llllll '

WI

7

”WW

I
|
|

626:
'THS:

 

SOME STUDIES IN
.HUMAN MATE SELECTION
AND HEREDITY 1N MICE

Thesis for the Degree of M- S.
DOROTHY PERMAR
‘929

.3

    

'50".
.o

>2“

- msgit'aww'

:
.'o

-.. .-¢.,..4,..... ,
,..,..,.-.A..-.....;u. .- -o... ... -
. a.o->4->o--n.4.._..l.. .... by .o-v . 0.. .
. .. .V». . .4.

m ti‘Wng; .- '. r -‘ = '

i _. _ . ,~ .‘ .1 ‘ L. ‘f,

I "' ' : fl; ' i: :g '

‘3‘? "I Hi In i; ,9: “a

h it ‘9' 6‘4» ' 1
f: " "*5, b: '. r‘.

L '. erMsj-fi. '

m:-
I’- ,

35}

I.

W! ."h

at»??? "mazmmmw’w

I . ‘. "
., nu. .
_)

‘fi. -' 3
w ;37“""

1

:gmeaWS-tcwqtm , W

.V‘ . ‘ N
. ‘ 9-6 ‘9
. 1" \ ft". ‘41,}.

3n'.-.---”
.n....:

o .mu-oow - .
, .u. v.. n o-

.. - v

... ‘- -

“‘5',“- .

. AWN?” -';'

any

. . '-!'Ifl.
~Q‘Jn'vi‘. “ '_I .' "“J.’ m4-

....

..._

T...
.,.-.._

9‘14; 3'

.1" -, '.
‘ - . . .
‘ A‘ ”-‘*Vt"qr.

,fl ‘ .- ‘;" ...
hm. m » . . . .-
. . . ~ s ,1 ‘x ' .

. ‘t ’ ‘b .. I. .‘...

.  §‘-”.’?‘-‘~.! 1 I

- . LWW“ .‘. 755' $8.57. .‘ .

.WN?» Us?"

‘a_.o.-

t,

I
4322;" -

93' W59,»

, -: .- - "
‘P _ _ .v'
4 .
- .-
‘.v .s "I . '-
1“ - a I ‘ . : I' J. '
u « _.
4 4 ._t . m . .' 431'.’
.1 ~ ‘ : A " ‘ L
. » l .
. , . _ n .
.
. I

.N

.u‘.

a
L

,.
‘

....;

‘ .VV‘W "a ‘0th '33-’44}...
f’e’fi"
v A.)\;*

Y“.

{‘3‘ A
T ‘5‘

f.

... .
O O O O. o
"-

.

>-_V
‘V
a l. 1.

- . .3 ‘V

 

(v ~-'.'w "~r'\ 'T‘T'WCI 1

DC ‘ ~ ‘

M -...J_J 8.1.. \JJ.L..4L) 1.
17‘0"“ ' 5mm YMMT’ ‘r
a»; ‘.,.A'.L_I - -¢s‘-L.LJ '4~1.L.41_4\J4.. .L $“

um

'ALQJ

1' ‘7‘... ‘ Trwfif I‘f V'T.\Y.:‘1
b.“v..'s 1"“. ‘) -....A.\JH

r3
v- 4
H—a
-4
‘J
L;-
H
C ’5

. . ~-1 . “ , I
r; L) .J- . ... .L ..4 4‘

'--".H-D’'."""""'w_:‘-'~ rm" f“"".‘ ‘1 ”“7? r1" 7.3 17?"
J.) Lk) A "‘ .L 4.4..L‘H4

'
$- ‘-

“TK‘W”IN 2'1 cvwr “1‘ (Nfi7’7"”'j I‘T w\‘>:wT.xT ‘“:*'i* "'f'1W'Vfi
‘.~+U ;‘»L‘,‘J1‘V MLAA .._.LJ |\/\J.J.JJJA4--‘I.J.J A'I ~"¢§L.— LIL; . \JIJJ.~‘ Mth-“J-.A.
"\.-9 "1' 4' "< ‘ -,_..4_‘ -‘ ' - r1.'4 “" “ ‘ r“v." " 1;.N‘WV1‘V" pw“
L l. s. '. :‘L— ~ ‘ [ . . , .l ~_p. ‘. .‘ K}.
.. 4.--..- -...4—‘,\.. $..J-_. *-.a."- - ..‘-. .5.-.4 -AJJ‘LHU .L
‘ — 1 -r‘fi Can t" W T"? i?
gJL-J.~JJ- ‘4... L4\/.~¢.4._‘I\JJ.J

V
&

L'J

DOILOTZ-fY I?BI{.7£-;l

.- v- ‘
1.413137 21) , 19:5

 

’T

r‘:" - r‘
.5" Y.‘ .7‘:

f ‘
§

177"": ”Mb
UL¢I£LJ$I$M
'5‘. "r1 ~ ' r'w‘fi <‘".‘**.r’s
L‘ALLJ- . ‘VJ-b ‘A‘J ‘ILJ—J—‘D
”“ “m '1' T
~ . I
*A‘.-\.L .L h D

f" i", ‘
A. IKv L‘
"'7‘,‘

1"

'fi' "r'fl‘TrV " ‘ ‘ "“ ‘
' I. I ,'
( . . , -
J 1-. .L-LV ULJ--L--\J 1.....-»
nr—rv 7'" ---,'1'vw
4 - | .
...,.).L.LJ .L‘i .. .4044

,P-1 w-
V41 ...'.LJ

‘ ‘q-~v

V\

5, - ...-A...J

~r‘. ‘ 1‘“
L41.-.‘ I

3.12 TB SELEC 'I‘ ION

Mate Selection

The purpose of this investigation was to determine
what kinds of peOple marry each other; whether pe0ple of
similar types tend to marry, or whether marriage is more
frequent between dissimilar individuals. It was here
possible to consider only a few characters, but it should
be recognized that there are innumerable characters which
QEILinfluence mate selection, provided it can be shown
that there is any definite selection exercised.

"Who marries who" is a question worthy of considera-
tion. Does a man choose for a wife a woman possessing
his own traits in a similar degree? Does a woman prefer
a husband who is much like herself? Or do men and women
marry persons of entirely different types? There are
many age old theories concerning this question, but few
efforts to prove or disprove these theories. And yet it
is important to know whether genius marries genius, or
whether a person of genius more often marries one of en-
tire lack of ability. Do persons of superior health
select for mates persons possessing this same attribute,
or do they marry persons inferior to themselves in this
respect? Do those of superior mentality marry each other,
or do they choose one of lesser mind for a mate? And if
extreme types gg_marry, are the offspring like one or the
other of the parents, or do they compose an "in between"
group? To state the question more generally, is there
a tendency toward the equalization of the human race,

toward the intermingling of the extreme classes of all

 

 

or some of the traits of this race to produce one inter-
mediate class of these traits; or is there a tendency to
produce classes which are more extreme than those we have,
eliminating gradation?

In this paper an attempt is made to answer a few of
these questions. To do this, it is necessary to consider
What factors may influence mate selection, and then, from
the data at hand, to determine whether or not these fact-
ors do influence selection, and if so, to what extent.

It is undoubtedly true that these factors are some-
what different in different countries and localities;
but we will consider those which might most frequently in-
fluence this choice in any locality where a considerable
degree of choice is permitted.

One factor Which should be considered is the sacial
status of the individuals marrying. In some EurOpean
countries this factor is undoubtedly effective; in this
country there may be some question as to its potency.
Another possible influence is that of family customs,
such as religion and occupation. -Personal traits and
talents, as for example, artistic ability and literary
interests, may be determining factors. Physical traits
of the individual, as the degree of pigmentation, size, and
health, should be considered as possible Operative ele-
ments. Or such circumstances as the age and education of
the persons marrying may be characteristics by which
selection is determined. There are innumerable factors
which may possible be influential, and which should be

recognized.

3.

From the data obtainable in the present investi-
gation, however, only a few of these factors may be stud-
ied.

The work was done at the Eugenics Record Office, loc-
ated at Cold Spring Harbor, Long Island, New York, under a
grant of money made by the Eugenics Research Association
of the Carnegie Institute of Washington. Appreciation is
expressed for the use of their archives, and for the many
courtesies and aiggestions which made the work possible.

The data were found in the Records of Family Traits.
These Records are printed blank forms which ask for certain
definite information concerning various members of a family.
The individuals to be described in the blank include child
number.1,2,3,4, etc., the father and mother of these child-
ren, the father's father, the father's mother, the mother's
father, and the mother's mother. There is also space for
information concerning the uncles and aunts of the child-
ren. The blank is made out mostly in tabular form, so
that it is perfectly clear what information is wanted.

Some of the questions for the father are stated as
follows:

Full name

Date of birth

Birthplace:town state or Country____
Education #7::
Residences, principal
Age at marriage .

