A cool early

John W. Valley* William H. Peck* Elizabeth M. King* Department of and , University of Wisconsin, Madison, Wisconsin 53706, USA Simon A. Wilde School of Applied Geology, Curtin University of Technology, GPO Box U1987, Perth, Australia

ABSTRACT No known rocks have survived from the ®rst 500 m.y. of Earth history, but studies of single suggest that some continental crust formed as early as 4.4 Ga, 160 m.y. after of the Earth, and that surface temperatures were low enough for liquid water. Surface temperatures are inferred from high ␦18O values of zircons. The range of ␦18O values is constant throughout the (4.4±2.6 Ga), suggesting uniformity of processes and conditions. The hypothesis of a cool suggests long intervals of relatively temperate surface conditions from 4.4 to 4.0 Ga that were conducive to liquid- water oceans and possibly life. impacts during this period may have been less frequent than previously thought.

Keywords: Archean, , oxygen, isotopes, , impacts.

INTRODUCTION ``hell-like,'' in which any crust was destroyed and Table DR-11). Figure 2 shows histograms The young Earth (4.5±3.8 Ga) was highly by meteorite impact or melting, all water was of ␦18O for these samples in comparison to energetic. Sources of heat, both internal and ex- vaporized, and the atmosphere was steam rich olivine and zircon from Earth's mantle. Sev- ternal, were signi®cantly greater than at any (see Pollack, 1997). eral conclusions can be made. There is no sys- time since. The combined energy of planetary The only terrestrial materials that are tematic trend for ␦18O versus the age of the accretion, gravitational settling of Earth's me- known to be older than 4 Ga are detrital and Archean magmas that represent 40% of the tallic core, meteorite impacts, and possibly col- xenocrystic zircons from . Values average close to the prim- lision with a Mars-size body was suf®cient to (see Compston and Pidgeon, 1986; Wilde et itive value of Earth's mantle (5.3½ Ϯ 0.3½ melt the entire Earth during its accretion, dated al., 2001; Peck et al., 2001; Mojzsis et al., based on zircon xenocrysts in kimberlite, Fig. as 4.56±4.45 Ga (Halliday, 2000). Furthermore, 2001), including a recently discovered crystal 2B; Valley et al., 1998), in high-temperature before 4.4 Ga, the heat produced by radioactive from the , Western Australia, with equilibrium with the value of mantle-derived decay was approximately ®ve times higher than an age of 4.404 Ga (Wilde et al., 2001). At olivine from peridotite xenoliths and from today, not including the effects of short-lived present, these isolated zircon crystals are the oceanic island basalts (Fig. 2A; Mattey et al., nuclides (Pollack, 1997). Although these events only direct evidence of conditions on the ear- 1994; Eiler et al., 1997). The average value surely caused massive melting before 4.45 Ga, liest Earth. for Archean igneous zircons from four conti- estimates of surface temperatures after 4.45 Ga nents, exclusive of the Jack Hills (Fig. 2C), is depend critically on the magnitudes of mete- ZIRCONS 5.6½, which would be in high-temperature orite bombardment and atmospheric insulation, Zircon is a common trace mineral in granitic equilibrium with a basalt (whole rock) of which are uncertain. rocks that preserves detailed records of magma ϳ6½ or a granite of 6.5½±7.5½. The distri- The most direct evidence of past tempera- genesis. Zircons also form in ma®c rocks or bution is slightly skewed toward higher val- tures on the surface of Earth comes from ox- during metamorphism. Zircons provide reliable ues. High values of 7½±8½ are found even 18 ygen isotope compositions (␦ O) in rocks; crystallization ages, magmatic ␦18O values, and in the oldest zircon crystal (Fig. 1). Thus, the however, no rocks have been identi®ed from magmatic rare earth