The agglomerate (b) is quite tumultuous, and here and there strikingly coarse. Some of its included blocks measure five feet in length. These fragments represent most of the varieties of the lavas of the district. Large slaggy masses are abundant among them, and sometimes exhibit the annelide-like elongation of the vesicles which I have referred to as occasionally displayed by the plateau-basalts. More than 60 feet of agglomerate are visible in vertical height from where its base is concealed by debris and vegetation to where its upper surface passes under a banded rock to be afterwards described. That this unstratified mass of volcanic detritus marks the site of a vent can hardly be doubted, although denudation has not revealed the actual walls of the chimney. The steep grassy slopes do not permit the relations of the rocks to be everywhere seen, but the agglomerate appears to pass laterally into finer, rudely-stratified material of a similar kind, which extends towards east and west as a thick deposit between the bedded basalts. Possibly denudation has only advanced far enough to lay bare the crater and its surrounding sheets of fragmentary material, while the chimney lies still buried underneath.

To the east or left of the agglomerate the detritus becomes less coarse, and shows increasing indications of a bedded arrangement. Close to the agglomerate the dip of the coarse tuff is towards that rock at about 10°. A few yards further east a sheet of very slaggy basalt is seen to lie against the tuff, which it does not pierce. The vesicles in this adhering cake of lava have been pulled out in the direction of the slope till they have become narrow tubes four or five inches long and parallel to each other. Some parts of this rock have a curved ropy surface, like that of well-known Vesuvian lavas, suggestive of the molten rock having flowed in successive thin viscous sheets down the slope, which has a declivity of about 30°. This part of the section may possibly preserve a fragment of the actual inner slope of the crater formed of rudely-bedded tuffs.

Continuing still eastward, we find the feebly stratified tuff (a) to be perhaps 200 feet thick. It forms a grassy declivity that descends from the basalt-escarpment above to the grass-covered platform which overlies a lower group of basalts. The visible portion of this tuff presents a thoroughly volcanic character, being made up of the usual dull dirty-green granular paste, through which are dispersed angular and rough lumps of slag and pieces of more solid basalt, varying up to a foot or two feet in length. These stones are generally disposed parallel to the indistinct bedding, but are sometimes placed on end, as if they had assumed that position on falling from an explosive shower. Among the smaller stones, pieces of a finely vesicular basic pumice are frequent and are among the most strikingly volcanic products of the deposit. From a characteristic sample of these stones, a thin slice was prepared and placed in Mr. Harker's hands. The following are his observations on it:—"A very compact dark grey rock, amygdaloidal on a minute scale. The lighter grey crust is probably due merely to weathering, and the specimen seems to be a distinct fragment, not a true bomb. The slice shows it to be essentially a brown glass with only occasional microscopic crystals of a basic plagioclase. It has been highly vesicular, and the vesicles are now filled by various secondary products, including a chloritic mineral, nearly colourless and singly refracting in thin section, and a zeolite."

Tracing now the tuff from the west or right side of the vent, we can follow it to a greater distance. No abrupt line can be detected here, any more than on the other side, between the agglomerate and the tuff. The latter rock extends under the overlying plateau of basalt, at least as far west as Portree Loch, a distance of fully a mile, but rapidly diminishes in thickness in that direction. Traces of what is probably the same tuff can be detected between the basalts at Ach na Hannait, more than three miles to the south (Fig. 305). It is thus probable that from the Portree vent fragmentary discharges took place over an area of several square miles.

Above the agglomerate of this vent two lavas may be seen to start towards opposite directions. One of these (c), already referred to, is a dull prismatic basalt with a slaggy bottom, its vesicles being pulled out in the direction of the general bedding of the section. It descends by a twist or step, and then lies on the inclined surface of the tuff which dips towards the agglomerate and seems to pass into that rock. Further east this basalt increases in thickness and forms the lowest of the basalt-sheets of the cliff. The lava that commences on the west side of the agglomerate (d) is a massive jointed basalt, which, though not seen at the vent, appears immediately to the west of it and rapidly swells out so as to become one of the thickest sheets of the locality. It lies upon the rudely-bedded tuff, and is covered by the other basalts of the cliff.

