Geochemiical Aspects of Atmospheric Evolution by Charles F

1194 EVOLUTION OF THE EARTH'S ATMOSPHERE PROC. N. A. S.

DR. COMMONER: Stereospecificity must have originated very early; various possible mechanisms which might either select or generate organic compounds with right-left specificity have been proposed, but in the absence of an established mechanism it is difficult to be at all precise about the time. DR. BRITTON CHANCE (University of Pennsylvania): I understand that the key to your hypothesis is that the modern catalyst developed last. I would like to ask what type of catalyst you imagine to operate in the primitive anaerobic metabolism which is considered as the first step. It seems to me that the development of catalysts at this point would be essential. DR. COMMONER: Perhaps I might restate the point made in my paper as follows: In the modern catalyst the active site represents an arrangement of only a few amino acid residues. M\Iy proposal was that, nonspecific protein synthesis being already present, there might appear, with some reasonable probability, among the numerous proteins produced, a few with the surface configuration of amino acids required for a particular catalytic effect. It is also possible that the probability of achieving the necessary amino acid arrangement might be enhanced by coordination with a metal. Hence, catalytically active proteins might appear although a mechanism for ensuring a highly specific amino acid sequence was absent. In effect, I suggest that catalytic activity of a protein is not necessarily dependent on its total amino acid sequence, and that catalytic activity is an earlier attribute of proteins than precise amino acid sequence. DR. CLOUD: I am afraid there is not time for further discussion of this paper.

GEOCHEMIICAL ASPECTS OF ATMOSPHERIC EVOLUTION BY CHARLES F. DAVIDSON UNIVERSITY OF ST. ANDREWS, SCOTLAND One of the major problems of geochemistry relates to the composition of ancient sedimentary rocks. It can be summarized in a single simple question. Are these early strata isochemical with the sediments originally deposited, or have they sometimes been profoundly modified by additions and abstractions of chemical elements during their postdepositional history? In the field of economic geology this issue is the basis of the long-continued controversies between syngeneticists and epigeneticists on the origin of strata-bound sulfide ore-bodies. The same arguments arise about the genesis of most deposits of dolomite, many kinds of iron ore, and stratiform occurrences of barites, celestine, and other minerals. Lately, a new aspect of this old dispute has become prominent with, on one side, the isochemical school which seeks to interpret the abnormal composition of some ancient sediments in terms of their deposition under an oxygen-free atmosphere, and on the other, the actualistic school which finds a uniformitarian explanation of com- positional abnormalities in the chemical effects of intrastratal waters, proof of extraordinary atmospheres being denied. Recent discoveries of unusual potassium-rich shales in Scotland and in Arizona present fresh grounds for discussions of this kind. In the Lower Cambrian fucoid Downloaded by guest on October 2, 2021 VOL. 53, 1965 N. A. S. SYMPOSIUM: C. F. DAVIDSON 1195

beds of the North-West Highlands of Scotland, a radiometric survey by Dawson of the Geological Survey of Great Britain has recently demonstrated the existence of a potassic shale horizon 10-15-ft thick carrying 10-12 per cent K20, underlain by a dolomitic shale 25-30-ft thick with 8.5 per cent K20, these strata having a strike- length of over 100 miles. The potassium is present as minute (<0.005 mm) crystals of adularia feldspar, almost free from sodium.' Simultaneously, Granger and Raup2 of the U.S. Geological Survey have shown that a 120-ft-thick siltstone horizon of the Upper Pre-Cambrian Dripping Spring quartzite of Arizona carries con- sistently, at widely separated localities, from 10.4 to 14.6 per cent K20, between 60 and 90 per cent of the rock being formed of sodium-poor adularia (Na2O = 0.18%). Other Proterozoic shales of similar chemistry have been described by Vaes3 from the copper mines of Katanga; and the Lower Cambrian siltstones of Georgia, at one time quarried near Cartersville as a source of potassium, are apparently of a comparable character.4 Less spectacular "authigenic" concentra- tions of adularia are on record from the Beltian dolomites of Montana and from Ordovician strata in Minnesota.5 There is no volcanicity associated with these adularia shales, and in most developments the strata would be classified by a hard-rock geologist as unmetamorphosed. It is beyond belief that such highly feldspathic sediments (120 ft thick in Arizona) could have accumulated by alluvial processes-a mode of origin also ruled out by their soda content and their very fine grain size. Equally incredible is the concept that they were formed by diagenesis under a potassic sea. In each known occurrence, however, they either underlie or are interbedded with a succession of dolomites; and this distribution pattern lends weight to the hypothesis that they originated by reaction between some common sediment and potassium-rich fractionates of intrastratal brines, possibly derived from the lixiviation of overlying evaporites which have not survived. The dolomitization may be attributed to the magnesian content of the same ground waters. In the context of the present paper, the adularia shales are relevant in that they give support to the supposition that widespread horizons of abnormal "sedimentary" rocks, unparalleled among modern sediments, may originate from postdepositional metasomatic changes involving ground waters. The anactualistic facies found ini old sedimentary sequences which have been claimed as evidence of former atmospheres-in particular, uraninite-bearing pyritic conglomerates, taconite-type iron ores, and extensive developments of dolomite-all seem capable of explanation in much the same manner. The ancient stratigraphical successions within which these rocks occur include those of southern Africa, possibly dating from as early as 3,400 million years B.P.; the Krivoi Rog series of the Ukraine, ranging upwards from 2,600 million years; the Huronian formation of Ontario, deposited more than 2,100 million years ago; and the Proterozoic strata of southern Karelia, from 2,000 million years onward. In the following pages the evidence relating to each no- tional anactualistic facies is reviewed, and some observations are added on red-beds and phosphorites which are also relevant in considering atmospheric evolution. The Evidence of Banket Conglomerates.-The main geochemical argument supporting the concept of an anoxygenic Pre-Cambrian atmosphere rests on an interpretation of the manner of mineralization of the uraninite-bearing pyritic quartz- pebble conglomerates known as bankets, which are found in each of the ancient Downloaded by guest on October 2, 2021 1196 EVOLUTION OF THE EARTH'S ATMOSPHERE PROC. N. A. S.