Occupations at successive ages
If dead, cause of death and age at death __>

 

 

 

Questions pertaining to personal traits were listed

in the form of a table. A part of one table is as follows:

 

Father Mother

 

Age at which description
is given

 

Adult or present height
in inches (without shoes)

 

Color of hair (before graying)
a.(albino) f.(flaxen), etc.

 

 

 

 

 

Most of these records were filled out by college men and
women, or by other persons interested in eugenics; and they
have been collected from all parts of the United States.
About eight thousand records were examined, and data were
used from the 2300 records which appeared to be written up
most completely and carefully. No other selection was made.

In most cases the person who had filled out the record
held the place in that record as one of the children (child
number 1, 2, or 3, etc.); in a few cases he held the place
of 'Father"or "Mother". In this study the married persons
used for comparison of traits were those described in the
record as "Father" and "Mbther"; so that in the majority
of cases the individuals had been described by one of
their children, and in a few cases by one of the persons
themselves. For this reason, the data given should be
accurate, at least in those instances where the trait
considered is an easily measurable one, as age and.height.

Data were collected for the following characteristics:
age, education, height, hair color, eye color, mental
ability, musical ability, drawing ability, mechanical ab-
ility, general bodily energy, condition of‘sight (at time

of marriage), and condition of hearing.

5.

Some of these characters may be easily measure, and
should give a fairly accurate indication of their importance
in mate selection. This is true of age and height, parti-
OUlarly, as they may be meaaired in years and inches without
question. Education may be slightly more difficult, since
the actual number of years spent in school, or the grade
which the individual finally reached, is not always in-
dicative of the amount of his education. However, when
nearly 2000 cases are considered, the results should be
of some value, for in the majority of cases the college
man or woman is better educated than the man or woman
whose schooling ended with the eighth grade of grammer
school, or perhaps earlier. Hair color and eye color are
difficult to measure without a color diart, and there were
undoubtedly many different ideas of blue and brown eyes
put into the records. There is one fact which should not
be overlooked, however; this is a comparison of two in-
dividuals; and since the two individuals concerned were,
in all cases, described by the same person, only one
person's classification of degrees of color enters into
any one comparison, a: that any difference indicated
between individuals is probably real. This is-true also
of the very poorly measured traits of mental, musicsl,
drawing, and mechanical abilities. The conditions of
sight and hearing were probably classified with reason-

able accuracy, and will be discussed later.

Age

An effort was made to determine whether or not age
influenced mate selection, and if so, how much. Do men
and women of about the same age most frequently marry,
or is a man of twenty-one as apt to marry a woman of
thirty or thirty-five as he is a woman of twenty or
twenty-one?

In the blanks to be filled out in the Records of
Family Traits, the information was asked for as follows:
under "Father" the question was put "Age at marriage";
under "Mother" it was "Date of marriage". The date of
birth was given in both cases. The age of marriage of
the mother was found by subtracting "Date of birth" from
”Date of marriage". The accuracy of the dates was check—
ed by subtracting the "Date of birth” of the father from
the "Date of marriage” of the mother, and comparing it
with the given "Age of marriage" of the father.

The age at which the two individuals were married
was given in 2212 cases. In tabulating the data the
individuals were classified as follows: those married at
sixteen years of age and under, those married at seventeen,
eighteen, nineteen,etc., years, up to thirty-eight years,
this latter group including those married at thirty-
eight years and over. Table I.

The coefficient of correlation between the ages of
the men and women married is r . l fi_§y_ equals

2
.57 i .0097. °y

C‘J
r-i

i:.

.-o .-

 

>

 

 

'

 

- A1-_

I

4“" r“-

 

H‘-

r’"‘1r“

 

 

 

 

- -~1>—---—4>-—~—>--—~
_l}—M 1}-__. *4} .—
- - A A. -L. AJL. -

 

 

 

.4 w. 9‘ o — fir.— o-1P—o—«ro. —

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

H

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

v . _ _
u |i A . A w
a A . A A A A
. LA A A A A .r A
m H _ . A A . A _ A . A A .
{RIII .l v A A . w H 1" o A AI
a H N A AH A .. A AH A . A . A A A A
In!!! .1 r} A! +1 . A lollA1 F A Li L. L t _F .v
if Am A AAAAHAANAAA AA
A. + A . A. A J. 1 1.1 A A LA A Ar 4 . A
M. An S a H A A AH Ar A A AH A Am AH _ L. H . AT- A A
a T Am a H AH _ W A AH AH A A A A A. A A rH A A m
. AH w _ _ L. AH! A J. . A H A Ht w H
u A. A n H H H An Am. A. H A H .9 Am A A A .H . A A .
r30 .1w 1 i . i + u: at i , . H i 1 H
m AH A.m w A m A m 5 Am Am H. AH A m A A A A A A A
utoAY'il. luluoiAbniiil : + .r L L _ 1m In» I” J. H» 4;“ r
a A . e a A 1 mH As Aw .m Am .m H AH A _ A
H . A w . A--;.“. h . h -: A v
0 AH m A N .m Am .m [A H Am Am A;H .m m n . A A
A r A L.r A A h. A A A __ A P
. A a t. -A . i A e--- .r 1.0.4.1...- int“! A
Am An A m e A n Ae mWy e AoH AmH As .Am Ae _n . A A A
r n It III 1 «r A A L in! lillt It}. a L J. Aw
@ r An n m As :H HH AmH ,Amm AeH .n m ,n A A .
TIO+I+ s riwvlilrit. 1r} i 14». . LI 9.» L. H!
m AH A m e a As As ANH AmH AHN Ha Asa Amn Am m AH A A
I.» A A rib; L H . III. A A. b
q 4 1 w _ l A _
Aw m. A n m “A Amt Am Am . on Am . AHm was A An .AH A
II. 1.9. . w w A : r J 1 . J7» L
.m A A m H a AN Ae As AOL AHm was As AH e A m
AI W . II? Ill!» 0” F . ilk: J L
As AH m m N Am Aw Am AmH AHa Aon ANH Am m A H.
A r a J. T L. J. p
m A n a As Am An _ Aem Ass Ann AAH AA A
w . w w atiTs .311:lhziL:ILI [A J 1L L:¥i
n m Ae . .m As A _mm Amm .Aam Amm AHH A
r A + t, .+ + . w- A _ A (w is
. Am An .m A .mH .nH Ame AAH Aom Am A
. A H A t A A w A- L A H
w H AH n .e w .oH AeH AAA .uH JAN“ AeH A
H .r A . A A A _ .
14 J . 11 H 4 J a
A As .AH A H Am An AuH . AAH An AH
A A. A A A A. . + A
A AH m .H AH Am H Am .m
.r fit A . L _ A
A m H A A 1IA A
7; an m Ha A-n A .. an Asa .ma «a AHn Age
1 H 4% . A

 

 

 

 

 

 

 

 

 

 

 

 

 

 

7.

The coefficient of correlation is 58 times its probable
error, and is significant. It may be concluded that men
and women near the same age more frequently marry than

men and women of widely different ages. age seems to play
an important part in the selection of a mate, which is

what would, after all, be eXpected.

Height

The measurement of height is quite as easily made as
that of age. In this case, the men were classified in
groups as sixty inches,and under, sixty-one inches, sixty-
two inches, sixty-three inches, up to seventy-eight inches.
The women were classified from fifty-eight inches and under,
up to seventy-two inches. Table II.

The heights of persons marrying were given in 1450
cases. The coefficient of correlation is .48 1 .013.
The coefficient of correlation is 37 times its probable
error, and is significant. Men and women of similar
heights more frequently marry than those of Opposite ex-
treme heights, and therefore height may be considered
as a factor in mate selection.

