Accretion of Moon and Earth and the Emergence of Life

Chemical Geology 169Ž. 2000 69±82 www.elsevier.comrlocaterchemgeo

Accretion of Moon and Earth and the emergence of life

G. Arrhenius a,), A. Lepland a,b a Scripps Institution of Oceanography, UniÕersity of California, La Jolla, San Diego, CA 92093-0220, USA b Institute of Geology, Tallinn Technical UniÕersity, EE-0001 Tallinn, Estonia Received 18 March 1999; accepted 9 June 2000

Abstract

The discrepancy between the impact records on the Earth and Moon in the time period, 4.0±3.5 Ga calls for a re-evaluation of the cause and localization of the late lunar bombardment. As one possible explanation, we propose that the time coverage in the ancient rock record is sufficiently fragmentary, so that the effects of giant, sterilizing impacts throughout the inner solar system, caused by marauding asteroids, could have escaped detection in terrestrial and Martian records. Alternatively, the lunar impact record may reflect collisions of the receding Moon with a series of small, original satellites of the Earth and their debris in the time period about 4.0±3.5 Ga. The effects on Earth of such encounters could have been comparatively small. The location of these tellurian moonlets has been estimated to have been in the region around 40 Earth radii. Calculations presented here, indicate that this is the region that the Moon would traverse at 4.0±3.5 Ga, when the heavy and declining lunar bombardment took place. The ultimate time limit for the emergence of life on Earth is determined by the effects of planetary accretion Ð existing models offer a variety of scenarios, ranging from low average surface temperature at slow accretion of the mantle, to complete melting of the planet followed by protracted cooling. The choice of accretion model affects the habitability of the planet by dictating the early evolution of the atmosphere and hydrosphere. Further exploration of the sedimentary record on Earth and Mars, and of the chemical composition of impact-generated ejecta on the Moon, may determine the choice between the different interpretations of the late lunar bombardment and cast additional light on the time and conditions for the emergence of life. q 2000 Elsevier Science B.V. All rights reserved.

Keywords: Moon; Early Earth; Emergence of life; Lunar bombardment; Earth satellites; Archean sediments; Chemofossils

1. Introduction al.,Ž. 1977 . It was later generally assumed that this protracted event must have affected the Earth vio- The manifestations of a late heavy bombardment lently and rendered our planet uninhabitable well of the Moon were first observed by Papanastassiou past 3.8 GaŽ 1 Ga-109 yr. , with major impacts and WasserburgŽ. 1971 and Tera et al., Ž. 1974 , its extending to 3.45 GaŽ. Fig. 1 . Such a bombardment significance was further discussed by Wasserburg et is thought to have severely affected the possibilities for life to have emerged on Earth in the earliest ) Ž Corresponding author. Archean Stevenson, 1988; Sleep et al., 1989; Maher E-mail addresses: [email protected]Ž. G. Arrhenius , and Stevenson, 1988; Chyba, 1993; Oberbeck and [email protected]Ž. A. Lepland . Fogleman, 1989, 1990. . In contrast, the record from

though the original sedimentary features bear an overprint of metamorphism and tectonism. The Isua and Akilia metasediments contain ubiquitous chemofossils with carbon isotopic composition suggestive of biochemically advanced microbial life forms Ž.Mojzsis et al., 1996 . Their presence indicates that the planet was not sterilized by impacts during the time periods of deposition of these sediments. This apparent lack of counterparts on Earth, to the late impact features on the Moon, together with new data obtained from the studies of the Martian meteorites, require a re-evaluation of the events in Earth±Moon space in this critical time interval for the emergence of life. The 3.86 Ga age of the Akilia formation determined by Nutman et al.Ž. 1997, 2000 , has been questioned by Kamber and MoorbathŽ. 1998 and Whitehouse et al.Ž. 1999 , proposing an age of 3.65 Ga for this formation. Regardless of ultimate consensus on the age difference between the Isua and Akilia formations, this difference would, in the present context, be insignificant relative to the spread in time of the late impacts on the Moon, 4.0±3.45 Fig. 1. Radiometric ages of samples from the lunar highlands, GaŽ. Fig. 1 , essential for the comparison with the illustrating the culmination of the late lunar bombardment around record on Earth. 3.8±3.9 Ga and tailing off toward 3.5 GaŽ. after Dalrymple, 1991 . The dashed line shows the minimum age of the Isua metasediments and delineates the overlap of the decaying late lunar bombardment with the early sedimentation record of Earth. Com- 2. The picket fence model and the sedimentary pare interpretations in Fig. 3. record

We are considering two possible explanations for the oldest known Isua and Akilia metasedimentary the apparent incongruence between the lunar and rocks from southern West Greenland, extending back terrestrial records. If the bombardment was caused in time beyond 3.75 GaŽ Fig. 2; Moorbath et al., by invading translunar objects, it is likely to have 1973; Rosing et al., 1996; Nutman et al., 1997. show been episodic. The effects could then have been sequences of banded ironstones without any clearly overlooked in the discontinuous geologic record in- identifiable disturbances that could have formed at vestigated from the early Archean. This concept is asteroidal impacts. Such disturbances, in the form of embodied in the ``picket fence'' modelŽ. Fig. 3 surge deposits, are exemplified by the Ordovician proposed by Mojzsis and ZahnleŽ Mojzsis et al., Lockne impact structure in SwedenŽ. Sturkell, 1998 1998. . Life could, in each such major occult impact and consist of coarse-grained rock fragments, trans- event, have been extinguished, only to arise again in ported and sorted by the strong tidal flow generated the intervening quiescent time intervals of a few ten by the impact. Crushing-, flow- and transport-effects or hundred million years. Or it could have found a at impacts of the size indicated by the lunar maria, niche for survival in the deep ocean or crust, and and scaled up for Earth's gravitation, would have have spread again from there, between each pair of been correspondingly violent. They would have left major assaultsŽ. Sleep et al., 1989 . records that are likely to be distinguishable from the Sequences of metasedimentary cherts and banded layer structure of the banded iron formation, even iron formations in the Isua Supracrustal BeltŽ. ISB , G. Arrhenius, A. LeplandrChemical Geology 169() 2000 69±82 71

