The ninety-centimetre lie

Eleven civic buildings, all completed since 2015, all on a plinth. The heights run from 600 millimetres to 1.4 metres and cluster hard at 900mm. When we asked the architects why, nine said some version of presence, two said drainage, and none could produce a drawing where the decision was made.

That last part is the finding. Not that the plinth is wrong — it is often right — but that in eleven out of eleven cases it arrived before the reason for it. We asked for the earliest model each practice still held. In eight the plinth is already there, at its final height, in a file dated before the first client workshop. In two it appears between the workshop and the planning submission. In one the practice could not find anything earlier than the tender set, which we have counted as unknown rather than as evidence.

A raised floor does real work. It keeps water out, it hides the servicing that would otherwise occupy a ceiling, it gives a level threshold across a sloping site, and it separates a building from a car park. Every one of those is a reason you can draw. What none of them explains is the clustering. Flood levels do not cluster. Site falls do not cluster. Service zones vary with the system.

The ramp is the tell. In all eleven buildings the ramp is longer than the stair, set to one side, and arrives at a different door. In seven it was added after planning. A plinth is a decision about who arrives at the front and who arrives at the side, and it is taken before anyone has said that out loud.

This is where the heights stop being an abstraction. At 900mm a ramp at the maximum gradient permitted for a public approach needs a little over eighteen metres of run, plus landings. Eighteen metres has to go somewhere. In four of the eleven it goes along the front, so the main elevation is read across a handrail. In five it goes down the side, out of the photographs. In two it was value-engineered into a platform lift, which is a cupboard with a button, and which was out of service on the day we visited one of them.

None of those outcomes was designed. They are all consequences of a height nobody can account for, resolved years later by whoever had the drawing open. The practices were not careless. Three of them are very good. The decision simply never presented itself as a decision, because it was made by a default rather than by an argument, and defaults do not generate minutes.

We put the clustering to four of the eleven practices. Two recognised it immediately and said, in almost the same words, that it is what a civic building looks like. One said their model template opens with the ground slab at that level and has done since a project in 2011. The fourth asked to see the survey before commenting and then did not comment.

The template answer is the honest one and the most interesting. A default inside a file is a brief that nobody wrote, agreed to, or can be held to. It survives staff turnover, it crosses between unrelated projects, and it is invisible in every drawing it produces, because a drawing shows what was decided and not what was never questioned.

The cover study carries the same 900mm. Cut through it and the plinth stops being a base and becomes what it is: a floor level that the ground has to climb to.

None of this is an argument against raising a floor. There are sites where 900mm is a flood level and buildings where the plinth is the only thing holding a terrible street together. It is an argument against drawing it first and justifying it afterwards, which is what the survey found in every case where the practice still had the early models.

The test is cheap. Open the earliest file. If the floor level is already there, it was not designed — it was inherited, and the ramp you are about to spend four months on is the interest payment.