How this site works

This is the page to read if you want to know whether to trust the rest of the site. It sets out what every page is held to, what the site refuses to do, and where it can be caught out.

Most of what follows is enforced in code rather than by good intentions. A page that breaks one of these rules does not get published looking slightly worse; it fails the build and does not get published at all. That distinction is the entire difference between a standard and a preference.

Every figure carries a citation, or the build fails

Any number in a canning context — a time, a temperature, a pressure, a pH, a headspace, an altitude — must carry a reference of the form [F-n.n] sitting next to it in the text. That reference resolves to an entry in the site's fact file, and each entry holds the verbatim source quotation, the document, the page, the URL, and the date the document was opened and confirmed.

The check runs on every build. It scans the text for figures in canning contexts and errors on any that has no citation within reach, and it separately confirms that every citation on the page points at a fact that actually exists. Both halves are necessary. An early version of the check accepted a citation holding two references as though it were no citation at all, while also skipping both references in the does-this-exist pass — which is the worst possible outcome for a gate like this: a figure that looks cited, passes the build, and points at nothing.

Why a gate in code and not editorial care? Because a wrong figure here is not a typo, it is a jar on somebody's shelf. Careful people do report their own figures as checked and turn out to be wrong about that — it is a well-documented property of proofreading your own work, not a character flaw — and the only defence that scales is a check that runs on every page, every time, and never gets tired or short of time.

This site publishes no processing schedules of its own

Not one time, one pressure or one headspace presented as this site's own. When a page reaches the point where you need a number, it names the tested document that holds it and links you there.

Two reasons.

A site that retypes somebody else's tables becomes a place where one typo can hurt somebody. The tables are long, repetitive, and full of near-identical figures that differ by jar size and elevation. Transcription errors in material like that are not hypothetical; they are the expected outcome at scale, and there is no version of this site where that risk is worth taking.

And it would be worse for you than the original. The tested tables are free, maintained by the people who ran the research, and updated when the research changes. A copy on this site would be a strictly worse artefact — later, staler, unauthoritative — offered to you for no reason except that it kept you here. What is missing from this subject is not another copy of the tables. It is a straight account of which copy is current.

The same logic governs whole clusters this site deliberately hands over. Where to get a dial gauge tested, the standard altitude charts, and whether bottled lemon juice belongs in canned tomatoes are all held thoroughly by Extension services, and the useful thing to do with them is to link out generously. That is not a concession. It is the cheapest available proof that the site is what it says it is.

"No tested process" means three different things

This is the standard that most separates this site from the pages around it, and the one it would be easiest to drop.

When a food or a method has no tested process, one of three quite different things is true.

Tested, and the answer was no. There is research behind it. A tested process was looked for, and the food or the method failed. This is the strongest claim in the set and the rarest.

Never tested at all. No laboratory has run the thermal process work. That is an absence of research, not a finding of danger, and the honest sentence is that nobody knows. It still means there is nothing to follow, because without the research there is no time and no pressure to give you — but it is not the same statement as the first one, and a reader who is told it is will eventually notice.

Tested once, then withdrawn. A published process existed and was later pulled. Shelled nutmeats are the clearest case: a process existed, it was withdrawn, and no replacement has appeared [F-14.10]. That is a stronger signal than never having been tested and a weaker one than a food that failed outright.

Almost every page in this niche collapses all three into "not recommended". That phrase is doing an enormous amount of unexamined work, and collapsing it costs the writer their most attentive reader — the one who correctly notices that "not recommended" and "will hurt you" are not the same sentence, and who starts discounting everything else on the page once they spot the elision. Every no-process notice on this site declares which of the three it is, and one that does not declare fails the build. So does one that says no without saying what to do with the food instead: a page that stops at "no" hands the reader straight back to the blog that says yes.

The three categories have a page of their own, with the documents behind each: tested and prohibited, never tested at all, and tested then withdrawn.

The currency stamp, and what it commits us to

Every page that rests on a document carries a small block naming that document, the date the document itself carries, the date this site last opened it, and any live caveat — a supersession, a contradiction with another source, a scope narrower than people assume.

That block is generated from one registry, not typed by hand, so a correction propagates to every page at once instead of being fixed in the place somebody happened to notice it.

