The manufacturer comes first
A specification is taken from the company that made the watch: its own product page, its own technical sheet, its own catalogue. Diameter, thickness, calibre, power reserve, crystal, water resistance and stated accuracy all come from there.
When a reference has been discontinued and the manufacturer no longer publishes it, the next source is an authorised dealer that prints the exact reference on the page. Below that, we stop rather than guess.
Every record lists the sources it was built from, with the date each one was consulted. They appear at the foot of the watch page, where you can check them.
One reference is one watch
A reference number is not a model name. SRPD55K1 and SRPD53K1 are the same case with different dials, and a specification copied from one to the other is simply wrong, even when the difference is invisible in a photograph.
So the catalogue is keyed to the exact reference. Specifications, photographs and the retailer link all belong to that reference and to nothing else.
The one exception is regional coding. The same watch is sold as J1 in Japan and K1 elsewhere, and Seiko and Orient both do this. Where a suffix identifies the same watch built to the same specification, it is recorded as an alternative reference on the same record. Where a suffix marks a genuinely different watch, it gets its own record or none at all.
What does not count as evidence
Some sources read like facts and are not. None of the following is treated as evidence for a specification:
- Marketplace listings, including Amazon and eBay. Sellers copy each other, and a listing describes what is being sold, not what was manufactured.
- Forum posts, subreddits and comment threads.
- Search-engine snippets and AI-generated summaries. A confident sentence is not a source; it is a sentence.
- Other watch sites, unless they are quoting a manufacturer document we can then read ourselves.
This is not a judgement about those places. It is that a number needs a document behind it, and none of them is one.
Unknown stays unknown
The rule that costs the most and matters the most: a field nobody has published stays empty.
Lug-to-lug is the usual case. Most brands do not publish it, and it is the measurement that actually decides whether a watch fits a wrist. It would be easy to estimate one from the diameter. We do not, because an estimate printed in the same table as a manufacturer figure becomes indistinguishable from it.
When a record is missing something material, the page says so by name — Bulova does not name the movement in the Surveyor, and the record says that rather than filling the gap. Unconfirmed fields are listed on the page itself, not buried in an internal note.
A gap you can see is more useful than a number you cannot check.
Where the photographs come from
A photograph is a specification too. The wrong image on the right reference is the same error as the wrong thickness, and it is harder to spot.
Every image file carries a provenance record: where it came from, whether it is the manufacturer's own photograph or an authorised retailer's, how its identity was established, and on what date. Images we cannot attribute are labelled as such and are never used to represent the product to search engines.
The order is strict. Manufacturer photography first, then an authorised retailer. An image whose provenance is unrecorded may still appear in a gallery, clearly marked, but it can never be the picture attached to the product data Google reads.
Amazon is a shop, not a source
Where a watch has a retailer link, we earn a commission if you buy through it. That is disclosed on every page that carries one, and it changes nothing about how the specifications above are established.
A retailer listing tells us three things and only three: whether the exact reference is being sold, whether it is in stock, and where the link goes. It never supplies a dimension, a calibre or a photograph for the catalogue.
And the link has to be the watch. A listing for the same model in a different reference does not get a button. No button at all is better than a button to the wrong watch — the reader who clicks it has been told something false about what they are reading.
What we have not done
Most of what is written here is analysis of published specifications, not a wearing report. Where that is the case the page says so, and the language stays honest about it: a watch is 38 mm across and 11.4 mm thick on paper, and we will tell you what that usually means on a wrist, but we will not tell you how it felt on ours.
That distinction is built into the data rather than left to the prose. A review marked as specification analysis cannot emit review structured data, no matter how complete it is. Only a watch that has actually been handled and worn can.
When we get it wrong
We will. A manufacturer updates a specification sheet, a reference gets a mid-production change, or we simply read a number wrong.
If you find an error, write to us with the reference and, where you can, the source. Corrections to a specification are made against the manufacturer document, not against the argument — being told we are wrong is not the same as being shown it, and only the second one changes a number.
Every page shows the date it was last updated, so you can see how old the data you are reading is.
