Why this matters before you keep reading
You're probably reading this near a bed. Maybe on a phone, in bed. Maybe with a child asleep in the next room. Maybe you've been thinking about replacing the mattress under you and don't know which questions to ask first.
The thing you sleep on, by the standards of human history, is very new.
For most of the time anatomically modern humans have existed — call it 200,000 years — the materials we slept on were grass, leaves, wool, feathers, straw, and animal hair. The chemistry was about as complicated as the chemistry of a wool sweater. The mattress as a complex synthetic chemical product is only about 60 years old. The flame retardants, the foam off-gassing, the microplastics, the adhesive emissions, the phthalates measured in toddler urine, the brominated compounds showing up in breast milk — none of that existed in human sleep environments before roughly 1960.
This piece is the foundational document behind everything else we publish at Embr Sleep. It's the answer to a question wellness-skeptical readers reasonably ask: if mattress chemistry is so concerning, why have we been doing it for so long? The honest answer is — we haven't. We've been doing it for sixty years out of two hundred thousand. The problem is recent. The regulation that drove it has come under serious scientific challenge. The economics that keep it in place are well-understood. And the industry is now — slowly, partially, unevenly — beginning to walk it back.
Whether you're a parent thinking about a crib mattress, a firefighter dealing with occupational chemistry exposures, someone with chemical sensitivities trying to figure out which brand to trust, or just a person who wants to sleep well in a bedroom you didn't design the chemistry of — the story below is yours.
It starts in a cave in South Africa, 200,000 years ago, with people who were already thinking about how they slept.
- The oldest known constructed human bedding is at least 200,000 years old, found at Border Cave in South Africa.
- By 77,000 years ago, people were deliberately adding insect-repellent plant leaves to their sleeping surfaces.
- Polyurethane foam — the material in most modern mattresses — was first commercialized in 1954.
- California's flame retardant standard for furniture (TB117) took effect in 1975 and became the de facto standard across North America.
- That standard was never shown, in peer-reviewed analysis, to produce a measurable fire safety benefit.
Grass bedding, deliberately constructed
The oldest known physical evidence of constructed bedding comes from Border Cave in South Africa. In 2020, a team led by Lyn Wadley at the University of the Witwatersrand published findings in Science showing sheaves of broad-leafed Panicoideae grasses laid down as sleeping and working surfaces, on top of ash layers that appear to have been deliberate site maintenance — periodically burning old bedding to manage pests and replace it with fresh material. The age of the deposits places this practice at at least 200,000 years before present — meaning anatomically modern humans, possibly their immediate ancestors, were already constructing sleep surfaces with chosen materials and a maintenance protocol. Peer-reviewed
A more famous find — and a more sophisticated one — comes from Sibudu Cave, also in South Africa, dated to about 77,000 years ago. At Sibudu the sleeping surface was sedge leaves and stems laid directly on the shelter floor, and crucially, the top layer was leaves from Cryptocarya woodii — the river wild quince — which contain natural insecticidal and larvicidal compounds. The choice was deliberate. People weren't just sleeping on what was available; they were selecting a top layer for pest control. A 2017 follow-up study by Lennox and colleagues found that hearths in the same site also contained charcoal from Tarchonanthus parvicapitulatus, an aromatic shrub still used in modern African camping for the same purpose.
Two things stand out. First, the earliest constructed sleep surfaces were already engineered systems — substrate plus comfort layer plus pest-control layer, with site maintenance built in. Second, the chemistry of the pest control was plant terpenes and resins — molecules that volatilize, perform their function, and degrade. Nothing accumulated. Nothing persisted. The end-of-life disposal was burning the old bedding, which is exactly what the Sibudu micromorphology shows happened on a routine cycle.
We're not the first species to figure out comfortable sleep. But we may be the first to design sleep surfaces that come with side effects. Peer-reviewed
Linen, palm, wool, feathers
By the time the first written records appear, the mattress had become a constructed textile object. Well-preserved Egyptian beds from royal tombs used wooden frames strung with woven palm fronds or rope mesh, topped with linen-covered cushions stuffed with wool, feathers, or straw. The Greeks adopted similar construction with leather webbing; the Romans elaborated it with bronze frames, down pillows for the wealthy, wool or straw stuffing for everyone else.
