Debate about ironstone: Has solute supply been surficial weathering, hydrothermal convection, or exhalation of deep fluids?

Michael M. Kimberley
Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University
Raleigh, NC 27695-8208, USA

NOMENCLATURE

Chemical sedimentary rocks are precipitates from a water body and/or diagenetic alterations of those precipitates. The rock types include limestone, dolostone, evaporites, chert, cherty ironstone (typically laminated), noncherty ironstone (typically oolitic), phosphorite, massive sulfide (where stratiform), and manganostone. Manganostone and ironstone are chemical sedimentary rocks which contain more than 15% Mn or 15% Fe, respectively (Kimberley, 1989a; Biihn et al., 1992). In a two-fold division of sedimentary rocks into chemical versus clastic types, the foregoing list may be expanded to include some paleosols such as bauxite and nickeliferous laterite. Ooids are characteristic of both noncherty and shallow-water cherty iron formations (Dimroth and Chauvel, 1973). Rock or sediment which contains a substantial proportion of ooids is termed oolite.

Discussion of chemical sedimentary rocks requires a clear distinction between lithological and stratigraphic characteristics. However, undifferentiated terms such as 'the cherts' commonly refer alternatively to varieties of chert within a single bed and to several beds of homogeneous chert. This confusion may be avoided by consistent usage of stratigraphic terms such as chert bed.  A stratigraphic unit which is dominantly composed of iron-rich chemical sedimentary rock (ironstone) is here called an iron formation, independent of chert content, laminated (banded) structure, or oolitic texture.

UNCERTAIN SOLUTE SOURCES FOR CHEMICAL SEDIMENTARY ORES

At present, chemical sedimentologists and ore-deposit geologists are sharply divided on the issue of solute supply by surficial weathering versus exhalation of deep fluids. Surficial weathering models invoke solute supply by soil formation and by chemical alteration of surficial marine sediment. Exhalation models typically invoke either deep-source magmatic devolatilization or shallower-source fluids such as seawater which has descended into the crust and ascended again due to a combination of convection, tectonic pumping, and/or gravity-driven flow. One purpose of this paper is to emphasize the differences among the weathering vs. shallow exhalation vs. deep exhalation schools of thought and explore the implications of adhering to any one of these three. In other words, the intent is to stimulate debate rather than pretend to resolve the issue cleanly. Simple solutions are presented, as in any debate, but these are intended primarily to guide data collection.

Resolution of the surficial vs. exhalative debate has broad implications for both mineral exploration and deduction of Earth history. Surficial proponents invoke a special climatic-topographic regime for each metalliferous stratigraphic unit. Deep-source proponents invoke a special tectonomagmatic regime. To appreciate the present polarization of concepts, one need only compare a pair of papers about comparable manganese formations within the August 1992 issue of Economic Geology, i.e. Biihn et al. vs. Frakes and Bolton.

Many low-temperature geochemists have published genetic models for iron formations because Precambrian examples are too large to ignore and because iron is the most abundant element in planet Earth (Anderson, 1989). The ongoing surficial vs. exhalative debate is proceeding like previous geological controversies, e.g. the plutonism-neptunism, uniformitarianism-catastrophism, glaciation-flood, and drifting continent-fixist continent controversies.

All sides of the surficial vs. exhalative debate about chemical sedimentary ores agree that iron formations record Earth's evolution. The proportion of cherty iron formations to other types of stratigraphic units has clearly declined through Earth history (e.g. fig. 1 of Kimberley, 1983). Moreover, there has been an apparent evolution of other types of ore (Hutchinson, 1992), and of rock types in general (Ronov, 1992). Chemical sedimentary ores have evolved more markedly than any other class of rock but explanation of that evolution is contentious. Surficial advocates favour evolution of Earth's surficial environment, e.g. an increasing oxidation state of the atmosphere, whereas exhalative advocates favour tectonomagmatic evolution related to cooling of our decreasingly radioactive planet.

Neither side of the surficial vs. exhalative debate has yet made convincing use of the most powerful approach to deduce rock genesis: actualism. Although the present can be an excellent key to the past, one must first find an appropriate modern process. Unfortunately, planetary evolution has diminished chemical sedimentation of metals on continental shelves so severely that the search for modern analogues has proven difficult for shelf deposits such as iron formations. To find an appropriate modern analogue, one may use the palaeoenvironment of an ancient deposit as a guide to select the modern environment which deserves closest scrutiny. This method is phrased in the following title.

THE PAST IS THE KEY TO THE PRESENT

Modern analogues of chemical sedimentary ores are scarce and difficult to locate. Nonetheless, their discovery is vital because sedimentary rocks inherently preserve only a small proportion of all the information which is necessary to understand sedimentary processes. To understand sedimentation thoroughly, one must observe it happening.

Being the most voluminous chemical sedimentary ores, iron formations have received the most palaeoenvironmental study. Cherty laminated (banded) iron formations apparently have accumulated in all marine environments, from turbidite-accumulating depths (Barrett and Fralick, 1985) to stromatolitic shallow shelves (Dimroth and Chauvel, 1973), but the great bulk of iron formations clearly have accumulated on continental shelves (Kimberley, 1978). Precambrian cherty iron formations were so areally extensive that if a good modern analogue were to exist, it surely would have been discovered by now. In contrast, noncherty oolitic iron formations have been much less extensive, more prolific through the Phanerozoic, more restricted to shallow water, and more clearly isolated from volcanic influences (Kimberley, 1989b).

Modern nonvolcanic shelves are either tectonically quiescent, as around most of the Atlantic, or tectonically active, as along the northern coast of South America where a broad transform fault separates the South American plate from the Caribbean. Tectonically quiescent Atlantic shelves lie closer to major research centres than do tectonically active shelves and have therefore been studied more thoroughly. Interesting discoveries are correspondingly more probable on tectonically active shelves and these areas deserve closest scrutiny, independent of one's preferences in the weathering vs. exhalation debate.

A prime feature of tectonically active shelves is diagenetic-metamorphic exhalation of aqueous and carbonaceous fluids. Coastal areas with broad transform fault zones are particularly exhalative, e.g. northeastern Venezuela. The Venezuelan island of Margarita exhibits ultramafic rocks which have become impregnated with exhalative carbonaceous volatiles. These isotopically identifiable volatiles have impregnated serpentinite bodies with a stockwork of magnesite veins and calcareous rocks with stockwork calcite (Abu Jaber and Kimberley, 1992a,b). Continuous exhalation of deep-source methane presently occurs here under a few metres of Caribbean seawater, half way between Punta Brasil and Punta La Horca on the northwestern coast of Cubagua Island. These shallow exhalative vents are anomalously rich in life forms, as are deep-sea vents elsewhere.

A search for iron-rich sediment on a modern shelf is enhanced by familiarity with the characteristic textures of ancient ironstone. Phanerozoic ironstone provides a particularly important clue in the internal structure of ancient ironstone ooids. This structure is similar to that of calcareous ooids and the shared morphology reveals that ooids must have formed at the sediment-water interface, given that concentric ooid growth has occurred without interference from adjacent grains. Any modern ferriferous ooids therefore must form at the very surface of the seafloor and be accessible to even a simple grab sampler.

The foregoing logic has inspired two decades of searching for ferriferous ooids on the VenezuelanColombian continental shelf, using inexpensive grab samplers from 5-m long fishing boats. Eventual discovery of ferriferous ooids has partially demonstrated that the past is indeed the key to the present but mostly has demonstrated the necessity of persistent plodding. For example, the initial discovery of a few ooids could not be pursued satisfactorily until a new type of grab sampler had been invented to handle dense sediment (Kimberley, 1990). Once collected, the silicate ooids have proven to be so finely crystalline that it has been necessary to learn transmission electron microscopy to study them properly.

The first thing that a geologist learns about sampling seafloor sediment is that it is inherently less efficient than sampling rock outcrop. Determination of offshore sample location was particularly time-consuming until the Global Positioning System became available. A remaining problem is the difficulty of penetrating the seafloor more than a few centimetres during reconnaissance grab sampling. This is particularly troublesome in the search for metals because the uppermost sediment typically is oxidized whereas most ancient sedimentary ore is chemically reduced. Deduction of how reducing diagenesis will affect observed seafloor sediment is not straightforward (Berner, 1980).

MALA PASCUA FERRIFEROUS OOIDS

The only known modern marine occurrence of iron silicate ooids extends from Cape Mala Pascua to El Fraile Point, Venezuela (Fig. 1). Ooids here are intermixed with a comparable proportion of similarly-sized peloids which also are composed of nearly pure iron silicate. If the past truly is the key to the present, the ooids at Mala Pascua merit preferential study because nearly all voluminous Phanerozoic iron deposits are at least partially oolitic. Odin et al. (1988a, p. 30) note that, 'The interpretation of the oolitic structure is still much debated. Although ferriferous ooids rarely constitute the major portion of the rock, they are highly conspicuous. ... the selective study of the oolitic grains is justified because the understanding of the environment of genesis of the whole formation largely depends on the interpretation of these concentric structures.'

Thin sections of the Mala Pascua ooids reveal extremely delicate oolitic layering (Fig. 2). As in typical ironstone, the oolitic layering is fully concentric and constitutes most or all of each ooid. A nucleus is apparent in less than half of the thin-section slices through the ooids. As in ancient ironstone, ooid nuclei locally consist of a calcareous shell fragment, a terrigenous grain (usually quartz), or a fragment of a previously-formed ooid. Layering is best observed under cross-polarized light or with a scanning-electron microscope (SEM). Thorough study of these grains will require a combined transmission and analytical electron microscope (TEM/AEM).

