Introduction
Much of the fast-flowing ice in Antarctica moves atop unlithified sediments whose physical properties regulate water flow and fast glacier slip (Flowers, Reference Flowers2015; Zoet and Iverson, Reference Zoet and Iverson2020). Recent model assessments have shown that a primary source of uncertainty in glacier flow and sea-level rise predictions stems from how the glacier basal boundary conditions are parameterized, specifically the parameterizations chosen to represent the myriad processes active in the basal zone (Brondex and others, Reference Brondex, Gillet-chaulet and Gagliardini2019; Seroussi and others, Reference Seroussi2024). Predictions of glacier flow sensitively depend on the physical properties of the underlying till, as it regulates the bed’s mechanical strength and the subglacial hydrology (Clark, Reference Clarke1987; Murray, Reference Murray1997; Tulaczyk and others, Reference Tulaczyk, Kamb and Engelhardt2000; Iverson, Reference Iverson2010; Zoet and Iverson, Reference Zoet and Iverson2020).
We use the term ‘till’ here and throughout as it is the term most utilized for subglacial sediments in Antarctica by the glaciology community; however, the term till implies a specific genetic origin; sediment deposited directly by ice. Therefore, basal till implies that the sediment was mobilized and deposited directly by the basal ice. In some (or perhaps many) instances, sediments may have been deposited by older geomorphic agents (i.e. deposited via marine processes) and have not yet been reworked by the glacier and thus are not technically ‘tills’. A more appropriate term would be subglacial sediments, but nonetheless, we retain and use the term ‘till’ here to align with the standard terminology in glaciology, even though it may not be accurate in all instances.
The dependence of ice flow sensitively depends on till properties, in part, from how the granular structure of till sets the basal resistance to gravitational driving stress (Iverson and others, Reference Iverson, Hooyer and Baker1998; Kamb, Reference Kamb2001; Hansen and others, Reference Hansen, Warburton, Zoet, Meyer, Rempel and Stubblefield2024; Madden and others, Reference Madden, Hansen, Zoet and Suckale2025). The structure of till also regulates water flow through the bed, thus setting the subglacial water pressure (Flowers, Reference Flowers2015; Warburton and others, Reference Warburton, Hewitt and Neufeld2023). While accurately representing the physical properties of till is important for modeling, these properties are often treated as tuning variables, even though they are intrinsic characteristics that can be measured. In part, treating these variables as tuning parameters arises from the limited information available on their physical properties in settings such as Antarctica. Here, we aim to establish bounds on a range of physical properties of subglacial sediments collected from the Antarctic continent (largely limited to terrestrial samples near the Transantarctic Mountains (TAM) and Victoria Land given their ease of collection compared with marine sediments), using a variety of measurement techniques, for use in models. Our goal here is not to interpret these results in the context of the various models that utilize them, but rather to provide values that can be used by other researchers.
Despite the importance of constraining the physical properties of subglacial sediments for glacial hydraulic and ice-flow models, geotechnical testing of Antarctic tills remains limited. A few studies have examined the geotechnical properties of subglacial till from Siple Coast ice streams (e.g. Kamb, Reference Kamb1991; Tulaczyk and others, Reference Tulaczyk, Kamb, Scherer and Engelhardt1998), but overall, the dataset remains sparse compared to those from other glaciated regions (Benn and Evans, Reference Benn and Evans2014). The difficulty of retrieving sediment from beneath the Antarctic ice sheet, hundreds to thousands of meters thick in some areas, contributes significantly to this issue. However, glaciers can transport till to the termini, where it is more easily collected. We obtained ten different till samples from the Byrd Polar Rock Repository and one till from the paleo-grounding zone of Bindschadler Ice Stream (Core NBP1502B KC14) (McGlannan and others, Reference McGlannan, Bart, Chow and DeCesare2017), which represent a range of locations and lithologies around Antarctica, though primarily found along the Transantarctic Mountain Range (Fig. 1). We conducted a suite of tests to characterize the properties of the tills: namely grain-size distributions (GSDs), frictional properties, hydraulic conductivities, coefficients of consolidation and mineralogic properties following common standards and methods (i.e. Lambe and Whitman, Reference Lambe and Whitman1979). In the following section, we describe how these physical properties inform glaciological investigations, followed by the methods used to measure them.
Shaded relief map of Antarctica showing the locations of the till samples collected from the Byrd Polar Rock Repository (PRR) and the Bindschadler (BIND) ice stream along with their sample numbers. The background color is bedrock elevation of Antarctica from BEDMAP2.
Grain-size distribution
The GSD of a sediment sample represents a combination of the mineralogical makeup, the physical history of the parent source materials and the transport history experienced prior to collection. For example, eroded sandstone is likely to produce a different GSD than granitic rock, owing to the greater diversity of mineralogic components in the granite compared to the relatively uniform composition of the sandstone. Additionally, in shearing subglacial till, force chains develop that disproportionately load certain grains. Despite the relatively low bulk effective and shear stresses, these localized stresses can reach sufficiently high values within a subset of grains to cause them to fracture (Iverson and others, Reference Iverson, Hooyer and Hooke1996). This process, termed ‘comminution’, breaks down grains of a certain size, often leading till to obtain a fractal grain-sized distribution that can be represented as follows:
\begin{equation}\mathbb{N}\left( d \right) = {\mathbb{N}_0}{\left( {\frac{d}{{{d_0}}}} \right)^{ - m}},\end{equation}where d is the grain size, d 0 is a reference-sized grain, 
${\mathbb{N}_0}$ is the number of grains of the referenced grain size and m is the fractal dimension. A fractal dimension of a mature sediment where particle loads within the deforming sediment are most supported is theorized to be ∼2.6, whereas most previously measured tills have an m closer to 2.9 (Hook and Iverson, Reference Hooke and Iverson1995).
Additionally, the processes of quarrying and abrasion produce two distinctly different initial particle-size distributions: quarrying produces larger clasts that can be progressively broken down in transport, whereas abrasion is highly effective at producing fine, silt-sized grains (Alley and others, Reference Alley, Cuffey and Zoet2019; Hansen and others, Reference Hansen, Brooks, Zoet, Stevens, Smith, Bate and Jahnke2023). Conversely, glaciers may also override preexisting sediments whose genesis is only minimally related to glacier erosion and whose initial GSD differs markedly from that produced in a purely glacial setting—for example, sediments originally deposited by marine processes that were later overrun by advancing ice. Many of the ice stream beds in Antarctica are hypothesized to have been open to the ocean at one point, with sediment deposition via marine processes and then later overrun by glaciers (Scherer and others, Reference Scherer, Sjunneskog, Iverson and Hooyer2004; Kingslake, Reference Kingslake2018), which may skew grain sizes to the finer side of the spectrum. Grain-size information, therefore, is of primary importance for understanding dynamical aspects such as the frictional and hydraulic response of glaciers, but it can also provide insight into the genesis of sediments within a region.
