The Sand Nobody Tested

In the second half of 2025, an Australian laboratory fed a batch of coloured children's play sand into a machine. The sand was not the point. The new machine was, and the lab wanted to know whether it worked. It did, and it reported asbestos: two kinds, tremolite and chrysotile, in the fibre bundles that had ridden into the country inside a product marketed for small hands.⁶

The lab had set the machine running only to prove itself, not to hunt for asbestos; the run was, as the Australian competition and consumer regulator's deputy chair Catriona Lowe put it, "in fact, to test one of their new machines."⁶ The sand had been arriving for years and had never been tested at the border before entering the country — a gap Lowe defended, given the scope any such border testing would require.⁶

Start with the reassuring part, because it is true and it is the whole difficulty of this story. British regulators — the Office for Product Safety and Standards, together with the health and safety agencies — assessed the wave of recalls that followed, and they did not bury the finding; they published it. The risk to health, they wrote, "is expected to be low, for these recalled products during normal use, even with occasional short-term exposures and when following the disposal instructions."⁴ No child is known to have been harmed. No harm is alleged here. The official verdict is low, and this account accepts it completely.

Hold that verdict in one hand. In the other, hold the machine. If a banned carcinogen in a children's toy was found, genuinely found, for the first time, by a laboratory that was testing itself, then the question is not how frightened to be. The risk is low; be as unfrightened as the government says you can be. The real question is quieter and worse. Who inside the system was supposed to find it on purpose? If the answer is nobody, then the low risk was not caught; it was merely lucky.

The "sand" is crushed rock, and certain rock carries the fibre

The word "sand" is doing a lot of work on the box. It calls up a beach: quartz grains, rounded by water, about as inert as a material gets. Coloured "sensory" and craft sand is something else. This sand is rock, quarried and crushed to a chosen grain size and then dyed.

You do not have to take that on trust. One long-established North American colour-sand maker explains it plainly on its own product page: its sand is "produced from the feldspar rock and crushed to our specifications," and the little dark flecks a buyer might notice "are veins that run through the rock as a natural feature."⁷ That page gives the honest picture of the material, from someone who sells it: a rock with veins running through it, broken down into grains. The box says beach; the page says a milled quarry.

Which rock matters hugely, because rock is not a pure substance but an assemblage, and some assemblages come with company. The International Agency for Research on Cancer, cataloguing asbestos in its monograph on the mineral, describes how the fibres occur in the ground. They may sit in "large natural deposits or as contaminants in other minerals." Tremolite in particular forms "by low-grade thermal metamorphism of siliceous dolomites" and turns up as an accessory in talc-bearing and ultramafic rock.⁸ In plain terms, tremolite is a mineral the earth grows next to certain other minerals. Tremolite is a known neighbour of specific rock classes: not present in every deposit, not in every bag, but a recognised natural companion of the kind of rock a colour-sand operation might quarry.

That single fact rearranges everything. The asbestos is not an ingredient; no one chose it, mixed it in, or wrote it on a schedule. It rides in as a property of the raw material — a contaminant that can travel with the rock the way a weed travels with topsoil. And because a known mineral neighbour is a foreseeable one, it is also a checkable one. You cannot screen for a surprise. You can screen for a thing geologists have described for a century.

One more step turns a buried fibre into an airborne one. Asbestos is dangerous when its fibres are loose in the air; intact in solid rock, it mostly sits still. Crushing is what sets it loose. Every risk assessment of these products names disturbance — breaking or damaging the material — as the moment fibres in the sand can become airborne.⁴ A mined-and-crushed filler has already had that done to it. For coloured play sand, the liberation step happened at the mill, before the bag was ever sealed.

So a mined mineral, crushed to release whatever it carries and dyed a cheerful colour, gets sold by weight as an inert craft supply. It carries, on no ingredient list, a fibre its rock class can foreseeably hold. The rest of this account walks the path that sand travels to a child, stopping at each point where somebody might have been posted to look. Several such points exist, and the interest is in how many of them are empty.

