Rit (Vísindafélag Íslendinga) - 01.06.1984, Blaðsíða 280

276
PÁLLIMSLAND
neglected in the calculation procedure but a plagioclase of An61 composi-
tion, accompanied by the olivine, is able to account for the compositional
diíi'erence. The total fit is poor, being greater than the standard deviation of
the whole rock analysis in case of nearly all elements. Jan 62, a slightly less
magnesian basalt and considerably richer in alkalies, is more likely pro-
duced by this primitive crystal fractionation. Small amounts of the plagio-
clase do, however, greatly help an otherwise poor total fit.
In Table 44B the calculations are continued towards the more Mg poor
basalts of this category but from the same “mother liquid”. The best fit is
now obtained using the same olivine but a more evolved titan-salite. The
compositional differences of the “mother and derived liquids” are consider-
ably increased from the previous case and one would thus expect both
minerals to change composition during this interval, but by doing so in the
calculations a poorer result is obtained. The equilibrium approximation of
the minerals to the “liquids” is thus clearly not good in the case of these
relatively low-K basalts. In agreement with this the fit obtained is generally
poor. In two cases out offour it is tolerable. Here the misfit is caused by Na,
Mg, Fe and A1 in one case and by Ca, Fe and Ti in the other. In the two
other cases, where the misfit is much higher, it is primarily caused by the
same components, i.e. Ca, Fe and Ti in one case and Na, Mg, Fe and A1 in
the other. In the case of the Na, Mg, Fe and A1 misfit, the calculated
“daughter liquid” contains too much Na, Mg and Al, but too little Fe when
compared to the “mother”. In the case of the Ca, Fe, Ti misfit the “daughter
liquid” is too poor in Ti and Ca, but one “daughter” is too poor and the
other too rich in Fe. The misfit in these cases is thus quite systematic. It may
thus be expected to be caused by some single and recognizable geochemical
process, but it cannot be “corrected” by changing the composition of one or
two subtracted minerals.
Jan 43 has here been selected as the “mother liquid” to the relatively K
rich group of the more magnesian basalts. Its derivation from the ankara-
mites is shown in Table 42. Table 45 shows the results of these calculations,
which can be continued towards a quite low Mg content. A titan-salite and
an olivine of Fo80 composition are the minerals fractionated in greatest
amount but these are accompanied by a titanomagnetite and a plagioclase
of An85 composition in smaller amounts. As the calculation is carried on
over a broad compositional spectrum the approximation to equilibrium
conditions is necessarily poor over a great compositional interval in the
calculations. The total fit obtained is generally good and the fractionate is
dominated by the clinopyroxene except in the case ofjan 43 towards Jan 9,
which is a step of small compositional difference. Here the clinopyroxene is
neglected and the olivine dominates the fractionate, which is quite small, or
5 per cent (giving 95 per cent liquid). The misfit in this case is primarily
caused by Ca being too low in the calculated “daughter”. In the rest the
clinopyroxene amounts to 54 to 77 per cent of the fractionate. This fractiona-