Evidence for freeze–thaw events and their implications for rock weathering in northern Canada: II. The temperature at which water freezes in rock

Many undertakings have used either a single value or a narrow window of temperatures as a threshold for the freezing of water within rock. These temperatures vary from 0 to ?5 C, with most windows being in the range ?1 to ?4 C. Based on thermal data, these thresholds are commonly used to ‘count’ the number of freeze–thaw events as a basis for determining freeze–thaw weathering. Data collected from northern Canada indicate that the temperature at which freezing occurs can vary substantially, even for the same site. Using exotherm and zero curtain observations from bricks, at angles of 90 and 45, aligned to the four cardinal aspects, the various temperatures at which water froze are shown. Bricks on the north and east commonly did exhibit freezing, based on exotherms, within the window ?1 to ?5 C, while data for the south and west aspects showed substantial variation, with freezes sometimes between ?6·4 and ?8·9 C. The data were evaluated for evidence of zero curtain effects (indicative of water freezing), but no unequivocal events could be found, and it is suggested that, at the scale of observation used here, they are unlikely. It would therefore appear that the use of thermal thresholds may not be meaningful for evaluation of freeze–thaw events. The available data also indicate many instances when temperatures went substantially sub‐zero (e.g. ?20 C) and yet no indication of water freezing occurred – most likely because there was no water available to freeze. This indicates that any form of freeze–thaw event counting, in the absence of some indicator of the presence of water and that it actually froze, is flawed, as thermal conditions alone are not adequate to indicate the occurrence or not of actual freeze–thaw weathering events. These data suggest that evaluations of freeze–thaw occurrence based simply on thermal thresholds may be substantially in error. Copyright © 2006 John Wiley & Sons, Ltd.     

Direct observation of frost wedging in alpine bedrock

Width and temperature of rock joints were automatically monitored in the Japanese Alps. Three years of monitoring on a sandstone rock face shows two seasonal peaks of joint widening in autumn and spring. The autumn events are associated with short‐term freeze–thaw cycles, and the magnitude of widening reflects the freezing intensity and water availability. The short‐term freezing can produce wedging to a depth of at least 20 cm. The spring events follow a rise in the rock surface temperature to 0 °C beneath the seasonal snowcover, and likely originate from refreezing of meltwater entering the joint. Some of these events contribute to permanent enlargement of the joint. Two other joints on nearby rock faces experience only sporadic widening accompanying freeze–thaw cycles and insignificant permanent enlargement. Observations indicate that no single thermal criterion can explain frost weathering. The temperature range at which wedging occurs varies with the bedrock conditions, water availability and duration of freezing. Copyright © 2001 John Wiley & Sons, Ltd.     

Experimental frost and salt weathering of chalk—II

Experiments are described in which chalk cubes were soaked in solutions of either sodium chloride, sodium sulphate, or magnesium sulphate at concentrations of 5·5 per cent and 12·5 per cent, or in a mixed solution of sodium chloride and sodium sulphate or in distilled water. After removal of excess liquid, the cubes were subjected to six freeze–thaw cycles with temperatures ranging from either +15 to ?10°C or +15 to ?30°C. The results confirm that frost weathering can be enhanced by the presence of certain salts, but the extent of weathering was much less than that previously reported for samples frozen totally immersed in the same liquids. Evidence is presented which suggests that salt crystallization is the major weathering process operative when non–immersed samples are frozen but a combination of frost and salt weathering operates when fully immersed samples are frozen.     

