SUbtMARY EMF measurements in cells Ag~AgCl~NaCl(~')~C_E_Memb~ne~NaCl(t~~~)~ AgClIAg using a new flow-cell zr&md and a membrane-strip method are reported. Elimination of concentration polarization effects in the flow-cell is demonstrated from the additivity of potentials of various concentration cells, and from comparison of the resuhs of the two measurement methods. Co-ion transport ntumbers (relative to the solvent) are calculated in the region 0.00 < p_PP -c 0.04. It is shown chat the flow-cell method used will give results with an accuracy between S",/, and 20% for P_! in this cast of high co-ion exclusion, and provides a re!iable and convenient method for selectivity determinations. MTRODUCTION The determination of the selectivitj of an ionic membrane by measuring the EMF of a concentration cell using the membrane as separator has become a routine procedure. Until recently comparisons between transport numbers measured with a Hittorf method and transport numbers cakutated from membrane potential and electro-osmotic flow measurements were unsatisfactory (1-8). This has led many investigators to use the Hittorf method rather than EMF measurements for selectivity determinations, Recently Dawson and Meares (9) obtained reasonable agreement ktweea the two types of measurement for a homogeneous cation-exchange membrane. Especially in the region of high co-ion exclusion both the EMF method and the Hittorf method can be expected to yield only moderate or low accuracy for the transport number of the co-ion. For instance when t+ = 0.99 a 1% error in the t+ determination of course results in a 100% error in t-. However, in this case f _ is the quantity of interest e.g. for co-ion mobility calculations and salt diffusidn predictions. Therefore in the case of anion excfusion, measurements ad theory should be compared on the basis of 1 -t+ rather than t+, and measurement Desalination, 11 (1972) 61-69