A small glacial cirque basin on Exmoor, Somerset

A small glacial cirque basin on Exmoor, Somerset Stephan Harrison*, Ed Anderson" and David G. Passmore" HARRISON, S., ANDERSON, E. & PASSMORE, D. G. 1998. A small glacial cirque basin on Exrnoor, Somerset. Proceedings oj the Geologists' Association, 109, 149-158. Geomorphological and sedimentological investigations at the Punchbowl on Exmoor, north Somerset, have demonstrated evidenceof cirque basin glaciation, some 30 km south of the accepted southernlimit of Anglian ice cover in this area. The Punchbowl is a steep-sided, north-facing basin that has a subdued arcuate ridge at its mouth. Analysis of rnacrofabrics, clast form and roundness and particle size on sediments forming the ridge indicate it to consist of glacial diamict deposited as a terminal moraine. Further evidence supportingthe presence of a former cirque basin glacier at this site is derived from basin morphologyand a striated sandstone boulder. Palaeoglacierreconstruction gives a surface area of 0.38 km-, a maximum thickness of 46.5 m and a maximum basal shear stress value of 51 kPa. The ELA of the former glacier has been calculated at 334 mOD. There is no evidence for multiple glaciations at this site, and, in the absence of dating controls on the moraine, the glacier is provisionally assumed to have formed during severe climatic conditions associated with the southerly advance of ice sheets during the Anglian cold stage or during the Dimlington Stadial. Development of a glacier at this location appears to have been a rare event during the Quaternary, and may have been facilitated by accumulation of windblown snow from the adjacent plateau. 'Centre jor Quaternary Science, Coventry University, Priory Street, Coventry CVI 5FB. "School oj Geography, Middlesex University, Queensway, Enfield EN] 4SF. tDepartment oj Geography, Newcastle University, Tyneside NEI 7RU.

1. INTRODUCTION At present, there is general agreement that the southernmost land-based glacial ice in the British Isles during the Quaternary was the Anglian glacial limit which runs from the northern coast of the Southwest Peninsula to the northern part of the Thames estuary (Bowen, Rose, McCabe & Sutherland, 1986; Jones & Keen, 1993). However, evidence is presented here of cirque basin glaciation in the Punchbowl on Exmoor, Somerset to the south of this limit. Straw (1995) has also highlighted the possibility that this site may have contained a small glacier; he wrote that, 'if it (the basin) had reached a size sufficient for occupation by a permanent snow patch within which some internal deformation was possible, it is, perhaps, Exmoor's sole instance of an incipient glacial corrie' (p.22). However, Straw (1995) did not present any evidence to support this idea. In this paper, detailed geomorphological and sedimentological evidence demonstrate the presence of former glacial ice at this location. From this evidence it is possible to reconstruct the dimensions of the former glacier, its basal shear stress and the Equilibrium Line Altitude (ELA).

2. DESCRIPTION The Punchbowl is an impressive north-facing basin which is carved into the broad plateau of Winsford Hill (426 mOD) (Figs 1,2 and 3). It is located at SS 883 345, some 2 km due west of the village of Winsford. The area is underlain by Proceedings of the Geologists' Association. 109, 149-158.

Upper Devonian Morte Slates and sandstones, and the siltstones and slates of the Pickwell Down Beds; the junction between these formations lies some 600 m north of the southern rim of the Punchbowl (Geological Survey of England and Wales 1969, Sheet 294, Dulverton). The floor of the basin is at 300 m OD and the steep backwall rises to 400 m 00 and is set at an angle up to 70°. Bedrock is exposed at several places on the backwall. The basin is up to 350 m wide and its downslope edge is marked by a subdued arcuate ridge up to II m high and 40 m wide which has been incised by a small stream (Fig. 4). This ridge is 110 m in distance from the base of the backwall. The basin floor is covered with thick vegetation and bog peat and the basin sides are also vegetation covered.

