There is a paucity of experimental data on calcite precipitation from waters at low ionic strength and low ratios of Mg/Ca and Sr/Ca, using controlled and constant precipitation rates. Such data are particularly needed for studies of speleothem geochemistry in relation to palaeoclimates.We carried out a series of experiments using a karst-analogue set-up in a chamber of constant temperature and 100% humidity. A steady flow of NaHCO3 and CaCl2 solutions at PCO2 around 10−3.2 were mixed just before passage through a tube (analogous to a soda-straw stalactite) and allowed to drip onto a surface, analogous to a stalagmite. Growth rates were comparable with linear extension rates of natural speleothems.Analytical spots gave reproducible analyses in later analytical cycles after ablation of surface calcite with Na and Mg contamination. Different crystals from the same experiment tended to show positive covariation of Na and Mg with negative covariation with Sr. This may be due to the presence of growth hillocks with vicinal faces with differential partitioning behaviour.The result for the partition coefficient for Mg (DMg) at 25°C is 0.031 ± 0.004, which is quantitatively in good agreement with the trends of previous workers. At 15°C, the result is 0.019 ± 0.003. The temperature dependency is higher than experimental data on seawater-analogue solutions, but lower than a previous estimate based on a comparison of speleothem chemistry with single water analyses.Data for DSr are mainly in the range of 0.057 to 0.078, with a possible weak dependency on growth rate, consistent with previous experimental work. Absolute values are higher than studies in Mg-free saline solutions, which is attributed mainly to salinity effects. Values of DSr are nevertheless somewhat lower than in natural caves, which may relate to crystal growth factors.Mg partition coefficient values should allow robust determination of solution Mg/Ca compositions in enclosed caves, which are at constant temperature on the decadal timescale. The inferred sensitivity of DSr to growth rate factors implies that Sr values should be interpreted more cautiously. Muted changes could relate entirely to growth rate variations, whereas changes of large magnitude imply a control by solution composition. The absence of local (tens of micron scale) antipathetic variations in Sr and Mg in studied natural speleothems, implies that intracrystalline zoning phenomena, if present, are on a finer scale in those natural materials compared with experimental products.