Before taking leave of the reader it seems advisable to indicate briefly the general geographical results of our two seasons of exploration in the interior of Spitsbergen, and to state what is now known about the structure of the surface of one of the most interesting areas of arctic land. On NordenskiÖld’s chart, the best map of Spitsbergen existing at the time when we began our labours, both Garwood Land and King James Land are described as covered with “inland ice.” Now, if the phrase “inland ice” merely means glaciers, so that it may be correctly applied to the glaciers of any district of snow-mountains, such as the Alps or Caucasus, it is a useless phrase, and ought to be abolished. Most persons of whom I have inquired receive from it a different impression, and judge it to be descriptive of a complete and continuous icy mantle enveloping a whole country, as Greenland, for instance, is enveloped. In fact, Nansen, in his book on Greenland, always uses the term “inland ice” to describe the great interior ice-covering. “Ice-sheet” is apparently a better descriptive term for such a mantle, and I shall accordingly so employ it. The term “inland-ice,” being essentially vague, should, I think, be erased from geographical literature, or only used as an indefinite term for the land-ice of an unexplored region, the exact nature of which is unknown. As long as a flowing body of land-ice is contained within definite watersheds and mountain ranges, it is a glacier and not an ice-sheet. The juxtaposition of no matter how many glaciers does not form an ice-sheet, but merely a glacial area. It is necessary to be thus particular in definition because, as has been stated above, neither Garwood Land nor King James Land, nor any large part of Spitsbergen, except New Friesland and North-East Land, is covered by an ice-sheet. They are all merely glacial and mountain areas. The discovery of this fact is the principal geographical result of our second expedition. That it is a not unimportant result I now proceed to demonstrate.

The old theory that glaciers not only polish but systematically excavate their beds is practically abandoned. Its supporters naturally considered that the larger the mass of ice the more vigorous would be its excavating action. A great arctic ice-sheet was regarded as an extraordinarily powerful excavator. We now know that moving land-ice does not so operate upon its bed, but, beyond polishing the surface of the rock it covers, has mainly a conservative effect upon it. In the case of a country like the interior of Greenland, wholly buried under ice, the buried land-surface undergoes modelling to a very slight degree, except round the coast. On the other hand, in the case of a glacial region, where mountains rise above the mean level, and where rock-faces are exposed to the rapid denudation that takes place at all snowy elevations, great developments of surface-formation are going forward. In the case of an ice-sheet, the forces acting on the land-surface are conservative; in the case of a glacial region, the acting forces are formative. Hence the immense importance of clearly distinguishing between these two types of ice-bearing country.

Without pausing to describe the particular places or views in Spitsbergen that suggested particular conclusions to my mind, let me rather, for briefness, indicate how it seems to me that one or two well-known mountain groups in Europe have been acted on by glaciers—for instance the Mont Blanc and Bernese Oberland ranges. Both, in their present developed condition, have been carved out of more solid masses which may be described as originally wrinkled plateaus, the original wrinkles having been approximately parallel to their length. Of course the denuding forces, whatever they were, operated simultaneously with the elevating forces; but the two may be considered separately for convenience’ sake, and we may speak of the plateau as first elevated and afterwards denuded. It must, however, be understood that during the earlier stages of the elevating process, water, not snow and frost, was the denuding agent. The culminating point of each plateau was approximately in the position of the highest point of the present ranges. The original main drainage must have run along the lines of the wrinkles; now, in both cases, it runs at right angles to that direction.

