Quaternary Science Reviews 22 (2003) 1659–1685 Subglacial till: the deforming glacier bed Jaap J.M. van der Meer a, *, John Menzies b , James Rose c a Department of Geography, University of London, Queen Mary Mile End Road, London E1 4NS, UK b Department of Earth Sciences, Brock University, St. Catharines, Ont., Canada L2S 3A1 c Department of Geography, University of London, Royal Holloway, Egham, Surrey TW20 0EX, UK Received 29 November 2002; accepted 24 April 2003 Abstract ‘‘Till is a sediment and is perhaps more variable than any sediment known by a single name.’’ R.F. Flint 1957 Glacial and Pleistocene Geology Tills are commonly classified according to the perceived process of deposition. However, it is increasingly recognised that this classification, which is mainly based on macroscopic field data, has severe limitations. At the same time the concept of the deforming glacier bed has become more realistic as a framework for discussing tills and their properties, and this (tectonic) concept is irreconcilable with the existing (depositional) till classification scheme. Over the last 20 years large thin sections have been used to study tills, which has provided new insights into the textural and structural properties of tills. These results have revolutionised till sedimentology as they show that, in the main, subglacial tills possess deformational characteristics. Depositional properties are rare. Based on this new insight the process of subglacial till formation is discussed in terms of glacier/ice sheet basal velocity, clay, water and carbonate content and the variability of these properties in space and time. The end result of this discussion is: till, the deforming glacier bed. To distinguish subglacial till from depositional sediments the term ‘tectomict’ is proposed. Within the single framework of subglacial till as the deforming glacier bed, many textural, structural and geomorphological features of till beds can be more clearly and coherently explained and understood. r 2003 Elsevier Ltd. All rights reserved. 1. Introduction Over the years till classification has become more complex. Whilst for many years it sufficed to recognise ablation tills and basal tills, there is now a whole array of till types. This is largely due to the activities of the INQUA Commission on Glaciation and its predeces- sors. Especially in the years when the Commission was chaired by Alexis Dreimanis huge progress was made in the description and understanding of tills. In the scheme outlined in Dreimanis (1988) the most common types of till to be classified are: lodgement tills, melt-out tills, flow tills, deformation tills and waterlain tills based on the processes whereby the tills are deposited. For lodgement tills, this process is the active lodging of particles on the glacier bed from the basal ice, caused by the forward movement of the ice overcoming the bond between ice and particles. Thus, the interpretation of a till as a lodgement till implies that the former ice body was active and temperate. Melt-out tills are produced by passive melting of ice, either from the top (supraglacial melt-out) or from the bottom (subglacial melt-out), where the resulting till is, by definition (Dreimanis, 1988), allowed to show some effect of the meltwater that is produced during the melting process. Interpretation of a deposit as a melt- out till implies that the ice body was passive. A similar process of sublimation is proposed for the formation of a rare variety of till in cold arid environments (Shaw, 1977). Lodgement and melt-out both involve the bond between particles and surrounding ice being released by melting of ice. The heat source for the melting is different, but the effect is the same. Another similarity between the two processes is that slabs of debris-bearing basal ice may be lodged and separated from the moving glacier. Consequently these can melt-out beneath actively moving ice (e.g. Menzies, 1989). ARTICLE IN PRESS *Corresponding author. Tel.: +44-20-7882-5403; fax: +44-20-8981- 6276. E-mail address: [email protected] (J.J.M. van der Meer). 0277-3791/03/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0277-3791(03)00141-0
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Quaternary Science Reviews 22 (2003) 1659–1685
Subglacial till: the deforming glacier bed
Jaap J.M. van der Meera,*, John Menziesb, James Rosec
aDepartment of Geography, University of London, Queen Mary Mile End Road, London E1 4NS, UKbDepartment of Earth Sciences, Brock University, St. Catharines, Ont., Canada L2S 3A1
cDepartment of Geography, University of London, Royal Holloway, Egham, Surrey TW20 0EX, UK
Received 29 November 2002; accepted 24 April 2003
Abstract
‘‘Till is a sediment and is perhaps more variable than any sediment known by a single name.’’
R.F. Flint 1957 Glacial and Pleistocene Geology
Tills are commonly classified according to the perceived process of deposition. However, it is increasingly recognised that this
classification, which is mainly based on macroscopic field data, has severe limitations. At the same time the concept of the deforming
glacier bed has become more realistic as a framework for discussing tills and their properties, and this (tectonic) concept is
irreconcilable with the existing (depositional) till classification scheme. Over the last 20 years large thin sections have been used to
study tills, which has provided new insights into the textural and structural properties of tills. These results have revolutionised till
sedimentology as they show that, in the main, subglacial tills possess deformational characteristics. Depositional properties are rare.
Based on this new insight the process of subglacial till formation is discussed in terms of glacier/ice sheet basal velocity, clay, water
and carbonate content and the variability of these properties in space and time. The end result of this discussion is: till, the
deforming glacier bed. To distinguish subglacial till from depositional sediments the term ‘tectomict’ is proposed. Within the single
framework of subglacial till as the deforming glacier bed, many textural, structural and geomorphological features of till beds can be
more clearly and coherently explained and understood.
r 2003 Elsevier Ltd. All rights reserved.
1. Introduction
Over the years till classification has become morecomplex. Whilst for many years it sufficed to recogniseablation tills and basal tills, there is now a whole arrayof till types. This is largely due to the activities of theINQUA Commission on Glaciation and its predeces-sors. Especially in the years when the Commission waschaired by Alexis Dreimanis huge progress was made inthe description and understanding of tills. In the schemeoutlined in Dreimanis (1988) the most common types oftill to be classified are: lodgement tills, melt-out tills,flow tills, deformation tills and waterlain tills based onthe processes whereby the tills are deposited. Forlodgement tills, this process is the active lodging ofparticles on the glacier bed from the basal ice, causedby the forward movement of the ice overcoming the
bond between ice and particles. Thus, the interpretationof a till as a lodgement till implies that the formerice body was active and temperate. Melt-out tillsare produced by passive melting of ice, either fromthe top (supraglacial melt-out) or from the bottom(subglacial melt-out), where the resulting till is, bydefinition (Dreimanis, 1988), allowed to show someeffect of the meltwater that is produced during themelting process. Interpretation of a deposit as a melt-out till implies that the ice body was passive. A similarprocess of sublimation is proposed for the formation ofa rare variety of till in cold arid environments (Shaw,1977).
Lodgement and melt-out both involve the bondbetween particles and surrounding ice being releasedby melting of ice. The heat source for the melting isdifferent, but the effect is the same. Another similaritybetween the two processes is that slabs of debris-bearingbasal ice may be lodged and separated from the movingglacier. Consequently these can melt-out beneathactively moving ice (e.g. Menzies, 1989).
0277-3791/03/$ - see front matter r 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0277-3791(03)00141-0
Deformation tills are defined as ‘homogenised,usually diamictic, sediment formed by glacially inducedshear of subsole material’ (Benn and Evans, 1997, p. 24).Thus they are formed by material being draggedforward by ice in the direction of glacier movementand this may be any pre-existing material, eitherlithified, or unlithified rock. It will also includedeformation of previously lodged particles which, insome publications has been called ‘(deformed) lodge-ment till’ and considered to be part of the lodgementprocess (Dowdeswell and Sharp, 1986) whereas othersdefine such material as deformation till (Hart, 1994,1995). The problem is not helped by the original‘definition’ of deformation till by (Elson 1961, 1989,p. 10) who ‘introduced the term deformation till to referto weak rock or unconsolidated sediment detached fromits source by subglacial shearing. The term was used todefine a spectrum of sediment types, including; (1)contorted sediment displaced only a short distance; (2)reoriented fragments of source material set in ahomogenised matrix; and (3) thoroughly homogenisedsediment in which all primary sedimentary structureshave been destroyed’ (from Benn and Evans, 1997, p. 24).As is evident, this was broad and encompasses all tillsshowing any macroscopic evidence of deformation.
The widespread use of terms such as lodgement till,melt-out till, flowtill, glaciomarine till suggests that theexisting classification is, at least to a degree, workableand widely acceptable. However, this classification hidesreal problems, which, for instance, arise from the factthat classification is based mainly on field criteria. Inparticular, attention has been given to structures ordirectional elements like clast fabric, or loosely definedparameters such as texture, consolidation and composi-tion that are discernable in the field and not based onlaboratory analyses. However, these criteria are notunique to specific varieties of till and experienced field
practitioners know that it is often impossible to come toa single conclusion (Table 1), regardless of whetherapplying single or multiple criteria. For example, drapedlamination structures in tills may be due to severaldifferent processes such as: sedimentation from melt-out, water deposition during a flow process, ordeformation whereby thick beds can be attenuated toform thin beds over clasts. Likewise, consolidation maybe due to glacier loading or drying. Although it isimpossible to differentiate which of these processes havebeen responsible by field study alone, they, and manylike criteria, are used as a basis for classification.
