Methods (extracted from original Readme file (.txt), which can also be found under Supplemental Files):

University of Victoria - Water samples were collected through CO2-flushed tubing into evacuated flasks until half-full. The water was equilibrated with the headspace and then removed. Noble gas samples were determined following a method similar to that described in (Hamme, R.C., and J.P. Severinghaus 2007) but with a helium rather than nitrogen balance gas. Gas samples were purified through a -90°C trap to remove water vapor and exposed to a hot getter to remove all but the noble gases. A calibrated aliquot of 38Ar was added along with compressed helium to bring the pressure back up. Samples were then measured for Ar isotopes and Ne/Ar and Kr/Ar ratios on a MAT 253 isotope ratio mass spectrometer. Absolute Ar concentrations were determined by Ar isotope dilution with the added 38Ar, while the ratio measurements were combined with the absolute Ar concentrations to yield Ne and Kr concentrations. Noble gas standards were calibrated relative to air with assumed dry mole fractions of 1.818e-5 for Ne, 9.34e-3 for Ar, and 1.141e-6 for Kr. N2/Ar measurements were determined following the method described in (Emerson et al. 1999). Gas samples were purified though a trap in liquid nitrogen to remove water vapor and carbon dioxide. Samples were then measured for N2/Ar ratios on a MAT 253 mass spectrometer. N2/Ar standard gases were calibrated relative to air with assumed dry mole fractions of 9.34e-3 for Ar and 0.78084 for N2.

Woods Hole Oceanographic Institution - Noble gas samples analyzed at Woods Hole Oceanographic Institution were determined following variants of the method described in (Stanley R.H.R., B. Baschek, D.E. Lott, and W.J. Jenkins 2009). Water samples were collected into stainless steel containers for cruises in 2008 and prior (Bermuda Atlantic Time-series Study cruises and CLIVAR I6S) or into crimped copper tubes for cruises occurring in 2009 and later. All the dissolved gas was extracted from the water and then purified through a cryotrap to remove water vapor and exposed to a hot getter to remove all the the noble gases. The noble gases were then frozen into two cryotraps, allowing each noble gas to be sequentially released for analysis in a quadrupole mass spectrometer. Noble gas concentrations were determined by peak height manometry for all gases and samples, except the most recent Kr and Xe data measured from the eastern tropical Pacific, which use a new isotope dilution method. Noble gas standards were calibrated relative to air with assumed dry mole fractions of 5.24e-6 for He, 1.818e-5 for Ne, 9.34e-3 for Ar, 1.141e-6 for Kr, and 8.7e-8 for Xe.

Scripps Institution of Oceanography - Water samples were collected through CO2-flushed tubing into evacuated flasks until half-full. The water was equilibrated with the headspace and then removed. Noble gas samples were determined following the method described in (Hamme, R.C., and J.P. Severinghaus 2007). Gas samples were purified through a -90°C trap to remove water vapor and exposed to a hot getter to remove all but the noble gases. A calibrated aliquot of 38Ar was added along with compressed nitrogen to bring the pressure back up. Samples were then measured for Ar isotopes and Kr/Ar ratios on a MAT 252 isotope ratio mass spectrometer. Absolute Ar concentrations were determined by Ar isotope dilution with the added 38Ar, while the ratio measurements were combined with the absolute Ar concentrations to yield Kr concentrations. Noble gas standards were calibrated relative to air with assumed dry mole fractions of 9.34e-3 for Ar, and 1.141e-6 for Kr. N2/Ar measurements were determined following the method described in (Kobashi, T., J.P. Severinghaus, and K. Kawamura 2008). Gas samples were purified though a trap in liquid nitrogen to remove water vapor and carbon dioxide and then through heated copper to remove oxygen. Samples were then measured for N2/Ar ratios on a MAT 252 mass spectrometer. N2/Ar standard gases were calibrated relative to air with assumed dry mole fractions of 9.34e-3 for Ar and 0.78084 for N2.

