Figure 3E illustrates locations of sample collection down the core. Twenty-one samples were chosen; one every 1 m (starting at core section 003, at 2.4 m depth) down to 10 m, then every 2 m down to 33 m. These included samples from either side of the four observed fiducial breaks at 4.75, 7, 19.9 and 23.2 m where orientation was lost (Figures 4A, B). An additional eight pairs of samples were collected between 33 and 43 m and the deepest sample, taken from the very bottom of the core at 58.75 m, are currently unoriented. The cores were cut by band saw into 5 mm slices perpendicular to the core axis in the Physics Antarctic Ice Lab at the University of Otago. Four oriented rectangular pieces approximately 30 mm × 35 mm were cut from each slice (Figure 4C). The samples have grain sizes ∼1 mm, so the rectangular pieces contain ample grains for robust statistical analysis. The cores were then returned to storage and the cut samples were placed in a nitrogen dry shipper at ∼−196°C, and transported to Otago Micro and Nano Imaging, University of Otago for cryo-EBSD analysis.

The process for cryo-EBSD sample preparation and analysis are outlined in Prior et al. (2015). A full cross-sectional orientation map was collected for each sample with a 20 µm step size, and two to four high resolution maps of smaller areas of interest were collected with a 5 µm step size. Whole maps were montaged using AZtec software (from Oxford Instruments, 2021), and data was then exported to software package HKL Channel 5 (from Oxford Instruments, 2021), where small (<4 pixel) errors with 1° misorientations within grains were removed. Data were cleaned using EBSDInterp 1.0 (Pearce, 2015), a MATLAB based program using band contrast to remove wildspikes and reduce noise introduced by non-indexed pixels. To do this, the software assigns non-indexed pixels or wildspikes with an orientation the same as at least 6 (with each pixel being surrounded by eight pixels in total) neighboring pixels. By restricting the software to only allowing a high number of neighbors before assigning an orientation, the production of artefacts in the maps is reduced. The MTEX toolbox for MATLAB (Bachmann et al., 2011) was used to quantify microstructural parameters and perform misorientation analysis.

Whole map secondary electron images were collected during EBSD acquisition in order to analyze bubble characteristics for selected samples. Secondary electrons (SE) are ejected from the sample as the microscope’s electron beam interacts with the surface. The SE are detected and creates an image that highlights the topography of a sample surface. For some images, each bubble was counted and classified as being on a grain boundary or within a grain. The electron images were then run through the ImageJ based processing package Fiji to analyze bubble size and shape statistics. Bubble analysis was done by identifying and tracing bubble outlines directly from the secondary electron image using the automated “Analyze Particles” function. To prevent any cracks or surface impurities in the sample being misidentified as bubbles, only particles with circularities >0.2 (how closely a particle resembles a perfect circle; calculated as 4 π ( A / P 2 ) where A = area, P = perimeter, and 1.0 is a perfect circle) and areas between 0.01 and 2 mm² were selected. All electron images were oriented in the kinematic reference frame to relate bubble shapes relative to flow. Bubble area, major and minor axis length and the angle between the major axis and the x_K axis of the image (normal to flow), were collected for each bubble. From the data, the equivalent radius ( = A / π ) and axial ratios ( d long   axis / d short   axis ) have been calculated.

At 4.75 m, an ambiguous break in the core resulted in a potential loss in orientation in the field (Figure 4A). To reconstruct the orientation of the core below this break, two samples; one directly above and one directly below the break were analyzed (samples 005 and 006; Figure 4B). The c-axis orientations were plotted to compare the angles of the c-axis maxima in the sample reference frame (Figures 4D,E). The angle between samples 005 and 006 c-axis maxima is defined as the rotation angle α. Using MTEX, EBSD data were rotated by α to return sample 006 to the correct orientation as defined by sample 005, as we assume that the orientations of the c-axis maxima in both samples are the same across a depth range of 10 mm (Figure 4F). For other breaks at 7 m (between 008 and L008) and 18–20 m (between 021 and 023), the same reconstructive techniques were used. A caveat of this reconstruction method is that α could be any one of two values, 180° apart. For this analysis the smaller α value is taken. Since all samples have CPOs with ∼180° rotational symmetry around the core, this uncertainty is less significant.

Using this methodology of reorientation, we have been able to restore the fiducial line on two locations of the core (breaks at 19.9 and 23.3 m) without the use of cryo-EBSD analysis. These breaks were noted as being potential breaks when collected in the field. Two thin (5 mm) core slices cut perpendicular to the vertical core axis either side of the possible breaks were placed on a light table between two orthogonal polarized sheets at the same orientation (fiducial line E-W oriented). The proportions of grains in extinction were not matching across the break. Both samples were rotated in the same direction until the bulk of grains were under extinction. The angle between the two was found and the lower sample was rotated by that angle to restore orientation. Both sets of samples had fine layering observed in the slices—this strengthens our restoration as the layers could be aligned across both samples after rotation into bulk extinction. This method was also used to show any changes in c-axis orientations down a 2–3 m section of the core between 30 and 32 m depth, by taking samples every ∼20 cm and comparing the bulk extinction angles and layer orientations of consecutive slices.