3. METHODOLOGY Geomorphic and sedimentological methods were used to determine the origin of the ridge at the front of the basin and the debris cover which mantles the adjacent plateau and valley sides. Detailed geomorphological mapping of the Punchbowl (see Figs I and 4) was carried out using enlarged base maps reproduced from the 1:10 000 Ordnance Survey map. The map was redrawn, scanned and digitized. Slope gradients were measured in the field using a clinometer and topographic profiles constructed from the I: 10000 map. Five sites were selected to investigate the composition of the basin ridge and the plateau and valley-side debris 0016-7878/98 $10·00 © 1998 Geologists' Association

s.

ISO

II ARRI S O N

tcr -v z..

," .' >0

»63 urn ) and sedi graph analysis « 63 urn). In addition , the sediment profile ex pose d at Site 3 was log ged using sta ndard lithostratigraphic techn iques.

4. RESULTS Morph ometric evidence sugg esting signific ant ero sio n of the Pun chbowl includes the stee p back wall and enclosed basin shape of the hollow. To the wes t, and also incise d into Winsford Hill are fo ur other north wa rd-facing tributary valleys . Each of these are cut into the Morte Slates and Pick well Down Beds; the latter rock s underlying their headwaters. These valleys are de scrib ed below.

Riscombe Combe Th is forms a semi-circular headed basin 300 m to the north wes t of the Punchb owl. Th e floor of the co mbe is smooth and rise s ge ntly from 270 m 00 to 300 m OO . Th e

152

S . H ARR IS O N ET AL .

.

\

J~ () _ .

__

I

I

I

R.\ o I

2(l(}m I

• Site 4 ._---.....

:?)

moraine ridge

~ stream cutting

)

Q

.::1~ marsh

rockwall trees

- - ..10 -

contours in metres

Fig. 4. Macrofabrics from the arcuate ridge and Site 3.

headwalls of the combe are steep (average angle of 32°) and rounded and rise to 370 m Ol) . The basin is up to 200 m in width and 350 m long. There is no stream occupying the floor at the present time.

rises at a spring at 350 m Of) and the mouth of the combe lies at about 280 m Ol) . The combe is about 300 m in length and 100 m wide at its widest part. Some 250 m further west is Ash Combe.

Unnamed combe

Ash Combe

This combe is located 400 m due west of Riscombe Combe. Its morphology is less rounded than Riscombe Combe and the floor is occupied by a small stream. The head of this

This basin is occupied by a stream and the smoothed. rounded basin form displayed by both Riscombe and the unnamed combes is disrupted at Ash Combe by headward

Key

Slabs

Sile 1.

%

. Mean roundness

60 oj 5 .0 • Standard

~~1~

deviation

0.08

40

n . 50

X

~

Site 2

0

'~(: 'h ·~%I·~

lOlh

80-1

Site lb

~

• •••:~~ :.:

.

\t : :~"

= 98% = 0.17

': ' .

= 0.07

Site 3

r

» o » r

~

,0 \

'

.. . -;'

. .

o ?J

oC

Site 0

SiteS

I

lA' A 'SA'SA'R wA'

I

tTl OJ

»

Vl

Z

m X = 0 37 50 .005

1

X . 0 38 50 .005

1

X . 0 38 5 .0 . 0.06

1

X. 02 5 .0 .0.04

l

X.

0 21

50 . 003

l

X.

0.33

5 0 .007

I

>< ~

0 0

iO

Site 1a

Site lb

Site 2

Site 3

Sile o

Site S

Fig. 5. Clast form and roundness diagrams from the five sites.

lJ> W

s.

154

H AR RI S O N

incision by the stream. As a result, the headwall is cut by two distinct strea m channels. The headwall rises to 350 m 00 and the combe is 400 m in length and up to 200 rn wide . Little Ash Combe lies 550 m to the west.

tcr

AL.

