Some of the ideas discussed in this blog are published in my new book called "The Stonehenge Bluestones" -- available by post and through good bookshops everywhere. Bad bookshops might not have it....
To order, click

Wednesday, 17 July 2019

Liverpool Land piedmont glacier apron

There is evidence in eastern Jameson Land of till having been laid down by westward-flowing glaciers.  These came from the Liverpool Land mountains and coalesced into a broad apron filling a wide transverse trough.  When the ice in the apron thickened sufficiently, flow was diverted northwards and southwards.

We have talked a lot about piedmont glaciers on this blog, in relation to glaciers decanting from troughs or fjords out onto adjacent lowlands — but not much about piedmont APRONS.  These are rather interesting, since they consist of  coalescing piedmont glaciers which spread laterally.

The Liverpool Land (East Greenland) example is an interesting one, and a number of authors have referred to the wide apron of glacier ice that filled the deep through valley connecting Hurry Fjord in the south to Carlsberg Fjord in the north.  This happened in the last glacial maximum (LGM) and probably in all preceding glaciations as well.  The glaciers flowing westwards came from the alpine mountain range of Liverpool Land.  On the east side of the mountains the glaciers coalesced on the open coast, and there may well have been a calving ice shelf.

According to the reconstructions, the coalescing piedmont glaciers filled the transverse trough and then flowed northwards and southwards towards the coast.  This must have happened in multiple locations in the arctic — and we can see examples today from Greenland, Arctic Canada and Arctic Russia.  We have referred in an earler post to Putorana:

Putorana — the mountain front runs across the photo, and the morainic loops show how the individual glaciers coalesced into an extensive ”glacier apron”.

Tuesday, 16 July 2019

Salisbury Plain LIDAR imagery

This is very impressive and informative.  Courtesy Mark Walters.  You need a good computer with a touch screen to get the best out of it.......

Will try to embed:

Friday, 12 July 2019

More on the Scoresby Sound piedmont glacier

In this interpretation it is argued that the whole of Jameson Land was deeply submerged beneath ice from the Greenland ice Sheet during the Salian Glaciation.  Glacial deposits referred to the "older drift" glaciation are dated to this phase, as are some of the glacial deposits found in the "driftless" area.  Other deposits in that area are assumed to be even older -- and are largely destroyed.  The "newer drift" deposits are dated to the LGM or Late Weichselian: these are closer to the coast, and suggest that the ice from the piedmont Glacier pressed inland up to altitudes around 250m, buttressed by cold and non-erosive ice formed on the Jameson Land uplands.  This ice cap covered and protected older deposits.  Other active glaciers flowed westwards, down from the accumulation areas on the mountains of Liverpool Land, maybe feeding a south-flowing glacier in Hurry Fjord..

This is an important paper from which I have quoted before -- check things out by doing a search for "Jameson Land."

Hakansson, L., Alexanderson, H., Hjort, C., Moller, P., Briner, J. P., Aldahan, A. & Possnert, G.: 2008.  Late Pleistocene ¨ glacial history of Jameson Land, central East Greenland, derived from cosmogenic 10Be and 26Al exposure dating. Boreas, Vol. 38, pp. 244–260.
10.1111/j.1502-3885.2008.00064.x. ISSN 0300–9483.

Previous work has presented contrasting views of the last glaciation on Jameson Land, central East Greenland, and still there is debate about whether the area was: (i) ice-free, (ii) covered with a local non-erosive ice cap(s), or (iii) overridden by the Greenland Ice Sheet during the Last Glacial Maximum (LGM). Here, we use cosmogenic exposure ages from erratics to reconcile these contrasting views. A total of 43 erratics resting on weathered sandstone and on sediment-covered surfaces were sampled from four areas on interior Jameson Land; they give 10Be ages between 10.9 and 269.1 kyr. Eight erratics on weathered sandstone and till-covered surfaces cluster around 70 kyr, whereas 10Be ages from erratics on glaciofluvial landforms are substantially younger and range between 10.9 and 47.2 kyr. Deflation is thought to be an important process on the sediment-covered surfaces and the youngest exposure ages are suggested to result from exhumation. The older (470 kyr) samples have discordant 26Al and 10Be data and are interpreted to have been deposited by the Greenland Ice Sheet several glacial cycles ago. The younger exposure ages ( 70 kyr) are interpreted to represent deposition by the ice sheet during the Late Saalian and by an advance from the local Liverpool Land ice cap in the Early Weichselian. The exposure ages younger than Saalian are explained by periods of shielding by non-erosive ice during the Weichselian glaciation. Our work supports previous studies in that the Saalian Ice Sheet advance was the last to deposit thick sediment sequences and western erratics on interior Jameson Land. However, instead of Jameson Land being ice-free throughout the Weichselian, we document that local ice with limited erosion potential covered and shielded large areas for substantial periods of the last glacial cycle.


