Inspired by the start of term at UCL, a new great cohort of students to teach and interact with, and the new module that I teach (‘Buildings, Organisations, Networks’ in the MSc ‘Advanced Architectural Studies’ – now called ‘Spatial Design: Architecture and Cities’), I have recently thought a lot about building typologies. So in essence, what makes a school a school, what makes a church a church, an office an office and what makes a shop a shop? How are these building typologies similar or different from one another?

At first, it seems obvious that a school must be a school because of its spatial layout and the structuring of space – classrooms, corridors, sports facilities, playgrounds, etc. Likewise, a church is a church because of its naves, aisles, colonnades and apses. An office is an office because of its arrangement of desks, meeting rooms, corridors, a reception, tea points, photocopy areas and watercoolers. And then a shop is a shop because of its display areas, window fronts, storage spaces, the cashier, etc. So it seems that building typologies are defined and easily recognised by their spatial arrangement of functions.

And certainly, we do recognise certain building types easily from their floor plan. Think of a traditional gothic cathedral, for example and we all have a clear picture of its spatial structure in front of our inner eye.

Floor plan of Notre Dame de Paris (Wikimedia Commons)

However, if we move away from the very specific architecture of the Gothic cathedral, and for instance look at basic early church buildings, the question at hand, i.e. recognising a building from its floor plan becomes a lot more difficult already, as can be seen in the floor plan of the Basilica Ulpia in Rome shown below.

Basilica Ulpia, Rome (Wikimedia Commons)

Of course, this was not only a church. In ancient Rome, the basilica was a multi-functional municipal building used as market hall, bank, exchange, court and a meeting point due to its traditional central location as part of the Roman Forum.

This would strengthen the argument that buildings used for a single function and purpose could be easily recognised by its spatial layout.

But does this hold? I felt like testing it, and my students (and colleagues at the Space Group of the Bartlett School of Graduate Studies, UCL) seemed good participants for an experiment like this. I therefore gathered floor plans of 18 different buildings (well, floors of buildings, to be precise), comprising 10 different building typologies (hospitals, libraries, museums, offices, research laboratories, schools, shopping malls, shops, supermarkets and universities), arranged them on two A3 sheets, roughly scaled to size and analysed according to their interior visibility relationship using Space Syntax methods. This highlights central areas with a high potential for usage in red and segregated areas with a low potential for usage in green and blue. In most cases furniture was taken out of the plans (apart from ceiling height furniture), and no labels or text was allowed of course. This is the result: lots of floor plans, lots of different shapes and structures, and actually very few hints to what they might be.

Building Typologies Exercise Sheet 1

Building Types Exercise Sheet 2

With informed guesses my colleagues managed to get approximately 50% right, although we all had a lot of fun, when our previous university office space, which was part of the sample and in which all of my colleagues would have worked for several months or years, was categorised not only as a university building, but also as a hospital, a prison or a home for the elderly. Maybe that also tells a lot about our previous workspace…

My students (a good majority of them with a first degree in architecture) identified 35% of buildings correctly, which I think was very well done given the complexity and difficulty of the task. Some of the most interesting results of this experiment included:

• The Research Laboratory in the sample (case C) was identified correctly by 65% of students.
• Museums, shopping malls, libraries, supermarkets and offices were identified correctly around 50% of the time.
• Hospitals (33% correctly identified), universities (4%), schools (0%) and shops (0%) were among the most difficult typologies to recognise.
• Obviously some building types are easier to recognise from floor plans than others.
• While one museum (case N – the MoMA) was 100% identified correctly, two other museums (case L – the Museo Nacional Rio de Janeiro and case F – Tate Modern) were mistaken for other buildings types to 100%, so obviously some specific layouts are easier to identify than others.
• Certain building types are commonly mistaken for one another, for instance supermarkets and libraries (large spaces with long shelves).

Some caveats of the experiment include a slightly biased and at times not very representative sample, the lack of definite scales and some inconsistencies of visual representation across the sample, but it certainly was and is an interesting experiment, which I am planning to repeat with next year’s students.

But why is it that we find it so hard to detect building typologies from floor plans? One of my favourite papers, ‘What do we mean by building function?’ written by my colleagues Bill Hillier, Julienne Hanson and John Peponis almost 30 years ago can shed some light on this. They argue:

“In general, buildings can be defined (…) as devices for making two kinds of interface:   one between inhabitants and visitors, and the other between different categories of   inhabitant. These interfaces are realised through some configuration of bounded,   convex and axial space, and some set of relations of integration and control. These   interfaces are what we mean by the global function of the building. It is this, we believe,   that we name – at once socially and spatially – when we say ‘school’ or ‘church’ or ‘hospital‘.” (Hillier et al 1984: 66)

So in essence, the churchness of a church and the schoolness of a school is defined by building function, and building function in turn is defined by the creation of an interface. In short, function is what people do in buildings. Therefore, it shouldn’t surprise us very much that we cannot detect building function from a floor plan, because a plan is devoid of people. So in the end it is people and their everyday activities, encounters and usage patterns that tell us what a building is.