by Jim Atkins, FAIA

Summary: In this, the fifth of the series, Jim Atkins takes us on a forensic exploration of the nation’s most famous house, including a few lessons on the structural properties of reinforced concrete, both initially and over time under the cyclical duress of sagging, cracking, and weathering—properties much better understood today than in 1935.

“The calculated stresses in the structure do not fall within the limits of those prescribed by accepted engineering practice … therefore the structure does not have a satisfactory factor of safety …”
—Metzger-Richardson Engineers, addressing Wright’s cantilever design, 1935

View of the terraces over Bear Run. Photo by James B. Atkins, courtesy of the Western Pennsylvania Conservancy.

As described last month in Part 4, construction finally began on Fallingwater after a much delayed start. Most of 1936 had passed without significant progress, beginning with Edgar Kaufmann’s doubts about the structure, followed by late changes by Wright in the supporting concrete bolsters, and further complicated by the difficulty in finding a competent contractor.

Finally, when work began in earnest the project was plagued with yet another challenge—one with potentially fatal consequences: a progressing failure in the building’s structure. Join us as we experience the ultimate challenge in constructing the country house known as Fallingwater.

A complex design
Perhaps the most striking single feature of Fallingwater is the cantilevered reinforced concrete balconies that extend out over the falls. Such cantilevers were a relatively new design element in the 1930s, and reinforced concrete technology itself had not progressed much beyond simple, direct bearing configurations. Given the limited use of such designs at that time, it is understandable that engineers may not have fully understood the forces imposed in a cantilever.

As the use of concrete has become more widespread over the years, more has been learned, and today there are many tools to employ, such as higher strength concrete, high-tensile reinforcing steel, and more advanced curing techniques to apply after the concrete has been placed in the formwork. Reshoring, or temporary bracing placed under a suspended concrete structure for a period of time after the formwork has been removed, usually requires its own engineering design. Reshoring prevents the concrete from moving, or in this case deflecting, until it reaches its ultimate design strength, typically calculated at 28 days after placement.

Concurrent with the construction of Fallingwater, the mushroom-shaped columns at the Johnson Wax Company building designed by Wright and the same structural engineer were being challenged by local authorities. In that case, they load tested a column to failure to verify its capacity. Kaufmann’s country house was not a publicly accessed building, and it was not in a city that required a building permit. There was no third-party authority to oversee and agree with this unique structural design.

It is easy to blame Wright for his failed cantilevers, but like the architects of today, he relied on technology available at the time to deliver his product. The awe of his work is that he actually produced such an innovative design with those materials in that time and place.

Frank Lloyd Wright and the Cantilever
By the time Wright designed the terraces at Fallingwater he was already well acquainted with the cantilever, having used it in his Robie house in 1906. Although the cantilever had come into its own in Asian bridge designs during the ancient times of Marco Polo, it was Wright’s manipulations of the form that raised it to the level that we respect and admire today.

Other notable Wright designs incorporating the cantilever include the Guggenheim Museum and the Marin County Civic Center. A lesser known, yet equally profound design is the 14-story Johnson Wax Research Tower in Racine, Wis., the second phase to its more popular sister, with its floors completely cantilevered from a 14-foot-diameter central core. And rising from the grave is the Massaro House on Lake Mahopac, 50 miles north of Manhattan, cantilevering 25 feet over the water. Designed by Wright in 1952, it remained unbuilt until completed by Thomas A. Heinz, AIA, in 2007. While the work is not recognized by the Frank Lloyd Wright Foundation as a Wright creation, the cantilever appears true to Wright’s original design.

The cantilever was considered by Wright to be the second principle of organic design, the regular grid being the first. He was obviously undaunted by the challenges of construction at Fallingwater since at least four more cantilevered designs sprang from his pencil in the years that followed. Some of his Massaro House drawings consist of 5-by-8-inch preliminary pencil drawings that had to be deciphered by Thomas Heinz, and one can only wonder as to the cantilever treasures possibly buried in archives or discarded in the trash can of his drafting room.