Bedrock freeze-thaw weathering regime in an alpine environment,colorado front range

Reduced major axis analysis is used to describe monthly temperature averages for daily maxima, minima, means and ranges at a sequence of bedrock microenvironments in the alpine zone of the Colorado Front Range. Seven thermistors buried at 1 cm in bedrock provide comparative data on easterly, southerly and westerly aspects, and also upon the impact of snow accumulation (?0.5m to ≥4.0m deep) against an east-facing rock wall. Intersite temperatures commonly differ by less than 5°C and, rarely, by more than 10°C. The freezing intensity of freeze-thaw cycles occurring within the confines of a seasonal snow patch rarely dropped to ?5°C, while at snowfree, vertical faces freezing dropped to ?5°C quite commonly. Comparison with laboratory established criteria for effective freeze-thaw weathering (abundant moisture and freezing to at least ?5°C) suggests that moisture rich microsites lack adequate freezing intensity, while adequately frozen sites lack moisture. Available data suggest that the overlap between freeze-thaw and hydration weathering requires careful re-evaluation.     

Simulation of moisture content in alpine rockwalls during freeze–thaw events

Rock moisture during freeze–thaw events is a key factor for frost weathering. Data on moisture levels of natural rockwalls are scarce and difficult to obtain. To close this gap, we can benefit from the extensive knowledge of moisture‐related phenomena in building materials, which is incorporated into simulation software, for example the WUFI® package of the Fraunhofer Institute of Building Physics. In this paper we applied and adapted this type of simulation to natural rockwalls to gain new insights on which moisture‐related weathering mechanisms may be important under which conditions. We collected the required input data on physical rock properties and local climate for two study areas in the eastern European Alps with different elevation [Sonnblick, 3106 m above sea level (a.s.l.) and Johnsbach, 700 m a.s.l.] and different lithologies (gneiss and dolomite, respectively). From this data, moisture profiles with depth and fluctuations in the course of a typical year were calculated. The results were cross‐checked with different thermal conditions for frost weathering reported in the literature (volumetric expansion and ice segregation theories). The analyses show that in both study areas the thresholds for frost cracking by volumetric expansion of ice (90% pore saturation, temperature < ?1 °C) are hardly ever reached (in one year only 0.07% of the time in Johnsbach and 0.4% at Sonnblick, mostly in north‐exposed walls). The preconditions for weathering by ice segregation (?3 to ?8 °C, > 60% saturation) prevail over much longer periods; the time spent within this ‘frost cracking window‘ is also higher for north‐facing sites. The influence of current climate warming will reduce effective frost events towards 2100; however the increase of liquid precipitation and rock moisture will promote weathering processes like ice segregation at least at the Sonnblick site. Copyright © 2016 John Wiley & Sons, Ltd.     

Temporal and spatial variations in periglacial soil movements on alpine crest slopes

This paper describes up to ten years of continuous monitoring of frost heave, creep and associated parameters on high mountain crest slopes in the Japanese and Swiss Alps, aiming to evaluate spatial and interannual variations in the rates and controls of soil movement. Shallow frost creep re?ecting diurnal frost heave activity dominates the crest slopes that lack a vegetation mat and have a thin debris mantle with good drainage. Seasonal frost heave activity can induce slightly deeper movement where ?ne soil exists below the depth reached by diurnal freeze–thaw penetration, although the shallow bedrock impedes movements below 20 cm depth. As a result, downslope velocity pro?les display strong concavity with surface velocities of 2–50 cm a^?1. The frost creep rates vary spatially, depending on the soil texture, slope gradient, frequency of temperature cycling across 0 °C and moisture availability during freeze–thaw periods. Soil movements recur in every freeze–thaw period, although with some interannual variations affected by the length of seasonal snow cover and the occurrence of precipitation during freeze–thaw periods. The Swiss Alps encounter more signi?cant interannual variations than the Japanese Alps, re?ecting the large variability of the annual snow regime. Copyright © 2005 John Wiley & Sons, Ltd.     

