After my article was published in ArchDaily this past January, I received a delightful email from an Architect in Massachusetts who had read my article 5 times.  It inspired him to think more carefully about his own tendencies to rely on technology to make his design decisions.  He also told me that my essay reminded him of Colin Rowe, and so began a personal research project to rediscover the pedagogy through which I was educated.  I knew that Rowe was a huge figure in architectural theory, and was specifically credited with creating and enforcing much of the undergraduate program at Cornell beginning in the 1960s, where I graduated in 2002.  So this became a research project to better understand Rowe’s ideas of contextualism, but also a research project into the very nature of how I was educated, with what biases, and how that bias affects the way in which I see Architecture.

First I read Collage City, probably Rowe’s most famous full book, published in 1978, and co-authored with Fred Koetter. Rowe argues against modern – utopian – idealism, and promotes a combination of historic reference and scaled construction of new buildings that would fit within the patterns of an existing city. I also read a book titled Reckoning with Colin Rowe, edited by Emmanuel Petit, just published this year. Here, an architectural elite – Rem Koolhaas, Peter Eisenman, and colleagues of Rowe from his Texas Ranger days and later – reflect on Rowe’s influence in their own careers and theories, and his influence on architectural discourse in general.

I will attempt to contemporize much of Collage City within the framework proposed by Emmanuel Petit in Reckoning. In the introductory Chapter “Rowe After Colin Rowe,” Petit proposes the intellectual framework for why an update to Rowe’s legacy is necessary. He posits that we need a new interpretation of Rowe to make his ideas more relevant today, creating a new framework for the architectural analysis of forms never imagined by Rowe.

After the introduction, the rest of the book is mostly devoted to the personal relationships of each contributor and Rowe, and to measuring Rowe’s influence within each contributor’s own architectural theory. In other words, the book is actually concerned with evaluating Rowe’s contributions within the context of only each contributor’s work, not as a guide for reinterpreting Rowe’s analytical methodologies for today’s design tools.

Though none of the contributors are so blunt, they offer mostly disregard for Rowe. But, Rowe’s contributions to Architectural discourse were undeniable to each of them, so how can we reckon with that? If nothing else, Rowe established the criteria for how Architects compare works, through abstraction of intent, form, and detail. Peter Eisenman says it succinctly on page 57, when quoting his then teacher, Rowe, imploring of him: “Tell me something about what you are looking at that you cannot see!”

Since Petit’s intent for the book is clear in his introduction, I’ll use his writing to ruminate. Broken into 4 sections, I’ll touch on these interrelated aspects of Rowe’s possible influence: Collage, Solid/Void, Computational Design, and Phenomenal Transparency.


When Rowe turned the argument of the transparency of form from the vertical plane (of his architectural analyses) into the horizontal plane in order to develop a theory for urbanism, he stopped using the term transparency and instead spoke of collage – yet nowhere did Rowe address the critical difference between the two notions. . . Indeed, not all collages hinge on phenomenal transparency in Geyorgy Kepes’s sense – that is, the interpenetration of figures without optical destruction of each other; collages are mostly based on formal ruptures and on the tearing of paper and the side-by-side gluing of incongruous geometries and figures. Unlike the conceptual device of transparency, collage involves the more classical rules of pictorial composition. [p6]

Here lies a fundamental interpretation, or misinterpretation, of Rowe’s methodologies. What is a collage, and what did Rowe mean by it? Petit posits that Rowe’s “collage” could be incongruous, without phenomenal transparency, defined as a pictorial composition. In Petit’s footnote for his statement, he mentions Rowe’s observed interest in a student’s work at the time, in which the composition did not reference transparency. I believe this to be an underestimation of Rowe’s more complex view of collage, especially if my own education decades later may have still been influenced by it.

Petit assumes a break in Rowe’s intellectual progression from the tradition of phenomenal transparency to the abstract concept of collage. But, what if collage was an extension of his earlier observation? Additionally, two contributors to the book believed that Rowe meant “montage,” not “collage.” We need to investigate what these words mean, and discover what our cultural associations are with each term. I’ll propose:

Montage ==> separate images placed in adjacency (as a video montage, linked thematically though not visually; sequential rather than simultaneous)

Collage ==> one image from separate images (as in superimposition, figural interpenetration)

Mosaic ==> a pattern from separate images (as in mosaic tile patterns)

For my definition, montage assumes no physical relationship between its adjacent parts, though each of the pieces may have a thematic relationship. This is a style in the manner of Tschumi or Eisenman, where any two ideas may be placed adjacent without regard to geometric continuity or legibility (deconstruction). In montage, the parts do not influence each other.

