Introduction
Two main types of sand silicification can be distinguished. On the one hand, in sandstones cemented during burial, compaction-pressure dissolution and increasing temperature hasten the deposition of silica and produce thick, fairly uniform rock units that correspond to the large clastic sedimentary series of the Early Mesozoic, such as the Permian Rotliegend and Permian Penrith sandstones and the Triassic Buntsandstein throughout northern Europe, and the Early Jurassic Navajo Sandstone in the USA. On the other hand, younger sandstones that were not deeply buried are variable in terms of facies, often tightly cemented, limited in thickness and lateral extension, and are essentially continental clastic series, such as the Sarsens in the UK, the white quartzitic sandstones in numerous Cenozoic formations in Western Europe, some quartzitic sandstone units from the Eocene Green River Formation in USA, and the diverse silicified intervals in the regolith in Australia and South Africa.
Most, if not all, silica-cemented Cenozoic continental sandstones were originally thought to have formed during warm and dry climates as part of a 100-year-old paradigm that assumed that the precipitation of silica in the surficial realm was triggered by evaporation. This idea of syn-sedimentary silicification was applied to sandstones as well as to silicification in lacustrine limestones and to silica pseudomorphs of gypsum. We realized that this hypothesis posed several problems, not the least of which was that any initial silicification/cementation would have generated a barrier to evaporation. This is what led us to consider that the silicification significantly post-dated deposition of the host sands and to look for an explanation for the silicification cementing the quartzitic sandstones in the Sables de Fontainebleau, which had never been buried to depths more than 50 m. Our model involves silicification in periglacial environments in the Paris Basin and it applies to sandstones, chert-type silicification in limestones, and to the famous gypsum pseudomorphs (Thiry and Milnes, Reference Thiry and Milnes2024a). To our knowledge, the association between silicification and periglacial environments has never before been considered and we think that it calls into question the silicification of many surficial formations that experienced periglacial conditions.
We have been able to demonstrate that silica cementation of the Sables de Fontainebleau occurred via groundwater flows, firstly by showing that the quartzitic sandstones are related to current landscapes and do not extend deeply into the Rupelian Sables de Fontainebleau (Oligocene, about 30 Ma old) stratum, making the silicification/cementation Quaternary (Thiry et al., Reference Thiry, Bertrand-Ayrault and Grisoni1988). Then we found evidence that silicification occurred during cold climatic conditions and recalled from geochemical studies that the precipitation of silica from solution could be triggered by cooling of groundwater (Thiry et al., Reference Thiry, Millot, Innocent and Franke2015, Reference Thiry, Liron, Dubreucq and Polton2017, Reference Thiry, Innocent, Girard, Milnes, Franke and Guillon2020; Thiry and Milnes, Reference Thiry and Milnes2024a). Our work was almost entirely focused on the Sables de Fontainebleau because detailed observations could be made in the extensive sand pits and the in exceptionally dense surveys and infrastructure works in the Paris region. Over decades, each of the steps in the development of the model were conceived, explored, tested, and assembled to indicate an association between silicification and periglacial environments in which hydrological dynamics were controlled by the presence of permafrost over several thousand years.
Although, we have not yet undertaken any specific research in this context, Quaternary periglacial alteration and associated silicification relating to groundwater outflows are likely to have occurred more widely wherever taiga and tundra developed during successive Pleistocene glacial stages in Europe. We expect that examples should exist to the south as far as Spain (and in mountain areas), and to the north in Germany, Central Europe, and Poland but probably not beyond because of cover by the polar ice sheet (see details in Supplementary Material).
Here we report on the geological setting for the Paris Basin quartzitic sandstones and discuss the silicification processes responsible for their formation in relation to their forms and relationships. We focus on the information that the forms and distribution of silicification may provide about the hydrological systems that are known to exist in permafrost landscapes. One objective is to explore whether an assessment can be made about the depth to which permafrost developed in the Paris Basin and so generate a basis on which a periglacial character can be demonstrated to apply to other formations.
Geological and geomorphological setting
General overview
The Paris Basin contains successive Cenozoic deposits of thick limestones overlying clays, sands, marls, and gypseous sediments. Overall, these are transgressive sedimentary sequences that advanced from the north and typically evolved towards brackish-water environments and later lacustrine sediments. The oldest sequences are restricted to the northern part of the basin while the most recent cover the southern part (Fig. 1A and C). The oldest formations in the central part of the basin (Paris area) have never been buried under more than 200 m of sediment. The Sables de Fontainebleau Formation in the south has generally never been covered by more than 20 m of limestone.
Geomorphological settings and site locations. (A) Geological map of the Paris Basin; the dotted rectangle outlines the main study area detailed in (B), trace A–B links to section in (C), A–E links to the section in Figure 2A, and C–D links to the section in Figure 2B. (B) Locations of sites mentioned in the text. Color shadings: orange = highlands ∼160 m asl through to progressively darker greens, which indicate river valleys ∼35 m asl. (C) Schematic cross-section through the superimposed plateaus; the Sables de Fontainebleau forms the escarpment between the Brie and Beauce plateaus; quartzites occur in all sand formations intercalated between limestones armoring the plateaus.
