EDITORIAL NOTE

John Hejduk’s ‘Texas Houses’ are a set of drawings for seven houses executed in graphite pencil 8H and 9H between 1954 and 1962.  Originally, Hejduk planned to work on one house per year over a ten-year period. All the houses are variations on the ‘nine-square grid’, a design methodology that Hejduk used as didactic device in his teaching at the University of Texas, Austin and later at Cooper Union. Hejduk conceived of the series as an effort to comprehend basic principles in regard to “architectural commitment.”  He insisted that the conceptual positions were evident in the architectural vocabulary of the drawings.  In 1980 Peter Eisenman curated an exhibition of these drawings at the Institute of Architecture and Urban Studies (IAUS) in New York and produced a small catalogue. For this publication, besides transcribing excerpts from a conversation with Hejduk, Eisenman penned the long essay ‘In My Father’s House Are Many Mansion’, a formal close reading of Hejduk’s houses which is republished here. This essay can be considered one of the most intense and rigorous examples of close reading in architecture. Originated in literary criticism, close reading consists in the careful interpretation of a text in which its formal characteristics are described in detail. As such close reading emphasizes a discrete passage over the general narrative. Close reading is a patient and at times obsessive search for the elemental aspects of language itself. Employed by the Russian Formalists and later by the American New Critics, close reading defamiliarize texts by suspending their obvious messages, while revealing their techniques and formal dynamics. In his approach to architecture, Eisenman embraced close-reading to suspend the more commonplace aspects of architecture — use and program — in order to focus on what he called work on language. With their rigorous and inventive form, Hejduk’s Texas Houses offered to Eisenman the ideal text upon which to exercise the art of close reading. Hejduk’s subtle moves in plan and elevation — which are clearly informed by a wide range of precedents — are unpacked by Eisenman’s precise wording. The challenge of close-reading is an old but still persistent problem for architecture: to what degree is possible to use words to describe the design moves of an architectural project?  In a world where architecture is reduced to slogans and the instantaneous consumption of images is its main modus operandi, close reading can be understood as a form of resistance in which the detail takes its revenge on the whole. Yet the whole is not forgotten or erased, it is deconstructed in all its infinitesimal components so that its ideological power is diminished, made visible, and perhaps momentarily tamed. Against reducing the discourse on architecture to strong and often predictable narratives — a tendency evident today — close reading offers the possibility of looking at things carefully, giving them the necessary time and space to reveal themselves. The republication of this essay is a teaser for a book that Pier Vittorio Aureli and Michael Robinson Cohen are preparing on John Hejduk’s houses, that focuses on the period from 1954 and 1974. In this period Hejduk produced an exceptional body of drawings defined by a sequence of ‘conceptual projects’: the Economy Houses, the Texas Houses, the Diamond Houses and the Wall Houses. This body of work represents one of the most important reflections on the nature of architectural form. 

Pier Vittorio Aureli and Michael Robinson Cohen

The structure of any discipline may be defined at two levels: first, those aspects of it which generally distinguish it from other disciplines; second, those which reveal it, in itself, to be that discipline and no other. Thus, although words might serve to distinguish prose and poetry from say music or sculpture, their use does not insure the presence of either prose or poetry, or distinguish between them. Equally, when words are put into an order—into sentences, according to the rules of a grammar, fulfilling all conditions of grammatically correct sentences—they do not constitute prose or poetry, nor do they necessarily help to distinguish between them. Even if such sentences are meaningful, this is still no guarantee that they are poetic. There remains unanswered the question of what ‘poetry’ or ‘prose’ is in itself.

A similar set of observations may be made concerning the distinction between building and architecture. While all building and architecture possess a set of elements in common (columns, walls, floors, doors, windows, rooms), which are not only necessary to their physical existence but also serve to distinguish them from other three dimensional geometric configurations or sculpture, these elements alone are not sufficient to distinguish between architecture and building. And since we know that all building is not architecture, we must look beyond the mere elements of building to find the critical distinction. All building and architecture also of necessity have some programmatic content—a use as well as a significance. Yet while this may necessarily distinguish building from pure geometry and even sculpture, it is not in itself sufficient to differentiate architecture from building.

