An interference record has a visibility of 0.30. Does that mean its two beams have a coherence magnitude of 0.30? Only under additional conditions. Beam imbalance and phase-independent background can reduce the recorded contrast even when the relationship between the optical fields has not changed.
Short answer: for a stationary two-beam scalar model with a linear detector, suitable matching polarisation and a complete phase sweep, visibility combines coherence, intensity balance and background dilution. Separate measurements of the two beam contributions and the background can identify those contributions under the model. A maximum and minimum alone generally cannot.
This guide uses original fictional intensity records and paper calculations. It provides no instructions for setting up or aligning a laser experiment. Begin with Wave Optics for phase and superposition and Learning That Transfers into Life for the learning purpose. The narrower task here is to distinguish a visible contrast measurement from the coherence claim someone wants to make.
eduKateSG · Wave Optics
Find your next learning step
Recover a modelled coherence magnitude from original calibrated two-beam records, separating beam imbalance, background, detector response, phase coverage and uncertainty.
ROUTE 1 · CHAPTERS 1–5
Define the optical measurement
State the observable, field model, extrema and three contrast factors.
ROUTE 2 · CHAPTERS 6–10
Calibrate and test the inference
Construct equivalent fringes, recover individual contributions and check closure.
ROUTE 3 · CHAPTERS 11–15
Challenge measurement assumptions
Check incompatible amplitudes, background, gain, nonlinearity and polarisation.
ROUTE 4 · CHAPTERS 16–21
Preserve scope and transfer
Test phase variability, sampling, uncertainty, a changed record and sensitivity.
ROUTE 5 · CHAPTERS 22–26
Practise and continue
Separate phase from magnitude, report evidence and follow the existing optics guides.
Chapter contents
Define the optical measurement · Chapters 1–5
1. Define the observable before interpreting the cause
2. State the two-beam model explicitly
3. Connect the correlation parameter with fields
Calibrate and test the inference · Chapters 6–10
6. Work a calibrated fictional record
7. Show why extrema alone do not identify coherence
8. Add separate background and beam records
9. Recover coherence and verify the reconstruction
10. Test a closure relation before interpreting the parameter
Challenge measurement assumptions · Chapters 11–15
11. Recognise a physically incompatible recovered magnitude
12. Correct background and imbalance in the right account
13. Check gain and offset invariance
14. Distinguish detector nonlinearity from a common gain
15. Keep polarisation overlap separate from scalar coherence
Preserve scope and transfer · Chapters 16–21
16. Interpret phase variability without overdiagnosing its cause
17. Check whether the phase extrema were actually sampled
18. Carry stated intensity bounds through the inference
19. Preserve shared calibration constraints
Practise and continue · Chapters 22–26
22. Distinguish the correlation magnitude from its phase
23. Build a calibrated evidence note
24. Choose a practical learning route
CHAPTER 1 OF 26 · Define the optical measurement
1. Define the observable before interpreting the cause
Fringe visibility is conventionally calculated as V = (I_max − I_min)/(I_max + I_min). The two quantities must be comparable intensities, measured under conditions where the selected maximum and minimum belong to the same relevant interference modulation.
A visibility of zero means the chosen record has no measured modulation under that definition. A visibility near one means its minimum is small relative to its maximum. Neither statement, by itself, identifies every optical and instrumental condition responsible for the result.
For this guide, imagine changing the relative phase at one observation point while the beam contributions, background and coherence remain fixed. We use the true maximum and minimum over a complete phase cycle. This avoids confusing an intensity change across a diffraction envelope with a phase modulation at fixed conditions.
A spatial image can require a more elaborate account because different positions can have different beam strengths, backgrounds or overlap. Taking the brightest and darkest pixels anywhere in that image is not automatically the same measurement. Define the observable first; otherwise a correct fraction may be attached to the wrong physical model.
CHAPTER 2 OF 26 · Define the optical measurement
2. State the two-beam model explicitly
Let I₁ and I₂ be positive individual beam intensities at the same observation point. Let B be a nonnegative phase-independent background contribution in the same linear measurement scale. Let g be the magnitude of the relevant normalised field correlation, with 0 ≤ g ≤ 1.
Under the supplied scalar model with suitable matching polarisation, write I(φ) = I₁ + I₂ + B + 2√(I₁I₂)g cos φ. A fixed phase offset can be included in φ. It changes where a maximum occurs, not the complete-sweep extrema.
