Heaven’s Gate (1980) Review
6/10
Wyoming in 1892. An association of cattle barons draws up a list of a hundred and twenty-five immigrant settlers to be killed, hires an army of mercenaries to do it, and the settlers organize to fight back. The 6 is a compromise between two honest reactions. The photography is among the most beautiful ever committed to a Western, the recreation of the period is total, and the massacre sequence is overwhelming. It’s also three and a half hours long with a protagonist nobody can locate and a love triangle that stops the film dead.
I refuse both of the usual verdicts on this film.
Calling it the worst film ever made gets the images wrong, and calling it a suppressed masterpiece ignores how long I spent waiting for Averill to want something. My gut reaction splits right down the middle: awe at what’s on the screen and a mounting irritation at a story that won’t start. It’s a magnificent, badly structured picture that cost a studio its independence.
What the Money Bought
The planned budget was around eleven and a half million dollars. The final cost was forty-four. Filming began in Montana in April 1979 and was five days behind schedule by the sixth day of shooting.
The excess is visible in the frame and that’s the uncomfortable part. Twelve hundred extras. A street set with houses on both sides torn down and rebuilt because Cimino wanted more distance between them. A vintage train trucked in from a Denver museum because it wouldn’t fit through the rail tunnels. Six weeks of training for the cast in skating, riding and shooting.
Vilmos Zsigmond photographed it, and the result is the argument for the whole enterprise. The light in this film is filtered through dust and smoke in nearly every exterior, and the effect is of looking at 1892 instead of at a reconstruction of it. So the real position is awkward. The waste was real, the perfectionism was indefensible as production management, and it produced images that a disciplined shoot wouldn’t have got. Both things are true and neither cancels the other.
Cimino had won the latitude. The Deer Hunter had taken five Academy Awards including Best Picture two years earlier, and United Artists handed him a Western and stood back. The autonomy that produced one film produced the other, and the industry drew exactly one conclusion from that.
Detail is the thing that money and time really buy, and it’s worth having. The difference between a period setting that convinces and one that doesn’t is a hundred small accuracies nobody consciously notices. The trap is that detail can’t replace structure. Cimino spent his resources on the surface of every scene and had nothing left over for the question of whose story this was, and no amount of accurate dust fixes that.
For a writer the equivalent is research. It’ll make your world solid and it won’t tell you what happens or who it happens to, and it’s seductive because it feels like progress.
Nobody Is the Protagonist
Kris Kristofferson plays Averill, a Harvard-educated marshal on the settlers’ side. He’s the nominal lead and he spends most of the film watching, drinking and arriving late to things.
Christopher Walken’s Nate Champion is the more interesting figure, a hired gun losing his stomach for the work, and he’s killed before the climax. Isabelle Huppert’s Ella runs a brothel and is positioned between the two men, and John Hurt drifts through as a wealthy alcoholic commentator with nothing to do.
The structural failure is that the film has a subject and no vehicle. The subject is class war on the American frontier. That’s real, important and well documented. The vehicle is supposed to be a triangle between three people, and the triangle carries no weight because Averill has no visible want.
Compare it to Shane, which the film consciously answers. Shane tells the same story in microcosm through a boy watching a man decide something. Heaven’s Gate has two hundred extras and a marshal who can’t make up his mind.
The supporting bench makes the waste sharper. Mickey Rourke, Willem Dafoe in his first film role, Terry O’Quinn, Brad Dourif as an organizer who’s the only character with a stated political position. Any one of them could have carried a thread and none of them is given one.
Kristofferson is the problem I can’t get past. He looks right in every frame and gives me nothing to hold onto, and I kept wishing the film had followed Walken’s Nate instead. Dafoe and Rourke passing through with nothing to do only adds to the waste.
The Massacre Works
The last hour is the film the whole production was reaching for. The mercenaries and the settlers meet on open ground and the sequence runs long, chaotic and without heroism.
Cimino stages it as a slow industrial process. Wagons are lashed into rolling barricades, positions are dug, men circle, and the camera holds on the mechanics instead of cutting to reactions. Nobody wins a duel. People die from a distance, incompetently, in dust.
That’s the correct way to shoot this particular history. The Johnson County War wasn’t a shootout, it was a siege that ended when the cavalry arrived to rescue the hired killers, and the film reproduces the moral shape of that exactly.
This hour earns most of the 6 on its own. I love how Cimino refuses anyone a heroic shot, and the dust, the confusion and the wagons grinding forward left me more shaken than any gunfight in a Western I can name.
It also justifies the running time locally. That’s the frustrating thing. When the film is doing this, the scale earns itself. It simply took two and a half hours to get here.
A climax built on process instead of personal combat is harder to write and more truthful. It requires the reader to understand the logistics. That’s why the setup must be laid in far earlier. The cost is that you can’t rely on a duel to carry the emotion, so the people inside the process have to have been established as individuals well before it starts. This is exactly where Heaven’s Gate is thin.
The lesson runs both ways. Build the mechanism and the characters, and the ending works. Build only the mechanism and you get a magnificent sequence about nobody in particular.
The Cuts and the Collapse
Cimino screened a version running around three hundred and twenty-five minutes to United Artists in June 1980. They refused to release it and considered firing him. He spent the summer cutting it to two hundred and nineteen minutes.
The premiere was at Cinema 1 in New York on 19 November 1980. Vincent Canby’s review in the Times was annihilating. During the intermission Cimino reportedly asked why nobody was drinking the champagne, and his own publicist told him that the audience hated the film.
UA pulled it after a one-week run. Cimino recut to a hundred and forty-nine minutes for an April 1981 release. That was also badly received and took under three and a half million dollars domestically.
The consequence for the industry was larger than the loss. Transamerica sold United Artists to MGM, and studios took production control back from directors. That closed the period of filmmaker autonomy that had produced Cimino’s own Deer Hunter two years earlier.
The Reappraisal and Its Limits
The restored version, reconstructed for the Criterion release from separation masters because the original negative had been cut up, has been treated far more kindly. Scorsese is among its admirers. The Times revisited it in 2013 and called it a film with greatness in it alongside long dead stretches. That’s the fair verdict and I’d keep both halves of it. The common defense, that critics in 1980 were reviewing the production coverage instead of the picture, is partly true. The film had been reported as a disaster for eighteen months before anybody saw it.
But the structural problems are in the film and not in the press. A viewer coming to the restored cut today with no knowledge of the history still waits a very long time for anything to happen, and still can’t say what Averill wants.
Where I land: watch it for Zsigmond, for the period, and for the last hour. Don’t expect the lost masterpiece, and don’t accept the story that it’s unwatchable. Both of those are things people say instead of watching it.
Reputation moves independently of the work and it takes decades to settle. A book received badly for reasons outside itself can recover, and the recovery doesn’t repair whatever was wrong.
Be suspicious of both directions. The pile-on is usually excessive and the rehabilitation usually overcorrects, because the second is a reaction to the first instead of a reading of the thing.
The practical version: when your own work is criticized, separate the objections that describe the work from the ones that describe its circumstances. Fix the first kind and wait out the second.
Should you watch Heaven’s Gate?
Watch the restored cut, and know what you’re getting: two hours of gorgeous, aimless period recreation, then an hour of the best staged massacre in American cinema. I land on a 6 because I can’t forgive the first part for how long it makes me wait, and I can’t dismiss the last part at all. Zsigmond’s light pulls the number up. Averill’s empty center and that dead love triangle drag it right back down.