Then what is architecture? Initially one might say it is all of those necessary conditions combined: building elements, program, meaning, etc., which distinguish building from music, poetry, sculpture, and geometry, together with some condition of sufficiency, which for now we will call architectural content itself. But the presence of the former is no guarantee of the latter. What then is this architectural content? Traditionally, it has been thought to be some combination of typical building elements, placed in some useful, expressive, and aesthetic context. This expressive aspect is often referred to as an architectural language. Most discussions of the language of architecture refer not so much to what it is, the intrinsic nature of its content, but rather to what it does, to its expressive and functional capacity.

Modernism (as opposed to modern architecture) for the first time proposed a clear distinction between the expressive content of a language and the nature of a language itself. Modernism understood that in any art there are two separate components to a language: one, a grammar and content which are extrinsic, that is, which refer outside themselves to the general condition of culture; two, an intrinsic condition, a structure and content which refer to the nature of the art itself. This second condition was seen to exist side by side with the extrinsic condition but ultimately also apart from it. Now, while Modernism, as it was manifested in most of the arts, also studied the expressive capacity of the discipline, in many cases as a social critique of the classical Western tradition, its primary focus was to isolate the internal structure of the art; it turned inward to the discipline itself to what could be described as ‘work on the language.’

In modern architecture the manifestations of the extrinsic condition are obvious. They were contained in both the polemical literature—the little magazines and manifestoes of the first quarter of this century —and in the buildings themselves, where the iconography of the machine prophesied a new social order. The intrinsic condition, however, is more difficult to articulate in architecture, precisely because the idea of an internal ‘work on the language’—on the nature of architecture, its own sufficiency—was never explicitly formulated, particularly by the Modern Movement, as something separate from the ideological content of architecture. In architecture ‘work on the language’ usually involved something else. On the one hand there were linguistic analogues, the idea that elements and formal relationships could be compared to phonemes and syntax; on the other hand there was the idea that classical elements—plans and facades—could speak of some extrinsic metaphoric or iconic content. In the latter case, rarely was the idea of any intrinsic content at issue. However, if one begins to look closely at the architecture of the early Modern Movement there are many examples of an activity which could truly be called ‘work on the language’ in the Modernist sense. In Europe, for example, and particularly in the work of Le Corbusier, certain fundamental ‘linguistic’ conditions were posited concerning the idea of a horizontal and vertical datum, as in the Maisons Dom-ino and Citrohan. In America, while the examples are more difficult to come by, certainly the work of the European ‘immigrant’ Mies van der Rohe exhibits tendencies which can only be described as internal to the work of architecture. For if anything can describe the early work of Mies in America, it is reductiveness, the attempt to reveal the essential elements of an architectural objecthood, which is an idea central to Modernism in general. For Mies, the minimalist aesthetic of “less is more” was an attempt to free the form from the program and thus from its cultural antecedents in order to look at the object itself. Mies’ details—his concern for how something is made—became the focus of this activity. Thus, in both Le Corbusier and Mies there can be seen an attitude which is not concerned so much with the use or meaning of the object or with the new use of the classical language as with the manifestation of the object itself, that is with the presentation of what can be called purely architectural content. The differences between this ‘linguistic work’ in Europe and America seemed to have to do with the nature of the elements that were chosen for the investigation. In Europe, it was the plane, whether vertical or horizontal, which was central to the work, while in America it was the volume, whether as defined by the outlines of mass and structure in the work of Louis Kahn or as articulated by the grid frame and the universal space in the work of Mies.

All this is by way of attempting to locate and understand what is at issue in the seven Texas Houses of John Hejduk. Far from attempting to define a pedigree or to establish ancestral origins, it is intended to frame the particular transformations in the conception of the object which take place in these houses from the first to the last, in order to determine how the subtle transitions manifest a kind of ‘work on the language.’ And while the parallels to Mies and to the other ‘American,’ Louis Kahn, are clear, they would tend if pursued only to obscure the differences between Hejduk’s ‘work’ and theirs, and thus the significance of the architectural content in Hejduk’s objects. While all three are involved in an initial reductivism, Mies and Kahn display an apparently resolved condition which produces a kind of reified essence. The seven Texas Houses reveal that Hejduk, on the other hand, is concerned with the process of reduction, of methodologically cutting down to produce a very complex essence.