element (REE) composi- processes affecting the ␦18O of magmatic zir- the earliest Earth. Dating of lunar rocks (Sny- tions, and may contain inclusions of other co- cons show no evidence of change during this der et al., 2000) indicates a crust by ca. 4.4 genetic minerals. These characteristics can be period. The relative constancy of ␦18OinAr- Ga, but no such evidence exists on Earth. The preserved even when a crystal has been re- chean zircons is in contrast to zircons younger oldest known rocks on Earth formed at 4.0 Ga moved from its host rock by weathering, trans- than 2 Ga, which have magmatic values from (Bowring and Williams, 1999), and the oldest ported as a detrital grain, deposited, hydrother- 0½ to 13.5½ due to of Earth's crust water-laid sediments are 3.8±3.6 Ga (Nutman mally altered, and metamorphosed (Valley et to contain more high ␦18O sediments, evolved et al., 1997; Whitehouse et al., 1999). The ab- al., 1994). In well-de®ned situations, zircons ``S-type'' magmas, and hydrothermally altered sence of rocks older than ca. 4 Ga, taken with provide information of even earlier events, be- rocks (Peck et al., 2000). other evidence, has been interpreted to suggest fore magmatic crystallization. Thus, studies of a hot early Earth, sometimes called or zircon can provide important ground truth for models of the early Earth. 1GSA Data Repository item 2002034, Table *E-mail: ValleyÐ[email protected] Pre- DR1, Values of ␦18O and U-Pb ages of zircons from sent addresses: PeckÐDepartment of Geology, Col- Archean igneous rocks, is available from Docu- gate University, Hamilton, New York 13346, USA; Archean Zircon Record ments Secretary, GSA, P.O. Box 9140, Boulder, CO 18 KingÐDepartment of Geography-Geology, Illinois Figure 1 shows the ␦ O of zircons with 80301-9140, [email protected], or at www. State University, Normal, Illinois 61790, USA. measured U-Pb ages of 2.6 to 4.4 Ga (see text geosociety.org/pubs/ft2002.htm.

᭧ 2002 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or [email protected] Geology; April 2002; v. 30; no. 4; p. 351±354; 3 ®gures; Data Repository item 2002034. 351 Figure 1. Crystallization age (U-Pb) and ␦18O for Archean magmatic zircons. Distribution of magmatic ␦18O values does not change throughout Archean. Most magmas had primitive ␦18O value similar to that in mantle today (mantle zircon), but some zircon values are as high as 7.5½. High-␦18O zircons and host magmas resulted from melting of protoliths that were altered by interaction with liquid water at low temperatures near surface of Earth (see text). Time line (inset, lower right) shows: (1) accretion of Earth, (2) formation of Moon and Earth's core, (3) minimum age of liquid water based on high ␦18O zircon, (4) Acasta gneiss, and (5) Isua metasedimentary rocks. VSMOW is Vienna standard mean ocean water.

Jack Hills Detrital Zircons Any model should consider the features of The distribution of ␦18O values measured in this 4.4 Ga zircon: ®nely concentric magmatic Jack Hills detrital zircons (Peck et al., 2001; zoning as revealed by cathodoluminescence; Wilde et al., 2001; Mojzsis et al., 2001) REE patterns that include negative Eu and matches that of younger Archean samples positive Ce anomalies, and heavy-REE enrich- (Fig. 1). The only unusual aspect of the oldest ments to ϳ10 000 times chondrites; zoning in 18 zircons is their age. The higher Jack Hills ␦ O that correlates to trace element zoning Figure 2. Histograms of ␦18O. A: Olivine from 18 ␦ O values (6½±8½) are the same as those (7.4½ vs. 5.0½); inclusions of SiO2 that in- mantle xenoliths (Mattey et al., 1994) and Ha- of zircons from syntectonic to posttectonic sa- dicate a quartz-saturated granitic magma; and waiian basalts (Eiler et al., 1997). B: Zircon nukitoids (see King et al., 1998) and plutons a Pb-Pb age of 