Three series of experiments were conducted to determine the relative bioavailability of various phosphate products with an emphasis on the defluorinated phosphates. In Series I, phosphorus in commercial mono/dicalcium phosphates, commercial dicalcium phosphates, commercial defluorinated phosphates, and experimental defluorinated phosphates were found to be 93, 81, 70, and 76% as available as that in dicalcium phosphate dihydrate, respectively, using 3-week bodyweight and toe ash in a nonlinear regression analysis. In Series II, phosphorus in a commercial defluorinated phosphate was compared directly with that in a commercial dicalcium phosphate using poult from 1 to 4 weeks of age. No significant differences were found between the two products which may be attributed partly to the feeding of the starter diet during the pre-experimental period. The commercial defluorinated phosphate was numerically less available than the commercial dicalcium phosphate in all cases. In Series III, phosphorus in various defluorinated phosphates was compared with that in a dicalcium phosphate dihydrate. Ash of dry unextracted tibia, ash of dry toe, and 0 to 3-week body weight gain as responses to added phosphorus followed a sigmoidal regression model while 0 to 2-week body weight gain followed an asymptotic model in Experiment 1. The defluorinated phosphate was found to be 53% as available as the standard. In Experiment 2, an experimental defluorinated phosphate from Series I which was found to be 76% as available as the standard was found to be 62% as available as the standard used in Series III, indicating differences in bioavailability of the two standards. In Experiment 3, two commercial defluorinated phosphates were found to have equal relative biological values, 67.2 and 67.6% as compared with the standard using poults from 0 to 2 weeks of age, and three experimental defluorinated phosphates 51.7, 74.0, and 88.2% with the product with the fine particles having the highest relative biological value. In cases where the sigmoidal regression gave a significantly better fit to the data than the asymptotic model, the relative biological values and the associated confidence limits from the asymptotic model were equal to those from the sigmoidal model. / Ph. D. / incomplete_metadata