Rit (Vísindafélag Íslendinga) - 01.06.1984, Blaðsíða 299

PETROGENETIC REEATIONSHIPS
295
obtained for Zr. The distribution coefficients for Rb between magma and
both alkali feldspar and biotite are higher than zero. These minerals occur
in the more evolved rocks and are used in the fractionation calculations.
Taking notice of this in calculating the Rb enrichment and using the
distribudon coefficients for Rb between both alkali feldspar and biotite and
magma, collected from the literature by Arth (op.cit.), Rb gives the same
results as both Y and Zr. The incompatible trace element content resulting
from the model fractionation is thus in reasonable agreement with the actual
content of the ankaramites and the less evolved basalts. In the case of the
more evolved basalts and the intermediate rocks, on the other hand, the
model fracdonation gives far too high incompatible element values.
Calculating for trace elements pardcipating in the crystallization (as in
the case of Rb in the most evolved rocks mendoned above) makes use of the
equation (Arth, op.cit.) Cl/Cí=1/F’+Ds(1— F’). This is calculated here for
Sr, which primarily enters the feldspars. The distribution coeílicients used
are those collected by Arth (op.cit.). In cases of minerals fractionated in the
model calculations, but not given distribution coefficients in the compilation
of Arth (op.cit.), the calculadons are carried out as if the distribution
coefficients were zero, as for instance for Sr in titanomagnetite and apatite.
The result of these calculations is shown in Fig. 130. As in the case of the
incompatible trace elements, the fit between observed and calculated values
is relatively good for the ankaramites and more primitive basalts (1 to 6).
The deviadon from the 1 to 1 line is in most cases small. In the case of the
more evolved basalts and the intermediate rocks (7 to 11) the deviation is
increased and the slope is quite different. In the ankaramites and the basalts
the minerals of greatest influence are the mafic ones. As a test on these, some
calculadons concerning the Ni behaviour have been undertaken. The
distribution coefficients of Ni between crystals and liquid are far from
constant. In the compiladon of trace element partitioning data of Irving
(1978), Ni in olivine and clinopyroxene is shown to be dependent on
crystallization temperature and even magma composition. Selecting from
this data Ni distribution coefficients that fit reasonably well to the Jan
Mayen condidons and calculadng the Ni content resulting from the model
fractionation for the ankaramites and basalts gives a reasonable fit to the
analysed Ni values of these rocks, as Fig. 131 shows. Here group 1 (the most
primidve one in composition) is calculated using a 1300°C distribution
coefficient and for the remaining groups a 1200°C coefficient is used. The
distribution coefficients for Ni between magma and plagioclase and
titanomagnetite are here taken as zero.
As the diagrams show, there is a fairly good agreement between the
results of the Ni calculations, which emphasize the mafic mineral fractiona-
tion, and the Sr calculations, which emphasize the plagioclase fractionation.
There is thus an agreement between the results obtained for incompatible
trace elements, on the one hand, and the “compatible” ones, on the other.