After core orientation, all core sections can be presented in one reference frame, relative to the fiducial line as oriented at the surface. Two reference frames are used to present data in this paper, outlined below and in Figure 5:

1. The “c-axis maxima” reference frame, where all samples are rotated by angle α so that the x_C axis for that sample is aligned with the c-axis maximum (Figure 5A). The angle α depends on the sample CPO and varies from 2–58° either clockwise or anticlockwise. This reference frame allows for best comparison of microstructural characteristics. The CPO strength and most microstructural data for each sample can be presented without being azimuthally oriented, e.g., sample 063 is unoriented but its microstructure is analyzed in this reference frame.

Vertical layering was observed in the surface blue ice during field reconnaissance in 2019 (Figure 6A). Layers were exposed over several meters, at two localities (Figures 6A,C). The strike of layers at the surface is rotated 30–60° clockwise of the flow direction with some variation (bend of layers visible in Figure 6A). One isoclinal fold with a vertical axial surface sub-parallel to mean layering was observed (Figure 6B). The most visible layers appear less cloudy than surrounding ice, and detailed observations show that these comprise multiple bands of cloudy and clearer ice with thicknesses of a few mm to cm (Figure 6D). Many near vertical fractures seen at the field site cross-cut these layers at a high angle (Figure 6C). Where fractures are ice filled, the ice is bubbly with distinct bubble trails radiating from a plane medial to the fracture (Figure 6E). Photo reconstruction of vertical layers documented during the 2018/2019 field season show the layers strike approximately 38° NE.

Televiewer analysis found seven sharp planar features and three indistinct planar features above 10 m depth, and eight indistinct planar features identified below 31 m, most of which are plotted on a stereonet relative to north and the kinematic reference frame (Figure 6F). All indistinct planes are near vertical, with strikes rotated 10–50° clockwise of the flow direction, at similar orientations to the layers oriented at the surface. The three shallow indistinct features in the televiewer record correspond to steep curviplanar clear ice layers in the core. The deepest of these corresponds to a clear layer seen in core section 006 (Figure 3A). Layers are more difficult to see in the core deeper than ∼8 m. Traces of layers are visible in some thin (∼5 mm) slices of the core. Layer traces observed in horizontal core slices under cross polarized light have similar orientations to the traces that would correspond to layers in the field and indistinct planar features in televiewer analysis. The layers as seen in thin slices are discussed further in section 4.3. Sharp planar features identified in televiewer analysis strike largely NW-SE and have shallower dip angles relative to layers. These sharp features correspond to planar fractures in the core.

Planar features (layers and fractures) can be correlated between the core and the televiewer down to core section 009, at a depth of ∼8 m. The correlation confirms the azimuthal orientations of the shallowest three azimuthal lines and validates restoration across the shallowest two ambiguous core breaks (see section 3.3, Figure 3). At greater depths, features cannot be seen in the core to fully validate azimuthal orientations unless in thin section. The traces of layering from horizontal core sections are, however, consistent with the orientations of indistinct planar features in the televiewer, suggesting that the calculated azimuthal orientation of the core is reasonable at all depths where this has been attempted.

The CPOs for six representative samples collected at different depths are presented in both the c-axis maximum and kinematic reference frames (Figure 7). References to rotation directions in the following descriptions are looking down for rotations around a vertical axis. All samples have a strong CPO (Figure 7A). Where c-axis clusters are aligned horizontally at a high angle to the ice flow direction (Figure 7D). The c-axis (0001) maxima have orientations clockwise of the pole to shear plane. The c-axis orientation of sample 003 is smeared clockwise from the primary c-axis cluster. The a- <11-20> and m-axis (10-10) orientations define vertical girdles normal to the c-axis cluster. Where there is a preferred orientation of a- and m-axes in the girdle, the axes are vertically aligned and sub-perpendicular (rotated up to 40° clockwise) to the shear direction (Figures 7B,C). The c-axis cluster are aligned horizontally at a high angle to the flow direction (Figures 7D). The orientations of the horizontal trace of layers identified under cross polarized light are also presented for each sample. In all cases, the c-axis maxima are rotated anticlockwise from the pole to layering (Figure 7D).

Figure 8 comprises orientation maps in the c-axis maximum reference frame, grain size and shape data for the six representative samples down the core. The maps have colors (inverse pole figure (IPF) legend in Figure 8A) that show for each pixel which crystal direction is parallel to the c-axis maximum (x_C axis) of the sample. Grains with c-axes aligned with the sample c-axis maximum are red. A small fraction of ‘oddly’ oriented grains (oriented differently from the main cluster), defined as grains whose c-axes are oriented >20° from the x_C axis, appear in the maps as isolated yellow, green, and blue grains (Figure 8A). A notable fraction (7.7%) of these grains are observed in the shallow sample 003.

