INTRODUCTION ANALYTICAL CHEMISTRY: THE CONTAMINATION PROBLEM ABSTRACT

Transcription

1 0. EAST PACIFIC RISE, GALAPAGS SPREADING CENTER AND SIQEIRS FRACTRE ZNE, DEEP SEA DRILLING PRJECT LEG 54: HGRMAGMAPHILE ELEMENTS A CMPARISN ITH THE NRTH ATLANTIC J. L. Joron 2, L. Briqueu, H. Bougault 4, and M. Treuil 2 ABSTRACT salt samples collected on DSDP Leg 54 are fresh, with some showing extensive Fe and enrichment. s from many sites were contaminated with,, and, to a lesser extent,, because they were powdered using a tungsten carbide shatter box onboard ship. /, /, and (for uncontaminated samples) / ratios are similar to those of North Atlantic basalts and chondritic meteorites; these ratios are probably characteristic of the primordial mantle. e / and / ratios indicate that the Leg 54 eastern Pacific basalts were derived from "depleted" mantle similar to the source of North Atlantic basalts drilled on DSDP Leg (22 N; Miocene crust) and Legs 5 through 5 (25 N; etaceous crust). INTRDCTIN DSDP Leg 54 was the first IPD leg in the North Pacific dealing with the oceanic crust. Several sites were drilled: the western flank of the East Pacific Rise (EPR) at 9 N (Figure ), Sites 0, (.5 m.y. old), (.2 m.y. old), 2, 8 (2 m.y. old), and 9 (4.5 m.y. old); the Galapagos spreading center at 0 N, 86, four holes at Site 4, and Site 5 (62 km north of the Galapagos spreading center axis); and the Siqueiros fracture zone, Site 7 (8 N, 5 m.y. old). s from this drilling program should give us a picture of the variability of basalts produced in this area and a basis of comparison with results obtained in the North Atlantic. ur investigations focused on /, /, and / ratios, shown to be constant for DSDP and dredged samples in the North Atlantic, and on the / ratio. In the North Atlantic, on the basis of / ratios, we can identify two main mantle sources for the basalts of the oceanic crust. e first source is characterized by a / ratio equal to 8, which represents the socalled "typical MRB tholeiites." salts drilled at Sites 95 and 96 on either side of the MidAtlantic Ridge during Legs and 46 have this ratio. e second source is characterized by a / ratio equal to 9 and represents the basalts of the oceanic islands and seamounts and also of parts of the ridges near these islands. salts drilled and dredged in the FAMS area at 6 N, at ntribution no. 50 du Departement ientifique, Centre céanologique de Bretagne. 2 Institut de Physique du Globe, boratoire de Geochimie, 4, place Jussieu, 7520 Paris Cedex 05, France. niversité des iences et Techniques du nguedoc, Place gene taillon, 00 Montpellier, France. 4 Centre céanologique de Bretagne, B.P., 2927 Brest Cedex, France. N, and near Iceland, during Leg, have this ratio (see Figure 2). e also deal with a contamination problem caused by powdering the samples in a tungsten carbide shatter box onboard the Glomar Challenger. ALTICAL CHEMISTR: THE CNTAMITIN PRBLEM Analytical procedures (Xray fluorescence [] both onboard ship and on shore and neutron activation analysis) have been described previously (Bougault, 7; and Bougault et al., 8). As on several previous legs, we have systematically analyzed trace elements in samples selected onboard ship for shipboard analysis. nfortunately, on Leg 54 the oil tank of the generator of the unit burned out. is represented the first major breakdown of our shipboard equipment since we started analysis onboard the Glomar Challenger (Leg ) and, consequently, most of our majoroxide results were obtained on shore. In addition, the agate mortar was cracked, pitted, and was therefore unusable. nsequently, we prepared the powders onboard ship using the shipboard tungsten carbide shatter box. e knew in advance that this procedure would lead to cobalt contamination but did not know the extent of the problem. Although some results may not be suitable for geochemical interpretation, we measured the same elements as on previous cruises, to determine what other elements (if any) were involved in the contamination. Hole 7 samples (ble b) provided a good opportunity to ascertain which elements were contaminated by use of the tungsten carbide shatter box, as some specimens were ground in this manner, others by means of the agate mortar. e former evinced a high tungsten concentration. e confirmed the anticipated contamination of cobalt: values as high as 80 ppm in contaminated samples occur, whereas the average in uncontaminated samples is about 46 ppm, a typical value in oceanic 725

2 to CΛ 0 N 9 N PT4SITE (49, 6' SIEIRS FRACTRE ZNE 04' Figure. Sites drilled on Deep Sea Drilling Project Leg 54. ntours in meters.

