Wait, what? An astronomy report shows a neat U-shaped dip centred at phase 0. The points from many observations seem to sit on one tidy curve. A student points at the graph and says, “So the telescope watched one complete orbit continuously, and the planet made exactly this same dip all the way around.”
That is a tempting reading of the picture, but the picture has been rearranged. A phase-folded light curve can take measurements collected at many different times and place observations from the same part of a repeating cycle together on a common phase axis. It is useful because repeated patterns become easier to see. It is dangerous only when a reader forgets that the horizontal axis is no longer ordinary clock time.
Quick answer
No. A phase-folded light curve is usually a transformed representation of measurements from repeated cycles, not one uninterrupted record of a single orbit. The learner’s job is to recover what was measured in ordinary time, what period was used to fold the data, which points came from different cycles, whether the cycles are actually consistent, and how strongly the folded pattern supports the scientific claim.
The owned learner job — and the boundary
This Reality Lab owns one evidence-transfer job: how to evaluate a phase-folded scientific light curve as a representation built from time-series evidence. It does not own the physics of exoplanet transits, stellar rotation, orbital mechanics, detectors, graph-reading in general, or statistical modelling. Those broader concepts remain with their existing Science owners.
For the general distinction between what is directly observed and what is inferred, route to How to Tell Observation, Inference, Prediction and Explanation Apart in PSLE Science. For the broader inquiry habit of checking information and methods before suggesting an improvement, use How to Evaluate PSLE Science Observations, Information and Methods. The series hub is PSLE Science Reality Lab.
Case file: seven nights become one cycle
Consider this original classroom dataset. A telescope measures the relative brightness of a star over several nights. Suppose a repeating dip seems to occur about every 2 days.
| Observation time (days) | Relative brightness | Cycle label added later |
|---|---|---|
| 0.05 | 1.000 | Cycle 1 |
| 0.42 | 0.999 | Cycle 1 |
| 1.96 | 0.991 | Cycle 1 |
| 2.06 | 0.992 | Cycle 2 |
| 2.39 | 1.001 | Cycle 2 |
| 3.98 | 0.990 | Cycle 2 |
| 4.08 | 0.993 | Cycle 3 |
| 4.44 | 0.999 | Cycle 3 |
| 5.97 | 0.992 | Cycle 3 |
| 6.07 | 0.991 | Cycle 4 |
In the original time series, those points are separated by hours or days. If a 2-day period is used, the data can be converted to a phase: where each observation sits within a repeating 2-day cycle. Measurements around 1.96, 3.98 and 5.97 days can land near the same phase even though they came from different dates. The graph can then place them near one another.
The folded plot is not fake. It is a useful re-expression of real measurements. But it is a new representation with a new question: do measurements taken at corresponding parts of repeated cycles line up?
Unfold the graph before you believe the story
A good evidence reader mentally performs an “unfolding” check. Ask:
- What quantity was actually measured?
- At what ordinary times were the measurements taken?
- What repeating period was assumed or estimated?
- How was clock time converted into phase?
- Were points averaged or binned?
- Did different cycles agree, or does the fold hide cycle-to-cycle differences?
- Does the folded pattern support the specific claim, or merely make it look tidy?
Observed, transformed, modelled and claimed
| Layer | What it means | What it does not automatically mean |
|---|---|---|
| Observed | Brightness measurements at recorded times | A planet was directly photographed going around the star |
| Transformed | Times are mapped to positions within an assumed repeating cycle | All points were taken during one continuous cycle |
| Summarised | Points may be binned, averaged or detrended for display | Every individual cycle had exactly the displayed shape |
| Modelled | A curve may be fitted to the phase-folded measurements | The model is itself an observation |
| Claimed | The pattern may support a periodic phenomenon | Every possible alternative explanation has been eliminated |
Why scientists fold time-series data
Repeating signals can be difficult to see when they are spread across a long timeline. If an event recurs once every few days, ordinary time-series data may contain many long stretches between events. Folding lets a scientist place corresponding parts of multiple cycles together. That can reveal whether a small dip or rise repeats at a consistent position in the cycle.
Current TESS archive products provide examples of this practice. MAST documentation describes phase-folded light curves for threshold-crossing events, together with model fits and unfolded views used for diagnostics. The recent TESS All-Sky Rotation Survey likewise supplies the time series and phase-folded views, with individual measurements and phase-binned summaries. Those two representations answer related but different questions.
The axis audit: “phase” is not “day”
A phase axis often runs through one repeating cycle, for example from 0 to 1 or from −0.5 to +0.5. Phase 0 and phase 1 can represent equivalent positions in consecutive cycles. They are not necessarily one day apart. The period could be hours, days or something else.
