Hmmm, I suppose I should've explained it better.

Air has a finite thermal capacity per volume. It happens to be around 1250J/m³.K (round figure).

i.e. it takes around 1250 Joules to raise 1m³ of air by 1°C.

So if the fan is moving 1.5m³/min of air through the radiator, then the air-flow has an inherent thermal capacity of around 31.25W/C, or a C/W of 0.032. i.e. the radiator CANNOT perform any better than the inherent thermal capacity of the air. Even if the radiator was 100% efficient, the C/W will never be lower than 0.032C/W.

Now if the radiator is 90% efficient at this air-flow speed, this means that the radiator will have a C/W of 0.032/0.9 =~ 0.036. i.e. the equilibrium point will be such that the water warm up by 0.036°C per Watt of heat input (from CPU + GPU + pump + whatever else is being water-cooled).

If we now add on a second radiator, and this increases the air-flow resistance such that the air-flow through the two radiators is now 1.1m³/min, then the inherent thermal capacity of the air is now 22.9W/C, or a C/W of around 0.044.

i.e. even if we achieved a 100% thermal transfer efficiency of heat from the fluid into the air using the two radiators, at best we will achieve a C/W of 0.044, or about 20% worse than the 0.036 of the single radiator.

Now as air-flow goes up, radiator's thermal transfer efficiency goes down. There is eventually a cross-over point where adding a second radiator in air-flow series to the first radiator does help, but for PC water-cooling radiators, that doesn't occur until you start stacking extremely noisy fans, or using obscenely noisy single fans (Delta FFB/TFB, etc).