It is obvious that in any effort to establish that a group of rocks is older than the very base of the sedimentary fossiliferous formations, we must somewhere find that group emerging from under the bottom of these formations. Until lithological characters are ascertained to be so distinctive and constant as to be comparable to fossil evidence for purposes of stratigraphical identification, we should not assume that detached areas of older rocks rising amid Palæozoic, Secondary or Tertiary formations are pre-Cambrian. We should, if possible, begin at the bottom of the Palæozoic systems and work backward, tracing each successive system or group as these rise from under each other, until we arrive at what appears to be the oldest traceable within the region of observation. It is clear that in the present state of knowledge we have no satisfactory means of identifying such successive systems in widely separated countries. All that can be attempted in the meantime is to ascertain the special types in each region, and to point out their general resemblances or contrasts to those of other regions. It is better to avoid confusion by refraining from applying the stratigraphical names adopted for the oldest rocks of one region to those of another geographically remote, though we may hope that eventually it may be possible to work out the equivalence of these local names.

In the British Isles, by much the most important region for the study of the oldest rocks is to be found in the north-west Highlands of Scotland. The very basement strata of the Cambrian system are there traceable for a distance of more than 100 miles, reposing with a strong unconformability upon all rocks of older date. They consist of dolomitic shales with Olenellus, resting upon a thick group of quartzites, full of annelid tubes. One of the most remarkable features of these ancient strata is the persistence of their component bands or zones which, though sometimes only a few feet thick, can be traced throughout the whole tract of country just referred to. For the study of the pre-Cambrian rocks this is an important point, for we can be quite certain that even where fossil evidence locally fails, the same basement members of the Cambrian system are persistent and lie directly upon the pre-Cambrian series.

Lewisian Gneiss. Ever since the researches of Murchison and Nicol in the north-west of Scotland, it has been known that two distinct systems of rock underlie the quartzites to which I have just alluded. Murchison regarded the upper of these as of Cambrian age, while he assigned the unconformable quartzites and limestones above it to the Lower Silurian period. But the recent discovery of the Olenellus zone intercalated conformably between the quartzites and the overlying limestones may be regarded as proving that all the rocks which underlie the quartzites and are separated from them by a strong unconformability must be pre-Cambrian. It is thus established beyond any reasonable doubt that two great pre-Cambrian systems of rock exist in the north-west of Scotland.

These two systems differ so entirely from each other that their respective areas can be defined with minute accuracy. The uppermost consists chiefly of dull reddish sandstones with conglomerates, and especially towards their base in Rosshire, some bands of dark grey shale, the whole having a thickness of at least 8,000 or 10,000 feet, though as both the base and the top of the series are marked by strong unconformabilities, the whole original thickness of deposits is nowhere seen. As these rocks are well developed around Loch Torridon, they were named by Nicol the Torridon Sandstone—a designation which has more recently been shortened into "Torridonian." The lower system is mainly composed of various foliated rocks which may be embraced under the general term "gneiss." These masses present the usual characters of the so-called "fundamental complex", "Urgebirge," or "Archæan Series" of other countries. The contrast between the thoroughly crystalline, gnarled, ancient-looking gneisses below, and the overlying, nearly horizontal Torridonian conglomerates, sandstones, and shales, which are largely made out of their debris, is so striking that every observer feels persuaded that in any logical system of classification they can not be both placed in the same division of the geological record. They are certainly both pre-Cambrian, but they must belong to widely separated eras, and must have been produced by entirely different processes. If it is proposed to regard the gneisses as "Archæan," we must refuse to include the Torridonian strata in the same section of pre-Cambrian time. But so much uncertainty exists as to the application of this term Archæan, examples are so multiplying wherein what was supposed to be the oldest and truly Archæan rock is found to be intrusive in rocks that were taken to be of much younger date, and there are such slender grounds for correlating the so-called Archæan rocks of one country with those of another, that I prefer for the present, at least, not to use the term at all. Let me very briefly state some of the main characteristics of the two sharply contrasted rock-systems of the north-west of Scotland.

