Jökull - 01.12.1985, Qupperneq 41

Fig. 8. Deposit of layered sands and
gravels, N end of Norðurdalur,
much contorted and showing signs
of frost wedging. See 80/9 in Fig. 6
for lower material and 80/10 for the
thinner covering layer of less well
sorted deposit. — 8. mynd. Setlög í
Norðurdal.
diverted away from the immediate area, to return to the
ice margin once the pressure was released. In the same
way, the high ground of Hafrafell and Staffell to the E
would have funnelled the ice stream, as can be seen
from the broad, swampy col between Hafrafell and
Rangárhnjúkur, but among all the signs of ice erosion in
this col, the glaciated surface on which they occur is cut
by fairly wide and steep-sided channels, (Figs. 2 and 7),
which contain peat deposits which have been drained
for cultivation. These could be channels carrying water
diverted away from the rock masses both to E and W,
cutting into the line of the dykes in the country rock,
(Fig. 3).
The northerly trend of channels along the edge of the
high ground of Fljótsdalsheiði and Fellaheiði is con-
tinued into the lowland to the N, characterised by lines
of rock ridges separated by long lakes and swampy
hollows. Lagarfljót flows through the E side of the
lowland in a probably adventitious course through the
ridges as far as Lagarfoss, and Jökulsá á Brú runs
parallel to the W in a similar situation.
Even on the top of the Fellaheiði plateau, on the E
side of which it appears that some of the channels end,
as mentioned above, there is evidence of water flow to
the NW of the lake Sandvatn stóra. This lake is at
present drained by the Rangá to the N, in a very
irregular course, but to the NW of the lake there is a
channel system cut in the central divide of the plateau.
The most noteworthy feature of this is the angular
notch, (Tröllagjót on the 1:100 000 map), with lowest
point about 590 m, from which the ground slopes fairly
gently W to Sandvatn at 569 m, but very steeply down to
the E, (Fig. 10). This notch cannot be associated with
any recent drainage from Sandvatn stóra because the
ridge falls below 590 m just to the N, and it must be
assumed that ice covered the ridge and the area which is
now the N part of Sandvatn when the notch was formed.
A col at 660 m on the ridge leads to Sandvatn Iitla lying
in one of the channels, which then leads into a wide,
deep trench, which eventually debouches into Jökul-
dalur, (Fig. 1). The Sandá stream, a tributary of which
could soon behead Rangá just N of Sandvatn stóra, is
cutting deeply into the older trench, the end of which
lies above the Giljahólar feature in Jökuldalur. From
the evidence of air photographs this trench could have
held a glacier tongue although this is not apparent on
the ground.
(c) Deposition. It is often difficult to interpret
deposits either by appearance or by their location in
occasional exposures, so as many as possible of the
samples collected have been subjected to mechanical
analysis. Even quite large samples, however, have such
a wide range of particle sizes that it is difficult to show
them adequately in some cases, while it is often very
difficult to separate out individual layers in some of the
stratifjed deposits of sands and gravels where layers are
thin. In the very poorly sorted deposits one large stone
can affect the analysis of quite a large sample.
Of the few large deposits of unsorted material, the
most important seems to be that found from the N end
JÖKULL 35. ÁR 39