The making of Mt Everest: channel flow and low-angle normal faults in the compressional Himalayan orogen

The Himalayan channel flow model is defined as the southward ductile flow of a weak, viscous, mid-crust layer between relatively rigid, yet deformable, crustal slabs above and below (e.g. Grujic et al. 1996, 2002; Beaumont et al. 2001, 2004; Searle et al. 2003, 2006; Law et al. 2004, 2011; Searle and Szulc 2005; Godin et al. 2006; Fig. 5). The mid-crust GHS slab is bounded by the MCT ductile shear zone, a south-vergent thrust along the base concomitant with an inverted metamorphic field gradient (Hubbard 1989; Searle et al. 2006, 2018; Jessup et al. 2008a, b), and the north-dipping STD low-angle normal faults concomitant with right-way-up metamorphic isograds along the top (Burchfiel et al. 1992; Searle et al. 1997, 2003; Law et al. 2004, 2011; Cottle et al. 2007, 2015; Searle 2010; Waters et al. 2019). Both the MCT and STD ductile shear zones were active simultaneously (see summary of chronology in Godin et al. 2006). Two end-member models for channel flow are the Poiseuille flow (‘pipe flow’) model, where the highest velocities occur in the middle of the channel, and the Couette flow model, where the highest velocities occur either along the top or the base of the GHS slab (Law et al. 2004, 2011; Grujic et al. 1996, 2002; Klemperer 2006). High-temperature ductile fabrics are commonly seen throughout the GHS above the MCT zone and beneath the STD zone. The migmatite–leucogranite zone along the upper part of the GHS records melting and deformation associated with southward-vergent channel flow (Searle et al. 2010).

In situ crustal melting to form migmatites and leucogranites is the key to the processes of mid-crust channel flow. Leucogranites are abundant along the upper part of the GHS and have the composition quartz–K-feldspar–plagioclase–muscovite–tourmaline–biotite–garnet, with some later intrusions containing magmatic cordierite (altered to pinite) and/or andalusite. The leucogranites are highly peraluminous and have a high boron concentration (abundant tourmaline) and high ⁸⁷Sr/⁸⁶Sr ratios, indicating a purely crustal melting origin (Harris and Massey 1994; Harris et al. 1995; Hopkinson et al. 2020). They are spatially related to a widespread sillimanite-bearing migmatite zone and consist of multiple phases of intrusions (Searle et al. 2010; Streule et al. 2010; Searle and Cottle 2023). The earliest evidence of crustal melting is small, plagioclase-poor, kyanite + quartz-bearing leucosomes within the GHS gneisses. These melts formed at c. 720–710°C and c. 10 kbar during the period 25–18 Ma (Iaccarino et al. 2015).

The GHS metamorphism evolved from late Eocene–Oligocene kyanite-grade metamorphism to lower pressure and higher temperature sillimanite-grade metamorphism. During the later stages cordierite formed at pressures <5 kbar. Most leucogranites were formed by muscovite or biotite dehydration reactions during the time period 25–15 Ma (Cottle et al. 2009a, 2015). Mid-crustal melting during the early to mid-Miocene likely triggered the large-scale ductile flow of a partially molten layer of mid-crust bounded by the MCT ductile shear zone below and the STD ductile shear zone and low-angle normal faults above. It must be emphasized that the Himalayan mid-crust channel flow model is a different process and is distinct from the lower crust flow model, which has been surmised to explain the eastward flow of thickened Tibetan crust (65–75 km thick) away from the Tibetan Plateau (e.g. Royden et al. 1997; Clark and Royden 2000; Klemperer 2006).

Geological mapping, combined with detailed thermobarometric studies across the GHS, has defined the thermal structure of the Himalaya. Along the base of the GHS, metamorphic isograds demonstrate a structurally inverted P–T profile across the MCT zone, with higher P–T grade rocks above lower P–T grade rocks. High-strain ductile shearing has condensed the isograds, showing that the southward extrusion of the hanging wall occurred along several zones of high strain (Hubbard 1989; Law et al. 2004, 2011; Jessup et al. 2006, 2008b; Searle et al. 2018). Along the top of the GHS, the mapping of metamorphic isograds demonstrates a structural right-way-up sequence of isograds, with decreasing metamorphic grade towards higher structural levels along the STD zone. In the Everest area, large-scale, low-angle normal faults cut and truncate the metamorphic isograds. The upper Qomolangma Detachment shows the largest jump in P–T conditions, with relatively unmetamorphosed Ordovician limestones overlying the metamorphic rocks of the Cambrian Yellow Band and the Everest Series (equivalent to the North Col Formation) schists and gneisses in the footwall (Searle 2003; Sakai et al. 2005; Jessup et al. 2006; Myrow et al. 2008).

Recognition of the southward extrusion of the Himalayan mid-crust initially came from the mapping of metamorphic isograds. In the NW Himalaya, Searle and Rex (1989) showed that right-way-up isograds along the footwall of the STD (or the Zanskar Shear Zone in the western Himalaya) could be spatially linked to inverted metamorphic isograds along the MCT zone in the Kishtwar Window. This geometry defined a large-scale SW-vergent fold of the metamorphic isograds. This fold nose has been eroded along most of the Himalaya, but in the western Zanskar–eastern Kashmir Himalaya the fold axis plunges to the NW, enabling the linking of the metamorphic isograds from the top (the Zanskar valley) to the bottom (the Kishtwar Window MCT zone) of the GHS slab (Searle and Rex 1989). Grujic et al. (1996, 2002) defined a similar geometry across the GHS in the Bhutan Himalaya.

