The three major rock types present in the ZSF, namely serpentinites, metagabbros and metabasalts,are studied from a geochemical point of view.
The main focus is on the magmatic genesis of the ZSF rocks and on their relationship. The peculiar difficulties of doing this kind of work in a polymetamorphic terrain are dealt with. In particular it is tried to shed some light on the geochemical effects of the metamorphic overprint. It is considered whether statistical or other methods can help to evaluate the relative mobility and immobility of diagnostic trace elements. In particular, in view of the wellknown problems of doint statistics with compositional data (closed data sets, where the components sum up to 100 %), the logratio method of J. Aitchison is advocated and used extensively.
The serpentinites can be explained as residual mantle after extraction of ca. 5 to 22 % melt in the spinel lherzolite field. The mode of melting seems to have been closer to batch melting than to fractional melting. A peculiarity are positive Ti and Nb anomalies and a negative Zr anomaly in the mantlenormalized spidergram. The Ti and Nb anomalies are explained by assuming that this mantle contains some exotic phase, perhaps Ticlinohumite. As a potential explanation it may be assumed that the serpentinites represent subcontinental mantle.
The metagabbros have been olivine + plagioclase cumulates, sometimes also clinopyroxene + plagioclase cumulates. Some highly differentiated members of this group are containing also FeTi minerals as cumulate phases. The chemical data are compatible with the assumption that these rocks were derived from those melts which belong to the restites now represented by the serpentinites. In particular the gabbros are displaying Ti and Nb anomalies complementary to the serpentinites. That they are also having a negative Zr anomaly then would speak in favor of a primary lack of Zr in the mantle source.
For the metabasalts, on the other hand, a magmatic relationship to either the metagabbros or to the serpentinites can be excluded, arguing from the lack of the aforementioned characteristic trace element anomalies. Furthermore, the lack of a negative Eu anomaly precludes a relationship to the gabbros as fractionated melt to cumulate.
The mantle source of the basalts must have been rather fertile, even more fertile than a TMORB source, as the REE and HFSE are rather enriched. On the other hand, the LILE are strongly depleted. This peculiar combination of enrichment and depletion might again point to subcontinental mantle, but this time lacking the exotic phase which has been present when the gabbroic melts were extracted.
These findings are not compatible with a MOR genetic model. Instead they are tentatively related to the framework of a simple shear rifting model as proposed by Wernicke (1981, 1985), where a lower plate is unroofed through extension along a detachment fault penetrating into the lithosphere. The ZermattSaas Fee ophiolite zone then is an example of a lithospheric ocean, as first proposed by Lemoine et al. (1987) for the Alpine ophiolites.