stratigraphical sequences mentioned above and in the undated Lower Proterozoic strata of Brazil. As is well known, these bankets form the gold-uranium ores of the Witwatersrand and of Serra de Jacobina, and the uranium ores (free from re- coverable gold) of Blind River and of southern Karelia. In most instances, the mineralization is found in or close to basal pebble-beds, overlying major unconformities and bottoming thick sequences of molasse-type sediments. According to syngeneticists, the bankets are isochemical with the pebble-beds as originally deposited, and much of the pyrite and uraninite still exhibits a pristine detritat form, though all the gold has been recrystallized. Since detrital uraninite does not survive in modern alluvials, the mineralogy of the conglomerate reefs has been explained on the grounds that they were deposited before an oxygenated atmosphere came into being. Nowhere, however, has it been possible to point to a primary source of uraninite whence the detritus could have been derived. There is no need to repeat the many arguments advanced elsewhere6 to illustrate the subjective nature of the observations on which these conclusions are based; but one line of evidence does merit emphasis. A reducing atmosphere could not affect the manner in which the processes of sedimentation are controlled by the force of gravity, and in a reducing environment the commonest refractory heavy minerals of Archaean shields, such as monazite, zircon, ilmenite, and rutile, must be as readily subject to alluvial concentration as in an oxidizing one. Anactualistic conditions, therefore, do not explain why, if the commercial banket ores are heavy- mineral alluvial fractionates, the tenor of uranium is almost invariably much in excess of that of thorium, this being the antithesis of what is found not only in all modern placers but also in the bedrock of all Archaean shields which have been mass-sampled. From the analyses listed in Table 1 it will be seen that the content of zirconium and titanium in uranium-rich bankets is no more than in an average sandstone. The absence of any significant concentration of zircon, ilmenite, and rutile demonstrates unequivocally that the mineralized reefs are not fossil placer deposits. They are not metamorphosed black sands. The heavy minerals present

TABLE 1 COMPOSITION (%) OF BANKETS COMPARED WITH AVERAGE SANDSTONES TiO2 ZrO2 Fe U30s SiO2 I 0.13 0.04 1.69 0.157 89.96 II 0.25 0.05 2.51 0.05 86.56 III 0.02-0.33 n.d. 0.05-0.75 0.08 85-95 1 0.36 0.016 2.10 0.035 84.99 2 0.46 0.024 2.16 0.073 78.67 3 1.42 0.027 7.60 0.56 75.04 4 0.60 0.095 7.92 0.164 75.44 A 0.15 0.016 0.57 n.d. 87.22 B 0.50 0.025 3.1 0.0001 78.33 C 0.42 0.051 1.30 n.d. n.d. D 0.50 0.030 2.82 n.d. 73.68 I-III. Uranium-rich bankets, Witwatersrand. Quoted from Davidson, C. F., Econ. Geol., 59, 169 (1964). 1-4. Banket matrix, 20-45% quartz pebbles removed, Blind River. Quoted from Pienaar, Geol. Surv. Canada Bull., 83 (1963). A. Average of 7 samples of uranium-free arenites, Matinenda formation, Blind River. Quoted from Pienaar, op. cit. B. Average sandstone. Quoted from Green, Bull. Geol. Soc. Am., 70, 1127 (1959): SiO2 from Clarke's average sandstone. C. Average Permian-Triassic sandstone of Central Europe (86 analyses for TiO2; 104 analyses for ZrO2 and Fe). Calculated from data of Hartmann, Geochim. Cosmochim. Acta, 27, 459 (1963). D. Average sandstone of Russian platform (423 samples composited from 5815 specimens, Sinian- Recent), quoted from Ronov, Khim. Zemnoi Kory (1963), vol. 1; ZrO2 (average of 262 Russian sandstones) quoted from Katchenkov, Khim. Zemnoi Kory (1964), vol. 2. Downloaded by guest on October 2, 2021 VOL. 53, 1965 N. A. S. SYMPOSIUM: C. F. DAVIDSON 1197

