Collectibles

There are 7 articles here , published between February 2026 and March 2026, and they answer questions rather than catalogue species: how to tell two similar things apart, what a particular mark or shape means, what to do once you know. They are the long form.

The faster route, when you have the thing in front of you, is one of the 1 free key for this section — 28 entries between them. A key does not ask you to describe what you are looking at in words; it asks a series of questions you can answer by looking, and the list shrinks as you go. Nothing is hidden behind the filter: every entry is written into the page, so the whole catalogue is readable with scripting switched off, and each one carries the single sentence that separates it from its nearest look-alike.

There is a third thing on the site worth knowing about from here, because it answers a question this section cannot. People arriving at an identification problem usually end up comparing apps, and the honest answer to "which one" depends on facts that change — what each costs this month, what it locks behind a subscription, who publishes it. Those pages exist under compare and alternatives, and not one figure on them is typed by hand: every price and every in-app purchase list is read from a dated App Store snapshot, and the build fails if a currency amount ever appears in the editorial text. It is the same discipline as the numbers below, applied to a subject that goes stale faster.

Reading a shortlist without being fooled by it

Whatever brings you here, the failure mode is the same and it is worth naming. A key, an app and a search all hand you a shortlist, and a shortlist feels like an answer. It is not: it is a set of candidates that survived the characters you happened to check, and the one at the top is usually the one that is common rather than the one that is right. That is often the same thing, which is exactly why it is dangerous when it is not.

Three habits fix most of it. Check a character you have not used yet — if the candidate survives a test drawn from a different part of the organism, the identification is much stronger than one built on colour alone. Check where you are: an enormous share of confident misidentifications are species that do not occur within a thousand miles of the observer, and geography alone settles them. And notice when the answer does not matter — a great many identifications are curiosity, where being wrong costs nothing, and the small number where it does cost something are worth treating completely differently.

Where the consequence of an error is medical — anything eaten, anything that bit somebody, anything a child had in their hands — no page here and no app is the right tool. Describe what you have to somebody qualified, and keep the specimen or a photograph. The keys are built to make that description accurate, which is a genuinely useful thing to be, and it is a different job from telling you it is safe.

One thing worth saying plainly about this section: there is no public occurrence database behind these subjects. For living things we can attach a real record — how many verified sightings exist, in which months, from where — because GBIF and iNaturalist publish exactly that. Nothing equivalent exists for the things catalogued here, so the keys carry the characters and the comparisons and stop there, rather than decorating the page with a number nobody can check. Where we cannot measure something, we say so instead of estimating it.

It is worth understanding why the gap exists, because it is not an oversight. Occurrence databases are built by people recording living things they encountered: a bird seen, a plant photographed, a specimen collected and catalogued by a museum. The institutions behind them — natural history collections, national surveys, millions of amateur naturalists — have spent decades pooling those records, and the result is one of the largest open scientific datasets in existence. Nothing comparable was ever built for objects, because objects are not observed in the wild in the same sense and there is no equivalent community pooling the finds.

The practical effect is that this section leans harder on the characters themselves. A key for living things can order a shortlist by how commonly each candidate is actually recorded, which is a genuinely useful prior. Here there is no such prior available, so the entries have to do all the work — which in turn means the diagnostic sentence on each one matters more, and a correction to it is worth more. If you know one of them is wrong, the contact page explains what makes a correction actionable.