This thesis explores the way in which power relations are constructed in French primary schools, by posing the question of the reality of the "power" of school principals. Power is mainly understood in the Croatian sense, as the control of areas of uncertainty by the various actors involved. French research is scarce on the issue of primary school principals, even though they abound on the issue of school heads. But primary school is not a legally autonomous institution like college or high school. It therefore has no legal and financial autonomy. The director is not a "headteacher", he does not have the administrative status. The emergence of primary school in France, analyzed historically in the first part, shows a school model under strong hierarchical supervision, in a highly centralized system with significant bureaucratic and regulatory legitimization. This legitimacy is challenged by the emergence of new actors and new policies, with a view to greater efficiency, which translates into a demand to work as a "project" at the local level. The mistrust of many teachers in the face of what is received as an injunction, has repercussions both on the close (NEI) and distant hierarchy (circulars, decrees, changes in policies depending on the electoral hazards) but above all on the direction school responsible for implementation. This is put in perspective with different sociological models (second part) and in particular with the sociology of Crozier's organizations related to more recent models (Mintzberg, Boltanski among others).The triangulation of the methods allows to obtain a vision and data at the same time generalist, via a survey by questionnaire (N = 5 747 among which 2 211 directors of school), singular by an ethnography integrated in the daily work of a director d and complementary from interviews with principals (N = 15) and with the immediate hierarchy (Inspectors, N = 5) and decision-makers, up to the ministerial cabinet. Epistemologically and methodologically, this thesis assumes the complementarity of quantitative and qualitative methods (Part Three). This work shows the tension, in the time of a change of frame of reference, between the educational freedom and the individualism which results from it and a call to the work "collective" according to the forms of the "New public management". It also shows the hesitations between this model of "autonomy" and the local and a centralized pyramidal model. He examines the paradox of a non-hierarchical, status-free school leadership and the resulting psychologization of relationships. We examine both the point of view of the directors and the point of view of the teachers on the particularly important issue analytically of the status of the first who cleave the functions. An examination of an important body of work shows the reasons and the consequences (fourth part). An integrated ethnography and interviews of directors (fifth part) bring out more precisely all the issues and power conflicts in a specific team; the occupation of the premises, the use of the school cooperative's money, the relations with the parents and the extracurricular for example, completes the portrait of the director in "marginal secant". Interviews with the superiors complete the situation of principals (Part Six). The director is undoubtedly a "secant marginal", but his power "is limited, sometimes disputed. Assigned to the "dirty work" of policing, he is at the center of games and challenges of complex alliances in and out of his school. It is finally examined whether and under what conditions it can be or become an "ingenious actor" capable of training a team beyond its only charisma.