Nobody can price, fund or value boxes. A two-day sampling survey turns them into counts, formats and condition grades that somebody can.
An archive described in adjectives gets quotes in adjectives; an archive described in numbers gets a price, a timetable and a second meeting.
Ask three scanning companies to quote for "about two hundred boxes of negatives and slides" and you should expect three very different numbers, each padded for the worst case. Ask them to quote for 134,340 frames in 502 folders, principally 6 x 7 transparencies, with a stated share mounted and a stated share needing cleaning, and the quotes have something to converge on. The archive has not changed. The description has.
This article sets out a way to get from the first description to the second in two working days, without cataloguing anything. Our angle is that a survey is a sampling exercise, not a small catalogue, and that a few hundred items chosen at random will tell you more than a month of opening the boxes that look interesting. We illustrate it with the one inventory we can speak about from experience: the published figures for the Robert Harper Photograph Library.
What does a survey need to find out?
Five things, each of which changes a price, a timetable or a risk.
| Question | Why somebody needs it |
|---|---|
| How many items, in what units? | Every quote, grant application and valuation starts from a count |
| In which formats and housings? | Each format needs a different capture set-up and enclosure |
| In what condition? | Dirt, decay and damage decide preparation time and urgency |
| How well labelled? | Captions on mounts and sleeves are the raw material of a catalogue |
| How many duplicates and variants? | Decides whether to scan everything or select first |
What it does not need to find out is what the pictures are of. Subject description is cataloguing, and it comes later, as our guide to cataloguing a photographic archive explains. A survey that drifts into admiring the contents will not finish.
How does the two-day method work?
Before you start: a table, good light, nitrile or clean cotton gloves, a 10x loupe, a ruler, sticky notes that touch only boxes and never film, a spreadsheet, and a way of generating random numbers. The Canadian Conservation Institute's handling advice applies throughout: hold items by the edges and never bend, fold or clip them.
Day one, morning: number every container. Walk the shelves and give each box, binder, drawer or folder a running number. Touch nothing inside. Record for each: its number, its type, any label, and its rough fullness. This list is the sampling frame, and it is worth having for its own sake.
Day one, afternoon: classify by opening and looking. Open each container long enough to record the dominant format (35mm, 120 roll film, sheet film, prints, Polaroids, mixed), the housing (mounts, sleeves, glassine bags, loose) and a quick count of units, such as sleeves or sheets. Do not count frames yet.
Day two, morning: sample. Use random numbers to pick containers: 30 is a workable figure for a few hundred. In each, take a fixed number of units, say every fifth sleeve up to four sleeves, and for every frame on them record the fields in the next section. That yields several hundred frames.
Day two, afternoon: scale up and write down. Average the sampled frames per unit, multiply by the units per container and the containers per format, and you have frame estimates by format with an honest margin. Summarise condition and labelling as percentages. Write one page.
Record what you did as well as what you found, so that a vendor or valuer can judge how far to trust it.
How big does the sample need to be?
Smaller than instinct suggests, as long as it is random.
For estimating a proportion, such as the share of frames that are captioned, the standard formula for a simple random sample at 95 per cent confidence gives:
| Margin of error | Items needed |
|---|---|
| Plus or minus 10 percentage points | about 96 |
| Plus or minus 7 points | about 196 |
| Plus or minus 5 points | about 384 |
| Plus or minus 3 points | about 1,067 |
Those figures hardly move whether the archive holds 20,000 frames or two million. What does move them is clustering: frames in the same box resemble each other, so 400 frames from four boxes tell you less than 400 frames from 40. Spread the sample across as many containers as time allows, and treat the margins above as a best case.
The discipline that matters is randomness. The boxes at the front, the ones with famous names on the spine and the ones somebody already opened are precisely the ones that will mislead.
What should be recorded for each sampled item?
Keep the form short enough to fill in for 400 frames without resentment.
- Container number and position in the container.
- Format and size.
- Type: negative, transparency, print, instant print, other.
- Colour or black and white.
- Film base, if it can be read. The Northeast Document Conservation Center notes that safety film is usually marked "Safety" on the edge, and that notch codes on sheet film help distinguish nitrate from acetate.
- Housing: mount type, sleeve material, loose.
- Condition grade, on a simple four-point scale: 1 good, 2 dirty or lightly scratched, 3 damaged but scannable, 4 at risk.
- Specific faults: dust, fingerprints, scratches, colour shift, mould, warping, odour.
- Labelling: none, job number only, caption, date.
- Duplicate status: unique frame, or one of a near-identical run.
- Under a loupe for prints: continuous tone or dots. The Image Permanence Institute's identification guides work at 10x and 50x; a survey only needs to separate photographic prints from printed pages.
Two of these are safety checks as much as data. Acetate film gives early warning by smell: in the NEDCC's six-level scale a vinegar odour appears early, before shrinkage and brittleness, and our article on vinegar syndrome explains why it spreads. Cellulose nitrate is a different order of problem. The Health and Safety Executive describes it as extremely dangerous, easily ignited and hard to put out. Kodak's nitrate production ended by format between 1933 and 1951, according to the NEDCC, so an archive that begins in the 1960s would not be expected to contain it, but a survey records what the edge markings say and does not assume. Anything suspected goes to a conservator, not back in the box.
