, except for the low values (<0.05 wt%) of K2O and P2O5 in BIR-1 (Tables T1, T2).
Previously published geochemical data of Subunits 4A and 4B include onboard X-ray fluorescence (XRF) (Leg 170) and inductively coupled plasma–atomic emission spectroscopy (ICP-AES) (Leg 205) analyses of a total of 52 samples for major and selected trace elements (Kimura, Silver, Blum, et al., 1997; Morris, Villinger, Klaus, et al., 2003). An additional 50 samples were powdered in an alumina ball mill and analyzed by XRF, inductively coupled plasma–mass spectrometry (ICP-MS), and instrumental neutron activation analysis (INAA) at Washington University in St. Louis (Missouri, USA). Results are reported here. A total of 33 samples from Leg 205 were analyzed for Sr and Nd isotopes by thermal ionization mass spectrometry (TIMS) at the National Oceanography Centre, Southampton (UK).
Major elements were determined for 34 samples from Leg 205 at Washington University by XRF analysis of fused glass discs prepared from preignited sample powders, using procedures described by Couture et al. (1993). Loss on ignition (LOI) values represent mass loss measured on sample powders ignited for 50 min at 950°C in a muffle furnace. LOI was <0.5 wt% for most samples but was 0.91–1.46 wt% for samples with Mg# (= Mg2+/[Mg2+ + Fe2+], where Fe2O3 is converted to FeO) >0.58 (one exception), which are clustered at the top half of Subunit 4B (~460–509 mbsf). Average sum of element oxides is 99.6 wt%, with all but three samples having totals between 99.5 and 100.5 wt%. Analyses of the reference basalts BHVO-2 and BIR-1 and gabbro JGB-1 agree with certified values within 1 or 2 , except for the low values (<0.05 wt%) of K2O and P2O5 in BIR-1 (Tables T1, T2).
Trace element abundances were analyzed by high-resolution ICP-MS (Table T2). Powders of rock samples and geochemical reference standards were dissolved with 5:1 HF:HNO3 in a microwave-assisted digestion system. Solutions were diluted with 1% trace metal grade HNO3 to attain a sample mass:dilution ratio of ~1:1000. For analysis of Sr, Ba, Y, and selected transition row metals, samples were further diluted to 1:5000 to reduce concentrations enough to preserve detection in pulse counting mode.
Instrument calibration for rare earth elements (REEs) and high-field-strength elements (HFSEs) (Zr, Hf, Nb, Ta, U, and Th) is performed using a rock reference standard solution, and the regression is forced through the origin after blank subtraction. This is a reliable method given the >6 orders of magnitude dynamic linear range of the ICP-MS (Jarvis et al., 1992; Eggins et al., 1997). Calibration with geochemical reference standards maintains approximately constant matrix conditions, including the presence of residual trace HF from sample digestion (McGinnis et al., 1997), and mitigates many of the issues created using calibration with near matrix-free synthetic cocktails of high-field-strength single-element standards. Elements that are not appreciably affected by oxide interferences and easily rinsed with a few percent HNO3 are calibrated against synthetic multielement solutions created from single-element standards. The choice of geochemical reference standards for some elements and synthetic calibration standards for others was made after extensive testing of both approaches. External oxide correction for the REEs Eu-Lu and HFSEs Hf and Ta is performed offline following methods modified from Lichte et al. (1987). Initial oxidation formation factors range from ~3% to <0.5% for the REEs, depending on element. In-run measurement of ThO/Th during analysis monitors changes in oxide formation and allows sample-to-sample oxide correction (Lichte et al., 1987); REE + Th oxide formation is approximately linear run to run (R2 > 0.97).
Typical instrument runs include measurements of procedural blanks, geochemical reference standards, and a drift solution interspersed frequently with sample solutions. Accuracy and precision based on replicate measurements of appropriate geochemical reference standards JGb-1 and BIR-1 are shown in Table T1.
Trace elements for 16 samples from Leg 170 (12 from Site 1039 and 4 from Site 1040) were measured by INAA at Washington University in St. Louis (Table T3). Powdered rock chips were encapsulated in high-purity silica tubing and subsequently irradiated at the University of Missouri research reactor in a thermal neutron flux of 5.15 x 1013 cm/s. Samples were radioassayed by gamma ray spectroscopy following the methods of Korotev (1996), and data were reduced using an updated version of the TEABAGS (trace element analysis by automated gamma ray spectroscopy) software developed by Lindstrom and Korotev (1982). Estimated accuracy and 1- precision for standard reference material JGb-1 is given in Table T1.
The coherence of data measured by multiple analytical techniques was evaluated by comparing ongoing, long-term analyses of standard reference materials at Washington University in St. Louis. Analyses of repeated determinations of reference basalts BCR-1 and NBS-688 by XRF and INAA methods at Washington University (Korotev, 1996; Couture et al., 1993) are in excellent agreement with certified values. Several trace elements of reference basalt JGb-1 have also been measured by both INAA and ICP-MS at Washington University and agree within 2- error (Table T1).
