Plant verification tracker

How tall it really gets, which is also the size its 3D model is built at.

How wide it grows, which is what spacing on a plan hangs on.

Where it can be expected to survive winter.

Full sun through full shade.

Cut flower, fragrant and pollinator: the three tags that drive recommendations, so they are held to sources.

When it actually starts and stops, read off named sources for a cold-winter garden rather than from a catalog's season word.

What the sources call the color. The color CODE the page is painted with is still nobody's to check, and the row below says so.

The one the drought-tolerant tag is derived from, so getting it wrong moves a recommendation.

The horizon the height and spread are read at. Where no publisher states one, the record says the number is a house convention instead of pretending it is measured.

Collected against named sources even though the catalog has no field for any of them, so the answer exists before the field does.

Every one of these plants carries a written list of what stayed unresolved. A check that finds nothing to disagree about has usually not looked hard enough.

Every built-in plant's scientific name was checked against the botanical registries; 42 were corrected in that pass.

A standing audit checks that a plant's own values agree with each other. It cannot catch a value that is consistent but wrong, which is exactly the gap the source checks exist to close.

The biggest one. Bloom time is regional: a UK source's bloom window is not a zone 5 bloom window, so the rule for reading sources has to be designed before collection starts.

Next in line after the current pass, partly because the drought-tolerant tag is derived from it.

Collected word for word from source pages, but deliberately stored as raw material rather than scored or trusted yet.

Type, pH and drainage have no field anywhere in the catalog; soil advice appears only inside some plants' curated notes, its source named. As a field it would be the highest-value addition the catalog could gain, and it comes once the claims already shipping are sound.

The clearest case of the difference: 124 plant pages print a sourced Safety note today, as prose. What no plant has is a yes-or-no toxicity field, so a page without a note is silent, not cleared, and the shipped notes are due an audit before any field is added.

Discussed in the curated notes on some plants, as a regional question, because it is one: a plant that is invasive in one region can be well behaved in another. A field would need a per-region answer, not a yes or no, so like native ranges it waits on a schema decision first.

Sources describe color in words, 'rose-pink', 'creamy white', never as a color code, so the code itself has nothing to be checked against, on any plant, ever. The color WORD can be checked, and on the 176 plants above it has been; you can also simply look at the photo.

No botanical source publishes the shape of a plant. Each model is drawn from a procedural archetype built at the plant's real mature size, and that size is exactly what the height and spread checks discipline.

'Dramatic' or 'elegant' is a matter of taste, and no source can confirm taste.

Out of 1,704 that carry care at all. Not one plant is left with nothing: a plant we cannot advise on says so instead of going quiet.

Their pruning advice is withheld and the plant says it is not settled yet. The reasons are listed below, one by one.

Advice is written against how a plant behaves, not against its name: a shrub that flowers on last year's wood needs different timing from one that flowers on this year's.

Serious means the timing is the whole point: get it wrong and you lose a year of flowers, and on the woody herbs a cut in the wrong place can kill the plant outright. The rest is watering, feeding and tidying, where being a few weeks out costs nothing.

Every single one of them rests on a named authority that publishes the rule. Not one rests on a guess, and none of them is a plant we are still unsure about, because those are held back instead.

Lavender and its woody relatives do not regrow from bare wood, so a cut below the leafy growth can end the plant rather than cost it a season. Their advice says exactly that inside the instruction itself rather than leaving it to be found out, and it rests on a university horticulture library and named university horticulturists rather than on an extension service.

The calendar a gardener actually sees is built against their own frost dates, and it drops a cut wherever the season is too short to carry it. So what any one garden is shown is that many or fewer, never more. We would rather publish the ceiling than a number that is only true in one zone.

Every page is listed by name further down this section. We are not going to summarize them as one kind of authority, because they are not one kind, and rounding them up into a tidier phrase would be the first dishonest sentence on this page.

49 plants.

39 plants.

39 plants.

39 plants.

39 plants.

18 plants.

12 plants.

12 plants.

12 plants.

11 plants.

10 plants.

10 plants.

10 plants.

8 plants.

7 plants.

3 plants.

3 plants.

3 plants.

2 plants.

2 plants.

2 plants.

1 plant.

1 plant.

1 plant.

1 plant.

Each one is sorted into a behavior class and given that class's advice. The file records that as exactly what it is, and so do we: a rule applied to a plant is not the same thing as a plant somebody looked up. It is why 90, not 1,704, is the number in this section.

That is the line we hold. Watering, feeding and tidying advice is derived for every plant, because being a few weeks out on those costs nothing. The advice that can cost you something is only ever shown where a source was read.

Done on 2026-08-13 and 2026-08-14, claim by claim, with the URL recorded next to each one. 34 separate pages back the 90 plants that carry a source.

The checkmarks above earn their name partly because an independent pass re-fetches every page and tries to refute the first. Nothing here has had that. It is a lighter method and we are not going to describe it as the same one.

What it establishes is that a rule matches what the authorities publish, so the advice is not built on a machine's memory. Our own source ledger records that limitation in those terms, and no care check moves a plant's count at the top of this page by so much as one.

On Snowmound (Spiraea nipponica 'Snowmound') the sources disagree about which wood it flowers on. We took the answer that is harmless if we are wrong: read it as last year's wood and a gardener prunes after flowering and loses nothing even if we are mistaken. The other way round, a late-winter cut takes the entire year's flowers. The disagreement is recorded rather than tidied away.

Shrubs whose genus carries no published pruning rule we could apply, so each one needs deciding by hand.

Plants that look like they flower on last year's wood, where being wrong costs the whole display, so the call waits for a source.

Clematis, which the RHS sorts into three pruning groups that cannot be read off a genus name.

Magnolias, where the authorities say an established plant should not be pruned at all unless it is necessary, so every routine cut is switched off for them.

Climbers in a genus that flowers on old wood, held for the same reason as the shrubs above.

A honeysuckle entry that names no species, and the species is what decides the answer.

A spirea entry that names no species, and the species is what decides the answer.

A plant that will not regrow from bare wood, where a wrong cut kills it rather than costing a season.