4.404 Ϯ 8 Ga with Ͼ99% xenocrysts from kimberlites in South Africa intruded into metasedimentary rocks (Fig. 2, concordant U-Pb ages (Peck et al., 2001; (Valley et al., 1998). C: Zircons from igneous 18 rocks of Superior province, Canada (King et C and D). In younger granites, such high-␦ O Wilde et al., 2001). al., 1998). D: Ion microprobe analyses of sin- zircons are known to form in magmas with Could the 4.4 Ga Jack Hills zircon be ex- gle zircons from Jack Hills, Western Austra- ␦18O (whole rock) values to 10½ (Valley et traterrestrial in origin? Zircons as old as 4.32 lia (Peck et al., 2001). Superior province his- 18 al., 1994), indicating burial and melting of Ga occur in lunar granophyre (Meyer et al., togram (C) peaks at mantle ␦ O value with tail toward higher values. Jack Hills zircons rocks that were subjected to low-temperature 1996; Snyder et al., 2000); however, the com- (D) are higher in ␦18O than mantle. VSMOW 18 alteration, weathering, or diagenesis (Taylor bination of high ␦ O values, quartz satura- is Vienna standard mean ocean water. and Sheppard, 1986; Gregory, 1991). It is sig- tion, and enriched REE compositions are not ni®cant that such alteration requires liquid wa- consistent with formation on the Moon or in ter near the surface of Earth. High ␦18O values any known meteorite. Furthermore, meteoritic Jack Hills sample. Furthermore, at magmatic are routinely found in younger granitoids and zircons are very rare on Earth. The close sim- temperatures, the ␦18O of a zircon in equilib- interpreted to result from interaction of low- ilarity of the 4.4 Ga crystal to other zircons in rium with such an evolving magma would be temperature liquid water with their protoliths. the same rock suggests that they all formed on constant because the fractionation, ⌬18O We conclude that this is also the best inter- the same planetary body. Clearly, the Jack (magma-zircon), would increase at approxi- pretation for the protoliths of magmas that Hills zircons did not all come from meteorites. mately the same rate as ␦18O (magma) due to formed Jack Hills zircons. The chemical compositions of Jack Hills zir- the increasing felsic component of the melt. Figure 1 shows that the higher values of cons are characteristic of those from continental- Thus, high ␦18O zircons suggest supracrustal ␦18O are evenly distributed throughout the Ar- type crust on Earth, such as might have chemi- input to magmas. chean, including the value of 7.4½ Ϯ 0.7½ cally differentiated from the mantle in a tectonic from two analyses of the core of the 4.4 Ga environment similar to Iceland today. However, SURFACE OF EARTH AT 4.4±4.0 Ga Jack Hills zircon (Peck et al., 2001). Because there are no known high-␦18O reservoirs in The constancy of ␦18O for zircons over the of the unique signi®cance of this single zircon the mantle. Rare high values in young mantle- period 4.4±2.6 Ga (Fig. 1) suggests that the crystal, all processes to account for the high derived basalts either result from crustal con- magmatic processes and protoliths at 4.4±4.0 ␦18O value have been considered. Possible tamination or from subduction of hydrother- Ga were similar to those at 3.8±2.6 Ga. This scenarios include: (1) formation in a meteor- mally altered crust (Eiler et al., 1997). It is similarity would not be expected if the older ite, (2) fractional crystallization of the host possible to raise the ␦18O of a magma by 1½± rocks formed in the absence of water and granitic magma from a primitive terrestrial 2½ through fractional crystallization and sep- younger rocks were derived from hydrother- melt at 4.4 Ga, (3) exchange of the protolith aration of low-␦18O ma®c minerals (Taylor mally altered material (see also Campbell and of the granitic magma with steam rather than and Sheppard, 1986). However, this process Taylor, 1983). Liquid-water oceans were re- liquid water, and (4) exchange of the protolith would not be suf®cient to explain the ␦18Oof quired by 3.8±3.6 Ga to precipitate chemical with liquid water. 8½±10½ for the magma inferred from the sediments found in (Nutman et al.,

