In the state of Vaud in Switzerland, the continuing education of primary school teachers faces many challenges, including that of taking into account the real activity of professionals. It seems, however, that the efforts made do not meet the expectations of the system, particularly in terms of improving student performances. Two factors could contribute to what we might call the relative "failure" of continuing education. The first is that training focuses on the teacher and less on teaching, analysed as a system. The second factor is that the tools offered by the training are often described as unusable or difficult for users to use. Sometimes considered by practitioners as additional prescriptions to their work, they function more like artifacts than real instruments, they are sometimes ignored, sometimes rejected.

These findings lead us to take an interest in the work developed by the sociology of organizations about the institution as the unit to consider in building the meaning of change (Gather Thurler, 2000). Indeed, some beliefs in the institution gives it a specific role, particularly from the point of view of its improvement through lasting changes. In addition, recent works on the management of schools in favour of academic success insist on taking into account (1) the way in which leadership focuses on students' learning (instructional leadership) and (2) the way in which this leadership is distributed among all the actors of the school (distributed leadership).

The research carried out so far in the Lausanne Lesson study Laboratory (3LS) provides a better understanding of the impact of LS on teachers' professional development in relation to students' learning. Now we want to broaden our understanding of LS by studying the role of trainers, teachers and principals during the implementation of a LS. Who provides leadership during a LS? In what way? Is this leadership about teaching-learning? And finally, how do actors collectively influence or even transform the culture of the institution?

Our presentation will analyze data collected during a lesson study implemented in an establishment. Our qualitative data have been collected in an inductive approach inspired by the grounded theory (Glasser & Strauss, 1967). More specifically, the observations in which we participated were supplemented by comprehensive interviews with the teachers and principals involved. These data were analyzed using the method of cross-induction categorization of variations and regularities (Glaser & Strauss, 1967; Maulini, 2013).