Industries / Food, farming and forestry / Food and beverage processing / Bakeries
Shelf life
Shelf life is the work of keeping bread, cakes and pastries soft, safe and free of mould and bacteria after they leave the oven, and of dealing with what goes unsold.
6
Milestones achieved, from 1853 to 2024.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1853
Staling is not drying out
People long thought bread went stale because it dried out. The French chemist Boussingault found that a loaf lost no more than 1% of its weight as moisture in the 48 hours after baking. He also found that a stale loaf became like new bread again when reheated in an oven, although reheating drove off more water, and that this could be done several times with the same loaf. Douglas Maclagan reported this in the Edinburgh Medical Journal in 1855, citing the Journal de Pharmacie of February 1853. A 2024 study calls the retrogradation of amylopectin "a major contributor to staling". In retrogradation, starch that swelled and softened in the oven slowly re-forms stiff crystals as the bread sits.
Credit
Boussingault
1918
Acid stops rope in bread
Rope is spoilage in which bacteria turn the inside of a loaf sticky and stringy, with a sour, rotting smell. Bakers had added lactic or acetic acid to dough by rule of thumb. Working at Harvard, E. J. Cohn, S. B. Wolbach, L. J. Henderson and P. H. Cathcart, two of them officers in the US Army's Sanitary Corps, showed that the rope bacterium, which they called Bacillus mesentericus, grows only when bread is not acid enough. pH is a scale of acidity on which lower numbers mean more acid. In their tests, inoculated loaves below about pH 4.8 stayed free of rope, while most less acid loaves roped. They advised bakers to keep dough acid enough and to check it with methyl red, a dye that turns red in acid.
LessonWartime loaves stretched with wheat substitutes were less acid than white wheat bread, and so more likely to rope.
Credit
Cohn, Wolbach, Henderson and Cathcart
1939
Propionate patented against mould
Charles Hoffman, Gaston Dalby and Thomas R. Schweitzer of the Ward Baking Company patented adding propionic acid or its salts, such as sodium or calcium propionate, to bread dough before baking to hold back mould. The US patent was granted on 18 April 1939, and one claim caps calcium propionate at about 0.5% of the flour's weight. A 2022 review says the bakery industry is now "heavily dependent on preservatives", generally the sodium, potassium or calcium salts of propionic and sorbic acid.
Credit
Hoffman, Dalby and Schweitzer, Ward Baking Company
1989
An enzyme slows staling
Tine Olesen of Novo Nordisk, the Danish enzyme maker, applied for a patent (Danish filing of 27 September 1989) on adding an exoamylase to dough to slow staling. An exoamylase is an enzyme that trims starch from the ends of its chains; the patent names a maltogenic amylase made by a Bacillus strain. The patent says the effect is believed to come partly from less retrogradation, and that an overdose is less likely to make bread gummy than with ordinary amylase. A 2015 review says Novamyl, Novozymes' maltogenic amylase, "is used commercially in the bakery industry for improved freshness".
Credit
Tine Olesen, Novo Nordisk
2017
Bread returns measured in Sweden
Under a take-back agreement, a supermarket pays a supplier only for what it sells and returns the rest. Mattias Eriksson, Ranjan Ghosh, Lisa Mattsson and Alisher Ismatov compared how Swedish supermarkets reject bread, fresh fruit and vegetables, and milk. For bread they found a full take-back agreement: the store paid only for bread it sold, and any bread still unsold three days before its best-before date went back to the bakery. Of the three foods, bread had the most waste and the most extensive take-back policy.
LessonThe authors conclude that waste rises when the business that causes it does not pay for it, as when a store orders bread but the bakery bears the cost of what is not sold.
Credit
Eriksson, Ghosh, Mattsson and Ismatov
2024
Ropy cakes make workers ill
On 12 May 2024, employees at 11 sites of an Ohio business ate cakes from a retail bakery at a celebration. Twelve had stomach symptoms between 25 minutes and 4.5 hours later. Investigators saw rope strings in the torn cake, and the Ohio Department of Health laboratory found Bacillus velezensis, a bacterium used in farming as a natural crop protectant. The report, published in Applied and Environmental Microbiology in March 2025, says this "appears to be the first mention" of the species as a cause of rope.
LessonThe authors call for study along the whole food chain, because spores of a crop-protection bacterium may survive milling and grow in moist cakes.
