A perturbation viscometer is a flux response technology device that measures gradients on the viscosity-composition curve of a binary mixture or more generally on the viscosity-composition surface. It provides a method of calculating relative gas viscosities very precisely. In conventional perturbation viscometry a small perturbation flow of one component of the gas mixture is added to a main flow of known composition flowing through a capillary. The pressure upstream of the capillary changes sequentially due to the change in flowrate and viscosity of the mixture. The ratio of these pressure changes gives the gradient of the viscosity-composition curve. Integration of the viscosity gradients across the composition range yields the relative viscosities of the mixtures. This present paper gives new theoretical treatments for large perturbation flows up to 150% of the main flowrate. Corrections to the theory for small finite perturbations are proposed that account for the differences observed when larger perturbations are used. These correction terms are used to demonstrate how the logarithmic viscosity gradient-composition curve can be found by adding progressively bigger perturbations to each of the two pure components. A second method is proposed which demonstrates how the viscosity ratios can be found directly from the large perturbation measurements. Experimental studies of the mixture argon-nitrogen at 25 degrees C and 1.1 bar are used to validate the new theoretical treatments. Results using both large and small perturbation measurements permit direct comparison between the different methods. In addition, the large perturbation data are used to define the optimum perturbation size for use with conventional finite difference theory. (c) 2005 Elsevier Ltd. All rights reserved.