H rǒthlisberger:

Before we start a heated discussion which will be difficult to stop, I have made a 1ittle drawing showing the transition zone from unfrozen bed to frozen, about which various questions may be raised:

The first question concerns differential shear. If we have bottom sliding further up- and no sliding down-stream, do we have, between these two zones, an area in the ice, which may be more or less extended, of discrete shear, or what could better be called a thrust plane or fault? We will then have to consider, as we did in the first session, water at the bed, what effect it may have in the various models, and where it goes. Incidentally, observations from the Axel Heiberg expedition, that have not been mentioned yet, show that water does come up exactly along the shear zones. Then we heard about the flow of ground water below the glacier bed. We may come back to discuss water in channels or sheets, and what its effects amount to. If we go further up-glacier, there may also be some intermediate storage, as is well-known from Antarctic radar records. Similar water bodies may also occur closer to the edge where we have a continuous flow and some storage in between. So there is a diversity in natural phenomena. but even more diverse are the methods applied to the study of them. Among them stands out continuum mechanics. which may be wholly or partially applicable. Then there is the approach referred to. among friends. as speculative discontinuum mechanics. We may even use observations, equally intuitively or speculatively interpreted. Please let us have more discussion of these points.

R L Hooke

I would like to try to clarify further my thoughts on shear and debris transport in glaciers, discussed at some length earlier in the week. and to invite further conment from participants.

Firstly, the existence of a foliation defined by alternating bands of debris-bearing and clean ice is not evidence for discrete shear along the debris bands. nor for shear being responsible for incorporation of debris into the ice. Two specific lines of evidence mitigate against such an interpretation:

• (1)Experimental data (Reference Hooke, Dahlin and KauperHooke and others 1972) suggest that debris concentrations in excess of about 10% by volume stiffen ice rather than soften it

• (2)Measurements and theoretical calculations indicate that foliation exposed at the surface near a glacier terminus is not. in general, parallel to the direction of maximumshe:3r strain-rate (Reference Hooke and HudlestonHooke and Hudleston 1978).

Secondly, two of tile most commonly cited lines of evidence in support of the shear hypothesis can be readily explained by alternative processes for which solid field evidence exists:

• (1)Down-glacier facing steps in a glacier surface are frequently the result of differential ablation. with ice down-glacier from the step melting faster due to a thin dirt cover.

• (2of factors affectin)Clean ice beneath debris-bearing ice in the type localities of the Thule-Baffin moraines can be explained without resorting to a thrust-type mechanism in which exceptionally high shear strain-rates occur across relatively thin zones (Reference HookeHooke l973)

Alternative explanations should be considered for other features which suggest thrusting in ice. For example, some up-glacier dipping discontinuities overlain by recumbent folds may actually represent the strongly attenuated lower limos of these folds.

In conclusion, the question is not one of whether shearing occurs as a normal process of glacier flow: it does. Nor is it one of whether zones of higher shear strain-rate in clean ice, which are a few centimetres in thickness, exist: they do (Reference Hudleston, Saxena and BhattacharjiHudleston l977). Nor is; t one of whether thrusting can occur. particularly in ice under low hydrostatic pressure near the glacier surface: it probably does. Rather the question is whether dirt is either incorporated into the ice by such shearing or, once incorporated, is transported to the glacier surface along such shear zones. This has not been demonstrated. To demonstrate it, we need, among other things, detailed velocity profiles across debris bands demonstrating that shear strain-rates are much higher in the dirt band than in the adjacent ice.

A Iken

You said that the observed thrust planes are usually not in the direction of maximum shear. If I remember rightly, Haynes from CRREL (Reference HaynesHaynes 1973) did laboratory experiments on fracture of ice and got the same result. For a sample under compression, the directions of maximum Compression and the smallest principal compression are at 90°, so maximum shear would be at an angle of 45°. From Haynes' results, however, it follows that the normal to the fracture plane ideally makes an angle of 29° with the tensile principal stress if the other stresses are compressive. I wonder whether this direction suits your observations better? My second point is that the shear fracture does not occur inside the dirt zone, but beside it, on top of it. So maybe it is a sequence of processes, that debris gets into a fresh shear plane which later is abandoned, and then a new one forms.

Hooke

This divergence of angle is more than I was referring to. What we find is that if you make measurements close to the glacier surface the foliation dips steeply up-glacier at the surface, but decreases in dip rapidly with increasing depth. We cannot come up with a specific angle between the foliation direction and the direction of maximum shear strain-rate as that angle varies in space. Near the base of the glacier the two are probably very nearly parallel. We have no measurements, but certainly feel that is the case. It is near the surface of the glacier that the two begin to diverge.

I think the second point is a valid one, but if the shear is above the dirt band, it is a question of what the role of that shear is in moving the dirt to the surface.

R A Souchez

I am working on the variations of 6D (deuterium) and SO¹⁸ in ice of a glacier basal sequence from Bylot Island in the Canadian Arctic. There are numerous dirt layers in the ice and it appears that previous investigators believe that the entire basal sequence was incorporated by freeze-on at the base of the glacier. If you study both δD and δD¹⁸, not simply δD¹⁸, you Can distinguish between ice that is refrozen at the bed and glacier ice. If you do that, you find that the ice which is refrozen at the bed is only a thin layer and the layers between are glacier ice. So you have a repetitive sequence of thin layers of ice which refroze on the bed and glacier ice. Looking closely at the structure you find very elongated folds which give rise to this repetitive sequence. This point favours major folding at the glacier base including glacial ice within the basal sequence. along the lines of Hooke’s idea.

Hallet

With reference to folding of basal ice, I think it would be appropriate to ask Dr John Shaw to express some of his ideas on the Subject

J Shaw

I have been wondering a little bit listening to the discussions what the motivation for studying glacial erosion and sedimentation is. I would have thought that much of the motivation would be to understand the landforms and sediments that result. But I have not noticed the glaciologists paying much attention to either landforms or sediments. I was very surprised to hear a point being brought up by several people that folding in the ice could possibly cause foliation; this is an idea that has been around for a long time and we are just recycling things. However, another important element is that, if we do get folding, we increase the debris volume over an area of the bed. Possibly this could directly lead to the formation of a landform. There are two ways in which it could not: if deposition occurred by lodgement on the bed and these folds were continuously dragged along, then the structures would be completely destroyed, and, alternatively, if the surface of the ice were to melt down and this soft sediment was let out at the surface, it would flow, and the structures would be destroyed. The final possibility is that we could simply melt out the ice in situ from a stagnant body. In this case the structures and the form will be preserved, but, of course, compacted.

Rogen moraines are an example of forms produced in this way. They appear in Sweden quite close to here [i .e. Norway]. The reason I bring them up is because I talked to several glaciologists at the beginning of the week and they did not know what they were. There are several important points about Rogen moraines. First of all, they are transverse ridges. Second, they tend to be "drumlinized" at their up-stream end. Third, they have flutings on the surface. Fourth, eskers which run through fields of these features tend to be of a younger age. Given these observations, the Swedish geomorphologists long ago decided that Rogen moraines in some way formed as a result of active ice. The proximal side of Rogen moraines is always extremely plane. and the distal side irregular. However, morphology is not sufficient to determine the mode of formation. We must also look at the internal structure. If we look at the internal structure of the moraines. we can recognize folded beds of Clearly differentiated till. Cross-cutting these folds are bands of sorted sediment. It seems that as we have folding of the till, but the stratified layers are horizontal. the stratified layers did not exist when the folding occurred. They must have developed subsequently. So we have folding occurring in the active stage, and then deposition by melt-out in an englacial position, probably caused largely by water flowing through a debris-rich ice system. At the time these observations were made. I knew of no modern-day analogue for Rogen moraines. However, some close similarities exist between the observations of David Croot on modern glacial processes and the above deductions based on Pleistocene landforms and sediments.

