For calibration of a large capacity gas flow meter, a sonic nozzle bank may be used as a reference system. International standards (ISO9300:1990) allow installation of a single nozzle in a meter tube as a flow transfer standard. For multiple nozzles in a single tube, the effect of interference between sonic nozzles and the chamber wall must be measured to predict the discharge coefficient of a nozzle array from those of single nozzles. The interference effect between neighboring nozzles can be additional error sources in mass flow measurement. Sonic nozzles with three different throat diameters (d = 4.3, 8.1, and 13.4 mm) were tested in a single meter tube in three geometrical arrangements. The mass flow rate was measured against a primary gas flow standard system. Three installation plates for sonic nozzles were made to vary the distance between nozzles and distance from the chamber wall. Discharge coefficients of the three individual nozzles were in agreement with the ISO recommended equation within +/- 0.2%. Discharge coefficients of the nozzle bank calculated from those of the individual sonic nozzle were the same as the direct measurements within +/- 0.098% at the 95% confidence level for all cases. For these experiments, the results were not influenced by the proximity of the tube wall or the interaction of the nozzles. (C) 1999 Elsevier Science Ltd. All rights reserved.