880, Beal described how he had increased the yield of corn on a large
scale. Two stocks of the same type of corn which had been grown a hundred

EARLY IDEAS ON INBREEDING AND CROSSBREEDING 11

miles apart for a number of years were planted together in alternate rows.
All of one stock grown in this field was detasseled and thus it could not be
self fertilized but could produce only hybrid seed. The tasseled stalks of the
other lot would still be pure bred as there was no foreign pollen to contami-
nate their ears and they could again serve as a parent to a hybrid. A small
amount of the first parental stock which furnished the detasseled stalks was
grown apart for future hybridization. The hybrid seed was planted, and
produced the main crop. Beal increased his yield by this method by as
much as 151 exceeds 100. This method and these results, it should be
emphasized, were published in 1880.

E. Lewis Sturtevant, the first director of the New York Agricultural
Experiment Station, made a number of studies of corn hybrids starting in
1882. His findings are interesting and important but not directly applicable
to heterosis. Singleton (1935) has called attention to this work and to the
excellent genetic research which the western corn breeders were carrying on
at this time — such geneticists as W. A. Kellerman, W. T. Swingle, and
Willet M. Hays. They anticipated many of Mendel's findings and described
dominance, the reappearance of recessives (atavisms), and even Mendelian
ratios such as 1 to 1 and 3 to 1. They were all concerned with practical
results. Hays (1889), in particular, tried to synthesize superior breeds of
corn by hybridizing controlled varieties.

Sanborn (1890) confirmed Beal's results and reported that his own
hybrid corn yielded in the ratio of 131 to 100 for his inbred. He also fol-
lowed Real's method of planting his parental stocks in alternate rows and of
detasseling one of them. He made an additional observation which we know
now is important:

It is this outcrossed seed which will give the great crops for the next year. It will be
noted that I gained twelve bushels i)er acre by using crossed seed. The operation is simple
and almost costless and will pay one hundred fold for the cost involved. The cross must be
made every year using nen' seed, the product of the outcross of two pure seed. (Italics C. Z.)

If our farmers had known of this discovery reported in 1890 they might
not have tried to use their own hybrid corn as seed.

Singleton (1941) also called attention to a pre-Mendelian interpretation
of hybrid vigor by Johnson (1891) which, in the light of our present knowl-
edge, deserves more than passing notice. We can state it in Johnson's own

words :

That crossing commonly gives better offspring than in-and-in breeding is due to the
fact that in the latter both parents are likely to possess by inheritance the same imperfec-
tions which are thus intensified in the progeny, while in cross breeding the parents more
usually have different imperfections, which often, more or less, compensate each other in
the immediate descendants.

We come next to a j)ublication of G. W. McClure (1892). This paper is
deservedly famous, and its many contributions are incorporated into our
modern genetics literature. Here we shall cite only the observations which
pertain to heterosis. McClure noted (1) that sterility and deformity often

12 CONWAY ZIRKLE

follow selfing, (2) that crossing imparts vigor, (3) that it is impossible
to tell in advance what varieties will produce corn of increased size when
crossed, (4) that what appears to be the best ear does not always produce
the largest crops, and (5) nearly all of the hybrid corn grown a second year is
smaller than that grown the first year, though most of it is yet larger than
the average size of the parent varieties.

McClure also called attention to the fact that our fine varieties of fruits
have to be propagated vegetatively, and hinted that the deteriorations of
the seedlings from fruit trees was not unrelated to a like deterioration which
occurred in the seedlings grown from hybrid corn.

The year following McClure's publication. Morrow and Gardiner (1893)
recorded some very pertinent facts they had discovered as a result of their
field experiments with corn. They reported that, "In every instance the
yield from the cross is greater than the average from the parent varieties:
the average increase per acre from the five crosses [they had made] being
nine and a half bushels." They noted further in a paper published later the
same year that, "It seems that cross bred corn gives larger yields at least
for the first and second years after crossing than an average of the parent
varieties, but how long this greater fruitfulness will last is undetermined."
Gardiner continued the work and in 1895 published the data he obtained
by repeating the experiments. He found that in four of six cases the yield
was greater in the cross, the average being twelve bushels per acre.

We now come to the great corn breeding research project which was
undertaken at the University of Illinois in 1895 by Eugene Davenport
and P. G. Holden. Both of these scientists had been students of Beal and
were interested in his work on inbreeding and cross breeding maize. We
are indebted to Professor Holden for an account of this work which he printed
privately in 1948. This account gives us valuable historic data not to be
found elsewhere, as most of the University of Illinois records were destroyed
by fire.

An intensive series of inbreeding experiments was undertaken by Holden,
and later on the inbred lines were crossed. Hybrid vigor was noted, and it
was found in addition that the crosses between different inbred lines differed
widely in their yield and in their general desirability. The main purpose of
the experiments was to find out how to use controlled crossing early and
effectively. After Holden left Illinois in 1900, the project was taken over by
C. G. Hopkins, a chemist, who was interested in increasing the protein con-
tent of maize. He hired as his assistant in 1900 a young chemist named
Edward Murray East, whom we shall hear about later.

Our account of the background of heterosis is coming to an end as the
beginning of the twentieth century makes a logical stopping point. We should
mention, however, the great hybrid vigor discovered by Webber (1900)
when he crossed a Peruvian corn, Cuzco, with a native variety, Hickory

EARLY IDEAS ON INBREEDING AND CROSSBREEDING 13

King. The average height of the parental stocks was 8 feet 3 inches while
the cross averaged 12 feet 4 inches, an increase of 4 feet 1 inch.

The next year Webber (1901) called attention to the marked loss of vigor
in corn from inbreeding. From 100 stalks of selfed corn he obtained 46
ears weighing 9.33 pounds, while from 100 stalks obtained from crossing
different seedlings he obtained 82 ears weighing 27.5 pounds. When he
attempted to "fix" his Cuzco-Hickory King hybrid by selling he got a great
loss of vigor and almost complete sterility, but when he crossed the different
seedlings there was little loss of vigor. He concluded that to fix hybrids
one should not self the plants.

In 1900, the discovery of Mendel's long-forgotten paper was announced.
Both Hugo de Vries and C. Correns, two of the three discoverers of Mendel,
published papers on Zea mays and all future work on Indian Corn was on a
somewhat different level.

SUMMARY OF KNOWLEDGE OF HYBRID VIGOR AT
BEGINNING OF 20th CENTURY

1. Inbreeding reduces vigor and produces many defective and sterile indi-
viduals which automatically discard themselves.

2. Cross breeding greatly increases vigor both in interspecific and inter- <
varietal hybrids. Crossing two inbred stocks restores the lost vigor and
frequently produces more vigor than the stocks had originally.

3. All inbred stocks do not produce the same amount of vigor when crossed. ^
Certain crosses are far more effective than others.

4. The simplest method of hybridizing Zea on a large scale is to plant two
stocks in alternate rows and to detassel one stock. The hybrid corn grown
from the detasseled stock produces the great yields.

5. Hybridization must be secured each generation if the yield is to be kept
up, although a second generation of open pollinated corn may still be
better than the original parental stocks.

6. In inbreeding, both parents are apt to have the same defects which are V
intensified in the offspring. The cause of hybrid vigor is that in crosses
the parents usually have different defects which tend to compensate for
each other in the immediate progeny.

7. The fact that hybrid vigor in Zea is not permanent but decreases if the
hybrids are open-pollinated, seems to be related to the fact that fruit
trees, whose desirable qualities are preserved by vegetative propagation,
produce seedlings which are inferior.

GEORGE HARRISON SHULL

Princefon University

Chapter 2

Beginnings of
the Heterosis Concept

The heterosis concept was first definitely recognized in the work with hybrid
corn. Before attempting to define this concept, however, we will take a brief
look at some of the observations of early workers which indicated the prob-
able presence of heterosis, and where recognition of heterosis as an important
biological principle might have been expected.

The first hybridizer of plants, Dr. J. G. Koelreuter, noted some impres-
sive examples of excessive luxuriance in his Nicotiana hybrids. These were
isolated observations which suggested no theory as to why these hybrids
should exceed their parents in size and general vigor. Koelreuter cannot be
said to have had a heterosis concept. Probably every conscious producer of
hybrids since Koelreuter's time has made similar observations of the exces-
sive vigor of some hybrids over their parents, so that such hybrid vigor has
ceased to cause surprise. But the general acceptance of hybrid vigor as a nor-
mal phenomenon did not estabUsh a heterosis concept. It was merely the
summational effect of oft-repeated experience.

Thomas Andrew Knight noted the deterioration of some of the old stand-
ard horticultural varieties, and concluded that such varieties have a natural
life-span and gradually decline as the result of advancing senility. He saw
that such decline makes it necessary to develop new varieties which will start
off with the vigor of youth. Although Knight himself produced many such
new varieties, some of which were produced by hybridization, it is not ap-
parent that he thought of hybridization as an agency for the production of
such new vigor. Although he advanced a theory concerning physiological
vigor and its decline, he did not recognize the heterosis concept.

Luther Burbank also produced numerous varieties, often following inten-

14

BEGINNINGS OF THE HETEROSIS CONCEPT 15

tional hybridizations, and it is easy to recognize heterosis as a potent factor
in the remarkable values displayed by many of these new varieties. But
while Burbank made great use of hybridizations in his plant breeding work,
he did not recognize hybridization, as such, as the source of the large size
and remarkable vigor of his new varieties. For him the role of hybridization,
aside from the bringing together of desirable qualities possessed separately
by the two chosen parents, was merely the "breaking of the types." In this
way the variability in subsequent generations was greatly increased, thus
enlarging the range of forms from among which to select the most desirable
for recognition as New Creations.

There are many other important observations and philosophical considera-
tions that bear a close relationship to our current understanding of heterosis,
and which antedated the recognition of heterosis. It would take us too far
afield, however, to discuss these related observations at length. We can
make only this passing reference to the highly significant work of Charles
Darwin in demonstrating that cross-fertilization results, in many cases, in
increased size, vigor, and productiveness as compared with self-fertilization
or with other close inbreeding within the same species.

Darwin did not recognize this increased vigor as identical with hybrid
vigor, nor specifically attribute it to the differences between the uniting
gametes. To him it only demonstrated a method which would inevitably
preserve by natural selection any variation that might occur — whether me-
chanical or physiological — which would make cross-fertilization more likely
or even an obligate method of reproduction. With heterosis established as a
recognized pattern of behavior, or type of explanation, we can now interpret
Darwin's demonstrated superiority of crossbreds as examples of the occur-
rence of heterosis. We may go even further and include the whole field of
sexual reproduction in showing the advantages of heterosis. These result
from the union of two cells — the egg and the sperm — extremely difTerentiated
physiologically, and in all dioecious organisms also dififerentiated genetically.

Let us briefly consider several investigations which foreshadowed the
procedures now used in growing hybrid corn — for somewhere in the course
of this work with corn the heterosis principle was first definitely recognized.

Two techniques are characteristically associated with the work of the
"hybrid-corn makers." Uncritical commentators have mistakenly considered
these techniques synonymous with the development of the hybrid-corn pro-
gram itself. These are (a) cross-pollination by interplanting two different
lines or varieties, and the detasseling of one of these lines which then sup-
plies the seed to be planted; and {b) controlled self-pollination.

In deciding what part these two methods played in the develoi)mcnt of the
heterosis concept, we must first consider why these methods were used by
various workers and how their use affected the experimental conclusions.

Dr. William J. Beal, of Michigan Agricultural College, apparently was

16 GEORGE HARRISON SHULL

the first to make extensive use of controlled cross-pollination in the breed-
ing of corn. Beal was a student of Asa Gray from 1862 to 1865, when the
latter was in active correspondence with Charles Darwin. Darwin was be-
ginning the studies on cross- and self-fertilization, which were reported in
1877 in an important book on the subject. It has been thought that Darwin's
views on the significance of crossbreeding may have been instrumental in
inciting and guiding Beal's experiments in the crossing of corn. There seems
to be no supporting evidence, however, for such a surmise.

Beal's lectures before various farmers' institutes stressed the importance
of being able to control the source of the pollen, so that the choice of good
ears in the breeding program would not be nullified by pollen from barren
stalks and other plants of inferior yielding capacity. On this point Professor
Perry Greeley Holden, for several years assistant to Dr. Beal, has stated that
controlled parentage, not heterosis, was the aim of the corn breeding pro-
gram at Michigan and at Illinois before 1900.

In 1895 Holden was invited by Eugene Davenport to become professor of

agricultural physics at the University of Illinois. Davenport also had served

for several years as assistant to Dr. Beal at Michigan. Like Holden, he was

very enthusiastic about the importance of Beal's program, so it was natural

that Davenport and Holden should agree that corn improvement be a major

undertaking of Holden's new department at the University of Illinois. On

initiating this work at the University of Illinois, they learned that Morrow

and Gardner already had tested Beal's variety crossing at Illinois before they

got there, and with confirmatory results. Concerning the motivation of all

this early work, both at Michigan and at Illinois, Holden says:

1. Hybrid corn [as we know it today] was unknown, not even dreamed of, previous to
1900. 2. Controlled parentage was the dominant purpose or object of this early corn improve-
ment work.

Holden thus makes it clear that while heterosis was at play in all of this early
work, it was not the result of, nor did it result in, a heterosis concept.

I refer next to the matter of inbreeding, which some writers have confused
with the crossing that has brought the benefits of heterosis. Enough selfing
had been done with corn prior to 1900 to convince all of those who had had
experience with it that it resulted in notable deterioration. The results of these
early observations are aptly summed up by Holden in the statement that
"Inbreeding proved to be disastrous — the enemy of vigor and yield." No-
where, so far as I have been able to determine, did any of the early inbreed-
ers discover or conceive of the establishment of permanently viable pure lines
as even a secondary effect of inbreeding.

In 1898 A. D. Shamel, then a Junior in the University of Illinois, offered
himself to Holden as a volunteer assistant without pay. He did so well that
when Holden severed his connection with the University in 1900, Shamel
was appointed his successor, and continued in this capacity until 1902. He

BEGINNINGS OF THE HETEROSIS CONCEPT 17

then transferred to the United States Department of Agriculture and did no
further work with corn. In Shamel's final report of his own corn experiments
(1905), he laid no stress on the positive gains which resulted from cross-
breeding, but only on the injurious effects of inbreeding. His "frame of ref-
erence" was the normally vigorous crossbred (open-pollinated) corn, and the
relation between self-fertilized and cross-fertilized corn was that of something
subtracted from the crossbred level, not something added to the inbred level.
The prime objective in a breeding program, he said, "is the prevention of the
injurious effects of cross-fertilization between nearly related plants or in-
breeding." In summing up the whole matter he said:

In general, ... it would seem that the improvement of our crops can be most rapidly
effected with permanent beneficial results by following the practice of inbreeding, or cross-
ing, to the degree in which these methods of fertilization are found to exist naturally in the
kind of plant under consideration.

This means, for corn, practically no self-fertilization at all, and makes it
obvious that, at least for Shamel, the heterosis concept had not yet arrived.

Edward Murray East was associated with the corn work at the University
of Illinois, off and on, from 1900 to 1905. He worked mainly in the role of ana-
lytical chemist in connection with the breeding program of C G. Hopkins
and L. H. Smith. He must have been familiar with the inbreeding work of
Shamel, if not with that of Holden. It is generally understood that he did
no self-fertilizing of corn himself, until after he transferred to the Connecti-
cut Agricultural Experiment Station in 1905. Some of his inbred lines at
Connecticut may have had the inbreeding work at Illinois back of them, as
he secured samples of seeds of the Illinois inbreds sent to him by Dr. H. H.
Love, who assisted him for one year and succeeded him at Illinois. But ac-
cording to his subsequently published records these older inbred lines did not
enter to any important extent into his studies in Connecticut.

As reported in Inbreeding and Outbreeding (East and Jones, pp. 123, 124),
"The original experiment began with four individual plants obtained from
seed of a commercial variety grown in Illinois known as Leaming Dent."
Table III (p. 124) presents the data for these four lines for the successive
years from 1905 to 1917, and clearly indicates that the selfing was first made
in 1905. East's work is so adequately presented in this excellent book that it
seems unnecessary to comment on it further here except to recall that, as
shown by his own specific statements, my paper on "The composition of a
field of maize" gave him the viewpoint that made just the difference between
repeated observations of heterosis and the heterosis concept. In proof of this
we have not only his letter to me, dated February 12, 1908, in which he says:
"Since studying your paper, I agree entirely with your conclusion, and won-
der why I have been so stupid as not to see the fact myself"; but we also
have the published statements of his views just before and just after the
publication of my paper. Thus, we read in his Conn. Agr. Exp. Sta. Bull. 158,

18 GEORGE HARRISON SHULL

"The relation of certain biological principles to plant breeding," which was
published in 1907, only a few months before I read my paper in his presence
in Washington, D.C., what seems like an echo of the final conclusion of
Shamel, above cited. In this bulletin East urged that "corn breeders should
discard the idea of forcing improvement along paths where nothing has been
provided by nature," specifically rejecting a program of isolation of uniform
types because of a "fear of the dangers of inbreeding," adding that he was
"not able to give a reason for this belief beyond the common credence of the
detrimental effects of inbreeding." He returned to this problem of the de-
terioration due to inbreeding in his Annual Report to the Conn. Agr. Exp.
Sta. for 1907-8, prepared in 19C8, with my paper before him. In this report
he says:

I thought that this deterioration was generally due to the establishment and enhance-
ment of poor qualities common to the strain. ... A recent paper by Dr. George H. Shull
("The composition of a field of maize") has given, I believe, the correct interpretation of
this vexed question. His idea, although clearly and reasonably developed, was supported
by few data; but as my own experience and experiments of many others are most logically
interpreted in accordance with his conclusions, I wish here to discuss some corroboratory
evidence.

We have thus far failed to recognize the existence of a general heterosis
concept among plant breeders, prior to the reading of my paper on "The
composition of a field of maize" in January, 1908, even when they were using
the methods of inbreeding and controlled crossing in which such a concept
could have developed. 1 must mention, however, a near approach to such a
concept from the side of the animal breeders. Before the American Breeders'
Association, meeting in Columbus, Ohio, 1907, Quintus I. Simpson, an ani-
mal breeder from Bear Creek Farm, Palmer, Illinois, read a paper which
definitely recognized hybridization as a potent source of major economic
gains beyond what could be secured from the pure breeds. The title of his
paper, "Rejuvenation by hybridization," is more suggestive of the views of
Thomas Andrew Knight than of the current students of heterosis, but the
distinction seems to me to be very tenuous indeed.

Although I listened with great interest to Simpson's paper, I do not think
that I recognized any direct applications of his views to my results with
maize. I was working within the material of a single strain of a single species,
and not with the hybridizations between different well established breeds to
the superiority of whose hybrids Simpson called attention.

Students may make varying estimates as to how closely the work of men
to whom I have referred approached the heterosis concept as we understand
it today. But there can be no doubt that there was a beginning of this concept
in the course of my own experiments with corn. At the beginning of 1907 I
had not the slightest inkling of such a concept. By the end of 1907 I had
written the paper that brought such concept clearly into recognition. At that
time I knew nothing of the work of Beal, Holden, Morrow and Gardner,

BEGINNINGS OF THE HETEROSIS CONCEPT 19

McCluer, Shamel or East, in the selling and crossing of the maize plant.
This will become obvious as I explain the motivation and plan of procedure
of my corn experiments.

Upon arriving at the Station for Experimental Evolution at Cold Spring
Harbor on May 2, 1904, 1 found the laboratory building unfinished. It was in
fact not ready for occupation until the following November. The potentially
arable portion of the grounds was in part a swampy area in need of effective
provision for drainage. The rest had been at one time used as a garden. But
it had lain fallow for an unknown number of years, and was covered with a
heavy sod that would need a considerable period of disintegration before it
could be used satisfactorily as an experimental garden. The total area avail-
able was about an acre.

In the middle of this small garden plot was a group of lusty young spruce
trees. These had to be removed in order to use the area for experimental
planting the following spring. The ground was plowed, disked, and planted
as soon as possible to potatoes, corn, sorghum, buckwheat, sugar beets, tur-
nip beets, and many kinds of ordinary garden vegetables. None of them
were designed as the beginning of a genetical experiment, but only as an ex-
cuse for keeping the ground properly tilled so it would be in best possible
condition for use as an experimental garden later. Due to this fact, no ade-
quate record was made of the origin of the several lots of seeds which were
planted. This is unfortunate in the several cases in which some of these cul-
tures did provide material for later experimental use.

There were two cultures of corn, one a white dent, the other a Corry
sweet corn. These two varieties were planted at the special request of Dr.
Davenport, who wished to have available for display to visitors the striking
illustrations of Mendelian segregation of starchy and sugary grains on the
single ears of the crossbred plants. I planted the white dent corn with my
own hands on May 14, 1904, and must have known at the time that the grains
came from a single ear. Although I have found no contemporary record to
that effect, I am now convinced from a well-remembered conversation with
Mrs. Davenport, that this ear of white dent corn came from the farm of her
father, Mr. Crotty, who lived near Topeka, Kansas.

When I was last in Ames, after almost forty years of devotion to other
lines of genetical experimentation, my memory played me false when Profes-
sor J. C. Cunningham asked me about the source of the foundation stock for
my experimental work with corn, and I told him that my studies on corn
began with some corn I had purchased in the local market as horse feed. I re-
peated the same unfortunate misstatement to several other highly reputable
historians of science. I deeply regret this error because these men were tr}-ing
so hard to get the record straight. My recollection was restored by iinding
the statement at the very beginning of the record of my formal corn studies

20 GEORGE HARRISON SHULL

under date Nov. 7, 1904: "Counted the rows on the ears of White dent corn
raised in Carnegie garden this year." In fact, as I think of it now, I doubt
that I could have bought white dent corn in the feed market of Long Island
at that time.

I planted the Corry sweet corn on May 17. On July 18 I bagged the corn
preparatory to making crosses between the two varieties. This crossing was
carried out on the Corry sweet on July 25, and the crosses for the reciprocal
combination were made on July 27 and 28. These were the first controlled
pollinations I ever made in corn, and they were not part of a scientific ex-
periment.

My interest in investigating the effects of cross- and self-fertilization in
maize arose incidentally in connection with a projected experiment with
evening primroses (Oenothera) to determine the effect, if any, of these two
types of breeding on the kinds and the frequencies of occurrence of mutations.
A critic of De Vries's mutation theory had urged that the mutations dis-
covered by De Vries in Oenothera lamarckiana were artifacts produced by
selfing a species which, in a natural state, had been always cross-fertilized. I
developed a program to put this question to a crucial test . Then, it occurred to
me that it would be interesting to run a parallel experiment to test the effects
of crossing and selfing on the expressions of a purely fluctuating character.
Since I had available this culture of white dent maize, I chose the grain-row
numbers on the ears of corn as appropriate material for such a study. The
Oenothera problems thus begun, continued to be a major interest throughout
my genetical career, but it is not expedient to pursue them further here. It is
important, however, to keep them in mind as a key to my motivation in
launching my studies with maize.

In this double-barreled exploration of the genetical effects of cross-fertili-
zation versus self-fertilization, I had no preconception as to what the out-
come of these studies would be in either the mutational or the fluctuational
field. Certainly they involved no plan for the demonstration of distinctive
new biotypes, nor any thought of the possible economic advantages of either
method of breeding. I was a faithful advocate of the early biometricians' slo-
gan: Ignoramus, in hoc signo lahoremus. Until the middle of summer of 1907,
certainly, I had no premonition of the possible existence of a heterosis prin-
ciple which would have important significance either scientifically or eco-
nomically. I was forced to recognize this principle by direct observations of
manifestations in my cultures which had not been anticipated, and there-
fore could not have been planned for.

Let us proceed then to a description of my experiments with corn which
forced the recognition of this important phenomenon. The culture of white
dent corn which we had growing, almost incidentally, on the Station grounds
that first year, showed no variations that seemed to indicate the presence of
any segregating characteristics. It appeared to be ideal material for the study

BEGINNINGS OF THE HETEROSIS CONCEPT

21

of fluctuations of so definite and easily observed a quantitative character as
the number of the rows of grains on the ears. The crop was carefully har-
vested and placed in a crib. On November 7, 1904, I counted the rows of
grains on every ear, with the result shown in figure 2.1. The 524 ears ranged
over the seven classes from 10-rowed to 22-rowed. The most populous classes

10 ...

.. 3

12 ...

. . 93

14 ...

. . 201

16 ...

. . 153

18 ...

. . 58

20 ...

. . 12

22 ...

.. 4

Total

. . 524

Mean,

14.827

± .061

c>,

2.082

± .043

c. v.,

14.02

±: .29

8

10

12 14

NUMBER

16 18 20

OF GRAIN ROWS

Fig. 2.1 — Frequency curve of grain-rows of 524 ears of white dent corn. The total progeny

of presumably a single ear of corn received from the Crotty farm near Topeka, Kansas, and

grown at the Station for Experimental Evokition in 1904.

were the 14-rowed with a frequency of 201, and 16-rowed with 153 individual
ears. The mean was 14.85 ± .06.

No photograph nor verbal description was made of the parent ear, since
there was no intention at the time of its planting to use it in a breeding ex-
periment. But its characteristics must have been accurately duplicated in all
of the crossbred families subsequently grown, as well as in most of the Fi hy-

22

GEORGE HARRISON SHULL

brids between the several selfed lines. From each of the grain-row classes,
several good ears were saved for planting in the spring of 1905, and the rest
was used as horse feed.

The plantings from this material were made on May 25, 26, 27, 1905, again
with my own hands, in the form of an ear-row planting. Two ears from each
grain-row class of the 1904 crop were used. The seeds were taken from the
mid-region of each seed ear. An additional row was planted from grains of
each of the two parent ears with 16 grain-rows. Only modified basal grains
and modified distal grains for the two halves of the same row in the field
were used. In Table 2.1 these cultures from modified grains are indicated by

TABLE 2.1

GRAIN-ROW COUNTS OF PROGENIES GROWN IN 1905 FROM PARENT

EARS SELECTED FOR DIFFERENT NUMBERS OF

GRAIN-ROWS IN NOVEMBER, 1904

Culture

Parental
Grain-
Rows

Frequencies of Progeny Grain-Row Numbers

Numbers

10

12

14

16

18

20

22

24

26

Totals

Al

10.4*
10 B*
UA
UB
14.4
UB
16 A
\6B
16.4bt
16.44
1656
16 Bp
18.4
18 B
20 A
20 B
22 A
22 B
22 CJ

8
11
12

3

7

1

3

4

3

2

3

3

'r

55
50
36
30
32
11
62
31
7

19

3

5

12

20

3

10

2

3

2

47
57
45
43
58
47
81
79
19
18
5

18
36
33
28
14
9
9
12

16
15
10
28
13
26
44
66

7
16

8
12
39
29
38
28
21
20
32

3

1

1

4

5

13

10

14

2

4

4

11

17

7

14
14
27
28
24

129

A2

1

135

A3

104

\4

108

AS

115

A6

A7and8

A9and 10. ..
All,

1

100

200

1

195

38

All.,

59

A12i

A 12..

23

49

A13

3

1

108

A14

89

A15

2

10
19
18
16

1
2
7

"3"

86

A16

79

A17

85

A18

2

1

1
1

81

A19

91

Totals...

61

393

658

468

203

71

15

3

2

1,874

* The significance of the A and B in this column involved the plan to use the A rows for selfing and the B
rows to be crossed with mixed pollen of plants in the corresponding A rows.

t The subscript b signifies the use for planting of only the modified basal grains of the given ear; and the sub-
script p refers to the planting only of modified grains at the "point" or distal end of the ear.

t C represents an added row grown to increase the probability of finding ears with still higher numbers of
grain-rows.

Ab and Bb for the basal grains, and Ap and Bp for the modified "point"
grains. A second row was planted from each of the two chosen ears having 16
grain-rows, and these additional rows (A8 and AlO) were detasseled, begin-
ning July 24, 1905, and received pollen from the intact plants in the corre-
sponding rows (A7 and A9) beside them.

In harvesting these two pairs of rows, one detasseled, the other intact, the

BEGINNINGS OF THE HETEROSIS CONCEPT 23

two rows from the same parent ear, through an oversight, were not kept
separate. No further detasseling was done. Since the self-fertilized plants
could not be detasseled and still utilized for selling, the method of controlling
cross-fertilization by detasseling would prove a distorting factor in comparing
the effects of selling and crossing.

Consequently, no detasseling was practiced in any of my subsequent ex-
perimental work with corn, but every pollination was controlled by bagging
with glassine bags and manipulation by hand. The bags were tied in place
by ordinary white wrapping-cord passed once around and tied with a loop
for easy detachment. Each plant was labeled at the time of crossing with a
wired tree-label attached to the stalk at the height of the operator's eyes,
and marked with the exact identification of the plant to which it was attached
and the source of the pollen which had been applied. On harvesting these
hand-pollinated ears, the label was removed from the plant and attached
securely to the ear, thus assuring that the ear and its label would remain
permanently associated. A third row (A 19) from an ear having 22 grain-rows
was added to improve the chances of finding ears with still higher numbers
of grain-rows.

In November, 1905, these 19 pedigree cultures were carefully harvested
by my own hands and the grain-rows counted, with the results tabulated in
Table 2.1.

The only observation noted on these 1905 cultures was that there was no
clear indication of mutations or segregations of any kind, but the aspect of
the field was that of any ordinarily uniform field of corn. Row counts did
show the expected indication of Galtonian regression, in that the parents
with low numbers of grain-rows produced progenies having lower numbers of
grain-rows than did the ears having higher than average numbers of grain-
rows. Thus, the two ears with 10 rows of grains each had the average of 13.2
rows of grains on their progeny ears. The two 20-rowed ears showed an aver-
age of 15.5 rows of grains on their progeny ears. The three 22-rowed parent
ears produced progenies with an average of 17.5 rows of grains.

The same general plan was followed in 1906, except that the pollen for
the crossbred cultures was no longer taken from the plants set aside for
selfing. The reason for this change, as specifically stated in my notes written
at the end of the 1906 season, being "to avoid the deleterious effects of self-
fertilization in the cross-fertilized series." This indicated that at the end of
19C6 I had only the concept held by Holden, Shamel, East, and all other
corn breeders who had had experience with the selfing of maize — that selfing
has deleterious effects, not that crossing has advantageous effects other than
the simple avoidance of the deleterious effects of selfing.

The new method of handling the crossbred cultures was to divide each
such culture by a marker set at the midpoint of the row. All the plants in
these rows were bagged. Mixed {)ollen from the plants in the first half of the

24 GEORGE HARRISON SHULL

row was collected and applied at the appropriate time to the silks of all the
plants in the second half of the row. Then the mixed pollen from the plants
in the second half of the row was applied in turn to the silks of all the plants
in the first half of the row. It was realized that this still involved a con-
siderable degree of inbreeding, but it seemed about the only way of carrymg
on a continuing program of crossing while still keeping the breedmg com-
pletely under the operator's control.

Two major observations made on the 1906 crop were: (1) that every one
of the seven families from selfed parents could be readily detected by their
less height, more slender stalks, and greater susceptibility to theattack ot
Ustila2o maydis. When the ears were harvested each lot was weighed and
it was found that cross-fertilized rows produced on an average about three
times as much grain as the self-fertilized. (2) The family A3, from a self-
fertilized ear having 12 grain-rows, was practically all flint corn, showing that
to be probably recessive. This occurrence of a rather obvious segregation in
the 1906 crop remained at the end of the season only an isolated observation
which led to no generalization. From the fall of 1905 until his retirement
in 1943 Charles Leo Macy assisted me in many of the technical details of my
experimental cultures. While I handled the planning and breeding operations
as well as the actual poUinations, Macy prepared the plants for selfing and
crossing, and counted the grain-rows and weighed the ear corn. The results
of these counts for the 1906 crop are given in Table 2.2. _

The following quotation from my notebook seems justified here, since it
includes the first formulation of the considerations and conclusions which
appeared in my report to the American Breeders' Association in 1908, on
"The composition of a field of maize":

^- A c;,, 1007 p'; in 1906) namely each self -fertilized row was

The same plan was continued, ^i" l^^^;^^ ^^^^^^^ row was divided in half,

the offspring of a smgle self-fertilized far, and each cross termzea ^^^^

each half coming from a single cross-fertilized e^^^' f^^^.^f" ^"J^'e ^Vhlr ear coming from the
the first half of the corresponding row of the precedmg > ear, the other ear c n g

second half. ... . mnA tv,o coif fprtili/pd rows being invariably

The obvious results were the same as in ^^O^, the self -fert^ed o^^ s g ..^

smaller and weaker than the corresponding """f "^^ ^iliz'-^j^^f l^^^ upon me

evidence on the self-fertilized A v^ry df erent explana ion of^ h^ P^.^.^^

by the fact that the several self- ertlllzedro^^s differ rom each ^^^^^^^^ elementary

morphological characteristics, thusindicatng that the, be^^^^ ^^^^^ ^^.^.^^ ^^^,

strains. The same point appeared last >ear in the c^se oi tne ^ ^^^^

almost a uniform flint corn, but the ^^^'^^^""^^f.^^^J^Z^^SSr^^^ that my corn-
It now appears that self-fertilization simply ^^'^f. ^ P^"^ > j^^' but between pure strains
parisons are not properly between "^^J^^f .^/^"^Jj^'^j^^^^^

and their hybrids; and that a well '•^g^^^^^f ,^^^ ^ °4e^^ ^ of 'corn must have as

its ?b.-t?^rn!?inraTc^ ^^s^ h^d ^^^^^^^^^^^^^ to\e most vigorous and

productive and give all desirable qualities of ear and grain.

The ideas in this quotation represent a discovery in complete disagree-
ment with my preconception that my white dent foundation stock, which
had been the progeny of a single ear, was essentially a ^^^^'^^^^y f'^l?^'^^'
I had before me seven distinct biotypes, clearly distinguishable m their sev-

BEGINNINGS OF THE HETEROSIS CONCEPT

25

eral morphological characteristics. They had been derived from seven sepa-
rate self-pollinations of sibs in a family which I had reason to think was
genetically homogeneous. This could not fail to make a great imj)ression.
Had these several pure-bred self-fertilized strains come from different
breeders and from more or less disconnected experiments, as did the selfed

TABLE 2.2

GRAIN-ROW COUNTS AND YIELDS OF EAR CORN IN CULTURES OF

WHITE DENT MAIZE GROWN AT THE STATION FOR

EXPERIMENTAL EVOLUTION IN 1906

Culture
Numbers

Parental
Grain-Rows

Frequencies of Progeny
Grain-Row Numbers

To-
tals

Weights
Lbs.Av.

Yield
Bu./A.

8

10

12

14

16

18

20

22

24

26

28

30

32

Al.l

10 selfed
10 crossed
10 crossed
12 selfed
12 crossed
12 crossed
14 selfed
14 crossed
14 crossed
16 selfed
16 crossed
16 crossed
16(22)X10
16^ crossed
16j, crossed

16p crossed

16 J, crossed

18 selfed
18 open-pol.
20 selfed
20 crossed
20 crossed
22 selfed
24 crossed
24 open-pol.
26 open-pol.
26 open-pol.
18 crossed
18(22)X10
14(22) XIO

4
'5'

36
3
2

13

1
"1

62

32

26

40

13

16

12

6

6

8

14

5

8

9

20

7

10

2

14
25
29
19
26
34
41
28
17
17
16
16
23
28
23

39

22

8
16
11

2
3
4
4
12
1

1

7
11

4
12

9
34
18
19
28
15
28
22
20
15

18

18

6
29
23

8
17
10
11
14

8

5

17
20
22

117
68
69
82

581
61/
107
59\
58J
74
471
61/
71
60
63

74

57

26
94

72

,t?l

55
91
66
60
68
78
72
92

A2.2i

1

1

1

6

1
15

7
12
17

1

11
11

3

9

5

5

18
18
21
20
17
25
17
11

9
25

6
26

A2.22

A3.3

21.6
65.8
33.6
61.3
29.6
58.3
22.1

37.7

A4.4i

1

A4.42

78.9

AS.5

4

44 9

A6.6i

A6.62

4
4

1
1
5

74.5

A7.7

59.1

A9.8i

.'^9.82

77.1

A19.9

44.5

A12i.lOi

AI22.IO2

A12i/"'

A122i"'

A13.12

1

2

3
19
10

13
13
13
24
11
17
14
21

2
9
4
5
4
7

18
6
6

19
7

9.6
58 3
23 6

56.3

24.1
57.3
32.6
34.6
40 6

52 9

A14.13

1

1

88.5

A15.14

6

46 9

A16.15i

1

A16.152

75. 1

A17.16

1

3
3
2

10

13

2

62 4

A18.17

4

89.9

A19.18

1

3

70.6

A19.19

7

5

82 4

A18.20

85.4

A16 21

"i

1
16
11

5
29
31

A19. 22 . .

A19. 23

46 1

71.6

Totals

9

58

334

543

469

323

183

89

36

17

3

2

1

2,067

lines available to Dr. East, the observation that they showed themselves to
be genetically distinguishable biotypes would have given no cause for the
special conclusions I drew from them. It would have been strange, indeed,
if strains thus derived from heterogeneous sources had not been genetically
different, one from another.

Comparison of the results for 1907, presented in Table 2.3, with those for
1906 in Table 2.2, shows a heavy accentuation of grain-row classes 8 and 10
and a marked decrease in classes 18 to 20, inclusive. There was also a sig-
nificant increase in all higher classes, with further extension of the range from
a maximum of 32 to about 40. The increase in the frequencies of the low

26

GEORGE HARRISON SHULL

grain-row classes was attributed in part to the fact that the 1907 season had
seemed less favorable in general than 1906.

It was also noted, as a possible contributory condition, that this was the
third season in which this corn was grown on the same area north of the
laboratory building, and that ''the yield may have been lessened by the
gradual accumulation of injurious substances in the soil." The fact that the

Fig. 2.2 — Young corn cultures growing in East Garden of the Station for Experimental
Evolution in 1911, illustrating that no two were alike despite their descent from a single ear
of 1904 by meticulously controlled pollinations that precluded the introduction of pollen

from any other strain of corn.

average grain-row numbers were not significantly different in the two years —
15.8 in 1906, 16.0 in 1907 — in fact a trifle higher in what was thought to have
been the poorer year, does not seem to support these suggested explanations
of the observed differences of distribution in the two years.

My contemporaneous notes proposed an additional explanation, namely,
that "each successive generation of close inbreeding still further reduces the
strains to their simple constituent biotypes, and as these are weaker than
hybrid combinations, this too would tend to lessen the vigor, and this
lessened vigor might readily be evidenced by a decrease in the average num-
ber of [grain-] rows and the total number of ears in the crop."

If we accept this latter suggestion as valid, it is clear that the occurrence

BEGINNINGS OF THE HETEROSIS CONCEPT

27

of essentially the same average numbers of grain-rows in the two years gives
only a specious indication of the relative climatic and soil effectiveness in
these two seasons. It must mean simjjly that the diminution of grain-row
numbers produced by increasing homozygosity happened to be balanced by
the increased frequencies in the higher classes, produced by the gradual ac-
cumulation by selection of more potent hybrid combinations.

TABLE 2.3

GRAIN-ROW COUNTS AND HEIGHTS OF PLANTS IN
THE CULTURES OF 1907

Pedigree

Grain-Rows
of p.\rents

Frequencies

DF Progeny Grain- Row Numbers

To-
tals

Av.
Ht

Numbers

8

10

12

14

16

18

20

22

24

26

28

30

32

34

36

38

40

IN

Ft.

Bl.l

10 selfed
10 crossed
10 crossed
8 selfed
12 selfed
12 crossed
12 crossed
14 selfed
14 crossed
14 crossed
16 selfed
16 crossed
16 crossed

16 crossed
I61 crossed
18 selfed
18 open-pol.
20 selfed
20 crossed
20 crossed
22 selfed
22 crossed
20 crossed
24 crossed
32 open-pol.
Branched ear
30 open-pol.
Branched ear
16 crossed
24 selfed
26 selfed

20

2

6

23

10

"1

25
22
28
48
21
1

' '4

20
21
18
17
18
16

7
23

5

5
19

2

2

5
5

1

68

501

57/

88

54

52

44

73

48!

43/

55

371

36/

311

37/

15

71

52

631

52/

45

63

62

50

41

58

58

49

64

36

12

7.25

B2i.2i

B22.22

9.00

B1.3

7.63

B3.4

4
19
14
29
15
18
26
8
9

13
15

1
13
14
15
19
19

9
14
15

10

9

5

22

21

16

10

6

9

7

1

3

9

5

14

14

5

6.25

B4i.Si

3

7

1
7
7

8 00

B42.S2

1
1
1
1

B5.6

8.50

B6i 7i . .

1
1

B62.72

9 67

B7.8

8 25

B8i.9i

7
9

1

1

7

15

11

18

13

9

17

22

7

6

13

12

17

15

5

3
3

1

1

1

3

6

13

16

17

17

17

16

8

8

17

12

17

7

3

3

B82 92 ...

8 75

BIO2 lOi. ..

6
11

BlOi IO2 .

8.67

B12.il

2

1

1

10

8

8

11

11

14

8

1

11

3

6

7

2

7.00

B13 12

9

21
8
3
4

8 33

B14.13... .

7.25

B15i 14i...

2

1
5
4
4
5
7
1
9

B152.142...

8 83

B16.15

7.00

B17.16... .

1
1

3

1

8 67

B15.17

8 83

B19.18

4
4

3

2

9 50

B20.19... .

2

9 50

B? 20. . .

6

2
3

5

8 33

B17 21... .

1

8 33

B?.22

BIS. 23

3
6
2

4
6

B20 24

8.00

B20.2S... .

1

2

1

1

7.83

Totals . .

62

150

204

236

282

228

189

108

49

22

6

3

3

1

1

1

1,545

A truer measure of the relative favorableness of the two seasons for growth
and productiveness of these cultures can be derived from a study of the
middle classes with 12, 14, 16, and 18 grain-rows. These grain-row classes
making up 80 per cent of the 1906 crop and 61.5 per cent of the 1907 crop,
must be relatively free from most of the distortion assumed to be produced
either by increasing homozygosity or by the accumulation of the more po-
tent hybrid combinations. If we average these four grain-row classes by them-
selves for the two years, we find that in 1906 their average was 15.5 grain-
rows, and for 1907 only 15.0, thus agreeing with my general impression
that 1907 was the less favorable year.

With the fundamental change in my understanding of the nature of my
corn population came a reorientation of the experiment. I found myself at

28 GEORGE HARRISON SHULL

the end of 1907 only ready to make a beginning on the problems of the rela-
tionship between pure lines and their hybrids, which I now saw was the cru-
cial field that needed exploration.

As a first step in that direction, but without as yet a full comprehension
of its importance, I made in July, 1907, pollinations between plants of C4,
which I later designated "Strain A," and a plant of C6, which later became
my "Strain B." I also made two sib crosses within these two strains. The
cross of Strain A X Strain B, which gave rise in 1908 to Fi family, D9, in-
volved an 8-rowed ear of the former strain (from an original selection for
12 grain-rows) and a 12-rowed ear of Strain B which had originated in a selec-
tion for 14 grain-rows. The near-reciprocal cross (Fi family, D13) resulted
from the application of pollen from a 12-rowed plant of Strain A to silks of
the same plant of Strain B, which supplied the pollen for the near- reciprocal
cross.

At the time when these two near-reciprocal crosses were made between
Strains A and B, the truth had not yet dawned upon me that I should do the
same with all of my other selfed families. Aside from these two sets of crosses,
the handling of the cultures was the same as in previous years. The results
of the grain- row counts are given in Table 2.4. Unfortunately, there was con-
siderable damage from crows, and failures to germinate for unknown reasons.
The missing hills were replanted on June 8, 1908, and all of the new plantings
made on this date seem to have reached maturity. To overcome the suggested
deteriorating effect of soil depletion, the cultures were grown this year on the
area east of the laboratory building (occasionally referred to in subsequent
notes as "Fast Garden").

In summarizing the results for the year 1908, it may be noted first that
the tendency to concentrate the frequencies of the grain-rows in the extremes
of the range, at the expense of those in the middle, has continued strongly.
As before, the most noteworthy concentration is at the lower extreme. All
classes below 16 are considerably stronger in 1908 than in 1907 and the
maximum frequency is now on 12 instead of 16. This is in part due to the fact
that several of the lower-class families were grown in duplicate. Between
classes 14 and 26 the relative strength of the classes was lessened in 1908.
Above class 24 the frequencies were increased, there being 84 ears above
class 24 in 1908 and only the equivalent of about 50 in the same region in
1907, when raised to the same total number. The highest number of grain-
rows noted was 42.

The important new features brought in by the near-reciprocal crosses be-
tween Strain A and Strain B and a sib cross in Strain A are presented in my
report to the American Breeders' Association at Columbia, Mo., in January,
1909, on "A pure line method in corn breeding." I find a discrepancy in that
the 78 ears produced by the sib cross weighed only 16.25 pounds instead of
16.5, as stated in my 1909 paper. Whether by an oversight or intentionally,

BEGINNINGS OF THE HETEROSIS CONCEPT 29

I cannot now determine, the corresponding sib crosses in Strain B were not
included in my 1909 report. The results were essentially the same as were re-
ported for the sib cross in Strain A. Selfed Strain B (see Table 2.4, family
C6.ll) showed average heights of plants 2.3 meters, and yielded 66 ears
weighing 13.0 pounds. The two sib crosses produced plants 2.5 meters tall
and yielded 89 ears weighing 28.5 pounds. Distribution of the grain-row
frequencies was closely similar in selfed and in sib-crossed Strain B, but sig-
nificantly higher in the latter:

_, . Totals Averages

Grain-rows 10 12 14 16 18

Selfed 2 20 26 17 1 66 1.3 8

Sib-crossed 3 15 45 18 8 89 14.2

There was abundant evidence that the sib crosses showed a greatly re-
stricted advantage over self-fertilization. It was also clearly indicated that

TABLE 2.4

GRAIN-ROW COUNTS, HEIGHTS, AND YIELDS OF
WHITE DENT MAIZE GROWN IN 1908

Pedigree

Gr.^in-Rows

of p.4rents

Frequencies of Progeny
Grain-Row Numbers

To-
tals

Av.

Hts.

IN

Dm.

Wts.

IN

Lbs.

Yield

Bu./

A.

8

10

12

14

16

17
11

6
11

3
28
25
13
34
16
14
39

4

6
17

9
22

3
tvs t

1

14
20

323

18

"1

1

5

3

1

2

2

1

11

4

21

12

4

23

15

17

19

20

26

4

30 d

2

19
31

244

1
20

"i

6

1

6
19
10
18
24
25
21
fBci
5

16
22

172

22

1

"7
11
3
13
19
16

It t(
10

14

7

91

24

8

16
21
24
) coi
10

4
3

60

26

"2

"4

"7

int;

16

2
31

28

"i

3

silks
18

2
24

30

i

sh
5

6

32

3

ort
6

9

34

i

er
9

3d

38

40

42

C3.1

C1.2

C2,.3,

C22.32

C1.4

C2i.5i

C22.52

C4.6

C4.7

C4.8

C4.9

C5i.ini....

C52.102....

C6.ll

C6.12I...,

C6.122

C6.13

C7,.14i....
C72.142....

C8.15

C9i.l6i....
C92.162....
C13.17....
CI22.I81...
C122.I82...
C13.19....
C14,.20i...

10 selfed
8 selfed
8 crossed
8 crossed
10 selfed
10 crossed
10 crossed
12 selfed
lOXsib
8 selfed
8X12
12 crossed
14 crossed
14 selfed
16Xsib
12Xsib
12X12
14 crossed
14 crossed
16 selfed
16 crossed
16 crossed
18 selfed
18 crossed
18 crossed
20 selfed
20 crossed
20 crossed
20 selfed
22 crossed
22 selfed
28 crossed
36? selfed
28(?)X26(?)
Branched ear

open-pol.
20 open-pol. t

52
51

6

6
28

9
12
11

8
65

"2
1

39

41

29

22

48

32

18

41

50

6

19

9

9

2

2

ii

13
2

14

12

12

9

3

32

19

2

64

31

17

20

4

11

56

18

9

31

4

6

6

6

8

2

"2

1
2

"5
1

'9
15
14
26
25
20
31
28
18
32
14
18
10
20
19
15
3
2

13
3
9

tha
3

n'h

usl

cs.

1

104
94
51
42
90
50
33
89
78
73
92
59
42
66
47
42
100
59
43
94
55
41
77
56
46
85
58
46
70
84
92
83

86

83
93

19.5
19.7

23.4

is'o

21.5

'i7'0
16.5
16.5
24.0
24.5
22.5
23.0
25.0
25.0
26.0
25.0
27.0
24.4
26.8
25.2
19.3
23.5
25.5
21.6
. . . .*

31.5
22.0
25.01
21.0/

22 0
20.81
14.0/
28.0
16.3
12.0
48.0
34.8

23 3
13 0
16.8
11.8
55 . 0
30.0

19 3
31.5
31.5

20 0
16.5
31.3
28.3
23 0
31.0
24.5
20.5
48.3
33.8
43.3

50.5

50 0
51.8

43.3
33.4

70.7

30.5

59.8

44.9
29.8
23.5
74.5
84.1
79.1
28.1
50.9
40.0
78.6
72.6
64.0
47.9
80.8
69.7
30.6
79.7
87.7
38.7

CI42.2O2.. .

76.4

C15.21... .

79.2

CI61.22. . .

41.8

C24.23....
CIS. 24....

2

82.1

52.4

C25.2S....
C19.26....

Grai

n-ro

74.4

C22.27....
C22.28....

1
1

11
9

83.9

86.2
79.5

Totals .

^S^

387

415

375

10

3

1

2,403

* The remaining nine rows were not measured and described, "for lack of time.'
lanlil^'' '''''"' ""'^"^ ^°"' ^^'' "^'^^ '■*' ''^'^^' ^""^ ^° '"""^ °^ ^''''"'' °^ ""'^'""^ °"'^' "'"^ twenty-rowed ear was used for

Fig. 2.3— Vegetative habits of Strain A (righl) and Strain B, drawn by J. Marion Shull
from a photograph taken in the summer of 1908. At upper right typical ears of these two
strains {Strain A at right) and between them their reciprocal Fi hybrids, each hybrid stand-
ing nearest to its mother type.

BEGINNINGS OF THE HETEROSIS CONCEPT 31

if the advantage consisted solely of the effects of heterozygosity, both Strain
A and Strain B were still a good way from being homozygous, Strain B being
as yet more effectively heterozygous than Strain A.

In the reciprocal crosses between these nearly homozygous strains A
and B, we have our first opportunity to arrive at an approximation to the
actual amount of heterosis. The most important new discoveries these
crosses made possible were: (1) As a result of such a cross it is possible to
completely cancel in a single year the accumulated deterioration which
had gradually accrued, although with lessening annual increments, over a
period of several years; and (2) the approximate identity of the results of the
reciprocal crosses gave assurance that the amount of heterosis resulting from
a given hybridization is a specific function of the particular genetical combi-
nation involved in the cross.

Several new cultures of yellow- and red-grained corn were added to my
experimental field in 1908, but these will not be followed here. They are
mentioned only because they were included in my numbered pedigrees, and
their omission in the following tables leaves a break in the series of numbered
families which might lead to some question as to the reason for the apparent
vacancies. The data from the 1909 cultures of white dent corn are presented
in Table 2.5.

The families grown in 1909, as tabulated in Table 2.5, fall into three major
classes: (1) Twelve families involve continuations of the original self- fer-
tilized lines, whose average yields range from 18.8 to 41.2 bushels per acre,
with the average for all twelve at 32.8 bushels per acre; (2) Twelve are con-
tinuations of crossbred families in which strictly controlled cross-fertiliza-
tions were made with mixtures of pollen taken from the other plants in the
same crossbred strain. These yielded from 58.1 to 83.3 bushels per acre with
the average of all at 73.3 bushels per acre; and (3) there were fourteen Fi hy-
brid families from crosses between pairs of individuals representing two dif-
ferent selfed lines. The yields of these range from 60.3 to 87.5 bushels per
acre, the average for all fourteen being 78.6 bushels per acre. As stated in my
1910 paper, the three highest yields of any of these cultures were from the
families produced by crossing representatives of different selfed strains (see
D8.13, D8.16, andD11.21).

Besides these, there were two cousin crosses involving matings between
different families of the same selfed line. These produced, respectively, 27.1
and 44.6 bushels per acre. One cross between two sibs in Strain A gave 26.0
bushels per acre. The other cross was two F2 families, each from crosses with
mixed pollen within one of the Fi families of my 1908 cultures. These Fo
families yielded 54.2 bushels per acre from the (A X B)F2, and 70.6 from
the (B X A)F2. These yields should be compared with those of the corre-
sponding Fi families grown in the same season, in which (A X B)Fi yielded
74.9 and 83.5 bushels in two different families, and (B X A)Fi produced
82.6 bushels per acre.

32 GEORGE HARRISON SHULL

In 1910 I was absent from the Station for Experimental Evolution during
the entire summer and my experiments with corn, evening primroses,
Lychnis, etc., were continued by an assistant, R. Catlin Rose, assisted by
Mr. Macy, who carried out the operations meticulously described by myself
in more than one thousand typewritten lines of detailed instructions.

The data on the white dent corn grown in 1910 are presented here in

TABLE 2.5

GRAIN-ROW COUNTS, HEIGHTS OF ST.ALKS, AND YIELDS OF
EARS OF WHITE DENT CORN IN 1909

Pedigree
Numbers

Grain-Rows
OF Parents

Frequencies of Progeny
Grain-Row Numbers

To-
tals

Hts.

IN

Dms.

Wts.

IN

Lbs.

Yield
Bu./A.

8

10

12

14

16

18

20

22

24

26

28

30

32

34

36

38

40

Dt.l

D2.2

D3i.3i. . .
D32.32. . . .

D4.4

D4 5

8 selfed
8 selfed
8 crossed
8 crossed
10X12
10X14
lOXsib
10 selfed
10 crossed
10 crossed
12 selfed
12Xcousins
12Xcousins
A selfed
AX20
AX22
AXB
AX20
AX 16
AXB

(AXB)Fisibs
12 crossed
12 crossed
B selfed
BXA
BX20

(BXA)Fisibs
14 crossed
14 crossed
16 selfed
16 crossed
16 crossed
18 selfed
20X16
20XA
16Xcousin
1 8 crossed
18 crossed
20 selfed
20X16
20 crossed
20 crossed
22 selfed
22 crossed
22 crossed
24 selfed
24 crossed
24 crossed
30 selfed
28 crossed
28 crossed
22 crossed
24 crossed

21
29
18
8
30

io

7
3
4
5
1
3
66

51
70
25
39
55

8
53
32
23
22
50
31
29

5

30
6

12

3

21

44

32

55

17

15

35

18

20

3

4

44

18

21

74

71

57

28

25

10

58

6

26

13

14

25

2

4

2

1

16

3

5

ii

1
4
1

4

1

'46
50

9
33
32

5
11
16
17
18

9
20
40
23
18
51
11
19
14

4
46

9
18
18
14
12

'3

4

102

105
55
50

106
63
96
98
44
41
94
50
53
74
96

102
31
60

115
86

108
51
51
40
86
80
84
48
41
81
35
48
73
96
85
36
47
51
87

113
52
67
91
32
44
97
34
21
68
39
44

100

104

18
20
21
22
20
24
17
19
24
24
18
19
19
17
24
26
24
26
28
27
25
25
23
26
28
28
27
28
29
24
25
26
20
27
24
23
28
28
24
28
30
29
26
25
27
23
27
27
25
29
28
29
29

24.0

24.8

21.0'i

22.5/

44-8

35.3

17.5

25.0

17. 3\

18. 3J

23,5

9.5
10,3

9.8
54 0
60.0
16,3

29 8
61,3
50.3
41 0
29,5

30 0
7.3

49.8

49 0

41,5

23,81

20 8/

21.0

22.51

24.0/

17.3

53.8

46 0

11.3

26.01

28.5/

20.3

63 3

29, 0\

34 8/

25,3

17,51

26.8/

28 0

14 81

12.8/

11 5

14,31

19,5/

37,3

53 , 5

53 0
33.7

59.2

60,3
80 0

D4.6

D4.7

DS1.81

D52.82. . . .

D6.9

D7.10i....
D7.102. .. .

D8.11

D8 1'

45
7
2

5
7

■2

"3
12

38

15

10

8

4

9

19

42

27

22

19

17

23

36

54

2

5

4

5

6

5

9

3

15

"8

6

15

43

' '4

5

4

27

36

12

19

30

12

5

22

2

' '4

18

"1

1

1

8

9

23

21

41

11

8

36

10

2

2

5

4

27

28

"3

"i

8

14

12

4

10

16

7

5

1

5

4

11

21

"5
7
4

"s

12
11
3
4
3
6
2
7

' '2
1

'2
5
3
7

12
5
7

'4

' 1
1
1
2

14
8
8
1

'2

"2
9

5
5

"1

li
1
3

4

■ ■

4
'2

26.0
56.4

59.7

35.7
27.1
27.6
18 8
80 4

D8.13

D8.14i....
D8 142

:::

1
2

84.0
74.9
70.8

D8.15

D8.16

D9.17

DIO1.I81. .
DIO2.I82. .
D 1 1 19

1

2
3
2

1
8

32

5
5

76.2
83.5
54,2

83^3'
25 9

1)11.20....
Dll M

19

82 6

87,5

D13.22....

D14i.23i.

D142.232

1
2

70,6
71.4

D15.24....
D16i.25i

1

37.0
80.0

D16'.252

D17.26....

33 8

D17 27

80 0

D17.28....
D17.29. .

77.3
44.6

DI81 3O1

79.4

DI81.3O2

D19.31

33 3

D19.32....

80.0

D20i.33i

76.5

D2O2.332

D21.34..

39.6

D22 35i

83.2

D22.352

1

D23.36....

41.3

D24 37i

71.4

D24.372

D'5 39

24.2

D26.40i

3

2

23

18

4

2

31

22

58.1

1)26.402

1)27 41

5
3

81.8

D28 42

73.5

Totals

214

570

846

588

497

341

261

123

73

48

36

24

16

5

6

3

4

S6S';

Without-

Se)jfftrti)«*

tjon.

live Greneyatiew'X

1 1 k I i

f7 libtr-u.

I

■'

I

1

^L

■t..

wm^

Bi.

Fig. 2.4 — An exhibit set up in the Genetics Department of Cornell University in 1910, dis-
playing materials grown at the Station for Experimental Evolution in 1909.

Fig. 2.5— The best eleven ears of the highest-yielding selfed line (F 29.70 in Table 2.7)
grown in 1911 {top row) ; the best eleven ears of the best Fi hybrid grown in the same year
(F 32.75 in Table 2.7) ; and the best eleven ears of a crossbred strain (F 55.84 in Table 2.7)
in which selling was completely prevented during five years. This shows the relative vari-
ability which is characteristic of these three types of families, the Fi being no more variable
than the inbred, while the crossbred is quite noticeably more variable.

34

GEORGE HARRISON SHULL

summary form. Some 73 ears were selected for planting, and 5,343 ears were
harvested. The complete grain-row distribution was as follows:

Grain-rows 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 Total

Frequencies ... . 401 812 1271 921 716476275 141 118 74 53 41 24 8 6 4 1 1 5343
Percentages. ... 7.5 15.2 23.8 17.2 13.4 89 5.2 2.6 2.2 14 10 0.8 0.5 0.2 0.1 0.1 00 0.0 100.0

To save space and still indicate as completely as possible the significant
results of these studies in 1910, the data from the several kinds of families
of white dent corn grown at the Station for Experimental Evolution that
year are presented in the form of averages in Table 2.6. The several quanti-
tative indicators of physiological vigor, namely, the average number of
grain-rows, heights of stalks, and bushels of ear-corn per acre, can be readily
compared as follows:

Types of Families

Inbreds selfed ....
InbredsXsibs. . . .

Crossbreds

Fi between inbreds
F2 from Fi selfed. .
F2 from Fi X sibs . .

No. of

Av. No. of

Families

Grain-Rows

10

12.6

8

13.7

11

16.9

6

15.2

11

13.3

11

13.5

Av. Heights
in Dms.

19.3
19.8

23.5
25.7
23.3
23.1

Av. Yields
in Bu./A.

25.0
28.7
63.5
71.4
42.6
47.9

Si.x interesting comparisons can be made among these summaries: (1)
comparisons between inbreds selfed and inbreds crossed with pollen from
one or more of their sibs; (2) comparisons between inbreds and crossbreds
in which selfing has been completely prevented, but which still represent a
(fairly low) degree of inbreeding; (3) comparisons between inbreds and their
Fi hybrids; (4) comparisons between the crossbreds in which selfing has been
prevented through six generations and the Fi hybrids in which five successive
generations of selfing have been succeeded by a single cross; (5) comparisons
between the Fi and the F2 hybrids of the inbreds; and (6) comparisons be-
tween F2 hybrid families produced by selfing the Fi and those F2 families
produced by sibcrosses in the Fi.

On making these comparisons we see that the evidence for residual hetero-
zygosity in the inbreds is indicated by excesses in the sibcrossed families of
the inbreds over the selfed inbreds of 8.7 per cent in grain-row number, 2.8
per cent in heights of stalks, and 14.7 per cent in yield of ear-corn. In the Fo
families (sections E and F, of Table 2.6) those produced from sibcrosses in
the Fi surpass those families produced from selfings in the Fi by 0.9 per cent
in grain- row number and 12.5 per cent in yield.

The average heights of stalks reverse the expectation by showing an in-
significantly less height from the sibcrossed matings than from the selfings,
the difference being 0.9 per cent. The contrast between the results of six
successive selfings and the continued prevention of selfing for the same six

TABLE 2.6

AVERAGE V.ALUES IN THE FAMILIES OF WHITE DENT MAIZE

GROWN IN 1910, GROUPED ACCORDING TO THE

TYPES OF MATING OF THE PARENTS

Pedigree
Numbers

Parental
Grain-Rows

Number
of Stalks

Av. No.
of Grain-
Rows

Heights
in Dms.

Wts. in
Lbs. Av.

Yields
Bu./A.

(A) Families from Inbreds Selfed

El. 16 . .

8 selfed
8 selfed
10 selfed
12 selfed
A(8) selfed
B(14) selfed
14 selfed
18 selfed
22 selfed

26. 28 selfed

■

57
83
79
80
75
53
66
82
62
72

10.0

9.0
11.1
12.3

8.8
12.9
13.8
15.2
17.9
15.2

17

18

20

17

16.5

24

23

19

19

19

9.8
22.0
18.3
11.4
9.1
7.3
16.3
15.3
11.0
17.5

24 4

E2.19

^9 6

E7.29

33 9

E9.32

20 9

E11.34

E19.47

E24.54

E26.56

E34.67

E36.71

18.1
11.0
25.8
22.9
19.2
34.2

Unweighted averages

71

12.6

19.3

10.7

25.0

(B) Families from Inbreds Pollinated by Sibs; Selfing Prevented

El. 17

lOXsibs
10 X sib
12 X sib
A(8)Xsib
B(12)Xsib
18 X sib
20 X sib
?(fasc.)Xsib

61

75
85
55
54
89
65
73

10.2
9.9
11.0
9.5
12.7
15.8
17.9
22.5

19
18
22
16
24
20
20
20

13.8
21.0
18.3
7.5
5.3
24.5
15.3
18.3

29 8

E2.20

39 5

E7.30

37 ^

E11.33

E19.48

E26.57

E34.68

E36.72

16.0
7.8
37.8
25.6
35.2

Unweighted averages

61

13.7

19.8

15.5

28.7

(C) Families from Parents Given Mi.^ed Pollen in Each Generation;
Selfing Prevented

E3.23 ...

8, 10 crossbred
10 crossed
12 crossed
14 crossed
16 crossed
18 crossed
20 crossed
20, 22 crossed
24, 20 crossed
32 crossed
32 crossed

88
65
91
94
95
202
100
45
69
56
99

9.5
10.3
13.2
13.7
14.9
16.0
18.5
20.0
24.2
19.2
26.2

22

22

24

27

28

22.5

23

21

22

24

23

30.8
31.0
51.0
49.0
48.8
76.8
35.8
26.3
24.5
22.5
39.0

49 9

E8.31

68 1

E18.46

E23.53

E25.55

E30.63

E33.66

E35.70

E37.73

E40.75

E40.76

80.1
74.5
73.3
54.3
51.1
83.3
50.7
57.4
56.3

Unweighted averages

91.3

16.9

23.5

39.6

•

63.5

TABLE 2.6 — Continued

Pedigree

Numbers

Parental
Grain -Rows

Number
of Stalks

Av. No.
of Grain-
Rows

Heights
in Dms.

Wts. in
Lbs. Av.

Yields
Bu./A.

(D) Fi Hybrids between Different Inbred Lines

E2.21

A(10)X16

A(10)XB

A(8)X10

A(8)XB

18X14

18X26 + (fasc.)

95
94
95
84
109
92

13.8
12.8
11.0
12.3
17.8
23.3

24
28

25
25
27
25

50.3
50.0
33.5
28.5
60.8
62.5

75.6

E2.22

76.0

E11.36

E11.37

E26.58

E34.69

51.5
48.5
79.6
97.1

Unweighted averages

93

15.2

25.7

47.6

71.4

(E) F2 Families from FiXSelf

E4.24

(lOXA)Fi selfed
(10Xl4)Fiselfed
(A X20)Fi selfed
(AX 22)Fi selfed
(AX 16)Fi selfed
(A XB)Fi selfed
(BXA)Fi selfed
(BX20)Fi selfed
(20Xl6)Fi selfed
(20XA)Fi selfed
(20 Xl6)Fi selfed

86
86
76
83
94
96
95
92
97
95
93

10.6
12.1
13.9
12.8
12.8
12,0
11.7
15.1
16.6
13.0
15.9

21

22
19.5

24
25
25
24

25
25
22
24

30.8
29.8
20.5
18.8
33.5
24.0
25.3
28.0
35.3
22.0
29.5

51.1

E5.26

49.4

E12.38

E13.40

E15.42

E16.44

E20.49

E21.51

E27.59

E28.61

E32.64

38.5
31.4
50.9
35.7
38.0
43.5
51.9
33.1
45.3

Unweighted averages

90.3

13.3

23.3

27.0

42.6

(F) F2 Families from FiXSibs

E4.25

(10Xl2)F,Xsibs

(10Xl4)FiXsibs

(AX20)FiXsibs

(AX22)FiXsibs

(AXl6)FiXsibs

(AXB)FiXsibs

(BXA)FiXsibs

(BX20)FiXsibs

(20Xl6)FiXsibs

(20XA)FiXsibs

(20Xl6)FiXsibs

85
83
80
96
95
93
80
93
89
92
97

10.7
12.2
14.2
13.4
12.3
11.8
11.6
15.5
17.2
13.7
15.4

21
22
21
25
23
24
24
25
25
23
21

31.3
35.0
28.8
27.0
37.3
21.0
23.5
31.8
37.3
30.0
25.3

52.5

E5.27

60.2

E12.39

E13.41

E15.43

E16.45

E20.50

E21.52

E27.60

E28.62

E32.65

51.3
40.2
56.0
32.3
42.0
48.8
59.8
46.6
37.6

Unweighted averages

89.4

13.5

23.1

29.8

47.9

BEGINNINGS OF THE HETEROSIS CONCEPT 37

years (sections A and C, Table 2.6) shows the latter in excess of the former
by 34.0 per cent in grain-row number, 22.1 per cent in height of stalks, and
154.2 per cent in per acre yields of ears. The superiority of the Fi hybrids
between different inbreds and the families in which selfing had been pre-
vented during six generations of controlled breeding (sections D and C,
Table 2.6), is indicated by an excess in heights of stalks of the Fi families
over the crossbreds, of 9.4 per cent, and in yields of ear-corn per acre of 12.3
per cent. But here there is a notable reversal in grain-row numbers. Not-
withstanding these proofs of the superior vigor of the F/s over the cross-
breds, the latter exceed the former in grain-row number by 10.8 per cent.

The reason for this reversal is easily recognized when we consider that
parents were selected in these studies for their grain-row numbers, with no
noticeable selection for heights and yields. In section D of Table 2.6, we note
that only one parent of any of the Fi families had a grain-row number in
excess of 18. The crossbred families ranged in parental grain-row numbers
from 8 to 32. Five of the families came from parents having more than 18
rows of grains.

To make a fair comparison between the two types of breeding in their re-
lation to grain-row number, it is necessary to use only the crossbred families
having parents with no more than 18 grain-rows. When we make such a limi-
tation, we find the average grain-row number for the remaining six crossbred
families is only 12.9. The grain-row average for the six Fi families, namely,
15.2, exceeds the crossbreds by 17.1 per cent. Limiting the other indicators
of physiological vigor to the same six crossbred families, we find that the F/s
exceed the corresponding crossbreds on the average by 6.3 per cent in height
of stalks and 7.0 per cent in yield of ear-corn.

In 1911 I was again in full personal charge of the corn experiments at
the Station for Experimental Evolution, and was able to expand the work
considerably, both quantitatively and in the types of matings studied.
We planted 84 cultures in the white dent series as well as 25 cultures of
other types of corn. The total number of white dent ears of which the grain-
rows were counted was 6,508 which showed the following frequencies:

Grain-rows 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 Total

Frequencies 267 767 1725 1298 931 683 363 164 114 95 65 23 7 3 3 6508

Percentages 4.1 11.8 26.5 19.9 14.3 10.5 5.6 25 1.8 1.5 0.9 0.4 0.1 0.1 0.1 99 9

In Table 2.7 the 1911 results are presented in condensed form. Families
are grouped in eleven sections representing fairly homogeneous groups,
mostly based on the types of matings involved. Sections D and E are both
made up of the same five families of F2 hybrids produced by selfing the same
number of different Fi's. For these families each seed ear was used to plant
two rows. The one row of each such family was grown with the other cultures,
as usual, in the East Garden. The second row of each of these families was

TABLE 2.7

AVERAGE GRAIN-ROW NUMBERS AND YIELDS PER ACRE OF WHITE

DENT MAIZE GROWN IN 1911 GROUPED ACCORDING TO THE

TYPES OF MATINGS OF THE PARENTS

Pedigree
Numbers

Parental Strains Involved

Number
of Stalks

Av. Num-
ber Grain-
Rows

Weights
in Lbs.

Yields
Bu./A.

(A) Families from Inbreds Selfed

FI6.681

E2.68.2

F29.70

F32.73

F34.76

F0.77

E19.79i

F47.792

F0.80

E24.82

F56.85

E36.92

F74.94

8 selfed

8 selfed

10 selfed

12 selfed

Strain A selfed

A from L. H. Smith

B selfed

B selfed

B from L. H. Smith

16 selfed

20 selfed

26, 28 selfed

*"Cobs" selfed

12
44
89
95
98
101
3
46
95
84
90
79
64

8.7 1.5 17.9

9.0 6.0 19.5

10.9 16.5 26.5

11.8 11.3 16.9

8.4 8.3 12.1

8.9 8.8 12.4

Not counted nor weighed

Not counted nor weighed

14.3 4.3 6.8

14.0 7.5 12.8

15.3 13.8 21.8

22.7 11.5 20.8

Not counted nor weighed

1

Unweighted averages (omit-
ting the three uncounted
families)

78.7

12.4

8.9

16.7

(B) Families from Parents Given Mixed Pollen in Each
Generation; Selfing Prevented

F23.69

F31.72

F46.78

F53.81

F55.84

F632.86

F66.87

F70i.91

F73.93

F76.96

8 crossed
10 crossed
12 crossed
14 crossed
16 crossed
18 crossed
20 crossed
22 crossed
24 crossed
32 crossed

71
95
92
97
101
105
99
63
68
94

10.4
10.7
12.2
13.7
15.2
18.2
19.4
22.3
23.8
25.2

30.3
30.3
44.5
40.8
33.0
42.5
40.0
20.8
34.5
50.5

60.2

45.5
69.1
60.0

46.7
51.8
57.7
45.9
72.5
60.4

Unweighted averages

88.5

17.0

36.7

57.0

(C) Fi Hybrids between Different Inbreds

F29.71

F32.74

F32.75

F54.83

(10Xl2)Fi
(lOXB)Fi
(10Xl6)Fi
(16X20)Fi

62
106
100
100

12.2
12.8
14.3
18.4

24.5
65.3
63.0
58.2

56.5
87.9
90.0
83.2

Unweighted averages

92

14.4

52.7

79.4

♦ This was a slightly fasciated brevistylis type, with silks about half as long as the husks. Usually it pro-
duced no grains except when given artificial help.

38

TABLE 2.7— Contimie

Pedigree

Number

Parental Strains Involved

Number
of Stalks

Av. Num-
ber Grain-
Rows

Weights
in Lbs.

Yields
Bu./A.

(D) F2 Families from Fi Selfed, Clrown in Annex No. 1

F21.24

F22.28

F36.31

F37.36

F58.54

(8X20)F, selfed
(8XB)F, selfed
(A XlO)Fi selfed
(A XB)Fi selfed
(20 XI 6) Fi selfed

69
61
99
93
103

13.8
13.4
11.3
11.8
16.2

23.0
31.3
33.3
17.0

54.3

47.6
73.2
48.0
29.3

47.5

Unweighted averages

83

13.3

31.8

49.1

(E) Same Families as in (D), but Grown in East Garden

F21.24

F22.28

F36.31

F37.36

F58.54

(8 X20)Fi selfed
(8XB)Fi selfed
(AX 10)F, selfed
(A XB)F, selfed
(20 Xl6)Fi selfed

98

101

98

76

97

13.4
13.4
11.1
11.0
16.8

36.0
56.0
31.3
15.3
34.3

52.5
79.2
45.9
28.7
50.8

Unweighted averages

94

13.2

34.6

51.4

(F) F2 Families from FiXsibs, All Grown in East Garden

F21.25

F22.29

F36.34

F37.37

F58.55

(8X20)FiXsib

(8XB)FiXsib

(AXlO)FiXsibs

(AXB)F,Xsib

(20Xl6)FiXsib

59
97
93
71
110

12.9
12.8
10.8
11.3
16.0

22.0
42.8
26.3
18.5
35.0

53.3
63.0
40.3
37.2
45.5

Unweighted averages

86

12.8

28.9

47.9

(G) Fs Families from F2 Selfed

F38.39

F40.42

F42.45

F44.46

F49.49

F51.52

F59.57t

F59.57

F61.59

F64.62

(A X20)F., selfed
(AX22)F.2 selfed
(A Xl6)Fo selfed
(A XB)F.. selfed
(16XA)F2 selfed
(16 X20)Fo selfed
(20 Xl6)Fo selfed
(20X16)F2 selfed
(20 XA)F2 selfed
(BX16)F.2 selfed

84
108

67

92
112

95
100
100
117
107

13.0
11.6
10.2
11.0
11.4
15.0
15.9
16.4
12.0
17.0

9.8
19.3

10.5

6.0

24.3

23.8

24.5

25.5

9.8

12.5

16.6
25.5
22.4
9.3
30.9
35.7
35.0
36.4
13.6
16.7

Unweighted averages

98.2

13.3

16.6

24.2

t This family was divided and this section was grown in the North Hill-field. All of the other families were
grown, as usual, in East Garden.

39

TABLE 2.1— Continued

Pedigree
Numbers

Parental Strains Involved

Number
of Stalks

Av. Num-
ber Grain-
Rows

Weights
in Lbs.

Yields
Bu./A.

(H) Fs Families from FiXSibs

F38.40

F40.43

F44.47

F49.50

F59.58

F61.60

F64.63

(AX20)F2Xsib

(AX22)F2Xsib

(AXB)F,Xsib

(16XA)F2Xsib

(20Xl6)F.2Xsib

(20XA)F,Xsib

(BXl6)F2Xsib

106
112

94
104

90
111
104

13.5
11.9
11.2
11.8
16.5
13.8
15.1

26.0
26.5
21.8
29.8
38.5
25.0
27.5

35.0
33.8
33.1
40.9
61.1
32.2
37.8

Unweighted averages

103

13.4

27.9

39.1

(I) Families from "Three-Way" and Iterative Crosses

F58.56

F74.95

F21.27

F22.30

F36.33

F27.38

F51.53

(20Xl6)FiX22

"Cobs"X(20Xl6)Fi

(8X20)FiX20

(8XB)F,XB

(AXB)FiXA

(AXB)F,XB

(16X20)F2X20

114
29
67

103
84
79

108

18.9
20.6
15.0
14.3
10.5
12.8
17.1

61.8
23.3
28.5
37.8
23.0
23.5
42.3

77.4
114.5
60.8
52.4
39.1
29.8
55.9

Unweighted averages (three-
way)

71.5

19.7

42.5

96.0

Unweighted averagesj (iter-
ative)

83.3

13.1

28.2

45.5

(K) Families from "Four-Way" Crosses, the So-called "Double-Cross"'

F21.26

F36.35

F69.66

F36 32§

(8X20)F,X(AXlO)Fi
(AX10)F,X(20X16)F,
(22X"Cobs")F,X(8XlO)Fi
(AXlO)FiX(AXB)Fi

67
106

75
102

12.7
12.8
16.3
11.2

28.5
47.0
58.5
45.5

60.8

63.3

111.4

63.7

Unweighted averages

87.5

14.3

44.9

74.8

(L) F3 Families from Four-Way F2 Crosses, and Imperfect
Iteratives of Same Form

F61.61

F38.41

F40.44

F44.48

F49.51

(20XA)F2X(BX16)F2

(AX20)F..X(AX22)F2

(AX22)F2X(AX16)F2

(AXB)F2X(16XA)F2

(16XA)F2X(16X20)F2

102
103
110

78
117

15.3
12.9
13.2
11.4
13.3

31.8
27.0
43.5
28.0
44.3

44.5
37.5
56.5
51.3
61.6

Unweighted averages

102

13.2

34.9

50.3

X Does not include F51.53.

§F36.32 is an imperfect 4-way, being partly iterative, involving only 3 inbreds.

40

BEGINNINGS OF THE HETEROSIS CONCEPT 41

planted in new plots of ground about one-fourth mile north of the original
Station grounds.

The purpose of this replication was to determine the degree of consistency
of results secured in these new locations with those recorded for the cultures
grown in the different conditions of soil, drainage, exposure, lighting, etc., in
the East Garden. Summaries of these two sections of Table 2.7 show the cul-
tures grown in the new plot with average grain-row number 1.29 per cent
higher than in the same families grown in the East Garden. However, the
East Garden cultures produced a higher average yield of ear-corn by 4.70
per cent.

Comparison between selfing and sibcrossing was made a subject of special
study in the inbred and Fi families in 1910. This was not continued in 1911
in the inbreds, but was given a further test in the derivation of the F2 families
from the Fi, and was carried forward to the derivation of F3 families from the
F2. These comparisons as they relate to Fi families are given in sections E
and F of Table 2.7. They show the F2 families derived from selfing their Fi
parents slightly superior to those Fo families produced from sibcrosses in
the Fi. This is indicated by an average grain-row number 3.1 per cent higher
and average yield 7.5 per cent higher in the Fo families from selfed Fi par-
ents, thus reversing the indications from the 1910 cultures.

The comparison of selfing versus sibcrossing in the production of the F3 by
these two methods of breeding in F2 can be derived from section G for selfings
and section H for the sibcrosses. Summaries of these two sections show a
superiority from sibcrosses of 0.4 per cent in average grain-row number and
61.6 per cent in yield. A part of this discrepancy is clearly due to the inclu-
sion of families in the selfed group which had no direct counterpart in the
sibcrossed group. If we limit the comparison to the families which are repre-
sented in both groups, we can avoid this cause of distortion. We then find
the sibcrossed families superior to the selfed by 1.5 per cent in grain-row
number, and 48.6 per cent in yields.

Comparative values between inbreds and crossbreds, as shown in sections
A and B of Table 2.7, and between crossbreds and Fi hybrids, are essentially
the same as in 1910. The ratios of inbreds, crossbreds, and Fi hybrids, with
respect to yields, is 0.29 to 1.00 to 1.22. Again the average grain-row number
is less in the Fi than in the crossbreds, and for the same reason. This particu-
lar group of Fi families came from parents with low average grain-row num-
bers, as compared with the broader parentage of the crossbreds.

The relationship of F.-s to Fo can now be noted by comparing the results
in sections G and H of Table 2.7, with sections D, E, and F. There are sev-
eral ways in which such comparisons can be made. Perhaps as good a way
as any is simply to combine all of the F2's together, regardless of the con-
siderations which led these to be tabulated in three separate sections, and
compare the results with all the F3 families of sections G and H likewise

42 GEORGE HARRISON SHULL

averaged in an undivided population. When treated in this way, we find that
the Fo's have an average grain-row number of 13.1 and average yields of 49.5
bushels per acre, while the F3 had an average of 13.4 grain-rows and pro-
duced an average of 30.4 bushels per acre. If we associate the average yield
of the Fi families, 79.0 with these values for Fo and F3, we see the beginning
of the characteristic curve in which the loss of yield from one generation to
the next is about twice as great as the loss for the next following generation .
It remains to consider the last three sections of Table 2.7, in which are

Fig. 2.6 — Total yields of ear corn of two selfed strains, Strain 16 and Strain 20, in the fore-
ground (exaggerated, of course, by foreshortening), and their Fi, F2, and F3 hybrids, left to
right, successively, in the background. As may be seen in Table 2.7, these yields, calculated
in terms of bushels per acre, are 12.76 and 21.82 for the two inbreds, and 83.21, 50.81,

and 36.43 for the three hybrid families.

included the results of more complicated crossing which had become possible
through the accumulation of simpler crossing in preceding years. In section I
are given two "three-way" crosses and four iterative crosses involving Fi
combinations and one iterative cross involving an F2 combination, each repre-
senting a cross between a hybrid and an inbred. As might be e.xpected, these
seven families although similar in form show no special consistency, since
they involve various combinations of five different inbreds and five different
hybrids.

In Table 2.7, section K, are presented what I believe to be the first "four-
way" or so-called "double crosses" ever made among inbreds. The elements
of one of these double crosses are shown in Figure 2.7. These double crosses
were made some five or six years before Dr. D. F. Jones pointed out the
potentialities of such crosses in producing hybridized seed corn at a price

X

X

10

(Ax 10) F,

Fig. 2.7 — One of the first four-zcay or double crosses ever grown from selfed strains of maize.
The single crosses for this double cross were made in 1909, the cross between the Fi's was
made in 1910, and the double-cross ear at bottom (G35.62) was grown in 1911 and grains

from it were used for i)lanting in 1912.

44 GEORGE HARRISON SHULL

that could make the pure-line method of corn production practical. No credit
is sought for the fact that I made these four-way crosses some years prior to
the similar combinations made by Dr. Jones. They are presented here only
because they belong in a historical account.

In the last section of Table 2.7 I have entered five families which have the
form of four-way crosses, but in which the single crossings used were Fo in-
stead of Fi. Only the first of these five families actually involved four differ-
ent inbreds, the others being partially iterative, in that only three inbreds
contributed to each. A comparison of the double crosses both of Fi and F2,
with the corresponding single crosses, is instructive. Comparison of the sum-
mary of section C with that of section K shows the double cross families
slightly inferior to the single cross families, as indicated by a 1 per cent higher
grain- row number and 6 per cent higher yield of the single cross families
over the double cross. Comparing sections L and E, it is to be noted that the
double cross retains the vigor of the F2, instead of declining to the vigor of
the F3 families produced by the usual methods, as seen in sections G and H,
Table 2.7.

In 1911 1 realized that the effective exposition of the important discoveries
we were making required photographs of prepared exhibits. A number of such
exhibits were set up and photographed, and have been presented in lantern
slides on many occasions. I have included the most instructive of these here.

Here the detailed account of these studies must end, for although they
were continued in 1912, 1 have been unable to locate the field and harvesting
notes including grain-row counts and weighings for the 1912 cultures. These
1912 cultures were especially designed to explore the evidences of Mendelian
segregations in the F2 and the F3 families, with respect to grain-row num-
bers and yields. They included 11 families of the breeding Fi X self, 8 families
of Fi X sib, 21 F2 X self, 10 Fo X sibs, and five families of F3 X self. There
was also an interesting pair of approximations to eight-way combinations or
quadruple crosses produced by reciprocal combinations of the four-way
crosses included in the 1911 cultures. While these had the form of quadruple
crosses, they were imperfect in that one of the inbreds was repeated, so that
only seven different inbreds were represented, instead of eight. This was in-
evitable since I initiated only seven inbred lines in the beginning of these
experiments.

The 1912 crop completed the experimental work with corn at the Station
for Experimental Evolution, and I spent the next year in Berlin, Germany.
In a lecture I gave at Gottingen about three weeks before the beginning of
the first World War the word heterosis was first proposed. I used the occasion
to discuss the bearing of the results of these studies on the practical work of
breeders of various classes of organisms, both plant and animal. I stressed
the point that the breeder should not be content, as had long been the case,
to seek merely to avoid the deterioration incident to inbreeding, but should

10

18

10

13

1904 1905 1906 1507 1908 1909 I'JiO igu 191.5

SELECTION FOR ROW-NIJHBER
IN CROSS-FERTILIZED MAIZE

"■?58=^

^;2^"-

»-^.

;o>*\

SELECTIOtJ FOR RaJ-WIKDER
IN SELF-FEHTILIZEU MIZE

1904

1905

1906

1907

1908

1909

1910

1911

1912

Fig. 2.8 — Diagrams of the progressive results of selection for grain-row number under the
two systems of breeding: selling completely prevented in the upper diagram; selling the
sole method of breeding in the lower. The numbers on the lines indicate the numbers of
rows of grains on the parent ears. The circles show by their position on the scale at left the
average grain-row numbers of the resulting [progenies.

46

GEORGE HARRISON SHULL

recognize in heterosis a potent source of practical gains, to be investigated,
understood, and utilized as a new tool in deriving from plant and animal
life their maximum contributions in the service of man.

Although no further experimental work was done with corn at the Station
for Experimental Evolution after 1912, I tried to resume the work in my
first two years at Princeton University, by planting 77 cultures of pedigreed

8

V, 7

u

■; i

&

^3^

«»

V *

f

V-

♦ • <

3

*"

z

^H*-^'

V,

'^ /' %fi> X ^h QA >S^

A

^^^

3 /^ X B

S 3.

Fig. 2.9 — Ears of my white dent "strain" of corn grown at Princeton University in 1916.
The ears, each typical of the progeny to which it belonged, are from left to right : SA, Shull's
Strain A; SA X BA, Fi hybrid between Shull's Strain A and Biakeslee's "branch" of the
same strain; BA X SA, reciprocal of the last; BA, Shull's Strain A, after two successive
sellings by Dr. A. F. Blakeslee; B.\ X B, Fi between Biakeslee's branch of Strain A and
Shull's Strain B; and SB, Shull's Strain B. About as much heterosis is shown by a cross be-
tween two sub-hnes of Strain A as between one of these sub-lines and Strain B, the impli-
cation being that something more specific may be involved in this example of heterosis than
the mere number of genetic difTerences. (Photo by W. Ralph Singleton in 1945.)

corn in 1916 and 65 in 1917. 1 used some of the materials from these cultures
for laboratory studies in biometry in my classes in genetics. The interesting
results shown in Figure 2.9 are from my 1916 crop at Princeton. The plantings
at Princeton were made late and the young plants were decimated by pigeons
and crows, so that some valuable connections were lost, and with them some
of my interest in their continuation.

As we all know, heterosis is not limited to corn, and my own interest in
the matter was in no wise restricted to its manifestation in corn. There were
examples presented in many other of my genetical experiments. I was par-
ticularly interested in the discovery of such special mechanisms as balanced
lethal genes in the Oenotheras and self-sterility genes in Capsella grandiflora

BEGINNINGS OF THE HETEROSIS CONCEPT

47

which, along with many types of asexual reproduction including partheno-
genesis, specifically enable the organisms possessing these special mecha-
nisms to maintain the full advantages of heterosis. On one occasion, one of
my new hybrid combinations in Oenothera happened to be planted through
an area in my experimental field where the soil had become so impoverished
that none of my other cultures reached their normal growth. Many of the

Fig. 2.10 — The Fi hybrids between a cultivated form of Helianthus anniius and a wild form
of the same species received from Kansas. This photograph, taken at the Station for Ex])eri-
mental Evolution in 1906, shows the author alfixing a glassine hag to a head of one of the
hybrid plants. The two parents of this hybrid averaged from 5 to 6 feet tall, while 51 of
these Fi hybrids, measured on August 28, 1906, ranged in height from 6.7 to 14.25 feet, the
average being 10.46 feet. This may be considered my iirst experience with hybrid vigor.

48 GEORGE HARRISON SHULL

plants remained rosettes or formed only weak depauperate stems. But this new
hybrid became a vigorous upstanding form in this impoverished area as well
as on better soil elsewhere. I recorded this as a notable example of making
heterosis take the place of manure or commercial fertilizers.

Figure 2.10 is a notable hybrid, which represents my first direct personal
contact with a recognized case of hybrid vigor. This hybrid resulted from a
cross I made in 1905 between the so-called "Russian" sunflower and the wild
Helianthus annuus of our western prairies. Both of these forms have been re-
ferred, botanically, to the same species. Both are of approximately equal
height, scarcely as tall as the six-foot step-ladder shown in the figure. The
tallest of these Fi hybrids was 14.25 feet in height.

Returning now to the question which I sidestepped in the beginning —
what we mean by the expression the heterosis concept — I suggest that it is the
interpretation of increased vigor, size, fruitfulness, speed of development,
resistance to disease and to insect pests, or to climatic rigors of any kind,
manifested by crossbred organisms as compared with corresponding inbreds,
as the specific results of unlikeness in the constitutions of the uniting parental
gametes.

I think the first clear approach to this concept was involved in a statement
which I have already quoted, that "a different explanation was forced upon
me" (in my comparisons of cross-fertilized and self-fertilized strains of
maize). That is, "that self-fertilization simply serves to purify the strains,
and that my comparisons are not properly between cross- and self-fertiliza-
tion, but between pure strains and their hybrids." Since heterosis is recog-
nized as the result of the interaction of unlike gametes, it is closely related to
the well known cases of complementary genes. It differs from such comple-
mentary genes, however, mainly in being a more "diffuse" phenomenon in-
capable of analysis into the interactions of specific individual genes, even
though it may conceivably consist in whole or in part of such individual
gene interactions.

H. K. HAYES

University of Minnesofa

Chapter 3

Development of
the Heterosis Concept

Hybrid vigor in artificial plant hybrids was first studied by Koelreuter in
1763 (East and Hayes, 1912). The rediscovery of Mendel's Laws in 1900
focused the attention of the biological world on problems of heredity and led
to renewed interest in hybrid vigor as one phase of quantitative inheritance.

Today it is accepted that the characters of plants, animals, and human
beings are the result of the action, reaction, and interaction of countless
numbers of genes. What is inherited, however, is not the character but the
manner of reaction under conditions of environment. At this time, when
variability is being expressed as genetic plus environmental variance, one
may say that genetic variance is the expression of variability due to geno-
typic causes. It is that part of the total variance that remains after eliminat-
ing environmental variance, as estimated from studying the variances of
homozygous lines and Fi crosses between them.

Early in the present century. East, at the Connecticut Agricultural Ex-
periment Station, and G. H. Shull at Cold Spring Harbor, started their
studies of the effects of cross- and self-fertilization in maize. The writer has
first-hand knowledge of East's work in this field as he became East's assist-
ant in July, 1909, and continued to work with him through 1914. In 1909,
East stated that studies of the effects of self- and cross-pollination in maize
were started with the view that this type of hiformation was essential to a
sound method of maize breeding. In addition to studies of maize, which is
normally cross-pollinated, East carried out studies in tobacco of crosses be-
tween varieties and species. This gave an opportunity of studying the effects
of self- and cross-pollination with a self-pollinated jilant. A 1912 j)ublication
of East and Hayes made the following statement:

The decrease in vigor due to inbreeding naturally cross-fertilized species and the increase
in vigor due to crossing naturally self-fertilized species are manifestations of one phenome-
non. This phenomenon is heterozygosis. Crossing produces heterozygosis in all characters
by which the parent plants differ. Inbreeding tends to produce homozygosis automatically.

49

50 H. K. HAYES

Several photographs from this bulletin are of some interest. A picture of
two inbred lines of maize and their Fi cross was one of the first published field
views of hybrid vigor from crossing inbred lines of maize. East told me that
such a demonstration of hybrid vigor would create a sensation if the material
had been grown in the corn belt.

Some Fi crosses between species and sub-species in tobacco gave large in-
creases in vigor. Some species crosses were sterile. Some varietal crosses
within species showed little or no increase in vigor, other crosses gave an aver-
age increase of 25 per cent in height over the average of their parents. A few
wide species crosses were very low in vigor. One such cross beween Nicotiana
tabacum and Nicotiana alata graiidiflora was sterile and very weak in growth.
Photographs of the parents and hybrids bring out the fact that a lack of vigor
in a few cases was known to accompany the heterozygous condition. Natural-
ly such undesirable combinations had little importance either to the plant
breeder or as a basis for evolution.

In 1910, G. H. ShuU summarized the effects of inbreeding and crossbreed-
ing in maize in a clear, concise, and definite manner. The student of heredity
in this early period had little conception of the complexity of inheritance.
Hybrid vigor was in many cases not clearly Mendelian. The term heterosis
was coined by Shull and first proposed in 1914. He used the term to avoid
the implication that hybrid vigor was entirely Mendelian in nature and to
furnish a convenient term to take the place of such phrases as "the stimulus
of heterozygosis."

At this time it was usually stated that increased vigor in hybrids was due
to a more rapid cell division as stimulated by the heterozygous condition of
the genotype. A. F. Shull in 1912 attributed the vigor "to the effect of a
changed nucleus and a (relatively) unaltered cytoplasm upon each other."

The purpose of this chapter is to discuss some phases of the development
of the heterosis concept since 1910. Three main topics will be presented cover-
ing utilization, breeding methods, and genetic concepts with particular ref-
erence to practical applications and to genetic explanations.

UTILIZATION OF HETEROSIS BY THE PRODUCER

The presentation of East and Hayes in 1912 emphasized the probable
practical value of heterozygosis. A review of experiments with maize was
made. In discussing Shull's (1909) plan for the use of single crosses between
inbred lines, it was stated that the procedure was desirable in theory but
difficult of application. At this early time the inbred lines of maize that were
available seemed so lacking in vigor that the use of Fi crosses between selfed
lines in maize for the commercial crop seemed impractical. Both Shull and
East believed that some method of direct utilization of hybrid vigor in maize
would be found.

One is inclined to forget that the inbred lines of maize of today are marked-

DEVELOPMENT OF THE HETEROSIS CONCEPT 51

ly superior, on the average, to those of 1910. Jones's discovery about 1917
of the double cross plan of producing hybrid seed in maize, and the subse-
quent proof by many workers that double crosses can be obtained that closely
approach the vigor of Fi crosses between selfed lines, furnished the basis for
the utilization of hybrid vigor in field corn. With sweet corn, however, F]
crosses between selfed lines are used very widely today for the commercial
crop.

East and Hayes emphasized that Fi crosses probably would be of com-
mercial value in some truck crops where crossing was easy. Eggplants, to-
matoes, pumpkins, and squashes were considered to offer promise for a prac-
tical use of such vigor. The writers also mentioned the fact that heterozygosis
had been used in vegetatively propagated plants, though not purposely, and
that it seemed feasible to make a practical application in the field of forestry.

The use of heterosis in practical plant and animal improvement has borne
out and surpassed these early predictions as shown in Table 3.1.

TABLE 3.1

USE OF HETEROSIS IN CROP PLANTS AND LIVESTOCK

Farm crops: Maize, sugar beets, sorghums, forage crops, and grasses

Horticultural crops: Tomatoes, squashes, cucumbers, eggplants, onions,

annual ornamentals
Silkworms

Livestock: Swine, poultry, beef and milk cattle

Vegetatively propa-
gated plants

In the corn belt of the United States nearly 100 per cent of all maize is
hybrid. Hybrid corn is rapidly being developed in other countries of the
world, and is one of the best illustrations of the practical utilization of mod-
ern genetics. Considerable evidence leads to the conclusion that heterosis can
be used extensively in farm crops, including such widely different plants as
sugar beets, sorghums, tobacco, forage crops, and grasses.

With horticultural plants, where the individual plant is of rather great
value, planned heterosis has proven worth while. First generation crosses
of tomatoes, onions, egg plants, cucumbers, and squashes have proven their
value and are being grown extensively by home and truck gardeners. Similar
use is being made of heterosis in some annual ornamentals.

Heterosis has become an important tool of the animal breeder. Its use in
silkworm breeding is well known. Practical utilization of hybrid vigor has
been made in swine and poultry, and applications are being studied with beef
cattle, dairy cattle, and sheep. A somewhat better understanding of the
effects of inbreeding and crossing by the breeder has aided in applications
with livestock. As in plants, inbreeding makes controlled selection possible,
while controlled crosses may be grown to utilize favorable gene combinations.

52 H. K. HAYES

METHODS OF BREEDING FOR HETEROSIS

In general there is a much closer relation between the characters of par-
ents and of their Fi crosses in self-pollinated plants than between the char-
acters of inbred lines of cross-pollinated plants and their Fi crosses.

Characters of Parents and Fi Crosses in Self-pollinated Plants
A recent study by Carnahan (1947) in flax, which is normally self-polli-
nated, may be used for illustrative purposes. Four varieties of flax were se-
lected to represent desirable parental varieties. Each was crossed with four
other varieties, of different genetic origin from the first group, to be used as
testers. Sufficient seed for Fi and F2 progenies was produced so that all

TABLE 3.2

PARENT AND Fi CROSSES, YIELD
IN BUSHELS PER ACRE*

Parent Tester Varieties

Varieties 5 6 7 8

16 14 17 13

1

19

31

25

22

19

2

18

24

26

19

20

3

13

26

24

20

18

4

17

22

21

20

19

* Parent yields outside rectangle, Fi crosses
within.

progenies could be planted in replicated, 8-foot rows at the rate of 200 seeds
per row. Combining ability was studied in Fi and F2 in comparison with the
parents for yield of seed, number of seeds per boll, number of bolls per plant,
weight of 1000 seeds, date of full bloom, and plant height.

As shown in Table 3.2, each Fi cross yielded more than its highest yielding
parent, although for one cross the difference was only slightly in favor of
the Fi. For an average of all crosses, the Fi yielded 40 per cent more than the
average of the parents, and the Fo, 26 per cent more. The lowest yielding
cross, 3X8, was produced from a cross of the two lowest yielding parents.
The highest yielding cross, 1X5, however, could have been selected only
by actual trial. It was obtained by crossing the highest yielding selected
variety with the second highest yielding tester variety.

There was excellent agreement, on the average, for each of the characters
studied between the average expression of the characters of the parents and
their Fi crosses. Carnahan concluded that for each character studied there
appeared to be a good relationship between the performance of the parents
and the average performance of their Fi crosses. The characters of the par-
ents in this study were as good or better indication of the combining ability
of a parental variety as that obtained from a study of average combining
ability in four crosses.

DEVELOPMENT OF THE HETEROSIS CONCEPT

53

Powers (1945) obtained also relatively good agreement in tomatoes be-
tween the parental yield of 10 varieties and that of all possible Fi crosses
between the 10 varieties (see Table 3.3).

Moore and Currence (1950) in tomatoes made a somewhat comparable
study to that of Carnahan with flax. They used two three-way crosses as
testers for a preliminary evaluation of combining ability of 27 varieties.
Based on this, eight varieties were selected that gave a wide range in aver-
age combining ability for sev^eral characters including early yield and total
yield. These varieties were crossed in all combinations, and yield trials of the

TABLE 3J

YIELD OF RIPE FRUIT IN GRAMS
IN TOMATOES (AFTER POWERS)

Yield of Ripe Fruits (per Plant)

Variety or Inbred

Variety or Inbred

9 Crosses (av.)

Grams

Grams

L. escidsntum

Bounty 4101

513+ 39

1280 + 53

4102

607+ 86

1267 + 46

4105

332+ 64

1081+33

4106

828 + 108

1236 ±45

Es.XL. pirn

4103

1066+159

1597 + 54

4104

808 + 114

1340 + 44

4107

801 + 111

1181+47

4108

857 + 108

1192+41

4109

1364+151

1968 + 46

4110

1868 + 149

2231 + 52

varieties and Fi crosses were made. There was relatively good agreement
between the early test for combining ability and the average yield of Fi
crosses, but the relationship did not seem superior to the varietal performance
as a means of predicting combining ability in crosses. In the studies by Carna-
han, Moore, Currence, and Powers the only means of selecting the most de-
sirable Fi cross was by actual trial.

Characters of inbred Lines and Their Fi Crosses in Maize

Numerous studies have been made with maize of the relation between
characters of inbred lines and of their Fi crosses. There usually have been
indications of significant correlations for most characters of inbred lines and
their Fi crosses. In most cases, however, the relationship was not very large
or highly important when one studied individual characters, or the more com-
plex character — yield of grain. The studies have been reviewed by numerous
workers (see Sprague, 1946b).

54 H. K. HAYES

Hayes and Johnson (1939) in Minnesota studied the relation between the
characters of 110 inbred lines of maize and their performance in top crosses.
The characters studied in selfed lines in replicated yield trials are given in
Table 3.4.

All possible correlations were made between the individual characters of
the inbreds and of these characters and the yield of grain of top crosses. The

TABLE 3.4

CHARACTERS OF 110 INBRED LINES IN
CORN CORRELATED WITH INBRED-
VARIETY YIELDING ABILITY

1. Date silked 7. Stalk diameter

2. Plant height 8. Total brace roots

3. Ear height 9. Tassel index

4. Leaf area 10. Pollen yield

5. Pulling resistance 11. Grain yield

6. Root volume 12. Ear length

TABLE 3.5

TOT.\L CORRELATIONS BETWEEN CHARACTERS OF 110 INBREDS,
LABELED 1 TO 12, AND YIELDING ABILITY OF INBRED-
VARIETY CROSSES DESIGNATED AS 15

Characters Correlated

2 3 4 5

6

7

8

9

10

11

12

15

1

0.51 0.61 0.48 0.65

0.62

0.55

0.38

0.37

0.22

0.07

-0.06

0,47

2

0.76 0.44 0.48

0.43

0.40

0.26

0.19

0,36

0.25

0.08

0.27

3

0.43 0.54

0.50

0.41

0.35

0.33

0.22

0.15

-0.01

0.41

4

0.50

0.44

0.48

0.40

0.29

0.18

0.20

0.08

0.29

5

0.76

0.51

0.60

0.41

0.21

0.15

0.04

0.45

6

0.55

0.74

0.39

0.29

0.19

0.03

0.54

7

0.54

0.24

0.27

0.21

0.15

0.41

8

Multiple value of R

0.26

0.22

0.20

0.07

0.45

9

for inbred-variety yield

0.20

-0.00

0.03

0.19

10

and twelve characters of

0.35

0.32

0.26

11

inbred = 0.67

0.64

0.25

12

0.28

Significant value of r for P of .05 = 0.19.
Significant value of ;■ for P of .01 = 0.25.

characters, in general, were those that were considered to evaluate the in-
breds in developmental vigor.

The total correlations between characters are summarized in Table 3.5.
Most correlations were significant at the 5 per cent or 1 per cent point ex-
cept the relation between ear length and other characters of the inbreds. All
relationships between the characters of the inbreds, including grain yield, and
the yield of top crosses were significant at the 1 per cent point except for
tassel index of the inbreds, and that was significant at the 5 per cent point.
The multiple correlation coefficient of 0.67 indicated that under the condi-
tions of the experiment about 45 per cent of the variability of inbred-variety

DEVELOPMENT OF THE HETEROSIS CONCEPT

55

yield was directly related to characters of the inbreds. These relationships
between the parents and their Fi crosses were somewhat larger than those
obtained by others with maize. Nevertheless, relationships were much
smaller than has been obtained in similar studies with self-pollinated plants.

Richey (1945b) compared the yield of inbred parents in the S3 and S4 gener-
ations of selling with the mean yield of their single crosses from data taken
by Jenkins and Brunson. Similar comparisons were made between the yield
in top crosses and the mean yield in single crosses (see Table 3.6).

Although for various reasons the r values are not strictly comparable, the
yield of inbreds was as strongly correlated with the mean yield of their
single crosses as the yield in top crosses was correlated with the mean yield
of single crosses.

TABLE 3.6

CORRELATION COEFFICIENTS FOR YIELDS OF

INBRED PARENTS OR TOP CROSSES WITH

MEAN YIELDS OF SINGLE CROSSES*

Hybrids
Correlated

Previous Generations
OF Inbreeding

WITH

S3t

84

Inbred parents
Top crosses

.25, .64, .67
.53

.41, .45
.53

* After Richey, after Jenkins and Brunson.
t S3 = three years selfed, etc.

Comparison of Methods with Self- and Cross-pollinated Plants

In self-pollinated plants it seems probable that the first natural step in
the utilization of heterosis normally may consist of the selection of available
parental varieties that in themselves produce the best combination of char-
acters. It seems important to continue breeding for the best combination of
genes that can be obtained in relatively homozygous varieties. Where hybrid
seed can be produced cheaply enough, or new methods can be found to
make crosses more easily, heterosis can be used to obtain from the hybrid an
advance in productivity over the homozygous condition.

In cross-pollinated plants two general methods of breeding for heterosis
are now being widely utilized. One consists, as in maize, of the selection with-
in and between selfed lines and the use of single, three-way, or double crosses
for the commercial crop. The second general method consists of selecting
or breeding desirable clones of perennial crops. These are evaluated for com-
bining ability by polycross, or other similar methods, and the desirable clones
used to produce Fi crosses, double crosses, or synthetic varieties.

56 H. K. HAYES

There seems to be some difference of opinion regarding the selection proc-
ess in its application to maize improvement. One school of thought practices
a somewhat similar method of breeding selfed lines as is used in self-pollinated
plants, with the viewpoint that controlled selection makes it possible to iso-
late in the inbred lines the genes for characters needed in the hybrids. Ap-
parently the relationship between the characters of inbreds and their Fi
crosses will become greater as inbred lines themselves improve. The other
extreme of viewpoint (Hull, 1945a) is that the greater part of hybrid vigor is
due to interallelic interaction of genes to such an extent that selection based
on appearance may be harmful. In a recurrent selection program Hull,
therefore, does not recommend selection for vigor of growth, although he
states that plants showing pest or weather damage should be avoided.

It is probable that differences between these two so-called schools may
have been overstated. Both believe that the actual test for combining ability
in hybrid combination is necessary. The stage in the breeding program when
such test should be made will depend on the material worked with and the
nature of the breeding program. In both cross- and self-pollinated plants an
actual trial will be needed to determine the combination that excels in
heterosis.

Where clonal lines can be propagated vegetatively, a method of selecting
for heterosis in alfalfa was suggested by Tysdal, Kiesselbach, and Westover
(1942), by means of polycross trials. The method is being used extensively
today with perennial forage crops that normally are cross-pollinated. The
writer is studying the method with early generation selfed lines of rye. With
perennial crop plants, selection for combining ability is made for heterozy-
gous parent clones. Where disease and insect resistance or winter hardiness
are important, it may be essential to insure that the clones used in the poly-
cross trials excel for these characters. Polycross seed is produced on selected
clones under open-pollinated conditions where the clones are planted together
at random under isolation.

In one study of progenies of eight clones by Tysdal and Crandall (1948)
yields were determined from polycross seed in comparison with top cross seed
when each of the clones was planted in isolation with Arizona common alfalfa
(see Table 3.7). The agreement for combining ability was relatively good in
the two trials.

An early suggestion of utilization of heterosis in alfalfa was by double
crosses, from single crosses between vegetatively propagated clones, without
entire control of cross-pollination. Synthetic varieties also have been sug-
gested as a means of the partial utilization of heterosis. In one comparison
the progeny of a synthetic combination of four clones of high combining
ability yielded 11 per cent more forage than a similar combination of four
clones of low yielding ability. A recent comparison of eight synthetics led
Tysdal and Crandall to conclude that the first synthetic and second syn-

DEVELOPMENT OF THE HETEROSIS CONCEPT

57

thetic seed progenies gave about the same forage yield. In this comparison,
heterosis continued through the second seed increase of the high yielding
synthetic.

Other Studies with Maize

Combining ability, that is ability to yield in hybrid combination, has been
shown by various workers to be an inherited character (Hayes and Johnson,
1939), (Cowan, 1943), (Green, 1948). It seems feasible to breed for high com-
bining ability as for other quantitative characters. In the breeding program

TABLE 3.7

FORAGE YIELDS OF POLY-
CROSSES COMPARED TO
TOP CROSSES OF THE
SAME CLONES*

Yield Relative to

Grimm

AS 100

Clone No.

Arizona

Polycross

Top Cross

1

121

130

2

111

122

3

101

117

4

'J9

103

5

97

105

6

96

101

7

89

101

8

76

101

* After Tvsdal and Crandall.

for the production of imj^roved inbred lines, it is often possible to select as
parents of crosses, select lines having high combining ability as parents of
crosses, in addition to selection for other characters that are desired. In
breeding for heterosis, however, it seems evident that genetic diversity of
parentage is equally as important as combining ability (see Hayes, and
Immer, 1942; Sprague, 1946b).

All relatively homozygous, inbred lines in maize are much less vigorous
than the better Fi crosses. It is apparent that heterosis is of great impor-
tance in crosses with inbred lines of maize.

Inbred lines that have undesirable characters may be easily imjjroved by
the application of any one of several methods of breeding. The breeder may
select for each problem the method or methods that seem to him most ap-
plicable. In breeding selfed lines the selection of parents that have comple-
mentary characters that together include the characters desired in the im-
proved inbred is a natural first step. Subsequent methods of breeding may

58

H. K. HAYES

be used according to the viewpoint of the breeder and the particular prob-
lem to be solved.

While combining ability is an inherited character, it seems of special in-
terest that single crosses of high X high combiners have not been greatly su-
perior in yield, on the average, to crosses of high X low. Both, however, were
clearly higher in yielding ability than low X low crosses (Johnson and
Hayes, 1940), (Cowan, 1943), (Green, 1948). An illustration from Johnson
and Hayes (Table 3.8) shows the type of results obtained. The crosses were
classified for yielding ability in comparison with recommended double
crosses of similar maturity.

Two recent studies in Minnesota may be used to illustrate other breeding
problems. A further study was made by Johnson (1950) of the combining
ability of F4 lines that were studied in earlier generations by Payne and
Hayes (1949). Yield relations in the double cross Min. 608 (A344 X A340)
(A357 X A392) are illustrated in Table 3.9.

TABLE 3.8

FREQUENCY DISTRIBUTION FOR YIELD OF SINGLE CROSSES

OF SIMILAR MATURITY IN COMPARISON WITH

RECOMMENDED DOUBLE CROSSES AS 0

Type of

Class Centers of —1 to —2, +1 to +2, etc. Times
THE S.E. OF A Difference

Cross

-7
-8

-5
-6

-3
-4

-1

-2

0

+ 1
+2

+ 3
+4

+ 5
+ 6

+ 7
+ 8

Total

Mean

Low X low .

1
3
1

1

....
"5

2

11

12

4 4

12
52
83

-0 5 + 0 7

Low X high

HighXhigh

1

6

8

16
35

9
20

5
4

1

+ 1.1 + 0.4
+ 1 1+0 2

TABLE 3.9

YIELD RELATIONS IN MIN. 608

( A334 X A340) (A357 X A392)

% M.

Yield
(Bu.)

A334XA357 and A392

A340XA357 and A392

19.6

18.5

66.8
62.4

Average

19.0

64.6

A357XA334and A340

A392XA334and A340

19.5
18.6

66.0
63.2

Average

19.0

64.6

Min. 608

19.0

64.0

DEVELOPMENT OF THE HETEROSIS CONCEPT

59

In these studies the usual method of predicting combining ability of a
double cross gave excellent agreement between both predictions and the
actual double cross yield.

The studies of the performance in early and later tests of F2 to F4 lines
from L317 X A116 when crossed with (A334 X A340) in comparison
with A357(A334 X A340) were carried out by Payne and Johnson. The
methods of comparing combining ability in different generations were
adapted by the writer, who alone is responsible for the conclusions drawn.
The lines were first placed in +1, — 1, etc. X L.S.D. at the 5 per cent point
with the performance of A357(A334 X A340) as 0. Classes for performance
of individual lines were made by adding the yield class of a line to its moisture
class with the sign of the latter changed.

The F2 and F3 crosses were both grown the same year, the F3 and F4 were
grown in different years, and the F4and the top crosses were grown the same
year (see Tables 3.10, 3.11, 3.12).

In these studies no new lines seemed markedly superior to A357 in com-

TABLE 3.10

COMBINING ABILITY RELATION OF F2 AND F3 LINES

OF (L317XA116) IN CROSSES WITH (A334XA340)

GROWN IN SAME TRIAL IN 1947

o
•-t

o

+2

1

0

2

2

1

1

-1

1

2

1

1

-2

1

2

3

-3

1

1

1

3

1

-4

?

2

-5

2

-7

2

+2 +1 0 -1 -2 -3 -4 -5
F2 crosses, performance classes

Total
1
6

5
6
7
4

2
2

3i

TABLE 3.11

COMBINING ABILITY RELATION OF F3 AND F4 LINES

OF (L317XA116) IN CROSSES WITH (A334XA340)

F3 GROWN IN 1947, F4 IN 1949

Total

o

o

en

+2

1

0

1

2

2

1

-1

5

-2

1

1

3

1

-3

1 1

2

3

-4

1

1

2

-5

1

1

-7

1

I

+2 +1 0 -1 -2 -3 -4
F4 crosses, performance classes

1
6

5
6
7
4
2
2

33

60

H. K. HAYES

bining ability with (A334 X A340). As A357 is rather outstanding in com-
bining ability the result may not be so surprising. There was much greater
relation between the combining ability of F3 and F4 lines and of F4 with top
crosses than between F2 and F3.

In an unpublished study of gamete selection, with a different but highly
desirable double cross, there was an indication that a lower yielding inbred
could be improved by an application of gamete selection (Stadler, 1944).
The study is from one phase of a breeding program to improve Min. 406.
The yield relations of inbreds in an average of single crosses are given in
Table 3.13.

Approximately 60 Fi plants of A25 X Golden King were selfed and top
crossed with A73 X A375. Thirty-two of the more desirable plants were se-
lected to study in yield trials. In this study both yield and moisture classes
of plus 1, plus 2, etc. X L.S.D. at 5 per cent were used around the mean of

TABLE 3.12

COMBINING ABILITY RELATION OF F4 LINES OF

(L317XA116) IN CROSSES WITH (A334XA340) AND

WITH GOLDEN KING. GROWN IN 1949

Total

In

A334
X

A340
Crosses

+ 2

1

+ 1

1

0

1 3

1

1

-1

1

-2

9

1

3

-3

3

4

1

-4

1

-5

1

1

+ 1 0 -1 -2 -3 -4 -5
In Golden King Crosses

1
1
6
1
13
8
1
2

33

TABLE 3.13

GAMETE SELECTION IN THE IMPROVE-
MENT OF MINHYBRID 406

(A25XA334)(A73XA375)

Av. OF Crosses

%M.

Bu.

A25XA73, A375

A334XA73, A375

A73XA25, A334

A375XA25, A334

24.6
24.7

24.6
24.7

76.2
79.4

74.8
80.8

Proposal for improvement of A25 and A73:
A25XG. King gametes
A73XMurdock gametes

DEVELOPMENT OF THE HETEROSIS CONCEPT

61

A25 X tester as 0. The results (see Table 3.14) indicate that gametes from
Golden King are a desirable source of improvement of A25 in crosses with
A73 X A375.

From this first trial three high and three low yielding lines were selected,
and selfed progeny grown in Si. Plants in each of the three Si high and three
low combining lines were selected, selfed, and again top crossed on A73 X
A3 75. The agreement for So and Si lines was very good (see Table 3.15).
It appears that gamete selection is an excellent breeding method for the
early selection of material to improve the specific combining ability of a
known inbred.

SOME GENETIC CONCEPTS OF HETEROSIS
It seems very evident to the writer that heterosis, the increased vigor of F]
over the mean of the parents or over the better parent, whichever definition
is "used, is not due to any single genetic cause. A brief summary of various

TABLE 3.14

DISTRIBUTION OF % MOISTURE AND YIELD OF 32 So
PLANTS OF A25XG. KING CROSSED TO A73XA375.
CLASSES OF L.S.D. 5% AROUND MEAN OF A25X
TESTER

%ear
mois.

+2
+ 1

1

3

2

-1

-2

2
3

8

5

5
3

— 2

-1

+ 1 +2

(mean of A25X
tester)

Yield
(mean of A25X tester)

TABLE 3.15

PERFORMANCE INDICES OF So AND Si LINES
FROM A25XG. KING WHEN CROSSED TO
A73XA375 TESTER AND COMPARED WITH
A25XTESTER

So

Si

Gamete

No. OF

Number

Si's

1947

1949

1949

19 H

+ 11

+ 19

+25

5

20 H

+ 14

+ 9

+ 14

7

36 H

+ 9

+ 16

+ 11

7

5L

-11

- 3

+ 5

7

29 L

-11

- 1

- 0

1

46 L

- 5

+ 1

+ 2

7

62 H. K. HAYES

theories advanced to explain heterosis seems desirable to set the stage for
later discussions. Bruce (1910) explained heterosis on the combined action
of favorable dominant or partially dominant factors, based as Richey (1945a)
has emphasized on mathematical expectations.

Keeble and Pellew (1910) used a similar hypothesis on a di-hybrid basis
to explain hybrid vigor in peas. East and G. H. Shull (1910-1914) believed
vigor was dependent on heterozygosis on the basis that the stimulus of hy-
bridity was not entirely Mendelian. A. F. Shull (1912) preferred the explana-
tion that heterosis was due to a stimulus resulting from a changed nucleus
on a relatively unaltered cytoplasm. Jones (1917) restated Bruce's concept
and added the concept of linkage.

Collins (1921) and Richey (1945) have pointed out that where large num-
bers of factor pairs are involved it would be very difficult to recover all fac-
tors in a favorable condition in F2, or in later segregating generations. With
multiple factors involved, however, linkage must of necessity make the re-
combination of factors more difficult. East (1936) presented a Mendelian
concept of the interaction of alleles at the same locus to explain heterosis,
where two alleles of a particular gene pair had each developed a divergent
physiological function. The writer believes he continued also to accept the
previous explanation that heterosis was dependent on the cumulative effect
of dominant or partially dominant linked genes.

Gustafsson (1947), Hull (1945a), Jones (1945), Castle (1946), and others
have emphasized the importance of interallelic action in relation to heterosis.
Castle has suggested also that the effect of interallelic action of a single pair
of genes "is similar to that of the killer mutation of Sonneborn, except that
the action induced in the dominant gene by its sensitized recessive, instead
of being harmful, in this case is beneficial."

In certain cases a homozygous recessive pair of genes may completely
modify the normal expression of either a homozygous or heterozygous or-
ganism. Homozygous dwarfs in maize condition such a result. A cross be-
tween two different dwarfs, however, releases the inhibition of each dwarf
and results in marked heterosis. Both dominant factors, where two dwarfs
are crossed, appear to be necessary to condition normal development. In this
case the dominant conditions of both factor pairs act as complementary fac-
tors for normal growth.

It is evident that genes are greatly affected in their expression by differ-
ences in both external and internal environment. Cytoplasmic inheritance of
male sterility may be used for illustrative purposes. Several cases of male
sterility in sugar beets and onions, for example, are known that are due to
maternal cytoplasmic inheritance which may be modified in expression by
the dominant or recessive condition of one or more factor pairs.

Recently Hsu (1950) at Minnesota has studied the effect of two pairs of
dwarf factors of maize in their homozygous dominant and recessive condi-

DEVELOPMENT OF THE HETEROSIS CONCEPT

63

tions, and also when heterozygous in near isogenic, homozygous, and highly
heterozygous backgrounds.

The factor pair for Didi was studied in the near isogenic background of
inbred A 188, that of D^dx in the near isogenic background of A95-344, and
both factor pairs were studied in crosses between A 188 X A95. Particular
attention was given to total dry matter produced at various periods of growth
under field conditions and to the growth in length of the coleoptile and meso-
cotyl under controlled laboratory conditions.

One comparison of the growth of the mesocotyl during a 12-day period
for DiDi and Didi on three different near isogenic backgrounds will be con-
sidered: the near isogenic background, A188, and the highly heterozygous
backgrounds of A 188 X A95 in the presence of D^D^ and D^d^, respec-
tively. While Di conditioned greater growth of mesocotyl in length than d\,
Dx conditioned less development of the mesocotyl in length than dx.

The mesocotyl length of six strains consisting of comparisons of DiDi
with Didi on three dififerent backgrounds was taken as 100. The comparisons
are summarized in Table 3.16 and in Figure 3.1.

It is apparent that the superiority of DiDi over Didi in mesocotyl length
becomes less in the highly heterozygous background than in the homozygous
background of A188. This may be more evident from the diagram in Fig-
ure 3.1.

TABLE 3.16

COMPARATIVE LENGTH OF MESOCOT-
\T. FOR SIX STRAINS OF CORN

Background

Percentage
Difference in

Mesocotyl

Length, DiDi

minus Didi

Percentage
Expression of
Background

A188

19

16

4

89

A188XA95Z?xZ?x...
^.\2,%X^9SDxdx....

101

no

It seems of some interest that the differences between DiDi and Didi were
smaller in the highly heterozygous background than in the homozygous
background, and that in the presence of D^x that the differences were
further reduced over those in the presence of DxDx. It may be well to recall
that dx conditioned greater length of mesocotyl than Dx-

Reference may be made to an explanation by Torssell (1948) of tlie decline
in green weight or length of stem in alfalfa in different generations of in-
breeding. It was not greatest in the first inbred generation. He suggests there
was a surplus of vigor genes in a heterozygous condition in the early genera-
tions of selfing, and that great loss of vigor was not observed until about I3

C/)

< -o

(/)

(A

105

(/)

Q

(/)

Ixl

O

tr

1-

Q.

X

o

z

UJ

h-

o

<

<

_J

1-

UJ

z

rr

UJ

o

or

UJ

CL

95

85

75

A D,D,

/

/

/

/

/

/

A'

A

B

i±

C

80 90 100 110

PERCENTAGE EXPRESSION OF
BACKGROUND

Fig. 3.1 — Relative expression of A A vs. A^i regarding final length of mesocotyl on vari-
ous backgrounds: (A, A188; B, A188 X A95-344 carrying AA; C, A188 X A95-344 D,d^).

DEVELOPMENT OF THE HETEROSIS CONCEPT 65

when selfing reduced the necessary genes below a stage needed by the or-
ganism. The following quotation from Thorssell emphasizes the viewpoint
that the relative importance of genes controlling heterosis is greatly in-
fluenced by other factors of the organism:

The cumulative effect of heritable characters, however, brings it about that develop-
ment, that is to say green weight, does not stand in arithmetical proportion to the number
of pairs of the dominant genes in question. From this it follows also that the said number can
be reduced within a certain limit without perceptible or any great influence upon green
weight. If this limit is exceeded, a considerable degeneration sets in.

The speaker has chosen to consider heterosis as the normal expression of
a complex character when the genes concerned are in a highly heterozygous
condition. x'Vs most normal characters are the end result of the action, reac-
tion, and interaction of countless numbers of genes, and as gene mutation
constantly occurs although relatively infrequently, it may be impossible to
obtain all essential genes in the most favorable homozygous state. After
selecting the best homozygous combinations, further vigor will be obtained
due to heterozygous combinations of factors. Dominance or partial domi-
nance seems of great importance as an explanation of hybrid vigor. In some
cases there may be extra vigor correlated with the heterozygous condition of
pairs of alleles. The types of response of inter and intra allelic factor interac-
tions are without doubt dependent upon both external and internal environ-
ment.

M. M. RHOADES

University of Illinois

Chapter 4

Preferential
Segregation in Maize

The outstanding example of the utilization of heterosis in plant improve-
ment is that of hybrid corn. Extensive studies on maize genetics have clearly
demonstrated that chromosome and gene segregation are in accordance with
Mendel's laws of segregation and recombination. It would appear, therefore,
that any unusual mechanism operating in maize to produce deviations from
normal Mendelian behavior should be worthy of our consideration, even
though the principles involved have no bearing on the nature or manifesta-
tion of heterosis. The purpose of this section is to present data on preferential
segregation in maize and to offer a tentative interpretation of this phe-
nomenon.

Two kinds of chromosome 10, the shortest member of the haploid set of
ten, are found in populations of maize. The common or normal type gives
typical Mendelian ratios when the two homologues are heterozygous for
mutant loci. The second kind of chromosome 10, which has been found in a
number of races from Latin America and the southwestern United States,
also gives normal Mendelian ratios for chromosome 10 loci in plants homozy-
gous for this chromosome. This second or abnormal kind of chromosome 10
differs from the normal chromosome 10 by a large, chiefly heterochromatic
segment of chromatin attached to the end of the long arm and also in the
chromomeric structure of the distal one-sixth of the long arm (see Fig. 4.1
and Fig. 1 of Plate I). As is illustrated in Figure 4.1 the chromomeres in this
region are larger and more deeply staining than are the correspondingly
situated chromomeres of the normal homologue.

Although normal Mendelian ratios are obtained for segregating loci in
chromosome 10 in both kinds of homozygotes, we were able to show in an
earlier paper (Rhoades, 1941) that preferential segregation occurs at mega-

66

n

Fig. 4.1 — Camera lucida sketch at pachynema of bivalent consisting of one normal and one
abnormal chromosome 10. Note the dissimilarity in chromomere pattern in the distal one-
sixth of the long arm. The identical chromomere pattern found in the remainder of the

chromosomes is not figured here.

Fig. 4.2 — Anaphase I of cell illustrated in Figure 4 of Plate I. Some of the disjoining dyads
are normal appearing while others have active neo-centric regions.

Fig. 4.3 — Metaphase I with eleven dyads. Five of the dyads have precocious neo-centro-
meres at sub-terminal portions of their long arms.

Fig. 4.4 — Anaphase II of cell illustrated in Figure 7 of Plate II. In some of the inverted
V-shaped monads the true centric regions are attracted toward the opposite pole.

68 M. M. RHOADES

sporogenesis in plants heterozygous for a normal and an abnormal type of
chromosome 10. Approximately 70 per cent of the functioning mega spores
possessed the abnormal 10 instead of the usual 50 per cent. The excess of
female gametes with the abnormal 10 was not due to lethal factors or to
megaspore competition. The disjunction of the two dyads comprising the
heteromorphic bivalent at anaphase I, and of the two monads of each dyad
at anaphase II, was such that an abnormal 10 chromosome tended to pass
with a high frequency to the basal spore of the linear set of four.

The factor or factors responsible for this preferential segregation reside
in the chromatin segments which differentiate the two kinds of chromosome
10. Whether the distal one-sixth of the long arm or the large heterochromatic
piece of extra chromatin carries the causative genes for preferential segrega-
tion has not yet been determined — since these two regions of the abnormal
chromosome 10 have never been separated by crossing over. The locus of
the gene R is in the long arm of chromosome 10. There is approximately 1 per
cent recombination between R and the end of the long arm in plants hetero-
zygous for the two kinds of chromosome 10; but every crossover distal to R
occurred to the left of the dissimilar chromomeres in the distal one-sixth of
the long arm. Apparently little or no crossing over takes place here, although
pairing at pachytene is intimate.

Strictly terminal chiasmata in the long arm have not been observed at
diakinesis in heterozygous plants. The close linkage of the R locus with the
extra segment of abnormal 10 is due to a suppression of crossing over in the
end regions of the long arm. E. G. Anderson (unpublished) has studied a re-
ciprocal translocation involving normal 10 with the break distal to R, and
found 5 per cent recombination between R and the translocation point.
There is an undetermined amount of crossing over between the translocation
point and the end of the chromosome. It should be possible to locate the re-
gion or regions in abnormal 10 responsible for preferential segregation by ob-
taining successively larger terminal deficiencies, but this has not been at-
tempted.

The dissimilarity in chromomere pattern in the distal portion of the long
arms of the abnormal and normal chromosomes 10, together with the lack of
crossing over in this region, suggest the possibility that the gene content may
not be identical in the two kinds of chromosome 10. Inasmuch as plants
homozygous for the abnormal chromosome 10 are not noticeably different in
growth habit and general appearance from sibs carrying only the normal 10,
it would appear that some kind of structural modification was responsible for
the suppression of crossing over. To assume that this distal region consists
of non-homologous loci in the two types of chromosome would mean that
plants with two abnormal 10 chromosomes would be homozygous deficient
for certain loci found in the comparable region of normal 10. This appears
unlikely.

PREFERENTIAL SEGREGATION IN MAIZE 69

That a structural difference, aside from the extra chromatin of abnormal
10, exists between the two kinds of chromosome 10 also is indicated by the
pairing relationships in plants trisomic for chromosome 10. in plants with
two normal and one abnormal chromosome 10, trivalent associations were
observed in 251 (60.2 per cent) among a total of 417 microsporocytes. When
a chain of 3 was found at diakinesis, the abnormal 10 occupied a terminal
position in 90 per cent of the cells. It was united with a normal chromosome 10
by a chiasma in the short arm. A univalent chromosome 10 was found at
diakinesis in 39.8 per cent of the pollen mother cells.

If pairing, as reflected by chiasmata formation, were random among the
three chromosomes, the ratio of normal -.abnormal chromosomes 10 in the
univalent class should be 2 : 1. Actually the unpaired chromosome was a nor-
mal 10 in 28 cells among a total of 166, while in the remaining 138 cells the
univalent was an abnormal 10. In individuals again trisomic for chromo-
some 10, but possessing one normal and two abnormal chromosomes, the
percentage of trivalent associations at diakinesis was 57.9 in a total of 513
cells. In the chains of 3, the two abnormal homologues were adjacent mem-
bers, joined by a chiasma between their long arms, in 70 per cent of the
cases. An unpaired chromosome 10 was found in 42.1 per cent of the micro-
sporocytes.

If pairing were random, two times as many abnormal lO's as normal lO's
should be found as univalents; but in a total of 216 cells an abnormal 10
was the univalent in 69, while a normal chromosome 10 was the univalent
in 147. Chiasma formation among the three chromosomes 10 of trisomic
plants clearly is not at random. There is a marked preference for exchanges
in the long arm between the two structurally identical homologues. If synap-
sis usually begins at the ends and progresses proximally, the non-random as-
sociations found in trisomic plants become understandable. Normal recom-
bination values for the li-gi and gi-R regions which lie proximal to R (see
Table 4.1 for gi-R data) indicate that any suppression of crossing over is
confined to the region beyond the R locus in disomic plants heterozygous for
the two kinds of chromosome 10. It is no doubt significant that differences
in chromomeric structure are not found in regions proximal to the R locus.

Inasmuch as the R locus is closely linked with the extra chromatin of ab-
normal 10, the ratio oi R:r gametes from heterozygous plants gives a good
approximation of the frequency with which the abnormal chromosome passes
to the basal megaspore. The genetic length of the long arm of chromosome 10
is such that at least one chiasma is found in the arm. If one chiasma invari-
ably occurs in the long arm of heteromorphic bivalents, each of the two dis-
joining dyads of anaphase I will possess one normal chromatid and one ab-
normal chromatid. Preferential segregation would be restricted to the sec-
ond meiotic division, and occur only if the orientation of the dyad on the
spindle of metaphase II were such that the abnormal chromatid passed to

70 M. M. RHOADES

the lower pole of the spindle. Normal segregation would occur in those mega-
sporocytes which had homomorphic dyads.

If the terminal segment of abnormal 10 determines preferential segrega-
tion, it follows that loci near the end of the long arm will be preferentially
segregated more frequently than loci further removed from the end of the
chromosome. From the data in Tables 4.1 and 4.2 it is evident that the dis-
tortion from a 1 : 1 ratio is greater for the R locus than for the more proximal-
ly situated gi locus. The li locus which is proximal to gi was less affected
than ^1.

Longley (1945) reported non-random segregation at megasporogenesis for
chromosome pairs other than chromosome 10 when one of the two homologues
had a prominent knob and the other was knobless. Segregation was random
for these heteromorphic bivalents in plants homozygous for the normal chro-
mosome 10, and non-random if abnormal 10 was heterozygous. He studied
preferential segregation of chromosomes 9 and 6. The data for chromosome 9
are the most instructive. Some strains of maize have a chromosome 9 with a
knob at the end of the short arm, others have a knobless chromosome 9. The
C, Sh, and Wx loci lie in the short arm of this chromosome, with Wx nearer
to the centromere. C and Sh are in the distal one-third of the short arm. Ap-
proximately 44 per cent recombination occurs between Wx and the terminal
knob — they approach independence — while C and Sh are 23 and 26 recombi-
nation units distant from the knob. *

When plants of knob-C/knobless-c constitution, which were also heterozy-
gous for abnormal 10, were pollinated by recessive c, 64 per cent of the func-
tioning megaspores possessed the C allele. The Sh locus, close to C, showed a
similar degree of preferential segregation in comparable tests, but the Wx
locus was little affected. Such a progressive decrease in effect is expected if
the terminal knob on the short arm is instrumental in producing preferential
segregation. The part played by the knob of chromosome 9 was wholly un-
expected. Obviously this heterochromatic structure can no longer be con-
sidered as genetically inert. The data on various loci in chromosomes 9 and
10 prove that the degree of preferential segregation of a locus is a function
of its linkage with heterochromatic regions which, in some way, are con-
cerned with non-random segregation.

The data presented above show that alternative alleles are not present in
equal numbers among the female gametes when abnormal 10 is heterozygous.
We have here an exception to Mendel's first law. Are deviations from Men-
del's second law, the independent assortment of factor pairs on non-homolo-
gous chromosomes, also occurring? This question is answered by Longley's
data where the C and R loci are both segregating preferentially. In separate
experiments he found the C locus was included in 64 per cent and the R locus
in 69 per cent of the functioning megaspores. Assuming that these percent-
ages hold in plants where both are simultaneously segregating, the observed

•fppl

2

* ^<

^ i'

•» ' .'

4

Plate I: Fig. 1 — Pachytene showing homozygous abnormal 10. Carmine smear. The proximal jjortion of
the extra chromatin is euchromatic as is a smaller distal piece. A large and conspicuous knob lies between
the two euchromatic portions. Fig. 2 — Metaphase I in microsporocyte homozygous for abnormal 10.
Carmine smear. The ten bivalents each have their true centric regions co-oriented on the spindle. The
onset of neo-centric activity is manifest in the second, sixth, and seventh bivalents from the right. The
third and fourth bivalents from the right are somewhat superimposed. Figs. 3 and 4 — .\naphase I in mi-
crosporocyte homozygous for abnormal 10. Carmine smear. Some of the dyads are undergoing a normal
anaphase separation while in others the neo-centric regions are pulling the ends poleward. Note that
the normal ap[)earing dyads are slower in their poleward migration. F'^ig. 4.2 is a drawing of Fig. 4 above.

Plate II: Figs. 1 and 2 — Metaphase II in plant homozygous for abnormal 10. Carmine
smear. Precocious poleward movement of neo-centric regions is clearly evident. One dyad
has a single neo-centric region (Fig. 4.5, dyad No. 8) while the left-most dyad has a neo-
centric region in both long arms (Fig. 4.5, dyad No. 7). This cell was figured in Rhoades
and Vilkomerson 1942. Figs. 3 and 4 — .Anaphase II in jilant homoz\gous for abnormal 10.
Carmine smear. Note that the rod-shaped monads with precocious neo-centromeres are
the first to reach the poles. Fig. 5 — Metaphase II in plant homozygous for abnormal 10.
Carmine smear. The only chromosome of the haploid complement which can be recognized
at metaphase II is chromosome 6 which has a satellite at the end of the short arm. In this
cell the chromosome 6 d\'ad is the second from the left. That the terminal chromosome of
the satellite is actuall\- a small knob is indicated by the formation of neo-centric regions al
the end of the short arm. Fig. 6 — Early anaphase II in plant heterozygous for abnormal 10.
Carmine smear. That the poleward movement of neo-centric regions is less rapid in hetero-
zygous than in homozygous abnormal 10 plants is indicated here by the relatively slight
attenuation of the rod-shaped monads. Fig. — 7 Late anaphase II in plant homozygous for
abnormal 10. Carmine smear. The previously greatlx- stretched rod monads with precocious
neo-centromeres have contracted. Note the inverted V-shaped chromatids. This is the same
cell shown in Figure 4.4. Fig. 8 — Side view of metaphase I in a normal plant showing the
fibrillar nature of the chromosomal fibers. Fi.xed in Benda, stained with haemotoxylin.
Paratiine section. The only chromosomal fibers j^resent are those formed by the true cen-
tromeres. Ordinarily chromosomal fibers are not evident in carmine smears since they are
destroyed by acetic-alcohol fi.xation and it is nccessar\' to use special techniques to demon-
strate them. Similar fibrillar chromosomal fibers are found at neo-centric regions when
proper fixation and staining methods are employed. Fig. 9 (top) — Polar view of meta-
phase I in normal plant. Fixed in Benda, stained with haemotoxylin. Parafhne section.
Note the arrangement of the ten bivalents on the ecjuatorial plate. This microsporocyte
was cut slightl}' above the metaphase plate. The next section, which includes the remaining
portion of this cell, is a cross section through the ten sets of chromosomal fibers.

.//

J

>

/

j>

1

/

' 2

'\ h I-
't J.^f

i . S

■*l!7

^.-■IH.- ;^^'4«H*'

#

8

//

T^ • > .'

A V V,

^ A.

' M» /•

••»

1

^lMjJ

TABLE 4.1

LINKAGE DATA FROM THE CROSS OF G r ABNORMMVg R
NORM.\L X gr d'd'

Linkage
Phase

Constitution of Chromosomes

Repulsion

(0)

G
r

(o)

g
R

(X)

G
R

(x)
S

r

Total

Ratio of
i?:r ON Ear

243
102
150
396
154
169
215
231

138
86

114
50
81
90
61
79

29
9

18
7

11
21
24
35

49
13
20
59
29
30
77
81

459
210
302
512

275
310
377
426

186/?:326r
136/?:319r
145/?:288r
169/?:588r
120/?:277r
127/?:223r
102/?:338r
133/?:358r

1660

699

154

358

2871

1118/?:2717r

%Rm total = 29.7 % g in total = 36.8

% i? in non-crossover classes = 29.6
% /J in crossover classes = 30.1

G — R recombination = 17.8%

29.2% R

TABLE 4.2

LINKAGE DATA FROM THE CROSS OF G r NORMAL/g R

ABNORMAL X gr

Linkage
Phase

Constitution of Chromosomes

Repulsion

(o)
G
r

(o)

g
R

(x)
G
R

(x)

g

r

Total

Ratio of
R\r (m Ear

12
38
35
39

87

96

86

107

13
29

21

1

6

7
9

113
169
161
176

182/?: 42r
188/?: 59r
230/?: 74r
241/?: 77r

124

376

96

23

619

841/?:252r

% r seeds in total = 23.8

% r seeds in non-crossover classes = 24.8

% T seeds in crossover classes = 19.3

G — R recombination = 19.2%

72 M. M. RHOADES

frequencies of F2 phenotypes can be compared with those calculated on the
assumption of independent assortment. The two values agreed very closely,
indicating little or no deviation from the law of independent assortment.
His data, from plants where loci in chromosomes 9 and 6 are both segregat-
ing preferentially, likewise permit such a conclusion to be drawn.

In my 1942 paper on preferential segregation the statement was made
that the chromosomes in plants with the abnormal chromosome 10 formed
extra chromosomal (half spindle) iibers at regions other than the true centro-
mere region. Rhoades and Vilkomerson (1942) found these supernumerary
chromosomal fibers were produced only in plants homozygous or heterozy-
gous for the abnormal 10, and that sister plants homozygous for the normal
10 had chromosomal fibers originating solely from the localized centric re-
gion in an orthodox manner (see Fig. 8 of Plate II). Although the abnormal
chromosome 10 was clearly responsible for the formation of these neo-centric
regions, they were not restricted to this chromosome since many of the non-
homologous chromosomes had supernumerary chromosomal fibers. The ab-
normal chromosome 10 is thus responsible for the formation of neo-centric
regions, as well as for preferential segregation. Since 1942, a considerable
body of data has been obtained bearing on the behavior of abnormal 10.
Some of the more pertinent observations have suggested a cytological mecha-
nism for the phenomenon of preferential segregation.

The unorthodox formation of supernumerary chromosomal fibers from neo-
centric regions is limited to the two meiotic divisions. (For a description of
normal meiosis in maize see Rhoades, 1950.) The first meiotic division is in
no way exceptional until metaphase I is reached. Normal appearing bivalents
are co-oriented on the spindle figure in a regular manner with the half spindle
fibers, arising from the true centric regions, extending poleward. Normally
these fibers effect the anaphase movement of the disjoining dyads with the
localized centromere region leading the journey to the spindle pole. How-
ever, in plants with the abnormal 10, chromosomal fibers arise from distal
regions of the chromosome while the bivalents are still co-oriented on the
spindle at metaphase I. The neo-centric regions are drawn poleward more
rapidly than the true centric regions. Consequently the distal ends, instead
of being directed toward the spindle plate during anaphase I, lead the way
to the pole.

The appearance of many disjoining dyads at anaphase I suggests that
their poleward migration is due largely, even exclusively, to the fibers origi-
nating from the neo-centric regions. The primary centric region appears to
play no active role even though it possessed chromosomal fibers at meta-
phase I when the tetrad (bivalent) was co-oriented. At mid-anaphase there is
no indication of the presence of these fibers in many of the dyads with the
precocious neo-centric regions.

Figure 4.5 and Figures 3 and 4 of Plate I illustrate some of the observed

4

ANAPHASE I DYADS

A

11

7 8

METAPHASE H DYADS

J ii I ^

12 13 »U ^15

10

i

12

ANAPHASE IE MONADS

t

16

Fig. 4.5 — All figures are from carmine smears of homozygous abnormal 10 plants. Figures
1-5 represent various configurations found at anaphase I. Figure 1 is a normal dyad with
chromosomal fibers formed only at the true centric region. In Figure 2, two arms have
formed neo-centric regions. The true centric regions appear to be inactive. Figure 3 shows a
dyad with two neo-centric regions and an active true centric region whose chromosomal
fibers are directed away from the nearest pole. Figure 4 is a dyad with a single neo-centric
region. In Figure 5 the two neo-centric regions are directed to opposite poles. Figures 6-7
illustrate various metaphase II dyads. The location of the equatorial plate is represented
by horizontal lines. Figure 6 is essentially normal with no formation of neo-centromeres.
Figure 7 is a dyad with two neo-centric regions directed toward opposite poles. There is a
single neo-centric region in Figure 8. Figure 9 is a dyad which is disjjlaced from the equa-
torial plate. The true centric region has divided to form two independent monads. Each
monad has formed two neo-centric regions which are oriented toward opposite poles. In
Figure 10 one of the monads has its two neo-centromeres directed to opposite poles. Fig-
ures 11-16 are illustrations of monads found at anaphase II. Figure 11 is a normally dis-
joining monad. In Figure 12 a single neo-centromere is evident. Figure 13 shows two neo-
centric regions. Figure 14 has a single neo-centromere which was active at metaphase II.
In Figure 15, chromosomal fibers have arisen from two neo-centric regions and also from
the true centric region. The true centric region and the neo-centromeres are acting in op-
posite directions. Figure 16 shows a monad with two neo-centric regions which are directed
toward opposite poles. This type of monad is derived from those shown in Figure 9.

74 M. M. RHOADES

anaphase I configurations. Chromosomal fibers may arise from one or both
of the long arms of each dyad at late metaphase or early anaphase I. Al-
though it was not always possible to differentiate between long and short
arms, the neo-centric regions in general appear to be confined to the long
arm. When both long arms of the two chromatids of a dyad possessed a neo-
centric region, the chromosomal fibers arising from these centric regions were
usually directed toward the same pole. Occasionally they were oriented to
opposite poles thus causing a great attenuation. In such cases, however,
those chromosomal fibers nearest to one pole were powerful enough to over-
come the oppositely directed force of the second neo-centromere. Despite the
great complexity of configurations at anaphase I resulting from interacting
and conflicting half-spindle fibers arising from both the true and neo-centric
regions, the end of anaphase I usually finds ten dyads at each pole. Some-
times, however, greatly stretched chromosomes undergo breakage. This
breakage doubtless accounts for the higher pollen abortion (about 10 per
cent) found in homozygous abnormal 10 plants as contrasted to the lower
(0-5 per cent) pollen abortion of normal sibs.

Even though one or two arms of some dyads are markedly stretched at
anaphase I, the ensuing telophase is normal. All four arms of each dyad con-
tract to form a spherical mass of chromatin which is loosely enveloped by
the lightly-staining matrical substance. The chromonemata uncoil during
interphase and early prophase II finds each daughter cell with ten, long X-
shaped dyads of typical appearance. The two chromatids comprising each
dyad are conjoined by the undivided primary centric region. There is no indi-
cation of neo-centric regions, although some of the long arms possessed chro-
mosomal fibers at the preceding anaphase.

The onset of metaphase II sometimes occurs before the dyads have under-
gone their usual contraction. Occasionally chromosomal fibers arising from
neo-centric regions in the long arms are found at late prophase II. These
precociously acting fibers produce an extension of the long arms before any
spindle is visible. This observation is of singular importance. Some authori-
ties believe that the centromere region is attracted (whatever this term may
signify) to the spindle pole. Here we have a movement produced by the
chromosomal fibers of neo-centric regions in the absence of an organized
spindle. The way in which these neo-centric fibers act can only be conjec-
tured, but no interaction between centric regions and spindle pole is essential.
It is, indeed, probable that the only role of a bipolar spindle is to provide a
structural frame which channels the chromosomes to the spindle poles.
Clark's (1940) studies on divergent spindles are pertinent in this respect.

The objection may be raised that the chromosomal fibers of neo-centric
regions are not comparable to those arising from the true centric region. I
do not believe this is a valid criticism. Not only are both kinds of fibers con-
cerned with chromosome movement, but, as will be shown in a later section,

PREFERENTIAL SEGREGATION IN MAIZE 75

the fiber-producing activity of the neo-centric regions is a ])r()<luct of the true
centric region.

The appearance of neo-centric fibers in prophase II is not the rule. Usually
the dyads come to lie with the true centric region on the spindle plate at
metaphase II before any pronounced activity of neo-centric regions is ap-
parent. Before the primary centric region divides, thus permitting a normal
anaphase, chromosomal fibers again arise near the distal ends of the long
arms of some dyads. These newly formed fibers move the long arms poleward
while the dyad is still held on the metaphase plate by the undivided true
centric region. This poleward movement is so rapid that the ends of the
chromosomes may reach the spindle poles before the true anaphase occurs.
Eventually the true centric region becomes functionally split, and the two
monads fall apart and pass poleward. It is evident from Figures 4.4 and 4.5
and Figure 7 of Plate II that the configurations of the disjoining monads
(chromatids) at anaphase II are greatly different from normal.

Neo-centric activity, as shown by formation of additional chromosomal
fibers, occurs in plants both homozygous and heterozygous for the abnormal
10, but it is much more striking in homozygous plants. Plants trisomic for
abnormal 10 were not greatly different from homozygous disomic sibs.

Precocious chromosomal fiber formation by neo-centromeres at metaphase
II appears in general to be confined to the long arms of the dyads, although
it is often difficult to differentiate between two unequal arms when one is
stretched poleward. Some chromosomes have arm ratios so extreme that
the distinction between long and short arms is clear, and in these chromo-
somes the precocious fibers at metaphase II arise from the long arms. It is
perhaps significant that, with the exception of the terminal knob on the short
arm of chromosome 9, all remaining knobs in our material were situated in
the long arms. (Chromosome 6 had two small knobs in its long arm but a
maximum of one knob was present in the other chromosomes.) Corre-
spondingly, only one of the two arms of any chromatid had neo-centric
activity at metaphase 11.^ The number of dyads with precocious spindle
fibers, as judged by the number of arms pulled poleward at metaphase II,
varied in different strains. The maximum number in some plants was seven,
in others five, etc. Plants with seven knobbed chromosomes had a maximum
of seven dyads with arms stretched poleward at metaphase II. Those with
four knobs had four such dyads. That is, a strong correlation exists between
knob number and the number of dyads with neo-centric activity at meta-
phase II.

A further observation of some interest was that in plants homozygous for
all knobs both homologous arms of a dyad usually were pulled poleward at
metaphase II; while in plants heterozygous for some knobs many of the
dyads had only one arm with neo-centric activity (see Figure 4.5 and Figures

1. With the possible exception of chromosome 6. See Figure 5 of Plate II.

76 M. M. RHOADES

1 and 2 of Plate II). It is not unreasonable to assume that dyads with both
homologous arms exhibiting neo-centromeres at metaphase II carried a knob
in each chromatid, while dyads with one neo-centromere consisted of one
knobbed and one knobless chromatid. Such heteromorphic dyads would arise
from heteromorphic bivalents by a crossover between the true centromere
and the knob. We believe that only knobbed chromatids have active neo-
centromeres at metaphase II, and that knobless ones are normal at this stage.
Unfortunately, knobs cannot be recognized at metaphase II, and the validity
of the above assumptions rests upon indirect but convincing evidence.

