The Yang version of Guliya goes back to “AD200” and is dated younger than Thompson et al 2004 (although a younger dating may also be in Thompson et al PNAS 2006). I did a quick comparison of the unarchived visual plot in Thompson et al (Science 1997) and compared it to the Yang version, yielding still more problems.
First, here’s the dO18 plot from Thompson et al 1997.
The original caption stated: “the Guliya record is also compared to the Vostok dD (blue), CH4 (red) and CO2 (black) [(13); circles, 1987 data; triangles, 1991 data] records (C), which display temporal continuity below isotope Stage 5d.” Thompson stated that dating was based on: “We used the apparent correlation between atmospheric CH4 levels and stadial and interstadial events inferred from d18O values”.
Next, I’m going to show a blow-up of the most recent portion of the above graphic, showing much higher dO18 values in an earlier portion of the Holocene, with the change in dO18 values from around -12 to -17 occurring about 6-8000 years ago.
Figure 2: Blow-up of Figure 3 from Thompson et al 1997.
Finally here is the data from Yang et al 2002 (kindly sent to me by email by Dr Yang some time ago) plotted with older to the right as in the Thompson graphic shown above. Observe the change between -17 values and -12 values, only this time the change takes place in the 3rd century AD. I’ll bet dollars-to-doughnuts that, if and when the samples are reconciled, we’re looking at the same transition in the two plots, merely dated differently. The reason for the enormous size of the difference is simply the exponential age models – so systemic age errors in the older parts of the core can be millenia, not simply 1-2%.