Annals of Glaciology 5 1984 © International Glaciological Society

GEOTHERMAL EFFECTS OF 18 ka BP ICE CONDITIONS IN

THE SWISS PLATEAU

by

Wilfried Haeberli

(Versuchsanstalt fur Wasserbau, Hydrologie und Glaziologie, ETH Zurich, ETH Zentrum, CH-8092 Zurich, Switzerland)

Werner Rellstab (lnstitut fur Geophysik, ETH Zurich, ETH Honggerberg, CH-8093 Zurich, Switzerland)

and William D. Harrison

(Geophysical Institute, University of Alaska, Fairbanks, Alaska 99701, U.S.A.)

ABSTRACT oil and gas, or studies concerning possible places Quaternary surface temperature and ice conditions for radioactive waste disposal may possibly help to have lowered today's temperatures in the uppermost verify palaeoclimatic reconstructions (Haeberli 1 to 2 km of the Earth's crust in the Swiss plateau 1982). The purpose of this paper is to introduce by about 5 to 6°C in regions of formerly temperate briefly three important aspects of geothermal ice-age glacier beds, and these effects were even greater on effects in the Swiss plateau: ice conditions (perma­ temperatures in formerly periglacial regions. Effects frost and gla~iers at 18 ka BP) discussed by Haeberli, of latent heat exchange during the formation and heat diffusion effects in bedrock discussed by thawing of ice-rich permafrost in high-porosity sedi­ Rellstab, and combined heat diffusion and latent heat ments enhance the effects of heat diffusion in low­ exchange effects in unconsolidated sediments which porosity rock. The influence of underground ice forma­ have undergone phase changes during the formation and tion, however, seems to be limited. This is due both thawing of permafrost, discussed by Harrison. to the limited thickness of high-porosity sediments and to the fact that high post-glacial surface tempera­ tures in the Swiss plateau shortened the thaw time of ice-rich permafrost which formed at the time of maximum ICE CONDITIONS AT 18 ka BP glaciation (18 ka BP). The greatest effects of ice con­ Around 18 ka BP, Switzerland was situated entirely ditions in 18 ka BP may therefore be expected outside within the zone of continuous permafrost distribution the plateau in Alpine valleys. Here, advection of cold (Washburn 1979). Periglacial ice wedges not only ice through glacier flow may have cooled high-porosity formed in loess deposits, but even in coarse gravels sediments of considerable thickness well below O°C. (e.g. Van der Meer 1980); perennially frozen sediments were deformed by glaciers in many places ("push mor­ aines" in the strict sense, i.e. deformed frozen sedi­ I NT RODUCTI ON ments, e.g. Schindler and others 1978, Van der Meer The temperature distribution in the Earth's crust unpublished, cf. Haeberli 1981) and the karst drainage is a result of (a) heat generation, thermal character­ in the limestone region of the Jura mountains was istics of rocks, and tectonic movements in the Earth's blocked by perennial frost (e.g. Barsch 1968, Haeberli interior, (b) the complex history of topography, sedi­ and others 1976). These examples provide morphological mentation, and erosion at the Earth's surface, and (c) evidence of permafrost with a high ice content and temperature variations at the air/ground interface of indicate that, during this time, extensive underground various frequencies and amplitudes. Among the tempera­ ice accumulated in unconsolidated sediments of peri­ ture variations during palaeoclimatic evolution, the glacial valleys and perhaps also in the limestone most prominent thermal effects in the uppermost kilo­ sediments of the Jura mountains (Fig.1). The depres­ metres of the crust were caused by two events: the sion of the permafrost boundary by more than 3 000 m pronounced depression of surface temperatures during with respect to today's conditions (Haeberli in the last ice age and the sharp rise in temperature press) corresponds to a lowering of the mean annual between the time of maximum glaciation (around 18 ka air temperature by at least 15°C compared to the BP) and the beginning of the thermally more stable present. Surface temperatures in the peri2lacial post-glacial period (about 10 ka BP). It is in the areas of the Swiss plateau were around -3 C and lower. uppermost kilometres that the studies on potential The equilibrium line on glaciers was depressed geothermal energy in the Swiss plateau are concen­ with respect to present conditions by about 1 200 to trated (Rybach and Jaffe 1976, Rybach and others 1 300 m. At the corresponding altitudes of about 1978). In addition, geothermal information from deep 1 000 to 1 500 m a.s.l. the mean annual air tempera­ drillings in connection with prospecting for salt, ture can be estimated at -10'e or colder. Surface

