A few miles to the east some remarkable felsitic rocks take the place of these vesicular lavas immediately below the pisolitic iron ore. I have not determined satisfactorily their relations to the surrounding rocks, and in particular am uncertain whether they are interbedded lavas or intrusive sheets. Dr. F. H. Hatch found that their microscopic characters show a close resemblance to the soda-felsites described by him from the Bala series of the south-east of Ireland. The slopes of Cader Idris are partly obscured with debris, from above which rises the great precipitous face formed by the escarpment of "porphyry," here intrusively interposed among the Arenig volcanic rocks. This enormous sill will be referred to a little further on in connection with the other intrusive sheets of the region. The remarkably cellular rock which forms the peak of Cader Idris is coloured on the Survey map as an intrusive sill of "greenstone," which in the Memoir is said to alter the contiguous slates and to appear to cut across them diagonally. I am disposed, however, to think that these appearances of intrusion are deceptive. On the southern declivity of the mountain this rock presents one of the most curious structures to be seen in the whole district. Its surface displays a mass of spheroidal or pillow-shaped blocks aggregated together, each having a tendency to divide internally into prisms which diverge from the outside towards the centre.[150] Some portions are extremely slaggy, and round these more solid portions finely crystalline parts are drawn, suggestive rather of free motion at the surface than of the conditions under which a subterranean sill must be formed. The idea occurred to me on the ground that while the band of rock marked as "greenstone" on the map is probably, in the main, an interstratified lava, there may nevertheless be basic intrusions along its course, as in the lower part of the mountain. The minute structure of this amygdaloid, as revealed by the microscope, shows it to be an epidiorite wherein the hornblende, paramorphic after augite, has been again partially altered along the margins into chlorite. [Footnote 150: This peculiar structure of the more basic Arenig lavas, where the rock looks as if built up of irregularly-spheroidal, sack-like or pillow-shaped blocks, will be again referred to in connection with the Arenig (and Llandeilo) lavas of Scotland and Ireland. It appears to be widely distributed, and especially in connection with the occurrence of radiolarian cherts. The black slate above the Cader Idris amygdaloid would, in a similar position in Scotland, be associated with such cherts, but these have not yet been noticed at this locality. With the spheroidal internally-radiating prismatic structure of the Cader Idris rock, compare that of the lava at Acicastello already noticed on p. 26.]