0.0

-r----.,

0. 1 0.2

/

topsoil

Little Ash Combe Th is basin , the most westerly of the four, also has the least rounded headwall; this and the sides having been incised by stream ero sion. Th e headwall rises to 350 m 00 and the mouth of the co mbe is at 300 mOD, some 350 m to the northeast. The co mbe is occ upied by a small stream and is up to 100 m in width. The geo morphology of the Punchbowl is shown in Figs I and 4. The arcuate ridge completely encloses the basin mouth exce pt where the stream has incised through it to form a v-shaped notch. There are no expos ures of bedrock in the ridge . The sediments exposed at Site 2 near the base of the ridge and Site I near the crest form a poorly-sorted, silty clay diamict with numerou s clasts. Elongate clasts display a prono unced preferred orienta tion with their a-axes pointing in a southwest (upslope) direction. Clasts from these sites have c:a ratios of 0.4 or less of 82% and 76% (Site I) and 78 % (Site 2) and they are subrounded to rounded (mea n roundness is 0.37, 0.38 and 0.38 respectively) (see Fig. 5). The clasts at Site 3 sho w a welldeveloped down slope macrofabric and are predomin antly 'slabby' in shape (98% have c:a ratios sOA ). The clasts are mainly angular or very angular (mean round ness of 0.2). A sectio n log from here (Fig. 6) show s 0.5 m of highly contorted sediments containing angular clasts overlying a similar, though uncontorted, unit some 1.2 m deep. Similar roundness and cla st form values are obtained from Site 4 (% c:a s O.4 is 98% and mean roundness of 0.2 1 (angular to subangular) althou gh it was not possible to obtain macrofabrics from this site. Differen ces between the sites are even more marked when usin g the percentage of c:a rat ios sO.3 as a discrim inator. In this case, the samples from Sites I and 2 on the arcuate ridge have c:a ratios of sO.3 of 50, 46 and 46%, which compares with 98% for both of the samples taken from sites above the Punchbowl (Site 4) and down slope of the ridge (Site 3). Differences in roundness of clasts bet ween the ridge sites and those outside the ridge are significa nt at p=O.OOl. Particle size analysis of the finer than 2 mrn frac tion does not distingu ish between the sediments (Fig. 7); percentages of silt and clay lie between 14 and 24%.

5. INTERPRETATION The depositi on al facie s cha racteristics outlined above clea rly dem onstr ate distinct sedi me ntologi cal co ntrasts between Sites I & 2 and Sites 3 & 4 which are probably related to differences in origin. It is proposed here that the sediment exposed at Sites I and 2 is a glacial diamict deposited at the margin of a small cirqu e glacier situated

involut ed angular grave ls and co bbles with silty sandy matrix

a ngular gravels and cobb les wit h silty sandy matrix

sa mple site

strea m channel Fig. 6. Section log from Site 3.

within the Punchbowl basin. In contrast, the deb ris cover which mantl es the adjacent plateau and valley sides is considered to be periglacial in origin. This interpretation is based on the following two arguments . \. Th e Pun chbowl has a constructional arcuate ridge at its mouth ; this is composed of matrix-supp orted diamict with numerous subrounded and rounded clasts. Similar sedimentary constru cti on al landfo rms are freq uent fea tures at the ablation margins of glaciers in contemporary settings, where the presence of rounded particles may be explained by the abra sion of clasts transported along subglacial pathways (Boulton, 1978). Therefore this feature is interpreted as a term inal/l ateral moraine ridge. In co ntrast, the adjacent slopes and plateau areas are mantled by a cover of cla st-supported angular gravels and cobb les with interstiti al fines. Significantly, the upper zone of this debris mantle is highly involuted (see Fig. 6). Conseq uently, this debris mantle is interpreted as

155

G LAC I A L C I R Q U E B A SI N . E X MOO R

--::-:--::-:-:-::--=-=-- - - - - - - - - - - - - - - - - - - - _ ----,

100 r

_________

"", _ . . _ 0.-

•••.••..••••.•••••.•.:.:~.2:·:-~= ·:~-"~·:~ : C C~': ... - - - -

~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~-~ ~~~~~:~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ -

-- - - ~ .:c. .:c.~:::/

- - - -

11

~

0

-

-/ "- -

,/

.'

