The Story of a Detail - Crown 515 Tile Set-Out

This is the story of forty millimetres.

Start with the pattern, because it is doing more than it lets on. Those are not squares. Each cube is drawn from three rhombus tiles, squashed squares with no right angles in them, one lit face and two shaded, locking together into a field that reads as a wall of stacked boxes. The cube is an illusion, and it is held together entirely by the joints.

Then look at where it sits. A street corner, beside a row of terrace houses that has stood there over 150 years, and the triangular tiled face lifts from just above the lane and cuts upward on exactly the angle of the old terrace roof behind it. The photograph is the top of that face, the tip, where the rake runs up and stops. And it is not a wall with tiles applied to it. The tiles were cut to that geometry first and the whole wall was set out around them, so the pattern met the edges of the building exactly. Set-out is the trade word for deciding, before anything is laid, precisely where every tile will land.

That face was Smart Design Studio's idea from the first conversation, and over three years we shaped the structure to sit inside it. What you are looking at is exactly what was drawn. Getting it there came close to costing the top of the building, and what saved it was a question nobody had gone up the scaffold to ask.

Yes, almost none, because it is all about the pattern. An illusion of cubes only survives while the geometry is exact. Let the joints drift by a millimetre or two and the three rhombi stop agreeing with each other, the light face and the shaded faces fall out of alignment, and the cubes quietly stop being cubes and become diamonds. Nobody looking at the wall would be able to say what was wrong with it. They would only know that it had stopped working.

So the joints had to hold to a tolerance measured in single millimetres, all the way up the wall. That is close to the limit of what tile and stone can do (they can be cut finer than that, they cannot be laid finer than that).

Nor can any of it be fudged afterwards. Follow the raking edge in the photograph and look at the tiles that meet it. Every one is a different shape, because every one meets that diagonal at a different point in the pattern. They are one-off cuts, made for a single position on a single wall.

Calida built the structure itself to those dimensions, well past what you would normally expect on site. The construction of the wall was never the difficulty. The difficulty lived in the joints.

It accumulates. A joint cannot close below a certain width or the tiles bind against each other, so in practice every one sits a little wider than the ideal. Multiply that by every row of cubes from the base of the wall to the top and the tiling finishes 40mm higher than the drawing.

On most walls you would never notice. On this one, where the top of the tiling had to land on a fixed point at the very top of the building, those 40mm were everything.

It could, and for longer than anyone wanted that was the only answer any of us had.

Working the set-out on site, Calida saw exactly what the accumulation meant. Lay the tiles the normal way and once the joints are built, the tiling arrives 40mm above the tip. To bring that last row back into line, the whole top of the building had to come up by 40mm.

Which meant going into a finished building and cutting the steel out of the roof.

And this is not a simple roof. Its geometry twists, so nothing in it repeats and every piece of steel in it is particular to the place it sits. It had taken us weeks to resolve that geometry in the first place. You cannot lift a roof like that the way you would pack up a beam (packing being shims, slid underneath to raise something a few millimetres). Move the top by 40mm and every member changes length and angle, so not one piece of what came out could go back in.

Which is a rebuild. Cut a finished roof off a building, fabricate another one, put it in its place. At height, on a live site, with the job stopped around it, to gain 40mm.

That is what we worked on. Drawing it, pricing it, working out the sequence. The number was large. Nobody wanted it, and nobody had anything better.

We didn't, and that is the part I find most interesting about it now. Nobody went up there to rethink anything.

Calida had a narrower question. Could the roof be packed up rather than cut apart? Could we win the 40mm without ever putting an oxy to the steel? That is what I was asked to site for.

So we went to the top of the scaffold, which is where you go when you want to look at the top of a building. Glenn and Ronald from Smart Design Studio, Matt and Vince from Calida, me, and plenty of others besides. And we spent the best part of an hour on the roof.

I do not think anything would have come of it any other way. You had to be standing in front of the real thing, with the people who knew it best, the work paused around us and a decision that could not wait.

It did, though it was not the answer we had gone up there for, and I want to be accurate about how it arrived.

I was doing the thing I find most useful in those moments, which is to build the wall in my head, tile by tile, and watch what happens. Laying from the base upward, arriving at the top, 40mm higher than the roof. Over and over.

And on one of those passes I stopped looking at the top. We had spent weeks working out how to move it. Nobody had gone back and asked whether it was the part that had to move. Me included. A wall has two ends. If the top could not come up, the bottom could go down.

We were standing at the very top of the scaffold, looking at the top of the wall. The answer was at the other end of it.

Because of what each edge costs to move. Look at the tip in the photograph, where the raking edge runs up and stops against the vertical. That point is the highest thing on the building, and moving it means cutting a roof apart. The bottom of the tiling comes down to the base of the wall, just above the lane, and moving that costs nothing at all.

Which comes down to a choice about where you start laying. Begin at the bottom, and you start from the cheap edge and drive every accumulated millimetre into the expensive one. Begin at the top, and you start from the expensive edge and let the tolerances settle at the bottom, where the base and the low parapet wall below it can quietly take up the difference. The same 40mm, moved from the one place that could not absorb it to the one place that could.

Once it was said, everyone saw it at once. We spent a few more minutes checking the base of the wall, the base came down by 40mm, and the tiling went ahead. Weeks of drawings for cutting a roof off and building a new one went in the bin, and the steel was never touched.

The roof, first. Nobody had to go back into a finished building and cut the steel out of it. That matters more than the money it saved, and it saved a lot of money. Every time you cut finished work apart and put it back, you hand back a known thing and get a slightly less known one in return. The structure Calida had built to those fine dimensions was left exactly as it was, still carrying what it was welded to carry.

Then the program. Raising the roof stopped the job, and no tile could go on the wall until the new steel was up. Starting at the top cost us a few minutes checking the base.

And the architect got the geometry they had drawn three years earlier, arriving on that corner at the angle of the terrace roof, with nothing about it renegotiated. The client paid for a wall. They did not pay for it twice.

I don't think so, and the move is not new either. Starting a set-out from the edge you cannot afford to move is old practice. Stonemasons do it, bricklayers do it, anyone hanging a ceiling in a room with an out-of-square floor does it. Asked the question cold in the office, I would have told you that in a sentence.

What was hard was seeing it there, and the reason it got seen has less to do with me than it looks. Matt and Vince were holding the build, which meant holding the risk of getting it wrong. Glenn and Ronald were holding the design, which meant holding what the wall had to look like when it was finished. I was holding a question about packing steel. It was the lightest thing anyone up there was carrying, and it was the only one that could be put down.

Any of them would have seen it from where I was standing. It was the others who put me there. First by asking me up, then by spending an hour taking the problem apart until the only thing left in it was which end of the wall we were arguing about.

Answers like that rarely arrive from one person alone at a desk. They come from the right people, standing in front of the real thing, at a moment when the decision cannot wait.

Behind every detail, a story. This one is about how long you can spend answering a question nobody has gone back and checked, and about the fact that the simplest answer is usually the last one you see, and is never found alone.

Architect: Smart Design Studio

Builder: Calida

Engineer: Cantilever Studio