Fig. 2. Time scale of events related to accretion of Moon and Earth and the emergence of life. The 3.8 Ga age line indicates the culmination of the late bombardment observed on the Moon. referred to above, are exposed in outcrops and in a pel, 1991. . The metasedimentary units are commonly series of several-hundred meter-long drill coresŽ Ap- intersected by intrusions of mafic rocks. Although individual 10±100 m sections of coherent sequences, represent comparatively long accumulation periods, no certain evidence has so far been found for catastrophic surge deposits. Reliable stratigraphic interpretation is, however, hindered by the tectonic overprinting; the formation is isoclinally folded. Therefore, it cannot be excluded that some of the sequences represent repetitions of a single unitŽ Ros- ing, 1998. . In the northeastern part of the ISB, the sequence of finely-laminated banded iron formation, contains strongly deformed 0.1±1 m thick layers, sandwiched between undeformed strata. These deformed packages probably mark tectonic shear zones Fig. 3. Three generalized interpretations of the late lunar bombardment data shown in Fig. 2. The concurrent record on Earth is that resulted from small-scale translations parallel to represented by the Isua sediments that have a minimum age of the bedding. Some of the thicker deformed packages 3.75 Ga. Retrieved segments of these deposits do not bear evi- Ž.Fig. 4 , display grading of the component rock dence of contemporaneous bombardment of Earth at the scale fragments, opening the possibility that they could be observed on the Moon. This suggests, either that the correspond- surge deposits formed by violent resuspension and ing time±rock units on Earth are missing or have not been found Ž.cf. Fig. 4 , or that the late lunar bombardment was restricted to sorting during resettling. If meteorite impact is as- lunar phase space.Ž Zahnle and Mojzsis, unpubl., reproduced in sumed as a cause of such implied surges, an en- Mojzsis et al., 1998. . hancement should be evident of the platinum group 72 G. Arrhenius, A. LeplandrChemical Geology 169() 2000 69±82

