Digitized by the Internet Archive

in 2011 with funding from

LYRASIS members and Sloan Foundation

http://www.archive.org/details/tobaccobreedingiOOhaye

THE CONNECTICUT

AGRICULTURAl. EXPERIMENT STATION

NEW HAVEN, CONN.
BULLETIN 176, MAY, 1913.

TABLE OF CONTENTS.

PAGE

Introduction, 5

Effects of Inbreeding in a Close Fertilized Species 6

Previous Work on Effects of Selection, 8

" " Inheritance of Size Characters, 10

" " Tobacco Breeding, 11

The Material Used .^ 13

The Methods Used, 14

Family (402X403) Havana X Sumatra, 15

(403X402) Sumatra X Havana, 18

Effects of Selection on the Havana X Sumatra Cross, . . 20

Quality of Cured Leaves, 26

Grain in Tobacco Leaves, 28

Texture Observations, 32

Results of Sorting Test, 34

Conclusions, 38

Family (403X401) Sumatra XBroadleaf, . 39

Inheritance of Leaf Number, 40

Shape and Size of Leaf, 42

Quality of the F3 Selections, 46

Conclusions, 48

Family (402X405) Havana X Cuban 49

Inheritance of Leaf Number, 51

Shape and Size of Leaf, 54

Inheritance of Quality, 56

Conclusions, 58

Interpretation of Results, 59

General Conclusions, 62

Literature Cited, 64

TOBACCO BREEDING IN CONNECTICUT.

By collaboration of H. K. HAYfes, Plant Breeder, Connecticut Agricul-
tural Station, E. M. East, Bussey Institution, Harvard Univer-
sity, and E. G. Beinhart, Assistant, Office of Tobacco Investiga-
tions, Bureau of Plant Industry, U. S. Department of
Agriculture.

INTRODUCTION.

The investigations, with which this paper deals, were com-
menced in the year 1908, and since that time have been carried
on in co-operative agreement between the Office of Tobacco Inves-
tigations of the Bureau of Plant Industry, United States
Department of Agriculture, Laboratory of Genetics of Harvard
University, and The Connecticut Agrictdtural Experiment Station.

The primary object of the work has been to study some of
the fundamental principles involved in tobacco breeding, with
the belief that a knowledge of these principles is absolutely
necessary if one is to build up a system of both practical and
scientific breeding.

It is self evident that the complex nature of the problems
involved makes it impossible to reach anything like a final
solution at present; this paper, therefore, is to be considered
in the nature of a report of progress. In it are described the
resiilts obtained during the past four years.

Connecticut Experiment Station, Bulletin 176.

Effects of Inbreeding in a Close-Fertilized Species.

Tobacco is a naturally close pollinated plant, although inter-
crossing through the agency of insects is probably somewhat
frequent. Observations on the earlier blossoms of the flower
head have convinced the writers that in many cases, at least,
fertilization of the pistil has taken place before the blossom
opens. In the later flowers the chances of intercrossing are
much greater, as the blossom often opens before fertilization
has been accomplished. It is evident that, as tobacco is a
naturally close-fertilized plant, it must be vigorous under self
fertilization, but some data on actual controlled inbreeding are
given to further substantiate this belief.

Darwin, in his classical experiments on inbreeding and cross-
breeding, found some types which were very vigorous when
continually self -fertilized.

Garner (1912) reports that a number of types have been inbred
under bags for six or eight years by the United States Depart-
ment of Agriculture without any observable change in vigor
or growth habit. A certain strain of our present Connecticut
Cuban shade type, now grown on one of our large plantations,
was inbred for a period of five years (1903-1908) by saving seed
from individual plants under a paper bag. Since that time
seed has been saved from desirable plants under cloth tent, the
chances, however, seeming very small that seed so produced
will be cross-fertilized. Instead of showing a loss of vigor due
to self-fertilization, this type seems more vigorous than in the
early years of its introduction.

The Sumatra type, which has been used as one of our parent
varieties, has been inbred for a period of seven years, without
giving any evidence of accumulated evil effects of inbreeding.

In a large series of generic crosses of Nicotiana the writers

Inbreeding In a Close Fertilized Species. 7

have observed a wide range of variation as to increased vigor
due to crossing. In some cases the first hybrid generation was
very vigorous while other species crosses were non-vigorous.

In a previous paper (Hayes, 1912) on variety crosses within
the species, five characters were measured in Fi and were com-
pared with the average of their parents for three sets of crosses.
These characters were height of plant, length, breadth and size
of leaf, and number of leaves per plant. All showed an increase
over the average of the parents, except in the number of leaves
per plant, which was almost exactly intermediate.

To quote from a previous paper (East and Hayes, 1912) :

"We believe it to be established that:

"1. 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 the same
phenomenon. This phenomenon is heterozygosis.* Crossing
produces heterozygosis in all characters by which the parent
plants differ. Inbreeding tends to produce homozygosis auto-
matically.

"2. Inbreeding is not injurious in itself, but weak types,
kept in existence in a cross-fertilized species through heterozy-
gosis, may be isolated by its means. Weak types appear in
self-fertilized species, but they must stand or fall by their own
merits."

The matter has been mentioned here because of its bearing
on the subject in hand. Houser (1911) has advocated the
system of growing first generation hybrid tobacco as a commer-
cial proposition. This was suggested for the heavy filler types

*Owing to the rediscovery of Mendel's law of inheritance, we now
know that many characters are separately inherited, and by the use of
descriptive factorial formulas the breeding facts are made clear. If a
certain character breeds true it is in a homozygous condition and each
male or female reproductive cell is supposed to bear some substance or
factor for the development of the character. If a cross is made between
two races which differ in a certain character we know that of the two
uniting reproductive cells, the one contains the factor for the contrasted
character and the other does not. The resulting plants of this cross
will not breed true in the next generation and they are said to be in a
heterozygous condition for the character involved. The amount of
heterozygosis produced by any cross depends on the number of gametic
factorial differences of the parent plants.

8 Connecticut Experiment Station, Bulletin 176.

of tobacco which are grown in Ohio. 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 compared with quality. Because of
the great importance of quality it seems much more reasonable
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, venation, and
number of leaves, together with that complex of conditions
which goes to make up quality, than by any other method.

Previous Work on Effects of Selection.

It is a well-recognized fact that among both plants and
animals no two individuals are exactly alike. This diversity
is due to two main kinds of variation:

1. Fluctuating Variations, such as size, shape, and number of
various plant organs, which are due to different conditions of
fertility, or to better positions for development. Such varia-
tions are not inherited.

2. Inherited Variations, which may be either large or small,
but are caused by some differences in the factors of inheritance
and are entirely independent of their surrounding conditions
for their transmission, although favorable environment is often
needed for their full development.

The real basis of the Mendelian conception of heredity is a
recognition of the fact that the appearance of a plant is not a
correct criterion of that particular plant's possibilities of trans-
mitting any particular quality, but that the breeding test is
the only real means of determining the plant's hereditary value.

By the universal adoption of Vilmorin's "isolation principle,"
in which the average condition of a plant's progeny is used as
the index of that particular plant's breeding capacity, breeders
have recognized these classes of variation.

A practical example demonstrating the truth of this classi-
fication is the work of Dr. H. Nilsson and his associates at
Svalof, Sweden. In 1891 a large number of heads from autumn
wheat varieties were collected and were separated into their
respective botanical and morphological groups, about 200

Previous Work on Effects of Selection. 9

groups in all being thus selected. In several cases certain
forms were found which had no duplicates, and in these cases
the individual form represented a group in itself. The following
season each group was given a separate plot and carefiil records
were made of the niimber of heads and plants which were the
ancestors of each plot.

A careful study of the resulting harvest showed that, of all
the cultures under observation, only those which originally
came from a single plant produced a uniform progeny (Newman,
1912).

The theoretical interpretation of this class of results was given
by Johannsen (1909) through his work with beans and barley.
This investigator found that a commercial variety was in reality
composed of different and distinct types which could be separa-
ted from each other by self-pollinating the individual plants
and studying their progeny. For example, he investigated the
character weight as applied to individual beans and found that
progress could be made when larger beans were selected from
the mixed commercial crop for several seasons. On the other
hand, after types comparatively homozygous had been isolated
by inbreeding, the same results were obtained in each isolated
line when large beans were planted as when the smaller ones
were used for seed — although fluctuation due to external
conditions still continued. This he explained as due to the fact
that environmental influences were not inherited but that a
plant simply transmits its inherent germinal qualities.

Certain corroborative results which show that fluctuating
variations are not inherited and that characters in a homozygous
condition are reproduced in practically the same degree gener-
ation after generation have been obtained by Barber (1907)
with yeasts; Pearl and Surface (1909) and Pearl (1912) with
poultry; East (1910) with potatoes; Hanel (1907) with Hydra:
Jennings (1908, 1910) with Paramaecium; Love (1910) with
peas, and Shull (1911a) with maize.

It is true that Castle (1911, 1912 a. b.) reports experiments
with a variable black and white coat color of the rat, in which
he shows that selection progressively modifies a character
which, in crossing with other types, behaves as a simple Mendelian
unit. These resiilts can be interpreted and, we believe, inter-
preted in a manner more helpful to practical breeding by assum-

10 Connecticut Experiment Station, Bulletin 176.

ing that although the coat pattern is transmitted as a single
unit, its development is affected by several other imit charac-
ters independent of the general color pattern in their trans-
mission. It may be that a few characters are so unstable that
they may be modified by selection after reaching a homozygous
condition, but so many thousand characters have been shown
to Mendelize and to breed true in successive generations when
in the homozygous state that for all practical purposes these
laws may be assumed to be universal in sexual reproduction.
Further reasons for this conclusion are given in the next few
pages.

Previous Work on Inheritance of Size Characters.

Since different degrees of expression of quantitative charac-
ters are inherited, as has been shown by Johannsen, and since
within an inbred line homozygous for a character, change can
seldom if ever be effected by selection, there seems good reason —
as stated before — for believing that size characters are inherited
in the same manner as qualitative or color characters.

The discovery of NHsson-Ehle (1909) that certain hybrids
are heterozygous for several inherited factors, either of which
alone is capable of producing the character, laid the foundation
for the proof of the generality of the Mendelian interpretation
of inheritance in sexual reproduction.

It was from similar facts that East (1910a) made the first
Mendelian interpretation of the inheritance of quantitative
characters by assuming absence of dominance and a multi-
plicity of factors each inherited independently and capable of
adding to the character, the heterozygous condition of any
character being half the homozygous.

In the last few years a number of investigations have been
made which show that linear or quantitative characters show
segregation. Some of the investigations which show segregation
in quantitative characters are as follows: Emerson (1910) for
shapes and sizes in maize, beans and gourds; Shull (1910,
1911b) for row classes of maize and for Bursa characters;
East (1911) and East and Hayes (1911) for height of plants,
length of ears, weight of seeds, and row classes in maize; Tammes
(1911) for certain characters of Linum forms; Tschermak

Previous Work on Inheritance of Size Characters. 11

(1911, 1912) for time of flowering in peas and for weight of
seeds; Hayes (1912) for height of plants, area of leaves, and
leaf number of tobacco; Davis (1912) for Oenothera characters;
Webber (1912) for plant characters of peppers; Belling (1912)
for plant characters of beans; McLendon (1912) for cotton char-
acters; Gilbert (1912) for characters of tomatoes; Heribert-
Nilsson (1912) for Oenothera characters; Phillips (1912) for
body size in ducks; Pearl (1912) for fecundity in fowls;
and Emerson and East (1913) for other characters of maize.

A few investigations which also comprise the Fs generation
show that in some cases forms breed true giving no greater
variability than the parent types. These results are of value
in any system of breeding which, in a large measure, deals
with size characters. Thus, by crossing two types which
differ in quantitative characters we may expect to obtain a
segregation in F2 and in Fs, some forms breeding true for some
characters and others again recombining the characters in which
they are heterozygous.

The possibilities of obtaining pure forms in F3 will, then,
largely depend on the number of character differences of the
parental types. A complete exposition of both theory and
practice when dealing with quantitative characters is given
in Research Bulletin No. 2 of the Nebraska Agricultural Experi-
ment Station entitled "The Inheritance of Quantitative Charac-
ters in Maize" by collaboration of Emerson and East (1913).

Previous Work on Tobacco Breeding.

There are two factors which must be reckoned with in any
system of breeding. These are heredity and environment.

Previous tobacco investigations have shown the great im-
portance of environmental conditions for both quality and
productivity. For example, Jenkins (1896) shows that on
similar land there are large variations in quality and yield due
to different systems of fertilization.

Selby and Houser (1912) have shown that the time of har-
vesting, after topping, has a great effect on both quality and
yield.

It has been stated by Frear and Hibsham (1910) that the

12 Connecticut Experiment Station, Bulletin 176.

climate of Pennsylvania has a much greater effect on the char-
acter of tobacco produced than either hereditary varietal differ-
ences or soil.

It is a well-known fact that tobacco harvested by the priming
method (picking individual leaves) has a different character
than when harvested by cutting the whole stalk. These few
illustrations, while in no way complete, indicate the great
importance of the environmental factor in tobacco breeding.

One of the earliest experiments on inheritance of tobacco
characters ever recorded was made by Naudin (Focke, 1881).
This careful experimenter crossed one variety which had lan-
ceolate leaves with a type which produced broadly oval leaves.
The plants, resulting from this cross were alike in all essential
features. In the second generation the differences were more
marked and many individuals were found which resembled the
parent types. Godron received two types of these hybrid
forms from Naudin, the one with small- leaves and the other
with broad leaves. Both forms bred true in later generations.

Since the year 1900 many attempts have been made to improve
the present types of tobacco by selection and crossbreeding.
Shamel and his co-workers have done an important work by
pointing out the value of selecting good type individuals for
seed plants, and the production of inbred seed by bagging the
seed head. Such methods have accomplished much by tending
to produce uniform and better races.

In regard to the benefits which may be obtained from hybrid-
ization and subsequent selection, our knowledge is very meagre.
On this subject Shamel and Cobey (1906) say:

"The best plan which can be followed in the case of crosses
is to grow 100 plants of each cross and carefully note the char-
acteristics of the hybrid plants. It will be found that there
will be considerable variation in the plants the first season.
Seed should be saved from those plants which are the most
desirable and which show the greatest improvement over the
native varieties. The next season a larger area can be planted
from this seed; and if the crop is uniformly of the type desired,
enough seed can then be selected the second season, to plant
the entire crop the third year."

This quotation certainly shows a lack of belief in the uni-
formity of the first hybrid generation, and on the other hand, no
conception of segregation in F2.

Previous Work on Tobacco Breeding. 13

Shamel (1910) also says:

"The writer believes that the two efficient means of inducing
variability as a source of new types are change of environment
and crossing. So far as the writer is concerned, the change
of environment — usually the growing of southern grown seed
in the north — is the most effective means of inducing varia-
bility."

Hasselbring (1912), however, gives experimental evidence from
a number of pure lines of tobacco which he grew both in Cuba
and in Michigan, and comes to the conclusion that there is no
breaking up in type due to changes of environment, and that
whatever changes take place affect all individuals of a strain
in a similar manner.

Some observations of the writers on the appearance of several
types grown in the Connecticut Valley from foreign seed serve
to corroborate Hasselbring's conclusions.

These few citations from previous investigators show that
there is no very definite knowledge of the manner of inheri-
tance of tobacco characters, and the writers hope that the
present paper may clear up some of the more important phases
of this subject.

The Material Used.

Four different types of commercial tobaccos furnished the
starting point for these investigations. They consisted of two
imported varieties tested for shade purposes, which prior to
1908, had been grown for a number of years in row selections
from selfed seed, and the two standard Connecticut types —
Broadleaf and Havana — which have been grown in Connecti-
cut since the early history of the tobacco industry. The follow-
ing descriptions give some of the more important features of
these types.

No. 401 Broadleaf.

