How we calculate

The scale

PAC and POD are relative numbers, both with sucrose at 100, and the app sums them over 1000 g of mix. The part that trips most spreadsheets is the basis: our coefficients are per 100 g of the product as you buy it, not per 100 g of dry matter. Dextrose monohydrate is the proof in our own library: 190 on a dry basis times 92% solids gives the 173 we publish. A table that mixes the two bases can be off by 43% on invert sugar and 11% on honey.

• PAC. Anti-freezing power, relative to sucrose = 100, summed over 1000 g of mix. It predicts how hard the tub sets at serving temperature.
• POD. Sweetening power, same scale and same sum. Sucrose is 100, dextrose 64 per 100 g as sold, fructose 170.
• Total solids, fat, MSNF, sugars, protein, water. Percentages of the mix, from the composition of each ingredient per 100 g of the product you actually buy.
• Sandiness. A check the ranges do not make: lactose above about 7% of the mass crystallises in storage and the spoon turns gritty. The app computes it per ingredient.

The ingredient data

The library holds 257 ingredients. Each one has its own page with the composition we use and the sources behind it, and you can read the whole set as a table in PAC and POD of sugars or browse it from the ingredient guide. When a number is contested we do not average the internet: we run two independent reviews of the same question, compare what each one can show its work for, and adopt one with the reason written down. Five rules came out of that work.

Starch hydrolysates (maltodextrin, glucose syrup, glucose powder)

PAC on a dry basis = 1.90 × DE. POD on a dry basis = 0.70 × DE. Both are then multiplied by the solids fraction of the product.

Anchored on dextrose, DE 100, which every table agrees is PAC 190 and POD 70 dry. A generic gelateria maltodextrin at DE 19 and 95% solids reads PAC 34.3 and POD 12.6 per 100 g. It also restores the ladder: DE 12 must sit below DE 18, which our own library had backwards until September 2026.

Alcoholic drinks

PAC = 743 per GRAM of ethanol per 100 g of product, added to the PAC of whatever sugar the drink carries. Alcohol by volume is converted to mass first: g of ethanol per 100 g = ABV × 0.789 ÷ density.

The common shortcut multiplies a factor by the alcohol degree, which treats a volume figure as if it were mass and inflates PAC by 20 to 45%. Rum at 40% ABV carries 33.4 g of ethanol per 100 g, not 40.

Honey

POD and PAC computed from the sugar composition of each variety (fructose, glucose, maltose, sucrose), not from the shortcut that treats honey as invert sugar.

By the shortcut, honey reads sweeter than sucrose per gram. By its own composition it does not: a jar is about 18% water, so 100 g of honey reads POD 91 to 95. The difference between varieties is small, PAC 139 to 144.

Lactose in dairy

Lactose is counted inside MSNF, never in the sugars column. For strained dairy we use the measured lactose of the product instead of the usual 54.5% of MSNF.

Straining takes the whey, and the lactose with it, while keeping the protein. Greek yogurt measures about 4 g of lactose per 100 g with MSNF near 14, which is 29% and not 54.5%. Using the generic fraction overstates the sandiness risk of every Greek yogurt recipe.

Invert sugar

The library line is invert SYRUP at 70% solids: POD 87.5 and PAC 133 per 100 g, from a dry basis of 125 and 190.

Trade tables publish a dry POD of 130, but none of them shows the arithmetic. The two sources that do land at 115 to 125 and at 119.5, and the average of the two sugars in our own library gives 120. We adopted 125, where the published ranges cross, and we say so. Commercial invert PASTE at 82% solids is a different product and would be a different line.

The target ranges

A recipe is judged against the range of its base type, not against a single ideal. There are nine types, each with its own window for solids, sugars, fat, MSNF, water, protein, PAC and POD, calibrated for a display case at -11 to -14 °C. Variegato is scored differently, on PAC per 100 g of water and on the Brix of the water phase, because a ripple sauce has to stay soft at -18 °C and is never churned. Every range, with its confidence and its sources, is published in balancing ranges, CSV included, under CC BY 4.0.

The ranges describe what a professional shop can serve, so a recipe outside them is not wrong by decree: it is a recipe that will behave differently in a case. The app says which indicator is out and by how much, and leaves the decision to you.

What we do when we are wrong

The library has been corrected in public more than once. In September 2026 maltodextrin went from PAC 21 to 34.3 because seven sources put it there and none supported the old figure; the alcohol rule was replaced because it treated volume as mass; honey stopped being read as invert sugar; and the lactose of Greek yogurt moved from the sugars column into MSNF, where every other dairy line already had it. Each change was simulated against all published recipes before it was applied, and the ones that moved were republished with the new numbers.

If you find a number here that your own measurements contradict, write to us with the source. That is how most of the corrections above started.