Well, between you all, you guys have just about cracked it, so at long last here's the conclusion to tonight's story....
I took the cover off the main panel and shut off the main GFI. Like all our main switches, this is double-pole, so it isolates the neutral from the incoming line. With my meter on the neutral and ground busbars (separate here remember!) I read almost a dead short. Pulling neutrals off the busbar one at time narrowed the short down to the ring circuit feeding the kitchen sockets.
My first thought then was a nail through a cable in one of the kitchen walls, espevcially as this hous belongs to a builder! I left an audible continuity tester connected at the panel and started pulling recepts off the wall, intending to disconnect the neutrals at strategic points to narrow down the location of the short.
At about the 2nd or 3rd outlet I unscrewed, however, the short disappeared as soon as I pulled it forward from the box (these are one piece outlet and faceplate combined - not separate).
Further examination revealed a cut in the insulation of the neutral where the wire was trapped between the side of the metal back box, the screw lug and one of the recept. mounting screws.
It turned out the sockets had been unscrewed and pulled forward to allow for decorating the walls, so it must have happened when someone was rather ham-fisted in screwing them back to their boxes.
Now if this were a PME system with the ground busbar bonded to the neutral, that parallel path would've tripped the GFI as soon as just a 60W light was turned on.
BUT.... The only ground return path here was via the local earth rod, which obviously had a much higher impedance than the solid neutral return.
So with a 3kW load, the parallel ground path was pulling less than 100mA. Any combination of appliances of around 5kW or so total resulted in more than 100mA flowing to the ground rod.