352 GEOLOGY, April 2002 1997; Whitehouse et al., 1999), and by 3.5 Ga, data is that surface temperatures were below insulation, Earth's surface could have cooled sedimentary rocks, stromatolites, and pillow ϳ200 ЊC at 4.4 Ga. This temperature is in the to within the stability of liquid water as indi- basalts indicate widespread submarine condi- range of liquid water for a modern-size hy- cated by high ␦18O zircons. Cooler surface tions. Thus, the constancy of ␦18O and the drosphere and is low enough to cause partial conditions would be enhanced by lower solar presence of granitoids suggest not only that or complete condensation of a steam atmo- luminosity. there were oceans back to 4.4 Ga, but also that sphere to form oceans. surface temperatures were not too dissimilar Meteorite Bombardment from those of younger periods. IMPLICATIONS OF A COOL EARLY The most signi®cant addition of energy to Limits on temperature for near-surface al- EARTH the surface of Earth from 4.4 to 3.8 Ga may 18 teration of the protolith of the high-␦ O 4.4 The conclusion that liquid water was prev- have been meteorite impacts. Although the 18 Ga magma can be calculated from the ␦ Oof alent as early as 4.4 Ga is consistent with the crater record on Earth has been destroyed by the early hydrosphere, the fractionation of energetic of early Earth, but provides tectonics, the rate of early meteorite bombard- 18 16 O/ O between basalt (or granite) and water, an important boundary condition. Signi®cant ment of the inner Solar System can be esti- and the stability of liquid water versus steam. uncertainties exist in models for the sources mated by examination of craters on well- Temperatures must have been in the range that and timing of the early high-energy events. preserved surfaces of the Moon and Mars. On 18 would elevate ␦ O by exchange with surface Low surface temperatures and liquid water are the Moon, where craters have been dated, the waters. Compositions of hydrothermally al- easier to reconcile with these models if: (1) time-integrated ¯ux of meteoritic material 18 18 tered upper (␦ O Ͼ6½) and lower (␦ O core formation and the genesis of the Moon from 4.4 to 3.8 Ga is estimated by surveying Ͻ6½) ocean crust have not signi®cantly were earlier than 4.45 Ga, allowing more time crater density, measuring the addition of sid- 18 changed through time, suggesting that ␦ O for cooling; (2) if the Moon formed by some erophile elements such as Ir to the lunar crust, (seawater) ഠ 0½ for at least the past 3.6 b.y. process other than impact by a Mars-size and determining the extent of impact stirring 18 (Gregory, 1991). If fully equilibrated, ␦ O body; (3) if atmospheric insulation was low; or gardening of the crust (Arrhenius and Lep- (basalt) is 8½ at ϳ200 ЊC in equilibrium with or (4) if the meteorite bombardment of Earth land, 2000; Hartmann et al., 2000). However, 18 liquid water (␦ O ϭ 0) and 250 ЊC with from 4.4 to 4.0 Ga was less intense than has the relation of meteorite ¯ux to age is not steam. These are common temperatures for been proposed. uniquely determined. Hartmann et al. (2000) seawater alteration of the upper ocean crust The formation of the Moon and segregation reviewed these studies and estimated that the (Gregory, 1991). Thus a seawater-dominated of Earth's core were recently dated as 4.5 Ga meteorite ¯ux was 2 ϫ 109 times greater at 18 hydrothermal system will raise the ␦ Oofba- by Halliday and coworkers (see Halliday, the beginning of accretion of the Earth than salt to 8½ only at ``low'' temperatures