Every property value is an answer to a longer question

The same property, from six sources, in one table

PubChem aggregates agency records for isopropanol. Its boiling-point entries read: 82.3 °C at 760 mm Hg from the Hazardous Substances Data Bank; 180.5 °F at 760 mmHg from CAMEO Chemicals citing NTP 1992; 83 °C from the ILO-WHO International Chemical Safety Card, with no pressure stated; 82.0–82.4 °C from the Joint FAO/WHO Expert Committee on Food Additives; 181 °F from OSHA and from NIOSH; and 82.3 °C at 760 mm Hg from the Department of Energy PAC database.

The density entries spread the same way and add a second trap: 0.785 at 68 °F, 0.78509 at 25 °C, 0.785 at 20 °C, 0.7809 at 25 °C — and, from the ICSC, “Relative density (water = 1): 0.79.” That last one is not a density. It is a dimensionless ratio, which is what a specific gravity is, and a ratio needs both temperatures to be meaningful. The same record supplies the reason it matters: a coefficient of expansion of 0.00107 per °C and a density correction of 0.00082 per °C. A hydrometer reading taken in a warm receiving bay and compared against a specification at 20 °C is not comparing like with like.

Even the reference value is a distribution

The NIST Chemistry WebBook gives isopropanol a normal boiling temperature of 355.5 ± 0.4 K, and states how that figure was produced: “Average of 102 out of 118 values.” Sixteen published values were excluded. The critical temperature is given as 509 ± 2 K from an average of 19 out of 20 values, and two separate fusion temperatures are listed, 185.75 K and 185.35 K, from different sources with an uncertainty assigned by NIST’s Thermodynamics Research Center.

This is what a well-curated property value looks like: a number, a unit, an uncertainty, a method label and a count. It is not a defect that the sources disagree; it is a defect to quote one of them stripped of all four. When a specification, a data sheet or a generated product page states a property as a bare number, the missing items are the ones that decide whether it can be compared against anything.

Write the property into a record that can be checked

NIST describes metrological traceability as a property of a measurement result: the result can be related to a reference through a documented unbroken chain of calibrations, each contributing to the measurement uncertainty, with a clearly defined property, a full description of the measurement system, a stated value with a documented uncertainty evaluation, and a complete specification of the reference at the time of comparison. Compiled literature values are not measurements of your lot, and a lot result is not a compiled literature value.

Our proposed field set for any numeric property is the same every time: the quantity and its units, the temperature and pressure or the reference-temperature pair, the concentration basis where the material is a mixture, the test method with its revision, the classification as typical, guaranteed or measured, and the document and revision it came from. Where an item is missing, record it as missing. Six fields is not bureaucracy; it is the difference between a value you can defend to a customer and a value you copied.

• Is this a specific gravity or a density, and at which two temperatures?
• At what pressure was a boiling point determined, and by which method?
• Is the figure typical, a guaranteed limit, or a measured lot result?
• Which document and revision does it come from, and when was that checked?

Requirement → evidence → decision boundary

An editorial checklist for your review—not a table of product specifications.

Revision record

1.0 · September 12, 2026 — Source-based editorial edition. No original testing or product qualification is claimed.

For commercial aerospace materials and purchasing reviews. No flight qualification, oxygen-service approval, propellant formulation or cleaning procedure is provided.