Then there is this -- the article itself is behind a paywall, but we get the gist of it from the abstract:

LARS RONNERT and MATS R. NYBORG, 1994.  The distribution of different glacial landscapes on southern Jameson Land, East Greenland, according to Landsat Thematic Mapper Data.
Boreas, Volume 23, Issue 4
Pages: 281-536
December 1994


Four geologic units previously mapped in southern Jameson Land. East Greenland (Funder 1978, 1990) are identifiable on a false colour composite of Landsat Thematic Mapper (TM) spectral hands TM5. TM4 and TM1. The area covered by the Weichselian glaciations has a fresh glacial morphology and a less developed drainage system than the older landscape. The Weichselian glaciers reached more than 200 m a.s.l. in the west. but only about 100 m a.s.l. in the east. A contextual analysis (local frequency and local orientation) was included in a Maximum Likelihood classification (M‐L) to map the extent of the Weichselian glaciations. Deposits correlated with the Saalian Scoreshy Sund glaciation are found on the central plateaux of Jameson Land. Landsat TM geological mapping of the surficial distribution of deposits from the Scoreshy Sund glaciation and of weathered Jurassic sandstone or deposits with a high percentage of such sandstone was done using a supervised Maximum Likelihood procedure. Except for the mapping of thc extent of the Late Weichselian Flakkerhuk glaciation, the Maximum Likelihood boundaries between units are in general agrecnient with earlier mapping or with the visual interpretation of the false colour composite. A strong vegetational influence. and similar spectral reflectance lrom deposits of different age due to similarities in lithological composition reduced the possibility of an independent remote sensing approach. Taking already existing general geological knowledge and chronology into account allowed successful Landsat TM geological mapping.

The Scoresby Sund piedmont glacier

The LGM maximum (Flakkerhuk) stage is shown by the location of the Kap Brewster moraine complex.  When the ice front was at that position Scoresby Sund and Jameson Land must have been covered by a very large piedmont lobe.  Evidence suggests that the ice was grounded all the way out to the Kap Brewster moraine. The Milne Land stage, around 10,000 years ago, represents a readvance which did not progress much beyond the trough exits.  Note from the text below that some researchers believe that Jameson Land was covered at the time of the LGM and others think it was ice-free!

On pondering about the nature of the Celtic Sea piedmont glacier, I am focussing more and more on Nordvestfjord, Scoresbysund and Jameson Land.  The parallels are striking.  In Nordvestfjord, a very large outlet glacier / ice stream like the Irish Sea Glacier, inputs from the flanks like those that came in from the Welsh ice cap and the Irish ice cap, a sudden deceleration / decanting of ice (Hall Bredning and St George's Channel), further inputs from the west (glacier ice from the Renland and Milne Land glaciers and in the case of the Celtic Sea piedmont, from southern Ireland) and finally a spreading of the ice southwards and eastwards over gentler terrain (Scoresbysund and Jameson Land in the case of Greenland and the Celtic Sea floor and the Bristol Channel floor -in the British case -- both probably above relative sea-level at the time -- so there would have been no ice shelves) -- except maybe off the set coast.

The Scoresby Sound piedmont was apparently grounded -- the water depth in the sound today is between 400m and 600m (much shallower water than in the feeder fjords).  Global sea level at the time was probably more than 100m lower than today.  But there were much greater isostatic depressions and rebounds here than in the Celtic Sea -- the implications need careful thought......

Running diagonally across this satellite image -- Nordvestfjord. the biggest of the outlet glacier routes.  Hall Breeding is speckled with icebergs.

Nordvestfjord landscape -- near the entrance of Gurreholmsdal. The trough walls are in places almost 2,000 m high -- and the fjord water is in places more than 1,500 m deep.  So the trough depth is c 3,500m or about 11,000 feet.

Fjord walls typical of the fjord system, near the Bear Islands.  Renland in the background  -- a "vertical landscape".....

Jameson Land -- a rolling lowland with ancient valleys and low hill masses.  For many years it was thought that the landscape bore no traces of glaciation.....  

Two images of the Jameson Land tundra landscape -- by contrast with the fjord landscape, this is dominated by wide horizons and "horizontal" features.