process, but was largely limited to the lunar orbit. This possibility may be evaluated against the back- ground of several current theories for the origin of the Moon, that can all be taken to provide more or less compelling support for such a conjecture. One currently popular scenario for the formation of the Moon assumes that Earth, in a relatively advanced state of accretion, including or followed by core formation, would have collided with a hypo- thetical planet moving in an Earth-crossing orbit and with a mass larger than that of MarsŽ Hartmann and Davis, 1975; Cameron and Ward, 1976; WankeÈ et al., 1984; Cameron and Benz, 1991; Cameron, 1985, 1997; Canup and Esposito, 1996; Ida et al., 1997. Ejecta from the impact would have been placed in prograde equatorial orbit around Earth and coalesced to form the Moon. In view of the fact that large impacts occur relatively late on the Moon without leaving an observed record on Earth, one may consider a scenario proposed by Canup and EspositoŽ. 1996 . They suggested that a massive proto-Moon coalesced from the inner part of the ejecta ring and receded from Earth due to tidal angular momentum transfer. The proto- Moon later overtook and collided with a series of Fig. 4. A rock sequence in the Isua banded iron formation which originally formed smaller moonlets. This scenario could possibly indicate a disturbance of the primary sedimentary requires a lapse time of about 400 million years structures by a seismic- or impact-related surge. The rock section, between the initial emplacement of the Moon and the seen in the photograph to be layered between sections of finely late bombardment. The delay time, aside from the laminated banded iron sediment, consists of rock fragments that become successively finer from leftŽ. bottom? to right Ž. top? . orbital distance from Earth of such swept up moon- Such grading may be interpreted as the result of violent suspen- lets, would depend on the rate of recession of the sion and sorting at settling of rock fragments, such as, after an proto-Moon from Earth, and thus, on tidal dissipa- earthquake or an impact. However, most of the Isua sediment tion. Bodily tides are generally considered of minor sequences bear signs of tectonic deformation, therefore, indicating importance in the present solid Earth, compared to that the described structure could also be the result of shear. The possibility that this is an impact-related deposit can be tested by the ocean, where dissipation is mainly restricted to analyses of the concentration of PGE that are characteristic com- shallow seas and regions around submarine ridges ponents of undifferentiated extraterrestrial materials. and seamountsŽ. Munk and Wunsch, 1998 . In the collisional theory for the origin of the Moon, consid- erable uncertainty is introduced for tidal dissipation and rate of recession, since the hydrosphere must elementŽ. PGE concentrations, in and above such have been totally evaporated, creating a thermal beds, but has not yet been observed. blanket perpetuating this state for a long time. On the other hand, extensive liquefaction of the outer layer of the Earth, creating a magma ocean, would have 3. Lunar bombardment restricted to the Moon? increased the dissipation factor. Other theories for the formation of the proto-Moon As another possibility we suggest that the late also offer possibilities for explaining late collision lunar bombardment was not a solar system-wide events occurring in lunar phase space. With regard to G. Arrhenius, A. LeplandrChemical Geology 169() 2000 69±82 73 formation and orbital evolution of the Moon, the extensively analyzed capture theoryŽ Gerstenkorn, 1955, 1969; Lyttleton, 1967; Singer, 1968, 1972; AlfvenÂ and Arrhenius, 1963, 1969; Urey and Mac- Donald, 1971; Kaula, 1971. deserves further consid- eration. The capture theory has, perhaps temporarily, become overshadowed by the collision theory. This invokes, as supporting evidence, material similarities between the Earth's mantle and the Moon, and a purported low probability of capture. Perhaps sub- liminally, it also draws on the computationally and visually attractive drama of Worlds in Collision. At closer scrutiny, the two former arguments against capture, however, assume diminishing importance. Ž. The low density of the Moon is comparable not only Fig. 5. Masses in grams of systems of secondary bodies around their magnetized primaries. The diagram shows the abnormally to the silicate mantle±crust of the Earth but in large masses of the retrograde Neptunian satellite, Triton, which is general, to circumsolar bodies beyond Earth, such as considered to be capturedŽ. McCord, 1966 , and of our present Mars and the asteroids. The similarity of oxygen Moon, which may likewise be captured from solar orbit. The isotope ratios of Moon and Earth, compared to Mar- original, ``normal'' satellites, postulated to have been swept up by tian meteorites, has been invoked as another material the Moon and causing the major impacts in late lunar bombardment, would have had a total mass estimated by interpolation to argument for the collision hypothesis. However, ma- be of the order 1021 ±1022 g, compared to 0.74=1026 g for the terials that isotopically are strikingly similar to those Moon.Ž. From AlfvenÂ and Arrhenius, 1972a . of the Moon and Earth are represented by certain types of meteoritesŽ. Clayton and Mayeda, 1996 that could have served as a source for the Moon and have also been invoked as a likely source material for the EarthŽ. Herndon, 1996 . sive study of the lunar orbital evolution by numerical The collision cross-section of Earth is about six integrationŽ. Goldreich, 1963, 1966 characteristically times smaller than that of a presumed dissipative leads to a high inclination of the initial near-Earth Earth ring region, and the capture probability from a orbit of the Moon. A high initial inclination of the corresponding encounter orbit is amplified by many proto-Moon can also be explained if it coalesced orders of magnitude by repeated passages. The im- from a debris ring around the EarthŽ Ward and probability argument would seem to apply to the Canup, 2000. . collision hypothesis more seriously than to capture, In the context of the present discussion of the late particularly when the efficient capture-dissipation lunar bombardment, the question of a capture vs. mechanism proposed by Kaula and HarrisŽ. 1973 is collision origin of the Moon is, however, of sec- taken into accountŽ. see also, Wood, 1977 . This ondary importance. The question of interest here is if mechanism involves collisions by the capturand with a time delay of several hundred million years can be particle swarms expected in near-Earth orbit, adding expected between emplacement of the Moon in significantly to energy loss by friction. Capture is near-Earth orbit and its interaction with original, furthermore, generally considered as the most proba- much smaller satellites at a relatively large distance ble mechanism for emplacement of many other satel- from the Earth. Such a delay would mainly depend lites with abnormal featuresŽ. Burns, 1977 , one of on the distance from Earth of such moonlets and the the most striking examples being Neptune's retro- rate of recession of the proto-Moon. Initially ignor- grade and massive moon, TritonŽ McCord, 1966; cf. ing the potentially delaying effect of Earth±Moon Fig. 5. . Finally, capture of the Moon would take resonancesŽ. AlfvenÂ and Arrhenius, 1969 or Moon± place from a highly inclined orbit. The comprehen- Venus resonancesŽ. Kaula, 1971 , the rate of increase 74 G. Arrhenius, A. LeplandrChemical Geology 169() 2000 69±82 in distance from Earth of the receding Moon is given the Earth, are considered in several theories. As an Ž.MacDonald, 1964 by example, the accretion theory developed by Safronov d R 3kmr5 Ž.1954, 1960, 1972 based on orbital dynamics, pre- s1r2 Ž.GM 11r2 Ž.1 dicts the formation of swarms of secondary bodies dtQ R around the planetsŽ. Safronov and Ruskol, 1977 but where R is the semimajor axis of the proto-lunar does not specify their radial distribution. Considering orbit, k the Love number and Q the specific dissipa- the seminal influence that Safronov's general theory tion factor for Earth, G the gravitational constant, M has had on developments in the West, the neglect of the mass, r the radius of Earth, and m the mass of their application to satellite formation in the litera- the proto-Moon. ture is notable. The characteristic time for a given R is More specific predictions about the radial distri- R 4p 5r3 Q bution of planets and satellites are based on the tss RMm13r27r62implied hydromagnetic control of circumstellar and r 1r2 Ž. dRdr ž/3 3kG circumplanetary plasma as observed in the planetary Integration of Eq.Ž. 2 , coupled with modern mea- ionospheres and in the protoplanetary circumstellar surements of the deceleration of the lunar longitude disk mediumŽ AlfvenÂ and Arrhenius, 1974, 1976; and thus, the slowing of the Earth, give the startling Montmerle and Andre,Â 1988; Hjalmarson and result of only 1.75 Ga for the evolution of the lunar Friberg, 1988; Bertout, 1988. . In the hydromagnetic orbit from close vicinity to the Earth to its present theory for secondary body formation, the emplace- position, in clear contradiction of geological evi- ment process is thought to be controlled by the denceŽ. MacDonald, 1964; Kaula, 1971 . Lambeck critical velocity, Vc , of ionization for the infalling Ž.1975 and Kaula and Harris Ž. 1973 have pointed to neutral component of the interstellar cloud plasma the uncertainly in the value of Q over time as a and by its interaction and partial co-rotation with the probable reason for this dilemma and the possibility magnetic field of the central bodyŽ. De et al., 1977 . that the controlling oceanic topography may have As verified by effects observed in the space medium, sr 1r2 been different in the past, with substantially lower VcŽ.2eV ionM a, where V ion is the ionization dissipationŽ.Ž. larger Q . Munk 2000 has furthermore potential of atoms with mass Ma and e the electron pointed to the likelihood that in near-interaction be- charge. The critical velocity effect was discovered in tween Moon and Earth, vorticity dissipation is likely laboratory plasma experimentsŽ Danielsson, 1970, to have been saturated, introducing an effective delay 1973. , showing selective ionization of neutral gas factor in the recession of the satellite, regardless of moving through a magnetic field. AlfvenÂ and Arrhe- its origin. niusŽ. 1974 visualized this effect as the physical If the late lunar bombardment was due to collision explanation of what AlfvenÂ had previously pointed of the proto-Moon with original moonlets, these out as a characteristic band structure of secondary must have been located at such a distance from Earth bodies around their magnetized primaries in the solar that the Moon traversed between emplacement, about systemŽ. Fig. 6 . The statistical significance of the 4.4 Ga and bombardment between 4.0 and 3.5 Ga. proposed band structure was investigated by Arrhe- Small bodies could have formed outside of a major nius and ArrheniusŽ. 1988 . The physical basis of the lunar embryo in a collisionally generated accretion band structure model led De, in 1972Ž. see De, 1978 diskŽ. Canup and Esposito, 1996; Ida et al., 1997 . to the prediction of rings inside the Roche limit of However, such moonlets close to Earth can not be Uranus; these rings were later discovered by Elliot et held responsible for late lunar bombardment since al.Ž. 1977 ; the relation of the prediction to the dis- the large exponent for R in Eq.Ž. 2 causes the covery is discussed by BrushŽ. 1996 . massive proto-Moon to traverse near-Earth space The observed parallelism in distribution of both comparatively rapidly, even if vorticity saturation is primary and secondary bodies in terms of gravita- taken into account. tional potential, together with the proposed explana- The formation of regular satellite systems, such as tionŽ. Fig. 6 , implies that the Earth would also have those of the outer planets, and by analogy, also of been originally endowed with a satellite system of G. Arrhenius, A. LeplandrChemical Geology 169() 2000 69±82 75

Fig. 6. Distribution of secondary bodiesŽ. planets and satellites around their primaries Ž Sun and the magnetized planets . . Orbital distances

Ž.rorb to the central bodies are normalized by relating them to the mass Ž.Mc of the central body as a measure of gravitational potential r energy, Egg, drawn as log E Ž.cm g . The individual systems are plotted in order of central body mass, normalized to the mass of the Sun Ž.M( . The distribution of secondary bodies is perceived as a series of bands, corresponding to actual groups of critical velocities for ionization of dusty gas falling into the circumsolar region from the interstellar source cloud. The ragged edges of the bands indicate uncertainly in their extension; the physical reason for the slope of the bands is discussed in AlfvenÂ and ArrheniusŽ. 1974, 1976 . In the present context, the diagram serves the purpose of illustrating the possibility that Earth may originally have had two sets of small satellites, swept up by the abnormally large Moon, during the evolution of its orbit. The location of the outer band corresponds to the approximate