Mala Pascua ooid layers are almost entirely composed of authigenic iron-rich silicate. Ooids sampled .to date have a micron-thick oxidized rim which includes some ferric hydroxide despite being dark green in hand sample. Very few ooids have any ferric-hydroxide layers well within their outer margin. The sub-rim layers in all ooids are composed of paler-green ferriferous silicate. The dominant X-ray diffraction (XRD) peak of this silicate is at 0.715 nm (7 Å) which is characteristic of the kaolinite-serpentine family. It is unlikely that the ooids also contain X-ray-amorphous nontronite but nontronite XRD peaks appear after heating of the ooids above 150°C within a pressure vessel.

Minor colour variation through the ooids is suspected to record inhomogeneity in the composition and oxidation state of the ferriferous silicate but this hypothetical inhomogeneity will only be revealed by TEM/ AEM of individual ooids. Existing samples represent just the upper 10 cm of seafloor sediment and deeper samples are needed to determine if the ratio.of Fe²⁺/ Fe³⁺ , in ooids varies with depth beneath oxidizing seawater.

Detrital chlorite is abundant within the green mud which lies on all sides except shoreward from the oolitic area at Mala Pascua. Chlorite also is a common mineral in the minor, clay-sized fraction of the oolitic sediment. However, no chlorite has been detected within the ooids themselves. As expected, the metamorphic chlorite exhibits sharp XRD peaks which coincide with some of the broad peaks of the oolitic serpentine-family (1 : 1 trioctahedral) mineral.

Naming of the authigenic serpentine- family mineral is not a prime concern of this paper but a brief comment is made here to avoid potential confusion. The mineral is here called either odinite or odinite-endmember berthierine, despite a suggestion by Odin et al. (1988a) that berthierine is absent from the modern seafloor. This issue is largely semantic. Four decades ago, all authigenic silicates of chlorite-like composition in ironstone were called chamosite, whether truly chlorite or actually serpentine-like, independent of oxidation state (Brindley and Youell, 1953). For the past decade, serpentine-like (1:1) phyllosilicates generally have been differentiated from iron-rich chlorite (chamosite) by the name berthierine (Brindley, 1982; Van Houten and Puruker, 1984). Both Mg-poor and Mg-rich 1:1 phyllosilicates have been called berthierine (e.g. Horvath and Gault, 1990). However, Bailey (1988, p. 238) recently has suggested that the iron-rich 1:1 phyllosilicates be subdivided into a magnesium-rich, ferric-iron-rich variety which never occurs as ooids, to be called odinite, and Mg-Fe³⁺- poor berthierine.

Bailey (1988) contends that odinite is compositionally very different from berthierine but he unfortunately relied entirely upon the berthierine analyses of Brindley (1982). Brindley (1982) is misleading because he arbitrarily eliminated all alkalis from his bulk analyses and reduced other components in some of them. In contrast, it has long been known from microprobe work that alkalis occur in berthierine (Kimberley, 1974; Maynard, 1986). Moreover, all known berthierine-rich ironstone contains both alkalis and ferric iron (Kimberley, 1989b). Bailey (1988) is misleading because he implies that berthierine exhibits negligible compositional variation. If the past is the key to the present, Mala Pascua ooids represent an extension of the berthierine range to an Fe³⁺- Mg-rich endmember. If one were to disassociate odinite from all the other iron-rich 1:1 phyllosilicates, collectively called berthierine, one would introduce as much confusion as if one were to disassociate the extremely magnesium-rich calcite of some modern sediments from the less magnesium-rich calcite which characterizes all ancient rocks.

In Table 1, two samples of Mala Pascua ooids are compared to the odinite of Odin (1988, p. 179) and to berthierine-rich ironstone from the Paz de Rio deposit in Colombia. More berthierine presently is being mined at the Paz de Rio deposit than at any other known deposit (Kimberley, 1980). In Table 2, the compositions of Table 1 have been converted into 'mineralogical components' using a computer program which calculates these components for any ironstone composition (Kimberley, in prep.). Paz de Rio berthierine clearly differs compositionally from the oolitic phyllosilicate which is exposed on the Mala Pascua seafloor (Table 2).

Mineralogical interpretation of the 'mineralogical components' of Table 2 awaits TEM/AEM analysis. If layer-to-layer variation exists in Quaternary and ancient berthierine, these layers presumably would confirm differing amounts of the components listed in Table 2.

Table 1. Composition (Wt. %) of Berthierine (including odinite endmember)

Note: MP42 and MP116 are ooids magnetically concentrated from Mala Pascua samples 42 and 116. These include a small amount of partially ferruginized calcareous fragments but no detrital clay. Odinite, from Odin (1988, p. 179), contains some detrital chlorite. Paz de Rio 165-112 and 173-78 are berthierine-rich, oolitic ironstone from the thesis collection of Kimberley (1974). "Brindley (1982)" is the average of 14 analyses of berthierine which Brindley (1982, Table 1) modified by omitting alkalis and reducing other elements somewhat arbitrarily.

Alternatively, these components may be partitioned among multiple discrete phases which have not been detectable by X-ray diffraction, both in the Quaternary and ancient samples. In particular, the glauconitic component may or may not constitute a separate phase in ferriferous oolite. Available microprobe analysis suggests that glauconite does not occur as a distinct phase (Maynard, 1986; Kimberley, 1974).

Mala Pascua ooids which are exposed to oxidizing seawater clearly exhibit a bulk composition which corresponds to the odinite of Bailey (1988), being rich in both magnesium and ferric iron (Table 1). However, the occurrence of this phyllosilicate as ooids (Fig. 2) contradicts Bailey (1988). Bailey (1988) assumed that he was analysing a single phase but that is not necessarily true. At Mala Pascua, it is possible that the exposed ooids represent partial oxidation of a more chemically reduced predecessor which may still occur within ooids buried under more than 10 cm of sediment.

Bailey (1988) and Odin (1988) believe that odinite disappears from the geological record by converting to chlorite under just a metre of sediment. However, many of their odinite-bearing grains appear to be faecal pellets which may include soil-derived chlorite. In contrast, the Mala Pascua ooids clearly have obtained their concentric structure by phyllosilicate growth on the seafloor.  Neither Bailey nor Odin have visited any of the sites from which odinite has been collected (Odin, 1988, p. 405). Odin (1988, p. 74) expresses surprise that Fe³⁺ dominates over Fe²⁺ in the 1: 1 phyllosilicate which others have collected for him from surficial, shallowwater sediment. Given the high oxidation state of the enveloping seawater, the real surprise might rather be the large proportion of Fe ²⁺ in this authigenic, seafloor silicate (Table 1).

Field relationships and depth-dependent diagenesis must be determined to understand the 1: 1 iron-rich phyllosilicate of exposed seafloor sediment. These relationships may be studied more readily at Mala Pascua than at any of the sites described by Odin (1988) because the authigenic phyllosilicate locally constitutes a larger proportion of the sediment at Mala Pascua than at any site in Odin (1988). The maximum elsewhere is less than 10% (Bailey, 1988). If the Mala Pascua sediment is indeed analogous to ironstone, the 1 :1 phyllosilicate presumably converts to a more ferrous and aluminous berthierine with burial rather than to chlorite. Deeper samples need to be collected at Mala Pascua to test Odin's (1988) contention that the serpentine-like structure disappears upon shallow burial. If the past is the key to the present, odinite does not necessarily disappear but may change progressively to ferrous-endmember berthierine.

Table 2. Components (Wt. %) calculated from compositions in Table 1

Model glauconite, (K+Na) _1.0 Al_0.25   Fe ³⁺_1.3 Fe ²⁺_0.3 Si_3.75 A1_0.25 0₁₀ (OH)₂.

Odinite-endmember berthierine, Fe ²⁺_0.3 Mg_1.0 Fe ³⁺_1.0 Al_0.2  Si_1.8 A1_0.2 0₅ (OH)₄.

Ferrous-endmember berthierine, Fe ²⁺_1.87 Mg_0.43 Fe ³⁺_0.23 Al_0.4  Si_1.5 A1_0.5 0₅ (OH)₄

Excess Fe²⁺ as pure 1:1 Fe-trioctahedral, Fe ²⁺₃ Si₂ 0₅ (OH)₄

Excess Mg as pure 1:1 Mg-trioctahedral, (lizardite) Mg₃ Si₂ 0₅ (OH)₄

Excess Mg as brucite, Mg(OH)₂

Kaolinite, Al₂ Si₂ 0₅ (OH)₄

Nontronite, Fe ³⁺₂ Si₂ 0₅ (OH)₄

Hematite, Fe₂O₃

Goethite, FeOOH

Excess Al as diaspore, AlOOH

Excess Al as Al₂O₃

Excess water, H₂O

Total components (wt. %)

There are persistent compositional differences between the berthierine of ironstone and the comparable seafloor phyllosilicate. No ancient berthierine ooids are known to be as rich in magnesium, as poor in aluminium, or as poor in phosphorus as are the ferriferous ooids of Mala Pascua (Table 1). However, these three differences may result from diagenesis. A hypothetical decrease in magnesium is the easiest to accept because calcareous sediment which initially was rich in highly magnesian calcite has consistently lost most of that magnesium upon conversion to Tertiary limestone (e.g. Milliman, 1974). Evidence for magnesium loss from Mala Pascua ooids comes from electron microprobe analysis of their outermost oolitic layers. These outermost layers consistently are less magnesian than the paler-green subjacent oolitic layers and also resemble ancient berthierine in enhanced aluminium and phosphorus concentrations. Unless one believes that chemical sedimentation of iron differs today from all of previous Earth history, one would tentatively consider the Mala Pascua ooids to be the surficial modern equivalent of normal berthierine ironstone ooids.

