The Order the Envelope Is Built In

Nothing on this list is irreversible because of the material. Each item becomes unrecoverable because something is placed over it, usually within hours or days of its being finished, and usually by a different trade. That makes the sequence itself a design consideration rather than an administrative one, which is how the records treat it.

It also means the useful question is not what could go wrong but when it stops being findable. The control layer record asks, of every junction, at which point in the construction programme it becomes permanently concealed. Answering that for each critical layer produces the map.

The masonry cavity wall record is unusually direct. Nearly every failure mode of that wall is created while it is open, mortar left on ties, interception lapped the wrong way, boards not held tight, outlets blocked, and none of it can be seen once the outer leaf is up. The record concludes that inspection while the void is open carries unusual weight here.

The interception record narrows the window further, to the courses rather than the leaf. Trays are installed within a wall as it is built, by a trade whose main task is laying masonry, and the published question asks how the trays and their stop ends will be inspected before the following courses conceal them. That is a window measured in courses, not in weeks.

The roof abutment record describes a sequencing problem that no amount of specification solves on its own. Provision for the concealed tray has to be made while the wall is being built, by a trade that will have left before the roof is weathered, and where it is missed the remedy involves taking part of the wall apart again.

The record's own conclusion about what to do with that is worth adopting generally: recording who provides the concealed item is more useful than specifying its material. Material choices survive a programme; scope assumptions do not.

The rooflight kerb record identifies an ordering failure that follows directly from convenience. Openings are often formed after the roof's field layers have been laid, which reverses the order in which the control layers should have been carried up the kerb, and the record states that when the aperture is formed is a design question rather than a programming one.

The warm deck record puts the same rule positively. On a warm deck every layer above the deck has to be turned up the kerb in the order it occurs in the field, and an aperture is where that order is most often broken. The record also notes that a vapour layer stopped at the deck turns the aperture into a route by which internal air reaches cold construction from behind, where nobody is looking for it.

In a built-up metal roof or wall, the layer carrying the air and vapour control duty is the first one fixed and the first one hidden. The roof record says the liner is asked to do unrelated jobs and only the visible one gets judged on site: as a ceiling it is assessed on line and finish, as the control element it is assessed on whether its laps are sealed and its perimeter closed.

The wall record adds that because the liner is complete and hidden as soon as it is finished, everything depending on it is settled before anyone can see the wall as a whole. Both records also note that each later operation can undo an earlier one, because fasteners for the outer sheet pass through the insulation, foot traffic can flatten the quilt, and any tape applied to the liner has to survive everything the following trades do above it.

The warm deck record generalises the pattern for a whole build-up. It is completed from below upwards, so each layer is covered by the next and cannot be revisited, and weather during the works and how the roof is kept dry before the membrane closes it therefore affect what ends up trapped inside the assembly.

The inverted roof turns the same fact into a handover question, because nothing below the surfacing can be inspected once complete. Its published questions ask how the concealed waterproofing will be checked before the roof is covered, and what record will exist afterwards, which is the only evidence available to a later owner or surveyor.

The trafficked deck adds that the surfacing is installed over a completed waterproofing layer, so the sequence itself is a hazard to the thing already finished, and how the layer is protected while the terrace is built is a real part of the assembly rather than site practice.

By the time an envelope is finished, the visible surfaces are the ones with the least to report. The window opening record notes that whichever layer is installed last is the only one anyone will ever see, and that a perimeter is normally judged on the appearance of that layer alone. The control layer record notes that a whole-building air test reads the envelope as a whole and cannot say which junction is leaking.

This is why the records place the weight they do on records made while things were open. A photographic record taken before closing, keyed to each elevation, is described as what a later air test result or damp investigation can be interpreted against, which is a different function from proving that the work was done well.

Some of these windows can be widened by decision. Forming a roof aperture before the field layers are laid rather than after is a choice. Agreeing that a following trade holds off until an inspection has happened is a choice. Grouping penetrations so that they are formed before a covering goes on rather than cut into it afterwards is a choice, and the roof records recommend it as materially easier than working back down through a completed build-up.

Where a window genuinely cannot be kept open, the records ask for the substitution to be explicit. If nobody can inspect it, decide now what record will stand in its place, who takes it, what it shows and where it is kept. That is the difference between a layer that was never inspected and a layer nobody can say anything about.