The Tacoma Narrows Bridge in Washington State, which collapsed in high winds shortly after it was built, November 1940

Throughout my adult life, I’ve crossed the Crown Point Bridge connecting Vermont with the Adirondacks across Lake Champlain. Built in 1929, it was one of the first continuous truss road bridges constructed in the US. Then governor Franklin Roosevelt addressed the celebration, which drew 50,000 farmers and their families from both sides of the lake; the new span heralded a “wedding after more than 150 years” between the two states, he said, and added that he was sure there “would be no divorce.” Airplanes crisscrossed the sky overhead, and according to the New York Times correspondent, one pilot “risked his neck in a sudden swoop beneath the arched roadway of the bridge, with less than a 100 feet headroom.” But the grand celebration was marred by at least one small failure:

A specially constructed ferryboat …was scheduled to be destroyed by dynamiting to symbolize the passing of an era, but something went wrong with the charge and the boat continued to float arrogantly when the throngs started homeward along roads choked with traffic.

Henry Petroski would not be startled by that small failure, nor by the larger failure of the entire bridge after eighty years. Indeed, he briefly describes in his engaging book the October 2009 inspection that discovered that the Crown Point Bridge was badly cracked—so badly that a few weeks later it was closed forever, and then demolished with high explosives to make sure it wouldn’t fall on passing boaters. In fact, I think it’s fair to say that no failure surprises Petroski. His classic first book, To Engineer Is Human (1985), whose title sets up twenty-seven years later this book’s pun, also dealt with failure, as do many of his columns in The American Scientist. “A single failure…is a source of knowledge we might not have gained in any other way.” They reveal “weaknesses in reasoning, knowledge, and performance that all the successful designs may not even hint at.” “The best way of achieving lasting success is by more fully understanding failure.”

It’s also the best way of entertaining an audience not necessarily gripped by engineering as a topic. Dams and bridges have their beauty—but collapsed dams and fallen bridges can be more fun to describe, albeit ghoulish fun. Reading these pages reminds us of how many spectacular failures have occupied the news pages for a week or two in our lifetimes: the Aloha Airlines flight where part of the fuselage peeled away, the collapsed cranes in midtown Manhattan, even the wreck of the Edmund Fitzgerald. (In recent weeks, of course, it has been the horrific collapse of a Bangladesh garment factory that has captured global attention, though that seems less a story of engineering stupidity than human cupidity—indeed, the day before the collapse, an engineer, noting a large crack in the structure, urged its evacuation.)

But it’s bridges above all that preoccupy Petroski: pedestrian bridges that have to be closed when the rhythm of foot traffic causes them to sway alarmingly; bridge cables that corrode, wire by wire, steadily lessening the “safety factor” of the span; bridges that collapse while under construction like the 1907 tragedy at the Quebec Bridge across the St. Lawrence1; or now-forgotten disasters like the 1967 collapse of the Silver Bridge connecting Point Pleasant, West Virginia, to Gallipolis, Ohio. That bridge, in use for half a century, went down ten days before Christmas, killing forty-six and severing US Route 35; LBJ immediately established a President’s Task Force on Bridge Safety, which went to work collecting eyewitness testimony and investigating the remains as they were dredged from the waters of the Ohio.

These forensic inquiries are rarely easy and straightforward—popular explanations talked of sonic booms. As Petroski writes, more likely possibilities included the fact that cars and trucks had tripled in weight since the road was built, adding big new stresses. As the months wore on, investigators closed in on their target, establishing that the suspension chain on the upstream side had failed first, and eventually establishing exactly what had failed. Attention became focused on the “eyebars,” the metal bars with a hole or “eye” at each end that could be linked together like a chain to hold up the bridge:

The Safety Board finds that the cause of the bridge collapse was the cleavage fracture in the lower limb of the eye of eyebar 330 at joint C13N of the north eyebar suspension chain.

The bridge was built in such a way that these metal rods were susceptible to corrosion and fatigue cracking, and couldn’t really be inspected; a bridge of the same design ninety miles upstream was dismantled as a precaution, and eyebars fell out of favor for suspension bridges as engineers returned to wire cables.


