This note looks as some time series data taken using a scope on BPM A15 during a single proton bunch store. This study shows that, in order to obtain a precise position measurement it is important to look only at the data in the neighbourhood of the bunch. If we integrate a full turn, the signal to noise is poor. I look at the stability of the position measurement over 763 turns and try two different methods of determining the position averaged over many turns: 1) Compute independent turn by turn Fourier transform and average the magnitudes from each turn. 2) Compute the Fourier transform over many turns. Both methods give similar signal to noise but the first method is simpler in other regards. The repeatability of the method is about 10% for single turn measurements and 1.2% for 64 turn averages.