SOLAS Atmosphere Testing & Gas Meter Calibration Guide

Almost every enclosed space fatality at sea involves an atmosphere that somebody believed had been tested. The instrument was on board, it was switched on, it showed a reading, and the reading was wrong — because the sensor had drifted since its last calibration, because the sample line had not purged, because the meter was measuring at the hatch rather than at the level where the person was going, or because the gas that killed them was one the instrument did not have a channel for. The equipment failed nobody. The method did.

This guide is about the instrument and the technique: what a multi-gas meter must be able to measure, the difference between zeroing, bump testing and calibration, how to sample a space before anyone goes into it, and how to read the result honestly — including the gas most four-channel meters on board cannot see at all.

SOLAS chapter XI-1 · safety officer’s guide

The meter does not tell you the space is safe. It tells you what reached the sensor.

A clean screen on a four-gas meter is not a clean atmosphere. It is four gases, measured wherever the sample was drawn from, by sensors that were last proved on a date somebody should be able to tell you.

Book a free demo Start free trial

Atmosphere testing before entry: what the rules require you to measure

Since 2016, SOLAS chapter XI-1 regulation 7 has required ships to carry an appropriate portable atmosphere testing instrument, capable as a minimum of measuring oxygen, flammable gases or vapours, hydrogen sulphide and carbon monoxide, together with the means to calibrate it. That is the four-gas meter every ship now has. The revised recommendations for entering enclosed spaces go further and expect personal gas detection to cover carbon dioxide as well, and to keep working in an oxygen-depleted atmosphere.

Oxygen

The first reading and the one that decides whether anything else matters. Depletion can come from rust, from an inerted or purged space, from cargo consuming oxygen, or from a leaking fixed extinguishing system.

LFL

Flammable gases and vapours

Reported as a percentage of the lower flammable limit. Remember that 100 per cent LFL for methane is only about 5 per cent by volume — a small number on the screen is a large quantity of gas.

H2S

Hydrogen sulphide

Heavier than air, so it pools low in the space. It deadens the sense of smell quickly, which is why nobody should ever be relying on smell as a secondary check.

Carbon monoxide

Odourless and produced by anything that has burned or smouldered. Common after a fire, in spaces adjacent to exhaust runs, and around cargoes that self-heat.

CO2

Carbon dioxide

Not on a standard four-gas meter. Organic cargoes consume oxygen and give off carbon dioxide, so a small fall in oxygen can sit alongside a large and dangerous rise in CO2 that nothing on board is measuring.

The acceptance criteria every atmosphere test is judged against

The revised recommendations set numbers rather than adjectives, which is a considerable improvement on “safe for entry”. If any one of them is not met, the space is unsafe and nobody goes in.

20.9 %

Oxygen, at or above

< 0.5 %

Carbon dioxide (5,000 ppm)

< 1 %

Of the lower flammable limit

< 50 %

Of the occupational exposure limit, toxic gases

Two points about the oxygen figure catch people out. It is not “above 19.5 per cent”, which is the number many crews carry in their heads from older guidance — it is normal atmospheric oxygen, 20.9 per cent. And it is a ceiling as well as a floor in practice: a reading above 21 per cent means the space is oxygen-enriched, which is its own serious hazard and usually means a leaking oxygen line rather than a good atmosphere.

Gas meter calibration, bump testing and zeroing: three different things

These three are used interchangeably on board and they are not interchangeable at all. Only one of them proves the instrument is still accurate.

Zeroing

Sets the baseline

Carried out in clean air, usually at switch-on. It tells the instrument what nothing looks like. It proves the sensors respond to clean air and nothing more — a dead sensor will zero perfectly.

Proves accuracy? No

Bump test

Proves it responds and alarms

A known gas is applied for fifteen to twenty seconds. It confirms gas reaches the sensors, that each channel responds, and that the audible, visual and vibrating alarms all work. It does not adjust anything.