The point I would like to make is that by studying Pleistocene sediments we have access to things that are probably inaccessible in the modern environment: the bottom of the ice sheet and processes which may not be occurring at present. I think that if the glaciologists would turn their attention to solving problems as to how such things as folding occur and the physics of melt-out, it would be helpful. The geomorphic conclusions presented are derived more on the methods of Sherlock Holmes than John Glen. Maybe the John Glens of this world can now apply their more quantitative skills to similar problems.

J T Teller

Roger Hooke stated that debris strengthens ice rather than softens it and therefore a debris-rich plane is not a likely avenue for shearing or thrusting. I would like to ask if a shear plane that was initiated in clean ice, but which progressively became enriched with debris. Might still continue to be the weakest plane within the glacier? That is, it seems that once movement along a shear plane develops. the minimum stress required for continued movement will coincide with that plane, even if debris is moving along it from a near-basal position to the glacier surface

Hooke

Firstly, a plane is a two-dimensional feature but ice crystals and debris particles are three-dimensional. This distinction may not seem important, but it has always been a stumbling block for me. A shear plane exists, for example, between two cards moving one over the other. In ice, however, the mechanical problem is easier to visualize if we think in terms of shear zones, a few millimetres to perhaps a few metres thick.

On the Barnes Ice Cap, Hudleston has measured crystal orientations in shear zones a few centimetres thick and finds strong single-maximum fabrics with c axes oriented perpendicular to the Zone. Such fabrics no doubt tend to soften the ice. This could possibly more than offset the hardening effect of any debris that might be present in the zone. (There has not been any debris in the zones we have studied, however.)

Of greater importance is the question of the role that such zones might play in entrainment and transport of debris. Regarding transport, our measurements, and I believe those of many others, have failed to demonstrate a significant difference in shear strain-rate across. debris-rich bands that have produced ice-cored ("shear") moraines at the glacier surface. As ice beneath such bands is not stagnant, shear zones are not necessary for formation of such moraines. Regarding entrainment by shear, so far as I know no one has presented a clear mechanical model for this process, so let me suggest one. Possibly ice surrounds a debris particle resting on the bed and then moves it forward over stagnant or slowly deforming ice. (If the latter ice is slowly deforming, the flow must be three-dimensional to avoid a space problem.) Detailed strain measurements would be necessary to demonstrate the reality of such a process and especially the role of Shear zones in it. If ice with a transverse flow component simply closes beneath the particle, as may occur during plucking, the process might better be called entrainment by plastic flow, thus contrasting it with entrainment by regelation.

R P Goldthwalt

I have a couple of observations to make. One is in connection with John Shaw's remarks. Near the margin of Casement Glacier, tunnels carne in behind what was essentially a roche moutonnee form or knob(Reference PetersonPeterson 1970). The tunnels intercepted an open space in the lee of the roche moutonnee over which ice was sliding at 2.9 cm d^-1. The ice looked like dirt, and indeed the bottom 0.1 to 0.2 m was till like debris. When we tried to work there it warmed up a little and the debris started to fall off on us. In such a situation on the lee side, you lose some of the diagnostic items that have been attributed to lodgement till. Lodgement type debris with oriented pebbles gets all mixed up as it peels off. As you go back from the margin you can imagine closure rates which might make large cavities impossible. What happens there I am not quite sure, but I am sure that the fabric under thin ice gets all mixed up over boulders and bumps.

I should bring up a second point which is pertinent, but old. We tunnelled into the Red Rock ice cliff (Reference GoldthwaitGoldthwait 1960) beneath 30 to 41 m of ice. In that tunnel we drove two shafts down and found that the first centimeter of ice above the bottom, which lay on miscellaneous rocks and vegetation, moved not at all. We put in steel pegs and measured flow rate. As we came up 20 mm above ground the ice moved a little (0.48 mm d⁻¹) and still more rapidly above (2.72 mm d⁻¹) at 5 m above ground). The surface of the glacier was moving at a rate of 12.1 mm d⁻¹. This differential motion is what I think of as an internal shear. Now at that particular point, more than 30 m back from the cliff under fairly thick ice, we had no significant amount of dirt except a few little lenses coming along in the lower 1.3 m of ice. We do not know where they tame from because this basal ice is obviously frozen to the ground beneath it and doing no erosion. Out in front, however, and in the outer part of tile tunnel and side tunnels, there were great intersecting folds of thin dirt bands and, in some cases, pods of dirt along the bands to as much as 10 m above the ice base here and 30 m in other places. Detailed petrographic work in the tunnel Shows that the dirt was accumulated in front of the ice cliff in snow-drifts which build up annually and into which a certain amount of dirt is incorporated. As the ice front advances it incorporates, this material as part of its body. On the inclined toe below the ice front, bands of dirt increase in slope as they come under the ice cliff. By the time this ice is overridden 8 m into the glacier, these bands have overturned and are well-folded. We believe the dirt rises and is accumulated and recycled right at the front of the glacier. We have no evidence on the basis of materials found that any of the debris came from far up-glacier, but it did come sometime.

We dated some of the organic debris in the pods and the age was around 4760 ± 220¹⁴.C a. On the stones underlying the glacier 16 to 30 m back, we found lichen which our lichen expert, Dr Wolfe, tried to revitalize. He felt that it was viable although we never got much growth. The material around those rocks dated around 200 years old, so presumably the advance occurred over here within the last 200 years.

D J Drewry

I would like to make additional comments to those of Roger Hooke in regard to whether shear planes bring up material, or whether material already in the ice encourages strong differential motion. There are one or two observations that we can take from ice sheets away from the very complicated marginal zone (where flow lines move up towards the surface and complications arise due to complex transverse and longitudinal strain).

Observations from Greenland and Antarctic ice cores show that very fine-grained sediment layers tend to be associated with very fine-grained ice having a high degree of c-axis verticality, and that there is strong differential motion associated with these layers. The fine-grained sediments have not been brought from the bed, which may be both several thousand metres deep and several hundreds of kilometres up-stream, but have been deposited on the ice-sheet surface (eventually forming deep isochronous horizons) by such activity as vulcanism. These fine-grained sediments locally enhance the ice creep rate. At Byrd station, for instance. there are thin bands of volcanic ash between 1200 m and 1800 m. Fabric studies of the ice column by Tony Gow show that, in the fine-grained debris band, the c axes have a very high vertical orientation and that there is strong differential movement in this zone. Nobody who has looked a t the Byrd care or the dynamics of Byrd station in relationship to the west Antarctic ice sheet would interpret these debris horizons and strong shear motion as having anything to do with material brought from the bed; it is a volcanic horizon enhancing creep in the ice sheet.

Hooke

Could you comment on the concentration of debris in those bands?

Drewry

The concentration was very, very low. Debris formed cloudy bands, containing volcanic glass of less than 5μ size. In this case we are not actually stiffeni.ng the ice but introducing extra dislocations which enable the ice to creep more rapidly

Röthlisberger

In relation to David Collins' concept of various melt-water routings through the glacier, I would like to point out that I expect the main drainage channels to be located at the bed. This can be inferred from the fact that the bed is undoubtedly the principal longitudinal discontinuitv extending over the whole length of the glacier, as well as from the consistent occurrence of heavy sediment load even in cases where the glaciers end in lakes or fjords. Indeed, the bulk of the melt water is successfully captured in sub-glacial intakes of existing hydro projects, although, as happened at Argentière, sometimes at changing locations of the bed. An englacial water conduit has nevertheless been reported temporarily, 0.5 m above the rock bed, at Argentière also. The englacial flow, however, occurred in this case in a severely disturbed zone where seracs are moving down a major rock step. It represents therefore an exception rather than the rule:

I would like to add that the picture outlined in connection with the plucking mechanism is part of the story only where subglacial drainage is concerned. At the large pressure fluctuations of the melt season, channels incised from below into the glacier sole can form over and over again in the same location relative to the glacier bed, so that water follows more or less the same course. During the steady water flow of winter, to the contrary, the channels are more likely to move along with the ice, thereby coming continuously into contact with new areas of the bed (provided there is sufficient bed slip). This might be one possibility for explaining tile high concentration of solutes in winter, and I am interested to hear from David Collins whether this idea is compatible with his findings or not.

