I found a very interesting article about Subarus in the US having a product fault on cars made between 1999-2002 that resulted in "external coolant leaks". See:
http://66.249.93.104/search?q=cache:p4TxfZ...uk&ct=clnk&cd=7
They brought out a "Cooling System Conditioner" which seemingly looked suspiciously like Holts Radweld...
So thoughts on my head gasket problem...
The Subaru problem was external, which I guess means it is from the cooling system into the outside world - into the engine compartment. This is the same pressure as the radiator so you can see how it is a cinch of a solution.. Also, the pressure in the cooling system would force water through the leak enabling a deposit of Radweld material (call it gunk!) to build up and seal the leak, which is how I assume Radweld works.
My problem is gases getting through from the high pressure of the cylinders into the cooling system - an "internal coolant leak". So, at least at present, there is no leakage from the coolant cylinder into the cylinders (although my garage tells me this is eventually what happens). So using Radweld would maybe not work because gases blowing through the leak would blow any coolant with Radweld out of the leak channel. So a thought..
I could use Radweld, run the engine for a while to stir it around into the cooling system and then just leave the car without running for a few weeks to allow capiliary action to carry the Radweld into the internal leak channel and solidify. OK, the Radweld is designed for sealing low pressure systems, but my leak is very small - I put about 1CM of water into the tank this morning after a journey of 140 miles yesterday. Also, the ability of the Radweld to seal will be dependent upon the size of the cross-section of the leak channel (measuring the pressure) and the length of the plug of Radweld filling the channel (measuring its adhesion to the inside of the channel). And if I leave it a few weeks, capiliary action may take the Radweld quite a long way down the leak channel. It may even be worth using a more diluted solution of Radweld so capiliary action can carry a small amount of Radweld gunk further before it sets (it does set does it?). With the replacement XM I could pretty much leave it as long as I like.
Any comments would be much appreciated, I'd be particularly interested in anyone who has used Radweld and knows what it does!