The model says that the two field contributions generate an interference term, while B adds without participating in that term. A third coherent field would not generally be represented by this simple background parameter. Its cross terms could alter the pattern.
We also require the detector’s reported scale to respond linearly and consistently to the relevant intensity. The original exercises supply that condition rather than infer it from a tidy-looking table. Beam strengths, background, polarisation and observation averaging must refer to the same conditions. Each requirement identifies something that a real measurement interpretation would need to check.
CHAPTER 3 OF 26 · Define the optical measurement
3. Connect the correlation parameter with fields
A compact mathematical description uses two complex field amplitudes u₁ and u₂. Their individual intensities are averages of |u₁|² and |u₂|² over the declared observation arrangement. Their averaged cross product contains information about their phase relationship.
Normalising the magnitude of that cross product by √(I₁I₂) gives g in the scalar account. The Cauchy–Schwarz inequality bounds it by one. A zero cross-product magnitude removes the average interference modulation; a magnitude at the upper bound permits the strongest modulation allowed by those beam intensities.
The averaging arrangement matters. A relationship can remain stable over one delay and observation window but not another. Treating g as an unrestricted quality label for a source discards that context. It describes the specified pair of fields under the stated conditions.
The derivation below does not need the full statistical machinery to be useful. Once the model is supplied, a learner can calculate its extrema and test what the measurements identify. Keeping the field definition visible prevents the parameter from being renamed contrast and assumed equal to a detector ratio before intensity balance and background have been considered.
Over a complete phase cycle, cos φ reaches +1 and −1. Define M = I₁ + I₂ + B and A = 2√(I₁I₂)g. Then I_max = M + A and I_min = M − A.
Conversely, M = (I_max + I_min)/2 and A = (I_max − I_min)/2. The first quantity is the midpoint of the modulation; the second is its amplitude. Neither is an individual beam intensity without further information.
Substituting into the visibility definition gives V = A/M = [2√(I₁I₂)/(I₁+I₂+B)]g. This equation states the inference problem clearly. The measured ratio contains g multiplied by another factor that depends on the individual beams and background.
The minimum remains nonnegative under the supplied conditions. Since 2√(I₁I₂) ≤ I₁+I₂ and g ≤ 1, the interference amplitude cannot exceed the two-beam sum, while B adds further nonnegative intensity. Nonnegative measured extrema alone, however, do not guarantee that a proposed set of individual beam measurements is compatible with those extrema. We will test that stronger consistency requirement later.
CHAPTER 5 OF 26 · Define the optical measurement
5. Separate three factors in recorded visibility
Write V = g × f_balance × f_background. Here f_balance = 2√(I₁I₂)/(I₁+I₂), and f_background = (I₁+I₂)/(I₁+I₂+B).
The balance factor lies between zero and one for positive beam intensities, reaching one when the beams match. The background factor is also at most one and reaches one when B = 0. Thus V ≤ g in this particular model.
Equality needs the relevant factors to reach one, apart from trivial zero-modulation cases. A statement that visibility directly measures coherence usually carries assumptions about balanced contributions and negligible or properly corrected background, in addition to the other model conditions.
The factorisation is explanatory rather than a universal correction recipe. Unknown polarisation mismatch, detector nonlinearity or additional coherent fields can invalidate the simple three-factor account. The right response to an unexplained contrast is to inspect the model and measurements, not to divide by guessed factors until a preferred coherence value appears. Every correction should correspond to independently supported information about the same observation conditions.
CHAPTER 6 OF 26 · Calibrate and test the inference
6. Work a calibrated fictional record
Supply individual intensities I₁ = 9 and I₂ = 1 in a common arbitrary linear scale. Let B = 2 and g = 0.60. These are original teaching inputs, not readings from an actual instrument.
The midpoint is M = 9+1+2 = 12. The modulation amplitude is A = 2√9 × 0.60 = 3.6. Therefore I_max = 15.6 and I_min = 8.4. Recorded visibility is 7.2/24 = 0.30.
The three factors explain the result: coherence magnitude 0.60, balance factor 6/10 = 0.60, and background factor 10/12 = 5/6. Their product is 0.30. A measured contrast of 0.30 has therefore arisen from a modelled correlation magnitude twice as large.
Check by reconstructing an intermediate phase. At φ = π/2 the cosine is zero and I = 12, the calculated midpoint. At φ = π the intensity is 8.4. These checks connect the extrema with the complete supplied sinusoid rather than treating the maximum and minimum as unrelated numbers chosen to make a formula work.
