Thirty-five azooxanthellate (non-photosynthetic) corals belonging to 18 species were collected at sites ranging from the Norwegian Sea to the Antarctic and of depths ranging from 10 to 5220 m. All specimens showed distinct, well-defined linear correlations between carbonate oxygen and carbon isotopic composition, with slopes ranging from 0.23 to 0.67 (mean 0.45 ± 0.9) and linear correlation r2 values that averaged 0.89. These pronounced isotopic disequilibria have, to date, rendered azooxanthellate corals unsuitable for use in paleothermometry. Most, but not all, of the heaviest skeletal δ18O values reached or approached equilibrium. If the isotopically-heavy ends of the δ18O vs δ13C regression lines reliably approximated isotopic equilibrium with seawater, these values could be used to estimate the temperature of the water in which the coral grew. The δ13C values of the heavy ends of each line, however, were always depleted compared to carbon isotopic equilibrium with ambient bicarbonate by varying amounts.
Despite the disequilibria, a reliable method for obtaining paleotemperature data was obtained. It was found that, if a δ18O vs δ13C regression line from an individual coral could be generated, the δ18Oarag value corresponding to δ13Carag = δ13Cwater and corrected for δ18Owater was a linear function of temperature: δ18O = −0.25 T(°C) + 4.97.