The Broadleaf variety produces long, pointed, drooping
leaves, averaging in length a little over twice the breadth, with
an average leaf area of about 9 sq. dcms. The ntmiber of leaves
per plant ranges from 16 to 23 and averages from 19 to 20,
The average height of plant is about 56 inches. This variety
sells for slightly more per pound than the Havana, and when

14 Connecticut Experiment Station, Bulletin 176.

used as a wrapper or binder is generally considered to give a
little better flavor to a cigar than the Havana type.

No. 402 Havana.

Havana produces medium length leaves, standing nearly
erect though drooping slightly at the tip. The average length
of the leaves is a little over twice the breadth. The number
of leaves per plant ranges from 16 to 25 and averages from
19 to 20. The average height of the plant is about the same
as the Broadleaf. This variety is well known as a wrapper
and binder tobacco.

No. 403 Sumatra.

This variety produces short, round pointed, erect leaves, a
little over half as broad as long, with an average leaf area of
about 3 sq. dcms. The upper leaves of this type are generally
narrow and pointed. The niimber of leaves ranges from 21
to 32 and averages from 26 to 28. The average height, when
grown under shade, is about 63^ feet. This variety produces
a larger percentage of wrappers than the Cuban type but the
quality is very inferior, being of a light, papery texture.

No. 405 Cuban.

The leaf of this variety averages about the same width as the
Havana, but is shorter and rounder. The position of the leaves
is nearly erect. The leaf munber ranges from 16 to 25 and
averages about 20 per plant. The leaves are somewhat larger
than those of Sumatra. This type is grown widely in the
Connecticut Valley under shade covering, and produces wrapper
tobacco of high quality.

The Methods Used.

As far as possible every precaution was taken to prevent
experimental errors. With the exception of a very few cases
the parental varieties have been grown from inbred seed, and
if, for various reasons, other seed has been used, the fact is
noted. Selfed seed has been obtained by covering the seed
head with a Manila paper bag, and crosses have been made in
the manner explained in previous papers (Hayes, 1912).

The Method. 15

Much efficient aid has been given by Mr. C. D. Hubbell of
The Connecticut Agricultural Experiment Station, who has
materially helped in taking data, shelling and filing seed, and
in the calculations. In the summer of 1912 Mr. A. F. Schulze,
of the Connecticut Agricultural College, assisted in the field
work.

We also wish to express our thanks to the Windsor Tobacco
Growers' Corporation and its manager, Mr. J. B. Stewart »
for so faithfully carrying out their part of the agreement by
which means we were enabled to obtain the accurate data
reported here.

As in previous work, each parental type has been given a
number. A cross between No. 402 Havana and 403 Sumatra
has been written (402X403), the female parent appearing
first. Later generations have been designated (402X403) — !,
(402X403) -1-1, and 403-1-2, which denote respectively
the second and third generations of a cross between Havana
and Sumatra, and the third parental generation of Sumatra.

The seedlings have been grown in sterilized soil. The steri-
lization of the beds has been accomplished by the use of steam
at a pressure of at least 70 pounds, as explained by Hinson and
Jenkins (1910). The actual sowing of the seed has always been
done by one of the authors.

The different families and selections have been marked in the
field by heavy stakes, to which wired tree labels were attached,
and a planting plan has always been kept on file showing the
exact location of the different selections. With this brief
description of methods used, we will take up the consideration
of the residts obtained, and for convenience each family will
be discussed separately.

Family (402X403) Havana X Sumatra.

A large number of crosses between tobacco varieties were
made by Shamel in 1903, and among these was one between
Havana as female and a small-leaved Sumatra type as male.
Shamel (1905) states that the male parent, which was descended
from Florida Stmiatra seed, had been grown in Connecticut
for two seasons and was partially acclimated. The Havana
parent was a type which had been grown for a number of

16 Connecticut Experiment Station, Bulletin 176.

years by Mr. D. P. Cooley of Granby, Conn. The cross was
grown at the Cooley farm in 1904 and 1905.

According to Shamel" the first hybrid generation grew some-
what more vigorously than the parent types and was rather
■uniform in its habit of development. The second generation
was thought to be no more variable than the first. Selected
plants of this generation were grown at the farm of Edmund
Halladay in Suffield in 1906.

It was the custom of the tobacco experts of The United
States Department of Agriculture, who at this time conducted
the work of tobacco breeding in Connecticut, to select desirable
field types, harvest the leaves from each seed plant separately,
and to base their judgment on the combined data from the
growing plants and the cured leaves.

After examining the data on the F3 generation collected in this
manner, Mr. Halladay and Mr. J. B. Stewart concluded that
one particular plant, bearing 26 short, round, pointed leaves
with short internodes between them, gave great promise of
becoming a desirable commercial type. Accordingly, Mr.
Halladay added one row of plants from inbred seed of this
individual to the two acres of experimental tobacco grown
by him in pursuance of a co-operative agreement with the
Department of Agriculture.

The plants in this row, numbered 2h-29 in accordance with
the Department nomenclature, grew comparatively uniformly
and several were inbred. In Mr. Halladay's absence, however,
Mr. Shamel and an employee of Mr. Halladay's, in reducing
the niunber of seed plants saved, topped all the plants except
a late one, which was afterwards inbred.

In view of Mr. Halladay's high opinion of this type, the seed
of this plant and that remaining from its parent were used for
planting in 1908, each generation being given a separate number.

The field in 1908 presented a fairly uniform appearance and
gave promise of producing a valuable wrapper tobacco. The
new type was named "Halladay Havana," in honor of Mr.
Halladay, who, in a large measure, was responsible for its
production. It averaged about twenty-six leaves per plant
and grew to about the height of Havana. The leaves were of
medium length, averaging slightly shorter than Havana; they
were fairly uniform in shape, with somewhat rounded tips.

The Halladay Havana. 17

The crop, when cured, lacked uniformity. Some leaves of
exceptionally fine quality were produced, but the general fault
of the crop was a lack of grain and too large a proportion of
the heavy leaves known to the trade as "tops."

From this 1908 crop one hundred seed plants were saved,
the leaves of each being carefully harvested, cured and fer-
mented. Mr. J. B. Stewart and one of the writers made care-
ful notes on the quality of these individuals, especial attention
being paid to the feature known as "grain." The plants showed
great variability; some of them had produced a fairly high
grade of wrapper tobacco, others exhibited rather poor quality.

In 1909, 'seed from twelve of the best of these plants was used
to continue our own experiments, but small amounts were also
distributed to a number of Connecticut farmers. In addition,
three acres were grown in Massachusetts. Some of these
results were very promising. At the Arnold farm in Southwick,
Mass., for example, a measured acre produced 3,000 pounds
and brought the grower over $700. Other results were less
favorable, but on the whole the experiment seemed worth
repeating on a larger scale.

Accordingly, about 125 acres of Halladay Havana were grown
in the Valley the following year and, while some men sold their
crops at a good price, the resrdts, in the main, were not en-
couraging. The chief faults mentioned by the buyers were
lack of grain, too large proportion of dark and heavy leaves,
and poor burn, although, in some cases, the burn was satis-
factory.

This was the status of the work on the Havana X Sumatra
cross when the data collected previously were turned over to
the writers in 1908. Shamel, who had been in charge of the
work up to this time, had come to the conclusion that the Halla-
day type was the result of a mutation. Apparently, he did
not lend his approval to certain biological beliefs current at
this time which indicated an alternative theory as an inter-
pretation of its origin. For example, he believed that in general
there was no greater variation in the second generation of a
cross than in the first, and that considerable progress could be
made by selecting good Fi plants, some of which would breed
true and give uniform progeny in F2.

The writers did not take this view of the problem. It was

18 Connecticut Experiment Station, Bulletin 176.

contrary to all modem ideas of breeding to expect a cross be-
tween two self-fertilized varieties to be variable in Fi. High
variability should occur in F2, due to the recombination of
Mendelian factors. New types should be produced in Fj
which could be reduced to an homozygous condition by selection
and thereby fixed.

It was not impossible that the many-leaved type could have
originated by mutation, but it appeared much more probable
that it had been produced by recombination of parental char-
acters. The type had the number of leaves and leaf shape
of the Sinnatra parent, combined with the habit of growth of
Havana, and a close approach to the Havana leaf size. Other
characters were in a somewhat intermediate condition; for
example, the crinkling of the leaf was apparently a blend of
the smooth Havana leaf with the much cnunpled Sumatra leaf.

Family (403X402) SumatraX Havana.

To test the hypothesis that the Halladay is a result of the
recombination of parental characters and can be reproduced
whenever desired, a cross was made in 1910 between Siunatra
female and Havana male. The Stimatra was a direct descendant
of the type used by Shamel in 1903 and had been grown from
inbred seed for a niunber of generations. The Havana was the
commercial variety grown at the Windsor Tobacco Growers'
Corporation in Bloomfield. Although this variety of Havana
was not exactly the same as that used by Shamel, it was the
same in all essential features, the probability being very large
that both types originally came from the same source.

The data on number of leaves per plant in this cross are given
in Table I. The Sumatra and Fi generation were grown at
New Haven in 1911; the Havana was grown at Bloomfield
from commercial seed of the same type as that used for the male
parent of the cross. The Fi generation was intermediate for
leaf ntimber and leaf size and was as uniform as the parental
types. The variability of the F2 generation for leaf number,
size, shape and height of plants was very large. Some types
were produced which coiild not be distinguished from pure
Siunatra; others had Sumatra leaf characters and Havana
leaf number; others resembled Havana in all features; and

Inheritance of Leaf Number.

19

^5

t^OO -H CO

C^ C<] T}H 05

>

^ M -h' -h'

o

CO CO CD O

1

QOr-<THQO

Oi-ii-;o

<!

19.8 ±
26.5 ±

23.3 ±

24.4 ±

"S

OiO (NO

lO (M »O00

0

1— 1 T-i ■■+1

1—1

•i-( -(N

CO

O

^

t^

CO

05

LO

00

<M

1— '

1— '

00

CO

lO

(N

1-1 -(M

1>

r-l 1-1 05

(N

(M Th

CD

»0(M05

(N

(M »0

CO

lO

t^iO ^

CO

(M

■(M t^

-*

1-1 CO >o 1—1

(N

T-l 1-H>.

"3

CO

coco I* CO

(N

1—1 CO

(N

O -0500

<M

1-'

TfH

1-1

(M

Tt<CO

(N

(M

CO

O

(M

(M^

<N

-*

(M

Oi

TfH

•OS

rH

'^

00

iM

1—1

7—1

iM

t^

CO

1—1

^

fofe

«. «..

c3 nj

C C

rt 05

> >

oi 03

c

XX

^

XX

rt S 2 S

C -iJ +j -u

03 TO cd oj

>BBB

" 3 ;3 S

20 Connecticut Experiment Station, Bulletin 176.

still others had the leaf size and growth habit of the Havana,
combined with the leaf ntimber of the Sumatra. These results,
illustrated in Plates I-IV, give conclusive evidence that the
Halladay type can be reproduced whenever desired.

Effects oj Selection on the "HavanaXSumatra" Cross.

Let us now consider the effects of three years of selection on
the Halladay strains of Shamel's cross. The purely genetic
resiilts of selecting for high and low leaf number are described
in another paper. The work is considered briefly at this point,
however, as the restdts have an important bearing on practi-
cal tobacco breeding. They show why the type lacked uni-
formity in 1908 and 1909, and hence the reason for its failure
as a commercial proposition. Further, they go far toward
indicating the proper procedure in obtaining results of economic
value after hybridization.

In brief, the method pursued in this selection experiment was
as follows :

Of the nine families with which the experiment was started
(Table H), eight were grown at the Krohn Tobacco Company,
in Bloomfield, in 1909, and the other (No. K) at a farm nearby.
These nine families were selected from the 100 seed plants of
Shamel's cross which were grown at the farm of Edmund Halla-
day, in Suf field, in 1908. From each of these families an inbred
plant was saved which bore a high leaf number, and another
with a low leaf number. These were made the basis of plus
and minus selections, which were grown the following year,
and from this time on seed plants with a high leaf niunber have
been saved from the high or plus selection, and seed plants
bearing a low leaf nmnber from the low or minus selection.

These results, given in Table II, include the selection number,
year grown, generation, number of leaves of parent, range of
variation for leaf number, total plants, and biometrical con-
stants, consisting of the mean for leaf number (A), and coeffi-
cient of variability (C. V.).

A consideration of these data shows that in one family. No.
27, no appreciable shift of the mean has been obtained, the
mean of the low selection for 1912 being 25.9='= .07, and that of
the high selection being 25.0 =1= .06.

Effects of Selection for Leaf Number.

21

<

H

<

CO

X

>
<

H

W

>■

d

CO Tfi »0 00 05 00

1— 1 C^ >— 1 i-H T-H t— 1

05 (Ml> »-H COO

■-H (M i-H (N rH CO

05IM

OSCDO^

00 CO c^ •* 1-H 05 00

i-H rH C^ rH (M ,-H 1-1

-H -H 41 -H -H 41

CO 00 TfH lO lO Tt<

T-H 1-; CO t^ o CO

IC InI UO lO CO CD

41 41 41 41 41 41

lO t^ CO CO t^ CO

CD O --J r-; i-H C35

CD CO CO CO -^ oi

41 41

cot^
coco

41 41 41 41

CO IMi-H (M
lO 1— ( CD lO
CO 00 lO CO

41 41 41 41 41 41 41

CO <-l (M •* 1> 00 00
!>; O CO O TJH I> CD

CO lO !>• lO CO lO CO

<i

CO 05 CO 00 00 ir^
oooooo

t^OOCD Oi lOCO
OOOOO'-i

00 05

oo

OOl>00

.-H i-HOO

t^ CO 00 CD t^ l>. CO
OOOOOOO

•H 41 41 -H •« 41

»ocD.-;c^coo

CD CD 00 05 00 05

41 41 41 41 41 41

T-4 (M 00 1> 1— 1 lO

>o>o loooio

(M (M (MCO(MCO

41 41

T-II>.

OJOO

41 41 41 41

»Oi-H00O
TJHCOCDCisi

41 41 41 41 41 41 41

OiCOiCOOC^ o o
lO CD CO 1> lO CO »0
C<1 <M C<) (M (N (N (N

3

o

05 T-H Tt< CO CO C<J

CO i-H 05 (M •* 00

lo ^ t^ »o 05 CO

COOOOOCOTt^
(M 1-H CO (M (M (M

CD O CO CO lO 00 i-H

05 >— 1 1— 1 1— 1 00 CO

CO Tf< Tf< O (N <N »0
■-I tH lO O (N <M O
CO CSI IM CO (M iM CO

(M CO CO '^JH Tt< Tf<

CO CO CO CO CO CO

1 1 1 1 1 1

CO --I CO lo T}H CO

(N (M (N 05 (M (M

(MCOi-HCDTfiCD iOTtH.-iOO(M(NCD

CO CO CO COCO CO CO CO CO (M CO CO CO
1 1 1 1 1 1 ! 1 i 1 1 1 1 1

'Hr-l(MlOCDOOThTt<d5T-lr-ICOCO

(MlM(M<M(Mi-l <M(M(MiM(M(M(M

7— 1 1— ( ^H 1-H 05 05 (N
CO CO CO CO <M (M CO
J 1 1 1 1 1 1

CD (M i-H (N ^ rH ,-H

(N (N (M (M (N (M (M

Tt< LO 00 --H IM CO

(M (M (NCOCOCO

COiOOOO (M(M
<M C<) (M CO CO CO

05 05 CO^OOOO
(M iM Cq (M (M CO CO

csi c^ ko 05 00 00 00

(N C<1(M (M (M (M <M

0

m

a

fl^fefep^fep^

fe fo Ph pL, Ph (i|

pH fe Ph fe Ph Ph plH

p^H fc p:< fe Ph Ph Ph

1

o

U

u
C4
d)

tH0050i-HC<1

1— 1 1— 1 O 1— t T-H tH
05 05 05 05 05 05

.-I0050'-I(M

1— 1 1— 1 O T-H i-H ,— 1
05 05 05 OS 05 05

(M r-l O 02 O I-H <N
1— 1 t-l T-H O I— 1 1— 1 1— 1

05 05 05 05 05 05 05

(M rH O 05 O i-H (N
1— 1 T— 1 rH O 1— 1 1— 1 T-l
05 05 05 05 05 05 05

6

(5-l)-l

(5-1)

5

(5-2)

(5-2)-l

(5-2)-l-3

(6-l)-2

(6-1)

6

(6-2)

(6-2)-l

(6-2)-l-4

I-H 1— 1 1— 1 7— 1

II II

t-H 1— 1 i-H 1— 1
11 II

(12-1
(12-1
(12-1
12

(12-2
(12-2
(12-2

III Mi;

22

Connecticut Experiment Station, Bulletin 176.