Frictional properties
The rate at which fast-flowing glaciers move is significantly influenced by the physical properties of the underlying till (Zoet and Iverson, Reference Zoet and Iverson2020; Hansen and others, Reference Hansen, Warburton, Zoet, Meyer, Rempel and Stubblefield2024). Till is typically characterized as a Mohr–Coulomb material (Iverson and others, Reference Iverson, Hooyer and Baker1998), meaning it exhibits a linear relationship between shear stress, 
$\tau ,$ and effective normal stress, N (overburden pressure minus basal water pressure) at the point of failure (e.g. plastic deformation). The shear strength of till can be expressed using the Mohr–Coulomb failure criterion:
\begin{equation}\tau = c + \left( {{P_{\text{i}}} - {P_{\text{w}}}} \right)\tan \left( \phi \right)\end{equation}
\begin{equation}\tau = c + N\tan \left( \phi \right)\end{equation} where 
$\tau $ is the shear strength, 
$c$ is the cohesion of the material, 
${P_{\text{i}}}$ is the pressure from the overlying ice, 
${P_{\text{w}}}$ is the water pressure, N is the effective normal stress 
$\left( {{P_{\text{i}}} - {P_{\text{w}}}} \right)$ and 
$\phi $ is the internal friction angle. Thus, a larger 
$\phi $ results in a ‘stronger’ till if all other factors are held constant. Because the friction angle directly affects the till’s resistance to shear forces, it plays an important role in determining glacier slip velocity (Zoet and Iverson, Reference Zoet and Iverson2020; Hansen and others, Reference Hansen, Warburton, Zoet, Meyer, Rempel and Stubblefield2024).
Previous studies have demonstrated Coulomb-plastic rheology in subglacial tills. Laboratory testing, including triaxial (Leeman and others, Reference Leeman, Valdez, Alley, Anandakrishnan and Saffer2016), ring-shear (Iverson and others, Reference Iverson, Hooyer and Baker1998; Damsgaard and others, Reference Damsgaard, Egholm, Piotrowski, Tulaczyk, Larsen and Tylmann2013) and direct-shear (Kamb, Reference Kamb2001) experiments, have shown that till deformation closely follows Coulomb’s model, where shear stress is linearly dependent on effective normal stress (Tulaczyk and others, Reference Tulaczyk, Kamb and Engelhardt2000; Kamb, Reference Kamb2001). When a till is sheared, resistance to plastic deformation arises from two processes that accommodate and resist the motion of the granular media: the grains within the matrix must both dilate and slide as they move past neighboring grains, during which microscopic asperities on their surfaces interact as they move relative to one another. Increasing N necessitates greater shear stress to accommodate further deformation in a manner that is linearly proportional to the increase in N. Additionally, as sediment is sheared from its initial packing state, the resistive drag evolves with strain until a displacement where a critical state is achieved (Lambe and Whitman, Reference Lambe and Whitman1979), representing the deformation necessary to reset grain contacts and the porosity of the matrix. As the sample is first deformed and strain accumulates, the strain–stress relationship is linear until a yield stress is reached, beyond which irrecoverable viscoplastic deformation occurs until the material reaches its failure strength. Upon further shearing, the stress reaches the residual strength, whereupon the ‘steady state’ resistive stress is obtained. The exact path to residual strength varies depending on whether the sediment was underconsolidated or overconsolidated prior to shearing; however, regardless of the initial conditions, the material should reach the same residual stress given sufficient strain.
Hydraulic conductivity
Water flow through a porous medium such as till is facilitated by various small-scale processes acting within pore spaces and along grain-water interfaces; however, it generally conforms to Darcy’s law, which states that water discharge is proportional to a pressure gradient and depends on material properties of the sediment (Flowers, Reference Flowers2015). In a subglacial setting, liquid water can be generated through basal melt or from the injection of meltwater generated higher in the ice column (e.g. via strain heating) or at the glacier surface. To conserve mass, water discharge at the bed must equal the rate of meltwater supply. Once at the base, water can take various pathways, such as through gaps at the ice-bed interface (Creyts and Schoof, Reference Creyts and Schoof2009), channels (Delaney and others, Reference Delaney, Tedstone, Werder and Farinotti2025), canals (Ng, Reference Ng2000) and porous flow through the till (Warburton and others, Reference Warburton, Hewitt and Neufeld2023). Some pathways facilitate efficient drainage systems where higher water discharge leads to lower water pressures, whereas others are inefficient and respond to higher discharges with increased porewater pressure (Flowers, Reference Flowers2015). In many of these inefficient instances, a limiting bound on water flow is that of Darcian flow through the porous till, where the physical structure of the till, in part, sets the hydraulic conductivity of the sediment. The Darcian flow can be expressed as follows:
\begin{equation}q = - K\nabla h,\end{equation}where the hydraulic head (h = z + Ψ) is a function of elevation head, z, and pressure head, ΨP/ρ wg, where P is the water pressure above some reference value. The hydraulic conductivity K = ρ wgk ij/μ is a tensor function that depends on the fluid density, ρ w, fluid viscosity, μ, and hydraulic permeability, k ij; however, k ij is usually assumed to be homogeneous within a given stratigraphic unit and thus K ij becomes a scalar K. Note that with strong fabric development in till, it is possible the tensor form of K ij should be retained.
Despite the importance of hydraulic conductivity in estimating water flow, it can vary by orders of magnitude (Flowers, Reference Flowers2015) due to the wide range of k stemming from variations in the GSDs and mineralogy in sediments, leading to markedly different results depending on the chosen value of K. The wide natural range of K provides substantial flexibility for modeling subglacial water flow; however, hydraulic conductivity is a physical property that can be measured.
Coefficient of consolidation
Variations in N alter the consolidation state of tills (Morgan‐Witts and others, Reference Morgan‐Witts, Hansen, Zoet and Haseloff2025). As N increases, tills consolidate, reducing sediment porosity, 
$\eta $ (often expressed as the void ratio, e = V v/V s, where V v is the volume of the void space and V s is the volume of the solid). A reduction in porosity also alters the hydraulic conductivity of the sediment, which in turn can influence the subglacial hydrologic system. Local porewater pressure within the sediment depends on the ability of the pore network to facilitate fluid flow and, consequently, plays a key role in regulating bed strength. Additionally, rapid changes in N can induce consolidation or dilation of the sediment skeleton on timescales too short for porewater to achieve equilibrium from flow into or out of pore spaces, leading to local variations in effective stress (N loc) that differ from the bulk or global N—particularly in zones of active dilation and contraction (e.g. within the deforming till layer). Understanding the relationship between N and porosity, and how the system transiently responds to sudden increases or decreases in N, is important for constraining ice–till interactions in dynamic environments, like near grounding zones.
