Every account of steam baking says the same thing: steam keeps the surface pliable so the loaf can expand before the crust sets. That is true as far as it goes. It is also incomplete, and the missing part is measurable.
Condensation is its own heat path
A 2020 review of heat transfer in baking ovens reports the breakdown for a commercial multi-deck oven: radiation 30 per cent, conduction 40 per cent, convection 15 per cent, and condensation 15 per cent.
Condensation appears there as a fourth category alongside the three modes of heat transfer taught in physics classes. That is not a taxonomic quirk. When steam condenses on a cool dough surface it gives up its latent heat of vaporisation, and latent heat is large. A gram of water condensing releases far more energy than a gram of air cooling by the same number of degrees.
The review provides a second configuration for comparison: a porous burner plate oven operated without added water, where radiation accounted for around 45 per cent and conduction and convection about 27.5 per cent each. No condensation term appears, because there was nothing to condense.
Most tellingly, the review reports that when water or a mist is introduced, the conduction, convection and radiation contributions all decrease, precisely because condensation has been added as an additional transfer channel. Steam does not intensify the other three paths. It substitutes for part of them with a more efficient one.
That reframes what a home baker is doing when they throw water into a hot oven. They are not just humidifying the cavity. They are switching on a heat delivery mechanism that a dry oven does not have.
What else is in the heat budget
The same review gives figures that put domestic ovens in perspective. Oven walls run 200 to 400 degrees Celsius with emissivity of 0.2 to 0.3. A burner plate surface can reach around 900 degrees, and in some configurations radiation accounts for 63 to 89 per cent of the total.
Low wall emissivity matters for anyone assuming that a hot oven radiates strongly. At 0.2 to 0.3, oven walls are poor radiators for their temperature. This is one reason a heavy preheated surface underneath the bread does so much: conduction from a solid at 40 per cent of the total is doing more work than the walls are.
The window in which steam can matter
The other half of the question is timing, and a 2025 study of temperature evolution during baking answers it.
That study tracked loaf dimensions during the bake. Lateral height stabilised at around 30 millimetres after 450 seconds. Central height continued rising until roughly 900 seconds, settling near 51 millimetres. In other words, expansion is essentially complete within about seven and a half to fifteen minutes depending on which dimension you measure, with the sides fixing first.
Steam can only assist expansion while expansion is happening. Introducing steam after the sides have set does nothing for volume, whatever it does for surface appearance. This is the quantitative basis for the standard instruction to steam at the start and vent partway through: the useful window has a measurable end.
What sets the crumb, and when
The same study identifies the temperatures at which the structure fixes, and they explain why the window closes when it does.
Dough viscosity turns from fluid to solid at around 50 degrees Celsius, with the range reported across the literature spanning 60 to 87. Starch gelatinisation showed a calorimetric peak at 76.0 degrees. Pores opened and became connected between 71 and 81 degrees. Full setting occurred at around 110 degrees. Meanwhile the interior temperature rose to a 100 degree plateau and stayed there.
Reading those in sequence gives the mechanism. The dough starts fluid and expandable. Between roughly 71 and 81 degrees the gas cells connect into an open network, which is the crumb structure being decided. By 110 degrees the structure is fixed. Everything a baker can influence about volume happens before the surface layers pass through that range.
Steam works on the surface, which is the region heating fastest and therefore passing through those temperatures first. Delaying surface setting is exactly the mechanism usually described, and the temperatures above are the ones being delayed.
What is not publicly documented
Here the literature runs out sooner than expected.
The obvious experiment, varying the amount of steam and measuring loaf volume, crust thickness and crust colour in the standard colour coordinates, has been done. A 2011 study in the Journal of Food Engineering by Schirmer and colleagues addresses the influence of the amount of steaming on heating kinetics and selected quality attributes of bread, and related work exists on crust temperature, water content and crust formation.
None of those is openly accessible. I could confirm that they exist, with their journals, volumes and page ranges, but not read their numbers. So any figure you encounter for how much a given quantity of steam changes loaf volume or crust thickness is not traceable to a source I can verify.
That gap is worth knowing about, because it means the practical guidance on steam quantity in circulation rests on craft consensus rather than on published measurement that a reader can check.
What follows for a home oven
The defensible conclusions:
Steam adds a heat path rather than merely adding humidity, and in a commercial oven designed for it that path carries about 15 per cent of the total. A domestic improvisation will not match that, but it is working on the same mechanism.
The window is the first several minutes. Measured lateral setting at 450 seconds and central setting at 900 gives the outer bound; steam introduced late is cosmetic.
Conduction is the largest single term in a deck oven at 40 per cent, and oven walls radiate poorly at emissivity 0.2 to 0.3. A preheated mass under the loaf plus steam at the start addresses the two largest levers available in a domestic oven.
And the amount of steam is the part nobody can currently justify with a public number. Enough to keep the surface from setting early, vented before the crust needs to dry, is as precise as the accessible evidence allows.
