the mass is remarkably disposed to separate into flat rhomboidal fragments, the surfaces of the laminæ of which are sometimes interspersed with a few small specks of white mica.
The rock constituting this summit, as may be satisfactorily ascertained by insensibly graduating specimens, is of the same nature with those already spoken of; though at first sight, and especially in particular parts, apparently very different. The hornblende gradually diminishing in its proportion, or being intimately blended with the substance of the felspar, often merely imparts an obscure shade of green to the whole mass; the presence of which colour principally assists the eye in recognizing the true nature of the rock; and but for which it might be confounded with a compact sandstone or felspar. The same observation holds, but still more strongly, with respect to the rock on which stands Roche Castle; a ruin situated to the north of the turnpike road, about half way between Haverfordwest and St. David's. This rock has, much more decidedly than Penberry, the character of a compact sandstone: but in its geographical relation to the surrounding country, it exactly corresponds with the preceding rocks, and with the numerous similar rocks of the neighbourhood. However this may be, the surface of the ground between Roche Castle and St. David's is scattered over with numerous large boulders, as they might be called, very closely resembling in their general character one or other of the three rocks already described; and all of them bearing strong marks of having been the result of chemical formation.
The rock of which Penberry is composed shews occasionally a slight tendency to concentric disintegration; and the external part