(uraninite, pyrite, molybdenite, and other sulfides) are species characteristically deposited from hydrothermal (i.e., hot water) solutions. What, then, is the origin of the mineralization? This question has recently been answered7 along strictly uniformitarian lines. The occurrence of these uraniferous conglomerates close to major unconformities within or bottoming deep confined basins of Proterozoic sediments is compatible with the view that the uranium has been leached by ground waters from the overlying stratigraphical sequence, the bankets marking the mature end-phase of a prolonged series of intrastratal migrations which the mineralization has undergone. Auriferous uranium deposits have been derived from the lixiviation of acid-intermediate volcanic-pyroclastic rocks, and uranium deposits devoid of workable gold from the leaching of granitic detritus. In each case ground waters have gradually and intermittently carried a load of mineralization downward to the lowest permeable horizon; and where there was a depth-source of heat they have moved outward, principally along the open conglomerate channels, to deposit the metals toward the cooler marginal zones. In all Proterozoic occurrences deposition of the uranium preceded one or more per- iods of regional metamorphism in which the uraninite was rejuvenated with release of radiogenic lead. Many Mesozoic and Cenozoic analogues are on record, but in these the distribution of mineralization has not always reached the mature basal- conglomerate end-stage seen in the ancient bankets. The validity of this explanation can be demonstrated by reference to the mineralization of the oldest bankets known, in the Dominion Reef of South Africa.7 The Dominion Reef System comprises 200-300 ft of basal quartzite with two mineralized conglomerate bands, succeeded by upward of 2,000-3,000 ft of rhyolitic lavas, rhyolitic ashes and tuffs, felsites, and andesites. If these volcanic rocks were in- itially disposed as a confined basin and if, like many similar Cenozoic formations, they were relatively rich in gold and uranium, ideal conditions must have been created for the leaching of the metals from the higher levels and their concentration in a basal conglomerate. The age of the mineralization in the latter has been con- vincingly established by concordant whole-rock uranium:lead datings at 3,100 t 100 million years, and a whole-rock rubidium:strontium age on the lavas has been reported at 3,000 (? 4 50) million years. The close agreement between these figures supports the thesis that the uranium was extracted from the volcanic rocks soon after their eruption. Old granite-gneisses underlying the Dominion Reef System have been dated by the rubidium: strontium method at 3,200-3,400 million years; but locally they have been mobilized to form "younger granites" dated at 2,700 1 55 million years, the sequence of events apparently being comparable to the mantled doming and rejuvenation found in Finland, Zambia, the Red Sea hills, and elsewhere. The Pb206 :Pb207 age of the Dominion Reef uraninite, freed from matrix, is 2,742 million years, much less than that of the whole-rock ore but agreeing well with the age of the "younger granites," suggesting that the uraninite was metamorphosed with expulsion of radiogenic lead at this time. An analogous history can be demonstrated for the auriferous bankets in Upper Witwatersrand strata, the plumbochemistry of these rocks strongly supporting the hypothesis that the gold-uraninite-pyrite mineralization was deposited from ground waters which acquired their load of metals by leaching the overlying Ventersdorp volcanics in Ventersdorp and possibly Transvaal times. During the Bushveld Downloaded by guest on October 2, 2021 1198 EVOLUTION OF THE EARTH'S ATMOSPHERE PROC. N. A. S.

