Judith,
About ice cores, there is a good overview of the Law Dome record and its gas age distribution at:
http://courses.washington.edu/proxies/GHG.pdf
Ice core CO2 is accurate to +/- 1.2 ppmv (1 sigma) for repeated samples of the same part of an ice core up to +/- 5 ppmv for different ice cores with extreme differences in snow accumulation and temperature.
The resolution of ice cores is less than a decade over the past 150 years (2 drillings at Law Dome), via 20 year over the past 1,000 years (Law Dome, downslope) to 560 years for the past 800,000 years (Dome C).
There is no measurable migration in the extreme cold inland ice cores, but there may be some theoretical migration in relative “warm” coastal cores. That gives a broadening of the resolution e.g. in the Siple Dome ice core from 20 to 22 years at medium depth and from 20 to 40 years at full depth (~70,000 years back in time).
Ice core CO2 are direct, be it smoothed measurements of ancient CO2 levels. Their main drawback is that the resolution worsens with the lower snow accumulation rate, but that allows to go further back in time. Another problem is to find out the exact timing between the age of the ice and the average age of the enclosed air.
Stomata data are proxies which have all the problems inherent on proxies: they grow locally on land, where there is a variable bias against “background” CO2 levels. That can be compensated by calibrating the stomata (index) data against direct measurements and ice cores over the past century. The main problem is that there is no guarantee that the local bias didn’t change over the centuries due to huge changes in land use, locally and in the main wind direction, over the centuries…
Thus if the stomata data give a different average CO2 level over the period of resolution of any ice core, the stomata data are certainly wrong…
BTW, your first reference mentioned Jaworowski, not the most reliable person about CO2 in ice cores:
http://www.ferdinand-engelbeen.be/klimaat/jaworowski.html