Grain size frequency distributions are presented in Figure 8B. The mean grain diameter ( D ¯ ) and the square mean root diameter D ¯ SMR   = ( ∑ ( D ¯ ) ) 2   , as defined by Fan et al., 2020, are calculated for all representative samples. While D ¯ is commonly used for grain size analyses (e.g., Jacka and Li, 1994; Obbard and Baker, 2007; Piazolo et al., 2013; Qi et al., 2017; Qi et al., 2019), D ¯ SMR reduces the bias introduced by larger grains when calculating an average, and reduces error as a result. The grain size distributions are skewed, with the majority of grains plotting toward finer grain sizes and a tail extending toward larger grain sizes. The mean and square root mean diameters are ∼900–1,200 µm in the representative samples.

Shape preferred orientations (SPOs), defined as the orientations of individual grain long axes, are compiled in Figure 8C. All samples have a relatively strong SPO at angles between 1 and 20° clockwise of the c-axis maximum (x_C axis). Samples 036, 047 and 063 (Figure 8C), as well as some samples between 5 and 30 m, have increasingly bimodal distributions of grain long axis orientations, with the second maximum oriented ±10–20° of perpendicular to the c-axis maximum. Grain axial ratio ( d L o n g   a x i s / d S h o r t   a x i s ) plots are skewed toward lower values, with some ratios extending toward values >3 (Figure 8D). Averages of the grain axial ratio in these samples are consistently 1.6–1.7. Grain boundary shapes in all samples are dominated by straight and slightly curved boundaries, with some larger grains displaying irregular grain shapes (Figure 8A). There does not appear to be any difference in the shapes of oddly oriented grains and the bulk of grains in most samples.

Layers of clearer ice were observed in full core sections taken above 8 m depth. Core section 006 contained one distinct layer dipping at a near vertical angle, and less obvious layers near parallel to this. This feature corresponds to an indistinct (or dark, not clearly defined) feature identified in televiewer logs (Figure 9A). The clear layer and near parallel layers are visible in the 5 mm horizontal slice cut to select cryo-EBSD samples (Figure 9B). The layer pattern is complex, and layer traces are rotated ∼40–60° clockwise from the flow direction. Layer traces in core slice 006 show variable thicknesses and slight bending on the centimetre scale. Isoclinal folding on the millimetre to centimetre scale has also been observed in a few slices down the core.

Cryo-EBSD analysis across these features show a subtle difference in microstructure between the layers (Figure 9C). The clear layer of ice had a significantly larger mean grain diameter of 1,410 µm. This layer has a lower bubble density relative to the rest of the ice. The c-axis maximum orientation varies slightly between adjacent layers by up to 14°. The layers illustrated as blue in Figure 9B have strong c-axis CPOs with few distinctly oriented grains, similar to the clear ice layer (Figure 9D). In contrast, the thicker layer illustrated as yellow in Figure 9B has a large number of oddly oriented grains, and a considerable resultant cluster of c-axes clockwise of the primary maximum (Figure 9D). The CPOs of the first three layers in Figure 9D show a small anticlockwise rotation left to right corresponding to anticlockwise rotation of the layers.

Intragranular lattice distortion occurs as a consequence of dislocation creep (Urai et al., 1986; Humphreys and Hatherley, 1996). It is possible to calculate the misorientation axis (rotation axis) and angle that maps the crystal orientation in one part of a grain onto the crystal orientation of another part of the grain (Wheeler et al., 2001). In typical EBSD maps, there are many pixels within a grain, and it is possible to calculate many misorientation axes between these pixels to investigate their orientation statistically and calculate an average dispersion axis for that grain (Reddy and Buchan, 2005; Michels et al., 2015). This approach is the basis of crystal vorticity axis (CVA) calculations (Michels et al., 2015; Kruckenberg et al., 2019) where the dispersion axes for multiple grains are calculated to give the aggregate dispersion axis, or vorticity axis, for a sample. This allows for evaluation of shear induced vorticity at the specimen scale, without having to rely on foliation or lineation as kinematic indicators (Michels et al., 2015). To contribute to a CVA using MTEX, grains require at least three indexed solutions (or pixels) and have a grain orientation spread (GOS) of >1°. For each Priestley sample, 50–130 grains met these conditions and were used to calculate CVA.

Intragranular distortion in the Priestley ice samples, much like other naturally deformed ice samples (Monz et al., 2021; Steinbach et al., 2017; Weikusat et al., 2017), comprises mostly discrete, sharp, low angle boundaries (Figures 10A,C) rather than continuous lattice bending. Grains oriented distinctly from the preferred crystallographic alignment direction have significantly less intragranular distortion with no apparent subgrains (Figures 8A, 10A). Some of these low angle boundaries may be subgrain boundaries, developed through recovery of dislocations within grains (Urai et al., 1986), but some may be there by chance, simply as a geometrical consequence of the very strong CPO (Wheeler et al., 2001). Distinguishing these requires comparison of the misorientations across neighboring pixels (neighbor-pair) with the misorientations from randomly picked pixels (random-pair) within the sample: the random-pair misorientations represent the geometrical consequence of the CPO (Wheeler et al., 2001).

