Systematic errors on the avalanche localization along a wire have been investigated and measured in a centroid chamber array described here. The use of discrete sampling by cathode strips may result in a systematic error which can be of the order of the chamber resolution. It is shown both by calculation and experiment that this error can be minimized by an appropriate choice of a "bias level" in the algorithm used for coordinate calculation. Another source of systematic error is also discussed.