Absolute uncertainties are hard to estimate. They depend upon what you measured and how well. You need to evaluate your own experimantal technique and equipment to do it. I would suggest trying to achieve two significant figures in estimating the relative error.