In order to indicate my meaning, it is not necessary to reconstruct entirely the original form of the plateau and its lines of drainage; one or two instances will suffice. In the case of the Mont Blanc range,[15] I suggest that originally there was a glacier with its head near the present summit of Mont Blanc, having for its left bank a ridge (or plateau-edge), now represented by the Aiguille du Midi and other aiguilles, the Aiguille Verte, the Aiguille du Chardonnet, and the Aiguilles DorÉes; whilst its right bank was approximately coincident with the modern watershed as far as Mont Dolent, except between Mont Blanc de Courmayeur and the Tour Ronde, where it has been denuded away. This ancient drainage system has been broken down, and now the snows of the upper reservoirs are all discharged by such glaciers as the Mer de Glace or the Glacier d’ArgentiÈre, which cut across one or other of these old containing ridges or plateau-edges. Similarly with the Bernese Oberland, I suggest that the original crinkled plateau was drained along depressions approximately parallel to its length, whereof one was a high glacier basin with its head near the top of the present Finsteraarhorn and flowing W.S.W. over the GrÜnhornlÜcke and the LÖtschenlÜcke and down the LÖtschenthal. The old watersheds to right and left of this glacier have been driven back by the general disintegration of the plateau-edge, and broken utterly down in various places, so that its snows are now drained away at right angles to its direction by the Great Aletsch and Walliser Viescher glaciers.

In fact, in these cases it is with the glacial drainage as in the Himalayas it is with the rivers. When the great Asiatic plateau was elevated, whereof Tibet alone retains anything approximating to the original surface condition of the whole, the drainage ran off along the hollows in the line of the crinkling of the surface coinciding with the strike of the strata. Now, however, by the operation of rivers eating their way back into the plateau at right angles to the strike of the strata, all the great rivers flow at right angles to their original direction. The Indus was originally a stream no bigger than the Swat River, flowing down the edge of the elevated region. It ate its way through the Nanga Parbat range into the depression which goes on to Gilgit, and thus it stole all the waters of the upper Indus of to-day, which in the remote past, I believe, discharged themselves (over a high region since excavated into mountain ranges) into the Kunar River, and before that into the Oxus. Similarly the Gilgit River has eaten back through the Rakipushi range and stolen the waters of the Hispar-Hunza valley and the Hunza stream has eaten back through the Boiohaghurdoanas range, and so reached the Kilik Pass. It is noticeable that, in each case, the river has broken its way through a range in the immediate proximity of its highest peak, that is to say, just where the fall and gathering of snow has been greatest and the denudation most energetic.

In the case of rivers the eating back process is well recognised and understood. It is not really the work of the river, but it is accomplished by the various forces of atmospheric denudation, by frost and thaw, by avalanches and so forth, all taking place about the head-waters of the stream. I suggest that, under the action of similar forces, glaciers likewise creep back, and that the modelling of snow-mountains out of high plateaus is largely due to this process. According to this theory, though glaciers do not excavate their beds to any great extent, they widen them by carrying away the results of atmospheric and other denudation, and similarly they eat back at their heads. The most striking examples of this process I have seen are in Garwood Land. There, far in the interior, are a series of cliffs, several hundred feet in height. What the origin of these cliffs may have been is immaterial to the question under consideration. They form the front of the remains of the old plateau, which is being and has been eaten away. At the foot of the cliffs are the snowfields of the great glaciers which flow thence in a south-east direction to the head of Wybe Jans Water. By the melting of the snows above the cliffs and on their ledges, and by the action of frost and thaw, the rocks are rapidly broken up. The dÉbris fall upon the glaciers below, and are carried away. If there were no glaciers in this position, the dÉbris would pile up, a slope would be formed, and would presently reach up to the top of the cliff, and protect it from further denudation. The presence of the glaciers below prevents the dÉbris from collecting. The cliff thus continues its existence, and merely moves backward by a steady progress, just as the cliff retreats over which Niagara falls. Where weaker rocks are encountered, or denudation is locally more energetic, the cliff eats backward more rapidly. An embayment is formed, which tends both to widen and to creep backwards, becoming in time a tributary valley. Of such valley heads which have crept back into the plateau we saw several examples; one in particular I remember in the midst of King James Land, which had annihilated a portion of a mountain range dividing two great glaciers, and had thereby caused what had originally been the chief nÉvÉ basin of one of these glaciers to drain into the other instead of down its own tongue. When two neighbouring embayments, reaching back from the lower level into a plateau, send arms to join one another, or meet obliquely, a nunatak is formed. The nunatak near our farthest point in Garwood Land was produced in this manner.