56

Downloaded from https://www.cambridge.org/core. 30 Sep 2021 at 11:07:59, subject to the Cambridge Core terms of use. Haebe rLi and others : Geothe nma L effects of 18 ka BP ice

o 50 100 km where t1 is the youngest of n intervals of time ti during which the deviation of the mean annual surface temperature Ti from today's is assumed to be constant, and K is thermal diffusivity. In the corresponding computer program, the number and length of the time intervals, surface temperatures during these inter­ vals, the length of the total time period under consideration, the heat diffusion coefficient, and the depth below the surface have to be selected. Present-day surface temperatures are estimated from ablatIon area of Ice age glaciers air temperature data at or close to the considered unconsolldated quater nary sedlments site. The past glacial conditions have to be selected OlD (hIgh Ice content when frOlen ' for each bore hole because subglacial temperatures consolidated older sedlments. rock may be different from periglacial temperatures. In D (Iow Ice content when frozen I connection with test runs using different models, limestone region ! karSI extensive literature searches were carried out to D blocked In permafrost) determine the best step function representing the palaeoclimatic evolution as reconstructed by geo­ Fig.1. Sketch map of ice conditions in th e Swiss logists and palaeontologists. For the last ice age, plateau around 18 ka BP. for example, the palaeotemperature curve is mainly based on a compilation by Rudloff (1980). Figure 2 ice temperatures were well below zero and a temper­ shows the section of the palaeoclimatic curve used in ate firn zone probably did not exist. From the well­ this paper. Figure 3 illustrates the relative import­ documented geometry of the piedmont glaciers in the ance of different time periods for present-day bore­ Swiss plateau (e.g. Jackli 1970, Keller and Krayss hole temperatures, calculated with a very detailed 1980), it can be inferred that the basal shear step function of palaeotemperatures (n = 36). The stresses were extremely low, as were mass balance calculation shows that the time period between 10 and gradients and flow velocities (cf. Blatter and 100 ka BP, roughly corresponding to the "last ice age", Haeberli 1984). The low glacier activity (cg. Meier has by far the most important effect for depths around 1961, Haeberli in press) makes it difficult to assume 500 to 2 000 m. It is followed in importance by the that these glaciers ever came close to steady-state conditions during the time available (cf. Fig.2), and the build-up mechanism of the large ice lobes within -6T a limited time remains an unsolved problem. It can 'C nevertheless be assumed that the glacier beds were 14 P temperate in a zone of limited extent near the ice 12 P fronts only, and that large glaciers may have trans­ 10 ported very cold ice from high altitudes to those 9 9 parts of their beds situated at the Alpine border of 8 the plateau, thus greatly enhancing depressions of 6 ground temperatures at 18 ka BP (Blatter and 4 Haeberl i 1984). In good agreement with palynological recon­ 2 structions, ice conditions during 18 ka BP in the Swiss plateau and relative displacements of perma­ frost and glacier boundaries reflect a very cold, dry climate during this time of maximum glaciation. Permafrost and glaciers of this time should not be 10 5 compared to conditions in maritime regions today (such as, for example, southern Alaska), since they Fig.2. Palaeoclimatic evolution during the last most closely resembled features of arid mountains 100 ka as assumed for the model calculations (e.g. in Asia, cf. Shih and others 1980). The reduc­ in the Swiss plateau. 6T-values are deviations tion in precipitation by a factor of three or even of the mean annual surface temperature from the more compared to values today (Haeberli 1982, in present, time is in a BP. p represents periglacial press) may be attributed to large-scale subsidence conditions and g subglacial conditions where (Manabe and Broccoli 1984). glacier beds were temperate. Present time is at right end of graph. GEOTHERMAL EFFECTS DUE TO HEAT DIFFUSION IN BEDROCK Palaeoclimatic corrections are applied to pro­ files drawn from bore-hole temperature observations -.o.r' .. (Bodmer unpubl ished), in connection with heat flow I"CI studies and research into the possibilities of ex­ ploiting geothermal energy in the Swiss plateau. In principle, thermal effects (6T* in Figs.3 and 4) due to palaeoclimatic changes of surface temperatures (6T in Fig.2) have to be computed as deviations of actual thermal conditions from steady-state condi­ tions. Oetail s of calculation and input data were described by Rellstab (1981 (revised 1982 )). Foll ow­ ing theoretical considerations hy Birch (1 948) and Carslaw and Jaeger (1959), a superpos i tion of error OL ~5 __ _ 20 25 km functions is used to calculate temperature T as a function of depth z and time t: 0 -10"y