Figure 10B shows the relative frequency of neighbor and random-pair misorientation angles. In all samples the neighbor and random-pair distributions for angles >10° are comparable, suggesting that these are geometrical consequences of the CPO. The tight peaks between 20 and 35° in all but the shallowest sample reflect the strong single c-axis maximum CPO. The wider range of misorientation angles in sample 003 relates to the population of differently oriented grains in this sample. In all samples, the neighbor-pair peak in the 5° bin is significantly larger than the random-pair peak, suggesting that there is a population of low angle boundaries that is not a geometrical consequence of the CPO. Mean grain orientation spread (GOS) values, which is the average misorientation between all pixels in a grain and the grain mean orientation, remain low in the representative samples with <1° for all samples. These values are low enough to be potentially affected by the limitations of the EBSD resolution (±0.5°).

The sample random-pair axis distribution and CVA are shown in Figures 10D, E. In the deepest four samples the CVA data show two maxima; one in the middle of the stereonets corresponding to the vertical direction and a second parallel to x_C, the c-axis maxima. These maxima could be related to subgrains that represent intragranular distortion or to low angle boundaries that are a geometrical consequence of the CPO. To distinguish between these options, the CVA data need to be compared with the random-pair misorientation axes plotted in Figure 10D. For the four deepest samples the vertical maximum seen in the CVA is absent or much diminished in the random-pair misorientation axis plot. In contrast the maximum parallel to x_C is well developed in both plots. These observations suggest that the CVA maximum parallel to x_C is a geometrical consequence of the CPO whilst the vertical maximum represents the true vorticity axis for the deformation of the ice within the shear margin.

The two shallower samples are different. In sample 003 the vertical maximum is present in both CVA and random-pair plots and the maximum parallel to x_C is absent from both. We cannot make any robust statements about vorticity in this sample. The CVA of sample 013 is similar to the deeper samples but the random pair misorientation axes subtly different. However, the vertical maximum is present in the CVA and absent in the random-pair plot, suggesting a vertical vorticity axis.

Small bubbles with diameters 100–500 µm are present throughout the Priestley Glacier ice core. These bubbles were counted manually in three samples to determine proportions of bubbles on grain boundaries or within grains: 77–87% of bubbles were found along grain boundaries. Secondary electron images of the bubble structure of sample 020 are presented in Figure 11, showing relatively spherical bubbles in the horizontal plane, with a small proportion appearing elongated or connected (Figures 11A,B). Bubbles can also cluster in and around grain boundaries (Figure 11B). It is worth noting that these are 2D assumptions of bubble orientation and 3D analysis of bubble shapes is ongoing. The 3D nature of bubbles may introduce bias when calculating shape preferred orientations or when locating bubbles at grain boundaries or in grains, as we cannot say what the bubble shape is below the analyzed surface.

Bubble diameter frequency diagrams are skewed toward lower bubble sizes, with all bubbles having diameters <1 mm (Figure 11C). The mean and square mean root diameter for bubbles remain constant throughout the majority of the core, between 200 and 240 µm. The deepest sample 063 displays decreased bubble diameters, with a significantly smaller d_SMR of 126 ± 4 µm. Peak bubble diameters for all samples are constant, between 80 and 140 µm. Bubble mean and square mean root axial ratios decrease slightly with depth, from ∼1.7 to 1.3 (Figure 11D). The diagrams are again skewed to the left with peak axial ratios 1.1–1.2 and have a small percentage of bubbles with axial ratios >5, presumably identifying connected bubbles as very elongate single bubbles.

There is a strong bubble long axis SPO generally aligning parallel to sub-parallel (∼10°) to the flow direction in the shallow core. In samples below 20 m the bubble SPOs are rotated clockwise toward the pole to shear direction (Figure 11E). All samples display a second preferred orientation normal to the primary preferred orientation. SPOs for bubbles with axial ratios >1.5 show strong preferred alignments of elongated bubbles. The elongate bubble SPOs for shallowest samples 003 and 013 are generally aligned parallel to flow in similar orientations to SPOs for all bubbles. In the deeper samples, the SPO of elongated bubbles are similar to the SPO for all bubbles (Figure 11E). All samples have multiple minor maxima across a range of orientations with the exception of sample 023, which has a significant spread of orientations.