Hejduk’s architectural ‘work’ as manifested in these houses can be broken down into two areas: first, the manner in which he uses horizontal elements—the site, the plan, and the roof—to imply concepts of space. Second, how he uses vertical elements—the columnar grid and the vertical surface—to reveal concepts of time. Unlike Colin Rowe’s conception of space or Siegfried Giedion’s concept of time, in Hejduk’s work space and time reveal intrinsic conditions of architectural content—perhaps aspects of the nature of architecture itself.

In Houses 1, 2, 3, and 5 this content is revealed through various articulations of the horizontal plane, while in Houses 4, 6, and 7 it is revealed through the frame and the vertical plane. The focus on the horizontal in the early houses is obviously influenced by the Texas landscape, the Texas prairie with its flatness articulated by low hills and tight gulleys. This must have had a profound impact on the Catholic ‘immigrant’ from the Bronx. This initial source is fused with the concept, both Wrightian and Miesian, of the roof as the dominant datum. The plans, on the other hand, again like Wright’s and Mies’s, are merely extrusions; they become the section and the section in turn becomes the facade.

House 1 is a progenitor. Its operative energies inform all the later houses. The entire figure is set on an implied horizontal plinth indicated by the height of the first  floor slab off the ground. Even though the columns are uninterrupted from ground to roof line (they pass outside the floor line), still a horizontal sandwich of space between floor and roof remains dominant as a datum. It has a Miesian quality even though the horizontals never extend beyond the column line. Entry is from above, through the garden, and one looks down on the roof as in Wright’s Prairie houses, which are apparently similar in conceptual in-tent. T’he difference between Hejduk’s concept of space and that of Mies is in the internal definition. Mies’s columns in his early European work are not on the periphery; rather the periphery, as in the Barcelona Pavilion, is exploded out. In his later American work, the periphery does become contained, but at the same time structure is eliminated from the interior and the specifics of program are minimized in favor of a more universal space. In Hejduk, the initial ground is nine squares ordered by sixteen square columns, which also define a center line as the intersection between bays. There are eight additional smaller square columns on each side located to define a stepped pyramidal shape extending out from each facade and forming a diamond by the intersection of their mid-points. The nine squares and the implied diamond can be seen initially as an overlay of Palladio’s Palazzo Thiene and Villa Rotonda. The house is thus not a simple set of geometric abstractions, but rather is impacted with a series of compressive and extensive energies. These are further developed in the unequal ground line which defines a secondary cross axis and a shear perpendicular the axis of movement. This secondary axis cues the major architectural proposition in House 1, which is that while the house seems biaxially symmetrical, it is not. The generic nine-square symmetrical plan is impacted with a series of programmatic asymmetries, which hint at the then contemporary dogma of form following function. These asymmetries are paired: the two side extensions enclose interior spaces; the two on the front and back extend as horizontal porches. The two on the side set back a half bay in from the corner and progressively recede as they extend out. A similar compression and extension occur on the front and the rear whose key lies in the nature and location of the two entry stairs. The one in the front extends out from the front porch, the one in the rear is contained within the porch; the front half of the house thus appears to be telescoped out, the rear half to be compressed. Unlike the De Stijl houses of van Doesburg, House 1 orthogonalizes the diamond, and unlike the Villa at Garches which creates a frontalized condition by compression, House 1 is frontalized by its additive extension.

House 2 is conceptually the same even though it appears quite different. It is both nine squares and a diamond or cruciform even though the latter figure is now contained within the square perimeter of the house. Thus there is distinction between center and edge. The implied plinth of House 1 now becomes materialized. The location of the interior doors sets up a secondary gridding which is similar to that in House I but this time runs across the entry axis rather than parallel to it. When read with the half bay which runs completely around the perimeter it creates simultaneously both a lamination of the interior, and a compaction of the exterior, and thus a conceptual density at the periphery.