River incision and subsequent erosion exposing a stepped series of plateau landscapes in the region are Plio-Pleistocene features (Lutaud, Reference Lutaud1948; Cholley, Reference Cholley1960; Larue and Etienne, Reference Larue and Etienne2002), like those in the neighboring Belgian and London basins (Green, Reference Green, Richards, Arnett and Ellis1985; Gibbard and Lewin, Reference Gibbard and Lewin2003). Cojan et al. (Reference Cojan, Brulhet, Corbonnois, Devos, Gargani, Harmand and Jaillet2007) used the elevations and ages of river terraces in combination with an evaluation of climatic and tectonic influences on erosional and depositional processes to demonstrate that the Early Cenozoic sedimentary fill in the Basin, and an extensive Mio-Pliocene piedmont surface developed on it, were significantly modified by Pleistocene river incision and cuesta retreat.
Our study focusses on the Sables de Fontainebleau (Rupelian age, Oligocene), which forms a regional scarp between the limestone-armored Brie and Beauce plateaus south of Paris (Fig. 1A). The formation is 50-m thick, generally unconsolidated sand in which there are commonly superimposed quartzitic sandstone pans and lenses. The sand is extensively mined for glass manufacture and is fine grained, uniform in terms of particle size, and essentially horizontally layered with rare scour and trough structures. Quartzitic sandstone pans hem the plateau and lenses in underlying unconsolidated sands crop out in valley sides (Fig. 2A). Erosion of the host sands during progressive incision and retreat of the Beauce Plateau resulted in the fracturing and collapse of the quartzitic sandstone pans and lenses and the accumulation of their fractured and jumbled remains in lag deposits with multi-meter-sized blocks that characterize the Sables de Fontainebleau landscapes (Fig. 2A).
Distribution of quartzitic sandstone pans and lenses in relation to outcrops. (A) Geological section through the Beauce Plateau (from borehole data and field survey); section A–E in Figure 1A. (B) Geological section through the dissected northeastern Beauce Plateau (based on borehole data, geological map survey by Denizot, Reference Denizot1970, and field survey by MT of the Darvault-Treuzy-Levelay area); note the Late Pliocene alluvium at the top of the Bute de Trin outlier, which testifies to the importance of Quaternary erosion; section C–D on Figure 1A. (C) Diagram showing sandstone bodies recorded in sand pit Bonnevault near Larchant (see location on Figure 1B) between 1977 and 2011. Sandstone pans crop out in the thalweg and are abundant and thick in the adjacent white sand. Smaller lenses disappear away from the thalweg where ferruginous sands occur beneath the limestone cover of the Beauce Plateau. Both white and yellowish ferruginous sands result from weathering related to the proximity of the outcrop.
The quartzitic sandstones are massive, horizontal, 1- to 4-m-thick slabs extending for 10–1000 m and exposed in outcrops on the edges of escarpments and valley sides that incise the Beauce Plateau. There is a much greater variety of sandstone characteristics and morphologies than are discerned in outcrop or in clasts in the lag deposits. This is not to say that such features are rare but ‘clean’ exposures are required to appropriately identify and describe them. Some convoluted quartzitic sandstone structures, in fact, are fragile and are not preserved in outcrop. Consequently, most of our observations were made in sand pits.
Silicification
Our analysis of all drillholes that intersect the Sables de Fontainebleau (more than 400), as well as detailed data from sections made during infrastructure (public) works and observations in sand pits, show that the quartzitic sandstone pans and lenses occur only in the zone of outcrops at the edges of plateaus and valleys (Fig. 2A and B): they do not extend more than 200–500 m behind the outcrop (Thiry et al., Reference Thiry, Bertrand-Ayrault and Grisoni1988, Reference Thiry, Millot, Innocent and Franke2015). In addition, there are some superimposed sandstone horizons (Fig. 2B and C). On this basis: (1) the close relationship between the sandstones and current geomorphology indicates that silica cementation occurred in relatively recent times, after valley incision (therefore in Plio-Quaternary times) and only in the vicinity of the zones in which they currently crop out; and (2) the occurrence of subhorizontal pans and lenses suggests that the silicification was related to groundwater levels (each silicified horizon corresponding to a groundwater level) and controlled by groundwater flowing towards springs at the base of sand slopes where dissolved silica was precipitated (Thiry et al., Reference Thiry, Bertrand-Ayrault and Grisoni1988).
The age of silicification was confirmed by dating ‘Calcites de Fontainebleau’ (large acute sand-calcite crystals) encased in the sandstones prior to or contemporaneously with the silicification. Calcites de Fontainebleau are specific indicators of cold environments and their ages (based on 14C and U–Th) are centered on glacial periods (Thiry et al., Reference Thiry, Millot, Innocent and Franke2015, Reference Thiry, Innocent, Girard, Milnes, Franke and Guillon2020), the most recent being the last glacial maximum (LGM) around 25,000 years ago. The calcites precipitated as cold surface water reached warmer terrain at depth.