The lifting up and rearranging of channels can lead to more severe changes, as in Argentiere where the main channel shifted from one side of the glacier to the other leaving the subglacial water intake without a source of supply. The ice thickness changed and so the whale system changed. Usually these changes happen in spring when the first melt water comes and the glacier gets a big push. Something similar also occurs locally, of course, and this may account for the highly variable concentrations of sediments with time. For example, new moulins develop and we get new crevasses, so at these places it is clear that when these events occur, we get a surge of fresh material

D N Collins

I would agree with much of what Hans Röthlisberger has just proposed. One thing to think about is the idea that at the bed not only is the glacier moving, and delivering sediment into small channels. But also that the channel system itself, particularly smaller channels, can move across the bed. At least channels which are incised upwards into the ice can move. However, I imagine that the main arterial channels win be incised down into the bed and therefore fixed in position. To actually explain how sediments and solutes vary with discharge, there must be a multi-component channel system at the bed. Perhaps big channels incised downwards are continuously scoured out of any sediment except during winter. In spring, sediment is flushed through and those channels never again contribute sediment during the season. But then the smaller channels have to be envisaged. These are always acquiring sediments and solutes as they migrate, and feeding them into the main channel from where they are effectively flushed from the system in summer.

I also like the idea of cavities at the bed which may change their sizes with differing water pressures in the main channels. Melt waters are led into temporary storage in cavities, and subsequently water returning into the main channel system has increased amounts. of solutes and sediment. I also agree with Hans Röthlisberger that englacial drainage in ice-walled channels above the bed is improbable. Where before I have talked about the difference between englacial and subglacial systems, my englacial system transports water which is not chemically enriched and includes the basal arterial channels in which I presume that further chemical enrichment does not take place.

Hallet

We have been mapping pro-glacial areas where the exposures have allowed us to decipher much about subglacial processes. Extensive networks of closely associated cavities and channels exist in these areas. In some places as much as 30 to 40% of the glacier sole was not in close contact with the bed. Therefore there is a tremendous potential storage of water under the glacier, and it seems from a number of lines of evidence that all these cavities at some time interconnect. Presumably as the water pressure drops or as the sliding velocity drops, the cavities will fill in with ice and can reform subsequently. Occasionally the areas where the glacier was in intimate contact with the bed seem to be flushed out, presumably as the water from the large-scale hydraulic network communicated with the basal film. Much of this story fits rather well with the solute record that Dr Collins presented.

Shaw

I would like to respond to Dr Röthlisberger’s and Dr Collins’ point that englacial channels are improbable. That may well be the case if the ice is active, but once the ice is stagnant, channels within the ice become very common. Much glacial sedimentation occurs at this stage.

The second point that I would like to make is that it seems that the people who are working with glaciological processes are looking at two scales: the very small scale on one hand, and a jump to the large scale of the whole ice mass. In doing so they neglect the transport of debris in the basal portion of the ice where we may have 40 to 60% by volume of debris. I would like to ask them why this is neglected, because so long as they neglect it they are not much use to those glacial sedimentologists and glacial geomorphologists dealing with depositional landforms.

Hallet

I would argue slightly with that. I think that for a glacial geomorphologist interested in erosional surfaces, it is quite appropriate to consider ice with sparse debris. Several of you probably have comments about the volumetric content of debris in basal ice, and others may want to comment on efforts on modelling debris-rich ice.

Drewry

One of the most valuable contributions to be made is communication between glacial geologists, glacial sedimentologists, and glaciologists to discuss what parameters are needed as input to models and how to measure them. When I started trying to find data on basal debris concentration, r found little of practical use. We have heard at this meeting many bulk values quoted, 0.1% up to 50 to 70% by volume. What we really require a.re vertical and horizontal variations and the disposition of debris; whether in stacked horizons, whether sediments extend for a considerable vertical distance within the ice. I would make a plea for glaciologically relevant measurements of sediment in ice masses.

D E Sugden

I would like to follow up what John Shaw said when he pointed to the difficulty of evolving effective links between those working on process studies under present glaciers and those working on landforms and sediments in areas formerly covered by ice in the Pleistocene. Two fields occur to me as being potentially very interesting: one is esker patterns. In many parts of Scotland it is quite common to find a very complex anastomosing pattern of eskers. When you map them in a bit more detail you find they seem to have a different morphology. Often in a whole system you might find one rather complex ridge with a very sharp-crested and clear form. The others may be successively less clear, with more signs of over-running and streamlining by ice. All I am suggesting is that perhaps these represent the channels that Hans Röthlisberger has been mentioning. An esker may build up in a channel at the end of one season: then the next season it may not reopen and another may open. "In the meantime the original esker deposit is beginning to become moulded by moving ice, and so on. Perhaps this line of attack might allow Pleistocene field studies to contribute to studies of melt-water flow in glaciers.

The other point I would like to make concerns water tables. In areas where down-wasting ice was a common feature, a lot of the landforms seem to be explained by geomorphologists and glacial geologists in terms of sediments deposited beneath a water table. These studies imply water is not necessarily flowing at the bottom of the glacier when there is a water table. Streams issuing from calving glacier snouts in Greenland fjords often emerge at the water level. I have seen this in many places in the South Shetland Islands, too. The channels emerge from the glacier at the water surface rather than the bottom of the glacier. Perhaps there is hope for interchange of ideas here

Hallet

I think the esker problem is one that is being explored rather fruitfully from a theoretical point of view. This is work that is being done by Professor Ron Shreve, who is considering the physics of tunnels in glaciers, calculating the equipotentials under the ice, and studying the different patterns of eskers. In places, eskers form complicated networks, and in others eskers occur by themselves. So the disposition and shape of eskers is a fascinating problem. I think we can look forward to some very interesting advances in that field following Shreve's introductory work on the subject (Reference ShreveShreve 1972).

J W Glen

I feel I should respond in some way to what John Shaw has been saying because I do not believe that glaciologists or ritvdellers are deliberately avoiding the evidence which glacial geologists want to put to us. I am continuously struck, not by the complexity of what we are asked to explain, but by the relative simplicity. For example, people talk about measuring the debris content in a glacier. Surely we would expect this to depend enormously on the nature of the rock over which the glacier was flowing, yet I do not see much discussion of that. Equally, the kinds of forms, be they erosional forms like roches moutennée, or depositional forms like eskers, or, dare I say it, drumlins (if they are a depositional form and not a depositional-erosional form. which is being debated), are relatively similar irrespective of very different types of rock over which the glaciers concerned have flowed. I for one got the message that these things were really rather similar irrespective of rock form and therefore what we had to do was to explain why that was so. That, I think, is what we have been trying to do. If we have got the wrong message I then tell us.

You say we work on a smaller scale. A thing like a roche moutennée I take to be what you mean by something on the small scale, and then, on a very big scale, I suppose you mean something like the Laurentide ice sheet. But I do not know what the phenomena are that we have to explain on some sort of intermediate scale in a way that we can go away and do it. Perhaps a valley glacier is one; plenty of people have worked on that. What is it that we are not doing?

