Bad Air: Why It's Invisible, and Why You Leave

Of every hazard covered on this site, bad air is the one where getting the arithmetic right matters least, and getting the response right matters most. You can't see it, smell it, or feel it clearly until it's already affecting the judgment you'd need to notice it — which is why the actual rule cavers use is short and doesn't depend on a number at all: if the air feels wrong, or someone's breathing or thinking seems off, you leave, immediately, and you don't try to calculate your way into staying. This article walks through what's actually happening when air goes bad underground, works a real example through the site's own Air & Ventilation Guide, and is honest about exactly where that tool's usefulness ends.

What's actually accumulating, and why it's invisible

Underground air quality problems are almost always about two linked things: carbon dioxide building up, and oxygen being displaced or consumed as it does. Both come from ordinary, unremarkable sources — decaying organic material, natural geological CO₂ seeps, simple human breathing in an enclosed space — and neither one announces itself the way a wet floor or a loose rock does. Elevated CO₂ is odourless. Reduced oxygen is odourless. The first reliable signal most people get is their own breathing working harder than the activity should explain, and by the time that's happening, the same gas is already starting to affect the clear thinking that would normally prompt someone to leave. That's the whole reason this hazard gets treated with more caution than its actual frequency would seem to justify: it's not that it's common, it's that it quietly disables your own warning system while it's happening.

Worked example: a small, still chamber

Take a genuinely plausible scenario: a low, roughly 30 cubic-metre chamber with essentially no air movement, a party of four resting there at light activity for two hours — waiting out a survey team ahead, say, or taking a break. Run those numbers through the site's airVentilation() function, the same one behind the calculator: at light activity, the standard per-person CO₂ production rate used by the tool is 0.0075 cubic metres per hour. Four people, over two hours, produce 4 × 2 × 0.0075 = 0.06 cubic metres of CO₂ in total. Spread across the 30 cubic-metre chamber and converted to parts per million, that adds roughly 2,000ppm on top of the tool's assumed 400ppm ambient background, for an estimated concentration of about 2,400ppm — which the tool labels "caution," its middle tier between "adequate" and "poor."

That's the entire calculation: a production rate, multiplied by people and time, divided by the space they're in. No part of it is complicated. What matters is what it does, and doesn't, tell you.

The part of the model that deliberately doesn't move

Here's the detail worth understanding rather than glossing over: run that identical scenario again, this time entering a healthy air speed through the passage — say 3 metres per minute instead of a dead-still zero — and the tool's estimated CO₂ concentration doesn't change at all. It's still around 2,400ppm, still "caution." That isn't a bug; it's a deliberate modelling choice, and the tool's own FAQ says so directly: the CO₂ estimate always models a worst-case scenario where the air in that specific chamber is completely still, regardless of what airflow figure you enter elsewhere in the form. The airflow inputs feed a separate, genuinely useful calculation — air changes per hour through the passage generally — but the CO₂ build-up figure is intentionally decoupled from it, because the point of a worst-case planning estimate is to flag "this is the kind of space where you'd want to check," not to quietly tell you a breeze fixes everything.

In other words: the number is conservative on purpose, and it's supposed to be. A real chamber with genuine airflow will very often clear CO₂ far faster than this static model assumes, exactly as the tool's own guidance says — but the tool has no way of confirming that's actually happening in your chamber, on your day, so it doesn't pretend to.

The other half of the tool: airflow, not build-up

The calculator actually runs two linked but separate calculations, and it's worth understanding the first one on its own terms rather than only through the CO₂ figure. Given a passage's cross-section — say 1.0m wide by 1.5m high, for a cross-sectional area of 1.5 square metres — and a measured or estimated air speed through it, the tool multiplies the two to get a volume of air moving per minute. At 3 metres per minute, that's 1.5 × 3 = 4.5 cubic metres per minute, or 270 cubic metres an hour. Divided by the 30 cubic-metre reference chamber from the earlier example, that's 9 air changes per hour — a genuinely useful, independent figure describing how thoroughly a space's air turns over, and one that comes from an actual air-speed reading (an anemometer, or a careful estimate) rather than an assumption.

That figure is real and useful on its own terms, for understanding a passage's general ventilation character. What it deliberately isn't wired into, as covered above, is the CO₂ build-up estimate for a specific chamber — the tool keeps the two calculations separate rather than letting a healthy air-changes figure quietly override a worst-case build-up warning it can't actually verify applies to that exact chamber.

What "adequate," "caution," and "poor" actually mean

It's worth being precise about what these three labels are and aren't. They describe where a modelled, worst-case-still estimate lands relative to widely used indoor air-quality reference points — nothing more. A result labelled "adequate" is not a safety certification and isn't the tool telling you a space is fine to occupy indefinitely; it means the stagnant-air model, for the inputs you gave it, didn't cross the tool's first reference threshold. A result labelled "poor" isn't a diagnosis of your actual air right now either; it means the modelled worst case crossed a higher one. None of the three labels know anything about the real airflow moving through your actual chamber today, whether the passage's biological activity has changed with the season, or how you or your party are actually feeling right now. Treating any of the three as a green or red light in place of a real reading is exactly the mistake this planning tool is built to help you avoid, not encourage.

What a real gas meter gives you that this can't

A personal CO₂ or oxygen meter, carried into a system with a known or suspected bad-air history, gives you something a planning estimate structurally cannot: an actual, on-site, real-time reading of the air you're standing in, right now, today. That's the tool that makes an actual go or no-go call underground. This calculator's honest job is entirely upstream of that moment — helping you think through, before you ever leave the surface, whether a planned stop in a small, poorly-known chamber is worth extra caution or a different plan — and it hands off completely to a real meter, and to your own and your party's judgment, the instant you're actually there.

Why the same trip can behave differently next time

One more reason a modelled estimate, or even last month's fine trip report, isn't a durable guarantee: the biological and geological sources that produce bad air aren't constant. Organic decay rates shift with temperature and moisture through the seasons. Flood events wash fresh vegetation into a system and can change a chamber's decay load overnight. Natural geological CO₂ seeps can vary with groundwater conditions. A chamber that was genuinely fine on every previous visit isn't a chamber that's provably fine on this one — it's a chamber worth the same care as an unfamiliar one, especially after unusual weather or a long gap since the last trip through. This is exactly the kind of local, current knowledge covered in more general terms in this site's broader article on foul air underground, which this piece deliberately narrows in on the calculator's own arithmetic instead of repeating.

The response doesn't need a number at all

Strip away every figure in this article and one instruction is left, and it's the only one that actually matters underground: unexplained heavy breathing, a headache that started down there, dizziness, or confusion is a leave-now signal, for whoever in the party is experiencing it, full stop — not a symptom to push through, not a cue to check a calculator, and not something to self-diagnose while the very thing being affected is your own judgment. Leave the area, move to good air, and if anyone still feels unwell once you're out, treat that as worth seeking medical help for rather than assuming it will pass on its own. Whatever a modelled estimate or even a real meter said an hour earlier doesn't override how a person actually feels right now.

The takeaway

The arithmetic behind a bad-air estimate is genuinely simple, and running your own trip's numbers through the Air & Ventilation Guide before you go is a worthwhile planning habit for a system with any known history. But the number it gives you back is a conversation-starter for extra caution, not a substitute for a real meter on site or for the one response that never needs a calculation behind it: if it feels wrong, you leave.