Other approaches for studying tills also introduceproblems. For the use of fabric properties and, inparticular, clast fabric statistics (vectorial or eigenvectorvalues) as genetic indicators (e.g. Rose, 1974; Dowdes-well and Sharp, 1986; Hart, 1994, 1995) assumes anumber of controls that are far from clearly understood(Bennett et al., 1999).
The inadequacy of field criteria for distinguishingbetween different types of till is demonstrated by thepublication of a number of different interpretations ofthe same data. For instance, lodgement tills changed tomelt-out tills in northern Germany (Ehlers, 1983), basaltills in coastal settings and containing molluscs havechanged to ‘in situ glaciomarine muds’ (McCabe andO’Cofaigh, 1995). Because of the equivocal criteria thereare many examples of interpretations that are highlycontested (for examples of such discussions see Malm-berg-Persson and Lagerlund, 1994; Rijsdijk et al., 1999;McCarroll, 2001). The point we wish to make is not tohighlight error or observer-bias, but to draw attention tothe limitations of the criteria used to define tills. For anattempt at objective recording of field data see Kr .ugerand Kjaer (1999).
The issue of till classification is further complicated bythe concept of a deforming bed beneath the glacier sole
ARTICLE IN PRESS
Table 1
Description of clast fabric of different till types by different authors
Author(s) Lodgement Subglacial
melt-out
Deformation Supraglacial
melt-out
Flow Sublimation
Bennett and
Glasser (1996)
Strong in the
direction of local
ice flow
May be strong in
the direction of ice
flow
Strong in the
direction of shear
Generally poorly
developed,
unrelated to ice
flow
May be strong,
related to palaeo
slope
Strong in the
direction of ice
flow
Boulton and
Deynoux (1981)
Strong flow
parallel peaks
Large-scale areal
consistency
Systematic areally
consistent
Difficult to resolve
systematic peaks
Peaks transverse
and parallel to flow
Dreimanis (1990) Well developed with parallel modes dominating, transverse
modes are also present, particularly in deformed parts of tills
Very variable, random to well defined
Hicock et al.
(1996)
Strongly clustered
and parallel to ice
flow
May be clustered,
parallel to ice flow
Bimodal to
multimodal
Rapid spatial
variation
Note that all descriptions strongly overlap and thus are not diagnostic. Similar overviews can be produced for all other till characteristics.
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–16851660
overlying undeformed glacial deposits (Boulton, 1987;Hart and Rose, 2001) and the fact that tills have beendiscussed in terms of different zones of deformation. Forinstance, Boulton (1987) has introduced the concept of‘a’ and ‘b’ horizons, Menzies (1989) has introduced ‘H’,‘M’ and ‘Q’ beds. In the latter classification, H beds aredefined by their hydraulic conductivity with subtypesranging from low conductivity in bedrock (Ha), to highconductivity in sediments overlying bedrock with lowconductivity (Hd). Meanwhile M beds are deformingbeds and Q beds are mixtures of H and M. Neither ofthese terminologies has yet been incorporated effectivelyin till classification.
In addition to the issues identified above a number ofadditional questions remain: (i) Does the fact that adeforming bed is inferred, imply that the resulting till is,by definition, a deformation till? (ii) Does a deformingbed obliterate all traces of particle release? (iii) Is itpossible, as suggested by Ruszczyska-Szenajch (2001),that glacier bed deformation cannot be described orexplained by existing till definitions?
In this paper we aim to demonstrate how micro-morphology, a laboratory method, may be used incombination with field observations to establish adifferent interpretation of subglacial tills. This newinterpretation is based on tectonic or structural criteriaalong with sedimentary properties, rather than solelysedimentary criteria. The implication of this work is thatsubglacial till is a tectonic deposit and not simply asedimentary deposit (van der Meer, 1993b). Thisdifferentiation is of profound importance for theunderstanding of contemporary subglacial depositionalprocesses and palaeoglaciological reconstructions. Inthis paper we seek to justify this contention by: (i)demonstrating the extensive body of work on which webase our ideas; (ii) using only field and laboratory dataand their interpretation, regardless of whether these fitexisting glaciological or geological theory or models,and (iii) developing a conceptual model of all subglacialtills as being the result of a deforming glacier bed.
2. Till micromorphology
2.1. Methodology
Micromorphology is the microscopic examination ofthe composition and constituent structural elements oflithified and unlithified earth materials (van der Meer,1987a). Samples are selected in the field or from cores torepresent either a particular structure, a particularmaterial unit or boundaries between material units.Where known, the orientation of samples and theirrelation to ice flow direction are noted. The sampleswhich are impregnated and used for micromorphologyare preferably of the dimensions 8� 14 cm, although
samples of 6� 10 cm samples are used regularly. Largethin sections have the important advantage that theyprovide a larger sample and provide a better representa-tion of macroscopically visible structures, includingstructural discontinuities like fissility, than small thinsections. We have found that standard petrographic thinsections are inadequate.
Whenever possible, samples are collected in the fieldin metal boxes of the dimensions given above. Thesesamples are taken to the laboratory where they are driedslowly, impregnated, cut and mounted following proce-dures described in Murphy (1986), van der Meer(1993b), Lee and Kemp (1993), Carr and Lee (1998)and Menzies (2000b). The thin sections are analysedwith a Petroscope (an amended microfiche reader inwhich the optics have been adopted to thin sectionanalysis) and with a low-magnification petrologicalmicroscope. Description follows the classifications ofBrewer (1976), van der Meer (1993b, 1997a, 2000) andMenzies (2000b).
Because the samples are undisturbed, the distributionand true orientation (Stroeven et al., 2002) of allparticles in the sediment, including those that are notvisible with the naked eye can be observed anddetermined. Furthermore because of the application ofa polarising microscope, it is possible to use the opticalproperties of particles to derive information such as thestress history of the material (see below). No otherlaboratory technique, including SEM (which does notallow use of optical properties), provides such detailedinformation about in situ diamictic materials.
2.2. Till plasma and birefringence—description and
definitions
Microscopic analysis of a large number of thinsections (see below) has resulted in the recognition ofan array of microstructures (Fig. 1). These includemicrostructures related to brittle deformation, ductiledeformation, mixed brittle and ductile (polyphase)deformation and to plasmic fabrics (Fig. 2). Thesemicrostructures are deformational in origin and notsedimentary. Macroscopic sedimentary structures in tillsmay be depositional, developed locally as lenses andbeds of sorted sediment by running water, but sedimen-tary microstructures in till are mainly inherited struc-tures which take the form of intraclasts that have beenincorporated by deformation. In addition to thestructures mentioned above, there are water-escapestructures (WESs) and sediments, clay coatings (cutans)lining porewalls and geochemical precipitates, all ofwhich tend to be limited in extent. Furthermore all theseadditional structures are related to percolating water orlocally channelised water, not to sedimentation from ice.The type of structures evident in glacial diamictons andthe general lack of sedimentary microstructures is a
ARTICLE IN PRESSJ.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–1685 1661
ARTICLE IN PRESS
Fig. 1. Diagram showing the range of microstructures recognised in tills (from Menzies, 2000b). S-matrix refers to the organisation of plasma,
skeleton grains and voids (Brewer, 1976).
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–16851662
ARTICLE IN PRESS
Fig. 2. A selection of micrographs to demonstrate the variation of structures that are common in glacial sediments at the microscopic level. The
examples are arranged in an ascending order of complexity and intensity of deformation. All examples are taken from vertical thin sections, but
orientation relative to the azimuth or ice direction is not always known in samples from cores, while micrographs may be oriented obliquely. (A).
Grain crushing in matrix-supported till, along a fissure which is part of a marble bed structure; see Fig. 6. This is a rare microstructure and only
observed occasionally; it points at deformation under dry conditions (see Hiemstra and van der Meer, 1997). Thin section R.974 from Wijnjewoude,
The Netherlands; polarised light; field of view, 18.0mm. (B). Grain crushing in clast-supported sands underneath a thrust plane in a Saalian
(penultimate glaciation) push moraine. Compare with Fig. 2A for the difference in crushing in the absence of fine-grained material. Thin section
Mi.1, Wilsum, Germany; polarised light; field of view 8.0mm. (C) Lineation—top left to bottom right, the arrows indicate some—interpreted as
shears, in tillite of Sirius Group of Tertiary age. This is a common microstructure. Thin section C.337 from core, Mt. Feather, McMurdo Dry
Valleys, Antarctica; plane light; field of view 18.0mm. (D) Grain lineations in diamict produced by grounded ice. Prominent grain lineations,
interpreted as shears, from top left to bottom right. This is a variation on the microstructure shown in Fig. 2C. Note that the same orientation is
present in the fine-grained material. This direction is also present in some of the finegrained inclusions. Thin section C.485, Cape Roberts core CRP-
1, Antarctica, Miocene; plane light; field of view 18.0mm. (E) Turbate (rotational) structure in till delineated by dark coloured material swirling
around a clast; note dark tail extending to the right above the clast. Turbate structures are the most common of all microstructures in glacial
sediments and adequately explain geochemical anomalies. Thin section Mi.694 from the Camcor River, Ireland, Midlandian; plane light; field of view
3.5mm. (F) Turbate structure, mixing of matrix-rich and clast-rich bands; coarse grains circle around dark matrix-rich pebbles. Thin section Mi.694,
Camcor River, Ireland, Midlandian till; plane light; field of view 9.7mm. (G, H) Pressure shadow; in plane light showing by two triangular patches of
matrix adjoining an elongate clast, while in polarised light it shows by low birefringence in triangular patches; highest birefringence in lines
delineating triangular patches. Clearly developed pressure shadows are rare as the thin section has to be in the right orientation, partly observed
pressure shadows are more common. Thin section R.676B, Amersfoort, The Netherlands, Saalian till; field of view 11.2mm. (I) Fissility
demonstrated by a horizontal fissure, the top of which is outlined by Fe and clay. Fissility is also under the microscope a common structure. Note
clean grains in the fissure. See text for discussion. Thin section R.866, Weiteveen, The Netherlands; Saalian till; plane light; field of view 7.0mm. (J)
Unistrial plasmic fabric, in this case the result of shearing, showing by elongate thin lines of high birefringence. Unistrial plasmic fabric is not so
common in diamictic materials, but more common in fine-grained material. Thin section Mi.25, Hainem.uhlen, Germany, tectonic lamination in
Elster till; polarised light; field of view 11.2mm. (K) Plasmic fabric development depends on the presence of clay (see text), as is clear from the
presence of birefringence in the matrix rich band and its absence in the sand bands. Thin section Mi.25, Hainem.uhlen, Germany, tectonic lamination
in Elster till; top to the left; polarised light; field of view 18.0mm.
