Big bridges are not as big as tunnels, dams, and canals, but they hold a special place in the human imagination. It is not just that the experience of crossing a bridge is memorable (in a different way, so is driving through a tunnel), or that a bridge, like a dam, performs an important economic function. The Suez Canal, over one hundred miles long, is an extraordinary engineering achievement, but it is, after all, a long and deep ditch. Great bridges like the Brooklyn or the Golden Gate, on the other hand, are not merely larger-scale versions of interstate overpasses, they are in a class by themselves.

Whether or not you know anything about bridge design, you can’t fail to be touched by their form. It is certainly possible for a large bridge to be ugly. The Quebec Bridge holds the record as the longest steel cantilever bridge in the world, but, to my eye, it has an awkward shape. It lacks the brooding presence of the second-place runner-up, the Firth of Forth Bridge in Scotland, which Alfred Hitchcock used to such good effect in The 39 Steps. The best bridges have an evocative beauty that is unlike anything else. They can be stolidly robust or soaringly graceful. They are utilitarian, but unlike transmission towers or fiber-optic cables, they have character. They also manage to give a sense of place to their surroundings—rural or urban. Transportation devices, civic symbols, landmarks, and sculptures—bridges are all of these.

Bridge building has a long history. Crude cantilever bridges of overlapping logs were known in ancient China, and the Mesopotamians and Egyptians built bridges of corbeled stone. It was the Romans, however, who perfected the technique of building stone arches, which they used to great effect in bridges, as well as in aqueducts and amphitheaters. Surviving Roman bridges in Italy, Spain, and France attest to the skill of these engineers. Then bridge building stagnated for several hundred years. Medieval bridges such as the Pont d’Avignon or the Ponte Vecchio in Florence are attractive but do not fundamentally improve on Roman techniques. Nor do Renaissance bridges such as the Pont Neuf in Paris or the Rialto Bridge over the Grand Canal in Venice. Like their Roman antecedents, these stone-arch bridges rarely exceed one hundred feet in span.

Bridges are sometimes ranked by their overall length, by the height of their towers, or by their height above water level, but the measure that really counts is the span—the clear distance between supports. Achieving as big a span as possible is often a major goal of bridge designers. This is partly a practical question. There are advantages to eliminating intermediate piers that are expensive and complicated to build (especially with underwater foundations) and that interfere with river traffic. But it is also the challenge of defying gravity. Spanning greater distances is a distinct measure of engineering prowess. Just as the early airplane builders were challenged to create machines that could fly farther and farther nonstop, bridge designers ventured to span greater and greater distances.

Beginning in the eighteenth century bridge builders were able to achieve larger spans chiefly thanks to the new methods of mathematical analysis that allowed a more effective use of materials. Waterloo Bridge, completed in 1817, consisted of nine stone arches, each with a 120-foot span. As early as the 1750s, several timber bridges were built in Switzerland with spans as great as 200 feet. As Henry Petroski points out in Engineers of Dreams, the longest wooden bridge was in the United States: the aptly nicknamed Colossus was built over the Schuylkill River at Philadelphia in 1812 by the engineer Lewis Wernwag. The trussed arch spanned 340 feet.

Wood has one obvious disadvantage—it is flammable. The Colossus was destroyed by fire in 1838, but by then there was another lightweight alternative to masonry: iron. The first cast-iron bridge was built across the Severn River in England in 1781 (it is still standing). It spanned only 100 feet, but less than twenty years later a 236- foot cast-iron railroad bridge was built by the Scotsman Thomas Telford across the mouth of the River Wear near Newcastle. In 1850, Robert Stephenson, who had helped his father George to build the first practical steam locomotive, the Rocket, designed a wrought-iron tubular bridge that spanned 459 feet across the Menai Strait in Wales. Wrought-iron chains made possible suspension bridges (the first such bridge was built as early as 1741), which were able to achieve yet larger spans. Wire cable was first used in place of the heavier chains in 1825 in Lyon; the first American suspension bridge to use strands of wire was the replacement for the Colossus, which was built in 1842 by Charles Ellet, a native Pennsylvanian.

