Staling is not spoilage.

Most bread is thrown away while it is still perfectly safe to eat.

That sentence is worth sitting with, because almost every commercial decision about bakery shelf life is made as though the opposite were true. The date on the pack is set by microbiology. The complaint that ends the repeat purchase is about texture. Those two things run on different clocks, and the second one is nearly always the faster.

A loaf does not become unacceptable because something grew on it. It becomes unacceptable because the crumb is firm, the surface is dry and the eating quality has gone. Ask a consumer why they threw bread away and they will say it went hard, not that it went off.

Staling is not the bread drying out

The most common explanation is also the wrong one. Bread does not stale because water leaves it.

Seal a loaf in a bag that loses nothing at all and it will still stale, on more or less the same schedule. Weigh it before and after and the total water is much the same. Something else is happening, and it is happening to the starch.

What the oven did, and what the shelf undoes

In the oven, starch granules take up water and gelatinise. The ordered crystalline regions inside them melt, the polymer chains uncoil, and the whole system becomes a soft, hydrated, disordered gel held inside a protein network. That disordered state is what soft bread is.

It is also not the state the starch wants to be in. Left alone, the chains slowly find each other again and reassemble into ordered regions, squeezing out the water they were holding as they do. That is retrogradation, a major contributor to crumb firming alongside moisture redistribution and other changes in the bread matrix.

Two clocks, not one

Starch is two polymers and they retrograde on completely different timescales.

Amylose is the linear one. It reassociates quickly, largely as the loaf cools, and it is what sets the initial structure of the crumb. By the time bread reaches ambient temperature that process is essentially finished. It is not what a baker means by staling.

Amylopectin is the branched one, and it is slow. Its short outer chains recrystallise over hours and days at ambient temperature, and this is the process that firms the crumb across shelf life. When someone says a loaf staled, this is the mechanism they are describing.

The practical consequence is that the thing you can still influence on day three is amylopectin, and anything that slows its recrystallisation buys real shelf life in the sense the customer cares about.

The water moves without leaving

Water does not disappear during staling but it does not stay where it was either.

As the starch recrystallises it releases water into the crumb. At the same time water migrates outward from crumb to crust, following the difference in water activity between them. So the crumb firms and the crust, which left the oven crisp, softens and toughens. Two opposite defects, one movement of water, and neither one visible on a weight check.

This is also why the crust of a part baked loaf behaves so differently from the crust of a fully baked one. The interface has had a different history.

The fridge makes it worse, and the freezer stops it

Retrogradation is temperature dependent in a way that catches people out.

It is fastest at temperatures just above freezing. Deep in the freezer it effectively stops, because the water is immobilised and the chains cannot move to find each other. At warm ambient it is slower than in a fridge. So the refrigerator, which people reach for instinctively to keep bread longer, is close to the worst place to put it, and the freezer is the best.

That single fact explains a great deal of frozen dough and part baked practice. It also explains why a product that cycles repeatedly through the zone just above freezing ages faster than one held steady, which is the subject of a separate note on freeze thaw.

Why reheating works, and why it does not last

Warming stale bread makes it soft again. Heat melts the amylopectin crystals that formed, and the crumb returns to something close to its original state.

It does not last. Once the loaf cools, recrystallisation restarts from where it left off, and it restarts faster than the first time because nucleation sites are already present. The bread will be worse on the second cooling than it was before it was warmed.

For a bakery this is useful in two ways. It confirms the mechanism, because nothing else would reverse on heating and return on cooling. And it sets the limit of what reheating can do for a product that has to survive several days.

Which means shelf life is two problems

Everything above has nothing to do with mould.

Microbial shelf life and eating quality shelf life are separate problems, with separate mechanisms and separate answers, and confusing them is the most common mistake in a bakery specification.

A mould inhibitor, however good, extends the date that can legally and safely be printed on the pack. It does not touch the crumb. If the product is rejected on texture on day four and the pack says day eight, the inhibitor bought four days of waste rather than four days of shelf life.

The reverse is equally true. An anti staling approach that leaves the product microbiologically exposed has extended nothing, because the date has to be set by the earlier of the two limits.

The useful specification names both, separately, with a target for each.

What actually slows it

Three routes are used industrially and they work in different places.

Enzymes, principally amylases, shorten the outer chains of amylopectin so that the chains have less to recrystallise with. This is the most direct attack on the mechanism itself rather than on its consequences.

Hydrocolloids and fibres hold water in the crumb and raise the viscosity of the aqueous phase, which slows the migration and the mobility that recrystallisation depends on. They also change how the crumb feels immediately, which matters because softness on day one is part of what is being defended.

Emulsifiers, and specifically those that complex with amylose, change how the starch sets in the first place rather than how it ages afterwards.

None of these is a single answer, and they are not interchangeable. An enzyme approach and a hydrocolloid approach can produce the same firmness reading on day five and completely different products to eat.

What this changes about a specification

Set two dates, not one. A microbiological limit and a texture limit, measured separately. If they are far apart, one of them is being paid for and not used.

Measure firmness across the whole shelf life, not at the end. The curve matters more than the endpoint, because the customer meets the product somewhere in the middle.

Ask where the product will be stored. A specification that assumes ambient and a chain that runs chilled are describing different rates of the same process.

Expect the two halves to interact. Moisture retention and water activity both matter. A softer crumb does not by itself establish microbial stability, so texture and microbiological shelf life need to be validated together.

If you are working on bread shelf life and want the texture half of it treated as seriously as the microbiological half, tell us what you are making.