Unusual nickel and copper to noble-metal ratios from the Råna Layered Intrusion, northern Norway

SARAH-JANE BARNES

Barnes, S.-J.: Unusual nickel and copper to noble-metal ratios fromthe Råna Layered Intrusion, northem Norway. Norsk Geologisk Tidsskrift, Vol. 67, pp. 215-231. Oslo 1987. ISSN 0029-196X.

Rocks from the Råna Layered Intrusion usually have low concentrations of noble metals but have normal Ni and Cu concentrations. Consequently the rocks have higher Cu/Ir, Ni/lr, Cu/Pd and Ni/Pd ratios than extrusive rocks from the literature. If a small amount of sulphide was removed from the magma prior to its emplacement at Råna the magma would be depleted in noble metals relative to Ni and Cu, and rocks forming from it would have high Ni and Cu to noble-metal ratios.

S.-J. Barnes, Norges geologiske undersøkelse, Postboks 3006, N-7001 Trondheim, Norway. Present address, Sciences de la Terre, Universite du Qw!bec, 555 Boulevard de I'Universite, Chicoutimi, G7H 2Bl, Canada.

The recent boom in the exploration for new plati­ deformation in the area (Boyd in prep.) at a num deposits has produced numerous Os, Ir, Ru, depth equivalent to 2-3 kb, and was subsequently Rh, Pt, Pd and Au (hereafterreferred to as noble deformed and metamorphosed along with its metals) analyses. Exploration programs rarely country rocks at 6-10kb and 600-700°C (Barnes tindplatinum deposits, but the noble-metal analy­ 1986). Together with the host rocks the intrusion ses that are gathered can be useful in providing was tectonically emplaced into its present position information on the geochemical cycle of these during the Scandian phase ( 400 Ma.) of the Cale­ elements in different geological environments. donian Orogeny as part of the tectonic unit ter­ This information can be used in developing med the Upper Allochthon (Boyd in prep.). models as predicative tools for exploration (e.g. Mineral and whole-rock analyses suggest that Keays et al. 1982; Naldrett et al. 1984). the parental magma of the intrusion was a high­ Rocks from the Råna Layered Intrusion of Mg tholeiite, which was slightly enriched in the northem Norway have exceptionally low Ir, Pt, highly incompatible elements such as Cs, Rb, K, Pd and Au contents relative to their Ni and Cu Ba, Hf, Zr and LREE relative to moderately contents. This paper will: (a) Document the Ni, incompatible elements such as Se, Y and HREE Cu and noble-metal concentrations of the Råna (La/LaN = 2.5) (Bames 1986 and Boyd in prep.). Layered Intrusion; and (b) consider bow the com­ The exa et tectonic setting of the Råna intrusion bined effects of partial melting, sulphide seg­ is unclear, but on the basis of the clinopyroxene regation and silicate crystallization could have analyses and using Leterrier's et al. (1982) dis­ produced rocks with such unusual compositions. crirnination diagrams, the intrusion formed in 'a distension zone' (Bames 1986). This includes ocean ridges, both in major oceans and back­ are basins, ocean islands and passive continental Geological setting margins. The spinel analyses indicate affinities The Råna Layered lntrusion lies 20 km southwest with abyssal tholeiites (Bames 1986). The fact of the town of Narvik in northem Norway (inset that the Råna intrusion formed at a depth equiv­ on Fig. 1). The intrusion belongs to the dass alent to 2-3 kb in graywackes suggests that it did of layered mafic intrusions intimately associated not form at an ocean ridge or an ocean island in with orogenesis, e.g. the 'Younger' Gabbros of a major ocean. Possibly it formed at an ocean NE Scotland, La Perouse in Alaska, and the ridge in a back-are basin that was choked with Caledonide gabbros of Maine. It intruded gray­ sediment, or at a passive continental margin. wackes and pelites of the Narvik Group after The intrusion may be divided into three zones the first and second phases of Caledonide (Boyd in prep; Bames 1986); a lower ultramafic NORSK GEOLOGISK TIDSSKRIFT 67 (1987) 216 S. -J. Barnes

SHOW ING

PLAGIOCLASE-BRONZITE CUMULATE �- t.:...:�:�l OUARTZ - BEARING PLAGIOCLASE- BRONZ ITE OR 180� PLAGIOCLASE-OLIVINE CUMULATES � - OLIVINE CUMULATES OR BRONZITE CUMULATES 2km c=J KYANITE-GARNET-BIOTITE SCHIS T � THRUST CONTACT

Fig_ l. Simplified geological map of the Råna Layered lntrusion; modifiedafter Boyd & Mathiesen (1979) to show the Iocations of Eiterdalen, Rånbogen and Tverrfjell.

zone of olivine and bronzite cumulates, a middle Sciences Associates of the U .K. The noble metals maficzone of plagioclase--olivineand plagioclase­ were determined by instrumental neutron acti­ bronzite cumulates, and an upper quartz-bearing vation (INAA) after preconcentration of the zone of plagioclase-bronzite cumulates with inter­ noble metals into a nickle sulphide bead; the stitial quartz. Rocks fromfour localities at Råna­ Tverrfjell samples were analyzed at the University Bruvann, Eiterdalen, Rånbogen, and Tverrfjell­ of Toronto, by the author; the other samples were (Fig. l) have been analysed for the noble metals, analyzed at Becquerel Laboratories of Toronto. Ni, Cu and Co. Rocks from Rånbogen and Bru­ Co was determined by INAA at Becquerel Lab­ vann are in the ultramafic zone and consist of oratories of Toronto. Table l lists estimates of olivine and bronzite cumulates containing sul­ the precision and accuracy of the analyses. phides. Bruvann is Norway's largest nickel deposit, and the results for this locality are pre­ sented separately (Boyd & Mathiesen 1979; Boyd et al. in press). The rocks from Eiterdalen come Results from the contact betwe�n mafic zone and the The Ni, Cu, Co, S and Au contents of these country rocks and are metagabbros containing rocks are greater than the detection levels. The sulphides. The rocks from Tverrfjell come from distribution of these elements is log-normal; three traverses across the upper portion of the therefore, in addition to the arithmetic means and ultramafic zone and most of the mafic zone. the ranges, the geometric means, which may be more representative, are presented in Table 2. Ru and Rh are present at less than detection level in all the rocks and will not be discussed further. Analytical methods Some samples from each locality have Os, Ir, Pt

Ni, Cu and S were determined by XRF, on and Pd contents at levels less than detection level. pressed-powder pellets, by Midlands Earth Therefore, the mean for each element cannot NORSK GEOLOGISK TIDSSKRIFT 67 (1987) Ni & Cu to noble-metal ratios 217

Table I. Precision and accuracy of analyses.

