Published figures we checked and rejected
Every figure in the ingredient database comes from a primary document: a manufacturer’s data sheet, a standard, a paper. Reading those documents turned up numbers that cannot be true, such as a saponification value no triglyceride can have, a refractive index no oil has, and columns printed against the wrong labels. A table copied from somewhere else would have shipped every one of them.
What was printed, and why it was not used
| Oil | Source | What it printed | Why it was not used |
|---|---|---|---|
| Andiroba Oil | Araujo-Lima et al. 2018 | Saponification value 59.5, iodine value 196.6 | The two rows sit against swapped labels. 59.5 is impossible for a C16/C18 triglyceride and would under-dose the lye about threefold. |
| Sacha Inchi Oil | Biomolecules 2023 | Saponification value 123.1, iodine value 65.2 | The fatty-acid profile in the same paper implies about 192 and 200. The profile is sound and agrees with the source used; the physicochemical row is not. |
| Coffee Seed Oil | Dong et al. 2021 | Saponification value 88.49–230.04 | Four extractions of the same beans. No oil has that spread. |
| Prickly Pear Seed Oil | Thamer and Alshargabi 2026 | Saponification value 160.2, iodine value 101.35, refractive index 1.132 | No vegetable oil has a refractive index of 1.132. The two values set only the minima; the Moroccan measurements are the typical figures. |
| Watermelon Seed Oil | Han et al. 2024 | Iodine value 63.65 for an oil that is 73.7 % linoleic acid | Arithmetically impossible: that composition implies about 136. Cited only for the disagreement. |
| Camelina Oil | Foods 2024 | “Saponification number” 28–59 | Impossible for a triglyceride. |
| Duck Fat | Kowalska et al. 2020 | Palmitic acid 46–47 % | Nearly double every other published figure. Rejected as an outlier, and cited so it stays traceable. |
| Dika Butter | Matos et al. 2009 | Saponification value 189.9–277.7 | Eight extracts of two seed lots. The pressed oils’ own compositions imply about 242, and no fat spans that range. |
| PlantFAdb, 1984 entry | Saponification value 243.0 and iodine value 8.7, beside lauric 5.7, palmitic 33.9 and oleic 24.5 % | The two values fit dika fat; the fatty-acid row describes no lauric fat. Cited for the values only. | |
| Green Tea Seed Oil (Camellia sinensis) | Sarmah and Das 2018 | Specific gravity 0.82–0.88 | No vegetable oil is that light: they run 0.91–0.92. |
| PlantFAdb, “Camellia sinensis” entries | Oleic acid 76–87 % | That is the composition of Camellia oleifera oil, and several of those datasets’ own titles concern other species. |
Tables read against the wrong labels
Two published tables had their label and value columns out of step in the PDF: Malacrida and Jorge’s Table 3 for passionfruit seed oil and Raziq et al.’s Table 5 for watermelon seed oil. Read as printed, the first makes passionfruit oil 73 % α-linolenic acid. Both were re-paired against the papers’ own saturated, monounsaturated and polyunsaturated subtotals before anything was used.
A claim that did not survive checking
Cranberry seed oil is sold on a “1:1:1 omega-3/6/9 balance”. The measured oil is 34.9 % omega-3, 38.5 % omega-6 and 20.4 % omega-9, and the producer’s own specification says the same. The record gives the measurement, not the claim.
How a figure gets checked
A saponification value and a fatty-acid profile are not independent: the chain lengths fix the average molar mass, and the molar mass fixes the value. Every oil page compares the two and says so when they disagree, and the calculator works from ranges rather than a single confident number. The formulas page sets out each equation and its source, and SAP values explained says why published tables disagree. What soapmaking books print is checked the same way: the books against the sources.
Found a figure that looks wrong, or a better primary source? Corrections with a link to the document are welcome at support@soapformulator.com.