I spent two winters in St. Petersburg sorting index cards on a long oak table, each card bearing the name of an element, its atomic weight, its valency, its known compounds. The arrangement consumed me. By February of 1869 I had tried every ordering principle I could think of — alphabetical by Latin name, grouped by color of flame test, sorted by density. None of them worked. Then, almost by accident, I laid the cards out by ascending atomic weight and noticed something staggering: the properties repeated.
Prediction Through Absence
The most radical move in the entire history of classification was not what Mendeleev included — it was what he left out. Where the weights did not fit the pattern, he left gaps. He predicted three undiscovered elements in 1871, describing their densities, melting points, and oxide formulas with uncanny precision. Eka-aluminum was found in 1875 and renamed gallium. Eka-boron became scandium in 1879. Eka-silicon turned out to be germanium, isolated in 1886. Each discovery confirmed that the 18-column structure was not a filing convenience — it was a map of reality itself.
A law of nature is not a description of what we have already seen; it is a constraint on what we are permitted to discover next.
— Dmitri Mendeleev, Journal of the Russian Chemical Society, 1871
This is the quiet power of a well-designed system: it holds space for the unknown. The periodic table did not merely organize the sixty-three elements known in 1869. It encoded the grammar of atomic behavior so faithfully that physicists could later explain why the pattern existed — electron shells filling in quantized steps — without Mendeleev ever suspecting that mechanism. He had captured the syntax of matter in a grid.