Cornell collage allows for phenomenal transparencies in the work because of its nature to allow materials to layer, implying a third dimension. A form or edge may disappear and continue in another “space,” allowing for continuous visual reinforcement and a reduction of neurological processing to determine the makeup of the form in its entirety. For example a strip of red paper can continue the length of a work, disappearing and reappearing to the artist’s desire. Montage seems much more discordant, as if thrown together without pattern manipulation.

There is also the opportunity for an examination of history within collage. Layering has a time component to its construction that cannot be edited out, whereas adjacent fragments do not. Taken a step further, a montage potentially could be “final,” or “finished.” Collage could potentially be layered in perpetuity. Collage is all about the underlying relationships between forms over time.


Rowe’s artistic notions of figure/ground and poche appear out of place with regard to today’s actual politics of private ownership, limited public land control, and decentralized and international decision making. [p8]

In Shanghai, the complexity of the vertical forms and zoning (as if several cities were stacked on top of one another at variable scales) seems to transcend any of Rowe’s theories of urban architecture; the latter’s horizontalism (even when extruded into the third dimension) needs to be updated with a theory of the volumetric. [p9]

There are actually two issues at play here, that are discussed frequently in the book:

(1) Rowe’s “opposing values” are too simple – Rowe often creates opposing values to analyze Architecture. He utilizes exaggeration of two opposites to prove a point, a major opposition being solid (black) vs void (white). We may never know whether Rowe meant this literally, as a simplification of an abstract idea, or whether he may have thought there would be an opportunity for “gray”. To use Petit’s example, the verticality of the Pudong district (Shanghai) is obvious and extensive, with cities seemingly stacked upon cities, into the sky. A two-dimensional figure ground study would show relatively small solids in an expansive sea of white: the avenues are up to 10 lanes wide and the buildings rarely touch the street, as they are set behind expansive “drop off circles” and back-turning podiums. The new forms may yet prove Rowe was correct with his analytical methodology.

(2) Collage is two-dimensional – As discussed, a Cornell collage may inherently imply a third dimension. The ideas of a “Collage City” might propose that there is an opportunity in Pudong to link these towers/forms at other levels. For example, there is the potential for: intelligent underground spaces for metro tunnels and service corridors; elevated parks and recreation links; and, for parts of the historic urban fabric to have been integrated.

Indeed, zoning doesn’t allow for a solid/void interpretation when we design, it is much more concerned with massing strategies than public space definitions. But, maybe zoning should incorporate some of those lessons. Perhaps it is important to discuss how figure/ground can be utilized by zoning or urban planners. Maybe podiums occupy a gray area, neither street confirming nor open space? How do we classify parking podiums in solid/void opposition? 6-story internalized Asian shopping malls? Skyways? Underground passageway networks?


Lynn found in Bataille an apologist of the artistry of devious form and the devious body, which he deemed irreducible to ideal types and therefore inaccessible to Rowe. In this, the computer was described as an enabler of transformational processes, which were generative of a new universe of creative alternatives. [p12]

 With the rapidly evolving digital technologies of form generation in architecture, Rowe’s theories seem to no longer provide much of a mechanism to recognize the inherent logic of form per se; his once energizing ability to read and compare antecedents across historical time seemed to recycle the same precedents and arguments and appear to have run dry in comparison to the morphological potential emerging from digital culture. [p15]

Rowe never made any explicit claims about the repercussions of digital technology as a ubiquitous design tool for architects – even in the 1990s.

