This paper focuses on any biases that might emerge during a graded summative evaluation. After having gone over the functions assigned to this process, and recalled the variety of disruptions to the process brought to light by previous research, we wanted to better understand how the process was actually put in practice, in particular in ÉCLAIR institutions (écoles collèges lycées ambition innovation réussite – primary and secondary schools ambition innovation success, a combination of words which in French spells éclair or flash). To this effect we tried to disrupt the didactic relationship by giving fifteen Quatrième (Year 9) classes from a variety of secondary schools a standard maths topic, answering the requirements of the academic institution for this level. By studying the reactions of students and teachers during the different phases of this external evaluation we were thus able to understand the way they usually functioned. We have brought to light, in particular in ÉCLAIR institutions, a number of concessions susceptible of distorting the evaluation process. These manifestations of a joint reciprocal game at the heart of classroom interaction sometimes give the evaluation process as its only goal that of maintaining the didactic relationship to the detriment of any other function and particularly to that of eventually measuring the students’ achievements.