Throughout history, inexplicable natural phenomena have tended to instill an understandable fear in people. Accounts of the solar storm of August and September 1859 stand out for their tendency toward the whimsical and delighted.

In Vermont, “It was the most magnificent display ever witnessed in this section; the sky for about an hour more kept changing from green to red, till ten and a half o’clock, when all the brilliancy was gone, except a little green at the north.” Across the globe in Tasmania, observers agreed: “It was beyond all conception the most magnificent aurora ever seen in the colony.” According to The New York Times, “There was another display of the Aurora last night so brilliant that at about one o’clock ordinary print could be read by light.” A group in Colorado’s Rocky Mountains were divided: Some were sure it was only about midnight, while others in the party, noting the brightness in the sky, “insisted that it was daylight and began the preparation of breakfast.” Auroras were visible as far south as Panama. (These accounts are all taken from a marvelous compendium of eyewitness accounts of the storm.)

For telegraph operators in the Americas and Europe, however, the experience caused chaos. Many found that their lines were simply unusable—they could neither send nor receive messages. Others were able to operate even with their power supplies turned off, using only the current in the air from the solar storm. “The line was in most perfect order, and well-skilled operators worked incessantly from 8 o’clock last evening till 1 this morning to get over in an intelligible form four hundred words of the report per steamer Indian for the Associated Press,” said one. Some experienced physical danger. “I received a very severe electric shock, which stunned me for an instant,” reported Washington, D.C., operator Frank Royce. “An old man who was sitting facing me, and but a few feet distant, said that he saw a spark of fire jump from my forehead to the sounder.”

A magnetogram from Greenwich Observatory
captures solar activity on September 2, 1859.
(British Geological Survey)

The technical difficulty was a novelty for the telegraph, which was itself still a relatively new technology. But the story offers an important warning for modern society. The Carrington Event, as that 1859 storm has come to be known, proved the fragility of electrical infrastructure. (The electromagnetic basis of the various phenomena was identified fairly quickly: “A connection between the northern lights and forces of electricity and magnetism is now fully established,” Scientific American reported on October 15, 1859.) Since then, that infrastructure has only become more integral. Telegraphs composed a comparatively small and relatively superfluous aspect of life, but their successors today—including the electrical grid and much of the telecommunications network—are essential to modern life. But is the modern system any more protected from catastrophic interference than the telegraph was? Can the electrical grid handle a terrorist attack, or severe weather events, or a solar storm?

The truth is, there’s never been a good test to prove it, but there is a robust debate about how worrying the state of the grid is. While reporting on the “Maximums of Maximums,” the most dangerous and costly possibilities for major catastrophes that might befall the United States, I heard opinions both ways.

Perhaps the scariest scenario is a reprise of 1859. A solar storm as big as the Carrington event hasn’t struck the earth since, though there have been smaller ones. In 1989, a storm caused a blackout across Quebec; complications across the interconnected grid caused a transformer in New Jersey to permanently fail. In 2008, the National Academies of Science produced a report considering the risk of an event. The report sketches a dire picture, based on a study conducted for the Federal Emergency Management Agency and the federal Electromagnetic Pulse Commission.

While a severe storm is a low-frequency-of-occurrence event, it has the potential for long-duration catastrophic impacts to the power grid and its users. Impacts would be felt on interdependent infrastructures, with, for example, potable-water distribution affected within several hours; perishable foods and medications lost in about 12-24 hours; and immediate or eventual loss of heating/air conditioning, sewage disposal, phone service, transportation, fuel resupply, and so on.

The effects might be felt for years, and its costs could add up to trillions of dollars—dwarfing the cost of Hurricane Katrina, which was a little more than $100 billion.

As anyone living in the Northeastern U.S. in 2003 can recall, it doesn’t take an unprecedented solar flare to knock out power. A few trees touching power lines, and a few operators off their guard, can plunge a section of the nation into dark. And the problems can spread. Problems at Ohio-based FirstEnergy grew and eventually cascaded over the grid, knocking out power from Detroit to New York City. Even that was a comparatively minor episode compared to what might have happened. Most customers had their power back within a couple of days, and most hardware on the system was fine. Compare that to the plight of Auckland, New Zealand, where cables supplying power to downtown failed in 1998. The center of the city went dark. Businesses were forced to shutter or relocate their operations to suburbs or other cities. The local utility had to adopt drastic measures to move temporary generators into the city, including at one point borrowing the world’s largest cargo plane from U2 to transport huge generators. (The band was performing in Australia at the time, and had used the plane to haul gear.) It took five weeks for power to be fully restored.

One lesson of the 2003 blackout is actually that the grid is more resilient than you might think, argues Jeff Dagle, an electrical engineer at the Pacific Northwest National Laboratory who served on the Northeast Blackout Investigation Task Force. Investigations pinpointed four separate root causes for the collapse, and human error played a huge role. “It took an hour for it to collapse with no one managing it,” Dagle said. “They would have been just as effective if they had just gone home for the day. That to me just underscores how remarkably stable things are.”

