Danforth: Is Twinning Hereditary ? 
so is suggested by the following rea- 
soning. 
It is found,® for example, that of 
37,621 pairs of twins born in Germany 
and France, there were 13,315 cases in 
which one twin was a boy, the other a 
girl. Now an interpretation that may 
be put on such data is this: These 
13,315 cases represent half of the 
biovular twins—since such twins have 
equal chances of being of the same or of 
opposite sex. This means that of the 
whole number roughly 26,630 cases 
represent biovular twins and the re- 
maining 10,991 cases represent uniovular 
twins, from which it appears that 29+ 
% of all twin cases are uniovular.’ 
The textbooks of obstetrics published 
in Germany and France, basing their 
statement on the relations of foetal 
membranes, generally give the number 
of uniovular twins as about 15%. 
Here is a discrepancy that seems to 
have been generally overlooked, but 
it is very probable that the difference 
between 15% and 29% represents the 
number of cases in which uniovular 
twins develop in separate sets of foetal 
membranes. 
POSSIBILITY OF OTHER KINDS 
There is another point that cannot 
well be overlooked in this connection. 
Prof. Thorndike*® found that when the 
degree of similarity between the two 
members of different pairs of twins is 
measured and plotted for a large number 
of cases the resulting curve is smooth 
and not two-humped as might have been 
expected. The mode falls at a point 
higher than that for comparisons be- 
tween ordinary brothers and sisters, but 
considerably below the point represent- 
ing identity. If twins fall only into 
the two classes usually postulated it is 
difficult to see how such a result could 
be obtained. Similar study of other, 
and if possible, more extensive material 
is greatly needed. 
It may ultimately be possible to show 
that Thorndike’s smooth curve repre- 
6 Simon Newcomb, op. cit. 
199 
sents the leveling effect of like environ- 
ment reacting on biovular twins and of 
somatic variation affecting uniovular 
twins, but it may also be that the two 
types mentioned do not represent all the 
classes of twins, for it must be admitted 
that theoretically there are other possi- 
bilities. One such possibility is sug- 
gested by the work of Boveri’ and others 
on the eggs of bees and sea-urchins. 
It was found by these investigators 
that the entrance of the sperm to the 
egg occasionally stimulates a precocious 
division of the latter so that the sperm 
nucleus is able to unite with only one 
half of the original egg nucleus, leaving 
the other half to develop (in these lower 
forms) parthenogenetically. If such a 
condition were to arise in man, the 
second half of the egg nucleus might, so 
far as is known, be fertilized by one of 
the innumerable superfluous sperm cells, 
in which case we would perhaps get a 
pair of twins derived from one egg 
and two sperms. Such three-germ twins 
might even be of opposite sex, yet they 
should be more similar than ordinary 
brothers and sisters. The at present 
puzzling distribution of twins in certain 
families could be explained very well on 
this assumption, but such a postulate 
lacks proof, and one must proceed cau- 
tiously in introducing new hypotheses. 
FREQUENCY OF TWINS 
For the student of heredity it is 
always desirable to know the “normal 
incidence”’ of the character under inves- 
tigation. Knowing this it is possible to 
calculate, on the assumption that the 
characteristic appears fortuitously, the 
probability that it will be found once, 
twice or oftener in groups of a given size. 
With this information at hand one may 
then determine whether the character- 
istic regularly appears in certain families 
enough oftener than the laws of chance 
would explain to warrant regarding it as 
hereditary. 
The approximate incidence of twin 
births as a whole is easily obtained. 
7 Miss Margaret V. Cobb has recently applied the same reasoning to American data with 
similar results. 
Science, N.S., Vol. XLI, No. 1057, pp. 501, 502. 
April 2, 1915. 
8 Archives of Philosophy, Psychology and Scientific Methods, No. 1, 1905. 
9 This work is scattered through numerous journals dealing with experimental embryology, etc. 
