How can astronomers discover a planet they cannot see directly? One answer begins with a graph. If a planet passes between its star and our telescope, it blocks a tiny fraction of the star's light. Repeated, carefully measured dips can reveal an orbital period, a size ratio and clues about the planetary system. The event is called a transit, and the brightness-against-time graph is a light curve.
This is a wonderful example of why mathematics is important in science. Ratios turn a brightness dip into a radius comparison. Periodic patterns identify repeated orbits. Geometry explains why most orbital alignments do not produce transits from Earth. Kepler's third law connects period with orbital scale when the star's mass is known. Statistics separates a possible signal from noise and competing explanations. The mathematics does not make a candidate automatically become a confirmed planet; it helps astronomers state what the data support and what still needs checking.
A quick map of the mathematics
- Relative flux compares measured brightness with a baseline outside transit.
- Transit depth is the fractional drop in flux during a transit.
- Area ratio gives the useful approximation depth ≈ (planet radius/star radius)².
- Period is the time between corresponding repeated transits.
- Duration depends on orbital speed, stellar size, planet size and the path across the stellar disc.
- Kepler's third law links orbital period and semi-major axis when masses are accounted for.
- Signal-to-noise reasoning combines measurement precision and repeated events.
- Uncertainty separates a point estimate from the range compatible with data and assumptions.
The ideas connect to comparing percentages fairly because a transit depth needs a clear baseline, and to sports statistics because repeated measurements, variation and context matter more than a single impressive number.
A light curve is a measurement story
A light curve places time on the horizontal axis and brightness, or flux, on the vertical axis. Astronomers often normalise the out-of-transit brightness to 1.000. A reading of 0.990 then represents one per cent less measured light than the chosen baseline. The graph is not a drawing of the planet's physical path. It is a record of how detected light changes with time.
Every point comes from an instrument and a reduction process. A camera records counts in pixels. Software corrects detector effects, estimates background light, locates the star and sums or models its brightness. The atmosphere may affect ground observations. Space telescopes avoid much atmospheric variation but still have pointing changes, detector behaviour and cosmic-ray events. Mathematics enters before the final graph appears.
Absolute brightness is often less useful than relative change
For a transit, the key measurement is a small change relative to the same star's normal level. If Fout is baseline flux and Fin is flux during transit, the fractional depth is:
δ = (Fout − Fin)/Fout.
If the baseline is normalised to 1, this becomes δ = 1 − Fin. A depth of 0.01 is 1%, 0.001 is 0.1% and 0.0001 is 0.01%. Converting carefully between fraction, decimal, percentage and parts per million prevents dramatic errors.
Did You Know? One percentage point is not one per cent
If normalised brightness falls from 100.0% to 99.0%, the drop is one percentage point and also one per cent of the original baseline. In other contexts those phrases differ. A fall from 50% to 49% is one percentage point but a two per cent relative decrease. Astronomy rewards precise denominator language for the same reason everyday data does.
Transit depth and the area-ratio model
Imagine a uniformly bright circular star with radius Rs and an opaque circular planet with radius Rp. If the planet passes fully in front of the star and does not cover the edge, the blocked fraction of area is:
δ ≈ πRp²/πRs² = (Rp/Rs)².
The π cancels. Taking a square root gives Rp/Rs ≈ √δ. The light curve therefore gives a radius ratio, not an absolute planet radius by itself. To estimate Rp in kilometres, astronomers also need an estimate of Rs.
Worked example: a one per cent dip
Suppose δ = 0.0100. Then Rp/Rs ≈ √0.0100 = 0.100. The planet's radius is about one tenth of the star's radius under the simplified assumptions. Its circular cross-sectional area is about one hundredth of the star's.
This is a useful place to resist linear intuition. A planet with half the radius blocks one quarter, not half, of a uniformly bright disc. If radius ratio doubles, area ratio quadruples.
Worked example: a shallow dip
Suppose the baseline is 1.00000 and the in-transit level is 0.99960. The depth is 0.00040, or 0.040%, or 400 parts per million. The radius ratio is √0.00040 ≈ 0.020. The planet's radius is about two per cent of the star's radius in the simple model.
