What shipped
Newest first. Short entries — what you can now do, and whether it fully works yet. The longer stories about how any of it went live on the blog, and the how-does-this-work pages are in help.
-
By the time you can see leaf miner tunnels, spraying is too late
A new guide on leaf miners — the pale winding tunnels inside a leaf — and it argues with the advice on nearly every other page about them. **The grub is not on the leaf. It is inside it**, feeding in the layer between the top and bottom surfaces, which is why the tunnel looks drawn on under a film. University of Minnesota Extension states the consequence outright: a pesticide **"prevents adults from laying eggs but does not kill larvae that are already feeding within plant leaves."** **So by the time you can see a tunnel there is nothing useful to spray.** The larva sits behind a wall of leaf for about two to three weeks and then drops into the soil. That is why people spray this pest again and again and nothing changes — nothing was going to. The only stage a spray reaches is a small fly in April that nobody notices. **And whether it matters at all depends on which part of the plant you eat.** Minnesota again: "Leafminer activity has little impact on plant growth but can be destructive to vegetables grown for edible greens." On a beet grown for the root, or a tomato, mined leaves cost you almost nothing and stripping them off does more harm than the grub did. On spinach and chard the damage is not a symptom of a hurt harvest — **it is the harvest**. The page ends by refusing to give a threshold for when damage is "bad enough" to act on, because we could not find one published for home gardens and an invented percentage would be worse than the judgment you can make standing at the plant.
-
Finish a job the chat asked for and you land back in the conversation
**The chat asks you to photograph a leaf. You say remind me later. It becomes a task.** Then you do it — and until now that was the end of the line. The photo filed itself and the conversation that asked for it sat there, three rows up the same history, none the wiser. **Now there is a way back.** Finish a job that came out of a conversation and you get one more button: back to the conversation. It opens where you left off with the photo already attached, above a line saying which day it was and what you had been asked. **The date is the point, not an apology for it.** Somebody who just did a chore is not carrying a three-week-old conversation in their head, and a wall of old text with no explanation reads as a mistake. Saying *picking up from 4 August* turns the age into the thing that is useful: the app kept hold of a question you had stopped carrying. **It offers rather than jumping.** You might be halfway through a list of jobs, and being thrown onto another screen you did not ask for is how you lose your place. **And the photo is attached, not sent.** It waits in the box for you to send it, because the app writing a message in your voice and posting it as though you had typed it is not something we are willing to do to make a flow one tap shorter.
-
Five pot sizes were wrong because we read the wrong line
Maryland Extension publishes container sizes two ways: a rough sorting list that puts a dozen crops in a bucket together, and a proper table with one row per vegetable. We had been reading the bucket. **Summer squash is the one worth knowing about** — we said 4–6 gallons, the table says 10. Four gallons of soil under a summer squash is a plant that wilts every afternoon in July. Radish went the other way: we said 1–3 gallons, the table says one pint. Bush beans 4–6 → 2, cabbage 4–6 → 5, Swiss chard 1–3 → 2. The bucket was not even consistent with itself — beans appear in two different size groups on the same page, and chard appears twice in one list. Where a source names your crop directly, that is the number we use.
-
Four blank rows that were never blank
Garlic, sweet corn, kohlrabi and onion all have no container size on their page. Not because nobody checked — each one has a careful, re-verified record of exactly which university sources were read and why none of them answers the question. Alabama Extension tells you to put four to six kohlrabi in a five-gallon pot, which is a shared pot for several plants, not a size for one; dividing it out would invent a number nobody published. Garlic has two real sources that disagree. **All of that reasoning existed and none of it reached the page.** Now each of those four says plainly that the line is missing on purpose. The same was true of winter chill hours on currant, elderberry, goji, honeyberry and mulberry — the exact five crops that feature was built for. A new check refuses to let it happen to a sixth.