00 Ol T-H O CO Cl T— I >

COOliOiOiMC<)TtHl'*i(MTtl^(M00

i(M(M(Mr0^(MCCil(M'-<iMT-i(M(M00;T-i»OCO(M(Ni

<Ml-0>OlOl^--HOOTtH

-H -H •« -« -H -H -H

(M ^ •* Oi 00 t^ 00

r-^ O CO CO lO CO rH

-h' -h' -h' -h' -h' i

00 t- TtH t^ GO oo
(MOCOIMGOC^

-H" -H* -h'

CO 0(M

CDiOCOI>t^CiC0t^'— lit^iOt^iOCDCOO lOCDCOCOOCD

i>i>ooi^oiTtiGoooQO oot^05CDooaicol'OC5'-ii>.ooi>. i>iooo
OOOOOi-HiMOO O00000r-(,0'-H--H000 ooo

-H-H-H-li4l-H-H-H4|-H-|]-H4i-H-tt4l -H-U-l]-f|-H-H|-tl-tl-H

rJHi— ICOOlO-tOOOOSOOt^CDCOOSCOtMt^ 03C^i— it^Ii— lOsirflCOOl

OOiOTtHCDTt((McDOQO|'-Ht^COO'-iiOCO OrHOlOSOkO CDCOCD
00(N(Ma3CO'^(N^l> C^^^hOOC»1> ^coooocoi>- COOCD
iM<MiM(M<Mi-<'-iCOC^(MC^(MCO(M'-H CO (NiM(M(MCO(M

bo

n1

«

>

I r r r r r r i ^ri r r r r r r i i i i i i i i i i i r

PnfoflHfopI-lpHpHfefe

fo fo fo ft, plH fe Ph

pL| pL, pL| Ph plH plH

fl-lpHfl-l

(M'-H0050^'-i(M(M (Ml— i0050'-H(M i— lOOiOi— iiM

T— (i-Hi-HOT-HrHi-HT— It— It— It-Ht-HOt— li-Hi— t T-HrHOT-^t— 1»— I

O^CiG50iGiCiO^C50^ O^ Oi <^ <^ Oi Oi Oi Oi Oi O^ O^ O^ O^

(Mt-< o

1— I T— I rH
05 05 05

1-1 CO 1-H 00 (M (N

(M(M (M (M (M
I I I I I

CO

I

COCO

rH .-f 1-1 (M (M<M ;

Effects of Selection for Leaf Number.

23

>
6

COOO^
CO CO (M (M

CO (M t^OCOO CO
(M 05 (M ^ CO iM (M

-H -H -H -H

CO(M (M 00

COCO oi^
coaJ cDi>

41 41 41 41 41 41 41

l>- (N CO (M 00 CO >o
t- "O 03 CO OJ O i-H
1> CD CO -^ 1> CD 1>

<

l-H r-IW 00
1— 1 T— 1 o O

00 !> 05 CO rH 1:^ t^
O O O 1-1 .-H o o

41 41 fl 41

T^^Oi COOO

cqrH coio

41 41 4] 4i 41 41 41

lO lO -^ CO 00 O 00
^ CO ^ <M cq T)H (M
(M Oq (M (M (M (M (M

o

»0 -* .-H O

GO (MOO 00
(M(M{M

05 0-1 GO 1-1 O ^ CO
OO^ CO coo 05
Cq(M^ ^ (M05

0 §
bo rt

t^lMCOCO

(M CO coco

OO (M 1-1
(M (M(M<M

00 Oi C5 ^ C31 CI O
<N (M (M 05 Cd 05 CO
1 1 1 1 1 1 1

1> O --1 O C3 t^ GO

T-H C5 05 05 rH ,-1 rH

Leaves of
Parent.

CO t^O o
(M (M coco

TfH 05 -co • CO CD
05 05 -05 ■ 05 05

d
o

la
S
s
o

pLHfefLH'fi,

fe fo Ph pC fe fe fo

o

1..

ai

OOr-KM (MrHOOO'^fM
Or-^r-H,-H ^r-(^OrH.^,-l

6
J5

77

(77-2)
(77-2)-l
(77 -2) -1

(K-l)-l-2

(K-l)-l

(K-1)

K

(K-2)

(K-2)-l

(K-2)-l-6

24 Connecticut Experiment Station, Bulletin 176.

All other plus selections except (73 — 2)— 3 — 3 and (K — 2)
— 1 — 6 have given a change toward the high leaf condition.
These selections gave about the same average leaf nvunber
as in 1909. In some strains the mean has been gradually
shifted, as in the plus selection of family 76, which gave pro-
gressive changes from a mean leaf value in Fs of 24.1 ±.11 to
24.4±.07 in Fe, then to 26.1 ±.08 in Ft, and finally in Fg
to 26. 9 ±.07. Other families, as Nos. 5 and 6, gave a large
change in mean due to the first year of selection but in later
generations have given no further changes due to continued
selection. In general, the results have been what one would
expect if selection simply isolated homozygous types from a
heterozygous population.

Selection for low leaf number has caused decreases in (5 — 1) — 1,
(K— l)-l-2 and (77-1) — 1-2, and slight decreases in
(6-l)-2, (73-l)-2-l and (76-l)-l, but of such a small
nature that little dependence can be placed upon them. A
negative effect is shown in case (41 — 1)— 2.

In previous papers we have shown that the number of leaves
per plant is a very stable character and, as such, little affected
by environment. That selection has made various degrees
of change in the mean of some types and no change in others,
we believe to be due to the fact that some selections, as for
example No. 27, were in a pure or nearly homozygous condition
in 1909, while others were heterozygous for different numbers
of factors for leaf number.

General field notes on the Halladay types, which were grown
in 1912, are given in Table III. Three different observations
on these types were made : general vigor, shape of leaf, and
leaf character, whether smooth or crinkled. Of the fourteen
selections given in this table, three were classed as very vigorous,
seven as having good vigor, three as of fair vigor, and one as
non-vigorous. As to shape, eleven have broad round tipped
leaves, one has broad leaves with a pointed tip, and two from
family No. 77 have leaves which resemble the Havana in shape.
Considering fullness between the veins, one selection has very
crinkled leaves, eight have crinkled leaves, two have slightly
crinkled leaves, and three are classed as smooth-leaved types.

Leaf Characters of Halladay Havana.

25

It is of interest to know that the leaf character and also the
length of internodes of No. 77 closely approach the type of the
Havana parent.

TABLE III.

General Characteristics of Selected Halladay Strains
Grown at Bloomfield in 1912.

No.

General
Vigor

Shape of Leaf

Type of Leaf

5-2-1-3

Very good

Broad, round tip

Slightly crinkled

6-2-1-4

Good

" pointed"

11 11

12-1-1

Very good

" round "

Smooth

12-2-1

a u

(f It 11

Very crinkled

27-1-1

Good

a u li

Crinkled

27-2-1

u

li u a

u

41-1-2

Fair

U li u

a

41-2-1

((

u u u

li

73-2-3-3

Good

U li 11

u

76-2-1-1

«

11 « li

li

77-1-1-2

Fair

Fairly broad, pointed tip

Smooth

77-2-1

Good

(I 11 li »

11

K-1-1-2

Poor

Broad, round tip

Crinkled

K-2-1-6

Good

li 11 11

11

Some data obtained on comparative leaf length of these
Halladay types are given in Table IV. This table gives the
average number of leaves per plant, by actual count, the total
yield of cured tobacco on an acre basis, and the number of pounds
of tobacco in each leaf length class. This, of course, does not
give the number of leaves of each length, as it naturally takes
more 12-inch leaves than 20-inch leaves to weigh a pound.
However, a general idea of the average length of leaves of a
selection can be obtained by this means.

This table shows that leaf length is not very closely correlated
with number of leaves per plant. For example, selection
(73 — 2)— 3 — 3, which averaged 26.7 leaves per plant, produced
only 256 pounds of 18-inch tobacco, while selection (12 — 1) — 1,
which, averaged 29.1 leaves per plant, produced 1,162 pounds
of 18-inch tobacco. (K— 1) — 1— 2, which averaged 21.5
leave , produced only 113 pounds of 20-inch length, while
(K— 2) — 1 — 6, which originally, in 1908, came from the same
plant as (K— 1) — 1 — 2, and which averaged 22.8 leaves per
plant, gave a production of 944 pounds of 20-inch length.

26

Connecticut Experiment Station, Bulletin 176.

TABLE IV.
Comparative Length of Leaves of the Halladay Strains in 1912.

Yield in Pounds for Leaf Length Classes

Average

Yield

in Inches

No.

No. of
Leaves

per
Acre

per Plant

12

13

14

15

16

17

18

20

5-2-1-3

29.0

2813

87

126

199

349

538

709

730

75

6-2-1-4

30.5

2822

82

109

181

245

402

588

872

343

12-1-1

29.1

3370

38

131

142

254

363

812

1162

467

12-2-1

29.0

3085

143

152

253

423

629

755

677

52

27-1-1

25.9

2766

86

138

146

300

483

628

833

150

27-2-1

25.0

2736

95

65

93

180

356

495

909

543

41-1-2

27.4

2196

72

93

125

175

271

430

800

234

41-2-1

26.9

2694

101

112

160

220

351

523

971

257

41-2-3-2

17.-8

1936

115

122

199

263

323

355

462

97

73-2-3-3

26.7

2645

229

229

379

512

617

423

256

76-2-1-1

26.9

2721

126

119

204

356

566

672

634

'44

77-1-1-2

18.4

2271

35

40

97

137

185

316

638

823

77-2-1

25.8

2341

64

47

88

138

219

359

942

483

K-1-1-2

21.5

2332

84

65

142

239

343

524

822

113

K-2-1-6

22.8

2740

62

49

86

216

210

392

781

944

Havana

*20.0

2119

56

57

86

128

191

284

570

747

*Estimated.

The largest amount of tobacco by weight was produced in
the 18-inch class by ten of the selections, in the 17-inch class
by two, in the 16-inch class by one, and in the 20-inch class by
two selections. The Havana grown for comparison also pro-
duced the greatest amount of tobacco in the 20-inch class.

Quality of Cured Leaves.

The data already submitted have shown that by 1912 several
types markedly different in leaf number have been produced.
Though it is less easy to demonstrate by concrete figures, these
types also differ in vigor, shape of leaf, plant height, etc. This
fact is of practical importance and gives conclusive evidence
for believing that the Halladay type, as grown commercially
in 1908-1910, was not the uniform type which it was, in general,
considered to be. May not these facts explain the reason for
the commercial failure of the Halladay by showing that the

Quality of Cured Leaves. 27

type, as a whole, was in a heterozygous condition and, therefore,
could not give tobacco uniform in quality. That some growers
were favorably impressed and others less so may then be en-
tirely due to the fact that some grew favorable types, and others
types which, from a commercial standpoint, were very inferior.
It was for this reason, justifiable from the commercial point
of view, that the culture of the Halladay was dropped.

From 1909 to 1911 inclusive, no data were taken on the cured
leaf of the Halladay, as our sole aim was to study the effects
of selection on the field habit. In 1912, however, the tobacco
was harvested, cured, fermented, and assorted, to determine
if certain selections had come to be better than the others and
if any gave promise of commercial value. Because the season
of 1912 was a dry one and not very favorable for tobacco, the
crop, as a whole, was of inferior quality. A small plot of com-
mercial Havana of the same type as that grown by the Windsor
Tobacco Growers' Corporation was grown on the same field,
however, and was cured, fermented, and assorted in the same
manner as the experimental tobacco. By this method we
were able to obtain some idea of the comparative value of our
selections, using Havana as the standard.

However, it should be noted that on account of practical
difficulties the time of harvesting the various pickings was
not always at the proper degree of ripeness. For example,
the first and third pickings should probably have been made a few
days earlier, but for unavoidable reasons this was impossible.
Further, some selections were a few days earlier in maturity
than others, and as all selections were harvested on the same
day, some may have received more favorable treatment. This
was partly corrected by making a larger picking, that is, by
taking more leaves from the very mature types at an early
picking than were taken from the later maturing types at the
same picking.

The method of harvesting tobacco by the "priming" method
is well known (see Stewart, 1908) and will be mentioned only
briefly here. Four pickings were made of our experimental
tobacco, as follows: About 5 leaves were harvested at the
first picking, 5 to 8 at the second picking, 7 to 12 at the third
picking, and all remaining leaves of commercial size at the last
picking. The leaves of each picking were then tagged with the

28 Connecticut Experiment Station, Bulletin 176.

selection number and carried to the barn, where they were
strung and hung on laths, from 36 to 40 leaves to the lath, with
a tag containing the selection number attached to each lath.

The curing season was somewhat wet and at two different
times it was necessary to dry out the tobacco by firing, which
was accomplished by building charcoal fires in small stoves.

After the tobacco was cured it was taken down when in "kase,"
that is, when just damp enough to be pressed in the hands with-
out breaking the leaves. The leaves from each lath, with tag
attached, were tied into hands, and the tobacco then placed
in a "bulk" to go through a period of fermentation. The
experimental tobacco was not fermented sufficiently for com-
mercial use, but the fermentation tended to even up the colors
so that the tobacco could be assorted with better judgment.

After the tobacco had remained in the bulk for about four
weeks it was removed and all of each selection placed together,
the different pickings being kept separate. Four hands of
the first three pickings of the different selections were drawn
at random and were examined for quality by three tobacco
judges. The same hands were carefully examined by the writers
for "grain" and "texture."

The total crop of tobacco was then sized by the usual method.
This consists in separating the leaves into different lengths,
from 12 to 20-inch classes being made. This work was done
by girls under our supervision.

After the tobacco was sized it was assorted into grades as in
comrnercial practice. The actual work of assorting was done
by experienced sorters, and the different lengths and grades were
weighed in pounds and ounces.

''Grain'' in Tobacco Leaves.

The presence of small pimple-like projections scattered over
the cured leaf of tobacco is called "grain." It is a well-known
fact that all tobacco does not exhibit this tendency in the same
degree. In some cases the grain is large and easily seen, and
in other cases small and scarcely visible to the naked eye.

One of the tobacco experts who kindly examined our Halladay
selections made the criticism that the "grain" was over-devel-

Grain In Tobacco Leaves. 29

oped, and another expert expressed the opinion that the selec-
tions, as a whole, were lacking in grain. This fact is mentioned
to show that the ideals of some of the best growers differ on
this matter. Both men desired grain in the leaves, but one
preferred large pimply grains, easily seen, and the other a fine
grain, scarcely distinguishable.

Sturgis (1899) found by microscopical examination that the
grain of tobacco leaves was due to a crystalline deposit of some
material, the compound being, in his opinion, calcium oxalate.
Contrary to expectations, he found no increased deposit due
to heavy liming of the soil but he did find that the thinner
leaves which were produced under shade apparently contained
it in smaller amounts.

If grain is calcium oxalate and as such of no value for burning
qualities, it is very probable that it does not deserve the impor-
tance that it generally receives, although, as Connecticut growers
generally consider the presence of grain to be an indication of
quality and as tobacco buyers as a rule make it a factor in their
judgment of the crop, it becomes necessary to consider its
production. From the writers' standpoint a fine-grained
wrapper leaf presents a more handsome appearance than leaf
with larger grains, although the final test of any quality depends
upon the demand of the consumer.