That depends if we are to analyze the forms literally, or the processes which create the forms. I’d argue it’s the process that we should take into consideration since we do not yet have the arc of history to judge the newer forms, and Rowe has plenty to say about that. I’ll let Rowe defend himself here, where indeed he does mention parametrics in 1978’s Collage City. and deems such architecture as “science fiction”:

“In so far as methodology, systems analysis and parametric design are elevated to be important pursuits, science fiction may present itself as an academicized version of what modern architecture was, anciently, supposed to be. But science fiction, like old-fashioned modern architecture, has also a less rigorous more poetic face. This is the familiar involvement with images conceived to illustrate science and then their advertisement as proofs of the designer’s all-relevant objectivity. . . The search for system, afterall, is very like the old academic thing – platonic certainty in brave new disguise; . . . rather than being conscious of morals, it is apt to be success-oriented. . . The results of science fiction, whether systemic or neo-futurist, usually suffer from the same conditions which plague the ville radieuse – disregard for context, distrust of the social continuum, the use of symbolic utopian models for literal purposes, the assumption that the existing city will be made to go away; and, if the ville radieuse is now supposed to be evil, productive of trauma and disorientation, it is not easy to see how science fiction, which would seem to compound the ills, is in any position to alleviate the problem.”

Rowe expresses alarm when he shares a quotation from the 1968 White house announcement of the National Goals Research Staff later in Collage City:

“An extraordinary array of tools and techniques has been developed by which it becomes increasingly possible to project future trends – and thus to make the kind of informed choices which are necessary if we are to establish mastery over the process of change.”

and adds himself:

“In so far as the form of the future depends upon future ideas this form is not to be anticipated.”

This is a great parallel to the incorporation of algorithmic modelling, fashionable amongst academics today. Surely, he might say, this is history repeating itself. There is a ridiculousness to somehow expect our “new” version of scientific reasoning to be somehow vastly more intellectual than prior attempts at controlling human nature. Scientific arrogance seems to have a way of repeating itself; Rowe was very fearful of the Architectural pursuit of “mastery over the process of change.”

Algorithmic design is reminiscent of the omniscient modern/utopian urbanist defining the future. With no ability to re-incorporate the future itself, research and conclusion is forever a frozen moment of time. In other words, the definition of utopia is a fallacy. How can you create utopia, if only a day later, that utopia is now history, and you need to construct a new vision for utopia? In essence, time always moves, and it is naive to somehow say that well, today, we know everything about humanity and have the perfect set of data, and now we can design the ultimate environment for happiness. But, in a year or two, you will have learned there are changes that need to be made, and how do you then incorporate those changes? Therefore utopia is a sham since it doesn’t assume any conceptual room for evolution or morphogenetic progression.

Why wouldn’t inputs change over time? There can be no perfect, unchanging, unflinching symbolic human being, or for algorithmic design, some sort of perfect set of data. History, adaptation, and evolution need to be a part the equation for asuccessful buildings and cities. Parametric design is a useful design tool, but it is no panacea.

I believe Petit would agree, here he finishes the thought:

Today’s formal avant-garde appears to eschew all historical content and instead, sees architecture as the edited instantiation of the inner structural logic of algorithms – often “fed” with the information of Big Data. . . . A more discerning understanding of history makes clear that the notion of historical precedent cannot simply be eliminated from the lexicon of architecture with the hope that in its wake, a space would open up for the genuinely new and unprecedented. And this is where Rowe can suddenly be made very relevant again. [p16]


. . . phenomenal transparency being “an argument between a real and ideal space” – is seen as the engine that produces formal complexity. For this intricate morphology to be readable then, it has to be disassembled into a series of discrete formal fragments.

Today, all buildings have turned literally transparent as wireframes on computer screens, and so the generator is mostly volumetric to start with, and can be thought of, and represented as, a diaphanous and approximate sphere of relationships rather than a rigid (fragmentary) geometry. An example of such a volumetric diagram can be found in UN Studio’s Mobius House, which builds on the twisted and involuted logic of two lines that contain volumes of space and intersect in specific points. . . no standpoint in space affords the “total” view of the diagram: it requires a volumetric reconstruction; the formal clues for such a mental construction are distributed through space rather than compressed into the flat surface of a facade. [p17]

From UN Studio’s website: “The organizational and formal structure of the private house is based on a double-locked torus, the mobius loop. The intertwining trajectory of the loop relates to the 24-hour living and working cycle of the family, where individual working spaces and bedrooms are aligned but collective areas are situated at the crossing points of the paths.”

For Rowe, the problem may be that there isn’t an historic precedent to constructing a space solely on the happenstance of a scientific equation. But, the evidence of a “human diagram” can begin to unfold the next generation of Rowe’s original thoughts.

