Sie sind auf Seite 1von 29

Scaling Up and Down: Extraction Trials in Architectural Design

Author(s): Albena Yaneva

Source: Social Studies of Science, Vol. 35, No. 6 (Dec., 2005), pp. 867-894
Published by: Sage Publications, Ltd.
Stable URL:
Accessed: 08-01-2018 21:10 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

Sage Publications, Ltd. is collaborating with JSTOR to digitize, preserve and extend access to
Social Studies of Science

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
ABSTRACT How do architects imagine, see and define a distant object that
to become a building? How does it become knowable, real? To answer th
questions, I follow architects as they fabricate models and scale them up a
different rates of speed. Instead of being a logical, linear procedure for g
new object that becomes progressively more knowable, ascending from th
to the concrete, scaling is a versatile rhythm, relying on surges, 'jumps' an
By focusing on the most frequently repeated moves such as 'scaling up', '
scale', 'scaling down', and describing their cognitive implications, I depict
architects involve themselves in a comprehensive dialogue with materials
Their material dialogue takes into account dispositions, resistance, stabilit
properties that change proportionally with scale. In the scaling venture, t
alternative states of the building are simultaneously achieved and maintai
of being 'less-known', abstract and comprehensive; and a state of being
known', concrete and detailed. After multiple up and down transitions be
small- and large-scale models, the building emerges, becomes visible, mate
real. These scaling trials bring the building into existence.

Keywords architecture, building, design process, ethnography, models, re


Scaling Up and Down:

Extraction Trials in Architectural Design
Alhena Yaneva

November 2001, Rotterdam, In the middle of an architectural office, on a

huge table, various scale models of a building, parts and detailed variations
are installed: a mini-exhibition space lit by neon light; a solemn spectacle
waiting to be discovered by invited visitors. Reproduced in various material
samples, colours and shapes, models are maintained in this particular
arrangement during the whole process of architectural design (Figure 1).
'This is the Whitney project', Rem Koolhaas tells the visitors to his
office as they view a colourful assemblage on the table.1 These models
illustrate different facets of the building; visualizing scenarios, issues and
possibilities that have been tested. No single starting point triggering a
linear series of models or elements can be found, but this is not a chaotic
assembly of scattered leftovers from the conception process. What we see
on the different parts of the table are diverse concentrations of models,
intensities of detail, variations and images. Separated by different spatial

Social Studies of Science 35/6(December 2005) 867-894

? SSS and SAGE Publications (London, Thousand Oaks CA, New Delhi)
ISSN 0306-3127 DOI: 10.1177/0306312705053053
www. sagepublications. com

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
868 Social Studies of Science 35/6

The Table of Models'; Office for Metropolitan Architecture, Rotterdam. (Photograph:
Albena Yaneva)

intervals, they all form a network of points and passages presenting

different vantage points on the same building. They all expose (in a
particular geometric configuration) a stabilized state of the Whitney
After my first visit to the Office for Metropolitan Architecture (OMA)
on this early November afternoon in 2001,1 joined the Whitney team and
followed the architects' discussions as they worked on the project. Gaining
ethnographic access to this field required me to 'live' in the architectural
office for a while, confronting various visual enigmas. One of these enigmas
had to do with the rhythm of scaling. I decided to follow the small material
operations of scaling in order to make this enigma ethnographically de
scribable: in what follows, I will try to help the reader also to 'see' these
operations in order to understand what the designers do when they
conceive a building. By following the particular rhythm of scaling in
design, we will have the chance to observe the appearance of a building as
it emerges from the architects' hands: a building that is made knowable
and real as scales are shifted.
Scaling can be considered as an experimental situation in the sense
that it is subjected to constant and well-equipped observation of possible
consequences of acting on scale models; it is an apparatus for conducting,
recording and interpreting the results of manipulating selected features of
models. When conducting such experiments, architects scale up and down
in order to see what might follow; they do so either as an exploratory move
by probing in trial-and-error fashion, or as a systematic test aiming for an
intended outcome to be confirmed or disconfirmed. These tests aim at
probing parameters and realities connected to the building's particular

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
Yanev . Scaling Up and Down 869

mission.2 In such experiments with scaling, the tenuous and minute moves
with various tools and models, the intelligibility of materials and the
actions of the architects are all made observable. As I shall show, the
rhythm of scaling relies on procedures for partial seeing: scoping, rescaling,
extending and reducing the material features of scale models. Architects
attempt to resolve problems by using tricks of the trade such as stepping up
the scale, disguising and revealing various aspects, and inspecting and
overseeing those aspects. Scaling requires special equipment, instruments
and embodied routines for manipulating models, as well as meticulous
work with foam and paper for seeing and defining details. Through such
practices, a building can be conceived in thought and brought into exist
ence. It is generated through numerous techniques of projection (Blau &
Kaufman, 1989) and translation.3 The models are scaled and re-scaled, not
according to the architect's mind's eye (Akin & Weinel, 1982), but accord
ing to numerous material formations, practices and relations among archi
tects, consultants, models, cameras and images in a complex visual field.
Over the past 20 years, science and technology studies (STS) have
closely followed scientists, engineers and physicians in and out of their
workplaces; but architects have not been followed as their practices move
from the model shop to the panel presentation for the client, and even
tually to the construction site.4 Some recent STS research on design
practices has analysed visualization and emphasized its social and complex
dynamics. However, such research focuses mainly on engineering design
(Ferguson, 1992; Bucciarelli, 1994; Henderson, 1999; Vinck, 2003). A few
studies have dealt with design in architectural firms, but explored their
activities from a more traditional sociological perspective.5 In this paper,
the architectural office will be studied in the same way that STS has
approached the laboratory (Latour & Woolgar, 1979; Lynch, 1985; Knorr
Cetina, 1999). By following particular scaling moves as they run up and
down, I aim to expose the materialization of successive operations, and to
trace the developing appearance of the building. I will compose a story
about gradations (nuances of size and degrees of presence) in numerous
architectural objects; gradations through which architectural practices play
out on a battlefield full of unknown internal streams, orders and disorders,
flows and synchronization moves, polemics among architects, visual puz
zles, and attempts to resolve disputed states of affairs through visual
instruments and convincing images.
My sources about scaling are conversations among architects engaged
in building models for a new exhibition hall at the Whitney Museum of
American Art in New York (at the time, the design of the museum
extension was commissioned to Rem Koolhaas), interviews with architects
and a rather dilettante personal participation in model fabrications. The
present paper does not invite us to imagine an architectural office. Instead,
by meticulously reconstructing participants' discussions and actions, and
by depicting concrete manipulations with materials and scaling instru
ments, it attempts to bring us into the office and to follow the work of the
Whitney team as it conceives and designs a new exhibition hall.

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
870 Social Studies of Science 35/6

A possible approach to use for following scaling in action would be a

mathematical analysis of all technical steps in a scaling algorithm that
translates paper into foam. Instead of taking this approach, I will describe
some frequently repeated moves, such as 'scale up', 'jump', 'scale down'.
Their successive repetition and redundancy compose a rhythmic conduit
through which the building develops. That is how scaling is treated in this
study. A static notion of scale understood as a proportional relationship
between the world of models and the external real world (Boudon, 1971,
1992), or a metric relationship (Licklider, 1966; Dupire et al., 1981;
Boudon, 1999), is insufficient to grasp architectural scaling, nor is a
phenomenological point of view relevant, if it is conceived as a way to
understand scale as a fleeting subjective feeling of harmony, proportion
and composition (Orr, 1985). The reason for defining the scaling venture
as a rhythm is the fact that it designates an ordered variation in a series of
moves performed with different intensities and speeds. It is not a schematic
repetition, as compared with a process or flow, but develops as a movement
that can repeat itself in a regular beat: in, up and down; strong and weak;
long and short sequences. A rhythmic sensation of motion comes out of
these successive moves.6
Following science studies and cognitive anthropology, we can assume
that much of the internal organization and operation of architectural
cognition can be directly observed in the activities of scaling as they relate
to the social and material environment of the architectural office.7 Swept
up in a rhythm, progressing in free flowing irregular up and down move
ments, architectural cognition involves interactions among architects, scale
models and scoping instruments. At the start of this process, architects
need to conceive of the distant and unfamiliar object (the building) in a
way that enables them to define it with precision and to enable its
realization; they begin with fuzzy approximations (small scale models) of
this object fabricated according to few known parameters, and then they
are supposed to obtain new information - even though they do not know
and cannot yet understand exactly what they need to know - in the course
of their practices. The tiny material operations of'scaling up', 'jumping the
scale', 'rescaling' and 'going down in scale' enable architects to think of the
building and to gain new knowledge about it. Knowing through scaling is
an integral aspect of architectural practice.8 Regardless of the particular
data obtained through scaling, two different presentational states of a
building are maintained simultaneously, so that it always exists as a little
known, abstract and fuzzy object, and at the same time a well-known,
concrete and precise object, instead of progressing in a linear fashion from
a state of zero information to a completely known and defined object.9 The
particular material arrangement of models on the table (Figure 1) corre
sponds to a stabilized, frozen picture of many intermediary presentational
states. Paradoxically, what results is that architects do not convert in
determinate, complex and incoherent information into determinate and
coherent objects.10 Designing a distant building requires knowing it more
and knowing it less at the same time; taking account of it with small and