The oldest gneiss of that region was originally called "Lewisian" by Murchison, from its large development in the Island of Lewis, and I think it would be, for the present at least, an advantage to retain this geographical appellation. At first this "fundamental gneiss" was thought to be a comparatively simple formation, and the general impression probably was that it should be regarded as a metamorphic mass, produced mainly from the alterations of very ancient stratified rocks. Its foliation-planes were believed to be those of original deposit which by terrestrial disturbance had been thrown into numerous plications and corrugated puckerings. But a detailed study of this primeval rock has revealed in it a far more complicated structure. The supposed bedding-planes have been ascertained to have nothing to do with sedimentary stratification, and the gneiss has been resolved into a complex series of eruptive rocks, varying from a highly basic to an acid type, and manifestly belonging to different times of extrusion. With the exception of one district, to which I shall immediately refer, no part of the whole region yet examined has revealed to the rigid scrutiny of my colleagues of the Geological Survey, any trace of rocks which can be regarded as probably of other than igneous origin. It is true that our researches have been hitherto confined to the mainland of Scotland, the large area of the Outer Hebrides, which consists of similar gneisses, remaining to be explored. It is therefore possible that indisputable evidence of an ancient sedimentary series through which the gneiss was originally protruded, may yet be discovered in the unexplored islands. But taking the gneiss as at present known in Sutherland and Rosshire, we find it to be generally coarse in texture, rudely foliated, and passing sometimes into massive types in which foliation is either faintly developed or entirely absent. Much of this gneiss is considerably more basic than the more typical rocks to which the term gneiss was formerly restricted. It consists of plagioclase felspar with pyroxene, hornblende, and magnetite, sometimes with blue opalescent quartz, and sometimes with black mica. These predominant minerals are segregated in different proportions in the different bands, some bands consisting mainly of pyroxene or hornblende, with little or no plagioclase, others chiefly of plagioclase, with small quantities of the ferro-magnesian minerals and quartz, others of plagioclase and quartz, others of magnetite. This separation of mineral constituents can hardly be attributed to mere mechanical deformation. It rather resembles the segregation layers which may be studied in intrusive sills and other deep-seated masses of eruptive material, and which are obviously due to a process of separation that went on while the igneous magma was still in a liquid or viscous condition. At the same time it is manifest that extensive dynamical changes have affected the rocks since the appearance of this original banded structure.

There is further evidence that beside the original eruptive masses, which for want of any means of discriminating their relative dates of protrusion must in the meantime be regarded as belonging to one eruptive period, other portions of igneous material have been subsequently and at successive epochs, after the first mechanical deformations, injected into the body of the original gneiss. These consist of dykes of basalt and dolerite, followed by still more basic peridotites and picrites, and lastly by emanations from a distinctly acid magma in the form of granites. The oldest or doleritic dykes form a wonderful feature in the gneiss, from their abundance, persistence and uniformity of trend in a west-northwest direction. They have no parallel in British Geology until we reach the crowded dykes of older Tertiary time.

Throughout this remarkable complex of eruptive material, though its different portions present many features that may be compared with those of intrusive bosses and sheets belonging to later geological periods, there is no trace of any superficial volcanic manifestation. No tuffs or agglomerates or slaggy lavas have been detected, such as might serve to indicate the ejection of volcanic materials to the surface. All the phenomena of the Lewisian gneiss point to the consolidation of successively protruded portions of eruptive material at some depth within the crust.

Nevertheless it may yet be possible to show that these deep seated masses have been injected into rocks of older date and of sedimentary origin, and that they have communicated with the surface in true volcanic eruptions. I have already alluded to one limited area where various rocks exist, distinctly different from the prevalent types in the Lewisian gneiss. In the area which is traversed by the long valley of Loch Maree in western Rosshire, there occur clay-slates, fine mica schists, graphitic schists, and saccharoid limestones. These rocks remind us of some of the prevalent members of a series of metamorphosed sediments. The minerals enclosed in the marbles are just such as might be expected in the metamorphic aureole of a granite boss, piercing limestone. But the relations of this group of rocks to the ordinary gneiss of the region are not quite so clear as could be desired, though they seem to point to these rocks being surrounded by and enclosed within the gneiss.

The detailed field-work of the officers of the Geological Survey has made known the remarkable amount of mechanical deformation which the various rock-masses composing the Lewisian gneiss have undergone. These rocks have been compressed, crushed, and drawn out, until what were originally massive crystalline protrusions have been converted into perfect schists. The dykes of dolerite have been transformed into hornblende-schists and the granitic pegmatites have been reduced to a kind of powder which has been rolled out so as to simulate the flow-structure of a lava. There is evidence that most, if not all, of this dynamical change was effected long before the deposition of the Torridonian series, for the latter rests in nearly horizontal sheets, with a strong unconformability upon the crushed and sheared gneiss.

Torridon Sandstone. This group of rocks covers only a limited area in the north-west of Scotland, but it must once have spread over a far more extensive region. It reaches a thickness, as I have said, of 8,000 or 10,000 feet, and consists almost wholly of dull, purplish-red sandstones, often pebbly, and bands of conglomerate. Dark grey shales, already alluded to as occurring towards the base of the series, are repeated also in the highest visible portion, and have yielded tracks of what seem to have been annelids and casts of nail-like bodies which may have been organic. I have said that the Torridonian deposits which were classed by Murchison as Cambrian, have been proved by the discovery of the Olenellus zone in an unconformable position above them, to be of pre-Cambrian age. Except along the line of disturbance to which I shall immediately refer, these strata are quite unaltered. Indeed, in general aspect they look as young as the old red sandstones with which Hugh Miller identified them. It is at first hard to believe that such flat undisturbed sandstones are of higher antiquity than the very oldest Palæozoic strata which are so generally plicated and cleaved.