NGU std Accepted Accepted Accepted XRF MESA n=4 l sigrna value1 NIM-D value2 BCRl value2

Cu 52.5 1.5 53.4 12 lO 16 20 Ni 51 1.9 51.9 2064 2050 11 lO 38 5.4 103 367 s 200? 400 INAA Becquerel Co 27.9 25

INAA UNIVERSITY OF TORONTO SARM-7 l Sigrna Accepted n=4 Value3 Os ppb 73 17 63 Ir 69 3 74 Ru 423 20 430 Rh 253 40 240 Pt 4093 358 3740 Pd 1495 68 1530 Au 356 17 310

1 NGU's accepted value (Faye pers. comm. 1985). 2 Govindaraju ( 1984). 3 Steele et al. (1975).

Table 2. Average chalcophile and siderophile element content of rocks from the Råna layered intrusion.

Tverrfjell n = 115 Rånbogen n =7

Geometric Arithmetic Geometri c Arithmetic mean mean Range mean mean Range

Ni ppm 350 443 36-1615 1800 4557 600-19,900 Cu 90 110 3-320 450 730 138-2383 Co 60 73 22-154 240 485 140-1910 1000 1441 0-6959 23,000 s 86,300 3504-405,816 os• ppb <2 <2 3 4 <3-12 rr• 0.05 0.06 <0.01-Ø.35 0.8 l <0.3-5 Ru

Eiterdalen n =7

Geometric Arithmetic mean mean Range

Ni ppm 2500 3447 340-7730 C u 900 961 11-3846 Co 250 337 75-716 40,000 64,600 413-163,488 s Os* ppb <3 <3 rr• 0.6 0.7 <0.3-1.4 Ru

• Some sarnplescontained Os, Ir, Pt and Pd at less than detection levels; therefore the means for these elements were estimated using log-probability graph paper. Bury (1975) p. 279 eq. 8.8. NORSK GEOLOGISK TIDSSKRIFT (1987) 218 S. -J. Barnes 67

L. ._. - PARTIAL CHROMITE "U \ a.. MELTING TREND � • REMOVAL

OLIVINE REMOVAL • ·;..:_. . OLIVINE - 6 ADDITION 10 CHROMITE 5 ADDITION 10 4 10 3 10 INTRUSIONS 2 10 1 10 ° HIGH HgO BASALTS 10 � 1 CALC-ALKALINE 10- � OCEAN FLOOR BASALTS

Fig. 2. The fields of Ni/Cu vs Pd/Ir for various rocks from the sources listed in the appendix. R = mean of the Rånbogen samples,

E = mean of the Eiterdalen samples, T = mean of the Tverrfjell samples. Note that the mean of the Eiterdalen and Tverrfjell samples falls within the field of high-MgO basalts, which is consistent with the development of these rocks from a high-MgO tholeiite as suggested by their whole-rock chemistry. The Rånbogen samples fall on the olivine and chromite enriched field, which is consistent with the fact that these rocks are olivine and pyroxene cumulates with minor chromite.

be calculated directly. Fortunately the geometric necessarily valid for all rock types. Firstly, it mean may be estimated by means of log-prob­ assumes that all the noble elements partitioned ability paper and the arithmetic mean calculated into a Cu-Fe-Ni sulphide liquid containing 36- from well-established statistical formulas (Bury 38% sulphur in the igneous phase of the rock's 1975: 279 eq. 8.8). This technique was used for history and secondly, that since then the S, Ni, estimating the geometric and arithmetic means Cu and noble-metal contents of the rock have not for Os, Ir, Pt and Pd at each locality. been disturbed. This may be true in sulphide-rich The Råna rocks contain sulphides; because sul­ rocks under certain metamorphic conditions. In phides usually contain high concentrations of sulphide-poor rocks, however, �h, Pt and Pd may noble metals, it is common practice to recalculate occur as compounds of As, Sb, Se, Hg and S, noble-metal, Ni and Cu values to 100% sulphides while Os, Ir and Au are often found in metallic (e.g. Naldrett 1981). However, it should be form. The noble metals may not have partitioned remembered that recalculation to 100% sulphides into a Cu-Fe-Ni sulphide liquid. Furthermore, involves a number of assumptions which are not during sea-ftoor alteration and subsequent meta- NORSK GEOLOGISK TIDSSKRIFT 67 (1987) Ni & Cu to noble-metal ratios 219

OLIVINE SULPHIDE REMOVAL "C Ø CALC- ALKALINE a.. ÅDDITION l' - 'l � OCEAN FLOOR z • >(, • CHROMITE REMOVAL BASALTS CHROMITE "-.. OLIVINE REMO AL ADDITION V Ei)_ ,., ,... .- ")'"' ' _",...... / . '\· ...... · . /' ... _. / / -·-·- ·-· ·-. i .� : / OPH/OLITES Cu-RICH SUL PH/DE VE/NS

Cu/lr

Fig. 3. Cu/Ir vs Ni/Pd for various rocks from the sources Iisted in the appendix. R = mean of the Rånbogen samples, E = mean of the Eiterdalen samples,T = mean of the Tverrfjell samples. Note that all the Råna rocks have high Cu/Ir and Ni/Pd ratios relative to extrusive rocks, which suggests that sulphides were removed from the magma prior to the formation of these rocks.

morphism sulphur may be mobile and hence the Råna rocks (R, E and T) with the other rock either be enriched or depleted relative to its types in Fig. 2 (the sources for the fields outlined igneous values. In order to avoid the potential in Figs. 2 and 3 are listed in the appendix and errors inherent in recalculating to 100% the rationale behind the diagrams is outlined in sulphides, this work will consider Ni, Cu, Co and Barnes et al. submitted). The similarity in Ni/Cu the noble metals in terms of whole-rock values ratio between high-MgO basalts and the Råna and ratios. The reader is referred to Boyd et al. rocks is consistent with the whole-rock and min­ (in press) for a study which considers the Bru vann eral geochemistry, which suggests that the Råna sulphide deposit at Råna, in terms of 100% sul­ magma was high-MgO tholeiite. phides for the sulphide normalized approach to The Pd/Ir ratio of the Rånbogen samples is treating noble-metal data. lower than the Pd/Ir ratio of the Eiterdalen The Ni/Cu ratios for the three Råna localities samples, which in turn is lower than the Pd/Ir vary from 3.5 to 6.5, which is lower than those ratio of the Tverrfjell samples (Table 2). The found in primitive rocks such as komatiites, higher Eiterdalen and Tverrfjell samples have Ni/Cu than the Ni/Cu ratio found in evolved rocks such and Pd/Ir ratios similar to high-MgO basalts and as ftoodbasalts, and similar to Ni/Cu ratios found consequently on a plot of Pd/Ir versus Ni/Cu plot in high-MgO basalts. Compare the positions of in the high-MgO basalt field (E and T in Fig. 2). 220 S. -1. Barnes NORSK GEOLOGISK TIDSSKRIFf 67 (1987)

