Batiza, R., Storms, M.A., and Allan, J.F. (Eds.), 1995Proceedings of the Ocean Drilling Program, Scientific Results, Vol. 142

10. DATA REPORT: CHEMICAL ANALYSES OF BASALTIC ROCKS:AN INTERLABORATORY COMPARISON1

Kurt Boström2 and Wolfgang Bach3

ABSTRACT

Samples of MORB basalts obtained during Ocean Drilling Program Leg 142 were supplied as unknown materials to fourdifferent land-based laboratories and analyzed by inductively coupled plasma-optical emission spectrometry (ICP-AES) and byX-ray fluorescence (XRF), partly to test the quality of the ship's XRF facility and partly to compare data by ICP-AES and ashore-based XRF laboratory.

The comparisons show that contamination of samples on the ship due to grinding in alumina mills is negligible, and that theship's XRF laboratory produces good data for all major elements except phosphorus and for several trace components. How-ever, some trace elements such as Ba are difficult to determine well by XRF at concentrations frequently found in MORBbasalts.

INTRODUCTION SAMPLE DESCRIPTIONS AND PREPARATIONS

A major purpose of Leg 142 of the Ocean Drilling Program(ODP) was to study alteration processes in mid-oceanic ridge basalts(MORBs) and other rocks that could explain the genesis of hydro-thermal solutions and the subsequent deposition of metal-rich activeridge sediments, which have been known since 1963 (Skornyakova,1964; Arrhenius and Bonatti, 1965; Boström and Peterson, 1966,1969). These sediments are rich in Fe, Mn, Ba, Cu, and Ni, so it iscritical to understand the geochemical pathways of these elements(Boström et al., 1973). However, successful studies of hydrothermalleaching processes require high-quality analyses (Rona et al., 1980;Arvanitides et al., 1990).

No altered rocks were discovered during Leg 142, but a fair quan-tity of homogenous MORB material was obtained, making it possibleto conduct an interlaboratory comparison of various analytical meth-ods and to test the quality level of the analyses produced in the ship'sX-ray fluorescence spectrometry (XRF) laboratory. XRF is the most-used method for geochemical studies aboard JOIDES Resolution,whereas inductively coupled plasma-optical emission spectrometry(ICP-AES) was the method used by most of the land-based laborato-ries.

We will report on comparisons of shipboard XRF data with re-sults from four laboratories, involving (1) shore-based XRF studies,(2) studies using sequentially reading ICP-AES instruments, and (3)studies using multichannel (simultaneously reading) ICP-AES in-struments. A comparative study involving the use of inductively cou-pled plasma-mass spectrometry for the determination of many tracecomponents and rare earths will be reported elsewhere.

batiza, R., Storms, M.A., and Allan, J.F. (Eds.), 1995. Proc. ODP, Sci. Results,142: College Station, TX (Ocean Drilling Program).

Department of Geology and Geochemistry, Stockholm University, 106 91 Stock-holm, Sweden.

Institut fur Geowissenschaften und Lithospharenforschung, Justus-Liebig Univer-sitat, 6300 Giessen, Federal Republic of Germany. (Present address: Geo ForschungsZentrum Potsdam, Projektbereich 4.2, Telegrafenberg A50, 14473 Potsdam, FederalRepublic of Germany.)

The samples consist of typical MORB rocks collected during Leg142; for descriptions, see Storms, Batiza, et al. (1993). In the follow-ing we will use standard ODP sample notations.

Most samples had been dried, crushed, and ground aboard shipand were therefore not further treated, except for drying at 100°C asrequired. However, samples received as rock pieces had to becrushed. This was done in a plattner mortar made of hard steel to less-en the impact from contaminants. This steel consists of about 99%Fe, 0.5% Mn, and 0.5% C, whereas Ni, Co, Cr, Zr, W, and Mo con-tents are low. The samples were kept in a heavy plastic bag to furtherlessen the contamination risk during this crushing, as was doneaboard ship. Subsequent grinding was done in a Retch Mikroschnell-mühle, which has a cylindrical grinding well and pestle of agate.

One sample from Section 142-846A-1M, about 400 g, consistedof sand- and pebble-sized basalt pieces heavily contaminated by drill-ing oils, fragments of drill bits, and other metallic matter. Repeatedwashings in an ultrasonic bath with organic solvents (methanol, ace-tone, and xylole), and the removal of magnetic particles using a handmagnet produced a cleaner fraction. This material was then wetsieved, producing a fraction larger than 2 mm in diameter. This +2-mm fraction was handpicked under a binocular microscope to deletecontaminants not removed by the previous treatment. Simultaneous-ly, the +2-mm material was divided into a very clean-looking, largelyglassy fraction, and a fraction less glassy and fine grained. The latterfraction contained a few thin specks (about 1 mm in diameter or less)of reddish brown and white matter, possibly representing ferric hy-droxide and opaline silica.