Trends of quantitative CPO strength and orientation, grain size and shape parameters have been plotted against depth in Figure 12. Grain long axes are consistently oriented similarly to the c-axis maximum and both are at a high angle anticlockwise to shear (or the flow direction, 180° in this reference frame). The plot of mean c-axis maxima and vector averages of SPO in Figure 12A shows an oscillatory pattern of orientations down core. From the surface to 16 m, the c-axis maxima and SPO average rotates from approximately parallel to the pole to shear (orthogonal to the flow direction), to ∼30° clockwise to the pole to shear at 9–10 m, then back to shear-parallel at 16 m. This pattern repeats from 16 to 33 m. The SPO average lies 1–20° clockwise of the c-axis maxima at all depths; maximum mismatch occurs when the c-axis maxima lie ∼30° clockwise to shear. Bubble SPO averages change with depth, rotating from ∼10° clockwise of the flow direction to 80° clockwise of the shear (flow) direction, near parallel to the pole to shear (Figure 12A). Layer orientations plot across a range of azimuths approximately perpendicular to the c-axis maximum. In all cases the pole to layer traces are clockwise of the c-axis maximum.

Through the whole dataset, the percentage of oddly oriented grains decreases from 7% to <1% with a few jumps at ∼9 and 17 m, then increases from ∼22 m toward the deepest sample, with 4% odd grains (Figure 12B). The amount appears to decrease again to <1% below 40 m. The quantitative fabric strength, defined by the M-index (Skemer et al., 2005) and c-axis eigenvalues (Woodcock, 1977; Mainprice et al., 2015), are calculated for all samples and plotted against depth (Figures 12C,D). The M-index defines the difference between a theoretically uniform orientation distribution function (ODF, a quantitative description of the CPO; Bunge, 1982) and the sample random-pair misorientation distribution. Values range from 0 (random fabric) to 1 (single crystal fabric). M-index values in the Priestley samples increase from the near surface down to 10 m, then remains relatively constant as depth increases (Figure 12C). Eigenvalues are calculated from the second order orientation tensor of the c-axis distribution and can be used to quantitatively describe the strength of a samples c-axis distribution. In this study, the three eigenvalues have values E₁>E₂>E₃, where E₁+E₂+E₃ = 1. The primary eigenvalue E₁ is also lowest in the top 10 m, increasing to a constant value of 0.9–0.95 below 10 m. Eigenvalue E₃ has constant low values of ∼0.01–0.02 below 10 m with values that rise slightly in the upper 10 m. E₂ has an inverse pattern to E₁. With maximum values of ∼0.07, and minimum values ≥ E₃. Samples with lower E₁ and comparatively higher E₂ and E₃ values are those with a population or smear of grain lattice orientations outside of the primary c-axis maximum (Figure 12D). Sample 063 has similar M-index values and eigenvalues to samples from 40 m depth.

Graphs of and D ¯ S M R against depth for all samples in Figure 12E show little significant change in grain size, and this appears to extend to the bottom of the core. Grain axial ratios in Figure 8 plot around 1.6–1.8 (Figure 12F). Bubble diameters are relatively constant down the core. Bubble axial ratios are variable, and the dataset is too sparse to identify any patterns.

The percentage of grains with GOS values >2° and grains with GOS values >5° gives a measure of how many grains contain subgrains (Figure 12G). Although there is some variability, the percentage of grains with increased misorientations are largely constant with depth. In the top ∼10 m, this plot has an inverse trend to the percentage of oddly oriented grains (Figure 12B). It is worth noting that the percentage of grains with higher degrees of misorientation are <10% throughout the core; across all samples, relatively few subgrains are observed.

The intensities of the CVA and random-pair misorientations need to be used together to infer vorticity related to intragranular deformation. The only significant maxima in both parameters, at all depths, are sub-vertical or close to parallel with the c-axis maximum (Figures 10D,E). The values of both CVA and random-pair misorientations parallel to the c-axes are similar in all samples and there is no consistency in whether CVA or random-pair misorientations have larger values in this direction (Figure 12H). It is thus likely that the apparent CVA parallel to c-axis maxima is a geometrical consequence of the strong CPO (see section 4.2). In contrast, the CVA in the vertical direction is significantly larger than the random-pair misorientations in nearly all samples below 10 m (Figure 12H), suggesting a vertical CVA relates to intragranular deformation. At depths shallower than 10 m, CVA and random-pair misorientations have similar vertical values.