The correlation here is greater than that obtained
by Pearson. In his study of 200 cases, he found
r . .0951 I .0473. His conclusion is that "although the
probable error is about half’the coefficient of correla-
tion, it is unlikely that the latter can really be zero....
we are Justified in considering that there is a definite

amount of assortive mating with regard to height going

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

000000000000. 00100 0000.000 0.040 000-4100 00010.. H
A
.H H flu.
r. #001001r Lj AW I .004 Iv 00004
H H A A Hp
00 A l 00 00.. - 00 00.40601
A 4A-0 a n H n H _ H A as
.A.00-A-- 0. LA. 00 00 AT 000 0A.. 00 IIIII 0000 00000.00 - A1000 0A.. 00 0 0 .0A
_ A
A n. ._. o m H. H. H H A A m)
00 .F - L ,00%0
r0 0 0 #00 .H! 0100f 0M .A
_ \4 \l
H A A . o 3. HH H A... o H H H on
.00. At. 0000 .l 01.: 4:0. 00
u A... u .A. H ”H H an _.. H 2 u u H A.)
+ - 00r. 000 0
H H q n 3 HH .. an at an a. AH a m H H on
w a 0 HH .n . 1. He on a: aH e H a.
n o . nzA HaA .. a. a. a. an nH a H H a.
A00.0 l fi00 -0 00.00. 0 00- 00A- 0 00 0.0 T000 70-0 0:0 0 0.0 00 0001.00 000A. H 00 0A
H n . AH aH m. 9H an n2 9. AH 0H 0 H n»
. 0 00 0- 0.0-00.1- 0 0 .. 0 0 0
A s .0 A o .H 22 N. H n a H .3. m. a n H. .1...
AT0 0r. 4.00
H H H -. n ...H a a... 3H HH H. H n AH A. Ho
00-- L1 00 - 0- 000-0 1 0 a 0
A H a. «H A! H H 0 J H. a ..... n H... 9.. .
0- - 0.0A .0 0.1000A I- 0 AA00 0
H n H H H H a H as
I0 0 L100 0.. 0.00.00 000A. 0040 0 01.00 0.0100001.
H e n. .... c. H H H a. no
H0 000...-.- 0...- .. - 0. -- .0- 0A.
3.... E. as. 3. MA. 3A. 4...... HA. 2. 1. 3b a- A...) H... H.) 3A .5 H» on
J-
.4113
. f .0 .
.410 4.). 1 0,. 1.. a x... j..w.4w..a .0. .) 0. .4... 4.00- 44 x0444.

I”? won

on in the middle class.”

Education

It is perhaps to be eXpected that the amount of
education of individuals will affect mate selection.
The data given in these Records of Family Traits may
throw some light on the question. The amount of educa-
tion of both parents was definitely stated in 1988 cases.
Certain grouping of individuals was necessary, as diff-
erent kinds of schools may, in the end, turn out persons
of essentially the same degree of learning.

The types of schools were grouped as follows:
Common school plus high school = high school
Common school plus business college a high school
Common school plus normal school = high school
High school plus normal school = college
High school plus college = college

The records which for "Education" gave only normal
school, business college, etc. without designating common
school or high school as preceding this, were not included
for there was no way of determining whether or not the

individual had attended high school. Table III.
Table III. Amount of Education.

 

: None E?fi??? sghggl College Total

 

 

 

 

 

 

None 10 14 1 4 29
Common‘school ‘016 659 164 99 918
High school 4 185 285 226 698
College 0 44 54 245 343
Total 30 882 502 574 i988

 

 

 

 

 

 

 

 

The coefficient of contingency, C = _§_§HE_ , is

found to be .671. The greatest possible value for C fbr
a symmetrical 4 x 4 fold table is .866. {See footnote).

The coefficient of correlation is .53 t .107.

Education, then, is indicated as being influencial
in mate selection.

1'he nest question is, just how much effect does this
factor have within each diass of persons? It will be
necessary to consider each class separately.

Upon a closer examination of the table it will be
noticed that there is an.extraordinarily large number of
college and high school persons, much larger than is
found in a random sample of the pepulation of the United
States. It will be recalled that the Records of Family
Traits from which these data were collected were filled
out by college men and women and by other persons inter-
ested in eugenics, and that the individuals here com-
pared are, in most cases, the parents of the recorder.
This will eXplain the reasan for the large percent of
well educated persons found here; for it is reasonable
to assume that in a group of parents of college students,
there are more high school and college graduates than in
the parents of all the pe0ple of the country between the

ages of eighteen and twenty-two.

Footnote: This and all of the following values were ob-
tained from G. Udney Yule - Introduction to the Theory
of Statistics, 7th edition, pp. 55, 55.

10

It is impossible to tell from the data here given,
how much this unfortunate selection affects our conclu-
sions. It is probable, however, that the mate selection
found here is strongly indicative of the selection for

the entire pOpulation.

Uneducated men:

We will begin with the group of uneducated men --
that is, men with no school training -- and see from
which classes they have selected their wives; which will
tell also, of course, which classes of women have select-
ed this type of man.

There are 1988 women in the entire p0pulation stud-
ied; 29 of them are without education, Which is 1.45%
of 1988. (Table III). If there are 1.46% uneducated
women in the total poPulation, and if marriage takes
place at random, one would eXpect to find 1.46% of un-
educated women among the wives of uneducated men. There
are 50 uneducated men, and of the 50 wives of these men,
10 are uneducated, or 55.55%. The probable errors of
these percents, the difference, and the probable error
of the difference are as follows:

Observed percent of uneducated women 33.33% t 5.80%
among the wives of uneducated men:

 

EXpected percent of uneducated women 1.46% i .18%
among the wives of uneducated men:
Difference 51.87% t 5.50%

The difference is five and one-half times its

probable error, and is therefiare significant. And we

11.

have found that there are 51.87% t 5.80% more uneducated
women among the wives of’uneducated men than was expect-
ed. The selection here is positive and rather extensive.
Secondly, how frequently do uneducated men marry

women of common school education? Of the entire female
pepulation of 1988, there are 918 having a common school
education, or 46.17%. With random marriage there should
be 45.17% of the wives of the 30 uneducated men with a
common school education. .There are 16 such wives, Which
is 53.33% of so.

Observed percent of women of common school 55.55% t 6.14%
education among wives of uneducated men:

Expected percent of women of common school 46.17% 1 .76%
education among wives of uneducated men:

Difference 7.15% i 5.15%

 

The difference is not significant. There is about
the eXpected number, assuming random marriage, of women
of common school education among the wives of uneducated
men.

Thirdly, do women of high school education number
largely among the wives of this group of men? There are
698 women of high school education in the total papulation
of 1988, which is 55.11%. Four of them are wives of un-
educated men. Then only 15.55% of the wives of uneducated

men are of high school education.

 

Expected percent of high school 35.11% 1 .72%
women among the wives of uneducated men:
Observed percent of high school 15.55% t 4.18%
women among the wives of uneducated men:
Difference 21.78% i 4.24%

12.

The difference is 21.75% i 4.24% and is significant.
High school women are evidently selected against by un-
educated men, or vice versa.

The fact that n3 college women have married uneducated
men is conspicuous. Since 17.26% of the total pOpulation
of women are of college education, the same percent, in
random marriage, should be the wives of uneducated men.
But we obviously do not have random marriage. We find
that the 17.26% of expected college women, and the 21.78%
of the exPected high school women, who are missing among
the wives of uneducated men, are replaced by the 51.87%
of uneducated and the 7.16% of common school women who
are in excess of the eXpected percents from these two

latter groups.

Common School Men:

The second group of men to be considered is that of
men of common school education. How frequently do men
of common school education marry women having no educa—
tion? As noted above, in any group of men we would
eXpect to find 1.46% of their wives uneducated if there
were random marriage. There are 882 men of common
school education, and of their wives, 14 are uneducated.
This is 1.59%. '

Observed percent of uneducated women 1.59% f .28%
among wives of common school men:

 

Expected percent of uneducated women 1.46% i .18%
among wives of common school men:
Difference .13% i .33%

15.

The difference is very small, and is not significant.
We may conclude, then, that uneducated women are selected
neither for nor against by men of common school education.

How frequently d) men of common school education
marry women of common school education? We would eXpect
46.17% of the wives of men of common school edication to
have this same degree of education. Actually there are
72.45% of such wives.

Observed percent of common school women 72.45% i 1.01%
among wives of common school men:

Expected percent of common school women 46.17% - .76%
among wives of common school men:

Difference 25. 25% I 1.25%

 

The difference is large, and is significant. There
are many more women of common school education among the
twives of men of common sdiool education than would be
expected in random marriage.

The number of women of high school education marry-
ing men of common school education gives a different
result. Among the wives of this group of men we may
eXpect 55.11% to be of high school education if we have
random marriage. Instead of this, we find only 20.97%.

Expected percent of high school women 55.11% i .72%
. among wives of common school men:

Observed percent of high school women 20.97% + .55%
among wives of common school men:

Difference 14.14% t 1.14%

 

The difference, 14.14%, is significant, beiig
thirteen times its probable error. There are fewer

wives having high school education than are expected.

14.
If there were random marriage, we should find among

the wives of common school men 17.26% of college education.

Actually we find only 4.99%.

 

Expected percent of college women among 17.26% i .57%
wives of common school men:

Observed percent of college women among 4.99% i .49%
wives of common school men: fl
Difference 12.27% t .75%

The difference is sixteen times its probable error
and is significant.

.It is evident that menof common school education
select women of similar education for wives, rather than

women of superior learning.

High School Men:

such selection is also carried out by men of the
high school group. There are,in the pOpulation studied,
502 men of high school education, one of whom has
married an uneducated woman.

