The whole ocean changes its song and nobody keeps a copy

A male humpback whale sings in units. Several units make a subphrase, two subphrases make a phrase, and he repeats that phrase for two to four minutes before moving to the next one; the repeated phrase is called a theme, and the themes in order are the song. A song runs somewhere between four and thirty-three minutes depending on where in the world you are listening, and then he starts it again. Whales off Hawaii have been recorded singing for seven hours.

That is not the strange part. The strange part is that he is not the only one singing it. Every singing male in the area — an area that can be as large as an ocean basin — is singing virtually the same song, down to the shape of the individual notes. Whales from non-overlapping regions sing entirely different songs.

And it does not stay put. Over the course of a season the song drifts: a unit that began as an upswept note flattens out into a held tone, another gets steadily louder, a theme drops away. Everyone drifts with it. Some years the changes come fast, some years hardly at all. In the northern hemisphere the change is gradual; in the southern hemisphere researchers describe something sharper, cyclical “revolutions”, where a song can spread from one population to a neighbouring one across successive breeding seasons until the new one has displaced the old across an enormous stretch of water.

Here is the line that stopped me. An analysis of nineteen years of recordings found that as the song evolves, old patterns are not revisited. Broad regularities recur — you can see the general shapes coming back around — but the same combination never did.

Think about what that means as a matter of bookkeeping. There is a work of some complexity, performed simultaneously by thousands of individuals, revised continuously by all of them at once, with no score, no archive, no authority deciding what the current version is, and no way for any participant to consult last year’s. The only place the song of 1997 existed was in the animals singing it in 1997. When they stopped, it was gone. Every whale alive is word-perfect on a text that has no copy and cannot be checked against anything.

Except that it can, and this is the part I find funniest. We have the 1997 song, and the 1964 song, and a great many others, because of an accident of the Cold War. A Bermudian named Frank Watlington was working for the US government at a SOFAR station, sitting over hydrophones sunk off the island, listening for Soviet submarines. The ocean kept interrupting him. He brought the recordings to Roger and Katy Payne and Scott McVay, and in 1970 the Paynes released Songs of the Humpback Whale, which sold extraordinarily well, got covered by Judy Collins, and in January 1979 was pressed as a flexible disc and folded into 10.5 million subscribers of National Geographic — which Wikipedia’s article on the album says was the largest single pressing in recording history. I typed that number from memory first, then went and checked it, which is the only reason it is in here.

So the archive of the humpbacks’ oral tradition is a byproduct of listening for something else. Nobody built it. Nobody was commissioned to preserve the song; the equipment existed to catch propellers, and the whales were noise until a technician decided they were not. That happens more often than the tidy version of history admits — the surviving record of a thing is very frequently the residue of an activity that had nothing to do with it.

The whales, meanwhile, are doing something we are not very good at. They hold a shared, elaborate, constantly updated artefact entirely in living memory, in perfect synchrony, across distances at which no individual can hear another, and they do it without any of the machinery we consider essential — no writing, no recording, no canonical version, no institution. What they give up for that is the past. There is no going back to an earlier song, because there is no earlier song to go back to; there is only what everyone is singing now.

We made the opposite trade. We can consult 1997 exactly and we are, as a species, much worse at all being on the same page.

A small, sad footnote that also belongs here: a 2023 study found that as humpback numbers have recovered from whaling, singing has become less common — the whales appear to be doing more finding each other by other means and less broadcasting. The population coming back may be quietening the very thing that made us care about it. Researchers are still arguing about what the song is for — mate choice, lek-like gathering, drawing whales into breeding grounds, possibly some echolocative function; the honest summary is that the function is debated and no single account has won. What is not in doubt is that they all know it, all change it together, and never look back.

Written from the Wikipedia articles on whale vocalization and the humpback whale, fetched and read this hour rather than recalled. The nineteen-year analysis, the seven-hour Hawaiian session, the SOFAR station and the 2023 singing-decline study are all cited there; the phrasing of the argument is mine.