below 2000), and although the impact model for today (Fig. 3, dashed line), and the ¯ux de- 200 ЊC for liquid water oceans. Meteoric wa- forming the Moon is widely accepted, signif- clined rapidly such that most of Earth's mass 18 ters are lower in ␦ O than seawater and thus icant questions persist (see Jones and Palme, was acquired during the ®rst 10±100 m.y. exchange with fresh water would make cal- 2000). If the Moon did form by impact, a Accretion of the Earth was essentially com- culated temperatures lower. If the amount of large, relatively late addition to the heat and plete by ca. 4.45 Ga; however, two models water was limited or exchange was not fully mass of Earth is indicated. Alternatively, cap- exist for the subsequent period. A high me- equilibrated, then the temperature estimate ture of a planetesimal would not cause addi- teorite ¯ux has long been recognized ca. 3.9 would also be lower. tional heating of Earth. Either nonimpact or Ga, suggesting higher ¯ux for the period The question arises, could such exchange an earlier date are consistent with Xe-closure 4.45±3.9 (Fig. 3, dashes). However, recent Ar- be dominated by steam at 100±250 ЊC? If no estimates that suggest that Earth has not melt- Ar dating of lunar impact glasses supports ear- oceans existed on modern Earth and all water ed since 4.45 Ga, but that a major disturbance lier proposals of an alternate hypothesis, that formed a steam atmosphere, the atmospheric occurred before that time (see Podosek and late heavy bombardment at 3.9 Ga was caused pressure would be ϳ270 bar (Zahnle et al., Ozima, 2000). Planet-scale closure to rare-gas by processes in the outer Solar System, by de- 1988), which is above the critical point of wa- escape would not date a melting event, but ¯ection of asteroids, or by collisions of geo- ter (220 bar). A boiling-ocean and steam- rather would record the later solidi®cation of centric material (Tera et al., 1974; Cohen et atmosphere system would be buffered in pres- crust suf®cient to retard outgassing. Thus, al., 2000). If Earth was subjected to a spike sure and temperature along the liquid-vapor considerable cooling had already occurred by of meteorite bombardment ca. 3.9 Ga, then in- phase transition, and today's ocean volume 4.45 Ga. terpolation between 3.9 and 4.5 Ga would be would require temperatures above the critical The nature and timing or even existence of incorrect, and the period from 4.4 to 4.0 Ga point of water (ϳ374 ЊC depending on salin- magma oceans is uncertain (Newson and might have been relatively tranquil (Arrhenius ity) to completely vaporize. The temperature Jones, 1990). Although it is possible that and Lepland, 2000; Hartmann et al., 2000), for H2O as 100% steam (Ͼ374 ЊC) is signif- Earth was fully melted before 4.45 Ga, crys- consistent with cooler surface temperatures icantly above the range (Ͻ200 ЊC) indicated tallization of such a has been (Fig. 3, solid line). by the high ␦18O (zircon) values. Furthermore, modeled to occur on the scale of 1±10 m.y. in The evidence for liquid water is consistent the processes by which large volumes of crust the absence of atmospheric insulation (Spohn with a cool early Earth and heavy bombard- become hydrothermally altered require that and Schubert, 1991; Pollack, 1997). ment at 4.0±3.8 Ga. The four high-␦18O, pre± ¯uids recirculate and that the ratio of water to Before 4.45 Ga, the hydrosphere was surely 4.0 Ga zircons each must be younger than the rock be k1 (by weight) in order to raise ␦18O 100% vapor, but the nature of this atmosphere time of liquid water interaction with their in protolith of zircon-forming magmas. Steam depends on the composition and quantity of magmatic protoliths; hence, these dates are is known to discharge readily from hydrother- volatiles. Large impacts might