Pattern-hunting that half worked

Start in 1817, and be honest about what a chemist actually had. About fifty substances that nobody could break down, arrived at over two centuries by a mixture of assay, accident and electrolysis. No reason to think that was all of them, and no reason to think it was not. Atomic weights, but contested ones. Dalton had given chemistry a way to weigh atoms relative to one another, and it worked — except that the answer depends on a formula you have to assume before you start. Is water HO or H₂O? Get that wrong and oxygen’s weight is out by a factor of two. Different laboratories used different assumptions and got different tables, and this stayed unresolved for forty years. No atom in the physical sense at all. No electron, no nucleus, no shells. An atom is a piece of stuff with a weight and a set of habits. So the question of the act is not "why is the table shaped like that". It is much more basic: is there any order here at all, or is this simply a list of the things the Earth happens to be made of? In 1817 that is a completely open question, and a serious chemist could reasonably have answered no.

Johann Wolfgang Döbereiner was professor of chemistry at Jena — a practical man whose best-known invention was a pocket lighter that ignited hydrogen on a sponge of platinum, and whose lectures Goethe used to attend. In 1817 he noticed something about three substances. Lime, strontia and baryta behave alike; that was already known. What he added was that the weight of strontia sits almost exactly halfway between the other two. He published it, and nothing happened. One coincidence about three substances is not an argument, and he knew it. Twelve years later, in 1829, he came back with a paper offering several more of them, and gave the rule a name that has stuck: triads.

Then thirty years pass, and the reason is more embarrassing than mysterious. Nobody could agree on the atomic weights. If you sort a list by a quantity that half the profession disputes, you produce an arrangement that half the profession rejects — and you cannot even tell whether a discrepancy is a fact about nature or a fact about your neighbour’s assumptions. That logjam broke in September 1860, at a congress in Karlsruhe called specifically to sort it out. It did not, quite, in the room; but a pamphlet handed out at the door as the delegates left set out one consistent system of weights so plainly that people went home and adopted it. Lothar Meyer said later that the scales fell from his eyes. And then, within six years, four independent people arranged the elements: Alexandre-Émile Béguyer de Chancourtois wound them round a cylinder in 1862, William Odling tabulated them in 1864, Newlands published his octaves in 1865, and Lothar Meyer plotted atomic volume against weight in 1864 and got a curve with peaks in it. Four at once is what usually happens when a blockage clears. The idea was not waiting on a genius. It was waiting on the numbers.

John Alexander Reina Newlands was an analytical chemist in London, mostly employed by the sugar trade. His mother was Italian, and in 1860 he went and fought with Garibaldi before coming home and getting on with it. His move was this. Put every known element in order of increasing atomic weight, and number them 1, 2, 3, 4. Then: the eighth resembles the first. The ninth resembles the second. He called it the law of octaves, and he meant the musical analogy seriously — in a note to Chemical News in 1865 he compared it to the eighth note of a scale. The numbering is the part worth pausing on. Newlands assigned every element an ordinal, and those ordinals are, for the light elements, exactly the atomic numbers. He had no idea what they counted, and neither did anyone else for another forty-eight years. But he wrote them down.

Look at where it breaks, because the reason is the whole shape of the modern table. Rows of seven work because the second and third periods hold eight elements each. Lithium to neon is eight; sodium to argon is eight. Step eight places along and you land in the same family, necessarily. The fourth period holds eighteen. So after calcium the rhythm changes, and an arrangement built on a fixed step of eight cannot follow it. What is filling those extra ten places is a row of transition metals — which is the same ten elements that broke Döbereiner’s nitrogen triad, showing up again in a different disguise. Newlands could not know that. Nothing in an atomic weight announces that the period is about to get longer. His rule is right, over the range where the period is eight, and he had no way to find out that the range ends. There is one more thing the simulation shows, and it is not something Newlands could have used but it is worth noticing. In the fourth row, sorted honestly by weight, nickel comes before cobalt. Later in the table, iodine comes before tellurium. Both of those are wrong, and it will take until 1913 to prove they are wrong.

Professor G. C. Foster humorously inquired of Mr. Newlands whether he had ever examined the elements according to the order of their initial letters.

1. 1817Döbereiner notices strontia’s weight sitting halfway between lime and baryta. Nobody follows it up, including him, for twelve years.
2. 1829The triads paper: the alkali metals, the alkaline earths, the halogens and the sulphur family, all obeying the same arithmetic.
3. 1860Karlsruhe. One consistent set of atomic weights, handed out as a pamphlet at the door.
4. 1862De Chancourtois winds the elements round a cylinder — the telluric screw, published without its diagram, and consequently unreadable.
5. 1864Odling tabulates; Lothar Meyer plots atomic volume against weight and gets a curve with peaks.
6. 1865Newlands numbers every element and states the law of octaves in Chemical News.
7. 1866The Chemical Society hears the paper, jokes about alphabetical order, and declines to print it.
8. 1869Mendeleev leaves the boxes empty.
9. 1887The Royal Society awards Newlands the Davy Medal.