I was hit at an estimated 40mph on A9 at North Kessock in 1983. Fortunately I threw the bike over towards the EMPTY outside lane and cartwheeled down the offside of the car. The core impact was between my bum and the A pillar. I watched the white lane line go past just below my head, really glad I wasn't wearing a helmet, which would have caught the tarmac and probably broken my neck.
I emerged with a chipped spine (lower lumbar) which delivered around 5 years of surprise spasms leaving me immobile in pain on the floor, but gradually the body has rounded that nicked edge and the issue has gone away.
A few months earlier though a close friend, was killed about half a mile to the North, only minutes after we'd spoken, as he passed me taking photos from the new Kessock Bridge. He was I understand decapitated by the roof edge of the car driven by a drunken man, undertaking the driver who had changed lane to overtake the cyclist.
Those with Heriot-Watt (or STV) connections can help here, as back in 1999 there was a programme called Action2000, and one edition featured a demonstration by the Heriot-Watt engineering team of a rear end impact using a remote control Mini Metro and a crash test dummy on a lightly supported bike. I had this on a VCR cassette, but it has disappeared from the chaos of a fire some years ago, when the tapes were stored on the other side of the wall from the burning pan on the cooker!
The video is very informative. It shows the massive damage to the car roof when a flailing body hits it, and how, when no intervention takes place the cyclist's body will be flying up and rotated to get sliced by the roof edge, with the exact point determined by the impact speed and linked parameters.
It follows from this that there may be some moves a cyclist can make, even when a rear-end impact is inescapable. One may be to go foetal, with, if possible feet and legs towards the initial impact as a shock absorbing element, and head with neck bundled in against the chest and arms over the back of the head. Ideally you then slide over the bodywork of the car and the windscreen/airbags are a softer place to hit. A second strategy may be to use the forces sent through the bike to launch you upwards in a more controlled way to clear the roof of the car that hits you. This still leaves the problem of where you then land.
I've also walked away from a couple of T-bonings, both at relatively low speed. In one case, tucked in and rolling over the front of the car, I managed to inflict over £600 worth of damage to the car, & bent the front forks in the bike, (I had a spare set).
I note that the LAS paramedics include training for at least 8 ways to dismount/depart from a moving bike without harmful end results. Is there a case for advanced cycle training to include falling off, against the similar requirement for a parachute jump, learning how to hit the ground roll and walk away, as this is almost a guaranteed event for all but the most perfect landings