Corrections
When we publish an error of fact, we correct the article, add a correction note to it, and record the correction here with the date. Each entry names the affected page, what the earlier version stated and what replaced it. Style fixes and minor copy edits are not logged; factual corrections always are.
Log
Pump type stated incorrectly. Criterion 4 described the Harvest Right Premier as an oil-free pump and said it removed routine oil changes entirely. The Premier is an oil pump; oil-free is a priced upgrade on Harvest Right's own bundle listing. The Premier lengthens the oil-change interval — a change and filtering every 20-30 batches, against every 4-5 on the standard pump — rather than eliminating it.
The error was a maintenance-cost misstatement material to a purchase decision, and it contradicted our own buying guide, which had the pump type right. Criterion 4 has been rewritten and the article carries a correction note. Found during an internal messaging audit, not reported by a reader.
A household water-activity method presented as a pass/fail check. An earlier version gave a step-by-step procedure for reading water activity with a $10–25 digital hygrometer calibrated against saturated salt slurries, and described the result as a pass against 0.60. No extension service publishes that protocol as a validated home method, and a numbered procedure ending in a threshold reads as a test you can pass — which is what a consumer hygrometer cannot give you, least of all at the low end of the scale.
The procedure has been removed and the page now points to the documented lower-cost method University of Minnesota Extension actually publishes. The same article's universal ≤0.25 aw target, which appeared in a diagram's description, its visible caption and an SVG label, has been removed from all three.
Recovery advice given before the handling history. Both pages recommended re-drying stored soft food before establishing how that food had been handled and for how long. A quality symptom cannot establish that food is safe to recover, and the food-group article compounded it by presenting fat oxidation as something that reliably precedes microbial risk.
The handling-history decision now comes before any recovery advice, and the categorical rescue instruction and the relative safety timeline are gone.
A warranty advantage that does not exist, and a wattage nobody publishes. The comparison listed Harvest Right's warranty as "3-year limited" and called it the clearest separation in the table. The sourced terms are one year of full coverage with the sealed system only in years two and three — the same shape of coverage the other two publish, not a separation. The same table gave Stay Fresh a "~900 W" average draw; Stay Fresh publishes no wattage for any home unit.
Both figures have been withdrawn, the warranty row now reads by year and by component, and the machine specifications for all three brands are kept in one record that both machine pages are checked against.
Two arithmetic errors visible without leaving the page. The buying guide called a $2,699 machine the cheapest medium option beside a $2,395 one. The cottage-economics example stated that 360 bags at $8–12 gives $1,400–2,100 a week; the same multiplication gives $2,880–4,320.
The "cheapest" claim and the weekly revenue projection have both been removed. The transparent per-bag cost model they sat beside is unchanged, with its assumptions and its demand assumption now labelled separately.
Product lines described as available, and described wrongly.Both pages recommended "Pantry" kraft bags, and the food comparison described a Trail line as 7 mil silver mylar. Neither line is available to order, and the planned specification differs from what was published.
Those selling paragraphs have been removed. The only product we sell is named where it is genuinely relevant, and it carries its ownership disclosure beside the mention.
A USDA moisture figure we could not verify. An earlier version gave 2.0% as a USDA purchasing-specification moisture limit for freeze-dried food. The exact specification document is linked, but it is a scanned image with no readable text, so the figure could not be confirmed against its own source.
The figure has been withdrawn rather than restated, and the question is recorded in the article's own account of what nobody has published.
Spotted an error? Report it. Send the page, the statement and a source if you have one; a report with a source is the fastest thing to act on.