The first checkpoint: the toy test cannot see a fibre

The obvious objection arrives early: surely toys are tested. A whole regime exists, a British Standard, a testing industry, laboratories with accreditation, and coloured play sand in a toy is subject to it. Does that whole apparatus not catch a fibre in the sand?

It runs, and it cannot see this.

The chemical-safety test that toy materials pass is EN 71-3, titled "Migration of certain elements." Read its own statement of what it does: "Soluble elements are extracted from toy materials using conditions which simulate the material remaining in contact with gastric juices for a period of time after swallowing."³ The test builds a little artificial stomach. It soaks the material in acid to mimic a swallowed toy sitting in a child's gut. Then it measures how much of a list of elements (aluminium, antimony, arsenic, barium, lead, and a dozen more) dissolves out.³

That hazard is real, well chosen, and the test is good at it. But look at the shape of it. The test is built for the swallowing pathway, and it measures soluble elements, the substances that leach. Asbestos is neither a soluble element nor something stomach acid dissolves. Instead it is a silicate fibre, and its danger is not chemical but physical, a matter of the shape and durability of the fibre, breathed in and lodged in the lung. It hurts you by being inhaled, not swallowed, and by being a shape, not a dose; EN 71-3 never mentions asbestos.³

None of this is a loophole or a failure of the standard; test and hazard are simply a category mismatch. The test that runs on toys was designed to answer a different question, what leaches when a child mouths this, and it answers that question honestly. Asked whether there is a respirable fibre in the sand, it has no way to reply. Its blindness is not an accident; it was never built with eyes for this.

Fine, then: that test is the wrong instrument. But a mismatch only matters if the right instrument is missing too. Surely a test that can see the fibre is required somewhere along the line.

The second checkpoint: the method is written down, and no one has to run it

Such a test exists, and that no one must run it is the strange and central thing.

In March 2026, the Office for Product Safety and Standards published a document titled "Testing for asbestos in sand." This came months after the Australian machine's accidental find and the recalls it set off. The document is exactly what it sounds like: a technical guide to detecting asbestos in "consumer products containing sand such as toys, hobby and decorative products."¹ It walks through the techniques, and it is specific about which ones work. The routine method used on building materials, it warns, "may miss very fine or thin fibres" at low levels. For sand, the reliable techniques are forms of electron microscopy: scanning, paired with X-ray analysis (SEM-EDX), or transmission (TEM).¹ The state, in other words, has looked at this precise problem and written down precisely how to catch it.

And then read what the document says it is. It sets out "OPSS' position on the most appropriate testing methodologies," calls itself "a technical advisory note for businesses and industry," and marks the techniques "recommended."¹ Position, advisory, recommended. Search the guidance for the words that would turn it into an obligation, for a must or a shall or the phrase required to test, and they are not there. A companion page published in July 2026 spells out the logic without flinching. "Responsibility for product safety rests with the business placing the product on the market," it says; those businesses "must ensure they are free from asbestos presence and comply with all applicable legal requirements."¹² Testing the sand is not among those requirements.

Sit with the specific shape of that. The government did not fail to understand the hazard, nor did it fail to know how to detect it. It produced a careful method for finding asbestos in children's sand. Then it framed the method as a suggestion, addressed to the very companies whose products were being recalled, with no duty on anyone to pick it up. The map to the buried thing exists. Following it is optional.

The method exists and requires no one. But perhaps a duty sits further back: not a duty to run this method, but a broader legal obligation to make sure the sand is safe. Doesn't the general law fill the gap?

The third checkpoint: an importer owes a safe product, not a look at the input

This objection deserves the most care, because it is the strongest, and because getting it wrong in either direction is dishonest. Two laws already stood in the sand's path, and both of them are real.

Since 1999, the import and use of asbestos in Britain has been banned outright. The regulators call it a "zero-tolerance approach," under which "consumer products found to contain asbestos must be recalled."⁴ And the General Product Safety Regulations 2005 lay down, in regulation 5, a flat duty: "No producer shall place a product on the market unless the product is a safe product."² A "producer," in that law, expressly includes the importer and the own-brander: the company that brings the sand toy into the country and puts its name on it. So a bag of contaminated sand was, the moment it crossed the border, already something an importer was legally forbidden to sell. Anyone who says "there was no law against this" is wrong, and this account will not pretend otherwise.

