There is certainly strong evidence that speed is a factor in a snake. Bailey have done quite a bit of research in to this and I am sure that they are not alone. A couple of years ago they had a model car and caravan running on a rolling road at the NEC show. They could simulate caravan loading and speed. This was a really good demonstration and graphically showed that the loading and speed were the main factors in an outfit becoming unstable. It also needed an input to set the snake in motion. With the model this was a small flick of the hand, but it was explained that in the real world this could be a gust of wind, being overtaken or a small swerve, all made worse if travelling downhill. Up to a certain speed, dependant on loading, the caravan would correct itself but as the speed became higher the wiggle on the caravan became larger until a speed was reached where the caravan became completely unstable and the outfit flipped. This really showed that at a critical speed, if there is a snake, there is no chance of recovery. This was a really eye opening demonstration and I wish Bailey would show it again.
I had a horrible journey up to Caen from Southern Brittany last Monday. I've never towed in a headwind like it. At one point I thought the car was breaking down but it was just the increased drag. I kept my max speed all the way to 50 and at times slower but some gusts still blew us about a bit. Definitely pays to be careful in high winds.