Proves accuracy? Partly

Calibration

Corrects the drift

Span gas of known concentration is applied and the instrument is adjusted to read it correctly, compensating for the sensitivity that every sensor loses over its life. Takes a minute or two and produces a pass, a pass with adjustment, or a fail.

Proves accuracy? Yes

The failure that hides in plain sight

A sensor that has drifted low still switches on, still zeroes, and still shows a plausible number. Nothing about the instrument looks wrong. Only gas of a known concentration reveals it — which is the entire reason both the bump test and the calibration exist, and why neither can be replaced by an inspection of the meter.

Gas meter calibration intervals and what a failed bump test means

The intervals below are the accepted working practice; the maker’s manual for your instrument is what governs, and several flag administrations and company systems set something tighter.

Before each day’s use

Bump test

Generally accepted best practice, and the standard an auditor will hold you to if your own procedures do not say something stricter. Before any enclosed space entry, it is not optional whatever the calendar says.

At least every 6 months

Full calibration

The common maker recommendation. Also required immediately after a failed bump test, after an impact or a fall, after exposure to a high concentration, and after any sensor is replaced.

Per the maker

Sensor replacement

Electrochemical cells and catalytic beads have a finite life that runs whether or not the instrument is used. A meter that lives in a locker ages at much the same rate as one in daily service.

A failed bump test is a straightforward outcome: the instrument goes out of service until it has been calibrated and passes. It does not get used for one more entry because the calibration kit is ashore. And the span gas has its own expiry date printed on the cylinder — calibrating against gas that has passed its date proves nothing, so the cylinders belong in the same schedule as the instrument, alongside the ship’s other pressure vessels and gas cylinders.

Atmosphere testing sequence: sampling a space you have not entered

The whole point of a pre-entry test is that it is carried out from outside. That means a sample line and an aspirator or pump, and it means the sequence matters more than the instrument.

Ventilate first, then test

Test after ventilation has run, not during the first minutes of it, and never as a substitute for it. A space that only passes while the fan is running has told you something important about itself.

Zero and bump in clean air

Done outside the space, upwind of any exhaust or vent outlet. Zeroing the meter in air that is already contaminated sets the baseline wrong in the most dangerous possible direction.

Purge the sample line fully

The reading lags the sample by the time it takes gas to travel the hose plus the sensor’s own response time. On a long line that is minutes, not seconds. The maker states the figure; use it, and wait for the reading to settle rather than for it to appear.

Test at several levels

Top, middle and bottom as a minimum, and at any pocket or recess the plan shows. One reading at the hatch is a reading about the hatch.

Record what was measured, not that it was

Actual figures for each gas at each level, with the time, on the entry permit. “Tested and found safe” is not a test result and will not survive an investigation.

Keep testing after entry begins

Conditions change as people disturb sludge, scale and residues. Everyone entering carries a personal monitor, and the attendant watches the space rather than the paperwork.

Why gases stratify, and why one atmosphere test is not a test

Gases do not mix themselves evenly in a still space. They settle according to their density, which means the atmosphere at the hatch and the atmosphere at the tank bottom can be entirely different things.

Hydrogen sulphide and most hydrocarbon vapours are heavier than air and collect at the bottom, which is also where oxygen depletion is usually worst and where a person is going to end up. Methane is lighter and gathers under the deckhead. A meter lowered to the working level and no further can produce a perfectly acceptable set of readings for a space with a lethal layer sitting a metre below the last sample point.

Gas sensor types, what they measure and what fools them

Knowing which sensor is behind each channel changes how you read the screen, because each technology fails in a characteristic way rather than at random.

