Abstract
The concepts of rigid body and reference frame are so deeply intertwined in newtonian physics that it is impossible to distinguish clearly between them. As a matter of fact, a reference frame is a rigid body taken as a basis for surveying. Consequently, the so called 'angular velocity' can be understood as a quantity relating two arbitrary reference frames. But it also appears as a typical concept in rigid-body kinematics. The purpose of the paper is to try to clarify both aspects, which the authors believe are obscure and mishandled in most textbooks on mechanics, by using a quaternion formulation for rotations as a pedagogical alternative to the usual matrix version.