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–1685 1663
strong indication that most basal till shows very littleevidence of sedimentation, and no evidence of theprocesses by which particles were set free from the ice.Known exceptions are basal melt-out tills still in contactwith ice (Lawson, 1979; Kr .uger and Kjaer, 2000) whichhave a restricted extent and a very low preservationpotential.
Within the known array of deformational micro-structures the plasmic fabric is the part of the till thatcarries the strongest genetic signal. The plasmic fabric isdefined as ‘birefringence models of the plasma, based onthe optical properties of the particles as well as theoptical properties caused by the orientation of particlesrelative to each other’ (based on Brewer, 1976, p. 305).Plasmic fabric is a microstructure that cannot be seen inthe field, not even with a strong handlens as it makes useof the optical properties of particles and requirespolarised light to be detectable. Individual clay particlesare too small to be observed with an ordinarypetrological microscope, and are examined as groups
(domains) that form characteristic patterns. Theseplasmic fabric patterns can be considered as consistingof pseudoparticles which are distinguished by theircombined birefringence. Imparted stresses (and theresulting strain) imposed on the deposit will reorientthe elongate clay particles (Morgenstern and Tchalenko,1967; Maltman, 1987; Hiemstra and Rijsdijk, in press)and different types and levels of stress will form differentplasmic fabrics. Although sharing microstructures theplasmic fabric pattern formed in mass-wasting (e.g. aflowtill, Lachniet et al., 2001; Menzies and Zaniewski,2003) is different from the plasmic fabric pattern formedunderneath an active glacier (Hiemstra, 2001; Hiemstraand Zaniewski, submitted), or from the plasmic fabricpattern formed by sedimentation (van der Meer andWarren, 1997). The subglacial plasmic fabric patternwill be illustrated and explained below.
Some types of plasmic fabric are also described by theterminology of structural geology. For instance, elon-gate plasmic fabrics (unistrial plasmic fabric) may show
ARTICLE IN PRESS
Fig. 2 (continued).
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–16851664
characteristics that are similar to discrete shears orcrenulation cleavage. However, plasmic fabric terms arepurely descriptive and do not relate to genesis, asindicated by the fact that not all unistrial plasmic fabricsare formed by shear. A pattern formed by unistrial aswell as masepic plasmic fabrics has been interpreted byvan der Wateren et al. (2000) and Kluiving et al. (1999)as a typical S/C fabric (van der Wateren et al., 2000, p.267) which means that material foliation (S-plane) is cutby shear band cleavage (C-planes).
Unfortunately the plasmic fabric is not alwaysdetectable. The main reason for this is that there is notenough clay in the material (Fig. 2K) and a clay-freematerial will not show a plasmic fabric. It is obviousthat there must be an—as yet unknown—minimumquantity of clay in a sediment for a recognisable plasmicfabric pattern to be developed irrespective of thestrength of the stress field, although clearly the natureof the plasmic fabric will be dependent on the type andmagnitude of that stress field. No matter how strong the
stress field, a pure sand will never show a plasmic fabric.Additionally, the plasmic fabric may be hidden behindother substances. For instance, a high primary ironcontent in the rocks that constitute the material, andiron and/or manganese staining will both obscure thebirefringence. This applies for instance to thin sectionsof tills from Breidamerkurj .okull and Myrdallsj .okull inIceland, which are derived from basalts and conse-quently strongly opaque in all grain sizes (Rose,personal observation; Simons, 1999). Finely dividedcarbonate particles (micrite) will undo the polarisingeffect of the microscope, thereby making any birefrin-gence undetectable. Etching the thin section by immer-sion in HCl to remove carbonates before putting on thecoverslip may help, but as the acid also affects the resinthe overall quality of the thin section deteriorates.
It is essential to consider all these factors whenevaluating till plasmic fabric. Comparison of thinsections of tills may reveal that birefringence is absentin some samples and present in others leading to the
ARTICLE IN PRESS
Fig. 2 (continued).
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–1685 1665
inference that reorientation of clay has not occurred inone group and hence that this till has not been subjectedto a stress field (Kluiving et al., 1999; Khatwa andTulaczyk, 2001). However, as this apparent lack ofbirefringence may be equally caused by lack of clay, Feand Mn staining, or micrite content, such an interpreta-tion could be erroneous and all micromorphologicalproperties of the till would need to be taken intoconsideration.
Fortunately, in addition to plasmic fabrics, there areother microstructures which do not depend on polarisedlight for their detection (Figs. 1 and 2). For instance,rotational structures (Figs. 2E and F), which are visiblein plane polarised light, are attributed to deformationalprocesses (van der Meer, 1993b, 1997a). Nevertheless, asa consequence of their process of formation suchstructures are often associated with plasmic fabrics(van der Meer, 1993b), especially with a skelsepicplasmic fabric which is ascribed to rotational processes.Thus it is possible to determine whether a till has beensubjected to stresses other than those associated with(re-) sedimentation, without the need to detect birefrin-gence.
It is also essential to recognise that plasmic fabrics canbe created by processes other than subglacial deforma-tion. For instance, mineralogical properties such as theswelling of clays by repeated incorporation and loss ofwater in the crystal lattice will form a skelsepic plasmicfabric (Dalrymple and Jim, 1984). For this reason(amongst many others) it is important to sample wellbelow the zone of soil processes as this is the zone wheremoisture variability is highest. The low hydraulicconductivity of the matrix of most glacial diamictonsand consequently the low variability in moisture contentimplies that regular swelling of clays is unlikely to be thecause of a plasmic fabric in till.
Caution should also be taken when comparing thestrength of plasmic fabrics in tills from differentlocalities. To prevent the introduction of variables otherthan strength, this should always be done at the samemagnification and the same amount of illumination.Independent, quantitative comparison of plasmic fabricstrength is now possible by the use of image analysis(Zaniewski, 2001).
2.3. Scope of database and conceptual rational of this
research
Over a period of >35 yr we have studied a largenumber of thin sections of tills from many differentlocalities and environments. These range geographicallyfrom the Arctic to the Antarctic, stratigraphically from‘just-formed’ material, to Precambrian (Menzies,2000a). Thus, the model presented below is not justbased on the interpretation of inferred properties of ‘old’tills, but includes observations from underneath active
glaciers. The visible appearance of tills investigatedranges from structured to massive and, as far as we areable to determine, the samples we have studied cover thefull range of glacial environments. Our experience alsoincludes extensive studies of thin sections of non-glacialmaterials including marine, fluviatile, aeolian, periglacialand mass movement (van der Meer, 1992, 1996a; van derMeer et al., 1996; Rose et al., 2000; Menzies andZaniewski, 2003), and this has contributed significantlyto our ability to determine whether structures, orcombinations of structures, are subglacial or otherwise.In addition to our own studies, there has been muchresearch and many publications on thin sections fromnon-glacial environments. Of particular note is thework on all varieties of slope deposits by Bertran andTexier (1999) who demonstrate that different slopeprocesses are characterised by different suites of micro-structures (see also Hiemstra and Zaniewski, submitted).The microstructures depicted in Fig. 1 are based on ourown extensive studies, details of which can be found inTable 2.