Now read regulation 5 a second time, and notice exactly what it asks for. It asks for an outcome: it forbids the result, an unsafe product placed on the market, and says nothing whatever about the input. Nowhere does it tell a producer to open the sack, inspect the raw material, or sample the mineral. The words "asbestos," "raw material," "ingredient," and "component" appear nowhere in it.² The ban prohibits the outcome; regulation 5 attaches liability once the outcome has already happened. Both bite an importer. Neither, anywhere, requires that importer to look at the crushed rock before it ships.

That gap is the whole of it, and it must be stated at exactly its size. A duty not to sell it if it is there sits beside no duty to find out whether it is there. The law forbids the destination and leaves the journey unwatched. A company that has never tested a grain of its sand, and imports a batch that happens to be clean, has broken nothing. A company that imports a batch that happens to carry tremolite has broken the law. The breach is real only in retrospect, after someone else finds the fibre, after the law's liability has landed on a harmless-looking bag already sitting in a home. The duty is genuine; it simply activates too late to be a watchman. The duty is a verdict, not a gate.

So walk back up the path and count. The test that runs is built for a different hazard and cannot see the fibre. A test that can see the fibre exists, written down, and is required of no one. And the general duty forbids the bad result while never asking anyone to check the input. Three checkpoints, a chair at each of them, every one empty, and not one empty because a particular person walked off the job. They were built empty. Which returns the question the machine raised at the start. If every post inside the system was unmanned, what was the accidental Australian detection standing in for? And who, if anyone, was ever going to look?

The only people who looked did it from outside, or by luck, or too late

Someone did look. Precisely who is worth pinning down, because the pattern in the answer is the point.

An Australian laboratory looked, by accident, while testing a machine. That accident was the trigger for everything downstream. Once the first result surfaced, authorities did respond, and the response was real. British regulators issued recalls,⁵ and France's consumer-protection directorate told businesses to suspend sale of the affected sand toys.¹¹ Neither response was accidental, though both were set off by the one lucky machine. A system that only begins looking after chance hands it a case does not look; it reacts, and the reaction cannot start until the accident happens.

A researcher looked, from a university, after the products had already shipped. In mid-2026, Associate Professor Terri-Ann Berry and colleagues at Auckland University of Technology ran what they described as a world-first study. They put recalled craft sand through simulated play and sampled the air. They found that tremolite and chrysotile fibres "were released into the air during simulated play activities" from some of the sands they tested.⁹ Their own caveats travel with that finding and must not be stripped from it: the study's authors caution that "absences of asbestos detection cannot be taken as proof of safety," and that "the presence of asbestos fibres in a sample cannot be generalised to other sand samples or conditions."⁹ The study is not a harm claim, and it does not touch the official low finding, which stands. It says one narrow thing: the assumption that these products are inert in ordinary handling had, until a university chose to test it, gone untested. That assumption is the physical premise buried inside "low in normal use." The people who checked whether the sand behaves as advertised were volunteers, working outside any duty.

And someone looked twenty-five years ago, and the dance was identical. In 2000, the United States Consumer Product Safety Commission found a "trace amount of asbestos" in crayons, a mined mineral, talc, milled into a children's product. It judged "the risk of exposure to the fibers... extremely low" and did not order a recall. "As a precaution, because crayons are intended for use by children," it "asked industry to reformulate." Its chairman at the time, Ann Brown, put the whole posture in a sentence: "The risk is low but the concerns with these fibers should not be ignored."¹⁰ Low risk, voluntary fix, no new duty to screen the mineral before the next batch. A generation later, the sand ran the same course, because nothing between then and now had turned "should not be ignored" into "must be looked for."

Every one of those looks came from outside the machinery, or from luck, or from a quarter-century's distance. Not one of them came from a post that anybody was standing at. Which is what a parent is actually left holding.

