A sepia photograph said to date from the 1880s shows a woman holding something to her ear that looks unmistakably like a mobile phone. You have probably seen a version of it. The picture is a staged modern joke dressed in old clothes, but the caption does all the work, and thousands of people share it as a genuine curiosity. A few hundred argue about it in the comments. Almost nobody spends the two minutes it would take to find out.

That is the uncomfortable truth about historical misinformation: it rarely needs to be clever. It needs only a plausible date, a grainy texture and an audience in a hurry. What follows is a working checklist — the kind of thing an editor keeps in mind when a strange photograph lands on the desk — for testing images, quotations and claims before you pass them on.

Separate the picture from the caption

The caption is a claim. The image is evidence. Your first job is to keep them apart, because most fakes are not forged images at all: they are real images with an invented story attached. A photograph of a flooded street in 1910 becomes a photograph of a flooded street in 1953; a portrait of an unknown soldier becomes a portrait of a famous one.

Start with a reverse image search. Crop the picture down to a face, a building or a distinctive object, and search each crop separately — a whole scene can return nothing while a fragment returns the original. Then look at the earliest appearance you can find, not the most popular one. Ask two questions: does that earliest version predate the date in the caption? And does it come with a different description? A museum catalogue, a newspaper archive or a national collection that labels the picture differently is close to a verdict.

Be wary of file metadata. Social platforms strip it, and a creation date proves nothing either way. If someone offers you an EXIF timestamp as proof of a photograph's age, they have misunderstood what it shows.

Read the photograph for physical clues

Old photographs are full of small, datable facts. You do not need to be a specialist to notice them; you need to look at the things a forger or a careless captioner would not bother to fix.

• Technology. Colour processes existed from the early twentieth century but were uncommon until mid-century, so a vivid colour street scene labelled 1910 deserves suspicion. Sharpness matters too: a photograph cannot be crisper than the optics and film that made it.
• Lettering. Shopfronts, street signs, posters and vehicle liveries are gold. Typefaces date, spellings date, telephone numbers change length, and a single readable brand name can place a picture within a few years.
• The built world. Overhead wires, road markings, kerb design, window glass, lamp posts and the make of a car are all clues. So is the absence of something you would expect.
• People. Clothing, collars, hair, teeth, shoes and posture are useful. So is behaviour: early cameras demanded long exposures, which is why sitters look stiff. Recreations tend to get the costume right and the body language wrong.
• The frame. Cropping is the cheapest deception available. Whatever sits just outside the picture often destroys the story inside it.

The anatomy of a suspicious quotation

Quotations fail in a few recurring ways. The most common is the drift from idea to sentence: someone summarises a writer's argument, a later writer puts the summary in quotation marks, and within a decade the author is saying something he never wrote in those words.

The line about the triumph of evil requiring only that good men do nothing is attributed to Edmund Burke constantly. It does not appear in his works. The sentiment is Burkean; the sentence is not Burke's.

Watch for these signals:

• Attribution to a category rather than a person — "a wise man once said", "historians agree", "an old proverb".
• No title, no chapter, no date, no speech, no letter. A real quotation can usually be located.
• Language that is too modern, too neat, or too flattering to the reader's own side.
• The word "attributed" doing heavy lifting, then quietly disappearing when the line is shared again.
• A genuinely famous name attached to a sentiment that suits the sharer's argument. Marie Antoinette and "let them eat cake" is the classic case: the phrase appears in Rousseau, written when she was a child in Austria, and no contemporary source places it in her mouth.

To check, take a distinctive phrase — five or six words, not the whole sentence — and search for it inside full-text archives: book collections, newspaper databases, digitised letters and out-of-copyright libraries. If every hit is a quotation website quoting another quotation website, you have found a rumour, not a source.

A ten-minute verification routine

1. Write the claim in one sentence, in your own words, including the date, the person and the place.
2. Reverse-search the image, including crops. Find the earliest version.
3. Look for an institutional source: a museum, national archive, university collection or major newspaper archive.
4. Search the caption's key phrase rather than the picture. Fakes are often exposed by earlier debunkings.
5. For a quotation, find the work. Read a page of it, not just the line.
6. For a claim about a "first", check what came before it — the word "first" is usually doing more work than the evidence can bear.
7. Ask what would change your mind, and whether you found anything that could.

Dates, numbers and the trap of "firsts"

Superlatives are where invented history thrives. "The first photograph of a smile." "The only surviving recording." "Banned by the government." Each is easy to assert and slow to disprove, which is precisely why it spreads. Over-precise figures — a death toll of 4,182, a fortune of £2.7 million — usually signal a number that has been laundered through several retellings, while round numbers often signal an estimate that has lost its label.

Also check simple chronology, which catches more fakes than any expert knowledge. A photograph cannot show an event that happened before the technology existed. A letter cannot mention a battle that had not yet been fought. A person cannot have said something in 1860 if they were born in 1871. And age is not accuracy: a genuine 1910 postcard can carry a false caption, and an authentic photograph can be dated wrong by decades.

The tricks that fool careful people

Colourisation and digital sharpening are the quiet offenders. When a black-and-white image is coloured or upscaled, the software fills gaps with plausible detail, and that invented detail then gets read as evidence — a badge, a button, a face. Staged recreations are another: reenactment photographs, film stills, museum dioramas and period dramas regularly circulate without their labels. So do real photographs from the wrong country, the wrong decade, or the wrong war, and screenshots of screenshots that have lost the article, the date and the correction.

None of this requires malice. Someone finds a striking picture, adds the date they were told, and posts it. The error enters the record looking exactly like the truth.

Before you share

You do not have to become an archivist. You have to be honest about what you actually know. If you found a source, name it in the post: "Imperial War Museums dates this to 1917." If you did not, say so — "I couldn't trace the original" is a perfectly good caption, and a far better one than a confident falsehood. When a picture is impressive but unverified, describe what you see rather than what you have been told it shows.

Keep a small toolkit and use it the same way every time: a reverse image search, one full-text archive, one or two museum catalogues, and a habit of reading past the first page of results. Most of all, treat a striking historical image the way you would treat a striking financial offer or a striking medical claim — with interest, and with one more question than the person who sent it to you.

Photo: Firmbee / Pixabay