What does an inventory look like once the counting is done?
The Robert Harper inventory is published in full, so it can stand as a worked example of the output. We describe only what it records.
| Measure | Figure |
|---|---|
| Inventory lines (folders) | 503 |
| Of which analogue | 502, holding 134,340 frames |
| Of which digital | 1 line: 42,803 NEF, TIFF and JPEG files |
| Unique photographs | 95,487: 52,684 analogue and 42,803 digital |
| Median analogue folder | 218 frames |
| Analogue folders under 100 frames | 117 |
| Analogue folders over 1,000 frames | 2 |
| Largest analogue line | 4,005 Polaroid prints |
| Share of analogue frames in the ten largest folders | about 9 per cent |
| Folders where shots equal frames | 48 |
| Folder recorded as already scanned | 1, of 842 transparencies |
The counts are the archive's own and have not yet been independently reconciled. Even so, three practical readings fall straight out of them.
The material is evenly spread. With a median of 218 frames and only two folders above a thousand, there is no single mass to do first. A pilot batch of almost any 20 folders would be representative in size, which makes pilots cheap to design.
Selection is the biggest lever. The inventory distinguishes shots from frames: 134,340 analogue frames against 52,684 unique analogue photographs, or about two and a half frames per shot. Scanning every frame and scanning one per shot are jobs of very different size, and the decision belongs to the owner, not the vendor.
Format is the main unknown. The inventory gives the principal analogue format as 6 x 7 colour transparency and records specific formats against only a handful of lines. One folder, a Cinzano shoot, shows why that matters: its description lists 10 x 8 inch transparencies, 5 x 4 inch film, 6 x 7 film and 19 Polaroids in a single job. A format-by-format sample of the kind described above is exactly what would close that gap.
The same counts feed the planning tools. The digitisation storage calculator turns frame counts by format into terabytes, and the scan resolution calculator shows what each format yields at a given setting.
Who uses the numbers, and for what?
| Reader | What they take from the survey |
|---|---|
| Digitisation vendor | Counts by format and mount, preparation load, pilot design |
| Funder | Scale, urgency from the condition grades, a credible plan |
| Valuer | What exists and how usable it is |
| Insurer | A schedule of what is held and how it is housed |
| Museum or library | Storage, conservation and documentation cost |
The last line is not hypothetical. The V&A's collecting policy says the total cost of an item, including conservation, documentation, imaging and storage, should be considered before acquisition. An owner who arrives with those quantities already estimated has done the institution's first week of work for it. The same file supports a valuer's questions and an underwriter's, and it is the natural appendix to a digitisation tender brief.
A survey also surfaces the questions that are not about film at all, such as what paperwork came with the pictures. That is the subject of model and property releases in older archives.
The next step for this archive
The complete inventory is our own first pass: every folder, with shot and frame counts, searchable and downloadable. It is a count, not yet a survey in the sense described here. Condition, film base, mount types and labelling have not been sampled and published, and more than half of the archive is still on film and has yet to be scanned.
Scanning and cataloguing specialists who would like to propose a sampling survey or a pilot batch will find what we are looking for on the digitisation partners page.
Frequently asked questions
How do you survey a photographic archive?
In two passes. First count and classify every container: boxes, binders, drawers, by format and housing. Then open a random sample of containers and record, for a fixed number of items in each, the format, film base, condition, labelling and whether frames are duplicates. Multiply the sample averages by the container counts to get estimates for the whole.
How many items do I need to sample?
Fewer than most people expect. Around 96 randomly chosen items give an estimate of any proportion, such as the share of frames with captions, to within ten percentage points at 95 per cent confidence. About 384 give five points. The figure barely changes with the size of the archive, provided the sample is truly random.
What should a digitisation vendor be told before quoting?
Counts by format and by mount type, the share that is sleeved or mounted, a condition summary, the amount of dust or damage, how items are labelled, whether everything or only selected frames is to be scanned, and the output specification. Without those a vendor has to price for the worst case.
Who should carry out an archive survey?
A careful non-specialist can do the counting and sampling with a written protocol, gloves and a loupe. Identification of film base, assessment of deterioration and anything involving suspected nitrate should be checked by a photograph conservator or an experienced archivist. Two people working together are faster and make fewer recording errors.
Is a survey the same as a catalogue?
No. A survey estimates what is there and what state it is in, at the level of the collection. A catalogue describes items or groups one by one so that they can be found. The survey comes first because it tells you how big the cataloguing job is.
Sources
- Northeast Document Conservation Center: Leaflet 5.1, A short guide to film base photographic materials
- Health and Safety Executive: INDG469, The dangers of cellulose nitrate film
- Image Permanence Institute: Identification of 20th century photographic processes (webinar slides)
- Canadian Conservation Institute: CCI Notes 16/4, Care of Black-and-White Photographic Prints
- [V&A Collecting Policy 2025 (signed off by the Board of Trustees, January 2025)](https://vanda-production-assets.s3.amazonaws.com/2025/08/26/13/42/22/c50582b5-a8fa-4591-9d49-ee29a828ac9d/V&A Collecting Policy 2025.pdf)
- Archive.London: Robert Harper Photograph Library, complete inventory
Published by Archive.London. Corrections are welcome under our editorial policy.