The theme of the compression and extension of space continues with House 3. It refers back to House 2 and forward to House 5. It is the last plinth house, the last house where the base is a more dominant (larger) element than either the roof or the columns. In the manner in which the lines of the columns are drawn and continued from the underside of the roof plane over the solid plinth to the ground it is like House 2; there is in this gesture a compaction of vertical and horizontal elements, a transparency in which there are both piloti and plinth. Again, the horizontal plane of the plan is the dominant element, but several different energies are also operative. The nine ‘squares’ are now articulated by sixteen void ‘slots’ which replace the formerly solid columns. These slots are defined by H-shaped sections which have metal flanges capping the solid bearing walls and a glass web. While the cruciform center has disappeared volumetrically, it now subtly reemerges in the similarity in the size of the five central bays. The center bay has two A dimensions, the four center bays on the periphery have A and B dimensions and the four corner bays have two B dimensions. Thus each bay position in plan is denoted by the different size relationships of its sides. Equally, it is the only house which articulates a difference between servant and served space. Because of the U shaped interior corridor enclosing a court, a peripheral ring of rooms can be read. This corridor ring is interrupted by the living room, causing a secondary set of parallel and perpendicular energies to be read: like House 2, there is a lamination from front to back, but here, it is manifested in the solid-void distinction made between the front-back and two sides. The front and back are articulated by void verandahs lined with seven square solid columns; the sides are almost a mirror image of the front: now the walls are solid and the columns are ‘void.’ In both front and side the columns and slots similarly extend down in a transparent overlay on the plinth, to create a continuous articulating band on the perimeter. Adding to this is a peripheral rotation initiated by the asymmetric location of the front and back stairs. This sets up simultaneously an internal shear and an external pinwheel. At the same time, the location of a column in the center of the front and back facade tends to ‘pin’ or deny the rotational tendencies.

House 5 follows conceptually from House 2 and House 3. It is the conclusion of the investigation of the horizontal dimension, while its site plan contains a first signal of the transition to the later houses. It anticipates, in its abandonment of symmetrical laminations for a dynamic asymmetry, the forces released in the site plan of House 4. But it is in the object itself that House 5 marks a most profound change. It is here that the Miesian condition of the roof and the frame is confronted. It is at once the most Miesian and the least Miesian of the houses. This paradox signals the arrival of Hejduk’s work beyond Mies, to a point where Mies’s work had conceptually been leading. First, the cellular compartmentalizing of the first three houses is replaced by a ‘universal’ space, punctuated by four internal columns, where only the trace of the former wall divisions is scored in the floor. Second, where in the first three houses the roof line was punctured by successive clerestory projections and there was no spandrel line, now the roof becomes inviolate and a secondary horizontal datum is provided by a continuous spandrel line, this time below the roof. With the introduction of the spandrel the plane of the glass becomes something different from that of Mies. While the elevation remains static and classical with respect to the plan, the solid functional elements play across this plane like the colored squares of an early Mondrian interrupting the grid lines. Here these solids become grid elements themselves, thickening and thinning as they pinwheel off an empty center. The thin intermediate mullions on the exterior frame move in their asymmetrical locations not so much to set up lines of force from side to side but rather a peripheral syncopation of movement akin to Broadway Boogie Woogie—a compaction of time (from early to late Mondrian) and space (now a dense peripheral membrane). But finally, it is in the conception of the roof that House 5 seems to move beyond Mies’s development of the same element. For Mies the roof plane was the dominant datum. It was usually expressed quite literally as a hovering umbrella. It relied on strong cantilevers or exposed roof trusses, from which the umbrella-like form was suspended. The glass in most cases was seemingly dropped from the roof to create an envelope, and the columns, rather than extending down from the roof, were the elements which supported the glass skin. In Hejduk’s Houses 1 and 2 the roof line was punctured by a clerestory roof projection. Both were cruciform in shape, and both of their projections conformed to the definition of the major public living space. The cruciform in House 1 corresponded to the peripheral extension of the diamond on the exterior, suggesting a contrapuntal pressure from the interior which pushed upward and outward at the same time as the roof could be seen to push down. In House 2, while there was pressure up through the roof, there was a containment on the periphery. House 3 suggested a different condition of these vertical spatial pressures. Its two rectilinear clerestory projections conformed to a progression created across the central bay from front to back, while a depression in the horizontal plinth in the center of the house made for movement up and down from a dual datum-the two sandwiching horizontals provided by the roof and the plinth. In House 5 it is significant that the ground plinth which had defined the limits of Houses 2 and 3 is gone. Now the plinth is the ground line of the entire site. And now the square columns become the same width as the line of the roof and thus visually connected to it. The spandrel line now appears to support and give definition to the glass envelope so that a delicate levitational quality is obtained from the roof-column umbrella, which seems to exert a downward force on the internal space as it comes to rest on the ground. Where the concept of the roof umbrella is reductive and progressively more obvious in Mies, in Hejduk the same concept is developed but in more subtle terms. This subtlety has the effect of impacting the meaning of each articulation of column and wall not only with the mere expression of literal structure but also with the manifestation of architectural content itself.