The sub-parallel layers with isoclinal fold structures, anastomosing layer patterns and local low angle truncations are typical of planar structures in high strain zones in rocks (Berthe et al., 1979b; Lister and Snoke, 1984; Toy et al., 2012). High shear strain will cause planar features of a wide range of orientations to become sub-parallel (Ramsay, 1980) and in a high shear strain zone of a glacier; original sub-horizontal annual layering and ice filled fractures (veins) will all become sub-parallel (Hooke and Hudleston, 1978; Hudleston, 2015; Jennings and Hambrey, 2021) while non-planar original features may be strained to generate new sub-parallel planar layers (see Figure 14 in Hudleston, 2015). The sub-parallel vertical layers in the Priestley core are consistent with the location in the lateral shear margin, however the strike orientation of the layers is rotated clockwise of that expected. Parallelism of layers is achieved with shear strains of around 10 (Norris and Cooper, 2003; Hudleston, 2015). At shear strain rates of 6 × 10⁻¹⁰ s⁻¹ (Still et al., this volume) and a site ice velocity of 90 m per year, a shear strain of 10 is achieved in approximately 50 km of downstream motion. Based on our observations and measurements it is highly likely that the ice at the Priestley Glacier has experienced this much finite strain. However, a homogenous shear strain of 10 would result in a finite extension direction, that would approximate the parallel layer strike orientation, <10° (Ramsay, 1980) from the shear direction (flow direction). Layer strikes are mostly 30–60° clockwise of flow and vary depending on location and depth. These variations are best explained by inhomogeneity in the shear leading to local rotations or by generations of new layer/foliation structures that represent less finite shear strain.

CPOs that develop as a result of deformation share the symmetry of the deformation kinematics (Wenk and Christie, 1991). The observed pattern of strong horizontal single maxima with a clockwise (looking down) smear of differently oriented grains has monoclinic symmetry, with a vertical 2-fold rotation axis and a horizontal mirror plane (applying these symmetry operators does not change the pattern). This is consistent with the vertical crystal vorticity axes (CVAs) constrained for all samples deeper than ∼10 m (Figure 12H). Deformations with a significant component of simple shear have monoclinic symmetry, with the vorticity axis in the shear plane and perpendicular to the shear direction (Michels et al., 2015). The observed CPOs are all consistent with shear deformation with a vertical shear plane and a horizontal shear direction (Figure 13).

All laboratory experiments where ice deformation is dominated by simple shear (Kamb, 1972b; Bouchez and Duval, 1982; Li et al., 2000; Wilson and Peternell, 2012; Journaux et al., 2019; Qi et al., 2019) result in a strong maximum (M1 max) of c-axes aligned normal to the shear plane. Some experiments also have a secondary maximum (M2 max) in the plane that contains the shear plane normal and the shear direction but rotated away from the primary (M1) maximum in the direction opposite to the sense of vorticity (Figure 13 schematic in middle section). The angle between the M1 and M2 maxima decreases with strain (Qi et al., 2019) and is also affected by deformation conditions, such as temperature and stress (Qi et al., 2019). The preferred a-axis orientations resulting from shear (Journaux et al., 2019; Qi et al., 2019) lie in the shear plane, perpendicular to the shear direction at low strain and parallel to the shear direction at high strain. The Priestley Glacier shear margin CPOs have characteristics very similar to CPOs from shear experiments. This is shown in Figure 13, where two representative CPOs from the Priestley data set are plotted for comparison with experimental data.

Single and double cluster fabrics are both observed in the Priestley Glacier ice core. A weak M2 maximum seen in sample 003 is geometrically similar to both the high stress, high strain experimental sample PIL144 and relatively low strain, low stress sample PIL94 deformed at a relatively warmer temperature (Qi et al., 2019). Deep sample 047 has a similar CPO geometry to experimental sample TGI1.96, deformed at a low stress at a higher temperature than that observed in the Priestley Glacier margin (−7°C, compared to ∼ −20°C; Figure 3B). Experimental samples were deformed to lower shear strains than those expected for the Priestley shear margin (Figure 13), and these strains were achieved at significantly higher strain rates (1 × 10⁻⁴ to 1 × 10⁻⁷ s⁻¹, compared to 6 × 10⁻¹⁰ s⁻¹ at the field site).

There are few measured ice CPOs from samples with well constrained deformation kinematics, very few from shear zones and even fewer from lateral shear margins. Natural shear zone CPO data that can be compared with our data are plotted in Figure 13. The best constrained data from natural ice shear zones are from basal shear of the Barnes ice cap (Hudleston, 1977; Hudleston, 1980). This is the only data set for natural ice deformation where the samples can be linked to both kinematics and a progression in strain. The data for ice deformed at −10°C match the results from up-strain laboratory experiments (Bouchez and Duval, 1982; Journaux et al., 2019; Qi et al., 2019) in having a primary (M1) maximum normal to the shear plane and a secondary maximum that rotates towards the M1 maximum with increasing shear strain. Data from very coarse-grained ice in a marginal shear zone of Storglaciären have two maxima with the M1 maximum close to normal to the shear plane and a weak M2 maximum (Monz et al., 2021). These data also show the a-axes to be aligned normal to the shear direction in the shear plane. Some samples from near the base of Talos Dome (Montagnat et al., 2012) have tight vertical maxima, related to a dominance of shear on a horizontal plane, with oddly oriented grains rotated 40°–50° from the maximum. Published CPO data from lateral shear margins are from boreholes with no azimuthal orientation information. CPOs from lateral shear zones in the margins of the Whillans ice stream have two horizontal maxima, one being more intense than the other (Jackson and Kamb, 1997; Jackson, 1999). Recent data from the margin of the Jarvis Glacier, Alaska show a preponderance of horizontal c-axes (Gerbi et al., 2021).

