Saunders, A.D., Larsen, H.C., and Wise, S.W., Jr. (Eds.), 1998 Proceedings of the Ocean Drilling Program, Scientific Results, Vol. 152 425 35. XRF ANALYSES OF VOLCANIC ROCKS FROM LEG 152 BY LABORATORIES IN EDINBURGH AND COPENHAGEN: IMPLICATIONS FOR THE MOBILITY OF YTTRIUM AND OTHER ELEMENTS DURING ALTERATION 1 Lotte Melchior Larsen, 2 J. Godfrey Fitton, 3 John C. Bailey, 4 and Jørgen Kystol 2 ABSTRACT Neighboring (parallel) samples from 72 lava flows, splits of identical samples from five lava flows, and splits of 10 ship- board powders from Ocean Drilling Program Leg 152 were analyzed by X-ray fluorescence (XRF) in both Edinburgh and Copenhagen. The agreement between the two data sets is very good, although systematic differences are apparent in some cases. Correlation between the results for the identical powders is very high. Correlation between the results for the parallel samples is somewhat lower, mainly because of element redistribution during secondary alteration, but also due to primary lava inhomogeneity. The elements show differences in the degree of scatter due to alteration, with K, Rb, and Ba being the most mobile, and Si, Sc, Al, Fe, Zn, V, and Nb the least mobile. Y is immobile in the Site 917 lavas but distinctly mobile in the Site 918 lavas where it is both leached and precipitated locally. The mobility of Y is related to secondary breakdown of clinopyrox- ene at Site 918, in contrast to Site 917 where the clinopyroxene is generally fresh. The hydrothermal alteration conditions at the two sites appear to have been different. INTRODUCTION During Ocean Drilling Program (ODP) Leg 152 to the southeast Greenland Margin, a succession of subaerial lava flows was drilled. The lava flows are part of the thick submerged seaward-dipping re- flector sequence (SDRS) along the margin. The lavas were cored at Sites 915, 917, and 918 and were sampled for geochemical analysis, mainly X-ray fluorescence spectrometry (XRF), in both Edinburgh and Copenhagen. This paper compares the results from the different laboratories. Moreover, the analysis of both identical and nonidenti- cal powders from the same lava units illustrates the effects of primary compositional variability and secondary alteration processes in the volcanic successions. SAMPLING AND SAMPLE TREATMENT From each of 72 lava flows two separate (parallel) samples were taken, whereas single samples from each of five lava flows were shared. Furthermore, splits of 10 powders prepared on the ship were analyzed for trace elements in both Edinburgh and Copenhagen. In 46 lavas, the parallel samples were taken within less than 10 cm from each other, and often in direct contact. In 16 lavas, the samples were taken within 10 cm to 1 m from each other, and in 10 lavas the samples were taken more than 1 m apart, the maximum distance be- ing 18.3 m in a 55-m-thick lava flow (Unit 917-52). The results show no dependence on the distance between the parallel samples. The data sets from Edinburgh and Copenhagen are published in Fitton et al. (Chap. 28, this volume) and L.M. Larsen et al. (this vol- ume), respectively. The analytical procedures used in Edinburgh and Copenhagen are very similar, major elements being determined on fused glass discs and trace elements on pressed powder pellets. The major differences are as follows. (1) The Edinburgh samples were crushed in an agate Tema barrel, whereas the Copenhagen samples were crushed in a tungsten carbide ball mill. (2) The Edinburgh glass discs are made with lithium tetraborate and a sample:flux ratio of 1:5, whereas the Copenhagen glass discs are made with sodium tetrabo- rate and a sample:flux ratio of 1:7, with Na 2 O determined by atomic absorption spectrometry (AAS). (3) In Edinburgh, the major ele- ments are calibrated on natural rock standards, whereas in Copen- hagen, the major elements are calibrated on synthetic standards. LABORATORY COMPARISON In general, the agreement between the two data sets is very good, as shown by the data plots (Fig. 1) and the correlation coefficients and regression line slopes (Table 1). For most elements the results plot close to a 1:1 line with very high correlation coefficients for the identical powders (>0.99) and somewhat lower correlation coeffi- cients for the parallel powders from the same lava flow units. Differ- ences due to calibration effects are noted for Ba, La, Nb, and Sc, with line slopes for identical powders of 0.89, 0.83, 1.07, and 0.81, respec- tively. Differences due to blank or background effects are noted for P 2 O 5 and Zn. The results for Co (Fig. 1) show that the tungsten car- bide ball mill used for crushing the samples in Copenhagen has intro- duced 5-10 ppm Co in the samples. Results for the parallel powders plot along similar regression lines to the results for the identical powders, but the scatter is larger (small- er correlation coefficients, Table 1). The larger scatter is caused by a combination of primary heterogeneity in lava compositions and ele- ment redistribution during secondary alteration processes. LAVA HETEROGENEITY Primary heterogeneity was evident during sampling of two lava flows. Unit 917-21 is a very olivine-rich lava, and the two samples contained visually different proportions of olivine. Unit 917-54 is a dacite with frequent inclusions of basalt. The parallel samples from these two lava flows show significant compositional differences, and they have been omitted from the calculations discussed below. Pri- 1 Saunders, A.D., Larsen, H.C., and Wise, S.W., Jr. (Eds.), 1998. Proc. ODP, Sci. Results,152: College Station, TX (Ocean Drilling Program). 2 Geological Survey of Denmark and Greenland, and Danish Lithosphere Centre, Thoravej 8, DK-2400 Kobenhavn K, Denmark. lml@geus.dk 3 Department of Geology and Geophysics, Grant Institute, University of Edinburgh, West Mains Road, Edinburgh EH9 3JW, United Kingdom. 4 Institute of Geology, University of Copenhagen, Oster Voldgade 10, DK-1350 Kobenhavn K, Denmark.