Ancient Air
The bubbles are the whole point.
When snow falls on the Antarctic plateau, it doesn’t melt. It piles up, year on year, and under its own weight it compacts into firn and then into ice, and as it seals, it traps little pockets of the air that was moving through it. A core drilled two miles down will give you ice that fell as snow eight hundred thousand years ago, and inside it, bubbles of the sky from that day. Melt it in a vacuum chamber and you can measure the carbon dioxide, the methane, the isotopes. It’s how we know what the atmosphere was like before there were thermometers. I’ve worked in a core lab in Colorado for six years, and I’ve never stopped finding that beautiful: a breath of air older than any human language, coming out with a little hiss in a steel box.
Last year we got a grant for a new extraction line with very fast, very sensitive pressure transducers. The idea was to measure gas release from individual bubbles, bubble by bubble, as they opened. My colleague Wren Abernathy built the software. It records each bubble’s release as a pressure trace, a tiny wiggling line, sampled forty thousand times a second.
It was Wren who first realized you could listen to it.
“Sound is pressure,” she said. We were in the lab at nine at night, eating cold noodles, and she had her headphones around her neck. “That’s all sound is. Little changes in air pressure. And we’re recording the exact pressure of the air in the bubble as it comes out, very fast. So I wondered.” She held the headphones out.
What she’d done was take the trace from a single bubble as it opened, about thirty milliseconds of data, and play it as audio. I listened. A click and a hiss. “So?” I said.
“So every bubble’s a sample,” she said. “Of the air, at the moment it sealed. Including whatever pressure waves were moving through it. If there was a sound in that air, a little bit of it gets trapped.” She grinned. “It’s a recording. A really, really bad one. Thirty milliseconds at a time.”
I told her that was ridiculous, that the physics didn’t work, that the signal would be destroyed in compaction. She said probably. Then she said she’d been running segments through overnight and stitching the traces together in depth order, thousands of bubbles, one after another.
“Listen to this one,” she said. “It’s from about forty thousand years ago.”
It was wind.
I don’t know how else to describe it. It was faint and broken up into a crackle of tiny fragments, like a radio between stations, but under the crackle there was a sound you’d recognize anywhere: wind over open snow, rising and falling, that long hollow moan.
I took the headphones off. “That’s noise,” I said. “That’s the extraction line.”
She played me the line with no sample in it. Silence. She played me ice from a lab freezer we’d made from tap water. Silence. Then she played the forty-thousand-year-old segment again, and it was wind, on a plateau, on some afternoon in the Pleistocene, when there were mammoths in Europe and people in Australia and nobody in the Americas at all.
We sat in the lab until two in the morning listening to it, not saying much. It’s the most beautiful thing I’ve ever heard.
We spent three months on it, mostly at night, mostly off the books. Wren improved the stitching. I ran sample after sample. We went deeper and older: seventy thousand, a hundred and ten thousand years. It was always wind. Sometimes faint. Sometimes a storm. Once, in a segment around eighty thousand years old, a sound like a long slow crack, ice shifting somewhere far off, that Wren said gave her chills.
And then she found the other sound.
It was in the hundred-and-ten-thousand-year segment, at first. Very low. Under the wind. A slow swell and fade, about every nine seconds, like the wind was gusting in a regular pattern. Wren filtered out the high frequencies and slowed it down and played it for me, and it was clear.
In. Five seconds. Out. Four.
“It’s breathing,” she said.
I said it was a pressure oscillation in the ice sheet. A seiche. Some kind of standing wave. She said maybe. She said, “But listen to how far away it is.”
It was very far away. You could hear the distance in it, the way you can hear that thunder is miles off.
We tracked it forward. Through the ages. It wasn’t difficult once you knew what to listen for. It was there in every segment we checked, under the wind, slow and steady, nine seconds a cycle. In the hundred-and-ten-thousand-year ice it was faint as a rumor. In the seventy-thousand, louder. In the forty-thousand, the wind we’d fallen in love with, it was there underneath, closer. A little closer in every segment, all the way up the core.
Wren made a graph. Amplitude against age. It was a smooth curve. It rose, slowly, over a hundred thousand years, the way the sound of a train rises as it comes up the line.
“It’s not getting louder,” Wren said quietly. “It’s getting nearer.”
Our core only went up to about 1800, because the top of the core is firn and doesn’t trap bubbles cleanly. In the 1800 ice the breathing was very loud. It was so close that you could hear texture in it. A wet catch at the bottom of each in-breath. It was enormous. You could feel, even through the bad stitching and the crackle, that whatever made that sound was big in a way that didn’t fit on the plateau. That didn’t fit anywhere.
Then Wren did something I didn’t know about until afterward. She took a sample of last winter’s snow from the surface pit at the drill site, which we had in the freezer for a different project, and she pressed it, and froze it under pressure until it trapped bubbles, and ran it through the line.
She was at her desk in the morning when I came in. She was wearing the headphones. She’d been there all night. She didn’t take them off when I said hello. She just held one side away from her ear so I could hear, and it was the breathing, and it was close. It wasn’t on the plateau anymore. It wasn’t far away at all. It was the loudest thing in the recording. Louder than the wind.
In. Five seconds. Out. Four.
“This is last winter,” she said. “This is the air last January.”
I said there had to be a mistake.
“I know,” she said. “So I tested it.” She pointed at a small steel canister on the bench, the kind we use for calibration samples. “That’s the lab. I took a sample of this room’s air at four this morning and ran it through the line.”
She put the headphones on my head.
There was no wind in the lab sample. Just the hum of the building, the faint thrum of the freezers, a clatter that was somebody’s chair. And under that, very close, so close that it was hard to separate from the hum, a slow, enormous swell and fade.
In. Five seconds.
Out.
I’ve tried to think of an explanation for what Wren found that isn’t the obvious one. I’m a scientist. That’s my job. I can come up with seiches, building resonance, an artifact of the software that scales with sample freshness. I’ve been trying to come up with them for six weeks.
Wren isn’t trying. She quit last month. She moved back to her parents’ place in Nebraska, and the last time we talked on the phone she said that she can’t sleep, because once you know the rhythm you can hear it anywhere if it’s quiet enough. In the hum of a refrigerator. In the sound of the furnace. In the wind outside. She says it’s in everything, now. It’s in the air.
She said, “It took a hundred thousand years to get here. It’s been coming across the ice that whole time. And it’s here now.” She said, “I keep wondering what it does when it stops coming.”
I’ve stopped working nights. I don’t listen to the samples. But the lab’s very quiet in the early morning before anyone else comes in, and I’ve noticed that the big ventilation system breathes in a slow cycle, the way they do, pulling and pushing the air through the building.
It’s about nine seconds. Five in, and four out.
I asked facilities how it’s programmed. The man there looked at his screens for a while and then said, puzzled, that it isn’t set to any cycle at all. It’s supposed to be constant.