These +2-mm fractions were subsequently rinsed for 15 minutesat 50°C in dilute hydrochloric acid (1 part concentrated HC1 + 3 partsdistilled water), to which reducing hydroxylamoniumchloride hadbeen added. This treatment did not etch the uncontaminated rock andglass surfaces, but removed all reddish brown specks. The sampleswere then rinsed with distilled and deionized water, dried, and furtherground for chemical analysis. The resulting very glassy fraction waslabelled 1M-6 very clean rock, and the less glassy material 1M-6clean rock. Representative subsamples of these fractions were sent toWolfgang Bach, Justus-Liebig University, Giessen (JLUG). The re-maining fraction was kept at Stockholm University (SU).

75

DATA REPORT

One sample presented a special sampling and preparation prob-lem, namely an alumina grinding puck that was broken aboard shipand contaminated Sample 142-846A-1M-2, 0-35cm. This puck wastherefore analyzed to study the possible effects of this contamination.All loose dirt was removed from the puck, after which a 5-mm-thickwafer was cut from the puck, using a petrographic saw. The edges ofthe wafer, which were stained gray, were then removed with a car-bide-tipped chisel. The wafer was easily crushed in a plattner mortarand subdivided in two samples, one for SU and one for JLUG. TheSU sample was further ground in the Mikroschnellmühle. Because ofthe exceptional hardness of alumina, all agate surfaces were some-what scratched, which caused some contamination of the aluminasamples (see below).

ANALYTICAL PROCEDURES AND RESULTS

XRF Studies

The samples at JLUG were treated in the same way as those stud-ied by Bach (1991) and those studies in the ship's laboratory (see"Explanatory Notes" in Storms, Batiza, et al., 1993). These analysesare labeled as JLUG in the tables.

ICP-AES Studies

Samples sent to SU were treated by two methods: (1) a metaboratedigestion technique for the determination of major and some tracecomponents (Sr, Zr, Cr) and (2) an HF-HClO4-digestion method forthe determination of most trace elements (Burman et al., 1977, 1978;Burman and Boström, 1979). The latter method fumes off SiO2 andresults in a solution 10 times richer in trace element than by the me-taborate digestion. The instrument at SU was a 10-year-old ARL3520 sequentially reading ICP-AES unit, which has since been re-placed. These analyses are labeled SU in the tables.

Subsamples were also sent to SGAB in Luleá, northern Sweden(LU), an analytical company that specializes in the determination ofelemental compositions of natural materials. The metaborate diges-tion method used at LU is identical to that at SU. However, at LU theobtained solutions are used for all elemental determinations; that is,no HF-HCIO4 digestion is used for the trace element determinations.Analytical results are labeled as LU in the tables.

Special aliquots of the metaborate-digested solutions at SU weresent to the company Höganàs Eldfast in southern Sweden and ana-lyzed in an ICP-AES unit (model SPECTRO) with 40 fixed channels.The solutions were 10 times more diluted than all other solutionsused for ICP-AES at SU and LU to avoid clogging of nebulizer andtorch systems in the Höganàs spectrometer. These analyses are la-beled SU-H in the tables.

The results of all analyses are given in Tables 1-8.

Accuracy and Precision

To test the performance of our methods we used the geostandardBHVO-1 for tests of accuracy and a MORB sample (G-224) from theEast Pacific Rise at 23°S for reproducibility tests (Table 1); the rec-ommended data for BHVO-1 derive from Govindaraju (1989). Theaccuracy test using BHVO-1 indicated quite satisfactory results forboth SU and LU data; the standard was supplied as an unknown sam-ple. Furthermore, the reproducibility of the G-224 rock data is verysatisfactory. It should be noted that the XRF values in Table 1 arebased on six separately produced glass discs that were analyzed re-peatedly. This avoids the common but erroneous procedure of usingthe same disc repeatedly over the years, which ignores the errors dueto variations in mounting method, etc.; that is, the procedure onlyshows variations in instrument performance. It is well known thatmany or perhaps most errors during analysis are caused by the non-instrumental steps (Flanagan, 1986; Hunt and Wilson, 1986).

Table 1. Comparison of analyses of standard basalts BHVO-1 and G-224.

SiO2

TiO2

A12O3

Fe2O3

MnOMgOCaONa2OK2OP,O 5

TotalCrZrSrBaCoCuNiSeVYYbZnLaGaRb

BHVO-1 SU

49.882.73

13.8812.440.1737.35

11.412.250.490.27

100.8730816539613543

13612433

31729

1.911517

—

BHVO-1 LU

49.002.81

13.9012.200.1737.39

11.402.350.420.31

99.9530216139213645

11711729

31127

29418

——

BHVO-1Rec value

49.942.71

13.8012.230.1687.23

11.402.260.520.27

100.53289179

13945

13612132

31728

2.0105

16——

G-224

Mean

50.361.19

14.9210.740.1788.31

12.742.290.0420.089

100.85373.4

71.974.4

—44.977.382.0

——

29.3_

78.0—

15.61.7

SD

0.170.010.090.070.0040.080.060.020.0050.0030.416.72.81.3—1.95.71.3——0.8—3.2—1.20.6

Notes: All major elements (SiO2-Total) in wt%; all trace elements (Cr-Rb) in ppm. — =not determined. SU = analyses made at Stockholm University, LU = analyses madeat SGAB, Luleá. Rec value = recommended values from literature (see text). G-224represents a fresh MORB basalt from EPR at 23°S and is used as a lab standard atJustus-Liebig University for continuous tests of drift and other instrument perfor-mances. Major components in G-224 have been determined in 83 cases; traces in 36cases.

