In 1950, the Gilbert U‑238 Atomic Energy Laboratory appeared on toy‑store shelves as an educational kit. This oddity resurfaces from time to time on social media; Facebook just served it up to me again. Back then it cost $49.50 (roughly $650 in today’s money), but nowadays—when it very rarely turns up—it can fetch $4,000–5,000 at auction.
Which reminds me: wouldn’t anyone like to buy me one for my name day? Just saying—András Day also falls on February 4… No, no, don’t buy one! I really can’t expect anyone to mess around with such dangerous junk—just hand over the $5,000 in cash and I’ll take care of the rest ;-). I did just find an eBay listing offering one for $13 plus shipping, but thankfully no one had lost their mind: the fine print explained that it wasn’t the kit, only a 2.5 × 3.5 inch image of it that you can stick on your fridge as a magnet—clever… I hope “fridge magnet” means there’s actually a magnet on it, though it might also mean you have to order that separately and glue it on yourself.
Gilbert U‑238 Atomic Energy Laboratory
I did, however, find an excellent three‑minute video in which Voula Saridakis, PhD, curator at the Museum of Science and Industry in Chicago, presents the kit created in 1950 by Alfred Carlton Gilbert—American athlete, magician, toy maker, and entrepreneur. The punchline was that A CHILD COULD EXPERIMENT WITH REAL RADIOACTIVE SOURCES.
Radiation is good for kids—it toughens them up early. Or does it? “What doesn’t kill you makes you stronger,” says folk wisdom—and as we know, “folk” is often synonymous with primitive and foolish. In this case, doubly so, and I’ll explain why.
What Was Actually in the Box?
The kit contained:
• Four glass jars holding ore samples with natural uranium content (mostly ²³⁸U): autunite, torbernite, uraninite, and carnotite.
• In addition, there were polonium‑210 (α‑ and β‑source), ruthenium‑106 (β‑source), and zinc‑65 (γ‑source), in the form of flat disks.
• A Wilson cloud chamber, allowing the observer to see alpha and beta particles (a small wire‑shaped ²¹⁰Po source was located at the center of the glass sphere).
• A modest‑quality spinthariscope (essentially the nuclear version of a “flea viewer”: a film‑canister‑sized device with a magnifying lens on one side and a zinc sulfide screen on the other, on which scintillations caused by radiation could be seen).
• An electroscope, for observing ionization effects.
• A battery‑powered, primitive “Geiger–Müller counter” that didn’t actually count anything; it merely indicated radiation via clicks in an earphone and flashes from a neon lamp.
• Red and green rubber balls with small rods, from which an alpha‑particle model could be assembled.
The accompanying manual even suggested exciting games such as “hide‑and‑seek” with radioactive sources: one child hides the sources in the room, the other finds them using the GM counter. Ahem. As I recall, adults played similar games in the forests around Chernobyl… You can’t start too early!
In his 1954 autobiography, Gilbert wrote that the government had encouraged development of the kit because it believed the “laboratory would help the public understand atomic energy and highlight its constructive aspects.” He also claimed the kit was harmless. Nevertheless, the press and social media periodically label it “the most dangerous toy of all time.”
Atomic Optimism and the Era Before Toy Safety
In the 1950s, few consumer‑protection laws regulated toy safety in the United States. Instead, manufacturers responded to public opinion and consumer tastes, which since World War II had been strongly pro‑science. Radioactivity was not widely regarded as dangerous at the time, although scientists were already aware of the risks. Recall that in the early 1900s radium was still considered a miracle cure: radium‑infused drinks were sold as medicinal products for all sorts of ailments, much like Béres Drops today. Children were even given radium slippers that glowed under the bed at night so they wouldn’t fear “monsters”—though arguably the slippers themselves were the monsters.
Only from the 1960s onward did books such as Rachel Carson’s Silent Spring (1962, Houghton Mifflin) appear, presenting the environmental and health hazards of chemicals to the general public. These led to congressional investigations and eventually to regulation. In the U.S., it was only in 1969 that a child‑protection law was enacted that also specified that items like dynamite and strychnine could not be sold as toys. Apparently they had been sold before? Good morning!
Of course, what seemed like victories for child safety were losses for science education. Chemistry experiment kits were radically simplified, and the substances they contained were either eliminated or heavily diluted.