If you’re like Petroski (or me) you find that kind of specificity (“eyebar 330 at joint C13N”) reassuring; in fact, the rational language of the engineer is one of the great American vernaculars, and perhaps the best part of To Forgive Design is Petroski’s account of one of the founts of that sort of analysis, the engineering departments of the nation’s great land-grant colleges. Petroski arrived in the 1960s at the University of Illinois, where his desk as a graduate student was in the Talbot Laboratory of the Department of Theoretical and Applied Mechanics. Academic offices occupied part of the building, but its huge atrium, the “crane bay,” was dominated by a three-million-pound capacity testing machine, essentially designed to put stress on metal or stone or any other material until it broke. Alarms would sound when testing was about to begin, alerting occupants of the lab

that something was about to fracture—releasing enormous amounts of energy in the form of an explosive noise accompanied by a recoil of the megamachine that made the entire building shudder.

The great testing machine—really, a failure machine—was the symbolic as well as the literal center of the enterprise,

always at the ready to apply real forces to confirm or contradict theoretical predictions of strength. It was the final arbiter between competing theories and contradictory predictions…. Failure was the incontrovertible counterexample to putative success.

If you want to understand what made postwar America the greatest design and engineering and manufacturing powerhouse the world has ever known, that hulking machine standing in a brick building on the edge of the prairie and attended by postdoc acolytes with slide rules at the ready seems like a reasonable symbol. Forget the virtual, the postmodern, the digital and pixelated; here’s a monster crushing concrete columns, figuring out the ways things work in the realest of worlds.

It was not an easy time to be a brush-cut engineer, however, even in the American heartland. Vietnam protesters, among many other targets, assailed their fellow students who aimed to work at companies tied to defense industries. Partly in response, engineers organized a new order for the profession, and began holding ceremonies where graduates took an oath to

practice integrity and fair dealing, tolerance and respect, and to uphold devotion to the standards and the dignity of my profession, conscious always that my skill carries with it the obligation to serve humanity by making the best use of Earth’s precious wealth.

Petroski applauds the effort, but then explains that it followed an older and humbler Canadian tradition, designed to instill in young engineers a strong sense that failure was always stalking their work. Not many years after the 1907 collapse of the Quebec Bridge, engineers in Canada aimed to improve the standards of their profession, and came up with not just an oath, but a ring ceremony—at first, in fact, the pinkie rings awarded the new engineers may actually have been cast from the bridge’s fallen iron girders. An anvil would be struck seven times, representing the driving home of a rivet, and then a Kipling poem, “Hymn of Breaking Strain,” would be read. It begins:

The careful text-books measure
(Let all who build beware!)
The load, the shock, the pressure
Material can bear.
So, when the buckled girder
Lets down the grinding span,
The blame of loss, or murder,
Is laid upon the man.

Not on the stuff—the man!

Petroski remains a son of the twentieth century, confident that the creed of the engineer is unchanged and sufficient. You could prevent failure, he writes, but only if you could “freeze technology at its present stage. Everything made subsequently would have to be designed and produced exactly like what had already been demonstrated to have been successful.” Instead, “endless improvement is what engineering is all about,” even if that inevitably leads to some failures (from which we then hope to learn). And in this enduring chain of progress, “good war stories are never obsolete.”

But what if, in fact, the old war stories are becoming obsolete? The engineer, like the insurance agent, is hampered by the fact that his skill depends on the earth behaving in the future as it has in the past. As Petroski writes,

Since it is future failure that is at issue, the only sure way to test our hypotheses about its nature and magnitude is to look backward at failures that have occurred historically. Indeed, we predict that the probability of occurrence for a certain event, such as a hundred-year storm, is such and such a percentage, because all other things being equal, that has been the actual experience contained in the historical meteorological record.

That record, however, is now shattered. In the course of Petroski’s lifetime, and all of ours, we’ve left behind the Holocene, the ten-thousand-year period of benign climatic stability that marked the rise of human civilization. We’ve raised the global temperature about a degree so far, but a better way of thinking about it is: we’ve amped up the amount of energy trapped in our narrow envelope of atmosphere, and hence every process that feeds off that energy is now accelerating. For instance, this piece of simple physics: warm air holds more water vapor than cold. Already we’ve increased moisture in the atmosphere by about 4 percent on average, thus increasing the danger both of drought, because heat is evaporating more surface water, and of flood, because evaporated water must eventually come down as rain. And those loaded dice are doing great damage. The federal government spent more money last year repairing the damage from extreme weather than it did on education.


Engineers try to cope with these changes, of course, especially those on the front lines. So, for instance, public works departments across the Northeast have begun replacing the pipe-like drains called culverts as fast as they can, swapping the size that the textbooks recommend for the much larger diameters the new rainfalls demand. “The books we’ve always used to design culverts, you can throw them all out,” Dave Wick, district manager of the Warren County Soil and Water Conservation District, recently said. “What was once called a 100-year event is now a 50-year event, and a 50-year event has become a 25-year event.” And indeed even those numbers may be optimistic—James Hansen, who retired earlier this spring from NASA, calculated recently that, as a summary put it, events like

the recent Texas heat wave, Moscow’s heat wave the year before, and the 2003 heat wave in Europe now occur twenty-five to fifty times more often than just fifty years ago.