Transient changes in the basal effective pressure, N, are common in Antarctic, Arctic and alpine glaciers, driven by processes such as tidal forcing or diurnal surface meltwater input to the bed, which alters subglacial water pressure. For marine-terminating glaciers, the overburden pressure near the grounding zone varies when tidal-induced flexure exerts varying pressure on the grounded ice, altering its contact with the bed. This tidal flexure produces a cyclic pattern in N that can extend many kilometers inland from the grounding zone (Walker and others, Reference Walker, Parizek, Alley, Brunt and Anandakrishnan2014). Most, if not all, Antarctic ice streams are marine-terminating and thus are potentially subject to such a response. These transient N have been shown to induce varying effects in the motion of glaciers (Anandakrishnan and Alley, Reference Anandakrishnan and Alley1997; Engelhardt and Kamb, Reference Engelhardt and Kamb1997; Kamb, Reference Kamb2001; Anandakrishnan and others, Reference Anandakrishnan, Voigt, Alley and King2003; Zoet and others, Reference Zoet, Anandakrishnan, Alley, Nyblade and Wiens2012). Morgan‐Witts and others (Reference Morgan‐Witts, Hansen, Zoet and Haseloff2025) examined the impact of cyclic N conditions on basal slip for soft-bedded glaciers and found a hysteretic relationship between basal drag and the imposed N. Different drag values were observed at the same N depending on whether the sample was experiencing an increase or a decrease in N; shear stress, τ, was higher during decreasing N, and lower during increasing N. By modeling the one-dimensional (1-D) Terzaghi consolidation equation, the hysteresis loops observed in the experiments could be replicated. Modeling the consolidation process requires only one experimentally determined parameter: the coefficient of consolidation, c v, which can be measured from oedometer tests. However, relatively small changes in c v result in substantial changes in the hysteresis loops and thus constraining the c v values for a range of tills is of importance in determining the dynamics for transiently varying N glaciers.
Lithological properties
The lithological properties of sediment can influence the mechanical and hydrologic properties as well as provide information on the provenance and pathways of how sediments were transported and the types of rocks they may have been derived from. For example, with clay-rich sediments, there are some forms of clay (e.g. smectite, vermiculite, montmorillonite) that swell when wet, giving rise to a change in mechanical properties depending on their saturation state. Other clays (e.g. smectite, illite) are known to host rate-weakening, unstable slip (produces icequakes), whereas other clays demonstrate rate-strengthening properties (promote stable sliding) (Carpenter and others, Reference Carpenter, Ikari and Marone2016). Thus, identifying the types and abundance of certain minerals within the sediment matrix can have bearing on the glacier’s mode of slip. Additionally, through reconstruction, the mineralogic properties of the sediment can provide insight into the rock types from which the clays are derived, which can otherwise be difficult to determine below an ice sheet.
Methods
Geology and sample selection
Most samples in this study come from the TAM. Broadly, the samples can be grouped into the Southern TAM (two), Central TAM (three), Southern Victoria Land (three), as well as one sample from Ross Island and one sample from Seymour Island along the Antarctic Peninsula. For most of the TAM sites, local bedrock consists of the Cambrian to Ordovician Granite Harbour Intrusive granitoids, the Devonian to Triassic Beacon Supergroup Siliciclastics and the Jurassic Ferrar Dolerite (Cox and others, Reference Cox2023). Local bedrock near Sample PRR-45326 (PRR is the code used for samples obtained from the Byrd Polar Rock Repository) is a narrow band of Archean to Paleoproterozoic metamorphic rocks called the Nimrod Complex (Goodge, Reference Goodge2020). Sample PRR-23842 was collected near Cape Royds along the lower flank of Ross Island, part of the late Cenozoic McMurdo Volcanic Group. Sample PRR-45325 was collected on Seymour Island, which consists of a Cretaceous–Paleogene sequence of clastic sedimentary rocks (Elliot and others, Reference Elliot, Askin, Kyte and Zinsmeister1994). One till from the paleo-grounding zone of Bindschadler Ice Stream (Core NBP1502B KC14) was collected from a water depth of 523 m b.s.l. from a subglacially deposited retreat ridge (McGlannan and others, Reference McGlannan, Bart, Chow and DeCesare2017). This core sediment was collected from the interval of 240–280 cm (base of the core). The samples analyzed here were, in some instances, subsampled from larger sample volumes at the Byrd Polar Rock Repository if a sufficient volume was present, but in instances where the sample volumes were small, the entire sample volume was sent and utilized in the tests. The samples were disaggregated by the time they were obtained at UW-Madison, and so all tests were completed on remolded samples.
Grain-size distribution
GSDs are a fundamental component of sediment. For grain sizes >2 mm, the GSD measurements were performed using stackable sieves with decreasing mesh sizes. To determine the GSD, the sediment was first oven-dried, and the total mass of the dry sediment was measured. If needed, the sediment was disaggregated with a mortar and pestle. The sediment samples were placed through a series of sieves with mesh sizes of 25, 16, 8, 4 and 2 mm. The sediment was placed in the sieve stack, and the stack was agitated. Grains smaller than the sieve opening passed through, until the grains can no longer pass through a sieve. The sediments retained on each sieve are weighed, and the percentage of weight retained can be determined. Sediments that passed through the 2 mm sieve were examined using a Malvern Mastersizer laser diffraction device. The laser grain-size analyzer employs laser diffraction to determine the size distribution of the grains. Approximately 0.25–0.5 g of sample was used for each laser diffraction measurement. The instrument measures how light scatters when a laser beam passes through a dispersed sample. Larger particles scatter light at smaller angles, whereas smaller particles scatter light at wider angles. This scattering pattern is then analyzed using Mie theory to calculate the particle-size distribution. Density was also measured for each sample and is detailed below. A plot of the retained grain percentage versus grain size is generated that allows an assessment of the uniformity of the material, and an estimate of m (Eqn (1)) that can assist with further classifications using USCS (Unified Soil Classification System) or AASHTO (American Association of State Highway and Transportation Officials) methods.