~.,--/~ ~

-:(~>. ;~ -

"

/.-. .- >- - - .. "..'

-

-

-

----

- - - - - - -

-- - - - - - - - - - - - - - - - - -- - - - - - -

Site %

===-- -_==========: =========: ==: =================== ~ - - - -- - - -- -- - - - - - - - - - - - - - - - - - - - --- - - -

- -

-

".

-

==L =================================================== --1- =------------------------------------------------- i . - -- - -- - -- _

-4

-3.5

-3

KEY

- - Site1 ................ Site 2 - - - Site 3 .. _ .. - S ite 4

-2.5

-2 -1.5

·1

·0.5

0

0.5

1.5

2

2.5

3

3.5

1 2 3 4

sic Ph i me d P h i me a n

14.42 18.37 24 .79 21.65

' 0.08 0 .18 0.4 0. 83

-0 .01 0 .4 1 0 .76 1.04

4

P hi s iz e

Fig. 7. Particle size curves of the finer than 2 mm fraction.

the product of prolonged perig lacial weathering of regolith and ill situ cryoturbation stresses . 2. Macrofabric analyses from the ridge diamict demonstrate that the component clasts possess strong fabrics related to the highly orientated clustering of a:b plane s parallel to the direction of flow (Fig. 4). Normalized eigenv alues are compatible with subglacial till fabrics recorded at contemporary glacial margins (Dowdeswell & Shar p, 1986; Hart, 1994; Benn, 1994, 1995) and are interpreted as having been formed by subglacial deformation of diamict during the accretion of the moraine ridge. The contrast in eigenvalu es between Sites 1 and 2 may be related to differences in shear strength and strain rate of the diamict during the final stages of subglacial deformation (cf. Dowdeswell & Sharp, 1986; Hart, 1994; Benn, 1995). The sediment profile at Site 3 also possesses a strong clast fabric, but since the a:b planes are orientated downslope at dip angles parallel to the slope gradient, the macrofabric is interpre ted to have formed in response to solifluction activity. Further evidence supporting the development of glacial ice at this site is demonstrated by the presence of a large sandstone boulder of local origin with a facetted and striated face which is resting against the northern side of the wall about 5 m to the west of the stream (Fig. 8) on the proxim al side of the ridge.

The depositional characteristics of the basin ridge demonstrate a glacial origin. However, its shape and position within the basin could encourage speculation that it is a protalus rampart. Indeed, Straw ( 1995) proposed this as a possible explanation for the basin ridge . Relict protalus ramp arts and moraine ridges can be notoriously difficult to differentiate (Shakesby & Matthews, 1993). Ballantyne & Kirkbrid e ( 1986) proposed several criteria to facilitate their distin ction. The points most relevant here are that protalus ramp arts are always located at the base of a talus slope and the rampart crest is usually within 30--40 m. of the talus foot; they are normally composed of coarse openwork debris with a variable amount of infill fines; rampart clasts are generally angular and slabby in form. Results of the current study demonstrate that the topogra phic location and sedimentology of the basin ridge fails to match any of these criteria . For instance, the ridge crest is over 200 m from the base of the steep northeast-facing slope which represe nts the only realistic site for significant talus accumulation; the ridge is composed of matrix-rich diamict and component clasts are generally subrounded. Therefore, on the basis of this evidence the idea that the Punchbowl basin ridge is a protalus rampart must be rejected . Development of the other valleys to the west of the Punchbowl can be seen in the context of varying amounts of snowblow when winds were from an easterl y direction and

156

S. H ARRIS ON ET AL.

Fig. 8. Boulder on the north side of the stream on the ridge showing facetted and striated surface.

blowing snow from the eastern and southeastern plateau areas of Winsford Hill. Their differing forms may reflect increasing snow starvation to the west (owing to capture of snow by the Punchbowl) and hence increasing fluvial activity at the expense of nival proce sses. The changing nature of the valleys cut into the north face of Winsford Hill may therefore be taken to represent the transition from valley form domin ated by glacial, nival and then fluvial processes with the Punchbowl probably having developed from a fluviall y-incised valley to its present glacial form.

