Also depends on number of significant digits being measureed in the environment you are in.
I've seen desk fans cause significant variation in output from scales when measureing injection moulded parts.
I've seen thermal effects (heating of the strian gauges) cause significant variation also with injection moulding.
I've seen steady rise in mass when dry materials are measured in humid atmosphere.
If the equipment is properly connected, then you need to examine the "system" you are using materials/method/operators/environment/equipment/maintenance to find where the deviations are originating.
__________________
Just an Engineer from the land down under.
It looks like noise (elctrical noise) How close are your cabling to large motors/machinary. It may be harmonics. Also check that all your earths are in place and tight. Your earth point on the scales may be on the same earth loop as the computer. It is better for the scales to have their own, separate, well grounded earth point. If you only have one transformer for your plant, check all cabling for earth and screening.