T(z,t) = T1[1 - erf(z!/4Ktl)] + T2[erf(z! /4Kt 1) - Fig.3 . Thermal effects 6T* of the palaeoclimatic evolution during different time periods on - erf(z!~)] .•. + Tn[erf(z!~) - (1) present-day temperature profiles. z is depth below surface; time is indicated in a BP for the time - erf(z;l4K t n)] periods considered.

57

Downloaded from https://www.cambridge.org/core. 30 Sep 2021 at 11:07:59, subject to the Cambridge Core terms of use. HaebeT'ti and otheT's: Geothe1'f11O.t effects of 18 ka BP ice

time period between 100 and 1 000 ka BP, an order of 20 ka BP, as indicated by the palaeotemperature curve magnitude which suggests an expression 1 ike "Quaternary (Fig.2) used for calculating ice-age effects on temp­ before the last ice age" for this period. These two erature distribution in the crust. most important time periods probably experienced com­ The warming-up of the surface to positive tempera­ parable temperature conditions. Older time periods are tures causes ice-age permafrost to become thinner of minor importance only and need not be considered, from above and below until it vanishes. (The thinning since input data are extremely uncertain for several effect from below is due to deep-seated geothermal (geological) reasons. The test runs also showed that heat.) The approximate time for complete thaw, and n = 15 was a minimum to give reasonable results, but the effect on temperatures in deeper layers, can be that high numbers of considered intervals did not assessed by a simple approach. The main approximation markedly change the results of the calculations. Fig­ is that sensible heat is neglected in the thaw time ure 4 gives the calculated influence of (spatially) calculation. In this limit the calculation becomes very simple: the temperature varies linearly above, o -2 -4 -6 -8°C in' in, and below the ice-rich permafrost, and at its base the heat flux has its geothermal value. The time t for permafrost of initial thickness Yo to thaw completely after the surface has warmed up to a positive temperature Tp can then be shown to be given by:

(2 ) 3

where Cl and C2 are defined by: (3) 5 and C2 = q/k . (4)

km 7 Here h is the latent heat per unit volume, q is the geothermal heat flux and k is the thermal conducti­ z vity of the thawed sediments. (More details are given in the appendix.) Given the approximations leading to Fig.4. Thermal effects 6T* of the palaeoclimatic evo­ Equation (2), it is probably val id to adopt for these lution on present-day temperature profiles for parameters the values recently determined for Prudhoe different ice-age surface temperatures. _1°C may Bay, Alaska, where the lithology is rather similar represent conditions of temperate glacier beds, and the heat flux has a fairly typical value (Lachen­ whereas -2 to _5°C is typical for periglacial bruch and others 1982). Then Cl ~ 2.2 ka °C- 1 and conditions and non-temperate glacier beds during C2 ~ 0.028°C m-l. the same time period. z is depth below surface. Figure 5 shows the resulting dependence of thaw time on initial permafrost thickness, assuming differ- varying surface temperatures during full glacial development: _1°C may be considered as a typical value for formerly temperate glacier beds (phase equilibrium temperature, cf. Harrison 1975), whereas -2 to _5°C is more likely to represent periglacial 10 0 e conditions and conditions in regions of non-temperate glacier beds. I t can be seen that, in the Swiss plat­ 12 eau, past cold periods during the Quaternary, and c especially the last ice age, have lowered ground 01 temperatures at a depth of about 1 000 to 1 500 m by 10 UJ 5°C or more. This corresponds to a heat flow correc­ tion of 0.25 to 0.5 heat flow units (HFU), or 20 to 30% of the observed values. The palaeoclimatic effect 8 t in regions of formerly temperate glacier beds is 1 to 3°C smaller than in formerly periglacial regions. In practice, it is probably impossible to observe this 6 temperature difference, because of the limited accur­ typical range of acy of geothermal measurements (temperature in deep bore holes, thermal characteristics of rock samples). Swiss Plateau Moreover, the effect can easily be offset by heat 4 convection due to circulating groundwater. The effect may, however, be enhanced by the temporary formation of ice rich permafrost in unconsolidated sediments. 2