Silica solubility also depends on temperature: it increases with temperature but decreases, and silica precipitates, when the temperature drops (Thiry and Milnes, Reference Thiry and Milnes2024a). This law has been long known and applied by geologists to deep hydrothermal systems in which water cools and precipitates quartz in veins as it rises to the surface (Fournier, Reference Fournier, Berger, Bethke and Robertson1985) and prevails in the deposition of silica sinters from hot springs in the temperature domain of 80–25°C (Herdianita et al., Reference Herdianita, Browne, Rodgers and Campbell2000).
Data for the encased Calcites de Fontainebleau point to triggering of silicification in the Sables de Fontainebleau by cooling of groundwater as it approached and was discharged through cold surficial sands during glacial periods (Thiry et al., Reference Thiry, Liron, Dubreucq and Polton2017). It is important to reiterate that the Fontainebleau quartzitic sandstone pans and lenses are groundwater-related features and differ in all morphological characteristics and geological setting from ‘silcretes’ that derive from pedogenic environments in which the precipitation of silica is commonly attributed to evaporation in hot and dry climates (Brückner, Reference Brückner1966; Milnes and Twidale, Reference Milnes and Twidale1983; Summerfield, Reference Summerfield1983).
Methodology
The approach we have taken in this study is to put forward a set of propositions on which basis our model for relating silicification to hydrology and environment can be explained.
Proposition 1. Silica responsible for cementation of the Sables de Fontainebleau was imported in shallow groundwater solutions. Because the solubility of silica in these waters is very low (about 10–20 mg/L SiO2) it would have been necessary to provide 50–100 m3 of water to cement 1 dm3 of sand to form quartzitic sandstone (Thiry et al., Reference Thiry, Bertrand-Ayrault and Grisoni1988). Thus, a huge volume of water is needed to form a 2-m-thick sandstone pan stretching laterally over several tens of meters. This silica supply involved discharges from an aquifer for several thousand years (Thiry et al., Reference Thiry, Bertrand-Ayrault and Grisoni1988).
Proposition 2. The primary morphologies of the quartzitic sandstones (representing zones of silicification) reflect the mechanisms in play at the times of their origins, particularly the flow paths of groundwater that carried and deposited the silica. The lenses could not have formed in a single step because the porosity would have decreased and the source solution would have diverted as soon as the sand became partly cemented, thus precluding tight cementation. Progressive centrifugal growth of the silicfied bodies, such as growth in concretions may explain our observations. Detailed meso-morphologies show that the quartzitic sandstone masses grew progressively by addition of centimeter-thick ‘waves’ or ‘sheets’ of silicified sand (Thiry and Maréchal, Reference Thiry and Maréchal2001). As the initial intergranular deposits of silica decreased the porosity and permeability of the sand, groundwater flows were diverted, and a new zone of silica cement was produced. The process repeated until there was a significant change in the geochemical or hydrodynamic environment. These ‘growth layers’ are sometimes visible on the surface of the sandstone masses and reflect gradual changes in groundwater flow paths (Fig. 3A).
Formation of quartzitic sandstone bodies via successions of envelopes of silicification. (A) Superimposed sheets of silicified sands showing centrifugal development of sandstone bodies (from sand pit Treuzy-Levelay). Numbers relate to successive growth stages. (B) Concentric envelopes of silicified sand that shape a spindle-like body (from Larchant). (C) Schematic for the formation of a concentric structure in relation to the incoming silica-laden groundwater flow. Numbers relate to successive growth stages. (D) Series of aligned spindle-like sandstone bodies, which take turns in space, apparently linked to the same groundwater flow that differentiates as the cemented, and therefore impermeable, bodies develop. Blue arrows show groundwater flow direction.
Proposition 3. Distinctive spindle-shaped quartzitic sandstone bodies display concentric sheets that overlap centrifugally from a central silicified body (Fig. 3B and C). These spindles are appendages down-gradient of massive, silicified slabs (Fig. 3D). Centrifugal growth indicates that the surface of the sandstone triggers the precipitation of silica. Because the sandstone conducted hot–cold better than the water-saturated aquifer, the connection of the spindle to a larger silicified mass embodied in cold terrain cooled the groundwater and caused the silica to precipitate on its surface.
Proposition 4. Whereas thick sandstone pans that hem the plateaus are cut by large vertical and horizontal fractures (Fig. 4A), those beneath a sedimentary cover or exposed in sand pits are never fractured, suggesting that the fractures relate to recent landform processes (Thiry et al., Reference Thiry, Panziera and Schmitt1984). Our suggestion is that thermal expansion and retraction linked to climate induced this fracturing (Thiry et al., Reference Thiry, Liron, Dubreucq and Polton2017). The highly thermally conductive quartzitic sandstone exposed at the land surface reacts directly to weather changes, whereas the lower part embedded in loose sand is less sensitive to climate variations, resulting in shear stresses between the upper and lower parts of the sandstone pan that induce horizontal fracturing (Fig. 4B). These fractures, as well as sandstone-dissolution features, are commonly coated by secondary silica that was deposited after exhumation of the slabs since it only affects outcropping slabs (Fig. 4A).
Fractures and silica coatings. (A) Sketch showing silica dissolution and precipitation along discontinuities (fractures and weathering features) in sandstones exposed in escarpments hemming the plateaus. (B) Thermal expansions and retractions linked to exposure induce the fractures in sandstones in plateau escarpments. After Thiry et al. (Reference Thiry, Liron, Dubreucq and Polton2017).