The definition of failure
The world’s most successful summer house was a failure; at least where its structure was concerned. Fallingwater experienced a structural failure in its concrete frame that would have eventually progressed to a catastrophic result. Structural failure in reinforced concrete can be progressive, and it typically begins with cracking followed by excessive deflection. This is precisely what occurred at Fallingwater while construction was still underway. Cracking and excessive deflection are by definition a structural failure because the design is not intended to behave in that manner.

Edgar J. Kaufmann jr., in his book, Fallingwater, A Frank Lloyd Wright Country House, vigorously defends what he calls “The Faults of Fallingwater.” He begins with the rationale: “Mistakes have plagued Fallingwater, yet the extraordinary beauty of the house and the delight it brought to the life of its inhabitants form the context in which its construction should be evaluated.”

If only architects today could be looked upon with such sympathy and understanding; if today our honest mistakes could be understood when compared to the benefits we achieve with the performance of our project. If only our clients would speak as Edgar jr. spoke: “No apologies are necessary for what he [Wright] achieved at Fallingwater.”

We are all aware that in this time such failures are not tolerated whatsoever. If Wright delivered such a product to Kaufmann today, he would likely be sued for loss recovery against his professional liability insurance policy, even if he and Kaufmann remained good friends. Wright did not carry professional liability insurance because it did not exist at the time. Ironically, until professional liability insurance was invented in the 1950s, no one appeared to be interested in getting into the architect’s pocket. Considering Wright’s history of building design failures, if his work were produced in today’s market, he may not experience the same level of historical prominence.

The structural failure that plagued Fallingwater is filled with complexities that may never be completely revealed or resolved. But like the Torre Pendente di Pisa, the ending is a happy one because in 2002 the progressing failure was ultimately arrested with modern structural technology. And like the famous Italian bell tower, the existing deflections in the floors of this famous house now only add to its rich history and intriguing notoriety.

Lightning and inevitable thunder
As the formwork placement continued on the cantilevered slab at the first floor, Kaufmann became concerned about the design. He hired the Metzger-Richardson Company, engineers he had used for many years that specialized in reinforced concrete design and construction. They provided new design drawings on August 10 requiring over twice the steel in Wright’s design.

New steel was ordered, and it was delivered on the site on August 15. The placement of the concrete for the cantilever was scheduled for August 19. Walter Hall, Kaufmann’s contractor, subsequently added steel to the structure, but he did not add the full amount called for by Metzger-Richardson. The final, in-place design was more than Wright called for but less than Kaufmann’s engineers required. It is convenient to conclude that this is why the cantilevers continued to deflect and were ultimately determined to be structurally inadequate.

Wright knew nothing of the added steel until he received a letter from Bob Mosher on August 25. When Wright learned of the additional engineer and their recommendation for added steel, he was outraged. His letter to Kaufmann followed.

If you are paying to have the concrete engineering done down there, there is no use whatever in our doing it here. I am willing you should take it over but I am not willing to be insulted.

Wright and Kaufmann remained close despite the occasional lightning and thunder. Photo courtesy of the Western Pennsylvania Conservancy.

Wright chided Kaufmann for his provincial attitude and overriding the architect’s decisions. He told him it was not too late to get another architect, and he called his apprentice Mosher back to Taliesin. His closing was equally acrimonious.

I don’t know what kind of architect you are familiar with but it apparently isn’t the kind I think I am. You seem not to know how to treat a decent one. I have put so much more into this house than you or any other client has a right to expect that if I haven’t your confidence—to hell with the whole thing.

Kaufmann came back at him with a letter that closely mimicked Wright’s.

If you have been paid to do the concrete engineering up there there is no use whatever of our doing it down here. I am not willing to take it over as you suggest nor am I willing to be insulted …

I don’t know what kind of clients you are familiar with but apparently they are not the kind I think I am. You seem not to know how to treat a decent one. I have put so much confidence and enthusiasm behind this whole project in my limited way, to help the fulfillment of your efforts that if I do not have your confidence in the matter—to hell with the whole thing …

He concluded with a postscript that suggested they stop writing letters and for Wright to come to Pittsburgh to clear the matter up. He expressed regret that Wright was calling Mosher back to Taliesin, and he said that, because of their past good relationship, he would consider that Wright’s letter had never been written. He initialed the postscript as if it were a separate letter.

