Williams have a unique rear end set up, with the rear wishbone and rear wing mounting to the same pylon. In early tests other drivers and media suggested the rear wing moving laterally and this was related to the lack of stiffness in this pylon. However Sam Michael has refuted these claims and looking at the rear wing set up, you can see where the movement came from.
Due to its low gearbox casing, the rear leg of the top rear wishbone mounts to a vertical metal extension of the gearbox. The structure then continues vertically in the form of a carbon fibre strut, which mounts the top rear wing. Normally teams will also support the beam wing with sizeable mount to the top of the rear impact structure. As part of Williams ‘waisted’ approach to structures at the back of the car, the beam wing ‘floats’ above the impact structure. Thus aero loads from this device pass upwards through the endplates into the top rear wing and consequently into the vertical pylon back to the chassis. In early tests the endplates were not able resist the lateral loads the beam wing exerted on the assembly and as a result the lower of the wing moved laterally. Williams had already accounted for this with a small metal stay (illustrated – yellow) between the vertical pylon and the beam wing. Such were the cornering loads around Valencia, this was unable to steady the wing. So Williams added links from the endplate to the diffuser. This greatly reduced the movement and no negative comments have merged about the structure around the back of the car since.