This long-term test of the MORB G-224 involved 83 complete de-terminations of the major components and 36 of the trace elements(Bach, 1991) and revealed a drift in the XRF instrument of both along-term and a short-term nature, which can be described as a long-wave sinus curve overprinted by a short-wave one. The short-termdrift (expressed as one standard deviation, SD) is fairly constant overextended periods of time, about 0.10% for SiO2,0.05% for Fe2O3, and0.01% for TiO2. This short-term noise scatters about another undulat-ing mean value, which is offset by as much as 0.2% (SiO2), 0.1%(Fe2O3), or 0.02 % (TiO2) after prolonged analytical activity.

At SU several geostandards are routinely used during all analyti-cal work in order to monitor drift, disturbances, and other uncertain-ties, a procedure we have used since 1967. For these reasons we alsoparticipate in collaborative interlaboratory studies to determine thebest data for new geostandards, the last of which involved the AilsaCraig granite and SY-4 (Boström, 1987; Boström, unpubl. data). Ourdata in Table 1 represents the test-run results during the period whenthe analyses reported in Tables 2-8 were performed. We have alsonoted that there may be a long-term drift in the obtained values in ad-dition to the short-term noise. Such long-term noise is particularlyeasy to recognize during determination of SiO2, for which deviationsof 0.2%-0.3% may occur. At SU these problems are partly caused bythe lack of thermal isolation around the spray chamber (sample intro-duction unit), but also can be attributed in part to the curvature in thecalibration curve for SiO2. Curved calibration curves such as the onefor SiO2 for Leg 142 are common for major components (see Storms,Batiza, et al., 1993, "Explanatory Notes" chapter, tables 2 and 3). Asan additional test, some samples were analyzed 3 or 4 times (see Ta-bles 3, 4, 6, and 7); these tests involved repeated metaborate and HF-HC1O4 digestions on fresh aliquots of the rocks in order to incorpo-rate the errors involved in the digestion steps. The resulting analysesshowed standard deviations that were similar to or less than the val-ues reported for the SU data in Table 5. The reproducibility and theaccuracy of the SU analyses are therefore quite good.

A little-considered complication in most analyses of rocks is thatmany results are dependent on the selected recommended values for

76

DATA REPORT

Table 2. Compositions of fresh and contaminated basalts from Hole 142-864A and contaminant (alumina grinding puck).

Sample:

Laboratory:

SiO2TiO2A12O3Fe2O3

MnOMgOCaONa2OK2Op 2o 5TotalSCrZrSrBaCdCoCuNiSeVYYbZnLaGan

142-864A-1M-2,0-35 cm fresh

SU

49.071.65

14.7011.610.1907.51

11.602.720.120.16

99.331140244113128146535727547

33734

3.296

3.7—1

LU

49.301.70

14.5011.600.1957.61

11.602.82BC0.17

99.52

237100127

14—

45677541

32338

4.588

4.5—1

142-864A-1M-2,0-35 cm contaminated

SU

44.701.41

21.3710.160.1706.84

10.652.400.130.15

97.981013242102116195827657242

29730

2.7843.4—1

LU

45.001.49

21.8010.800.1757.17

10.802.54BC0.15

99.94

25291

11318

38587137

28934

4.173

4.0

1

AP

SU

2.9(7)0.01

970.110.0020.200.170.10

0.04

757

782

1

JLUG

0.3

99

0.150.15

2265

501

Notes: All major elements (SiO2-Total) in wt%; all trace elements (S-Ga) in ppm. — =not determined. SU = analyses made at Stockholm University, LU = analyses madeat SGAB, Luleá, JLUG = analysis made at Justus-Liebig University, Giessen. Theapproximative value of A12O3 for the alumina puck (AP) is derived by difference(100% - other constituents); for details, see text. BC = determinations beyond thecalibration curve; n = the number of analyses.

geostandards. Thus, for instance, collaborative studies of the graniteAilsa Craig reveal that the mean SiO2 data obtained by ICP-AES,XRF, and by atomic absorption spectrometry (AA) may differ signif-icantly. A student t-test of the standard errors (that is, the standard de-viation divided by the square root of the number of determinations;

see Crow et al., 1960) suggests that these differences are significant.The mean for the SiO2 data by AA is distinctly smaller than that byICP-AES (-0.45%) and by XRF (Govindaraju, 1987). These poorlyunderstood problems are not exclusively occurring for silica, but doobviously cause a problem in the selection of the proper recommend-ed values (Abbey, 1991); all users of analytical data should be awareof these uncertainties.