Wright, never the fool, began his reconciliation in a letter on August 31. However, the letter was more explanatory than apologetic, and Wright continued to stress that, “the details must be mine.”

A man’s work is his honor … had I thought less of you I should have been more polite, not outrageous … Apologies are nothing to a man like yourself. But explanation seems to be in order. The atmosphere should be cleared. Lightning and inevitable thunder may help to clear it …

Multiple causes and effects
Many design issues contributed to the structural inadequacies. First of all, no reverse camber was cast into the cantilevers. Reverse camber involves raising the forms upward to counteract the natural “creep” of the concrete downward. For example, if a cantilever is curved upward like a cow’s horn when it is formed, when the concrete naturally deflects due to creep, it will, by design, ideally move downward to a level position.

This photo shows the apparent lack of planning in placing the temporary shoring for the cantilevered terrace. Photo courtesy of the Western Pennsylvania Conservancy.

There is no record of a reshoring design, and the supporting braces placed after the forms were removed appear to have been randomly placed, more of convenience than design. The random reshoring is visible in an image recording the original construction. It is also not known how long the reshoring remained after the concrete was in place. Removing the supports before the concrete reached its full design strength could have contributed to the distress experienced at Fallingwater.

Cracks initially began to manifest in the second story parapets. When Hall placed the concrete roof over the guest room terrace on October 29, cracks immediately appeared. Wright was concerned about the cracks, and he had Hall take measurements and send them to Glickman. Hall discounted the cracks as caused by an absence of expansion joints. He wrote, “These cracks mean nothing in my opinion …”

On the spot in Pittsburgh
Wright continued to assure Kaufmann that the cracks were typical in the concrete curing process. Throughout November and December, cracking and drooping of the cantilevers continued. Wright had Glickman feverishly working to determine the cause. Wright himself had contracted pneumonia and physically struggled through the end of the year.

Meanwhile, Kaufmann hired Metzger-Richardson to do a full structural analysis. They reported back that the cantilevers had been overstressed beyond design, and they recommended permanent supports be placed under them.

By January Metzger-Richardson had declared the terrace at Kaufmann’s bedroom in danger of catastrophic failure, and workers installed a supporting wall under it. Metzger-Richardson performed load testing on the cantilever using bags of sand and cement and cast iron pipes for weight. Many overhangs were determined to be safe, but the tests caused cracks in Liliane’s terrace. Meanwhile, back at Taliesin, Wright admitted to Glickman: “We are on the spot in Pittsburgh.”

It took until the end of January for Wright to learn of the load tests. When he did, he telegraphed Kaufman with more lightning and thunder, and Kaufmann returned the same. Volatility was not in short supply. As their tempers ultimately cooled, Wright sent a printed form for Kaufmann to sign. It stated that Kaufmann agreed to make sure that the architect’s instructions would be faithfully executed, with no exterior advice or criticism that would interfere with the architect’s authority until the building was completed. The form was never signed.

In spite of progressing deflection and cracking, Wright continued to assure Kaufmann that the design was adequate. Kaufmann was apparently persuaded, and he directed Hall to remove the shoring that had been placed under the west terrace by Carl Thumm at the beginning of the year.

The cantilevers would now rest unsupported, proudly extending over the streambed below as Wright originally conceived. These conditions would remain until forensic evaluations revealed fatal developing conditions in the late 1990s. Meanwhile, Edgar Kaufmann Sr. would fret and worry about the cantilevers the remainder of his life, having them constantly monitored and measured.

Repair and redemption
Edgar jr. gave the house to the Western Pennsylvania Conservancy in 1963. They continued to monitor the progressing deflection problem, and by 1994 the deflection ranged from 4 to 7 inches in the most extreme conditions. The conservancy was greatly concerned, and the firm of Robert Silman Associates was hired to evaluate the problem and design a fix. Interestingly, this would be the sixth Wright house the Silman firm had worked on for structural problems.

