Week 12: Heat from one side

The last lecture — conduction instead of an oven, the pita pocket as pure physics, nixtamalised maize, and an audit of what has been holding up your bread all semester

10 May 2027

Open the slides

Bread is older than ovens by several thousand years. The oldest charred bread crumbs anyone has found — Shubayqa 1, in north-eastern Jordan, around 14,400 years old — are flatbread, cooked on stone, and predate agriculture itself. Every apparatus this course has assumed since week 1 is a later invention. This hour is about what happens when you take it away.

Four breads, and the bench work for all of them is on the week 12 session page. What follows is why a disc of dough on hot metal behaves as it does, and — since this is the last hour — what has actually been holding your bread up for twelve weeks.

An oven is a box of hot air; a surface is not#

In an oven a loaf is heated from all directions at once, mostly by radiation from the walls and by convection in air that conducts heat poorly. That is why a 900 g tin loaf takes thirty-five minutes at 230 °C: the heat has to get to the centre through the dough itself, and dough is a decent insulator.

A flatbread is pressed against solid metal or clay. Conduction moves heat into it one or two orders of magnitude faster than hot air does, so the surface can run much hotter and the bake is over before the bread has time to dry out. The numbers are the argument.

Two consequences follow. Moisture loss is small, which is why a flatbread is soft and a crusty loaf is not — the loaf spends half an hour evaporating water off its surface and building a crust out of what is left. And the dough must be thin, because conduction into one face only reaches so far before that face burns.

The pocket is physics, not recipe#

A pita has nothing special in it. Flour, water, salt, a little yeast, a little oil — the week 1 formula rolled flat. The pocket comes from the geometry and the heat.

Lay a 3 mm disc on a 250 °C surface and the outer millimetre of each face passes 100 °C almost immediately. Its starch gelatinises and its gluten sets, and you now have two thin, cooked, gas-tight skins with cool wet dough between them. A few seconds later that middle reaches boiling. Liquid water becoming steam at atmospheric pressure expands about 1600-fold, and here the expansion is happening faster than the steam can diffuse out through set skins. Pressure builds, the skins are still hot and extensible, and they separate. The bread inflates like a bag.

Yeast has almost nothing to do with it. It contributes flavour and a slightly more extensible dough, which is why a good pita is fermented at all, but the lift is steam. The roti proves the point: no leavening of any kind, and it balloons over a flame in four seconds.

So the pocket depends on two variables only. The disc must be of even thickness, because steam will always vent through the weakest point, and a patch rolled thin sets early and holds while a thick patch stays soft and lets the pressure out sideways. And the surface must be genuinely hot. Below about 220 °C the skins take long enough to form that the middle is drying out while they do, and there is no longer enough free water to flash. That bread browns, stays flat and eats like a cracker. Both failures are diagnosable from the finished bread, which is a useful property in an assessment.

Naan and the yoghurt#

Naan is the enriched member of the family and belongs with week 9 rather than week 1. Fat — ghee, oil, sometimes milk — coats the gluten strands during mixing and interrupts the network, so the bread tears short and tender instead of pulling into chewy strings.

The yoghurt is doing three separate things. It is about 85% water, so it is part of the hydration. Its fat and milk solids enrich the dough and, because lactose and milk protein are both Maillard substrates, it browns faster and darker than a lean dough at the same temperature. And its lactic acid drops the dough pH to somewhere near 4.5, which weakens the gluten network slightly and gives a softer bite — the same acid-slackens-gluten effect that goes too far in an over-fermented sourdough is here being used on purpose, in a dough that will be in the oven inside two hours.

The tandoor supplies the rest. A clay cylinder over charcoal sits at 400 to 480 °C; the naan is slapped wet-side-down onto the wall and cooks by conduction from the clay, radiation from the walls opposite and convection up the chimney, all at once. Sixty to ninety seconds, blistered and charred in spots, and the char is not a defect but most of the flavour.

Masa: a bread with no gluten and no rye either#

Maize has no gluten-forming proteins and no useful pentosans. A corn tortilla should not be possible, and it is possible because of a process the Mesoamericans worked out at least 3,500 years ago.

Nixtamalisation: dried maize is simmered for thirty to fifty minutes in water with roughly 1% calcium hydroxide, then steeped for several hours and washed. The alkali, at around pH 12, dissolves the pericarp and the hemicellulose that binds the kernel together, partly gelatinises the starch, and lets the softened grain be ground wet into masa — a dough held together by gelatinised starch and solubilised cell-wall polysaccharides, with calcium ions cross-linking between them. No protein network anywhere.

The nutritional half is the more remarkable part. Maize niacin is bound as niacytin and is not absorbable; alkaline treatment hydrolyses that bond and frees it. The lime also adds a substantial amount of calcium and improves the balance of available amino acids. Maize travelled out of the Americas in the sixteenth century without the process attached to it, and pellagra followed it across southern Europe and later the American South. The populations that had kept the alkaline cook did not get it.

The flour tortilla is the control experiment sitting next to it: wheat, so gluten is available, plus 15 to 20% fat to keep that gluten short. It folds cold; a corn tortilla held up by starch alone cracks when you fold it cold and has to be warmed first.

What has been holding up your bread#

Twelve weeks, and the through-line was never gluten. It was the question of what is providing structure in this particular dough.

Gluten did the work in weeks 1 through 5, and most of the argument there was about hydration and how long to let the network relax. Week 6 changed the organism inflating it. Week 7 removed gluten entirely: a rye loaf stands on a pentosan gel, protected from its own amylase by acid. Week 8 added a gelatinised starch skin from the outside. Weeks 9 and 10 kept gluten but deliberately compromised it, with fat and then with sheets of butter. Week 11 dropped fermentation and used a chemical reaction that starts when it gets wet. Today it is steam and nothing else, and in the masa it is gelatinised starch with no protein involved at all.

There is no hierarchy in that list. A baker’s job is to identify which mechanism is operating in the dough in front of them, because that determines what mixing, resting and heat are supposed to achieve. Do that correctly and a formula you have never seen becomes readable.

Outline#

• Shubayqa 1 at 14,400 years: flatbread before agriculture, before ovens
• conduction into one face versus radiation and convection in a box of air
• surface temperatures of 250 to 480 °C, bakes of 60 seconds to 4 minutes
• the pita pocket: two set skins, a wet middle, and a 1600-fold expansion
• even rolling and a genuinely hot surface as the only two variables
• the three failures: half-pocket, too thick to flash, too cool to steam
• naan as enriched dough; yoghurt as water, fat, browning and acid at pH 4.5
• nixtamalisation at pH 12: cohesive masa, free niacin, and pellagra
• twelve weeks audited by mechanism — gluten, pentosans, starch, steam, chemistry