Frictional properties
To measure the frictional properties of the tills (
$\phi $ and c in Eqn. (2)), we used a direct-shear machine, shown in Fig. 2, following ASTM (American Society for Testing and Materials) standards D3080/D3080M. Following standard protocol, we removed grains larger than 1/10 of the smallest sample shear box dimension, which was 2 mm. The sieved till was then added to the shear box, which was filled to approximately three-fourths capacity. The shear box consists of an upper half and lower half that are initially locked together by removable pins shown in Figure 2. The frame is rigid to prevent distortion during shearing, and the shear box has dimensions of 10 × 10 × 2 cm. The shear box was gently agitated to ensure the sediment was evenly distributed throughout. The sediment-filled shear box was then placed into the direct-shear apparatus, and a normal load was applied to the sample by hanging weights from the lever arm. Each sample underwent ten separate tests with varying weights: 1, 2, 4, 5, 7, 8, 17, 20, 36 and 52 kg, which correspond to effective normal stresses of 10, 20, 39, 49, 69, 79, 167, 196, 353 and 510 kPa. Each time a load was added, the lever arm was leveled to ensure uniform weight distribution. Before the test begins, the shear box halves are unlocked by removing the pins so the lower half can be laterally displaced. Once displacement is initiated, the lower half of the shear box is displaced laterally with respect to the upper half at a constant rate of shearing via a horizontal ram (Fig. 2). A horizontal ram displacement speed of 5 mm min−1 was chosen for all tests. this speed was selected to ensure enough time for the sediment to fully dissipate excess pore pressure. While the sample is undergoing shear, the shear stress, lateral displacement and normal displacement are measured via a load cell and horizontally and vertically mounted linear variable differential transformers (LVDTs), respectively. The measured steady-state shear stress is plotted against the applied vertical normal load, and the slope of the best-fit line is equal to 
$\phi $, whereas the y-intercept is used to estimate c in Eqn. (2).
Annotated direct-shear machine. (a) This panel shows an annotated figure of the direct shear where (A) is the crossmember that supplies the normal load, (B) is the vertical LVDT that measures dilation and contraction of the sample during shear, (C) is the horizontal LVDT that measures the lateral displacement of the sample during shear, and the ‘S’ type force gauge that measures the resistance to shear, (D) is the lever arm upon which weights are hung that transfer the vertical load to the crossmember. The black box to the right of (A) is the shear box, which is schematically shown in panel (b). (b) Within the shear box, the horizontal ram supplies a horizontal stress that displaces the upper half of the sample past the lower half at a prescribed rate. This displacement causes a shear plane to develop within the sample (dashed red line), and the frictional resistance provided along the shear plane is measured. This process is then incrementally repeated for increasing normal stress loads.
Figure 2 Long description
An annotated direct shear machine setup. The crossmember labeled A supplies the normal load. The vertical LVDT labeled B measures dilation and contraction during shear. The horizontal LVDT labeled C measures lateral displacement and the ′S′ type force gauge measures shear resistance. The lever arm labeled D transfers vertical load to the crossmember. The shear box is to the right of A. The image B shows a schematic of the shear box. It includes a horizontal ram that applies horizontal stress, displacing the upper half of the sample past the lower half, creating a shear plane indicated by a dashed red line. Normal stress is applied from above and horizontal stress is applied from the side, with sediment filling the box.
Hydraulic conductivity
Compaction permeameters are a tool used to evaluate the hydraulic conductivity of porous materials at different compaction conditions. Fundamentally, the test is based on Darcy’s law. To determine the hydraulic conductivity, water is allowed to flow through a prepared sample of the sediment under constant or falling head conditions. The volume of water that flows through the sediment in a given amount of time is used to calculate the hydraulic conductivity following Eqn (3). The tests performed here were constant head tests, which refer to the constant water pressure gradient that drives flow through the sediment specimen. We followed ASTM standards D5856/D5084/D698/D1557/D2434/D2216 for each applicable step. First, a compaction mold of known dimensions is utilized to calculate the volume of the sediment sample that will be placed inside (Fig. 3). The mold used for these experiments is of the same dimensions as a standard Proctor mold, with an interior diameter of 4 in and a depth of 4.584 in. The empty mold is placed onto a steel base plate, which contains a cavity for a porous stone. The mold is held in place by two wing nuts and threaded rods that run parallel to the mold. A 2 in-long collar is placed on top of the mold. The empty mold, collar and plate are weighed. The sediment of interest is then prepared to the desired water content. In most cases, this is done with just enough water to maintain the sediment integrity. (For samples receiving a full compactive effort, additional water may be added to reach the moisture content corresponding to optimal dry density.) At this stage, a small subsample of the sediment is collected, weighed and oven-dried to determine water content, which is later used to estimate porosity. The sediment specimen is loaded into the mold in lifts and compacted to the desired value. Typically, porosity decreases with increased compaction effort. The majority of the sediment tested had significant percentages of clayey materials, so compaction efforts were minimal to moderate. Once the sediment is in the mold and compacted, the mold is weighed, and the thickness of the sediment is measured to determine the volume and mass of the sediment for subsequent density calculations. Another porous stone is placed on top of the sediment, finishing the sediment preparation. Next, the top of the sample chamber is placed onto the collar. The top cover is a steel plate with a valve connected to the sample chamber. The cover is screwed into place, sealing the sediment inside.
Annotated image of the compaction permeameter. (a) is the lower sample mold which holds the compacted sediment sample. (b) is the lower assembly with an entry valve upon which the porous plate and sample mold are placed. (c) are the thumb screws which are used to compress the upper sample mold and lid against the lower sample mold. (d) are the porous plates and O-rings. One porous plate is placed beneath the sample, and a second is placed atop it. (e) is the upper sample mold that is placed atop the lower mold once the sample is installed. (f) is the upper assembly with an exit valve. Once the sample is constructed and the permeameter assembled, a standpipe with a water-holding funnel (not shown) is connected via a tube to the lower valve in the lower assembly, and water is allowed to flow through the sample to the upper valve, where the discharge is measured.
Next, the constant head reservoir is prepared. A standpipe is used to raise a funnel some distance above the sediment specimen. This funnel delivers water to the sample through a tube. The funnel was modified via the drilling of two holes in its side, one to provide water from a faucet and another to allow excess water to flow out. This system allowed the water level to remain constant throughout the test, keeping the water pressure acting on the specimen constant (e.g. constant head). The vertical distance between the specimen and the water level in the reservoir is measured to determine the head acting on the specimen. The sample chamber is allowed to fill with water, and the valves are opened to allow water flow through the specimen. Once flow begins, a beaker is used to collect the outflowing water that has traveled through the specimen, and a stopwatch is used to record time since flow began. Using a graduated cylinder, the volume of water that was collected in the beaker can be measured. The hydraulic conductivity, K, was then estimated based on the time to flow and the volume of water that exited using Eqn (3).
Coefficient of consolidation
To determine the coefficient of consolidation, c v, for a sample, a 1-D consolidation test using incremental loading was employed using an oedometer apparatus (Fig. 4) (Lambe and Whitman, Reference Lambe and Whitman1979). Both the sample preparation and experimental testing complied with ASTM standard D2435/D2435M and are detailed below.
Oedometer apparatus. The main panel shows the oedometer apparatus where (a) is the vertical LVDT that measures the compactive state of the sediment throughout the test and (b) is the sample assembly in which a ring of sediment is compressed via the pneumatic press, which delivers a downward load via the horizontal crossmember. The inset figure shows the sample rings, inside of which the sediment samples are contained before the sediment-filled rings are inserted into the sample assembly.
