I had a piece in Scientific American over the weekend, a short news article on a cool theoretical physics paper that argues that no material that bends light can be perfectly transparent.

Normally I don’t have much “bonus info” for the shorter news pieces…but there’s a fun subtlety here.

I figured I’d start the piece by talking about some of the most transparent materials out there, like the Guinness world record, an aerogel designed to insulate windows, then compare them. I’d list how transparent each substance is, ending up with the new theoretical limit, far above any real material.

There was just one problem: there wasn’t a number to list.

You might think of transparency as a percentage, how much light gets through, the kind of thing you tweak on a slider in a program like Photoshop. Indeed, there’s a nice number like that in the Guinness world record, a range of 97% to 99%, with a single pane of glass, in comparison, listed at 92%.

But when I wanted to compare those numbers to water or air, I realized I was missing something: how thick is the glass or the gel? How thick is the air, or the water?

Really, the numbers I should be comparing are distances.

The attenuation length of a material is how long light travels before it runs out. It determines the depth of the aphotic zone, the part of the ocean where the water turns pitch black, with no light visible from the surface.

But you can’t see down to the aphotic zone, even in the clearest water. That’s because long before you reach the attenuation length, you reach the opalescence length. There, light still gets through, but it gets scattered in all directions. You get a diffuse glow, not a clear view of what lies beneath.

So I was hoping I could look up a table of different opalescence lengths, like you might see a conductivity table for different metals. But another subtlety reared its head: light has a wavelength.

Water is impressively transparent…to visible light. Arguably, it’s why we evolved to see that kind of light in the first place. It’s much less transparent to infrared light, or to X-rays, or any of the other lovely parts of the electromagnetic spectrum. If you want to see how transparent a material is, you have to look at a whole graph, attenuation lengths versus wavelengths. It’s a spiky mess, full of lines where the energy of the light is exactly right to wiggle an electron’s quantum state, absorbing more than it would in the rest of the spectrum.

So if you want to really compare how transparent things are, you have to break out those graphs and line them up. What the physicists I wrote about found is something they expect to be way down at the bottom of the plot, a smooth curve that sits below everything else, a theoretical limit more transparent at every point on the spectrum, but still not perfectly transparent.

But you still can’t compare that to the Guinness world record, because of surfaces! Attenuation lengths, opalescence lengths, these things measure how far light travels. But a surface between two materials, like the surface of a pane of glass, doesn’t just let light through. It also reflects it. So the Guinness world record is an achievement not primarily in attenuation length or opalescence length, but in reflection, carefully matching the properties of the material with the air around it so the light doesn’t bounce back. It’s playing a totally different game.

Amusingly, while working on this piece, I was also reading Andy Weir’s lesser-known book, Artemis. The book features a conspiracy centered around a material for fiber optic cables that is perfectly transparent, a Zero Attenuation Fiber Optic, or ZAFO. Weir gestures at the idea that the material might work analogously to superconductors, which genuinely have no resistance. If there’s one takeaway from the theory paper I covered, it’s that this analogy doesn’t work: there can be no such material.