The most striking property revealed by working withthin sections under a petrological microscope is thatthere is no such thing as a massive, structureless till; alltills possess microstructures. Almost of equal impor-tance is the fact that, apart from occasional WESs, themicrostructures in basal tills are typical of deforma-tional processes. Hitherto, microstructures that mayindicate how these sediments were released from glacierice have not been recognised. The consequence of theseobservations is that if all we can see in thin sections ofbasal tills are microstructures formed after deposition,then it is impossible to detect the process of deposition.If we accept that this is the case then it must be acceptedthat we cannot detect depositional processes such aslodgement or melt-out. Samples of (unconfined) supra-glacial melt-out tills show that these have a specific set ofmicrostructures, which are not recognised in subglacialtills (Simons, 1999; unpublished data). Thus, we cannotdetect subglacial melt-out tills microscopically and if wecannot find signs of either lodgement or melt-outprocesses by macro- or microscopic studies (see above)the logical conclusion is that there are no lodgement tillsor subglacial melt-out tills. To extend this line ofreasoning, if there are no lodgement or subglacialmelt-out tills, and all we can see is deformation, weare led to the conclusion that the material produceddirectly by glaciers constitutes a deforming bed. Giventhe range of samples that have been used in our studies(Table 2) the conclusion must be that all subglacial tillsfrom all places and all ages reflect deforming beds. Asindicated above the only exception to this rule is theoccurrence of subglacial melt-out tills in contact withglacier ice, which are of limited extent. To our knowl-edge there are no uncontested Pleistocene or oldersubglacial melt-out tills.
ARTICLE IN PRESSJ.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–16851666
We wish to address the issue that the selection ofsamples used in this study is biased. Although thismay (unintentionally) be the case for samples collectedby ourselves, many of the samples used in our re-search have been collected for us by others and this isreported in the acknowledgements at the end of thepaper. Furthermore, samples taken from coresallow little choice, and all that one can use is what isavailable. Finally, the collection is available for study,should others so wish, and samples from thecollections have been used and studied in a numberof international workshops (Amsterdam, 1993; Brock,St. Catharines, Canada 1997; Leeds, 1998; London,1998, 2000, 2002; T .ubingen, 2001; more will follow).In the meantime others have independently cor-roborated our findings of microstructures (e.g., Evans,
1998; Fuller and Murray, 2000, 2002; Lachniet et al.,2001).
This paper seeks to elaborate on the concept that allsubglacial tills are part of a deforming glacier bed. It isemphasised that our treatise is based on microstructuresobserved in thin section combined with our fieldexperience of glacial sediments and contemporaryglacial environments. This paper is not a discussionbased on the perceived character of tills, i.e. whether tillalways behaves as a Coulomb material or not.
3. The deforming glacier bed
Experiments underneath glaciers have demonstratedthe existence of deforming beds (see for instance papers
ARTICLE IN PRESS
Table 2
Overview of thin section collection on which this study is based
Area No. of thin
sections
Citation
The Netherlands 157 van der Meer (1987a,b,1993b, 1996b, 1997a); van der Meer and Laban (1990); van der Meer et al. (1983,
1985); Rappol (1983); Rappol et al., 1989; Hiemstra and van der Meer, 1997; van Beek, 1990; van Ginkel
(1991), Carr (1998)
Switzerland 52 van der Meer (1980, 1982)
UK 39 Carr (1998, 1999), Carr et al. (2000), Rijsdijk (2000), Menzies and van der Meer (1998), Croot et al. (1996)
Ireland 123 van der Meer et al., (1994a, b), Rijsdijk (2000), Verbers (1989), Bloetjes and van der Meer (unpublished data)
Sweden 18 Lagerlund and van der Meer (1990, and unpublished data)
Spitsbergen 24 Boulton et al. (1996, 1999), van der Meer (1997a), van der Meer and Solheim (unpublished data)
Germany 122 Hiemstra (1995), van der Meer et al. (submitted), Menzies and Maltman (1992), Menzies (2000b)
Argentina 4 Unpublished
Spain 23 Bordonau (1992), Bordonau and van der Meer (1994)
Greenland 2 Unpublished
Iceland 39 Simons (1999), van der Meer and Solheim (unpublished data)
Canada 145 Menzies (1986, 1991, 2002)
USA 66 Menzies and Woodward (1993), Menzies et al. (1997), Menzies and Zaniewski (2003)
Antarctic 170 van der Meer (2000), van der Meer and Hiemstra (1998), van der Meer et al. (1994, 1998), Hiemstra (1999,
2001), Zaniewski (1997), Stroeven et al. (1999, 2002), Wilson et al. (2001), Lloyd Davies et al. (unpublished
data)
Denmark 12 Unpublished
Glacial; non-till 176 van der Meer and Warren (1997), van der Meer and Solheim (unpublished data), van der Meer et al. (1992)
Non-glacial van der Meer et al. (1996) (fluviatile), van der Meer (1992, 1993a, 1995, 1996a), Menzies and Zaniewski
(2003)
Total 1274
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–1685 1667
in Hart and Rose, 2001) and that the velocity of the bedsis variable. The deforming bed is by now a wellestablished concept, its existence is no longer challenged(Murray, 1997). Today, only its thickness, its character-istics and how widespread it is and was is challenged(Clark, 1991; Ronnert, 1992; Hart, 1995; Boulton,1996a, b; Boulton et al., 2001; Hoffmann and Piotrows-ki, 2001; Piotrowski et al., 2001, 2002). Micromorphol-ogy is a relatively new technique when it comes tostudying deforming beds (van der Meer, 1993b, 1997a;Menzies, 1998, 2000b). However, it is no longer aquestion of whether micromorphology is able to identifydiagnostic fabric features (Boulton et al., 2001, p. 9).That ability has already been demonstrated (see forinstance Hiemstra and van der Meer, 1997and furtherreferences in Table 2).
Investigation of the deforming glacier bed requires anunderstanding of the factors that influence its velocity.This clearly requires knowledge of the properties of boththe glacier and the bed, especially glacier velocity, whichis linked with the temperature of the ice, and thecomposition and water content of the bed material.
3.1. The effects of glacier movement
In subglacial experiments it has been established thatup to 70% of the forward movement of the glacier canoccur in the bed; figures of this order have been recordedin the fast moving temperate glaciers of Iceland(Boulton, 1987). However in the slow moving, coldcontinental glaciers of the Tien Shan of northwesternTibet Plateau (Echelmeyer and Wang Zhongxiang,1987), bed movement accounted for only a minutefraction of the forward movement of the glacier.Numerous other examples replicate these findings, and
overall these experiments imply that glacier velocity canbe taken as a proxy for the velocity of movement in thesedimentary bed beneath the glacier (Table 3). For thesake of our argument it does not matter whether or notthere is perfect coupling between ice and bed, only thatthere is a coupling at this interface.
3.2. The effects of water within the bed
Water content of the deforming bed is the majorfactor controlling deformation, but a property that is farfrom simple both conceptually and with respect to insitu measurement. The following questions need an-swers: (i) How does water get to the till? (ii) How iswater distributed within the till? (iii) How does watermove through the till? (iv) What causes variability ofwater content (and consequently effective pressure) overtime and space and at different scales? Investigation ofthe possible impact of water content in subglacial tills byMenzies (1989) has resulted in the ‘H’ and ‘M’ bedsconcept.
Water content depends on ice temperature, both atthe glacier surface and at the base. High production ofmeltwater at the glacier surface that subsequentlyreaches the bed, may dissipate into the bed dependingon the presence of channels beneath the glacier, thepressure applied to the bed material, and the texture andstructure of the bed. Where well developed channelsystems exist at the base of the glacier, large areas of thebed will not be in contact with this source of water andthese areas will receive their water mainly from basalmelting. As these channels are known to shift over time,this property alone will mean that at any given timethere to be spatial or vertical differences in the watercontent in the glacier bed, as demonstrated by Boulton
ARTICLE IN PRESS
Table 3
Deformable bed data: basal sediment velocity (us), thickness (hs) and calculated strain rates (es) (from Menzies, 2002)
aEchelmeyer and Wang Zhongxiang (1987).bEngelhardt et al. (1978).cBoulton, (1979).dNow Whillans Ice Stream.eAlley et al. (1997).fEngelhardt and Kamb (1998).gTruffer et al. (2000).hBrown et al. (1987).
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–16851668
et al. (2001) in Iceland and by Fischer and Clarke (2001)in Canada.
The pressure (the combined pressure of ice over-burden and water pressure) applied to the glacier bedwill help in distributing water at and into the glacier bed.Studies in Iceland (van der Meer et al., 1999) havedemonstrated that subglacial water, which could notescape because the glacier toe was frozen to the base,caused the formation of a huge 3D array of water-escapeconduits which allowed the pressurised water to escapefrom up-glacier locations and created structures thatemanated from the base of the till. In order for thisprocess to take place the water must first have beenforced into and through the till bed (till bed andchannels in Alley, 1989). In this case the forcedevacuation of water was so effective that the glacierbed dried out and was subsequently fractured by theglacier (van der Meer et al., 1999), before resumingdeforming mode once again.
Careful study of sand inclusions in tills (Carruthers,1953; Eyles et al., 1982; Menzies, 1991) and of the base
of tills overlying sands and gravels (van der Meer, 1979,1980) show the widespread distribution of the effects ofthis process, indicating that the features are not limitedto the Icelandic example. These structures have a widerange of sizes, reflecting the range of scales of theevacuation process. The presence of silt/sand filled,WESs at the centimeter scale, indicate formation by thesame process, but at a much smaller scale (Fig. 3). Alongsimilar lines, Fuller and Murray (2002) ascribed smallsand lenses to canal or pipe flow in soft sediments. Onthe other hand hydro-fracturing as a result of high waterrecharge from the glacier does not necessarily lead tosediment intruding into fractures (Klint, 2001, 2003),but can still lead to systematic vertical and obliquefracturing of the till bed.
