“…The flow is isotropic if the Leidenfrost material levitates above a flat solid, but it can be rectified on asymmetric patterns, as shown in the context of flows against ratchets ( Jiang et al 1998, Yang et al 2004). For thin films (h < a), it was shown experimentally (using tracers) and numerically that the vapor flow is generally a lubrication flow (Cousins et al 2012, Dupeux et al 2011b, in which the gas is directed toward the deepest part of each tooth (Supplemental Video 13), then hits the step, and escapes laterally along it. Hence the pattern makes the vapor flow cellular (there is nearly no gas exchange between two successive teeth), and rectified, in the direction chosen by the levitating body: Owing to the vapor viscosity, this flow can drag the material above (Cousins et al 2012, Dupeux et al 2011b, Linke et al 2006, which suggests a traction force F scaling as (η v U/d )R 2 , where d is a typical vapor thickness (between h and a).…”