A second aspect of work on the elements of the language in the Texas Houses concerns the articulation of the vertical dimensions. Houses 4, 6, and 7 present this second aspect. The concern is now shifted from the horizontal development of space to the vertical development of volume. Even the two site plans (none exists for House 7) suggest a new concern for the relationship of the interior and exterior horizontal plane. Whereas in the first three houses the house was conceived of as a series of horizontal layers, albeit as centralized pavilions or nodes on a progressive series of axes, they received their energy from the ground as datum. Now Houses 4 and 6 not only work contrapuntally with their particular sites, but the datum is shifted from the horizontal to the vertical plane. Whereas the first three houses extended horizontally, their architectural energies turned inward, in Houses 4 and 6 the opposite condition occurs. The architectural energy is still, in a sense, turned inward but there is vertical enclosure. There is a further disjunction which reveals another change in the concept of the house-site relationship in Houses 4 and 6. In the first three houses, the house was a microcosm of the site, in a way recapitulating its organization. In Houses 4 and 6, while the energies extend out, they do so in a jarring way; they are not cohesive but disruptive. In House 4, for example, a minimal curvature appears for the first time at the corners of the paved forecourt. In its position and size it gives the sense of a compressed figure, blunted by the presence of the house, which acts as a buffer or counter-force to the directional movement of the entry drive. In House 6, a second kind of disturbance is produced by shifting the drive, forecourt, and garage-servant quarters off-axis from the body of the house. Instead of compression, a set of crisscrossing shears is set up which threatens to rip the plan of the ground into four parts: the shifting of the garage and forecourt off the axis of volume of the house creates an initial shear, inducing a piston-like energy parallel to the grain of the house; however, the alignment of the rear of the garage and the front of the house causes a secondary shear line perpendicular to the dominant grain of the house. These two energies create for the first time a countermanding spatial condition, in which the house is held together only by the secondary suggestion of a time warp.

The concept of trace also begins to appear both externally and internally with House 4. The internal plans of Houses 4, 6, and 7 must be seen in their relationship to the vertical surfaces, and thus rather than horizontals which layer space, these plans become elements in a three-dimensional volumetric cage. This is first seen in House 4 where the traces of the interior openings in the solid panels, the doors, are marked by vertical mullions on the facade. In House 6, the volumetric nature becomes obvious as the nine-square grid in plan also becomes the facade. Now, for the first time, there is a full three-dimensional lattice. But two gestures reveal that this lattice is more than a grid of neutral space, but also contains a series of positively charged volumetric elements. Four of the nine quadrants in plan and elevation are cut away in the lower left corner and articulated as a double height volume. This leaves a three story L-shaped volume in plan and in section as a residue. To underscore the existence of these interlocking volumes within the grid, the garage and servant quarters sits on the site as an exact, positive replica of the volume which has been cut out of the interior. Second, an L-shaped volume, the stair tower, is pulled from the volume to again replicate the L-shaped volume of the interior. Here an idea of transposition is added to the idea of trace as a means by which the time of process—the internal, architectural ‘history’ of the house—is recorded.