Moon±Earth distance, 3.8±3.5 Ga. The Moon is slowly receding and is presently located at log Eg 17.2, just below the letter ``R'' in ``EARTH''. The Moon±Earth resonances, where recession of the Moon could have been retarded in its early history are discussed in AlfvenÂ and ArrheniusŽ.Ž 1969 from AlfvenÂ and Arrhenius, 1972b . . possibly a dozen moonlets and inside the Roche our proto-Moon, regardless of its mode of formation, limit, perhaps also a ring. As is generally conceded, by capture or collision, receded toward its current 76 G. Arrhenius, A. LeplandrChemical Geology 169() 2000 69±82 location at 60 Earth radii as a result of transfer of have held ammunition for the late bombardment angular momentum by tidal exchange with Earth. It until perturbations of the characteristic long orbital would, in this process, have slowly approached and excursions from the libration points brought some of interacted with any normal satellites that may have these bodies into collision with the Moon. Small existed in its path, either by coalescence or ejection. amounts of dust are observed trapped around the Creation of the maria by late impact with pre-exist- Lagrangian points todayŽ Roach, 1975; Winiarsky, ing Earth satellites was proposed by KaulaŽ. 1971 . 1989. . Observable dust particles are probably contin- The probability of ejection by interaction with such ually removed from the libration region by the moonlets, can perhaps best be judged by the fact that Poynting±Robertson effect and radiation pressure, our solar system contains a large number of plane- and replenished by accumulation of interplanetary tary satellites that are considered to have been cre- dust. A dynamic calculation is needed to evaluate the ated by collisional accretion of smaller bodies. actual residence time of larger and thus, practically One group of Earth's original satellites would invisible solid bodies around the libration points. have been located in the Martian band, between 14 Assuming the direct applicability of Eq.Ž. 2 , an and 34 Earth radii, and another in the transposition estimate can be made from Eq.Ž. 3 of the radial of the Saturnian±Uranian band, between 2.9 and 6.3 distance range from Earth in which the receding Earth radiiŽ. Fig. 6 . The latter, together with the proto-Moon would have encountered the outer band inside-Roche ring transposed from Ariel and Mi- of small regular Earth satellites in a time period randa, would have been swept up first, when the around 3.8 Ga, 600 million years after the initial precursor of our present Moon receded from near- emplacement of the Moon in near-Earth orbit by Earth orbit and could have contributed to the fric- collision or capture, assumed to have taken place tional medium in the Kaula±Harris mechanism for around 4.4 Ga. capture assistance. In the hydromagnetic model for satellite formation, the putative moons that appear in 6.5 tbb600 R Earth's gravitational equivalent to the Martian satel- ss Ž.3 t lite band, would be the most likely candidates for i 4400ž/ 60 interaction with the proto-Moon when it, at the time r of the late lunar bombardment, had receded to 1 4± With t b being the orbital evolution time, about 600 1r2 of its present distance from Earth. Ma, up to the late lunar bombardment, the radial

With sufficiently close similarity of orbital veloci- distance, Rb, of the Moon from Earth at the time of ties at the time of the encounters, the collision bombardment is found to be 44 Earth radii. This velocities could, under conditions outlined by Canup distance value is probably an upper limit, since the and EspositoŽ. l.c. , have been close to the escape recession of the Moon could have been retarded by velocity for the Moon, around 2.4 kmrs. Collision spin-orbit resonances with EarthŽ AlfvenÂ and Arrhe- velocities of the order of 2 kmrs would have sub- nius, 1972a. andror by resonance between Moon stantial cosmetic effects on the surface of the Moon; and VenusŽ. Kaula, 1971 . ejecta with speeds below the escape velocity would The distribution of accreting material in the re- have been recaptured and not be able to overcome gion around 40 Earth radii extrapolated by hydro- the energy barrier for escape and diversion to Earth- magnetic considerations would thus, seem to support impacting orbits. the concept of the late lunar bombardment as a series Debris from collisions exceeding escape velocity of collisions, possibly accompanied by ejections, as may have remained in lunar intersecting orbits for an the receding Moon crossed the orbits of an outer set extended time before ultimate capture by the Moon. of original, proportionately small satellites of the Resonance locking in the Earth±Moon system may Earth. A potential impediment to such a hypothesis, have contributed to such delay, so far, of undeter- particularly in conjunction with capture, could result mined durationŽ. Canup and Esposito, l.c. . Metastable from interference of the proto-Moon with the origi- storage around the Lagrangian points, L4 and L5, nal small satellites already on its initial approach to 608 before and after the Moon in its orbit may also Earth during the early evolution of a capture orbit. G. Arrhenius, A. LeplandrChemical Geology 169() 2000 69±82 77

