This study was designed to determine whether regular Head Start teachers trained as "experimenters" could produce two different teaching modes (combinations of techniques) in two consecutive sessions, using the same lesson content. Each of six Head Start teachers was asked to teach a 10-minute classification lesson to four children, using a set of assorted objects and teaching in a specified mode (e.g., Giving to the Group). Then, with another group of children, the teacher was to replicate the lesson, shifting to her second targeted mode (e.g., Asking Individuals). The problem to be investigated was whether teachers trained to observe, classify, and modify their own teaching behavior would be able to shift from one mode to another, if lesson goal and materials were kept constant. Also of interest were questions related to variability of teacher behavior within treatments and the general problem of experimental control in classroom studies. Results indicated that despite individual variability the teachers were quite effective in altering their behavior on the experimental variable. They exhibited a high level of performance both in producing their assigned modes and in shifting from one distinct mode to another. (Author/CS)