Keenly possessed by the memory of these phenomena, I went recently to Grindelwald, and was immediately struck by the resemblance in character between the great bluffs of the Bernese Oberland—the Eiger, Mettenberg, and Wetterhorn—and the bluffs of Spitsbergen’s Sassendal. The latter, as we know, were formed, and are still in process of development, by means of the torrents draining the snowfields above, which eat away the plateau and cut back into it, thus carving out a row of flat-topped steep-fronted hills that jut forward into the ever-widening main valley. It seemed evident that the ancient Oberland plateau had been similarly cut down, the excavation not having been accomplished by the grinding action of glaciers pushing forward and filing down their beds, but by the action, first, of torrents, before the plateau was elevated above the snowline, afterwards of glaciers; both torrents and glaciers creeping backwards at their heads, where faces of rock are exposed to rapid atmospheric denudation, and the dÉbris that fall are transported to low levels by the movement of the flowing ice.

It was thus, I suggest, that the Upper and Lower Grindelwald glaciers and the Rosenlaui Glacier invaded the plateau and crept back into the heart of the mountain mass, isolating as high individual peaks the Wetterhorn and Schreckhorn. Originally they were “corrie glaciers,” plastered on to the north face of the plateau—just such glaciers, in fact, as is the Guggi Glacier, which lies in the hollow between the Jungfrau and the MÖnch. They have crept farther back than it, because they had the better start, but the Guggi Glacier now emulates their former vigorous initiative. The cliffs at its head are being continually broken and worn away by the action of frost. The rocks that fall from them either tumble on to the nÉvÉ and are carried down or roll into the bergschrund, and so get under the ice, where no doubt they are ground to dust, and may do some excavating in the process. That, however, can only be in the upper regions; lower down, the waters below the glacier are the excavating agent, rather than the glacier itself, except, perhaps, at the edge of some sub-glacial cliff beneath an icefall. In this way the rocks of the north face of the ridge between the Jungfrau and MÖnch are being eaten away, and the ridge itself is not merely being lowered, but its crest is being pushed backward towards the south. Every yard of its movement is made at the expense of the Jungfrau Glacier. Let the process go forward for a sufficiently long time, and the area now occupied by the upper basin of the Jungfrau Glacier will be occupied by a snow-basin lying at a lower level, and draining northward down the Guggi Glacier.

Similar, I suggest, was the development of what is now the Great Aletsch Glacier. Originally, according to this theory, the LÖtschen Glacier stretched back to the Finsteraarhorn, and had for its left bank a ridge parallel to, but south of, the range of which the Aletschhorn is now the culminating point. The Aletsch Glacier’s original head was on the south face of this range, but the glacier ate its way backwards, its head advanced to the north, finally broke its way right through the range and drew off a portion of the ice of the LÖtschen Glacier.[16] The snout of the LÖtschen Glacier was thus disconnected from its former nÉvÉ, and a pass (the LÖtschenlÜcke) was formed between them. The nÉvÉ, at what is now called the Place de la Concorde, flowed as a great icefall over the remnant of the old left bank of the original glacier. It no doubt deepened and widened the breach, and, as it did so, lowered the level of the snow in the upper reservoir, whose various branches were thus likewise enabled, each in its place, to creep backwards at the expense of the plateau. In this manner were formed the Ewig Schnee Feld, the Jungfrau Firn, and the other nÉvÉ tributaries of the present great glacier. The great icefall gradually diminished in turbulence as the cliff beneath it was broken and rounded away, till now it is merely represented by the crevassed area just below the Concordia Hut.

If there is any truth in the theory thus briefly propounded, in a form which must be considered altogether incomplete and preliminary, it follows that the distinction I have endeavoured to make between an icesheet and a congeries of glaciers is a distinction of the first importance; for under an icesheet none of the processes are going forward which are vigorously proceeding in a glacial region. The old idea of Spitsbergen was that its interior consisted of a great icesheet, fringed at the edge by a number of boggy valleys and green hillsides. Our explorations have shown the utter falsity of this conception. Let me now briefly indicate the outlines of the true geography of the main island.

