国立情報学研究所 - ディジタル・シルクロード・プロジェクト
『東洋文庫所蔵』貴重書デジタルアーカイブ

(since the ΣΔ, in spite of the fact that one of its fifteen factors is a whole number, is only 3·64); consequently
their ΣΔ has been printed in ordinary type, and it may be allowed that a close relationship exists between
them.

Let us first regard the table from the point of view of similarity. The lowest ΣΔ is that for Turfan and Korla,
which, though it contains fifteen factors, amounts only to 2·03. This remarkably low figure shows clearly that
the populations of the two localities are not merely closely related but practically identical. Two other ΣΔ fall
below 3, viz. Turfan and Khotan (2·98) and Kirghiz and Dolan (2·84), and indicate nearly as close a relationship
between these pairs of tribes relatively. (See Table 10, which shows more clearly the inter-relation of the various
tribes; names in italics indicate that the ΣΔ contains a Δ as factor which amounts to 1·00 or over with the
exceptions noted above.) If the ΣΔ is raised to 4, we find that the Dolan, Kelpin, and Kirghiz form a closely-
related group of three, that the Wakhi are related to Turfan, that Keriya pairs with Niya, and Bagh-jigda with
the Sarikoli. To anticipate, therefore, we have the nuclei of three groups, the Kirghiz group, the Turfan group,
to which one mountain people, the Wakhi, are closely related, and the Pamir group, with an outlying branch
at Bagh-jigda. If the ΣΔ be raised to 5, these groups materialize more distinctly. It is now seen that Turfan,
Korla, Khotan, and Wakhi are all inter-related by ΣΔ under 5; and that Turfan, Korla, and Hami stand in similar
relationship. Further that Sarikoli, Bagh-jigda, and Pakhpu constitute a similar group; while the Kirghiz-
Kelpin-Dolan group remains unchanged. For convenience the first group will be called the Desert group
(although it includes the Wakhi); the group containing the Sarikoli, the Pamir group; and the last, the Kirghiz
group. Now if we look at the Mastuji we see that they are related to the Pamir group through the Sarikoli, and
have, besides, affinities with the Kafir and Chitrali; again, it is apparent that the Pamir group is brought into
contact with the Desert group owing to a relationship existing between the Pakhpu and Hami, while Kok-yar
and Polur show a connexion with the Desert group through Khotan, and Karanghu-tagh through Hami. Other
close relationships are those between Karanghu-tagh and Nissa, and between Loplik and Charkhlik.

If the limit of the ΣΔ be raised to 6, we find that the Mastuji can be related with the Bagh-jigda as
well as the Sarikoli, and though their ΣΔ as regards the Pakhpu is over 6, being in fact 6·27, yet the excess
is so small that we may fairly regard them as belonging to the Pamir group. The ΣΔ for the Kafir and
Chitrali falls below 6, being 5·92, but the Δ for the nasal length is 1, and the relationship cannot, therefore,
be regarded as so close as that existing between each and the Mastuji. As regards the Desert group, we find
this now constituted by Turfan, Korla, Wakhi, Charkhlik, and Khotan, or by the first four and Hami. The ΣΔ
for Hami and Khotan is 6·62, and the factor which brings it above 6 is that for the facial index, 1·33, due to the
fact that the people of Khotan are considerably more euryprosopic. Further connexion between the Pamir and
Desert groups is seen in the relation of the Mastuji with the Wakhi and of the Sarikoli with Turfan. Kok-yar,
which has already been shown to possess affinities with Khotan, now becomes related to the Wakhi, and also
with the Pamir group through the Sarikoli; Keriya becomes attached to the Desert group through Khotan
and Turfan, while similar relationships are seen to exist between Nissa and Hami, between Karanghu-tagh and
the Loplik, and between the latter and the Chinese. The Kirghiz group also receive an addition in the Ak-su,
who become related to the Kelpin and Dolan, their ΣΔ as regards the Kirghiz themselves being only just over 6,
viz. 6·20. This excess is due solely to head-circumference, which, as has been shown, does not afford trustworthy
evidence. As yet this group has shown no relationship to either of the others, but, as a matter of fact, it lies far
nearer to the Desert group than to the Pamir group. The ΣΔ of Ak-su and Kelpin only just misses inclusion in
the present limit, being 6·20, and containing no factor over 0·90; ΣΔ for Keriya and Kelpin, too, is only 6·82,
though in this case Δ for the stature-span index amounts to 1·67. Other affinities, besides those already mentioned,
appear between the Pamir and Desert groups, though in each case the ΣΔ contains a Δ of 1 or over; they are the
following, the words in brackets indicating the particular in which Δ attains or exceeds unity: Pakhpu and
Karanghu-tagh (facial breadth), Pakhpu and Polur (facial breadth), Pakhpu and Nissa (facial breadth and stature-
span index), Sarikoli and Hami (nasal breadth), Sarikoli and Khotan (facial breadth). It is obvious from this that
the main particular in which the two groups differ is facial breadth, but that in the case of the Wakhi this
difference does not occur. It is interesting to note that the average of the ΣΔ of the Wakhi is the lowest, and
never attains 10.

One people, that of Faizabad, has not been related at present to any other, and it is a fact that in no case
has it a ΣΔ which does not contain a factor under 1. Its lowest ΣΔ are with Khotan (6·59, Δ for span and
stature-span index being over 1) and with Ak-su (6·73, Δ for facial breadth and head-circumference being over 1).