The Rånbogen samples have similar Ni/Cu ratios and sulphide liquid is lower than that of the noble to high-MgO basalts but lower Pd/Ir rati os; there­ metals and thus when sulphides segregate from a fore, on the Pd/Ir versus Ni/Cu plot, the Rån­ magma the sulphides tend to have lower Ni and bogen samples plot below the high-MgO basalts, Cu to noble-metal ratios than the melt from which but still within the field of layered intrusions. they segregated. Consequently any rock enriched The differences in the composition of the rocks in these sulphides will have lower Ni/Pd and Cu/ from the three localities may be explained by Ir ratios than the melt from which it formed. olivine and chromite crystallization. Both chro­ However, the decrease in Ni/Pd and Cu/Ir ratios mite and olivine tend to concentrate Ir over Pd in the sulphides is also dependent on the amount and Ni over Cu. Therefore, the accumulation of of sulphides that segregate (i.e. the R-factor) as chromite and olivine in a rock tends to displace discussed by Keays & Campbell (1981) for Cu the cumulate towards low Pd/Ir ratios and high and Pd and Campbell & Barnes (1984) for Ni and Ni/Cu ratios and the fractionated liquid towards Pt. If a small amount of sulphides segregate the high Pd/Ir ratios and low Ni/Cu ratios, as illus­ Ni/Pd and Cu/Ir ratios in the sulphides are lower trated on Fig. 2 and discussed elsewhere (Bames than that of the melt, but if a large amount of et al. sub). Furthermore, the position of the Rån­ sulphides segregates the Ni/Pd and Cu/Ir ratios bogen samples in Fig. 2 suggests that these rocks of the sulphides approach that of the melt. In any are enriched in olivine and possibly chromite rela­ event the Ni/Pd and Cu/Ir ratios of the sulphides tive to the high-MgO tholeiite from which the never exceed those of the melt from which they · rocks are formed, which is consistent with the formed because the partition coefficientof Ni and petrographic observations that these rocks are Cu into sulphides is lower than that of noble olivine and bronzite cumulates with minor chro­ metals into sulphides. Thus, sulphide accumu­ mite. The relative position of the Tverrfjell and lation lowers the Ni/Pd and Cu/Ir ratios of the Eiterdalen samples suggests either that the liquid rocks in which they accumulate. Since the Råna from which the Tverrfjell samples formed bad rocks have higher Ni/Pd and Cu/Ir ratios than fractionated chromite and/or olivine prior to the their proposed parent magma (high-MgO tholei­ formation of the Tverrfjell rocks, or that the ite) sulp bide accumulation or variations in the Eiterdalen rocks are enriched in olivine and chro­ amount of sulphide that accumulate (i.e. vari­ mite relative to the Tverrfjell samples. The Eiter­ ations in the R-factor) will not account for the dalen rocks consist of plagioclase, bom blende and high Ni/Pd and Cu/Ir ratios in the Råna rocks. sulphides and do not appear to be enriched in Since sulphide accumulation lowers the Ni/Pd olivine and chromite. Therefore the difference in and Cu/Ir ratios, sulphide removal from a silicate Pd/Ir ratio between the Eiterdalen and Tverrfjell liquid should raise the Ni/Pd and Cu/Ir ratios of rocks is attributed to olivine and chromite frac­ the silicate liquid (Campbell & Barnes 1984 make tionation prior to formation of the Tverrfjell a similar point for Ni/Pt). Thus the high Ni/ rocks. Pd and Cu/Ir ratios observed in the Råna rocks The Råna rocks have much higher Ni/Pd and suggest that prior to the emplacement of the Cu/Ir ratios than high-MgO basalts or any other magma at Råna sulphides segregated and were extrusive rock (Fig. 3). The rocks are cumulates, removed from the silicate liquid. These sulphides some of which contain olivine and chromite. The depleted the magma in noble metals relative to Ni and Cu to noble-metal ratios could have been Ni and Cu. The depleted magma was emplaced modified by olivine and chromite fractionation. at Råna and from it the Råna rocks developed. But, as can be seen in Fig. 3, no combination The idea that sulphides have been removed of olivine and chromite fractionation from a from the Råna magma prior to emplacement is common magma type could produce rocks with supported by the low Ni content of the olivines Ni/Pd and Cu/Ir ratios similar to those observed from the intrusion (Fig. 4). Many of the olivines at Rånabogen and Eiterdalen. Some other pro­ are depleted in Ni relative to 'normal layered cess has therefore occurred. Because these rocks intrusions' as defined by Fleet et al. (1977). The contain appreciable amounts of sulphides and removal of sulphides from the Råna magma sulphides are known to concentrate both noble prior to its emplacement would have depleted metals Ni and Cu, the most obvious process to the magma in Ni, and consequently olivines investigate is sulphide accumulation. The par­ that formed from this magma would be Ni tition coefficient of Ni and Cu between silicate depleted. NORSK GEOLOGISK TIDSSKRIFT 67 (1987) Ni & Cu to noble-metal ratios 221

.6

.5

.4

NORMAL LAYERED . 3 INTRUSIONS

NID %

.2

o

. 1

o 30 40 50 60 MGO%

Fig. 4. MgO versus NiO for theTverrfjell olivines. O = oikocryst, C = olivines in coronas, F = free olivines. Most of theTverrfjell olivines are depleted in NiO relative to the field of normal layered intrusions as defined by Fleet et al. (1977), which suggests that sulphides segregated from the magma prior to the development of these olivines. 222 S.-1. Barnes NORSK GEOLOGISK TIDSSKRIFf67 (1987)