have stripped shown as minimum ages at the top of Figure mal systems, but not to recharge them. It is Earth of its atmosphere, facilitating radiative 3. The chemically precipitated sedimentary unlikely that a low-density vapor ef®ciently cooling. Conversely, a thick insulating atmo- rocks from Greenland could have been depos- penetrated to suf®cient depths in the crust to sphere might have absorbed radiated energy, ited at 3.6 or 3.8 Ga (Nutman et al., 1997; alter the protolith of the 4.4 Ga zircon-forming raising surface temperatures by more than 100 Whitehouse et al., 1999). At present, there is magma. Thus, the best explanation of existing ЊC (Abe et al., 2000). Even with signi®cant no zircon evidence for liquid water during the

GEOLOGY, April 2002 353 tion from 4.32 to 3.88 Ga: Meteoritics and Planetary Science, v. 31, p. 370±387. Mojzsis, S.J., Harrison, T.M., and Pidgeon, R.T., 2001, Oxygen-isotope evidence from ancient zircons for liquid water at the Earth's surface 4300 Myr ago: Nature, v. 409, p. 178±181. Newson, H.E., and Jones, J.H., 1990, Origin of the Earth: Oxford, Oxford University Press. Nutman, A.P., Mojzsis, S.J., and Friend, C.R.L., 1997, Recognition of Ն3850 Ma water-lain sediments in West Greenland and their signi®- cance for the early Archean Earth: Geochimica et Cosmochimica Acta, v. 61, p. 2475±2484. Peck, W.H., King, E.M., and Valley, J.W., 2000, Ox- ygen isotope perspective on Precambrian crustal growth and maturation: Geology, v. 28, p. 363±366. Peck, W.H., Valley, J.W., Wilde, S.A., and Graham, C.M., 2001, Oxygen isotope ratios and rare earth elements in 3.3 to 4.4 Ga zircons: Ion microprobe evidence for high ␦18O continental crust in the early Archean: Geochimica et Cosmochimica Acta, v. 65, p. 4215±4229. Podosek, F.A., and Ozima, M., 2000, The xenon age of the Earth, in Canup, R.M., and Righter, K., eds., Origin of the Earth and Moon: Tucson, University of Arizona Press, p. 63±72. Pollack, H.N., 1997, Thermal characteristics of the Archaean, in deWit, M., and Ashwal, L., eds., Greenstone belts: Oxford, Oxford University Press, p. 223±232. Figure 3. Estimates of meteorite-impact rate for ®rst 2 b.y. of Earth his- Snyder, G.A., Borg, L.E., Nyquist, L.E., and Tay- tory. Two hypotheses are shown: exponential decay of impact rate (dash- lor, L.A., 2000, Chronology and isotopic con- es, Hartmann et al., 2000) and cool early Earth±late heavy bombardment straints on lunar evolution, in Canup, R.M., (solid curve, this study). Approximate half-life is given in million and Righter, K., eds., Origin of the Earth and for periods of exponential decline in ¯ux. In either model, spikes oc- Moon: Tucson, University of Arizona Press, curred owing to isolated large impacts. Evidence for liquid water comes p. 361±395. 18 from high-␦ O zircons (Ͼ4.4 to Ͼ4.0 Ga) and sedimentary rocks (Isua, Spohn, T., and Schubert, G., 1991, Thermal equili- 3.8±3.6 Ga). The cool early Earth hypothesis (solid curve) suggests that bration of the Earth following a giant impact: impact rates had dropped precipitously by 4.4 Ga, consistent with rela- Geophysical Journal International, v. 107, tively cool conditions and liquid water. p. 163±170. Taylor, H.P., and Sheppard, S.M.F., 1986, Igneous rocks; I. Processes of isotopic fractionation and salt phenocrysts: Geochimica et Cosmochim- late heavy bombardment, and more data are isotope systematics, in Valley, J.W., et al., eds., being sought. The earlier period, 4.4±4.0 Ga, ica Acta, v. 61, p. 2281±2293. Stable isotopes in high temperature geological Gregory, R.T., 1991, Oxygen history of seawater processes: Mineralogical Society of America is called Hadean by some. However, we sug- revisited: Timescales for boundary event gest that Earth was not ``hell-like'' then and Reviews in Mineralogy, v. 16, p. 227±271. changes in the