Polarisation: Light With an Orientation

Glare is not extra light

Start where anyone can check it. Glare off wet asphalt or a lake is not simply more light; it is light that has been sorted. Sunlight arrives with its electric field pointing every which way, changing orientation faster than any instrument follows. Reflection from a flat dielectric surface does not treat those orientations equally: the component lying parallel to the surface reflects more strongly than the component in the plane of incidence. At one particular angle — Brewster's angle, where the reflected and refracted rays are perpendicular — the second component vanishes from the reflection entirely, and the glare is fully oriented.

So the surface has performed a measurement of sorts. It did not add energy and it did not choose randomly. Its geometry — a plane, with a normal direction — decided which orientations left and which stayed, and the light that reaches the eye now carries that decision as a property. A polarising lens is a second, deliberately aligned surface in the same business: rotate it and the glare disappears while the rest of the scene stays. It does not dim. It combs.

The bookkeeping, named out loud

This volume's rule is to name the equations being borrowed. For fully polarised light the object is a two-component complex vector — the Jones vector — holding the amplitude and relative phase of the two transverse components. A polariser, a wave plate, or a rotation is then a 2×2 matrix acting on it, and a sequence of elements is a product of matrices. That is the whole formalism, and it is linear.

Real light is usually not fully polarised, so the working description is the Stokes vector: four real measurable intensities, with a degree of polarisation that can sit anywhere between zero and one. Its geometry is the Poincaré sphere — every pure polarisation state is a point on a sphere, linear states around the equator, left and right circular at the poles. Optical elements move points around on that sphere.

The sphere is worth pausing on, because it is where the chapter's real distinction lives. Any two points on a sphere are joined by a continuous path. There is no obstruction, no integer, nothing that cannot be undone by turning a wave plate. A polarisation state is a direction, and directions deform into one another freely. That is a statement about topology, and it is the reason a single beam's orientation, however precisely measured, is not yet a braid.

From an orientation to a twist

Give the two transverse components equal amplitude and a quarter-cycle phase offset and the field stops oscillating in a plane. It rotates as the wave advances, so the tip of the field vector traces a helix around the direction of travel, one full turn per wavelength. This is circular polarisation, and it is the second step: not a fixed orientation but a winding one.

Two registers to keep apart, and this is where careless writing about braided light usually goes wrong. The spiral is the field, not the path: nothing travels along the helix, and the energy moves straight down the axis while the orientation winds around it. And handedness — left or right circular — is a discrete label, but it is a label on a single strand. A twist is a local property of one thread. Chapter 7 said a braid is a property of nothing smaller than the group.

From a twist to a braid

The third step is to add strands. Three circularly polarised beams sharing one axis, each offset by a third of a cycle, project onto the page as a standard three-strand braid. Nothing is drawn by hand: the transverse position of strand n is a sine with phase 2πn/3, and the sign of its radial component decides which strand passes in front at each meeting. The crossings are the phase offsets.

What the second picture makes available is a different kind of quantity. The helix gave a continuous phase and a handedness. The braid gives a word: a finite sequence of crossings, each one over or under, read along the axis. Stretch the strands, bend the axis, change the amplitudes — the word does not change. Only cutting a strand or reversing a crossing changes it. That is the sense in which a braid carries information the geometry cannot smooth away, and it is the first time in this part that the sentence has been earned rather than asserted.

Chapter 9 takes the last step honestly. A single vortex beam carries an azimuthal phase winding whose topological charge is an integer, measurable and unchangeable by any continuous adjustment — a winding that lives in one beam and still cannot be deformed away. That is where the braid stops being a figure the author likes and becomes a number an instrument reports.

What the filter costs

One correction to the comb, because the chapter should not flatter its own metaphor. Filtering is not free. An ideal polariser placed in unpolarised light passes half the intensity, and a second polariser at angle θ to the first passes cos²θ of what reaches it. Energy is genuinely removed — absorbed or reflected away — not merely relabelled. The reason the lens still feels like a comb rather than a dimmer is that the light it removes is concentrated in the part of the scene you did not want, so contrast rises while total brightness falls.

That accounting matters beyond sunglasses. It is the same structure as every selective boundary condition in this volume: a geometric criterion applied to incoming energy, with a real cost, producing information that was present all along but unreadable in the mixture. Naming the cost is what keeps the phrase energy as incoming information a description rather than a slogan.