That these two basalts came out of this vent cannot be affirmed. If they did so at different times, their emission must have been followed by the explosion which cleared the funnel and left the central mass of agglomerate there. But that some kind of saucer-shaped depression was still left above the site of the vent is indicated by the curious elliptical mass of rock (e) that lies immediately above the agglomerate, from which it is sharply marked off. This is one of the most puzzling rocks in the district, probably in large measure owing to its advanced state of decay. It is dull-red in colour, and decomposes into roughly parallel layers, so that at a short distance it looks like a bedded tuff, or like some of the crumbling varieties of banded lavas. I could not obtain specimens fresh enough to put its nature and origin beyond dispute. Whatever may have been its history, this ferruginous rock rests in a flat basin-shaped hollow directly above the agglomerate of the vent. The form of this depression corresponds fairly well with what we may suppose to have been the final position and shape of the crater of the little volcano. The rock that occupies the bowl dies out towards the east on the face of the cliff, and the prismatic basalt (c) is then immediately covered by the rest of the basalt-sheets of the plateau (f). On the west side its precise termination is concealed by grass. But it must rapidly dwindle in that direction also, for not many yards away it is found to have disappeared, and the basalts (d and f) come together.

Though the decayed state of this rock does not warrant any very confident opinion regarding its history, I am inclined to look upon it as a deposit of much disintegrated volcanic detritus washed into the hollow of the old crater when it had become filled with water, and had passed into the condition of a maar. The peculiarly oxidized condition of its materials points probably to long atmospheric exposure, and an examination of the surrounding parts of the district furnishes more or less distinct evidence that a considerable lapse of time did actually intervene between the cessation of the eruptions of the Portree volcano and the next great basalt-floods of this part of Skye.

That volcanic eruptions from other vents continued after the Portree vent had become extinct is proved by the great sheets of basalt (f) that overspread it, and still bury a large tract of the fragmentary material which it discharged. At a later time a fissure that was opened across the vent, allowed the uprise of a basalt dyke (g), and subsequently another injection of similar material took place along the same line of weakness (h).

Before leaving this interesting locality we may briefly take note of the distribution of the ashes and stones ejected by the volcano, and the evidence for the relative length of the interval between the outflow of the lavas below and that of those above the tuff and volcanic conglomerate. These deposits may be traced in clear sections along the base of the cliffs for a mile to the west of the vent. They thin away so rapidly in that direction that at a distance of three-quarters of a mile they do not much exceed fifty feet in thickness. At Camas Bàn they consist mainly of a fine, dull-green, granular, rudely-stratified basalt-tuff, through which occasional angular pieces of different lavas and rough slags are irregularly dispersed. These stones occur here and there in rows, suggestive of more vigorous discharges, the layers between the platforms of coarser detritus being occupied by fine tuff. Some of the ejected blocks are imbedded on end—an indication of the force with which they were projected so as to fall nearly a mile from the crater.

The upper parts of the tuff pass upward into fine yellow, brown, and black clays a few feet in thickness, the darker layers being full of carbonaceous streaks. On this horizon the coal of Portree was formerly mined. The workings, however, have long been abandoned, and, owing to the fall of large blocks from the basalt-cliff overhead, the entrance to the mine is almost completely blocked up. One wooden prop may still be seen keeping up the roof of the adit, which is here a slaggy basalt.

To the east and south-east of the Portree vent, extensive landslips of the volcanic series and of the underlying Jurassic formations make it hardly possible to trace the continuation of the tuff-zone in that direction. To the south, however, at a distance of rather more than three miles, what is probably the same stratigraphical horizon may be conveniently examined from Ach na Hannait for some way to the north of Tianavaig Bay. At the former locality the calcareous sandstones of the Inferior Oolite are unconformably covered by the group of rocks represented in Fig. 305. At the bottom of the volcanic series lies a sheet of nodular dolerite with a slaggy upper surface (a). Wrapping round the projections and filling up the depressions of this lava comes a thin group of sedimentary strata from an inch or two to eighteen inches or more in thickness (b). These deposits consist of hardened shale charged with macerated fragments of linear leaves and other plant-remains, including and passing into streaks of coal, which may be looked upon as probably occupying the same horizon with the coal of Portree. But here, instead of reposing on a mass of stratified tuff, the carbonaceous layers lie on one of the bedded lavas. The tuff has died out in the intervening three miles, yet that some of the discharges of volcanic detritus reached even to this distance, and that they took place during the accumulation of these layers of mud and vegetation, is shown by the occurrence in the shales of pieces of finely amygdaloidal basalt, from less than an inch to six inches in length, likewise lapilli of a fine minutely cellular basic pumice, like some varieties of palagonite. The overlying dolerite (c) becomes finely prismatic at its junction with the sedimentary layers and has probably indurated them.