The final aspect of this work in the vertical dimension is manifested in the nature of the vertical divisions, the now totally ‘voided’ column which becomes a time trace of the former solid one. To understand this development in the later houses one must return to the first four houses (Houses 1, 2, 3, and 5). In House 1 all the columns were solid and square. They were, in a way, a system set free of the floor and ceiling in that the columns and wall panels always intersected a grid line at a mid-point or center line; it was the grid divisions of floor and ceiling themselves which defined the outer edges of the interior spaces. In House 2 the columns remained solid and square but they now left a trace on the floor and ceiling which articulated the nine square divisions; these divisions no longer defined the center line of the columns, but their outer edges. Thus, the columns, floor and ceiling became one system. However, the major wall panels were still set on the center lines of the floor grid so that a disjunction occured between two systems in the vertical dimension, one a wall-column system and one a column-floor system. House 5 completed the sequence of solid, square column houses. It reduced the complexity of the system by eliminating internal partitions. There were two clearly separate systems: first the sixteen square columns and their interval banding, a time ‘trace’ of the former wall divisions marked on the floor, and second, the glass envelope, now located off the center line and pressed to the forward edge of the column line. All marking of the former center-line system disappeared, and with it the idea of dimension as manifested in the binary system of center lines and edges began to recede as a concern.

These three houses were for Hejduk the substance of a purely architectural as opposed to pictorial expression precisely because the idea of materiality in the form of dimensions (there are no literal points in architecture) distinguished the development of the nine squares in plan as opposed to the nine squares in elevation. In fact, as Hejduk moves more and more toward a three-dimensional grid (an object which he apparently considered non-architectural in his early houses) he abandons one aspect of materiality and objecthood-connection and dimension-for another aspect-the nature of the vertical surface itself as manifested in traces and transpositions. The former aspect had been seen as a concept of space; the latter is related to a concept of time.

Interestingly, in House 3 the center line reappears not as a grid but almost as the residue of a grid—in the form of H-shaped columns. These suggest two further changes. There are now two sets of columns: void webs and solid squares. And while the floor banding introduced in House 2 remains (in the width of both the solid and void columns), there are now two types of walls. Walls which come off the closed side of the H (which are parallel to the plane of entry) are all the full width of the column and the floor bands. Panels which run orthogonal to these walls remain thin (the width of the panels in Houses 1 and 2). However, they are now moved from the center lines to either the forward or rear edge of the floor bands. Even the floor marking between the interval bands has changed. Instead of being a series of different sized grids, it is now drawn as mono-directional floor boards, parallel to the grain of the front and back porches, setting up a dominant grain in what would otherwise be a biaxial house. Thus, by House 3 there are two interlocking systems: a column system of presence and absence, and a panel system of expansion and contraction; the former has to do with time, the latter with space.

Houses 4, 6, and 7 develop the idea of presence and absence in a multi-story (as opposed to single story) house. While the matrix of grid intervals is reintroduced on the floor, the webs of the void columns of Houses 4 and 6 are the last vestiges of the center line. The internal vertical surfaces of Houses 4 and 6, when they are superimposed on the floor grid, are either solid (running parallel to the direction of entry) or void (running orthogonal to the direction of entry). Interestingly in both cases it is the void marking which leaves a pair of void traces, parallel to the entry movement, on the roof plane of each house. In both houses the interior panels are thin planes which align with the outside of the interval bands. The exterior glazing on the two ‘open’ sides of Houses 4 and 6 are also moved to the outside edge of the frame.

Thus house 4 (1956—1960), even though it predates House 5 (1960—1962), represents a conceptual and spiritual break with Houses 1, 2, 3, and 5. Not only does it manifest a methodological development in its introduction of vertical circulation—it is the first two-story house—but now and more importantly it seems a more personally assertive house. The language is somehow distorted—there is a time warp in the trace of the interior doors on the exterior intermediate mullions. The plan is no longer either Miesian or Palladian. In this sense, the purging of the influences that were present in Houses 1, 2, and 3 is complete. But it is still a transitional house since the De Stijl and Corbusian influences which will appear in the later houses—the free plan and the free facade—are not yet present. In a curious way it is closer to the pavilion-like spaces of Louis Kahn than to any other source. But it is its distinction from Kahn that should be noted. The grid divisions which are the width of the void columns create a grid-like interval or division of the square plan. The nine resulting squares are now compartmentalized volumes resembling the early Adler and Devore Houses of Kahn. But whereas in Kahn the bearing or point structure is the outer layer of each pavilion, creating a void in between, in House 4 both solid structure and void structure are overlaid in a single plaiding. Thus the interval between the volumes is either solid or void but in both cases ‘positive,’ and the outer edge of the volume of each quadrant is void or negative. It is this positive internal interval which, for the first time, defines each quadrant as a volume. It also sets up a contrapuntal cross-layering of presence and absence, for the intervals parallel to the plane of entry are essentially solid while the ones perpendicular to the plane of entry are void. Thus on the entry and rear surfaces the frame becomes a void, and on the sides it becomes a solid. This is a clear reversal of the Miesian proposition; but there is also a subtle recall of Mies in the definition of the horizontal planes. In the first three houses these surfaces were articulated in such a way that the horizontal surface—the plan—is Miesian, and their ends—the elevation—is not. Now, even with the addition of a third slab (more like the Dom-ino section than the Farnsworth House or Barcelona Pavilion) the horizontal plan surface is no longer Miesian. Nor is it laminated as before; rather the house is sliced into three vertical volumes by the two void slots that run from side to side.