Proposition 5. A synthesis of observations led to a model for the silicification on the edge of the Beauce Plateau (Fig. 5). Here, groundwater outflowing at the scarp foot, behind the spring, encounters a zone of cold ground against which silica precipitates (Fig. 5A). The establishment of persistent cold conditions would have led to deepening of the cold zone, then to the formation of permafrost. Deepening of the permafrost could have produced superimposed sandstone levels, with the youngest ones at the base of the succession. Erosion between glacial periods could have caused back-wearing of the scarp face into the plateau. The entire arrangement then would have moved towards the plateau and younger silicified levels eventually may have been interposed or even superimposed on ones formed during the previous cold period. Silica could also have precipitated at the top of the permafrost by cooling of infiltrating waters or groundwaters contained in the thawed active zone (Fig. 5B).
Hydrogeological and thermal model for silicification during Quaternary cold periods. (A) Groundwater silica precipitates by cooling at the base of cold ground or permafrost on the edge of the plateaus where calcite precipitates by warming of surface water at depth in subsoil. (B) Silica in surface water infiltrating in the vadose zone precipitates by cooling at depth in contact with permafrost. After Thiry and Milnes (Reference Thiry and Milnes2024a).
Comments on the hydrological characteristics of periglacial environments
Permafrost affects regional hydrologic regimes (Lemieux et al., Reference Lemieux, Sudicky, Peltier and Tarasov2008; Dobiński, Reference Dobiński2012). In particular, the lateral flow of groundwater is limited by permafrost, which acts like the lower limit of an impervious horizon, and the freezing of groundwater increases pore water volume and thus may increase pressure within the aquifer.
If permafrost develops at depth, the pressure increase may force laterally flowing groundwater downwards, even well below the water table. Under these conditions, groundwater flows are no longer driven by gravity and their pathways may become very tortuous and narrow, bypassing not only less-permeable aquifer zones but also frozen aquifer volumes (Yoshikawa and Kane, Reference Yoshikawa, Kane, Yang and D.L2021). When flows take place under pressure, they are squeezed against the frozen masses and may take on the form of a ‘roller coaster’ (Fig. 6).
Diagram of permafrost with variations that may be interpreted as linked to a latitudinal climatic gradient or an evolution of climate over time (cooling or warming). When permafrost thickens, part of the water in the aquifer freezes with a concomitant increase in volume and therefore pressure within the aquifer, possibly leading to fracturing of the permafrost. Groundwater flows vary depending on changing climate: (1) artesian rise during cooling periods, (2) infiltration during permafrost thawing periods. References: Lemieux et al. (Reference Lemieux, Sudicky, Peltier and Tarasov2008) and Dobiński (Reference Dobiński2012).
Even if the permafrost is thick and the water table is not recharged by infiltration, groundwater may continue to discharge following to an increase in volume of the aquifer due to freezing. Thus, under thick permafrost, groundwater may continue to discharge in valleys through unfrozen structures such as taliks. This is because the relatively warm water derived from deeper parts of the aquifer will prevent taliks from freezing as long as discharge rates are sufficient. If permafrost continues to develop, outflow structures in the valleys could gradually freeze, causing an increase in pressures within the aquifer to the point of generating hydraulic fracturing through which the water can again escape towards the surface. Thus, in northern Canada, water at 16.8°C has been recorded flowing into a river (Clark et al., Reference Clark, Lauriol, Harwood and Marschner2001). Elsewhere it has been shown that water rising to the surface comes from 600 m depth through permafrost (Andersen et al., Reference Andersen, Pollard, McKay and Heldmann2002). On the other hand, the groundwater level can fall during thawing of the permafrost through loss of volume by melting of the ice: infiltration can then penetrate deeper into the permafrost through unfrozen areas such as taliks and thermokarst features.
Above permafrost, there is usually an ‘active layer’ that thaws in summer and into which surface water infiltrates and shallow groundwater can form (Dobiński, Reference Dobiński2020). During warming periods, permafrost gradually thaws from above and a perennial unfrozen layer can form beneath the seasonally freezing and thawing active layer and may progressively thicken to up to several tens of meters. Depending on climatic conditions, the active layer plus the thawed layer may operate as a shallow, perched, supra-permafrost aquifer (Walvoord and Kurylyk, Reference Walvoord and Kurylyk2016). Infiltrating water can cool by 5–10°C from its temperature at the surface as it encounters the permafrost (Boike et al., Reference Boike, Roth and Overduin1998).
It is these specific hydrological arrangements in permafrost that provide a basis for interpreting particular configurations and morphologies of the quartzitic sandstones in the Sables de Fontainebleau and to which no special attention had been paid until now.
Results of investigations
Given that very large water volumes were necessary to nourish silica cementation of the Sables de Fontainebleau, then the quartzitic sandstone morphologies must reflect water flow paths. Furthermore, because the cemented sand had a higher thermal conductivity than the water-saturated sand aquifer, it ‘carried’ the cold front and thus induced silica precipitation and incremental centrifugal growth of the sandstone bodies. Interpreting these ‘fossilized’ flow paths makes it possible to unravel hydrological conditions prevailing during silicification.