Wire suspension bridges proved capable of spanning great distances. The first bridge to span more than 1,000 feet was built by Ellet across the Ohio River in 1849. That bridge was destroyed by wind five years later. It was rebuilt by John Augustus Roebling, who was to become the preeminent suspension-bridge designer and builder in the country. Roebling built a 1,057-foot suspension bridge at Cincinnati (now called the Roebling Bridge), the world’s first railroad suspension bridge over the Niagara Gorge, and his masterpiece, the 1,595- foot Brooklyn Bridge, which was scarcely begun in 1869, when Roebling died in an accident and his son Washington Roebling took over the work, eventually falling ill with decompression sickness suffered as a result of continuous work underground.


By 1883, when the Brooklyn Bridge was completed, the United States was in the middle of an unprecedented bridge-building boom, and for several reasons. Bridges are chiefly an urban phenomenon since they rely on heavy concentrations of population (lightly trafficked river-crossings can be made by ferry), and the last quarter of the nineteenth century as well as the first half of the twentieth coincided with an unparalleled expansion of America’s cities. Almost all these cities were beside rivers (or, like San Francisco, projected into bays), and the largest and most important city. New York, was surrounded on all sides by water. This guaranteed that bridge builders would be kept busy.

The size of most North American rivers posed a special challenge to bridge builders. Rivers such as the Mississippi, the Missouri, the Ohio, the St. Lawrence, and the Hudson were broader than most European rivers, and their being used for commercial navigation required bridges that both were tall and had long spans. Finally, until about 1920 Americans’ great desire for mobility was fulfilled chiefly by trains and resulted in railroad bridges like the Niagara Gorge (1854), the Poughkeepsie Bridge (1889), the Hell Gate (1917), and the Quebec Bridge (1918). After 1920, travel increasingly meant automobile travel, and bridges like the Delaware River Bridge (now called the Ben Franklin Bridge) linking Philadelphia and Camden, the George Washington Bridge, the Golden Gate, and the Tacoma Narrows Bridge were all designed to carry car traffic exclusively.

These conditions would have resulted, one way or another, in a flurry of bridge building, but the American boom also coincided with the advent of “structural steel,” i.e., steel specifically adapted for use in building. The first American steel bridge, across the Mississippi at St. Louis, was built by James Buchanan Eads in 1874. Structural steel was stronger and more flexible than either cast or wrought iron, and allowed a greater variety of design: it could be used to build arches, girders, cantilevers, and trusses, as well as high-strength suspension cables. So just when the American economy required bridges, the knowledge and techniques of bridge building reached unprecedented heights—or rather breadths. By 1950 the six longest spans in the world were all in the United States.

The great American bridge-building period lasted from the 1870s until the 1950s. It is the subject of Henry Petroski’s interesting new book, unfelicitously titled Engineers of Dreams. Petroski is the author of a well-received book, The Pencil, as well as The Evolution of Useful Things, and is chairman of the Department of Civil and Environmental Engineering at Duke University. He is obviously suited to explain the technical background of his subject, but his book goes much further. “To understand the works of engineers and engineering is to understand the material manifestations and progress of civilization,” he writes, and in his fascinating and lively account he places the achievements of the great bridge builders in their appropriate social, economic, and cultural setting.

Petroski concentrates less on theories of bridge designing, although these get their due, than on the bridge builders themselves, seeing the individual engineer as team leader, entrepreneur, and political operator. Bridge designers like Gustav Lindenthal (the Manhattan Bridge, the Queensboro Bridge, the Hell Gate Bridge), and Othmar Ammann (the George Washington Bridge, the Bronx-Whitestone Bridge) were not simply builders, they were also promoters who lobbied politicians as well as bureaucrats and financiers. Getting the commission was rarely easy. It was not simply that bridges were so expensive; there was also often competition, not only from other bridge builders but from tunnel builders. Competitive as this profession was, it was also curiously limited to a small cluster of technicians. Since most bridges involved teams of consulting engineers as well as chief designers, the small number of experts inevitably worked together in one or the other capacity. Although institutions such as the Rensselaer Polytechnic Institute were training engineers in the 1830s, the best way to learn about large bridges was by building them, and since the largest bridges were built by recognized engineers, formal education was augmented by a system of apprenticeship. This on-the-job training also accounts for the pragmatism and inventiveness of American bridge builders, who were never constrained by academic theories.