-
Garlic has a pot size now, and the reason it did not is worth telling
Garlic’s page said nothing about container size because our records said two university sources disagreed — Maryland asking for a 4–6 inch depth, and a garlic-specific source saying at least 8 inches. We do not pick a winner when two sources disagree, so the line stayed blank. **We went to check which source said the 8 inches, and there isn’t one.** All five of the extension pages behind our garlic record were read again, twice, two different ways. None of them mentions growing garlic in a pot. The one page containing the words "8 inches" says to plant cloves "6 to 8 inches apart and 2 to 3 inches deep" — that is how far apart, not how deep. Somebody wrote it down as a depth, and two later passes repeated it without checking. So garlic now reads 1–3 gallons, from Maryland, along with the part that actually decides whether you get a bulb: it needs 4–6 inches of depth too, and a wide shallow bowl will not do even if it holds enough soil.
-
How deep the pot needs to be, not just how much it holds
A pot can hold plenty of soil and still be too shallow. Three gallons in a wide, flat bowl grows no garlic bulb and no carrot — and until today we only ever told you the gallons. **Seven crops now show a depth as well**: carrot wants 10 inches, garlic 4–6, beets 8, and blueberries, raspberries and blackberries all want two feet. Every one of those numbers was already sitting in the extension pages we had read — the sentence that gives the gallons usually gives the depth in the same breath. We just had nowhere to put it. Most crops still show no depth, and that is deliberate: it is there where a university actually states one and where getting it wrong costs you the crop. We are not going to estimate the rest.
-
It recorded a plant you pulled out as one you planted
**"Pulled the dead plant out."** The app read the word *plant* and recorded a planting — a new plant in your garden’s history, dated the day one died. **That is worse than not understanding you.** A missed sentence costs you a fact; this one invents a fact, and every date the app works out for a plant hangs off its planting date. **The cause is that we had taught it verbs, and you name jobs by what you touched.** *"Pulling weeds."* *"Threw a sheet over them."* *"Stuck a cane in next to it."* Not one of those contains the word weed, cover or stake the way the app was listening for it. Across 20 such phrases it understood **three**. **The fix was not more phrases.** Counting up, those twenty sentences used 15 different verbs and only 7 different nouns — and the noun is the reliable part. Weeds mean weeding whether you pulled them, got them out, dug them up or cleared them. So the app now recognizes the *thing* as well as the doing. **3 of 20 became 17 of 20.** **And the false records got their own guard**, which does not depend on anyone thinking of the right sentence: no phrase about removing something can be recorded as planting, harvesting, transplanting or putting a tray outside — checked across 48 combinations, not a list of examples. Two of those combinations were wrong when we started. Understanding of what a gardener says is now **80%**, up from 73% this morning.
-
Look at a plant before you add it
Adding a plant, every crop in the list now has a **Look** button beside it — and the icons are finally there too. Tapping the name picks it as before; tapping Look opens a short page for that crop. **Its first line answers one question: can I plant this here, this week.** Not what the plant is — you can get that anywhere. Whether it belongs in *your* garden right now, worked out from your own last frost date. Yes, not yet with roughly when, or too late this year said plainly, because an annual is not coming back. Then the few numbers that decide it — room, pot size, and how deep the pot has to be. At the bottom, a button that brings you back to where you were with the crop already chosen. **That button is why the door exists**: wandering a catalog is only worth your time if it ends with something planted.
-
On the day of your first frost, the chat put it in the past
**"Will the tomatoes ripen before frost?"** turns on one number: how many days you have left. The chat is told that number, counted from your own frost date. **On the day itself it said "0 days ago."** Past tense, for something that had not happened yet — on the one night of the year when the answer matters most and the right move is to go out and cover things. It now says that is today. **A second one on the same line:** a plant with no bed and no spot recorded read as *"Tomato (tomato) — , in since July 15"*, with a comma sitting where a fact should be. A person skims past that. The thing reading it is not a person, and a hole that looks like a failed lookup is worse than one that says nothing was entered. **Both were found the same way** — by listing the situations that part of the app can be in and checking which ones the tests had never actually put it in. The eight tests covering it all passed, and all eight ran the same summer garden with everything filled in.