The next recognition of channel flow came with extensive P–T thermobarometry showing that the metamorphic grade increases up-section above the MCT and increases down-section beneath the STD normal faults, with the highest grade sillimanite + K-feldspar gneisses showing muscovite dehydration melting and migmatization towards the central or upper part of the GHS. U–(Th)–Pb dating of peak metamorphism and melting also showed that the highest metamorphic conditions occurred in the Oligocene and Early Miocene, with early kyanite-grade conditions at c. 40–30 Ma and later sillimanite-grade conditions with dehydration melting occurring generally between c. 35 and c. 15 Ma (Godin et al. 2006; Cottle et al. 2007, 2009a, 2015). The large-scale structure of the GHS is therefore mid-crust slab of high-grade sillimanite gneisses, migmatites and leucogranites bounded by an inverted metamorphic gradient below (the MCT zone) and a right-way-up gradient above (the STD zone). Both the MCT and the STD were active simultaneously during the Miocene between c. 35 and c. 15 Ma.

Large-scale seismic imaging of southern Tibet (the INDEPTH project) showed a mid-crust layer of partially molten weak crust (Brown et al. 1996; Kind et al. 1996; Nelson et al. 1996; Hauck et al. 1998; Alsdorf and Nelson 1999) bounded by a rigid lower crust below and a brittle deforming upper crust (Tethyan zone) above. A series of ‘bright spots’ along the upper margin of the mid-crust layer were interpreted as pockets of leucogranite forming at the present day (Searle et al. 2003, 2006; Gaillard et al. 2004). These ‘bright spots’ are present at relatively shallow levels (c. 13–15 km depth) and are equivalent to the depths of formation of the leucogranites in the Miocene present along the crest of the Himalaya (Searle et al. 2006, 2010). Beaumont et al. (2001, 2004) and Jamieson et al. (2004) used both geophysical (the INDEPTH seismic surveys) and geological constraints to numerically simulate the channel flow model. They proposed that surface denudation of the GHS was required to exhume the channel and therefore made a direct link between deep ductile deformation and ductile flow with surface processes, including climate, rainfall and erosion.

Field structural constraints are now described around the Everest massif that are linked with petrology, thermobarometry, strain and the U–(Th)–Pb geochronology, which enables determination of the precise timing of fabric formation, fault motion and leucogranite intrusion along the massif.

The classic trekking approach to Everest on the Nepal side lies along the Dudh Kosi river transect from Phaplu, or Lukla, north to the Khumbu glacier (Fig. 4). The geology, thermobarometry and geochronology of the GHS along this transect has been studied by Hubbard (1989), Carosi et al. (1998), Simpson et al. (2000), Searle et al. (2003, 2006) and Jessup et al. (2006, 2008b). The GHS includes all the metamorphic rocks structurally above the MCT zone and below the STD or Qomolangma Detachment, a map distance of c. 50 km. An inverted metamorphic isograd sequence and inverted P–T gradient above the MCT is well exposed around Kharikhola, south of Lukla (Hubbard 1996). The MCT zone is a 5 km thick high-strain zone with numerous kinematic indicators showing top-to-the-south shearing. Index minerals including biotite, garnet, staurolite, kyanite and sillimanite show an inverted metamorphism with both flattening (pure shear) and non-coaxial southward shear strain (Hubbard 1996; Searle et al. 2003; Law et al. 2004, 2011; Jessup et al. 2006, 2008a, b). Plots of P–T against distance above the MCT show an increase in both temperature and pressure (8–6 kbar, kyanite grade) across the MCT zone at Kharikhola and then near-isothermal conditions (745–620°C) across the GHS slab up to the Lhotse Detachment (Jessup et al. 2008a, b). Migmatites are present from south of Lukla through Namche Bazar, northwards to the Khumbu glacier, representing one of the thickest sillimanite-grade migmatite sections in the Himalaya.

Evolution of the GHS rocks includes early prograde metamorphism during crustal thickening, resulting in kyanite-grade assemblages (c. 680–550°C, 10–8 kbar) and later decompression-related metamorphism associated with sillimanite + K-feldspar-grade metamorphism, migmatization (750–650°C, 7–4 kbar) and the formation of leucogranite melts (Pognante and Benna 1993; Searle et al. 2003, 2006, 2010; Visona et al. 2012). Cordierite growth resulted from the breakdown of sillimanite and garnet. Sillimanite + cordierite-bearing augen gneisses near Pumori Base Camp have a U–Pb monazite age of 22.7 ± 0.2 Ma (Simpson et al. 2000), interpreted as representing the timing of the later higher temperature metamorphism. This led to widespread crustal melting and the formation of the main Everest leucogranites at 21.3–20.5 Ma (Simpson et al. 2000) and the Nuptse leucogranite at 23.6 ± 0.7 Ma (Jessup et al. 2008a, b). Larson et al. (2021) dated the Nuptse granite at c. 19–18 Ma (U–Pb; zircon, monazite, xenotime), with older monazite grains of 25–23 Ma interpreted as inherited ages. The leucogranites around the Everest massif at the northern end of the Dudh Kosi–Khumbu glacier profile contain both muscovite and biotite, garnet and tourmaline as common magmatic minerals, with rare early andalusite and late cordierite associated with sillimanite–biotite clusters. Tourmaline is very common, both in the migmatite leucosomes and leucogranites. Star clusters composed of tourmaline + quartz are common in many localities.

The SW face of Everest (Fig. 6a) shows two north-dipping, low-angle normal faults: the upper Qomolangma Detachment, which places unmetamorphosed fossiliferous Ordovician lime mudstones above the metamorphic calc-silicates of the Mid-Cambrian Yellow Band; and the lower Lhotse Detachment, which places the Everest Series schists (North Col Formation) above the high-grade sillimanite gneisses and leucogranites of the GHS (Searle 1999a,b, 2003; Searle et al. 2003, 2006). The Lhotse Detachment is a prominent brittle fault in some areas (e.g. Gyachung Kang), whereas in others (e.g. the SW face of Everest) it is a diffuse zone corresponding to the upper limit of leucogranite intrusions. The Everest Series rocks still preserve sedimentary textures (Myrow et al. 2008), but the metamorphism reached garnet, staurolite and andalusite grade (Jessup et al. 2008a, b). The P–T conditions of staurolite schists are 650–610°C and 6.2 kbar, with pressures decreasing up-section to c. 3 kbar beneath the Qomolangma Detachment (Jessup et al. 2008b). Limestones from the summit of Mt Everest contain relict fossils, notably fragments of trilobites, crinoid ossicles and ostracods (Gansser 1964; Searle et al. 2003, 2006; Sakai et al. 2005).

