Thanks to spider95 for running the thread and doing the stats.
I agree that this is a pretty worrying potential fault.
Just to add my thoughts FWIW, the water pump is usually reported to have failed due to its bearings. This seems to me to be a "how many times has it gone round" sort of failure rather than being time-related (or how about knackered-as-fitted
). I can understand a cam belt deteriorating with age, but I don't see that a set of bearings would necessarily be the same, except if they were made of an unstable plastic or something.
Assuming the bearings are metal, perhaps contamination is entering from somewhere or is there from new, or else they are not manufactured by drill and turning but are instead pressings (not having seen one I can't comment further). I would still hazard a guess that failures are almost exclusively number-of-rotations related. The differences in mileage of failure are probably due to the degree of built-in knackeredness of the pump.
My own Alfa (Oct2004 JTD GT) has done 43000 odd miles and is due for cam-belt change around the end of August, so I could be accused of "talking my book" here! However, in view of what I have just said, I am prepared to let it go to the next service and have the pump changed at the same time. My driving style is such that I usually keep the revs down to about 2000 in as high a gear as possible (diesels work better that way) so I should have some leeway.
An alternative scenario would be to have the pump changed (with the belt and all the gubbins!) at the lowest recorded mileage of failure which from spider95's excellent summary is 42000 miles. Practically this would mean the 36000 service. I don't see why AR shouldn't agree, if they could be persuaded that the water pumps are likely to be defective. Maybe it would make them fit better ones. It still shouldn't be necessary though.