Is a Ball recipe USDA-tested? The people who maintain the guide answer

Two categories, and why the word "tested" hides the difference

Almost every recipe you will meet falls into one of two boxes.

Research-tested. Laboratory work: pH measurement, heat penetration experiments with instrumented jars, and generally published methods so that somebody else can evaluate the data [F-15.1]. This is what the processes in the USDA guide and on NCHFP's site are built on, and it is what the phrase "research tested" means when a food scientist uses it.

Manufacturer-validated. A company's own internal process for checking that its instructions produce a safe and good product. Newell Brands, which owns the Ball home-canning line, put its position on the record: "We're confident in our recipe validation efforts and the testing and review we go through to ensure we provide safe and quality instructions and recipes for our canners" [F-15.1]. No data accompanied that statement [F-15.1].

Both of those are real work. They are not the same work, and the difference is not about competence. It is about whether the evidence is external and checkable, or internal and asserted. On most subjects that distinction is academic. On this one, where the failure mode is Clostridium botulinum in a sealed jar and the failure is silent, it is worth being able to name.

What the criticism actually consists of

It is worth being specific here rather than gesturing at "concerns", because the specifics are milder and more useful than the summary.

Iowa State University Extension, in an announcement quoted by the same article, listed its concerns with the publications as not providing testing analysis and data to verify the safety of the finished product, a lack of precise measurements, and wordy explanations, mistakes and missing information [F-15.1].

Only the first of those is a food-safety claim, and it is a claim about publication, not about outcome. The rest are editorial: a recipe can be safe and still be imprecisely written.

The concrete example of imprecision that circulates from that article is not Iowa State's. It comes from John Larsen, a home canner in Utah, and it is the difference between four cups of raspberries and four cups of crushed raspberries — two quite different quantities of fruit behind the same line of text [F-15.1]. Larsen also said, in the same piece, that Ball has not been the source of the most egregiously unsafe recipes he has seen in the wild [F-15.1]. Both halves of what he said belong on the page, and pages that carry the first without the second are misreporting him.

The case that runs the other way

If this page only reported the criticism it would be a hit piece with citations, so here is the other side, which is equally documented.

Presto publishes canning tables on its own website whose figures for green beans and for corn match the USDA 2015 guide exactly [F-15.2]. Not approximately — exactly. That is a manufacturer taking the tested figures and reproducing them faithfully, which is precisely what you would want a manufacturer to do. The only odd thing about it is that the page never cites USDA [F-15.2], so a reader has no way to know that the numbers are the federal ones.

The same company tests dial gauges free by mail, for four brands including two it did not make [F-3.6, F-15.2]. Annual gauge testing is a real requirement that most canners have no local way to meet, and a manufacturer is the reason it is meetable at all.

And on the same Presto page sit the company's own low-acid instructions for its digital canner, about which NCHFP has said it has no information on the type or extent of the research and cannot independently verify the company's statements [F-14.12]. All three of those facts are true at the same time, on one website, and any page that reports only one of them is telling you something other than what is there. The appliance question has its own page: Instant Pots and electric canners.

And the layer below manufacturers

Worth knowing about, because it is where the categories stop being subtle: retailers write food-safety content. A farm-supply chain, a sweetener company and a jar retailer have all been found ranking on questions about which canning method a food needs [F-15.4]. Search results move, and those particular pages have not been re-opened for this site since, so treat the pattern as the finding rather than any one position on any one day.

That is not the same problem as a manufacturer whose validation data is unpublished. A manufacturer has a product-liability interest in its instructions being right. A retailer's blog is marketing with a food-safety topic, and the pattern — rather than any single page — is the durable observation [F-15.4].

How to tell which category you are holding

Four questions, and none of them requires you to know any food science.

Does the source say what was measured? A research-tested process rests on pH values and heat penetration data [F-15.1]. A page that describes testing without naming a measurement is describing a review, not a test.

Can you find the same food in the tested sources? For most common preserves you can, free, at NCHFP or in the 2015 guide. Where a manufacturer recipe and a tested process cover the same food and give the same figures, the category question stops mattering. Where the tested sources have nothing for that food at all, that absence is itself the important finding, and it is what tested and prohibited, never tested at all, and tested then withdrawn is about.

Is the recipe precise enough to follow one way only? Crushed or whole, packed or loose, peeled or not. Ambiguity is not a safety failure by itself, but it is the mechanism by which two people follow the same recipe and produce two different jars [F-15.1].

Does the source tell you where its authority comes from? "USDA-tested" is a claim that can be checked. "Trusted by home canners for generations" is a claim about how long a brand has existed. Both may be true and only one of them is about the food.

None of that makes a manufacturer the enemy. It makes you someone who can read a label. That is the whole ambition of this page, and it is more useful than a verdict would have been — because the verdict changes when a company changes its practices, and the skill does not.