Logisticregression models were also used to analyse the farrow-ing rate within each treatment according to the optimalinsemination period (between 0 and 24 h before ovulationor not), and the occurrence or not of semen backflow orbleeding during insemination.Oestrus length, the time of ovulation, the number ofpiglets born, and the number of piglets born alive wereanalysed using the MIXED procedure, with the means com-pared by the Tukey-Kramer test. In the models used for theanalysis of oestrus length and time of ovulation, WOI andoestrus length were respectively used as covariates.