Silman determined that the cantilevers were in danger of imminent catastrophic failure. Heavy, unattractive steel trusses were installed in 1997 beneath the cantilevers to prevent the house from falling into the creek until remediation could be completed. The visual elegance of Fallingwater was now completely compromised.

The floor of the Great Room removed for the repair. Photo courtesy of the Western Pennsylvania Conservancy.

Stretching things a bit
The Silman firm designed a repair using post-tensioned cables attached to the sides of the existing beams within the floor cavity. In 2001, the 600 flagstones on the great room floor were carefully numbered and removed along with the redwood subflooring. The post-tensioning contractor VSL, a pioneer in post-tensioning technology in America, installed a bundle of thirteen one-half-inch steel cables alongside each beam.

Post-tensioning tendons protruding from the Great Room terrace. Photo courtesy of the Western Pennsylvania Conservancy.

One end of each tendon bundle was anchored in a concrete block attached to the beam. The other “live” end of the cables passed through another anchor block in the terrace wall at the unsupported end of the cantilevered beam. Each end of the tendon cables was lower at each end than in the middle, where a higher block was positioned, configuring the cables in an arch. The cables were encased in concrete.

Post-tensioned Concrete
Post-tensioning is a method of reinforcing concrete with high-strength strands (tendons), anchored at each end of a concrete beam and stretched, or tensioned, after the concrete has reached a specific strength. Fallingwater was repaired with an “unbounded” post-tension design where the tendons are coated with corrosion-inhibiting grease and encased in an extruded plastic protective sheathing. The anchorage at each end consists of an iron casting and a conical, two-piece wedge which grips the tendon. The tendons are stressed with hydraulic jacks on the “live end,” and the excess tendon end is removed and the pocket is grouted. Post-tensioned construction has been used throughout the world since the 1950s, and the technology was not available to Wright and Kaufmann when Fallingwater was constructed.

Hydraulic jacks were attached to the terrace wall end on the protruding cables. Over a two-day period in the spring of 2002 the cables were stretched with the jacks until the end of the cantilever rose back up three quarters of an inch from its deflected position. The cables were stretched until they exerted a force of 195 tons on each side of the beam. The cable “live” ends were anchored, the loose cable ends were removed, and the connector pockets were grouted and finished to appear like the original surface. The progressing and ultimate failure of Fallingwater had finally been arrested and resolved.

Meanwhile, the long-term creep in the concrete had left the cantilevers in a permanent deflected condition that could not be remedied by the post-tensioning. Moreover, it was not the intention of the engineers to completely correct the sagging, as it is truly an integral part of the creation of the house. Wright created the drooping cantilevers as much as he created the house itself. Apparently imperfection, in some cases, can become an integral part of perfection.

The failure of the cantilevers of Fallingwater fortunately did not bring down the house, and the iconic design is no less evident or truthful because of the flaw. Fallingwater gave Wright new life as an architect in his later years, and it put Kaufmann out in front of his stuffy elitists Pittsburgh contemporaries. The exclusivity of the Duquesne Club pales in comparison to Kaufmann’s preeminent position as Lord of Fallingwater.

It gave the Kaufmanns a wonderful place to be together as a family, and it gave the American public, and the world, an iconic masterpiece to be freely enjoyed and that will likely never be surpassed. It is a solemn and profound example of how perfection, in its purest form, can be made up of the perfect and the imperfect.

Next month
So ends the story of the country house, Fallingwater. It is an American residential icon and the greatest and most complete example of Wright’s work that is available today to the public. It is lovingly and carefully maintained and managed by the Western Pennsylvania Conservancy, and it is available for everyone to visit and experience.

Join us next month for the final part of this series as this author and his wife visit Fallingwater in late May 2009. The weather is ideal, and one can relate to Kaufmann as he came to his country house in the spring to get away from the hectic atmosphere of the store in busy Pittsburgh.

We were treated most graciously by the members of the Western Pennsylvania Conservancy, particularly Clinton Piper, a true professional and fitting host. Join us next month on a trip that any architect should seriously consider if he or she has a meaningful interest in American architecture.

Meanwhile, good luck out there.