There was no differentiation in Houses, 1, 2 and 3 between the ground plane, the plinth and the roof plane. With the introduction of the second floor in House 4, these three are now treated in the same manner—as equivalent horizontals. However, unlike Dom-ino and Farnsworth, which imply continuous and ideal horizontal extensions, these horizontals are merely the elements of a three-dimensional frame (which becomes complete in House 6 with the addition of a third floor). Rather than implying the extension of space, they suggest the containment of volume. This can be seen in the intermediate mullions on the vertical exterior surfaces. In their irregular asymmetrical locations, they initiate a contrapuntal relationship to the frame. They also are actual projections of each of the door openings on the interior, from the center to the periphery, thereby creating secondary wedges of space, which anticipate the similar development in Hejduk’s later Diamond houses. As such they suggest a peripheral syncopation not unlike Mondrian’s Boogie Woogie paintings. This peripheral energy also creates a difference in the center. Where the centers of Houses 1, 2, and 3 were main living spaces—they were open and by their clerestory projections suggested movement up through the roof—now the center is given over to circulation and is closed. Further, while there is literal movement up via the stairs, there is no clerestory projection (only a skylight). Thus, the upward movement of the first three houses from the ground through the roof is here conceptually contained and denied. Now the pressure is in the vertical plane, from the center to the periphery.

House 6 as a multi- (three) story house must be seen in relationship to House 4 and House 7. It continues the transition from a horizontal to a vertical datum—from the horizontal layering of space to the vertical containment of volume—and also shows the change from Hejduk’s thinking of 1954. The house is a three-dimensional grid. It has the same initial plan parti as House 4, nine square bays articulated by sixteen H-shaped slots at the corners. Again the intervals between the bays are treated as bands—void coming off the open side and solid coming off the closed side of the H-shaped slots. Both House 6 and House 4, since they have more than one floor, deal with a condition of vertical circulation. But here the similarities end. In House 6 the circulation is pulled from the main body of the House and articulated in a tower with a three-level bridge connection. This means that circulation with the main body of the house must take place from room to room. While the spaces in the cross-wall system of House 4 are made accessible by the central location of the stair, in House 6 this is not possible. Instead, a secondary circular stair is inserted to allow for connection between the two public levels.

In its irresolution of circulation, House 6 points up the extraordinary clarity of House 4. It is also a precursor of House 7. On first impression the plan and the elevation of both these houses appear to be the same—a three-dimensional grid matrix. But while in 1954, the three-dimensional grid appeared to Hejduk to be non-architectural, precisely because it did not acknowledge the plan—the difference between the ground line, the roof line, and the vertical plane—and because it did not acknowledge the idea of dimension itself, House 6 represents the transition from the dominance of the horizontal plan and roof ending in House 5 to the vertical dominance in House 7. Thus, House 7 is different volumetrically from both House 4 and House 6. While it presents two solid and void planes as in House 4, and a three-dimensional nine-square grid in both plan and elevation as in House 6, the volumetric energies at work from the inside to the outside operate in a totally different manner.