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
Yaneva: Scaling Up and Down 871

large scale models, with both abstraction and precision. The final building
is never present in any single state or model, but in what all of them
together project. That is why the building is a multiple object: a composi
tion of many elements; a 'multiverse' instead of a 'universe'.11
Some studies on engineering design treat the designed object to be the
result of a social process involving lengthy negotiations and contradictory
debates among participants, and argue that its final shape depends on
various modes of consensus (Bucciarelli, 1994). Accordingly, what comes
first in the process is a subjective agreement among participants about the
meaning of the artefact designed, and its realization is triggered only after a
shared vision is gained. Like Henderson (1999) and Law (2002), I
consider models as objects over which negotiations and conflicts take place
and treat architects as being implicated in a dialogue with concrete
materials, spatial figures, proportions, dispositions and shapes. That is, a
'reflexive conversation with the materials of the situation' (Sch?n, 1985),
rather than a question of inter-subjective agreement. In this conversation
designers make numerous moves that have unintended effects, with unex
pected problems and potentials. In their design meetings, architects dis
cuss concerns about scoping and rescaling the models; they 'lend' their
bodies to many visual instruments, which enables them to see and experi
ence the internal space, 'guided' by the inner logic of the foam construc
tions, and 'influenced' by many previous choices. They are also 'con
strained'12 by numerous requirements (client demand, city politics, site
specificity, users' expectations) and 'led' to solutions. Materials, scoping
instruments and new knowledge 'talk back' to the architects, and they are
prepared to listen, thus triggering reinterpretations of interim results.
These idioms ('talk back', and so forth) are commonplace in architects'
stories about different projects at OMA.13 Following architects' commu
nication with such objective materials allows us to gain access to forms of
cognition they deploy in the course of design work. Here, more than in any
other context, architects need to make clear to one another what it is they
do - what emerges from their hands - when they engage in design work.

How Does Scaling Begin?

Our attempt to follow the work of scaling is significant, not because the
scale models represent the evolution of architectural ideas or retrace a
chronology of material steps making up a design. It is important for
another reason: architects are implicated in the making of composite things
- the models. Anyone who has visited an architectural office would see that
models are important tools in architectural design.14
Models have a life of their own in the architectural office,15 together
with a number of more schematic presentations of the building such as
diagrams, sketches and technical drawings. All these visual representations
are not meant to transform a specimen into observable, standardized,
mathematically analysable and reliable data, as numerous science studies
have shown for science (Latour & Woolgar, 1979; Lynch, 1985; Galison,

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
872 Social Studies of Science 35/6

1995). That is, models cannot be treated as inscription devices that

visualize invisible substances.16 Instead, their purpose is to gather a num
ber of things - human and non-human actors, and their concerns, require
ments and disputes - and to 'accommodate' them into objects that can be
subjected to design experiments. By making models architects invent
objects, which have the properties of being composite and mutable.17
Instead of attributing a scrip-tion-like term18 to models, they will be
considered here as peculiar compositions of things that are manipulated in
the scaling process, and whose transformations cumulatively lead to the
building. Generated as physical approximations of the building, along with
affording the calculation of various types of numerical data, they aim to
realize the building through a joint venture.
Physical models are used in OMA as a major visualization tool for
project presentations. They facilitate communication, serve as 'social
glue'19 among architects, experts, clients and publics, and organize the
design process in the office and in networks of outside consultants and
experts.20 Moreover, as compared with many other architectural firms in
which models are built only at the final stage of a project, OMA fabricates
models at every step of the design process, along with two-dimensional
representations. These models are important tools for shared cognition:
architects think of the building by modelling, by cutting foam and paper
and using various scoping techniques. It is not a free intuitive creation of a
building shape generated 'out of the blue'. The first small models of the
Whitney building are produced according to few important 'con
straints'. These constraints are mainly negative - 'not to exceed the zoning
envelope', 'not to demolish the brownstones', 'not to damage the adjacent
buildings' - in the way they place limits on the process of experimentation
that we can witness in the office of Rem Koolhaas.
At the very beginning of the modelling process architects tend to enlist
the variety of things that have to be 'accommodated' by the Whitney
models: site location, programme, volume, city fabric, district fragmenta
tion, circulation, mechanicals, zoning envelope, artists' expectations, his
torical landmarks, museum philosophy, art display and community con
cerns. These constraints include requirements from the client that
architects take into account as 'givens', but also parameters established by
the architects themselves.
The shape of the first Whitney models is generated in response to all of
these constraints. For example, a small concept model with barely visible
figures takes into account the adjacent buildings, the tiny slot allocated for
the site, the eclectic features of New York city fabric, the dense network of
local districts, the zoning fragmentation, the variety of building heights, the
marks of history, the city politics and the neighbours' relationships. In the
model, heterogeneous parameters are fitted together so that the building
appears with a distinctive footprint, elevation and mechanical structure.
Modelling welds together of all these elements, no matter how diverse, into
a new gathering. Models are not projections or anticipations of the

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
Yaneva: Scaling Up and Down 873

Kunl? and Sho are discussing the shape of some small scale 'study models'. (Photo
graph: Albena Yaneva)

building; rather, they are new compositions shaped according to multiple

Once accommodated in a model, the site parameters are temporarily
forgotten. In what follows, we depict a peculiar versatile up-and-down
scaling flight as momentarily detached from the parameters. As the archi
tects lead the client toward a better visualization of the emerging building,
the reader will be guided through the puzzling visual procedures, with the
hope of finally 'seeing' a building.

Scaling Up and Down

Scoping in the Small Model
Zooming in on the table of models, one can notice the smallest proto
building placed there. A walk through the architectural office will allow us
to discover many similar small-scale models. Let us follow one of them
from the table of models, through the working tables, into the architects'
hands (Figure 2).
Since little is known about the new building, the small study model has
no 'real details'. The 'little' knowledge includes the parameters according
to which the model has been fabricated. Made by hand according to these
few restrictions, the small model is easy to shape quickly; it is a less precise,

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
874 Social Studies of Science 35/6

sketchy version of the building. Why do architects spend hours and hours
looking at this small piece of foam, turning it in their hands, meticulously
examining its corners and openings, positioning it in relation to different
objects at hand, passing it to each other, and inspecting it during disagree
ments and disputes? What are they able to see with its barely visible
features? What is this piece of foam telling them? How are they able to see
it? What is it that guides them to the building?
To answer such questions, I suggest that we follow the team as they
fabricate a large-scale model of an exhibition hall in the extension of
Whitney Museum of American Art; starting from that small model of the
building, modifying it, and subjecting it to numerous visual puzzles.
Architects use two parallel working tables. Various small models and
specific details are scattered on the first one, while a huge model under
construction is installed on the adjacent table. 'Crowds' of architects,
paper cuts-outs, drawings, foam pieces and instruments are gathered
around these scale models.
Architects use a particular instrument, called a modelscope,22 to look
inside the small model in order to see things that cannot be observed
directly from outside. To understand scaling and its cognitive implications,
it is important to consider how this instrument works, and what forms of
thinking are associated with it. When this miniature periscope is inserted
into the small model, it can function as design tool by providing visual
access to selective and realistic eye-level images. By doing so, it can
generate new information about the building and can enable architects to
conceptualize it with more detail, clarity and precision.
In this instance, the design task is to determine the position of a huge
escalator in the interior space of the model. Kunl? changes the place of
the escalator in the large-scale model, then asks the other members of the
team, 'Do you like it?' Nobody answers. He takes the modelscope and
the others keep encouraging him, 'scope it, scope it!' Kunl? moves to the
adjacent table, and then switches on the light source and adjusts the level
of illumination, as required for a comfortable viewing position. Then, he
carefully inserts the modelscope to inspect the small-scale model, and
checks the function of the focusing control (from 5 mm to infinity
according to a field of view of 60? for the small model 040) and adjusts the
orbital scan to achieve the required view. Kunl? now looks inside the small
model, and a deep silence follows (Figure 3).
As Kunl?'s eye inspects the interior space of the model, the eyes of the
others are looking in the direction of the scattered things around the
model, without fixing their glances. They are waiting for their turn. While
anticipating Kunl?'s reactions, they encourage him, 'ouyaou, ouyaou', as if
they were able to see inside the model along with him; as if they collectively
shared the result of his inspection. Kunl? adjusts the light guide connector,
situated inside the instrument's handle, in order to regulate the light. Then
he sets the orbital scan again and starts rotating the viewing direction with
regard to the handle, through a total arc of 360?. An orientation mark in
the image indicates the direction of view. He sees something. He says:

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
Yaneva: Scaling Up and Down 875

Kunl? is using a modelscope to inspect a small model. (Photograph: Albena Yaneva)

'Here is the northern part of the hall. Ouyaou, I see it, where are you,
stairs? ... I see a staircase, the two pieces of Hopper around? Mmm, here
it is then (a space for the elevator).' While Kunl? communicates his
impressions of what he sees with small gestures, the others also begin to
express reactions. As time goes on, their silent impatience is reinstated.
While the architects from the Whitney team gather around Kunl?, they
talk with tiny particles in the situation, and with the data obtained in the
scoping venture. Rather than coming to agreement, often discussed in
studies of design as preceding the artefact fabrication, the scaling team
engages in a dialogue with a dynamic assemblage of objective materials:
dispositions, objects they see inside the model, spatial transitions, material
properties of the foam, proportions and shapes. Scaling together means
scoping the models, entering into conversation with their barely visible
figures, and discussing with the team what is seen. It turns out to be as
important to design as the drawings and scale models themselves
(Bucciarelli, 1994). The experience of scoping the small models is very
mysterious. The scoping architect remains with a sense of confusion since
he has to see something on his own, being unclear just what he is supposed
to see. Only after numerous adjustments of the instrument are Kunl?'s and
the modelscope's eyes connected, able to see at that particular moment the
shady interior of the small-scale model of the museum exhibition hall. The
others can 'see' only partially by sharing collectively Kunl?'s experience.23
They find themselves in a state of impatience, just as Kunl? was a few

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
876 Social Studies of Science 35/6

moments earlier, before taking the instrument. As I join the group of

anxious architects waiting to see the model inside, Kunl? explains to me
what happens:
The modelscope just gives you a view that is like the scale of that model.
So, you get to express the space at that scale. It gives you the opportunity
to move around spaces you ordinarily can't get into and to see how they
look. It's a very useful tool ... We see at very small scales, and we are
getting to have tiny looks and private sets. We are able to see how space is
inside. (K1102)
Now I know why Kunl? is so slow in using the modelscope, how exactly
the others manage to see partially, and why they are so impatient to see
what he has seen. Kunl? not only is scoping the spaces, he is moving
around them, crossing the threshold of the exhibition hall after having
walked on the stairs, looking to find a suitable place for the escalator. The
modelscope as technique allows him to bridge the scale barrier, to reduce
his own human size and to think of the building by transporting his eye
directly into a model space. Minimized to the scale of the tiny model, he is
exploring these microscopic spaces like in Gulliver's travels,24 he 'enters'
the spaces and experiences them. That is how he is able to gain new
knowledge about the interior space of the building; knowledge expressed
not in facts, codes and numbers, but in terms of dispositions, arrange
ments and spatial transitions. Only after a scoping venture are Kunl? and
the members of the Whitney team able to 'know where' to put the
escalator, instead of'knowing that'.
After having 'walked' through the inside of the small model, Kunl?
puts the modelscope aside. Thus, his 'travel' is finished; he is again in the
architectural office, next to the small noisy group of impatient colleagues
eager to have the same experience. He explains to them that the position of
the red escalator has to be changed. Then, to my surprise, instead of taking
the modelscope, the group of impatient architects moves to the huge
model and start examining the large-scale interior of the exhibition hall,
the 'same' interior that Kunl? had seen moments earlier inside the small
scale model. Numerous material changes are performed on the model.
How does a semi-'blind' and constrained25 movement allow Kunl? to
'see' inside the tiny model and to make it visible for the others? How is that
he knows after inspection with the modelscope that the position of the red
escalator has to be changed? Where are the traces of this inspection kept
and inscribed? How is his knowledge articulated and cognitively shared by
the team? Why do architects move from one table to another one, im
mediately after the endoscope inspection of the small model is completed?
Why do they constantly go back to the tiny eye of the modelscope instead
of relying only upon their naked eyes? What is it that they take with them
while moving from the monocular inspection of the tiny model to the
binocular examination of the huge scale model? What travels across the
tables in the office? What passes among scaling actors? These questions
guide me in understanding the cognitive dimensions of the scaling venture.
This parallel work in different scales needs to be explained.

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
Yaneva: Scaling Up and Down 877

Since there is no visible trace from the act of scoping in the small
model, architects rely only upon the visual experience of the one user of
the instrument. His experience is not absolutely lonely, but actively shared
by the team. A scoping ou?6 movement follows: the knowledge about the
escalator gained by Kunl?, expressed in spatial dispositions, is immediately
transposed onto the huge physical model and discussed by the team. That
is how the use of the modelscope triggers numerous material alternations.
The escalator is placed in the middle of the exhibition hall (not in the
northern part of the building) in a way that enables museum-goers to enter
the gallery immediately; thus the space usually used for circulation is now
designated for art display. The decision to move the escalator is triggered
also by the museum requirement to have 'more space for the permanent
collection', as well as by the users' expectation of a larger building, the
architect's ambition to maintain historical continuity by providing similar
principle of circulation with the one of the old Whitney building, and by
the museum decision to accommodate art in support spaces.
Gathered around the huge scale model, architects discuss the new
escalator position and repeatedly rearrange its interior. Every new disposi
tion is checked out again with the modelscope. A member of the team
takes it, sets the integral light guide, adjusts the working length and
direction and the field of view, and looks inside, inspecting the spaces.
Then, a team member moves to the next table and suggests a new physical
adjustment in the large-scale model of the exhibition hall. After each new
arrangement on the large scale, architects go back to see it in the small
scale model; they trust the monocular image of the internal space obtained
with the tiny eye of the modelscope. Then, they return to the large-scale
model to make it binocular and to perform some adjustments.
Scoping in and scoping out the tiny model, the exhibition hall of the
Whitney is made bigger and bigger, allowing architects to see its 'inside
The bigger it gets, the more details and more interior you see, and you
start really looking at the way a surface meets a floor, or the way you detail
something around the window, and it gets much more refined as it gets
bigger. (Cl 102)

With every move of the modelscope, architects acquire more visibility to

the particular details of the exhibition hall, and get more data about it.
Although they can see the escalator inside the small model by scoping in
with the modelscope, they need to see it again in a more open space; that is
why they scope out> fabricate it with paper, and install it in the huge model
of the exhibition hall. By doing so, they transform the ephemeral experi
ence of seeing with the modelscope into physical arrangements of the
larger foam model, thus translating a lonely visual experience into a
collectively accessible material space. That is why every instance of model
scope inspection is followed by a series of operations with paper, scissors,
foam, cutters and paints.27 Thus, the escalators and the transparent stair
case are made out of paper and installed. Moreover, even miniature

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
878 Social Studies of Science 35/6

The interior of the exhibition hall as seen in the large-scale model. (Photograph:
Albena Yaneva)

paintings from the museum collection are placed on the walls of the model
and white plastic figures - future exhibition visitors - are painted in red;
that is, they are rendered more visible and real. These operations enable
architects to accurately arrange the interior architecture, by numerous
hand movements, to shape and produce its space (Figure 4).
The scale shift makes the model larger and more accessible for the
viewer's body; that is, it becomes large enough to simulate an interior eye
level view. It provides more visibility to concrete interior details. Architects'
corporal operations require less effort to see into the building; compared
with the scoping operation, their postures are less stooped, tense and
uncomfortable when assuming the appropriate viewing position. This
physical space manufactured with foam and paper becomes an object of
collective experience, which is visible for many actors at the same time
(Figure 5).
Only by following the material re-arrangements of the big model can
we become aware of what each architect has seen while inspecting the tiny
model with the scoping instrument. What a single architect sees is shared
with the others and changes the cognitive properties of the team.28 The way
he imagines the building, now, is made visible for the other architects by
the tentative movements of his hands that repeatedly change the escalator's
placement. Only when we follow architects' hands as they point to, and
work with, the huge model to transform its composition - taking the same

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
Yaneva: Scaling Up and Down 879