The chemistry of all of these was simple to the point of being almost unrelated to chemistry: cellulose (linen, cotton, palm fibre, straw, hay), keratin (wool, feathers, down), and collagen (leather webbing). Three classes of biopolymer, all part of the evolutionary environment of the body sleeping on them.
Three properties of this construction are worth noticing. It was breathable: linen and cotton ticking let air pass through; wool wicks moisture and regulates temperature. It was maintainable: the stuffing was removed and replaced as it compressed. And it was fire-resistant by the chemistry of the materials themselves: wool fibre has a limiting oxygen index of roughly 25% — well above the 21% oxygen in normal air — meaning it doesn't readily ignite or sustain a flame. It chars rather than melts, doesn't drip molten material, and self-extinguishes when the flame source is removed (Cardamone 2013; Guo et al. 2022). Peer-reviewed
The horsehair era
By the early 1700s a new premium construction had emerged in Western Europe: the horsehair mattress. Hand-tufted, edge-stitched, layered: a base of palmetto fibre or hemp for support, a thick middle of horsehair for springiness, a top layer of cotton or wool batting for comfort, all wrapped in heavy linen or cotton ticking and finished with hand-tied tufts.
Horsehair was — and remains — a remarkable material. It's springy. It's breathable. It wicks moisture. It chars rather than ignites. And it's durable on a timeline that's hard to comprehend now. A well-made horsehair mattress from 1900 can still be in service in 2026, with periodic re-tufting and re-covering. Two of the European companies still making horsehair mattresses today — Hästens in Sweden and Vispring in England, both founded in the 19th century — are over a century old in part because their construction outlasts the people who buy it.
Two other facts about this period reframe what the current mattress industry calls "innovation." The first innerspring mattress was invented in the 1860s by Heinrich Westphal in Germany — and he died in poverty without commercializing it. Natural latex foam was invented by Dunlop in England by 1929. The chemistry of a modern organic latex mattress is essentially identical to a Dunlop latex mattress of the 1930s: natural rubber, sulfur cure, soap as a foaming surfactant. The most "advanced" non-toxic mattress option available in 2026 uses 90-year-old chemistry.
The chemistry shift begins
Polyurethane was invented in 1937 by Otto Bayer at IG Farben in Germany. According to a 2023 review in Polymers, flexible polyurethane foam was first commercially introduced in 1954 (Cherednichenko et al. 2023). That's the moment the chemistry shifts. For the first time in 200,000 years of human history, the material people lay on while sleeping was a synthetic polymer made by reacting an isocyanate with a polyol — a chemistry that did not exist on Earth until the 1930s.
Polyurethane foam had one immediate, unavoidable problem: it's extremely flammable. Flexible polyurethane foam has a limiting oxygen index of only about 14–18% — meaning it ignites and sustains burning in normal atmospheric air. Once ignited, it produces large volumes of toxic smoke containing hydrogen cyanide, carbon monoxide, isocyanates, and aromatic hydrocarbons.
The industry response was not to reconsider the material. The response was to add chemicals that would slow the burn rate. According to a 2024 review in Journal of Fire Sciences, the first commercial flame retardants for polyurethane foam were developed in the early 1960s — about 60 years before the present (Levchik 2024). Two technologies converging at the same moment: a flammable synthetic foam, and a class of additive chemicals designed to suppress its ignition. The combination would become the standard mattress construction for the next sixty years. Peer-reviewed
The regulation that decided the chemistry
In 1972, California passed legislation requiring all upholstered furniture sold in the state to be "flame retardant." Implementation followed in 1975 with California Technical Bulletin 117 (TB117) — a mandatory standard requiring furniture components to pass small-scale flame and smoulder ignition tests (Damant 1995).
The fire-policy motivation was real. Smoking-material ignition of upholstered furniture was causing roughly 1,500 civilian deaths per year in the United States in the early 1970s. California's standard was an attempt to prevent that.
