The mechanisms involved in cloud formation affect the resulting cloud radiative properties. Therefore, accurate cloud formation models and related global warming models require freezing data for ammonium sulfate particles, a major constituent of upper tropospheric clouds. Homogeneous ice nucleation in aqueous ammonium sulfate particles has been studied by several laboratory groups; however, there is significant disagreement in the reported data. In the present work, the ice freezing temperatures are measured by sending (NH4)2SO4 particles, generated by vapor condensation, through a temperature-controlled flow cell, where ice formation is detected with infrared spectroscopy. To determine possible explanations for disparate results among several laboratories, the experimental methods used in this study are closely examined. Experiments are performed to preclude artifacts, and experimental conditions are varied to develop a complete understanding of processes involved in particle freezing. All experiments produce consistent results. Mass transfer calculations for the aerosol flow tube technique employed in the current study suggest that one frozen particle in 104 to 106 may scavenge sufficient water vapor from surrounding supercooled aqueous particles such that a single ice particle accumulates enough mass to yield a positive freezing event in the infrared spectrum. This result appears to partially explain the confusing and conflicting results coming from several laboratories employing different experimental methods to observe homogeneous ice nucleation in ammonium sulfate. Six lower limit J (cm-3 s-1) values are determined as 1.1¿106(233¿1 K,x=0.013¿0.001), 1.1¿106(231¿1 K,x=0.029¿0.004), 1.7¿106(227¿1 K,x=0.066¿0.015), 2.8¿106(226¿1 K,x=0.071¿0.018), 8.4¿106(224¿1 K,x=0.097¿0.034), and 8.4¿108(221¿1 K,x=0.153¿0.076), where the associated temperatures and solute mole fraction aqueous compositions (x) are noted in parentheses. ¿ 2001 American Geophysical Union