Hysteretic freezing characteristics of riparian peatlands in the Western Boreal Forest of Canada

Freezing characteristics were investigated for a sedge covered floating fen and spruce covered swamp located beside a shallow lake in the Western Boreal Forest of Canada. Thermal properties were measured in situ for one freeze‐thaw cycle, and for two freeze‐thaw cycles in laboratory columns. Thermal conductivity and liquid water content were related to a range of subsurface temperatures above and below the freezing thresholds, and clearly illustrate hysteresis between the freezing and thawing process. Thermal hysteresis occurs because of the large change in thermal conductivity between water and ice, high water content of the peat, and wide variation in pore sizes that govern ice formation. Field and laboratory results were combined to develop linear freezing functions, which were tested in a heat transfer model. For surface temperature boundary conditions, subsurface temperatures were simulated for the over‐winter period and compared with field measurements. Replication of the transient subsurface thermal regime required that freezing functions transition gradually from thawed to frozen state (spanning the ?0·25 to ?2 °C range) as opposed to a more abrupt step function. Subsurface temperatures indicate that the floating fen underwent complete phase change (from water to ice) and froze to approximately the same depth as lake ice thickness. Therefore, the floating fen peatland froze as a ‘shelf’ adjacent to the lake, whereas the spruce covered swamp had a higher capacity for thermal buffering, and subsurface freezing was both more gradual and limited in depth. These thermal properties, and the timing and duration of frozen state, are expected to control the interaction of water and nutrients between surface water and groundwater, which will be affected by changes in air temperature associated with global climate change. Copyright © 2009 John Wiley & Sons, Ltd.     

Surface‐climate attributes at Injisuthi Outpost,Drakensberg, and possible ramifications for weathering

Aerial and sub‐aerial climatic data were collected from a station at 1920 m a.s.l. in the Injisuthi region of the South African Drakensberg. Sensors monitored air temperature, soil surface and rock surface temperature, for two rock types, over the summer and winter of 2001/2002. Rainfall was measured from the summer of 2001 to January 2004. These are the first rock and soil surface‐climate data to be collected for an exposed site at this altitude in the area. Rainfall over the two calendar years 2002 and 2003 was found to be below estimates for the region, but patterns imply numerous rock wetting and drying cycles in summer. At the site, air, rock and soil temperatures differ considerably on a diurnal basis with respect to both absolute temperature and daily ranges. Mean rock daily ranges, as conducive to possible thermal fatigue, are found to be similar in the summer and winter periods. Of the two rock types monitored, the darker coloured basalt attained higher maximum and marginally lower minimum temperatures than the sandstone. Soil frost did not occur at 2·5 cm depth, but rock did reach below ?6 °C in winter. Both rock types maintain relatively high rock temperatures in winter (exceeding 25 °C), thus chemical weathering is probably only moisture restricted during this dry period. Findings highlight the importance of directly monitoring rock temperature when attempting to discern the rock weathering environment. Copyright © 2006 John Wiley & Sons, Ltd.     

WEATHERING BY WETTING AND DRYING: SOME EXPERIMENTAL RESULTS

A series of experiments on sandstone and dolerite was undertaken in an attempt to better understand the wetting and drying weathering process. As rock samples are frequently subjected to wet–dry cycles within the simulation of other weathering mechanisms (e.g. freeze–thaw), three common methods of moisture application were used and the influences of these evaluated. It was found that the method of moisture application could affect the nature of the weathering products resulting from wetting and drying. It was also observed that there were changes in the internal properties of the rock (e.g. porosity/microporosity) and that these could influence the synergistic operation of other weathering processes. Although not all of the observations could be explained, it is apparent that wetting and drying has both a direct and an indirect effect on the weathering of rock that has not been taken into account in simulations. Greater cognizance needs to be given to the role of this process both in the field and in laboratory simulations.     