Gabriele Uenzelmann-Neben (1992) Scoresby Sund, East Greenland: Structure and Distribution of Sedimentary Rocks.
Polarforschung 62 (1): 1-9, 1992 (erschienen 1993)

In Scoresby Sund, the Pleistocene was a time characterized by erosion. Thicker deposits can be found only at the valley mouths, high mountain plateaus and the coast of Jameson Land (FUNDER 1990). The occurrence pattern of Quaternary sediments documents a number of glacial-interglacial cycles. The oldest sediments (Lodin Elv formation) found on land are of Pliocene/Pleistocene age. The most expanded glaciation, the Scoresby Sund Glacial, occurred around 200 ka (FUNDER 1972, 1984, 1989) and represents an important stratigraphic marker (HJORT & MÖLLER 1991; HJORT & SALVIGSEN 1991). Sediments deposited during the Langelandselv Interglacial (Eemian 120 ka, FUNDER 1990) can be found only along the southwestern coast of Jameson Land. This interglacial showed higher temperatures than the present and can be correlated with isotopic stage Se (BÖCHER & BENNICKE 1991). The first two Early Weichselian glacials (Glaciation 1 and 2 during the Jameson Land Marine Episode) showed neither an ice cover of the outer fjord areas (FUNDER 1989) nor of Jameson Land (FUNDER et al. 1991). During the Flakkerhuk Glacial (late Early Weichselian, FUNDER et al. 1991) a thick floating glacier extended out onto the shelf while Jameson Land stayed uncovered (FUNDER 1989, 1990). Glaciation 4 (Late Weichselian, FUND ER et al. 1991) commenced about 20 to 13 ka. Until present, it has been considered to be a minor glacial stage. Thick glaciers terminated at the mouths of the western fjords into Hall Bredning and Scoresby Sund during the Milne Land Stage (11 - 9.5 ka, HJORT 1979). After a continuous retreat of the glaciers astandstill followed during the Rodefjord Stage (FUNDER 1971). The end of that phase resembles the present ice extent.

Some refs

Dowdeswell, JA, Villinger, H., Whittington, R.J. & Marienfeld, P. (1991): The Quaternary marine record in the Scoresby Sund fjord system, East Greenland: preliminary results and interpretation. In: P. Möller.; Ch. Hjort & O. Ingolfsson, eds., The last interglacial-glacial cycle: preliminary report on the PONAM fieldwork in Jameson Land and Scoresby Sund, East Greenland. Lundqua Reports 33: 149-156, Lund.

Dowdeswell, JA, Uenzelmann-Neben, G. & Whittington, R.J. (submitted): The Late Quaternary sedimentary record in Scoresby Sund, East Greenland. - Boreas.

Funder. S. (1971): Observations on the Quaternary Geology of the Rodefjord Region, Scoresby Sund. - Rapport Grenlands Geologiske Undersogelse 37, Copenhagen.

Funder. S. (1972): Remarks on the Quaternary geology of Jameson Land and adjacent areas, Scoresby Sund, East Greenland. - Rapports Gronlands Geologiske Undersogelse 48: 93-98, Copenhagen.

Funder. S. (1984): Chronology of the last interglacial/glacial cycle in Greenland: First approximation. - In: W.c. Mahaney (ed.), Correlation of Quaternary chronologies, 261-279, GeoBooks, Norwich,

Funder. S. (1989): Quaternary Geology of the ice-free areas and adjacent shelves of Greenland. - In: J.R. Fulton (ed.), Quaternary Geology of Canada and Greenland. Geological Survey of Canada, Geology of Canada No I.

Funder. S. (1990): Quaternary Map of Greenland, 1:500000, Sheet 12, Descriptive Text. - Geological Survey of Greenland, Copenhagen.

Funder. S.; Hjort, CH & Landvik, IY. (1991): Quaternary stratigraphy of Jameson Land - a first approximation. - In: P. Möller; Ch. Hjort & O. Ingolfsson eds., The last interglacial-glacial cycle: preliminary report on the PONAM fieldwork in Jameson Land and Scoresby Sund, East Greenland. Lundqua Reports 33: 171-176, Lund.

GEBCO (1980): General bathymetric chart of the oceans. - Chart No. 5.04, Canadian Hydrographie Service, Ottawa.

Hinz, K., Mutter, J.C., Zehnder, C.M. & NGT Study Group (1987): Symmetrie conjugation of continent-ocean boundary structures along the Norwegian and East Greenland Margins. Mar. Petr. Geology 4: 166-187.

