There are two kinds of light-the glow that illuminates, and the glare that obscures. - James Thurber
Some of you are probably old enough (like me) to remember when PolaroidR sunglasses first became available. They were a sensation. Glare light is reduced by more than the background light with the result that you actually have better vision-not just less light in your eyes. They work best in situations where the sun is more or less in front of you, and the combination of reflected light and near-specular scatter causes a bright glare that dominates your field of view. Because reflectance is a function of polarization, the glare light (which has been reflected once) often has a horizontal polarization component that is much stronger than the vertical component. Normal background light (which has been reflected many times, at many angles) is more evenly divided between the two polarizations. Polaroid sunglasses simply discriminate against the horizontal component to reduce the fraction of glare light. If you look at the blue sky - a source of scattered light - through your sunglasses, you will find that rotating the glasses 90 deg causes the sky to look dimmer (i.e., the sky has a strong vertical component). This effect is most strongly pronounced if you look in a direction perpendicular to the sun's rays. These two examples clearly demonstrate that scattering is polarization sensitive.
When light is scattered, its polarization - along with its amplitude and direction - is changed. The changes depend on the sample shape and material, as well as the polarization, amplitude, and direction of the incident beam. All three quantities must be considered when examining the effect of a sample on the reflected and transmitted light. A complete description of the polarization characteristics of an EM wave, before or after sample interaction, can be accomplished by straightforward measurements. Comparison of polarization amplitude and direction before and after interaction with the sample allows information about the sample material to be obtained. The trick is to relate the before-and-after changes to useful, or needed, sample characteristics.