Two types of disjoining monads are found at anaphase II, those which
are rod-shaped and those which are V-shaped. Monads which had one arm
extending poleward at metaphase II are rod-shaped. They are the first to
reach the pole. Indeed distal portions of such chromatids already had arrived
there during metaphase II owing to the early action of their neo-centromeres.
The V-shaped monads of anaphase II are derived from those chromatids
devoid of neo-centromeres at metaphase II. The poleward migration of some
monads is first begun by the chromosomal fibers emanating from the true
centric region, but shortly after anaphase is initiated chromosomal fibers
may arise from the ends of both arms. These terminally placed fibers, which
are directed to the same pole, propel their ends poleward with such rapidity
that the ends first overtake and then pass the centric region in the course of
anaphase migration. Consequently these monads reach the poles as inverted
V-shaped chromosomes (see Fig. 4.4). The spindle fibers from the true centric
region now are directed toward the spindle plate rather than to the pole — they
have reversed their orientation. This would be impossible if chromosomal
fibers were of a thread-like structure. It is more likely that these fibers repre-
sent nothing more than lines of force emanating from the centromere. In-
verted V-shaped chromatids are not invariably found at anaphase II.

Some monads have chromosomal fibers only at the true centric region and
move poleward in a normal fashion. Either neo-centric regions are not pres-
ent, or else arise too late to be effective. It should be emphasized that a funda-
mental distinction exists between the rod and inverted V chromatids found
at anaphase II. The rod-shaped monads come from dyads with neo-centric
activity at metaphase II. Their supernumerary chromosomal fibers arise
from one arm. Their sub-terminal location suggests they may arise adjacent
to the knob, but this is merely a conjecture. The later-formed extra chromo-
somal fibers of the inverted V chromatids, which are knobless, are terminal
and arise from both arms.

If a dyad is oriented on the spindle plate at metaphase II before the onset
of precocious neo-centromere activity, the supernumerary chromosomal
fibers arising from the knobbed arm of the chromatid situated slightly above
the spindle plate are directed toward the upper (nearest) pole, and those
from the bottom chromatid go to the lower pole — they are co-oriented (see

PREFERENTIAL SEGREGATION IN MAIZE 11

Fig. 4.3). No such regularity is found in those infrequently occurring dyads
which are longitudinally displaced from the spindle {)late at metaphase II.
Their true centric regions divide prematurely. Consequently, the two
chromatids of these displaced dyads no longer remain conjoined, but fall
apart to become independent monads which lie side-by-side, parallel with
the longitudinal axis of the spindle.

The neo-centric activity which these monads now manifest is similar to
that found at anaphase II for those monads derived from normally oriented
dyads lacking precocious neo-centromeres at metaphase II, in that neo-
centromeres may arise from the ends of both arms. When this occurs, the
orientation of the two neo-centromeres of each monad is usually to opposite
poles, but sometimes both ends of a monad are directed toward the same
pole. Although the monads from displaced dyads have neo-centromeres at
the end of each arm, one end being attracted to the nearest pole and the other
to the more distant pole, normal disjunction usually occurs. This requires
one monad to move away from the nearest pole toward which one of its ends
is attracted, and to pass to the more distant pole, while the other monad goes
to the nearest pole. It is difficult to interpret this phenomenon in terms of
strength of attraction as a function of distance from centromere to pole.

The formation of neo-centric regions requires the presence of the abnormal
chromosome 10. In its absence, no such regions are found. It appears highly
probable that heterochromatic knobs located on other chromosomes also are
concerned in the formation of precocious centric regions at both meiotic
metaphases, since the cytological observations show a correlation between
number of knobs and number of precocious centric regions. Knobless arms
later form neo-centric regions, but not until anaphase movement has already
been initiated by the true centric region.

It is possible that maize chromosomes possess latent centric regions which
are activated by the abnormal 10. It has been demonstrated, however, that
the true centric region is involved in the formation of neo-centromeres.
Plants homozygous for abnormal 10 and heterozygous for the long para-
centric inversion in chromosome 4, studied by McClintock (1938) and Mor-
gan (1950), were obtained. Both the normal and inverted chromosome 4
carried a large knob in the long arm which is included in the inverted seg-
ment. Single crossovers within the inversion give rise to two non-crossover
monocentric chromatids, one dicentric chromatid which forms a bridge at
anaphase I, and an acentric fragment. The knobbed acentric fragment lies
passively on the spindle with no indication of spindle fiber activity. Neo-cen-
tromeres arise from the same chromatin segments comprising the acentric
fragment when they constitute a portion of a whole chromosome 4. It fol-
lows that the true or primary centromere plays an essential role in the pro-
duction of neo-centromeres.

The localized centromeres of maize chromosomes are concerned with the

78 M. M. RHOADES

elaboration of fiber-producing material. Normally this unique substance is
confined to the true centric region, hence chromosomal fibers arise solely
from this part of the chromosome.

It is our belief: (1) that these centric regions produce an over-abundance
of fiber-forming material if abnormal 10 is present in the nucleus; (2) that a
portion of this substance escapes from the confines of the centric regions and
moves distally along the chromosome to produce supernumerary chromo-
somal fibers; and (3) that the knobs either stimulate centric activity or else
cause the excess fiber-forming substance to move preferentially along knob-
bearing arms so that neo-centric activity is first manifested by these arms.

The failure of the acentric fragment to form chromosomal fibers suggests
that the postulated movement of the material from the true centric region
occurs after crossing over has taken place. If it happened prior to pachytene,
the regions which later constitute the acentric fragments would receive some
of this fiber-producing substance which subsequently could form spindle
fibers. In support of the above interpretation is the observation that small
aggregations of a substance similar in appearance to that located in the true
centric region are sometimes found near the distal regions of some chromo-
somes at metaphase I and metaphase II. This observation is subject to vari-
ous interpretations. But in conjunction with the behavior of acentric frag-
ments, it strengthens the hypothesis that the production of neo-centromeres
is intimately related to the presence or activity of the primary centric region.
It is obvious that the presumed movement of the products of the centromere
along the arms of the chromosome has a bearing on the kinetic theory of Posi-
tion Effect.

Evidence has been presented that the abnormal chromosome 10 produces
the phenomenon of preferential segregation, and that it also causes the for-
mation of neo-centromeres. Are these two phenomena related — does prefer-
ential segregation occur as a consequence of neo-centric activity? While no
definite answer can be given at this time a tentative hypothesis has been de-
veloped. Sturtevant and Beadle (1936), seeking to account for the absence of
egg and larvae mortality following single crossovers in paracentric inversions
in Drosophila, postulated that the crossover chromatids were selectively
eliminated from the egg nucleus. The two spindles of the second meiotic divi-
sion in Drosophila eggs are arranged in tandem. Following a crossover within
the inverted segment, the tetrad at metaphase I consists of two non-crossover
chromatids, a dicentric and an acentric chromatid.

They assumed that the chromatin bridge arising from the dicentric chro-
matid, when the homologous centromeres pass to opposite poles at anaphase
I, ties its two centromeres together. The spatial arrangement thus produced
is such that the two monocentric chromatids lie nearer the two poles than
does the dicentric chromatid.

The persistence of this relationship into the second division results in a

PREFERENTIAL SEGREGATION IN MAIZE 79

non-random orientation on the metaphase II spindles. The monocentric, non-
crossover chromatids are free to pass to the two terminal poles, while the two
centromeres from the dicentric chromatid are directed to the two inner poles.
Consequently, at anaphase II the terminal poles each receive a non-crossover
chromatid. Since the egg nucleus arises from the innermost terminal pole it
would contain a non-crossover chromatid with a full set of genes. The cor-
rectness of this ingenious hypothesis was established by Darlington and La
Cour (1941) in Lilium and Tulipa and by Carson (1946) in Sciara.

It is possible that a somewhat similar mechanism is operating in Zea to
produce preferential segregation. In maize, as in Drosophila, the two
spindles of the second meiotic division of megasporogenesis are arranged in
a linear order. The basal megaspore of the linear set of four develops into
the female gametophyte, the remaining three aborting. We know that in
plants heterozygous for knobbed and knobless chromosomes, one arm of
some of the disjoining dyads at anaphase I possess precociously-acting
chromosomal fibers not present in the homologous arm. There is reason to
believe that the knobbed arms form precocious neo-centromeres while knob-
less arms do not. Owing to the rapidity with which neo-centric regions pass
poleward at anaphase I, those chromatids with neo-centromeres reach the
pole in advance of knobless arms lacking neo-centromeres. In a dyad con-
sisting of one knobbed and one knobless chromatid, the knobbed chromatid
would come to lie closer to the pole, while the knobless one would face the
spindle plate.

In order to account for preferential segregation, it is necessary to assume
that this orientation persists until the second metaphase, and that it results
in the knobbed chromatids facing the two terminal poles while the two knob-
less ones would be oriented toward the two inner poles. On such a mechanism,
preferential segregation would occur only when a crossover takes place be-
tween the knob and the true centromere in a heterozygous bivalent. The
extent of preferential segregation would be a direct function of the amount of
crossing over in the knob-centromere region.

Such an explanation can only be considered as a working hypothesis. It
can be critically tested, however, and such experiments are being conducted
by Jean Werner Morgan, who also participated in the studies reported here.
They include varying the crossover distance between knob and centromere
by translocation and inversion, testing for preferential segregation of hetero-
morphic chromosomes other than chromosome 10 in plants homozygous for
abnormal 10, determining neo-centric activity in chromatids with knobs in
both the long and short arm, etc. I prefer not to mention her incomplete
findings at this time, since to do so would detract from continued interest in
her work.

The phenomenon of preferential segregation is by no means confined to
maize. Sturtevant (1936) found a non-random segregation of three chromo-

80 M. M. RHOADES

somes IV in Drosophila. Bridges, in Morgan, Bridges, and Sturtevant (1925),
established that the distribution of the chromosomes in triploid Drosophila
was not according to chance. Beadle (1935) reported that crossing over in
triploid Drosophila near the centromere region between one member of at-
tached -X's and a free X chromosome was correlated with autosomal dis-
junction. Lower crossover values were found in 1X2,4 and XX L4 combina-
tions than in IX 1.1 and XX 2.4 gametes. This non-random distribution
indicates a correlated orientation of non-homologous chromosomes on the
equatorial plate.

In Sciara the paternal set of chromosomes moves away from the pole of
the monocentric spindle of the primary spermatocyte. The two sister .Y
chromosomes pass to the same pole at the second spermatocyte division
(Metz, 1938). Schrader (1931) observed a non-random orientation in Pro-
tortonia which led to selective distribution in secondary spermatocytes.
Catcheside (1944), in an analysis of Zickler's data on spore arrangement in
the Ascomycete Bombardia lunata, found that certain genes were prefer-
entially segregated. Not all of the above examples are strictly comparable to
the situations found in maize, Sciara, and Bombardia. In the latter cases a
specific spindle pole receives a certain chromosome or set of chromosomes,
while in the Drosophila cases particular chromosomes pass preferentially to-
gether, but presumably at random, to either pole.

The neo-centromeres arising from chromosome ends, reported in rye by
Prakken and Muntzing (1942) and Ostergren and Prakken (1946), closely
resemble those found in maize. In both maize and rye the neo-centric
regions are found on arms possessing knobs (heterochromatin), and the pole-
ward movement of neo-centromeres is precocious in both plants. Unfortu-
nately, nothing is known about preferential segregation in rye, but it should
occur if our hypothesis is correct.

R. A. BRINK

University of Wisconsin

Chapter 5

Inbreeding and Crossbreeding
in Seed Development

It is now generally recognized that the effects on growth of inbreeding and
crossbreeding are intimately interwoven in the whole complex fabric of
development and reproduction. Not only are the effects widespread and
often of major consequence in the economy of the organism, but sometimes
they are manifested in devious ways. Such is the case in the seed of flowering
plants.

The success or failure of seed development turns primarily, not on the
embryo which embodies the line of descent, but upon an accessory organ of
reproduction, the endosperm. The novel origin and sensitivity of this latter
tissue to changes in genetic composition render early seed development one
of the critical stages in the life cycle of flowering plants. My colleague, D. C.
Cooper, and I have been exploring these relations during the past decade. An
attempt will be made here to review some of the evidence upon which our
point of view rests, and to call attention to some of the broader implications
of the main facts.

As a means of bringing the important aspects of the problem in flowering
plants into focus, seed development in the angiosperms and gymnosperms
will be compared. Essential features of the general hypothesis by which we
have been guided will then be set fo"rth. The central role of the endosperm in
formation of the angiosperm seed and the responsiveness of this tissue to
variations in genetic composition will be illustrated by a consideration of the
immediate effects of self- and cross-fertilization in alfalfa. It will then be
shown that the means by which the embryo in the common dandelion, an
autonomous apomict, is nourished is of a type which would be expected
according to the hypothesis proposed.

* Paper from the Deparlment of Genetics, College of Agriculture, University of Wiscon-
sin, No. 432.

81

82 R. A. BRINK

An illustration will next be given of endosperm failure as an isolating
mechanism. Finally, the significance of the present results for the problem
of artificially rearing embryos whose development in the seed is blocked by
endosperm disfunction will be pointed out.

Complete literature citations are not given. These may be found in the
summary paper (Brink and Cooper, 1947) in which much additional evidence
bearing on the present thesis also is presented.

The endosperm is a special structure intercalated between the female
parent and the embryo, serving to mediate the relations between the two.
The tissue originates from the central cell of the female gametophyte, follow-
ing a fertilization distinct from that giving rise to the embryo. The secondary
fertilization is unusual in that two identical haploid nuclei of maternal origin
are united with one contributed by the pollen. The endosperm thus becomes
3x in chromosome number in contrast with the 2x condition of the embryo
and the mother plant, respectively. Endosperm and embryo carry the same
kinds of genes, but the genie balance may be unlike in the two tissues by
virtue of the double contribution to the endosperm from the maternal
parent. A further element of genetic heterogeneity in the seed arises from the
fact that nucellus and integuments, which are maternal structures, may
differ in genotype from the endosperm and embryo which they enclose,
since they belong to the previous generation.

These facts, of course, have long been known. Certain of their implica-
tions, however, are only now becoming apparent. Particularly is this true of
the secondary fertilization on which our attention will be focussed.

A word should be said at this point concerning the manner in which the
endosperm should be visualized. Many are familiar with the tissue only in
the mature seeds of species in which the endosperm persists as a storage
organ. This condition, well known in the cereals, for example, is exceptional
among flowering plants, and represents a secondary adaptation of signifi-
cance mainly for the future seedling. In most species the endosperm either
does not persist in the fully developed seed or occurs therein as a residue
only. On the other hand, the endosperm is regularly a prominent organ in
the juvenile seed. It is especially active directly following fertilization, during
what may be termed the lag phase of embryo growth. This period is seldom
longer than a few days, and varies according to the species. In spite of its
typically ephemeral character, the endosperm plays a critical role in (1)
transforming the mature ovule into a young seed and (2) nourishing the
embryo during its initial period of growth. We are here concerned with the
endosperm in these two relationships only.

THE SEED IN GYMNOSPERMS AND ANGIOSPERMS

It is helpful in understanding the significance of the secondary fertilization
to compare the circumstances of seed development in the angiosperms with

INBREEDING AND CROSSBREEDING IN SEED DEVELOPMENT 83

those in the other great class of seed forming plants, the gymnosi)erms. A
secondary fertilization does not occur in the gymnosperms. The endosi)crm
is a haploid tissue derived from the megas])ore by continuous cell division.
The tissue is a part of the gametophyte rather than an integral structure
distinct from both gametophyte and sporophyte, as in the angiosperms.

On the other hand, the endosperms in the two classes of seed plants have
an important common function, namely, nourishment of their respective
associated embryos. The genetic equipment with which the two kinds of
endosperms are furnished differs in a fundamental respect. That of the
gymnosperm is a sample half of the mother plant's inheritance, whereas the
angiosperm endosperm, being of biparental derivation, has two chances in-
stead of only one of receiving a physiologically effective genie complement.
Insofar as the two tissues are autonomous in their functional properties, the
angiosperm endosperm, therefore, is equipped to meet much more exacting
requirements than its counterpart in the gymnosperms. A summary review
of the differences in the gymnosperm and angiosperm ovules and seeds at
fertilization, and during the immediately subsequent period, shows the im-
portance of (or necessity for) a secondary fertilization in the flowering
plants in order to maintain continuity of the life cycle at this stage.

The differences between the mature ovules of gymnosperms and angio-
sperms which appear to have a direct bearing on the present problem may be
summarized as follows:

1. The seed coat in the gymnosperms approaches its mature size at the
fertilization stage. The angiosperm seed coat undergoes extensive growth sub-
sequent to fertilization. These facts are of interest in relation to the total
food requirements of the two respective classes of growing seeds and the
post-fertilization distribution of nutrients between the seed coat and the
enclosed tissues.

2. The female gametophyte in the gymnosperms is an extensively de-
veloped multicellular (multinucleate, in some higher forms) structure. Its
counterpart in the angiosperms typically consists of only seven cells. The
potential disadvantage of the extreme reduction of the female gametophyte
in the flowering plants will be considered below.

3. Generally speaking, the gymnosperm ovule is rich in food reserves,
whereas the angiosperm ovule is sparsely supplied. This means that in the
latter, the large volume of nutrients required for growth of the endosperm,
embryo, and seed coat must be moved in from other parts of the plant. In
the gymnosperms an extensive supply is directly at hand.

4. So far as may be inferred from the published accounts, fertilization in
the gymnosperms initiates a new cycle of growth in the embryo only. Other
parts of the ovule do not appear to be stimulated. Double fertilization in the
angiosperms, in contrast, not only marks the inception of endosperm and
embryo formation, but also incites pronounced mitotic activity and en-

84 R. A. BRINK

largement of the cells in the integuments. Thus, with the exception of the
nucellus which is broken down and absorbed by the rapidly expanding endo-
sperm, all the elements of the young seed which were previously quiescent,
suddenly spring into active growth following syngamy.

Consideration of these differences between the seeds of gymnosperms and
angiosperms led us some ten years ago to explore the hypothesis that the
secondary fertilization in angiosperms is essentially a means of enhancing
the competitive power of the endosperm relative to the maternal portions of
the seed — by conferring upon the endosperm the advantages of hybridity.
The nutritive requirements of the young seed suddenly are raised from a low
to a high level since fertilization starts a new cycle of growth in the massive
integuments. The nutrient supply, on the other hand, quickly falls to the
plane which can be maintained by movement of foods into the seed from
other parts of the plant as a result of exhaustion of the limited ovule reserves.

It seemed reasonable to assume that, within the seed, the incoming nutri-
ents would tend to be partitioned between the different tissues according to
the respective amounts of growth occurring in them. On this basis, the ex-
tensively developed integuments would consume the major portion. The
diminutive endosperm and embryo would receive but a small fraction of the
total. Under these conditions, failure of the young seed through starvation of
the embryo could arise, unless the endosperm — as the nutritive agent of the
embryo — were endowed with special properties which offset its initially small
size. It seemed essential that the endosperm, by one means or another, be
enabled to quickly acquire a position of physiological dominance in the
juvenile seed in order to insure continued development.

Two genetic characteristics of the endosperm suggest themselves as being
important in this connection. The first is the triploid condition of the nuclei.
Little is known of the physiological effects of ploidy in general, and virtually
nothing of its meaning in special situations of this kind. One suspects, how-
ever, that the endosperm gains some advantage from its extra chromosome
garniture, as such, in mediating the relations between the diploid maternal
parent and the young diploid embryo. It is also probably significant that,
whereas the embryo inherits equally from the two parents, two-thirds of the
endosperm's genie complement is derived from the plant upon which it is
nutritionally dependent and one-third of the complement from the male
parent.

Heterozygosis is the second characteristic of the endosperm which might
enhance the inherent physiological efficiency of this tissue. The possibility of
heterozygosity arises, of course, from the biparental origin of the endosperm
mother nucleus. The condition is realized in matings between genetically
different plants. Haploidy of the endosperm, as occurs in the gymnosperms,
appears to be genetically insufficient for seed development in the flowering
plants. Early post-fertilization circumstances, particularly the dependence

INBREEDING AND CROSSBREEDING IN SEED DEVELOPMENT 85

upon and competition for an outside nutrient supply in the latter, require
that the tissue shall share in the advantages of sexuality. The advantage
gained is not that of amphimixis in general, as in the embryo, but solely the
extra vigor of growth associated with the union of unlike nuclei in the mother
cell. Thus hybrid vigor in the endosperm has some claim to uniqueness. The
sole object gained by entry of a sperm into the nuclear makeup of this sterile
tissue is the added vigor of growth thus acquired. Some of the evidence by
which the validity of this point of view may be tested will now be considered.

INBREEDING AND CROSSBREEDING EFFECT ON
SEED COLLAPSE IN MEDICAGO SATIVA

Two classes of matings on seven alfalfa plants were carried out under
favorable growtli conditions in a greenhouse. After removal of the anthers
from the flowers used, a part of the flowers were pollinated with pollen from
the same respective plants. This constitutes the self-fertilized series. Other
flowers on the same plants were cross-pollinated, the pollen being derived in
each case from an unrelated plant within the group. These matings comprise
the cross-fertilized series.

Since alfalfa is regularly cross-fertilized, the second series of matings is
designed to maintain the level of heterozygosity normal to the endosperm
and embryo in this species. The enforced self-fertilization, on the other
hand, would be expected to reduce heterozygosity in the endosperm mother
nucleus and the zygote by 50 per cent. It is proposed to review the conse-
quences for seed development of this sharp reduction in heterozygosis.

Following the above two series of matings, the pistils were collected at 30,
48, 72, 96, 120, and 144 hours and imbedded in paraffin. After sectioning and
staining, data were taken on fertility of the ovules, frequency of fertile ovules
collapsing, number of cells in the embryo, and number of nuclei in the endo-
sperm. Detailed observations were made subsequently on growth of the
integuments.

Alfalfa was known previously to be partially self-incompatible. It was
not unexpected, therefore, to find that only 15 per cent of the ovules became
fertile after selfing in contrast to 66 per cent after cross-pollination. The new
fact which emerged was the much higher incidence of collapse of ovules sub-
sequent to fertilization in the selfed than in the crossed group. The data are
summarized in Table 5.1. Fertilization occurred within about 30 hours after
pollination under the prevailing conditions. It was somewhat delayed after
selfing. Little evidence of breakdown of the seeds was found at 48 hours. In
the 72 hour and subsequent collections, however, the phenomenon was com-
mon. The results presented in the table cover the period from 72 hours to
144 hours, inclusive, and are based upon 433 seeds and 1682 seeds in the
selfed and crossed series, respectively.

Growth of the young seed at this stage appears to be quite independent

86

R. A. BRINK

of that of its neighbors in the same ovary. Furthermore, the quickly succeed-
ing secondary effects of fertilization, such as enlargement of the surrounding
fruit, are at a minimum. Studies on the reproductive physiology of the flower-
ing plants are rendered difficult by the multiplicity of changes which are
eventually set in motion in the tissues of the seed, the fruit, and the maternal
plant following fertilization. The sequence and interrelations of the events
immediately subsequent to syngamy are simpler to analyze than those which
occur later, in view of the fact that each very young seed may be considered
to behave independently of the others.

The data in Table 5.1 show that, for each of the seven plants tested, the

TABLE 5.1

FREQUENCY OF FERTILE OVULES COLL.APSING IN SEVEN .ALFALFA
PLANTS FOLLOWING SELF- AND CROSS-FERTILIZATION. DATA BASED
ON COLLECTIONS AT 72, 96, 120, AND 144 HOURS .\FTER POLLINATION
(AFTER COOPER AND BRINK, 1940)

Self-fertiliz.'\tion

Cross-fertiliz.ation

Plant

No. of Fertile Ovules

Percentage
Collapsing

Plants
Crossed

No. of Fertile Ovules

Percentage

Selfed

Total

Collapsing

Total

Collapsing

Collapsing

A

B

C

D

E

F

G

37
37
20
17
39
109
55

9

19

7

7

8

39

19

24.3
51.4
35.0
41.2
20.5
35.8
34.5

AXB

BXC

CXD

DXE

EXA

FXG

GXF

Total. .

187
110
171
171
146
228
198

13
5

13
16
9
14
16

7.0
4.5
7.6
9.4
6.2
6.1
8.1

Total. .

314

108

34.4

1211

86

7.1

frequency of seeds collapsing is much higher in the selfed than in the crossed
series. The proportions vary in different individuals from about 3 to 1 to
over 11 to 1. On the average, approximately five times as many seeds con-
taining inbred endosperms and embryos collapse within the first six days
after pollination as in the crossbred group. Since other factors were not
varied, the decrease in survival in the selfed series must be attributed to
the inbreeding.

The evidence, both general and particular, points to the endosperm
rather than the embryo as the seat of the inbreeding depression effect. The
endosperm in alfalfa is free nucleate up to about 144 hours after pollination,
although it develops as a cellular tissue thereafter. Successive waves of
mitotic divisions traverse the tissue, the number of nuclei being doubled in
each cycle. Thus growth during this period proceeds at an exponential rate.

INBREEDING AND CROSSBREEDING IN SEED DEVELOPMENT 87

The concurrent development of the embryo, on the other hand, is relatively
slow. The zygote divides to form a two-celled proembryo. Successive divi-
sions of the apical cell give rise first to a six-celled proembr>'o and then to the
initials of the definitive embryo.

The pronounced difference in rate of development of the two tissues is il-
lustrated by the fact that at 144 hours the modal number of cells in the
embryo is only 16, whereas the typical number of nuclei in the endosperm
at this time is 128. Rapid and precocious development of the endosperm as
seen in alfalfa is characteristic of the angiosperms in general. The much
higher level of activity of the endosperm is presumptive evidence tlvit this
tissue, rather than the embryo, is especially subject to develo{)mental upsets
in the young seed. Data available in the present instance provide direct con-
firmation of this interpretation.

The comparative rates of growth of endosperm and embryo in the selfed
and crossed alfalfa series up to 144 hours after pollination are illustrated in
Figure 5.1. Not only are the values for the embryo low, but also there is
little difference between those for the inbred and crossbred series. The con-
clusion appears warranted that the direct effect of inbreeding on the embryo
at this stage, if indeed there is a demonstrable effect, is too small to account
for the high frequency of seed collapse. In contrast, there is a very sharp
decline in rate of nuclear division in the endosperm, following enforced self-
fertilization of this naturally cross-fertilized plant. The lower rate is shown
from the first division onward. There are about twice as many nuclei present
at 144 hours in the crossbred as in the inbred endosperms.

Due to the partial self-incompatibility in alfalfa, fertilization on the
average, is slightly delayed following selfing. A comparison of the rate of
growth of the two classes of endosperms independent of time as shown in
Figure 5.2, however, establishes the reality of the difference in rate of growth
between the inbred and crossbred endosperms. When the seeds are arrayed
in terms of cell numbers of the enclosed embryo, it is found that for all nine
classes occurring in the material the endosperms are more advanced in the
crossbred than in the inbred series. That is to say, the embryos at a given
stage of development have associated with them more vigorously growing
endosperms following cross-fertilization than after selfing. Moreover, the
decrease in size resulting from the inbreeding is so large that one is led im-
mediately to suspect that herein lies the primary cause of the frequent seed
collapse following selfing.

Why should impairment in rate of endosperm growth lead to arrested seed
development? The answer in the present case is clear. As was pointed out
earlier, double fertilization initiates not only endosperm and embryo develop-
ment, but also a new cycle of growth in the integuments. The latter compete
directly with the endosperm for the nutrients moving into the young seed.
If the endosperm is developing subnormally, a disproportionate amount of

88

R. A. BRINK

the incoming nutrients is diverted to the integuments. As a result this
tissue frequently becomes hyperplastic. The overgrowth in the case of al-
falfa characterizes the inner integument. As Dr. Cooper observed, it begins
at a point opposite the distal end of the vascular bundle where the concen-
tration of nutrients maybe assumed to be the greatest. The inner integument,
which is normally two cell layers in thickness, becomes multilayered and
somewhat callus-like in the region of the greatest mitotic activity. This pro-
nounced overgrowth of the inner integument quickly reacts upon the endo-
sperm, further impairing its development. In the seeds which fail, a complete

48 72 % 120

Time in hours after pollinatior\

m

Fig. 5.1— Increase in number of cells in embryo and in number of nuclei in endosperm
following self- {broken line) and cross-fertilization {continuous line). After Brink and

Cooper, 1940.

INBREEDING AND CROSSBREEDING IN SEED DEVELOPMENT

89

collapse of the endosperm then ensues. Significantly, breakdown of the
endosperm tissue begins in the region opposite the end of the vascular
bundle where the inner integument is especially hyperactive. Following col-
lapse of the endosperm, the young seed dies.

SEED DEVELOPMENT WITHOUT FERTILIZATION

There are a few species of flowering plants in which both endosperm and
embryo develop without fertilization. These so-called autonomous apomicts

1-celled
proembri/o

2-celled
proem bryo

3-celled
proembryo

4-celled
proembryo

5-ceiied
proembryo

6-celied
proembryo

2-celied
emb.*5usp.

3-ceiled
emb.-t-5u5p.

4-celled
emb.+sasp.

1 5elf-fcrtilization
cross-fertilization

0 10 20 50 40 50 60 70

Number of endosperm nuclei

Fig. 5.2 — Number of endosperm nuclei associated with proembryos and embryos at various
stages of development following self- and cross-fertilization. After Brink and Cooper, 1940.

should provide an independent test of the hypothesis that aggressive develop-
ment of the endosperm is requisite to seed development, and that the sec-
ondary fertilization is a device by which aggressiveness of the tissue is en-
hanced. On the basis of the reasoning applied to sexual species, one would
e.xpect to find in autonomous apomicts that the embryo is not basically de-
pendent on an active endosperm for its nourishment. So far as I am aware,
the evidence bearing directly on this question is limited to a single study
which Cooper and I carried out on the common dandelion, Taraxacum
officinale (Cooper and Brink, 1949).

The common dandelion is triploid (3x = 24). The regularity and abun-
dance of seed production in the plant is well known. A full complement of seed

90 R. A. BRINK

forms in the absence of pollination, as may be demonstrated easily by re-
moving the corollas and anthers — by cutting off the distal portion of the
head in the bud stage. Ordinarily the anthers do not open in the intact
mature flower.

The female gametophyte is formed without reduction in chromosome
number of the nuclei. Otherwise it is a typical eight-nucleate, seven-celled
structure lying in direct contact in the mature ovule with the innermost
layer of cells of the single thick integument. The polar nuclei fuse to give a
hexaploid primary endosperm nucleus. The single layer of cells comprising
the nucellus disintegrates during formation of the embryo sac.

Sexual forms of the common dandelion are not known to occur. Accord-
ingly another species, T. kok-saghyz, the Russian dandelion, was examined
as a control. T. kok-saghyz is diploid (2x = 16) and, since it is self-incom-
patible, requires cross-pollination for seed formation. A comparative study
of T. officinale and T. kok-saghyz was made with a view to discovering, if
possible, the means by which the former is enabled to dispense with the
secondary fertilization, which is essential to seed formation in the latter.
Heads were collected at four stages: late bud, just prior to anthesis, open
flower, and with seeds ranging up to six days of age. After sectioning and
staining, the number of cells in the endosperm and embryo was determined,
and observations were made on the amount and distribution of food ma-
terials.

Seed formation in T. kok-saghyz follows the course typical of the angio-
sperms. Endosperm and embryo development are initiated by double
fertilization. Subsequently, the two tissues grow very rapidly, and in tune
with each other. Cell number in the endosperm increases exponentially. The
endosperm, however, is somewhat less precocious than in most flowering
plants. The seed is mature 9-12 days after fertilization.

A markedly different set of relations present themselves in the seed of
the apomictic T. officinale. The seed in this species begins development when
the flowers are in the late bud stage. By the time the flowers open, there may
be 100 cells or more in the endosperm, the embryo, or in both tissues in some
seeds. A further significant fact is the extraordinary amount of variability
in the size ratios of endosperm and embryo from seed to seed of even age.
There is a positive relation between cell number in endosperm and embryo
over the period studied — as would be expected in view of the fact that in
most seeds both tissues are growing. As measured by the correlation co-
efficient, this value is low (r = .57) compared with that for T. kok-saghyz
(r = .76).

Average cell number in the embryo in relation to endosperm size is de-
picted for the two species in Figure 5.3. Cell number in the endosperm in-
creases geometrically, so that size of the tissue may be expressed appropriate-
ly in terms of division cycles. Embryo cell number, in contrast, increases

INBREEDING AND CROSSBREEDING IN SEED DEVELOPMENT

91

arithmetically. It will be noted from Figure 5.3 that the mean embryo cell
number in T. officinale, before the endosperm mother cell divides (0 cycle),
is about 16. The corresponding value T. kok-saghyz is 1. This is a reflection of
the fact that the embryo in the apomictic sj)ecies usually starts growth in
advance of the endosperm. Although they start from different levels, the two
curves are not greatly dissimilar. The embryo in the common dandelion, on
the average, is consistently larger in the young seed than that of T. kok-
saghyz, relative to given stages in endosperm develo])ment.

100

T. officinale

T. kok-soghyz

o

90

>-

oc

m

60

z

UJ

z

70

UJ

60

m

z

50

z>

■^

_J

40

-I

UJ

o

30

2

20

<

UJ

Z

10

4

K.SJ

0 12 3 4 5 6 7

NUMBER OF DIVISION CYCLES IN ENDOSPERM

Fig. 5.3 — Early growth of embryo of T . kok-saghyz and T. officinale in relation to endo-
sperm size. After Cooper and Brink, 1949.

More instructive than the mean values on which Figure 5.3 are based, is
the variability in the frequency distributions concerned. The data are sum-
marized in Table 5.2. A logarithmic scale was used in expressing embryo
sizes merely as a convenient way of summarizing the widely dispersed values.
As mentioned above, growth of the embryo during this period is approxi-
mately linear.

Table 2 reveals that the variability is low in embryo cell number at suc-
cessive stages of endosperm development in T. kok-saghyz. This means that
embryo and endosperm are closely synchronized in their growth in the sexual
species. The variability in embryo size in the apomict, on the other hand, is
enormous. For example, in seeds in which the endosperm is still at the mother
cell stage (0 cycle), the associated embryos are distributed over all size
classes from 1 to 128. The standard deviation for embryo cell number is

92

R. A. BRINK

15.6, a value equal to the mean. The range is even greater in the class of
seeds having 128-cell endosperms, and the standard deviation rises to 51
cells.

The extreme variability in embryo size for given stages of endosperm de-
velopment in T. officinale is a fact of cardinal importance in the present
analysis. Inspection of Table 5.2 reveals certain details which emphasize
the significance of the summary data on dispersion. Note, for instance, that

TABLE 5.2
DISTRIBUTION OF EMBRYOS BY CELL NUMBER RELA-
TIVE TO ENDOSPERM DIVISION CYCLE
(AFTER COOPER AND BRINK, 1949)

Endo-
sperm

Species

Total
Seeds
Ex-
amined

Embryo Cell Number — Logarithmic
Class Values

Stand-
ard

Division
Cycle

1

2

4

8

16

32

64

128

256

Devia-
tion

0

T. officinale
T. kok-saghyz

T. officinale
T. kok-saghyz

T. officinale
T. kok-saghyz

T. officinale
T. kok-saghyz

T. officinale
T. kok-saghyz

T. officinale
T. kok-saghyz

T. officinale
T. kok-saghyz

T. officinale
T. kok-saghyz

111
All

253

77

145

32

108

25

111
34

115
68

99

55

60
19

9

All

23
31

18

1

12

16

ii

57

66

38

7

1

15.6
0

1

11
46

11
31

6

37

70

55

50

6

1

13.6
0.5

2

7

23

43

IZ

9

1

17.0
0.2

3

6

22

2
5

4

19

3

9

27

4
24

27

25

12

1

21.1
0.7

4

4

1

39

2

23
40

7
10

1

40

14

2

19.2
1.9

5

6

50
4

31
41

8
3

23

5

24.2
4.1

6

1

1

46

4

17
16

13

29.9

9.0

7

2

1

28

3

51.0
16.7

among the seeds still in the endosperm mother cell stage (0 cycle) one con-
tains an embryo in the 128-cell class and seven have embryos in the 64-cell
class. Similar, although less extreme, cases occur in the 1 -cycle and 2-cycle
endosperm distributions. Study of the histological preparations shows that
the seeds in which the embryos are found are growing vigorously and appear
capable of completing development. This can mean only that either very
small endosperms in T. officinale are extraordinarily efhcient structures, or
embryo growth in this species is not dependent on an endosperm.

At the opposite corner of the table, on the diagonal, two seeds are entered
in the 7-cycle endosperm array in which the embryos are still in the one-cell
stage. These seeds also appeared to be healthy and capable of continued

INBREEDING AND CROSSBREEDING IN SEED DEVELOPMENT 93

development. These extreme examples {)oint unmistakably to the conclusion
that in the apomictic dandelion the endosperm, as the master tissue in the
young seed, has been disestablished. Embryo growth must be sustained by
other means.

The substitute arrangement for nourishing the embryo in T. officinale was
disclosed by a histological study of the ovules of this species and T. kok-
saghyz. Basically the structure of the ovule is the same in both. As the female
gametophyte expands, the nucellus disintegrates so that the gametophyte
comes to lie in direct contact with the endothelium which comprises the in-
nermost layer of cells of the massive integument. The endothelium [)ersists
and appears to function in the transfer of nutrients during the course of seed
development. In T. kok-saghyz the inner layers of integumentary cells ad-
jacent to the endothelium lose their contents during formation of the game-
tophyte, and contain shrunken and misshapen nuclei when the ovule is
mature. The cells of the integument immediately surrounding this depleted
region are densely cytoplasmic and possess well-defined nuclei. The outer-
most parenchymatous cells of the integument are highly vacuolate. The
single vascular bundle makes an arc about the greatest circumference of the
ovule in both species. Only limited amounts of stainable reserve food ma-
terials occur anywhere in the T. kok-saghyz ovule.

The T. officinale ovule differs conspicuously from that of T. kok-saghyz in
possessing an abundance of reserve food. The cells of the integument just
outside the endothelium enlarge as the ovule matures and become gorged
with a homogeneous material which appears to be proteinaceous in composi-
tion. This substance also extends between the cells at the outer edge of the
storage region proper.

This extensive prestorage of protein-rich food material in the integument
provides an explanation of the fact that embryo development in the apomict
may proceed normally in spite of very limited endosperm growth. The con-
ditions render superfluous an aggressively functioning endosperm. The
embryo draws directly on a food supply already at hand. From the physio-
logical point of view, the nutritive mechanism in the apomict is analogous to
that in the gymnosperms. In both these classes of plants certain of the
processes essential to seed development, which follow double fertilization in
sexual species of flowering plants, are pushed back into the ovule. The
secondary fertilization, which through its efifect on vigor of endosperm growth
may be looked upon as a means of offsetting the tardy provision of nourish-
ment for the embryo, thus can be dispensed with.

SEED DEVELOPMENT GRADE AND EMBRYO
GROWTH POTENTIALITIES

The conclusion that growth of the angiosperm seed is basically controlled
by the endosperm has an interesting corollary. That is, that the grade of seed
development attained after a given mating is not a definitive index of the

94 R. A. BRINK

intrinsic vigor of the embryo. This statement is not intended to imply that
the two phenomena are unrelated, but rather that they vary independently
of each other to a significant degree. Many interspecific matings, for example,
yield poorly developed seeds. Often the embryos in these seeds give rise to
relatively weak plants. Sometimes, however, the embryos within such seeds
are capable of forming plants of great vegetative vigor. In other words, the
fact that development of the seed is impaired, even to a degree that calls for
special methods of propagation, does not necessarily mean that the embryo is
intrinsically weak. The hybrid during the seed stage may merely be the
victim of a faulty endosperm. Only when released from this stricture can the
inherent potentialities of the new individual be expressed.

Two examples of such intrinsically vigorous hybrids in which the condi-
tions of seed development have been explored will be briefly mentioned. They
differ in the grade of seed development attained. Small but nevertheless
germinable seeds are formed in the one case, whereas in the other the embryo
egularly dies unless special precautions are taken to save it.

Cooper and I found that when the diploid (2» = 24) Red Currant tomato,
Lycopersicon pimpinelUfolium, is pollinated with a particular strain of L.
peruvianium., likewise a diploid, fertilization occurs with high frequency but
all the seeds collapse before the fruit is ripe. Seed development follows a
familiar pattern. The endosperm grows less vigorously than in normal L.
pimpinelUfolium seeds, and the endothelium enclosing it tends to become
hyperplastic. Endosperm cells become highly vacuolate and starved in ap-
pearance. Densely staining granules of unknown composition accumulate in
the chalazal region just outside the endosperm, suggesting that the latter
tissue is incapable of absorbing the available supply of nutrients. All the
seeds in the ripe fruit are shrivelled and incapable of germination.

Following the application of pollen from the same diploid strain of L.
peruvianium to a tetraploid (2w = 48) race of L. pimpinelUfolium, about
one-half the fertile ovules develop into small but germinable seeds containing
triploid embryos. The other seeds collapse at various stages of growth. Histo-
logical examination of the 4n L. pimpinelUfolium X 2n L. peruvianium seeds
shows retarded embryo development and a less rapid endosperm growth
than occurs in the normally pollinated tetraploid parent. The endosperm in
sixteen-day-old hybrid seeds lacks the rather densely packed starch reserves
characteristic of tomato seeds at this stage. The peripheral layers of endo-
sperm cells adjacent to the endothelium break down. An unusually large
cavity is formed in the interior of the tissue as a result of digestion of the
cells by the slowly differentiating embryo. Endosperm function is markedly
impaired in this cross, but in many seeds remains somewhat above the
threshold at which complete failure occurs.

The triploid plants resulting from germinable 4» L. pimpinelUfolium X
2w L. peruvianium seeds are extraordinarily vigorous. Although partially

INBREEDING AND CROSSBREEDING IN SEED DEVELOPMENT 95

sterile, they considerably exceed both the parents in capacity for vegetative
growth. The inference is clear that the genie combination resulting from this
cross yields markedly different results in the endosperm and the sister
sporophy te. The difference in part may be a consequence of the 2 : 1 balance
of L. pimpinellifolium and L. peruvianium genes in the embryo as compared
with the 4: 1 ratio in the endosperm. The important point, however, is that
the mechanism of seed formation in the flowering plants is such that the two
products of a given double fertilization may be quite differently endowed in
terms of the genes necessary to perform their respective functions.

The second example to be discussed in this connection will enable us to
visualize the limits which may be reached in endosperm disfunction with
retention of embryo viability.

Fertilization freely occurs when squirrel-tail barley, Hordeum jubatum is
pollinated by cultivated rye, Secale cereale. The resulting seeds all die, how-
ever, within less than two weeks. Space does not permit me to recount here
the steps leading to the breakdown. They have been described in detail else-
where (Cooper & Brink, 1944; Brink & Cooper, 1944). The endosperm early
becomes completely disorganized. Some of the embryos formed, however,
reach a stage previous to collapse at which time they may be dissected from
the seed and successfully reared on an artificial nutrient medium. A single
plant was grown to maturity from an embryo treated in this way. The
plant was thrifty, although sterile. Representatives of the parent species
grown under comparable conditions were not available, so that a valid com-
parison of relative vigor could not be made. The hybrid, however, appeared
to be intermediate in stature and number of tillers.

The extreme character of the endosperm disturbances in the H. jubatum X
S. cereale seed indicates that this hybrid could not arise under field condi-
tions. Although the embryo is demonstrably capable of continued develop-
ment its growth is terminated in the seed due to failure of the associated
endosperm. Death of the embryo, as an indirect result of endosperm disfunc-
tion following wide crosses, appears to be commoner than was thought before
the physiological implications of the secondary fertilization in flowering
plants were recognized. Realization of this fact has stimulated additional
interest in circumventing the phenomenon by excising such embryos from
the seed and rearing them artificially.

Artificial methods of cultivating embryos removed from abortive seeds
often have been used to extend the area within which gene transfers may be
effected. Numerous interspecific hybrids have thus been grown which other-
wise are not realizable. The nature of the general problem involved may now
be seen in somewhat broader perspective. Two points of particular interest
may be noted.

The first, briefly adverted to above, is that the frequency with which em-
bryos are formed following matings between distantly related plants is much

96 R. A. BRINK

higher than earlier believed. Various investigators have expressed the opinion
that the mere presence of growing pollen tubes in the style causes enlarge-
ment of the ovules. This view now appears to be incorrect.

On the other hand, there is a steadily increasing amount of evidence to
show that the incipient growth of the ovules, following many interspecific
matings which do not yield functional seeds, is a response to fertilization.
That is to say, the block in the reproductive cycle which was assumed to
intervene prior to fertilization actually occurs following syngamy. Embrycs
are formed in these cases, but they perish when the young seed fails to de-
velop. Some rather extreme examples of this phenomenon which have been
observed in our laboratory include Nicotiana ghitinosa X Petunia violacea,
N. glntinosa X Lycopersicon esculentum, and Medicago saliva X M. scutellata.

It is not to be inferred that all hybrid embryos of this general class are
capable of growing into mature plants. The fact that the seeds containing
them collapse is not proof, however, of intrinsic inviability. An unknown but
probably significant proportion of these novel zygotic combinations are po-
tentially propagable. The problem is to discover the means by which they
may be reared. This brings us to the second point — the nature of the problem
to be faced in growing very small excised embryos.

With few exceptions, the embryos which have been successfully culti-
vated artificially have been removed from the seed at rather advanced stages
of development. Unless they are multicellular and differentiation has at least
begun, the embryos usually do not grow on the media which thus far have
been devised. There are reasons for thinking that the nutritional require-
ments of these older embryos are simpler than those in a juvenile condition.
Histological evidence shows that at the early stages of seed development the
embryo is enclosed, or nearly enclosed, in the highly active, young endo-
sperm. The endosperm cells adjacent to the proembryo and the very young
embryo remain intact. A little later, as the embryo enlarges, these cells
begin to break down and their contents disappear. Eventually all the endo-
sperm tissue is consumed in most species.

One may infer from these facts that the embryo is dependent upon the
endosperm for certain metabolites which initially the embryo is quite in-
capable of synthesizing. The endosperm may be pictured as secreting the
needed materials at the early post-fertilization stage, and yielding them
later in a more passive fashion as the tissue becomes lysed. Meanwhile the
embryo becomes progressively less dependent upon the endosperm by acquir-
ing for itself the synthetic capabilities previously limited to the nurse tissue.
On this view the very young embryo is an obligate parasite on the endo-
sperm. Once past the state of obligate parasitism, growth of the embryo may
be effectively supported by comparatively simple nutrients such as may be
provided in artificial culture media.

Visualized in those terms, the problem of cultivating very young, excised

INBREEDING AND CROSSBREEDING IN SEED DEVELOPMENT 97

embryos resolves itself into the discovery of means of duplicating the un-
known but presumably special nutritive functions of the normal endosperm.
Two possibilities suggest themselves in this connection. One is to determine
natural sources of the special metabolites produced by the endosperm and
then add these materials to the nutrient medium. Van Overbeek (1942) ob-
tained significant improvement in the growth of small Datura stramonhim
embryos by supplying them with unautoclaved coconut milk. Blakeslee and
Satina (1944) later reported that the coconut milk could be replaced by un-
autoclaved malt extract. The other possibility is to cultivate the embr>^os
artificially in association with actively functioning endosperm tissue. Cur-
rent findings offer some encouragement that the latter procedure may prove
efficacious.

Dr. Nancy Ziebur, working in our laboratory, recently has shown that
the growth of very young embryos of common barley (0.3-1.1 mm. long)
may be greatly improved by surrounding them on a nutrient agar medium
with aseptically excised endosperms. The basic medium employed permits a
satisfactory growth of older barley embryos but does not yield transplantable
seedlings from embryos shorter than about 0.6 mm. except in conjunction
with endosperms. Coconut milk and malt extract are ineffective with barley
embryos. Water extracts of fresh barley endosperms gave positive, although
smaller effects than the intact tissue. Further exploration of the living endo-
sperm as a source of nutrients for very young, excised embryos should prove
rewarding. The interrelationships of these two tissues in the juvenile seed
give strong credence to this approach. The success which has so often at-
tended efforts to grow older embryos artificially on rather simple media may
have blinded us to the fact that the young embryo, divorced from the endo-
sperm, may have quite different requirements.

W. GORDON WHALEY

The Plant Research Insfifufe of fhe Universify of Texas and
the Clayton Foundation for Research

Chapter 6

Physiology of Gene
Action in Hybrids

The physiology of gene action in hybrids is not a subject apart from the
physiology of gene action in organisms in general. The approach to specific
problems of gene action is probably better made in non-hybrid organisms
than in hybrids. Hybrids do, however, represent one type of genetic situation
which in certain instances is particularly favorable for the study of gene
action. Most useful in this respect are those hybrids which exhibit the phe-
nomenon referred to, often rather loosely, as hybrid vigor. The terms hybrid
vigor and heterosis often are used synonymously. A more precise usage, and
one in accord with the original definitions, refers to the developed superior-
ity of hybrids as hybrid vigor, and to the mechanism by which the superior-
ity is developed as heterosis. By this definition, hybrid vigor is heterosis
manifest. Because in studies of growth and development it is often desirable
to distinguish clearly between mechanism and end result, this use of the two
terms will be followed in this chapter.

Heterosis has been the subject of many experiments and a great deal of
speculation on the part of geneticists. The concern has been mostly with the
genetic bases of heterosis, and relatively little attention has been given to the
physiological mechanisms involved. As a matter of fact, the literature on
heterosis mirrors faithfully the changing emphasis in genetics in the last two
or three decades. Practically all of the early investigations of heterosis had
to do with the comparison of mature characteristics of inbred lines and their
vigorous hybrids, and then with attempts to formulate genetic schemes in
explanation of the differences. Gradually, the focus of investigation has
turned to a study of developmental differences responsible for the hybrid
vigor, and more recently to the gene action bases of these developmental
differences.

98

PHYSIOLOGY OF GENE ACTION IN HYBRIDS 9^

It is a fair hope that from detailed studies of the nature and development
of heterosis, much will in time be revealed about specific gene action. Un-
fortunately, most of the studies up to the present time have been directed
to general rather than to specific considerations. It has been necessary to deal
in terms of size differences, yield differences, and growth rate differences, un-
til enough of the pattern should appear to indicate what specific physio-
logical considerations are likely to be involved in heterosis. Because we have
come only to this point and have proceeded but a little way in an analysis of
these specific physiological considerations, this chapter will have to deal
more with suggestions of the likely mechanisms than with data from investi-
gations of them.

It is neither possible nor desirable to separate wholly the consideration of
the physiological mechanisms of heterosis from the genetic bases. Our main
concern will ultimately be with the genes involved and the nature of their
action.

The word hybrid has no good, definitive genetic meaning. It can be used
with equal propriety to refer to organisms which approach complete hetero-
zygosity or to organisms which are heterozygous for only a small number
of genes.

There is at least a rough relationship between the amount of heterosis in a
hybrid and the extent of the genetic differences between the parents. Physio-
logical and morphological diversity are dependent both upon the number of
allelic differences between organisms and upon the nature of the action of
the particular genes among which these allelic differences exist. It is quite
possible that organisms differing by only a few genes may be more widely
separated in certain characteristics than are organisms differing by many
more genes — the actions of which are of less fundamental significance for
the control of the developmental pattern.

In our approach to questions of hybrid vigor, we may be concerned with
different degrees of hybridity. Consideration of this factor must involve not
only the number of genes but also the nature of the action of the particular
genes. Nor is this all, for the action of any specific allele is conditioned by the
genetic background in which it occurs in a particular individual. Hence, the
relations among genes may often be of critical importance.

Of tremendous import, too, are the interactions between the activities of
the genes and the environment. In speaking of hybrid vigor, we are general-
ly concerned with such characteristics as size and yield, but these are merely
end products of the metabolic processes. Patterns of these metabolic proc-
esses are set by the genes, but the processes themselves may be either ac-
celerated, inhibited, or otherwise modified by the effects of environmental
factors. Hybrids which are particularly vigorous under certain conditions
may show relatively little vigor under other environmental conditions. It is
true that the enhanced vigor of hybrids frequently gives to them a wide

TOO W. GORDON WHALEY

range of environmental adaptability. It is equally true that certain hybrids
exhibit vigor within only relatively narrow environmental limits. For lack of
evidence it must be assumed that the distinction lies in the differences be-
tween the patterns of hybridity and in the action of the genes responsible for
the hybrid advantages.

Any attempt to explain the genetic basis of heterosis must make initial
recognition of one fact. The phenomenon can involve only the recombination
of alleles already existing in the population or populations from which the
hybrid organisms have been developed; unless, by rare chance, mutation
should take place just prior to or just after the actual crossing. We are thus
concerned with an interpretation limited to different types of recombina-
tions, and to different kinds of gene action resulting from these recombina-
tions.

GENETIC MECHANISM OF HETEROSIS

Consideration of the characteristics of dominance and heterozygosity has
been of primary importance to investigators concerned with interpretation
of the genetic mechanism of heterosis. Jones's dominance of linked factors
hypothesis (1917) probably is still the most popular explanation of the
genetic basis of heterosis.

Dobzhansky (1941) and his co-workers, and many others, have recorded
that in most species there has been, in the course of evolution, accumulation
of deleterious recessive characters, which when homozygous reduce the
efficiency of the organism — but which in the heterozygous condition are
without efficiency-reducing effects. This revelation calls for a reshaping of no-
tions regarding the nature of the favorable effects of the dominant alleles, but
does not otherwise modify the structure of the explanation. The favorable-
ness of the action of many of the dominant alleles probably is not the result
either of directional mutation producing more favorable dominants or of
selection tending to eliminate the unfavorable dominants. Instead, it may
be due to the accumulation in populations of deleterious recessive mutations.
These, if their effects are not too deleterious, often can be piled up in sig-
nificant numbers.

The piling-up of such deleterious recessives is probably one of the reasons
why heterosis is a much more important phenomenon in such a plant as corn
than it is, for example, in the tomato. Corn has been handled for hundreds
or even thousands of years in a manner that has made possible the accumula-
tion in populations of relatively large numbers of deleterious recessive modi-
fiers. The tomato is more than 90 per cent self-pollinated, and any great
accumulation of deleterious modifiers is unlikely. Corn populations char-
acteristically contain thousands of individuals, and wind pollination makes
for maintenance of heterozygosity. In tomato, the effective breeding popula-
tion size approaches one, and deleterious mutations would tend to become

PHYSIOLOGY OF GENE ACTION IN HYBRIDS 101

homozygous with sufficient frequency to bring about the elimination of
many of them.

As a matter of observation, it would seem that a comparison of the occur-
rence and degree of heterosis in different species, along with a consideration
of the reproductive mechanisms in the various species, supports the proposal
that heterosis in many cases is the result of the covering uj) in the hybrids of
deleterious recessive alleles with a consequent return to vigor. The often
stated argument that hybrids of corn, for instance, frequently are more
vigorous than the original open-pollinated populations from which the in-
breds used in their production were derived, has no validity with respect to
this situation. In the production of the inbreds there is invariably a reassort-
ing of the alleles of the open-pollinated populations.

It is highly improbable, however, that dominant alleles operating either
because of certain inherent favorable characteristics of their own, or simply
to prevent the deleterious activity of recessives, present the only genetic
basis of heterosis. Dominance is by no means the clear-cut feature described
in Gregor Mendel's original paper. The dominance of a particular allele may
be conditioned by the environment, or it may depend upon the genetic
background in which the allele exists. A completely dominant effect of one
allele over another, in the classic sense of our utilization of the word domi-
nance, is by no means universal.

Rather unfortunately the so-called heterozygosity concept of heterosis has
usually been introduced as being in opposition to the dominance explanation.
Because the concepts of the features of dominance and recessiveness early
put them into rigid categories, it has been difficult to postulate how a hetero-
zygous condition with respect to one or more genes could render an organism
more vigorous than the homozygous condition, usually of the dominant
alleles.

Evidence of significance for the interpretation of the importance of hetero-
zygosity in heterosis has been accumulated slowly. There is now, however, a
fairly long list of instances in many different species in which the heterozy-
gous condition for certain alleles is known to be superior to either the homo-
zygous recessive or the homozygous dominant condition (Stubbe and
Pirshcle, 1940; Singleton, 1943; Karper, 1930; Robertson, 1932; Robertson
and Austin, 1935; Gustafsson, 1938, 1946; Nabours and Kingsley, 1934;
Masing, 1938, 1939a, 1939b; Rasmusson, 1927; and Timofeef-Ressovsky,
1940.

The accumulation of data on these cases followed a long period during
which all the investigations reported seemed to indicate no marked differ-
ences between organisms heterozygous for certain alleles and those with the
dominant homozygous condition for these same alleles. At least, in no in-
stance, was there any marked superiority referable to the heterozygous
condition. Most of the genes involved in the more recent findings have been

102 W. GORDON WHALEY

catalogued as having at least moderately deleterious effects in the mutated
state. The characteristics controlled by them include: chlorophyll deficien-
cies, modifications of leaf form and pigmentation, stalk abnormalities, flower-
ing pattern, and time of flowering.

The extent to which the actual nature of the genetic situation has been
analyzed varies, but in several of the cases it seems clear that the mutation
of a single gene is involved and that the Fi hybrids are heterozygous only
with respect to the alleles at this particular locus. The amount of heterosis
manifest also varies greatly. Because of experimental differences, no accurate
comparisons can be made, but in some instances the amount of hybrid vigor
appears to be nearly comparable to that which occurs in crosses involving
large numbers of allelic differences. The situation appears to be one in which
a mutation takes place, and the mutated allele is definitely deleterious when
homozygous. In individuals heterozygous for the particular gene, there ap-
pear none of the deleterious effects. Instead, a definite heterotic effect ap-
pears. Dominance is of no apparent importance, and the distinction between
the vigorous hybrids and the less vigorous non-hybrids rests upon hetero-
zygosity.

Jones (1944, 1945) has reported several cases of what he has called heter-
osis resulting from degenerative changes. He first suggested that these cases
represented instances of heterosis with a genetic basis in the heterozygosity
of certain of the mutated genes. More recently (private communication)
Jones has concluded that these cases involve more than single gene differ-
ences, and that the results may be explained on the basis of an accumulation
of favorable dominant effects.

The case of a single locus heterosis reported by Quinby and Karper (1946)
involves alleles which do not produce any detectable deleteriousness, but in
certain heterozygous combinations produce hybrid vigor comparable in
amount to that in commercial hybrid corn. Quinby and Karper have referred
the hybrid advantage in this case to a stimulation of meristematic growth in
the heterozygous plants.

All of these instances involve specific allelic interactions and not superior-
ity resulting from heterozygosity per se — as was postulated by some of the
earlier workers concerned with the genetic interpretation of heterosis. These
examples contribute to the increasing realization that the phenomenon of
dominance is perhaps of less importance with respect to heterosis than has
been supposed. There is no a priori reason why the interaction of a so-called
recessive allele and a so-called dominant allele should not give results differ-
ent from and metabolically superior to those which are conditioned by either
two recessives or two dominants.

This situation bears closely upon the interpretation of heterosis set forth
by East in 1936. East postulated that at the loci concerned with the
mechanism of heterosis there might be a series of multiple alleles — with the
combinations of different alleles giving results metabolically superior to

PHYSIOLOGY OF GENE ACTION IN HYBRIDS 103

those determined by the combinations of like alleles, and with no considera-
tions of dominance being involved. In the light of existing evidence it seems
a safe assumption that a considerable portion of hybrid vigor is the result
of allelic interaction between different alleles at the same locus. Although the
evidence as yet is scanty, it is certainly pertinent to suggest that some
heterosis may result from the interaction of alleles at different loci, when
such alleles are brought into new combinations in the hybrids.

Most of the recent studies of the relation of heterozygosity to heterosis
have been concerned with the results of the action of single genes. Such
studies have emphasized that heterosis need not have its basis in the action
of large numbers of genes but can be, and apparently frequently is, a result
of the combining of different alleles of a single gene. Any considerable amount
of hybrid vigor resulting from the action of single genes would seem to indi-
cate the involvement either of multiple effects of single genes or of genie
action in the control of relatively fundamental metabolic processes. Both are
likely probabilities.

The metabolic system of any organism which grows and functions in a
satisfactory manner is an exceedingly complicated mechanism with a great
number of carefully balanced, interrelated processes. The mutation of any
gene which has control over any of the key processes or functions will almost
certainly be reflected in a number of processes and activities. For example, if
a change in the character of some fundamental enzyme system is involved,
either the addition or subtraction of a functional step, or of a substance
produced at a particular developmental stage, would be likely to enhance or
inhibit a number of important processes in the general metabolism of the
organism.

The equilibrium factor in genie action is obviously a consideration of
great importance. If a mutation disturbs this equilibrium after it has become
fairly well established through selection and elimination processes, the con-
sequences may reduce the organism's vigor. If, in a hybrid, the mutation is
then brought together with the original wild type or normal allele, the sum
total of the actions of the mutated allele and the original allele may well be
such as to exceed that of two copies of the original allele in the production
of vigor in the organism.

When we give attention to physiology of gene action in hybrids which are
heterotic, we must concern ourselves with all of these considerations in-
cluding the fact that a single gene, the mutation of which affects some
processes in a sufficiently fundamental stage of the organism's formation,
may well have a greater end effect than a number of genes whose functions
are concerned with more superficial developmental processes.

SEED AND EMBRYO DEVELOPMENT

The literature on heterosis contains a number of discussions concerning
the relation between seed and embryo size and heterosis (Kiesselbach, 1926:

104 W. GORDON WHALEY

Ashby, 1930, 1932, 1937; East, 1936; Sprague, 1936; Copeland, 1940; Mur-
doch, 1940; Kempton and McLane, 1942; Whaley, 1944, 1950).

Most of the investigations have dealt with mature seed and embryo size.
The evidence shows that in many instances hybrid vigor is associated with a
high embryo weight. In some cases the initially high-weight embryo is found
in a relatively large seed. There is, however, by no means a consistent correla-
tion between either high embryo weight or large seed size and heterosis.

The results of studies on corn inbreds and hybrids in our own laboratory
(Whaley, 1950) seem representative of the general findings. Among some ten
inbred lines there occurred a great deal of variation from one line to another
as to both embryo weight and seed weight. There was somewhat more varia-
tion with respect to embryo weight. Among the Fi hybrids, all of which
exhibited considerable vigor under central Texas conditions, there were a
few with embryo weights which exceeded those of the larger-embryo parent.
For the most part, the embryo weights were intermediate, and in one or two
cases they were as low as that of the smaller-embryo parent. The weight of
the seed tissues other than the embryo tended to follow that of the pistillate
parent, but was generally somewhat higher. Double crosses which had vigor-
ous Fi hybrids as pistillate parents characteristically had large seeds with
what were classified as medium-weight embryos.

The few reports, such as Copeland's (1940), concerning the development
of embryos in inbred and hybrid corn, suggest that at the earlier stages of
development some hybrid vigor is apparent in the hybrid embryos. The
observations of hybrid vigor during early development of embryos and the
absence of any size advantage at the time of seed maturity are not necessari-
ly conflicting. In most plants, embryo and seed maturation represent fairly
definite stages at which a certain degree of physiological maturity and of
structural development has been attained. It is probably to be anticipated
that even though certain heterotic hybrids show early embryo development
advantages, these advantages may be ironed out by the time the embryo
and the seed mature. The size of both the embryo and the other seed tissues
is conditioned not only by the genotype of these tissues themselves, but also
by the nutritional background furnished them by the plant on which they
grow.

It is quite possible that this genotype-to-background relationship is an
important consideration in the determination of whether or not hybrid vigor
is exhibited in the development of the embryo and seed. The background
provided by the pistillate parent might be such as to preclude the develop-
ment of embryo vigor, even though the embryo genotype were of a definitely
heterotic constitution. The fact that hybrid vigor is apparent during certain
stages of embryo and seed development may or may not be related to an
embryo or seed size advantage at maturity. Because of this, it seems doubt-
ful that embryo or seed size is a reliable measure of hybrid vigor; and that

PHYSIOLOGY OF GENE ACTION IN HYBRIDS 105

the rate of development during the embryo and seed maturation i)eriod is
of any critical importance with respect to the development of hybrid vigor
during post-embryonic growth.

EARLY SEEDLING GROWTH AND HETEROSIS

There have been few studies of early postgermination growth in plants in
relation to heterosis. It would seem that the usual failure to find higher
growth rates during the grand period of growth, or longer continued growth
periods in heterotic hybrids, would suggest that the answer to the develop-
ment of hybrid vigor lies for the most part in the early postgermination
growth stages. The work of Ashby and his co-workers (Ashby, 1930, 1932,
1936; Hatcher, 1939, 1940; Luckwill, 1937, 1939) emphasized that the hybrid
advantage in their materials was either present in the resting embryo or be-
came manifest in early postgermination growth. Its development was defi-
nitely not a characteristic of the later growth phases. This observation has
now been made for many cases of hybrid vigor (Whaley, 1950). There are
some instances in which hybrid vigor seems to be the result of longer-con-
tinued growth on the part of the hybrid. These probably have a dilTerent
explanation from the majority of cases.

We have been concerned lately in our own laboratory with an analysis
of the early postgermination growth of corn inbreds and single and double
cross hybrids (Whaley, 1950) . Studies of growth during the first ten to twelve
days after germination have revealed that the hybrid advantage is largely
the result of the heterotic hybrid plants reaching a high growth rate earlier
than do the inbreds. Almost without exception, the development of the hy-
brid advantage takes place very rapidly in the early stages of germination
and growth. Rarely have we seen evidence of the hybrids having higher
growth rates during any later part of the developmental cycle. Neither are
the hybrid growth periods extended appreciably beyond those of the in-
breds. In most instances the hybrids mature somewhat more rapidly than
the inbreds — a fact of common observation among plant breeders.

Since the attainment of the maximum growth rate takes place more
quickly during the early stages of development, the hybrids do have a longer
maximum growth rate period. During this period the early advantage is
compounded, to give a considerably greater maturity advantage. When
both the inbred lines and the hybrids used in our studies are considered, it is
apparent that the rapid attainment of high early growth rates is correlated
with relatively low embryo weights. This apparent higher efficiency of small
embryos and its importance in relation to hybrid vigor requires further study.

On the basis of the data at hand one can suggest that the hybrid advantage
lies in the more rapid unfolding of certain metabolic processes, a suggestion
which receives support from the recorded studies of later growth.

106 W. GORDON WHALEY

LATER GROWTH AND HETEROSIS

It is unfortunate that most studies of the physiology of heterosis have been
confined to the later growth period, and consequently do not include that
part of the growth cycle during which the important differences seem to be
developed. Nonetheless, we can learn much from these studies of later
growth as to the nature of the physiological differences which may furnish
bases for the development of hybrid vigor.

The early experiments on physiological differences between inbreds and
hybrids were concerned mostly with the responses of the inbreds and the
hybrids to different soil conditions. A few examples will serve to indicate the
type of investigation and the character of the results. Hoffer (1926) deter-
mined the amounts of the constituents of the ash of heterotic hybrid corn to
be generally intermediate between those of the parental types. He noted that
iron and aluminum were present in the ash of the hybrids in smaller amounts
than in the inbreds. His studies showed that although there were marked
differences in the absorption of iron and aluminum in different soil types the
vigorous hybrids tended to absorb less of both these elements than the less
vigorous inbred lines.

In the same year Kiesselbach (1926) reported distinct differences in water
requirements between selfed lines of corn and their heterotic F; hybrids. The
low productivity inbreds had much higher water requirements than the
vigorous Fi hybrids, when water requirements were calculated on the basis of
either water absorbed per gram of ear corn or water absorbed per gram of
total dry matter. Barley inbreds and heterotic barley hybrids were shown
by Gregory and Crowther (1928, 1931) to make distinctly different responses
to various levels of available minerals. These investigators postulated that
heterosis in barley might be directly related to differences in the ability of
the hybrids and the inbreds to use certain nutrients. This suggestion has had
a fairly adequate test, particularly with reference to nitrogen and phos-
phorus nutrition.

The work of DeTurk et al. (1933), Smith (1934), Lyness (1936), Harvey
(1939), Burkholder and McVeigh (1940), and Rabideau et al. (1950), has
provided a fairly adequate picture of the relation of phosphorus and nitro-
gen nutrition to the development of hybrid vigor. Smith demonstrated dis-
tinct differences among inbred corn lines with respect to phosphorus nutri-
tion, noting that these differences were most apparent when the phosphorus
supply was limited. He postulated that the higher phosphate utilization effi-
ciency of the hybrids might be referred, at least in part, to the dominant in-
heritance in them of a much branched root system. Later studies have shown
that the root growth pattern is certainly important in relation to heterosis.

Smith noted particularly that when inbred lines were inefficient in the
utilization of phosphorus or nitrogen, crossing them failed to produce hybrids
showing any evidence of physiological stimulation resulting in the more

PHYSIOLOGY OF GENE ACTION IN HYBRIDS 107

efTective use of these elements. Lyness (1936) studied heterotic Fi hybrids
resulting from crosses between a low phosphorus-absorbing capacity inbred
and a high phosphorus-absorbing capacity inbred. He found the heterotic
Fi plants to have high phosphorus-absorbing capacity. These results sug-
gested that phosphorus-absorbing capacity in corn, in some instances at
least, acts genetically as a dominant factor. Lyness also noted the relation-
ship between high phosphorus absorption and the extent of root develop-
ment. He supposed that the extent of root development might be responsible
for varietal differences in phosphorus absorption, a supposition which is sup-
ported by later studies. The work of DeTurk et at. (1933) suggested that more
than simply phosphorus-absorbing capacity is involved. This work revealed
that the actual phosphorus content patterns of two Fi hybrids of corn were
quite different. By estimating the amount of phosphorus in various chemical
fractions, De Turk and his coworkers were able to demonstrate marked phos-
])horus pattern difTerences and to associate these pattern differences with
various phosphate fertilizer treatments.

In our laboratory we have made a study of the phosphorus-absorbing ef-
ficiency of corn inbreds and hybrids, and have attempted to correlate the
findings of this study with developmental changes in the vascular system
and with general growth (Whaley et al., 1950; Heimsch et al., 1950; Rabideau
et al., 1950). The data indicate that heterotic hybrids definitely absorb more
radioactive phosphorus than their inbred parents. This advantage in ab-
sorption on the part of the hybrid is associated with more rapid early de-
velopment, with earlier attainment of maturity, and with certain features of
vascular organization. The greater absorption can be referred at least in
part to better early development of the root system in the hybrids, and to a
generally higher level of metabolic activity which presumably creates a
greater phosphorus demand. The greater absorption of phosphorus by the
hybrids is certainly one of the factors which compounds the heterotic effects,
but it seems doubtful that it is a primary factor in the development of hybrid
vigor.

Harvey's (1939) studies of nitrogen metabolism among inbreds and hy-
brids of both corn and tomato showed differences from one line to another
with respect to the ability to use nitrate and ammonium nitrogen. The ex-
periments were of such a nature as to make it clear that such differences in
nutritional responses were results of differences in genetic constitution. The
behavior of hybrids produced from the inbreds reflected a combination of the
characteristics of the inbreds. Significantly, Harvey's study revealed that not
only did differences exist among his inbreds and hybrids with respect to the
ability to use different types of nitrogen, but that there were distinct genetic
differences in the responses of the plants to various levels of nitrogen avail-
ability.

Somewhat similar differential responses to potassium availability were

108 W. GORDON WHALEY

revealed by Harvey's studies on tomato inbreds and hybrids. Burkholder and
McVeigh (1940) have also noted differences in responses of corn inbreds and
hybrids to various levels of available nitrogen. These investigators corre-
lated apical meristematic development, and the differentiation of the vascu-
lar system with the level of nitrogen nutrition, and the efficiency of different
lines and hybrids in utilizing the available nitrogen. Their results indicate
that hybrid vigor, involving superiority in the production of dry matter
and the differentiation of organs, was not correlated with greater growth and
development of the vascular system.

There definitely are vascular organization differences between the heterot-
ic hybrids and the inbreds in the material we have studied. These vascular
organization differences seem not to be the result of differences in mineral
absorption and distribution, but rather to be one of the factors responsible
for the differences in absorption and distribution. All the evidence seems to
indicate that the greater absorption of minerals by heterotic hybrids can be
referred to better developed root systems in the hybrids, probably also to the
presence of more efficient transport systems, and to a generally higher level
of metabolic activity.

Recently we have undertaken a rather extensive analysis of both the
morphological and physiological characteristics of a tomato cross in which
there is marked heterosis. We have found no significant differences between
the inbreds and the hybrids as to total phosphorus content of the leaves,
stems, or roots. There is some suggestion that the phosphorus content of the
organs of the hybrids reaches a higher level earlier in growth than it does in
the inbreds. Neither do the hybrid plants have any consistent advantage
with respect to nitrogen content.

Analyses of the starch content of the leaves and stems suggest that the
hybrid plants may have a slightly higher starch content than the inbreds
during the early growth stages. In terms of average figures over the whole
growth period, however, there are no marked differences between the in-
breds and the hybrids. The same appears to be true of the sugar content.
The hybrids have a somewhat higher sugar content, at least in the leaves,
early in development. During the greater part of the growth cycle the hy-
brids do not have significantly more sugar than the inbreds. The only clear
difference found between the inbreds and the hybrids is in the catalase ac-
tivity of the shoot tips, the hybrids having an appreciably greater index of
catalase activity than either of the inbred parents. The catalase activity
differences are associated with much more active meristematic growth in the
hybrid plants.

THE ROLE OF SPECIFIC SUBSTANCES IN HETEROSIS

Evidence for another sort of physiological differences possibly involved in
heterosis is furnished by the work of Robbins (1940, 1941a) in assaying the

PHYSIOLOGY OF GENE ACTION IN HYBRIDS 109

growth-promoting activities of extracts from inbred and hybrid corn grains.
Robbins' evidence indicates that a substance or substances, which he has
designated as factor Z, may be synthesized in greater amounts by the hy-
brids than by the inbreds. He has stated that factor Z can be fractionated into
Zi, which is hypoxanthine; and Z2, a still unidentified fraction. Robbins'
work suggests that among the advantages possessed by heterotic hybrids
may be the ability to synthesize certain growth substances which the in-
breds either cannot synthesize or cannot synthesize as well.

Further evidence of a slightly different nature is provided by the root
culture work of Robbins (1941b) and of Whaley and Long (1944). Robbins
used cultures of a strain of Lycopersicou pimpiiieUijolium Mill., a strain of
L. esculeutum Mill., and their Fi hybrid, in solutions supplemented by thia-
min, thiamin and pyridoxine, or thiamin, pyridoxine, and nicotinamide.
Robbins found that the Fi roots grew much more rapidly and produced
more dry matter than those of either parental line. He was able to show
further that one parental line made a greater response to the presence
of pyridoxine than did the other, while the roots of the second parental line
made a greater response to nicotinamide than those of the first. This suggests
the combination of complementary factors from the parents in the hybrid.
Whaley and Long (1944) obtained essentially the same results with a cross
involving two inbred lines of L. esculeutum.

In the University of Texas tissue and organ culture laboratory, we have
been exploring certain aspects of this problem. While the results are not suf-
ficiently complete for publication, some facts are already clear. Among the
roots of many inbred lines of tomatoes which we have been culturing, there
are marked differences in growth responses associated with the availability
or non-availability of thiamin, pyridoxine, niacin, and certain other sub-
stances. These differences appear definitely to be inherited and they can be
studied in either the inbred lines or hybrids.

It is still too early to say what the inheritance pattern is, but consideration
can be given to some aspects of the growth response patterns. One of the
most significant revelations is that the responses of most of the roots to a
specific substance are conditioned not only by the availability of that sub-
stance, but by the availability of the other substances and by the gen-
eral composition of the culture medium. Heterosis in tomato root cultures
is, like heterosis in whole plants, definitely relative, and conditioned, not
only by the environment, but, with respect to any specific gene action, by
the background of other gene actions taking place in the developing or-
ganism.

Heterosis in tomato root cultures is definitely related to the inheritance of
the capacity to synthesize or utilize such substances as thiamin, ])yridoxine,
and niacin. This is not to suggest that heterosis in whole plants of tomato
may have its basis in the genetic recombination of factors concerned in the

no W. GORDON WHALEY

control of thiamin, pyridoxine, or niacin metabolism. In intact plants, it is
likely that the green parts supply these substances to their own tissues and to
the roots, in amounts satisfactory for growth and development. The root
tissue responses, however, are definitely heterotic in certain instances, and
these mechanisms merit examination.

It seems pertinent to explore the role of these B vitamins in growth and
development. Thiamin appears to be a metabolic requirement for all types of
cells. Its metabolic activity apparently revolves around a role in enzyme
systems. Thiamin pyrophosphate is the co-enzyme of the enzyme pyruvate
carboxylase (Lohmann and Schuster, 1937). The enzyme carboxylase occurs
in many plant tissues. The possible biochemical basis of thiamin action in
plants has been set forth in some detail by Bonner and Wildman (1946),
Vennesland and Felsher (1946), and Bonner and Bonner (1948). It is assumed
that thiamin represents a step in the development of co-carboxylase which is
active in one or more of the decarboxylating enzyme systems of the respira-
tory mechanism.

Pyridoxine also has an enzymatic role, apparently being important for its
conversion to pyridoxal phosphate, which is a co-enzyme of one or more of
the reactions in the nitrogen metabolism of the plant (Bonner and Bonner,
1948). As a co-enzyme active in nitrogen metabolism reactions, pyridoxine
may be of extreme importance in amino acid-protein building, and hence
active in conditioning fundamental growth activities.

Similarly, niacin activity is enzymatic in character. Niacin appears to be
involved as a constituent of the nucleotide cozymase, and possibly of tri-
phosphopyridine nucleotide. Cozymase is a co-enzyme for a whole series of
dehydrogenase enzymes, including alcohol dehydrogenase, malic dehydrog-
enase, and glutamic dehydrogenase (Bonner and Bonner, 1948).

The genetic background of thiamin, pyridoxine, and niacin metabolism is
thus a genetic background concerned with basic components of the plant's
enzyme systems. Heterosis, which rests upon recombinations concerned with
thiamin, pyridoxine, or niacin metabolism, quite obviously rests upon recom-
binations which are concerned with the acceleration, inhibition, or blocking
of specific stages or developed substances in the basic enzyme system.

A considerable amount of supporting evidence for the involvement of such
fundamental enzyme and other growth substance activities in the develop-
ment of heterosis has been coming for some time from the work on Neuro-
spora. In many heterocaryons of Neurospora, increased growth responses
directly suggestive of heterosis have been observed. In a number of instances
(Beadle and Coonradt, 1944), the growth responses depend upon the two
types of nuclei in the heterocaryon — each carrying wild type alleles of de-
leterious mutant genes carried by the other nucleus. Such instances represent
essentially the same situation as the recombination of favorable dominant
alleles in normally diploid organisms.

PHYSIOLOGY OF GENE ACTION IN HYBRIDS 111

In one case reported by Emerson (1948) a different situation obtains. A
mutant strain of Neurospora which requires sulfonamides for growth at cer-
tain temperatures will grow satisfactorily in the absence of sulfonamides,
provided that the concentration of available p-aminobenzoic acid is held at a
particular level. Either higher or lower concentrations of p-aminobenzoic acid
result in growth inhibitions. Emerson has made heterocaryons between a
mutant strain carrying the sulfonamide-requiring gene {sfo) and a gene which
prevents the synthesis of p-aminobenzoic acid (pab), and a strain carrying
sfo and the wild type allele (+) of pab. The resultant heterocaryons grow
vigorously on the minimal medium (without sulfonamides), whereas strains
carrying sfo and pab, or sfo and +, make no appreciable growth on the
minimal medium. Emerson's explanation of the growth of the hetero-
caryons is that it results from a balance between the production of p-amino-
benzoic acid by one of the types of nuclei and the absence of production of
p-aminobenzoic acid by the other type of nucleus; so that the total produc-
tion of p-aminobenzoic acid is sufBcient for growth but still within the range
tolerated by strains carrying sfo. Heterosis-like effects of this sort are sugges-
tive of the instances of heterosis related to the heterozygosity of particular
genes in diploid organisms.

We thus have in Neurospora, heterosis-like effects assignable both to a
recombination of dominant alleles basis and to a heterozygosity basis. More
important for this discussion is the fact that these instances are all concerned
with facilitation in the hybrid of the production or utilization of substances
which are components of the basic enzyme or other growth substance pat-
tern of the organisms.

Various investigations of heterosis in Drosophila, while for the most part
not concerned with specific growth substances, have nonetheless assigned
manifestation of heterosis to a background in the fundamental biochemical
activities of the organisms. Inasmuch as these investigations are discussed in
detail in another chapter, they will not be treated here.

THE PHYSIOLOGICAL BASIS OF HETEROSIS

From consideration of the pertinent data, a definite pattern emerges.
This associates the development of heterosis with the ability of the hybrid
to synthesize or to utilize one or several specific substances involved in the
fundamental growth processes of the organisms. Nutritional factors, water
absorption factors, and the other more gross considerations with which in-
vestigators have been particularly concerned seem to be secondary factors^ —
perhaps responsible for compounding the heterotic effects but probably not
responsible for their initial development. Much of the evidence agrees with
the assumption that the primary heterotic effect is concerned with growth
substances whose predominant activity is registered in the early part of
the developmental cycle; in plants, especially in early postgermination

112 W. GORDON WHALEY

growth. Into this category fall the enzymes, the auxins, and the other "phys-
iological key" substances.

Many heterotic hybrid plants seem to gain their advantage within the first
few hours after germination. This advantage may not be shown as statistical-
ly significant until it has been further heightened by subsequent growth.
The primary growth activities during this period are those involved in the
unfolding of the enzymatic pattern; the mobilization, transformation, and
utilization of stored materials, and the building up of active protoplasmic
synthesis. It seems definitely to be here that the hybrid advantage lies. By
the time growth is well under way, the hybrid advantage is already well
developed.

Structural diflferences between inbreds and heterotic hybrids shown by the
studies of Burkholder and McVeigh (1940), Weaver (1946), and the members
of our laboratory (Whaley et al., 1950; Heimsch et al., 1950; Rabideau ef al.,
1950) are apparently to be regarded as results of heterosis rather than as
causal factors. The evidence suggests that heterosis is concerned primarily
with growth processes and that differentiation activities are most likely in-
volved secondarily rather than primarily. What seems to be indicated is the
assignment of the physiological basis of heterosis to the activity of one or
more of the so-called physiologically active substances involved in early
growth.

Much of the apparent hybrid vigor is assignable to these activities only in
a secondary fashion. Once the advantage of a larger number of growing
centers or of heightened meristematic activity is established, the greater
availability of nutrients, the greater amount of protoplasm involved in
further protoplasm building, and other general advantages tend to increase
the initial differences. To the general evidence in favor of this supposition
can be added the specific evidence of the few cases in which the physiological
action of particular alleles is known. Where these alleles in combination are
responsible for heterosis, they have — when studied in sufficient detail —
invariably been shown to be alleles whose action involves basic enzyme or
other growth substance activity.

If we are to make significant headway in understanding the physiological
mechanism of heterosis, we shall have to concentrate on a detailed study of
the developmental physiology of early growth. Much of the general knowl-
edge we already have can contribute toward this understanding if we trans-
late it into terms signifying that when we speak of quantitative differ-
ences— size, yield, or of rate difTerences — we are really concerned with differ-
ences in the level of metabolism. We must recognize that these differences in
the level of metabolism are bound to vary against different environmental
backgrounds, and where the particular genes involved are associated with
different genetic backgrounds.

Our approach to the heterosis problem has been complicated by common

PHYSIOLOGY OF GENE ACTION IN HYBRIDS 113

insistence upon attempts to find a single genetic mechanism. It has suffered,
too, from faikire to recognize that between the gene and the final mature
organism there lies a system of developmental processes of great complexity.
The complexity of this system is formidable but it surely can be analyzed,
at least with respect to its most significant features, if it is taken part
by part.

SUMMARY

The evidence relating to heterosis suggests that the phenomenon is to be
explained genetically in terms of various recombination effects. In some cases,
dominance is the important consideration, while in other cases, hetero-
zygosity must be considered. In any event, it is the resulting specific gene
action which lies at the basis of the physiological advantage or advantages
which give rise to hybrid vigor. One or many genes may be involved. Con-
siderations of genetic balance and genotype-environment balance are im-
portant. Probably most cases of heterosis are to be explained physiologically
in terms of differences in the more fundamental aspects of the metabolic pat-
tern, particularly those concerned with enzyme, auxin, and other growth
substance activity in plants and with enzyme and hormonal activities in
animals.

To clarify the mechanism further, studies must be concerned primarily
with the genetics and physiology of early development. We have been con-
cerned with mature characteristics of size and yield, with the inheritance of
so-called quantitative genes, and with analyses by the classic methods of
genetics. These studies have brought us close enough to an understanding
of the phenomenon of heterosis to indicate that its further analysis by
techniques now at hand will uncover facts of tremendous importance for
genetics, physiology, and other studies of development, some of them con-
siderably afield from heterosis itself.

WILLIAM J. ROBBINS

Columbia University and New York Bofanical Garden

Chapter 7

Hybrid Nutrifionol
Requirements

Hybrid vigor has been recognized for more than a century. It has been con-
sidered from a genetic, morphological, developmental, physiological, and
commercial standpoint. Although a great deal of information has been ac-
cumulated about the phenomenon, we are still unable to define exactly why
a hybrid grows better than the parents from which it comes.

It is obvious that the cause is physiological — the hybrid functions more
effectively or for a longer period of time, and accumulates a greater mass of
cell substance. Its metabolic efficiency is greater (East, 1936). It would be
illuminating if we could locate specifically the physiological processes which
are responsible for the greater vigor of the hybrid — recognizing that they may
be numerous and complex rather than single and simple, and that they may
not be the same for all examples of hybrid vigor.

For many years I have been interested in the factors which determine why
one plant species, variety, or strain grows slowly in a given environment
where another flourishes. I have dealt mainly with microorganisms, especial-
ly the filamentous fungi, because the external env ironment can be more easily
controlled and photosynthesis is not a complicating factor. From my ex-
perience, as well as from the work of others, it is clear that in many instances
growth — the accumulation of cell substance — is limited by the efficiency of
the organism's metabolic machinery, especially the activity of one or more
enzyme systems. Whether this concept can be applied also to the phenome-
non of hybrid vigor is still to be determined. However, it is a hypothesis
which deserves exploration.

Let us begin with a simple example of growth-limitation. Aspergillus niger
grows well in a liquid medium of sugar, mineral salts, and asparagine. In the
same medium Phycomyces Blakesleeanus will not grow at all.

114

HYBRID NUTRITIONAL REQUIREMENTS 115

Does Phycomyces fail to grow in the basal solution because of the absence
of something essential which it needs for growth, or because of the presence
of something detrimental? Does Aspergillus niger grow in the basal solution
because it does not need to be furnished with the "essential" substance, or
because it is more resistant to the supposed injurious ingredient?

For the example cited, we have a definite and well demonstrated explana-
tion. Phycomyces fails to grow in the basal medium because it requires the
vitamin, thiamine — which it is unable to make from sugar, mineral salts, and
asparagine. Aspergillus niger also needs thiamine, but it constructs the vita-
min from the elementary materials present in the basal solution. In this in-
stance, therefore, the failure to grow is due to the lack of something es-
sential for growth; namely, thiamine, the precursor of co-carboxylase.

This is not an isolated example. Many species of fungi grow slowly, or not
at all, in a basal medium because of their inability to make one or more of the
essential metabolites. These metabolites may include various vitamins,
purine and pyrimidine bases, amino acids, fatty acids, or substances as yet
unidentified.

ESSENTIAL METABOLITES-RELATION TO GROWTH

It may be assumed that the complex chemical compounds which make
up the cell substance of a living organism are constructed by the organism
from simpler compounds. A series of intermediate chemical compounds are
formed between the original simple foods and nutrients and the final product,
cell substance. This step-wise progression from simple to complex is made
possible by a series of enzymes, also made by the organism, which operate on
each stage as that stage is completed. Although synthesis is likely to be
emphasized in considering growth, there are other subsidiary processes —
necessary concomitants for the building up of new cell substance. The cata-
bolic processes of digestion and respiration also occur in steps, and are made
possible by the action of a series of enzyme systems.

Any substance playing a necessary part directly or indirectly in the chain
of reactions which end in the synthesis of new cell substance is an essential
metabolite. Unless each essential metabolite, each chemical substance in the
step-wise process of growth, each enzyme which facilitates the chemical re-
actions concerned, is made within the organism or supplied from without, the
series is interrupted. New cell substance is not made, and growth does not
occur. If not enough of an essential metabolite is made, growth will be
slowed.

Of course, this is an oversimplified statement of a very complicated
process. The reactions concerned in growth probably do not occur in a
straight line. Some steps may be bypassed and side reactions may occur, all
of which may affect the speed and character of the growth which results.

It would be difficult to estimate the number of essential metabolites in-

116

WILLIAM J. ROBBINS

volved in the growth of even the simplest organism, or to put a limit on the
number for which some organism may not eventually be found to exhibit a
deficiency.

Some species or strains exhibit a complete deficiency for one or more
essential metabolites. They are unable to synthesize any of the substances in
question and do not grow unless the substances are supplied in the medium
in which they are cultivated (Robbins and Ma, 1942). Others suffer from
partial deficiencies, that is, they grow slowly in the absence of a particular

Fig. 7.1 — Growth affected by complete and partial deficiencies for essential metabolites.
Fungi grown on mineral-dextrose medium containing asparagine and purified agar and
supplemented as follows: (1) no addition; (2) thiamine; (3) pyridoxine; (4) biotin; (5) thia-
mine and pyridoxine; (6) thiamine and biotin; (7) pyridoxine and biotin; (8) all three vita-
mins. .\bove, Ceratosiomella midtiannulata, complete deficiency for pyridoxine, partial for
thiamine; below, C. microspora, complete deficiency for thiamine, biotin, and pyridoxine.

essential metabolite but more rapidly if it is added to the medium (Fig. 7.1).

For example, the clone of excised tomato roots, with which we have
worked for many years, suffers from a complete deficiency of thiamine and a
partial deficiency of pyridoxine. It will not grow unless the medium contains
thiamine or its equivalent. When pyridoxine is added to a medium contain-
ing thiamine, the growth of the excised roots is markedly increased.

In a sugar, mineral-salt solution, the growth of our clone of excised tomato
ri)ots is limited by its ability to synthesize thiamine. In a thiamine solution,
growth is limited by the ability of the roots to synthesize pyridoxine (Robbins,
1946). We have not been able to define what limits the growth of the root
in a solution which contains both thiamine and pyridoxine. Other examples

HYBRID NUTRITIONAL REQUIREMENTS 117

of partial deficiencies could be cited. Their effect is to decrease the rate of
growth but not to inhibit it entirely.

As a result of investigations which have extended over the past decade or
two, we know of many examples in which poor growth or failure to grow in a
specific environment is due to the inability of the organism to synthesize
adequate quantities of one or more essential metabolites. The metabolic
machinery lacks a part, or some part works slowly, with the result that the
organism does not make sufficient quantities of one or more growth essen-
tials, and unless supplied with the missing materials from without, grows
slowly, or not at all.

Not all instances of failure to grow or of poor growth in a given environ-
ment are explainable on the basis of deficiencies of essential metabolites. In
some instances growth may be limited by autogenic growth inhibitors.

AUTOGENIC INHIBITORS

Zalokar (1948), Emerson (1947, 1948), and others have described a mutant
strain of Neurospora which grows poorly at high temperatures. Growth oc-
curs if sulfonamide is added to the medium. One might conclude that
sulfonamide acts for this organism as an essential metabolite. It appears,
however, that this mutant produces growth inhibitors which are antagonized
in some way by the sulfonamide. This seems to be an example of poor growth
caused by the accumulation of autogenic growth inhibitors, and not because
of the lack of an essential metabolite.

Information on the role of autogenic inhibitors in limiting growth is less
specific and more difficult to obtain than evidence for the limitation of growth
due to a deficiency of an essential metabolite. How commonly do internally
produced inhibitors reduce growth? What is the nature of these substances?

From the investigation of antibiotic substances we know that many organ-
isms form metabolic products, highly inhibitory for organisms other than
themselves. Do they also produce substances which limit their own growth?
The role of autogenic inhibitors in limiting growth deserves much more
attention than it has received.

It is well known that minute amounts of specific chemical compounds
materially modify the amount and nature of growth in plants. Zimmerman
and Hitchcock (1949) treated Kalanchoe plants with small amounts of the
ortho, para, and meta forms of chlorophenoxyacetic acid. The para form
caused the apical meristem to develop into a spathe-like organ which could
be cut off and rooted. It had little resemblance to Kalanchoe. The ortho and
meta forms of this compound did not have this effect. This modification was
not a mutation. The effect wore off as the chemical in the plant disappeared,
and the Kalanchoe eventually returned to its normal growth pattern. If the
change had been permanent, we would have been inclined to call it a muta-
tion and look for a genie explanation ; i.e., look for a gene which controlled the

118 WILLIAM J. ROBBINS

production of para-chlorophenoxyacetic acid. We might say that this com-
pound and the Kalanchoe plant acted temporarily as linked genes.

Many other kinds of abnormal growth in plants are probably the result of
the effect of minute amounts of specific chemical compounds. Insect galls
are characterized by an abnormal but specific growth pattern superimposed
on normal tissue by the presence of a foreign living organism. It seems very
likely from the observations of Boysen Jensen that the abnormal growth of
insect galls is caused by specific chemical compounds produced by the larvae
which inhabit the galls.

It must be emphasized that growth is an extremely complex process, not
just a series of chemical reactions. To consider it as such is admittedly an
oversimplification giving no thought to the organization in which these re-
actions occur, or to the structural elements, physical processes, and chemical
reactions which must play a role.

The concept of growth as a series of catalyzed reactions is useful and
stimulating, however, in considering the role of essential metabolites —
especially enzymes — and the action of inhibitors and minute amounts of
specific chemical compounds.

HYBRID VIGOR

Some years ago I attempted to determine whether hybrid corn contains a
greater quantity of substances which stimulate the early growth of Phyco-
myces Blakesleeanus than the inbred parents. The effect of extracts of air
dry grains and of partially germinated grains of the hybrid corn and its in-
bred parents was determined on the growth of Phycomyces in the presence of
thiamine (Robbins, 1940, 1941a).

When compared on the basis of extract per grain, I found that the extracts
of the grains of the hybrid corn gave a greater dry weight of mycelium of
Phycomyces than those of either of the inbred parents (Fig. 7.2). The stimu-
lating material seemed to be present in both the embryo and the endosperm.
Since the solution in which the beneficial effects of the extracts were exhibited
contained sugar, asparagine, mineral salts, and thiamine, it appeared that
the effect was produced by unidentified growth substances. These were
termed for convenience, factor Z.

After estimating the amount of factor Z present — from the effects of the
extracts of the corn grains on the early growth of Phycomyces in the presence
of thiamine — the following generalities seemed permissible. The amount
of factor Z increased with the time of the germination of the corn grains, at
least up to seventy-two hours' germination. The quantity of Z was greater
per endosperm than per embryo, and was greater in the grains of the hybrid
than in those of either parent. The amount of thiamine and its intermediates
in the embryo and endosperm of the grains of the hybrid and its parents
was not correlated with the amount of factor Z, nor did the amount of biotin
in the extracts appear to be correlated with the amount of factor Z.

HYBRID NUTRITIONAL REQUIREMENTS

119

These results suggest that there is present in the grains of corn, material
which stimulates the early growth of Phycomyces in the presence of thiamine,
and that there is more of this material per grain in heterotic hybrids than in
those of the inbred parents.

Interpretation of these results depends in part on the identity of factor Z.

100

90

80

O

D
_1

u

O

>-

H

>-
tr
a

70

60

50

40

30

20

10

025

0.5
ML. CORN GRAIN EXTRACT

Fig. 7.2 — Increase in dry weight of Phycomyces produced by extracts of air dry grains of

maize. Extracts added to medium of sugar, minerals, asparagine, and thiamine. A = line

4-8; B = line 187; C = 985, 4-8 X 187; D = 995, 187 X 4-8. 1 ml. extract = 1 grain.

Unfortunately, we do not know what factor Z is. We succeeded in dividing it.
We demonstrated that factor Z is multiple, and separated it into a fraction
adsorbed on charcoal, factor Zi, and a filtrate fraction, factor Z2. Factor Zi
was identified as hypoxanthine. Factor Z2 may be a mixture of amino acids.
Although this problem is left in an uncertain and unsatisfactory condi-
tion, it suggests a line of attack. This would be an investigation of heterosis
by studying the efifect of extracts of parents and of heterotic hybrids on the
growth of other organisms. This may serve as a means of bioassay for favor-
able or unfavorable growth factors.

120

WILLIAM J. ROBBINS

Vigor in Heterocaryons

Observations of Dodge (1942) on heterocaryosis in Neurospora are of
interest to the general problem of heterosis. Dodge inoculated three petri
dishes, one with his Dwarf 16 strain of Xeurospora telrasperma, one with race
C-8, and the third with mixed mycelium or conidia of both the dwarf and the
C-8 races. He observed that the mycelium of the mixed culture grew much
more rapidly and produced more abundant conidia than the mycelium of
either the dwarf or the C-8 races (Fig. 7.3).

Fig. 7.3 — Heterocaryotic vigor in Neurospora telrasperma. Growth in 34 hours at room

temperature in petri dishes. The myceUum of the two heterocaryotic races {16 -\- C 4 and

16 + C8) has nearly covered the medium in the dishes; C4 and C8 have not grown halfway

across the medium and Dwarf 16 has made no visible growth.

When two races of Xeurospora telrasperma are grown together, there is a
migration of nuclei through the openings at the points of hyphal anas-
tomoses. The races need not be of opposite sex. After nuclear migration, the
cells of the resulting mycelium are heterocaryotic. They contain two kinds of
haploid nuclei. The greater vigor of the mixed culture referred to above ap-
pears to be the result of the presence in a common cytoplasm of two kinds of
nuclei.

Heterocaryotic vigor does not always accompany heterocaryosis. Dodge
(1942) observed heterocaryotic vigor when the two races, Dwarf 16 and C-4,
were grown together. But heterocaryosis for races C-4 and C-8 did not result
in increased vigor in the mixed culture. Not all dwarf races act as race 16
does. Some of them evidence heterocaryotic vigor with both C-4 and C-8,

HYBRID NUTRITIONAL REQUIREMENTS 121

others with C-4 but not with C-8, and still others develop none with either
C-4 or C-8.

Dodge has suggested that the heterocaryotic hybrid may synthesize a full
quantity of growth substances or essential metabolites. Whereas the growth
of each of the parents is limited by their inability to synthesize adequate
quantities of one or more essential metabolites.

Dwarf 16, for example, may be able to make adequate quantities of essen-
tial metabolites 1, 2, 3, and 4, but unable to construct enough of 5, 6, 7,
and 8. On the other hand, race C-4 may be unable to synthesize enough of
1, 2, 3, and 4, but be capable of producing an adequate supply of 5, 6, 7,
and 8. When nuclei of the two races are brought together in a common
cytoplasm, the essential metabolites synthesized by one of the nuclear com-
ponents supplement those synthesized by the other component. The hetero-
caryotic mycelium is then supplied with adequate quantities of all the
essential metabolites necessary for rapid growth.

We have tried to test this hypothesis by supplementing with various
substances the medium on which race 16 and other dwarf races were grown.
If it were possible to increase materially the growth rate of the dwarf race by
supplements in the medium, without introducing the heterocaryotic condi-
tion, the limiting factors for dwarfness could be identified and the stimulus
involved in the heterocaryotic condition identified.

A basal agar medium containing mineral salts, dextrose, asparagine, neo-
peptone, and thiamine was supplemented by a mixture of purine and pyrim-
idine bases; by a vitamin mixture containing PAB, calcium pantothenate,
inositol, nicotinic acid, pyridoxine, riboflavin, thiamine, guanine, hypoxan-
thine, and 2-methyl-l, 4-naphthohydroquinone diacetate; by malt extract,
casein hydrolysate, cow's milk, dried yeast, choline, a-tocopherol, hemin,
oleic acid, ascorbic acid (filtered sterile), coconut milk, Taka-diastase
(filtered sterile), water extracts of the mycelium of Neurospora, liver ex-
tracts (both filtered sterile and heated), adrenal cortical extract (unheated),
estrogenic substance, progesterone, anterior pituitary extract, posterior
pituitary extract, whey, or potato extract.

None of the substances or combinations of them as used increased the
growth rates of any of the dwarf races to an extent adequate to explain
heterocaryotic vigor. Some beneficial effects, usually noted only in older cul-
tures, were obtained from cow's milk and from liver extract. These efifects
were not sufficiently marked to suggest that either supplement supplied the
missing factors.

We were unsuccessful, therefore, in defining the factors limiting the
growth of the dwarf races and conversely those effective in inducing more
rapid growth in the heterocaryotic mycelium.

Our failure may be explained in various ways. We may not have included
in our various supplements the missing essential metabolites. These metabo-

HYBRID NUTRITIONAL REQUIREMENTS 123

lites may be non-diffusible or very labile substances such as enzyme pro-
teins, which could only be introduced into the cell through inserting a nucleus
and its genes. The original hypothesis may be in error. We may not be
dealing with limiting quantities of essential metabolites but with inhibitors.
We might assume that the growth of one or both of the parents is limited by
autogenic inhibitors, and the presence of both kinds of nuclei in a common
cytoplasm results in the neutralization in some fashion of the inhibitors.

Emerson (1948) has succeeded in producing heterocaryons in which one
kind of haploid nucleus neutralizes the effect of the other. The augmented
growth of the heterocaryon, as compared to that of strains which are
homozygous, reminds one, says Emerson, of instances of single gene heterosis
in maize reported by Jones.

The importance of internal factors in heterosis is suggested by the results
I obtained on the growth of the excised roots of a heterotic tomato hybrid
and its inbred parents (Robbins, 1941b). The hybrid roots and the roots of
the two inbred parents were grown in liquid culture which contained mineral
salts and cane sugar. This basal medium was supplemented with thiamine,
with thiamine and pyridoxine, and with thiamine, pyridoxine, and nicotina-
mide.

Growth of the roots of the hybrid exceeded that of either of the inbred
parents in all three types of media (Fig. 7.4). Growth of one parent was im-
proved by the addition of pyridoxine to the thiamine solution, but a further
supplement of the medium with nicotinamide had little effect. Growth of the
second inbred parent was little affected by the addition of pyridoxine to the
thiamine medium, but was improved by the further addition of nicotinamide
to the thiamine and pyridoxine solution.

These results suggest that the greater vigor of growth of the heterotic
hybrid is determined in part by its greater ability to synthesize pyridoxine
and nicotinamide. That is evidently not the whole story, because its growth
exceeded that of the inbred parents in media containing all three vitamins.

Although heterosis may be considered and should be considered from the
genetical standpoint, it should also be studied from the physiological stand-
point. I have suggested that it may be important to devote attention to the
question of v/hat the internal factors are which limit growth, what they are
in inbreds, and how they are removed in heterotic hybrids. We should con-
sider in such investigations the role of essential metabolites, of growth in-
hibitors, and of other specific chemical compounds which materially modify
growth. Microorganisms might be utilized as tools for the detection of growth
stimulators or growth inhibitors.

EDGAR ANDERSON

Missouri Bofanical Garden

and

WILLIAM L. BROWN

Pioneer Hybrid Corn Company

Chapter 8

Origin of Corn Belt Moize and Its

Genetic Significance

Several ends were in view when a general survey of the races and varieties of
Zea mays was initiated somewhat over a decade ago (Anderson and Cutler,
1942). Maize, along with Drosophila, had been one of the chief tools of mod-
ern genetics. If one were to use the results of maize genetics most efficiently
in building up general evolutionary theories, he needed to understand what
was general and what was peculiar in the make-up of Zea mays. Secondly,
since maize is one of the world's oldest and most important crops, it seemed
that a detailed understanding of Zea mays throughout its entire range might
be useful in interpreting the histories of the peoples who have and are using
it. Finally, since maize is one of our greatest national resources, a survey of its
kinds might well produce results of economic importance, either directly or
indirectly.

Early in the survey it became apparent that one of the most significant
sub-problems was the origin and relationships of the common yellow dent
corns of the United States Corn Belt. Nothing exactly like them was known
elsewhere in the world. Their history, though embracing scarcely more than
a century, was imperfectly recorded and exasperatingly scattered. For some
time it seemed as if we might be able to treat the problem only inferentially,
from data derived from the inbred descendants of these same golden dent
corns. Finally, however, we have been able to put together an encouragingly
complete history of this important group of maize varieties, and to confirm
our historical research with genetical and cytological evidence.

An even approximate survey of Zea way.s-as-a-whole remains a goal for

124

ORIGIN AND SIGNIFICANCE OF CORN BELT MAIZE 125

the distant future, but our understanding of Corn Belt dent corns is already
more complete than we had originally hoped. Since our evidence is detailed
and of various kinds, it may make the presentation somewhat easier to follow
if we give a brief description of the pre-hybrid commercial yellow dents of
the United States Corn Belt, review their history in broad outline, and then
proceed to an examination of the various kinds of evidence on which these
generalizations have been built.

Corn Belt dents, the commercial varieties which dominated the chief
centers of corn production in the United States for over half a century pre-
ceding the advent of hybrid corn, were variable open-pollinated varieties.
They varied from plant to plant, from field to field of the same variety, and
from variety to variety. Figure 8.1, based upon an examination of a field of
Golden Queen, one of the lesser known of these varieties, will indicate the
kind of variation which characterized the fields of that day.

In spite of this variation, or one might almost say, impressed on top of it,
was a remarkably persistent combination of generally prevalent characters.
Considered from plant to plant or from field to field, as individuals, these
varieties seemed ephemeral and unimportant. Seen as populations, as col-
lections of inter-breeding individuals, the Corn Belt dents as a whole were a
well-marked and definite entity, particularly when contrasted with maize
in other parts of the world. They tended to have one well-developed ear, fre-
quently accompanied by a small ear at the node below this primary one.
The ears had large, nearly cylindrical cobs with red or reddish glumes. The
usually golden yellow kernels, pronouncedly dented at the tip, had a peri-
carp frequently roughened by tiny wrinkles. They were set in from 14 to 22
straight rows with little external indication of the fact that the rows were in
pairs. The mathematical perfection of the ear was frequently lessened by a
slight tendency for the whole ear to taper toward the apex, and for the row-
ing of the kernels and the diameter of the cob to be somewhat differentiated
in its lowermost quarter.

Characteristically, the plant on which this ear was borne had a single, up-
right stem, leaves with tight sheaths and strong, arching blades, and a
heavy, many -branched tassel. Kernel color was remarkably standardized,
a faint flush of coppery red in the pericarp and a yellow endosperm, combin-
ing to give varying shades of deep, golden color. Epidermal color was ap-
parent on the culm and leaves at the base of the plant, but seldom or never
were there to be found the brilliant reds, dark purples, and other foliage
colors which are so characteristic of maize in various parts of Latin America.
While there was some variation in anther color and silk color, pinks and dull
reds were commonest though greens and bright reds were not unknown.

As we have shown elsewhere (Anderson and Brown, 1950) there cannot
be the slightest doubt that these widespread and standardized Corn Belt
varieties were the creation of the nineteenth century. They came in large part

20

18

OS.
Ul

C9

16

%

^

^•V i. *-

14

12

8-9

10-

■11 12-

13

14-15

KERNEL WIDTH

TASSEL BRANCH
NUMBER

GLUME LENGTH
IN MM.

PITH WIDTH
IN MM.

EAR LENGTH
IN CM.

• 11-20

• 8-9

•4-9

•26-30

• 21-23

4io

4io-ii

• 22-25

#^OVER 24

• 11-13

• 12-15

• 14-21

Fig. 8.1 — Pictorialized diagram showing relationship between numbers of rows of kernels,

kernel width, tassel branch number, glume length, pith diameter, and ear length in an

open-pollinated sample of Golden Queen dent corn.

ORIGIN AND SIGNIFICANCE OF CORN BELT MAIZE 127

from crosses between White Southern Dents, mostly of Mexican origin, and
the long, slender Northern Flints which had dominated the eastern United
States for at least some hundreds of years preceding the discovery of
America. While these two complexes were of primary importance in the crea-
tion of Corn Belt corn, it should be pointed out that germ plasm of other
types of maize has undoubtedly filtered into Corn Belt mixtures. Compared
to Southern Dents and Northern Flints, these certainly are of minor im-
portance. There are, nevertheless, to be found among dent inbreds of the
Corn Belt certain strains which exhibit Caribbean influence and others
which seem to contain germ plasm of southwestern United States or western
Mexican varieties.

Although the following discussion does not go into detail regarding the
influence of these secondary sources of germ plasm on Corn Belt corn, the
effects of such influences are important and we have already made small
beginnings at studying them. The Northern Flints are in some ways strik-
ingly similar to the common yellow flints of the Guatemalan highlands, strik-
ingly unlike most Mexican maize. They are one of several cultural traits
which apparently spread from the Mayan area to the eastern United States
without leaving any clear record of the route by which they came. In their
general appearance, as well as in technical botanical details, the Northern
Flints were very different from the Southern Dents. The hybrid vigor which
resulted from mixing these diverse types was soon noted by alert agricultur-
ists. While some of the blending of flints and dents may have been haphazard
and accidental, much of it was directed and purposeful. The benefits to be
gained were listed in public, and the exact effects of continued mixing and of
backcrossing were discussed in detail as early as 1825 (Lorain, 1825). This
intelligent, controlled hybridizing proceeded for at least a half century until
the new yellow dents were so ubiquitous and everyday that their very origin
was forgotten.

For theoretical reasons this neglect of historical tradition was unfortunate.
Maize breeders have not understood that the heterosis they now capitalize
is largely the dispersed heterosis of the open-pollinated flint-dent mongrels.
Maize geneticists are for the most part unaware that the germ plasm they
use for fundamental generalizations is grossly atypical of germ plasms in
general. We shall return to a detailed discussion of these two points after
referring briefly to the evidence concerning the origin of Corn Belt maize.

Though there is abundant evidence that our Corn Belt dents came from
mixtures of Northern Flints and Southern White Dents, the evidence con-
cerning these two regional types is very one-sided. The Northern Flints
(Brown and Anderson, 1947) were remarkably uniform from place to place
and from century to century. The archaeological record is rich going back to
early pre-Columbian times and there are numerous naive but accurate de-
scriptions of these varieties in colonial accounts.

128 EDGAR ANDERSON AND WILLIAM L. BROWN

The Southern Dents (Brown and Anderson, 1948) are much more vari-
able. For over a century their variability has been stressed by all those who
have discussed them. The samples which we obtained from the South differed
from field to field, and from variety to variety. For an accurate understanding
of them and their history, we would like many more archaeological specimens
than we have for the flints, and many more colonial descriptions. Instead, we
have as yet no archaeological record, merely two accounts in early colonial
times — one from Louisiana and the other from Virginia. There is one passing
mention in a pre-revolutionary diary, and then a truly remarkable discussion
by Lorain in 1825. Finally, the United States Patent Office report for 1850
gives us, for region after region, a detailed picture of the extent to which this
purposeful mixing had proceeded by that time.

To summarize the historical evidence, the Northern Flints were once the
prevailing type of maize throughout the eastern United States (Brown and
Anderson, 1947) with an archaeological record going back at least to a.d.
1000. There is as yet no archaeological evidence for their having been pre-
ceded in most of that area by any other type of maize, or of Mexican-like
dents having been used there in pre-Columbian times. The Northern Flints
belong to a type of maize rare or unknown over most of Mexico, but common
in the highlands of Guatemala. The Southern Dents, on the contrary, obvi-
ously are largely derived from Mexican sources, and by 1700 were being
grown as far north as Louisiana and Virginia (Brown and Anderson, 1948).
As to how and when they spread north from Mexico, we have no evidence
other than the negative fact that they are not known archaeologically from
the eastern United States, and are not represented in the collections of early
Indian varieties from that region.

As early as 1800, the benefits of crossbreeding these two different types of
maize were appreciated by at least a few experts. By 1850 the process was
actively under way from Pennsylvania to Iowa, and south to the Gulf states.
By the '70's and '80's, a new type of corn had emerged from this blending,
although crossing and re-crossing of various strains continued up to the ad-
vent of hybrid corn. During the latter half of the process, the origin of Corn
Belt dents from 50 to 100 generations of selective breeding of crosses of
Northern Flints and Southern Dents was almost completely forgotten. Hav-
ing at length resurrected the evidence (Anderson and Brown, 1950) for this
mingling of two fundamentally different types of maize, we shall now turn
to the genetical and cytological evidence which first called the phenomenon
to our attention and led us to search for historical proof.

CYTOLOGY

The most important cytological contribution on the origin of Corn Belt
maize is found in a comparison of the numbers and distribution of chromo-
some knobs in the Northeastern Flints, open-pollinated varieties of Southern

ORIGIN AND SIGNIFICANCE OF CORN BELT MAIZE 129

Dents, and inbred strains of Corn Belt dents. As has been shown previously
(Longley, 1938) and (Reeves, 1944), chromosome knobs may be an im-
portant tool in studying relationshii)S in maize. Our work with North Ameri-
can corn not only supports this contention, but suggests that knob data may
be even more important than has previously been supposed.

The 8-10 rowed flint and flour varieties of New York, Pennsylvania, and
New England are nearly knobless. In the material we have examined, they
have 0 to 2 knobs. These observations are in agreement with Longley's
earlier conclusions that maize varieties of the northern Indians were char-
acterized by having few knobs. Longley's material, however, included no
strains from northeastern United States — the area in which the flint an-
cestors of Corn Belt corn were highly concentrated. It is interesting, more-
over, to note that varieties from this segment of North America have even
fewer knobs than do the strains from most Northern Plains Indian tribes.

In contrast, many more knobs were to be found in the open pollinated
varieties of Southern Dent corn. In these strains we have found numbers
ranging from 5 to 12, for those varieties representing the least contaminated
segment of present-day Southern Dent corn. These cytological data are in
complete agreement with the known facts regarding the history of Northern
Flints and Southern Dents.

There seems little doubt that the Gourdseed-like Dents' of the southeast-
ern United States have stemmed directly from Mexico wiiere morphological-
ly and cytologically similar corns can be found even today. Likewise, we
have found in highland Guatemala varieties of maize with ear character-
istics strikingly similar to Northern Flints and with as few as three knobs.
Insofar as cytology is concerned, therefore, it is not at all difficult to visualize
a Guatemalan origin for Northeastern Flint corn. The Corn Belt inbreds
with which we have worked (Brown, 1949) have knob numbers of 1 to 8.
The distribution of numbers in these strains is almost exactly intermediate
between that of Northern Flints and Southern Dents (Fig. 8.2). This evi-
dence, based on a character which certainly has not been intentionally
altered by selection, strongly fortifies the archaeological and historical facts
pointing to a hybrid origin of Corn Belt dent corns.

GENETIC EVIDENCE

The genetical evidence for the origin of Corn Belt maize from mixtures
of Northern Flints and Southern Dents is of various kinds. In its totality, it
is so strong that, had we not been able to find the actual historical evidence,
we could have determined what had happened from genetic data alone. In
the first place we have demonstrated, by repeating the cross, that it is pos-
sible to synthesize Corn Belt dents from hybrids between Southern Dents

I The name "Gourdseed" has been used since colonial times to describe the extremely-
long seeded, white Southern Dents, whose kernels are indeed not so diflferent in appear-
ance from the seeds of gourds of the genus Lagenaria.

130

EDGAR ANDERSON AND WILLIAM L. BROWN

and Northern Flints. Our experiments in crossing a typical white gourdseed
from Texas and a typical yellow flint from New York State are now only in
the third generation and are being continued. However, it is already evident
that some of the segregates from this cross are within the range of varia-
tion of Corn Belt dents (Fig. 8.3).

In spite of the 50 to 100 generations of mixing which has taken place, the
characters of Northern Flints and Southern Dents still tend to be associated
in Corn Belt dents. Anderson (1939) has shown that in crosses between species

10

OLD SOUTHERN OENTS

10

>
u

Z 0
lit

3

a

^20
li.

10

DERIVED SOUTHERN OENTS

CORN BELT INBREDS

10

1

NOR

1

THERN ruiNTS

0

1 ' 2

' 3

4 '

5 '

6

7

8 •■

9 '

10 '

II

' 12

"I

NUMBER OF CHROMOSOME KNOBS

Fig. 8.2 — Frequency distribution of chromosome knobs in Northern Flints, Southern

Dents, and Corn Belt inbreds.

or between races, all the multiple factor characters which characterize each
are partially linked with one another and tend to remain associated, even
after generations of controlled breeding. More recently he has used this
principle in the development of the method of extrapolated correlates (Ander-
son, 1949) by which the original characteristics can be deduced from the mix-
tures even when previously unknown.

Using this method in a relatively crude form, we were able (in advance
of our historical evidence) to demonstrate (Brown, 1949) in Corn Belt in-
breds, the association of low knob numbers, flag leaves, cylindrical ears, few
tassel branches, and flinty kernels — all characteristics which typify the
Northern Flints. Similarly, it was possible to show the association among
these 98 Corn Belt inbreds of high knob numbers, no flag leaves, tapering
ears, dented kernels, and many tassel branches — a combination of char-

ORIGIN AND SIGNIFICANCE OF CORN BELT MAIZE 131

acters which is typical of the Southern Dents. As a matter of fact, by this
technique Brown predicted the knob numbers of the Northern Flints, even
when that fact was unknown to us.

The association of characters in actual open-pollinated fields of Corn Belt
dents is so complex that one might suppose any study of it would be hopeless.
However, from a study of character association in an open-pollinated field

Fig. 8.3 — Corn Belt Denl-like segregates from an Fo generation of cross of Longfellow

Flint X Gourdseed Dent.

of Golden Queen Dent (Fig. 8.1) we were able to demonstrate the association
of: (1) wide kernels, (2) low row numbers, (3) short glumes, (4) few tassel
branches, (5) long ears, and (6) narrow central pith in the ear — all of these
characterizing Northern Flints. The opposing combination: (1) narrow
kernels, (2) high row numbers, (3) long glumes, (4) many tassel branches,
(5) short ears, and (6) wide central pith also tended to be associated and is
characteristic of Southern Dents. In other words, some of the characters
which went in together from flints and dents were still in this open-pollinated
variety tending to stay together on the average. The existence of such char-
acter complexes has been appreciated by experienced corn breeders, though
apparently it has never been commented on in print. Of course, corn breed-
ers and corn geneticists differ in their endowments for apprehending such

132 EDGAR ANDERSON AND WILLIAM L. BROWN

phenomena in advance of the published facts, and the existence of these
strong linkages has been more apparent to some than to others.

WIDTH OF CROSS

The demonstration that Corn Belt dents largely are derived from hy-
bridization between Southern Dents and Northern Flints is of particular im-
portance because this is such a wide cross. Our evidence for this assertion is
largely morphological, though there is supporting evidence from cytology
and genetics.

In nearly all species of cultivated plants there are conspicuous differences
in color and shape. These differences give the various cultivated varieties of
a species a false aspect of difference from one another, and from their wild
progenitors. False, because these differences are usually due to a few genes, if
not being actually monofactorial. The striking differences between such
varieties are therefore no true indication of the distinctness of their germ
plasms.

On the other hand, there are subtle differences in form, proportion, and
indument which, though difficult for a novice to apprehend, are more like
the differences which distinguish distinct species of the same genus. These
taxonomically important differences have proven valid criteria for indicating
the diversity of germ plasms. So it has been proven that the subtle taxonomic
differences between the Old World and New World cottons are much more
representative of the genetic diversity and relationships of these two groups
of varieties than are the conspicuous differences in color and leaf-shape which
are found within each group. In the Cucurbits the striking differences in
color and form of fruit, which differentiate the varieties of Cucurbita Pepo
and of C. moschata, are superficial compared to the taxonomically significant
features which separate these two groups. The latter, moreover, have been
proved to be a significant index of genetic diversity, either between these
two groups of Cucurbits or in assaying the variation within C. Pepo itself
(Shifriss, 1947) (Whitaker and Bohn, 1950).

The difficulty in relying upon such taxonomic criteria is that the method
is highly subjective. Taxonomy is of necessity still more of an art than a
science. This means that one must personally examine the evidence if his
opinion is to be worth anything. It also means that the worker's opinion is
worth no more than his understanding of the taxonomic entities included in
his judgment. However, until more objective criteria are evolved for this
field, we shall have to use fairly traditional taxonomic methods for want of
anything better. Accordingly, the senior author has for two years spent one
day a week in a technical, agrostological, herbarium survey of all the grasses
conceivably related to Zea mays — all the genera in the tribes Andropogoneae
and Maydeae. With that background, his judgments may well be mistaken
but they are certainly informed.

From this point of view, the variation within Zea mays is without parallel,

ORIGIN AND SIGNIFICANCE OF CORN BELT MAIZE 133

not only in the cultivated cereals but in any other domesticated plant or
animal. There are such superficial characters as aleurone color, pericarp
color, plant color, carbohydrate composition, and such amazing single factor
differences as tunicate and teopod. In addition, there are a whole battery of
characters which are difficult to work with genetically, but which are the
kinds of differences that agrostologists find significant in the deployment of
species and genera: spikelet shape and venation, spikelet arrangement,
rachis morphology, pubescence, leaf-shape, internode proportions, etc. Using
such criteria, the hybridization of the Southern Dents and the Northern
Flints represents the mingling of two basically different germ plasms.

For evidences of relationship, the male inflorescence of maize (the tassel)
is of particular importance. Inflorescence differences generally have proved
to be of primary taxonomic importance in the Gramineae. Variation in the
male inflorescence of Zea would likely be less obscured by domestication than
the female inflorescence (the ear) which has been deliberately selected for
various peculiarities. The entire male inflorescence of the Southern Dents
has been extensively modified by condensation (Anderson, 1944), a sort of
fasciation which telescopes adjacent nodes, and in the ear produces increases
in row number. It is an abnormality conditioned by at least two pairs of
recessive genes and its expression is certainly modified by still other genes.

Tassels of the Northern Flints are without any condensation. Though
condensation modifies the general aspect of the tassel, it is relatively super-
ficial. The presence of so much condensation renders difficult the demonstra-
tion of a much more fundamental difference. The central spike of the North-
ern Flints is decussately arranged. That is, the pairs of spikelets are in alter-
nate whorls of two; whereas the spike of the Southern Dents (allowing for
the modifications produced by extreme condensation) is fundamentally in
whorls of 3, or mixtures of whorls of 3 and whorls of 2. The rachis of the
Northern Flints is slender with long internodes, that of the Southern Dents
is short and flattened (Fig. 8.5). Pedicels of the upper spikelets always are
long in the Northern Flints. In the Southern Dents they may be so short that
one cannot distinguish the normally pedicellate spikelet from its sessile
partner.

Correlated differences are seen in the ear. That of the Northern Flints has
a narrow central pith and is long and slender, characteristically with 8-10
rows. The ear of the Southern Dents is short and thick with a wide central
pith, and with from 16 to 30 or more rows. Pairing of the rows is markedly
evident in the Northern Flints, even when they are pushed closer together
in those occasional ears with 10 or 12 rows (Fig. 8.4). There is little or no row
pairing in the Southern Dents. The kernel of the Southern Dents is long, flat,
and narrow. Its largest diameter is near the base. By contrast, the kernel of
the Northern Flints is wider than it is high, and is considerably thicker
at the apex than it is at the base.

The ear of Zea mays is terminal on a secondary branch, which is hidden by

.^ r^

-n

-.», -iii*; iafe.

S-2

M.

S-l

N-3

N-2

N-4

t- '■

S-4-

*•-.

S-3

Fig. 8.4 — Typical ears (/), shanks (2), and seeds {3 and 4) of Northern Flint (iV), and

Southern Dent (5).

Fig. 8.5 — Typical plants, tassels, and staminate spikelets of Northern Flint and Southern

Dent.

136 EDGAR ANDERSON AND WILLIAM L BROWN

its specialized leaves or husks. When dissected out, these ear shoots (or
shanks) are diagnostically different in Northern Flints and Southern Dents
(Fig. 8.4). In the former they are long, with elongated internodes which are
widest between the nodes, and which have a smooth surface upon drying. In
the latter they are very short, frequently wider at the nodes than between
them, and have a characteristically ribbed surface upon drying.

The leaves of the Northern Flints are long and slender and frequently a
light green. Those of the Southern Dents are proportionately wider and
shorter and are often dark green. They are set upon culms whose internodes
are proportionately longer and more slender in the Northern Flints, and less
prone to become greatly shortened at the internodes immediately above the
ear.

If we ignore such abnormalities as differences in carbohydrate composition
and condensation, these two races of Zea mays still are widely different from
one another — as compared to differences between their wild relatives in the
Andropogoneae or the Maydeae. The differences in internode pattern and
proportion and in leaf shape are similar to those frequently found between
species of the same genus. The differences in pedicellation of the upper spike-
let would be more characteristic of genera and sub-genera. On the other
hand, in the whorling of the central spike (whorls of 2 versus whorls of 3)
is the kind of difference which would ordinarily separate genera or even
groups of genera. On a par with this difference are those in the cupule (the
bony cup in which the kernels are attached in pairs). They are so difficult to
observe that we cannot discuss these until the general morphology of this
organ has been described. If we sum up the morphological evidence, it is clear
that the fundamental differences between the Northern Flints and the
Southern Dents are similar to those which differentiate distantly related
species (or even genera) among related wild grasses. There is every morpho-
logical indication, therefore, that we are dealing with two fundamentally
different germ plasms.

The cytological facts reported above lend further weight to the conclusion
that the Northern Flints are basically different from the Southern Dents.
The former have chromosomes which are essentially knobless at pachytene.
The latter average nearly one knob per chromosome (Fig. 8.2). Heterochro-
matic knobs are known in other grasses besides Zea mays. In these other
genera, their presence or absence, from such evidence as is available, seems
to be characteristic of whole species or groups of species. Such a difference
between the Flints and Dents indicates that we are dealing with two funda-
mentally different germ plasms. It has been shown in Guatemala (Mangels-
dorf and Cameron, 1942) and in Mexico (Anderson, 1946) that the varieties
with many knobs are morphologically and ecologically different from those
with low numbers of knobs.

A further indication that these two germ plasms are physiologically dif-

ORIGIN AND SIGNIFICANCE OF CORN BELT MAIZE 137

ferent is given by their pachytene behavior. The pachytene chromosomes of
the Northern Flints are easy to smear and give sharp fixation images. South-
ern Dents are more ditlficult to smear. The chromosomes do not spread out
well and do not stain sharply. This is not a result of differences in knob num-
ber, since some of the Mexican Dents with few knobs are equally difilicult to
smear. Whatever the physiological significance of this reaction, it is direct
evidence for a difference in the chemistry of the germ cells. Again such dif-
ferences in stainability are more often met with, between genera, than they
are in different strains of the same species.

There is genetic evidence for the difference between Southern Dents and
Northern Flints, in the behavior of crosses between them. The F/s are fully

TABLE 8.1

PERCENTAGE OF STERILE OR BARREN PLANTS IN

GOURDSEED, LONGFELLOW, AND Fa GENERATION

OF CROSS GOURDSEED X LONGFELLOW

Total
Number
of Plants

Gourdseed

Longfellow

Fo Gourdseed X Longfellow

Sterile

Normal

or

Ear

Barren

37

63

2

98

52

48

fertile and exhibit extreme hybrid vigor. The Fo's show a high percentage of
completely barren plants — plants which formed ears but set little or no seeds,
either because of sterility or because they were too weak to mature success-
fully— and plants which managed to set seeds, though their growth habit
indicates fundamental disharmonies of development.

Table 8.1 shows the percentages of good ears and plants which were either
without ears or on which the ears had failed to set any seed, for Gourdseed-
Dent, Longfellow Flint, and their F2, when grown in Iowa. Like Southern
Dents generally, the Gourdseed is less adapted to central Iowa than is Long-
fellow Flint. An F2 between these two varieties, however, has a much greater
percentage than either parent of plants which are so ill-adapted that they
either produce no visible ear, or set no seed if an ear is produced. Similar
results were obtained in other crosses between Northern Flints and Southern
Dents, both in Missouri and in Iowa. From this we conclude that they are
so genetically different from one another that a high percentage of their F2
recombinations are not able to produce seed, even when the plants are care-
fully grown and given individual attention.

SUMMARY
The common dent corns of the United States Corn Belt were created
de novo by American farmers and plant breeders during the nineteenth cen-

138 EDGAR ANDERSON AND WILLIAM L BROWN

tury. They resulted in a large measure from deliberate crossing and re-
crossing of two races of maize (the Northern Flints and the Southern Dents)
so different that, were they wild grasses, they would be considered as totally
different species and might well be placed in different genera. The origin of
two so-different races within cultivated maize is an even larger problem and
one outside the scope of this discussion. It may be pointed out parentheti-
cally that the Tripsacum hypothesis (Mangelsdorf and Reeves, 1945) would
not only account for variation of this magnitude, it would even explain the
actual direction of the difference between these two races of maize. However,
the relation between maize and Tripsacum on any hypothesis is certainly a
most complicated one (Anderson, 1949). It would be more effective to post-
pone detailed discussions of this relationship until the comparative morphol-
ogy of the inflorescences of maize and of Tripsacum is far better understood
than it is at present.

SIGNIFICANCE TO MAIZE BREEDING

Derivation of the commercial field corns of the United States by the de-
liberate mingling of Northern Flints and Southern Dents is a fact. Unfortu-
nately, it is a fact which had passed out of common knowledge before the
present generation of maize breeders was educated. From the point of view
of practical maize breeding, either hybrid or open-pollinated, it is of central
importance. Briefly, it means that the maize germ plasms now being worked
with by plant breeders are not varying at random. They are strongly
centered about two main centers or complexes. Such practical problems as
the development and maintenance of inbreds, the detection of combining abil-
ity, and the most effective utilization of hybrid vigor need to be rethought
from this point of view. Detailed experiments to provide information for such
practical questions already are well under way. While these experiments are
not yet far enough along to give definite answers, they have progressed far
enough to allow us to speak with some authority on these matters.

HETEROSIS

The heterosis of American Corn Belt dents acquires a new significance in
the light of these results, and practical suggestions as to its most efficient
utilization take on a new direction. We are immediately led to the hypothesis
that the heterosis we are working with is, in part at least, the heterosis ac-
quired by mingling the germ plasms of the Northern Flints and the Southern
Dents.

Insofar as hybrid vigor is concerned, the hybrid corn program largely has
served to gather some of the dispersed vigor of the open-pollinated dents.
Preliminary results indicate that this has not been done efliciently in terms
of what might be accomplished with somewhat more orientation.

The early days of the hybrid corn program were dominated by the hy-

ORIGIN AND SIGNIFICANCE OF CORN BELT MAIZE 139

pothesis that one could inbreed this vigorous crop, identify the inferior
strains in it, and then set up an elite cross-pollinated germ plasm. This hy-
pothesis was clearly and definitely stated by East and Jones {Inbreeding and
Outbreeding, 1919, pp. 216-17).

Experiments with maize show that undesirable qualities are brought to light by self-
fertilization which either eliminate themselves or can be rejected by selection. The final re-
sult is a number of distinct types which are constant and uniform and able to persist in-
definitely. They have gone through a process of purification such that only those individu-
als which possess much of the best that was in the variety at the beginning can survive. The
characters which they (pure lines) have, can now be estimated more nearly at their true
worth. By crossing, the best qualities which have been distributed to the several inbred
strains can be gathered together again and a new variety recreated. After the most desirable
combinations are isolated, their recombination into a new and better variety, which could
be maintained by seed propagation, would be a comparatively easy undertaking.

Though other corn breeders and corn geneticists may not have committed
themselves so definitely in print, such a notion was once almost universal
among hybrid corn experts. Modified versions of it still influence breeding
programs and are even incorporated in elementary courses in maize breeding.

The facts reported above would lead us to believe that heterosis, having
resulted from the mingling of two widely different germ plasms, will probably
have many genes associated with characters which in their relatively homo-
zygous state are far from the Corn Belt ideal of what a corn plant should look
like. It is highly probable that much of the so-called "junk" revealed by in-
breeding was extreme segregants from this wide cross, and that it was closely
associated with the genes which gave open-pollinated dents their dispersed
vigor. It is significant that some very valuable inbreds (L317 is a typical ex-
ample) have many undesirable features. For this reason, many such inbreds
are automatically eliminated even before reaching the testing stage.

If one accepts the fact that Corn Belt dents resulted from the compara-
tively recent mingling of two extremely different races of maize, then on the
simplest and most orthodox genetic hypotheses, the greatest heterosis could
be expected to result from crosses between inbreds resembling the Southern
Dents and inbreds resembling the Northern Flints. If heterosis (as its name
implies) is due to heterozygous genes or segments, then with Corn Belt corn
on the whole we would expect to find the greatest number of differing genes
when we reassembled two inbreds — one resembling the Northern Flint, the
other resembling the original Southern Dent.

Theory (Anderson, 1939a), experiment (Anderson, 1939b; Brown, 1949),
and the results of practical breeding show that linkage systems as differenti-
ated as these break up very slowly. On the whole, the genes which went in
together with the Northern Flints still tend to stay together as we have
demonstrated above. This would suggest that in selecting inbreds, far from
trying to eliminate all of the supposed "junk," we might well attempt to
breed for inbreds which, though they have good agronomic characters like
stiffness of the stalk, nevertheless resemble Northern Flints. On the other
hand, we should breed also for those which resemble Southern Dents as close-

140 EDGAR ANDERSON AND WILLIAM L. BROWN

ly as they can and still be relatively easy to grow and to harvest. It would
seem as if the opposite generally has been done. A deliberate attempt has
been made to produce inbreds which look as much as possible like good Corn
Belt maize in spite of being inbreds.

There are, of course, practical necessities in breeding. In this direction the
work of corn breeders is a remarkable achievement. Strong attention to lodg-
ing resistance, to desirable kernel shapes and sizes, and to resistance to
drought and disease has achieved real progress. The inbred-hybrid method
has permitted much stronger selection for these necessary characters than
was possible with open-pollinated maize. Most Corn Belt dents now plant
well, stand well, and harvest well.

Perhaps partly because of these practical points there has been a conscious
and unconscious attempt on the part of many breeders to select for inbreds
which are like the Corn Belt ideal in all characters, trivial and practical
alike. The corn shows are now out-moded, but corn show ideals still influence
corn breeding. For instance, there has been an effort to produce plants with
greatly arching leaves, whose margins are uniformly ruffled. Such characters
are certainly of a trivial nature and of secondary importance in practical pro-
grams. Any potential heterosis closely associated with upright leaves, yellow
green leaves, tillering, or blades on the husk leaves has seldom had a chance
to get into inbreds where it could be tested on a basis of achievement. It
would seem highly probable that, in not basing the selection of inbreds more
soundly on performance, we have let much potential heterosis slip through
our sieve of selection.

Heterosis Reserves

These considerations lead us to believe that there is probably a good deal
of useful heterozygosis still ungathered in high yielding open-pollinated
varieties. There is also a distinct possibility that still more could be added
by going back to the Northern Flints and Southern Dents with the specific
object of bringing in maximum heterozygosity. From our experience it is
more likely that superior heterosis is to be found among the best flints than
among the best dents. On the whole, the Northern Flints have been farthest
from the corn breeders' notion of what a good corn plant should look like.
Flint-like characteristics (tillering, for example) have been most strongly
selected against, both in the open-pollinated varieties and the inbreds derived
from them.

Several of the widely recognized sources of good combining inbreds are
open-pollinated varieties with a stronger infusion of Northern Flints than
was general in the Corn Belt. This is particularly true of Lancaster Surecrop,
the excellence of whose inbreds was early recognized by several breeders in
the United States Department of Agriculture. In our opinion, it is probable
that the greater proportion of flint germ plasm in Lancaster Surecrop has

ORIGIN AND SIGNIFICANCE OF CORN BELT MAIZE 141

made it an outstandin<f source of inbreds of proven highly sj)ecific combining
ability when used with other Corn Belt inbreds. This is not an isolated ex-
ample, and even more extreme cases could be cited. We think it is a reason-
able working hypothesis that Northern Flint varieties of superior productiv-
ity might be efficient sources of improved heterozygosity for the United
States Corn Belt.

Morphological Characters as Related to Heterosis

To put this hypothesis in different language, morphological characters, if
carefully chosen, may be used as criteria of specific combining ability in Corn
Belt inbreds. Before presenting data bearing directly on this hypothesis, two
points need to be emphasized and discussed: (1) the effective selection of
morphological criteria, and (2) the relativity of all measures of effective
combining ability.

Previous studies (Kiesselbach, 1922; Jenkins, 1929; and others) have indi-
cated that the only positive correlations between the morphology of inbreds
and their combining ability are those involving characters of the inbreds
which are indicative of plant vigor. Reference to these investigations shows
that the characters chosen were such superficial measurements as date of
silking and tasseling, plant height, number of nodes, number of ears, ear
diameter, etc. Unfortunately, the morphology of the maize plant is not a
simple matter. It is so complex that one needs technical help on morphology
quite as much as he would in biochemistry were he studying the concentra-
tions of amino acids in the developing kernel.

Accordingly, we first familiarized ourselves thoroughly with the technical
agrostological facts concerning the detailed gross morphology of grasses in
general and Zea in particular. Just as in the case of a biochemical study of the
kernel, we found that further original research was necessary if the investiga-
tion was to be carried on effectively. We have accordingly undertaken de-
tailed studies of internode patterns and branching of the inflorescence; the
venation, size, and shape of the male spikelet, the development of the husk
leaf blades, the external anatomy of the cob, and the morphology of the
shank. Some of these investigations are still continuing, and must continue
if inbred morphology and combining ability are to be effectively correlated.

It is impossible to produce an absolute measure of combining ability.
When one speaks of combining ability of two inbreds, he always refers to
their behavior with each other compared to their behavior with certain other
inbreds or open-pollinated varieties. This is such a relative measure that the
scoring of a particular Fi cross as very low or very high in combining ability
might depend solely upon our previous experience with the two inbreds. We
may illustrate this point with an extreme example. Let us suppose that we
have inbreds IF and 2F derived directly from Northern Flints, and inbreds
lOD and IID derived from Southern Dents. Were we to cross IF X 2F and

142 EDGAR ANDERSON AND WILLIAM L. BROWN

lOD X IID we would expect relatively little heterosis within either of the
crosses. Accordingly, when we crossed 2F X IID we would rate this cross
as having high specific combining ability. On the other hand, had we origi-
nally crossed 2F X lOD and IID X IF, then there would probably have
been almost equally great heterosis in each of the crosses. Had these been
used as a basis for comparing the heterosis of 2F X IID, then our notion as
to the amount of heterosis in these crosses would have been very different
than it would have been had comparisons been made with IF X 2F or
lOD X IID.

If the germ plasms of the two main races of maize involved in Corn Belt
dents are still partially intact as a result of linkages, it should be possible to
classify inbreds on the basis of morphological differences according to their
flint and dent tendencies. If this can be done, and if genetic diversity is im-
portant in bringing about a heterotic effect in hybrids, one should be able to
predict with some accuracy the relative degree of heterosis to be expected
from crossing any two inbred lines. With this hypothesis as a background, a
series of experiments was started three years ago to determine whether or not
hybrid vigor in maize, as expressed in terms of grain yield, could be predicted
on the basis of morphological differences of inbreds making up the Fi hybrids.

Fifty-six relatively homozygous inbred lines consisting of eighteen
U.S.D.A. or experiment station lines, and thirty-eight strains developed by
the Pioneer Hi-Bred Corn Company were scored for the following character-
istics: row number, kernel length, denting, development of husk leaf blades,
number of secondary tassel branches, glume length, and chromosome knob
number. For each of these characteristics the two extremes in the eastern
United States are to be found in Southern Dents and Northeastern Flints.
At least twelve plants of each of the fifty-six inbreds were scored, and these
scores were then averaged to give a mean value for the line. The resulting
means were translated into numerical index values, in which a low value
represents Northern Flint-like tendencies, and a high value Southern Dent-
like tendencies. For example, the mean row number values for the inbreds
studied ranged from 11.2 to 19.5. These were arranged in the following index
classes.

1 2 3 4 S 6 7

11.2-11.7 11.8^12.3 12.4-12.9 13.0-13.5 13.6-14.1 14.2-14.7 14.8-15.3

8 9 10 11 12 13 14

15.4-15.9 16.0-16.5 16.6-17.1 17.2-17.7 17.8-18.3 18.4-18.9 19.0-19.5

Index values for the other characteristics were arranged similarly, and
from the individual characteristic inbred indices (each being given equal
weight) a total "Inbred Index" was determined as is shown by example in
Table 8.2.

After index values had been determined for the inbreds, single cross combi-

ORIGIN AND SIGNIFICANCE OF CORN BELT MAIZE

143

nations were made and these tested for yield. In 1948, sixty-six single crosses
were grown in yield tests in Iowa and in Illinois. Each Fi hybrid was repli-
cated six times in each test. At the end of the season, yield of grain was de-
termined on the basis of 15 per cent moisture corn. Actual yields in bushels
per acre and morphological differences of the inbreds involved in each of the
crosses were then plotted on a scatter diagram as shown in Figure 8.6. It will
be noted that although the observations exhibit considerable scatter, there is
a tendency for grain yields in single crosses to increase as the morphological
differences between the inbreds making up the crosses become greater.
Actually the correlation coefficient between yield and index differences in
this case was r = +.39.

The experiment was continued in 1949, in which 100 Fi hybrids were
tested for yield. In this experiment three characters only were used to deter-

TABLE 8.2
INBRED INDICES BASED ON SEVEN CHARACTERS

Inbreds

Row

No.

Kernel
Length

Dent-
ing

Husk
Leaves

Tassel
Branches

Spikelet
Length

Chromo-
some
Knobs

^ , J Sums of 7
Inbred ^.„
^ , Differences
Index .,, ^ „.

without Signs

Hy

Oh40b..
MYl...

9

2

14

14

8

11

4

4
14

14

1
14

5

4

14

6
1
6

12

3
9

f±

IV/ 30

82 /

mine the index of relationship between the inbreds used. These were row
number, kernel length, and degree of development of husk leaf blades.
Elimination in this experiment of certain morphological characteristics used
previously was done largely to facilitate ease and speed of scoring. It had
been determined previously that, of the several characteristics used, those
having the highest correlation with yield were differences in row number,
kernel length, and husk leaf blades. There was likewise known to be a rather
strong association between each of these characteristics and tassel branch
number, denting, glume length, internode pattern, and chromosome knob
number. Therefore the scoring of these three characteristics probably covers
indirectly nearly as large a segment of the germ plasm as would scores based
on all seven characteristics.

The 1949 tests in which each entry was replicated six times in each loca-
tion were again grown both in Iowa and Illinois. Yields from these tests,
plotted against index differences of the inbreds, are shown in Figure 8.7.
As in the previous year's data, a pronounced tendency was shown for hybrids
made up of inbreds of diverse morphology to produce higher grain yields than
hybrids consisting of morphologically similar inbreds. The correlation co-

1948

.

;

#

•

115

"

•

•

•

u

\

Mk

•

W

•

V

DC

no

.

•

•

#

(J

•

•

•

•

•

•

•
•

•

#

^

•

•

•

•

•

D

105

•

•

•

•
•

•

•

•
•

•

•

•

OQ

• •

•

•

•

•

•

1

•
•

•
•

•

#

•

Q

100

-

•

•

•

•

J

• •

•

•

•

LiJ

• #

•

>

95
90

•

•

85

•

0 5 10

NDEX DIFFERENCE

r=+.38

Fig. 8.6 — Scatter diagram depicting relationship between grain yields of 66 single cross
hybrids and morphological differences of inbred parents of the hybrids. Explanation in text.

ORIGIN AND SIGNIFICANCE OF CORN BELT MAIZE

145

efficient between yield and index differences is r = +.40, a significant value
statistically.

In terms of practical corn breeding, the distribution of single crosses in
Figures 8.6 and cS.7 is of particular significance. If these observations are
critical (we have produced a repea table result) it means that one could have
eliminated from the testing program the lower one-third of the crosses on the
basis of index differences, without losing any of the top 10 per cent of the
highest yielding hybrids. In the case of the 100 hybrids in Figure 8.7, one could
have eliminated from testing 35 per cent of the crosses, thereby permitting the
inclusion of 35 additional hybrids in this particular testing area. If further

1949

130

125 •

120 ■

3 lis
CD

I I 10

a

LJ 10^

LJ

q:
u

>

00

95

• • • •

0

• • • • •

:.• •

• •

• •

5 10 15 20

INDEX DIFFERENCE

25

r= +.40

^x

Fig. 8.7 — Scatter diagram depicting relationship between grain yields of 100 single cross
hybrids and morphological differences of inbred paren ts of the hybrids. Explanation in text .

146 EDGAR ANDERSON AND WILLIAM L BROWN

experiments show that the method is reliable, such a procedure should
expedite most corn breeding programs.

Our method of scoring does not take into account the variation brought
about by the infusion of germ plasm other than that from Northern Flints
and Southern Dents. Perhaps this is one reason why we have not ob-
tained higher correlations between differences in inbred morphology and
yield. There are a few inbreds in the Corn Belt which appear to be affiliated
either with Caribbean flints or the Basketmaker complex. Scoring of such
inbreds on a scale designed for Northern Flints and Southern Dents un-
doubtedly leads to conflicting results. It is hoped that experiments now in
progress will aid in clarifying this situation.

SIGNIFICANCE OF FLINT-DENT ANCESTRY
IN CORN BREEDING

The Flint-Dent ancestry of Corn Belt maize bears upon many other breed-
ing problems besides those concerned with heterosis. Its widest usefulness is
in giving a frame of reference for observing and thinking about the manifold
and confusing variation of Corn Belt maize. When one becomes interested
in any particular character of the corn plant, he no longer needs to examine
large numbers of inbreds to understand its range of variation and its general
over-all direction. He merely needs to examine a few inbreds, and a Northern
Flint and a Southern Dent. A good part of the variation will then be seen to
fall into a relatively simple series from an extreme Northern Flint type to the
opposite Southern Dent extreme, with various intermediates and recombina-
tions in between. This is quite as true for physiological or biochemical char-
acters as for glumes, lemmas, or other morphological characters. One is then
ready to study further inbreds with a framework in his mind for sorting out
and remembering the variation which he finds.

The actual breeding plot efficiency of this understanding will be clearer if
we cite a practical example. Now that corn is picked mechanically, the size,
shape, texture, and strength of shank are important. When maize was picked
by hand, the hand had a brain behind it. Variations in ear height, in the
stance of the ear, and in the strength and shape of the shank were of minor
significance. Now that machines do the work, it is of the utmost practical im-
portance to have the shank standardized to a type adapted to machine
harvesting. When this necessity was brought to our attention a few years
ago, there were few published facts relating to variation in the shank. Exam-
ination of a few inbreds showed that though this organ varied somewhat
within inbreds, it varied more from one line to another than almost any
other simple feature of the plant. We accordingly harvested typical shanks
from each of 164 inbreds being grown for observation in the breeding plots
of the Pioneer Hi-Bred Corn Company. We also examined a number of
Northern Flints, and had they been available, we would have studied the

ORIGIN AND SIGNIFICANCE OF CORN BELT MAIZE

147

shanks on typical Southern Dents. However, simply b}- using the hyj)othesis
that one extreme would have to come in from the Northern Flints, the other
from the Southern Dents, we were able within one working day to tabulate
measurable features of these shanks and to incorporate all the facts in a

<

X
h-

o

z

LJ

<

I-

o

I-

'4

4

V

)l

i

V

4

265

i V
4 4

4 4

U 4

4

'^4 4

V

u

'i

4

V

t

210

4 '* ^ 4

V /
^4

4

\

4

/

U 4

'i

u

w

4

)r

W'

4^4 ^i 4 ^i

4

u

%

^w

135

i, ^

V

V

V

*-

V

^

¥

t )i

^ 4

i

a
^

4

•

60

10

15

20

WIDTH OF MID-INTERNODE OF SHANK

LENGTH OF

MAXIMUM WIDTH -

CONDENSED

LONGEST INTERNODE

MINIMUM WIDTH

INTERNODES

• 0-15

• 0

\o

¥ 16-35

i 1-3

■* 1-2

436- +

i4- +

• 3■-^

Fig. 8.8 — Pictorialized diagram showing relationship in 164 Corn Belt inbreds of the fol-
lowing shank characters: total length, width of mid internode, length of longest internode,
maximum width minus minimum width, and number of condensed internodes.

