The bacterialcountswerelog10-transformedpriorto
analysistoensurenormalityofdata.Statisticalanalysesof
growthperformanceweresimilartothatinexperiment1.
The effectofweaningage(3,4and5weeks)wastestedin
a randomisedcompleteblockdesignwithgender,ancestry
and weaningweightasblockingfactors.Thepairmean
wastheexperimentalunitforgrowthperformanceana-
lyses.Forbacterialcounts,theindividualpigwasthe
experimentalunit.Theresultswerepresentedasleast
squaremeans7SEM. TukeyStudentizedRange(HSD)test
wasusedformeansseparation.Theprobabilitylevelthat
denotessignificancewas Po0.05.