orogeny the uraninite was metamorphosed, with liberation of radiogenic lead. A fuller discussion of the genesis of these and other western bankets has been published elsewhere.7 In the two Russian fields of Krivoi Rog and southern Karelia, whole-rock lead-uranium datings of the polymineralic matrix of the bankets give ages much younger than those of the underlying basements, clearly indicating that the initial uranium mineralization was not of a detrital character. Phanerozoic equivalents of the banket formations are to be seen in the Permian Verrucano conglomerates of Central Europe, in which basal pebble-beds are locally mineralized with uraninite, pyrite, and other sulfides. Wherever the strata are somewhat metamorphosed, the ore minerals exhibit the allegedly "detrital" textures reported from Witwatersrand ores. In essence, the mode of origin of these rocks is not much different from that of the many fields of Cretaceous-Eocene uraniferous lignites from the Dakotas and Wyoming, where the uranium has been leached from overlying tuffs and lavas, from granitic detritus, or both, and sorbed by coal seams from descending solutions. Had the coals not been present to inter- cept this uranium, it could under favorable hydrological-structural conditions have found its way downward to a basal conglomerate. Deposits such as the uranium- rich arkosic sands of Eocene age in the Shirley Basin of Wyoming may represent a young phase in a yet unfinished sequence of intrastratal migrations of which the ultimate result, not everywhere reached, will be a basal banket. A basal conglomerate deposit of Eocene age, which rests on Pre-Cambrian granites and schists, was worked from 1954 to 1964 in the minor Copper M\'ountain district (Owl Creek Mountains) of Central Wyoming.8 The hypotheses that the bankets are fossil placer deposits and that a reducing atmosphere existed at the time of their formation have become linked in mutual support against a wind of criticism which neither of them can withstand alone. If, therefore, it is accepted that the mineralization could have been emplaced by a hydrothermal ground-water mechanism similar to processes operating in the Phanerozoic eras, the concept that an anoxygenic atmosphere existed in Proterozoic times is reduced to the level of a problematical postulate for which firm geochemical evidence has yet to be found. Observations on Iron Ores. The origin of the immense developments of stratiform iron ores in Pre-Cambrian rocks, estimated by N. AM. Strakhov to amount to 3.15 trillion tons, has never been satisfactorily explained. Great ore deposits mostly similar to the banded taconites of the Lake Superior region are found in amost all ancient stratified successions, ranging from the Fig Tree Series of Swaziland dated at >3,400 t 300 million years to some Russian fields of magnetite-quartzites in northern Kazakhstan, Sayan, and M\Ialyi Khingan, immediately underlying fossili- ferous Cambrian strata and classified as uppermost Riphaean. The geochemical differences between these ancient siliceous iron formations and the minette-type sedimentary ores deposited in Phanerozoic times have led many geologists to attribute the peculiarities of the former rocks to one or more anactualistic causes, such as intensified volcanicity in Pre-Cambrian eras, a greater proportion of acid juvenile waters of volcanic origin when the oceans were supposedly smaller than they are today, an increase in the content of carbon dioxide in the contemporary atmosphere, and the existence of an atmosphere which was devoid of free oxygen. Claims that these taconite and allied deposits are attributable to an abnormally Downloaded by guest on October 2, 2021 VOL. 53, 1965 N. A. S. SYMPOSIUM: C. F. DAVIDSON 1199

high incidence of volcanicity can readily be dismissed, since the measured stratigraphical sequences of Proterozoic rocks, studied in greatest detail in the U.S.S.R.,9 display no greater proportion of lavas and pyroclastics than is found in younger strata. Claims invoking an abnormal atmosphere must be examined against a geochronological background. Potassium: argon datings on micaceous schists interbedded with magnetite-hematite-quartzites in eastern Sayan give an age of 600 millionyears, and inwestern Sayan 933 millionyears.10 Like manychertyiron formations of much earlier date these Upper Riphaean deposits are characterized by a high tenor of manganese. Roughly contemporaneous with these strata and with the magnetite-quartzites of Malyi Khingan in Siberia, one finds oolitic chamosite ores of minette type in the Chinese province of Honan, overlying a glauconitic sandstone dated at 1,162 million years; and also, in northern China, extensive oolitic hematite ores in quartzite, older than 880 million years, but younger than 1,040 million years (K-Ar dates on glauconite).9 The occurrence of siliceous iron formations in rocks of about the same age as strata which yield minette ores clearly con- tradicts the hypothesis that the genesis of the former is due to an anactualistic atmospheric control. It is noteworthy that the bedded iron ores in the relatively unmetamorphosed Pretoria Series (Transvaal System) of South Africa, which is at least 2,000 million years old, are mainly hematite-chamosite or magnetite-chamosite oblites akin to the Ordovician Wabana ores of Newfoundland and to other Phanerozoic formations. Since many taconites are younger than the Pretoria Series, the concept that they provide evidence of a reducing atmosphere must be invalid. A feature of the vast literature on Pre-Cambrian iron formations is the agreement among most authorities that the iron compounds were deposited contemporaneously with the sediments in which they are now found; but for one particular group of Proterozoic ores this tenet has been questioned. The great bedded, tabular, or lenticular deposits of siderite occurring among the Lower Riphaean (ca. 1,400- 1,700 million years) dolomites and limestones of Asia9 have long been the focus of argument between syngeneticists and epigeneticists. Typical of these rocks are the large ore fields at several stratigraphic levels (>1,263 million years, by K-Ar dating on overlying Middle Riphaean glauconite) in the Bakal'sk district of the southwestern Urals," where an epigenetic hypothesis is supported by the observations that the siderite pseudomorphs Collenia algae and other sedimentary structures in the limestone, that there are relics of unaltered dolomite within ore, and that the ores are generally similar to the famous metasomatic siderite-ankerite mineralization of Alpine age in the Erzberg of Austria. Like the siliceous iron formations, these metasomatic ores are usually high in manganese and low in phosphorus. The nature and origin of the metasomes which gave rise to these iron carbonate deposits is unknown; but it is a reasonable working hypothesis that the ores were formed by reaction between limestones or dolomites and hot intrastratal brines rich in iron chlorides.12 Evaporite deposits have not survived in Pre-Cambrian formations because of the metamorphism to which these strata have been subjected; and, by analogy with Phanerozoic events such as the formation of ground-water brines to a depth of 2-3 miles below the Cambrian evaporites of the Russian platform, it may be suggested that the Pre-Cambrian salt deposits were removed to Downloaded by guest on October 2, 2021 1200 EVOLUTION OF THE EARTH'S ATMOSPHERE PROC. N. A. S.