148 EDGAR ANDERSON AND WILLIAM L. BROWN

pictorialized scatter diagram (Fig. 8.8). Using the method of Extrapolated
Correlates we were able to reconstruct the probable shank type of the
Southern Dents. (We later grew and examined them and verified our predic-
tions.) We arranged most of the facts concerning variation in shank type in
United States inbreds in a single, easily grasped diagram. All the technical in-
formation needed as a background for breeding was available after two days'
work by two people. Without the Northern Flint-Southern Dent frame of
reference for these miscellaneous facts, we might have worked around the
problem for several breeding seasons before comprehending this general, over-
all picture.

SUMMARY

1. Archaeological and historical evidence shows that the common dent
corns of the United States Corn Belt originated mainly from the purposeful
mixing of the Northern Flints and the Southern Dents.

2. Cytological and genetic evidence point in the same direction and were
used in the earlier stages of our investigations before the complete historical
evidence had been located.

3. The Northern Flints and Southern Dents belong to races of maize so
dififerent that, were they wild grasses, they would certainly be assigned to
different species and perhaps to different genera. Such cytological and
genetical evidence as is available is in accord with this conclusion.

4. The significance of these facts to maize breeding problems is outlined.
In the light of this information, the heterosis of Corn Belt maize would seem
to be largely the heterosis acquired by mingling the germ plasms of the
Northern Flints and Southern Dents. It is pointed out that most breeding
programs have been so oriented as to be inefficient in assembling the dis-
persed heterosis of the open-pollinated varieties of the Corn Belt. The possi-
bility of gathering more heterosis from the same sources is discussed and it is
suggested that more might be obtained, particularly among the Northern
Flints.

5. Morphological characters of dents and flints, if carefully chosen, should
be useful criteria for specific combining ability. The problem of selecting
such characters is described. Two seasons' results in correlating combining
ability and flint-dent differences are reported. They are shown to be statisti-
cally significant and of probable practical importance.

6. The practical advantages of understanding the flint-dent ancestry of
Corn Belt maize are discussed and illustrated by example. In brief these facts
provide a "frame of reference" for detecting, organizing, and understanding
much of the manifold variability in Corn Belt maize.

ADRIANO A. BUZZATI-TRAVERSO

Universiia, Isfituto di Gene/ico, Pavia, Ifaly

Chapter 9

Heterosis in
Population Genetics

Population genetics is the study of the genetic structure of populations.
Such a statement may look at first to be a truism, a tautology. The subject
matter of our research becomes very intricate, however, as soon as we try to
specify what we mean by the above definition. The terms "genetic structure"
and "population" may have different meanings according to what we are
willing to indicate by such words. It therefore seems convenient to start
with an analysis of the terms we are using. Such discussion will give us a
chance to see how the problem of heterosis is intimately connected with the
general theme of population-genetical studies. A few experimental data will
be used to illustrate such points.

Let us consider first what we mean by population. If we take a dictionary
definition, we find in Webster's that population is "all the people or in-
habitants in a country or section." It means, in this sense, the sum of indi-
viduals present at a certain moment over a more or less arbitrarily limited
territory. But this definition does not correspond to the requirements of our
studies, as I have tried to show elsewhere (Buzzati-Traverso, 1950). Such a
definition is a static one, while the population, as considered in the field of
population genetics, is a dynamic concept. We are interested not in the
number of individuals present at a certain time in a certain place and their
morphological and physiological characteristics. Instead, we are concerned
with the underlying mechanisms which bring about such characteristics, and
the particular size the population reaches at any particular moment. Since
such mechanisms depend upon the numerical dynamics of the population
and upon heredity, it follows that our concept of population is typically
dynamic. On this view, then, a population is an array of interbreeding indi-
viduals, continuous along the time coordinate.

149

150 ADRIANO A. BUZZATI-TRAVERSO

Consideration of a population as a phenomenon continuously occurring in
time makes it impossible for the experimental student of population genetics
to get a direct and complete picture of what is occurring within a population
at any particular moment. We can attempt to collect data on the population
under study only by freezing such flowing processes at particular time in-
tervals. Collecting observations on a population at different times gives us
a chance to extrapolate the direction and rate of the processes that have
occurred within the population during the time elapsed between two succes-
sive sets of observations. If the samples studied are large enough and give an
unbiased picture of the total population at the time when the sample is being
drawn, this experimental procedure ma}^ give us a fairly adequate idea of
what is going on within the array of interbreeding individuals continuous
along the time coordinate. That sum of individuals at a definite time, which
one usually means by population, is of interest to the population geneticist
only as an index of the particular evolutionary stage reached by the array of
interbreeding individuals. Since there are actual breeding and genetic rela-
tionships between the individuals of any such array, of any such population,
the population can be considered as the natural unit of our studies.

If we consider now what we mean by "genetic structure," our task be-
comes much more complex. At first we could assume that the genetic struc-
ture of a population could be properly described in terms of the gene frequen-
cies present at a certain time within a population. But this is only part of the
picture.

For the total description of the genetic structure of a population we have
to consider not only the frequencies of existing genes, but how these are
fitted within the chromosomes, how these allow the release of variability by
means of recombinations, how large is the amount of new variability pro-
duced by mutations, and several other factors which we cannot analyze now.
In a few words, the study of population genetics aims at the knowledge of the
breeding system of populations. This, as we shall see, is a rather difficult task
because of the complexity of factors responsible for the origin and evolution
of such systems.

EVOLUTIONARY FACTORS INVOLVED

When we take into consideration a species or a natural population at a
certain stage, we have to assume that such a natural entity is the product of
a series of evolutionary factors that have been at work in previous times and
that some, or all of them, are still operating on the population while we are
studying it. This means that we should try to explain the genetic structure of
the population in terms of such evolutionary factors.

Now, if we are willing to examine the nature of the known evolutionary
agencies, we conclude that these can be classified into two types. On one side
we find, in sexually reproducing organisms, a limited number of chromo-

HETEROSIS IN POPULATION GENETICS 151

somes, linkage between genes, sterility mechanisms, mating discriminations,
devices favoring inbreeding, and other conservative forces that aim at the
preservation of certain constellations of genes over a large number of genera-
tions. On the other side we find mutation pressure, recombination between
chromosomes, recombination among genes due to crossing over, outbreeding
devices, migration pressure, and other revolutionary forces that aim at the
production of genetic novelt3^

It seems reasonable to maintain that, at any particular time, a species or
a natural population can be considered as a sort of compromise between the
two conflicting forces — a compromise that is brought about through the
action of natural selection. In other words, the fine adjustment or adaptation
of a population to its environment is the expression of such compromise. At
any particular time the terms of the compromise between the conflicting
forces are always different as compared to other moments, as the compro-
mise itself is a dynamic process.

In order to reach the highest possible level of adaptation with respect to a
certain set of environmental conditions, natural selection is discriminating not
only for or against a certain individual genetic constitution, but for or against
a group of individuals, as w-ell. Sometimes selection acts at the level of the in-
dividual, sometimes it operates at some higher level. If we consider a genotype
that insures resistance against an infectious disease, present in a certain area of
distribution of a species, it wall be obvious that an individual carrying it shall
directly benefit by it. But if we consider a genotype producing fecundity
higher than the average of the population, this will be selected by the mere
fact that a larger number of individuals having such genetic constitution will
be present in the next generation. These, in their turn, shall have a chance of
being represented in the next generation greater than that of individuals
having a less fertile genotype. The individual itself, though, obtains no direct
advantage from such selection.

The next extreme condition we can consider is the one occurring when the
advantage of the individual is in conflict with the advantage of the group.
This is the case, for instance, of a genotype that would extend the span of
life far beyond the period of sexual activity — or higher fertility linked with
antisocial attitudes in the case of man. In both cases, natural selection favor-
ing the preservation of the group will discriminate against the individual. A
similar mechanism must have played a great role in various critical periods
of organic evolution. When intergroup selective pressure is in the opposite
direction from intragroup selection, a sort of compromise has to be reached
between the two conflicting tendencies. This can be reached in many differ-
ent ways that are best illustrated by the great variety of life histories and
mating systems to be found in the living world.

Those factors which we have classified as conservative tend to j)roduce
genetic homogeneity, or what is technically known as homozygosis. Factors

152 ADRIANO A. BUZZATI-TRAVERSO

that we have named revolutionary tend to produce genetic heterogeneity or
heterozygosis. Thus we come to the conclusion that the mentioned compro-
mise brought about by selection consists of the pursuit of an optimum level
of hybridity with respect to the conditions under which the organism lives.
Such a hybridity optimum is the product, not only of the mutation rate and
selective value of single genes, but also depends largely upon the genetic sys-
tem and the mating system — the breeding system — of the considered species
or population.

The genetic structure of natural populations cannot be solved only in
terms of individual variations observable in the group. Instead, it must be
integrated into a unitary research on changes in gene frequencies as related
to the underlying breeding systems. This is why we are justified in consider-
ing the natural population as a unit, since individual variations must be
referred to the genetic balance of the whole aggregate of individuals.

What is that hybridity optimum I was speaking about but heterosis? How
else could heterosis be defined in population problems other than that type
and amount of heterozygosity that gives the population or the individual the
best adaptive value with respect to the conditions in which the organism
lives? With this view, then, it becomes feasible to analyze experimentally
what morphological and physiological characteristics of the hybrids produce
the better adaptation.

MECHANISMS WHICH PROMOTE HYBRIDITY

In studying how heterosis mechanisms are brought about in living crea-
tures, we may attempt a sort of classification of the devices present in plants
and animals insuring hybridity. Starting from the most complex and proceed-
ing to the less complex cases, we can distinguish three types of mechanisms:
(1) mating systems, (2) chromosome mechanisms, and (3) gene effects.

We will not discuss in detail all the devices insuring hybridity found in
plants and animals. We will mention a few, in order to show how many differ-
ent paths have been followed in evolution to reach the same sort of results.

Under the heading "mating systems" we may mention homo- and hetero-
thally among fungi; monoecism and dioecism, incompatibility mechanisms,
and heterostyly among flowering plants. Here, in some cases such as Primula
scofica, there is close relation between the variability of ecological conditions,
and, therefore, of selection pressure and the efficiency of the incompatibility
mechanisms. Other species of this genus present in England are character-
ized by heterostyly and incompatibility devices to insure the occurrence of
outcrossing, apparently necessary to meet the requirements of varied eco-
logical conditions. Primula scofica, living in a very specialized ecological
niche, shows that such a mechanism has broken down. In fact, it looks as if
the requirements of a constant environment are met better by populations
genetically less diversified.

HETEROSIS IN POPULATION GENETICS 153

Among animals, the largest ])art of which are not sessile and therefore not
bound to the ground, the differentiation into two sexes offers the best solution
to the problem of insuring a wide range of crossing among different geno-
types. But even here we see that special behavior patterns have been de-
veloped for this purpose. These may be courtship relationships, sexual selec-
tion, dominance relationships among a group of animals, or protandry
mechanisms, where the presence of two sexes in hermaphrodites could reduce
the amount of outcrossing and therefore endanger the survival of the species.
Even among parthenogenetic animals, such as Cladoceran Crustacea, the ap-
pearance of sexual generations after a long succession of asexual ones seems
to depend upon extreme environmental conditions. For its survival, the
species must shift over to sexual reproduction in order to obtain a wider
range of genetic combinations, some of which might be able to survive under
the new set of conditions.

At the level of the chromosome mechanisms, several examples of perma-
nent hybrids are known well enough to be sure that they play an im-
portant role for the survival of some flowering plants. In animals, too, some
similar mechanism may be present. In a European species of Drosophila
which we are studying now^, Drosophila subobscura, one finds that practically
every individual found in nature is heterozygous for one or more inversions.
It looks as if the species were a permanent hybrid.

Rarely, though, one finds individuals giving progeny wath homozygous
gene arrangement. Such cases have been observed only three times: once
in Sweden, once in Switzerland, once in Italy; and they are very peculiar
in one respect. The three homozygous gene arrangements are the same, even
though the ecological and climatic conditions of the three original popula-
tions were as different as they could be. It looks as if the species could
originate only one gene arrangement viable in homozygous condition, and
that this may occur sporadically throughout its vast distribution range
(Buzzati-Traverso, unpublished).

At this level too is the fine example of heterozygous inversions from the
classical studies of Dobzhansky (1943-1947). They have demonstrated that
wild populations of Drosophila pseudoobscura show different frequencies of
inversions at different altitudes or in the same locality at different times of
the year. Variation in the frequency of inversions could be reproduced ex-
perimentally in population cages by varying environmental factors such as
temperature. It is shown in such a case that natural selection controls the in-
crease or decrease of inversions determining an interesting tyj)e of balanced
polymorphism. Finally, according to the investigations of Mather (1942-
1943) on the mechanism of polygenic inheritance, it appears that linkage rela-
tionships within one chromosome, even in the absence of heterozygous inver-
sions, tend to maintain a balance of plus and minus loci controlling quantita-
tive characters.

154 ADRIANO A. BUZZATI-TRAVERSO

We come then to the third level, that of gene effects. Here it is well known
that heterozygotes for a certain locus sometimes show a higher viability or a
better adaptation to the environment than either homozygote. The most
extreme examples are those of the widespread occurrence of lethals in wild
populations of Drosophila, noted in the next section.

Every population of plants and animals that has been studied from
the genetic viewpoint has proved to be heterozygous for several loci. We
have now at our disposal a large series of data showing that the phenomenon
of genetic polymorphism is frequent in plants, animals, and man. These offer
to the student of evolutionary mechanisms the best opportunities to test his
hypotheses concerning the relative importance of selection, mutation pres-
sure, migration, and genetic drift as factors of evolution. Wherever we find
a well established example of balanced polymorphism, such as that of blood
groups and taste sensitivity in man, it seems safe to assume that this is due to
selection in favor of the heterozygote. How this selection actually may pro-
duce an increase in the chances of survival of the heterozygote, as compared
to both homozygotes, is an open question. When the characters favored by
natural or artificial selection are the result of several genes in heterozygous
condition, the analysis becomes very difficult indeed, as the experience of
plant and animal breeders clearly shows.

EXPERIMENTS WITH HETEROSIS

The importance of the problem of heterosis forpopulation-genetical
studies is clearly shown, not only by such general considerations and by the
few examples mentioned, but also by the everyday experience of people
interested in such lines of work. I have come across problems involving
heterosis several times and shall describe some results we have obtained
which may be of interest for the problem under discussion, especially at the
level of single gene differences.

Several Drosophila workers have been able to show the occurrence of
heterosis in the fruit flies. L'Heritier and Teissier (1933), Kalmus (1945), and
Teissier (1947a, b) have shown that some visible recessive mutants of Dro-
sophila melanogaster such as ehony and sepia have a higher selective value in
heterozygous condition than either of the corresponding homozygotes under
laboratory conditions. Dobzhansky and collaborators in Drosophila pseudo-
obscura, Plough, Ives, and Child, as well as other American and Russian
workers in Drosophila melanogaster, have shown that recessive lethals are
widely spread in natural populations. It is generally accepted that such genes
are being maintained in the population because the heterozygotes are being
selected. Teissier (1942, 1944) has brought similar evidence under labora-
tory conditions for Drosophila melanogaster.

It has been shown in several populations of species of the genus Drosophila
that heterozygous inversions are being selected, under natural and ex-

HETEROSIS IN POPULATION GENETICS 155

perimental conditions. It seems, however, that the study of selection in
favor of the heterozygote for single loci deserves more careful analysis. The
whole problem of heterosis for several genes affecting quantitative characters
will be solved, I think, only when the more simple cases of heterosis where
single gene differences are involved shall be cleared up. I have been lucky
enough to come across some useful experimental material for the purpose.

For a number of years I have kept about one hundred different wild
stocks of Drosophila melanogaster coming from diflferent geographical locali-
ties. Such stocks were maintained by the usual Drosophila technique of
transferring about once a month some 30-40 flies from one old vial to a new
one with fresh food. About twice a year I look at the flies under the micro-
scope. Since all such stocks were wild type, no change by contamination was
expected, as these stocks were phenotypically alike. Contamination by mu-
tants kept in the laboratory could not have produced any appreciable result,
owing to the well known fact that both under laboratory and natural condi-
tions mutants are generally less viable than the normal type. To my sur-
prise, however, I happened to observe at two different times, in two different
wild stocks, that a fairly large number of the flies showed an eye color much
lighter than the normal. These two mutants proved to be indistinguishable
recessive alleles at the same locus in the third chromosome. The presence
of the homozygotes has been checked at different times over a number of
years.

In the summer of 1947 while collecting flies in the wild for other purposes,
I found in the neighborhood of Suna, a small village on the western shore of
the Lake Maggiore, in Northern Italy, several individuals of both sexes show-
ing the same eye color. From these a homozygous stock for such mutant was
obtained. Crossing tests proved that it was another allele of the same locus as
the above mentioned. The occurrence of several individuals mutant for an
autosomal recessive within a free living population was remarkable enough.
But finding that the same gene was concerned as in the laboratory stocks, I
suspected that such a mutant might have a positive selective value, both
under laboratory and natural conditions.

I began an experiment to check this point. Two populations in numerical
equilibrium were started, applying the method previously used by Pearl for
the study of population dynamics of Drosophila, described in detail else-
where (1947a). Sixteen light-eyed individuals, eight males and eight females,
were put together in one vial with sixteen wild type flies. The gene frequency
at the beginning of the experiment was therefore .5. Under the experimental
conditions the population reached an equilibrium in respect to the amount
of available food at about 700-900 flies per vial. After about twenty genera-
tions, assuming that each generation takes 15 days, the frequency of recessive
homozygotes was about 40 per cent. Assuming random mating within the
population, taking the square root of .40 one gets a gene frequency for the

156 ADRIANO A. BUZZATI-TRAVERSO

light-eyed gene of about .63. Since in both parallel populations the gene fre-
quency was similar, one could conclude that selection had favored the mutant
type, shifting its frequency from .5 to .63 in the course of about twenty
generations.

Such an experiment did prove that the mutant gene had a positive selec-
tive value. It was impossible to know whether in the long run it would have
eventually eliminated its normal allele from the population. At this stage, I

0.875^

Fig. 9.1 — Variation in the frequency of the Hght-eyed gene in selection experiments. In the

abscissae is the number of generations, in the ordinates the gene frequency. Each line

represents a single experiment on an artificial population.

have begun a new experiment along the same lines, but with different gene
frequencies to start with. Two populations were started with 2 males and 2
females of the mutant type, plus 14 males and 14 females of the normal type.
Two populations were started with 16 mutant and 16 wild flies, and two
populations with 28 mutant and 4 wild type flies.

I had, therefore, at the beginning of the experiment six populations. Two
had a gene frequency of the light-eyed mutant approximately equal to .125.
Two had a gene frequency of .5, and two had a gene frequency of .875.
Figure 9.1 shows the result of such an experiment after about fifteen genera-
tions. Crossings of wild type males, taken from the populations, with homo-
zygous recessive females showed that there was no significant departure from
random mating within the population. The gene frequencies indicated on the

HETEROSIS IN POPULATION GENETICS

157

orclinates were obtained by taking the square root of the observed frequencies
of homozygous recessives.

The following conclusions can be drawn: (1) the three experimental popu-
lations, each being run in duplicate, have reached the same gene frequency
at about the .579 point; (2) natural selection has been acting on the three
populations producing the same end results, irrespective of the initial gene
frequency; (3) natural selection has been acting in favor of the heterozygous
flies; and (4) the homozygous mutant seems to be slightly superior in its
survival value to the homozygous normal allele.

It was of considerable interest to determine whether the intensity of selec-
tion operating in the three experiments was the same. Since the three experi-
mental curves (each being the mean of the two duplicate populations) could
not be compared directly. Dr. L. L. Cavalli elaborated a mathematical
analysis of the problem (Cavalli, 1950). The function of gene frequency linear
with time F, when the heterozygote is at an advantage, is given by:

Y = Qe^Og p-\-pe\0g q -\0g[pe-p] ,

where p and q are the gene frequencies at the beginning of the experiment in
a random breeding population, and pe and qe are the equilibrium frequencies.
By means of this function it is possible to transform the experimental curves
to linear ones. Results can then be plotted graphically for the three experi-
ments. Fitting straight lines with the method of maximum likelihood, one
obtains the following values for the constants of the linear regression equa-
tion:

Experi-
ment

Initial Gene
Frequency
(Observed)

Slope

Position

Initial Gene

Frequency

(Theoretical)

1

2

3

.500
.125
.875

.0879
.0631
.0726

+ 1.21
- .41

+ .27

.425
.100
.830

The position is the transformed value of the initial gene frequency which
is given in the last column, and is in good agreement with the experimental
value. If one tests the parallelism of the three regression lines so obtained, one
gets a chi square of 4.0 with two degrees of freedom. Parallelism therefore
seems to be satisfactory. This implies that the intensity of selection is inde-
pendent of initial conditions.

If we take these results together with the two independent occurrences of
the same mutant gene in different genotypical milieus, it seems safe to main-
tain that such a gene has a positive selective value with respect to its normal
allele, and that selection is acting mainly through a typical heterosis mecha-
nism. It is worth while to stress that this gene was found both in natural and

158 ADRIANO A. BUZZATI-TRAVERSO

experimental conditions. The exceptional occurrence of many mutant indi-
viduals in a free living population can be accounted for by assuming that
they have a higher selective value.

BASIS FOR SUPERIORITY OF THE HETEROZYGOTE

It would be interesting to try to find out how selection discriminates
against both normal and mutant homozygotes. I am just beginning to attack
this problem.

Dr. E. Caspar! has some interesting results on a similar problem, and I
wish to thank him for permission to quote them (Caspari, 1950). In free
living populations of the moth Ephestia kuhniella, this author has observed
a balanced polymorphism, whereby individuals having brown colored and
red colored testes occur in various numbers. The character brown behaves as
a complete dominant with respect to red. The polymorphism seems to be
determined by a higher selective value of the heterozygote. It has been pos-
sible to show that the heterozygote is equal or superior to the homozygous
recessive and the latter is superior to the homozygous dominant with respect
to viability. It was found that, while the heterozygote is equal or superior to
the homozygous dominant, the homozygous brown is superior to the homozy-
gous red with respect to mating activity. The dominance relationships of such
two physiological characters are therefore reversed.

There is no decisive evidence for heterosis for any of the characters
studied. The recessive for the testis color acts as dominant with respect to
viability, and the dominant testis color acts as dominant with respect to
mating behavior. The net result is a selective advantage of the heterozygote
that can account for the observed polymorphism. This seems a good ex-
ample of how a heterosis mechanism can be determined by the behavior of
two visible alleles in heterozygous condition. It is hoped that similar analyses
will be developed for other cases of balanced polymorphism.

The search for clear-cut examples of heterosis depending on single genes
seems to me the most promising line of attack on the general problem under
discussion. If I could find another gene behaving in a way similar to the one
I have studied in Drosophila melanogaster, and could study the interaction of
the two, it would be possible to go a step further in the analysis of heterosis
mechanisms. The evidence derived from such single genes being favored in
heterozygous condition is likely to be very useful in more complex condi-
tions where the action of several genes is involved.

When we come to consider the selective advantage of polygenic charac-
ters, even in such an easy experimental object as Drosophila, the problem
becomes very entangled indeed. In recent years I have been studying, for
example, a number of quantitative characters being favored by natural
selection in artificial populations in numerical equilibrium, such as the ones
I have been speaking about. I have set in competition at the beginning of one

HETEROSIS IN POPULATION GENETICS 159

experiment two stocks differing for visible mutants. One stock was while-
and Bar-eyed, the other stock was normal for both characters. The two
stocks differed, too, in a number of quantitative characters such as fecundity,
fertility, rate of development, longevity, and size.

After about thirty generations the two mutant genes had been wiped out.
This could have been expected on the basis of previous data of L'Heritier and
Teissier on the elimination of such genes in artificial populations. At that
time, however, I did not discard the populations, but kept them going for
some seventy more generations. All the individuals present in the popula-
tions were phenotypically normal. But testing from time to time the values
of the above mentioned characters, I could establish that natural selection
was continuously operating and favoring higher fecundity, higher fertility,
higher longevity, and quicker developmental rate throughout the four years
that the experiment lasted. At the end, the flies present in the population
were superior by a factor of more than six to the mean of the considered
characters in the two original parental stocks. When I measured such values
in the Fi hybrids between the two stocks I could observe values higher than
those obtained after more than one hundred generations of selection.

The selection experiment could then be interpreted in two different ways.
Either (a) selection had picked up a new genotype made out of a new com-
bination of polygenes derived from the two parental stocks, or (b) selection
had just preserved by means of a heterosis mechanism a certain amount of
heterozygosity, which was at its highest value at the beginning of the experi-
ment. The fact that in the course of the experiment the factors had been
steadily improving seemed to be against hypothesis b, but I could not be sure
that was the case.

I then set up a new selection experiment, whereby I put in competition the
original stock white Bar with the normal type derived from the population
which had been subjected to natural selection for more than one hundred
generations. The result was clear. The genes white and Bar were elimi-
nated in this second experiment at a much higher rate than in the first ex-
periment. In the first experiment the gene frequency of the gene Bar after
ten generations had dropped from .50 to .15. In the second experiment, after
as many generations, the Bar gene frequency had dropped from .50 to .03.
It seems that the genotype produced by a hundred generations of natural
selection under constant conditions was so much better adapted to its en-
vironment that it could get rid of the competing genes with much greater
ease than the original wild type flies. But could it not be that all or at least
part of this result could be accounted for by the action of some heterosis
effect?

Another example of a similarly puzzling condition is an experiment on
artificial populations under way now in my laboratory. I would like to find
out whether it is possible to produce so-called small mutations or polygenic

160 ADRIANO A. BUZZATI-TRAVERSO

mutations with X-rays, and whether an increase in the mutation rate may
speed up the evolutionary rate under selection pressure.

For this purpose I have set up four artificial populations starting from an
isogenic stock of Drosophila melanogaster. One of these is being kept as con-
trol while the other three get, every two weeks, 500, 1000, and 2000 r-units
respectively. At the start, and at various intervals, I am measuring fecun-
dity, fertility, and longevity of the flies. The few data so far collected show
clearly that in the irradiated populations the percentage of eggs that do not
develop is much higher than in the control. This is due to the effect of
dominant and recessive lethals. But the startling result is that the fecundity,
measured by the number of eggs laid per day by single females of the irradi-
ated populations, is higher than in the control series. Probably X-rays have
produced a number of mutations for higher fecundity which have been ac-
cumulated by natural selection in the course of the experiment. But, are spe-
cific mutations for higher fecundity being produced, or am I dealing with
heterosis phenomena dependent upon nonspecific mutants?

These few examples from my own experience with population-genet ical
studies show, I think, how important the heterosis phenomenon can be in our
field of work. Both in natural and artificial populations, heterosis seems to
be at work, making our analysis rather difficult, but stimulating as well.
Closer contacts between students of selection and heterosis in plant and
animal breeding and students of evolutionary problems are to be wished.
Let us hope that a higher level of hybridization between various lines of
investigation might become permanent, since it surely will make our studies
more vigorous and better adapted to the requirements of a rapidly growing
science.

HAROLD H. SMITH

Cornell Universify

Chapter 10

Fixing Tronsgressive Vigor
in Nicofiona Rustica

Hybrid vigor has been observed to varying degrees among certain inter-
varietal hybrids of the self-pollinated cultivated species Nicotiana rustica L.
(Bolsunow, 1944; East, 1921). In experiments undertaken to obtain a larger
;V. rustica plant giving increased yield of nicotine, it was reported (Smith
and Bacon, 1941) that inbred lines derived as selections from hybrids among
three varieties exceeded the parents and Fi's in plant height, number of
leaves, or size of the largest leaf.

The general experience of breeders of self-pollinated plants has been that
improved varieties can be developed through hybridization followed by selec-
tion and inbreeding, to fix desirable transgressive characteristics. Yet it is
difficult to find data from which quantitative relationships of parents, Fi, and
transgressive inbred can be adequately evaluated; as from replicated and
randomized experiments in which the generations have been grown at the
same time under comparable conditions. In view of the increasing number
of reports on hybrid vigor in self-pollinated crop plants and its suggested
utilization (Ash ton, 1946), it was considered opportune to present relevant
data accumulated on N. rustica.

Since methods of partitioning phenotypic variance have become generally
available there was additional interest in making further study of the iV.
rustica material. Breeding results obtained in advanced selections could be
related to the heritability estimated from data on early generations.

MATERIALS AND METHODS

Four varieties of Nicotiana rustica were used in these experiments. Three

of them — hrasilia strain 34753, Olson 68, and tall type have been described

* Published as Paper No. 261 , Department of Plant Breeding, Cornell University, Ithaca
New York.

161

162 HAROLD H. SMITH

in Smith and Bacon (1941). The fourth was received originally from the
director of the Tabak-Forschungsinstitut, Baden, Germany, under the name
of texana, a designation which we have retained. It is a small, early-maturing
type. The four parental varieties were of highly inbred stocks maintained by
the Division of Tobacco, Medicinal and Special Crops of the United States
Department of Agriculture. The earlier part of the breeding program was
carried out while the writer was associated with this organization.

The advanced selection, designated Al, used in these experiments has a
complex genetic history of crossing, backcrossing, and inbreeding. This can
be briefly summarized by stating that its ultimate composition was, on an
average, 60 per cent 34753, 22 per cent Olson, 12 per cent tall type, and 6
per cent texana. About 82 per cent of the Al genotype was, on chance alone,
contributed by the two most vigorous parents, 34753 and Olson 68. This
calculation does not take into account any differential effect of selection on
changing the frequency of genes introduced from diverse parental origins.
Observation of the Al phenotype led us to believe that selection had further
increased the proportion of genes from the two most vigorous parents.

In 1947 the four parents, the six possible F/s, the three double crosses,
and the F4 generation (preceded by three generations of inbreeding) of line
Al were grown in a randomized complete block design with fifteen plants in
each plot and replicated six times. In 1949 the two most vigorous varieties
(Olson 68 and 34753), the Fi, F2, backcrosses of the Fi to each of its parents,
and the Fe generation of line Al were grown in a randomized complete block
design with twenty plants in each plot and replicated eight times.

Measurements were made on plant height, number of leaves or nodes, and
length of the largest leaf. In addition, data were taken on the width of the
largest leaf, number of days from planting to appearance of the first flower,
and total green weight of individual plants.

Typical plants of Olson 68, 34753, the Fi between these two varieties, and
selection Al are illustrated in Figure 10.1.

EXPERIMENTAL RESULTS

Data obtained from the 1947 and 1949 plantings are summarized in
Tables 10.1 and 10.2, respectively.

Phenotype-Genotype Relations

Preceding further biometrical analysis of the data, tests for evidence of
differential environmental effects and genetic interactions were made. For
the former, the relation between genotype mean and non-heritable variabil-
ity was determined by comparing means and variances of the parents and Fj
(1949 data. Table 10.2). For the characters plant height and leaf length, the
variances were unrelated to the means and the parental variances were not
significantly different^from each other. For node number, however, the

Fig. 10.1 — Typical field-grown (1949) ]ilants of Nicotiana riistica. Left to right: Olson 68,
brasilia strain 34753, Fi Olson 68 X 34753, and selection A^Fe). The scale shown at the

left is in inches.

TABLE 10.1

PLANT CHARACTERISTICS IN PARENTAL VARIETIES, HYBRIDS,
AND AN INBRED SELECTION OF NICOTIANA RUSTICA*

Type

Plant
Height

Leaf
Number

Leaf
Length

Mean
Leaf
Width

Mean
Days

TO

Ma-
ture

Gen-
era-
tion

Mean

Total

within
Plot

Mean

Total

within

Plot

Mean

Total

within

Plot

Mean
Green

Wgt.

Per
Plant

d.f.

Var.

d.f.

Var.

d.f.

Var.

Pi

Olson 68 (/4 )

in.

49.9

29.0

46.7

33.6

73
84
83
83

17.8
20.0
30.3
11.1

18 8
15.8
16 0
12.7

72
82
83
83

4.68
7.96
1.36
1.04

in.

11.7
8.7
6 2
6 6

72
82
83
83

1.07
1.43
1.24
1.01

in.

8.7
7.5
5.5
5.4

70.1
66.6
48.0
40.9

lbs.
1 51

34753 (B)

0 89

Tall type (C)

0 SI

Texana (D)

0 48

Average

01sonX34753UX5). .

OlsonXtall(^XC)

Olson Xtexana {A XD) .

34753 Xtall(BXC)

34753Xtexana(SX£>).
Tall Xtexana (CXC) . .

Average

iAXB)X(CXD)

{AXC)X(BXD}

{AXD)XiBXC)

Average

39.8

19.8

15.8

3.76

8.3

1.19

6.8

56.4

0.85

Fi

48.5
42.9
40.1
47.1
40.3
44.2

74
80
81
84
83
84

58.3
25.4
20 8
45.2
28.1
29.3

16.8
13 0
11.2
16 6

14.4
15.5

74
80
81
84
83
83

7.18
7.74
4 99
1.77
6 34
1.49

10 8

10.1

10.6

7.8

8 7

7.7

74
80
81
84
83
83

1.92
1.82
1.49
1.37
2.30
1.88

8.7
9.6
9.7
6.7
7.4
7.0

75.0
65.1
70.6
50.5
60.0
51.8

1.47
1.16
1.13
0.76
0.93
0.95

43.8

34.5

14.6

4.92

9.3

1.80

8.2

62.2

1.07

FiXFi

41.9
39 6

42.5

75
82
81

25.1
61.1
39.9

14.0
12.4
13.9

74
79
80

10.60

9.29

10 80

8 8

9 8
9.7

74
79
80

2.36
3.11
2.50

7.7
8.8
8.7

61.9
66.0
60.4

0.86
0.99
1.06

41.3

42.0

13.4

10.23

9.4

2.66

8.4

62.8

0 97

Selection Al

F4

54.9

78

78.8

19.9

77

S.28

10.4

77

1.51

8.0

79.6

1 83

Least si
5%le
l%le

?nificant diff. at

vel

2.68
3.56

1.22
1.62

0.89
1.19

0.76
1.00

4.11
5.46

0 25

vel

0 34

* Summary of 1947 data.

164

HAROLD H. SMITH

means and non-heritable variances were linearly related for both 1947 and
1949 data, and the parental variances were significantly different.

Tests to reveal the presence or absence of non-allelic interactions were then
made according to the method proposed by Mather (1949). Results are
shown in Table 10.3. No significant deviations from zero were found if the
level of significance was taken as P ^ .01. In each test, however, the P values
for number of nodes were less than for plant height or leaf length, possibly
owing to non-additive gene effects.

It was concluded, on the basis of these tests, that for the two characters

TABLE 10.2

PLANT CHARACTERS IN THE TWO MOST VIGOROUS VARIETIES OF

N. RUSTIC A, THEIR F,, F2, AND FIRST BACKCROSS PROGENY

AND IN SELECTION AKFe)*

Gen-
era-
tion

Pi
P2
Fi
F2
Bi
B2
Fe

Type

Olson 68

34753

Olson 68X34753...
(Olson X34753) self.

FiXOlson 68

FiX34753

Selection Al

Plant Height

Mean

in.
47.8
28.7
43.2
40.6
47.3
36.2
55.6

Total with-
in Plot

d.f.

141
143
140
149
149
148
133

Var.

15.46
22.63
39.18
99.19
40.28
101.50
29.77

No. OF Nodes

Mean

24.
21.

22.
23.
24.
21.

31.0

Total with-
in Plot

d.f.

136
106
110
119
138
117
126

Var.

3.
10.

8.

45
10
60

10.52
10.49

45
44

Leaf Length

Mean

in.
11
10.
11
11
11
10.8
12.0

Total with-
in Plot

d.f.

142
127
131
142
144
135
141

Var.

0.68
0.81
0.63

08
10

0.95
0.69

Least significant diff. at

5% level 2.55

1% level 3.42

1.37
1.83

0.49
0.66

* Summary of 1949 data.

TABLE 10.3
SCALING TESTS FOR AVERAGE ADDITIVENESS OF GENE EFFECTS*

Character

Test A

Dev.

Var.

Plant height. 3.6 2.86
No. nodes...' 2.8 1.50
Leaf length.. 0.9 0.47

Test B

Dev.

Var.

Test C

Dev.

Var.

2. 13. 03-. 04 0.54.13 0. 25'. 80-. 81 -0.515.48 0. 13 .89-. 90
2. 30. 02-. 03-1. 11. 71 0.84. 40-. 41 3.5 6.18 1.40. 16-. 17
1.32,. 18-. 19, 0.00.47 0.00 1.00 i 0.5 1.77 0.38 . 70-. 71

* Based on 1949 means.

FIXING TRANSGRESSIVE VIGOR IN NICOTIANA RUSTICA 165

plant height and leaf length, the data, as taken, could be used without
serious error in partitioning the variance of segregating generations. For
node number it was indicated that some correction should be made with
the data before further biometrical analysis was undertaken.

Mather suggested that difficulties of the sort encountered in these data
with node number may be overcome by finding a transformation of scale on
which they would be minimized. The transformations \/X, X'^, X^, and
\/a -\- bx on the individual measurements were made. In the latter transfor-
mation b is the linear regression coefficient and a the intercept. Also, for
^ya + bx — K, \/—K was taken as —\/K. In some cases the transforma-
tions reduced the departure from the preferred relationship in one test, only
to make the transformed data less preferable by another test. No transforma-
tion tried resulted in a consistent improvement over the original scale, and
consequently none was used.

It is evident that the significantly different variances in node number of
the two parental types were due mainly to different interactions between
genotype and environment. From previous experience we know that under
ideal conditions of growth, Olson 68 and strain 34753 show approximately the
same variability. The adverse weather conditions of the 1949 season were ob-
served to have a more deleterious effect on leaf number in strain 34753. Con-
sequently it was considered that the greater variability of this variety, com-
pared to Olson 68, could be attributed to a greater phenotypic interaction
between genotype and environment. In view of these relationships, the
analysis of the data on node number was approached in another way, as
mentioned below under "Partitioning Phenotypic Variance."

First Generation Hybrids

Deviations of the Fi means from mid-parent values (arithmetical average
between parental means) can be used to estimate the preponderance of
dominant gene effects, acting in one direction, at loci by which the parental
complements differ. Mid-parent values were calculated from the 1947 da-
ta on the four original varieties. The results for each line are summarized in
Table 10.4. The data shown were obtained by first calculating the difference
between the Fi mean and the mid-parent (Fi — MP) for each cross, then tak-
ing the average of the differences for each group of three F/s involving the
parent variety under consideration. The ratio of the deviation of the Fi from
the mid-parent to half the parental difference, Fi — MP/\{P2 — Pi) , is a meas-
ure of the relative potence (Mather, 1949; Wigan, 1944) of the gene sets. Po-
tence ratios, calculated from averages, are shown in Table 10.4. For plant
height and leaf length the Fi means fall, on an average, about .6 of the dis-
tance from the mid-parent toward the larger parent. For leaf number the Fi
means fall, on an average, about .7 of the distance from the mid-parent
toward the smaller parent.

166

HAROLD H. SMITH

The Fi's were taller and had larger leaves, on an average, than the mid-
parent. It was concluded, therefore, that a preponderance of dommant+
genes was mvolved in determining differences in plant height and leaf length.
In the development of the parent varieties, selection resulted in the accumu-
lation of dominant-f modifiers, as is usually the case in naturally cross-
pollinated plants.

The result with the character leaf number was different in that the Fi had
fewer leaves, on an average, than the mid-parent. Evidently, in the evolution
of the varietal gene sets, there had been accumulated a preponderance of
recessive+ modifiers (or dominant genes ior fewer leaves) at the loci by which

TABLE 10.4

DIFFERENCE BETWEEN THE Fi AND MID-PARENT (Fi - MP) AND THE
POTENCEt RATIO IN INTERVARIETAL HYBRIDS. BOTH V.ALUES ARE
EXPRESSED AS THE AVERAGE FOR EACH VARIETY IN CROSSES WITH
THE OTHER THREE VARIETIES*

Variety

PL.4NT Height

No. Leaves

Leaf Length

Fi-MP

Potencef

Fi-MP

Potence

Fi-MP

Potence

Olson 68

34753

in.

+0.7
+9.1

+2.6
+3.8
+4.0

+0.10

+ 1.26
+0.46
+0.68
+0.62

-3.2
+0.1
-0.9
-1.1
-1.2

-1.62
+0.09
-0.87
-0.53
-0.73

in.

+0.9
+0.3
+0.8
+ 1.1
+0.8

+0.43
+0.33

Tall . . .

+0.63

Texana

+0.97

Average ....

+0.59

* 1947 data. _ _ _

t Potence = Fi-MP/kiPr-Pi).

the parents differed. There can be little doubt that selection for mafty leaves
was practiced in producing the parent types. This is especially true for Olson
68 which was developed from hybrid origin by the late Mr. Otto Olson
(Smith and Bacon, 1941) by selection for plants yielding large amounts of
nicotine. In crosses with Olson 68, the Fi was consistently below the mid-
parent. This result, interpretable as due to an accumulation of a preponder-
ance of recessive genes for the character favored by selection, might be ex-
pected occasionally in naturally self-pollinated plants. Dominance is of less
importance here than in cross-pollinated organisms, since selection is largely
a matter of sorting out superior homozygous combinations.

The 1949 results (Table 10.2) on Olson 68 X 34753 were consistent with
those of 1947 discussed above.

Double Crosses

The three possible double crosses involving all six Fi hybrids of four varie-
ties were grown in 1947 in order to obtain evidence on genie interactions by
comparing experimental results with predicted values. The latter were made

FIXING TRANSGRESSIVE VIGOR IN NICOTIANA RUSTICA

167

in the manner employed in corn breeding, namely Jenkins' method, in which
the average of the four Fi's not contributing to the double cross was used.
These comparisons are shown in Table 10.5 for the three plant characters
studied. The differences between observed means and j)redicted values in the
nine comparisons made were all within the limits required for odds of 19: 1.
It was concluded that the double cross means for plant height, number of
leaves, and leaf length in iV. rustica could be predicted with a high degree of
precision by Jenkins' method. The results indicated that there were no

TABLE 10.5
COMPARISON BETWEEN PREDICTED AND OBSERVED VALUES FOR PLANT
HEIGHT, NUMBER OF LEAVES, AND LEAF LENGTH IN THREE DOU-
BLE CROSSES INVOLVING FOUR VARIETIES OF N. RUSTICA

Double Cross

Observed

Predicted

Difference,
Obs.-Pred.

Plant height {in.) :

{AXB)X{CXD)

{AXC)X{BXD)

{AXD)X{BXC)

41.9 + 2.68
39.6 + 2.68
42.5±2.68

42.6+1.34
45.0 + 1.34
44.0±1.34

-0.7+3.00

-5.4 + 3.00
-1.5±3.00

Average

41.3

43.8

-2.5

No. leaves:

{AXB)X{CXD)

{AXC)X{BXD)

{AXD)X{BXC)

14.0+1.22
12.4±1.22
13.9+1.22

13.8 + 0.61
15.0 + 0.61

14.9 + 0.61

+0.2 + 1.36
-2.6 + 1.36
-1.0 + 1.36

Average

13.4

14.6

-1.2

Leaf length {in.) :

{AXB)X{CXD)

{AXC)X{BXD)

{AXD)X{BXC)

8.8 + 0.89
9.8 + 0.89
9.7 + 0.89

9.3 + 0.44

9.2 + 0.44

9.3 + 0.44

-0.5 + 0.99
+0.6 + 0.99
+0.4±0.99

Average

9.4

9.3

+0.1

A , B, C, D represent the parent varieties as shown in Table 10.1.

marked interactions between the genes or gene sets from the four varieties
when combined in a variety of associations.

To illustrate this point, let us assume that each parent is homozygous for
a different allele at each of two independent loci so that A = XX YY, B =
X'X'Fip, C = X^XWW\ and D = X^XWW^. The Fi's represent six dif-
ferent combinations of these alleles. Each double cross contains all four alleles
of each locus in four particular combinations. For example, the i)opulation
{AXB)XiCX D) is 1/4A'X2 _^ \/^xX' + \/^X^X- + \/AX'XHoi the X
locus and 1/4FF2 + 1/477^ + 1/47^72 + \/4:Y^YHor the Y locus. Sixteen
different combinations of alleles at the two loci are possible in this double
cross. Accurate prediction of the double cross value on the basis of only four
of these combinations, namely: Fi's AXC, AXD, BXC, and B X D,

168

HAROLD H. SMITH

indicates that the other 12 possible combinations do not introduce any sig-
nificant non-additive effects.

Another indication that epistatic effects were unimportant in the in-
heritance of plant height, leaf number, and leaf length was afforded by the
evidence that the means of the double crosses did not differ significantly from
each other (Table 10.1).

The average variance of the double crosses was greater than that of the
parents or Fi's (Table 10.1), as would be expected from segregation.

Partitioning Phenotypic Variance, Heritability, and
Number of Effective Factors

Estimates of the magnitude of the non-heritable variation (0%), in popu-
lations involving Olson 68 and 34753 (1949 data), were obtained by taking

TABLE 10.6

ESTIMATES OF COMPONENTS OF \\\RIABILITY, NUMBER OF EFFECTIVE
FACTORS (Ai), HERITABILITY, AND GAIN FOR PLANT HEIGHT, LEAF
LENGTH, AND NUMBER OF NODES IN THE X. RUSTICA CROSS OLSON
68 X BRASILIA, STRAIN 34753*

Character

0

0

"l

A'l

Herit-
ability
Per
Cent

Gain

Plant height

Leaf length

Node number. . .

25.76 + 15.3
0.71+ 0.45
7.38+ 4.38

67.32 + 53.5
1.04+ 1.05
8.16 + 13.00

113.20 + 71.3
0.22+ 0.69
2.20+ 8.11

0.81
1.38
0.83

54.9
11.2
12.4

1.74
0.91

2.42

* 1949 data.

an average of the total within plot variance of the non-segregating families —
Pi, P2, and Fi. As shown in Table 10.6, the values obtained were 25.76 for
plant height, 0.71 for leaf length, and 7.38 for number of nodes.

The following symbols are used for the components of heritable variance
(total phenotypic minus environmental) : alb = variance depending on addi-
tive gene effects, a|) = variance depending on dominance. The heritable
variance of the F2 was calculated and equated to: 1/2o-g + l/4o-|). The
pooled heritable variance of the two first backcrosses was equated to
l/2crG + l/2o-z). Solving for 0%, the values obtained were 67.32 for plant
height, 1.04 for leaf length, and 8.16 for number of nodes. Values for 0%,
as calculated by substitution, were 113.20 for plant height, 0.22 for leaf
length, and 2.20 for number of nodes.

In view of the influence on node number of a differential interaction of the
two parental genotypes with environment, an additional way of approach-
ing an analysis of the data on this character was tried. If a simple relation

FIXING TRANSGRESSIVE VIGOR IN NICOTIANA RUSTICA 169

between the environmental variances of the Pi, P2, and Fi is assumed, so
that a% of the Fi = 1/2((t% of Pi + al of P2), then a% of the Fi = 6.78.
The environmental variance of Bi may then be equated to 1/2 (variance of
Pi+ variance of Fi), which is 5.12. By a similar relation, the environmental
variance of B2 is equal to 8.44. The pooled heritable variance of Bi + B2, i.e.,
l/2ah + 1/2(7?,, may be equated to: (W.49 - 5.12) + (9.45 - 8.44). This
gave 6.38. The heritable variance of the F2, i.e., l/2ao + l/4o-i), may be
equated to (10.52 — 6.78). This gave 3.74. Solving: <t% = 10.56 and <to =
2.20. The former, (xjj, has a somewhat larger value than that obtained by
the original analysis (8.16, Table 10. 6j; the latter, cc, is the same.

Heritability of a character was estimated as the ratio, expressed in per
cent, of the variance component due to additive, fixable gene effects (aa) to
the sum, <r% -\- a]) -\- a%. Heritability of plant height was calculated to
be 54.9 per cent, of leaf length 11.5 per cent, and of node number 12.4 per
cent.

Estimates of the number of effective factors (Ki) were made on the as-
sumptions inherent in the equation A'l = (Pi — P^^l^a'a. The values ob-
tained (Table 10. 6 j were 0.81 for plant height, 1.38 for leaf length, and 0.83
for number of nodes. These estimates were undoubtedly too low, due in part
to non-isodirectional distributions of + and — genes in the parents. Ex-
perimental evidence of non-isodirectional distribution was afforded by the
fixing of transgressive characteristics in inbred selections following hybridiza-
tion between varieties. Some -\- genes were contributed by each parent, and
consequently could not have been concentrated in one. Linkage in coupling
phase and/or differences in magnitude of effect of the individual genes or
gene blocks might also have contributed to the low estimates of the number
of effective factors.

In the absence of data on Fa's, biparental progenies, and double back-
crosses (Mather, 1949), the errors of the estimates of 0%, a\,, and 0% for each
character were computed as follows. From the eight replications, four means
were calculated by grouping replications 1 and 2, 3 and 4, 5 and 6, and 7
and 8. The standard error of the four independent means was then obtained
(Table 10.6). These errors are maximum estimates since there was a pro-
nounced gradient of environmental effects from replication 1 to replication 8.

Mather (1949) is in the process of making an extensive biometrical genetic
analysis of plant height in a Xicotiana rustica cross, and it was of interest to
compare his published results with corresponding statistics presented in this
study. From his data so far reported, the average values (mean of 1946 and
1947) for components of variance for plant height are: 9.30 for 0%, 9.25 for
(To, and 18.05 for (7%. The heritability calculated from these estimates is 44.1
per cent. The results reported in this discussion are similar in that heritabil-
ity is high and <!% has about twice the value of a]}.

170 HAROLD H. SMITH

Results of Selection

The result of selection for tall plants with many, large leaves can be seen
by comparing the means of Al with those of the parental and hybrid genera-
tions in Tables 10.1 and 10.2.

From the 1947 data it is evident that in the F4 generation of selection Al a
significant increase had been obtained over the parents and Fi's in plant
height and green weight. This was accompanied by a lengthening in time
required to reach maturity. With regard to this latter character, it was noted
that the average time for reaching maturity in five of the six Fi's was later
than the average of their respective parents. This is contrary to the usual
result in first generation hybrids of certain other plants, as maize and toma-
toes; and, where early maturity is an important economic character, would
generally not be considered a manifestation of hybrid vigor, at least in a
"beneficial" sense.

The number of leaves in selection Al was significantly higher (P < .05)
than in any of the Fi's, and all but the most vigorous parent, Olson 68. Leaves
of the selection were shorter than the parent with the longest leaves (Olson
68), not significantly different from the three Fi's that involved this parent,
and longer than in the other three parents and three Fi's.

The 1949 data (Table 10.2) corroborated the 1947 results. There was a
significant increase (P ^ .01) in plant height and in number of nodes over
the two main parents and their Fi. Number of nodes, rather than of leaves,
was used since it is a more reliable criterion of the same character. As in
1947, there was a less marked effect of selection on leaf length, though there
appeared to be an increase in Al from the F4 to the Fe. For this character
the selection was superior to 34753 and the Fi, but not significantly different
from Olson 68, although a close approach to significance at the 5 per cent
level of probability was reached.

The total within plot variances of selection AKFe) for plant height, num-
ber of nodes, and leaf length were in no case significantly higher than for the
more variable parent. It was deduced, therefore, that the inbred selection
had reached relative homozygosity.

The general conclusions were that an inbred selection had been produced
which had increased plant height, more nodes, heavier green weight, and a
longer growth period than any parent or Fi. Length of leaf had been main-
tained at least at the level of the best parent variety.

It was also noted, though no quantitative data were taken, that selection
Al had markedly less vigorous sticker growth at topping time than any of the
other varieties or hybrids. This is an important agronomic character.

Heritability and Gain
One of the objectives in conducting these experiments was to attempt to
determine to what extent the progress realized in actual selection experi-

FIXING TRANSGRESSIVE VIGOR IN NICOTIANA RUSTICA 171

ments could be related to the heritability of a character as determined from
F2 and first backcross data.

Results on the three main characters studied were similar in that there
was no indication of complex genie interactions, and that estimates of the
number of effective factors were low and of the same order of magnitude in
each. If we wish to assume that the "reach" or selection differential (in terms
of standard deviations) was the same for each character, and this is approxi-
mately correct though exact records on this point are lacking, then the gain (in
terms of standard deviations) due to selection should be roughly pro])ortional
to the heritability. The gain was calculated as the difference between the
mean of selection Al and the mid-parent value, divided by the standard
deviation of the F2 (1949 data. Table 10.2).

The relationships between heritabilities and gains can be observed by com-
paring the last two columns in Table 10.6. With regard to plant height and
leaf length, both heritability and gain are higher in the former character;
though the gain is less in plant height than would have been anticipated from
the relative heritabilities. Some possible explanations for this latter result
could be that the selection differential for plant height was lower than for leaf
length, that there was a relatively more rapid reduction in heritability, or
that an approach to a physiological limit for tallness was made.

The gain in node number is disproportionately high in relation to its
heritability. Some possible explanations for this result could be that the
selection differential was higher, that there was a genetic correlation with
plant height, or that the selected character was determined by a preponder-
ance of recessive genes (see Fi result), and individual plants selected for high
node number were largely homozygous for recessive+genes.

DISCUSSION

The experimental results have shown that first generation crosses among
different varieties of Nicotiana riistica exhibit different degrees of character
expression ranging from the smaller parent value to above the larger parent.
By selection and inbreeding it was possible to develop an essentially true-
breeding improved line which exceeded the best Pi or Fi in most character-
istics measured.

This same type of result has also been obtained in our experience with the
commercial species, N. tahacum, and it may be generally characteristic of
self-fertilized plants, as, e.g., Phaseolus ■vulgaris (Malinowski, 1928), soy-
beans (Veatch, 1930), and Galeopsis (Muntzing, 1930).

Crossbreeding
There have been relatively few fundamental changes in the standard
domestic varieties of N. tabacum over a long period of years, except for recent
development of types resistant to destructive diseases (Garner, 1946).

172 HAROLD H. SMITH

Houser (1911) originally suggested the use of first generation intervarietal
tobacco hybrids on a commercial scale to increase yields. He presented breed-
ing results on cigar filler types, dating back to 1903, in which the hybrids
outyielded the parent types by as much as 57 per cent. Plant breeders in
various tobacco-growing areas of the world have observed hybrid vigor
among first generation hybrids of commercial varieties (Ashton, 1946), and
have suggested its use in practice to increase production. Currently, con-
sideration is being given to improving the yield of flue-cured varieties by
this method (Patel et al., 1949).

The results of Hayes (1912), Hayes, East, and Beinhart (1913), and East
and Hayes (1912) showed that by intervarietal hybridization, selection, and
inbreeding the number of leaves, an important factor in yield of tobacco,
could be fixed at a level exceeding the parents or Fi. Regarding the use of
Fi hybrids on a commercial scale, they stated (Hayes, East, and Beinhart,
1913),

While it is doubtless true that by this method the yield could be somewhat increased,
the yield factor, for cigar wrapper types at least, is only of secondary importance com-
pared with quality. Because of the great importance of quality it seems much more reason-
able to suppose that further advance can be made by the production of fixed types which in
themselves contain desirable growth factors, such as size, shape, position, uniformity, vena-
tion, and number of leaves, together with that complex of conditions which goes to make
up quaHty, than by any other method.

The problem of producing higher yielding varieties of N. tabacum with
acceptable quality characteristics of the cured leaf remains today. Kosmo-
demjjanskii (1941) bred four families from the cross Dubec 44XTrebizond
1272, two Russian varieties of .Y. tabacum, which, he reported, were uniform
for morphological characters and flavor and maintained transgression in
plant height and number of leaves to the F; generation.

While first generation hybrids between selected parents may be of use as
a temporary measure to improve self- fertilizing crop plants, it would appear,
in so far as can be generalized from the results on Nicotiana, that production
of fixed types with favorable transgressive characteristics offers a better long-
time solution. Within any one type of tobacco, such as flue-cured, there are
currently available a number of high quality inbred varieties which, though
of similar phenotype, may be expected to differ by genes of a multifactorial
system affecting size characteristics (Emerson and Smith, 1950). Selections
from intervarietal crosses may be expected, therefore, to yield fixed types of
increased size without presenting undue difficulties to the breeder attempt-
ing to maintain quality.

In order to discuss the hereditary basis for experimental results on hetero-
sis and inbreeding, current concepts of the genetic and evolutionary mecha-
nisms involved are briefly presented. In the evolution of naturally crossbred
organisms, mutation and selection result in the accumulation of dominant
favorable genes, hidden deleterious recessives, and alleles or complexes of

FIXING TRANSGRESSIVE VIGOR IN NICOTIANA RUSTICA 173

linked polygenes which give heterotic effects as helerozygoles. Heterosis is
explained genetically as due to the accumulated effect of the favorable domi-
nants and/or coadapted heterozygous combinations. It is an adaptive evolu-
tionary phenomenon (Dobzhansky, 1950).

Selfing

In naturally selfed populations there are accumulated, for the most part,
favorable genes that are either dominant, recessive, or lacking in strong
allelic interactions. Dominance is of little evolutionary significance, and
hence a preponderance of favorable dominant genes is not to be expected.
Furthermore, there would ordinarily be no adaptive significance to favorable
heterozygous combinations. One possible exception is suggested by Brieger's
(1950) demonstration that "if survival values for both homozygotes should
be below 0.5 (compared to the heterozygote value of 1.0) in selfed populations,
a final equilibrium is reached with all three genotypes remaining in the
population." Such a condition might have adaptive value in maintaining
variability in selfed organisms. Hybrid vigor in self-pollinated plants, in
view of the above considerations, is a chance manifestation, an "evolutionary
accident" causing luxuriant growth (Dobzhansky, 1950), and not an adap-
tive product of mutation and selection.

However, from published data on crosses within selfed species of culti-
vated plants, it appears that hybrid vigor is of frequent rather than chance
occurrence. Reported results with flax (Carnahan, 1947), wheat (Harrington,
1944), barley (Immer, 1941), tomatoes (Larson and Currance, 1944), egg-
plants (Odland and Noll, 1948), and soybeans (Weiss, Weber, and Kalton,
1947) all demonstrated that hybrid vigor is characteristic of Fi's. If these
data constitute a representative sample, then, although hybrid vigor is an
evolutionary accident in naturally selfed species, it is not a genetical accident.

The result may be interpreted genetically as follows: Selfed species are
purged of deleterious genes by selection. Different varieties within the
species have accumulated different alleles all of which control "non-defec-
tive," slightly different physiological reactions. The combination of divergent
alleles in heterozygous condition may, more frequently than not, act as East
has suggested in a complementary manner to produce a more efficient physio-
logical condition. This is expressed phenotypically by the hybrid manifesting
more vigorous growth than midway between the homozygotes. Subsequent
selection and inbreeding, however, would permit an accumulation of the most
favorable alleles or gene complexes in the homozygous condition.

As a simplified schematic example, let us assume that two varieties. Pi
and P2, differ by three alleles or linked polygene complexes: X' is dominant
and favorable for vigorous growth, F^ is a favorable recessive, and at the Z
locus the product of the heterozygous condition is above the mean of the

174 HAROLD H. SMITH

homozygotes. The composition of parents, Fi, and selected inbred is shown
below with arbitrary "size" values assigned to each.

Pi = XKX' (4) + y V (4) +Z'Z' (2) =10

Po = X'X^ (2) + Y^V (2) -\-Z'-Z-' (6) =10

Fi = X'X' (4) + I'l ¥' (2) ^Z'Z- (5) =11

sel. = X'X' (4) + Y'V (4) -^Z-Z'- (6) =14

Although the difficulty in selecting superior inbreds would become
greater with increasing numbers of effective segregating units, the following
advantages of selfed over crossbred systems would enhance the opportunity
for success: (1) lack of deleterious recessives, (2) less preponderance of
dominant favorable alleles, (3) homozygous pairs of alleles are superior, as a
result of an adaptive evolutionary process, to heterozygous combinations.
Naturally inbred organisms are products of historical evolutionary processes
in which harmonious systems of homozygous loci have been selected to
attain optimum adaptation. These considerations favor the expectancy
and practicability of obtaining maximum advance through selection and
inbreeding with self-fertilized organisms.

SUMMARY

There were two general purposes in conducting these experiments: First,
to demonstrate that by selection following intervarietal hybridization in a
self-fertilized organism, inbreds could be produced which transgressed the
character expression in parents and Fi; secondly, to investigate the relation
between estimated heritability and the actual results of selection.

An inbred selection of Nicotiana rustica which transgressed the Pi and Fi
characteristics in plant height, node number, and leaf length was obtained.
The heritabilities for these three characters were calculated to be 54.9 per
cent, 12.4 per cent, and 11.2 per cent, respectively. The gains (in terms of
standard deviations) due to selection were 1.74, 2.42, and 0.91, respectively.
Some possible explanations for the lack of direct proportionality between
heritability and gain were discussed.

The number of effective segregating factors for each of the three characters
studied was estimated to be of the same order of magnitude and relatively
few. Non-isodirectional distribution of -|- and — genes in the parent varieties
contributed to an underestimation of this number.

Non-allelic interactions were apparently not an important source of
variation, as indicated by scaling tests and evidence from double cross means.

Reasons for expecting greater advances by selection and inbreeding, as
contrasted to the use of first generation hybrids, in naturally self-fertilizing
genetic systems were reviewed.

PAUL C. MANGELSDORF

Harvard University

Chapter 1 1

Hybridization in
the Evolution of Moize

All varieties and races of maize so far studied prove upon inbreeding to con-
tain numerous heterozygous loci, and all respond to inbreeding with a marked
decline in vigor and productiveness. Since contemporary maize is both
heterozygous and heterotic, it is probable that the factors which have been
responsible for bringing about the present conditions are also factors which
have played an important, if not the principal role, in the evolution of maize.

All of the steps involved in the evolution of maize are not yet known.
Archaeological remains have told us something of the early stages of maize
under domestication, and we can draw additional inferences about its original
nature from its present-day characteristics. Our knowledge of the nature and
extent of its present variation, although far from complete, is already sub-
stantial and is growing rapidly. By extrapolating forward from ancient
maize, and backward from present-day maize, w^e can make reasonably valid
guesses about some of the intermediate stages and about some of the evolu-
tionary steps which have occurred in its history.

The earliest known archaeological remains of maize, as well as the best
evidence of an evolutionary sequence in this species, occur in the archaeo-
logical vegetal remains found in Bat Cave in New Mexico in 1948. This ma-
terial which covers a period of approximately three thousand years (from
about 2000 B.C. to a.d. 1000) has been described by Mangelsdorf and Smith
(1949). It reveals three important things: (1) that primitive maize was both
a small-eared pop corn and a form of pod corn; (2) that there was an intro-
gression of teosinte into maize about midway in the sequence; (3) that there
was an enormous increase in the range of variation during the period of ap-
proximately three thousand years resulting from teosinte introgression and
interracial hybridization.

175

176 PAUL C. MANGELSDORF

INTERRACIAL HYBRIDIZATION IN MAIZE

For additional evidence on interracial hybridization in maize we may
turn to existing races of maize. Among these the Mexican races are of par-
ticular interest and significance, not because maize necessarily originated in
Mexico, since there is considerable evidence that it did not, but because
Mexico is a country where primitive races, which in other places are to be
found primarily as archaeological remains, still exist as living entities. It
is possible in Mexico to find all stages between ancient primitive races and
modern highly-developed agricultural races. One has only to place these
racial entities in their proper sequence in order to have at least the outline
of an evolutionary history.

Wellhausen et al. (1951) have recently made a comprehensive study of the
races of maize of Mexico. They recognize twenty-five distinct races as well
as several additional entities which are still somewhat poorly defined, but
some of which may later be described as races. They divide the known races
into four major groups as follows:

Group No. Races

1. Ancient Indigenous 4

2. Pre-Columbian Exotic 4

3. Prehistoric Mestizos 13

4. Modern Incipient 4

Origin of Mexican Races of Maize

Ancient Indigenous races are those which are believed to have arisen in
Mexico from the primitive pod-pop corn similar to that whose remains were
found in Bat Cave in New Mexico. The races in this group are called in-
digenous not because they necessarily had their primary origin in Mexico,
but because they are thought to be the product of indigenous differentiation
from a remote common ancestor. The differentiation is assumed to have re-
sulted from independent development in different localities and environ-
ments with hybridization playing little if any part.

Four races of the Ancient Indigenous group — Palomero Toluqueno, Arro-
cillo Amarillo, Chapalote, and Nal-tel — are recognized. All of these, like their
primitive ancestor, are pop corn. Two of the four — Chapalote and Nal-tel —
are forms of pod corn. All have small ears, and all are relatively early in
maturity.

Pre-Columbian Exotic races are those which are believed to have been
introduced into Mexico from Central or South America before 1492. Four of
these races — Cacahuazintle, Harinoso de Ocho, Oloton, and Maiz Dulce —
are recognized. The evidence for their antiquity and exoticism derives prin-
cipally from two sources: all have South American counterparts; all except
Maiz Dulce have been parents of hybrid races, some of which are them-
selves relatively ancient.

HYBRIDIZATION IN THE EVOLUTION OF MAIZE 177

Prehistoric Mestizos,^ thirteen in number, are races which are believed to
have arisen through hybridization between Ancient Indigenous races and
Pre-Columbian races and hybridization of both with a new entity, teosinte.
The term })rehistoric rather than pre-Columbian is used for this grouj) be-
cause, although all are prehistoric in the sense that there is no historical evi-
dence of their origin, it is not certain that all are pre-Columbian.

Modern Incipient races are those which have come into existence in the
post-Columbian period. These races, of which four are recognized, have not
yet reached a state of genetic equilibrium. They are recognizable entities but
are still changing.

The seventeen races comprising the two last groups all appear to be prod-
ucts of hybridization, either between races in the first two groups, or between
these races and teosinte. In several cases, secondary and even tertiary hy-
bridization seems to have occurred.

That a race is the product of previous hybridization seems highly prob-
able when the following four kinds of evidence are available.

1. The race is intermediate between the two putative parents in a large
number of characteristics.

2. The putative parents still exist and have geographical distributions
which make such hybridization possible and plausible.

3. Inbreeding of the suspected hybrid race yields segregates which ap-
proach in their characteristics one or the other of the two putative parents —
in some cases both.

4. A population quite similar to the race in question can be synthesized
by hybridizing the two putative parents.

Wellhausen el al. (1951) have presented all four kinds of evidence for the
hybrid origin of a number of the present-day Mexican races. They have pre-
sented similar but less complete evidence for the remainder.

The variety Conico, for example, which is the most common race in the
Valley of Mexico, is clearly the product of hybridizing the ancient Palomero
Toluqueno with the exotic Cacahuazintle. Conico is intermediate between
these two races in many characteristics. The two putative ancestral races still
are found in isolated localities in the Valley of Mexico. The race is interme-
diate in its characteristics between the two suspected parents. Inbreeding
yields segregates which almost duplicate in their characteristics one of the
parents — Palomero Toluqueno. Segregates approaching the other suspected
parent, Cacahuazintle, also result from inbreeding but this parent is never
exactly duplicated. Obviously the race has become something more com-
plex than a mixture of equal parts of two earlier races. Nevertheless the
crossing of Palomero Toluqueno and Cacahuazintle still produces a hybrid
which in many respects is scarcely distinguishable from the suspected hybrid
race. The data in Table 11.1 show that Conico is intermediate between Palo-

1. Mestizo is the Latin-American term for a racial hybrid.

178

PAUL C. MANGELSDORF

mero Toluqueno and Cacahuazintle in a large number of characteristics. They
also show how closely a recently-made hybrid of these two ancient races re-
sembles the suspected hybrid race, Conico. Ears of the three races and the
hybrid are illustrated in Figure 11.1.

The hybrid race, Conico, has in turn been the ancestor of still more complex
hybrid races. A Modern Incipient race, Chalqueho, which has originated in his-
torical times in the vicinity of the village of Chalco in the Valley of Mexico,

TABLE 11.1
COMPARISON OF CONICO WITH ITS PUTATIVE PARENTS*

Characters

Ears and plants:

Ear diameter, mm

No. rows grain

Width kernels, mm

Thickness kernels, mm

Diameter peduncle, mm

Length ear, cm

Height plant, cm

Tillering index

Pilosity score

Internal ear characters:

Ear diameter, mm

Cob diameter, mm

Rachis diameter, mm

Length kernels, mm

Estimated rachilla length, mm

Cob/rachis index

Glume/kernel index

Rachilla/kernel index

Pedicel hairs score

Rachis flap score

Races

Palomero
Toluqueno

37.1
21.8
4.6
2.8
8.0
9.8
175
.26
3

34.0
19.5
10.4
11.4
.4
1.88

.40
.04

0
0

Fi
Hybrid

Conico

45.2
18.6
6.8
3.6
9.2
11.8
200
.35
4

45.1

15.7

7.4

3.9

9.8

12.6

193

.22
3-4

42.4

19.0

9.6

14.8
1.6
1.98
.32
.11
2-4

2-3

Cacahua-
zintle

53.2
16.2
9.8
5.3
10.6
14.7
210
.39
4

47.0
27.7
11.7
14.0

3.6

2.37
.57
.26

4

3

* After Wellhausen el al.

is undoubtedly the product of hybridizing Conico with Tuxpeiio, a pro-
ductive lowland race of the Prehistoric Mestizo group. Since Tuxpeno is
itself a hybrid, the postulated pedigree for Chalqueno, which is shown in
Figure 11.2, becomes quite complex.

In the pedigree of Tuxpeno a distinction has been made (by employing
different styles of type) between the facts which are well-established and
those which are largely based upon inference. There is little doubt that
Conico is a hybrid of Palomero Toluqueno and Cacahuazintle, or that Chal-
queno is a hybrid of Conico and Tuxpeno. There is little doubt that Tuxpeno
is a hybrid derivative of Tepecintle, but it is not certain that the other par-
ent is Olotillo, although this is the best guess which can be made with the

Fig. 11.1 — Ears of the Mexican maize races Palomero Toluqueno, Conico, and Cacahua-
cintle. Conico is intermediate between the two other races and is thought to be the product

of their hybridization.

CHALQUENO

CONICO

PALOMERO TOLUQUENO

"•CACAHUACINTLE

TUXPENO

OLOTILLO

^TEPECINTLE

HARINOSO FLEXIBLE

TEOSINTE

HARINOSO DE GUATEMALA

^TEOSINTE

Fig. 11.2— The postulated geneology of the Mexican race Chalqueno. Parts of the geneal-
ogy not well established by experimental evidence are shown in Italics.

180 PAUL C. MANGELSDORF

evidence now at hand. That Olotillo and Tepecintle are both hybrid races
involving teosinte is even more difficult to prove, although data on chromo-
some knobs presented by Wellhausen et al. tend to substantiate such a con-
clusion.

There is at least no doubt that interracial hybridization has been an im-
portant factor in the evolution of maize in Mexico. Has this hybridization
produced heterosis, or has it merely resulted in Mendelian recombination?
The extent to which the suspected hybrid races remain intermediate be-
tween the two putative parents suggests that natural selection (operating in
a man-made environment) has tended to preserve the heterozygote and to
eliminate the segregates which approach homozygosity. It is at least certain
that the hybrid races are intermediate between their putative parents in
their characteristics to a remarkable degree and that they are highly hetero-
zygous. Even in the absence of natural selection favoring the more heterozy-
gous individuals, there would seem to be a tendency for repeated interracial
hybridization to create an ever-increasing degree of heterosis. This is the
consequence of the fact that maize is a highly cross-pollinated plant, and
that heterozygosity does not diminish after the Fo in cross-fertilized popula-
tions in which mating is random.

Wright (1922) has suggested that the vigor and productiveness of an Fo
population falls below that of the Fi by an amount equal to l/« of the dif-
ference between the production of the Fi and the average production of the
parental stock, where n is the number of inbred strains which enter into the
ancestry of a hybrid. The formula is also applicable to hybrids in which the
parental stocks are not inbred lines, but are stable open-pollinated varieties
in which random mating does not diminish vigor. It is, of course, not ap-
plicable to hybrids of single crosses which are themselves subject to dimin-
ished vigor as the result of random mating.

Hybrid Vigor in Advanced Generations

The rate at which hybrid vigor diminishes in a population after the F2 gen-
eration is related to the proportion of outcrossing. This is true whether hybrid
vigor depends upon heterozygosity or upon the cumulative action of dominant
genes, and irrespective of the number of genes involved and the degree of
linkage. With complete selffiig the amount of hybrid vigor retained is halved
in each succeeding generation. With complete outcrossing the amount of
hybrid vigor falls to one-half in the F2 and thereafter remains constant. With
a mixture of selfing and outcrossing an intermediate result is to be expected.
This can be calculated from the following formula presented by Stephens

(1950):

h = l[{\-k)h'-Vk].

In this formula h is the proportion of Fi vigor retained in the current gen-
eration, // is the proportion retained in the preceding generation, and k is

HYBRIDIZATION IN THE EVOLUTION OF MAIZE 181

the proportion of outcrossing. The formula is based upon the assumption that
gene action is, on the average, additive.

It is obvious (according to this formula) that the percentage of hybrid
vigor retained in later generations of a cross will approach but never fall be-
low kjl. Since the value of k in the case of maize lies usually between .9 and
1.0, it is apparent that the amount of hybrid vigor retained in later genera-
tions of maize crosses will (with random mating) seldom fall below the
one-half, which is characteristic of the Fs-

There are experimental data which tend to show that advanced genera-
tions of maize crosses behave approximately as would be expected from the
formulae of Wright and Stephens.

Kiesselbach (1930) compared the Fi,F2,andF3of 21 single crosses with the
parental inbred lines. The average yield of the inbreds was 24.0 bushels. The
average yield of the Fi was 57.0 bushels. The theoretical yield of the Fo is
40.5 bushels. The actual yield was 38.4 bushels which does not differ signifi-
cantly from the theoretical. The yield of the F3 was 37.8 bushels which is
almost identical to the F2 yield.

Neal (1935) compared the yield in Fi and F2 of 10 single crosses, 4 three-
way crosses, and 2 double crosses. The theoretical reduction in yield be-
tween the Fi and F2 in these three groups (based upon Wright's formula)
should have been 31.1 per cent, 21.0 per cent, and 15.2 per cent respectively.
The actual reduction was 29.5 per cent, 23.4 per cent, and 15.8 per cent. The
agreement could scarcely have been closer.

There is abundant evidence from maize crosses to show that equilibrium
is reached in F2, and that in the absence of selection there is no further reduc-
tion in yield in the F3. Data from the experiments of Kiesselbach (1930),
Neal (1935), and Sprague and Jenkins (1943) are summarized in Table 11.2.

The data so far presented are concerned with crosses of inbred strains. Do
hybrids of open-pollinated varieties behave in the same w^ay? Since open-
pollinated varieties, although not homogeneous, are stable in productiveness
they should behave in crosses in the same way as inbred strains. Data from
advanced generations of topcrosses presented by Wellhausen and Roberts
(1949) indicate that they do. The theoretical yields of the F2 of a topcross
can be computed from a formula suggested by Mangelsdorf (1939).

W'ellhausen and Roberts compared the Fi and F2 generations of 31 dif-
ferent topcrosses each including the open-pollinated variety Urquiza and
two inbred lines of unrelated varieties. The latter were in all cases first-gener-
ation selfs. The mean yield of the 31 Fi hybrids (in terms of percentage of
Urquiza) was 132 per cent. The mean yield of the corresponding 31 F2 hy-
brids was 126 per cent. Since the yields of the first-generation selfed lines
entering into the cross is not known, it is impossible to calculate with pre-
cision the theoretical yield of the F2. However, it is known that good homozy-
gous inbreds yield approximately half as much as open-pollinated varieties

182

PAUL C. MANGELSDORF

(Jones and Mangelsdorf, 1925; Neal, 1935) which means that inbreds, selfed
once and having lost half of their heterozygosity, should yield 75 per cent as
much as the open-pollinated varieties from which they were derived. Assum-
ing that the single-cross combinations involved are at least equal to the top-
cross combinations — 132 per cent — we compute the theoretical F2 yield of
the topcrosses at 117 per cent, which is considerably less than the 126 per
cent actually obtained in the experiments. From the results it can be con-
cluded that hybrid combinations including open-pollinated varieties of maize
retain a considerable proportion of their vigor in advanced generations.

There is also some evidence to indicate that the amount of heterosis which
occurs when open-pollinated varieties are used in hybrid combinations may be

TABLE 11.2

SUMMARY OF EXPERIMENTS DEMONSTRATING EQUILIBRIUM
REACHED IN Fo AND NO ADDITIONAL YIELD REDUC-
TION IN F3 OF MAIZE CROSSES

Class of Hybrids

No.
Hybrids
Tested

Yield in Per Cent of Fi

Investigators

Fi

F2

Fs

Kiesselbach, 1930

Neal, 1935

Single crosses
Single crosses
3-way crosses
Synthetics

21

10

4

5

100
100
100
100

68.0
70.5
76.6
94.3

66.0

75.7

Neal, 1935

Sprague and Jenkins, 1943. . . .

75.8
95.4

Total and averages

40

100

76.9

78.2

considerably higher with Latin-American varieties than with varieties com-
monly grown in the United States. Wellhausen and Roberts report single
topcrosses yielding up to 173 per cent of the open-pollinated variety and
double topcrosses up to 150 per cent. A recent report from the Ministry of
Agriculture of El Salvador (1949) shows four different hybrids between open-
pollinated varieties yielding about 50 per cent more than the average of the
parents. Such increases are not surprising, since the varieties used in the
experiments are quite diverse, much more so than Corn Belt varieties.

All of the data which are available on the yields of advanced generations
of maize crosses, whether the parents be inbred strains or open-pollinated
varieties, tend to show that a substantial part of the hybrid vigor charac-
teristic of the Fi is retained in subsequent generations. Thus maize under
domestication is potentially and no doubt actually a self-improving plant.
Distinct more-or-less stable varieties or races evolve in the isolation of
separated regions. Man brings these varieties or races together under condi-
tions where cross-fertilization is inevitable, and a new hybrid race is born.
Repeated cycles of this series of events inevitably lead to the development,

HYBRIDIZATION IN THE EVOLUTION OF MAIZE 183

without any direct intervention of man, of more productive races. If, in addi-
tion, natural selection favors the heterozygous combinations as it does in
Drosophila (Dobzhansky, 1949), then the retention of hybrid vigor in ad-
vanced generations of maize crosses will be even greater than that indicated
by the experimental results.

INTER-SPECIFIC HYBRIDIZATION OF MAIZE AND TEOSINTE

Superimposed upon these evolutionary mechanisms, at least in Me.xico
and Central America, is a second kind of hybridization which involves the
introgression of teosinte into maize. The importance of this evolutionary
factor would be difficult to overemphasize, for as Wellhausen el al. have
shown all of the more productive races of maize of Mexico show evidence of
past teosinte introgression.

The genetic nature of teosinte need not enter into the present discussion. Dr.
R. G. Reeves and I concluded some years ago that teosinte is not, as many
botanists have supposed, the ancestor of maize, but is instead the progeny
of a cross of maize and Tripsacum. This conclusion has not yet been ex-
perimentally proven, and although there is much evidence to support it, it is
by no means universally accepted by other students of corn's ancestry. For
the purpose of this discussion we need not debate this particular point, since
we need only to recognize that there is a well-defined entity known as teo-
sinte which occurs as a weed in the corn fields of central Mexico and as a wild
plant in Southwestern Mexico, Guatemala, and Honduras.

Teosinte is far more common than formerly supposed. Twenty -five years
ago its occurrence was known in only three or four localities in Mexico. Since
then, numerous additional sites have been described in Mexico and Guate-
mala, and recently a locality in Honduras has been added (Standley, 1950).

Teosinte is the closest relative of maize. It has the same chromosome num-
ber (ten) as maize, and hybridizes easily with it to produce hybrids which are
completely fertile, or almost so. The chromosomes of corn and teosinte are
homologous to the extent that they pair almost completely. Crossing over
between teosinte and corn chromosomes is of the same order as crossing over
in pure corn (Emerson and Beadle, 1932).

Present-Day Hybridization
Since both teosinte and maize are wind-pollinated plants and since they
hybridize easily, it is almost inevitable that hybridization between the two
species should occur in any region where both are growing. There is no doubt
that such hybridization is constantly occurring, and that it has been going
on for many centuries. Fi hybrids of com and teosinte have been collected
in both Mexico and Guatemala. They are especially common in Central
Mexico where teosinte grows as a weed. In 1943, 1 obtained some data on the
extent to which hybridization occurs near the village of Chalco where teosinte
is a common weed in and around the corn fields. In a field where teosinte oc-

184 PAUL C. MANGELSDORF

curred abundantly as a weed permission was obtained from the owner to tag
and harvest 500 consecutive plants. Of the 500 plants tagged, 288 proved to
be maize, 219 were teosinte plants, and 3 were Fi hybrids. Of the 288 ears
classified as maize, 4 showed definite evidence of contamination with teosinte
in earlier generations. In addition, one ear was found in an adjacent row (not
part of the sample of 500 plants) which was identical in its characteristics
with a first backcross to teosinte.

The plants in this field therefore furnished unmistakable evidence of hy-
bridization, both present and during the recent past, between maize and
teosinte. One plant out of every 167 plants in the field was a vigorous Fi hy-
brid shedding abundant pollen which became part of the general pollen mix-
ture in the field. The Fi hybrids themselves, in spite of their vigor, have a low
survival value. The Mexican farmer makes no distinction between teosinte
and the Fi hybrids. Both are left standing in the field when the corn is har-
vested. The pure teosinte disperses its seeds which are enclosed in hard bony
shells, and a new crop of teosinte plants appears the following spring. But
the Fi hybrids have no effective means of seed dispersal, and their seeds, only
partially covered, are quite vulnerable to the ravages of insects and rodents.

Both maize and teosinte are quite successful in occupying distinct niches
in Mexican corn fields. The one, a cultivated plant, depends for its survival
upon its usefulness to man. The other, a weed, depends for survival upon its
well-protected kernels and its efficient method of dispersal. There is no such
niche for the Fi hybrid. It is discarded by man as a cultivated plant, and it
cannot compete with teosinte as a weed. "Finding no friend in either nature
or man" (to use Weatherwax's apt description) the Fi hybrids would be of
no evolutionary significance were it not for the fact that they hybridize with
both parents. Thus there is a constant introgression of teosinte into maize and
of maize into teosinte. In the vicinity of Chalco, in Mexico, this process has
gone on so long and the teosinte has become so maize-like in all of its charac-
ters, that maize and teosinte plants can no longer be distinguished until after
the pistillate inflorescences have developed. The teosinte of Chalco has "ab-
sorbed" the genes for hairy leaf sheaths and red color characteristic of the
maize of the region. Individual plants of teosinte have been found which have
the yellow endosperm color of corn, although teosinte is normally white-
seeded.

The introgression of teosinte into maize in Mexico today is an established
fact. The question is how long this process has been going on and whether it
is strictly a local phenomenon or whether it has affected the maize varieties
of America.

Practically all students of maize and its relatives recognize that teosinte
varieties differ in the degree to which they have become maize-like. Longley
(1941), for example, considers the teosinte of Southern Guatemala to be the
least maize-like and that of Mexico the most maize-like.

HYBRIDIZATION IN THE EVOLUTION OF MAIZE 185

Rogers (1950) has shown that teosinte varieties differ quite markedly in
their genes governing the characteristics in which maize and teosinte differ,
especially characters of the j)istillate inflorescence, tillering habit, and re-
sponse to length of day. He attributes these differences to varieties in the
type and amount of maize germplasm which has become incorjiorated into
teosinte.

If teosinte varieties differ in the amount and kind of maize contamination
which they now contain, it is difficult to escape the conclusion that maize
varieties must likewise differ in the amount of teosinte contamination. There
is little doubt that maize varieties do differ in this respect.

Ancient Hybridization

The prehistoric maize from Bat Cave has already been briefly mentioned.
The earliest Bat Cave corn, dated at approximately 2000 B.C., shows no
evidence whatever of teosinte introgression. Beginning about midway in the
series (which would be about 500 b.c. if the sequence were strictly linear but
which, according to unpublished radio-carbon determinations made by Libby ,
is probably somewhat later) cobs make their appearance which are scarcely
distinguishable from the cobs which we have produced experimentally by
crossing corn and teosinte. \\'eatherwax (1950) regards this evidence of teo-
sinte introgression as far from conclusive, and it is, of course, quite impossible
to prove that a cob a thousand years or more old is the product of hybridiza-
tion of maize and teosinte. Nevertheless, it is true that teosinte introgression
produces certain definite effects upon the cob, as some of us who have studied
the derivatives of teosinte-maize crosses on an extensive scale are well aware.

When it is possible to duplicate almost exactly in experimental cultures
specimens found in nature, the odds are at least somewhat better than even
that the resemblance between the two specimens is more than coincidence.
There is little doubt in my mind that the later Bat Cave corn is the product
of contamination with teosinte. Certainly it differs from the earlier Bat Cave
corn quite strikingly, and it is exactly the way in which it would be expected
to differ if it is the product of teosinte introgression.

Significance of Chromosome Knobs

Mangelsdorf and Reeves (1939) suggested some years ago that the deeply
staining heterochromatic knobs, characteristic of the chromosomes of many
varieties of maize, are the result of the previous hybridization of maize and
teosinte, or more remotely of maize and Tripsacum. There has been much
indirect evidence in support of this hypothesis (especially Mangelsdorf and
Cameron, 1942; Reeves, 1944), and the recent studies of Wellhausen ei al.
on Mexican races of maize provide additional evidence of this nature.
Chromosome knob number in Mexican races is closely correlated with the
characteristics of the races. The four Ancient Indigenous races, assumed to
be relatively pure corn, have an average chromosome knob number of 4.2.

186

PAUL C. MANGELSDORF

The four Pre-Columbian Exotic races, also believed to be relatively free
from contamination, have an average chromosome knob number of 4.3. The
thirteen Prehistoric Mestizos and the four Modern Incipient races (all except
one of which are assumed to involve teosinte introgression) have chromo-
some knob numbers of 7.1 and 8.0, respectively.

It is interesting to note that in races for which hybridization is postulated
the hybrid race, although usually intermediate in chromosome knob number
between its two putative parents, resembles most closely the parent with a
high knob number. For the eleven hybrid races for which chromosome knob
numbers are available, not only for the hybrid races but for the two suspected
parent races, the data (Table 11.3) are as follows: the average of the lower-

TABLE 11.3

CHROMOSOME KNOB NUMBERS OF MEXICAN HY-
BRID RACES OF MAIZE AND OF THEIR
PUTATIVE PARENTS*

Hybrid Race

Tabloncillo

Comiteco

Jala

Zapalote Chico. .
Zapalote Grande

Tuxpeno

Vandeiio

Chalqueno

Celaya

Conico Norteno.
Bolita

Averages . . .

Race

7.6
5.6
7.5
11.7
7.4
6.1
8.1
6.8
8.5
8.0
8.6

7.8

Parents

Lower

low
5.0
5.6
5.5
7.0
6.3
6.1
1.0
6.1
1.0
7.6

5.1

Higher

8.0

.0
.6
.0

,7
.0

7.
7.
9.

11.
9.
7.4
6.1
7.6
8.5

11.7

8.5

* Data from Wellhausen et al.

numbered parent was 5.1 knobs, of the higher-numbered parent, 8.5 knobs,
of the hybrid, 7.8 knobs. The fact that the average knob number in the
hybrid races approaches the average knob number of the higher parents
suggests, perhaps, that natural selection has tended to retain the maximum
amount of teosinte introgression and hence the maximum number of knobs.

The Effects of Hybridizing Maize and Teosinte
There is no doubt that maize and teosinte are hybridizing in Mexico and
Central America today, and there is at least a strong indication that they
have done so in the past. What have been the effects of that hybridization?
One valid way of determining what happens when teosinte introgresses
into maize is to produce such introgression experimentally. This has been
done on an extensive scale by crossing an inbred strain of maize, Texas 4R-3,

HYBRIDIZATION IN THE EVOLUTION OF MAIZE 187

with four varieties of teosinte, and by repeatedly backcrossing (three times
in most instances) the hybrids to the inbred strain, retaining various amounts
of teosinte germplasm through selection. The end result is a series of modified
inbred strains approximately like the original 4R-3 — all relatively isogenic
except that parts of one or more chromosome segments from teosinte have
been substituted for homologous parts from maize.

That the substitution involves chromosome segments or blocks of genes
and not single genes is strongly indicated by the fact that the units have
multiple effects and that there is breakage within the units in some cases,
although in general they are transmitted intact. Their mode of inheritance
and their linkage relations can be determined as though they were single
genes. Yet each of the units affects many if not all of the characters in which
maize and teosinte differ. The block of genes on chromosome 3, for example,
although inherited intact as a single hereditary unit, affects number of ears,
size of ear, number of seeds, size of seeds, number of rows of grain, staminate
spikelets on the ear, and induration of the rachis. In addition it has a con-
cealed effect, discussed later, upon such characters as response to length of
day and the development of single spikelets. The block of genes on chromo-
some 4 has practically the same effects in somewhat greater degree, but this
block shows definite evidence of breakage or crossing over which is of the
order of 30 per cent.

These blocks of genes are not random samples of teosinte germplasm, but
represent definite genie entities which are transmitted from teosinte to maize
in the process of repeated backcrossing. Different varieties of teosinte yield
comparable if not identical blocks of genes, and the same variety of teosinte
in different crosses does likewise. Regardless of the amount of introgression
of maize which teosinte has undergone in its past history, and regardless of
the differentiation which has occurred between varieties of teosinte, there are
still regions in all varieties of teosinte, perhaps near the centromeres, which
have remained "pure" for the original genes.

Effects in Heterozygous Condition
When these blocks of genes are introduced into maize they have profound
effects which differ greatly in the heterozygous and homozygous condition.
Since maize and teosinte represent completely different morphological and
physiological systems (especially from the standpoint of their pistillate in-
florescences and their response to length of day), this substitution, of seg-
ments of chromatin from one species for homologous segments from the
other, represents a drastic interchange of parts comparable, perhaps, to in-
stalling a carburetor or other essential part from one make of car into an-
other. In the Fi hybrid of corn and teosinte where the blocks of genes are
heterozygous, there is no particular functional difficulty. Here the two com-
plete systems are operating simultaneously and the result is a vigorous hy-

188

PAUL C. MANGELSDORF

brid, vegetatively luxuriant, potentially capable of producing great numbers
of seed. Measured solely by total grain yield, the Fi hybrid does not exhibit
heterosis since its grain yield is considerably less than that of corn, but meas-
ured in terms of number of seeds, or number of stalks, or total fodder, the
hybrid certainly exhibits heterosis.

In the modified inbred in which a block of genes from teosinte has been
substituted for a block of genes from maize, the situation is quite different.
There are no functional aberrations so long as the block of genes from teosinte
is heterozygous. Under these circumstances it has very little discernible

Fig. 11.3 — Ears of a teosinte-modified inbred strain 4R-3 which are isogenic except for an

introduced block of genes from chromosome 3 of Florida teosinte. The ear at the left lacks

the block of teosinte genes, the center ear is heterozygous for it, the ear at the right is

homozj'gous for it. Note the high degree of dominance or potence of the maize genes.

HYBRIDIZATION IN THE EVOLUTION OF MAIZE

189

effect. Figures 11.3, 11.4, and 11.5 show ears of corn heterozygous for blocks
of genes from chromosomes 3 and 4 respectively, compared to "pure" corn
in the same progeny. The blocks of genes from corn are much more "potent"
(a term proposed by Wigan, 1944, to describe the integrated dominance
effects of all genes) than the block of genes from teosinte, at least in the
striking characteristics which differentiate the two species. This is in itself a
noteworthy phenomenon since corn is not strongly "dominant" or more po-
tent than teosinte in the Fi hybrid, where both species contribute more or
less equally.

Fig. 11.4 — These ears are the exact counterparts"]<)f those in Figure 11.3 exce])t that the
block of teosinte genes was derived from chromosome 4 of Florida teosinte.

-<

t

•-^i

\ r

m

Fig. 11.5 — When the inbred 4R-3 is crossed with No. 701 the liybrid ear illustrated above
(left) is produced. When a modified strain of 4R-3 (right) which has had three blocks of genes
from Durango teosinte substituted for corresponding maize genes is crossed with No. 701,
the hybrid (center) is much more maize-like than teosinte-like. The hybrid, being multiple-
eared, bears a substantially greater number of seeds than either parent and in one experi-
ment was appreciably more productive.

HYBRIDIZATION IN THE EVOLUTION OF MAIZE 191

The reason for the strong jjotence of maize over teosinte in blocks of
genes introduced from teosinte into maize, is to be found in a phenomenon
termed "antithetical dominance" which has been postulated by Anderson
and Erickson (1941) on theoretical grounds. These writers assumed that in
species hybrids such as that between maize and Tripsacum, the Fi would be
intermediate but that backcrosses to either parent would strongly resemble
the recurrent parent. The basis for this assumption is that the possibilities
for successful recombination of two such different systems is remote.

The conception of antithetical dominance has some relationship to
Richey's opinion (1946) that dominance in some cases is no more than a con-
dition where one allele is capable of doing the entire job, or most of it, while
the other allele merely stands by. According to this interpretation, genes are
not favorable because they are dominant, but are dominant because they are
favorable. They reveal their presence by doing something.

There is, in any case, little doubt that something of the general nature of
antithetical dominance or the kind of dominance postulated by Richey is
involved in the teosinte-maize derivatives. Both teosinte and maize are
about equally potent in the Fi hybrid, but a small amount of teosinte germ-
plasm incorporated into maize in the heterozygous condition is definitely
lacking in potence.

Effects in Homozygous Condition

Since a block of teosinte genes introduced into maize is largely recessive
in its effects when heterozygous, its effects should become much more ap-
parent in the homozygous condition. This is indeed the case. The ear on the
right in Figures 11.3, 11.4, and 11.5 illustrates the effects of one or more
blocks of teosinte genes incorporated in a homozygous condition in the inbred
strain 4R-3.

The combination of traits from corn and teosinte which occurs in these
homozygous teosinte derivatives is characterized by a distinct lack of har-
mony in the development of the pistillate inflorescence. The husks are too
short for the ears, the glumes are too small for the kernels and tend to con-
strict the growing caryopses producing misshapen kernels. The vascular sys-
tem is inadequate for the number of kernels borne on the ear, and there are
many shrunken kernels as well as numerous gaps where no kernels have de-
veloped. Germination of the seeds is often poor, and viability of short dura-
tion. Homozygous combinations of this kind obviously have a low survival
value. Indeed it has been difficult to maintain some of them in artificial
cultures.

These unfavorable effects of teosinte introgression in the homozygous con-
dition may be nothing more than the result of substituting parts of one well-
integrated system for corresponding parts of another. They may, however,
also involve "cryptic structural differentiation" of the kind suggested by

192 PAUL C. MANGELSDORF

Stephens (1950) for species of Gossypium, although the extent of this cannot
be great, otherwise some combinations would be lethal. But whatever the
cause, there is little doubt about the reality of the unfavorable effects.
Therefore, if the repeated hybridization of corn and teosinte which has oc-
curred in the past has had any permanent effect, one of two things or both
must have happened: (1) The undesirable effects of teosinte have become
recessive as the result of natural selection for modifying factors. (2) The
regions of the chromatin involving teosinte genes have been kept heterozy-
gous. There is some evidence that both may have occurred.

There is some evidence, by no means conclusive, that maize varieties of
today have absorbed teosinte germplasm in the past and are now bufered
against the effects of teosinte genes. There is at least no doubt that when the
same variety of teosinte is crossed on a series of maize varieties, considerable
variation is displayed by the Fi hybrids in the potence of the maize parents.

In general, varieties which show some evidence of previous contamination
with teosinte are more likely to produce maize-like Fi hybrids than those
which do not show evidence of such contamination. Corn Belt inbreds as a
class produce the most maize-like Fi of any of stocks tested. Figure 11.6 illus-
trates a case where a South American stock (an inbred strain derived from
the Guarany corn of Paraguay) is less potent in crosses with two varieties of
teosinte than is a North American stock (a genetic tester). I also have ob-
served that blocks of teosinte genes introduced into an inbred strain of
Guarany by repeated backcrossing have a greater effect than these same
blocks introduced into Texas 4R-3 or Minn. A158, both of which seem al-
ready to contain appreciable amounts of teosinte.

If the increased potency of teosinte-contaminated maize proves to be gen-
erally true, then the reason for it is that there has been a selection of modify-
ing factors which have tended to suppress the most unfavorable conspicuous
effects of the teosinte introgression. Otherwise, varieties of maize containing
teosinte germplasm should produce hybrids which are more teosinte-like,
rather than more maize-like, than the average. This is convincingly demon-
strated experimentally by crossing the original inbred 4R-3 and one of its
modified derivatives with the same variety of teosinte (Florida type). The
results are illustrated in Figure 11.7.

The Fi of 4R-3 X teosinte is a typical Fi hybrid, intermediate between
its parents. It has both single and double spikelets and, although the fact is
not revealed by the illustration, it has approximately the same type of re-
sponse to length of day as does maize. In marked contrast, when a derived
strain of 4R-3 (in which a block of teosinte genes on chromosome 3 has been
substituted for a corresponding block of maize genes) is crossed with the
same teosinte, the Fi hybrid is scarcely distinguishable in its pistillate spike
from pure teosinte. Furthermore, it has teosinte's response to length of day.
Plants of this hybrid started in the greenhouse in February did not flower

HYBRIDIZATION IN THE EVOLUTION OF MAIZE

193

until the following October and November. This derivative of a maize-teo-
sinte hybrid, therefore, carries at least two concealed characteristics of teo-
sinte: single spikelets and response to length of day. Genes for these two char-
acters do not express themselves in the derivative itself, but their presence
becomes immediately apparent when the derivative is crossed with teosinte.
The situation is comparable to the concealed genes for hair color and texture

b^

i

, c

%

^

Fig. 11.6 — A North .\merican stock is more potent in crosses with Nobogame teosinte (A)
and Durango teosinte (C) than the Guaranj' corn from Paraguay {B and D). This is at-
tributed to previous introgression of teosinte accompanied by the evolution of modifier

complexes in North American varieties.

in persons who are completely bald. The genes are there but have no oppor-
tunity to express themselves.

Since varieties of maize which appear to be the product of previous teo-
sinte contamination, such as those of the Corn Belt, behave quite differently
in crosses from stocks known to be contaminated, there is at least an indica-
tion that such contamination has become modified through selection acting
upon the modifier complex. More data are obviously needed on this problem.

A second question which arises in considering the effects of the natural
hybridization of com and teosinte is whether there is any mechanism which

194

PAUL C. MANGELSDORF

tends to maintain the maize-teosinte loci in a perpetual state of heterozy-
gosity. It already has been shown that homozygous teosinte genes in the
maize complex are decidedly deleterious. Therefore, if the teosinte genes are

^*"«*W*^iW-

D

Fig. 11.7 — When the inbred 4R-3(^) iscrossed with Florida teosinte (C), thcFi hybrid ears
(B) are maize-like in having four-ranked ears, some double spikelets, and partially naked
seeds. When a teosinte-modified strain of 4R-3 (D) is crossed with Florida teosinte (F),
the Fi hybrid (E) is much more teosinte-like. The spikes are two-ranked, single, and the
seeds are completely enclosed. The teosinte derivative obviously carries "concealed" genes

for these teosinte characteristics.

to survive their deleterious effects, they must be modified through selection
or the genes must be maintained in a more or less heterozygous state. It may
be assumed that the latter mechanism would operate only if heterozygosity
for a group of maize-teosinte genes confers a distinct selective advantage
making the heterozygous combination superior, not only to the homozygous
teosinte genes (as it obviously is) but also to the corresponding homozygous
maize genes.

HYBRIDIZATION IN THE EVOLUTION OF MAIZE 195

Data are available both from my experiments and those of R. G. Reeves
(1950), conducted independently, to indicate that heterozygosity for a block
of teosinte genes does sometimes confer a selective advantage. In 1944, in my
experiments, five Corn Belt inbred strains were crossed with the Texas in-
bred 4R-3, as well as with foar modified strains of 4R-3 in which teosinte
genes had been substituted for maize genes. The four modified strains may be
briefly described as follows:

No. Blocks Teosinte

Strain Genes Variety

Modified 4R-3 Strain A 2 Florida

Modified 4R-3 Strain B 2 Florida

Modified 4R-3 Strain C 3 Durango

Modified 4R-3 Strain D 3 Durango

The Fi hybrids were grown in 1945 in two replications in a modified Latin-
Square yield test. Several hybrids were omitted for lack of sufficient seed.
The results are shown in Table 11.4.

TABLE 11.4

AVERAGE YIELDS IN BUSHELS PER ACRE OF HYBRIDS OF

CORN BELT INBREDS WITH TEXAS 4R-3 AND ITS

TEOSINTE-MODIFIED DERIVATIVES

4R-3 OR Deriv.^tive

Corn Belt Inbreds

K1S5

38-11

L317

701

CC24

4R-3 (check)

108.6

102.6

126.6*

94.2

93.0

85.2

99.0

100.2
87.0

100 2

Modified Strain A

88 8

Modified Strain B

82.8
75.6
57.0

109.8
66.0
71.4

78 6

Modified Strain C

Modified Strain D

97.8
146.4*

92.4
79.8

* Difference probably significant.

Of the 17 hybrids tested, only 3 proved to be better than the correspond-
ing checks in total yield, and in only 2 of these is the difference significant.
Although the data are not extensive, there is some indication that the Corn
Belt inbred strains used in these experiments differ in their ability to "com-
bine" with the teosinte derivatives.

Perhaps more important than total yield, from the standpoint of selective
reproductive advantage, is total number of seeds per plant (Table 11.5).
Here 6 of the 15 hybrids for which data are available were superior to the
checks, 4 of these significantly so.

These results, so far as they go, are in agreement with the recently pub-
lished results of Reeves (1950). Reeves tested 49 modified 4R-3 lines in hy-
brids with a common tester. He found none significantly better than the
check in yield, although several were superior in heat-tolerance. Reeves

196

PAUL C. MANGELSDORF

found, however, that when teosinte germplasm was introduced into another
inbred strain, 127C, the results obtained in the hybrids were somewhat dif-
ferent. In 1946, 6 hybrids out of 25 were better than the check, 3 of them
significantly so. In 1947, 15 hybrids out of 49 were better than the check,
6 of them significantly so. Reeves suggested that the difference between 4R-3
and 127C in their response to teosinte introgression lies in the fact that
4R-3 already contained considerable amounts of teosinte germplasm while
127C does not. The suggestion is supported by differences in the morphologi-
cal characteristics of the two lines.

There was also an indication in Reeves' experiments that the entries with

TABLE 11.5

AVERAGE NUMBERS OF SEEDS PER PLANT IN HYBRIDS OF

CORN BELT INBREDS WITH TEXAS 4R-3 AND ITS

TEOSINTE-MODIFIED DERIVATIVES

Corn Belt Inbreds

4R-3 OR Derivative

K1S5

38-11

L317

701

CC24

4R-3 (check) ....

849
756
937
1419*
770

636

925

1132
1095

1179

Modified Strain A

807

Modified Strain B

809'
573

1107*
746
843

Modified Strain C

Modified Strain D

1696*
1811*

885
864

* Difference probably significant.

teosinte genes made their best showing in 1947, a season of severe drought.

Considering all of the results together it may be concluded that: (c)
blocks of teosinte genes in the heterozygous condition do in some instances
improve the total yield of the plants which contain them ; (b) even more fre-
quently do such blocks of genes increase the total number of seeds produced ;
(c) there is some evidence that the teosinte derivatives impart resistance to
heat and drought to their hybrids.

In those crosses in which the heterozygous combination is superior to
either of the homozygous combinations, a block of maize genes or a block of
teosinte genes, natural selection would undoubtedly favor, at least initially,
the heterozygous combination. If the block of genes were one involving the
region of the centromere where crossing-over is reduced, it is quite possible
that the block of genes would be retained more or less intact for a consider-
able number of generations. The maintenance of heterozygosity through
natural selection also would be promoted if, as in the case of Drosophila
studied by Dobzhansky, one set of genes is superior in adapting the organism
to one kind of environment while the other set contributes to adaptation

HYBRIDIZATION IN THE EVOLUTION OF MAIZE 197

to a wholly different environment which the organism also encounters pe-
riodically.

It cannot be proved that such a situation exists in the case of maize which
has become contaminated with teosinte, but it is quite possible that it does.
For example, human selection when practiced would tend to favor the larger-
seeded, larger-eared individuals with a minimum of teosinte contamination.
Natural selection would favor the individuals with the larger number of
seeds, hence those with an appreciable amount of teosinte contamination.
These two forces operating simultaneously or alternately would tend to per-
petuate the heterozygote. Similarly, if maize germplasm were superior in
seasons of excessive moisture and teosinte germplasm in seasons of drought
(for which there is some evidence), there would be a tendency for natural
selection to perpetuate heterozygous combinations. It cannot be demonstrated
that any of these hypothetical situations actually exist. There is no doubt,
however, that present-day maize is highly heterozygous, and there is more
than a suspicion that repeated hybridization with teosinte has been respon-
sible for part of the heterozygosity.

DISCUSSION

The present-day heterozygosity of maize may involve a variety of differ-
ent factors and forces which have operated during its past history. Two of
these are now reasonably clear: interracial hybridization, and introgression
of teosinte into maize.

When interracial hybridization occurs, hybrid vigor not only manifests
itself in the first generation, but also persists in part through an indefinite
number of subsequent generations. Maize under domestication is, therefore,
potentially a self-improving plant. The evidence from Mexican races of
maize indicates that repeated interracial hybridization has been an extremely
important factor in the evolution of maize in Mexico. There is every reason
to believe that the situation in Mexico, so far as interracial hybridization is
concerned, is typical of other parts of America.

The second factor, introgression of teosinte, which is believed to have
played an important role in the evolution of maize, is not so easily demon-
strated. There is no doubt, however, that teosinte is hybridizing with maize
in Guatemala and Mexico today, or that this hybridization has occurred in
the past. It would be surprising indeed if such hybridization had no effect
upon the evolution of maize. There is every indication that it has had a pro-
found effect. All of the most productive modern agricultural races of maize
in Mexico show evidence of contamination with teosinte, not only in their
external characters, but also in their internal cytological characteristics.

It can be shown experimentally that teosinte germplasm, when introduced
into maize, may sometimes have a beneficial effect when heterozygous, but
is always deleterious when homozygous. Therefore it follows that after maize

198 PAUL C. MANGELSDORF

and teosinte have hybridized, and after there has been an introgression of
teosinte into maize: (1) the teosinte genes must be eliminated or, (2) their
effects must be changed through the accumulation of a new modifier com-
plex, or (3) they must be kept in a heterozygous state. There is evidence, but
not final proof, that both of the two last-named factors have operated during
the evolution of maize. Interracial and interspecific hybridization accom-
panied by sustained heterosis are therefore regarded as two important fac-
tors in the evolution of maize.

SUMMARY

1. Evidence is presented to show that both interracial and interspecific
hybridization, accompanied by heterosis, have been factors in the evolution
of maize.

2. The races of maize of Mexico are cited as an example of interracial hy-
bridization. Of the 25 Mexican races described by Wellhausen et al., 14 are
considered to be the products of interracial hybridization.

3. The hybrid vigor, which occurs when races of maize are crossed, is
capable of persisting in part in subsequent generations. Maize under domesti-
cation is therefore potentially a self-improving plant.

4. Interspecific hybridization of maize and teosinte is occurring in Gua-
temala and Mexico today, and there is evidence — archaeological, morphologi-
cal, and cytological — that it has occurred in the past.

5. Introgression of teosinte into maize in experimental cultures is some-
times beneficial when the teosinte genes are heterozygous, but is always
deleterious when they are homozygous.

6. It, therefore, seems probable that the persistence of teosinte germ-
plasm in races of maize has been accompanied either by development of
modifier complexes which have made the teosinte genes recessive in their
action, or by the maintenance of a continued state of heterozygosity.

7. The possibility that heterozygosity in maize has been preserved by
natural selection as it has been in Drosophila is discussed.

STERLING EMERSON

California Insfifufe of Technology

Chapter 12

Biochemicol Models
of Heterosis in Neurosporo

Some of the things that have been learned about gene controlled reactions
in Neurospora can be used in forming a picture of how individual genes con-
tribute to heterosis. I wish to consider especially those examples which indi-
cate that heterozygosity at a single locus may influence the growth of an
organism to a considerable extent.

It should be noted at the beginning, however, that one is not justified in
assuming that the situations found in Neurospora are necessarily similar to
those occurring in the higher organisms in which heterosis is ordinarily
studied. It may be unwise to assume that any two organisms are essentially
similar. There are special reasons for caution in making comparisons between
Neurospora and higher plants and animals, since the nuclear and chromo-
somal basis for the expression of heterosis is so dissimilar. On the other hand,
there is a considerable accumulation of information about the parts played
in the physiology and biochemistry of Neurospora by individual genes
(Beadle, 1948; Horowitz, 1950) and, with proper caution, we may assume
that some of this information may have rather broad application.

In any haploid organism, such as the ascomycetous fungus Neurospora,
in which there is a single set of genes in each nucleus, such phenomena as
dominance, heterozygosis, and heterosis cannot occur. There is, however,
a condition known as heterocaryosis which permits a loose approximation
to each.

CHARACTERISTICS OF HETEROCARYONS

The plant body of Neurospora can be said to be made up of cells, but they
are very different from the cells of higher plants. In the first place, the cells
contain a large and variable number of nuclei in a common cytoplasm. The
so-called cells themselves are not as discrete as cells are generally supposed

199

200

STERLING EMERSON

to be. The walls between them have perforations which permit both cyto-
plasm and nuclei to move from cell to cell. If all nuclei are identical, their
movement and distribution is probably of minor importance, but if they are
not identical there may be effects of considerable consequence arising from
irregularities in nuclear distribution.

There are two ways in which a mixture of different kinds of nuclei within
a single cell may come about. In the growth resulting from a sexually pro-
duced ascospore, or from a uninucleate asexual microconidium, all nuclei
are directly descended from a single haploid nucleus. Barring mutation, they
should all have the same genetic constitution. After the growth has become

Strain X

Fig. 12.1 — Heterocaryon formation resulting from hyphal fusion (a diagram).

multinucleate, if a mutation should occur in one nucleus, the descendants of
that nucleus would then have a different genetic constitution from the re-
maining nuclei in the common cytoplasm, and a condition of heterocaryosis
would exist. The second way in which heterocaryons arise is from the direct
fusion of branches or hyphae of different strains, with the subsequent in-
termingling of their nuclei. By the latter method, heterocaryons of pre-
determined genetic constitution can be made at will.

The controlled production of heterocaryons is shown diagrammatically in
Figure 12.1. Strain X is represented as having black nuclei to distinguish
them from the nuclei of strain Y, which are pictured as being white. After
fusion between hyphae, nuclei of strain Y may migrate into cells of strain X,
and those of X into Y. It is possible that different hyphal tips, growing from
this common mass of cells, will have different relative numbers of the two
sorts of nuclei, as illustrated by the ratios 1:7, 1:1, and 7:1 in three of

BIOCHEMICAL MODELS OF HETEROSIS IN NEUROSPORA

201

the hy})hal tips. To prove that two kinds of nuclei were present in the same
cells of such heterocaryons, Beadle and Coonradt (1944) cut ofT single hyphal
tips, transferred them to fresh medium, and then identified two kinds of
nuclei in the resulting growth by genetic test.

Where there is freely branching filamentous growth, as in Neurospora, it is
possible for the two types of nuclei in a heterocaryon to become sorted out

Fig. 12.2 — Somatic segregation of dissimilar nuclei in the formation of conidia (a diagram).

purely as a matter of chance, as illustrated in a schematic way in Figure 12.2.
This diagram actually represents an erect fruiting branch, or conidiophore,
on which the asexual spores are born. The conidia of Neurospora have
variable numbers of nuclei, but generally more than one. Dodge (1942)
proved that two kinds of nuclei were present in the same cell of a heterocar-
yon by growing cultures from single conidia, and then showing by genetic
test that some of these cultures had both types of nuclei. In some instances
he was able to distinguish the heterocaryotic and both homocaryotic types
in culture derived from single conidia by their mon:)hological characteristics.
The essential differences between Neurospora and higher organisms with

202 STERLING EMERSON

respect to heterosis result from the points just noted. In a diploid which is
heterozygous for a single gene pair, both alleles are present in the same nu-
cleus and in equal dosage. Whereas in the corresponding haploid heterocar-
yon, the two alleles are present in different nuclei, and the relative propor-
tions of the two alleles vary with the frequencies of the two types of nuclei.
All cells of a diploid heterozygote have the same genetic constitution, but
there can be a considerable variation in genetic constitution in different parts
of a heterocaryotic individual. Interactions between alleles, by which I mean
such things as the expression of dominance, must result from the ability of
genes to act at some distance in heterocaryons, in which there is no possibility
of an intimate association of alleles within a nucleus (Lewis, 1950). It is
considerations such as these that show that dominance and heterosis-like
effects in Neurospora are only approximations to the phenomena as known in
diploid organisms.

HETEROSIS IN HETEROCARYONS

An enhancement of growth, closely simulating heterosis, in heterocaryons
of Neurospora teirasperma was reported by Dodge in 1942. In this paper he
distinguished between heterocaryotic vigor and the hybrid vigor of diploid
organisms along much the same lines as I have just done. He suggested that
the heterocaryotic vigor observed might be the result of complementing
growth factors whose production was controlled by the two types of nuclei
(Robbins, 1950). It was later (Dodge, Schmitt, and Appel, 1945) demon-
strated that genes responsible for enhanced growth segregated and recom-
bined in a normal fashion. These studies showed that genes residing in differ-
ent nuclei, but in a common cytoplasm, can cooperate in establishing condi-
tions favoring rapid growth, and that a condition resembling hybrid vigor
occurs.

Meantime, Beadle and Coonradt (1944) had reported on heterocaryons
between pairs of mutant strains of Neurospora crassa, each of which is unable
to synthesize a particular vitamin or amino acid. Each mutant strain by itself
is unable to grow unless supplied with its specific growth requirement, but
nine heterocaryons involving different combinations of seven mutant strains
grew at rates approximating that of wild type without the addition of growth
factors. The authors conclude that the wild type allele is dominant to the
mutant allele in each of the examples studied.

Beadle and Coonradt note further that in such heterocaryons, in which
there is the opportunity for great diversity in the relative numbers of the
two types of nuclei in different hyphal tips, those tips having the most favor-
able proportions of nuclei should grow most rapidly. Conversely, rapidly
growing hyphae should have the two sorts of nuclei in roughly optimal pro-
portions. In heterocaryons involving pairs of mutant strains, Beadle and
Coonradt found nuclear ratios varying between approximately 1 : 1 and al-
most 20: 1. They interpreted these results to mean that the wild type alleles

BIOCHEMICAL MODELS OF HETEROSIS IN NEUROSPORA 203

of different mutant genes have different degrees of dominance. A strongly
dominant wild type allele will need to be present in relatively few nuclei — say
one in twenty.

A heterocaryon between two mutant strains could grow at the maximum
rate over a large range of nuclear proportions, provided the wild type alleles
concerned were both strongly dominant. A weakly dominant wild type
allele, on the other hand, must be present in a large j)roportion of the nuclei —
say nineteen of twenty — to ensure vigorous growth. Heterocaryons in which
the wild type alleles concerned are both weakly dominant could never result
in vigorous growth, since the two wild type alleles cannot both be present
in excess, one being in one type of nucleus and the other in the remaining
nuclei.

HETEROSIS DUE TO HETEROZYGOSITY AT ONE LOCUS

The heterosis effect in heterocaryons studied by Beadle and Coonradt re-
sults from the mutually complementary nature of the nuclei involved. For
each deleterious mutant allele in one nucleus there is the corresponding
favorable and dominant wild type allele in another. In contrast to these
there are other heterocaryons (briefly reported in Emerson, 1947) in which
the nuclei differ in only one gene, yet which still show the heterosis effect.
Heterocaryons in which some nuclei carry the dominant allele and some the
recessive are superior to homocaryons, all of whose nuclei have the dominant
allele, or all the recessive.

Heterocaryotic Suppression of the Sulfonamide-requiring Character

Most of the heterocaryons of this sort that have been found so far have
involved the so-called sulfonamide-requiring mutant strain. At 35° on mini-
mal medium, this strain makes extremely poor growth, but it does keep
creeping along. After varying lengths of time, it frequently happens that the
growth will change to a rapid vigorous type. Growth curves of six cultures
which have reverted to something approaching wild type growth are shown
in Figure 12.3. When the mycelium had reached the end of the growth tubes,
inocula from the newest growth were introduced into fresh tubes containing
minimal medium, resulting in the growth curves shown in the upper part of
the figure.

From these curves it can be seen that the reverted type of growth usually
persists through a conidial transfer. After the mycelium had reached the end
of the second tube, conidia were removed and used in outcrosses to wild type
to determine the genetic constitution of their nuclei. These tests showed
that each of the six cultures represented in Figure 12.3 was a heterocaryon.
One type of nucleus present in each heterocaryon was identical to those in
the original sulfonamide-requiring strain. The second type of nucleus in each
also carried the sulfonamide-requiring gene, sjo (in one instance, that de-
rived from culture number 1 in Figure 12.3, the sJo gene itself was somewhat

en

bo On

>. ni

^ S

^ eg

O n!
u

-2 Pi

-a o

.S -^

o -

OS f

CI- «

c/i C

2 ^

3 S

o 3

c ^

3 M

cr u

11 o

en

cj .5

c —

'^ o

>

O

O

o

c

3

O

o

BIOCHEMICAL MODELS OF HETEROSIS IN NEUROSPORA

205

modified), and in addition a second mutant gene, S, which was presumably
responsible for the change in growth (Table 12.1).

The new mutants appearing in the heterocaryons have been called sup-
pressors because they overcome the deleterious effect of the sulfonamide-
requiring gene in heterocaryons. Actually they are not like the usual suj)-
pressors, because in homocaryotic strains which also carry the sulfonamide-
requiring gene they do not result in wild type growth.

Growth characteristics of strains homocaryotic for four of these suppres-
sors, with and without the sulfonamide-requiring gene, are represented in

TABLE 12.1

DISTRIBUTION OF NUCLEI IN
THE HETEROCARYONS REPRE-
SENTED IN FIGURE 12.3

From Culture
Tube Number

Nuclei

sfo.

+

sfo,S

1.

2.
3.
4.

3

6

15

8
8
1

5
5
2
1

5

1

6

14

Figure 12.4. From these growth curves it can be seen that wild type (+, +)
is neither inhibited by sulfanilamide in a concentration of 2 X 10~^ M, nor
stimulated by />-aminobenzoic acid in a concentration of 10"'* M when grown
at 35°, and is only slightly inhibited by sulfanilamide at 25°. At 35° growth of
the sulfonamide-requiring strain (sfo, -\-) is stimulated by sulfanilamide and
inhibited by ^-aminobenzoic acid, though neither substance has an appre-
ciable effect at 25° in the concentrations used.

The suppressor from tube 1 (+, ^-1) does not grow at 35°, and grows slow-
ly on all media at 25°. The suppressor from tube 2 (+, 5-2) differs from wild
type principally in taking longer to attain its maximum growth rate, though
there is also some stimulation by sulfanilamide at 35°. When combined as a
double mutant with the sulfonamide-requiring gene (sfo, 5-2), it almost ap-
proximates the growth of wild type. The suppressor from tube 4 (+, 5-4)
differs from wild type in being stimulated by />-aminobenzoic acid and in-
hibited by sulfanilamide, the inhibition being stronger at 25°. In combination
with the sulfonamide-requiring gene (sfojSA) it resembles the sulfonamide-
requiring strain itself except that there is a long lag phase on sulfanilamide
at 35°, and inhibition at 25°. The suppressor from tube 6, either alone

HOURS

Fig. 12.4 — Growth curves of suppressor strains in absence and presence of the sulfonamide-requiring gene (sfo)
at 35° and 25° on minimal medium (light Hne), 10"* M />-aminobenzoic acid (line of open circles — PABA), and

2 X 10"" M sulfanilamide (dotted line -SA).

BIOCHEMICAL MODELS OF HETEROSIS IN NEUROSPORA 207

(-|_^ 5.6) or in combination with the sulfonamide-requiring gene (sfo, 5-6),
grows very poorly at 35°.

Of those illustrated, suppressors numbered 4 and 6 are perhaps the most
significant to the present discussion. When combined with the sulfonamide-
requiring gene (sfo, SA and sfo, S-6), neither grows well on minimal medium
at 35°. Yet heterocaryons between either of these double mutants and the
sulfonamide-requiring strain are enabled to grow quite well under those
conditions. In these heterocaryons the sulfonamide-requiring gene is present
in all nuclei, in some of which it is combined with a suppressor. The suppres-
sor is not capable of overcoming the ill effects of the sulfonamide-requiring
gene when present in all nuclei, but is effective when present in only some

of them.

Biochemical Basis for the Sulfonamide-requiring Character

This seeming paradox becomes less important once the nature of the reac-
tion controlled by the sulfonamide-requiring gene is understood (Zalokar,
1948, 1950; Emerson, 1950). The diagrams in Figure 12.5 illustrate some of
the important reactions involved. There are a large number of amino acids,
vitamins, components of nucleic acid, and so on, that are essential to growth.
But we shall consider only two amino acids, methionine and threonine, and
the vitamin />-aminobenzoic acid. Para-aminobenzoic acid is involved in a
number of reactions essential to growth, one of which is the final step in the
synthesis of methionine from homocysteine. Wild type carries out all essen-
tial reactions and produces all essential growth factors, with the exception
of biotin which must be supplied to all strains.

The reaction governed by the sulfonamide-requiring gene has not yet been
identified, but we know quite a little about it. It requires the presence of both
homocysteine and />-aminobenzoic acid. Presumably homocysteine is used as
a substrate in this reaction, and />-aminobenzoic acid, or a derivative, is
needed as a catalyst. The reaction either results in the destruction of threo-
nine or else interferes with its normal utilization, so that the sulfonamide-
requiring strain has too little threonine for growth. W^e also know that more
homocysteine is required for this deleterious reaction than for the syn-
thesis of methionine, and that in the presence of limiting amounts of homo-
cysteine, the synthesis of methionine goes on without any interference with
the utilization of threonine.

Furthermore, the deleterious reaction requires larger amounts of ^-amino-
benzoic acid than are needed for all essential reactions combined. Only about
half as much is needed in the synthesis of methionine, about a quarter as
much in the production of purines, and very much less still for other essen-
tial, but still unidentified factors. Both wild type and the sulfonamide-requir-
ing strain produce about one hundred times as much />-aminobenzoic acid as
is needed for all essential reactions.

We know of three ways in which the deleterious reaction leading to threo-

Wild Type

Homocysteine

CH2-SH

CH2
"^ CH-NHz

COOH

Methionine
CH2-S-CH3

CH2

CH-NH2

COOH

COOH

^-Aminobenzoic
acid

GROWTH

CH3
^CHOH

CH-NH2
COOH

Threonine

Sulfonamide Requiring, Homocvsteineless

H-9a

*- .»-.■■-

•- c?/f(yr ^

Fig. 12.5—
by genes of

CH3

CHOH

CH-NH2

COOH

Certain biochemical reactions essential to growth in Neurospora as influenced
the sulfonamide-requiring strain (E-15172), the homocysteineless strain (H-98),
and the aminobenzoicless strain (1633).

208

SulfonoTiide Reg uirin g

CH2-SH ,

CH2 ^''''
CH-NHz
COOH "^2

CH2-S-CH3
GHz

CH-NHg
COOH

UNITS

UNITS^ UNITS

COOH

1/2 UNIT

CH3

CHOH

CH-NHz

COOH

GROWTH

Sulfonamide Requiring, Aminobenzoicless

CH^-S-CH3

CH2-SH
CH2
CH-NH2
COOH

CH3
^CHOH

"^'cH-NHz
COOH

Fig. 12.5 — Continued

GROWTH

209

210 STERLING EMERSON

nine deficiency can be prevented by genetic means. The simplest is of course
by introducing the wild type allele of the sulfonamide-requiring gene, but the
other two are of more interest. One of these is by introducing a genetic block
to the synthesis of homocysteine. Mutant strain H-98 blocks the terminal
step in the synthesis of homocysteine. In the double mutant — sulfonamide-
requiring, homocysteineless — there is no interference with the availability of
threonine for growth, since the deleterious reaction does not take place in
the absence of homocysteine. In the absence of homocysteine, however, there
can be no synthesis of methionine, so that the double mutant fails to grow
because of a methionine deficiency. The double mutant will grow if supplied
with exactly the right amount of methionine — more inhibits growth, because
methionine is degraded to homocysteine which then supports the deleterious
reaction (Zalokar, 1950).

The remaining method is to introduce a genetic block to the synthesis of
/»-aminobenzoic acid. In the double mutant — sulfonamide-requiring, amino-
benzoicless — there is again no interference with the utilization of threonine
since there is no />-aminobenzoic acid to catalyse the deleterious reaction.
There is again a deficiency for methionine, because />-aminobenzoic acid is
needed in its synthesis. There is also a deficiency of />-aminobenzoic acid for
other essential processes. The double mutant will grow if supplied just the
right amount of /(-aminobenzoic acid to satisfy the essential requirements,
but not enough to stimulate the deleterious reaction (Zalokar, 1948).

Model Heterocaryons

It can be seen that the simple sulfonamide-requiring mutant on the one
hand, and the two double mutants on the other, have different deficiencies.
One produces methionine and ^-aminobenzoic acid, but not enough threo-
nine. The others produce sufficient threonine, but no methionine, and in one
case, no ^-aminobenzoic acid. In heterocaryons between the simple and
double mutants, the two types of nuclei should complement each other in
the production of essential growth substances. If the nuclear ratios can be so
adjusted that the different substances are produced in appropriate amounts,
vigorous growth should result. Heterocaryons involving the simple sulfona-
mide-requiring mutant and the double mutant sulfonamide-requiring, amino-
benzoicless have resulted in vigorous growth (Emerson, 1948) in every test
so far made. Growth curves of some of these heterocaryons are illustrated
in Figure 12.6.

Growth of these heterocaryons is usually not maintained at a constant
rate. Growth may stop completely after a time, or it may nearly stop and
then start again. This is believed to be due to fluctuations in the ratio of the
two kinds of nuclei in the advancing hyphal tips. Apparently there must be
many times as many double mutant nuclei as simple sulfonamide-requiring
nuclei to result in a favorable combination. This is not surprising since the

'— I 1) C

oB I

« <U r-

[1- X! 00

sy3i3Piiiim

212 STERLING EMERSON

sulfonamide-requiring strain produces something in the order of one hun-
dred times as much />-aminobenzoic acid as is required for essential reactions,
or about fifty times as much as is required for the reaction which makes
threonine unavailable for growth.

Limited direct tests of nuclear frequencies in such heterocaryons indicate
that nuclei carrying only the sulfonamide-requiring gene are much less fre-
quent than those carrying the aminobenzoicless gene as well. In one test of
about one hundred nuclei, all proved to be double mutants. In another test,
conidia from heterocaryons were transferred to fresh growth tubes which
contained a concentration of sulfanilamide sufficient to inhibit growth of the
double mutant very strongly and still be favorable to the growth of the simple
sulfonamide-requiring mutant. Only one of five such transfers grew — again
suggesting that simple sulfonamide-requiring nuclei were infrequent.

If in order to have rapid growth there must be many double mutant nuclei
and few simple mutants, it is not surprising that vigorous growth should
cease rather suddenly. Ryan, Beadle, and Tatum (1943) have shown that
growth substances can be transported for a distance of about one centimeter
in the mycelium of Neurospora. One sulfonamide-requiring nucleus at a dis-
tance of about a centimeter from the tip might supply enough ^-aminoben-
zoic acid for the growth of that tip. But as the tip grows, that nucleus might
easily be left behind. A deficiency of />-aminobenzoic acid would then de-
velop in the tip, and growth would be arrested unless a nucleus of the proper
constitution happened to migrate into the tip.

Attempts to obtain rapidly growing heterocaryons involving the sulfona-
mide-requiring mutant and the sulfonamide-requiring, homocysteineless
double mutant were unsuccessful. It may be that it is impossible to have a
nuclear ratio which will produce sufl&cient, but not too much methionine,
and at the same time sufficient threonine for the requirement of the hetero-
caryon.

Interpreting Suppressor Heterocaryosis Based on Model Experiments

The heterocaryons between the sulfonamide-requiring mutant and its
double mutants with aminobenzoicless and homocysteineless were set up as
models which should duplicate the behavior observed in the sulfonamide-
requiring strain when suppressor mutations occurred, provided the interpre-
tation placed on them was correct. For this purpose, the results obtained
were gratifying. We should like to know just where each of the suppressor
mutations studied fits into the biochemical scheme, but at present it can be
shown only that they fit in a general way.

Four suppressors in the first lot of six (those illustrated in Fig. 12.4),
which are the only ones that have been studied in any detail at all, appar-
ently represent mutation at four different loci, though almost no direct tests

BIOCHEMICAL MODELS OF HETEROSIS IN NEUROSPORA 213

for allelism are available. The inference that they are distinct genes is based
on the data summarized in Table 12.2.

The reactions controlled by the suppressor genes have not been identified.
Suppressor SA is stimulated in growth by additional /)-aminobenzoic acid,
and is inhibited considerably by sulfanilamide at concentrations twenty
times less than that required to inhibit wild type. It is possible that a de-
ficient amount of />-aminobenzoic acid is produced by this mutant, which
would make it approximate the condition in one of the model heterocaryons.
Growth of suppressor S-2 is somewhat stimulated by sulfanilamide (Fig.
12.4) and by threonine, in this respect resembling the sulfonamide-requiring
mutant which it "suppresses." It is even more stimulated by the purine,

TABLE 12.2

EVIDENCE SUGGESTING THAT SUP-
PRESSORS 5i, S2, Si, AND Se ARE
DIFFERENT GENES

Suppressor

Second

Division

Segregation

Relation
to 1633

Genetically
Independ-
ent of

Si

25%

50%

0%

60%

none

allele ?

none

none

52

Si

s,

s.

Si

adenine, as shown by the growth curves in Figure 12.7. It was previously
known that in the presence of methionine, adenine reduces the normal re-
quirement for />-aminobenzoic acid to about one-tenth its usual value. This
suggested that the production of adenine also requires /)-aminobenzoic acid.
The reaction controlled by this suppressor may thus be closely related to
that controlled by the sulfonamide-requiring gene. No clues have turned up
to indicate how the reactions governed by the remaining suppressor muta-
tions may be related to these.

In the living cell of Neurospora the reactions which are influenced in one
way or another by the amount of available /»-aminobenzoic acid must be fairly
numerous. The production of adenine and methionine requires the presence
of this vitamin as does the reaction in the sulfonamide-requiring mutant
which makes threonine unavailable.

Strauss (1950) has studied a mutant strain (44602) which requires pyri-
do.xine unless grown at high pH with ammonia as nitrogen source. He found
that under the latter conditions it is inhibited by methionine, and that this
inhibition is reversed by sulfanilamide, as if />-aminobenzoic acid were re-
quired for the inhibition. Still another interrelationship has been found by
Shen (1950) in studies of a mutant strain (84605) which requires sulfur in a

214

STERLING EMERSON

form at least as reduced as thiosulfate. At 35° it has no other requirement,
but at 25° it needs reduced sulfur, generally supplied as the amino acid
cysteine, and also tyrosine. When methionine is supplied as the source of sul-
fur at 25°, growth is strongly inhibited by choline. Under these conditions,
choline does not inhibit at 35°, but there is an unexpected stimulation in
growth by ^-aminobenzoic acid at that temperature.

Mutant strains have been reported on two occasions which require either
choline or />-aminobenzoic acid — choline may be the source of the methyl

to
ce

HOURS

Fig. 12.7 — Growth curves of suppressor mutant strain S-2 on minimal medium, on threo-
nine (5 mg/100 ml), on methionine, and on purines (5 mg/100 ml each adenine sulfate and

guanine hydrochloride) at 35°.

group of methionine. Strehler (1950) has reported a strain which requires
either methionine or /?-aminobenzoic acid. There is also a suggestion that
/>-aminobenzoic acid may be involved in the metabolism of lysine. In Neuro-
spora this is suggested only because the double mutants between the sul-
fonamide-requiring strain and two different mutants which are unable to
synthesize lysine do not grow on any combination of growth factors we have
tried. In bacteria a strain has been found which requires either lysine or
/»-aminobenzoic acid as a growth factor (Koft el al., 1950), strengthening the
supposition of a similar interrelationship in Neurospora.

These observations are referred to at this time because they indicate that
there are a large number of metabolic reactions that are in one way or an-
other related to the availability of />-aminobenzoic acid. These reactions
must themselves be interrelated in the sense that an upset in one of them

BIOCHEMICAL MODELS OF HETEROSIS IN NEUROSPORA 215

may have a strong effect on one or more of the others, possibly through
changing the availability of /?-aminobenzoic acid or a derivative. The model
heterocaryon experiments described earlier show that it is possible for one
mutation to cause an upset in one reaction and thus be detrimental to growth,
and for a second mutation to restore conditions favorable to growth by actu-
ally interfering with a different reaction which is itself essential to growth,
but which is interrelated with the first reaction. In the reactions related to
the metabolism of ^-aminobenzoic acid, there is sufficient complexity to ac-
count for the occurrence of a large number of different suppressors of the sul-
fonamide-requiring character.

DISCUSSION

It has been shown that increased vigor can result from heterocaryosis in
which the two kinds of nuclei differ by only one pair of alleles. This may be
true only under very special conditions such as have been present in the
examples discussed. On the other hand, it is possible that the necessary con-
ditions may be met with rather frequently in Neurospora, as suggested by
the following examples.

In mutant strains which have specific requirements for particular amino
acids, it is commonly found that their growth is inhibited by the presence of
other amino acids which do not ordinarily interfere with growth. Some mu-
tants which require an outside source of threonine are strongly inhibited by
methionine, (Teas, Horowitz, and Fling, 1948). Mutants specifically requir-
ing lysine are inhibited by arginine (Doermann, 1944), and so on. In each of
these instances, the inhibition by a particular amino acid is competitively
antagonized by the specific amino acid required by the strain in question.
The growth of these mutants should be favored by a reduction in the amount
of the inhibiting amino acid, as would occur if some of the nuclei carried a
genetic block to its synthesis.

In extreme cases, the specific requirement for an amino acid may not re-
sult from a failure in its synthesis, but from an oversensitivity to the in-
hibiting amino acid. Thus, the sulfonamide-requiring strain can be said to
be oversensitive to homocysteine in a way that leads to a requirement for
threonine. One of the lysineless mutants (33933) seems to be oversensitive to
arginine in much the same way. Heterocaryons having the lysineless gene in
all nuclei, some of which also carry a genetic block to the synthesis of ar-
ginine (from strain 36703), make considerable growth on minimal medium,
whereas neither the lysineless nor the double mutant does (Fig. 12.8).

Mary B. Mitchell (personal communication) recently observed that the
stock cultures of certain lysineless mutants (4545, 15069, and 33933) had
become less sensitive to inhibition by arginine. Tests of these showed that
they were heterocaryons, some of whose nuclei were unchanged. Some car-
ried mutant genes which lowered the sensitivity to arginine inhibition while

216

STERLING EMERSON

leaving the requirement for lysine. These heterocaryons were more vigorous
than the original lysineless strain, but no more vigorous than the pure double
mutant strains extracted from the heterocaryons.

In studies on reverse mutation in a leucineless strain (33757), Ryan and
Lederberg (1946) found that heterocaryons, whose nuclei differed only in

'

;|

•150

/

.

/ 2',

•

*> i

/ f]

•

o /

/'''

■

TJ /

/

'1 t

ex:

■100

*» /
4. /

UJ

■^/

•/ /

h-

/t/

UJ

'p'

-J
_j

•
/

/ //J •

S

/ /

/^ /^

50

/

/

//

y/

/y

>

*••/

^•^x

,-'

.y^ X

, --

*• *

-^l

^^•-^

^ .'y^ ^^% _ .

50

i6o

HOURS

Fig. 12.8 — Growth curves of heterocaryons between lysineless (/y, +) and lysineless,
arginineless (/y, arg) strains of Neurospora at 35° on minimal medium. Curve 1 : heterocar-
yon in which both nuclear types were of mating type A; curves 2 to 5: heterocaryons made
up of nuclei of different mating types (/v, +, A and /v, arg, a) — cf. Beadle and Coonradt

' (1944).

that some carried the wild type allele and some the mutant allele of the
leucineless gene, almost invariably reverted to the homocaryotic condition.
By the time growth had reached the end of a tube containing minimal me-
dium, nothing but wild type nuclei remained. In tubes containing limiting
concentrations of leucine, nothing but leucineless nuclei were present after a
short period of growth. This was under conditions where the growth rate of
the leucineless strain is considerably less than that of wild type. Under both
of these conditions, the heterocaryon is at a strong disadvantage compared
to its components. It is not known whether or not there is a particular con-
centration of leucine which would favor the heterocaryon.

BIOCHEMICAL MODELS OF HETEROSIS IN NEUROSPORA 217

Houlahan and Mitchell (1948) have studied the interactions of mutant
strains involved in the metabolism of i)yrimidines and lysine. A pyrimidine-
less mutant (37301) has a specific requirement for pyrimidine. There is a
suppressor of this mutant which enables it to grow without added pyrimidine,
unless arginine is also added, whereupon the pyrimidine requirement is re-
stored. One lysineless strain (33933) can utilize a-amino adipic acid in place
of lysine. As a double mutant with the pyrimidine suppressor, it can still
use a-amino adipic acid, but requires four times as much as the simple lysine-
less strain unless small amounts of arginine, or an arginine precursor, are
added. The double mutant combining this lysineless with the pyrimidineless
mutant is unable to use a-amino adipic acid unless a small amount of lysine
is added — arginine is ineffective in this instance. A second lysineless mutant
(454.5), which has a specific requirement for lysine and which secretes pyrimi-
dines into the medium, behaves in a predictable fashion as a double mutant
with pyrimidineless, or its suppressor, but not as the triple mutant lysineless,
pyrimidineless, suppressor of pyrimidineless. Instead of requiring only lysine
for growth, this triple mutant also requires pyrimidines and arginine. This
example is cited as another in which metabolic interactions may be as com-
plex as in those discussed earlier which depend in one way or another on
/>-aminobenzoic acid.

Applicability to Classical Heterosis

Observations relating to one-gene heterosis in higher plants are discussed
in other papers in this series (Crow, Hull, Jones, and Whaley). Studies of
Neurospora heterocaryons have shown that a very similar phenomenon oc-
curs under certain special physiological conditions. In a particular genetic
background, the amount of an essential metabolite normally produced has
deleterious consequences which are removed by reducing the dosage of a
gene responsible for the production of that metabolite. This reduction was
brought about through heterocaryosis in the studies reported, but it should
also result from heterozygosis under similar physiological conditions. There
is nothing in the studies of heterocaryosis in Neurospora to suggest that one-
gene heterosis is of general occurrence and importance, or that other examples
should have similar biochemical backgrounds.

TH. DOBZHANSKY

Columbia University

Chapter 13

Nature and
Origin of Heterosis

Exploitation of heterosis in cultivated plants and animals is to date by far
the most important application of the science of genetics in agricultural prac-
tice. It is therefore unfortunate that few of the studies so far made on
heterosis go beyond crudely empirical observations and descriptions and
that little effort is being made to understand the underlying causes of the
phenomena involved. Such an understanding is needed particularly because
the advances of general genetics make it evident that several quite distinct,
and even scarcely related, phenomena are confused under the common label
of heterosis or hybrid vigor.

In what follows, an attempt is made to indicate briefly what seem, to the
writer, promising lines of approach to a classification and study of the various
kinds of heterosis. The tentative nature of the classification here suggested
is fully realized. But it is believed that this classification may nevertheless
serve a useful function if it directs the attention of the students of heterosis
to factors which are only too often overlooked.

MUTATIONAL EUHETEROSIS

Perhaps the simplest kind of true heterosis — euheterosis — is that which
results from sheltering of deleterious recessive mutants by their adaptively
superior dominant alleles in populations of sexually reproducing and cross-
fertilizing organisms.

Although only a small fraction of the existing species of organisms have
been investigated genetically, it is reasonable to assume that mutational
changes arise from time to time in all species, albeit at different rates. Fur-
thermore, a great majority of the mutations that arise are deleterious, and
lower the fitness of their carriers to survive or to reproduce in some or in all

218

NATURE AND ORIGIN OF HETEROSIS 219

environments. This deleterious character of most mutations seems surpris-
ing, especially because in modern biology the process of mutation is regarded
as the source of the raw materials from which evolutionary changes are con-
structed.

A little consideration shows, however, that the ada])tively negative char-
acter of most mutations is by no means unexpected. Indeed, since every mu-
tation has a finite probability to occur in any generation, the mutants which
we observe in our fields and laboratories must have arisen many times in the
history of the species. The rare mutants which confer adaptive advantages
on their possessors in the environments in which the species normally lives
have had the chance to become established in the species populations as
components of the normal species genotype. In a more or less static environ-
ment, the genotypes of most species are close to the upper attainable level of
adaptedness.

The above argument may seem to prove too much. In the absence of use-
ful mutants, evolution would come to a standstill. The paradox is resolved
if we recall that the environment is rarely static for any considerable periods
of time. Furthermore, most living species occur not in a single but in several
related environments. Genotypes which are adaptively valuable in a certain
environment may be ill adapted in other environments, and vice versa. It
should be possible then to observe the occurrence of useful mutations if we
place the experimental organisms in environments in which their ancestors
did not live.

Progressive improvement of domesticated animals and plants in the hands
of breeders constitutes evidence that useful mutations do occur. The genetic
variants which are being made use of by breeders have arisen ultimately
through mutation. These mutations have been arising from time to time, be-
fore as well as after the domestication. But while they were deleterious in
the wild state, some of them happened to be suitable from the standpoint of
the breeders. They were useful in the man-made environment or they were
useful to man. Favorable mutations can be observed also in wild species,
provided that the latter are placed in unusual external or genetic environ-
ments. This has been demonstrated in experiments of Spassky and the writer
on Drosophila pseudoobscura. Several laboratory strains of this fly were sub-
jected to intense selection for fifty consecutive generations, and improve-
ments of the viability have been observed in most of them.

Many, perhaps most, deleterious mutants are nearly or completely reces-
sive. Others are more or less dominant to the "normal," or ancestral, state.
The fate of the dominant deleterious mutants in jwpulations of sexually re-
producing and cross-fertilizing species is different from that of the recessives.
By definition, deleterious mutants in wild species lower the fitness of their
carriers to survive or to reproduce, and in cultivated species impair the
qualities considered desirable by the breeders. Natural and artificial selec-

220 TH. DOBZHANSKY

tion will consequently tend to lower the frequency, or to eliminate deleteri-
ous mutants.

Selection against a dominant deleterious mutant is, however, a far more
efficient process than that against a recessive mutant. This is because dele-
terious recessive mutant genes are sheltered from selection by normal domi-
nant alleles in heterozygotes. Deleterious dominants are eliminated by selec-
tion within relatively few generations after their origin. Deleterious reces-
sives accumulate in heterozygotes until their frequencies become so high that
recessive homozygotes are produced. Dominant alleles are not intrinsically
beneficial, and recessives are not necessarily deleterious. But at any one time,
we find in cross-fertilizing populations more deleterious recessives than dele-
terious dominants, because the former are not eliminated by selection as
promptly as the latter.

Analysis of wild populations of several species of Drosophila has revealed
extensive infestation of the germ plasm by deleterious recessive mutant genes.
According to the unpublished data of Pavan and collaborators, 41 per cent
of the second chromosomes in Brazilian populations of Drosophila willistvni
are lethal or semilethal when homozygous. Among the remainder, 57 per
cent are sublethal when homozygous. Furthermore, 31 per cent of the second
chromosomes make the homozygotes completely sterile in at least one sex,
32 per cent retard the development, and 16 per cent cause various visible
abnormalities. Comparable figures for the third chromosomes are 32 per
cent of lethals and semilethals, 49 per cent subvitals, 28 per cent steriles,
36 per cent retarded, and 16 per cent containing visible mutants. Since
every fly has two second and two third chromosomes, it is easily seen that a
great majority of individuals in Brazilian populations carry several deleteri-
ous variants in heterozygous condition.

The mass of deleterious recessives carried in normally breeding natural
populations has no disastrous effects on the average fitness of members of
such populations. This is because the frequency of recessive homozygotes
found in a population at equilibrium is equal to the number of the corre-
sponding recessive mutants that arise in every generation. The loss of fitness
caused in a normally breeding population by dominant and by recessive mu-
tants is thus proportional to the frequency of the origin of these mutants by
mutation.

The situation changes completely if a normally crossbred population is
subjected to inbreeding. For inbreeding renders homozygous many reces-
sives that would remain sheltered in heterozygotes under normal crossbreed- y
ing. These recessives become suddenly exposed to natural, or to artificial,
selection. The loss of fitness in inbred lines of normally cross-fertilized species
is the consequence. Conversely, the heterosis observed in the progeny of
intercrossed inbred lines is the outcome of restoring the normal reproductive
biology and the normal population structure of the species.

NATURE AND ORIGIN OF HETEROSIS 221

BALANCED EUHETEROSIS

Balanced heterosis is due to the occurrence of a rather special class of
mutations and gene combinations, which confer on heterozygotes a higher
adaptive value, or a higher agricultural usefulness than is found in the cor-
responding homozygotes.

The conditions most frequently found in heterozygotes are either domi-
nance and recessiveness, when the heterozygote is more or less similar to one
of the homozygotes, or phenotypical intermediacy between the homozygotes.
A heterozygote may, however, be in some respects phenotypical ly more ex-
treme than either homozygote. Thus, a heterozygote may be more viable,
more productive, or otherwise exceed both homozygotes in some positive or
negative quality. This condition is sometimes spoken of as overdominance
(Hull).

Although overdominance is, by and large, an exceptional situation, it is of
particular interest to a student of population genetics, and especially to a
student of heterosis. Suppose that a certain gene is represented in a popula-
tion by a series of alleles, A\ A^,A^ . . . which are deleterious in homozygous
condition, AKi\ A-A^, A^A^ . . . , but which show a relatively higher fitness
in heterozygotes A^A'^, A'^A^, A~A^ . . . , etc. Natural or artificial selection
would preserve in the population all the variants A\ A", A^ . . . , regardless
of how poorly adapted the homozygotes may be. In fact, one or all homozy-
gotes may be semilethal or even lethal, and yet selection will establish an
equilibrium at which every one of the variants will be present with a definite
frequency. This equilibrium can easily be calculated if the selective dis-
advantages of the homozygotes, compared to the heterozygotes, are known.
The resulting situation is referred to as balanced polymorphism.

Balanced polymorphism may be produced by mutations in single genes,
provided that the heterozygotes exhibit overdominance in fitness in some
environments. This has been demonstrated, among others, by Gustafsson
and Nybom. They observed several mutations in barley that were deleterious
in homozygotes, but produced heterozygotes superior to the ancestral "nor-
mal" homozygotes. Ford and others showed that certain color variants in
butterflies, which are inherited as though caused by a single genetic change,
are maintained in natural populations by the same mechanism.

Detailed data are available on balanced polymorphism in several species
of Drosophila, in which natural populations are very often polymorphic for
gene arrangements in some chromosomes. These gene arrangements differ in
inversions of blocks of genes. Thus, in certain populations of Drosophila
pseiidoobscura from Southern California, at least 70 per cent of the wild indi-
viduals are inversion heterozygotes. In populations of Drosophila willisloni
from central Brazil (Goyaz), an average individual is heterozygous for as
many as nine inversions, and very few individuals are homozygous.

222 TH. DOBZHANSKY

Now, it has been shown by observation both on natural and on experi-
mental populations of some Drosophila species, that the heterozygotes for
the naturally occurring inversions possess considerable adaptive advantages
over the homozygotes. For example, taking the adaptive value of the
heterozygotes for ST and CH inversions in Drosophila pseudoobscura to be
unity, the adaptive values of the ST/ST and CH/CH homozygotes are
about 0.8 and 0.4 respectively. Further, it has been shown that the heterosis
in the ST/CH heterozygotes occurs only if the constituent chromosomes are
derived from the same population, or from populations of nearby localities.
Chromosomes with the same gene arrangements, ST and CH, derived from
remote localities (such as Central and Southern California, or Southern
California and Mexico) exhibit little or no heterosis.

This finding is most compatible with the assumption that the over-
dominance in fitness observed in the heterozygotes is the property not of a
single gene locus, or of a chromosome structure, but rather of integrated sys-
tems of polygenes. Such polygenic systems are coadapted by natural selec-
tion to other polygene complexes present in the same populations. The role
of the chromosomal inversions in the formation of the heterotic state of bal-
anced polymorphism is due to the suppression of crossing over caused by
most inversions, at least in Drosophila. Elimination of crossing over prevents
the breakup of the adaptively integrated polygene complexes which are
carried in the chromosomes involved.