6. PALAEOGLACIER RECONSTRUCTION The reconstruction of the surface area and morpholo gy of the Punchbowl glacier was based on the evidence of the geomorphic field mapping and followed the same procedure outlined by several authors when reconstructing Younger Dryas cirque glaciers in the British upland s (e.g. Sissons, 1974, 1980; Gray, 1982; Thorp, 1986; Ballantyne, 1989). Delimitation of the former ablation margin along the line of the terminal moraine ridge crest was straightforward; however, reconstru ction of the former limits in the headwall area was more difficult because of the absence of trimline evidence. Consequentl y. the former margin of the glacier accumulation zone had to be extrapolated from the ablation area. Assuming that the former accumulation zone extended up to 30 m below the top of the backwall slope (cf. Gray, 1982), the former surface area of the glacier was estimated to be 0.38 km-'. The glacier contours were reconstru cted by analogy to the characteristic contour pattern of contem-

porary glaciers. Former maximum glacial thickness was estimated at 49.5 m by assuming that the present morphology of the basin floor has changed little since glaciation and superimposing elevation profiles of the reconstructed glacier and present land surface . Maximum basal shear stress was calculated to be 51 kPa (using an F factor of 0.8) employing the relation ship outlined below: 1 =pgh

sin a

where 1 is sheat stress, p is ice density, g is acceleration due to gravity, h is ice thickness and a is slope gradient. Although the value of basal shear stress must be regarded as approximate (bec ause the shear stres s relationship assumes zero basal slip or subglacial deformat ion), it falls within the range of 50-ISO kPa regarded as characteristic of contemporary glaciers (Paterson, 1994). An estimate of the former ELA for the glacier was established using the area-weighted mean method outlined by Sissons (1974 ). This procedure is based on two assumptions: (I ) during the glacial maximum the glacier was in equilibrium; (2) both the accumulation and ablation gradients have a linear relationship with altitude . Providing that these assumptions can be satisfied then the position of the ELA can only be related to the distribution of altitude on the glacier surface. Using this technique the ELA was calculated to be 334 mOD . Howe ver, this approach may be flawed for two reasons. First, at present, there is no absolute evidence that the glacier was entirely constrained by the basin and it may have been an outlet of a small ice cap positioned over Winsford Hill (similar relationships are

G L AC IA L C I R QU E B ASIN , E X MOO R

observed by Gellatley, Whalle y & Gordon, 1986). Second , small glaciers do not necessaril y have ELAs; in some years they have positive mass balances and in other years they display negative mass balances. Howe ver, by adopting a simple systems approach to glacier dynamics, the calculation of an equilibrium line does prov ide a useful index of the relationships between a glacier, altitude and climate. The reconstructed Punchbowl glacier is shown in Fig. 9. The devel opment of glacial ice within the basin was probably partially controlled by the redistributi on of plateau snow by wind. However, it is impossible to quantify the influence of snow -blow since the regional annual precipitation at the time of glacier development is not known .

7. AGE AND PALAEOENVIRONMENTAL SIGNIFICANCE At pre sent , the age of the Punchbowl moraine ridge is unknown . The age of Pleistocene moraine ridges can be determin ed by dating material incorp orated within a diamict or associated glacigenic sediment, or by establishing local