Whether at one time the whole island was enveloped in an icesheet which was gradually withdrawn from the west towards the east, or whether the west part of the island has merely been longer raised above the sea than the east part, I do not attempt to determine. At any rate, it seems to be a fact that the forces of denudation have been longer at work, or, at least, more vigorously at work, all down the west part of the island, and that the resulting mountain formation is most developed in the west, and becomes continually less developed as you proceed toward the east. All down the western region you find highly specialised mountain-forms—peaks and ranges of considerable abruptness and marked individuality. As you advance eastward the mountains become generally more rounded, till the original plateau-form, and even parts of the undenuded plateau itself, are encountered.

Bearing in mind this general structure of the land-surface, it will now be easy to describe the character of different parts of the main island. The whole of the north coast, as might be expected, bears evidence of a more rigorous climate than districts further south. This was specially noticed by us when proceeding down Wijde Bay, at whose mouth the snow lay down to sea-level in the month of August, whilst, twenty miles in, the snowline was almost 1000 feet above sea-level. The northern rim, therefore, may be regarded as a separate geographical division. At the north-west angle of the island is a region of very bold mountains and large glaciers. It is well represented by the beautiful and often described Magdalena Bay. Nothing is known about the interior south-east of it, but some old Dutch charts mark a valley leading from the extremity of Mauritius or Dutch Bay up to a sequestered lake in the hills. Whether the draughtsman intended his winding valley and river to represent a glacier and the lake a snowfield, or whether a true lake and river existed here in the eighteenth century, can only be settled by some one going to look.

Passing southward down the west coast, we come to the seven parallel glaciers ending in the sea, known to the whalers as the Seven Icebergs. These all appear to flow down from a high common snowfield which stretches east toward Wood Bay and south almost to the head of Cross Bay. South-eastward this high plateau is broken by a series of nÉvÉ-valleys, the chief of which discharge themselves towards Ekman and Dickson bays. Their general direction is south-south-east. South of this plateau region comes the mountainous area of King James Land, whose character has been described in this volume. The main watershed runs north and south. A series of parallel glaciers drain south-south-east from it to Ice Fjord. The valley system on the west is less regular, but the glaciers are equally numerous and fine.

The deep north-and-south depression filled by Wijde Bay and Dickson Bay is bordered on the west by a range of mountains, a group of which intrude between and divide the bays. Some of these are of striking form, but no one has ever been amongst them or accurately determined their position. East of the two bays comes the plateau region. Its edge is cut up by a few deep valleys, down which the icesheet of New Friesland sends glacial tongues to Wijde Bay, but east of Dickson Bay the marginal valleys are longer, and no glaciers come out of their mouths. The portion of the plateau between Dickson and Klaas Billen bays is a good deal cut up by deep valleys, such as the Rendal, the Skans valley, and the Mimesdal (all well known to geologists), but there are no large glaciers found upon it. Further east comes a great glaciated area approximating to an icesheet in appearance, but with many exposed faces and peaks of rock. From it several large glaciers flow into the sea, namely, the glacier that ends in the head of East Fjord of Wijde Bay, the glacier that fills a wide valley debouching into Hinloopen Strait opposite the South Waiigat Islands, some more glaciers that empty into Bismarck Strait and that neighbourhood, the series of great glaciers at the head of Wybe Jans Water, and the NordenskiÖld Glacier (specially explored by us) near the head of Klaas Billen Bay. All these glaciers are divided from one another by more or less well-marked watersheds.

The neck of Spitsbergen, which may be defined as bounded on the north by a line from the mouth of NordenskiÖld Glacier to Wiche Bay, and on the south by the Sassendal and the depression across to Agardh Bay, is a district that would well repay exploration, and is easily accessible from the Post Glacier at the head of Temple Bay. Nowhere are better illustrated than here the phenomena of mountain formation by plateau degradation under the action of rivers and glaciers. In the east are the remains of an ice-sheet; in the west are deep and wide glacier and river valleys. Between the two are many mountain ranges, and some peaks of considerable height and abruptness.