To prepare the samples, the tills are sieved to include only particles with a diameter less than 2 mm, following the standard. Deionized (DI) water is added to the sieved sediment and mixed until uniform. The moistened sediment is organized into a mound, taller in height and wider in diameter than the ring used to load the sediment into the consolidation apparatus (Fig. 4). The ring, which is 2.5 in diameter with a height of 1 in, must be weighed before filling with sediment. The sharp end of the ring is used to cut into the mound, such that the entire ring is filled with sediment and that the ring is overflowing on every side. A wire saw is then used to level the top and bottom of the ring, ensuring that the sediment is flush with the plane of the ring. The excess sediment cut from the ring is saved on preweighed trays to be used for calculating the volumetric water content (VWC) change throughout the experiment, using the following equations:
\begin{equation}{w_0} = \frac{{{M_{{\text{T}}0}} - {M_{\text{d}}}}}{{{M_{\text{d}}}}}\end{equation}
\begin{equation}{\text{ }}{w_f} = \frac{{{M_{Tf}} - {M_d}}}{{{M_d}}}\end{equation}where w 0 is the initial water content, (%), w f is the final water content, (%), M T0 is the moist mass of specimen before the test, (g), M Tf is the moist mass of specimen after the test, (g), and M d is the dry mass of specimen (g). When the VWC samples (at least three) have been weighed and the ring cleaned so that no excess sediment remains on the outside, the ring is weighed to obtain the initial wet mass of the sediment. Filter paper is placed on both sides of the ring, ensuring that the side of the ring with the cutting edge has no filter paper extending past the internal edge of the ring. Moistened porous stones are placed on the other side of the filter paper on both ends, and the ring is loaded into the oedometer apparatus (Fig. 4).
In these tests, a till is subjected to different increasing vertical loads while situated in a rigid ring to prevent lateral distortion. An LVDT records vertical displacement (consolidation of the sample) over time as the incremental loading is applied. Till samples were subjected to increasing loads between 5 and 400 kPa (see results Table 4 for specific values). (Note: the 5 kPa seating stress is not intended to consolidate the sample, but instead to prevent expansion from the added water.) The load is increased at 24 hour intervals, according to the ASTM standard, to allow for the sediment to consolidate fully at that stress step. When all increasing steps have been completed, the sediment is ‘unloaded’ by decreasing the load in the reverse order that it was loaded in, also in 24 hour intervals. Throughout the duration of the experiment, the water level in the apparatus will drop due to evaporation. Water is added when the water level drops below 1 in from the top of the sample chamber to ensure that the till remains fully saturated. Following the end of the experiment, the ring is removed from the apparatus, all filter paper is removed, the final thickness of the sediment is measured and the final wet mass is recorded. VWC samples and the sediment-filled ring are then placed in an oven for at least 24 hours to ensure that all water is evaporated from the samples. The dry masses for all VWC samples and the sediment-filled ring are recorded.
In order to estimate c v, the vertical displacement data collected from the LVDTs are used to calculate axial strain, ε, continuously over the duration of the experiment, which is plotted against the square root of time (s1/2).
\begin{equation}\varepsilon = \frac{{\Delta H}}{{{H_0}}}*100,\end{equation}where ε is the axial strain percent (–), ∆H is the change in height (cm) and H 0 is the initial height (cm). To calculate the c v, which is done for each consolidation step, the Taylor method of calculation is used (Taylor, Reference Taylor1948). This method involves plotting axial strain, ε, against the square root of time (s1/2), for each load increment. A best-fit line is calculated for the early linear time readings (i.e. red line in Fig. 5), and extrapolated back to t = 0, which represents 0% primary consolidation. A second line is drawn from t = 0 so that the x-intercept is 1.15 times larger than the first (red) line. The intersection of this line with the square root of the time gives the axial strain at 90% primary consolidation (ε 90), which is then used to calculate the coefficient of consolidation, c v, in cm2 s−1, using Eqn (6).
\begin{equation}{c_v} = \frac{{TH_{D50}^2}}{t},{\text{ }}\end{equation}Taylor method. The x-axis is the square root of time, and the y-axis is the axial strain. This method is used to find the c v value. The best-fit line is calculated for the early linear time readings (i.e. red line) and extrapolated back to t = 0. The second line (i.e. black line) is drawn from t = 0 so that the x-intercept is 1.15 times larger than the red line. The intersection of this black line with the x-axis gives the axial strain at 90% primary consolidation (ε 90).
where T is a dimensionless time factor equal to 0.848, t is the time corresponding to 90% consolidation, t 90, and H D50 is the length of the drainage path at 50% consolidation, equal to the full specimen height. A consolidation coefficient is calculated for each increasing load increment.
Void ratio
For each step in the consolidation process where the sample consolidated sufficiently to measure a c v, the void ratio, e, the ratio of void volume to the volume of solids (Lambe and Whitman, Reference Lambe and Whitman1979), was also calculated in order to estimate the porosity, 
$\eta ,$ versus N using the following series of (7a–7d):
\begin{equation}{V_{\text{s}}} = \frac{{{M_{\text{d}}}}}{{{\rho _{\text{d}}}}},\end{equation}
\begin{equation}{H_{\text{s}}} = \frac{{{V_{\text{s}}}}}{A},\end{equation}
\begin{equation}e = \frac{{{H_{\text{f}}} - {H_{\text{s}}}}}{{{H_{\text{s}}}}},\end{equation}
\begin{equation}\eta = \frac{e}{{1 + e}},\end{equation} where V s is the volume of solids, M d is the dry mass of the solids, ρ d is the till density, H s is the height of solids, A is the specimen area and H f is the final height of the sample at the given step. The void ratio, e, at the beginning of the experiment was also calculated by using the initial height of the sample before the experiment started in place of H f, and 
$\eta $ is the porosity of the sample.
Till density, ρ d, was determined using a pycnometer-based method. A dry and empty pycnometer was weighed, then partially filled with deionized water, and a known mass of till was added. The pycnometer was swirled to combine the water and till, and air bubbles were removed using a vacuum pump for at least 5 min or until no more bubbles could be seen rising to the top of the pycnometer. The pycnometer was filled with water until the meniscus was equal to the fill line, and the total mass was measured. Using the measured mass of the pycnometer and till, the volume of the water in the pycnometer was determined and subtracted from the total volume to determine the volume of till present in the pycnometer, from which the density of the till is estimated from the till’s known mass and volume.
