Guides

Why density tables are anchored at 20°C — and when it's worth correcting for temperature

A hydrometer reading and a reference table only agree if they were both taken at the same temperature. Here's how big that gap actually gets, and a way to decide fast whether it's worth chasing.

By D. Shin · Published September 3, 2026

Every density table on this site — NaOH, hydrochloric acid, sulfuric acid — carries the same quiet footnote: values at 20°C. It's easy to skim past that until the day a hydrometer reading doesn't match the paperwork. A tote that sat on a delivery truck through a July afternoon, a sample pulled off a heat-traced line, a graduated cylinder in a warm mixing room — any of these hands you a density that's correct for the temperature it was actually taken at, and slightly wrong for the 20°C table you're about to read it against. This guide covers how far off that gets you, which direction it goes, and — the part that actually saves time on a busy shift — when it's worth doing anything about it at all.

Why every table says 20°C

Twenty degrees Celsius isn't a law of physics; it's a convention, chosen because it happens to be a good one. It sits close to typical indoor and lab ambient temperature across most of the world, so a routine sample often needs little or no correction just by virtue of where it was measured. It's also the temperature most volumetric glassware — Class A flasks, burettes, pipettes — is manufactured and certified to deliver or contain its stated volume at, so a density figure and the glassware used to prepare the solution are already speaking the same language. And because it's close to universal across the aqueous-density literature, a number read off this site's tables lines up directly against a supplier spec sheet, an older printed Baumé chart, or a value in the CRC Handbook, with no unit or temperature conversion in between.

It isn't the only convention in use, though, and mixing them up is its own quiet source of error. Petroleum measurement standardises on 60°F (15.56°C) instead — see the volume-correction discussion in reading a tank dip chart — and a fair amount of physical-chemistry reference data (equilibrium constants, conductivities) is tabulated at 25°C. None of these is wrong; they're just different agreements. The first thing worth checking on any spec sheet or old chart is which one it's using before assuming it matches the calculators on this site.

What's actually expanding

There's no chemistry trick here — it's the same thermal expansion that makes a thermometer work. Warm a liquid and its molecules gain kinetic energy, jostle a little further apart on average, and the same mass spreads out over slightly more volume. Density is mass divided by volume, so as volume creeps up, density creeps down. It happens in pure water on its own, before any solute is even added, and it's easy to see because pure water's density is one of the most precisely measured physical constants there is:

Two things worth noticing in that table. First, the scale: even a 40°C swing only moves pure water's density by about 1.5%. This is not a correction that turns a passing sample into a failing one very often. Second, the rate isn't constant — the drop from 20°C to 30°C is about 0.25%, but the drop from 50°C to 60°C is nearly double that, around 0.48%. Water's thermal expansion coefficient itself grows as it warms, so the correction gets a bit steeper the hotter you go, not a flat percentage per degree all the way up.

Dissolve something in that water and the exact coefficient shifts — a concentrated acid or caustic solution doesn't expand at precisely water's rate, because the dissolved ions change how tightly everything packs together, the same packing effect that makes the density-vs-concentration curve on the NaOH, HCl and sulfuric acid calculators curve instead of running straight. But the shift in coefficient is generally modest, not a different order of magnitude — real measurements on concentrated aqueous acids and bases land in the same neighbourhood as plain water, on the order of a few tenths of a percent of density per 10°C in the room-temperature-to-hot-process range. Close enough to water's own numbers above to use them as a working estimate when you don't have a temperature-specific table for your exact solution in hand — which, for most shop and lab reference tables, you won't.

What that error actually costs you

Take the tote example from the NaOH concentration calculator: a hydrometer reading of 1.30 g/cm³ resolves to 27.41% NaOH against the 20°C table. Now suppose that reading wasn't taken on a room-temperature sample, but straight out of a tote that had spent the afternoon on a truck at close to 35°C. Using water's coefficient as a stand-in — roughly 0.03% of density per °C averaged across that 15°C span — the sample's actual density at 35°C would run about 0.45% below what the same solution reads at 20°C: close to 1.294 g/cm³ instead of 1.30. Treat that warm reading as if it were already at 20°C and interpolate it against the table, and you get back roughly 26.9% — about half a percentage point light, on paper, for a tote that's genuinely 27.41%.

