The layered ultrabasic rocks of Rhum, Inner Hebrides

Abstract The Tertiary volcanic centre of Rhum is best known for the ultrabasic rocks which outcrop over 9 square miles of mountain country. A. Harker (1908) and subsequent investigators attributed their formation to a series of sill-like injections of magma already partly or completely differentiated into allivalitic and peridotitic material. The object of the present investigation was to describe this interesting assemblage of rocks in more detail and to reconsider its manner of origin in the light of modern petrological research, especially recent work on layered intrusions. The ultrabasic and associated basic rocks were re-mapped over an area of 3 square miles, including Hallival and Askival, the highest mountains of the island. Within a ring-shaped fracture which separates the igneous centre from the Torridonian sediments, there is visible a thickness of 2600 ft. of rock consisting of magnesian olivine, bytownite-felspar, diopsidic pyroxene and chromespinel, the minerals being arranged in layers which differ considerably in thickness and in the proportion of the constituent minerals, but yet dip regularly at about 15° towards a centre lying to the west of the area mapped. The variation in the layers is rhythmic rather than cyclic, and in this respect is comparable with the Bushveld, Skaergaard and Stillwater basic layered intrusions. The broad rhythmic pattern consists of an olivine-rich rock passing gradually upward into a plagioclase-rich rock, termed peridotite and allivalite respectively by Harker. These names are retained but their meaning is more precisely defined. A new rhythm begins abruptly with a fresh olivine-rich rock, the spinels being concentrated at the beginning and the clinopyroxenes near the end of a rhythm. Fifteen major rhythmic units have been recognized and traced over great lateral distances; each has a distinct character yet possesses a majority of characters common to the series. The best developed rhythm , unit 10, has been studied in detail as a type, the others being considered only in so far as they diverge from the standard pattern. The range in mineralogical composition is from a rock containing 80 % of olivine to a pure felspar rock, the clinopyroxene and spinel usually constituting 5 to 15 %. Though rare concentrations of spinel occur there are no monomineralic olivine, pyroxene or spinel rocks. The olivine, plagioclase felspar and clinopyroxene from an allivalite of the type unit were chemically analyzed. The olivine is Fa14, the plagioclase An84 and the pyroxene Ca44Mg49Fe7. There is no appreciable variation in the composition of the unzoned cores of minerals within the layered series, apart from an oscillation in plagioclase composition between An83 and An88. Zoning is practically absent from minerals of the upper seven units, but normal zoning is developed lower in the series, the range being An84 to An44 in the lowest unit. In view of the felspar composition it is suggested that Harker’s term allivalite be extended to include rocks in which the felspar, present in excess of olivine, has a composition An80 to An100. Most of the area originally mapped by Harker as eucrite sheets has been shown to form the three lower units of the layered series. Two broad sheets mapped earlier as allivalite are now recognized as later intrusions of a fine-grained olivine-gabbro. The lower of these sheets contains numerous inclusions, many ultrabasic, which led Bailey (1945) to support the separate injection hypothesis of Harker for the whole series, as he believed the sheet to be an allivalite of the ultrabasic series. No inclusions are found in the layered series of this area, such streaks and lenticles as are found being identified as the result of auto-brecciation, frequently associated with slumping. The textures indicate that the layers formed from a body of magma through the bottom accumulation by sinking of crystals whose ultimate size, shape, orientation and distribution were governed by the order of crystallization and relative densities of the minerals, and perhaps also by the winnowing action of currents within the liquid and the occasional release of volatiles and influx of fresh magma associated with surface vulcanicity. The interprecipitate liquid has played an important role, adding material to the margins of the primary precipitate crystals and sometimes forming poikilitic growths where no primary crystals of the particular mineral existed. The ultimate size of minerals with a poikilitic habit is a function of the number of centres of growth developed, which is related to the ease with which diffusion could operate in the interstitial liquid. Some peridotites contain poikilitic plates of felspar and pyroxene 3 cm in diameter each enclosing about 10000 olivine crystals. Fluctuations in the rate of deposition of layers would affect the extent to which diffusion could occur between interstitial and overlying liquid, and they are believed to be responsible for variety in proportions of minerals, including the development of monomineralic felspar layers, and also for the non-systematic variation in the composition of the minerals, usually evidenced by normal zoning. The lower units are considered to have been deposited faster than the upper units, while the rate of deposition within each unit apparently decreased from bottom to top. Periods of non-deposition resulted in the upward growth from the top surface of the accumulating crystals of long, branching olivine crystals forming the ‘harrisite structure’. The composition of the magma has been estimated from the composition of interstitial liquid trapped at certain horizons, and from comparison of this with the associated undifferentiated gabbros. The composition thus obtained is that of a tholeiitic basalt, and the minerals of the layered series are such as would form during the early stages of the crystallization of many known basalt magmas. It is suggested that the sub-crustal chamber in which this magma slowly cooled was connected with the main volcanic conduit, and that intermittent surface extrusion occasionally removed liquid from the chamber, while fresh material replaced it from below. Such replenishment would account for the great thickness of ultrabasic material crystallized within a narrow temperature range and for the absence of overlying, lower-temperature layered rocks. Subsequent to consolidation of the layered series in depth a ring-shaped fracture developed and the layered rocks, probably representing only a small portion of an extensive complex, were driven up to their present high level with the help of the lubricating action of what is now the structureless marginal gabbro.