As has already been mentioned, some of the parent plants i
of our 1909 selections were examined for grain because it was
believed that the Halladay Havana, as a whole, lacked in this
particular. We have therefore considered this character in
our experimental work in 1912.

Before the tobacco was sized and after fermentation had
taken place, four hands containing approximately forty leaves
each were drawn at random from the first three pickings of
each selection and were examined for grain. The method
followed was an arbitrary one. Seven general classes were
made ; those leaves which had a maximum amount of grain
were placed in Class 1, and those in which no grain could be
distinguished were placed in Class 7. Obviously the remaining
classes ranged in value from maximtim to minimum grain
production. The results are given in Table V.

30

Connecticut Experiment Station, Bulletin 176.

TABLE V.
Variation in Grain of Halladay Strains in 1912.

Grain Classes

No

Leaves

Picking

Mean

Class

Plant

1

2

3

4 5

6 7

5-2-1-3

30.5

1

15

20

45

46 19

3.23

From good

2

2

7

27

49 22

4 .

3.85

grained plant

3

1

4

15

26 67

37 .

4.77

in 1908

Total

18

31

87

121 108

41 .

3.97

6-2-1-4

29.1

1

16

28

52

27 23

3.09

From fair

2

6

25

27

68 22

3.50

grained plant

3

2

6

21

46 54

2i .

4.38

in 1908

Total

24

59

100

141 99

21 .

3.66

12-1-1

29.1

1

2

5

24

59 40

16 .

4.21

2

5

18

31

54 37

10 .

3.84

3

2

7

30 63

52 2

5.04

Total

7

25

62

143 140

78 2

4.37

12-2-1

29.0

1

13

24

41

33 18

2 .

3.19

2

7

22

46

56 16

3.35

3

1

11

25

13 63

27 .

4.48

Total

21

57

112

102 97

29 .

3.68

27-1-1

25.9

1

4

18

23

49 39

9

1

3.92

From fair

2

5

20

52 47

23 .

4.43

grained plant

3

10

40 64

41 .

4.88

in 1908

Total

4

23

53

141 150

73

i

4.42

27-2-1

25.0

1

10

18

48

47 25

3.40

as 27-1-1

2

5

13

53

61 16

3.47

3

1

8

28

68 36

is .

4.16

Total

16

39

129

176 77

18 .

3.69

41-1-2

27.4

1

18

27

47 34

19

1

4.08

From good

2

3

15

30

52 38

9 .

3.91

grained plant

3

in 1908

Total

~8"

T7"^

41-2-1

26.9

1

31

55 27

4 .

3.62

as 41-1-2

2

8

19

33

57 30

4 .

3.54

3

4

41 92

15 .

4.78

Total

16

36

68

153 149

23 .

4.02

73-2-3-3

26.7

1

4

11

22

28 21

6 .

3.75

From good

2

8

15

36

49 18

9 .

3.60

grained plant

3

2

16

45 60

31 .

4.66

in 1908

Total

i2

28

74

122 99

46 .

4.07

76-2-1-1

26.9

1

6

26

43

44 54

20 .

3.90

2

5

15

31

46 42

10 .

3.91

3

1

25

60 46

20 .

4.39

Total

11

42

99

150 142

50 .

4.05

Grain In Tobacco Leaves.
TABLE V — Continued.

31

No.

Leaves

per
Plant

Picking

Grain Classes

Mean
Class

12 3 4 5 6 7

77-1-1-2

18.4

1
2
3

Total

8 28 52 51 13 . . .

13 34 41 43 4 . . .

6 40 74 23 . . .

21 68 133 168 40 . . .

3.22
2.93

3.87
3.32

77-2-1

25.8

1
2
3

Total

25 36 37 40 4 . . .

5 20 52 57 19 . . .

.... 4 37 64 10 .

30 56 93 134 87 10 .

3.44
3.42
4.70
3.54

K-1-1-2

21.5

1
2
3

Total

13 31 42 42 16 . . .
9 15 35 54 23 9 .

26 65 59

22 46 77 122 104 68

2
2

3.12
3.65
5.24
4.02

K-2-1-6

22.8

1
2
3

Total

4 17 38 45 33 3 .

5 12 44 35 15 3 .
1 11 35 69 36 7 .

10 40 117 149 84 13 .

3.68
3.46
3.94
3.72

Havana

20.0

1
2
3

Total

36 37 37 24 10 1 .
36 53 37 17 2 . . .

8 29 29 37 28 12
80 119 103 78 40 23

4
4

2.57

2.28
3.68
2.92

82-2-1

From poor

grained plant

in- 1908

26.7

1
2
3

Total

6 19 55 52 15 .
3 20 41 51 32 '
.... 2 21 70 58
9 41 117 173 105

i

5

6

4.35
4.62

5.28
4.76

A consideration of this table brings some interesting facts
to light. It will be seen that in general there is less grain in
the upper leaves — that is, the later pickings — than in the lower
leaves. On comparing the results obtained from the experi-
mental selections with the Havana selection grown on the same
field, we observe that although the Havana was variable in
this character it had a larger amount of grain than the other
selections. This, however, we know is due to the fact that
each individual "grain" of the Havana was larger than in the
other selections, our classes representing total grain production
and not closeness of grain.

In the first column of the table, under the selection numbers,
the "grain" condition of the 1908 ancestral parent plant is
given when known. Of the sixteen selections given in the table
only eight can be considered under this head, and in one of the
eight no third picking was examined, so only seven cases remain
for discussion. Of these seven, three descended from plants

32 Connecticut Experiment Station, Bulletin 176.

classed as having good grain, three from fair-grained plants,
and one from a poor-grained plant. Those descending from
good-grained plants have means of 4.02, 4.07 and 3.97; those
from fair-grained plants have means of 3.66, 4.42 and 3.69;
and the selection descending from the poor-grained plant has
a mean of 4.76.

Of course it woiild not be fair to lay very much stress on thes^
results, it being probable that all tobacco has the ability to
produce some grain. Our results simply indicate that some
types, under favorable conditions, produce more grain than
others. As such is the case, it seems only fair to conclude that
different degrees of grain production are inherited.

Texture Observations.

The same leaves which were examined for grain were also
classed as to texture. In this work grain received no weight,
and the following brief descriptions give an idea of the character-
istics of each class.

Class I — Included those leaves having a dry nature, lacking
in oils and gums, with a body so thick as to render it
too heavy for the best wrapper leaf.

Class II — Included those leaves of a semi-dry nature,
apparently having no more oil than those of Class I,
but more gum. The body stiff but sufficiently elastic
as to allow its use for wrapper purposes.

Class III — Included those leaves most desirable for wrapper
purposes, the oils and gums being present in sufficient
quantity and accompanying a medium body, resulting
in a leaf of good elasticity, soft but firm handling qualities.

Class IV — Included those leaves of medium body and
the gum content, but with excessive amount of oils,
giving the leaf a coarse appearance with a tendency to a
"rubbery" nature.

Class V — Included those leaves of excessive oil and gum
content with a medium to heavy body, resulting in a
texture of a decided "rubbery" nature.
Of the classes here given Class III is most desirable from a
wrapper standpoint and Classes I and V least desirable.

The results given in Table VI show that many of the selec-
tions have a much greater percentage of leaves in Class III
than Havana, while other selections have a smaller percentage
of leaves of good texture than Havana.

Texture of Leaves.

33

<!
H

<

a

<

>^ <

m o

H

y,
w

^ TO

o >

OSOOCDOSi— lCOOiOOJlOOTt<CO'*'<*l

t^t^T-lr}<COTtH-*'*l«O00'— H^i-HCOi-H

t^OiiOi— iTtHCOCSOii— li— I00O5CSCCIO

1— IrHC^i— It— ll— li— (I— It— ICSIi— li— (1— It— li— I

kC(Mait^O00<M(MO00e0CDT-i0000

T-H(M(MCqcOiOCOiOCOC55C^O'^'£>'*
0(MCO»00500>0001>-CDOOI:^T— lOOO

(MIMt— lOq T— { T-lT— lT-lr-(i-li— It— It— I

OicoooioiocooikOi>corOT-<iooo

•Tfl -T-l -co • -(M • -T-i -Oi

•00 -lO • -(M • -t^ -to

■(M -CD •!> • -O • -T-l -CD

•00 -T— I . •T— I . -00 •(M

CO '

I (M

I I I

fOTtI

I I I i I I I g

iiMT-i(MT-i(M(M(MT-i(M^ljq c3

77 I I I I I I I I I I 1 I >

I I ^ M ^

a

iOcDT-iT-i(NC<l'*-*l>lr^l>l>)

34 Connecticut Experiment Station, Bulletin 176.

These data were taken in such a manner that any possible
correlation with the grain classes of the previous discussion
could be determined, and while no correlation coefficients have
been figured we feel justified in concluding from inspection that
there is no correlation between grain and the characters here
discussed.

While there was no great difference between the selections in
texture, there is no question but that some selections were better
than others, and several of them gave a somewhat larger per-
centage of better leaves than the Havana.

Results of Sorting -Test.

The results of the actual sorting test are given in Table VIII.
For convenience they are calculated to an acre basis, since by
this means one can easily compare the value of one selection
with another. During the actual sorting, the various lengths
of each picking were kept separate, but for convenience they
are grouped in the table.

The tobacco was sorted into five different grades: Light
Wrappers, Medium Wrappers, Dark Wrappers, Binders and
Tops. The Light Wrappers comprise those leaves which have
a light even color and thin texture with good body and good
vein. Medirmi Wrappers are a little darker and heavier than
the Light Wrappers but must also have good texture and vein.
Dark Wrappers are heavier than Medium Wrappers and of a
darker color. A great many leaves, which under ordinary
circumstances would have been classed as Mediums, are placed
in the Dark Wrapper class because of white veins. Binders are
thin leaves which are either off -colored, have white veins, or
have a tear in them, such faults not permitting them to be
graded as Light Wrappers. Tops are heavy, dark, oily leaves.

Table VII gives the prices used in computing the comparative
values. These figures were obtained by consulting tobacco
men who handled primed sun-grown tobacco in 1911 and 1912,
and taking the averages of the prices so obtained. These
prices refer to the packed value after fermentation.

The computations for actual packed value were made as
follows: First, the yield per acre for a perfect stand of plants
was calculated from the healthy plants in a measured row.

Results of Sorting Test.

35

Second, the total amount and percentage of each grade was
figured to this basis by utihzing the actual sorting data. It was
then assumed that these grades could be sold at the prices
quoted in Table VII.

TABLE VIL
Prices Per Pound Used in Computing Values.

Grade

Prices per Pound for Leaf Lengths and Grades

12 in.

13-14 in.

15-20 in.

Light wrappers

Medium "

Seconds

Dark wrappers

Tops

20 cents
10 "

8 "
8 "
5 "

30 cents
18 "
10 "
10 "

7 "

80 cents
50 "
22 "
25 "
12 "

Deductions were made for harvesting an extra number of
leaves, as many of the selections produced a larger number of
eaves per plant than Havana. These deductions were made as
follows :

Taking an actual case, for example (5 — 2) — 1—3 averages
29 leaves per plant, by count, and our standard Havana averages
about 20 leaves. If we assume that all leaves have an equal
weight, 9/29 of 2,813 pounds of tobacco, or 873 pounds must be
handled because of the nine extra leaves. One of our best-
known growers said that it actually cost him 28 cents per pound
to put primed Havana into bales. Thus, the extra cost of
handling nine leaves, after growing, and fertilizing the land,
would be about 20 cents a pound, and for 873 pounds would
amount to $174.60.

If we take the Havana, which averages about 20 leaves per
plant, as the standard, and compare its relative value with that
of (5 — 2) — 1—3, we must first deduct $174.60 from the packed
value of (5 — 2) — 1—3. Assuming the value of Havana as
100, we can then obtain relative values of our other selections
by dividing their packed value, after deducting the extra cost
for larger leaf number, by the calculated packed value of Havana.
Relative values so computed appear in the last column of Table
VIII.

36

Connecticut Experiment Station, Bulletin 176.

> <u

(D>.

eO'-ICOI>i-HCOCOOOCOiM'-il>t^lCICOO

00t^t^TtHlOl>C0IM>-HT)H'*Tt1i-H^(MO

OCOTtHOi-nOOOCOT-Ht^T-iT^T-ICOCOO

ft

TjHOi-Ht^(M(Mi01>l>005rtHI>iOiOO

co»ocoicr^oo^coooo5i-H(Mcocoo5>o

lMC^(MCO(Mi-iOO(MOO'*COCOCO(Mi-HCO

c
n
o

Ph

OOT^^O(^J(Mt^COt^OcOCO^COOGOTJ^O
lOOOOOiOOlCiCOLOClCOCOOO'—COTjH
COI>00i-H>^COCOl>C<)00t>t^COiOI>

OS ft

4J

d
m
o

PL,

(MCO00i-i(MC3Cq'*»Ot^C000Tt<COO500

iOt^t^COCOOCOi0005COi-H05CDCOGO

'd
a
n
o

a,

oi>iLO>ococoo5iococoTt<Tt<a>^t>iM
oa)coo-^i>o^oiOT-H03ooc<iror-i

O'*a300CO>OOTfi(M(Mt^^C0C0i>-C0

■a
.S
m

<u

111

lClOiO-*OOiOCO"^COCO»OOi(M(Ml>00

»0t^C0iOC0(M(MGC'-iC5O(MC0C0Q0(N
(M<MC^(MCOCOC0C0COiM(MC<)^<M(MO5

o

G0r-(C0iOiO05O'*'(Mi01>^OOt>i0
'-Ht^OlOOCOOO'-iCO'-HCOiCKMOOTtiOOOO

t^t^(X)t^aiGoi>ocoi>ioioco":n>nH
1—1

3 ft

4J

V

o

<u

PL,

c<j'*icoo30ocoooioocot-i^05coc;

COOi<MOOCOCC|(MOOrt<T-H,-H,-i(MCO

"2 CO CD I> OO (N O -* T^^ Ci O <r> O 05 lO iC o
3 1> CO 00 (M (M O Tti lO (M T}< CO TJH CO 00

O t-H (M i-H r-f
Oh

S ft

bift

4J

a

o
u

t^l>OOOIr-0»01>I>00(NOOO»OiO

CM^'-Hr-l(M.-i(Mi-ii-<i-H(M'-<(MC<I

13

d
o

a.

--HC^l iOiO^I:^OCO(MCOCO^iO(X)t^'-l
l>00t^CDi— it^-^LOCOO»OOOTt<OT-HO
(MiOiOCO-^cOIMiOCOCO^TtH-^iOCOC^

Yield in

Pounds

per

Acre

CO (M O »0 CO COCOrt^COiOi— ir-i,-ic^005

T-HC^I>00CDCOO5O>COTt<(Mt^TtlCOTtHr-l

oooocooi>r^'-HCD02coi>(Mfocot^'-<

(M(MCOCOC^(M<M(M'-H(M(M(M(M(M(M(N

Mean
Leaf
Produc-
tion

OiOi-HO05O-*O500I>O5Tt<00lOQ0O

05 0C35a)ioiot^coi>cocoooiOr-ic^o
(Nco(^^(^^<M(^^(M(^^'-l(^^(^^,-l(^^(^^(M(^^

6

2:

5-2-1-3

6-2-1-4

12-1-1

12-2-1

27-1-1

27-2-1

41-1-2

41-2-1

41-2-3-2

73-2-3-3

76-2-1-1

77-1-1-2

77-2-1

K-1-1-2

K-2-1-6

Havana

Results of Sorting Test. 37

A glance at the percentages of Light Wrappers received shows
in no case a very favorable result. Selection (27 — 2) — 1,
which gave a relative value of 157.3, leads all the selections
by producing a total of 24.7% of Light Wrappers. As the
Havana which was grown on the same field produced only
9.5% Light Wrappers, the results seem more favorable. There
is certainly a wide range of value for these Halladay selections.
The poorest, (73 — 2)— 3 — 3, which also was the selection which
produced the shortest leaves of the lot, had a relative value
of 74.2, while the most favorable, (K — 2) — 1 — 6, gave a relative
value of 162.6 as compared with Havana.