The unintended consequence — which would not become clear for decades — was that California's standard, in a market the size of California's, became the de facto national standard. A Duke University team led by Heather Stapleton documented that 85% of US residential couches contained chemical flame retardants at concentrations greater than 1% by weight (Stapleton et al. 2012). The same pattern played out in children's products. A 2011 Environmental Science & Technology study found chemical flame retardants in 80 of 101 baby products tested — nursing pillows, baby loungers, changing pads, car seats, sleep positioners (Stapleton et al. 2011). Peer-reviewed
The replacement cycle: how the industry restructured around foam
The chemistry shift isn't the whole story. The other half is the business model that grew up around the new materials.
A well-made horsehair mattress from 1900 could last fifty years. A natural latex Dunlop mattress from the 1930s often lasted twenty-five to forty years. Mattresses were built to be repaired: the ticking was opened, the stuffing was refreshed, and the mattress went back into service. Sleep surfaces were inherited.
That entire economic model collapsed when foam took over. Polyurethane foam, once compressed and cycled for ten thousand nights of body weight, body heat, and body moisture, doesn't recover. It permanently flattens. There's no way to "refurbish" a foam mattress because there's nothing to refurbish — you can't refluff polyurethane the way you can refluff wool. The product is structurally disposable.
This changed the industry. Mass-market foam mattresses are typically sold with ten-year warranties, but the Better Sleep Council — itself a mattress industry trade association — recommends replacement every seven to ten years. Most consumers replace closer to seven. That's a turnover cycle three to seven times faster than the horsehair construction it replaced. Faster turnover means more units sold per customer over a lifetime, which means a much larger consumer industry than the durable-goods market that came before.
The chemistry problem and the durability problem are the same problem with two faces. You can't have the cheap, mass-market, accessible mattress industry without the foam. You can't have the foam without the flame retardants. You can't have the seven-year replacement cycle without the durability collapse. The whole structure holds together as a system, and the system has no internal incentive to change. Inferred
The chemistry comes due
What changed after 2000 wasn't the mattress. It was the science.
Biomonitoring studies — measuring chemical concentrations in human bodies — started finding pentabromodiphenyl ether (penta-BDE), the dominant flame retardant of the 1980s and 1990s, in breast milk, blood, and adipose tissue at levels that were doubling every five years. In 2005, the manufacturer voluntarily withdrew penta-BDE from the US market. Industry pivoted to chlorinated organophosphate alternatives — TDCIPP, TCEP, TCIPP — and by 2010, those compounds dominated foam flame retardant chemistry (Stapleton 2012; Cooper et al. 2016).
The replacement chemistry turned out to have its own problems. TDCIPP went onto the California Proposition 65 list as a carcinogen in 2011. TCEP was added in 2014. A 2024 review in Environmental Research synthesized 48 studies and found associations with pregnancy loss, altered gestational duration, lower birthweight, thyroid dysfunction in both mothers and newborns, child metabolic dysregulation, impaired neurodevelopment, and altered immune response (Shahin et al. 2024).
And in 2025, a team at the University of Toronto led by Miriam Diamond ran the test that should have been run decades ago: they took 16 new children's mattresses available on the Canadian market, measured the semi-volatile organic compounds emitted at room temperature and at body temperature with body weight applied, and published the results in Environmental Science & Technology (Vaezafshar et al. 2025). They detected 21 semi-volatile organic compounds across four chemical classes. One mattress exceeded Canada's regulatory limit for di-n-butyl phthalate. Five had concentrations above 0.1% for phthalates regulated in children's toys but not in mattresses. One mattress contained TCEP — a flame retardant whose use has been prohibited in Canada since 2014. Peer-reviewed
In 2013, California revised TB117 to a smoulder-resistance standard (TB117-2013) that can be met without flame retardant chemicals. In 2020, AB 2998 prohibited flame retardants above 1000 mg/kg in juvenile products, upholstered furniture, and mattresses. A 2024 study by the Green Science Policy Institute found these regulations worked: flame retardant levels in California furniture and children's products dropped sharply (Gill et al. 2024).
The fix worked. But it only worked for new purchases in California, after 2014, in households that could afford to replace their furniture. Everywhere else, and in every older piece of furniture, the chemistry from the previous fifty years remains.
