A distinction between "optical activity" and "circular polarization" is
important to emphasize. In the first case the incoming light, linearly
polarized in one plane, comes out as linearly polarized in another plane.
This is also called "rotation of the plane of polarization". It is not a
case in which the next particle of the escaping light (or the next rope
segment in empty space, if you wish) is rotated more than the one before
it. Each layer of transparent material turns the plane of polarization a
little more.
Circular polarization, as stated by many, calls for two linearly polarized
waves (mutually perpendicular and of equal amplitude) which are out of
phase by 90 degrees. Linearly polarized light can be mathematically
described as two circularly polarized waves of equal amplitude (left and
right being in phase with each other) but that does have any physical
significance. There is only one basic kind of polarization, linear.
Neither circular nor elliptical polarization is possible when the linear
polarization is not allowed.
A mathematical decomposition of a plane wave into two components is
artificial in an isotropic material, such as glass. A simple analyzer
can be used to extinguish the outgoing light. But in some materials the
mathematical decomposition becomes much more than the artifact. This
happens when speed of light of one component is different from that of
the other. In that case the escaping components are phase-shifted, in
proportion t the traversed thickness, and the total extinction of light
by a simple analizer is no longer possible.
The real conceptual question is why is the speed of ligh in one plane
different from the speed of light in another? The old idea of Huygens,
according to which waves are absorbed and reemitted (from layer to
layer), has something to do with this. But how? Speed of light in glass
is about 2/3 of c (a well known factoid). And it can be expressed in
terms of the epsilon, or terms of the index of refraction, which are
also experimentally measurable material constants. I do not remember
seeing calculations of the speed of light in terms of specific
parameters of the medium through which it travels, for example H2O
versus SiO2. How do molecular potentials, or electrons densities, etc.,
contribute to a difference between c and v? Can n=1.3 be calculated,
for water, in terms of its atomic structure? That would be a big jump
above the level of factoids.