Effect of seasonal freeze–thaw process on spatial and temporal distribution of soil water and its infiltration to recharge groundwater

Clarifying the distribution and dynamics of soil moisture during the freeze–thaw process is crucial for surface ecology and is an objective requirement to investigate the mechanism of changes during the groundwater recharge process in a freeze–thaw zone. Based on the monitoring data of soil moisture and temperature in the Changbai Mountain area, the freeze–thaw process is classified into four periods. This study investigates the hydrothermal migration processes during different periods. The simultaneous heat and water model is used to simulate and analyse the infiltration of soil moisture into groundwater under five precipitation insurance rates. The results are as follows: (1) The smaller the soil depth, the stronger is the correlation between soil temperature and air temperature during the freeze–thaw process. (2) The redistribution of soil moisture before and after freeze–thaw is significantly affected by the soil texture, and soil permeability affects the recharge of soil moisture from the upper region to the lower region during the thawing period. (3) Groundwater receives vertical infiltration recharge mainly during non-freezing and is supplied by freezing and snowmelt recharge during the stable thawing period. The percentage of soil water infiltration during the stable thawing period in the total annual infiltration increases gradually with the precipitation insurance rate.     

Rock thermal data at the grain scale: applicability to granular disintegration in cold environments

Consideration of the mechanisms associated with the granular disintegration of rock has been limited by available data. In most instances, both the size of the transducer and the nature of the study have negated any applicability of the resulting data to the understanding of grain‐to‐grain separation within rock. The application of microthermocouples (≤0·15 mm diameter) and high‐frequency logging (20 s intervals) at a taffoni site on southern Alexander Island and from a rock outcrop on Adelaide Island (Antarctica) provide new data pertaining to the thermal conditions, at the grain scale, of the rock surface. The results show that thermal changes (ΔT/t) can be very high, with values of 22 °C min^?1 being recorded. Although available data indicate that there can be differences in frequency and magnitude of ?uctuations as a function of aspect, all aspects experienced some large magnitude (≥2 °C min^?1) ?uctuations. Further, in many instances, large thermal changes in more than one direction could occur within 1 min or in subsequent minutes. These data suggest that the surface grains experience rapidly changing stress ?elds that may, with time, effect fatigue at the grain boundaries; albedo differences between grains and the resulting thermal variations are thought to exacerbate this. The available data failed to show any indication of water freezing (an exotherm) and thus it is suggested that microgelivation may not play as large a role in granular breakdown as is often postulated for cold regions, and that in this dry, Antarctic region thermal stress may play a signi?cant role. Copyright © 2003 John Wiley & Sons, Ltd.     

Characteristics of land surface heat and water exchange under different soil freeze/thaw conditions over the central Tibetan Plateau

Freezing and thawing processes at the soil surface play an important role in determining the nature of Tibetan land and atmosphere interactions. In this study, land surface water and heat exchanges under different freezing and thawing conditions over the central Tibetan Plateau were investigated using observations from the Coordinated Enhanced Observing Period/Asia‐Australia Monsoon Project on the Tibetan Plateau, and the Simultaneous Heat and Water Model. During the freezing and thawing stages, significant diurnal variation of soil temperature resulted in a diurnal cycle of unfrozen water content at the surface. Radiation and energy components and evapotranspiration averaged over four freeze/thaw stages also changed diurnally. On average, the surface albedo (0·68) during the completely frozen stage was sharply higher than those during the freezing, thawing, and completely thawed stages due to the snow cover. The Bowen ratios were 3·1 and 2·5 in the freezing and thawing stages, respectively, but the ratio was only 0·5 in the completely thawed stage. Latent heat flux displayed distinctly better correlation with unfrozen soil water content during the freezing and thawing stages than during the completely frozen and thawed stages. This implies that the diurnal cycle of unfrozen soil water, resulting from diurnal freeze/thaw cycles at the surface, has a significant impact on latent heat flux. A surface energy imbalance problem was encountered, and the possible sources of error were analysed. Copyright © 2011 John Wiley & Sons, Ltd.     

