By the 1860s chemists knew about sixty elements, and the list looked like a jumble. Some were metals, some gases, some violently reactive, some nearly inert. Several people noticed patterns. Johann Döbereiner pointed out families of three similar elements in 1829, and John Newlands suggested in 1865 that properties repeated every eighth element, an idea his colleagues mocked. Nobody had yet produced an arrangement that could guide new discoveries.

Mendeleev's gamble

In 1869 Dmitri Mendeleev, a chemistry professor in St Petersburg writing a textbook, arranged the known elements in order of their atomic weights and grouped them by their chemical behavior. Others had tried that much. His boldest move was leaving gaps for elements not yet discovered and predicting their properties. Where the pattern called for an element no one had found, he left an empty box and described what would fill it. He also moved a few elements out of weight order when their behavior plainly belonged elsewhere, trusting chemistry over the scales.

He was not alone. Julius Lothar Meyer in Germany reached a similar table independently, publishing his version in 1870. Mendeleev's fame rests mostly on his predictions. Three of his missing elements were found within twenty years: gallium in 1875, scandium in 1879 and his "eka-silicon," which was found in 1886 and named germanium. Fairness requires the other half of the story. Some of his predictions failed, including elements lighter than hydrogen that do not exist, and he was slow to accept the noble gases when they turned up in the 1890s. His table was a powerful tool, not an oracle.

Why the pattern works

Mendeleev did not know why his table worked. The answer came from inside the atom. In 1913 the young British physicist Henry Moseley used X-rays to measure the charge of each nucleus and showed that the true order of the table is by atomic number, the count of protons, not by atomic weight. This fixed the few spots where Mendeleev had swapped elements by instinct. Moseley was killed at Gallipoli in 1915, aged 27.

Electrons are arranged in shells around the nucleus, and the outermost shell does most of the chemistry. The table's rows are called periods and its columns are called groups. Elements in the same group have the same number of outer electrons, which is why they behave alike. Moving along a row, each element has one more proton and one more electron than the last. Moving down a column, atoms tend to get larger, because each new row adds another shell of electrons.

Reading the map

The table works like a map. Metals fill the left side and the center, and most elements are metals: shiny, good conductors, easy to shape. Nonmetals sit at the upper right. Along the staircase between them lie the metalloids, such as silicon, with properties in between, which makes them useful for electronics.

The columns have family characters. Group 1, the alkali metals, are soft metals that react with water, and going down the group they react more and more violently: lithium fizzes, sodium skates about and melts, potassium bursts into lilac flame. Group 17, the halogens, including fluorine and chlorine, are reactive nonmetals that grab electrons. At the far right, group 18 holds the noble gases. Noble gases have full outer shells, so they rarely react with anything. When Lord Rayleigh and William Ramsay discovered argon in 1894, it did not fit any existing column, and an entire new group had to be added. The table absorbed the shock and became stronger for it.

Worked example: testing eka-silicon

In 1871 Mendeleev described the element that should sit below silicon. He predicted an atomic weight of about 72, a density of about 5.5 grams per cubic centimeter, a grey metallic look and a high melting point. He expected it to form an oxide and a compound with chlorine that would be a liquid boiling below 100 °C.

In 1886 Clemens Winkler, analyzing a silver ore from a mine in Saxony, found a new element and named it germanium after his country. Its atomic weight came out at about 72.6 and its density at about 5.35. It was greyish white, and its chloride was a liquid boiling at about 86 °C. The match was close enough that chemists who had treated the table as a filing system began to treat it as a law of nature.

The lesson goes beyond chemistry. Mendeleev could have published a tidy table that only sorted what was already known. Instead he wrote down specific numbers that could have proved him wrong. When an idea risks failure and survives, people have much better reason to trust it.