Estimated radioactivity
The Gilbert U‑238 Atomic Energy Laboratory kit included several radioactive materials, but publicly available descriptions usually list only the source types and isotopes, not their activities. I can only estimate that the natural mineral samples had activities of roughly 50–100 kBq, except for the uraninite sample, which could have been an order of magnitude higher. Estimating the activity of the flat artificial sources is even harder, since we don’t know how much material was deposited on each disk. I’ve worked with similar low‑activity sources and would guess these too were on the order of 50–100 kBq. According to recollections, the GM counter “clicked and crackled” vigorously in their vicinity.
A GM counter fundamentally detects beta and gamma radiation (alpha only with special thin‑window designs). A beta particle or gamma photon emitted in a decay can wander off in thousands of directions or be absorbed (there is self‑shielding even within the source). So it doesn’t necessarily reach the GM tube—and even if it does, the tube may not detect it. Detection efficiency can theoretically be around 20% for beta and 1% for gamma, but many metal‑walled tubes absorb beta more strongly. As a very rough rule of thumb: if a roughly point‑like source causes a GM counter to click about 10 times per second (10 cps), that corresponds to roughly 50–300 kBq for a gamma source, or about 3–5 kBq if beta dominates. This also suggests that the activities were in the estimated range. (For clarity: kBq = kilobecquerel = 1,000 becquerels. One becquerel is an activity of one decay per second.)
Time: The Great Decontaminator
It should be added that radionuclides decay over time; activity decreases exponentially. The half‑life of ²³⁸U is 4.5 billion years, so the activity of natural minerals has effectively not changed over the past 75 years. The half‑life of ²¹⁰Po is about 138 days, so the polonium in the cloud chamber and on one of the disks has completely decayed by now (within 7.5 years its activity dropped to one‑millionth). Thus, the nice cloud chamber in surviving kits today is dead—purely decorative. The half‑life of ¹⁰⁶Ru is 373.6 days, that of ⁶⁵Zn 244 days, so those too are long gone from kits that appear at auction.
Would It Be Legal Today?
In any case, according to today’s regulations (Hungarian Atomic Energy Authority Decree 3/2022 (IV.29.), Annex 1), the original activities were about five to ten times the exemption level (10 kBq). And there were eight such sources in the box! Today, without a permit from the Hungarian Atomic Energy Authority, one could neither sell nor even possess such a product. As for the actual hazard: it is indeed negligible as long as the sources remain sealed. Based on available data, the dose from one hour of play can only be estimated. For gamma sources, “looking at” them from 10–30 cm yields on the order of 0.1–1 µSv/h, while “direct contact” (holding them) could reach 10–100 µSv/h; for beta radiation, “looking at” might give 10–100 µSv/h, and holding them could result in skin doses up to 1 mSv/h. By comparison, a dental X‑ray or a Budapest–New York flight is about 100 µSv (mainly from cosmic radiation)—so this is not dramatic.
In mineral samples, alpha radiation dominates, which is about 20 times more biologically damaging than the other two, but alpha particles cannot even escape the glass (the GM counter detects secondary beta and gamma radiation from uranium decay products). So that’s fine too—as long as everything stays in the jar.
So is it dangerous or not? As I said: not dangerous—as long as the sources are sealed, meaning the radioactive material cannot smear onto the skin, get into the eyes, or be swallowed by a child. But what if a child does swallow it? If the source is not sealed and contamination occurs, local skin doses could reach on the order of 100 mSv/h.
The Real Problem: Internal Exposure
The main risk is typically internal exposure. If the mineral crumbles and dust escapes the container and enters the body (via inhalation, ingestion, or rubbing into eyes or wounds), it causes internal dose. A single poppy‑seed‑sized piece of uraninite (1–10 mg), by my rough estimate, could deliver a committed equivalent dose of 20–200 µSv from ²³⁸U and its decay products (²¹⁰Po, ²¹⁰Pb, ²²⁶Ra). That doesn’t sound like much, since natural potassium‑40 in the body of a 30–40 kg, 10‑year‑old child already contributes about 50–100 µSv per year (100–200 µSv/year in adults, varying by individual—more in muscular people). The small snag is that ⁴⁰K is evenly distributed throughout the body, whereas that poppy‑seed‑sized uranium mineral could sit in one spot, say the appendix.
