Which document is current, and how to tell

The same safety panel, in two versions

"For Safety's Sake" is the panel in the front matter of the 2015 guide that tells you when you may eat low-acid home-canned food without boiling it first. It is one of the most consequential paragraphs in home canning, and it is the passage most often quoted second-hand.

NCHFP publishes the same panel as a web page. The web page ends by saying it was extracted from the 2015 guide. Read side by side, the two texts differ in at least five places [F-14.6, F-11.4]:

• The web version adds a CDC note that is not in the print guide, telling you that CDC's recommendation is to discard any home-canned food that might contain botulism toxin [F-11.4].
• The web version defines what boiling means — liquid actively forming large foamy bubbles that break all over the surface. The print guide never defines it [F-11.4].
• Condition one is expanded. Print: "Food was processed in a pressure canner." Web: processed in a pressure canner operated according to the procedures in the USDA guidelines [F-11.4].
• Condition five is rewritten. Print: the jar lid is firmly sealed and concave. Web: firmly sealed and indicates a vacuum seal is present [F-11.4].
• Condition eight gains a clause. Print: no unnatural or 'off' odors can be detected. Web: the same, plus no mold is present [F-11.4].

The web version also says "elevations" where the print says "altitudes" [F-11.4].

None of those edits is sinister and none of them makes the food less safe. Two of them are improvements. But they are edits, they are undated, and the page carrying them describes itself as an extract of a document that does not contain them.

That is the whole problem in one paragraph. If you quote the web condition five and cite it to the 2015 guide, you have published a sentence that is not in the document you named — and you had every reason to think it was, because the page said so.

Why the lag exists, and why it is not carelessness

A fixed publication and a live website update on different clocks. The publication needs a revision cycle. The website needs somebody to edit a page. When research finishes in between, the website moves first.

The research centre says this out loud. On its own "Equipment and Methods Not Recommended" page, directly beneath a verbatim reproduction of the 2015 line saying steam canners are not recommended, it explains that the research has since been done and adds that integrating the advice into the USDA guide "will take some time" [F-6.4].

That is not an organisation that has forgotten to update a page. It is an organisation that updated the page, left the superseded sentence standing above the update so you can see both, and told you which way the difference runs. The full version of that case is on steam canning: what changed, and when, which is the cleanest single example of the whole pattern.

Where the lag becomes a real disagreement

Not every difference is a lag. At least one is two current federal texts saying different things about the same question, with neither citing the other.

USDA 2015 makes the boil before eating conditional on eight things you must be sure of [F-11.3]. CDC, last updated in February 2026, tells every reader to boil all low-acid home-canned food, with no conditions offered at all [F-11.5, F-14.7]. Both documents are current. Both are federal.

There is a whole page on what to do about that — do I have to boil my home-canned vegetables — and the short version is that the disagreement resolves in your favour cheaply, because boiling costs you ten minutes [F-11.3] and settles the question either way.

Other places where the guidance runs out or disagrees with itself: lid preheating, where the guide assumes a step the manufacturer no longer recommends [F-14.2]; lid shelf life, answered two ways in one paragraph [F-14.3]; and the dial-gauge tables, which simply stop at 8,000 feet with no federal process above that [F-14.4].

A small thing about how the document was made

While comparing the print text against the web text, one detail turned up that is worth a sentence and not more.

The text layer of the published 2015 PDF, on printed page 1-23, still carries an uncorrected typesetting note addressed to someone named Ken, about how the process tables on that page were sized [F-14.13]. It does not appear when you look at the page. It is only in the file's text layer, which is where you end up if you search the document rather than flip through it.

It is not a safety finding and it says nothing whatever about the guidance, which is correct on that page. It is included here for one reason: it is checkable, and it is the kind of thing you only find by opening the actual file. A page that has read the PDF and a page that has read a summary of the PDF look identical from outside. Small verifiable details are how you tell them apart — including on this site, which is why it is printed rather than kept as a private proof.

The method: four questions for any page you find tomorrow

This works on any canning page, including this one.

One. Which document, and which part of it? Not "the USDA" — the Complete Guide to Home Canning, 2015 revision, printed page such-and-such. If a page cannot tell you the page number, it is quoting somebody who was quoting somebody. That is not always wrong, but it is not checkable, and on this subject checkable is the whole point.

Two. Is that the printed text, or the web version of it? They are different objects. Open the NCHFP page and the PDF side by side if the sentence matters. If the page you are reading treats nchfp.uga.edu and the 2015 guide as one thing, it has not compared them.

Three. What is the date on the text, not on the page? A site can print today's date at the top and quote a document from 2015. Both facts are useful and they are not the same fact. The date that matters is the one on the source.

Four. Does the page tell you where its source runs out? This is the one that separates good pages from confident ones. Tested guidance has edges — foods with no process, altitudes above the tables, questions two agencies answer differently. A page that presents the guidance as seamless has either not reached an edge or has walked past one without noticing.

If a page passes all four, you can generally trust the rest of it. If it fails the first one, nothing else on it is checkable, however right it sounds.