Vörös A. szerk.: Fragmenta Mineralogica Et Palaentologica 11. 1983. (Budapest, 1983)

Fig. 6 Typical chemical variations of different garnet crystals expressed in terms of pyropealmandine-grossular end-molecules. Arrows indicate changes in composition proceeding from the center to the margin even more strikingly (Fitton 1972). These facts were considered, of course, as one of the best evidences in favour of the cognate origin by the same authors. This interpretation is farther strengthened by the results of Irving & Frey (1 978), according to which the wide variation of the garnet/liquid distribution coefficients for minor elements is presumably related to bulk compositional variations. As garnet is an alumina-rich phase, it could be expected, assuming a cognate origin for them, that the volcanic host rocks are peraluminious and often high in potassium. This is frequently the case indeed, e. g. in Japan, Spain, England, Lipari and Crimea (GILL 1981). However, garnetiferous andésites in Slovakia and Hungary are dy-normative and medium-K (with the exception of Nagybörzsöny which has a negligible quantity of normative corundum; Table 3). Hungarian garnetiferous andésites carry abundantly H^O-bearing silicate minerals such as hornblende and biotite just like those of Japan, New Zealand, Crimea, Kamchatka, Spain an Clear Lake and unlike those of England, Slovakia and Lipari (GILL 1981). Nevertheless, the observation that garnet does not coexist with clinopyroxene (i. e. in lavas with < 58 % Si0

2

) could be strengthened and this may also be used as an argument against the accidental origin. Indeed, as garnet megacrysts are clearly in disequilibrium with their host (acid or basic), there is no reason why they should be invariably absent from more basic andésites if they come from metamorphic basement or country rocks.