It has already been mentioned that before the tobacco was
sorted it was examined by three tobacco men. These three
men examined the same hands which had been used for the
grain and texture results, each working independently and
without prejudice of any kind other than some diversity of
opinion as' to what constitutes an ideal tobacco. None of the
three men were very favorably impressed, the general criticism
of each being that the tobacco lacked a bright finish. The
different selections, however, were given relative placings,
at our request. After the placings had been roughly made,
each man was then given the second picking of the six selections
which, in preliminary judgment, were rated the highest. With
these second pickings final placings were then made, and the
results are given in the table below, the gradings being placed
in sequence with the better type at the top.

Judge 1

Judge 2

Judge 3

(77-l)-l-2

(K-2)-l-

-6

(K-2)-l-

-6

Havana

(73-2)-3-

-3

Havana

(77-2)-l

Havana

(27-2)-l

(76-2)-l-l

(77-2)-l

(77-2)-l

(K-l)-l-2

(77-l)-l-

-2

(41 -2) -1

(6-2)-l-4

(27-2) -1

(76-2)-l-

-1

It will be noted that (K — 2) — 1— 6 appears first twice and
it is also of interest to know that this selection gave a high
relative value by the sorting test. Commercial Havana ranks
second twice and third once. The only other selection which
appears three times in the judges' table is (77 — 1) — 1— ^.
(27 — 2) — 1 , which gave the second highest relative value,
appears twice in this table. ' •■'

38 Connecticut Experiment Station, Bulletin 176.

As the crop was of such an inferior nature no hard and fast
conclusions can be drawn as to the commercial value of the
selections. It is encouraging that under similar conditions
several types gave much higher relative values than Havana.

Conclusions.

Our results show conclusively that the Halladay Havana was
not a mutation or sport, but that it resulted from a recombi-
nation of parental characters, in which the number of leaves
and leaf shape of the Sumatra were united with the leaf size
and habit of growth of the Havana. That the general Halladay
Havana type as it appears in the field can be reproduced when-
ever desired is an undoubted fact.

The apparent uniformity of the Halladay type in 1908 has
proved to be of only superficial nature. By selection we have
been able to produce several strains which differ very widely
in number of leaves, leaf size and vigor. In other families
of this cross, selection has as yet given no results of appreciable
value. It seems only fair to conclude that by selection we have
been able simply to isolate different lines that approach a
homozygous condition, and that in those cases where selection
has given no results the lines were already in a nearly homozy-
gous condition.

Quality of cured leaf is, without a doubt, due to both external
and internal factors. Environment, of which we may mention
physical characters of soil, moisture, temperature and soil
fertility, and methods of handling, such as time of harvesting,
are of great importance. These may be roughly classed as exter-
nal factors.

In our experiments we have eliminated, as far as possible,
unfavorable external factors, but the total elimination of un-
favorable conditions is a physical impossibility. All that we
have been able to do is to give all selections as nearly an equal
chance under as favorable conditions as possible. The relative
values of the experimental selections were compared with
Havana grown under similar conditions. Assuming the value
of Havana as 100, the experimental types have ranged in value
from 74.2 to 162.6.

Previous experiments have shown that the Sumatra parent
lacks wrapper quality when grown in Connecticut. It has,

Sumatra-Broadleaf Cross. 39

however, a large leaf number and a good leaf shape. The
Havana parent, while widely grown, is not an ideal type. The
leaf is too pointed in shape and there are also possibilities of
improving its quality. A leaf which is of intermediate weight
between Sumatra and Havana and which shows the bright
appearance and elasticity of the Havana parent would be of
commercial value. Nearly all of our Halladay strains have
good leaf size and an improved leaf shape. Some of the types
are very inferior in quality, others are of intermediate value,
and a few closely resemble Havana. The better selections will
be further tested as they show promise of being of commercial
value.

Family (403X401), Sumatra XBroadleaf.

In 1909 a cross was made between Sumatra (403) and the
Connecticut out-door type of tobacco known as Broadleaf
(401). The Sumatra had been grown under tent from inbred
seed for four years and appeared uniform. The Broadleaf
parent was a commercial variety, and as seed of the same type
has proved very uniform we feel justified in saying that this
cross was made between types which, as to external characters,
were in a nearly homozygous condition.

The objects of this cross were to study the inheritance of
certain characters as a check on the Halladay Havana results,
and to produce a type of tobacco which had the desirable quality
of the Broadleaf parent together with more desirable mor-
phological characters, and it was thought that a recombination
of factors from both the Broadleaf and the Sumatra might
furnish such a variety. The leaves of the Broadleaf are long
and drooping, and for this reason the tobacco is hard to cul-
tivate and harvest. The shape of the leaf, with its narrow pointed
tip, is such that considerable waste is made in cutting wrappers.
A shorter, rounder, more erect leaf of as good quality as the
Broadleaf would be of material value. It has not been pro-
duced as yet but the results are of interest as some facts of
importance have been obtained.

The first generation of the cross together with its parents
was grown in New Haven in 1910, though a few plants of the
Fi generation were also grown in Bloomfield.

40 Connecticut Experiment Station, Bulletin 176.

In 1911 the parents and two F2 generations were grown in
New Haven and large cultures of three F2 generations were
grown in Bloomfield. It was our intention to harvest the
Bloomfield selections and to examine them for quality, but
there was a heavy hail storm a few weeks before harvesting time,
and as only about half the leaves were worth harvesting, the
tobacco was sold in the bundle and no actual sorting data were
taken. However, some leaves were of good quality.

A number of F3 generations were grown in Bloomfield in
1912, and others, together with further generations of the
parents, were grown in New Haven. The Bloomfield selections
were assorted in the same manner as the Halladay Havana types.

Inheritance of Leaf Number.

The inheritance of number of leaves per plant for this family
has been considered in a previous paper (Hayes, 1912) and the
Fi and Fa hybrid generation results were then given.

Table IX gives the results of three generations of the parents,
the first generation of the cross which was grown in New Haven,
two F2 generations, and nine F3 generations, which were grown
in Bloomfield. This table gives the number of leaves of the
parent, the total number of variates, the means, and the co-
efficients of variability.

The Broadleaf parent (401) has shown little variation in
mean leaf number in the three years grown, the means being
19.2 ±.05 leaves in 1910 and 19.9 ±.07 in 1912. The coefficient
of variability is slightly higher in 1912 than in 1910.

The mean leaf number of the Sumatra variety was 28. 2 ±.08
in 1910, 26.5 ±.11 in 1911, and 26.2 ±.12 in 1912. The dupli-
cation of the results in the last two years indicates an error of
counting in 1910, since such an error might arise by not
discarding the three basal leaves uniformly as was done in the
later years.

The coefficient of variability for the Sumatra parent was
5.27±.21 in 1910, 6.64±.28 in 1911, and 8.28±.32 in 1912.
The cause of this rise in variability in 1912 is not clear. It
may be due to a small mutation in one of the germ cells of
the 1910 plant that gave rise to the 1911 population. The
population in 1912 would then be the F2 generation of the

Sumatra-Broadleaf Cross.

41

w

Q

<
O

X

H

<!

s

^ X

1-3 T*

>

1-100 (M
C<)iM CO

050-*
^COiM

rHi-IOOTf<lOO'-ICOCDl>CO
(Ni-lrHCOCOlOCO^^COlM^

■H fl -11

^^ ■« ^

O'^CO
OiOO

■H4J^41414J^4]4f41-H-H

^Olr-H'^TtlOOlOTMQOlO'-H'O
l.0 051000rHCOt^»0'-l'*t^<M

lOCOGO

lO CO CD

i000050001>-Ot>C»QOGO

T-H I— 1 1— 1 r— 1

<

GO'-i (N
O^-ir-i

OOO

t^C0C0(MOOOTt<Tt<T-iai(N

OOO^Ht-IIMt-I^Ht— li— lO^H

-H -H 41

■ti -H -H

-H-H-H-tl-H-li-ll-D-ll-H-W-H
or-'OiciocO'-n>ooi:^ioco

CO CO CD

C5 05 O
1— 1 .— 1 1— {

COCCI(M(N^cDCOCO'-i'-iC^O

o

OlQi-H OCOiO
lOIMiO lO O TfH

T-< t—i T^ 1— 1 T— 1 1— 1

OCq(MOt^(M^05COOCS>iO
iOOCOI>OOOiOOiOiO(MOO

T-lr^COrHlM T-l C^ rH (M

en ni

24-31
23-31
21-32

(MiMco

<N (M (M
t^COt-

CDC^iOOlCCCOt^OOGOt^CO
C^COCO(MC^CO(MCO(MiM(M(M

on>i:^cot^cia3oooi>cDt>

m a

> is

t-5

•05 05
•(M(M

■0(M

.lO^LOINOOCDLOi-Or^GOkO

.2

i

a

o

P^fCPu

pCfCPM

fefefofeplHfLHfXHClHpHpLHflHfe

a
&
2

o

Ot-i (M
I— I 1— 1 ,— (

03 (33 CJ

.— 1 T— 1 ,— (

OCiO

I— It— It— (^Hi— It— (1— 1^1-^7— It— ll— (
OCiOCi0^0^000030^03

6
2

403, Sumatra

403-1

403-1-2

401, Broadleaf

401-1

401-1-1

II

^ OiNTti

2 Tf t^ 00 T-l T-4 T-H lO CD 00

y 1 M II 1 1 1 1

■r;^,_icoT-(T-(r-( ,-!,-( T-H COCO CO
o 1 II 1 II 1 11 II

:^m mmmmmmmmmm
*

42 Connecticut Experiment Station, Bulletin 176.

mutating germ cell with a normal cell. On the other hand,
though, we have data on another cross that indicate that the
field environment has but little effect in determining the number
of leaves, it may be that this effect is somewhat greater on
the Sumatra variety with its different habit of growth.

Cross (403X401) has been designated as B in Table IX, and
as such it will be described in the text. An inspection of the
table will show that the first generation of the cross is no more
variable than the parents, although intermediate in leaf number,
whereas the F2 generations, B — 1 and B— 3, of which large
cultures were grown, are extremely variable, giving coefficients
of variability of 8. 99 ±.11 and 9.51 ±.10, and ranging in value
from the leaf number of the Broadleaf to that of Sumatra.

Of the nine F3 generations, B — 1 — 8 has a mean for leaf
number of 26. 3 ±.20, which is about the same as Sumatra,
while the remainder show means of intermediate value, although
that of B-3-8, 20. 6 ±.12, is only slightly greater than the
Broadleaf parent.

B — 1 — 14 shows a coefficient of variability of 7.18 ± .46, which
is only slightly higher than the parents. This same selection
was also grown in New Haven and gave a coefficient of vari-
ability of 6. 44 ±.27. For this reason, if one is to attach any
value to this biometrical constant, it seems only fair to con-
clude that this type is in a homozygous condition for leaf number.
B — 1 — 10 also proved rather uniform since it had a variability
coefficient of only 7. 75 ±.30. These two types were both of
intermediate value for leaf number.

On the other hand, five of the remaining populations have
coefficients of variability of practically the same value as the
F2 generation, and two show an intermediate value. This
difference in the variability of Fs populations grown from
individuals from various F2 classes is exactly what shotild be
expected if several Mendelian factors have recombined in the
F2 generation.

Shape and Size oj Leaf.

In the data on inheritance of leaf size in cross B, which were
given in an earlier paper, there were no F2 plants mth as large
an average leaf area as the extreme variates of the Broadleaf.

Sumatra- Broadleaf Cross,

43

CD O ^ O O

(N CO C^l lO CO

>

-h" -h" -h" -h' -h'

fa

(j

O (M 00 --HO

COCDi-HgO'^

W

cot^coocJ

►J

1— t

Q

<

O

m

PQ

050 OOlOCi

X

<

i-H C<J i-H CO --H

<i

4]" -n" i -h' -h'

a

10 03 (MCOiO

<

C0 1> C0O5CO

»o ■* »o ■^ -^

CZ2

.H

H

OOOt^i-i

O

lO »0 lOO TtH

T^

H

T— 1 1— ( 7— 1 I— 1 d

X

o

^

Oi

• • • 1—1 •

"^-^

CO

t/2

CO

o

CD

Pi

O

CO
CO

T— 1

■ 1—1 •

fe

to

o

(M

CO(M(N

.

o

w

CD

1-1 •

X

Cfi

s

t^

OCOOO 05C0

PQ
<

2

o

lO

CO (M

td

«

TjH

coco (M co»o

<
Ph

Q
iz;

^

1

I— 1

■<^ 1—1 CO 1—1 1—1
t^ C-1 Tt< lO o

Xi

lO

'^ CO CO 1-1-*

4J

00

lO CD to O t^

T*<

1— 1 ■* 1— 1 CO lo

<

D

U

XO

(M i-liO^ "0

n

TjH

TjH (MCO

O

(N

•CO • C30

H

^

•1-1 . Ttl

«!

05

(M -(Mt^.

P4

CO

1-H

W

Iz;

CO

• ■ -i-KN

M

CO

O

c

_o

P^

d

^

piTpcfofl^fe

f^H

o

O

X

^^ S

OOO^i-i

w

?, ^

1—1 1—1 1—1 1—1 l—(

Q

0) o

Gi O^ Ci 05 C5

2

1—1 1—1 1—1 1—1 1—1

W

H

M-ipq

Q

Cd " 11

W

0!i

2 aj II

PQ

6

403, Suit
401, Bro;
(403X40
B-1
B-2

44 Connecticut Experiment Station, Bulletin 176.

This was explained by the fact that the environmental conditions
for F2 were poorer than the parents or Fi had enjoyed. While
no statistical records were taken, the large size of leaves of
numerous plants of several of our F3 generations grown at
Bloomfield in 1912 has shown this explanation to be the correct
one.

Size of leaf, as perhaps should be expected, is greatly influenced
by environment, which made proper analysis of our breeding
results a difficult task; but shape of leaf, which is the basis
of our next study, is fortunately less subject to such modification.

The method of determining leaf shape which has been used
is called breadth index. It is obtained by dividing the breadth
by the length and expressing the result in per cent.

The same variates which showed no distinct segregation in
leaf size have been considered, the results of this method of
treatment appearing in Table X. The middle leaf of each
plant was used in computing breadth index.

The table shows that the average breadth index of the Sumatra
is 53. 5 ±.19, which means that, on the average, the breadth
of leaf of the Sumatra is a little more than half the length.
The Broadleaf gave an index of 47. 9 ±.20, and the Fi generation
an index of 53.2 ±.18. The indexes of the two F2 generations
are shown by the table to be 49.3 ±.35 and 46.5 ±.19. The
conditions for the F2 generations were very unfavorable and
the indexes are smaller than one would expect. That there
is some sort of segregation of leaf shape seems very evident,
as the coefficients of variability of the F2 are much larger than
those of the parents, or Fi.

Table XI gives comparative results for length of leaf of the
F3 selections grown at Bloomfield in 1912. This table gives the
average number of leaves per plant, by actual count, the yield
of cured tobacco per acre, and the number of pounds of cured
tobacco of leaf length classes, which range from 12 to 20 inches.
It is regretted that no Broadleaf selection was grown to compare
with the hybrids.

Sumatra-Broadleaf Cross.

45

TABLE XL

Comparative Length of Leaves of the F3 Generations of Cross
(403X401), Sumatra XBroadleaf.

No.

