Commercial spinning rotor gauges (SRGs) use a special linearization procedure to compensate for the pressure dependence of the rotor deceleration rate in the transition regime from 0.1 to 100 Pa. These procedures have been found to have large errors above 10 Pa [J. Setina and J. P. Looney, Vacuum 44, 577 (1993)], but can be significantly improved. An extensive set of experimental data of rotor deceleration rate versus gas pressure up to 130 Pa was acquired. A group of six SRGs and four gases (N-2, Ar, He and H-2) were used in the study. Temperature measurements were also included to account for heating effects. The data show that the differences between rotor/thimble combinations are large enough to cause differences of several percent if one uses the linearization functions currently in use without adjustable parameters. To get the best accuracy, one parameter is left in our linearization procedure to be determined by calibration. We call it the Knudsen length of the rotor/thimble assembly. This is the second calibration constant of a SRG that needs to be determined for accurate pressure measurements above 1 Pa. A method to determine the second calibration constant at 100 Pa is proposed. The first calibration constant is the already well-known rotor accommodation coefficient and is determined at pressures less than 0.1 Pa. The error in reading of the new linearization procedure with the two calibration parameters is less than 2% over the entire range from molecular regime up to 130 Pa.