I've spent a lot of years on the emergency response side of this industry, and if there's one number that gives workers false confidence faster than any other, it's the oxygen reading on a four-gas monitor.
You turn on the meter, it warms up, and there it is: 20.9%. Green light. Everybody relaxes. The hole watch signs the permit, the crew climbs in, and nobody thinks about that number again.
That reaction is exactly backwards, and I want to walk you through why.
Why oxygen is the first reading — and what it actually tells you
Oxygen is the first thing you test in a confined space, and there are two good reasons for that.
The first is obvious: you can't breathe without it. Normal air is 20.9% oxygen, and in British Columbia anything below 19.5% is legally an oxygen-deficient atmosphere. Below that line, judgment and coordination start to slip before you feel anything wrong — which is a terrible combination when you're twenty feet down a manhole.
The second reason is one a lot of people miss: your other sensors depend on it. The combustible gas sensor in most monitors is a catalytic bead, and it needs adequate oxygen to burn the sample and give you an honest LEL reading. In an oxygen-deficient space, your flammable gas reading can be dangerously low — not because the gas isn't there, but because the sensor can't see it properly. So oxygen isn't just the first reading. It's the reading that tells you whether you can trust everything else on the screen.
The math nobody does: 1.4% is 14,000 ppm
Here's the part I want you to sit with.
The "acceptable" oxygen range everyone quotes — 19.5% to 20.9% — sounds narrow. It's a 1.4 percentage point window. Barely a flicker on the display.
Now convert it. One percent of an atmosphere is 10,000 parts per million. That means the gap between "perfectly normal air" and "legally oxygen deficient" is 14,000 ppm. Fourteen thousand parts per million of something has to be going on in that space before your oxygen sensor even reaches its alarm point.
And it's actually worse than that. When a foreign gas leaks into a space, it doesn't just push out the oxygen — it dilutes the entire atmosphere, and oxygen is only about a fifth of it. Run the numbers and you need roughly 67,000 ppm of a displacing gas — nitrogen, argon, carbon dioxide, whatever — to drag your oxygen reading from 20.9% down to 19.5%.
Now compare that to what actually kills people in confined spaces:
- Hydrogen sulphide is immediately dangerous to life or health (IDLH) at 100 ppm.
- Carbon monoxide is IDLH at 1,200 ppm.
Read those numbers again next to 67,000. A space can hold ten times the lethal concentration of H2S, and your oxygen sensor will not move. Not a tenth of a percent. The display will say 20.9%, the green light will stay on, and the atmosphere will still drop you before you reach the ladder.
That's the whole point: the oxygen sensor is a bulldozer, and toxic gases are needles. It measures in thousands of ppm. H2S kills in tens.
A normal oxygen reading clears exactly one hazard
So what does 20.9% actually tell you? It tells you the space isn't oxygen deficient or oxygen enriched at the location and height you sampled. That's it. It does not tell you:
- That there's no H2S pooling at the bottom (it's heavier than air — it settles)
- That there's no CO mixed evenly through the space (it's nearly the same density as air — it goes everywhere)
- That there's no flammable gas sitting near the top
- That conditions won't change the moment someone starts welding, pumping, or disturbing sludge
This is why proper pre-entry testing means all four readings — oxygen, LEL, H2S, CO — taken at multiple levels, top to bottom, before anyone breaks the plane of the opening. And it's why BC treats a confined space with a normal oxygen reading and untested toxics as an unknown atmosphere, not a safe one. An atmosphere with an unknown concentration of a dangerous substance is treated as IDLH until you prove otherwise. The burden of proof is on the tester, not the space.
What this means for your permit
If your entry permit has one line for "O2 %" and a checkbox, it's not protecting your crew — it's documenting your assumptions. A proper permit forces the sequence: test oxygen first, confirm the meter can be trusted, then test flammables and toxics at every level, then record it all with times and locations before a single boot goes in.
I've seen what happens when that sequence gets skipped. The rescuer who goes in after a downed worker becomes the second patient more often than anyone wants to admit, and in almost every one of those incidents, somebody looked at an oxygen reading and called it good.
20.9% is where the testing starts. It's never where it ends.