Greinar (Vísindafélag Íslendinga) - 01.01.1977, Síða 125

123
Secondly, still more strange are the deep foci, down to 700
km, confined to volcanic island arcs with a deep trench in
front.
The earthquakes at these depths demand, together with some
movement, also strength. But this may be very low, 10—100
kg/cm2, for the stress or stress drop indicated by the release
of strain energy in earthquakes, is in this range.
Further, strength exists regionally at depths of 300—400
km, as shown by the fact that shield areas retain individuality
of seismic character to such depths (88); without strength,
there would be reason to expect that these shield roots would
have been spread. They indicate that the strength has per-
sisted for at least the last 1000 My.
These cases of regional strength at great depths in areas
of particular surface characters, demonstrate that there can
have been no, and are not at present any such general flows
which could explain the undulations of the geoid, as have been
proposed by Runcorn (89); see also further below.
We now return to the question about the intermediate and
deep earthquakes, and propose to enlighten it by way of the
undulations of the geoid.
These can be explained principally in three ways: 1) 1%
horizontal deviations from a constant density in individual
earth shells or, wrhat may mean the same, 100°C temperature
deviations (93, 94). 2) Flow in the mantle (89). 3) Mantle
strength of about 100 kg/cm2 (84).
Case 2) we seem to have been able to exclude above. As to
case 1), one must ask, whether it is reasonable to expect a
greater horizontal homogeneity in density than nfc 1%. At most,
a cosmogonical background for ideas about the earlier state
of the earth could justify such an expectation. For this reason
the strength in case 3) cannot at the present stage be calcu-
lated with any certainty. The value of 100 kg/cm2 is calcu-
lated by exclusion of case 1), and thus provides a maximum
value for lower mantle strength. But if there is strength at
all, then density inhomogeneity can exist, and case 1) might
play a role. Hence, with reference to our above findings about