Expected percent of uneducated women 1. 46% 1 .18%
among wives of high school men:

Observed percent of uneducated women .20% 3 .15%
among wives of high school men:

 

fir

Difference . , 1.26% 1 .22%
The difference is 5.7 times its probable error,
and is significant. High school men tend to select
against uneducated women for wives.
The expected number of women of common school
education among the wives of high school men surpasses

the actual number:

15.

Expected percent of common school 46.17% t .75%
women among wives of high school men;

Observed percent of common school 52.67%u; 1.55%
women among wives of high school men: _

Difference 13. 5o% . 1.54%
The difference is significant, being nine times its
probable error. High school men select against women of
- lesser education.
But high school men more frequently marry high school

W omen e

H-

Observed percent of high school women 56.57% 1.48%

among wives of high school men:

.9,

Expected percent of high school women 55.11% - .72%
annng wives of high school men:

Difference 21.26% t 1.64%

The difference is twelve times its probable error
and is significant. The selection is positive, and quite
extensive.

The selection is seen to be negative for the marriage
of high school men with college women.

Expected percent of college women among 17.26% .57%'

wives of high school men:

H-

Observed percent of college women among 10.75% a .94%
wives of high school men:

 

Difference . 6.51% z 1.09%
The difference is nearly six times its probable
error and is significant.
It may be concluded that high school men select for
wives women of their own educational class rathem than

those of classes above or below it.

16.
College Men: _ .

The last group is that of men having a college educa—
tion. Do men of college education marry women who are
uneducated? The date here give results which are unex-
pected in the light of what has Just been found to be
true of all the other groups. The expected percent of
uneducated wives in this group if there were random
marriage is 1.46%. There are found to be four such wives,
which is .7%.

Expected percent of uneducated women 1.45% i .15%
among wives of college men:

Observed percent of uneducated women .69% I .24%
among wives of college men:

 

Difference .77% : .30%

While this observed percent is smaller than the
expected, the difference is less than three times its
probable error, and is not significant. The date do not
snow the exPected significant selection against uneduc-
ated women by college men. The writer finds nothing in
the data to account for this circumstance.

The selection by these men against women of common

school education is about what would be anticipated.

 

Expected percent of common school 45.17% +— .75%
women among wives of college men:

Observed percent of common school 17.25% 1 1.05%
women among wives of college men: __

Difference 25.92% : 1.32%

The difference is significant.

High school women numb er more frequently as the

wives of college men.

l7.

 

Observed percent of high school 59.58% 1 1.55%
women among wives of college men:
Expected percent of high school 55.11% 1 .72%
women among wives of college men:
Difference 4.27% 1 1.55%

The difference is positive, showing more than the
exPected number, andis nearly three times its probable
error. The deficiency of uneducated and common school
women found in the wives of college men is partly made
up by the excess of high school women.

' The difference between the expected and observed
percents of college women among'the wives of this group

of men makes up the remainder of the preceding defic-

 

iencies.

Observed percent of college women 42.68% i 1.42%
among wives of college men: g

Expected percent of college women 17.26% i .57%
among wives of college men: ,

Difference 25.42% 1 1.55%

The difference is eighteen times its probable error
and is significant. College men frequently select
college women for wives.

Conclusion:

Education is a very effective factor in mate select-
ion. It affects all classes of persons similarly. Men
and women tend to choose for mates persons of their own
educational class rather than persons of a widely diff-

erent degree of schooling.

18.

Hair Color

Does hair color in any way influence mate selection?
The color of the hair, before graying, was given for 2074
sets of married persons. 'In the Records of Family Traits
the question was etated,
Color of hair (before graying). a.(albino,white),f.{flaxen),
y.br.(yellow-brown), l.br.(light brown), m.br.(medium-
brown), d.br.(dark brown), bl.(black), cl.r.(clear red),
d.r.(dark red), d.br.r.(dark brown red).

_ For purpose of this work, these classes were grouped

_as follows:

f., y.br. - blond

l.br., m.br. - brown

d.br., bl. + black

cl.r., d.r., d.br.r. s red

The main purpose of this grouping was to attain
greater accuracy in classification. It is obvious that
data collected in this manner will not be uniform for
such characters as color of hair and eyes. It was be-
lieved that by thus grouping the date, the smaller
variations in the Opinions of different shades of'color
might be eliminated. For instance, ore person might
classify a man's hair as light brown, while another per-
son might put the same man's hair under the class of
medium brown. However, it is not likely that is a
hundred persona classified this particular hair any of
them would make the gross error of calling it flaxen, or
black, or red; at least, such cases would probably be so

few as to make little difference in so large a group.

19.

Another factor which greatly decreases the inaccuracy
of the classification of this character is that stated
above: the two individuals compared were, in all cases,
described by the same person, and so only one person's
ideas of degrees of color enters into any one comparison.

While the data have not been collected in an entirely

satisfactory manner, the results should give some fair

indication of how important hair color is in mate selection.

 

 

 

 

 

 

Table IV. Hair Color MEN
Blond Brown Black Red Total
Blond 7 27 67 5 104
53 Brown 20 217 417 15 667
:3 Black 59 571 742 42 1214
Red‘ W‘ 5 19 65 2 89
Total 91 654 1289 60 2074

 

 

 

 

 

 

 

 

The coefficient of contingency is 0.081, the highest
possible value being 0.866. Table IV.

In this case 0 is so low that there is no indication
that hair color is effective in mate selection.

f When these data are treated as those given above for
education, it is found that in only one case is there a
significant difference between the observed number and
the number expected in any one group assuming that hair
color does not affect mate selection, and that marriage
is at random. That is, the difference betweenthe observed
and the expected is in all but one case considerably less
than three times its probable error. There are 10.19% i

2.9% fewer brown haired women snong the wives of blond

_ 20.
men than is expected in random marriage.
Conclusion.

It may be concluded that the color of hair has very
little to do with determining "who marries who"; it is not

an effective factor in mate selection.

Eye Color

The color of the eyes of the married persons was given
for 1941 marriages. In the records eye color was classi-
fied in seven groups: p.bl.(pa1e blue), d.b1.(dark blue),
bl.br.(blue with brown spots). l.br.(light brown).
d.br.(dark brown), bl.(black), y.bl.(yellow-b1ue,gray or
green), r.br.(reddish brown).

Since it is difficult to distinguish between such

colors as pale blue and dark blue, and black and dark

brown, the classes were grouped as follows:

 

 

 

 

 

Table V. Eye Color

Blue Brown Black Total
Blue 657 112 259 1028
Brown 158 56 55 269
Black 574 68 202 644
Total 1189 356 516 1941

 

 

 

 

 

 

 

peble,'deble, yeble ’ 131110

blebre. lebre, rebre‘ brown

d.br., b1. . black:

The coefficient of contingency is 0.127, and the
highest possible value for C for 5 x 5 fold table is
0.816. (See Footnote). Eye color apparently has little
Note: G. Udney Yule - Introduction to the Theory of Stat-

ietics.

 

I u
0
. u
o e A
o .
n
e I
a
.
0 ~
g
—
, a

, 21.
effect in mate selection.

The data may be treated in the same manner as those
given above for education. We may find the expected per-
cent of any type of woman in sad: group, assuming that for
this factor marriage takes place at random, and then find
the actual percent. If the difference between these two
percents is more than three times its probable error it
may be considered significant.

Group I. Blue eyed men:

 

Observed percent of blue eyed women 55.26% 1 .94%
among wives of blue eyed men:
Expected percent of blue eyed nomen 52.96% t .77%
snong wives of blue eyed men:
Difference 2.50% t 1.21%
Not significant.
Expected percent of brown eyed women 15.85% t .55%
among wives of blue eyed men:
Observed percent of brown eyed women 15.29% i .6876

among wives of blue
Difference

Expected percent of
among wives of blue

Observed percent of
among wives of blue

Difference

Group II.

Expected percent of

eyed men:

 

Not significant.

black eyed wommn 55.18% I .74%
eyed men:
black eyed women 51.45% i .88%

eyed men:

 

1.75% t 1.15%
Not significant.

