It's pretty clear that this is the Sigma: http://www.canonrumors.com/forum/index.php?action=dlattach;topic=1280.0;attach=166893;image
There is a noticeable drop in sharpness and contrast.
Alan, your tests really are great, and it's incredibly interesting to see.
It's funny how the line pair tests show such a drastic increase in resolution on the Sigma 150-600C over the 400DO+TC, when these images show the Sigma at a disadvantage.
(Unless I'm totally wrong in which case these results would be quite shocking.)
It reminds me of my own comparison between the Samyang 800mm Mirror lens on Full Frame and the Canon 400f5.6 on a crop body.
Detail between the two lenses is almost identical, but the Canon lens has much better colour transmission (and obiously the mirror lens is manual focus only).
Alan, if you don't mind humoring a skeptic next time you're in the mood for comparing lenses, what I'd really like to see is a full frame image from the Sigma 150-600C downsampled to the same size as a 1.6x center crop from the bare 400DO without any TC (which should effectively emulate the 7D2).
In the TDP tests the 7D2 on a 400mm lens gives longer effective reach while only giving up a slight edge in terms of center sharpness when compared to the 1DS on the Sigma 150-600C.
The 5DS has the same pixel density as the 7D2 so cropping a 5DS image to the same resolution as a 7D2 image should be a fairly precise simulation of the crop body (if only we could get a 7D with no Aliasing filter).
Given the good results shown this far I don't expect the center crop of the 400DO to match the Full Frame results of the Sigma, but if it's close enough then once you factor in the increased AF speed and smaller overall package, cropping still seems like a better option than buying a longer third party lens.
On top of that, if you're on Full Frame then tracking BIF at 400mm will also be much easier than using 600mm glass.