A Home Cloud Chamber Revisits Rutherford's Gold Foil Experiment
September 2026
Ernest Rutherford. Bain News Service, undated (circa 1920s by visual estimate), Library of Congress, restored. Public domain.
In 1909, Hans Geiger and Ernest Marsden, working in Ernest Rutherford’s laboratory at the University of Manchester, aimed a beam of alpha particles at thin metal foils and counted how many scattered back toward the source. Rutherford had asked them to check specifically for large-angle scattering, not expecting to find much of it. What they found instead, a small but unmistakable fraction of alpha particles bouncing back at steep angles, had no explanation in the physics of the day.
A century later, a science creator on TikTok pointed a home cloud chamber at an americium-241 button (a stand-in for Rutherford’s radium source) and a sheet of gold leaf, and captured something that looks, at a glance, like the same result: alpha tracks passing straight through, and every so often, one appearing to head back the way it came. Whether that glance holds up, and what it would take for it to, is worth working through directly.
@emyeukhoahoc6
Original video: @emyeukhoahoc6 on TikTok, “This Is What It Looks Like When Atoms Collide (2).” If the embed above doesn’t load, watch it directly at that link.
What a cloud chamber actually shows
A cloud chamber is a sealed container holding air saturated with alcohol vapor, chilled until the vapor sits right on the edge of condensing. A charged particle passing through ionizes the gas molecules along its path, and the supersaturated vapor condenses onto that trail of ions before it condenses anywhere else. The streak isn’t the particle. It’s a visible record of the particle’s path, left behind by the ionization it caused, and it fades within a second or two.
The short, dense tracks in the video are the signature of alpha particles, each two protons and two neutrons bound together, most likely launched from the decaying americium-241 button the video shows. Americium-241’s half-life is 432.2 years, a standard figure cross-checked against the World Nuclear Association’s and CDC’s published values, so the button won’t run dry within a human lifetime. The video’s own “about 432 years” is consistent with that figure, which is the reason a transcription slip in this piece’s own early drafts, misheard as “amaryseum,” could be caught and corrected (see the notes on sources below).
A track being straight and long mostly reflects how alpha particles lose energy: continuously, through many small ionizing collisions with electrons along the way, until they run out of energy and stop. That’s a different mechanism from a rare, hard deflection off a nucleus, and it’s worth keeping the two apart rather than treating “the track looks unobstructed” as evidence about what’s inside the atom.
The original experiment, and what forced a new model
Before 1909, the working picture of the atom was J.J. Thomson’s “plum pudding” model: a diffuse cloud of positive charge with electrons distributed through it. That picture allows alpha particles to be nudged off course by many small, soft interactions as they pass through a foil, but nothing in it can turn a fast, heavy alpha particle sharply backward. A diffuse charge just isn’t concentrated enough anywhere to do that.
Diagram of the Geiger-Marsden apparatus, from Geiger and Marsden’s 1913 paper, “The Laws of Deflexion of α Particles through Large Angles.” Public domain.
Geiger and Marsden’s gold-foil measurements found alpha particles deflected by about 90 degrees or more at a rate on the order of 1 in 10,000 to 1 in 20,000, depending on the specific foil and the angle cutoff used in a given run, small but far too large to explain as the sum of many small nudges. Rutherford described his own reaction to that result in a widely quoted account, commonly dated to a 1936 lecture at Cambridge:
“It was quite the most incredible event that has ever happened to me in my life. It was almost as incredible as if you fired a fifteen-inch shell at a piece of tissue paper and it came back and hit you.”
Rutherford worked out, in 1911, that the structure consistent with that scattering data was a compact, positively charged nucleus at the atom’s center, with the rest of the atom essentially open space around it, and that no diffuse-charge model could produce a comparable rate. That reasoning, not any single event, is what the plum pudding model couldn’t survive: a measured rate of hard deflections, compared against a formula, and found to require a concentrated charge.
