In March 1869, Dmitri Mendeleev presented to the Russian Chemical Society a way of laying out around sixty elements then known. He ordered them by atomic weight and broke the line into rows so that elements with similar chemical behaviour fell into the same columns. Where the pattern called for an element that no one had found, he did not close the line up. He left the space empty and described what should eventually fill it.

The arrangement is now so familiar that the gaps are easy to overlook. They are the part worth dwelling on. Other people had already noticed that the elements repeat their properties at regular intervals. What Mendeleev did with the repetition was different in degree: he trusted the pattern enough to bet against the existing inventory, and to say, in print, that the inventory was incomplete in specific places by specific amounts.

What the table was built on

The organising idea was periodicity. Run through the elements in order of increasing atomic weight and their properties do not change steadily; they cycle. A reactive light metal is followed some steps later by another reactive light metal, a pungent gas by another pungent gas. Mendeleev arranged the rows so that these recurrences stacked into vertical groups, lithium above sodium above potassium, and so on.

Atomic weight was the only quantity he had to order by. He did not know what caused the periodicity, because the structure of the atom was still decades away. He was working from measured weights and observed chemistry, and arranging both until the columns made chemical sense. In a few cases the weights and the chemistry disagreed, and he chose the chemistry, placing an element where its behaviour fit even when its measured weight argued for somewhere else.

The gaps and the predictions

For three of the empty spaces, Mendeleev went further than leaving a blank. He gave the missing elements provisional names built from the Sanskrit prefix eka, meaning one, so that eka-aluminium meant the element one place below aluminium. For each he estimated an atomic weight, a density, and a set of chemical properties, reading them off the neighbours that surrounded the hole.

The estimates were close enough to be checked. Eka-aluminium arrived in 1875, when the French chemist Paul-Émile Lecoq de Boisbaudran isolated a new metal he called gallium; its density, once measured carefully, matched Mendeleev’s figure. Eka-boron arrived in 1879 as scandium, identified by Lars Fredrik Nilson in Sweden. Eka-silicon arrived in 1886 as germanium, found by Clemens Winkler in Germany, and its properties tracked the prediction closely enough that the agreement is still the standard example taught from the episode.

Three confirmed predictions within seventeen years did more for the table than any argument could. A scheme that merely tidied the known elements would have been one convenience among several. A scheme that named absent elements and described them before anyone had seen them was harder to dismiss.

He was not the only one, and that matters

Mendeleev did not invent the idea that the elements form a periodic series. Alexandre-Émile Béguyer de Chancourtois had plotted them on a spiral in 1862. John Newlands, in London, had published his “law of octaves” in the mid-1860s, noting that every eighth element resembled the first, and had been mocked for it. The German chemist Lothar Meyer was developing his own table at almost the same time as Mendeleev and arguably had the periodicity of physical properties drawn more cleanly.

What separated Mendeleev’s version was not priority but nerve. Newlands had let the known elements set the length of his pattern. Mendeleev let the pattern set the length, and treated mismatches as evidence that an element was missing or that a measured weight was wrong, rather than as evidence against the scheme. The historian Eric Scerri, who has written the standard modern account of the table’s development, has argued that this willingness to subordinate the data to the system, and then to stake testable predictions on it, is the better explanation for why Mendeleev’s name attached to the result.

Where the 1869 table was wrong

The story is usually told as a clean run of successful forecasts. It was not clean. Mendeleev predicted elements that were never found, including light elements he expected to sit before hydrogen, tied later to his mistaken belief in a luminiferous ether. Several of his other interpolations did not pan out as described.

Two larger problems sat outside the 1869 table entirely. The noble gases, argon and its relatives, were discovered in the 1890s and formed a whole column that the original scheme had not anticipated; Mendeleev was at first reluctant to accept them. And the ordering principle itself was incomplete. A handful of pairs, tellurium and iodine among them, sat in the right chemical place but the wrong weight order. The reason became clear only in 1913, after Mendeleev’s death, when Henry Moseley showed that the elements are properly ordered not by weight but by nuclear charge, the atomic number. The table Mendeleev built by chemical intuition turned out to be tracking a physical quantity he had no way to measure.

That is the honest shape of it. The gaps were the achievement, and the parts he filled wrongly belong in the same account as the parts he filled before anyone else could.

The part worth keeping

The detail in the headline is fair: by 1869 the chemistry of the elements had become ordered enough that a missing piece could be described before it was found. That is a particular kind of confidence, and it is rarer than the finished table makes it look. It requires a pattern reliable enough to extrapolate from, and someone prepared to be checked against an empty square.

Mendeleev was prepared to be checked, and within his lifetime three of the squares filled in close to where he had said they would. The ones that did not fill in, and the deeper rule that only emerged after he was gone, are part of why the table is interesting rather than merely useful. It was right enough to predict, and wrong in ways that pointed at the physics underneath.