Samples were prepared for analysis at the Southampton Oceanography Centre (SOC; UK). All samples were sketched and described before being sectioned and trimmed for powdering. Weathered edges and saw marks were removed by grinding before the samples were washed and scrubbed in deionized water and then oven dried (~110°C). Density was measured on a representative fragment of the sample following Archimedes's principle by weighing it in air and then suspending it in a beaker of water. Following oven drying, samples were reduced to a coarse grit using a fly-press, with the sample wrapped in clean paper to reduce contamination and sample loss. Samples were then reduced to a very fine powder suitable for chemical analyses by grinding in a clean tungsten carbide shatterbox.
Major and trace element analyses of whole-rock samples were conducted at the Ronald B. Gilmore X-Ray Fluorescence Laboratory at the University of Massachusetts—Amherst (USA). Major element analyses (Si, Ti, Al, Fe, Mn, Mg, Ca, Na, K, and P) were conducted on fused La-bearing lithium borate glass disks using a Siemens MRS-400 multichannel X-ray spectrometer following methods modified from Norrish and Hutton (1969). Trace element analyses (Nb, Zr, Y, Sr, Rb, Th, Pb, Ga, Zn, Ni, Cr, V, Ba, Ce, and La) were obtained from pressed powder pellets on a Philips PW2400 sequential spectrometer. Intensities were corrected for background interferences and variations in mass absorption coefficients following methods modified from Norrish and Chappel (1967). Mass absorption coefficients for elements with shorter and longer wavelengths than the Fe absorption edge were estimated following the techniques of Reynolds (1967) and Walker (1973), respectively. Estimates of the precision and accuracy of these analytical procedures are described in Rhodes and Vollinger (2004).
Carbon and sulfur concentrations were determined on a split of the sample powders (~5 g) at the Department of Geology, University of Leicester (UK) using a Leco C-S analyzer and corrected for instrument blanks. Individual samples were weighed into crucibles, with the addition of iron and tungsten chip accelerants prior to loading into a radio-frequency induction furnace. The system was initially purged with oxygen, which then continued to stream throughout the combustion process. Power is applied until the accelerants are molten; carbon is given off as CO2 and CO and sulfur is given off as SO2. A dust filter prevents silicates (a potential infrared wavelength overlap) from entering the infrared cells. The combustion gases are passed through a drying tube of magnesium perchlorate and then passed to the SO2 infrared cell. Once determined, the gases pass through a platinized silica gel catalyst to convert any CO to CO2. Any SO2 is trapped as SO3. The CO2 content is then determined in the CO2 infrared cell, and all gases then vent from the system. The infrared absorption cells use a tungsten filament as the source, heated to ~850°C. The output from the cells is monitored at 4 Hz and converted from an analog signal to a digital signal, and the areas of the peaks are integrated. These values are then corrected for sample weight, blank value, and calibration factors to calculate total carbon and sulfur concentrations. The lower limit of detection for carbon is 10 ppm with a precision of ±5%, and for sulfur the lower limit is 10 ppm with a precision of ±8%.
Ferrous iron concentrations were determined at SOC by titration with a standardized KMnO4 solution following an adapted "Platt" method. Sample powder (0.5 g) was placed in a clean polypropylene bottle to which a mixture of H2SO4 and HF was added and then simmered for 20 min to ensure full decomposition. This rock solution was then added to a large volume (~300 mL) of deionized water that had been pre-prepared with 10 mL H2SO4 and 10 mL saturated boric acid solution (the latter to render any excess HF inoperative), stirred, and titrated against the standard solution of KMnO4. Most samples were analyzed twice, and duplicate determinations of FeO are generally within ±0.02 wt% (all within 0.05 wt%).