Petit starts to answer the question:

. . . this does not mean that architecture has disengaged from its history but that Rowe’s model of precedent and formal analysis needs updating. To fill the conceptual void, architectural culture in the early 2000s switched to a type of discourse that foregrounded the search for a terminology of aesthetic effects, but this trend, too, is running out of steam because it is fundamentally aimed at producing a mere inventory of sensations without relation to the broader disciplinary context and history. The digital means of form generation, analysis, and representation should make it possible to read in a different way in relation to precedents, and to therefore realize that the term amorphic is used by analysts of form when they are unable to explain the relation of that form to the system of historical analysis used.

“phenomenal transparency” is futile to analyze a whole branch of space conceptions, which have a new relevance today [Saarinen’s TWA Terminal, much of Zaha Hadid]. Their diagrams are volumetric in the sense that they are irreducible to any two-and-a-half dimensional spatiality and thus resist the logic of cubist montage on a flat surface with intimations of depth. But these bulging, winding, interweaving, looping, bending, flexing, twisting, and bowing morphologies, which have taken center stage in contemporary design, do not have to be deemed “unreadable,” they need their own discourse of precedent to be constructed.  [p18]

Petit has a very good point here. And, again, I was hoping the theorists in the book would expand upon this investigation, though they did not. The Cubist reference, though, may be limiting. Though Rowe’s era of design was mostly orthogonal, or employed a much simpler vocabulary when using curves, there is no reason why collage and phenomenal transparency can’t work as analytical tools for projects without straight lines.

He kind of answers it here:

If phenomenal transparency is going to remain a useful analytical category for architects, it needs to be recast with certain modifications: the analyst has to be thought of as internal to the volumetric diagram, and the environment has to be conceptually unraveled from the inside out rather than with parallel striations in abstract space. . . . expression of phenomenal penetrability. . . unlike the concept of transparency in cubism, digital culture has brought back to the table a characteristic endemic to literal transparency in the physical world – the morphological transition from high to low resolution in the passage through a medium for example, by offering ways to conceptualize the conversion of precise geometric articulation into a mere building. The bulge indicated the thatness of a volumetric diagram over the whatness of it. [p20]

Diagramming of dynamic forces can help to analyze these complex forms, and how their geometries speak and mimic each other in a similar manner to phenomenal transparency.

Perhaps phenomenal transparency was really only a half-measure, a reduction of a truth, a way to make orthogonal arrangements more legible by the user. Geometric continuities between spaces allows for a reduced processing of visual stimuli for the user, and therefore a more comfortable, legible, experience. But, what if Architecture were to incorporate the dynamic use of space in the Architecture itself? Would dynamic diagram become the new tool of precedent study, an intellectual extension of the studies of Rowe? In other words, Rowe proposed geometry to be the continuity between spaces, which it still is, whether straight or curved. But, in addition, the movement of people and the programmatic uses of spaces can also provide a continuous link between non-orthogonal spaces, a layering of human use which inspires form. A good historic example might be Frank Lloyd Wright’s Guggenheim Museum, and a current example might be BIG’s La Maison des Fondateurs.

Dynamic diagramming would include: circulation, views, and compression/release forces. If something is volumetric for the sake of only its own form, and not diagrammable with any such applicable forces, it may be purely stylistic or fetishistic, and would fail such an intellectual architectural analysis. (Though, of course, there will always be opportunities for aesthetic and subjective judgment.)

Another logical extension of Rowe’s phenomenal transparency may be to examine designs by their consistency of pattern and rules, a continuity of algorithmic intent. Again, the link to Rowe’s original idea is to offer a reduction in neurological processing for the user, an ability to move through a sequence of spaces with fluidity. An architecture such as the Getty Center can be entirely understood by typical phenomenal transparency, which is what makes the campus so unique and whole. But, a more contemporary design such as UN Studio’s Galleria or Zaha Hadid’s Guangzhou Opera House – both their exteriors and interiors – requires a more complex analysis where we can examine the parametric tools used to design it, and the resulting forms.

This type of extension of Rowe’s original methodology is the type of reinterpretation the contributors to Reckoning should have considered. Not that Rowe was simply wrong or naive. Rather, that as design has progressed, his filter couldn’t approximate all that could be imagined.