Evidence from unusual morphologies of sandstone masses
Non-horizontal sandstone slabs were occasionally observed in Sables de Fontainebleau sand pits. An oblique sandstone slab was exposed and blasted out in 1995 in a sand pit at La Chapelle-la–Reine (Fig. 7A). An approximately 1.5-m-thick subvertical slab was also encountered in 2013 in a sand pit in Larchant. In addition, a vertical sandstone wall over 2-m thick and cut by vertical fractures (Fig. 7B) can be distinguished on a flat area at around 100 m altitude in the lower part of the rock lag that covers the western escarpment of the Rocher de la Reine south of Arbonne-la-Forêt. It is inconceivable that such subjoined, vertical, and aligned blocks could have been erected by solifluction of blocks broken away from a horizontal slab. We argue that these oblique and vertical sandstone ‘dikes’ were cemented in situ. The regularity of the cemented volumes suggests that they relate to discontinuities (breakages, faults) in permafrost through which warm groundwaters from depth were directed, thus promoting thawing of permafrost on its walls and the precipitation of silica against the contacts with the newly established cold front.
Sandstone bodies with unusual morphologies. (A) Superimposed sandstone slabs and an intercalated slab inclined at around 20°; arrows indicate slabs and their orientations; sand pit La Chapelle-la–Reine, operating front in 1995. (B) Wall-like alignment of five vertical sandstone blocks (block 2 has collapsed) that match an in situ ‘wall’ in sands; Arbonne-la-Forêt. (C) Pillars and pendants between two superimposed sandstone slabs; Boigneville. (D, E) Sandstone partitions formed by interweaving of horizontal and vertical structures that surround and delimit unconsolidated sand pockets, which emptied when released by sand pit operation; Buthiers.
Other forms of sandstone with a pronounced vertical or oblique morphology are relatively common in sand pits as well as in outcrops. A rock shelter in Boigneville has ‘pendants’ and ‘pillars’ that connect two superimposed sandstone slabs (Fig. 7C). A sand pit at Rochefort, near Buthiers, contains the most spectacular morphologies. These are extremely tortuous bodies with relatively narrow curved and anastomosed silicified walls, 2–30 cm thick, that sometimes intersect, and which delimit and surround decimetric to metric volumes of uncemented sand. During sand extraction, the uncemented sand spilled out, leaving hollows that expose the complexity of the sandstone structures (Fig. 7D and E). These morphologies and the arrangement of the sandstone walls represent permeable passages that channeled water flows through impermeable domains. This particular sand pit is at the very southern extension of the Sables de Fontainebleau. It is topped by a thick limestone cover and is the only known site where a silicified level has been found 7 meters below the current water table and, moreover, below the current valley bottom.
Fracturing of the quartzitic sandstone is restricted to the thick pans that hem the plateaus (Thiry et al., Reference Thiry, Panziera and Schmitt1984), indicating that it relates to recent landforms. In many places the fractures are coated with a bright white, glossy and silky-looking, silica skin (Fig. 8A). Silica skins may cover large areas of fractures up to 250 m2. Sometimes these silica skins also coat more irregular, rounded surfaces that were the original surface contours of the sandstone pans and lenses (Fig. 8B). These later deposits have clear geotropic characters (oriented according to gravity) and relate to water percolating downwards through sand, along fractures in the sandstone, over the surfaces of the sandstone lenses and the interior walls of pseudokarstic dissolution hollows.
Secondary silica precipitated in a vadose environment; 1 = primary quartzite, 2 = secondary silica. (A) Silica skin on a vertical fracture wall; Noisy-sur-Ecole. (B) Silica skin enwrapping pseudokarst; Buno-Bonnevaux. (C) Silica ‘puppets’ with botryoidal envelopes featuring successive overlapping concretionary growths (view upside down); sand pit Maintenon.
The curious silica concretionary ‘puppets’ in the Sables de Fontainebleau, also named ‘gogottes’ (Guillemin, Reference Guillemin1978), were collected in sand pits where the Sables de Fontainebleau rests directly on Cretaceous chalk. They display botryoidal envelopes featuring successive overlapping concretionary growths (Fig. 8C). Their orientation when in situ, particularly the downwards-projecting character of the concretionary growths, reflects a geotropic process linked to waters percolating down through the vadose zone towards the water table.
Evidence from particular arrangements of sandstone masses
The sand pit in Treuzy-Levelay is outstanding for its sandstone morphologies, unique sequence of slabs, and more massive sandstone bodies (Fig. 9A). It has at least four superimposed sandstone horizons with each showing specific morphological characters (Fig. 10).
Overview and schematic showing arrangement of sandstone morphologies in the Treuzy-Levelay sand pit. (A) The yellow sands of the N–S section result from collapse of the summit sands, which are oxidized. The E–W working face in the foreground is formed of white sands that separate sandstone masses darkened by lichens. (B) Seemingly regular and elongated slabs on the N–S working face. (C) Cylindrical bodies on the E–W working face that intersects the elongation of the silicified bodies. The letters (a–f) refer to the images (A–F) in Figure 10; the letter (g) refers to (B) in Figure 11.