form deep intrastratal chloride solutions capable, under orogenic conditions with a high geothermal gradient, of giving rise to ferruginous brines by reaction with basic igneous rocks or with iron ores of sedimentary origin. In a siliceous environment extensive redistribution of syngenetic mineralization and deposition of epigenetic bedded ores could occur in this manner. Many so-called "exhalative" iron ores typified by the Lahn-Dill deposits of Germany can be regarded as Phanero- zoic analogues if it is accepted that they have not originated contemporaneously with the lavas, as generally supposed, but have been derived from them as a result of their later lixiviation by hot saline brines. The keratophyric-spilitic character of basic igneous rocks associated with these bedded iron ores is in keeping with this explanation. Markovsky'3 notes that "almost all the Lower Proterozoic iron ore deposits of the U.S.S.R. occur in a highly typical close association of iron quartzites with effusive assemblages of spilite-keratophyre." The possibility therefore exists that the rocks referred to as iron formations range from sedimentary to metasomatic in character, and that their predominance in Pre- Cambrian strata can be attributed in part to the greater opportunities for intrastratal collection and redistribution of iron in these old metamorphic sequences. The hypothesis that they originated by anactualistic weathering in an atmosphere devoid of oxygen'4 is gainsaid by quantitative considerations. The ratio Fe2O3: FeO in the average igneous rock has been calculated to be 0.81 (Clarke and Washington), 0.83 (Vogt), and 0.82 (Grout, for the Canadian shield). The ratio Fe2O3:FeO in magnetite is 2.22, and in the primary "unoxidized" magnetite-quartzites of the Kursk Magnetic Anomaly it is 2.71. The mass of Pre-Cambrian iron formations, on Strakhov's calculations, is 3.5 X 1012 tons, containing, say, 2 X 1012 tons of magnetite (including hematite calculated as magnetite). To derive this magnetite by weathering processes from an equivalent amount of iron compounds present in "average igneous rock" needs the addition of about 2 X 1011 tons of oxygen. To form the almost ubiquitous hematite still more is necessary. Where could this immense amount of oxygen have come from if not, directly or indirectly, from an oxygenated atmosphere? The Relevance of Red-Beds. Desert sandstones and marls stained red or yellow with hematite and goethite are more abundant in Phanerozoic than in Proterozoic formations; and the supposed absence of such red-beds from strata more than a milliard years old has been interpreted in some quarters as an indication that continental sediments older than 1,000 million years were deposited under reducing conditions. No valid basis for this claim exists. The age of the most ancient red-beds varies greatly from one region to another. In the Balkan and Ukrainian shields, K-Ar datings by Polkanov and Gerling on authigenic micas in marls have given values for "the oldest reds" of 1,185 million years in Sweden, 1,280-1,310 million years in Finland, and 1,345-1,485 million years in the Ukraine.'5 These ages are in good agreement with datings on transgressive rapakivi granites. Red-beds in the Uyansk series and the Uchursk series of the Siberian platform (the latter with gypsum and salt pseudomorphs) have been dated at 1,600 and 1,500 million years, respectively, by K-Ar determinations on igneous rocks and glauconites.9 In British Guiana the Roraima red-bed formation, interpreted a decade ago as "possibly Triassic continental," is now known to be more than 1,675 ± 100 million years old; red-beds have been recorded from the Lorrain Downloaded by guest on October 2, 2021 VOL. 53, 1965 N. A. S. SYMPOSIUM: C. F. DAVIDSON 1201

formation of the Huronian in Ontario, older than 2,000 million years;25 and it has been claimed that red-bed strata occur in the much older Muruwa formation of Guiana, dated at around 2,500 million years. 16 Sandstones and marls whose continental character is attested by dune bedding, mud cracks, and rain pits abound in most major sequences of Proterozoic strata, and in any one succession the red-beds lie amid the younger horizons. But since red-beds and continental grey-green beds are often demonstrably contemporaneous, clearly the color distinction cannot be attributed solely to atmospheric causes. A much simpler explanation for the stratigraphical distribution pattern can readily be found in the circumstance that most Proterozoic sediments other than the youngest formations are at least in the green schist facies of metamorphism where, in the presence of magnesian ground waters, hematite and goethite pigments dis- appear on the formation of chlorite. There is, therefore, good reason to suppose that red-beds were once much more abundant throughout Pre-Cambrian strata than they are today, and that deductions about the Proterozoic atmosphere based on their present distribution are probably fallacious. Deductions from Dolomite.-The view is widely held in Russian literature that the dominance of dolomites among carbonate sediments of Proterozoic age is attributable to the existence of a coeval heavy atmosphere greatly enriched in carbon dioxide, this concept apparently stemming from experimental observations that a high pressure of carbon dioxide is needed before dolomite will precipitate from water saturated with calcium and magnesium carbonates. Western authorities, on the other hand, tend to agree that "most stratified dolomites are replaced limestones, formed under a variety of conditions ranging from penecontemporaneity with sedimentation to late diagenesis through the action of either connate magnesium-bearing brines or meteoric waters."'17 On the latter explanation, the prepon- derance of dolomites among ancient sediments is merely a reflection of their age, by virtue of which they have become more subject to intrastratal metasomatism by magnesian solutions. Among the many arguments which can be marshalled against the mutually dependent hypotheses of primary dolomite formation and a carbon dioxide atmosphere, the following merit consideration. First, in most ancient fields of carbonate rocks relics of primary limestone per- sist. For example, in the Transvaal dolomites of South Africa and the Lomagundi dolomites of Rhodesia (>2,000-2,200 million years), developments of magnesium- poor limestone, with occasional seams of dolomite and chert, are found in lenses up to more than a hundred feet thick and several miles in length. In the Huronian of Ontario (>2,100 million year), the Bruce limestone and the Espanola limestone are dolomitized only in some thin but persistent bands; and in the Jatulian dolomites of Karelia (>1,700 million year) there are many irregular limestone horizons up to 6 ft thick (averaging in seven analyses 55% CaO, 0.4% MgO).18 Generally, impervious marls and limestones associated with marls are poor in magnesium, suggesting that permeability to magnesian metasomes was a salient factor in dolomite formation. Attempts to explain the presence of limestone rather than dolomite as being due to variations in salinity at the time of deposition' are not at all convincing, particu- larly as the limestone developments often transgress stratigraphical horizons; and this explanation is apparently negatived by studies on the boron content of Downloaded by guest on October 2, 2021 1202 EVOLUTION OF THE EARTH'S ATMOSPHERE PROC. N. A. S.