Sediment mineralogy
Collected samples were analyzed with a Rigaku Rapid II at the Bailey XRD Lab, University of Wisconsin-Madison. This X-ray diffraction (XRD) uses a Mo source X-ray with a 0.71 Å Kα wavelength and a modified beam stop, which is specialized for identifying and differentiating clay minerals with their large (001) d-spacing in sediment samples. The analyzed samples were loaded into 1 mm polyamide tubing. This tubing contributes minimal background to the XRD patterns. Bulk <2 mm sediments were weighed alongside their respective clay fraction to quantify the relative clay mineral percentages for each bulk sample. A sediment column and centrifuge were used to extract the fine sediments with an application of Stokes’ law. Once XRD spectra were collected, the TOPAZ software was used for Rietveld refinement in order to quantify the relative phase percentages.
Results
Grain-size distribution
The grain sizes were measured for all 11 samples (Fig. 6), and their m values and GSD are shown in Table 1 and Figure 6. The full table of grain sizes for particles smaller than 2 mm can be found in the supporting material.
Grain-size distribution. (a) Plots the grain size against the normalized number of particles for the full range of grain sizes. The slope of this log–log plot is used to estimate the m value. (b) Grain size versus cumulative percent for a range of till samples. All grain sizes below 2 mm were measured using a laser diffraction device, while those grains larger than 2 mm were measured using a series of sieves.
Figure 6 Long description
A showing a log scale plot with the horizontal axis labeled, Grain size left parenthesis millimeter right parenthesis, ranging from 10 superscript negative 4 to 10 superscript 2. The vertical axis is labeled, Normalized number of particles, ranging from 10 superscript negative 4 to 10 superscript 12. A legend lists PRR 45326, PRR 22508, PRR 23658, PRR 45325, PRR 23943, PRR 22507, PRR 22506, PRR 13709, PRR 23928, PRR 23842 and Bind. Multiple dotted curves descend from upper left to lower right. Points are plotted across the full grain size range, with denser point clusters between about 10 superscript negative 3 and 10 superscript 0 on the horizontal axis. B showing a line graph with the horizontal axis labeled, Grain size left parenthesis millimeter right parenthesis, ranging from 0 to 25. The vertical axis is labeled, Cumulative percent, ranging from 0 to 1. A legend lists PRR 45326, PRR 22508, PRR 23658, PRR 45325, PRR 23943, PRR 22507, PRR 22506, PRR 13709, PRR 23928, PRR 23842 and Bind. Multiple curves rise steeply near grain size 0 and then level off toward 1 as grain size increases. Several curves reach about 0.9 by grain size near 5 and most curves approach 1 between about 10 and 25 on the horizontal axis.
Grain-size distributions.
Table 1 Long description
Grain size percentages are listed by sieve size for columns PRR minus 22506, PRR minus 23842, PRR-23658, PRR-22508, PRR-23943, PRR-45325, PRR- 23928, PRR-45326, PRR-13709, PRR-22507, and Bind. For 25 mm, values are 0.0 for all columns except PRR-23943 at 1.8, PRR-23928 at 0.9, and Bind at 11.81. For 16 mm, values range from 0.0 to 10.0 across PRR columns, with PRR-22507 at 10.0 and Bind at 3.83. For 8 mm, PRR values range from 0.0 to 12.7, with PRR-22507 at 12.7 and PRR-13709 at 0.0, and Bind at 4.43. For 4 mm, PRR values range from 0.0 to 17.2, with PRR-22507 at 17.2 and PRR-13709 at 0.0, and Bind at 3.84. For 2 mm, PRR values range from 0.0 to 19.5, with PRR-22507 at 19.5 and PRR-13709 at 0.0, and Bind at 1.18. Under 2 mm has values of 64.2, 64.7, 62.7, 65.7, 67.5, 97.6, 93.1, 67.8, 100.0, 40.6, and 74.9 for Bind. The final row labeled m (fractal dimension) lists 2.52, 2.53, 2.83, 2.41, 2.66, 2.90, 2.84, 2.97, 2.61, 2.69, and 2.88.
* All grains smaller than 2 mm were analyzed using laser diffraction, and their percentages are included in the supplemental data.
Hydraulic conductivity
The hydraulic conductivity of the measured sediments varies by four orders of magnitude for the porosities at which they were tested. The measured values are shown in Table 2.
Hydrologic properties.
Table 2 Long description
Columns list sediment sample, dominant sediment type, test type, head in centimeters, porosity in percent, average K in centimeters per second, standard deviation, and range. All rows use the constant head test type, with head values from 80.25 to 82 centimeters and porosity from 30.42 to 50.23 percent. PRR 13709, clay, has head 80.25, porosity 50.23, average K 2.37 times 10 to the minus 5, standard deviation 7.66 times 10 to the minus 6, range 1.53 times 10 to the minus 5. PRR 22506, sand, has head 80.5, porosity 36.45, average K 5.38 times 10 to the minus 4, standard deviation 1.37 times 10 to the minus 4, range 3.80 times 10 to the minus 4; PRR 22507, sand, has head 82, porosity 34.85, average K 8.80 times 10 to the minus 4, standard deviation 1.78 times 10 to the minus 4, range 5.57 times 10 to the minus 4. PRR 22508, sand, has head 81.8, porosity 34.93, average K 9.18 times 10 to the minus 5, standard deviation 2.47 times 10 to the minus 6, range 5.64 times 10 to the minus 6; PRR 23658, sand, has head 80.5, porosity 33.22, average K 3.79 times 10 to the minus 3, standard deviation 4.00 times 10 to the minus 4, range 1.02 times 10 to the minus 3. PRR 23842, sand, has head 80.4, porosity 34.45, average K 2.39 times 10 to the minus 4, standard deviation 3.39 times 10 to the minus 5, range 9.80 times 10 to the minus 5; PRR 23928, sand, has head 81.7, porosity 35.05, average K 5.09 times 10 to the minus 3, standard deviation 1.00 times 10 to the minus 3, range 2.48 times 10 to the minus 3. PRR 23943, clay, has head 80.4, porosity 41.28, average K 5.34 times 10 to the minus 6, with standard deviation and range shown as em dashes; PRR 45325, clay, has head 80.5, porosity 47.54, average K 6.08 times 10 to the minus 6, with standard deviation and range shown as em dashes. PRR 45326, clay, has head 81, porosity 30.42, average K 2.84 times 10 to the minus 4, standard deviation 3.20 times 10 to the minus 5, range 8.64 times 10 to the minus 5.
Note: ‘—’ represents tests where only one repetition was done, and therefore standard deviation and range are not applicable.
Frictional properties
The direct-shear tests measured shear stress, lateral displacement and normal displacement. From these, we plotted the shear stress against the lateral displacement and identified a range where the shear stress remained relatively constant (Fig. 7a). The mean shear stress, τ, and normal stress, σ n, were then computed based on the steady-state shear stress measurements. These values were plotted for the ten masses/loads that were tested, and a line of best fit was applied to the data, shown in Fig. 7b. The equation of this best-fit line is essentially Eqn (2), where the friction angle can be found from the slope of the fitted line and the cohesion is the y-intercept. From this, we estimate the frictional properties shown in Table 3 for nine tills. (Note the full suite of tests was not conducted on all tills due to an insufficient volume of available sediment in some instances.)