The main Everest leucogranite has been mapped towards the west in the peaks of Lingtren, Pumori and Cho Oyu, to the south along the base of Nuptse and Ama Dablam, and to the east into the Makalu–Baruntse massif (Fig. 4). The leucogranites form dominant sills emplaced into the sillimanite gneisses, with occasional cross-cutting dykes feeding structurally higher sills (Searle 1999a). One large xenolith of sillimanite gneiss is completely enclosed by leucogranite on the south face of Lingtren (Fig. 6b). On Everest, the upper intrusive contact of the Everest–Nuptse leucogranite shows a ragged, uneven margin with a few thin dykes and sills emanating up into the Everest Series schists. Elsewhere, the contact is a sharp normal fault, termed the Lhotse Detachment (Searle 1999a, 2003), which usually marks the upper limit of leucogranite intrusions. Several prominent calc-silicate bands run from the South Col across the SW face of Everest within the Everest Series (Fig. 6a). In the Western Cwm, these calc-silicate bands define south-vergent chevron folds exposed along the upper part of the north faces of Lhotse and Nuptse. The south-vergent folds in the Western Cwm of Everest are truncated above by the low-angle normal fault of the Qomolangma Detachment.

The Khumbu glacier cuts across the Everest–Nuptse leucogranite where the granite has the form of a bulbous termination of a sill rather than an intrusive pluton (Fig. 7a). Foliation in the sillimanite gneisses wraps around the granite with a southerly vergence and a prominent zone of high strain defines the Lhotse ductile shear zone structurally beneath the Nuptse granite. Simpson et al. (2000) dated early monazite growth and high-temperature metamorphism (kyanite and sillimanite grade) at 32.2 ± 0.4 Ma from a sample collected from Kala Patar, SW of Everest Base Camp (Fig. 4). A granitic augen gneiss (K-feldspar + plagioclase + biotite + sillimanite + cordierite + muscovite) collected from the south ridge of Pumori gave a U–Pb monazite age of 22.7 ± 0.2 Ma, whereas a sample of the main Everest granite gave a monazite and xenotime age of 21.3–20.5 Ma (Simpson et al. 2000). The Nuptse granite contains biotite, muscovite and tourmaline, in addition to quartz, K-feldspar and minor plagioclase, and has a U–Pb age of 19–18 Ma, with older inherited monazites at 23.6 ± 0.7 Ma (Jessup et al. 2008b; Larson et al. 2021). A network of pegmatite and aplite dykes emanates from the upper contact of the Nuptse granite, beautifully exposed on the ‘Spider Wall’, the SW face of Nuptse (Searle et al. 2003).

The high volatile component of the pegmatites resulted in metasomatic growth of K-feldspar and magma injection preferentially followed foliation-parallel flow along the sills (Searle 1999a, b; Weinberg and Searle 1999a). Some leucogranite sills have been mapped for >50 km from the Khumbu region of Nepal northwards along the flanks of the Rongbuk glacier and valley in Tibet.

The P–T conditions and microstructures across the Everest Series and the upper levels of the GHS have been recorded in detail by Jessup et al. (2008a, b). The Everest Series schists from the south face of Lhotse and Nuptse have the assemblage: staurolite + garnet + muscovite + biotite + quartz + plagioclase ± cordierite ± andalusite, with early staurolite replaced by later cordierite. At the lower structural levels of the Everest Series, at an altitude of 5600 m, sillimanite appears as fibrolite intergrowing with biotite roughly at the location of the Lhotse Detachment.

Metamorphic isograds along the top of the GHS slab are telescoped and flattened, with rocks showing different P–T histories across the Lhotse Detachment ductile shear zone. Searle (1999a) and Searle et al. (2003, 2006) also described a south-vergent ductile thrust (the Khumbu thrust) underlying the Nuptse granite, with the metamorphic foliation wrapped around the bulbous leucogranite. In general, south-vergent, thrust-related fabrics have been overprinted by north-vergent, normal-sense (‘extensional’) fabrics superimposed during exhumation beneath passive roof faults.

The Imja Khola (valley) separates the Everest–Lhotse–Nuptse massif to the north from Ama Dablam peak to the south (Fig. 4). Mapping shows that the Nuptse granite continues south to Ama Dablam and further south to the peaks of Tramserku and Kangteiga. A large leucogranite sill on the north face of Ama Dablam, presumed to be the same age as the Nuptse granite, shows a set of dykes that intrude upwards into the host sillimanite gneiss (Fig. 7b). These dykes are all bent towards the north along the footwall of the Lhotse Detachment, which outcrops along the south face of Lhotse across the valley. These dykes reveal the southward flow of granite magma along the giant sill complex structurally beneath the STD. Larson et al. (2021) dated the main Nuptse granite body (U–Pb monazite) at 19–18 Ma and a structurally lower granite from Nuptse Base Camp at 17.1 ± 0.2 Ma. The fact that leucogranite sills are exposed on many of the higher peaks further south in the Khumbu region suggests that multiple sills make up all the granite sheets exposed. The granite sills are intruded into, and parallel with, the metamorphic fabric in the host sillimanite gneiss. A few later dykes cross-cut the ductile fabrics, as seen along the Rongbuk valley. In general, leucogranite intrusion in the Everest region occurred via multiple batches of magmas, sometimes intruded over a very short time interval.