Petroski writes lucidly and at length about the various technical issues that affected the design of bridges, but he does not neglect aesthetics. The chief question is whether a bridge should merely reflect its structural function, or whether adornment should not also be a part of bridge design. Of course, some people thought that bridges were merely mechanical expressions of the utilitarian principle. William Morris called the Forth Bridge “that supreme specimen of ugliness.” Benjamin Baker, the designer of the bridge, replied that he had been explicitly concerned with its aesthetic effect. Most of the early engineers distinguished between bridge construction and bridge architecture. Gustav Lindenthal, for example, ridiculed the notion that “correctly designed structures have an innate architectural beauty, requiring no adornment, unless perhaps that of a well selected color of paint.” These engineers often worked with architects: Lindenthal with Henry Hornbostel on the Hell Gate Bridge between Manhattan and the Bronx; Ralph Modjeski with Paul Philippe Cret on the Delaware River Bridge; Ammann with Aymar Embury on the Bronx-Whitestone Bridge. Ammann himself was dissatisfied with his George Washington Bridge precisely because the masonry cladding of the steel towers (designed by the architect Cass Gilbert) had to be left off for economic reasons. But Le Corbusier called the George Washington “the most beautiful bridge in the world” precisely because it was unadorned. Later, perhaps influenced by modernist architects, engineers tended to favor this view, and bridges like the Tacoma Narrows, the San Francisco—Oakland Bay, and the Verrazano-Narrows are devoid of decoration. Still, perhaps the best evidence that bridges can benefit from ornamentation is the wondrous Golden Gate Bridge: both the characteristic color and the Art Deco treatment of the towers and other details were the work of the consulting architect, Irving F. Morrow.

The great American bridge-building episode recalls an earlier spectacular burst of construction. During a one-hundred-year period, from the middle of the twelfth century to roughly the middle of the thirteenth, eleven small cities in northern France—Sens, Noyon, Laon, Paris, Bourges, Chartres, Rouen, Rheims, Le Mans, Amiens, and Beauvais—undertook to build great cathedrals. As among bridge builders, there was competition, over the height of the nave, said to represent closeness to God. Started in 1163, Notre-Dame de Paris established the record with its vault 115 feet above the floor. Thirty-one years later the builders of Chartres pushed upward to 120 feet. In 1221, the cathedral at Amiens added another 20 feet. When the builders of neighboring Beauvais decided to rebuild the choir of their cathedral, they made it almost 160 feet high, and created a structure that was lighter and more graceful than anything heretofore built. It took twenty-five years to complete the choir (1247–1272), longer than with most cathedrals, which attested to the complexity of the design, but when it was done it was the tallest structure in the world. In 1284, a large part of the great vault collapsed.

One has the sense of the cathedral builders ambitiously pushing—and surpassing—the limits of stone, mortar, and gravity. (Beauvais was the last of the cathedrals. The choir was eventually rebuilt, with extra piers and reinforced vaults, but it took forty years; the nave itself remained unbuilt.) One might expect that modern bridge builders would be immune to the vicissitudes of trial and error, but this was not so. Petroski writes of three disastrous bridge collapses, starting with the Tay Bridge near Dundee, Scotland, a two-mile-long railroad bridge, the longest in the world. Two years after it opened, on December 28, 1879, during a winter storm, more than a dozen spans collapsed, carrying with them the Edinburgh—Dundee train. All seventy-five passengers were killed. A commission of inquiry found that the designer of the bridge, Sir Thomas Bouch, was at fault for, among other things, neglecting to take into account the pressure of the wind, which in the broad expanse of the Firth of Tay was considerable. The replacement bridge was designed to resist fifty-six pounds of wind pressure per square foot, but it was not designed by Sir Thomas, who had died four months after the inquiry at the age of only fifty-eight. His demise was undoubtedly “hastened,” as the Victorians would say, by the ignominy of the Tay collapse.