Brown eyed men:

blue eyed women 52.96% t .77%

among wives of brown eyed men:

Observed percent of blue'eyed women 47.46% I 2.16%
among wives of brown eyed men:

 

—__7

5.5 A : 2.29%
Not significant.

Difference

Observed percent of brown

among wives of brown eyed men:

Expected percent of brown

among wives of brown eyed men:

Difference

eyed women 25.75% 1 1.88%
eyed women 15.85% 1 .55%

 

9.88% 1 1.94%

The difference is five times its probable error,

and is significant.
Expected percent of black

eyed women 55.18% 1 .74%

among wives of brown eyed men:

Observed percent of black
among wives of brown eyed

Difference

Group III. Black eyed men:

Expected percent of blue eyed women
among wives of black eyed men:

28.81% 1 1.96%

4.57% t 2.09%
Not significant.

eyed women
men:

Observed percent of blue eyed women

among wives of black eyed
Difference

Expected percent of brown
among wives of black eyed

Observed percent of brown
among wives of black eyed

Differs nce
Observed percent of black
among wives of black eyed

Expected percent of black
among wives of black eyed

Difference

52.96% 1 .77%
men: 50.19% + 1.48%

 

2.77% I 1.67%
Net significant.

eyed women 15.85% t .55%
men:
eyed women 13.66% I .94%

men:

 

5.19% 1 1.08%
Not significant.
eyed women 59.14% 1 1.42%
men:
.74%

eyed women 55.18% ;

men:

 

5.96% : 1.60%

The difference is 5.7 times its probable error,

and is significent.

22.

23.

The difference is three or more times its probable
error in only two cases: that for marriages of brown eyed
men with brown eyed women, and that for marriages of black
eyed men with black eyed women. In both of these cases
there are more than the eXpected number of such marriages.
In the case of the marriage of blue eyed men with blue
eyed women, we find more than the eXpected number of blue
eyed women among wives of blue eyed men, but the difference
between the observed and expected is not significant. In
all other cases -- that is, in all cases except those of
the marriage of similar types with each other -- we find
that the observed number is less than the expected. In
none of these cases, however, is the difference signifi-
cant.

Conclusion:

The data above given do not prove that marriage
takes place at random so far as the eye color of the
persons marrying is concerned; neither do they prove
conclusively that matings are definitely determined by
oye color in many cases. ’However, it indicates that eye
color may possibly influence mate selection to a limited
degree.

This is in accordance with Pearson's conclusion of
the effect of eye color in Mate Selection. ”The corre-
lation between husband and wife for two very divirgent
characters" (referring to height and hair color) "is
thus shown to be about .1, or is 25 percent greater than
is reguired between first cousins°by the law of ancest-

° .. r , ‘ :‘-\ r
ll 10‘. QOQe ifOCe, J01. 042., £0 4:100"

24.

ral heredity."

General Bodily Energy
'“_In the hecord of Family Traits the degrees of
enerty, or activity, were classified as follows:
General bodily energy. 1. (very inactive). 2.(ordi-
nary). 5. (exceptionally energetic). Table VI.
Table VI. General l1‘odily Energy

 

 

 

 

 

 

 

 

 

 

 

l 2 5 Total

‘1 7 14 9 50

2 14 612 592 1018

5 21 458 656 1155
Total 42 1084 1057 2185

 

This is a very difficult character to classify,
and the accuracy of the data is questionable. However,
it is interesting to see whether or not there is any
indication of selection for this character in the data
obtained.

The actuel percent of group 1 woman found among
the wives of group 1 men exceeds the percent expected
in random selection by 15.29% 1 5.85%. The difference
is significant.

The difference between the expected and the observed
percents of group 2 women among the wives of group 1 men
is not significant.

Group 5 women number among the wives of group 1 men
about as expected in random selection.

There is apparently no selection for or against

 

_ 25.
women of group 1 by group 2 men.

The observed percent of group 2 women among the wives
of group 2 men exceeds the expected by 9.85% 1 1.25%.

This is significant. ,

The selection of group 5 women by group 2 men is
negative. The difference is 9.64% 1 .25% and is signi-
ficant. The expected is greater than the observed.

The marriage of group 1 women and group 5 men seems to
berandom:

Expected group 1 women among 1.57% 1.17%
wives of group 5 men:

Observed percent group 1 women .85% 1.18%
among wives of group 5 men:

 

Difference .52% 1 .25%

The difference is not significant. In view of the
seeming selection in the other groups, we might well
expect selection here.

PCroup 5 men seem to select against group 2 women;
the expected percent of group 2 women among the wives of
group 5 men exceeds the observed by 9.55% I 1.25%.

-4 positive selection is seen for the marriage of
group 5 women and group 5 men. The observed exceeds the
eXpected by 13.07% s 1.25%.

The coefficient of contingency'is .26, the highest
possible value being .816.

The coefficient of correlation is .194 I .014.

The data indicate the marriage of similar types to
be slightly more general than the marriage of extreme~

. types, but the nature of the data does not warrant any

26.
definite conclusion concerning the effect of the char-

acter of general bodily energy in mate selection.

mental Ability.

The Records of Family Traits gave data concerning
the general mental ability of the individuals married.
The data, however, are far from being satisfactory. It
is extremely difficult to accurately measure mental abil-
ity by the most carefully arranged mental tests; but far
less desirable are data obtained from the estimate of
the mental ability of any set of parents by the children
of those parents. It is readily admitted that the pres-
ent data can do no more, at best, than give a very rough
estimation of whether unusually capable and quick in-
dividuals tend to marry persons who are also of active
mind, or whether marriage much more frequently takes
place between persons of widely different ability.

The degrees of ability were classified in the Re-
cords as follows:

General «lentai ability. 1. (poor; failure to advance at
school). 2. (medium to good). 5. (exceptionally good).

 

 

 

 

 

 

Table VII. Mental e ility
l 2 5 Total
11 pg, 8 2 .49
2 12 857 467 1516
5 4 226 595 825
Total 20 1071 1064 2155

 

 

 

 

 

 

 

‘The coefficient of contingency is .595. The high-
est possible value is .816.
The coefficient of correlation is .56 1 .012.

_ 27.
Assuming that marriage takes place at random and
that this character does not influence selection, we may
find here certain eXpected numbers of individuals for
each group. These numbers may be compared with the
actual number found in each group, as has previously been

done for the other characters.

I

 

Observed percent of 1 women among 23.00% 1 6.00%
the wives of 1 men:

Expected percent of 1 women among .65% z .11%
the wives of 1 men: 4_k

Difference 19.55% i 6.00%

The difference is more than three times its probable
error and is significant; but the numbers are very small;

there are only 23 group 1 men to consider.

+

Expected percent of 2 women among 61.06% - .74%
the wives of 1 men:

Observed percent of 2 women among 60.00% x 2.22%
the wives of'l men:

 

Difference 1.06% z 2.52%
Hot significant.

EXpected perceit of 5 women among 58.28% I .69%
the wives of 1 men:

Observed percent of 5 women among 20.00% i 6.00%
the wives of 1 men:

 

Difference _ 18.28% I 6.03%
The difference is three times it probable error
and is significant.

Observed percent of 1 women among .74% i 1.74%
the wives of 2 men:

Expected percent of 1 women among .65% i .11%
the wives of 2 men:

Difference .09% t 1.70%
Rot significant.

28

Observed percent of 2 women among the 78.15% 1 .7l%

wives of 2 men:

+

Expected percait of 2 women among the 61.06% _ .74%

wives of 2 men:
Difference

. The difference
is significant.

E1pected percent of
the wives of 2 men:

Observed percent of
the wives of 5 men:

Difference
The difference

and is significant.

Expected percent of 1 women among .65%

the wives of 5 men:

Observed percent of 1 women among .19%

the wives of 5 men:
Difference
The difference

and is significant.

 

17.09% i 1.01%

is 16 times its probable error and

5 woman among 88.28% I .69%

5 women among 21.1% I .81%

 

17.18% : 1.08%

is sixteen times its probable error

.11%

'4-

0+

.08%

 

.48% x .14%

is 5.4 times its probable error

EXpected percent of 2 women among 61.06% 1 .74%
the wives of 5 men:
Observed percent of 2 women among 45.89% I 1.01%

the wives of 5 men:
Difference
The difference

istsignificant.

the wives ofp5 men:

17.17% : 1.25%

is 15 times its probable error and

Observed percent of 5 women among 55.92% 2 1.0193
5 women among 58.28% I .69%

Expected percent of
the wives of 5 men:

Difference

 

17.64% i 1.22%

29.

The difference is 14 times its probable error and
is significant.

There is a significant excess of’group 1 women
among the wives of group 1 men, of group 2 women among
the wives of group 2 men, and of group 5 women among the
wives of group 5.men.

There is a significant lack of group 5 women among
the wives of group 2 men, ef group 1 women among the
wives of group 5 men, of group 2 women among the wives of
group 5 men, and of group 5 nomen among the wives of group
1 men.

The other differences are not significant.

The data indicate the marriage of similar individ-
uals to be more frequent than of dissimilar individuals,
but, as previously pointed out, any definite conclusions
based on these data would be unsound, as the data are

not sufficiently accurate.

Musical Ability
ability in vocal music was classified as follows:
1. (poor), 2. (Medium to good), 5. (exceptionally good).
Estimates of such ability were given in this manner for
I the classification

1992 marriages. No tests were given;

was purely an estimate.