An example of how drainage affects the behaviour ofthe till can be seen in Fig. 4 stemming from the study inIceland referred to above. The creation of preferentialpathways for water by the development of WESs has ledto an immediate thickening of the till. At one point thishas tripled the thickness of the till (Fig. 4, at 9m) over
ARTICLE IN PRESS
Fig. 3. La Tuffi"ere, S. of Fribourg, Switzerland. W .urm (Last Glaciation) glacigenic sequence of tills overlying wastage phase glaciofluvial gravels.
(A) The base of the till overlying the gravels shows small WESs. In the field these are only visible as whisps of silt. (B) Thin section R.250 from the
base of the exposure depicted in A shows a micro-water-escape structure with associated grain separation. In subglacial tills this is one of the few
depositional microstructures and it relates to the role of water at this scale. Plane light, field of view 5.6mm.
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–1685 1669
the width of the WES, while at 19m it developed a morecomplex wedge-shaped structure that reaches seventimes the usual thickness of the till. This is only one ofa number of such observations, including more tillwedges in the Iceland study (Kilfeather, 2002). Similarcomplex lower boundary structures, including tillwedges have been described from other areas as well(van der Meer, 1979, 1980). For a discussion of howchanges in water content result in changes in tillthickness see Menzies (1989) and also Kjaer et al. (2003).
Water content also varies with the dilatancy of till.Anyone with experience of temperate glaciers is familiarwith water-saturated diamictons around and beneaththe glacier margin that liquefy on touch (Fig. 5). In thisdilated state large quantities of water are present in thetill (Murray and Dowdeswell, 1992), and it is the waterthat keeps apart (dilates) the sediment particles. Thislayer is variable in extent and thickness (van der Meer,1997b), and unless it extends to lithified bedrock, itoverlies dryer till, demonstrating that at any one time atill body beneath a glacier may demonstrate a suite ofcompletely different conditions of water content.
Another factor that influences the water content of tillis the presence of well developed, open, interconnectedfissures (Klint, 2001, 2003). These enhance the move-
ment of water throughout the till, compared to move-ment through intergranular, and not necessarilyinterconnected, pores. Evidence for these interconnectedfissures in tills are fissility (Fig. 2I) and marble-bedstructure (van der Meer, 1993b; Hiemstra and van derMeer, 1997; Menzies and Zaniewski, 2003). Fissility isthe result of shearing—a direct product of a deformingbed. There is no reason to assume that fissility observedin a Pleistocene till would retain the dimensionsoriginally acquired within the deforming bed. Pressureswithin the subglacial system would change their state,either keeping them closed by overburden pressure or, ifthey were invaded by meltwater, forcing them open.Clean quartz grains inside fissures (Rappol, 1983, Fig.88) indicate, at least temporary periods of current flow.Exposures of fissile tills show that these structures arestill preferential pathways for water and dissolvedmatter.
Marble-bed structure, where the deposit consists of anumber of spherical aggregates that behave like ballbearings (van der Meer, 1993b) has been observed anumber of times in thin sections (in the study collection,Table 2) of tills of different localities. This structureconsists of small fractures which, in the best documentedcase (Hiemstra and van der Meer, 1997), evolve fromangular elements at depth in the till profile to well-rounded structures closer to the glacier base (Fig. 6).This gradual change reflects the increase in deformationtowards the glacier contact. In the case documented byHiemstra and van der Meer, this structure revealed thecomplications of water movement through, and watercontent of a till bed over time. The marble-bed structureand the associated in situ crushed grains were inter-preted as resulting from pulsed water movements withalternating wet and dry conditions.
Localised, temporary freezing may occur in the bed,which should lead to further complications. We do notwant to pursue this further in this paper.
3.3. The effects of bed composition
Clay content is a critical factor in the behaviour of theglacier bed, and in many glaciotectonic structures claybeds or concentrations form the plane of d!ecollement,along which movement is concentrated (van der
ARTICLE IN PRESS
Fig. 4. Change in till thickness over WESs (see text); forefield of Sl!ettj .okull, Iceland.
Fig. 5. Portal of Turtmanngletscher, Valais, Switzerland. Clean ice
overlies corrugated (macro slickensided) till which demonstrates the
variable water content and related behaviour in the glacier bed. Solid
arrows point at wet or flowing till, while open arrows point at fissures.
The till could not be samples for thin sectioning as it liquefied on
touch. Glacier movement left to right.
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–16851670
Wateren, 2002). This is caused by water content in theclay, or electrochemically bonded water and the smoothtabular form of the clay particles which become alignedwith parallel surfaces. In matrix-rich sediments silt-sizeparticles also facilitate movement. Part of the rotationalstructures that occur so widely in tills consist of siltparticles that are orientated parallel to the surface oflarger particles. Clays in deformed glacier beds are thepre-requisite for the development of a skelsepic plasmicfabric, which is formed by clay particles lining uparound and adhering to a larger particle. Clearly, ease ofdeformation is directly related to an increase in claycontent (Menzies, 1989, 2000b, Fig. 1), and thisrelationship is reinforced by the water-holding capacityof the clay size material.
The effect of a change in clay content and composi-tion upon bed deformation processes can be seen inFig. 7. This figure relates to widespread evidence from
northern Germany and The Netherlands where twodifferent till types occur side by side (Meyer, 1987;Rappol, 1987) with a sharp contact. On one side is aclayey-silty till (up to 45% clay), with a high proportionof smectite, on the other is a silty-sandy till (up to 20%clay) without smectite (Rappol et al., 1989). Erratics inboth tills have been transported considerable distancesfrom southern and central Scandinavia, yet there is nosign of mixing. The tills differ widely in every respect:clast provenance, carbonate content, as well as theparticle size properties mentioned above. It is suggestedthat the reason why these tills do not mix, despite theirlong transport distances, is that their clay (andcarbonate; see below) content determines their porewater pressure, cohesion and inherent strength, ensuringthat they did behave independent of one another despitethe fact that they exist side by side, and are both part ofthe deforming glacier bed. In laboratory experiments
ARTICLE IN PRESS
Fig. 6. Wijnjewoude, The Netherlands, till profile (Saalian) with position of thin section samples (Hiemstra and van der Meer, 1997). X-rays of two
of the samples clearly show the outline of fractures in the till (in white). This is a clear demonstration of the development of the marble bed structure
from angular at depth to more rounded higher up. It is a structure that has been observed a number of times in subglacial tills from different localities
and places. Denser pebbles show up dark. The till symbols in the diagram at left represent the descriptions to their right.
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–1685 1671
Iverson et al. (1996) demonstrated the mixing of tills in adeforming bed, however, the field evidence presentedabove demonstrates that it does not always occur.
The capacity of clay to absorb water means that waterwill penetrate in between clay particles. Given a highenough water pressure this movement contributes to thedispersion of clay particles leading to the formation ofilluviated clay cutans in subglacial tills (Menzies, 2000b)as in perpetually submerged riverbeds (van der Meer,1993a). Additionally, clay mineralogy plays a rolethrough the capacity of certain clays to incorporatewater into their crystal lattice. Thus, the presence ofswelling clays like smectites will contribute to a higherwater content. The presence of swelling clays will alsomean that the water content in a deforming bed willchange less rapidly as this water is more difficult toremove than absorbed water. As stated above, thewetting and drying of swelling clays will affect thepresence and strength of the plasmic fabric, althoughthis process may be most effective after deglaciationwhen the till can be dried more effectively.
An as-yet under evaluated parameter that maycontribute to the mobility of the deforming bed issecondary carbonate (Fig. 8) within the material. It isknown that carbonates are precipitated directly ontobedrock underneath glaciers, both at the present day
(Hallet, 1976, 1979; Hallet et al., 1978; Fairchild et al.,1994) and during the Pleistocene (Bjærke and Dypvik,1977). Despite their often stromatolitic nature and theoccasional presence of organic components like pollen,the subglacial origin of these carbonates is not doubtedand these should be differentiated from allogenic
ARTICLE IN PRESS
Fig. 7. Urk, The Netherlands; boundary between two widely different Saalian tills (see text). (A) View of the whole thin section, arrows point at the
boundary. (B) Sharp, undulating boundary between two till beds. Note the lack of mixing between two, compositionally completely different till
beds. Thin section O.817, Urk, The Netherlands; plane light; field of view 18.0mm.
Fig. 8. Secondary carbonates (greyish colours) following the till
structure, in this case a marble bed structure. Thin section O.818,
Urk, The Netherlands, Saalian till; plane light; field of view 18.0mm.
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–16851672
carbonates formed since deglacierisation, which are notrelevant to this paper. Authigenic carbonate is likely tohave had a stiffening effect on the material andconsequently is likely to have influenced the deformingbed.