And here's the part that's hardest to absorb. A 2011 review by Vyto Babrauskas and colleagues — Babrauskas was one of the leading fire scientists in the United States, formerly at the National Institute of Standards and Technology — concluded that TB117 had never been shown to produce a measurable fire safety benefit (Babrauskas et al. 2011). The regulation that drove fifty years of flame retardant chemistry into mattresses, sofas, baby products, and household dust did not, in the peer-reviewed analysis, save lives in proportion to the chemistry it deposited into human bodies. The fix for cigarette-ignition fires turned out to be reduced-ignition-propensity cigarettes — a tobacco-product regulation, not a furniture regulation (Gann et al. 2020). Peer-reviewed
What this history actually means
The historical arc is short, and the moral isn't subtle.
For 200,000 years, humans slept on plant fibre, animal hair, feathers, and wool. The construction was breathable, maintainable, compostable, and chemically simple to the point of being almost invisible to the body. Mattresses lasted decades. They were repaired, refilled, and inherited.
For roughly 60 years, that's no longer been true. The mattress became a synthetic foam, treated with synthetic chemistry, sold on accelerated replacement cycles, and disposed of into landfills where it persists for decades. The chemistry problem the rest of our research describes — flame retardants in body fluids, microplastics in brain tissue, phthalates in toddler urine, foam VOCs in bedroom air — is not a permanent feature of human sleep. It's a feature of a particular six-decade industrial choice that the industry has no internal incentive to reverse.
The strongest version of the Embr Sleep position is this: we are working in an interim era. The historical default was clean materials and durable construction. The recent default has been synthetic materials, chemical management of the safety problems those materials created, and accelerated replacement cycles that make the whole system financially viable for everyone except the customer. The future default — and we'll know in fifteen or twenty years whether the industry actually gets there — will probably be a return to the historical pattern, with modern improvements: organic latex with mechanical bonding, wool fire barriers, organic cotton covers, no adhesives, certifications that actually mean something, and replacement cycles measured in decades rather than years.
Embr Sleep is developing tools designed to address the chemistry gap that exists right now — for the people sleeping tonight on the mattress they already own, in a bedroom that contains the chemistry of the last sixty years of industrial choices. That's a response to a manufactured problem, not a permanent feature of sleep. Inferred
We think the right way to talk about this is honestly: there was nothing wrong with how humans slept for most of the last 200,000 years. The problem is recent. The fix is partly behavioural (buy better when you can, ventilate, wash bedding), partly regulatory (TB117-2013, California AB 2998, Canada's CEPA flame retardant group, the EU's REACH SVHC list), and partly engineered.
But the deepest fix is the historical one. The mattress doesn't need new chemistry. It needs less of it.
Frequently asked questions
When did mattresses start containing chemicals?
The chemistry shift began in 1954, when flexible polyurethane foam was first commercialized as a mattress material. Until then, mattresses were made of natural fibres — cotton, wool, horsehair, feathers, straw, or natural latex — that contained essentially no synthetic chemistry. Flame retardants began appearing in foam mattresses in the early 1960s, accelerated dramatically after California's TB117 standard took effect in 1975, and dominated US mattress construction from roughly 1980 to 2014.
What did people sleep on before modern mattresses?
For most of human history, sleep surfaces were grass, leaves, animal skins, straw, hay, wool, feathers, or natural latex — sometimes layered, sometimes loose, but always natural fibre. Wealthy households in 18th- and 19th-century Europe slept on horsehair mattresses with cotton or wool batting tops, often hand-tufted in linen ticking that could last decades with re-stuffing and re-covering.
Why don't modern mattresses last as long as old ones?
Polyurethane foam compresses permanently under the cycles of body weight, body heat, and body moisture. There's no way to refurbish a foam mattress because there's nothing to refurbish — you can't refluff polyurethane the way you can refluff wool batting or horsehair. The product is structurally disposable. A horsehair mattress from 1900 can still be in service today; a foam mattress from 2018 is usually due for replacement by 2025. The shorter lifespan isn't an accident — it's what makes the modern mass-market mattress industry economically possible.
How long does a mattress last?
A modern foam or hybrid mattress typically lasts seven to ten years before noticeable sag, compression, and material breakdown make replacement worth considering. A historical horsehair mattress could last fifty years or more with periodic re-tufting and re-covering. A modern certified-organic natural latex mattress falls somewhere in between: typically fifteen to twenty-five years if maintained well.
