The search of the sea floor for missing Malaysia Airlines flight MH370 is nearly concluded (at the time of writing) but no trace of the plane has been found, leaving open the question of whether the sea floor search has been conducted in the right place. Meanwhile, several more parts of the plane have now been found in the general vicinity (SW Indian Ocean) of Ile de la Reunion where the flaperon first washed up in July 2015. This talk will present the results of an attempt to re-assemble the found parts of the plane, using a high-resolution global ocean model (that assimilated all available satellite altimetry) to back-track the plane parts to the time of the crash, thereby providing an independent estimate of the location of the crash. Ocean turbulence and various uncertainties might seem to make this a futile attempt but we find that this is not completely true. In particular, there is much uncertainty over the drift characteristics of the flaperon. To reduce that uncertainty we constructed a replica, fitted it with GPS, and deployed it alongside oceanographic drifters to measure the direction and magnitude of its windage coefficient and response to wave forcing.