Last and First Men · essays & theory

A reading · through the lens of theory

A camera tilts slowly up the edge of a concrete fin. Cloud moves behind it. The shot holds past the point where you have finished looking, and somewhere in that extra time you notice you have no idea how big the thing is. Three metres? Thirty? Nothing in the frame will tell you. No person, no path, no railing, no sign.

That withholding is the film's founding decision, and everything else follows from it.

The objects are real and locatable. They are spomenici, the memorial complexes commissioned across Tito's Yugoslavia from the 1960s on to mark WWII battlefields and sites of fascist atrocity, by architects and sculptors including Bogdan Bogdanović, Dušan Džamonja, Miodrag Živković and Vojin Bakić. Jóhannsson gathered the footage over several years of trips through Bosnia, Croatia, Serbia and North Macedonia. He tells you none of that. The film never identifies, locates or dates a single structure. What it gives you instead is Tilda Swinton's voice, speaking as a collective "we": these are the Eighteenth Men, the last human species, dying on Neptune two billion years from now, transmitting backward to us.

Deleuze has a name for what that operation does to an image. He calls it crystalline description. An ordinary description points at an object that stays put underneath it. A crystalline description composes its object and then replaces it, so the thing on screen loses the right to be itself. Say Neptune over spalled Yugoslav concrete for seventy minutes and never cut away, and the concrete simply cannot hold its own identity. Notice that the film contains no visual effects and no future imagery whatsoever. Its science fiction is produced entirely by grammar.

Here is why the Deleuzian lens earns its place, rather than being draped over the film afterward. The obvious complaint about Last and First Men is that it is a slideshow with a voice-over. Deleuze's whole account of postwar cinema is a theory of exactly that condition. In what he calls the movement-image, someone perceives a situation and what they perceive tells them what to do; the cut serves the doing. After the war, he argued, a cinema appears in which perception no longer prolongs into action, and the protagonist becomes a voyant, a seer who can only watch and endure. Deleuze's term for the resulting image is the opsign: a pure optical and sound situation, something you can only look at and listen to because no action is available to discharge it into. This film takes that to a limit. There is not even a body left to be the seer. The position has been handed upward to the dying speakers, who can describe and transmit and nothing else, and then handed to you.

The spaces are what Deleuze calls espaces quelconques, any-spaces-whatever: places that have stopped functioning as settings for anything. More precisely, this is the disconnected variety. The editing groups shots by shape, rhythm and the cadence of the narration rather than by geography, so the fragments never assemble into a map. You cannot walk from one image to the next. And duration is the real cutting instrument. Shots are held long enough that the eye gives up reading them as photographs of objects and starts reading them as events.

Withholding scale does something specific to the frame. Deleuze distinguishes the relative out-of-field, the room next door that a pan could reach, from the absolute out-of-field, something outside the image that no camera movement could ever arrive at. Neptune is an absolute out-of-field sustained for a whole feature. La Jetée found this first: a future that exists only as a voice narrating in the past tense over photographed present-day matter. Jóhannsson takes the device and slows it almost to stillness. Where Marker moved by still frames, this film moves by tilt and tempo.

The camera itself has been de-peopled. No tourists, no caretakers, no roads, no signage. That erasure of the visitor is what lets the framing read as survey rather than authorship, and the debt is to Michael Snow's La Région Centrale, where an unmanned instrument rotates across unpeopled terrain. Deleuze would call this the photogramme: vision freed from any human center, matter registering matter.

What the voice does to these images is the film's real invention, and Deleuze's late vocabulary fits it almost too well. He describes the lectosign, an image that has lost its sensory-motor obviousness and must be read like a text whose key is missing. Its strongest form is the stratigraphic image: an emptied landscape, layered like rock, spoken over by something no longer present. He derives it from Straub and Huillet, where a voice of the buried past passes over depopulated ground. Jóhannsson inverts the polarity. The buried thing here is a future. The spomeniks were built in a deliberately abstract, non-figurative idiom precisely because a multi-ethnic federation could not afford any one nation's iconography, and that enforced abstraction is now the exact reason they photograph as spacecraft. The socialist future failed, and its failure is what makes it look interplanetary.

So the film contains two missing peoples, which is rare. Deleuze's phrase for modern political cinema is that "the people are missing" — the political film can no longer speak on behalf of a constituency that exists, and must work in the gap where one has yet to be invented. The audience the spomeniks were built to address dissolved in the 1990s. The humanity Swinton claims to speak for will never exist at all. Two absent constituencies address each other straight over our heads. What she performs is not dialogue and not a documentary voice-over but what Deleuze calls a speech act, a voice caught in the act of legending, of telling a people into being. Swinton withholds affect entirely; the delivery is measured, evenly stressed, closer to recital than report. There is no grief in the voice. All the grief is in the sentences.

Sound and image here are not two channels of one report. Deleuze calls this heautonomy: each becomes a self-standing whole, touching the other only across a gap. The score never illustrates. Long low brass, wordless choral writing hovering between liturgy and drone, tape-degraded textures so that even acoustic sources arrive weathered. Most of the film's sense of enormous time is carried by harmonic rhythm, by nothing you can see. That is Jóhannsson's specific contribution, and it comes from his other career. A composer who had spent the decade scoring Sicario and Arrival made a film in which duration is composed rather than cut. The work began as a live performance with 16mm projection, orchestra, chorus and narrator in the room. It was authored as a score first and a film second, and it never quite stops behaving like one.

The time it produces is what Deleuze calls sheets of past: not a dated flashback but coexisting strata you move among. 1941 and its atrocities. The concrete optimism of the 1970s. The wars that ended the federation. The weeds and spalling of 2020. And a vantage two billion years out from which all of that, including us, is sediment. The film's most radical move happens in its first sentence, when our present is placed in the past tense and left there.

Its ancestry is honest and specific. Les statues meurent aussi invented the narrated-object essay and argued that an object severed from its living culture has died; that thesis is transposed intact onto the spomeniks. Nuit et brouillard supplies the slow lateral track over depopulated ruin, held until concrete becomes a witness, with a composed literary text in an actor's cool voice rather than a reporter's. Kubrick supplies the monolith grammar, a featureless slab framed by vertical movement that refuses to state its size, and Ligeti's choral mass standing in for synchronous sound. Tarkovsky proves the underlying wager: alien register comes from tempo, not effects, and derelict terrain becomes another planet purely through duration. What Jóhannsson adds is 16mm grain treated as content. The photochemical image decays in the same way the monument does. Both are attempts to fix something; both are losing.

What it did to narrative is simply to remove it. No protagonist, no conflict, no scene, and in place of drama a single sustained address. It also proves something useful to anyone with no money: deep time does not require rendering. It requires patience, a composer, and the discipline to keep people out of the frame. The film premiered in 2020, two years after Jóhannsson died at forty-eight, finished by collaborators including Yair Elazar Glotman. A third orphaned future, sitting quietly beside the other two. The film does not mention it and does not need to.

The transmission is described as both a plea for help and a warning, and the film never decides which. That may be the honest ending. What could we possibly do with it?

Watch it again and resist the urge to look the monuments up. The not-knowing is the film. And watch the sky, which is the co-subject of half the running time: pale mass against moving cloud. The cloud is the only clock in the frame. The concrete is doing exactly what the cloud is doing. It is only slower.