The probability for such premature destruction of marauders, they would be expected to consist of these bodies would be decreased by the high inclina- undifferentiated cometary- or meteorite-type mate- tion of a capture orbit. However, model calculations rial, conferring their characteristic excess of PGE to of interactions at this stage would be desirable. the ejecta, with allowance for dilution with lunar The Martian surface, if represented by the mete- target material. If they were to be identified with orite, ALH 84001, has not been melted since its partly differentiated asteroids, some impact material crystallization time, 4.5 Ga. ago. The surface layer of would be depleted in PGE, but other samples, en- all of the terrestrial planets would have been melted riched in metal, would have a correspondingly en- if the projectiles causing the bombardment of the hanced PGE content. If, however, such a signature is Moon had come from an external sourceŽ Stevenson, found to be consistently absent in the samples repre- 1988. . Extended observations from Mars could senting the collision events, the most likely interpre- therefore contribute to the interpretation of the late tation would be that the impacts were caused by lunar bombardment as a parochial lunar event or metal-depleted low density materials of the kind that alternatively as an invasion, affecting the entire inner formed the Moon and possibly original Earth satel- solar system. lites. An integrating studyŽ. Ryder and Mojzsis, 1998 of the total surface accumulation of PGE over lunar history, indicates that this amount is only 1.3% of 4. Evidence from terrestrial sediments what would be expected if the source had been meteoriticŽ. asteroidal impactor material with a high PGE content. Again, this suggests that the late lunar All current theories of lunar origin and collision bombardment was caused by projectiles with compo- history, suffer from dynamic difficulties and remain sition similar to the moon rather than asteroids. speculative. However, there are possibilities for veri- The necessarily speculative views of the hardships fication of the circumstances around the late impacts. or even survival of life during the late lunar bom- One consists of detailed examination of long se- bardment may now be replaced by searches in the quences of the oldest laminated sediments, searching oldest preserved records on our planet. The manifes- for and identifying depositional irregularitiesŽ Stur- tations of planetary impacts may probably be found kell, 1998. , and meteorites, along with interplanetary even in older samples from ancient terraces on Mars. dust embedded in the sedimentsŽ Schmitz et al., Life on Earth, instead of as is often quoted, originat- 1997. . Quantitative element and mineral analyses of ing 3.5±3.8 billion years ago, is found to have such sediments give information about their content already developed to a high degree of autotrophic of extraterrestrial matter, and if the total rate of sophistication before these age limits. sedimentation can be estimated, the rate of accretion of the cosmic component can also be determined Ž.Chyba, 1991 . Analyses by Ryder and Mojzsis Ž.1998 of the metasediments from Akilia Island, off 6. Planetary accretion, the early surface state and the coast of southern West Greenland, show no the emergence of life indication of enhancement of PGE above the average for crustal terrestrial rocks. The processes that formed our planet and modified it in its initial stages, must also have determined the thermal state of the planetary surface and atmo- 5. Evidence from late lunar impact ejecta sphere. These developments predate the period of late heavy bombardment of the Moon, discussed A second approach toward verification involves above. The planetary accretion process and the initial analyses of such lunar samples that are chronologi- state of the Earth, have been subjects of intense cally identified with, and serving as the basis for scientific inquiry in recent years, and it is generally conclusions about the late bombardment on the agreed that the heat balance would have been con- Moon. If the colliding objects were interplanetary trolled mainly by the rate of growth of the planet by 78 G. Arrhenius, A. LeplandrChemical Geology 169() 2000 69±82 infall of extraterrestrial matter. If, as assumed in cooling, following removal of the presumed dense, many theories, all of the planetary source material thermally insulating proto-atmosphere. Such a sec- was emplaced around the Sun at one point in time ondary atmosphere would provide a medium for the Ž.``instant nebula'' theories , the accretion rate would eventual emergence of life; its source and composi- have increased exponentially with growing radius of tion are consequently of importance in this context. the planet, because of its concurrent increase in Different sources have been proposed for such a gravitational attraction. post-cooling atmosphere, surviving in chemically The result would be an accretional runaway catas- modified form today. One is an assumed ``late ve- trophe with an initial Earth, covered by a magma neer'' of cometary material; another is the solidify- ocean, extending perhaps to the depth of the upper ing volatile-laden upper mantle, releasing gases mantle and obviously detrimental for an early devel- originally dissolved in equilibrium with a dense opment of a hydrosphere and of life. High accre- proto-atmosphere. After arrival of a comparatively tional temperatures would also lead to the early high transparency of the atmosphere in the infrared removal of iron, which would sink toward the core, region, radiative cooling would rapidly allow a solid heating the planet even further by the release of the crust to form, and internal heat could, as today, be gravitational potential energy of the falling metal. In relatively efficiently transferred to the surface by the absence of such a development, metallic iron upwelling of molten rock in mid-ocean ridge sys- would remain distributed in the outer layers of the tems. planet and would serve as a potential reductant and In an intermediate accretion scenario, developed hydrogen generator from water and hydroxyl ion as by WetherillŽ. 1994 , calculations starting with al- discussed below. ready formed planetesimals, lead to a distribution of With the high temperature of the planet, its orbits with a wide range of eccentricities and inclina- volatiles would have been evaporated into a dense tions. In this case, planetesimals in near-Earth orbits atmosphere which would form an effective blanket, would have been captured at a high rate, leading to delaying radiative cooling for an extended time, as an adolescent Earth at high surface temperature. The on Venus. Protagonists of such scenarios therefore remaining population of planetesimals of approxi- propose that this heavy primitive atmosphere was mately lunar size and in highly eccentric and in- removed, e.g. hydrodynamically, with the aid of clined orbits, would have been captured at a lower hydrogen as a carrier gas, propagated by strongly rate, each with catastrophic local or regional effects enhanced UV radiation from the early SunŽ Hunten, but because of their small size, incapable of neces- 1993. . Such a process may also explain the critical sarily melting the entire planet. features of noble gas abundances in the present The heterogeneous accretion theory originally atmosphereŽ. Pepin, 1991 . Erosion of the atmosphere suggested by EuckenŽ. 1944 , and further developed by the impact of large bodies could also have effec- by Turekian and ClarkŽ. 1969 , specifically avoids the tively stripped off an early atmosphere but only catastrophic melting of the entire planet as a result of before Earth had accumulated a gravitational mass core formation by gravitational segregation in the comparable to MarsŽ Matsui and Abe, 1986; Ahrens already completed planet; a fateful consequence of et al., 1989. . Consequently, these processes could homogeneous accretion. The heterogeneous scheme only have been effective in the early stages of solar assumes that an early accretional episode involved and planetary evolution. After formation of the iron-rich material, spatially or temporally separated planet, stripping of the proto-atmosphere would, in the space medium. Later episodes would then add however, also be likely to accompany a collision the silicate-rich space condensates making up the event of the magnitude invoked in the currently main mass of the Earth, contained in its mantle and popular theory for the origin of the Moon and in- similar in density to Mars, the Moon and asteroids. volving a colliding planet larger than Mars, as dis- Various processes have been proposed that could cussed above. have led to such separationŽ e.g. Orowan, 1969; The stripping models also introduce the need for a Danielsson, 1973; Canup and Esposito, 1996. . Con- secondary atmosphere that could arise after radiative siderations of the evolution of the Earth's core based G. Arrhenius, A. LeplandrChemical Geology 169() 2000 69±82 79 on isotopic measurements of Hf and WŽ Halliday et al., 1996. would also be displaced from Earth to its dispersed precursor medium if the fractionation took place there. The actual occurrence of such fractionation in the inner solar system is suggested by the large density differences between, on one hand, Earth and Venus; on the other, Mars, Moon and the asteroids. Mercury represents material with particularly high density, presumably richer in iron than the other planets. In the case of the Moon, its proposed collisional removal from the mantle of a differentiated Earth has been suggested as a reason for its low densityŽ. WankeÈ et al., 1984 , but this does not explain the large density differences between the other bodies in the inner solar systems. AlfvenÂ and ArrheniusŽ. 