Our knowledge of the rare rock type eulysite (A. J. Erdmann, 1846) comes almost wholly from Swedish occurrences amt hitherto no examples have been recognized from Britain. It is therefore of interest to place on record a new Occurrence of eulysites and related rocks from among Lewisian rocks of the Glenelg area, Scotland.

The rocks shortly to be described come from the vicinity of Totaig on Loch Duich between Totaig and Letterfearn, little more than half a mile from the first described occurrence of eclogite in Britain. These eulysitic rocks crop out on the hillside immediately south-west of Druideig Lodge. About 200 yards south-west of the lodge coarse garnet-paragneisses form outcrops associated with grunerite- and cummingtonite-garnet-gneisses, and a similar interbanded association is found at higher levels.

Aconspicuous feature of all published diagrams, showing the results of the determination of a number of plagioclase felspars by the Fedorov method, is the pronounced scattering of the poles. They lie not only on the curves of the determinative stereogram, but on either side of them, and, in fact, occupy a belt, varying in different instances, and in different positions along the length of the same curve, from 3 to 8 mm. in width. These belts usually contain, but are not symmetrically disposed to, the curves.

The recovery of these two meteoric stones is due to the energy and resource of Mr. R. Bedford of the Kyancutta Museum, South Australia. They have been sent by him for description and placing on record, and he has very kindly allowed portions to be retained for the British Museum collection of meteorites. I also owe my thanks to him for the detailed information he has supplied respecting the finding of the specimens.

Inquiry into this feature was stimulated by a reasonable doubt concerning the reality of the immiscibility gap in the clinoenstatite-diopside series. The existence of this gap was first suspected by Sosman and Asklund, and the identity of the gap with the composition field of the hornblendes has recently been suggested by Kennedy, whose comparative study of hornblende and pyroxene composition engaged statistical methods applied in the first instance by Asklund.