-
Blight is two diseases, and they need opposite responses
A new guide on tomato blight, and the shared name is the problem. **Early blight and late blight are different organisms that demand opposite tempos**, so a gardener who reads the wrong half either panics over a slow problem or shrugs at one that takes the planting in a week. **The tell is rings.** University of Minnesota on early blight: "Larger spots have target-like concentric rings," starting on "older foliage near the ground" and working up. Late blight makes "large, dark brown blotches with a green-gray edge" — no rings — **anywhere on the plant, including new growth at the top.** In muggy weather late blight also grows a thin white powdery layer. Early blight never does. **But the row that changes what you do is where each one spends the winter.** Early blight overwinters "in infected plant debris and soil" — it is yours, it comes back, and you manage it across seasons with mulch, rotation and resistant varieties. **Late blight mostly is not in your garden at all.** It survives a northern winter only "if protected in potato cull piles"; otherwise it blows in from the south or arrives on shipped-in transplants. Resting the bed does almost nothing about it. **And late blight is not even a fungus** — it is a water mold, *Phytophthora infestans*, which produces thousands of airborne spores per lesion in under five days and can cause "total crop failure if untreated". **Minnesota's instruction for it is not a product. It is a phone call** — contact the extension team, because "late blight is an important region-wide issue and neighboring crops are at risk". That is an unusual thing to ask of a home gardener, and it is the clearest signal that this half of the name is not an ordinary garden problem.
-
Stripes mean the grubs are in your roots, spots mean they are not
A new guide on cucumber beetles, and the two that share the name look almost identical. Both a quarter of an inch, both yellowish. **One has three black stripes down its back, the other has twelve black spots and a black head.** Standing over a squash plant that reads as a difference that does not matter. **Underground it is the whole story.** UC IPM on the striped beetle's larva: it **"feeds exclusively on cucurbit roots and can damage these crops when infestations are heavy."** The spotted beetle's larva feeds on corn, small grains, beans, sweet pea and grasses, and does "not usually cause noticeable damage in gardens." **So the marks tell you where the next generation is.** Stripes mean grubs in the roots of the plant you are standing over. Spots mean they are in the sweet corn or the lawn, and whatever is wilting your cucumber is not their doing — which sends you looking for the real cause instead of blaming the beetle you can see. **And we will not tell you which beetle spreads bacterial wilt**, because our source does not. UC IPM says only that cucumber beetles "may also spread cucumber mosaic virus or wilts in cucurbits" — no species named, and wilts in the plural. Plenty of pages assign it confidently; that confidence is not in the publication. **There is also no spray recommendation here**, because there is none on their page for home gardens. UC IPM says management of cucumber beetles is difficult, and what they do recommend is cloth over the seedlings at emergence, taken off when the plants can tolerate the damage.
-
The app can now tell you nothing is wrong with your tomato
Ask GardenTrack why your tomato leaves are curling and until today it had nothing written down for that, so it said so rather than guessing. **Now it has — and the answer is that usually nothing is wrong.** Leaflets cupping **upward** and rolling inward, starting on the lower older leaves while the rest of the plant looks well, is physiological leaf roll. University of California IPM is unusually blunt about it: **it is not caused by a pathogen, no action is needed, and it will not harm the plant** — nor the fruit that forms later. It follows heat, a wet spell, hard pruning or a knock to the roots, and it passes as soils dry. **The direction is the whole diagnosis**, so that is written down too. Leaves rolling **downward** with twisted, strappy growth are herbicide drift instead, and that one does not recover. That takes us to **86 problem descriptions**, covering a bit over three-quarters of the times a plant in our database names a problem. **This one is the first where the useful answer is "leave it alone"** — which is a thing a plant-problem app should be able to say, and most cannot, because a list of diseases has nowhere to put it.
-
The app only understood you once you had stopped working
**"My lettuce is going to seed."** That is the sentence, and the app filed it as a plain note — no event, nothing your records could ever count. Say *"gone to seed"* instead and it understood perfectly. Every phrase the app knew had been written down in the past tense: *watered the beans*, *pruned the plum*, *gone to seed*. **That is how you talk when you have finished. It is not how you talk with a hose in your hand.** Measured across 21 jobs: 100% understood in the past tense, **71% while you are still doing them.** **One of those was worse than a missed note.** Say *"treating the roses for aphids"* and the app recorded that you had **seen** aphids — the fact that you had done anything about them was gone, and the record said close to the opposite. **And taking a cover off was recorded as putting one on.** *"Taking the fleece off tonight"* landed as covering the bed. All fixed, along with *picking*, *cutting back*, *putting stakes in*, *sowing*, *thinning out*, *starting seeds indoors* and the rest. **The general understanding of what a gardener says went from 73% to 75%** on our own test set, and the mid-job phrasings went from 71% to 100%.