Shaw

Yes, you are working on a small scale at the scale of a roche moutennée, and an even smaller scale when you are considering the effect of the single particle eroding the bed. And yes, you are dealing at a large scale with ice sheets such as the Laurentide ice sheet. An example of the intermediate scale is exactly what you mentioned, that is the drumlin. For a long time geomorphologists have been considering drumlins in terms of flow patterns in the ice. I think we have very good evidence that there are complicated flows associated with drumlins, but all we can do is establish a kinematic argument for such flows. I would hope the mechanicists could at least tell us if these flows are reasonable, and that is what I hoped they would do.

T J Kemmis

I would like to provide a partial answer to Dr Glen's question about what to investigate at the intermediate scale. To date we have looked at the scale of the individual process, but not at the intermediate scale of processes in combination. For an example, why not look at processes occurring in combination under temperate ice conditions? We know that under such conditions a large number of processes can take place. And yet we know temperate glaciers do not all produce the same product. Obviously there are different sets of processes taking place in various glaciers. We need to know what sets of processes may occur essentially simultaneously, what may not, and why.

S Haldorsen

I am always a little disappointed when I listen to lectures concerning very clean ice moving upon a very clean bed. In that connection, I have a comment on the peper of Bernard Hallet. Glacial abrasion does not only include abrasion of bedrock, but also abrasion of material in drift and already deposited till. When you are out in the field, you commonly find that the thickest and most massive tins in many cases are dominated by abraded clast material and such till usually is classified as lodgement till, i. e. till deposited from a sliding glacier. It is quite clear that much of the abrasion occurred in situ, just before, during, or after deposition. The very massive structure shows that the material hardly was deposited from debris-poor ice. In such cases I find it difficult to apply the idea that a debris-rich ice is not abrading.

Hallet

I would like to respond to that by saying that it really depends on your perspective. I have worked mostly in alpine areas, and there it is quite appropriate to look at ice that is relatively clean. The field evidence for this is clear. One can look at glaciers that are retreating and. except for terminal moraines where there is a bit of debris, only widely scattered rock fragments can be found on extensive bedrock exposures. Furthermore, on those rock. Surfaces extremely delicate features (like fragile spicules of subglacial carbonates) occur that would be readily destroyed by pro-glacial waters. By the fact that they are preserved almost intact, you can practically preclude the removal of any coarse debris. Looking at the occasional scattered rock fragments on a pro-glacial rock surface, and knowing something about the sliding velocity of the glacier and the retreat rate, one can figure out concentrations of debris in the basal ice. I can assure you that you have to stretch it to get a by volume in some of these glaciers. Now I am not saying that we do not have glaciers with thick debris bases, and I am not even saying that they cannot abrade. In fact one of the intriguing peculiarities of the mechanics of abrasion is that as glacier sliding slows down. the drag imparted by rock debris decreases. Slow sliding appears possible even when the glacier bed is entirely covered with debris. Perhaps this could lead the discussion in another direction. What is it that glacial geologists would like theoreticians to model in terms of the mechanical properties of materials at the base of a glacier? Should we be thinking of a till layer that is very soupy, as I have heard mentioned? What is known about the mechanical properties of this material?

D G Croot

Following on from Bernard Hallet's comment, I would like to refer to my own work on glacier surges in Spitsbergen. I hesitate to suggest that surging glaciers are everything we want the glaciologist to model, but I would like to say firstly that most of the Pleistocene sequences which the glacial geologists deal with tend to be in lowland areas, not Arctic or alpine high-altitude cirque glaciers or small valley glaciers. Secondly, a Simple question I would like to put to glacial mechanists or glaciologists is: how do we explain some of the thrust and fold structures observed at the snouts of surging glaciers? These features are very common, and often include sorted, bedded sediments which are totally undisturbed. I have seen "shear zones" 0.3 m wide with beds completely intact within them, with 30 m or so of ice beneath. I cannot envisage differential movement within that shear zone; it would completely destroy the bedding which cuts across the lineation of the shear zone. If I were to go into a Pleistocene environment and find a till sequence containing such a zone of bedded material, it would be very difficult to interpret in terms of the glaciological conditions that formed it if I had not already seen it in a present-day environment. Having seen it, I can perhaps try and interpret a Pleistocene sequence, but very often in a Pleistocene situation, as in the Laurentide ice sheet, we have no model on which to work. Perhaps someone would like to try and explain in glaciological terms how this kind of zonation of debris can occur.

M Seppälä

We glacial geomorphologists would be very grateful for even more detailed information of the glacial erosion processes under the present ice sheets to help us interpret the features found under glaciated surfaces and to reconstruct the conditions under the past continental ice sheets. There are still too many unknowns in the mathematical formulae to make this possible.

Glen

I think some of the problems which are involved here are extremely complicated and this is the reason why there are the unknowns. For example, the whole question of surging glaciers is incomprehensible: they do not form a single class. There seem to be surging glaciers of all different sorts, valley glaciers and wide fronts of ice caps. Sometimes there seem to be a lot in one place, as I think in the Pamirs, sometimes there are just one or two. They have posed very many questions, but I think we must contemplate that there may be many different causes of surging glaciers. If that is so, it is very difficult to use the evidence to interpret Pleistocene deposits. To take the particular question, that Dr Croot asked, of how you get folds: one quite good way of getting folds is to have a layer of rheologically different material and compress it. Reference RambergRamberg (1964) did work compressing samples of differing mechanical properties, and showed that if they differed by more than a certain amount, then when you compressed a planely-layered sample, folds developed. I think Roger Hooke’s remark about debris layers strengthening ice may be sufficient reason why if ice is compressed it should then fold.

On the other hand, there are things which we have suggested which I think are capable of experimental tests, but as far as. I know have not been proven. For example, ever twenty years ago I made some postulations about the way in which stones move in ice as an attempt to understand till fabrics (Reference Glen, Donner and WestGlen and others 1957), and suggested that the longitudinal tendency of stones in till was due to the stones undergoing a rolling action, with the stones spending a much longer time in the sub-horizontal position than in the vertical. This certainly is what you would expect from a continuum mechanics theory of a rigid body embedded in a shearing material. If that shearing material was also being compressed, you would expect the axes to be tilted slightly away from the horizontal, very much like the till fabrics found. I am told we never actually see the stones doing the turn-over. I think this is an interesting question and would love to know whether people have ever found stones moving away from this position and over. It ought to be visible; it is only a small fraction doing it, but a small fraction should be doing it. Similarly the transverse fililximulllvas interpreted in that old paper of mine as being due to collisions between stones. It ought to be more common when the stone density is higher. Again, I think some field measurements on this sort of thing would be helpful. Now I am not suggesting that that is the be-all and end-all of till fabrics. I am just saying that that is an attempt I made to try and solve a problem, and I hoped to see more papers which pulled it to pieces than I have seen in the last twenty years.

D E Lawson

I have measured pebble orientations in glacier ice (Reference LawsonLawson 1979) and found that some pebbles within basal ice do have near-vertical orientations, probably 1 in 200, or 1 in 300. The remainder were very close to horizontal or within 10 to 30° of horizontal.

