Baking Soda and Baking Powder Are Not Versions of Each Other, and One of Them Needs an Acid

They sit next to each other on the shelf in nearly identical boxes, both white powders, both there to make things rise, and the recipe often calls for both at once without ever saying why.

They are not two brands of the same idea. One is a single chemical that does nothing on its own; the other is a small kit with everything already in it. Swapping one for the other produces a flat, soapy, oddly brown result, and once the difference is clear that stops being a mystery and starts being arithmetic.

Baking soda is one ingredient and it is only half a reaction

Sodium bicarbonate is alkaline. Put it in the presence of an acid and some moisture and it reacts immediately, producing carbon dioxide, which is the lift.

No acid, no reaction. What you get instead is undissolved alkali sitting in the crumb, which tastes soapy and faintly metallic and is the flavor everybody recognizes and cannot name.

So a recipe calling for baking soda must contain something acidic, and it always does if you look: buttermilk, yogurt, sour cream, molasses, brown sugar, honey, citrus juice, vinegar, coffee, or natural cocoa powder. The acidic dairies are the most common of these and the most commonly substituted away by someone using regular milk, which breaks the recipe in a way that looks like bad luck.

Cocoa is the trap inside the trap. Natural cocoa is acidic and will react with soda. Dutch-process cocoa has been treated with alkali and will not, so a recipe written for one and made with the other loses its lift entirely.

Which is worth settling before the butter is creamed rather than after the cake comes out flat. The tin says which one it is, somewhere small on the back, and reading it takes less time than greasing the pan.

Baking powder is baking soda that brought its own acid

It is sodium bicarbonate plus one or two dry acids plus a little cornstarch to keep them apart and dry in the tin.

Which means it works in anything, acidic or not, and it is what a recipe uses when there is nothing sour in the bowl. It is also weaker per spoonful, because most of what is in the tin is the acid and the starch rather than the active ingredient — roughly three to four times as much baking powder is needed to do what a given amount of soda does.

Almost all of it sold now is double-acting: a small amount of gas is released when it gets wet, and the larger share only when it gets hot. That second stage is what gives a batter some tolerance for sitting on the counter while the oven finishes preheating.

Which is why recipes call for both

A batter with buttermilk in it has more acid than a sensible amount of soda can neutralize. The soda is there to react with that acid and to balance the pH; the powder is there to supply the rest of the lift independently.

Put another way: the soda is doing chemistry on the batter, and the powder is doing the leavening. Seeing both on an ingredient list is not redundancy, it is a recipe that was tested.

It is also why I only had one of them, so I used more of it ruins a buttermilk cake twice over. Double the powder and the acid in the buttermilk still has nothing to meet: the crumb comes out tangy and stubbornly pale. Double the soda and there is more alkali than the batter can use, and it goes soapy and dark instead. Two different bad cakes from one act of confidence.

Soda is also a browning agent, and that is half of why it is there

Raising the pH speeds up the browning reaction between proteins and sugars considerably. A cookie with soda in it is darker and more deeply flavored than the same cookie with powder, at the same bake time.

This is why soda appears in recipes that are not obviously in need of lift, and it is the same trick used outside baking. A pinch in the pan takes onions to mahogany in a fraction of the usual time. A little in the water softens dried beans and chickpeas faster, though it is age and acid rather than salt that made them stubborn in the first place. A small amount on sliced beef before a stir fry keeps the proteins from bonding tightly and reads as tenderness.

In every one of those cases the dose is tiny and the margin is narrow. Past a certain point the food tastes of soap and takes on a faintly slippery texture, and there is no correcting it afterward.

Too much of either has its own signature

Too much soda: soapy, metallic, a coarse open crumb, excessive browning, and sometimes an odd color — cocoa goes reddish, blueberries go green — because the pigments are pH sensitive.

Too much powder: a bitter edge from the acid salts, and a batter that rises fast and then collapses because the structure never set while the gas was still there. The crater in the middle of a cake is often exactly this.

Too little of either: dense, pale, heavy, and usually blamed on overmixing when the tin was the problem.

Timing is different for the two of them

A soda-leavened batter starts working the moment the wet and dry meet, and a good share of the gas is gone within ten or fifteen minutes. It goes in the oven promptly. Mixing a pancake batter with soda in it an hour ahead is throwing the lift away.

A double-acting powder batter can wait, because the main event is heat-triggered. This is why some pancake and waffle recipes tolerate resting and others do not, and it is entirely down to which tin was used.

Resting has its own benefit either way — the flour finishes hydrating and the structure relaxes — so the ideal is a powder-leavened batter rested, or a soda-leavened one mixed at the last minute.

Both go stale, and the test takes ten seconds

Baking powder loses strength on the shelf, faster once opened and faster in a humid kitchen. Drop half a teaspoon into hot water: it should fizz enthusiastically and immediately. A lazy fizz means a flat cake.

Soda lasts far longer but absorbs odors and moisture. Test it with a spoonful of vinegar; it should erupt.

Neither test costs anything, and both are worth doing before a recipe rather than after it disappoints.

And a tin of baking powder costs a couple of dollars. The butter, eggs and flour it is about to flatten cost several times that, and the oven hour is not refundable at all. When the fizz looks half-hearted, buy the new tin and stop thinking about it.

Substituting, with the arithmetic

Powder from soda: a teaspoon of baking powder is approximately a quarter teaspoon of baking soda plus half a teaspoon of cream of tartar. This works and it is single-acting, so it goes in the oven immediately.

Soda from powder: it takes three or four times as much powder, which usually brings a bitter taste and a lot of extra starch. Possible, rarely good.

The better substitution is to change the acid rather than the leavener. A recipe calling for buttermilk and soda works with regular milk soured with a spoonful of lemon juice or vinegar, because what it actually needed was an acid in the right place, not that specific carton.

The quantities are more standardized than they look

Roughly a teaspoon of baking powder per cup of flour, or a quarter teaspoon of soda per cup of flour where there is enough acid to meet it.

Which is a useful sanity check on any recipe you are scaling, adapting or reading skeptically. Far more than that and the thing will taste of its leavening; far less and it will be a brick. Sugar, fat and flour all move around a good deal between recipes; the leavening barely does, and a number well outside that range is usually a typo rather than a technique.

Part of Recipes & Technique.