Reconstructions of monthly mean sea-level grid-point pressure from station pressure data over the Arctic region using an objective technique have revealed two serious errors in the gridded data. First, all gridded data (from the U.S. Historical Weather Map series) for the Arctic Ocean area away from the North Atlantic sector appears to be between 4 and 6 mb too high before about 1931. The semi-permanent Arctic high produced in the early historical map series was the result of a lack of basic station data, and the belief amongst many North American meteorologists of the 1920s and 1930s of the existence of a polar or glacial anticyclone. The second error affects only north-western North America for the years 1899-1909. Here pressure is again too high compared with reconstructed data. The cause of this error is not immediately apparent, but is probably the result of an erroneous correction to sea level for stations in the region. Both errors seriously affect the homogeneity of the gridded Arctic series. The objective scheme used to reveal the errors, can also be used to derive correction factors for the gridded data prior to 1931.