This paper presents an analysis of a teaching experiment with seven high achieving upper secondary school students in Sweden focusing on the de Villiers' discovery function of proof. The aim of the experiment was to test if it is possible for students to get insights to, and use, this function. The data consists of a tape recorded introductory pass, students' group work and the final discussion together with the students. The results show that the students did get some insights about the function. However, it was difficult for the students to construct the original proofs in order to use them to discover new results. The paper also shows that the function of discovery needs to be explored and clarified, as there are different interpretations of it in our field.