Non linear growth functions were fitted to weight-age data of 40 Santa Inês and Bergamacy lambs. The Hartley and Bartlett tests indicated that the data presented heteroscedastic variance, so the growth functions were fitted through weighted minimum square method by the inverse variance weight in each sampler using the WEIGHT option in PROC MODEL (SAS INSTITUTE, 1996). The functions were compared using the biological interpretation of the parameters and the quality indicators of fit (adjusted determination coefficient, mean square error and Durbin-Watson test). The von Bertalanffy and Gompertz functions provided the best fittings, whereas Richards, Logistic and Brody functions turned out not to be suitable for describing growth behavior, became estimates based on them do not agree with observation.
growth functions; weighted regression; lamb