That is why the first representation check is almost embarrassingly simple: read the horizontal-axis label and its definition. A beautifully smooth U-shape can still be misunderstood if the reader silently replaces “phase” with “time”.
Worked case 1: three dips, one folded dip
Suppose three brightness dips occur at days 2.0, 4.0 and 6.0. Each is about 1% deep. A 2-day folding period places all three near phase 0. A graph then shows one dense dip.
Valid conclusion: the measurements contain dips that recur at roughly corresponding phases when a 2-day period is used.
Invalid conclusion: the telescope observed one single dip continuously for six days. The folded representation combined evidence from different cycles.
Worked case 2: the average hides a changing event
Cycle A has a 0.8% dip, Cycle B a 1.3% dip and Cycle C a 0.9% dip. A binned phase-folded curve may show a clean average near 1.0%. If the reader sees only the summary, the variation among cycles can disappear.
A better evaluation asks for the individual points or separate cycles. If the scientific claim is “the event has a stable depth every cycle,” variation is directly relevant. If the claim is merely “a recurring dip exists,” moderate cycle-to-cycle variation may not destroy the pattern.
Worked case 3: the wrong folding period can manufacture confusion
Imagine the true repeating interval is close to 2 days, but a student folds the data at 1 day. Dips that belong together may be split across different phases or unrelated points may overlap. The resulting graph could look noisy, doubled or strangely symmetric.
The key point is not that there is one magic period hidden in every dataset. It is that the appearance of a phase-folded plot depends on the period used to construct it. The period is part of the method and must be checked, not treated as invisible background.
Worked case 4: a gap disappears after folding
A telescope cannot observe a target for several hours during one cycle. In a second cycle it obtains measurements during those missing phases. After folding, the combined plot looks well filled. That can be scientifically useful, but a reader must not infer that every phase was observed during every cycle.
The folded graph can repair visual coverage of the cycle by combining occasions; it cannot rewrite the original observation schedule.
Worked case 5: two possible periods
An original signal has two similar features within a repeating pattern. A computer analysis suggests one strong period, but a period twice as long also produces an orderly fold. A current MAST rotation product explicitly shows phase-folded views at twice a detected period to help identify half-period aliases. That is a useful reminder: a clean fold is evidence to evaluate, not a certificate that the chosen period is uniquely correct.
Representation check: raw time series beside folded view
Whenever possible, place two views side by side:
- Unfolded time series: preserves when each measurement happened, gaps, drift and changes across cycles.
- Phase-folded view: emphasises what repeats at corresponding positions in a chosen period.
Neither is automatically “the real graph” and the other fake. Each preserves some information and compresses other information. Strong scientific communication tells the reader which transformation was applied.
Method check: five questions before trusting the fold
- Period source: Was the period independently known, estimated from the same data, or chosen after trying many possibilities?
- Pre-processing: Was a trend removed? Were instrumental systematics corrected?
- Binning: Are displayed points individual observations or summaries of many observations?
- Coverage: How many cycles and how many measurements contribute at each phase?
- Consistency: Do separate cycles or observing sectors show a similar pattern?
Alternative explanations live outside the tidy curve
A recurring-looking brightness pattern may have more than one possible explanation depending on the system: an orbiting body, an eclipsing binary star, stellar rotation and spots, instrument systematics, contamination from a nearby source or an incorrect period. This page does not teach how to diagnose each possibility. Its job is narrower: do not allow the transformation itself to eliminate alternatives that the evidence has not tested.
A phase-folded curve can organise evidence for comparison with explanations. It does not replace the rest of the evidence chain.
What would strengthen a claim based on the folded curve?
- The same feature recurs in multiple independent cycles.
- The unfolded time series shows events at the times predicted by the proposed period.
- Separate observing sectors or sessions produce a consistent pattern.
- Individual data points, not only a smoothed summary, support the feature.
- Instrument and background checks do not reveal a simpler artifact explanation.
- A physically appropriate model predicts additional evidence that is later observed.
What would weaken it?
- The pattern appears only after aggressive averaging and vanishes in individual measurements.
- One cycle contains the whole effect while other cycles do not.
- A slightly different period destroys the supposed recurrence and there is no independent period evidence.
- Gaps or removed points coincide with the most important part of the claimed signal.
- A nearby source or instrument behavior can reproduce the same pattern.
- The public graphic omits the definition of phase, binning or preprocessing.
How far can the conclusion travel?
If a well-documented dataset repeatedly shows a brightness dip at consistent phase, a careful conclusion can say that a periodic pattern is present under the stated analysis. Moving from that pattern to “there is definitely a planet,” “the orbit has exactly this shape,” or “every cycle is identical” requires additional evidence and modelling.