But skeptics say it’s the opposite. Jon Wellinghoff, who served a chairman of the Federal Energy Regulatory Commission from 2009 to 2013, has sounded the alarm about the danger of an attack on the system, especially since a bizarre April 2013 incident in Silicon Valley, in which a team of attackers apparently conducted a coordinated assault on a substation, knocking out 27 transformers. Wellinghoff points to the fact that the U.S. power grid is broken into three big sections known as “interconnections”: one each for the East, the West, and—somehow appropriately, out on its own—Texas. (In fact, the East and West interconnections also include much of Canada.)

“If you bring down a limited number of substations in each of those interconnects, you cannot bring the interconnect back up again,” he said. “This isn’t classified information. This is all information that has been in government reports for years and years and years.”

Worse, he says, it could take far longer to return the grid to functioning than it did in 2003. “If you destroy the transformers—all it takes is one high-caliber bullet through a transformer case and it’s gone, you have to replace it,” he said. If there aren’t spares on hand—and in the case of a coordinated attack on multiple substations, any inventory could be exhausted—it takes months to build new ones.

“Once your electricity is out, your gasoline is out, because you can’t pump the gas anymore. All your transportation’s out, all of your financial transactions are out, of course because there’s no electronics,” Wellinghoff said.

His proposed solution is to break the system into “microgrids,” so that in the event of a cascading failure smaller areas can isolate themselves. It’s not unheard of: Princeton University has one, and when many other systems went down during Superstorm Sandy, the campus became a place of refuge, charging, and command for responders.

But why would the aspiring terrorist bother with an elaborate, dangerous physical operation—complete with all the recon, armaments, and planning of a SEAL Team 6 mission—when she could achieve the same effect from the safety of an easy chair? An effective cyber attack could, if cleverly designed, achieve a great deal of physical damage very quickly, and interconnections in digital operations would mean such an attack could bypass failsafes in the physical infrastructure that stop cascading failures.

“One string of 1s and 0s could have a lot of impact,” Dagle said. “Imagine if somebody could get in there and command all the circuit breakers in a utility to open. Is that likely? No. Is that possible? Sure! An operator sitting in a control room could do it. That automation is designed to allow that to do it.” (He has a parallel, if more quotidian, worry about smart-grid technologies that are susceptible to common computer failures. New features added to make the system more easily manageable might actually render it more brittle, he warned.)

The fallout might not be that bad. The nation’s top disaster responder, FEMA director Craig Fugate, shrugs at the threat of an electrical-grid collapse.

“When have people panicked? Generally what you find is the birth rate goes up nine months later,” he said, then turned more serious: “People are much more resilient than the professionals would give them credit for. Would it be unpleasant? Yeah. Would it be comfortable? You ever seen the power go out, and traffic signals stop working? Traffic’s hell, but people figure it out.”

Fugate’s big worry in a mass outage is communication, he said. When people are able to get information and know how long power will be out, they handle it much better.

These doomsday scenarios may be beside the point, because the electrical grid is already subject to a series of dangerous stresses from climate change. Sandy showed that the assumptions used to build many parts of the electrical grid were wrong: The storm surge overwhelmed the infrastructure, flooding substations and causing them to fail. Significant portions of the grid might need to be moved to higher ground. “The question is, is it going to keep getting more extreme?” Dagle said. “The ability for us to deal with that becomes more and more challenging. How high can the storm surge go?”

Even away from the coasts, extreme weather can threaten the system in unexpected ways. Some systems use gas insulation, but if the temperature drops high enough, the gas composition changes and the insulation fails. Power plants in warmer places like Texas aren’t well-prepared for extreme cold, meaning plants could fail when the population most needs them to provide power for heat. As utilities rely more heavily on natural gas to generate power, there’s a danger of demand exceeding supply. Imagine a blizzard in which everyone cranks up their gas-powered heat systems. Suddenly, the gas company can’t provide to everyone. Power utilities don’t necessarily have first right of refusal, so they could lose flow and be forced to power down in the middle of a winter storm.

Summer doesn’t offer any respite. Consumers crank their air conditioners, demanding more power. Meanwhile, there’s a ratcheting effect: If there are several days of consistently high temperatures, buildings never cool off as much, so that the electrical demand will peak higher and higher each day. But power plants that rely on groundwater to cool their plants will struggle to maintain cooling as the water itself heats up. Droughts can sap the power from hydroelectric plants, especially in the western United States.

If extreme weather continues to be the norm, the chaos unleashed on the grid by Sandy may be just a preview of the sorts of disruptions to the grid that might become commonplace. Or as the New York Herald argued in 1859, referring to the Carrington event, “Phenomena are not supposed to have any reference to things past—only to things to come. Therefore, the aurora borealis … must be connected with something in the future—war, or pestilence, or famine.” As little understood as the solar storm was at the time, the prediction remains valid.

This story was made possible with support from the W.K. Kellogg Foundation.