This intercalation of a shaly and coaly band among the lavas can be followed northward along the coast. In some places it has been invaded by dykes, sills, and threads of basalt on the most remarkably minute scale, of which I shall give some account in Chapter xlii. (see Fig. 321). North of Tianavaig Bay—that is, about three-quarters of a mile nearer to the Portree vent—a perceptible increase in the amount of volcanic material is observable among the shales and leaf-beds. Not only are lapilli of basic pumice abundant, but the volcanic detritus has accumulated here and there in sufficient amount to form a band of dull greenish-brown tuff.

These coast-sections in the neighbourhood of Portree afford additional illustrations of the characteristic fact, on which I have already insisted, that the interstratifications of sedimentary material in the basalt-plateaux frequently terminate upward in leaf-beds, thin coals, or layers of shale, full of indistinctly preserved remains of plants. As I have endeavoured to show, this vegetation, which was undoubtedly terrestrial, probably grew not far from the sites where its remains have been preserved. Leaves and seeds would naturally be blown or washed into pools on the lava-fields, and would gather there among the mud and sand carried by rain from the surrounding ground. Such a topography and such a sequence of events point to intervals of longer or shorter duration between the successive outpourings of basalt. It was probably during one of these intervals of quietude that the crater of the Portree volcano became a maar and was finally silted up.

Reference has already been made to a conspicuous mass of agglomerate which occurs at the east end of the island of Canna, and marks the site of an important volcanic vent belonging to the Small Isles plateau. A portion of it projects from the grassy slopes, and rises vertically above the beach as a picturesque crag, in front of the precipice of Compass Hill (Fig. 306). But the same rock may be traced southward to the Coroghon Mòr, and north-westward in the lower part of the cliffs to a little beyond the sea-stack of An Stòll. It has thus a diameter of at least 3000 feet. Westward it passes under the conglomerate described in Chapter xxxviii. Its eastern extension has been concealed by the sea.

The materials that fill this vent consist of a typical agglomerate composed entirely, or almost entirely, of volcanic detritus. The embedded blocks vary up to eight feet in diameter or even more. They are chiefly fragments of various basalts and andesites, generally vesicular or amygdaloidal. Some of these, which have evidently been broken off from already consolidated lavas, are angular or subangular in shape, and their steam-holes are cut across by the outer surfaces of the stones. Where they consist of calcite, zeolite, etc., the amygdales so exactly resemble those of the bedded basalts of the plateaux that, as already remarked, we must believe them to have been already filled by infiltration before the disruption of the rocks by volcanic explosions. Other blocks are true bombs, with a fine-grained crust outside and a more cellular texture inside, the vesicles of the outer crust being sometimes dragged round the surface of the stone. The variety of materials included among the ejected blocks and the abundance of pieces of the red bole which so generally separates the plateau-basalts indicate that a considerable thickness of bedded lavas has probably been broken through by the vent.

Beside the volcanic materials, occasional angular pieces of red (Torridon) sandstone may be observed in the agglomerate. The paste is a comminuted mass of the same material as the blocks, tolerably compact, and entirely without any trace of stratification.

The actual margin of this vent has nowhere been detected by me. We never reach here the base of the volcanic series, for it is sunk under the sea-level. On the other hand, the upper limits of the agglomerate have been partially effaced or obscured by the conglomerates which overlie it. From the breadth of ground across which the agglomerate can be followed along the shore, the vent might be regarded as having been perhaps not less than three-quarters of a mile in diameter. But there is the same difficulty here as at the Strath vent in Skye in determining the actual limits of the volcanic funnel. Possibly there may have been more than one vent in close proximity. Even if there was only one, the existing agglomerate may include not only what filled the chimney, but also a portion of what had accumulated round the orifice and formed the external cone. That the volcano continued for some time in vigorous eruption may be judged from the amount of material ejected from it, the large size of its blocks, and the distance to which they were sometimes thrown.