If diagenetic alteration of Mala Pascua phyllosilicate occurs progressively, as expected, differentiation of multiple mineral names would be arbitrary and it would be preferable to describe each silicate composition in terms of end members, as is commonly done for other phyllosilicate groups. Mineralogists often select endmember compositions which are pure and simple but which do not occur naturally and cannot be produced as stable experimental products. Hewitt and Abrecht (1986) discredit such endmembers by noting that they cannot be used to represent biotite compositions unambiguously.

Like biotite, berthierine is best described as a combination of observed endmember compositions. The worldwide compositional similarity of modern marine 0.7 nm ferriferous silicates is attributed herein to maximum oxidation without breakdown of the 0.7 nm lattice. It is possible that deep sampling at Mala Pascua will reveal a progressive change to a more chemically reduced 0.7 nm silicate which more closely resembles ancient oolitic berthierine. Until this possibility is tested, it is recommended that odinite be considered an endmember of the berthierine family rather than a distinct mineral.

Considering all available analyses of modern 0.7 nm ferriferous silicate, odinite-endmember berthierine has a composition of roughly Fe ³⁺_1.0 Fe ²⁺_0.3 Mg_1.0 Al_0.2  Si_1.8 A1_0.2 0₅ (OH)₄. This odinite-endmember composition is useful for description of some ancient berthierine, e.g. Cretaceous Mg-rich berthierine which is  Fe ²⁺_1.87 Fe ³⁺_0.23 Mg_0.43 Al_0.4  Si_1.5 A1_0.5 0₅ (OH)₄ (Mandarino and Anderson, 1989, p. 273). Bailey (1988) has proposed an alternative odinite model formula, i.e. Fe ³⁺_0.784 Fe ²⁺_0.279 Mg_0.772 Mn_0.015 Ti_0.016 Al_0.556   Si_0.778 A1_0.212 0₅ (OH)₄.  However, this alternative is excessively precise, considering that it is based upon a single sample of a potentially reactive marine precipitate which was acid-washed prior to chemical analysis.

SEDIMENTARY ENVIRONMENT OF THE MALA PASCUA OOLITE

Phanerozoic ironstone presents a paradox of chemically reduced minerals intermixed with fossils of oxygendemanding animals. The age-old search for modern marine ferriferous ooids has been driven by the knowledge that this paradox of simultaneous oxidation and reduction on the ancient seafloor would best be resolved by finding a modern analogue. A modern analogue of ironstone would not only provide prediagenetic ooids to resolve questions about diagenetic alteration but it should provide insight into iron supply and the setting of chemical sedimentation. The inability of geologists to resolve the ironstone setting from the ancient record is best illustrated by the diversity of published settings in Fig. 5 of Odin et al. (1988a, p. 45). The prime lesson from this diversity is that some types of sedimentary rock preserve only a small proportion of the information which is needed to deduce their palaeoenvironment. Without knowledge of modern sedimentary environments, geologists generally would have little success at deducing palaeoenvironments.

The setting of Mala Pascua ooids provides several clues, the most obvious of which is that this is not a setting in which one would expect calcareous ooids to form. A previous attempt to explain the oxidationreduction paradox of ironstone had invoked sedimentation of calcareous ooids on an oxidizing seafloor, followed by diagenetic replacement of those ooids by chemically reduced, iron-rich groundwater (Kimberley, 1979). Diagenetic replacement is still invoked by an NSF reviewer who considers Mala Pascua to be of only marginal interest when he writes, "... the Mala Pascua outcrop contains the usual minerals of the verdine facies ... The single difference (interesting originality of the outcrop) is the presence of an oolitic structure in some grains. This might well be interpreted as the result of the 'verdinization' of calcareous oolites (a hypothesis which was supported as the exclusive one by MMK in the past). This hypothesis would be supported by the known occurrence of calcite and iron oolites described by Pujos (Bordeaux) from the nearby French Guyana platform."

Pujos (in Odin et al., 1988b) has reported calcareous ooids which formed on the outer Guyana shelf at the end of the last glaciation (17,000 yr BC) and which have been slightly stained with iron oxides (up to 2% Fe₂0₃). The iron-stained Guyana sediment apparently contains much less iron than world-average crust and can hardly be cited as a modern analogue of ironstone. In contrast, the Mala Pascua sediment clearly records an influx of voluminous dissolved iron to the seafloor (not the subseafloor) and probably is a modern analogue of ironstone. The calcareous-oolite-covered seafloor of Guyana lies about 100 km off the coast and underlies 130 m of seawater. Ferriferous-oolite-covered seafloor at Mala Pascua lies as close as 3 km to shore and underlies 35-40 m of seawater. Longshore drift of abundant suspended sediment at Mala Pascua is incompatible with sedimentation of calcareous oolite, just as it presently impedes sedimentation of calcareous oolite everywhere else on the inner shelves of North and South America. The foregoing disparaging comment to NSF about oolitic texture, based on Odin et al. (1988b), is inconsistent with the aforementioned quotation from Odin et al. (1988a, p. 30) who note that, "... the selective study of the oolitic grains is justified".

Pure marine chemical sediment of all types preferentially accumulates on banks which rise above detritus-accumulating depressions. Broad Bahaman banks promote precipitation of calcareous ooids through evaporative concentration of seawater solutes (e.g. Bathurst, 1975). In contrast, the ferriferous oolite at Mala Pascua rests on banks which are too small to have had any topographic influence on seawater composition. The ferriferous oolite occurs sporadically on bank tops from Cape Mala Pascua westward to El Fraile Point, a distance of 6 km along this E-W-trending coastline (Venezuelan topographic map #7548). Intervening ooid-poor sediment ranges from terrigenous mud to calcareous shell hash to quartzose sand. Quartzose sand which lacks ooids consistently separates oolitic sediment from the cliff-faced Venezuelan shore. However, ferriferous oolite lies as close as 3 km to shore north of El Fraile Point and within 5 km of shore north of Cape Mala Pascua. Ooid abundance diminishes in deeper-water sediment, both that which lies farther offshore and that which fills slight depressions within the ooid-covered banks.

On all sides except shoreward of the Mala Pascua oolitic area, there is green mud which lies at an equal to slightly deeper water depth. This green mud has no obvious properties which distinguish it from the green mud which has been found to extend over much of the continental shelf of northeastern Venezuela. The green hue is largely due to phaeopigments, i.e. degradation - products of chlorophyll (Kimberley et al., 1988). Upon burial, this sediment surely will become black shale; most of Earth's shallow marine black shale probably was green mud prior to burial.

An appreciation for the geometry of the oolitic sediment bodies at Mala Pascua must await seismic profiling. Determination of sediment age must await radiometric age dating. Seismic profiling will require an expensive marine expedition but age dating on existing samples would be straightforward because the ferriferous-silicate voids are accompanied by calcareous shell fragments which have been partially replaced by the same ferriferous silicate which constitutes the ooids. The calcareous shells should be amenable to 14C dating because iron mineralization probably has occurred within the Holocene.

The assumption of a Holocene age of mineralization is based on the fact that the ferrous-iron-bearing oolite underlies only 35-40 m of water depth and, if older than Holocene, would have been exposed to atmospheric oxidation during the latest glaciation. According to the eustatic sealevel curve of Fairbanks (1989), the oolite-covered banks have been covered by seawater for less than 10,000 years. By invoking local tectonic activity, one might hypothesize a pre-Holocene Quaternary age but this is unlikely, given that the oolite presently exhibits only 5 m of relief across an area of 6 km (E-W) by 2 km (N-S).

Partially ferruginized calcareous fragments are abundant within Mala Pascua oolitic sediment whereas similar calcareous fragments in nearby ooid-free sediment are not ferruginized. Echinoid spines constitute a significant proportion of these partially ferruginized fragments. Local SCUBA divers report that spiny echinoids are particularly abundant on the oolitic seafloor.

THE NORTH COAST-COCHE FAULT UNDER MALA PASCUA OOLITE

Seafloor topography off Cape Mala Pascua is characterized by an anomalous concentration of scarps which may be fault planes. The oolite lies at the intersection of a major E-W strike-slip fault zone and a minor orthogonal fault zone. The E-W fault zone is a component of the Caribbean-South American plate boundary and extends under water for hundreds of kilometres from Venezuela to Trinidad. This fault zone has been named the North Coast-Coche Fault Zone by Robertson and Burke (1989). Coche island (100 km west of Mala Pascua) is the only place where this fault and its Riedel (additive) shears are exposed on land (Kimberley and Llano, 1991). Here the fault has brought Mesozoic meta-morphic rock into contact with Quaternary fanglomerate. A sharp increase in chemical alteration toward the main fault plane on Coche records substantial fluid migration along that plane.

Exhalative input of iron and silica into Coche fanglomerate has produced veins of intergrown goethite and chert. Some of the ascending iron-rich solutions did not reach the Earth's surface and terminated in mushroom-shaped masses of goethite within the porous fanglomerate (Kimberley, 1989b). These 'mushrooms' are tens of centimetres in diameter and are concentrated at a stratigraphic horizon which runs E-W across the middle of Coche island.

POTENTIAL SOURCES OF IRON FOR MALA PASCUA OOIDS

As within ancient ferriferous ooids, the concentric layering within Mala Pascua ooids clearly records growth at the sediment-water interface. Nonetheless, ooid concentricity is not proof, by itself, of pre-burial growth because concentric layering can form diagenetically as in bauxite (Schellmann, 1969). Within bauxite, diagenetic concentric layers locally truncate earlierformed concentric layers whereas this type of truncation has never been reported from any ferriferous ooid (Bhattacharyya and Kakimoto, 1982; Kimberley, 1979). Ferriferous ooid layering therefore is not diagenetic. If the ferriferous minerals also are not diagenetic, specifically not a replacement of oolitic aragonite, then ooid growth has required an iron-rich, chemically reduced water mass within a water body which otherwise was generally oxidizing, given the oxygen requirements for life forms which are fossilized within most Phanerozoic oolitic iron formations. This oxidizing-reducing paradox has long been recognized as pivotal to the debate about oolitic ironstone (Hallimond, 1925).