negligible effects BUFFERI NG OF THE THPERATURE RESPONSE OF DEEPER ROCKS • BY ICE-RICH PERMAFROST Vo Porous sediments at the surface can have an im­ ,1 ,2 ,3 .4 ,5 ,6 km portant effect on the response of deeper rocks to surface temperature change. Th e source of the effect Fig.5. Thaw time t for ice-rich permafrost as a func­ is the high latent heat exchan ge required for the tion of initial thickness Yo and post-glacial sur­ formation and thawing of ice-rich permafrost in face temperature. During the disappearance of porous sediments. In the Swiss plateau, with post­ of ice-rich permafrost, the surface temperature of glacial surface temperatures around +8 to +10°C, the Earth can be considered to remain at O°C. Thus formation of ice-rich permafrost in periglacial the palaeoclimatic curve represented in Figure 2 areas must have occurred for the last time about may have to be changed in cases of long thaw times.

58

Downloaded from https://www.cambridge.org/core. 30 Sep 2021 at 11:07:59, subject to the Cambridge Core terms of use. Haebe rLi and other>s : GeothemlCLL ef f ects of 18 ka BP ice

ent post-glacial ground surface temperature values. Ba10baev V T, Devyatkin V N, Kutasov I M 1978 Con­ To assess the effect of relict permafrost on the evo­ temporary geotherma1 conditions of the existence and lution of temperatures in deeper layers during the development of permafrost. In Pe mlCLfrost Second post-glacial period, it is reasonable to assume that InternationaL Co nfe rence . USSR Co ntribution. the presence or absence of permafrost determines the Washington, DC, National Academy of Sciences: 8-12 surface boundary condition to which these temperatures Barsch D 1968 Perig1azia1e Seen in den Karstwannen respond. While permafrost is present, this temperature des Schweizer Juras. Regio Ba siZiensis 9( 1) : 115-134 is O·C; after thawing it is the actual post-glacial Bau1in V V, Dubikov G I, Uvarkin Vu T 1978 The main surface temperature (e.g. +8 to +10'C in the Swiss features of the structure and development of the plateau). Only if the permafrost persists for an permafrost of the west Siberian plain. In Pe mla­ appreciable fraction of the post-glacial period will frost Second Inte rna tiona~ Conference . USSR Con­ the ,permafrost buffering effect be significant. From tribution. Washington, DC, National Academy of Figure 5 it can be seen that negligible effects only Sciences: 63-67 should occur when, for example, Y (100 m and Birch F 1948 The effects of P1eistocene climatic Tp ) 2'C, but that significant ef~ects may occur when variations upOD geotherma1 gradients. Ame rican Vo = 400 m and Tp = 8·C. Jou rnal of Science 246: 729-760 B1atter H, Haeber1i W 1984 Modelling temperature distribution in Alpine glaciers. Annals of C~aci ­ OISCUSSIONS AND CONCLUSIONS ology 5: 18-22 Thermal effects of Quaternary temperature and ice Bodmer P Unpublished Beitrage zur Geothermie der conditions in the Swiss plateau produced by heat Schweiz. (PhD thesis, Eidgenossische Technische diffusion in low-porosity bedrock have lowered today's Hochschu1e, Zurich, 1982) rock temperatures by the order of 5 to 6'C at about Cars1aw H S, Jaeger J C 1959 Conduct ion of heat in 1 to 2 km depth in regions of former temperate glacier solids . Oxford, Clarendon Press beds; in former perig1acia1 regions, the temperature Haeberli W 1981 Ice motion on deformable sediments. reduction is somewhat greater. Eff'ects of latent heat Jou rna ~ of C~acio~ogy 27(96): 365-366 exchange during the formation and thawing of ice-rich Haeberli W 1982 Klimarekonstruktionen mit permafrost in high-porosity sediments are superimposed G1etscher-Permafrost-Beziehungen. Universitat on the effects of heat diffusion. With perig1acia1 BaseL . Mate rialien zur Physiogeognzphie 4: 9-17 surface temperatures of -3'C and less, ice-rich Haeber1i W In press Permafrost-glacier relation­ permafrost in unconso1idated sediments of unlimited ships in the Swiss Alps - today and in the past. thickness could have reached a thickness of 200 m or In Fourth