We describe how reference filter characteristics (unexposed weights) can influence instrument-reported attenuation (thereby black carbon concentrations) using the OT21 (SootScan Optical Transmissometer), in the case of Emfab filters. Reference filters (unexposed filters) with higher unexposed weights (pre-weights) compared to that of sample filters are found to reduce the instrument-reported aerosol sample attenuation measurement; reference filters weighing less than the sample filter results in the reverse outcome. This sensitivity can lead to significant under- and over-estimation by the instrument (estimated mean absolute error: similar to 10 and 16 ATN units for infrared (IR) and ultraviolet (UV) channels, respectively, in our tests, which represent 44% and 15% errors, respectively). We find that the error is linear with respect to difference in the unexposed filter weights of the sample and reference filters. We report two equations (one for IR, one for UV) for correcting the Transmissometer output based on experimental data, which allow use of any unexposed Emfab filter as the reference filter; application of correction parameters to data here reduces the error to 2 and 3 ATN units, respectively, for IR and UV (10% and 3% errors, respectively).