|

Aug 14, 2026

A Practice of Edges

Almost everything that determines whether a building is any good happens at an edge. The site boundary, the reveal, the threshold, the junction, four scales, one problem.

Landscape & Urbanism

Masterplanning

An article by

a black and white photo of a woman standing in front of a window

Camille Dubois

Associate, Interiors

There is a habit in architectural education of teaching buildings as objects: a mass, a volume, a thing with a shape. It is a reasonable place to start and a bad place to stay, because almost nothing about how a building is experienced happens in the middle of it.

It happens at the edges. Where the site stops and something else begins. Where the wall ends and the sky starts. Where you pass from outside to in. Where the timber meets the steel and two different trades, on different days, had to agree.

These read as four separate problems at four scales. They are the same problem, and a practice that gets good at one tends to get good at all of them.

The edge of the site

Every site has a boundary and the boundary is almost always where the interesting information is. What is beyond it, what the neighbours can see, where the sun comes from, which way people already walk across the plot.

The common failure is to treat the boundary as the limit of the project — to design inward from it, resolving the site as a self-contained composition. This produces buildings that are coherent in plan and mute in the street, because every decision about the edge was made from the inside looking out.

At a pavilion in an Amsterdam park, the root protection zones of mature plane trees covered most of the plot, and the diagonal desire line that thousands of people walk each day crossed it. Both are edge conditions in the sense that matters: things outside the brief that determined the plan. The columns land on screw piles positioned in the gaps between root zones, which is why the structural grid is slightly irregular. We did not choose that geometry. The site did.

The edge of the building

The most consequential dimension in most facades is depth, and it is the first thing lost when a budget tightens.

A window reveal at 300mm and the same window at 180mm are not a subtle difference. Depth produces shadow, shadow produces a facade that changes across the day, and a flush window produces a wall with holes in it that looks identical at eight in the morning and six in the evening. The cost difference is small. The visual difference is most of what people mean when they say a building looks cheap.

The same logic governs the roof. At the pavilion the roof oversails generously on all four sides, which does the practical work of keeping rain off the open two-thirds, and does the compositional work of giving the building a decisive top. Buildings that stop abruptly at the roofline read as fragments of something taller. The eaves is an edge and it deserves the attention normally given to the entrance.

The edge of the room

Threshold is the edge at human scale, and it is where housing succeeds or fails.

What lies between the public street and a private front door, and how many gradations does it pass through? Street, shared ground, semi-private threshold, private interior — four steps, each legible. Housing that works has this gradient. Housing that does not has a door onto a corridor, and no amount of quality inside the apartment repairs it.

The pavilion makes the argument in its simplest form. Two-thirds of the building has no walls. There is enclosed accommodation pushed to one edge, and the rest is covered, open, and unconditioned — neither inside nor outside but a third thing, which is the part people actually use. A café operator briefed for summer trading now works year-round, because the covered edge is comfortable in rain, and rain is what Amsterdam mostly does.

The edge of the detail

At the smallest scale, an edge is a junction: two materials, two trades, two tolerances that do not match.

Timber moves and steel does not. Concrete is poured to ±15mm and glazing is manufactured to ±2mm. Every junction between them is a negotiation, and the drawing either resolves it or leaves it to be resolved on site by somebody with a grinder and a deadline.

The discipline is to decide in advance which element is dominant and which accommodates — and then to express the accommodation rather than conceal it. A shadow gap is not decoration. It is an honest admission that two things arriving from different factories will not meet perfectly, and a designed tolerance is invisible in a way that a mastic joint never is.

At the pavilion every connection is bolted rather than glued or welded, so the structure can be taken apart at the end of its permission. That requirement forced every junction to be visible and considered, because you cannot hide a connection you intend somebody to undo. The building is better for the constraint, which is generally how constraints work.

Where this stops being useful

A focus on edges can become a mannerism. There is a strain of contemporary detailing — everything floating, every element separated by a shadow gap, no two materials permitted to touch — that has stopped being about resolution and become a style, and an expensive one. Sometimes two things should simply meet.

It can also become an excuse for neglecting the middle. A building can have superb junctions and a bad plan, and the plan matters more. If the choice is between a well-resolved facade and a section that gets daylight into the middle of the floor plate, the section wins every time and it is not close.

The edges are where the care shows. They are not where the quality comes from.

|

Aug 14, 2026

A Practice of Edges

Almost everything that determines whether a building is any good happens at an edge. The site boundary, the reveal, the threshold, the junction, four scales, one problem.

Landscape & Urbanism

Masterplanning

An article by

a black and white photo of a woman standing in front of a window

Camille Dubois

Associate, Interiors

There is a habit in architectural education of teaching buildings as objects: a mass, a volume, a thing with a shape. It is a reasonable place to start and a bad place to stay, because almost nothing about how a building is experienced happens in the middle of it.

