The first source-to-sea descent of the Amazon River, completed by Polish kayaker Piotr Chmielinski and San Francisco newspaperman and rafting novice Joe Kane over the winter of 1985–86, was a six-month epic that claimed one of the last great adventure prizes. Chmielinski and Kane were two of just four people left standing after the 12-person expedition was held up at gunpoint by Shining Path rebels, consumed by infighting, and ultimately abandoned by its self-serving South African leader, François Odendaal. Kane’s 1989 book, Running the Amazon, was both a classic and a bestseller.

But now it turns out that the trip may no longer count as the first descent. New evidence, currently under review by a respected scientific journal, makes the case that the tributary where Chmielinski and Kane put in, the Apurimac, isn’t the true source of the river. That’s a bit like saying Hillary and Tenzing climbed the wrong mountain. It also means that the first descent of the Amazon may once again be up for grabs. And in a race that’s reminiscent of the classic era of exploration, two rival paddlers are battling it out to the finish.

It might seem unbelievable that the Amazon’s source could have been overlooked for this long, but the starting point of a river is a somewhat arbitrary distinction, defined as the place from which water travels the greatest distance to reach the ocean. It was only in 2000 that an international team of scientists definitively declared the snowmelt on Peru’s 18,363-foot Nevado Mismi, above the Apurimac, to be the Amazon’s starting point. Except that it probably isn’t.

Last winter a San Diego whitewater-kayaking guide and former neuroscientist named James “Rocky” Contos, 41, was using Google Earth to plan a trip to Peru, to run the five major headwaters that feed the Amazon. As he plotted each of their lengths via GPS, he decided that the Mantaro, north and west of the Apurimac, was the real source. “I was looking at the maps and realized that the Mantaro is 50 miles longer than the Apurimac,” says Contos. He did a bit more research and learned that much of the Mantaro had never been paddled.

Contos immediately went to work on an article about his finding, submitting it last spring to the peer-reviewed British geography journal Area, which should make a decision on whether to publish the report in the next few months. He also rebooted his Peru plans to run the Mantaro and then keep going 4,000-plus miles down the Amazon to the Atlantic, capturing the first descent before anyone else knew it was back in play. Contos had the skills and mojo to pull it off—he’d notched numerous first descents in Mexico and had also run the Colorado River through the Grand Canyon, twice, without a permit—but not the funds to support such a lengthy endeavor.

So he started looking around for potential paddling partners who might share costs for sections of the trip. In March, he came across the website for an expedition billing itself as the Amazon Express. The leader, West Hansen, a 50-year-old marathon flatwater paddler and social worker from Austin, Texas, was organizing a two-month source-to-sea sprint down the river. Contos reached out over email, and the two began swapping information. Initially, Contos described only his earlier plan to run the Amazon’s tributaries, which made him an attractive potential teammate to Hansen, who was looking for whitewater specialists to help get him through the Class V rapids at the top of the Apurimac.

Hansen declined to comment for this story, citing an agreement with National Geographic, which signed on to sponsor his trip this past summer, but Contos provided what he says is a complete record of their correspondence. In an email dated March 23, Hansen told him he could probably fund most of the trip in return for “your company and gear collaboration.”

Not long after that, Contos felt compelled to open up about the Mantaro, insisting that Hansen keep the news secret. Hansen found the claim credible but wasn’t sure he wanted to change his plans. “While I’m very interested in departing from the true source of the Amazon,” he wrote to Contos on April 2, “the 30-to-50-mile difference in the Mantaro and Apurimac will have little bearing upon my accomplishment to any but a select few.”