It should be noted that adaptively integrated polygene complexes can be
maintained in crossbreeding populations with the aid of genetic mechanisms
other than chromosomal inversions. Any factor which restricts or prevents
crossing over in chromosomes, or parts of chromosomes, can accomplish the
same biological function. Localization of chiasmata may be such a factor.
If, for example, chiasmata are found chiefly or exclusively at some definite
points in a chromosome, the genes carried in the sections which intervene
between these points are inherited in blocks. Such gene blocks may act
exactly as gene complexes bound together by inversions.

Balanced heterosis differs profoundly from mutational heterosis. The
latter is due simply to the sheltering of deleterious recessive mutants by
their dominant alleles. Balanced heterosis is a result of overdominance. Mu-
tational heterosis is a protective device of a sexual species with a certain
population structure against the mutation pressure. Balanced heterosis is an
evolutionary contrivance that permits maintenance in a population of a mul-
tiplicity of genotypes that may be adaptive in dififerent ecological niches
which the population occupies.

LUXURIANCE

Mutational and balanced heterosis resemble each other in one important
respect — both are normal adaptive states attained in outbred sexual species

NATURE AND ORIGIN OF HETEROSIS 223

as a result of an evolutionary history controlled by natural or by artificial
selection. The normal heterotic state can be disru])ted by sudden inbreeding,
which is evidently a disturbance of the reproductive biology to which the
species is adjusted. The heterotic state can also be restored by intercrossing
the inbred lines. This is true heterosis, or euheterosis. Euheterosis is a form
of evolutionary adaptation characteristic of sexually reproducing and cross-
fertilizing species.

Numerous instances are known, however, when hybrids between si)ecies,
neither of which can be regarded as inbred, are larger, faster growing, or
otherwise exceeding the parental forms in some quality. Similar luxuriance is
observed in some hybrids between normally self-fertilizing species, races, or
strains. This kind of luxuriance of hybrids cannot be ascribed to sheltering of
deleterious recessive mutants, because the latter are sheltered in the parental
populations. It is also unlikely to arise from overdominance since, at least in
wild species, natural selection would be expected to have induced such bal-
anced heterosis in the parental species or strains.

Luxuriance is, from the evolutionary standpoint, an accidental condition
brought about by complementary action of genes found in the parental form
crossed. Two sets of facts are important in this connection. First, in cases of
luxuriance there is usually no indication whatever that the luxuriant hybrids
would prove adaptively superior in competition with the parental forms in
the natural habitats of the latter. Second, luxuriance appears to be more
frequently encountered in domesticated than in wild species.

It stands to reason that increase in body size, or in growth rate, is by no
means always an adaptively superior change. To equate size with vigor, fit-
ness, or adaptive value would be a height of anthropomorphic naivete. The
rate of growth and the size attained by an organism in its normal environ-
ments are evidently controlled by natural selection. Excessive as well as de-
ficient sizes are adaptively about equally disadvantageous. The checks upon
excessively rapid growth and excessive size are, however, very often relaxed
under domestication. In man-controlled environments those qualities often
become desirable from the standpoint of the breeder if not from that of the
organism. Luxuriance is, really, pseudoheterosis.

DONALD F. JONES

Connecf/'cuf Agricultural Exper'imenf Sfaiion

Chapter 14

Plasmagenes and
Chromogenes in Heterosis

The word heterosis is essentially a contraction of the phrase stimulus of
heterozygosis. It was first used by G. H. Shull (1914). The concept of a
stimulation resulting from the genetic union of unlike elements was de-
veloped by East (1909). Previous to the Mendelian conception of units of
heredity, it was generally considered by plant and animal breeders that the
invigorating effect of crossing unlike varieties of plants and breeds of live-
stock was due to the correction of imperfections that existed in both parental
types. This idea is clearly stated by Samuel Johnson in the second edition of
his book How Crops Grow (1891).

The early recordings of instances of hybrid vigor and the various means
of accounting for this phenomenon have been stated and restated so many
times that there is no need or useful purpose in repeating them here. Excel-
lent reviews of the literature are readily available (see especially East and
Hayes, 1912; Jones, 1918; East and Jones, 1919; East, 1936; and Whaley,
1944).

THE EXPRESSION OF HETEROSIS

At the present time, the term heterosis designates the increased growth or
other augmented action resulting from crossing, however it is produced. As
generally used, it is essentially synonymous with hybrid vigor. Heterosis has
two general modes of expression. In one, there is an increase in size or num-
ber of parts. This is usually the result of a greater number of cells and a faster
rate of cell division and cell activities. This results in an improvement in gen-
eral well-being of the organism similar to the result of being placed in a more
favorable environment. Such luxuriance may be accompanied by partial or
complete sterility in diverse crosses.

A somewhat different manifestation of heterosis is an increase in bio-

224

PLASMAGENES AND CHROMOGENES IN HETEROSIS 225

logical efficiency, such as reproductive rate and survival ability. This may
even be shown with a reduction in productiveness as measured by economic
characters. Some confusion has arisen by not distinguishing clearly between
these two different manifestations of heterosis.

In addition to these two general types of heterotic effects, there may also
be a reduction in both growth and survival ability; in other words, hybrid
weakness or a reversed or negative heterosis. This effect is much less com-
mon and is seldom found in cultivated plants and domesticated animals.

TYPES OF GENE ACTION

An understanding of the mode of action of heterosis has now resolved into
a study of the nature of gene action. The genes usually used to illustrate
Mendelism are the loss variations that have a major effect such as the inabil-
ity to produce some essential substance. This results in a block in the normal
chemical processes, finally resulting in an individual of greatly altered ap-
pearance, size, or ability to survive. The effect ranges in intensity from a com-
pletely lethal condition at some stage of development, up to individuals that
differ only slightly in appearance from normal with no appreciable reduction
in growth or survival ability. Such genes are illustrated by the long lists of
Mendelizing characters now tabulated for maize, Drosophila, mice, and many
other animals, plants, and lower organisms.

DOMINANT AND RECESSIVE GENES

In these cases, the normal allele is usually designated by a capital letter,
with the mutant, deficient allele denoted by the corresponding lower case
letter. In comparison with the normal allele, the recessive mutants are de-
ficient in some respect. In their inability to produce certain specific sub-
stances, as shown in the haploid Neurospora by Beadle and his co-workers,
they are referred to as .4 -less, 5-less, C-less, etc. In diploid organisms A is
usually completely dominant over a; that is, one A allele functions as well
or nearly as well as two.

There is no question that the accumulation in a hybrid of the normal
alleles of this type results in heterosis. In the simplest example of a cross of
^-less by B-less (aaBB XAAbb) the hybrid offspring are all AaBb, and
essentially normal for whatever effect A and B have. But since the mutant
recessive alleles of this type are so drastic in their effect, most of these
deficiencies are removed by natural selection in all species whether self-
fertilized or cross-fertilized. Therefore they have little part in the heterosis
that is shown by these organisms when crossed. Furthermore, genes of this
type are eliminated when naturally cross-fertilized species, such as maize,
are artificially self-pollinated. Yet such inbred strains show the largest
amounts of heterosis.

There is evidence, as will be shown later, that there are many genes of this

226 DONALD F. JONES

type having small effects that are not eliminated by natural or artificial
selection either in the wild or under domestication, and that these deficiencies
or degenerative mutants do have a large part in bringing about reduced
growth. Before presenting this evidence, there are other types of gene action
that should be considered.

CHROMOSOMAL DELETIONS
In addition to the recessive mutant alleles that are deficient as compared
to their normal alleles, there are also chromosomal deletions which result in
the complete elimination of the normal locus. Large deletions are usually
lethal and are quickly eliminated. Small deletions that cannot be detected
cytologically are haplo-viable, and may persist indefinitely if they are closely
linked with essential loci. Changes of this type have been demonstrated by
McClintock (1931) and by Stadler (1933). They cannot be readily distin-
guished from recessive mutants of the .4 -less type. In fact there may be no
difference. In practically all cases they show varying amounts of germ cell
abortion, and do not mutate back to normal. Deletions of this type are
designated Ao.

DOMINANT UNFAVORABLE GENES

In many cases of deletion the heterozygote, or the hemizygote, is visibly
and unfavorably altered from normal, in which event the genes involved are
listed as dominant, and if partially viable they can bring about negative
heterosis or hybrid weakness. It is not known whether all dominant unfavor-
able genes are deletions of this type, but as far as their effect on heterosis is
concerned it makes little difference whether or not they are. An illustration
of this type of gene action may be seen in a cross of Ragged and Knotted
maize plants. Both of these genes result in a marked reduction in growth in
the heterozygous condition. They are not completely lethal in the homozy-
gous dominant condition, but seldom produce seed or pollen. When both
dominant genes are present together in the heterozygous condition, there is
a marked reduction in size, rate of growth, and reproductive ability as com-
pared with either parental type.

Tunicate, teopod, and corn grass are also dominant genes that reduce
grain yields in both the homozygous and heterozygous condition. They are
probably reversions to a primitive condition which in suitable genetic com-
binations maybe favorable to survival in the wild. Dunn and Caspari (1945)
describe many structural abnormalities in mice that seem to be due to dele-
tions having a dominant effect in the hemizygote. Some of these counteract
each other and tend to restore a more normal condition, while others accumu-
late unfavorable effects. A similar situation has been reported in Drosophila

by Stern (1948). .

In addition to recessive deletions with a dominant effect m the hetero-
zygote, there are also dominant inhibitors that have no indication of being

PLASMAGENES AND CHROMOGENES IN HETEROSIS 227

deletions, but do prevent other genes from liaving their usual exi)ression.
Most of these inhibitors control color characters and are usually not involved
in heterosis. If they were, there would be more negative heterosis than actual-
ly is found.

GENES WITHOUT DOMINANCE

Unlike the visible Mendelizing genes with their clear-cut dominance and
unfavorable action of one or the other allele, there are many genes that dif-
ferentiate size or number of parts, time of flowering and maturing. These are
the genes usually involved in normal variation. They are the ones the plant
and animal breeder are mainly concerned with and could expect to have a
major effect on heterosis. Since neither member of an allelic pair can be con-
sidered abnormal or deficient, both are designated with a capital letter with
some prefix to differentiate them, as for example A and A\

Genes of this type usually have simple additive effects such as the F endo-
sperm color gene in maize, in which each allele adds a definite increment in
total carotene content. Such additive genes without dominance are used to
interpret the inheritance of quantitative characters which have been shown
to segregate and recombine in a Mendelian manner.

No clear distinction can be made between the A a and .4^1' types of genes
and this has led to much confusion. The first class shows complete or nearly
complete dominance. The second shows no dominance or very little domi-
nance, but one type integrates into the other. The principal question at issue
is whether either type shows over-dominance, or in other words, an interac-
tion between alleles such that Aa > A A or aa or AA^ > A A or .4 '.4'. Before
considering the evidence for or against over-dominance, two remaining types
of genes should be considered.

CHROMOSOMAL REARRANGEMENTS

By chromosomal rearrangements such as inversions and translocations,
genes without alteration are placed in different spatial relations with other
genes. In their altered position they have different effects. Dobzhansky and
his associates have studied many geographical races of Drosophila that differ
by chromosomal rearrangements. Crosses between these chromosomal types
from the same region exhibit heterosis, whereas the same chromosomal type
from different regions do not show such a high degree of heterosis. This
seems not to be a position effect, but is the result of an accumulation of gene
differences that are protected from random distribution by the prevention of
crossing over in hybrids of different chromosomal types.

COMPOUND GENES AND GENES WITH MULTIPLE EFFECTS
In many organisms, loci are known which have different effects on differ-
ent parts of the organism. In maize the .1 , P, and R genes have been studied
in considerable detail by Stadler and his co-workers. These loci each have a

228 DONALD F. JONES

series of alleles that produce characteristic color patterns and intensities of
colors in different parts of the plant such as culm, leaf sheath, leaf blade,
glumes, anthers, silks, cob and pericarp, and endosperm. They may be con-
sidered either as genes located so closely together that they never show-
crossing over, or compound genes with multiple effects. Without going into
the evidence for or against these two hypotheses, it is obvious that compound
genes can have an important part in heterosis if they control growth proc-
esses. More information is needed on the specific effect of compound genes.

In Godetia a series of multiple alleles has been described by Hiorth (1940)
that is often cited as an illustration of an interaction between alleles produc-
ing an effect analogous to heterosis. Actually these are color determiners that
control pigment production in different parts of the flower quite similar to the
A, P, and R loci in maize. Each allele has a different manifestation, and all
tend to accumulate color in the heterozygotes.

The familiar notation of a chromosome as a linear arrangement of loci,
each of which is the site of a single gene with one effect function, is probably
an oversimplification of the actual condition. It is difficult to see how an
organism could have originated in this way. It is more likely that a chromo-
some is an association of primitive organisms of varying types and functions.
These primitive organisms found it to be an advantage in the evolutionary
process to become associated in some such process as the colonial organisms
now exhibit. This association has undergone very great modification and
ramifications, but the compound genes may be vestigial structures of such
an association, differing greatly in size, arrangement, and function. Many of
them still retain some independence, and when removed from their normal
position in the chromosome could function as plasmagene or viroid bodies.

These compound genes may undergo mutation and possibly recombina-
tion or reorganization within themselves, but crossing over takes place for
the most part only between these compound structures. Compound genes
also arise by unequal crossing over and duplication of loci are shown by the
Bar eye gene in Drosophila and others of similar type.

In addition to compound or multiple genes, there are single genes with
multiple effects. Many of these are important in growth processes and are
illustrated by chlorophyll production in maize studied by H. L. Everett
(1949). One major gene is essential for the production of carotene. In the
recessive condition the seeds are pale yellow in color, in a normal, dark yel-
low seeded variety. The young seedlings grown from these pale yellow seeds
are devoid of chlorophyll. The recessive allele is therefore lethal. By using the
pale yellow endosperm as a convenient marker and crossing with a number of
standard field corn inbreds, it has been found that these inbreds differ widely
in their normal chorophyll mechanism. Many of them have genes that can
restore normal chorophyll production without restoring the production of
carotene in the seed. Other genes restore chlorophyll production only partial-

PLASMAGENES AND CHROMOGENES IN HETEROSIS

229

]y (see Table 14.1). Hybrid combinations that bring these genes together are
appreciably more efficient in chl()ro])hyll production than combinations that
lack some of them. However one of these dominant alleles has a suppressing
eflfect on chlorophyll development. The combination of all of these chloro-
phyll genes so far studied is not the most productive. There are many genes
of this type that block chemical syntheses, that are not lethal in the usual
genetic assembly, but which combine to give a cumulative efficiency in most
cases.

Lethal genes which show complete dominance of the normal allele would
have no effect on heterosis other than to reduce the number of offspring. Such

TABLE 14.1

GENES CONTROLLING CHLOROPH\TL PRO-
DUCTION IN MAIZE*

Ch

Ch

Ch

Seed Color

Chlor. Grade

Viability

Pale

Albino

Lethal

—

+

_

Pale

Virescent

Lethal

—

+

Pale

Light green

Normal

—

+

+

Pale

Light green

Normal

+

—

—

Yellow

Light green

Normal

+

+

—

Yellow

Med. green

Normal

+

+

+

Yellow

Dark green

Normal

+

+

Yellow

?

Normal

* Data from H. L. Everett.

genes would be just as effective in the homozygous as the heterozygous con-
dition. Genes that have any part in the type of heterosis that is manifested
in increased growth must be viable and have some degree of dominance. In
other words, Aa must be greater than | AA. Aa may even be greater in
its effect than A A or aa in which case theoretically there is over-dominance,
but very little specific evidence is available to show that such a situation
actually exists.

I can see no way in which it is possible to separate over-dominance from
a stimulus of heterozygosis. They seem to be different ways of saying the
same thing. The essential point at issue at the present time is whether or not
over-dominance actually occurs, and if so, how important this is in the
total amount of heterosis in addition to the known accumulation of favorable
dominant effects.

INTERACTION BETWEEN ALLELES

Evidence has been presented from many sources bearing on the problem
of over-dominance and interaction between different alleles. Much of the
argument is based on mathematical treatment of data that require many as-
sumptions. What is needed is more specific evidence where the effect of

230 DONALD F. JONES

multiple genes can be ruled out. Very few specific examples of single gene
action are available.

In one case studied by the writer there is clear evidence for an interaction
between alleles (Jones, 1921). A mutation in a variety of normally self-ferti-
lized tobacco changed a determinate plant into an indeterminate, non-
flowering variation. It was a change in the normal response to the summer
day length period. The mutant plants failed to flower in the normal growing
season and continued in a vegetative condition. Reciprocal crosses between
the mutant and normal types both grew at the same rate as the normal
plants showing complete dominance of the normal growth rate. The hetero-
zygous plants continued their vegetative growth longer and produced taller
plants with more leaves and flowers than the normal homozygous plants.
This result I consider not to be heterosis, since there was no increase in
growth rate. It is merely an interaction between alleles to produce a result
that is different from either parent. There are undoubtedly many allelic
interactions of this type. Whether or not they can be considered to contribute
to heterosis is largely a matter of opinion.

Other cases in corn where heterosis resulted from degenerative changes
(Jones, 1945) were at first assumed to be single allelic differences, since they
originated as mutations in inbred and highly homozygous families. The de-
generate alterations were expressed as narrow leaves, dwarf plants, crooked
stalks, reduced chlorophyll, and late flowering. All of these mutant variations
gave larger amount of growth in a shorter period of time and clearly showed
heterosis.

The further study of this material has not been completed, but the results
to date indicate that the differences involved are not single genes. Both the
extracted homozygous recessives and the extracted homozygous dominants
from these crosses are larger than the corresponding plants that originally
went into the crosses.

This indicates quite clearly that the visible changes were accompanied or
preceded by other changes with no noticeable effects, but which are expressed
in growth rates. A more complete summary of these results will have to
wait until all of the evidence is at hand. It is a simple matter to extract
the homozygous recessives from these crosses, but it is difficult to extract
the homozygous dominants. Many of the self-fertilized plants proved to be
heterozygous.

GENES CONTROLLING GROWTH

Additional evidence that there are a large number of genes having small
effects on growth without visible morphological changes is becoming clearly
apparent from a backcrossing experiment now in progress. Several long
inbred lines of corn, one of which is now in the forty-first generation of con-
tinuous self-fertilization, were outcrossed to unrelated inbred lines having
dominant gene markers which could be easily selected. The markers — red

PLASMAGENES AND CHROMOGENES IN HETEROSIS 231

cob, yellow endosperm, and non-glossy seedlings— were chosen because Ihey
had little or no effect on growth of the plant.

The first generation outcrosses showed the usual large increases in size of
plant, time of flowering, and yield of grain that is expected in crosses of un-
related inbred strains of corn. The hybrid plants were backcrossed as seed
parents with pollen from the inbred with the recessive gene marker. In every
generation, plants with the dominant gene marker were selected for back-
crossing. These plants have now been backcrossed six successive times. Many
progenies have been grown. They are all heterozygous for the gene marker
plus whatever neighboring regions on the same original chromosome from
the non-recurrent parent that have not been lost by crossing over.

The plan is to continue the backcrossing until no measurable differences
remain between the backcrossed plants and the recurrent parent, or be-
tween the two classes of backcrossed individuals in the same backcrossed
progeny, those with the dominant marker and those with the recessive
marker. When the point is reached where no differences can be detected, the
plan is to compare successive earlier generations from remnant seed to pick
up whatever single gene differences there might be that could be measured
and detected by their segregation.

So far both classes of backcrossed plants in nearly all progenies are taller
and flower earlier, showing that they have not been completely converged to
the parental type (see Table 14.2). The differences are small and not statisti-
cally significant in the tests so far made, but are nearly all in the direction of a
heterotic effect. As yet there are not sufiicient data to base final conclusions.
It is hoped that the comparison of the two classes of backcrossed progeny
with the original recessive parent will permit a distinction between the favor-
able action of dominant genes and an interaction between heterozygous
alleles. Also that it may be possible to show whether or not there is any
residual cytoplasmic effect, since some of the outcrossed plants have the same
cytoplasm as the dominant gene marker and some do not.

Important facts do stand out clearly from this experiment. Since heterosis
still remains after these many generations of backcrossing, it shows clearly
that these three chromosome regions selected as samples have an appreciable
effect on growth. Since the gene markers themselves have no effect on
growth, as far as this can be determined in other material, these three regions
are random selections for growth effects. This indicates quite clearly that
there are genes in all parts of the chromosomes that contribute to normal
growth and development. While the evidence so far available does not per-
mit a clear separation between the effects of an accumulation of favorable
genes as contrasted to an interaction between alleles, or between genes and
cytoplasm, the results show that there are many loci involved in the heterotic
effect in addition to the dominant gene markers.

This follows from the evidence at hand. If the heterosis now remaining

232 DONALD F. JONES

were due solely to the interaction between the dominant and recessive mark-
ers, there would have been a rapid approach to the level of vigor now re-
maining. If it were due to a larger number of genes distributed rather evenly
along the chromosome, the reduction in heterosis would be gradual, as it
has proved to be. Small amounts of heterosis may persist for a long time un-
til all of the genes contributing to it are removed by crossing over.

A recent experiment by Stringfield (1950) shows a difference in produc-
tiveness between an F2 selfed generation and a backcross having the same
parentage. The amount of heterozygosis as measured by the number of
allelic pairs is the same in both lots. In the backcross there are more indi-
viduals in the intermediate classes with respect to the number of dominant

TABLE 14.2

INCREASE IN HEIGHT OF PLANT IN SUCCES-
SIVE BACKCROSSED GENERATIONS HET-
EROZYGOUS FOR A DOMINANT GENE
MARKER

Number of

Per Cent Increase in Height

Generations
Backcrossed— >

4

5

6

20yX243Y

20yXP8 Y

20pX243P

243glX20Gl

6.7
1.9
6.6

2.2
2.3
3.0

1.5

1.2

1.1

-1.3

genes. This indicates a complementary action of favorable dominant genes.
Gowen et al. (1946) compared the differences in egg yield in Drosophila
between random matings, 47 generations of sib mating, and homozygous
matings by outcrossing with marker genes. The differences are significant,
and indicate a large number of genes having dominant effects on the repro-
ductive rate.

INTERACTION OF GENES AND CYTOPLASM

The suggestion has been made many times that heterosis may result from
an interaction between genes and cytoplasm. Within the species, differences
in reciprocal crosses are rare. In commercial corn hybrids, reciprocal differ-
ences are so small that they can usually be ignored. Evidence is accumulating
that there are transmissible differences associated with the cytoplasm, and
that these must be considered in a study of heterosis. Small maternal effects
are difficult to distinguish from nutritional and other influences determined
by the genotype of the mother and carried over to the next generation.

The cross of the two different flowering types of tobacco previously cited
shows a maternal effect. The first generation cross of the indeterminate or

PLASMAGENES AND CHROMOGENES IN HETEROSIS 233

non-flowering type as seed parent grows taller than the reciprocal com-
bination, and flowers later. These difi"erences are statistically significant.

Reciprocal crosses between inbred California Rice pop, having the small-
est seeds known in corn, with inbred Indiana Wf9 having large embryos
and endospersms, show differences in early seedling growth and in tillering.
Inbred Wf9 produces no tillers. California Rice, also inbred, produces an
average of 4.1 tillers per plant. The first generation cross of Rice popXWfQ
averages 1.0 tillers, while the reciprocal combination under the same condi-
tions produced 2.2 tillers per stalk. In this case the non-tillering variety,
when used as the seed parent, produces more than twice as many tillers.
This seems to be a carry-over effect of the large seed. Tillering is largely de-
termined by early seedling vigor. Anything that induces rapid development
in the early stages of growth tends to promote tillering.

PLASMAGENES AND CHROMOGENES

In addition to these transitory effects there are many cases of cytoplasmic
inheritance. Caspari (1948) has reviewed the evidence from fungi, mosses,
the higher plants, and from Paramecium, insects, and mammals to show
that many differences do occur in reciprocal crosses and that they persist into
later backcrossed generations. Reciprocal differences in the amount of
heterosis have been demonstrated in Epilobium (Michaelis, 1939) and in
mice (Marshak, 1936).

Cytoplasmic pollen sterility has been found in Oenothera, Streptocarpus,
Epilobium, flax, maize, onions, sugar beets, and carrots. In every case that
has been adequately studied, the basic sterility remains the same in repeated
generations of backcrossing, but the amount of pollen produced varies in
different genotypes. There is an interrelation between plasmagenes and
chromogenes determining the final result (Jones, 1950).

In maize the amount of pollen produced ranges from 0 to 100 per cent.
Only by suitable tests can these cases of full fertility be recognized as having
any cytoplasmic basis. Interest in this problem now centers on the effect of
these cytoplasmic differences on heterosis.

A series of standard inbreds have been converted by crossing these onto
suitable sterilizer stocks, and backcrossing a sufficient number of generations
to re-establish completely the inbred, and maintaining the inbred in a sterile
condition by continuous backcrossing. It has been found necessary to select
both the cytoplasmic sterile seed parent individuals and the individual fertile
pollen parents for their ability to maintain complete sterility both in inbreds
and in crosses. In some lines it has proved to be impossible to establish com-
plete sterility, but the majority are easily sterilized and maintained in that
condition.

A comparison of fertile and sterile progenies in inbreds, in single crosses of
two inbreds, and multiple crosses of three and four inbreds, shows that this

234

DONALD F. JONES

cytoplasmic difference has no appreciable effect on size of plant as measured
by height at the end of the season, in days to silking, or in yield of grain.
The results are given in Table 14.3. With respect to pollen sterility-fertility,
the cytoplasm has no effect on heterosis.

In the conversion of standard inbreds to the cytoplasmic sterile pollen
condition, it has been found that many of these long inbred strains, presum-
ably highly homozygous, are segregating for chromogenes that have the abil-
ity to restore pollen fertility. In normally fertile plants these genes have no
way of expressing themselves. They are not selected for or against unless
they contribute in some way to normal pollen production. It is one more

T.\BLE 14.3
A COMPARISON OF FERTILE AND STERILE MAIZE PLANTS

Fertile

Sterile

5 Inbreds

7 Crosses of two inbreds. .
7 Crosses of two inbreds .
3 Crosses of three inbreds.

1 Cross of three inbreds. . .
3 Crosses of four inbreds

5 Crosses of four inbreds. .

2 Crosses of four inbreds .

72.3
102.6

58.5
111.7

99.1
123.9

61.1
115.8

70.1 Height of Stalk

97.7 Height of Stalk

58.3 Days to first silk
108.9 Yield, bushels per acre
103.3 Yield, bushels per acre
119.0 Yield, bushels per acre

64.5 Yield, bushels per acre
117.3 Yield, bushels per acre

14 Crosses, average yield. . . .

102.8

102 . 6 Yield, bushels per acre

source of evidence to show that there is a considerable amount of enforced
heterozygosity in maize. Even highly inbred families remain heterozygous.
This has been shown to be true for other species of plants and animals.

SUMMARY

Specific evidence from a study of chlorophyll production in maize and
from similar studies in Neurospora, Drosophila, and other plant and animal
species proves conclusively that there are numerous mutant genes that re-
duce the ability of the organism to grow and to survive. Such genes exist
in naturally self-fertilized and cross-fertilized organisms and in arti-
ficially inbred families such as maize. The normal alleles of these mutant
genes show either complete or partial dominance, and any crossbred indi-
vidual contains a larger number of these dominant, favorable alleles than
any inbred individual.

Evidence from Nicotiana shows that there is an interaction between di-
vergent alleles at the same locus such that the heterozygote produces a larger
amount of growth and a higher reproductive rate than either homozygote.
There is no increase in growth rate and this instance is considered not to be
heterosis. The assumption of an increased growth rate, or true heterosis, in
such allelic interactions is not supported by specific evidence that cannot be

PLASMAGENES AND CHROMOGENES IN HETEROSIS 235

interpreted in other ways. The experimental evidence to date does not dis-
tinguish clearly between a general physiological interaction and a specific
contribution from favorable dominant effects. More evidence on this point is
needed.

Backcrossing experiments in maize, where dominant gene markers are
maintained in a heterozygous condition, show heterosis continuing to the
sixth generation. The approach to the level of growth activity of the recur-
rent inbred parent is so slow as to indicate that every region of the chromo-
somes, divisible by crossing over, has an effect on growth.

The growth rate in these backcrossed generations is maintained at a level
appreciably above the proportional number of heterozygous allelic pairs.
This effect can be interpreted in a number of ways other than a general
physiological interaction, such as enforced heterozygosity, and the comple-
mentary action of dominant genes at different loci.

There is no way known at the present time to distinguish clearly between
the accumulation of favorable dominant effects of compound or multiple
genes at the same loci and a general physiological interaction or over-
dominance.

Reciprocal crosses differ in many species, resulting in appreciable diver-
gence in the amount of growth, and these differences have a cytoplasmic basis.
The evidence from maize, however, shows clearly that cytoplasmic pollen
sterility has no effect on size of plant, time of flowering, or productiveness.

M. R. IRWIN

Universify of Wisconsin

Chapter 15

Specificity
of Gene Effects

If an attempt were made to survey all the possible ramifications suggested
by the title of this paper, it should include much of the published work in
genetics. It is of course a truism to all students of genetics to state that some
sort of differential specificity towards the end product must exist between
allelic genes or their effects could not be studied. It would be very interesting
as a part of this discussion to attempt to trace the change in concepts held by
various workers during these past fifty years concerning the nature and paths
of action of the gene. However, beyond a few remarks, such considerations
are hardly within the scope of this chapter.

Since the effects of genes can be recognized only if there are differences in
the end product, it is quite natural that the differences in the experimental
material first subjected to genetic analyses should have been those which
were visible, as differences in form, color, etc. Although the pendulum has
swung somewhat away from intensive investigations of such hereditary char-
acteristics, it should be emphasized that by their use the underlying mecha-
nisms of heredity have been elucidated.

Major attention was given by most of the investigators during the first
quarter of this century to the effects of respective genes upon individual
hereditary characters. In some quarters there was an oversimplification in
the interpretation of the relation of the gene to the character affected by it.
Gradually, however, the concept has become clearer that the majority of
hereditary characters — even many of those which had previously appeared
to be most simply inherited — are affected by many genes.

An early observation of gene specificity, too long neglected by all but a few
geneticists, was that made by Garrod in 1909 (see 1923 edition of Inborn

* Paper No. 433 from the Department of Genetics, University of Wisconsin.

236

SPECIFICITY OF GENE EFFECTS 237

Errors of Metabolism) on the inability of some humans to break down homo-
gentisic acid (2,5-dihydroxy])henylacetic acid), resulting in the disease known
as alcaptonuria. Observations reported by Gross (1914) indicated that this
affliction was due to the lack of a ferment (enzyme) in the serum of alcapto-
nurics, whereas the enzyme capable of catalyzing the breakdown of homo-
gentisic acid was demonstrable in the serum of normal individuals. As
Beadle (1945) has stated, no clearer example exists today that "a single gene
substitution results in the absence or inactivity of a specific enzyme and that
this in turn leads to the failure of a particular biochemical reaction." (The
writer distinctly remembers that, while he was a student in a class in physio-
logical chemistry, the instructor paid considerable attention to the chemical
explanation of alcaptonuria, but none at all to its hereditary nature.)

Another example of gene specificity and also of gene dosage is that of yel-
low endosperm in corn and the content of vitamin A reported by Mangelsdorf
and Fraps (1931). Their study showed that the amount of vitamin A in the
endosperm of white corn was almost negligible, but that the presence in the
endosperm of one, two, or three genes for yellow pigmentation was accom-
panied by corresponding increases in the amounts of the vitamin.

GENE EFFECTS IN A SERIES OF REACTIONS

There are numerous examples which have shown that many genes con-
tribute to the development of a heritable character. Thus, in corn there are
many genes which affect the development of chlorophyll. Each recessive
allele, when homozygous, allows the formation of only partial pigmentation,
or in extreme cases no pigmentation at all, and the seedlings are albino. It is
generally believed that the majority, if not all, of these different genes for
albinism affect different steps in the process of chlorophyll development. A
breakdown of the process at any one of these steps results in albinism of the
seedling. Haldane (1942) has likened the complexity of such a synthetic proc-
ess to the activity of an equal number of students as there are genes, "en-
gaged on different stages of a complicated synthesis under the direction of a
professor, except that attempts to locate the professor have so far failed. Or
we may compare them to modern workers on a conveyor belt, rather than
skilled craftsmen each of whom produces a finished article."

One of the earliest examples of the physiological bases of the specificities
which are the final gene products is that of the chemical analyses of genetic
variations in flower color. These studies were carried out in England by sev-
eral workers. See reviews by Beadle (1945), Beale (1941), Haldane (1942),
Lawrence and Price (1940) for the general results and references to specific
papers.

Mention will be made here of only one of the many investigations which
have defined in chemical terms the hereditary differences in pigmentation.
Anthocyanin is one of the five types of pigments concerned in flower color,

238 M. R. IRWIN

and its presence or absence in several species is genetically determined. One
way in which anthocyanin may be modified is by the degree of oxidation of
the prime ring. According to Beale (1941) in the two genera Lathyrus and
Streptocarpus, the hydroxyl group is at position 4' in the pelargonidin type,
at positions 3' and 4' in the cyanidin types, and at 3', 4', and 5' in the delphin-
idin types. The more oxidized pigments are usually dominant to the less
oxidized types. Thus flowers with genes AB and -46 will be of the delphinidin
type of pigment, those with aB of the cyanidin type, and those with ab of the
pelargonidin type.

These and other extensive chemical studies on the anthocyanin pigments
genetically modified in various ways are dramatic examples of the specifici-
ties of gene effects. The analogy drawn above between the various genes and
students working on a complicated synthesis becomes a little more clear in
relation to flower pigments, since considerable information is available as to
what some of the genes accomplish.

A further example of the effect of many genes upon a character is that of
eye color in Drosophila melanogaster. Between twenty-five and thirty genes
are known to modify the brownish-red color of the wild-type eye. There ap-
pear to be two independent pigments, brown and red, concerned in the devel-
opment of the wild-type eye, each of these being affected by specific genes.
Certain components of the brown pigment are diffusible from one part of the
body to another, and hence are more readily subjected than others to chem-
ical analyses.

The details of these analyses are presented in other review articles (Beadle,
1945; Ephrussi, 1942a, 1942b). Briefly, dietary tryptophan is converted to
alpha-oxytryptophan by a reaction controlled by the wild-type allele of the
vermilion gene (?). This substance is oxidized further to kynurenine (the so-
called i'+ substance). By virtue of the activity of the normal allele of the cin-
nabar gene, kynurenine is further oxidized to the or substance, which may
be the chromogen of the brown pigment (Kikkawa, 1941). The production of
either brown or red eye pigment can be blocked by genes at the white eye
locus, thus indicating that such genes act on a common precursor of the red
and brown pigments.

Mention should be made of the relation between the original designation
of certain of the genes for eye color and their presently known effects. Thus,
the eyes of flies with the mutant alleles bw bw are brown. But it is now known
that this pair of alleles, instead of being concerned with the production of
brown pigment, restricts the development of red pigment and thus we see
only the brown color. Similarly, the four gene pairs whose mutants modify
the red coloration do so by virtue of their effect on the brown pigment, not
upon the red.

Wheldale (1910) proposed four decades ago that genetic characters were
the resultant of a series of reactions, and that if a break in the chain occurred,

SPECIFICITY OF GENE EFFECTS 239

the series of steps would have proceeded only to that j)oint. Following the
initial work in Neurospora by Beadle and Tatum (1941) on mutants which
blocked certain metabolic processes, this type of approach has expanded
enormously and profitably. Attention can be called here to but one very sig-
nificant example of this kind of experimental study in microorganisms. A
report by Srb and Horowitz (1944) shows clearly how many genes act in the
synthesis of arginine. Of fifteen mutant strains studied, there were seven dif-
ferent steps represented in the synthesis of arginine. One of the forms grew
only if arginine was supplied. Two others required either arginine or citrul-
line, and these two strains were genetically different. Four other strains,
genetically different from the first three strains and from each other, would
grow if arginine, citrulline, or ornithine were provided. For a diagrammatic
representation of these steps, see Beadle (1945).

DIRECT EFFECTS OF GENES

The preceding examples are but a few^ of the many which could be cited to
illustrate the gene specificities in the development of a genetic character
which involves the successive activities of many genes. Are there any genetic
characters which may be the immediate products of the causative genes? An
example almost unique in higher plants is that of the waxy gene in corn
(Collins, 19C9) in its effects upon the starch of the pollen grain and the endo-
sperm reserves. As is well known, the starch granules in the pollen grains
bearing the waxy gene are stained reddish-brown with iodine, as are the
endosperm reserves of waxy seeds, in contrast to the typical blue reaction of
the starch granules of non-waxy pollen and of the endosperm reserves of
non-waxy seeds. Following studies of the physiological effects of the waxy
gene, Brink (1929) proposed that this gene has its effect on the enzyme amy-
lase which functions directly in the synthesis of starch.

Another class of hereditary characters which in some respects appears to
satisfy some of the criteria for a direct effect of the causative genes is that of
the antigenic characters of the red blood cells of animals. With only rare
exceptions, to be considered later in more detail, each of the known antigenic
substances has appeared in the cells of an individual only if one or both
parents also possessed it. If there is but a single pair of contrasting charac-
ters, each is expressed in the heterozygote. Further, the cells which give rise
to the hematopoietic tissue from which the red blood corpuscles are derived
are laid down shortly after the first division of the fertilized egg. The possibil-
ity cannot be excluded, of course, that there is a chain of reactions within
each cell leading to the formation of the antigen, but no block in such a chain
of reactions has yet been observed. There are two statements concerning the
cellular antigens which are of interest: (1) the antigenic substance must be
located at or near the surface of the cell in order to be detectable, and (2)
there is no known effect of the environment upon them.

240 M. R. IRWIN

We should avoid misunderstanding about the meaning of the terms com-
monly used in immunological literature. For example, the word antigen was
originally defined as any substance which, when introduced parenterally into
an animal, would invoke the production of antibodies. This definition would
now be extended to include any substance which will react visibly with an
antibody. And an antibody would be defined as a constituent of the serum
which reacts with an antigen in any of several ways. The circle of reasoning
here is obvious. However, insofar as chemical studies of various antigens have
contributed to an understanding of their specificities, the specificities have
always been associated with structural differences of the antigenic sub-
stances. On the other hand, the reasons underlying the specificities in re-
activity of the antibodies are almost completely unknown, although it is
known that the antibodies are intimately associated with the globulins of the
serum, and in fact may constitute the gamma globulins of the serum.

CELLULAR ANTIGENS IN HUMANS

As our first example of these antigenic substances, let us consider the well
known and extensively studied O, A, B, and AB antigenic characters, or
blood groups, of human cells. Following their discovery by Landsteiner
(1900, 1901), it soon became clear that these substances were gene controlled.
At the present time, the theory of three allelic genes, as postulated by Bern-
stein (1924) on statistical grounds, is generally accepted. The two other the-
ories proposed for their inheritance — independent and linked genes, respec-
tively— are fully discussed by Wiener (1943). Landsteiner noted that the
serum of certain individuals would agglutinate (clump) the cells of other
individuals, and from this observation the reciprocal relationship between the
presence and absence of each antigen and its specific antibody has been
elucidated.

A or B Antibody

Antigen on of the

Group the Cells Serum

O None Anti-A, Anti-B

A A Anti-B

B B Anti-A

AB AB None

It may readily be seen that the presence of an antigen, as A, on the cells
is accompanied by the presence of the antibody (anti-B) for the contrasting
antigen, as B, in the serum, and vice versa. If both antigenic characters are
found on the cells, as in AB individuals, the serum contains no antibodies.
While if neither A nor B is present on the cells, the serum contains both anti-
A and anti-B.

These phenomena pose the question whether the genes producing the cel-
lular substances also have an effect on the antibodies of the serum. That is,
does the gene which is responsible for the O antigen (which is definitely an
entity but is less reactive than A and B) also effect both anti-A and anti-B

SPECIFICITY OF GENE EFFECTS 241

in the serum — while in individuals with substance A, only anti-B is found;
in those with B, only anti-A is present; and in AB individuals the effects of
the respective genes on the antibodies are somehow neutralized?

Before attempting to answer this question, it will be advisable to review
the present knowledge of the chemistry of the A and B substances of human
cells. See Rabat (1949).

These antigens (blood groups, cellular characters, antigenic factors, etc.)
are found in nearly all the fluids and tissues of the human body. They also
are widely distributed throughout the animal kingdom. The A substance or
an A-like substance has been found, for example, in hog gastric mucosa, in
the fourth stomach (abomasum) of the cow, and in swine pepsin, while both
A and B substances have been noted in the saliva and stomachs of horses.
Following chemical fractionations, principally of horse saliva and hog gastric
mucosa, various investigators have obtained preparations with activity re-
lated to the A substance. These preparations have been largely polysac-
charide in nature. In addition to the polysaccharides, even in the purest
preparations, some workers have noted traces of amino acids.

At present, while it appears that both the A and B substances of human
cells may be classed as nitrogenous polysaccharides, no information is avail-
able as to the structural differences between them. Our knowledge of such
specificities rests entirely upon the technics of immunology, that is, by the
interaction of either naturally occurring antibodies (as anti-A and anti-B),
or immune antibodies, with the respective substances A and B.

The antigenic substances A and B of human cells are complex polysac-
charides, while the antibodies are modified globulins, or are found in serum
protein very closely related to the globulin fraction. If the gene which effects
antigen A is responsible also for the B antibody, and that for antigen B for the
A antibody, it would seem that here is a clear-cut case of pleiotropic effects of
the respective genes. This explanation runs into difficulties in AB individuals
which, on this proposal, should have both kinds of antibodies but actually
have none. In contrast, a current explanation of the reciprocal presence of the
antigenic substance of the cells and the antibody for the contrasting sub-
stances is that the antibodies for both substances (A and B) are normal con-
stituents of human serum. Production of the antibodies would then be con-
trolled by a gene or genes at another locus than that having to do with the
cellular substances, ij genes were involved in their production. If an indi-
vidual carries the gene for A, and hence has A substances widely distributed
throughout his body, the A antibodies are presumed to be absorbed from the
serum, and of course the B antibodies are left. Also, an individual with the
B substance would absorb the B antibodies, and the antibodies to A would
remain, while both anti-A and anti-B would be absorbed in an AB individual.
Other hypotheses are given by Wiener (1943). Unfortunately, no experimen-
tal test of the correctness of this or other hypotheses is likely.

242 M. R. IRWIN

Landsteiner and Levine (1927) announced the discovery in human cells of
a new pair of contrasting antigens, called M and N. These were detectable
only by the use of immune sera produced in rabbits, as was another antigenic
factor called P. The heritability of the M and N substances is adequately
explained by the assumption of a single pair of allelic genes, and the sub-
stance P appears to be dominant to its absence.

Another antigenic factor in human blood which has aroused wide interest
is the recently discovered Rh substance, or complex, as it might be termed.
In 1940, Landsteiner and Wiener (1940) reported that a new antibody, de-
rived from a rabbit immunized with the erythrocytes of a rhesus monkey,
was reactive with the cells of about 85 per cent of the white population of
New York. They gave the name Rh (a contraction of rhesus) to this agglu-
tinable property of human cells. As Boyd (1945) aptly states:

The technic of testing for the new factor was difficult, the best available serums were
weak, and had it not been for a remarkable series of discoveries which followed in the next
few months, the Rh factor might have aroused no more interest than its practically still-
born brethren. . . .

The Rh factor was shown to be involved in previously unexplained com-
plications following transfusions (Wiener and Peters, 1940), but is most
widely known for its role as the etiologic agent in the majority of cases
of hemolytic disease of the newborn. The proposal was first made by
Levine and Stetson (1939) that an antigen in the fetus, foreign to the mother
and presumably transmitted by the father, could pass through the placenta
and immunize the mother. Later studies implicated the Rh factor as the
foreign antigen, and showed that the antibodies developed in the mother may
pass back through the placenta and affect the red blood cells of the fetus,
before or following birth. Although the majority of cases of hemolytic disease
of the newborn may be justly ascribed to Rh incompatibility between the
father and mother, there is no satisfactory explanation as to why only about
one in forty of such potentially dangerous combinations leads to morbidity.

There exist several subgroups, or subtypes, of the Rh complex, and inves-
tigations as to their respective specificities occupy the center of interest of
many workers at the present writing. There are two schools of thought as to
the mode of inheritance of these subgroups, which also involves the terminol-
ogy to be used in their identification (see Strandskov, 1948, 1949, for leading
references). One explanation is that the various subtypes are manifestations
of a series of multiple allelic genes, the other that they are the result of the
action of respective genes at three different but closely linked loci. It is not
within the province of this chapter to discuss the arguments for and against
these two proposals. However, it should be stated that the genetic results
under either explanation are essentially the same.

One of the most pertinent statements which can be made about these
various antigenic substances of the erythrocytes is that they are detectable

SPECIFICITY OF GENE EFFECTS 243

no matter in what gene complex they may occur. That is, other genes than
the causative ones have no measurable influence upon their expression. A
possible exception to this statement might be proposed for the A and N char-
acters, res])ectively, since each is somewhat less readily agglutinated when in
the heterozygote, AB and MN, than when either occurs singly.

THE HYBRID SUBSTANCE IN SPECIES HYBRIDS
Until the early part of this century, most of the workers in immunology
had reached the conclusion that the specificities obtained in immunological
reactions were primarily if not entirely concerned with proteins. Therefore,
the finding by Heidelberger and Avery (1923, 1924) that the immunological
specificities of the pneumococcal types were dependent upon polysaccharides
was indeed a forward step in our understanding of the chemical nature of
biological specificity. It is a pleasure to acknowledge that this work of
Heidelberger and Avery convinced the writer that immunological technics
should be a useful tool in studying genetic phenomena. Also, although at that
time pollen differing in gene content seemed (and still does) to be promising
experimental material, the species and species hybrids in pigeons and doves
produced by the late L. J. Cole were tailor-made for further studies.

Pigeon-Dove Hybrids

The first step was to determine whether the cells of one species could be
distinguished from those of the other. In brief, all the comparisons by im-
munological technics, between any pair of species of pigeons and doves, have
resulted in the ability to distinguish the cells of any species from those of an-
other, and to show that each species possessed antigenic substances in com-
mon with another species, as well as those peculiar to itself — those species
specific. A dozen or more kinds of species hybrids have been obtained in the
laboratory, and in general, each kind of hybrid has contained in its cells all
or nearly all of the cellular substances of both parental species. One such
species hybrid is that obtained from a mating between males of an Asiatic
species, the Pearlneck (Slreplopelia chinensis) and the domesticated Ring
dove females {St. risoria). The corpuscles of these hybrids contained all the
substances common to each parental species, but did not contain quite all the
specific substances of either parental species. Further, the cells of these
hybrids did possess a complex of antigenic substances not found in the cells
of the parents. These relationships are presented in Table 15.1 and are given
in diagrammatic form in Figure 15.1. This new antigen has been called the
"hybrid substance," and it has been ])resent in every hybrid produced be-
tween these two species.

Upon repeatedly backcrossing these species hybrids and selected back-
cross hybrids to Ring dove, ten antigenic substances which differentiate
Pearlneck from Ring dove have been isolated as probable units. That is, a

244

M. R. IRWIN

backcross bird carrying any one of these unit substances, when mated to a
Ring dove, has produced approximately equal proportions of progeny with,
and without, the particular substance in their blood cells. These substances
peculiar to Pearlneck, as compared with Ring dove, have been called d-1,
d-2, d-3, d-4, d-5, d-6, d-7, d-8, d-11, and d-12. Each of these is distinct from
the others (Irwin, 1939) both genetically and immunologically. Thus it ap-
pears that a gene or genes on each of ten of the thirty-odd pairs of chromo-

TABLE 15.1

ANTIGENIC RELATIONSHIPS OF THE BLOOD CELLS

OF PEARLNECK, RING DOVES, AND

THEIR HYBRIDS

Immune Serum

Pearlneck .
Pearlneck .
Pearlneck .

Ring dove.
Ring dove.
Ring dove .

Fi.
Fi.
F,.
Fi.

Absorbed by
Cells of

Ring dove
Fi

Pearlneck
Fi

Pearlneck
Ring dove
Pearlneck and
Ring dove

Agglutination Titers
WITH Cells of

Pearlneck

23040

11520

90

15360
0
0

15360

0

7680

0

Ring Dove

23040
0
0

15360

3840+
180

15360
3840+
0
0

23040

11520

0

15360
3840+
0

15360
3840+
7680
360+

" -o ^'^ o"q a o

.<>%»o • Pearlneck . <, o.-/

On a oQf, a n M /i 1% n ^ O O O O « O

I

A a. H.Q a a q q •

I

t\\ Pearlneck "aV-sfCs;

Common

Common

Ringdove

1

i

Pearlneck

Ringdove

Common-

Ringdove^ P'N. x R,D.-F,

Hybrid

Fig. 15.1 — Diagrammatic representation of the antigenic relationships of the Pearlneck,

Ring dove, and their hybrids.

SPECIFICITY OF GENE EFFECTS 245

somes of Pearlneck produce effects on cellular antigens which differentiate
Pearlneck from Ring dove. Although the cellular substances particular to
Ring dove, in contrast to Pearlneck, have not been obtained as units, the
available evidence indicates strongly that a gene or genes on nine or ten
chromosomes of Ring dove produce antigenic effects which differentiate that
species from Pearlneck.

The question may well be raised as to what this recital of antigenic char-
acters in man and doves, which in general illustrates gene specificity in the
production of cellular antigens, has to do with the general topic of heterosis.
The so-called hybrid substance has one word (hybrid) in common with the
term hybrid vigor, and suggests a possible relationship of the two terms.

The hybrid substance seemingly represents a departure from the hypoth-
esized direct action of a gene on the antigenic substance, in that it appears
to result from the interaction of two or more genes in the species hybrids to
produce some antigenic substance different from any detectable in either
parent. With but one exception proposed by Thomsen (1936) in chickens, and
for which another explanation will be considered shortly, a hybrid substance
has thus far been found only in species hybrids.

Mention should be made of the technics required for the detection of the
hybrid substance. Briefly, if an antiserum prepared against the cells of an
individual, whether a species hybrid or not, would be absorbed by the cells of
both its parents and would then react with the cells of the individual, but not
with the cells of either parent, there would be evidence of a different anti-
genic substance in the homologous cells — those used in the immunization.
(If an antigen were recessive, it would be present in the heterozygote, and
presumably could absorb its specific antibody.)

Domestic Fowl Hybrids

As stated above, Thomsen (1936) reported that within each of two families
of chickens there was a different antigenic substance present than was found
in the parents. Attempts in our laboratory by Mrs. Ruth Briles to duplicate
this finding were without success, but a very interesting and quite unex-
pected observation was made which may be the explanation of Thomsen 's
finding. If an antigenic substance were present in an individual different from
that possessed by either parent, immunization of either parent (as #1) with
the cells of this individual might engender antibodies against the new sub-
stance. Absorption of such an antiserum by the cells of the other parent (as
#2) should remove all antibodies except those formed against the new or
hybrid substance, and such a reagent should be reactive only with the cells
containing the new substance. This was the procedure followed by Thomsen,
except that his tables do not show that the cells of the two parents were used
as negative controls in the tests made after the various absorptions.

Immunizations of each of the parents of a family of chickens against the

246 M. R. IRWIN

cells of one of the offspring, or the pooled cells of two or more, were made by
Mrs. Briles. Following the absorption of the antiserum obtained from either
parent by the cells of the other, it was noted that the absorbed antiserum was
at least weakly reactive with the cells of the individual from which the anti-
serum was obtained. That is, such an antiserum would not react (agglutinate)
with its own cells before absorption with the cells of the mate, but after such
absorption it definitely would agglutinate the cells of the individual from
which it was derived.

To use a concrete example, bird R614 (containing Bi antigen) was im-
munized with the washed cells of R2C43, to produce B3 antibodies (Briles,
McGibbon, and Irwin, 1951). After this antiserum from R614 was mixed for
absorption with the washed cells of R622 (having B3 antigen in its cells and
having been immunized to produce Bi antibodies), all cells containing the Bi
antigen were reactive with it, including those of R614 itself. Thus it appears
that the antibodies to Bi which were circulating in the serum of R622 were
also attached to the surface of the red blood cells and were transferred to the
antiserum from R614 during the absorption process. It was possible to dem-
onstrate that, after washing the cells of R622 in saline, the saline contained
antibodies, even after nine successive washings. Hence, unless the cells of
both parents were used as controls in comparable tests for the presence of a
hybrid substance, agglutination of any cells could be explained as due to a
transfer of antibody from the blood cells to an antiserum. Unfortunately,
such controls are not given in Thomsen's paper, and the possibility cannot be
eliminated that the reactions obtained by him were due simply to segregation
within the various families of an antigenic character of one of the parents.
This possibility was mentioned by Thomsen (1936), but was not considered
applicable to his experiments.

Hybrid Substances

Returning to the hybrid substance for which there is definite evidence, it
should first be stated that such a substance has not been found in all kinds of
species hybrids, as may be seen from the data given in Table 15.2. It has been
reported from our laboratory in hybrids between Pearlneck and Ring dove,
the pigeon {Columba livia) and Ring dove, the Mallard {Anas platyrhynchos)
and Muscovy duck (Cairina moschata), but not in the hybrids between the
triangular spotted pigeon (C. guinea) and livia. Irwin (1947) gives the spe-
cific references to pertinent articles.

A hybrid substance has been detected but not previously reported in
hybrids from malings between the Philippine turtle dove (St. dussumieri)
and Ring dove, the dwarf turtle dove {St. Jmmilis) and Ring dove, the
Oriental turtle dove {St. orientalis) and Ring dove, and the band tail pigeon
{C. fasciata) and livia. No such substance has been observed in the hybrids
between the Senegal dove {St. senegalensis) and Ring dove, an African dove

SPECIFICITY OF GENE EFFECTS

247

(St. semitorquata) and Ring dove, the Senegal dove and the Cape turtle dove
{St. capicola), the spot wing pigeon (C. maailosa) and livia, and between the
Grayson dove {Zenaidiira graysoni) and the common mourning dove {Zen.
macroura). It is possible that a hybrid substance does exist in these latter
species hybrids, but the same technics by which it was observed in the other
species hybrids failed to demonstrate its presence in them.

Three different fractions of the hybrid substance have been demonstrated
in the hybrids between Pearlneck and Ring dove (Irwin and Cumley, 1945),
by virtue of a frequent association of each fraction with one or more antigens

TABLE 15.2

TESTS FOR HYBRID SUBSTANCES IN THE CELLS OF

VARIOUS SPECIES HYBRIDS

Antiserum to

Fi — Pearlneck X Ring dove

Fi — C. //zi/aX Ring dove

Fi — St.dussumieriXRing dove. . . .

Fi — St. humilisXRing dove

Fi — St. orientalisXRing dove

Fi — C. fasciataXlivia

Fi — Mallard X Muscovy

Fi — St. senegalensisXRing dove. . .
Fi — St. semitorquataXRing dove. .

F: — Senegal X-S/. capicola

Fi — C. tnaculosaXlivla

Fi — C. guineaXlivia

Fi — Zenaidura gray soniX Zen. ma
crmira

Absorbed by Cells of

Parent 1

Parent 2

Pearlneck

livia

dussumieri

h II mil is

orientalis

C. fasciata

Mallard

St. senegalensis

St. semitorquata

Senegal

C. maculosa

C. guinea

Zen. graysoni

Ring dove
Ring dove
Ring dove
Ring dove
Ring dove
livia

Muscovy-
Ring dove
Ring dove
St. capicola
livia
livia

TLen. macroura

Re.actions of Pa-
rental AND HYBRID

Cells with the

Respective

Reagents

Par-

Par-

ent 1

ent 2

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

Hy-
brid

+ +

+ +

+

+

+

+

+ +

0

0

0

0

0

0

peculiar to Pearlneck. Thus one fraction called dx-A was always associated
in the backcross hybrids with the d-1 1 substance, dx-B seemingly was loosely
linked with the d-1 character and with certain others as well — thereby pro-
viding strong evidence that on several chromosomes of Pearlneck there are
duplicate or repeat genes — and dx-C was always associated with the d-4
antigen. The pertinent reactions which show these specificities are given in
Table 15.3 and are represented diagrammatically in Figure 15.2.

Because of the constant association of the dx-A and dx-C fractions with
the d-11 and d-4 substances, respectively, one cannot be certain that these
two fractions, although antigenically distinct from the d-1 and d-11 specific
characters, are not simply a new specificity conferred upon the specific char-
acters by some sort of rearrangement of the specific substances following the

248

M. R. IRWIN

interaction of the causative genes. This question cannot be completely an-
swered until either a genetic separation has been observed, as between the
dx-A and d-11, or the chemical separation into two distinct substances has
been done. On the other hand, the dx-B fraction has been separated from
each of the species specific characters to which it presumably is loosely
linked, thereby showing that this fraction of the hybrid substance is an
antigenic entity.

The reagent which interacts with the hybrid substance (hybrid antiserum

TABLE 15.3

TESTS FOR SIMILARITIES AND DIFFERENCES OF THE COM-
PONENTS OF THE "HYBRID SUBSTANCE" OF THE SPECIES
HYBRID BETWEEN PEARLNECK AND RING DOVE

Reactions of Different Cells with Anti-Fi Serxjm

Cells

Absorbed

by Cells

of Both

Pearlneck

and Ring

Dove

Absorbed by the Cells of Both Pearlneck and Ring Dove,
in Combination with Others as Listed

d-1
(dx-B)

d-4

(dx-C)

d-11
(dx-A)

P.N.
^iR.D.

Pgn
^iR.D.

Sen.

Aus.
cstd.

Pearlneck

Ring dove

Fi— P.N./R.D

d-1 (dx-B). . . .

d-4 (dx-C) ....

d-11 (dx-A). . .

Fi-Pgn/R.D

Senegal

Australian crested.

0
0

++

+

+

++

+

+

±

0
0

++

0

?

++
+

0

+

0
0

++
+

0

++
+

0

+

0
0

-1-
+

-f

0

0

0

0

0
0
0
0
0
0
0
0
0

0
0

+ +
+

-1-

+
0
0
0

0
0

++
+

-f-

++
+

0

±

0
0

+ +

+
-f

+
-1-
0

Column

2

3

4

5

6

7

8

9

Symbols: -f -f = marked agglutination; -|- = agglutination; ± = definite but weak agglutination; ?
doubtful reaction; 0 = no agglutination — at the first dilution of the serum cell mixture.

absorbed by the cells of both Pearlneck and Ring dove) will also agglutinate
the cells of various species. Thus in the genus Streptopelia, there were five
species {capicola, dussutnieri, kumilis, orientalis, and senegalensis) other than
Pearlneck and Ring dove whose cells were reactive, and one {semitorquata)
with nonreactive cells. Within the genus Columba, the cells of one species
(rufina) likewise reacted with this reagent, but those of seven other species
(fasciala,flavirostris, guinea, livia, maculosa, palumbus, and picazura) did not.
And of twelve species tested in other genera within the Columbidae, only
three species from Australia (Australian crested dove, or Ocyphaps lopkofes,
the bronze wing dove or Phaps chalcopiera, and the brush bronze wing dove,
or Phaps elegans) possessed reactive cells. In Table 15.3, the Senegal cells are
representative of the parallel reactions of the five species of the Streptopelia,

SPECIFICITY OF GENE EFFECTS

249

as are those of the Australian crested of the equivalent reactivities of the
three Australian species.

Although the reagent for the hybrid substance did not agglutinate the
cells of livia, it invariably clumped those of the hybrids between pigeon
(livia) and Ring dove. As previously reported (Irwin and Cole, 1936), these
hybrids also contain a hybrid substance. Because of the cross reactions exist-
ing between these two hybrid substances, a certain degree of similarity can
be assumed. That the fraction in the hybrid substance of the Fi-Pearlneck X

DIAGRAMMATIC REPRESENTATION OF THE'HYBRID SUBSTANCE"
OF THE HYBRID BETWEEN PEARLNECK AND RINGDOVE

VPEARLNECK>-

^^■•^V-VV:-

COMMON- >•:•.■•. '.<,"•.•••:

RINGDOVE

O O O O O

0 0 O O O

O O O ^ Q

o O O OO

O 0 0 0 o

n PEARLNECK
^i RINGDOVE

HYBRID SUBSTANCE

(d-ll)

Cd-i) (d^)

M

O 0« OO

O O o o o
0 O o 00
O 0 £7 oO
0 0 O Oft
o a o OO

0 0 a 00
0 a aao

G O 0 QtJ
O 0 0 00

dx-A

d<-B

dii-C

MrBRIO SUBSTANCE ASSOCIATED

WITH SPECIFIC PEARLNECK SUBSTANCES

RELATED SUBSTANCES

Fig. 15.2— The separation into constituent parts of the hybrid substance of the species

hybrid between Pearlneck and Ring dove.

Ring dove, which is primarily if not entirely responsible for the cross reac-
tions, is dx-A may be deduced from Table 15.3, in that this fraction (associ-
ated with d-ll) is the only one which will exhaust the antibodies from the
reagent for the cells of the pigeon-Ring dove hybrid (column 5). Also, in
unpublished tests the reagent for the hybrid substance of the pigeon-Ring
dove hybrid (anti-hybrid serum absorbed by the cells of pigeon and Ring
dove) did not react with Pearlneck cells, but reacted strongly with d-ll cells,
presumably by virtue of their content of the dx-A fraction, and not definitely
with cells carrying dx-B or dx-C. If the dx-A hybrid substance of the species
hybrids between Pearlneck and Ring dove were partially or largely a rear-
rangement of an antigenic substance, in this case d-ll, which is species spe-
cific to Pearlneck — since the Ring dove is a common parent of the two kinds

250 M. R. IRWIN

of species hybrids — that specific substance (cl-11) should be detectable in the
cells of livia.

To date, reasonably extensive tests (unpublished) have not shown that the
cells of livia contain more than a trace of an antigenic substance related to the
d-11 of Pearlneck. Whatever the relationship of the genes in Pearlneck (as-
sociated with those on a chromosome effecting the d-11 specific substance)
and livia, respectively, which presumably by interaction with a gene or genes
from Ring dove in the two species hybrids effect a common fraction of the
two hybrid substances, they are not associated with genes which produce
similar antigenic patterns in the two species. On these grounds, it would seem
unlikely that the hybrid substances in these two kinds of species hybrids are
merely a different arrangement of a species specific antigen.

The question is pertinent as to whether such reactivities in the cells of
these other species, as Senegal and Australian crested, are themselves an in-
dication of antigenic response to gene interaction within each species, or the
more direct product of a gene. This cannot be answered directly. But, as
given in Table 15.3, the fact that absorption of the reagent for the hybrid
substance by fractions dx-A, dx-B, or dx-C removes the antibodies for the
cells of Senegal indicates that there is some common constituent of these
three fractions related to, if not identical with, a reactive substance in Senegal
cells. However, only the dx-A fraction removes the antibodies for the cells of
the Australian crested dove. Further, absorption by the cells of the pigeon-
Ring dove hybrid also removes the antibodies from this reagent for the cells
of the Australian crested dove.

The hypothetical explanations could be advanced, (1) that the antigenic
substances in Senegal and the Australian crested dove, themselves being dis-
tinct, but both related to the hybrid substance in Pearlneck-Ring dove
hybrids, are the result of a genie interaction. But there is no evidence for such
an assumption. Also, (2) the argument could be advanced that the relation-
ship between these substances in Senegal and Australian crested, and in the
respective species hybrids, is fortuitous, simulating the occurrence of the
Forssman antigen in many species of animals and plants, including bacteria
(Boyd, 1943). That is, the antigenic substances involved (related in some
manner to the hybrid substance) may be gene controlled in each of the re-
lated species, since indistinguishable substances to those of Senegal were
found in four other species of Streptopelia, capicola, dussumierl, humilis, and
orienlalis, and to those of the Australian crested dove in two species of an-
other genus, Phaps chalcoptera and Phaps elegans, but the antigenic similarity
to the hybrid substance is by virtue of some related antigenic component.
Various ramifications of these and other explanations would be purely
speculative.

The hybrid substance, as it has been observed in the cells of various species
hybrids in birds, simulates for cellular antigens the expression of heterosis in
plants and animals. That is, it appears as the resultant of an interaction be-

SPECIFICITY OF GENE EFFECTS

251

tween genes. One may well ask if there is any other manifestation of heterosis
in these species hybrids and backcross hybrids. Extensive measurements of
eight body characteristics, as over-all length, extent, width of tarsus, width of
band, length of wing, beak, middle toe, and tail, were made over a 7)eriod of
years under the supervision of L. J. Cole. The differences in the averages of
these various characteristics between the parental Pearlneck and Ring dove
species, as yet unpublished, were statistically significant, and the averages of
the measurements of these characteristics in the species hybrids showed them
to be in general intermediate between those of the parental species. Thus
there was no evidence of heterosis in any external characteristic of the species
hybrids, and no correlation with the hybrid substance of the blood cells.

CELLULAR CHARACTERS WITHIN A SPECIES

The finding that one or more genes on each of nine or ten pairs of chromo-
somes of Pearlneck had effects on the species specific antigens of the blood
cells of this species made plausible the belief that many more genes than
commonly believed would have effects within a species making for individual-
ity of the cellular patterns. Acting on this assumption, a series of exploratory
tests w^ere made in experimental animals, principally in cattle and chickens.
For example, following the transfusion of the blood of a young cow into her
dam, an antibody was obtained from the serum of the recipient which re-
acted (produced lysis of the reacting cells upon the addition of complement
to the serum-cell mixture) with the cells of some individuals, but not with
those of others. The reactive substance was called A.

The objective was to be able to detect each antigenic factor separately, ac-
cording to the following criterion. The reactive cells from any individual
should remove the antibodies from the reagent specific for those cells, when
added in excess to the reagent. However, if there were antibodies in the re-
agent which recognized two or more distinct blood factors, any such ab-
sorption with cells containing only one such substance would remove only a
part of the antibodies. Those remaining would still be reactive with all cells
containing the substance corresponding to the unabsorbed antibody.

To this criterion was added that of genetics for a single character, using
the gene-frequency method since controlled matings were not possible. A
typical example of the analysis is that for substance A, as follows:

Typk ok
Mating

Number of offspring

With
Antigen A

Lacking
Antigen A

AXA

AX-

217

76

0

23
51

-X-

41

252

M. R. IRWIN

These results illustrate the observation that an individual has any cellular
character recognized to date only if one or both parents possessed it. Also,
each behaved as if it were a dominant to its absence.

From further isoimmunizations in cattle, and from immunizations of rab-
bits, various antisera have been obtained which detect other antigenic factors
of cattle cells. Each of these has been subjected to the criteria of both genetics
and immunology for a single character, as described in reports by Ferguson
(1941), Ferguson et al. (1942), and Stormont (1950). At present, about forty
different reagents are regularly used in typing cattle cells.

Other Antigens in Cattle

As stated above, the first substance detected in cattle cells was named A.
The next was called B, the next C, . . . Z. That called A' implies no relation-
ship to A, nor B' to B, etc. Each of these antigenic factors is therefore recog-
nized independently, and when subjected to an analysis of gene frequency,
each has behaved as expected if effected by a single gene in comparison to its
absence.

However, some definite associations have been noted among them. For
example, Ferguson (1941) reported that the C and E factors were not inde-
pendent, for only C occurred alone, whereas E was present always with C,
and such cells therefore had CE. It was postulated that there were three al-
lelic genes involved, one for the components C and E together, one for C
alone, and a third for the absence of both C and E.

It was later noted by Stormont that certain additional antigenic factors
appeared only if one or more other components also were present. For ex-
ample, the substance B occurs alone, as does that called G. But a third factor
called K has never been observed unless both B and G were also present.
(A possible exception to this rule was noted shortly after these factors were
first demonstrable, and a weak reaction at that test with the reagent for the
G substance was probably incorrectly recorded.) This association of K with
B and G has been noted in over eighteen hundred animals of more than six
thousand tested. Hence the combination of the BGK factors has always oc-
curred as a unit, and it has also behaved as a unit in the progeny of individ-
uals possessing it. A compilation of some unpublished data has yielded the
following information:

Number of Offspring

Type of Mating

With
BGK

Without
BGK

BGKXBGK

BGKX-

151

185

0

44
137

— X —

160

SPECIFICITY OF GENE EFFECTS 253

Notwithstanding the fact tliat H, (1, and K are recognized sei)arately by-
respective reagents, these data, and the observation that K has occurred
only with both H and G, are strong evidence for tlie conclusion that B, G, and
K in the cells behave as a unit.

Further, offsj)ring of some individuals ])ossessing B and G (BG) in their
cells have given only two classes of offspring, those with B and those with G,
as would be expected if the causative genes were alleles. But another type of
BG individual has produced offspring of two quite different types — those
with both B and G (BG) and those with neither, as if a gene producing B and

TABLE 15.4

THE DISTRIBUTION OF THE CONSTITUENTS OF

THE "B" COMPLEXES IN THE OFFSPRING

OF SELECTED SIRES

Sire

Antigenic
Complex

Number
of Off-
spring

Antigenic
Complex

Number
of Off-
spring

H-1

H-4

H-5

H-6

H-7

H-11

H-19

G-19

BbgioiTza'

Bb02A'E3
BbOiY2D'

BbOi
Bbgke'.
Boia'
Bgy2e;

BlE'i

25
35
26
15
14
31
19
8

BoiYjA'
Bo3J'k'
Bo3j'k'

BgYjE'i

BgyoeJ

B03I'k'

Bb
Bb

23
31
24

23
15
23
13

7

G together was allelic to one not effecting either B or G. These combinations
of antigenic substances, as BG and BGK, have been called antigenic com-
plexes.

There are two series of such complexes, called the B and C series, respec-
tively. In the B series there are twenty-one of the forty-odd antigenic char-
acters which are associated in various conbinations. At least seven of these
may appear singly, as was described for B and G. The other fourteen have
been found only in various antigenic complexes, each of which may be made
up of from two to eight of the twenty-one characters. The majority of these
twenty-one characters do not occur at random in a complex with each of the
others. As was stated above, the character K has always been found with B
and G, but it has never occurred with I, with which it appears as a contrast-
ing substance. In contrast, either B or G may be present in a complex with I.
No separation of the antigenic characters of a complex has ever been ob-
served in the cells of the offspring of an individual possessing it. A few ex-
amples are listed in Table 15.4 from more complete data given in a paper by
Stormont, Owen, and Irwin (1951). All present evidence makes it seem some-
what more reasonable to assume that each antigenic complex is produced by
a single gene than by linked genes. The various antigenic complexes in each

254 M. R. IRWIN

of the two systems, or series, would then be produced by a series of multiple
alleles. The possibility of pseudo-alleles cannot be eliminated, but for the
present may be assumed not to be a complicating factor.

If the assumption be granted that a single gene controls an antigenic com-
plex, as BGK, what explanation or explanations can be proposed for the dif-
ferent antigenic specificities of this and other complexes, and, in turn, what
can be inferred from such an explanation as to the action of the causative
gene?

Antigens of Pneumococci

By virtue of the ability to attach simple chemical compounds to proteins,
thereby preparing conjugated antigens with specifically reacting components
of known constitution, there has emerged from such studies the realization
that a so-called single antigenic substance may engender a multiplicity of
antibodies of varying specificities (see Landsteiner, 1945, for a critical review
and references). A pertinent example of this sort may be found in the anti-
genic relationship existing between type III and type VIII pneumococci.
Cross reactions between the respective antisera (produced in horses) and the
two types of pneumococci have been observed, implying to them some sort of
antigenic similarity.

As is well known, the specificities of the pneumococcal types depend upon
the carbohydrates of the capsules (Heidelberger and Avery, 1923, 1924).
Thus, the carbohydrate of type III has been found to be a polyaldobionic
acid (Reeves and Goebel, 1941). The understanding of the structure of the
polysaccharide of type VIII is not as complete as for type III, but about 60
per cent of the molecule of the carbohydrate of type VIII appears to be
aldobionic acid. Cross reactivity may therefore be expected between the
soluble specific substances of types III (S III) and VIII (S VIII), by virtue
of the presence in each of multiples of the same aldobionic acid as a structural
unit. It is probable that the serologically reactive unit in each of these two
types is a larger portion of the polysaccharide molecule than a single chemical
structural unit. Type S VIII also contains approximately two glucose mole-
cules for every aldobionic acid residue, thereby presumably accounting for at
least a part of the specificity of type VIII in contrast to type III. Thus it may
be seen that serological cross reactions may be expected when the antigenic
substances under comparison are closely related chemically. Also to be ex-
pected is the ability to distinguish between such substances, as was actually
possible in the case of types III and VIII (Heidelberger, Kabat, and Meyer,

1942).

Genetic Significance

The above example may be combined with other findings in the field of im-
munochemistry to allow the statement that antigenic substances of related
but not identical chemical constitution may — but sometimes do not — incite
the production of cross reacting antibodies. From the serological point of

SPECIFICITY OF GENE EFFECTS 255

view, a pertinent question concerning tiiese antigenic complexes in cattle is
whether the cells which react with the B reagent, or with any other specific
reagent, do so by virtue of the presence of a specific reacting substance in a
single antigenic molecule, or otherwise? Does the comj)lex BOX, for exam])le,
represent (1) three different and separate antigenic substances? Or does it
represent (2) a single antigenic substance with (a) a possible common base
and three more or less different reactive groups accounting for B, G, and K,
respectively, or (6) a single substance capable of inciting many specificities of
antibodies, of which those for B, G, and K represent only a part of the re-
activities of the spectrum of antibodies which may be produced? A combina-
tion of (a) and (b) also may be a possibility.

At present, very little experimental evidence is available concerning the
adequacy of any one or combination of the above possibilities to explain the
antigenic relationships of the components of the antigenic complexes of cattle
cells. Tests are under way to determine whether the reactive substance called
B, for example, is the same in all cells in which it appears, whether singly or
in an antigenic complex.

In terms of the action of the causative genes, apart from the possibilities of
linkage and pseudo-allelism, the question seems to resolve itself around two
main aspects: (1) Do the genes controlling an antigenic complex, as a single
gene for BGK, have separate specificities for B, G, and K, or (2) does this
gene produce a single substance with no such separate specificities, and the
similarities between such a complex as BGK and BGIY, are due primarily if
not entirely to the general similarities in their chemical structure. The writer
is inclined to adopt a combination of these two possibilities as a current work-
ing hypothesis. No matter what may eventually prove to be the correct in-
terpretation of antigenic structure of the complexes, and the action of the
controlling genes, it appears that these studies have given some insight into
the complexities of the gene products and perhaps also of the causative genes.

The studies of the specificities of the gene products — the antigens of the
blood cells of cattle — and the resulting inferences of the structure of the genes
themselves, may not be directly related to the over-all heterosis problem.
Nevertheless the writer is convinced that somewhat comparable specificities
might well be obtained in plants, in which attempts are currently in progress
to measure various aspects of the genetic bases for heterosis. Just how useful
an additional tool of this sort would he is only a guess.

CARL C. LINDEGREN

Southern Illinois Universify

Chapter 16

Genetics and Cytology
of Soccharomyces

In the middle of the last century, Buchner ground up yeast cells and proved
that the cell-free filtrate contained a substance capable of fermenting sugar.
This experiment settled a heated controversy between Liebig and Pasteur
concerning whether or not living structures were essential to fermentation.
The substance responsible for the fermentation was called an enzyme, the
word being derived from the Greek and meaning "in yeast." Since that time,
yeast has been the organism of choice for experimenting in enzyme chemistry
because of the abundant supply obtainable from breweries and from factories
producing bakers' yeast. The biochemistr}^ of fermentation has provided the
foundation for our present understanding of the biochemistry of respiration
and of muscular contraction — two of the fundamental problems which have
intrigued biologists. It has led to an understanding of vitamins and through
them to an understanding of chemotherapy.

BIOCHEMICAL DEFECTS AS GENE MARKERS
The work of Beadle and Tatum has popularized the generally accepted
view that enzymes are derived somehow or other from genes. Their work
initiated a new interest in biochemical genetics. They showed that the in-
activation of a specific gene caused a deficiency which could be met by sup-
plying a specific chemical. Vitamins, amino acids, purines, and pyrimidines
were the substances chosen in this analysis. They used the fungus, Neuro-
spora, because its life cycle had been thoroughly worked out by B. O. Dodge
and because the Lindegrens had shown by genetical analysis that it contained
conventional chromosomes on which genes, arranged in linear order, could
be mapped by the standard procedures used in studying corn and the fruit

fly-

256

GENETICS AND CYTOLOGY OF SACCHAROMYCES 257

YEAST GENETICS

Until 1935, yeasts were considered to be devoid of sex and, therefore, un-
suitable for genetical analysis. At that time, Winge showed that the standard
yeast cell carried two sets of chromosomes — one contributed from each
parent — and was, therefore, a typical hybrid. The hybrid yeast cell produces
four spores, each with a single set of chromosomes. Each of these spores is a
sex cell. By fusing in pairs they can produce the standard (hybrid) yeast cell
and complete the life cycle. In this laboratory it was shown that the spores
are of two mating types, and that each spore can produce a culture each cell
of which can act as a sex cell, like the original spore. Mass matings between
two such spore-cultures result in the production of fusion cells, from which
new hybrids are produced by budding.

This work made it possible to study the inheritance of biochemical de-
ficiencies in the organism on which classical enzyme study is based, and to
attack the problem of the relation of genes to enzymes in this fruitful mate-
rial. We have related specific genes to several of the most thoroughly studied
classical enzymes: sucrase, maltase, alpha methyl glucosidase, melibiase, and
galactase.

The principal advantages of yeasts for biochemical genetics are:

(1) Yeast enzymes have been the subject of intensive biochemical study.

(2) Techniques for studying respiration and fermentation are based prin-
cipally on work with yeast and thus especially adapted to this organism.
Yeasts grow as free cells rather than as mycelial matts and, therefore, can be
subdivided any number of times without injury, thus simplifying weighing
and dilution of the cells.

(3) Large quantities of cells are available from industrial sources or can be
grown cheaply and quickly and are easily stored in living condition.

(4) A variety of genes concerned with the differential utilization of nu-
merous monoses as well as di- and poly-saccharides are available.

(5) A polyploid series of yeast cultures is now available: (a) haploid cells,
each containing a single set of chromosomes, (b) diploid yeast cells, each con-
taining the double number of chromosomes, (c) triploid, and (d) tetraploid
cells (made available by our recent discovery of diploid gametes [Lindegren
and Lindegren, 1951]).

(6) With free cells it is possible to study competition between genotypes
and to observe the advantages or disadvantages in controlled environments.
The populations involved are enormous and the life cycles short, so it is pos-
sible to simulate natural selection in the laboratory. Experiments of this type
have enjoyed an enormous vogue with bacteria, but it has not been possible
to distinguish gene-controlled variation from differentiation. Eor this reason,
experiments with bacteria cannot be interpreted in terms of the comparison
between gene-controlled and other types of inherited characteristics.

258

CARL C. LINDEGREN

CHROMOSOMAL INHERITANCE

In our selected breeding stocks of Saccharomyces, irregular segregations
do not occur very frequently. In maize or Drosophila a similar frequency of
irregularity would not be detectable since tetrad analysis is not possible in
these forms. Using regularly segregating stocks of Saccharomyces we have
mapped four and possibly five chromosomes for genes controlling the fer-
mentation of carbohydrates and the synthesis of various nutrilites (Fig.
16.1).

HI

AN

PN

AD I

IN

PY

TH

24

23

22

10

30

24
AD 2

26

ME

l_

40

45

I

22

PB

22

UR

Fig. 16.1 Chromosome maps of Saccharomyces.

Chromosome I, PN (pantothenate), centromere, ADl (adenine), IN
(inositol), PY (pyridoxine), and TH (thiamin).

Chromosome II, centromere, G (galactose), AD2 (adenine), ME (meli-
biose).

Chromosome III, centromere, a (mating type).

Chromosome IV, centromere, PB (paraminobenzoic acid).

Chromosome V, centromere, UR (uracil).

Chromosomes IV and V may or may not be different; UR and PB have
not been used in the same hybrid.

HI (histidine) and AN (anthranilic acid) are linked to each other (24
morgans) but have not yet been located on a chromosome.

DIRECT TETRAD ANALYSIS

The focal point in the life cycle is the reduction division, at which the
chromosomes of a diploid cell are sorted out, and the haploid sex cells (such as
sperm, eggs, pollen, or yeast spores) are produced. Each diploid parent cell
divides twice to produce a tetrad of four haploid sexual nuclei. This process
is substantially the same whether a single yeast cell produces four spores or a
cell in the testis produces four sperm. In yeast, however, each of the four
spores of a single tetrad can produce clones which are available for individual
study, and the reduction division can be analyzed directly instead of by in-
ference.

GENETICS AND CYTOLOGY OF SACCHAROMYCES 259

Many yeast hybrids have been produced by mating sex cells carrying
chromosomes marked with biochemical mutant genes. The tetrads from these
hybrids have been analyzed by growing clones from each of the four spores
of a single ascus and classifying each of the spore-cultures. These exj)eriments
are direcl tests of the Memieliau theory. They have shown that excei)tions to the
Mendelian theory occur more frequently than was hitherto supposed.

CURRENT STATUS OF IRREGULAR MENDELIAN SEGREGATION

Tetrad analysis of triploid and tetraploid yeasts has revealed that some of
the irregular (not 2 : 2) segregations in hybrid asci arise from the fact that one
or both of the parents is diploid (Lindegren and Lindegren, 1951; Roman,
Hawthorne, and Douglas, 1951). Roman, Hawthorne, and Douglas have
concluded that all irregular segregations in Saccharomyces arise from the
segregation of triploid or tetraploid zygotes. We have recently completed the
analysis of segregation in diploid hybrids heterozygous for both MA /ma and
MG/mg. This analysis revealed that in many asci in which segregation of
MA /ma was 2:2 {MA MA ma ma), segregation of MG/mg was 1:3 {MG mg
mg mg). This finding excludes the possibility that the hybrid was either
triploid or tetraploid since segregation of both genes would have been equally
affected. The phenomenon has been explained as conversion of the MG gene
to mg in the zygote. This conclusion is further supported by evidence indicat-
ing that both genes are in the same linkage group.

One hypothesis of the nature of the gene developed during the study of
irregular segregation seems to have some merit. This is the proposal that the
gene is a complex of many more or less loosely connected molecules rather
than a single macromolecule. In this view, the gene is composed of a series of
identical sites around the periphery of a more or less cylindrical chromosome.
These sites may be extremely numerous since they are of molecular dimen-
sions around the periphery of a thread easily visible under the microscope.
At these sites identical agents responsible for the action of the gene are
located.

GENE DIVISION

The concept of the gene as a bracelet of catalysts arranged on the outside
of the chromosome simplifies the concept of gene reproduction. When one
conceived of genes as macromolecules arranged like beads on a string, it was
difficult to understand how all the genes on a chromosome could divide
simultaneously. If, however, there are thousands of loci and chromogenes at
the site of a single gene on the outside of an otherwise inert chromosome which
is composed principally of skeletal material, any longitudinal splitting of the
chromosome will partition two qualitatively equivalent parts which may or
may not be quantitatively equivalent. The restoration of balance by inter-
dependence of the autonomous organelles may make precise division unneces-
sary.

260

CARL C. LINDEGREN

EXTRACHROMOSOMAL INHERITANCE

When a pure haplophase culture of red yeast (adenine dependent) is
planted on an agar plate, both red and white colonies appear. When the white
colonies are subcultured, only white colonies appear. The red cells when
planted on a second plate continue to produce both red and white colonies.
The white colonies are stable variants derived from red. Bacterial variations
of this type are ordinarily called gene mutations, but bacteriologists have
been unable to test their so-called mutations by breeding experiments except

While Normal
(Adenine Independent)

Red Original Culture
(Adenine Dependent)

White Normal
(Adenine Independent)

Red Subculture
(Adenine Dependent)

White Variant
(Adenine Dependent)

Red Subculture White Variant White Subculture

Normal X (Adenine Dependent) (Adenine Dependent) (Adenine Dependent) X Normal

Hybrid

(White Culture, Adenine Independent)

ASCI Analyzed
White (Adenine Independent)
White (Adenine Independent)
Red (Adenine Dependent)
Red (Adenine Dependent)

Hybrid

(White Culture, Adenine Independent)

ASCI Analyzed
White (Adenine Independent)
White (Adenine Independent)
Red (Adenine Dependent)
Red (Adenine Dependent)

Fig. 16.2 — Inheritance of pink versus white colony in Saccharomyces. The white mutants
derived from pink produce pink offspring and are indistinguishable from the original pink

genetically.

in a few cases. The white cultures have lost their color but they are still
characteristically adenine dependent like their red progenitor. Breeding ex-
periments (Fig. 16.2) have been carried out with the white yeast cultures
derived from the red. When the derived white cultures were used as parents,
they produced precisely the same kind of offspring as the original red culture
from which they arose. This proves that the change from red to white did
not affect a gene. The change from red to white may, therefore, be called a
differentiation since it occurs without gene change.

The phenomenon of Dauermodifikation which was first described by JoUos
(1934) has thus been confirmed in yeast genetics. The stable change from

GENETICS AND CYTOLOGY OF SACCHAROMYCES 261

red to white resembles the discontinuous variations which occur in the vege-
tative cycles of bacteria. Hybridization experiments have revealed that the
origin of white cultures does not involve gene change. This phenomenon in
yeast, called depletion mutation, is identical with Dauermodifikation in Para-
mecia. Since neither involves gene change, both are equivalent to differentia-
tion.

It is not possible to study Dauermodifikationen using the classical objects
of genetical research, maize and Drosophila, since each generation of these
higher organisms is produced sexually — a process during which Dauermodi-
fikationen revert to normal. The stable variants in vegetative cultures of
yeast, which revert to normal (produce only normal offspring by sexual re-
production) have no parallel in maize and Drosophila. This points up a
striking disadvantage of maize and Drosophila — that they cannot be propa-
gated vegetatively. One cannot be certain that the characteristic variations
in flies, which occur when they hatch on wet medium or are subjected to
shock treatment, would be lost on vegetative cultures unless one were able
to propagate the bent wings or other peculiarities asexually, possibly in tis-
sue culture.

THE AUTONOMOUS ORGANELLES OF THE YEAST CELL

In addition to the chromosomes (Lindegren, 1949) there are other perma-
nent structures in the yeast cell which never originate de novo (Lindegren,
1951). They have the same type of continuity in time as chromosomes but
are less precisely partitioned than the latter.

The Cytoplasm

The cytoplasm is a limpid fluid which is transmitted to each daughter cell.
It is rich in RNA but varies in basophily and contains the mitochondria,
usually adhering to the surface of the centrosome or the nuclear vacuole.

The Mitochondria
The state of the mitochondria varies from highly refractile lipoidal struc-
tures, sharply defined from the cytoplasm to less refractile organelles with
somewhat irregular boundaries.

The Centrosome

The centrosome is a solid and rigid structure which stains with acid
fuchsin but does not stain with basic dyes. This highly basic organelle may
contain some of the basic proteins which Caspersson and Mirsky have found
in chromosomes. The centrosome is always attached to the nuclear vacuole
and is the most rigid structure in the cell as revealed by its behavior following
shrinkage of the cell. It never originates de novo and plays a leading part in
budding, copulation, and meiosis.

262 CARL C. UNDEGREN

The Centrochromatin

The centrochromatin is a basophilic, Feulgen-positive substance closely
attached to the basic centrosome (probably by an acid-base reaction). Some
portion of it is usually in contact with the nuclear vacuole. It is partitioned
between the cells following budding by a direct division controlled by two
tiny centrioles. In the resting cell it may assume a spherical form and cover
most of the centrosome. In division it is usually present in the form of a long
strand. The centrosome and centrochromatin have been identified with the
nucleus by several workers, but this view has been criticized by Lindegren
(1949), Lindegren and Rafalko (1950), and Rafalko and Lindegren (1951).
The filament often bends on itself to assume a V- or U-shape. In some
preparations it appears to be composed of numerous small particles, but this
is due to poor fixation and is especially prevalent in preparations fixed with
alkali. The view that the centrochromatin is a single filament external to
the centrosome is supported by a multitude of observations on well-fixed
cells. Centrochromatin is probably homologous to the heterochromatin of
higher forms differing only in being carried on the centrosome rather than
the chromosome.

The Nuclear Membrane and the Chromosomes

The nuclear vacuole contains the chromosomes and the nucleolus. The
chromosomes are partitioned between mother and bud vacuole in a precise
orderly manner without recourse to a spindle. The wall of the nuclear vacuole
does not break down at any time in the life cycle; it is a permanent cellular
structure.

The Cell Membrane and the Cell Wall

The cell membrane is a permanent cell structure. The cell wall appears to
be formed de novo in the spores, but it may depend on the cell wall sur-
rounding the ascus for its origin.

BUDDING

Figure 16.3-1 shows a cell in which the acidophilic centrosome attached
to the nuclear vacuole is surrounded by the darkly staining cytoplasm. A
band of basophilic centrochromatin is securely applied to the side of the
centrosome and is also in contact with the nuclear vacuole. Greater differen-
tiation often reveals a small centriole at each end of this band. The nuclear
contents are unstained.

Figure 16.3-2 shows the first step in the process of budding. The centro-
some produces a small conical process which forces its way through the
cytoplasm and erupts into the new bud shown in Figure 16.3-3.

Figure 16.3-4. The nuclear vacuole sends out a long, slender process which
follows the centrosome into the bud. Although the cell wall is not visible in
these preparations it must be assumed that the cell wall never ruptures but is

.Mk

i

7

8

10

12 .

Fig. 16.3 — Behavior of the centrosome and centrochromalin during budding.

264 CARL C. LINDEGREN

extended to enclose the bud at all times. The vacuolar process follows the
external surface of the centrosome into the bud, lying between the cell wall
and the centrosome.

Figure 16.3-5 shows a cell in which the bud vacuole has received its two-
stranded chromosome complex. This is an exception to the rule that the
chromosomes usually are completely destained in the differentiation by
iron alum.

In Figure 16.3-6 the bud vacuole is lobed. This is a rather common phe-
nomenon. The cytoplasm has passed into the bud and completely surrounds
the centrosome and the bud vacuole. The extension of the centrochromatin
along the surface of the acidophilic centrosome has begun.

Figures 16.3-7, 16.3-8, and 16.3-9 show cells in which the separation of
the centrochromatin has been completed with mother and bud held to-
gether by the centrosome.

In Figure 16.3-10 the division of the centrosome is complete, but both
centrosomes are near the point of budding. In Figure 16.3-11 the bud cen-
trosome has reached the distal end of the bud while the mother cell centro-
some still lies in the neighborhood of the point of budding. In Figure 16.3-12
both centrosomes have reached the distal ends of the cells and are prepared
for the formation of the next bud.

CONCLUSIONS CONCERNING EXTRACHROMOSOAAAL INHERITANCE

Cytological examination of the yeast cell shows that many of its organelles
may have the same integrity and continuity in time that characterize the
chromosomes — they cannot arise de novo. In the yeast cell there are seven
or eight such "continuous" organelles. The cell membrane, the nuclear
membrane, the centrosome, the centrochromatin, the cytoplasm, the mito-
chondria, and the chromosomes are permanent cell structures. Because they
apparently divide in a manner which does not provide for precise trans-
mission of specific portions to each daughter cell, it appears that the other
components differ from the chromosomes in a significant manner — they are
probably homogeneous, or their heterogeneity is simple, possibly a few
different types of dipolar molecules held together in a specific manner.

There is no reason to assume that any one of these components is of more
importance, or directs the "activities" of any one of the other components.
The cell can function only if all its component parts are present in proper
structural correlation and in adequate amounts. There is no reason to as-
sume that any one of these components is unique in the manner in which it
reproduces itself. The present hypothesis proposes that they all reproduce
by the simple accretion of molecules like those which they contain, and it is
their association with each other in an adequate milieu which provides the
molecules necessary for their increase in size. Each of the different organelles
is rate limiting in growth. When any one is present in less than the minimal

GENETICS AND CYTOLOGY OF SACCHAROMYCES 265

amount, the other organelles cannot obtain the supply of molecules necessary
for maintenance and increase until the amount of the deficient organelle has
increased.

The chromosomes differ from the other permanent organelles in their high
degree of linear heterogeneity. It is this characteristic which has given them
the spurious appearance of "controlling" other cellular activities. Mutations
with which we are familiar in the laboratory constitute defects or deletions
in the extraordinarily heterogeneous chromosomes. The deficiency in the
organism caused by the defect — the deletion of the contribution ordinarily
made by the intact region of the chromosome — becomes apparent only be-
cause the rest of the chromosome produces sufficient materials to enable the
defective cell to continue to grow in its absence, although in a manner differ-
ent from that which was previously characteristic.

Any transmissible defect in a homogeneous structure like the cell wall,
the cell membrane, the nuclear membrane, the centrosome, or the centro-
chromatin would result in total failure of the organism to survive and bring
all vital activity to a halt. The survival of the defective mutants in their
altered condition due to the defect in the chromosome (which has been called
a mutant gene) has led to the view that genes are different from other cellu-
lar components since they can reproduce variations in themselves. This is an
incorrect point of view. It is more proper to say that when a defect or dele-
tion occurs in a small segment of a chromosome, the rest of the organism can
carry on, albeit in a changed condition due to the absence of the contribution
previously made by that region, now called the gene. This denies the im-
portance of the ordinary mutations encountered in the laboratory as factors
for progressive evolution, and implies that progress in evolution must occur
in some other way.

It may be that progressive evolution occurs more frequently as the result
of changes in the chromosomes than of other organelles. But the present hy-
pothesis does not exclude the possibility that advances in evolution can
occur by ''progressive" changes in the composition of any one of the eternal
organelles such as the nuclear membrane or the centrosome. The condition
for the perpetuation of any change would be that the mutated organelle
could be provided with the materials necessary for its continuance by the
cell as a whole in its surrounding environment at the time of its occurrence.
On this hypothesis, progressive changes in evolution are not confined to any
single cellular component, but constitute a potential of every component of
the cell. Although progressive changes of the different substances compris-
ing the chromosome may not occur significantly more frequently than
changes in the substances making up the other organelles, more changes may
occur in the chromosomes in toto because a change in each individual com-
ponent of the extraordinarily heterogeneous chromosome registers as a sepa-
rate change.

266 CARL C. LINDEGREN

In many types of organisms the chromosomes are always separated by the
nuclear membrane from the cytoplasm. The mitochondria (like the chromo-
somes) are relatively non-homogeneous, but apparently the balance of their
activities is not so critical since no specific devices appear to be required to
limit their reproduction or activity. The cytoplasm is probably heterogene-
ous also, with every separate eternal component having the same continuity
in time as the chromosomes. However, it comprises substances transmitted to
the daughter cells in a manner which is apparently subject to control by the
environment, and this may constitute the basis for differentiation. In the
germ line, the entire cytoplasmic potential must be maintained. In fact, the
main function of the germ line under this hypothesis would be to maintain
an intact cytoplasm. The integrity of the chromosomes is usually provided
for in either the somatic or the germinal tract. Defects in the extra chromo-
somal apparati are reconstituted in an outcross, thus differentiating so-called
cytoplasmic from genie inheritance.

H. H. PLOUGH

Amhersf College

Chapter 17

Genetic Implications of
Mutations in S. Typhimurium*

The contribution that an account of studies in bacterial genetics can make
to the problem of heterosis must be indirect, since actual sexual or other
fusion in bacteria has not been observed and the weight of evidence is against
the view. Even the very interesting genetic evidence of recombination dis-
covered by Tatum and Lederberg (1947) in the K12 strain of the colon
bacillus, and now being developed by the capable studies of Lederberg (1947,
1949) and others, is still susceptible of other interpretations. Diploid strains,
if they occur at all, are certainly so rare as to be unimportant in the produc-
tion of hybrid vigor in bacterial populations.

The apphcations of bacterial genetics to the problem of heterosis must be
rather in the information they make available concerning the kinds and fre-
quencies of gene mutations, and the ways in which they interact with each
other within populations. It has been generally recognized by geneticists
only recently that the bacteria are excellent material for studies of these
problems, though bacterial mutation was first mentioned by Massini in 1907,
and distinctive and precise food requirements for bacterial strains have been
known since 1913 (Hinselwood, 1946). Studies in the genetics of bacteria
have, of course, been greatly stimulated by the pioneer work on mutations
in fungi by Thom and Steinberg (1939), and particularly on Neurospora by
Dodge, by Lindegren, and by Beadle (1949) and his associates, as well as by
the important work on yeast as presented in Dr. Lindegren's chapter.
Long before the currently enlarging wave of interest in bacteria as objects
of genetic study, Gowen had shown that mutations of the same order of
frequency as in higher plants or animals were induced by radiation in Phy to-

* This research was supported bva grant from the Atomic Energy Commission, Division
of Biology and Medicine #AT (30-1) -930.

267

268 H. H. PLOUGH

monas (1945). He and Zelle had indicated the genetic basis of virulence in
Salmonella (Zelle, 1942).

ADVANTAGES OF SALMONELLA FOR GENETIC STUDIES

I became acquainted with the Enterobacteriaceae and particularly with
the pathogenic forms in Zinsser's laboratory at the Columbia Medical School.
My own realization that Salmonella offered excellent material for studies in
microbial genetics was heightened when, as an Army bacteriologist in the
Philippines, I had to diagnose enteric infections. I found most of the Salmo-
nellas which Flexner first described from Manila still present in the islands.
More than 140 strains or species of Salmonella are recognized which are dis-
tinguishable by a common pattern of fermentation reactions (dextrose and
maltose- AG, lactose and sucrose-negative, citrate and H2S positive). Each
one has been shown by the serological studies of White (1929), Kaufmann
(1944), or Edwards and Bruner (1942) to have a very precise and readily
separable antigenic constitution.

The antigens are determined by agglutination studies using serums from
different rabbits immunized to one or another of the major strains. They
fall into two distinct groups: the somatic (0) antigens associated with the
surface protein layers, and the flagellar (H) antigens determined by proteins
of the flagella. Each of these groups is known to be compound, with
some twenty separate O antigens — each strain may carry three or four
(O) antigens — and eight or ten different specific (H) antigens as well as cer-
tain alternative and non-specific phases of the latter. Thus each strain can
be shown to have a distinctive and readily determinable antigenic constitu-
tion (S. typhimurium is I, IV, V, XII — i, 1, 2, 3). The whole group naturally
falls into a tree-like pattern very like the evolutionary trees made for fami-
lies of animals or plants on the basis of structure.

Tatum's (1946) discovery that mutagenic agents (including radiation and
nitrogen mustards) could induce mutants of colon bacteria having constant
growth factor requirements more limited than the parental organism, just
as with Neurospora, has re-emphasized the one gene-one enzyme hypothesis.
It has strengthened the idea of bacterial evolution developed by Lwoff (1943)
that the parasitic forms have been derived from the less exacting hetero-
trophic organisms by successive losses of synthetic abilities. Thus it gives
added meaning to the tree-like interrelationships suggested by the antigenic
analyses.

Soon after the war our Amherst group entered on an intensive study of
induced biochemical and antigenic mutations in the food poisoning organ-
ism, Salmonella typhimurium. It was our hope that this organism would
prove more favorable for genetic studies than E. coli, not only for the analysis
of the mode of action of genes, but for evidence on the genetic nature of type
specificity, virulence, and their bearing on evolutionary relationships.

GENETIC IMPLICATIONS OF MUTATIONS IN S. TYPHIMURIUM 269

METHODS OF INDUCING AUXOTROPHIC MUTATIONS

The strains of Salmonella typhimuriimi which we have used are two: 519
received from the New York Salmonella center at Beth Israel Hospital, and
533 (lie) from Gowen.

Our method for isolating mutations to specific food or growth factor re-
quirements by penicillin screening is that of Lederberg and Zinder (1948)
and of Davis (1949) with some additions of our own. S. typhimurium is a
heterotrophic organism of the least exacting sort. Cultures will grow on a
basic medium containing ammonium sulphate, sodium chloride, potassium
phosphate buffers, with traces of other metallic ions, and glucose added as
an energy source. Better growth is obtained with a supplementary nitrogen
source, such as asparagin, and a further energy source, citrate, but these are
not essential. Thus the organism synthesizes all its own food comj)onents,
coenzymes, and growth factors, as well as the enzymes necessary for food
and energy tranformations.

Suspensions are subjected to radiation by X-rays (up to 100,000 roent-
gens) or ultraviolet light (up to 3,600 ergs per mm.-), and are then transferred
to an enriched nutrient broth for 24 hours. The broth stimulates active
division of all organisms. These are centrifuged off, washed, and reinoculated
for 24 hours into the basic or minimal medium containing 100 units per ml.
of penicillin. This stops the divisions, and progressively kills the organisms
which divide actively.

These organisms which penicillin screens out are called prototrophic
(Lederberg), and they are, of course, the unchanged originals. Any mutated
organisms w^hich now require some specific nutrilite will not divide on the
basic medium, and so they are not affected by penicillin. These are now
auxotrophic organisms (Davis), and they are isolated by plating on complete
agar, and identified by paper disc inoculations on successive plates of basic
medium with single nutrilites added — amino acids, nucleic acid fractions, or
vitamins, as shown in Figures 17.1 and 17.2. These methods are described
in more detail by Plough, Young, and Grimm (1950).

AUXOTROPHIC MUTATIONS FROM RADIATED LINES
I shall cite only one set of isolations from such a radiation experiment, the
data for which are given in Table 17.1. Suspensions from an unradiated con-
trol and from seven successively increased X-radiation dosages were run
through the penicillin screening, and 500 auxotrophic mutants isolated. Of
these a total of 459 were recovered and their specific requirements deter-
mined. Although the control had been derived from successive single colony
isolations within 3 days of the tests, still 5 per cent of the isolated strains were
mutants — indicating that spontaneous mutation occurs and accumulates in
stock strains.

From the major strain used (#533), 234 strains out of the 459 isolated

5 ML

5 ML

CENTRIFUGE

24-HOUR 17 ML

AGAR SLANTS NACL

CELLS
IN NACL

0 1 ML

I ML

COMPLETE
BROTH

ISOLATIONS

If

3 ML

10* 100 UNITS

DILUTION PENICILLIN
IN MIN. MED.

COMPLETE
AGAR

MINIMAL
BROTH

IN 3 ML
OF NACL

COMPLETE
BROTH

Fig. 17.1 — -Diagram showing methods for the production of radiation-induced auxotrophic
mutations in Salmonella and for their isolation by screening through minimal medium

containing penicillin.

PAPER DISC METHOD FOR TESTING BIOCHEMICAL MUTANTS

TEST STRAIN

NEEDLE TRANSFER

24-HR BROTH DISTILLED Hg 0

PAPER DISKS

SALTS +GLUCOSE +
AGAR + NUTRILITE

Fig. 17.2 — Diagram showing method for determining the particular nutrilite required by
the auxotrophs isolated as in Figure 17.1. .\ series of Petri plates is used, each containing

a different test substance.

GENETIC IMPLICATIONS OF MUTATIONS IN S. TYPHIMURIUM

271

were auxotrophic mutants, among which 17 different auxotroj^hic mutants
occur according to tests of the specific nutrilite required. A summary of these
requirements with the numbers of each is given in Table 17.2. Tlae most
frequent auxotroph is the one requiring cysteine. The next most frequent is
the histidine auxotroph, and so on down the list to one which has a double
requirement of both valine and isoleucine for growth. Only two auxotrophic
mutants require substances other than amino acids. One must be supplied

TABLE 17.1

FREQUENCIES OF AUXOTROPHIC MUTATIONS IN S. TYPHIMURIUM
AFTER X-RADIATION AND PENICILLIN SCREENING

1

X-Ray Dosage
and Time

2

%
Bacteria
Surviving

3

Total

No.

Tests

4

Total

No.

Mutants

5

%
Mutants

6

No.
Different
Mutants

7

%
Different

Mutants

Strain 533

I Controls

II 11,400 R4 min

III 17,100 R6 min

IV 22,800 R 8 min

V 28,500 RIO min

VI 34,300 R 12 min

VII 45,600 R 16 min

VIII 57,000 R 20 min

100

40

25

14
6

1.5
0.9

135
62
86
41
25
94
99
50

7

16
18
11
19
64
72
34

5.1
25.9
20.9
26.8
76.0
68.1
72.6
68.0

3
4
4
3
6
8
17
12

2.2

6.4

4.7

7.3

24.0

8.4

17.2

24.0

Totals

459

234

IX II + III + IV

189
268

45
189

23.8
70.5

11
43

5.8

X V+ VI 4- VII 4- VIII. . .

16.0

Strain 519
XI 45,600 R 16 min

25

100

22

22.0

9

9.0

with adenine, and others (not found in this experiment) must have either
guanine or thiamin in the medium.

In our published report of these data (Plough, Young, and Grimm, 1950,
Table 3) we listed a number of additional strains showing alternative re-
quirements. Davis (personal communication) retested a number of these
and found them to be mixtures of single autotrophs. We have just completed
an extensive recheck of all strains listed originally as alternates, and now
confirm his results except for the three types of alternates listed in Table
17.2 (Plough, Miller, and Berry, 1951).

MUTATION FREQUENCY AND X-RAY DOSAGE

One of the most interesting results of this experiment is the clear relation
between the frequency of auxotrophic mutants and the X-ray dosage. This
is shown in Table 17.1, column 5, lines II- VIII, and I could add to the data
from other experiments. The numbers of tests vary for the different radiation

272

H. H. PLOUGH

dosages, and some of the values are less significant statistically, but the per-
centage of mutants is significantly higher at the higher dosages. This is em-
phasized by lines IX and X where the sums of the first three and the last
four values are compared. The same conclusion is evident from inspection
of column 7 in the table, where the numbers of different mutants at the
successive dosages are shown. Nearly three times as many were isolated from
the upper group as from the lower.

TABLE 17.2
KINDS OF AUXOTROPHIC MUTATIONS IN S. TYPHIMURIUM

No.

Strain 533
Single Amino Acids

No.

Strain 519
Single Amino Acids

105

55

15

14

5

5

4

2

1

Cysteine

Histidine

Leucine

Proline

Tyrosine

Threonine

Methionine

Vahne

Arginine

8

3

3

3

1

1

1

Histidine

Cysteine

Methionine

Proline

Leucine

Tryptophane

Phenylalanine

5

Nucleic Acid Fraction
Adenine

1

4

Multiple Amino Acids
Valine and isoleucine
Unanalyzed

21

1

Alternative A niino Acids
Cysteine or methionine
Tyrosine or tryptophane

1

Alternative A mino Acids
Cysteine or Methionine

1

Tyrosine or phenylalanine

Line XI in the table shows the result of one radiated series made on a
different initial strain, #519. Comparison of the column 5 and column 7
totals with line VII above, shows that this strain is much more resistant to
radiation than is strain #533. It is clear that comparisons of the mutagenic
effects of radiation dosage must always be made between samples from the
same strain.

The data in Table 17.1, column 5, are graphed in Figure 17.3. Compari-
son of the percentages of mutants at successive dosages shows a positive
correlation, though rather far from a straight line curve. As the penicillin
screening method involves a 24 hour growth in complete broth, and another
24 hours in minimal medium with penicillin, it might be expected that the
final percentage of mutant strains would not bear the direct relation to dos-
age shown in tests of mutations produced in germ cells in sexually reproduc-
ing organisms. Indeed Davis, in his account of the penicillin screening method

GENETIC IMPLICATIONS OF MUTATIONS IN S. TYPHIMURIUM

273

as used in E. coli, stated "... the method as developed so far does not appear
to yield quantitative survival of mutants." Such a statement assumes that
the penicillin screening may be expected to be complete, which in fact is not
true. Rather penicillin acts, as do all antibiotics, in a progressive fashion ac-
cording to a typical logarithmic killing curve. If two or more mutant cells

PERCENT
MUTATIONS
100

80

60

40

20

DATA OF TABLE 1
A = I - VllI
O - IX- X

o

4 6 8 10 12 14 16
MINUTES OF RADIATION

20

Fig. 17.3 — Graph showing the relation between percentage of mutations isolated and
X-ray dosage in minutes (2850 R per minute).

appear in a growing wild type population, they will increase logarithmically
and form smaller less numerous clones. As the penicillin acts, the far more
numerous parent clones will be logarithmically reduced in numbers, while
the mutant clones exposed will have reached a level which may be main-
tained during the 24 hour period of penicillin action. It is clear that if a
sample is taken, and plated at any point short of the complete killing oflf of
the wild type, we may expect frequencies showing the same order as in the
original population, although the mutant percentages are greatly magnified.

274

H. H. PLOUGH

An actual test of artificially made mixtures of the parent strain and one
cysteine requiring mutant as screened by the media is shown in Table 17.3.
The data show that a mixture of 90 per cent wild and 10 per cent mutant
still gives a greater number of wild survivors after penicillin screening than
does a mixture having 10 per cent wild and 90 per cent mutant. For the
actual experiments reported in Table 17.1 the proportion of mutants to un-
mutated wild type even after 24 hours of growth in complete broth is one
in many thousands, rather than 10 per cent to 90 per cent. So it seems justi-
fied to consider the percentage of mutants and wild type as an index of muta-

TABLE 17.3

EFFECT OF GROWTH IN COMPLETE MEDIUM FOLLOWED
BY PENICILLIN SCREENING ON ARTIFICIAL MIX-
TURES OF CONTROL (533) AND A CYSTEINE AUXO-
TROPH (533-169)

Mixture

Percentages
Original Mixture

Percentages

After 24 Hrs.

IN Broth

Percentages

After Subsequent

Screening

533

533-169

533

533-169

533

533-169

A

90

10

70

30

8

92

B

50

50

33

67

2

98

C

10

90

5

95

2

98

tion frequency in comparing X-ray dosages. The trend in Figure 17.3 sug-
gests a sigmoid curve rather than a straight line as Hollaender (1948) has
shown for ultraviolet induced visible mutations in fungi. Essentially the
same interpretation can be drawn from a comparison of the number of differ-
ent mutations found at the successive X-ray dosages. Much more extensive
data are now available showing the relation between mutation frequency
and both X-radiation and ultraviolet dosages and they will appear in another
publication. In general they all bear out the conclusion that the frequency
of auxotrophic mutations is directly correlated with radiation dosage as is
true for gene mutation in other organisms.

A rather interesting result of comparison of these percentages of mutants
present after penicillin screening is that the most frequent class changes from
the lower to the higher dosages. Thus after 11,000 roentgens, a cysteine auxo-
troph is the most frequent, while after 57,000 r it is a histidine requirer.
Perhaps we are dealing with a specific effect of dosage or conceivably with
a differential effect of wave length, but until the complex nature of the cys-
teine mutants are more fully understood it is unwise to attempt too definite
an interpretation.

RECOMBINATION TESTS IN SALMONELLA

Much interest has been excited among geneticists as well as bacteriologists
by Lederberg's proof that mixtures of multiple mutant stocks of the K12

GENETIC IMPLICATIONS OF MUTATIONS IN S. TYPHIMURIUM 275

strain of E. coli give rise to new strains having the auxotrophic mutants in
new combinations. These initial observations have been repeated in differ-
ent combinations and amply confirmed by the observations of many other
investigators. As Lederberg has suggested, these results are most reasonably
interpreted as due to bacterial union like a sexual fusion of gametes, fol-
lowed by an immediate reduction process involving segregation and genetic
recombination, suggesting linkage in a single chromosome system. More
recently Lederberg (1949) has found evidence of what appears to be a diploid
strain which gives highly aberrant segregation ratios. These require assump-
tions of such an extremely complex and involved type of chromosome
interchange that it becomes questionable whether some other explanation is
not after all more probable.

In S. typ/iinmrium we now have more mutant strains carrying single
auxotrophic genes or multiple combinations of these than in any other
bacterial species except E. coli. This makes it especially important to test the
theory with our strains. Accordingly Miss Marie McCarthy has been mixing
these in varying combinations, and then plating out in heavy suspensions on
base medium supplemented so as to show up the transfer of one or more re-
quirements from one to the other original combination.

Although more than a hundred such tests have been made and carefully
checked, the results have been unequivocally negative until very recently.
This work will be reported in detail in a later publication, but I will describe
it briefly here. Multiple strain #519-38-94-41 requiring tryptophane, me-
thionine, and histidine was mixed with #533-486-96-85 requiring leucine,
threonine, and arginine. On plating in appropriate media it was found that
in addition to the original parental combinations several colonies each gave
strains requiring two new sets of requirements. Recombination No. 1 re-
quired tryptophane, leucine, and threonine. Recombination No. 2 needed all
six amino acids: tryptophane, methionine, histidine, leucine, threonine, and
arginine. These new stocks have been retested, and there can be no question
of the fact that we have here two recombinations of the original stocks used.
Other recombinations have now appeared but reciprocal classes are never
found. Thus we have in Salmonella confirmation of the recombination results
found by Lederberg in the K12 strain of E. coli. In view of the irregularity of
such results both in E. coli and in Salmonella, it would seem wise to suggest
that some alternative explanation may yet prove to be more satisfactory than
recombination or chromosomal crossing-over.

BIOCHEMICAL STUDIES OF AUXOTROPHIC MUTANTS

The Neurospora studies of Eeadle and his associates as well as those of
Lindegren (1949) on yeast have made it evident that in studying the action
of auxotrophic mutants we are many steps closer to the initial determinative
activities of the genes themselves than is ordinarily true for characters in the
higher plants and animals. When a series of auxotrophic genes can be shown

276

H. H. PLOUGH

to block successive steps in the syntheses of particular amino acids or vita-
mins or more complex products, the one gene-one enzyme hypothesis offers
the most satisfactory preliminary explanation, even though the presence of
the particular enzyme as a gene product has not been demonstrated. Each
set of auxotrophic mutants offers data on the chain of synthetic processes to
some essential substance, and thus becomes a challenging biochemical prob-
lem. It is significant that many of those already studied in the fungi have
also been uncovered in E. coli, but every organism shows individual differ-
ences. So far in Salmonella we have investigated the biochemical steps in
only two such series of auxotrophs, but many others await study especially
as new mutants are added.

TABLE 17.4

UTILIZATION OF SULPHUR COMPOUNDS BY VARIOUS
AUXOTROPHS OF 5. TYPHIMURIUM

Strain

NaiSOi

NajSiOs

NajS

Cysteine

Cysta-
thionine

Methio-
nine

Block
in Fig. 17.4

1. Original 533

2. 533-575

+

+
+

+
+
+

I 1 -f + + +

+
+
+
+
+

+

+

+
+

None
7-1-2

3. 533-526

7

4. 533-452

5-h2

5. 533-P249

6. 533-535

4
2

The first of these sets of interacting synthetic steps which we have studied
is the cysteine-methionine auxotroph series. These mutants fall into many
of the same gradations described by Lampen, Roepke, and Jones (1947) for
E. coli, by Emerson (1950) for Neurospora, and by Teas (1950) for B.
subtilis. We have tested all of the apparent cysteine or methionine requirers
for their ability to reduce inorganic sulfur compounds as well as to utilize
organic precursors of methionine. The wild type strains can reduce sulphate,
sulphite, or sulfide, and can grow with no other source of S. It has been
shown, however, that none of the apparent cysteine requirers can reduce
sulphate, but some can reduce sulphite and some sulfide. Many, however,
must have cysteine or cystathionine (kindly supplied by Dr. Cowie) and
others require methionine as such.

A summary of representative mutants isolated as cysteine or methionine
requirers and their abilities to grow on various compounds as the sole source
of S is given in Table 17.4. This can be visualized as in Figure 17.4 in terms
of a succession of steps, each catalyzed by an enzyme controlled by a gene
which is inactivated by the mutation numbered in parentheses. Such a
straight line series appears to run in the direction of the arrows from sulphate
to protein. When a mutation occurs, as at (5), it must be assumed that growth
requirements will be satisfied by any compound succeeding the break in the

GENETIC IMPLICATIONS OF MUTATIONS IN S. TYPHIMURIUM 277

synthetic chain, unless a second mutation has occurred. This does not hold
for methionine which cannot be utilized in mutants #2 and 4 (Table 17.4).
Such a result suggests that cysteine is enzyme controlled through a gene
which is inactivated by the mutation numbered in parentheses. Cysteine is
ordinarily made from methionine (as has been shown for the mammal) and so
the reverse dotted arrow marked (1) is shown in the figure. It is hardly likely
that a second mutation is indicated for the mutants cited as showing two
blocks, but rather that certain mutations cause inhibition of more than one
enzyme system. A more comprehensive scheme for the cysteine-methionine
synthesis based on the Neurospora work has been given by Emerson (1950).
It is certain that more is involved in the series of reactions shown in Figure
17.4 than the furnishing of essential sulphur for cysteine and methionine.

^ Protein

(7) (6) (5)
SO4 »> SO3 *► S Cysteine

HOMOSERINE
(1)/ \ (4)

y Cystathionine

? ^ d) \

Protein -< Methionine < Homocysteine

Serine

Fig. 17.4 — Possible chain of reactions involving sulphur-containing compounds. (Mutant
blocks indicated by numbers in parenthesis.)

Sulphate, sulphite, and suliide, as well as cysteine itself, may act as H
acceptors, cooperating with dehydrogenases involved in the respiratory or
energy producing activities of the organism. That the organism reduces more
sulphate than is necessary for the S required in the amino acids is indicated
by the fact that Salmonella forms a readily testable excess of H2S. We are
attempting to trace the course of the sulphur by the use of the radioactive
isotope S^''. Last summer Dr. T. P. Ting and the writer were able to show that
(NH4)2S*04 is taken up by the wild type 533 organisms and not at all by a
cysteine requiring mutant, thus confirming our growth tests (see also Cowie,
Bolton, and Sands, 1950).

We hope to continue this work using labeled sulphur in sodium sulfide or
barium sulfide, which should be utilized by wild and mutants number (7),
(6), (5) (Fig. 17.4). Finally it should be possible to determine by quanti-
tative tests how much S^* is combined into bacterial protein and how much
passed out in HoS. Comparisons between different strains in oxygen utiliza-
tion are being made with the W'arburg respirometer. As already shown in

278

H. H. PLOUGH

Table 17.3, some cysteine requiring strains will overgrow the parent and this
may be due to differences in energy requirements.

The second set of steps in synthesis being studied concerns the adenine
requirer. Here we appear to have rather more definite information than was
described by Guthrie (1949) for the purine auxotrophs of E. coli. It has been
shown that the Salmonella auxotroph utilizes adenine and hypoxanthine, but
not guanine and xanthine. Of the nucleosides and nucleotides only adenosine
and adenylic acid are used, and much more of the latter is required for com-
parable growth than of adenine. Thus it appears that in purine metabolism,
Salmonella and an animal like Tetrahymena (Kidder and Dewey, 1948)
show almost opposite requirements, for the bacteria do not convert adenine
to guanine. Preliminary studies by Mrs. Helen Y. Miller demonstrate a spar-
ing action for adenine utilization by the amino acid histidine. This suggests

N==C-NH2

HC— NH

HC— NH

H-C

C-NH

N

— C-N

Adenine

^

CH

C— N
CH2

o=c

^

CH

-^-

C — N
CH2
H2N-CH

/-

:cH

COOH

Imidazole Pyruvic Acid

+
Pyridoxamine

COOH
Histidine

Fig. 17.5 — A Possible relation of Adenine to Histidine synthesis (after Broquist and Snell).

that for this organism as with Lactobacillus (Broquist and Snell, 1949) the
purine is a precursor of histidine, probably by the utilization of the imid-
azole ring through pyruvic acid, and the transaminating action of pyridoxa-
mine (Figure 17.5). While these facts have been revealed by a study of the
adenine mutant alone, further gene changes and their reactions with the
histidine auxotrophs already available should help clarify some of the inter-
actions of purines and amino acids in the bacterial cell.

ALTERATION OF ANTIGENIC SPECIFICITY

The auxotrophic mutations reveal a series of biochemical steps or trans-
formations common to whole groups of organisms. Antigenic analysis, on the
other hand, has revealed precise specific or strain differences which are as
distinctive as the form or structural differences of complex animals and
plants. This has been clearly demonstrated by the blood group analysis
presented in the studies of Irwin and his colleagues. The specificity is no less

GENETIC IMPLICATIONS OF MUTATIONS IN S. TYPHIMURIUM 279

sharp in the antigenic analyses of Salmonella. The tree-like relationship
which they suggest was our chief stimulus to a study of bacterial genetics in
this organism.

Preliminary tests of all of the auxotrophic mutants made by Miss Dorothy
Farley show that they are unchanged antigenically. Not only the specific
antigens, but the agglutination titers are the same as the original strains.
This has been confirmed by reciprocal absorption tests, as well as by precipi-
tation, and inhibition of agglutination using supernatants from boiled cul-
tures. Thus it appears that the loss of ability to synthesize a particular amino
acid in no way alters the antigenic configuration. Apparently if proteins are
formed at all they take on the antigenic configuration of the cytoplasm al-
ready there. The auxotrophic mutants and the antigenic patterns fall into
two quite independent systems so far as present evidence goes. This seems
to be true also for variations in or loss of virulence. The relation of the auxo-
trophic mutants to virulence for mice is being studied in detail by Gowen
and his associates and will be reported separately, but so far at least it ap-
pears that there is no relation between virulence and the biochemical re-
quirements of the strain.

It was originally and is still our hope to be able to induce antigenic
variants by radiation, but so far such attempts have given negative results.
We have inoculated radiated suspensions into one end of U tubes of semi-
solid agar containing low concentrations of O serum from a rabbit im-
munized against the specific strain, and the organisms grow through the
medium. When agar containing specific H serum is used, however, the
organisms grow only at the site of inoculation. If antigenic mutants had
occurred we would expect that the homologous serum would act as a screen
to block off the original and let the mutants through, just as the penicillin
does for the auxotrophs. The result simply means that we have not found any
antigenic mutants following radiation. Perhaps we should not expect any.
Antigenic mutants have been induced in several bacteria by other meth-
ods, especially by McCarty (1946) in the pneumococcus, by Bruner and
Edwards (1947) in Salmonella, and by Boivin (1947) in E. coU. The pneu-
mococcus method is not applicable to Salmonella, and the Boivin method in-
volving exposure of the organism to autolysates of rough variants of other
strains gives negative results. Tests using similar culture filtrates have
been unsuccessful in altering the antigenic constitution of our organism. On
the other hand, Miss Farley has made use of the Edwards technique of grow-
ing an auxotrophic mutant in a semi-solid medium containing homologous O
serum previously absorbed with a related organism which lacked one of the
major antigens, XII (and in another case lacked V but carried an additional
antigen XXVII). By this method two successful transformations of type
have been secured out of several tried. Both of these transformations were
performed on an auxotrophic mutant (519-PlO) requiring histidine.

280 H. H. PLOUGH

Preliminary tests showed that these strains were antigenically similar and
gave the same agglutination titer with homologous serum as the parental
wild types — (I) IV, V, XII, for the O antigens. The parentheses indicate that
(I) is very weak or absent. The first case is typical. Specific serum from ani-
mals immunized by #519 was absorbed with a suspension of organisms of
#527, an unnamed strain known to have 0 antigens IV, V only. After it was
passed through semisolid agar containing the absorbed serum now carrying
XII antibodies only, 519-PlO was retested and shown now to give agglutina-
tion at a very low titer (1/320 instead of 1/10,000) compared with the origi-
nal. Further testing has demonstrated that this strain retains the two major
0 antigens (IV and V), but has lost XII. Thus it has been transformed to IV,
V like strain #527. Further tests on differential media prove that the strain
is unchanged as an auxotrophic mutant, and still cannot grow unless the
medium contains histidine (519-10).

In the other case 519 0 serum was absorbed by S. schleissheim (V, XII,
XXVII). The mutant after growing through the absorbed serum failed to
agglutinate in XII serum, and had a higher titer in XXVII than S. schleiss-
heim. Thus the changed mutant has lost XII and taken on antigen XXVII.
It still retains its histidine requirement.

Thus we have two independent cases of the alteration of antigenic speci-
ficity by the Edwards method of passage through specific serum. Here again
the evidence indicated no relation between antigenic configuration and the
biochemical requirements. We are now exposing these antigenically altered
strains to further radiation with the idea of building up multiple auxotrophic
stocks combining the two major systems of mutations. These can then be
used for more conclusive tests of possible fusion and recombination. How-
ever, this demonstration that antigenic mutants can be induced by specific
serum adds to the possibility that mutual interaction of genes or gene
products between organisms in mixtures may give a more acceptable
explanation of the recorded cases of recombination in bacteria, than does one
based on genetic analogies with higher forms.

SUMMARY

An account has been given of the results of X-radiation of suspensions of
the two strains of Salmonella lyphimurium, and the isolation of strains with
specific nutrilite requirements (auxotrophic mutants). These strains are iso-
lated by the Davis-Lederberg method of growth for twenty-four hours in
enriched broth, followed by twenty-four hours in minimal broth containing
100 units per ml. of penicillin. The method screens out the unmutated organ-
isms according to a logarithmic survival curve, and preserves the mutant
bacteria.

Successive tests show a relation between X-ray dosage and the percentage
of recovered auxotrophic mutants, and also between dosage and the number
of different mutants.

GENETIC IMPLICATIONS OF MUTATIONS IN S. TYPHIMURIUM 281

In all, 249 separate auxotrophic mutants, of which 20 are different, were
isolated out of 459 tests. Most of these showed requirements for single amino
acids, but a few required the purine base adenine, and others showed alterna-
tive, and a small number, multiple requirements.

A large number of tests involving growth of multiple mutant stocks in
mixtures followed by re-isolations have been made to test for possible fusion
and recombination as reported by Lederberg and others in the K 12 strain
of E. coli. Recombination has been found but it is unlike that in sexually
reproducing organisms.

Detailed studies of the different auxotrophs requiring cysteine or me-
thionine show a step-like series beginning with loss of ability to reduce in-
organic sulphate, and continuing to the loss of ability to form methionine.
Many of these mutational steps are explainable as due to the inactivation of a
specific enzyme, but several require a complex pattern of chemical interac-
tions.

Similar studies of the adenine auxotroph suggest that adenine may be a
source of histidine.

Tests have been made to determine if antigenic specificity can be altered
by radiation, with negative results. However, an auxotrophic mutant has
been antigenically altered in two difTerent cases by the Edwards technique of
passing through absorbed immune serum. In each case, one of the 0 antigens
was removed, and in one case another 0 antigen was added. In both cases the
biochemical requirement of histidine was retained.

It appears that the auxotrophic and antigenic series represent two quite
different and unrelated sets of mutations.

JAMES F. CROW

Universify of Wisconsin

Chapter 1 8

Dominance
and Overdominance

Since the first attempts to explain hybrid vigor and the deleterious effects
of inbreeding in Mendelian terms, there have been two principal hypotheses.
Both were advanced early, and though each has had its ups and downs in
popularity, both have persisted to the present time. The first hypothesis is
based on the observed correlation between dominance and beneficial effect
(or recessiveness and detrimental effect). Inbreeding uncovers deleterious
recessives, and typically results in deterioration.

With hybridization, some of the detrimental recessives brought into the
hybrid zygote by one parent are rendered ineffective by their dominant
alleles from the other, and an increase in vigor is the result. If the number
of factors is large, or if there is linkage, the probability becomes exceedingly
small of a single inbred line becoming homozygous for only the dominant
beneficial factors. Consequently, there should be a consistent decrease in
vigor with inbreeding, and recovery with hybridization. This idea has been
called the dominance or the dominance of linked genes hypothesis.

The alternative theory assumes that there is something about hybridity
per se that contributes to vigor. In Mendelian terms this means that there
are loci at which the heterozygote is superior to either homozygote, and that
there is increased vigor in proportion to the amount of heterozygosis. This
idea has been called stimulation of heterozygosis, super-dominance, over-
dominance, single gene heterosis, cumulative action of divergent alleles, and
simply heterosis.

In accordance with the title of this discussion I shall use the words domi-
nance and overdominance for the two hypotheses. This leaves the word
heterosis free for more general use as a synonym for hybrid vigor (Shull, 1948).

* Paper No. 434 from the Department of Genetics, University of Wisconsin.

282

DOMINANCE AND OVERDOMINANCE 283

In most situations, the hypotheses of dominance and overdominance lead
to the same expectations. In either case there is a decrease of vigor on in-
breeding and a gain on outcrossing. Wright (1922c) has shown that with the
dominance hypothesis the decline in vigor is proportional to the decrease in
heterozygosis, regardless of the relative number of dominant and recessive
genes and of the degree of dominance. The same decline in vigor with de-
creasing heterozygosity is true with overdominance.

It is usually impossible in a breeding experiment to differentiate between
true overdominance in a pair of alleles, and pseudo-overdominance due to
the effects of two pairs of alleles closely linked in the repulsion phase. Only
in special circumstances, such as when a mutation has recently occurred in
an isogenic stock, can the experimenter be reasonably certain that the effect
is due to a single allelic difference. Furthermore, there is the possibility of
heterosis due to borderline situations, such as might arise in pseudoalleles
with a position effect, which could not even theoretically be classified as due
to dominance of linked genes or overdominance. Finally, it should be noted
that the various hypotheses may not be equally important in all situations.
For example, it is reasonable to expect that overdominance would be more
important in determining differences between inbred lines of corn pre-
viously selected for general combining ability than in lines not so selected.

If the two hypotheses are not mutually exclusive, neither are they col-
lectively exhaustive. There is no reason to think that multiple factors are
any less complex in their interactions than factors concerned with qualitative
differences. With the number of genes involved in heterosis, and with the
complexity of interactions known to exist in cases where individual gene
effects have been isolated and studied, there must surely be all sorts of com-
plex interactions in heterosis. Therefore no single theory can be expected to
account for the entire effects of heterosis. Although it is difficult to separate
by statistical methods the effects of dominance and epistasis, it may be
possible to construct simple models which are of some utility.

DOMINANCE

Davenport (1908) was the first to point out the now well-recognized fact
that in most cases the dominant character is beneficial to the organism pos-
sessing it, while the recessive has a weakening effect. He noted that this could
help explain the degeneration that usually follows inbreeding. Davenport
was thinking of relatively few factors with individually large effects, whereas
at present, more emphasis is given to multiple factors. But he was close to
the ideas now held.

Keeble and Pellew (1910) found that hybrids between two pure varieties
of peas were taller than either parent. In this case, two different dominant
factors were involved — one resulting in longer internodes and the other in-

284 JAMES F. CROW

creasing their number. Here only two gene pairs were involved, but it was
mentioned that similar systems might hold for more complex cases.

A more general development of the dominance hypothesis was given dur-
ing the same year by Bruce (1910). He designated the frequencies of domi-
nant and recessive alleles as p and q in one breed and P and Q in the other.
The array of individuals in the two groups will then be {p'-DD -\- IpqDR -\-
q^RRY and {PWD + 2PQDR + Q-RRY, where D and R are the dominant
and recessive alleles and n is the number of factor pairs involved.' If these
two populations are crossed, the mean number of homozygous recessive loci
is nqQ, whereas the average number for the two parent populations is
n{q- -f Q~)/2. The former is the geometric mean of the two parental recessive
genotype frequencies while the latter is the arithmetic mean. Since the geo-
metric mean is always less than the arithmetic, the number of homozygous
recessive loci will always be less in the hybrid population than the mean
number in the two parent populations. If either or both the parent popula-
tions are inbred the decrease will be greater.

Bruce then said:

If, now, it be assumed that dominance is positively correlated with vigor, we have the
final result that the crossing of two pure breeds produces a mean vigor greater than the col-
lective mean vigor of the parent breeds. ... I am aware that there is no experimental evi-
dence to justify the assumption that dominance is correlated with a "blending" character
like vigor; but the hypothesis is not an extravagant one, and may pass until a better takes
the field.

The average proportion of recessive homozygotes in the parents, which is
(9^ + '2")/2, may be rewritten as qQ -{- {q — Q)~/2. This is always larger
than qQ, the proportion in the hybrid, unless q and Q are equal. Although
Bruce didn't mention this, after one generation of random mating the propor-
tion of recessives in the hybrid population becomes {q + QY/'^ = (lQ'\~
(q — QY/4-, which shows that half the gain in vigor is lost as soon as ran-
dom mating begins.

Bruce concentrated his attention on the decrease of homozygous reces-
sive loci in the hybrid, and postulated a correlation between recessiveness
and deleterious effect. He could have used the same algebraic procedures to
show that crossing produces an increase in heterozygous loci, and thus based
a theory of hybrid vigor on overdominance. He showed remarkable foresight
in choosing the former, at a time when he had no evidence of a correlation
between dominance and beneficial effect.

1. The notation used by Bruce implies equal frequency of dominant and recessive
alleles at all loci. This assumption is not at all necessary for the argument, and I think
that what Bruce really meant was

Yl {p]DD-^2p,q,DR-^qfRR).

DOMINANCE AND OVERDOMINANCE 285

Objections to the dominance hypothesis were made largely on two
grounds. First, if vigor is not a product of heterozygosity as such, it should
be possible by selection to obtain individuals which are homozygous for all
the beneficial dominant factors, and hence have the same vigor as hybrids.
Secondly, in the Fo of a cross between two inbred strains there should be a
skew distribution of the trait being measured — since the dominant and re-
cessive loci would be distributed according to the expansion of (3/4 + 1/4)",
where n is the number of factors.

These objections were largely removed when Jones (1917) pointed out
that, with linkage, the consequences of the dominance hypothesis were
much closer to those postulating superior heterozygotes. If a detrimental
recessive were linked with a favorable dominant, the heterozygous chromo-
some would be superior to both homozygotes, and the linked combination
might not break up readily. Later, Collins (1921) showed that with a large
number of factors, regardless of linkage, the skew distribution disappears.
The probability of getting all the beneficial dominants into one homozygous
strain becomes vanishingly small, so the objections hold only if a small num-
ber of factors is assumed.

Most of the mutations known in Drosophila and elsewhere are recessive,
and practically all are in some way deleterious. Even if dominant and re-
cessive mutations were occurring with equal frequency, the deleterious mu-
tations in a population at any time would be mostly recessive, since the domi-
nants would be rapidly eliminated. It is to be expected — and it has been often
observed — that at most unfixed loci the recessive is deleterious in compari-
son with its dominant allele.^

Almost thirty years ago Sewall Wright (1922c) wrote:

Given the Mendelian mechanism of heredity, and this more or less perfect correlation be-
tween recessiveness and detrimental effect, and all the long-known effects of inbreeding — -
the frequent appearance of abnormalities, the usual deterioration in size, fertility, and con-
stitutional vigor in the early generations, the absence of such decline in any one or all of
these respects in particular cases, and the fixation of type and prepotency attained in later
generations — are the consequences to be expected.

It has been shown many times that populations actually contain a large
number of detrimental recessives — sufficient to account for a large decline in
vigor on inbreeding. In Drosophila pseudoobscttra, Dobzhansky et al. (1942)
found that almost every fly examined had at least one concealed lethal. Fur-
ther evidence that at least some heterosis is due to dominant favorable genes
is provided by the experiments of Richey and Sprague (1931) on convergent
improvement in corn.

2. I consider the statement that a dominant is beneficial and the statement that a reces-
sive is deleterious as meaning the same thing. Since a geneticist ordinarily can study gene
effects only by substituting one allele for the other, he cannot distinguish what each factor
is doing individually or whether it is harmful or beneficial except relative to its allele. That
is, he can only tell what the effect of the substitution is.

286 JAMES F. CROW

OVERDOMINANCE

The concept of a stimulating effect of hybridization began independently
with Shull (1908, 1911b) and East (1908). It was assumed that there was a
physiological stimulus to development which increased with the diversity
of the uniting gametes — with increasing heterozygosis. East (1936) elabo-
rated the idea further by postulating a series of alleles each having positive
action functions, and with these functions to some extent cumulative. As the
alleles became more and more divergent in function, the action was postu-
lated to become more nearly additive in the heterozygote.

At the time when East and Shull first formulated the hypothesis, there was
no direct evidence of any locus at which the heterozygote exceeded either
homozygote. For a number of years, overdominance as an explanation of
heterosis largely was given up because of the failure to find such loci.

Stadler (1939) pointed out that in certain of the R alleles in corn a situa-
tion obtains in which certain heterozygotes have more areas pigmented than
either homozygote. He suggested that genes acting in this manner could re-
sult in overdominance for such characters as size and yield. Other such loci
are known in corn.

There are now several cases in the literature of single genes with heterotic
effects. In most of these it is not possible to rule out the possibility of close
linkages giving pseudo-overdominant effects. In particular, many cases may
turn out to be pseudoallelism, but the consequences for the animal or plant
breeder would not be changed.

Several workers (Teissier, 1942a; Robertson, unpublished) have found per-
sistent lethals in Drosophila population cage experiments. If these are not
due to individually heterotic loci, extremely close linkage must be postu-
lated. Also certain recessive genes, such as ebony, come to an equilibrium
with their normal alleles in population cages. One of the most convincing
cases is that of the eye color mutant described by Buzzati-Traverso in this
volume. This mutant persists in the population, and was found in three in-
dependent stocks. It is quite improbable that in each of these cases the gene
happened to be linked in the repulsion phase with another harmful recessive.

The idea of superior heterozygotes has been upheld by Hull (1945) who
suggested the word overdominance. Hull's original argument for overdomi-
nance is a simple one. He noted that in most cases the hybrid between two
inbred maize lines has a greater yield than the sum of the two inbreds. This
would not be possible with dominant genes acting in a completely additive
manner — unless it were assumed that a plant with no favorable dominants
had a negative yield.

The validity of this argument depends on the unimportance of epistasis
in corn yields. Evidence on this point is very incomplete and somewhat con-
tradictory. Neal (1935) reported that the Fo yields were almost exactly inter-
mediate between the Fi and the average of the parents. This would suggest tha t

DOMINANCE AND OVERDOMINANCE 287

epistatis is not important or else that there is some sort of cancelling out of
various effects. On the other hand, Stringfield (1950) found that in many
cases backcrosses showed consistently higher yields than the F2. This sug-
gests some sort of interaction, as if some of the gene combinations selected for
during the inbreeding process were active in the backcross, but were broken
in the F2. None of these data give any evidence as to the importance of
epistasis in determining the difference between an inbred line and a hypothet-
ical line with none of the favorable dominants, since the data do not extend
into this range. It is in this range where non-additivity might be expected
to be most pronounced.

Hull's second argument is based on results obtained by the technique of
constant parent regression. The regression of Fi on one parent, with the other
parent held constant, has different expectations when there is overdominance
than when there is dominance. With overdominance the regression may be
negative when the constant parent is high-yielding, so the regression surface
is different from that expected with dominance. In this volume Hull gives
data which conform with this expectation.

Overdominance is not the only possible explanation of such results, as
Hull has pointed out. In addition, the constant parent regression technique,
or any technique making use of yield data on inbred lines, is complicated by
the difficulty of obtaining consistent results with inbreds. Another possi-
bility is that the factors responsible for yield in inbreds are largely different
genes from those determining the yield in the hybrids. This possibility will
be considered later.

For these reasons it is still not possible to be sure of the importance of
overdominance from Hull's methods. They are at least strongly suggestive,
and recent data from Robinson et al. (1949), obtained by an entirely differ-
ent procedure, also gave evidence of overdominance.

MAXIMUM HETEROSIS WITH THE DOMINANCE HYPOTHESIS

In this discussion several assumptions are made. Most of these have been
implicit in most discussions of heterosis, but it is best that they be clearly set
forth at the outset. The assumptions are:

1. Genes concerned with vigor are dominant, and in each case the domi-
nant allele is beneficial and the recessive deleterious. This is an assumption
of convenience which does not alter the essential nature of the hypothesis.
The conclusions still hold if dominance is not complete. Also there are loci
in which the recessive is advantageous or in which the heterozygote is inter-
mediate; but these are of no consequence for heterosis and therefore can be
omitted from the discussion.

2. There is complete additivity of effects between loci — no epistasis.

3. There are no barriers to recombination that prevent each gene from
reaching its own equilibrium frequency independently of other loci.

288 JAMES F. CROW

4. The gene and phenotype frequencies of the parent population are at
their equilibrium values.

5. Increased vigor results in, and can be measured in terms of, increased
selective advantage, though the selection may be natural or artificial. This
assumption restricts the discussion to those cases in which heterosis results
in changes in the same direction as selection had previously been acting.
Such an assumption appears to be valid for yield characters in field crops,
and for viability and fertility as is measured in Drosophila population
studies. It is highly questionable for such things as increase in size of hybrids
between wild varieties or species, where natural selection pressure may well
have been toward an intermediate size.

Under this assumption the increase of vigor on hybridization depends di-
rectly on the number of loci which are homozygous recessives in the parent,
but which become heterozygous in the hybrid. The individual or population
of maximum vigor is one in which every allelic pair contains at least one domi-
nant. The actual attainable heterosis would be less than this in any particu-
lar case.

Consider the case of complete dominance. The recessive phenotype is as-
sumed to have a selective disadvantage of s. That is, the dominant and re-
cessive phenotypes are surviving and reproducing in the ratio of 1 to 1 — 5.
The rate of mutation from A to a is u per gene per generation. Reverse muta-
tion will be ignored as it can be shown to have a negligible effect on the
equilibrium gene frequency attained.

Genotype A A A a a a

Frequency P 2Q R

Selective value 1 1 1 — 5

F-\-2Q-\-R= I

Under these assumptions, the frequency of gene A will be P -j- ^, while
the frequency of a will he Q -\- R. With mutation from A to a at rate u, the
frequency of .4 will be reduced in one generation by u{P -f- Q) and the
frequency of a increased by the same amount. Likewise, due to the effect
of selection, the frequency of a will be decreased by sR. Therefore the gene
ratio, {P -j- Q)/(Q + R), will change in one generation due to the effects
of mutation and selection to

iP + Q){l-u)

{P-hQ)u-\-Q-\-R- sR'

When equilibrium is reached the gene frequency will no longer change from
generation to generation which, stated algebraically, is

P^^ ^ {P + Q){l-u)

Q^R {P-\-Q)u-\-Q-\-R- sR-

DOMINANCE AND OVERDOMINANCE 289

This has the sokition, R = u/s. (For a more pedantic demonstration of this,
see Crow, 1948.)

The average reduction in selective value of the population due to a detri-
mental factor will be the product of the selective disadvantage of the factor
and the proportion of individuals possessing the factor. This amounts to
(s) (ti/s), or, simply, u, the mutation rate. Hence, the effect of a detrimental
gene on the selective value of the population is equal to the mutation rate to
that gene, and is independent of the selective disadvantage which that factor
causes, as was first pointed out by Haldane (1937). This fact, which at first
appears paradoxical, is readily understandable when one notes that a mildly
deleterious mutant persists much longer in the population, and hence affects
many more individuals than one which has a greater harmful effect.

The total effect on the population of all the loci capable of mutating to
deleterious recessives is simply the sum of the individual mutation rates as
long as the gene effects are additive. If there are n such loci with an average
mutation rate of u, the net reduction in selective value due to all homozygous
detrimental recessives at all loci in which they occur is nil. This is also ap-
proximately correct if the factors are multiplicative, provided the individual
effects are small.

The product ml is probably in the vicinity of .05 (Crow, 1948). This means
that if all the deleterious recessives were replaced by their dominant alleles,
the selective advantage of an equilibrium population would be increased by
about this amount. This could be considered as the maximum average im-
provement in vigor, as measured in terms of selective advantage, that could
occur due to hybridization. This means that the dominance hypothesis can-
not, under the conditions postulated, account for average increases of more
than a few per cent in vigor.

There are several reasons why the 5 per cent figure given above may be
too large. One is that many deleterious factors considered to be recessive
may not be completely recessive. Stern and Novitski (1948) and Muller
(1950) have shown that the majority of lethals and detrimentals that occur
in laboratory cultures of Drosophila are not completely recessive. Even if the
detrimental effect of the heterozygote is much less than that of the homozy-
gote, the greatest selection effect will still be on heterozygotes because of
their much greater frequency in the population. Thus, from the population
standpoint, these factors would be acting more like dominants than reces-
sives. This means that each locus would have a detrimental effect of 2u in-
stead of u (since a dominant gene would be responsible for twice as many
"genetic deaths" as a recessive), but the locus would be unimportant for
heterosis. Since the n in the formula refers only to the number of loci which
are capable of mutating to a completely recessive allele, its value may be
smaller than previously assumed and the product nu proportionately less.

It has been assumed that the parent populations are at equilibrium be-

290 JAMES F. CROW

tween selection and mutation pressures. This assumption probably is not
strictly correct for any population. Any equilibrium involving occurrences
as rare as mutations must be slow of attainment. Hence many if not most
populations must not be at equilibrium. Probably the most common way
in which a population gets out of equilibrium is by an alteration of the breed-
ing structure or population number so that the effective amount of inbreeding
is changed. If the change in population structure is such as to increase the
amount of homozygosity, a new equilibrium is reached comparatively rapidly
through the elimination by selection of the recessives which have been made
homozygous. On the other hand, if the change in population is such as to
decrease the amount of homozygosity a new equilibrium is attained only
through the accumulation of new mutations. This is an extremely slow
process.

Since the return to equilibrium is much slower when the population
changes in the direction of less inbreeding, it follows that most populations
which are out of equilibrium will be out in the direction of having too few
detrimental recessives. Therefore the effect of fluctuations in population
size and breeding structure will be on the average such as to increase the
fitness of the population. For this reason, the average loss of fitness per locus
is probably less than the mutation rate. Fisher (1949) has pointed out that
if the yield of a crop is near a "ceiling," the relative effect of each factor con-
ditioning yield becomes less. There will be a similar tendency for the popula-
tion to be out of equilibrium because of the slowness of occurrence of the
mutations required to bring the population to the new equilibrium level.

Another factor also pointed out by Fisher is that complete lethals and
highly deleterious factors contribute to the mutation rate but, at least in
grain crops, have no appreciable effect on yield since they are crowded out
by other plants.

All of these factors make the 5 per cent figure an overestimate, so it should
be regarded as a maximum. The true value may be much less. In this con-
nection Fisher (1949) said:

... it would appear that the total elimination of deleterious recessives would make less
difference to the yield of cross-bred commercial crops than the total mutation rate would
suggest. Perhaps no more than a 1 per cent improvement could be looked for from this
cause. Differences of the order of 20 per cent remain to be explained.

These considerations make it difficult to explain, in terms of the domi-
nance hypothesis, cases in which two equilibrium populations produce hy-
brids with considerable heterosis, or in which crosses between inbred lines
average appreciably more than the randomly mating populations from
which they were derived.

This discussion is relevant only when the character is measurable in terms
of selective value. For yield characters subject to any high degree of artificial
selection an increase in yield is probably accompanied by a greater proper-

DOMINANCE AND OVERDOMINANCE 291

tional increase in selective value. Thus any conclusions about maximum i)ro-
portional increase in selective value would hold a forliori for yield. Fisher
(1949) reaches a similar conclusion when he says: "If the chance of survival is
equated to the yield, as is reasonable with grain crops."

Another assumption is that the hybrids are compared with equilibrium
populations. There is room for question, particularly with domestic plants
and animals, as to whether selection has been occurring long enough and its
direction has been consistent enough for a gene frequency equilibrium to have
been attained. Another point that must be remembered in discussions of
maize is that commercial hybrids are not random combinations of inbred
lines, but highly select combinations. An average hybrid may have a yield
very close to that of a randomly mating population. Thus the argument of
this section may not be relevant for corn. But it can hardly be true that the
high yield of certain corn hybrids is due to the elimination of deleterious
recessives during inbreeding.

The quantitative limit placed on average improvement on hybridization
with the dominance hypothesis does not hold for overdominant loci. A locus
at which the homozygote A A has a selective disadvantage of .y with respect
to the heterozygote, and the homozygote A' A' has a disadvantage of /, will
come to equilibrium with gene frequency of A equal to t/(s + 0, and the
frequency of A' equal to s/(s + /) (Wright, 1931b; Crow, 1948). The average
reduction in selective advantage of the population due to the two homozy-
gous genotypes comes out to be st/{s + /). The loss in fitness of the popula-
tion is of the order of magnitude of the selection coefficients, as Haldane
(1937) has first shown, whereas with a detrimental recessive, the loss is of
the order of the mutation rate. Hence a single overdominant locus has a
tremendously greater effect on the population fitness than a single locus with
dominance or intermediate heterozygote. If such loci are at all frequent they
must be important. The question is: how frequent are they?

Even with overdominance it is difficult to understand large average in-
creases in selective advantage of hybrids between equilibrium populations.
Such populations should be somewhere near their optimum gene frequencies,
which means that the hybrids would be about the same as the parents. It
may be that, on the average, hybrids do not greatly exceed their parents in
selective advantage, and that the cases of increased size observed in variety
crosses and occasionally in species crosses are nothing but luxuriance. If so,
they are much less difficult to explain.

As Bruce showed in 1910, if the parents differ at all in gene frequencies,
the hybrids will be more heterozygous. If both parents are at equilibrium
they should have, for additive genes, approximately the same frequencies.
But what differences there are — due to chance, for example — will amount to
much more in an overdominant than in a dominant locus because the former
has a gene frequency much nearer .5.

