Within each 30 d period, data collected during the last 2 weeks of treatment were analyzed using MIXED for analysis
of repeated measures and GLM for non continuous variables (SAS, 1999). Cow was used as the experimental unit for all
variables analyzed except for DM intake where the experimental unit was pen. Dependent variables were milk yields,
milk components and vaginal temperatures. Independent variables included treatment, parity, period, and the appropriate
interactions. Cow nested within sequence was random, with others fixed. The level of significance was accepted as P 0.1) then they were removed from the model.