The relevant technical facts are as follows.
The date was 30 June 1994, the airfield Toulouse.
The test flight was about a fairly minor upgrade to the flight controls software. All the manoeuvres involved had been flown the previous day perfectly satisfactorily with the CG at the forward limit and crewed by the same captain and flight test engineer.
For the accident flight the loading was adjusted to give the aft CG. Limit
The flight was to include several takeoffs with the autopilot engaged at unstuck to check its performance with and without engine failures. The captain let the co-pilot (not a test pilot) carry out the takeoff for the accident takeoff.
The trim was set to 2 nose up when it should have been set to zero.
There was a small amount of over rotation beyond the target attitude.
The Captain got on with engaging the autopilot, bringing one engine to idle and turning of the hydraulics from the idle engine to simulate the system circumstances of an engine shut down. This was all completed by about 135 kts.
At 129 kts the captain realised something was wrong, as the speed should have been increasing not decreasing very quickly.
By 100 kts (18 below VMCA) he had got the autopilot off and applied full forward stick
Minimum speed reached was around 80 kts, max bank around 110 deg. With both engines at idle the aircraft was recovered back under control but needed 400 feet more to recover and was still 15 deg nose down at impact.
An autopilot mode asking for 2000 feet was left selected from the previous circuit
End of facts
My understanding (which may be oversimplified or even incorrect as I have never flown an Airbus) is that the mode of the autopilot that was trying to get to 2000 ft (a mode that would not normally be selected at take off) was responsible for the speed reduction as it was asking for a climb and the mode concerned did not care about speed.
Nick Warner, the captain and CTP, was a good friend of mine. The flight was at the end of a very long and complicated day where he had been required to do a succession of important non flying duties as well as dealing with customers and flying other test sorties before the accident one. The enquiry concentrated on his excessive workload that day and the associated human factors issues.
Having been a CTP for a manufacturer myself, please believe me when I say anybody who looks at the above and just sees simple pilot errors does not understand the business.