Every Oil Blend Has Its Own Lye Requirement

The spread between recipes is thirteen per cent

Coconut oil saponifies at 0.178 and olive at 0.134 — a third more alkali per gram of fat. Any recipe is a weighted average of its oils, so the average value of a blend moves with the recipe and not with the total weight. Castile at 0.1340 and the three-oil blend at 0.1509 differ by 12.6 per cent, which on a kilogram batch is 16 grams of sodium hydroxide. Sixteen grams is far more than the margin any soap has.

This is why substituting one oil for another in a recipe is not a free swap. Replacing the coconut in a Bastille with an equal weight of olive drops the average value and leaves the original lye quantity in surplus, which pushes a five per cent superfat towards zero and past it. Change an oil and recompute; the calculator exists precisely so that the recomputation costs nothing and therefore actually happens.

A five per cent supplier variation eats the whole superfat

Saponification values are nominal. A published 0.134 for olive oil is an average over crops, pressings and refining routes, and a real batch can sit three to five per cent either side. Take a kilogram of Castile computed at 0.134 with five per cent superfat — 127.3 g of lye. If the oil you actually bought saponifies at 0.141, the true requirement is 141 g and your recipe is 6.6 g short: the bar comes out at 9.7 per cent superfat, oily and soft but harmless.

The error in the other direction is the one that matters. If the same oil saponifies at 0.127 instead of 0.134, the true requirement is 127 g and you have added 127.3 — your five per cent superfat has become minus zero point two per cent. There is now more alkali than fat, and the surplus stays in the bar as free lye. That is the failure mode this craft has to respect: the safe direction is to under-dose, and the reason every row's value is editable in the tool is so you can enter what your supplier's sheet says rather than what an average says.

Potassium hydroxide is not sodium hydroxide by another name

Bar soap uses sodium hydroxide and liquid soap uses potassium hydroxide, and the quantities are not interchangeable. Potassium is the heavier atom: converting a sodium requirement to a potassium one multiplies it by 56.1056 divided by 39.9971, or 1.4027. Then there is purity — potassium hydroxide flakes are typically ninety per cent, the rest being water and carbonate — so the figure is divided again by 0.90.

The two factors compound. The Bastille blend at three per cent superfat takes 142.2 g of sodium hydroxide and 221.6 g of potassium — 1.56 times as much alkali by weight for the same kilogram of oils. Anyone reading a bar recipe and reaching for potassium flakes at the printed quantity has under-dosed by more than a third, and the batch will not saponify. The reverse mistake over-doses by the same factor, which is worse.

Frequently asked questions

What superfat should I use?

Five per cent is the usual starting point for a bar and three for liquid soap, where a higher figure clouds the finished product. Above eight per cent a bar goes soft and can develop rancid spots as the free oil oxidises; below three the margin for a supplier's variation disappears. On a kilogram of Castile, moving from zero to five per cent withholds 6.7 g of lye out of 134 — the whole safety margin is smaller than a teaspoon, which is why it is worth measuring on a scale that reads to a tenth of a gram.

How much water does a batch need?

Water is the solvent and not a reagent, so it does not enter the stoichiometry at all — it only decides how the batch behaves. Three conventions are in use and the tool accepts all three: a water-to-lye ratio, most often 2:1; a percentage of the oil weight, typically 33 to 38 per cent; or a target lye-solution concentration, usually 28 to 35 per cent. They are different expressions of the same number, and a recipe that quotes one can be converted to any other.

Can I trust an online recipe without recalculating?

Recalculate it, every time, and treat a mismatch as a reason to stop. A published recipe carries the author's oils, the author's supplier and often an unstated superfat, and any of the three can move the lye by ten per cent. Running the same oil percentages through the calculator takes a minute and either confirms the published figure or tells you the recipe has been transcribed wrongly somewhere between the author and your screen — which, on a page that has been copied from blog to blog, is not rare.