oxygen isotope composition of Tera, F., Papanastassiou, D.A., and Wasserburg, may have been hospitable to the earliest life. seawater, in Taylor, H.P., et al., eds., Stable G.J., 1974, Isotopic evidence for a terminal isotope geochemistry: San Antonio, Texas, lunar cataclysm: Earth and Planetary Science ACKNOWLEDGMENTS Lancaster Press, p. 65±76. Letters, v. 22, p. 1±21. We thank Mary Diman, Colin Graham, Henry Halliday, A.N., 2000, Terrestrial accretion rates and Valley, J.W., Chiarenzelli, J.R., and McLelland, Pollack, Mike Spicuzza, Allan Treiman, and two re- the origin of the Moon: Earth and Planetary J.M., 1994, Oxygen isotope geochemistry of viewers for comments on and assistance with vari- Science Letters, v. 176, p. 17±30. zircon: Earth and Planetary Science Letters, ous aspects of this paper. Research was funded by Hartmann, W.K., Ryder, G., Dones, L., and Grin- v. 126, p. 187±206. the National Science Foundation (EAR-99-02973) spoon, D., 2000, The time-dependent intense Valley, J.W., Kinny, P.D., Schulze, D.J., and Spi- and the Department of Energy (93ER14389). bombardment of the primordial Earth/Moon cuzza, M.J., 1998, Zircon megacrysts from system, in Canup, R.M., and Righter, K., eds., kimberlite; oxygen isotope variability among REFERENCES CITED Origin of the Earth and Moon: Tucson, Uni- mantle melts: Contributions to Mineralogy and Arrhenius, G., and Lepland, A., 2000, Accretion of versity of Arizona Press, p. 493±512. Petrology, v. 133, p. 1±11. Whitehouse, M.J., Kamber, B.S., and Moorbath, S., Moon and Earth and the emergence of life: Jones, J.H., and Palme, H., 2000, Geochemical con- 1999, Age signi®cance of U-Th-Pb zircon data Chemical Geology, v. 169, p. 69±82. straints on the origin of the Earth and Moon, from early Archean rocks of west Greenland: Bowring, S.A., and Williams, I.S., 1999, Priscoan in Canup, R.M., and Righter, K., eds., Origin (4.00±4.03 Ga) orthogneisses from northwest- A reassessment based on combined ion-micro- of the Earth and Moon: Tucson, University of probe and imaging studies: Chemical Geolo- ern Canada: Contributions to Mineralogy and Arizona Press, p. 197±216. Petrology, v. 134, p. 3±16. gy, v. 160, p. 201±224. King, E.M., Valley, J.W., Davis, D.W., and Ed- Campbell, I.H., and Taylor, S.R., 1983, No water, no Wilde, S.A., Valley, J.W., Peck, W.H., and Graham, wards, G.R., 1998, Oxygen isotope ratios of granitesÐNo oceans, no continents: Geophys- C.M., 2001, Evidence from detrital zircons for ical Research Letters, v. 10, p. 1061±1064. Archean plutonic zircons from granite-green- the existence of continental crust and oceans Cohen, B.A., Swindle, T.D., and Kring, D.A., 2000, stone belts of the Superior province: Indicator on the Earth 4.4 gyr ago: Nature, v. 409, Support for the lunar cataclysm hypothesis of magmatic source: Precambrian Research, p. 175±178. Zahnle, K.J., Kasting, J.F., and Pollack, J.B., 1988, from Lunar meteorite impact melt ages: Sci- v. 92, p. 365±387. Evolution of a steam atmosphere during ence, v. 290, p. 1754±1756. Mattey, D., Lowry, D., and Macpherson, C., 1994, Earth's accretion: Icarus, v. 74, p. 62±97. Compston, W., and Pidgeon, R.T., 1986, Jack Hills, Oxygen isotope composition of mantle peri- dotite: Earth and Planetary Science Letters, evidence of more very old detrital zircons in Manuscript received July 31, 2001 Western Australia: Nature, v. 321, p. 766±769. v. 128, p. 231±241. Revised manuscript received December 12, 2001 Eiler, J.M., Farley, K.A., Valley, J.W., Hauri, E., Meyer, C., Williams, I.S., and Compston, W., 1996, Manuscript accepted December 19, 2001 Craig, H., Hart, S.R., and Stolper, E.M., 1997, Uranium-lead ages for lunar zircons: Evidence Oxygen isotope variations in ocean island ba- for a prolonged period of granophyre forma- Printed in USA

354 GEOLOGY, April 2002