The next great Scottish railroad bridge was the Firth of Forth, designed by Benjamin Baker and Sir John Fowler and completed in 1890. With the memory of the Tay Bridge still fresh, the engineers paid particular attention to strength and rigidity. “The bridge was built ‘straddle legged’ not only to achieve a great stiffness against the wind,” writes Petroski, “but also to look as if it did.” The Forth Bridge was one of the first examples of a new principle in bridge building: the cantilever. In a cantilever bridge, the spans are built out from a central support in two directions, one side counterbalancing the other. Unlike an arch bridge, a cantilever bridge requires no large temporary bracing, which reduces costs considerably, especially for large spans.

In fact, the principle of the cantilever had been used almost twenty years earlier, by Eads in St. Louis. Petroski devotes an interesting chapter to describing how Eads incorporated cantilever principles in the construction of the St. Louis Bridge (now called the Eads Bridge), the first bridge across the Mississippi, completed in 1874 by Andrew Carnegie’s Keystone Bridge Company. The three arches of Eads’s steel bridge each spanned slightly more than 500 feet, but the Forth Bridge was in a different league, with two 1,710-foot spans. That made it the longest-spanning bridge in the world, surpassing the 1,595-foot Brooklyn Bridge.

Different periods have been dominated by different structural types. Arched bridges had their day, so did truss and girder bridges (until the Tay collapse). The Forth Bridge made cantilever bridges popular around the turn of the century; they seemed cheaper and stiffer than suspension bridges which, it was believed, were not suited to the heavier loads of contemporary trains. Thus, when a railroad bridge was built across the St. Lawrence at Quebec City, it was decided that it should be a cantilever. Its span of 1,800 feet would surpass even that of each part of the Forth. That was no accident, as Petroski notes—the piers were originally to be 1,600 feet apart, but were set farther, at the suggestion of the chief engineer. The bridge was Canadian, but both the chief engineer and the contractor—from Phoenixville, Pennsylvania—were American. During construction, on August 29, 1907, the south arm of the bridge, which was projecting about 600 feet out above the river, suddenly collapsed. About seventy-five workers were killed.

The Quebec Bridge was eventually completed. It is credited not to the original chief engineer, Theodore Cooper, but to a team that included Ralph Modjeski, a flamboyant American engineer who went on to design two notable suspension bridges: the Delaware River Bridge and the Poughkeepsie Bridge. The Quebec disaster shook the bridge-building world, not only because the sixty-eight-year-old Cooper was considered the most eminent engineer of his day, but also because the accident cast doubt on long-span bridges in general, and on the cantilever type in particular. The Queensboro Bridge, a cantilever type designed by Lindenthal, was already underway, but it would be another twenty years before another major cantilever bridge would be undertaken in the United States.

The collapse of the Quebec Bridge, and Theodore Cooper’s subsequent retirement from active practice, “greatly influenced the bridgescape across our rivers and the bridgeline of our cities to become what we know today,” writes Petroski. One of the consequences of the Quebec accident was the rise of the suspension type. There were exceptions, such as Lindenthal’s Hell Gate Bridge and Ammann’s beautiful Bayonne Bridge, in New Jersey, in both of which the deck is hung below a steel arch; but on the whole the next generation of major bridges were all suspension bridges.