Freeze–thaw processes degrade post-fire water repellency in wet soils

Wildfires are a cause of soil water repellency (hydrophobicity), which reduces infiltration whilst increasing erosion and flooding from post-fire rainfall. Post-fire soil water repellency degrades over time, often in response to repeated wetting and drying of the soil. However, in mountainous fire-prone forests such as those in the Western USA, the fire season often terminates in a cold and wet winter, during which soils not only wet and dry, but also freeze and thaw. Little is known about the effect of repeated freezing and thawing of soil on the breakdown of post-fire hydrophobicity. This study characterized the changes in hydrophobicity of Sierra Nevada mountain soils exposed to different combinations of wet–dry and freeze–thaw cycling. Following each cycle, hydrophobicity was measured using the Molarity of Ethanol test. Hydrophobicity declined similarly across all experiments that included a wetting cycle. Repeated freezing and thawing of dry soil did not degrade soil water repellency, but freeze–thaw cycles decreased hydrophobicity in wet soils. Total soil organic matter content was not different between soils of contrasting hydrophobicity. Macroscopic changes such as fissures and cracks were observed to form as soil hydrophobicity decayed. Microscopic changes revealed by scanning electron microscope imagery suggest different levels of soil aggregation occurred in samples with distinct hydrophobicities, although the size of aggregates was not clearly correlated to the change in water repellency due to wet–dry and freeze–thaw cycling. A 9-year climate and soil moisture record from Providence Critical Zone Observatory was combined with the laboratory results to estimate that hydrophobicity would persist an average of 144 days post-fire at this well-characterized, typical mid-elevation Sierra Nevada site. Most of the breakdown in soil water repellency (79%) under these climate conditions would be attributable to freeze–thaw cycling, underscoring the importance of this process in soil recovery from fire in the Sierra Nevada.     

Tafoni development in a cryotic environment: an example from Northern Victoria Land,Antarctica

Tafoni are a type of cavernous weathering widespread around the world. Despite the extensive distribution of the tafoni, their genesis is not clear and is still a matter of debate, also because they occur in such different climatic conditions and on so many different types of substrate. Geomorphological characterization of more than 60 tafoni in three different Antarctic sites (two coastal and one inland) between 74 and 76° S with sampling of weathering products and salt occurrences are described together with thermal data (on different surfaces) and wind speed recorded in different periods of the year in a selected tafone close to the Italian Antarctic station. The aim of this present study is to provide further information to help understand the processes involved in the growth of tafoni in a cryotic environment, and the relationship of these processes to climate, with particular attention to the thermal regime and the role of wind. The new data presented in this paper suggest that there is no single key factor that drives the tafoni development, although thermal stress seems the most efficient process, particularly if we consider the short‐term fluctuations. The data also confirm that other thermal processes, such as freezing–thawing cycles and thermal shock, are not really effective for the development of tafoni in this area. The wind speed measured within the tafoni is half that recorded outside, thus favouring snow accumulation within the tafoni and therefore promoting salt crystallization. On the other hand, the wind effect on the thermal regime within the tafoni seems negligible. While both salt weathering and thermal stress appear active in this cryotic environment, these are azonal processes and are therefore active in other climatic areas where tafoni are widespread (such as the Mediterranean region). Copyright © 2007 John Wiley & Sons, Ltd.     

A new soil freeze/thaw discriminant algorithm using AMSR‐E passive microwave imagery

The soil freeze–thaw controls the hydrological and carbon cycling and thus affects water and energy exchanges at land surface. This article reported a newly developed algorithm for distinguishing the freeze/thaw status of surface soil. The algorithm was based on information from Advanced Microwave Scanning Radiometer Enhanced (AMSR‐E) which records brightness temperature (Tb) in the afternoon and after midnight. The criteria and discriminant functions were obtained from both radiometer observations and model simulations. First of all, the microwave radiation from freeze–thaw soil was examined by carrying out experimental measurements at 18·7 and 36·5 GHz using a Truck‐mounted Multi‐frequency Microwave Radiometer (TMMR) in the Heihe River of China. The experimental results showed that the soil moisture is a key component that differentiates the microwave radiation behaviours during the freeze–thaw process, and the differences in soil temperature and emissivity between frozen and thawed soils were found to be the most important criteria. Secondly, a combined model was developed to consider the impacts of complex ground surface conditions on the discrimination. The model simulations quite followed the trend of in situ observations with an overall relation coefficient (R) of approximately 0·88. Finally, the ratio of Tb18·7H (horizontally polarized Tb at 18·7 GHz) to Tb36·5V was considered primarily as the quasi‐emissivity, which is more reasonable and explicit in measuring the microwave radiation changes in soil freezing and thawing than the spectral gradient. By combining Tb36·5V to indicate the soil temperature variety, a Fisher linear discrimination analysis was used to establish the discriminant functions. After being corrected by TMMR measurements, the new discriminant algorithm had an overall accuracy of 86% when validated by 4‐cm soil temperature. The multi‐year discriminant results also provided a good agreement with the classification map of frozen ground in China. Copyright © 2011 John Wiley & Sons, Ltd.     

