I have a theory/hypothesis I would like to share for debate.
The theory applies to all Gyro's but is more prone to happen to shortly-coupled side-by-side machines.
We all know that a rotor must have a positive angle of attack into the apparent wind in order for the auto rotational forces to work. Without a positive angle of attack the rotor will lose RPM and will not fly positively. Once in the air, it is only possible to achieve this state by 'unloading' the rotor. (Negative G's - absolutely forbidden of course)
The moment-arm of these new-generation gyro's is very close to the Center of Gravity resulting in the nose popping up far quicker than in the tandems during take-off. Perhaps the wider, flattish-bottomed, fuselage's/cockpits generate their own lift and contribute to the cause by adding to total lift up to a point that the rotor takes over all the lift - thereafter it (the fuselage) is pure drag. In a typical take-off roll, the stick is full back and the rotor is loaded as the drag increases with forward speed. The rotor speeds up to the point that the nose lifts and we check forward on the stick to hold the nose down and to avoid getting airborne behind the power curve. As the nose drops again, the stick position may remain the same relative to the cockpit resulting in the angle of attack becoming too small. This is the same as taking off with the stick forward which we all know WILL result in blade-flap, regardless of rotor RPM. The disc happily skips along as the airspeed increases rapidly because there is no drag. The rotor is unloaded at this point. We then decide to take-off because we have adequate speed and come back on the stick. The rotor goes from an unloaded condition to fully loaded in an instant, the rotor speeds up rapidly. As a result of the high airspeed the advancing blade is exposed to a lot more airsflow than the retreating blade during this transition. The advancing blade 'flies' happily while the retreating blade 'stalls' and BAM, roll over to the left (in an anti-clockwise rotation) is the result.
Disclaimer: I am particularly poor at trying to draw things on a computer so please forgive the amateurish effort above. I also hasten to add that the gyro represented in the picture is a generic creation of my own and is NOT intended to represent any specific make or model of gyro. It is for illustration purposes only. (Before some agents/manufacturers throw a hissy-fit again)
I have often heard pilots say that the wind must have thrown them over on takeoff. The wing tip speed (lift generating portion) of a typical rotor is in the region of 650km/h or 400mph at typical flight speed. A cross-wind of at least this speed will be required to throw it over. Anything less will push the gyro off course downwind but will not flip it over - or am I wrong here?
If my hypothesis proves to be reasonable, then it follows that the safest method of takeoff is to ensure that the rotor is loaded throughout the takeoff roll until the gyro is airborne, whether it is behind the power-curve or not. In other words, you must feel rotor drag throughout the takeoff roll. Once the wheels are off the ground, blade-flap cannot happen in steady flight. The best method then would be to push the nose forward, gain adequate airspeed and only then begin the climb out.
Avoid climbing out steeply at the gyro's best angle of climb, stick to best rate of climb. The reason for this is, again, more pronounced in shortly-coupled side-by-side gyro's as opposed to tandems. If the engine fails while in a nose-high attitude, the airspeed will decay very rapidly. Pushing the nose forward to regain flying speed (45mph+) may take too long to avoid making contact with terra firma before the gyro is 'flying' again and it may not be possible to round-out, flare and land normally. In other words you may be forced to 'mush' onto the ground hard, resulting in all sorts of gyroscopic effects taking over and spoiling your day.
If you fly a side-by-side. Try this. Climb to a safe altitude (1500' agl or more). Do your HASSEL/HASELL checks of course. Establish a powered climb at 50mph and then chop the throttle on yourself. Delay pushing the nose hard forwad for a second or two - simulating a real life WTF? situation. Check airspeed and altitude. Now push the nose forward and look for 65mph airspeed. Once you get to 65 check your altitude again. The difference is how high above the ground you need to be to survive an engine out after takeoff (assuming you still have adequate runway ahead of you or a safe LZ in your path when it happens). You may be shocked at what you find. I was!

Fly safe and let me have your thoughts and arguments for and against my theory please.
Rgds
Len