3 HGRMAGMAPHILE ELEMENTS Leg 95&95A* inner floor fracture zone A ^intersection Figure 2. (ppm) versus (ppm) for basalts drilled and dredged on Leg. Symbols for this and Figures through 8 are as follows: : Hole 0 ; + : Hole ; : Hole 2 ; Δ : Hole ; D : Hole 4 and 4A;M:Holes4nd4C; * :Hole5; ir : Hole 7; : Holes 8 and 8 A ; and : Hole 9 A. tholeiites. rrespondingly, very high tantalum concentrations ( to 5 ppm) occur in contaminated samples, whereas the average contents in uncontaminated samples is 0. ppm. obium is also subject to contamination but to a lesser extent. alues of about 9. ppm occur in contaminated samples, whereas the average value in uncontaminated samples is 7.2 ppm. e other elements investigated,,,,,,,,, and showed no evidence of contamination. ese observations are confirmed on the one hand by samples of Holes 0 through 5 (bles lb, 2b, b, 4b, and 2b), all of them ground with the tungsten carbide shatter box, and on the other hand by the samples of Holes 8 and 9 (bles 5a, 6b, and 7b) which were ground with the agate mortar. us, it appears that the tungsten carbide shatter box used on Leg 54 causes considerable contamination of cobalt and tantalum, and, to a lesser extent, niobium. is evaluation is, of course, made by comparison with the concentrations of uncontaminated samples presently studied, which are ~ ppm, ~ 0. ppm, and ~ 7 ppm. ntamination of and results from impurities in the tungsten, that of cobalt because this metal is used as a binder for the tungsten carbide. RESLTS e tables are arranged not by hole or site numbers but according to drilling area. bles through 7 are concerned with the western flank of the EPR, Holes 0,,, 2, 8, 8A, and 9, respectively. ble 8 through 2 relate to the Galapagos spreading center Holes 4, 4A, 4B, 4C, and 5, respectively. ble relates to the Siqueiros transform fault (Hole 7). In the first place, all of the samples have low loss on ignition, and are considered reasonably fresh. nly one significant instance was found, in Hole 8 (bles 6a and 6b), of alkali metals being sensitive to alteration effects (Bougault et al., 8; Joron et al., in press). From the oxide analyses in 7 samples it appears that the basalts of this hole are homogeneous. Alkalimetal concentrations are low ( < 0.2%, < ppm, and < ppm), except for Sections 8A4, No. 9 and 8A2, No. 4 ( : 0. and %, : 6.7 and.9 ppm, and : 0.28 and ppm, respectively). is enrichment results from a moderate alteration effect, comparable to alteration observed in Hole 95 of Leg (Bougault et al., 8). is variation is by no means to be interpreted as the result of magmatic processes. ith respect to both major oxide and trace elements, despite the number of holes drilled on the western flank of the EPR, the range of basaltic types is rather limited. Choosing 2 as an indicator, three types of hygromagmaphile element increases can be distinguished: Type (Holes 8, 8A, and 2) with 2 about. per cent; Type 2 (Section, No. of Hole 2, Holes 0,, and ) with 2 about 2 per cent and Type (first two samples