Tafoni on coastal slopes,South Devon,U.K.

Cavernous weathering (tafoni development) occurs on coastal slopes in greenschist bedrock at elevations up to 40 m above sea level. The freshly weathered surfaces of the cavern interiors are irregular in morphology, discordant to major rock structure, formed by substantially weakened rock and associated with granular weathering debris. The weathering debris contains soluble elements in proportions similar to those present in seawater, and the penetration of elements associated with sea salts into the weathering surface to estimated depths of at least 0·1–0·2 m is indicated by the presence of chlorine. Scanning electron microscopy and microprobe analyses suggest that rock breakdown occurs principally through limited chemical weathering at grain boundaries. The mechanism for the emplacement of marine salts within sheltered rock surfaces in the tafoni is postulated to be a combination of dry deposition under turbulent atmospheric conditions and wetting by coastal fog.     

Rates and processes of periglacial solifluction: an experimental approach

Cold room physical modelling of periglacial solifluction processes on an experimental slope of 12° is described, and data on soil temperatures, surface frost heave, thaw consolidation, downslope soil movement and porewater pressures over seven freeze–thaw cycles are presented. These data are analyzed in the context of laboratory determination of the rheometry of the experimental soils at high moisture contents. It is concluded that the observed thaw-induced solifluction represents pre-failure soil shear strain and results from loss of strength due to the combined effects of raised porewater pressures during thaw consolidation and upward seepage pressures as water flows towards the surface away from the thaw front. An investigation of the rheometry of thawing soils offers the prospect of an analytical model to predict rates and depths of periglacial solifluction. © 1997 by John Wiley & Sons, Ltd.     

Effects of sand-mulch thickness on soil evaporation during the freeze–thaw period

Control of evaporation from seasonally frozen soil is an important method for alleviating water shortages in arid and semi-arid areas. To investigate the inhibition of soil evaporation by sand and the major factors that influence soil evaporation, a series of field experiments with five sand-mulch thicknesses (0 cm, bare soil [BS], 1 cm [T1], 2 cm [T2], 3 cm [T3] and 4 cm [T4], with an average diameter of 1 mm) were conducted during the freeze–thaw period in Northern China. Soil evaporation characteristics in the three freeze–thaw stages were revealed and the major factors influencing soil evaporation were analysed using grey correlation analysis. The results showed that the cumulative soil evaporation decreased with increasing sand-mulch thickness during the freeze–thaw period, and only small differences in soil evaporation were observed between the T3 and T4 treatments. The reduction in soil evaporation under different sand-mulch thicknesses was 19.2–62.6% in the unstable freezing stage (P1), 2.0–28.3% in the stable freezing stage (P2) and 4.8–20.4% in the thawing stage (P3). In P1, solar radiation was a major factor influencing soil evaporation in all treatments and vapour pressure was a major factor in the sand-mulch treatments, and the influence of relative humidity on soil evaporation decreased in the T4 treatment. During the coldest P2, solar radiation was lowest so that relative humidity and wind speed became the more dominant influence factors on soil evaporation in all treatments, and surface soil water content was a major factor in the sand-mulch treatments. In P3, average air temperature and solar radiation were major factor influencing soil evaporation in all treatments and vapour pressure was a major factor in the BS and T1 treatments, whereas water surface evaporation was the major factor in the T2, T3 and T4 treatments. The results suggest that the addition of sand mulch in agricultural fields may be a beneficial practice to reduce water stress in arid and semi-arid areas.     

