2.3 Testing: instruments, response time and the order
A gas meter is not a smoke detector. It reads the air that reaches its sensors, at the moment it reaches them, in the order the sensors can cope with. Almost every bad confined space reading is a good instrument being asked the wrong question — at the wrong depth, for too short a time, or in the wrong sequence.
1 The order, and the reason for it
29 CFR 1910.146(d)(5)(iii) is one sentence: when testing for atmospheric hazards, test first for oxygen, then for combustible gases and vapors, and then for toxic gases and vapors. 1926.1204(e)(3) repeats it word for word for construction, and the ventilation-only paragraphs — 1910.146(c)(5)(ii)(C) and 1926.1203(e)(2)(iii) — add that the instrument must be a calibrated direct-reading instrument.
Appendix B(5) to 1910.146 explains why the order is that order, and it is worth knowing rather than memorizing:
- 1Oxygen first — “because most combustible gas meters are oxygen dependent and will not provide reliable readings in an oxygen deficient atmosphere.” A catalytic-bead sensor burns the sample on a hot bead; with too little oxygen, there is nothing to burn and the reading falls, not rises.
- 2Combustibles next — “because the threat of fire or explosion is both more immediate and more life threatening, in most cases, than exposure to toxic gasses and vapors.”
- 3Toxics last — “if tests for toxic gasses and vapors are necessary, they are performed last.”
On a modern four-gas meter all four sensors read at once, so the order looks like a formality. It is not. The order tells you which reading invalidates the others: an oxygen reading below the acceptable range makes the %LFL reading on the same screen untrustworthy, and a crew that does not know that will read 2 percent LFL in a nitrogen-blanketed vessel and think the space is lean.
2 Evaluation testing and verification testing are two different jobs
Appendix B opens by saying atmospheric testing is required for two distinct purposes, and separating them clears up a lot of confusion about who is qualified to do what.
| Evaluation testing | Verification testing | |
|---|---|---|
| Question it answers | What hazards can exist or arise in this space, and what should the acceptable entry conditions be? | Are the acceptable entry conditions present right now? |
| Equipment | Sufficient sensitivity and specificity to identify and evaluate any hazardous atmospheres that may exist or arise | The equipment the permit specifies |
| Who | Done by, or reviewed by, a technically qualified professional — the appendix names the OSHA consultation service, a certified industrial hygienist, a registered safety engineer, a certified safety professional or a certified marine chemist as examples | The person designated to test or monitor, trained under 1910.146(g) |
| Output | The entry procedure and the stipulated acceptable entry conditions for that space | Results recorded on the permit “in the space provided adjacent to the stipulated acceptable entry condition” |
That last cell is a small piece of design advice with real force: the reading belongs next to the limit it is being compared against, so that nobody has to remember what good looks like. It is why the atmospheric log in the SteelToeTools generator prints the acceptable entry conditions above the readings and flags each row against them.
3 How long you hold it there
Appendix B(3), Duration of testing: measurement of values for each atmospheric parameter should be made for at least the minimum response time of the test instrument specified by the manufacturer. That is the whole requirement, and it is the one most often broken by people doing everything else right.
Response time is not a fixed number. It depends on the sensor, and it grows with everything between the atmosphere and the sensor: a sampling probe, a length of tubing, a hand aspirator, a pump that has to draw the sample the length of the hose before the sensor sees anything. A meter with a 30-second response time and 10 feet of sample line takes considerably longer than 30 seconds to tell you the truth. Glance at the display for two seconds and you have measured the air in the tube.
4 Stratified atmospheres: a reading at the rim describes the rim
Gases sort themselves by density. Methane rises and collects under the roof; hydrogen sulfide, propane, gasoline vapor, argon and carbon dioxide sink and pool at the bottom. Air trapped between them can be perfectly normal. So a single reading taken through the open hatch is a statement about one horizontal slice of the space and nothing else.
Appendix B(4) gives the procedure: when monitoring for entries involving a descent into atmospheres that may be stratified, the atmospheric envelope should be tested a distance of approximately 4 feet (1.22 m) in the direction of travel and to each side. And if a sampling probe is used, the entrant's rate of progress should be slowed to accommodate the sampling speed and detector response.
Read that as a description of how a descent actually happens: test four feet ahead and four feet to each side, move, wait for the instrument to catch up, test again. A ladder descent into a 20-foot manhole is therefore five or six tests, not one. The crew that takes one reading at the top and climbs down at normal speed has tested the atmosphere they were already standing in.
5 Calibration, bump testing and the zero
The standard's word is calibrated, and it appears in both ventilation-only paragraphs. Three distinct operations sit behind it, and they are not interchangeable:
- Calibration — adjusting the instrument against a certified gas of known concentration, on the manufacturer's schedule. It is a maintenance operation with a record.
- Bump test, or function check — applying a known gas before use to confirm that every sensor responds and every alarm activates. It proves the instrument still works today; it does not adjust anything.
- Fresh air zero — establishing the baseline in known-clean air, which means away from the space, not standing over the open hatch. Zero the instrument in the exhaust plume of your own blower and every subsequent reading is wrong by that amount, in the reassuring direction.
A meter that has never been bump tested is an object that displays numbers. The failure mode that matters is silent: an oxygen sensor at the end of its life, a combustible sensor poisoned by silicone or by a previous exposure to a high concentration, a filter blocked with sludge. None of those makes the screen blank; all of them make it read low.
6 Timing, and the entrant's rights
Two timing rules are easy to miss. 1910.146(d)(5)(i) requires the space to be tested to determine whether acceptable entry conditions exist before entry is authorized to begin. 1926.1204(e)(1) is tighter for construction: test conditions before changes to the space's natural ventilation are made, and before entry is authorized. In other words, on a construction site you read the space as you found it, before the blower starts — because a reading taken after ten minutes of fresh air tells you nothing about what the space does when the fan stops.
Then there are three rights that belong to the entrant, and they are duties on the employer, not courtesies:
| Right | General industry | Construction |
|---|---|---|
| Observe the pre-entry and any subsequent testing | 1910.146(d)(5)(iv) | 1926.1204(e)(4) |
| Have the space reevaluated in their presence when they have reason to believe the evaluation may not have been adequate | 1910.146(d)(5)(v) | 1926.1204(e)(5) |
| Be given the results of any testing immediately | 1910.146(d)(5)(vi) | 1926.1204(e)(6) |
The second one is the strongest sentence in the paragraph. A worker who says “I do not believe that reading” has, by regulation, triggered a retest in their presence. Not a discussion about it. And the person best placed to know that a reading is wrong is usually the one who has been in that space before.
- Oxygen, then combustibles, then toxics — 1910.146(d)(5)(iii), 1926.1204(e)(3). Oxygen goes first because most combustible sensors are oxygen dependent; a low oxygen reading invalidates the %LFL beside it.
- Appendix B separates evaluation testing (done or reviewed by a technically qualified professional, to set the acceptable entry conditions) from verification testing (are those conditions present now, recorded next to the limit).
- Hold each reading for at least the manufacturer's minimum response time — App. B(3) — and remember that a probe or sample line lengthens it.
- For a descent into a possibly stratified atmosphere, test about 4 ft in the direction of travel and to each side, and slow the descent to the sampling speed — App. B(4).
- Calibration, bump test and fresh-air zero are three different things; the sensor failures that matter are silent and read low.
- On construction work, test the space before changing its natural ventilation — 1926.1204(e)(1). And an entrant who doubts a reading is entitled to a reevaluation in their presence — (d)(5)(v) / (e)(5).
Free educational content — not OSHA-authorized training, no certificate or card issued. Follow your employer's program and the standards cited.