Contamination Resulting from Grinding

Table 2 shows the compositions of the alumina puck and uncon-taminated and contaminated rock (Section 142-846A-1M-2). The SUvalues for the contaminant show slightly higher values for SiO2 thanthe JLUG data. This difference is probably caused by contaminationof the SU sample during grinding in an agate mill, but this error is oflittle consequence for the following discussion.

The contaminated rock shows a considerable increase in aluminaand barium, and significant drops in silica, iron, calcium, magne-sium, vanadium, yttrium, ytterbium, zinc, and lanthanum concentra-tions. A simple conservative mixing model shows that thecontaminated rock consists of some 8%-9% of material from thepuck, with 91%-92% fresh rock as the balancing fraction; that is, theconcentrations of the elements silica to lanthanum listed above arelowered because of the admixed alumina.

These contamination effects are easy to spot, particularly for Aland Ba. All basalts from Section 142-846A-1M-2, 0-85 cm, form ahomogeneous group (see Tables 3,4, and particularly 5); even an ad-mixture of a few percent from the puck would cause a significant in-crease in the A12O3 and Ba values. A comparison of mean data fromTables 4 and 5 (Table 4 for materials not processed in alumina grind-ers and Table 5 for rocks mostly processed in alumina grinders) clear-ly indicates that no contamination of significance took place duringgrinding of the rocks during Leg 142, except for Sample 142-846A-1M-2, 0-35 cm.

Variations in Data Between Different Laboratories

The overall agreement between most data in Tables 3-8 rangesfrom satisfactory to excellent. Seven different 142-846A-IM sam-

Table 3. Compositions of bulk samples of Core 142-864A-1M basalts.

Sample:

Laboratory:

SiO 2

TiO2A12O3Fe 2 O 3

MnOMgOCaONa 2OK2OP 2 O 5TotalSCrZrSrBaαiCo

CuNiSe

VYYb

ZnLan

1M-3, 0-35

SU

49.921.67

14.3811.790.1947.53

11.612.720.1470.16

100.121140261108135

156432697550

32935

392

34

1M-3, 0-35

LU

49.701.71

14.5011.700.1977.66

11.802.81

BC0.16

100.29—

225101125

13—

39976441

32637

49241

1M-5, 118-135

SU

49.721.66

14.2011.780.1937.44

11.582.710.1570.16

99.61120261108134

146533697450

33335

394

44

1M-5, 118-135

LU

49.201.68

14.4011.500.1947.57

11.502.83

BC0.16

99.07

22610012213

43787241

32136

489

41

1M-6, 0-75

SU

49.851.66

14.2511.830.1957.49

11.632.700.1470.16

99.911115280109134

156433697350

33334

392

34

1M-6, 0-75

LU

49.301.68

14.4011.500.1937.56

11.602.81

BC0.16

99.22

248100122

14

43647641

32237

4

8541

1M-6, 75-142

SU

49.361.67

14.4111.860.1967.55

11.712.740.1450.16

99.801116264110136

146533707551

33735

393

34

1M-6, 75-142

LU

49.701.70

14.5011.600.1967.64

11.702.86

BC0.16

100.09—

225101124

15

42647442

32538

480

41

Notes: All major elements (SiO2-Total) in wt%; all trace elements (S-La) in ppm, based on number of analyses (n). — = not determined. BC = determinations beyond the calibrationcurve. SU = analyses made at Stockholm University, LU = analyses made at SGAB, Luleá.

77

DATA REPORT

Table 4. Mean data for specially treated large samples of basalts from Section 142-864A-1M-6.

Sample:

Laboratory:

SiO2TiO2A12O3Fe'03MnOMgOCaONa2OK2OP2O5TotalSCrZrSrBaCdCoCuNi

SeVYYbZnLaGaNbRbn

SU

49.831.70

14.9912.180.1977.78

12.172.810.1170.17

101.941210263113135139436697550

33135

395

7———

3

1M-6, 0-150,

LU

49.201.68

14.5011.600.1957.70

11.702.81BC0.17

99.59—

25910012314

48617541

324374

774

—1

very clean rock

JLUG

49.971.61

14.4111.660.197.54

11.692.680.120.15

100.02

257108124

6969

35.6_94

17.42.51.66

SD

0.270.010.070.060.000.040.050.020.010.000.48

311

11

0.3

1

0.60.50.4

SU

49.311.68

15.0012.170.1977.83

12.152.770.1370.17

101.41123027411513316

39687452

33736

3977

3

1M-6, 0-150,

LU

49.201.67

14.5011.600.1947.70

11.702.80BC0.17

99.56

241100124

1610351667741

32136

467

4

1

clean rock

JLUG

50.171.62

14.4711.670.197.56

11.752.670.120.15

100.35—

260107124

——

7169

—

35.1—

95—

16.92.61.66

SD

0.060.010.030.030.000.030.040.020.000.000.10

311

21

0.7

1

0.60.80.7

1M-6,WR, Glass

SL

49.971.64

14.2911.620.207.27

11.732.580.130.12

99.55—

236111122(8)

——

7573

—35737

—90

——3.1—2

Notes: All major elements (SiO2-Total) in wt%; all trace elements (S-Rb) in ppm, based on the number of analyses («). — = not determined. BC = beyond calibration curve. SU =analyses made at Stockholm University, LU = analyses made at SGAB, Luleà, JLUG = analyses made at Justus-Liebig University. The triplicate SU analyses represent three sep-arate digestions of the rocks; the six JLUG analyses represent three separate sample mounts, each analyzed in duplicate. All standard deviations (SD) refer to JLUG data. SL =Data obtained in ship's laboratory during Leg 142 and represents the mean of two determinations. BC = determination beyond the calibration curve.

