Birds (n 169) which varied in age, live weight, nutritional history, physiological state and genotype were slaughtered and analysed for total body water. Before slaughter, birds were injected with the water isotopes tritiated water (TOH) or deuterium oxide (D2O), or both, to determine TOH space or D2O space, or both, as estimates of total body water in vivo. 2. At the mean total body water of all birds determined by desiccation, of 1096.4 (SD 424.1) g, TOH space and D2O space overestimated total body water by 10.4 and 8.5% respectively. The difference between the isotopes was significant (P less than 0.05). 3. Based on recovery of isotope it was postulated that the main reason for the observed overestimation of total body water in vivo was incomplete recovery of isotope due to the vacuum sublimation technique. The mean recovery (%) of added isotope to whole blood after vacuum sublimation was 93.0 (SD 2.6) and 92.4 (SD 5.5) of the theoretical concentrations of TOH and D2O respectively. 4. Nevertheless, accurate prediction of total body water was obtained from regression equations which included live weight and isotope-dilution space. Values required logarithmic (base 10) transformation before derivation of linear and multiple linear regression equations, and the precision of prediction was determined by the residual standard deviation (RSD). 5. Total body water could be predicted with nearly equal accuracy from live weight or isotope-dilution space (RSD 0.025 and 0.020 respectively). Prediction of carcass protein was more accurate from live weight (RSD 0.033) than from TOH space (RSD 0.036), and inclusion of both variables resulted in only a marginal decrease in RSD to 0.031. 6. The prediction of carcass fat and energy was markedly improved by the inclusion of isotope-dilution space in conjunction with live weight compared with live weight alone. 7. The relations show the developmental nature of body composition of domestic fowl given diets adequate in nutrients. The prediction equations demonstrate the precision possible for studies in which estimates of body composition in poultry are required without slaughter.