Hjort, Ch. (1979): Glaciation in northern East Greenland during Late Weichselian and Early Flandrian. - Boreas 3: 281-296.

Hjort, Ch. & Möller P. (1991): Glacial deposits on the interior plateaux of Jameson Land, East Greenland. - In: P. Möller; Ch. Hjort & O. Ingolfsson (eds.), The last interglacial-glacial cycle: preliminary report on the PONAM fieldwork in Jameson Land and Scoresby Sund, East Greenland. Lundqua Reports 33: 7-22, Lund.

Hjort, Ch. & Salvigsen, O. (1991): The channel & tor-landscape in southeastern Jameson Land, East Greenland. - In: P. Möller; Ch.

Hjort & O. Ingolfsson (eds.), The last interglacial-glacial cycle: preliminary report on the PONAM fieldwork in Jameson Land and Scoresby Sund, East Greenland. Lundqua Reports 33: 23-26, Lund.


Another interesting paper:

Funder, S. & Hansen, L. (1996): The Greenland ice sheet - a model for its culmination and decay during and after the last glacial maximum. Bulletin of the Geological Society of Denmark, Vol. 42, pp. 137-152. Copenhagen, 1996-02-01.


Onshore studies at the mouth of Scoresby Sund have shown that the large grounded outlet glacier which filled the fjord basin during LGM, the Flakkerhuk stade, was thin and less than 400 m thick at the fjord mouth (Mangerud & Funder 1994; Tveranger, HoumarkNielsen, Løvberg & Mangerud 1994). It probably had its front on the "Kap Brewster sedimentary ridge", a 20 km wide, 175 m high, and more than 30 km long ridge of Quaternary sediments which has been located from air gun and bathymétrie data at the fjormouth (Fig. 4 and Dowdeswell, Uenzelmann-Neben, Whittington & Marienfeld 1994). This is in agreement with ice core studies on a local ice cap which showed that the Inland Ice during LGM was drained through the deep fjord troughs and never invaded the adjacent mountain plateaus at c. 2000 m a.s.l. (Johnsen, Clausen, Dansgaard, Gundestrup, Hansson, Jonsson, Steffensen & Sveinbjømsdottir 1992). Recent seismic studies have indicated that a similar but smaller moraine-like ridge occurs at the mouth of Kong Oscar Fjord to the north of Scoresby Sund, but is lacking from other fjords (Hubberten, Grobe, Jokat, Melles, Niessen & Stein 1995). Coring on the shelf and its edge at the mouth of Scoresby Sund and at Hochstetter Forland show maximum fluxes of terrigenous material and pulses of IRD indicating the presence of ice bergs and melt water from glacier fronts on the shelf between 16 and 21 ka. This was correlated with LGM in Scoresby Sund, the Flakkerhuk stade (Nam, Stein, Grobe & Hubberten 1995; Stein, Nam, Grobe & Hubberten in press). The history of the glacier in Scoresby Sund is discussed further below.

Scoresby Sund between 16 and 10 ka

The Scoresby Sund drainage system is the largest single outlet from the eastern margin of the Inland Ice. Owing to work during the PONAM Project this area has the most detailed record of events during and after LGM (Funder, Hjort & Landvik 1994), and current field work has added to this. The position of ice margins and key-14C dates are shown on Fig. 4, and details of the 14C-dates are given in Table 1. As noted above, a large outlet glacier filled the fjord system during LGM probably with its front on the Kap Brewster sedimentary ridge, between 16 and 21 ka (Fig.4). Six thousand years later, during the Milne Land stade, the glacier fronts had receded c. 150 km and uncovered the outer fjord basin. Current studies on western Jameson Land show that the glacial history during and after LGM began when the glacier in Scoresby Sund ran along the present coastline and deposited proglacial sediments along the margin and in blocked river valley basins. Next, during the peak of glaciation, the glacier expanded inland over the ice dammed lake basins, up to 15 kilometres from the coast and deposited thin and discontinuous till over the area. After this, the ice melted, maybe with lakes dammed against its margin. In the final collapse much of the land based ice was transformed to dead-ice as seen from the numerous kames. The retreating ice front was followed by marine water, and the present pattern of fluvial drainage was established with marine limit at c. 70 m above sea level in western Jameson Land. During the Milne Land stade the fjord glaciers advanced to positions on the western margin of the Scoresby Sund basin, but did not reinvade western Jameson Land (Fig. 4).

