For many years we have lived with the statement that the system resistance is proportional to the flow rate squared i.e. that it is a quadratic relationship. Whilst this is almost true for small changes in the flow, its shortcomings are beginning to be realised in an era of inverters and other methods of wide ranging speed control. System air velocities are often in the transitional zone, such that Reynolds Number effects have to be considered. The wider the range of flows, the bigger the observed discrepancies. This paper gives the theoretical reasons for the breakdown of the ‘square law’. It also notes that there is no intrinsic loss coefficient for a given duct fitting. This document is available in English.