Friction measurements. (a) Displacement series plotted against the measured shear resistance, 
$\tau $. This test is conducted entirely at one 
${\sigma _N}$ value. Initially, there is an increase in 
$\tau $ as the sediment texture evolves until a critical state is reached around 9 mm of displacement. Once this steady state is achieved, a 
$\tau $ value is recorded for the given 
${\sigma _N}$. After which 
${\sigma _N}$ is increased and the test is run again to measure another 
$\tau $. (b) Plot of all the 
${\sigma _N}$ versus 
$\tau $ values for a single till. The data from each individual test (blue dots) are fit with a best-fit line (red line), whose slope is equal to the tangent of the friction angle and whose y-offset is the cohesion value.
Frictional properties of the tills.
Table 3 Long description
Measurements are listed for each till sample, giving friction angle in degrees and cohesion in kilopascals. PRR -22506 has friction angle 33.61 and cohesion 3.09; PRR -22507 has 33.30 and 6.25; PRR -22508 has 31.56 and 9.09. PRR -23658 has 33.36 and 0.49; PRR -23928 has 30.99 and 5.70; PRR -23943 has 27.36 and 9.68. PRR -45325 has 32.43 and 5.90; PRR -45326 has 31.75 and 5.11; Bindschadler IS has 30.57 and 11.16. Friction angles span from 27.36 to 33.61 degrees, and cohesion spans from 0.49 to 11.16 kilopascals.
Coefficient of consolidation
The calculated c v for each step of the consolidation process is displayed in Table 4. Within each individual till, the consolidation coefficient varied by less than two orders of magnitude. Throughout all the tills, however, the consolidation coefficient varied within several orders of magnitude. The dependence of c v on hydraulic conductivity is the likely reason for the large range of c v values measured between the different samples.
Consolidation coefficients and porosity for measurable steps for Antarctic tills.
Table 4 Long description
Columns list till sample, load in kilopascals, cv in square centimeters per second, density in grams per cubic centimeter, and porosity. For sample 13709, density is 2.49 throughout; porosity goes from 0.52 at the initial row to 0.29 at 400 kilopascals, and cv ranges from 1.66 times ten to the minus three at 60 kilopascals to 7.34 times ten to the minus four at 400 kilopascals. For sample 22507, density is 2.64; porosity is 0.40 at the initial row and remains 0.40 at 30 kilopascals, then declines to 0.29 at 400 kilopascals, while cv ranges from 1.36 times ten to the minus two to 3.17 times ten to the minus two across measured loads. For sample 22508, density is 2.57; porosity declines from 0.34 at the initial row to 0.19 at 300 kilopascals, and cv spans from 3.1 times ten to the minus two at 30 kilopascals to 9.66 times ten to the minus four at 60 kilopascals. For sample 23928, density is 2.61; porosity goes from 0.31 at the initial row to 0.26 at 400 kilopascals, with cv values including 9.46 times ten to the minus two at 5 kilopascals and 4.23 times ten to the minus three at 400 kilopascals. For sample 23943, density is 2.62; porosity declines from 0.36 at the initial row to 0.25 at 400 kilopascals, and cv ranges from 7.33 times ten to the minus four at 60 kilopascals to 6.81 times ten to the minus three at 5 and 400 kilopascals. For sample 43526, density is 2.6; porosity stays at 0.28 from the initial row through 75 kilopascals, then declines to 0.18 at 400 kilopascals, while cv ranges from 2.25 times ten to the minus four at 5 kilopascals to 4.81 times ten to the minus three at 400 kilopascals. For Bindschadler IS, density is 2.57; porosity is 0.43 at the initial row and at 50 kilopascals, then declines to 0.21 at 400 kilopascals, and cv ranges from 2.90 times ten to the minus five at 75 kilopascals to 1.65 times ten to the minus four at 400 kilopascals.
Note: Spaces marked with ‘—’ represent the initial porosity of the sample before any consolidation occurs.
In these tills, there was no clear relationship between GSD and c v. The tills dominated by sand-sized particles generally had no noticeable consolidation in the first two load steps (5 and 12.5 kPa) but began to consolidate at 30 kPa. Alternatively, the clay-dominated tills consolidated noticeably at a seating stress of 5 kPa but then did not consolidate greatly until at least two stress steps later.
Sediment mineralogy
Four mineral phases dominated the clay size fraction of these till samples including illite, montmorillonite, kaolinite and muscovite (Fig. 8). Additionally, the three primary clay minerals in each sample were normalized with respect to illite, montmorillonite and kaolinite and plotted on a ternary diagram with the associated site next to the data point (Fig. 9). All the samples had relatively low kaolinite compared to the two smectite clays (illite and montmorillonite). Muscovite is included in these plots as it is present in significant amounts in the clay-sized fraction but was not considered for the ternary plot as it is not a true clay mineral and was likely not altered from other primary mineral phases, like the other three phases. The clay fractions from these sites vary significantly in the amount of expansive versus nonexpansive smectite.
Bar graph showing the relative percentages of illite, montmorillonite, kaolinite and muscovite within the till samples. These percentages were Rietveld-refined from XRD data using TOPAZ software.
Ternary diagram using the normalized percentages of montmorillonite, illite and kaolinite. Note that PRR-23928 and PRR-23943 are superimposed on the same approximate position in ternary space.
Figure 9 Long description
The ternary diagram displays normalized percentages of kaolinite, illite and montmorillonite. The three axes represent kaolinite, illite and montmorillonite, each ranging from 0 to 100 percent. Several data points are plotted, labeled with identifiers such as PRR-45236, PRR-22508 and PRR-23928. The diagram shows the distribution of these minerals in various samples, with some points clustered near the illite and montmorillonite axes, indicating higher concentrations of these minerals compared to kaolinite.
Montmorillonite and illite are both 2:1 clay minerals, but montmorillonite has a great expansive capacity due to its relatively weak and exchangeable interlayer bonds, while illite is considered relatively rigid and nonexpansive due to its stronger interlayer bonds. Like illite, kaolinite is also nonexpansive due to its rigid interlayer bonds (Hamza and others, Reference Hamza, Hussein and Mahmoud2023; Zhang and others, Reference Zhang, Peng, Wang, Zhang and Luo2025). These largely varied quantities of expansive montmorillonite and nonexpansive illite, kaolinite and muscovite will have an impact on how the bulk sediment mixtures behave in subglacial environments, potentially affecting their ability to host rate-weakening slip (Carpenter and others, Reference Carpenter, Ikari and Marone2016). Note that XRD was not completed on the Bindschadler marine sediment sample due to limited access to the XRD device.