 

 

 

 

 

Table VIII' Musical Ability MEN
1 2 5 Total
1 522 232 32 586
53 e a 415 625 105 1145
C: -
‘ z 109 102 50 261
Total “AA 959 187 one

 

 

 

 

 

 

50.

The coefficient of contingency is .252, the highest
possible value being .816.

For the marriage of group 1 women and group 1 men
the observed number exceeds the number expected in
case of random marriage by 8.64% I 1.24%“ which is signi-
ficant. indicating the selection of similar types.

. The the marriage of group 2 women with grOUp 1 men,
the expected exceeds the observed by 8.45% I 1.57%.

The difference is significant, indicating that group 1
men select against group 2 women for wives, or vice versa.

The eXpected number of'group 5 women among the wives
of group 1 men exceeds the observed by .25% I .84%. The
difference is not significant.

Group 2 men select against group 1 wome.. The ob-
served percent is 5.25% z 1.16% less than the eXpected.
The difference is significant.

Group 2 men select more frequently group 2 women
for wives. The observed percent of such wives exceeds
the expected by 7.69% 1 1.25%. The difference is signi-
ficant.

The expected number of'group 5 women among the wives
of group 2 men exceeds the observed by 2.47% I .85%.

The difference is a little less than three times its
probable error, but may possibly be considered signifi-
cant.

Group 5 men select against group 1 women, the ex-

_ 51.
pected number exceeding the observed by 12. % 1 2.02%.
The difference is significant.

The eXpected percent of group 2 women among the wives
of group 3 men exceeds the observed by 1.33% i 2.21%.

This is not significant.

Group 5 women hunt or largely among the wives of
group 3 men, the observed exceeding the eXpected by 13.63% i
2.20%, and is significant.

The coefficient of correlation is .15 i .012.

The data indicate that persons of similar ability in
vocal music tend to marry a little more frequently than
persons of dissimilar ability, but the nature of the data
permits no definite conclusion to be drawn. In spite of the
unsatisfactory data, it is probable that musical talent

and interest do play some part in mate selection.

Drawing and Coloring

Ability in drawing and coloring was classified 1.
(poor), 2. (medium to godi), 5. (exceptionally good).
Again there were no tests given, and the value of the
data is questionable due to the probable inaccuracy
of classification in a number of cases. However, the
data should serve to indicate whether or‘not persons
of great interest and activity in such art tend to marry
each other more frequently than they marry persons
entirely lacking in such interest and ability.
Table IX.

32.

 

 

 

 

 

 

Table IX. Drawing and Coloring
l 2 5 total
1 646 245 65 954
2 272 55 55 650
5 74 72 42 188
Total 992 J 642 158 1792

 

 

 

 

 

 

The coefficient of contingency is .29, the highest
possible value being .816.

The coefficient of correlation is .255 j .015.

The observed group 1 women found among the wives of
group 1 men exceeds those expected in case of random marri-
age for this character by 11.89% I 1.5%. The difference
is significant.

The expected group 2 women among the wives of group
1 men exceeds the observed by 8.85% 1 1.2%. The difference
is significant.

The emiected group 5 women among the wives of group 1
men exceeds the observed by 5.04% i .74%. The difference
is significant.

The OXPEOtOd groupl women among the wives of group
2 men exceeds the observed by 15.07% i 1.5%. The difference
is significant.

The observed group 2 women among the wives of group
2 men exceeds the expected by 14.55% i 1.5%. The difference
is significant.

The observed group 5 women anwng the wives of group

+

2 men exceeds the eXpected by .72% - .94%. The difference

is not significant.

53.

The expected group 1 women among the wives of group 5
men exceeds the observed by 13.36% 1 2.7 %. The difference
is significant.

The expected group 2 women among the wives of group 5
men exceeds the observed by 2.73% 1 2.6%. The difference
is not significant.

The observed group 5 women among the wives of group
3 men exceeds the expected by 16.08% 1 2.3%. The differ-
ence is significant.

The date might seem to indicate that mate selection
is influenced by interest and ability in drawing and
coloring, but this cannot be considered conclusive, due

to the inaccurate nature of the date.

Mechanical ability.

This character was listed simply as "mechanical
skill", to be classified in the same manner as musical
ability: 1. (poor). 2. (Medium to good). 5. (Exception-
ally good).

There were 1698 marriages so classified. Theco-
efficient of contingency is .186, the highest possible
value being .816. _

The coefficient of correlation is .15 i .016.

Following are listed the significant differences found
between the percents of any group of women among the wives
of any group of men eXpected in random marriage, and the

percents actually observed:

54.

Group 1 women among the wives of group 1 men:

Observed 70.67% 1 1.48%
Expected 57.48% g .81%

 

Difference 13.19% 1.68%

I4

The difference is 8 times its probable error, and
is significant.

Group 2 women among the wives of group 1 men:

 

Expected 36.16% 1 .74%
Observed 22.44% i 1.35%
Difference 15.72% t 1.54%

The difference is 9 times its probable error, and
is significant.

Group 2 women among the wives of group 2 men;

Observed 40.82% i 1.15%
Expected 36.16% ; .7g%

 

Difference 4.66% f 1.36%
The difference is 5 times its probable error, and
is significant.
Group 1 women among the wives of group 5 men:

.81%
1.62

Expected 57.48%
Observed 49.66

I
I
Difference 7.82% 1 1.81%

The difference is 4 times its probable error, and is
significant.

The other differences were small and less than three
times their probable error. The data indicate that mech-
anical ability may have a slight influence in mate selec—

tion; that persons of similar ability marry more often

than would be expected in random marriage. But the un-

55.
satisfactory classification of individuals prevents any
conclusions being drawn when the selection is no more de-

cisive than it is in this case.

Sight and Hearing

The condition of sight was classified in three groups:
1. (blind), 2. (imperfect;wears glasses), 5. (strong).

The condition of hearing was classified in three groups:
1. (deaf), 2. (defective), 5. (strong).

Only those cases were used where the age at Wiich
the defect, if any, was acquired was given; and if the
defect occured after the time of marriage, the case was
not used unless the condition before that time was given.
This necessary selection of cases has ruled alt many of
group 1 in both sight and hearing; for in many cases a
person was classified as blind or deaf after the age of
marriage, but no statement was made as to whether or not
there was a defect in the sight or hearing before marri-
age.

The date here given may be relied upon as being
accurate, for this is not an estimate of slight degrees
of a cnndition, such as the estimate of hair color, for
instance. If a man's hearing is imperfect enough for him
to be classified in group 2, it is probably imperfect
enough to affect mate selection is imperfect hearing does

affect it.

56.

These persons who were blind in only one eye, or

deaf in only one ear were here placed in group 2.

 

 

 

 

 

TABLE x. Conditi n of s ght. MEN
1 2 5 Total
1 o 0 0 o
E 2 0 69 177 246
o r-
7 3 1 167 1403 1571
Total 1 236 1580 1817

 

 

 

 

 

 

 

Date for condition of sight:

Observed % group 2 women among wives 29.25% 1 2.03%

of group 2 men:

EXpected % group 2 women

of group 2 men:

Difference

The difference is 7

is significant.

Expected % group 5 women

of group 2.men:

Observed % group 5 women

of group 2 men:

Difference

The difference

is significant.

Expected % group 2 wousn

of group 5 men:

Observed % group 2 women

of group 5 men:

Difference

is 7

among wives 13.53% i .53%

 

15.70% 1 2.1 %

times its probable error, and

among wives 86.46% t .55%

among wives 70.76% 1 2.02%

 

15.70% i 2.1 %

times its probable error, and

among wives 15.55% t .55%

among wives 11.2 % 1 .55%

 

2.33% t .75%

37.
The difference is 3 times its probable error, and

is significant.

l+

Observed % group 3 women anong wives 88.79% .53%

of group 3 men:

EXPGOtOG % group 3 nomen among wives 86.46% 1 .53%
of group 3 men:

 

Difference I 2.33% t .75%

The difference is 3 times its probable error, and
is significant.

It is indicated by the data that the condition of
sight does affect mate selection: a person having defect-
ive sight is more apt to marry someone whose sight is '

also defective than he is to marry someone whose sight is

 

 

 

 

 

normal.
Table II. Condition of Hearing: AMEN
WF““‘——

1 2 3 Total
1 2 1 O 3
E 2 0 0 50 50

§
“" 3 3 68 1872 1943
Total 6 69 1922 1996

 

 

 

 

 

 

 

a comparison of the observed percents with those
expected in random marriage for the condition of hearing
gives no significant diffmrences. There are too few de-
fectives among data collected in the manner in which
this was collected - from among the parents of college
students, for the most part - to give a true ssnple of

the papulation for such characters.

