If you have taken only a single measurement then often you will use +/- 1 scale division of your measuring instrument as an estimate of uncertainty. However, when using a stopwatch with a precision of 0.01s this will give an unrealistic estimate of uncertainty. Human reaction time is at least 0.1 s and probably more like 0.2s for most people. Therefore in your case I would estimate the uncertainty as 0.2 s and determine percentage uncertainty using (0.2/9.5)x100 = 2.1%. If you time for a greater number of oscillations then clearly you will measure a larger time than 9.5 s so your percentage uncertainty would decrease.