Discussions and conclusions
Overall, the samples exhibit a range of values, with some properties (e.g. c v and hydraulic conductivity) varying more widely than other properties (e.g. density, frictional properties and grain-size maturity). The maturity of the tills here shows m values that fall near the values of other tills (Hooke and Iverson, Reference Hooke and Iverson1995). The m values range between 2.52 and 2.90 with a mean of 2.7 ± 0.18, indicating these tills are slightly more ‘mature’ than other tills measured outside of Antarctica, where m was closer to 2.9. The hydraulic conductivity of the samples (for the tested porosities) spanned three orders of magnitude, from 3.3 × 10−3 to 6.6 × 10−6 cm s−1, with a mean value of 1 × 10−3 cm s−1. The frictional properties of the till demonstrate a range of friction angles between 27.3° and 33.6°, with a mean of 31.7 ± 1.9°. Cohesion ranged between 0.5 and 11.2 kPa with a mean of 6.3 ± 3.3 kPa. The coefficient of consolidation at the 120 kPa step (or the nearest step to 120 kPa) ranged over three orders of magnitude between 5.82 × 10−5 and 9.03 × 10−2 cm2 s−1 for the different samples. This range of values is large and likely reflects variability in GSD across the samples and their hydraulic conductivities. This implies that grounding-zone drag, which depends sensitively on c v, could vary significantly from glacier to glacier (see Morgan‐Witts and others, Reference Morgan‐Witts, Hansen, Zoet and Haseloff2025), thereby altering the timescale and lag of the drag as N varies from tidal forcing. This variability depends on the degree of homogeneity in grounding line sediments in Antarctica and elsewhere. Sediment density ranged between 2.49 and 2.64 g cm−3 with a mean of 2.59 ± 0.05 g cm−3.
For the mineralogy, montmorillonite and illite are both 2:1 clay minerals, but montmorillonite has a high expansive capacity due to the relatively weak, exchangeable interlayer bonds, whereas illite is comparatively rigid and nonexpansive due to stronger interlayer bonding. Kaolinite, although a 1:1 clay mineral, is likewise nonexpansive because of its rigid interlayer structure (Hamza and others, Reference Hamza, Hussein and Mahmoud2023; Zhang and others, Reference Zhang, Peng, Wang, Zhang and Luo2025). Variations in the relative abundances of expansive montmorillonite and nonexpansive illite, kaolinite and muscovite will influence the bulk behavior of sediment mixtures in subglacial environments.
These data define a range of properties for the tills analyzed. Most samples originate from outlet glaciers and ice streams along the TAM, reflecting sampling accessibility, as only a limited number of studies have recovered subglacial material directly through the ice (e.g. Engelhardt and Kamb, Reference Engelhardt and Kamb1997; Tulaczyk and others, Reference Tulaczyk, Kamb, Scherer and Engelhardt1998). As a result, the dataset is biased toward terrestrially derived tills that likely reflect local bedrock sources, rather than recycled, polycyclic sediments. Tills formed from reworked marine material may differ systematically from those derived directly from hard bedrock, and targeted sampling of marine sediments would provide a useful comparison. Nonetheless, these measurements establish a baseline for Antarctic till properties that can be expanded as new samples become available.
Within this framework, the samples are predominantly terrestrial, with one exception from the Ross Sea sector obtained from a paleo-grounding-zone wedge of the Bindschadler Ice Stream. This marine-derived sample is distinguished by higher clay content, lower permeability and a broader porosity range across the applied stresses relative to the terrestrial samples. These characteristics are consistent with subglacial sediments recovered from neighboring Whillans Ice Stream, where boreholes indicate clay-rich, fine-grained, water-saturated tills with high porosity and low effective stress (Engelhardt and others, Reference Engelhardt, Humphrey, Kamb and Fahnestock1990; Tulaczyk and others, Reference Tulaczyk, Kamb, Scherer and Engelhardt1998). The similarity between these upstream measurements and the grounding-zone sample, despite its location well beyond the modern ice margin, suggests a high degree of regional consistency across the Siple Coast. This supports the interpretation that large portions of the region are underlain by compositionally similar, fine-grained sediments and justifies the use of comparable parameter ranges in ice-flow models of these systems.
This motivates the use of marine cores collected proximal to modern grounding zones to constrain subglacial properties in regions where direct subglacial access remains limited or unpublished. Such studies are widespread, including high-priority sectors like the Amundsen Sea, where studies have largely focused on reconstructing paleo-climate and ice-sheet history (e.g. Ó Cofaigh and others, Reference Ócofaigh, Evans, Dowdeswell and Larter2007; Smith and others, Reference Smith2011, Reference Smith2014, Reference Smith2017; Clark and others, Reference Clark2024). While additional geotechnical testing of these marine sediments would provide valuable constraints, similar to those presented here, the existing datasets, especially GSDs, already offer a useful baseline for defining parameter space in modeling studies. For example, Clark and others (Reference Clark2024) present data from cores in the Amundsen Sea that consist of diamictons characterized by high clay content, abundant fines and dispersed lithic fragments, similar to the Siple Coast till we present. Although their precise origin, whether deposited as traction till during basal slip or as melt-out from debris-rich ice, remains uncertain, their physical characteristics broadly resemble those inferred for subglacial tills in marine-terminating environments.
Given the geologic history of West Antarctica, where ice has overridden a former marine basin (e.g. Scherer and others, Reference Scherer, Aldahan, Tulaczyk, Possnert, Engelhardt and Kamb1998; Kingslake, Reference Kingslake2018), the presence of marine-derived, typically fine-grained sediments is expected. At the same time, terrestrial samples remain useful analogs where basal ice interacts with and actively erodes exposed bedrock. Evidence for heterogeneous basal conditions continues to accumulate, for example at Thwaites Glacier, where bedrock highs are interspersed among sediment-filled basins (e.g. Muto and others, Reference Muto, Anandakrishnan, Alley, Horgan, Parizek, Koellner, Christianson and Holschuh2019; Zeising and others, Reference Zeising2026). As constraints on local geology improve, these distinctions will refine interpretations of till genesis and distribution. For now, the ranges reported here, together with prior measurements from Siple Coast ice streams (e.g. Tulaczyk and others, Reference Tulaczyk, Kamb, Scherer and Engelhardt1998; Kamb, Reference Kamb2001), provide a reasonable basis for parameterizing subglacial sediment properties in regions where direct observations remain sparse.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/jog.2026.10208.
Acknowledgements
We thank Peter Sobol and Neal Lord for the help with the geotechnical equipment. We thank Elmo Rawling III for help with grain-size analysis. We thank Philip Bart for providing sediment from the paleo-grounding zone of Bindschadler Ice Stream (NBP1502B KC14). This work was partially supported by NSF OPP award number 2048315 awarded to L.K.Z. We thank the reviewers for the helpful comments.
Open access