Of the ten houses projected in 1954, House 7 was the last one done (1962—1963). It also was the last of the pencil drawings. There is no site plan; rather the house itself implies what the site plan might be. And there are only six plans and four elevations and an unfinished axonometric projection. The vertical surfaces of House 7 suggest a different conceptual spectrum from the previous houses. In the two elevations perpendicular to the plane of entry (the side elevations), because the ground line is expressed in the frame and the vertical and horizontal edges at top, bottom, and side are the same, the vertical surface becomes a dialectical screen: both a plane (whose linear elements are read as the residue of an erosion) and a frame (whose grid is seen as the product of an addition of linear elements). The two elevations parallel to the plane of entry exhibit for the first time a consciousness of the vertical surface as something more than a frame. These two surfaces are conceptually solids which are progressively eroded from their base through a punctured plane in the middle story to a frame at the top, or, reading in reverse, a movement from most open on the top to most closed on the bottom—an upside-down piloti structure. Thus, the front and back facades can be read from bottom to top as A, wall; B, pier; C, column, while the two side elevations are read, depending on how the spandrels are interpreted as either ABA or ABABA. Thus, the originally neutral nine-square condition in plan becomes articulated in elevation.

Equally, this neutral plan becomes in reality a series of layers which act on the exterior elevations as a centrifugal vortex from a complete solid in the basement releasing itself in section to a point condition at the top. This is no longer a horizontal sandwich of space, nor even a three-dimensional grid. It is conceptually two vertical planes containing a vortex of centrifugal action which erodes the horizontal layers. Thus, it is a precursor of Hejduk’s later wall houses, where the front and back walls will compress together, seemingly into one plane (but actually two distinct surfaces), and the centrifugal movement in the form of an exterior stair will move to the periphery. In this sense House 7 can be said to be an initial destruction of the box masked by the two eroding walls.

This concept of the vertical surface as a mask is also implied by a distortion of scale. This distortion suggests for the first time in Hejduk’s work what can be called a self-referential condition of the vertical plane. For the mask, in its alteration of scale, speaks neither of the interior nor the exterior site, nor of any metaphorical resonance; rather it reveals the physical presence of the plane itself. On initial appearance, the front and back of the house appear to be three stories, but in fact the three horizontal layers of column, pier, and wall cleverly mask six internal levels. The three-foot spandrel divisions, both void and solid, are the agents of this masking. Sometimes the top of the spandrel aligns with the floor line, other times the bottom is in alignment. This fluctuation creates an ambiguity concerning where the actual floor line is and how many there are. Thus, as has been said, the facade openings reflect neither internal accommodation, external context, nor metaphoric significance. They refer merely to conditions of the plane itself, introducing through the act of erosion the concept of progressive time. Further, the insertion of the spandrel voids creates in the vertical dimension a piston-like energy in each of the three bays: in the rear the central bay seems compressed and the side bays released, while in the front the addition of a void spandrel in the two side bays creates a compression in these two bays, causing the central bay, which otherwise is in a neutral condition, to take on qualities of release. Compression and release play contrapuntally against both the neutrality of the nine-square frame and the progressive erosion from the base to the top. Thus the subtly positioned solids and voids become in effect vertical traces of the void columns of House 6, now projected horizontally onto the facade which is read as almost a negative of the plan. They have nothing to do with Gestalt figure and ground perception, but rather internalize the significance of the vertical plane, thereby creating a purely architectural content.

While these first seven houses of John Hejduk are a synthesis of abstraction and reality, while they use a reduced vernacular of the twentieth century in a sixteenth century manner, they are at once prisoners of the past with an authenticity and objecthood of the present. As in Proust’s Remembrance of Things Past, they contain a conceptual overload, a density and compaction of themes, which produces a clarity of another kind. Within all of the necessary conditions for building they reveal a parallel energy: the existence of architecture itself.

In the end these houses do not bring us to confront classicism or modernism as a decisive issue, or cause us to distinguish between Europe and America, Mies and Le Corbusier, but rather they articulate an intrinsic architectural language without which architecture cannot be said to exist.

First published in: Kenneth Frampton (editor), John Hejduk: 7 Houses (New York: Institute for Architecture and Urban Studies, 1980). © CCA

AUTHOR

Peter Eisenman is an architect and architectural theorist who is internationally renowned for both his building designs and his writings. He is founder of Eisenman Architects and Professor in Practice at the Yale School of Architecture. Eisenman’s latest book is Rewriting Alberti, forthcoming in September 2025.