Thursday, 11 July 2019

Strange tale from the Mediterranean

Thanks to Dave for alerting me to this very strange and somewhat garbled article in The Independent — by David Keys,  one of our more gullible archaeological journalists:

I have no ideas how serious or well-founded this may be — it may just be the latest hoax from a group of researchers desperate for some headlines.......

Will do some research and report back......... in the meantime, if anybody out there knows more, and would like to share info, please post your comments.

Giant marble pyramid-shaped island complex rising from sea uncovered, revealing secrets of ancient Greece’s origin

Exclusive: Thousands of years of history being unlocked in the Aegean isles which could provide groundbreaking knowledge of ancient civilisations.


Archaeologists now believe that, in order to construct the complex, early Bronze Age Greeks embarked on at least 3,500 maritime voyages to transport between 7,000 and 10,000 tonnes of shining white marble from one Aegean island to another.

Each return voyage would have required up to 24 crew members to paddle for around five hours.

“It is by far the largest prehistoric marine transport operation that has ever come to light anywhere in the world,” said Dr Julian Whitewright, a leading maritime archaeologist at the University of Southampton.

“It demonstrates quite clearly just how important, and integral to their culture, seafaring was to these early Bronze Age Aegean people.”

The voyages – totalling around 45,000 miles – allowed the architects to construct what is thought to have been a huge religious sanctuary consisting of up to 60 marble buildings, which were constructed specifically to glisten in the sun.

..........the architects “terra-formed” the pyramid-shaped island “mini-mountain”, known in recent centuries as Dhaskalio (possibly just meaning “islet”), to create around 1,000m of artificial terracing, arranged in six “steps” on its steep slopes.

These roughly six-metre wide terraces appear to have been built specifically to accommodate all the buildings. The summit itself was not initially built on – but instead had a small, probably sacred, open area where votive offerings may have been deposited.

Friday, 5 July 2019

More on ice gradients in the Celtic Sea arena

A winter photo of the Malaspina Glacier, the word's biggest piedmont glacier at the present time.

I have mulled over this topic before, because of the apparent mismatch between modelling work and "ground truthing" in which field evidence is crucial.  The diagram above shows assorted ice sheet long profiles, including some which accord reasonably well with the theoretical ice profile of an equilibrium glacier or ice sheet, and others (the red and green lines) which suggest that in some cases extremely low or "flat" profiles can be observed or deduced.  The diagram is from a 1973 article by Matthews, in which he wondered why glaciers with such low profiles could still move and be capable of eroding and transporting erratics and morainic materials.  See refs below.

The perfect piedmont glacier -- there are quite a few of them in North Greenland and Arctic Canada.  Generally the gradient of the feeder glacier is steeper, and the gradient on the piedmont part very flat..... 

The flat profiles come from the SW part of the Laurentide ice sheet and the "piedmont" glaciers flowing from the Cordilleran ice sheet. Lionel Jackson and colleagues have worked in areas affected by these glaciers, and have come to the view that gradients were often only about half as steep as those deduced by earlier researchers.  Without going into detail about shear stresses, basal sliding and internal deformation, there seems to be a rule that piedmont glaciers can have very low profiles as the ice spreads laterally across open country.  That makes sense,  and measured gradients are usually below 8m per kilometre and sometimes as low as 1m per kilometre.  

If we translate that into some crude figures for the Celtic Sea lobe, and if we assume for the moment that the long profile should be oriented from NE towards SW, as assumed by most researchers over the past 20 years, that means there could have been a grounded snout at -150m on the continental shelf edge and an ice surface at +250m about 400 km away in St George's Channel (if the gradient was 1m per km).  At the other extreme, with a gradient of 8m per km, the ice surface in St George's Channel would have been at +3050m -- more or less as I have previously assumed in earlier posts.   If we take a more modest figure of 4m per km, assuming the ice was dynamic enough to cross the Celtic Deep and still push southwards for 400 km, we get figures of -150m at the snout and +1,450m at the St George's Channel constriction point.

If the new suggestion from Scourse et al (2019) is to be taken seriously, glacially-forced isostatic depression on the shelf edge around 27,000 years ago must imply thicker rather than thinner ice, grounded over a substantial area.  No doubt the modelling people are hard at work on this issue as we speak...........

Whatever the models throw up, we have to explain the presence of a Devensian ice edge at about 300m on Carningli and about 250m on the north face of Mynydd Preseli -- and at assorted altitudes below 100m around the south Pembrokeshire coast.  Can these ice edge altitudes be accommodated in a scenario involving a very low long profile for the piedmont glacier in the Celtic Sea arena?  Watch this space......