It happens at the edges. Where the site stops and something else begins. Where the wall ends and the sky starts. Where you pass from outside to in. Where the timber meets the steel and two different trades, on different days, had to agree.

These read as four separate problems at four scales. They are the same problem, and a practice that gets good at one tends to get good at all of them.

The edge of the site

Every site has a boundary and the boundary is almost always where the interesting information is. What is beyond it, what the neighbours can see, where the sun comes from, which way people already walk across the plot.

The common failure is to treat the boundary as the limit of the project — to design inward from it, resolving the site as a self-contained composition. This produces buildings that are coherent in plan and mute in the street, because every decision about the edge was made from the inside looking out.

At a pavilion in an Amsterdam park, the root protection zones of mature plane trees covered most of the plot, and the diagonal desire line that thousands of people walk each day crossed it. Both are edge conditions in the sense that matters: things outside the brief that determined the plan. The columns land on screw piles positioned in the gaps between root zones, which is why the structural grid is slightly irregular. We did not choose that geometry. The site did.

The edge of the building

The most consequential dimension in most facades is depth, and it is the first thing lost when a budget tightens.

A window reveal at 300mm and the same window at 180mm are not a subtle difference. Depth produces shadow, shadow produces a facade that changes across the day, and a flush window produces a wall with holes in it that looks identical at eight in the morning and six in the evening. The cost difference is small. The visual difference is most of what people mean when they say a building looks cheap.

The same logic governs the roof. At the pavilion the roof oversails generously on all four sides, which does the practical work of keeping rain off the open two-thirds, and does the compositional work of giving the building a decisive top. Buildings that stop abruptly at the roofline read as fragments of something taller. The eaves is an edge and it deserves the attention normally given to the entrance.

The edge of the room

Threshold is the edge at human scale, and it is where housing succeeds or fails.

What lies between the public street and a private front door, and how many gradations does it pass through? Street, shared ground, semi-private threshold, private interior — four steps, each legible. Housing that works has this gradient. Housing that does not has a door onto a corridor, and no amount of quality inside the apartment repairs it.

The pavilion makes the argument in its simplest form. Two-thirds of the building has no walls. There is enclosed accommodation pushed to one edge, and the rest is covered, open, and unconditioned — neither inside nor outside but a third thing, which is the part people actually use. A café operator briefed for summer trading now works year-round, because the covered edge is comfortable in rain, and rain is what Amsterdam mostly does.

The edge of the detail

At the smallest scale, an edge is a junction: two materials, two trades, two tolerances that do not match.

Timber moves and steel does not. Concrete is poured to ±15mm and glazing is manufactured to ±2mm. Every junction between them is a negotiation, and the drawing either resolves it or leaves it to be resolved on site by somebody with a grinder and a deadline.

The discipline is to decide in advance which element is dominant and which accommodates — and then to express the accommodation rather than conceal it. A shadow gap is not decoration. It is an honest admission that two things arriving from different factories will not meet perfectly, and a designed tolerance is invisible in a way that a mastic joint never is.

At the pavilion every connection is bolted rather than glued or welded, so the structure can be taken apart at the end of its permission. That requirement forced every junction to be visible and considered, because you cannot hide a connection you intend somebody to undo. The building is better for the constraint, which is generally how constraints work.

Where this stops being useful

A focus on edges can become a mannerism. There is a strain of contemporary detailing — everything floating, every element separated by a shadow gap, no two materials permitted to touch — that has stopped being about resolution and become a style, and an expensive one. Sometimes two things should simply meet.

It can also become an excuse for neglecting the middle. A building can have superb junctions and a bad plan, and the plan matters more. If the choice is between a well-resolved facade and a section that gets daylight into the middle of the floor plate, the section wins every time and it is not close.

The edges are where the care shows. They are not where the quality comes from.

|

Aug 14, 2026

A Practice of Edges

Almost everything that determines whether a building is any good happens at an edge. The site boundary, the reveal, the threshold, the junction, four scales, one problem.

Landscape & Urbanism

Masterplanning

An article by

a black and white photo of a woman standing in front of a window

Camille Dubois

Associate, Interiors

There is a habit in architectural education of teaching buildings as objects: a mass, a volume, a thing with a shape. It is a reasonable place to start and a bad place to stay, because almost nothing about how a building is experienced happens in the middle of it.

It happens at the edges. Where the site stops and something else begins. Where the wall ends and the sky starts. Where you pass from outside to in. Where the timber meets the steel and two different trades, on different days, had to agree.

These read as four separate problems at four scales. They are the same problem, and a practice that gets good at one tends to get good at all of them.

The edge of the site

Every site has a boundary and the boundary is almost always where the interesting information is. What is beyond it, what the neighbours can see, where the sun comes from, which way people already walk across the plot.

The common failure is to treat the boundary as the limit of the project — to design inward from it, resolving the site as a self-contained composition. This produces buildings that are coherent in plan and mute in the street, because every decision about the edge was made from the inside looking out.