The Whitney team working on the spatial arrangement of a large-scale model.
(Photograph: Albena Yaneva)

paper figures, manipulating the space - are we able to see how architects
think together.
However, scoping in and scoping out describe two different settings
where other real agencies than humans (with their intentions and isolated
individual minds) take part and shape complex metaphysical imbroglios:
stairs, escalators, foam materials, foam cutters and recalcitrant models. In

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
880 Social Studies of Science 35f 6

the two settings of partial visualizing into the model and through the
models, cognition is a complex social phenomenon, distributed among
individuals, model scope, team, visual puzzles, and materially shape-able
model. The passage from the setting 1 in which a single architect scopes,
many others react and subsequently lend their bodies to puzzling scoping
procedures, to the setting 2 in which many architects stand together at the
side of the model, inspect it with naked eyes and collectively transform its
material body, is not a transition from individually to a socially structured
experience, from a self-contained disembodied technology of cognition to
a collectively and publicly shared one. These settings differ only by the
distinctive way of distributing the action of scaling.
The task of positioning the escalator is organized in such a way that
architects use their visual language more often than their verbal expres
sions; therefore the main resource for communicatively mediating the
performance is also predominantly visual. No single architect occupying a
central position in the process supplies verbal directives. Instead, the
architects communicate in a centre-less heterogeneous network that in
cludes the materials at hand. When viewed as the externalization of an
individual cognitive process, the gestures of the impatient architects be
come entangled with those of the person inspecting the model, and provide
an additional network for mutual actions among members of the scaling
team. That is what makes the inspection of the model always a collectively
shared experience.
By following the scoping in and out procedures that architects deploy
on a daily basis to see a building, we can find two main actors simultane
ously present in the architectural work: the small-scale model and the
large-scale model.

We did small models with different cuts to see how it looks. But since the
small model is too tiny to think about the possibilities of the internal
space, we have to build the big one. People think that it's a lot of effort to
build the huge models but I am glad it works ... And also, we weren't able
to resolve the circulation problem in the small model. I did a box, a small
jewel box, and Rem liked it, but it wasn't sufficient to resolve the
circulation problem. So, we shifted the scale. (SI 102)

If the small model of Whitney is undefined and abstract, deploying rough

figures and approximate relationships, the large-scale model is meticulous
and enriched with more data and concrete details. While the former can
'evoke things and make broader assumptions', the latter visualizes sizes,
shapes and precise positions. The details come out only after numerous
repeated procedures with the small model and the standardized paper and
foam materials. Since small models lack sufficient visibility of concrete
details (such as transitions, escalators, thresholds and stairs), architects
scale-up to define and clarify more aspects of the building interior. Such
repetition with scale variations is how working with the small model leads
to detailed relations incorporated into the large one. Scale variations do
not change in a random way, nor do they follow strict metric rules. The

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
Yaneva: Scaling Up and Down 881

scale moves 'up' as architects develop a larger gathering of things. Conse

quently, the large scale model is more powerful, not because of an inherent
superiority of size, but because it has the ability to capture more para
meters and concerns, to sum up more requirements and limitations, to
reflect more details, corners and finishing, to enrol more viewers, to enable
more bodies to gather around it, to mobilize the public awareness better, to
provoke more violent disputes or to trigger more unpredictable actions.
That is why the small and the large scale do not differ only in size, in some
neutral or absolute sense, but in their distinct capacity to capture heteroge
neous actors in a model.
From the very beginning architects do not understand what designing
this particular escalator position means. That is why the artistry of thinking
architecturally of space seems elusive and mysterious, as though burdened
with epistemological paradoxes.29 Without knowing the spatial features
they are looking for in the new (or, rather, not-yet-existent) exhibition hall,
or what exactly they need to know in order to be able to conceive of those
features, architects take a plunge into the scaling circuit and rely only on a
few stable parameters. Thus, the fundamental features of the building are
grasped only in the process of doing - by scoping in and scoping out,
architects gain knowledge about the building. The modelscope provides
them with a direct30 access to an unknown (and sometimes disputed) state
of affairs at small scale, supplying resolutions of the particular design issues
that are then brought 'up' to a larger-scale. Two arrangements of foam
models are kept on two adjacent tables in the office. They account for two
distinct states of the building. One table contains tiny fuzzy and abstract
models, which present a state at which little is known of the building.
Fewer actors are being mobilized in the model, and the modelscope allows
architects to gain more information about it. A second table, situated
nearby, contains larger and more precise scale models of the same build
ing: paper and foam figures, cutting instruments, glue and drawings. This
table provides a distinct presentational state of the building - a state at
which more is known about it and more actors have been gathered by it.
These two tables are part of a rich network of mutual representational
dependencies. Each borders the other and is part of a continuum through
which the scaling venture takes place. The states of 'knowing less' and
'knowing more' of the building are simultaneously maintained within the
cognitive unit of the Whitney team.
The table of models described at the beginning of this paper is not just
a peripheral detail of the office interior, randomly chosen for the purposes
of an introduction. Tables form an important environment that organizes
the team's cognitive activities in the Office for Metropolitan Architecture.
It is impossible for architects to imagine the building without all the
models and try-outs on the tables. All changes are made with the materials
kept on the tables, as the building emerges out of the multitude of
presentational states. Just as they make the Whitney team's cognitive moves
visible, the models also render the building accountable: exposing options,
possible scenarios, failures and decisions. As architects move from one

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
882 Social Studies of Science 35/6

table to another, they pass from the small model to the large one, from the
tiny detail to a larger spatial arrangement of the exhibition hall, from the
disposition of the escalator to the overall circulation principle, mechanical
engineering and the philosophy of artistic display. That is, they move from
a small, stabilized composition of things towards a composition of a larger
scope, with greater cognitive and representational power.

Speeds of Scaling Up
Architects use the expression 'jumping up the scale' to describe the move
of passing suddenly to a much larger scale and a larger gathering of

We started from the small ones, and I jumped up the scale. It's dangerous
to shift to the big scale. If we start from the big one, we will lose our
concept. If we start from the huge model, it's dangerous because we will
be lost in details. (SI 102)

The 'jump' is rapid - an almost impulsive and radical shift in scale, not a
slow and gradual one, and so it can become risky. The 'jump' can be
dangerous, according to Shiro, because as architects go into a more refined
version of the building, they risk 'losing' the coherence of the small model
- the main features of the building (the so-called 'concept'31). The 'jump'
also can mean that suddenly knowing more about the building can make it
impossible to maintain a 'knowing-less' state. Its logic can be dispersed in
numerous practical details; 'disperse' meaning that when the particular
elements are more visible and articulated on the large-scale model, the
main idea - those less well-defined, but key, features of the building that
make it function - can become lost. What has to be retained when passing
suddenly from the small to the large scale, is the consistency of the whole
assemblage, representing a state of the building when only 'few things are
known'. In addition, what is retained is the possibility to return to a smaller
scope of gathering.
'Jumping' the scale can be a risky move, because it interrupts the
gradual process of slow-scale doubling.

Of course there are some times when you 'jump'. In the Whitney project
we did such a thing. So, we do something and then, we go to a really big,
big scale to show some details that are really important. For example, for
the scheme 'A' of the Whitney project, it was the windows' shape, and we
wanted to show how it looks with the colour graphs, and we did a really
big model - 1:25. These are really big models. Like a person, like that...
(he shows a human size model with his hand). And you can see how the
detail of the glass is, and then we went back. Sometimes it's important to
do these jumps. But usually we are going up slowly. (El 102)

The 'jump' accelerates the visualization process. It allows one to gain

better visibility of internal displays and spatial arrangements, to inspect
them with precision and foresight. In the 'jump', materials and dispositions
can 'talk back' to architects. When they step back, they reinterpret what

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
Yaneva: Scaling Up and Down 883

they see or reframe the problem to be solved. By doing so, architects

constantly go down to refine the small model instead of simply progressing
slowly towards a bigger and bigger version of the building, increasing
precision, and attaining more and more knowledge about it.
This 'jump' is reminiscent of the movement of'partial seeing' with the
modelscope inside the shady space of the small foam model. The 'jump'
permits sudden visibility of the higher scale, just as the modelscope
provided Kunl? with the possibility of directly seeing the interior space at a
microscopic level. Both of these moves supply zooming views of particular
spatial arrangements: if the modelscope allows virtual travelling through
the spaces, then the 'jump' to the higher scale allows an architect to peer
into a human-sized model, a 'doll house'. If the modelscope goes down
into the tiny unreachable space, and makes its experience visually possible,
the sudden scale 'jump' goes up, to the building, and makes it possible to
experience it physically.
In the move of scaling up there is no reference to the parameters
according to which the first models of Whitney have been fabricated -
parameters such as existing site conditions, adjacent buildings, city fabrics,
urban density and district fragmentation. Thus, the upward move produces
a double detachment - from these parameters, and at the same time from
the small-scale model. The large model is brought into existence by
reference to the small one, but not all of the initial parameters are
incorporated into its fabrication; the small model points to the large one.
In this way, there is a particular moment in which models refer only to
each other and trace a circular trajectory (instead of adopting an external
factor as a centre of meaning). This circularity is important, as it provides
possibilities for re-examining again and again the different presentational
states of the building before further development and definition.