Once engineers understood how to stiffen the structure, it turned out that suspension bridges were capable of extraordinary spans, as well as carrying the heavier loads of truck and car traffic. In 1926, Modjeski’s Delaware Bridge spanned 1,750 feet; the George Washington Bridge, designed by Ammann, doubled this span; and in 1937, the main span of the Golden Gate measured an astonishing 4,200 feet. The consulting engineers included Ammann and Leon Moisseiff, an experienced engineer who had been involved with many major bridges. During the late 1930s, Moisseiff was engaged as the chief consulting engineer on a 2,800-foot suspension bridge over the Tacoma Narrows. The bridge was unusual in having an exceptionally narrow roadway (only 39 feet wide, compared to the George Washington’s 106 feet). The long suspension bridges of the 1930s were famous for undulating in high winds, and sometimes required stiffening or restraining devices to be installed after they were completed. So when high winds created a wavelike motion in the roadway of the Tacoma Narrows Bridge, drivers were not alarmed. The bridge was nicknamed Galloping Gertie and became a tourist attraction. However, on November 7, 1940, a wind of about 40 miles per hour exacerbated the wavelike motion; the roadway began to twist like a propeller and shortly fell into pieces. Fortunately there were only two people on the bridge, an engineer and a reporter, and both reached safety; but the reporter’s dog, left in the car, could not escape.

The dramatic collapse of the Tacoma Narrows Bridge (which was captured on film) reminded engineers, who obviously needed reminding, that they had to take into account dynamic as well as static forces. (Curiously, John Roebling had written about the effect of wind on bridges years before, and his Niagara Gorge bridge included cable stays below the deck to damper its force.) There was some resistance to applying the lessons of aerodynamic engineering to something as solid as a bridge. According to Petroski, even as talented an engineer as Ammann was reluctant to have bridge designs subjected to wind-tunnel testing. Still, the tendency to design thinner and more delicate suspension bridges abated: Ammann’s own Verrazano-Narrows Bridge (with a main span 60 feet longer than that of the Golden Gate), and the younger engineer David Steinman’s elegant Mackinac Bridge in Michigan, would both include stiffening features. “The aesthetic of light and slender had been replaced with one of strong and solid,” Petroski concludes.

The author cites a chilling study that has identified a regularity in bridge collapses—roughly every thirty years. (The Tacoma Narrows Bridge collapse of 1940 was followed by two bridge failures of innovative steel box girder bridges in 1970, in Wales and Australia.) The explanation for these recurrent failures seems to be a combination of hubris, a failure of engineers of different generations to communicate with one another, and the development of new and untested materials, bridge types, and methods of analysis. Petroski speculates that if the thirty-year theory is correct, the next victim will be “a cable-stayed bridge”—a suspension bridge whose cables radiate directly from one central mast and not from two or more towers. The type was popularized in the 1950s in Germany as a solution to rebuilding war-damaged bridges on old piers. It was assumed, at first, that cable-stayed bridges, which are considerably lighter than suspension bridges, could span only about 1,200 feet (the span of the 1987 cable-stayed Sunshine Skyway Bridge across Tampa Bay). But by 1991, the longest cable-stayed bridge in Europe, the Queen Elizabeth II Bridge across the Thames, had a main span of almost 1,500 feet, and this year the cable-stayed Pont de Normandie over the Seine was opened. Its main span is 2,800 feet. There have even been plans for a Danish cable-stayed bridge almost 4,000 feet in span.

It appears that the lead in bridge building has passed back to Europe, at least for the moment. This is partly because fewer new bridges are needed in the US, as well as owing to the strained economic situation of many of our large cities. But there may be another reason. A large bridge is always a public structure; it seems to affirm a public optimism. The absence of large bridge projects in the last thirty years is probably an indication of the current lack of enthusiasm among the American public for large-scale technology in general, particularly in public works.

Undoubtedly, great bridges will one day again be built in the United States. Meanwhile, the challenge remains to preserve the legacy of great bridges that has been given us. There is much to be done. Petroski writes that roughly one out of five of American bridges has some structural defect, an alarming figure. Maintenance is crucial, and not just for immediate practical reasons. Preserving bridges is at least as important as preserving architectural monuments for, as Petroski amply demonstrates in this timely book, the dreams of engineers are also our own.

This Issue

November 16, 1995