What the video shows, and what it doesn’t prove
Geiger and Marsden’s setup used a collimated beam: every alpha particle started out traveling in essentially the same direction, toward the foil. That single fact is what makes a backward-moving particle meaningful. If every particle begins by moving one way, and one is later found moving the other way, something turned it around, and there’s only one candidate nearby able to do that.
A bare button source doesn’t work that way. Americium-241 in an uncollimated setup emits alpha particles into a wide angle, not a defined beam. In a chamber built around a source like that, some tracks will appear to point back toward the source simply because that’s the direction they happened to be emitted in, never having been deflected by anything at all. A straight track that looks like it’s heading back the way it came, viewed on a flat video frame, could be exactly that: ordinary, undeflected emission, seen from an angle that makes it look like a reversal. That’s a real limit, but not the only one worth checking, since a cloud chamber (unlike Geiger and Marsden’s fluorescent screen) draws the whole path, so a genuine deflection would ideally show up as a visible bend at the point the particle actually hit something, not just an oddly-aimed straight line.
Watching the clip’s own frames closely, frame by frame, around the moment it calls out: the “backward” streak sits immediately beside the source’s own foil-wrapped housing, not as a track that visibly travels out to the gold leaf and bends at a resolvable point on its surface. There’s no observable kink and no deflection vertex to point to, just a bright, straight streak right next to where the alpha particles originate. That’s consistent with an ordinary, undeflected particle that was simply emitted in that direction to begin with, and it’s not what documented evidence of a scatter would need to show. Without collimation to define an incident direction, and without a visible kink to substitute for one, this home setup can’t tell a genuine large-angle scatter apart from a particle that started out going that way, and the real Rutherford backscatter rate (order 1 in 10,000 to 1 in 20,000 for a thin gold foil) is rare enough that “every so often” in a short clip from an undirected source is, if anything, more consistent with ordinary geometry than with nuclear scattering.
That’s the plain answer to what the video shows: real alpha particles, genuinely ionizing a gas and leaving visible trails, which is worth seeing on its own terms. It isn’t, even qualitatively, a demonstration of the specific effect that overturned the plum pudding model, because neither of the two things that could have made a backward-looking track meaningful, a defined beam direction or a visible deflection point, is present in the footage. What did the actual work in the original experiment wasn’t the existence of a backward-moving particle. It was a collimated beam, a counted rate, and a comparison against a formula, none of which a bare source and a phone camera supply.
None of that makes the video worthless, and it isn’t new in the sense of being a novel apparatus. Home-built diffusion cloud chambers, using dry ice and a smoke-detector source, have been a classroom demonstration for decades. What’s genuinely new is how easily a short video can put a working one in front of a wide audience with no lab access required. That’s a real and worthwhile thing. It just isn’t the same claim as recreating Rutherford’s evidence.
One number in the video worth checking
The narration states that 99.4% of a gold atom’s mass sits in its nucleus. That’s the right idea in the wrong ballpark, and it’s worth showing the actual arithmetic rather than repeating the figure. A gold-197 atom has 79 electrons, each with a mass of 0.00054858 atomic mass units, against a total atomic mass of 196.966569 u:
$$\frac{79 \times 0.00054858}{196.966569} \approx 0.00022 = 0.022\%$$
Electrons account for about 0.022% of a gold atom’s mass, which puts the nucleus at about 99.98%, not 99.4%. The qualitative point the video makes, that the nucleus carries essentially all of an atom’s mass in essentially none of its volume, is correct, and it’s the reason a concentrated charge can turn an alpha particle around at all. The specific number just needed re-deriving rather than repeating.
The point of pairing these two
The gap between this video and Rutherford’s actual evidence isn’t a matter of degree, fewer particles counted, a little less rigor, same basic idea. It’s a missing ingredient: collimation. Once a source fires in every direction at once, a backward-looking track stops being informative, because ordinary, undeflected particles already point every which way. The 1909 experiment worked because Rutherford’s apparatus made a backward particle mean only one thing. A home chamber, however striking to watch, hasn’t reproduced that condition, and describing a visually similar clip as showing “the same phenomenon” would repeat the exact mistake that calling it “the same measurement” already was, just one level further down.