The pieces of Torridon Sandstone were no doubt derived from the extension of that formation underneath Canna. On the opposite island of Rum, where these pre-Cambrian red sandstones are copiously developed, they form the platform through which the Tertiary volcanic series has been erupted. The several remaining outliers of the bedded basalts, referred to in a previous chapter (p. 215 and Fig. 267) as visible on the west side of this island, show that the basalt-plateau of Small Isles, which once covered that area, rested immediately on the inclined edges of the Torridon Sandstones. Probably the same structure stretches westward under Canna and Sanday. No traces of any Jurassic strata have been detected beneath the volcanic rocks of Rum, though they are so well developed a few miles to the east in the island of Eigg. Either they were not deposited over the pre-Cambrian rocks of Rum, or they had been removed from that ancient ridge before the beginning of the Tertiary volcanic period. Certainly I have not detected a single recognizable fragment of any Jurassic sedimentary rock in the agglomerate of Canna.

This Canna vent exhibits, better than is usually shown, the occurrence of dykes and irregular injections of lava through the agglomerate. A large mass of a finely columnar basalt runs up from the beach at Garbh Asgarnish. A similar rock forms several detached crags a little further south, particularly in the headland of Coroghon Mòr and the island of Alman. Here the basalt is beautifully columnar, its slender prisms curving from a central line until their ends abut against the agglomerate. The truly intrusive character of this basalt is well shown on the southern front of Coroghon Mòr, and on the northern face of Alman, as represented in the accompanying diagrams (Figs. 307 and 308).

Although there is no conclusive evidence that these intrusions belong to the time of the activity of the vent, yet they differ so much from the ordinary dykes (one of which also cuts the agglomerate and ascends through the conglomerates and basalts above), are confined so markedly to the vent and its immediate proximity, and resemble so closely the basalt-injections of other vents, such as those of the Carboniferous and Permian necks of Scotland, that they may with probability be regarded as part of the mechanism of the Canna volcano.

Though the form and size of the vent of this volcano cannot be precisely defined, the upper part of its agglomerate, as we have seen (ante, p. 219), is dovetailed in the most interesting way with the series of coarse conglomerates, which indicate strong river-action in this part of the volcanic area during the time of the eruption of the plateau-basalts.

The agglomerate vents described in the foregoing pages as occurring in Antrim and among the Inner Hebrides all appear either in the midst of the plateau-basalts or in close proximity to them. Before quitting the Scottish examples, I may refer to some that rise through much more ancient formations at a distance from any portion of the volcanic plateaux, and yet may with probability be assigned to the Tertiary volcanic period.

During the progress of the Geological Survey through the district of Applecross, in the western part of the mainland of Ross-shire, and far away from the basalt-plateau of Skye, Mr. John Horne[306] has found two small necks rising on each side of a line of fracture, through gently inclined Torridon Sandstones. They are conspicuous from a distance by the verdure of their slopes, in contrast with the brown tints of the surrounding moorland. The larger of the two necks measures about 180 by 150 feet, and abruptly truncates the beds of Torridon Sandstone, which as they approach it assume a bleached aspect and become indurated. The material filling this vent is an agglomerate made up mainly of pieces of Torridon Sandstone and grit which, though generally small, occasionally measure a foot across, and in one case were found to reach a length of four feet. They are not as a rule markedly altered, but some of them have acquired a glazed or vitreous texture. Besides these fragments of the general rock of the district, there occur abundant lapilli of a basic volcanic rock, found by Mr. Teall to consist of porphyritic felspar, extremely minute acicular microlites of felspar, somewhat irregular transparent spaces now occupied by a yellowish-green substance, and interstitial matter. At the south end of the vent a small mass of decayed basalt appears to pierce the agglomerate.

Though there is no indication of the age of these necks, they agree so closely in general character with known vents of the Tertiary volcanic plateaux that there cannot be much hesitation in regarding them as dating from the same great period of basalt-eruption. But no relic now exists anywhere around of lavas or tuffs ejected from them. They rise on the bare Applecross hills, 1000 feet above sea-level, two miles from the shore, and about ten miles from the nearest outlier of the basalt-plateau in the Dùn Can of Raasay. If they once discharged streams of lava that united with the rest of the plateau, the total destruction of this lava affords another impressive picture of the waste which the volcanic rocks of the Inner Hebrides have undergone.