The west face of Everest, as viewed from the summit ridge of Lingtren and the West Col of Everest, shows low-angle normal faults dipping at a gentle angle (between 5 and 15°) to the north (Fig. 8a). The Everest Series is c. 800 m thick between the two faults. The upper Qomolangma Detachment is a clear structural break, with a significant jump in P–T conditions from relatively unmetamorphosed sedimentary rocks on the summit above the Qomolangma Detachment to metamorphic calc-silicates of the Yellow Band below. These Cambrian limestones, c. 510 Ma, were metamorphosed to P–T conditions of c. 440°C and 3–4 kbar along the top of the GHS during the Miocene (Sakai et al. 2005; Jessup et al. 2008a, b; Streule et al. 2012). Below the Yellow Band, black slaty rocks are mainly staurolite ± andalusite pelites of presumed Cambrian–late Neoproterozoic age, metamorphosed in the Miocene during the Himalayan event. The Lhotse Detachment is more diffuse, with a gradual change from lower level sillimanite gneisses upwards to staurolite and andalusite schists without any granitic melts. In some sections, this structural horizon marks the upper limit of leucogranite intrusions.

The peak of Gyachung Kang (7922 m), west of Everest, very clearly shows both the Qomolangma and Lhotse detachments as clean-cut faults (Fig. 8b). The summit rocks are Ordovician sedimentary rocks, similar to the Everest summit rocks (Searle et al. 2003, 2006; Myrow et al. 2008). The Everest Series rocks structurally below are less thick than on the west face of Everest, but are still at least 600 m thick. The lower parts of Gyachung Kang are massive leucogranites, which are the western extension of the Everest–Nuptse granite sill complex and most likely the same age, c. 19–18 Ma, with the oldest monazites dated at 25–23 Ma possibly recording the earliest granite melts in this locality.

The structure of the Everest and Gyachung Kang massifs continues westward to the south face of Cho Oyu (Fig. 9) and further west to the peak of Shisha Pangma (Searle et al. 1997). The view from the summit shows the gentle north-dipping Qomolangma Detachment extending along the Rongbuk glacier for >70 km to the north. The black rocks of the Everest Series are structurally thinned out towards the north as the Qomolangma and Lhotse detachments merge into a single large ductile shear zone. On Cho Oyu, the Qomolangma Detachment crops out just below the summit with Ordovician black shales, similar to the Everest summit rocks, below. The Everest Series rocks on Cho Oyu are thicker than those on Gyachung Kang and appear to progressively extend downwards to the sillimanite gneisses of the GHS, demonstrating right-way-up metamorphic isograds beneath the STD. Leucogranite sills are abundant along the base of the south face of Cho Oyu, extending south to Gokyo, but become progressively thinner and less common with structural height. There does not appear to be an abrupt faulted contact in the Cho Oyu south face profile, as seen for the Lhotse Detachment at Gyachung Kang.

The east (Kangshung) face of Everest in Tibet is the most remote face of the Everest massif, accessible only by a long trek from Kharta village in the north. The Kangshung face of Everest is >3.5 km high and rises from the Kangshung glacier, which flows east and then south into the upper drainage of the Arun River (Fig. 10a). The Arun River turns south and flows through a deep, impenetrable gorge east of the Makalu–Chomolonzo massif. Samples collected from the first ascent of the Kangshung face (Venables 1989) show garnet + tourmaline leucogranites intruding dark sillimanite ± cordierite gneisses, similar to the Nepal side. Leucogranite sills decrease in abundance structurally upward and disappear completely above the South Col. On Pethangtse, east of Everest (Fig. 4), the deepest structural levels of the massif show spectacular outcrops of large boulders of metamorphic rocks (biotite, sillimanite and cordierite gneiss) intruded by older leucogranite sills completely enclosed within the younger Everest leucogranite (Fig. 10b). U–(Th)–Pb dating of metamorphic zircon, monazite and xenotime shows two periods of ‘peak’ metamorphism: one at 38.9 ± 0.9 Ma and a later one at 20.8 ± 0.8 Ma (Cottle et al. 2009a, b). The main Everest leucogranite has been dated at c. 16.7 Ma in the Kangshung valley and two subsequent phases of leucogranite emplacement were dated at 15.2 ± 0.2 and 12.6 ± 0.2 Ma (Cottle et al. 2009a, b). These younger leucogranites cross-cut the main fabrics associated with the southward channel flow.

The massive 3200 m high granite wall of the north face of Chomolonzo (7804 m) shows a giant leucogranite sill, c. 2 km thick, intruded into dark sillimanite + cordierite gneisses of the upper GHS (Fig. 10c). The gneisses are exposed along both flanks of the Kangshung glacier and the Everest East Base Camp site below the Kangshung face. The Chomolonzo leucogranites are extremely inaccessible, but samples collected from the Kangshung valley contain cordierite (pinite) and rare andalusite, as well as garnet, tourmaline and muscovite as magmatic phases. These cordierite-bearing leucogranites are similar to those exposed on the south face of Makalu on the Nepal side (Streule et al. 2010; Searle 2015). The Makalu and Chomolonzo leucogranites are the same massif. P–T estimates from sillimanite-grade metamorphism on the south flank of Makalu are 713°C and 5.9 kbar, with a secondary lower pressure cordierite overprint at 618°C and 2.1 kbar (Streule et al. 2010, 2012). U–(Th)–Pb dating shows that the main phase of leucogranite formed by muscovite dehydration melting on Makalu was between c. 24 and 21 Ma, whereas the most recent cordierite leucogranites, associated with the Barun migmatites, were formed at 700°C and c. 4 kbar between 16.0 ± 0.6 and 15.6 ± 0.2 Ma (Streule et al. 2010). At least six phases of cross-cutting leucogranites can be demonstrated from the Barun glacier, west and south of Makalu and Chomolonzo (Streule et al. 2010).

The summit rocks of Mt Everest are Ordovician lime mudstones (Fig. 11a). They dip gently (15–10°) to the north and show sedimentary features and fossils, including crinoid ossicles, fragments of trilobites and ostracods (Gansser 1964; Searle et al. 2003; Sakai et al. 2005). Samples from the summit area show sedimentary textures with a weak metamorphic overprint or recrystallization, whereas the Yellow Band is clearly a metamorphic marble (Fig. 12). Myrow et al. (2008) correlated the Yellow Band on Everest with the Chiatsun Group in Tibet and proposed an age of Early to Mid-Ordovician based on nautiloid, brachiopod and conodont assemblages (Yin and Kuo 1978; Burchfiel et al. 1992; see also Stouge et al. 2021; Zhan et al. 2014).

