The engineer put in charge of saving the Leaning Tower of Pisa summed up the problem with a confession: “No matter how many calculations we made,” John Burland told Scientific American, “the tower should not have been standing at all.”

By 1990, the tilt had crept past 5.5 degrees — the top overhanging the base by some 15 feet — and the medieval masonry of the first storey was approaching the stress at which towers simply burst. Italy closed the monument to the public, cleared the ground beneath the overhang in case it fell, and appointed a commission of 13 experts, led by Burland, a soil-mechanics professor from Imperial College London.

Their assignment came with a poison pill: save the tower — without straightening it. A vertical Tower of Pisa is just a tower. The lean was the asset; the lean was also the killer.

Eight hundred years of falling slowly

The tilt was never an accident that happened to the tower. It is the tower. Begun in 1173 on a floodplain, the 14,000-ton marble campanile stands on a foundation only three meters deep, pressed into soft silt and clay — with slightly more sand under the north side and softer clay under the south. It began sinking southward before the builders reached the third storey.

Construction paused for a century (wars, fortunately — the soil consolidated under the load), and when work resumed, the masons built the upper floors with one side taller than the other to compensate, giving the tower its subtle banana curve. Every subsequent century added lean, and every attempt to fix it made things worse — most catastrophically in 1934, when Mussolini’s engineers pumped grout into the foundation and the tower lurched further south.

The counterintuitive fix

Burland’s committee spent years and several near-disasters (temporary lead ingots stacked on the north side as a counterweight; an anchor scheme that triggered a terrifying overnight movement) before committing to the least intuitive plan on the table: to pull the tower back, dig the ground out from under the side it was leaning away from.

The method, soil extraction, works like removing sand from under one edge of a brick. Starting in 1999, 41 extraction holes were drilled at half-meter spacing under the north edge of the foundation, and augers withdrew soil literally by the trickle — dozens of tons in total, removed a few liters at a time while instruments watched the tower’s response day by day. Gravity did the rest: the tower settled gently northward into the voids, hinging back toward vertical. Steel cables harnessed the structure as insurance, and drains fixed a subtler discovery — the north side’s water table rose with every rain, lifting that flank and worsening the lean each winter.

By 2001, the tower had been eased back about 44 centimeters — some 17 inches — returning it to its inclination of the 1830s. The reduction of half a degree is invisible to a tourist, which was precisely the specification: the stresses in the masonry dropped decisively, and the lean survived to the naked eye. On June 16, 2001, eleven years after the gates shut, the tower was handed back to Pisa, and engineers declared it safe for the next 300 years.

The tower that kept going

Then came the epilogue nobody ordered. The annual measurements kept showing movement — northward. The tower continued straightening itself for more than a decade after the engineers left, adding another 4 centimeters before finally settling; Burland got the news in a 4 a.m. phone call and pronounced the bonus unsurprising — the drained, relieved foundation was still slowly finishing the job. Monitoring today shows the tower stable, leaning about as much as it did two centuries ago.

The moral of the missing soil

The Pisa rescue has become a classic of engineering not for its scale but for its restraint. The failed fixes of 1838 and 1934 shared a philosophy: attack the tower — brace it, grout it, reinforce it — and each nearly killed it. The fix that worked never touched the building at all. Burland’s team treated the monument as untouchable and negotiated entirely with the ground, removing perhaps a truckload of soil, spoonful by monitored spoonful, from the correct side of a 14,000-ton problem.

Eight centuries of lean, arrested by subtraction. The tower still tilts, exactly as its economy requires, photographed daily by tourists holding it up with an outstretched palm — most of them unaware that the real trick was performed underground, and that the most famous structural failure on Earth is now, thanks to 17 quiet inches, one of the most closely monitored and most stable old buildings in Italy. It should not have been standing at all. It now has a booking, in writing, for the year 2300.