A column about curious stories from the worlds of science, technology, and film. Today — tartrazine, aka E102, aka FD&C Yellow No. 5: the yellow dye from American chips and soda that, in 2024, suddenly turned out to be the most effective known way of making living tissue transparent. First at Stanford — on mice and supermarket chicken. And now — in an ordinary kitchen, in a replication more compelling than any pop-sci piece: with a failed solution, a crystalline crust, and chicken that looks like a gummy bear.
Let me begin with a child’s question: why is meat opaque? Water is transparent, protein in solution is transparent, fat is translucent — mix them in the right proportions, and you get chicken. The answer is not chemistry but optics: scattering. Inside tissue lie billions of microscopic boundaries between the water-based medium (refractive index around 1.33) and lipid-protein structures such as membranes and myofibrils (1.43–1.53). At every boundary, light bends slightly, and after hundreds of such collisions, a beam turns into fog. In most tissues, the scattering coefficient is 10–1000 times greater than the absorption coefficient — meaning meat is opaque not because “something eats the light,” but because light ricochets around inside it like a pinball.
The most honest everyday analogy is snow. Every snowflake is transparent ice, yet a handful of snow is white and opaque: the ice–air boundaries are to blame. In meltwater, where those boundaries have vanished, the same water remains — and suddenly it is transparent. Tissue “melts” in exactly the same way if you match the refractive indices of its components. The whole problem is how to do that without killing the tissue.
The classic solution is to flood the tissue with an agent with a high refractive index: glycerol, fructose, sucrose. Here physics sticks out its foot, and this is where things get interesting. To raise water’s index from 1.33 to the necessary 1.45, you need 11.6 molar glycerol — practically pure glycerol. Hence the familiar rituals of “tissue clearing”: 80-percent solutions, dehydration, sample shrinkage, dead tissue at the end. For a living organism, this is not a method but an obituary.
The team of Zhixiao Ou, Guosong Hong, and Mark Brongersma at Stanford came at it from the counterintuitive side: strongly absorbing molecules dissolved in water raise its real refractive index. This is not esoterica but the Kramers–Kronig relations — a fundamental property of causality linking absorption and refraction spectra. The sharper the absorption resonance in the blue part of the spectrum, the more strongly the refractive index is lifted in the red and infrared. A dye that “devours” blue light works in red light as a fine refractive-index engineer — without absorbing it.
The authors tested 21 candidate dyes. The choice fell on the absolute champion of the blue region. Tartrazine’s absorption peak is 428 nm, which is why it is yellow; above 600 nm, it is virtually transparent. Its efficiency — the increase in refractive index per unit of concentration — is one and a half dozen times higher than glycerol’s, while the serendipitously discovered Dye-21 for near-IR is a hundred times higher. One caveat: the window of transparency opens in the red region of the spectrum, and you cannot overdo the concentration — if the index overshoots, scattering returns, and the transmission curve predictably falls. Physics does not forgive excess, and that is beautiful: the effect has a steep, testable dosage.
And crucially: transparency arises not because the dye swallows stray rays, but because there are fewer scattering events in the tissue itself. Fluorescent markers still glow through cleared tissue, a hydrogel with silicate beads produces a sixtyfold increase in transmission in the red region, and wave simulations show a wavefront passing through the medium as though through clean water.
The protocol is astonishingly simple. A 0.6 M tartrazine solution is rubbed into the skin of an anesthetized mouse. Ten minutes later, windows open: on its abdomen, you can see a living mouse’s intestines — peristalsis in real time, fluorescently tagged intestinal neurons; on its scalp — blood vessels in the brain; on its paw — muscle sarcomeres. The window lasts ten to twenty minutes, then the dye diffuses back out and the tissue clouds over again. Wash it off, and you can repeat the process. Reversibility is the key word here: conventional clearing agents killed the tissue, while tartrazine simply lets it go.
With chicken, there is a demo delivered with a laboratory sense of humor: beneath a two-millimeter slice of chicken breast, they placed a printout bearing the word “Stanford” and drowned it in solution. A couple of hours later, details five micrometers across can be read through the meat at a depth of one millimeter — light-microscope resolution through a layer that, that very morning, was “frosted glass.” Tartrazine diffuses an order of magnitude faster than the sucrose rituals: two hours for a two-millimeter slice versus a day.
Now for the backstory, without which the story is incomplete. Tartrazine was synthesized in 1884 in Basel by the dye chemist Johann Heinrich Ziegler — the company was called Bindschedler'sche Fabrik für Chemische Industrie, the paper came out in 1887, the BASF patent in 1885; at first it dyed wool, later food. Today’s list of names: FD&C Yellow No. 5, E102, Acid Yellow 23. One hundred and forty years of an honest food career without a single explosion.