Sensor	Usually measures	What fools or degrades it	Practical consequence
Catalytic bead	Flammable gases as per cent LFL	Needs oxygen to burn the sample on the bead; poisoned by silicones, sulphur compounds, lead and halogenated vapours	Under-reads badly in an inerted or oxygen-depleted space — the worst possible place to be optimistic
Infrared	Hydrocarbons and carbon dioxide	Dirty or fogged optics; condensation in the sample line	Works without oxygen, which is why it is the right choice for inerted tanks
Electrochemical	Oxygen, carbon monoxide, hydrogen sulphide	Finite cell life that runs whether used or not; cross-sensitivity between toxic gases; extremes of temperature and humidity	A cell near the end of its life reads low and gives no warning that it is doing so
Photo-ionisation	Volatile organic compounds at low ppm	High humidity; a lamp that has become dirty or weak	Useful for cargo vapours a four-gas meter will not see, but it is an addition rather than a replacement

The maker’s manual for your instrument is authoritative on sensor type, cross-sensitivity and life. Two meters that look identical may not carry the same sensor behind the same channel label.

The catalytic bead problem in oxygen-depleted and inerted atmospheres

This is the single most dangerous technical trap in shipboard gas testing, and it is worth stating plainly. A catalytic bead sensor measures flammable gas by burning a small sample of it on a heated element. Burning requires oxygen. In a space that has been inerted, or one where oxygen has been consumed by rust or cargo, there is not enough oxygen for the reaction — so the sensor reads low or reads nothing, in an atmosphere that may be thick with flammable vapour.

The revised recommendations address this directly by expecting detection equipment to be capable of operating in oxygen-depleted atmospheres. In practice that means an infrared sensor for the flammable channel where inerted spaces are in scope, or a separate instrument for those spaces. If your ship inerts cargo tanks and carries only catalytic bead instruments, that is a gap worth raising before somebody finds it the hard way — and cargo tanks, pump rooms and slop spaces carry their own regime on top of this, set out under the MARPOL annexes.

A calibration date is only useful if somebody sees it coming

Marine Inspection holds every meter, sensor and span gas cylinder against the vessel with its bump test log, calibration certificate and next-due date — and warns before the instrument that will be used for the next entry falls out of date.

See a walkthrough Start free trial

The CO2 channel most four-gas meters do not have

A standard four-gas meter measures oxygen, flammable gas, carbon monoxide and hydrogen sulphide. It does not measure carbon dioxide, and electrochemical cells are not the technology for it — carbon dioxide measurement needs an infrared sensor, which on most ships means a second instrument rather than a fifth channel.

That matters most with organic cargoes. Grain, timber, wood pellets, fishmeal and similar cargoes consume oxygen and give off carbon dioxide as they respire. Because the two move together, a modest fall in oxygen can sit alongside a large rise in carbon dioxide — and carbon dioxide is odourless, tasteless and dangerous at concentrations well below the level at which the oxygen reading looks alarming. Effects can begin at one or two thousand parts per million; the eight-hour exposure limit is 5,000 ppm, which is the same 0.5 per cent that the entry criteria set.

Where it bites

Holds and adjacent spaces carrying organic cargo, spaces next to a discharged fixed carbon dioxide system, and any compartment where something has been fermenting or decomposing.

What to check

Whether the ship actually has a carbon dioxide measuring capability at all, whether span gas for it has been supplied, and whether it is in the same calibration schedule as the four-gas meters.

Why it is missed

Because the four-gas meter shows a clean screen. Nothing on it is wrong; the gas that matters simply is not among the four it measures.

Continuous monitoring and personal gas detection during entry

A pre-entry test describes the atmosphere at one moment. Entry changes it. Disturbing sludge releases hydrogen sulphide, opening a section of pipework releases whatever was in it, and a ventilation fan that stops changes the space within minutes.

Everybody who enters carries personal gas detection, and it is worn in the breathing zone rather than clipped to a bag at the bottom of a ladder. The attendant at the entrance keeps continuous communication and watches the readings as well as the person. And the entry permit has a stated validity — a maximum of eight hours under the revised recommendations — after which the space is re-tested and a fresh permit issued rather than the old one extended.

Gas meter calibration records and the entry permit

Two records carry the weight here, and inspectors and auditors ask for them in the same breath.

