Your pictures are highly deceptive. They imply ice core resolution that is simply impossible in any sample other than a modern (uncompacted) one. The first one overlays measurements so you can't even see if there any ice measurements underneath (not that they matter). The second is pretty lacking, I'm sure you can find better. In the meantime, here's a better source, showing error bars on most of the samples. Note the absence of a hockey stick.
It's not a surprise that ice core CO2 data from older periods has decreased temporal resolution. But the concentrations of CO2 measured in the bubbles are accurate.
Where's your figure? Here's another one (with error bars; this might be the one you were planning to post):