Evidence-aware family history
How to identify people in old family photos without guessing
A name is only useful when you can explain why it belongs to the person in the photograph. This guide helps you gather clues, test candidates, and record an honest level of confidence.
The rule before you begin
Uncertain identifications should remain uncertain. Do not convert a family resemblance, an old caption, or an online hint into a fact without testing it. It is better to preserve a well-documented possibility than attach the wrong name to a person for future generations.
Before researching
Start with three separate columns
Create a simple research note for the photograph. Keeping facts, interpretations, and questions apart prevents a promising theory from quietly becoming “known family history.”
Observed
What is visibly present or written on the photograph.
Inferred
What a clue may suggest, with the reason and limitations.
Unknown
What still needs evidence, comparison, or a record search.
The method
Work from the photograph outwards
Preserve the original context
Before separating a photograph from its album, envelope, frame, or box, record exactly where it was found. Photograph the album page and nearby images. Note who owned the collection and how it reached you. This provenance can be more useful than the image itself.
Evidence note: Record the source as a fact: “Found in Mary Evans’s album, page 14,” rather than “This must be Mary’s family.”
Make a careful digital copy
Scan or photograph the front and back at the highest practical quality. Include the full edges, mounts, studio marks, handwriting, and any damage. Keep an untouched master file before cropping, restoring, sharpening, or colourising a copy.
Evidence note: A photographer’s address, postcard format, paper type, or faint inscription may narrow the date or place later.
Transcribe every written clue
Write down names, initials, dates, locations, studio details, military units, and partial words exactly as they appear. Preserve uncertain letters with a question mark instead of silently correcting them. Search spelling variants only after keeping the original transcription.
Evidence note: Separate what the photograph says from what a relative remembers about it. Both matter, but they are different kinds of evidence.
Estimate a date range, not a single year
Use the photograph’s format, mount, studio history, clothing, hairstyles, uniforms, visible technology, and the apparent ages of known people to build an overlapping date range. Fashion alone rarely proves an exact date, especially when clothes were reused or inherited.
Evidence note: Write “probably 1910–1920 because…” and list the clues. Avoid “taken in 1914” unless a dated source supports it.
Build a shortlist from the family tree
List relatives who were alive in the estimated period, connected to the photograph’s place or owner, and approximately the right age. Include spouses, siblings, neighbours, and close family friends where the collection suggests they belong.
Evidence note: A shortlist is a set of candidates, not an identification. Do not eliminate someone only because one estimated clue seems imperfect.
Compare photographs cautiously
Look for several stable features across images: ear shape, hairline, eye spacing, nose structure, chin, posture, scars, jewellery, and relationships within a group. Allow for age, expression, camera angle, focal length, retouching, and image damage.
Evidence note: Resemblance can support a theory, but it should not stand alone. Automated face matching can suggest comparisons; it cannot turn an uncertain match into a fact.
Test the theory against records
Check censuses, civil registrations, parish registers, directories, electoral rolls, military records, newspapers, passenger lists, school records, wills, and addresses on certificates. Look for evidence that the candidate could have been in that place at that time.
Evidence note: Also seek contradictions. A candidate who had died, emigrated, or was documented elsewhere may need to be removed from the shortlist.
Ask relatives independently
Show the photograph without leading the witness. Ask what they notice, whether the setting is familiar, and who they think each person might be. Record their wording, the date, and how they knew the family. Where possible, ask people separately before sharing earlier answers.
Evidence note: “Aunt Joan identified him in 1998” is useful evidence. It is stronger when you also record Joan’s relationship to the person and whether she knew him directly.
Write a confidence-based conclusion
Summarise the strongest supporting and conflicting evidence. Use a label such as confirmed, probable, possible, unlikely, or unidentified, and explain why. Add source references so another relative can review the same trail.
Evidence note: If the clues do not converge, keep the person unidentified. An honest question mark protects the family record from a confident mistake spreading into copied trees.
Recording your conclusion
Use words that show how certain you are
Confirmed
Direct identification is supported by reliable documentation or a strong chain of independent evidence with no material conflict.
Probable
Several independent clues agree and alternatives have been tested, but direct proof is absent.
Possible
The candidate fits some clues, while important evidence is missing or another candidate remains plausible.
Unidentified
The available evidence does not justify naming the person. Preserve the clues and revisit the photograph when new material appears.
A useful identification note
“Probably Thomas Morgan, photographed between 1908 and 1914.”
Then explain the basis: the album owner, studio dates, estimated age, confirmed comparison image, matching address, and any remaining contradiction. Name every source another researcher would need to check.
Common mistakes
Clues that deserve caution
- Assuming every photograph in a family album shows a blood relative.
- Treating handwriting on the back as contemporary with the photograph without checking the ink, wording, or writer.
- Using clothing, apparent age, or facial resemblance as a single decisive clue.
- Copying a name from an online tree that has no cited source for the image.
- Colourising, cropping, or enhancing the only scan before preserving an untouched copy.
- Sharing a tentative identification without the confidence label and evidence trail.
Questions people ask
A little more clarity
Can clothing identify the exact year of an old photograph?
Usually not by itself. Clothing can help establish a broad period, but garments were reused, altered, inherited, or worn after fashion changed. Combine clothing with the image format, studio, known ages, place, and records.
Is facial recognition reliable for old family photographs?
It can help organise possible comparisons, but it should not be treated as proof. Age, pose, lighting, damage, retouching, and low resolution can all affect a match. Use it only alongside documentary and contextual evidence.
What should I write on the back of a photo if I am not certain?
Avoid writing directly on the original if possible. Add an archival note or digital record that states the proposed name, confidence level, reasons, source, and date of the conclusion—for example, “Possibly Ellen Hughes; identification unconfirmed.”
A careful next step
Preserve the question as carefully as the answer.
A photograph does not need a confident name to be worth keeping. Record what you know, show what you infer, and leave space for the evidence that may surface later.