All down to the odd pitch-roll coupling effects most have seen on the Lynx - and possibly down to a combination of the high AoA that the retreating side of the disc can operate at (thanks to Berp blades) and the high effective hinge offset that gives such excellent roll and pitch response.
Would that have also caused the following, where if accelerating from the hover or slow forward speed, purely by using the trim switch, the Lynx would roll in the opposite direction to what you would expect for inflow roll?