Beltian illites,19 samples from limestone and dolomite formations alike indicating a palaeosalinity comparable to that of the present-day oceans. Second, the incidence of dolomitization in a group of Proterozoic rocks is controlled by the same lithological characteristics as it is in younger formations. Just as coarse-grained limestones of Phanerozoic age tend to be less susceptible to magnesium metasomatism than fine-grained, so in 50 chemical analyses of Jatulian dolo- mitesl8 the CaO: MgO ratio (1.40 in pure dolomite) falls steadily from a mean value of 1.56 in the coarse-grained rocks to 1.35 in the fine-grained. Again, in Jatulian strata stromatolite bioherms are formed of dolomite but the country-rock of these organic reefs is often limestone, reflecting the similar preferential development of dolomitization commonly found in coralline algae of more recent formations. In some Beltian bioherms, however, one finds the reverse situation of limestone stromatolites in a dolomite matrix,20 a difference perhaps attributable to variations in permeability and texture. These observations suggest that the abundance of dolomite among Proterozoic strata is due to the mineral being of postlithification metasomatic origin, thus afford- ing no support for the postulate that a heavy carbon dioxide atmosphere survived in Upper Proterozoic and even into Palaeozoic times. Perhaps it is significant that in the latest monograph on the Proterozoic of the U.S.S.R.,9 all mention of this hypo- thetical heavy atmosphere has been omitted. Inferences from Phosphorite Deposits. Weighty circumstantial evidence on the tenor of carbon dioxide in ancient atmospheres can be deduced from the stratigraphical distribution of sedimentary phosphorites.2' On the well-substantiated theories of Kazakov, the mode of formation of these deposits is as follows. Phos- phatic phytoplankton which develop by photosynthesis in the uppermost 50 meters of the ocean sink after death into deeper waters and decay, so releasing carbon dioxide into solution and permitting the phosphate to be dissolved. This takes place at depths below 200 meters, where the partial pressure of carbon dioxide is up to 12 X 10-4 atmospheres. Marine phosphorites are precipitated when an upsurge of these cold deep waters saturated with carbon dioxide and phosphate ascends into the region of the continental shelf, the rise in temperature and a concomitant fall in partial pressure of carbon dioxide to about 3 X 10-4 atmospheres leading to a precipitation of calcium phosphate, in waters saturated with calcium carbonate, at depths between 50 and 200 meters. The control of precipitation of phosphorite by partial pressure of carbon dioxide is thus extremely delicate; and in the existence of ancient phosphorites within limestone or dolomite environments there is, therefore, an indication that throughout the stratigraphical range of these rocks, the content of carbon dioxide in the near-surface waters of the sea (and in the contemporary atmosphere) was never much higher than it is today. On uniformitarian reasoning, the occurrence of ancient phosphorites in marine strata also demonstrates the contemporary photosynthetic development of phytoplankton. Information on the distribution of phosphorites in relatively old rocks is to be found mainly in Russian and Chinese literature.9 The suggestion that a heavy atmosphere persisted as late as Palaeozoic times is refuted by the presence of im- portant deposits of phosphorite in sediments of Cambrian age, in the Karatau and in Central Kazakhstan with the U.S.S.R., and in many economic occurrences marking the Lower Cambrian marine transgressions on the Chinese platforms. Downloaded by guest on October 2, 2021 VOL. 53, 1965 N. A. S. SYMPOSIUM: C. F. DAVIDSON 1203