International Conference on Pe mlaf rost, more at 18 ka BP in the Swiss plateau; in most places, Fai r>banks, Alaska , 1983 . Proceedings however, the thickness of unconso1idated sediments is Haeberli W, Schneider A, Zoller H 1976 Der "See­ much less than 200 m. For typical values of 50 m the wener See": refraktionsseismische Untersuchungen thaw time of ice-rich permafrost is of the order of an einem spatg1azia1en bis fruhho10zanen Bergsturz­ 200 to 300 a, which is too short to be important •. Stausee im Jura. Regio Ba sUiensi s 17(2): 133-142 Effects due to the former existence and buffering Harrison WD 1975 Temperature measurements in a influence of ice-rich permafrost in the Swiss plateau temperate glacier. JournaL of GLacioLogy 14(70): need therefore be considered only in situations which 23-30 Jackli H 1970 Die Schweiz zur letzten Eiszeit. hav~ exceptionally thick (>100 m), high-porosity sedlments. The fact that the limit to the thickness In Atlas der> Schweiz Blatt 6 . Wabern-Bern, of ice-rich permafrost is given by the lithology of Eidgenos s ische Landestopographie the plateau rather than by the pa1aeotemperature Ke11er 0, Krayss E 1980 Die letzte Vor1andvereisung makes it difficult to check pa1aeoc1imatic recon~truc­ in der Nordosts chweiz und im Bodensee-Raum (Stadi­ tions by observations of corresponding temperature a1er Komp1ex Wurm - Stein am Rhein). EcLogae Ge o­ anomalies. On the other hand, this also means that Logicae HeLvetiae 73( 3): 823-838 geocryo10gica1 evidence of the formation of cold Lachenbru ch A H, Sass J H, Marshall B V, Moses T H Jr ice-rich permafrost is not necessarily contradicted 1982 Permafrost, heat flow, and the geothermal by the lack of pronounced temperature anomalies in regime at Prudhoe Bay, Alaska. Journal of Ceo­ deep bore holes within the Swiss plateau. physica~ Re seamh 87(B11): 9301-9316 The situation is obviously different in somewhat Manabe S, Broccoli A J 1984 Ice-age climate and co1der regions with thicker deposits of unconso1i­ continental ice sheets: some experiments with a dated, porous sediments. There, even relict perma­ general circulation model. AnnaLs of GLacioLogy 5: frost may be found which must have survived many 100-105 thousands of years (Ba10baev and others 1978, Bau1in Meier M F 1961 Mas s budget of South Cascade Glacier, and others 1978). In Switzerland, the largest thermal 19 57-60. US Ceo~ogica~ SU'Y'Vey . Professional Pape r> effects may be expected in deep Alpine valleys where 424-B : 206-211 the basal temperature of ice-age glaciers was well Rellstab W 1982 Der> Einfluss des Paleok~imas auf below O·C (B1atter and Haeber1i 1984), and sediment da s Tempe nz tu rfe~d in der Schweiz . Zurich, thickness can be considerable (cf. Aric and Eidgenossische Technische Hochschu1e. Institut fur Steinhauser 1976). Geophysik (Interner Arbeitsbericht) (1981, revised 1982) Rudloff H v 1980 Das Klima - Entwicklung in den letzten Jahrhunderten im mitteleuropaischen Raume ACKNOWLEDGEMENT S (mit einem Ruckblick auf di e postglaziale Periode). The authors are grateful to Or H Roth1isberger, In Oeschger H, Messerl i B, Svil ar M (eds) fus Dr A Iken and Jurg A1ean (VAW/AG / ETHZ) for reading KLima - Analysen und ModeLLe , Geschichte und the manuscript critically. Pame1a A1ean assisted in Zukunft . Ber1 in, Springer: 125-148 preparing the manuscript and Werner Nobs completed Rybach L, Jaffe F C 1976 Geotherma1 potential in the drawings. Switzerland. In Second United M7 tions Symposium on the DeveLopment and Use of Ceothe mlal Resoumes, San Fnznci sco, California, USA , 20- 29 ML y 1975. Proceedings Vo~ 1 . Washington, DC, US Government REFERENCES Printing Office: 241-244 Aric K, Steinhauser P 1976 Geophysika1ische Unter­ Rybach L, Bodmer P, Pavoni N, Mue11er S 1978 Siting suchung des Innta1-Untergrundes bei Thaur, ost1ich criteria for heat extraction from hot, dry rock: von Innsbruck. zeitschrift fur GL e tsche rkunde application to Switzerland. Pure and Appli ed Ge o­ und GL aziaLgeoLogi e 12(1): 37-54 physi cs 116: 1211-1224