University of Washington - Water samples were collected through CO2-flushed tubing into evacuated flasks until half-full. The water was equilibrated with the headspace and then removed. Neon samples were determined following the method described in (Hamme, R.C., and S.R. Emerson 2004). A calibrated aliquot of 22Ne was added to the sample flasks before sampling. Gas samples were purified though a trap in liquid nitrogen to remove water vapor and carbon dioxide and then through an activated charcoal trap in liquid nitrogen to remove argon and heavier gases. Samples were then measured for Ne isotopes on a quadrupole mass spectrometer. Absolute Ne concentrations were determined by Ne isotope dilution with the added 22Ne. The spike aliquot was calibrated relative to air with assumed dry mole fractions of 1.818e-5 for Ne. Ar concentration and N2/Ar ratios were determined by two different methods. Samples collected in 2001 and earlier were determined following the method described in (Emerson et al. 1999). Gas samples were purified though a trap in liquid nitrogen to remove water vapor and carbon dioxide. Samples were then measured for N2/Ar and O2/Ar ratios on a MAT 251 mass spectrometer. For the samples collected near Bermuda in 2001, the O2/Ar ratio measurements were combined with absolute O2 concentrations determined by Winkler titration to yield Ar concentrations. More recent Ar concentration and N2/Ar ratio measurements were determined following the method described in (Emerson, S., T. Ito, and R.C. Hamme 2012). Gas samples were purified though a trap in liquid nitrogen to remove water vapor and carbon dioxide and then a calibrated aliquot of 36Ar was added. Samples were then measured for Ar isotopes and N2/Ar ratios on a Delta X/L isotope ratio mass spectrometer. Absolute Ar concentrations were determined by Ar isotope dilution with the added 36Ar. Ar and N2/Ar gas standards were calibrated relative to air with assumed dry mole fractions of 9.34e-3 for Ar and 0.78084 for N2. Through rigorous method inter-comparison and repeated laboratory comparison of oxygen concentration determined by isotope dilution and Winkler titration, Ar concentration samples analyzed by this 36Ar isotope dilution method have been found to be 0.7% too low. Accordingly, the Ar concentration and Ar saturation anomaly data from this method have all be increased by 0.7% in this database. Kr/Ar samples were determined following a method similar to that described in (Hamme, R.C., and J.P. Severinghaus 2007). Gas samples were purified through a -90°C trap to remove water vapor and exposed to a hot getter to remove all but the noble gases. Compressed nitrogen was added to bring the pressure back up. Samples were then measured for Kr/Ar ratios on a MAT 253 isotope ratio mass spectrometer. Noble gas standards were calibrated relative to air with assumed dry mole fractions of 9.34e-3 for Ar, and 1.141e-6 for Kr.

Quality control - Samples measured at University of Victoria, Scripps Institution of Oceanography, and University of Washington were nearly all collected in duplicate. For these samples in this database, only data where both duplicates were analyzed successfully and where their standard deviation was less than three times the pooled standard deviation are included. Noble gas duplicates were required to be within 0.93% of each other for Ne, within 0.28% for Ar, and within 0.35% for Kr. Similarly N2/Ar duplicates were required to be within 0.17% of each other. Both duplicates are present in the database. The exception to this is the N2/Ar data collected in 2007 in the Labrador Sea and analyzed at Scripps Institution of Oceanography. These samples were not collected in duplicate but are present in the database. For samples collected at the Bermuda Atlantic Time-series Study and in the Southern Ocean that were analyzed at Woods Hole Oceanographic Institution, we binned the data by depth for each cruise and removed samples that were outside three times the standard deviation of samples within each depth bin. For the 2010-2011 Atlantic GEOTRACES transect samples that were analyzed at Woods Hole Oceanographic Institution, we simply removed data where the Ne saturation anomaly was less than -10% or larger than 5%. For the 2013 Pacific GEOTRACES transect samples that were analyzed at Woods Hole Oceanographic Institution, we removed Xe measurements analyzed by peak height manometry, retaining only those analyzed by isotope dilution.

NaN = missing data.