Mean
Leaf
Produc-
tion

Yield

in

Pounds

per Acre

Yield in Pounds for Leaf Length Classes
in Inches

12

13

14

15

16

17

18

20

B-1-4

B-1-7

B-1-8

B-1-10

B-1-12

B-1-14

B-3-5

B-3-6

B-3-8

22.0
21.5
26.3
23.1
23.7
21.8
21.7
22.5
20.6

2030
2476
2579
2517
2405
2629
3206
2927
2566

130
63

305
41
46

58
36

220
126
291
133
101
159
152
173
154

295
213
410
233
150
265
190
203
190

350
281
388
388
261
361
262
275
298

350
352
298
484
362
520
410
323
361

330
399
276
443
421
392
512
405
425

299
567
410
653
545
583
982
643
669

55
475
201
142
519
350
698
845
434

In considering these results it is important to note that only
medium size and large leaved plants were used as parents of
the F3 generations. There is considerable variation in leaf lengths,
as shown by this table. Thus, B — 1 — 4 produced a large number
of leaves on classes 15 and 16. B — 1 — 8 and B — 1 — 14, while
producing the greater weight of leaves on class 18, also pro-
duced a large number of leaves on classes 15 and 16. B— 3 — 6
is the only selection which produced the most leaves by weight
in class 20. The selections, then, show considerable variation
in leaf length when compared with each other and show that
there are probably a number of factors affecting leaf size.

Some general notes on the leaf conditions of these F3 genera-
tions of cross B are given in Table XII. Three general features
— uniformity, color of leaves and type of leaf — were con-
sidered. Uniformity refers to the leaf characters of the selection
as a whole. Those marked "good" in the table were uniform
in all characters, while the remainder showed considerable
variation. These facts are mentioned here, as our results point
to the conclusion that the different characters, such as leat
number, shape of leaf and type of leaf, in which the parents
differ, are in a large measure inherited independently. One
other purpose was to determine if any single external character
could be correlated with quality.

46

Connecticut Experiment Station, Bulletin 176.

TABLE XII.
General Notes on the Leaf Condition of thf F3 Generations of
Cross (403X401), Sumatra XBroadleaf.

No.

Uni-
formity

Color of Leaves

B-1-4

Good

B-1-7

Fair

B-1-8

Good

B-1-10

Fair

B-1-12

Fair

B-1-14

Good

B-3-5

Fair

B-3-6

Fair

B-3-8

Fair

Light green

Medium green

Light green

Medium green to bluish

Somewhat bluish

Medium green

Light to medium green

Medium to dark green

Medium green

Type of Leaf

Moderately crinkled

Smooth to crinkled

Very crinkled

Slightly crinkled

Leaves mostly smooth

Slightly crinkled
Moderately crinkled
Moderately crinkled

Quality oj the Fz Selections.

Data on texture and grain were not taken for the F3 Sumatra
XBroadleaf crosses, with the exception of two selections which
were examined for grain, the leaves being classified into seven
grain classes as for the Halladay types. The selections used
were B — 1 — 10, which proved uniform for number of leaves
per plant, giving a variability coefficient of 7.75 ±.30, and
B — 1 — 7 which was not uniform for leaf number and which
gave a variability coefficient of 10. 14 ±.34.

If there were a correlation between grain and leaf number we
should expect the classes for B — 1 — 10 to be more uniform than
those for B — 1 — 7. A glance at Table XIII indicates that such is
not the case, since both selections were about equally variable
and both have a large amount of grain. At the same time
it is realized that the method of determining grain is exceedingly
arbitrary.

TABLE XIII.
Comparison of Grain of B — 1— 7 and B — 1 — 10.

No.

Leaves
per Plant

Picking

Grain

Classes

1

2 3

4 5

6

B-1-7

21.5

1
2
3

Total

37

32

32

101

41 42

51 40

39 53

131 135

25 13

26 11
23 10

74 34

4
4

B-1-10

23.1

1
2
3

Total

30
29
59

35 40

40 46

44 44

119 130

31 10
26 9
34 5
91 24

1
1

2

Sumatra-Broadleaf Cross.

47

>

>
<

Q

CO CO I> CD >0 '^l

ci ft

Q 2

2 P<

OI> ■ 00 00 CO C<1 Oi 05

O5(M»Ol>-i-^'*0000^
COCOI>00I>O300I>t^

g

COi-H • rt< CO i-H CO O <M

Oh

OCO • t^ CO 05 1-1 CO 05

COCO • <M c^ CO CO CO (M

■3

c
o

Oh

(MOl'^OCOOOOOCOO
<Mt-iiM02CO(MC5CO>0
COOOiMCDcOOOSOSt^

rtfOO

• C<) 00 lO CO 00 1^^

t^t^>oi>-ooa5(Mi— 103

OO'OCOOC^JfM'-iOiiO
(M'^CO-^^COOO'^CO

(M (M • COt^OO 05 TJ^
OO ---HOOOO

O kC • O -* CO ^ 05 (M

00C0O2(Mt^t^00O5CO
COOOcDOOOOJiOOl^
'^lOTtHiOtQiMiOCOTjl

>Ha,

Oh

Oc005t:^»005COI>CO
Cot>t^T-iO(MO(McO
0^»OiCTtHCO(MOiO
(M(M(N(M(M(MCOiM(M

OIOCO'-I1>OOI>IOCD

O IM T)H

TjHI>0Oi— 1>— li— (lOCOOO

I I I I I M I I

,-lrH.-H^,-HT-ICOCOCO

I I I I I I I M

pq pq pq pq pq pq pq pq m

pq

(4

48 Connecticut Experiment Station, Bulletin 176.

Table XIV gives the sorting test and relative values of the
Fs selections. The ^aeld ranged from 2,030 pounds per acre
in B — 1—4 to 3,206 pounds in B — 3 — 5. This seems to be good
evidence that a selection can be produced which would give a
much higher yield per acre than the commercial Broadleaf now
grown. The success of our experiment does not depend so
largely on 3deld factors as it does on quality values, however,
and on this subject no very definite conclusions can be drawn
until the selections are more uniform for external plant characters
and have been tested for quality another season.

B — 1 — 4 has about the same relative value as the Havana
type given in Table VIII, the relation of B — 1 — 4 to Havana
being 105.1 to 100. For the relative values given in the last
• column of Table XIV, B — 1 — 4 has been used as the standard
(100), the actual prices for grades being assiuned to be the
same as for the Halladay types which were given in Table VII.
B — 1 — 14 gave about the same relative value as B — 1 —4, although
it gave a 3deld of 2,629 pounds per acre while B — 1— 4 only
gave a yield of 2,030 pounds. B— 3 — 5 gave the highest ATield,
/ and also the highest relative value of any of the selections.

The attempt to discover some external character or characters
which are correlated -y^ith quality has not, as yet, proved suc-
cessful. It seems very probable that, although it may be neces-
sary to have all characters in a nearly homoz^^gous condition
in order to produce tobacco that is of uniform quality, this
is not because there is a close relation between quality and any
one external character. If the type is in a complex hybrid
condition, variation in time of maturity, venation, etc., will
be the rule. Such conditions will not be favorable to producing
a uniform quality of tobacco.

Conclusions.

The results obtained from the Broadleaf XSiunatra cross
show that, as a rule each character, such as leaf size, leaf shape,
number of leaves and type of leaf, are inherited independently.
Hence the difficiilty of producing a uniform strain after crossing
will depend largely on the gametic condition of the parents.
If the parents differ in a large number of factors the difficulties
will be much greater than if there are but a small number with
which to deal.

Havana-Cuban Cross. 49

The really important feature is that there is a segregation
of quantitative characters in the F2 generation of tobacco crosses
and that some segregates will breed true in F3. As this is the
case, there seems to be no need of using a different method
when working with quantitative characters than for qualitative
or color characters.

Since quality of cured leaf depends on many factors, external
as well as internal, it is probably unreasonable to expect a single
external character to be closely correlated with quality, but
as homozygosis produces uniformity in both quantitative and
qualitative characters it must tend to produce uniform quality.

The important matter in practice is simply to grow a sufficient
ninnber of F3 and later generations to run a fair chance of
testing out all the combinations of factors possible to the parental
varieties used.

Family (402 X 405J , Havana X Cuban.

This cross was made in 1909 between strains of Havana
and Cuban which had been grown for several years from inbred
seed. The Pi generation of the Cuban parental type given in
the- tables was not grown from inbred seed of a single plant,
but from commercial seed saved under tent covering. The
plants from which this seed was saved were grown from seed
of direct descendants of the inbred Cuban type used as the
male parent. The Pi generation of Havana given in our tables
was also grown from commercial seed.

This cross has been designated as C in our discussion. The
parents and different generations of this cross have been grown
under shade covering at the Windsor Tobacco Growers' Cor-
poration in Bloomfield, with the exception of C — 1 — 5 and
C — 1 — 6, which were grown outdoors on the same field as the
Halladay and F3 Broadleaf selections. The conditions for
this cross grown under cloth shade are more uniform than for
the previous experimental selections which were grown in the
open, due to the protection the covering affords from heavy
winds and storms.

The parents and Fi were grown in 1910, further generations
of the parents and F2 in 1911, and the third generation of parents
and five F3 generation families in 1912.

50

Connecticut Experiment Station, Bulletin 176.

pa

O
X

<

<
>
<

X

n

W

>■ •

00fOTt<

t^iooo

(NCOr-l

^ '^ t^ O ■* C<1 CO
(M lO CO CO -* t> lO

41 41 41

41 41 41

00 1>.O3
C55 GO »0

CO GO^

41 41 41 41 41 41 41

O ^ 1> iM O O t>

^< 00 CO O C^ O 1— 1

CO lO ^ C5 i-H O CO

»-( 7-H ,—1 ,-1 ,—1

ooo

Oi-tO

t^ CO 00 Oi CO GO lO

O --H 1-H O '-KM --H

41 41 41

1-1 (M(M

^ M -i\

00 CO Til

oiooi

41 41 ^ 41 ^ fl 41

00 05 1> '^ CD O i-H

oi o oJ 00 CO 00 c5

.-KM rH r-l(N (N (N

o
H

OTt< O

fO oo

O (M M (M GO lO T-l

IC Ol T—i •*■*■* O

COOOt^

(M(N(M(MCO<M(MfOCO(M
1 1 1 1 1 1 1 1 J 1 1

t^CDiOiO^COiCOiMCO

PhPhPLi

O'-i (M

1— I 1— I T— I

05 05 05

03 r-l

-9 I

I I

lO lO iC

ooo

0hPhP-i

fo pL, fe fin fo pL, Ph

0^(M OT-KMfMIMC^KM

1— (i-Hi— ( 1— It— (1— li— It— li— It— I
02O505O2O305O1O5O5O3

o

^j7

(M (M(M

ooo

<MCOrt< lOCO

; I I I I I I
'OOOOOO

Havana-Cuban Cross. 51

Inheritance of Leaf Nvimher.

The inheritance of number of leaves per plant is given in
Table XV. The Cuban selection gave a range of variation of
16 to 25 leaves in 1910 and from 17 to 25 in 1912. The mean
number of leaves per plant was 19. 9 ±.08 in 1910, 20.6 ±.07
in 1911, and 20.9±.07 in 1912. There has been a slight pro-
gressive change in leaf number for the three years, but whether
this is due to an actual germinal change or to unavoidable errors
in our leaf counts is impossible to say. No wide changes are
shown by the coefficients of variability, which were 7. 53 ±.28
in 1910, 5.29 ±.23 in 1911, and 6. 17 ±.24 in 1912.

The Havana selection gave a mean of 19. 8 ±.07 leaves in
1910, 20.3 ±.10 in 1911, and 19.4±.05 in 1912. This selection
shows no great change for leaf ntunber. The coefficient of
variability shows considerable variation, as it was 6. 98 ±.27
in 1910, 8.87 ±.35 in 1911, and 4.59 ±.18 in 1912.

The Fi gave about the same mean and variability coefficient
a,s the parent types, the mean being 19.8 ± .07 and the coefficient
of variability 6. 10 ±.24.

If the parents both contained the same inherited factors
for leaf number, which one might expect from their having
about the same average number of leaves per plant, no increased
variability over Fi should be obtained in F2. The range of
variation, 14 to 33 leaves, and the coefficient of variability
of the F2 generation, 15. 84 ±.54, both show that such' is not
the case. Plants appeared which bore a higher and also a
lower number of leaves than in Fi.

The counts of leaf number for the five F3 generations show
•conclusively that tiie increased variability in F2 was a germinal
one. These five F3 selections were grown from F2 plants which
bore 20, 20, 22, 28 and 30 leaves respectively. Progeny from
one of the 20-leaved F2 plants, C — 1— 3, gave rather uniform
results in Fs, the mean being 18. 4 ±.09 and the coefficient of
variability 9. 02 ±.36. Progeny from the other 20-leaved parent
plant, C — 1— 2, and also the 22-leaved plant, C — 1— 6, gave
means of about 20 leaves per plant and large variability coeffi-
cients, 14.67 ±.67 and 16.17±.56 respectively.

The two remaining selections, C — 1— 4 and C — 1 — 5, with
coefficient of variability values of 11. 20 ±.44 and 10.00 ±.72
-were more variable than the Fi and less variable than the F2.

52

Connecticut Experiment Station, Bulletin 176.

iz;

<
m

O
X

<

<
>
<

o
X

o

■*OOC0(N(M0500t^

C<j(NC<;r-j(NCO(N(NC<J

>

-h' -h* -h" -h' -h' -h' -h' -h" -h'

o

•^COOlOSCOQOOiMO

OT-^OOt^T-JTiHOO

COiOiO-<^iOCicOCOI>

iccoco'^cococoooco

I— IrHrHrHi-HlMCqi— li— 1

<i

-h' -h' -h' -h' ■«" -h' -h' -h" -h'

•<d^cDcoTt^co(^^lOooo

ioioo6oc>T-Ht>Iioi>

TjHTtHlOCDlOlOlO-*-*

"tS

OOOOOIMCOIMO

o

lO'O'OiO'OOi— l-^O

H

I— IT-Hi— li-Hi— li— li— iT-Hi— 1

(M

• .1-1

1>

OS

• 1-1

CO

o

1-1 05

■ 1—1

<©

CO

rfH^fH

1-100

n

CO

.-ICO ■

o

o

CDOi-lO 00

i

CO

iCl> i-lCO

U

t-

COOCOCDiO

(N

K!

in

U3CO (NCO •

Hi

■*

i-H

05 1—1 10 TfH 0 1— 1 CO

ti

»o

•1-1 (NcOi-i

^

T-l

Oi-^CO -aUXiOOSrfl

lO

cDio i-icq

•S

CS

00

0000 ■

00 t^ CO 0 10

pq

■*

CO'*

COCO eoiQ

lO

03TJ^

coio -coo

rt<

CDt>

1-1 1—1

10 10

(M

05,-l

1-10

100

-*

(M(M

1—1

(M1-I

C5

■*co

• 1—1

coco

CO

CO

• i-H

CO

t«

is c

O rt

C.2

4) -t-3

CL|CL|ClHCL|pL<fefJHpHpLH

o

''S

OIMO(NOt-1(N(M(M

s^ ^

1— ll— Ii-(t— li— li— li— li— li— 1

4) o

OOiOiOOC^OOiOS

o

rt II

G c ^

rt 1— 1 oj 1-1 lo

^ !-§ 1^ 77^

K707X 1 1 '

1.^ i-H 1-1 i-H 1— I

•. ' •» ' ^

(N(M":iiOo 1 1 1 1

Havana-Cuban Cross.

