I have a firm belief in experimentation.
Air, water or fancier stuff- it doesn't matter: what works best for some might not work best for you.
It is very hard to quantify ALL the variables involved in cooling a computer, period. That said, many people still try
Bill A. is doing some fantastic stuff with water cooling testing and benchmarking: it has to be the best work I have seen done on the subject, period.
And yet even that doesn't answer EVERY question
YET!
Most commonly the best results are found with the radiator (whatever kind you use) immediately before the waterblock; order of other components like reservoir/pump seems less important.
It seems that having the coolant as cool as possible when it hits the cpu is best
and rad-->waterblock may be the best way to do that....usually.
But the difference I've seen reported is usually minimal, or nonexistent. In my own case I get 1 or 2c difference, MAX.
So for me the experiment of switching around was worth it, others get different results.
As long as water cooling remains as much Art as Science you will find few truly definitive answers