292 JAMES F. CROW

POPULATION VARIANCE

The same considerations which show that an overdominant locus has a
much greater effect on average population fitness than a dominant locus also
show that an overdominant locus has a much greater effect on the population
variance. If the selective values of the three genotypes, AA,Aa, and aa are 1,
1, and 1 — s respectively, the frequency of aa genotypes is u/s and the aver-
age selective value 1 — u. The variance in fitness will be sii. On the other
hand, with an overdominant locus where the fitnesses of the three genotypes
are 1 — ^, 1, and 1 — ^, the mean fitness is 1 — s/2. The variance in fit-
ness is 5-/4.

The ratio of these variances is s/4u, which means that an overdominant
locus causes a population variance s/4:U times as great as that resulting from
a recessive locus of the same selective disadvantage. If 4u is 10"^, this
amounts to 100 for ^ = .001, or is 1000 for s = .01. This makes an over-
dominant locus with these selective values equivalent to 100 or 1000 ordinary
loci in its effect on the population variance. Haldane (1950) has emphasized
the importance of loci with adaptively superior heterozygotes in increasing
the variance of natural populations.

From this we must conclude that there doesn't have to be a very high
proportion of overdominant loci for overdominance to be the most important
factor in the genetic variance of the population. If much of the genetic vari-
ance of a population is due to overdominance, this would explain the great
slowness of selection. Characters with high genetic determination but low
parent-offspring correlation might be due to this cause.

The facts of hybrid corn also are consistent with this. Ordinary selection
has not been effective. Yet there is a great deal of variation in an open-
pollinated variety. It has been relatively easy to find combinations of inbred
lines that have yields well above the open-pollinated averages. There appears
to be a relatively high degree of genetic determination of yield, but relatively
low heritability. These results are not impossible with dominant genes, es-
pecially with epistasis, but are precisely what would be expected if some of
the variance were due to overdominant loci.

A population with many overdominant loci is always well below its maxi-
mum possible fitness. It is expected that such factors could eventually be
replaced in long evolutionary periods. This might occur by an appropriate
mutation, by duplication, or by modifiers. Or a population with too many
overdominant loci might disappear due to inter-population competition. But
at any particular time, a population may have a small proportion of such loci,
and it does not require many for these to be the major source of variation.

DO THE SAME GENES DETERMINE VARIATION IN
INBREDS AND HYBRIDS?

The rarer a recessive phenotype is in a population, the greater will be its
relative increase in frequency on inbreeding. If the frequency of the recessive

DOMINANCE AND OVERDOMINANCE 293

gene is q, the frequency of recessive homozygotes in a randomly mating
population is (f. With increasing amounts of inbreeding, the frequency
changes from q- to q. The smaller the value of q, the greater is the ratio of
q to q~. If a gene is highly deleterious it will be very rare in the population.
Hence the genotypes which are most deleterious are those which have the
greatest relative increase in frequency on inbreeding.

These relationships are brought out in the following figures, based on a
mutation rate of 10~^. The ratio given is the ratio of homozygous recessives
in a homozygous population as compared with one which is mating at
random.

Selective disadvantage (5) 0001 .001 .01 .1 lethal

Gene frequency (?) 1 -032 .01 .003 .001

Ratio (g/5=) 10 32 100 316 1000

This means that highly deleterious recessives, which ordinarily have an
effect on the population only of the order of the mutation rate, become much
more important wdth inbreeding and may become the major factors in deter-
mining the fitness of an inbred population. This might to some extent be
offset by selection during the inbreeding process, but such selection would be
directed against factors which are of no consequence in a more heterozygous
population.

The detrimental recessive factors referred to here include the lethals and
semilethals (such as chlorophyll deficiencies) that show up during inbreed-
ing. But more important are the larger number of factors, not individually
detectable, which collectively result in the loss of vigor with inbreeding de-
spite rigorous selection.

On the other hand, the major part of the variance of a non-inbred popu-
lation may well be determined by genes of intermediate frequencies, from
.1 to .9. The effect of such factors in determining the population variance
in fitness would change only slightly with inbreeding.

As an example, consider a hypothetical population mating at random
whose variance is made up of two components. Ninety per cent of the vari-
ance is due to relatively common loci with gene frequencies of the order of .5.
The other 10 per cent is due to loci with recessive gene frequencies of the
order of .01 or less. Now when this population is inbred without selection,
the variance due to the common genes will not change greatly but the vari-
ance due to the recessive loci will increase by a hundred fold or more. Thus
the factors which originally contributed only 10 per cent to the variance
may now contribute over 90 per cent of the variance between the various
inbred lines derived from the population.

Gene frequencies of the order of .5 might result from several causes. They
might be genes which are advantageous in one geograj)hical location and
disadvantageous in another so as to form a cline. Or there might be seasonal
differences in selective value. They may be due to complex interactions with
other loci or be of extremely small selective advantage or disadvantage. But

294 JAMES F. CROW

another explanation is selective superiority of heterozygotes (Haldane, 1950),
at least for those factors of importance in heterosis.

If yield is determined entirely by dominant factors, the correlation be-
tween inbreds and their hybrids should be positive. If it is due to over-
dominant loci, the correlation should be generally positive, though there
would be negative correlations between yield of hybrids and inbreds when
the other inbred is constant and high yielding. If both factors are involved
and overdominant loci are relatively important in hybrids while dominants
are important in inbreds, the correlation would approach zero. The experi-
ence of corn breeders has been that selection for yield during inbreeding is
relatively ineffective, and that the correlation of hybrid with inbred yield,
though positive, is small.

With overdominant loci the effect of a certain percentage increase in
heterozygosity is to cause the vigor to increase by a certain amount. De-
creasing the heterozygosity by the same percentage would cause a decrease
of approximately the same amount. On the other hand, with dominant loci,
making the original equilibrium population more heterozygous would cause
a very slight increase, whereas making the population more homozygous
would have a decreasing effect of a much greater amount. Therefore it is
easier to account for inbreeding depression by dominant loci than to account
for increase in vigor on hybridization above the level of a random mating
population.

I should like to suggest the following interpretation of the effects of in-
breeding and hybridization: The deleterious effects of inbreeding and the re-
covery on hybridization are mainly due to loci where the dominant is favor-
able and the recessive allele so rare as to be of negligible importance in a non-
inbred population. Variance of a non-inbred population, and hybrid vigor
when measured as an increase over an equilibrium population, are deter-
mined largely by genes of intermediate frequency, probably mostly over-
dominants.

OVERDOMINANCE AND GENE ACTION

In order to have overdominance it is not necessary that the immediate
gene products of the heterozygote exceed in quantity or variety those of
either homozygote. At the level of the immediate gene product, or any inter-
mediate state, the effect of the heterozygote may be intermediate between
the two homozygotes and still result in a greater final result. Any kind of
situation in which something is produced for which an intermediate amount
is optimum could be such that the heterozygote is nearer this optimum than
either homozygote.

A model for such cases is found in the sulfanilamide-requiring strain of
Neurospora reported by Emerson (1948). When this mutant is present the
heterokaryotic state of the suppressor gene results in more nearly the opti-

DOMINANCE AND OVERDOMINANCE 295

mum amount of para-amino benzoic acid than either homokaryon. Other
cases, less known biochemically, may be similar.

I think that it is doubtful whether such a system would persist for long
evolutionary periods. Alleles of intermediate productivity could arise and
replace the originals. Also modiiiers altering the expression of the homozy-
gotes would have considerable selection pressure. Or if the alleles were anti-
morphic, the situation might be resolved by duplication, as Haldane (1937)
has suggested. It is significant that the system reported by Emerson is not
one which is ordinarily of importance, but acts only in the presence of the
sulfanilamide-requiring mutant.

A form of gene action that appears more likely to account for instances
of overdominance is one in which the two alleles differ qualitatively or each
does something that the other fails to do. Instances of mosaic dominance
provide excellent examples. This has been demonstrated for the scute series
of bristle characters in Drosophila and for color pattern in beetles (Tan,
1946). Other examples are provided by the ^4 and R loci in maize.

Similar examples of physiological mosaic dominance are found where the
heterozygote apparently produces something approximating — at least
qualitatively — the total effect of the two homozygotes. An example is rust
resistance in flax, where each strain is resistant to a certain rust but the hy-
brid is resistant to both (Flor, 1947). By the usual tests for allelism, the two
resistance factors are alleles. Another series of examples is found in the
blood group antigens in man, cattle, and elsewhere. In almost every instance
the heterozygote has all the antigenic properties of both homozygotes
(Irwin, 1947). The presence of both the normal and abnormal types of
hemoglobin in humans heterozygous for the gene for sicklemia provides
another example (Pauling, 1950).

Many instances of overdominance may have a similar explanation. This
is the kind of action that East (1936) postulated in his discussion of heterosis
due to cumulative action of divergent alleles. It is not necessary that the
effects be completely cumulative; only that the net effect on the phenotype
be greater in the heterozygote than in the homozygote. Any system in which
the alleles act on different substrates to produce the same or different prod-
ucts, or convert the same substrate into different products — neomorphs, in
Muller's terminology — could result in overdominance.

Any of the examples listed above may turn out to be closely linked genes
(pseudoalleles) rather than alleles. In most cases it is impossible to distin-
guish between these alternatives. If the overdominance effect is due to
linked genes, eventually a crossover should result in a situation where the
desirable effects could be obtained in a homozygous individual. If there are
position effects, it may be that no homozygous arrangement is as advanta-
geous as one which is heterozygous. Unless there are position effects, it does
not seem likely that heterosis due to pseudoallelism would persist for any

296 JAMES F. CROW

great length of time, but in any particular population such factors might be
important.

IS INCREASED SIZE ADAPTIVE?

The foregoing arguments are based on the assumption that heterosis is
measurable in terms of increased selective advantage. The selection may be
natural or man-imposed. This assumption would appear to be reasonable for
such factors as fertility and resistance to disease. It also would apply to in-
crease in size or yield, if the direction of selection in the past were in this di-
rection, as in corn. However, it is questionable whether the increase in size
that is sometimes observed in variety hybrids is really adaptive.

Mather (1943) and especially Dobzhansky (1950) have emphasized that
increased size does not necessarily result in increased fitness in natural popu-
lations. Dobzhansky proposed the words euheterosis and luxuriance, re-
spectively, for increased selective advantage and for mere non-adaptive in-
crease in size. In these terms this discussion has dealt entirely with eu-
heterosis.

If euheterosis occurs in species or variety crosses, it is very difficult to
explain. It raises the troublesome question: How can the hybrid between
two well adapted strains be better adapted than its parents when there has
been no selection in the past for its adaptation? It may be that euheterosis
is developed only under some form of selection, as in the inversion heterozy-
gotes studied by Dobzhansky, or in the series of hybrids between inbred lines
of corn selected for combining ability.

If large size is not advantageous, luxuriance may be due to the covering
of recessive factors which were acting as size bottlenecks and had been
selected into the population because of this. Each of the parents might have
its growth limited by or held in check by a series of factors, and if some of
these were recessive, increased size would be found in the hybrids.

SUMMARY

Since the earliest attempts to explain hybrid vigor in Mendelian terms
there have been two principal hypotheses. The first of these is the domi-
nance hypothesis. This notes the observed correlation between recessiveness
and detrimental effect and attributes the increased vigor of heterozygosity
to the covering of deleterious recessive factors by their dominant alleles.
The alternative hypothesis, the overdominance hypothesis, assumes that
heterozygosity per se is important — that there exist loci at which the hetero-
zygote is superior to either homozygote.

It is clear that the dominance hypothesis is adequate to explain the de-
terioration that results from inbreeding and the recovery of vigor on out-
crossing, but it is difficult to explain how the hybrids could greatly exceed in
fitness the equilibrium populations from which their parents were derived.
The overdominance hypothesis demands the assumption of a kind of gene

DOMINANCE AND OVERDOMINANCE 297

action known to be rare, but it is pointed out that if only a small proportion
of the loci are of this type, these may nevertheless be the major factor in the
population variance.

The following interpretation is suggested: Inbreeding depression and re-
covery on crossing are mainly the result of loci at which the favorable allele is
dominant and the recessives are at low frequency. On the other hand the
variance of heterozygous populations and the differences between different
hybrids are due mainly to loci with intermediate gene frequencies. It appears
likely that such loci are due to selectively superior heterozygotes, but there
are several other possibilities.

LEROY POWERS

USDA, Bureau of Plant Industry

Chapter 19

Gene Recombinotion
and Heterosis

This article will be confined primarily to the tomato {Lycopersicon) genetic
work which has a bearing on gene recombination and heterosis. The barley
(Hordeum) genetic research which is discussed briefly was conducted at the
University of Minnesota. The tomato genetic research which constitutes the
bulk of the material discussed was conducted at the United States Horticul-
tural Field Station, Cheyenne, Wyoming.

With the present available methods of analysis it is difficult in quantita-
tive inheritance studies to distinguish between blocks of fairly closely linked
genes and individual pairs of genes. This has been shown by the work of
Jones (1917), Warren (1924), Mather (1942, 1949), and Straus and Gowen
(1943). Consequently, in this article where the two genetic systems are not
distinguishable the term pairs of genes will be employed. Mather (1949) has
used the term ejffedive factor to depict such a genetic situation.

MARKER GENES AND LINKAGE IN BARLEY

Powers (1936) has shown that in a cross between Bl {Hordeum deficieiis)
and Brachytic {Hordeum vulgare) the Fi, which is a two-row barley, gave a
greater yield of seed per plant than either the two-row or six-row parents.
Then, weight of seed per plant shows heterosis. The data on marker genes
and linkage in barley presented have some bearing upon whether any of the
advantages of the Fi hybrid attributable to heterosis can be recovered in
inbred lines through gene recombinations.

The deficiens (two-row) character was found to be differentiated from the
vulgare (six-row) character by one pair of genes designated as Vv, and the
brachytic character from the normal character by one pair of genes designated

298

GENE RECOMBINATION AND HETEROSIS

299

as Brbr. Using these symbols, the genotype of the Fi is VvBrbr. The Vv gene
pair is carried on chromosome 1 and the Brbr gene pair on chromosome 7.

Table 19.1 gives the comparative effect upon four quantitative characters
of genes associated in inheritance with Vv and vv and VV and vv, as deter-

TABLE 19.1

COMPARATIVE EFFECT UPON FOUR QUANTITATIVE CHARACTERS OF

GENES ASSOCIATED IN INHERITANCE WITH Vv AND vv, AND

VV AND vv; F-, GENOT\TES OF A BARLEY HYBRID

Weight

3F Seed*

Spikes per Pl.ant*

Height of Plant*

Length

3F Awn*

Genotvpe

Vv-vv

vv-vv

Vv-vv

VV-VV

Vv-vv

vv-vv

Vv-vv

VV-VV

BrBr

Brbr

brbr

-2.22
-2.98
-1.88

-3.44
-3.74
-2.74

1.72
0.94
0.13

0.21

0.39

-0.94

1.54
2.08
1.03

0.64

1.41

-0.68

16.58

16.42

1.95

7.. SO

9.20

-6.68

* Weight of seed per plant is expressed in grams, spikes per plant in number, height of plant in inches, and
length of awn in millimeters.

TABLE 19.2

COMPAR.VTIVE EFFECT UPON FOUR QUANTITATIVE CHARACTERS OF

GENES ASSOCIATED IN INHERITANCE WITH Vv AND VV, AND

VV AND vv; F. GENOT\TES OF A BARLEY HYBRID

Weight (

3F Seed*

Spikes per Plant*

Height of Plant*

Length

OF Awn*

Genotype

Vv-vv

VV-VV

Vv-vv

vv-vv

Vv-vv

vv-vv

Vv-vv

VV-VV

BrBr

Brbr

brbr

1.22
0.76
0.86

-3.44
-3.74
-2.74

1.51
0.55
1.07

0.21

0.39

-0.94

0.90
0.67
1.71

0.64

1.41

-0.68

9.08
7.22
8.63

7.50

9.20

-6.68

* Weight of seed per plant is expressed in grams, spikes per plant in number, height of plant in inches, and
length of awn in millimeters.

mined by differences between means of F2 plants. In every case, the differ-
ences between Vv and vv are greater than the differences between VV and vv
for spikes per plant, height of plant, and length of awn. With the exception
of the comparison between VV and vv within the brbr genotype, the differ-
ences are in favor of the two-row (Vv and VV) segregates as compared with
the six-row (vv) segregates. Within the brbr genotype, iJt) plants exceed the VV
plants for all three characters. As regards weight of seed per plant in every
case the six-row plants outyielded the two-row plants whether heterozygous
deficiens or homozygous deficiens. However, the differences between vv and
Vv were less than those between vv and VV.

The data of Table 19.2 show that for all characters the Vv plants give an

300 LEROY POWERS

increase over the VV plants, and with the exception of the Brbr genotype for
height of plant and length of awn, the differences of Vv-VV are greater than
the differences for VV-vv.

These facts concerning the data reveal that Vv is associated with an in-
crease in all four quantitative characters. For spikes per plant, height of
plant, and length of awn this increase results in heterosis.

Hypotheses for Difference in Vigor

If the increase noted is due solely to an interaction between V and v such
as is depicted by East's physiological hypothesis, then it would not be pos-
sible to obtain homozygous lines possessing any of this increase. However, if
the heterosis noted is due to a combination of favorable and unfavorable
genes linked with V and v, it should be possible to obtain lines in which
some of the favorable genes are recombined. These lines should show some
increase in the four quantitative characters studied. In the event that linkage
of genes favorable and unfavorable to an increase in the quantitative charac-
ters was found to furnish the most logical explanation, an intraallelic interac-
tion such as depicted by East's physiological hypothesis still may be having
some influence as the two systems are not mutually exclusive.

Tables 19.1 and 19.2 show that Vv results in an increase of all four charac-
ters: weight of seed per plant, number of spikes per plant, height of plant,
and length of awn. This fact is most simply explained by assuming the pro-
duction of a favorable growth-promoting substance which influences all of
them. Then such being the case, on the basis of East's (1936) physiological
hypothesis, V and v supplement each other, resulting in greater development.
Next consider the development of the lateral florets which determines the
number of rows of kernels per spike (two-row or six-row spikes). The Vv
segregates are two-row types, whereas the vv segregates are six-row types.
Hence, as regards the character number of rows of kernels per spike, the in-
teraction between V and v is such as to prohibit the development of the
lateral florets, resulting in a two-row barley spike rather than a six. Summing
up, on the basis of the physiological hypothesis, in the case of four quantita-
tive characters the interaction between V and v is such as to stimulate de-
velopment. In the case of number of rows of kernels per spike the interac-
tion is such as to prohibit development of the lateral florets. From physio-
logical genetic considerations such a pleiotropic effect seems rather im-
probable.

Explaining the heterosis associated with Vv plants on the basis of linkage,
a simple interpretation would be that the favorable linked genes and their
alleles interact according to Jones's (1917) hypothesis to produce a substance
favorable to growth processes, resulting in the heterosis noted; and that V
is dominant to v resulting in Vv (Fi) plants having two-row barley spikes.
This explanation does not require the assumption that V and v stimulate

GENE RECOMBINATION AND HETEROSIS

301

growth in one character and inhibit it in another, and hence is more in ac-
cord with modern physiological genetic concepts.

The article by Powers (1936) furnishes additional information j)ertaining
to gene recombination and heterosis. If genes other than Vv are responsible
for the heterosis noted, then F2 plants having a genotype identical to the Fi
generation should give a somewhat lower yield than the Fi. Since the Fi
plants were not grown in a randomized experiment with the F2 plants, the
comparison must be made through the Bl parent. As compared through the
Bl parent an actual reduction of one gram in yield of seed per plant was
found (Powers, 1936). This reduction could be due to genes carried on chro-
mosome 1, as are V and v, or to genes carried on other chromosomes. In
either event, theoretically some of the genes favorable to increased weight

TABLE 19.3

COMPARISON BETWEEN PARENTS AND F2 PAREN-
TAL GENOT\TES FOR WEIGHT OF SEED
PER PLANT IN A BARLEY HYBRID

Total

Weight of Seed per Plant in Grams

Number of
Plants

Fs

Parent

Differ-
ence

/

78 and 266*

64 and 63 f .

3.9
4.5

1.9
4.0

2.0
0.5
1.5

28.189
0 761

Interaction

5 807

* VVBrBr, genotype of Bl parent, two-row normal.

t vvbrbr, genotype of Brachytic parent, six-row Brachytic.

of seed per plant that resulted in the heterosis noted in the Fi population
must be capable of recombination.

Even though some of the genes favorable to increased growth can be re-
combined, the yield of the lines in which the favorable genes have been com-
bined depends upon the nature of the interaction of the genes. The weights
of seed per plant of parents and F2 plants of the parental genotypes are given
in Table 19.3. From this table it can be seen that the F2 plants of the VVBrBr
genotype gave an increased yield of 2.0 grams per plant over the Bl parental
plants having the same genotype. However, the F2 plants of the same geno-
type as the Brachytic parent gave an increase over this parent of only 0.5
grams per plant, which is not statistically significant. The interaction of 1.5
grams (Table 19.3) is statistically significant. This means that a preponder-
ance of the genes favorable to increased weight of seed per plant must have
entered the cross from the Brachytic parent. The balance of the unfavorable
genes that entered the cross from the Bl parent did not cause a correspond-
ing decrease in weight of seed per plant of the F2 plants possessing the vvbrbr
genotype.

302

LEROY POWERS

In this same study (Powers, 1936) found that the greater the number of
genes in the genotype tending to increase a character the greater is the effect
of any given gene. It is apparent that it is not possible to definitely predict
the yield of seed per plant resulting from recombining genes favorable to
growth because of the interactions noted. Either a greater or smaller increase
than expected may be obtained. Such interactions of genes would affect the
yield of plants in which the favorable genes were recombined, and hence the
feasibility of obtaining inbred lines equaling or excelling the Fi hybrid. In
some cases the probability of getting the desired results would be increased
and in other cases decreased; depending on the type of interallelic and intra-
allelic interactions of the genes.

GENE RECOMBINATIONS DIFFERENTIATING WEIGHT PER LOCULE
WHICH EXCEED HETEROSIS OF F, POPULATION

The data for weight per locule of fruit for the Porter X Ponderosa to-
mato hybrid and parental populations grown at Woodward, Oklahoma, in

TABLE 19.4

.ARITHMETIC AND LOGARITHMIC MEANS FOR
WEIGHT PER LOCUI.E OF PORTER X PONDERO-
SA TOMATO HYBRID AND PARENTAL POPULA-
TIONS*

Mea\

P0PUL.\TI0N

Arithmetic

Logarithmic

Porter

10.2
11.8
14.4
13.5
13.7
9.8

1.018253 + 0.012325

Bi to Porter ....

1.070936 + 0.009939

F,

Fs

Bi to Ponderosa

Ponderosa

1.168729 + 0.010134
1.128481+0.011879
1.124941+0.012651
0.982054 + 0.011845

* Grown at Woodward, Oklahoma, in 1041; original data taken in grams
and transformed to logarithms to obtain the means and standard errors of the
logarithms.

1941 (Powers, Locke, and Garrett, 1950) will be analyzed to determine
whether in Fo and backcross populations gene recombinations are occurring
which exceed the heterosis of the Fi population.

The means for weight per locule calculated on both the arithmetic and
logarithmic scales are given in Table 19.4. Weight per locule is greatest for
the Fi population, and the means of the Bi to Porter, F2, and Bi to Ponderosa
populations are larger than the means of the Ponderosa and Porter parents,
but smaller than the mean of the Fi population. The only means not showing
significant differences are the means of Porter and Ponderosa, and the means
of the F2 and Bi to Ponderosa populations. Hence, in these hybrid popula-
tions weight per locule definitely shows heterosis on either scale.

GENE RECOMBINATION AND HETEROSIS

303

The frequency distributions for weight per locule for the Porter X Pon-
(lerosa hybrid and parental populations are given in Table 19.5. This table
shows that the F2 and Bi to Ponderosa populations have plants falling into
classes of greater value than 1.51 1883, the last class in which Fi or Ponderosa
plants occur. There are nine such F2 i)lants and three such Ei to Ponderosa
plants. If no recombination of genes to produce plants with weight per locule
greater than the Fi plants is possible, these plants with values greater than
any individual of the Fi population must be chance deviates. Moreover, the
chance deviates must be those plants in the F2 population having the I'^

TABLE 19.5

OBTAINED FREQUENCY DISTRIBUTIONS FOR WEIGHT PER LOCULE
OF TOMATO FRUITS FOR PORTER X PONDEROSA HY-
BRID AND PARENTAL POPULATIONS*

Upper Limit of Class in Logarithms or Grams

Popula-
tion

0
d

d

00

d

On
rs

d

00

q

0

o\

so
On

q

00
0

00
•*

00

-0

00

0

10

00

m

0
0

00

00
00

OS

00
IT)

0

00
10

10

0

0
00
00

o>

Total
Plants

Porter... .

1

4

27

80

98

20

2

232

Bi to Por-

ter

1

3

13

54

81

102

80

49

35

16

11

1

1

1

448

Fi

1
13

6
31

22
68

35
8?

49

81

37
63

34
47

23
?4

13
17

4

11

2

8

3
3

4

1

4

?

?

1

233

F2

453

Bi to Pon-

derosa. .

1

6

17

29

45

71

72

62

52

26

19

18

1

4

4

4

1

1

1

434

Pondero-

sa

10

21

28

25

16

18

9

4

6

2

3

1

1

1

145

* Grown at Woodward, Oklahoma, in 1941; original data taken in grams and transformed to logarithms to
obtain the means and standard errors of the logarithms.

genotype or a very similar genotype. The probability of their being chance
deviates possessing the Fi or similar genotypes can be determined.

The mean of the logarithms of the Fi population is 1.168729, and the
standard error of a single determination is 0.123426. Calculations (for meth-
od see Powers, Locke, and Garrett, 1950) show that only 0.3 per cent of such
a genotypic population would be expected to have a value greater than
1.511883. The following tabulation shows the theoretical number of gene
pairs differentiating the parents, the theoretical percentage of the j)opula-
tion of the F2 or Bi to Ponderosa populations possessing the same genotyi)e
as the Fi, the theoretical number of plants of the Fi genotype in a popula-
tion of 453 F2 plants and in a population of 434 Bi to Ponderosa plants, and
the theoretical number of plants of the Fi genotype in the F2 population
and in the Bi to Ponderosa population expected to exceed a value of 1.511883.

An examination of the data opposite one pair of genes in the tabulation be-
low shows that only 0.68 Fo plants would be expected to exceed a value of

304

LEROY POWERS

1.511883, whereas 9 plants did so (see Table 19.5). The same comparison
for the Bi to Ponderosa population is 0.65 expected and 3 obtained. Also, a
study of the tabulation below reveals that with an increased number of
gene pairs diflferentiating the parents the odds become even greater against
those plants which exceed 1.511883 being chance deviates.

It remains to be seen whether plants of the Fi genotype plus plants of
genotypes which might have similar effects, but do not possess recombina-
tion of favorable genes in excess of the total number of favorable genes car-

Number

Pairs of

Genes

Per Cent
Population
OF Fi Geno-
type

Number of Plants
OF Fi Genotype

Number of Plants
Expected To Ex-
ceed A Value of
1.511883

F2

Bi

F2

Bi

1

50.00

25.00

12.50

6.25

3.12

226.50

113.25

56.62

28.31

14.17

217.00

108.50

54.25

27.12

13.56

0.68
0.34
0.17
0.08
0.04

0.65

2

3

0.33
0.16

4

0.08

5

0.04

ried by the Fi, could be responsible for the results noted. The result would be
to increase the proportion of the F2 and Bi to Ponderosa populations fluctu-
ating around means very similar in magnitude to that of the Fi population.
The extreme case (but highly improbable) would be to have all of these two
populations made up of such plants. On this basis and on the basis that the
parents are differentiated by one pair of genes, the number of plants of the
F2 population expected to exceed 1.511883 is 1.36, and for the Bi to Pondero-
sa population is 1.30. The number of plants obtained (Table 19.5) is 9 for the
F2 population and 3 for the Bi to Ponderosa population. Furthermore, the
Bi to Porter population had 1 plant in a class beyond that in which any Fi
plants occurred.

The analysis can be carried further. For the F2 population the number of
plants expected to exceed 1.562293 is 0.3223 and the number obtained is 3.
Whereas the values for the Bi to Ponderosa population are 0.3087 and 1,
respectively. Also, the frequency distributions (Table 19.5) in general do
not support the supposition that over one half of the plants of the F2 and Bi
to Ponderosa populations are fluctuating around a mean as great as that of
the Fi generation. Again with an increase in number of gene pairs differentiat-
ing the parents, the odds against the plants exceeding 1.562293 being chance
deviates become even greater. It is evident that the data are not in accord
with the assumption that plants of the Fi genotype have the greatest weight
per locule. This is true regardless of the number of gene pairs differentiating
the parents. Therefore, some of the plants falling in classes having values

GENE RECOMBINATION AND HETEROSIS 305

greater than 1.511883 must have genotypes composed of more favorable
genes than the Fi, and therefore recombinations of genes to produce plants
having a greater weight per locule than the Fi plants have occurred.

Whether inbred lines retaining this increased weight per locule can be
established is dependent upon the number of gene pairs differentiating the
parents and linkage relations (Jones, 1917). Close linkage of genes favorable
to increase in weight per locule would favor recombination. Whereas close
linkage of genes favorable to increase in weight per locule with those not
favorable would hinder recombination and hence reduce the chances of ob-
taining inbred lines retaining some or all of the advantages attributable to
heterosis.

The data furnish evidence concerning the number of gene pairs differen-
tiating weight per locule. From Table 19.5 it can be seen that the plants of
the F2 generation falling beyond the value 1.511883 are distributed over four
different classes, and those of the Bi to Ponderosa population falling beyond
this same value occur in three different classes. The behavior of these plants
cannot be explained on the basis of five or more independently inherited
pairs of genes, as there are too many of these plants falling beyond the
1.511883 class. In addition, the weights per locule of those falling in these
classes are greater than can be explained on the basis of chance deviation.

Further, to account for the plants of the F2 and Bi to Ponderosa popula-
tions falling in those classes beyond 1.511883, on the basis of five or more
pairs of independently inherited genes differentiating the parents, it would
be necessary to assume that 50 per cent or more of the plants were fluctuating
around a mean greater than that of the Fi generation. Since the means
(Table 19.4) of the F2 and Bi to Ponderosa populations are less than the
mean of the Fi, these populations cannot have a greater majority of the
plants fluctuating around a mean larger in magnitude than that of the Fi
plants. This deduction is confirmed by the frequency distributions of Table
19.5, as both of these populations have a greater percentage of their plants
in lower classes of the frequency distributions than does the Fi population.
Powers, Locke, and Garrett (1950) have shown that the data give a good fit
to frequency distributions calculated on the assumption that the parents are
differentiated by three pairs of genes.

Here, proof of recombination of genes to produce plants in the F2 and Bi
to Ponderosa populations with greater weight per locule than Fi plants is
fairly conclusive. Also, since the number of gene pairs or closely linked blocks
of genes is few, it should be possible by selection to establish inbred lines re-
taining this advantage.

MAIN AND COMPONENT CHARACTERS

The data from the parental and hybrid populations of tomatoes on the
main and component characters provide information concerning the rela-
tions between gene recombination, dominance, and heterosis.

306

LEROY POWERS

Weight of Fruit and Its Component Characters

The data on weight per locule, number of locules, and weight per fruit for
the Porter X Ponderosa hybrid and parental populations are given in Tables
19.4 and 19.6. On the arithmetic scale, smaller numbers of locules show par-
tial dominance. On the logarithmic scale the means of the Fj and F2 popula-
tions are not significantly different from the average of the means of the
Porter and Ponderosa populations. The mean of the Bi to Porter population
is not significantly different from the average of the means of the Porter
and Fi populations. The mean of the Bi to Ponderosa population is not sig-

TABLE 19.6

THE ARITHMETIC AND LOGARITHMIC MEANS FOR NUMBER OF LOCULES

AND WEIGHT PER FRUIT OF PORTER X PONDEROSA TOMATO

HYBRID AND PARENTAL POPULATIONS*

Number of Locules

Weight per Fruit

PoPUL.iTION

Arith-
metic

Logarithmic

Arith-
metic

Logarithmic

Porter

2.1
3.1
4.5
4.7
7.1
10.0

0.30707210.002151
0.468411+0.008158
0.637265 + 0.007663
0.628793+0.012522
0.82940410.007738
0.983292 + 0.017094

21.5
36.6
65.0
63.5
97.3
97.7

1.326101+0.012358

Bi to Porter

F:

Fo

Bi to Ponderosa . .
Ponderosa

1.539833 + 0.010394
1.806845 + 0.009416
1.762614 + 0.013078
1.95443010.013269
1.965097 + 0.008750

* Grown at Woodward, Oklahoma, in 1941; original data taken in numbers and grams and transformed to
logarithms to obtain the means and standard errors of the logarithms.

nificantly different from the average of the means of the Fi and Ponderosa
populations. Hence, on the logarithmic scale there is no dominance, and the
data indicate that the genetic variability follows the logarithmic scale. In
other words, the effects of the genes differentiating weight per locule are
multiplicative. This is true of both the intraallelic and interallelic inter-
actions.

Thus on the logarithmic scale number of locules shows no dominance,
weight per locule shows heterosis (Table 19.4) and the two combine addi-
tively to give weight per fruit. For weight per fruit the Fi indicates partial
dominance of greater weight per fruit, the Bi to Ponderosa complete domi-
nance, and the Bi to Porter no dominance. On the arithmetic scale the two
component characters unite multiplicatively, and the Fi indicates partial
dominance of greater weight per fruit, the Bi to Ponderosa complete domi-
nance, and the Bi to Porter partial dominance of smaller weight per fruit.
Then it is clear that regardless of scale, one of the component characters
shows some degree of dominance, the other heterosis. They combine to pro-
duce the main character which in turn shows some degree of dominance.

GENE RECOMBINATION AND HETEROSIS 307

Powers, Locke, and Clarrett (1950) found the number of major gene j)airs
differentiating number of locules to be 3. Since weight per locale was found
to be differentiated by 3 pairs of major genes, a comparatively few (prob-
ably 6) pairs of major genes differentiate weight per fruit. Hence, the number
of major gene pairs responsible for heterosis of weight per locule is no greater
than the number of major gene pairs responsible for no dominance of num-
ber of locules and partial or complete dominance of weight per fruit on the
logarithmic scale. Then, in this study the number of pairs of major genes
differentiating the character has no bearing on whether the hybrid popula-
tions will show no dominance, partial dominance, complete dominance, or
heterosis.

From these results it follows that in this material recombination of genes
to retain the advantages of heterosis is no different than recombination of
genes to combine desirable characters. Furthermore, these data furnish
rather convincing evidence that dominance and heterosis are different de-
grees of expression of the same physiological genetic phenomena, as was
postulated by Powers (1941, 1944).

Main and Component Characters of 45 Hybrids Produced
by Crossing 10 Inbred Lines of Tomatoes

Table 19.7 summarizes the dominance relations of the main and com-
ponent characters of 45 hybrids produced by crossing 10 inbred lines of
tomatoes.

The percentage values given in Table 19.7 were calculated from data pre-
sented in a previous article (Powers, 1945). The reader is referred to this
article for the experimental design, a description of the material, and meth-
ods. Here, only the method of compiling the data need be given. All of the
values of this table with the exception of those listed under heterosis were
calculated from the formula 100[2Fi/(Pi + A)]. The percentages listed
under the column headings "heterosis" were calculated from the formula
lOO(Fi/Pi) and 10G{Fi/F2), respectively. Fi is the mean of the Fi popula-
tion, Fi the mean of the parent with the smaller value, and A the mean of
the parent with the larger value. The 11 characters listed in Table 19.7 were
originally expressed in the following units of measurement: Spread of plant
in inches, yield per plant in grams, number of fruit that ripened per plant,
height per plant in inches, weight per locule of the fruit in grams, number of
days from first fruit set to first fruit ripe, number of days from first bloom
to first fruit set, weight of fruit in grams, number of days from seeding to
first bloom, number of locules per fruit, and number of days from seeding
to first fruit ripe.

The odds against any value belonging in an adjacent classification (col-
umn) are greater than 19: 1 with the exception of the two values designated
with an asterisk. Even for these two values the odds against their deviating

308

LEROY POWERS

more than one class are greater than 19: 1. When interpreting the data it is
necessary to have in mind that parental percentage values would have fallen
into the complete dominance columns, the Pi value into the first such col-
umn, and the Pi into the second such column. Also, it should be kept in mind
that the values listed in Table 19.7 are for the different Fi hybrids, and with
the exception of the values listed under the columns headed "heterosis" are
percentages based on the means of the two respective parents. The percent-
ages listed under the heterosis columns are based on the mean of the parent
that fell into the adjacent complete dominance columns.

TABLE 19.7

PERCENTAGE RANGE IN EXPRESSION OF DOMINANCE FOR
DIFFERENT CHARACTERS OF Fi TOMATO HYBRIDS*

Dominance

Character

Heter-
osis

Com-
plete

Partial

None

Partial

Com-
plete

Heter-
osis

Spread of plant . ....

114
166
172
112
109

122

Yield, ripe fruit per plant. . .

106
99

100
98

99

117
142
104

103

171

Number, ripe fruit per plant.

78

96
70

93

155

Height of plant

121

Weight per locule

119

Period, first fruit set to first
fruit ripe

90

80

'95'
69

93

Period, first bloom to first
fruit set

75
95
89
73

95

125*

VVpie^ht ner fruit

53
99
79

96

102

100*

96

Period, seeding to first bloom .

Number of locules per fruit. .

Period, seeding to first fruit

ripe

* As measured by percentages of averages of values of parents and percentages of parental values.

If dominance and heterosis are different degrees of expression of the same
physiological genetic phenomena, then the different genotypes, as represented
by the different Fi hybrids, might be expected to show ranges in expression
of a given character from different degrees of dominance to heterosis.

Every character listed in Table 19.7 except number of days from first fruit
set to first fruit ripe, in the different hybrids, ranges from some degree of
dominance to heterosis. Yield in grams of ripe fruit per plant, depending
upon the genotype (Fi hybrid), varied from no dominance through all classes
to heterosis for increased yield. Number of ripe fruit per plant and height of
plant varied through all classes from partial dominance of a decrease in mag-
nitude of these two characters to heterosis for an increase. Weight of fruit
in grams, number of days from seeding to first bloom, and number of locules
per fruit varied from no dominance to heterosis for a decrease of these char-
acters. Considering all of the characters there is a continuous array of values

GENE RECOMBINATION AND HETEROSIS 309

(that is values in all classes) from heterosis for decrease of a character to
heterosis for increase of a character, depending upon the character and geno-
type (Fi hybrid).

The most logical conclusion from these figures is that dominance and het-
osis to a considerable extent are different degrees of expression of the same
physiological genetic phenomena. This hypothesis is greatly strengthened by
findings of Powers (1941) that whether a character shows dominance or het-
erosis in some cases is dependent upon the environment and in other cases
upon the genotype. As pointed out previously, gene recombination in rela-
tion to heterosis is no different from combining any two desirable characters
by recombination of genes. A study of the component characters of the main
characters given in Table 19.7 offers further evidence in support of this
contention.

Yield of ripe fruit as determined by weight of fruit in grams is dependent
upon number of fruits that ripen and weight per fruit. The first of these
component characters, depending upon the Fi hybrid being considered, varies
from partial dominance of fewer number of ripe fruits to heterosis for an in-
creased number of ripe fruits. The second component character varies from
no dominance to heterosis for smaller weight per fruit. They combine mul-
tiplicatively, and in many cases result in heterosis for yield of fruit (Table
19.7 and Powers, 1944). Here again, then, is a case involving combination
of characters to produce heterosis. To retain some of the benefits of heterosis
in inbred lines would involve recombination of the genes differentiating the
two component characters.

In turn the number of fruit that ripens is dependent to a large extent at
Cheyenne, Wyoming, on earliness of maturity, number of days from seeding
to first fruit ripe (Powers, 1945). Earliness of maturity varies from partial
dominance of fewer days from seeding to first fruit ripe to heterosis for the
shorter period. The component characters of earliness of maturity are period
from seeding to first bloom, period from first bloom to first fruit set, and pe-
riod from first fruit set to first fruit ripe. Number of days from seeding to
first bloom varies from no dominance to heterosis for the shorter period.
Number of days from first bloom to first fruit set varies from complete domi-
nance of the longer period to heterosis of the shorter period. Number of days
from first fruit set to first fruit ripe varies from partial dominance of the longer
period to complete dominance of the shorter period.

Weight per fruit is dependent upon weight per locule and number of
locules per fruit. Weight per locule varies from partial dominance for less
weight per locule to heterosis for greater weight per locule. Number of
locules varies from no dominance to heterosis for fewer locules. On the
arithmetic scale these two component characters combine multiplicatively
so that weight per fruit varies from no dominance to heterosis for less weight
per fruit.

310 LEROY POWERS

From the above, as was found true for yield per plant, the heterosis noted
for earliness of maturity results from the combination of component charac-
ters which in certain Fi hybrids may themselves exhibit heterosis. The same
is true for weight per fruit. In other words, the study of genetics of heterosis
has been somewhat simplified by breaking the main characters down into
their component characters. Also, as before, the study shows that gene re-
combination to retain some or all of the increase of the Fi hybrid over the
parents is dependent upon the same physiological genetic phenomena as are
involved in attempting to combine two or more desirable characters into a
single inbred line.

RECOVERING INBRED LINES RETAINING ADVANTAGES
ATTRIBUTABLE TO HETEROSIS

The physiological genetic phenomena that hinder or aid, by the recombi-
nation of genes, the recovery of inbred lines retaining some or all of the
advantages attributable to heterosis are the same as those emphasized by
Jones (1917) and East (1936). These are the number of gene pairs differen-
tiating the parents, linkage relations of the genes, pleiotropy, and the inter-
action of the genes as determined by the measurement of end products, both
interallelic and intraallelic. This genetic information can be obtained only
by rather detailed genetic studies. With the quantitative characters such
studies are expensive and time consuming. Hence, very few such studies have
been made with tomato hybrids. Powers, Locke, and Garrett (1950) and
Powers (1950b) have made a gene analysis for some of the main characters and
their more obvious components. Even though the gene analysis for number
of days from seeding to first fruit ripe has been completed for only one of
the four crosses to be considered, this character and weight per locule will be
treated as component characters of yield of ripe fruit per plant in the section
dealing with number of pairs of genes differentiating the parents.

Number of Gene Pairs Differentiating Parents

In considering the bearing that number of gene pairs differentiating the
parents has upon gene recombination and heterosis, just two characters will
be considered: weight per locule and number of days from seeding to first
fruit ripe. That both of these characters have an effect upon yield of ripe fruit
should be kept in mind during the analyses and discussions which follow.
Also, other component characters listed in Table 19.7 could be studied. How-
ever, the additional information gained would not justify the time and space
required, as the fundamental principles involved can be brought out from
an analysis and discussion of the data for the two characters chosen. The
number of gene pairs (effective factors; Mather, 1949) differentiating
weight per locule has been determined for all the hybrid populations listed
in Table 19.8. For days from seeding to first fruit ripe the number of gene

GENE RECOMBINATION AND HETEROSIS

311

pairs (effective factors) differentiating the parents has been determined for
the Porter X Ponderosa hybrid populations only.

In discussing the bearing the number of gene pairs differentiating the two
parents has upon gene recombination and heterosis, information concerning
phenotypic dominance of the characters for the hybrid populations is neces-
sary and will be derived by studying the means of the parental and hybrid

TABLE 19.8

MEANS FOR WEIGHT PER LOCULE AND NUMBER OF DAYS FROM SEED-
ING TO FIRST FRUIT RIPE WITH TYPE AND NUMBER OF GENE PAIRS
DIFFERENTIATING THE PARENTS FOR WEIGHT PER LOCULE*

Danmark X

Danmark X

JOHANNISFEUER X

Porter X

Red Currant

JOHANNISFEUER

Red Currant

Ponderosa

Population

No. of

No. of

No. of

No. of

Weight

Days

Weight

Days

Weight

Days

Weight

Days

per

From

per

From

per

From

per

From

Locule

Seeding

Locule

Seeding

Locule

Seeding

Locule

Seeding

(Gm.J

to Fruit
Ripe

(Gm.)

to Fruit
Ripe

(Gm.)

to Fruit
Ripe

(Gm.)

to Fruit
Ripe

Pit

0.45

156.9

4.61

164.9

0.44

126.0

10.2

147.7

Bi to Pi

0.97

155.0

6.72

165,0

1.04

123.1

11.8

152.0

F,

2.33

153.8

7.96

165,6

2,70

118.9

14.4

149.6

F2

2.12

156.6

8.35

166.4

2.12

125.5

13.5

155.0

Bi toPz

4.82

159.7

8.32

167.6

4.48

124.7

13.7

168.8

P2t

10.36

169.8

9.92

170.0

6.20

136,1

9.8

204.8

Type and num-

ber of pairs

of genes

Minor

Major

Minor

Major

Major

40 -f

2 or

3

40+
Major
2 or
3

3

8

* For the hybrid populations of Danmark X Red Currant, Danmark X Johannisfeuer, Johannisfeuer X
Red Currant, and Porter X Ponderosa.

t Pi is Red Currant, Johannisfeuer, Red Currant, and Porter, respectively.
t P2 is Danmark, Danmark, Johannisfeuer, and Ponderosa, respectively.

populations given in Table 19.8. The means for weight per locule of tomato
fruits and number of days from seeding to first fruit ripe together with the
type and number of gene pairs differentiating the parents for weight per
locule for the hybrid populations of Danmark X Red Currant, Danmark X
Johannisfeuer, Johannisfeuer X Red Currant, and Porter X Ponderosa are
given in Table 19.8.

The first two hybrid populations were grown at Cheyenne, Wyoming, in
1938, the third hybrid population at the same location in 1939, and the last
hybrid whose means are listed in the extreme right hand column of Table
19.8 was grown at Woodward, Oklahoma, in 1941. The means of this table

312 LEROY POWERS

were taken from the following publications: Powers and Lyon (1941),
Powers, Locke, and Garrett (1950), and Powers (1950a). The data will be
analyzed to obtain information concerning the recombination of the genes
differentiating weight per locule and number of days from seeding to first
fruit ripe. Also, the data will be studied to ascertain the probable bearing
this information has upon the production of inbred lines, by gene recombina-
tion, that retain some or all of the advantages attributable to heterosis of
yield of ripe fruit per plant which the hybrid populations would be expected
to exhibit.

On the arithmetic scale the Danmark X Red Currant populations show
partial phenotypic dominance for smaller weight per locule. The parents of
the Danmark X Red Currant hybrid were found to be differentiated by a
large number of gene pairs (probably more than 40) which individually had
minor effects. From these results it is evident that, if somewhere near one-
half of the genes for smaller weight per locule in the Danmark X Red Cur-
rant hybrid populations had entered the cross from one parent and the
balance from the other parent, smaller weight per locule would have shown
heterosis. Some of the genes must be linked because the parents have a
haploid chromosome number of 12. In fact, since 40 or more pairs of genes
are differentiating the parents, it seems highly probable that a system of
linked polygenes is involved. With 40 pairs of genes differentiating the par-
ents in the F2, to recover an individual possessing all of the genes for in-
creased weight per locule (without linkage) would require a population of
lO^'* individuals. The size of such a population can be appreciated by con-
sidering the fact that 10" is 100 billion. The bearing this has upon the
feasibility of recovering from segregating populations inbred lines retaining
much of the advantage that might be exhibited by Fi hybrids is apparent.

The Red Currant parent which possesses small weight per locule also
possesses earliness of maturity. Hence, some of the genes tending to increase
weight per locule are almost certain to be located on the same chromosomes
with a non-beneficial gene or genes tending to increase the time required for
maturity. However, due solely to the large number of gene pairs differentiat-
ing weight per locule, with no close linkage, pleiotropy, or unfavorable in-
terallelic and intraallelic interactions of the genes, only a comparatively
small amount of the increased weight per locule of the Danmark parent could
be combined with the earliness of maturity of the Red Currant parent by
selection in the F2 or backcross populations.

Weight per locule and earliness of maturity have a material influence on
yield of ripe fruit per plant (Powers, 1945). In some crosses (see Tables 19.7
and 19.8) greater weight per locule is at least partially dominant. Since the
shorter period for days from seeding to first fruit ripe for the Danmark X
Red Currant cross shows heterosis (Table 19.8) the hybrid populations would
be expected to show heterosis for yield of ripe fruit per plant in crosses hav-

GENE RECOMBINATION AND HETEROSIS 313

ing such a polygenic system conditioning weight per locule, provided greater
weight per locule was at least partially dominant, and provided the genes
for increased weight per locule and shorter period from seeding to first fruit
ripe were divided between the two parents. The analyses and discussions in
the immediately preceding paragraphs show that in such an event it would
be almost impossible to obtain inbred lines which through gene recombina-
tion would retain any appreciable amount of the yield of the Fi hybrid.

On the arithmetic scale the Johannisfeuer X Red Currant populations
show partial phenotypic dominance of smaller weight per locule with the
exception of the Bi to P2 which indicates no dominance. The parents of the
Johannisfeuer X Red Currant hybrid populations were found to be differ-
entiated by a large number of gene pairs (probably more than 40) each of
which individually had minor effects and in addition by a few gene pairs
(probably 2 or 3) having major effects. In these hybrid populations the total
effect of the minor genes was greater than the total effect of the major genes.
Again the shorter period from seeding to first fruit ripe showed heterosis.

With the number and type of gene pairs conditioning weight per locule
found for the Johannisfeuer X Red Currant hybrid, and provided the genes
differentiating weight per locule exhibited at least partial dominance, as is
indicated for the Danmark X Johannisfeuer populations, certain parental
combinations of the genes would result in the hybrid populations showing
heterosis for increased yield of fruit per plant. Since comparatively few ma-
jor gene pairs differentiate weight per locule, it should be possible by re-
combination of genes through selection in F2 and backcross populations of
such a cross to combine into inbred lines some of the increased yield at-
tributable to heterosis.

The Danmark X Johannisfeuer hybrid populations show partial pheno-
typic dominance for greater weight per locule, and complete dominance for
shorter period from seeding to first fruit ripe. Two or three major gene pairs
were found to be differentiating weight per locule. For weight per locule and
number of days from seeding to first fruit ripe, dominance is such that had the
genes tending to increase each of these two characters been divided between
the two parents, the hybrid populations would have shown heterosis for both
component characters. Likewise, if the above conditions had been fulfilled,
yield of ripe fruit per plant would have shown heterosis in the hybrid popu-
lations.

The Porter X Ponderosa hybrid populations showed at least partial genie
dominance for weight per locule (Powers, Locke, and Garrett, 1950). The
parents were found to be differentiated by three pairs of genes and the genes
tending to increase weight per locule were distributed between the two par-
ents. As was to be expected, the hybrid populations showed heterosis for in-
creased weight per locule. Period from seeding to first fruit ripe showed al-
most if not complete dominance for the shorter period from seeding to first

314 LEROY POWERS

fruit ripe. The number of major gene pairs found to be differentiating the
parents was eight. Due to the magnitude of the work involved it was not
possible to measure yield of fruit, but in all probability the hybrid popula-
tions of this cross would have shown heterosis for yield of ripe fruit per plant.
In such an event it seems highly probable that some and perhaps a con-
siderable amount of the increase in yield attributable to heterosis could be
obtained in inbred lines through recombination of genes.

Considering the data for all the crosses listed in Table 19.8 the informa-
tion may be summarized as follows: In the Danmark X Red Currant cross
a large number of gene pairs differentiates the parents and individually the
genes have minor effects. The same is true of the Johannisfeuer X Red Cur-
rant cross with the exception that two or three pairs of genes have major
effects. In both the Danmark X Johannisfeuer and the Porter X Ponderosa
crosses weight per locule is differentiated by a comparatively few pairs of
genes having major effects. It is apparent that in the Porter X Ponderosa
cross it should be possible by selection in the segregating populations to ob-
tain by recombination of genes inbred lines equaling if not excelling the Fi
fruits in weight per locule.

The discussions treating weight per locule and number of days from seed-
ing to first fruit ripe as component characters of yield of ripe fruit per plant
reveal that the recombination of genes to retain some or all of the advantages
of the Fi hybrid is analogous to recombination of genes for the purpose of
combining desirable characters.

Linkage Relations

Linkage may be an aid or a hindrance to gene recombination. The data
in Table 19.9 were computed to facilitate a consideration of the manner in
which different linkage relations may affect recombination of genes.

Certain assumptions were essential to a calculation of the data. First, it
was assumed that the coefficient of coincidence is 1. Since in most cases there
is interference, to assume a coefficient of coincidence of 1 is to err on the
conservative side. For example, all the values given in the second row head-
ing (with the exception of the first and last) would increase as the coefficient
of coincidence became smaller. The reverse is true of the figures in the third
and fourth columns. The frequencies listed in the second, third, and fourth
columns of Table 19.9 are the theoretical number of individuals in the F2 pop-
ulation carrying the 12 plus genes in the homozygous condition. The cross-
over values expressed as decimal fractions are assumed to be equal for the
different sections of the chromosomes delimited by any two adjacent genes.

The conclusions to be drawn from the theoretical data of Table 19.9 are not
invalidated by these assumptions. They merely serve the purpose of allowing
the calculation of theoretical values for illustrative purposes. Other assump-
tions such as different values of crossing over for the various sections of the

GENE RECOMBINATION AND HETEROSIS

315

chromosomes and diflferent numbers of genes, combinations of genes in the
parents, and number of linkage groups would not alter the conclusions to be
drawn. In the illustration chosen only two linkage groups are shown and each
has three pairs of genes. Also, the top row of genes represents the gamete from
one parent and the lower row of genes the gamete from the other parent. In
all three assumed cases, 3 plus and 3 minus genes entered the cross from
each parent.

It is evident that innumerable plausible cases could be assumed, but the
fundamental principles derived from a consideration of the theoretical values
given in the table would not be altered. One further assumption should be

TABLE 19.9

THEORETICAL NUMBER OF INDIVIDUALS IN THE F2 POPU-
LATION THAT CARRY 12 PLUS (+) GENES WHEN THE PAR-
ENTS ARE DIFFERENTIATED BY 6 PAIRS OF GENES, EACH
OF 2 CHROMOSOME PAIRS CARRYING 3 PAIRS OF GENES*

Cross-

Linxage Relations in Fi (Number per Million)

over
Value

(^^^)(;;;)

(^i;)(;;^)

(^;i)(;i;)

0.000...
0.075...
0.225...
0.375...
0.450...
0.500...

62,500

33,498

8,134

1,455

523

244

0.000

1.448

57.787

188.596

234.520

244.141

0.000000

0.000063

0.410526

24.441630

105.094534

244.140625

* The crossover values for each section of the chromosome being equal and of the magni-
tude shown.

mentioned. In every case the plus genes are assumed to give an increase in
some desirable quantitative character and, comparatively, the minus genes
a decrease. Finally, in the table two extreme situations are shown, namely
that in which there is no crossing over and that in which the two sections
of the chromosome between adjacent genes show 50 per cent of crossing over.

The data in the second column apply to that situation in which all of the
plus genes occur in one member of the homologous chromosomes in each of
the two pairs of chromosomes depicted. In the case of 50 per cent of crossing
over or independent inheritance, only 244 individuals in a million of the F2
population possess all twelve plus genes. The number of such individuals
among a million F2 individuals increases with a decrease in the percentage
of crossing over until with no crossing over 62,500 individuals in a million
possess all six pairs of the plus genes in the homozygous condition.

The data in the third column apply to that situation in which two plus
genes are linked with one minus gene in one member of a chromosome pair
and two minus genes with one plus gene in the other member of the same

316 LEROY POWERS

chromosome pair. In this column the situation is reversed as compared to
column two. Again 50 per cent of crossing over gives 244 individuals among
a million in the F2 possessing all twelve plus genes. This decreases with a de-
crease in the percentage of crossing over until with no crossing over no indi-
viduals in the infinite F2 population contain more than eight plus genes.
However, since two of the plus genes are carried on the same chromosome
in each of the two linkage groups, an increase in the linkage intensity results
in an increased number of individuals in the F2 population possessing all
eight plus genes in the homozygous condition.

Here, then, is a case in which close linkage facilitates recombination of
desired genes up to a certain number, and from a practical standpoint further
advances by selection in that generation are impossible. Also, it would be
difhcult to make further advances by continued selection in later genera-
tions. In the F2 population with a crossover value of 0.075 the frequency of

the ( T T _ ) ( T T _ ) genotype e.xpressed as a decimal fraction is

0.183024 and of the (t t 7)(i t l) genotype is 0.014840.

To obtain some F3 families derived from F2 plants of the latter genotype
would require growing at least 300 selections in the F3 generation. To sepa-
rate the F3 families derived from the F2 plants of the former genotype from
those derived from the latter genotype would require an adequately replicated,
well designed experiment. Anyone who has worked with the quantitative char-
acters either in genetics or plant breeding realizes the difficulties besetting such
a task. After such F3 families had been determined, only 25 per cent of the in-
dividuals would be of the (TTT)(TT_) genotype. These would have
to be tested in the F4 to separate them from F4 families derived from F3 plants
of the (i i 7) (i t l) and the (^ ]j] ~) (^ ][] ~) genotypes. Even

with the small number of genes assumed in the above example, it would not be
a simple matter to make progress by continued selection in later generations.
The addition of a few more genes having the plus and minus genes alternating
on the same chromosome would make further progress by continued selection
in generations later than the F2 practically impossible. From the above it is
apparent that any series of plus genes being adjacent without minus genes in-
tervening would facilitate recombination of desirable genes in the F2 genera-
tion. It seems that in actual genetic and plant breeding materials many such
combinations do exist.

The figures in the fourth column of Table 19.9 are the theoretical fre-
quency distributions for that situation in which the plus and minus genes
alternate on the chromosome. Again the number of individuals expected in
the F2 generations possessing all twelve plus genes decreases rather rapidly
with a decrease in the percentage of crossing over. Even in the case of 50 per

GENE RECOMBINATION AND HETEROSIS 317

cent of crossing over it is doubtful whether it is possible for the plant breeder
or geneticist to isolate individuals from the F2 population carrying twelve
plus genes.

The data in Table 19.9 emphasize that even with the probably over-
simplified genetic situation depicted it is not possible to recover in a single
individual all of the genes favorable to the production of a desirable charac-
ter for which the F2 population is segregating, unless the favorable genes are
located on the same chromosome and immediately adjacent to each other
without unfavorable genes intervening. If any of the favorable genes are
adjacent to each other without unfavorable genes intervening, then decided
advances can be made by selection in the F2 populations up to a certain point.
Beyond that point further selection in the F2 will have no effect, and selec-
tion in advanced generations does not offer much promise. The most difficult
situation is that in which the linkage relation is such that the favorable and
unfavorable genes alternate on the chromosome and the number of such
linkage groups is at a minimum for the number of gene pairs involved.

For the sake of clarity of illustration only three linkage relations were
shown. However, it is apparent that undoubtedly in the material available
to plant breeders and geneticists, the possible different kinds of linkage rela-
tions are almost innumerable. Some will aid the investigator in obtaining the
desired recombination of genes and others will be a decided hindrance. In
the cases of undesirable linkage relations it will be almost impossible for the
breeder to obtain individuals possessing recombinations of genes making
that individual equal to or superior to the Fi for the character exhibiting
heterosis. On the other hand, desirable linkage relations may make it pos-
sible to obtain the recombination of genes sought even though a large num-
ber of gene pairs differentiates the parents used in hybridization.

Pleiotropy, and Interallelic and Intraallelic Interactions

Powers, Locke, and Garrett (1950) have made a rather detailed genetic
study of eight quantitative characters in hybrid and parental populations
involving the Porter and Ponderosa varieties of Lycopersicon esculentum
Mill. The characters studied and the indicated number of major gene pairs
differentiating the parents are as listed immediately below.

Character Gene Symbols

Percentage of flowers that set fruit FifiFifiFifiFifi

Period from seeding to first fruit ripe:

Seeding to first bloom BihiB-ibiBibz

First bloom to first fruit set SiS\S2Si,SiSi

First fruit set to first fruit ripe RxfiRiTi

Weight per fruit:

Number of locules LcilciLcildLcilci

Weight per locule WiWiWiW'zWzWz

With most quantitative characters it is difficult to distinguish between
pleiotropy and linkage. It seems highly probable that linkage instead of plei-

318 LEROY POWERS

otropy produced the relations noted by the above authors between the four
series of genes F/, Ss, Rr, and Lclc with the exception of the Ff and Ss re-
lation, because all the associations noted are those expected on the basis of
linkage. If pleiotropy were involved, such relations would be coincidental,
which for all these gene series is highly improbable. However, as pointed out
by Powers, Locke, and Garrett (1950) some of the genes of the Ff and Ss
series must be identical, as percentage of flowers that set fruit has an effect
on period from first bloom to first fruit set. The Lclc and Ww series of genes,
differentiating number of locules and weight per locule, respectively, were
independent as regards linkage and pleiotropy. In these studies pleiotropy
was not of major importance.

Phenotypic and genie dominance furnish some information concerning
the interallelic and intraallelic interactions of the genes. That genie domi-
nance is dependent upon the genotypic milieu was pointed out by Fisher
(1931) and many others (Dobzhansky, 1941). Hence both interallelic and
intraallelic interactions as measured by end products are second order inter-
actions, genes X genes X the environment.

Any of the interactions of genes noted as affecting any of the component
characters dealt with in the study by Powers, Locke, and Garrett (1950)
were interactions of genes differentiating yield of ripe fruit per plant. With
this fact in mind, it is interesting to note the interactions of the genes differ-
entiating the component characters. The intraallelic and interallelic interac-
tions of the Ff gene series were such that genie dominance was intermediate.
The intraallelic and interallelic interactions of the Bb series of genes were
such that one of the six dominant genes shortened the period from seeding to
first bloom as much as all six, which shows that both dominance and epistasis
were complete. For the 5^ series and Rr series of genes, genie dominance was
complete. Also, the effects of the gene pairs were cumulative.

Had the dominant genes of the Ss series entered the cross from one parent
and the dominant genes from the Rr series entered the cross from the other
parent, the Fi hybrid would have shown heterosis for earliness of maturity.
Porter would then represent an inbred line which by recombination of
genes retained the earliness of maturity of the Fi hybrid. Genie dominance
was partial for genes (LciLc^) tending to produce fewer locules per fruit and
for the (Lcs) tending to produce more locules per fruit. A series of genes such
as Lci and Lc2, some entering the cross from one parent and some from the
other, would produce an Fi hybrid showing heterosis for fewer locules per
fruit. On the other hand a series of genes such as (Lcs), some entering the
cross from one parent and some from the other, would produce heterosis for
more locules per fruit.

Finally, for the Wiv series of genes, genie dominance was partial for in-
creased weight per locule and the effects of the gene pairs were cumulative.
As regards this character, both parents did contribute genes for increased

GENE RECOMBINATION AND HETEROSIS 319

weight per locule, and the Fj hybrid did show heterosis for increased weight.
Also, as has been shown in the F2 and Bi and P2 populations some individuals
were obtained having greater weight per locule than the Fi plants and this
greater weight per locule proved to be due to recombination of favorable
genes.

Also, the interallelic interactions of the genes as determined by the inter-
relations of the component characters are of interest because of the informa-
tion they provide concerning recombination of genes and heterosis. The
effects of the Bb series of genes, the ^'^ series, and the Rr series, respectively,
were found to be cumulative. On an average the S genes would be e.xpected
to shorten the period from first bloom to first fruit set less in the presence
of the R genes than in the presence of the r genes — if the physiological reac-
tions affecting these two component characters that were instigated by the
environment were the same as those instigated by the 6'5 and Rr gene series.
That such was the case seems probable from the results of Goldschmidt's
work (1938) with phenocopies. In fact it seems almost axiomatic that this was
the case, because the second order interaction (Ss gene series X Rr gene
series X environment) was such that, on an average, when the ^5 series
responded to a given environment by shortening the period from first bloom
to first fruit set the Rr series in the same plant tended to produce a longer
period from first fruit set to first fruit ripe. Then the effects of these two
series of genes were less than additive as regards the dependent character pe-
riod from seeding to first fruit ripe

About the same situation existed in respect to the Lclc series and the Ww
series of genes in that greater number of locules, on an average, was ac-
companied by less weight per locule. This type of interallelic interaction
would tend to decrease the possibility of obtaining inbred lines combining
desirable characters. This would be particularly true of the interallelic inter-
action between the Ss and Rr gene series, because a shorter period from first
bloom to first fruit set tended to be accompanied by a longer period from
first fruit set to first fruit ripe

The data do not furnish any evidence concerning that type of intraallelic
interaction postulated by East's (1936) physiological hypothesis, other than
to say that no cases of overdominance were found. This would indicate that
probably overdominance does not play a predominant part in the produc-
tion of heterosis in the tomato hybrids studied.

A. J. MANGELSDORF

Experiment Sfation, Hawaiian Sugar Planters Assodaiion , Honolulu, T.H.

Chapter 20

Gene Interaction in Heterosis

Sugar cane behaves very much like corn in its reaction toward inbreeding
and outcrossing. Although the sugar cane flower is normally provided with
both male and female organs, male sterility is not uncommon. Among the
varieties that produce an abundance of pollen, many are partially or highly
self-sterile. As a consequence, cross-fertilization by wind-borne pollen is the
rule in sugar cane, as in corn. When sugar cane is subjected to self-pollina-
tion, the usual result is a reduction in seed setting and a marked reduction in
the vigor of the offspring.

The sugar cane breeder enjoys one great advantage over the corn breeder:
sugar cane can be propagated asexually. Each node on the stalk is provided
with a bud and with a number of root primordia. In field practice, stalks of
the selected variety of sugar cane are sectioned into cuttings of two or more
internodes each. These cuttings are then placed horizontally in furrows and
covered lightly with soil. In due course the cutting sends out its roots, the
buds develop into shoots, and a new plant is established.

Were it possible to apply this procedure to corn, and thus to perpetuate
outstanding individuals from whatever source, it is unlikely that the corn
breeder would have felt obliged to resort to the laborious procedures now
employed.

When sugar cane varieties are propagated by cuttings, the traits by which
we are able to distinguish one variety from another maintain their integrity
through many cycles of clonal propagation. This is true not only of morpho-
logical traits, but also of physiological traits.

Sugar cane has a number of relatives growing in the wild, some of which
may be ancestral to the original cultivated forms. Wild Saccharums are wide-
ly distributed in the tropical and sub-tropical regions of the Old World, from
central Africa through Asia and Malaya, to and including the Indonesian
and many of the more westerly Pacific islands. This heterogeneous array of

320

GENE INTERACTION IN HETEROSIS 321

wild forms has been somewhat arbitrarily classified into two great groups — -
the S. spontaneum group and the S. robiistum group. Each of these groups
comprises a diversity of tyj)es which differ among themselves in morphology
and in chromosome number. The members of the sponlanenni group have
slender stalks; they are often strongly stoloniferous. The members of the
robust U7n group have hard, woody stalks, sometimes of good diameter; sto-
lons, if present, are not strongly developed.

The original cultivated varieties likewise may be classified into two great
groups. The first of these comprises a number of slender varieties which ap-
pear to be indigenous to India, and which have been lumped together under
the name S. Barberi. Certain of the Barberi varieties bear a striking resem-
blance to the wild spontaneums of that region.

The New Guinea region is the home of a group of large-stalked tropical
cultivated varieties of the type which Linnaeus named S. officinarum. The
wild form most closely resembling S. officinarum and possibly ancestral to
it is S. robustum, which is indigenous to that region.

In the closely related genus Sorghum, the difference between varieties
having pithy stalks containing but little sugar, and varieties with sweet
juicy stalks, has been shown to be determined by a single major gene. In
Saccharum the change from the dry, pithy, low-sucrose stalks of the wild
forms to the juicy, high-sucrose stalks of the cultivated varieties appears to
have been brought about by several, but perhaps by no more than three or
four major gene changes.

The cultivated and wild forms also differ in genes for stalk size. In crosses
between the two, the genes responsible for the slenderness of the wild forms
show a high degree of dominance.

A striking feature of this multiform genus is the prevalence of inter-
fertility among its members. Widely divergent forms can be crossed without
undue difficulty. The resulting hybrids are rarely completely sterile; they
are often highly fertile. The explanation is presumably to be sought in the
polyploidy which is characteristic of both the wild and the cultivated forms.
They range in chromosome number from 24 to 80 or more pairs. It appears
that once the minimum chromosomal complement needed to produce a func-
tional zygote has been supplied, there is considerable latitude in the number
and in the assortment of chromosomes that can be added without impairing
the viability, or even the fertility of the hybrids.

Since the breeder is as yet unable to create superior genes at will, he is
obliged to content himself with developing new combinations of the genes
available in whatever breeding material he may be able to assemble. The
sugar cane breeder is fortunate in having in the wild relatives of sugar cane a
reservoir of genes for disease-resistance and hardiness. Those are traits that
had to some degree been lost in the course of domestication. Considerable

322 A. J. MANGELSDORF

use has already been made of the wild forms. The important varieties today
are almost without exception complex hybrids that include in their ancestry
representatives of both the S. officinarum and the S. Barberi groups of cul-
tivated varieties, together with representatives of one or both of the wild
species.

Thus the sugar cane breeder has been exploiting, to the best of his ability,
the advantages that heterosis has to offer. He is, however, acutely aware that
a better understanding of the genetic basis of heterosis is prerequisite to its
more effective utilization. Since he suffers the disadvantage of isolation from
the centers of research, he cherishes such rare opportunities as he may have
to peer over the shoulder of the research worker, to whom he must look for
new facts that may lead to a better understanding of the mechanism of gene
action and thus, of heterosis.

Recently some of us who are engaged in sugar cane breeding in Hawaii
formulated a number of postulates with the object of providing a basis for
discussing heterosis and related matters. These postulates have been ex-
cerpted or inferred from the published literature and from correspondence
with workers engaged in genetic research, whose helpful suggestions are
gratefully acknowledged.

Although the evidence supporting these postulates is sometimes meager,
and sometimes capable of other interpretations, we have deliberately phrased
them in a categorical vein in the belief that they might thus better serve
their primary purpose — that of provoking a free exchange of ideas.

POSTULATES RELATING TO INCIDENCE OF LESS FAVORABLE ALLELES

1. Naturally self-fertilized populations tend to keep their chromosomes
purged of all alleles other than those which in the homozygous condition
interact to best advantage with the remainder of the genotype and with the
existing environment^ to promote the result favored by natural selection (or
by human selection). This does not imply that any single population will con-
tain all of the best alleles existing in the species. Selection can make a choice
only between the alleles present in the population.

2. In addition to their prevailing (normal, plus, or wild type) alleles, cross-
fertilized organisms such as corn and sugar cane carry in the heterozygous
condition, at many loci, recessive alleles which in the homozygous condition
would be inferior in their action to that of their normal or prevailing partners.

3. These less favorable alleles may be thought of as belonging to one of two
classes, which, although differing in their past history, may have similar
physiological consequences: (a) fortuitous, resulting from sporadic mutation,
and representing the errors in the "trial and error" of the evolutionary proc-
ess; or {b) relic, representing the residue of what were once the prevailing

1. The term environment is here used in a broad sense to mean the sum-total of the ex-
ternal influences acting upon the organism, including its nutrition.

GENE INTERACTION IN HETEROSIS 323

alleles but which, in the course of evolution or under a changed environment,
have been displaced, to a greater or lesser degree, by still better alleles.

4. The prevailing allele at a given locus has reached its pre-eminent posi-
tion through the sifting action of natural selection over many generations.
Given a stable environment, further improvement, through mutation, at that
locus would long since have materialized if the chances for such improvement
were high. It is not strange that random mutation should only rarely be able
to produce a superior new allele. Nevertheless, once the possibilities for im-
provement through recombination of existing genes have been exhausted,
further evolutionary progress will be contingent upon just such an event,
however rare its occurrence may be.

5. Whether dominant or recessive, and whether in a naturally self-ferti-
lized or naturally cross-fertilized population, a substantially superior mutant,
once established in the population, is destined to increase in frequency and to
become the prevailing allele in the population.

6. A deleterious dominant is doomed to eventual extinction. In a cross-
breeding population of sufficient size a deleterious recessive may persist in-
definitely, its incidence, except for random drift, being determined by the
balance between its elimination by selection and the rate at which it recurs by
mutation.

7. The best allele for one environment may not be best for another envi-
ronment. The burden of less favorable alleles which cross-fertilized organisms
carry along generation after generation is not an unmitigated liability. It
serves as a form of insurance by providing a reservoir of adaptability to
changing conditions.

ROLE OF LESS FAVORABLE ALLELES

Turning now to the role of these less favorable alleles in the heterosis
phenomenon as manifested in naturally cross-fertilized organisms we may
formulate a second group of postulates:

1. At many and perhaps at most loci, .la is as good or nearly as good as
A A, and both A A and A a are better than aa.

2. There may be a few loci where aa is better than .LI or .la. This is par-
ticularly likely to be the case for loci affecting traits which are advantageous
under domestication, but disadvantageous in the wild under natural selec-
tion.

3. There may, for all we know, be occasional loci where .LI' is better than
A A or A' A' (overdominance).

4. There may be many regions in the chromosomes which behave as though
A A' were better than /L4 or A' A'. With deleterious recessive alleles in the
heterozygous condition at many loci, it seems almost inevitable that some of
these will be closely linked in the repulsion phase, as for example Ab/aB,
which in the absence of crossing over would behave as a single locus, the

324 A. J. MANGELSDORF

heterozygous condition of which is superior to either homozygote. It is to be
expected that such a linkage will eventually be broken. However, there may
be regions in the chromosomes, such as the centromere region, for example,
where crossing over is reduced, and where a group of genes may act indefi-
nitely as a single gene. We may for convenience designate the effect of such
reciprocal apposition of favorable dominants to their less favorable reces-
sives as a pseudo-overdominance effect. It will be noted that such a balanced
defective situation conforms with the dominance and linkage hypothesis ad-
vanced by Jones as an explanation of the heterosis phenomenon.

5. Even in the absence of linkage, an overdominance type of reaction (but
resulting from pseudo-overdominance) must assert itself whenever each of
the two members of a pair of gametes is able to supply the favorable domi-
nant alleles required to counteract the less favorable recessives carried by
the other member of the pair. The likelihood of success in retaining, in suc-
cessive generations of selfing, all of the favorable dominants heterozygous
in Fi, and eliminating all of the less favorable recessives, diminishes ex-
ponentially with increasing numbers of loci heterozygous in Fi. It would
seem that naturally cross-fertilized organisms which carry, at many loci,
deleterious recessives of low per locus frequency in the population could
hardly fail to manifest a pseudo-overdominance type of response to inbreed-
ing and outcrossing.

6. From an evolutionary standpoint, it may be important to distinguish
between the consequences of (a) true overdominance (heterozygosis at the
locus level) and {b) pseudo-overdominance (heterozygosis at the zygote level
resulting from the reciprocal masking of deleterious recessives by their
dominant alleles). From the standpoint of the breeder who is of necessity
working against time, this distinction may have little practical importance
if many loci are involved in the pseudo-overdominance effect. A breeding
plan designed to deal efficiently with one of these alternatives should be
effective also in dealing with the other.

7. Whether due to true overdominance or to pseudo-overdominance, the
widespread if not universal occurrence among naturally cross-fertilized or-
ganisms of an overdominance type of response to inbreeding and outcrossing
poses a problem which the breeder cannot afford to disregard.

8. Neither overdominance nor pseudo-overdominance can be called upon
to explain the differences in vigor between different varieties of wheat, beans,
sorghums, and other self-fertilized forms. Such differences are determined by
genes in the homozygous state, as are also the differences between homozy-
gous inbred lines of corn.

ROLE OF LIMITING FACTORS

A consideration of the role of limiting factors in quantitative inheritance
leads us to a third group of postulates:

GENE INTERACTION IN HETEROSIS 325

1. The adequacy of a diet is determined not by those constituents which
are present in ample amounts, but by those which are deficient to the point of
acting as limiting factors. Similarly the excellence of a genotype is deter-
mined not by its strongest but by its weakest links. The term weak link as
here employed refers to a gene pair at a particular locus which at some mo-
ment in the life of the organism proves so inadequate in performing the task
required of that locus as to act as a limiting factor — a bottleneck in an essen-
tial physiological process. A bottleneck efifect may result from a deficiency of
an essential gene product or from an excess of a gene product.

2. At each moment throughout its life the physiological processes of even
the most vigorous organism are held down to their prevailing rates by bottle-
necks or limiting factors. We are merely rephrasing a genetic axiom when
we say that a bottleneck in the physiological reaction system is neither purely
genetic nor purely environmental. The physiological bottleneck at any given
moment results from the interaction of a particular locus (which we may for
convenience refer to as the bottleneck locus) with the remainder of the geno-
type and with the environment of that moment. When we speak of an en-
vironmental bottleneck, we are merely focusing attention upon the environ-
mental component of the genetic-environmental bottleneck. When we speak
of a bottleneck gene, we are referring to the genetic component of the genetic-
environmental bottleneck.

3. The value of an otherwise perfect diet would be seriously impaired by
the omission of a single essential element. Similarly an otherwise superior
genotype could be rendered mediocre or worse by a single bottleneck. A po-
tentially superior genotype is unable to manifest its potentialities so long as it
is being throttled by a genetic-environmental bottleneck. A breeder looks at
the bottleneck and sees the need of a better allele at the bottleneck locus.
An agriculturist looks at the same bottleneck and sees the need for correcting
its environmental component. Bottlenecks relating to climatic limitations
usually can be most economically dealt with by breeding.- On the other hand,
bottlenecks resulting from nutritional deficiencies can often be advantage-
ously dealt with by correcting the environment.

4. The substitution, at a bottleneck locus, of a better combination of al-
leles^ will result in an improvement in yield providing that no other limiting
factor, genetic or environmental, asserts itself before an appreciable gain has
been realized.

5. The substitution of potentially better alleles at loci other than bottle-
neck loci cannot substantially improve yields any more than the addition of
calcium to the diet of a plant or an animal can relieve the effect of a phos-
phorus deficiency in that diet. We take it for granted that each essential

2. This rule is not without exceptions. For example, a bottleneck resulting from a
deficiency of rainfall can sometimes be economically eliminated by irrigation.

3. As already indicated, the best combination of alleles may be AA, Aa, or aa depend-
ing upon the particular locus.

326 A. J. MANGELSDORF

chemical element has its specific role to perform in the physiological reaction
system. Similarly we accept as well established the thesis that gene action is
likewise specific — that a particular gene can perform its particular function,
and that function only. Nevertheless we sometimes engage in speculations
which ignore these convictions and which appear to assume that genes affect-
ing quantitative characters such as yield are freely interchangeable, one with
another, and that one yield gene can serve as well as another, regardless of
its locus or function.

6. A bottleneck locus may act as such throughout the life of the individual
or it may act as a limiting factor only for a short period and under specific
conditions, such as drought, nitrogen deficiency, or excessively high or low
temperatures. Under a varying environment the bottleneck of one moment
may be superseded by a different bottleneck at the next moment.

7. The physiological bottleneck may be ameliorated or removed by correct-
ing the particular feature of the environment contributing to the bottleneck.
In the examples cited above this would entail supplying moisture, or nitrogen,
or lowering or raising the temperature. Or the bottleneck may be ameliorated
or removed by substituting a more effective allele at the bottleneck locus,
providing that such an allele is available.

8. As already indicated, the amelioration or removal of a bottleneck, either
by improving the environment or by substituting a better allele at the
bottleneck locus, will permit a rise in the rate of the essential physiological
processes. This rise may be small or it may be large, depending upon the
point at which the next ensuing bottleneck begins to make itself felt. The
substitution of a more efficient allele at a bottleneck locus in a certain geno-
type, under a particular environment, may result in a large gain. The substi-
tution of the same allele in a different genotype or under another environ-
ment may result in little or no gain. It is not strange that difficulty should
be encountered in analyzing the inheritance of genes affecting yield and other
quantitative characters which are subject to the influence of a varied and
fluctuating array of genetic-environmental bottlenecks.

9. A diet that is low in calcium may supply calcium at an adequate rate so
long as growth is being retarded by a lack of phosphorus. But once phos-
phorus is supplied at an adequate rate, calcium deficiency becomes a bottle-
neck which limits the rate of growth. Similarly a mediocre gene m at one
locus may be adequate (not a bottleneck) so long as the rate of physiological
activity of the organism is being throttled by environmental limitations or
by a bottleneck gene at some other locus. But once the other genetic-environ-
mental limiting factors have been removed, the mediocre gene m is unable
to handle the increased load and becomes the bottleneck in the reaction
system.

10. The maximum vigor or yield possible under a given environment will
be attained when the organism is endowed with the best available allele or

GENE INTERACTION IN HETEROSIS 327

combination of alleles at each bottleneck locus. There are presumably many
loci that never act as bottlenecks in any part of the reaction system affecting
vigor or yield, no matter which allele or combination of alleles happens to
occupy such a locus.

11. The difference between the weakest inbred and the most vigorous hy-
brid is merely one of degree. Each represents an integration of the many
genetic-environmental bottleneck effects under which it has labored. The
weak inbred has been throttled down by one or more bottlenecks to a low
level. The superior hybrid is able to go much further, even attaining what we
might concede to be extreme vigor. But both the weak inbred and the vigor-
ous hybrid have throughout their lives been held down to their respective
levels by their genetic -environmental bottlenecks.^

MISCELLANEA

The fourth and last group of postulates comprise a heterogeneous popula-
tion randomly listed as separate topics for discussion.

1 . If each step in a complex physiological process such as photosynthesis is
conditioned by the action of a specific gene, and if each successive step in the
chain of reactions is contingent upon the successful completion of the pre-
ceding steps, it follows that in attempting a biomathematical analysis of the
inheritance of quantitative characters such as yield we may not be justified
in assuming, as a basis for our calculations, that each of the genes concerned
is independent in its action.

2. Since our efforts to "improve" the genotype are constantly being
thwarted by bottleneck genes, we may be tempted to damn all such genes as
inventions of the Devil. No doubt there are many defective genes that would
have to be classed as liabilities under any normal environment. But certainly
there are many bottleneck genes that are indispensable to survival — genes
that act as governors in regulating physiological reactions and in fitting the
organism to its particular ecological niche. A mouse or a moss can survive and
reproduce where larger organisms would perish. And a mouse which, as a
result of changes in certain of its adaptive bottleneck genes attained the size
of a rat, might find itself at a disadvantage in a community of normal mice.

3. If we are correct in assuming that even a single major bottleneck locus
can act as a limiting factor in the development of an otherwise superior geno-

* Certain of the foregoing postulates pertaining to the role of bottleneck genes in quanti-
tative inheritance may be guilty of gross over-simplification. So complex is the physiological
reaction system of even the simplest organism that we are only now beginning to gain an
inkling of the extent of its complexity. These postulates may also be guilty of exaggeration.
Because we believe that the action of limiting factors in quantitative inheritance has not re-
ceived the attention that it deserves, we have intentionally stressed the importance of the
bottleneck locus, even at the risk of over-emphasis. Furthermore, we have pictured the
limiting factor at a given moment as pertaining to a single bottleneck locus. This may or
may not be the rule. It would not be difficult to imagine a bottleneck which pertains to sev-
eral loci and which could be relieved or eliminated by substituting a more effective allele at
any one of these loci.

328 A. J. MANGELSDORF

type, it is hardly to be expected that the phenotype of an inbred line will
afford a wholly reliable indication of its breeding potentialities in hybrid
combinations.

4. We need to keep in mind the limitations that pertain to a rating for gen-
eral combining ability. The best "general" combiner thus far discovered in
corn is not so general in its combining ability as to be able to combine to ad-
vantage with itself or with any other genotype that happens to be afflicted
with the same bottleneck genes. At best, a rating for general combining abil-
ity can represent nothing more than an average arrived at by lumping a given
population of specific combinations. An average derived from a different pop-
ulation of specific combinations could result in quite another rating.

5. If aseriesof inbreds^,5, etc., be crossed with a tester inbred T, we ob-
tain the hybrids AT, BT, etc. The yield of AT will be determined by the
bottleneck genes in the A T genotype. The yield of BT will be determined by
the bottleneck genes in the 5 T genotype. The test cross can tell us which lines
combine to best advantage with the tester line, but it cannot reasonably be
expected to tell us more than that. It cannot, for example, tell us with cer-
tainty what we may expect from A X B. Both A and B may combine to
advantage with T, but if A and B each happen to be afflicted with one or
more of the same bottleneck genes (not present in T) the yield of the cross
AB will suffer.

6. The failure of a cross between two convergently improved lines to equal
the cross between the two original lines from which they were derived cannot
be taken as critical evidence for the existence of an overdominance mecha-
nism. The benefits which convergent improvement seeks to achieve can be
vitiated if a recessive bottleneck gene b, present in only one of the original
parent lines, should become homozygous in both convergently "improved"
lines. Selection exercised with the object of preventing such an occurrence
may be ineffective if b becomes a bottleneck only under the enhanced rate of
physiological activity of the A{B) X B{A) hybrid.

7. During recent years several examples of heterosis reported in the litera-
ture have been attributed to the effect of heterozygosity at a single locus.
When the amount of heterosis is substantial, it should be possible to verify
the validity of the hypothesis by breeding tests. If the two parents are really
isogenic, except for the heterosis locus H, and if HiH2 individuals are more
vigorous than either homozygote, then by selfing only the most vigorous in-
dividuals in each generation it should be possible to retain in one-half of
the population the original vigor of Fi even after many generations of selfing,
and such a line should continue indefinitely to segregate HiHi, HiH-i, and
H<iH2 individuals in a 1:2:1 ratio.

8. East describes the effect of heterosis as "comparable to the effect on a
plant of the addition of a balanced fertilizer to the soil or to the feeding of a
more adequate and more chemically complete diet to the animal." The simi-

GENE INTERACTION IN HETEROSIS 329

larities noted by East between the beneficial effects of heterosis and those of
improved nutrition are more than coincidental. The first prerequisite for en-
hanced well-being is the removal of the bottlenecks that stand in the way-
This can sometimes be accomplished by improving the nutrition, sometimes
by substituting more efficient alleles at the bottleneck loci, and sometimes by
both.

9. The term heterosis remains ambiguous in spite of the many attempts to
define it. It continues to have different meanings for different workers.

10. If heterosis is to be measured by comparing performance of offspring
with performance of parents, then the higher the standing of the two parents
in the scale of measurement, the lower the degree of heterosis to be expected
in their offspring. Conversely the lower the standing of the parents, the great-
er the heterosis to be expected. (Exceptions to the latter rule will occur when
both parents owe their enfeeblement to the same bottleneck genes.)

1 1. Success in crop and livestock production depends largely upon the skill
of the grower in detecting, diagnosing, and correcting the environmental com-
ponents of the bottlenecks affecting yield. Success in developing higher yield-
ing genotypes depends largely upon the ability of the breeder to substitute
more effective alleles at the bottleneck loci, and to accomplish this without
establishing new and equally serious bottlenecks at other loci.

GORDON E. DICKERSON

UnivarsHy of Missouri

Chapter 21

Inbred Lines for Heterosis Tests?

The justification for considering heterosis tests in breeding work rests on the
mode of action and interaction of the genes responsible for genetic variability
in the material available to the geneticist. The nature of this genetic variabil-
ity may vary widely between species or populations in response to differences
in the degree of inbreeding and kind of selection, natural or imposed, that
has characterized the population over an extended period. For any given
trait or combination thereof, structure of genetic variation will depend upon
how consistent, intense, and prolonged selection has been.

It follows that choice of the system of mating and selection appropriate
for most rapid improvement in economic attributes of any given plant or
animal population should be guided by as complete knowledge of the kind
of genetic variation in the population as analysis of all available data affords.
The discussion which follows is an attempt (1) to interpret the evidence
presently available concerning the sort of genetic system which underlies
important economic traits, using swine as the example; and (2) to compare
expected effectiveness of several alternative breeding methods.

NATURE OF GENETIC VARIATION IN ECONOMIC TRAITS

Types of association between the genotype and its phenotypic expression
have been classified logically as intra-allelic and inter-allelic. The former
includes all degrees of dominance or levels of expression for the heterozygote
relative to the corresponding homozygotes. The concept of heterozygote ad-
vantage or overdominance differs from the usual ideas of dominance in that
each gene is visualized as exerting certain dominant favorable effects lacking
in its allele. Inter-allelic gene action or epistasis includes all effects of a gene
in one set of alleles on the expression of genes in other sets of alleles. Comple-
mentary, inhibiting, duplicate dominant, and duplicate recessive gene inter-
actions are extreme examples.

330

INBRED LINES FOR HETEROSIS TESTS? 331

By definition, epistasis is universal in the sense that expression of every
gene is to some degree dependent on and modified by the effects of genes in
other sets of alleles. Epistasis would include fixed multij)licative or propor-
tional effects of each gene on the expression of non-allelic genes. Such
epistasis, although unlikely to be important, would be of a highly predictable
sort and would disappear if phenotypes were measured in log scale units.
A potentially much more important sort of epistasis would be that involved
whenever a phenotypic maximum is associated with an optimum genetic
intermediate (Wright, 1935). Here a given gene may have either a positive
or a negative selective value, depending on whether the individual's average
genotype is above or below the optimum genetic intermediate.

Some of the evidence concerning the kind of genetic variability with which
we must deal in seeking to improve economic characters of swine has been
considered earlier (Dickerson, 1949, 1951) and may be summarized here as
follows:

Inbreeding and Crossbreeding Effects

Proportion of heterozygous loci has a major influence on total perform-
ance, affecting most the highly important but lowly heritable characters for
which selection has been consistently in one direction. Take for example, an
intra-season comparison of 538 inbred and 325 linecross litters from the same
lines in four projects of the Regional Swine Breeding Laboratory (Dickerson
et al., 1947). This showed a decline in performance per 10 per cent increased
inbreeding of litter amounting to 2.6 and 7.8 per cent, for litter size at birth
and weaning, respectively; 2.6 per cent for pig weight at 154 days of age; and
11.4 per cent for total weight of litters at 154 days. Similar estimates per 10
per cent increased inbreeding of dam, based on sixty-three inbred and fifty
linecross dams at the Iowa Station, were 2.1 and 5.0 per cent for litter size
at birth and weaning; 1.6 per cent for pig weights at 21 days; and 5 per cent
for total weight of litters at 154 days.

Results from studies of regression of j)erformance on inbreeding of dam
and litter within line and season (Blunn and Baker, 1949; Stewart, 1945;
Comstock and Winters, 1944; and Hetzer et al., 1940) agree quite well with
the figures given. Inbreeding of dam and litter together greatly depresses
prolificacy, suckling ability, pre- and post-natal viability and growth rate,
and particularly their product — total litter weight. Inbreeding effects on
carcass composition, body conformation, and efficiency of food utilization
were relatively minor (Dickerson et al., 1946).

The results of the earlier crossbreeding exi)eriments have been summarized
by Lush (1939) and Winters (1936). When the mean of the two purebred
stocks crossed is compared with the crossbred litters, the results of many ex-
periments summarized by Carroll and Roberts (1942) indicate that the
average performance of crossbred individuals is increased about as much as
it would be by a 10 per cent reduction of inbreeding (see Table 21.1). More

332

GORDON E. DICKERSON

recent studies of crossbreeding using inbred strains (Hazel el al., 1948; Sierk,
1948) verify the earlier conclusions.

Some degree of dominance is the most obvious genetic mechanism by
which change in heterozygosity from inbreeding or crossbreeding would affect
the level of performance. Inbreeding decline due to dominance would be a
function of 2q{l — q)k f, where q is frequency of the dominant allele, / is
Wright's inbreeding coefficient, and k is the degree of dominance (Hull,
1945) defined in terms of phenotypic scale as (2 Aa-AA-aa)/{AA-aa).

TABLE 21.1

RESULTS OF CROSSBREEDING EXPERIMENTS SUM-
MARIZED BY CARROLL AND ROBERTS (1942)

Factors of
Production

No. of
E.xpts.

Mean of Two
Pure Breeds

Mean of
Crossbreds

Relative Per-
formance of
Crossbreds
with Purebreds
= 100

No. pigs per litter

Birth weight of pigs (lbs.) .

Survival ability (%)

Weaning wt. of pigs (lbs.) .

Weaning wt. of litters

(lbs.)*

12

6

15

15

13
9
6

32
32

9.74
2.77

76.3

32.5

235.6

1.381
374.1

1.359
345.4

9.48
2.79
80.2
33.12

254.1

1.436
368.6

1.381
344.3

97.3
100.6
105.1
101.8

107.9

Av. daily gain (lbs.)

Feed for 100 lbs. gain (lbs.)
Danish pig-testing sta-
tions:

Av. daily gain

Feed per 100 lbs. gain . .

104.0
98.5

101.5
99.7

* From the original publications of these experiments.

If genetic intermediates in one or more primary functions produce maxi-
mum performance, the increased total genetic standard deviation (V 1 + /)
associated with inbreeding would tend to increase the average deviation from
optimum genotype and hence depress performance roughly in proportion to
(vl + / — !)• Inbreeding alone would not alter mean level of performance
without dominance, if only epistatic factors of the complementary or dupli-
cate sort were involved.

Inbreeding depression and crossbreeding advantage indicate some degree
of dominance or of genetic intermediate optimum, but, alone, they fail to dis-
tinguish between the two or to indicate the probable degree of dominance.

EfFectiveness of Selection within Inbred Lines

Selection within mildly inbred lines has been only slightly effective. De-
cline in performance with mild inbreeding (2 to 4 per cent per generation)
has been much the same as would have been expected from inbreeding with-
out selection. These statements are based largely on a study' of time trends

1. To be published in more detail, separately.

INBRED LINES FOR HETEROSIS TESTS?

333

in litter size and growth rate in 49 inbred lines from five projects with an
average of 9 seasons per line (see also Dickerson, 1951). In Figure 21.1 the
average actual linear time trend (solid line) is negative for both litter size at
weaning and for pig weight at 154 days of age. An estimate of the eflfective-
ness of selection was made by adjusting the time trends for the effect of the
increased inbreeding, using corrections derived from the intra-season com-
parison of inbreds and linecrosses from the same inbred lines involved in the
time trends. The adjusted time trend (dashed line) indicates that selection

.75

/

.50

■

H

/'