Of course, exposure from natural background radiation, internal ⁴⁰K, food, and building materials adds up to about 3–4 mSv/year, depending on where one lives (with radon contributing one‑half to one‑third). Added to this is another significant and growing source: medical imaging (CT scans are brutal—2–12 mSv per exam). Less well known is that, at least here, exposure from industrial sources and nuclear accidents (e.g., Chernobyl) contributes only a few tenths of a percent overall—though locally, near accident sites, it can be dominant. Around nuclear power plants, typical annual public doses are only a few tens of microsieverts (0.01–0.03 mSv). But if a child eats uranium, that adds to everything else. So: don’t eat uranium!
What We Know—and Don’t Know—About Low Doses
The physiological effects of acute medium and high doses (≳0.1 Sv) are well documented. For example, a worker receiving 1 Sv in an accident would almost certainly be hospitalized with radiation sickness. The symptoms and course of radiation sickness at various doses are well known. By contrast, the effects of chronically received low doses are highly uncertain. If the same worker accumulates 1 Sv over 50 years, they might get away with it—but cancer risk increases: it’s not certain that cancer will occur, but it’s not certain that it won’t. In this range, deterministic effects (like radiation sickness or tissue necrosis) do not dominate; instead, stochastic processes do—primarily mutations caused by direct or indirect ionization of DNA molecules. The body’s repair mechanisms usually fix these errors, but not always perfectly, so errors accumulate over time. Cumulative damage increases the probability of malignant cancers. In principle, it may also increase the risk of heritable genetic changes via germline mutations—but evidence for this is very limited.
While for large doses there is a more‑or‑less definable, though not sharp, threshold above which harm occurs, for low doses there is none: risk is roughly proportional to dose. That is, the more radiation someone receives, the higher the chance of developing cancer—but this does not mean that those who receive little will certainly be fine, or that those who receive more will certainly get cancer. It’s a lottery. The scientific community generally applies the linear no‑threshold (LNT) model, which assumes risk is directly proportional to dose. In reality, the exact shape of the dose–response curve is uncertain, especially at very low doses, so radiation protection takes a conservative approach and extrapolates linearly. Even today, there is no full consensus on the biological effects of very low dose‑rate, chronic ionizing radiation.
As for the Atomic Energy Laboratory: although it entered popular lore as “the most dangerous toy ever made,” this is simply not true. The product was not successful; only about 5,000 units were produced in 1950–1951 before manufacturing ceased. The reason at the time was not concern over radioactivity, but the high price. Gilbert later felt the product had been poorly positioned: it was better suited to those with some prior education. The A.C. Gilbert Company offered many other experiment kits, mainly chemical ones, aimed at younger audiences.
The real most dangerous games
If one wants to know what the most dangerous toy of the past century really was, one should look for products that statistics show actually caused many injuries and were therefore banned.
Jarts (Lawn Darts)
At the top of the list are Jarts (or Lawn Darts). This popular outdoor game of the 1980s was essentially a garden version of darts. It consisted of four large (~30 cm), heavy (~150 g), metal‑tipped darts with plastic fins, and two target rings. Players were supposed to toss the darts into the rings, the idea being that they would stick into the ground. The problem was that children hit each other instead of the rings. In the U.S. between 1980 and 1988, 6,100 children were treated in emergency rooms for dart‑related injuries. Three children died after being struck in the head. Under pressure from parents and advocacy groups, the U.S. Consumer Product Safety Commission banned the product, and later that same year (1988) Canada imposed a full ban as well. In the EU, lawn darts can technically be sold, but only with blunt plastic tips or in magnetic versions.