-
It is a calcium problem, and the calcium spray cannot reach the fruit
A new guide on blossom end rot — the sunken brown patch on the bottom of a tomato — and it takes apart both the popular answer and the popular correction. **Everyone says calcium deficiency. Half right.** NC State: the rot "is caused by a lack of calcium in the developing fruit." So the fruit really is short. **But your soil usually is not** — Clemson notes it "can occur even when calcium levels in the soil are adequate." Calcium travels to the fruit dissolved in water, so when watering swings between dry and drenched the delivery stops while the supply sits there untouched. **And the spray sold for it cannot work.** Clemson publishes it as a named gardening myth with two separate mechanisms, either enough on its own: "Calcium that is present in leaves is not able to move through the phloem into the fruit," and **"the skin of the fruit is not able to directly absorb foliar-applied calcium."** Spray the leaves, it stays in the leaves. Spray the fruit, it does not go in. There is no route from the bottle to the tissue. **The overcorrection is wrong too, which is why the page does not stop at "it is watering."** NC State does recommend lime or gypsum where a soil test shows genuine deficiency. The rule is test first, and most of the time the test says you did not need it. **One surprise worth knowing:** too much water causes it as well. NC State — "either an inadequate or excess moisture stress favors blossom-end rot development." The target is steady, not more.
-
Ask what a problem looks like and you get a sourced answer, with the source
**"What does powdery mildew look like?"** now comes back with text we wrote off a named source — a university extension service or a plant clinic — and it tells you which one, with a link. We hold descriptions for 85 of the problems our plants get, which covers just under three-quarters of the times a plant in our database names a problem. **Ask about one we have not written up and it says so.** That is the part worth having: an answer from memory reads exactly like an answer from a source, and only one of them can be checked. **What it will not do is guess between two things that look alike.** Ask about "mildew" and you get both the powdery and the downy kind, and it says plainly that it cannot rank them. These descriptions say what each problem looks like on its own; none of them says what tells it apart from the one beside it. **We tested what happens when a description gets used to tell things apart, and it made the answer worse** — it pushed the model off an answer it already had right. So it says it cannot, which is slower and true. **A note on how this nearly did not happen.** The descriptions had been written, sourced and checked, and nothing in the app read a single one of them. Written is not shipped, and from the outside those look identical.
-
The chat could lose your task list and answer as if the week were empty
**"What should I do this week?"** is the first thing most people type, and the chat answers it from the same list your Today screen shows — worked out on your phone and sent along with your question, so the two screens cannot disagree. **If that list could not be built, the chat said nothing about it.** Not an error, not a warning — the list simply was not there, and a missing list looks exactly like an empty one. So it would tell you there was nothing due, in a week that was full. That is the worst kind of wrong answer: confident, specific, and about the one question people ask most. **It now says so.** If it cannot see your list it tells you plainly and sends you to Today, which builds the list on your phone and is almost certainly still fine. Anything you asked for by name still comes through either way — those are yours, and they do not depend on the rest being worked out. **The part worth explaining:** the app already knew this was happening. It had been writing a note about it to a log on your phone that nobody can open, which felt like handling the problem and handled nothing. Telling the developer is not telling the person reading the answer.
-
The cucumber tell does not work on a squash
Our cucumber yellow-leaves guide is built on one shortcut: if the yellow patches have straight edges that stop dead at the veins, it is downy mildew and you are done diagnosing. Minnesota Extension calls those spots **"angular and bounded by leaf veins."** People kept arriving at that page with a squash or a zucchini, so we went and read Minnesota's separate diagnostic for summer squash. **The cause list is identical — the same nine, right down to the order they matter in.** Powdery mildew, cold, squash bug, virus, vine borer, bacterial wilt, spider mites, downy mildew, aphids. Nothing on one list is missing from the other. **But the description of downy mildew is not the same, and it is the one that matters.** On squash they write it as "Pale green to yellow spots on upper surface of leaves, later turn brown." No angles. No veins. So the single most useful thing you can do with a yellow cucumber leaf is not what they tell a squash grower to look at — check the underside for fuzz instead, which both pages agree on. Powdery mildew reads differently too. The guide now has a squash and zucchini section saying exactly that, cited to the squash page rather than stretching the cucumber page's citations to cover a crop they were never about.