Numerical modelling partial melting of the mantle (greater than l100°C), Cu-Fe-Ni or Pd sulphides would melt. On the basis of the comparison of the Ni/Cu and So it could be argued that all the Cu and Pd from Pd/Ir ratios with rocks from the literature the the mantle could be release to the melt, and the Råna rocks appear to have formed from a high­ partial melting equation reduced to C1 = C0/F, MgO tholeiite, and on the basis of the Ni/Pd and i.e. magmas from low degrees of partial melting, Cu/Ir ratios some sulphides appear to have been such as alkali basalts, should contain the highest removed from the magma prior to its emplace­ concentrations of Cu and Pd. However, alkali ment at Råna. The variations in the Ni and Cu to basalts do not contain the highest Pd and Cu noble-metal ratios between the three localities values (Barn es et al. 1985). Thiscould be because, may be explained by olivine and chromite frac­ although all of the sulphides melt, not all the tionation. In the following section these ideas will sulphur can dissolve in the silicate magma. Hence be tested by attempting to numerically model the there is a sulphide liquid present in the residuum processes involved. However, because many of which will act as a host for the chalcophile the parameters in the equations used below are elements and have the same effect as a residual poorly constrained, the model outlined is just one mineral. In order to solve the partial metting possible solution. The purpose of the modelling equation, the amount of sulphide liquid left in the is to illustrate the trends in the Ni and Cu to residuum must be decided. This is governed by noble-metal ratios during partial melting, crystal the equation: fractionation and sulphide segregation. The present Ni and Cu to noble-metal ratios of X,= (Co-(S X F)) X 2.7/(1-F) (2) the rocks is the product of all the previous pro­ X, = the weight percentage sulphides remaining cesses that led to the formation of the rocks. S = the solubility of the sulphur in the magma, The minimum number of processes that can be 2. 7 converts the concentration of S to sulphide, considered would be: partial melting of the assuming 37% sulphur in the sulphides. mantle, followed by intrusion of the resulting In this work S concentration in the mantle is magma in to the crust (during this process the assumed to be 0.035%E (Sun 1982), and the composition of the initial magma may have solubility of sulphur in the magma 0.155% (the changed due to crystal fractionation), and finally mid point between the extremes listed by Wend­ the formation of a cumulate and crystallization. landt (1982)). A plot of the degree of partial The effect of each of these processes is examined melting versus the weight percentage sulphides below. in the residue based on equation 2 and these assumptions is shown in Fig. 5. Equation 2 Partial melting assumes that the solubility of sulphur is constant over a range of degrees of partial melting, which The concentration of an element in a melt it is probably not. However, in the absence of any assuming batch partial melting is governed by the equation data to guide us this equation serves as a first approximation. On the basis of a solubility of C1 = C0/(F + D -FD) (1) sulphur of 0.155% at !east 23% partial melting is C1 = concentration of the element in the liquid required for all the sulphides in the mantle to C0 = concentration in the source dissolve (Fig. 5). This calculation is consistent F = the weight fraction of the solid that has with a similar one presented by Morgan (1986). (If melted the minimum and maximum sulphur solubilities D = the bulk partition coefficient between the listed by Wendlandt (1982) are used, 70% and n 14% partial melting respectively is required for restite and liquid = L X;D;, where X; = all the sulphides in the mantle to be dissolved.) i=! For the present calculation at greater than 23% the weight fraction of each mineral present partial melting when D = the concentration of in the residue, D;= the partition coefficient O Cu or Pd (and by analogy, Rh, Pt and Au) in the between the mineral and the melt. liquid = 1/F, i.e. there is a steady decrease in Cu The only minerals that will remain Cu and Pd and Pd as the degree of partial melting increases during partial metting of the mantle would be (Fig. 6A), but the ratio of Cu/Pd remains constant sulphides. At the temperatures achieved during (Fig. 6B). NORSK GEOLOGISK TIDSSKRIFf 67 (1987) Ni & Cu to noble-metal ratios 223

0.100 CONCENTRATION IN THE MELT Ni F VS % SULPHIDE IN RESTITE 2000 p pm

O.o75 1500

0.050

%SUL. 500

0.025

o o o 50 o 10 20 30 40 50 F Fig. 6A. Plot of concentration of Ni, Ir, Pd, Cu in the melt vs 0+------+------+-L-----� degree of partial metting. Cu and Pd are principally held in o 10 20 30 sulphides and hence are released to the mett as the sulphides % PARTIAL MELTING dissolve, after all the sulphides have dissolved these elements are diluted by further partial metting by the volume effect. Ni Fig. 5. Degree of partial metting vs the % sulphides in the restite. Assuming 350 ppm sulphur in the mantle (Sun 1982) and Ir in contrast are compatible with the residue and hence and a sulphur solubility of 1550 ppm in the mett (mid-point of their concentrations increase with the degree of partial metting. solubilities given by Wendlandt (1982)), 23 % partial metting is required before all the sulphides are released. } F vs MELT RATIOS Ni/lr 1500 Cu/lrNi/Pd 150 150 At less than 23% partial melting the situation Cu/Pd 30 is more complex because the concentrations of Cu and Pd in the melt are proportional to both ! 1/F and 1/D (eq. 1). A partition coefficient of Ni/lr Cu/lr 1000 1000 for noble metals between silicate liquid and Ni/Pd 100 sulphides has been assumed for the present cal­ Cu/Pd 10020 culation. Estimates for the partition coefficientof Ni/Cu 0·02 noble metals between silicate liquid and sulphide X 103 vary from 100 (Ross & Keays 1979) to 100,000 ! Ni/lr 500 (Campbell & Bames 1984). A partition coefficient Cu/lr 50 of 1000 has been used in this work because, if Ni/Pd 50 higher values are used, practically no noble metals Cu/Pd 10 are released to the magma as long as any sulphides Ni/Cu 0·01 remained in the mantle. In other words, magmas formed from less than 23% partial melting of the mantle, such as continental fiood basalts, should o contain

Table 3. Parameters used for the partial melting modelling. Partition coefficients*

Residue Sulphides Olivine Other silicates Degree sul Minerals % ol wt % sil Ni C u Co PGE Ni Cu Co Pd Ir Ni Cu Co Pd Ir

l 70 0.0913 30 250 250 50 1000 16.8 o o 5 o o o 5 5 70 0.077 30 250 250 50 1000 16.8 o o 5 o o o 5 lO 70 0.058 30 200 200 50 1000 11.6 o o 5 o o o 5 15 70 0.037 30 150 150 50 1000 7.5 o o 5 o o o 5 20 70 0.0083 30 100 100 50 1000 5.4 o o 5 o o o 5 25 70 30 3 o o 5 o o o 5

• Naldrett & Duke (1980); except Ir into olivine (see text for discussion). NORSK GEOLOGISK TIDSSKRIFT 67 (1987) Ni & Cu to noble-metal ratios 225 partial melting of a fertile mantle (Table 3). The F vs CUMULATE RATIOS Råna rocks have base- to noble-element ratios one to two orders of magnitude higher than these. 1.00

Cumulate processes The rocks at Råna are cumulates containing sul­ 0.75 phides. The simplest assumption is that if partial Tv melting alone cannot account for the Ni and Cu z to noble-metal ratios of the Råna rocks, then w partial melting followed by cumulate devel­ L..(!)o a.. opment and crystallization may account for the ::J composition of the rocks. Therefore the effect of (.) 0.50 the cumulate process on the Ni and Cu to noble­ metal ratios must be considered. In this work z ...... Rayleigh fractionation rather than batch seg­ regation will be used to model the cumulate w (!) 0.25 process, because batch segregation has been dealt z < with exhaustively in the form of the 'R-factor' I u in various publications (e.g. Keays & Campbell 1981). In any event, both batch segregation and Rayleigh fractionation will produce the same 0+------r------� trends although with different absolute values for 1 10 100 the composition of the cumulate. The con­ % CRYSTALLIZATION centration of an element in the average cumulate Fig. 7. Plot of variations in (Cu or Ni)/noble-element ratios in during Rayleigh fractionation is governed by the the cumulates with the degree of crystallization of the melt. equation: Tv = curve, assuming 0.39% sulphides in the cumulate, the amount present in the average Tverrfjell sample, Et = curve assuming 17% sulphides in the cumulate, the amount present

in the average Eiterdalen sample, Rå = curve assuming 23%

Cc = concentration of the element m the sulphides in the cumulate as there is in the average Rånbogen cumulate sample.

