I spoke to my friend Iliah Borg about this and this is what he said...
The issue should be the curvature of the rear element.
http://forums.dpreview.com/forums/read.asp?forum=1026&message=6422541
New IR filter: In something of a departure, the SLR/n uses a dye-based IR
filter, rather than the dichroic type used by the 14n and most other digital cameras. The dye-based approach sacrifices about a half-stop in light sensitivity, but reduces optical artifacts caused by internal reflection with the dichroic filter. (The problem being most apparent with lenses with elements that projected back into the camera body. The 60mm Micro-Nikkor was apparently particularly prone to producing a bright glint in the center of the frame when stopped down, the result of internal reflections in the near-IR range of wavelengths.) The long pass-band (e.g., visible light) "tail" of the dye-based filter's spectral response also apparently helps offset the higher red-sensitivity of the sensor itself, producing more balanced output levels between the red, green, and blue sensors. In the raw silicon, there's about a two-stop difference in sensitivity between the red and blue channels. When you combine this with the heavy color cast of incandescent light sources, the total variation in response can reach four or even five f-stops. You'd thus expect the dye-based IR filter to be particularly helpful when shooting under incandescent light sources, and the SLR/n does indeed appear to do a better than average job with such lighting.
As far as I see it, the closer is the rear element the less light fall-of
between reflections, and the more pronounced is the effect. the more curved is the rear element the more light goes off-sensor and the less pronounced effect follows.
--