If the stellar radius estimate has uncertainty, the planet-radius estimate inherits it. A beautifully measured depth does not remove uncertainty in the star.
Why the simple depth formula is only a first model
Stars are not uniformly bright discs. Their centres usually appear brighter than their edges, an effect called limb darkening. A planet crossing the bright centre blocks more light than the same planet covering a dimmer region near the limb. Detailed transit models include limb-darkening laws and fit the whole curve rather than using only a flat-bottom depth.
Other light can dilute the transit. If an unresolved neighbouring star contributes flux, the planet blocks the same host-star area but the combined baseline is larger. The observed fractional dip becomes shallower, making the radius ratio appear smaller if dilution is ignored.
Starspots and faculae change stellar brightness. A planet crossing a dark spot may temporarily block less light, creating a bump within the transit. Stellar rotation can move active regions into and out of view. These are not reasons to abandon the method; they are reasons to model the data honestly.
Grazing transits
If the planet clips only the stellar edge, it may never be fully superimposed on the disc. The light curve can look V-shaped rather than having a clear flat bottom. Radius, path and limb darkening become more strongly entangled. A shallow V-shaped event might also be an eclipsing binary blended with another star.
“Depth equals squared radius ratio” is therefore a good teaching approximation with named conditions. Responsible mathematics says when an equation is useful and when a more complete model is required.
Period comes from repeated timing
If corresponding dips occur at times t1, t2, t3 and the orbit is stable, the period P is approximately the time between them. With many events, fit a linear ephemeris:
tn = t0 + nP,
where n is an integer transit number. Plotting observed times against n gives slope P and intercept t0. Using many transits reduces the influence of one noisy timing estimate.
NASA's light-curve explanation notes that elapsed time between transits gives the orbital period. NASA's What’s a transit? also explains that the brightness dip and repetition carry information about the planet and orbit.
Worked example: fit the period
Suppose transit centres occur at days 2.4, 7.9, 13.4 and 18.9. Consecutive differences are 5.5 days each, so P = 5.5 days. If the final centre were 19.0 instead, the individual intervals would be 5.5, 5.5 and 5.6 days. A line fitted across all four times would estimate a period near 5.53 days rather than treating every difference as exact.
Missing a transit
If observations contain gaps, two recorded dips 11 days apart could represent a period of 11 days or a shorter period whose intermediate event was missed. Other data and continued observation resolve aliases. Period-finding is a reasoning problem, not merely subtraction.
Folding a light curve
Phase folding stacks data from many orbits onto one cycle. Choose a trial period P and reference time t0. For each observation time t, compute phase:
φ = fractional part of ((t − t0)/P).
Phase runs from 0 to 1. Points separated by one whole period receive the same phase and align. A true repeated transit becomes clearer because data from many events reinforce the shape; unrelated noise tends to scatter.
A wrong period leaves a clue
If the trial period is slightly wrong, successive transits drift in phase and the folded dip broadens. If the chosen period is twice the true value, alternate transits may appear at the same phase while others appear halfway around the cycle. If it is half the true period, two different orbital locations are combined. Folded plots are powerful diagnostic views, but they can also hide time-dependent changes.
NASA's TESS tutorial How to Find an Exoplanet demonstrates working with real light curves and folding data at a candidate period. It is a useful next step after a hand-built example.
Do not average away the question
Binning phase values can make a curve easier to see. Larger bins reduce scatter but blur ingress and egress, the entry and exit slopes that carry geometric information. Display both individual points and binned summaries where possible. A smooth line without raw variation can make evidence look more certain than it is.
Kepler's third law connects period and orbit size
For a planet of mass Mp orbiting a star of mass Ms, Newton's form of Kepler's third law is:
P² = 4π²a³/[G(Ms + Mp)],
where a is the semi-major axis and G is the gravitational constant. If the planet is much less massive than the star, Ms + Mp ≈ Ms. Then:
a ≈ [GMsP²/(4π²)]^(1/3).
