We have noted interesting behaviour of our data at the lunar limb, link here. It seems that our data show a convergence of intensity ratios, between certain symmetric points on and near the limb, towards unity. We saw the same in model data.
The model we use is based on the Hapke 1963 BRDF (i.e. reflectance) model and looks like this:
BRDF ~B(phase)*S(phase)* 1/(1+cos(e)/cos(i))
where e and i are the angles of emission and incidence. For a given instant the phase is the same everywhere (almost) so the only angular dependency that remains is the last term above. On the limb we can have various values of i but only one value of e - namely pi/2 - recall that i is the angle between local normal vector and the Sun, while e is the angle between local normal vector and the direction towards the observer.
cos(pi/2) is 0, of course, so the last term above is unity everywhere on the lunar limb. The ratio of the reflectance in two points on the limb therefore reduces to unity. The ratio of two limb intensities should therefore be equal to the ratio of the albedos at those two points.
So, there is no clue here as to why we observe, and the model gives, intensity ratios near unity along the limb. Mystery remains!
Our observation of unit intensity ratio certainly is consistent with Minnaerts reference to Schoenberg's statement - namely that the 'intensity along much of the bright limb is a constant value'. Intensity ratios between limb points would give unity.