The Everest summit rocks are cut by penetrative deformation fabrics of variable intensity, with deformation temperatures (c. 250–500°C) inferred from the associated microstructures generally increasing structurally downwards (cf. Sakai et al. 2005; Jessup et al. 2006; Corthouts et al. 2015; Larson et al. 2020, 2022). Apatite fission track ages of the Everest Series and summit Ordovician limestones are up to 30.5 ± 5.1 Ma, older than the GHS (16.3 ± 0.8–3.8 ± 0.4 Ma) (Streule et al. 2012). This age is consistent with a syn-kinematic rutile age, suggesting that the Qomolangma Detachment was active at 16.3 ± 5 Ma (Larson et al. 2020). The base of the summit pyramid (the Third Step on the North Ridge route) is composed of a 60 m thick prominent bed of microbial thrombolite, made up of tightly compressed algal mats (Myrow et al. 2008).

The Qomolangma Detachment separates relatively unmetamorphosed limestones of the summit from metamorphic rocks of the Yellow Band and Everest Series beneath. The Yellow Band marbles, structurally beneath the Qomolangma Detachment, outcrop between c. 8348 and 8520 m altitude on the north face of Everest. A calc-phyllite sample collected by Lawrence Wager from the base of the Yellow Band has the assemblage quartz + plagioclase + muscovite + calcite + biotite and a similar sample from Changtse peak has P–T conditions estimated at 440 ± 40°C and 3–4 kbar (Wager collection housed in the University of Oxford Natural History Museum, details in Waters et al. 2019). At the Dzachha Chu locality, NW of Everest, the brittle detachment merges with a ductile shear zone up to 800 m thick above the upper limit of the leucogranite sheets (Cottle et al. 2007, 2011; Waters et al. 2019).

The upper limit of leucogranite intrusions is well exposed along the lower cliffs of Changtse, above Everest Advanced Base Camp (old Camp III) on the upper East Rongbuk glacier (Fig. 11b). The North Col of Everest is permanently snow-covered, but is likely to be underlain by the lower part of the Everest Series. Both the Yellow Band carbonate and the upper part of the Everest Series have protolith ages of Early to Mid-Cambrian, with the youngest detrital zircons at 526 Ma (Myrow et al. 2008). ⁴⁰Ar/³⁹Ar muscovite ages from the Yellow Band have two age peaks of 33.3 and 24.5 Ma (Sakai et al. 2005), ages that are surprisingly older than the granite intrusion ages. These ages may have been affected by excess Ar and may therefore be geologically unreliable. Zircon and apatite fission track ages from the Yellow Band gave ages of 14.4 ± 1.4 Ma (Sakai et al. 2005). Larson et al. (2022) presented three U–Pb calcite ages, with large uncertainties from veins intruding rocks below the summit, of 55 ± 28, 11 ± 26 and 45 ± 5.4 Ma. It is not clear what these ages represent; they may possibly relate to the timing of fluid flow, but they do not provide any reliable age date for the timing of deformation.

The Lhakpa-la is the high pass (6800 m) that connects the upper part of the East Rongbuk glacier (Fig. 13a) to the Far East Rongbuk glacier next to the peak of Kharta Phu. Samples collected by Lawrence Wager from the Lhakpa ridge in 1933 in the footwall of the Qomolangma Detachment have the assemblage: garnet + staurolite + sillimanite + biotite + plagioclase + quartz + ilmenite and have ‘peak’ P–T conditions of c. 5.5–6.5 kbar at 630°C (Waters et al. 2019). Late growths of andalusite and pseudomorphs of pinite after cordierite overgrow and post-date the deformation fabrics. The shear fabrics in these rocks can be seen to curve into alignment with a ductile shear zone along a prominent calc-silicate band that may correspond to the Yellow Band. The Qomolangma Detachment is interpreted as following the upper contact of this band (Fig. 13a). Above the Qomolangma Detachment, relatively unmetamorphosed sedimentary rocks show a gentle northerly dip (10–20°), with the bedding abruptly cut by the Qomolangma Detachment. The Lhotse Detachment is not obviously present in the outcrops below the Lhakpa-la, where the strain seems to have dissipated into a ductile shear zone. The lower outcrops above the east bank of the East Rongbuk glacier are composed of leucogranites showing a similar ragged upper margin to those in the Western Cwm on the Nepal side of Everest.

The Qomolangma Detachment dips at 10–15° north on the summit pyramid of Everest and flattens towards the north of Changtse and the Lhakpa-la (see structure contour map made from Google Earth imagery with the digital elevation model, in Waters et al. 2019, based on the geological map of Searle 2003). The Qomolangma Detachment and footwall gneisses and leucogranites are exposed for 40 km north of Everest as the normal fault dips at <10° and is often horizontal, until the Qomolangma Detachment dips more steeply to the north beneath the sedimentary rocks of the Tibetan Plateau. Ductile fabrics in the sillimanite gneisses and calc-silicate marble are parallel to the Qomolangma Detachment. U–Pb monazite ages from the footwall of the Qomolangma Detachment at the Hermit's Gorge location show the peak timing of metamorphism at c. 24 Ma and melting to form leucogranites at c. 20.4 Ma (Cottle et al. 2015). Mottram et al. (2019) presented U–Pb titanite ages from five samples of metamorphic rocks (calc-silicates and biotite amphibolites) from the Hermit's Gorge and the valley to the north, which range from 19.1 ± 1.6 to 15.2 ± 0.9 Ma. ⁴⁰Ar/³⁹Ar muscovite ages of c. 15.5–14.2 Ma and U–Th–He apatite dates of c. 14.5–11 Ma (Schultz et al. 2017) attest to extremely rapid exhumation and cooling during channel flow.

