One of Aksel Wiin-Nielsen's early contributions in the field of dynamical meteorology was his suggestion to solve the advection problem in a semi-Lagrangian manner. In this presentation I will take his Tellus (1959) paper as a starting point and give an overview of how influential that way of solving the transport problem turned out to be many years later. In recent years the so-called on-line chemical coupling into dynamical atmospheric models may revitalise the semi- Lagrangian idea: one possibility to be presented here is to set up models including both an Eulerian and a fully Lagrangian representation of all the prognostic variables. In this way the numerical dispersion and diffusion can be reduced and replaced by a more physically based dissipation. The performance of the new hybrid Eulerian-Lagrangian locally mass conserving scheme is tested in a 2D dynamical model including full chemistry.