Table 5. Mean composition of all basalts from Core 142-864A-1M.

SiO2

TiO2

A12O3

Fe2O3

MnOMgOCaONa2OK2OP2O5TotalSCrZrSrBaCdCoCuNiSeVYYbZnLaNb1

SU

Mean

49.651.67

14.5111.900.1957.58

11.772.740.1420.162

98.64115326611013414.47433727444.6

33535.62.79

954.6—

23

SD

0.270.030.360.230.0040.170.270.060.0170.0060.99

4816342.2

173210.560.50.0421.7

LU

Mean

49.441.69

14.4711.590.1957.63

11.662.82

BC0.164

99.62

237100.212314.1

44717341.1

32336.94.3

834.4—

7

SD

0.230.020.050.070.0010.060.100.02

0.0050.44

130.611.3

41340.320.80.290.1

SL

Mean

49.911.64

14.3111.620.207.32

11.712.550.140.105

99.50

237111122

75*74*

35536.3

91*

2.912

SD

0.120.010.070.050.010.110.050.030.020.0110.25

511

22

90.6

4

0.2

Notes: All major elements (SiO2-Total) in wt%; all trace elements (S-Nb) in ppm,based on number of analyses (n). — = not determined. SU = analyses made atStockholm University, LU = analyses made at SGAB, Luleá. SD = standard devia-tion, BC = determinations beyond the calibration curve. * = mean data obtained inthe ship's laboratory (SL) based on 11 determinations (one outlier deleted in eachcase).

pies were analyzed by three participating laboratories, of which 23determinations were made at SU, 7 at JLUG, and 12 at LU. The re-sults show that these basalts form a very homogeneous group, as isalso shown by the ship's laboratory (SL) data.

Some trends in the data should be pointed out, however, so thatanalytical procedures can be improved. Several SU analyses showsomewhat high oxide sums, exceeding 100%. This is not necessarilywrong because the recommended value for BHVO-1 is 100.30%. In-deed, oxide sums exceeding 100% should be expected for some ba-salts when total iron is reported in three-valent form, because about90% of all iron generally is present in divalent form in basalts. How-ever, there is probably some source of error when these oxide sumsexceed 101.00%. This problem may be related to difficulties in get-ting stable silica determinations, as discussed above, and also to thefact that the SU data for Fe2O3, MgO, CaO and Cr seem to be system-atically somewhat larger than corresponding data from LU, JLUG, orSL. These problems are caused partly by the lack of thermal stabili-zation of the spectrometer's nebulizer unit, and partly by the fact thatthe sequential read-out system in the SU monochromator ARL 3520prevents the use of internal standards to correct for short-term noise.However, in some cases monochromators are useful, because theypermit fast selections of alternative spectral lines if a preferred ana-lytical line has an interference. For many trace elements this unit hastherefore proved excellent. This old instrument has recently been re-placed with a new one, which unfortunately prohibits further tests.

Major elements and most traces for SU-H samples (Tables 7 and8) show good agreement with many other determinations, except forBa, which could not be measured with confidence. The reason for thisis that no HF-HC1O4 dissolved solutions could be used owing to thecorrosion risk for the plasma torch.

The same analyses also report some rather high Ni values, whichprobably represent real variations and not errors. Thus, sporadic de-terminations in the ship's laboratory also yielded high values for theelements Cu, Ni, and Zn (see Table 5). Studies of the basalt pebblesbefore analysis showed that some of them had tiny pores containing

78

DATA REPORT

Table 6. Composition of bulk samples of major aliquots of basalts fromSection 142-846A-4Z-1.

Laboratory:

Sampledepth (cm):

SiO2

TiO2

A1,O3

Fe,O 3

MnOMgOCaONa2OK2OP2O5

TotalSCrZrSrBaCdCoQ iNiSeVYYbZnLaKbn

SU

0-35

49.351.78

14.4712.350.2037.49

11.652.790.1250.17

100.381200244117131

156734657350

34737

397

5_4

LU

0-35

49.701.81

14.4012.000.2027.51

11.602.88

BC0.17

100.29

208107121

13

43557142

33734

587

4

1

SU

Piece 2

49.731.73

14.3312.090.2007.30

11.462.760.1600.18

99.941210227117125

146635647248

34235

396

7

1

LU

Piece 2

49.501.80

14.4012.100.2037.50

11.502.87

BC0.19

100.11

192108120

12

44577141

33640

444

4

1

SL

Piece 1

49.781.75

14.1011.960.217.27

11.502.650.150.11

99.46

201121120

7173

37140

92

3.11

Notes: All major elements (SiO2-Total) in wt%; all trace elements (S-Nb) in ppm,based on number of analyses («). — = not determined. SU = analyses made atStockholm University, LU = analyses made at SGAB, Luleà. SL = data obtained inthe ship's laboratory during Leg 142. BC indicates determinations beyond the cali-bration curve.

