candyman and merc your ideas are close, but not quite home yet. i will try to explain more to make my point stronger and more clear, this is easy to get confused on.
IMOG, I understand your point, but if I can get the same number of molecule (of water) to be cooled per second, but I can also get each single molecule to be cooled twice as long as normal with my 2 parallel h-c, that way the delta between inlets and outlets would be increased, thus the delta between the water and the water-block would be increase also, thus better performances...
this is slightly off base, but we are talking about the difference between 2 rads in series or 2 in parallel right? So, no molecules will be getting cooled any longer in parallel than they would in series. the water doesnt slow down as it goes through the rads in parallel, the flow rate is lower through each rad because the volume of water moving through the rads is lower, but the speed with which the water moves is ~equal to that of the rads in series. So there will be no difference in water speed or the time in which molecules spend in the radiators. the rest you are right about on higher dT's being advantageous.
turbulence is rough flow. instead of the same molecules staying against the walls of the material exchanging heat, turbulence causes the molecules closer to the middle of the flow, which are cooler and have a higher dT to the material, to come into contact more with the walls of the material.
I agree on this one with Mercury Lenth, another point to note is that usually the higher the difference between temperatures of 2 items the more heat is transfered, so having the two radiators in parallel means both get the hottest water possible, yielding the biggest heat transfer delta. Lastly about turbulance, yes fast flow mean more turbulance but with smooth surfaces, such as the pipes the difference is minimal. For big turbulance differences you would need some kind of ridges of uneven surfaces on the pipes, then I would reconsider this but on this one I would cast my vote for parallel setup.
you are right again about higher dT's being good. (of course the difference between parallel and series may be minimal but it is a small point to consider still.) However, I am sorry if i implied that higher flow increased turbulence, although true, this isnt really what i meant. What i meant was that higher flow and turbulence are both factors which contribute to more frequent and varied contact between the molecules of a liquid and a solid, and to better heat transfer by twofold - through higher number of molecules touching the walls, higher frequency with which the molecules touch the walls. also, there is considerable turbulence in pipes that bend like that of a radiator, but thats something i will overlook.
BTW, if we are voting i think i would vote for series so you don't have the splitters and extra interfaces. either way though, the difference will be negligble if both would be done properly.
hope that helps.
as always, if im wrong somewhere, call me on it.
side-note: on skifletchs note on maximum temperature differences throughout a loop i think he mentioned he thought the differences in water temp was 2-5C. i have read that this is actually .2-.5C and i wonder if that isnt what he read also, it wouldnt surprise me. I know this is far, far closer to what could be observed... anyone have a linky to someone observing their inline water temperatures? the reason why this is FAR more reasonable could be easily explained rationally, but its always nice to see actual readings too.