Interesting, I haven't noticed any color shift when using my Schneider 58 mm XL on 4*5. But I do indeed typically use a dedicated center filter with that lens.
Would mind sharing with us the physics underlying the appearance of a color shift?
I use a 55m Grandagon on a 6x12 camera and when I don't use a center filter there is a color shift - usually to blue. I am not sure there is a whole lot of research on this, but my educated guess says it seems to be because of underexposure - variations in exposure cause color shifts. There is most likey a transmission problem with the film because the incoming light is at such an oblique angle and the blue layer for color film is at the top. Film is also most sensitive to blue. The underexposure emphasizes this problem. (I use a 1.5 stop neutral density center filter and it almost completely eliminates the problem.)
In mark's case, he seems to be getting a green/magenta shift. The bayer pattern is mostly green. If the pixels are not physically flat, this may be giving the green shift toward the edge. The white balance may be trying to compensate so there is a magenta shift toward the center. The reason his dedicated center filter may not be working is the surface of the CCD causes a greater fall-off effect (this is why I asked him the difference with the filter off). Phase One in their instruction recognize the problems of the cosine 4th law (that is the reason for natural vignetting), but have decided to invent a new term called "lens cast."
But why not the Canon DSLR? Short focal length lenses are a reverse telephoto design so they can be far enough away from the CCD to allow the mirror to rise. So at a comparable angle of view, you are not actually getting light hitting the CCD at a similar angle of incidence as you would from a veiw camera. Also since the effect is predictable, DSLR manufacturers can compensate electronically to a certain degree - most DSLRs know what lens is attached. In fact lens data is stored in the exif file. Not something that can be done with the view cameras or Hasselblad Arc Bodies.
So it appears the cosine 4th law works. When using wide angle lenses, the term "vignetting" is applied. So far in this thread I have not seen anything new that would suggest another cause, but I am game if someone can come up with another explination.