Figure 3. ²³⁴U(T) vs. age plot for nine of the samples from this study that are thought to have undergone U loss or gain. Filled symbols represent the measured values and are off the figure for Site 1005 samples. Curves are labelled with sample identifiers and are derived by progressively subtracting/adding U from/to the measured (²³⁰Th/²³⁸U) ratio. The lower panel shows an approximation of sea-level constructed by scaling the marine-oxygen isotope curve (Imbrie et al., 1984) so that the most recent deglaciation corresponds to 120 m of sea-level rise. Because the oxygen isotope curve also contains a temperature component, this is not an accurate indicator of past sea level but it is useful to highlight the times in the past at which sea level has come close to its present-day value. These periods are indicated by gray shading and marine-isotope-stage numbers on the lower panel, and gray bars on the upper panel. Because the samples are known to have formed during sea-level highstands, the intersection of the sample curves with highstand intervals enables each sample to be corrected for U loss or gain and assigned to a highstand event (see text for full discussion). All five samples from Site 1005 showed loss of U, with the uppermost sample having lost the most (percentage loss is shown after the sample identifier). The remaining four samples showed slight U gain. The intersections between the curves and highstand events define a broad trend toward lower ²³⁴U(T) with increasing age as indicated by the dashed arrow. This is due to alpha recoil of ²³⁴U from the U-rich aragonite material analyzed here. Of the nine samples not on this figure, four yielded Holocene ages and five were sufficiently old that no useful age information could be gleaned from the U-Th system.