A clarification on C/W numbers and their accuracy:
Those numbers are actually quite accurate- for the parts they test.
In a water cooling system you do add complexity with a larger number of parts, but you also add a number of different ways to improve efficiency:
With a well designed system using a high flow pump with plenty of head capacity, a large heater core and minimal tubing of 1/2" or larger size I would expect to get the WW's C/W down a good bit. Add LOUD fans and it will come down still more.
When I crank my fan voltages up to 12v my C/W is around .08.
(NOTE that MY numbers can not be accurately compared to those done with any other system or a die simulator- my temp readings may be off and my figures for overclocked watts are almost certainly off
)
But the point still remains: with a heatsink there are few ways to improve C/W- lapping, thermal compound and faster fans are it.
With water cooling you can improve: pump, tubing, radiator, fans, lapping and thermal compound.
All the ways to improve the system BESIDES THE CPU's HEAT REMOVAL DEVICE make it possible for the cooling system to have a considerably lower C/W than the water block does.
Hey, I even have a system with a Koolance block that gets a C/W around .18, lol. And that block is pretty....poor