53

X

m

<

<
n

O
X

<

>

o
X

O

CD CD CO IC tx3i 03 CD O

>

■H^l^^^^l^■H^^^^

6

,-HC01>'r-I^C0C0lCl'*

l^COCD'^OOiOOSCDCO

,— IrH,— 1^1— l(Ml— l^(M

coioio-^t^dt^cob-

OOOOOOOOO

<

•H-H-H-H-H-H-H-H-H

l^OSCOQOiOOC^i-ICO

COTti>0-*COt>.iOCO"3

1

OOOOOC^IOrHO

H

1— IT-H,— iT-Hi-HirHl— ll— It— 1

- 1— 1

1—1
i-H

•TtH

o

l-(

•00

05

GO

IM

(N

Q

00

CO

t^CO T-ICOO
(M CO tH T-i

_g

1>

rH coo

o lo 00 ■* c<i
coTft co^

i

CO

N CO Tt< GO CO C<) i-< CO 1-1

T^coiOi-iTtfcoco-^co

O

lO

CO »Ci GO 1> 1>(M CO to 00
<McOiO00(MCO-<l<'*»O

Tj<

T-lt^CDT-iCOO(MCOCO
TjH iMTtH ^ (M CO

CO

• <NCOTt< ■ COiO • ri^

2a
S.2

Ph Pu pC Pm pC fe Ph fe fe

O

en ^

OtNOC^JOi-iC^KNIM

1— It— (i— (T— It— li— iT—li— It— 1

o

O

03 II

03 1—1 03 1—1 lO

Cq(M»OiOo 1 1 1 1

Th ^ Tt< ^ wU U O O

54 Connecticut Experiment Station, Bulletin 176.

The means for leaf number were 26.6=t=.16 and 28.0±.28.
Thus, from crossing two types ibearing an average of about
20 leaves per .plant, a new type has been produced with a larger
leaf number. ; ;

: Size and Shape of Leaf.

It was pointed out in an earlier paper that Cuban and Havana
have about the same average leaf width but that Havana has
somewhat longer leaves than Cuban. The breadth indexes
of the parental varieties and crosses are given in Table XVI.
As in the other cross, the middle leaf of each plant was used
for these computations. The Havana leaf is shown to be pro-
portionally much narrower for its length than the Cuban.
The Fi was ;of intermediate value for breadth index, and in
F2 there was an increase of variability. The F3 strain, C — 1 —2,
bred comparatively uniformly for the Cuban shape of leaf,
giving a mean breadth index of 57. 5 ±.23. This is slightly
lower than the index of the 1910 Cuban selection, which is
58.3='=.16, but the difference between these values is slightly
less than four times the probable error. The parent F2 plant
of C — 1 — 3 resembled Havana in all particulars and the progeny
was of Havana type in both leaf size and breadth index value.
The breadth index of C — 1— 4 was also of Havana type, and the
coefficient of variability showed this selection to be uniform
in leaf shape.

Table XVII gives the inheritance of leaf size for this cross.
For this work, the areas of the fourth leaf from the bottom,
the middle leaf, and the last leaf at the top below the bald sucker
were taken. The area of leaf used in the table is the average of
these three mieasurements. |

The table shows that in 1910 the average Havana leaf area
was greater than the Cuban and that the Fi generation had
nearly as large an average leaf area as Havana. The average
leaf area of the F2 generation was slightly greater than in Fi
and the variability was also much greater.

It is true that none of the shade selections grew as vigoroush^
in 1912 as in previous years, but this does not explain the pro-
portionally greater decrease in leaf size of the Havana as com-
pared with the Cuban. It is of interest to know that selection
C — 1— 3, which was not very variable for leaf number and

Havana-Cuban Cross.

55

which was of uniform leaf shape, gave a variabiHty coefficient
of about the same value as the parental selections. The coeffi-
cient of variability of C — 1— 2 was only slightly greater than
that of the parents, while C — 1— 4 seemed to be more variable.

It should be mentioned that the coefficient of variability is
not a very safe criterion by which to judge when dealing with a
character such as area of leaves. It is to be expected that a
selection which is heterozygous in other plant characters will
be more variable in a character such as leaf area than a com-
pletely homozygous selection, as stimulus to development is
greater in a heterozygous than in a homozygous state, and
when segregation is taking place some plants of a generation
are homozygous and others complex hybrids.

The comparative length of leaves of the parents and Fs
generations is given in Table XVIII. As in previous tables
of this kind, one must remember that these computations are
made on the acre basis and that the figures in the table under
the heading "leaf classes in inches" refer to pounds and not
to number of leaves.

TABLE XVIII.

Comparative Length of Leaves of the Parents and F3 Generations
OF Cross (402X405), Havana X Cuban.

No.

Mean Leaf
Produc-
tion

Yield in
Pounds
per Acre

Leaf Classes in Inches

12

13

14

15

16

17

18

20

405-1-1*

402-l-lt

C-1-2

C-1-3

C-1-4

C-1-5

C-1-6

20.9
19.4
19.7
18.4
26.6
28.0
20.1

1493
1508
1635
1369
2036
1709
2206

186
51

102
44

98

193
29

137

36

6

100

151

142
208
183
33
51
168
214

350
113
218
120
93
200
351

328
164
295
153
93
302
292

218
273
355
127
206
369
411

76
499
279
517
556
469
538

17i

66

339

1032

101

151

i

*Cuban.

t Havana.

This table shows that the Cuban produces a larger percentage
of short leaves than the Havana. C — 1— 2, which it will be
remembered was of Cuban shape except that is leaves average
slightly larger, shows a population similar to 405 — 1 — 1. C — 1 —

56 Connecticut Experiment Station, Bulletin 176.

3, the Fs Havana type, shows a population more nearly like
Havana. Selection C — 1— 4 is of interest as it produced a
much larger number of leaves per plant than the other shade
selections. It also produced a large proportion of leaves of
20 inch length, averaging 1032 pounds per acre. The results
given for C — 1 — 5 and C — 1— 6 should not be given much
weight in the discussion of comparative leaf lengths as they
were grown out of doors. The interesting feature of these
results is that one of the five F3 generations closely resembled
the Havana parent in leaf size and shape while another F3
generation produced leaves that were of the shape and size
of the Cuban parent.

Inheritance of Quality.

The results of a sorting test for quality are given in Table
XIX, and the prices per pound which were used in computing
relative values 'are given in Table VII. It is, of course, true
that the selections which were grown under shade are worth
more per pound than the prices used indicate; however, for
our purposes these prices are probably as valuable as any other.
No corrections were made for leaf number except for C — 1— 4,
which produced 26.6 leaves per plant, this being reduced to a
20 leaf basis. The fourth picking of C — 1 — 5 was lost, so the
figures given for this selection represent the first three pickings
only. Selection C — 1— 6 was weighed before sizing and the
yield given in the table is correct. During the warehouse work
the third picking of C — 1 — 6 was mixed with a Broadleaf selection.
The Broadleaf selection was discarded, but in the case of the
C — 1 — 6 the value per pound of the third picking was estimated,
as we knew the actual value of the first, second, and fourth
pickings.

The results of this sorting test throw some light on the problem
of quality inheritance. Both parental varieties in this cross are
tobaccos which produced a good quality of wrapper leaf. The
percentages of light wrappers are 31.9 for 405 — 1 — 1, Cuban,
and 39.8 for 402 — 1 — 1, Havana. For the computation of the
relative values, Havana is again taken as the standard and
the ratio of the shade selection 402 — 1 — 1 to the out-door
Havana given in Table IX is 118.3:100.

Havana-Cuban Cross.

57

X

f^ X

^ >
fa <
offi

w

>
>

<

w

Q

.2 ta

Q2

§"^-3 5

<U (d'O o

Oh

O O lO I>- Oi O i-H

OiMOOOSOO^

rfH 00 CO CO '^ O »0

CO CO Ci T— I ^ lo lo

1— I 1-H rH i-H 1—1 CO

»00(MI>^ COO

o cqt>^ Tf^oco

t^ lO I> iM 1-H CO 00

'^ '^ ■* CO ^ CO 1-1

CO O 05 CO 1-H ■* 00

IM I> lO O CO CO CO
00 1-H (M Oi lO CO c^

CO CO CO CO lO '^ lo

C5 00 1-H O lO O (M

!>. 1-H O 1-H lO l> Ci

t>0-*0 OCO 00
•^ CO CO CD 05 00 T-H

CO 00 iCi 00 CO 05 CO
030CO (M CO OO
"^i lO CO Tj< O IXM
1-H I-H T— 1 1— I cq T— ( c^

05 ■* t^ ^ CD O I-H

I 1 (MCO-^mcO

^^ I I I I I

I I rH^rH^^

J^s ' ' ' ' '

ctiO

GO

03 --H

%^.

ffiS

bfi^

G 03

^x

^

.Q

,

ol

1

H

<M

O

'^i

G

Clj

G
>

ri

03

cfi

ffi

j2

;3

O
O

t«

-d

_G

=1

1

OS

lO

G

o

w

^

03

G

ffi

03

U

o3
,G

M

-n

G
o3

rt

G

o

oi
>
o3

G

ffi

03

>

G

11

<u

>

is

4)

oj

a-!

o,

CO

a

o

u

tH

II

o

J3

OH

58 Connecticut Experiment Station, Bulletin 176.

That the increase of leaf number does not cause an increase
of dark and top leaves is clearly shown by selections C — 1— 4
and C — 1 — 5. These selections both produced a high percent-
age of light wrappers and gave a high relative value.

The yields of the shade tobacco are much less than they
would be if they were grown 'in the open, as the shade covering
produces a thin leaf. A sample of Havana shade-grown light
wrappers was shown to a well-known buyer who was in the
warehouse when the experimental tobacco was being assorted
and he was asked what they were. He immediately replied,
"A fine quality of Havana." On the other hand, an out-door
Cuban selection retained its distinctive character, although
the percentage of dark leaves was greater and the leaves were
heavier in the out-door tobacco. Thus we must come to the
conclusion that quality, while decidedly affected by environment,
is nevertheless greatly dependent on heredity.

The relative value of C — 1— 6 is only 86.1 although this
selection gave a yield per acre of 2,206 pounds. This seems
most easily explained by the fact that this selection was in
a heterozygous condition for many characters. The variation
in leaf number per plant was very high, as is shown by Table
XV, and we know from observation that the variation in
leaf shape and size was also very large. Hence, though some
leaves of this selection were of high quality, the percentage was
very low, and a large percentage of off-colored and dark leaves
was produced. These results show that uniformly high quality
cannot be expected if many characters are in a heterozygous
condition.

Conclusions.

The results obtained from this cross show clearly that an
external similarity of size characters in tobacco varieties does
not necessarily mean a genetic similarity. Havana and Cuban
both produced about the same average number of leaves per
plant, yet when they were crossed together an increased vari-
ability occurred in F2. The five F3 generation selections show
that this increased variability was germinal, two of the five Fs
selections giving; a much higher leaf average than the parents.

Similar results have been obtained frequentl}^ in inheritance

Interpretation of Relults. 59

of qualitative characters. The general basis of the Mendelian
conception of heredity depends on the fact that the somatic
appearance of a plant is not a correct expression of its breeding
nature. Of two red-flowered plants in the second generation
of a cross between white and red-flowered races in which com-
plete dominance is the rule, the one may breed true for the
red color, giving only red progeny, and the other may give both
red and white progeny. Advances may be disguised and may
appear in crosses as well as simple recessives, although
advances due to crossing are as a rule less frequent than simple
recessives. In such cases as the purple aleurone color of maize,
which depends on the presence of at least two color factors
we may receive purple aleurone seeds on crossing white races if
one white race contains one of the necessary color factors and
the other white race contains the other. That similar results
are obtained when dealing with size characters and that in
both quantitative and qualitative characters it is impossible
to know the germinal characters except by a breeding test
seems further proof of the belief that both are inherited in a
similar manner.

The results of the sorting test of the parents and third genera-
tion crosses show that heterozygosis affects quality and that
uniformity of external characters tends to produce uniformity
of quality in the cured leaves. Some of the hybrids gave in-
creased yields and good quality and look promising from a
commercial standpoint. It will be necessary, however, to con-
tinue the selections in row cultures until all characters are in a
homozygous condition or nearly so.

Interpretation of Results.

In a previous paper (Hayes, 1912) the data obtained from the
first and second hybrid generations of size studies of tobacco
were given a strict Mendelian interpretation by assuming a
multiplicity of factors, each inherited independently and capable
of adding to the character, the effect of the heterozygous condition
of each factor being half the homozygous. The data on the
third generations and on the Halladay reported in this paper
show no need of a change of interpretation.

In order that the above interpretation may be justified,

60 Connecticut Experiment Station, Bulletin 176.

certain results must be obtained. The first generation of a
cross between two homozygous varieties which differ in a
quantitative character, such as number of leaves per plant, must
be of intermediate value and no more variable than the parents ;
the F2 generation shoiild give an increase in variability and, when
sufficient individuals are studied, should give a range of vari-
ability equal to the combined range of the parents. Certain
selected F2 plants should breed true giving no greater variability
than the parents; others should give a variation as great as
the F2 generation, and others should give variabilities inter-
mediate between the value of the Fi and F2. All of these con-
ditions are fulfilled in our crosses.

The exact number of factors involved in any cross is difficult
of determination, due to the obscuring effects of fluctuating
variability. It might be possible to determine the number
accurately by growing the parents, the Fi and Fo generations
and a large number of F3 generations under as uniform environ-
mental conditions as possible. But even when only a limited
number of F3 generations are grown, it is possible to obtain an
approximate idea of the factorial condition.

For the sake of illustration, let us first consider the inheritance
of leaf number in the cross between Sumatra and Broadleaf
given in Table IX. In this cross the parents differ by about
six leaves per plant, the Broadleaf producing an average of
about 20 leaves and the Sumatra an average of about 26 leaves.
The Fi generation was of intermediate value and no more variable
as determined by the coefficient of variability than the parents,
while the F2 generation gave a range of variability equal to the
combined range of the parents.

Of the nine F3 generations, B ~ 1 — 14 is comparatively uniform.
Only 56 variates of B — 1 — 14 were grown at Bloomfield, the
calculated coefficient of variability being 7.18=^.46, but 131
variates of this same selection were grown in New Haven and
a variabiHty coefficient of 6. 44 ±.27 was obtained. Considering
the large probable errors of these determinations it seems only
fair to conclude that the coefficients of variability are really
identical and that B — 1 — 14 is in a homozygous condition for
leaf number. B — 1 — 10 is also rather uniform giving a vari-
ability coefficient of 7. 75 ±.30. Of the remaining selections,

Interpretation of Results. 61

four show coefficients of variability slightly greater than in F2,
one has about the same coefficient value as F2 and two are of
intermediate variability.

The results of this cross can be explained by supposing that
the parental varieties are each pure for the same basal factorial
formula for 20 leaves and that in addition the Sumatra has
three independently inherited factors, each adding two leaves
when homozygous and one when heterozygous.

Our gametic conditions for Broadleaf will be 20 aabbcc and
for Sumatra 20 AABBCC. The Fi formula will be 20 AaBbCc
or 23 leaves, and in F2 there will be a germinal variation from
20 to 26 leaves. With these gametic formulas we should expect
one out of every eight F3 generations to breed true. Of the nine
F3 generations given in Table IX, one gave a coefficient of vari-
ability of about the same value as the parents. That the F3
generations gave different averages for leaf number may be
seen by consulting our results.

All crosses cannot be explained in as simple manner as this
one. In the case of inheritance of leaf number of cross (402 X 405)
Havana X Cuban, the conditions are apparently more complex.
Here both parents and Fi gave an average of about 20 leaves
per plant and about the same coefficients of variability. The
F2 generation was very variable, and of the five F3 generations
grown two proved as variable as the F2, two were of inter-
mediate variability, and one showed a coefficient of variability
slightly larger than the parents or Fi. As selections were grown
in F3 which gave higher and lower leaf averages than the parents,
the variability of F2 must have been germinal. As only about
150 variates were counted and only five F3 generations grown
it is impossible to say definitely how many factors are involved.

If we suppose our parental formulas for leaf number to be
14 AABBCC and 14 DDEEFF, we will obtain a condition in
Fi of 14 AaBbCcDdEeFf or 20 leaves, and a germinal variation
of 14 to 26 leaves in F2. While this hypothesis may not be
correct, the resiilts can be explained by some such means.