1973, 1974 have empha- Fig. 7. Accretional model for Earth, based on the assumption of sized control of the early accretional process by the formation of the core Ž.;0.5 R[ from material initially emplaced orbital dynamics of the charged solid particles in the in Earth's accretional feeding zone in the solar nebula. Accumula- dusty plasma observed around young stars. They also tion of this mass would exponentially increase the gravitational attraction of the planetary embryo and end with a catastrophic avoided the hypothesis of an ``instant nebula'' that runaway accretion phase. After temporary depletion of Earth's assumes all source material for the planets, at one feeding zone, accretion of the core would be followed by accumu- single time, to be present in the circumsolar region, lation of the mantle by gradual infall and sweep-up of material after falling in from the interstellar source cloud for from the interstellar source cloud. Observations of T-Tauri stars our solar system. As observed in young stellar ob- support the notion of such episodic infall of matter onto their circumstellar discsŽ. see text . This model provides rationales for a jects surrounded by protoplanetary discs, such infall separate accretional composition of core and mantle, and for a low of condensates from the molecular source cloud, average temperature of the planetary surface during accumulation does not appear to be constant, but episodicŽ Lada, of the mantle. It therefore, also permits gradual retention of liquid 1988. . The variability of infrared luminosity in such water and emergence of life during the late stages of Earth objects, thought to result from variations in material accretion.Ž. From Arrhenius et al., 1974 . supply, can now, for obvious reasons, only be recorded on a human time scale. It is, however, also reasonable to assume lower frequency events of en- the average temperature of the Earth's surface main- hanced infall of material into the solar nebula. Such tained low; a ``hot spot accretion'' of a type advo- episodic emplacement events would have decisive cated by SchmidtŽ. 1944 and Vinogradov et al. effects on the orbital dynamics of the accreting Ž.1971 . In this extreme, the surface layer of the system. growing Earth may never have been entirely molten Elaborating on the hypothesis of heterogeneous at any given time. A hydrosphere could begin to accretion, AlfvenÂ and ArrheniusŽ. l.c. assumed an develop as soon as the gravitational field permitted initial accretion of the present metallic core of the the retention of liquid water at a size of the planetary EarthŽ. Fig. 7 . Later condensate emplacement embryo somewhat larger than the coreŽ. Fig. 7 . episodes would, with this gravitational center in Heterogeneous accretion of the planets is also place, lead to the efficient capture of new material by favorable for models seeking to maximize the rate of the Earth-embryo without an opportunity for this generation of methane and hydrogen from water and material to form large separate planetisimals. After carbon dioxide. This would be effectuated by the runaway formation of the core, the mantle of the oxidation of metallic iron forming a dispersed minor Earth could, under such conditions, have been grow- component in the silicate mantle. The production of ing as a result of a large number of minor impacts, hydrogen, methane and ammonia by this mechanism each causing local melting and degassing, but with would favor the formation of reduced organic com- 80 G. Arrhenius, A. LeplandrChemical Geology 169() 2000 69±82 pounds of biogenic importance. The destruction and BostromÈ and two anonymous reviewers, and are escape rate of reducing gases would also be mini- greatly appreciated. Generous research support was mizedŽ. Miller and Lyons, 1998 . The reduction pro- provided by NASA's Office of Space Science by cess would, in such models, be localized in the grants, NAGW-1031 and NAG5-4563, and by the temporary hot spots generated by impacts, separated Marianne and Marcus Wallenberg Foundation. in space and time. In heterogeneous accretion scenarios with early accretion of an iron-rich core, followed by comparatively slow growth of a silicate mantle at low aver- References age surface temperature, the planet could have already become habitable during or soon after plane- Ahrens, T.J., O'Keefe, J.D., Lange, M.A., 1989. Formation of tary formation, up to 4.5 billion years ago. This atmospheres during accretion of the terrestrial planets. In: would permit several hundred million years for the Atreya, S.K., Pollack, J.B., Matthews, M.S.Ž. Eds. , Origin and emergence of life, which is first seen in already Evolution of Planetary and Satellite Atmospheres. Univ. of enzymatically sophisticated form in rocks older than Arizona Press, Tucson, AZ, pp. 328±385. Alfven,Â H., Arrhenius, G., 1963. The early history of the Moon 3.75 billion years. Heterogeneous accretion theories, and the Earth. Icarus 1, 357±363. thus, connect with biopoesis scenarios that for intu- Alfven,Â H., Arrhenius, G., 1969. Two alternatives for the history itive reasons, wish for the longest possible time of the Moon. Science 165, 11±17. available for the emergence of life on Earth. It is Alfven,Â H., Arrhenius, G., 1972a. Origin and evolution of the worth emphasizing, however, that there is no factual Earth±Moon system. The Moon 5, 210±230. Alfven,Â H., Arrhenius, G., 1972b. Structure and Evolutionary or even speculative evidence, for the length of time History of the Solar System. Reidel, Dordrecht, Netherlands, required for the origin of lifeŽ. Orgel, 1998 . 276 pp. Alfven,Â H., Arrhenius, G., 1973. Structure and evolutionary history of the solar system, III. Astrophys. Space Sci. 21, 117± 176. Acknowledgements Alfven,Â H., Arrhenius, G., 1974. Structure and evolutionary history of the solar system, IV. Astrophys. Space Sci. 29, 63±159. Alfven,Â H., Arrhenius, G., 1976. Evolution of the Solar System, All of the considerations above, illustrate that NASA SP-345. U.S. Government Printing Office, Washing- time is of essence. The contributions by Professor ton, DC, 399 pp. Wasserburg and his collaborators to precision timing Appel, P.W.U., 1991, The Isua iron ore deposit at Isukasia, West Greenland. Open File Ser. 31, Geol. Survey of Greenland. and elucidation of processes on Earth and Moon and Arrhenius, G., De, B.R., Alfven,Â H., 1974. Origin of the ocean. In: in interplanetary space, have been of fundamental Goldberg, E.D.Ž. Ed. , The Sea 5 Wiley, New York, pp. importance for understanding events in the most 839±862. distant past. With this paper, we wish to express our Arrhenius, G., Arrhenius, S., 1988. Band structure of the solar thanks for the inspiration and insights that he has system; an objective test of the grouping of planets and satellites. Astrophys. Space Sci. 144, 357±372. provided and doubtlessly, will continue to give. Bertout, C., 1988. Observations of young stellar objects. In: The authors wish to thank Kevin Zahnle and Dupree, A.K., Lago, M.T.V.T.Ž. Eds. , Formation and Evolu- Stephen Mojzsis for extensive discussion of the range tion of Low Mass Stars. Kluwer Academic Publishing, Boston, of subjects at hand, and Peter Appel for supporting pp. 45±64. and coordinating our field studies as part of the Isua Brush, S.G., 1996. A History of Modern Planetary Physics. Cambridge Univ. Press, Cambridge, 800 pp. Multidisciplinary Research Project in southern West Burns, J.A., 1977. Orbital evolution. In: Burns, J.A.Ž. Ed. , Plane- Greenland and for discussion of the results. We are tary Satellites. The Univ. of Arizona Press, Tucson, AZ, pp. also indebted to Stephen Moorbath, Minik Rosing, 113±156. Chris Fedo, Martin Whitehouse, Victor McGregor Cameron, A.G.W., 1985. Formation of the prelunar accretion disk. and Alan Nutman for valuable views on tectonic, Icarus 62, 319±327. Cameron, A.G.W., 1997. The origin of the Moon and the single sedimentary and chronological relationships in the impact hypothesis. Icarus 126, 126±137. Isua complex. Constructive review comments were Cameron, A.G., Ward, W.R., 1976. The origin of the Moon. received from Dr. Christopher McKay, Prof. Robert Lunar Sci. 7, 120±122. G. Arrhenius, A. LeplandrChemical Geology 169() 2000 69±82 81

