Roanoke Farms, 1938: The Government Killed 211 of These 384 Photographs

In April 1938, the federal government paid for this photograph. Then someone in Washington put a hole punch through the negative.

The photographer stepped closer and took the painted child alone. That frame was killed too — no hole. It just never got printed.

One frame of the wall was kept. Captioned, printed, filed: brick wall, Fredericksburg, Virginia.

The wall advertised a photo studio: you take 'em — we develop 'em. The office in Washington was about to choose which of his pictures would exist.

The Library says it plainly: rejected images were classified as killed. Sometimes a hole was punched through them. Usually, it says, they were near duplicates.

The arithmetic of the killed file is still being done. A recent survey of the online negatives counted four thousand two hundred thirty-three punched holes.

Each one now carries a sentence written in our lifetime: a hole punch made by FSA staff, to indicate that the negative should not be printed. This record also names the frame's kept sibling.

In 2020, the Library went back and studied a single assignment, frame by frame. The photograph on the cover of its own report is one of the killed.

April 1938. Roanoke Farms — a federal resettlement project in Halifax County, North Carolina. Farmers line up for cotton seed they are buying cooperatively.

Certified seed, weighed out sack by sack. This is what Washington sent John Vachon to photograph — twenty-three years old, a year past file clerk.

Vachon's caption calls Nat Williamson the first Black farmer in the United States to receive a loan under the federal tenant purchase program. Here he reads over the paperwork with an official.

Williamson, his wife, and one of their children, at the log house in Guilford County.

Mrs. Williamson, on her own steps. Selected for the file.

Inside a resettled house at Enfield: the parlor stove, the mantel, a china horse. Selected.

The project built its houses from prefabricated panels, raised by the families themselves.

Lumber stacked for the next ones. All of this entered the file.

The community center on the project's Black settlement. The project was split in two: the white half was called Roanoke Farms; the Black half, Tillery.

Every roll came back to a typed caption sheet. A caption meant a picture was selected. And where a caption should be, again and again, one typed word.

The tally for this one trip: three hundred eighty-four photographs. One hundred seventy-three selected. Two hundred eleven killed. Sixty-eight of the killed carry a punched hole.

The verdicts sit side by side. The office kept: sharecropper and sharecropper's dog.

Every frame where the woman herself fills the picture was killed.

In the last of them she holds a child. That negative took the punch.

The wife of a resettled farmer at her washline: kept. The next press of the shutter, nearly identical: punched.

Whittling. The wide frame kept, the close one punched.

The fertilizer signs on a barn wall: one for the file, one with a hole through the middle.

The Pepsi-Cola barn: kept, and punched.

The furniture store that was also the funeral director's: kept, and punched.

Farm women on a porch in Halifax County. One frame kept; its neighbor punched.

Then there are the strips where nothing was kept. The wife of a former tenant farmer, on the porch with her husband. Killed.

Killed — and this time the punch fell at her collar, just under her chin.

And again. She is in the file three times. No one printed her once.

Why a hole punch? Because of how film is stored: strips of five, killed and selected side by side. The hole marked a kill.

Some kills were not even whole pictures: half a repeated frame, half a bare light bulb — and the catalog admits it is guessing which half was meant.

The photographers knew. Edwin Rosskam, on tape in 1965: Roy did an awful lot of what he called picture-killing. And the method — a punching of holes through negatives — was barbaric to me, because there is no way of telling, no way, what photograph would come alive when.

The Smithsonian's interviewer stopped being neutral: a lot of people share your opinion on that too. I'm one of them.

Ben Shahn remembered a storefront he had photographed in Marked Tree, Arkansas — prices marked on the glass. He went looking for the negative during the war. He punched a hole through it. When I was looking for it, in 1943 or '44, it didn't exist any more.

Sometimes the paperwork itself slipped. On this strip, the caption sheet lists one frame as killed — and the punch went through its neighbor. The office was moving fast.

The frame beside it — same man, same baskets — is the one the sheet meant, and it stayed in the file.

Arthur Rothstein kept his temper. Occasional negatives Roy didn't think were suitable — maybe they were duplicates, maybe it wasn't the best. And then: we don't do this today. We save everything.

The punching stopped in mid-1939, as far as the Library can tell, after the photographers protested. The killing didn't.

In 1973, Stryker put his name on a book. The introduction, by Nancy Wood, says he killed about a hundred thousand, by punching holes. It is arithmetic, not a count — the shot total minus the negatives kept.

The Library counted the actual holes: four thousand two hundred thirty-three. And Stryker, on the same page, calls the file narrowed down to one hundred seventy thousand — no punch mentioned.

At the end of his life, Stryker gave the Smithsonian eight hours of interviews about the file. The hole punch never comes up. Nobody asked. He never said.

The office kept receipts. Every selected picture got a typed caption card; the Library still uses them as its shelf list. This card belongs to a photograph the caption sheet says was killed.

Main Street, Chocowinity: selected.

Its neighbor was marked killed on the sheet — and printed anyway, filed, given a card. One frame out of three hundred eighty-four crossed back.

The road into town took the punch.

And the inventory has gray cells. Of Vachon's three hundred eighty-four negatives, only two hundred sixty-five exist at all today.

The grandfather: selected, printed, still here. Thirty-four selected frames are gone — one keeper in ten, not even on the 1944 microfilm.

The farm boy with the horse, killed —

— and the farm boy, kept. When both survive, you can watch the choice. For eighty-five kills here, nothing survives to watch — discarded, or handed back.

In 2020 the Library categorized every kill it could see. Near duplicates, mostly. A few camera-conscious; two or three flawed. And thirty to thirty-five: inexplicable. I saw no intent to conceal anything.

The girl at the washline. The punch fell in the sky above her.

The woman in the sun hat. Killed, no hole, no explanation.

The little girl at the door.

A woman bends over the child at her side.

Two girls in front of a house in Beaufort County.

A child crossing the street in Washington, North Carolina. Punched.

A mother and three children at their front door. Killed on sheet film — no hole, just a stamp on the envelope, and sixty years in the dark.

People have gone looking for censorship in the kills. An archivist surveyed six series with strong potential for politically troublesome material, and reported: not much evidence of cleansing.

Vachon photographed the drinking fountain on the courthouse lawn at Halifax — a man drinking, and the sign nailed to the tree above him. This frame was selected.

He photographed the corner again — a man resting on a wagon under the same tree. Punched.

And again, closer to the sign. Punched.

The fourth frame was killed without a hole. The one the file kept shows the sign as plainly as any of them. Nothing here was hidden. Somebody chose fast, and moved on.

The Library's verdict on its own file: killing is not identical to hole-punching. And the kills show not concealment but — I see hasty work: a government agency processing images at a pace nearly as rapid as journalism.

Some of what the office kept is as strange and fine as anything it killed. The scarecrow made the file.

Rosskam, 1965: the historical view is so different from the contemporary view — it is impossible to foresee what the historical view is going to be. Sixty-eight thousand times, somebody guessed.