157

morphostratigraphic relation sh ips. Howe ver, inspection of exposed sections within the ridge revealed an absence of material suitable for dating, and its location south of the Anglian ice sheet limit undermines the application of standard morphostratigraphic method s. Furthermore , the sedimentary sequence beneath the basin floor rule s out establishing the minimum age of the infill by pollen stratigraph y (c. J. Caseldine , per s. comm .). However, consideration of the geomorphic evidence does shed some light on the likel y glacial history of The Punchbowl. Two points are particul arly important. I. At the mouth of the Punchbowl is a single arcuate moraine ridge with a distinct crest and there are no other recessional or advance ridges within the basin. Beyond the terminal moraine, the landscape reflects a periglacial legacy and there appears to be no ev idence of former glaci ation here. Therefore, the moraine ridge appears to be the result of a single glacial advance , though it may be an accretional feature related to more than one glacial phase. It is highly unlikely to be the product of multiple glaciations since there is no evidence that the basin was occupied by glacial ice on several occasio ns durin g the Quaternary. 2. The southerly location of the Punchbowl and the low ELA of the reconstructed glacier suggests that the development of glacial ice here was a rare event during the Quaternary and was probabl y initiated by the severe climatic conditions which fuelled the southerly adva nce of ice sheets during the Anglian cold stage and the Dimlington Stadia!. In co mparison to the Dimlin gton Stadial ice sheet ELA calculated by Boulton, Smith, Jones & Newsome (1985), the reconstructed Punchbo wl ELA appears anomalously low, implying that glacial ice could not have formed here during the Dimlington Stadi a!. Howe ver, the presence of large plateau areas above the site implies that the growth of glacial ice could have been considerably aided by the accumulation of windbl own snow. Therefore the redistribution of snow into a north-facing basin during the Dimlington Stadial may have been sufficient to trigger the development of glacial ice. If the moraine is of Anglian age, the absence of thick mass-wasting deposits within the basin suggests that during later cold stages the Punchbowl was occ upied by a protective cover of nival ice which prevented infillin g of the depres sion.

8. CONCLUSIONS x

-~ )SO

Altit: mOD

Fig.9. Reconstruction of the Punchbowl glacier.

v Geom orphological and sedimentological investigations at the Punchbowl have demo nstrated the presence of a former cirque basin glacier located 30 km to the south of the accepted southern limit of Quaternary ice cover in this part of the British Isles. The absence of evidence for multipl e glaciations in the Punchbo wl, coupled with intensive periglacial modification of debris cove r mantling the adjacent plateau and valley sides, suggests formation of glacial ice

158

S . HARRISO N F:T A L .

her e was a localized a nd rare e vent. N o d at in g co ntro ls are availa ble fo r sedi me nts fronting o r infi lli ng the basin; ho we ver, in view o f the so uthe rly locat ion of the g lac ie r. it is co ns ide re d m o st likely to have form ed during the extreme climatic cond itio ns ass oc ia ted w ith the m aximum advanc e of ei the r the A nglian or Dimlingt on Sta dial ice shee ts . While rel ati vel y low ELA va lues for the recon structed Pu nc hb o wl g lacie r would seem to pr eclude a la te r dat e for thi s fea ture, th e g ro w th of glacial ice during th e Dimlington Stadial m ay have been facilitated by accum u la tio n of w ind blown snow from the adjacent plateau. The former pre sence of g lac ia l ice at rel at ively low altitudes o n Exmoor sugges ts that sim ila r ice masses should have de vel op ed o n the higher a nd m ore we sterly plateaus of Dartmoor a nd B odmin Moor.

These find ing s suggest tha t othe r landforms o n th e upl an ds of so uth west En gl and may need re-evaluating in te rms of a pot ential g lac ia l legac y in the la nd scap e .

ACKNOWLEDGEMENTS The a uthors w is h to th an k Jan Foster and tw o a no ny mous referee s who made useful com me nts on an earl ie r d raft of thi s paper; the farmer at Withycombe F arm for access to the si te and Te ssa Kin gsley wh o helped wi th fie ld work. A preliminar y poster ve rsio n of thi s paper was presented at the British Geomo rpho lo gical Re se ar ch Group Annual Confe ren ce at Dundee in September 1997.