minerals that looked like sulfides. Because these sulfides occur veryirregularly, there is also reason to expect sporadic jumps in the datafor Cu, Ni, and Zn. These conclusions are corroborated by two spe-cial analyses (SU-Spec) in Tables 7 and 8, which were done to showthat the variations are real and not the result of freak events during theanalytical process.

The LU data are generally very good for major components, judg-ing from the agreement with other analyses. These results are expect-ed in view of the fact that this laboratory consistently uses internalstandards in the samples to correct for both short-term and long-termdrift. However, the data for K are far too low, obviously because of apoor calibration curve, which has not worked well at low concentra-tions; in the tables this is marked BC, indicating that the values fallbeyond the lower end of the calibration curve. Some trace elementsare likewise remarkably variable; data for Cu and Zn can be surpris-ingly low compared with those from other laboratories, and the SDdata for Co, Cu, Ni, and Zn show more scatter than the correspondingSU data; these analytical problems are also revealed by the compila-tion in Table 1. These problems at LU probably resulted from too in-tensive processing of samples during the metaborate fusing step;studies at SU suggest an almost systematic loss of Zn and possiblyalso disturbances in the Cu analyses as a consequence of faulty me-taborate fusions.

The JLUG data are too few to justify more sweeping generaliza-tions, but agreement with the other determinations is good. The datafor Zn and Cr agree well with those from SU.

The SL data (Storms, Batiza, et al, 1993) show very good agree-ment with most other results for the major elements, except phospho-rus, which is consistently low compared with data from SU, LU, andJLUG, and the MgO data, which appear to be too low. Furthermore,many trace component determinations appear quite good; data forCu, Zn, and Ni, for instance, agree with those from SU, but data forCr are consistently low and values for V are commonly high. Never-theless, our overall impression is that the ship's XRF facility is excel-lent for the determination of most major components in rocks, as wellas for the determination of some trace elements. Also, the quality of

Table 7. Composition of bulk samples of major aliquots of basalts from Section 142-864A-5Z-1.

Laboratory:

Sample depth (cm):

SiO2

TiO2

A12O3

Fe2O3

MnOMgOCaONa2OK2OP,O 5

TotalSCrZrSrBaCdCoCuNiSeVYYbZnLaNbn

SU

0-5

49.881.83

14.2512.590.2077.22

11.542.800.1230.18

100.621190226120129

147036676950

35437

398

4—4

LU

0-5

49.101.85

14.1012.300.2047.21

11.402.88BC0.18

99.24—

176110119

13

45546941

34341

585

4—1

SU

24-30

49.441.85

14.7412.740.2007.33

11.662.890.1300.18

101.161210210122125

126634646545

34934

3977—1

LU

24-30

49.801.86

14.4012.300.2077.34

11.502.93BC0.19

100.60

153113120

12

41586341

34442

477

5—1

SU-H

24-30

49.931.78

14.0212.200.2007.02

11.342.740.1370.18

99.55—

185108ND

8*—

5269

16648

28739

81——1

SUSpec

24-30

49.611.85

14.7412.740.2007.33

11.662.890.1300.18

101.33—

211125144

13—

3164

21143

34737

83——1

SL

24-30

49.641.80

13.9712.220.217.02

11.412.620.140.12

99.14—

176123118

———

7065

—366

41—

95—2.31

Notes: All major elements (SiO2-Total) in wt%; all trace elements (S-Nb) in ppm, based on number of analyses («). — = not determined. SU = analyses made at Stockholm Univer-sity; SU-H = data from Höganas Eldfast; LU = analyses made at SGAB, Luleà; SL = data obtained in ship's laboratory during Leg 142. BC indicates determinations beyond thecalibration curve. * = All SU-H data were determined in a major element program only, to avoid HF corrosion on the plasma torch in the spectrometer used in this case. All Ba datain these determinations are too close to the detection limit to be valid, because of the strong dilution of these samples.

DATA REPORT

Table 8. Compositional variations of bulk samples of major aliquots ofbasalts from Section 142-864B-2W-1.