Drewry

The study of glaciomarine sedimentation has involved several disciplines which have remained quite separate in their interests and methodologies. Sedimentologists have looked at the petrographic characteristics of continental and deep-ocean sediments; glaciologists have concentrated on looking at ice and have ignored the sediments; and the oceanographers have been essentially in the background as far as looking at any interactions between ice or the sediments. It is now time that we brought these three fields together to interpret the very large volume of glaciomarine sediments that occur around the world. Some 10% of the world's oceans today have glaciomarine sediments forming in them. During large-scale ice expansions of the Pleistocene, that area was probably doubled to 50 to 60×l0⁵ km². For the geologist (to go further back in time) the most likely evidence of former glacials will be materials preserved in the oceans. Subaerial weathering and surface geological processes are likely to strip away most of the terrestrial evidence of former glaciations. The items that I would highlight for essential understanding of processes involved in glaciomarine sedimentation are, first, evaluation of the varying roles of ice masses, such as ice shelves, ice streams, and tidewater glaciers. We have little quantitative data in regard to mass flux, velocities, thickness, dynamics, and thermodynarnics in general. Secondly, we have little useful information about the sediment content of any of these ice masses that can be used for modelling purposes. Thirdly, we need to understand the physics of melting and freezing and the release of sediments when ice enters Shallow water, in the case of the grounding line of an ice shelf, or deep water, as in the case of a calved iceberg. Here, surely, is an area where theoretical glaciology can help considerably our understanding. Let us hope that the pursuance of iceberg conferences can lead to practical investigations along these lines. Fourthly, one of the major problems arising from the discussions at this meeting relates to water issuing from beneath ice streams and tidewater glaciers near to the grounding line. Where does this water go? How much sediment is in suspension? What are the density contrasts that govern whether the water moves up to the surface, along the bed, or is inter-stratified? This problem needs examining in a fjord environment, as well as in large embayments such as the Ross and Weddell seas in Antarctica. It is here that the oceanographers can play an important role so that sediment transport by these waters can be fitted into a unified theory. Fifthly, although we can now provide reasonably realistic two-dimensional models, the next step, and Ross Powell's beautiful three-dimensional diagrams must surely point in this direction. is to consider realistic three-dimensional analyses. Finally, we must look at processes through time. How, for instance, do frontal oscillations and mass variations in ice sheets and glaciers affect the production of sediments in an oceanic environment? I would now like to ask people for their suggestions for future research in glaciomarine sedimentology, which, I believe, will have an increasingly significant contribution.

W H Mathews

I would like to respond to your last remark. Having worked in glaciomarine sediments, I find that they can be an Extreme source of frustration. Indeed, I have one student that I can think of particularly who just about chucked in the sponge and went back to some other activity because of the difficulty of distinguishing glaciomarine sediments from till. All I could say to give her some kind of support was "join the club". We need criteria: geochemical. fabric, something of this sort, that help us La go to a site and say with some confidence "this is glaciomarine, that is till". I can only cite as an example one that comes very close to home, the excavation for our geology office building at the University of British Columbia, in which three of us who had some experience with both glaciomarine and glacial sediments examined it, proclaimed it glaciomarine, and then did a fabric analysis and found that it was beautifully developed - as well developed as you would get in any till. Criteria are what we need first of all.

O ORHEIM

There was one point I wanted to make concerning David Drewry's very episodic model of iceberg sedimentation. He showed a slide of an iceberg that had rolled 90° to expose a cliff of debris and from this he suggested that iceberg sedimentation was episodic. I would make the point that even if you turn an iceberg on the side you have got approximately 8/9ths remaining underneath the water, so most of the sediments are going to continue to melt from the sides at much the same rate as from the base. Therefore most of the sediment is going to come out at the same rate as before, unless you turn the iceberg 180° so the sediment lies on the top; then for a little while you will not nave sediment melting out at all.

I think the grounding line is exceedingly important and we can expect, with a high degree of confidence, that practically all the sediments that are not incorporated in the ice will be deposited at the grounding line. There is good evidence to indicate that for an ice sheet most sediments are not in the ice but underneath it. However these sediments are transported, whether by melt water, traction, or whatever, they are going to be deposited near the grounding line. Changes in grounding-line position, e.g. following sea-level changes, will then move sediments with it.

The thermal effects of grounding ice shelves are also important. When an ice shelf goes aground, the bottom temperature will fall below the melting point. For a glacier that was 200 m thick, with little surface melting and mean annual surface temperature of −20°C, the temperature at the base would typically be −12°C. If we suddenly ground an ice shelf by changing sea-level. then the temperature has to change inside the ice shelf and, of course. Bottom freezing occurs incorporating sediment which up to that point had not been frozen. It is almost impossible to calculate the basal freezing rates. People have tried to solve how sediments melt in the Arctic when permafrost is uncovered. That seems almost an intractable problem, and going the other way is even worse. For a selected combination of sediment and (saline) water, and making some assumptions, we found that if it takes a time, t years, to change linearly from 0 to −12°C, the freezing depth in metres is approximately 0.01 t, in other words if the change takes 1000 years. 10 m of permafrost will form. This suggests that sea-level changes of that sort of order and much longer can actually manage to freeze a considerable amount of sediments beneath an ice shelf. The phenomenon is an interesting one and it would be very useful to see the modellers tackle it and solve the heat-flow problem.

Drewry

I agree with Olav Orheim that the grounding-line zone is extremely important. Present knowledge shows that the grounding line of large ice shelves is highly complicated. This means that simple two-dimensional models will be inadequate where there is high differentiation of the ice-sheet margin. For instance, the Ross and the Filchner ice shelves are the world's largest ice shelves. On one side of the Ross Ice Shelf we have a range of mountains with ice discharging from the ice sheet in outlet glaciers. On the other side we have an ice sheet which is grounded principally below sea-level where ice flow is differentiated into a series of ice streams and grounded domes and ridges. The ice streams. are moving at the order of hundreds of metres a⁻¹. Thermal calculations indicate that the bottom of the ice streams are at the pressure-melting point and are sliding. The intervening ice domes and ice ridges are frozen to bedrock, so flow here is extremely small. Thus the ice that is important for sedimentation beneath the floating Ross Ice Shelf is almost entirely contributed by ice streams. The position of the ice streams in the ice shelf is governed by relative geometry and discharge. If these ice streams are crucial to the ice shelf and hence the sedimentation, we are looking at a zone which is comparable to normal sliding, temperate glaciers having sediments at the bed. Therefore differentiation of the grounding line and the varying processes which operate there make modelling difficult, but the available data provide us with clear boundary conditions and focus upon ideas regarding sedimen -water-ice interactions.

Mathews

I would like to comment, Dr Drewry, on your discussion of the ice streams coming off from Marie Byrd Land. On the British Columbia coast and also on the Norwegian coast, we find trenches leading across the continental shelves which may very wen be an expression of Pleistocene ice streams. The big question I want to raise is "is a shelf a necessary adjunct to these or could these be simply streams coming out and terminating in perhaps the deep water at the edge of the continental shelf?"

Drewry

The ice streams that we observe in Antarctica are associated, in most cases, with bedrock channels. Some of them are fairly shallow and have a relief amplitude of only a few tens of metres to maybe 100 to 200 m. Usually these channels are eroded by the ice in soft semi-consolidated sediments. In the case of ice confined by rock outcrops, it is certain that outlet glaciers are producing very deep trenches (such as at Byrd and Beardmore glaciers, etc.). These continue for long distances, even continuing out onto the continental shelf. This is common in the Ross and Weddell sea embayments. Channel formation is thus dependent on how transient the ice streams are and how much time there is available to erode channels. There is an enigma, however. We see channels extending over several hundreds of kilometres which relate to the ice streams, but there are some ice streams that have no bedrock channels:

Orheim

I do not think you meant to say that ice streams were common in the Weddell Sea. In fact, the bathymetric data show only one large over-deepened channel, although people who have postulated surging, or a collapsing west Antarctic marine ice sheet, have published several channels going out from the Weddell Sea. Therefore our concept of grounding in the Weddell Sea gives ice of fairly uniform thickness instead of highly complicated as in the Ross Sea.

Iken

My question refers to Dr Orheim's earlier remarks on the temperature at the bottom of grounded ice shelves. I was surprised by your basal temperature estimate. White Glacier on Axel Heiberg Island fits your example nicely. It is about 200 m thick and the mean annual temperature is about −20°c, but the glacier is sliding over the bed. Now of course the glacier is on a slope, so the analogy is not perfect, but bottom temperature appears to depend on velocity. Could this not also be true for ice shelves, so the grounding line might be at the melting temperature instead of cold?