A line drawn from the head of Van Keulen Bay to Whales Bay forms the southern limit of the next region to the south—the region that I call Adventure Land, using the old name which in the case of Advent Bay has been clipped of its last syllable in the present century. It is a country of boggy valleys, rounded hills, and relatively small glaciers. Originally it was one large plateau formed of soft, almost horizontally bedded rock, except along its west margin. It has therefore been penetrated by wide valleys radiating in all directions and cut down almost to sea-level. A range of rather fine peaks lies along the west coast; behind them are some large glaciers descending north into Green Harbour and south to the mouth of Low Sound. Then the undulating country begins. Several valleys lead inland from Coles Bay, whilst from Advent Bay starts the Advent Vale with its many branches. From Low Sound a series of boggy valleys strike in to north and south. At the north angle of its head opens the deep valley of the Shallow River (after the Sassendal the largest valley in Spitsbergen), whose upper part has never been explored. The eastward prolongation of Low Sound, which was known to the Dutch as Michiel Rinders Bay is very poorly charted, but we know that at its north angle there is a secluded inner harbour, with a big ramifying valley leading back from it, while at its extreme east corner three large glaciers debouch together. One of these probably connects by a high snowfield with the head of Strong Glacier descending to Whales Bay.

Last comes the south division of the island, over which we had a panoramic view in 1897 from the summit of Mount Hedgehog. Unfortunately a roof of cloud covered the glaciers, and we could only see tops of mountains rising clear above it. The north-west angle of this region was explored in 1897 by Mr. Victor Gatty,[17] who found it to consist of a ring of snowy mountains surrounding the nÉvÉ of the Fox Glacier, which discharges into the so-called Recherche Bay. A gap or col, south-east of Dunder Bay, separates this group from a range of hills running for some distance south along the coast, and called Roebuck Land. The extremity of these hills abuts against the right foot of Torell Glacier, one upper bay of which rests against the hills immediately south of Recherche Bay, whilst another stretches inland to the east as far as the main watershed of the island. There are one or two other approximately north and south ranges of hills lying west of this watershed. East of it the plateau-character resumes its predominance. The southernmost part of the island, south of Horn Sound, is dignified by the boldest mountain range in the country, that of the Hornsunds Tinder, which lie west of the watershed, and run almost due north and south. East of them are at least two lower parallel ranges, beyond which the ice-covered country seems to dip to the sea.

Of the other islands in the Spitsbergen group, North-East Land is the largest. It is known, from Baron NordenskiÖld’s exploration, to be covered with a true icesheet, the edge of which descends to the sea all along the south-east coast. The north coast and the small islands off it altogether resemble the northern belt of the west island. The west belt is a low undulating region, from which the icesheet has retreated in relatively recent times.

In the sea east of Spitsbergen are two islands whose existence has long been known. They were named Wiche Land, after an old navigator. Walrus hunters have landed on them, but they were first really explored in 1897 by Mr. Arnold Pike.[18] The west island, now called Swedish Foreland, has a high flat-topped backbone. The east island, King Karl’s Land, consists of two hills, about 1,000 feet high, united by a low flat isthmus. There is no ice-sheet on either island and only small unimportant glaciers.

I have never landed on Barents or Edge Islands, though I have seen them from east and from west. Neither possesses an icesheet. Both are practically devoid of glaciers down their west coast, and have large glaciers in the east. The whole of the south-east of Edge Island is occupied by a great glacier ending in the sea. Barents Land has several sharply pointed peaks, but the Edge Island hills are mainly flat-topped, like those along the east coast of the main island.

Prince Charles Foreland now alone remains to be considered. It is very badly represented on the existing chart. At its southern extremity is an isolated hill. Then comes a very flat plain of about fifty square miles, raised but a few feet above sea-level. North of it is a mountain range consisting of fine, sharp snow-peaks. It is cut off on the north by a deep depression, running in a south-west direction from Peter Winter’s Bay, which, though marked south of St. John’s Bay on the chart, lies some miles north of it. North of Peter Winter’s Bay and Valley the mountain range is continued; but the peaks, though fine in form, are not so high as those of the south group, but they send down eastward an almost uninterrupted series of glaciers into Foreland Sound. Further north are yet lower snowy hills, which end in the bold headland called Bird’s Cape or Fair Foreland.