-
The fusarium in your tomato bed cannot touch your beans
A new guide on fusarium wilt, and it exists to correct one belief that costs people a whole bed. **Gardeners meet this disease and conclude the soil is finished.** It is a reasonable thing to think and it is wrong. UC IPM, on its own tomato page: the disease is caused by **"a fungus that infects tomatoes only."** NC State states the general rule — "Fusarium species are often host-specific, attacking only a single host species or group." **So there is no such thing as "fusarium in my soil" in the way people mean it.** There is a tomato fusarium and a watermelon fusarium and a pea fusarium, and they are not interchangeable. The one that killed your tomatoes last August is still in that bed and it can do nothing at all to beans, lettuce, corn, brassicas or carrots. That is the difference between losing a crop and losing a bed. **It also means "fusarium wilt" is not really one disease** — it is a family of host-locked diseases that look alike and share a name. **And the page refuses to tell you it is fusarium rather than verticillium**, because UC IPM calls the two "extremely difficult to distinguish," noting only that fusarium browning "tends to be darker." Any page handing you a confident visual test for telling them apart is claiming more than the source does. It mostly does not matter: the response is the same either way — a resistant variety, marked F or FF, and grow something unrelated in that ground.
-
Sixteen frost jobs now name a temperature — including the orange tree, eventually
A job that said **"cover plants when frost is forecast"** was true and useless — it is the sentence you already knew. We hold a real temperature for sixteen crops that get that job, and it was sitting in the data with no way onto the card. **All but one said it straight away:** cover these beans when nights head toward 32°F, which you can check against tonight's forecast. **The sixteenth was the orange tree, and it is the reason this took a second pass.** The only figure we held for orange was 20°F — and that number is the temperature the tree is **still fine at**, not the temperature that damages it. Printed as a trigger it would read *"cover it when nights head toward 20°F"*, which tells you to wait for a night the tree does not mind. **The exact opposite of what the source says, on the one plant that prompted the work.** So the app said nothing, which was right but not good enough — a missing number is a thing to go and find, not a thing to explain. **It has been found.** Georgia extension publishes what actually happens to a citrus in the cold, and it depends on which part of the tree you mean: the fruit freezes at **26–28°F** if it stays there for a few hours, while the wood on a sweet orange takes until the low-to-mid 20s. The fruit is what you lose first and what ends your harvest, so that is the number the job now names. **The 20°F did not go anywhere** — it moved to the line that answers the other question, where the page now reads *"survives winter down to 20°F"* instead of implying damage begins there. One number, two questions, and it was only ever wrong about which one it answered. **A second kind of number had the same problem in a different direction.** Some frost figures only apply while a tree is in blossom — apple's is one. Said flatly in November, it would send somebody out to cover a bare dormant tree. Those now say *while it is in blossom* or they do not appear. **So the rule is: a number becomes a sentence only when we know which question it answers.** Anything we cannot read that way produces no sentence at all — the job keeps its plain wording, which is worse copy and true copy. **We would rather tell you less than tell you backwards.** The same pair of misreadings reached a plant page a week ago and was caught there too, which is why the fix this time went into the shared code both screens read from rather than into the screen. **The number was right every time. The words in front of it were wrong.**
-
A potted tree was being told to dig into a bed
**"Water. Pinch off suckers. Check for pests."** That is what a plant's own page said, and those three phrases are the whole of what it said. Meanwhile the daily run-through had been printing *"Check the leaves for aphids, scale and mites"* since July. Same app, same plant, two answers depending on which screen you opened. **The plant page now carries the instruction, the rhythm and the reason.** Every job says what to actually do, repeating jobs say how often, and one tap explains the word — because "pinch out suckers" assumes you know what a sucker is, and the sentence that says where to find them had been written and shown to nobody. **And it now reads three things about the plant it had been ignoring:** where it is sitting, whether it is mulched, and what you already logged. That last one is why a job goes quiet after you do it instead of sitting there all week. **The one that mattered most: a pot is not a bed.** "An inch of water a week" is a depth of rain over open ground and it means nothing over a container. A potted plant now gets a question — *lift the pot, a dry one is light* — and a stopping point: **water until it runs out of the drainage holes, then stop.** You never have to know how many gallons your pot holds, which is the point, because almost nobody does. **One bug fell out of it.** The check for "has this only just been planted out?" subtracted two dates inside a single calendar year, so a tree planted two summers ago read as ten days old every spring and was told to check a root ball that stopped existing two winters back. Before its own planting date the subtraction went negative, which also counts as "under two weeks". It affected 233 of 1,160 crop-days we simulated.