Orheim

I do not think grounded Antarctic ice shelves could be sliding. The situation is quite different. There is practically no surface melting and a surface mass balance of 0.5 to 0.6 m a⁻¹ brings the cold surface temperature deeper into the ice.

Could I make the general comment that in our attempt to understand glaciomarine sediments in the Antarctic and parts of the Arctic, we are looking at exceedingly small pin-points on the ocean bottom? We have been fortunate in the Norwegian expeditions to have obtained much, seismic evidence together with cores, so we are beginning to reconstruct some kind of regional picture. In the Arctic there are certain areas which have been very heavily studied, but even the most studied areas have nothing like the evidence available above water. I think we have to be exceedingly careful in claiming that, based on a few cores here and there, we know the story. We all recognize this, but it does not hurt to say it again.

V Kotlyakov

I would like to thank the organizers of this symposium for selecting very interesting and important topics which bring together glaciologists and geologists. I am a glaciologist and I will say a few words from the point of view of present-day glacier investigations. Many of the papers delivered considered new concepts and approaches to the studies of glacier and glaciomarine deposits. The studies of glaciomarine sediments, which incorporate the long history of glaciers, although not yet finally clarified, are of the greatest significance. In particular, there are many data obtained in the western hemisphere and in our country testifying to the existence of vast ice sheets over areas that are now water-covered. In our studies, we proceed from the assumption that a number of methods should be applied simultaneously for the study of glacial erosion and sedimentation. This enables us to conduct some comparatively accurate analyses. The results of such studies are presented in the paper by Serebryanny and Orlov, which unfortunately was not delivered here as the authors are now working on an expedition in Spitsbergen. The balance method occupies the basic position in the investigation of sedimentation and debris transport. For some glaciers of the Soviet Union we have calculated all the components of debris input and output with reasonable accuracy. This is most important for the Central Asian glaciers where, as you have seen from the slides of Dr Chizhov, many glaciers have a very thick debris cover. In this respect, special attention should be paid to surging glaciers in whose regime the debris balance seems to play an important role. One of our theories states that overloading of glaciers by debris may work as the triggering mechanism for surges. From direct surveys, we know now that many end moraines of the Central Asian glaciers, which were recently believed to have formed thousands of years ago, were generated by big surges only a few decades ago. The problems of glacier surges are very important for glacial geology, and that is one of the reasons for their being included in the programme of Soviet astronauts' studies at the Salut 6 orbital station.

Finally, remote radio echo-sounding of glaciers has recently become a very efficient method for studying glacier structure and geological activity, as can be seen from several papers here. We now pay much attention to the development of radio echo-sounding equipment in the same way as does the Scott Polar Research Institute. In the near future we hope to have equipment which will allow us to echo-sound temperate glaciers of various thicknesses and to get reliable data on their internal structure. That is very important for this point of view.

Glen

There is one thing that impressed me very much in looking at Dr Chiznov's photographs, and that was the way in which his glaciers were not flowing over the full width of the valley. They were flowing through broad moraine walls and outside the moraines there was a further valley with streams flowing along it. My memory is that this is also what I have seen for other Central Asian glaciers on the south side, from Pakistan and India. I wonder if this is itself, perhaps, a feature which is peculiar to surging glaciers? I have no experience on which to say this, and wonder if either Dr Chizhov or somebody else could comment on it because it could be a feature which, if found in relic form, might help us to tell if we were looking at the relics of a surge.

O P Chizhov

I cannot answer this question. The valley is very large and was not made by previous ice-age glaciers, but is tectonic. The glacier lies in the bottom of that great valley and the lateral moraines were made by the glacier when it surged and the ice surface bulged. Before the surge, the glacier surface in the lower zone is down-wasted with large elevated lateral moraines. It gives the appearance of the glacier not filling the valley. But during the surge the glacier flows over the greater part of the valley width, and the appearance of the surface is entirely different. I do not know the exact cause.

Hallet

Dr Glen's question is a very intriguing one because some glaciers that are clearly non-surging show this pattern. In fact during his presentation, Dr Small showed a beautiful picture of Glacier de Tsidjiore Nouve with the moraines clearly encasing the glacier, some distance from the valley wall.

Croot

The surging glaciers in Spitsbergen which I have examined occupy the full valley width when they surge. However, with a year or two of ablation, because of the distribution of debris-rich and debris-poor ice, one attains a profile which would appear from the air to give the valley-within-valley situation described by Dr Glen and Dr Chizhov. Valley-side streams coming down towards the lateral margins produce erosional valleys, by incising themselves between the valley side and the lateral debris-rich ice margin giving exactly the situation described. although the original surge did fill the valley.

D M Mickelson

I have benefited a great deal from the discussions we have had this week. and I think both glacial geologists and glaciologists here have profited from this interchange. One concern that I have, and something that I think glacial geologists and glaciologists should keep in mind, is the temporal aspect of the processes we have been considering. For instance, if we glacial geologists look at a stratigraphic section anywhere back from the terminal moraine, we are certainly looking at a point in the landscape which has undergone a number of processes during time. In fact the evidence that we have may not be a complete record of the processes that have been on-going. I had a feeling a number of times this week that most of the discussion was revolving around an ice mass that was out at its maximum position and people were trying to look at what was happening in certain places. We really have to keep in mind that the zones are moving across the landscape, and it is only by recognizing that and keeping in mind this temporal aspect, that we will be able to understand from the glacial geology what was happening in the ice mass itself.

I have an intuitive feeling that, for the most part, in marginal zones, sometimes only 100 in wide and sometimes, in the case of ice sheets, 10 km or more wide, you have primary deposition taking place. We can look at some places well back from the margin and see evidence that we have a period of time when deposition took place, a period of time when erosion took place, and then deposition again as the ice margin retreated

One question that arose in discussion was whether in marginal areas it is possible to preserve very delicate things like organic deposits if the toe of the ice was not frozen. It has been suggested a number of times by glacial geologists that it is necessary to have the glacier frozen to its bed to preserve these kinds of deposits. I would suggest that it is not necessary, and that in marginal areas, whether or not the bed is frozen, you are quite likely to preserve pre-existing organic materials. I would cite the Two Creeks Forest in Wisconsin as a case in point, where we nave over a dozen localities, all within a few kilometres of the outer front, and where, because of the [evidence of] surrounding vegetation, beetle remains, and a variety of other things, we must have had warm-based ice.

Kemmis

It seems to me that many of the problems discussed here arise from the incomplete and inadequate questions asked and the lack of perspective given to the answers. The shear zone-debris band problem discussed all week is a case in point. A very simple question is asked: "does shear take place in these zones?" as if that is the only factor involved. People then answer this question with a simple yes or no based on field observations or laboratory calculations. But the problem is greater than simply "does shear occur?" There must be a host of factors affecting whether or not shear may take place in the ice, factors such as thickness of the debris band, spacing of the bands (if important), particle size and concentration in the debris band, ice dynamics (i.e. velocity, thermal regime variations at the bed, ice thickness, etc.), and so on. Until each of us includes information on these factors with our yes or no, that is, until we provide a perspective for our answer, we will never be able to satisfactorily answer such questions as: "does shear take place in these zones, and if so, what factors are important to its occurrence and where is such shear likely to occur?"

The same inadequacies in outlining a problem apply to the study of glacier beds comprised of sediment. A question was asked earlier, should we model till beds as "muddy"? But till beds are much more complicated than that: Tills may vary greatly in particle size and distribution from the very cobbly tills at HardangorjØkulen to the matrix-dominated tills of the Laurentide ice sheet in midwestern U.S.A. So particle size is an important factor to consider in modelling. Moisture content and consolidation characteristics (in the geotechnical sense) are additional factors channels to consider. Until our questions include analysis] of all the important factors it will be impossible to arrive at meaningful, comprehensive answers.