C1 = concentration of the element in the liquid

F = the weight fraction liquid remaining n

= average cumulate D LX;D; as in equation l. the with degree of crystal­ i= l lization assuming 23%, 17.5% and 0.39% sul­ phides in the cumulate (these are the percentages From this equation it can be seen that the con­ of sulphides present in average Rånbogen, Eiter­ centration of an element in the cumulate is con­ dalen and Tverrfjell samples respectively). Note trolled both by the degree of crystallization (1 - that although at two of the localities in the cumu­ F) (i.e. '!ow much of the magma has crystallized) lates contain large amounts of sulphide, the Ray­ and by D, the bulk partition coefficient between leigh model does not require that the liquid the magma and the cumulate. Noble metals and segregates these sulphides from equal volumes of Ni and Cu have extremely high partition coef­ magma and cumulate. Consider for instance the ficients from silicate magma into sulphides, so D Rånbogen locality, which contains 23% sul­ is strongly influenced by how much sulphide is phides. If it is assumed that sulphide saturation present in the cumulate. Further, the partition takes place when the magma contains 0.26% sul­ coefficients for the noble metals into sulphides phur (maximum estimate of sulphur solubility are much greater than that of Ni and Cu. Conse­ listed by Wendlandt 1982), then approximately quently the Ni and Cu to noble-metal ratios of the 0.70% sulphide liquid may form within the cumulate changes depending on the percentage magma. This sulphide might then accumulate at sulphides present in the cumulate. Fig. 7 shows various places in the magma chamber, in struc­ the change in Ni and Cu to noble-metal ratios for tural traps or by gravity settling, to form a rock NORSK GEOLOGISK TIDSSKRIFT 67 (1987) 226 S. -J. Barnes containing more than 0.7% sulphide. The pres­ factor of approximately 0.9 (assurningF < 0.9) to ence of the rock containing 23% sulphides 6.5 x 103 by the formation of cumulates similar requires only that the volume of magma from to those observed at Eiterdalen and Rånbogen. which the sulphide was collected be at least 33 The Cu/Pd ratios of rocks from Eiterdalen and times the volume of the present cumulate (23/ Rånbogen are 84 x 103 and 91 x 103, an order of 0.7), i.e. that 97% of the magma has not crystal­ magnitude higher than either partial melting or lized. Fig. 7 indicates that the change in Ni and crystallization alone would produce. At Tverrfjell Cu to noble-metal ratios increases rapidly from the observed Cu/Pd ratio is 11 x 103, while the l% to 10% crystallization and then levels out calculated product of partial melting and sulphide at approximately 0.95 for the two localities that fractionation is 7.2 x 103 x 0.359= 2.58 x 103 contain large percentages of sulphides. In contrast (assuming 50% crystallization based on whole­ the change in Ni and Cu to noble-metal ratios is rock geochemistry, Barnes 1986). Sirnilar cal­ fairly constant from l% to 10% crystallization at culations on the other Ni and Cu to noble-metal about 0.2 and then increased rapidly to 0.74 at ratios reveal the same result; namely that cumu­ 90% crystallization for the low percentage of sul­ late processes decrease the Ni and Cu to noble­ phides present at Tverrfjell. metal ratios of the cumulate, which is the exact If only partial melting and cumulate processes opposite effect on the Ni and Cu to noble-metal have affectedthe Ni and Cu to noble-metal ratios ratios to that needed for explaining the ratios (nc/nm) at Råna, then in the Råna rocks. This result is fairly easy to understand, because noble metals have higher nc/nm(observed)= nc/nm(partial melting) partition coefficients into sulphides than Ni and x nc/nm(cumulate process) Cu, and therefore sulphides must have lower Ni For the sake of developing a model it has been and Cu to noble-metal ratios than the magmas assumed that the Ni and Cu ratios of the Råna from which they segregate. Comequently any magma were originally dose to that of a high­ rock containing these sulphides will have lower MgO basalt (see Table 4). The author does not Ni and Cu to noble-melting ratios than the magma wish to imply that the Råna magma ratios were from which they form. Thus partial melting fol­ exactly these, but they will be used to illustrate lowed by crystallization cannot alone account for the point. The assumed initial magma Cu/Pd is the Ni and Cu to noble-metal ratios observed at 7.2 x 103 and the Cu/Pd ratio is changed by a Råna.

Table 4. Model composition of the Råna rocks. Ni Ir Description ppm Cu Co ppb Pt Pd Au Initial mett, based on a 20% partial melt of mantle 628 135 99 1.17 26 18.7 3.7 Melt after removal of 0.3% sulphides 401 86 85 0.058 1.3 0.93 0.18 Model of Eiterdalen consisting of 17.5% sulphides, 82.5% liquid, F = 0.90 3765 807 540 0.58 13 9.3 2 Model of Rånbogen consisting of 23% sulphides, 77% olivine, F = 0.90 3952 837 616 0.58 13 9.3 1.8 Melt after removal of 15% olivine and 0.15% sulphide 249 125 91 0.136 6.8 4.9 0.96 Model ofTverrfjell consisting of 0.39% sulphides, 30% liquid, 20% olivine and 50% plagioclase and pyroxene 429 141 102 0.27 13 9.3 1.8 NORSK GEOLOGISK TIDSSKRIFT 67 (1987) Ni & Cu to noble-metal ratios 227

Sulphide segregation % SULPHIDES vs MELT RATIOS As outlined above sulphide accumulation 1000 decreases the Ni and Cu to noble-metal ratios in cumulates and the Ni and Cu to noble-metal ratios of rocks from Råna require a process that increases the Ni and Cu to noble-metal ratios. This leads to the suggestion, despite the presence of sulphides in the Råna cumulates, that they formed from a silicate liquid from which the sul­ z 100 phide liquid had been removed prior to its w emplacement at Råna. O._o a.. During Rayleigh fractionation the concen­ '::J tration of an element in the melt is governed by u the equation:

z (4) ...... 10 w Fig. 8 shows the change in Ni and Cu to noble­ o metal ratio in the magma as sulphides are z

