Arctic. Atlantic. Long ago, the two oceans existed in harmony, with warm and salty Atlantic waters gently flowing into the Arctic. The layered nature of the Arctic — sea ice on top, cool freshwater in the middle and warm, salty water at the bottom — helped hold the boundary between the polar ocean and the warmer Atlantic.

But everything changed when the larger ocean began flowing faster than the polar ocean could accommodate, weakening the distinction between the layers and transforming Arctic waters into something closer to the Atlantic. This process, called Atlantification, is part of the reason the Arctic is warming faster than any other ocean.

“It’s not a new invasion of the Arctic,” said Yueng-Djern Lenn, a physical oceanographer at Bangor University in Wales. “What’s new is that the properties of the Arctic are changing.”

Satellites offer some of the clearest measurements of changes in the Arctic Ocean and sea ice. But their records only go back around 40 years, obscuring how the climate of the ocean may have changed in prior decades.

“To go back, we need a sort of time machine,” said Tommaso Tesi, a researcher at the Institute of Polar Sciences-CNR, Italy.

In a paper published Wednesday in the journal Science Advances, Dr. Tesi and colleagues were able to turn back time with yard-long sediment cores taken from the seafloor, which archived 800 years of historical changes in Arctic waters. Their analysis found Atlantification started at the beginning of the 20th century — decades before the process had been documented by satellite imagery. The Arctic has warmed by around 2 degrees Celsius since 1900. But this early Atlantification did not appear in existing historical climate models, a discrepancy that the authors say may reveal gaps in those estimates.

“It’s a bit unsettling because we rely on these models for future climate predictions,” Dr. Tesi said.

Mohamed Ezat, a researcher at the Tromso campus of the Arctic University of Norway, who was not involved with the research, called the findings “remarkable.”

“Information on long-term past changes in Arctic Ocean hydrography are needed, and long overdue,” Dr. Ezat wrote in an email.

In 2017, the researchers extracted a sediment core from the seafloor of Kongsfjorden, a glacial fjord in the east end of the Fram Strait, a gateway between the Norwegian archipelago Svalbard and Greenland, where Arctic and Atlantic waters mingle.

The researchers sliced up the core at regular intervals and dried those layers. Then came the painstaking process of sifting out and identifying the samples’ foraminifera — single-celled organisms that build intricate shells around themselves using minerals in the ocean.

When foraminifera die, their shells drift to the seafloor and accumulate in layers of sediment. The creatures are crucial clues in sediment samples; by identifying which foraminifera are present in a sample and analyzing the chemistry of their shells, scientists can glean the properties of past oceans.

The team’s original idea was to reconstruct the oceanographic conditions of a region that contained both Arctic and Atlantic waters, going back 1,000 to 2,000 years. But, in the slices of the core dating back to the early 20th century, the researchers noticed a sudden, massive increase in the concentration of foraminifera that prefer salty environments — a sign of Atlantification, far earlier than anyone had documented.

“It was quite a lot of surprises in one study,” said Francesco Muschitiello, an oceanographer at the University of Cambridge and an author on the paper.

The sheer amount of sediment was so high that the researchers could assemble a chronology of past climate down to five- or 10-year increments. Additionally, a molecular biomarker could pinpoint a specific year, 1916, when coal mining began in Kongsfjorden. Since the foraminiferal shift occurred just before this marker, the researchers estimate Atlantification began around 1907, give or take a decade.

When the researchers compared the data from their paleoclimate model with others to see if they overlapped, they found existing climate models had no sign of this early Atlantification. The researchers suggest a number of possible reasons behind this absence, such as an underestimation of the role of freshwater mixing in the Arctic or the region’s sensitivity to warming.

Dr. Lenn, who was not involved with the research, sees a difference between this early Atlantification and the present, rapid Atlantification, which is largely driven by melting Arctic sea ice. “It’s too soon after the start of the industrial revolution for us to have accumulated excess heat in the planetary system for it to be anthropogenic at that point,” Dr. Lenn said.

The authors are not sure of the precise reasons behind the early Atlantification. If human influences are the cause, then “the whole system is much more sensitive to greenhouse gases than we previously thought,” Dr. Muschitiello said.

In another possibility, earlier natural warming may have made the Arctic Ocean much more sensitive to the accelerated Atlantification of recent decades. “Could it be that we destabilized a system that was already shifting?” Dr. Tesi said.

This is the maddening mystery of any paleoclimate model. “None of us were there,” Dr. Lenn said, laughing.

Although this is true of humans, it is not true of corals in the Fram Strait. The long-lived animals record changes in climate and other parameters, making them excellent sentinels of climate history. Dr. Tesi hopes to study the strait’s cold-living corals next, to see what insight they may offer into the Atlantic’s usurpation of the Arctic.