In the Mala Pascua region, the iron-rich water mass apparently occurred locally because the ferriferous ooids have formed locally, rather than having been supplied by erosion of some distant source. There is no known potential source rock which contains ferriferous ooids within several hundred kilometres (Van Houten, 1992). Metamorphic rock underlies the drainage basins of all creeks which bring clastic sediment to the shelf around Mala Pascua (Ave Lallemant, 1991). However, the ooids clearly have not been metamorphosed, given their broad X-ray diffraction peaks and their association with partially ferruginized recent fossils.

The iron-precipitating solution at Mala Pascua may well have had a hydrothermal source, given that Pliocene dacite plugs are scattered 25-50 km WSW of Mala Pascua. These shallow young intrusives are each so small (< 1 km2) that they cover less than 1 % of the land area but they have been interpreted to overlie more extensive magma chambers (Zambrano and Seijas, 1966; Santamaria and Schubert, 1974; Vierbuchen, 1984). Hydrothermal alteration plumes are widespread from the region of dacite plugs to Cape Mala Pascua. Seacliffs of the cape itself exhibit hydrothermal plumes between the Medina and Medinita beaches. Similar hydrothermal alteration is exposed within the town of Carupano, 27 km west of Mala Pascua.

Both sporadic and continuous exhalation continue to characterize the region landward and seaward of Cape Mala Pascua. Sporadic exhalation of toxic volatiles at sea causes massive fish kills along the Mala Pascua coast (Ernst, 1886). The recurrence interval of fish kills in El Bichar Bay of Coche island approximates a decade (Kimberley, 1989b). Continuous exhalation of boiling water on land occurs about 30 km SW of Mala Pascua, 5 km WSW of the town of El Pilar. Native sulfur continuously precipitates from the boiling effluent. A wide variety of other hydrothermal precipitates occur in ejected blocks which surround the boiling vents.

If the supply of dissolved iron to Mala Pascua has come from deep sources, as advocated herein, that supply probably has occurred during seismic rupturing of the seals on overpressurized reservoirs of metalliferous saline fluids. The potentially catastrophic supply of fluids from such reservoirs has been described by Cathles (1990) and will not be elaborated herein. Between rupture events, seals apparently reform and convection within sealed volumes may produce more metalliferous fluid by driving water through hot rocks. Powerful earthquakes which could rupture reservoir seals occur frequently in the Mala Pascua region (Perez and Aggarwal, 1981; Fiedler, 1972).

Granitic rocks would be a good potential supplier of dissolved iron at temperatures greater than about 500°C because introduced water could not be consumed to produce a hydrous mineral and the oxidation state could be too high for a purely ferrous mineral (J.A. Speer, pers. comm., 1992). In addition to a potential granitic source, Kimberley (1989b) had considered ophiolite and new basaltic crust but mafic rocks more readily precipitate iron as amphibolite, hence inhibiting its fractionation into an escaping fluid. Moreover, there is no general association of ophiolite with noncherty ironstone or of basalt with cherty ironstone. The primary role of ultramafic rock in the genesis of ironstone probably has involved heat transfer, e.g. from an upwelling mantle under a pullapart basin along a continental-edge transform.

SOURCE OF THE MALA PASCUA IRON AND IMPLICATIONS FOR PETROLEUM

The ferriferous ooids at Mala Pascua appear to have formed by direct precipitation onto the seafloor. The initial precipitate probably was more chemically reduced than the odinite-endmember berthierine which has been sampled to date within the upper 10 cm of the seafloor (Table 1). Although partially ferruginized aragonitic fossils occur within the oolitic sediment, there is no evidence that the ferriferous ooids are replacements of aragonitic ooids (cf. Kimberley, 1979). Moreover, there is no evidence of replacement of either a mixture of kaolinite and ferric oxides (cf. Bhattacharyya, 1983, 1989) or of ferric clay minerals (cf. Odin et al., 1988). The aforementioned field relationships point to an exhalative source for both the Mala Pascua oolite and for deeper-water concentrations of peloidal glauconite that occur elsewhere on the Venezuelan continental shelf (Kimberley, 1989b).

An exhalative model to explain high concentrations of both glauconite and odinite-endmember berthierine carries implications for the origin and migration of petroleum in coastal Venezuela. In general, the production of petroleum from sedimentary kerogen requires both heat and metal catalysts (Mango, 1992a). Both would be supplied by a metalliferous fluid which has risen from an amphibolite-grade metamorphic zone into a sedimentary sequence. Convection within a geopressurized sedimentary reservoir could affect all kerogen within the reservoir, not just the kerogen-rich 'source beds' which usually are emphasized (e.g. Waples, 1985). The composition of the resulting petroleum would partially depend upon the metal ratios within the saline fluid (Mango, 1992b). Being immiscible with a saline fluid, petroleum would preferentially collect at the top of a reservoir where it could escape separately during local rupturing of the reservoir seals. Subsequently formed petroleum probably would differ in composition due to progressive degradation of the kerogen and a change in reservoir boundaries.

If berthierine, glauconite, and petroleum are genetically associated, then the times of maximum petroleum generation within a sedimentary basin may be approximated by the ages of the beds which are richest in authigenic iron silicates, typically glauconite. Additional work is needed to evaluate this putative association. In the meantime, it is noteworthy that sporadic seepage of petroleum has been observed by local inhabitants at Cape Mala Pascua. One of the world's greatest petroleum reservoirs (El Furrial) lies 115 km to the south (Carnevali, 1988). However, only natural gas has yet been reported in any known documentation of the northeastern continental shelf of Venezuela, including the Mala Pascua area (Robertson and Burke, 1989).

If there is sufficient similarity between ancient ironstone and subsurface Mala Pascua (yet to be sampled), one may conclude that the ongoing processes at Cape Mala Pascua are generally related to those which have formed noncherty oolitic ironstone. Demonstration of an exhalative origin for oolitic ironstone would be revolutionary, given that Villain (1902) has been the only specialist on Phanerozoic ironstone who previously has advocated an exhalative source. Recent reviews on oolitic ironstone typically do not even mention the possibility of a metamorphic-exhalative source (e.g. Young and Taylor, 1989; Odin et al., 1988a).

GLOBAL DISTRIBUTION OF QUATERNARY FERRIFEROUS OOIDS

The only serious prior claim of marine ferriferous oolite involves grains which underlie just a couple of metres of seawater in Indonesia (Allen et al. 1979). However, the Indonesian grains are composed entirely of iron hydroxide and resemble concentrically-laminated grains which are similarly concentrated about 100 km west of Cape Mala Pascua, on the southern coastal plain of Margarita Island, Venezuela. In contrast, all known marine noncherty oolitic iron formations include at least some authigenic silicate (0.7 nm berthierine or its compositional equivalent, 1.4 nm chamosite) (e.g. Van Houten and Purucker, 1984; table 2 of Kimberley 1989b). The goethitic Indonesian 'ooids' are either slightly submerged terrestrial grains or became oxidized on the seafloor. In either case, the supply of dissolved iron probably was exhalative, given the frequent exhalation of petroleum through the deltaic seafloor around the Indonesian oolite (Kimberley, 1989b).

Goethitic 'ooids' locally cover the coastal plain of Margarita Island but are not readily attributable to surficial weathering processes because they also occur deep within unweathered rock. Where most abundant, the Margarita 'ooids' appear to have precipitated from fluids which have risen along a fault between Cretaceous serpentinite and Tertiary sedimentary rock. A km-high cliff along this fault zone provides adequate gravitational potential for groundwater to percolate down from high-standing serpentinite and then rise under the adjacent coastal plain without any thermal forcing. In contrast, the Mala Pascua region clearly exhibits thermally driven exhalation.

FIRST STEP BACKWARD INTO EARTH HISTORY: PLIOCENE OOLITIC IRONSTONE

Deeper sampling of ferriferous oolite at Mala Pascua would address the issue of ooid diagenesis. By looking at the next older deposit of ferriferous ooids, at 5 Ma, one finds that diagenesis is complete. This Pliocene noncherty oolitic iron formation, between the Black Sea and Sea of Azov, is one of the most voluminous noncherty iron formations on Earth and it shares all the basic characteristics of voluminous noncherty deposits which have formed since the end of the Archean (table 2 of Kimberley, 1989b).

Voluminous noncherty iron formations of all ages share stratigraphic, textural, and compositional properties. Noncherty iron formations generally occur within condensed sequences. The corresponding palaeotopography would have resembled the modem Kerch-Taman peninsular shelf between the Black and Azov Seas or the shallow banks of the headland off Cape Mala Pascua. Oolitic layering generally is most delicate within silicate-rich ooids but may be almost as delicate within oxide-rich ooids. Ferrous iron is abundant wherever there is a large proportion of detrital clay in the interstices between ooids. Apatite mostly occurs in ooid interstices and rarely as the purely phosphatic ooids which characterize some phosphorite. Noncherty ironstone averages about 1 % P₂0₅. Siderite typically is present, especially in silicate-rich ironstone, and locally is the dominant mineral. Siderite invariably is diagenetic, e.g. occurring as rhombs which cut across primary oolitic layering. Carbon in the siderite of ironstone and in the rhodochrosite of manganostone consistently contains a higher proportion of ¹²C than does the carbonate carbon in marine limestone or dolostone (Hangari et al., 1980; Okita and Shanks, 1992).

Detrital quartz in ironstone generally is more angular than in average sandstone. Detrital quartz may form the core of some ooids but does not appear within the majority of ooids. A more common type of ooid core is a fragment of a broken ooid. Noncherty ironstone commonly is rich in metazoan fossils despite the fact that its high proportion of ferrous iron records the complete absence of dissolved oxygen during iron-mineral precipitation. Echinoderm fragments are commonly partially replaced but siliceous fossils are not. The opposite tendency for replacement characterizes nonoolitic glauconitic ironstone.

