This paper presents a complete analysis of the theory of an instrument to measure the diffusion coefficients in liquid mixtures based upon the phenomenon of Taylor dispersion. The analysis demonstrates that it is possible to design an instrument that operates very nearly in accordance with the simplest mathematical description of the dispersion of a solute pulse in a fluid in laminar flow within a straight, circular cross-section tube. The small departures of a practical instrument from the ideal are evaluated as corrections by means of a general perturbation treatment that allows them to be examined one at a time. The corrections considered include the effects of the finite volume of the injection pulse, the finite volume of the concentration monitor, the coiling of the tube, and the nonuniformity and noncircularity of the cross section, as well as the variation of the fluid properties with
composition. All the equations necessary for the design of an instrument of this type, and for the evaluation of experimental data free from significant systematic errors, are presented.