There is widespread occurrence of secondary carbo-nates in tills (van der Meer, 1982; Menzies and Brand,submitted). Concretions of carbonates may occurthroughout the till, but are regularly seen at the contactwith underlying sands and gravels. In the till thesecondary carbonates form coatings along fissures(Fig. 8) as well as cemented patches in the till mass. Itis always assumed that in Pleistocene tills these wide-spread secondary carbonates are caused by groundwateractivity of postglacial age, despite till matrices having alow hydraulic conductivity. However, recent work onstained thin sections of till from within a drumlin in NewYork State has established that secondary carbonateswere deposited within a till penecontemporaneouslyduring a readvance across deltaic sands and gravels(Menzies and Brand, submitted; see also discussion inFairchild et al., 1994).
Further changes in mobility of the glacier bed may beinduced by changes in the substrate over which theglacier moves. It has long been known (Geikie, 1863)that tills change their lithological properties includinggrain size as the glacier and the glacier bed move acrossdifferent lithologies. Change of composition caused bymovement from a fine grained onto a coarse grainedsubstrate will affect subglacial drainage, porewaterpressure, effective stress levels and thus sedimentmobility and consequently glacial behaviour (van denBerg and Beets, 1987). However the same change inmobility may come about because of glacial erosion orglaciotectonics. Glacial erosion may change the glacierbed abruptly from one grain size to another, with theconsequent effect on drainage and behaviour. Glacio-tectonics may change the bed in the most dramaticfashion with the removal or inclusion of rafts or beds ofsediment, thereby introducing a different lithology intothe subglacial domain.
Taking all factors together we must conclude thatsubglacial conditions, and consequently subglacialbehaviour, must have been extremely variable in bothtime and space (Menzies, 1989; Menzies and Shilts,2002); a conclusion that has been reached on differentgrounds by others (Piotrowski and Kraus, 1997;Knight, 2002) though leading to different models.Depending on the uniformity of the bed and ofmeltwater availability zones of the subglacial domainmay have been uniform, whereas variations in meltwaterinput could produce variations in bed mobility on verysmall scale (Boulton et al., 2001; Fischer and Clarke,2001). A texturally highly variable bed will, even underuniform meltwater availability, lead to a highly variablebed mobility.
4. A deformation model—properties and process
4.1. The local scale
When combining our observations on the widespreadpresence of a deforming bed with the factors that affectthe mobility of the bed, we arrive at the followingdescription of the characteristics of the deforming bed inspace and time. This description begins with Fig. 10Aderived from Alley (1989). This diagram shows howdeformation changes from sliding and ploughing on theice-bed contact, through pervasive shearing to discreteshearing at the base of the deforming bed. It should benoted that this diagram has no scale, neither horizontal,nor vertical; we will come back to this later. For thepurposes of our thesis we assume the till is diamicticwithout textural or structural discontinuities and thatthe stress is applied continuously. Using these assump-tions we can now deduce how two of the mostprominent factors that influence bed mobility: waterand clay content, affect this diagram (Fig. 10B).Although the deforming bed is a complex temporaland spatial rheological system, in which water and claycontent act in relation to stress levels we have for thesake of simplicity to treat these factors as if they areindependent of each other.
When water content is low (Fig. 10Bi), sliding at thebed contact will be low, the zone affected by deforma-tion will be thin and will be characterised by discreteshearing. The shears will delineate a 3D pattern which isangular at depth with increasing roundness towards theglacier sole (Fig. 6; Hiemstra and van der Meer, 1997). Itis proposed that this pattern is developed by the gradualupward progression of (sub-)horizontal shears, followedby Riedel and anti-Riedel shears. Further movementalong these shears is considered to lead to rounding asprotuberances are broken off, producing a marble-bedstructure (van der Meer, 1993b) upon which, in effect,the glacier is being carried along ‘on a bed of ballbearings’. This proposition is substantiated by the in situproduction of crushed sand grains, for which low watercontent is a prerequisite (Hiemstra and van der Meer,1997). Furthermore, it has been demonstrated that in-between the shears deformation is non-existent (Men-zies, 2000b). Microscopically this stage can be recog-nised by the outlines of the marble-bed structure and thein situ crushed grains (Fig. 2A). The latter should not beseen as the result of sand ‘grain bridges’ which ontheoretical grounds were assumed by Hooke andIverson (1995), but which have never been observed inany of our thin sections. Under the microscope skeletongrains in tills are not in contact with each other, there isalways a minute, but still buffering layer of matrix inbetween.
With increasing water content, sliding along the bedwill become more important, but the marble-bed
ARTICLE IN PRESSJ.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–1685 1673
structure can no longer develop as individual particlesbecome more mobile. The deforming bed will initiallydecrease in thickness as a function of attenuationassociated with higher mobility. In well sorted (drained)sediments pervasive shearing will become dominant andthe zone of faulting will become thinner. The zone ofpervasive shearing will change from the marble-bedconfiguration, to one that consists of rotational struc-tures (Figs. 2E and F), in which small particles swirlaround larger particles and become oriented parallel tothem. This shows as galaxy structures, rotationalstructures and a skel- to lattisepic plasmic fabric(Fig. 1). We can also expect the development of shears,initially as short shears which show either as short grainlineations (Figs. 2C and D), or as varieties of masepicplasmic fabric. Additionally, we can also expect thedevelopment of pressure shadows. This is supported byobservations of Fuller and Murray (2002) who studied athin deforming bed related to a documented surge. Theydescribe many of the above microstructures in a thin, upto 20 cm thick deforming bed at a high water content.Note that within the same thin sections they alsodescribed units and patches of well-sorted fine sands,which they interpreted as the result of running water incanals and pipes in soft sediment.
Further increase in water content will see thedevelopment of unistrial plasmic fabrics (Fig. 2J), asshearing intensifies individual shears become longer.This microstructural development does not persistindefinitely. As described above, the dilatant uppermostpart of a till bed immediately beneath a glacier has beenobserved to contain so much water that the materialliquefies when a stress is applied, and consequently theparticles start to behave independent of each other. Inthis state, which is also developed near the margin oftidewater glaciers (van der Meer and Hiemstra, 1998;van der Meer, 2000), the dilated state of the sedimentwill only allow the development of similar, but isolatedstructures. Thus we see first an increase in intensity(both in number and size) of microstructures with anincrease in water content, followed by the microstruc-tures becoming single and isolated with a furtherincrease in water content. The latter is a consequenceof the water separating sediment particles and prevent-ing their interaction.
4.2. Role of clay in the development of deformational
structures
If clay content is taken as the only variable, a differentarray of deformation structures will be developed(Fig. 10Bii). With a low clay content, we expect shearingto be the main process, and detection of this shearingwill depend on the presence of strain markers. Shearingmay be visible as discrete shears if silts are present, andintergranular movement can be expected in sands.
Evidence of intergranular deformation has been ob-served in coarse sand directly underneath and in contactwith a major thrust in a push moraine (Kluiving, 1989).A thin section of these sands showed silt flakes to havebeen produced by tectonic attrition (Fig. 2B). Theseobservations differ from those on crushed grains formedunder conditions of a low water content (see above) inthat this material is clast supported without a matrix, sothat there are only open pores. However, because of alack of strain markers, sands may appear to beunaffected and massive, and deformation may only berecognised by careful micromorphological analysis ordeformed bedding structures.
With increasing clay content, deformation will occurmore easily, rotational structures, plasmic fabrics andpressure shadows (Figs. 2G and H) will be formed andbecome widespread. Although the likely expectation isthat increased clay content would lead to pervasivedeformation only with restricted linear shearing atdepth, the fact is that as clay content increases, thehydraulic conductivity of the as-yet unstructured sedi-ment decreases. With decreasing hydraulic conductivity,and no change in pressure, it will become more andmore difficult for the water to move through thesediment. Consequently we can expect the water toexploit any preferential pathway and, in the absence ofthe latter, develop new pathways. This has the potentialto lead to the development of interconnected shearplanes, and this process can be expected to speed up aswater pressure increases instead of remaining constant.The shear planes developed in this way would at thesame time accommodate deformation of the sedimentand evacuation of water. The end product of thisprocess is a fissile till. The lenticular elements betweenfissures (shears) have been observed to vary fromcentimetres to metres in size and this spacing mostlikely reflects water pressure and clay content duringformation (Fig. 9). It has been observed that thelenticular units between fissures may show only a minoramount of microstructural development (Rappol, 1983).When present, microstructures in between the planesof weakness may have evolved before the fissile structurecame into being, reflecting a two-step development. Inthis case the microstructures have been preservedbecause further deformation was concentrated alongthe shears. Thus, an increase in clay content doesnot automatically lead to deformation throughoutthe sediment body, but may lead only to localiseddeformation.
In reality it will not be so easy to separate changes inwater content from changes in clay content (Table 4)because the effects are similar and because of theintricate relationships between the two. As outlined inSection 3.3, increasing the clay content automaticallyleads to increasing the moisture content. Because smallpores form a stronger bond between sediment and water
ARTICLE IN PRESSJ.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–16851674
than large pores, water is not evacuated rapidly. In asubglacial environment with a continuous supply ofwater we can expect a clayey bed to be more mobile (butnot necessarily to the same depth; Boulton, 1996b) thanan adjacent more sandy bed, all other parameters beingequal. Given the great range of combinations of grainsize and water content we can expect the deforming bedto be extremely variable laterally. Depending on howquickly both variables can change, we may expect lateralchanges in a range of tens of centimetres (Menzies et al.,
1997), to thousands of metres. Because this is easier todepict in a variation of ‘H’, ‘Q’ and ‘M’ beds (Menzies,1989), Fig. 10 changes from the vertically variable Alleydiagram (Fig. 10A) into the horizontally variable bedsdiagram (Fig. 10C).