Recurrence of the foregoing characteristics from the end of the Archean to the Pliocene indicates that noncherty iron formations have shared similar origins. Genetic processes should be well recorded by the Pliocene Kerch iron formation, given its similarity to older noncherty iron formations (Yakontova et al., 1985). That genesis has been attributed to exhalation because Kerch lies near the centre of a region of ongoing exhalation (Pavlov, 1990; Kimberley, 1989b). Fluids presently are rising at least 8 km through the crust there and are producing voluminous mud volcanoes (Shnyukov et al., 1971, 1986). The Pliocene Kerch iron formation is thickest where it fills mud-volcano depressions (Shnyukov et al., 1971, 1986).

If Pliocene exhalation has produced the Kerch iron formation, that exhalation should have occurred as readily through the land surface as through the ironstone-accumulating seafloor of the ancient Sea of Azov. Evidence of land-based exhalation is found in Pliocene karst fillings west of the Black Sea. Pliocene concretionary ironstone and manganostone are reported to be "common" precipitates within karst caverns in this region (Bleahu, 1989, p. 248). Acidic exhalative fluids probably dissolved the limestone caverns before metalliferous exhalative fluids filled them with the subsurface equivalent of manganiferous Kerch ironstone. Exhalative dissolution of karst caverns is a more widespread phenomenon than is generally appreciated by North American geologists (Tsykin, 1989; Kimberley, 1992).

RELATIONSHIP OF NONCHERTY IRONSTONE TO PHOSPHORITE

Other than phosphorite itself, noncherty ironstone is the only sedimentary rock type which is consistently enriched in phosphorus. Noncherty ironstone locally grades to phosphorite and the origins of these two rock types apparently are related (Kimberley, 1989b). Most phosphate specialists attribute the source of phosphorus to fluvial input into the oceans and concentration of that fluvial-marine phosphorus is attributed to intense upwelling (e.g. Burnett and Riggs, 1990; Compton et al., 1990). However, the common association of phosphorite with authigenic iron silicates (mostly glauconite) cannot be explained by upwelling because seawater is too impoverished in iron (Glenn and Arthur, 1990). Nonetheless, the combined popularity of the upwelling model for phosphorite and the soil-erosion model for noncherty ironstone induced Madon (1992) to attribute gradational ironstone-phosphorite to a coincidence of these completely unrelated processes.

The rarity of both ironstone and phosphorite among common sedimentary rocks makes the coincidence of unrelated genetic processes extremely improbable. Moreover, sources like iron-rich soil or normal seawater would produce more gradational contacts than are generally observed between beds of siliciclastic sediment and beds of either ironstone or phosphorite. Both ironstone and phosphorite are attributed herein to precipitation from concentrated solutions which acquired their solutes by high-temperature dissolution deep beneath the edge of a continental block rather than the weathering surface of that continent. Resolution of this surficial vs. deep issue would help elucidate the origin of Earth's metazoans because their appearance coincided with widespread sedimentation of phosphorite and glauconite (Cook, 1992; Brasier, 1992; Kimberley, 1989b).

Both ironstone and phosphorite commonly are associated with carbonaceous shale. Enrichment in organic carbon generally is attributed to enhanced preservation under a stagnant, oxygen-depleted water mass (de Graciansky et al., 1984). However, the study of modern, oxygen-depleted basins reveals that the sedimentation of organic carbon depends more upon variation in organic productivity than degree of oxygen depletion (Pedersen and Calvert, 1990). Earthquake-induced exhalation of anoxic, nutrient-rich fluids could affect both organic productivity and oxygen depletion but would have a greater probable effect on productivity. Modern exhalation of methane preferentially occurs in tectonically active areas that have fault-bounded, restricted basins, e.g. the Black Sea and the Cariaco Basin of coastal Venezuela. Ongoing tectonism in these two regions produces both topographic impedance of marine circulation and steeply dipping faults which release volatile nutrients from deep reservoirs (Kimberley, 1989b).

RELATIONSHIP OF NONCHERTY TO CHERTY IRON FORMATIONS

Cherry iron formations differ from noncherty iron formations stratigraphically, texturally, and compositionally. The largest cherty iron formations are orders of magnitude more voluminous than the largest noncherty iron formations. The most voluminous cherty iron formations accumulated during the Precambrian whereas the most voluminous noncherty iron formations accumulated during the Phanerozoic. Cherty iron formations apparently covered significant portions of Precambrian continental shelves and are commonly associated with dolostone, quartz-rich sandstone, and black shale. Noncherty iron formations are more commonly associated with limestone than dolostone. Cherty iron formations locally grade to other types of chemical sedimentary rock bodies, typically formations of chert or dolostone, whereas noncherty iron formations more commonly grade to clastic sedimentary rock units.

Shallow-water cherty iron formations include rip-up clasts whereas rip-up clasts are rare in noncherty iron formations. This difference is attributable to an initially gelatinous state of cherty ironstone whereas noncherty ironstone presumably has been granular like the modern Mala Pascua sediment.

The oolitic texture which is so characteristic of noncherty ironstone also occurs in shallow-water cherty ironstone but oolitic texture does not predominate within any voluminous cherty iron formation. In fact, it is completely absent from most cherty iron formations. Nonetheless, the global total of cherty oolitic ironstone probably exceeds that of noncherty oolitic ironstone because the collective volume of cherty iron formations is so much larger than that of noncherty iron formations.

Compositional differences between cherry and noncherty ironstone include a typically higher concentration of both aluminum and phosphorus within the latter. Silicon is more concentrated in cherty ironstone, as expected from the chert content, but silicate-rich noncherty ironstone also records precipitation from a highly siliceous fluid. Similarities between cherty and noncherty ironstone include the average concentration of iron and typical ratios of ferric-to-ferrous iron. Moreover, neither type exhibits any age dependence for the ratio of ferric-to-ferrous iron, from the Archean to the Pliocene. This lack of age dependence for the oxidation state casts doubt upon models of atmospheric evolution based upon the evolution of iron formations (e.g. Towe, 1983).

Rhythmic lamination (banding) is ubiquitous in cherty ironstone but virtually absent in noncherty ironstone. Some mm-scale lamination within iron formations of Western Australia is correlative across 130 km (Morris, 1993; Trendall, 1983). Morris (1993, p. 264) notes, "Three major sources for the iron and silica in the ocean have been proposed. Firstly, the direct chemical erosion of the continents; secondly, the continental sediments in the oceans; and thirdly, the submarine reactions of outpouring lavas and the concomitant hydrothermal output of 'hot-spots' or MOR." However, all three of these are rejected herein in favour of a deeper source.

Some of the lamination (banding) within iron formations is attributable to precipitation along a long-lasting marine interface between underlying iron-rich seawater and overlying iron-poor seawater (fig. 17-10 of Kimberley and Kimberley, 1992). However, most ironstone lamination, particularly where areally extensive or enclosing oolitic layers, is attributable herein to deep-source exhalative fluids which exploded into the atmosphere, typically through an ocean, and rained precipitates onto the Earth's surface. Hypothetical atmospheric precipitates which fell onto a water body would have produced the laminated chemical sediment which is characteristic of voluminous Precambrian iron formations whereas atmospheric precipitates onto land would have produced nonlaminated sediment.

Several Precambrian iron formations include a nonlaminated, more oxidized facies which is loosely called 'iron ore' because it hosts the best ore bodies. The 'iron ore' facies commonly is attributed to postdepositional weathering of cherty laminated ironstone but its origin remains unclear (e.g. Morris, 1985; Goldring, 1991). In this paper, most 'iron ore' is attributed to a 'raining' of exhalative precipitates onto exposed land. Some 'iron ore' may well be weathered laminated ironstone but most is considered to be a terrestrial facies of cherry ironstone which never was laminated.

The most voluminous noncherty iron formations are tiny compared to typical cherty iron formations. The smaller volume of noncherty iron formations and their smaller proportion of chemically precipitated silica are both attributed herein to less potent exhalation of cooler fluids, given that a high concentration of dissolved silica requires a very high temperature (Fournier and Potter, 1982).

Cherty and noncherty iron formations collectively represent the most voluminous marine formations which contain ¹²C -enriched carbonate minerals. This enrichment in ¹²C is attributed to a high-temperature exhalative component because carbon dioxide becomes more enriched in ¹²C at high temperature in the presence of hydrocarbons. Isotopic fractionation generally diminishes with temperature so convective mixing of hydrocarbons and carbon dioxide within a geopressurized reservoir would produce ¹²C -enriched carbon dioxide.

MODERN IRON-MANGANESE MINERALIZATION

Voluminous cherty iron formations have not accumulated on Earth since the late Precambrian-early Palaeozoic transition. However, small cherry iron formations have continued to form through the Phanerozoic so a modern cherty analogue remains a possibility. Phanerozoic cherty iron formations are more commonly associated with sulfide ore bodies than are Precambrian examples. Hence, to use the past as a key to the present, one might examine the recent "cataclysmic hydrothermal venting on the Juan de Fuca Ridge" off Washington State (Baker et al., 1987).

The most voluminous body of modern iron-rich seawater is the Orca Basin under the northern Gulf of Mexico (Sheu and Presley, 1986). This nonvolcanic basin underlies 2 km of normal seawater and contains a thousand times more dissolved iron than does average seawater. Just below the interface between iron-poor and iron-rich seawater, the solute content of manganese is 10⁵ times that of average seawater (Sheu et al., 1988). The reason for abundant iron-manganese solutes in Orca seawater is not yet known and deserves additional study by those interested in cherty iron formations.