of Hole ) with 2 about 2.5 per cent. ithin Type, four samples (Sections 9, 9, 94, and 95 in Hole 2; ble 2a) have higher A 2 (5.8% instead of 4.2%) and (74 ppm instead of ppm) than other samples within this group. Enrichment in plagioclase and clinopyroxene phenocrysts could account for the variations observed, but,,, and are the same as in other samples of the group. and are enriched, and is depleted (and remains somewhat problematic). ithin Type 2, some significant differences can be observed; that is, Section, No. of Hole 2 and Hole 0 have 2 contents of about.92 per cent, Hole 2 contents of about 2.05 per cent, and Hole 2 contents of about 2. per cent. ese differences are confirmed by other elements (Figures, 4, and 5). salts having at least two different compositions were obtained in Hole 9 (4.5 m.y. old); they appear to be similar to Type basalts. From the Galapagos rift, samples recovered in Holes 4, 4A, 4B, and 4C appear to be identical. ey are characterized by low A 2 values (2.6%), high values (5.%), and very low strontium concentrations ( ppm) as compared with typical values of oceanic tholeiites. 2 is about per cent. Two main groups are present in Hole 5 drilled at 62 km north of the spreading center: they have 2 values of 0.98 per cent and.9 per cent (one other sample has 2 =.%). A feature in common with the Site 4 samples is the low value of strontium (50 ppm). From the data available, this low strontium concentration seems to be a characteristic feature of the Galapagos rift. 727

4 J. L. JRN, L. BRIQE, H. BGALT, M. TREIL TABLE la Hole 0 (EPR, 9 N, ~.5 m.y. old): Major xides AI2 Mn Mg Ca K2 P25 LI 4,08 5 (No. 4),20 7, TABLE lb Hole 0 (EPR, 9 N, ~.5 m.y. old): Trace Elements Ct l 4,08 Section 5, (No. 4),20 7, TABLE 2a Hole (EPR, 9 N, ~.5 m.y. old): Major xides Si 2 AI2 Mn Mg Ca θ2 P25 LI 2,57, 0 (No. 5),7275 (No. ), 720, 48 4,47 (No. ) 4,26 BIT 0 (No. 2) TABLE 2b Hole (EPR, 9 N, ~.5 m.y. old): Trace Elements 2,57,0 (No. 5),7275 (No. ), 720 (No. 7), 48 4,47 (No. ) 4, 26 BIT 0 (No. 2) ? Hole 7 which was drilled in the Siqueiros fracture zone consists of one single homogeneous unit. Although the A 2 content is rather low (.%), concentrations (~ 25 ppm) correspond to a typical value in oceanic tholeiites. ese samples are similar to Type basalts, as defined by the two uppermost samples of Hole on the western flank of the EPR. HGRMAGMAPHILE ELEMENTS In all samples recovered by drilling in the Atlantic cean (as well as those recovered during the FAMS operation), for each pair,, and, one element has been shown not to fractionate with respect to the other (Bougault et al., in press; Joron et al., in press). us, the ratios of these elements are constant or nearly constant for each pair, and close to chondritic ratios. Because of the contamination problem, and data are available only for two EPR sites (Holes 8, 8A, and 9A) and for the Siqueiros fracture zone (Hole 7). e parallel lines on Figure 6 show a major portion of the range of North Atlantic data (absolute values go up