A final example of not considering all the relevant factors is that of the preservation of organic deposits and buried soils, just mentioned by Dave Mickelson. Organic deposits or buried soils may indeed be "soft" when subjected to a triaxial test. And one way to preserve such material might be to freeze it beneath the "cold" snout of a glacier, thus increasing its strength. But is shear strength the only variable which is important? Organic soils commonly have low permeabilities and virtually no discontinuities such as joints. They therefore are probably very difficult to pluck and may persist as easily beneath a thin film of melt water under a temperate glacier as under a frozen toe. So, until we ask questions that encompass all of the factors relevant to a problem, and until we put our answers into the perspective of the system from which our answers came, it will be very difficult for us to understand the spectrum of conditions over which different glacial processes may occur.

J Ehlers

I should like to ask a few questions which might be answered one way or the other by the glaciologists present here. One of the main questions seems to be: how did the inland ice manage to get so rapidly from the centres of the glaciation to the marginal areas, for instance from Scandinavia to north Germany? If we estimate the time the inland ice had to travel this distance, we see that it must have proceeded with an average speed of roughly 50 to 100 m a⁻¹. Allowing for halts, it may have advanced at an even faster speed. During its advance many interesting things happened and some of these created the phenomena Wickham and I described this morning. The ice dynamics are by no means solved yet. Did the processes which were envisioned by Boulton play an important role? While I think they offer a very interesting solution to the speed question, the field evidence makes it unlikely that they played such an important role as in Iceland where he made his observations (Reference BoultonBoulton 1979).

Another important question may be: what happened to the melt water? In some cases it was confined to tunnels under the ice and led to the creation of tunnel valleys. Mickelson told me that in the Laurentide glaciation this process took place during the whole Pleistocene whereas in our northern European glaciation it was restricted to the Elsterian, our oldest has offered a possible solution to that problem by suggesting that during the later glaciations we had permafrost in the region, and thus no such in-cutting could occur. That point is worth discussing: during the Saalian and Weichselian (Illinoian and wisconsinan) glaciations the process of sheet-like accumulation of melt-water sands prevailed, while the incision of channels or tunnel valleys took place on a much smaller scale or was completely missing. Without permafrost it would be hard to solve the question of how this sheet-like sand accumulation could take place. In permeable deposits, as they occur in northern Germany, water either infiltrates into the ground or collects in streams. Run-off which is not concentrated in streams only happens on impermeable ground. Such were the conditions when permafrost prevailed, and consequently the sheet-like sand deposits could be formed. These are some of the questions which still remain for me.

Hallet

I think some of those questions are going to remain for all of us, but it seems to me that you have really pointed to the kind of information useful to theoreticians in terms of modelling the subglacial system. If you have tunnel valleys and if much of the subglacial drainage is actually concentrated in these valleys then it is quite a different problem from considering subglacial drainage by a uniform flow of water through a porous till. If we are trying to understand the subglacial water pressure, and especially the pore pressure in a till, which figures in a very important way in determining whether or not till will deform, it is quite critical that we be able to say something meaningful about the flow of water. So the tunnel valleys are beautiful evidence that you are dealing with channellized water very much as we would expect on a completely different scale from some of the work on valley glaciers.

One thing that I found very interesting about your map of tunnel valleys is that if you took that map and reduced it by two orders magnitude, you would get almost exactly the same type of map that we have compiled for bedrock areas in front of a present-day glacier. The drainage networks appear irregular with non-arborescent and non-converging channels.

Glen

I was very impressed by Dr Ehlers figure of 100 m a⁻¹ and wonder if he could say whether this is a firm figure and also whether it applies to both kinds of advance, the one with channels and the one without.

Ehlers

It refers to the advance without big channels and is an estimate, not a firm figure. I have taken into account the Weichselian (Wisconsinan) interstadial deposits in Sweden and Norway, and if you do that you see that the ice had to advance to northern Germany and to retreat again with that very high speed.

Glen

I think it is important to get a feel for these figures because they give people who are attempting to model past ice sheets something to go on, just as the models can feed back and tell people what to look for. If it is something like 100 m a⁻¹ advance, then we are talking about something which is much more like a surging-glacier advance than anything else we look at today. Perhaps some of these questions about what surging glaciers do become more relevant. Of course, it is not a valley-glacier advance we are talking about, but on a broad front. Some of the features which are most distinctive of current surging glaciers are, I suspect, due to the fact that they are valley glaciers. But it does give added impetus to looking at what glaciers which are advancing at speeds of the order of 100 m a⁻¹ are doing.

Goldthwait

There is one answer to John Glen that he may not be aware of. We are possibly in the most fortunate position in Ohio to determine how fast the last Wisconsin glacier which reached Ohio was moving. We have a combination of two things; wood just beneath the drift, still more or less in place, and wood in the lowermost till of the Wisconsin in time. I have had this wood dated at over 50 places and calculated the net rate of advance between half a dozen points down each of two lobes. The rates vary between something on the order of 20 m a⁻¹ and a little in excess of 100 m a⁻¹. The lowest figure is for the outermost portion of the advance, when it got almost to Cincinnati (Reference GoldthwaitGoldthwait 1958).The controls are figures from the north side of lake Erie which come largely from the area of Dreimanis' work, from Cleveland, from intermediate central Ohio, and at the terminal position. Depending on the lobe, the terminus was reached at around 21 500 BP for the outermost moraine.

Mathews

Dr Goldthwait, you are not the only people who have wood. What I will come out with is almost a repeat of what you said. In tile Vancouver area, we have evidence that tile ice was still somewhere to the north of Vancouver about 18000 BP. It moved southward to the southern end of the Puget Sound lobe, a distance of 350 km, by sometime between 15000 and 14 000 BP. In other words, the average velocity was just about 100 m a⁻¹. If you want figures, John, there you have it. Like Dr Goldthwait's, that is a pretty firm number.

Hooke

The hypothesis of instantaneous glacierzation, presented by Ives and 'Andrews, is perhaps pertinent to this discussion. Snow-banks and snow-patches in hollows gradually expand because summer melt is insufficient to remove all the snow in such hollows. Thus instead of ice from the northern side of Lake Erie, in Dick Goldthwait's situation, actually having to advance across the lake and move all the way down through Ohio, ice could have been formed in Ohio and in the Lake Erie basin by accumulation in pro-glacial snow banks, and the glacier grows in place.

Glen

Surely they did not claim it at the southern margin? They claimed it for the middle, at the beginning.

Hooke

That is correct. They claimed it for the central part. But a similar process contributed to the advance of temperate valley glaciers under the climatic conditions found in northern Scandinavia in the early 1900s (W Karlén, personal communication).

Another very brief comment: moraines formed in localities where ice did not expand when it came out of narrow valleys, but instead continued to go straight, are quite common along the east face of the Sierra Nevada, in California. I do not think anybody has proposed that any of those glaciers were surging glaciers, but this is a fairly common moraine form.

Several people have made comments about the interchange that does not take place between modelers and glacial geologists. The modellers were accused of ignoring the glacial evidence. I plead guilty to some of that, but at the same time I think that it is incumbent upon the glacial geologist to make a real effort to understand glaciological theory. Not necessarily so they can apply it, but so they know the important parameters to measure, and what critical observations could be made to test the theories. You do not have to be a theoretician or modeller to look at the model, see what the assumptions are, and try to develop experiments that can test some of these models. The interchange obviously has to take place in both directions.

