Samples for this dataset were separated into different worksheets depending on the sample collection method (worksheets have been combined into one dataset; "Collection_Method" column indicates the source worksheet).

"CTD" samples were collected from 20L Niskin bottles associated with CTD casts by slowly (<1 L/min) draining the sample via a sampling tube from the spigot on the Niskin bottle through a dry, 25 g aliquot of acrylic fiber impregnated with MnO2 (Moore, 1976). These "Mn fibers" quantitatively sorb Ra isotopes from the aqueous phase.

"Core-top waters" were collected by gravity siphoning water overlying sediment cores that were collected as push cores by HOV Alvin, filtering the samples through 0.45 um cellulose filters, then slowly (< 1 L/min) passing the effluent over 25 g Mn fiber.

"Alvin Slurp" samples were collected using the slurp gun on HOV Alvin. These sample waters were stored onboard Alvin until they were recovered on the ship and slowly (< 1 L/min) passed over 25 g Mn fiber.

Finally, "Alvin Niskins" were recovered from one or several of the 1.25 L Niskin bottles on Alvin. On the ship, these water samples were drained from the Niskin bottles and slowly (<1 L/min) passed over 25 g Mn fiber.

After passing the water samples over the Mn fibers, the fibers were rinsed with Ra-free fresh water, then dried using a compressed air stream to a suitable humidity (Sun and Torgersen, 1998). The fibers were then counted immediately on a Radium Delayed Coincidence Counter (Moore and Arnold, 1996) for total Ra-224 and Ra-223 activity. The fibers were counted again after 3 weeks to measure the supported Ra-224 activity from any sorbed Th-228 on the fibers. This activity is subtracted from the total Ra-224 activity derived from the initial measurement to compute the excess Ra-224 activity (the activity reported in the dataset). Ra-226 activities were measured by sealing the Mn fibers in air-tight cartridges for ~1 week and measuring the accumulated Rn-222 (daughter product of Ra-226) on a radon emanation line (Peterson et al., 2009). Fibers were then counted an additional time on the Radium Delayed Coincidence Counter after ~1 year from collection for Ra-228 (as the change in Th-228 activity from the 3-week measurement; Moore, 2008). Analytical uncertainties are based on counting statistics (as 1-s standard deviation of the total counts logged, propagated through the activity computations).