Examined Type I error rates and power for four tests for treatment control studies in which a larger treatment mean may be accompanied by a larger treatment variance and examined these aspects of the independent samples "t" test and the Welch test. Evaluated each test and suggested conditions for the use of each approach. (SLD)