A previous study by the authors indicates that historical wind speeds may exhibit a statistically significant trend attributable to a changing climate. While data quality control was scrutinized, uncertainties from several causes were identified. These include changes of observation averaging time, data standardization for terrain effects and other large scale cyclical effects such as El Nino. Each of these factors deserves investigation. As a following up study, this paper focuses on the first factor. The wind speed archived by Environmental and Climate Change Canada is reported as a "nominal" hourly mean wind speed at each hour. This nominal hourly mean wind speed is not an average over an entire hour. The most widely used dataset, HLY01, began by recording the one minute mean wind speed before the top of each hour, but changed to recording the two-minute mean wind speed before the top of each hour in 1985. This would indicate that the wind speed data prior to 1985 could be biased compared to those after 1985. This paper studies several stations in Ontario to investigate the impact of the change of averaging time on the detection of a trend in historical extreme wind speeds. The ratio of one-minute wind speed to two-minute average wind speed is treated as a random variable and Monte Carlo simulation is employed to randomly correct the one-minute wind speed to two-minute mean wind speed before extracting the annual maximum wind speed. Trend analysis is then applied to detect trends in the corrected extreme wind speeds. The impact of the change of averaging time on detected trends is discussed.