The period from the light curve therefore helps estimate orbital scale when stellar mass is known. It does not give distance from period alone; the star's mass is part of the relationship.
Solar-unit shortcut
For objects orbiting a one-solar-mass star, if P is measured in years and a in astronomical units, the relation is approximately P² = a³. A period of 0.125 years gives a ≈ (0.125²)^(1/3) = 0.25 AU. Check: 0.125² = 0.015625 and 0.25³ = 0.015625.
NASA JPL's Exploring Exoplanets with Kepler uses light curves and Kepler's third law in a student lesson. It is an official educational source and makes clear which quantities are measured and inferred.
Scale is not habitability
An orbital distance estimate does not automatically tell us whether a planet is habitable. Stellar luminosity and spectrum, atmosphere, albedo, rotation, geology and other factors matter. Even “within a habitable zone” normally refers to conditions under which liquid water could be possible under model assumptions, not proof that life exists.
Transit duration contains geometry and speed
For a central transit in a circular, edge-on orbit, a rough duration is the distance travelled across the stellar diameter divided by orbital speed. If planet size is small compared with the star, distance is about 2Rs. Orbital speed is approximately 2πa/P. Thus:
Tduration ≈ (P/π)(Rs/a).
A more complete equation includes Rp, impact parameter, inclination and eccentricity. Still, the rough formula exposes the relationships: a larger star lengthens the crossing, faster orbital motion shortens it and a path near the edge crosses a shorter chord.
Chords and impact parameter
Let b represent the projected distance of the transit path from the stellar centre in units of stellar radius. For a small planet and a straight projected path, the chord length is about 2Rs√(1−b²). At b = 0, the planet crosses a diameter. At b = 0.8, the chord factor is √(1−0.64) = 0.6, so the route across the disc is only 60% of the central chord.
Duration, ingress shape and depth help constrain b, but parameters can be correlated. A model may trade a more grazing path against other quantities and still fit noisy data. Confidence intervals and joint parameter plots are more informative than one best-fit number.
Transit probability is geometric
Most planets do not transit from our viewpoint because their orbital planes are not aligned to carry them across the stellar disc. For a circular orbit with random orientation, a rough transit probability is:
Pr(transit) ≈ (Rs + Rp)/a.
Close-in planets are therefore more likely to transit than distant planets. This creates a selection effect: a transit survey's detected sample is not a simple miniature of all planetary systems.
Worked comparison
If a star has radius 0.005 AU and the planet radius is small, a planet at 0.05 AU has transit probability about 0.005/0.05 = 0.10, or 10%. A planet at 1 AU has probability about 0.5%. These are rough values under circular-orbit assumptions.
The first planet is also likely to transit more frequently because its period is shorter. Detection favours frequent, deep, well-aligned signals around stars whose measurements are precise. Population studies correct for these efficiencies rather than treating raw detections as unbiased counts.
Why non-detection proves little by itself
If no transit is seen, a planet may be absent, too small, too long-period, hidden in noise or simply misaligned. This is a general statistical lesson: a method observes only the cases to which its geometry and sensitivity respond.
Noise is more than random scatter
Photon arrival has statistical variation often approximated by Poisson behaviour. Detector electronics add read noise. Background light contributes counts. Pointing motion moves a star across pixels with slightly different sensitivity. Stellar granulation, pulsation and magnetic activity can create real brightness variation.
Some noise is approximately independent from one measurement to the next; other noise is correlated over time. If 100 independent observations are averaged, standard error can shrink by roughly a factor of √100 = 10. Correlated variation does not necessarily average down that quickly. Treating every point as independent can exaggerate confidence.
Signal-to-noise for repeated transits
In a simple model with transit depth δ, per-point standard deviation σ and N independent in-transit points, a detection signal-to-noise estimate scales like δ√N/σ. Doubling depth doubles this ratio; quadrupling the number of points doubles it. This square-root law explains why repeated observation helps but with diminishing returns.
Suppose δ = 0.001, σ = 0.002 and N = 100. Then the simplified signal-to-noise estimate is 0.001×10/0.002 = 5. If the points are strongly correlated, the effective N is smaller and the true significance is lower.