D

38.
Conclusions.

Since these Records of Family Traits were made out
largely by college students, or by other persons suffici-
ently weel educated to be interested in the field of
eugenics, and concern, in the majority of cases, the
family of the recorder, any conclusions reached or any
“condition indicated can be said to be true only of what
is often called the "upper strata" of our civilization.
Whether or not these same indications apply to persons of
a group inferior mentally and physically to this om we
are not in a position to say.

The data have given definite proof that mate selec-
tion is affected by at least three characters, age,
height, and education, the first and third having, perhaps,
a greater effect than the second.

Persons of about the same age much more frequently
marry than persons of widely different ages.

Persons of similar heights tend to marry each other
somewhat more frequently than persons of widely different
heights.

A person of extensive education usually selects for
a mate someone possessing an education nearly equal to
his own, rather than a person of limited education.

Some other characters have been indicated as affect-
ing or not affecting mate selection, but the accuracy of
the data is not sufficiently satisfactory to permit any

positive conclusions.

39.

Hair color does not appear to have 31y influence in
mate selection. It appears that persons of similar eye
color somewhat more frequently marry than persons of
dissimilar eye color. Similarly energetic and active
persons tend to marry each other. Persons of’about the
same degrees of mental ability are apt to marry each
other. It is indicated that musical ability and interest
is effective in selection. Drawing and coloring apparent-
ly have some influence. Mechanical skill seems to have
nothing to do with mate selection. The condition of
sight is apparently effective. But as pointed out above,
the data for these characters are not entirely satisfactory,
and the evidence of selection cannot be considered decisive.

It should be noted that in no case is there an in-
dication that types dissimilar for any character tend to
marry. In many cases there is an indication that similar
types of individuals tend to marry each other, and in the
study of three characters it has been shown with reason-
able certainty that there is assortative mating.

The ultimate effect of such selection cannot be
pr0phesied. Are we heading toward a better civilization,
or is there danger of more widely different groups of
people than we already have - groups that will constantly
struggle for supremacy?.And would such a struggle help

to make a better civilisation?

ACKNOWLEDGEMENTS

Appreciation is eXpressed to Dr.
C.B. Davenport and Dr. H.H. Laughlin
of the Eugenics Record Office, Cold
Spring Harbor, Long Island, New Y0rk,
for the use of the Archives and for
their cooperation in this work.

Appreciation is also eXpressed to
Dr. H. R. Hunt, Head of Department of
Zoology and Geology, Michigan State
College, for his many helpful sugges-
tions and criticisms throughout the
organization of this paper.

“T II

_-

I--

The Genetic Character of
Flexed Tail in Mice History
The first flexed tailed mice appeared in the
winter of 1927 in the rodenicolony of Michigan State
College. Two albino male mice having this peculiar-
ity were discovered by Dr. H. 3. Hunt among animals
descended from stock supplied by Dr. W. E. Castle.
The first such mouse had an S shaped tail which
was directed first anteriorly
over the back of the mouse, and

then posteriorly. Figure l was "_£;;“‘..‘-E=

not drawn directly from the mouse,

 

but illustrates the approximate shape
of the tail. There is no record of Fig. 1
the type of flexure found in the second mouse.

These two mice were bred with females from a1-
bino stocks, and the Fl's were inbred; the result was
the production of more flexed tailed mice. The problem
of determining the genetic character of this type of
tail was undertaken in the fall of 1927.

Description of Flexed Tail

Figure l and plates 1, 2, 3, etc. illustrate the
appearance of the flexed tail in the adult. The tail
is found to be very stiff at the point of bending, and
also, in some cases, in places where the tail is not

visibly bent. There seems to be no constant form or

forms of flexure; the tail may bend in any direction

any number of times; the flexures may be very angular,

or they may be curvilinear; or the tail may take the

form of a Spiral. In some animals the tail appears to

be of normal length; in others it is conSpicuously short.

A number of adult animals were found in which there was

no visible flexure in the tail, but upon passing the tail

between the fore-finger and thunb, and bending it slight-

ly at the same time, there could be detected an unquestion-

able stiffness not feund in normal mice. In some cases

this stiffness was well marked; in others it was uncertain.
Abnormalities of other sorts were frequently notic-

ed. A defective eye was found among some of the general

stock of the colony; and frequently appeared in the flex—

ed animals. Several flexed tailed animals were observed

to have dorsal enlargements of the head in front of the

ears, suggesting a hydrocephalic condition. These ani-

mals never lived to maturity. ‘In newborn litters of

Fé's nearly all of the flexed tailed mice, and occasion-

ally a mouse that appeared to have a normal tail, ex-

hibited a distinct paleness of color, suggesting anemia.
In very young flexed tailed mice, the tail usually

does not have sharp bends, but rather several waves. The

tails of adult mice frequently are bent very sharply.

It would be an interesting problem to study the chaige

of form of the tail from birth to maturity. The flexed

tail apparently does not change shape appreciably after

the mouse is fully grown.

3.

Dr. Ernst Blank, in his paper of "Die Khickschwgnze
der nguee~,° describes a tail abnormality which he calls
"broken tail". He found tails of different degrees of
bending, some bent to the extreme of a right angle, others
visibly straight but more or less stiff when run through
the fingers. Investigation showed this to be due not to
a true break, but to a fusion of the vertebrae. The bone
regions of two adjacent vertebrae approached each other
on one side of the tail, pushing the growth zone and
entire intervertebral apparatus in the Opposite direction.
This condition produced a bend in the tail.

MicroscOpic centions of different ages embryos of
our own flexed tailed mice have, so far, revealed noth-
ing of these processes, but it seems probable that the
cause of the flexure is the same as that described by
Dr. Blank. Out own microscOpic investigations are insuff-

icient to confirm or to contradict Blank's conclusions.

Procedure

The flexed tailed mice used in these experiments
'were mostly Fz's and Fs's from the original flexed mice.
They were crossed with straight tailed mice from the
regular stock cages. The straight tailed mice used did
not, so far as we knew, carry flexed tail - that is, no
flexed tailed animals had been observed in the stock.

The Fl's obtained were mated together. In some
cases two or three females were placed in the same cage

°Archiv fur Entwicklungsmechanik der Organismen. V01. 42,
1916“. 17 , ppe 555-406e

4.

with one male, and when this was done the females were
isolated before the birth of a litter. In most cases
only one female and one male were placed in a cage, and
in these crosses in the earlier part of the work the
female was not always isolated when pregnant, but the
litter was born in the cage with the male. In all of
the later experiments all pregnant females were isolat-
ed.

In the majority of the crosses of F1 x F1, the
young were counted at the time of weaning - between
three and four weeks after birth. The animals should,
of course, be counted promptly at the time of birth.

It is usually, but not always, possible to distinguish
straight and flexed tails when the animals are very
young. 1'he policy of counting the animals within 48
hours after birth has been followed in all of the crosses
of flexed x flexed, the condition of the tail being
checked at the time of weaning. -

Flexed mice were mated with flexed mice to deter—
mine whether or not the character bred true. Diagrams
of the tail of the parents were made at the time of
mating, and later compared with diagrams of the tails
of their progeny to determine if the type of flexure
of the parents appeared to be transmitted to the off-

spring, or if one type seemed to dominate another.

Results

Crosses of flexed x straight to obtain Fl's:

From crosses of flexed x straight tailed animals

5.

142 Fl's were obtained. All of these Fl's were straight
tailed. (Table 1). The flexed tailed here behaved as
though it might be a Hendelian recessive.

Crosses of F1 x F1 to obtain F2's:

In most cases the Fl's crosses were from the same
parents, and frequently from the same litter. F2 young
were obtained from 30 crosses of F1 x F1. (Table 2). A
total of 1065 Fg's was obtained, 936 of which were
straight tailed and 129 flexed, a ratio of 7.26 : l.00.°
Crosses of Flexed x Flexed:

From 16 crosses of flexed x flexed 137 young were
obtained, all of which had a flexed tail. It should be
noticed that in crosses 101 and 102 (Table 3) one parent
was classified as having a gtiff tail rather than a
flexed tail, but that all of the young were flexed tail-
ed, indicating that this stiffneSs is genetically simi-
lar to the flexed condition.

The following plates give diagrams drawn from some
of the parents and their progeny. Plates 1, 2, 3, 4
and 5.

° Dr. H. d. Hunt, working with mice from the same stock,
obtained a ratio of 6.8 : 1.0 from 1141 Fz's counted.
From a back-cross of flexed tailed parent x F1 generation,
he obtained a ratio of 1.4 straight to 1.0 flexed in 266
young. The back-cross shows the same deficiency of flex-

tailed animals as the crosses of F1 x F1. This is in
accordance with the findings given above.