Cameron, A.G.W., Benz, W., 1991. Origin of the Moon and the Kamber, B.S., Moorbath, S., 1998. Initial Pb of the Amitsoq single impact hypothesis. Icarus 92, 204±216. gneiss revisited: implications for the timing of early Archaean Canup, R.M., Esposito, L.W., 1996. Accretion of the Moon from crustal evolution in West Greenland. Chem. Geol. 150, 19±41. an impact-generated disk. Icarus 119, 427±446. Kaula, W.M., 1971. Dynamical aspects of lunar origin. Rev. Chyba, C.F., 1993. The violent environment of the origin of life: Geophys. Space Phys. 9, 217±238. progress and uncertainties. Geochim. Cosmochim. Acta 57, Kaula, W.M., Harris, A.W., 1973. Dynamically plausible hypothe- 3351±3358. ses of lunar origin. Nature 245, 367±369. Chyba, C.F., 1991. Terrestrial mantle siderophiles and the lunar Lada, C.J., 1988. Infrared energy distributions and the nature of impact record. Icarus 92, 217±233. young stellar objects. In: Dupree, A.K., Lago, M.T.V.T.Ž. Eds. , Clayton, R.N., Mayeda, T.K., 1996. Oxygen isotope studies of Formation and Evolution of Low Mass Stars. Kluwer Aca- achondrites. Geochim. Cosmochim. Acta 60, 1999±2017. demic Publishing, Boston, pp. 93±109. Dalrymple, G.B., 1991. The Age of the Earth. Stanford Univ. Lambeck, K., 1975. Effects of tidal dissipation in the oceans on Press, 474 pp. the Moon's orbit and the Earth's rotation. J. Geophys. Res. 80, Danielsson, L., 1970. Experiment on the interaction between a 2917±2925. plasma and a neutral gas. Phys. Fluids 13, 2288±2294. Lyttleton, R.A., 1967. Dynamical capture of the Moon by the Danielsson, L., 1973. Review of the critical velocity of gas±plasma Earth. Proc. - R. Soc. A296, 285±292. interaction: Part I. Experimental observations. Astrophys. MacDonald, G.J.F., 1964. Tidal friction. Rev. Geophys. 2, 467± Space Sci. 24, 459±485. 541. De, B.R., Alfven,Â H., Arrhenius, G., 1977. The critical velocity Maher, K.A., Stevenson, D.J., 1988. Impact frustration of the phenomenon and the origin of the regular satellites. In: Burns, origin of life. Nature 331, 612±614. J.A.Ž. Ed. , Planetary Satellites. The Univ. of Arizona Press, Matsui, T., Abe, Y., 1986. Formation of a `magma ocean' on the Tucson, AZ, pp. 472±491. terrestrial planets due to the blanketing effect of an impact-in- De, B.R., 1978. A 1972 prediction of Uranian rings, based on the duced atmosphere. Earth, Moon, Planets 34, 223±230. AlfvenÂ critical velocity effect. Moon Planets 18, 339±342, McCord, T.B., 1966. Dynamical evolution of the Neptunian sys- ŽThe paper was originally submitted to Icarus in 1972 under tem. Astron. J. 71, 585±590. the title ``On the possibility of the existence of a ring of Miller, S.L., Lyons, J.R., 1998. Organic shielding of greenhouse Uranus'' but was rejected. Further details and comments are gases on early Earth. Science 279, 779. given in BrushŽ.. 1996 . Mojzsis, S.J., Arrhenius, G., McKeegan, K.D., Harrison, T.M., Elliot, J.L., Dunham, E., Mink, D., 1977. The rings of Uranus. Nutman, A.P., Friend, C.R.L., 1996. Evidence for life on Nature 267, 328±330. Earth before 3800 million years ago. Nature 384, 55±59. Eucken, V.A., 1944. UberÈ den Zustand des Erdinnern. Naturwiss. Mojzsis, S.J., Krishnamurthy, R., Arrhenius, G., 1998. Before 14, 112±121. RNA and after: geophysical and geochemical constraints on Gerstenkorn, H., 1955. UberÈ Gezeitenreibung beim molecular evolution. In: Cech, T., Gesteland, R., Atkins, J. Zweikorperproblem.È Z. Astrophys. 36, 245±274. Ž.Eds. , The RNA World II. Cold Spring Harbor Laboratory Gerstenkorn, H., 1969. The earliest past of the Earth±Moon Press, Cold Spring Harbor, pp. 1±47. system. Icarus 11, 189±207. Montmerle, T., Andre,Â P., 1988. X-rays, radio emission, and Goldreich, P., 1963. On the eccentricity of satellite orbits in the magnetism in low-mass young stars. In: Dupree, A.K., Lago, solar system. Mon. Not. R. Astron. Soc. 126, 257±288. M.T.V.T.Ž. Eds. , Formation and Evolution of Low Mass Stars. Goldreich, P., 1966. History of the lunar orbit. Rev. Geophys. 4, Kluwer Academic Publishing, Boston, pp. 225±246. 411±439. Moorbath, S., O'Nions, R.K., Pankhurst, J.R., 1973. Early Archean Halliday, A., Rehkamper,È M., Lee, D.-C., Yi, W., 1996. Early age for the Isua iron formation. Nature 245, 138. evolution of the Earth and Moon: new constraints from Hf-W Munk, W., Wunsch, C., 1998. Abyssal recipes: II. Energies of isotope geochemistry. Earth Planet. Sci. Lett. 142, 75±89. tidal and wind mixing. Deep-Sea Res. 45, 1977±2010. Hartmann, W.K., Davis, D.R., 1975. Satellite-sized planetesimals Munk, W., 2000. Personal communication. and lunar origin. Icarus 24, 504±515. Nutman, A.P., Mojzsis, S.J., Friend, C.R.L., 1997. Recognition of Herndon, J.M., 1996. Substructure of the inner core of the Earth. G3850 Ma water-lain sediments in West Greenland and their Proc. Natl. Acad. Sci. U.S.A. 93, 646±648. significance for the early Archaean Earth. Geochim. Cos- Hjalmarson, A., Friberg, P., 1988. Radio andŽ. sub millimeter mochim. Acta 61, 2475±2484. observations of the initial conditions for star formation. In: Nutman, A.P., Bennett, V., Friend, C., McGregor, V., 2000. The Dupree, A.K., Lago, M.T.V.T.Ž. Eds. , Formation and Evolu- early Archaean Itsaq Gneiss Complex of southern West Green- tion of Low Mass Stars. Kluwer Academic Publishing, Boston, land: the importance of geology in interpreting geochronologi- pp. 65±92. cal and isotopic data on early terrestrial evolution. J. Geol. Hunten, D.M., 1993. Atmospheric evolution of the terrestrial Soc.Ž.Ž London submitted for publication . . planets. Science 259, 915±925. Oberbeck, V.R., Fogleman, G., 1989. Estimates of the maximum Ida, S., Canup, R.M., Stewart, G.R., 1997. Lunar accretion from time required to originate life. Orig. Life Evol. Biosphere 19, an impact-generated disk. Nature 389, 353±357. 549±560. 82 G. Arrhenius, A. LeplandrChemical Geology 169() 2000 69±82

