Two Yellow Lines Three Metres Apart

You know how when you measure forge temperature you’re really just picking one wavelength and hoping the steel radiates predictably? That pyrometer uses a red filter—maybe 650 nanometres—to isolate a slice of the thermal spectrum. But the full radiation curve contains information I’m throwing away. What if I could see all of it at once?

There was a triangular prism sitting in a box of reclaimed optics on my bench. Estate sale find, mixed in with lenses and mirror blanks. Held it up to the overhead LED and a rainbow splashed across the wall. Just refraction through glass, basic physics, but I hadn’t actually looked at dispersed light in years. Started wondering: what does sodium vapour actually look like when you split it into components? Not “yellow-orange” as a vague colour impression, but as discrete wavelengths I could measure.

Found an old DSLR tripod, a single-edge razor blade, and some black cardboard. Built a slit by taping the razor to the cardboard with a 0.3 mm gap—narrow enough for sharp lines, wide enough to let photons through. Aimed it at the sodium street lamp two blocks over. Put the prism behind the slit. No eyepiece, no camera, just stood there in the dark garage squinting at the dispersed light on the far wall.

Two yellow lines. Not one continuous yellow blob—two distinct emission lines, maybe half a millimetre apart at three metres distance. That’s the sodium D doublet: 589.0 and 589.6 nanometres, separated by 0.6 nm. Fraunhofer mapped these in 1814 and used them as the standard for measuring refractive index in optical glass. They’re in every street lamp, every kitchen salt flame test, every solar absorption spectrum. I’d been filtering them out with the pyrometer’s red glass to avoid seeing them. Now I wanted nothing but.

Tried a compact fluorescent bulb next—one of those spiral tubes people phased out when LEDs got cheap. CFL phosphors are excited by mercury vapour discharge, and mercury has a distinctive line spectrum. Slit, prism, dark room. Four clean lines appeared: violet at 405 nm, blue at 436 nm, green at 546 nm, yellow at 577 nm. Not a smooth rainbow. Discrete photon energies where electrons drop between quantum states. You’re looking at orbital mechanics, not colour.

That’s the moment where it flips from “I’m separating colours” to “I’m watching atoms emit photons at exact frequencies determined by their electron shell structure.” Same light source you’ve walked under a thousand times, but now you can see the quantum signature. It’s not wavelengths—it’s strontium versus barium versus copper, each with a fingerprint you can measure.

Prisms disperse nonlinearly, which is annoying. Blue spreads out more than red because refractive index varies with wavelength. A diffraction grating would give linear dispersion—constant nanometres per millimetre across the whole spectrum—but gratings require precise alignment and I didn’t have one. The prism distorts your wavelength scale. You calibrate against known lines (sodium D, mercury green) and interpolate everything else. It’s workable, just tedious.

Slit width matters more than I expected. Too wide and the spectral lines blur together into mush. Too narrow and you don’t have enough light to see anything. The sweet spot for handheld observation is around 0.2–0.4 mm, which is roughly the thickness of five sheets of paper. Also learned that slit rotation matters—the long axis has to be parallel to the prism’s dispersing edge or you get skewed, ugly lines. Spent twenty minutes wondering why my sodium lines looked like diagonal smears before I noticed the razor blade was tilted.

Pointed it at the Sun for exactly three seconds before remembering that’s a terrible idea without a proper solar filter. Did get a glimpse of dark absorption lines mixed into the continuous spectrum—those are Fraunhofer lines proper, where elements in the solar atmosphere absorb specific wavelengths. But also telluric lines from Earth’s oxygen at 759 nm and 686 nm, which look identical to solar absorption unless you know what you’re looking at. That’s the trap for beginners: not all dark lines come from the source. Some come from the air between you and it.

Tried flame tests with kitchen chemicals. Sodium chloride gives intense yellow (589 nm, predictably). Potassium chloride from salt substitute shows deep red at 766 nm, right at the edge of visible. Copper sulfate from the pool supply aisle burns blue-green around 515 nm with a secondary yellow line at 578 nm. Strontium from a flare I tore apart: orange-red at 606 nm. You’re burning elements and reading their electron transitions. Same technique they used in the 1860s to discover cesium and rubidium—light emission before anyone understood quantum mechanics.

Wanted to photograph the spectra with the DSLR. That’s where it got frustrating. Camera sensors have Bayer filters—each pixel is masked with red, green, or blue filter glass in a checkerboard pattern. The camera blocks two-thirds of the spectrum at every pixel to reconstruct colour. When you’re trying to measure specific wavelengths, that’s useless. The 589 nm sodium line lands on some green pixels, some red pixels, all at different sensitivities, and the demosaicing algorithm smears it into an interpolated mess. Silicon sensors also drop off hard below 400 nm (UV) and above 700 nm (near-infrared), so you’re only seeing a truncated window of what’s actually there.

For quantitative wavelength measurement, you’d need a monochrome sensor with no filters, or a scanning spectrometer with a proper diffraction grating and calibrated detector. Or just film, which has surprisingly good spectral linearity if you’re patient enough to develop it. But I’m not doing spectral analysis for publication—I’m looking at light sources and learning which ones emit continuous blackbody spectra (incandescent bulbs, hot metal) versus discrete line spectra (sodium lamps, mercury discharge, flame tests). The distinction is immediate and visceral once you’ve seen it through a prism.

There’s something disorienting about realizing that “yellow light” isn’t a single thing. Incandescent bulbs produce yellow by emitting all wavelengths from 570–600 nm with roughly equal intensity—a smooth continuous curve. Sodium lamps produce yellow by emitting two wavelengths at 589.0 and 589.6 nm and nothing else. They look identical to your eye. The prism reveals they’re completely different physical processes.

Still haven’t measured the forge light. That’s what started this, technically. Blacksmith’s “dull red” heat at 850°C emits peak thermal radiation around 3.4 micrometres in the infrared—totally invisible—with a long tail extending into the red-orange visible range. A prism would show continuous blackbody emission skewed toward red. Not interesting. But the flame around the hot steel? That’s coal combustion, molecular emission bands from C₂ radicals, sodium contamination from coal ash, calcium from limestone flux. Might show discrete features overlaid on the thermal background. Worth checking.

A prism, a razor blade, and a dark room. That’s all the equipment you need to see quantum mechanics with your naked eye.