The Orca Basin exhibits an areally extensive interface (400 km²) between underlying iron-rich, phosphorusrich seawater and overlying normal seawater. The interface is stabilized by the high density of underlying hypersaline water. Upwelling could not occur across this stable interface but uniform precipitation could occur along it. Such uniform precipitation could account for some of the lamination of cherty iron formations but not within the shallowest-water (oolitic) facies because wind-driven mixing would have precluded maintenance of a stable interface there.

A hypothetical deep iron-precipitating interface surely would have been populated by Archaebacteria as along the modern Orca interface (Dickins and Van Vleet, 1992). Bacteria would have enhanced mineral precipitation along this interface, independent of water depth to the interface (Robbins et al., 1987). The foregoing interface-precipitation and atmospheric-precipitation models may be contrasted with one of photooxidative precipitation of dissolved iron following upwelling into the photic zone (Anbar and Holland, 1992). Upwelling is inherently less spatially uniform than either of the foregoing models and any upwelling model is therefore difficult to reconcile with the extensively thin laminae of cherty ironstone in Western Australia.

At the southeastern end of the Gulf of Mexico, Quaternary manganese mineralization on Grand Cayman Island deserves scrutiny because it occurred during or immediately following dolomitization (Jones, 1992). Voluminous dolostone sedimentation commonly preceded voluminous cherty ironstone and manganostone sedimentation in the Early Proterozoic (Gross, 1965, p. 91; Schissel and Aro, 1992). However, young dolostone is exceedingly rare and so an association of Late Pliocene-Quaternary dolostone with manganese mineralization deserves attention. If one prefers an exhalative origin for iron formations, one would attribute the combined influx of manganese and magnesium to fluid exhalation in the tectonically active Cayman area. However, one must recognize that diagenetic processes have controlled the mineralogy and areal distribution of Cayman mineralization just as diagenetic processes are controlling the fate of exhalative iron-manganese mineralization in the Lake Malawi rift system of eastern Africa (Williams and Owen, 1992).

Lake Malawi contains ferriferous silicate ooids which are comparable to the Mala Pascua ooids but are more oxidized, being composed of nontronite instead of odinite-endmember berthierine (Miiller and Forstner, 1973). Although the marine setting and 0.7 nm mineralogy of Mala Pascua oolite are more characteristic of oolitic ironstone, Lake Malawi oolite more closely resembles oolitic ironstone in its abundance of phosphate, i.e. 0.5% P₂0₅ which is about twice that of Mala Pascua ooids (Table 1). An alumina content of about 5% A1₂0₃ is remarkably consistent among Lake Malawi, Mala Pascua, and various odinite localities. The source of Lake Malawi iron apparently has been the same as for marine ironstone and that source has been deduced to be exhalative-hydrothermal by Miiller and Forstner (1973). The 700-m deep Lake Malawi receives hydrothermal silicon and fluorine as well as iron and manganese (Williams and Owen, 1992). The ultimate heat source probably is mantle upwelling into stretched lithosphere (McKenzie, 1978).

SOLUBILITY DEPENDENCE ON TEMPERATURE, SALINITY, ACIDITY, AND ORGANIC IONS

The solubility of metals generally is enhanced by an increase in temperature, ionic strength, acidity, and/or organic complexing. Fluid inclusions suggest that hot, hypersaline solutions may contain iron, manganese, and zinc in the weight percent range (Rankin et al., 1992). Experimental studies confirm the importance of iron-chloride complexes but have not yet achieved such high solubilities (Fein et al., 1992).

Organic complexing may be important at low temperatures for aluminum (Palmer and Wesolowski, 1992) and at moderate temperatures for trace metals such as gold and mercury (Myakori kiy and Dmitriyev, 1992). Fluorine complexes aluminum at high temperature and is the preferred carrier of aluminum in the high-temperature exhalative model advocated herein, given the aforementioned evidence of fluorine exhalation in Lake Malawi.

Lake Malawi and other exhalative sites should be compared to sites of modern low-temperature mineralization. Metalliferous acidic saline groundwater occurs extensively in southern and western Australia (Long et al., 1992a). The solute ratios of this brine resemble those of seawater rather than evaporated freshwater. The solute source locally may be Holocene bedded halite (Handford, 1991) but elsewhere is attributed to marine aerosols (Long et al., 1992a). Some of the Australian brines dissolve iron from shallow aquifers and rise into playa lakes, producing millimolar concentrations of dissolved iron within the water bodies (Long and Lyons, 1992). Voluminous jarosite (KFe₃[S0₄]₂[OH]₆) and chert precipitate onto the lake floors and within pores of adjacent sediment (Long et al., 1992b). Deeper dissolution of iron is providing ferriferous groundwater (> 50 mg1^-1 Fe) to a South Australia bay where a resulting, 20-cm thick deposit of modern sediment averages 40% Fe (Ferguson et al., 1983). However, even this optimal case of gravitydriven groundwater supply has been shown by Ferguson et al. (1983) to be inadequate to produce a noncherty iron formation like the voluminous Pliocene Kerch deposit. A more potent process such as hydrothermal exhalation apparently is required.

The high concentration of dissolved iron in southern and western Australia has been attributed uniquely to organic decay within soils (McArthur et al., 1991) but some other factor, e.g. high salinity of the groundwater, must be operative or else similarly metalliferous groundwater would be found extensively elsewhere in the world, in fresh groundwater of similar oxidation state. The Australian brines offer a better modern analogue for sedimentary uranium deposits and copperlead-zinc deposits than for noncherty iron formations,  given up to 12.3 mg 1^-1 U, 38.5 mg 1^-1 Pb and 3.5 mg l^-1 Cu in the brines (Giblin, 1987; Giblin and Dickson, 1992). Moreover, iron formations are generally not stratigraphically correlative with evaporites whereas fluid-inclusion studies on sedimentary Cu-Pb-Zn deposits commonly attribute mineralization to lateral brine migration from salt deposits. Distant lateral migration of brine is invoked by many geologists to explain Cu-Pb-Zn beds, disseminated veins, and karst infillings (e.g. Williams-Jones et al., 1992). Alternatively, the saline ore fluids may have migrated subvertically from geopressurized reservoirs during earthquake rupturing of reservoir seals (e.g. Cathles, 1990). In either case, the temperature during initial dissolution of Pb-Zn generally is estimated to be 100-250°C (Turner, 1992). In contrast, the temperature for initial dissolution of ironstone iron is estimated herein to exceed 400°C.

If the present is the key to the past, hypersaline ironrich lakes in Australia contradict existing iron-formation models which invoke precipitation within hypersaline lakes because existing models envision an extremely high pH whereas the Australian lakes are extremely acidic (Kempe and Degens, 1985). The Australian playa lakes have a pH of about 3 because hydroxyl ions are precipitating as iron hydroxides (McArthur et al., 1991).

Only one iron formation is known to contain abundant pseudomorphs of evaporite minerals (Martini, 1990). This 2.2 Ga iron formation is closely associated with noncherty iron formations and overlies a well-developed paleosol. All three characteristics probably resulted from trapping of exhalative ferriferous fluid within a playa lake. The associated noncherty iron formations share all of the recurring characteristics of Phanerozoic noncherty iron formations despite having formed during the Early Proterozoic (Schwiegart, 1965; Wagner, 1928). The associated paleosol is controversial (Retallack, 1986) and may reveal that some Precambrian paleosols record an enhancement of surficial weathering by the exhalative supply of reactants and nutrients onto extensive areas of the Earth's surface. The prime evidence for an exhalative influence on several Precambrian paleosols is the association of these paleosols with stratiform phosphatic uranium mineralization (Kimberley, 1992).

CONCLUSIONS

Our understanding of iron formations presently is minimal and will remain so until the ongoing debate stimulates additional study of modern analogues. Significant evidence has been provided by geochemical indicators within ancient rocks such as stableisotope ratios and rare-earth-element patterns. However, multiple explanations for these data and the possibility of diagenetic alteration combine to inhibit deduction of genetic processes. Even geologists with intimate knowledge of the geochemical data are alternating between advocating the radically different processes of surficial weathering versus hydrothermal exhalation (e.g. Holland, 1984 vs. Anbar and Holland, 1992). The prime lesson from this confusion is that chemical sedimentary rocks inherently provide little clear information about metal-accumulation processes relative to modern sites of analogous mineralization. An indication of what could happen for ironstone is provided by recent progress on massive-sulfide sedimentation through comparison of ancient deposits with modern analogues (e.g. Herzig et al., 1991).

ACKNOWLEDGEMENTS

I thank Fundacion La Salle de Ciencias Naturales for logistical support during twenty field trips in coastal Venezuela over the past eight years.

REFERENCES

Abu-Jaber N.S. and Kimberley M.M. (1992a) Origin of ultramafic-hosted magnesite on Margarita Island, Venezuela, Mineralium Deposita, 27, 234-241.

Abu-Jaber N.S. and Kimberley M.M. (1992b) Origin of ultramafic-hosted vein magnesite deposits, Ore Geol. Rev., 7,155191.

Anbar A.D. and Holland H.D. (1992) The photochemistry of manganese and the origin of banded iron formations, Geochim. Cosmochim. Acta, 56, 2595-2603.

Anderson D. (1989) Theory of the Earth. Blackwell Scientific Publications, Boston, 366 pp.

Ave Lallemant H.G. (1991) The Caribbean-South American plate boundary, Araya Peninsular, Eastern Venezuela. In:

Trans. 12th Caribbean Geol. Conf., (ed. by D.K. Larue and G. Draper), pp. 461-471. Miami Geological Society, South Miami, FL.

Bailey S.W. (1988) Odinite, a new dioctahedral - trioctahedral Fe 3'-rich I: 1 clay mineral, Clay Minerals, 23, 237-247.