Be clear about what the check date claims. It does not say the document was published on that date. It says a person went to that URL on that date and confirmed the quoted words still read that way, in that place. That is a claim about this site's own diligence, made in public, and it is deliberately easy to falsify. If a source changed last month and the check date printed here is older, the site is wrong on its own terms and you can see it without leaving the page.

Which is the point. A site whose whole claim is that it knows which text is current fails the moment a text changes and it does not. Re-checking the sources is not maintenance on this site — it is the product. The order the re-checks run in is fixed in advance: the pages most likely to move first, meaning steam canning and electric multi-cookers, then any revision of the USDA guide, then the lid-preheating position, then every Extension page quoted here word for word.

Unsafe methods are described, never instructed

Open-kettle canning, oven canning, dry canning, dishwasher canning, sun canning, turning a hot jar upside down to make it seal: this site explains what each one is and why the physics does not work. That explanation is necessary. A reader who does not understand the mechanism will be re-recruited by the next confident page they meet, because "don't" without "here is why" is just one voice against another.

What the site does not do is lay any of them out in a shape you could follow. Quoting a dangerous instruction in order to take it apart is allowed and often required — the green beans page is built on that method from top to bottom, and the page on open kettle, oven and the rest takes one mechanism at a time. Quoting an instruction in a form that reads as a procedure is a build error, even inside a criticism, even in quotation marks, even with a warning above it.

A seal proves a vacuum, not safety

A lid that pings and holds is evidence that a vacuum formed inside the jar. It is evidence of nothing else. In particular it says nothing about whether the contents ever reached a temperature that destroys Clostridium botulinum spores, because a vacuum forms from cooling and contraction and does not care how hot the food got or for how long.

Any sentence on this site implying otherwise is a build error in any wording, and the check for it exists because that implication is the single most common dangerous sentence in the whole subject. It is usually not even stated outright. It arrives as a reassurance at the end of a paragraph about something else.

Judgement and fact are marked differently

Where a page reports what a document says, it says "the guide says" and quotes it. Where a page makes a judgement — that one source is more useful than another, that a particular argument is weak, that a risk is worth taking seriously — it says "I".

You should never have to work out which one you are reading. Doing that consistently is most of what "trustworthy" means on a site like this.

Related, and worth stating: the site is legibly on the tested-method side of this subject and does not pretend to be neutral about it. It also does not sneer at the other camp, and where a contrarian argument is well made it gets engaged on the merits rather than dismissed. Where the reporting on something is thin, the site says the reporting is thin instead of borrowing outrage it cannot source.

What the site is not

It is not written by a food scientist, a Master Food Preserver or an Extension agent, and it is not a substitute for the guidance it quotes. See about, which is blunt about this.

It carries no sponsored content and no paid placement. The pages that tell you not to buy or not to do something carry no book link at all. The full ownership disclosure sits in the footer of every page, and privacy covers the one measurement tag the outbound book links are built to carry.

Two small structural choices follow from all of this. The site emits no recipe markup, no FAQ markup and no how-to markup, because it publishes no recipes and marking a page up as one would be a lie told to a machine rather than a person. And the emergency block about botulism symptoms appears on only the small number of pages where it belongs, because a warning printed on forty pages is a warning nobody reads.

Corrections

Send them. Specifically, and in rough order of how much they are worth:

A source that has moved. If a document this site quotes has been revised, re-dated or taken down, that is the most valuable message the site can receive, because it breaks the site's central claim at the exact place it is made. Include the URL and what changed if you know.

A quotation that no longer reads as quoted. Same thing at sentence level. Documents get edited without announcements.

A figure attached to the wrong citation. The gate proves that every figure has a citation and that the citation resolves. It cannot prove the citation is the right one for that figure. That gap is real and a reader with the source open is the only thing that closes it.

A missing condition. The most common way a true page becomes a dangerous one is by dropping a condition — an altitude adjustment, a time ceiling, a jar size the research covered. If a page here states a source's position with fewer conditions than the source does, that is a defect.

A sentence that reads as a procedure. If any page here reads as though it is telling you how to run a process rather than where to get one, say so, even if the underlying facts are right.

What happens next: the source gets re-opened, the page gets fixed, the check date changes, and the change is visible in the currency stamp on every page resting on that document. Corrections are not made quietly and then forgotten about; the dates are the record.

Contact has the address and what to include. I would rather be corrected in public than be right in private, and on this subject that is not a nice sentiment — it is the only workable arrangement.