Half a point on a caustic tote rarely changes a decision. But the same mechanism, applied to a different solute, doesn't stay small. Ethanol's thermal expansion coefficient runs at roughly four times water's, which is precisely why a spirit hydrometer ships with its own correction table clipped to the case — the same 15°C offset that cost the NaOH example half a point can move an uncorrected ABV reading by two or three points, enough to fail a label spec outright. See why alcohol plus water is less than the sum for that solute's numbers specifically, and run one through the ethanol-water dilution calculator to see how far off a warm or cold reading actually lands. The direction is symmetric either way: a warm sample under-reads concentration against a 20°C table, a cold sample over-reads it — worth remembering when a reading comes back suspiciously off and the sample's temperature hasn't been checked yet.

When the correction is worth doing

It matters

• Custody transfer. Anywhere a volume gets multiplied by a density to invoice a mass — a tanker of acid metered by volume, a delivery reconciled against a bill of lading — a few tenths of a percent is real money at scale, and it's exactly the gap that shows up as an "unexplained" shortfall between two honest readings taken hours apart.
• Lab prep and titration standards. A molarity solution made up in 20°C-certified glassware, using a density figure read at a different temperature, carries that mismatch into every downstream calculation using it as a standard.
• Anything with a tight spec band. If the tolerance on a QC check is a few tenths of a percent, thermal drift alone can be the whole margin.

It doesn't

• Routine dosing and shop-floor checks. A brewery CIP wash targeting 2% w/w caustic, dosed by weight on load cells, doesn't care whether the density used to convert a target weight into a tank level is nominally off by a few tenths of a percent — the wash works the same either way. See the CIP example on the NaOH calculator page for the full case.
• A field check that a tote is roughly what the label says. Confirming an unlabeled drum is "somewhere around 25% and not 50%" doesn't need temperature correction; a titration or a certified assay does, but that's a different measurement entirely, covered in dilution math that survives the shop floor.
• General-purpose dilution work through the solution dilution calculator, where the target is usually a band, not a certified figure to four significant places.

The two ways to actually correct for it

The simple fix, and the one that covers most situations: let the sample reach the table's reference temperature before you read it. For a small lab sample, that's fifteen or twenty minutes on the bench or in a room-temperature water bath; for a full tote, it can mean waiting until morning rather than dipping a hydrometer into it straight off the truck. It costs time and nothing else, and it sidesteps the whole calculation — a 20°C sample read against a 20°C table needs no correction at all. If a reading absolutely must be taken warm or cold, at minimum log the actual sample temperature next to the number, so anyone reading the paperwork later knows it wasn't a clean 20°C figure.

The rigorous fix, for when you can't wait: apply a volume correction factor (VCF) — a multiplier, looked up from a published table by product density and observed temperature, that converts a reading taken at whatever temperature it happened to be measured at back to the reference temperature. This is the same tool covered in more depth in reading a tank dip chart, where a warming tank gains apparent volume through the same mechanism this guide has been describing for density — the petroleum industry's API MPMS Chapter 11.1 tables are the standard reference there, and equivalent temperature-correction tables exist for other industrial liquids. Like the water table above, a VCF isn't one flat percentage-per-degree; it's read off the table at the specific product density and temperature in question, because the correction itself changes shape as both of those change. If you're already gauging tank volume rather than a bench sample, the tank volume calculator and strapping chart calculator are the tools that reading eventually feeds into.

A rule of thumb worth carrying around

Twenty degrees Celsius unless the spec sheet says otherwise — and it's worth a five-second check of which convention a given chart is actually using before comparing it to this site's tables. Expect on the order of a few tenths of a percent of density change per 10°C for an aqueous solution near room temperature, growing a bit steeper the hotter it gets, and roughly double that for a solute like ethanol whose expansion runs faster than water's. If the number only has to be roughly right, let the sample settle to room temperature and stop thinking about it. If a contract, a certificate or a tight spec is riding on it, note the actual temperature and correct for it properly — the arithmetic is small, and skipping it is exactly how two honest readings a few hours apart end up looking like a discrepancy that never really existed.