Experiments were carried out for functional evaluation of water-based roller ball. In an ideal pen, ink consumption y was considered to be proportional to writing distance M. Experimental apparatus was built, pen function was evaluated with the usage conditions as a noise factor, and the S/N ratios of four manufacturers'pens were compared. The results made it clear that functional evaluation could not be carried out on the basis of the S/N ratio alone, but suggested that pen fuction could be evaluated in a comparatively short time by comparing the results of 100-meter writing tests. It became evident that function should be clearly defined in terms of the contribution and variation of each evaluated condition to the S/N ratio and the linearity L, considering the conditions under which the user will use the pen, and that the evaluation should be made by comparison with the defined function.