28 field tools for working out what you just found

Why a filter beats a wall of photographs

Search "caterpillar identification chart" and you get a page of photographs with names under them. That works if your caterpillar happens to be one of the first dozen shown, and fails completely otherwise, because scanning pictures is a linear search through a space that is not organised the way your observation is. You know the thing was green, about four centimetres, and had a horn on its rear end. A photo grid makes you check that against sixty pictures one at a time.

A key inverts it. You state what you observed, and everything that contradicts the observation disappears. Three well-chosen filters usually cut a list of forty candidates to two or three, and at that point the remaining work is comparing your specimen against a short, specific description rather than against a mood board.

That is the whole design principle of this section. Each tool asks only for features a person without equipment can genuinely observe — colour, count, shape, size against a coin, the substrate it was found on — and never for characters that require a microscope, a spore print left overnight, or a key you would need training to read.

Everything renders before the JavaScript does

The filters are JavaScript, but the data is not. Every entry in every key is written into the page as plain HTML: names, scientific names, sizes, ranges, the diagnostic sentence and the look-alike warning. Turn scripting off and you lose the ability to narrow the list, and nothing else. That matters for three groups of readers — people on locked-down networks, people using a screen reader that gets confused by dynamic filtering, and the search and language-model crawlers that do not execute scripts and would otherwise see an empty page.

It also means you can use a key with no signal. Load the page while you still have reception, and the whole catalogue is already in your phone.

Where the entries come from, and where the limits are

Each tool lists its sources at the bottom of its own page, and they are the obvious authorities for the subject: the ASPCA toxic plant database, USDA and Forest Service field keys, state extension services, the Mohs scale and standard determinative mineralogy, the World Meteorological Organization cloud atlas, regional natural history museums. Where two sources disagree, the more conservative reading wins — a key that under-warns is worse than one that over-warns.

Three limits are worth stating in the same breath. These keys cover common species in North America and Europe, and they say so; a key with forty entries cannot represent a continent with four hundred candidates, so a specimen that matches nothing is more likely to be outside the catalogue than to be exotic. Several groups — fungi, wild berries, spiders, snakes — carry a safety banner because a wrong identification there has consequences that a wrong maple does not, and in those groups the tool is designed to produce a shortlist for an expert rather than a verdict. And where an entry mentions a risk to a person or an animal, the correct next step is a poison centre or a doctor, not a web page.

How to get a usable answer in about a minute

1. Measure something. Almost every key has a size filter, and size is the single most discriminating character available to somebody without a hand lens. Put a coin, a car key or your thumb next to the specimen before you photograph it.
2. Start with the filter you are most certain about, not the one at the top. If you are sure about the number of toes and unsure about the colour, set the toes first and look at what survives.
3. Stop filtering when the list gets to five. Reading five short descriptions is faster and much more reliable than adding a fourth filter you are guessing at, and guessed filters are how you eliminate the right answer.
4. Read the "confused with" line. It is there because the most common identification error in each group is a specific one, and knowing which pair you are choosing between is most of the work.
5. If nothing matches, remove a filter rather than adding one. An empty result almost always means one observation was wrong, not that you have found something undescribed.

Which tool answers which question

The catalogue divides into two kinds of thing, and it is worth knowing which you need before you open one.

A determination key takes an observation and returns a shortlist. You tell it that the print had four toes and claw marks and was six centimetres long, and it removes everything that contradicts you. Keys work on things that exist in discrete kinds — species, minerals, cloud genera, point types — and they are the right tool when the question is "what is this".

A calculator takes a measurement and returns a number. Trunk circumference to tree age, a frost date to a sowing calendar, spot price to melt value. Calculators work on continuous quantities and they are the right tool when the question is "how much" or "when". They also carry their assumptions in the open: the tree age tool prints its error bars, and the coin tool asks you for the spot price rather than fetching a number nobody will check.

Several subjects appear in both forms, deliberately. The rock key names a rock; the Mohs scale tells you what the scratch test you just did actually means. The plant toxicity checker names a species; the seed calendar tells you when to sow one.

The safety flags are not boilerplate

Five of these tools carry a red banner at the top, and each one is there because a wrong answer in that subject has a consequence a wrong maple does not.

• Mushrooms. The death cap and destroying angel cause around 90 % of fatal mushroom poisonings worldwide, they look like ordinary field mushrooms, and the illness appears to resolve on day two before the liver fails on day three. That tool is built to produce a shortlist for an expert and it never authorises eating anything.
• Wild berries. Colour rules do not work — there are excellent red berries and lethal black ones — and the toxins that matter give no warning taste.
• Snakes. Most bites happen to people who were handling, killing or moving the animal. The tool is designed to be used from several metres away.
• Caterpillars. Roughly two dozen North American species sting, and the ones that look softest — the puss caterpillar in particular — are the worst.
• Plants around pets. The severity column exists because "toxic" is applied identically to a sore mouth and to kidney failure in eighteen hours.

Where an entry names a risk to a person or an animal, the correct next step is a poison centre, a vet or a doctor. In the United States that is (800) 222-1222 for people and (888) 426-4435 for animals; in the UK, 111 and your own vet. A web page is a good place to find out what something is and a bad place to decide what to do about it.

Nothing here sends your input anywhere

Every filter, every calculation and every search box on this site runs in your browser on your device. There is no server call when you tick a box, nothing is logged about what you filtered, and no result is transmitted. That is a consequence of how the tools are built rather than a policy statement: the data is already in the page, and filtering it is a few lines of JavaScript operating on the DOM.

Two practical benefits follow. The tools work with no signal once the page has loaded, so a key opened at the trailhead still works in the valley. And they keep working if this site ever stops being maintained, because a saved page is a complete copy of the tool.

What the paid apps charge

The other half of this question is what the identification apps in the store cost, and it is harder to find out than it should be: Apple lists in-app purchase names and prices but not the billing period, so a low tier is frequently weekly rather than one-off.

We publish the whole table — every in-app purchase tier for the major identification apps, read from the App Store and dated — because the single most useful fact in that market is which apps have nothing inside them to buy at all. The answer turns out to correlate almost perfectly with who publishes them: the free ones come from universities, public research institutes and non-profits.

What the app adds

A key needs you to describe the thing. Orvik does the opposite: it reads the photograph and proposes the species, then shows the near-misses so you can check the reasoning rather than take the answer on faith. The two work well in sequence — photograph it in the app to get a name, then bring the name back to the relevant key here and confirm it against the diagnostic features. An identification that survives both is one you can act on.

Everything on this page stays free and stays in the browser regardless. The tools are not a trial, they do not ask for an email, and none of them sends your input anywhere: the filtering happens on your device, which is why they keep working when the signal does not.