Polarisation Experimentation

In desperate search of an emergency blip while it's pouring down outside I came across this strange phenomenon. When you shine a LED light at a mirrored surface the spectrum of polarised  light you obtain off the back of a CD is very different from the normal ones you get when the light is pointed straight at  the reverse of a CD. I am not sure why this happens. The spectrum of light changes normally when you alter the angle of incidence but it never seems to create a similar pattern to the one I have obtained here. I wonder if anyone can throw 'light' on this mystery.