100 not both conclusions can be correct, so some other process must have disturbed the Ir relative to the other elements at Tverrfjell. 50 The most obvious difference between the Tverrfjell locality and the other two is that at Tverrfjell the rocks appear to have formed from Cc/Cl a more fractionated magma than at the other two localities, e.g. the highest forsterite content observed in the olivines at Tverrfjell is 87% while 10 at Rånbogen it is 90%. Ir tends to behave in a compatible fashion during crystallization. If 15% olivine were removed from the initial magma ( containing approximately 18% MgO) prior to its emplacement at Tverrfjell, the MgO content of the magma would fall to 12% and consequently the forsterite content of the olivine would fall to approximately 87%. The Ir might possibly be removed at the same time by chromite frac­ W 00 �F M � � � � tionation. A satisfactory match between the Fig. 9. Plot of concentration of the elements in the cumulate divided concentration of the element in the liquid vs the weight observed and modelled rocks at Tverrfjell can be fraction liquid remaining (F). obtained by assuming 0.15% sulphide and 15% olivine removal prior to emplacement at Tverrfjell Followed by formation of a cumulate with the mineral proportions presently observed to Pd in the magma= 11.4/0.93 (Tables 2 and 4). in the rocks (Table 4). This implies an F of 0.92 (Et PGE in Fig. 9). U sed in a similar manner, Ni and Cu give a slightly lower estimate of F of 0.9 (Et Ni and Cu in Fig. 9). The results of the modelling are listed in Table Summary 4. The segregation of small amounts (0.15-0.4%) of Similar calculations at Rånbogen indicate that sulphides prior to emplacement of the magma at approximately 0.3% sulphides were removed Råna may account for the Råna Layered Intrusion prior to emplacement of the magma at this having such unusually high Ni and Cu to noble­ locality. The rocks are principally olivine cumu­ metal ratios. This conclusion is supported by the lates and in modelling the cumulate allowance olivine compositions observed at Råna; most of was made for the presence of the olivine which the olivines are depleted in Ni relative to normal increases the Ni content of the rocks. A satis­ layered intrusions (Fig. 4). If some sulphides had factory match was obtained between the observed segregated from the magma prior to its emplace­ and model rocks (Table 4). ment at Råna, the magma would have been At Tverrfjell the situation is more complicated. depleted in Ni and any olivines that formed in The degree of crystallization has a marked effect equilibrium with this magma would also be Ni on the Ni and Cu to noble-element ratios (Fig. depleted. 7). On the basis of whole-rock geochemistry dis­ The unusual Ni and Ci to noble-metal ratios cussed elsewhere (Barnes 1986), approximately and the variations into Ni and Cu and noble-metal 50% crystallization has occurred. The change in concentration between the different localities at Ni and Cu to noble-metal ratios due to cumulate Råna may be modelled as shown schematically in processes is therefore approximately 0.35 (Fig. Fig. 10 and Table 4. Approximately 20% partial 7). On the basis of this assumption the Cu/Pd and melting of the mantle took place to produce a Ni/Pd ratios indicate that approximately 0.15% MgO-rich tholeiite which intruded into the lower sulphides were removed from the magma prior to crust. Some of this magma intruded in the Eiter­ emplacement at Tverrfjell (Fig.8). In con trast the dalen and Rånbogen areas; en route to these Cu/Ir and Ni/lr ratios indicate approximately localities approximately 0.3% sulphides segre­ 0.5% sulphides have been removed. Obviously, gated. An olivine-sulphide cumulate consisting NORSK GEOLOGISK TIDSSKRIFf 67 (1987) Ni & Cu to noble-metal ratios 229

FLOW DIAGRAM SHOWJNG THE DEVELOPMENT OF THE RÅNA Appendix: sources for Figs. 2 and 3 (contd) JNTRUSJON SUL PH!DES Mantle Rock type Source Location

South USA spinel lherzolite Jagoutz (1979) Ronda spinel lherzolite Stockman (1982) Ronda harzburgite Stockman (1982) West USA harzburgite dts1 Govindaraju (1984) West USA harzburgite pcc1 Govindaraju (1984) Thetford harzburgite Oshin & Crocket (1982)

Komatiites 15'1• MgO Kam bal da olivine spinifex Keays et al. (1982) THOLEIITE Kam balda dunite Keays et al. (1982) Kam balda sulphide-rich Keays et al. (1982) peridotite 20'1• PARTIAL �LTING Mt Edward sulphide-rich Naldrett (1981) Of" THE peridotite MANTLE Lunnon sulphide-rich Naldrett (1981) peridotite Fig. JO. Model for the development of the Rånbogen, Eiterdalen Munro olivine spinifex Crocket MacRae and Tverrfjell samples. & (1986) Munro B-zone Crocket & MacRae (1986) Langmuir sulphide-rich Green & N aldret! of 23% sulphide and 77% olivine developed at peridotite (1981) Texmont B-zone sulphide Barnes & Naldrett Rånbogen and a cumulate containing 17% �ul­ rich (1987) phides developed at Eiterdalen. When modelling Dundonald B-zone sulphide Barnes & Naldrett the Tverrfjellrocks, in addition to the removal of rich (1987) sulphides, olivine and possible chromite have to Hart B-zone sulphide Barnes & Naldrett rich (1987) be removed from the magma prior to emplace­ Pipe sulphide-rich Naldrett (1981) ment at Råna. For the sake of completeness a peridotite model for the Bruvann Ni sulphide is also shown Hi tura sulphide-rich Papunen (1986) in Fig. 10; the data for this were taken from Boyd peridotite et al. (in press); the removal of 0.5% sulphides prior to emplacement of the magma at Bruvann High-MgO Lavas Katinq komatiitic basalt Barnes et al. (1982) could account for the unusual composition of the Donaldson komatiitic basalt Barnes et al. (1982) sulphides at this locality. Fred's flow komatiitic basalt Crocket & MacRae (1986) Acknow/edgements.- This investigation was funded by LKAB, Theo's Row high-MgO Crocket & MacRae NGU and by a postdoctoral fellowship from NTNF. I thank K. tholeiite (1986) S. Heier and R. Boyd for initiating the project; E. W. Sawyer Australia high-MgO basalt Redman & Keays and T. A. Karlsen without whom no samples could have been (1985) collected and no map produced; T. Boassen of IKU for assist­ Kanichee high-MgO Naldrett (1981) ance with the microprobe analyses; A. Hemming and E. W. tholeiite Sawyer for dratting the diagrams. A special thanks to A. Kor­ neliussen and to the NTNF committee for their hospitality. Ca/c-a/kaline Rocks West USA andesite, AGV-1 Govindaraju (1984) Australia basalt Redman & Keays Appendix: sources for Figs. 2 and 3 (1985)