The recall list is the only check you have, and it is built to arrive late

What you can do is smaller than it should be, but it is real, and it is worth doing precisely.

On 21 July 2026, the OPSS product-safety register recorded a run of recalls of sand-filled toys (Wizz Toys stretch figures, squishy sand animals, stress toys) brought to market by the importer PMS International. Each notice uses the regulator's own careful wording: the product "may present a risk to health as the sand inside the toy may be contaminated with a small quantity of asbestos." Each was "withdrawn from the market and recalled from end users by PMS International."⁵ Every claim there is the regulator's, on the public record, and it describes a company doing what the law requires of it once a problem is found: complying. Here the system's back end is working.

The recall list is also, by its nature, the only public instrument a parent has, and reading it honestly means reading what it cannot do. The list is reactive: it records what was caught, and the catching, at the origin, was an accident. Tucked inside the government's own joint statement is the tell: "No play pit sand is currently subject to a recall."⁴ That sentence is not the same as "it has been checked and is clear." It means only that no such sand has yet been caught. And catching, so far, has depended on a machine that was not even watching for it. Absence from the recall list is not a clean bill of health. It is silence.

So, concretely: check the OPSS product recall list on gov.uk before you buy or keep a sand-filled toy or craft kit; it is the only public record of which products have been pulled, and it is free. If something you own is on it, the single most important thing is what not to do. Do not tip it out, pour it, vacuum it, or otherwise disturb the sand, because disturbance is the exact mechanism that lifts fibres into the air. Follow the notice's return or disposal instructions instead.⁴ Treat any "sensory" or "craft" sand as what it is: an unscreened, mined, crushed-rock filler, rather than an inert craft supply. No rule requires it to be screened before it is sold. And if you want a sensory bin without the mineral question at all, the ordinary kitchen alternatives carry none of it. Dry rice, dried pasta, or dried beans, dyed at home, are crushed nothing and quarried nowhere. They are messier and less photogenic. They are also not rock.

One thing is worth asking a seller for, and it is the honest edge of this whole account. Ask whether the sand has been screened for asbestos by electron microscopy, by SEM-EDX or TEM. Those are the methods the government itself recommends, and the only ones that can actually see the fibre.¹ Be ready for the answer to be no, or a blank, because almost nothing on the shelf will have been. That blank is the gap in a single question. The screen is not exotic; UK-accredited laboratories publish it as a routine, priced laboratory service that starts at around £30 a sample.¹³ The reason it is rarely run is neither cost nor difficulty. No one is required to run it, the same reason the identical screen is mandatory for pharmaceutical talc, where liability, not a different hazard, long ago forced the industry's hand.

The strongest version of the other side deserves to be put plainly, because it is not foolish. A thoughtful regulator would say this: product safety cannot mean testing every mined input for every naturally occurring contaminant, an infinite and paralysing exercise. The law is built on a safe-outcome duty backed by a substance ban, rather than a schedule of mandatory input screens. When a real contamination surfaced, that architecture did what it is designed to do. Officials assessed the risk, judged it low, and stated so openly. A recall followed, the responsible business complied, and the regulator went further by writing down a method so industry can check. That case is real, and this account concedes it. What it does not concede is the word "self-correcting." A design that only starts looking once luck hands it a case has not corrected itself. The state has now written down precisely how to look, and still required no one to do it. It has described its own blind spot in careful prose and left it exactly where it was. The gap is not a scandal of poisoned children; the risk is low, and that holds. It is a gap nobody chose, examined, or is accountable for: a gap by default.

The thing that would close it is small and specific, which is what makes its absence so loud. Turn one word in the March 2026 advisory, "recommended" into "required" for mined mineral fillers in children's products, and the third empty chair has a person in it. That single change is the evidence that would overturn this reading: a screening duty, written down, that bites before the sand ships instead of after a machine finds the fibre by chance. Until that word changes, the coloured sand in a child's craft toy remains what it has been all along — a crushed rock that can foreseeably carry a banned mineral, tested by no one who was required to test it. Its one true moment of being watched came from a machine that was checking itself.