Question about temps and C/W

I was just wondering how people seem to think that they are getting all these load temps lower than about 40-45 degrees.

Looking at the water block ratings on the main site, the white water, the best block only has a rating of .19 c/w. So a CPU putting out about 100watts ( a average overclock) would be 19 degrees hotter than the water going into it. Say the water is 25 degrees, thats still a temp of 44 degrees. With the atlantis( I have this block) it would be 51 degrees.

My question is how are people with just rad cooling, getting load temps below 40 degrees????

Inaccurate thermometers.

Edit: Also, CPU power dissipation is rarely known with any accuracy to speak of.

Second edit: And, the thermal conduction path down the CPU pins and through the socket, may have a C/W that is only five times higher than that of the watercooling system.

Taking just the water block into consideration is only looking at part of the system.

Radiator efficiency and airflow are also critically important.

My system:
XP1600@1915mhz
Cascade water block (slightly better than the White Water)
Eheim 1250
Big (6"x12") heater core
Ambient temp (measured in front of rad.) 30.6c
cpu temp (on die) 41c
cpu watts (by New Watt 3- not exact, but close) 89.06

C/W=0.117

The ratings also state that a flow rate of 1 gpm is used for all tested blocks- while this is good for comparisons, it does not reflect what is running through someone's system, and has a MAJOR impact on performance! It is probably very close to real-world flow rate, but low.

There are many things to consider in a water cooled system, and a good water block is important, but far from the only factor.

Edit- A note on temperature readings:
It is very true that readings are NOT consistent between bios versions, motherboards and systems so comparison is not an exact thing. However, we need something to go by and few of us can afford the test equipment that would be need to verify all system statistics. My system also has a DigiDoc5 thermal probe mounted against the core: reading is 35c with an on-die temp of 41c; I'd call it close enough.