Remember too that most of the earlier samples are from high latitudes, usually Antarctica, where the low temperature and large cold Ocean surface area act as a CO2 sink (Henry's Law). Whereas the more recent measurements are from Mauna Loa in Hawaii, which is much warmer and where warmer Ocean surface temperatures act as a CO2 source (Henry's Law again). Pre-direct measurement data is generally from ice core data, which has a host of issues on top of the cold temperatures under which that atmosphere was "trapped" (hint... it's not very trapped). For a much more reliable CO2 proxy reconstruction, you could use plant stomata records from sediment samples of lakes. Though these records are patchy and difficult to find. When they are analyzed, it shows that atmospheric CO2 levels fluctuate significantly over time and are not correlated in a causal way to temperature (though the reverse is correlated, i.e. temperature effects CO2 levels).