Morphologies of sandstone lenses in the Treuzy-Levelay sand pit; blue arrows indicate water flow directions. (A) Horizontal sandstone slab with tapered ends. (B) Fusiform and slightly curved lenses. (C) Front view of (B) and sketch of the water flow around the lenses. (D, E) Aggregation of sandstone spindles of various orientations into compact masses. (F) Detail of silicified layers stacked on a vertical surface suggesting laminar flow along the quartzite wall. The inset sketch illustrates water flow paths related to cementation of the spindles. Location of images shown on Figure 9C. After Thiry and Milnes (Reference Thiry and Milnes2024b).
The sandstone lenses exposed in the N–S pit face (Fig. 9B) are 1–2 m thick with a flat upper surface whereas the lower surfaces have large, N–S-elongated undulations. Some of the lenses have tapered endings pointing towards the sandy escarpment to the north and others point towards the plateau located to the south. The upper horizontal surface may correspond to a paleo-groundwater table whereas the metric-sized spindle-shaped elongations reflect the flow paths of waters that fed the silicified bodies.
The slightly deeper sandstone slabs exposed in the E–W pit face (Fig. 9C, location a) are thinner (0.2–0.5 m thick), flat with rounded ends and surface indents that are elongated south–north. Their flat, triangular, spindle-like morphologies suggest horizontal laminar flows close to a groundwater table (Fig. 10A).
Deeper sandstone masses (Fig. 9C) are more substantial and include elongated and spindle-shaped forms directed towards the escarpment to the north (Fig. 10B and D). In addition, spindles are not horizontal but dip and straighten (Fig. 10B) with orientations that can diverge to form sheaves that amalgamate and fit together (Fig. 10D and E). If these spindles relate to groundwater flows, they were not ‘free’ and governed by gravity, but ‘forced’ and confined between zones of impermeable sands.
The separation of large sandstone masses at lower levels by comparable volumes of uncemented sand (Fig. 9C) raises questions. Given that the cementation was consequent upon ‘forced’ groundwater flows, it is odd that they were not deflected laterally. They appear to have been confined to specific corridors between masses of impermeable sand. Nevertheless, the arrangement of the silicified bodies along a regular surface, slightly sloping towards the west, suggests that the groundwater flow corridors were probably regulated by a common base level. The sandstone laminae superimposed on the vertical walls of certain sandstone bodies (Fig. 10F) indicates that laminar flows occurred between the silicified masses and impermeable sand and that silica was precipitated against the already silicified zone.
Several sandstone masses show odd arrangements compared with the elongated spindles described above (Fig. 11A). Successive steps in their silicification show geotropic arrangements related to water percolating through a desaturated sand above the water table after cementation of the amalgamated masses (Fig. 11B–D). The elongated ‘gullet’ that terminates several geotropic masses probably corresponds to an infiltration path with progressive spreading of large volumes of water downwards and could mark a talik structure with silica precipitating against a cold front at depth (Fig. 11D).
Complex silicified masses in the Treuzy-Levelay sand pit. (A) Isolated mass with the appearance of a seahorse on the western face of the sand pit. The mass has rotated a few degrees counterclockwise by sliding down the sand slope. Stacking of the silicified masses points to geotropic arrangements (arrows). (B) Front view (to east) of an inclined fan-like sandstone mass leaning against a large volume of amalgamated sandstone (Figure 10B, location g); note the successive ledges that overlap up slope. (C) Cross view (to south) of the fan-like sandstone arrangement. Note anchoring of the structure on the amalgamated sandstone by an elongated ‘gullet’. (D) Sketch of the relationships between the sandstone masses and an interpretation of hydrothermal conditions that prevailed during silicification. The relative position of the two sandstone structures shows that the geotropic structure formed subsequently to emplacement of the main mass.
Discussion
Sandstone morphologies versus periglacial dynamics
Our suggestion is that the morphologies of the silicified sand masses viewed in terms of hydrological flows, together with the spatial arrangements of the various sandstone bodies, make it possible to formulate a chronological sequence of phreatic flow regimes in context of the dynamics of a developing permafrost and gradual settlement of periglacial conditions in the landscape (Fig. 12). As glacial conditions became established, shallow ground cooled but did not freeze. The water table remained free, and groundwater flowed horizontally towards springs. Silica precipitated when cooled sands encroached on the water table, which is how the horizontal sandstone slabs formed (Fig. 12A).
Diagram of thermal and hydrological arrangements along valley sides when the cold environment developed (modified from Thiry and Milnes, Reference Thiry and Milnes2024b). (A) The cold front, beneath the permafrost or not, induced silicification of the sand into horizontal slabs. (B) With downward growth of the permafrost, springs in the valleys remain functional, but groundwater flow is forced against the base of the impermeable permafrost. (C) With even deeper permafrost, some flow paths may be isolated within frozen sands. (D) With deepening of permafrost below valley level and the increase in pressure due to groundwater freezing, hydraulic fracturing can occur; the resulting artesian groundwater flows can reach the land surface. (E) With permafrost thawing, the water table dropped due to melting of the aquifer ice. Here, talik structures intersect the permafrost and account for geotropic forms of silicification.
