Glæshus Tower came to us through an invited competition in late February, four weeks after the studio opened. The brief was twelve storeys of workspace on a corner site in central Malmö, with stepped terraces facing south. We submitted in April, and were told the news on a Friday afternoon over a poor phone line.

First decisions: brass, glass, winter sun

The first sketches are about the cladding. We are testing brushed brass mullions against low-iron glass, with samples sitting on the window sill in the studio so that we can watch them through a week of weather. The brass we are after is a warm grey-pink that the foundry warned us about but no sample has yet properly shown.

The geometry of the stepped terraces is being set by a 1:50 cardboard model held up to the courtyard sky every afternoon between two and four. The steps will not be readable until the building is occupied in 2028, but we can already feel the angle of light they are made for.

Project update: Glæshus Tower

We are often asked why the studio works with such a narrow palette. Limewashed brick, oiled oak, raw concrete, brushed steel, a few weights of linen for curtains. The list is short and we have written it down. It is taped to the wall above the model stand.

The honest answer is that we are not interested in invention for its own sake. We are interested in attention. A short list of materials forces us to look harder at each one, to know how an oak floor takes a low sun in February and a high sun in July, to know which limewash will chalk in the rain and which will hold.

A quiet building is not an empty building. It is a building that does not ask for the visitor’s attention before the visitor has decided to give it. The proportions are settled. The light is settled. The door handle sits where your hand expects it. Nothing is shouting.

We want the building to be the second thing you notice, after the weather.

This takes more drawing, not less. Restraint is expensive. The detail that disappears takes three weeks and four prototypes. The wall that looks effortless is built by a mason who has been laying brick for forty-one years and who left us a long voicemail about the mortar mix.

Every project starts with a list of what we are removing. No skirting boards. No fluorescent lighting. No ceiling grids. No interior fluorescence of any kind. No “feature wall.” The list grows as the building grows, and by the time a project goes to site the list is often longer than the specification.

A house we are drawing for a client on Bornholm has only seven different materials in it so far. We counted. The client noticed first. She said the rooms already felt longer than they were, that walking through the model was like walking along a beach.

On quiet architecture

Daylight is the only material in a building that arrives free, changes every minute, and cannot be returned to the supplier. We treat it as the most important specification in a project and we draw it before we draw walls.

Step one: walk the site at the worst hour

Every project begins with a site visit between 14:00 and 16:00 on the shortest plausible day of the year. We bring a folding stool and a paper compass. We sit. We watch where the light is, and more importantly where it is not. The early sketches are not of the building. They are of the shadow it will cast at four in the afternoon in January.

The window openings are drawn before the floor plan. This sounds backwards. In practice it means the rooms are organised around the few moments of usable winter light, and the rest of the plan settles itself around those moments.

Step three: model in cardboard, not the screen

Daylight simulation software is useful for compliance and almost useless for design. A 1:50 cardboard model held to the actual sky in the studio courtyard will tell you in thirty seconds what the screen tells you in three days. We still use the software. We do not trust it before we have held a model.

A drawing can lie about light. A cardboard model in a real window cannot.

Designing with daylight, not against it

On the second week of the studio, the joiners arrived with four drawing boards we had ordered before we even had chairs. A friend asked, politely, why we were spending the budget that way. The boards are heavy. They take up floor space. The pencils need sharpening. It is a fair question and the answer is longer than I expected.

A pencil line takes about a second. A line in the model takes a tenth of that. The pencil is slow on purpose. It will not let you commit to a wall before you have looked at it. It will let you change your mind in the middle of a line. The model will not.

The drawing is for the building, not the client

We do not show the hand drawings to clients. They are the first conversation the project has with itself, and they belong to the studio. By the time a client sees a drawing it has usually been through six pencil iterations and three models. The pencil drawings are not deliverables. They are the building thinking out loud.

If you cannot draw it by hand, you do not yet understand it. If you can, you may begin in the model.

The friend who asked, in the second week, has now visited the studio twice more. The second time was last week. She stood at one of the boards, with a 4B pencil and a section through a stair we had not yet started, and the question was hypothetical.

Field notes, project updates, and slow essays from the studio.

Project update: Glæshus Tower

Earlier entries

On quiet architecture

Designing with daylight, not against it

Why we still draw by hand