Where the catalogue stops
Every printed maximum is a maximum under conditions. Knowing which conditions, and what it costs to leave them, is most of the job.
A system catalogue is a careful document. Every figure in it was reached by testing a specimen of a stated size, with stated glass, stated hardware and a stated load. The numbers are real.
They are also conditional, and the conditions travel with them. A maximum sash width is valid at an assumed weight. A watertightness class belongs to one specimen at one dimension. A chart axis is where the chart ends, which is not the same as where the assembly ends.
Most of the difficulty in facade work sits in that gap — between what a document permits and what a building asks for.
Three ways a project leaves the chart
It asks for less. On a house above Beirut, the requirement was a sliding assembly with as little visible aluminium as the structure would allow. The catalogue draws thirty millimetres. The project needed twenty.
Twenty millimetres is not a smaller version of thirty. Everything that lived inside those ten millimetres — drainage path, interlock, the fixing for a concealed balustrade — has to go somewhere else or stop existing. Each of those is a separate decision with a separate consequence, and none of them appears in a chart.
It asks for more. On the coast at Amchit, the main elevation carries a fixed pane of five by three metres. That single panel passes 1.1 tonnes.
At that weight glass is no longer glazing. It is a structural element that has to be craned, set, packed and held, and the tolerance that matters is the one in the workshop rather than the one on site, because there is no reveal left to absorb an error. The frame around it was engineered so that the exceptional panel was not exceptional to install.
It asks for something that does not exist. On a project where acoustic separation was the client's reason for coming to us at all, no available frame reached the performance at the proportion required. We developed one, proved it, and then repeated it more than five hundred times across two thousand square metres.
Repetition is the real test of a developed detail. A frame that works once is a prototype. A frame that works five hundred times was resolved before the first one was cut.
The judgement, not the number
None of this argues against catalogues. We work with European system houses precisely because their documents are rigorous, their testing is independent and their tolerances are honest.
The judgement is knowing when a project has left the tested envelope, saying so plainly, and being able to explain what changes as a result — what has to be verified, what carries risk, what the alternative would cost.
The unhelpful answers are at both extremes. A company that will only build what is printed cannot serve architecture that is trying to do something. A company that will agree to anything has not read the document carefully enough to know what it is agreeing to.
The useful position is narrow and it is where we try to stand: read the conditions, know exactly where they end, and be honest about what happens on the other side.