3

y

a,

/

Q

.25

-

/

H

/

Z

/■

<

/

W

/

:*

/

CO

/

O

00

-^^-

0.

~~~~"

^^^^

.25

^^^^^-^

TIME IN YEARS

Fig. 21.1 — Linear time trend within mildly inbred strains for pigs weaned per litter and
154-day weight per pig. Solid line is actual trend, dashed line is trend adjusted for effect
of inbreeding trend to non-inbred basis, and the top broken line indicates mean superiority

of selected parents.

has failed to improve genetic merit for litter size and has allowed growth
rate to decline, although selection of parents per year has averaged about
.6 pigs for size of litter weaned by the dam and sixteen pounds for pig
weight at 154 days (top broken line).

These results must be accepted with caution, because time trends can be
influenced by trends in nutrition, parasites, disease, management, or other
factors. Also, the correction for inbreeding effects may have been underesti-
mated. It seems clear that improvement has been at best only a small frac-
tion of what would have been expected from the heritability of these traits
and the amount of selection practiced for each. Evidence from comparison of
intra -breed linecrosses with representative purebreds is meager but does not
suggest any major improvement. Intra-herd comparisons of viability and
growth rate of progeny from inbred and from representative purebred boars

334 GORDON E. DICKERSON

(Hazel et al., 1948) likewise have shown little advantage accruing from
selection during development of the inbred lines.

The apparent inability of selection to offset the decline in performance
from mild inbreeding casts doubt on the assumption that epistasis or ordi-
nary dominance (between none and complete) can account for the major
influence of inbreeding on performance in swine. Unless one assumes a pre-
ponderance of tight repulsion phase linkages, selection should have increased
the frequency of favorable dominant genes. Similarly, under epistasis in
which the genetic intermediate is optimum, selection should have prevented
fixation of the more extreme homozygous combinations, particularly if a
rather large number of loci determine the genetic range for each primary
function.

The type of genetic mechanism that would most surely produce an in-
breeding decline relatively unresponsive to selection is heterozygote superior-
ity (^ > 1). Here selection would maintain gene frequency near some inter-
mediate equilibrium value, rather than move it toward fixation of any one
allele (qa smaller). Linear regression of genotype on phenotype (heritability)
would be lower than for lesser degrees of dominance, making selection rela-
tively ineffective. Inbreeding depression for dominance, which is propor-
tional to 2 5'a(1 — Qa) kf, would increase with k, particularly since qa would
be smaller and Qa (1 — q) larger than under partial or complete dominance.

"Controlled" Selection Experiments

Results have been published from two ''controlled" experiments on selec-
tion with minimum inbreeding in swine. In both the Illinois study of growth
rate (Krider et al., 1946) and the Alabama study of feed efficiency (Dickerson
and Grimes, 1947), the high and low selection lines separated appreciably
and significantly. However, it is difficult to judge from the time trends
whether the difference came partly from improvement in the high line or
almost entirely from decline in the low line. Taken at face value, the time
trends indicate that the separation was due to decline in growth rate of the
low line in the Illinois experiment, but that efficiency increased in both lines
in the Alabama study.

In these experiments, the low line involved a reversal in the usual direction
of selection. This amounted to assigning new selective values to genes affect-
ing growth and feed utilization, and hence selection might be expected to be
unusually effective for the first few generations in moving toward some new
equilibrium. In both experiments, selection was most effective in the first
generation.

In Goodale's (1938) and in MacArthur's (1949) selection for size in mice,
there is no question that a steady increase in size was produced. However,
these experiments with adult size in mice are not directly analogous to those

INBRED LINES FOR HETEROSIS TESTS? 335

with prolificacy, viability, and growth rate in swine, for several reasons. First,
the history of selection prior to the beginning of the experiment presumably
had not been consistently positive for adult size in mice, as it was for pro-
lificacy, viability, and rate of growth in swine. Second, selection for increased
size of the organism may be quite different from selection for a further in-
crease in efficiency within the same adult body size. Adult size is generally
highly heritable but not consistently selected for in either direction in farm
animals. The steady decline in effectiveness of selection without reduction in
variability for size in MacArthur's study suggests approach to an equilibrium
similar to that postulated for total performance in swine.

Heritability Estimates

Heritability, the portion of observed variance linearly associated with
genotype, ranges from about 10 to 50 per cent for individual characters of
economic importance. But heritability is found to be lower for the highly
important characters such as prolificacy and viability, for which selection
has been appreciable and always in one direction, than for traits such as
carcass composition or external dimensions, for which selection has been mild
or in opposite directions in different portions of the breed or during different
periods of time. Heterozygote superiority is more likely to be important for
genetic variability in the highly important characters, since selection would
have had greater opportunity to fix those genes whose homozygotes were
equal or superior to alternative genotypes at the same locus, leaving at
intermediate frequencies (larger qA[i — Qa]) genes exhibiting heterozygote
advantage.

Ineffectiveness of selection for heritable traits suggests that degree of
dominance may be higher and heritability lower for total performance than
for its individual components. This has been shown for grain yield and its
components in corn by Robinson el al. (1949) and by Leng el al. (1949). In
swine, Cummings el al. (1947) found heritabilities of 22 per cent for size
of litter at birth, 40 per cent for survival from birth to weaning, but only 7
per cent for total litter weight at weaning. Heritability of total weaning
weight jumped from 7 to 59 per cent when effects of size of litter at birth
and of survival were held constant. These results could have arisen from
negative genetic-physiological or from high positive environmental correla-
tions, or both, between numbers per litter and weight per pig at weaning.

Positive estimates of heritability for economic characters may be obtained,
even though selection is ineffective due to heterozygote advantage. If ^ > 1
and rates of reproduction were proportional to phenotypic levels, equilibrium
frequency for the more favorable allele would be 9.4 = (1 + k)/2k. At this
point, the linear regression of genotype on phenotype in an unselected popu-
lation would be zero, and all intra-allelic genetic variability would be due to

336

GORDON E. DICKERSON

dominance deviations (Fig. 21.2). Here both paternal |-sib correlation and
regression of progeny on parent would yield zero estimates of heritability, if
only dominance were involved.

Equilibrium gene frequency actually will be determined by degree of
dominance expressed in terms of relative selective values or reproductive
rates {k') rather than in terms of relative performance levels {k) of the sev-
eral genotypes. Conceivably, k' could be either larger or smaller than k. If
culling is mild, difference in reproduction rates will be smaller between Aa
and A A and larger between A A and aa than if proportional to phenotypic
levels, and effective k' will be smaller and equilibirium qA larger. Conversely
if phenotypic selection is intense, differences in reproduction rates between

V^ = 2q (l-q)[(l+k)^ - 2kq (2+k-kq)] d^

2 ,2

V^= 2q(l-q)[uk (l-2q)] ^ d

AA-aa

0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
FREQUENCY OF MORE FAVORABLE ALLELE (q^)

Fig. 21.2 — Total variance in phenotype (V„) and portion linearly associated with genotype
(Vq) in a random breeding population for a single chromosomal unit and heterozygote ad-
vantage of jfe = 2, at varying frequencies for the more favorable of two alleles.

INBRED LINES FOR HETEROSIS TESTS? 337

AasLndAA will be magnified and those between. 1.1 and aa minimized, mak-
ing k' larger than k and equilibrium (ja smaller than (1 + k)/2k. The larger
the number of genes controlling genetic variation in the basis of selection,
the less difference intensity of culling will introduce between k' and k.

Estimates of heritability from regression of offspring on parent will in-
crease positively as equilibrium q becomes larger than (1 + k)/2k, and as-
sume larger negative value as q becomes smaller than (1 + k)/2k. Larger
positive heritability estimates based on paternal |-sib correlation will be
obtained as q becomes either larger or smaller than (1 + k)/2k, since this
method estimates fraction of the phenotypic variance linearly associated
with genotype regardless of the sign of the regression of ofTs})ring on parent
(Fig. 21.2).

It seems clear that positive estimates for heritability of individual char-
acters do not rule out the possibilities (1) that heterozygote advantage ob-
tains, especially for net selective advantage or total performance; and (2)
that effectiveness of selection may be only a small fraction of that indicated
by the estimates of heritability for individual characters. More attention
needs to be given estimates of heritability for total performance indices and
their components.

Negative Genetic Correlations between Components
of Total Performance

Existence of negative genetic correlations would correspond to hetero-
zygote superiority. This is in the sense that an increase in frequency of
genes with partially or completely dominant favorable effects on one char-
acter would amount to a decrease in frequency of their alleles having partial-
ly or completely dominant favorable effects on one or more other characters.
This involves the reasonable assumptions that genes have manifold end
effects and that selection maintains at intermediate frequencies where
contribution to genetic variability is larger — only those genes having domi-
nant favorable but recessive unfavorable effects on performance. Mac-
Arthur's (1949) experiment provided ample evidence that selection for a
single character (adult size) produces many important changes in other
characters.

Direct evidence for negative genetic correlations is not plentiful. Much
data must be analyzed to estimate genetic correlation with precision, par-
ticularly when the traits correlated are of low heritability. Also, it is difficult
to avoid bias from environmental correlations. If leaner hog carcasses are
considered desirable, the genetic correlations of .3, .7, and —.7 for ratio of
fat to lean cuts with 180-day weight, daily gain, and feed requirements per
unit of weight gain found in a study of Iowa Record of Performance data
(Dickerson, 1947) need to be considered. In the same and in another study
(Dickerson and Grimes, 1947) evidence for genetic antagonism between

338 GORDON E. DICKERSON

good milking ability and rapid, economical fattening in swine is presented.

Other reasons for expecting negative genetic correlations that might be
mentioned are (1) lower heritability for total performance than for its com-
ponents, as outlined previously, and (2) approach to some physiological
maximum, where increase in one function must necessarily reduce others, as
in division of nutrient energy available between milk production and fleshing.

Negative genetic correlation, in some degree, is maintained by the
process of selection itself and would disappear if selection were relaxed.
Animals mediocre in any one respect are retained as breeders only if superior
in several other characters. Thus selection leads to a negative correlation be-
tween characters among the animals selected as parents. To a much lesser
degree, these negative relationships would appear among the progeny, where
fresh selection would magnify them again. Such negative character relation-
ships may explain in part the discrepancy between rates of improvement
"expected" and obtained, and could exist quite apart from any real heterozy-
gote advantage.

Analogy between Results with Corn and with Swine

In both corn and swine, (1) inbreeding has been slight during domesticated
history, until recently at least, (2) degree of heterozygosity exerts a major
influence on performance, (3) effectiveness of continued phenotypic selection
is questionable in stocks with a long history of selection for the same complex
of characters in which further improvement is sought.

Hull (1945) has postulated overdominance or heterozygote superiority,
with additive interaction of non-alleles, to explain corn breeding results. He
does so on the basis that (1) yields of parental, Fi, Fa, and backcross popula-
tions are linearly related to proportion of loci heterozygous (Neal, 1935), (2)
yields are usually less than one half as large for homozygous lines as for
their Fi crosses, (3) regression of Fi yield on parental inbred yield among Fi
crosses having one parent in common often is zero or negative for the higher
yield levels of the common parent. Robinson et al. (1949) have obtained esti-
mates indicating heterozygote advantage {k = 1.64) for grain yield but only
partial to complete dominance for components of yield. Crow (1948) has
shown that under complete dominance {k — 1 ) of favorable genes combining
additively, average superiority of maximum hybrid over the variety at
equilibrium gene frequency would be the product of mutation rate and
number of loci, or less than 5 per cent, whereas potential hybrid advantage
under some degree of heterozygote advantage (^ > 1) at even a small pro-
portion of loci could be many times greater, in agreement with results al-
ready obtained

The impossibility of accounting for the 15 to 25 per cent advantage of
better corn hybrids over open-pollinated varieties through complete domi-
nance of favorable genes combining additively can be demonstrated (Dicker-

\f^

PS

<

o

<

H

<
>

Q
<

Q

K

e

u

O

a

Id.

luT

o

Eh

.9 .8 .7 .6
FREQUENCY OF MORE FAVORABLE ALLELE (q)

.5

Fig. 21.3 — Potential hybrid advantage per locus {Aa — Fi) as a proportion of observable

inbreeding decline {f^ — p), for varying degrees of dominance and frequencies of the more

favorable allele in a population of homozygous (/ = 1) lines and their Fi crosses.

340 GORDON E. DICKERSON

son, 1949) from the average ratio per locus of maximum potential superiority
of hybrid over average of all possible Fi's to the observable mean advantage
of the Fi's over the inbred lines themselves (Fig. 21.3). Using (1 — q) for
frequency of the less favored allele, /for Wright's inbreeding coefficient, k for
degree of dominance, as before, and d = {A A — aa)/2:

H = Maximum heterozygote =C -\- {l -\- k) d

Fi = Mean of Fi crosses = C-{-2q[\ + k{l-q)]d

P = Mean of inbred lines = C -\-2q [\ -j- k (1 - q) {I - f) ] d

(H-Fi) {k-l)[l-2q(l-q)]-h2{l-q)^~

Hence ,

Fi-P 2kjq{\-q)

Underpartialor complete dominance, equilibrium (1 — 9) = {k —\)/2k =
0, except for reverse mutation pressure. When the parental lines are homozy-
gous (/ = 1), mean (1 — q) lies between .05 and .1 and mean inbred yield is
about 40 per cent of that for Fi crosses, the maximum increase of hybrid over
average Fi would lie between 3 and 7 per cent. There is little reason to sup-
pose that present better hybrids approach the maximum. The potential maxi-
mum increase over open-pollinated varieties increases rapidly with degree of
heterozygote advantage {k), approaching {Fi — P) or about 50 per cent in
corn yield.

The evidence thus far obtained in swine suggests that the genetic basis of
variation in net productivity is fundamentally quite similar to that in corn.
This indicates challenging possibilities for increasing productivity of swine
by utilizing potential heterosis.

Heterozygote Advantage for Single Loci and Chromosome
Segments in Other Species

Dobzhansky (1949) has shown experimentally that natural selection
favors individuals heterozygous for inversion chromosome segments in
Drosophila. He also has shown that the resulting equilibrium between fre-
quency of alternative homologous segments fluctuates with locality and sea-
son of the year, depending on relative selective advantage of alternative
"homozygous" segments. He postulates natural selection for increased co-
adaptation between alternative segments in heterozygotes within each inter-
breeding population. Demonstration of heterozygote advantage at individual
loci would be difficult, since any one locus usually would account for only a
small part of the total variability in selective value or in any complex charac-
ter. However, some cases of presumably single gene mutations exhibiting
heterozygote advantage have been reported (Jones, 1945; Gustafsson, 1946,
1947). The "yellow" gene of the agouti series in mice (Danforth, 1927;
Dickerson and Gowen, 1947) provides a classic example of manifold effects

INBRED LINES FOR HETEROSIS TESTS? 341

of genes and of heterozygote su{>eri()rity in food utilization, if not in selective
value.

It seems inevitable that manifold effects of genes and equilibrium between
frequencies of alternative alleles are common{)lace, with relative selective
values shifting with the characters given emphasis in selection at each stage
of development from conception through maturity.

EFFECTIVENESS OF METHODS OF SELECTING
FOR MAXIMUM HETEROSIS

The evidence presented provides several related assumptions concerning
the nature of genetic variability in economic characters of swine as the basis
for considering how selection for maximum heterosis can be made most effec-
tive. These are: (1) Heterozygote advantage (^ > 1) is important for total
performance when its components are characters that have had consistently
positive selective values, although lesser degrees of dominance may obtain
for individual characters. (2) Average gene frequency approaches an inter-
mediate equilibrium near q.i = (1 + k)/2k, whose value and stability depend
on the intensity, consistency, and duration of selection. (3) Performance
levels attainable by selection in outbred populations are far below the maxi-
mum heterozygote, because more than one-half of the individuals are homo-
zygous at each locus. (4) Inbreeding decline may be considered as due largely
to the reduced number of genes useful to the species that can be carried by
the more homozygous individuals, rather than to fixation of unfavorable re-
cessive genes.

Under these assumptions, any method of selecting for maximum perform-
ance will involve (1) selection for maximum proportion of heterozygous
loci in crosses of complementary strains, and (2) selection based on progeny
tests of individuals or lines in crosses. These methods are indicated only
when individual and family selection become relatively ineffective, because
the intensity of selection per unit of time is much lower for selection based on
test-cross progeny performance.

Importance of Recurrent Selection to Achieve Maximum Heterosis

Hull (1945) has emphasized the great importance of utilizing cumulative
gains from recurrent selection for heterosis in crosses, rather than relying on a
single selection among Fi crosses of a group of homozygous lines. This prin-
ciple may be illustrated by contrasting the observed distribution for number
of heterozygous loci in a population of Fi crosses among inbred lines with the
potential range (Fig. 21.4). It can be shown that the standard deviation
in proportion of heterozygous loci is:

-4

\2q{\-q){\-J)[\-2q{\-q){\-J)]

342 GORDON E. DICKERSON

within linecrosses, and

<THb

n

among linecrosses, where/ is the inbreeding of the population of lines and n
is the efifective number of segregating chromosomal units.

Range in degree of heterozygosity among all Fi crosses of a population of
lines for likely values of n {n = 100 and/= 1 in Fig. 21.4) is small rela-

.3 .4 .5 .6 .7 .8 .9 1.0

PROPORTION OF LOCI HETEROZYCX)US

Fig. 21.4 — Frequency distribution for proportion of 100 loci heterozygous when k = 2 and
initial q = .75 within lines 50 per cent inbred (.4), between Fi crosses of homozygous lines
or in a non-inbred population (B), between Fi crosses of hnes inbred only 50 per cent (C),
and in a cross between complementary strains (^i = .95, go = .15) attainable only through
recurrent selection for cross performance {D).

tive to the potential range. Hence recovery of Fi crosses much above the
average for all Fi's or for non-inbred stock cannot be expected. Inbreeding
provides a means for steadily reducing the proportion of heterozygous loci.
What is needed is recurrent selection in complementary strains to make them
steadily approach opposite extremes in gene frequency at each locus exhibit-
ing heterozygote advantage. The best Fi of a population of 100 Fi crosses
would average about 2.6 am above the mean, whereas the best 1 of 10
would average about 1.54 am above the mean and cumulative selection of
the best 1 of 10 in each of 10 recurrent cycles of selection would amount to
choosing Fi's that were 10 (1.54) am = 15.4 am above the original mean.

Homozygous Tester versus Reciprocal Selection
Hull (1945) has proposed recurrent cycles of selection in crossbred material
based on progeny test in crosses with a single homozygous line (alternatively,
with two related lines or their Fi) as a method of producing highly comple-

INBRED LINES FOR HETEROSIS TESTS? 343

mentary lines to be used in production of commercial hybrids. Comstock
et al. (1949) have compared expected effectiveness of Hull's plan with that
for reciprocal selection for cross performance between two foundation stocks
of divergent origin, avoiding inbreeding in both stocks. The}' point out that
the potential limits of improvement are the same for the two methods, except
for loci exhibiting only partial dominance {k < 1), where the use of a tester
homozygous for any of the less favorable alleles would reduce potential hy-
brid performance. Existence of important epistatic effects also would tend to
make limits lower for use of a homozygous tester.

There is no reason to expect initial cross performance to differ between
reciprocal and homozygous tester selection, other than because of the per-
formance of the inbred tester line itself (inbreeding effects on maternal en-
vironment of the litter in swine). If anything, it would be easier to find a
population differing materially in gene frequency at individual loci from a
homozygous tester than to find two similarly complementary non-inbred
populations.

Relative rates of improvement expected from the two plans depend on (1)
selection pressure applied, and on (2) size of regression of gene frequency in
the material under selection on performance of test-cross progeny. Hull's
homozygous tester plan limits selection to only one of the parental stocks.
Hence selection applied will be only | as great as in reciprocal selection.
However, as long as frequencies of the more favorable alleles {q) are any-
where near their expected equilibrium of (1 + k)/2k, progress toward com-
plementary gene frequencies (toward maximum proportion of heterozygous
loci in the cross) per cycle of selection will be far less for reciprocal than for
homozygous tester selection.

Comstock (1949) has shown that improvement in performance of random
crosses between two segregating populations per generation of selection, at a
given locus, is:

AP.= [^qA^ + k-2kq.i) -{-Aq2{l + k-2kqi) - Ik- Lq,- Aq^] d (1)

The change in gene frequency at a given locus within each of the two selected
populations (A^i and Ag2, respectively) will be determined by (1) the in-
tensity of selection based on the test-cross progeny means {s = selection dif-
ferential in a units), (2) the correlation between qi and the mean progeny
performance (P), and (3) the size of aq^ among the tested individuals, as
follows:

Avi = s r,,,.a„ = , Cov,...^ 5^^ q,{\-q,){\ + k -2kq.^ (2)

0" P 2<T p

Hence, as Comstock indicates, improvement in cross performance from
one cycle of reciprocal selection and for any one locus is:

APr = ^[qi{\-q,)(l + k-2kq,)^-+q2i\-q2)il + k-2kq,)^ ^^^

— 2k'Aqi-\q-2

344

GORDON E. DICKERSON

However, equation (3) is not applicable for evaluating Hull's (1945) plan
of recurrent selection for cross performance with a homozygous tester line.
Here, cross performance will improve as qi -^ 0 for loci at which the more
favorable homozygote (A A) is fixed in the tester, and as Qj -^ 1 for loci at
which the tester is homozygous for the less favorable allele (aa). li qr repre-

TABLE 21.2

MEANS, VARIANCES, AND COVARIANCES FOR GENOT\TES OF

SELECTED POPULATION AND PHENOTYPES OF TEST-CROSS

PROGENIES FROM HOMOZYGOUS TESTER

Selected Population

Mean Phenotypes of Progenies
FROM Homozygous Tester

Genotypes (qs)

AA at gj. of
Loci (Gi)

aa at (1— gj-) of
Loci (Gy)

AA

Means

Dev.

Freq.

1.0
l-qs

2d

i\-q,)(l-k)d

qhr

(l-f^M
(l-9,)(l + ^)rf

q^i^-qr)

Aa

Means

Dev.

Freq.

.5

.5-9.
29.(1-9.)

{?>^k)d

{\-qi){\-k)d
2q^{\-qi)qT

{^^k)d

{\-q,){\^k)d
29,(1 -9,) (l-9r)

aa

Means

Dev.

Freq.

0

-qs

(1-9.)^

-q,{\-k)d
{l-q.)'qT

0

-q^'^^k)d
(.^-q^m-qr)

Means

9.

[l + k+qiil-k)]d

9,(1 + *M

Variances

2

qi{\-qi){\-k)H^
2

9;(1 -9,) (1 + ^)^2

2

Covariances (9,, • G)

qi{\-qi){\-k)d
2

9,(l-9y)(l-H^)^
2

sents the proportion of loci segregating in the stock under selection that are
homozygous A A in the tester, then it can be shown (Table 21.2) that aver-
age progress in cross performance per locus is:

AP,= [qT•^qA'^-k) + {\- qr) '^qJ{\ + k)]d (4)

For loci fixed .4.4 in the tester:

A^, = s- rpo^' >'G,„,-(rgi = —rr-—'q^ (I- q,)(l- k) (5)

^ (T p

Similarly for loci fixed aa in the tester:

sd

Aqj

S ' r p(; • y a ,, ' a

., ^,, . Vil-qr) 'qj{\-qj)il-\-k) (6)

Z, (7 p

INBRED LINES FOR HETEROSIS TESTS? 345

From equation (4) we can now express average progress per locus from selec-
tion for cross performance with a homozygous tester as:

M\ = [ qfq.(i-q.) ( 1 - ^)-+ ( 1 - qr^Qj ( 1 - 9.) d + ^) ^1 ^ (7)

z a j>

Rate of improvement in cross performance from reciprocal selection (equa-
tion 3) approaches zero as gene frequencies approach the equilibrium ex-
pected if rates of reproduction of individuals were directly proportional to
their phenotypes (i.e., 9 = [1 + k]/2k). Hence, progress from reciprocal selec-
tion may be expressed more usefully in terms of the deviation of gene fre-
quencies from (1 + k)/2k, as follows:

^''' = £['■('-'■> Gr - "')' + ^= ( ' - '') (Vr - *'T

4 k-'

— 2k 'Aqi' Aqod (8)

Comparisons of expected progress per generation from homozygous tester
and from reciprocal selection may be made from equations (7) and (8), re-
spectively. The comparison may be visualized by plotting rate of improve-
ment against deviations of gene frequencies from an initial equilibrium value
of (1 + k)/2k, using qi and q-i for the two populations under reciprocal selec-
tion, and qi and qj for loci that are AA and aa, respectively, in the tester, for
homozygous tester selection.

In Figure 21.5, it is assumed that k = 2, and qi is shown approaching
0 {k -\- \):ik — 1) times as fast as q, approaches 1. Actually, qt would
move more slowly than Qj at first because {Aa — A A) = {k — \)d and
(.4 a — aa) = (1 + k)d. However, A^,- increases as qi falls from .75 toward
.5 because of the increased variance of qi and consequent increase in genetic
variance and in covariance with progeny means, and then Aqi declines as
qi moves from .5 toward 0. There is a steady decline in Aq, as qj rises from
.75 toward 1.0.

Under reciprocal selection, if q\ and q^ are near an equilibrium of (1 + k)/2k
at the outset, initial progress will be slight compared with that from homozy-
gous tester selection and will not equal APh until qi and q^ differ, in opposite
directions from (1 -\- k)/2k, by an average of about .50. Only during the late
generations of selection will reciprocal selection surpass homozygous tester
selection in effectiveness.

Another possible disadvantage of reciprocal selection is that gene fre-
quencies at most loci for which k> \ may be somewhat below (1 -\-k)/2k.
This will occur if the advantage oi Aa over .4.4 and aa individuals in rate of
reproduction is made greater by intensive individual or family selection than
it would be if reproductive rates were directly proportional to phenotypic
levels of performance. This would amount to increasing the effective degree

346

GORDON E. DICKERSON

of heterozygote advantage from k to k' , and hence making actual equilibrium
q nearer to .5 (i.e., ^ = [1 + k']/2k').

If actual equilibrium frequencies for the more favorable allele are generally
below (1 + k)/2k in both populations, reciprocal selection will tend to raise
both qi and 92 toward (1 + k)/2k, but at an ever decreasing rate, until q
chances to go beyond (1 + k)/2k in one of the populations. However, q\ and
92 are unlikely to be equal, even when both are smaller than (1 + k)/2k. If
qi > q-i, then qi will be closer than qi to (1 + k)/2k and will move faster in
that direction (A^i > d^q-y). Consequently gi will become larger than (1 + k)/
2k and direction of Aq2 will be reversed without reducing A^i to zero. Only

Qj and qj = .75
q. and q, = .75

Fig. 21.5 — Mean performance of crosses (upper) and rates of progress per cycle (lower) for

homozygous tester (solid lines) and for reciprocal (broken lines) selection when k = 2 and

q = .75, as qi and ^i approach 0 and qj and ^2 approach 1.

INBRED LINES FOR HETEROSIS TESTS? 347

then can the slow-starting reciprocal selection begin moving qi and 52 toward
opposite extremes.

Use of Partially Inbred Tester Lines

In large animals, or even in poultry, discussion of selection utilizing homo-
zygous tester lines is still largely academic. Few very highly inbred and
usable lines of swine and chickens exist. However, there are many partially
inbred lines of swine and poultry whose average cross performance has been
or is now being tested. These partially inbred lines should be extremely use-
ful in overcoming the initial disadvantage of reciprocal selection, because in-
breeding will have pushed frequencies of individual genes in these lines much
further away from equilibrium than in non-inbred stocks. Since effectiveness
of reciprocal selection (equation 8) increases with

even a moderately inbred line used as one of two populations under reciprocal
selection would materially increase initial progress per cycle from Aqi.

Of course, further selection within the inbred line itself on the basis of
cross performance would be relatively ineffective (A92 small) until the selec-
tion on cross performance has had time to shift qi at individual loci in the
non-inbred population away from (1 + k)/2k in the opposite direction from qi.
It might be wise to ignore cross performance in selecting replacements
within the inbred line for a number of cycles to allow time for qi to make this
shift at loci where initial ^'i and 92 chance to deviate from (1 + k)/2k in the
same direction. Beyond this point, progress from reciprocal selection between
the partially inbred and the non-inbred populations should approach and
finally exceed that from selection for cross performance with a homozygous
tester.

In selecting a partially inbred line for use in reciprocal selection, one in-
stinctively would choose a line known to be superior in its average cross per-
formance and in its usability as an inbred strain. This seems desirable to
assure that the line carries at high frequencies any genes whose favorable
effects on total performance are incompletely dominant. In addition, it
would be helpful to try a number of different partially inbred lines in crosses
with a given non-inbred stock, choosing finally for reciprocal selection the
line showing best initial cross performance. Diversity of origin and previous
crossing data would of course aid in selecting the lines more likely to be
initially complementary to a given non-inbred stock.

Presumably, both initial cross performance and rate of progress from
reciprocal selection are likely to be greater if the two populations are of dif-
ferent breeds. However, Dobzhansky's (1949) finding of greater heterozygote
advantage from alternative homologous chromosome segments within a
single population than in crosses between non-interbreeding populations of

348 GORDON E. DICKERSON

Drosophila suggests the need for further investigation of the importance of
diversity of origin for attainment of maximum heterosis in crosses.

Use of an Fi Cross as the Tester

Hull (1945) also has suggested selection to complement the Fi cross of two
homozygous lines as a means for developing new lines to replace the poorer
ones presently used in successful corn hybrids. Here, expected rate of im-
provement in performance of the 3-line cross (AP/) would be a composite
of that expected from selection for cross performance with a homozygous
tester, and with a non-inbred strain in which gene frequency is ^ at each
locus:

AP/= [q\'^qd\-k)-\-{\-qT)''^qM + k) + 2qA\-qT)'^qf\d (9)

where qr is average proportion of loci homozygous for the more favorable
allele in the lines represented in the Fi tester; qi, qj, and q/ are average fre-
quencies of the more favorable alleles at loci that are A A, aa, and Aa,
respectively, in the Fi tester. The Fi tester is A A at qr, aa at (1 — 5^)^ and
Aa at 2^1.(1 — qr) of the loci. Hence,

Aqi = ^ qrqii^ - 9,)(1 -k) , A9, = :r— (1 - 9r)9;(l - qj)(l-\-k)

and

sd

A?/ = 7i V2gr(l - 9r) -9/(1 - (//) .

Substituting in equation (9), we obtain as estimated progress per cycle:

Ai'/= \q%qS\-q){\-ky-^{\-q^yq^{\-q){\-^ky~

sd^- (10)
+ [2q^{\-q^)]^/'^-qf{\-q;)]j—

^ (J p

Apparently one might expect that selection to complement an Fi tester (of 2
homozygous lines) would be about one-half as effective as selection to fit a
single homozygous tester.

In selection for complementary strains in livestock, the Fi tester may be a
cross of two partially inbred lines, M and X. Selection of a population, L, to
complement M-N would tend to improve the L{M'N) cross at a rate inter-
mediate between one-half that for reciprocal selection (APr)/2 and that for
use of an Fi cross of homozygous lines as the tester (AP/), depending on the
degree of inbreeding in lines M and N .

If M and .Y were being selected to complement each other, gene frequency
in the {M'N) linecross tester would tend to be lowered from equilibrium
(1 -f k)/2k toward ^ as the limit. Consequently, rate of improvement in the
L{M-N) cross from selecting L to fit Af •A'^ should approach that expected
from selecting in population L to complement a non-inbred tester in which

INBRED LINES FOR HETEROSIS TESTS? 349

^ = .5 at each segregating locus. Progress per cycle when q^ = .5 should
approach

qiii - q\) '^ — .

lap

Since qi would be increased above initial equilibrium of (1 + k)/2k, maxi-
mum progress per cycle should be

{\-]-k){k-\) sd-'

W lap'

and rate of progress would decline as ^i became larger than (1 -(- k)/2k. The
maximum rate of progress, then, for selecting population L to complement
the cross of two highly complementary strains M and iV, is expected to be
little more than one-half that for selection to complement a homozygous
tester.

Other Considerations

Under heterozygote advantage and selection toward complimentary
strains by either the reciprocal or the homozogous tester method, the strains
themselves may be expected to decline in performance for characters that are
depressed by inbreeding. The less favorable allele would tend to become
fixed at about half of the loci segregating in the foundation stocks. The effec-
tiveness of this sort of selection in moving gene frequencies toward opposite
homozygous extremes in the complementary strains would be greater for
those traits in which heterozygote advantage (^ > 1), and hence inbreeding
depression, is larger. That portion of the inbreeding depression arising from
loci at which there is no heterozygote advantage {k < 1) would not be pro-
duced by selection for cross performance without inbreeding, because selec-
tion would favor the dominant allele in both strains. Therefore, any serious
decline in performance of the strains themselves, while under selection for
cross performance, is indicative of heterozygote advantage and should be
accompanied by compensatory improvement in performance of the cross.

In order to develop complementary strains whose own performance would
make them usable in commercial production of crosses, some compromise
may be necessary between selection based on test-cross and on individual
performance. There is much opportunity for selection in choosing young
breeders, especially males, to be tested in the strain-cross. Individual selec-
tion for characters little affected by inbreeding would be least apt to impair
the effectiveness of the complementary selection. Some selection for indi-
vidual performance characters important for both the strains and their
cross may become necessary to prevent fixation of rare genes with major
detrimental effects in the homozygote, but advantageous in the heterozygote.
Selection for fertility and maternal influences (e.g, hatchability, prolificacy,
or suckling ability) in test-cross matings should help maintain usable strains.

350 GORDON E. DICKERSON

SUMMARY
Genetic Variability in Economic Characters of Swine

1. Inbreeding and crossbreeding effects indicate that degree of hetero-
zygosity exerts a major influence on the important performance characters,
and that a high degree either of dominance or of epistacy due to deviations
from an optimum genetic intermediate, or both, characterizes genetic vari-
ability in performance.

2. Relative ineffectiveness of selection within mildly inbred strains makes
ordinary dominance or epistasis doubtful as an explanation of inbreeding de-
cline, and suggests heterozygote advantage for net desirability in prolificacy,
suckling ability, viability, and growth rate.

3. "Controlled" selection experiments with swine show that high and low
lines for growth rate or feed utilization can be separated, but indicate little
improvement of high line over foundation stock, particularly for net per-
formance in all characters.

4. Lower heritabilities and larger inbreeding declines for characters long
and intensely selected in one direction, compared with those selected toward
an intermediate or in varying directions, indicate a higher degree of domi-
nance for the former.

5. Some sort of negative relationship between components of total per-
formance is indicated by lower heritability for total performance than for its
component characters and by direct estimates of correlation. This would
correspond to heterozygote superiority, in that increased frequency of genes
with dominant favorable effects on one character would constitute decreased
frequency of their alleles having dominant favorable effects on other char-
acters.

6. The genetic basis of performance appears to be similar in corn and in
swine, as indicated by natural degree of inbreeding, extent of inbreeding
decline in performance, and the effectiveness of phenotypic selection. Ordi-
nary dominance is inadequate to account for heterosis already achieved in
corn, and by analogy, is unlikely to be adequate in swine.

7. Examples of manifold effects and heterozygote advantage for specific
chromosome segments or loci support their inferred importance for quantita-
tive economic characters.

Methods of Selecting for Maximum Heterosis

1. Intensity of selection per unit of time is lower when based on progeny
performance in test-crosses than when based on individual and family per-
formance. Hence, methods of selecting for maximum cross performance be-
tween complementary strains are indicated only when individual and family
selection have become relatively ineffective, and when there is evidence for

INBRED LINES FOR HETEROSIS TESTS? 351

important heterozygote advantage with attendant intermediate equilibrium
gene frequencies.

2. Cumulative gains from recurrent selection pressure are necessary to
obtain efficiently crosses heterozygous for anywhere near the potential maxi-
mum proportion of loci, since distribution of Fi crosses within any generation
is narrow relative to the potential range when numbers of loci are large.

3. Expected effectiveness of reciprocal recurrent selection between two
populations and recurrent selection for cross performance with a homozygous
tester may be compared as follows:

a. They are alike in potential limits of cross performance for loci exhibiting
heterozygote advantage, but use of a homozygous tester would be more
likely to limit ultimate cross performance if partial dominance or special
epistatic effects were important.

b. They would be similar in initial cross performance, except that it should
be easier to deliberately select a stock differing materially from a homo-
zygous tester in gene frequency at individual loci than to select two
equally complementary non-inbred stocks.

c. As long as gene frequencies in the selected populations are anywhere near
their expected equilibria, improvement in cross performance per cycle
will be far greater for the homozygous tester than for the reciprocal selec-
tion plan. The difference between progenies from A and a gametes under
selection approaches zero as frequency of A in the non-inbreed tester
approaches an equilibrium of (Aa — aa)/(2 Aa — AA—aa), but discrimi-
nation between A and a gametes under selection is maximum when the
tester is homozygous aa or A A.

d. Rate of progress from reciprocal selection accelerates as the difference in
frequency of homologous chromosomal units in the two populations be-
comes larger, and surpasses homozygous tester selection when qi — q^
exceeds about .5.

4. Use of a partially inbred line as one of the two populations in reciprocal
selection would greatly increase progress in early cycles, since individual
gene frequencies will be further away from equilibrium in inbred strains than
in non-inbred stocks.

C. R. HENDERSON

Cornell Universify

Chapter 22

Specific and General Combining Ability

By general combining ability we mean the average merit with respect to
some trait or weighted combination of traits of an indefinitely large number
of progeny of an individual or line when mated with a random sample from
some specified population. The merit of the progeny is measured in some
specified set of environmental circumstances. If maternal effects are present,
we must specify that the tested individuals are males. If the tested individu-
als are females, the merit of the progeny is a function of both general com-
bining ability and maternal ability.

General combining ability has no meaning unless its value is considered in
relationship to at least one other individual or line and unless the tester
population and the environment are specified. For example, suppose two
dairy bulls used concurrently in an artificial breeding ring each have 500
tested daughters, and that it can be assumed that the cows to which the two
bulls were mated were a random sample of cows from herds using artificial
breeding. Suppose that the mean of the butterfat records of the daughters
of the first sire is 410 pounds and of the second sire is 400 pounds. Five hun-
dred tested daughters are sufficient to reduce the sampling variance of the
progeny mean to a negligible amount. Consequently the general combining
ability of the first sire is 410 — 400 = 10 pounds better than that of the sec-
ond in this particular population and in this set of environmental circum-
stances. The general combining abilities of the two sires might differ by more
or by less than 10 pounds if they were used in some other region where both
the genotypes of the cows to which they were mated as well as the environ-
ment could be quite different from those of the test.

SPECIFIC COMBINING ABILITY

We shall define specific combining ability as the deviation of the average
of an indefinitely large number of progeny of two individuals or lines from

352

SPECIFIC AND GENERAL COMBINING ABILITY 353

the values which would be expected on the basis of the known general com-
bining abilities of these two lines or individuals and the maternal ability
of the female parent. As Lush (1948) has pointed out, apparent specific
effects, or what animal breeders usually call nicking, also can be a conse-
quence of Mendelian sampling, of inaccurate estimates of the additive genetic
values of the two parents, and of environments affecting the progeny which
are different from the average environments in which the general combining
abilities and the maternal abilities were estimated.

Genetically, specific combining ability is a consequence of intra-allelic gene
interaction (dominance) and inter-allelic gene interaction (epistasis). We shall
assume in this paper that we can estimate only the joint effect of dominance
and epistasis. As an illustration of specific combining ability let us suppose
that we know that the general combining ability with respect to weight in
swine line A is +10 pounds at 154 days, and that the general combining
ability plus maternal ability of line B is -|-5 pounds at 154 days. Then if an
indefinitely large number of progeny of the cross A X B has a mean of +7
pounds, the specific effect for this cross is 7 — 10 — 5 = —8.

SELECTION FOR GENERAL AND SPECIFIC
COMBINING ABILITY

Under some circumstances selection would be largely for general com-
bining ability, and in other circumstances for a combination of general and
specific combining ability. For example, those selecting sires for use in a
large artificial breeding ring are interested primarily in obtaining sires with
the highest general combining ability with respect to the population of cows
and environments in which the bulls are to be used. On the other hand, those
wishing to employ crosses among inbred lines for commercial use select for a
combination of general, maternal, and specific effects.

Now let us consider some of the problems involved in selecting for general
and specific combining ability. There are reasonably good solutions to some
of these problems, but almost none for others. Some of the questions which
are involved are :

1. Given a particular set of records how can one best estimate the general
combining abilities of individuals, families, or lines, and how can one best
estimate the value of the progeny of a specific cross between families or in-
bred lines?

2. What proportion of the breeder's resources should be put into a testing
program? For example, if he is dealing with inbred lines, what proportion of
his resources should be employed in the making of lines and what proportion
in testing them for general and specific combining ability?

3. Having decided on the size of the testing program, what kind of tests
should be made? For example, should lines be tested in topcrosses or in line
crosses or in some combination of these two procedures? Also what use should

354 C. R. HENDERSON

be made of a sequential type of testing in which some lines are discarded on
the basis of a very preliminary and inaccurate test?

4. What relative emphasis in selection should be placed on general as
compared to specific combining ability?

5. How much inbreeding should be done in the making of lines? How fast
should the lines be made?

Obviously a complete discussion of all these problems and their possible
solutions in the time at our disposal is impossible. Consequently we shall
discuss primarily the problem of estimating general combining abilities of
lines and individuals and of estimating the values of specific crosses among
lines, given a particular set of records. In addition, since estimates of the
variances play an important role in these selection methods, we shall discuss
briefly the problem of estimating variance components from the results of
line-cross tests.

So far as estimation of general combining abilities of individuals is con-
cerned, the methods to be presented here are essentially those of the selection
index. It will be shown that no assumption of normality of distributions is
required; that joint estimates of general combining abilities and certain
parameters such as the population means, the yearly effect, the age and in-
breeding effect, can be obtained; and that certain short-cut computational
procedures are sometimes distinctly advantageous. An application of the
principles of the selection index to estimation of general combining abilities
of lines or families also will be presented. Finally it will be shown that appli-
cation of the selection index need not be restricted as it has been to selection
for additive effects, but can be applied equally well to joint selection for
specific effects and general combining ability. The selection index approach
to appraising crosses can, under some circumstances, be much more efficient
than selection based on the mean of the progeny of a particular cross.

ESTIMATION PROBLEMS IN SELECTION
Before turning to selection for general and specific combining abilities let
us consider the type of estimation problem which is involved and some gen-
eral solutions to it. Later the manner in which the solutions can be applied
to our present problem will be discussed. Our estimation problem can be
stated in this way. We have a sample of N observations, yi, y2, • • • , ^a^,
from which we wish to estimate ^i, 62, . . . , dg. The y's are assumed to have
a multivariate distribution (precisely what distribution need not be specified
for the present) with means, biXu-\- ^2-V2i + • • • + bpXpi, and variance-
co variance matrix,

Tiie b's are fixed parameters such as the population mean and the regres-
sion of y on age of the dam, and x is an observable parameter, the first sub-
script denoting with which b it is associated, the second subscript with

SPECIFIC AND GENERAL COMBINING ABILITY 355

which sample observation. As an illustration, .Vi might be associated with bi,
the population mean. Then Xn would have the value 1 in each observation;
Xii might denote the inbreeding coefficient of the dam.

Now comes the really crucial part of the model. 'J'he ^'s are regarded as
having some multivariate distribution with means zero and variance-co-
variance matrix,

Also the d's and ^''s are regarded as having a joint distribution with covari-
ances aekyi- The way in which this problem differs from the ordinary estima-
tion problem in statistics is that here we wish to estimate the values of indi-
vidual ^'s which are regarded as a sample from some specified population.

Selection for Additive Effects in the Normal Distribution
What is the ''best" way to estimate the 0's? Suppose that they represent
additive genetic values of individuals and that any linear function of the y's
is normally distributed. Lush (1948) has shown that, subject to the normality
assumptions, improvement in additive genetic merit of a population through
selection by truncation of the estimates (indexes) of additive genetic values
is maximized by choosing that index which has maximum correlation with
additive genetic value. This principle has been used in the index method of
selection by Fairfield Smith (1936), Hazel (1943), and others. These workers
have shown that the index can be found in a straightforward manner pro-
vided certain variances and covariances and all of the 6's, the fixed elements
of the model, are known.

The values of Kei which maximize ree where 6 — KeiWi + . . . + KbnWn
are the solution to the set of simultaneous equations (1). The w's are the
^''s corrected for the fixed elements of the model such as the population mean
(not the sample mean). Thus Wi = yi — biXn — ■ ■ . —b,,Xpi.

ei y^y-i ' B2 y.^ ' ' e^ y.,y\ J/.."

(1)

K„,(T + K„„a -|- . . . + K„..(T-^ = o-„ „

d\ y^yx ^' y^'Ux *-^ l^X yN''

Selection when Form of Distribution is Unspecified
and b's Are Unknown
Maximization of ree is a satisfactory solution to the problem of selection
for additive genetic values under the normality assumption and the as-
sumption of known b's,. Is a comparable solution available when nothing is
known of the distribution or of the b's? So far as I am aware there is not.

356 C. R. HENDERSON

Consequently let us consider some other criterion of a "best" index. We shall
use as our criterion of "best" that index from the class of linear functions of
the sample which is unbiased (coefBcients of all b^s, = 0 in Ed) and for which
E{d— Oy is a minimum. E denotes expected value. Consequently E{d — dY
denotes the average in repeated sampling of the squared deviations of the
index of 6 about the true value of d. When the i's are unknown, the same
criterion of best is applied to them, that is, minimum E{b — bY for unbiased
estimates {Eb = b) which are linear functions of the sample. It turns out
that minimization of E{d — Oy and maximization of ree lead to identical in-
dexes. Hence the assumption of normality is not essential to construction
of selection indexes as now used.

It must be obvious that the selection index method just described is very
laborious when a number of different 6 need to be estimated, for the solution
to a set of simultaneous equations is required for each 6. In practice this diffi-
culty is avoided to a certain extent by choosing arbitrarily only a few sources
of information to be employed in selection. This is not a wholly satisfactory
solution, for in most cases if the number of different indexes is not to be en-
tirely too large, information must be rejected which could add at least a
little to the accuracy of the index.

By means of a simple modification it becomes necessary to solve only one
set of equations no matter how many B are estimated from a particular set
of data, and precisely the same index as in the conventional method is ob-
tained. Using the same notation as before, the index for 6 is now

where the Cs are the solution to a set of equations identical to set (1) except
that the right members are W\, . . . , Wn rather than a^y^, . . . , agy^.. Con-
sequently once the C's are computed, any number of 0's can be estimated
simply by taking the appropriate linear function of the C's.

More tedious computations result if the b's are not known. One solution
is of the following general form. In order that each d be unbiased it is neces-
sary that the K's have these restrictions imposed:

A'l.Vn + A'2.V]2 + • • • + K^Xix = 0

A'i.r2i + A'2.Y22 + . . . + A.vA'l.V = 0

(2)

A'l-Vp, -f A'2-Vp2 + • • • + A.Y.TpTv = 0

Subject to these restrictions the values of the A's which minimize E{d — dY
can then be found.

If we wish to obtain estimates of the i's which are unbiased and have
minimum E(J) — bY, we impose the restrictions of equations (2) except that

SPECIFIC AND GENERAL COMBINING ABILITY 357

the right member of the equation pertaining to the particular b to be esti-
mated is 1 rather than 0.

An easier solution to the problem of unknown 6's often can be obtained
by regarding the model as,

where the d are independently distributed with mean zero and variance al^
and the z's are observable parameters. For example, ^i might rei)resent the
general combining ability of inbred line A, di the general combining ability
of line B, and di a specific effect peculiar to the cross A X B. The observable
parameters z would have the following values: Zi = 1 when line A is one
of the parents, = 0 otherwise; Z2 = 1 when line B is one of the parents, = 0
otherwise; and Zs = 1 when ji is an observation on the cross A X B ox
B X .1, = 0 otherwise. Now the joint estimates of 6's and 0's are the joint
solution to the subsets of equations (3), (4), and (5).

<^i< +C2V. +• • • +C,^_,^. = y,- b,x^, - ... - 6^x,,

'. . (3)

C,a +C,(7 +...+C,.(72 =y -hx -. . .-h X

1 i/,y_Y ' 2 y„yx A y^ ' -V 1 UV p p.-v

1 1 j/j9[ A y^v'i

(4)

"?

where

b,Sx\ + . . . + bpSx.Xj, + ^i^.vi si + • . . + ^,5.vi2, = 5.viy

'. . (5)

Si5:*;iXp + . . . + 6p5.Vp + diSXp zi + . . . + ^,>SXp z, = Sxj,y ,

S.v;=2:#. 5.v,.v.= 2:^?^, etc.

These equations can be solved by the following steps. First solve for the C's
in equations (3). The results will be in terms of the sample observations and
the 6's. Second, substitute values of these C's in equations (4) to obtain ^'s
in terms of the sample and the b's. Third, substitute these values of the ^'s
in equations (5) and solve for the 6's. Fourth, substitute the computed values
of the S's in (4) and solve for the ^'s.

358 C. R. HENDERSON

An alternative computational procedure which is less laborious when the O's
are few in number, and in particular when the 5's are uncorrelated, involves
joint estimation of the i's and 9's by solution of equations (5) to which are
added equations (6).

(6)

b,Sx,z,-\-. . .+ b,SXj,z,-\-di{Sz,z,+ a'')-\-. . .-^d,(Szl+a"')=Sz,y
where

,-'MI = IIV,l

an

d

SxiZi = > — -^, etc

These equations are simply least squares equations (the d's are regarded
as fixed rather than having a distribution) modified by adding a"-' to certain
coefficients.

SELECTION BY MAXIMUM LIKELIHOOD ESTIMATES

Now let us assume that the ^'s have the multivariate normal distribution
and that the errors are normally and independently distributed. What are
the maximum likelihood estimates of the d's and 6's? It just so happens that
the estimates which are unbiased and which have minimum E(d — 6)- and
E(b — b)- for the class of linear functions of the sample are also the maxi-
mum likelihood estimates. Consequently the estimation procedure we have
described can be seen to have the following desirable properties: unbiased-
ness, maximum relative efficiency of all linear functions of the sample, maxi-
mization of genetic progress through selection by truncation when the dis-
tributions are normal, properties of maximum likelihood estimates when the
distributions are normal, and equations of estimation which can be set up in
a routine manner.

Unknown Variances and Covariances

An important problem in selection remains unsolved and perhaps there
is no practical solution to it. What should be done if the variances and co-
variances are unknown? If our sample is so large that estimates of the vari-
ances and covariances can be obtained from it with negligible errors, we can
use these estimates as the true values. Similarly we may be able to utilize
estimates obtained in previous experiments. But if there are no data available
other than a small sample, the only reasonable advice would seem to be to
estimate the variances from the sample, perhaps modifying these estimates

SPECIFIC AND GENERAL COMBINING ABILITY 359

somewliat if they appear totally unreasonable. At any rate the estimation
procedure serves to point out what additional information is needed if an
intelligent job of selection is to be accomplished.

SELECTION FOR ADDITIVE GENETIC VALUES IN INDIVIDUALS
As our first application of the methods described above, consider the esti-
mation of additive genetic values of individuals with respect to a single trait
(the single trait might be net merit) from a set of records all made in the
same herd or flock. It will be assumed for the present that the population
mean is known and that records can be corrected satisfactorily for all non-
random environmental factors. For example, the records might represent
all of the 305 day, mature equivalent butterfat records made in a herd during
the past ten years. It is desired on the basis of these records to decide which
cows should be culled, which heifers should be selected for replacements, and
which bull calves should be grown out for possible use as herd sires.

In the usual approach to this selection problem by use of the selection
index, one would decide what particular subset of the records would con-
tribute most to the estimate of the value of each animal under consideration
and would then construct separate indexes. The method to be presented here
employs all available records in estimating the value of each animal. That is,
no prior decision is made concerning which records to use to construct the
index for each animal, but instead all available ones are used.

The first step in the procedure is the computation of what Emik and Ter-
rill (1949) have called a numerator relationship chart and Lush (1948) has
called genie variances and covariances for all animals whose records are to be
used in the index or whose breeding values are to be estimated. In terms of
Wright's (1922) coefficients of relationship and inbreeding, the genie variance
of the iih animal is 1 -|- Fi, where Fi is the inbreeding coefficient of the ith
animal, and the genie co variance between the ith. andjth animal is

i?oV(l+F,)(l-|-Fy),

where i?,, is the coefficient of relationship between the two animals. The nu-
merator relationship or genie eovariance, which we shall denote by a.y, is the
numerator of the fraction representing relationship. That is

7? - ^''

Vd+FJd-fF,)

The computation of ||a,_,|| is a routine procedure if it is done systematically as
described by Emik and Terrill and by Lush.

Next we need an estimate of heritability of the trait, and if more than one
record is available on a single animal, as would be true of butterfat produc-
tion, an estimate of repeatability. Now let yi, yz, ■ ■ • , yp be the mean of the
■Hi records of each of p animals, these records having been corrected for

(7)

360 C. R. HENDERSON

non-random environment and expressed as deviations about the population
mean. The next step is to solve the following set of equations for Ci, . . . , Cp.
In these equations h denotes heritability and r denotes repeatability.

Cianh-\-C-2 (FoJH j + • • • -hCpG^ph = 5*2

* •

Cia^pJi -\-C2a2ph + . . . +Cp [Fp -\ 'A = Vp

If all available records are to be used in the estimation procedure just de-
scribed, the number of equations to be solved for the C's is large. It might ap-
pear, in fact, that the number is too great for the method to have any value.
However, the equations are ideally suited to an iterative solution. The reason
for this is that the diagonal elements of the left members of the equations are
very large compared to the off-diagonal elements thereby making the itera-
tive solution a particularly rapid one. On the basis of our experience with a
few herds a solution to sufficient accuracy can be obtained in three or four
rounds of iteration.

Once the C's have been computed the estimate of gt, additive genetic value
of the iXh. animal, is

gi = h (Cifli, -\-C2a2i + . . . +Cpap,) .

If the iih animal had one or more records included in the computation of
the C's the estimate can be computed more easily, for

-^ ^ . _^ 1 + ("> - 1) >' - n,h

The estimate of the real producing ability of a tested animal is even more
simple to express. The estimated real producing ability is

y'i—Ci .

It should be pointed out that this estimate differs from the one presented by
Lush (1945) since his method does not utilize records on relatives.

Valuable characteristics of the method just described, in addition to its
ease of computation and its use of all available information, is that the inclu-
sion of the records of the contemporaries of the ancestors of the animals being
appraised automatically eliminates the troublesome problem of what effect
selection has had on the phenotypic and genetic variances of the selected
group of ancestors. Also changes in additive genetic variances and covari-

SPECIFIC AND GENERAL COMBINING ABILITY 361

ances effected by inbreeding are automatically taken into account. If selec-
tion is intense, the samj^le mean may considerably overestimate the i)o[)ula-
tion mean appropriate for subtraction from the records. The safest procedure
is to regard ^t as unknown and to estimate it by the procedure described
earlier (equations 3, 4, 5) . It is also of interest to note that joint estimation by
this method of such factors as environmental trends and age effects automati-
cally eliminates biases in the estimates resulting from use of selected data.

SELECTION FOR GENERAL COMBINING ABILITY
IN TOPCROSS TESTS

When it comes to estimation of the general combining abilities of inbred
lines or of the values of specific crosses, apparently no application has been
made of the selection index method. This failure may have been due to diffi-
culty in obtaining the estimates of the needed variances and covariances,
failure to see that the method was applicable, or the opinion that since inbred
lines can be carefully tested more efficient but complex methods of appraisal
are not worth the extra computational labor. We propose to show here how
the methods can be applied to such selection problems, to indicate some situ-
ations in which it may result in considerably more efficiency in selection than
the use of the straight means of the lines or crosses as the criteria of selection,
and to present some approximate solutions which are relatively easy to
compute.

Let us consider first one of the most simple tests of lines, the topcross test.
In this test a random sample of individuals from each of several lines is mated
to a tester population, and measurements are taken on the resulting progeny.
If only one trait is considered important, the lines are usually rated according
to the means of their topcross progeny. This method of ranking is as good as
any, provided either that the same number of progeny is obtained for each
line or that the sampling errors of the line means are negligible. Seldom, at
least in large animal tests, would either of these conditions hold. Accidents
usually preclude attainment of equal numbers, and sampling errors are usual-
ly large. If sequential testing is done, numbers would always be unequal. By
sequential testing we mean here that lines are given a preliminary test, and a
certain fraction of those performing worst are discarded. Then the remaining
lines, accompanied perhaps by some new lines, are given another test, and so
on through any number of cycles desired. The lines surviving several such
tests would obviously have larger numbers of progeny than the new lines, and
it would be a very inefficient procedure to disregard the results of prior tests
on the older lines when choosing between them and the newer, less well-
tested lines.

The way in which the lines should be ranked on the basis of all information
is analogous to choosing between individuals with different numbers of rec-
ords. In the latter case both repeatability of single records and the number of

362 C. R. HENDERSON

records need to be considered; in the former case the genetic differences
among Hnes, the environmental variance, and the number of progeny. Also in
both cases consideration of the genetic covariances between individuals or
between lines increases the accuracy of the ranking.

Assuming that the population mean is known and that it and non-random
environmental factors have been subtracted from the means of the progeny
of the various lines, the estimate of gi, the general combining ability of the
iih line, is

£.=C,a- --{-... -\-C a -

°i 1 BiVi P BiVp

where yi, • ■ ■ , Jp are the corrected means for the p tested lines and the C's
are the solution to a set of equations with

as coefificients in the left members and corrected y\, . . . , y? as the right mem-
bers. Computation of o-^^j, and o-oiSi requires good estimates of

Ik II

and of al. Assuming that the corrected mean of a particular topcross is
yi = gi'\- ^i, and that the errors are independent with common variance af
we have the following variances and covariances

a- = a- -]-a-/n. cr- — a~

a-- = a (where i 7^ j) a- = a (i ^ j)

ViVj aivj -' viOj Qiuj ■'

Frequently good estimates of n and non-random environmental factors are
not available and consequently must be estimated from the topcross data.
For example, it is very likely that the environment is not the same from test
to test and must be taken into account if the data from several tests are to be
combined into a "best" index. In such cases the method of equations (5) and
(6) can be employed to distinct advantage unless

is too difficult to compute. To illustrate this method as applied to topcross
data we shall assume that y^j, the record of the/th progeny of the ith line,
can be represented by

yu = byXuj -\- hoX'iij + gi -\- e,j .

bi and 62 are examples of fixed parameters, gi is the general combining ability
of the /th line, and d, is a random error. Assuming that the gi are distributed
with means zero and known variance-covariance matrix.

SPECIFIC AND GENERAL COMBINING ABILITY 363

and that the dj are mde{)endently distributed with means zero and common
variance a;, the estimates of the b's and g's which are "best" by the criterion
used in this paper are the solution to the following equations:

i>i

(8)

^i-vi;,. + b.x.^ + ip ("p+ <T?ff^^) + ^ ^pff^cr"' = >'

Dots in the subscripts denote summation over that subscript, and a'' denotes
an element of

cr

The above procedure for appraising lines on the basis of topcrosses assumes
either that the lines are homozygous or that only one progeny is obtained
from each randomly chosen male. If these assumptions are not correct, the
procedure is moditied to take into account intra-line variances and covari-
ances and the number of progeny per male.

What are the consequences of appraising lines on the basis of the arithmet-
ic average of their respective progeny as compared to the more efficient
method just described? First, the errors are larger than necessary. Second,
selection of some small fraction of tested lines will tend to include a dispropor-
tionately large number of the less well-tested lines. The more efficient meth-
od discounts the higher averages in accordance with the number of tested
progeny and the relative magnitudes of a^ and a;.

What if the number of lines tested is large and certain lines are related?
This means that a large matrix,

11%,, II.

has to be inverted and then a large set of simultaneous equations solved.
What approximations might be employed in the interest of reducing compu-
tations? For one thing, we might ignore the covariances between the g's,
thereby reducing the inverse matrix to l/a^, in the diagonal elements and 0
in the off-diagonal elements. Also if we know /i and non-random environmen-

364 C R. HENDERSON

tal factors well enough, further simplification is possible. Let Wi be the cor-
rected mean of the progeny of zth line. Then

p = 7^

Hi e

This result is a straightforward application of the principles of the selection
index.

It must be quite apparent that efficient appraisals of the general combin-
ing abilities of lines depend on knowledge of the variances and covariances of
general combining abilities and of the variance of error. It hardly seems like-
ly that estimates of the line variances and covariances can be obtained with
accuracy comparable to estimates of additive genetic variances and covari-
ances with respect to individuals. The latter estimates are based on studies of
heritability and on the known facts of the hereditary mechanism. In the case
of inbred lines, however, the sample of different lines tested is usually so
small as to make the estimates of al less reliable than we should like. A way
around this difficulty in the case of traits for which heritability is well known
is to compute the expected variances and covariances based on knowledge of
(tI in the original population from which the lines were formed, the inbreeding
of the different lines, and the relationships between pairs of lines. It seems
likely that such estimates would be more reliable when the number of lines
is small than would estimates arising from the actual line tests. We cannot
be any more precise regarding this point until methods are developed for
placing confidence limits on estimates of variances and covariances arising
from non-orthogonal data.

SELECTION FOR GENERAL COMBINING ABILITY, MATERNAL ABILITY,
AND SPECIFIC ABILITY IN LINE CROSS TESTS

If we wish to estimate the general combining ability of lines relative to the
population from which the lines themselves can reasonably be regarded as a
random sample, line crosses give, for fixed size of testing facilities, more accu-
rate estimates than do topcrosses. The reason for this is that we obtain from
each cross estimates of the general combining abilities of two or more lines.
Also, line crosses enable one to estimate differences in maternal abilities un-
confounded with differences in general combining abilities and to appraise the
values of specific crosses. In those species for which hand mating is the cus-
tomary procedure, little more labor is required for line cross than for topcross
tests. The estimation of line and line cross characteristics from line cross data
is no different in principle from what we have already described with respect
to estimation of additive genetic values of individuals or general combining
abilities of lines. As before, we wish to obtain unbiased and most efficient esti-
mates of certain genetic values. For the sake of simplicity of presentation we

SPECIFIC AND GENERAL COMBINING ABILITY 365

shall confine ourselves to discussion of the analysis of single crosses. Applica-
tion of these principles to multij)le cross data involves no new principles.

Let us consider first what type of model might be reasonable for a single
cross. It is not too difficult to suppose that the value of a particular observa-
tion on a single cross is the sum of the general combining ability of the male
line, the general combining ability of the female line, a maternal effect coming
from the line used as the female, a specific effect due to dominance and
epistasis and peculiar to the particular cross, non-random environmental ef-
fects, and a multitude of random errors such as Mendelian sampling and the
environment peculiar to the particular progeny on whom the record is taken.
More complicated models could of course be proposed, but the one which we
have just described would seem to account for the major sources of variation
among crosses. Furthermore it is amenable to mathematical treatment. Put-
ting the above description in a mathematical model we have

Jiik = ^i-^iu'^- + boXi^jk + g, + gj + W; -f s,j + eijk ,

where yuk is the observation on the ^th progeny of a cross between the ith
line used as a male parent and thejth line as a female parent, the 6's and a;'s
are related to the mean and other non-random environmental factors as de-
scribed in the model for the topcross test, giigj) is the general combining abili-
ty of the ith(yth) line, nij is an effect in addition to the additive genetic value
which is common to all progeny of they'th line used as a female parent, 5,, is
an effect over and above the additive genetic and maternal effects and which
is common to all progeny of the cross of the ith line by thej'th line or of the
yth line by the ith line, and Cijk is a random error associated with the particu-
lar observation.

In this model the gi are regarded as having some multivariate distribution
with means zero and variance-covariance matrix,

11 '.,-., II-

The wy, Sij, and e^j are all regarded as independently distributed with means
zero and variances o-^, o",, and al, respectively. It is of course conceivable that
the variances of the m, and sa and the covariances between them vary with
the inbreeding and relationships of the lines. Also gi and ntj may be correlat-
ed. In the absence of any real knowledge concerning such covariances we
shall ignore them for our present purposes. If, however, something is known
about these covariances, the estimation procedure can be modified to take
them into account. The procedure should also be modified if the lines are not
homozygous and each parent has more than one progeny.

A single cross test can supply answers to the following questions with re-
spect to the lines tested:

1. What are the best estimates of the relative values of the tested lines

366 C. R. HENDERSON

when used as the male parent in topcrosses on the population from which
the lines are regarded as a sample?

2. What are the best estimates of the relative values of the tested lines as
female parents in crosses with males from the above population?

3. What are the best estimates of the relative values of specific single
crosses among the tested lines?

Suppose that tin progeny of the cross tth line of male by 7th line of female
are tested (nn can be zero for some crosses) . Now the easiest way to estimate
the value of the ith line as a male parent is simply to compute the mean of
all progeny of the line when used as the male parent. This simple procedure,
however, fails to take into account the distribution among lines of the mates
of males of the /th line, the covariances among the general combining abili-
ties of lines, the consequences of specific effects, the size of the error variance,
and the number of progeny tested. Furthermore, since the zth line is used also
as the female parent in certain crosses, something can be gained by employ-
ing the measurements on these progeny. Estimation by the general procedure
we have described takes into account all of these factors. Similarly the easiest
way to estimate the maternal ability of thejth line is to compute the mean
of all progeny out of females of the /th line, but the most efficient procedure
takes into account the same factors as are needed in efficient estimation of
general combining ability. Finally the easy way to appraise the value of a
particular cross is merely to find the mean of all progeny of the specific cross
(if that cross has been tested). This latter estimate is subject to large sam-
pling error since it would seldom be feasible to test many individuals of the
numerous possible crosses among even a few lines. The error of estimation can
be materially reduced by utilizing the fact that the true merit of a cross is a
function of the general combining abilities of two lines, the maternal ability
of the female line, and the specific effect peculiar to that cross and to its re-
ciprocal. The method to be described places the proper emphasis on estimates
of general and maternal abilities and on the progeny averages of the specific
cross and its reciprocal. The procedure also enables estimates to be made of
the value of a specific cross even though that particular cross has not been
tested.

The major step in these efficient estimation procedures is the setting up
and solving of a set of simultaneous equations in the 6's, ^'s, ?n's, and s's.
These equations are as follows:

^1 2 y^ 2 -VluA- + ^2 23 2 2 •^■l'J>^"2 0-i- + 2^ gi i^l'.. + -^'l. I. )

i j k i i k i ( 9 )

and similarly for the bo equation.

i<; i J k

SPECIFIC AND GENERAL COMBINING ABILITY 367

biixn.. +^-1.1.) +^2(a-2i.. +:t2.i.) +^i(«i. + ".i + ff^cr*^)

= 3'!.. +>'.!.

and similarly for the other gi equations.

^l-Vl.l. + ^2-V2.1. + gl«.l+2^ gi«.l+ Wi (;/.!+ Cr!/ffl,)+ 2 5i,M,i = J.l.

and similarly for the other ;«; equations.

^(■Vn2. +-Vi21.) +6.2(.V212. +-V22I.) + (gl+ g2) ("12 + W21) +/«1«21

+ W2H12+ ■Jl2(«12+»21 + ae/(rJ = yi2. +3'21.

and similarly for the other Sij equations.

These equations are not particularly difficult to solve, for each Sij can be
expressed as a function of ya., y;,-., 5i, 62, Qi, Qj, and mj. Utilizing this relation-
ship the equations can be reduced to a set involving none of the ,Sij. Also an
iterative solution is usually easy because of the relatively large diagonal co-
efficients. Once the estimates of gi, nij, and s^ are obtained it is a simple mat-
ter to evaluate the lines and crosses. The estimate of the value of a line as the
male parent in topcrosses is Qi, and the estimate of its average value as the
female parent is ^i + ifu. The value of a single cross is estimated simply as
9 1 + di + '^ij + •^u- It is appropriate to add the estimates in this manner be-
cause they have the desirable property of invariance.

If solution of the large set of simultaneous equations required for most ef-
ficient appraisal of lines is considered too burdensome, certain approximate
solutions can be employed. An approximation suggested by the common
practice in construction of selection indexes is the choosing of certain infor-
mation most pertinent to the particular line or cross to be appraised. For ex-
ample, the estimate of g, might be based entirely on y^.. and y.,., each cor-
rected for the b's as best can be done with the information available regarding
their values. As a further simplification it might be assumed that the gi are
uncorrelated and have common variance o-|. Similarly w, might be estimated
entirely from y,. . and y.,. . These approximate solutions are

g. =C ,a +C><7
fh ■ =C ,(j -\-C„(j ,
where the C's are the solution to

368 C. R. HENDERSON

The variances and covariances needed in this approximate solution can be
computed easily from a^, a^, a;, and a^. Approximate values of §ij can then
be obtained by substituting the approximate bi, hi, Qi, and rhj in equations (9) .

ESTIMATION OF VARIANCES OF GENERAL, MATERNAL,
AND SPECIFIC EFFECTS

As mentioned earlier, one might take as the additive genetic variance and
covariance among the lines the theoretical values based on relationships
among the lines, degree of inbreeding among the lines, and the genetic vari-
ance in the original population from which the lines came. It is necessary even
then to estimate al^, af, and a;. It is well known that methods for estimating
variance components are in a much less advanced stage than estimation of
individual fixed effects. It is seldom possible to obtain maximum likelihood
estimates. Consequently many different methods might be used, and the
relative efficiencies of alternative procedures are not known.

We shall consider as desirable criteria of estimation procedures for vari-
ance components ease of computation and unbiasedness. If the single cross
experiment is a balanced one, that is if there are the same number of observa-
tions on each of the possible crosses, it is not difficult to work out the least
squares sums of squares for various tests of hypotheses, regarding the line
and cross line characteristics as fixed. Then assuming that there are no co-
variances between the various effects and interactions, one can obtain the ex-
pectations of the least squares sum of squares under the assumption that the
effects and interactions have a distribution (Henderson, 1948). In case the
experiment is not a balanced one, it is still possible to obtain least squares
tests of hypotheses and to find expectations of the resulting sums of squares-
This, however, is ordinarily an extremely laborious procedure (Henderson,
1950).

A much easier procedure is available. It probably gives estimates with
larger sampling variance, although that is not really known, and gives almost
exactly the same results in the balanced experiments as does the least squares
procedure. This involves computing various sums of squares ignoring all cri-
teria of classification except one, taking expectations of these various sums of
squares, and solving the resulting set of simultaneous equations. The latter
procedure will now be illustrated for single cross data in which we wish to
obtain estimates of the variances pertaining to general combining ability,
maternal ability, specific effects, and error. It will be assumed that the only
fixed element in the model is /x. Now let us compute certain sums of squares
and their expectations. These are set out below.

Total: i^(X;2: Vy?,,)=>/..(M^ + 2a^ + ai + a^4-ap

SPECIFIC AND GENERAL COMBINING ABILITY 369

Sires: £ f Y) -'-') = ;/.. (m" + ^D + S — (<t"' + <t2 + a^) +5^2,

where 5 denotes number of different lines used as the male line.

n-..
'I

where d is the number of different lines used as the female line.
Crosses: e( V -^^-^^■'■•^-') = «•• (m^ + 2a^ + a'-')

C<T-

where c denotes the number of different crosses (regarding reciprocals as one
cross)

Correction Factor: -E (^-— ) = "••M"+ ^ (;/, -f ».,) Vp/M..

2

e

The above sums of squares and expectations are quite easy to compute and
once this is done all one needs to do is to subtract the correction factor and
its expectation from the other sums of squares and expectations and solve the
resulting set of four equations for a\, cr^,, a,, and af.

FURTHER RESEARCH NEEDED
If maximum progress through selection for general and specific combining
ability is to be attained, much additional research is needed. From a statisti-
cal standpoint we need to know if an index based on minimization of E{6—dY
comes close to maximizing progress through selection by truncation when the
distributions are not the multivariate normal. If such an index does not do
so, we need to know what practicable index or indexes will. Further, if
nothing is known of the variances and covariances needed in construction of
indexes or if there are available only estimates with large sampling errors, we
need to know if the index based on the assumption that the estimate is the
true value is best from the standpoint of maximizing genetic progress. Final-
ly, much more work is needed on the problem of estimating variance and co-
variance components and placing confidence limits on such estimates.

370 C. R. HENDERSON

Although there is a considerable body of literature on heritability esti-
mates, we need more accurate estimates of the heritabilities of most traits of
economic importance. Also almost nothing is now known about genetic cor-
relations between traits, about genetic-environmental interactions, and
about the magnitude of genetic differences among herds. Estimates of these
genetic parameters are essential to intelligent selection for additive genetic
values. In the case of inbred lines, little is known concerning the variances
of general and specific combining abilities. The work of Sprague and Tatum
(1942) with corn and Henderson (1949) with swine illustrates the types of
estimates which are badly needed in selecting for general and specific combin-
ing abilities from the results of line cross tests.

Finally, well designed experiments are needed to test how closely predic-
tions made from indexes or other selection procedures check with actual re-
sults.

L M. WINTERS

University of Minnesota

Chapter 23

Rotational Crossbreeding

and Heterosis

It is well for all of us, including our most eminent scientists and philosophers,
to reduce our thinking to relatively simple terms. Genetics is, after all, basi-
cally rather simple. A fertilized zygote results from the union of two germ
cells, each of which carries a haploid number of chromosomes, and a haploid
number of genes which are resident in the chromosomes. By the very nature
of the procedure, genes are paired which are alike or not alike. As the pairing
of similar genes is increased, the population approaches increased purifica-
tion. As the pairing of dissimilar genes increases, the resulting population be-
comes more heterozygous. Increased heterozygosity has been generally as-
sociated with increased vigor which is generally spoken of as hybrid vigor.

PLANNING THE MINNESOTA EXPERIMENTS

I believe the best way to discuss rotational crossbreeding is to relate briefly
how the system was developed. When I was asked in 1928 to head the re-
search in animal breeding at the University of Minnesota, I brought with me
several proposed projects. One of these was a study of crossbreeding swine.
A review of the literature of crossbreeding experiments conducted previous
to 1928 shows that for the most part they were small-scale experiments.
When the data were all put together, however, the evidence was in favor of
crossbreeding. Yet, the general sentiment at that time among the stockmen
was overwhelmingly opposed to the practice of crossbreeding. The statement
frequently heard was that crossbreeding was quite satisfactory for the pro-
duction of one crop, but all of the crossbreds must be marketed because it
was absolutely disastrous to use any of the crossbreds as breeding animals.
By 1928, it was quite evident that corn breeders were revolutionizing the
system of breeding corn, and that hybridization was to become the rule rath-

371

372 L. M. WINTERS

er than the exception in the production of commercial corn. Wright (1922)
had several years previously published what has since turned out to be a
classic: The Report of the U.S.D.A. Studies of Inbreeding and Crossbreeding
with Guinea Pigs. Why then should the situation be different in livestock
than in corn and in guinea pigs? W^as it true that livestock failed to respond
to crossbreeding, and if so, why? A likely explanation appeared to be that our
breeds of livestock were not truly comparable to inbred lines of guinea pigs
and corn because they did not possess sufficient genetic purification.

I am sorry now that I did not record in advance of this experiment the re-
sults that I expected to derive. Had I recorded them, they would have been
something like this: The crossing of the breeds of livestock will result in a
slight increase in vigor. The increase will be so slight that it is scarcely worth
while for the commercial producer, in contrast to the more simple procedure
of grading or the maintenance of a registered herd. Most of my severe critics
regarding crossbreeding will be quite surprised to read this statement. As
nearly as I can tell at this time, there were two major reasons for my belief.
The first was the accumulation of the continued absorption of a large amount
of teaching toward that end. The second was the general belief on the part
of geneticists that our breeds of farm animals had not been sufficiently puri-
fied nor separated genetically to yield hybrid vigor when crossed.

If I had recorded all of my thoughts, they would have included this reser-
vation: If the crossing of the breeds does result in increased performance suf-
ficient to make crossing worth while, then the standard advice that had been
given through the years regarding the use of crossbred females for breeding
must be erroneous. This reservation was based on the results that Wright had
previously obtained in his use of crossbred guinea pigs as dams, and from the
information already available regarding the production of hybrid corn. At
this time there was no thought regarding continuous crossbreeding by either
rotation or crisscross breeding. The objectives were merely to find out if there
was any advantage in crossing the breeds for the market production of swine.
If there was an advantage in crossing the breeds for market production of
swine, was there then any advantage in retaining these crossbreds to become
future parents?

In planning the experiment, provision was made whereby as nearly as
possible the same genetic material was put in the crosses as was produced
in the purebreds. In planning the use of crossbreds as parents, the original
plan called for the use of both crossbred females and crossbred males. My
senior officers informed me that they were willing to go along with me quite
a way in this crossbreeding study, but that when it came to the use of cross-
bred males, that was going just a bit too far and I would have to compromise.
I compromised on this point all too willingly. How I have wished, during the
last few years, that I had insisted on carrying out my original plan of
using both crossbred females and crossbred males in the experiment. But little

ROTATIONAL CROSSBREEDING AND HETEROSIS 373

did any of us realize at that time that within twenty-two years we would be
in the midst of a flourishing hybrid boar business. None of us know much
about their true merits and demerits. Nevertheless, the crossbred boars were
not included in the experiment.

Experimental Results

The experiment did proceed as planned for the production of first cross
offspring from the mating of purebred females of one breed to purebred males
of another breed. Crossbred females were then retained as breeding animals
to be mated in one case to a boar of one of the two parental breeds and in the
other case to a boar of a third breed. The results of this experiment showed
that there was a very definite advantage in the production of firstcross pigs.
There was a slightly greater advantage in the production of backcross pigs
(that is, where crossbred females were mated back to a boar belonging to one
of the parental breeds). There was a still greater advantage where the cross-
bred females were mated to a boar of another breed.

As I mentioned before, you bring together either genes that are alike or
genes that are not alike. There appears to be very little likelihood of bring-
ing about any more heterozygosity as a result of a three-breed or a four-breed
cross than there is in a two-breed cross. The advantages derived from the
backcross and from the cross to a third breed appear therefore to have been
derived from the fact that the female parents were crossbreds or in a more
hybrid state than their purebred half sisters. Why should this be the case?
The female produces the eggs, carries the fertilized eggs, and develops them
to the point where, after a period of about 114 days, they are ready for birth
and then nurses the little pigs for another 56 days. In general, the advantage
derived from the crossbred female in contrast to the purebred female is
about equal to that derived from having the progeny crossbred in contrast
with having progeny that are purebred.

The above are the general deductions that I made at the close of our cross-
breeding experiment. Now I am not so certain that this interpretation is ab-
solutely correct. The reason for my questioning is that recently I had a long
visit with one of the largest hybrid seed corn producers in this nation. He is a
man who has had many years of experience in the field. lie told me that he
had not yet seen a single cross of hybrid corn that was as useful for com-
mercial corn production as the double hybrid. He elaborated further to the
efifect that the single cross hybrid often would yield as heavily as the double
hybrid, but that under adverse environmental conditions the double hybrid
fared better. This he attributed to the fact that the double hybrid de-
veloped from four inbred lines possessed greater genetic diversity toward
adversity. This appears to be somewhat in contrast to the experimental re-
sults and interpretations of those results in some of the ])resent-day funda-
mental studies of Urosophila genetics. Undoubtedly, with time and more

374 L. M. WINTERS

experimental results, we will be in a better position to bridge this gap. My
experiences have convinced me, however, that it is a mistake not to take
seriously the observations made by competent practical men in the field of
operations. I am inclined to believe that very often these men see more, al-
though they measure less accurately, than we in the field of research, with
our eyes glued carefully to the job of measuring certain details.

In this experiment we used four standard measurements for appraisal of
the pigs' worth. These were: number of pigs born alive, number of pigs
weaned, rate of gain, and feed per unit of gain. Since then we have added a
fifth measure — appraisal of body form on the basis of judgment. When we
took the first four factors and compared the performance of the crosses with
the comparable purebreds, we obtained an advantage of the crosses over the
purebreds of 6.3 per cent for the first cross, 7.5 per cent for the backcross,
and 11.7 per cent for the three-breed cross. This was obtained by throwing
the four factors together as equal in importance.

By another method of comparison, wherein more factors were thrown into
the pool, we obtained an advantage for the first cross of 7 per cent, the back-
cross 6 per cent, and the three-breed cross of 17 per cent. If, however, we
were to take litter weight at weaning, which in one sense is comparable to
yield in corn, we would have an advantage of the first cross of approximately
25 per cent, the backcross 39 per cent, and the three-breed cross 61 per
cent. If we were to take total litter weight at the close of the experiment, the
advantages of the crosses would be still greater. In my opinion, total litter
weight as a sole measure of merit exaggerates the diliference. On the other
hand, I do not consider the method we have used entirely satisfactory. I do
not know of an entirely satisfactory measure of performance in livestock.
We in the livestock field need to do a great deal in the matter of perfecting our
methods of measurements. The important question to the practical man is
whether one procedure is better than another, rather than whether this pro-
cedure gives me exactly 20 per cent or 18 per cent increase.

ROTATIONAL CROSSBREEDING

On the basis of these results, we developed and put forward our plan of
rotational crossbreeding. Even at the time that I started to analyze the data,
I did not believe that our three-breed cross had given us any worth-while
advantage over the single cross. I mention this merely to show how strongly
entrenched the old teaching had become regarding the limitations of cross-
breeding in livestock production. The results of the experiment were, how-
ever, very definite. I calculated and recalculated, and the results were always
essentially the same- — the three-breed cross possessed distinct advantages
over the first cross and over the backcross.

Simple calculation shows that, on the average, the first cross will possess
50 per cent of the chromosomes, or more properly speaking, linkage groups

ROTATIONAL CROSSBREEDING AND HETEROSIS 375

of breed 1, and 50 y^er cent of breed 2. The second year, wherein three breeds
are used, the resulting pigs will, on the average, possess: 25 per cent of
the chromosomes from breed 1, 25 per cent from breed 2, and 50 per cent
from breed ^. The third year, the pigs will possess 62.5 per cent of the
chromosomes from breed 1 , 12.5 per cent of breed 2, and 25 per cent of breed
3. The fourth year, the pigs will carry, on the average 31.25 per cent of the
linkage groups from breed 1, 56.25 per cent from breed 2, and 12.5 per cent
from breed 3. The lifth year, the pigs will possess, on the average: 15.63 per
cent from breed 1, 28.12 per cent from breed 2, and 56.25 per cent from breed
3. From that time on, they will remain in a continuous cross, in about that
general state of equilibrium, but the percentage of relationship to the differ-
ent breeds will change.

On the basis of these calculations, we advocated rotational crossbreeding.
Some of our critics could not understand how we felt Justified in recommend-
ing rotational crossbreeding when our experiments had been carried only to
the three-breed cross. Calculations showed so clearly that if the three-breed
cross was good, then the continuous cross, by rotation, could not help being
successful, insofar as the system of breeding was concerned. On the basis of
the theory I have always contended that there was very little advantage in a
four-breed cross. Now, however, I am not so sure that that is correct, if we
are to take seriously what my commercial hybrid corn producer told me
regarding the merits of the double cross of corn in contrast to the single
cross. There may be merits in the four- or even the five-way cross that are
not generally revealed in short-time experiments.

We have recommended rotational crossbreeding for commercial swine
production, and it seemed, on the basis of theory again, that the rotational
scheme of crossing had a particular aptitude for swine production, and was
perhaps questionable with other classes of four-footed farm animals. The
reason for this is that in swine it is possible for the commercial producer to
turn a generation every year if he so desires. I have, however, a number of
friends who are breeding commercial flocks of sheep after this general pattern
with remarkably good results. If you look at their flocks with the strictly
commercial viewpoint, they do not have the variance that most critics of the
plan have contended would result. Further than that, the experiments con-
ducted by the United States Department of Agriculture with beef cattle and
dairy cattle have shown that the same basic principles ap})ly to these classes
of livestock as in swine. Dairymen have perhaps been more reluctant to de-
part from the purebred philosophy of breeding than any other group of live-
stock breeders. Yet by a strange coincidence, the experiments of the United
States Department of Agriculture are showing a greater increased yield as
the result of crossing dairy cattle than the crossing of any of our other species
of farm animals. Their data show an increase of 25 per cent in milk and 32
per cent in butterfat yield.

376 L M. WINTERS

ROLE OF INBRED LINES

The next logical question then is: Where and how do inbred lines enter
this general picture? I cannot see that it changes the picture appreciably
unless perhaps it gives an added reason as to why four or five inbred lines
may (theoretically speaking) prove of advantage over the three-line rotation-
al cross. We have now carried the continuous rotational cross of three inbred
lines in two series of crossings to the seventh continuous generation of crossing.
We have several others in the sixth, and several in the fifth generation. The
comparative results of the different line crosses have been remarkably similar
and uniform from generation to generation.

I have already given the average increased performance of the different
breed crosses as being 6.3 for the first cross, 7.5 for the backcross, and 11.7
for the three-breed cross. What then are the increases obtained from crosses
of inbred lines? By the same method of comparison used in breed crosses,
except in this case including an estimate on type, we obtained an average of
approximately 12 per cent increased performance for the crossing of inbred
lines belonging to the Poland China breed, and an increased performance of
18 per cent when we crossed the Minnesota No. 2 with our inbred Poland
China lines, and 20 per cent when we crossed Minnesota No. 1 with Minne-
sota No. 2 or crossed Minnesota No. 1 with our inbred Poland China lines.
This is an increased performance over the performance of the inbred lines.

I am constantly asked what the comparative performance of our crosses
of inbred lines with the performance of outbred stock is. By the best methods
with which we have been able to make comparisons to date, the increased
performance of our crossbred lines in comparison to the performance of the
old-line breeds is an increase of about 20 to 25 per cent. One of these compari-
sons was made with outbred stock from our own University of Minnesota
purebred herds. The other comparison is with the performance of purebred
herds as given by Lush and Molln (1942). I do not regard either of these
comparisons as entirely adequate, and again I will frankly state that I do
not know how to make a comparison that will be entirely adequate. I would
be much obliged if someone would present me with a plan by which a satis-
factory comparison can be made.

I cannot conceive of any sampling method (sampling of the breeds) that
will constitute an adequate sample of the breeds for comparative purposes,
unless we go far beyond any funds that I can conceive of being made avail-
able for this purpose. Field trials such as have been conducted for compari-
sons with corn have been advocated. Some of the corn breeders inform me
that they are not at all satisfied that these field trials are adequate. One
reason is that yield is not a sufficient measure. Many farmers have told
me that our own estimates of the advantages of crossing both the standard
breeds and the use of our inbred lines is in error, due to an underestimate
rather than an overestimate of the benefits.

ROTATIONAL CROSSBREEDING AND HETEROSIS 377

Contrary to expectations, our three-line crosses have not given us as much
increased performance over the two-line crosses as I expected on the basis of
the results with breed crosses and theoretical expectations. I do not know
the cause, but I am inclined to believe that it is due to inadequate sampling,
and that as our samples become larger the advantages of the three-breed
continuous cross will become more pronounced.

FARM APPLICATIONS

How does this work out on farms? The records on one of the large farms
with which I am working show that their percentage of survival from the
purified lines (230 litters) is 75 per cent, whereas their survival from the
crosses of lines (248 litters) is 92 per cent under the same conditions. The
survival of crossbred pigs out of crossbred sows is 91 per cent, but the cross-
bred gilts weaned an average of 9.1 pigs, to 8.3 for the first cross pigs and 7.2
for the purified lines.

This discussion would not be complete without reference to hybrid boars
and how they are entering the picture. I have not seen sufl&cient data to al-
low me to appraise properly the advantages and disadvantages of the so-
called hybrid boar, but he does seem to be proving popular with a number of
farmers. If, then, the hybrid boar is here to stay, what is his place in rota-
tional breeding? In my opinion, it will not change the basic situation ma-
terially, except that at least six inbred lines will be needed to produce the
boars for the rotational crossing of the production of commercial stock. In
this case, we will then use hybrid boar of lines 1 and 2, the following year
hybrid boar of lines 3 and 4, the next year hybrid boar of lines 5 and 6, and
then we will go back to 1-2, to 3-4, and to 5-6 in rotation.

In thinking about rotational crossing, we need to keep in mind that it is
merely a procedure whereby we are able to maintain our breeding females
(and perhaps our breeding males), as well as the offspring, in a relatively
permanent hybrid state. It in no way affects the basic concepts of hybridiza-
tion. It is just a means of utilizing hybridization, and if at some future date our
methods of production change, as for instance the general development of
so-called pig hatcheries, then we may well find some other method of cross-
breeding better suited to the swine industry.

E. L PINNELL

E. H. RINKE

and

H. K. HAYES

Universify of Minnesofa

Chapter 24

Gamete Selection for
Specific Combining Ability"

Gamete selection as a breeding method was designed for more efficient
sampling of open-pollinated varieties. It was suggested by Stadler in 1944.
The method was outlined in detail by Stadler (1945) and preliminary data
presented. Hayes, Rinke, and Tsiang (1946) proposed that the same technic
could be used to select gametes from such sources as synthetic varieties,
single or more complex crosses, and inbred lines. They discussed gamete
selection in its relation to the improvement of a particular double cross
combination.

MATERIALS AND METHODS

In 1945, three double crosses, Minliybrids 602, 607, and 406, were selected
for a method study in gamete selection. Single cross performance data shown
in Tables 24.1 and 24.2 indicate that A344 is low in combining ability in
Minhybrids 602 and 607, and that the same is true for inbreds A25 and A73
in Minhybrid 406.

A344 was crossed to Minnesota #13 (Morris strain) and to 8 inbred lines
namely, Oh51A, A97, 1234, A315, A348, A367, A396, and 111. 4226 as sources
of gametes. The inbreds were selected because of their diversity of origin
and good general combining ability. In addition, A367 had yielded well in
specific tests with A357, A385, and A392. A315 and A348 had performed well
in crosses with A392. The remaining five inbreds had not been crossed to
A357, A385, or A392 in previous years. A25 was crossed with Golden King
and A73 with Murdock. In 1946, individual Fi plants of these crosses were

* Paper No. 2591 of the Scientific Journal Series, Minnesota Agricultural Experiment
Station.

378

GAMETE SELECTION FOR SPECIFIC COMBINING ABILITY 379

selfed and crossed to the opposing single cross parent of the double cross.
Table 24.3 gives the number and type of test crosses made.

Approximately 60 inbred X variety Fi plants from each of the three sources
were selfed and crossed to the testers. Selection of the better Fi plants at
harvest reduced the number for further study to 35, 32, and 38 as listed in
Table 24.3. P:xperiments 1 and 2 were tested in separate randomized blocks

TABLE 24.1
PERFORMANCE OF A344 AND A334*

Inbred s

Av. OF Crosses

% Moist.

Bu.

A344XA357, A392, A385

A334XA357, A392, A385

28.3
30.1

69.8
84.3

* As indicated by average moisture and yield when crossed in
non-parental single cross combinations of Minhybrid 602 (A344X
A334) (A357XA392) and Minhybrid 607 (A344XA334) (A357X

A38.S).

TABLE 24.2
PERFORMANCE OF A25, A334, A73, AND A375*

Inbreds

Av. OF Crosses

% Moist.

Bu.

A25 XA73, A375

24.6
24.7

24.6

24.7

76 2

A334XA73, A375

A73 XA25, A334

79.4
74 8

A375XA25, A334

80.8

* As indicated by moisture and yield in non-parental single
crosses of Minhybrid 406 (A25XA334) (A73XA375J.

using two replicates at each of three locations in central Minnesota. Data
from the two testers were averaged to give a total of 12 replicates as a test
for each gamete. Experiments 3 and 4 were also grown in randomized blocks
with three replicates at each of four locations in southern Minnesota. One
location of experiment 3 was discarded and one replicate of experiment 4 was
abandoned prior to harvest.

On the basis of yield trial results in 1947, gametes were selected from all
varieties and from two inbreds, for use as parents in the development of
new lines by straight selfing and by backcrossing to the sampler inbred.

Study of the performance indices and agronomic characters of the test
crosses led to the selection of three Golden King gametes and four Murdock

380 E. L. PINNELL, E. H. RINKE, AND H. K. HAYES

gametes as higher in general desirability than the sampler inbred parent. In
addition, gametes of low yield potential but of relatively satisfactory agro-
nomic characters were selected from both varieties, three from Golden King
and two from Murdock.

Fo populations were grown in 1948 from the selfed ears of the twelve Fi
plants selected in the above manner. Visual selection was practiced for desir-
able plant and ear characters, and these individual plants were crossed to
the appropriate single cross tester. Remnant seed of the test crosses of the
selected Fi plants was used to make a direct yield comparison in 1949 with
the test crosses of the selected F. plants. Two randomized block experiments
were made at each of three locations in southern Minnesota with three

TABLE 24.3

GAMETE SOURCES, TESTERS USED, AND TEST CROSSES

MADE IN SELECTING GAMETES FOR IMPROVEMENT

OF INBREDS ASM, A25, AND A73

Experi-
ment
Number

1,

2

3

4

Inbreds

A344
A344

A344*
A344

A25

A73

Gamete Source

Tester

Morris 13 gametes

Inbred gametes

A344

Morris 13 gametes

Inbred gametes

A344

Golden King gametes

A25

Murdock

A73

Number

of
Crosses

A357XA392

A357X.\392

A357XA392

A357XA385

A357XA385

A357XA385

A73XA375

A73XA375

A25XA334

A25XA334

35
8
1

35
8
1

32
1

38
1

* Same plants as in experiment 1.

replications per location. One experiment consisted of the crosses of 6 Fi
plants (remnant seed) from A25 X Golden King, and 34 F-, plants by the
tester compared with the cross of A25 X tester. The other included test crosses
of 6 Fi and 37 Fo plants from A73 X Murdock gametes in comparison with
A73 X tester.

EXPERIMENTAL RESULTS
The Morris strain of Minnesota 13 has been a very outstanding open-
pollinated variety in Central Minnesota for many years. Yield trial data
from plants of A344 X Morris 13 crossed with the testers show that a large
proportion of Morris 13 gametes have higher yield potential than A344.
Table 24 4 gives the distribution of yield and moisture data obtained from
thirty-five test crosses of A344 X Morris 13. Sixteen of the thirty-five gametes
gave test-cross vields significantly in excess of A344 X tester. Although not
higher in yield, five other gametes may be considered superior to A344 be-

GAMETE SELECTION FOR SPECIFIC COMBINING ABILITY

381

cause of their significantly lower moisture content at harvest and yields not
significantly different in test crosses from A344 X tester.

The eight inbreds tested as possible sources of germ plasm for the improve-
ment of A344 were A97, A315, A348, A367, A396, OhSlA, Ia234, and 111.4226.

TABLE 24.4

DISTRIBUTION OF PER CENT MOISTURE AND
YIELD OF 35 F, PLANTS OF A344 X MORRIS U
CROSSED TO S.C. TESTERS A357 X A392 AND
A357 X A385*

^(Mean A344
X tester)

% Ear +1

3

5 1

Moisture _,
-2

2 3
3

6

2

6 2
2

-2 -1
Yield

+ 1

+2 +3

(Mean A344 X tester)

* Classes separated by one or more than one LSD (5 per cent) from the
mean of A344Xtester.

TABLE 24.5

DISTRIBUTION OF PER CENT MOISTURE AND YIELD OF
TEST CROSSES OF 8 INBRED LINES AS SOURCES OF GAM-
ETES. CROSSES ARE OF THE TYPE (A344 X INBRED) X TEST-
ERS*

^(Mean of A344
X tester)

%Ear

+2
4-1

2

1 1

1 1

Moisture

-1

1

1

-2 -1
Yield

4-1 4-2 -H3 4-4 4-5

i

(Mean of A344 X tester)
* Classes separated by one or more than one LSD (5 per cent) from the mean of A344 Xtester.

The distribution for yield and moisture of test crosses of A344 X Inbred is
given in Table 24.5. Three inbreds, Ia234, A396, and A97, demonstrated
yield potential significantly higher than A344.

Golden King and Murdock are old, well-adapted varieties formerly
grown extensively in southern Minnesota. Test crosses of thirty-two Golden
King gametes X A25 and thirty-eight Murdock gametes X A73 are shown in
Tables 24.6 and 24.7, respectively.

No gamete from Golden King exceeded A25 significantly in test-cross
yields. However, eight not different in yield were significantly earlier, and

382 E. L. PINNELL, E. H. RINKE, AND H. K. HAYES

are considered superior in yield performance on the basis of maturity. Eight
gametes from Murdock demonstrated yield potential greater than A73, as
indicated by significantly higher yields in crosses. In addition, fourteen
gametes not different from A73 in yield were significantly earlier in maturity.
The proportion of promising gametes extracted from the three varieties
and the eight inbreds is summarized in Table 24.8. About 25 per cent of the
total number tested were superior to the sampler inbred in yield potential.
Another 25 per cent would be considered desirable parents because they had
a yield potential equal to and a maturity potential which was significantly
earlier than the sampler inbreds.

SELECTION OF GAMETES AS PARENTS AND TESTS
OF F2 PROGENIES

Years of testing at Minnesota have led to the conclusion that, in general,
there is a direct relation between yield and moisture content at husking
among hybrids of equal genetic desirability. On this basis the combining
ability of inbred X gamete plants was determined by considering both yield
and moisture percentage at husking. They were effectively placed on a com-
parable basis by calculation of a performance index using the test cross of
the sampler inbred as 100 for both ear moisture and yield. For example, if
the moisture percentage of an A25 X Golden King plant (in test cross) was
93.5 as compared with A25 X tester, and its yield was 106.5 per cent, its per-
formance index would be -f 13. Where the comparative moisture percentage
is higher than the yield percentage the index becomes a negative value.

The performance indices for the selected gametes for both 1947 and 1949
trials and similar data for the F2 plant progeny tests appear in Tables 24.9
and 24.10. The tests of F2 plants from gametes of both high and low yield
potential were made as explained in "Materials and Methods". This wa scar-
riedout by selfing selected F2 plants and also crossing them with the appro-
priate tester, and again comparing the results with the crosses of the appro-
priate inbred with the same tester. Agreement between the two tests of the
gametes was very good except for Murdock gametes numbered 12 and 49.

On the average there was good agreement between Fi and F2 progeny
performance. Tables 24.11 and 24.12 show that there is evidence of segrega-
tion for yield factors within almost all of the F2 families tested.

Mean performance of the F2 progeny from the high testing gametes was
little different from the Fi for either ear moisture or yield (Table 24.13).
However, the F2 progeny from the gametes of low yield performance ex-
ceeded the Fi parent plant in yield performance on the average. This indi-
cates that visual selection of plants within the F2 populations was more
effective among the progenies arising from the gametes of low yield per-
formance than for those F2 plants that were selected from high performing Fi
crosses (gamete X inbred).

TABLE 24.6

DISTRIBUTION OF PER CENT MOIS-
TURE AND YIELD OF 32 Fi PLANTS
OFA25 X GOLDEN KING CROSSED
TO A73 X A375*

+2
% Ear +^

1
3

2

Moisture _.
-2

I

8
5

5
3

1 +1 +2

Yield

♦Classes separated by one or more LSD (5 per
cent) from the mean of A25Xtester.

TABLE 24.7

DISTRIBUTION OF PER CENT MOIS-
TURE AND YII:LD of 38 Fi PLANTS
OF A73 X MURDOCK CROSSED TO
A25 X A334*

+3

1

+2

+ 1

1

1

-1

5

7

3

-2

1

2

9

3

-3

2

1

2

%Ear
Moisture

-2 -1 -fl +2
Yield

* Classes separated by one or more LSD (5 per
cent) from the mean of A73Xtester.

TABLE 24.8

NUMBER AND SOURCE OF GAMETES SUPERIOR
IN PERFORMANCE TO SAMPLER INBREDS

Source of
Gametes

Sampler
Inbred

Total

Gametes

Tested

Higher in
Yield and
as Early or
Earlier in
Maturity

Not Dif-
ferent in
Yield but
Earlier

Morris 13

8 inbreds

Golden King. . .
Murdock

A344
A344
A25
A73

35

8

32

38

16

2*

0

8

5

0

8

14

Total....

113

26

27

* An additional gamete higher in yield was also later in maturity.

383

TABLE 24.9

PERFORMANCE INDICES OF TEST CROSSES OF

SELECTED Fi PLANTS AND F. PROGENY

FROM A25 X GOLDEN KING

Perform.«ice Index

Fi Plant
Number

1947

1949

Number

OF Fz

Plants

Fi

Fi

F2

19H*. ...

20 H

36H

5L

29 L

46L

+ 11
+ 14
+ 9
-11
-11
- 5

+ 19
+ 9
+ 16
- 3

- 1

+ 1

+25
+ 14
+ 11
+ 5
- 0
+ 2

5
7
7
7
1
7

H = high- and L = low-performing gametes.

TABLE 24.10

PERFORMANCE INDICES OF TEST CROSSES OF

SELECTED Fi PLANTS AND Fj PROGENY

FROM A73 X MURDOCK

Performance Index

Fi Pl.\nt
Number

1947

1949

Number

OF Fa

Plants

F,

F,

F2

12H

14 H

49 H

50 H

6L

35 L

+ 18

+25
+20
+29
-10
- 4

- 3

+33
+ 1
+ 19
-23
-24

- 6

+27
+ 5
+ 18
-10

- 7

8
5
6
7
5
6

384

TABLE 24.11

FREQUENCY DISTRIBUTION OF PERFORMANCE IN-
DICES OF 1-2 PROGENY PLANTS FROM A25 X
GOLDEN KING AROUND MEAN PERFORMANCE
OF A25 X TESTER

Fi Plant

Indices for F2 Plants

Number

-15

-S

+ 5

+ 15

+25

+35

19 H

20 H

36 H

5L

29 L

46 L

1

3
1
1

2

3
2

5

1

3
4

1

2
2

1

2

TABLE 24.12

FREQUENCY DISTRIBUTION OF PERFORMANCE
INDICES OF F2 PROGENY PLANTS FROM A73 X
MURDOCK AROUND MEAN PERFORMANCE OF
A73 X TESTER

Fi Plant

Indices for ¥2 Plants

Number

-15

-5

+ 5

+ 15

+25

+ 35

12 H

14 H

49 H

50 H

6L

35 L

5

2

2

1

2

2
3

1

2

2
1

1

1
2
2

1
2

2

2
1

TABLE 24.13

COMPARISON OF Fi PLANTS AND THEIR F2 PROGENY IN 1949 TEST
CROSS PERFORMANCE FOR EAR MOISTURE, YIELD,
AND PERFORMANCE INDEX

Parents

No.
Plants
Tested

Ear Moisture %

Yield in Bu.

Performance
Index

F,

F2

Fi

F2

Diff.

Fi

F2

Diff.

Fi

F2

Diff.

A25XG. KingH.. ..

A25XG. KingL

A73XMurdockH....
A73XMurdockL....

3
3
4
2

19
15
26
11

20.3

23.3
21.8
22.8

20.2
23.8
21.1
22.4

-0.1
+0.5
-0.7*
-0.4

54.0
52.9
56.0
41.7

54.5
56.3
55.0
48.7

+0.5

+3.4**
-1.0
+ 7.0**

+ 14.4
- 2.1
+ 12.5
-23.6

+ 15.8
+ 3.3
+ 9.6
- 8.6

+ 1.4
+ 5.4
- 2.9
+ 15.0

* Exceeds 5% point of significance.

** Exceeds 1% point of significance.

385

386 E. L PINNELL, E. H. RINKE, AND H. K. HAYES

DISCUSSION

Almost 50 per cent of the gametes studied showed evidence of having
combining ability in excess of the sampler lines. The gametes chosen as par-
ents appear to furnish a desirable source of germ plasm for a selection pro-
gram designed to improve the yield potential of A344, A25, and A73 in
combining ability in specific crosses.

Where a high combining varietal gamete is chosen for an inbred selection
program, the Fi plant of which it is a parent represents a high X low type of
cross so far as combining ability is concerned. To the extent that such Fi
plants are comparable to crosses of inbreds, the breeding results should be
similar to those from crosses of inbreds differing in combining ability. At
Minnesota (Hayes and Johnson, 1939), crosses of highXlow combiners have
given F5 lines ranging from high to low, but with a good proportion of high

combiners.

Whether selection of gametes should be followed by test controlled selec-
tion in the F2 is an important question. In these studies more than 50 per
cent of the F2 plants from high combining gametes tested at least ten per-
formance index units higher than the sampler inbred. Thus without further
test crosses, the chances of choosing high combining F2 plants would still
have been very good. The number of Fo plants that could be handled in test
crosses was quite limited. This may make for greater difficulty in recovering
or improving the agronomic type of the sampler lines. It was very evident
from field observations that the proportion of agronomically desirable F3
lines appeared lower than usually found from crosses of highly selected

inbreds.

The greater effectiveness of visual selection among the F2 progenies of the
low testing gametes is at this stage only an interesting development. Only a
small proportion of the plants arising from the low testing gamete parents
exceeded the sampler inbred in performance by a significant amount.

It was not possible to determine by visual examination which F2 popula-
tions were derived from high gametes and which from low gametes, although
there were wide differences in plant type between populations.

Gametes from eight inbred lines compared fairly well with varietal
gametes from Morris 13, in offering promising sources of germ plasm for the
improvement of specific combining ability of A344. Where a breeder has
available large numbers of inbred lines of diverse origin the use of test
selected inbred parents rather than varietal gamete parents may be the more
feasible approach. Selection for characters other than yield would presum-
ably be done more economically. The same advantage can be claimed for the
use of complex crosses of inbreds. On the other hand, utilization of varietal
gametes in improvement work does not "use up" inbreds so far as their
combination in hybrids for commercial use is concerned. Lines recovered

GAMETE SELECTION FOR SPECIFIC COMBINING ABILITY 387

from crosses of inbreds may be more restricted than their inbred parents in
commercial use because of relationship. It seems probable to the writers that
the method of gamete selection is worthy of considerable use for further
selection of material from open-pollinated, desirable commercial varieties.
Studies of lines recovered from selected varietal gametes will have to be
carried to F5 or later generations to determine if the large amount of out-
cross testing is justified economically. The writers would like to emphasize
the importance they attach to method studies of the type presented here.
New ideas in breeding must be explored constantly if continued progress is
to be made in corn improvement.

SUMMARY

Since 1945, a program has been underway at Minnesota to attempt im-
provement of Minhybrids 602, 607, and 406 by the method of gamete selec-
tion. The hybrid pedigrees are respectively: (A344XA334) (A357XA392),
(A344XA334) (A357XA385), and (A25XA334) (A73XA375). Detailed
studies of the non-parental single crosses among the inbred parents of each
hybrid led to the conclusion that A344 in Minhybrids 602 and 607, and A25
and A73 in Minhybrid 406 were low in combining ability.

A344 was crossed to the Morris strain of Minnesota 13 and to eight in-
breds of diverse origin. A25 was crossed to the Golden King variety and A73
to Murdock. (Inbred X gamete) (tester) crosses were made using the oppos-
ing single cross parents as testers. These were compared with the appro-
priate cross of inbred X tester. Yield trial performance was obtained from a
total of 113 gametes, 35 from Morris 13, 8 inbreds, 32 from Golden King,
and 38 from Murdock.

Sixteen gametes from Morris 13, three from the inbreds, and eight from
Murdock gave significant increases in yield over the test crosses of the checks
A344 and A73. Five gametes from Morris 13, eight from Golden King, and
fourteen from Murdock were not significantly different in yield but were
significantly earlier so that yield performance could be considered better
than the checks on the basis of ear moisture at harvest. These varieties and
the three high testing inbreds thus appear to be good sources of gametes for
improving the relatively low performing inbreds in specific combining ability
for yield.

Both high and low testing varietal gametes were selected for use in a study
of the development of new inbreds. From the crosses, A25X Golden King and
A73X Murdock, selected F2 plants X the approj)riate tester were compared
with the progeny of their Fi parental plants when crossed on the same tester.
While there was excellent agreement, on the average, for combining ability
in the Fi and F2, there was evidence of segregation for combining ability from
almost all of the twelve Fo families which were studied. Visual plant selection
within the F2 generations appeared to be effective in increasing yield per-

388 E. L. PINNELL, E. H. RINKE, AND H. K. HAYES

f ormance of the plants from the low testmg gametes, but appeared to have no
effect in further increasing the yield performance of the F2 plants from the
high testing gametes.

The economic feasibility of Fo plant testing in a gamete selection program
is questioned.

SHERRET S. CHASE

Iowa Sfafe College

Chapter 25

Monoploids in Maize

Haploid sporophytes have been reported in jimson weed (Blakeslee et al.,
1922), cotton (Harland, 1920), tobacco (Chipman and Goodspeed, 1927),
evening primrose (Gates, 1929), maize (Randolph, 1932a, 1932b), wheat
(Gaines and Aase, 1926), rice (Ramiah et al., 1933) tomato (Lindstrom,
1929), pepper (Christensen and Bamford, 1943), and in many other genera
which have been subjects of cytogenetic study.

A haploid organism, strictly speaking, is one which has only one set of
chromosomes, that is, one genome per cell. In the common usage of botanists,
geneticists, and others, a sporophy te originating by reduced parthenogenesis
or by an equivalent process, and consequently carrying the reduced or gamet-
ic complement of chromosomes in each cell instead of the normal zygotic
complement, is referred to as a haploid.

Thus the term, as applied to a sporophyte, has come to carry the connota-
tions of both parthenogenetic origin and gametic chromosome number, and
the actual genomic condition tends to be ignored. Many so-called haploids are
actually diploids or polyploids. Thus the haploids of common wheat are
triploids, since the parent species, Triticum vulgare, is a hexaploid. To em-
phasize the fact that the haploids of maize carry only one set of chromosomes
per cell, that is, only one chromosome of each type instead of the normal pair,
the alternate term monoploid is used here to designate these aberrant plants.

In normal sexual reproduction in maize the pollen tube penetrates the
eight nucleate embryo sac. One of the two male gametes released fuses with
the egg nucleus to form the zygote, while the other fuses with the two polar
nuclei to form the primary endosperm nucleus. In the abnormal type of re-
production giving rise to monoploid sporophytes the ])rocesses apparently
are the same except that for some reason the first male gamete fails to fuse
with the egg nucleus and is lost. The egg nevertheless is activated and devel-

* Journal Paper No. J-1906 of the Iowa Agricultural Experiment Slalion, Ames, Iowa.
Project 1201.

389

390 SHERRET S. CHASE

ops into an embryo. Evidence for this is indirect — monoploid embryos are
found in kernels having normal (3n) endosperm. It is possible that some or all
monoploids arise from reduced cells of the embryo sac other than the egg,
from the synergids perhaps.

As tools for experimental research monoploids offer many possibilities: in
the cytological field for studies of the meiotic distributions of unpaired chro-
mosomes, non-homologoiis synaptic relations of the chromosomes and me-
chanics of chromosome doubling; in the genetic field for direct observation of
mutational effects, measurement of mutation rates, studies of cytoplasmic
effects, and biochemical investigations; in the agronomic field for the produc-
tion of diploid, homozygous stocks directly from the monoploids. The follow-
ing discussion is concerned primarily with my own investigations into the
latter possibility.

A monoploid carries in each of its cells, or nuclei, only one chromosome of
each type. Thus if the chromosome complement of any cell can be doubled,
the affected cell and any derivative of it will consequently be both diploid and
homozygous. If such homozygous diploid sectors include the reproductive tis-
sues, meiosis should then be normal and the gametes produced functional.
Thus such plants can produce diploid progeny — homozygous diploid progeny
if the individual is successfully self pollinated — since every gamete of the
plant is genetically equivalent to every other gamete. In a monoploid without
diploid sectors, since the chromosomes lack synaptic mates, meiosis is highly
irregular. Only rarely are functional gametes carrying the full complement of
chromosomes produced. If two of these rare functional gametes from a single
monoploid do fuse in syngamy, the zygote produced will be diploid, and
homozygous, unless the gametic chromosomes were subject to chromosomal
aberration during the irregular meiosis.

Production of homozygous diploid progeny from monoploids results in the
fixation of a single gametic complex. In any population, desirable gametes are
more frequent than desirable zygotes. Foi example, if one has on hand an
individual heterozygous for three pairs of genes and wishes to obtain from it a
definite homozygous product by selfing, one individual in sixty-four of the im-
mediate self progeny (Si) will, on the average, carry the desired genotype.
One gamete in eight, extracted as a monoploid and then converted into a
homozygous diploid, will furnish the same genotype (see Fig. 25.1).

Successful production of homozygous diploids in quantity from mono-
ploids depends upon the adequate solution of two main problems. The first
of these is the production and recognition in the seedling, or in earlier stages,
of large numbers of monoploids. This problem has been solved to the extent
that thousands of monoploids can be produced with relatively small expendi-
ture of effort. The second problem is that of deriving self, and consequently
homozygous, diploid progeny from the monoploids isolated. This problem
also has a practical, though partial, solution.

MONOPLOIDS IN MAIZE

391

MONOPLOIDS IN MAIZE

It has been known for some time that monoploids occur naturahy in maize
in measurable frequency. Data of Randolph (Randolph and Fischer, 1939)
and of Einset (1943) suggest that monoploids occur spontaneously at a rate
of about one per thousand. Data of Stadler (unpublished) indicate a rate of
about one per hundred in a genetic stock. At the start of the studies reported

\ cT

? \

ABC

ABe

AbC

Abc

aBC

aBc

ahC

abc

ABC

ABC

ABe

AbC

Abe

aBC

aBe

abC

abc

ABC

ABC

ABC

ABC

ABC

ABC

ABC

ABC

ABC

ABe

AbC

Abe

aBC

aBc

abC

abe

A Be

A Be

ABc

ABc

ABc

ABc

ABe

ABe

ABc

ABC

ABc

AbC*

Abe

aBC

aBe

abC

abc

AAbbCC*...AbC...AbC

AbC

AbC

AbC

AbC

AbC

AbC

AbC

AbC

ABC

ABc

AbC

Abc

aBC

aBc

abC

abc

Abe

Abe

Abe

Abc

Abe

Abc

Abc

Abc

Abc

ABC

ABe

AbC

Abe

aBC

aBc

abC

abc

aBC

aBC

aBC

aBC

aBC

aBC

aBC

aBC

aBC

ABC

ABe

AbC

Abc

aBC

aBc

abC

abe

aBc

aBe

aBe

aBe

aBe

aBe

aBc

aBc

aBc

ABC

ABc

AbC

Abc

aBC

aBe

abC

abc

abC

abC

abC

abC

abC

abC

ahC

abC

abC

ABC

ABc

AbC

Abc

aBC

aBc

abC

abc

abc

abe

abc

abc

abc

abc

abc

abc

abc

* Desired homozygous individual, j of gametes and ,'4 of zygotes.

Fig. 25.1 — Efficiency of Monoploid Method Compared with Selfing to Si for Obtain-
ing Homozygous Individual AAbbCC from Heterozygous Parent AaBbCe.

here it was assumed that naturally occurring monoploids would furnish a suf-
ficient supply at a rate of occurrence of the order of one in one or two thou-
sand plants of a progeny provided the bulk of the diploids could be screened
out by some simple device during the seed or seedling stages. This has proven
feasible.

It was also assumed that some method for inducing doubling of the mono-
ploid chromosome complement would have to be developed. Though this still
appears desirable and possible, artificial induction of chromosome doubling
has not been necessary in order to obtain diploid self progeny from a portion
of the monoploids. The reason for this is that the fertility of the plants is in-

392 SHERRET S. CHASE

creased naturally by spontaneous doubling of the chromosome complement.
About 10 per cent of untreated monoploids have yielded successful self
progeny, largely as a result of this spontaneous somatic diploidization.

Since monoploids are for the most part of maternal origin, these plants
should resemble their seed parents. Thus the search for monoploids is greatly
facilitated if one looks for them among the progeny of markedly dissimilar
parents. If one crosses a purple maize stock as pollen parent onto plants
which lack this color and then finds non-purple seedlings in the progeny, one
has reason to think these aberrant plants may be monoploids. In practice, the
marker phenotype is used to indicate the diploid plants. These are discarded
as recognized. Morphological and cytological tests are used for positive recog-
nition of the monoploids.

In brief, the techniques used in isolating monoploids are as follows. The
stock from which one wishes to obtain monoploids is pollinated with pollen
from a genetic marker stock. The marker may carry the purple plant color
genes (.4i A^B PI R) or brown (ai A^ B PI R), purple plumule (.4 Pui Pn-i),
or any suitable complex of marker genes not carried by the seed stock. The
ears at harvest are checked for kernels resulting from accidental self or cross
pollinations. This check is made possible by using marker stocks which carry
endosperm marker genes as well as plant marker genes. The markers which
have been used, as appropriate, are purple aleurone {AiA^AzCR i Pr) , red
aleurone (.4 1 ^4 2 .-1 3 C i? / pr) , starchy endosperm {Su) , and yellow endosperm
(F).

The kernels riot showing the endosperm marker phenotype are discarded
(if the pollinations have been carefully made few discards are necessary).
Then the kernels saved are germinated and a check made of the embryos or
seedlings for the plant marker phenotype. All showing this character are dis-
carded. The remainder are transplanted after first taking from each a root tip
or two for cytological study. A second screening of? of diploids is carried out
after the first seedling leaves of the putative monoploids are fully extended.
Those having the first leaf as long as the comparable leaf of the seed parent
are almost without exception diploid and are therefore discarded. The true
monoploids are then recognized by chromosome number determinations.
Errors in classification at each stage result primarily in loss of monoploid
plants. Consequently monoploid frequencies as reported are likely to be less
than the actual frequencies of occurrence.

The putative monoploids screened o&. as a result of the genetic check in-
clude the actual monoploids and also diploids of the following types: diploid
hybrids mutant for marker genes, hybrids carrying strong color suppressor
genes, hybrids in which disease (generally fungus infection) has suppressed
the development of the color phenotype, and a few maternal diploids. Occa-
sionally paternal monoploids also are produced. These may be recognized

MONOPLOIDS IN MAIZE 393

when the hybrid phenolype is unUke that of either the pollen or the seed
parent, as is the case in crosses in which the brown marker stocks are used as
pollen parents. In such crosses, maternal monoploids of the progeny should
resemble the seed parent. Paternal monoploids should be brown (green at
early stages) and the hybrids purple. The particular brown stocks used carry
recessive markers, liguleless or japonica. These also serve to mark the very
rare paternal monoploids.

When the monoploids reach the reproductive stage the practice has been
to self these plants if any self pollen is shed, to cross them by other mono-
ploids shedding excess pollen, or to pollinate them by diploids if self pollen is
lacking.

FERTILITY OF MONOPLOIDS

The estimate of the fertility of monoploids, based on the assumption of 10
chromosomes distributed independently at meiosis, is one normal egg in
1024. That is, if abundant normal pollen were used in pollination these plants
should set one good kernel in 1024 ovules. Actual fertility of the monoploids
studied has been much higher than this, in spite of the fact that the amount of
pollen used has often been scant. Little is known of the mechanics of meiosis
in maize monoploids. Studies of the reactions of unpaired chromosomes at
meiosis suggest that monoploid meioses may produce some functional gam-
etes with structurally altered chromosomes (Kostoff, 1941). A proportion of
the syngamic products in such cases would consequently be structurally
heterozygous. If the reproductive tissue of a monoploid becomes diploidized
before meiosis is initiated the gametes produced should all be structurally
normal and strictly equivalent genetically. Some progenies were checked to
determine the extent of chromosome aberration. The percentage of non-
viable (actually, non-stainable) pollen produced by the monoploid deriva-
tives was used as an indication of chromosome abnormalities. Among the
progenies of diploid seed parents by monoploid pollen parents about 1 per
cent had 10 per cent or more bad pollen. Among the progenies of monoploid
seed parents by diploid pollen parents about 8 per cent had 10 per cent or
more bad pollen. Among the progenies of monoploid by monoploid, 17 per
cent had 10 per cent or more bad pollen. In the latter two classes, both of
those in which monoploids were used as the seed parents, the monoploids
thus used were those which had shown no evidence of diploidization in the
tassels.

In a group of 298 monoploids, 282 matured. Of these 139 shed some pollen,
68 formed kernels, and 34 yielded successful self progeny. The fertility of this
group of plants and of the whole series to date was far in excess of that ex-
pected of maize monoploids on theoretical grounds. The difference can be
ascribed largely to spontaneous doubling of the chromosome complement in
cells giving rise to reproductive tissue (Chase, 1949b).

394

SHERRET S. CHASE

PARTHENOGENESIS

A number of interesting facts have come out of studies of the frequency of
reduced parthenogenesis in maize. One unanticipated fact has been that of
the effect of the male (pollen) parent on parthenogenesis. Although this
parent does not contribute its genes to the maternal monoploid, the particu-
lar pollen parent used in any cross does have an effect on the rate of occur-
rence of maternal monoploids (Chase, 1949a). In Table 25.1 the results of

TABLE 25.1

FREQUENCIES OF OCCURRENCE OF MONOPLOIDS FROM SEVERAL

INBREDS AND HYBRIDS WHEN INBREDS A385 AND

38-11 WERE USED AS THE POLLEN PARENTS

A385 AS Pollen Parent

38-11 AS Pollen Parent

Seed Parents

Number of
Progeny

No.
»

Freq.
per
Thou-
sand

Av.

Number of
Progeny

No.
n

Freq.
per
Thou-
sand

Av.

Os420

M14

WF9

1,715

2,074
1,792
1,839
6,238
5,148

5,068

1
0
0
0
0
2

1

.58

.00
.00
.00
.00

.38

.20

.29
.00
.19

4,909
2,738
5,065
2,322
6,648
3,554

4,868

9

5

6

3

11*
12

2*

1.84
1.83
1.19
1.29
1.66
3.38

.41

1.83

W22

1.24

Os420/M14. . . .

WF9/W22

(Os420/M14)/
(WF9/W22)..

2.52

Averages ....

.17

1.66

Golden Cross
Bantam

12,324

2

.16

6,638

20

3,01

Known to be too low.

paired crosses involving two different pollen i)arents, inbreds A385 and
38-11, are summarized. Both of these inbreds carry the purple plumule
marker system. From the genetic point of view A385 is the more satisfactory
of the two. That is, in its hybrids the marker phenotype is generally well
developed. In the hybrids of 38-11 the phenotype is often obscure. Conse-
quently few monoploids were lost by misclassification in the progenies of
A385, whereas a considerable number may have been lost in those of 38-11.
In spite of this the data show 38-11 to be ten or more times as effective as
A385 as a stimulator of parthenogenesis. This effect seems to be general. That
is, the several dent stocks and also the sweet corn hybrid show about the

MONOPLOIDS IN MAIZE

395

same proportionate effect of the j)ollen parents. Other data involving other
crosses and data taken in other seasons are in agreement with tliose summa-
rized here.

Data summarized in Table 25.2 are presented to show the variation in
monoploid frequency dependent on the seed parent. Summaries are given of
frequencies in crosses in which a single pollen parent, a brown marker, was
used. The differences, expressed in terms of frequencies per 1000 seedlings and
also as the frequency per seed parent, are quite striking. The rate of partheno-
genesis seems to be roughly proportional to the intensity of agronomic selec-
tion to which these various stocks have been previously subjected.

TABLE 25.2

FREQUENCIES OF OCCURRENCE OF MONOPLOIDS IN SOME

DENT STOCKS WHEN CROSSED BY A UNIFORM

POLLEN PARENT

Seed Parent

Pollen

Number
PTQgeny

No.
n

Freq. per
Thousand

Freq. per

Seed

Parent

Lancaster

Reid's

Stiff Stalk Synthetic (SSSo)t

Early Synthetic (ESo)

Dent Inbreds and Hybrids!

N*

N

N

N

N

N

10,173

14,650

91,125

8,226

4
11

90
10

176

.39

.75

.98

1.13

1 . 35

1.45

.12

.38
.37
.36

Stiff Stalk Synthetic (SSSi)t

121,764

.51

Averages

1.01

.35

* Brown, liguleless stock, a B PI C R" Pr Ig; Randolph 43687-1.

t Original and first cycle Stiff Stalk Synthetic.

i 1947 data (Chase 1949), averages of frequencies per thousand.

Other data, including that from sweet corn varieties, hybrids and inbreds,
bear out this relation. A likely explanation, other things being equal, is that
the frequency of occurrence of viable monoploids is correlated inversely with
the frequency of lethal genes in the source stocks. That the frequency of
lethal genes in a stock is not the sole basis of differences between stocks be-
comes evident when one compares stocks which have been subject to an
equivalent degree of selection.

It also becomes evident that there is another genetic basis for differences
in rates of parthenogenesis when one analyzes the frequency of occurrence of
monoploids as a function of the individual seed parent plants. In Table 25.3
summaries are given of the numbers of monoploids per seed ])lant in crosses
in which the Stiff Stalk Synthetic variety was used as the seed parent. The
distribution of none, one, two, and three monoploids per parent is about what
one might expect on a chance basis. But the likelihood of getting five, six, and
seven monoploids per seed plant by chance in three, three, and two cases
respectively in a sample of 1065 parent plants is remote. The likeliest expla-

396 SHERRET S. CHASE

nation is that certain genotypes favor parthenogenesis. Whether this is a
function of the sporophyte or of the gametes is not certain. It appears more
likely that the efifect originates in the individual gametes (eggs).

Emerson (unpublished) and Lindstrom (unpublished) and others have
attempted to stimulate parthenogenesis in maize by the application of hor-
mones and other chemicals to the ovules before or during fertilization. The
results were uniformly discouraging. Randolph (1932b) found a number of

TABLE 25.3

DISTRIBUTION OF MONOPLOIDS PER SEED
PARENT, STIFF ST.ALK SYNTHETIC

Number of

Monoploids Number of Seed

per Seed Parents in

Parent Each Class

0 776

1 195

2 60

3 19

4 7

5 3

6 3

7 2

Total 1,065

TABLE 25.4

MONOPLOID FREQUENCIES AMONG THE

PROGENIES OF MONOPLOID

DERIVATIVES

Seed Parents

Pollen
Parent

Number of
Progeny

Number
of Mono-
ploids

Freq. per
Thousand

H159

(H15/H25),S,....
(H19/H25),Si....
(H152/H143)

V
V
V
V

1,716

1,792

537

550

15

14

5

10

8.70

7.81

9.34

18.18

monoploids in material which had been subjected to heat treatments de-
signed to induce polyploidy. Though it is a question whether the heat induced
parthenogenesis, this type of treatment should be repeated on material in
which the natural rate of parthenogenesis is known. In connection with the
general monoploid study reported here a number of special treatments have
been tried. Among these are hormone treatments, X-radiation of pollen,
intergeneric crosses, pollination with pollen from tetraploid maize, and de-
layed pollination. These experiments are incomplete.

There is presently available one method by which high rates of partheno-
genesis can be had. This is by selection of the pollen and seed parents used in

MONOPLOIDS IN MAIZE 397

a cross. As shown in Table 25.4, monoploid derivatives are particularly
favorable parthenogenetic stocks. In this series of crosses the stock V used as
the pollen parent is a purple marker which is better than average as a stimu-
lator of parthenogenesis. The seed parent in each case was a monoploid de-
rivative; either a homozygous diploid (HI 59), or a single cross hybrid
between two monoploids (H152 H14v3), or an advanced generation of such a
hybrid.

The average frequency ])er 1000 for the stock from which H15y was de-
rived (the Stifif Stalk Synthetic) is about 1.21. In each case the frequency of
parthenogenesis is higher than that of the stock or stocks from which the
monoploid derivatives were obtained. The hybrid H152/H143 and the fre-
quency of monoploids in its progeny are particularly interesting in that HI 52
was a monoploid extracted from Inbred P39 and H143 a monoploid from
Inbred P51. Thus the cross of the two is the single cross hybrid Golden Cross
Bantam, based on monoploid parents. Normal Golden Cross Bantam crossed
by marker stock V has a monoploid frequency of about 4.00 per thousand. A
high rate of parthenogenesis is characteristic not only of the four stocks listed
in Table 25.4 but of all monoploid derivatives adequately tested.

H159 not only has a high rate of parthenogenesis among its progeny but
also a high degree of fertility among the monoploids produced. Of the 15
monoploids obtained from the cross with stock F, 12 were grown to maturity.
All of these had one or more diploid sectors in the tassel and all set good seed.

On the average about one monoploid in ten is self fertile — in the sense that
it yields a successful homozygous diploid progeny. One would like to obtain
diploid self progeny from all monoploids. Since any increase in the rate of
somatic diploidization should result in increased fertility, a number of treat-
ments with polyploidizing agents have been tried. Colchicine, as used,
brought about an increase in fertility but injury to the plants killed so many
that no over-all gain was effected. In these treatments, solutions of approxi-
mately .5 per cent aqueous colchicine were injected into the scutellar nodes of
the monoploid seedlings. It is possible that use of more dilute solutions in-
jected repeatedly would be more effective.

Podophyllin, as a saturated aqueous solution, produced drastic stunting
and inhibition of the development of the ears and tassels. Heat treatment,
tried on a very minor scale, seemed to be about as effective as colchicine and
had the same disadvantage. In this problem, as in that of increasing the rate
of parthenogenesis, genetic methods seem to offer the best available solution.
That is, stocks derived from self fertile monoploids are better sources of self
fertile monoploids than the stocks from which the original monoploids were
obtained.

Synthetic varieties that combine high monoploid frequency, high mono-
ploid fertility, and high general agronomic desirability can probably be de-
veloped from homozygous diploids, both sweet and dent, already on hand.

Fig. 25.2 — Sweet corn monoploid sporophyte derived from Golden Cross Bantam.

Fig. 25.3— Ears of homozygous diploid dent (H502) and inbred WF9. H502 is a Stiff Stalk
Synthetic derivative. The ears shown are from plants of the first diploid generation.

MONOPLOIDS IN MAIZE 399

Such synthetic varieties should be 10-20 times more effective as sources of
new homozygous diploid lines than the better heterozygous stocks already
tested.

About fifty homozygous sweet corn diploid stocks and about fifty homo-
zygous dent stocks have been developed at Ames during the past two years of
exploratory work. These are being tested for combining ability in comparison
with related inbred lines. Though there is no reason a priori to expect that
these lines will be better than average combiners, there is reason to think they
should carry well balanced genetic systems, since passage through the si)oro-
phyte phase as a monoploid involves drastic selection against lethal and sub-
lethal genes. In appearance the homozygous lines seem better than average
unselected inbreds in general vegetative vigor.

CONCLUSIONS

• It has been demonstrated that homozygous diploid stocks of maize can be
produced from monoploid sporophytes. The method as now developed is
practical from the point of view of the plant breeder as an alternate to in-
breeding for the production of homozygous lines. As a method of gamete se-
lection it offers unique possibilities. Improvements now being attempted
should increase the efficiency of the procedure very considerably. It is not
known yet whether the homozygous lines produced will prove to be better or
poorer or equal to unselected advanced generation inbred lines on the average
in respect to combining ability.

G. F. SPRAGUE

USDA and Iowa State College

Chapter 26

Early Testing and Recurrent Selection

It appears desirable to review the history of corn breeding very briefly in
order that early testing and recurrent selection may be placed in their proper
perspective. The first breeding method used in corn was undoubtedly mass
selection. The fact that the corn ear is large, and that harvesting for a long
period of time was essentially a hand operation, provided excellent oppor-
tunities for selection to be practiced on ear length, diameter, and kernel char-
acteristics. This type of selection undoubtedly was practiced from the be-
ginning of the domestication of the corn plant until well into the twentieth
century. This type of selection was quite effective in modifying ear and kernel
characters even though it provided no opportunity for parentage control.
Variation in ear size, etc., due to soil fertility were assumed to be genetic.

Varietal hybridization was the next breeding procedure tried. The results
obtained in some cases were very promising, but no extensive use was made of
the method. Varietal hybrids, however, did provide source material from
which many of the widely grown varieties were isolated.

The ear-to-row method of breeding was suggested by C G. Hopkins of the
Illinois Station in 1896. This procedure, as the name implies, involved select-
ing a group of ears, planting these ear-to-row and obtaining information on
performance. In such tests, marked differences in yield were obtained among
the ears tested. This method was tried rather extensively, but was finally
abandoned when it became apparent the cumulative improvement in yielding
ability was not realized.

The ear-to-row breeding method provided for selection on the basis of the
visual characters of the original parent ears and some measure of performance
based on the progeny of the selected ears. Opportunities for genetic control
were limited, and the original high yielding progenies were hybrids of un-
known ancestry which could not be duplicated. The ear-to-row method of
breeding was quite effective in modifying chemical composition, plant and

400

EARLY TESTING AND RECURRENT SELECTION 401

ear height, and leaf area. These characters, for which selection was efifective,
differ from yield in that the genetic basis is undoubtedly much less complex
and environmental variability less likely to lead to mistaken classifications.
We now know that the plot technics used in these ear-to-row trials were quite
inadequate, and some of the failure to achieve improvements in yields must
certainly be ascribed to this cause. Many of the modifications of the ear-to-
row method of breeding which were introduced to minimize inbreeding prob-
ably had an opposite effect, and the rate of inbreeding was actually in-
creased. On the basis of data now available, it is impossible to fully assess the
relative importance of various causes resulting in the ineffectiveness of this
method in increasing yields.

SELECTION WITHIN AND AMONG INBRED PROGENIES
The next method tried, and the one still used most extensively, involved
selection wdthin and among inbred lines and the evaluation of the lines re-
tained in hybrid combinations. Some of the early work which served as a
foundation for this breeding method has been reviewed in other chapters of
this book. Extensive breeding programs were established at the various sta-
tions in the early 1920's, and a large percentage of the lines now used in the
production of commercial hybrids had their origin in this early work.

In the earlier days of these programs any inbred line which could be main-
tained was considered to have potential value. As the work progressed it be-
came apparent that inbred lines must meet certain minimum standards of
performance as lines in order to merit testing in hybrid combinations. Studies
were undertaken by Jenkins (1929) and somewhat later by Hayes and John-
son (1939) to determine which, if any, characters of the inbred lines were cor-
related with yield in hybrid combinations. In the studies reported by Jenkins
correlations were used to measure the relationship between (1) various char-
acters of the parental inbreds and the same character in their Fi hybrids, and
(2) between characters of the parental inbreds and the means of the same
characters for all of their crossbred progeny. The results obtained under 1
and 2 were somewhat different. In the first case, none of the characters of the
parental inbreds were closely related to the yield of their Fi hybrids. The cor-
relations reported ranged from — .10 to -|-.24. The correlations between yield
of the parents and yield of their Fi hybrids were .14 and .20. Multiple cor-
relations considering various grouping of characters of the inbreds and the
yield of their hybrids ranged from .20 to .42.

In the second series which involved characters of the parental lines and the
means of the same characters for all crossbred progeny, the correlations ob-
tained were materially larger. With different groups of material the correla-
tions involving yield ranged from .25 to .67. In some cases the degrees of free-
dom were few and the relationshij) therefore poorly determined. A weighted r
calculated for the entire series was .45. The difference between these two

402 G. F. SPRAGUE

series can be readily accounted for by the assumption of epistasis, though no
claim is made that this is the only or even the correct explanation. Where the
correlations involve some character of the inbred parent and the same char-
acter in their Fi crosses, epistatic effects would be expected to be at a maxi-
mum. When a character of the parent is correlated with the mean of all cross-
bred progeny opportunity would be provided for a considerable degree of can-
cellation of the epistatic effects.

The results reported by Hayes and Johnson are more directly comparable
with Jenkins' group 2. Various characters of the inbred parent were corre-
lated with the yield of their topcross progeny. The correlations for individual
characters ranged from .19 to .54, and the multiple r for 12 characters of the
inbred parent and yield of the topcross progeny was .67.

As a result of these studies some investigators have decided that the cor-
relations were too low to provide a wholly satisfactory basis for prediction,
and the only safe measure of the worth of an inbred line was to evaluate it in
hybrid combinations.

EARLY TESTING

Since the characteristics of the inbred lines did not provide an adequate
index as to the value of a line, and since this value must be determined by
crossbred progeny tests, it seemed advisable to determine whether crossbred
performance could be evaluated at an earlier stage of inbreeding. Several
lines of reasoning suggested that this might be feasible and desirable. First
the ear-to-row tests with all of their limitations suggested that there were
marked differences in yielding ability between individual carefully selected
open-pollinated ears. The genotype of such high yielding ears was modified
or diluted in ear-to-row testing procedure, but the identity of these individual
ears could readily be maintained by self-pollination. Second, it appeared logi-
cal to assume that a potential ceiling was established for any derived line at
the time of the selfing of the So or F2 parent plant. This ceiling is established
by the genotype of the parent plant and the most desirable combination of
genes which can be isolated from this gene sample.

The small population commonly grown from each selfed ear, the hindrance
of linkage in preventing random recombination of genes, and the limited ef-
ficiency of visual selection would all operate to render the probability of
isolating this most desirable gene combination very unlikely. The effort ex-
pended in growing and continued inbreeding and selection of strains having
the less desirable genotype might represent a considerable waste. Third, if
facilities were limited, as they always are, greater progress might be achieved
by the early discarding of the less desirable genotypes and the growing of
larger progenies of the more desirable genotypes in the early generations of
selfing when variability and the efficiency of visual selection would be ex-
pected to be at a maximum.

Before these ideas could be put to a test, data were presented by Jenkins

EARLY TESTING AND RECURRENT SELECTION 403

(1935) which seemed to lend considerable support to the general ideas men-
tioned above. Remnant seed of 14 lines from the variety Lancaster and 14
lines from the variety lodent representing eight generations of selfing were
chosen in Jenkins' study. These 28 lines represented a random sam[)le of the
lines from these two varieties which had survived the eight generations of
inbreeding. Two sibs were chosen to represent each generation, one repre-
senting a selected ear in the direct line of descent and the second a discarded
sib. These 56 ears were grown ear-to-row, and pollen from 10-12 plants of
each line were mixed and applied to ten ear shoots of the tester variety Krug.

Due to variation in stands and the unfavorable season neither the sam-
pling of plants within a strain nor the topcross parent was as adequate as
planned. Only 12 of the lines originally chosen were represented in each of the
eight generations of selfing. The yield trials of the topcrossed progeny were
grown in 1932. Information on several important problems is presented in
this paper, but the items of most importance in the present discussion deal
with the performance of the lines after successive generations of selfing. In
the lodent series, represented by seven lines, the mean square associated
with generations was not significant. In the Lancaster series, represented
by five lines, the variation associated with generations was significant but
there was no indication of a consistent trend.

On the basis of these results Jenkins concluded that, "The inbred lines
acquired their individuality as parents of top crosses very early in the
inbreeding process and remained relatively stable thereafter." Since this
paper was published, several people have assumed that the stability men-
tioned by Jenkins was synonymous with homozygosity, and therefore experi-
ments demonstrating segregation in F2 or F3 were disproof of this stability.
However Jenkins took particular pains to point out that the stability he was
assuming did not arise from homozygosity, but was a sampling phenomenon.
This sampling stability, if confirmed, makes the early testing procedure
even more attractive, but stability is neither assumed nor required as a pre-
requisite for early testing.

Results from Early Testing
Experiments on early testing were started in Missouri in 1935. However
due to unfavorable seasons, no critical data were obtained until 1938. The
experiments were continued in Iowa in 1939 and subsequent years. The re-
sults of these studies were summarized in 1946 (Sprague, 1946). Some 167
selected So plants from a strain known as Stiff Stalk Synthetic were self pol-
linated and outcrossed to the double cross tester parent Iowa 13. The yields
of the test crosses ranged from 61.8 to 100.8 bushels per acre. Four of the
test crosses were significantly lower yielding than the synthetic parent, and
two were significantly higher yielding than the tester parent. The plants
chosen for selfing represented a carefully selected group on the basis of

404

G. F. SPRAGUE

phenotypic desirability. The wide range in topcross yields obtained is evi-
dence of the poor relation between phenotype and performance in hybrid
combinations.

The frequency distribution of topcross performance was subjected to two
types of samplings. In one sample the Si lines representing the best 10 per
cent of the population were grown, and individual plants again self pollinated
and outcrossed to the tester parent, Iowa 13. The distribution of the So and
Si topcrosses are illustrated in Figure 26.1. (The So topcross yields have been
adjusted to the So topcross level on the basis of the performance of the tester

50

40

o

o

lit

30

20

10

/

\

/

\

A

/

\ y

s
\

/

/

/

/

\

^v.

/

/

\

^

/

f -

o

/

/

f

5

e
z

:

\

V

\

V
\

/

^

/^

\

V

5>

/
/

e

\

^-^

\
\

62.5 67.5 72.5 775 825 87.5 92.5

YIELD IN BUSHELS PER ACRE

97.5

102.5

107.5

Fig. 26.1 — A comparison of the frequency distributions of 167 topcrosses of So plants {solid
line) with a series of topcrosses of Si plants {doited line), representing the highest j'ieiding

10% of the original So population.

parent, Iowa 13.) The distribution of the Si topcross yields clearly indicate
that the So plants exhibiting high combining ability transmitted this char-
acteristic to their Si progeny. Segregation within progenies was quite ap-
parent, indicating that opportunities for additional selection existed.

A group of twelve lines was chosen which provided a seriated sampling of
the frequency distribution of So topcross yields. These were grown in 100
plant progenies, and an attempt was made to self pollinate 25 of the better
plants in each progeny and to outcross these to the tester parent. Because of
differences in time of pollen shedding only 6 of the 12 lines chosen were
finally used (Table 26.1).

Significant differences in yielding ability were obtained within each of the
six Si families. The range in yield was of about the same magnitude in each
family, suggesting that the So plants having the highest test cross perform-

EARLY TESTING AND RECURRENT SELECTION

405

ance were no more heterozygous than the So plants having poor test cross per-
formance. The distributions arising from the four highest yielding families
were not significantly different, but were significantly different from the
distributions arising from the two lowest yielding families. These same general
types of results were obtained when stalk breaking was considered.

Finally, three of the lines arising from selected sample when in the S3 gen-
eration of inbreeding were compared with five standard lines. These eight
lines were crossed in all possible combinations and compared in rej)licatcd
yield trials (Table 26.2). The S3 lines, as a group, were superior to the stand-

TABLE 26.1

FREQUENCY DISTRIBUTION OF ACRE YIELDS IN BUSHELS FOR

20 Si TOPCROSSED PROGENIES DERIVED FROM

6 So LINES (SPRAGUE, 1946)

Family

Yield

Distribution

OF 1942 Acre Yields in Bushels

No.

1940

1942

87.5

92.5

97.5

102.5

107.5

112.5

SSS 278

100.8
92.9
92.9
82.5
73.5
64.9

105.9
104.6
102.2
103.3
94.1
97.3

3
2
6
6

5
9

5

8

10

5
3
5

9

7
3
8

3

SSS 295

SSS 393

1

2
1

SSS 130

1

SSS '27

4

1

8
5

SSS 407

TABLE 26.2

RELATIVE AVERAGE PERFORMANCE OF
STANDARD AND NEW S3 INBRED LINES
OF CORN BASED ON SINGLE CROSS YIELD
TRIALS (SPRAGUE, 1946)

Relative Performance
AS Measured by

Inbred
Design.\tion

Yield in Bu.
per Acre

Root
Lodging
Per Cent

Stalk
Breaking
Per Cent

L317

78.4

79.2

87.5
78.8

72.7

11.8
8.0
8.0
1.1
5.6

2.7

187-2

0.9

\VF9

0.7

38-11

1.2

Oh67A

2.1

Average

79.3

6.9

1.5

SSS211-.S00

SSS 278-161

SSS 507-193

86.7
81.0
89.1

3.2
1.2
3.6

0.5
1.5
0.8

Average

85.6

2.7

0.9

406 G. F. SPRAGUE

ard lines in yield, and in resistance to root lodging and stalk breaking. On the
basis of these results it was suggested that early testing might be a valuable
tool in a breeding program. However it was pointed out that the method
might be of limited value under some conditions. This warning has to some
extent been ignored and some have assumed that the early testing procedure
is useful at any stage of the breeding program and with any parental ma-
terial.

Additional trials of the early testing procedure have been conducted by
Dr. John Lonnquist (1950) of the Nebraska Station. In this experiment a
series of selected plants from a strain of Krug were self pollinated and out-
crossed to a series of plants of the same variety. When test cross performance
data were available two samplings were made. One consisted of the group of
lines exhibiting the highest topcross yields and the second group those
exhibiting the lowest topcross yields.

In each group in subsequent generations selection was practiced in both
directions. In the high group the phenotypically most desirable and least
desirable plants were self pollinated and outcrossed to the tester. In the low
group again the most and least desirable plants were selfed and outcrossed.
This plan had to be modified somewhat as inbreeding progressed, since seed
was not always obtained on the least desirable plants. The group actually
used were the least desirable plants which could be propagated. After each
test cross generation the selection of lines to be continued was based on
combining ability. The single cross WF9XM14 was substituted for Krug
as the tester parent after the original series of test crosses.

The results obtained during the first four selfed generations clearly indi-
cate that topcross combining ability can be readily modified by a combination
of selection and testing (Fig. 26.2). In the high group selected for high
combining ability, the average topcross yields of all lines represented in-
creased from 98.6 to 107.5 bushels. In the high group selected for low com-
bining ability after the Si yields decreased from 98.6 bushels to 93.3 bushels.
In the low group selected for high combining ability after the Si generation
yield increased from 85.9 to 94.0. Where selection was practiced for low
combining ability in each generation, yields decreased from85.9to77.9bushels.

Thus selecting for high combining ability for three additional generations
when the original lines exhibited poor combining ability produced S4 lines
which were not significantly different in combining ability from those of the
high group selected for a similar period for poor combining ability. Selection
in the low group therefore would be largely wasted effort. Continued selec-
tion and testing after the Si would be most profitable for only those lines
exhibiting the highest Si topcross combining ability.

Limitations of Early Testing
Three papers have been published which are somewhat critical of the value
of early testing. These will be reviewed briefly. Payne and Hayes (1949)

EARLY TESTING AND RECURRENT SELECTION

407

have presented data on a comparison of combining ability in F2 and F3 lines
of corn. On the basis of these comparisons they concluded that early testing
was of doubtful practical value. The material used in this study was 30
selfed ears from early segregates from the single cross A116XL317. Each of
the 30 selfed ears was grown ear-to-row and pollen from approximately 30

IK)

105

100-

THREE HIGH COMBINING S, LINES (MEAN)

-THREE LOW COMBINING S, LINES (MEAN)

\

\

60

75

\

^Jc—

r

GENERATIONS of INBREEDING

Fig. 26.2 — The effects of visual selection and testing for coml)ining ability during four

generations of selling in the variety Krug.

plants in each progeny was bulked and applied to the four inbreds chosen as
testers; A334, A357, A340, and A392.

In addition, five individual plants selected at random were also out-
crossed to the same four testers. The test crosses arising from the bulked
pollinations were considered as representing a random sample of the gametic
production of the individual Fo plants and the five individual test crosses as
samples of the F3 progenies. Adequate seed was obtained from 26 of the origi-
nal 30 families. Within the different tester groups correlations between F2
and F3 test cross means ranged from .51 to .76.

408

G. F. SPRAGUE

Payne and Hayes stated that:

The extent of relationship between the performance of F2 test crosses and of the per-
formance of their F3 progenies in test crosses leads the writers to conclude that in these
studies there was some doubt of the practical value of early testing for combining ability
as a means of selecting desirable sources of F3 lines. By a test however of relatively few
F3 lines it was possible to select F3 lines that seemed to be a desirable source for improving,
or substitution for certain inbred lines in Minhybrid 608.

It may be well to emphasize again that the only claim made for early test-
ing was that it enables the separation of a population into two groups on the
basis of combining ability. Also, continued selection in the more desirable
group will yield a larger number of high combining lines than will the less
desirable group or a random sample of lines selected solely on the basis of

TABLE 26.3

FREQUENCY DISTRIBUTIONS OF YIELD IN BUSHELS PER ACRE FOR

1 TO 3 F3 PROGENIES DERIVED FROM F. LINES OF A116 X L317 CROSSED

WITH 4 DIFFERENT TESTERS (AFTER PAYNE AND HAYES, 1949)

Tester

F2

Distribution of Fa Test Cross Yields
IN Bu. PER Acre

Number of
Test Crosses

P.^RENT

47.5

52.5

57.5

62.5

67.5

72.5

Yielding 60.0
Bu. OR More

A334

Higher 50%
Lower 50%

Higher 50%
Lower 50%

Higher 50%
Lower 50%

Higher 50%
Lower 50%

3
4

10
11

2

5

5
9

5

7

14
2

1
9

16
10

9
3

5
1

10
6

7
6

7

6

2

4

1

21

A340

A357

3
"3"

1

3
9

10

19
3

17

A392

1

"1

2

1
5

17

27
17

phenotype. The frequency distributions of test cross combining ability for F2
and Fg progenies seem to fulfill this claim very nicely. In the table that fol-
lows, each F2 distribution has been divided into the higher yielding 50 per
cent and the lower yielding 50 per cent. The distribution of F3 test crosses for
each of these subgroups was taken from their paper. The results are pre-
sented in Table 26.3.

The writer would conclude from these distributions that the testing of F2
would have been a desirable practice. Within each test cross series it would
have permitted of the discarding of a considerable number of lines. If the
number of F.s's to be tested had been held constant and all of the lines to be
tested derived from the higher yielding F2 subgroup, even greater progress
might well have been expected.

The results obtained in this study are exactly those to be expected under
the postulates of early testing. Early testing obviously cannot be used as a

EARLY TESTING AND RECURRENT SELECTION

409

substitute for the more refined tests possible when the lines are more nearly
homozygous. This limitation was clearly outlined in the 1946 paper (Sprague,
1946).

Data have also been presented by Singleton and Nelson (1945) which
they interpret as demonstrating the inefifectiveness of early testing. In the
stud}' reported, forty-eight ears were chosen from the variety Whipple early
yellow. These were grown ear-to-row and one self made within each lot. The
selfed plants were also outcrossed to the inbred line P39. Selfing was continued
for three generations. In each generation the plants chosen for selfing were
outcrossed to P39. At the end of this period of selfing and testing, ten lines
were chosen for this special study. By using remnant seed, test crosses were

TABLE 26.4

ANALYSIS OF VARIANCE FOR YIELD. 1940 AND 1941
(AFTER SINGLETON AND NELSON, 1945)

Source of Variation

DF

MS

F

Blocks

Years

8
1
9
9

72

1194.81

275362.56

230.49

722.66

411.08

2.91*
66 99**

Varieties

56

VarietiesX Years

1 76

Var.XBlks.XYrs

Generations

Linear

3
1

1

1
3

27

27

238

2163.57
4586.95
1170.96
732.79
296.04
885.76
212.88
150.80

14.35**
30 42**

Quad

7.75*

Cubic

4 86

Generations X Years

1.96

Generations XVar

Gen.XVar.XYrs

Error Term

5.87**
1.41

produced involving So, Si, S2, and S3 generations. No data are given in the
publication but an analysis of variance for the two-year test period is pre-
sented in Table 26.4.

At least two points concerning the analysis are worthy of mention. First
there were no significant differences among the ten lines studied. In view of
the extensive testing back of the group of lines chosen, and because they \\ere
selected to be very similar in yield, it is not surprising that the early testing
procedure failed to disclose differences. The early testing procedure is cer-
tainly not suited to the measurement of very small differences. However the
degree of genetic uniformity with respect to combining ability would normal-
ly not be expected in sampling with open-pollinated or Fo populations.

The second comment bears on their interpretation of improvement in
combining ability during the course of inbreeding. The appropriate test of
significance in this case depends upon the specific question the data are
asked to answer. If conclusions are to be confined to the particular lines used,

410 G. F. SPRAGUE

then the variation associated with generations is correctly judged significant.
If, however, the experimental material is assumed to represent a random
sample of lines and therefore typical of lines in general, the appropriate test
indicates generations to be non-significant. Since no yield data were presented,
no test of significance can be calculated for the linear component of genera-
tions. Their results, as presented, have little bearing on either early testing
or the effectiveness of visual selection in modifying combining abiUty during
the course of inbreeding.

Richey (1945, 1947) has presented a re-analysis of Jenkins' (1935) data on
combining ability after successive generations of inbreeding and reached
conclusions differing from those presented by Jenkins. He questions the
stability of combining ability and the effectiveness of early testing in provid-
ing a satisfactory criterion of combining ability when the lines approach
homozygosity. He also presents some information on tester parents and their
effectiveness in revealing segregation. This latter is a very important prob-
lem but will not be discussed here.

We return to the first criticism raised by Richey, namely that lines do not
reach stability early in the course of inbreeding. To demonstrate his ideas,
Richey has combined the eight generations into pairs, thus providing four
groups. Then by selecting certain inbreds he has shown by graphs that,
visually, quite different slopes are obtained over the period under study.
Other groupings than those used by Richey may be selected with equal
validity. These different groupings show quite an array of slopes upon visual
inspection. However if one extends the original analysis of variance pre-
sented by Jenkins separating generations into a linear and remainder com-
ponent, the linear component is not significant. This, of course, does not
prove that trends are absent. It does indicate that such trends as may exist
are small in comparison with the random variation.

Richey 's second criticism deals with the effectiveness of early testing as a
measure of combining ability as the lines approach homozygosity. He con-
cludes that early testing would have been quite ineffective. The real basis
for the evaluation of any breeding or testing system depends upon the lines
which are produced or revealed which have sufficient value for use in com-
mercial hybrids. Of the twenty-seven lines on which Jenkins presented data,
two lines of the Lancaster series have been of sufficient value to be used ex-
tensively. These are L289 and L317. These two ranked in the upper half of
the lines tested and would have saved under an early testing procedure.

Two other lines have been used to a limited extent. One of these, 1224,
exhibited the highest yields in the lodent Si test cross series and would cer-
tainly have been saved. The other line L304A was in the upper 50 per cent of
the Si Lancaster series. If early testing had been used with this material,
saving the upper 50 per cent of each frequency distribution, no commercially
useful lines would have been discarded. The early discarding of the remain-

EARLY TESTING AND RECURRENT SELECTION 411

ing lines would have resulted in a very great saving of time and nionejy- us
compared with testing at a more advanced stage of inbreeding.

RECURRENT SELECTION

Superficially recurrent selection has a considerable resemblance to the
ear-to-row method of breeding. However recurrent selection differs in several
important respects. It provides for a much more accurate genetic control, and
the plot technic can be modified to give any desired degree of accuracy. Our
use of the recurrent selection technic was a direct outgrowth of the work on
early testing. It appeared logical to assume that if the individual So plants
selected on the basis of test cross performance were a superior group, inter-
crosses among this group to provide source material for a new cycle of selec-
tion would minimize certain of the limitations arising from continued
selfing. Accordingly a group of the best lines from the early testing series
were intercrossed to provide material for the evaluation of this method.

Somewhat earlier, studies were started to compare the relative efficiency
of recurrent selection and inbreeding in isolating material having a high oil
percentage. At the time the work was started we were of the opinion that
this was a new idea. It was some time later that we discovered that essential-
ly the same ideas had been published independently by East and Jones (1920)
and by Hayes and Garber (1919). In neither of these cases was any extensive
use made of the method and no critical data were published. The first de-
tailed description of recurrent selection was made by Jenkins (1940). The
breeding procedure did not receive a name however until Hull (1945) pub-
lished his article dealing with recurrent selection for specific combining abil-
ity.

Because of the shorter time period required per cycle we have much more
information on recurrent selection as a method for modifying chemical
composition than we have for the modification of combining ability (Sprague
and Brimhall, 1949). We shall report in some detail only one study — that
contrasting recurrent selection and inbreeding in modifying oil percentage in
corn. The source material for this study was obtained from Si ears from re-
ciprocal backcrosses involving the single cross wxOs420Xlll. High Oil. Indi-
vidual plants were self-pollinated in each backcross population and analyzed
individually for oil percentage in the grain. The five ears having the highest
oil percentage in each population w^ere planted ear-to-row the following
season and all possible intercrosses made among the ten progenies. Equal
quantities of seed from each cross were bulked and used as source material
for a new cycle of selfing, analyzing, and intercrossing.

A duplicate planting of the ten ears mentioned above was made in 25
plant, ear-row progenies. The phenotypically most desirable plants in each
progeny were self-pollinated. At harvest time approximately five ears were
saved and analyzed individually for oil content of the grain. The two ears

412 G. F. SPRAGUE

of each family having the highest oil percentage were again grown in progeny
rows for continued inbreeding and selection. When the analyses were avail-
able the sibling progeny having the lowest average oil percentage was dis-
carded. The two selfed ears having the highest oil percentage in the selected
sibling were used to propagate the family. This process was continued
through five generations. The general procedures used in selection, with the
exception of the chemical analyses, are essentially those commonlv employed
in the development of inbred lines by the standard method.

It should be emphasized that the time requirement, number of pollina-
tions and analyses, land requirements, and selection differentials were es-
sentially the same for the recurrent and the selfing series. The relative
efficiencies of the two methods therefore should be directly comparable.

RECURRENT SERIES

The results from the recurrent series will be presented first. The material
from the 111. High OilXw^xOs420 series has been carried through two cycles
after the original selfings. The frequency distributions are shown in Figure
26.3. The distribution presented for the original population is a composite
for the two backcrossed populations. The solid vertical line represents the