to 7.8 ppm for and 9 ppm for at N); data of Leg 54 plot on the same line. All and data can be used since no contamination has been observed for these elements. As with, all the data of Leg 54 plot in the range defined by the samples of the Atlantic cean (Figure ). n the diagram (Figure 4) the data plot on a line the slope of which is a little higher than the average slope of the Atlantic. Since this difference can result from a small systematic difference of 2 to ppm in or 0. to 0.2 ppm in, we do not state that the Leg 54 line is distinct from the North Atlantic line. 728

5 HGRMAGMAPHILE ELEMENTS TABLE a Hole 2 (EPR, 9 N, ~ 2 m.y. old): Major xides Si 2 AI2 θ Mn Mg Ca 2 P25 LI 5, 226 (No. 5) 7,0 7,2025 7, 8992 (No. 7) 7, 22 Section 72, (No. ) 85, 226 (No. 5) 85,25 (No. 4) 9,707 (No. 4c) 92, 58 (No. 6) 9, 994 (No. 6c) 94,4 (No. b) 95, 7275 (No. 6), 925 (No. ) 0,2? TABLE b Hole 2 (EPR, 9 N, ~ 2 m.y. old): Trace Elements XRl XRl XR 7,0 7, 2025 Section 8992, (No. 7c) 7, 22 Section 72, (No. ) 85, 226 (No. 5) 85, 25 (No. 4) 9, 707 (No. 4c) 92^ 58 (No. 6) 9,994 (No. 6c) 94,4 (No. b) 95, 7275 (No. 6), 925 (No. ) ; TABLE 4a Hole (EPR, 9 N, ~.2 m.y. old): Major xides Si 2 A 2 Mn Mg Ca 2 P2θ5 LI 5 CC, (No. 4),284 (No. 5) Section 7, (No. ) 7, 47 (No. 7) 8, (No. 6) TABLE 4b Hole (EPR, 9 N, ~.2 m.y. old): Trace Elements \ 5 CC, (No. 4), 284 (No. 5) Section 7, (No. lb) 7, 47 (No. 7) 8, (No. 6) Figure 5 is a plot of versus ; only slight fractionation is observed between these elements, confirming the hygromagmaphile behavior of (Bougault, 7b). e slope found is. 0~ 2, very close to the value obtained at 22 N and 6 N in the Atlantic: (Bougault et al., 8). ese results confirm that there is little or no fractionation between such pairs of elements as,, and. us, we conclude that the ratios are distinctive of the mantle prior to differentiation. Since these ratios are the same, or nearly the same, in the Atlantic and Pacific oceans, and in addition are similar to chondritic ratios, it is very probable that they are characteristic of the earth itself. ith respect to two hygromagmaphile elements whose partition coefficients are different by an order of magnitude, for instance and, it has been shown (Bougault et al., in press,2) that the / ratio may vary widely. Being very similar to the Sm/ ratio used by Shilling (), it is a function of mantle differentia 729

6 J. L. JRN, L. BRIQE, H. BGALT, M. TREIL TABLE 5a Hole 8 (EPR, 9 N, ~ 2 m.y. old): Major xides Si 2 AI2 Mn Mg Ca 2 P25 LI 54 ; 8 (No. 4) 54, 7880 (No. 9) 62, (No. ) TABLE 5b Hole 8 (EPR, 9 N, ~ 2 m.y. old): Trace Elements XRl Ci Λ 54, 8 (No. 4) 54, 7880 (No. 9) 62, (No. ) ' TABLE 6a Hole 8A (EPR, 9 N, ~ 2 m.y. old): Major xides Si 2 A 2 Mn Mg Ca 2 P25 LI, 22 (No. ), 2729 (No. 4),9294 (No. 0), 2022 (No. ) 2,5 (No. 6), 7274 (No. b) 4, 6770 (No. 9) 2, (No. 4), (No. 2a) 4, 226 (No. 4a), 0709 (No. 8) 5, 5 (No. 9) 52, 889 (No. 5e) 5, 8 (No. ) 54, 8486 (No. 6), 525 (No. 8) 7,96 (No. 6e) TABLE 6b Hole 8A (EPR, 9 N, ~ 2 m.y. old): Trace Elements, 22 (No. ), 2729 (No. 4), 9294 (No. 0), 2022 (No. ) 2,5 (No. 6), 7274 (No. b) 4, 6770 (No. 9) 2,5 (No. 4), (No. 2a) 4, 226 (No. 4a), 0709 (No. 8) 5, 5 (No. 9) 52,889 (No. 5e) 5, 8 (No. ) 54, 8486 (No. 6),525 (No. 8) 7, 86 (No. 6e) D ( J n n 4 tion and partial melting and can vary within a single hole (ood et al., in press). For this reason, we believe that this ratio by itself is not sufficient indication of the heterogeneity of different sources in the mantle. Figure 7 shows where the samples of Leg 54 plot in comparison with the data obtained in the North Atlantic. All the Leg 54 samples fall in what has so far been considered to be a "depleted" field. But again, this is not sufficient reason to deduce that these basalts were produced from a "depleted" mantle. A good example is given by some samples of Hole 9 (North Atlantic) which show a "depleted" character (high / and Sm/); but 70

7 HGRMAGMAPHILE ELEMENTS TABLE 7a Hole 9A (EPR, 9 N, ~ 4.5 m.y. old): Major xides Si 2 A 2 Mn Mg Ca 2?2θ 5 LI, (No. ) 2, 8789 (No. 2) Section, (No. 7b) 2, 8 (No. ) 22, 46 (No. 2) Section, (No. ) TABLE 7b Hole 9A (EPR, 9 N, 4.5 m.y. old): Trace Elements, (No. ) 2,8789 (No. 2) 22, 46 (No.2), 69 (No. ) Section, (No. 7b) :<i: <:f, TABLE 8a Hole 4 (Galapagos Spreading Center, 0 N 86 ): Major xides Si 2 Al 2 θ Mn Mg Ca 2 P25 LI 5, 94 (No. 2) 5, 5 (No. 5) 5, 6 (No. ) 52,56 (No. 2b) 5, 202 (No. 2) 54, 25 (No. ), 26 (No. 5), 07 (No. 0c) 7, TABLE 8b Hole 4 (Galapagos Spreading Center, 0 N 86 ): Trace Elements Sal Intcrv 5, 94 (No. 2) 5, 5 (No. 5) 5, 6 (No. ) 52,56 (No. 2b) 5, 202 (No. 2) ( , 25 (No. ). 26 (No. 5) , 07 (No. 0c) , " ii. in when we consider / (discussed below), these basalts are clearly derived from the same "undepleted" source as the other portion of Hole 9. s from the North Atlantic at 6 N, N, and 6 N all lie on a single line on a diagram, the range of variation being 0.2 to 7.8 ppm for, and.5 to 50 ppm for. Seeing that this wide range of concentrations produces no change in the ratio between the two elements signifies that does not fractionate with respect to. But for samples from 22 N (young crust) and 25 N (old crust), a different line is found; ranges from to ppm, and from.0 to 4. ppm (Figure 8). is difference in slope and ratio of is attributed to a different property of the mantle between 22 N and 25 N on the one hand and 6 N, N, and 6 N on the other. It reflects either different values for / in the mantle sources or different partition coefficient ratios for these elements in the two areas. Leg 54 data plot very nearly on the trend defined at 22 N to 25 N in the Atlantic. ey would thus correspond to a mantle "depleted" in lanthanum. Since the /, /, and / ratios are constant in both the Atlantic and Pacific oceans, we deduce that these ratios are characteristic of the primordial mantle. If the two fundamental / values which have been found are confirmed by further studies, it may be possible to conclude that sources in the mantle are not completely uniform in composition and that 7