REFERENCES ANDR EWS, J. T. & SHIMIZU, K. 1966. Three-dim ensional vector technique for analyzing till fabris: discussion and FORTRAN program. Canada Department of Mines and Technical Surveys Bulletin,8,1 51- 165. BALL ANTYNE, C. K. 1982. Aggregate clast for m charac teristics of dep osits at the margins of four glacie rs in the Jotunheimen Massif, Norway. Norsk Geografisk Tidsskrift, 36 , 103-11 3. - - 1989. The Loch Lomond Readvance on the Isle of Skye. Scotland: glacier reconstruction and palaeoclimatic implications. Journal of Quaternary Science, 4, 95- 108. - - & KIRKBRIDE, M. P. 1986. The charac teristics and significance of some Lateglacial protalus ramp arts in upland Britain. Earth Surface Processes and Landf orms , 11, 659-67 1. BENN, D. I. 1994 . Fabri c shape and the interpretation of sedimentary fabric data. Journal of Sedimentary Research. 64, 9 10-9 15. - - 1995. Fa bric signatu re of subglac ial till deform ation , Breidamerkurj (kull , Iceland. Sedimentology. 42 , 735-747. BOULTON, G. S. 1978. Boulder shapes and grain size distributions as indica tors of trans port paths through a glac ier and till genesis. Sedimentology, 25, 773-799. , SMITH, G. D., JONES , A. S. & NEWS OME , 1. 1985. G lacial geo logy and glaciol ogy of the last mid-latitude ice shee ts . Journal of the Geologi cal Society, London, 142, 447--474. BOWEN, D. Q., ROSE, J., MCCABE , A. M. & SUTHERLAND, D. G. 1986. Correlation of Quaternary glacia tions in England, Ireland, Scotland and Wales. Quaternary Science Reviews . 5, 299-340. DOWDESWELL, J. A. & SHARP, M. 1986. Characterisation of pebble fabrics in modem terr estri al g lacige nic sed iments. Sedimentology. 33, 699- 7 10.

GELLATL EY, A. E, WHALLEY, W. B. & GORDON, J. E. 198fJ. Topographi c control over recent glacier changes in southern Lyngen Peninsula, North Norway . Norsk Geografisk Tidsskrift, 40 (4) , 211-2 18. GR AY, J. M. 1982. The last glacie rs (Loch Lomond Advance) in Snowdon ia. North Wales. Geological Journal. 17, 111- 133. HART, J. K. 1994. Till fabrics associated with deformab le beds. Earth Surface Processes and Landforms. 19, 15- 32. JON ES, R. L. & KEEN. D. H. 1993. Pleistocene Environments in the British Isles. Chapman & Hall, London. MARK , D. M. 1973. Analysis of axial orientation data including till fabris, Geological Society of America Bulletin. 84 , 1369- 1374. PATERSON, W. S. B. 1994. The Physics of Glaciers . Pergamon, Oxford . POW ERS, M. C. 1953. A new roundn ess scale for sedimentary particles. Journal of Sedimentary Petrology. 23. I 17- 119. SHAKESBY. R. A. & MAlTHEWS, J. A. 1993. Loch Lomond Stadial glacier at Fan Hir, Mynydd Du (Breco n Beacons). South Wales: c ritical ev ide nce and palaeoclimat ic implications . Geological Journal, 28, 69-79. SISSONS, 1. B. 1974. A lateglacial ice-cap in the Central Gramp ians, Transactions ofthe Institute of British Geographers . 62, 95-1 14. - - 1980. Palaeoclimatic inferences from Loch Lomond Advance glaciers. In (Lowe, 1. J., Gray, J. M. & Robinson, J. E.: eds) Studies in the Lateglacial of North-West Europe. Pergamon. Oxfo rd, Pl'. 3 1--44. STRAW, A. 1995. Aspects of the geomorphology of Exmoor. In (Binding, H.: ed.) The Changing Face of Exmoor. Exmoor Books, Tiverton. THORP, P. 1986. A mountain icefield of Loch Lomo nd Stadial age. western Gram pians, Scotland . Boreas. 15, 83-97.

Manuscript received 5 December 1997; typescript accepte d 30 J anuary 1998