The first problem for engineers, of course, is that the changes you need to make to deal with these shifts—bigger culverts, different pavement that won’t buckle in the heat, more water storage behind higher dams—all lead to more expense. The Crown Point Bridge, for instance, was lost due to engineering failure after eighty years; now a new one stands in its place, presumably good for a similar term. No one can really complain. But Vermont lost many more bridges in 2011, some of them covered bridges that had taken everything nature threw at them for centuries, only to be swept away in the record rainfalls that accompanied Hurricane Irene.

As they’re replaced, prudence demands that they be built much longer than their predecessors, so that a similar flood won’t wash away the banks that hold them—in fact, a state report on the flooding specified exactly that measure. But a longer bridge is a more expensive one and, as Petroski points out, length adds stress that will lead to more deterioration unless a span is carefully maintained. (And, of course, it makes it harder to rebuild the covered bridges that are highly useful to state tourism officials.) Hurricane Sandy, a year after Irene, demonstrated far more fundamental weaknesses in our technological civilization—a subway system, say, is an engineering marvel right up to the moment when it fills with seawater.

A deeper problem, however, is that there’s no new normal to aim for, no way to reestablish the textbook formulas that served us well. We’ve increased the temperature one degree so far, but the same climatologists who predicted that rise also tell us that unless we can quickly break our addiction to fossil fuels, we can anticipate four or five degrees as the century wears on. Each increment adds new energy to the system, and at the upper boundaries engineering as we’ve known it becomes very nearly impossible. If the atmosphere is 8 percent wetter instead of 4 percent, what kind of bridges do we build? Where do we put roads so they’re not washed away? What, if anything, should we be building in those zones (like, say, Manhattan) that are only a few meters above a rising sea? If the danger of forest fire grows constantly, what kind of building codes do we need for construction in the West?

These are problems not just for the engineer, but for his constant companions, the bond salesman and the insurer. Each of these has to assure himself that new projects are within some new margin of safety so that their investments and estimates of risk make economic sense; if they don’t, the cost of building bridges will reach the level where we go straight back to ferries.

If the engineering goals of the past were to build longer bridges and higher skyscrapers and cheaper, more graceful structures generally, are those still sensible goals? Or in a more difficult world, might we choose a different set of targets, and in the process change many professions, engineering included? You could argue, I think, that a less hospitable earth might, in many places, dictate a design style geared toward the squat, the durable, the hardy. Instead of a few big, inherently vulnerable structures (giant power plants, say, that could be taken out by a flood or a storm), engineers are increasingly interested in “distributed generation,” the idea of a thousand or a million rooftops covered with solar panels and each feeding into a grid. Such new arguments present engineering challenges of their own—how do you store the power when the sun doesn’t shine, or redirect it from sunny places to dark ones? But it’s clearly less vulnerable to catastrophic failure.

In fact, most of what Petroski describes are all-or-nothing failures. The bridge works, or it falls. But in a tougher world we’ll need more structures and systems that can fail gradually, where problems don’t immediately ramify into catastrophes. If a flooding river washes away my house and with it my solar panels, I’ll be miserable, but the whole East Coast doesn’t go black. By Petroski’s standards such a loss would be an obvious failure. But by the standards of a far more turbulent and difficult world, it may also be a kind of success. The Army Corps of Engineers worked the better part of a century to make sure the lower Mississippi never flooded, but when record volumes of water came down the river in the spring of 2012, they were able to blow up some levees and inundate farmland, sparing cities. That was an engineering accomplishment as much as a setback.

This kind of thinking extends beyond threats to physical objects, of course. Twice Petroski mentions in passing the current financial crisis, but in neither case, oddly, does he bring up the phrase that became emblematic of the saga: “too big to fail.” In political terms, that apparently meant “we must bail them out.” But to any thinking person watching from the sidelines, there was another obvious implication. Anything too big to fail was too big. Period. The remedy was to make it smaller—a thousand small banks, not six big ones.2 So far those big banks have used their political clout to resist such rearrangement, just as the big energy companies have fought small energy sources. But if Petroski’s account proves anything, it’s that the forces of the real world may eventually prevail on even the mightiest structures.