59

Downloaded from https://www.cambridge.org/core. 30 Sep 2021 at 11:07:59, subject to the Cambridge Core terms of use. HaebeY'Zi and ot he r's: Ceo t he r>ma L effects of 18 ka BP ice

Schindler C, Rothlisberger H, Gyger M 1978 Glaziale T + Stauchungen in den Niederterrassen-Schottern des Aadorfer Feldes und ihre Deutung. Ec Logae Geo­ l ogicae He l veti ae 71(1): 159-174 Shih Ya-feng, Hsieh Tze-chu, Cheng Pen-hsing, y, Li Chi-chun 1980 Distribution, features and vari­ ations of glaciers in China. Int er>na t iona l Associ­ ...... at ion of Hyd r'ol ogi caL Sciences PubLication 126 0) Cl) (Workshop at Riederalp 1978 - WOY'ld GLacie r' Inven­ C 0 tO r'yJ : 111-116 lo.... lo.... CU '+- Van der Meer J J M 1980 Different types of wedges Q) CU Vo in deposits of WUrm age from the Murten area (west­ .0 E ern Swiss plain). EcLogae Geologicae Helvetiae lo.... Q) Q) 73(3) : 839-854 () 0.. Van der Meer J J M Unpublished The Fribourg area, Switzerland - a study in Quaternary geology and soil development. (PhD thesis, Universiteit van Amsterdam, 1982) I Washburn A L 1979 Geoc Y'yology - a sUr>Vey of pe r>i­ V2 glacial pr'ocesses and enviY'Onments. London, Edward 1 Arnold --1-

APPENDIX A rigorous calculation of the effect of ice-rich permafrost on deep temperature response would require a complete, simultaneous calculation of the tempera­ z ture fields above, in, and below the permafrost, using appropriate Stefan boundary conditions at the ice interfaces. However, a simple approach for dealing Fig.1A. Thawing ice-rich permafrost. For explanation with the permafrost separately is possible if sensible see text. heat is neglected. This is equivalent to assuming an infinitely rapid temperature response, and in this limit temperature varies linearly above, in, and below temperature. This tends to hold ir. -; . ,;- -saturated sedi­ the permafrost, as shown in Figure lA. Yo is the ments of high porosity. Ha..Iever, the neglect of the initial permafrost thickness at time t = 0, and Y1 and sensible heat below the permafrost is a more delicate Y2 the thicknesses that have thawed from top and matter, because the heat flux q will not be the bottom, respectively, at time t. Products of the thaw steady-state geothermal value implied by the rates and the latent heat per unit volume h are given approximation, but a transient value that should be by heat f1 uxes calculated in the course of a complete solution of the problem. Nevertheless it is clear that the exact dYl kT value of q will not be important as long as most of h (Al) the thawing occurs from the top : dt Y1 and » dt dt dY2 h q , (A2) From Equations (AI) and (A2) this is true if -dt - kT where k is the thermal conductivity of thawed perma ­ » 1, frost, T is the surface temperature, and q is the qYI heat flux into the base of the permafrost. Integration gives whi ch is not violated seriously enough to affect the conclusions in the main text.

YI - t (A 3) ~h

and Vl ~t (M) h

Thaw.ing is compl ete when

YI + Y2 = Yo, (AS) with VI and Y2 given by Equations (A3) and (M). Solution of Equation (A5) for the t ime when this occurs gives Equation (2) in the main text. The val idity of the key ap prox imation, the neg­ lect of all sensible heat, needs to be considered. This heat above and in the permafrost can be neglected as long as (pc) ~T « h, where (pc) is volumetric heat capacity and ~ T the difference be­ tw een surface temperature T and initial permafrost

60

Downloaded from https://www.cambridge.org/core. 30 Sep 2021 at 11:07:59, subject to the Cambridge Core terms of use.