Paleotemperature reconstructions of Quaternary climate are difficult to obtain for terrestrial regions due to the lack of long, continuous, well-dated records. U-series-dated speleothems formed at isotopic equilibrium can be used to provide records of paleotemperatures based on calcite-water oxygen isotopic fractionation. The calculation of paleotemperatures using requires knowledge of the d¹⁸O of the drip waters from which they were formed. The isotopic composition of cave dripwaters have been found to be equal to the average annual d ¹⁸O and d D of precipitation (Yonge et al .1983). Fluid inclusions in spelothems preserve the d D of primary drip waters; therefore d ¹⁸O can be calculated using the relationship d D = 8d ¹⁸O + 10 valid for meteoric waters (Dansgaard 1983). An improved method of extracting fluid inclusion water from speleothem calcite allows the measurement of d D for sub-microliter volumes of water. This can give a resolution of less than 500 years for a fast growing speleothem.

Two speleothems from Reed's Cave, Black Hills, S. Dakota gave dissimilar records of d ¹⁸O of calcite, but recorded similar temperature changes for the time period 62,000 to 49,000 years ago. This indicates that d ¹⁸O records for some speleothem calcite may not give direct information of climate variations. Although the early part of sample 99902 was coeval with the record in sample 20000, the former record shows very little response to climate in contrast to the high resolution oxygen isotope for 20000. A full record over the Wisconsin glacial period from sample 99902 gives an interglacial to glacial temperature decrease of 10 ° C in the mid-continent. This is consistent with groundwater and faunal temperature reconstructions from the Colorado Plateau where temperature decreases are estimated at 5-10 ° C (Anderson et al. 2000). Our record also shows that significant (up to 80% of interglacial-glacial temperature difference) short-term temperature variations can occur on the continents.