Leucogranite sills within the Greater Himalayan Sequence metasedimentary rocks exposed along the sides of the Rongbuk valley are parallel to the foliation and the overlying detachment for most, or all, of their exposed lengths. In Hermit's Gorge, three sets of leucogranites have been mapped and dated (Fig. 13b; Searle et al. 2006; Cottle et al. 2015). The early set 1 sills are folded and deformed (Fig. 14a, b). The set 2 sills are parallel to the ductile fabrics in the Hermit's Gorge, but, to the north, the set 2 leucogranites are dykes that cross-cut the ductile fabrics. The youngest set 3 dykes cross-cut both sets of earlier leucogranites and ductile fabrics. At higher structural levels, the brittle Qomolangma Detachment fault cuts all the fabrics and leucogranites in the footwall. Dating by U–(Th)–Pb isotope dilution thermal ionization mass spectrometry and laser ablation multi-collector inductively coupled plasma mass spectrometry show that the early set 1 folded leucogranites are 16.4 Ma and the later set 2 and 3 sills and dykes are within error at 15.6–15.4 Ma (Cottle et al. 2015). Hodges et al. (1992, 1998) dated the ‘Rongbuk granite’ (sample R113) at 16.67 ± 0.04 Ma and described the granite as cutting the STD. Murphy and Harrison (1999) also dated a mylonitized granite from the lower Rongbuk valley at 16.2 ± 0.8 Ma by ²³²Th/²⁰⁸Pb dating, but could not find any granite cross-cutting the STD.

During our mapping along the Rongbuk glacier and valley, we found that all the leucogranites are truncated by the overlying brittle STD at higher elevations and that none intruded up into the unmetamorphosed hanging wall rocks. Strain decreases up-section from the sillimanite gneisses and leucogranite intrusions to the Qomolangma Detachment, but, as seen at location ABC and the Lhakpa-la on the north side of Everest, all the leucogranites are confined to the GHS structurally beneath the STD.

The Rongbuk leucogranites contain varying proportions of garnet, muscovite and tourmaline, with minor biotite, similar to the leucogranites on the south side of Everest and at Ama Dablam, Lingtren and Gyachung Kang (Searle 2003). Magmatic andalusite has been reported from two-mica leucogranites and tourmaline leucogranites from the lower Rongbuk valley (Visona et al. 2012), as well as other localities along the GHS and in the north Himalayan domes. Andalusite and cordierite are indicative of low-pressure melting (<4 kbar), in common with the more widespread garnet + tourmaline + muscovite leucogranites (Searle et al. 2010; Streule et al. 2010).

Microstructures and quartz c-axis fabrics from three vertical transects across the Greater Himalayan Sequence gneisses and schists in the footwall of the Qomolangma Detachment unequivocally demonstrate a top-down-to-the-north (equivalent to bottom-up-to-the-south) sense of shear (Law et al. 2004, 2011), compatible with the channel flow model. The quartz fabric opening angles indicate steep field gradients in increasing deformation temperatures with depth beneath the detachment, with inferred temperature gradients (from north to south) of c. 370, 385 and 420°C km⁻¹ in the Northern, Rongbuk Monastery and Hermit's Gorge transects, respectively (Law et al. 2004, 2011; locations shown in Fig. 4). A thermal gradient of c. 310°C km⁻¹ has been estimated by Cottle et al. (2007, 2011) in the footwall to the STD in the Dzakuu Chu transect, located 50 km NE of Everest, using metamorphic mineral assemblages and Raman spectrometry on carbonaceous material.

More recently, data from the Dzakaa Chu, Northern and Hermit's Gorge transects have been re-examined by Waters et al. (2018) and, for the Northern and Hermit's Gorge transects, integrated with microstructural and petrological data from samples collected by Lawrence Wager in 1933 and placed into a structural framework defined by the structure contours on the overlying detachment. This re-examination, although indicating more modest thermal gradients for the transects of c. 200°C km⁻¹, still attests to the exceptionally high strains (with a significant component of pure shear; Law et al. 2004, 2011; Jessup et al. 2006) that have accumulated in the immediate footwall to the STD during the southward extrusion of the Greater Himalayan Sequence beneath the STD. Hydrogen isotope studies combined with Ar–Ar geochronology on the Northern transect indicate that meteoric water infiltrated down to mid-crustal levels during high-temperature shearing on this strand of the STD at 15 Ma (Gébelin et al. 2013, 2017).

The Ama Drime massif is the northeastern extension of the GHS in the Everest region (Fig. 2) and represents the deepest exposed sections of the Indian plate crust in the central Himalaya. The massif is a NNE-plunging antiform, bounded by two large-scale NNE–SSW-striking, outward-dipping ductile shear zones overprinted by steeply dipping normal faults: the Ama Drime detachment in the west and the Nyonno-Ri detachment in the east (Jessup et al. 2008a; Cottle et al. 2009a). Footwall rocks in the core of the Ama Drime massif include granulite facies migmatitic orthogneisses with pods of eclogite, calc-silicates, marble and quartzite intruded by leucogranites. The hanging wall rocks are migmatitic orthogneisses of the GHS. The steep normal faults bounding the margins of the Ama Drime massif offset the earlier ductile fabrics of the STD zone and are therefore related to uplift of the Ama Drime footwall rocks after motion along the STD ceased. The uplift was likely the result of deep-level thrusting beneath the Ama Drime massif.

Similar to other compressional core complexes, these normal-sense ductile shear zones relate to uplift and exhumation of the footwall and not to lowering of the surface elevation, crustal thinning or orogenic collapse (Searle and Lamont 2019). The ductile shear zones are most likely spatially associated with the STD, but were folded by culmination and uplift above a deeper level thrust. There is a kinematic continuum between the Ama Drime massif bounding shear zones and the brittle faults. The steep brittle normal faults are related to orogen-parallel extension and the development of the north–south aligned graben systems across southern Tibet, including the Dinggye rift (Burchfiel et al. 1992; Kali et al. 2010) east of Ama Drime (Fig. 2). These north–south aligned graben do not extend south of the STD into the GHS, affect only the upper crust and have relatively minor amounts of east–west extension. It is unclear how east–west extension in the upper crust of the Tethyan Himalaya is compatible with north–south compression in the GHS.