Quite extensive developments in Sinian and Riphaean strata are recorded from western Hunan, northern Kwiangsi, and elsewhere in China,22 as well as from upward of a dozen regions in the Soviet Union9 including the Lower Riphaean trans- gression in the Yenisei region dated at 1,400 million years. Low-grade phosphatic shales of great thickness (300-800 meters) occur in the uppermost Proterozoic of the Lake Baikal region; and in the dolomites of the upper Krivoi Rog series in the Ukraine a thin phosphatic horizon with uraniferous apatite (about 1% U) accepted as being of sedimentary origin has given, in four assays, a Pb2M:Pb20 age range of 1,780-1,860 million years. Since part of the uranium may have been introduced from ground waters, this must be regarded as a minimum age. Apparently all of these strata are marine (many of them contain glauconite), and they show no lithological resemblance to the Eocene lacustrine phosphorites of problematical origin recently reported from Wyoming and Utah.23 Deposits of phosphate rock reputedly of "Archaean" age include apatite-bearing marbles, potentially of commercial grade, developed at several localities in the pre- Sinian paragneiss complex of the South China platform.22 Some of these metamorphic assemblages may be of a relatively late date; but the commercial deposits of Slyudyanka in eastern Sayan, in the Baikal region of the U.S.S.R., are un- doubtedly very old. They comprise six or seven apatite-bearing horizons up to 10-12 meters in thickness, the richest carrying 4-5%, P205 in the form of fluor- apatitic diopside-quartzites and diopside-marbles, interbedded with other un- doubted paragneisses; and they underlie an aggregate thickness of Proterozoic and Eocambrian sediments estimated at not less than 10 km and possibly up to 20 km thick, containing six major unconformities. Potassium: argon dating has been com- plicated by a superimposed Palaeozoic orogeny, but values of 2,330-2,500 million years obtained on pyroxenes and amphiboles of ortho-amphibolites perhaps date the Archaean metamorphism. It seems a reasonable deduction that phosphatic phytoplankton were in existence far back in the Pre-Cambrian. Conclusion.-These observations based on the stratigraphical record contradict the belief that a primaeval reducing atmosphere persisted for much of Pre-Cambrian time. Such an atmosphere may well have been lost in the protogeological period, before the earliest known sediments were formed. On the uniformitarian reasoning that the events of the past have been governed by the chemical and physical laws of the present, progressive changes of atmospheric composition must of course be envisaged; and, in particular, a gradual build-up of oxygen, from photosynthetic and photochemical reactions involving carbon dioxide and water, cannot be dis- puted. The existence of immense hematite deposits in the form of banded iron- stones, jaspilites, and ferruginous shales as far back as the Fig Tree series of Swazi- land (3,400 1 300 million years), in a sedimentary succession which also bears the first problematical vestiges of life,24 is however compatible with the presence of free oxygen in surface or intrastratal waters at a very early date; and there seems to be no evidence at all to favor the view that the metabolism of the stromatolites and oncolites found in the carbonate rocks of the Bulawayan (?2,700 million years), Ventersdorp (?2,300 million years), and Transvaal (>2,000 million years) sys- tems, and in nearly all later carbonate sequences of Pre-Cambrian age, was any different from that of their Palaeozoic and modern analogues. The distribution of Downloaded by guest on October 2, 2021 1204 EVOLUTION OF THE EARTH'S ATMOSPHERE PROC. N. A. S.