8 J. L. JRN, L. BRIQE, H. BGALT, M. TREIL TABLE 9a Hole 4A (Galapagos Spreading Center, 0 N 86 ): Major xides Si 2 A 2 Mn Mg Ca 2 P25 LI 4,5 (No. ) 4, 47 (No. 6), 9095 a Sediment: hydrothermal deposit. TABLE 9b Hole 4A (Galapagos Spreading Center, 0 N 86 ): Trace Elements D 4, 5 (No. ) 4, 47 (No. 6) TABLE 0a Hole 5B (Galapagos Spreading Center, 0 N 86 ): Major xides Si 2 A 2 Mn Mg Ca 2 P25 LI 5, 04 (No. 2) 5, 48 (No. 6) 5,554 (No. 7) 5, 6770 (No. 9) 5, 59 (No. 5) 52, 6 (No. ) 52, 5256 (No. 6), 05 (No. 5), 7580 (No. ), 020 (No. 6), 8 (No. 9) , TABLE 0b Hole 5B (Galapagos Spreading Center, 0 N 86 ): Trace Elements L 5, 04 (No. 2) 5, 48 (No. 6) 5,554 (No. 7) 5, 6770 (No. 9) 52, 6 (No. ) 52, 5256 (No. 6), 05 (No. 5), 7580 (No. ), 020 (No. 6) they are not differentiated from the primordial mantle in the same way. e different values found for such ratios as / and Sm/ are a consequence of both heterogeneous mantle sources and variable partial melting processes. CNCLSIN Grinding samples in a tungsten carbide mortar introduces considerable contamination of and, and, to a lesser extent,. /, /, and / ratios have the same single values close to chondrites as in the North Atlantic and are probably characteristic of a primordial mantle. e / ratio, in agreement with the / ratio, shows that Leg 54 basalts derive from a socalled depleted mantle. e / ratio is close to the value obtained at 22 N and 25 N in the Atlantic but different from that at 6 N, N, and 6 N in the North Atlantic ("undepleted" mantle). REFERENCES Bougault H., 7a. Major elements: analytical chemistry on board and preliminary results. In Aumento, F., Melson,. G., et al., Initial Reports of the Deep Sea Drilling Project, v. : ashington (.S. Government Printing ffice), p. 52., 7b. First transition series elements: fractional crystallization and partial melting. In Aumento, F., Melson,. G., et al., Initial Reports of the Deep Sea Drilling Proj 72

9 HGRMAGMAPHILE ELEMENTS TABLE lla Hole 4C (Galapagos Spreading Center, 0 N 86 ): Major xides S2 AI2 Mn Mg Ca K2 2 P25 LI 2, 85 (No. 2) Section 2, (No. 7), 6 (No. 2), 47 (No. 4) TABLE lib Hole 4C (Galapagos Spreading Center, 0 N 86 ): Trace Elements Lu u 2, 85 (No. 2) Section 2, (No. 7), 6 (No. 2), 47 (No. 4) TABLE 2a Hole 5 (Galapagos Spreading Center, 60 km from the axis): Major xides Siθ2 A 2 Fe; Mn Mg Ca K2 P2< ^5 LI 7,47 (No. 7) ,885 (No. ) ,46 (No. ) ,5254 (No. 7a) , 5559 (No. 7a) ,9095 (No. 8b) ,05 (No. 5) 2.0 8, 050 (No. 2) ,2028 (No. 5) , 59 (No. 9) , 00 (No. 5) ,687 (No. 7) , 5 (No. ) , 62 (No. 9) TABLE 2b Hole 5 (Galapagos Spreading Center, 60 km from the axis): Trace Elements C o 7,^7 (No. 7) 7, 885 (No. ) 7, 46 (No. ) 72, 5254 (No. 7a) 72, 9095 (No. 8b) 8, 050 (No. 2) 9, 2028 (No. 5) 9, 59 (No. 9) 9, 00 (No. 5) 92, 687 (No. 7) 9, 5 (No. ) 9, 68 (No. 9) o ect, v. : ashington (.S. Government Printing ffice), p Bougault H., Treuil M., and Joron J. L., 8. Trace elements in basalts from 22 N and 6 N in the Atlantic ocean: fractional crystallization, partial melting, and heterogeneity