Was sleep actually better 200 years ago than it is now?
No, and we're not arguing that. Pre-industrial sleep had real problems: vermin, dampness, inconsistent temperature, lice, bedbugs, infectious disease transmitted on shared sleep surfaces. The historical construction was clean in chemistry terms; it was not clean in biology terms. Modern sleep is on net safer in most ways that aren't chemistry. The argument of this piece is specifically about chemistry and durability, not about whether the 1700s were preferable to the 2020s.
Why are flame retardants used in mattresses?
Modern polyurethane foam is extremely flammable. Its limiting oxygen index is roughly 14–18%, meaning it ignites and burns readily in normal air. Once ignited, foam burns fast and produces large volumes of toxic smoke including hydrogen cyanide and carbon monoxide. Starting in the 1960s, the industry added chemical flame retardants to slow ignition rather than reconsider the material itself. California's 1975 TB117 standard effectively mandated these chemicals across the US market for the next four decades.
Is California TB117 still in effect?
The original TB117 is no longer in force. California revised the standard to TB117-2013, which can be met without flame retardant chemicals. In 2020 California went further with AB 2998, prohibiting flame retardants above 1000 mg/kg in juvenile products, upholstered furniture, and mattresses. But mattresses and furniture made before 2014 still contain the original chemistry, and that legacy material is still in many homes.
What should a person sleeping on a foam mattress right now do?
Realistically: don't panic and don't replace it before you can afford to. Ventilate the bedroom well, wash bedding in hot water weekly, vacuum and dust-wipe floors regularly, and when replacement time comes, choose a certified-organic option or at minimum one with full material disclosure. See our off-gassing guide and our memory foam chemistry piece for practical next steps.
Selected citations
- Wadley L et al. (2020). "Fire and grass-bedding construction 200 thousand years ago at Border Cave, South Africa." Science.
- Wadley L. (2012). "Two 'moments in time' during Middle Stone Age occupations of Sibudu, South Africa." Southern African Humanities.
- Lennox S et al. (2017). Charcoal from Tarchonanthus at Sibudu Cave. Quaternary International.
- Cardamone JM. (2013). "Flame resistant wool and wool blends."
- Guo S et al. (2022). "Efficient suppression of flammability in wool fabrics via chelation with ferric ions." Applied Surface Science.
- Cherednichenko KA et al. (2023). "Polyurethane Foam Synthesis: An Overview." Polymers.
- Shen J et al. (2024). "Recent advances in flame retardancy of polyurethane foam." European Polymer Journal.
- Levchik SV. (2024). "Polyurethane flame retardants: a review." Journal of Fire Sciences.
- Damant GH. (1995). "Cigarette ignition of upholstered furniture: TB117 history." Journal of Fire Sciences.
- Damant GH & McCormack JA. (1983). "The role of cigarettes in upholstered-furniture fires." Journal of Fire Sciences.
- Babrauskas V et al. (2011). "Flame retardants in furniture foam: benefits and risks." Fire Safety Science.
- Stapleton HM et al. (2011). "Identification of flame retardants in polyurethane foam collected from baby products." Environmental Science & Technology.
- Stapleton HM et al. (2012). "Novel and high volume use flame retardants in US couches." Environmental Science & Technology.
- Cooper EM et al. (2016). "Identification of organophosphate flame retardants in polyurethane foam recovered from couches." Environmental Science & Technology.
- Shahin S et al. (2024). "Organophosphate ester flame retardants and maternal-child health: systematic review." Environmental Research.
- Vaezafshar S et al. (2025). "Are Sleeping Children Exposed to Plasticizers, Flame Retardants, and UV-Filters from Their Mattresses?" Environmental Science & Technology 59(16): 7909–7918.
- Gill B et al. (2024). "Flame retardant chemical decreases in California furniture and children's products." Chemosphere.
- Rodgers K et al. (2021). "Decreases in flame retardant dust concentrations in California homes." Environmental International.
- Gann RG et al. (2020). "Reduced-ignition-propensity cigarettes and fire fatalities." Fire and Materials.
- Attfield KR et al. (2025). "Biomarker decline post TB117-2013 furniture replacement." Environmental Pollution.