Frost sorting on slopes by needle ice: A laboratory simulation on the effect of slope gradient

Following a previous attempt to reproduce miniature sorted patterns on a level surface, we report the results of a full‐scale laboratory simulation on frost sorting produced by needle ice activity on inclined surfaces. Four models, with different slope gradients (5°, 7°, 9°, 11°), were designed. Stones 6 mm in diameter placed in a grid covered 20% of the surface of frost‐susceptible water‐saturated soil. These models were subjected to 20–40 freeze–thaw cycles between 10°C and ?5°C in 12 hours. The evolution of surface patterns was visually traced by photogrammetry. Needle ice growth and collapse induced downslope movement and concentrations of stones. A model produced incipient sorted circles on a 5° slope, whereas it resulted in three distinct sorted stripes on a 7° slope. The average diameter or spacing of these forms is 9.7–19.4 cm, comparable to those in the field dominated by diurnal freeze–thaw cycles. Surface parallel displacements of stone markers were traced with motion analysis software. The observed downslope stone displacements agree with those expected assuming that surface soil and stones move by repeated heaving perpendicular to the surface and vertical settlement due to gravity, although the growth of curved needle adds complexity to the overall displacements. Copyright © 2017 John Wiley & Sons, Ltd.     

Origin of honeycombs and related weathering forms in Oligocene Macigno Sandstone,Tuscan coast near Livorno,Italy

Two types of cavernous‐weathering features are exposed in the Oligocene Macigno Sandstone along 5 km of the Tuscan coast south of Livorno, Italy. Honeycomb cells (type 1 features) are typical closely spaced, more or less circular pits of centimetre scale that have been eroded 2 to 6 cm below the general surface of bedding planes or joints. ‘Aberrant honeycomb’ cells (type 2 features) are highly elongate, polygonal, or irregular ?at depressions of decimetre scale surrounded by walls rarely higher than 2 cm, some of which pass into long, free‐standing walls or tendrils. Thus, not all type 2 ‘honeycomb’ cells are fully enclosed. We measured the geometry of 551 honeycomb cells and examined various rock properties (microscopic texture and fabric, mineralogy, porosity, permeability, and chemical composition) to isolate factors that control the size, shape, distribution, and pattern of the honeycombs. Our goal was to narrow potential origins of the features and to understand their formation. The ubiquitous occurrence of sea salt in the honeycombs and scanning electron microscope evidence of physical weathering of silicates, especially micas, favours an origin for the honeycombs chie?y by salt weathering. Honeycombs do not form in siltstone, iron‐oxide‐impregnated sandstone, calcite‐cemented concretions, or in case‐hardened joints. Thus, salt weathering of type 1 and 2 honeycombs is not effective in very low permeability rocks. We propose for type 1 honeycombs that seawater is drawn into micropores of the sandstone and evolves into self‐organized diffusion cells (Turing patterns). Selective evaporation at the stationary nodes of diffusion cells, which form at the same site over time, leads to the precipitation of salt, then grains spall off, and pits are formed. The deepest pits (>40 mm) formed where Turing patterns consistently formed at the same sites. Although the walls are more porous and weathered than the host sandstone, they become selectively case hardened by an unidenti?ed component of low abundance. Initial honeycomb cell shape and gravity locally in?uenced type 1 honeycomb shapes. We suggest that type 2 honeycombs develop where diffusion‐controlled Turing patterns lead to case‐hardening along linear trends; gravity and rock fabric are important locally in in?uencing the orientation of the walls. Only type 2 cells are forming today, suggesting recent environmental changes. Gravity is not a fundamental control on honeycomb shape; in places it is a contributing factor. Pre‐existing depressions (quarry tool marks) have strongly in?uenced honeycomb shape locally. Copyright © 2004 John Wiley & Sons, Ltd.