The large proportion of Torridon Sandstone blocks in these two Applecross necks suggests, however, that the orifices never became active volcanic vents. They may have been mere spiracles, or blow-holes, where the funnels drilled by explosive vapours were filled up with the debris of the rocks that were blown out. But that lava did rise within them is shown by the basic lapilli in the agglomerate, and by the basalt which in both vents has found its way up the chimney.

In the island of Raasay Mr. Teall, during the summer of 1894, observed a group of curious neck-like masses of breccia which pierce the Torridon Sandstone near Brochel (Fig. 309). The blocks in them are large angular unaltered pieces of the surrounding sandstones and shales, sometimes ten feet or more in length, and the matrix is sometimes pure crystalline calcite like Iceland spar. The breccia is generally coarsest towards the outer margin. But though the Lewisian gneiss exists immediately below the thin cake of Torridonian strata, not a fragment of it could either Mr. Teall or I, when I visited the locality with him, find among the components of the breccia. Nor did we detect any trace of volcanic material. The general ground-plan of these masses is elliptical, the most northerly measuring 30 yards in diameter. Where the junction of the breccia with the Torridon strata can be seen it is a nearly vertical one, the sandstones and shales being much jumbled and broken, but not sensibly indurated. This little cluster of patches of breccia can hardly be due to local crushing of the rocks. Their definite outlines and composition seem rather to indicate spiracles of Tertiary time, which never became vents erupting lava or ashes. The absence of fragments of the underlying gneiss may be accounted for if we suppose that the orifices were completely cleared out by the violence of the explosions and were afterwards filled up by the falling in of the walls of the higher parts now removed by denudation, which consisted of Torridon Sandstone and shale.[307]

Further research may detect at still greater distances from the basalt-plateaux ancient volcanic necks that might, with more or less probability, be referred to the Tertiary period. As an instance of this kind, I refer to the neck at Bunowen, County Galway, recently described by Mr. M'Henry and Professor Sollas. Though so remote from the Tertiary basalt-plateaux, the rock of this boss is an olivine-basalt presenting a close resemblance to some of the rocks of Antrim.[308]

As a final illustration of Tertiary volcanic vents I will now describe the Faroe group already alluded to (vol. i. p. 63, vol. ii. p. 256). It was almost by a kind of happy accident that these vents were discovered. Noticing at a distance of a mile or more from the deck of a steam-yacht that the base of the great basalt cliffs on the west side of Stromö were varied by what looked like agglomerate, I steamed inshore, and was delighted to find, as the vessel drew near to the cliff, that the agglomerate assumed definite boundaries and occurred in several distinct patches, until at last it presented the unmistakable outlines of a group of vents underlying and overspread by the bedded basalts of the plateau. Favoured by an unusually calm sea, I was enabled to boat into every nook and round every buttress and islet of this part of the coast-line.

The basalt-plateau here presents to the western ocean a nearly vertical escarpment which must reach a height of at least 1000 feet (see Fig. 328), and displays a magnificent section of the bedded lavas. The lower part of this section shows chiefly the banded structure already described, the layers of different consistency being etched out by the weather in such a way as to give them the look of stratified rocks. In the upper part of the precipice columnar and jointed or prismatic sheets are more common, but the most prominent band is the great sill, to which further reference will be made in the next Chapter.

In the course of the gradual retreat of the cliff, as the waves tunnel its base, and slice after slice is detached from its vertical front, a group of at least five small vents has been uncovered lying along a nearly north and south line. Of two of these a segment remains still on the cliff-wall and passes under the basalts; the others have been dissected and half cut away from the cliff, while groups of stacks and rocky islets of agglomerate may mark the position of others almost effaced. The horizontal distance within which the vents are crowded is probably less than half a mile, but the lofty proportions of the precipice tend to lead the eye to underestimate both heights and distances.

The agglomerate is a thoroughly volcanic rock, consisting of large and small blocks of various basalts, among which large slags are specially conspicuous, the whole being wrapped in a granular matrix of comminuted volcanic detritus. The arrangement of this material is best seen in the fourth vent (Figs. 310 and 311). In this characteristic volcanic neck (b in Fig. 311) the boundary walls, as laid bare on the face of the precipice, are vertical, and are formed of the truncated ends of the banded lavas (a a) which have been blown out at the time of the formation of the orifice. The visible diameter of the vent was roughly estimated by me to be about 100 yards. No appreciable alteration was observed in the ends of the lavas next the vent.

