We start from the assumption that school mathematics knowledge could be better explained if social practices were considered to be generators of knowledge. This perspective changes the way we look at what school mathematics knowledge is and what it takes to teach and learn it. In this article, we will present a teaching situation about periodic functions, which was designed with this perspective in mind. The design was based on the assumption that the scientific notion of periodic function is related with the social practice of prediction. In the situation, prediction as a social practice is transformed into a situational line of argument which redefines that which is periodical. The situation brings into play meanings for the repetition of a movement, which takes place in time in the context of graphs of functions. Our analysis of the situation will focus on the prediction tools that participants generated in order to define that which is periodical. We will conclude with some implications of our observations for the teaching of mathematics.