Laboratory:

Sample depth (cm):

SiO2

TiO2

A12O3

Fe2O3MnOMgOCaONa2OK2OP2O5TotalSCrZrSrBaCdCoCuNiSeVYYbZnLaNbn

SU

16-18

49.841.75

14.9812.130.2007.49

12.102.780.1600.18

101.611200249114127

13633167

50329

343

917—1

LU

16-18

49.401.70

14.5011.700.1997.52

11.802.79BC0.18

99.87—

24110112212

44546841

327364

444—1

SU-H

16-18

50.281.64

14.3811.690.1907.26

11.772.650.1400.19

100.19—

23696

—7*

3970

23644

290

80——1

SU-Spec

16-18

_

—

—

25311412217

2867

25743

32433

90

—1

SL

16-18

49.901.65

14.1911.570.207.20

11.762.570.170.12

99.33—

226113124

7366

34237

86

2.91

Notes: All major elements (SiO2-Total) in wt%; all trace elements (S-Nb) in ppm. SU =analyses made at Stockholm University; SU-H = data from Höganàs Eldfast; LU =analyses made at SGAB, Luleá; SL = data obtained in the ship's laboratory duringLeg 142. BC = determinations beyond the calibration curve; n = number of analy-ses. — = not determined. * = All SU-H data were determined in a major elementprogram only, to avoid HF corrosion on the plasma torch in the spectrometer used inthis case. All Ba data in these determinations are too close to the detection limit tobe valid, because of the strong dilution of these samples.

the data produced in the ship's XRF facility is the same as that frommany shore-based laboratories. Similar conclusions were reached byHergt and Sims (1994), and appear reasonable in view of recent stud-ies by Normand et al. (1989).

A further reassuring fact derives from the comparison of these SLresults with data from the shore-based studies; the differences in data,for instance for silica or alumina, rarely amount to more than 0.5%and 0.2%, respectively. Many other interlaboratory studies, on theother hand, reveal numerous cases where corresponding differencesmay reach several percent, and for many trace elements it is not un-usual for the highest value reported to be several times larger than thesmallest value (Gladney et al., 1990; Gladney and Roelandts, 1988a,1988b, 1990).

One weakness of XRF is the restricted capacity to determinemany trace components at low concentrations, which are of interestfor the study of many alteration processes as discussed above. Thiswas demonstrated in a study of 1580 XRF-determinations of 13 traceelements in 20 geostandards (Boström and Bach, this volume). De-terminations of Ba were therefore aborted during an early phase ofLeg 142.

RECOMMENDATIONS

Statistical tests should be used more widely in the ship's XRF lab-oratory. The uncertainties in the calibration curves (e.g., as Sy x)should be explicitly studied before other final results are evaluated,as pointed out above and in Flanagan (1986). Such routines are gen-erally followed in many chemical laboratories for all sets of data.Similar conclusions were reached by Dick, Erzinger, Stokking, et al.(1992). They pointed out that the interlaboratory variability for sam-ples from ODP/DSDP Hole 504B is surprisingly high, even whenonly XRF analyses are considered. For instance, Ti/Zr ratios seem-

ingly differ significantly with depth in the lower part of the hole, avariability that is caused by systematic differences in the determina-tion of Ti and Zr in different laboratories. Although all of the data areprecise, the accuracies obviously vary considerably. However, accu-racies are rarely reported in geoscientific papers. This makes combin-ing different data sets difficult, because these data sets must have thesame precision and accuracy. This also applies to DSDP/ODP holes,which are revisited over a considerable time span. Consequently,when rock analyses are made at sea, it is critical to (1) carefully doc-ument the analytical precision and (2) perform analyses of widelyused reference rocks, such as BHVO-1 and BCR-1, and report thesefindings in the Initial Reports volume, as was done during Leg 136(Shipboard Scientific Party, 1992) and Leg 142 (Storms, Batiza, etal., 1993).

The advantages and drawbacks of introducing a faster and moreversatile analytical system (e.g., an ICP-AES unit) has been dis-cussed elsewhere in this volume.

ACKNOWLEDGMENTS

We want to thank all the marine technicians aboard JOIDES Res-olution for cordial assistance and good-humored patience in sampleprocessing and distribution during Leg 142 , as well as for help in theship's computer laboratories. We also want to thank Dr. J. Allan forkindly supplying us with an advanced copy of the article by Hergt andSims (1994), which we were unaware of when this study was started.

The preparations and analyses at SU were made by Mrs. BirgittaBoström, but assistance was provided also by Mr. Jan-Erik Anders-son at Höganàs Eldfast, who generously let us use his SPECTROunit, and by Mr. Christer Pontér and his associates at SGAB. This re-search was supported by a grant from the Swedish Natural SciencesResearch Council to one of us (K.B.). We want to convey our sincerethanks for all this assistance and support.

REFERENCES*

Abbey, S., 1991. What is a "Method"? Geostand. NewsL, 15:191-194.Arrhenius, G.O.S., and Bonatti, E., 1965. Neptunism and volcanism in the

ocean. In Sears, M. (Ed.), Progress in Oceanography (Vol. 3): New York(Pergamon), 7-21.

Arvanitides, N., Boström, K., Kalogeropoulos, S., Paritsis, S., Galanopoulos,V.P., and Papavassiliou, C, 1990. Geochemistry of lavas, pumice andveins in drillcore GPK-1, Palaea Kameni, Santorini. In Hardy, D.A.,Keller, J., Galanopoulos, V.P., Flemming, N.C., Druitt, T.H. (Eds.), Theraand the Aegean World, HI (Vol. 2): London (The Thera Foundation),266-279.

