Rit (Vísindafélag Íslendinga) - 01.06.1984, Síða 269

PETROGENETIC REEATIONSHIPS
265
practice, but theoretically it is, on the other hand, more easily done. As the
computer program used does not exclude negative subtraction, as is bound
to happen frequently in this case, the chrome-spinels have been omitted
from these calculations. As the chromian diopside alone is not able to
account for the Cr changes observed, this chrome-spinel omission is the
primary cause for the poor Cr fit of the calculations. The calculated
derivative is in most cases too rich in Fe and too poor in Al. The subtracted
minerals are thus too poor in Fe and too rich in Al. In spite ofa rather small
misfit in this case, this poses a problem. Around 75 per cent of the
fractionate is a chromian diopside and in the pyroxenes of the Jan Mayen
rocks, Fe and A1 increase and decrease together. The problem can therefore
not be eliminated by selection of another slightly different pyroxene. The
olivine used is rich in iron relative to the pyroxene and contains no A1 so that
increasing its share in the fractionate results in a correction of the Al- and Fe
misfit, but simultaneously increases the misfit of other elements according to
the rules of the calculations. The Fe problem may be at least partly ascribed
to oxidation, as the liquids and minerals are treated in the calculations as if
all iron was FeO, which is a simplification made in opposition to known
facts. A very minor misfit appears in the Na- and especially K values.
Mostly, both elements are slightly too low in the calculated derivative.
Apparently the fractionate is therefore too rich in both elements. This is a
paradox as both the subtracted minerals are without K, the olivine is
without Na and the pyroxene is poor in Na.
If all primitive basalt magma is to be produced in this manner, by the
fractionation process, then fractionation within the ankaramites is a major
operation in terms of mass. Fractionating from a Mg rich ankaramitic
liquid, as for instance Jan 166 (in one step) to the ankaramitic basalt Jan 26,
leaves just over 40 wt. per cent of the original mass as liquid of Jan 26
composition.
As the crystal fractionation is calculated further towards the more basaltic
compositions the fit becomes poorer (see 166 to 26 in Table 40). Other lines
of evidence have already shown that at this stage the clinopyroxene changes
composition towards the titan-salites and the olivine becomes more fayalitic
(ankaramitic crystallization in Fig. 126). Table 42 shows the result of this
crystallization, in terms of fractionation from a Mg poor ankaramite (Jan
10) to the most Mg rich basalts, i.a. Jan 26, the most primitive liquid to
reach the surface in a totally liquid state. This is carried out by subtracting a
pyroxene intermediate between the chromian diopsides and the typical
titan-salites and an olivine of Fo87 composition. Adding a trace amount of
titanomagnetite to the subtracted minerals gives this fractionation almost a
perfect fit. Such an early titanomagnetite is probably not unrealistic. In the
case of Jan 26 the misfit is within the analytical error of the whole rock
analysis. In the case ofjan 43 and 73, K and A1 show misfit slightly in excess
of the analytical error. The sum of the squares of the differences between