The instrument record

Per instrument, by serial number: bump test log with dates, calibration certificates, sensor replacement history, span gas cylinder expiry dates and the next calibration due. Held in the ship’s planned maintenance system rather than a drawer, so the due date surfaces on its own.

The entry permit

Actual readings for each gas at each level tested, with times, the instrument used, who tested, the ventilation arrangement, the stated validity period and the attendant’s name. Figures, not ticks.

The commonest audit finding in this area is not a missing instrument. It is an entry permit with a signature where the readings should be, and a meter whose last calibration certificate nobody on board can produce.

What changed for atmosphere testing in 2026

The carriage requirement itself is unchanged: SOLAS chapter XI-1 regulation 7 has required a portable atmosphere testing instrument since 2016 and still does. What changed is the guidance around it, and the change is substantial enough that ships are working through it now.

Resolution MSC.581(110) took effect on 3 December 2025 and replaced the long-standing recommendations for entering enclosed spaces. For atmosphere testing specifically it brings in numeric pre-entry criteria, expects personal gas detection for every entrant covering oxygen, carbon dioxide, flammable gases and toxic gases, expects that equipment to be capable of operating in oxygen-depleted atmospheres, and caps the entry permit at eight hours. It also asks for a ship-specific enclosed space register and an enclosed space emergency response plan.

These are recommendations rather than a new SOLAS regulation, but they reach ships through the safety management system, which means an auditor can ask for them. The register and the carbon dioxide capability are the two items most likely to be absent on a ship that has not looked at this since 2016.

Atmosphere testing and gas meter calibration: frequently asked questions

How often should a gas meter be bump tested?

Before each day’s use is the generally accepted best practice, and before any enclosed space entry regardless of when the last one was done. A bump test takes fifteen to twenty seconds and confirms that gas reaches the sensors and that the alarms work.

How often does a gas meter need full calibration?

At least every six months is the common maker recommendation, and your instrument’s manual governs. Calibration is also required immediately after a failed bump test, after an impact or a fall, after exposure to a high gas concentration, and after any sensor is replaced.

What is the difference between a bump test and a calibration?

A bump test proves the instrument responds to gas and that its alarms work, but adjusts nothing. A calibration applies span gas of known concentration and adjusts the instrument to read it correctly, correcting the sensitivity every sensor loses over time. Zeroing does neither — it only sets the clean-air baseline.

What oxygen level is required before entry?

At or above 20.9 per cent, which is normal atmospheric oxygen. A reading above about 21 per cent indicates oxygen enrichment, which is a separate and serious hazard usually caused by a leaking oxygen line.

Why does a four-gas meter not measure carbon dioxide?

The four standard channels are oxygen, flammable gas as per cent LFL, carbon monoxide and hydrogen sulphide, and carbon dioxide needs an infrared sensor rather than an electrochemical cell. On most ships that means a separate instrument, which also needs its own span gas and its own calibration schedule.

Can a catalytic bead sensor be trusted in an inerted tank?

No. It measures flammable gas by burning a sample on a heated bead, and burning needs oxygen. In an inerted or oxygen-depleted space it reads low or reads nothing even when flammable vapour is present. Use an infrared sensor for those spaces.

Where in the space should the atmosphere be tested?

At several levels — top, middle and bottom as a minimum, plus any pocket or recess. Gases separate by density, so hydrogen sulphide and hydrocarbon vapours collect low while methane gathers high, and a single reading taken at the hatch says nothing about the tank bottom.

How long is an entry permit valid?

The revised recommendations cap it at eight hours. Beyond that the space is re-tested and a fresh permit issued rather than the existing one extended, and the atmosphere is monitored continuously throughout the entry in any case.

Make every reading a record, and every calibration date visible

Marine Inspection keeps the meters, the sensors, the span gas and the entry permits against the vessel — readings captured at the space on one phone, bump test and calibration history per serial number, and alerts before an instrument falls out of date across the whole fleet.

Book a free demo Start free trial