The Ama Drime massif is composed of similar lithologies and is of a similar metamorphic grade to the GHS. Sillimanite-grade gneisses and migmatites similar to the GHS from the main Everest massif overlie a structurally deeper orthogneiss that contains boudins of mafic eclogites that have been transformed to granulites (Lombardo and Rolfo 2000; Groppo et al. 2007; Cottle et al. 2009b; Corrie et al. 2010). The granulitized eclogites have the mineral assemblage garnet + clinopyroxene + plagioglase + hornblende + biotite + quartz + orthopyroxene + ilmenite, with symplectites after omphacite, which reveals P–T conditions of c. 750°C and 0.8–0.7 GPa (Groppo et al. 2007). The mafic igneous protolith has U–(Th)–Pb ages of 986.6 ± 1.8 Ma, whereas the host felsic orthogneiss is 1799 ± 9 Ma (Cottle et al. 2009a), showing that Paleoproterozoic to Neoproterozoic rocks occur in the Lesser Himalaya south of the MCT, within the MCT zone (the Ulleri augen gneiss) and in the GHS north of the MCT. Lu–Hf garnet ages of 37.5 ± 0.8–36.0 ± 1.9 Ma constrain the age of eclogite facies metamorphism (Kellett et al. 2019). Monazite and xenotime ages show that the granulite metamorphic overprint and anatexis occurred at <13.2 ± 1.4 Ma (Cottle et al. 2009a). Leucogranite dykes dated at 11.6 ± 0.4 Ma cross-cut the north–south-striking fabrics and are compositionally similar to the leucogranite dykes along the Rongbuk valley.

Along the Himalaya, crustal thickening and shortening occurred within the Indian plate following the final closure of the Tethys Ocean at c. 50 Ma (Green et al. 2008) and the India–Asia collision. Eocene–Oligocene kyanite-grade rocks record burial to depths of 30–45 km. Staurolite and kyanite assemblages were replaced by sillimanite + muscovite and sillimanite + K-feldspar assemblages, with partial melting forming widespread migmatites. U–(Th)–Pb dating of zircon, monazite and xenotime in the Everest region show that sillimanite-grade metamorphism lasted at least 20 Myr and propagated northwards with time (Cottle et al. 2009a). In the deepest structural levels in the Kangshung valley, ‘peak’ metamorphism has been dated at 38.9 ± 0.9 and 28.0 ± 0.8 Ma, with two phases of leucogranite intrusion at 15.2 ± 0.2 and 12.6 ± 0.2 Ma (Cottle et al. 2009a). The later leucogranites cross-cut the north–south fabrics in Ama Drime massif, suggesting that deformation occurred before 15–12 Ma.

Later brittle faults bound the Ama Drime massif on both flanks and are related to the young uplift of the massif. At the structurally deepest crustal levels of the Ama Drime massif to the north, granulite–upper amphibolite facies orthogneisses contain boudins of retrogressed eclogites. The eclogite event (>580°C, 1.5 GPa) occurred at 37.5 ± 0.8 Ma and was overprinted by granulite facies metamorphism (>750°C, 1.0–0.8 GPa) at 13.2 ± 1.4 Ma (Cottle et al. 2009a). The Late Eocene eclogite facies event relates to major Himalayan crustal thickening (c. 70 km) and not to Early Eocene plate margin subduction as recorded by the Kaghan (Pakistan) and Tso Morari (Ladakh) eclogites.

The transition from staurolite- and kyanite-grade metamorphism to sillimanite-grade metamorphism triggered decompression-related muscovite and biotite dehydration and the production of widespread partial melts. Along the Dudh Kosi transect south of Everest, the migmatites make up a mid-crustal layer c. 15–20 km thick above the MCT zone. Melting peaked during the early to mid-Miocene, with large leucogranite sills emanating from the deeper migmatite zone and intruding along foliation-parallel sills. In the Everest area and along the Rongbuk glacier, several thick sills have been mapped for >70 km across-strike (Searle 2003). The larger sills, such as the Everest–Nuptse granite, flowed southward above ductile thrusts (e.g. the Khumbu thrust) and ballooned outward in southward-verging folds (e.g. the Nuptse folded leucogranite sill).

Field observations and mapping around the Everest massif and surrounding mountains provide solid evidence for horizontal channel flow along the GHS. Restoration of the upper Qomolangma Detachment (Fig. 15) shows at least 100 km of southward flow of the footwall gneisses and leucogranites of the GHS, similar to the offset along the Annapurna Detachment in western-central Nepal (Searle 2010). The pressure estimates from rocks around the leucogranites at Everest Base Camp on the Nepal side are compatible with depths of 15–20 km. These rocks must have moved southward beneath the fixed roof fault of the Qomolangma Detachment in order to juxtapose next to the unmetamorphosed Ordovician sedimentary rocks on the summit of Everest.

It seems likely that partial melting of the GHS during the Early Miocene initiated channel flow and that flow ended when the system ran out of melt at c. 13–11 Ma. The ages of the leucogranites define the time frame of channel flow. Himalayan channel flow is therefore time-restricted (between c. 21 and 11 Ma) and also spatially restricted (the upper part of the mid-crust). The source of heat for melting is almost certainly internal radiogenic heating as a result of the shallow level of partial melting, the high U–Th content of the source rocks (Haimanta Group pelites) and widespread in situ melting in the migmatite zone. The heat source is unlikely to be shear heating along the MCT because the melting zone is structurally much higher (>5 km) within the GHS slab. The shallow level and widespread extent of sillimanite-grade metamorphism along the top of the GHS is also regional low-pressure metamorphism and not contact metamorphism.

The channel flow process of mid-crustal flow is now very well constrained in the Himalaya, but may be unique to this mountain range. From 11 Ma, brittle thrusting within the Lesser Himalaya thrust sheet and along the Main Boundary Thrust is compatible with the critical taper model. We suggest that all the geological data support Miocene ductile channel flow within the GHS mid-crust and that critical taper models are compatible with younger brittle thrusting along the Lesser Himalaya above the Main Boundary Thrust.