4.3. Spatial and temporal variability
Fig. 10D tries to visualise such a spatially variabledeforming bed. In this figure we have simplified the
ARTICLE IN PRESS
Fig. 9. The spacing of fissility in till most likely reflects water pressure and clay content during formation. (A) Small-scale fissility delineated by open
fissures and Fe staining in Drenthe till (Penultimate Glaciation). Neu Wulmstorf, N. Germany. (B) Large-scale fissility in Midlandian (Last
Glaciation) till from Ringaskiddy, S. Ireland (knife for scale).
Table 4
Relation between microstructures and water and clay content, assuming constant pressure
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–1685 1675
ARTICLE IN PRESS
Fig. 10. Diagram depicting till, the deforming bed. (A) So called Alley diagram (after Alley, 1989) showing the main types of subglacial deformation.
(Bi) Changes in style and intensity of deformation with increasing water content. (Bii) Changes in style and intensity of deformation with increasing
clay content. (C) Translation of variable styles and intensity in deformation in ‘H’, ‘Q’ and ‘M’ beds (after Menzies, 1989). (D) Theoretical
distribution of ‘H’, ‘Q’ and ‘M’ beds under a glacier, showing the spatial variability of the deforming bed. (E) Mapping the distribution of ‘H’, ‘Q’
and ‘M’ beds over time shows the temporal variability of the deforming bed ((i) and (ii) are two steps in time).
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–16851676
ARTICLE IN PRESS
M
Q HM M
Q QQ M
Q MH Q
Q HM H
H
H
H
M
M
Q
Q
M
H
H
M
Q
Q
time
ice
ice
spatial variability
temporal variability
(D)
(E)
(Ei)
(Eii)
Fig. 10 (continued).
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–1685 1677
depiction of the deforming bed by representing it aseither ‘H’, ‘Q’ and ‘M’ beds. The indication ‘Q’ or ‘M’represents the whole range of deformational processes atthat locality. It should be emphasised that this is atheoretical model.
To create a sense of variability in the deforming bed,we have indicated the presence of either an ‘H’, ‘Q’ or‘M’ bed at the grid intersections. Note that there is stillno scale to the figure, which we will address later. Thediagram depicts a flat contact between glacier and ice,and a deforming bed split into ‘H’, ‘Q’ or ‘M’ classes.We envisage this deforming bed as a matrix of variationsin composition, water content, shear strength (oreffective stress) and applied shear stress levels interact-ing with variations in thickness and velocity. Thevariations in velocity mean that flow lines across thissurface will not be parallel, but will show a complicatedpattern as well (Kjaer et al., 2003). The usual way toaccommodate differences in velocity is to changethickness. In combination with a steady flux thissuggests that locally the fastest flowing parts of thebed would also be the shallowest. In reality we canassume that accommodation may occur either through adifferential flux of material, or through adaptation ofthe bed form. The latter means that the contact betweenice and bed cannot remain horizontal but will developundulations which will be the start of bedforms, eitherflutes or drumlins (Rose and Letzer, 1977; Boulton,1987; Menzies, 1987, 1989; Rose, 1987). These wouldthen be the result of inherent bed instability. Theseinstabilities explain the existence of drumlinised tillwhere there is no apparent point of initiation for theselandforms (Kjaer et al., 2003). It can similarly explainthe formation of flutes that lack a starting boulder attheir up-glacier end (Menzies and Shilts, 2002). Thiswould explain the universal development of flutes anddrumlins without a point of initiation (Rose and Letzer,1977; Rose, 1987, 1989a, b; Hindmarsh, 1998, 1999;Menzies and Shilts, 2002; Kjaer et al., 2003). Drumlinswith a stratified sand and gravel core, with the coreacting as an aquifer node causing an increase in tillstrength and consequently a reduction in mobility,readily fit into this scheme. Based on this relationbetween bed mobility, water and clay content, it isreasonable to expect a textural difference between till indrumlins and till in interdrumlin depressions. Suchtextural differences would require a large database forinvestigation and consequently this effect has, hitherto,not been effectively demonstrated. However, new datafrom County Clare, Ireland, appear to demonstrate sucha systematic textural difference (Oscar Bloetjes, pers.comm.).
The situation depicted in Fig. 10D can also berepresented in a map (Fig. 10Ei). In this figure thedifferent types of beds have been depicted. The linesdelineating the different types of bed have been drawn
out in the direction of ice flow. Because bed conditionswill not remain the same over time in any one position,this map will change with time (Truffer et al., 2001). Inparticular, because of seasonal variations in watercontent, bed conditions will not remain the same, eventhough texture might not change. Alternatively, condi-tions may change because of changes in the texture ofthe bed due to erosion of new lithologies. This can takeplace at the base of the till without the ice being in directcontact with the new lithology because rotationalstructures in a deforming bed may scavenge the sub-till sediments (van der Meer, 1997a), thereby exposinganother material. The mobility of particles in a bedconsisting of rotational structures means that thecomposition of the whole deforming bed will change,downstream of a change in bed texture. This will onlyapply if the deforming bed encompasses the fullthickness of the till bed. Indeed, changes in moisturecontent of tills beneath ice mean that the thickness of thedeforming layer will also change. This will mean, inconsequence, that scavenging at the base may betemporary. Given enough time, the map depicted inFig. 10Ei may change into the situation depicted inFig. 10Eii. It should be emphasised that both aremoments in time and deformation will only ceasefollowing deglaciation.
Depending on the time involved and the intensity ofdeformation, a till may acquire a massive appearancedevoid of any macroscopic structures. Changes inglacier movement and the intensity of bed deformation(velocity and/or thickness) do not mean that the visibleappearance of the till bed will change; it may simplyremain massive and structureless. However, it isimportant to stress that the latter only applies to themacroscale and the record of change will be recorded inthe microstructures.
Nevertheless, the appearance of a till may change ifthe texture, water content, or scavenging activitychanges, resulting in, respectively, the development ordestruction of fissility or the incorporation of rafts ofsediment into the till. This variability of the deformingbed implies that a till may change from a ‘mature’,massive, to an ‘immature’ structure due to the incor-poration of allochthonous material, indicating thatobserving a ‘massive’ till is not necessarily the same asobserving the end member, as well as emphasising theunsuitability of these anthropomorphic adjectives (Men-zies and Zaniewski, 2003). Finally, it should be realisedthat some structures in the deforming glacier bed are notjust the passive result of subglacial conditions, but dothemselves affect these subglacial conditions, i.e. theyare back-coupling. As such, we can point at glaciotec-tonic fractures which are themselves related to waterrecharge from the glacier and drainage conditions belowthe glacier (Klint, 2003). After their formation suchfractures will strongly influence water movement
ARTICLE IN PRESSJ.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–16851678
through the till bed, adding to the complexity of thedeforming glacier bed in time and space.
5. Thickness of the deforming glacier bed
Following the above the horizontal properties of thedeforming bed are reasonably well understood and canvary between centimetres and thousands of metres andthe properties of the till may vary from extremelyhomogeneous to highly variable. This can be substan-tiated by all studies of till beds: some are extremelyhomogeneous over long distances, while others show ahigh variability. There is clearly less information andunderstanding about the thickness of the deforming bedand this topic has been the subject of discussion(Boulton, 1987; Menzies, 1989; Hart, 1998; Boultonet al., 2001) and recent work on temperate glaciers(Humphrey et al., 1993; Hooke et al., 1997; Fuller andMurray, 2000, 2002; Truffer et al., 2000). The deformingbed at Breidamerkurj .okull had a thickness of only70–80 cm although on theoretical grounds Boulton andDobbie (1993) maintained that it can vary in thicknessbetween 4 and 47m. Other field studies also indicaterelatively thin deforming beds; a few centimetres at most(Table 3; see also discussion in Piotrowski et al., 2001).However, although values reported from beneathexisting glaciers do not indicate a great thickness, it ispossible that these are not representative and thicknessesare greater elsewhere, or beneath Pleistocene glaciers asindicated by deformation structures. Until measure-ments are available to resolve this problem, we must usestructures in Pleistocene till beds to estimate (minimum)values for deforming bed thickness. The latter isinherently difficult, as will be shown in the followingexample. Although it is possible to provide moreexamples, that should be the topic of a follow-up paper.