Baker E.T., Massoth G.J. and Feely R.A. (1987) Cataclysmic hydrothermal venting on the Juan de Fuca Ridge, Nature, 329, 149-151.Barrett T.J. and Fralick P.W. (1985) Sediment redeposition in Archean iron formation: examples from the Beardmore-Geraldton greenstone belt, Ontario, J. Sedim. Petrol., 55, 205-212.

Bathurst R.G.C. (1975) Carbonate Sediments and their Diagenesis, 2nd edn. Elsevier, Amsterdam, 658 pp.

Berner R.A. (1980) Early Diagenesis: a theoretical approach. Princeton University Press, Princeton, NJ, 241 pp.

Bhattacharyya D.P. (1992) Secular variation and chemistry of the REE in Phanerozoic oolitic ironstones, Geol. Soc. Am. Abstr. Prog., 24, A214-A215.

Bhattacharyya D.P. and Kakimoto P.K. (1982) Origin of ferriferous ooids; an SEM study of ironstone ooids and bauxite pisoids, J. sedim. Petrol., 52, 849-857.

Bleahu M. (1989) Paleokarst of Romania. In: Paleokarst; a systematic and regional review (ed. by P. Bosak, D. Ford, J. Glazek and I. Horacek), pp. 231-251. Elsevier, Amsterdam, 725 pp.

Brasier M.D. (1992) Nutrient-enriched waters and the early skeletal fossil record, J. Geol. Soc. London, 149, 621-629.

Brindley G.W. (1982) Chemical compositions of berthierines - A review, Clays Clay Miner., 30, 153-155.

Brindley G.W. and Youell R.F. (1953) Ferrous chamosite and ferric chamosite, Miner. Mag., 30, 57-70.

Bdhn B., Stanistreet I.G. and Okrusch M. (1992) Late Proterozoic outer shelf manganese and iron deposits at Otjosondu (Namibia) related to the Damaran oceanic opening, Econ. Geol., 87, 1393-1411.

Burnett W.C. and Riggs S.R. (eds) (1990) Phosphorite Deposits of the World, vol. 3: Neogene to modern phosphorites. Cambridge University Press, NY, 464 pp.

Carnevali J.O. (1988) El Furrial oil field, northeastern Venezuela: First giant in foreland fold and thrust belts of Western Hemisphere, Bull. Am. Ass. Petrol. Geol., 72, 168.

Cathles L.M. (1990) Scales and effects of fluid flow in the upper crust, Science, 248, 323-329.

Compton J.S., Snyder S.W. and Hodell D.A. (1990) Phosphogenesis and weathering of shelf sediments from the southeastern United States: Implications for Miocene 513C excursions and global cooling, Geology, 18, 1227-1230.

Cook P.J. (1992) Phosphogenesis around the Proterozoic Phanerozoic transition, J. Geol. Soc. London, 149, 615-620.

de Graciansky P.C., Deroo G., Herbin J.P., Montadert L., Miiller C., Schaaf A. and Sigal J. (1984) Ocean-wide stagnation episode in the Late Cretaceous, Nature, 308, 346-349.

Dickins H.D. and Van Vleet E.S. (1992) Archaebacterial activity in the Orca Basin determined by the isolation of characteristic isopranyl ether-linked lipids, Deep-Sea Res., 39, 521-536.

Dimroth E. and Chauvel J.-J. (1973) Petrography of the Sokoman iron formation in part of the central Labrador Trough, Quebec, Canada, Bull. geol. Soc. Am., 84, 111-134.

Ernst A. (1886) On some interesting cases of migrations of marine fishes, on the coast of Venezuela at Carupano, Nature, 33, 321-322.

Fairbanks R.G. (1989) A 17,000-year glacioeustatic sea-level record: influence of glacial melting rates on the Younger Dryas event and deep-ocean circulation, Nature, 342, 637-642.

Fein J.B., Hemley J.J., D'Angelo W.M., Komninou A. and Sverjensky D.A. (1992) Experimental study of iron-chloride complexing in hydrothermal fluids, Geochim. Cosmochim. Acta, 56, 3179-3190.

Ferguson J., Burne R.V. and Chambers L.A. (1983) Iron mineralization of peritidal carbonate sediments by continental groundwaters, Fisherman Bay, South Australia, Sediment. Geol., 34, 41-57.

Fiedler G. (1972) La liberacion de energia sismica en Venezuela, volumenes sismicos y mapa de isosistas. Mem. Fourth Geol. Congr. Venezuela, Boletin de Geologia, Spec. Pub[. Ministerio de Minas e Hidrocarburos de Venezuela, 4, 2441-2462.

Fournier R.O. and Potter R.W. (1982) An equation correlating the solubility of quartz in water from 25° to 900°C at pressures up to 10,000 bars, Geochim. Cosmochim. Acta, 46, 1969-1973.

Frakes L. and Bolton B. (1992) Effects of ocean chemistry, sea level, and climate on the formation of primary sedimentary manganese ore deposits, Econ. Geol., 87, 1207-1217.

Giblin A.M. (1987) Applications of groundwater geochemistry to genetic theories and exploration methods for Early Tertiary sediment-hosted uranium deposits in Australia, Uranium, 3, 165-186.

Giblin A.M. and Dickson B.L. (1992) Source, distribution and economic significance of trace elements in groundwaters from Lake Tyrell, Victoria, Australia, Chem. Geol., 96,133-149.

Glenn C.R. and Arthur M.A. (1990) Anatomy and origin of a Cretaceous phosphorite-greensand giant, Egypt, Sedimentology, 37, 123-154.

Goldring D.C. (1991) Significance of pre- or syntectonic origin for certain iron ores hosted in banded iron formation, Trans. Inst. Min. Metall., Sect. B 100, B148-B158.

Gross G.A. (1965) Geology of Iron Deposits in Canada, vol. 1: General geology and evaluation of iron deposits. Geological Survey of Canada, Econ. Geol. Report no. 22, 181 pp.

Hallimond A.F. (1925) Iron ores: bedded ores of England and Wales, petrography and chemistry. Spec. Rep. Miner. Res. Mem. Great Britain Geol. Surv., 29, 129 pp.

Handford C.R. (1991) Marginal marine halite: sabkhas and salinas. In: Evaporites, Petroleum and Mineral Resources (ed. by J.L. Melvin), pp. 1-66. Elsevier, Amsterdam.

Hangari K.M., Ahmad S.N. and Perry E.C. Jr. (1980) Carbon and oxygen isotope ratios in diagenetic siderite and magnetite from Upper Devonian ironstone, Wadi Shatti district, Libya, Econ. Geol., 75, 538-545.

Haq B.U., Hardenbol J. and Vail P.R. (1987) Chronology of fluctuating sea levels since the Triassic, Science, 235, 11561167.

Herzig P.M., Hannington M.D., Scott S.D., Maliotis G., Rona P.A. and Thompson G. (1991) Gold-rich sea-floor gossans in the Troodos ophiolite and on the mid-Atlantic ridge, Econ. Geol., 86, 1747-1755.

Holland H.D. (1984) The Chemical Evolution of the Atmosphere and Oceans. Princeton University Press, Princeton, NJ, 582 pp.

Horvath L. and Gault R.A (1990) The mineralogy of Mont Saint-Hilaire Quebec. The Mineralogical Record, 21, 284-368.

Hutchinson R.W. (1992) Mineral deposits and metallogeny: indicators of Earth's evolution. In: Early Organic Evolution (ed. by M. Schidlowski et al.), pp. 521-545. Springer-Verlag, Berlin, 555 pp.

Jones B. (1992) Manganese precipitates in the karst terrain of Grand Cayman, British West Indies, Can. J. Earth Sci., 29, 1125-1139.

Kempe S. and Degens E.T. (1985) An early soda ocean?, Chem. Geol., 53, 95-108.

Kimberley M.M. (1979) Origin of oolitic iron formations. J. sedim. Petrol., 49, 111-131.

Kimberley M.M. (1989a) Nomenclature for iron formations, Ore Geol. Rev., 5, 1-13.

Kimberley M.M. (1989b) Exhalative origins of iron formations, Ore Geol. Rev., 5, 13-145.

Kimberley M.M. (1990) A hand-operated, 180° rotating-scoop sediment sampler, Continental Shelf Res., 10, 1157-1169.

Kimberley M.M. (1992) Significance of Precambrian paleosols. In: Early Organic Evolution (ed. by M. Schidlowski et al.), pp. 115-129. Springer-Verlag, Berlin, 555 pp.

Kimberley M.M. and Kimberley S.J. (1992) Guide to The Dynamic Earth, 2nd edn. Wiley and Sons, New York, 327 pp.

Kimberley M.M., Abu-Jaber N. and Llano M. (1988) Presencia de feopigmentos y silicatos ferriferos (glauconita) en los fangos verdes de los alredadores de Margarita. United Nations (FAO) Symp., Congreso Theroamericano y del Caribe, Venezuela, Abstracts Vol., p. 13.

Kimberley M.M. and Llano M. (1991) Structural lineaments in the Margarita-Araya region of Venezuela: boundaries of ecological environments, Mem. Soc. Ciencias Natur. La Salle, 51,25-41.

Long D.T., Fegan N.E., Lyons W.B., Hines M.E., Macumber P.G. and Giblin A.M. (1992a) Geochemistry of acid brines: Lake Tyrrell, Victoria, Australia, Chem. Geol., 96, 33-52.

Long D.T., Fegan N.E., McKee J.D., Lyons W.B., Hines M.E. and Macumber P.G. (1992b) Formation of alunite, jarosite and hydrous iron oxides in a hypersaline system: Lake Tyrrell, Victoria, Australia, Chem. Geol., 96, 183-202.

Madon M.B.H. (1992) Depositional setting and origin of berthierine oolitic ironstones in the Lower Miocene Terengganu Shale, Tenggol arch, offshore peninsular Malaysia. J. sedim. Petrol., 62, 899-916.