In the inheritance of leaf shape of the cross between Havana
and Cuban, the conditions are very simple. The data from this
cross are given in Table XVI. The Fi generation is shown to be
intermediate in leaf shape and in F2 there is segregation. Of
the three F3 generations given in the table, all are comparatively

62 Connecticut Experiment Station, Bulletin 176.

uniform, two having the Havana leaf shape and one the Cuban
leaf shape. Two other F3 generations were grown and although
no statistical results can be given we know by observation
that one selection had the Cuban leaf shape and the other
had a variable leaf shape. These results can probably be ex-
plained by the use of a single factor.

It is not assumed that the factorial formulas here given are
necessarily correct, as the conditions may be of a more complex
nature, but we wish to show that some such mathematical
description simplifies the breeding results in a manner that is
helpful in actual practice.

General Conclusions.

Our results show that the Fi generations of size crosses in
tobacco are as uniform as the parents and of an intermediate
value; that there is an increase of variability in F2 and where
sufficient variates are studied, a range of variation equal to the
dombined range of the parents; that certain F2 individuals
breed true in F3, and that others give variabilities ranging in
value from the parents to that of the F2 generation.

These results can be explained in essentially the JMendelian
manner — by the segregation of potential characters in the
germ cells and their chance recombination — therefore, from
the plant breeding standpoint there seems good reason for
believing that quantitative characters are inherited in the
same manner as qualitative characters.

The production of fixed forms which contain certain desirable
plant characters is not, however, a simple problem, due to the
large number of factors in which plants of different races differ
and because a superficial resemblance does not necessarily
mean a genetical resemblance. It is necessary to grow large
F2 generations and to save seed from those plants which most
nearly conform to the desired type. Progeny of these Fo plants
should be grown in row tests in F3 and selection continued in
later generations until the desired form has been obtained.
The length of time which it takes to produce a uniform type
will depend largely on the number of variates which can be
grown in F2 and the number of row tests which can be gro-^Ti
inF,.

General Conclusions. 63

Quality of cured leaf is a complex character and due to many
conditions, environmental as well as inherited. There is also
the added difficulty that the quality of leaf must conform to
the trade ideals. The experiments here reported indicate that
a good quality of leaf can more generally be expected in a
hybrid, if the parents are both of high quality, than if one parent
is a good variety and the other somewhat lacking.

It shoiild be realized that the production of improved cigar
wrapper types is not an easy problem and that desirable results
cannot be obtained without the outlay of considerable time
and money.

64 Connecticut Experiment Station, Bulletin 176.

LITERATURE CITED.

BARBER, M. A.

1907 On Heredity in Certain Micro-Organisnis.

Kansas Univ. Science Bull. 1, vol. 4, 48 pp.

BELLING, JOHN

1912 Second Generation of the Cross between Velvet and Lyon
Beans.

Florida Agr. Expt. Sta. Report for 1911: 83-103.

CASTLE, W. E.

1911 Heredity in Relation to Evolution and Animal Breeding.

184 pp. New York.

1912 a The Inconstancy of Unit Characters.

Amer. Nat. vol. 46: 352-362.
1912 b Some Biological Principles in Animal Breeding.

Amer. Breeders' Mag., vol. 3, No. 4: 270-282.

DARWIN, CHARLES

1876 The Effects of Cross and Self Fertilization in the Vegetable
Kingdom.

482 pp. London.

DAVIS, BRADLEY M.

1912 Genetical Studies in Oenothera, III.

Amer. Nat. vol. 46: ill-Ml.

EAST, E. M.

1910 The Transmission of Variations in the Potato in Asexual
Reproduction.

Connecticut Agr. Expt. Sta. Report 33: 119-160.

1910 a A Mendelian Interpretation of Variation that is Apparently

Continuous.

Amer. Nat. vol. 44: 65-82.

1911 The Genotype Hypothesis and Hybridization.

Amer. Nat. vol. 45: 160-174.

1913 Inheritance of Flower Size in Crosses between Nicotiana
Species.

Bot. Gaz. vol. 55: 177-188.

Literature Cited, 65

and HAYES, H. K.

1911 Inheritance in Maize.

Connecticut Agr. Expt. Sta. Bull. 167. 142 pp.

1912 Heterozygosis in Evolution and in Plant Breeding.

U. S. Dept. of Agr., Bureau of Plant Industry Bull 243.
58 pp.

EMERSON, R. A.

1910 Inheritance of Sizes and Shapes in Plants.
Amer. Nat. vol. 44: 739-746,

FOCKE, W. O.

1881 Die Pflanzen-Mischlinge.

569 pp. Berlin. (Borntraeger.)

FREAR, WILLIAM AND HIBSHMAN, E. K.

1910 The Production of Cigar-Leaf Tobacco in Pennsylvania.
U. S. Dept. of Agri. Farmers' Bull. 416: 5-24.

GARNER, W. W,

1912 Some Observations on Tobacco Breeding.

Amer. Breeders' Report, vol. 8: 458-468.

GILBERT, A. W.

1912 A Mendelian Study of Tomatoes.

Amer. Breeders' Report, vol. 7: 169-188.

HANEL, E.

1907 Vererbung bei ungeschlechtlicher Fortpflanzung von Hydra
grisea.

Jenaische Zeitschrift fiir Naturwissenschaft, vol. 43: 321-
372.

HASSELBRING, H.

1912 Types of Cuban Tobacco.

The Botanical Gazette, vol. 53: 113-126.

HAYES, H. K.

1912 Correlation and Inheritance in Nicotiana Tabacum.
Connecticut Agri. Expt. Sta. Bull. 171. 45 pp.

HERIBERT-NILSSON, N.

1912 Die Variabilitat der Oenothera Lainarckiana und das Problem
der Mutation.

Zeitschrift fiir Induktive Abstammungs und Vererbungslehre,
Band 8, Heft 1 u. 2: 89-231.

66 Connecticut Experiment Station, Bulletin 176.

HINSON, W. M. and JENKINS, E. H.

1910 The Management of Tobacco Seed Beds.

Connecticut Agri. Expt. Sta. Bull. 166. 11 pp.

HOUSER, TRUE.

1911 Comparison of Yields of First Generation Tobacco Hybrids
with Those of Parent Plants.

Amer. Breeders' Report, vol. 7: 155-167.

JENKINS, E. H,

1896 Some Results of Experiments with Tobacco Fertilizers for the
Five Years, 1892-96.

Connecticut Agr. Expt. Sta. Report 20: 310-321.

JENNINGS, H. S.

1908 Heredity, Variation and Evolution in Protozoa, II.

Proceedings of American Philosophical Society, vol. 47:
393-546.
1910 Experimental Evidence on Effectiveness of Selection.
Amer. Nat. vol. 44: 136-145. «

JOHANNSEN, W.

1909 Elements der exakten Erblichkeitslehre.

515 pp. Jena (Fischer).

LOVE, H. H.

1910 Are Fluctuations Inherited?

Amer. Nat. vol. 44: 412-423.

McLENDON, C. A.

1912 Mendelian Inheritance in Cotton Hybrids.

Georgia Expt. Sta. Bull. 99: 141-228.

NEWMAN,' L. H.

1912 Plant Breeding in Scandinavia.
193 pp. Ottawa.

NILSSON-EHLE, H.

1909 Kreuzungsuntersuchungen an Hafer und Weizen.

Lunds Universitets Arsskrift, N. F. Afd. 2, Bd. 5, Nr. 2:
1-122.

PEARL, R.

1912 Mode of Inheritance of Fecundity in the Domestic Fowl.
Maine Agri. Expt. Sta. Bull. 205: 283-394.

Literature Cited. 67

and SURFACE, F. A.

1909 A Biometrical Study of Egg Production in the Domestic Fowl.

U. S. Dept. of Agri., Bureau of Animal Industry Bull.
110, Part 1. 80 pp.

PHILLIPS, J. C.

1912 Size Inheritance in Ducks.

Journal Exp. Zoology, vol. 12, No. 3: 369-380.

SELBY, A. A. and HOUSER, TRUE

1912 Tobacco Culture in Ohio.

Ohio Agri. Expt. Sta. Bull. 238: 263-359.

SHAMEL, A. D.

1905 Tobacco Breeding Experiments in Connecticut.

Connecticut Agri. Expt. Sta. Report 29: 331-342.

1910 Tobacco Breeding.

Amer. Breeders' Report, vol. 6: 268-275.

SHAMEL, A. D. and COBEY, W. W.

1906 Varieties of Tobacco Distributed in 1905-6 with Cultural
Directions.

U. S. Dept. of Agri., Bureau of Plant Industry Bull. 91.
38 pp.

SHULL, GEO. H.

1910 H} bridization Methods in Corn Breeding.

Amer. Breeders' Report, vol. 6: 63-72.

1911 a The Genotypes of Maize.

Amer. Nat. vol. 45: 234-252.
1911 b Defective Inheritance — Ratios in Bursa Hybrids.

Verb. Naturf. Ver. Briinn, Bd. 49: 1-12. (The Mendel
Festband.)

STEWART, J. B.

1908 The Production of Cigar Wrapper Tobacco under Shade in the
Connecticut Valley.

U. S. Dept. of Agri., Bureau of Plant Industry Bull. 138.
31 pp.

STURGIS, W. C.

1899 On the So-Called "Grain" of Wrapper Tobacco.

Connecticut Agri. Expt. Sta. Report 23: 262-264.

68 Connecticut Experiment Station, Bulletin 176.

TAMMES, TINE

1911 Das Verhalten fluktuierend variierender Merkmale bei der
Bastardierung.

Rec. Trav. Bot. Neerl. vol. 8, Livre 3: 201-288.

TSCHERMAK, ERICH VON

1911 Uber die Vererbung der Bliitezeit bei Erbsen.

Verhandl. Naturf. Ver. Briinn, vol. 49: 1-23.

1912 Bastardierungsversuche an Levkojen, Erbsen und Bohnen mit
Riicksicht auf die Faktorenslehre.

Zeit.schrift fiir Induktive Abstammungs — und Vererbungs-
lehre, Band III, Heft 2: 81-234.

WEBBER, HERBERT J.

1912 Preliminary Notes on Pepper Hybrids.

American Breeders' Report, vol. 7: 188-199.

PLATE I.

x>

rt

M

CU

be

a

u

■V

>

^

fl

oj

oj

'S,

'a

^

0)

0)

P^i

72

aj

>

CIj

03

>

O

oi

C<l

> ?

PLATEIII

rt o^.g

&- ^ rt

-»-* (D

"3 4) _Q

O d

'rH CO ri

o t/". r3

bog rf

(Nl

:= rt S

2

+-> '^-^ Tl!

,

d <^ a

K

- -„c3

0)

^fo g3

(1) Cu

,402
the
mat

^ >

d

fe

rf

s

fc*

CO

'^

f1^

a

^ OJ

feg

pj

- d

A OJ

--I >

.2 ^

gffi

o

-* ^

0) ^

^E

(D

^

•ji cd

[/) si;

9 S

"^

(Nl

^ffi

oJ 0^

fl.-i

- d d <u

<A

d

O CO O i-^

PLATE III.

rm

™ I— I

^ ,7, a>7r

lu '•-> '^

^ rt <u

fi ^ o

p . ^

o g g

- R ^ ^

-r § ^ >

'^ (Ti "^ s

03 '"'

O J-' ■

PLATE IV.

1

bo

CD

^

•C^

a;

Oh

p

3

p^

U3

-l->

O

_

V.

0)

rt

rt

^

C

T)

cJ

^

'■+-1

Q

>

15

ffiK

^ <U 0 >

<

(/2

o

1

xr.

o

O

a;

rt

fi

^

(=:

f-i

s-i

oj

>

PI

(U

o

cS

p<

ffi

>.

c

S-.

■*^

03

'd

O

^

'Ph

c-c

rt

as

m

PI

ro

ti!"

^

'^

> ON

u

c

oi

r-H

rt

s

rt

ffi

pT

>>a>

,—1

p

oi

>

^j^

.C/3

>

t3

c3

og

05

rt

K

O

PI

0)

^

^

(U

o3

4)

2;

CO

W

PLATE V.

At left, 12-1-1, a vigorous strain and at right, K-1-1-2,
non-vigorous strain of Hallacla}' Havana.
Bloomfield, 1912.

PLATE VI.

O N
D-ON

O C O

>

m

rt

<u a3 o
^ -^ 3

a '^

O (1)--;
-tJ Cij "rt

<D — o
't, ^

■Ji G 'A

— o -;

^1 s '^

PLATE VII.

1/1 'Xj %-,

, -co •

^ of^

_U ^ CTn

^ S

'^. o -

Broa

f 22.
aven

tion
leaf
af SI

a; ^ffi

-^^

cS 'O 0)

-2 <"

ener
Broa
ate 1

■^ M-i f^

"f^ 2

-c! 0) ,.■

-^ -H

cd fi ;$;

c i3 -2

S rt"^.

rt oj !=!

^ (D t^

W S**:;

■o a r!

« ^^

AC'"
,-( aj o
o a>^

droop

gave

aC.V

aj 'Tl! Tj

^ C C

g.QXi

t; ai rt

CTn

03 .5
t/3 ft
CU O

0)

?? j:

ai

. OJ

0-'^

m

cS >

•^

ft <u

o

• i=l

-^

S <u

o

ftO

PLATE VIII.

('403x401)-l-6, an Fg generation of a cross between Sumatra
and Broadleaf which gave a mean leaf number of
23.9 ± .08 and a C.A^ of 6.61 ± .23. The size of leaf is as
yet very variable. New Haven, 1912.

PLATE IX.

o

"^

;-!

<y

fl

M-l

OJ

<A

o

(U

fl

<Li

to

s

T-!

c

a)

C3

^n

-M

cfl

^

;-.

6/1

di

>

<<

PLATE X.

ci

(-•

^

[o

'■J

s

eg

0^

o

' — '

•^

-;-*

K*

•^

ci

„

"cd

g3

"5

<D

d

•*_

^

Ti

GC

"rt

O

c

V

<u

bJD

s

'-4-^

K

^

rt

o

rt

fc

3
>

S

^'

^ .^ E ^^ (->

ro CD

X o S ^ 5

■^ O _ ^H

• ^-.^

7-^- ^^se-^ataappT?

0)

cu

o

,E|

(XI

M

*J

rC

p

-*->

o

o

OS

'o

t/2

CD

CD

tJ,'^

(33

C

>-l

P

CD

>

G

<

^

CN!

o

C5

On

c3

^

>

CIS

-;:J

"S

K

_3

cC

.

a

fxpq

PLATE!XI.

(X 0)

41

cu

H

GO

J^

(M

H

03
(1)

■H

T— 1

!>

o

m

On

0/

1^1

T-l

Tl

a

o

a

PiTJ

OJ

-u

1

O

aM

^

H

'n;

c3

■^

^

D

03

i^

03

o

<U

"-w

tr.

03
>

01

CD ^

o!

o

K

-H

TIj

O

o

fo'

c

o3

<1)

OJ .

p

0)

n

OJ

o

T) CM

01

K*.

03

w ON

cm'

f-;

(/I
o3

J3
>

o

jH s !=:

^ o3 oj

03^rQ

3 a

O (D

SO

p

a

if

o
oi

11

p

o

bfl

O

r^

fQ

<1)

Oi

<

p.

^H

01

OJ

^

rH

v^

tfl

o

rC

O 3 ON

PLATE XII.

-l->

><

0)

OJ

o

(-<

c

o

•d

r-(

53

fe

03

a

1 — 1

6

03

,

S3 S <^

C M ^

M ,-1 c

t/3 >— I

o

CD

03
bo C

0) "<

fc

o3 <D" 03

<D b/3 3

i^ 03 ?^

(D 03 ^

*« ^- s

2 -Q

^ (Nl -^

C Q

^QQ- ^ OR

5 -'i ^'

University of
Connecticut

Libraries

39153029221233