Credit
Donald E. Brannen, Greene County Public Health, with the Ohio Department of Health laboratory
Watch for
Cultured wheat replacing propionate
Corbion says its Verdad Essence WH150 extends the mould-free shelf life of white and wheat breads, buns, flatbreads and tortillas. It is a fermented wheat powder, labelled as cultured or fermented wheat, sold as a drop-in alternative to synthetic preservatives such as calcium propionate. Corbion pairs it with its Natural Mold Inhibition Model, software that it says predicts preservation results early in recipe work. The announcement gives no count of mould-free days.
Achieved whenan independent laboratory publishes mould-free days for the same bread made with it and with calcium propionate at stated doses.
Claimed by
Corbion, September 2026
Enzyme claims 34 fresh days
IFF launched POWERFRESH ACE 2000, an enzyme for keeping industrial bread fresh, in the US market. It says the enzyme gives white and whole-wheat bread better softness, resilience and cohesiveness "even at lower dosage levels compared to market alternatives", and that in "company-led trials" bread kept its structure and look for up to 34 days.
Achieved whenan independent study measures crumb firmness over storage for bread made with it and with an existing maltogenic amylase, at stated doses.
Claimed by
IFF, August 2025
Open
01
Mould days against the date
A public test of how many days packaged breads stay free of visible mould, set beside the best-before date printed on the bag. Unopened packs of widely sold loaves, rolls and wraps are kept at two room temperatures and photographed by fixed cameras every few hours until the first spot of mould. The photos, labels and timings are published as an open dataset.
The briefCredit · unclaimed02
A staling meter for bakers
A low-cost firmness tester for bread slices, with an open parts list, 3D files and software, checked against a lab texture analyser. Firmness, the force needed to press a slice, is the usual measure of staling, so the tester would let a small bakery or a school measure how fast its bread stales and whether an anti-staling enzyme helps.
The briefCredit · unclaimed
Open 01 · the brief
Mould days against the date
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public test of how many days packaged breads stay free of visible mould, set beside the best-before date printed on the bag. Unopened packs of widely sold loaves, rolls and wraps are kept at two room temperatures and photographed by fixed cameras every few hours until the first spot of mould. The photos, labels and timings are published as an open dataset.
Why it is still open
Consumer tests have looked at storage or taste, not at mould against the date. Which? compared five ways of storing a homemade white loaf in 2019, and CHOICE in Australia rated 30 white breads in 2019 on taste and nutrition. Published work on finding mould uses close-up images: a 2023 study classified bread photos taken through a 50x clip-on phone lens, and its authors call visual checks and expiry dates "imprecise". A web search found only small home experiments comparing brands.
Built with what exists
Raspberry Pi computers with camera modules, or old phones running a timer app; a cabinet with a temperature logger; Label Studio, an open-source labeling tool, for marking the first frame with mould; an AI model such as Claude Fable 5.2 (Note: "Three working devices"), or an open object detector such as YOLO, as a second reader; Zenodo for the data.
First step
Buy three packs each of five breads, set one camera over them on a shelf with a temperature logger, and photograph them every six hours for two weeks.
Done looks like
At least 30 products bought in at least two countries, three packs of each, kept at 22 °C and at 27 °C (180 packs). For each pack, two people agree the hours to first visible mould, set beside its printed date, and all photos are published. A model trained on part of the photos flags the first mould within 12 hours of the people's call on at least 90% of the packs it has not seen.
Open 02 · the brief
A staling meter for bakers
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A low-cost firmness tester for bread slices, with an open parts list, 3D files and software, checked against a lab texture analyser. Firmness, the force needed to press a slice, is the usual measure of staling, so the tester would let a small bakery or a school measure how fast its bread stales and whether an anti-staling enzyme helps.
Why it is still open
Research papers measure bread firmness on lab texture analysers such as Stable Micro Systems' TA.XT Plus, pressing a 36 mm round probe into a slice about 20 mm thick. Low-cost compression testers exist: a 2026 HardwareX design built around an Arduino was checked against a commercial testing machine on silicone samples. Searches of GitHub, Europe PMC and the web found other prototypes for chips and fruit, but none tested on bread or released for it.
Built with what exists
A 5 kg load cell and an HX711 amplifier board; an Arduino or Raspberry Pi Pico; a stepper motor and lead screw, or a hand lever; a 3D-printed 36 mm probe; Python for logging; a texture analyser borrowed from a university food lab for the check.
First step
Build a bench version from a load-cell kit and compare it with a university food lab's analyser on ten slices.
Done looks like
Readings on at least 60 slices from three kinds of bread, taken 1, 3, 5 and 7 days after baking, agree with a TA.XT-type analyser within 10% on average. All files are published under an open licence, and two people outside the team build the tester and repeat the result.