Where the Data Comes From

This page exists so that any number on this site can be traced back to the reference that published it, and so that the handful of decisions made on top of those references are visible rather than buried.

The three sources, and what each one supplies

The full detail is in the table below, but the division of labour is simple.

Atomic weights and isotopes come from the National Institute of Standards and Technology's compilation Atomic Weights and Isotopic Compositions with Relative Atomic Masses, which is the NIST publication of the standard atomic weights set by IUPAC's Commission on Isotopic Abundances and Atomic Weights. Two weights — argon's and lead's — are taken from the commission directly instead, for the reason given further down. Every isotope listed on an element page — its mass number, its relative atomic mass, its uncertainty and its natural abundance — is from that table, unaltered, with a single exception. Carbon-12 has no uncertainty because there is none to have: it is the nuclide that fixes the unified atomic mass unit, so its mass is 12 exactly, and the reference table says so by printing zeros where a measured spread would go. Reproducing those zeros as though they were a spread would misstate the one number on this site that is true by definition, so carbon-12 appears as 12.

Physical and chemical properties come from the PubChem periodic table published by the National Center for Biotechnology Information. Melting and boiling points, density, electronegativity, atomic radius, ionisation energy, electron affinity, oxidation states, standard state and year of discovery are read from it — with three of those properties excepted, listed below, where the published determination has moved and the property table has not.

Configurations, blocks and discovery credits come from an open factual dataset of the elements. Only its numeric and categorical fields were taken; every prose field of that dataset was stripped out before it entered this site, so no third-party writing appears anywhere here.

All three are frozen snapshots, captured on one date and committed to the site's source. No page performs a lookup when you load it. That means the number you read today is the number that was checked, and it also means the snapshots go stale — they are re-captured deliberately, and the capture date is printed below rather than implied.

Decisions made on top of the sources

Six things on this site are editorial choices rather than facts read from a reference. They are listed here because a choice presented as data is the quiet kind of inaccuracy.

Elements 109 to 118 are coloured as unknown. Reference tables assign them predicted categories based on their position — element 118 as a noble gas, element 117 as a halogen. Those predictions may well be right, but no chemistry of any of these elements has been experimentally characterised, and a colour on a periodic table reads as a statement of fact. Elements 104 to 108 keep their transition-metal category, where the d-block assignment is on firmer ground.

Halogens are coloured as reactive nonmetals. "Halogen" names a group — column 17 — and the colour system on this site encodes chemical category. The halogen hub exists and is built from group membership; it simply does not get a colour of its own.

Fourteen elements use a conventional atomic weight. Hydrogen, lithium, boron, carbon, nitrogen, oxygen, magnesium, silicon, sulfur, chlorine, argon, bromine, thallium and lead have interval standard atomic weights, because their isotopic composition genuinely varies between natural sources. A calculator needs one number, so the calculators use the conventional atomic weight IUPAC publishes for that purpose. The interval is printed on the element page and the conventional value is named beside it. The step that assembles the element dataset refuses to write a file at all if an interval ever turns up without a conventional value listed against it, so that list cannot go out of date without somebody noticing.

Argon and lead follow CIAAW rather than the NIST table. These two are the one place where the sources above genuinely disagree. NIST's compilation still carries argon at 39.948(1) and lead at 207.2(1), the values that stood before the commission reassigned them: argon to [39.792, 39.963] in 2017, lead to [206.14, 207.94] in 2020. Both changes were made because the natural variation in those elements is wider than the measurement uncertainty on any single sample — the argon in air is overwhelmingly argon-40 that potassium-40 decay has poured into the atmosphere since the Earth acquired one, and lead's isotope mix depends on how much uranium and thorium happened to sit in the rock it was extracted from. IUPAC's commission is the body that sets the values NIST republishes, so where the two disagree its assignment is the one used here. The captured NIST file is left exactly as it was served; the correction sits beside it, dated, and names the string it replaces.

Three properties are taken from the paper that determined them. The property table above is the source for every physical quantity on this site except three, and the test applied to those three was not "is there a better number somewhere" but "does the tabulated figure contradict what has actually been established". The densities of osmium and iridium are the clearest case: the tabulated pair sits 0.15 g/cm³ apart with no uncertainty attached, which is around six times the separation that lattice-parameter work has since established, and a card carrying it would put the two elements' own pages in the position of arguing about a margin the card denies. Francium's first ionisation energy is tabulated as an extrapolation down group 1 that predates the spectroscopy, and it lands so close to cesium's that the one genuinely interesting thing about francium's ionisation energy disappears. Electron affinity accounts for the last two: fluorine's is quoted here from the photodetachment measurement rather than the older figure, and astatine's from the 2020 experiment at CERN that measured it for the first time, on an isotope that survives about seven hours. Every number the property table carried for astatine before that was somebody's calculation. Nothing else has been substituted, and each substitution is recorded against the exact string it replaces, so a future capture that has caught up stops the dataset from being assembled at all rather than quietly keeping a correction that is no longer a correction. Francium's electronegativity is not on that list, and the reason is on its own page.

Element names follow the American spellings used by the primary property source — aluminum, cesium, sulfur. The IUPAC and British forms are carried on each element record and are wired as permanent redirects, so /elements/caesium/ and /elements/sulphur/ both land on the element rather than on a dead end. Sixteen such redirects exist, including the Latin forms behind the symbols that do not match their English names.

What "radioactive" means here

An element is marked radioactive on this site when it has no stable isotope. That is thirty- seven elements: technetium, promethium, and everything from polonium upward.

The definition is worth stating because it has edges. Bismuth-209 is observed to alpha-decay with a half-life around 10¹⁹ years and is not marked radioactive here, because the reference tables treat it as primordially stable and it is listed with a full natural abundance. Potassium and rubidium both contain long-lived radioactive isotopes in their natural mixture and are likewise not marked, because both also have stable ones. Where that nuance matters for a particular element, its own page says so.

Separately, thirty-four elements have no isotope with a measurable natural abundance. That is not quite the same as "synthetic": astatine and francium occur in vanishingly small quantities in natural decay chains but have no measurable abundance to report.

Where the hazard flags come from

The hazard vocabulary is eleven fixed terms, assigned per element from standard safety references: GHS classifications as published in supplier safety data sheets, the International Agency for Research on Cancer's monograph classifications for the carcinogen flag, and the NIOSH Pocket Guide for the acute-toxicity and asphyxiant flags.

Two rules govern how they are used. The definition of each term is written once and is identical on every page that shows it — a safety definition reworded page to page to look less repetitive would be a worse thing than a repeated one. And the absence of a flag is not a safety claim: an element with no flags means no term in this vocabulary applies to it, not that it is harmless.

Corrections

If a number here disagrees with a source you trust, the contact page is the place to say so. Quote the element and the value and it will be checked against the reference. Corrections to data are the most useful mail this site receives, and the sources named above are the ones a correction will be checked against.