Permafrost developed as glacial conditions continued. Flow paths persisted due to a constant flow of warm water to spring outlets. However, descent of the impermeable permafrost forced flows around irregularities in the frozen front. It is along these ‘perturbed’ flows that convergent and inclined spindles were formed. (Fig. 12B).
As the permafrost thickened, phreatic flows were eventually forced into pressurized pipes (taliks) within the permafrost (Fig. 12C). At this point, previously formed sandstone masses may have been partially encased in permafrost. Depending on flow rate and the temperature of the water, a dynamic equilibrium (which could move laterally mainly depending on flow rate variations) became established and silicification of sand could have occurred against the cold quartzitic sandstone wall simultaneously with thawing of the adjacent permafrost. Laminar flows of varied orientations formed around the quartzitic masses and resulted in thin, silica-indurated layers like those in Figure 10F.
With further deepening of the permafrost, the groundwater outflows to the valleys gradually froze and closed. In the confined aquifer, freezing of the upper part of the groundwater increased its volume, leading to excess pressures, with a concomitant rise in hydrostatic level, leading to hydraulic fracturing and ultimately to artesian springs to the plateau surface (Fig. 12D). Complex, narrow flow paths in all directions were opened via the fracture network where silica precipitated in contact with the permafrost. Such phenomena explain the formation of the steeply dipping slabs (Fig. 7A and B) and the spectacular tortuous morphologies with silicified walls surrounding volumes of sand that were initially frozen (Fig. 7D and E). The same is probably true for structures that hang down from or form pillars between relatively regular slabs.
At the end of the glacial period, permafrost thawed from above due to global warming and from below due to the contribution of geothermal heat. A residual permafrost remained at depth and the hydrostatic level dropped by melting of the aquifer ice (Fig. 12E). Surface waters moved to depth through karst pipes in the limestone cover and cooled on contact with the permafrost, leading to silica precipitation that formed geotropic quartzitic sandstone masses within the sand (Figs. 8 and 11). These geotropic silicifications are witnesses to an upper thawed zone and possibly indicate the depth to the upper surface of the permafrost, which actually increases during climate warming.
Regional distribution of quartzitic sandstone morphologies in the Paris Basin
Horizontal sandstone slabs predominate in the north, along the Beauce Plateau escarpment, under a reduced limestone cover or even without a limestone cover (north of the Fontainebleau Forest). More irregular sandstone masses, with curved shapes and concavities are common farther south where there is an approximately 10-m-thick limestone cover along the valleys that incise the Beauce Plateau (Boigneville, Treuzy-Levelay). The more exotic morphologies are mainly found in the southernmost and deepest sites of the Sables de Fontainebleau in the Buthiers area (Fig. 1).
This regional distribution results from the conjunction of two independent factors—the depth of the Sables de Fontainebleau aquifer and the depth of the permafrost (Fig. 13). Thus, on the northern edges of the Beauce Plateau, where the Sables de Fontainebleau are at shallow depths, silicification produced mainly horizontal slabs at the water table. If the permafrost reached greater depths, with groundwater confinement and excess pressure, the sand was completely frozen and the excess pressure phenomena was eventually imposed on the Calcaire de Brie aquifer beneath. On the other hand, in the southern areas, where the Sables de Fontainebleau are buried under 20–30 m of limestone, possibly below the level of the valleys, it was mainly the conditions of aquifer confinement and pressurization that predominated with the formation of sandstone ‘shells’ and silicified sand ‘walls’ developed in cracks induced by hydraulic fracturing.
Geological sketch showing the deepening of the cold front over time (after Thiry and Milnes, Reference Thiry and Milnes2024a). (1) No permafrost at the start of the glacial period, which leads to formation of horizontal sandstone slabs linked to the water table. (2) Thick permafrost during the glacial maximum, with freezing of groundwater outflows and pressure increase in the aquifer leading to hydraulic fracturing and the formation of irregular quartzitic masses.
Extent of permafrost development
There are few tools that provide direct evidence of the presence, extent, and depth of permafrost during Quaternary glaciations. Most of the current estimates come from paleoclimatic simulation models (Levasseur et al., Reference Levasseur, Vrac, Roche, Paillard, Martin and Vandenberghe2011; Bertran et al., Reference Bertran, Stadelmaier and Ludwig2022). Isotopic analysis of modern groundwaters makes it possible, in certain cases, to demonstrate that they were not recharged during the last glaciation and thus to deduce that these regions were either covered by an ice cap or affected by a continuous and relatively thick permafrost (Jiraková et al., Reference Jiraková, Huneau, Celle-Jeanton, Hrkal and Le Coustumer2011). Likewise, patterned ground (with decametric polygonal structures), which characterizes soils subject to seasonal frost to a depth of 3–4 m over long periods, does not provide any indication of deeper frozen zones or their extent (van Vliet-Lanoë and Lisitsyna, Reference van Vliet-Lanoë, Lisitsyna, Paepe and Melnikov2001; Thiry et al., Reference Thiry, van Oort, Thiesson and van Vliet-Lanoë2013; Vandenberghe et al., Reference Vandenberghe, French, Gorbunov, Marchenko, Velichko, Jin, Cui, Zhang and Wan2014; Murton, Reference Murton, Murton, Bradshaw and Pillans2025). It has also been suggested that landforms resulting from solifluction may be used as proxies of the depth subjected to freeze–thaw in former periglacial landscapes (Matsuoka, Reference Matsuoka2001; Whiteman, Reference Whiteman2024) and at best may indicate a minimum depth reached by an active layer, above permafrost. But for all the above, there is no indication of the depth reached by permafrost.
