This study describes some of the differences in the way primary teachers work with different school subjects. The activities during reading and arithmetic instruction in six third grade, inner-city, self-contained classrooms were observed, tape-recorded, and analyzed using three systems of observation. A total of five hours of instruction in reading and three hours of arithmetic instruction were submitted to a series of content analyses. The major differences discovered included a marked "negativity" during reading instruction. This reason was hypothesized to lie with the intrinsic difficulties in teaching the subject relative to the more "transparent" characteristics of arithmetic. In addition, reading was also characterized by a greater degree of "control" by the teacher and at the same time, arithmetic instruction was characterized by more student-initiated interactions. The discovery of inner-teacher differences correlated with differences in the primary curriculum calls into question the interpretation of criteria from observational systems such as Flanders' Interaction Analysis Categories as measures of internal conditions (e.g., personality) in the teachers. The meaning of at least some criterion activities is more probably that these reflect some external forces acting on the classroom, particularly the curriculum. (Author/RC)