The turn came in 2007. The University of Southampton published a study in The Lancet involving 297 children — 153 three-year-olds and 144 eight-year-olds — six weeks on identical-looking drinks: some with mixtures of artificial colors and the preservative sodium benzoate, some without. Double-blind, commissioned by the Food Standards Agency for 750 thousand pounds. The conclusion: the mixtures were associated with increased hyperactivity. Not one dye was banned — but the European Union required six azo dyes, including E102, to carry the warning “May have an adverse effect on activity and attention in children.” Formally, it was a label; in practice, a sentence: a warning on a package of candy can kill sales faster than a ban. Manufacturers in Europe switched en masse to other dyes; and multinational chip brands, Hacker News regulars have noted, do without E102 entirely in their European versions.
And now, irony upon irony: a dye that survived on the American market within FDA-approved limits turns out, seventeen years after Southampton, to be an optical material. EFSA, as the Stanford paper notes, found no side effects or systemic toxicity at doses of up to 2 grams per kilogram of body weight — for laboratory optics, that is a ready-made pass: high solubility in water, minimal reactivity, a narrow resonance in the blue region, zero absorption in red. All the properties of an ideal clearing agent had been selected for decades by food regulators. For free.
That, really, is the hook. The author of the blog Chill Physics Enjoyer received a link to the Science paper from his friend Adaobi Adibe and decided to do a warm-up before a “more ambitious replication” — it has been promised, we are watching. The plan: chicken breast, three solutions — a zero control, ~0.19 M, and ~0.56 M. The powder was >80% pure, from APC Pure: the actual molarity could be up to 15% lower than nominal, and the author honestly calculates moles from the nominal mass, noting the correction. Kitchen, scales, tap water.
The first discovery: the journal protocol does not scale to a tabletop. The paper’s authors mixed 0.62 M at 40 degrees on an orbital shaker; at home, the 0.56 M solution crystallized by evening into a solid yellow crust — the concentrated batch failed. What remained was the middle one: a 5-mm slice in ~0.19 M for six hours at room temperature, with a control slice in water beside it. The result is honest: partial transparency, not Stanford-grade. The upper layers began to resemble a gummy bear (in the original: gummy bears, “it makes you want to chew it, if not for the risk of catching salmonella”), and a 5×5 mm grid can be read through a two-millimeter cut — the thing that, in the paper, shone through as the word “Stanford.” The second run was an honest repeat: because of the crystallization, only the control and 0.19 M remained, twelve hours in sealed tubes — a nervous eye on food safety instead of the earlier carefree attitude.
What I like about this story is not the experiment itself but its mode of honesty. The author did not announce, “I replicated the discovery of the year”: he showed the before-and-after image, published a table of concentrations with error margins, and separately spelled out where his protocol diverged from the journal’s — no shaker, no heating, lower concentration, longer exposure. The difference from the laboratory is not “the kitchen is worse”; it is a diffusion front that, without shaking and heating, lumbers along at its leisure. Reproducibility as a sport, not as a slogan.
It gets better from here, but not immediately. The clinic is a long way off: human skin is ten times thicker than mouse skin, and Ou directly calculates that “minutes will turn into hundreds of minutes” — you cannot fool diffusion. The FDA approves tartrazine as a food additive, not as a window into the abdominal cavity; allergic reactions are documented; Stanford already has a patent application hanging over the method. Guosong Hong lists the dreams: veins visible without ultrasound (IVs for elderly people are a pain all their own), simpler laser tattoo removal, early diagnosis of skin cancer. Dermatologist Rajan Kulkarni, who was not involved in the work, delivers a verdict I would gladly sign beneath any high-profile paper: “As a proof of concept — fantastic. Clinically — we have yet to see.”
My conclusion is this. For one hundred and forty years, chemistry sought dyes to make the world brighter. It turned out they were there to make it more transparent in red light, and for that the dye merely had to be what it had always been: soluble, harmless in sensible doses, and desperately yellow. Kramers–Kronig physics, nearly a century old, had been waiting inside a bag of chips — nobody had simply asked it to come out. And now it is being tested in a kitchen, with chicken from the nearest supermarket, and to my mind that is the best possible scenario for pop science: not “scientists were surprised,” but “you can be surprised too, provided you bring scales and a table.”
Special respect to Adaobi Adibe — a rare case where the phrase “look at this article” ends with chicken in a dye solution and an entry in a lab notebook.