Mantle Rock type Source Ocean floor basalts Location Jane de Fuca basalt 010 Crocket & Teruta (1977)? Troodos olivine basalt Becker Lesotho garnet lherzolite Mitchell & Keays & (1981) Agiorgitis (1978) DSDP-37 olivine basalt Crocket Teruta World spinel lherzolite Mitchell & Keays & (1981) (1977) 230 S. -J. Barnes NORSK GEOLOGISK TIDSSKRIFr 67 (1987)

Appendix: sources for Figs. 2 and 3 (contd) Appendix: sources for Figs. 2 and 3 (contd)

Mantle Rock type Source Mantle Rock type Source Location Location

Thetford basalt Oshin & Crocket Sudbury sulphide veins Hoffman et al. (1986) (1979) Intrusive portions of ophiolites Donaldson sulphide veins Dillon-Leitch et al. Thetford ultramafic to Oshin & Crocket (1986) mafic rocks (1982) Kambalda sulphide veins Lesher & Keays Ronda gabbro Stockman (1982) (1984) Oman chromitite Page et al. (1982) Unst chromitite Gunn et al. (1985)

Flood basalt related West USA basalt BCR-1 Govindaraju (1984) References West USA diabase W-1 Stockman (1982) Barnes, Sarah-Jane 1986: Investigation of the potential of the Duluth sulphide-rich Naldrett (1981) Tverrfjell portion of the Råna intrusion for platinum group gabbro element mineralization. Norges geologiske undersøkelse, Great Lakes sulphide·rich Naldrett (1981) rapport 169 pp. gabbro 82.201, Barnes, Sarah-Jane & Naldrett, A. J. 1987: Fractionation of lnsizwa sulphide-rich Lightfoot et al. the platinum-group elements and gold in some komatiites of gabbro (1984) the Abitibi Greenstone Belt, northern Ontario. Economic Talnakh sulphide·rich Nal drett (1981) Geology 165-183. gabbro 82, Barnes,Sarah-Jane, Naldrett, A. J. & Gorton, M. P. 1985:The Layered Intrusions origin of the fractionation of platinum group elements in Bush veid chills Sharpe (1982) terrestial magmas. Chemical Geology 53, 303-323. Jimberlana peridotites to Keays & Campbell Barnes, S. J. & Naldrett, A. J. 1985: Geochemistry of the J-M gabbros (1981) reef of the Stillwater complex, Minneapolis Adit Area. I Montcalm gabbro sulphide- Naldrett (1981) Sulfide chemistry and sulfide-olivine equilibrium. Economic rich Geology 80, 627-645. Ken bridge gabbro sulphide· Naldrett (1981) Barnes, S. J., Coates, C. A. J. & Naldrett, A. J. 1982: Petro­ rich genesis of Proterozoic nickel sulphide-komatiite association. Sudbury gabbro sulphide· Naldrett (1981) The Katiniq Still, Ungava, Quebec. Economic Geology 77, rich 413-429. Delta peridotites to Clark et al (1985) Boyd, R. (in prep.) The geology and petrology of the Råna gabbros nickeliferous Layered Intrusion, Nordland. Norges geol­ Siika-Kam gabbro sulphide- Papunen (1986) ogiske undersøkelse rapport. rich Boyd, R. & Mathiesen, C. O. 1979: The nickel mineralization Kotalahi gabbro sulphide- Papunen (1986) of the Råna mafic intrusion, Nordland, Norway. Canadian rich Mineralogist 17, 287-298. Laukunkan gabbro sulphide· Papunen (1986) Boyd, R., McDade, J., Millard, H.T. & Page, N. (in press).: rich Platinum and palladium geochemistry of the Bruvann nickel­ Vammala gabbro sulphide· Papunen (1986) copper deposit, Råna, North Norway. Norsk Geologisk rich Tidsskrift. Koillisma gabbro sulphide· Lahtinen (1985) Bury, K. V. 1975: Statistica/ models in applied sciences. Wiley, rich New York. 625 pp. Espedalen gabbro sulphide- Naldrett (1981) Campbell, I. H. & Barnes, S. J. 1984: A model for the geo­ rich chemistry of the platinum group elements in magmatic sulfide Vakker lei gabbro sulphide- Thompson et al. deposits. Canadian Mineralogist 22, 151-160. rich (1980) Clark, T., Belander, M. & Giovenazzo, D. 1985: Precious metals in New Quebec. Ministere de l'Engergie et des Res. Platinum reefs Quebec. Document de promotion. 16 pp. Stillwater JM-reef Barnes & N aldrett Crocket, J. H. 1981: Geochemistry of the platinum-group (1984) elements. Canadian Institute Mining and Metallurgy, Special Busveld UG-2 reef McLaren & de Issue 23, 47-64. Villiers (1982) Crocket,J. H. &MacRae, W. E. 1986: Platinum-group element Gain (1985) distribution in komatiitic and tholeiitic volcanic rocks from Bush veid MR reef Steele et al. (1975) Munro Township, Ontario. Economic Geology 81, 1242- Lac des Iles Roby Zone Naldrett (1981) 1251. Crocket, J. H. &Teruta, Y. 1977: Palladium, iridium and gold Cu-rich sulphides contents of mafic and ultramafic rocks drilled from the Mid­ Rathburn Lake sulphide veins Rowell & Edgar Atlantic Ridge, leg 37, Deep Sea Drilling Project. Canadian (1986) Journal Earth Sciences 14, 777-784. NORSK GEOLOGISK TIDSSKRIFT 67 (1987) Ni & Cu to noble-metal ratios 231