population mean and the dotted vertical line the mean of the selected sample.
These selected ears were grown in ear-row progenies the following year and
all possible intercrosses made by hand. Bulked seed from these intercrosses
provided the source material for the next cycle of selfing and selection. The
mean of the first cycle population was essentially the same as the mean of
the selected parents — the full selective advantage of the parents had been
retained. In the second cycle population the mean was further shifted to the
right by an amount equal to 2.1 class intervals, but still failed to equal the
mean of the selected parents by an amount equal to 1.1 class intervals.
The mean of the original population was 7.2 per cent of oil. The mean of
the second cycle population was 10.5 with the extreme deviate at 13.5.

The ranges and standard deviations of these three populations are of some
interest in indicating any changes in genetic variability. Considering first
the range: in the original population the range was from 4.5 to 10.5, in the
first cycle 5.5 to 12.5, and in the second cycle 7.5 to 13.5 — a difference of 6,
7, and 6 class inter\^als respectively. The first cycle had the greatest, the
original population intermediate, and the second cycle the smallest standard
deviations. The fact that the second cycle exhibited the smallest standard
deviation may indicate some loss in genetic variability. However 65 per
cent of the selective advantage of the parents was retained indicating that a
considerable amount of genetic variability exists.

SELFING SERIES
The selfing series presents a strikingly different picture. The results are
presented graphically in Figure 26.4. The values plotted for the Si generation

EARLY TESTING AND RECURRENT SELECTION 413

represent the oil percentages of the original selfed ears. Two lines were lost
during the course of inbreeding because of failure to produce any pheno-
typically desirable plants. The eight lines remaining however represent eight
of the ten lines comprising the recurrent selection series. The values pre-
sented for the So generation represent the mean of all ears of a particular
family which were analyzed. In S3 to S5 the value plotted represents the

60

30

60

0

> 30

0 O
6

60

30

/«'• Cyc/e

^'^ Cycle

j.o 40 SO ttO 70 eo 9.0 zoo no
Oil Parcznfogc In Groin

12 O 13.0

Fig. 26.3 — .\ comparison of the frequency distributions of oil percentage in the corn kernel,
in the original population, Illinois High Oil X wxOs420, and after one and two cycles of

recurrent selection.

mean for the sibling population in the direct line of descent. If the highest
values in each generation had been plotted instead of the means, the picture
would have been essentially the same except that the fluctuation from gen-
eration to generation would have been increased. The eight lines exhibited
somewhat different patterns during the course of inbreeding. Si.x of the eight
lines exhibited an increase, and two a decrease in oil j)ercentage. There does
not appear to be any consistent trend within the families from generation to
generation. It would appear that chance has played a very important role in
spite of the intensive selection practiced .

414

G. F. SPRAGUE

Comparisons between the two systems of breeding may be made in a num-
ber of ways. Selection during inbreeding is normally practiced within and
among families. If only the two families having the highest oil percentage
were retained and these compared with the mean of the second cycle popula-
tion, the differences are very slight but in favor of the selfing series. If these

S2 S3 S4 S5

Gana fO'Hona of I nbnaedinq

Fig. 26.4 — A comparison of mean oil percentages in the corn kernel from the recipro-
cal backcrossed Illinois High Oil X wxOs420 during five generations of inbreeding and

selection.

two lines are compared with the extreme deviate of the recurrent series the
lines are lower in oil by nearly three per cent. If the comparison is made
between the mean of the S5 lines and the mean of the second cycle population
the lines are again lower, the contrast being 7.5 and 10.5 per cent of oil
respectively.

Any comparison involving these two series must also take into account
the time at which the comparisons were made. In the selfing series, genetic

EARLY TESTING AND RECURRENT SELECTION 415

variability, and therefore opportunity for selection, would be largely ex-
hausted after five generations of selfing. For reasons mentioned earlier, it is
assumed that a considerable degree of genetic variability remains in the re-
current series. The disparity between the two systems would therefore be
expected to increase with additional generations of selfing and cycles of selec-
tion.

Recurrent selection has been practiced for oil percentage in two additional
populations. One series had its origin in an F2 population of the single cross
I198XHy. This population started with a much lower average oil percentage,
but the effectiveness of selection was essentially the same as in the 111.
High OilXwxOs420 series.

In a third series a strain known as Stiff Stalk Synthetic served as parental
material. This material also has been divided into a selfing and a recurrent
series to supplement the material already presented. This experiment has not
yet been completed. The difference between the two series, in so far as data
are available, closely parallels the wxOs420Xlll. High Oil series already dis-
cussed.

Data on the effectiveness of recurrent selection in modifying combining
ability are still quite limited. One such comparison is shown in Figure 26.5.
The original stock used was the Stiff Stalk Synthetic, and the double cross
Iowa 13 was used as the tester parent. The yields for the two years were not
greatly different, but to facilitate a direct comparison the lower frequency
distribution has been displaced to the right so that the yield of Iowa 13 for
the two years falls on the same ordinate. Stands were somewhat variable in
the test crosses comprising the first cycle. The effect of this variation was
minimized by adjusting all yields to an average stand by means of a covari-
ance analysis. This adjustment reduced the range in yields so that the con-
trast between the two frequency distributions does not necessarily present a
true picture of the relative variation in the two populations.

RECIPROCAL RECURRENT SELECTION

A modification of the recurrent selection scheme has been suggested by
Comstock et al. They have designated this procedure reciprocal recurrent
selection. Under this modification two diverse foundation sources, A and B,
are to be used. Individual selected plants in A are self-pollinated and out-
crossed to source 5 as a tester parent. Similarly selected plants from source
B are self-pollinated and outcrossed to source .1 as a tester. When test cross
data become available, a group of selfed ears from source A having the best
test cross performance are recombined to produce ^1^. AB^ population is
formed in a similar manner. A^ and B^ then serve as source material for a
new cycle of selfing and test cross evaluation followed by the intercrossing
of the most desirable plants. No data are yet available from either their
experiments or ours using this method.

416

G. F. SPRAGUE

In the original paper by Comstock et al. (1949) a comparison is presented
of improvement limits of three definite breeding procedures. These were (1)
selection based on general combining ability using at least two single crosses
as testers, (2) recurrent selection for specific combining ability as proposed
by Hull (1945), and (3) reciprocal recurrent selections. The assumptions on

75

60

ORIGINAL POPULATION

62.5 675 72.5 775 82.5 87 5 92.5 97.5 102.5

YIELD IN BUSHELS PER ACRE

75

60

45

30

15

IOWA 13

QL —

FIRST CYCLE

625 675 72.5 77.5 825 87.5

YIELD IN BUSHELS PER ACRE

92.5

97.5

Fig. 26.5 — A comparison of the frequency distributions for yield in bushels per acre for top
crosses from the original Stiff Stalk Synthetic and after one cycle of recurrent selection.

which these comparisons were based were stated by Comstock et al. and
will not be repeated here. The conclusions reached are briefly as follows:

1. When dominance is incomplete methods 1 and 3 are essentially equal
and superior to method 2.

2. If over-dominance is of major importance methods 2 and 3 will be
essentially the same and superior to method 1.

3. When dominance is complete all three methods would be rather
similar.

Thus method 3, reciprocal recurrent selection, would appear to be the

EARLY TESTING AND RECURRENT SELECTION 417

safest and most efficient method to use with our j)resent lack of knowledge
concerning the relative imjiortance of partial dominance, dominance, and
over-dominance in determining combining ability.

In the discussion presented so far no emphasis has been placed upon choice
of testers. It is obvious that either early testing or recurrent selection can be
carried out giving special emphasis to either general or specific combining
ability depending upon the tester parent chosen. In the experiments involv-
ing oil percentage of the grain this problem does not arise. In the e.xperiments
involving test crosses for yield evaluation, double crosses or open-pollinated
varieties have been used as tester parents thus giving special emphasis to
general combining ability.

SUMMARY

In the data which have been presented bearing on early testing, the
method has demonstrated all of the characteristics which have been claimed
for it. This is not to be interpreted as meaning early testing is the ideal corn
breeding method and equally applicable under all circumstances. It is useful
under some conditions. The ideal method of corn breeding probably is still
to be devised.

Recurrent selection has been found to be quite effective in modifying the
chemical composition of the corn grain. Tests of this method in modifying
combining ability have been less extensive. Here again this method may not
be equally valuable under all conditions and circumstances, but on the basis
of results to date it is certainly deserving of more extensive use.

E. J. WELLHAUSEN

Rockefeller Foundation, Mexico City

Chapter 27

Heterosis in a New Population

Data recently presented by Mangelsdorf and Smith (1949) indicate that corn
was being grown in what is now southwestern United States and Mexico at
least four thousand years ago. The corn grown in these prehistoric times was
both a pod corn and a pop corn of relatively low yield capacity. Today in
this same area an enormous variation exists. Direct derivatives of the ancient
low yielding pod-pop type still can be found on a very limited scale in certain
areas of Mexico, but these low yielding ancient corns now have been replaced
largely by more vigorous and productive types.

Tremendous changes have been brought about in both type and yield
capacity since ancient times. The modern varieties of Mexico have a yield
capacity many times more than the ancient types. On the high plateau of
Mexico a variety known as Chalqueno, whose pedigree in part can be traced
back to an ancient pop corn, has yielded up to 125 bushels per acre. If the
various evolutionary processes and the kinds of gene actions involved in the
development of such high yielding varieties from the low yielding prehistoric
types were known, we would certainly have a better understanding of the
phenomenon of heterosis.

It is the purpose of this chapter to present, first, what seems to have been
involved in the development of the modern, relatively high yielding varieties
over a period of about four thousand years; and second, a discussion of the
methods used and results obtained in the further improvement of some of the
modern varieties in a short period of six years.

Perhaps the title might best have been "Heterosis in an Old Population"
in the sense that the Mexican corns as a whole are much older than those in
the United States. However, from the standpoint of modern corn breeding,
it is a new population in that it involves new material on which to try the
modern techniques of corn breeding developed in the United States. The suc-

418

HETEROSIS IN A NEW POPULATION 419

cesses and failures of standard techniques in this new i)opulation, together
with certain modifications that are being tried, will be discussed.

HETEROSIS IN NATIVE OPEN-POLLINATED VARIETIES

The first obvious step in any breeding program in a new area is adequate
testing of the varieties at hand. In the early years of the program, therefore,
considerable time was devoted to a study and classification of the ])resent-day
varieties in Mexico (Wellhausen el al., in collaboration with Mangelsdorf,
1951). Evidence presented in this report strongly indicates that many factors
have been involved in the evolution of corn in Mexico, the most important
of which are repeated here as follows:

1. Varieties in the ancient pod-pop corn type were probably at first chiefly
brought about through mutation and by a partial release from the pressure
of natural selection by man. There are four ancient races in Mexico which
definitely trace back to a common parent. Where this common parent origi-
nated is still unknown. All have a sufficient number of different characters
to warrant their classification as separate races, yet they all have a number of
characters in common; namely, all are pop corns, two of the four are pod
corns, all are early maturing, all have a low chromosome knob number, and
all are relatively low in yield capacity compared to modern varietal stand-
ards. Since no record of the common ancestor is available, no direct compari-
sons can be made of the yield capacities of the ancient indigenous races as
they exist today in Mexico and of their common ancestor. Judging from the
Bat Cave material (Mangelsdorf and Smith, 1949) it is not at all unlikely
that considerable increase in yield capacity was brought about through gene
mutation alone.

2. It is distinctly evident from a study of the various collections that some-
time during the history of the Mexican corns there was an influx of e.xotic
types from countries to the south. As a result of the introgression of the an-
cient indigenous types into the exotic types, and vice versa, many new varie-
ties and races came into existence.

3. Superimposed upon the above two evolutionary mechanisms was the
introgression of teosinte germplasm. If Mangelsdorf and Reeves (1939) are
right in their theory that teosinte originated as a cross between corn and
Tripsacum, then this teosinte germplasm is largely Tripsacum germplasm.
Practically all the modern more-productive types of corn contain some teo-
sinte germplasm.

4. The fourth important factor in the evolution of corn in Mexico has been
the geography of Mexico itself. Mexico is a mountainous country with many
different climates and geographically isolated valleys. Corn is grown from
sea-level up to 10,000 feet elevation under a wide range of rainfall conditions.
In some areas rainfall is limited to five to ten inches for a i)eriod of four
months. Other areas receive up to 100 or more inches in a period of six to ten

420 E. J. WELLHAUSEN

months. Different temperatures due to changes in elevation and different
amounts of rainfall may occur in areas separated only by a few miles. Such
conditions are conducive to the development of many different varieties of
corn.

As a result of the above evolutionary factors operating over a period of at
least four thousand years, there is a greater variation in the corns of Mexico
today than in any country in the world. Without doubt the greatest single
factor in the development of the modern high yielding agricultural types in
Mexico has been the introduction of exotic types from the south. These exotic
types were largely big-grained flour corns, which no doubt brought in a series
of genes for higher yield that had not existed in Mexico before.

The various processes and types of gene action involved in the develop-
ment of higher yielding varieties from the reciprocal introgression betwieen
the indigenous and exotic types, plus introgression of teosinte, are not easily
explained.

Gene Combinations

These processes probably involved a gradual sifting of the gene combina-
tions brought together by hybridization, and continuous backcrossing or re-
hybridization of resulting hybrids or segregants. The complex pedigree of
some of the modern high yielding varieties in Mexico, taken from Wellhausen
elal.m collaboration with Mangelsdorf (1951), are shown in Figures 27.1-27.4.
In these pedigrees each product of the indicated hybridization between two
different races, or species in the case of teosinte, was higher yielding or better
adapted to its native habitat than either one of the putative parents. For ex-
ample, in Figure 27.1 Conico is a better corn than either Palomero Toluqueno
or Cacahuacintle, and Tuxpeno is a more productive corn than either Olotillo
or Tepecintle. Chalqueno, which is somewhat more recent in the evolutionary
scale, is more productive than either Conico or Tuxpeno.

This does not necessarily mean that the same races crossed today would
all show considerable heterosis in Fi. As a matter of fact many of the crosses
indicated in the diagrams have been made and studied. In certain cases the
Fi hybrid, when tested in the environment best suited to one or both parents,
showed considerable heterosis. In other cases it was no better than the better
parent or was intermediate between the two parents. In some crosses the Fi
was definitely unadapted to the environment of either parent.

In the natural development of higher yielding corns from the intercrossing
of different races, there were no doubt many instances in which the Fi hy-
brids that first occurred between a native and an introduced variety were
very poorly adapted to native conditions and showed no heterosis. A 50 per
cent random dosage of an introduced variety is often more than sufficient to
completely upset the physiology of a native variety that has adapted itself
to a fixed environment over a long period of natural selection. Under natural
conditions, however, any crossing that might take place between two varie-

HETEROSIS IN A NEW POPULATION

421

ties is purely at random and not complete. Hybrid plants that appear in a
field of native corn in the succeeding generation, therefore, might be widely
scattered. But no matter how little seed these Fi plants may produce, if their
flowering periods coincide, then germplasm would be passed on to the native
variety through backcrosses.

Thus by repeated backcrossing and the sifting action that always takes
place through natural and artificial selection, certain genes from an intro-
duced population may be readily transferred to a native population. These
might be additional favorable yield genes that express themselves in the na-
tive gene complex, or they might be other genes which permit the fuller ex-
pression of the yield genes which the native variety already contains, or both.

r CON ICO

CHALQUENO <

PALOMCRO TOLUQUE/^0

CACAHUACINTLE

TUXPENO

OLOTILLO

r HARINOSO FLEXIBLE

^TEOCINTLE

■HARINOSO DE GUAT?

L TEPECINTLE ^

L TEOCINTLE

Fig. 27.1 — Prol)ahle origin of Chalqueno.

p CONICO

CONICO NORTE NO <

PALOMERO TOLUauENO

CACAHUACINTLt

CELAYA

r HARINOSO n EX ISL f

OLOTILLO

«

TUMPENO

TEOCINTLE
r HARINOSO DE CUAT ?

TEPECINTLE

"

■ TEOCINTLE
rCHAPALOTE

rREVENTADOR

*

TABLONCILLO

TEOCINTLE

HARINOSO DE

OCHO

Fig. 27.2 — Probable origin of Conico Norleno.

422

E. J. WELLHAUSEN

A classical example of such introgression is the introgression of genes from
teosinte into corn, a process which is still taking place in many areas of
Mexico. Teosinte grows as a weed in the corn fields of certain areas. Also the
Mexican farmers in some areas have been known to plant teosinte in their
corn fields based on a belief that such a practice would make their native
corns more drought resistant. The Fi hybrids between corn and teosinte are
very small-eared, and ears are difficult to collect because of their verj^ brittle
rachis. As such, the F/s have no value in artificial selection. However, the Fi's
shed pollen about the same time as the native corn variety, and a large num-
ber of backcrosses result with the corn parent as the female. Some of these are
unconsciously selected as seed for the following year since they cannot be

(NAL - TCL

BOL I TA -

ZAMLore CHicc

^T£PECINrL£

■ HARINOSO Oe GUAT ?

■TEOCINTU

REVENTADOR

CHAPALOre

I TEOCINTLE

^TABLONCILLO

^HARINOSO DC OCHO

Fig. 27.3 — Probable origin of Bolita.

[OLOTILLO

r TUXPCNO

VANDENO •

^TEPECINTLE

ZAPALOTE GRANDE .

ZAPALOTE CHICO

rHARINOSO FLEXIBLE

\-TEOCWrLE

tHARINOSO DE GUAT. ?

TEOCINTLE

,NAL -TEL

■ TEPECINTLE

HARINOSO DE GUAT. ?

TEOCINTLE

HARINOSO DE GUAT.

Fig. 27.4 — Probable origin of Vandeiio.

HETEROSIS IN A NEW POPULATION 423

separated at time of harvest from the non-hybrid grains. These backcrosses
then, bring about a second generation of backcrosses.

Through this sifting action certain genes from teosinte, such as those that
condition greater drought resistance when in combination with the native
corn gene-complex, become fairly well established. In a dry area any genes
bringing about greater drought resistance would immediately affect the
yield, and gene frequency for such characters would increase in the popula-
tion through subsequent natural and artificial selection. Thus a new, im-
proved higher yielding population under dry conditions may be brought
about which will rep)lace all other populations in its range of best adaptation.
It must have been in this way that the old superstition of a greater protection
from drought by interplanting corn with teosinte arose. How many other
factors of survival value were obtained from teosinte is difficult to ascertain.

This is probably the most common manner in which higher yielding varie-
ties for specific areas, especially the old corn areas, were built up. However,
as old land wore out, new methods of corn cultivation and new areas in which
corn could be produced were constantly sought. New environments for corn
production thus came into existence throughout the years. These new envi-
ronments often consisted of the artificial or natural drainage of old lake beds
which brought into cultivation highly fertile areas with high water-holding
capacity. In such areas, due to reserve moisture in the soil from the previous
rainy season, corn could be planted as much as two months ahead of the be-
ginning of the normal rainy season. This provided a six or eight month grow-
ing season instead of the usual four to six months. It was in such new environ-
ments, where plants could develop and fully express their yield capacity, that
certain Fi hybrids presented a much higher degree of heterosis and adapta-
tion than in the native habitat of either one of their immediate parents.

Hybridization

Perhaps the outstanding example of a highly productive hybrid race that
has developed in a relatively new environment in the central high plateau of
Mexico is a late maturing race called Chalqueno. This race (Fig. 27.1)
probably came into existence through the hybridization of the two distinct
races, Conico and Tuxpeno. Conico is an early maturing corn that originated
on the high plateaus of Central Mexico from the hybridization of an ancient
indigenous high altitude pop corn called Palomero Toluqueno and an exotic
race called Cacahuacintle. Tuxpeno, the other parent of Chalquefio, is a
cylindrical dent adapted to the lowland coastal areas of Mexico. It probably
came into existence through the intercrossing of two prehistoric races,
Olotillo and Tepecintle, which in turn probably were derived from two differ-
ent exotic flour corns through the introgression of teosinte.

Chalqueno, in the areas where it is grown today, is much more productive
than either of its putative parents Conico and Tuxpeno. In the highland re-

424 E. J. WELLHAUSEN

gions to which Conico is best adapted, neither Chalqueno nor Tuxpeno will
mature, and in addition Tuxpeno is very heavily attacked by rust. On the
other hand, in the lowland areas where Tuxpeno is best adapted, Chalqueno
produces very little and Conico produces practically nothing. A similar rela-
tionship has held with an artificial cross between Conico and Tuxpeno in
which half the germplasm was from one parent and half from the other. The
artificial hybrid, as one might expect, was not as well adapted to the Chal-
queno area as Chalqueno itself. Chalqueno over a period of many years of
gene sifting no doubt concentrated those favorable growth genes best suited
to its present environment and practically eliminated the frequency of other
genes which were unnecessary or deleterious to its maximum development.
The hybrid, nevertheless, was much superior to the two parents in the envi-
ronment best suited to Chalqueno.

Here then is an example of the process which often happens in nature or
in planned breeding programs. A hybrid between two parents adapted to
widely different environments shows no hybrid vigor, as measured in yield,
over either one of the two parents when grown in the native habitat of either
parent. But in a new environment different from the one under which either
parent developed, the hybrid may show extreme vigor. It is highly probable
that many new varieties came into existence when new areas of land were
brought under cultivation or when the native inhabitants migrated to new
areas. Very often several different corns were brought together in these new
environments which were not as well adapted as the hybrids that resulted
between them. It is also highly probable that the first varietal hybrids were
not as good on the whole as the varieties that developed from them through
successive generations of backcrossing and gene sifting.

From a study of the origin and development of the various types of corn
in Mexico, it seems that the most important factor in the evolution of the dif-
ferent productive races has been the gradual accumulation of favorable
growth or yield genes in combination or balance with the proper "governing
or regulating" genes in each specific environment. Maximum yield in a spe-
cific environment is not only dependent on favorable genes for such quantita-
tive characters as ear and kernel size, number of ears, leaf area or photosyn-
thetic efficiency, but also on genes which govern such functions as maturity,
disease and drought resistance, or general adaptation.

The latter group might well be genes which some investigators have
termed botlleneck genes. Such genes may inhibit the full expression of certain
quantitative yield genes and thus prohibit the organism from reaching its
maximum production allowed by a specific environment. An increase in yield
capacity, therefore, often may not involve the accumulation of more favor-
able yield genes, but rather the removal of certain bottleneck genes. In new
environments these bottleneck genes might be relic genes carried over from
their old native habitat where they existed because they had survival value.

HETEROSIS IN A NEW POPULATION 425

It would seem, therefore, that heterosis is the result of the combined effect
of individual gene action (for growth) plus the interaction of all genes within
the genotype (its genetic environment) in relation to the sum-total of all ex-
ternal influences acting upon the organism and governing expression of its
gene complex. If heterosis in a cross between A X Bis measured as the excess
vigor or yield over the average of A and B, there may be no heterosis in the
environment to which ,4 or B is best adapted as shown with Chalqueno. Yet
in some new^ environment different from that in which A and B developed,
the excess vigor may be great, even exceeding that of A and B in their native
environment, if such comparison could be made. Certainly the genotype is
no different in any of the areas. The difference must be due to different inter-
actions between over-all gene action and environment. It is no wonder that so
many different ideas exist when it comes to the explanation of heterosis. It
has no simple explanation.

IMPROVEMENT THROUGH BREEDING

In the evolution of corn in Mexico, the different varieties and races were
brought together in a haphazard manner. Relatively few of the total combi-
nations of races and varieties possible have been made, and when two varie-
ties or races come together in a specific region by chance, there is no reason
to believe that the particular combination was the best that could have been
made for the area. Although some fairly productive varieties did develop,
especially in the more fertile areas with higher rainfall, the possibilities of
further over-all improvement are astonishing and offer a challenge to modern
corn breeders.

The cooperative corn improvement program of the Mexican Government
and the Rockefeller Foundation was begun about six years ago. Its objective
was to provide higher yielding varieties or hybrids for the many different en-
vironments in the main corn producing areas. In most of Mexico, selection
has been going on for many centuries for adaptation to low soil fertility and
extreme climatic conditions. A variety or strain capable of producing some-
thing in the years of extreme drought or early frost was highly prized by the
native Indians, even though it produced only a little more in good years. Low
yields meant hunger, but a crop failure meant starvation. With modern
means of transportation, crop failures in a region no longer mean starvation
for the people of that region, and low fertility can and must be remedied be-
fore corn production can be greatly increased. The breeding program, there-
fore, was geared to the development of productive varieties or hybrids adapt-
ed to the average climatic conditions of a particular region, and a level of
fertility that could permit maximum production under these average condi-
tions. More productive varieties would pay the cost of soil improvement and
pave the way for a generally higher level of corn production.

The program also is based on a gradual improvement year by year. As

426 E. J. WELLHAUSEN

soon as a variety or hybrid was found to be superior to native varieties in a
particular area, it was increased and distributed, although it still may have
had many minor defects. When better varieties or hybrids become available,
they are substituted for those previously released in the increase and distribu-
tion program.

The different steps involved in this gradual improvement program for any
particular area have been as follows:

1. Variety testing. In this way good open-pollinated varieties were some-
times isolated for immediate distribution and as basic material for the breed-
ing program to follow.

2. The improvement of the best native open -pollinated varieties through
the formation of synthetics which could be propagated through open-pollina-
tion.

3. The formation and distribution of double cross hybrids which were good
not only as double crosses but also as synthetics in advanced generations. For
this purpose one generation selfed lines were used.

4. Finally, after many of the farmers have learned how to use hybrid corn,
greater emphasis may be devoted to the formation of more specific, higher
yielding uniform hybrids with highly selected and proven inbred lines.

Areas in Which Improvement Work Has Been Concentrated

The methods used and results obtained may be understood more clearly
if the areas in which improvement programs were initiated are identified.
Although corn is grown everywhere in Mexico, the most important commer-
cial corn growing areas are found on the central plateau between 18 and 22
degrees latitude. It is within this area that the breeding work has been con-
centrated. To facilitate the work still further, the area was divided into five
zones on the basis of elevation as follows:

Zone 1 — 2200-2600 meters elevation

a) Late varieties with six months' growing season planted under irrigation

b) Early varieties with four months' growing season under natural rainfall conditions
Zone 2—1800-2200 meters

Zone 3—1400-1800 meters
Zone 4—1000-1400 meters
Zone 5 — 0-1000 meters

The main breeding stations for these five zones are located in Zone 1 at
2200 meters, Zone 3 at 1600 meters, and Zone 4 at 1200 meters. With these
three main stations and with the cooperation of farmers in making yield tests
in outlying regions, it has been possible to cover the central plateau and cer-
tain tropical areas fairly completely.

Utilization of Good Native Open-pollinated Varieties

Since it is entirely possible that the original gene populations in the many
different isolated valleys of a particular zone were not the same, one might
e.xpect to find a different variety in each valley as a result of natural and arti-

HETEROSIS IN A NEW POPULATION 427

ficial selection. Often this was precisely the case. Out of 240 samples of corn
collected in Zones 2 and 3 from areas with elevations ranging from 1400 to
2000 meters, and tested under conditions representative of these two zones,
15 varieties were outstanding. These consisted of three early, five medium,
and seven late maturing varieties in relation to the normal growing or rainy
season of the area in which tests were made. The best early varieties (Well-
hausen, 1947) yielded from 15 to 49 per cent more than the average of 26
varieties of the same maturity. The best medium maturing varieties yielded
from 30 to 54 per cent higher than the average of 57 varieties of medium ma-
turity. The best late varieties yielded from 25 to 60 per cent more than the
average of the 62 late varieties of similar maturity included in the tests.

A similar situation was found to exist in the valleys at higher elevations
in the high plateau (Zone 1) among both the late and early varieties. A late
variety commonly grown under irrigation in the fertile valleys of the State
of Hidalgo was found to yield about 20 per cent more than a variety called
Chalco widely used for irrigation plantings in the States of Mexico and
Puebla (Wellhausen and Roberts, 1948). The best early varieties for Zone 1
were found in the State of Mexico, north of Mexico City, and these yielded
from 15 to 20 per cent more than the average early varieties grown in Zone 1 .
On the whole, from 15 to 20 per cent increase in yield often could be obtained
in some parts of all zones through the wider distribution of the best open-pol-
linated variety found within each zone.

Fundamental Methods in the Formation of Good Inbred Lines

Steps 2 to 4, as outlined above, imply the formation and use of inbred lines,
and the degree of improvement one might expect depends upon the isolation
of good, vigorous, disease-resistant lines that combine well with each other
when used in synthetics or hybrids. In the formation of lines we have, in gen-
eral, adhered to the following principles:

1. The use of diverse varieties adapted within a particular zone.

2. Rigid selection based on vigor and desirable agronomic characters.

3. Tests for combining ability after one generation of selfing.
Approximately 35 different varieties of corn belonging to eight different

races have been inbred in the improvement program for central Mexico.
These eight races are listed as follows in order of adaptation to elevation:

Zone to
Which
Race Adapted

Conico 1

Chalqueno 1

Conico Norteno 2

Celaya 3 and 4

Bolita 3 and 4

Tabloncillo 3 and 4

Vandeno 4 and 5

Tuxpeno 5

428 E. J. WELLHAUSEN

All are races which have come into existence in more recent times and all
have rather complex pedigrees as shown in Figures 27.1-27.4. So far only the
outstanding varieties of each race have been inbred. It was concluded early
in the breeding program that inbred lines for immediate use in a particular
zone might best be obtained from the high yielding varieties adapted to that
zone. It is entirely possible that in the future better hybrids may be obtained
through the utilization of a wider base of germplasm. The latter procedure
would require more time and more adequate testing. It would involve the ex-
traction of the favorable yield factors from several races and their conver-
gence into a synthetic variety or hybrid along with the proper governing genes
for a given environment.

In the inbreeding program it has become apparent that vigor in the origi-
nal plant or first generation selfed progeny (S]) can be used to a considerable
extent as a measure of the number of favorable vield genes with which the
particular plant or line has been endowed. But vigor becomes less and less
useful as an indicator of the number of yield genes in a particular line in the
second, third, or fourth generations of inbreeding because of a greater num-
ber of bottleneck genes that are fixed with successive generations of inbreed-
ing. In advanced inbred generations, two lines might be greatly different in
vigor, yet equal in combining ability with a specific tester, because of simi-
larities in their yield gene complex. As lines, they may be different in relative
vigor because of certain different specific bottleneck genes or loci that mask
the expression of genes or loci for growth and development.

The Mexican corn program not only involves the actual improvement of
corn, but also the development and training of young corn breeders. In order
to demonstrate more vividly that considerable effort may be saved by select-
ing only the best and most vigorous plants or lines for further work, several
students visually classified the inbred lines available after one generation of
selfing from five different varieties into four classes on basis of vigor. The
most vigorous lines were classified as A lines. Those somewhat less vigorous
or desirable were classified as B, and so on, with the least vigorous lines classi-
fied as Z).

Although classifications were originally made into four categories, the D
lines were discarded without further consideration. Of the remaining lines
saved from each of the five different varieties, 3-6 per cent in each were clas-
sified a.s A, 12 to 15 per cent as B, and the remaining 79 to 85 per cent as C.
All were topcrossed to a common tester, but only the topcrosses involving A
and B lines were finally included in yield tests because of lack of space. Com-
parative yields of the A and B topcrosses are summarized in Tables 27.1-27.3.

Table 27.1 shows the results obtained with A and B lines from a variety
known as Leon CrioUo topcrossed on the variety Urquiza and tested at three
locations. At each place the average yield of the topcrosses involving the A
lines was slightly higher than those involving B lines. But what is more im-

HETEROSIS IN A NEW POPULATION

429

portant, the percentage of lines increasing the yield over the tester parent was
much higher among the A lines than among the B lines.

Table 27.2 shows the performance of A and B lines from Urquiza top-
crossed on Leon CrioUo when tested at two different locations. Again the
average yield of the A topcrosses was higher than that of the B topcrosses,
and the percentage of topcrossed lines yielding more than the tester was
higher among the .1 lines than among the B lines.

The same thing was true in A and B lines obtained from three other varie-
ties when topcrossed on Urquiza, as summarized in Table 27.3. These data
definitely indicate that the probability of obtaining good general combiners

TABLE 27.1

COMPARATIVE YIELDS OF A AND B Si LINES

OF LEON CRIOLLO IN TOPCROSSES

WITH URQUIZA AT 3 LOCATIONS*

Per Cent of

Average

Lines In-

Yield of

Location

Class

No. of
Lines

Topcrosses

in % of

Check

creasing

Yield 25%

or More

of Check

I

A

31

116

32

B

60

113

23

II

A

33

122

39

B

64

108

20

Ill

A

33

92

18

B

45

90

14

* Variety Urquiza was used as check.

TABLE 27.2

COMPARATIVE YIELDS OF A AND B Si LINES

OF URQUIZA IN TOPCROSSES WITH LEON

CRIOLLO AT TWO LOCATIONS*

Average

Per Cent of

No. of
Lines

Yield of

Lines In-

Location

Class

Topcrosses

creasing

in % of

Yield of

Check

Check

I

A

13

111

69

B

33

97

40

II

A

19

102

63

B

56

90

28

*Leon Criollo was used as check.

430

E. J. WELLHAUSEN

as measured by the testers used, is higher among the more vigorous and
agronomically better lines.

The experiment has been carried somewhat further. Inbreeding was con-
tinued in the various Si families originally classified a.s A, B, and C. After
three generations of selfing, the advanced lines on hand were topcrossed on
two different testers and the resulting topcrosses were tested for yield. On the
basis of average topcross performance, certain of the advanced inbred lines
were selected as worthy of keeping for further work. The number selected
from each of several varieties together with their classification is given in
Table 27.4.

TABLE 27.3

COMPARATIVE YIELDS OF A AND B S, LINES OF

MICH. 21, PUE. 16, AND MEX. 39 IN

TOPCROSSES WITH URQUIZA*

Average

Per Cent of

No. of
Lines

Yield of

Lines In-

Variety

Class

Topcrosses
in ^0 of

creasing
Yield of

Check

Check

Mich. 21

A

21

110

76

B

21

105

67

Pue. 16

A

22

85

9

B

10

74

0

Me.x. 39

A

11

101

55

B

14

92

29

* Urquiza was used as check.

TABLE 27.4

NUMBER OF LINES SELECTED FROM FOUR
VARIETIES ON BASIS OF TOPCROSS TESTS
AFTER THREE GENERATIONS OF IN-
BREEDING AND THEIR CLASSIFICATIONS
IN Si

Variety

No. OF

Original
Si Lines
Inbred

No. OF
S3 Lines
Selected

No. Originally
Classified as
A and B in Si

A

B

Mich. 21

Mex. 39

Leon CrioUo. . . .
Chalqueno

219

131

218

67

9

2

5
22

8

1

4

15

1
1
1

7

HETEROSIS IN A NEW POPULATION 431

As pointed out before, in the first three varieties listed above, 3-6 j)er cent
of the lines were classified as A , 12-15 per cent as B, and the rest were classi-
fied as C. In Chalqueno, the Si families were discarded more heavily and the
67 Si families selected for further inbreeding were classified as follows: 16^,
35B, and 13C. It is clearly evident that by far the majority of S-> lines saved
for further study after testing in topcrosses came from the Si families origi-
nally classified as A .

In the early stages of the program, about five hundred plants were inbred
in each variety. At harvest, about two hundred, or 40 per cent, were selected
for further work. The above data indicate that, as far as the varieties listed
were concerned, the same results might have been obtained with a more
drastic elimination of lines at the beginning of inbreeding.

Selection of testers for use in the isolation of good inbred lines is always a
problem. In Mexico, chief concern in the early stages of the breeding program
was the isolation of lines with good general combining ability. For this pur-
pose the inbred lines in each zone were topcrossed on at least two different
testers, usually unrelated adapted open-pollinated varieties. The good com-
biners, when selected on average topcross performance, often were disap-
pointing when crossed inter se. In those zones where two varieties were avail-
able which, when crossed, produced a desirable hybrid agronomically, a re-
ciprocal method of testing was used. With this method, inbreds from variety
A were topcrossed on variety B, and inbreds of variety B were topcrossed on
variety A. Good combiners thus isolated from variety A were crossed with
good combiners from variety B to form single crosses and subsequently
double crosses. This method of double cross formation was more efficient
than the recombination of lines with so-called general combining ability from
the above method. Where a good single cross of first generation inbred lines
was available that could be used as a tester to isolate inbred lines which com-
bine well with it, this cross proved more efficient than either of the above
two methods in the formation of good double crosses.

Utilization of Semi-inbred Lines in Synthetics and Hybrids

Almost from the beginning of hybrid corn production, breeders have
sought to discover methods of utilizing superior inbred strains in more or
less permanent combinations. In a country such as Mexico, where the majori-
ty of the farmers will not readily adopt a practice of securing new hybrid seed
for each planting, superior synthetic varieties would have real advantage.

As Sprague and Jenkins (1943) pointed out, four factors operate to deter-
mine the yield of advanced generations of hybrids: (1) the number of lines
involved, (2) the mean yield of these lines, (3) the mean yield of all of their
possible single crosses, and (4) the percentage of self pollination. Since maize
is almost wholly cross-pollinated, the last factor may be largely ignored.

432 E. J. WELLHAUSEN

Which of the remaining three factors is the most important has not been de-
termined to date.

Wright (1922) has shown that with random mating the vigor and produc-
tiveness of an F2 is less than that of the Fi by an amount equal to 1/nth of
the difference between the Fi and the average of the parental lines where n
is the number of parental lines involved. These theoretical conclusions of
Wright are adequately supported by experimental data from maize (Neal,
1935; Kiesselbach, 1933; Wellhausen and Roberts, 1949).

In the past, in estimating the number of lines to use in a synthetic, appar-
ently it was assumed that the Fi mean in Wright's formula could be taken as
a constant value regardless of the number of lines involved. If this assump-
tion were correct, then the more lines involved the higher would be the yield
of the resulting synthetic. In actual practice, however, synthetics with a
large number of lines have yielded little more than the open-pollinated varie-
ties adapted to the same area. As indicated by Kinman and Sprague (1945),
the assumption of a constant mean yield for all Fi combinations seems un-
warranted. In any series in inbred lines, there are some that combine better
than others, and it is much easier to obtain four inbred lines that yield well
in all possible combinations than ten or sixteen. Therefore, to bring about the
highest mean yield of all possible single crosses, the use of relatively few lines
is indicated.

It can be shown by holding the Fi yields as a constant that a better syn-
thetic might be made with four more productive lines than with eight less
productive. For example, assuming the mean yield of all single crosses in-
volved to be 120 per cent, an F2 of a synthetic involving four lines with a
mean yield of 80 per cent will be 110 per cent, while a synthetic involving
eight lines yielding only 30 per cent will yield 109 per cent in F2. Kinman and
Sprague (1945) concluded that in general the most efficient number of lines
to be included in a synthetic will vary with the range in combining ability
among the inbreds available as parents. However, on the basis of their study,
four to six lines appeared to be the most efficient number, the smaller number
being most efficient when more productive lines, yielding at least 75 per cent
of open-pollinated variety, were involved.

Theoretically, therefore, the best synthetics would result from the use of
four to six lines which are as productive as possible and which are good com-
biners inter se. Certain practical aspects also must be taken into considera-
tion. If the inbred lines that are combined into a synthetic are greatly differ-
ent in type and maturity, the resulting Fo may be extremely variable for
these two characters. and may require considerable selection before distribu-
tion as a new variety. Variation is often a serious objection for many farmers
who have become used to their more-uniform highly selected old varieties.
The resistance of farmers to synthetic varieties which are variable and which

HETEROSIS IN A NEW POPULATION 433

are not strikingly higher yielding is often so great that it is difficult to obtain
wide scale distribution and use.

Considerable success was achieved in the formation of synthetic varieties
from double topcrosses in the early stages of the Mexican program, as de-
scribed by Wellhausen and Roberts (1949). Nevertheless, in consideration of
all factors, it seemed that the most logical procedure would be the formation
of good double cross hybrids which would also make good synthetics in ad-
vanced generations. In this way, the more progressive farmers could take ad-
vantage of the higher yield capacity of hybrids, while less progressive farmers
would still benefit by planting the advanced generation progeny. Also much
wider use of improved seed could be obtained more easily and rapidly since
hybrids always make a better showing than synthetics. With time, through
education, demonstration, and the formation of better hybrids, the demand
for hybrid seed could be increased gradually, and use of advanced generation
seed gradually would decrease. Some of the hybrids obtained for Zones 1 and
3 and their relative value as synthetics will be discussed below:

Results in Zone 1

In Zone 1, for March and April plantings under irrigation or under condi-
tions where subsoil moisture is sufficient for germination, the race Chal-
queno has become widely distributed. Some of the results obtained in the
attempt to improve this long season race are given in Table 27.5.

The variety listed as Chalco in Table 27.5 represents an average variety
of the race Chalqueho, and its yield for the sake of comparison has been
taken as 100. Variety V-7 is one of the best varieties found in the race, and
has been widely distributed as an improvement of the common variety
Chalco. Its two-year average yield in comparison to Chalco was 118 per cent.

Hybrid H-2 is a cross between two composites, one of which (Hgo. Comp.
1) was made up of a composite of five first generation selfed lines (Si) from
the variety V-7, that were good in topcrosses with the variety Urquiza, of
the race Conico Norteiio. The other (Urq. Comp. 1) was made up of a com-
posite of five Si lines from the variety Urquiza, selected on the bases of their
combining ability in topcrosses with a variety similar to Chalco. Urquiza
was the only variety of the different races tried in Chalqueho territory that
was fairly well adapted. Hybrids between two such composites are of in-
terest because the parents may be propagated as open-pollinated synthetic
varieties, thus eliminating the necessity of forming single crosses and the
maintenance of four inbred lines, as is the case in a double cross.

The yield of hybrid H-2 was only slightly higher than the open-pollinated
variety V-7. Work is under way to find out how much this hybrid might be
imi)roved through reciprocal recurrent selection.

The three-way hybrid H-1 was made by using a single cross involving an

434

E. J. WELLHAUSEN

Si line of Urquiza (Urq. 54) and an Si line of Chalqueno (Hgo. 3-5) as the fe-
male parent, with a very vigorous first generation selfed line Mex. 37-5 also
from Chalqueno as the pollinator. This hybrid involving one Urquiza line
and two Chalqueno inbreds has been one of the best combinations to date
of Urquiza and Chalqueno lines. In general, hybrids of inbred lines from
Chalqueno and inbred lines of Urquiza have been disappointing in comparison
with the yield of the variety V-7, although the Chalqueno lines in the crosses
were selected on the basis of combining ability with the variety Urquiza,
and Urquiza lines on the basis of their combining ability with a variety of
Chalqueno. None of the hybrids, including H-1, approached very closely the

TABLE 27.5

RELATIVE YIELDS IN PER CENT OF CHALCO FOR THREE

LONG SEASON VARIETIES AND HYBRIDS

IN 1948 AND 1949

Variety or

Pedigree

Relative Yields
m Per Cent

Hybrid

1948

1949

Average

Chalco

Open-pollinated
Selected O.P.
Hgo. Comp. 1 X
Urq. Comp. 1

(Urq.54XHgo.3-5)X
Mex. 37-5

100
117
120

134

100
120
125

130

100

Variety V-7

Hybrid H-2

Hybrid H-1

118
122

132

yield, ear size, and depth of grain of the best 10 per cent of the plants in the
variety V-7.

If a hybrid could be made in which 80-90 per cent of the plants ap-
proached the yield of the best 10 per cent of the plants in the variety V-7,
it would be an excellent hybrid. If such a hybrid is to be obtained, another
line of approach seems warranted. The best approach to this immediate ideal
is probably through the recombination of lines from V-7 or other varieties
within the race Chalqueno rather than bringing in outside germplasm, since
no race or variety has as yet been found which in Fi crosses with V-7 has
been as good as V-7. The question of selecting the proper tester for use in
isolating the high-combining genotypes, inter se, needs further study. Per-
haps a tester made up as a double cross or as a synthetic of the most vigorous,
most agronomically desirable, highest yielding, and best combining Chalque-
no lines would be the one to use. Against this nucleus of concentrated adapted
germplasm, other lines from V-7 or lines from other varieties of Chalqueno,
regardless of vigor, could be tested for their combining ability. Those that
increased the yield of the tester could then be pooled, and this pool could be

HETEROSIS IN A NEW POPULATION 435

concentrated further by another round of selfs and outcrosses to the tester.
The objective would be to concentrate the factors which complement those
of the tester to bring about the ideal hybrid. If successful, the final hybrid
could consist of a cross between two pools of germplasm, rather than indi-
vidual inbred lines, with the tester used as the female parent if so desired.

The race Conico (Figs. 27.1 and 27.2) is almost universally distributed in
Zone 1 where it is used for June plantings at the beginning of the rainy
season. With respect to general plant characters it is probably the poorest
corn in Mexico. The plants are sparsely leafed, have a poor tassel, and have
an extremely weak root system. As a normal practice, dirt is hilled around
each individual stalk to keep the plants upright until harvest. Its one out-
standing feature is its ability to grow and develop grain of high test weight
under relatively low temperature conditions.

To improve this race, incorporation of germplasm from another race with
a strong root system seemed highly desirable. The materials most likely to
be of use were the earliest varieties of the race Conico Norteno. For the pur-
pose of improving the corn varieties for the rainy season of Zone 1, therefore,
two varieties were selected. One, designated as Mex. 39, was one of the best
yielding varieties of the race Conico at 2200 meters elevation. The other,
designated as Leon I, was a good yielding early variety of the race Conico
Norteno commonly grown in Zone 2 at 1800 to 2000 meters elevation. This
second variety was a little late in maturity in Zone 1 and was not as well
adapted to the cool growing season as was Conico. Selected first generation
selfed lines from Mex. 39 were topcrossed on Leon I and selected Si lines
from Leon I were tested with Mex. 39. The best combiners with the respec-
tive testers under conditions best suited to Conico were then crossed in all
possible combinations, and tested for yield. From these results a double
cross that might also make a good synthetic was predicted, made, and tested.
The pedigree of this double cross, its yield relative to Mex. 39, and its prob-
able yield in F2 are given in Table 27.6.

The female parent of this double cross consisted of a single cross between
two Si lines from Leon I (Race Conico Norteno). The male parent consisted
of a cross between an Si line and a composite of four Si lines all from Mex. 39
(Conico). This pollinator also is being propagated and used as a synthetic
variety with good results.

Although based on only one year's results, the data in Table 27.6 indicate
substantial differences in yield capacity between the hybrid and the open-
pollinated variety Mex. 39. Perhaps what is more important at the moment
is that the advanced generation progeny of this hybrid shows promise of
being substantially superior in yield capacity to Mex. 39 which is one of the
better varieties of the race Conico. The actual yields of the double cross and
the six possible single crosses between the four parental lines, together with
the open-pollinated variety Mex. 39 as check in Table 27.6, were all deter-

436

E. J. WELLHAUSEN

mined in the same experiment under the same conditions. In calculating the
yield of the F2 generation, however, the average yield of the four parental
lines was estimated as 70 per cent of Mex. 39. In view of the fact that the Si
lines involved were the most vigorous within their respective varieties, this
estimation is conservative. Actual yields of the F2 generation in comparison
with Mex. 39 and the double cross hybrid are not yet available.

Results in Zone 3

Zone 3 comprises the corn belt of Mexico and produces more commercial
corn than any other area. In the eastern part of Zone 3, an area commonly
referred to as the Bajio, the race Celaya (Fig. 27.2) is widely distributed

TABLE 27.6

YIELD AND PER CENT DRY MATTER OF A DOUBLE CROSS

HYBRID COMPARED TO THE OPEN-POLLINATED

VARIETY MEX. 39 AND TO ITS F. YIELD*

Hybrid

(LI 27 X LI 193) X Mex. 39-26 X Mex. 39

Comp. 1

Mex. 39 O.P. (check)

Av. of all possible singles

Av. of parental lines (est. 70% of check) . .
F2 (calculated by Wright's formula)

Av.

c^

Yield
Kg. /Ha.

JO

D.M.

3795

71

2568

69

3506

70

1798

3079

Yield in
% of
Check

148

100

136

70

120

* Calculated from average yield of the three inbreds and one composite and their
six possible single crosses.

and is apparently a recent introduction into the area. The predominating
corn in the Bajio area at one time must have been the race Conico Norteno.
But as the root rot organism populations built up, the varieties of Conico
Norteno rapidly dropped in yield because of their almost complete root rot
susceptibility. Celaya probably originated in the State of San Luis Potosi
which is adjacent to the Bajio, but at a lower elevation than is common in
Zone 3. When introduced into the Bajio, it was a late variety and was first
grown only by farmers who had irrigation and could plant corn considerably
ahead of the rainy season. As irrigation farming increased, the race Celaya
became very popular and widespread. Selection pressure operated in the
direction of earliness, and certain varieties were developed which were more
productive than Conico Norteno on the better soils and under conditions of
a normal rainy season. The rainy season begins about July 1 and ends in
October.

In the western part of Zone 3, the predominating types are varieties of
the race Tabloncillo and inter-mixtures of Tabloncillo, Celaya, and a third

HETEROSIS IN A NEW POPULATION

437

complex from the mountains in southern Jalisco. The breeding work for
Zone 3 was concentrated in the Bajio and had the following objectives:

1. The development of hybrids and synthetics liigher yielding than
Celaya, with a high degree of root and ear-rot resistance, which have general
adaptation to Zone 3 under irrigation.

2. The development of early drought resistant hybrids and synthetics
for Zone 3, which are adapted to the normal rainy season of the area, and
which have a high degree of root and ear-rot resistance.

Since the variety Celaya already had considerable root and ear-rot re-
sistance, it was used as basic breeding material in the attempt to attain the
above objectives.

One of the first attempts at the im[)rovement of Celaya involved the pro-

TABLE 27.7
YIELDS OF TWO DOUBLE TOPCROSSES IN COMPARI-
SON TO THE OPEN-POLLINATED
VARIETY CELAYA

Hybrid

Pedigree

Yield
Kg. /Ha.

Yield in

% of

Celaya

H-305

H-301

Celaya*

(M30-60XGto. 59A)X
(LII123Xjal.35)

(LII123XGto. 59A)X
(M30-33Xjal. 35)

Open-pollinated variety

4845
4411
4092

120
108
100

* The variety of Celaya used as a check in this and in the following tables
is the one being maintained by the Agricultural Experiment Station in Leon,
Gto.

duction of double topcrosses. In this method, two different high yielding
varieties adapted to Zone 3 were used as testers. One of these, designated as
Gto. 59A, was one of the better late varieties found within the race Celaya.
The other, designated as Jal. 35, was a high yielding variety obtained from
southern Jalisco. This latter variety apparently was derived from a highly
heterogeneous mixture of several races that came together in southern Jalis-
co: Tabloncillo, Celaya, and the Jaliscan mountain complex (Wellhausen et
al.). Inbred lines (Si) from Gto. 59A and other varieties of the race Celaya
were topcrossed with Jal. 35, and Si inbred lines from Jal. 3>S and similar
varieties were topcrossed on Gto. 59A. These topcrosses were then tested
in three locations in Zone 3 with Celaya as a check. As a result of these tests,
the ten best topcrosses with Gto. 59A were selected and each crossed with
each of the ten best topcrosses with Jal. 35. Subsequently, the resulting
double topcrosses were tested at two different locations in Zone 3. Finally,
two double topcrosses were selected and released for commercial produc-
tion. The yields of the two double topcrosses for commercial production to-
gether with the yields of Celaya are given in Table 27.7. These comparative

438

E. J. WELLHAUSEN

yields are based on an average of four replications in each of 20 experiments
in two different localities and are highly significant statistically.

The double topcross hybrid H-305 yielded 20 per cent more than Celaya
and was equal to Celaya in maturity. Hybrid H-301 yielded only 8 per cent
more than Celaya but was about ten days earlier. In normal years this is a
decided advantage in adaptation to the variable rainy season of the Bajio.
The major portion of the 2000 hectares planted in 1948 for hybrid seed pro-
duction in the Bajio was used for the production of these two double top-
cross hybrids. According to a formula presented by Mangelsdorf (1939), the
gain in yield of these two double topcrosses in F2 would be about half of
what they showed in Fi over Celaya. The results of an experiment set up to
measure the difference in yield between the Fi and F2 generations of eight
different double topcrosses made up as described above are given in Table

TABLE 27.8

COMPARISON OF Fi AND F, GENERATION YIELDS OF EIGHT
DOUBLE TOPCROSSES IN PER CENT OF CELAYA

Genera-

Double Topcrosses

Average

tion

1

2

3

4

5

6

7

8

Fi

F2

Celaya . . .

132
123
100

115
104
100

111

70

100

99

92

100

94

93

100

101
117
100

98
114
100

89
114
100

105
103
100

27.8. The data are based on an average of four replications in each of two
locations.

According to the data presented in Table 27.8, the differences in yield
between the Fi and F2 generations were statistically significant only in the
double topcrosses 3 and 8. In topcross 8 the difference was in favor of the Fo
generation. Some of the double topcrosses undoubtedly held a certain ad-
vantage in yield over Celaya in the F2, but further data are needed before
accurate conclusions can be drawn with respect to the comparative yield
capacities of the Fi and F2 generations of double topcrosses.

Although some improvement was achieved over Celaya by means of the
multiple topcross method, the direct recombination of lines from Celaya has
given better results. Celaya definitely offers the best breeding material for
Zone 3. Crosses between Celaya and other races in Zone 3 such as Conico
Norteno and Tabloncillo have been disappointing. The race Conico Nor-
teiio introduces some earliness, but it also contributes susceptibility to root
rot and ear-rots. The race Tabloncillo of western Mexico behaves in a
similar manner. In addition, it introduces certain undesirable ear char-
acters.

HETEROSIS IN A NEW POPULATION

439

As sliown in Table 27.9 some outstanding hybrids have been attained
through a recombination of lines from Celaya. The Celaya lines used in the
hybrids of Table 27.9 were isolated by topcrossing with a variety of the
race Conico Norteno and with the variety Jal. 35. The two hybrids, H-309
and H-307 were made with Si lines, while H-310 was made with S2 lines.
Both H-309 and H-307 gave slightly better all around performance in yield
and in disease resistance than did H-310, but H-309 was the best of the three
in yield, disease resistance, and general agronomic characters. Its yield, as

TABLE 27.9

YIELDS OF DOUBLE CROSSES MADE FROM Si AND S2 LINES

COMPARED WITH THE AVERAGE YIELD

OF CELAYA, H-301, AND H-306*

Hybrid

Pedigree

Yield
Kg. /Ha.

%of
Av. of
Checks

H-309. . .
H-307 . . .
H-310. . .

(C 123XC 243) X(C 90XAg. 172)
Av. of checks

(Ag. 172XC 79)X(C 67XC 90)
Av. of checks

(M30-60-3XC 243-2-2) X(Ag. 172-2XG 61-5-4)
Av. of checks

4549
3703

4032
3375

4437
3760

123
119
118

* An average of two years' data from two locations.

shown in Table 27.9, was 23 per cent higher than an average of the three
varieties used as checks. The checks, in addition to Celaya, included two
double topcross hybrids (H-301 and H-306) which brought the level of
yield of the checks up to 107 per cent of Celaya. The actual difference be-
tween H-309 and Celaya, therefore, would be somewhat greater than 23
per cent.

Hybrid H-309 should also make a good synthetic because: (1) the average
yield of the six single crosses possible between the four inbred lines is very
nearly the same as the yield of the double cross H-309, and (2) because the
four Si lines were among the most vigorous obtained from Celaya. The actual
yield relationship between Fi, F2, and F3 progenies, where available for a
series of double crosses in a randomized block experiment with eight replica-
tions, is given in Table 27.10.

All yields in Table 27.10 are expressed in per cent of Celaya. All double
cross hybrids with the exception of No. 1, involve Si lines from varieties of
the race Celaya. Hybrid No. 1 was made with S2 lines from varieties of the
race Celaya. Hybrid No. 2 contains the same lines as H-309 in Table 27.9,
but in a slightly different combination. This combination has resulted in

440

E. J. WELLHAUSEN

slightly higher double cross yield, but should have no effect on yields of
the F2.

As shown in Table 27.10, the F2 generation progeny of the better double
cross hybrids with Si lines retained a substantial advantage in yield (12 to
20 per cent) over the open-pollinated variety Celaya. Hybrid No. 1 with S2
lines was included to see if it would actually show a greater drop in yield
between Fi and F2 than the others.

From Table 27.11 it is evident that the F2 or F3 yields of the double

TABLE 27.10

YIELDS OF Fi, F,, AND F3 GENERATION PROGENY OF

SEVEN DOUBLE CROSS HYBRIDS IN PER CENT

OF THE VARIETY CELAYA

No.

1.
2.
3.
4.
5.
6.
7.
8.

Pedigree

(Gto61-5-4XAg. 172-2) X(M30-6O-3XC 243-2-2)

(C 1 23 X C 90) X(C 243 XAg. 172)

(C123XM30-60)X(C90XC243)

(C 123XC 90)X(C 243XM30-60)

(C 123XC 90)X(C 243XAg. 32)

(C 123XC 90)X(Ag. 32XAg. 172)

(C 123XC 243) X(L II 67 XL II 90)

Celaya

Fi

F2

118

103

140

112

139

107

150

118

149

119

146

120

127

100

100

101

111

101
100

L.S.D. = 12.5%

TABLE 27.11
YIELD OF Fi, F2, AND F3 PROGENIES IN PER CENT OF Fi

No.

Pedigree

Fi

F2

F3

1. ...

(Gto61-5-4XAg. 172-2) X(M30-60-3XC 243-2-2)

100

87

86

2....

(C 1 23 X C 90) X (C 243 X Ag. 1 72)

100

80

3. ...

(C 1 23 X M30-60) X C 90 X C 243

100

•11

80

4. . ..

(C 1 23 X C 90) X (C 243 X M30-60)

100

79

5.. ..

(C 1 23 X C 90) X ( Ag. 32 X C 243)

100

79

6. ...

(C 123XC90)X(Ag. 32 XAg. 172)

100

82

7....

(C 1 23 X C 243) X (L II 67 XL II 90)

100

80

crosses with Si lines was consistently about 80 per cent of the yield in Fi,
whereas the F2 of the double cross with S2 lines was 87 per cent of the Fi.
This is not significantly higher, but also not significantly lower as one might
expect on the basis of the lower yields of the S2 lines.

The few F3 yields available were not greatly different from those of the F2.
The assumption that in general, barring selection, there is no further reduc-
tion in yield beyond Fo has been adequately supported by experimental
data. Sprague and Jenkins (1943) tested the Fi, Fo, F3, and F4 of one 24 line
and four 16 line synthetics in various districts in Iowa. There was little dif-

HETEROSIS IN A NEW POPULATION 441

ference between the various advanced generations, the 1^2, F3, and F4 yield-
ing 94.3, 95.4, and 95.1 per cent as much as the Fi respectively. Kiesselbach
(1933) compared the yield of the F2 and F;; generations of 21 single crosses.
The yield of F2 and Fg was approximately the same, being 38.4 and 37.8
bushels per acre respectively. Wellhausen and Roberts (unpublished) com-
pared the average yields of the Fi, F2, F3, and F4 generations in 18 topcrosses.
The average yield was 10.8, 9.9, 9.6, 9.8 kilos per plot for the Fi, F2, F3,
and F4 generations respectively.

In the attempt to obtain still greater yield over Celaya, approximately
1000 So and S3 lines were crossed with the single cross C 67 X C 90. Both lines
in this single cross were first generation selfs from Celaya, and the majority
of the lines crossed with it were from varieties of this race. It is of interest
to note the kind of lines that gave the highest yields in combination with
C 67 X C 90. Among the ten that were finally selected as the best combiners
with C 67 X C 90, three were from the variety Jal. 35, one from a variety
from the State of Coahuila in northern Mexico, and the rest were from
Celaya. The total number of lines included from Jal. 35 and from other va-
rieties not classified as belonging to the race Celaya were relatively small
compared to the total number of Celaya lines involved in the test. However,
four of the ten best combiners with respect to yield came from varieties out-
side the race Celaya. This is in line with the belief that the possibility of ob-
taining high yielding hybrids is greater in the combination of lines from dif-
ferent varieties than in the combination of lines from the same or closely re-
lated varieties. Nevertheless, hybrids obtained from a recombination of
Celaya lines were satisfactory in yield and generally more disease resistant
and more acceptable from an agronomic standpoint than hybrids between
Celaya and non-Celaya lines.

In the yield test results of all possible single crosses between the ten
selected good combiners with the tester C 67 X C 90, the two lines C 110-3
and C 126-5 from Celaya (the same variety from which the tester lines were
obtained) were of considerable interest. These two lines were not only good
combiners with the single cross tester C 67 X C 90, but also combined well
with each other. The single cross C 110-3 X C 126-5 was among the highest
of all the 45 possible single crosses among the ten selected lines.

The tester single cross which was made up of two average Si Celaya lines
would tend to isolate genotypes which contribute the greatest number of
additional yield factors to its own genotype. These genotypes could be very
much alike or greatly different. Apparently the two genotypes represented
by the lines C 110-3 and C 126-5 were greatly different both genotypically
and phenotypically. In ear type they seemed to be opposite extremes in the
range of segregation among Celaya lines. As shown in Figure 27.5, C 110-3 is
a line with a fairly long 8-rowed ear, and phenotypically appears to be a
segregant in the direction of Tabloncillo which is one of the probable pro-

442

E. J. WELLHAUSEN

genitors of the race Celaya. The other, C 126-5, has a fairly short ear and a
high row number and apparently is a segregant in the direction of Tuxpeno
which is the other probable progenitor of the race Celaya.

If selection for ear type, using Celaya as the ideal, had been a factor in the
development of inbred lines, then both C 110-3 and C 126-5 probably would
have been discarded. Selection for type may be a mistake in those varieties

i '

FiG. 27.5— Typical ears of the two inbreds C 110-3 {left) and C 126-5 {right). Both are from
the race Celaya, which probably originated from the hybridization of Tabloncillo and
Tuxpeiio. C 110-3 phenotypically appears to be a segregant in the direction of Tabloncillo
and C 126-5 appears to be a segregant in the direction of the tropical many-rowed cylindri-
cal dent Tuxpefio.

which apparently have not reached equilibrium, or in which segregants close-
ly resembling one putative parent or other appear. It may be an especially
bad practice if the lines from the same variety are to be recombined into
hybrids. In the recombination of lines from the same variety, it remains to
be seen whether good hybrids can be more readily made by a recombination
of lines which phenotypically are opposite extremes, or from those lines
which resemble more closely the type of the variety from which they came.
Probably both types are needed.

Hybrids and synthetics developed from Celaya lines were well adapted
to regions with supplemental irrigation, and to certain of the regions in Zone

HETEROSIS IN A NEW POPULATION 443

3 where rains are generally well distributed throughout the rainy season.
However, in many areas of Zone 3, the corn is often subjected to long periods
of drought. Since drought generally reduces the total length of time for
growth, varieties are needed which are not only drought resistant, but also
earlier in maturity than Celaya.

So far, no good hybrids earlier than Celaya have been obtained from a
recombination of early Celaya lines. It became necessary to look elsewhere
for material which would give the desired earliness and drought resistance
when combined with Celaya. The two races Conico Norteno and Tabloncillo,
which overlap Celaya in its distribution in Zone 3, were found to be un-
desirable because of their high susceptibility to both root and ear-rots, al-
though they were early in maturity. In the search for suitable material, a
race called Bolita, found in a small valley in Oaxaca about 500 miles from the
Bajio, has shown considerable promise. It probably originated in the Valley
of Oaxaca through the hybridization of Tabloncillo and an early maturing
tropical race called Zapalote Chico (Fig. 27.3). The Valley in Oaxaca where
Bolita probably originated has the same elevation and has a climate similar
to parts of the Bajio. Bolita, when grown in the Bajio, was found to be early
maturing, very resistant to ear-rots, and generally resistant to root rots. Its
yield capacity, however, was considerably below that of Celaya in years
with good rainfall distribution.

Through a method of reciprocal testing of lines of Bolita with Celaya and
lines of Celaya with Bolita, Si lines of Bolita were isolated, which when
combined with certain Si lines of Celaya, produced hybrids superior to both
Bolita and Celaya in the drier areas of Zone 3. One of these hybrids, made
with a single cross of two Si Celaya lines (C 90 X C 67) as a female parent
and a synthetic of four Si Bolita lines as a pollinator, is now being pro-
duced for large scale testing. Preliminary data obtained on this hybrid,
called Celita, are given in Table 27.12.

In the first three localities where Celita was tested, the rainfall was either
well distributed or supplemented by one irrigation in a period of extreme
drought. Under these conditions as evident in Table 27.12, Celita was about
equal in yield with the standard variety Celaya. But at Irapuato under ex-
treme drought conditions, Celaya yielded only 741 kilos per hectare (about
12 bushels per acre) while Celita yielded 1441 kilos per hectare, or about 23
bushels per acre. Also as indicated in Table 27.12 by the differences in per
cent dry matter at harvest, Celita was considerably earlier in maturity than
Celaya. Celita is also fairly resistant to root rots and much more resistant
to ear-rots than the best hybrids made with Celaya lines.

It appears, therefore, that the hybrid Celita, under conditions normal
for Celaya, is equal to it in yield, but under severe drought conditions it is
greatly superior. This hybrid also is superior to Bolita under both normal
and dry conditions although the data are not presented in the table. Here

444

E. J. WELLHAUSEN

then is another case where a hybrid between Si lines of two different races
under one set of conditions is no better than the better of the two parents,
but, under a different set of conditions, is superior to both.

Double cross hybrids made from Si lines of Celaya in combination with
Si lines of the race Conico Norteiio have in general given good results in
Zone 2, with yields ranging from 20 to 25 per cent higher on the average
than the native varieties commonly grown in the area.

TABLE 27.12

YIELD OF CELITA AND PERCENTAGE DRY MAT-
TER AT HARVEST COMPARED TO CELAYA AT
FOUR DIFFERENT LOCATIONS, 4 REPLICATIONS
EACH

Yield Kg. /Ha.

% Dry Matter

Locality

Celita

Celaya

Celita

Celaya

Vista Hermosa. . .

Guadalajara

Leon

3793
4069
4273
1441

3806

3760

4223

741

68
80

75
81

60
69
66

Irapuato

77

In the tropical areas (Zones 4 and 5) hybrids were under test for the first
time in 1950. These involved principally combinations of Si and S3 inbred
lines from the races Tuxpeno and Vandeno (Fig. 27.4).

Lines Selfed Once versus Lines Selfed More Than Once
in Hybrid Formation

The use of Si lines in the early stages of a breeding program has many
advantages. It means that testing for combining ability can begin in the
first generation of selling. It can, in fact, begin with selected open-pollinated
plants which may be simultaneously selfed and crossed. Lines thus isolated
in a breeding program where uniformity is not of prime importance can be
utilized immediately in the formation of hybrids and synthetics. Since Si
lines are more vigorous than advanced generation selfed lines, they also have
a definite advantage in the formation of synthetics. It has never been defi-
nitely determined whether high yielding hybrids can be obtained more
readily with homozygous lines than with heterozygous lines. Jenkins (1935)
has shown that crosses of lines selfed only once are on the average as pro-
ductive as crosses involving the same lines selfed six to eight generations.
This may indicate, as Jenkins suggests, that the effects of selection are al-
most exactly balanced by the loss of good genes through the rapid attain-
ment of homozygosity.

Some data have been accumulated to date in the Mexican program

HETEROSIS IN A NEW POPULATION 445

which may have some bearing on the relative value of Si lines versus more
homozygous lines in the formation of hybrids.

Preliminary data on the relative combining ability between Si and the S3
lines selected from each Si are available from topcrosses to the same tester.
Each topcross with an Si line was tested for yield in the same experiment
with the corresponding topcrosses involving the lines obtained from that Si
after three generations of selling and selection for desirable agronomic char-
acters. The number of S3 lines in each Si family varied from one to sixteen,
some families having a larger number of desirable S3 lines with respect to
agronomic characters than others.

A frequency distribution of the differences in topcross yields in per cent
between Si line topcrosses and the average of the S3 line topcrosses within
each family is given in Table 27.13. The differences are expressed as Si minus

TABLE 27.13

DISTRIBUTION OF DIFFERENCES IN TOPCROSS YIELDS BE-
TWEEN Si LINES AND THE AVERAGE OF THE S3 LINES
WITHIN EACH FAMILY (Si - AVERAGE OF Sa's)

MINUS

Class center

Frequency

75
1

70 65

60

55

50

1

45 40 35
2 2

30
4

25
6

20
3

15
9

10

12

5
12

PLUS

Class center

Frequency

0

25

5 10
18 18

15

3

20

3

25
3

30 35 40

4 4 2

45

50
4

55
1

60

65

1

Number of observations = 138 Mean = +0.90

the average of the S3 within the respective Si family. The class mid-points,
therefore, range from 0 to 65 per cent positive and from 0 to 75 per cent
negative, with class intervals of 5 per cent. A positive difference means that
the Si topcross yield exceeded the average of the S3 topcrosses. A negative
difference indicates that the average of the S3 topcrosses was higher than
the Si topcross within the same family. It is evident from Table 27.13 that
the distribution of the differences approaches very closely that of a normal
curve. That is, there were as many cases in which the Si exceeded the aver-
age of the S3 as there were cases in which the average of the S3 exceeded the
Si. The mean difference between the 138 pairs was +0.90 per cent. These
data indicate that visual selection in advanced selfed generation progeny
based on agronomic characters is largely at random with respect to com-
bining ability. If visual selection in successive generations of inbreeding had
been effective in increasing combining ability, then the above curve would
have been skewed in the direction of the negative differences.

Upon further inbreeding of Si lines at random without selection for com-

446 E. J. WELLHAUSEN

bining ability, one would expect to end up with about as many advanced
generation inbred lines which exceed the Si in combining ability with a
specific tester as lines which were below that of the Si. In other words, the
distribution in relation to the Si yield would follow that of a normal curve.
In Table 27.14 are given the distributions of the yields of Si and S3 top-

TABLE 27.14

FREQUENCY DISTRIBUTIONS OF Si AND S3 LINE-TOPCROSS YIELDS OF 12

FAMILIES. CLASS IN WHICH Si LINE-TOPCROSS OF EACH FAMILY

F.ALLS IS INDICATED BY NUMBER IN BOLD FACE TYPE

Class Mid-
points (Yield
of Topcrosses
in % of
Checks) 65 75 85 95 105 115 125 135 145 155 165 175 185 195 205

Family and

Frequency

Hgo. 9-4 12 7

Ch. II 148 12 1 3

M. 37-5 12 113 1

Hgo. 4-5 11 4 2 3 3 1

Hgo. 1-5 1 111

Hgo. 3-4 1 2

Hgo. 2-3 12

Ch. II 187 12

Hgo. 1-8 111

Hgo. 6-11 1113 1

Hgo. 3-5 12 2 11

Ch. IV 146 2 1

crosses of 12 families from the same race of corn (Chalqueno). All of the Si
and S3 lines within each family were topcrossed on an Si line from the race
Celaya, and all topcrosses were tested under the same conditions. The class
in which the Si topcross of each family fell is indicated by the number in bold
face type. It is evident from the table that the number of S3 lines from each
family tested are insufficient to show a normal distribution. However, in
nearly every case where more than three S3 lines were available for com-
parison, some were no better, some were significantly better, and still others
were significantly worse than the respective Si line in combining ability. It
appears, therefore, that in certain cases considerable increase in the yield of
specific combinations involving Si lines can be obtained through further in-
breeding and selection for specific combining ability.

Further evidence that better yields can be obtained through the substitu-
tion of S2 or S3 lines in a specific Si combination is presented in Table 27.15.

This table is divided into two parts. In column A is given the yield of each
specific single cross made with Si lines. In column B is given the yield of each
corresponding single cross of two advanced lines selected from the Si lines
given in column A in the second or third generation of inbreeding. Selection

HETEROSIS IN A NEW POPULATION

447

of the So or S3 lines in each case, however, was not based on test crosses for
specific combining ability with the other line involved. The advanced lines
used were picked from among their sister lines largely on the basis of desir-
able agronomic characters. It may be seen in Table 27.15 that, although
in many instances the differences were small, the yield of the crosses of S2
or S3 lines exceeded that of the corresponding cross with Si lines in every

TABLE 27.15

YIELDS OF SINGLE CROSSES BETWEEN
Si LINES COMPARED TO YIELDS OF
SINGLE CROSSES BETWEEN TWO
LINES*

Yields of

Yields of

Crosses

Crosses

between

Differences

Crosses

between

S2 and/or S3

A-B

Si Lines

Lines

(A)

(B)

1

5935

6873

-938

2

6074

6630

-556

3

6340

6560

-220

4

5172

6514

-1342

5

5056

6479

-1423

6

5669

6306

-637

7

5588

6259

-671

8

5970

6202

-232

9

5970

6190

-220

10

6005

5935

+70

11

5334

5843

-509

12

4535

5577

-1042

13

5368

5473

-105

14

5542

4964

+578

N = 14.

Mean difference = 517.6 + 147 kilos (or

9.2 per cent).

* Derived from the respective Si's after two or three
generations of inbreeding and selection for agronomic char-
acters only. (Kilos per hectare.)

case except two. The average difference between the fourteen paired crosses
was 9.2 per cent. It is highly probable that a greater increase would have been
obtained had the various S2 or S3 lines also been picked on the basis of tests
for specific combining ability. However, since selection can make a choice
only between the alleles present in a particular Si, a point of diminishing re-
turns may be rapidly reached upon straight selling. Experiments are under
way to determine in what generation of selfing maximum gains may be
reached.

The data in Table 27.14 are of further interest from the standpoint of rela-
tionship between the combining ability of Si lines and the advanced genera-

448 E. J. WELLHAUSEN

tion selfed lines obtained from them. In this table it is evident that in those
families where the Si lines were poor combiners with a given tester, the S3
lines obtained from them on the average also tended to be poor. In those
families where the Si lines were good combiners, the S3 lines obtained from
them were also good. A correlation coefficient of 0.69 was obtained between
the topcross yields of Si lines and the average topcross yields of the S3 lines
derived from each. This highly significant correlation coefficient, based on the
same 138 pairs whose differences were distributed as shown in Table 27.13,
indicates a high degree of relationship between the performance of Si and the
average performance of lines obtained from each through subsequent genera-
tions of inbreeding. It seems, therefore, that tests for combining ability in the
So or Si generation would serve to separate the families that are good com-
biners from families that are poor combiners with respect to a given tester in
the early stages of the inbreeding program.

CONCLUSIONS AND SUMMARY

Although the yield results are based on relatively few years' data, it is evi-
dent that the methods used in the improvement of corn in Mexico during the
six years that the Mexican program has been under way have given excellent
results. In some areas considerable improvement in corn yields was obtained
by the wider distribution of certain good native, open-pollinated varieties
that had developed in isolated areas through chance hybridization and subse-
quent natural selection. In areas where two different adapted varieties were
available which expressed a certain degree of heterosis when crossed, the
formation of double topcrosses offered a means of rapid improvement over
the native varieties.

It has also been shown that excellent three-way or double cross hybrids
can be made from first generation selfed lines. Some of these same double
crosses in advanced generations have made good synthetics. This means that
those farmers who cannot or are unwilling to plant newly crossed seed every
year may still have a 12-20 per cent advantage in yield over their native
varieties. Hybrids made from crosses of two synthetics, each consisting of a
pooled set of closely related Si lines that combined well with a different
pooled set of related Si lines, have shown some promise in the greater sim-
plification of hybrid seed production. In this way the maintenance of lines for
hybrid seed production can be greatly simplified.

The use of first generation selfed lines in the early stages of a new breeding
program obviously has many advantages. Whether such lines can be main-
tained for a reasonable period of time without much change in combining
ability remains to be determined. So far through a composite sib method of
propagation they have been maintained reasonably "pure."

Data have been presented which indicate that hybrids made from more
homozygous lines might be superior to those made with lines selfed only once.

HETEROSIS IN A NEW POPULATION 449

It remains to be shown whether more uniform hybrids actually would be
superior to hybrids made with Si lines in a country such as Mexico. Variations
in climate from year to year in any one valley in Mexico are usually extreme.
Under such conditions, a high degree of uniformity in a hybrid may actually
be a detriment over a period of years.

The problem of what tester or testers to use in isolating lines of high com-
bining ability continues to be a difficult one. Usually the tester chosen de-
pends upon the use to be made of the lines. Judging from the segregants ob-
tained upon inbreeding in some of the races of maize, a point may have been
overlooked in the selection of lines and testers. This appeared to be especially
true in those races where it was necessary to recombine lines from the same
race to obtain immediate improvement. In some of the races upon inbreed-
ing, especially Bolita, Chalqueno, and Vandeno, inbred line segregants often
appeared which were very similar to the putative parents of the particular
race (Wellhausen et al. in collaboration with Mangelsdorf, Fig. 98). If these
races had reached equilibrium on an individual gene loci basis, one would not
expect to get the parental types in subsequent inbred generations from 500
ears selfed at random in the original population. It appears, therefore, that
many of the modern races in Mexico are not in equilibrium on an individual
gene loci basis, but consist of blocks of genes in equilibrium with each other.
Although it is difficult to estimate the age of some of the modern varieties,
these gene blocks obtained from the various ancestors seem to have persisted
more or less intact through many generations.

If blocks of germplasm as received from various ancestors are still intact
in some of the modern high yielding races, then it may not be as difficult as it
once seemed to reconstruct a hybrid that would approach the yield of the
ideal plant in a particular variety by the recombination of inbred lines from
that variety. Isolation of good lines for such recombination may involve dif-
ferent procedures. A method based on selection for origin and type, with sub-
sequent crossing to an unrelated variety or varieties for the determination of
combining ability, may not be the best procedure.

Selection for vigor and type in an environment best suited to a race such
as Chalqueno, which probably originated from the hybridization of two dif-
ferent races neither of which is adapted to the environment best suited for
Chalqueiio, would eliminate those segregants in the direction of either one of
the putative parents. It is probable that with the elimination of such segre-
gants, many genes are discarded that are needed to reconstruct the ideal
chance hybrids which often appear in a particular variety or race through
open pollination. Selection for vigor and type also would tend to select those
genotypes which are similar, and more nearly like those of the variety from
which they came, than the extreme segregants.

Tests for combining ability of a group of lines from the same variety, based
on crosses with an unrelated variety or varieties, tend to select those geno-

450 E. J. WELLHAUSEN

types which combine well with the particular tester, but do not differentiate
lines that combine well among themselves. Thus it seems that new methods of
isolating the good combiners must be sought in those races of hybrid origin in
which improvement is desired through inbreeding and recombination of lines
within the race.

FRED H. HULL

Florida Agriculfural Experimenf Sfation

Chapter 28

Recurrent Selection
and Overdominonce

For many breeders, in considering problems that lie ahead and methods
of meeting them, the main problem is whether to continue with varieties or
breeds, or to work with inbred lines and Fi crosses. Behind this question are
the problems dealing with the relative importance of general and specific com-
binability, or of prepotency and nicking:

Is the yield gain of hybrid corn due mainly to selection within and among inbred lines,
or to selection among Fi crosses of inbred lines?

Is it due to improved frequencies of dominant favorable genes in elite inbred lines which
are parents of elite-yield hybrids?

Is selection within and among inbred hnes to accumulate higher frequencies of domi-
nant favorable genes many times more powerful in one cycle without recurrence, than is
selection without inbreeding through many recurring cycles?

To what extent may higher levels of specific combinabihty be reached by recurrent
selection?

How may heritability of specific combinability be evaluated?

Why have the less favorable alleles of vigor genes been retained in such high frequencies?

May selection for general combinability and selection for specific combinability some-
times have counter effects on gene frequencies?

Does superiority of Fi crosses of inbred lines over varieties or breeds necessarily depend
on overdominance?

If this choice of problems is approximately correct, the research emphasis
may begin to shift from effects of inbreeding to effects of selection.

EARLY EXPLANATIONS FOR HYBRID CORN
East and Emerson in an early paper considered the theoretical problem of
recovering two traits together from a crossbreeding po]Hilation in which the
frequency of each trait was 1/1000, and the two were independent. The au-
thors offered two solutions: first to select at the rate of one per million in one
generation, and second at the rate of one per thousand in two generations re-
currently, first for one trait and then for the other. It is clear now that selec-

451

452 FRED H. HULL

tion for both traits together each time, with normal distribution, would pro-
vide theoretical recovery in two generations at the rate of 1 per 400 or less.
Multiplication of selection differentials in recurrent selection was sufhciently
understood at the inception of hybrid corn. Nevertheless, hybrid corn has
been developed with virtually no use or benefit from recurrence of selection.
Hybrid corn is almost wholly an empirical development, but I think we may
now consider applications of genetic science to improve the process.

Recurrent selection (Hull, 1945a) was meant to include reselection genera-
tion after generation, with interbreeding of selects to provide for genetic
recombination. Thus, selection among isolates, inbred lines, or clones is not
recurrent until selects are interbred and a new cycle of selection is initiated.
Recurrent selection for specific combinability would seem to require a special
breeding plan to provide heritability through successive cycles.

Shull's original plan for developing superior corn hybrids was designed for
maximum immediate employment of specific combinability. Selection was
mainly among specific Fi crosses of lines which had been isolated and stabi-
lized by inbreeding, thus providing repeatability of crosses. This plan was
consistent with a theory of heterozygosis of a degree here termed over-
dominance. Shull's plan did not involve recurrent selection to accumulate
higher frequencies of favorable genes in successive cycles.

The apparent heterozygosis which Shull proposed to use was interpreted
by Jones about ten years later as the expectation of repulsion phases of ran-
dom linkages of dominant favorable factors and recessive less favorable al-
leles. This interpretation was particularly attractive because it seemed to
eliminate any necessity of accepting overdominance. Overdominance is a
contradiction of the time-honored principle that purity of blood is to be
sought and maintained. Vigor was no exception to the old principles of like
begets like and breed the best to the best. Moreover, the postulated linkage rela-
tions would appear to be inevitable where many loci are involved.

In the decade following appearance of the Jones hypothesis, most corn
breeders began more intense selection for vigor within and among lines during
the inbreeding process, and selection among lines for general combinability.
Most of the very considerable success of hybrid corn came quickly after these
modifications of Shull's method were adopted. Selection within and among
inbred lines to improve frequencies of dominant favorable factors became the
guiding principle for developing superior hybrids of corn, other crops, and of
livestock. Selection among specific Fi crosses was retained as a final step, but
with very little verbal emphasis.

Initial successes with hybrid corn (which so far have not been greatly sur-
passed) were obtained with inbred lines which were, for the most part, iso-
lated directly from the open-pollinated varieties. Corn breeders then had at
least two alternatives for further work. Empirically, the choice might well
have been to continue isolation and testing of additional new lines from the

RECURRENT SELECTION AND OVERDOMINANCE 453

same sources, abandoning recurrent selection entirely. Usually, successful
but mysterious processes are not modified on theoretical grounds alone. How-
ever, most of us, and myself most of all I suspect, chose the alternative course
without question. New lines for a second cycle of selection were isolated from
crosses of elite first-cycle lines. Since it was soon apparent that second-cycle
lines as a group were a vast improvement over first-cycle lines, it was clear
that we were on the right track. Recurrent selection for higher frequencies of
dominant favorable genes was fulfilling expectation admirably. That it had
failed in ear-to-row selection (progeny testing without inbreeding) meant
that "selection within and among inbred lines" was the key. Apparently the
protagonists of "early testing" have not fully appreciated this latter i:)oint.

DISAPPOINTMENT WITH SECOND-CYCLE HYBRIDS

My first susj)icion that all was not well was aroused by disappointing
yield performance of second-cycle hybrids in 1941. The first reaction then
was to conclude that heterosis might involve complex gene interactions to a
greater extent than I had supposed. Cytoplasmic-nuclear interactions could
not be ruled out entirely. But no thought of heterozygosis, of overdominance,
was entertained at all, so thoroughly had I been weaned from it.

In 1942 w^e began the process of separating Florida inbred lines into two
permanently distinct groups on the basis of combining values with two single
cross testers which were thought to make a good double cross. Subsequent
breeding operations after the initial separation were to consist of isolating
new lines within each group from crosses of the older lines within the group.
New lines were to be stabilized by at least three self-pollinations with ac-
companying selection for vigor and type, and then tested for combinability
with the reciprocal group. This, of course, was reciprocal recurrent selection
without early testing. I still adhered firmly to the efficacy of "selection within
and among inbred lines."

Segregation of the breeding mass into two permanently distinct reciprocal
groups, first of all, did not cost anything. A search for satisfactory substitutes
for each of the four master tester lines was well in order. It seemed that the
necessity of recovering specific combinability again as the last step of each
breeding cycle might be avoided to some extent. Possibly higher levels of
specific combinability might be accumulated.

Two years later, after interviews with a number of other corn breeders,
it seemed that a still higher rating might be in order for specific combinabil-
ity. Second and third-cycle hybrids were not much superior to first-cycle
hybrids in yield of grain. Recurrent selection for general combinability was
not proving to be very effective.

An early test of recurrent selection for specific combinability seemed de-
sirable. One way to intensify the process already in operation was to adopt a
more specific tester. This was done by abandoning the reciprocal feature of

454 FRED H. HULL

the plan — by reducing one of the reciprocal groups to the single-cross tester
alone. That tester is to be continued indefinitely. Another way to intensify
the operation was to increase the frequency of recurrence of selection. This
was done by adopting the general principle of early testing, by abandoning
the inbreeding interphase of each cycle, by testing So plants rather than S3
lines or higher. Inbred lines, including the tester lines, of the second reciprocal
group were intercrossed to provide one crossbred group of So plants. Re-
peated selection within this crossbred group for combinability with the per-
manent unrelated tester is the proposed plan. It is only for practical reasons
that one homozygous line is not employed as the tester for field corn. With
sweet corn a line tester might well be used.

The working definition of specific combinahility employed in designing the
foregoing breeding plans was about thus: that part of the genetic superiority
of specific Fi crosses of homozygous lines which is not transmitted into or
through general recombinations. The concurrent definition of general com-
binability then is: that part which is transmitted into and through general
recombinations. That these definitions are perhaps inadequate for analyses of
variance does not necessarily mean that they are not admirable for the other
purpose.

ShuU, East, and others who isolated inbred lines and crossed them discov-
ered that inbreeding did little or no irreparable harm to the germ plasm.
Gametes of inbred lines hardly differ basically from gametes of crossbred
varieties. The inbreeding effect is very nearly or entirely a zygotic phenome-
non. Vigor genes in both homozygous and heterozygous associations were
obeying Mendel's first law of non-contamination. All of this was an important
discovery.

ShuU in addition invented selection for specific combinability, which was
certainly something new under the sun; yet to be generally recognized as one
of the great inventions. Shull was led, I suspect, to this invention by the
empirical evidence before him, not by considering the more abstract concept
of heterozygosis. Shull must have recognized very soon that reconciliation of
his invention with his knowledge of genetics required heterozygosis, and per-
haps the more inclusive heterosis.

RECURRENT SELECTION FOR SPECIFIC COMBINABILITY

A little more than thirty years later the inevitable invention of recurrent
selection for specific combinability was made from matter-of-fact empirical
considerations as outlined above. Again it seemed necessary soon afterwards
to embrace some theory of heterozygosis for reconciliation with genetics.
The breeding plan was presented (Hull, 1945a) with confusing emphasis on
the abstract concept of overdominance, I fear, and too little emphasis on the
actual motivation.

May it be said now that the first proposal was to determine with direct

RECURRENT SELECTION AND OVERDOMINANCE 455

tests if higher levels of specific combinability could be accumulated by recur-
rent selection. There is no need to await incontrovertible evidence of over-
dominance; indeed even if it were in hand the direct test would still be
needed.

The second proposal was that if recurrent selection for specific combinabil-
ity should be important, selection within and among inbred lines had been
greatly over-emphasized. The inbreeding interphase could be abandoned.
This would provide an enormous saving in time and otherwise, particularly
with poultry and other livestock. Curiously, some reviewers have described
the proposed breeding plan as a "laborious method."

Grain yield of corn depends appreciably on resistance to new and sporadic
diseases, insects, and adverse environmental complexes. Here it would seem
that overdominance is not likely, but that selection within and among inbred
lines is yet of real value. Significant resistance where it exists will eventually
be identified in continuing a stable line. Selected crosses will be generally
superior insofar as the several resistances are dominant and matched com-
binations are found. Here again I am not certain that rapidly recurring
progeny tests without inbreeding may not be equally or more effective in the
main. One resistant line among some hundreds of susceptible ones in an epi-
demic provides a striking field illustration — perhaps a deceptive one.

Breeding plans to accumulate specific combinability may be designed in
many ways, the better ones to be determined by actual tests. Testers might
best be the male parent of the hybrid in some cases, or the female parent in
others. The inbreeding interphase may be omitted or included in any prac-
ticable degree. It has been thought that the problem of the preceding para-
graph might be met well enough by direct selection in the crossbred lot and
selection among So testcrosses. But in some cases there might be an advan-
tage with Si or So testcrosses. With So or S2 some of the selection may be for
general combinability, for higher frequencies of genes which are favorable in
any combination.

The early view (Hull, 1945a, Proposition 7) was that where aA is generally
intermediate to aa and AA, A should be in high frequency, in improved
varieties. Not much further opportunity for improving combinability would
remain.

Crow's viewpoint, as he has presented it here, seems to be that without
overdominance long continued selection in any form would have carried
favorable alleles to high frequency in equilibrium with reverse mutation,
where heterozygosity is infrequent and heterosis not large.

If recent shifts of environment or of emphasis in artificial selection should
have provided important loci with intermediate gene frequencies, Crow's
argument may not be germane. Here I may venture an opinion (Hull,
1945b) that without overdominance rapidly recurring mass or ear-row selec-
tion should continue to surpass contemporary selected Fi crosses of homozy-

456 FRED H. HULL

gous lines. Or we may consider the more efBcient technic of recurrent testing
of controlled testcrosses of So plants with the parent variety and recombin-
ing the better ones into an improved variety. We know this will not work, al-
though it has not been fairly tried. Finally, in modern corn breeding the same
technic with S4 and higher lines has been extensively advanced through at
least two cycles. Most corn breeders will admit that a general recombination
or synthetic blend of parent lines of present elite-yield hybrids would hardly
yield more than a random blend of parent varieties of today or of 50 years
ago.

A few recombinations of lines selected wholly for general combinability
have been reported with significantly higher yields than improved varieties.
This result I will attempt to show later is a different matter, fully consistent
with overdominance theory.

It seems likely that improvement of general combinability, accumulation
of dominant favorable genes with respect to grain yield, in the field corn of
our central Corn Belt in the past fifty years has been hardly significant except
for that depending on disease resistance, resistance to lodging, to ear drop-
ping, etc. Almost any one of the common breeding technics is quite efifective
with general improvement of morphology of the corn plant, or with oil and
protein of the grain. Genetics of vigor would appear to differ in some impor-
tant respect from genetics of the other traits.

Overdominance has seemed the more likely, but I have never meant to in-
sist that the existence of every other alternative had been disproven. Refrac-
tory repulsion linkage has seemed insufficient alone to explain the apparent
degree of overdominance in corn (Hull, 1945a).

The main point now is accumulation of general combinability with recur-
rent selection. It is axiomatic with most of us, including the corn breeders,
that general combinability is the first consideration, despite the evidence
cited here. This kind of evidence has been largely ignored and almost taboo.

Comstock et al. (1949) have proposed Reciprocal Recurrent Selection to
obtain maximum utilization of general and specific combinability together.
In this they have accepted that specific combinability might be accumulated
in successive cycles, and that the inbreeding interphase could be abandoned
entirely. This variation of the general plan was compared on theoretical con-
siderations with selection in a crossbred for combinability with a homozygous
tester. Now, since a homozygous tester is clearly impracticable in many cases
and heterozygosity would impair efficiency of a tester except for reciprocal
selection, there is an advantage in the reciprocal plan which the authors did
not record.

It has never been my intent, however, to attempt to rule out judicious
reciprocal selection. We have crossed each of the two tester lines of corn to a
goodly number of unrelated strains, and have backcrossed in bulk to each
tester line separately. The two lots are being held in separate crossbreeding

RECURRENT SELECTION AND OVERDOMINANCE 457

reserves with nominal selection for agronomic type. If either tester line should
develop a serious fault, or if the present main selection for specific combinabil-
ity should seem to reach a ceiling, reversal of selection would seem almost
inevitable. A tester would be chosen from the current crossbred and the two
bulk backcrosses would furnish a reciprocal crossbred to reverse the process,
temporarily at least.

An accessory operation with bulk backcrosses is hardly practicable with
livestock. But here the tester would be one inbred line which would need
to be 50 per cent inbred for equal efficiency with the single cross of homozy-
gous lines employed as the corn tester. The tester should be the male parent
of the improved hybrid in livestock to avoid any impairment of the female
function by inbreeding.

Beginning with a partly inbred or non-inbred stud flock or herd as the
tester, and continuing mild inbreeding, it is inevitable that choices among
young males for herd sires of the stud herd would depend partly on their
testcross progeny. Sufficient vigor must be retained in this herd to provide
satisfactory sires of commercial hybrids. The problem is real and obvious
enough, but I have thought the details must await a demonstration that
specific combinability can be accumulated in important amounts by recur-
rent selection. For an early test the more homozygous tester is probably to be
preferred. If uniformity of the product is of some moment, the operator of
reciprocal selection may expend considerable effort for it. Such expenditure
might be avoided by partial inbreeding of one of the groups.

The two breeding plans, selection in a crossbred to a homozygous tester
and reciprocal selection between two crossbreds, are the extremes of recur-
rent selection for specific combinability. Between these we may have any
practicable degree of inbreeding of one of the groups at the start, or subse-
quently. Inbreeding restricts reciprocal selection but, aside from that, the
reciprocal feature may be varied at will. I do not know what factors may
determine the more efficient plans except that general combinability with
respect to vigor is probably not an important one. Nor is it likely to be im-
portant to choose an inbred tester with above-average general combinability.

PHYSIOLOGICAL NATURE OF OVERDOMINANCE
Overdominance has been defined (Hull, 1946a) as aA > AA, which is a
sufficient definition for present purposes. However, there may be some value
in considering what the underlying physiology may be. Heterozygosis as con-
sidered by Shull and his early contemporaries is entirely or very nearly the
same concept. Fisher (1918, 1932) has discussed this concept more gen-
erally as super-dominance. Some recent writers have employed heterotic al-
leles or heterotic interaction of alleles as a modern form of heterozygosis. But
since any degree of dominance of the more favored allele is essentially a
heterotic interaction, heterotic alleles does not necessarily imply a A > AA.

458 FRED H. HULL

In the current sense that any interaction of alleles is dominance, aA > AA
is overcomplete dominance, overdominance. In a similar sense all inter-
actions of non-alleles are epistasis. Dominance and epistasis differ in dis-
tribution on chromosomes, but not necessarily in underlying physiology so
far as I can see. Overepistasis would excite no particular comment.

Dominance and epistasis are no more fundamental properties of genes than
is interaction a property of a unit of nitrogen or phosphorus. These fertilizer
elements may exhibit an interaction in plant growth if made available to a
living plant, or they may seem to act independently. One quantity of nitro-
gen may be adequate for the needs of the plant. Adding the same quantity
again may produce no further effect. There is an interference or decreasing
returns interaction.

East (1936) has discussed dominance as a decreasing returns or interfer-
ence interaction of active alleles Ai and Ai in the homozygote. The amount
by which the two together failed to do twice as much as either alone was a the
dominance effect — a loss which could not explain heterosis. East then pro-
posed that if Ai should develop by successive steps to ^44 (analogous to re-
placing successive parts of one bag of nitrogen above with phosphorus until
there is one of phosphorus and one of nitrogen) of a different quality, Ai and
^4 might interfere very little or none in ^1^44. The principle as East states it
is: "The cumulative action of the non-defective allelomorphs of a given gene
approaches the strictly additive as they diverge from each other in function."

The effect of the phosphorus and nitrogen together is the sum of their
separate effects — no interference. Dominance by interference disappears
when Ai and ^44 are independent in functions, leaving yl 1^4 superior to either
AiAi or .44.44. Now it must be clear that any deviation of AiAi from the
mid-point between the two homozygotes must be interpreted as dominance
of ^ito .4 4 for the .4 1 functionor dominanceof .44to^4ifor the.44 function or
both. If the primary dominance in each case is complete, .4i.4 4 will just equal
the sum of .4i.4i and .44.44 in total effect beyond a neutral aa.

Overdominance may occur when: (1) aa is neutral and aA is nearer to an
optimum dose of A than is AA, (2) A' and A are both active for separate
supplementary functions and each is dominant to the other for its own
function (cf. East, 1936), (3) A' and A are both active for separate primary
functions, and the primary functions interact to produce an effect greater
than those of either A'A' or AA (Hull, 1945a).

Pseudo-overdominance may occur when A and B are linked: (1) with no
epistasis, aB and Ab combinations simulate the second case above, (2) with
positive epistasis aB and Ab simulate the third case.

If {aB X Ab) is superior to both (ab X AB) and (AB X AB), selection
may tend to tighten the repulsion linkage until ab and AB disappear and the
paired blocks are hardly distinguishable from alleles with primary over-
dominance.

It is clear enough that the frequency of heterozygotes is greater and of

RECURRENT SELECTION AND OVERDOMINANCE 459

homozygotes less for any locus with multiple alleles present in a crossbreed-
ing population. If heterozygosity should be of general advantage, multiple
alleles would provide more heterosis. East was at some pains to explain the
development of A 4 from A 1 by successive steps to the end of a superior hetero-
zygote. He apparently did not accept that hetero zygote superiority might be
general, with multiple alleles affecting vigor. I do not accept it either as a
likely proposition.

It seems likely that production of grain, meat, eggs, or milk may consist
of main effects and interactions of many, perhaps most, of the genes of the
plant or animal. Main effects must be of many kinds and magnitudes. Where
inbreeding depression and heterosis are evident there must be bias of positive
dominance or interactions of alleles to provide a gain in heterozygotes over
the arithmetic mean of homozygotes. Whether the interaction is basically a
stimulation of unlike alleles in the heterozygote, an interfering depressing in-
teraction in the top homozygote, or some other kind of interaction is an im-
portant problem in gene physiology. Present concern, however, is only with
the magnitude and frequency of the effect without regard to its basic physi-
ology.

Various writers have noted that dominance is not an absolute property. If
the phenotype is fruit size, degree of dominance is hardly the same for both
diameter and volume. The same genes might also affect stem length and ex-
hibit a third degree of dominance there. Gene effects are often greatly subject
to environmental fluctuations and to presence or absence of genes at other
loci.

Within reasonable limits of soil fertility and climate, grain yield of selected
homozygous corn is about 30 per cent of the yield of crossbred corn. Seventy
per cent of the apparent yield of crossbred corn consists of dominance effects
and perhaps of interactions of dominance with other gene effects. The 30 per
cent yield of homozygous corn consists of main effects and epistatic interac-
tions of main effects.

One difficulty in resolving the present situation without regard to how it
may have evolved is that the absolute zero of the genetic yield range cannot
be easily estimated. However, it might be assumed that it is less than zero on
our data scale. More specifically, the homozygotes with more than two-thirds
of the concerned loci aa or less than one-third AA may be inviable or have
an average yield potential of zero. The 100 per cent of measured yield then
would represent only the upper two-thirds of the total genetic range. With
dominance of high yield complete at each locus and the foregoing assumption
the present situation is adequately explained without resort to epistatis or
overdominance.

LINEARITY OF INBREEDING DEPRESSION AND HETEROZYGOSITY
Any appreciable degree of interaction of dominance with other gene ef-
fects might be detectable in a non-linear relation of inbreeding depression to

460 FRED H. HULL

predicted frequency of heterozygosity in succeeding generations of inbreeding.
Since the considerable body of data on inbreeding effects on yield of corn
fails to show any such non-linearity at all, I have been inclined to dismiss in-
teraction of dominance with other gene effects. Since, in addition, back-
crosses of Fi's to homozygous parent lines fail to show significant non-line-
arity I have been inclined to dismiss epistasis in general as an appreciable
part of the explanation of the disparity of yields of homozygous and cross-
bred corn.

Overdominance alone is an adequate explanation of the disparity. Pseudo-
overdominance from random linkage is not an adequate explanation by itself
since the totals of gene effects are independent ( f linkage (Hull, 1945a).

REGRESSION OF Fi YIELD ON YIELDS OF PARENT LINES

Corn breeders have frequently chosen a small sample (usually 10) of in-
bred lines and have made all or most of the specific crosses. Comparable
yield records on parent lines and Fi's have become available now in 25 sets
of data. F2 records are included with 3 of them. None of these data are
mine. Some of them were analyzed in part by simple regression of yield of Fi
on yield of parents, which would seem to provide the significant information
from the general combinability viewpoint. Interaction of parents is mostly
neglected.

Within each column or each row of a (10 X 10) table as described are nine
Fi's or nine F2's with one common parent. The common parent is the tester
of the other nine lines. Each line serves as the tester of one such group.

On the assumption that the partial regression of offspring on parent with-
in a group having one common parent is a relative measure of heritability
within the group, or of efficiency of the common parent as a tester, it has
seemed worth while to calculate all of the regression coefficients for individual
columns of the twenty-five Fi and three Fo tables. We tacitly accept that
yield may be a heritable character. Beyond this we need no fine-spun theory
nor any genetic theory at all to warrant direct regression analysis of the data.
However, Mendel's final test of his theory was with backcrosses to aa and
A A separately. He noted essentially that with completely dominant charac-
ters the expected regression of offspring phenotype on gene frequency of par-
ent gamete was unity with the aa tester and zero with the ^.4 tester. We may
be dealing with multiple factor cases of such testcrosses and of course with
different degrees of dominance at the several loci. The significant differentia-
tion of our homozygous testers may be in relative frequencies of aa and ^.4 at
the fli, ai, a^ — a„ loci.

Results with the first two examples are shown in Table 28.1. Yield of the
tester parent (P) is in bushels per acre. Directly below are the partial regres-
sion coefficients (bp) for the respective testers. Since there are apparently
negative trends of bp with respect to P, the second order regression (62) of

RECURRENT SELECTION AND OVERDOMINANCE 461

bp on P has been calculated. The second order regression function has been
solved for the special case bp = 0, to obtain an estimate of Fc the critical
value of P where the regression surface is level and heritability is zero.
The third summary in Table 28.1 is for average yields in six states of the

TABLE 28.1

REGRESSION OF YIELD OF Fi AND F. CORN HYBRIDS ON YIELD OF
INBRED PARENTS WITHIN GROUPS HAVING
ONE COMMON PARENT

Yield of parents (P) is recorded in bushels per acre, with the partial regression coefficient
(bp) below each one for the group of which it is the tester. The second order regression 62 is
regression of bp on F. Critical P (Fc) is estimated value of F for bp = 0.

Stringfield, G. 11. Unpublished. Ohio Agr. Exp. Sla. and USD A*

F 14 28 30 46 51 55

bp(FO .68 .41 .31 .22 .07 .05

bp{F.;) .55 .45 .33 .24 .26 17

Mean bp{Fi) .29, (F-,) .33; 6.,(F,) -.014, (Fo) -.008; Mean P 37; P,. 58; Mean F, 97;

Mean F2 70.

KinmuH and Spragiie, Agron. Jour. 1Q45*

F 3 15 20 26 28 28 32 39 40 50

6p(Fi) .63 .75 .84 .69 .13 .30 .25 .39 .22 .01

6p(F.,) .26 .36 .42 .69 .24 .29 .37 .58 .54 .47

Mean ip(F,) .42, (F.) .42; ^,(Fi) -.016, (F2) +.005; Mean F 29; P. 54; Mean Fi 80;

Mean F. 51.

USD A and State Regional Tests, Midseason 1943; Iowa, Kans., III., Ind., Ohio, Fenn. F values
from Kinman and Sprague above; their Fi's included here*

bp{Fi) -.05+.11+.08-.13-.20-.11 + .12-.01-.18
Mean bp -.01; 62 -.004; Fc 25.

* Sources of data.

same Fi's as those of Kinman and Sprague in Iowa. The Iowa test included
parent lines and F2's as well as Fi's. The third summary has been made with
Iowa records on parent lines. An analysis was made also of the Fi records for
each state separately with the same values of P. Regression trend was posi-
tive for the Indiana data, thus failing to support any theory of dominance of
high yield. Regression trends for the other four states were negative with esti-
mates of Pc all lower than the one for Iowa.

The eighteen other sets of data not summarized in the table are from Min-
nesota, Iowa, Illinois, Ohio, New York, and North Carolina. They are be-
lieved to be generally independent genetically and ecologically. Regression
trends are positive in eight cases. Taking the five cases summarized together
in Table 28.1, as five separate ones, we have seventeen with negative regres-
sion trend to eight with positive. Estimates of Pc for the seventeen negative
trends are near to or within the range of data as in Table 28.1 for each case
but one. With one of the least extensive tests the estimate of Pc is roughly
12 times the top inbred line, thus agreeing nicely with incomplete dominance.

Insofar as regression trends are due to heterozygosity they may be expect-

462 FRED H. HULL

ed to disappear with inbreeding of the crosses. The first two examples in
Table 28.1 are the two more extensive of the three cases which include F2,
and it is apparent that the negative trend of Fi has decreased or become posi-
tive in F2. It is positive in F2 of the third case also with a strong negative
trend in Fi.

The regularity of regression trends apparent in the first two examples in
Table 28.1 is by no means so readily apparent in any of the other twenty-
three cases. The eight cases with positive trends do not appear worse in this
respect than the others.

The possibility that the 10 inbred lines of Kinman and Sprague do not
comprise a representative sample has been tested by dividing the 10 into two
groups of 5 each in various ways. This provides a 5 X 5 table in each case
with a unique sample of 25 F/s from the total of 45. These 5X5 tables do
not have vacant cells which arise when one parent line is included on both
margins of a table. Each tester in one group is rated with the same five lines
in the other group. Estimates of b^ and Pc from such 5X5 tables have con-
sistently substantiated those reported in Table 28.1, for the 10 X 10 table.

Analyses of six of the twenty-five cases have been done also with loga-
rithms of P and Fi records, with results generally in agreement with those of
the original data.

Most or all of the individual values of bp and 62 are not statistically sig-
nificant. The distribution of the twenty-five 62's is distinctly bi-modal. Eight
are positive indicating dominance of low yield, one is negative and small
enough to indicate intermediate dominance of high yield. Sixteen are nega-
tive and decidedly in the overdominance range. No explanation of the bi-
modality is apparent now. The eight positive values of 62 are in some degree
suspect since they are inconsistent with so many facts. All of these tests could
be repeated with the same unique samples of genotypes insofar as the parent
lines were homozygous and are still available. We need more comprehensive
and precise data.

Present evidence from regression analysis is slightly in favor (2 to 1) of the
conclusion that a zone of nearly level regression, nearly zero heritability,
exists near the upper end of the range of present data. This conclusion would
be more consistent with the failure of selection for general combinability if
it should be that selection for specific combinability should favor aA over
AA, and thus tend to degrade gene frequencies below that equilibrium where
heritability and regression change from positive to negative.

GENETIC INTERPRETATION OF THE REGRESSIONS

The problem of genetic interpretations of bp and 62 may be approached
first with the simpler case of no epistasis. Consider the multiple gene set
aiAi to OnAn- Let {I — v) and v be relative frequencies of a and A in the
gametes of P, with respect to the n loci, and w similarly for Pj. The product

RECURRENT SELECTION AND OVERDOMINANCE 463

of the two gametic arrays provides expected frequencies of aa, aA, A A in
Fi(P, X Pj) with respect to n loci.

Fi'^n (l — r) {1 — iv) aa-{-n[v {\ —w) + w ( 1 — v) ] aA -\-nvwA A
Define' phenotypes:

aifll C2O2 • • • Ondn = T

a A = T -\- d -^ kd

A A = T -\-2d

Fi = niv-\-w){d-^kd) -n2iw{kd) +T (1)

This is the regression of Fi phenotype on gene frequencies of parents and is
independent of degree of inbreeding of parents.

If P, is homozygous it has ;/(! — v)aa and nvAA loci.

P, = nvi2d) +T, v= {P,-T)/n2d

(2)
Pj=nw{2d) ^-T , w= {Pj-T) /ti2d

Substituting for v and iv in (1)
Fi= {\ + k + kT/nd) (P,-\-Pj)/2-ik/ n2d) iP,Pj)-kr~/ n2d-kT (3)

This is the regression of Fi phenotype on phenotypes of homozygous parents,
the equation of a surface in three dimensions, Fi, P,, Pj. The surface is a
plane if 62 = k/n2d is zero, if ^ = 0, if there is no dominance, no inter-
action of P, with Pj. Then, Fj = (P, + P,)/2, by setting ^ = 0 in (3).
Taking P, constant as the common tester of one column of the regression

table,

Fi= [i(l + yfe) -k(P,-T)/n2d]Pj + C' (4)

bp is the coefficient of Pj, within brackets,

bj,= i-k/n2d)P,-j-^{\-^k) -{-kT/)i2d . lik = 0, bp=\

Regression of bp on P, is 62 = —k/u2d. Since P, == iiv{2d) -\- T, bp =
i(l + k) - kv. If bp - 0,

v^ {\Jrk)/2k (5)

With no dominance,

^ = 0, I' =1/0 at equilibrium

1. T, d, and kd are defined here in units of bushels/acre or pounds/plot, for example.
Then, k = kd/d is in units of (bu./A)/(bu./A), likewise bp, but 62 is in units of l/(bu./A),
making the whole term boPiFj in bu./A.

In terms of selective values it is convenient to define ds in terms of number of progeny
surviving to breed. Then, ks may be greater or less than k, depending on artificial breeding
plan. If roan in Shorthorn cattle is intermediate, k is essentially zero, but if roan is favored
in artificial selection over red and over white, kg > I and there is overdominance with re-
spect to artificial selective values.

With corn yield no single locus is identified, no heterozygote may be favored to pro-
vide ke > I, except that k > I. Then, ks may depend on gene frequency and on rate of
culUng.

464 FRED H. HULL

Complete dominance

^ = 1 , V = I at equilibrium

k = 2 , V = 3/ 4 at equilibrium

For the more general case where Pi and Pj are (not inbred) individuals in
a crossbreeding population, equations paralleling (2), (3), and (4) are second,
fourth, and second degree, respectively. The simplification obtained with
homozygous parents is reduction of the three functions to first, second, and
first degree, respectively, by removing dominance effects (allelic interactions)
from parent phenotypes Pj and Pj. Mendel found the simplification obtained
with homozygous parents to be of considerable value in his early studies of
monogenic inheritance.

The Mendelian model (2), (3), (4) may be complicated with innumerable
kinds of interactions (epistasis) by simple, compound or complex transforma-
tions (log, anti-log, exponential, etc.) of (2), (3), and (4). It is not intended
to imply, however, that interactions of alleles must precede interactions of
non-alleles in living organisms.

The estimate of bp for any tester parent line is independent of gene fre-
quencies of the other parent lines with respect to dominance interactions. If
obtained estimates of bp for the same tester with samples of weak and strong
lines respectively should differ significantly, the necessary interpretation
would seem to require some kind of interaction other than between alleles,
or that the lines were not strictly homozygous.

Interpretation of bp, b^, etc., by the Mendelian model presented here will
not be biased by linkage of two loci if frequencies of ab, aB, Ab, and AB do
not deviate significantly from expectation from random recombination of
gene frequencies of the two loci with respect to all of the parent lines. Free
assortment of the two loci is then effectively simulated. But any union of two
unlike gametes must contain some cases of repulsion linkage close enough to
retain the aB and Ab combinations in high frequency through several gen-
erations. A sample of lines all derived directly by selfing from one heterozy-
gous parent plant may well contain many cases of repulsion linkage to simu-
late overdominance. This effect would not be counterbalanced by high fre-
quencies of coupling linkage of other pairs of loci. Lines within each of the
25 samples reported here are in most or all cases no more closely related than
plants within one or more varieties.

Variations of d and k from locus to locus would contribute to total vari-
ance, but would not seem to impair seriously the validity of the estimates of
regression coefficients, nor of Pc when bp — 0.

When all loci are aa or all loci are A A, Pi = Pj = Fi = F2 = x. With
this restriction (3) becomes a quadratic with roots equal to the phenotypes
at the two limits. The difference is n{2d), the genetic range, the denominator
of 62 = —k/nld. Values of k, calculated thus, for the nine cases where parent

RECURRENT SELECTION AND OVERDOMINANCE 465

and Fi yields are strictly comparable and b-i is negative, are: 2.25, 1.50, 1.88
(2.18), i.83, 1.78, 2.45, 1.41, 2.25, 1.69. The 1.88 (2.18) entry is Fi and F2 re-
spectively of Stringfield's example. Table 1. The value 1.09 from F2 data was
doubled to correct for the effect of inbreeding.

If these independent estimates of k should be unbiased operationally, we
must still be cautious in attempting any unique physiological interpretation.
All of the several types of apparent overdominance listed here and others too
may be operating in corn yield.

Estimates of backcrosses 5, and B, may be written by inspection of (1)
and (2). Fi is transformed to F2 (by selfing Fi) by multiplying the coefficient
of each k term in (1) by h. This provides three linear sets Fi, Fo, bar P; Fi,
Bi, P„-and F:, Bj, Fj, on the assumption of no epistasis. Fi, F2, and P are
alike in gene frequency. They differ only in frequency of heterozygosity.
Differences in the backcross comparisons arise from both gene frequency and
frequency of heterozygosity.

GRAPHIC TRANSFORMATIONS TO REMOVE EPISTASIS
Where the two intervals in any one of the three comparisons are not equal,
epistasis may be suspected and a transformation of data may help to elimi-
nate some of its effects. No transformation of the corn yield data would be
warranted by all of the considerable amount of published data I have found,
since the data fit the linear hypothesis very closely with F2 and backcross
comparisons.

Where transformation is clearly indicated, I may suggest a graphical de-
termination of the best function. Plot the data. Pi, Pj, P, Fi, F2, Bi, Bj, and
B on the vertical axis, and the same values on the horizontal axis linearly
with no dominance, with any arbitrary scale. If the plotted points do not
seem to provide a smooth curve, move Fi to the right a trial distance. Move
F2, B^, Bj, and B the same direction one-half as far. Move to the right or left
(Fi twice as far as the others each move) until the best fit visually to a
smooth curve is found as the best transformation function. The only excuse
for suggesting such a crude process is that if it is carefully carried out with
good data the function is so much more refined than any arbitrarily chosen
function for the purpose of correcting a complex of different kinds of epistasis
together.

The transforming function determined by the above process with all avail-
able data on grain yield of corn would not differ sensibly from a straight line.
From this I have said earlier that epistasis is unimportant in corn yield. Con-
siderable amounts of increasing and decreasing returns types of epistasis may
be effectively balanced, of course. In that case, epistasis would provide no ex-
planation of the disparity of inbred and crossbred yields.

MAXIMUM YIELDS FROM CROSSING HIGH BY LOW?
For four loci with v and w = f , the gametes are aiAiAsAi, Aia^A^Ai,
AiAiazAi, AiAiAidi. Equations (1) or (3) with appropriate substitutions

466 FRED H. HULL

calculate the mean of the 16 Fi combinations of four gametes of equal gene
frequency. Deviations of the individual Fi's from the mean are not predict-
able from parent phenotypes. They are due to specific combinability arising
from varying frequencies of heterozygosity. No more than two loci can be
heterozygous in this example. But iiv= f , w = j, six of the sixteen Fi's are
heterozygous at all four loci. In the event of overdominance ^g of high X low
combinations may exceed the best high X high combination. If 1 < ^ < 2,
and V = I, the mean of high X high is greater than the mean of high X low.
From the general combinability viewpoint we see only the difference of
means. Selection of the very few elites among specific Fi's would, however,
find them more frequently in high X low combinations. Hayes and Immer
(1942, Table 21) present data of Johnson and Hayes which seem to agree
with this interpretation in that the mean of high X high is best, but the
highest specific combination is more likely in high X low.

EQUILIBRIUM FREQUENCIES OF GENES

We may substitute for v in equation (1) the mean gene frequency of a
group of lines or of a variety, a general tester, to be held constant. Then if
V is less than (1 + k)/2k, and ^ > 1, regression of Fi on w is positive. Selec-
tion for general combinability with the same tester should continue to fixa-
tion of A except for reverse mutation. But if selected lines are recombined
for each cycle and the recombination is the tester for the next cycle, selection
comes to equilibrium when gene frequency of the tester reaches (1 -\- k)/2k,
short of fixation if ^ > 1.

If concurrent with the foregoing process there should be selection of the
high specific combinations (high X low) with lower gene frequencies, the
combined effect on gene frequency may be nil. It may even be to degrade
gene frequency when gene frequency is so near the equilibrium that herita-
bility of general combinability is weak. From this view we may expect in the
event of overdominance to find the equilibrium zone nearer the upper end of
the range of data, providing some degree of positive heritability, some de-
gree of positive regression of Fi's on inbred parents.

Ear-to-row selection should have progressed toward equilibrium gene fre-
quencies except for the counter effect of selection of superior plants within
ear-rows and within recombinations, selections of elite specific combinations
of two gametes with above-average heterozygosity and lower gene frequency.

Modern corn breeding is failing largely beyond the first cycle for the same
reasons that caused the failure of ear-to-row selection, except that inbred
lines provide for a more efficient identification of elite specific combinations
which may have the lower gene frequencies.

The whole of the evidence fits the generalized Mendelian model neatly
enough if we may accept overdominance and otherwise proceed without
prejudice to those conclusions more consistent with the data.

RECURRENT SELECTION AND OVERDOMINANCE 467

In familiar theory, selective advantage of a heterozygote leads to an
equilibrium gene frequency in natural selection, where every individual
leaves progeny (no culling) in proportion to fitness or where fitness is in fact
fertility or more specifically, number of offspring surviving to breed. We
must distinguish now between k for a physical trait, kna for natural selection
of the same trait, and k, for artificial selection. Since there is little apparent
difference between bushels per acre and potential number of offspring surviv-
ing to breed, it may be supposed that k and kns are about the same for yield
factors in corn. But if ^ > 1, artificial selection including strong culling may
make k^ appreciably greater than k, and (1 + ks)/2ks appreciably less than
(1 + k)/2k. The expected effect of any single cycle of artificial selection is
to shift gene frequency towards (1 + k^/2ks, '\i k > 1. The operator's suc-
cess (measured by k^) in culling out homozygotes will improve as gene fre-
quency approaches § and frequency of aA approaches maximum. The limit
is reached when ks is infinite, and gene frequency is (1 + oo)/2co or |; e.g.,
as when saving only roan Shorthorns for breeding stock. The roans then have
infinitely more progeny than whites or reds, which have none.

It does not seem likely that the limit equilibrium oi q = \ can be reached
or maintained with multigenic complexes such as corn yield, because of ina-
bility to cull absolutely all homozygotes. On this theory, strong selection
will seem to degrade vigor. Relaxation of selection may allow vigor of the
corn variety to improve. But there may be important loci where overdomi-
nance does not obtain, which tend to obscure the overdominance effect.

If artificial control should maintain fertility continually proportional to
the physical trait where k > 1, gene frequency should progress to equilibrium
at (1 + k)/2k; cf. recurrent selection for general combinability for corn
yield. The population mean is maximum for the physical trait when q =
(1 -f k)/2k.

If overdominance should be important in vigor of cattle at a number of un-
fixed loci and a herd is close to (1 + k)/2k for those loci, mild culling of fe-
males would tend to raise gene frequencies above (1 + k)/2k. Strong culling
of males might have the opposite effect. Founding an elite herd with choice
females from many herds and an expensive bull might be more likely to de-
grade gene frequency below optimum in the event of overdominance. The
offspring of the choice animals might be disappointing aside from expected
regression towards the mean of the breed.

EFFECTIVENESS OF RECURRENT SELECTION
Most of the selection practiced with plants and animals is recurrent. Ex-
ceptions are selection among homozygous lines or among clones. Inbreeding
may curtail the efficiency of recurrent selection by lengthening the cycle.
Selection within inbred lines during the process of inbreeding is recurrent but
inefficient to the extent that freedom of recombination is curtailed. I have

468 FRED H. HULL

suggested before that breeders of self-fertilized crops might find greater effi-
ciency in more frequent recombinations. It was to emphasize these considera-
tions that the term recurrent selection was introduced. The sense of recurring
back to the same tester was never intended.

Breeders of open-pollinated corn need to save no more than 1 ear from
500 or more to plant the same acreage again. If selection is only 20 per cent
effective, the net effect in ten years is {yqY^. The number of corn plants
grown in the world in one year is roughly (10)^'. In 100 million times the
world acreage of corn there might be one plant as good as the farmer's whole
field after he has done 10 to 12 years of recurrent- selection. That this seem-
ingly fantastic theoretical concept is essentially correct is supported very
well, I think, by results of selection for oil and protein of the corn kernel in
the well-known Illinois experiments and in many other less well documented
cases with animals, too. East has proposed that selection for oil and protein
in corn might be more efficient with inbred lines. However, East proposed
that Si lines from the selected ears after chemical analyses be recombined for
another cycle of selection. He employed inbreeding only to avoid open-pol-
lination of the ears to be analyzed. It is unthinkable that East meant to pro-
pose that selection within and among inbred lines for oil or protein without
recurrence of selection should be the more effective process.

Open-pollinated corn varieties of 50 or 30 years ago were actually pretty
good, in yield and in many other respects. The selection differentials by which
they were isolated were probably enormous. Nevertheless, specific combina-
tions of inbred lines are sometimes 20 to 30 per cent above the varieties in
yield. That this gain is mainly due to higher frequencies of dominant favor-
able genes in the elite inbred lines isolated from only a few hundred without
recurrence of selection is really inconceivable.

A single corn plant in the variety is a product of two gametes. An Fi of
two homozygous lines is a product of two gametic types. The plant and the
Fi are genetically the same in mean, variance, and expectation of homo-
zygosity in advanced generations as well as the first. It should not be diffi-
cult, if asexual propagation were possible, to isolate from the single plants
clones that are easily superior to the present elite Fi's. That the reservoir of
specific combinability in corn is far from exhaustion by present hybrids is
evident in comparisons of Fi's with the range of individual plants in varieties.
The animal breeder may look upon a family of full sibs (from four grand-
parental gametes) as a double cross of unselected but homozygous lines, for
a rough estimate of possibilities with hybrids. But, aside from that, the
breeder of open-pollinated corn was selecting among specific combinations of
two gametes the same as in selection among Fi's. Continued selection within
varieties might have degraded gene frequency below (1 + k)/2k at any locus

2. Cf. Huxley, Genetics in the 20th Century, p. 595. "Recurrent selection," natural or
artificial, is designed to multiply improbabilities; requires heritability in the strictest sense.
Selection among inbred lines may go on and on without "recurrence."

RECURRENT SELECTION AND OVERDOMINANCE 469

where k > \, thus providing the ])ositive mild regression of offspring on par-
ent, the heritability which so many have taken as strong evidence against
k> I.

Many traits of the corn plant are mostly independent of genes concerned
with yield. Many others may be optimum for yield at intermediate points
genotypically as well phenotypically. It should hardly seem surprising if,
subsequent to intense selection for yield, we should find evidence of inter-
mediate gene frequencies and very little inbreeding depression or heterosis
with such characters. An intermediate optimum may place some premium on
aA, but hardly to the extent of explaining the evident heterosis of corn
yields, so far as I can see.

Evidence cited here of overdominance in the genetics of grain yield of corn
consists of:

1. Failure of mass selection and ear-to-row selection beyond the level of
the adapted variety.

2. Crossbreeding recombinations of parent lines of elite hybrids yield
little more than the original varieties.

3. Hybrids of second-cycle and third-cycle lines yield little more than those
of the first cycle.

4. Homozygous corn yields 30 per cent as much as heterozygous corn.

5. No evidence of epistasis in corn yield.

6. Regression analyses of yields of Fi's and inbred parents indicate a zone
of nearly level regression near the upper end of the range of present data,
where it might be predicted with the kind of artificial selection which has
been practiced, and in the event of overdominance.

7. There is some evidence that selection for general combinability alone
with respect to yield is effective and this too is consistent with the expectation
of overdominance theory.

8. The fact of hybrid corn is hardly to be explained as other than a result of
selection for specific combinability, which in turn is manifestly dependent on
heterozygosity of corn yield genes.

My proposal (Hull, 1945a) that recurrent selection for specific combina-
bility be given a trial was made on the assumption that recurrent selection
for general combinability or for accumulation of dominant favorable genes
had been fairly tried in mass selection and subsequently. The tentative con-
clusion was that varieties (and breeds perhaps) were near equilibrium, with
mean gene frequencies approximately at (1 + k)/2k. Regression analyses a
little later indicated that the corn samples were below equilibrium. Since then
it has been proposed orally many times that two parallel breeding plans re-
stricted respectively to specific and to general combinability might well be
run with corn and with small laboratory animals as pilot experiments. I have
later come to believe that recurrent selection among homozygotes might
also provide results of considerable theoretical interest.

470 FRED H. HULL

Present-day corn breeding is done in three steps: selection among inbreds
based on their own phenotypes; selection among inbreds for general combina-
bility ; selection among specific Fi's of the remaining inbreds. These steps are
the three processes of the preceding paragraph. The corn breeder applies the
three processes in the order named to the same stock, then recombines the
elite lines and begins the cycle again. The present proposition is to apply the
three processes separately to parallel stocks, and thus attempt to learn which
ones are primarily responsible for superior hybrids.

RECURRENT SELECTION AMONG HOMOZYGOTES

This process can be done effectively enough in corn, perhaps with S2 lines.
Two selfings would amount statistically to reducing the degree of dominance
to one-fourth of the original value. One-half of the Si lines could be discarded
in the first comparison. About fifty S2 lines should be retained in the recom-
bination. Selection within ear-rows should be rigidly excluded.

There is no reason to suppose that a physiological barrier would be reached
short of the level of elite hybrids. Recurrent selection towards an extreme has
been very effective with many characters where not much dominance is ap-
parent. In noted cases no limit of genetic variance has been reached. What
genetic limit might be reached with vigor or yield genes of corn when the con-
fusion of dominance is artificially eliminated is to be explored. Theoretically,
this process of recurrent selection should be much superior to any non-recur-
rent selection among gametes or doubled haploids.

RECURRENT SELECTION FOR GENERAL COMBINABILITY

Strictly, the tester should be the variety. So plants or S„ lines are to be
testcrossed with several plants of the variety. The So plant must be selfed at
the same time. Parents of elite testcrosses are recombined into an improved
variety which becomes the tester for the next cycle. If gene frequency of the
variety is improved to approach (1 + k)/2k, where ^ > 1, heritability will
approach zero and the variety mean its maximum. If pseudo-overdominance
from repulsion linkage is important the equilibrium may advance to higher
levels as recombinations occur. But, aside from that, we have now no experi-
mental verification of a selection equilibrium, and a test would seem desir-
able. Concurrent selection for specific combinability should be strictly avoid-
ed in this test.

RECURRENT SELECTION FOR SPECIFIC COMBINABILITY

This process has been adequately described both here and earlier (Hull,
1945a). From the theoretical viewpoint it would be best to use a homozygous
tester and avoid selection within the crossbred except that based on testcross
performance. The purpose is to determine first how much specific combina-
bility may be accumulated in early cycles and eventually to determine where
this process may reach physiological or genetic limits.

RECURRENT SELECTION AND OVERDOMINANCE 471

Now if we are convinced that overdominance is not very important and
that, perhaps for other reasons too, selection for general combinability will
eventually win, or at least not lose, we may proceed at once with recurrent
selection for general combinability to render hybrid corn obsolete. Some of us
may find it necessary to include an inbreeding interphase between cycles.
Breeders of livestock may as well return to improvement of pure breeds by
progeny testing. We will run these pilot tests merely for the sake of verifica-
tion.

But if it should seem likely that recurrent selection for specific combina-
bility may win, the breeder of livestock may begin now with recurrent recip-
rocal selection for specific combinability. For my part, I would choose two
crossbreds for the start and would begin mild inbreeding in one of them which
would become the stud herd. On one side of this is the Comstock plan with
no inbreeding in either herd. On the other side we might choose a line with
50 per cent inbreeding at the start and practice reciprocal selection along with
continued mild inbreeding. Evaluation of these alternatives of the reciprocal
plan with small laboratory animals, along with the other two main plans,
would be of considerable interest theoretically. The cost might be minute in
comparison with the total of wasted effort in current breeding practices.

Recurrent selection for general combinability without the inbreeding in-
terphase is a fairly obvious technic which has been employed and described
variously. The first discussion of it from the overdominance viewpoint with
the restriction against selection for specific combinability was that of Hull
(1946b). Since then I have continued to urge parallel tests with fast breeding
species as pilot experiments. Recurrent selection for superior homozygotes is
proposed here for the first time, I believe.

Reciprocal selection for specific combinability was a counter proposal to
me of several corn breeders in 1944 and later, when I proposed selection in a
crossbred for combinability with a fixed tester, a homozygous line or Fi of
two homozygous lines.

For simplicity of illustration we may consider a 4-factor example with gene
frequency in a homozygote or gamete (v or w) taking values, 0, |, f , f , |.
Gene frequencies intermediate to these values may occur in heterozygotes
and in whole populations. Let us take ^ = 2 for the degree of dominance as
suggested roughly for corn yield by estimates reported here. Then regression
of ofi'spring phenotype on gene frequency of parent in any column of the
(5 X 5) Mendelian checkerboard is bp = ^(3) — 2v, where v is gene frequen-
cy of the common parent of the column. Substituting the five values of v pro-
vides the five values of bp, 1^, 1, |, 0, — |, for the five columns. Heritability
changes from positive to negative where v ^ (I -\- k)/2k = f . These values
of bp for the given values of v are the same for any number of loci. In any
case the zone of near-zero heritability for one locus is relatively broad on both
sides of the critical value of zero. Reciprocal selection between two crossbreds

472 FRED H. HULL

is at equilibrium for one locus when gene frequencies are (1 + k)/2k in both,
and k> 1. It is conceivable that gene frequencies of the two crossbreds may
wander in the zone of low heritability through many cycles of reciprocal selec-
tion, but they must eventually separate on opposite sides to approach aa and
AA respectively with increasing velocities. When the two gene frequencies
are on opposite sides of the equilibrium initially, reciprocal selection will tend
to drive them farther apart. If they are on the same side both will tend to ap-
proach equilibrium. Comstock's statement here that the one nearest equilib-
rium may approach it more rapidly and continue beyond to reverse the trend
of the other, thus obtaining a quick separation, seems good. I had overlooked
this point and hope it may be experimentally verified.

Gametes with critical gene frequencies in the present model are aAAA,
AaAA, AAaA, AAAa. A general tester composed of the four homozygous
lines producing these four gametes respectively will provide zero heritability.
So also will a crossbred tester for every locus where gene frequency is f . One
of the homozygous lines alone as a specific tester provides mean bp = 0 —
[f + 3( — 1)]/4. But here the individual values of bp for each locus are at
maximum, | for the aa locus, and — | for each A A locus, providing maximum
heritability in selection to a homozygous tester.

Defining phenotypes of aa, aA, A A alternatively as 1 — .y, 1 — hs, 1, pro-
vides bp = \ — h — {I — 2h)v. Then with // = — | for the same degree of
dominance as the present model, bp = ^3) — 2v again. The only inconsist-
ency between the two systems of defining phenotypes which may be encoun-
tered here, I think, is failure to distinguish between physical values and selec-
tive values, e.g., body weight and number of offspring surviving to breed.

It seems fairly clear that overdominance of the degree considered here may
provide considerable variation of heritability within a finite sample, a herd
or a variety on one farm. Mean bp may be positive and fairly large, yet bp =
0 near the upper range of gene frequency in the sample. Moreover, the degree
of dominance for selective values might be appreciably greater than for the
physical trait. For these reasons, selection indexes made up with average
heritabilities of physical traits could be misleading.

Parallel operations of the foregoing breeding plans with heavy dosages of
mutagenic agents in addition might provide significant information on pro-
gressive improvement, where the objectives respectively are the superior
homozygote, the mean of the population, and the superior heterozygote.
This proposal will be subject to criticism by those who are convinced that it
is only in gene-by-gene analysis that real advances in knowledge of genetics
can be obtained. I have no quarrel with that viewpoint except that where
many genes with minute efliects may be involved the gene-by-gene approach
still seems fairly remote.

Recurrent selection in prolific species such as corn, chickens, mice, and
Drosophila may soon build up very large selection intensities, perhaps to re-

RECURRENT SELECTION AND OVERDOMINANCE 473

cover high frequencies of rare natural or mutant alleles. Chemists have em-
ployed high pressures and temperatures to obtain reactions of great interest.
They have concentrated rare elements and rare isotopes by various ingenious
processes. With selection intensities and mutation rates well above natural
values it might be possible to obtain estimates of the minimum ratio of selec-
tion to mutation for survival or improvement of the variety or breed.

JOHN W. GOWEN

Iowa Stafe College

Chapter 29

Hybrid Vigor in
Drosophila

Experience has defined hybrid vigor as the evident superiority of the hybrid
over the better parent in any measurable character as size, general vegetative
vigor, or yield. For any one species it is left for us to show that, within the
possible crosses of pure lines, hybrid vigor actually exists and what particular
morphological and physiological characters express it best.

With this in mind, investigations with wild-type Drosophila of diverse
geographical origin were begun in 1934 and continued to date. The group
working on this problem has included Dr. Leslie E. Johnson, Dr. F. S. Straus,
Miss Janice Stadler, Dr. S. Y. Loh, and myself. The material reported here
is the result of our joint efforts. To specify the problem of hybrid vigor, five
characteristics were chosen for investigation in eight inbred lines of Drosoph-
ila and a hybrid between two of the lines. The characteristics chosen were
egg production throughout the full life of the fly, the days the females laid the
eggs, the hatchability of the eggs, and the duration of life of the males and
females in each line.

To determine egg production, a pair of flies of a particular line was placed
in a quarter-pint milk bottle sealed with a paraffin paper cap on which was
placed a disk of nutrient banana agar colored with charcoal. The female laid
her whole day's egg output on this disk when it was properly seeded with
yeast and a little acetic acid. The caps, a sample shown in Figure 29.1, were
changed daily and the eggs were counted for each day.

The characteristic performances of the different pure lines and the hybrid
are shown in Table 29.1.

The average egg production for the different inbred races varied from 263
to 1701 eggs. There is some correlation between the intensity of the inbreed-
ing and the production of the particular race. Ames I and II are less inbred

474

HYBRID VIGOR IN DROSOPHILA

475

races than Inbred 92 or Homozygous. Correlation exists between the egg pro-
duction of the race and its fitness to survive as judged by its duration of life
as measured by the survival of either males or females.

The hybrid race came from the cross Inbred 92 and Ames I. The mean
productions of the parents were 389 and 1000 eggs respectively over the life-

FiG. 29.1 — Photograph of laying cap with eggs and some hatched larvae.

time period. The mean production of the hybrid was 2034 eggs, or 203 per
cent greater than that of its high producing parent, Ames I, and 422 per cent
greater than that of its other parent. Inbred 92. The hybrid showed more
eggs than any of the pure races. The excess of the hybrid over the pure par-
ents is greater in this particular experiment than in several other similar ex-
periments.

The females of the different races showed average egg laying periods of
17.2 to 46.0 days. The hybrid lays eggs nearly as long as the higher producing

Alb

JOHN W. GOWEN

purebreds, 43.4 days, but does not exceed the range. The parents entering
into the cross for this hybrid laid eggs for an average of 38.4 and 17.2 days
respectively. Hybrid vigor is only 113 per cent for the length of the produc-
tive period.

The physiological fitness of the purebred races, as measured by their dura-
tion of life, ranged from 22 days to 58 days. The average life of the hybrid
was 52.2 days. The hybrid's parents lived for 38.7 and 50.2 days respectively.
This character showed little hybrid vigor — 104 per cent.

These results show that egg production is the most favorable of the
Drosophila characters analyzed for the study of hybrid vigor. The lifetime

TABLE 29.1

CHARACTERISTIC VARIATION OF DROSOPHILA RACES
IN EGG PRODUCTION AND DURATION OF LIFE

Survival of

Days Females

Survival of

Egg Production

Females in

Males in

Laid F.ggs

Race

Days of Life

Days of Life

No.

Mean

No.

Mean

No.

Mean

No.

Mean

Ames II

56
56
56
56

1701

1511

814

389

48
41
45
43

56.1
51.5
48.4
33.4

56
56
56
56

46.0
40.0
35.4
17.2

30
39

43
42

58.7

Ames II

53.6

Princeton

46.7

.->Inbred92

44.0

Florida-45

54

610

49

28.5

54

22.4

52

32.6

Oregon R-C-44.. ..

54

413

49

36.4

54

28.7

48

35.5

Swede-b-40

53

398

50

26.7

53

16.5

50

35.7

Homozygous 42 . . .

54

263

54

22.7

54

16.7

51

27.9

Ames I

54

1000

51

50.9

54

38.4

50

49.6

Hybrid

54

2034

52

50.0

54

43.4

51

55.4

Pooled Variance

d/f537

236847

d/f537

179.5

distributions of egg productions for the inbred and hybrid races are shown in
Figure 29.2.

Newly hatched females require a short period after emergence for maturing .
Heavy egg production begins on the fourth day and rises rapidly to a maxi-
mum in early life. From the high point, production gradually declines. The
rate of this decline varies with the different races. The average slope is shown
by straight lines.

Drosophila egg production presents a single cycle as contrasted with the
series of cycles or egg clutches observed in the egg production in certain other
forms — the domestic fowl or the fungus fly, Sciara. This fact makes Drosoph-
ila egg production an easier character to study. The egg yield curve is deter-
mined by the initial high point in production and the rate of loss in produc-
tivity with age.

The form of the egg production curve in Drosophila fits in with Ashby's
hypothesis of metabolic reserves being responsible for hybrid vigor. The hy-

100

10

^

^

^

^

^

VMES n

s

}

k

<

Q

(E
UJ

a.

Q

UJ

o

a
o

(T
Q.

If)
O
O
UJ

JU

10
1

Ar-

=^

V5

S

,ME

5 H

NT*

>

r

100

10

>

^

V

PRII

ICET

ON

\/

vw

^

\

100

<

UJ

10

(^

k

INB

RED

-92

A

^

\

JU

K

V

^

FL

ORIC

A

lU
1

\

100

10

r

\

OREGON

V

%

\

V

100

10

f

f^

\

SWEDE

1

\N^

s

\

100

100

10

r

N

^

VfV

AMES 1

V\

\^

100

10

r

^

^

V

^

HYB
1 li

?1D

"^

•^

's

0 20 40 60 80 0 20 40 60 80

DAYS OF LIFE (FEMALE)

Fig. 29.2 — Lifetime daily egg production of different races of Drosophila melanogasler,

number of initial females tested, 53 to 56.

478 JOHN W. GOWEN

brid has a higher initial production than its parents, or, for that matter, any
of the purebred races. The hybrid expends its metabolic reserves less rapidly
than either of its parents. Taking all inbred races together, the hybrid utilizes
its reserves at slightly less than the average rate. The hybrid is chiefly char-
acterized by its large initial production. Examination of the pure lines indi-
cates that there are slight differences in the rate of expenditure of the initial
reserve, even when the obviously different Inbred 92 is not considered.

WHAT IS HYBRID VIGOR?

These results show that the vigor of the hybrid is greatest for lifetime egg
production, 203 per cent; is less if length of egg laying period is considered,
113 per cent; and is still less with life span as the character, 104 per cent.
What is the explanation of egg production's high heterosis? Egg production
is a character which is in turn dependent on other component characters. A
simple breakdown would be, lifetime egg yield is determined by the length
of egg laying period, the decline (slope) representing the loss of ability to
produce eggs with age, and maximum egg production at the initial phase of
the egg laying cycle. Length of egg laying period has already been shown to
have 113 per cent heterosis. The slopes of the decreasing egg yields with ad-
vancing ages are Inbred 92 — 0.17, Ames I — 0.06, and their Fi hybrids — 0.05.
The hybrids show heterosis in that their egg productions decline less rapidly
than their best parent, but the heterosis is only 121 per cent.

Maximum productions, as judged by the three highest days' average
yields of the strains, are Inbred 92, 40.7; Ames I, 52.7; and Fi hybrid, 81.4
eggs, or the heterosis is 154 per cent. The highest of the component heterosis
values is only about half of that noted for lifetime egg yields. It seems not
unlikely that if the division into components could be carried further, it
would be found that the heterosis values would approach closer and closer
to 100. The results consequently argue for heterosis, as the result of the com-
bined action of two or more groups of distinct characters which, when jointly
favorable, and as frequently truly multiplicative in action, lead to heterosis.

Analysis of the variation in egg production between races — the heritable
fraction controlled through inbreeding — as contrasted with the variability
within races — the fraction due largely to environment — shows that about
56 per cent of the lifetime egg production is fixed within the races and 44 per
cent is due to gene segregation, environment, etc.

Consideration of the individual records further support this view. Con-
trasting the performances of the individual females within the hybrid groups
with those in the different inbred races shows that the hybrid has no females
with greater production than those of the inbred races. The hybrids are good
because, on the average, all members of the cross are good producers. The
hybrids include one female laying 3083 eggs and twenty-seven others laying
between 2000 and 3000 eggs. The Ames I has one female laying 2016 eggs;

HYBRID VIGOR IN DROSOPHILA 479

Ames II, in two similar experiments, has two females laying 3168 and 3108
eggs and thirty-two others laying between 2000 and 3000 eggs. The other
pure races have no individuals laying more than 2000 eggs. Hybrid vigor con-
tributes consistently high performance to all individuals rather than excep-
tional performances to a few. It is the consistency of high performance which
calls for explanation.

MEASURES OF HETEROSIS

As lifetime egg production is a difficult character with which to work, a
less tedious measure of productivity was sought. The character chosen was
daily egg yield 5, 6, 7, 8, and 9 days after the female emerged from the pupa.
These records are at the general maximum of the female's productive life.
The correlation with lifetime production is high.

Chromosomal and Cytoplasmic Basis for Hybrid Vigor

The possibility of creating homozygous races of Drosophila through out-
crossing offers a unique opportunity for analyzing the causative agents be-
hind hybrid vigor. Hybrid vigor has been postulated as due to differences in
allelic genes and to differences in the cytoplasms which combine at fertiliza-
tion. The reduction in yield of inbred races is accompanied by increasing
identity in both the combining alleles and the cytoplasms which combine to
form succeeding generations. Both these factors have been invoked to explain
the low yield of such inbred races. Production of homozygous tya^ through
outcrossing furnishes a contrast between these two possible causes of low pro-
ductivity. The allelic genes are made homozygous so that any undesirable re-
cessive gene would have full expression in the different races and thus lower
the yield. The cytoplasms which combine are diverse and as such should
give high yield to the individuals if hybrid vigor is an expression of differences
in combining cytoplasm. This contrast is shown as follows:

Genes tend Cytoplasms tend

toward toward

Inbreeding

(^ Gametes
9 Gametes

d^ Gametes
9 Gametes

identity identity

Homozygous by outcrossing
identity diversity

The effects of genes as contrasted with the effects of the cytoplasm may be
measured by comparing inbred performance with that of a race made homo-
zygous for the same genes. Table 29.2 shows this comparison.

For Princeton 1, the inbred progeny resulting from brother-sister mating
for 28 generations had an egg production of 73.6 eggs per day over the test
period. The homozygous Princeton 1 race, coming from the outcross breeding

480

JOHN W. GOWEN

system, had an average egg production of 62.7 eggs, or, the homozygous pro-
duction was 10.9 eggs less than the inbred. The differences of the different
inbred-homozygous comparisons range from —22.0 to +10.6 eggs. In nine-
teen comparisons the inbred races produce more than the homozygous. In
five instances the homozygous races yield more than the inbred. Of the nine-
teen trials in which the homozygous races had less production than the in-

TABLE 29.2

Vx\RIANCE ANALYSIS FOR PROGENY OF IN-
BRED (BROTHER BY SISTER) AND OUT-
CROSS (LEADING TO HOMOZYGOSIS) MAT-
ING SYSTEMS

Race

Mean Difference

Value F

Princeton

1

-10.9

5.9*

Princeton

1

- 2.8

.3

Princeton

1

- 9.9
+ 6.6

3.7

Princeton

8

2.1

Princeton

8

- 6.3

- 1.3
+ 6.0

2.9

Princeton

8

.1

Princeton

10

1.9

Princeton 10

+ 0.1

.0

Princeton 10

-15.9
-f- 0.8

9.6**

Princeton 10

.1

Florida

2

- 7.3

.8

Florida

2

- 8.7

3.4

Florida

2

-20.5

20.1**

Florida

2

-16.6

8.3**

Florida

5

- 5.5

.8

Inbred

3

+ 10.6

3.1

Inbred

4

-19.2

7.9**

Inbred

4

- 9.6

2.6

Inbred

4

- 9.3

1.6

Inbred

4

- 6.6

.7

Inbred

9

-13.7

4.4

Inbred

9

-21.1
-16.2
-22.0

7.6*

Inbred

9 . .

3.9

Inbred

9

7.9*

LE^e ....

Avera

- 8.4

breds, there are four differences which are highly significant and three differ-
ences that are in the significant range when account is taken of chance varia-
tions. In no instance was the homozygous egg yield significantly larger than
that of the inbreds. The data were consistent in showing the homozygous
poorer in egg production than the inbred, even though the particular homo-
zygous is only a sample of the germ plasm of the highly inbred strain.

The average difference between the homozygous and inbred progeny is
— 8.38. Considering each observation as equivalent, the probability that the
homozygous are on the whole poorer producers than the inbreds is well be-
yond the 1 per cent range by the test.

HYBRID VIGOR IN DROSOPHILA

481

There are three major hypotheses to account for the vigor of race crosses.
One hypothesis assumes an as yet unexplained physiological stimulation re-
sulting from the union of gametes of unlike origin. The second hypothesis at-
tributes hybrid vigor to the union of gametes carrying difTerent favorable
dominant genes for vigor, which cover up defects which may exist in each of
the original parent races. The third hypothesis also depends on genie action.
It assumes that the vigor of the hybrid comes from the association of unlike
alleles brought in from the two parental races, these unlike alleles are postu-
lated as contributing different, as well as like, chemical or physical stimula-
tions favorable to the vigor of the hybrid. The results of these experiments
presented in Table 29.2 are in favor of a genie basis rather than a physio-
logic stimulation as the cause of hybrid vigor, since throughout this work,
diverse cytoplasm has shown less yield than like cytoplasm when put on a
background of homozygous or inbred inheritance.

INBREEDING EFFECTS ON HETEROSIS AS RELATED
TO DEFECTIVE GENES

The creation of homozygous types tests the parent race for heterozygosity
of particularly undesirable genes, lethals, and semi-lethals. Table 29.3 shows
the results obtained in mating the homozygous races.

TABLE 29.3

GENOTYPES OF INBRED RACES FOR VIABLE, LETI|^,
AND RECESSIVE VISIBLE ALLELES

Race

Princeton
Inbred. . .
Florida. . .
Florida. . .

Line

1

9

5
2

No.
Lethal

9

0

98

0

Lethal

7

17
20
12

Recessive
Visible

0
0

2
2

Total
Isolation

16

17

118

14

The lethals observed were all in chromosomes 2 and 3. They range in fre-
quency from 17 per cent for one race to 100 per cent for another. The visible
recessives picked up were also semi-lethal. The mathematical model em-
ployed in inbreeding calculations postulates random recombination and fer-
tilization. Conclusions are misleading when these postulates are not met. The
above evidence for mechanisms to maintain heterozygosity in races even
though the matings are of relatives as close as continued full brother X sister
seems unmistakable. The defective genes are in the races. Residual defective
genes can contribute both toward and away from greater heterotic effects in
particular crosses. As these defective genes arise ultimately by mutation and
as the number of the genes is large, the ultimate possible genetic changes are
appreciable and may be an important force toward heterosis.

482

JOHN W. GOWEN

EFFECT OF THE GENOME COMPONENTS ON HYBRID VIGOR

The combining capacity of a genome may be analyzed into its components
— the individual chromosomes. To make this analysis, Dr. Straus in our genet-
ics laboratory carried through duplicated experiments based on the cross of
Inbred 92 and Ames I. This cross, as noted previously, showed high hybrid
vigor. The following data were taken from his thesis (1942).

The results showed no cytoplasmic effects. Effects of reciprocal crosses
also were found negligible.

The first step in these investigations required that 8 possible homozygous
lines be created for the first, second, and third chromosomes. About 98.5 per
cent of the genes would be homozygous in each of the eight types. Crosses
of the eight different homozygous lines will give all the other types ranging
from those heterozygous in one chromosome pair to those heterozygous for
each chromosome. The productivities of these 27 different types together
with their chromosomal constitution are as follows:

Heterozygous for 3 chromosome pairs
Heterozygous for 2 chromosome pairs
6 chromosome combinations

Heterozygous for 1 chromosome pair
12 chromosome types

Homozygous
8 types

Average Daily

Type

Egg Yield

1

76.9

2

64.7

3

64.4

4

51.5

5

65.5

6

66.5

7

62.9

8

55.2

9

52.7

10

55.8

11

60.6

12

46.5

13

53.6

14

35.3

15

56.1

16

56.2

17

52.4

18

51.7

19

41.7

20

45.0

21

51.3

22

40.4

23

36.3

24

37.0

25

35.9

26

27.8

27

31.9

The analysis of the variance of the 1440 daily egg productions in this com-
pletely balanced factorial experiment shows that the difference between
chromosome effects makes the most important contribution to variation.

From type 20 to 27, all types are homozygous or are of zero heterozygosity.
Types 8 to 19 have one chromosome heterozygous. Since each chromosome

HYBRID VIGOR IN DROSOPHILA

483

enters in equal frequency, this means that one-third of the genes are on the
average heterozygous. Types 2 to 7 have 2 chromosomes heterozygous, or the
average of these types is two-thirds heterozygous. Type 1 is completely
heterozygous or 100 per cent. Plotting the average egg production for the
four groups shows the effects of different degrees of heterozygosity on the
hybrid vigor.

The property of additivity of the heterotic gene effects would seem to be
the logical explanation for this linear relation and also for the absence of in-
teractions between the genes of the different chromosomes. It must be real-
ized, however, that the chromosomes themselves represent interacting gene
effects which give the block reactions. The trend so far considered is an aver-
age trend, each point, except that for the completely heterozygous, being
based on several types. Interaction — combination effects which are larger or
smaller than the sum of the chromosomal effects separately — may exist. Such
effects, it is true, must be in opposite directions and equal. The factorial de-
sign of the experiment facilitates evaluation of these interactions. The data
following gives the three levels of effect each of the chromosomes can assume,
together with the two and three chromosome interactions.

VARIANCE ANALYSIS OF CHROMOSOMAL EFFECTS

Source of Variation

I chromosome

II chromosome

III chromosome

I and II chromosomes

I and III chromosomes . . . .

II and III chromosomes. . .
I, II, and III chromosomes

d/f

2
2
2
4
4
4
8

Mean
Square

574

1916

1010

81

62

116

103

Apportion-
ment of
Variation
Per Cent

11

44

22

0

0

1

22

The effects of the direct order actions of the first, second, and third
chromosomes are highly significant. None of the interactions show large vari-
ations. Nor are any of these interactions in excess of what would be expected
from random differences. Apportioning the variance to its various chromo-
somes, 11 per cent is attributable to the first chromosome, 44 per cent to the
second, and 22 per cent to the third.

Analysis of this material shows that the hybrid vigor of the egg yields re-
ceives a significant contribution from the heterozygosity of each chromosome
pair, and that none of the chromosome interactions are significant. The ho-
mozygous chromosomes of the two parental inbreds do not differ in either
their direct or interaction effects from zero.

The linearity of the effect on egg yield and the absence of interactions show
that the chromosomes with their contained genes behave as integrated units

484

JOHN W. GOWEN

— much like major genes — with given degrees of dominance. Within each
group the genes may have any known type of gene action so long as the quali-
ty of additivity of their effect between chromosomes is maintained.

EFFECT OF CHROMOSOME LENGTH ON HYBRID VIGOR

The analysis of variance shows that the effects of the three chromosome
pairs differed widely. These differences could be due to differential numbers
of gene loci within the separate chromosomes, to varying magnitude of gene
effects, or to both. The data do not allow us to positively distinguish between
these hypotheses. Proportionality between the effects of the three chromo-
somes and their sizes would favor the first interpretation.

There are several different measures of chromosome size. These measure-
ments of the different chromosomes may be compared with their heterotic
effects in a least square test.

The proportionality between heterotic effects and chromosome lengths
was as follows:

Metaphase length

Salivary length

Salivary bands

Per cent visible loci

Cross over length

Observed heterotic ^7 i„„

pffprt J 7 day..

^^^^^^ \A11 data

Chromosomes

I

II

III

1.56

2.21

2.80

220

460

485

1024

2134

2077

.69

1.00

.77

.62

1.00

.98

248

386

325

192

305

174

Closeness of Agreement

7-Day Data

X

X

XX

XX

XX

All Data

X
X
X

XX
X

X significantly closer 1 : 20 or
X X highly significant 1 : 100

Excellent agreements are observed between per cent of visible loci or the
crossing-over units with the heterotic effects of the chromosomes. Less agree-
ment is noted between the number of bands in the salivary chromosomes and
the heterotic effect. The metaphase lengths of chromosomes or the physical
lengths of the salivary gland chromosomes are less closely related to heterotic
effects. All comparisons of chromosome sizes with heterotic effect give excel-
lent to fair correlations. In general, the heterotic effect is distributed accord-
ing to random distribution of several genes to the various chromosomes. This
favors the view that the heterotic effect is due to many gene pairs in each
chromosome, rather than to one having a specific additive phenotypic effect.
These genes would be randomly distributed to the different loci within the
chromosome.

GENOME CONTRIBUTIONS TO HYBRID VIGOR

Average combining ability of one inbred line when mated to several lines
is called general. The genomes of an inbred line can be regarded as uniform

HYBRID VIGOR IN DROSOPHILA

485

and good or bad according to the genes which they contain. These genes
could be additive in effect making the genomes of uniform effect with other
inbred lines. Specific combining ability represents unlike combining ability of
the genomes from one race with those of a succession of other races. This
variation in hybrids could be due to different allelic distributions as comple-
mentary or epistatic reactions of the different gene combinations with which
the given genome was combined. The relative effects of general vs. specific
combining ability for a particular group of crosses may be measured in data
containing all possible combinations between a series of different inbreds.
Table 29.4 presents the egg productions for the possible hybrids of five inbred
races.

TABLE 29.4

EGG PRODUCTIONS OF 5 INBRED RACES AND THEIR CROSSES

AVERAGE YIELDS FOR 5, 6, 7, 8, AND 9TH DAYS

OVER 4 EXPERIMENTS

Fem.^le

Male Parent Race

P.\RENT Race

A

B

C

D

E

Total

A .

2509.0

2681.0
2712.8

"3215^2'
2498.2

3479.4

3427.4
3298.8

2503.8
1822.2
3116.0
3447.6

11173 2

B

2908.6
1804.8
2321.4
2109.8

10871 0

C

2827.8
3485.6
1908.2

11047 4

D

13467 8

E

3301.0

9817 2

Total

10144.6

10728.6

11107.2

13506.6

10889.6

56376.6

Inbred Race
Yields

2595.2

2586.4

1996.6

2173.4

1859.4

Table 29.4 shows 14 of the race hybrids have higher average yields than
their inbred parent races. The average hybrid produced 2818.8 eggs, the aver-
age inbred 2242.2, or the increase over the average inbred was 25.7 per cent.
These data serve to re-emphasize the fact pointed out earlier, that charac-
ters built up of components of lesser characters generally show more heterosis
than observed for each of the components taken separately.

The individual race crosses differed in their ability to unite into favorable
hybrids. Race D is evidently high in its general combining ability. The other
four races show about equal combining ability. Race D has this high general
combining ability even though its own productivity is rather low — 2173.4 —
eggs as against 2595.2 for another of the races.

For individual flies the range in egg production was from 0 to 146 eggs.
The zero egg producing flies are an important class which give an insight into
female sterility. We have dissected over 300 such flies. These observations
have led to the opinion that this class is the result of a variety of causes and