Scaling Down
Reaching finer details through scaling up does not naturally guide archi
tects, slowly and linearly, to an ever-larger version of the building, where
more and more becomes known about it. Scaling up is immediately and
reversibly followed by scaling down.
When all the escalators of the new exhibition hall of Whitney are
decided upon, constructed in paper and painted red, they are definitively
placed inside the large-scale model of the exhibition hall. Then, the
architects return to the adjacent table to translate the changes that have
been made in the large model onto the small one: tiny red paper escalators
are carefully manufactured by Torsten and Shiro and placed in the small
scale model so that it is revised with new details and becomes a jewel-like
replica of the huge model. However, the small scale of the model doesn't
allow complete translation of all the details from the large model. Nor is a
complete reproduction needed. Some figures are simply sketched onto the
small one: the escalators are indicated as red endings, and the elevator is
indicated by a transparent Plexiglas box in the middle of the hall. In this

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
884 Social Studies of Science 35/6

way, the small and large models are mutually revised and enriched with
new details from latest developments in the project. The small-scale model
is then closed and placed back on the table, shaped proportionally and
adjusted to other visual representations. It is made compact and easily
movable from one place to another in the office, and from the architect's
office to the client.
All these briefly drawn operations are aimed at scaling down the
model. Scaling up scatters the original unity of the small-scale model,
while scaling down incorporates a patchwork of the complex scattered
details back into that unity. Scaling down is performed as a skilful arrange
ment of these disparate pieces. To achieve this patch-up, a specific tiny
movement of 'taking down the change' is realized. We can see this move
ment in another design task, which consists of determining the shape and
the position of the windows in the new Whitney extension.

We had this idea of the windows' shape and particular position, and we
wanted to see whether it works well, whether it's really good; because in
the small model we liked it, but we weren't sure if it will look good in the
big one. It wasn't only that we weren't sure that it will look good, but we
weren't sure that it would look convincing. So, we wanted to convince
other people that it was looking good. So, we built the big model, and
then we took this detail and we brought it back to the small model, and we
said: 'Okay, that's how we are going to determine it.' (El 102)

With the large model, architects can see whether the windows of the new
Whitney building 'work well'. Every part of the building has to be inte
grated with numerous other interior features, such as lighting, air
conditioning, circulation and infrastructure, mechanical structure, mate
rial properties, as well as with architects' worries, clients' concerns and
users' anticipations. As the windows' position is being defined on the large
scale model, a new adjustment, a new 'good fit'32 among the various
elements is obtained, and then translated to the small model. Once
brought back to the small model, it is then pushed again towards the large
The larger and more differentiated model does not differ in
evolutionary fashion from the small one; rather, it is a tool fo
better, gaining new knowledge, enrolling more actors and refin
small-scale model. Although it is a mediator in the scaling proces
final goal, it is not an ephemeral visual device. It is kept on the
models' along with numerous small-scale models, drawings and c
and foam and paper try-outs. Although stabilized in a given shap
them is completely defined, and any of them can be materially
thus triggering a chain of modifications.
As the design process develops, the scales are shifted and new
the building are gained:

We work on a model and a drawing at the same time, sometime

drawing will tell you more than the model and you go back and f
between the two. And then, you go back to the model that tell

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
Yaneva: Scaling Up and Down 885

something different and you have to change the drawing. And I think it's
the same for the large scale and the small scale. If we get further on in the
process, in design development, we know the shape, we know where the
floor levels are, we know where the windows are, but then you start to look
at more interior spaces and you might do a much larger model which is
proportional to the space. But that may affect the smaller one. And you
might say this window has to be like this to get this kind of light and that
means changes, and we have to take it back down and to see how it looks.
So, it's back and forth between the scales. (Cl 102)

As architects shift scales they enter into dialogue with objective materials,
far from any mental models (Gorman, 1997), that compel them and 'tell
them more' about the building, offer resistance, opposition and set up
tensions within. By doing so, they acquire more knowledge about shapes,
dispositions, locations; again, this is not knowledge of facts, but rather
knowledge about spatial transitions, not 'knowing that', but 'knowing
where'. In the translation from the small to the big, a special connection is
maintained between the two models that makes it possible for changes in
the large model to 'affect the smaller one'. Architects 'take the changes
back down' to the small model and update it. That is, more data are being
transmitted to the small model, but always schematically, so it can account
for an abstract and broad-spectrum method for presenting the state of the
building. Moving up and down in scale lets us discover two hologram-like
faces of the building: one small, vague and data-poor, the other large,
detailed and data-rich; being maintained as such, they make it possible for
the building to emerge in the architectural office.
Models are considered as small and large, respectively abstract and
concrete, as they treat compositions of things in a rather different way.
While the large model closely deals with the things - recalcitrant material
properties and adjustments - the small one stands apart from them. No
effort of translation is needed to understand the position of a window, an
escalator or a plug in the large model of the exhibition hall. However, the
meaning of the small model can be grasped only by calling to mind a few
evocative features of the building, and tracing out connections between
them. While the small model, as a first approximation of the building, has
the purpose of facilitating knowledge, inquiry and speculation, the large
model is associated with practical concerns. Therefore, since the small
model is employed simply as a means to encourage more thinking, it is
considered abstract; as the big model is used as a means to define figures of
the building beyond itself, it is a concrete presentation of the building.
However, the development of the practical cognitive power of the large
model does not weaken the abstract properties of the small one. Made for
the sake of knowing more of the building, the abstract model is pushed
towards the large one, in order to facilitate concrete achievements; regard
less of what these achievements turn out to be, the small model remains
abstract, as a tool for defining and perfecting the building.
Detached from the site parameters, the small- and the large-scale
models mutually inform each other and are simultaneously modified.
Jointly replicating and referring to each other in their redundancy, together

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
886 Social Studies of Science 35/6

they constitute a circuit: when the small model is no longer needed

because its job has been completed, it is scaled up and transformed into a
large one; when the large model accomplishes its function, it is necessary
to return to the small one. In this circuit, one can observe an important
degree of abstraction from the building 'programme';33 the building is
rendered diffuse, nearly atmospheric, and mundane; it is lost in transit.
Likewise, some problematic issues are dissolved in the scaling.34
By defining a multitude of technical details of materials, proportions
and sizes, and by supplying numerous 'convincing images'35 of the build
ing, every scale shift reduces uncertainty about the future building. It helps
architects avoid possible failures that can occur when a sudden scale up is
performed at the end.36 As the scale goes up and down, architects witness
that they are dealing with a 'defined' building; a building that is simultane
ously 'less-known' and 'well-known', abstract and concrete. The continuity
of scaling activity across settings relies on the flexible variability in their
structuring, and builds a continuum in which overall and detailed, abstract
and concrete are no longer distinguishable.