• (1) Mt Everest shows two large-scale, low-angle ductile shear zones that merge to the north into one shear zone with a brittle low-angle fault (the Qomolangma Detachment) above. The Qomolangma Detachment divides hanging wall sediments of Ordovician and younger age above from the staurolite- to sillimanite-grade Everest Series gneisses below. The lower Lhotse Detachment separates Everest Series gneisses above from leucogranites and sillimanite-grade migmatites below. The Lhotse Detachment is a clear brittle fault in some places (Gyachung Kang), but elsewhere is a diffuse shear zone (SW face of Everest and Cho Oyu).

• (2) Protolith rocks of the GHS in the Everest–Ama Drime massif are Paleoproterozoic to Cambrian, dominantly metasedimentary rocks, with prominent igneous protoliths, including Proterozoic and Cambrian orthogneisses. Similar Proterozoic rocks are therefore present both below and above the MCT, including in the Ama Drime Range.

• (3) Metamorphism of the Everest–Ama Drime massif shows peak sillimanite-grade metamorphism on the Nepal side of Everest at c. 32.2 Ma and slightly younger on the Tibet side, spanning c. 24–15.2 Ma, from U–Th–Pb monazite and titanite dates. Sillimanite–muscovite assemblages were replaced by sillimanite + K-feldspar migmatites and the formation of in situ leucogranite melts.

• (4) Tourmaline + garnet + muscovite leucogranites have been intruded as elongated sill complexes at least c. 70–120 km long (north–south), as mapped along the Rongbuk valley and continuing south across the migmatite zone of the upper Khumbu (Nepal) side. U–Th–Pb monazite ages from the Everest granites around the Nepal side Base Camp are c. 21.3–20.5 Ma, slightly older than the leucogranites at higher structural levels in the Kangshung and Rongbuk valleys in Tibet. The Nuptse–Ama Dablam leucogranites were intruded at c. 19–17 Ma. Along the Rongbuk valley, early leucogranites are folded in with host sillimanite gneisses and calc-silicates and are dated at c. 16.4 Ma. Later leucogranites intruded as dykes, cross-cutting the regional ductile fabrics, and are dated at 15.6–15.4 Ma. All the leucogranites are cut by the highest brittle, low-angle normal fault, the Qomolangma Detachment.

• (5) The leucogranites were derived from in situ partial melting of sillimanite + muscovite pelites and migmatites. Although there are isotopic differences between the source pelites and leucogranites, there is widespread field evidence for in situ migmatization. Batch melting resulted in multiple intrusions separated by short time periods. Sheeted sill complexes transported melts for >100 km by horizontal melt transport within sillimanite-grade host rocks.

• (6) The Ama Drime massif is the northern extension of the GHS in the Everest area and is composed of similar metamorphic rocks and leucogranites to the GHS. The protoliths are similar to, and older than, the main GHS as exposed, extending down to Paleoproterozoic rocks. Late Eocene eclogites are preserved as pods enclosed in later granulites. These eclogites are related to Late Eocene crustal thickening and not to Early Eocene, pre- or syn-collision slab break-off, as recorded by the Kaghan (north Pakistan) and Tso Morari (Ladakh) eclogites. Metamorphic ages and leucogranite dates are slightly younger than the GHS rocks on Everest, showing a northward younging of metamorphism from deeper to shallower levels.

• (7) The Mt Everest massif and the Ama Drime massif are located at the top of the GHS mid-crustal slab. The GHS slab extends southwards and structurally down to the MCT at Karikhola, south of Lukla, where a narrow zone of inverted metamorphism records syn- to post-metamorphism ductile shearing and inversion of the isograds. The timing of metamorphism youngs towards the south in Nepal and towards the north in the Ama Drime massif, consistent with the folded isograd model (Searle and Rex 1989). Leucogranites are common along the top of the GHS, but are not present along the MCT zone to the south.

• (8) Regional mapping of structures, strain analysis, thermobarometry, timing of metamorphism and leucogranite formation in the Everest–Makalu region all support the channel flow model – that is, the southward extrusion of a mid-crustal layer of high-grade and partially melted rocks during the Oligocene and Early Miocene. Himalayan mid-crustal channel flow is distinct and completely different from the lower crust flow model proposed to explain the geology of Eastern Tibet. It remains unclear how north–south compression along the Himalayan mid-crust GHS is partitioned with upper crustal east–west extension in south Tibet. The north–south normal faults in southern Tibet do not cross the STD into the GHS, so extension is restricted to the upper c. 10–12 km of crust. GPS data show that north–south convergence is still occurring along the Himalaya and earthquakes, such as the 2015 Gorkha earthquake (Elliott et al. 2016), reveal that the Himalayan range remains under compressional forces today.

Thanks to Jamie McGuinness for his photographs from the summit of Everest, Stephen Venables for his photographs from the Kangshung face, Marko Prezelj for his photographs of Gyachung Kang and Leo Dickinson for his photographs taken from a balloon flying over Everest. Thanks also to David Hamilton, Jon Tinker and Kenton Cool for collecting samples from the south face and summit of Everest. We thank Rob Simpson, Mike Streule and Roberto Weinberg for discussions in the field, Dave Waters for the thermobarometry, and Randall Parrish, Steve Noble and Matt Horstwood for their U–(Th)–Pb geochronology expertise.

MS: investigation (equal), writing – original draft (lead), writing – review and editing (lead); JC: investigation (equal), writing – original draft (supporting), writing – review and editing (supporting); MJ: investigation (equal), writing – original draft (supporting), writing – review and editing (supporting); RDL: investigation (equal), writing – original draft (supporting), writing – review and editing (supporting).

This work was supported by a NERC (UK) Senior post-doctoral fellowship grant to MPS and National Science Foundation (USA) grants EAR-0207524 to RDL and MPS, EAR-0911561 to MJJ and EAR-1119380 to JMC.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

All data are available from authors.

In Figure 12 the altitude was corrected from 8969 m to 8689 m.