The performance of the differential pulse-width pair Brillouin optical time domain analysis (DPP-BOTDA) is evaluated experimentally using either the gain from log normalization or linear normalization for the subtraction of traces collected with pump pulses of slightly different pulse widths. Using pump pulses widths of 43 ns and 40 ns, amplified Brillouin time domain probe traces were obtained for 10 km of standard single mode fiber. Two hotspots of length 30 cm and 6 m, separated by more than the spatial resolutions of the individual pulses and kept in a temperature controlled hot bath facility, were interrogated with temperature variations from 5 to 70°C, having probe signal gain of ~ 40% at the Brillouin Frequency Shift (BFS). This research work demonstrates, for the first time, that the use of linear gains for the subtraction step in creating the Brillouin gain spectrum, produces results for small to medium Brillouin frequency shifts (≤30 MHz), that deviate from the results of the subtraction of the logarithmic gains by as much as 2 MHz (~ 2°C), particularly for hotspots of the order of the spatial resolution of the DPP-BOTDA. For hotspots longer than the spatial resolution of the technique, the difference between results of the two processing methods show BFS deviations only at the end of the hotspots.