-
The plants go in the bed before we ask how big it is
**"I built a new raised bed and put a row of lettuce, spinach and kale in it"** now draws one bed with three plants inside it, straight away. **It used to arrive a turn late.** The first card showed three loose plants and a Save all button; the bed only turned up after you answered how big it was. **And that button was live** — taking it left you with three plants sitting nowhere and a bed that had to be joined to them afterwards. **The two are not the same kind of thing.** That the plants went in the bed is something you already said. How big the bed is, is a detail we can ask for after. Holding the first back until we had the second meant showing you a version of your own sentence with half of it missing.
-
The straight-edged patches are a cucumber thing
A new guide on downy mildew, and it takes apart the one shortcut every page about it teaches. **Look for patches with straight edges that stop dead at the leaf veins.** On a cucumber that is exactly right. NC State describes "angular, yellow to brown lesions on top of the leaves that are restricted by leaf veins." **Then everyone applies it to every crop, and NC State says plainly that it does not hold:** "identification of downy mildew can be more difficult in watermelon and cantaloupe than in cucumber since **lesions do not always have the characteristic angular shape**." So the best shortcut in cucurbit diagnosis is crop-specific, and using it on a watermelon will talk you out of a correct answer. **The reason the descriptions differ is that the organisms differ.** Cucurbit downy mildew is *Pseudoperonospora cubensis*. Basil downy mildew is *Peronospora belbahrii* — a different genus, not a variant. They share a common name and very little else. It splits again inside the cucurbits: one clade more often infects watermelon, pumpkin and squash, another cucumber and cantaloupe. **What works on every crop is turning the leaf over.** Fuzzy growth underneath the discolored patch — purple-gray on cucurbits, near-black on basil. On basil especially, because from above it looks exactly like a plant that wants feeding, and the fertilizer does nothing. **And one fact that inverts the usual advice:** on cucurbits in a cold-winter garden it is not in your soil. NC State says it "can only overwinter on plants in sites with warm weather such as the Southern United States and greenhouses" — it arrives on the wind each year. The exact opposite of fusarium wilt, which waits in the ground for years.
-
Three green caterpillars, one name, and the spray only works on one of them
A new guide on cabbage worms, which is not an insect. **It is at least three** — the imported cabbageworm, the cabbage looper and the diamondback moth larva. University of Minnesota describes all three on one page, and reading their descriptions together shows the problem with every identification guide ever written for them: **"velvety green", "pale green", "light green."** Appearance is the one thing these three share. **So the test is not looking. It is touching.** Minnesota separates them by behavior. The looper has no legs in the middle of its body, so it humps its back and inches along — nothing else here moves that way. The imported cabbageworm "moves sluggishly when prodded." The diamondback larva **"wiggle vigorously when touched"** and is tiny, up to a third of an inch. Three seconds and a finger. **And it is not pedantry, because the name decides whether your spray works.** Minnesota on Bt: use it "when diamondback moths are the dominant caterpillar found, or when other caterpillars are found while they are still young," because it is "highly effective on small caterpillars." The common factor is size. **So finding a full-grown looper in August and reaching for Bt is the right product at the wrong moment** — and the same spray three weeks earlier is the recommendation. That is how a good product gets written off as useless. The page also declines to give a threshold for how many caterpillars are too many, because we could not find one published for home gardens.