486

JOHN W. GOWEN

definitely differs genetically and otherwise from that of the flies which pro-
duce even one egg in their lifetimes. For this reason and the fact that hetero-
sis is a phenomenon of quantitative inheritance, we have excluded such flies
from consideration in these studies.

The general analysis of the variations within these hybrid egg yields is
presented in Table 29.5.

TABLE 29.5
DISTRIBUTION OF VARIANCE IN EGG PRODUCTION

Source of
Variation

Desig-
nation

d/f

Mean
Square

Components of
Variation

Per
Cent
Con-
trib-
uted

Total

5624
3

24
4
19

1
4

72
12
96

288
5125

Experiments. . .
Races

E
R

116461

37405
12630
36811

147779
6665

5800

18202

767

830
510

W-f-11.2ERA-|-280EA+56ER

+ 1400E
W-fll.4ERA-(-57ER+45RA

-f-225R

7.3
15 4

Inbreds

Hybrids

Inbreds vs.
Hybrids. . . .

Ages

Exp X Races ....

ExpXAges

Races X Ages. . .
Exp X Races X

Ages

Residual

A

ER
EA
RA

ERA
W

W+11.4ERA+285EA+45RA

+ 1125A
VV-I-11.2ERA+56ER
W+11.2ERA+280EA
\V-(-11.4ERA+45RA

\V-I-11.2ERA
W

1.2

9.8
6.8
0.2

3.1
56.2

100.0

Table 29.5 presents data on the factors which may be of importance in the
interpretation of heterotic effects. The percentage contribution of each factor
is shown in the right hand column. The largest contribution, residual, is made
by the variation within flies of a given age. It is half of the total observed.
This variation shows what minor differences in seemingly constant condi-
tions can be responsible for differences in egg yields.

Differences in races represent the next most significant contribution to
yield variations — 15 per cent. Major contributor to this effect is the differ-
ence between the productivities of the inbred parent races and their hybrids.
These differences may be looked upon as the effects of additive genetic fac-
tors for yield, and the effects of specific gene combinations leading to the ex-
pression of dominance, overdominance, or epistacy in the phenotypes.

A point of currently even more importance brought out by these data is
the dependence of yield on the close interrelation of environment and geno-
type. The interaction of experiment X race accounts for 9.8, and experiment,
race, and age, 3.1 per cent of the variation. The total is 12.9 per cent. Even
with great care to closely control conditions both within and between experi-

HYBRID VIGOR IN DROSOPHILA 487

ments, the environment is sufficiently important to the yield of the particular
race to account for nearly as much of the total yield variation as race alone.
With widely ranging environments, given genotypes may show much more
variation in phenotypic expression. The interaction terms show that genes in
quantitative inheritance are not stable in their effects. In one condition the
phenotypic reaction, in some degree, could be such as to suggest recessive
action; in another dominant, in another additive or epistatic.

These interpretations may be brought out by another analytical approach.
In analyzing data of this kind it has been customary to neglect the genotypic
environmental interactions. This neglect finds expression in the models
adopted to explain the yield. For the data above it is sometimes assumed that
yield, yuk, may be accounted for by a basic value common to all crosses, the
mean; deviations due to additive general combining ability of the different
races ga, gb, etc.; deviations due to specific combining ability, Sab, Sac, etc.,
such as dominance variations, and epistatic effects common only to members
of that particular cross; reciprocal effects, Tab vs tba, etc., of any differences be-
tween members of reciprocal crosses; and a term representing residual varia-
tions, Bahk, Back ctc, duc to unknown causes. These variables are set up in the

linear equation:

yuk = ni + gi + gj-\- Sij-\- rij-\- Cijk

Analysis of the data for the contributions of these variables to the yield
variajice gives these results:

EXPERIMENT 35
General combining ability. . . . 11.3%
Specific combining ability. ... 9.7

Reciprocal effects 2.3

Residual effects 76. 6

Two sets of experiments are available. One is for five and the other for six
inbred line hybrids. The test as presented above shows that 11.3 per cent of
the variance is due to differences in general and 9.7 per cent to differences in
specific combining ability. Differences in reciprocal crosses account for 2.3
per cent. Experiment 36 shows similar contributions attributable to general
and to specific combining ability, but the effect of reciprocal crosses is insig-
nificant. The two experiments are concordant in showing that general and spe-
cific combining ability account for most of the variation attributable to known
causes. In both cases general combining ability is somewhat more important
to productivity than specific combining ability.

These results from Drosophila are entirely without any previous selection
for combining ability. They are comparable to the observations which were
obtained in corn when combining ability was tested for the early crosses of
inbred lines. It is significant that Sprague's analyses of such crosses show gen-
eral combining ability twice as important as specific. This difference is like
that of Drosophila but gives even more emphasis to general combining
ability.

488 JOHN W. GOWEN

In later corn hybrids, the products of more stringently selected inbreds,
the emphasis was reversed. The specific combining ability was zero to five
times as important as the general. Improved utilization of hybrid vigor has
seemingly selected and fixed general combining ability in the approved in-
breds. Further progress is dependent on specific combining ability. One com-
parison weakens this evidence. On exactly the same ten inbred combinations
one set grown at Ames, the other at Davenport, the specific combining ability
was five times that of the general at Ames, while at Davenport the general
and specific were identical. A place X genotype interaction in the general-
specific combining ability similar to that observed above for Drosophila egg
yield is also important even in these highly selected lines.

SIGNIFICANCE OF ENVIRONMENT-GENOTYPE
INTERACTION IN HETEROSIS

An experiment by Dr. Loh evaluating the significance of early or late test-
ing furnishes data on the part played by environment in the stability of the
hybrid phenotypes (1949). Fifty full brother X sister lines were formed from
each of three wild stocks having different geographical or chronological ori-
gin. Each line was then doubled and mated full brother X sister for as long as
possible, or until 37 generations were reached. The average productivity of
the initial lines crossed to the same synthetic strain at the start of the in-
breeding was Ames 1947, 179 ± 2; Ames 1943, 176 ± 3; and Amherst 1947,
166 ± 3 eggs for the 5, 6, and 7th days after the hybrid females hatched. All
surviving inbred lines were crossed to the same synthetic stock, and the hy-
brid females tested at the 8, 9, 16, 23, and 30th generations. The results were
consistent for the three stocks. The egg productions of the hybrids declined
2.4 eggs per generation on the average. This result was surprising, but may
possibly be accounted for by the fact that the inbred lines surviving in the
three stocks were becoming more and more like the synthetic tester due to
the fact that they were cultured on the same media and in the same way. The
favorable gene differences between the crossed lines became less each genera-
tion and resulted in a progressive lowering of hybrid yield. As the generation
times were confounded with time of year, it was also possible that the egg
yields showed some effects of the progressive changes in season.

The surviving inbred lines were tested for egg yield on the 21, 26, and 31st
generations. They showed an average decrease in egg yield of 4.3 eggs per
generation. This decrease was greater than that observed for the inbreds X
synthetic cited above. This was not entirely unexpected, although it did in-
dicate continued and persistent heterozygosity in the inbred lines to a much
greater extent than was sometimes realized. The inbred lines produced 20 to
40 per cent less than the hybrids. The differences became greater as the in-
breeding advanced.

The inbreds of the 15, 24, and 34th generations were crossed in all possible

HYBRID VIGOR IN DROSOPHILA 489

ways. The line crosses were 17, 30, and 62 per cent better than their inbred
parents. They were also 2 to 4 per cent better than the inbred X synthetic
crosses. Figure 29.3 shows these trends for the three types of progenies.

Coeflicients were calculated for the like cross performances at different
generations. The synthetic X inbred lines had correlations for the 1, 8, 9, 16,
23, and 30th generations. Like numbers of comparisons were available in the
reciprocal crosses, inbred lines X synthetic. The correlations were similar for
generations and their reciprocals. The average for the thirty comparisons was
— 0.01. In terms of the data, the synthetic X inbred line cross of one genera-
tion gave no information on the relative performance of the same cross in a
succeeding test. The hybrids showed random variation within themselves,
but at the same time averaged out to be distinctly better than the inbreds.

The inbred lines of the 21, 26, and 34th generations were crossed in all
possible ways. Again the correlations between the productivities of the like
crosses in different generations showed variation. The average correlation
was 0.25. The performance of the cross uniting two of these inbred lines did
have some predictive value for the performance of like crosses made subse-
quently. Again these hybrids showed most of the variation within the crosses
to be random, but that the yield level of the hybrids was significantly better
than the inbred parent lines. The over-all value for larger yield came as a
consequence of the cross rather than as an effect of specific cross differences.

This fact is brought out in another way. The inbred crosses were analyzed
for general and specific combining ability, as described earlier, for the three
different generations of inbreeding, 15, 24, and 34. The average results were:

General combining ability 12%

Specific combining ability 5

Residual variation 83

These results are comparable with those presented earlier. General and
specific combining ability can be estimated for each line in the particular
crosses and experiments. The values can be compared as between the differ-
ent generations, to determine how consistent in combining ability is the be-
havior of each line. The correlations for these comparisons were as follows:

15 and 24 15 and 34 24 and 34

General combining abilities —0.02 —0.27 —0.17

Specific combining abilities 0.13 0.02 0.26

These correlations are so small as to indicate that combining abilities are
not consistent from one generation (in this case also season) to the next. The
hybrids are uniformly better than the inbreds in yield, but again the geno-
typic system does not appear to have a fixed reaction. The explanation of this
fact appears in Table 29.4 where a high experiment genotype interaction was
observed. It means that each genotype may react differently to different en-
vironment. As these environments change from place to place, season to sea-

490

JOHN W. GOWEN

son, and even between simultaneous carefully controlled experiments, it is
not surprising that the general and specific inheritance effects show varia-
tions. A particular fitting of strains to place and season, etc., appears essen-
tial for highest yields. In view of this conclusion, it is important to remember
that this effect is within hybrids, and that hybrids, in general, are distinctly
better than the inbreds (see Fig. 29.3).

Through the kindness of Dr. G. F. Sprague, making available certain of
his extensive data on Fi crosses of some 62 inbred lines of corn, we have been
able to extend this analysis and compare the stability of general and specific

O

z>
o
o

CL

o

<

UJ

140

20

100

80

INBRED X
SYNTHETIC y

INBRED LINE
CROSSES

■•••s.x

INBRED LINES

15

20

25

30

35

GENERATIONS OF BROTHERxSlSTE R MATING

Fig. 29.3 — Changes in productivity with advancing generations of brother X sister mat-
ing of inbred Unes, inbred line crosses, and inbred X s_vnthetic.

combining ability in the two species. The trials were conducted yearly from
1940 to 1948. The Fi hybrids were planted in ten different areas chosen as
representative of the different climatic conditions of Iowa. Any one trial may
contain all possible single crosses of 4 to 14 inbred lines. The trials contain
many individuals, and are replicated several times so the record for the Fi is
an average of numerous Fi's instead of an individual as in the data on Dro-
sophila. As would be expected, a particular cross was occasionally lost from a
test. When this happened, the missing plot value was calculated from the
mean and the general combining abilities of the lines entering the test. The
specific value was considered zero. The data for the general and specific com-
bining abilities of the lines in the remaining plots were used for further study.
Our study considered the first order values, as these are the only values which
have operational significance in breeding for heterosis.

HYBRID VIGOR IN DROSOPHILA 491

General and specific combining abilities are strictly applicable to the par-
ticular experiment from which they are calculated. The values for the dif-
ferent lines vary with the group of lines from which they are calculated. This
is a serious defect, for the results have no significance unless they may be used
for the prediction of the future performance of the particular line or line com-
bination. As the interest is in the operational use of these parameters in guid-
ing breeding work, the theoretical objections to comparing successive values
for general or specific combining abilities are outranked by the practical con-
sideration. This study measures the repeatability of the estimates of general
and specific combining ability for particular inbred lines when the crosses are
grown in different locations, different years, and in different combinations.

Sixty-two inbred lines were the parents of the Fi crosses. The data include
451 determinations of general and 2033 estimates of specific combining abil-
ity. As pointed out above, these determinations are not of equal weight be-
cause of differences in numbers and lines in the different Fi hybrid tests.
However, for the purposes of this comparison they are regarded as of equal
weight, since it is on this basis that the data will be used for guiding future
breeding operations.

The intraclass correlation between the repeated tests of the general com-
bining abilities was 0.29; that for the specific combining abilities of the re-
peated crosses of the pairs of lines was 0.27. These correlations are definitely
higher than those observed for the Drosophila data. They are high enough to
be of reasonable importance in practice. The data for the general combining
abilities become of somewhat greater value when the determinations are re-
stricted to particular regions of the state, the over-all correlation becoming
0.31. When arranged within years but allowing free range over the 10 differ-
ent geographical regions of the state, the over-all correlation becomes 0.53.
The specific combining abilities do not show an equal improvement in predic-
tive values when subdivided by these categories. Specific combining abilities
drop when the data are subdivided by geographical regions of the state, the
over-all correlation becoming 0.18. When the subdivision is made by years,
the over-all correlation becomes 0.34.

These results reemphasize the effects of the environmental-genotypic in-
teractions on performance as discussed earlier. The corn hybrids are fitted to
the geographical regions of the state by selection of the place of planting for
season of maturity. Little or no selection is possible for fitting the plantings
to the vagaries of the different years. The effects are noted in the intraclass
correlations. Double selection for genotypic environmental correlation when
the data are subdivided by years leads to definitely increased correlations for
the general combining abilities of the particular lines and to slightly increased
correlations for the specific combining abilities of these same lines. Where the
years X genotypic effects are allowed to express themselves, the correlations
are no greater than those of the whole or are reduced.

492 JOHN W. GOWEN

SUMMARY

Consideration of egg production and other component characters in
Drosophila melanogaster shows that hybrids are uniformly better producers
than inbreds even though the inbreds be the parents of the hybrids. The
hybrids themselves are not exceptional in production when contrasted to the
best random bred individuals. Rather, hybrid vigor contributes consistently
high performance to all individuals rather than very superior performance to
a few.

Lifetime egg productions show greater heterosis than any of the com-
ponent factors which ultimately determine it. Length of egg laying period has
113 per cent heterosis, maximum egg production 154 per cent, and resistance
to decline in vigor, as measured by egg production with advancing age, 120
per cent, while the over-all character lifetime egg yield has 203 per cent
heterosis. Heterosis appears to be a consequence of the combined action of
two or more groups of distinct and more elementary characters which when
jointly favorable lead to generally high yields.

Tests show that hybrid vigor is attributable to nuclear contributions of the
two parents rather than to possible cytoplasmic differences in the uniting
gametes. Inbred races frequently contain or soon attain mechanisms to slow
down or prevent reaching complete homozygosis through continued close in-
breeding. Lethal genes, deficiencies, or defective genes residual in all stocks
or acquired through mutation, balance to prevent free interchange of genes
within chromosome groups, and thus retard or stop the formation of the
homozygous types. In the light of these results, mutations as a heterosis
mechanism assume much greater importance than ordinarily supposed.

When the egg yields were analyzed by the degree of heterozygosity it was
found that flies homozygous for all loci in chromosomes I, II, and III or 0
heterozygous, produced 38.2 eggs on the average. Those heterozygous for
one-third of the unlike parental genes in the cross produced 51.5 eggs on the
average. Those heterozygous for two-thirds of the unlike parental genes laid
62.6 eggs, and those heterozygous for all unlike parental genes, three-thirds
heterozygous had a mean yield of 76.9 eggs. The differences are additive,
about 12.9 eggs being added with each increase of one-third of the genes
heterozygous. The additivity of the mass gene effects would suggest addi-
tivity of the individual gene actions on egg yield. This is an important point
but does not necessarily follow, because the dominance or recessiveness or
interallelic interactions could be balanced by the mass of gene pairs compris-
ing one-third of the heterozygous loci.

Study of the contributions to the heterosis made by the different chromo-
somes shows that they are all first order contributions, there being no inter-
action between chromosome pairs. Comparison of the heterosis attributable
to the different chromosomes with different measures of the numbers of gene
loci which they contain, shows that as the method of chromosome measure-

HYBRID VIGOR IN DROSOPHILA 493

ment approaches what a})pears to be the likely loci number, the better this
method agrees with the heterosis which is observed when the chromosome is
made heterozygous. The evidence favors several to many gene pairs per chro-
mosome as necessary for the heterotic effects.

Heterotic effects of parental genomes as shown by a series of Fi hybrids
were analyzed. For the individual the most significant contribution to varia-
tion was that due to a large number of unanalyzed causes. This component
contributed over half of the total variation. Differences due to races contrib-
uted 15 per cent, w'hile those due to race-experiment and age interactions, 13
per cent. The interaction term shows that genes in quantitative inheritance
are not stable in their effects. In one condition the genes could react as reces-
sives; in another as dominants; in a third, show epistacy.

The dependence of yield on the interrelation of environment and genotype
is of even greater importance. The model customarily chosen to represent
genetic and environmental effects ordinarily considers the interactions of
these terms zero when in truth they may be quite large. The data on both
Drosophila and corn general and specific combining abilities of inbred lines
show these interactions to be of major importance. Further progress in the
utilization of heterosis appears to lie in the adjustment of the hybrid genotype
to the environment.

R. E. COMSTOCK

and

H. F. ROBINSON

Norih Carolina Sfaie College

Chapter 30

Estimation of Average
Dominance of Genes

This discussion will center around three experimental procedures used at the
North Carolina Experiment Station for investigating the degree of dominance
involved in the action of genes that affect quantitative characters of economic
plants. The objective is twofold: (1) to outline and, in so far as possible, eval-
uate these methods; and (2) by example, to point up the role of statistics in
genetical research.

Basic criteria for the usefulness of a projected experiment are: (1) Will
data obtained provide a logical basis for inference relative to the research
objective? (2) Will the random variability in the experiment be of an order
that will permit satisfactory certainty of conclusions? The latter has obvious
statistical overtones, but statistics is not always deeply involved in the
former. In genetic work, random variability in the experimental material is
generated in part by the genetic mechanism, and can therefore be used as
a basis for inference in genetic problems. Hence statistics plays an inescap-
able role in both aspects of the evaluation of many genetic experiments.

Examination of any proposed basis for inference must obviously center on
the premises involved and the validity of deductions predicated on those
premises. We will turn first, therefore, to description of the experiments and
the logical basis for the estimates they are designed to provide.

THE EXPERIMENTAL DESIGNS

The designs of each of these experiments have two aspects: (1) the genetic
background and (2) the field arrangement of the material on which data are
collected.

494

ESTIMATION OF AVERAGE DOMINANCE OF GENES 495

Experiment I

The experimental material is produced from matings among plants of the
F2 generation of a cross of two inbred lines. Each plant used as a pollen parent
is mated with n seed parents, no seed parent being involved in more than one
mating. Thus, if sm pollen parents are used there will be smn seed parents
used in smn matings. The progenies of these smn matings comprise the ex-
perimental material. All parent plants are chosen from the F2 population at
random.

Pollen parent and seed parent plants will for brevity be referred to in what
follows as males and females, respectively. A group of n progenies having the
same male parent will be referred to as a male group.

The field arrangement of the material is based on division of the sm male
groups into 5 sets each of which contains mn progenies in m male groups.
Each set of progenies comprises the material for a distinct unit of the experi-
ment and is planted in a randomized block arrangement having mn entries
and r replications. Thus the total field arrangement is composed of s inde-
pendent units, each unit being devoted to a different set of progenies. Data
on characters of interest are collected on k plants per plot.

Experiment I!

This is a modification of Experiment I that can be used when dealing with
multi-flowered plants. The foundation stock is again the F2 generation of a
cross of inbred lines. In this case, however, a set of mn progenies is produced
by making all of the mn possible matings of w. males and n females chosen at
random from the F2 population. With annual plants this can be done (and
the progenies kept distinct) only if more than one pistillate flower per plant
is available. It could not be done, for example, with single-eared corn.

The field arrangement is as described for Experiment I, the sets arising
from the mating plan being maintained intact in the units of the field struc-
ture of the experiment.

Experiment III

The experimental material is produced from backcross matings of F2
plants to the two inbred lines from which the F2 was derived; the F2 plants
are used as pollen parents. A set of progenies is made up of the 2n progenies
obtained from backcrossing n random F2 plants to each of the parent in-
breds. The number of inbred plants used in production of each backcross
progeny is important only with respect to insuring sufficient seed.

The total experimental material consists of s sets of n pairs of progenies.
The members of each pair have the same F2 (male) parent but different in-
bred parents. The two inbred parents are, of course, the same for all pairs of
progenies.

The field arrangement is analogous to that for Experiments I and II. The

496

R. E. COMSTOCK AND H. F. ROBINSON

unit in this case is a randomized block arrangement {2n entries and r replica-
tions) of r plots of each of the progenies of a set.

ANALYSIS OF DATA

The appropriate variance analyses for the data of the three experiments
are outlined in Tables 30.1 to 30.3. The expected value (the value that

TABLE 30.1
VARIANCE ANALYSIS (EXPERIMENT I)

Source of Variance

d.f.

m.s.

Expectation of m.s.*

Sets

5-1

s{r-\)
s{m — \)
sm{n — \)
s{mn — \){r-

-1)

Replications in sets

Males in sets ... ....

Mu
Ml.

My,

a^+ra^f+rnal

Females in males in sets. . . .
Remainder among plots. . . .

a'+raj
a"

* a^ is the error variance among plots of the same progeny (due in part to
soil variation among plots in the same block and in part to variation among
plants of the same progeny).

aj is progeny variance arising from genetic differences among female
parents.

al, is progeny variance arising from genetic differences among male parents.

TABLE 30.2
VARIANCE ANALYSIS (EXPERIMENT II)

Source of Variance

d.f.

m.s.

Expectation of m.s.*

Sets

5-1

i(r-l)

5(m-l)

5(w-l)

5(w-1)(k-1)

simn-l)(r-l)

Replications in sets

Males in sets

A/21
M22
A/ 2.3

M24

cr- -\-rann + rn af,,

Females in sets

Males X females in sets. . . .
Remainder among plots. . .

a'+ra7,n-\-rm(j}
<j''--\-raj,n

* cr^m is progeny variance arising from interaction of genotypes of male
and female parents. Other symbols are defined in Table 30.1.

would be approached as a limit if the amount of data were made infinitely
large) is listed for each mean square to be used in interpretations.

In order to specify the significance of components of the total variance of
which estimates can be used for inferences about dominance, some additional
symbolism must first be established. Consider the three genotypes possible at
a locus where there is segregation between two alleles. Let the difference in
effect of the two homozygous genotypes on a measured character be 2u and
the deviation of the effect of the heterozygous genotype from the mean effect
of the homozygous genotypes be au. Note that u and au have the same sig-
nificance as d and h, respectively, in the symbolism used by Fisher et al.
(1932). Also, they have the same significance as d and k in the symbolism

ESTIMATION OF AVERAGE DOMINANCE OF GENES

497

employed earlier in the Heterosis Conference. The symbols u and an are
used here for consistency with usage in articles by Comstock and Robinson
(1948) and Comstock et al. (1949). Let the number of segregating gene pairs
that affect a particular character be symbolized as N, and a numerical sub-
script to w or a specify the locus to which the symbolized quantity is relevant.
Thus 2^3 is the difference in effect of the two homozygous genotypes of the
third locus and a^Uh is the dominance deviation for the fifth locus.

Now granting validity of several assumptions (to be listed and discussed
later) a;,, aj, af„f, and a'lt have genetic meaning as set out in Table 30.4.

TABLE 30.3
VARIANCE ANALYSIS (EXPERIMENT III)

Source of Variance

d.f.

m.s.

Expectation
of m.s.*

Sets

s-1
sir- 1)

kn-1)
sin -I)
s{2n-l)ir-l)

Replications in sets

Tnhrpfi linp in sets

Fo parent in sets

Fo parent X line in sets

Remainder among plots

il/32

M33

<T''+2r<xl,
ff'^+rali
a2

* at is progeny variance arising from genetic differences among F2 (male)

parents
a?

i,i is progeny variance arising from interaction of genotypes of F2 and in-
bred parents.

THE ESTIMATE OF AVERAGE DOMINANCE
The magnitude of a measures the degree of dominance in the action of any
one pair of genes, being related to qualitative classification of dominance as
follows:

Class of

Numerical

Dominance

Value of II

No dominance

a = 0

Partial dominance

0<a<1.0

Complete dominance

a=1.0

Overdominance

a>1.0

However, a problem arises concerning the best way to represent the average
dominance for all loci with a single number. An obvious possibility is the
unweighted mean of a's for all gene pairs. On the other hand, it can be argued
that a mean in which individual a's are weighted relative to the importance
of loci would be more useful. This in turn raises the question of how the rela-
tive importance of loci should be measured. However, the matter will not be
pursued further, since the experiments under consideration offer no choice of
measure to be estimated.

The estimate that can be made is of

a- =

2^2

or

498

R. E. COMSTOCK AND H. F. ROBINSON

a^ is a weighted average of all a-s, weighting being relative to the square of u
(one of the possible measures of the importance of loci), a- and a can exceed
unity only if one or more individual a's are larger than one, but values of a^ in
excess of one do not exclude the possibility of partial dominance at numerous

TABLE 30.4

GENETIC NATURE OF COMPONENTS
OF PROGENY VARIANCE*

Compo-
nent

Experiment

/

i2M2

//

i2M2

i2«^

III

52«2

i2aV

* Summation is in all cases over loci, i.e.,
2;«2 = («?-F«i+ . ..ujf) and
Zaht^ = (af «f +4ul+ ... a%Uff)

loci. On the other hand, a- will not be less than one unless dominance is less
than complete at one or more loci, but values less than one do not insure
absence of overdominance at all loci.

Experiment I

In accordance with the mean square expectations of Table 30.1 we can
estimate

al^ as (Mn-Mii) / rn

and

0-2 as iMi2—Mu)/r

and from Table 30.4 we see that in this experiment
and

(r2 = 121*2

in o

1 m 16

Hence,

2[(«+l)Mi2-Mu-wMi3] . ^. ^ ,2a2«2 .,
IS an estimate of .„ ^ = a^

Mn —Mn

Experiment II

Note first from Table 30.4 that in this experiment a^ = o-/. If the experi-
ment is designed with m = n (this will be assumed in what follows since it is
a rational procedure where possible) this means that the expectation of Mn

ESTIMATION OF AVERAGE DOMINANCE OF GENES 499

and Af 22 (see Table 30.2) are equal and hence that the two mean squares may
be pooled.' Let the pooled mean square be symbolized by M20. Then

(M20 - M23) / rn estimates aj = (t;^

and

(M23 - M24) / r estimates aj^

In this experiment (see

Table 30.4)

aj = al = li:u^

and

'^'fr. = ^^^'^'

It follows that

2na

^2z-M2i) .. . ^a^u"^

,, 7T estimates ^ . = a- .

M20 — M23 2w-

Experiment III

Following Tables 30.3 and 30.4 we see that

(M„ — M,J /2 r estimates a^ = iSw^

^ 31 33 m °

(Afg, — M33) / r estimates a^^ = jZa^w^

so that

M32— M33 .. . -2
■77 —— estimates a^

M31 —M33

ASSUMPTIONS

Evaluation of procedures described above should obviously begin with
examination of assumptions underlying derivations of mean square expecta-
tions listed in Tables 30.1 to 30.3 and genetic interpretations placed on vari-
ance components in Table 30.4. Premises involved in the derivation of mean
square expectations were as follows:

1. Random choice of individuals mated for production of ex-
perimental progenies.

2. Random distribution of genotypes relative to variations
in environment.

3. No non-genic maternal effects.

The first of these can be assured easily in the conduct of the experiment.

The second is equally easy to assure in so far as environmental variations

within the experiment are concerned. On the other hand, the environments

encountered in an experiment conducted within the confines of a single year

1. By taking an unweighted mean since degrees of freedom will also be equal when
m = n.

500 R. E. COMSTOCK AND H. F. ROBINSON

and location do not constitute a random sample of the environments that
occur within the wider limits of time and space for which we would like ex-
perimental findings to apply. The consequence of this is that, if interaction of
genotype with environment is a source of variation, each mean square arising
from variation among progenies will contain some variance from such inter-
action. Thus, to have been rigorously correct, the expectations of all mean
squares between progenies should have included terms recognizing contribu-
tions from this source. Separate estimation of the genetic and interaction
components of mean squares between progenies could not be effected with
data collected in a single year and location. If the ratio of these two sorts of
variance is constant for the several mean squares, and there is no obvious
reason why it should vary, the presence of interaction variance does not bias
the estimates of a^ since numerator and denominator are affected propor-
tionately. Nevertheless this constitutes a possible weakness of the methods
but one which, if important, can be corrected by replication of all progenies
over years and locations.

There are many characters and organisms for which it appears safe to
assume maternal affects are absent or of no consequence. This assumption
must be viewed with some suspicion when dealing with seedling characters
of plants or any character for which there is any hint that cytoplasmic in-
heritance may be operating, and it is definitely not tenable for pre-maturity
characters of mammals. Maternal effects do not contribute to the pertinent
mean squares in the variance analysis of Experiment III and only to M22 in
that of Experiment II. Thus these two experiments are useful in the presence
of maternal effects, though if II is used 2«^ must be estimated from M21
instead of jointly from Mi\ and M^i-

Assumptions involved in deriving the genetic interpretations of variance
components are as follows:

1. Regular diploid behavior at meiosis.

2. Population gene frequencies of one-half at all loci where
there is segregation (not necessary for Experiment III).

3. No multiple allelism.

4. No correlation of genotypes at separate loci. This implies
no linkage among genes affecting the character studied or that,
if linkages exist, the distribution of genotypes is at equilibrium
with respect to coupling and repulsion phases.

5. No epistasis, i.e., the effect of variation in genotype at any
single locus is not modiiied by genes at other loci.

In accord with the first of these, usefulness of the procedures described is
limited to studies with diploids or amphidiploids in which multivalent meiotic
associations are entirely absent or are absent in meiotic divisions giving rise
to fertile gametes.

ESTIMATION OF AVERAGE DOMINANCE OF GENES 501

Save for deviations due to natural selection, the second assumption is as-
sured by the fact that the population used is an F2 of a cross of homozygous
lines. Moreover, natural selection strong enough to have more than a trivial
effect on gene frequencies can only occur if the development of a moderately
large proportion of F2 plants is so slow or aberrant as to prevent their effec-
tive use as parents. Thus a good stand of usable plants constitutes insurance
that this assumption is satisfied.

Number three is also assured by the origin of the populations. Multiple
alleles in an Fo of homozygous lines can result only from mutation, and in the
light of present knowledge of mutation rates would be expected very infre-
quently.

On the other hand, complete validity of the fourth assumption is improb-
able. Present day geneticists are in general agreement that quantitative char-
acters, and particularly physiologically complex ones such as yield, are influ-
enced by many genes. If that is so, there may well be linkages among some of
the genes affecting any single character. Furthermore, specific linkage rela-
tionships in an Fi of homozygous lines must be in either the coupling or repul-
sion phase, and equilibrium between the phases cannot occur in the Fo. In
fact the approach to equilibrium in later generations is rather slow unless
linkage is very loose (see Wright, 1933).

The effect of linkage is to cause upward bias in estimates of a. Thus
Comstock and Robinson (1948) in discussion of Experiments I and II and
Robinson et al. (1949) in discussing results obtained using Experiment I in-
dicated that values of a larger than one can result either from true overdom-
inance or from repulsion linkage of genes that are completely or partially
dominant to their alleles. The same conclusion can be inferred from Mather
(1949).

The situation can be summarized in another manner by stating that values
of a in excess of unity do not distinguish true overdominance in the action of
alleles from what Mangelsdorf has termed pseudo-overdominance or over-
dominance at the gamete level. However, in defense of the procedures under
discussion, it must be emphasized that knowing one or the other of these two
phenomena is at work is an advance over being uncertain as to whether either
is operative. On the other hand there is good reason to attempt to distinguish
which is responsible if estimates of a by the methods described are much
greater than one. One source of such supplementary information is an exten-
sion of Experiment III to be considered briefly in the next section.

The assumption of no epistasis is no more realistic than that of no linkages.
It has been pointed out (Comstock and Robinson, 1948) that epistasis prob-
ably causes upward bias in the estimates of a, but that the amount of bias
may not be large. Subsequent investigation of several simple epistatic models
with respect to expected values of estimates of a from Experiments I and II
have turned up nothing to change that ])oint of view. It must be emphasized

502 R. E. COMSTOCK AND H. F. ROBINSON

that the matter has not been considered exhaustively, and the possibility
remains that in some materials epistasis would be responsible for serious
overestimation of a by the methods being discussed.

The authors' knowledge of the situation may be summarized briefly as
follows. It appears possible that with complete dominance the rule, a = 1.0,
epistasis might bias estimates upwards by as much as .10 to .25. This cannot
be considered serious against the background of an actual estimate of 1.6 as
reported for grain yield in corn by Robinson et al. (1949). On the other hand,
genetic models can be specified in which the consequences of epistasis would
be serious, but to date no such models have been discovered that seem likely
to have reality in nature.

Much investigation of the epistasis problem remains to be done. Theoreti-
cal studies of a variety of epistatic models are needed as a basis for under-
standing (1) how and to what extent inferences based on expectations de-
rived from non-epistatic models may be in error, and (2) how epistasis may
be measured and characterized experimentally. Equally important are ex-
perimental investigations of the role of epistasis in inheritance of quantita-
tive characters of various organisms. The problem in this connection is one
of knowing how to obtain critical information. The most familiar approach is
that of studying the regression of character measurements on levels of
homozygosity as represented at the extremes by inbred lines and Fi's and at
intermediate levels by Fa's and various sorts of backcrosses. While this ap-
proach has admitted shortcomings, it has not been exploited to the limit of its
usefulness. Other possibilities are suggested by Mather (1949).

EVALUATION OF THE ROLE OF LINKAGE

It was pointed out above that repulsion linkages are a source of upward
bias in estimates of a. In fact if a moderately high number of genes is postu-
lated, one finds on careful examination that estimates in excess of one seem
inevitable unless dominance at the locus level is considerably less than com-
plete. From the point of view of breeding methods it then becomes important
to distinguish between true overdominance and pseudo-overdominance. Par-
ticularly is this true if the latter is to any important degree a consequence
of linkages that are loose enough to allow their effects to be dissipated by
recombination in a few generations of random breeding, as opposed to the
rather durable associations that appear to be postulated by Anderson (1949).

If the assumption that frequencies of genes at all segregating loci are one-
half were tenable for generations beyond the F2, any of the three experiments
would provide a basis for obtaining information on the role of linkage. The
procedure would be to compare estimates obtained as described with others
obtained when parents were taken from an advanced generation (produced
under random mating, not with inbreeding) rather than from the F2. In fact
one might systematically repeat the experiment using each successive genera-

ESTIMATION OF AVERAGE DOMINANCE OF GENES 503

tion as it became available. Then if loose linkages were of much importance
in the first estimate, one would antici]:)ate a downward trend in the estimates
of a as more and more advanced generations were employed. Natural selec-
tion too weak to have much effect on results when the F2 is used could be the
source of significant changes in gene frequencies over a period of generations.
Hence the effects of recombination and of shifting gene frequencies would be
confounded in trends observed using either Experiment I or II.

Fortunately, E.xperiment III does not depend on any assumption about
gene frequencies. Letting q symbolize the population frequency of any gene
and 1 — ^ that of its allele, the genetic interpretation of al, and 0-^,; can be
expressed more generally than in Table 30.4 as ^2^(1 — q)u'' and l.qil —
q)a-ii~, respectively. One possible weakness of the proposal is apparent. If
shifts in gene frequency are variable by loci the weighting of individual c's in
a- is shifted slightly since it is now relative to ^(1 — q)u' rather than ur.
However, barring shifts greater than .2 which are unlikely unless a gene has
a very important effect, shifts in weights will be of minor magnitude since
q{\ — q) varies only between .21 and .25 as q varies from .3 to .7. Further-
more, shifts in weight are not a source of bias unless degree of dominance
(size of a) is correlated with importance of the gene. While this weakness
should not be overlooked, it appears of minor consequence. A partial check
could be made by accumulating seed of each generation for a yield compari-
son of the successive generations. If major gene frequency trends have oc-
curred at important loci they should be evidenced in higher yields by the
later generations.

The suggested extension of Experiment III is intrinsically the same sort of
technic as Mather (1949) has outlined for investigating linkage effects on
genetic variances.

DERIVATION OF GENETIC INTERPRETATIONS OF COMPONENTS
OF VARIANCE BETWEEN PROGENIES

The genetic constitution of a?„ and al,i of Experiment III will be derived
as examples. Derivations for components of the other two experiments are
given elsewhere (Comstock and Robinson, 1948). Initial assumptions will in-
clude only the following: regular diploid behavior at meiosis, no multiple
allelism, no epistasis. Restrictions are not being placed on gene frequencies
or linkage. To that extent the derivations to be given below are more general
than those cited above which assumed absence of linkage and gene frequen-
cies all equal to one-half.

The population sampled in Experiment III is outlined in Table 30.5. It
consists of an infinity of pairs of backcross progencies, one pair for each vari-
able parent that might be chosen from the F2 or a later generation from cross-
ing the two homozygous lines. Expected genetic values of each progeny are
indicated symbolically. Because all progenies must be of finite size, there will

504

R. E. COMSTOCK AND H. F. ROBINSON

be a sampling deviation between the actual and expected values of the prog-
enies. The variation among expected values is the variation due to genetic
differences among parents, and hence that to be considered in evaluating o-^
and al^l. c7^„ is the progeny variance due to genetic differences among the vari-
able parents, i.e., the variance of the pair means indicated in the next to the
last column of Table 30.5. a;„i, the progeny variance from interaction of geno-
types of the variable and homozygous parents, is one-half the variance of the
pair differences indicated in the final column.-

TABLE 30.5
POPULATION SAMPLED IN EXPERIMENT III

Variable

Homozygous Parent

Mean

Differ-

Parent*

Line A

Line B

ence

1
2
3

s

Xas

A'm

Xb2
Xb3

Xbs

X2

Xz

Xs

A
A

A

* The one chosen from F2 or later generation of the cross between lines
A and B.

s symbolizes an infinitely large number.

X's are expected genetic values of progenies, subscripts indicate parentage
of individual progenies.

Xi = mean of Xai and Xbi [where the subscript / identifies the variable
parent, e.g., Xi = (.Vqi + A'6i)/2].

Di = Xai — Xbi-

Now note that a pair mean or difference is the sum of contributions from
individual loci. Let

Xij be the contribution of the 7th locus to the ith pair mean, and
d ij be the contribution of the 7th locus to the /th pair difference.
Then

.Y,- = .^',1 -|- .v,2 -\- . . . Xij^

Di = d,i + dr2-\- ■ ■ ■ d,jy

where .Vis the number of contributing loci. Then the variances of pair means
and differences must be as follows:^

<t\. ( = a- ) = > cr- . -|- 2 7 (T .,

A m ' ^^ I] ' ^^^ X]k

] i, k

D lilt ^^^ d] ^^ djk

(1)

(2)

i.k

2. It is well known and easily verified that in the analysis of variance of any 2 X .^
table, the interaction variance is one-half that of the pair differences.

3. Since the variance of the sum of any number of variables is the sum of the variances
of those variables plus twice the sum of all covariances among them.

ESTIMATION OF AVERAGE DOMINANCE OF GENES 505

where alj is the variance of contributions from thejth locus to pair means,

(Txjk is the covariance of contributions from the _;th and ^th loci to pair
means,

alj is the variance of contributions from thej'th locus to pair diflferences,

crljk is the covariance of contributions from they'th and ^th loci to pair dif-
ferences, and 2^ indicates summation over all pairs of loci.

;, k

From equations (1) and (2) it is apparent that general expressions for al^ and
alti can be written as soon as we know (i) the variance of contributions of
any single locus to pair means and differences, and (ii) the covariance of con-
tributions of any two loci to pair means and differences.

With respect to two loci, there will be ten types of variable parent (when
classification is by types of gametes giving rise to the parent plant). Table
30.6 lists these, together with their frequencies in the population and the

TABLE 30.6

FREQUENCIES OF VARIABLE PARENT TYPES AND CONTRI-
BUTIONS* OF INDIVID U.\L LOCI TO EXPECTED
GENETIC V.-VLUES OF PROGENIES

FREQ.t

Homozygous Line , Mean

Difference

Variable
Parent

Bibi/Bibi

61B2/61B2 ;

i

di

di

XI

Xi

1st locus

2d locus

1st locus

2d locus

B\Bi/B\Bi

^2

Ml

OiUi

aiui

M2 I (Ml + ai«l)/2

{uo+a2Ui)/2

ui—aiui

aiUi—Ui

BiBi/B,h

ipr

«t

(au-u)/2

au

{u+au}/2\{u + au)/2

au/2

u~au

—u

Bibi/B\bi

r2

u

—u

au

au (u-\-au)/2

{au-u)/2

u—au

— u—au

B\Biih\Bi

2ps

(u+au)/2

au

(au-u)/2

u au/2

{u+au)/2

u

au —u

BxBi/hbi

ipt

iu+au)/2

(au-u}/2

(au—u)fl

(u + au)/2 au/2

au/2

u

—u

B\bi/b\Bi

Irs

{u+au)/2

{au-u)/2

{au-u)/2\ {u-}-au)/2 au/2

au/2

u

—u

B\bi/b\b2

2rt

(.u+au)/2

— u

{au—u)jV au auj2

(au~u)/2

u

—u—au

biBi/bxBi

52

au

au

— u 1 u (au—u)/2

{u+au)/2

u + au

au —u

biBz/bibi

2si

au

(au—u)/2

-u 1 {u+au)/2 {au-u)/2

au/2

u + au

-u

bib^/bxbi

/2

au

— u

— u 1 au [ {au—u)/2

(au-u)/2

u+au

—u—au

* Coded by subtraction of mi + =i (or m -\- zi) where z is the contribution Irom the locus when homozygous
for the b allele.

t On the basis that frequencies in which BiBi, Bibi, biBt, and bibi gametes are produced in the generation
preceding that used for variable parents are p, r, s, and t, respectively, p + r -{■ s + t = 1.0.

t For ease of printing, subscripts to a and « are omitted in rows beyond the first. However, the subscript
used in the first row of each column applies throughout the column.

contributions of the two loci to the expected genetic values of progenies and
to pair means and differences. As is evident from genotypes indicated for the
homozygous lines, the initial linkage phase assumed is repulsion. The re-
quired variances and co variances can be worked out directly from informa-
tion in the table. For example, the variance of contributions from the Lst
locus to pair means is

(p-ij^2pr+ r^) (u^ + a^u^) y i -{- (Ip s -{- 2pt -\- 2 r s + 2 rt) a\ul/ 4

+ (5'^ + 2 5/ + /'^)(aiMi-Mi)V4- (2x,)2

506 R. E. COMSTOCK AND H. F. ROBINSON

and the covariance of contributions to pair means from the 1st and 2d

loci is

p~{jii-\-aiUi) iu2 + a2U2) / ^-\-2p r (7<i + ai«i) (02^2) /4 + . . .

... + /- (ajWi — Ml) (a2M2 — U2) /4 — (IfXi) (2x2) .

The algebraic reductions are tedious, particularly for the covariances, and
will not be written out. However, the final expressions, for both the repulsion
and coupling phase, are listed in Table 30.7.

TABLE 30.7

VARIANCES AND COVARIANCES OF SINGLE

LOCUS CONTRIBUTIONS TO PAIR

MEANS AND DIFFERENCES

Initial Linkage Phase

Item

Coupling

Repulsion

9
<^ii

crio

(Txii- ■ ' ■
<>

o-di

<^<i2

0"dl2- • • ■

h(P+r)(s+t)ul
UP+sKr+Oul
\{pt — rs)thU2

2{p+r)(s+t)a\u\
2{p+s){r+l)a^nl

2(pt — rs)aiUia2U2

^ip+r)(s+tM

Up+s){r+t)i4

^ipl-rs)uiUn

2(/'+r)(5+/)(7?«?

2{p+s)ir+l)alul

2{rs—pi)aiUia2U2

Note now that if the frequencies of Bi and B^ (in the population from which
the variable parents are taken) are symbolized as qy and q2, then

p+ r = qi 5+/ = 1 - 9i

P+ s = qo r + / = 1 - (/2

(7'~.= lq.{l- q.)u'~

crj. = 2^.(1 - q.)ahC-.
Substituting in equations (1) and (2), we have

and

In general
and

A =

(3)

and

2,U,

0-2 = 1 0-2 = 'V 0.(1 - o.) a-Ji'.-\-2y^ {pt- rs) ..a.u.a,

> i.k

T

4-2^ {rs -pt) ^^a.u.a

(4)

k'^k

l,k

ESTIMATION OF AVERAGE DOMINANCE OF GENES 507

r

where ^^ indicates summation over all pairs of loci for which the initial

r

linkage phase was coupling and > , summation over pairs for which the

'77k

initial phase was repulsion.

When the associations between alleles at two loci are at equilibrium with
respect to coupling and repulsion phases, either because the loci are not
linked or because there has been sufficient opportunity for recombination

P = qjqh r = qj{\ - qf,)

s= (l-q,) qk 1= {\-q,)i\- q,)
and [pt — rs) — 0. Thus assuming no linkages (3) and (4) reduce to

as indicated in the preceding section. If, in addition, gene frequencies at all
segregating loci are assumed to be one-half, o^^ and a\a reduce to the values
assigned them in Table 30.4.

If there are linkages and equilibrium has not been reached, {pi — rs) will
be negative if the initial phase was repulsion, positive if the initial phase was
coupling. Thus covariances from repulsion and coupling linkages will tend to
cancel in al,. In fact if one assumes that enough loci are involved so that the
number of linked pairs must be high and that there is no reason why the
closer linkages should be predominantly in one phase, one is tempted to con-
clude that the sum of co variance will not be very important in a%.

On the other hand the covariance term is always positive in al,i, being a
function of (pt — rs) for coupling and of (rs — pt) for repulsion.^ Thus pres-
ence of any linkage, regardless of whether the two phases are equally frequent ,
will cause o-^, to be greater than I,q(l — q)ahi", except in the improbable
event that a for either or both members of pairs of linked loci is zero. And un-
less all linkages were in the coupling phase (in which case the ratio of a;, to
^2^(1 — q)u- would be the same as of o-^, to Zq(l — q)a-u^ and hence the
ratio of o-^ to ala unaffected by the linkages) crl,i/2al would overestimate
Zq(l — q)a?u-/'2,q(\ — g)^- so long as equilibrium in linkage associations had
not been attained through recombination. However, as stated in the preced-
ing section, the linkage bias becomes progressively smaller as equilibrium is
approached.

For purposes of illustration, consider application of the formulae in a
simple hypothetical situation. Assume that Experiment III is applied as
first described, with variable parents taken from an Fo, and that the quanti-

4. This assumes generality of dominance of the more favorable allele — that a will al-
most always be positive.

508 R. E. COMSTOCK AND H. F. ROBINSON

tative character to be studied is affected by seven pairs of genes that are dis-
tributed as follows in the parent lines:

Line A —BiBi bo bo b^ bz b^ biB-^B^B^B^ b-, b^
Line B—b^b^BoBoB^BiBiBib-^b-.b^bJi^B-,

with w's and a's having the following values:

Locus

1

2

3

1

4

2

5
1

6

7

u. . .

...1

2

2

1

a. . .

...6

.6

.8

.8

.8

1.0

.8

Note that less than complete dominance has been assumed for every locus.
Gene frequencies should be one-half in an F2, so ^^2^(1 — q)n" becomes
|Sm"^ and 2^(1 — q)ahi- becomes \l^aht^. Substituting numerical values of u
and a listed above, we obtain

and

2^(1 -(/) a%2 = 2.57

Now assume the following recombination values for })airs of loci:

Recombination

Pair

Va:

lue {v)

1 and 2

.3

3 and 4

.2

5 and 6

.1

5 and 7

.2

6 and 7

.1

All others

.5

Thus the seven loci fall in three groups that are either on three separate
chromosomes or, if on the same chromosome, far enough apart to allow free
recombination. In an F2 the values of p, r, s, and / will depend on v, the re-
combination value, and the original linkage phase as follows:

p r s t

Coupling (l-t')/2 v/2 v/2 {l-v)/2

Repulsion.... v/2 (l-i')/2 (l-i')/2 v/2

Hence {pi — rs) takes the following values:

Coupling. .
Repulsion .

.1

.2

.3

.5

.20
-.20

.15
-.15

.10
-.10

.0
.0

ESTIMATION OF AVERAGE DOMINANCE OF GENES 509

Substituting these and the numerical values of the a's and 7<'s, we find

Locus Linkage

Pair Phase (pt — rs)UiUk (pt — rs)aiUiakUk {rs — pt)aiUjakUk

1 and 2 Repulsion -.2 .0720

3 and 4 Coupling .3 .1920

5 and 6 Coupling . 4 . 3200

5 and 7 Repulsion -.15 .0960

6 and 7 Repulsion -.40 .3200

.05 .5120 .4880

With these three sums and the values found above for \'^q{\ — q)u- and
2^(1 — q)a~u^ we compute

ct2 = 2.00+ ( -.05) = 1.95

and

Thus, while

(T^^^^ = 2.57 + 2 (.5 12) +2 (.488) =4.5 7

., ^ Sa^^ ^ ^^7_ ^
'^' 2m2 2(2.0)

the experiment would estimate

flnj_ ^ _Jj^i_ =11-
2(t2 2 (1.95) • ' ■

Put differently, the estimate of a^ provided by Experiment III would in this
case have positive bias in the amount 1.17 — .64 = .53.

The foregoing example is given only to clarify the meaning of the formulae,
not to suggest the amount of bias that may actually be present in practice.
The actual bias with any specific material would depend on the amount of
linkage and the relative prevalence of coupling and repulsion phases. How-
ever, the bias can only be positive and may range from a negligible to a large
amount depending on the prevalence of repulsion linkage. While such bias
detracts from the described estimate as a criterion of average dominance at
the locus level, it is worth emphasizing that it represents a pseudo-overdomi-
nance effect which if persistent (due to closeness of linkages responsible) has
much the same significance for short-run breeding practice as true overdomi-
nance. If the bias declines fairly rapidly as opportunity is provided for re-
combination, Experiment III offers a means of measuring that decline and
thereby gaining an idea of the extent to which apparent dominance stems
from linkage relationships that are loose enough to allow a near approach to
equilibrium of linkage phases within a moderate number of generations.

AMOUNT OF DATA REQUIRED

An exhaustive consideration of this problem would require more space
than can be devoted to it here. Detailed discussion will therefore be limited
to one specific question. Let P symbolize the probability of an estimate of a

510 R. E. COMSTOCK AND H. F. ROBINSON

that is significantly^ greater tlian one. The question to be considered is as
follows: Assuming a particular value (> 1.0) for a how much data is required
if P is to be one-half? Procedure and the argument involved will be given in
detail for Experiment III ; only comparative results will be indicated for the
other two.

If the values of o-^ and alj listed in Table 30.4 are substituted into the ex-
pectations of M.31 and M-^-y of Table 30.3, we have

£(M3i) =0-2 + ^2^2

4

Note that when 1u~ — 1,a-u~, i.e., when a = 1.0, the two e.xpectations are
equal. But if a > 1.0, which means Za-u- > i)//-; then EiMs^) > E{M-ix).
Also, the estimate of a will exceed one only where M32 > Mn- It follows that
a one-tailed test of the hypothesis that E{M^<^ — E{Mu ^ 0 is also a test
of the hypothesis that d ^ 0. Since both mean squares are functions of ran-
dom variables (fixed effects do not contribute to either of them) the variance
ratio test, the F test, is applicable and P is equivalent to the probability that
the test ratio, M'i2/M;ii, will exceed Fa, where a is the probability level of the
test.

Let EiMn)/E(M-ii) = </>. If 0 = 1.0, Ms-jMu will be distributed in
samples in the same manner as F, otherwise it will be distributed as </)/^, i.e.,
M-ii/Mn for any probability point in its distribution will be exactly 4> times
the value of F for the same point in the F distribution. Thus the probability
of a sample value of M32/ M31 equal or greater than Fa is the same as that of
a sample value of F equal or greater than Fa/<1>. When degrees of freedom are
equal for the two mean squares, as will always be true in Experiment III, the
50 per cent point of the F distribution is 1.0. Hence P will be one-half when
the amount of data is that for which Fa (the lowest value of Msa/M^i to be
considered significantly different from one) is equal to 0.

We now must know the magnitude of </> when a is not unity.

^ £_(M32^ _ 4o-H:_^_a:M^
"^'EiMsJ' 4(r2+r2M^-

It varies with r, the number of replications in the experiment; with the ratio
of I,a~u' to 2/r which is a-; and with the ratio of a- to 2w^. Let c = o-yZw-.
Then

4c-hra2

0 =

4c+ r

Number of replications is subject to the will of the experimenter, but c and a

5. In the statistical sense, that the i)robabiHty of the observed or a larger estimate as
a consequence of random sampling is small.

ESTIMATION OF AVERAGE DOMINANCE OF GENES

511

are not. The logical procedure is to compute <^ for various combinations of
values of r, c, and a. This is tedious but very useful if the three items are
varied over rational ranges. A set of values for 4> is presented in Table 30.8.
Choice of rational values for a presented no difficulty since, in this connec-
tion, we are not so much concerned with its actual value as with the smallest
for which sufficient data to make P = .50 are not beyond the reach of the

experimenter.

TABLE 30.8

VALUE OF 0 FOR r = 2 AND VARYING
V/VLUES OF c AND a

a

c

Expt.

.25

.50

1.00

2.00

4.00

TTT

1.2

1.29

1.22

1.15

1.09

1.05

1.4

1.64

1.48

1.32

1.19

1.11

1.6

2.04

1.78

1.52

1.31

1.17

2.0

3.00

2.50

2.00

1.60

1.33

TT

1.4

1.27

1.17

1.09

1.05

1.03

1.6

1.44

1.27

1.15

1.08

1 04

2.0

1.80

1.50

1.29

1.15

1.08

T

1.4

1.13

1.10

1.06

1.04

1.02

16

1.21

1.15

1.10

1.06

1.03

2.0

1.38

1.28

1.18

1.11

1 06

Appropriate values for c will vary with the experimental material. The
range listed in the table was chosen for application to work with grain yield
of corn, a" is plot error variance which, judging from experience, will usually
be between 50 and 160 when yield is measured in bushels per acre.*^ This cor-
responds to a range of about 10 to 18 per cent for the coefficient of variation
if mean bushel yield is 70. 2m- is twice the additive genetic variance in the
Y'l population used. Robinson el al. (1949) worked with three F2 populations
and reported .0056 as an estimate of the average amount of additive genetic
variance where yield was measured as pounds per plant. Converted to
bushels per acre this figure becomes 78.4. More recent work at the North
Carolina Experiment Station has yielded estimates of the same order of mag-
nitude. From these results it appears that additive genetic variance will in
many cases be between 20 and 100 and hence that Zm^ will be between 40
and 200. The extreme values for c, if 0-- and '^u"- are within ranges suggested
above, ^ are 50/2(iO = .25 and 160/40 = 4.0.

6. In work at the North CaroHna station it has been quite close to 50.

7. Note that the suggested range for cr^ is off-center upwards and that for ^.ti^ is off-
center downwards with respect to estimates from North Carolina data. This was done
deliberately in an effort to be on the safe side. EfTiciency of the experiment suff'ers from
large <r^ or small Sm^.

512

R. E. COMSTOCK AND H. F. ROBINSON

All values of ^ listed in the table are for r = 2. However, the effect of mul-
tiplying r by any constant is the same as dividing c by the same constant.
Hence, <^ for c = 1 and r = 8 is the same as for c = .25 and r — 2; 0 for
c = 4 and r = 4 is the same as for c — 2 and r = 2; etc.

Table 30.9 lists the approximate degrees of freedom required for Msi and
Mzi if F.05 is to equal 0 so that P will be .50. As an example to clarify the
significance of this table, assume that c = 1.0, a = 1.4, and r = 2. Then if
the data provide 142 degrees of freedom for both Mzi and M32, the probabili-

TABLE 30.9

.\PPROXIAJ.ATE DEGREES OF FREEDOM* RE-
QUIRED TO MAKE P = .50 IN
EXPERIMENT III

c

a

.25

.50

1.00

2.00

4.00

1.2

168

275

555

1460

4550

1.4

45

72

142

360

995

1.6

23

34

63

150

450

2.0

10

14

24

50

134

♦Obtained assuming normal distribution of Fisher's z and em-
ploying the facts that <tv = i(l/f\ + I (fi) (where h and J2 are de-
grees of freedom for the two mean squares) and F = e-'.

ty of the estimate of a being significantly greater (at the 5 per cent point)
than one is one-half. Degrees of freedom can be related to amount of data as
follows. Suppose that n, the number of progeny pairs per set, is 8. Then de-
grees of freedom will be 7/8 the number of progeny pairs, and assuming two
replications, r = 2, degrees of freedom will be 7/32 the number of plots in the
experiment. The 142 degrees of freedom indicated in the specific instance
singled out above would require data on a total of about 650 plots.

An obvious question is whether increasing replications is as effective as in-
creasing the number of progeny pairs. Consider the case where c = 4.0 and
a — 1.6. Degrees of freedom required are 450 when r = 2. But remembering
that multiplying r by a constant has the same effect on 0 as division of c by
the same constant, we see that with four replications degrees of freedom re-
quired would be only 150. Thus with two replications a total of about 2056
plots would berequired, whereas with four replications only about 1370 would
be needed. The same is not true for the entire area of the table. Careful in-
spection will show that when c is 1.00 or less, doubling the number of progeny
pairs is more effective than increasing replications from two to four. But
when c is 2.0 or greater, the opposite is true.

Also pertinent are (1) the effect on P of increasing data above amounts in-
dicated in Table 30.9, and (2) the probability of an estimate of a that is less

ESTIMATION OF AVERAGE DOMINANCE OF GENES 513

than one even though the true value exceeds one. F becomes about .75 if the
data are doubled, between .85 and .9 if the data are tripled, and about .95 if
the data are quadrupled. With the degrees of freedom indicated, the proba-
bility of an estimate less than 1.0 for a is in all cases close to .05, and that of
an estimate significantly less than one is much smaller. This is an important
point since it means a very small chance of erroneously concluding that a is
less than one if its real value is greater thaji one by any very important
amount.

The general point to note is that the amounts of data indicated in Table
30.9 are moderate for any combination of c ^ 1.0 and a ^ 1.4. In addition,
it is not prohibitive when both a and c are (within the ranges considered)
either large or small. Actually, as indicated by earlier references, estimates
of c for corn yield from data collected to date at the North Carolina Experi-
ment Station have been somewhat less than .50.

An exact F test of the hypothesis that o ^ 1.0 is not provided in the vari-
ance analysis of either Experiment I or II. In both instances there is a func-
tion {R) of three mean squares that provides an approximate F test. They
are given below. Remember for Experiment II that we are assuming m = n

Experiment R

I R^ = {2n+i)Mn/{iMn+2nMn)

II Ri = {2n+\)M^/{Mio+2nM<,i)

and using M20 to symbolize the mean of M 21 and ^22- As was true for the test
ratio of Experiment III, the expectations of numerator and denominator are
equal in both of these ratios w-hen a — 1.0, but when a > 1.0 the expectation
of the numerator exceeds that of the denominator. Also, the estimate of a is
greater than one only when the test ratio is greater than one. Values of (f> for
Experiments I and II in Table 30.8 are the ratios of expectations of numera-
tor and denominator in these test ratios. As suggested by relative sizes of 4>
for the three experiments, more data are required in Experiment II than in
III, and still more are required in I. However, the degrees of freedom sup-
plied are greater relative to numbers of plots used than in III so differences
in data required cannot be judged properly in terms of the 0's.

The data requirement cannot be determined as accurately as for Experi-
ment III, primarily because degrees of freedom that should properly be as-
signed to the denominators of the test ratios cannot be known exactly
though they can be approximated by the method of Satterthw-aite (1946).
For the same reason, determination of the approximate data requirement is
more time-consuming. Attention will therefore be confined to the three situa-
tions indicated below. Degrees of freedom for Experiment I refer to the mean
square, Mn, and for Experiment II to M23. In both cases, n was assumed to
be 4.0. Thus in II, progenies per set would be 16 as was assumed for Experi-
ment III. This would make degrees of freedom for M 23 be 9/32 of the number

514

R. E. COMSTOCK AND H. F. ROBINSON

of plots, lir = 2. If male groups per progeny set are 4.0, in Experiment I, as
in the work of Robinson et al., there would also be 16 progenies per set and
degrees of freedom for M12 would be 12/32 of the number of plots.

Experiment III is obviously the most powerful and I the least powerful of
the three. In the three cases examined, the plot requirement for I is from ten
to twelve times that of III. Experiment II is intermediate, requiring from
two to four times the data needed in III. It may be of interest that in the

c

a

Degrees of Freedom
Required

Plots Required
IF r = 2

E.xpt. I

II

III

I

II

III

.50
1.00

.25

1.4

2.0
1.6

1525
440
480

315

120

60

72
24
23

4066
1173
1280

1120
426
213

329
110
105

work reported by Robinson et al. (1949) in which Experiment I was used in
studying corn yield there were about 500 degrees of freedom for Mn- The esti-
mate of a was 1.64 and, by the approximate F test, was just significant at the
5 per cent point.

Before leaving the subject, it should be noted that the problem of data re-
quired has been dealt with under the original assumptions. If what have been
called estimates of a are biased upward by linkage or epistasis, their expected
values are larger than a, and the foregoing has relevance to the expected
values of the estimates rather than to a itself. To exemplify, suppose that a
were 1.2, but as a result of bias from epistasis and linkage the expected value
of the Experiment III estimate were 1.2. Then assuming c = .25 and r = 2,
the probability of the estimate being significantly above 1.0 would be .50 if
the data furnished 168 degrees of freedom (Table 30.9), the same number re-
quired if a were 1.0 and the estimate unbiased. Thus, we see that the proba-
bility of an estimate significantly greater than one is a function of the expect-
ed value of the estimate rather than of a when the two are not equal. The
corollary, that an estimate (obtained as described) significantly greater than
one is not final proof of overdominance at the locus level, has been indicated
in preceding sections.

CONCLUDING REMARKS

To attempt a general discussion of what has been presented appears un-
wise. It would almost certainly lead to some unnecessary repetition and could
do more to confuse than to clarify. However, certain comments seem in order.

With regard to the experiments themselves, III appears definitely the
most useful (1) because it is the most powerful, and (2) because it can be em-
ployed to learn something about the effect of linkage on the estimate of a.

ESTIMATION OF AVERAGE DOMINANCE OF GENES 515

It should not be necessary to comment on the role of statistics in the de-
vising and evaluation of schemes for investigating the inheritance of quanti-
tative characters. If the importance of statistics in this area of research has
not been adequately demonstrated by the foregoing, general statements
could hardly be expected to be convincing. The point to be emphasized is
that more theoretical investigation of experimental technics in quantitative
inheritance is badly needed. For example, insofar as the three experiments
considered here are concerned, more information is needed on the biases re-
sulting from various sorts of epistasis. It is possible that such biases are
greatest in Experiment III and would detract from its apparent superiority.
It is also possible that the biases from epistasis differ between the experi-
ments and that the differences vary with type of epistasis. In that event,
comparison of results from two or more of the experiments could conceivably
contribute to our knowledge of epistasis.

Investigation of the power of a variety of technics used in quantitative
genetic research also would be fruitful. The intent is not to imply that there
are no such procedures for which the power is known within satisfactory
limits, but only to point out that there are some for which this is not the
case. For example, mention has been made of the use of parent, Fi, F2, and
backcross means for investigating ep)istasis, but to the authors' knowledge
there is nothing in the literature concerning amount of data required to in-
sure that the chances of erroneous conclusions from such a study would be
small.

Equally important is continued search for useful technics and procedures.
It is entirely possible that approaches may thereby be discovered which are
more efficient than any presently known. As a case in point, at the time the
work described by Robinson et at. (1949) was planned we had not thought of
the procedure designated here as Experiment III which, so far as we know,
has not been previously described as a technic for investigation of dominance.
Judging from findings of the preceding section, the same amount of work
using the latter procedure would have provided considerably more precise
estimates.

While attention herein has been devoted to estimation of average level of
dominance, the experiments described provide other information as well. The
data collected can be used also for estimation of additive genetic variance,
variance due to dominance deviations, and the genetic and phenotypic co-
variances and correlations of pairs of characters.

LITERATURE

No attempt has been made to cite all of the various publications that in
one way or another were stimulatory to the above discussion, since a careful
attempt to assign credit where due would have made the manuscript consid-
erably longer. Most interested readers will be familiar with relevant litera-

516 R. E. COMSTOCK AND H. F. ROBINSON

ture, but examples will be given here of papers that might have been cited.

The utilization of genetic variance component estimates is illustrated by
numerous publications, for example, Baker et al. (1943). The composition (in
terms of additive genetic variance and variance due to dominance deviations)
of the estimable genetic variance components in the sort of population on
which Experiment I is based is known generally and is indicated by Lush
et al. (1948).

An experiment very similar to II but not designed with as specific informa-
tion about dominance as its objective has been reported by Hazel and
Lamoreux (1947).

The general pattern for genetic interpretation of variance components
arising from Mendelian segregation was set in such papers as those by
Fisher (1918), Fisher et al. (1932), and Wright (1935).

Other procedures for estimation of dominance have been described by
Fisher et al. (1932), Mather (1949), and Hull (in this volume).

Bibliography

Anderson, Edgar, 1939a The hindrance to gene recombination imposed by Hnkagc: an

estimate of its total magnitude. Amer. Nat. 73:183-188.

, 1939b Recombination in species crosses. Genetics 24:668-698.

, 1944 Homologies of the ear and tassel in Zea mays. Ann. Missouri Bot. Gard.

31:325-342.
, 1946 Maize in Mexico: a preliminary survey. Ann. Missouri Bot. Gard. 33:147-

247.

, 1949 Introgressive hybridization. 109 pp. John Wiley & Sons, Inc., New York.

-, and William L. Brown, 1950 The history of common maize varieties in the

United States corn belt. Jour. New York Bot. Gard. 51:242-267.
, and Hugh C. Cutler, 1942 Races of Zea mays. I. Their recognition and classitica-

tion. Ann. Missouri Bot. Gard. 29:69-88.
, and Ralph O. Erickson, 1941 Antithetical dominance in North American

maize. Proc. Nation. Acad. Sci. U.S.A. 27:436-440.

Anonymous, 1949 Informe de los Anos 1944-1948, Centro Nacional de Agronomia. EI
Salvador.

Aristotle, History of animals. Tr. by D'Arcy W. Thompson. 633 pp. Oxford Press,
Oxford. 1910.

AsHBY, E., 1930 Studies in the inheritance of physiological characters. I. A physiological
investigation of the nature of hybrid vigour in maize. Ann. Botany 44:457-467.

, 1932 Studies in the inheritance of physiological characters. H. Further experi-
ments upon the basis of hybrid vigour and upon the inheritance of eiiftciency index and
respiration rate in maize. Ann. Botany 46:1007-1032.

— , 1936 Hybrid vigour in maize. Amer. Nat. 70:179-181.

, 1937 Studies in the inheritance of physiological characters. HI. Hybrid vigour in

the tomato. Pt. I. Manifestation of hybrid vigour from germination to the onset of

flowering. Ann. Botany (New Series) 1:11-41.
AsHTON, T., 1946 The use of heterosis in the production of agricultural and horticultural

crops. Imper. Bur. Plant Breed, and Genetics, School of Agric, Cambridge, England.

30 pp.
Baker, A. L., and J. R. Quesenberry, 1944 Comparison of growth of hereford and Fi

hereford-shorthorn heifers. Jour. Animal Sci. 3:322-325.
Baker, Marvel L., L. N. Hazel, and C. F. Reinmiller, 1943 The relative importance

of heredity and environment in the growth of pigs at different ages. Jour. Animal Sci.

2:3-13.
Barthollet, S., 1827 Phenomenes de Facte mysterieux de la fecondation. Memor. Soc.

Linn. Paris. 1:81-83.
Barton, A. A., 1950 Some aspects of cell division in Saccharomyces cerevisiae. Jour.

Gen. Microbiol. 4:84-86.
Beadle, G. W., 1935 Crossing over near the spindle attachment of the X chromosomes

in attached-x triploids of Dt osophila mclanogasicr . Cienetics 20:179-191.

, 1945a Biochemical genetics. Chem. Rev. 37:15-96.

, 1945b Genetics and metabolism in Neurospora. Physiol. Rev. 25:643-663.

517

518 BIBLIOGRAPHY

Beadle, G. W., 1948 Some recent developments in chemical genetics. Fortschr. Chemie

organ. Naturst. 5:300-330.
, 1949 Genes and biological enigmas. Pages 184-245; 316-317. Science in Progress.

6th Series. Yale University Press: New Haven.
, and V. L. Coonradt, 1944 He\.eroca.Tyosis in N euros por a crassa. Genetics 29: 291-

308.
, and E. L. Tatum, 1941 Genetic control of biochemical reactions in Neurospora.

Proc. Nation. Acad. Sci. U.S.A. 27:499-506.
Beal, William J., 1876, 1877, 1881, 1882 Repts. Michigan State Board Agric.
, 1878 The improvements of grains, fruits, and vegetables. Rept. Michigan State

Board Agric. 17:445-457.

, 1880 Indian corn. Rei)t. Michigan State Board Agric. 19:279-289.

Beale, G. H., 1941 Gene relations and synthetic processes. Jour. Genetics 42 : 197-214.
Bernstein, Felix, 1924 Ergebnisse einer biostatistischen zusammenfassenden Betrach-

tung iiber die erblichen BJutstrukturen des Menschen. Klin. Wochenschr. 3, II: 1495-

1497.
Blakeslee, a. F., J. Belling, M. E. Farnham, and A. D. Bergner, 1922 A haploid

mutant in the jimson weed. Datura stramonium. Science 55:646-647.
, and S. Satina, 1944 New hybrids from incompatible crosses in Datura through

culture of excised embryos on malt media. Science 99:331-334.
Blunn, C. T., and M. L. Baker, 1949 Heritability estimates of sow productivity and lit-
ter performance. Jour. Animal Sci. 8:89-97.
Boivin, a., 1947 Directed mutation in colon bacilli by an inducing principle of desoxyri-

bonucleic nature: its meaning for the general biochemistry of heredity. Cold Spring

Harbor Symposia Quant. Biol. 12:7-17.
Bolsunow, I., 1944 Zur Untersuchung der Heterosis bei Nicotiana rustica L. V. Fort-

gesetzte Selbstbefruchtung und Ernteertrag. Zeitschr. Pflanzenzucht. 26:223-244.
Bonner, James, and Harriet Bonner, 1948 The B vitamins as plant hormones. Vita-
mins and Hormones. Academic Press, Inc.: New York. Vol. 6:225-275.
, and S. G. Wildman, 1946 Enzymatic mechanisms in the respiration of spinach

leaves. Arch. Biochem. 10:497-518.
Boyd, W. C, 1939 Blood groups. Tabulae Biologicae 17:113-240.
, 1943 Fundamentals of immunology. 446 pp. Interscience Publishers, Inc.: New

York.

, 1945 Rh blood factors: An orientation review. Arch. Path. 40:114-127.

Brieger, F. G., 1950 The genetic basis of heterosis in maize. Genetics 35:420-445.
Briles, W. E., W. H. McGibbon, and M. R. Irwin, 1950 On multiple alleles effecting

cellular antigens in the chicken. Genetics 35:633-652.
Brink, R. A., 1929 Studies on the physiology of a gene. Quart. Rev. Biol. 4:520-543.
, and D. C. Cooper, 1940 Double fertilization and development of the seed in an-

giosperms. Bot. Gaz. 102 : 1-25.
, , 1944 The antipodals in relation to abnormal endosperm behavior in

II ordeum jubatum- X Secale cereale hybrid seeds. Genetics 29:391-406.
, , 1947 The endosperm in seed development. Bot. Rev. 13:423-541.

Broquist, H. D., and E. E. Snell, 1949 Studies of the mechanism of histidine synthesis

in lactic acid bacteria. Jour. Biol. Chem. 180:59-71.
Brown, William L., 1949 Numbers and distribution of chromosome knobs in U.S. maize.

Genetics 34:524-536.
, and Edgar Anderson, 1947 The northern flint corns. Ann. Missouri Bot. Gard.

34:1-28.
. , 1948 The southern dent corns. Ann. Missouri Bot. Gard. 35:255-268.

BIBLIOGRAPHY 519

Bruce, A. B., 1910 The Mendelian theory of heredity and the augmentalion of vij^or.
Science 32:627-628.

Bruner, D. W., and P. R. Edwards, 1947 Changes in the nonspecific antigens of Sal-
monella. Jour. Bact. 53:359.

Burger, J., 1809 Naturgeschichte, Culture und Benutzung des Mais. VV'ein.

BuRKHOLDER, Paul R., and Ilda McVErGH, 1940 Growth and differentiation of maize in
relation to nitrogen supply. Amer. Jour. Hoi. 27:414-424.

Buzz.vti-Traverso, a., 1947a Genetica di i)opolazioni in Droso[)hila. V. Selezione nalu-
rale in popolazioni artiticiali di Drosophila melanogastcr . Memor. 1st. Ital. Idrobiol.
4:41-62.

, 1947b Su alcuni casi di evoluzione in bottiglia. Memor. 1st. Ilal. Idrobiol. 4:115-

120.

, 1950 Genetic structure of natural populations and interbreeding units in the hu-
man species. Cold Spring Harbor Symposia Quant. Biol. 15: 13-23.

Carnahan, H. L., 1947 Combining ability in flax (Liniini usitatissimiim). M.S. Thesis.
Univ. of Minnesota.

Carroll, W. E., and E. Roberts, 1942 Crossbreeding in swine. Illinois Agric. Expt. Sta.
Bull. 489.

Carson, H. L., 1946 The selective elimination of inversion dicentric chromatids during
meiosis in the eggs of Sciara impatiens. Genetics 31 :95-113.

Caspari, E., 1948 Cytoplasmic inheritance. Advances in Genetics Vol II: 1-66. Academic
Press, Inc.: New York.

, 1950 On the selective value of the alleles Rt and rt in Ephcstia kuhniella. Amer.

Nat. 84:367-380.

Castle, W. E., 1946 Genes which divide species or produce hybrid vigor. Proc. Nation.
Acad. Sci. U.S.A. 32:145-149.

Catcheside, D. G., 1944. Polarized segregation in an Ascomycete. Ann. Botany (New
Series) 8:119-130.

Cavalli, L. L., 1950 The analysis of selection curves. Biometrics 6:208-220.

Chase, S. S., 1949a Monoploid frequencies in a commercial double cross hybrid maize,
and in its component single cross hybrids and inbred lines. Genetics 34:328-332.

, 1949b Spontaneous doubUng of the chromosome complement in monoploid sporo-

phytes of maize. Proc. Iowa Acad. Sci. 56:113-115.

Chipman, R. H., and T. H. Goodspeed, 1927 Inheritance in Nicotiana tabacum. VIII.
Cytological features of purpurea haploid. Univ. California Publ. Bot. 11:141-158.

Christensen, H. M., and R. B.amford, 1943 Haploids in twin seedlings of peppers. Capsi-
cum annuum L. Jour. Hereditj' 34:99-104.

Clark, Frances J., 1940 Cytogenetic studies of divergent meiotic spindle formation in
Zea mays. Amer. Jour. Bot. 27:547-559.

Collins, G. N., 1909 A new type of Indian corn from China. U.S. Dept. Agric. Bur. Plant
Ind. Bull. 161.30 pp.

, 1921 Dominance and the vigor of first generation hybrids. Amer. Nat. 55:116-133.

CoMSTOCK, R. E., and H. F. Robinson, 1948 The components of genetic variance in popu-
lations of biparental progenies and their use in estimating the average degree of domi-
nance. Biometrics 4:254-266.
, , and P. H. Harvey, 1949 A breeding procedure designed to make maxi-
mum use of both general and specific combining ability. Agron. Jour. 41:360-367.

, and L. M. Winters, 1944 A com{)arison of effects of inbreeding and selection on

performance in swine. Jour. Animal Sci. 3:380-389.
Cooper, D. C, and R. A. Brink, 1940 Partial self-incompatibilily and the collapse of
fertile ovules as factors affecting seed formation in alfalfa. Jour. Agric. Res. 60:453-472.

520 BIBLIOGRAPHY

Cooper, D. C, and R. A. Brink, 1944 Collapse of the seed following the mating Horde-

umjubatum X Secale cereale. Genetics 29:370-390.
, , 1949 The endosperm-embryo relationship in an autonomous apomict,

Taraxacum officinale. Bot. Gaz. 111:139-153.
COPELAND, F. C, 1940 Growth rates in inbred and hybrid corn embryos. Collecting Net

15:169.
Cowan, J. R., 1943 The value of double cross hybrids involving inbreds of similar and

diverse genetic origin. Sci. Agric. (Ottawa) 23:287-296.
CowiE, D. B., E. T. Bolton, and M. K. Sands, 1950 Sulfur metabolism in Escherichia

coH. Jour. Bact. 60:233-248.
Crabb, a. Richard, 1947 The h3^brid corn makers: prophets of plenty. xxv-)-331 pp.

Rutgers University Press: New Brunswick, New Jersey.
Crow, James F., 1948 Alternative hypotheses of hybrid vigor. Genetics 33:477-487.
, 1952 Dominance and overdominance in heterosis. Heterosis, Chapter 18. Iowa

State College Press: Ames, Iowa.
CuMMiNGS, J. N., L. M. Winters, and H. A. Stewart, 1947 Heritability of some factors

affecting productivity of brood sows. Jour. Animal Sci. 6:297-304.
Cunningham, J. C, 1948 Maize bibliography for the years 1888-1916. Contributions

Iowa Corn Res. Inst. 3:67-189.
Danforth, C. H., 1927 Hereditary adiposity in mice. Jour. Heredity 18:153-162.
Darlington, C. D., and L. La Cour, 1941 The genetics of embryo sac development. Ann.

Botany (New Series) 5:547-562.
Darwin, Charles, 1868 The variation of animals and plants under domestication. J.

Murray: London.
, 1877 The etYects of cross and self fertilization in the vegetable kingdom. viii-f482

pp. D. Appleton and Company: New York.
Davenport, C. B., 1908 Degeneration, albinism and inbreeding. Science 28:454-455.
Davis, B. D., 1948 Isolation of biochemically deficient mutants of bacteria by means of

penicillin. Jour. Amer. Chem. Soc. 70:4267.
, 1949 The isolation of biochemically deficient mutants of bacteria by means of

peniciUin. Proc. Nation. Acad. Sci. U.S.A. 35:1-10.
DeTurk, E. E., J. R. Holbert, and B. W. Howk, 1933 Chemical transformations of

phosphorus in the growing corn plant, with results on two first generation crosses. Jour.

Agric. Res. 46:121-141.
DeVries, Hugo, 1907 Plant breeding: comments on the experiments of Nilsson and Bur-
bank. xvi+360 pp. The Open Court Publishing Company: Chicago.
Dickerson, G. E., 1947 Composition of hog carcasses as influenced by heritable difYer-

ences in rate and economy of gain. Iowa Agric. Expt. Sta. Res. Bull. 354. 31 pp.
— , 1949 Importance of heterosis for total performance in animals. Proc. Eighth

Internationa] Congress of Genetics: Abstr. 560. (Suppl. Vol. of Hereditas, Berlingska

Boktryckeriet, Lund, Sweden.)
, 1951 Effectiveness of selection for economic characters in swine. Jour. Animal

Sci. 10:12-18.
, and J. W. Go WEN, 1947 Hereditary obesity and efficient food utilization in mice.

Science 105:496-498.
, and J. C. Grimes, 1947 Effectiveness of selection for efficiency of gain in Duroc

swine. Jour. Animal Sci. 6:256-287.
, J. L. Lush, M. L. Baker, J. A. Whatley, Jr., and L. M. Winters, 1947 Perform-

ance of inbred lines and line-crosses in swine. Jour. Animal Sci. 6:477.
, • , and C. C. Culbertson, 1946 Hybrid vigor in single crosses between in-

bred Unes of Poland China swine. Jour. Animal Sci. 5:16-24.

BIBLIOGRAPHY 521

DoBZHANSKY, Th., 1941 Gcnctics and the origin of species. Second Edition. 446 p]). Co-
lumbia Univ. Press: New York.

, 1943 Genetics of natural populations. IX. Temporal changes in the composition

of populations of Drosophila pscudoobsciira. Genetics 28:162-186.

, 1947a A directional change in the genetic constitution of a natural poi)ulation of

Drosophila pscudoobscura. Hereditj^ 1:53-64.

1947b Genetics of natural populations. XIV. A response of certain gene arrange-

ments in the third chromosome of Drosophila pscudoobscura to natural selection.

Genetics 32: 142-160.
, 1949 Observations and experiments on natural selection in Drosophila. Proc.

Eighth International Congress of Genetics: 210-224. (Suppl. Vol. of Hereditas, Ber-

lingska Boktryckeriet, Lund, Sweden.)
, 1950 Genetics of natural populations. XIX. Origin of heterosis through natural

selection in populations of Drosophila pseudoobsctira. Genetics 35:288-302.
, A. M. HoLz, and B. Spassky, 1942 Genetics of natural populations. VIII. Con-

cealed variability in the second and fourth chromosomes of Drosophila pscudoobscura.

Genetics 27:463-490.
Dodge, B. O., 1942 Heterocaryotic vigor in Neurospora. Bull. Torrey Bot. Club 69:75-91.
, Mary B. Schmitt, and Anita Appel, 1945 Inheritance of factors involved in one

type of heterocaryotic vigor. Proc. Amer. Phil. Soc. 89:575-589.
DoERM.ANN, A. H., 1944 A lysineless mutant of Neurospora and its inhibition by arginine.

Arch. Biochem. 5:373-384.
Dunn, L. C, and E. Caspari, 1945 A case of neighboring loci with similar effects. Genet-
ics 30:543-568.
East, Edward M., 1907 The relation of certain biological principles to plant breeding.

Connecticut Expt. Sta. Bull. 158. 93 pp.
, 1908 Inbreeding in corn. Rept. Connecticut Agric. Expt. Sta. for 1907. Pp.

419-428.
, 1909 The distinction between development and heredity in inbreeding. Amer.

Nat. 43:173-181.

, 1921 A study of partial sterility in certain hybrids. Genetics 6:311-365.

, 1936 Heterosis. Genetics 21:375-397.

, and H. K. Hayes, 1912 Heterozygosis in evolution and in plant breeding. U.S.

Dept. Agric. Bur. Plant Indust. Bull. 243. 58 pp.

and D. F. Jones, 1919 Inbreeding and outbreeding: their genetic and sociological

significance. 285 pp. J. B. Lippincott Co.: Philadelphia and London.

-, 1920 Genetic studies on the protein content of maize. Genetics 5:543-610.

Edwards, P. R., and D. W. Bruner, 1942 Serological identification of Salmonella cul-
tures. Kentucky Agric. Expt. Sta. Circ. 54. 35 pp.

Einset, J., 1943 Chromosome length in relation to transmission frequency of maize tri-
somes. Genetics 28:349-364.

Emerson, R. A., and G. W. Beadle, 1932 Studies of Euchlaena and its hybrids with
Zea. II. Crossing-over between the chromosomes of Euchlaena and those of Zea. Zeit-
schr. indukt. Abstamm.- u. Vererbungsl. 62:305-315.

, and H. H. Smith, 1950 Inheritance of number of kernel rows in maize. Cornell

University Agric. Expt. Sta. Mem. 296. 30 pp.

Emerson, S., 1947 Growth responses of a sulfonamide-requiring mutant strain of Neuro-
spora. Jour. Bact. 54:195-207.

, 1948 A physiological basis for some suppressor mutations and possibly for one

gene heterosis. Proc. Nation. Acad. Sci. U.S.A. 34:72-74.

— , 1949 Competitive reactions and antagonisms in the biosynthesis of amino acids

by Neurospora. Cold Spring Harbor Symposia Quant. Biol. 14:40-47.

522 BIBLIOGRAPHY

Emik, L. O., and C. E. Terrill, 1949 Systematic procedures for calculating inbreeding

coeti&cients. Jour. Heredity 40:51-55.
Ephrussi, Boris, 1942a Analysis of eye color differentiation in Drosophila. Cold Spring

Harbor Symposia Quant. Biol. 10:40-48.
, 1942b Chemistry of "eye color hormones" of Drosophila. Quart. Rev. Biol. 17:

327-338.
Everett, H. L., 1949 A genie series controlling chloroplast pigment production in diploid

Zea mays. Proc. Nation. Acad. Sci. U.S.A. 35:628-6.34.
Fairfield Smith, H., 1936 A discriminant function for plant selection. Ann. Eugenics

7:240-250.
Ferguson, L. C, 1941 Heritable antigens in the erythrocytes of cattle. Jour. Immunol.

40:213-242.
, Clyde Stormont, and M. R. Irwin, 1942 On additional antigens in the erythro-
cytes of cattle. Jour. Immunol. 44:147-164.
Fisher, R. A., 1918 The correlation between relatives on the supposition of Mendelian

inheritance. Trans. Roy. Soc. Edinburgh 52: Pt. 2, 399-433.

, 1931 The evolution of dominance. Biol. Rev. Cambridge Phil. Soc. 6:345-368.

, 1949 The theor\- of inbreeding. 120 pp. Oliver and Boyd: Edinburgh.

, F. R. Immer, and Olof Tedin, 1932 The genetical interpretation of statistics

of the third degree in the study of quantitative inheritance. Genetics 17:107-124.
Flor, H. H., 1947 Inheritance of reaction to rust in flax. Jour. Agric. Res. 74:241-262.
FocKE, VV. O., 1881 Die Pflanzen-Mischlinge. 569 pp. G. Borntraeger: Berlin.
Gaines, E. F., and H. C. Aase, 1926 A haploid wheat plant. Amer. Jour. Bet. 13:373-385.
Gallesio, G., 1813 Teoria della riproduzione vegetal. 136 pp. Vienna.
Garner, W. W., 1946 The production of tobacco. 516 pp. Blakiston Co. : Philadelphia.
Garrod, a. E., 1923 Inborn errors of metabolism. 2nd Edition. 216 pp. Oxford Univ.

Press: London.
Gartner, C. F., 1827 Correspondenz in der Flora. Bot. Zeitschr. 10:74.
, 1849 Versuche und Beobachtungen iiber die Bastarderzeugung im Pflanzenreich.

791 pp. Stuttgart.
Gates, R. R., 1929 A haploid Oenothera. Nature (London) 124:948.
Goldschmidt, Richard, 1938 Physiological genetics. 375 pp. McGraw-Hill Book Com-
pany, Inc.: New York and London.
GoODALE, H. D., 1938 A study of inheritance of body weight in the albino mouse by selec-
tion. Jour. Heredity 29:101-112.
GowEN, J. VV., 1945 Genetic aspects of virulence in bacteria and viruses. Ann. Missouri

Bot. Card. 32:187-211.
, and Leslie Johnson, 1946 On the mechanism of heterosis — metabolic capacity

of different races of Drosophila melanogaster for egg production. Amer. Nat. 80: 149-179.
-, J. Stadler, and L. E. Johnson, 1946 On the mechanism of heterosis — the chromo-

somal or cytoplasmic basis for heterosis in Drosophila melanogaster. Amer. Nat. 80:506-

531.
Gray, Asa, 1858 Action of foreign pollen upon the fruits. Amer. Jour. Sci. and Arts (Series

2) 25:122-123.
Green, J. M., 1948 Inheritance of combining ability in maize hybrids. Jour. Amer. Soc.

Agron. 40:58-63.
Gregory, F. G., and F. Crowther, 1928. A physiological study of varietal differences

in plants. I. .\ study of the comparative yields of barley varieties with different manur-
ing. .\nn. Botany 42:757-770.
, , 1931 A physiological study of varietal differences in plants. II. Further

evidence for the differential response in yield of barley varieties to manurial deficiencies.

Ann. Botanv 45:579-592.

BIBLIOGRAPHY 523

Gross, Oscar, 1914 Uber den Einfluss des Blutserums des Normalen und des Alkaptonuri-

kers auf homogentisinsiiure. Biochem. Zeitschr. 61:165-170.
GusTAFSSON, Ake, 1938 Studies on the genetic basis of chlorophyll formation and the

mechanism of induced mutating. Hereditas 24:33-93.
, 1946 The effect of heterozygosity on variability and vigour. Hereditas 32:263-

286.
, 1947 The advantageous efifect of deleterious mutations. Hereditas 33:573-575.

Guthrie, R., 1949 Studies of a purine-requiring mutant strain of Escherichia colt. Jour.
Bact. 57:39-46.

Haldane, J. B. S., 1937 The efifect of variation on fitness. Amer. Nat. 71 :337-349.

, 1938 Essay "The Biochemistry of the Individual," in "Perspectives in Biochem-
istry." Cambridge Univ. Press: London.

, 1942 New paths in genetics. 206 pp. Harper and Brothers: New York.

, 1950 Equilibrium under natural selection (abstr.). Heredity 4:138-139.

Harland, S. C, 1920 A note on a peculiar type of "rogue" in Sea Island cotton. Agric.
News Barbados 19:29.

Harrington, J. B., 1944 Intra-varietal crossing in wheat. Jour. Amer. Soc. Agron. 36:
990-991.

Harvey, Paul H., 1939 Hereditary variation in plant nutrition. Genetics 2f :437-461.

Hatcher, E. S. J., 1939 Hybrid vigour in the tomato. Nature (London) 143:523.

, 1940 Studies in the inheritance of physiological characters. V. Hybrid vigour in

the tomato. Pt. III. A critical examination of the relation of embryo development to
the manifestation of hybrid vigour. Ann. Botanj' (New Series) 4:735-764.

Hayes, H. K., 1912 Correlation and inheritance in Nicotiana tabacum. Connecticut Agric.
Expt. Sta. Bull. 171. 45 pp.

, E. M. East, and E. G. Beinhart, 1913 Tobacco breeding in Connecticut. Con-
necticut Agric. E.xpt. Sta. Bull. 176. 68 pp.

, and R. J. Garber, 1919 Synthetic production of high protein corn in relation to

breeding. Jour. Amer. Soc. Agron. 11:309-319.
, and F. R. Immer, 1942 Methods of plant breeding. 432 pp. McGraw-Hill Book

Company, Inc.: New York and London.

-, and I. J. Johnson, 1939 The breeding of improved selfed lines of corn. Jour. Amer.

Soc. Agron. 31:710-724.
, E. H. Rinke, and Y. S. Tsiang, 1946 Experimental study of convergent improve-

ment and backcrossing in corn. Minnesota Agric. Expt. Sta. Tech. Bull. 172. 40 pp.
Hazel, L. N., 1943 Genetic basis for constructing selection indexes. Genetics 28:476-490.
, and W. F. Lamoreux, 1947 Heritability, maternal effects and nicking in relation

to sexual maturity and body weight in White Leghorns. Poultry Sci. 26:508-514.
-, A. L. MussoN, and J. L. Lush, 1948 Comparisons of inbred Poland China, Land-

race and purebred boars on Iowa farms. Jour. Animal Sci. 7:512-513.
Heidelberger, M., and O. T. Avery, 1923 The soluble specific substance of pneumococ-

cus. Jour. Exptl. Med. 38:73-79.
, , 1924 The soluble specific substance of pneumococcus. Second Paper. Jour.

Exptl. Med. 40:301-316.

-, E. A. Kabat, and M. Mayer, 1942 A further study of the cross reaction between

the specific polysaccharides of type III and VIII penumococci in horse antisera. Jour.
Exptl. Med. 75:35-47.

Heimsch, Charles, Glenn S. Rabideau, and VV. Gordon Whaley, 1950 Vascular de-
velopment and differentiation in two maize inbreds and their hj^brid. Amer. Jour. Bot.
37:84-93.

Henderson, C. R., 1948 Estimation of general, specific, and maternal combining abilities

524 BIBLIOGRAPHY

in crosses among inbred lines of swine. Unpublished Ph.D. thesis. Iowa State College

Library, Ames, Iowa. 199 pp. .
^ 1949 Estimation of general, specific, and maternal combmmg abihties m crosses

amonginbredlinesofswine. Jour. Animal Sci. 8:606.

^ 1950 Estimation of genetic parameters (abstr.). Ann. Math. Statistics 21 :309.

Herbe'rt, W., 1837 Amaryllidaceae. 428 pp. London. . ^ , , , . ,

Hestrin S and Carl C. Lindegren, 1950 Carbohydrases in Saccharomyces haploid

stocks of defined genotype. I. Fermentation and hydrolysis of alphaglucosides by yeast

6233. Arch. Biochem. 29:315-333. r-u j- a fV,..

Hetzer, H. O., W. V. Lambert, and J. H. Zeller, 1940 Influence of mbreedmg and other

factors on litter size in Chester White swine. U.S. Dept. Agric. Ore. 570 10 pp.
HiLDEBRAND, F., 1868 Einige Experimente und Beobachtungen, etc. Bot. Zeitschr. 26.

"^21— 328
HiNSHELWOOD, C. N., 1946 The chemical kinetics of the bacterial cell. 284 pp. Oxford

Univ. Press: London. . ^ > .•

HiORTH, G., 1940 Fine Serie multipler allele fiir Blutenzeichnungen bei Codetta amoena.

Hereditas 26:441-453. . . u j- • . T..nr Hpr^Hifv

Hodgson, R. E., 1935 An eight generation experiment in inbreeding swine. Jour. Heredity

HorrER^^G. N., 1926 Some differences in the functioning of selfed lines of corn under
varying nutritional conditions. Jour. Amer. Soc. Agron. 18:322-334.

HorEN^P. G., 1948 Corn breeding at the University of Illinois 1895 to 1900. u+lO pp.
Privately published by the author: Charlevoix, Michigan. , ,. . , ,

HOLLAENDER, A., 1948 Mechanism of radiation effects and the use of radiation o- Produc-
tion of mutations with improved fermentation. Ann. Missouri Bot. Card. 32 : 165-^8^

HOROWITZ, N. H., 1950 Biochemical genetics of Neurospora. Advances in Genetics. Vol.
111:33-71. Academic Press, Inc.: New York.

HOULAHAN MARY B., and H. K. Mitchell, 1948 Evidence for an interrelation in the
rtabolism of lysine, arginine. and pyrimidines in Neurospora. Proc. Nation. Acad. Sci.

HousER^T.!i9n'^Comparison of yields of first generation tobacco hybrids with those of
parentpiants. Amer. Breeders Rept. 7:155-167. i .„H Hwarf

Hsu, K. j!, 1950 Comparative studies in corn of the development of normal and dwarf
plants. Ph.D. thesis. University of Minnesota.

Hull, Fred H., 1945a Recurrent selection for specific combining ability in corn. Jour.

_^Tl9'45b t:rzeTeitts cooperation, News Letter 19:21-27. Dept. Plant Breeding,
Cornell Univ., Ithaca, New York.

1946a Regression analyses of corn yield data, (abstr.) Genetics 31 .219.

; 1946b Overdominance and corn breeding where hybrid seed is not feasible. Jour.

Amer. Soc. Agron. 38:1100-1103. ,. 98 Tnwa State
, 1952 Overdominance and recurrent selection. Heterosis, Chapter 28. Iowa State

College Press: Ames, Iowa. . . t_ i ^^^c

IMMER, F. R., 1941 Relation between yielding abiUty and homozygosis m barley crosses.

Tour. Amer. Soc. Agron. 33:200-206 , , ., ^ *v. -ja-

Irwin, M. R., 1939 A genetic analysis of species differences in Columbidae. Genetics 24.

^^^^^'947 Immunogenetics. Advances in Genetics Vol. 1:133-160. Academic Press,

— .^andl. j'coLK, 1936 Immunogenetic studies of species and of ^Pef^ hybrids from
the cross of Columha livia and Streptopdia risoria. Jour. Exptl. Zool. 73:3Uy 316.

BIBLIOGRAPHY 525

, and R. W. CuMLEY, 1945 Suggestive evidence for duplicate genes in a species
hybrid in doves. Genetics 30:363-375.

Jenkins, M. T., 1929 Correlation studies with inbred and crossbred strains of maize
Jour. Agric. Res. 39:677-721.

, 1935 The effect of inbreeding and of selection within inbred lines of maize upon
the hybrids made after successive generations of selling. Iowa State College Tour Sci
9:429-450. fe j • •

, 1940 The segregation of genes affecting yield of grain in maize. Jour. Amer Soc
Agron. 32:55-63.

Johnson, I. J., and H. K. Hayes, 1940 The value in hybrid combinations of inbred lines of
corn selected from single crosses by the pedigree method of breeding. Jour Amer Soc
Agron. 32:479-485.

Johnson, R. T., 1950 Combining ability in Zea mays as related to generations of testing,
selection of testers and characters of the inbred lines. Ph.D. thesis. University of Minne-
sota.

Johnson, S. W., 1891 How crops grow. Orange Judd & Co.: New York.
JoLLOS, Victor, 1934 Dauermodifikationen und Mutationen bei Protozoen Arch f
Protistenk. 83:197-219.

Jones, D. F., 1917 Dominance of linked factors as a means of accounting for heterosis
Genetics 2:466-479.

, 1918 The effects of inbreeding and crossbreeding upon development. Connecticut
Agric. Expt. Sta. Bull. 207. 100 pp

-, 1921 The indeterminate growth factor in tobacco and its effect upon develop-
ment. Genetics 6:433-444.

, 1944 Equilibrium in genie materials. Proc. Nation. Acad. Sci. U.S.A. 30:82-87.

, 1945 Heterosis resulting from degenerative changes. Genetics 30:527-542.

, 1950 The interrelation of plasmagenes and chromogenes in pollen production in
maize. Genetics 35:507-512.

, 1952 Plasmagenes and chromogenes in relation to heterosis. Heterosis, Chapter 14.
Iowa State College Press: Ames, Iowa.

, and P. C. M.^NGELSDORF, 1925 The improvement of naturally cross-pollinated

plants by selection in self-fertilized lines. Connecticut Agric. Expt. Sta. Bull. 266. 69 pp.

Kabat, E. a., 1949 Immunochemical studies on blood group substances. Bact Rev 13 •

189-202. ...

Kalmus, H., 1945 Adaptive and selective responses of a population of Drosophila melano-
gaster containing e and e+ to differences in temperature, humidity and to selection for
developmental speed. Jour. Genetics 47:58-63.

Karper, R. E., 1930 The effect of a single gene upon development in the heterozygote in
Sorghum. Jour. Heredity 21:187-192.

K.4UFFMANN, F., 1944 Die Bacteriologie der Salmonella Gruppe. 393 pp. Edwards: Ann
Arbor, Michigan.

Keeble, F., and C. Pellew, 1910 The mode of inheritance of stature and of time of flow-

ermg m peas {Pisum sativum). Jour. Genetics 1:47-56.
Kempton, J. H., and J. W. McLane, 1942 Hybrid vigor and weight of germs in the seeds

of maize. Jour. Agric. Res. 64:65-80.
Kidder, G. W., and V. C. Dewey, 1948 Studies on the biochemistry of Tetrahymena.

XIV. The activity of natural purines and pyrimidines. Proc. Nation. Acad. Sci. USA

34:566-574.

Kiesselbach, T. a., 1922 Corn investigations. Nebraska Agric. Expt. Sta. Res. Bull. 20.
151 pp.

526 BIBLIOGRAPHY

KiESSELBACH, T. A., 1926 The comparative water economy of selfed lines of corn and
their hybrids. Jour. Amer. Soc. Agron. 18:335-344.

^ 1930 The use of advanced generation hybrids as parents of double-cross seed corn.

Tour Amer Soc Agron. 22:614-626.

--— ,'l933 The possibilities of modern corn breeding. Proc. World's Grain Exhibition
and Conference 2:92-112.

KiKKAWA, H., 1941 Mechanism of pigment formation in Bombyx and Drosophila. Genet-
ics 26:587-607. , r * i

KiNMAN MURRAY L., and G. F. Sprague, 1945 Relation between number of parental
lines and theoretical performance of synthetic varieties of corn. Jour. Amer. Soc.

Agron. 37:341-351. , ^ j ^- ^c

KNIGHT, THOMAS ANDREW, 1799 An account of some experiments on the fecondation of

vegetables. Phil. Trans. Roy. Soc. London. Pp. 195-204. „ ^, ^ x^a

KOELREUTER, J. G., 1766 Vorlaufigen Nachricht von einigen das Geschlecht der Pflanzen

betreffenden Versuchen und Beobachtungen. 266 pp. Leipzig.
KOFT B W E Steers, and M.G.Sevag, 1950 Replacement by D-lysine of ^aminoben-

zo;c acid'ls a growth factor for Lactc^acUlus arabinosus 17-5. Arch. Biochem. 28:144-

KoInicke, Fr., 1876 IJber einige Erscheinungen in Okonomisch-botanischen Garten.

Bonn, Niederrhein. Gesell. Sitzgsb. Pp. 47-48. . u u -^ ^f A7,v-.

KOSMODEMJJANSKI, V. N., 1941 (Transgression in different generations of hyb"ds of ^..o-

tiana tabacum.) All Union Mikojan Research Institute of the Tobacco and Makhorka

Industry. Krasnodar 143:3-28. (In Russian.)
KOSTOFF D 1941 The problem of haploidy (Cytogenetic studies m Nicotiana haploids

and their bearing to some other cytogenetic problems). Bibliogr. Genetica 13:1-148.
KRZDER J L: B. W Fairbanks, W. E. Carro.., and E. Roberts, 1946 Effectiveness of

selecting ior rapid and for slow growth rate in Hampshire swine. Jour. Animal Sci.

Lampen ^J O , R. R. ROEPKE, and M. J. Jones, 1947 Studies on the sulfur metabolism
TE:cierickia coli. III. Mutant strains of E. coli unable to utilize sulfate for their com-
nlete sulfur requirements. Arch. Biochem. 13:55-66.

LANDSTEiNER, K., 1900 Zur Kenntnis der antifermentativen lytischen und agglutinieren-
den Wirkungen des Blutserums und der Lymphe. Zentralbl. ^akt 27 :357--362^

^ 1901 Ueber Agglutinationserscheinungen normalen menschlichen Blutes. Wien.

Klin Wochenschr. 14:1132-1134.

-,'l945 The specificity of serological reactions. Second Edition. 310 pp. Harvard

-^t^v'Z^^^^^r^r observations on individual differences of human
blood Proc. Soc. Exptl. Biol. New York 24:941-942. ,, ^ • .^ u.

and A S Wiener, 1940 An agglutinable factor in human blood recognized by

immune sera for Rhesus blood. Proc. Soc. Exptl. Biol. Ne^v York 43 : 223.

LARSON, R. E., and T. M. Currence, 1944 The extent of hybrid vigor in F. and F. gen-
erations of tomato crosses. Minnesota Agric. Expt. Sta. Tech. Bull. 64. 32 pp^

Lawrence, W. J. C, and J. R. Price, 1940 The genetics and chemistry of flo^^er colour

Lec^Th.: 184^ d! Ta' fL'ondatfon naturelle et artificielle de v.g.taux et de I'hybridation.

Led!L''e''rgT'i947 Gene recombination and linked segregations in Escherichia coli.

Genetics 32:505-525. ,• c . A^pr is-QQ-100

1949 Segregation in Escherichia coli. Records Genetics Soc. Amer. 18.99 100.

; and N. ZiNDER, 1948 Concentration of biochemical mutants of bacteria with peni-
cillin. Jour. Amer. Chem. Soc. 70:4267-4268.

BIBLIOGRAPHY 527

Leibowitz, J., and S. Hestrin, 1945 Alcoholic fermentation of the oligosaccharides. Ad-
vances Enzymol. 5:87-127.

Leng, E. R., C. M. VVoodworth, and R. J. Metzger, 1949 Estimates of hcritability and
degree of dominance in certain quantitative characters of corn and soybeans. Records
Genetics Soc. Amer. 18:101.

Levine, p., and E. M. Katzin, 1940 Isoimmunization in pregnancy and the varieties of
isoagglutinins observed. Proc. Soc. Exptl. Eiol. New York 45:343-346.

, and R. E. Stetson, 1939 An unusual case of intragroup agglutination. Jour. Amer.

Med. Assoc. 113:126-127.

Lewis, E. B., 1950 The phenomenon of position effect. Advances in Genetics Vol. Ill:
73-115. Academic Press, Inc.: New York.

L'Heritier, Ph., and G. Teissier, 1933 Elimination des formes mutantes dans les popu-
lations de Drosophiles. Cas des Drosophiles "ebony." Compt. Rend. Soc Biol 124-
882-885.

Lindegren, Carl C, 1949 The j-east cell, its genetics and cytology. 384 pp. Education-
al Pubhshers, Inc.: St. Louis, Missouri.

^> 1951 The mechanics of budding and copulation in Saccharomyces. Exptl Cell
Res. 2:305-311.

, and Gertrude Lindegren, 1947 Depletion mutation in Saccharomyces. Proc.

Nation. Acad. Sci. U.S.A. 33:314-318.

1951 Tetraploid Saccharomyces. Jour. Gen. Microbiol. 5:885-893.

— , and Margaret Rafalko, 1950 The structure of the nucleus of Saccharomyces

bayanus. Exptl. Cell Res. 1:169-187.
LiNDSTROM, E. VV., 1929 A haploid mutant in the tomato. Jour. Heredity 20:23-30.
LoH, S. Y., 1949 Early testing as a means of evaluating Fi heterosis between inbred lines of

Drosophila melanogastet . Ph.D. Thesis. Iowa State College Library, Ames, Iowa.
Lohmann, K., and P. Schuster, 1937 Uber die co-carboxylase. Naturwiss. 25:26-27.
LoNGLEY, A. E., 1938 Chromosomes of maize from North American Indians Jour '\gric

Res. 56:177-195.

, 1941 Knob positions on teosinte chromosomes. Jour. Agric. Res. 62:401-413.

, 1945 Abnormal segregation during megasporogenesis in maize. Genetics 30:100-

113.

LoNNQuiST, John, 1950 The effect of selection for combining ability within segregating

lines of corn. Agron. Jour. 42:503-508.
Lorain, John, 1825 Nature and reason harmonized in the practice of husbandry. 563 pp.

Carey and Lea: Philadelphia.
Luckwill, L. C, 1937 Studies in the inheritance of physiological characters. IV. Hybrid
vigour in the tomato. Pt. 2. Manifestations of hybrid vigour during the flowering period.
Ann. Botany (New Series) 1:379-408.

, 1939 Observations on heterosis in Lycopersicum. Jour. Genetics 37:421-440.
Lush, J. L., 1945 Animal breeding plans. Third Edition. 443 pp. The Collegiate Press,
Inc.: Ames, Iowa.

, 1948 The genetics of populations. Unpublished manuscript. Ames, Iowa.

, W. F. Lamoreux, and L. N. Hazel, 1948 The heritability of resistance to death

in the fowl. Poultry Sci. 27:375-388.

, and A. E. Molln, 1942 Litter size and weight as permanent characteristics of
sows. U.S. Dept. Agric. Tech. Bull. 836. 40 pp.

' P- S. Shearer, and C. C. Culbertson, 1939 Crossbreeding hogs for pork pro-
duction. Iowa Agric. Expt. Sta. Bull. 380. Pp. 83-116.
LwoFF, A., 1943 L'evolution physiologique, etude des pertes de fonctions chez les micro-
organismes. 308 pp. Hermann et cie.: Paris.

528 BIBLIOGRAPHY

Lyness, a. S., 1936 Varietal differences in the phosphorus feeding capacity of plants.

Plant Physiol. 11:665-688.
MacArthur, John W., 1949 Selection for small and large body size in the house mouse.

Genetics 34:194-209. • ^ ,.;„„

Mccarty, M., 1946 Chemical nature and biological specificity of the substance mducmg

transformation of pneumococcal types. Bact. Rev. 10:63-71.
McClintock, Barbara, 1931 Cytological observations of deficiencies involving known

genes, translocations and an inversion in Zea mays. Missouri Agric. Expt. Sta. Res. Bull.

' 19^8 ' The fusion of broken ends of sister half-chromatids following chromatid

breakage at meiotic anaphases. Missouri Agric. Expt. Sta. Res. Bull. 290. 48 pp.
McCluer G. W., 1892 Corn crossing. Illinois Agric. Expt. Sta. Bull. 2. Pp. «^-l"l-
Malinowski, E., 1928 A peculiar case of heterosis in Phaseolus vulgarzs. Zeitschr. indukt.

Abstamm.- u. VererbungsL, Supplementband 11:1090-1093.
Mangelsdorf, p. C, 1939 Use of multiple top-crosses in southern corn improvement.

Mimeo Rept. First Southern Corn Improvement Conf. 1:33-35. _
and James W. Cameron, 1942 Western Guatemala, a secondary center of origin

of cultivated maize varieties. Bot. Mus. Leafl., Harvard Univ. 10:217-252. _

and G. S. FRAPS, 1931 A direct quantitative relationship between vitamm A m

■ corn and the number of genes for yellow pigmentation. Science 73:241-242. _
^ and R. G. Reeves, 1939 The origin of Indian corn and its relatives. Texas Agric.

Expt Sta. Bull. 574. 315 pp.
_^ ^ 1945 The origin of maize, present status of the problem. Amer. Anthro-
pologist 47:235-243. , • •

^ and C. E. Smith, Jr., 1949 New archaeological evidence on evolution in maize.

Bot. Mus. Leafl., Harvard Univ. 13:213-247. • • a „ ^„

Marshak, a. G., 1936 Growth differences in reciprocal hybrids and cytoplasmic influence

on erowth in mice. Jour. Exptl. Zool. 72 :497-510.
MASiNG R A, 1938 Increased viability of heterozygotes for a lethal in i?....M^/a^e/a^^-

gastel. Compt. Rend. (Doklady) Acad. Sci. URSS 20:173-176.

■^ 1939a Different viability among flies of Drosophila melanogaster heterozygous for

leth'als Compt. Rend. (Doklady) Acad. Sci. URSS 23:835-838.
1939b Analysis of vitality of flies of Drosophila melanogaster heterozygous for

leth'als arisen in nature. Compt. Rend. (Doklady) Acad. Sci URSS 25:64-6 .
Mather, K., 1942 The balance of polygenic combinations. Jour. Genetics 43.309-336.

1943 Polygenic inheritance and natural selection. Biol. Rev. 18:^/ 04.

' 1949a The genetical theory of continuous variation. Proc. Eighth International

Congress of Genetics 376-401. Suppl. Vol. of Hereditas, Berlingska Boktryckeriet, Lund,

Sweden

, 1949b Biometrical genetics. 158 pp. Dover Pubhcations: New York.

Mendel, Gregor, 1865 Versuche uber Pflanzen-Hybriden. Naturf. Ver. in Brunn Verh.

Met!,' C.^W., 1938 Chromosome behavior, inheritance, and sex determination in Sciara.

Mi^nTELirP. 'l9S''?ber den Einfluss des Plasmons auf die Manifestation der Gene.

Zeitschr indukt. Abstamm.- u.Vererbungsl. 77:548-567.
-_-, 1940 tJber reziprok verschiedene Sippen-Bastarde bei EpiloUurr. hnsutum 78:

Moore', J. F., and T. M. Currence, 1950 Combining ability in tomatoes. Minnesota

Affric Expt. Sta. Tech. Bull. 188. 22 pp. r. ^- ,c.

MORGAN, D T., JR., 1950 A cytogenetic study of inversions in Zea n^ays. Genetics 35.

153-174.

BIBLIOGRAPHY 529

Morgan, T. H., C. B. Bridges, and A. H. Sturtevant, 1925 The genetics of Drosophila.

Bibliogr. Genetica 2:1-262.
Morrow, G. E., and F. D. Gardner, 1893 Field experiments with corn. lUinois Agric.

Expt. Sta. Bull. 2. Pp. 173-203.
, , 1893 Field experiments with corn. Illinois Agric. Expt. Sta. Bull. 2. Pp.

333-359.
MuLLER, H. J., 1950 Radiation damage to genetic material. Amer. Sci. 38:33-59.
MuNDKUR, B. D., 1949 Evidence excluding mutations, polysomy, and polyploidy as

possible causes of non-Mendelian segregations in Saccharomyces. Ann. Missouri Bot.

Card. 36:259-280.

, 1950 Irregular segregations in yeast hybrids. Current Sci. 19:84-85.

, and Carl C. Lindegren, 1949 An analysis of the phenomenon of long-term adap-
tation to galactose by Saccharomyces. Amer. Jour. Bot. 36:722-727.
MiJNTZiNG, A., 1930 Outlines to a genetic monograph of the genus Galeopsis with special

reference to the nature and inheritance of partial steriHty. Hereditas 13:185-341.
Murdoch, H. A., 1940 Hybrid vigor in maize embryos. Jour. Heredity 31:361-363.
Nabours, R. K., and L. L. Kingsley, 1934 The operation of a lethal factor in Apotettix

eurycephalits (Grouse locusts). Genetics 19:323-328.
Naudin, C, 1865 Nouvelles recherches sur I'hybridite dans les vegetaux. Nouv. Arch.

Mus. Hist. Nat. Paris 1:25-174.
Nawaschin, S., 1899 Neuen beobachtungen liber befruchtung bei Fritallaria tenella und

Lilium Martagon. Bot. Zentralbl. 77:62.
Neal, N. p., 1935 The decrease in yielding capacity in advanced generations of hybrid

corn. Jour. Amer. Soc. Agron. 27:666-670.
Odland, M. L., and C. J. Noll, 1948 Hybrid vigor and combining ability in eggplants.

Proc. Amer. Soc. Hort. Sci. 51:417-427.
Ostergren, G., and R. Prakken, 1946 Behavior on the spindle of the actively mobile

chromosome ends of rye. Hereditas 32:473-494.
Ovid (tr. by Brookes More), 1922 Metamorphoses. 38 pp. Cornhill Publishing Co.:

Boston.
Patel, M. S., p. H. Harvey, and W. C. Gregory, 1949 Hybrid tobacco?— Why not? Res.

and Farming, Spec. Tob. Issue, North Carolina Agric. Expt. Sta.: page 15.
Pauling, Linus, Harvey A. Itano, S. J. Singer, and Ibert C. Wells, 1949 Sickle cell

anemia, a molecular disease. Science 110:543-548.
Payne, K. T., and H. K. Hayes, 1949 A comparison of combining abihty in Fa and F3 lines

of corn. Agron. Jour. 41 :383-388.
Phillips, R. W., W. H. Black, Bradford Knapp, Jr., and R. T. Clark, 1942 Cross-
breeding for beef production. Jour. Animal Sci. 1 : 2 13-220.
Plough, H. H., H. Y. Miller, and M. E. Berry, 1951 Alternative amino acid require-
ments in auxotrophic mutants of 5. typhimurium. Proc. Nation. Acad. Sci. U.S.iV. 37:

640-644.

, H. N. Young, and M. Grimm, 1950 Penicillin screened auxotrophic mutations in

Salmonella typhimurium and their relation to X-ray dosage. Jour. Bact. 60:145-157.
Powers, LeRoy, 1936 The nature of the interaction of genes affecting four quantitative

characters in a cross between Hordeum deficiens and vulgare. Genetics 21:398-420.
, 1941 Inheritance of quantitative characters in crosses involving two species of

Lycopersicon. Jour. Agric. Res. 63:149-174.
, 1944 An expansion of Jones's theory for the explanation of heterosis. Amer. Nat.

78:275-280.

, 1945 Relative yields of inbred lines and Fi hybrids of tomato. Bot. Gaz. 106:247-

268.

530 BIBLIOGRAPHY

Powers, LeRoy, 1950a Determining scales and the use of transformations in studies on

weight per locule of tomato fruit. Biometrics 6:145-163.
, 1950b Gene analysis of weight per locule in tomato hybrids. Bot. Gaz. 112:163-

174.
— ■ , L. F. Locke, and J. C. Garrett, 1950 Partitioning method of genetic analysis

applied to quantitative characters of tomato crosses. U.S. Dept. Agric. Tech. Bull. 998.

56 pp.
, and Channing B. Lyon, 1941 Inheritance studies on duration of developmental

stages in crosses within the genus Lycopersicon. Jour. Agric. Res. 63:129-148.
Prakken, R., and A. Muntzing, 1942 A meiotic peculiarity in rye, simulating a terminal

centromere. Hereditas 28:441-482.
QuiNBY, J. R., and R. E. Karper, 1946 Heterosis in Sorghum resulting from the heterozy-
gous condition of a single gene that affects duration of growth. Amer. Jour. Bot. 33:

716-721.
Rabideau, Glenn S., W. Gordon Whaley, and Charles Heimsch, 1950 The absorption

and distribution of radioactive phosphorus in two maize inbreds and their hybrid.

Amer. Jour. Bot. 37:93-99.
Rafalko, Margaret, and Carl C. Lindegren, in press Cytological observations of

copulation and sporulation in Saccharomyces.
Ramiah, K., N. Parthasarathi, and S. Ramanujam, 1933 Haploid plant in rice (Oryza

saliva). Current Sci. 1:277-278.
Randolph, L. F., 1932a The chromosomes of haploid maize with special reference to the

double nature of the univalent chromosomes in the early meiotic prophase. Science 75:

566-567.
, 1932b Some effects of high temperature on polyploidy and other variations in

maize. Proc. Nation. Acad. Sci. U.S.A. 18:222-229.
, and H. E. Fischer, 1939 The occurrence of parthenogenetic diploids in tetraploid

maize. Proc. Nation. Acad. Sci. U.S.A. 25:161-164.
Rasmusson, J., 1927 Genetically changed linkage values in Pisum. Hereditas 10:1-150.
Raut, Caroline, 1950 Effect of concentration of pantothenate on selection of a mutant

for pantothenate synthesis in Saccharomyces cerevisiae. Genetics 35:381-395.
Reed, Ollie E., 1946 Is the crossbred dairy cow on the way? Country Gentlemen, June,

116:15, 61-64.
Reeves, R. E., and \V. F. Goebel, 1941 Chemoimmunological studies on the soluble

specific substance of pneumococcus. V. The structure of type III polysaccharide. Jour.

Biol. Chem. 139:511-519.
Reeves, R. G., 1944 Chromosome knobs in relation to the origin of maize. Genetics 29:

141-147.
, 1950 The use of teosinte in the improvement of corn inbreds. Agron. Jour. 42:

248-251.
Rhoades, M. M., 1942 Preferential segregation in maize. Genetics 27:395-407.

, 1950 Meiosis in maize. Jour. Heredity. 41:59-67.

■ , and Hilda Vilkomerson, 1942 On the anaphase movement of chromosomes. Proc.

Nation. Acad. Sci. U.S.A. 28:433-436.
Richey, F. D., 1945a Bruce's explanation of hybrid vigor. Jour. Heredity 36:243-244.
, 1945b Isolating better foundation inbreds for use in corn hybrids. Genetics 30:

455-471.

, 1946 Hybrid vigor and corn breeding. Jour. Amer. Soc. Agron. 38:833-841.

, 1947 Corn breeding: gamete selection, the Oenothera method, and related miscel-

lany. Jour. Amer. Soc. Agron. 39:403-412.
, and G. F. Sprague, 1931 Experiments on hybrid vigor and convergent improve-

ment in corn. U.S. Dept. Agric. Tech. Bull. 267. 22 pp.

BIBLIOGRAPHY 531

RoBBiNS, William J., 1940 Growth substances in a hybrid corn and its parents. Bull.

Torrey Bot. Club 67:565-574.

, 1941a Factor Z in hybrid maize. Bull. Torrey Bot. Club 68:222-228.

■ — , 1941b Growth of excised roots and heterosis in tomato. Amer. Jour. Bot. 28:216-

225.
, and Roberta Ma, 1942 Vitamin deficiencies of Ceratostomella and related fungi.

Amer. Jour. Bot. 29:835-843.
— ■ — , 1946 A report on the growth of excised tomato roots. Jour. Arnold Arboretum 27 :

480-485.

, 1952 Hybrid nutritional requirements. Heterosis, Chapter 7. Iowa State College

Press: Ames.

Roberts, H. F., 1929 Plant hybridization before Mendel, vii+374 pp. Princeton Uni-
versity Press: Princeton.

Robertson, D. W., 1932 The effect of a lethal in the heterozygous condition on barley
development. Colorado Agric. Expt. Sta. Tech. Bull. 1. 12 pp.

, and W. W. Austin, 1935 The effect of one and of two seedling lethals in the

heterozygous condition on barley development. Jour. Agric. Res. 51:435-440.

Robinson, H. F., R. E. Comstock, and P. H. Harvey, 1949 Estimates of heritability and
the degree of dominance in corn. Agron. Jour. 41 :353-359.

Rogers, J. S., 1950 The inheritance of inflorescence characters in maize-teosinte hybrids.
Genetics 35:541-558.

Roman, H., D. C. Hawthorne, and H. C. Douglas, 1951 Polyploidy in yeast and its
bearing on the occurrence of irregular genetic ratios. Proc. Nation. Acad. Sci. U.S.A.
37:79-84.

Ryan, F. J., G. W. Beadle, and E. L. Tatum, 1943 The tube method of measuring the
growth rate of Neurospora. Amer. Jour. Bot. 30:784-799.

, and J. Lederberg, 1946 Reverse mutation and adaptation in leucineless Neuro-
spora. Proc. Nation. Acad. Sci. U.S.A. 32:163-173.

Sag.aret, a., 1826 Considerations sur la production des hybrides. Ann. d. Sci. Nat. 8:
294-314.

Sanborn, J. W., 1890 Indian corn. Rept. Maine Dept. Agric. 33:54-121.

Satterthwaite, F. E., 1946 An approximate distribution of estimates of variance com-
ponents. Biometrics Bull. 2:110-114.

Schrader, F., 1931 The chromosome cycle of Protortonia primiiiva (Coccidae) and a con-
sideration of the meiotic division apparatus in the male. Zeitschr. Wiss. Zool. 138:386-
408.

Shamel, a. D., 1905 The effect of inbreeding in plants. U.S. Dept. Agric. Yearbook for
1905. Pp. 377-392.

Shen, San-Chiun, 1950 The genetics and biochemistry of the cysteine-tyrosine relation-
ship in Neurospora. Ph.D. Thesis California Inst, of Tech., Pasadena.

Shifriss, 0., 1947 Developmental reversal of dominance in Cucurhita pepo. Proc. Amer.
Soc. Hort. Sci. 50:330-346.

Shull, a. F., 1912 The influence of inbreeding on vigor in Hydatina scuta. Biol. Bull.
24:1-13.

Shull, G. H., 1908 The composition of a field of maize. Rept. Amer. Breeders' Assoc. 4:
296-301.

, 1909 A pure-line method in corn breeding. Rept. Amer. Breeders' Assoc. 5:51-59.

, 1910 Hj'bridization methods in corn breeding. Amer. Breeders' Mag. 1 :98-107.

, 1911a Experiments with maize. Bot. Gaz. 52:480-485.

, 1911b The genotypes of maize. Amer. Nat. 45:234-252.

, 1914 Duplicate genes for cajjsulc form in Bursa bursa-pastoris. Zeitschr. indukt.

Abstamm.- u. Vererbungsl. 12:97-149.

532 BIBLIOGRAPHY

Shull, G. H., 1921 Estimating the number of genetic factors concerned in blending in-
heritance. With rejoinder by VV. E. Castle. Amer. Nat. 55:556-571.

, 1922 tjber die Heterozygotie mit Riicksicht auf den praktischen Ziichtungserfolg.

Beitrage z. Pflanzenzucht 5:134-158.

, 1946 Hybrid seed corn. Science 103:547-550.

, 1948 What is "heterosis"? Genetics 33:439-446.

SiERK, C. F., 1948 A study of heterosis in swine, (abstr.) Jour. Animal Sci. 7:515.

— , 1948 A study of heterosis in swine. Ph.D. Thesis, Univ. of Minnesota Library,

MinneapoHs.

Simpson, Q. L, 1907 Rejuvenation by hybridization. Rept. Amer. Breeders' Assoc. 3:
76-81.

Singleton, W. Ralph, 1935 Early research in maize genetics. Jour. Heredity 26:49-59;
121-126.

, 1941 Hybrid vigor and its utilization in sweet corn breeding. Amer. Nat. 75:

48-60.

, 1943 Breeding behavior of C30, a diminutive P39 mutant whose hybrids show in-
creased vigor. (Abstract) Genetics 28:89.

■ , and O. E. Nelson, Jr., 1945 The improvement of naturally cross-pollinated

plants by selection in self-fertilized lines. IV. Combining abihty of successive generations
of inbred sweet corn. Connecticut Agric. Expt. Sta. Bull. 490. Pp. 458-498.

Smith, H. H., and C. W. Bacon, 1941 Increased size and nicotine production in selections
from intraspecific hybrids of Nicotiana rustica. Jour. Agric. Res. 63:457-467.

Smith, S. N., 1934 Response of inbred lines and crosses in maize to variations of nitrogen
and phosphorus supplied as nutrients. Jour. Amer. Soc. Agron. 26:785-804.

Sprague, G. F., 1936 Hybrid vigor and growth rates in a maize cross and its reciprocal.
Jour. Agric. Res. 53:819-830.

, 1946a Early testing of inbred lines of corn. Jour. Amer. Soc. Agron. 38:108-117.

— , 1946b The experimental basis for hybrid maize. Biol. Rev. Cambridge Phil. Soc.

21:101-120.

, 1942-1949 Annual reports of corn breeding investigations conducted by the Divi-
sion of Cereal Crops and Diseases. Bur. Plant Indust. S.A.E. and the Iowa Agric.
Expt. Sta.

, and B. Brimhall, 1950 Relative effectiveness of two systems of selection for oil

content of the corn kernel. Agron. Jour. 42:83-88.

and M. T. Jenkins, 1943 A comparison of synthetic varieties, multiple crosses,

and double crosses in corn. Jour. Amer. Soc. Agron. 35:137-147.
, and L. A. Tatum, 1942 General vs. specific combining ability in single crosses of

corn. Jour. Amer. Soc. Agron. 34:923-932.
Sprengel, C. K., 1793 Das entdeckte Geheimniss der Natur im Bau und in der Befruch-

tung der Blumen. 433 pp. Berlin.
Srb, a. M., and N. H. Horowitz, 1944 The ornithine cycle in Neurospora and its genetic

control. Jour. Biol. Chem. 154:129-139.
Stadler, L. J., 1933 On the genetic nature of induced mutations in plants. Missouri Agric.

Expt. Sta. Res. Bull. 204. 29 pp.
, 1939 Some observations on gene variability and spontaneous mutation. Spragg

Memorial Lectures, Michigan State College. Pp. 1-15.
— , 1944 Gamete selection in corn breeding, (abstr.) Jour. Amer. Soc. Agron. 36:

988-989.
, 1945 Gamete selection in corn breeding. Maize Genetics Cooperation News Letter.

19:33-40.
Standley, Paul C, 1950 Teosinte in Honduras. Ceiba 1:58-61.
Stephens, S. G., 1950a Factors affecting the genetic complexity of a partially outcrossed

population. Proc. Second Cotton Improvement Conference, Biloxi, Mississippi.

BIBLIOGRAPHY 533

, 1950b The internal mechanism of speciation in Gossypium. Bot. Rev. 16 : 1 15-149.

Stern, C, 1948 Negative heterosis and decreased effectiveness of alleles in heterozygotes.
Genetics 33:215-219.

, and Edward Novitski, 1948 The viability of individuals heterozj-gous for re-
cessive lethals. Science 108:538-539.

Stewart, H. A., 1945 An appraisal of factors afifecting prolificacy in swine. Jour. Animal
Sci. 4:250-260.

Stormont, Clyde, 1950 Additional gene controlled antigenic factors in the bovine ery-
throcyte. Genetics 35 : 76-94.

, R. D. Owen, and M. R. IRW^N, 1948 Gene action on cellular characters in cattle

(abstr.) Genetics 33:126.

J , , 1951 The B and C systems of bovine blood groups. Genetics

36:134-161.
Str.'^ndskov, H. H., 1948 Blood group nomenclature. Jour. Heredity 39: 108-112.
, 1949 Recent views on the genetics of the Rh-Hr blood factors. Bull. New York

Acad. Med. 25:249-255.
Straus, F. S., 1942 The genetic mechanism of heterosis as demonstrated by egg produc-
tion in Drosophila melanogaster . Ph.D. Thesis, Iowa State College Library, Ames, Iowa.
, and J. W. Gowen, 1943 Heterosis: its mechanism in terms of chromosome units in

egg production of Drosophila melanogaster . (abstr.) Genetics 28:93.
Str.\uss, B. S., 1950 Studies on vitamin Be requiring mutants of Neurospora crassa.

Thesis, California Inst, of Tech., Pasadena.
Strehler, B. L., 1950 The replacement of para-aminobenzoic acid by methionine in the

growth of a Neurospora mutant. Jour. Bact. 5Q: 105-1 11.
Stringfield, G. H., 1950 Heterozygosis and hybrid vigor in maize. Agron. Jour. 42:145-

152.
Stubbe, H., and K. Pirschle, 1940 tJber einen monogen bedingten Fall von Heterosis bei

Antirrhinum majus. Ber. Deut. Bot. Ges. 58:546-558.
Sturtevant, a. H., 1936 Preferential segregation in triplo-IV females of Drosophila

melanogaster . Genetics 21 : 444-466.
, and G. W. Beadle, 1936 The relations of inversions in the X chromosome of

Drosophila melanogaster to crossing over and disjunction. Genetics 21 : 554-604.
Tabernaemontanus (Jakob Dietrich [Theodorus] of Bergzabern), 1588. Neuw

Kreuterbuch, etc., Franckfurt am Mayn.
Tan, C. C, 1946 Mosaic dominance in the inheritance of color patterns in the lady-bird

beetle, Harmonia axyridis. Genetics 31:195-210.
Tatum, E. L., 1946 Induced biochemical mutations in bacteria. Cold Spring Harbor

Symposia Quant. Biol. 11:278-283.
, and J. Lederberg, 1947 Gene recombination in the bacterium E. coli. Jour. Bact.

53:673-684.
Teas, H. J., 1950 Mutants of Bacillus subtilis that require threonine or threonine plus

methionine. Jour. Bact. 59:93-104.
, N. H. Horowitz, and M. Fling, 1948 Homoserine as a precursor of threonine and

methionine in Neurospora. Jour. Biol. Chem. 172:651-658.
Teissier, G., 1942a Persistance d'un gene lethal dans une population de Drosophiles.

Compt. Rend. Acad. Sci. (Paris). 214:327-330.
, 1942b Vitalite et fecondite relatives de diverses combinations genetiques compor-

tant un gene lethal chez la Drosophile. Compt. Rend. Acad. Sci. (Paris) 214:241-244.
-, 1944 Equilibre des genes lethaux dans les populations stationnaircs panmictiques.

Rev. Scient. 82:145-159.

-, 1947a Variation de la frequence du gene sepia dans une population stationnaire

de Drosophiles. Compt. Rend. Acad. Sci. (Paris) 224:676-677.

534 BIBLIOGRAPHY

Teissier, G., 1947b Variation de la frequence du gene ebony dans une population sta-

tionnaire de Drosophiles. Compt. Rend. Acad. Sci. (Paris) 224:1788-1789.
Thom, C, and R. A. Steinberg, 1939 The chemical induction of genetic changes in fungi.

Proc. Nation. Acad. Sci. U.S.A. 25:329-335.
Thomsen, O., 1936 Untersuchungen liber erbliche Blutgruppenantigene bei Hiihnern, II.

Hereditas 22:129-144.
Timofeeff-Ressovsky, N. W., 1940 Allgemeine Erscheinungen der Gen-Manifestierung.

Handbuch d. Erbbiologie des Menschen, Berlin.
Torssell, Robert, 1948 Different methods in the breeding of lucerne. In Svalof 1886-

1946. Pp. 237-248. Carl Bloms Boktryckeri, Lund, Sweden.
Tylor, E. B., 1865 Researches into the early history of mankind and development of

civ'ilization. 388 pp. London.
TYSD.4L, H. M., and B. H. Crand.all, 1948 The polycross progeny performance as an in-
dex of the combining ability of alfalfa clones. Jour. Amer. Soc. Agron. 40:293-306.
■ , T. A. Kiesselbach, and H. L. Westover, 1942 Alfalfa breeding. Nebraska Agric.

Expt. Sta. Res. Bull. 124. 46 pp.
U.S. Commissioner of Patents, 1851 Report for the year 1850, Part II, Agriculture,

Washington.
VAN Overbeek, J., M. E. CoNKLiN, and A. F. Blakeslee, 1942 Cultivation in vitro of

small Datura embryos. Amer. Jour. Bot. 29:472-477.
Veatch, C, 1930 Vigor in soybeans as affected bj' h}bridity. Jour. Amer. Soc. Agron. 22 :

289-310.
Vennesland, B., and R. Z. Felsher, 1946 Oxalacetic and pyruvic carboxylases in some

dicotj-ledonous plants. Arch. Biochem. 11:279-306.
ViLMORiN, H., 1867 Sur la fecondation du Mais. Bull. Soc. Bot. France 14:246-249.
Warren, D. C, 1924 Inheritance of egg size in Drosophila melanogaster . Genetics 9 :41-69.
Weaver, Harry Lloyd, 1946 A developmental study of maize with particular reference

to hybrid vigor. Amer. Jour. Bot. 33:615-624.
Weatherwax, Paul, 1950 The history of corn. Sci. Month. 71 : 50-60.
Webber, H. J., 1900 Xenia or the immediate effect of pollen in maize. U.S. Dept. Agric.

Div. \'eg. Phys. Bull. 22.

, 1901 Loss of vigor in corn from inbreeding, (abstr.) Science 13:257-258.

Weidenhagen, R., 1940 Handbuch der Enzymologie. Leipzig.

Weiss, M. G., C. R. Weber, and R. R. Kalton, 1947 Early generation testing in soy-
beans. Jour. Amer. Soc. Agron. 39:791-811.
Wellhausen, E. J., 1947 Comparacion de variedades del maiz obtenidas en el Bajio,

Jalisco y en la Mesa Central. Folleto Tecnico No. 1, Oficina de Estudios Especiales,

S.A.G., Mexico, D.F.
, and L. M. Roberts, 1948 Rocamex V-7, una variedad sobresaliente de maiz para

sembrarse de riego en la Mesa Central. Folleto de Divulgacion No. 3, Oticina de Estudios

Especiales, S.A.G., Mexico, D.F.
, , 1949 Methods used and results obtained in corn improvement in Mexico.

Iowa Agric. Expt. Sta. Res. Bull. 371:525-537.
, , E. Hernandez X., en colaboracion con P. C. Mangelsdorf, 1951 Razas

de Maiz en Mexico, su origen, caracteristicas y distribucion. Folleto Tecnico No. 5,

Oficina de Estudios Especiales, S.A.G., Mexico, D.F.
Whaley, W. Gordon, 1944 Heterosis. Bot. Rev. 10:461-498.

, 1950 The growth of inbred and hybrid maize. Growth 14:123-155.

, Charles Heimsch, and Glenn S. Rabideau, 1950 The growth and morphology

of two maize inbreds and their h\'brid. Amer. Jour. Bot. 37:77-84.
, and Alice L. Long, 1944 The behavior of excised roots of heterotic hvbrids and

their inbred parents in culture. Bull. Torrey Bot. Club. 71:267-275.

BIBLIOGRAPHY 535

Whatley, J. A., Jr., m42 Influence of heredity and other factors on 180-day weight in

Poland China swine. Jour. Agric. Res. 65:24^-264.
Wheld.\le, M., 1910 On the formation of anthocyanin. Jour. Genetics 1:1,^,?-158.
Whitaker, T. \V., and G. W. Bohn, 1950 The taxonomy, genetics, production and uses

of the cultivated species of Cucurbita. Econ. Bot. 4:52-81.
White, P. B., 1929 The Salmonella group. In a system of bacteriology. Med. Res. Council

(Great Brit.) 4:86-158.
VViEGMANN, A. F., 1828 tJber die Bastarderzeugung im Pflanzenreiche. Braunschweig.

40 pp.
Wiener, A. S., 1943 Blood groups and transfusion. Third Ed. 438 pp. Charles C.

Thomas: Springfield.
, and H. R. Peters, 1940 Hemolytic reactions following transfusions of blood of

the homologous group, with three cases in which the same agglutinogen was responsible.

Ann. Internal Med. 13:2306-2322.
WiGAN, L. G., 1944 Balance and potence in natural populations. Jour. Genetics 46:150-

160.
Wince, 0., 1935 On haplophase and diplophase in some Saccharomycetes. Compt. Rend.

d. Lab. Carlsberg Ser. Physiol. 21:77-112.
, and C. Roberts, 1948 Inheritance of enzymatic characters in yeasts and the

phenomenon of long-term adaptation. Compt. Rend. d. Lab. Carlsberg Ser. Physiol.

24:263-315.
Winkler, H., 1930 Die Konversion der Gene. 186 pp. Jena.
Winters, L. M., D. L. Dailey, P. S. Jordan, O. M. Kiser, R. E. Hodgson, J. N. Cum-

MiNGS, and C. F. Sierk, 1948 Experiments with inbreeding swine. Minnesota Agric.

Expt. Sta. Bull. 400. 28 pp.
■ , P. S. Jordan, and R. E. Hodgson, 1944 Preliminary report on crossing of inbred

hues of swine. Jour. Animal Sci. 3:371.
, O. M. Kiser, P. S. Jordan, and W. H. Peters, 1935 A six years' study of cross-

breeding swine. Minnesota Agric. Expt. Sta. Bull. 320. 18 pp.
, , , , 1936 Crossbred swine for greater profits. Minnesota Ext.

Div. Spec. Bull. 180. 12 pp.

Wright, Sewall, 1922a Coefficients of inbreeding and relationship. Amer. Nat. 56:330-
338.

, 1922b The effects of inbreeding and crossbreeding on guinea pigs. I. Decline in

vigor. II. Differentiation among inbred families. U.S. Dept. Agric. Tech. Bull. 1090.

, 1922c The effects of inbreeding and crossbreeding on guinea pigs. III. Crosses be-
tween highly inbred families. U.S. Dept. Agric. Tech. Bull. 1121. 60 pp.

, 1931a Inbreeding and recombination. Proc. Nation. Acad. Sci. U.S..\. 19:420-

433.

, 1931b Evolution in Mendelian populations. Genetics 16:97-159.

, 1935 The analysis of variance and the correlations between relatives with respect

to deviations from an optimum. Jour. Genetics 30:243-256.
Zalokar, Marko, 1948 The p-aminobenzoic acid requirement of the "sulfonamide-re-

quiring" mutant strain of Neurospora. Proc. Nation. Acad. Sci. U.S. .A. 34:32-36.
, 1950 The sulfonamide-requiring mutant of Neurospora: Threonine-methionine

antagonism. Jour. Bact. 60:191-203.
Zelle, M. R., 1942 Genetic constitutions of host and pathogen in mouse tj^phoid. Jour.

Infect. Dis. 71:131-152.
Zimmerman, P. W., and A. E. Hitchcock, 1949 Formative effects of several substituted

phenoxy acids applied to Kalanchoe. Contrib. Boyce Thompson Inst. 15:421-427.
Zirkle, Conway, 1935 The beginnings of plant hybridization. 231 pp. University of

Pennsylvania Press: Philadelphia.

Index

Acentric chromatid, 78

Acentric fragment, 78

Adaptability, hybrids, 99

Adaptedness, upper attainable level, 219

Adaptive evolutionary phenomenon, 173

Additive gene effects, 168

Additive genetic variance, 511

Additivity of heterotic gene effects, 483

Adenine relation to histidine synthesis, 278

Adenine requirer, 278

Agglutination titer, 280

Albinism, 237

Alcaptonuria, 237

Alcohol dehydrogenase, 110

Alfalfa cross- and self-fertilization, 81

Alleles, 481

divergent, 173,458

fortuitous, 322

interaction between, 229-30, 234

of intermediate productivity, 294

less favorable, role of, 323-24

pseudo and multiple, 254

relic, 322
Allelic genes, 236, 240
Allelism, multiple, 500
Aluminum in hybrids, 106
American Corn Belt dents, heterosis of,

138-46
Anaphase, 75

Anaphase I of cell illustrated, 67
Anaphase I dyads, 73
Anaphase II, 76, 77, 79
Anaphase II of cell illustrated, 67
Anaphase II monads, 73
Angiosperm endosperm, 83
Angiosperm ovule, 83

Angiosperms

female gametophyte in, 83

fertilization, 83

mature ovules, 83

secondary fertilization, 84

seed, 82-85

seed coat, 83

seed development, 81

Animal inbreeding, 5-7
Anthocyanin, 237-38

Antibiotic substances, 117

Antigenic characters, 239, 240

Antigenic complexes, 253

Antigenic relationships, 244

Antigenic specificity, alteration of, 278-80

Antigenic variants induced by radiation,

279
Antigens, 240, 242, 268

in cattle, 252-54

of pneumococci, 254

Artificial hybrid, 424

Artificial selection, 219-20

Asexual reproduction, 47

Autogenic growth inhibitors, 117

Autonomous apomicts, 89

Autonomous organelles of the yeast cell,

261-62
Auxins, 112

Auxotrophic mutants

biochemical studies, 275-78

Edwards technique of growing, 279

frequencies in S. typhimurium, 271

kinds in S. typhimurmrn, 272

methods of inducing, 269

from radiated lines, 269-70

of S. typhimurium, 281
Auxotrophic organisms, 269

Auxotrophs

determination of particular nutrilite re-
quired, 270
utilization of sulphur compounds, 276
Average dominance, estimate of, 497-99
Average dominance of genes, estimation of,
494^516

Average hybrid, 291

Backcrosses, 423
Backcrossing, 231, 379, 421

in maize, 235
Bacterial genetics, 267
Bacterial mutation, 267
Balanced defective, 324
Balanced euheterosis, 221-22

Balanced lethal genes
in Capsella grandiflora, 46
in Oenotheras, 46

Balanced polymorphism, 221

537

538

INDEX

Barren stalks, 16

Biochemical basis for sulfonamide-requiring
character, 207-10

Biochemical defects as gene markers, 256
Biochemical deficiencies, 257
Biochemical genetics, 256

Biochemical models of heterosis in Neuro-

spora, 199-217
Biochemical reactions essential to growth

in Neurospora, 208-9

Biochemical studies of auxotrophic mu-
tants, 275-78

Biotypes, 20, 24
variation in, 26

Bottleneck genes, 325, 327, 424, 428

Bottleneck locus, 325

Bottlenecks

environmental components of, 329
genetic-environmental, 325
physiological, 325, 326

"Breaking of the types," 15

Breeding, practical necessities, 140

Breeding plans
reciprocal selection between two cross-

breds, 457
selection in crossbred to homozygous

tester, 457

Breeding plot efficiency, 146

Breeding procedures, improvement limits,
416

Breeding records

of human beings, 2

of primitive deities, 2
Breeding stations in Mexico, 426
Breeding systems, 414
Budding, 262-64

Carboxylase, 110
Catalase activity, 108
Cattle cells, 252

Causative genes

action of, 255

direct effect of, 239

Cell membrane and cell wall, 262

Cellular antigens in humans, 240-43

Cellular characters within a species, 251-55

Centrochromatin, 262

Centromere region, 72, 74

Centrosome, 261

Character complexes, importance in maize
breeding, 131
extrapolated correlates, 131

Chemical analyses of genetic variations in
flower color, 237

Chemistry of A and B substances, 241

Chiasmata formation among three chromo-
somes, 69

Chlorophenoxyacetic acid, 117
ortho, para, and meta forms, 117

Chlorophyll production in maize, 228-29

Choline, 214

Choosing testers, 449

Chromatin bridge, 78

Chromogenes and plasmagenes in heterosis,
224-35

Chromomere pattern, 68

Chromonemata, 74

Chromosomal and cytoplasmic basis for hy-
brid vigor, 479-81

Chromosomal deletions, 226

Chromosomal effects, variance analysis of,
483

Chromosomal fibers, 72, 74
formation, 75

Chromosomal inheritance, 258-59

Chromosomal rearrangements, 227

Chromosome doubling, 390-92

Chromosome knob number, 419

Chromosome knobs
distribution in Northern Flints, Southern

Dents, and Corn Belt inbreds, 130
significance, 185-86

Chromosome length, effect on hybrid vigor,

484
Chromosome maps of Saccharomyces, 258
Chromosome mechanisms, 153
Chromosome movement, 74

Chromosome 10, 67, 69
abnormal, 68, 72
kinds, 66

Chromosomes

block transfer, 187
breakage, 74

of maize and teosinte, 183
and nuclear membrane, 262

Co-carboxylase, 110

Colchicine, 397

Combining ability, 62, 141, 148, 330-51,
386,399,408,409, 431,485
general, 328, 364

general and specific, 352-70, 485-90
and morpholog}', correlation of, 143-45
and morpholog>', experimental results,

143-45
specific, 352-53
testing for, 444
tests for, 449

Commercial corn growing areas in Mexico,

426
Commercial corn hybrids, 232

Commercial corn production, 373

INDEX

539

Commercial hybrids, 291
origin of, 401

Complementary genes, 48

Compound genes and genes with multiple
effects, 227-29

Conidia, formation, 201

Conidiophore, 201

Constant parent regression, 287

Controlled cross-pollination in breeding

corn, 16
Controlled parentage, 16
Controlled self-pollination, 16
Convergent improvement, 328
Corn {See also Maize)

grown in prehistoric times, 418
improvement through breeding, 425-48
inbred lines, 280
selfing and crossing results, 36
Corn Belt dents
association of characters in
inbreds, 131

open-pollinated varieties, 131
derived from hybridization between
Southern Dents and Northern Flints,
132
origin of, 124-48
historical and archaeological evidence,
148
taxonomically important differences, 132
width of cross, 132-37
Corn Belt inbreds, 147
characteristics, 130

distribution of chromosome knobs, 130
Corn Belt maize
archaeological record, 127
Caribbean influence, 127
cytology, 128-29
description, 125
historical record, 127
origin, 127

genetic evidence, 129-32
and genetic significance, 124-48
understanding variability in, 148
variation, 146

Corn belt of Mexico, 436

Corn breeding
history, 400-406
pure line method, 28
steps, 470

Corn breeding methods

ear-to-row, 400

mass selection, 400

selection within and among inbred lines,
401-2

varietal hybridization, 400
Corn evolution in Mexico, 419
Corn hybrids, commercial, 232
Corn improvement program of Mexico, 425
Corn production, pure line method, 44
Corn races in Mexico, 427

Corn and swine inbreeding
degree of heterozygosity, 338
effectiveness of continued phenotypic se-
lection, 338

Corn, white dent, 21

Cousin crosses, 31

Cozymase, 110

Cross-fertilization, 15, 16, 17, 45
in alfalfa, 81
in maize, 49

Cross-fertilization versus self-fertilization,
20

Cross performance, 347
improvement, 351
initial, 343, 351
limits, 351

Cross-pollinated and self-pollinated plants,
comparison of methods, 55

Cross-pollination, 15

Crossbred performance, 402

Crossbreeding, 171-73

early ideas on, 1-13

effect on seed collapse in Medicago saliva,
85-89

effects on growth, 81

and inbreeding effects, 331-32, 350

livestock, 372

in seed development, 81-97

USDA experiments, 375

Universitv of Minnesota experiments,
371-76"

Crossbreeding, rotational

advocation, 375

farm applications, 377

and heterosis, 371-77

inbred lines, 376-77

for swine, 375

versus inbreeding, 7
Crossbreeding swine, 371-76

University of Minnesota experimental
results, 373-74

Crossing, 12, 139, 382
advantageous effects, 23

Crossing inbred lines of tomatoes, 307-10
Crossing over, 68, 222

in maize X teosinte, 187
Cysteine requirers, 276

Cytology and genetics of Saccharomyces,
256-66

Cytoplasm, 261

and genes, interaction, 232-33, 479
Cytoplasmic inheritance, 266

Dauermodifikation, 260-61

Decarboxylating enzyme systems of the
respirator}- mechanism, 110

Dehydrogenase enzymes, 110

Deleterious dominant, 323

Deleterious recessives, 220, 323

540

INDEX

Depletion mutation, 261

Detasseling, 15

Deterioration, 16

De Vries' mutation theory, 20

Diakinesis, 69

in heterozygous plants, 68

Dianthus hybrid, 9

Dicentric chromatid, 77, 78

Differentiation, 260

Diploid behavior at meiosis, 500

Diploid heterozygote, 202

Diploids, 389

Direct tetrad analysis, 258-59

Disjoining monads, types, 76

Dispersed heterosis, 127

Divergent alleles, 173, 458

cumulative action of, 282

interaction between, 234

Divergent spindles, 74

Domestic f ow 1 hybrids, 245-46

Dominance, 11, 100, lOl, 167, 284, 332,
338, 350, 353
to account for heterosis, 350
analvsis of, 495
complete, 288, 416, 497
degree of, 336, 494
as explanation of hybrid vigor, 65
of growth factors, 60
and heterosis, 307, 494-516
incomplete, 416
by interference, 458
in maize X teosinte, 188-92
andover dominance, 282-97
partial, 65, 497
phenotypic and genie, 318
and scales of measurement, 306, 313
in self-pollinated plants, 166
of tomato hybrids, 308

Dominance of genes, estimation of, 494-516

Dominance and heterosis as expression of
same phvsiological genetic phenomena,
307, 309

Dominance hypothesis, 230, 284
objections to, 285

Dominance of linked genes hypothesis, 282

Dominant alleles, 101

Dominant, deleterious, 323

Dominant favorable genes, 451

Dominant gene effects, 165

Dominant genes, 225-26

Dominant and recessive lethals, 160

Dominant unfavorable genes, 226-27

"Double-cross," 40, 42, 43, 166-68

Double cross hybrids, 433, 444

Double cross yields versus single cross
yields, 58

Double fertilization, 9
and embryo development, 87
and endosperm development, 87
and growth in the integuments, 87

Drosophila melanogaster
hybrid vigor in, 474-93
egg production, 474, 476
egg production curve, 476
lifetime daily egg production of different

races, 477
races, variation in egg production and

duration of life, 476

Drought resistance, in maize X teosinte,

196-97
Dwarf races, growth rates, 121
Dwarfs, 62
Dyads, 75

Early testing, 402-11
limitations, 406-11
as a measure of combining ability, 410

Early testing and recurrent selection, 400-
417
results, 403-6
as a tool in a breeding program, 406

Ear-row planting, 22

Ear-row selection, 456

Edwards technique of growing auxotrophic

mutants, 279
Effective factor, 298

Egg production, 478

distribution of variance, 486

range, 485
Eight-way combinations, 44
Elementary strains, 24
Elite germ plasm hypothesis, 138-39
Embryo development, 104
Embryo growth, lag phase, 82
Embryo and seed development, 103-5
Embryo size as related to seed size and
heterosis, 103-4

Embrj'os

artificial methods of cultivating, 95

of barley, 97

development in inbred and hybrid corn,
104

early growth in relation to endosperm
size, 91

frequency of formation, 95-96

growing small excised, 96

growth rates, 87

vigor, 94

Endogamy, 5

Endosperm, 81

genetic characteristics of triploid condi-
tion of the nuclei, 84
growth rates, 87
heterozygosis, 84
hybrid vigor, 85

INDEX

541

nature, 9

nutritive functions, 97

Environment-genotj'pe interaction in het-
erosis, 488-91, 493

Environmental variances, 169

Enzyme system, 103

Enzymes, 112, 115, 256

alpha methyl glucosidase, 257
galactase, 257
maltase, 257
melibiase, 257
sucrase, 257

Epistasis, 330, 331, 334, 353, 402, 458, 465,
486, 500, 501, 502

biases resulting from, 515

in corn yields, 286

graphic transformation removal, 465

Epistatic effects, 168

Equilibrium factor, 103

Equilibrium frequencies, 346
of genes, 446-67

Equilibrium gene frequency with over-
dominance, 463-64

Essential metabolite, 115, 116, 117
relation to growth, 115-17

Euchlaena {See Teosinte)

Euheterosis, 218-23, 296

balanced, 221-22

mutational, 218-20

Evening primroses, 20

Evolution {See also Introgression)
of maize, 176

Evolution of corn in Mexico, 425

factors involved, 419
Excised tomato roots, 116

heterotic vigor in, 122

Exogamy, 4

Experimental designs, 494-96

Experimental procedures used at North
Carolina Experiment Station, 494-96
analysis of data, 496-97
experimental designs, 494-96

Extrachromosomal inheritance, 260-61,
264-66

Extrapolated correlates, method of, 130,
148

Female gametophyte, 90

Fertilization
in the angiosperms, 83
in the gymnosperms, 83

First generation hybrids, 165-66, 174

First generation intervarietal hybrids, 172

First generation selfed lines, 448

Flagellar antigens, 268

Flint corn, 24

Flint-Dent ancestry in corn breeding, sig-
nificance, 146-48
Flour corns, 420

Fodder, maize X teosinte, 187-88
Forssman antigen, 250
"Four-way" crosses, 40, 42, 43

Frequency distribution of two types of
samplings, 404

Fruit, weight of and its component char-
acters, 306-7

Fungi growth, 115

Galtonian regression, 23

Gamete selection, 60, 61, 382-86, 399
method, 387
sources of

synthetic varieties, 378

single or more complex crosses, 378

inbred lines, 378
for specific combining ability, 378-88

Gamete selection program, feasibility of,
388

Gene action, 326
and overdominance, 294-96
types of, 225

Gene combinations, 420-23

Gene-controlled characteristics
chlorophyll deficiencies, 102
flowering pattern, 102
leaf form and pigmentation, 102
stalk abnormalities, 102
time of flowering, 102

Gene division, 259

Gene dosage, 237

Gene effects, 154
in a series of reactions, 237-39
specificity of, 236-55

Gene frequency, 341
equilibrium, 336

Gene interaction, 245, 330
in heterosis, 320-29

Gene pairs differentiating parents, 310-14
Gene recombination and heterosis, 298-319
Gene specificity, 237, 325
Genes, 265

accumulation of favorable dominant ef-
fects and general physiological inter-
action, 235

affected by environment, 62

complementary, 48

compound and with multiple effects,
227-29

controlling growth, 230-32

and cytoplasm, interaction, 232-33, 480

direct effects, 239-40

without dominance, 227

dominant and recessive, 225-26

equilibrium frequencies, 446-67

estimation of dominance, 494-516

542

INDEX

Genes — Continued

interallelic interactions, 319

manifold effects of, 340^41

pairing of, 371

in the synthesis of arginine, 239

transferred to a native population, 421
General combining ability, 352, 487-91

definition, 454

recurrent selection for, 470

General and specific combining ability,

352-70, 487
General and specific combinability, relative

importance of, 451, 487
Genetic correlations, negative, between

components of total performance,

337-38
Genetic implications of mutations in S.

typhimurium, 267-81
Genetic intermediates, 332, 334

optimum, 331
Genetic interpretation of regressions, 462-

65
Genetic interpretations of components of

variance between progenies, 503-9
Genetic mechanisms and heterosis, 100-103
Genetic nature of components of progeny,

variance, 498
Genetic and phenotypic covariances and

correlations, 515
Genetic structure of natural populations,

152

Genetic variability, 331
in economic characters of swine, sum-

mar}-, 350
Genetic variation in economic traits, nature

of, 330-41
Genetical combination, 31
Genetics and cytology of Saccharomyces,

256-66
Genie inheritance, 266
Genome components, effect on hybrid

vigor, 482
Genome contributions to hybrid vigor,

484-88
Genotype-to-background relationship, 104
Genotypes, 219, 336, 344, 425
Genotypes of inbred races for viable, lethal,

and recessive visible alleles, 481
Germ plasm, inbreeding effect on, 454
Glutamic dehydrogenase, 110
Golden Queen Dent, 131
Gourdseed Dents, explanation of name, 129
Governing genes, 428
Graphic transformations to remove epista-

sis, 465
Growth affected by deficiencies for essential

metabolites, 116

Growth curves of heterocaryons, 211
Growth of heterotic hybrid, determination,

123
Growth-limitation, 114

Growth in plants
abnormal, 118
modifying amount and nature, 117

Growth rates of dwarf races, 121
Growth rates of hybrids, 105
Growth requirements for hybrids, 121
Growth responses with thiamin, pyridoxine,
and niacin, 109

Guatemalan flints, 127

Gymnosperms
female gametophyte in, 83
fertilization in, 83
mature ovules of, 83
seed in, 82-85
seed coat in, 83
seed development in, 81

Gymnosperm ovule, 83

Haploid, 389

Haploid organism, 199, 389

Haploid sporophytes, 389

Hemizygote, 226

Heritabilities, 350

Heritability, 161, 174, 460
estimates of, 335-37, 352-70
for individual components, 335
for total performance, 335

Heritability and gain, 170-71

Heritability of specific combinability, 451

Heritable variance, 168

Heterocaryon formation, 200

Heterocaryons

characteristics of, 199-202

controlled production, 200

growth curves of, 211, 216

heterosis in, 202

model, 210-12

between sulfonamide-requiring mutant
and its double mutants, 212

vigor in, 120-23
Heterocaryons of Neurospora, 110, 199-

217
Heterocaryosis in Neurospora, 120
Heterocaryotic suppression of sulfonamide-
requiring character, 203-7
Heterocaryotic vigor, 202
Heterochromatic knobs, 136
Heterochromatin, 80
Heteromorphic chromosomes, 79

Heterosis, 14-48, 282, 425, 454

of American Corn Belt dents, 138-46
amount, 31
applicability, 217

INDEX

543

under asexual propagation, 320
biochemical models in Neurospora, 199-

217
breeding for in cross-pollinated plants,

55,400-417
breeding methods, 52-61, 400 417

breeding for in self -pollinated plants, 55

breeding for in vegetative ly propagated
plants, 56

and chromosomes, 492

in component traits, 303, 492

and cytoplasm, 492

development of concept, 49-65

dispersed, 127

and dominance, 224-35, 307, 494-516

and Drosophila, 111, 475

and early growth, 105

environment-genotype interaction in,
488-91

evaluation of, 329

example of utilization, 66

experiments with, 154-57

as explained genetically, 173

expression of, 224-25

first use of term, 50

gene interaction in, 320-29

genetic basis, 100, 101

genetic concepts, 61-65

and genetic mechanisms, 100-103

in heterocaryons, 202

due to heterozj-gosity at one locus, 203-
15

heterozygosity concept, 101

importance of internal factors, 123

inbreeding effects as related to defective
genes, 481

and interracial and intra specific hybridi-
zation, 198

and later growth, 106

and linkage, 224-35, 285

in maize, 1, 53

in maize variety hybrids, 182

in maize X teosinte, 183-84

measures of, 479-81

Mexican corns, 418-50

in native open-pollinated varieties, 419-
25

natural mechanisms for maintaining ad-
vantages of, 46-47

nature and origin, 218-23

in a new population, 418-49

as observed in pre-Mendelian research,
1-14

physiological basis of, 111-13

physiological mechanism of, 112

plasmagenes and chromogenes in, 224-35

in polj-genic characters, 159

in population genetics, 149-60

potential, 140

potential maximum, 340

practical use of, 44

as related to embryo and seed size, 103-4

as related to heterozygosity, 103

resulting from degenerative changes, 102

reversed or negative, 225

and rotational crossbreeding, 371-77

scientific basis of, 1-2

in self-pollinated plants, 52

single locus, 102

stimulus of heterozygosis, 49

in sugar cane, 322

in tomato root cultures, 109

as tool of the animal breeder, 151

usage, 98

use of in

farm crops, 51

horticultural crops, 51

silkworms, 51

livestock, 51

vegetative ly propagated plants, 51
utilization, 50-51, 55, 56

Heterosis concept, 16, 17, 18
beginnings of, 14-48
defined, 48
in work with hybrid corn, 14

Heterosis development

as affected by nutritional factors, 111
as affected by water absorption factors,
HI

Heterosis and dominance, as expression of
same physiological genetic phenomena,
307, 309

Heterosis and gene recombination, 298-319

Heterosis, maximum, 326-37

with the dominance hypothesis, 287-91
methods of selection for, 350-51

Heterosis and morphology, 141-46

Heterosis in Neurospora, biochemical mod-
els of, 203

Heterosis principle, 20

Heterosis reserves, 140-41

Heterosis, single gene, 155

Heterosis tests, inbred lines, 330-51

Heterosis theories

genetic explanation, 62
interallelic action, 62

Heterotic locus, 328

Heterotic hybrid, determination of growth,
123

Heterotic hybrids, 119

Heterotic h3'brids and inbreds, structural
differences, 112

Heterotic tomato hybrid, 123

Heterotic vigor in excised tomato roots, 122

Heterozygosis, 49, 152, 454
stimulus of, 224, 229, 282

Heterozygosity, 100, 102, 332, 373
degree of, 350

enforced in maize, 180-81, 193-96
range in degree of, 342
related to heterosis, 103
single locus, 199

544

INDEX

Heterozygosity and inbreeding depression,
linearity of, 459-60

Heterozygote

basis for superiority of, 158-60
physiological superiority of, 158
selective advantage of, 467

Heterozygote advantage, 347, 349, 350
for single loci and chromosome segments,
340-41
Heterozygote, diploid, 202

Heterozygous combinations, adaptive sig-
nificance, 173

Heterozygous condition, effects in, 187-91

Heterozygous inversions, 154

Heterozygous loci, 341

Heterozygous populations, variance, 297

Hexaploid, 389

Histidine, 278

Histidine synthesis relation to adenine, 278

Homocysteine, 215

Homozygosis, 49, 151

Homozygosity, 290, 403

Homozygotes, recurrent selection among,

470
Homozygous abnormal 10 plants, 73
Homozygous condition, effects in, 191-97
Homozygous diploid dent, 398
Homozygous diploid lines, 399
Homozygous diploids, 390
Homozygous gene arrangements, 153
Homozygous lines, 448
Honozygous mutant, 157
Homozygous normal allele, 157

Homozygous tester versus reciprocal selec-
tion, 342-47

Homozygous tester lines, 347

Human endogamy, 2-4

Athenians, 3-4

Greeks, 3

Hebrews, 2-3

Nordics, 4
Human inbreeding, 2-5

Hybrid advantage, 105

potential, 339
Hybrid, average, 291
Hybrid, commercial, 291
Hybrid combinations, selection, 27

Hybrid corn, 15, 16, 118

and Darwin's influence on W. J. Beal, 10

development of embryos, 104

double hybrid, 373

early explanations, 451-53

single cross, 373

vegetative habits of, 30
"Hybrid-corn makers," 15
Hybrid embryos, 96

Hybrid growth, 114

Hybrid and inbred corn grains, growth -
promoting activities of extracts from,
109

Hybrid nutritional requirements, 114-23

Hybrid plants, analyses of starch content of
leaves and stems, 108

Hybrid, reconstructing, 449

Hybrid results, 433

Hybrid substance, 243-51
fractions of, 247
in species hybrids, 243-51
tests for, 247
tests for similarities and differences, 248

Hybrid vigor, 14, 15, 98, 108, 112, 118-23,
127, 218, 224, 282, 424, 492

in advanced generations, 180-83

in artificial plant hybrids, 49

as associated with high embryo weight,
104

cause of, 13

chromosomal and cytoplasmic basis for,
479-81

from crossing inbred lines of maize, 50

definition, 474

discovery by Mendel, 7

in Drosophila, 474-93

effect of chromosome length, 484

effect of genome components, 482

of egg yields, 483

in the endosperm, 85

as evolutionary accident, 173

explained by dominance or partial domi-
nance, 65

explanation of, 478-79

factors effecting, 475

in field corn, 51

first described, 7

genome contributions to, 484—88

hypotheses to account for, 282, 481, 492

in maize, 142

necessary preliminary to, 4

in production of geniuses, 4

in plants, 2, 7-8

as result of allelic interaction, 103

summary of early knowledge, 13

in sweet corn, 51

in Zea mays, 10

Hybrid vigor development, relation of
phosphorus and nitrogen nutrition 106

Hybrid weakness, 225

Hybridity

mechanisms which promote, 152-54, 476
optimum, 152

Hybridization, 423-25
ancient, 185

natural, maize X teosinte, 183-84
present-day, 183-85

of Southern Dents and Northern Flints,
132-33

INDEX

545

Hybridization in evolution of maize, 175-98

discussion, 197-98

summary, 198
Hybridization and inbreeding, effects of,

294
Hybridization in maize, 9
Hybridization tests on corn, 429-47
Hybridized seed corn, 42
Hybridizing maize and teosinte, effects of,
186-87

Hybrids, 424

adaptability of, 99-100

barley, 301

differences in water requirements, 106

first generation, 165-66, 174

first generation intervarietal, 172

growth rates, 105

growth requirements, 121

intervarietal, 161

physiology of gene action, 98-113

pre-Mendelian, 11

presence of iron and aluminum, 106

response to soil conditions, 106

Hybrids and inbreds
correlation between, 294
gene determination of variation, 292-94

Hybrids, second-cycle, 453-54

Hybrids and synthetics developed from
Celaya lines, 442

Hypoxanthine, 109, 119

Inbred corn, development of embryos, 104

Inbred and hybrid corn grains, growth-pro-
moting activities of extracts from, 109
"Inbred Index," 142

Inbred Unes, 386, 401, 403
of corn, 230
formation of, 427-31
for heterosis tests, 330-51
methods of improving, 57
in reciprocal selection, 347
recovering of, to retain advantage at-
tributable to heterosis, 310-18
of tomatoes, crossing of, 307-10
undesirable characters, 57

Inbred parents, partial regression of yield
on yield of, 460-62

Inbred races, genotypes for viable, lethal,
and recessive visible alleles, 481

Inbred selection, 140, 163, 170

Inbreds

crossed, 34

residual heterozygosity in, 31

selfed, 34

Inbreds and heterotic hybrids, structural
differences between, 112

Inbreds and hybrids
correlation between, 294
gene determination of variation, 292-94

Inbreeding, 161, 414
dangers of, 18
degree of, 23

deterioration incident to, 44
different generations of, 63
early ideas on, 1-13
effect on growth, 81
effect on seed collapse in Medicago saliva,

85-89
to improve cattle breeds, 6
injurious effects of, 17
reciprocal recurrent selection with, 457
records of, 5

results with corn and swine, 338-40
in seed development, 81-97
too-close, 2

Inbreeding, animal, 5-7

Inbreeding versus crossbreeding, 7

Inbreeding and crossbreeding effects, 81-97,
331-32, 350

Inbreeding and crossbreeding maize, 12

Inbreeding decline, 332, 341, 350

Inbreeding depression, 297, 349
effect, 86
and heterozygosity, linearity of, 459-60

Inbreeding effect
on germ plasm, 454

on heterosis as related to defective genes,
481

Inbreeding experiments of swine, results,
331-32

Inbreeding, human, 2-5

Inbreeding and hybridization, effects, 294

Induction of polygenic mutations, 160

Inheritance of quantitative characters, 327

Insect galls, 118

Interaction between alleles, 229-30

Interaction of genes, 245
and cytoplasm, 232-33

Interallelic association, 330

Interallelic gene interaction, 63, 319, 353

Interallelic and intraallelic interactions and
pleiotropy, 317-19

Intercrossing, 420

Intergroup selection, 151

Interracial hybridization in maize, 176-83

Interspecific hybridization of maize and
teosinte, 183-97

Interspecific hybrids, 95

Intervarietal hybridization, selection, and
inbreeding, 172

Intervarietal hybrids, 161

potence in, 166
Intraallelic association, 330
Intraallelic gene interaction, 353

Intraallelic and interallelic interactions and
pleiotropy, 317-19

546

INDEX

Intra-breed linecrosses, comparison with

representative purebreds, 353
Introgression, 419, 421

of genes, 422

teosinte X maize, 185
Inversion, 77, 222
Iron in hybrids, 106
Irregular Mendelian segregation, 259
Iterative crosses, 40, 42

"Junk" inbred, importance of, 139

Kalanchoe plants, 117

Kinetic theory of position effect, 78

Knob number, 129

Knobbed chromosomes, 75

Knob less chromosome 9, 70

Kynurenine, 238

Lancaster Surecrop, 140
Leaming Dent, 17

Leaves and stems of hybrid plants, analyses

of starch content, 108
Lethal genes, 229, 476
Lethals, incompletely recessive, 289
Linear heterogeneity, 265

Linearity of inbreeding depression and
heterozygosity, 459-60

Line cross tests, selection for general com-
bining ability, maternal ability, and
specific ability, 364-68

Lines selfed once versus lines selfed more
than once in hybrid formation, 444-48

Linkage, 327, 402
effect of, 501

and gene recombination, 314
and heterosis, 285
multiple factor, 130
role of, 502-3

Linkage and marker genes in barley, 298-
302

Linkage and pleiotropy, 317-18

Linkage relations, 314-17

Linkage systems, 139

Locus, heterotic, 328

Luxuriance, 22-23, 224, 291, 296

Main and component characters of to-
matoes, 305-10

Maize

archaeology, 175, 185

backcrossing in, 235

breeding, importance of character com-
plexes, 131

chlorophyll production in, 228-29

chromosomes, 185-87

chromosomes, localized centromeres of,
77

cross- and self-fertilization in, 49

heterosis in, 53

history, 176

hybrid vigor in, 142

hybridization, 9

hybridization in evolution of, 197-98

interbreeding and cross breeding, 12

interracial hybridization in, 176-83

introgression with teosinte, 185

male inflorescence of, 133

Mexican races, origin of, 176-80

monoploids in, 389-99

plants, comparison of fertility and steril-
ity, 234

pod corn, 175, 176

pop corn, 175, 176

preferential segregation in, 66-80

races, 148

races, crossed, 177, 178, 182

and teosinte hybridization, effects of
186-87

and teosinte hybrids, 183-84

and teosinte interspecific hybridization,
183-97

and Tripsacum, relation between, 138

white dent, 29, 35, 38

Male inflorescence of maize, modification
by condensation, 133

Malic dehydrogenase, 110

Manifold effects, 350

Marker genes and linkage in barley, 298-302

Mass matings, 257

Maternal ability, 352

Maternal effects, 352, 499-500

Maternal monoploids, 393

"Mating systems," 152

Mating types of spores, 257

Maximum yields from crossing high by low

combinations, 465-66
Mechanical picking, breeding for, 146

Mechanisms which promote hybridity, 152-
54

Megasporogenesis, 79

occurrence of preferential segregation, 66,
68

Meiosis, 390

diploid behavior at, 500

mechanics of, in maize monoploids, 393

Meiotic divisions, 72
Mendelian ratios, 11
Mendelian sampling, 353, 365
Mendelian segregation, 19
irregular, 259

Mendelian theory, direct tests of, 259
Mendel's first law, 70
Mendel's Laws, rediscovery of, 49
Mendel's laws of segregation and recom-
bination, 66

INDEX

547

Mendel's second law, deviations from, 70

Meristematic growth, 102, 108

Metabolic activity in enzyme systems, 110

Metabolic reactions, 214

Metabolites, essential, 115, 116, 117

Metaphase I, 78

Metaphase I with eleven dyads, 67

Metaphase II, 74, 75, 76, 78, 79

Metaphase II dyads, 73

Metaxenia, 10

Methionine requirers, 276

Mexican commercial corn growing areas,

426
Mexican breeding stations, 426

Mexican corns, 175-98, 418-50
Bolita, 427, 443, 449
Cacahuacintle, 420, 423
Celava, 427, 436, 437, 438, 440, 441, 442,
443, 444, 446
crosses between Conico Nortefio and

Tabloncillo, 438
hvbrids and synthetics developed from,
'442
origin of, 436
Celita, 444
Chalco 433 434
Chalqueno, 418, 423, 424, 427, 431, 433,

434, 446, 449
Conico, 420, 423, 424, 427, 435
Conico Norteno, 427, 435, 436, 438, 439,
443,444

probable origin of, 421
Leon CrioUo, 429
Olotillo, 420, 423
Palomero Toluqueno, 420, 423
Tabloncillo, 427, 438, 443
Tepecintle, 420, 432
Tuxpeno, 420, 423, 424, 427, 442, 444
Urquiza, 429, 430, 433, 434
Vandeno, 427, 444, 449
probable origin of, 422
Mexican corn belt, 436
Mexican corn, evolution of, 419, 424, 425
Mexican corn improvement program, 425
Mexican corn program, 428, 448
Mexican corn races, 427
Mexican open-pollinated varieties, 419
Mexican races of maize, origin of, 176-80
Mitochondria, 261
Model heterocaryons, 210-12
Modifying genes in maize X teosintc, 191-

92
Monoploid sporophyte, 389, 398, 399

Monoploids

distribution, 396

fertility, 393, 398

frequency of occurrence, 394-95

in maize, 389-99

method of isolating, 392
origin, 389
reproduction, 390
study of

in the agronomic field, 390

in the cy to logical field, 390

in the genetic field, 390

Morphological characters as related to
heterosis, 141-46

Morphology of inbreds, technique of study-
ing, 141
Mosaic dominance, 295
Multiform genus, 321
Multiple alleles, 102, 228, 254
Multiple allelism, 500
Mutant genes, survival and growth, 235
Mutant strains, 214
Mutant, superior, 323

Mutants, 218

in S. typhimunum, 281

Mutation, 103

Mutation frequency and X-ray dosage,

271-74
Mutation, random, 323
Mutation theory, De Vries, 20
Mutational euheterosis, 218-20

Mutations

deleterious character of, 219
genetic implications in S. typhimurium,
267-81
Mutations isolated and X-ray dosage, rela-
tion between, 273

Native open-pollinated varieties, utilization

of, 426-27
Natural populations, genetic structure of,

152
Natural selection, 15, 151, 157, 219-20
Negative heterosis, 225
Neo-centric activity, 75, 77
Neo-centric fibers, 75

Neo-centric regions, 72, 74

formation of, 77
Neo-centromeres, arising from chromosome

ends, 80
Neomorphs, 295

Neurospora

biochemical models of heterosis, 199-217
biochemical reactions essential to growth,

120, 208-9
cells of, 199-201

and higher organisms, difTerences be-
tween, 201-2
Nicking, 353, 451
Nicotiana hybrids, 14

548

INDEX

Nicotiana rustica

experimental results, 162-71

experiments with, 161-62

fixing transgressive vigor in, 161-74

plants of, 163
Nitrogen metabolism

of corn and tomato, 107

of the plant, 110
Nitrogen nutrition

relation of hybrid vigor development, 106

use by hybrids and inbreds, 106

Non-allelic interactions, 164, 174
Non-heritable variability, 162
Non-random segregation at megasporo-

genesis, 70
Normal segregation, 70
Northeastern Flint corn of Guatemalan

origin, 129
Northern Flints, 127, 129, 131

characteristics of, 136

distribution of chromosome knobs, 130

historical record, 128

kernel of, 133

pairing of rows, 133

seeds of, 134

shanks of, 134

and Southern Dents, crosses between, 137

and Southern Dents, differences between,
136

typical ears of, 134

typical plants, tassels, and staminate
spikelets of, 135
Nuclear membrane and chromosomes, 262
Nucleotide cozymase, 110

Oenothera, 20

One gene heterosis, 203-15, 217

One gene-one enzyme hypothesis, 268, 276

Open-pollinated varieties, continuing im-
portance in inbreeding program, 140
Outcrossing, 406

Overdominance, 221, 229, 282, 323, 330,
338, 416, 451-73, 492, 497

in cattle, 467

equilibrium gene frequency with, 463-64

physiological nature of, 457

in tomato hybrids, 319

in yield of corn, 469
Overdominance and dominance, 282-97
Overdominance and gene action, 294-96
Overdominance and recurrent selection,

451-73
Overdominant loci, 291
Overepistasis, 458

Para-aminobenzoic acid. 111, 207, 210, 214

Parthenogenesis, 47, 389, 394-99
artificial induction, 396
effect of male parent, 394

effect of seed parent, 395

in monoploid derivatives, 397

Partial dominance as explanation of hybrid
vigor, 65

Partitioning phenotypic variance, 161

Partitioning phenotypic variance, herita-
bility, and number of effective factors,
168-70

Paternal monoploids, 393

Penicillin, 269

Penicillin screening, 272-74

Performance index, 382, 386

Performance indices, 284-85

Permanently viable pure lines, 16

Phenotype, 336, 344, 463

Phenotype-genotype relations, 162-65

Phenotypic and genetic covariances and

correlations, 515
Phenotypic and genie dominance, 318
Phenotypic variance, partitioning, 161
Phosphorus-absorbing capacity in corn, 107

Phosphorus nutrition
relation of hybrid vigor development, 106
use by hybrids and inbreds, 106
Photosynthesis, 327
Physical vigor, 34

"Physiological key" substances, 112
Physiological mechanism of heterosis, 112
Physiological mosaic dominance, 295
Physiological nature of overdominance, 457
Physiological superiority of heterozygote,
158

Physiology of gene action in hvbrids, 98-

113
Pigeon-dove hybrids, 243-45
Plant breeding, selection, 15
Plant hybridization, 9
first record of, 9

Plant hybrids, 7-13

Plasmagenes and chromogenes, 233-34
in heterosis, 224-35

Pleiotropy and interallelic and intraallelic

interactions, 317-19
Pleiotropy and linkage, 317-18
Pneumococcal types, specificities of, 254
Pneumococci, antigens of, 254
Pod corn, 418
PodophylUn, 397
Pollen abortion, 74
Polycross trials, 56
Polycrosses, forage yields compared to top

crosses of the same clones, 57

Polygenes, 173

INDEX

549

Polygenic characters

heterosis in, 159

selective advantage of, 158
Polygenic inheritance, 153
Polygenic mutations, induction of, 160
Polymorphism, 154
Polyploid series, 257
Polyploids, 321, 389
Pop corn, 418

Population

adaptation to environment, 151

definition, 149

evolutionary factors responsible, 150-52

genetic structure, 150

as natural unit, 150
Population genetics of heterosis, 149-60
Population variance, 292-94
Populations, natural, genetic structure, 152
Position effect, kinetic theory of, 78
Postgermination growth in plants, 105
Potassium availability, studies on tomato

inbreds and hybrids, 107-8
Potence in intervarietal hybrids, 166
Potence in maize X teosinte, 188-91
Potential heterosis, 140
Prolificacy, 335
Predicting combining ability, 59

Preferential segregation
factors responsible, 68
and neo-centric activity, 78

Preferential segregation in maize, 66-81

Pre-Mendelian hybrids, 11

Prepotency, 451

Post-Mendelian investigations, 2

Proembryo, 87

Progeny variance, genetic nature of com-
ponents, 498

Progressive evolution, 265

Propagation, vegetative, 12, 13

Prophase II, early, 74

Prophase II, late, 74

Prototrophic organisms, 269

Pseudo-alleles, 254, 295

Pseudoheterosis, 223

Pseudo-overdominance effect, 325, 458, 501

Pure lines and their hybrids, relationship
between, 28

Pure line method in corn breeding, 28

Pure line method of corn production, 44

Pure strains, 24

Pyridoxal phosphate, 110

Pyrimidineless mutant plus suppressor al-
lows growth, 217

Pyruvate carboxylase, 110

Quadruple crosses, 44
Quantitative inheritance, role of limiting
factors, 324-27

Random mating, 432

Recessive characters, deleterious, 100, 323
Recessive and dominant lethals, 160, 480
Recessive genes, 225-26
Recessiveness and deleterious effect, corre-
lation between, 284

Recessives, 11

Reciprocal cross, 28

Reciprocal effects, 487

Reciprocal recurrent selection, 415-17

with inbreeding, 457
Reciprocal selection, 351
Reciprocal testing, method of, 443
Recombination, 267

Recombination tests in Salmonella, 274-75
Reconstructing a hybrid, 449
Recurrent selection, 351, 411-12, 451-73

effectiveness of, 467-70

for general combinability, 470

among homozygotes, 470

meaning, 452

as a method for modifying chemical com-
position, 411

as a method for modifying combining
ability, 411

in modifying combining ability, 415

for specific combinability, 454-57, 470-73
Recurrent series, 412

Recurrent selection and early testing, 400-
417

Recurrent selection and inbreeding in modi-
fying oil percentage in corn, 411-12

Recurrent selection and overdominance,
451-73

Red blood cells, 239

Regression, partial, offspring on parents,
463

Regression trends, 461

Regressions, genetic interpretation, 462-65

Relic genes, 424

Reproduction, asexual, 47

Repulsion linkages, 502

Reversed heterosis, 225

Residual heterozygosity in inbreds, 31

Rh substance, 242

Root cultures, 109-10

Root growth pattern, relation to heterosis,
106

Rotational crossbreeding, 374-77
and heterosis, 371-77

Rust, 424

550

INDEX

Saccharomyces

chromosome maps of, 258
genetics and cytology of, 256-66

Salmonella

advantages for genetic studies, 268
recombination tests, 274-75

5. typliimurium, genetic implications of mu-
tations in, 267-81

Scales of measurement as related to domi-
nance, 313
Scaling tests, 162, 164, 174
Second-cycle hybrids, 453-54

Secondary fertilization, 82, 90, 93
in angiosperms, 84
in flowering plants, 95

Seed development, 104

in angiosperms and gymnosperms, 81

early stages of, 96

without fertilization, 89-93

grade and embryo growth potentialities,

93-97
inbreeding and crossbreeding in, 81-97

Seed in gymnosperms and angiosperms,
82-85

Seed and embryo development, 103-5

Seed size as related to embryo size and
heterosis, 103-4

Seedling growth, 105
and heterosis, 105

Seeds, number in maize X teosinte, 196

Segregating factors, number of, 174

Segregation, 24

Segregation for yield factors, 382

Segregations, irregular, 258

Selecting for maximum heterosis, effective-
ness of methods, 339-51
Selection, 14, 161, 341, 406, 414

for additive effects in normal distribution,

355
for additive genetic values in individuals,

359-61
for cross performance, 349
against a dominant deleterious mutant,

220
effectiveness of, 333, 335
effectiveness of within inbred lines, 332-

34
as an estimation problem, 354-58
when form of distribution is unspecified,

355-58
for general and specific combining ability,

353-54
for general combining ability, maternal

ability, and specific ability' in line cross

tests, 364-68
for general combining ability in topcross

tests, 361-64
homozygous tester versus reciprocal,

343-47

index method, 355, 356

on individual performance, 349

ineffectiveness of, 335, 350

by maximum likelihood estimates, 358-59

natural and artificial, 219-20

natural, for heterozygosity, 180-81

performance levels attainable by, 341

reciprocal or homozygous tester method,

349
results of, 170

for specific combinability, 454
on test-cross, 349
with unknown variances and covariances,

358-59
use of all records in, 359

Selection experiments, 156
controlled, 334-35
with swine, 350

Selection for general and specific combining
ability, need for additional research,
369-70

Selection within and among inbred prog-
enies, 401-2

Selection of inbreds on performance, 140

Selection index, 354
modified, 356

Selection for maximum heterosis, methods

of, 350-51
Selection, reciprocal, 351

Selection, recurrent, 351
and early testing, 400-417
for maximum heterosis, 341-42
and overdominance, 451-73

Selective advantage of a heterozygote, 467

Selective advantage of polygenic characters,

158
Self-fertiHzation, 16, 17, 45, 48, 139

in alfalfa, 81

versus cross-fertilization, 20

in maize, 49
Self-fertilized populations, 322
Self-pollinated plants, 161
Self-pollinated and cross-pollinated plants,
comparison of methods, 55

Selfed lines, first generation, 448
Selfing, 12, 16, 23, 173-74, 379, 382, 390,
402, 406, 409

Selfing and crossing corn, results, 36

Selfing and loss of vigor, 13

Selfing series, 412-15

Selfing and sibcrossing, comparison be-
tween, 41

Semi-inbred lines in synthetics and h}--
brids, utilization of, 431

Senility, 14

Sequential testing, 361
Shank, importance in modern corn breed-
ing, 146-17

INDEX

551

Sib crosses, advantage over self-fertiliza-
tion, 29
Sicklemia, 295

Single cross

distribution, 145

estimation of the value of, 367
Single cross yields versus double cross, 58
Single gene heterosis, 155, 282
Single locus heterosis, 102
Somatic antigens, 268
Southern Dents, 127, 129, 131

characteristics of, 136

distribution of chromosome knobs, 130

ear of, 133

historical record, 128

kernel of, 133

and Northern Flints, crosses between, 137

and Northern Flints, differences between,
136

seeds of, 134

shanks of, 134

typical ears of, 134

typical plants, tassels, and staminate
spikelets of, 135
Species specific, 254

Specific combinabiiity, 452

heritability of, 451

recurrent selection for ,454-57, 470-73

working definition, 454
Specific combining ability, 352-53, 487-91

gamete selection for, 378-88
Specific and general combinabiiity, relative

importance, 451
Specific and general combining abihty,

352-70
Specificity of gene effects, 236-55
Spindle, 74

Spores, mating types, 257
Sporophyte, 389

Standard inbreds, 233
conversion, 233

Statistics, role in genetical research, 494

Sugar cane

cross-fertilization in, 320
male sterility in, 320
propagated asexually, 320
reaction toward inbreeding and out-
crossing, 320
self-pollination in, 320

Sulfonamide-requiring character
biochemical basis for, 207-10
heterocaryotic suppression of, 203-7

Sulfonamide-requiring strain, 212, 215
Sulfonamides, 111

Sulphur compounds, utilization by various
auxotrophs, 276

Sulphur containing compounds, reactions,
277

Super-dominance, 282
Superior mutant, 323

Supernumerary chromosomal fibers, 72

unorthodox formation, 72
Suppressor heterocaryosis, 212-15
Suppressor genes and sulfonamide utiliza-
tion, 213
Suppressor mutant strains, growth curves,

214
Suppressor mutants, 203
Swine and corn inbreeding
degree of heterozygosity, 338
effectiveness of continued phenotypic
selection, 338
Swine, heterozygote advantage for single
loci and chromosome segments, 340-41
Swine inbreeding experiments, results, 331-

32
Synapsis, 69, 115
Synthetic varieties, 432
formation, 433

Synthetics, 448

estimating number of lines to use, 432
propagated through open-pollination, 426

Teosinte {See also Maize)
ecology, 183
variation, 184-85

Teosinte germplasm, 419
Terminal chiasmata, 68
Test crosses, 408

Tester lines, partially inbred, use of, 347-48
Tetrad analysis, 258
Thiamin pyrophosphate, 1 10
Thiamine, 118
"Three-way" crosses, 40, 42
Tobacco, crosses between varieties and
species, 49

Tomato hybrids

number of fruit that ripens, 309

weight per fruit, 309

yield of ripe fruit, 309
Tomatoes, main and component characters,

305-10
Tomato roots, 116
Topcross combining ability, 406
Transgressive characteristics, 172
Transgressive inbred, 161
Translocation point, 68
Triphosphopyridine nucleotide, 110
Triploids, 389
Tripsacum, 419
Tripsacum hypothesis, 138
Trivalent associations, 69

Univalents, 69

552

INDEX

Ustilago maydis, 24

Variability, 49

Variance components, estimating, 368

Variance components, genetic interpreta-
tions, 503

Variances of general, maternal, and specific

effects, estimation of, 368-69
Variety crossing, 16

Variety testing, 426

Vascular organization, 108

Vegetative propagation, 12, 13

Viability, 335

Vigor

hypotheses for difference in, 300-302
loss of and selling, 13
reestablished by outcrossing, 7

Vigor in original plant as measure of number
of favorable yield genes, 428

Vigorous hybrid and weak inbred, difference
between, 327

Virulence, 279

Water requirements of hybrids, 106
Waxy gene, 239

Weak inbred and vigorous hybrid, differ-
ence between, 327

White dent corn, 32

frequency curve of grain rows, 21
White dent maize, 29

average grain-row numbers, 38

average values in the families of, 35

yields per acre, 38

Xenia, 8, 9
history of, 10
in maize, 10

X-ray dosage and mutation frequency,
271-74

Yeast genetics, 257

Yeast, advantages for biochemical genetics,

257

Yield loss from one generation to the next,
42

Yield of seed, prediction of, 302

Yielding abilitv, correlations indicative of,
54

Zea {See Maize)

Zea mays

double crosses, 378

gamete selection for specific combining
ability, 378-88

single cross performance data, 378

test crosses, 379
Zygote, 371

[

PRINTED H

N USaJI