Metro III ZK-POA was scheduled to depart from Auckland (AKL) at 21:00 on an NZ Post return flight to Blenheim (BHE). The payload comprised 1790 kilograms (kg) of courier packs and parcel mail. Loading of the freight was delayed, being completed at about 21:15. The crew ordered 570 litres (about 1000 pounds or 450 kg) of additional fuel and instructed the refueller to put it all into the left wing tank, rather than put half of the ordered amount into each tank, as was company practice. Refuelling was completed at 21:30. Two minutes later the flight taxied out to runway 23R. Takeoff was performed at 21:36. The aircraft climbed to an altitude of FL220 with the autopilot engaged for the climb and cruise. At cruise altitude the climb power remained set for about 15 minutes in order to make up some of the delay caused by the late departure. At about 22:12:28, after power was reduced to a cruise setting and the cruise checks had been completed, the captain noted some fuel imbalance and decided to carry out further fuel balancing. He said, “We’ll just open the crossflow again ..sit on left ball and trim it accordingly”. The captain repeated the instruction 5 times in a period of 19 seconds, by telling the first officer to, “Step on the left pedal, and just trim it to take the pressure off” and “Get the ball out to the right as far as you can ..and just trim it”. The first officer sought confirmation of the procedure and said, “I was being a bit cautious” to which the captain replied, “Don’t be cautious mate, it’ll do it good”. These instructions resulted in the aircraft being flown at a large sideslip angle by the use of the rudder trim control while the autopilot was engaged. Forty-seven seconds after opening the crossflow, the captain said, “Doesn’t like that one mate.. you’d better grab it.” Within one second there was the aural alert “Bank angle”, followed by a chime tone, probably the selected altitude deviation warning. Both pilots exclaimed surprise. The autopilot had probably disengaged automatically because a servo reached its torque limit, allowing the aircraft to roll and dive abruptly as a result of the applied rudder trim. The crew was unable to recover control from the ensuing spiral dive before airspeed and g limits were grossly exceeded, resulting in the structural failure and in-flight break-up of the aircraft. The flying conditions of a dark night with cloud cover below probably made it more difficult to notice the developing upset at an early stage. The break-up probably started when the aircraft was quite high, perhaps FL199. The wreckage was scattered over a large area of at least two to three square km of rural farmland. PROBABLE CAUSE: The investigation report did not contain a probable cause paragraph as recommended in ICAO Annex 13.