The short answer
A precise disagreement can be real without its explanation being unique. The historical question was not simply whether two numbers differed, but which additional observations could distinguish the explanations.
Evidence and original analysis
Hold the observation fixed
Invented ideal two-component volume model.
| Candidate | Component density | Volume fraction fitting 2.04 |
|---|---|---|
| A | 3 | 4% |
| B | 4 | 2% |
| Shared conclusion | A heavier component is compatible | Identity is not selected |
Chronology
A timeline of the turning points
Retrospective account
Discrepancy and alternatives explained.
Original sensitivity test
Change the proposed constituent, keep the measured mixture.
The discrepancy was an opening
Rayleigh described nitrogen preparations whose densities disagreed beyond expected experimental errors. He considered a heavier atmospheric constituent and an alternative involving dissociated nitrogen. An eight-month storage comparison tested the latter expectation. Ramsay and Rayleigh then pursued isolation, first separately and later together.
The density result motivated further discrimination. It did not by itself state the identity and amount of every constituent. Keeping the rejected alternative in the narrative shows what the follow-up observation was meant to test.
Sources for this section: [1]
Two invented explanations
Assume fixed temperature and pressure, ideal additive component volumes and a two-component mixture. In common arbitrary mass-per-volume units, let the base gas density be 2 and measured mixture density 2.04. Candidate A has density 3. Find its volume fraction f.
Repeat with candidate B of density 4, leaving every other assumption unchanged. Do both explanations fit? What independent information would be needed before selecting one candidate? These are fictional gases and values, not reconstructed atmospheric measurements.
Sources for this section: [1]
Reference answer
For A, 2(1−f)+3f=2.04, giving f=0.04. For B, 2(1−f)+4f=2.04, giving f=0.02. Thus a 4% fraction of A and a 2% fraction of B produce the same measured mixture density under this model.
The fit cannot choose the identity. An independent test sensitive to composition, or independently established component properties, is needed. Even a perfectly precise density would leave this particular ambiguity; reducing measurement error does not remove an unspecified component.
Sources for this section: [1]
Precision and identification are different achievements
The calculation does not diminish precise measurement. It identifies the next question that measurement leaves open. The historical account’s subsequent isolation work is therefore a separate evidential step, not a decorative confirmation of an identity already read directly from a weighing.
Sources for this section: [1]
Continue the chronology
Related history guides
Biography context
Explore United Kingdom through connected lives
Connected biographies
The researcher and the record
Quick answers
Common questions
- Are these historical argon percentages?
- No. Both candidates and all numerical densities are fictional.
- Would a more precise 2.04 choose A or B?
- No. Both models give exactly 2.04 under their stipulated assumptions.
Primary and institutional sources
How this history was checked
Selected historical passages are distinguished from the original teaching model. Invented inputs are not reconstructed experimental data. This article was reviewed on 31 October 2026.