phosphorites suggests that by about 2,000 million years ago the carbon dioxide content of the oceans and atmosphere was not much more than it is today. More information on the chemistry, lithology, and geochronology of Pre-Cam- brian sedimentary rocks has been published in the last decade than ever before, and within this framework there is clearly a need for a closer alignment between hypotheses on atmospheric evolution and the evidence presented by the only witnesses, the rocks themselves. However, as Charles Lyell observed in his Princi- ples of Geology, "The great contrast in the aspect of the older and newer rocks ... appeared formerly one of the strongest grounds for presuming that the causes to which they owed their origin were perfectly dissimilar from those now in operation- but this incongruity may be the result of subsequent modification." Those geologists who advocate Pre-Cambrian departures from Lyellian uniformitarianism must beware lest their evidence be explicable in terms of metasomatic processes which are equally manifest in formations of Phanerozoic age. 1 Bowie, S. H. U., Sum. Progr. Geol. Surv. Gt. Brit. 1963, (1964), p. 75. 2 Granger, H. C., and R. B. Raup, U.S. Geol. Surv. Bull., 1168 (1964). 3Vaes, J. F., Ann. Musge Roy. I'Afrique Centrale, Ser. Geol., 8, no. 43 (1962). 4Shearer, H. K., Georgia Geol. Surv. Bull., 34 (1918). 6 Gruner, J. W., and G. A. Thiel, Am. Mineralogist, 22, 842 (1937). 6Davidson, C. F., Mining Mag. (London), 88, 73; 92, 152; 100, 92; 101, 30; 102, 84, 149, 222; 105, 88; 107, 158 (1953-1964); Davidson, C. F., Econ. Geol., 52, 668; 53, 887; 54, 1316; 59, 168 (1957-1964). 7Davidson, C. F., Trans. Inst. Mining Met., 74, 319 (1965). 8 Guilinger, R. R., personal communication, May 1964. 9Keller, B. M., ed., Stratigraphy of the U.S.S.R.: Upper Precambrian (Moscow: Gosgeol- tekhbidat, 1963) (Russian). 10 Klyarovsky, V. M., Trans. 11th Session Commission .... on Absolute Age, 296 (1963) (Rus- sian). 11 Burgelya, N. K., in Reviews of the Metallogeny of Sedimentary and Sedimentary-Metamorphic Rocks (Moscow: Gosgeoltekhizdat, 1962) (Russian). 12 Davidson, C. F., Mining Mag. (London), 110, 176, 244 (1964). 13 Markovsky, A. P., Geological Structure of the U.S.S.R. (Moscow: Gosgeoltekhizdat, 1958), vol. 1, p. 139 (Russian). 14 Lepp, H. S., and S. S. Goldich, Econ. Geol., 59, 1025 (1964). 15Polkanov, A. A., and E. K. Gerling, in Geology and Geochronology of the Precambrian (Moscow: Nauka, 1964), p. 181 (Russian). I' Cannon, R. T., Nature, 205, 586 (1965). 17Carozzi, A. V., Microscopic Sedimentary Petrography (New York: John Wiley and Sons, 1960), p. 284. 18 Sokolov, V. A., Geology and Lithology of the Middle Proterozoic Carbonate Rocks of Karelia (Moscow: Academy of Sciences, 1963) (Russian). 19 Reynolds, R. C., Geochim. Cosmochim. Acta, 29, 1 (1965). 20 Ross, C. P., U.S. Geol. Surv. Profess. Papers, 346 (1963). 21 Davidson, C. F., Mining Mag. (London), 109, 205 (1963). 22 Tupitzyn, N. V., ed., Elements of the Tectonics of China (Moscow: Gosgeoltekhizdat, 1962) (Russian trans. from Chinese). 23 Love, J. D., U.S. Geol. Surv. Profess. Papers, 474-E (1964). 24 Ramsay, J. G., Trans. Proc. Geol. Soc. S. Africa, 66, in press. 25Frarey, M. J., Geol. Surv. Canada Map, 32 (1962), (Bruce Mines, Ontario). Downloaded by guest on October 2, 2021 VOL. 53, 1965 N. A. S. SYMPOSIUM: A. G. FISCHER 1205

(Discussion of Dr. Davidson's paper) DR. R. B. HARGRAVES (Princeton University): I would like to enter an appeal on behalf of the opposition, comprising most of the geologists who have worked in the Witwatersrand. First, I would say that Professor Davidson provides us with two extreme alter- natives; namely, either the gold and uraninite is detrital or it must be hydrother- mally produced. I think it is fair to say that the fundamental fact about these gold and uranium deposits is the incredibly detailed fidelity with which they follow sedimentary guides. It seems to me the rocks in which these deposits occur are gravels. Gravels today are known to contain much of the gold in this country. I think it makes it easier if one assumes that at least some,if not all, of the gold and uraninite was already present in these gravel deposits. Then, if the gold and if the uranium could only be syngenetic under a reducing atmosphere, I do not think that there is any valid alterna- tive to that hypothesis. DR. DAVIDSON: I can only say, sir, that there is no known geological environment from which detrital gold, uraninite, and pyrite devoid of other heavy minerals can be obtained as alluvial deposits. If Dr. Hargraves can suggest how such deposits could form syngenetically in the Dominion Reef, the southeast of Finland, Blind River, and so forth-if he can suggest any geological environments from which such associations can be derived syngenetically, then he will have made a very great step forward. DR. CLOUD: I am sure we could have a very lively discussion on these matters, but we are short of time.

FOSSILS, EARLY LIFE, AND ATMOSPHERIC HISTORY BY ALFRED G. FISCHER

DEPARTMENT OF GEOLOGY, PRINCETON UNIVERSITY The Fossil Record.-From the paleontological standpoint, earth history is sharply divided into two parts-the vast Pre-Cambrian, with a sparse fossil record, and the Phanerozoic, beginning with the Cambrian period some 600 million years ago, with its rich documentation of life (Fig. 1). The only fossils which have been found widespread and abundant in the Pre- Cambrian are stromatolites: head-shaped or branched, laminated structures, generally composed of calcite or dolomite, but in some cases siliceous, and generally attributed to the Cyanophyta (also known as Myxophyta or blue-green "algae"). They appear to be present in most Pre-Cambrian limestones and include the oldest fossils known to date, from the African Bulawayan,' about 2.7 billion years old. Like most fossil stromatolites, those from the Bulawayan show no cellular detail; their interpretation as fossils rests on three lines of evidence: we know of no inorganic processes which form such structures; cyanophyte colonies produce structures of this type today;2, 14 and well-preserved stromatolites of various ages have yielded microscopic cellular detail which confirms their organic origin. Thq Downloaded by guest on October 2, 2021