Lesson 6 of 6 · Measurements that could distinguish models
Measurements that could distinguish models
10 minute lesson · 3 checks along the way
Your goal for this lessonChoose complementary observations, track shared calibration errors, and distinguish particle-search candidates from discoveries.
Follow a linked name to explore its archive card. “Back to your lesson” brings you to the same part of the reading.
Words with a dotted underline open a short definition right here in the lesson.
Ask where predictions separate
Two models that reproduce one graph need not make the same predictions everywhere. A useful follow-up measures an observable on which their predictions differ by more than the achievable uncertainty. Repeating the same measurement can improve precision, but a second method may test an assumption that the first cannot. Designing that comparison begins with a stated model prediction, a measurable quantity, and a realistic account of instrumental and astrophysical uncertainties.
Euclid combines weak-lensing shapes with galaxy clustering and distance information. The geometry of expansion and the growth of structure provide complementary tests: in a specified gravity model they are related, so a model that fits distances must also account for growth. provide a characteristic clustering scale. contain information about galaxy motions and growth; they are physical effects, not simply an instrument malfunction.
Euclid is secured to its launch adaptor before launch in 2023. The white enclosure contains the telescope; dark solar panels run along the right side. This is the observing hardware, not an image of the matter distribution that its survey reconstructs.
Look closely
What can this hardware photograph establish, and which course claims require the measurements the telescope later collects?
See what to notice
It shows the spacecraft and visible telescope enclosure, linking the survey to a real instrument. It cannot by itself measure galaxy shapes, distances, structure growth, or dark-matter abundance; those require calibrated observations and analysis.
Check 1 of 3
10 pointsUse the reading above. You can try again without losing points.
Separate random scatter from a shared offset
Suppose two fictional, fixed models predict a dimensionless observable: A predicts 1.00 and B predicts 1.08. A measurement gives 1.02 with 0.02 from statistical noise alone. Its central value is one stated uncertainty above A and three below B. Under these restricted assumptions, A is closer. That arithmetic does not give the probability that either entire model is true.
Now discover a calibration correction that raises the central value by 0.04, to 1.06. It is then three of the original statistical uncertainties from A and one from B. The ranking reverses before considering uncertainty in the correction itself. Collecting more observations with the same unknown offset would not guarantee a valid conclusion. An independent calibration and an honest combined uncertainty budget are essential parts of the experiment.
A shared correction can reverse the ranking
The fictional comparison holds the two predictions fixed and displays only the original statistical uncertainty. A complete analysis must include uncertainty in the correction and any model parameters. Distance in standard uncertainties is not a model probability.
Open full-size diagram in a new tab: A shared correction can reverse the rankingCheck 2 of 3
10 pointsUse the reading above. You can try again without losing points.
Work through a decision
Keep particle experiments and sky surveys distinct
An underground dark-matter experiment asks a different question from a lensing survey: can a proposed particle interaction produce a detectable signal in a carefully shielded target? A result with no convincing signal constrains particular combinations of particle mass, interaction, and other model assumptions. It does not eliminate every possible dark-matter explanation. Background radiation, detector response, event selection, and the number of hypotheses tested must be characterized.
In September 2026, the LZ collaboration reported one unusual nuclear-recoil candidate in a search allowing broader interaction models. Its reported was 2.6 standard deviations after accounting for the search across possibilities. The collaboration did not claim a dark-matter discovery. Such a statistical statement describes compatibility with a background test under its assumptions; it is not the probability that the background hypothesis is true, nor an automatic probability that the event was .
Build a decision that another observer can test
For the fictional A-versus-B problem, prioritize a calibration measurement that can establish whether the suspected offset is real. Then add an observable whose predictions differ for a separate physical reason, ideally with substantially different systematic errors. Report the measured value, uncertainty convention, corrections, and assumptions. This makes a disagreement traceable instead of leaving a future team to guess how an impressive significance figure was produced.
Dark-sector research progresses through this combination of sky maps, time-dependent cosmological tests, and laboratory searches. Particle experiments probe interactions; lensing probes gravitational effects; distances and growth constrain cosmic models. A convincing explanation must connect its predictions to the appropriate evidence. Finding agreement across methods would be more powerful than treating a single unusual event, colorful image, or unexpectedly precise number as the end of the investigation.
Check 3 of 3
10 pointsUse the reading above. You can try again without losing points.
Put it together
Your lesson progress
Each check connects to a part of the story. Complete all 3 to earn your 25-point lesson bonus.
Go to the source
Explore the evidence behind this lesson. These optional readings and videos open in a new tab.
- Read: Euclid: geometry and growth probesHow weak lensing, clustering, and redshift-space distortions supply complementary information.
- Read: LZ collaboration: current resultsDated September 2026 candidate and its global significance; no discovery claim.
- Read: Berkeley Lab: LZ candidate contextPrimary laboratory account. Interpret significance as a conditional test, not the probability a hypothesis is true.