At a pavilion in an Amsterdam park, the root protection zones of mature plane trees covered most of the plot, and the diagonal desire line that thousands of people walk each day crossed it. Both are edge conditions in the sense that matters: things outside the brief that determined the plan. The columns land on screw piles positioned in the gaps between root zones, which is why the structural grid is slightly irregular. We did not choose that geometry. The site did.

The edge of the building

The most consequential dimension in most facades is depth, and it is the first thing lost when a budget tightens.

A window reveal at 300mm and the same window at 180mm are not a subtle difference. Depth produces shadow, shadow produces a facade that changes across the day, and a flush window produces a wall with holes in it that looks identical at eight in the morning and six in the evening. The cost difference is small. The visual difference is most of what people mean when they say a building looks cheap.

The same logic governs the roof. At the pavilion the roof oversails generously on all four sides, which does the practical work of keeping rain off the open two-thirds, and does the compositional work of giving the building a decisive top. Buildings that stop abruptly at the roofline read as fragments of something taller. The eaves is an edge and it deserves the attention normally given to the entrance.

The edge of the room

Threshold is the edge at human scale, and it is where housing succeeds or fails.

What lies between the public street and a private front door, and how many gradations does it pass through? Street, shared ground, semi-private threshold, private interior — four steps, each legible. Housing that works has this gradient. Housing that does not has a door onto a corridor, and no amount of quality inside the apartment repairs it.

The pavilion makes the argument in its simplest form. Two-thirds of the building has no walls. There is enclosed accommodation pushed to one edge, and the rest is covered, open, and unconditioned — neither inside nor outside but a third thing, which is the part people actually use. A café operator briefed for summer trading now works year-round, because the covered edge is comfortable in rain, and rain is what Amsterdam mostly does.

The edge of the detail

At the smallest scale, an edge is a junction: two materials, two trades, two tolerances that do not match.

Timber moves and steel does not. Concrete is poured to ±15mm and glazing is manufactured to ±2mm. Every junction between them is a negotiation, and the drawing either resolves it or leaves it to be resolved on site by somebody with a grinder and a deadline.

The discipline is to decide in advance which element is dominant and which accommodates — and then to express the accommodation rather than conceal it. A shadow gap is not decoration. It is an honest admission that two things arriving from different factories will not meet perfectly, and a designed tolerance is invisible in a way that a mastic joint never is.

At the pavilion every connection is bolted rather than glued or welded, so the structure can be taken apart at the end of its permission. That requirement forced every junction to be visible and considered, because you cannot hide a connection you intend somebody to undo. The building is better for the constraint, which is generally how constraints work.

Where this stops being useful

A focus on edges can become a mannerism. There is a strain of contemporary detailing — everything floating, every element separated by a shadow gap, no two materials permitted to touch — that has stopped being about resolution and become a style, and an expensive one. Sometimes two things should simply meet.

It can also become an excuse for neglecting the middle. A building can have superb junctions and a bad plan, and the plan matters more. If the choice is between a well-resolved facade and a section that gets daylight into the middle of the floor plate, the section wins every time and it is not close.

The edges are where the care shows. They are not where the quality comes from.

What are you building?

Housing, a school, a clinic, a warehouse you've just bought. Start anywhere.

  • Maison des Tilleuls
    Harbour Commons
    Atlantic Workshop
    low angle view photography of a gray building
    Atlantic Courtyard
    House of Light
    Northbank Commons
    man walking on library
    Asteria Courtyard Homes
    The Listening Room
    Porto Passage
    Green Line Offices

What are you building?

Housing, a school, a clinic, a warehouse you've just bought. Start anywhere.

  • Maison des Tilleuls
    Harbour Commons
    Atlantic Workshop
    low angle view photography of a gray building
    Atlantic Courtyard
    House of Light
    Northbank Commons
    man walking on library
    Asteria Courtyard Homes
    The Listening Room
    Porto Passage
    Green Line Offices

What are you building?

Housing, a school, a clinic, a warehouse you've just bought. Start anywhere.

  • Maison des Tilleuls
    Harbour Commons
    Atlantic Workshop
    low angle view photography of a gray building
    Atlantic Courtyard
    House of Light
    Northbank Commons
    man walking on library
    Asteria Courtyard Homes
    The Listening Room
    Porto Passage
    Green Line Offices

Newsletter

Stay in touch

New journal pieces, projects reaching completion, and roles we're hiring for.

No spam. Unsubscribe anytime.

Newsletter

Stay in touch

New journal pieces, projects reaching completion, and roles we're hiring for.

No spam. Unsubscribe anytime.

Newsletter

Stay in touch

New journal pieces, projects reaching completion, and roles we're hiring for.

No spam. Unsubscribe anytime.

Create a free website with Framer, the website builder loved by startups, designers and agencies.