A threshold does not create truth
Detection pipelines may use thresholds to select candidates for further checks. A score above threshold is not automatic confirmation; a score below it is not proof of absence. Thresholds balance missed signals against false alarms and depend on the search space, noise model and validation plan.
Detrending without deleting the transit
Long-term brightness drift can obscure a short transit. Detrending estimates the slowly varying baseline and divides or subtracts it. Methods include polynomials, splines, moving filters and models tied to spacecraft behaviour. Every method has a scale or flexibility choice.
If the trend model is too rigid, residual drift remains. If it is too flexible, it may absorb part of the transit and make the event shallower. Analysts often mask candidate transits while fitting the baseline, test multiple settings and inject synthetic signals to measure recovery.
The danger of seeing what the pipeline expects
After processing, a clean transit-like curve may seem inevitable. Yet each step—outlier removal, aperture choice, detrending, period search and folding—changes the data view. Reproducible work records settings and checks whether the signal appears under reasonable alternatives.
This is a useful lesson for student graphs. “I cleaned the data” is not enough. State which points were removed, why, and whether the conclusion changes when they remain.
False positives and competing explanations
An eclipsing binary star can create periodic dips. If its light blends with a brighter star, a deep stellar eclipse may look like a shallow planetary transit. Instrumental systematics can repeat with operational cycles. Starspots can create quasi-periodic variation. A genuine planet around a nearby contaminating star may be assigned to the wrong target.
Astronomers examine odd and even transit depths, secondary eclipses, centroid shifts, colour dependence, stellar properties and follow-up observations. Radial-velocity data may estimate mass or expose a stellar companion. High-resolution imaging can find nearby sources. Statistical validation compares planetary and false-positive scenarios.
Radius is not mass
A transit primarily constrains radius relative to the star. Two objects with the same radius may have very different masses and compositions. Calling every transit-sized body “Earth-like” because its radius is similar to Earth's would ignore atmosphere, mass, temperature and environment.
Confirmation uses converging evidence
The correct question is not “Does this graph look planet-shaped?” It is “Which model best explains all relevant observations, and what alternatives remain plausible?” Mathematics helps calculate likelihoods and parameter ranges; scientific judgement designs the tests.
Transit timing variations
If transit centres do not follow tn = t0 + nP exactly, plot O−C: observed time minus calculated time. A structured pattern may indicate gravitational interaction with another planet, orbital changes or an incorrect period. Random scatter may reflect timing uncertainty.
Suppose a period estimate is too long by 20 seconds. After 100 orbits, the predicted transit is about 2,000 seconds late, more than 33 minutes. Small period errors accumulate. This is why a long time baseline greatly improves ephemerides.
Timing needs a standard
Times recorded in different time zones or reference systems cannot simply be mixed. Astronomical timing may correct for light-travel effects and use precise standards. For classroom data, use one stated time unit and origin. The deeper lesson is universal: a number without its reference system may be unusable.
Uncertainty travels through the calculation
If δ is measured with uncertainty and Rp/Rs = √δ, the radius-ratio uncertainty is not the same numerical size. For small errors, differentiation gives:
σ(Rp/Rs) ≈ σδ/(2√δ).
If δ = 0.0100 ± 0.0004, then √δ = 0.100 and the approximate ratio uncertainty is 0.0004/(0.2) = 0.002. So Rp/Rs ≈ 0.100 ± 0.002 before adding stellar-radius uncertainty and model systematics.
Correlated parameters
Stellar radius, limb darkening, impact parameter and radius ratio may influence similar parts of the curve. Their errors are not always independent. Adding separate error bars by a simple root-sum-square formula can be misleading when covariance matters. Modern fits often sample a joint posterior distribution to show parameter relationships.
Students can learn the core idea without advanced statistics: vary inputs across plausible ranges, recompute outputs and report how much the conclusion changes. Sensitivity analysis is better than false precision.
