The Elements

Cohort map

Each element's ten closest physical matches, drawn on the periodic table. Every element has its own list, and the lists overlap.

Open the cohort map →

Element k (other elements, 0–117; the query is always added)

Second ring

How the similarity works

Each element is described by ten physical features: melting point (K), boiling point (K), density (log10 of g/cm³), molar heat (J/(mol·K)), Pauling electronegativity, and five atomic radii in pm (empirical, calculated, covalent single, van der Waals, metallic). Missing values are filled with the population median (synthetic, not observations); each feature is z-scored and each element's vector scaled to unit length, and similarity is the cosine between vectors. Scores are computed once from the full-precision matrix and shown to six decimals.

Five of the ten features are correlated radius measures, so atomic size can carry extra weight, although each feature's actual contribution varies by element. Median filling makes sparsely measured elements look more alike than their observed data show; check the coverage count. Values for superheavy elements are often predictions. A high score is resemblance in these numbers, not chemical interchangeability. Negative scores mean opposite sides of the population average in this standardised space, not chemical opposition.

Excluded from the vector: atomic number and mass, table position, period, group, block, category, phase, the text summary, and electron affinity and ionisation energies (units not documented upstream).