Shaw

For a moment we were back to talking about the stability of large ice sheets and surges, and got away from historical evidence of glacial erosion and sedimentation. Maybe we should stick to that for a moment. This morning Dr Glen said in response to Mr Dardis lecture on drumlins that he did not believe in the convection theory and that it was inappropriate to apply to landforms. Dr Glen referred us back to geomorphologically-based and geologically-based theories which some geomorphologists find wanting, We are just throwing the baton back-wards and forwards. Could Dr Glen tell us how thinks we might cooperate to solve such problems as the origin of drumlins? As far as I know, no theoretician has modelled that problem in detail?

Glen

My reason for saying what I said this morning was precisely the point which Roger Hooke was making a moment ago, that we need to make sure that the postulations made by glacial geologists are physically reasonable. I do not believe it is physically reasonable to have convection, by which I mean a flow of ice driven in a vertical cell by the differential temperature and therefore different buoyancy of ice, on a scale of a drumlin field. I am not sure it is reasonable anywhere, but I am prepared to argue it about Antarctica. I am not prepared to argue it about the drumlin field, because I am absolutely sure that if you work out the Rayleigh number you are going to find that convection is quite out of the question. The viscosity of ice is far too high for the very small temperature differences to be able to drive something on this small scale. My purpose was to say I do not think we understand sufficient about the other possible theories of drumlins to rule them out of court and do a Sherlock Holmes argument, which I think is what you are talking about. "When you have considered all the other possibilities, the one remaining, no matter how improbable, must be true." I do not think we have reached that point. I do think we want to get information about what a drumlin is. Is it a uniform body? Is it something which always has a similar structure inside? Or is it only something which has a similar shape or form, but in different places has very different internal structures? I rely on the geologists to tell me the answer to those questions. We can only explain drumlins if we know what exactly it is we are explaining. If they do consist entirely of till material, then certain theories are possible. If they do not, or if inside them there are sometimes complicated structures, then perhaps we should say that the similarity between things with differing structures shows that the form has been developed by the glaciers on various other different structural situations. Therefore, as some people have suggested, they may be an erosional form, not a depositional form, an erosion of deposits. I am not an expert on drumlin theory, but I know some of these ideas are going around. Some people say it is due to the rheology of the till material when it has water in it. Some people say it is due to certain areas of the bed being frozen and other areas unfrozen. Some say it is due to deposition around some kind of knob, be it a rock knob, a previous till knob, or even a bit of frozen ground. These seem to be the things we can postulate about. I am not a drumlin theorist, but those are the questions we should be debating, not what I believe to be physically impossible solutions.

Shaw

Let me, as a dull Dr Watson in the presence of the rather eloquent Sherlock Holmes, try to express my problems. Roger Hooke is saying that the glacial geologist or the glacial geomorphologist has to be able to handle the mechanical theory. To establish that rather complicated flows are required to form drumlins took four years of intensive geological work. To ask us at the same time to handle the mechanics in an original fashion is too much. In fact when I spoke to people about this problem here they said, "Ah yes, that is a very difficult problem". Now I wish that someone who feels that this is a difficult problem would take it on.

Orheim

My comment goes back to convection theory and also to those glacial geologists who have liked to invoke the concept of surging to explain anything they cannot explain by any other means. I would make a plea that glacial geologists and glaciologists, when trying to explain phenomena, should never invoke a physical process they cannot demonstrate happening today. We know, for example, some glaciers are surging; we know nothing about surging of large ice sheets. Yet that is a very heavily discussed topic. There are other processes where one goes into the esoteric, along the lines of what John Glen was saying. Any model that cannot be tested in the field should not be used.

Orheim

I was not trying not to be provocative. I do think though that it is very important that you try to test models and, as Roger Hooke was saying, a model should clarify your thinking, you do not have to understand the model perfectly, but a good model focuses your thinking and improves field experiments. For example, we have discussed bed roughness for 20 years, since Hans Weertman's paper on sliding, yet exceedingly few glacial geologists or glaciologists have gone out and measured it. There are many obvious field tests to check whether models are applicable. The mathematics for the modeller gets more intractable as you get field evidence, but that is not the fault of the glacial geologist. He can turn the problem back to the modeller.

D A Fisher

I want to point something out that comes from ice-core work whenever ice cores have penetrated the layer that was dropped during the last glacial period, the Wisconsin or Würm. There are some unique characteristics of this ice that suggest that the type of ice you are dealing with as an ice sheet might not be quite the same stuff you are looking at today on the surface. The unique characteristics are that Wisconsin ice is very dirty in microparticles, 500 to 600 parts per billion by weight, and the ice crystals are an order of magnitude smaller than the ice crystals in the Holocene or in the pre-Wisconsin. The third characteristic which seems to be unique is that the Wisconsin ice is rheologically softer. There are measurements from the Barnes Ice Cap on Baffin Island, the Devon Island ice cap, and tile Agassiz ice cap on Ellesmere Island which show this. On the Barnes Ice Cap, where the drill hole goes through Wisconsin ice it bends more quickly. In the case of Devon Island and Agassiz, where we measured bore-hole closures, one can pick out the ice due to the Wisconsin glaciation because the bore hole closes much faster; this means that in various forms of creep the ice is weaker. Therefore the modellers may want to consider a different rheology, other than just due to temperature. for old ice.

Lawson

We have been discussing the difference between the way glaciologists and glacial geologists are looking at problems. My training is more as a sedimentologist, and I see the problem as something that spans the two. Really what we are looking at is a complex sedimentary environment. For the glacial geologists to interpret their sediments, they need criteria based upon the characteristics of debris fanning in active glacial environments, with their relationship to debris properties and the glacier's mechanisms of entrainment, transport, and deposition well defined. Investigators of active glaciers, as well as theoreticians, must present their results in these terms so that glacial geomorphologists can go from the sediments as they find them and interpret, as fully as possible, the glacialogical and sedimentological mechanisms that formed them.

Hallet

Considering the range of interest in glacial erosion and deposition, which spans from very idealized theoretical glacial mechanics to very detailed unravelling of complicated glacio-stratigraphic relations, it is not surprising that, at times, communications may be difficult between scientifically distant researchers in this field. This conference, however, has gone a long way in the direction of sharing central ideas, delineating key problems that remain to be solved, and hopefully catalyzing collaborative work in research on glacial erosion and deposition. Thank you very much for your participation in the discussion.

L.W Gold

We have now come to the end of a successful symposium and five full days of interesting contributions and discussions. I would like to take this opportunity to thank on your behalf the individuals that made this possible. First, the Papers Committee under Garry Clarke. This includes David Drewry, Dick Goldthwait, Bernard Hallet, Hans Röthlisberger, and Johan Ludvig Sollid. To them I express our appreciation for putting together an excellent programme. They also are working hard as Scientific Editors, along with Ailsa Macqueen, the House Editor, in putting together the proceedings, of this meeting. Next, I want to express our thanks to those individuals that made this all possible: the local arrangements committee under the skilful direction of Olav Orheim; the individuals responsible for the field trips and post-symposium excursions - Inge Aarset, Hans Holtedahl, Olav Kjeldsen, Olav Liest∅l, Gunnar ∅strem, Johan Ludvig Sollid, Leif S∅rbel; the group that have worked quietly and efficiently running the projector and looking after the lights - Kjell Kjenstad, Noritz R∅yr. Pál Strandvik; the people that looked after our various needs such as travel arrangements, photocopying, answering our questions - Beverley Baker from our Cambridge office and Annemor Brekke from the Norsk Polarinstitutt. And behind all of them, of course, has been Hilda and to her I give our special thanks. I wish also to recognize and express our appreciation to the Norsk Polarinstitutt and the University of Oslo for joining with the Society in sponsoring and running this symposium.

Finally, I want to thank all of you for your participation, particularly the authors for their clear presentations within the allotted time so that there was a good opportunity for discussion. There was a real meeting of minds during those discussions and I know they are going to continue this evening, during the post-symposium tours, and whenever there is an opportunity to meet in the future.