What holds up unchanged since 1909 is the reasoning: a diffuse charge cannot turn a fast alpha particle around, so whatever explains the real, measured backscatter rate has to be concentrated. That argument doesn’t need this video to be true. It needed Geiger and Marsden’s actual apparatus, and it still rests on that, not on anything a phone camera can add.
Notes on sources, and how firm each claim is
Video source and correction. The video was reviewed directly from TikTok, and its audio was transcribed locally (not published) to check specific claims against independent references, per this site’s re-derive-don’t-transcribe standard. The transcription software misheard “americium” as a nonsense word close in sound. The video’s own “about 432 years” is consistent with the accepted 432.2-year half-life of americium-241, which points toward a mishearing rather than an unrelated isotope, though it isn’t independent lab proof of which isotope is actually in the button.
Half-life. 432.2 years, cross-checked against the World Nuclear Association and CDC.
The large-angle deflection rate. Commonly cited secondary figures for Geiger and Marsden’s gold-foil results range roughly from 1 in 10,000 to 1 in 20,000 for deflections of about 90 degrees or more, depending on foil and angle cutoff; stated here as a range rather than a single figure, after an earlier draft used an unverified 1-in-8,000 figure (from a platinum-derived number, not checked against gold specifically) and a second draft asserted a single unhedged 1-in-20,000 figure without a primary citation. Both caught during two rounds of adversarial review before publication.
Rutherford’s quote. The wording used here matches the version most consistently attributed to Rutherford, commonly dated to a 1936 address to the Cambridge Philosophical Society and recounted in secondary histories including the American Physical Society’s own piece on the discovery. No primary transcript of the lecture was located, so the attribution rests on secondary sourcing rather than a first-hand printed record.
Mass-in-nucleus figure. Recomputed directly from gold-197’s atomic mass and electron rest mass, both standard published constants, rather than taken from the video or from memory.
Images. Rutherford portrait: Bain News Service, Library of Congress, public domain, restored; the Library of Congress records no confirmed date, and “circa 1920s” here is a visual estimate from the Commons restoration record, not a documented date. Geiger-Marsden apparatus diagram: from Geiger and Marsden’s 1913 paper, “The Laws of Deflexion of α Particles through Large Angles,” per the file’s own Wikimedia Commons description. Both public domain, via Wikimedia Commons.
Adversarial review, three rounds. This piece went through three independent adversarial passes (CLAUDE.md item 12) before publication. The first found the draft overclaiming the video as equivalent, evidence-grade proof of the original result: calling a single reversed streak a “direct hit,” treating straight air tracks as confirming an empty atom, and using an unverified deflection-rate figure. The second pass, run to confirm that fix actually held, found the deeper problem underneath it: even the softened claim that the video shows the same qualitative phenomenon as Rutherford’s result doesn’t hold on collimation grounds alone, because an uncollimated source can produce backward-looking tracks through ordinary emission geometry with no scattering involved. The third pass confirmed the collimation argument itself but flagged that it wasn’t the whole story: a cloud chamber, unlike Geiger and Marsden’s screen, draws the full path, so a genuine deflection could in principle still show up as a visible bend at the foil, and the piece hadn’t checked for one. That was answered by going back to the actual footage rather than a fourth round of review: frame-by-frame, the “backward” streak sits immediately beside the source’s own housing, with no track visibly traveling to the foil and bending at a resolvable point. No kink, no deflection vertex, which is what settles the question the third pass raised. The “What the video shows, and what it doesn’t prove” and “The point of pairing these two” sections reflect all three rounds; none of the three drafts each review actually looked at survives into what’s published here.
Categorization. Proven, for the historical physics: Rutherford’s nuclear model is undisputed, century-repeated, textbook nuclear physics. The video is presented as a real demonstration of alpha-particle tracks, not as evidence of the specific scattering effect the historical experiment established.