The only tangible elements currently available to indicate that permafrost existed at a certain depth are provided by karst cavities. When water-filled karst cavities freeze, pure water is incorporated into the ice and the residual water is enriched in calcium which ultimately precipitates as calcite crystals. The isotopic composition of these calcite crystals shows that they come from water concentrated by frost and therefore testify that permafrost had reached these cavities (Žák et al., Reference Žák, Urban, Cilek and Hercman2004; Richter et al., Reference Richter, Schulte, Mangini, Erlemeyer and Erlemeyer2010). These cryocalcites (calcites formed by freezing) record permafrost sinking and are an archive of the minimum depth that the permafrost reached (Žák et al., Reference Žák, Richter, Filippi, Zivor, Deininger, Mangini and Scholz2012; Töchterle et al., Reference Töchterle, Baldo, Murton, Schenk, Edwards, Koltai and Moseley2024). But fewer than 10 cryocalcite karst cavities have been recognized to date in Central Europe. Moreover, not all cryocalcites indicate deep permafrost. Indeed, calcite silts composed mainly of aggregates of 3- to 5-μm euhedral crystals have been recorded in alluvial fans in southwest France and are considered to be cryocalcites precipitated by ice freezing in platy soil structures (Bertran et al., Reference Bertran, Couchoud, Charlier, Hatté, Lefrais, Limondin‐Lozouet and Queffelec2023). They occur in shallow deposits and may date to the last glacial maximum but cannot be used as a proxy for permafrost.
In the Sables de Fontainebleau, curved quartzitic sandstone spindles and arrangements of oblique and vertical slabs, as well as quartzitic sandstone forming an envelope around volumes of non-cemented sand, allows us to argue that permafrost reached at least the depth at which these sandstone morphologies occur and that overpressures existed in the water table, which was confined by freezing. Thus, with the quartzitic sandstones we have one more tangible element to suggest that the ground was frozen to at least the depth of such unusual sandstone masses. The deepest record (50 m) occurs in Buthiers where the maximum depth of the Sables de Fontainebleau occurs (Fig. 13). Although these observations remain sporadic, like cryocalcites that indicate frozen karst cavities, they constitute valuable anchor points to clarify areas likely to have been frozen to depth and provide a new method for assessing the minimum depths to which permafrost may have extended.
Conclusions
A model for groundwater silicification within the Sables de Fontainebleau in sub-contemporary landscapes in the Paris Basin was first proposed 35 years ago but has since been continually updated and recently refined with reference to studies that support silica precipitation during periglacial climate conditions. As well as the conditions under which silica was precipitated, our investigations indicate a link between the various distinctive morphologies of silicified sandstone bodies in the Sables de Fontainebleau and the flow paths of groundwaters that imported the silica. We consider the silicified masses to be fossilized groundwater flow paths from an evolving periglacial landscape. Moreover, we have evidence that permafrost extended to a depth of 50 m in part of the Basin.
We visualize a gradual settlement of periglacial conditions in the landscape, with cooling of the shallow ground, progressive descent of impermeable permafrost as glacial conditions persisted, thickening of the permafrost, and pressurization of groundwater in the phreatic zone when groundwater outflows to the valleys froze and closed. Later thawing of the permafrost with a reduction in hydrostatic level accompanied climate warming, although a residual permafrost remained at depth.
In this context, the morphologies of the silicified masses are proxies for the paleohydrology that prevailed during silicification and constitute a new basis for determining the depths reached by permafrost at the time of silicification. It is possible that our description of these structures will help in establishing benchmarks for estimating the depths to which permafrost extended on a regional scale, and possibly on a continental scale.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/qua.2025.6.
Acknowledgments
The observations and data in this manuscript were assembled from information gathered while organizing field trips in the Fontainebleau region, namely: (1) field trip at Buthiers with Association des Naturalistes de la Vallée du Loing et du Massif de Fontainebleau, Sunday, November 5, 2017 (Thiry, Reference Thiry2017); (2) field trip in Fontainebleau area with AIG‐11, Applied Isotope Geochemistry Conference, September 23, 2015 (Thiry et al., Reference Thiry, Millot, Innocent and Franke2015); and (3) field trip at Treuzy-Levelay with Astro Geo Workshop, Thursday, April 11, 2024 (Thiry et al., Reference Thiry, Milnes and Grimaud2024). We acknowledge the support of the organizations that sponsored these productions. The authors would also like to acknowledge the positive remarks and suggestions of two reviewers that have strengthened our interpretations, Prof. Julian Murton from University of Sussex, UK, and Prof. David Parcerisa from Universitat Politècnica de Catalunya at Manresa, Spain.
Statements and declarations
The authors declare they have no financial interest directly or indirectly related to the work submitted for publication.
