When the wooden forms came off Fallingwater's first cantilevered terrace in 1936, the concrete deflected an inch and three quarters, far more than a design this ambitious should have allowed. Mendel Glickman, the engineer Frank Lloyd Wright had trusted with the calculations, is said to have exclaimed on hearing the news, "Oh my God, I forgot the negative reinforcing." He had left out steel needed to resist tension in the top of the beam, a basic omission dressed up in an otherwise daring design.
The contractor building the house, himself a trained engineer, had caught a version of the problem earlier. Looking at Wright's drawings, he calculated the specified eight reinforcing bars were at least fifty percent under strength for a cantilever of this span, and he raised the alarm with the client, Edgar Kaufmann Sr. Rather than confronting Wright directly, the contractor quietly doubled the reinforcement before the concrete was poured, protecting the house without waiting for an argument he expected to lose.
He should have expected the argument, because Wright got wind of the change anyway and was furious, not grateful. He wrote to Kaufmann in terms that made the stakes plain: "I have put so much more into this house than you or any client has a right to expect that if I haven't your confidence, to hell with the whole thing." It fell to Kaufmann to calm his architect down so construction could continue, a strange position for a client whose engineer had just quietly saved the project from its own architect's mistake.
Doubling the steel did not fully solve the problem, because the contractor had not built in the pre-cambers, a slight upward curve poured into the formwork to offset expected sag, that Wright's own office had specified. The terraces still deflected once they carried their own weight, just less than they would have otherwise, and the sagging continued slowly for decades: by the late 1980s the main terrace had dropped as much as seven inches from level. In 2001, the engineer Robert Silman completed a six-month, four-million-dollar repair, threading post-tensioning steel cables through the original 1936 beams to arrest the movement for good, without changing anything visible about the house Wright had designed.
Historians have mostly been gentle with Wright about all of this. Michael Sorkin argued the flaw does not diminish what may be the best house ever designed, and Robert A. M. Stern made a similar point about experimental architecture more generally: working at the edge of what is structurally possible means accepting real risk, and Fallingwater is still standing, drawing visitors by the hundred thousand each year, seventy years after it nearly did not.