of the upper mantle. In Melson,. G., Rabinowitz, P. D., et al., Initial Reports of the Deep Sea Drilling Project, v. : ashington (.S. Government Printing ffice), p Bougault H., Cambon P., rre., Joron J. L., and Treuil M., in press. Evidence for variability of magmatic processes and upper mantle heterogeneity in the axial region of the midatlantic ridge near 22 N and 6 N. Tectonophysics. Bougault H., Joron J. L., Treuil M., in press. Alteration, fractional crystallization, partial melting, mantle properties from trace elements data recovered in the North Atlantic. 2nd Ewing Symposium, montdoherty Geological bservatory, 8. Joron J. L., Bollinger C, Quisefit J. P., Bougault H., and Treuil M., in press. Trace elements in basalts at 25 N old crust, in the Atlantic cean: Alteration, mantle and magmatic processes. In Donnelly, T., Francheteau, J., Bryan, 7

10 J. L. JRN, L. BRIQE, H. BGALT, M. TREIL TABLE a Hole 7 (Siqueiros fracture zone, 8 N 'v 5 m.y. old): Major xides Siθ2 A 2 Mn Mg Ca 2 P25 LI 9, 8 (No. 4b) 92, 5 (No. ) 9, 9 (No. 2) 94, 59 (No. 8a) 95, 8689 (No. 7) 0, 67 (No. 4a), 92 (No. la), 4 (No. 4) 0, 000 (No. 8) 0, 48 (No. 9c) 04, 8 (No. Ik) 04, 4 (No. le) 05, 8284 (No. 4b) TABLE b Hole 7 (Siqueiros fracture zone, 8 N >v 5 m.y. old): Trace Elements XR XRl G XRl XRl Λ Λ liu ' 9, 8 (No. 4b) 92, 5 (No. ) 9, 9 (No. 2) 94, 59 (No. 8a) 95, 8689 (No. 7) 0,67 (No. 4a),92 (No. la), 4 (No. 4) 0, 000 (No. 8) 0, 48 (No. 9c) 04, 8 (No. Ik) 04.4 (No. 0c) (No. 4b) IS n // / / / / A~ë,A y +// / 20 / / Figure. (ppm) versus (ppm) : the two parallel lines define the range where the North Atlantic data plot. Symbols as for Figure 2.., Robinson, P., Flower, M., Salisbury, M., et al., Initial Reports of the Deep Sea Drilling Project, v. 5, 52, 5, Part 2: ashington (.S. Government Printing ffice). hilling, J. G.,. Icelandic mantle plume: geochemical study of Reykjanes Ridge, Nature, v. 2, p Figure 4. (ppm) versus (ppm) : the two lines define the range where the North Atlantic data plot. Symbols as for Figure 2. ood D. A., aret J., Bougault H., rre., Joron J. L., Treuil M., Bizouard H., Norry M. J., Hawkesworth C. J., and Roddick J. C, 9. e petrology, geochemistry and mineralogy of North Atlantic basalts: a discussion based on IPD Leg. In Luyendyk, B. P., Cann, J. R., et. λ.jnitial Reports of the Deep Sea Drilling Project, v. : ashington (.S. Government Printing ffice), p

11 HGRMAGMAPHILE ELEMENTS i f / + + / / / /+ / α / / / Figure 7. (ppm) versus (ppm) : the different segments are related to the North Atlantic data ; one single hole can plot on two of these lines (see text). Symbols as for Figure (x0 ) Figure 5. (ppm) versus (ppm) : symbols as for Figure 2. e two lines are a tentative envelope. T 0 A 20 / 0 Figure 6. (ppm) versus (ppm) : the two parallel lines define the range where the North Atlantic data plot, üf is specifically related to 25 N old crust in the Atlantic (Legs 5 and 52). Figure 8. (ppm) versus (ppm) : ranges where the North Atlantic (N.A.) data plot. Symbols as for Figure 2. 75