Two kinds of nitrogen had different weights
In the early 1890s, Lord Rayleigh was weighing gases with enough care to notice a small nuisance: nitrogen taken from air was heavier than nitrogen made from chemical compounds. The difference was about half a percent. He eventually found that the air sample contained another gas, argon. What looked like an error in a familiar measurement became a way to discover an unfamiliar element.
The comparison sounds easy: fill a glass globe, weigh it, and subtract the weight of the empty globe. A gas, however, changes its density when its temperature or pressure changes. Air also pushes upward on the globe during weighing. Rayleigh controlled these effects and compared the working globe against a closed globe of similar external volume. A tiny change in the laboratory air then affected both sides of the balance similarly.
Changing the preparation made the discrepancy larger
One route started with ordinary air and removed its oxygen. Another used reactions that supplied nitrogen from ammonia or other compounds. At first, a preparation that mixed these origins gave only a small discrepancy. Rayleigh then made the nitrogen entirely from chemical sources. The difference grew, which made it easier to investigate.
His 1895 account lists a mean mass of about 2.3102 grams for atmospheric nitrogen in the globe and 2.2990 grams for chemically prepared nitrogen under the standard conditions of that experiment. That is roughly eleven milligrams: small beside the glass apparatus, large enough to repeat reliably. The important comparison was between gases occupying the same volume at the same temperature and pressure.
A little hydrogen left in the chemical sample could have made it lighter. Rayleigh checked different preparations instead of trusting one chemical recipe. He also considered whether the lighter gas contained nitrogen molecules broken into smaller units. After months of storage, the gas still gave the same density when he restored the filling conditions. Attention shifted toward a heavier constituent in the atmospheric sample.
Sources: [1]
The gas that survived two ways of removing nitrogen
Rayleigh returned to an experiment Henry Cavendish had reported in 1785. Electric sparks make nitrogen react with oxygen; an alkaline liquid removes the resulting nitrogen oxides. With stronger electrical equipment, Rayleigh could consume the familiar nitrogen and concentrate what remained.
Ramsay used a different route. Hot magnesium takes up nitrogen, leaving a small residue that resists the treatment. Both routes led to the same heavy, remarkably unreactive gas. Agreement between these preparations mattered: a contaminant introduced by one apparatus would have been a much less convincing explanation of the repeated result.
The residue had its own pattern of bright lines in an electrical discharge. These spectral lines supplied a further way to recognize it. Density had supplied the clue; separation and physical tests made the new substance available for investigation. The name argon referred to its reluctance to react.
A small residue explained the extra mass
Argon accounted for roughly one percent of the atmospheric gas, and it was substantially denser than nitrogen. Mixing a little of that heavier gas into nitrogen was enough to raise the average density by the observed fraction. Isolation had supplied a concrete substance whose properties explained the original weighing result.
Its scarcity also made separation laborious. To obtain a useful quantity of argon, the experimenters had to remove much larger quantities of nitrogen. Concentrating the residue turned a small effect in an average into a sample they could examine directly.
A discovery shared across two Nobel Prizes
Rayleigh and Ramsay announced argon in 1894 and published their fuller investigations afterward. In 1904, Rayleigh received the Physics Prize for work on gas densities and the discovery of argon. Ramsay received the Chemistry Prize for the inert gaseous elements and their position in the periodic system. The two awards recognized different parts of a closely connected investigation.
A label on a sample bottle can hide an assumption about its contents. Once the half-percent discrepancy could be reproduced, asking how the sample had been prepared gave a promising direction for the next experiment. Air itself supplied the unexpected material.