This is a general PSLE Science habit: the conclusion should travel only as far as the evidence chain reaches.
Tempting reasoning that fails
- “The graph is smooth, so the phenomenon itself was smooth.” The display may include folding, binning, averaging or a model overlay.
- “All the points near phase 0 happened at the same time.” They may come from many different cycles.
- “One clean folded graph proves the period.” Period choice is part of the analysis and may need independent checks.
- “Phase 0.5 means half a day.” Phase is a fraction or position in the chosen cycle, not a time unit unless converted using the period.
- “A fitted curve is what the telescope saw.” The fit is a model built from the measurements.
PSLE-style transfer case: the insect activity recorder
This is original practice written for transfer, not a reproduced examination question.
A light sensor outside a garden records moth activity for six nights. A researcher suspects a pattern repeating every 24 hours. She converts every clock time to a position in a 24-hour cycle and plots all six nights on one phase graph. The graph has a strong peak near phase 0.8.
Question 1: Does the peak prove all six nights had exactly the same number of moths at that time?
Explained answer: No. Measurements from different nights have been placed at corresponding phase positions. The combined peak can show a recurring timing pattern without proving identical counts on every night.
Question 2: What original evidence should be checked?
Explained answer: The separate night-by-night time series, including gaps and the actual counts, should be checked to see whether the pattern appears across nights rather than being dominated by one night.
Question 3: What would weaken the claim of a 24-hour repeating pattern?
Explained answer: If only one night contains the peak while the others show no similar increase, the combined folded plot may give an exaggerated impression of consistency.
Delayed independent return
- Why can measurements from different dates appear beside one another on a phase-folded graph?
- What information is easiest to lose when ordinary time is replaced by phase?
- Why should individual cycles be checked even when the folded graph looks clean?
- How can an incorrect period affect the representation?
- Why is a model line different from an observation?
Self-check: folding aligns repeated positions in a chosen cycle; ordinary timing and cycle identity can be compressed; cycle-by-cycle evidence tests repeatability; the period controls where points land; a model is an explanatory or predictive construction fitted to evidence.
Practice: unfold these claims
- A graph says “phase-folded at 3.2 days.” What must you know before interpreting phase 0.25? The phase convention and period; 0.25 is a quarter-cycle position, not automatically 0.25 day.
- Black points are labelled “phase-binned medians” and grey points “individual measurements.” Which preserves more detail about spread? The individual grey measurements.
- Three observing sectors show the same dip at the same phase. Does that strengthen recurrence? Yes, because the pattern appears across separated observations, though alternative explanations still need checking.
- The folded curve looks perfect, but the unfolded curve contains one enormous instrumental jump at every event time. What should happen? Investigate the instrumental explanation before treating the fold as evidence of the proposed phenomenon.
Parent and tutor teaching guide: fold a paper timeline
Draw a long strip marked 0 to 8 days. Put a small circle at days 1.9, 3.9, 5.9 and 7.9. Then cut or physically fold the strip into four 2-day sections and stack them. The circles line up. Ask the learner: “Did the four circles happen at one time?” Of course not. They align because we changed the representation.
Next, move one circle slightly. The folded stack now shows spread. That makes a powerful bridge to scientific graphs: folding is useful precisely because it lets us compare corresponding parts of repeated cycles, including their disagreements.
For a small group, give one learner the role of time keeper, another representation checker, and another claim limiter. The time keeper reconstructs when observations happened. The representation checker describes how the graph was transformed. The claim limiter writes the strongest conclusion that does not exceed the evidence.
Authoritative sources and curriculum frame
- SEAB — 2026 PSLE Science syllabus: the current examination assesses the 2023 Primary Science syllabus and includes interpreting and analysing information, evaluating observations, information and methods, and communicating explanations and reasoning.
- MOE — 2023 Primary Science Teaching and Learning Syllabus: includes communicating, evaluating and defending ideas with evidence, use of multiple representations, and healthy scepticism toward observations, methods, processes and data.
- MAST / TESS Science Data Products Description: documents phase-folded light curves, model overlays, sector views and unfolded light curves in TESS data-validation products.
- MAST — TESS All-Sky Rotation Survey: provides current time-series and phase-folded data products, including individual measurements and phase-binned summaries.
The quiet habit to keep
When a scientific graph looks unusually tidy, do not distrust it merely because it was transformed. Instead ask the better question: what transformation turned the original observations into this view, and what information did that transformation preserve or compress? A phase-folded light curve becomes powerful when you can mentally unfold it again.