-
Two universities disagree about how big a raspberry pot needs to be, and now the page says so
Oregon State says a raspberry wants a 20 to 30 gallon container. The University of Maryland puts brambles in a bucket starting at 16 inches, about 10 gallons. **Both are real extension services, both are talking about the same plant, and they are roughly three times apart.** We have known that for a while. It was written in a curator's note, in a file nobody outside this project reads. **The page said 20 to 30 gallons and looked settled.** A number with a single citation under it reads as a fact that somebody checked — which it was — and gives no hint that a second institution checked the same thing and got a different answer. **Now the disagreement is on the page, next to the number.** We still lead with Oregon State's figure, because a gardener needs one number to act on and the bigger pot is the one that fails safe. But the sentence says the other exists. **Blackberry carries the identical wording, deliberately** — the same disagreement described twice in two different ways would drift apart over time, and two slightly different accounts of one fact read like two facts. This is the same machinery that put garlic's depth warning on its page this week: a short condition a reader actually sees, rather than a note only a curator reads. **The gap between what we know and what the page says was the whole problem, and it is not fixed by knowing more.**
-
Warmth is not comfort for seedlings — it is a shorter time at risk
A new guide on damping off, the thing that takes a tray of seedlings overnight, and it is the last of six new plant-problem pages. **The seedlings that fell over are dead and there is no treatment**, which sounds like bad news and is actually the most useful sentence about it. Every rival page has a treatment section. There is nothing honest to put in one. So the entire job moves to before you sow. **And the best line in the source is one almost nobody quotes**, because it does not read like disease control. Wisconsin says to germinate seeds at high temperatures **"so that seedlings rapidly grow out of the stage."** Read that carefully: warmth is not about comfort or speed for its own sake. The seedling is vulnerable for a window after it emerges, and a fast-growing seedling **spends fewer days in that window.** The disease has a clock and you are trying to beat it. **It is also several different organisms sharing one name** — Wisconsin lists Pythium, Rhizoctonia and Fusarium, "including (but not limited to)". That is the third time in this set that one name covered several things. **Unusually, here it does not matter which you have**: none can be treated, they all like cool wet soil, and the prevention is identical. Narrowing it down would need a laboratory and would change nothing. The rest is short — fresh pasteurized mix rather than last year's tray, do not overwater, and thin crowded seedlings early.
-
We were wrong about the other garden planning apps
Our home page told you that garden planning apps are a winter puzzle, and that **"come planting season they just make you feel behind."** The implication was that they hand you a drawing and stop. **That is not true, and we should not have said it.** Seedtime's own site offers "a daily or weekly task list that automatically compiles from your gardening calendar plan." Several others ship a calendar too. They do not stop at the drawing. The same block claimed your history is "locked up the moment you stop paying" — **a claim about other companies' pricing that nobody here had actually checked.** Also gone. **The honest version turned out to be better for us, which is the annoying part about holding a false one:** it hid the thing we actually do differently. Their reminder is counted forward from the date you said you planted, so **it arrives on schedule whether the seed came up, drowned, or never went in.** Nothing is watching the plant. Ours is watching. We hold **229 checkpoints across 149 crops** for a sign that should have appeared by now — first true leaves, first flower, first fruit — so the app can tell you when the thing that was supposed to happen didn't. **A date cannot see that.** We would rather be accurate about a competitor and win on the real difference than be flattering to ourselves and win on a made-up one.
-
Where you put the word "down" decided whether the job counted
**"Laid woodchip round the beds" was logged as mulching. "Laid down woodchip" became a plain note.** Same job, same material, same person — and the only difference was which side of the woodchip the word "down" sat on. **We had it filed as a missing word, and it was not.** Woodchip was already on the list. Every pattern for mulching wanted the material first and the direction second, so any sentence that put the direction in front fell straight through. Fixing the one phrase would have left "put down bark" and "tipped a barrow of compost out" broken right beside it. **We tested it on nine sentences written before the fix, and it got seven.** The two it missed were the same shape a third time — "laid the cardboard down", with the material in the middle — so that closed too. **One thing fell out of it that was worse.** "Took the compost out of the bag" was being recorded as putting plants outside to harden off — a real event, on a real date, for a plant nobody carried anywhere. Compost does not harden off. **And two of them we are leaving alone, on purpose.** "Cleared the bed" and "emptied the pot out" stay notes, because a bed is a place and not a job: clearing one could be weeding, digging, or pulling out a crop that had finished. We would rather keep your words than pick one of the three and be wrong. That is written into the tests now, so nobody quietly changes their mind about it later.