What Scaling Attains

We have seen that models 'jump' from scale to scale, without referring to
the site parameters as a defining reality or a transcendental centre of the
scaling enterprise. Thus, the small and large models exist together, guiding
the architects from one to the other, and crystallizing in a circuit. In this
circuit, there is no stable distinction between small and large, real and
What is visible in the circuit is a double movement in time. There is
not a chronological succession of past and present images of the building.
This movement can be defined only according to an actual present, from
which it derives in an absolute and simultaneous way through the process
of scaling. The small- and the large-scale models are correlated, but not in
simple relation of past and present versions; instead, they are simultane
ously connected, with each following the other. The past co-exists with the
present in a perpetual recurrence: the present state of the building follows
from an immediate spatially adjacent past; the past state is the particular
present state of the building that existed a few minutes ago, at a different
scale. As the present state of the building proceeds, the past state is
preserved on the adjacent table; a small model emerges into a large one, a
large model retires in the small. The temporal sequence is not chrono
logical, in a linear sense. Time is recreated and reinvented through a circuit
of scaling, each moment of which refers back to its, simultaneously
present, past states. This contrasts with the chronologically successive
moves of an evolutionary design process.37
Each move is tentative, and the circuit does not run in a senseless
fluctuation, as a game having no end other than itself. The scaling does not

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
Yaneva: Scaling Up and Down 887

move in a random and chaotic way, nor is it a matter of pure routine; it has
definite cumulative effects. Something new emerges out of the circuit: a
projected reality - the new building. This reality becomes visible in the
redundancy between 'knowing-less' and 'knowing-more': abstraction and
concretization, idea and multiple pragmatic details. It emerges as scales are
shifted between small and large models; one pushes to the other in a long
lasting game, and all of them make the building happen. Thus, scaling up
and scaling down are not successive moves, but parallel states, each
containing the other and referring to it. Instead of emerging in a propor
tional relationship to site parameters with a definite referent or 'content',
the building is defined in scaling trials; as it passes through these trials, it
becomes more and more visible, more present, more material, real. 'Scal
ing' is not a way to fit into reality; rather, it is a conduit for its extraction.
Scaling implies seeing in different scales, through a variety of presenta
tional states. The building is simultaneously present in all of these states: it
appears as less-defined and well-defined at the same time. Architectural
design develops, at a given moment and for a certain span of time, through
a circular generative regime instead of a linear process of varying possible
designs and selecting solutions, which subsequently generates the building
- a process of 'punctuated evolution'38 or a process in which successive
artefacts follow one another 'along trajectories'.39

How Scaling Ends

The scaling venture is long lasting, but not infinite. Scales vary until they
are 'stabilized' at a certain level of definition of the building. Then the
architects stop scaling and 'fix' the building.

I wouldn't say that there is a constant variation of scales. Not constantly

.. .You are going basically up and down all the time until you stabilize it in
one scale. In this project it's 1:50. And then, you can develop the project
in that scale. And sometimes you are going up, you jump to a bigger scale
to check some details, some corners, it's always details. You have even the
details for the furniture. In the big model you see the chairs and every
thing. But then, you are going back to the small model and you are
designing again. So, there is one level when you stop and you stabilize
things. (El 102)

At a given moment in the process, a few models are detached from the
scaling circulation network. They are stabilized in a certain profile and start
working on their own, taking new, independent, and straightforward linear
paths of development. Thus, the scaling process ends up with 'stabiliza
tion'.40 Contrary to all expectations, the scaling venture fails to have as its
upshot a huge detailed 'realistic model of the whole'. The final product of
architectural design is neither the building nor a mock-up sample of the
building in scale 1:1. It is instead that particular assembly of a few 'one
shape models' detached from the scaling continuum and its circular
network. This is what I found on the 'table of models' in the early
afternoon of November 2001 (Figure 1). It was not a bunch of successors

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
888 Social Studies of Science 35/6

and predecessors, but a myriad of presentational states generated, col

lected and stabilized in the office through time. This is the Whitney
building: the building is ubiquitous in the scaling operations, and is not
specifically located in any of them. Resulting from a rhythm with fine
undertones of variation and distance, acceleration and slowing down, it
appears as something quasi-unreachable and at the same time ever-present
in all models and states: a multiple, cumulative object visible through all of
them and in the movements connecting them.

I am indebted to the people from the Office for Metropolitan Architecture in Rotterdam,
and especially to Ole Scheeren, Erez Ella, Carol Patterson, Torsten Schr?der, Sarah
Gibson, Kunl? Adeyemi, Shiro Agata, Shohei Shigematsu, Olga Aleksakova and Rem
Koolhaas for welcoming me in the office and letting me follow them at work; they devoted
a lot of time and patience to me and my questions. For their comments on earlier versions
of this project, I acknowledge Bruno Latour, Peter Galison, Lorraine Daston and Joel
Snyder. Albena Yaneva is a grantee of the Graham Foundation for Advanced Studies in the
Fine Arts, grant # 03069

1. This research is based on a two-year ethnographical observation in the Office for

Metropolitan Architecture (OMA) of Rem Koolhaas in Rotterdam.
2. In a recent study, John Law shows how aircraft designers engage in a sequential process
of prototyping and testing to learn what is acceptable for establishing the best wing
design, in terms of relatively stable and determinate shape. Through different tests of
the extent to which the wing passes through vertical gusts of wind, the way it bounces
up and down, and the way it experiences turbulence, they find a strategy for modelling
the factors that might affect gust response. These factors refer to a variety of external
realities: the weight leads into the realm of bureaucratic politics, the size of the wing, to
the Russians (the need for short take-off from camouflaged airstrips), and so on (Law,
3. The term of 'translation' is used in science studies to designate, with all its linguistic
and material connotations, the displacements realized by actors whose mediation is
indispensable for any action to occur. Instead of maintaining the rigid opposition of
content and context, words and world, the 'chain of translations' points to the
meticulous work though which actors modify, displace and translate their various
interests in practice, multiplying the mediators instead of demystifying the pretensions
of science as critical sociology does (Latour, 1990, 2001: 33-82). Translations,
transfiguration, transformation, transfer - any of these terms refer also to the multiple
procedures through which a building is brought into existence. They sit happily over
the blind spot between architectural drawings, models and diagrams and their object -
the building. However, it remains unclear 'how things travel and what will happen to
them on the way to the final building', and ... 'the transmutation that occurs between
drawing and building remains to a large extent an enigma' (Evans, 1997: 160; Allen,
4. Michel Call?n argued for the importance of an STS study of architecture focusing on
the materiality of design as a world of graphs and strategies of visualization, grounded
in negotiations (Call?n, 1996). Other authors addressed criticisms to this programme
(Raynaud, 2001), but did not suggest empirical alternatives.
5. They focused on the social underpinning of design and production activities (Blau,
1984), or analysed the products of architectural design as socially constructed in
negotiations among architects and an array of contributors (Cuff, 1991).
6. On the notion of rhythm in architecture see Itten (1975) and Greene (1976).

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
Yaneva: Scaling Up and Down 889

7. Compare this with studies that consider design as a work of the brain when, in a
mysterious moment of inspiration, delirium and concealment, an image of the building
appears in a flash (Boyd, 1965; Akin & Weinel, 1982).
8. A fascinating study on the cognitive dimension of engineering, conducted by Vincenti
(1990), noted that engineering design knowledge is acquired in a day-to-day enterprise
according to a systematic experimental methodology. That is how engineers acquire
empirical data needed to carry out design, since the theoretical methods cannot supply
the requisite data.
9. Architectural design is not a gradual step-by-step transfer from one scale to another,
developing towards a ratio of 1:1 (Boudon, 1972); rather, discontinuity and versatility
are its main figures (Schatz & Fiszer, 1999). It relies on surges, breaks, sudden 'jumps'
and meticulous inspections, repetitions and returns; it sets into play simultaneously
different sized actors and several scales, many of which persist throughout all the stages
of the project, regardless of their precision. This story of discontinuity to some extent
follows recent studies on engineering design that treat it as a messy non-linear process,
full of unforeseen pitfalls and unpredicted actions (Henderson, 1999) - a maze, or
complex multidimensional web of interconnections, moving toward a final well
designed product (Bucciarelli, 1994).
10. This view differs from that in studies in the philosophy of technology, which describe
the genesis of technical objects as concretization, that is, as ascending from the abstract
to the concrete; from an unpredictable abstract object, open to its environment,
towards a closed, predictable, differentiated and concrete object (Simondon, 1989).
11. On the building as a 'multiverse', not socially constructed but a stabilized gathering of
few models, tentatively adjusted together and composed in a whole, see Yaneva (2005).
12. Design studies considered constraints as a 'primary generator' triggering a process of
architectural exploration that led to a conjectural solution (Darke, 1979). Although
explicitly articulated, constraints and boundaries, which are so critical for engineering
design, are not considered inflexible; rather, they are subject to change and negotiation
(Bucciarelli, 1994). Science studies also have argued that scientific experimentation
runs according to 'multiple constraints', considered as material obstacles that shape
and delimit action in experimentation (Galison, 1995, 1998).
13. See Koolhaas & Mau (1995) and the Office of Metropolitan Architecture & Koolhaas
(2004). Koolhaas analyses different scale-projects at OMA, investigating how they are
proportionally applied to different-sized cities and urban spaces, as well as how they
generate multiple contents.
14. However, there are few accounts of scale models in architecture, as compared with the
enormous amount of writing on architectural drawing (Porter, 1979; Blau & Kaufman,
1989; Robbins, 1985, 1994; Evans, 1997).
15. Models often travel outside the architectural office to gain powerful allies among
clients, sponsors and future users, community groups and city planning commissions.
They are supposed to express concerns, expertise, opinions and expectations, which are
furthermore taken into account in design. Thus models incorporate not only a variety
of technical concerns, but also a range of other viewpoints.
16. On the dual existence of molecular models in chemistry as quasi-inscriptions and anti
inscriptions at the same time, see Francoeur (1997, 2000). like models of molecules,
architectural models are submitted to various manipulations, assembled, probed and
measured, and are used to gain knowledge about spatial arrangements. However, they
do not reveal properties of structures to ascribe them to hidden phenomena (such as
molecules in chemistry stand); they all work together in a common visual space to
'obtain' a building.
17. They tend to develop different characteristics as the result of a change in the initial
composition; like mutants, models are physically distinct variants of the same species,
the building. None of them is identical to another; each of them is a distinct
composition of things. All together they shape and contain the building.
18. 'Scription' terms include in-scription (Latour & Woolgar, 1979), and con-scription
(Henderson, 1999) and pre-scription.