The kinematics of the Priestley shear margin, as measured by stakes spaced 50–100 m, is dominated by simple shear (Still et al., this volume). The measured c-axis maxima oscillate twice down depth between normal to the shear plane as defined by the velocity data (the vertical plane containing the velocity vector: Figure 1), and ∼30° clockwise to the pole to shear direction. The simplest explanation of this is that the individual CPOs have undergone rigid rotations that postdate their formation, or that rigid rotation processes operate faster than the CPO evolution can respond to changing kinematics. Such rotations are also needed to explain the orientation of layering as outlined in section 4.1. The range of clockwise rotations of layering, from shear plane parallel, is similar to the range of clockwise rotations of c-axis maxima, from shear plane normal. Moreover, the pattern of layer trace orientation with depth in the top 12 m is similar to the pattern of c-axis maximum orientations (Figures 6, 12). Where CPOs are measured across layers (Figure 9), the c-axis maxima rotate with layer orientation. These observations suggest that layer and c-axis maxima orientations are linked. Layer traces always lie anticlockwise of the c-axis maxima, by less than 90°, most commonly by about 70°; this relative orientation is consistent with sinistral shear with the shear plane normal to the c-axis maximum.

The structures that give rise to the rotations of CPOs and layers could include folding and bending of the layers as seen in the field (Figure 6) and in the core (Figure 9). Rotations of CPO with the relatively small orientation changes of layers has been documented in ice deformation experiments (Wilson, 1983; Wilson et al., 2003) and has been reported for natural quartz CPOs (Carreras et al., 1977) and is observed in layered sections of the Priestley ice core (Figure 9). Rotation could also be related to localization of deformation. The localized deformation could be ductile, and evidence for such structures is seen in the low angle truncations of layers (Figure 6) or brittle, such as the dominant set of NE to NNE fractures upstream of the borehole. Blocks of ice between the localized zones undergo rotation (Platt and Vissers, 1980; Lister and Snoke, 1984). The rotations required to explain the CPOs would be around a vertical axis, consistent with localization of horizontal motion on vertical planes.

The changes in CPO orientation with depth (Figure 14) would require heterogeneity in the shear kinematics on the meter to 10 m scale. Such heterogeneity should be expected; it is commonplace in highly anisotropic rocks from shear zones (Berthe et al., 1979a; Platt and Vissers, 1980; Lister and Snoke, 1984; Goscombe and Passchier, 2003; Toy et al., 2012) and such complexity of structure has been described in glaciers (Kamb, 1959; Hambrey, 1977; Hooke and Hudleston, 1978; Hudleston, 2015; Jennings and Hambrey, 2021). Such heterogeneity may integrate over the scale of many tens of meters to approximate the broad kinematics of the shear margin. GNSS positioning data from the Priestley Glacier shear margin (Still et al., this volume) show that the kinematic fields calculated from the measured velocities of neighboring triangles of marker stakes differ (e.g., the extension direction changes) in a manner more complicated than a simple increase in simple shear strain rate towards the margin. These data may reflect the short-range complexity within the shear margin kinematics we infer from the CPO and layer data.

Dominant grain long axis (SPO) orientations almost exclusively lie clockwise of the c-axis maximum (Figures 12, 14). The difference in azimuthal orientation between the c-axis maximum and SPO increases by up to 20° at depths where the c-axis maximum azimuth is at a large angle clockwise from the x_K axis. The grain long axis orientations are not oriented ∼45° anticlockwise of the c-axis maximum as expected from experiments by Journaux et al. (2019). Secondary maxima of grain long axes, where they occur, lie 60–90° anticlockwise of the c-axis maximum and would be consistent with the finite extension direction expected for simple shear on the plane normal to the c-axis maximum, corresponding to a shear strain of ∼2. In the deepest three samples where bubble long axis orientations were measured, averages lie ∼80° anticlockwise of the c-axis maxima, corresponding to a finite shear strain of ∼4. Neither grain shapes nor bubble shapes are likely to record finite strain as both can be reset during deformation; grains by recrystallisation (Lister and Snoke, 1984) and bubbles by a drive to remain spherical (Hudleston, 2015; Dadic et al., 2019). Bubbles in the upper ∼10 m of the core could have complex histories (Dadic et al., 2019) related to seasonal temperature gradients (Figure 3D).

The dominant orientation of grain long axes 1–20° clockwise of the c-axis maxima (approximately normal to the shear plane) is unexpected. The relative proportion of shear normal extension to simple shear (Still et al., this volume) is unlikely sufficient to explain this (Fossen and Tikoff, 1998). It is possible that there are recrystallization processes that give unexpected long axis orientations. Another possibility here is that the long axes reflect the cyclic tidal flexural strains that occur in addition to continuous shear at this site (Still et al., this volume).