Completeness turns detections into population evidence
A catalogue of detected planets is shaped by the survey. Suppose a survey watches 10,000 suitable stars and finds 50 planets of a particular size and period. Dividing 50 by 10,000 gives 0.5%, but that is not automatically the fraction of stars hosting such planets. Only a fraction of orbits transit, and the pipeline misses some transits that do occur.
Imagine the geometric transit probability is 10% and injection tests show the pipeline recovers 50% of aligned planets in this category. A very simplified overall detection probability is 0.10×0.50 = 0.05. If every star were equally suitable and each hosted at most one relevant planet, 50 detections would then suggest roughly 50/0.05 = 1,000 underlying planets among 10,000 stars, or about 10%. Real occurrence-rate studies model star-by-star sensitivity, false positives and parameter uncertainties, but the example shows why raw counts understate hidden populations.
Injection and recovery
To estimate completeness, analysts add synthetic transit signals with known properties to real or realistic data, run the detection pipeline and record which signals are recovered. Recovery fraction depends on depth, period, duration, stellar noise and data gaps. This tests the whole workflow rather than assuming a nominal noise level predicts performance perfectly.
The exercise also guards against an appealing but circular method: selecting only obvious signals and then claiming the method is highly effective. A fair test includes difficult and missed cases whose truth is known.
Reliability is a different question
Completeness asks what fraction of real target signals are found. Reliability asks what fraction of reported detections are real. A permissive threshold may raise completeness while lowering reliability. Tightening it may do the opposite. Survey design balances both according to the scientific goal and follow-up resources.
This distinction appears throughout education and decision-making. A screening test, spam filter or search system has both missed cases and false alarms. One accuracy percentage rarely describes both.
Careful surveys therefore publish selection rules and sensitivity information alongside the exciting catalogue, allowing later researchers to reproduce population estimates.
A safe student investigation with open data
Begin with a synthetic light curve so the hidden truth is known. Create times from day 0 to day 20 in steps of 0.02 day. Set baseline flux to 1. Add a transit of depth 0.01 lasting 0.15 day every 3.2 days. Then add small random noise, for example standard deviation 0.002.
Plot the raw time series. Search trial periods near 3.2 days and fold the curve. Estimate the median out-of-transit flux and in-transit flux, calculate depth and infer radius ratio. Compare the answer with the value used to generate the data.
Make it a fair experiment
- Keep the original noiseless curve.
- Change only one factor at a time: depth, noise, period or missing observations.
- Repeat random noise several times rather than trusting one lucky simulation.
- Record the method used to estimate baseline and transit centre.
- Test whether a wrong period can produce a misleading folded pattern.
- State when the box-shaped model no longer resembles a real transit.
After mastering synthetic data, use NASA's Exoplanet Watch analysis guidance to understand how telescope images become light curves. Treat it as a research workflow, not a race to announce a discovery.
How this connects to school mathematics
Ratio and percentage
Transit depth is a relative change. Radius ratio comes from a square root. Converting 400 parts per million to 0.0004 and 0.04% is essential numerical literacy.
Graphs and functions
Students read axes, identify baselines, compare shapes and recognise periodic structure. A light curve makes “function of time” concrete.
Geometry
Circles, areas and chords explain depth and duration. Inclination changes the projected path even when the orbit itself is unchanged.
Algebra
Rearranging δ = (Rp/Rs)² and Kepler's law develops control of powers, roots and units. An equation becomes a relationship to interrogate, not a template to fill.
Statistics
Noise, repeated measurements, outliers, confidence and alternative hypotheses determine whether a pattern is persuasive. This is the bridge from textbook calculation to evidence.
Common misconceptions worth correcting
“A one per cent dip means the planet is one per cent as wide”
Depth follows area ratio in the simple model. A one per cent depth gives a radius ratio near ten per cent.
“Every planet transits its star”
Only systems aligned suitably from our viewpoint transit. Geometry strongly selects which planets the method can detect.
“A repeating dip proves a planet”
Eclipsing binaries, blends, stellar activity and instruments can imitate signals. Candidates need validation or confirmation using additional tests.
