In the half-century covered by the 1895–1946 graph CO2 rose by only 15 ppmv (5.3%). In the half-century covered by the 1957–2008 graph CO2 rose by 70 ppmv (22.5%), i.e., a 4.2× greater CO2 forcing.
I do not doubt that rising CO2 level contributed to warming, but the similarity of the two graphs, despite the huge dissimilarity of the two CO2-forcings, obviously doesn't support the case for CO2 being the “principal control knob” for climate.
The “control knob” wasn't methane, either. In the half-century covered by the 1895–1946 graph methane (CH4) rose by 0.25 ppmv (29%), and in the half-century covered by the 1957–2008 graph CH4 rose by 0.57 ppmv (47%).
Here's a follow-up article:
You can easily reproduce these two graphs (very closely) at any web site which can produce graphs of HADCRUT3 temperatures, e.g.,