Studies have shown that dislocation creep is the main deformation mechanism in polar ice sheets (Duval et al., 1983; Weertman, 1983; Shearwood and Whitworth, 1989; Duval et al., 2010). The CPO of polycrystalline ice is initiated by this process (Wilson et al., 2014). Microstructural analyses on deep Antarctic ice cores describe changes in recrystallisation mechanisms with depth (Gow and Williamson, 1976; Lipenkov et al., 1989; Azuma et al., 2000; Obbard and Baker, 2007; Durand et al., 2009; Weikusat et al., 2009). Typically, grain size increases linearly with depth in shallow ice (Alley et al., 1986; Thorsteinsson et al., 1997; Durand et al., 2006), interpreted as the result of grain growth. A transition is observed, below which grain sizes remain constant as depth increases, indicating grain growth is balanced by grain nucleation, that occurs by subgrain rotation (SGR) recrystallization. Corresponding CPOs are strong single maxima. Transitional depths vary in different deep ice cores (Duval and Castelnau, 1995; Faria et al., 2014a). A second transition occurs to a basal warm ice facies, at depths where temperature increases to ≥10°C. Basal ice is very coarse grained with irregular, often amoeboidal grain shapes and lobate grain boundaries, interpreted as the result of dynamic strain induced grain boundary migration (De La Chapelle et al., 1998). CPOs in the warm ice are typically multimaxima (Hellmann et al., 2021) but can be poorly defined due to poor grain statistics (Hudleston, 2015) and recent work (Monz et al., 2021) has suggested that these are, given a shear dominated kinematic regime, probably double maxima.

The Priestley ice microstructures, with relatively regular sub-polygonal grain shapes and strong single maximum CPOs, are most comparable to the constant grain size zones of deep boreholes. Low angle subgrains, though proportionally minor, attest to the operation of a recovery and SGR process. Irregular grain boundaries suggest that strain-induced GBM also operated. GBM operating alone will lead to grain growth and requires a nucleation mechanism to maintain grain sizes (Urai et al., 1986). There is no evidence of “core and mantle” grain structures in the Priestley Glacier samples, where small, recrystallized grains are arranged around large, strained parent grains during continuous rotation recrystallization (White, 1976). Such structures are seen readily in low temperature ice deformation experiments from Fan et al. (2020) but are not readily observed as higher strain microstructures (Qi et al., 2019; Fan et al., 2021) that may represent a steady state.

Clockwise smears of the c-axes and oddly oriented grains could represent nucleation during GBM (Halfpenny et al., 2006; Urai et al., 1986). Oddly oriented grains at Talos Dome (Montagnat et al., 2012) are related to localized GBM (discontinuous recrystallization) events. The oddly oriented grains represent a poorly developed M2 maximum; development of this maximum would be related to the operation of GBM (Qi et al., 2019). Samples with oddly oriented grains have lower relative frequencies (>10%) of low angle grain boundaries. This is seen in samples at a range of depths, and varies between different layers (Figure 9), but is most clear in the upper 10 m. While there appears to be a slight inverse relationship between the amount of oddly oriented grains and percentages of high GOS grains (>2° and >5°), this does not necessarily extend to greater depths; the percentage of grains with high GOS values remains largely constant with some variability as depth increases (Figure 12). This variability may come from less robust statistics, as samples do not contain abundant grains with high degrees of misorientation, or subgrain boundaries. A slight increase in grains with high intragranular misorientation at depth would suggest an increase in strain energy due to high dislocation density, that could drive fast rates of grain boundary mobility despite a decrease in temperature (De La Chapelle et al., 1998; Fan et al., 2021). The temperature nears 0°C in the topmost section of the core (Figure 3B) which may encourage GBM recrystallization and growth of oddly oriented grains in summer months.

The microstructures and CPOs of the Priestley samples contrast with those from the Whillans ice stream margin (Jackson and Kamb, 1997; Jackson, 1999), the only other samples collected from a cold polar lateral shear margin. The Whillans samples are coarser grained, with much more irregular grain boundaries and have clear double maxima CPOs, all suggesting a dominance of GBM processes. The in situ temperatures are slightly colder (−22°C vs. −19°C) and strain rates an order of magnitude faster (5 × 10⁻⁹ s⁻¹ vs. ∼6 × 10⁻¹⁰ s⁻¹) at Whillans. Given that lower temperatures and faster strain rates will correspond to higher stresses one would expect Whillans ice stream samples to be finer-grained (Jacka and Li, 1994) relative to Priestley and to have microstructures inferring a greater significance of SGR versus GBM (Qi et al., 2017; Qi et al., 2019; Fan et al., 2020). This disparity may be explained by the Priestley microstructure having been modified by higher stress cyclic deformation processes associated with tidal flexure as documented by diurnal strains in phase with tidal elevation at this site (Still et al., this volume).