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
890 Social Studies of Science 35/6

19. According to Henderson the visual communication (expressed in sketches and

prototypes) is the 'glue' that holds engineering groups together (Henderson, 1999).
20. Various actors are enlisted at different stages of architectural design: structural and
mechanical engineers, experts, stage designers, client representatives, future users, and
so on. Although these consultants are not directly mobilized in the scaling venture,
some of their concerns, expectations and requirements are taken into account while
architects scale up and down. However, it is beyond the scope of this essay to depict
their specific modes of participation.
21. On the term of'composition' versus 'construction' see Latour (2003). Nothing is more
convincing for showing the composite character of the building than a scale model in
which a variety of concerns and requirements have to be fitted together.
22. The real name of this instrument is a borescope. It is primarily designed for the
observation and inspection of the inside of machines, equipment and structures. Since
in the architectural office it is used to inspect the interior of the scale models, it is
called 'modelscope'.
23. The attachment of a miniature video camera or a special viewing adaptor would allow
architects to attend the endoscopie examination simultaneously with the operator
Kunl?, and can enable a filmed real-time movement through and around the model.
Perceived movements can also be simulated by the simple operation of panning and
tracking, which also can be photographed.
24. The awareness of our own physical size and bulk in relation to that of a scale model is
known in architectural theory as 'Gulliver Gap' (Porter & Neale, 2000).
25. It is a constrained experience, because seeing in the small model requires efforts with
the body and straining the eyes. By bending his own body to the model's folds, passing
through numerous trials, it is as though the scoping architect reduces himself to the
scale of the small model. Doing so, he scales down, plunges inside the building, and
stands at the same level. This differs from presenting the building from a dominant
position, which normally states that phenomena have been investigated in a laboratory
and are under control (Latour, 1984; Sibum, 1992; Schaffer, 2004). Thus, seeing the
building requires a series of material operations to adjust the architect to the model
size, instead of a mysterious work of his brain.
26. The terms of'scoping in9 and 'scoping ouf are borrowed from the computer language
of'zooming in\ 'zooming out and 'fit into image'.
27. The material realization of the huge model requires more accurate procedures of
cutting, pasting and modelling. These procedures all require exact measurements for
reproducing shapes, sizes, openings and proportions. In the model-making process,
architects 'follow the drawings with precision', cut the paper plans and sections, apply
them meticulously on the foam blocks; the sequence of operations has to be performed
with rigorous accuracy rather than through a random exercise with materials and
shapes. A detailed description of these operations is beyond the scope of this paper.
28. As shown by Hutchins, the cognitive properties of the group differ significantly from
the cognitive properties of an individual member (Hutchins, 1991, 1995). On cognition
as embedded in social practices and distributed within group activities, see also Lave
29. By following architecture students as they learn to design, Sch?n (1985) defines an
epistemological paradox of the architectural studio: on the one hand, students need to
learn a new competence, and they do not initially understand what they need to learn;
on the other hand, they can only educate themselves by beginning to do design work.
30. Sometimes the viewing experience is mediated by a camera, with tiny manoeuvrable
lenses, which is able to enter the model and to document the static interior of the
model in the form of a moving image. If the first reason why architects enter the small
model is to experience the space and to use this immediate knowledge for the physical
transformation of the model, the second reason is to obtain images of the building that
will be closer to the ways people will experience it.

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
Yaneva: Scaling Up and Down 891

31. Architects call a 'concept' the main idea of the building, taken in its relationships with
the client demand, the city, the urban fabric, and the broader social, political and
cultural context.
32. Following Alexander's definition of successful design as a 'good fit' (Alexander, 1964),
we can consider scaling up as an attempt to make a myriad of small elements and
micro-equipment fit together and 'work well'. While the 'good fit' at large scale provides
incentive for scaling down - when successful it is taken down and 'determined' on the
small scale - a 'misfit' provides an incentive for making radical changes to the small
model. Thus, scaling up and down aims at neutralizing the incongruities that cause a
'misfit'. The follow-up of scaling operations gives us the possibility to experience
cognitively the sensations about what architects call a 'good test'.
33. Architects call a 'programme' the content of the building - the internal distribution of
spaces according to functional needs, general scope and insertion in reality.
34. To solve a 'problem' means in architectural terms: 'isolating' the problematic feature,
transporting it between scales, zooming and reducing, pushing and defining, thus 'dis
solving' it through travel. By 'dis-solved' I mean 'solved' in a particular way. I use the
twofold meaning of 'solution' - as the resolution of a difficulty and as two substances
mixed together and uniformly dispersed. As scales are shifted the problematic feature is
solved in the same way one substance is dissolved in another and the building becomes
35. Images are regularly produced at every scale-up and scale-down move; the images serve
as protocols for carefully maintaining the traces of the scaling experiences, investing the
trials with materials and shapes; documenting new data about the building. On the flat
surface of multiple collages, montages and drawings one can find imprinted the 'faces'
of the three-dimensional models, the traces of their movements and transformations.
Thus, the pure and formal redundancy of models cannot occur without the expressive
images, which capture their meaning and transmit information. The building can be
interpreted through the material body of these images that make it presentable for
client and publics.
36. The 'ability to anticipate' errors in size and relationship is considered important in the
scaling venture (Licklider, 1966). Examples from the history of technology highlight
that building successful full-sized machines was one of the challenging intellectual
problems with which early modern engineering confronted pre-classical mechanics in
the 17th century. All attempts to scale-up models of machines and to enlarge them in
correct proportion used to fail because scaling was only approached mathematically,
without taking into account the properties of the materials; namely, the robustness and
resistance of materials, qualities that diminish proportionally when a device is enlarged
(Popplow, 2003). Likewise, the difficulty of scaling-up models of vessels in naval
architecture in the 18th century was defined as the incapacity of rational mechanics to
describe or predict ship behaviour. Only the skilful model-maker was able to foresee
the displacement, stability, weathering qualities and other essentials of a large ship,
instead of using pure calculation (Schaffer, 2004).
37. According to evolutionary theories (Forty, 1986; Petrovski, 1992, 1994, 1996; Basalla,
1988; Pye, 1988) a new design product follows from earlier products through
successive functional changes. To elucidate the multiplicity of technical tools and the
drive for their improvement, these theories argue that novelty appears among
continuously evolving artefacts. They explain how the new design object comes into
being in relation to an external factor (social context, cultural atmosphere, economic or
political factors, society), always being the starting point of a new process of series of
transformations - that is, a linear and temporal succession of finished and limited
38. According to which the development of an artefact - understood in relation to its
ancestors and successors - follows a staircase-like evolution of sequences of change and
continuity (Bijker, 1995: 88).
39. See what Latour (1989: 322-52) has called 'diffusion model of technology' as
compared with the 'translations model'.

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
892 Social Studies of Science 35/6

40. I am following here the actors' definition of stabilization, although the term was used in
laboratory studies (Latour & Woolgar, 1979). Architects define stabilization as a
momentary pause in the scaling up and down process; a clarification of the building
profile slowing down the versatile scaling course.