Oberbeck, V.R., Fogleman, G., 1990. Impact constraints on the Sleep, N.H., Zahnle, K.J., Kasting, J.F., Morowitz, H.J., 1989. environment for chemical evolution and the continuity of life. Annihilation of ecosystems by large asteroid impacts on the Orig. Life Evol. Biosphere 20, 181±195. early Earth. Nature 342, 139±143. Orgel, L.E., 1998. The origin of life Ð how long did it take? Stevenson, D., 1988. Greenhouses and magma oceans. Nature Orig. Life Evol. Biosphere 28, 91±96. 335, 587±588. Orowan, E., 1969. Density of the Moon and nucleation of planets. Sturkell, E.F.F., 1998. Impact related Ir anomaly in the Middle Nature 222, 867. Ordovician Lockne impact structure, Jamtland,È Sweden. GFF Papanastassiou, D.A., Wasserburg, G.J., 1971. Lunar chronology 120, 333±336. and evolution from Rb-Sr studies of Apollo 11 and 12 sam- Tera, F., Papanastassiou, D.A., Wasserburg, G.J., 1974. Isotopic ples. Earth Planet. Sci. Lett. 11, 37±62. evidence for a terminal lunar cataclysm. Earth Planet. Sci. Pepin, R.O., 1991. On the origin and early evolution of terrestrial Lett. 14, 281±304. planetary atmospheres and meteoritic volatiles. Icarus 92, 2± Turekian, K.K., Clark, S.P., 1969. Inhomogeneous accumulation 79. of the Earth from the primitive solar nebula. Earth Planet. Sci. Roach, J.R., 1975. Counterglow from the Earth±Moon libration Lett. 6, 346±348. points. Planet. Space Sci. 23, 173±181. Urey, H.C., MacDonald, G.J.F., 1971. Origin and history of the Rosing, M., Rose, N.M., Bridgwater, D., Thomsen, H.S., 1996. Moon. In: Kopal, Z.Ž. Ed. , Physics and Astronomy of the Earliest part of the Earth's stratigraphic record: a reappraisal Moon. Academic Press, New York, pp. 213±289. of the )3.7 Ga IsuaŽ. Greenland supracrustal sequence. Vinogradov, A.P., Yaroshevskii, A.A., Il'in, N.P., 1971. Physico- Geology 24, 43±46. chemical model of element separation in the differentiation of Rosing, M., 1998, pers. comm. mantle material. Trans. R. Soc.Ž. London 268, 409. Ryder G., and Mojzsis S.J., 1998. Accretion to the Earth and Wanke,È H., Dreibus, G., Jagoutz, E., 1984. Mantle chemistry and Moon around 3.85 Ga: What is the evidence. Eos. Transac- accretion history of the Earth. In: Kroner, A., Hanson, G.N., tions, U31-11, AGU Fall MeetingŽ.Ž. Abstract 79 F48 . Goodwin, A.M.Ž. Eds. , Archaean Geochem. Springer, New Safronov, V.S., 1954. On the growth of planets in the protoplane- York, pp. 1±24. tary cloud. Astron. Zh. 31, 499. Ward, W.R., Canup, R.M., 2000. Origin of the Moon's orbital Safronov, V.S., 1960. Accumulation of planets of the Earth's inclination from resonant disk interactions. Nature 403, 741± group. Vopr. Kosmog. Akad. Nauk SSSR 7, 59. 743. Safronov, V.S., 1972. Evolution of the preplanetary cloud and the Wasserburg, G.J., Papanastassiou, D.A., Tera, F., Huneke, J.C., formation of the Earth and planets. Jerusalem, Israel Program 1977. Outline of a lunar chronology. Trans. R. Soc.Ž. London for Scientific Translations, NASA Technical Tanslation; F-677. 285, 7±22. Safronov, V.S., Ruskol, E.L., 1977. The accumulation of satel- Wetherill, G.W., 1994. Provenance of the terrestrial planets. lites. In: Burns, J.A.Ž. Ed. , Planetary Satellites. The Univ. of Geochim. Cosmochim. Acta 58, 4513±4520. Arizona Press, Tucson, AZ, pp. 501±512. Whitehouse, M.J., Kamber, B.S., Moorbath, S., 1999. Age signifi- Schmidt, O.Y., 1944. Meteoritic theory of the origin of the Earth cance of U-Th-Pb zircon data from early Archaean rocks of and planets. Dokl. Akad. Nauk SSSR 45, 245. West Greenland Ð a reassessment based on combined io-mi- Schmitz, B., Peucker-Ehrenbrink, B., Lindstrom,È M., Tassinari, croprobe and imaging studies. Chem. Geol. 160, 201±224. M., 1997. Accretion rates of meteorites and cosmic dust in the Winiarsky, M., 1989. Photographic observations of the cloud in

early Ordovician. Science 278, 88±89. the neighbourhood of libration point L5 of the Earth±Moon Singer, S.F., 1968. The origin of the Moon and its geophysical system. Earth, Moon, Planets 47, 193±215. consequences. Geophys. J. R. Astron. Soc. 15, 205±226. Wood, J.A., 1977. Origin of Earth's Moon. In: Burns, J.A.Ž. Ed. , Singer, S.F., 1972. Origin of the Moon by tidal capture and some Planetary Satellites. The Univ. of Arizona Press, Tucson, AZ, geophysical consequences. The Moon 5, 206±209. pp. 513±532.