 

Table 1. Flexed x Straight to Obtain Fl's

 

 

 

 

 

 

 

 

 

I
‘ I
1 Cross 1 Number of animals ~
g , )4
2 3 3 6
z 6 I 36
' 7 I 2
12 ' 7
. 13 3 60 I 18
I 13 2 61 I 21
‘ 18 . 10
19 i 14
56 8
58 6 .
64 3 ‘
72 4
I 67 7 ,
i I
1
Total ‘ 142 all Straight Tailed
Table 2. x F to Obtain F '8:

Cross no. Stra exe a
7 '8 F '8

20 , 31 5 34

 

- TOtRl Gad , "an finne- I

7.

 

 

 

Table 3. Flexed x Flexed
Cross Number of Young
10 5
11 8
14 7
74 12
80 10
81 19
86 8
87 3
93 10
94 3
95 16
96 3
100 13
101 5 (stiff x flexed pareits)
102 5 (stiff x flexed parents)
103 10
Total 137 (All young flexed tailed)

 

Interpretation of Data

In the F1 generation. obtained from crossing a
flexed tailed animal with a straight tailed animal, the
flexed character behaves as a Hendelian recessive - it
does not appear in F1.

When flexed is crossed with flexed the character
again acts as a Mendelian recessive - it breeds true.
The fact that a flexed tailed animal, when crossed with
a stiff tailed animal, gives all flexed tailed progeny
(crosses 101 and 102, table 3) supports the assumption
that a stiff tailed animal is genetically similar to a
flexed animal, and such animals are counted as flexed
in counting the Fz's.

The ratio in F2 is yet to be eXplained. It is
neither 3 : 1 nor 15 : 1, but is between the two, 7.26

8.
straight : 1.00 flexed.

If this character were a simple Mendelian recessive
caused by a single factor we would expect to find a ratio
of 3 : 1, or 798.7 straight to 266.3 flexed. Instead, we
get 936 straight to 129 flexed, a ratio of 7.26 : 1.00.
The deviation is 137.3 1 9.5. The numbers obtained are
not in a simple 3 : 1 ratio.

If this character were caused by two recessive fact-
ors acting together the eXpected ratio would be 15 : 1 or
998.4 : 66.6. The deviation is 62.4 i 5.3. We are not
dealing with a 15 : 1 ratio.

The ratio obviously does not approximate 9 : 7.

If we consider that there are three pairs of factors
concerned, and assume a modified 27:9:9:9:3:3:3:1 ratio,
the theory still does not fit; for by grouping 27:9:9;9
and 3:3:3:1, the ratio is 54 : 10, or 5.4 : 1.0, the ex-
pected numbers being 898.6 straight to 166.4 flexed. The
deviation is 37.4 i 7.9.

Let us assume the presence of two factors, one of
which acts as an inhibitor to the appearance of the flex-
ed condition.

A - straight (dominant) B - inhibitor
a - flexed (recessive) b - non-inhibitor

The zygotes formed in F1 are:

1AABB + 2AABb + ZAaBB + 4AaBa + laAbb + 2Aabb + laaBB +
BaaBb + leabb. Now if B inhibits the flexure entirely,
the ratio is 15:1. We do not have a 15:1 ratio. If

BB inhibits the appearance of the flexure entirely and

Bb permits it to appear only as a slight flexure, and

bb permits sharp flexure in the presence of as, the
ratio would be 13 : 3, or 4.3 : 1.0. The deviation from
this in our count is 72 I 8.6. The deviation is more
than three times its probable error indicating that this
is not a ratio of 4.3 : 1.

No satisfactory explanation for our ratio has been
found. We may have a character influenced by more modify-
ing factors. There is also another possibility of diff-
erential mortality, either pro-natal or post-natal. As
observed above, there are numerous abnormalities in the
stock, and the flexed young are frequently of anemh:
appearance at birth. Before any definite conclusions as
to the cause of the 7.26 : 1.00 ratio may be reached, a
study must be made of both pro-natal and post-natal mortal-
ity. The young must be counted at birth, and at weaning,
and the pro-natal mortality must be determined, perhaps
by counting corpora lutea.

To turn now to the study of the inheritance of flex-
ure type, it will be helpful to consult the diagrams
made of the tails of two flexed parents and their off-
spring. This is not the ideal way to study inheritance
of type, since two types are introduced into the cross,
one from eigher parent; but it was undertaken along with
the endeavor to determine if the character bred true.

In cross 10 (Plate 1) it will be noticed that the
flexure in the parents was not extreme. Four out of
the five offspring have a very extreme bend at the base

of the tail in about the same direction; the fifth

resembles the parents.

10.
Cross 11 diaplays very little similarity in the

offspring, although the parent types are not widely
different.

In cross 14 young 1, 2, 3, 4, 6 and 7, seem to re-
semble the male parent, while number 5 is more like the
female parent.

Very little resemblance is noticed among the young,
or between the young and the parents in cross 81. The
same is true of the other crosses here shown. Whihe
some resemblances are present, they are neither numerous,
nor striking. The similarity observed in some of the
crosses, however, suggests that flexure type may be in-

herited.

Conclusions

The conclusion to be drawn from the data collected
are:

A cross of flexed x straight gives, in F1, all
straight tailed animals, which is what would be expect-
ed if flexed tail is a mendelian recessive.

a cross of flexed x flexed gives all flexed animals,
which is also in accordance with the idea that flexed
tail is a Mendelian recessive.

Mice having stiff tails which do not show bending
bred as flexed tailed animals, giving all flexed progeny
in a cross of stiff x flexed. Stiff tail is genetically

similar to flexed tail.

ll.

Crosses of flexed x straight give in F2 a ratio of
7.26 : 1.03. This cannot be eXplained by a single factor,
the flexed condition being recessive to normal; nor by
modified ratios produced by the Operation of two factors.
The explanation for the 7.26 : 1.00 ratio may have numer-
ous modifying factors, or a differential pre-natal or
post-natal mortality.

Types of flexure may possibly be inherited, but no

such conclusion can be drawn from the data at hand.

flute I

01

Cross 11

Trrcnts

 

‘i'omlg

d:
is?

 

 

Cl]

CO

t—«l
H

Yelnlg

 

§

 

1r-.. _
Lbiii’lr“.
V

III

32.3

C]

(:1

8
MO 1‘33]-

3011331

 
 
 

W

W
14 W

he

’JOTS

[‘0

Ci

()1

CD

CO

(0

flats IV

1r 9 n p
.i. 0 ill. ’.'
SJ

% Dorsal

Jot bent, but
verv stiff entire length

 
    
 
  

Tot bent, but
fl: st proxim 1 end

J- -',
C L ..'- .L ..

*1"! _‘ _D_ 7'
-. Jan 06’ Ill

Cross 103
Jarents

 

01

CO

Cross 103 ,
i'arents 1cull-T

‘—
V

1‘0

(:1

7 % Dorsal

r r‘ T‘T‘f“:“"*'rm
‘LC;-‘.U n 4414.)») \-T.L—J4---'JJ.9A

3"”.‘riT/t‘."“1’\'t T? 'Z.“'_"_'.“'-‘\"'_“« run
£13-..) -\».1_4V.L.—1.. .Lk. «Ll Jud -4 . ~..~-5.L_o’— Tdhfilu *{J

""1: T '1) Tjrm ";:"“*‘('r;_"“; '7“, PF‘.‘
1111. do .LL. 1;. 1.1, -LU.-‘-1..~:;'\/';L C- uUULC'G I All?)

'“'“"~T "Tn" Iran cw'wrtl“ HGT‘TV‘T‘ 7‘. . .-,- Y
GUI, 4 \J' 11' ' ' n " i

‘J LJ'V'LL‘ a--.L\Jii J“*I baae.-‘a.a v ALLJJJ, .L. b- a..a..ee.

777‘” "1'7 ‘ o."/v"‘,."1—/‘,“r‘i '-,-r r *‘T ‘ riw‘:““",’“f-'f'/\."
"“"‘LL (W ‘ W * L D 01 CT W~W*v“eim
1---..La-- .u , LJL. ~ :J... Leuka ---- .. .i.' 1.1. L1-.-)- 1 -._.sv u a. -1. U .1.

raw-ii "r,"-:*' 0“ ”WI“ iron-‘1‘”
1...; .4...1...-. ‘-.‘ 1.-.. u ;_i{UaJlx.fiu-‘.ie

ROOM ust mm

 

7"

mt . . .
n F
K .

. \

.31“. e.
"5"“).

h: ET“? ‘ .,

5):. I'J{89_&;\:“Wrxn&yez.ghxfi
"s- .. cit
‘ 5', t\_&'}jsa - ,
tk' wenfi
:3. ' 3’ ~

IF

.isfifuxfi
‘pn’i’. 1;?

hv‘ ‘SI.
-.l
’ *N 30"
v!!?“‘ e
R
.".r,{‘ .
.' .0
.

‘ (

é,
. -. 2;; l‘