Akin, ?. & E.F. Weinel (eds) (1982) Representation and Architecture (Silver Spring, MD:
Information Dynamics).
Alexander, Christopher (1964) Notes on the Synthesis of Form (Cambridge, MA: Harvard
University Press).
Allen, Steve (2000) Practice: Architecture, Technique and Representation (Amsterdam: Overseas
Publishers Association).
Basalla, George (1988) The Evolution of Technology (Cambridge: Cambridge University
Bijker, Wiebe (1995) Of Bicycles, Bakelites, and Bulbs. Toward a Theory of Sociotechnical
Change (Cambridge, MA & London: MIT Press).
Blau, Eva & Edward Kaufman (eds) (1989) Architecture and its Image (Montreal: Canadian
Centre for Architecture).
Blau, Judith (1984) A Sociological Perspective on Architectural Practice (Cambridge, MA: MIT
Boudon, Philippe (1971) Sur l'Espace Architectural. Essai d'?pist?mologie et d'Architecture
(Paris: Dunod).
Boudon, Philippe (1972) LaVille de Richelieu (Paris: AREA).
Boudon, Philippe (1992) Introduction ? l'Architecturologie (Paris: Dunod).
Boudon, Philippe (1999) 'The Point of View of Measurement in Architectural Conception:
From the Question of Scale to Scale as Question', Nordic Journal of Architectural
Research 12(1): 7-18.
Boyd, Robin (1965) The Puzzle of Architecture (Carlton, Australia: Melbourne University
Bucciarelli, Louis L. (1994) Designing Engineers (Cambridge, MA: MIT Press).
Call?n, Michel (1996) 'Le Travail de la Conception en Architecture', Situations, Les Cahiers
de la Recherche Architecturale 37(1): 25-35.
Cuff, Dana (1991) Architecture: The Story of Practice (Cambridge, MA: MIT Press).
Darke, Jane (1979) 'The Primary Generator and the Design Process', Design Studies 1:
Dupire, Alain, Bernard Hamburger, Jean-Claude Paul, Jean-Michel Savignat & Alain
Thiebaut (1981) Deux Essais sur la Construction (Bruxelles: Mardaga).
Evans, Robin (1997) Translations from Drawing to Building and Other Essays (London:
Architectural Association).
Ferguson, Eugene (1992) Engineering and the Mind's Eye (Cambridge, MA & London: MIT
Forty, Adrian (1986) Objects of Desire (New York: Pantheon).
Francoeur, Eric (1997) 'The Forgotten Tool: Design and Use of Molecular Models', Social
Studies of Science 27(1): 7-40.
Francoeur, Eric (2000) 'Beyond Dematerialization and Inscription: Does the Materiality of
Molecular Models Really Matter?', Hyle - International Journal for Philosophy of
Chemistry 6(1): 63-84.
Galison, Peter (1995) 'Context and Constraints', in J.Z. Buchwald (ed.), Scientific Practice.
Theories and Stories of Doing Physics (Chicago, IL & London: University of Chicago
Press): 13-41.
Galison, Peter (1998) Image and Logic. A Material Culture of Microphysics (Chicago, IL &
London: University of Chicago Press).
Gorman, Michael (1997) 'Mind in the World: Cognition and Practice in the Invention of
the Telephone', Social Studies of Science 27: 583-624.

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
Yaneva: Scaling Up and Down 893

Greene, Herb (1976) Mind and Image: An Essay on Art & Architecture (Lexington, KY:
University Press of Kentucky).
Henderson, Kathryn (1999) On Line and On Paper: Visual Representations, Visual Culture, and
Computer Graphics in Design Engineering (Cambridge, MA: MIT Press).
Hutchins, Edwin (1991) 'The Social Organization of Distributed Cognition', in L. Resnick
& J. Levine (eds), Perspectives on Socially Shared Cognition (Washington, DC: APA
Hutchins, Edwin (1995) Cognition in the Wild (Cambridge, MA: MIT Press).
Itten, Johannes (1975) Gestaltung^- und Formenlehre: Mein Vorkurs am Bauhaus und Sp?ter
(New York: Van Nostrand Reinhold).
Knorr-Cetina, Karin (1999) Ep?temic Cultures: How the Sciences Make Knowledge
(Cambridge, MA: Harvard University Press).
Koolhaas, Rem & Bruce Mau (1995) Small, Medium, Large, Extra-Large (Rotterdam: Office
for Metropolitan Architecture).
Latour, Bruno (1984) Pasteur: Guerre et Paix des Microbes, Suivi de Irr?ductions (Paris: La
Latour, Bruno (1989) La Science en Action, Introduction ? la Sociologie des Sciences (Paris: La
Latour, Bruno (1990) 'Drawing Things Together', in M. Lynch & S. Woolgar (eds),
Representation in Scientific Practice (Cambridge, MA: MIT Press): 19-68.
Latour, Bruno (2001) UEspoir de Pandore: Pour une Version R?aliste de VActivit? Scientifique
(Paris: La D?couverte).
Latour, Bruno (2003) 'The Promises of Constructivism', in D. Hide & E. Selinger (eds),
Chasing Technoscience. Matrix for Materiality (Bloomington, IN: Indiana University
Press): 27^7.
Latour, Bruno & Steve Woolgar (1979) Laboratory Life (London: SAGE Publications).
Lave, Jean (1988) Cognition in Practice: Mind, Mathematics, and Culture in Everyday Life
(Cambridge: Cambridge University Press).
Law, John (2002) Aircraft Stories: Decentering the Object in Technoscience (Durham, NC: Duke
University Press).
Licklider, Heath (1966) Architectural Scale (New York: Brazillier).
Lynch, Michael (1985) 'Discipline and The Material Form of Image: An Analysis of
Scientific Visibility', Social Studies of Science 15: 37-66.
Office for Metropolitan Architecture & Rem Koolhaas (2004) Content (K?ln: Taschen).
Orr, Frank (1985) Scale in Architecture (New York: Van Nostrand Reinhold Company).
Petrovski, Henry (1992) The Evolution of Useful Things (New York: Vintage Books).
Petrovski, Henry (1994) Design Paradigms. Case Histories of Error and Judgment in
Engineering (Cambridge: Cambridge University Press).
Petrovski, Henry (1996) Inventing by Design, How Engineers Get from Thought to Things?
(Cambridge, MA: Harvard University Press).
Popplow, Markus (2003) 'Models of Machines: A "Missing link" Between Early Modern
Engineering and Mechanics?', Preprint no. 225 of the Max-Planck Institute of the
History of Science in Berlin.
Porter, Tom (1979) How Architects Visualize (London: Studio Vista).
Porter, Tom & John Neale (2000) Architectural Supermodels. Physical Design Simulation
(Oxford & Boston, MA: Architectural Press).
Pye, David (1988) The Nature and Aesthetics of Design (London: The Herbert Press).
Raynaud, Dominique (2001) 'Comp?tences et Expertise Professionnelle de l'Architecte
dans le Travail de Conception', Sociologie du Travail 43: 451-69.
Robbins, Edward (1985) 'Drawing and the Social Production of Architecture', in PL. Knox
(ed.), The Design Professions and the Built Environment (London & Sydney: Croom
Helm/New York: Nichols Publishing Company): 42-61.
Robbins, Edward (1994) Why Architects Draw? (Cambridge, MA & London: MIT Press).
Schaffer, Simon (2004) 'Fish and Ships: Models in the Age of Reason', in S. de
Chadarevian & N. Hopwood (eds), Models: The Third Dimension of Science (Stanford,
CA: Stanford University Press).

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to
894 Social Studies of Science 35/6

Schatz, Fran?oise & Stanislas Fiszer (1999) 'Dealing with Space: Tales and Scales in
Architectural Design', InnehM 1: 43-59.
Sch?n, Donald (1985) The Design Studio: An Exploration of its Traditions and Potentials
(London: RIBA Publications).
Sibum, Otto (1992) Shifting Scales: Microstudies in Early Victorian Britain. Unpublished
manuscript, Max-Planck Institute for the History of Science.
Simondon, Gilbert (1989) Du Mode d'Existence des Objets Techniques (r??dition avec postface
et pr?face) (Paris: Aubier).
Vincenti, Walter G. (1990) What Engineers Know and How They Know It: Analytical Studies
from Aeronautical History (Baltimore, MD & London: The Johns Hopkins University
Vinck, Dominque (ed.) (2003) Everyday Engineering: An Ethnography of Design and
Innovation (Cambridge, MA: MIT Press).
Yaneva, Albena (2005) 'A Building is a Multiverse', in B. Latour & P. Weibel (eds), Making
Things Public (Cambridge, MA: MIT Press).

Albena Yaneva completed her doctoral thesis on ethnography of art

installation in the Centre de Sociologie de l'Innovation, ?cole des Mines de
Paris. She has worked at the Max-Planck Institute for the History of Science
in Berlin and the Department of the History of Science at Harvard
University. She is currently the director of The Gallery of Research (Galerie
der Forschung), Austrian Academy of Sciences, Vienna.

Address: The Gallery of Research (Galerie der Forschung), Austrian Academy

of Sciences, Wollzeile 17, 1010 Vienna, Austria; email:

This content downloaded from on Mon, 08 Jan 2018 21:10:58 UTC
All use subject to