“The light curve directly gives kilometres”
It gives dimensionless ratios and timing. Absolute planet size depends on stellar properties; orbital scale depends on stellar mass and model assumptions.
“More decimal places mean greater certainty”
Precision of display is not uncertainty. Measurement noise, stellar parameters and model choices limit justified digits.
Guidance for students
Always label time units and brightness normalisation. Calculate depth as a fraction first, then convert to percentage or parts per million. Draw a small geometry diagram and write the assumptions beside the formula. If the transit is grazing or blended, do not force the central, uniform-star model.
Use multiple representations: time-series plot, phase-folded plot, data table and equations. Each reveals a different feature. Keep raw points visible when showing a smoothed curve. When a result surprises you, check the baseline, period alias and units before inventing an exotic explanation.
Treat public data with care. A school analysis can reproduce an established signal and still be excellent science learning. Do not label a new exoplanet from a single graph; use “candidate feature” and explain what follow-up would be needed.
Guidance for parents and educators
Exoplanet transits can connect percentages, circles and graphs without requiring advanced astrophysics. Begin with paper discs covering a larger circle, then move to area ratios and a synthetic time series. Let students predict before calculating.
Praise questions about uncertainty and alternatives. “What else could cause the dip?” is not negativity; it is scientific maturity. Avoid claiming astronomy careers follow automatically from one topic. The transferable skills—data handling, modelling, coding, careful writing and collaboration—support many educational and career pathways.
For younger learners, use simple ratios and repeated patterns. For older learners, add Kepler's law, curve fitting, residuals and detection bias. Depth should grow with readiness, not with sensational language.
Frequently asked questions
Can a backyard telescope observe a transit?
Some relatively deep transits around bright stars can be observed with suitable equipment and careful technique. Precision, calibration and safe, appropriate supervision matter. NASA's Exoplanet Watch provides structured guidance.
Why does the light curve slope during entry and exit?
The planet gradually overlaps more or less of the stellar disc during ingress and egress. Limb darkening also shapes the slope.
Does transit depth stay exactly constant?
It may vary because of noise, starspots, dilution, instrument behaviour or real system effects. Apparent changes require careful testing.
Can the method find a planet's atmosphere?
Transit spectroscopy compares wavelength-dependent changes as starlight passes through an atmosphere. It is more demanding than a simple broadband light curve and requires specialised observations and models.
Why are short-period planets easier to find?
They transit more often within a fixed survey duration and are geometrically more likely to align. Their signals may also accumulate more repeated events.
What is the difference between validation and confirmation?
Usage can vary, but confirmation often means decisive independent evidence or dynamical measurement, while statistical validation shows false-positive scenarios are sufficiently unlikely. Read the specific study's definition.
Can a transit tell whether a planet has life?
No. A transit can provide size and orbital information, and specialised spectra may constrain atmospheric constituents. Evidence of life requires much more and must rule out non-biological explanations.
What is the best first calculation?
Measure a normalised depth, convert it among fraction, per cent and parts per million, then take its square root to estimate radius ratio under the simple model.
Useful next reading
- NASA explains the basic geometry in What’s a transit?.
- See how elapsed time between dips yields period in NASA's Light Curve of a Planet Transiting Its Star.
- Try an official classroom pathway with NASA JPL's Exploring Exoplanets with Kepler.
- Learn the data workflow through Exoplanet Watch.
- Strengthen denominator habits with comparing percentages fairly.
The larger lesson
Transit astronomy works because a tiny change can be meaningful when measurement, geometry and probability are handled well. A shallow dip becomes a radius ratio; repeated timing becomes a period; a period and stellar mass become an orbital scale. Each inference has assumptions and inherited uncertainty.
The happiest part of the story is not that one formula reveals a hidden world. It is that several modest mathematical ideas cooperate. Graphs show patterns, ratios connect areas, algebra rearranges relationships, statistics weighs evidence and models expose limits.
That is the importance of mathematics in science. It lets us reason beyond what our eyes can see while remaining honest about what the data do not yet prove. A distant planet may be invisible as a dot, yet its regular shadow can become a careful, testable argument.