A till of Saalian (penultimate) Glaciation from TheNetherlands (Emmerschans, Pit ‘de Boer’, 1:25,000 mapsheet 17H; coordinates: 259.30/536.25). Fig. 11 shows adeformed sand lens at the base of the till. Particle byparticle lodgement cannot account for the shape of thelens, it has been elongated from an initially morespherical form. Similar inclusions have been describedfrom many other sites in the world (Kr .uger, 1979). Inthis case, the till bed is ca 3m thick but this is anunderestimate because there is no Last Interglacial(Eemian) soil in the till indicating that erosion hastaken place, probably under periglacial conditionsduring the Last Glaciation (van der Meer and Lager-lund, 1991). Consequently we have to add a minimum of2m to the thickness of the till, bringing it to at least 5m.Above the deformed sand lenses the till is well-structured and fissile (Rappol, 1983).
It could be possible to argue that the sand lens hasbeen deformed in a 30 cm layer reflecting its verticalextent, but that the diamicton that constitutes theremaining part of the sequence has been lodged particleby particle. Not only would this be special pleading, butit also makes the sequence more complicated requiring achange from lodgement to deformation, and back tolodgement. Such a variability is not explained in ourmodel, because the latter only deals with variation indeformation.
However, the fissile structure of the till can beexplained by deforming bed processes and the micro-structures observed in thin section are typical of adeforming bed. Thus we believe that the whole >5m oftill was formed as a deforming bed. It is important torehearse the alternative hypotheses: (i) The deformingbed has never been thicker than the 30 cm to cover thesand lens, and that this deforming bed has movedupward to form the overlying diamicton in an accre-tionary sequence. The implication of this scenario is thatthere is sedimentation in between the ice and thedeforming bed and that particles are being lodgedbecause the ice cannot overcome the friction against amoving, lubricated bed. (ii) A glacier deformed the sandlens, after which deformation ceased and was replacedby particle-by-particle lodgement of all or part of theoverlying till sequence. Following this hypothesis it isconceivable that the overlying till could have beenconstructed by alternation of deformation and lodge-ment. The latter scenario would explain the presence ofdeformational microstructures, but not the absence ofany major structural boundaries.
In addition to the above information it is important tonote that the sub-till sediments consist of uniform, finesands with frost wedges, which have been plasticallydeformed to a depth of several decimetres (van der Meerand Kars, 1995), and of sand, silt and clay laminatedsediments which show evidence of intense deformationover a thickness of up to 5m and interfingering with the
ARTICLE IN PRESS
Fig. 11. Emmerschans, The Netherlands; deformed sand lens near the
base of the Saalian till, has been incorporated from underlying bed. See
text for discussion of what this implies for deforming bed thickness in
general.
J.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–1685 1679
base of the till (see also Rappol, 1983). The 5mthickness gives an indication of the depth to whichsubglacial deformation can extend.
6. Tectomict
To many, sediments are deposits that are made up ofparticles of pre-existing rocks or their alterationproducts. At the same time, the term carries the notionof these particles having been deposited by water, wind,ice, air or gravity. However, in light of the foregoingdiscussion of till as a deforming bed, the question needsasking: ‘is subglacial till a sediment?’. To place thisquestion in context it is important to recall that not allaccumulations of pre-existing rock material are depos-ited by water, wind, ice, air or gravity. Peats grow insitu, as do certain limestones. Fault gouge is notdeposited by water, wind, ice, air or gravity, althoughit is composed of particles of pre-existing rocks. Thereare three types of sediments: (i) those that are depositedby water, wind, ice, air or gravity, (ii) those that grow insitu, and (iii) those that are the result of structural,tectonic (kinetic) processes. It is our contention thatsubglacial tills fall into the third category, togetherwith fault gouge. This is not a new notion as alreadyin Pettijohn (1949), in his book ‘Sedimentary Rocks’under the heading ‘Cataclastic breccias and conglomer-ats’ stated: ‘Formed in an analogous matter is till,which is the product of glacial action. This material isbasically an extensive gouge, caused by grindingalong the base of an overthrust ice sheet. Petrographicand structural similarities and mode of origin justifythe grouping together of the glacial and tectonicmoraines.’
At a higher level, these structural sediments can begrouped together with cataclastic, metamorphic rocks.In order to avoid the confusion over the traditionalconcept of sediments we propose to use the term‘tectomict’ for sediments such as subglacial till that arethe result of structural processes.
7. Conclusions
We conclude that:
* Extensive use of till micromorphology has establisheda range of microstructures which demonstrate that allsubglacial tills, including macroscopically massiveones, are former deforming glacier beds. Hitherto,much consideration of the existence of former orpresent deforming beds, has ignored these observa-tions. Because all subglacial tills are deforming glacierbeds, it follows that classical till varieties such aslodgement till and subglacial melt-out till are unlikely
to occur, and we are of the opinion that they do notexist. Debris release from ice can be discussed interms of lodgement or melt-out processes, buthitherto, subglacial sediments (except melt-out tillsin contact with glacier ice) produced by theseprocesses have yet to be recognised.
* Deformational microstructures demonstrate that theintensity of deformation is influenced by glaciervelocity, water content, clay content, possibly deposi-tion of secondary carbonates and to a lesser extent byclay mineralogy.
* A combination of spatial changes in water and claycontent results in a strongly diversified deformingbed, continuously changing its configuration overspace and time. However, because the changes inwater content can be modelled, it is possible topredict the till resulting from glaciation for anyknown sedimentary sequence. Viewing subglacial tillsas spatially and temporally highly variable deformingbeds and discarding till varieties like lodgement andsubglacial melt-out is not a step back in under-standing tills, instead it leads to much better under-standing of the complexity of till structures andproperties, and allows prediction of till propertieswith a precision hitherto unknown. It diminishes thenumber of till varieties but increases the number ofrecognisable processes and the factors that influencethem.
* Understanding subglacial tills in all the complexitiesof a spatially and temporally highly variable deform-ing bed instead of understanding them in depositionalprocesses increases our understanding of the pro-cesses of till formation.
* Processes active within deforming beds explain:
(i) the close associations of tills of markedly differentcomposition and without apparent mixing,
(ii) geochemical anomalies,(iii) the development of fissility in till,(iv) the development of deformation macrostructures
such as shears, folds and fractures,(v) the development of deformation microstructures in
till including birefringent plasmic fabrics andmarble-bed configuration,
(vi) the development of internal and lower boundariesof till beds, including till wedges.
* Both (iii) and (iv) have important consequences forthe hydraulic properties of many tills.
* Differences in bed velocity under constant fluxconditions lead to differences in thickness of thedeforming bed. One way to accommodate this isthe formation of glacier bedforms which respond tothe variations in stress at the glacier bed. Because thisautomatically leads to streamlining, the final result isa drumlin or a flute field, and the consequence of
ARTICLE IN PRESSJ.J.M. van der Meer et al. / Quaternary Science Reviews 22 (2003) 1659–16851680
changing glacier or deforming bed properties aresuperimposed bedforms.
* Subglacial tills are not depositional but structuralsediments and are defined as tectomicts.
* All till properties and their development, for instanceporosity or internal and external boundaries have tobe redefined following the concept of the deformingbed.
* Given the origin of subglacial tills as deforming beds,the nature and amount of glacial erosion has to beredefined, which will lead to a reassessment of glacialtransport and deposition.
Acknowledgements
We would like to thank the technicians that over theyears have provided us with thin sections of goodquality, despite the difficult nature of the material:Frans Backer, Candy Kramer, Jerry Lee, AdrianPalmer, John Taylor, Cees Zeegers. Likewise thestudents and colleagues who have provided us withsamples, the opportunity to examine their findings andmany challenging discussions (please note that inclusionin this list does not automatically imply agreement withour views): Marcel Bakker, Peter Barrett, Ernst vanBeek, Oscar Bloetjes, Jaume Bordonau, GeoffreyBoulton, Simon Carr, Tha.ıenne van Dijk, DietrichEllwanger, Ed Evenson, Maarten van Ginkel, VicGostin, John Hiemstra, Hans H .ofle, Aoibheann Kil-feather, Kurt Kjaer, Johannes Kr .uger, Cees Laban,Erik Lagerlund, Mark Lloyd Davies, JuhaPekkaLunkka, Danny McCarroll, Herman M .ucher,Sandra Passchier, Ross Powell, Mike Prentice, JorgeRabassa, Martin Rappol, Kenneth Rijsdijk, ChristianSchl .uchter, Vincent Simons, Anders Solheim, MirandaSteward, Arjen Stroeven, Anja Verbers, Johan Visser,Willie Warren, Dick van der Wateren, Colin Whiteman,Grant Young, Kamil Zaniewski. Similarly the manydifferent funding agencies that have sponsored ourresearch.
We would like to profoundly thank the reviewers ofthis paper, Jan Piotrowski and Tavi Murray. Theirprofessional approach and very critical comments havehopefully led to a better paper, even if it does not takeaway all their problems.
Last, but not least, we would also like to acknowledgethe participants in the International Workshops on theMicromorphology of Glacial Sediments (Amsterdam,Brock, London, T .ubingen), the students in the NERCsponsored M.Sc. in Quaternary Science at RoyalHolloway University of London, as well as those inother (short) courses for continuously asking ‘but is it alodgement till?’. This question has now given them morethan they bargained for.
References
Alley, R.B., 1989. Water pressure coupling of sliding and bed
deformation: II. Velocity-depth profiles. Journal of Glaciology