Rit (Vísindafélag Íslendinga) - 01.06.1984, Page 129

mineralchemistryandrelationships
125
some authors (Kuno, 1960; Le Bas, op.cit.) classify the Skaergaard rocks as
being high-alumina in nature. In the case of Hawaiian clinopyroxenes
Fodor et al. (1975) show that Na increases with differentiation in the alkalic
and nephelinic rock suites, while in the tholeiitic rock suite, Na is fairly
constant in the Ca rich clinopyroxenes, which on the other hand are richer
in Na than the Ca poorer clinopyroxenes. This is the case in the tholeiitic
rock suite of Thingmuli as well, as shown by Carmichael (1967). The
behaviour of Na in clinopyroxenes during differentiation af rock suites
therefore seems to be more complicated than that of Mn. It is probably
highly dependent on the crystallization of the clinopyroxene relative to
feldspars. The decrease of Na in the ferroaugites (occurring as micropheno-
crysts in the most salic Jan Mayen trachytes) relative to the salites richest in
Na (occurring in less salic trachytes and tristanites) is probably caused by
early alkali feldspar crystallization (occurring as much larger phenocrysts).
Fe203■ The microprobe does not distinguish between valency state of
multivalent elements. There is thus no other way than to determine these
elements as the total of all or both valency states. In microprobe analyses
iron is thus usually reported as FeO total. The actual amounts of both Fe203
and FeO can thus only be found by indirect methods such as by calculation.
Usually the calculations are based on theoretical assumptions as for exam-
ple the most popular one, which assumes that the stoichiometry of the
crystal is perfect and analytical deviations from stoichiometry result from
undetermined Fe3+. This means that all the small errors in the analysis,
such as undetermined minor and trace elements, deviation in accuracy and
so on will seriously affect the results. These errors may in cases accumulate
in the calculations to produce very poor values. This leads to generally
much less reliable values for Fe3+ and Fe2+ than for the other components.
This can though in cases be qualitatively tested by plotting the calculated
Fe203 values against some parameter which gives defmite trends when
plotted against other elements. The calculated Fe203 values do not usually
form as clearly a dcfmed trend, which means that the values are more erratic
and deviations from the actual Fe3+ contents are much greater than for the
other determined elements. Conclusions based on such Fe203 calculated
values are thus not reliable.
The calculated Fe203 concentrations of the clinopyroxenes of the Jan
Mayen rocks plotted against Mg/Fe ratio (Fetot) of the clinopyroxenes (Fig.
80) show considerably more scatter than the oxides shown in Figs. 78 & 79.
A pattern is however clear. At all Mg/Fe ratio values it shows a more or less
even scatter from <0.5 wt. per cent up to an upper level, which varies with
Mg/Fe ratio. This upper level increases from ~1.5 wt. per cent in the most
magnesian chromian diopside towards a maximum of ~8 wt. per cent in the
titan-salites and then decreases again towards ~3 wt. per cent for the titan
poor salites and ferroaugites. The plutonic xenoliths, with the exception of
the wehrlite, have lower Fe203 contents than the clinopyroxenes ofcompar-