OK, so we're back to the question of whether T leads CO2 in the ice cores, the skepics favourite talking point. The std.answer is "OK, so there is a lead (maybe) but..." (Stoat passim). The "but" is a good enough answer, and I suspect most people skip over the (maybe). But its important, because the lag/lead is not at all easy to establish, because the T comes from the ice (via d-o-18 proxy) and the CO2 comes from the bubbles. 800y+/-600 is the current "best guess", but not certain.
Thanks to GH for pointing out Loulergue et al. [Meh, dead link. Try http://www.clim-past.net/3/527/2007/cp-3-527-2007.pdf perhaps] who propose that the T lead is much smaller or even that CO2 leads: Our results reveal an overestimate of the Δage by the firn densification model during the last glacial period at EDC. Tests with different accumulation rates and temperature scenarios do not entirely resolve this discrepancy. Our finding suggests that the phase relationship between CO2 and EDC temperature inferred at the start of the last deglaciation (lag of CO2 by 800±600 yr) is overestimated and that the CO2 increase could well have been in phase or slightly leading the temperature increase at EDC.
Now... I wouldn't get carried away by this, since this isn't definitive either - just interesting at this stage. And I have done no more than skim it. Its worth a skim, just to get some idea of how hard this relatively simple thing is to establish. Note that the paper is under review in an open-access journal, so jump in if you think you're hard enough!
ISTM it would present a few new problems if CO2 did lead. What would have caused it to rise then? And why would that coincide with orbital cycles?
But while you are revisiting this general question, is there anything to be teased out of the reverse process? ie does temp start to fall and then ~800 years later CO2 falls in a mirror image of the rise?
how do we *know*, that after CO2 goes up, it acts in the positive feedback in the further temperture increase?
Is it not possible, that CO2 simply goes *always* up with the temperature increase without significant effect? Can we *rule out* this possibility?
[Only if you're able to repeal the laws of physics -W]
I think it's because the initial forcing - the orbital fluctuation - is not enough, on its own, to account for the temperature rise. Add on the CO2 rise and the two together are.
William: somebody has asked me this, & I don't know the answer: can you help?
"...The problem I have with this is that the forced increase in temperatures will lead to further increases in CO2 - a perfect, and direct, feedback mechanism - but shouldn't this mean that the graphs from the Vostok Ice Core should show exponential curves at each major upturn in temp and CO2? I'm sure that audio feedback follows this exponential curve, and I was under the impression that any direct feedback loop would do the same. So why do the Vostok graphs show linear increases in temperature and CO2?..."
BTW: On the Loulergue et. al. paper, they have done some interesting re-working of the data using spikes, which required changes in the firn densification formula (I think that's right): they may need to justify their changes better. At the moment, it looks too much like the method fits the required results too comfortably, but this is just the impression of a dliettante.
[I don't have a really good answer to the feedback stuff - the real answer is that its not fully understood. Another one is that its probably/possibly related to upwelling of ocean water releasing CO2, which might set a maximum rate of release of CO2.
L et al is in review; it will be interesting to see what comments come out. The main reason to link to it is not to say "this is correct" but to point out that "there is always exactly 800 year lag and everyone agrees" is wrong -W]