Mandarino J.A. and Anderson V. (1989) Monteregian Treasures: the minerals of Mont Saint-Hilaire, Quebec. Cambridge University Press, NY, 281 pp.

Mango F.D. (1992a) Transition metal catalysis in the generation of petroleum and natural gas, Geochim. Cosmochim. Acta, 56,553-555.

Mango F.D. (1992b) Transition metal catalysis in the generation of petroleum: A genetic anomaly in Ordovician oils, Geochim. Cosmochim. Acta, 56, 3851-3854.

Martini J.E.J. (1990) An Early Proterozoic playa in the Pretoria Group, Transvaal, South Africa, Precambr. Res., 46, 341-351.

Maynard J.B. (1986) Geochemistry of oolitic iron ores, an electron microprobe study, Econ. Geol., 81, 1473-1483.

McArthur J.M., Turner J.V., Lyons W.B., Osborn A.O. and Thirwall, M.F. (1991) Hydrogeochemistry on the Yilgarn Block, Western Australia: ferrolysis and mineralisation in acidic brines, Geochim. Cosmochim. Acta, 55, 1273-1288.

McCaig A.M. (1988) Deep fluid circulation in fault zones, Geology, 16, 867-870.

McKenzie D. (1978) Some remarks on the development of sedimentary basins, Earth Planet. Sci. Lett., 40, 25-32.

Morris R.C. (1985) Genesis of iron ore in banded ironformation by supergene and supergene-metamorphic processes - a conceptual model. In: Handbook of Strata-bound and Stratiform ore deposits, vol. 13 (ed. by K.H. Wolf), pp. 73-235. Elsevier, Amsterdam.

Morris R.C. (1993) Genetic modelling for banded iron-formation of the Hamersley Group, Pilbara Craton, Western Australia, Precambr. Res., 60, 243-286.

Miiller G. and Forstner U. (1973) Recent iron ore formation in Lake Malawi, Africa, Mineral. Deposita, 8, 278-290.

Myakon'kiy A.G. and Dmitriyev S.N. (1992) Compounds formed by gold and mercury with organic acids in Cheleken-Peninsular hot brines. Geochem. Int., 29, 129-134.

Odin G.S. (ed.) (1988) Green Marine Clays. Elsevier, Amsterdam, 445 pp.

Odin G.S., Knox R.W.O'B., Gygi R.A. and Guerrak S. (1988a) Green marine clays from the oolitic ironstone facies: habit, mineralogy, environment. In: Green Marine Clays (ed. by G.S. Odin), pp. 29-52. Elsevier, Amsterdam.

Odin G.S., Mackinnon I.D.R. and Pujos M. (1988b) The verdine facies off French Guiana. In: Green Marine Clays (ed. by G.S. Odin), pp. 105-130. Elsevier, Amsterdam.

Okita P.M. and Shanks W.C. (1992) Origin of stratiform sedimenthosted manganese carbonate ore minerals: examples from Molango, Mexico and TaoJiang, China. Chem. Geol., 99,139-164.

Palmer D.A. and Wesolowski D.J. (1992) Aluminum speciation and equilibria and aqueous solution: II. The solubility of gibbsite in acidic sodium chloride solutions from 30 to 70°C,

Geochim. Cosmochim. Acta, 56, 1093-1111.

Pavlov D.I. (1990) Possible genetic association of Fe sedimentary deposits with oil-bearing basins. Abst. 13^th Int. Sedimentology Congr., pp. 172.

Pedersen T.F. and Calvert S.E. (1990) Anoxia vs. productivity: What controls the formation of organic-carbon-rich sediments and sedimentary rocks?, Bull. Am. Ass. Petrol. Geol., 74, 454-466.

Perez O.J. and Aggarwal Y.P. (1981) Present-day tectonics of the southeastern Caribbean and northeastern Venezuela, J. geophys. Res., 86, 10791-10804.

Rankin A.H., Ramsey M.H., Coles B., Van Langevelde F. and Thomas C.R. (1992) The composition of hypersaline, ironrich granitic fluids based on laser-ICP and synchrotron-XRF microprobe analysis of individual fluid inclusions in topaz, Mole granite, eastern Australia, Geochim. Cosmochim. Acta, 56,67-79.

Retallack G. (1986) Reappraisal of a 2200 Ma-old paleosol near Waterval Onder, South Africa, Precambr. Res., 32, 195-232.

Robbins E.I., LaBerge G.L. and Schmidt R.G. (1987) A model for the biological precipitation of Precambrian iron formations: B, Morphological evidence and modern analogs. In: Precambrian Iron Formations (ed. by P.W.U. Appel and G.L. LaBerge), pp. 97-139. Theophrastus, Athens, 674 pp.

Robertson P. and Burke K. (1989) Evolution of southern Caribbean plate boundary, vicinity of Trinidad and Tobago, Bull. Am. Ass. Petrol. Geol., 73, 490-509.

Ronov A.B. (1992) Chemical evolution of sedimentary and magmatic rocks in the Earth's crust. In: Early Organic Evolution (ed. by M. Schidlowski et al.), pp. 546-549. Springer-Verlag, Berlin, 555 pp.

Santamaria F. and Schubert C. (1974) Geochemistry and geochronology of the southern Caribbean-northern Venezuela plate boundary, Bull. geol. Soc. Am., 7,1085-1098.

Schellmann W. (1969) Di Bildungsbedingungen sedimentarer Chamosit-und Hamatit-Eisenerz am Beispiel der Lagerstatte Echte, N. Jb. Miner. Abh., 111, 1-31.

Schissel D. and Aro P. (1992) The major Early Proterozoic sedimentary iron and manganese deposits and their tectonic setting, Econ. Geol., 87, 1367-1374.

Schweigart H. (1965) Genesis of the iron ores of the Pretoria Series, South Africa, Econ. Geol., 60, 269-298.

Sheu D.-D. and Presley B.J. (1986) Formation of hematite in the euxinic Orca Basin, northern Gulf of Mexico, Mar. Geol., 69, 309-321.

Sheu D.-D., Shakur A., Pigott J.D., Wiesenburg D.A., Brooks J.M. and Krouse H.R. (1988) Sulfur and oxygen isotopic compositions of dissolved sulfate in the Orca basin: implications for origin of the high-salinity brine and oxidation of sulfides at the brine-seawater interface, Mar. Geol., 78, 303-310.

Shnyukov E.F., Sobolevsku Yu V., Gnamenko G.I., Naumenko P.I. and Kumnu V.A. (1986) Mud Volcanoes in the KerchTaman Region (in Russian). Kiev Nauk Dumka, 152 pp.

Shnyukov E.F., Naumenko P.I., Lebedev Yu S., Usenko V.P., Gordievich V.A., Yukhanov I.S. and Shiritsa A.S. (1971) Mud Volcanism and Ore Formation (in Russian). Kiev Nauk Dumka, 332 pp.

Tow K.M. (1983) Precambrian atmospheric oxygen and banded iron formations: A delayed ocean model, Precamb. Res., 20, 161-170.

Tendall A.F. (1983) The Hamersley basin. In: Iron formation: facts and problems (ed. by A.F. Trendall and R.C. Morris), pp. 69-129. Elsevier, Amsterdam, 558 pp.

Tsykin R.A. (1989) Paleokarst of the Union of Soviet Socialist Republics. In: Paleokarst; a systematic and regional review (ed. by P. Bosak, D.C. Ford, J. Glazek and I. Horacek), pp. 253295. Elsevier, Amsterdam, 725 pp.

Turner R.J.W. (1992) Formation of Phanerozoic stratiform sediment-hosted zinc-lead deposits: evidence for the critical role of ocean anoxia, Chem. Geol., 99, 165-188.

Van Houten F.B. (1992) Review of Cenozoic ooidal ironstones, Sediment. Geol., 78, 101-110.

Vierbuchen R.C. (1984) The geology of the El Pilar fault zone and adjacent areas in northeastern Venezuela. In: The Caribbean-South American Plate Boundary and Regional

Tectonics (ed. by W.E. Bonini, R.B. Hargraves, and R. Shagam), pp. 189-212. Mem. Geol. Soc. Am., 162, 421.

Villain F. (1902) Le gisement de minerai de fer oolithique de la Lorraine. Dunod, Paris, 179 pp.

Wagner P.A. (1928) The iron deposits of the Union of South Africa. Mem. geol. Surv. Sth Afr., 26, 268 pp.

Waples D.W. (1985) Geochemistry in Petroleum Exploration. Int. Human Res. Develop. Corp., Boston, MA, 232 pp.

Williams T.M. and Owen R.B. (1992) Geochemistry and origins of lacustrine ferromanganese nodules from the Malawi Rift, Central Africa, Geochim. Cosmochim. Acta, 56, 2703-2712.

Williams-Jones A.E., Schrijver K., Doig R. and Sangster D.F. (1992) A model for epigenetic Ba-Pb-Zn mineralization in the Appalachian thrust belt, Quebec: evidence from fluid inclusions and isotopes, Econ. Geol., 87, 154-174.

Yakontova L.K., Andreeva N.Ya., Cipurskij S.I. and Naumenko P.I. (1985) New data on the mineralogy and formation conditions of the Kerch iron ore (in Russian), Miner. Zhurnal (Kiev Nauk Dumka), 7, 29-43.

Young T.P. and Taylor W.E.G. (1989) Phanerozoic Ironstones. Spec. Publ. Geol. Soc. London, 46, 251 pp.

Zambrano E. and Seijas F. (1966) Mapa geologico de la region norte central del estado Sucre, 1:50,000. Ministerio de Minas e Hidrocarburos, Venezuela.

Manuscript received 7 December 1992; revision accepted 15 October 1993

End of Text