strip comprises eight microtiter wells (300 ml total volume per well) which are detachable into single CO2-trap wells. They can be reassembled in a 96-well plate carriage to be read when required. All samples (with and without CO2-traps) were then sealed and introduced into a TAM Air calorimeter (TA Instruments Sollentuna,sweden) and heat production and CO2 evolution were measured over 5 h at 25 C. A schematic representation of one reaction vessel with a CO2-trap is given in Fig.1. Heat outputwas recorded between 1.5 h and 5 h as the calorimeter requires at least 45 min until the heat signal can be considered to be correct. For CO2 measurements, the CO2-traps were read immediately before and after 5 h incubation at 25 C with a plate reader (SPECTRAmax® Plus384, Molecular Devices, Wokingham, U.K.) at 572 nm. A calibration curve of absorbance () versus headspace equilibrium CO2 concentration (y) in reaction vialswas fitted to a power decay model (R2 ¼ 0.99) as follows: y ¼ 3.19x3.0. Preliminary work showed that CO2 production rates were constant throughout the 5 h incubation interval (data not shown). In total 96 samples were examined but only 24 samples could be analyzed simultaneously as only three TAM air calorimeters with 8 channels each were available. Therefore, the experiment was staggered over 4 days, i.e. samples and treatments were randomized and incubated in four separate calorimetry runs. The mean values are reported for the four technical replicates of each treatment (i.e. Milli-Q water with or without CO2-traps, Dglucose with or without CO2-traps) from the six long-term field experiments. The Student's t-test for independent variables was used to calculate differences in heat output rates between samples without and with CO2-traps as well as for differences in calorespirometric ratios between samples amended with Milli-Q water or glucose, respectively (Statsoft, 2000). Heat output ranged between 17 and 152 mJ g1 soil in Milli-Q water amended soils and increased to 306e1129 mJ g1 soil in