Dillon-Leitch, H. C. H., Watkinson, D. H. & Coats, C. 1986: bution of gold, palladium and iridium in some spinel and Distribution of platinum-group elements in the Donaldson gamet lherzolites. Implications for the nature and origin of West Deposit, ·Cape Smith Fold Belt, Quebec. Economic precious metal-rich intergranular components in the upper Geology 81, 1147-1159. mantle. Geochimica Cosmochimica Acta 45, 2425-2442. Fleet, M. E., MacRae, N. D., Herzberg, C. T. 1977: Partition Morgan, J. 1986: Ultramafic xenoliths: dues to earth's late of nickel between olivine and sulfide. Contributions Min­ accretionary history. Journal Geophysica/ Research 91 Bl2,

eralogy & Petro/ogy 65, 191-197. 12375-12387 o Gain, S. B. 1985: The geological setting of the platiniferous Naldrett, A. J. 1981: Platinum-group element deposits. UG-2 chromitite layer on the farm Maandagshhoek, Eastem Canadian Institute Mining and Metallurgy, Special Issue 23, Bushveld. Economic Geology 80, 925-943. 197-232. Govindaraju, K. 1984: 19 84 compilation of working values and Naldrett, A. J. & Barnes, S.-J. 1987: The fractionation of sample description for 170 international reference samples of the platinum-group elements with special reference to the mainly silicate rocks and minerals. Geostandards News/euers composition of sulphide o res. Fortschift Mineralogy Petro/ogy VIII. Special Issue July. 64. Green, A. H. & Naldrett, A. J. 1981: The Langmuir volcanic Naldrett, A. J. & Duke, J. M. 1980: Platinum-group metals in peridotite-associated nickel deposits. Canadian equivalent of magmatitic sulfide ores. Science 208, 1417-1428. the Western Australian occurrences. Economic Geology 76, Naldrett, A. J., Duke, J. M., Lightfoot, P. C. & Thompson, J. 1503-1523. F. H. 1984: Quantitative modelling of the segregation of Grønlie, A. (in press). Platinum-group minerals in the Lil­ magmatic sulphides: An exploration guide. Canadian Institute lefjellklumpen nickel-copper deposit, Norway Norsk Geol­ Mining Metal. Bull. 1-11. ogisk Tidsskrift. Oshin, I. O. & Crocket, J. H. 1982: Noble metals in the Thetford Gunn, A. G., Leake, R. C. & Styles, M. T. 1985: Platinum­ mines ophiolite Quebec, Canada. Part I. Distribution of group element mineralization in the Unst ophiolite, Shetland. gold, iridium, platinum and palladium in the ultramafic and Mineral Reconnaissance Programme Rep. British Geo/. Surv. gabbroic rocks. Economic Geology 77, 1556-1570. 73, 117 pp. Oshin, I. O. & Crocket, J. H. 1986: Noble metals in the Thetford Hoffman, E. L., Naldrett, A. J., Van Loon, J. C. & Hancock, mines ophiolite Quebec, Canada. Part Il. Distribution of R. G. V. 1979: The noble-metal content of ore in the Levack gold, silver, iridium, platinum and palladium in the Lac de West and Little Stobie mines, Ontario. Canadian Mineralogist I'Est volcano-sedimentary section. Economic Geology 81, 17, 437-451. 931-945. Jagoutz, E., Palme, H., Baddenhausen, H., Blum, K., Dreibus, Page, N. J., Cassard, D. &Haffty, J. 1982: Palladium, platinum, G., Spette), B., Lorenz, V. & Wanke, H. 1979: The abun­ rhodium, ruthenium and iridium in chromitites from the dance of major, minor and trace elements in the earth's Massif du Sud and Tiebaghi Massif, New Caledonia. Econ­ mantle as derived from primitive ultramafic nodules. Proc. omic Geo/ogy 77, 1571-1577. 10th, Lunar Planetary Science Conference, 2031-2050. Papunen, H. 1986: Platinum-group elements in Svecokarelian Keays, R. R. & Campbell, I. H. 1981: Precious metals in the nickel-copper deposits Finland. Economic Geology 81, 1236- Jimberlana Intrusion, Western Australia: Implications for 1241. genesis of platiniferous ores in layered intrusions. Economic Redman, B. A. & Kayes, R. R. 1985: Archaean basic volcanism Geology 76, 1118-1141. in the eastem goldfields province, Yilgam Block, Western Keays, R. R., Nickel, E. H., Groves, D. l. & McGoldrick, P. Australia. Precambrian Research 30, 113-152. J. 1982: Iridium and palladium as discriminants of volcanic­ Ross, J. R. & Keays, R. R. 1979: Precious metals in volcanic exhalative hydrothermal and magmatic nickel sulfide miner­ nickel sulphide deposits in western Australia I. Relationship alization. Economic Geology 77, 1535-1547. with the composition of the ores and their host rocks. Lahtinen, J. 1985: PGE-bearing copper-nickel occurrences in Canadian Minera/ogist 17, 417-435. the marginal series of the early Proterozoic Koillismaa Lay­ Rowell, W. F. & Edgar, A. D. 19 86: Platinum-group miner­ ered lntrusion, northern Finland. Geologica/ Survey Finland alization in a hydrothermal Cu-Ni sulfide occurrence, Bulletin 333, 165-179. Rathburn Lake, northeastern Ontario. Economic Geo/ogy Lesher, C. M. & Keays, R. R. 1984: Metamorphically and 81, 1272-1277. hydrothermally mobilized Fe-Ni-Cu sulphides at Kambalda, Sharpe, M. 1982: Noble metals in the margin rocks of the Western Australia. InBuchanan, D. L. &Jones, M. J. (eds.): Bushveld Complex. Economic Geo/ogy 77, 1286-1295. Sulphide deposits in mafic and ultramafic rocks, London, Steele, T. W., Levin, J. & Copelowitz, I. 1975: The preparation Institute of Mining Metallurgy, 62-69. and certification of a reference sample of a precious metal Leterrier, J., Maury, R. C., Thonon, P., Girard, D. & Marchal, ore. National Institute of Metallurgy Report 1696. M. 1982: Clinopyroxene compositions as a method of identi­ Stockman, H. W. 1982: Noble metals in the Ronda and Jose­ fication of the magmatic affinities of paleo-volcanic series. phine peridotites. Unpubl., Ph.D., thesis, Mass. Inst. Tech. Earth Planetary Science Letters 59, 139-154. 236 pp. Lightfoot, P. C., Naldrett, A. J. & Hawkesworth, C. J. 1984: Sun, Shen-Su, 1982: Chemical composition and origin of the The geology and geochemistry of the Waterfall Gorge section earth's primitive mantle. Geochimica Cosmochimica Acta 46, of the Insizwa Complex withparticular reference to the origin 179-192. of the nickel-sulfide deposits. Economic Geology 79, 1857- Thompson, J. F. H., Nixon, F. & Siverten, R. 1980: The geology o 1879. of Vakkerlien nickel prospect Kvikne, Norway. Geo/ogica/ McLaren, C. H. & De Villers, J. P. R. 1982: The platinum­ Survey Finland Bulletin 52, 3-22. group chemistry and mineralogy of the UG-2 chromite layer Wendlandt, R. F. 1982: Sulphur saturation of basalt and andes­ of the Bush veid Complex. Economic Geo/ogy 77, 1348-1366. ile melts at high pressures and temperatures. American Min­ Mitchell, R. H. & Keays, R. R. 1981: Abundance and distri- eralogist 67, 877 885.