The agglomerate is coarsest in the centre, where huge blocks of slaggy lava lie imbedded in the amorphous mass of compacted debris. On either side of this structureless central portion the agglomerate is distinctly stratified from the walls towards the middle, at angles of 30° to 35°. Even from a distance it can be observed that the upper limit of the agglomerate is saucer-shaped, the sloping sides of the depression dipping towards the centre of the neck at about the same angle as the rudely-stratified agglomerate underneath. From the bottom of this basin to the sea-level may be a vertical distance of some 30 yards. The basin itself has been filled up by three successive flows of basalt, of which the first (c) has merely overflowed the bottom, the second (d), entering from the northern rim of the basin, extends across to the southern slope, while the third (e), also flowing from the north, has filled up the remainder of the hollow and extended completely across it. The next succeeding lava (f) stretched over the site in such a way as to bury it entirely, and to provide a level floor for the piling up of the succeeding sheets of basalt.

The second vent, which is represented in Fig 312, exhibits the same features, but with some additional points of interest. It measures roughly about 20 yards in diameter at the sea-level, rises through the same group of banded basalt (a a), and is filled with a similar agglomerate (b). Its more northerly wall is now coincident with a line of fault (h) which ascends the cliff, and probably marks some subsidence after the eruptions had ceased. The southern wall shows that a dyke of basalt (g) has risen between the agglomerate and the banded basalts, and that a second dyke (g′) traverses the latter at a distance of a few feet. In this instance, also, the upper surface of the agglomerate forms a cup-shaped depression which has been filled in by two successive streams of lava (c, d). Among the succeeding lavas (e) the prominent sill (f) has been intruded, to which further allusion is made on p. 323.

These necks are obviously volcanic vents belonging to the time of the basaltic eruptions. They have been drilled through the basalts of the lower part of the cliff, but have been buried under those of the central and higher parts. The arrangement of their component materials in rude beds dipping towards the middle of each vent shows that the ejected dust and stones must have fallen back into the orifice so as to be rudely stratified towards the centre of the chimney, which was finally closed by its own last discharges of coarse detritus. The saucer-shaped upper limit of the agglomerate seems to indicate, as has been suggested above in the case of the Portree volcano, that after the eruptions ceased each vent remained as a hollow or maar on the surface of the lava-fields. And the manner in which they are filled with successive sheets of basalt shows that in course of time other eruptions from neighbouring orifices gave forth streams of lava which, in flowing over the volcanic fields, eventually buried and obliterated each of the vents.

In the destruction of the precipice some of the vents have been so much cut away that only a small part of the wall is left, with a portion of the agglomerate adhering to it. The third neck, for instance, affords the section represented in Fig. 313, where the horizontal sheets of basalt (a) have still a kind of thick pellicle of the volcanic detritus (b) adhering to what must have been part of the side of the orifice of eruption. The waves have cut out a cave at the base, so that we can, by boat, get behind the agglomerate and see the margin of the volcanic funnel in the roof overhead.

The fragment of geological history so picturesquely laid bare on the Stromö cliffs presents a significant illustration of what seems to have been a frequent, if not the normal type of volcanic vent in the Tertiary basalt-plateaux. By the fortunate accident that denudation has not proceeded too far, we are able to observe the original tops of at least two of the vents, and to see how such volcanic orifices, which were doubtless abundant all over these plateaux, came to be entombed under the ever-increasing pile of accumulating basalt.

There is still one feature of interest in these cliff-sections which deserves notice here. Every geologist who has studied the composition of the basalt-plateaux has remarked the comparatively insignificant part played by tuffs in these volcanic accumulations. Hundreds of feet of successive basalt-sheets may often be examined without the discovery of any intercalation of fragmental materials, and even where such intercalations do occur they are for the most part quite thin and extremely local. I found it impossible to scale the precipice for the purpose of ascertaining whether around the Stromö vents, and connected with them, there might not be some beds of tuff interstratified between the basalts. If such beds exist, they can only be of trifling thickness and extent. Here, then, are examples of once active vents, the funnels of which are still choked up with coarse fragmentary ejections, yet from which little or no discharge of ashes and stones took place over the surrounding ground. They seem to have been left as crater-like hollows on the bare surface of the lava-fields.