-
Which way the leaves roll tells you whether anything is wrong
A new guide on curling tomato leaves, and it opens somewhere no other page does: **with the possibility that nothing is wrong.** Every page on this symptom is a list of problems, and **the most likely answer is not a problem.** UC IPM on physiological leaf roll, in three consecutive sentences: **"This is not caused by a pathogen." "No action is needed." "It will not harm plants."** They add there is "no damage to the fruit that develop later." It comes on in wet or hot spells and eases as soils dry. **And the thing that separates it from real damage is not a color or a spot. It is a direction.** Clemson describes physiological leaf roll as "an upward cupping of leaflets followed by an inward lengthwise rolling of the leaflets toward the mid-vein," starting on the lower leaves. Herbicide injury is the opposite: **"downward rolling of leaves and twisted growth,"** often with white split stems and deformed fruit — and that damage cannot be reversed. **Up, from the bottom of an otherwise healthy plant, is nothing. Down, with twisting, is drift.** That is an unusually clean test for a garden symptom, and it decides between doing nothing at all and having a real problem. A virus is the third possibility and reads differently again — it cups the NEW leaves, pales them, puts purplish veins underneath and stunts the whole plant, rather than curling the old leaves on a plant that is otherwise thriving.
-
A bucket was never too shallow. That is not what stops it
New guide, on what a five-gallon bucket will not grow. **The thing everybody repeats about buckets turns out to be false.** Maryland Extension sorts container plants into three sizes, and the deepest one — tomatoes, peppers, eggplant, cucumbers, winter squash — asks for 12 to 16 inches. A standard bucket stands about fourteen. It is inside that band. **What actually stops a bucket is two different things that usually get called one.** Root volume, which a bigger container fixes completely. And the plant itself — how wide it grows, whether a twelve-inch base holds it up, whether you can still lift it. Nothing fixes that second one, and a bigger container makes it heavier. So fruit trees are out. Maryland puts a dwarf one at 25 to 30 gallons of potting mix and a shrub at 25 — five or six buckets' worth **in one container**, because roots cannot cross from one bucket to the next. The distinction we are most pleased with: **a fifty-foot chayote does not need a fifty-foot container.** A vine sprawls from roots that stay put, so how wide a plant grows rules out a tree and says nothing about a squash. Half the lists of what-not-to-grow-in-a-container treat those as the same fact. And the part where we mark our own homework: we hold a container size for 143 crops, 26 of them need more than a bucket, and **only 4 of those 26 carry a citation.** The other 22 are our estimate. So the guide argues from how big the plant gets — sourced on 91% of our records — rather than from gallon figures we cannot stand behind, and it says which is which on the page.
What you cannot grow in a 5-gallon bucketWhat to grow in a 5-gallon bucket
-
Seven fruits will never fruit alone, whatever size pot you buy
Last of four new guides on growing in containers, and this one starts somewhere odd on purpose. Every article about fruit in pots opens with what size pot. **That is the third question.** **A container garden is a one-plant garden.** Nobody puts two apple trees on a balcony — you buy one thing, in one pot, and find out three springs later whether it works. So the first question is whether one plant can fruit on its own. **Seven of the 26 fruit crops we hold cannot.** Apple, sweet cherry, European pear, persimmon, kiwi, honeyberry and mulberry all need a different variety flowering nearby. One on its own blooms every spring and gives you nothing, in any pot, forever. Every one of those seven carries citations — apple alone is Clemson, Michigan State, Minnesota and Utah State. What makes it cruel is that the failure looks exactly like patience. Three years is a normal wait for a young tree, so nothing in the first three springs tells you anything is wrong. **And here is why citrus is the classic patio fruit, which we had never seen explained.** Lemon, lime, orange, mandarin and grapefruit all need only 0 to 100 hours of winter cold — Mississippi State, all five. A peach wants 200 to 1,050. The one thing a pot does that the ground cannot is move, and citrus is the fruit you can carry indoors for winter without costing yourself the crop. **We left out the table you were expecting.** Pot size is cited on 4 of our 26 fruits. A pot-size table would have been 22 estimates wearing four citations, and you would have had no way to tell which was which.
Which fruit grows in a containerWhat to grow in a 5-gallon bucket