# Carbon-monoxide detection inspection log **Version:** 1.0 **Date:** 2026-08-22 **Type:** Instrument --- ## 1. What this is A walk of the carbon-monoxide detection devices in the building, recorded device by device on a sheet printed from the device inventory. Some brand standards require the log itself and not only the work; whether yours does is a question for your standard. Either way the page is what survives the walk. The sheet is carried. It is filled in standing at the device. ## 2. Two documents, not one The **inventory** lists the devices that exist: location, unit, identifier, type, the test method that device's instructions require, install date, and end-of-life date where the device has a fixed service life. It changes only when a device is added, moved, or removed, and each change is dated and initialled. The **inspection record** is this sheet: one walk, one line per device. The sheet is printed from the inventory. It is never copied forward from a completed log. A copied log inherits every line from the last one, including devices that came off the wall in a renovation and were never taken off the list. Those lines keep reading as inspected and fine for as long as someone keeps copying. Printing from the inventory does not stop that by itself; it moves the failure to a place where somebody has to write something false rather than simply not notice. ## 3. The sheet **Header — machine-printed at print time, not hand-written** Inventory version · date the inventory was printed from · number of rows printed · route order used. **Header — hand-written** One block per session, because a walk can span shifts: date · start time · end time · inspector name · initials. Panel on test — time and initials. Panel restored — time and initials. **Rows — pre-printed, one per device, numbered, in fixed route order** | # | Location / unit | Device ID | Type | Required test method | End-of-life | Open since | Mark | Time | Init | Follow-up / re-visit | |---|---|---|---|---|---|---|---|---|---|---| | 1 | | | | | | | | | | | | 2 | | | | | | | | | | | | 3 | | | | | | | | | | | The left seven columns arrive filled. The inspector writes four: mark, time, initials, and a follow-up reference where required. **Marks — one letter, legend printed on the sheet** **P** tested, passed · **B** battery replaced, then tested, passed · **R** device replaced, then tested, passed · **F** failed, out of service · **N** not found at the printed location · **A** no access. **N** and **A** are different findings. **N** means the device is not where the inventory says it is; it opens an inventory change — found and corrected, replaced, or the line retired with a date and a reason. **A** means the inspector could not get to it, an occupied room or a locked space; it opens a re-visit date on the same row, not an inventory change. Without **A** on the form, an inspector who cannot enter a room has two exits: report a device missing that is fine, or write a pass they did not perform. Most people write the pass. **Footer** Count by mark, each written in its own box · panel restored confirmed · inspector signature · received by (inventory holder) signature, date. ## 4. What makes an incomplete sheet visible **A blank row is a blank row.** Rows are pre-printed from the inventory, so a device nobody reached leaves an empty numbered line. It cannot leave nothing. With hand-written rows a skipped device produces no row at all, and no row is invisible. **Times must be non-decreasing down the page and inside the session times.** Rows print in route order, so a real walk produces times that only go forward, each at or after that session's start and at or before its end. Anyone can check that without knowing the building. A sheet written up afterwards has to invent a rising sequence that also fits the session blocks and the panel times. **The counts are checked by a second hand.** Row count is machine-printed in the header; the inspector counts marks into the footer boxes; the inventory holder checks the two against each other on receipt and signs. Nobody certifies their own arithmetic. **A sheet with a panel on-test time and no restore time cannot be filed.** The restore time must be at or after the last device time on the page. A system left on test shows up as a missing entry above a signature. **End-of-life and open-since print on the row.** A device whose printed end-of-life date has passed contradicts a **P** on its own line, without anyone having to remember the date. An **F**, **N** or **A** left unresolved prints again with the date it opened, so an old finding does not look like today's. ## 5. Intervals, standards and numbers from outside Monthly is a convention, not a finding. So is any figure for placement, test medium, record retention, or service life. These come from local code, a brand standard, and the instructions for the specific device, and they do not agree across jurisdictions or equipment. This instrument states none of them. Set the interval and thresholds from your own requirements and your own measured failure and replacement rates, and name the source beside each number. The house has a standing example of the alternative: a par level built on a room count taken from a status report instead of the room list made every par on seven floors wrong in the same direction at once, and nothing on the page showed it. ## 6. Before first use - Was the inventory built by walking the building, rather than from a drawing or an earlier list? Who holds it, and who may change it? - For each device type here, does the test button verify the sensor or only the horn and electronics, and what does a full test require? - Which devices are combination smoke/CO units, and how does this log stay consistent with the fire log recording them? - At your occupancy, can every guest-room device be reached in one interval? If not, what is the re-visit rule for **A**? - Who receives a failed device, within how long, and on what sheet? ## 7. What this instrument does not do It does not tell you what your jurisdiction requires and it is not a compliance document. It records that a test happened; it does not make the test correct, and a button pressed without knowing what the button proves yields a clean log and an unsafe building. It does not find a device that is present, working, and mounted in the wrong place — the sheet assumes the inventory is right about where devices belong. It cannot see a walk performed carelessly. The time-order and count properties raise the cost of a false page; a determined person can still fabricate a route that survives them. The sheet publishes blank. Every property's filled sheets are its own.