W. H. Mathews (1974) Surface profiles of the Laurentide ice sheet in its marginal areas. Jnl of Glaciology,  Volume 13, Issue 67 1974 , pp. 37-43.  Published online: 30 January 2017.


Jackson, L.E., Jr., Little, E.C, Leboe, E.R., and Holme, P.J., 1996: A re-evaluation of the paleoglaciology of the maximum continental and montane advances, southwestern Alberta; in Current Research 1996-A; Geological Survey ofCanada, p. 165-173.

Controversy over the age of the maximum known advances of continental and montane glaciers in southwestern Alberta rests largely on the paleoglaciology of the maximum montane piedmont glacier. Reconstruction of it, based upon features that mark the former glacier surface in the Foothills, indicate that the glacier had a maximum surface gradient of 0.4-0.6° - less than half the value assumed by previous workers. The continental ice sheet that was roughly coeval with the piedmont glacier and coalesced with it in the Foothills. Paleoglaciological evidence favours a Late Wisconsinan age for maximum extent of continental and montane ice.

Was there a Devensian ice stream in the Celtic sea arena?

Ice streams around the northern sector of the Laurentide Ice Sheet, shown with relation to reconstructed ice sheet edges.  The Hudson Bay Ice Stream appears to have flowed in a distinct channel while bounded by more sluggish ice.  The only ice stream that appears to have flowed as a lobe beyond the ice sheet edge was the McClure Ice Stream, but it too flowed through a highly constrained trough between islands like a great outlet glacier.  It was probably bounded by local ice caps. It does not provide a good analogue for the Celtic Sea arena.

On the matter of the name we should give to the 80,000 sq km ice mass that occupied the Celtic Sea arena in the Late Devensian, this very comprehensive article is of great interest and relevance.  Scourse et al  (2019) have suggested that the Hudson Bay ice lobe is a reasonable comparator for the BIIS that they consider pressed all the way out to the shelf edge — but the authors of this article suggest that nearly all of the identified ice streams around the edges of the Laurentide Ice Sheet were typographically constrained — in other words, they flowed in great troughs bounded by uplands.  The Hudson Bay Ice stream was no exception, and it did not apparently extend beyond the edge of the ice sheet itself at any stage.  

Reference (recommended reading!)

Margold, M., Stokes, C.s & Clark, C. (2015). Ice streams in the Laurentide Ice Sheet: Identification, characteristics and comparison to modern ice sheets. Earth-Science Reviews. 20. Volume 143, April 2015, Pages 117-146



This paper presents a comprehensive review and synthesis of ice streams in the Laurentide Ice Sheet (LIS) based on a new mapping inventory that includes previously hypothesised ice streams and includes a concerted effort to search for others from across the entire ice sheet bed. The inventory includes 117 ice streams, which have been identified based on a variety of evidence including their bedform imprint, large-scale geomorphology/topography, till properties, and ice rafted debris in ocean sediment records. Despite uncertainty in identifying ice streams in hard bedrock areas, it is unlikely that any major ice streams have been missed. During the Last Glacial Maximum, Laurentide ice streams formed a drainage pattern that bears close resemblance to the present day velocity patterns in modern ice sheets. Large ice streams had extensive onset zones and were fed by multiple tributaries and, where ice drained through regions of high relief, the spacing of ice streams shows a degree of spatial self-organisation which has hitherto not been recognised. Topography exerted a primary control on the location of ice streams, but there were large areas along the western and southern margin of the ice sheet where the bed was composed of weaker sedimentary bedrock, and where networks of ice streams switched direction repeatedly and probably over short time scales. As the ice sheet retreated onto its low relief interior, several ice streams show no correspondence with topography or underlying geology, perhaps facilitated by localised build-up of pressurised subglacial meltwater. They differed from most other ice stream tracks in having much lower length-to-width ratios and have no modern analogues. There have been very few attempts to date the initiation and cessation of ice streams, but it is clear that ice streams switched on and off during deglaciation, rather than maintaining the same trajectory as the ice margin retreated. We provide a first order estimate of changes in ice stream activity during deglaciation and show that around 30% of the margin was drained by ice streams at the LGM (similar to that for present day Antarctic ice sheets), but this decreases to 15% and 12% at 12 cal ka BP and 10 cal ka BP, respectively. The extent to which these changes in the ice stream drainage network represent a simple and predictable readjustment to a changing mass balance driven by climate, or internal ice dynamical feedbacks unrelated to climate (or both) is largely unknown and represents a key area for future work to address.