Bach, W., 1991. Geochemie und Petrogenese von Ozeanbodenbasalten desOstpazifischen Rückens 6°-30°S [Diploma (M.S.) thesis]. Justus-LiebigUniv., Giessen.

Boström, B., 1987. Report on Ailsa Craig Granite. In Govindaraju, K. (Ed.),Compilation Report on Ailsa Craig Granite AC-E with the Participationof 128 GIT-IWG Laboratories. Geostand. NewsL, 11:203-255.

Boström, K., Joensuu, 0., Moore, C , Boström, B., Dalziel, M., and Horow-itz, A., 1973. Geochemistry of barium in pelagic sediments. Lithos,6:159-174.

Boström, K., and Peterson, M.N.A., 1966. Precipitates from hydrothermalexhalations on the East Pacific Rise. Econ. Geol., 61:1258-1265.

Boström, K., and Peterson, M.N.A., 1969. The origin of aluminium-poor fer-romanganoan sediments in areas of high heat flow on the East PacificRise. Mar. Geol., 7:427-447.

Burman, J.O., and Boström, K., 1979. Comparison of different plasma exci-tation and calibration methods in the analyses of geological materials byoptical emission spectrometry. Anal. Chern., 51:516-520.

*Abbreviations for names of organizations and publications in ODP reference lists fol-low the style given in Chemical Abstracts Service Source Index (published byAmerican Chemical Society).

80

DATA REPORT

Burman, J.O., Boström, B., and Boström, K., 1977. Geochemical analysis byplasma spectroscopy. Geol. Foeren. Stockholm Foerh., 99:102-110.

Burman, J.O., Pontér, C , and Boström, K., 1978. Metaborate digestion pro-cedure for inductively coupled plasma-optical emission spectrometry.Anal. Chern., 50:679-680.

Crow, E.L., Davis, F.A., and Maxfield, M.W., 1960. Statistics Manual: NewYork (Dover).

Dick, H.J.B., Erzinger, J., Stokking, L.B., et al., 1992. Proc. ODP, Init.Repts., 140: College Station, TX (Ocean Drilling Program).

Flanagan, FJ., 1986. Reference samples in geology and geochemistry. Geo-stand.Newsl, 10:191-264.

Gladney, E.S., Jones, E.A., Nickell, E.J., and Roelandts, I. 1990. 1988 com-pilation of elemental concentration data for USGS basalt BCR-1. Geo-stand. Newsl., 14:209-359.

Gladney, E.S., and Roelandts, I., 1988a. 1987 compilation of elemental con-centration data for USGS BIR-1, DNC-1 and W-2. Geostand. Newsl.,12:63-118.

, 1988b. 1987 compilation of elemental concentration data forUSGS BHVO-1, MAG-1, QLO-1, RGM-1, SCo-1, SDC-1,SGR-1 andSTM-1. Geostand. Newsl., 12:253-262.

-, 1990. 1988 compilation of elemental concentration data forCCRMP reference rock samples SY-2, SY-3 and MRG-1. Geostand.Newsl., 14:373-458.

Govindaraju, K., 1989. 1989 compilation of working values and sampledescription for 272 geostandards. Geostand. Newsl., 13 (spec. iss.)

Govindaraju, K. (Ed.), 1987. 1987 compilation report on Ailsa Craig GraniteAC-E with the participation of 128 GIT-IWG laboratories. Geostand.Newsl., 11:203-255.

Hergt, J.M., and Sims, Jr., D.R., 1994. Assessment of the precision of Leg135 shipboard XRF analyses and the contamination introduced by crush-ing in tungsten carbide. In Hawkins, J., Parson, L., Allan, J., et al., Proc.ODP, Sci. Results, 135: College Station, TX (Ocean Drilling Program),925-929.

Hunt, D.T.E., and Wilson, A.L., 1986. The Chemical Analyses of Water:Cambridge, UK (Royal Society of Chemistry).

Normand, M.D., Leeman, W.P., Blanchard, D.P., Fitton, J.G., and James, D.,1989. Comparison of major and trace element analyses by ICP, XRF,INAA and ID methods. Geostand. Newsl., 13:283-290.

Rona, RA., Boström, K., and Epstein, S., 1980. Hydrothermal quartz vugfrom the Mid-Atlantic Ridge. Geology, 8:569-572.

Shipboard Scientific Party, 1992. Explanatory notes. In Dziewonski, A.,Wilkens, R., Firth, J., et al., Proc. ODP, Init. Repts., 136: College Station,TX (Ocean Drilling Program), 9-26.

Skornyakova, I.S., 1964. Dispersed iron and manganese in Pacific Oceansediments. Int. Geol. Rev., 7:2161-2174.

Storms, M.A., Batiza, R., et al., 1993. Proc. ODP, Init. Repts., 142: CollegeStation, TX (Ocean Drilling Program).

Date of initial receipt: 28 June 1993Date of acceptance: 11 May 1994Ms 142SR-113

S1