AFGHANISTAN

69

the Aksu, and part to the west down the Ab-i-Panja. Thus
the glacier, or the great bed of glaciers and snowfields
which feeds the Aksu, feed also this northern affluent of
the Wakhan ; and the same gigantic but indefinite snowcapped
 watershed of the Nicolas range which forms the
glacial sources of the Aksu and Wakhan on the one side,
forms those of Lake Victoria on the other. So that the
glaciers of the Nicolas range in about East longitude
74, become the sources of the main affluents of the
Oxus, excepting the southern head of the Wakhan.
Whether the glaciers of Nicolas or those at the head
of the southern affluent of the Wakhan are the mightiest,
is another question. It is one which, at any rate, cannot
be decided by a comparative estimate of the capacity of
the various channels through which the glacial streams
work their way to the great river.
From about the longitude of Kala Panja, where Afghan
territory narrows to about ten miles of width, the boundary
bends south-west, and still following the axis of the Hindu
Kush, runs for nearly another hundred miles to the Dorah
Pass, leading from Ishkashim in Badakshan to Chitral. It
is in this section of its length that the highest peaks are
found, and even beyond this section, southwards from the
Dorah, where a long spur defines the boundary for another
fifty or sixty miles. Parting the Chitral River from an
important affluent called the Arnawai or Bashgol, in Eastern
Afghanistan, we have as magnificent a natural barrier on
the north-west frontier of Kashmir as can be found in
the width of the Himalaya. It is true that between the
trijunction of the three boundaries of Russia, China, and
India, to the junction of the Bashgol River of Kafiristan with
the river of Chitral, two principal passes cross the Hindu
Kush, besides several of minor importance. These two are
the Baroghél (12,500 feet) and the Dorah (14,800 feet), and
one of them (the Baroghél) presents no great obstacle for
some six months of the year when not under deep snow ; but
the difficulties to passage presented by the routes in which
these passes are incident is not to be measured by the
altitude of the passes themselves, or by their own inherent