Bindeez (Aqua Dots)
Second place could go to Bindeez (also known as Aqua Dots). These are still sold today, even here. It’s a craft kit containing lots of colorful beads and a board. Beads are arranged on the board according to patterns. The most popular versions today use small cylindrical beads that are fused with a warm iron (iron‑on beads such as Hama or IKEA Pyssla), but non‑iron versions also exist, where the design is fixed by spraying with water. The original Bindeez worked this way. They were marketed by the Canadian toy company Spin Master. When sprayed with water, the adhesive coating on the beads dissolved and then dried, sticking the beads together. The product was banned in the U.S., Canada, the EU, and Australia in 2007 because children swallowed the beads. Not because they choked—but because they fell into comas from severe GHB overdose. The adhesive contained 1,4‑butanediol, which is metabolized into gamma‑hydroxybutyric acid (GHB), a narcotic. In small doses GHB is used medically (e.g., for treating alcoholism or sleep disorders) and is also known as a party drug causing euphoria and disinhibition; in larger doses it causes nausea, dizziness, loss of consciousness, and death. There was no GHB in the adhesive—but no one considered that butanediol would be converted in the child’s liver (by the same enzymes that metabolize alcohol: alcohol dehydrogenase first oxidizes butanediol to gamma‑hydroxybutyraldehyde, then aldehyde dehydrogenase converts it to GHB). The beads were recalled worldwide after children lost consciousness. An unconscious game indeed…
Cabbage Patch Kids (The “Eating” Doll)
Third place might go to the Cabbage Patch Kids. These cloth dolls with plastic heads were first produced by Coleco Industries in 1982, inspired by rare handmade dolls by Martha Nelson. Xavier Roberts, a 21‑year‑old art student, modified Nelson’s dolls just enough to avoid being sued when he began selling them in a Cleveland toy store. Initially, other manufacturers rejected them as too ugly. Coleco, successful in electronic toys, took the plunge and made them a global hit. So much so that in 1983, riots broke out in the U.S. as parents fought to buy one for their children. Coleco upped the ante with a talking version featuring a sound chip, touch sensors, microphone, and short‑range 49 MHz AM transceiver. It detected how it was played with and said things like “Hold my hand!” Sometimes it asked for a drink, and when several dolls were nearby, they interacted and encouraged group play. After Coleco went bankrupt in 1988, the dolls passed to Hasbro, then to Mattel in the mid‑1990s. Hasbro switched from cloth to vinyl and reduced the size from about 40 cm to under 35 cm. Mattel tried adding gimmicks: some dolls played water games, swam, brushed teeth—or ate. That’s where trouble began. Mattel implemented eating by installing small metal rollers in the doll’s mouth, driven by an electric motor. Plastic “snacks” were stored in the doll’s backpack. When a child put a piece into the mouth, the doll “swallowed” it, and the food magically reappeared in the backpack. The mechanism could be disabled by removing the backpack. The problem was that it couldn’t distinguish between plastic food and a child’s hair or fingers. After multiple injuries (broken fingers, scalping), the product was recalled. “The real horror doll—from Mattel.”
Hungary’s Tiki‑Taki and the Myth of the “Ban”
In Hungary, the most dangerous toy was probably the Tiki‑Taki, popular in the 1970s. It consisted of two hard plastic balls connected by a string with a ring. By moving the wrist, players made the balls collide, producing a rapid clacking sound. I recently read in a tabloid that it was banned because its “machine‑gun‑like” sound made police think we were under attack by imperialists—a quaint image of the communist years. But: 1) Hungary never had communism; we tried to build “socialism,” unsuccessfully. 2) We really did fear imperialists, not terrorists—those we supported (including Palestinians and Western communists and separatists, who vacationed here and received training, weapons, and explosives). 3) The Tiki‑Taki was not banned because of its sound. It was first classified as a dangerous device in the U.S. by the FDA. The reason was that the balls sometimes struck children’s wrists or hands, causing bruises, fractures of fingers and wrists. The greatest danger was that the hard acrylic balls fatigued over time and suddenly shattered, sending shards into children’s faces and eyes. The long string also posed a strangulation hazard, and yes, the noise could cause hearing damage over time.
The Safety Label Is Written in Court Records
What these toys have in common is not that “everything used to be more dangerous,” but that designers consistently underestimated children’s creativity. These toys became dangerous not because they were designed by evil people, but because adults tried to model children using adult logic—assuming “they obviously won’t do that.” But they will. They’ll try it, eat it, take it apart, throw it at a sibling, wrap it around their neck, stick it where it doesn’t belong. Not out of malice, but because this is one way learning happens. Strange as it may seem, a not‑entirely‑dim child learns even when parents and schools try their hardest to prevent it.
Today’s toy safety standards—choking tests, chemical compliance lists, age ratings—are largely the hard‑won lessons of past accidents, written in blood, tears, and court records. Behind every rounded edge and every “not for children under 3” label lies a statistic—and often a child who was in the wrong place at the wrong time, near the wrong toy. The next time someone sighs nostalgically, “back in our day…,” let’s remember this too.