CHAPTER 7 OF 26 · Calibrate and test the inference
7. Show why extrema alone do not identify coherence
Consider a separate example with B = 2. In one model, I₁ = 9, I₂ = 1 and g = 1. The midpoint is 12 and the modulation amplitude is 6, giving extrema 18 and 6.
In another model, I₁ = I₂ = 5 and g = 0.60. The midpoint is again 12 and the modulation amplitude is again 6. The full intensity function over phase is identical, not merely its maximum and minimum.
The two models have different coherence magnitudes and beam balances. No increased precision in that same phase-dependent intensity record can distinguish them without additional information. More accurate measurement of an underidentified observable does not create a missing independent relationship.
This counterexample does not say coherence cannot be measured. It says that the available measurement must constrain the individual contributions or supply a justified balance condition. It also shows why a smooth, repeatable fringe trace is not automatically a complete diagnosis. A repeatable function can be compatible with more than one physical parameter combination under the same broad model.
CHAPTER 8 OF 26 · Calibrate and test the inference
8. Add separate background and beam records
Return to the first calibrated example. Suppose the fictional record supplies J₀ = 2 with neither beam contribution present, J₁ = 11 with only beam one present and J₂ = 3 with only beam two present.
Under the declared stable-background and unchanged-beam conditions, B = J₀, I₁ = J₁−J₀ = 9 and I₂ = J₂−J₀ = 1. The subtraction removes the common background from each individual-beam reading.
| Supplied record condition | Linear reading |
|---|---|
| Background only, J₀ | 2 |
| Beam one plus background, J₁ | 11 |
| Beam two plus background, J₂ | 3 |
| Both beams, complete-sweep maximum | 15.6 |
| Both beams, complete-sweep minimum | 8.4 |
These records add information that the full fringe trace alone lacked. Their interpretation still depends on the conditions matching: changing which beam is represented must not change the other beam’s strength, detector response or background in an unmodelled way. This is a supplied paper calibration account, not instructions to alter a live optical setup.
CHAPTER 9 OF 26 · Calibrate and test the inference
9. Recover coherence and verify the reconstruction
From the measured extrema, A = (15.6−8.4)/2 = 3.6. From the background-corrected individual records, √(I₁I₂) = √9 = 3. Therefore g = A/[2√(I₁I₂)] = 3.6/6 = 0.60.
In one line, g = (I_max−I_min)/[4√((J₁−J₀)(J₂−J₀))], provided the model holds and both corrected beam intensities are positive. The denominator uses individual contributions, not the total fringe midpoint.
Rebuild the forward record after inference. The recovered I₁, I₂, B and g give I(φ) = 12 + 3.6 cos φ, whose extrema match 15.6 and 8.4. Check both the modulation and midpoint, because fitting only one part can conceal a mismatch in the background or total intensity.
A good result names the quantity recovered: the magnitude of the specified normalised scalar correlation under the declared observation model. It does not establish the correlation’s phase, a complete spectrum, a coherence length or the reason that g is below one. Those are additional questions requiring different information.
CHAPTER 10 OF 26 · Calibrate and test the inference
10. Test a closure relation before interpreting the parameter
The background and individual-beam readings predict a two-beam midpoint J₁ + J₂ − J₀. The complete-sweep extrema predict a midpoint (I_max+I_min)/2. Under the supplied stable linear model, these two values must agree.
For the worked record, 11+3−2 = 12 and (15.6+8.4)/2 = 12. This closure check is independent of g because the interference term averages symmetrically between its extrema.
If the two midpoints disagree beyond the stated uncertainty, at least one condition of the account is incompatible with the record. Background drift, changed beam contributions, incomplete phase sampling, detector response or transcription could be involved. The discrepancy alone does not identify which explanation is correct.
Passing closure is useful but not a complete instrument certification. Several errors could compensate numerically. Treat it as one necessary consistency check within this model, alongside positivity, the g bound, measurement comparability and the actual method. A good audit reports what each check tests instead of describing one successful equality as proof that every hidden assumption is true.
CHAPTER 11 OF 26 · Challenge measurement assumptions
11. Recognise a physically incompatible recovered magnitude
Keep J₀ = 2, J₁ = 11 and J₂ = 3, but supply extrema 20 and 4. Their midpoint remains 12, so the closure relation passes. Their amplitude is eight.
The individual beam contributions still permit a maximum scalar modulation amplitude of 2√(9×1) = 6. Recovering g would give 8/6 = 4/3, exceeding the model’s bound of one. The record is incompatible with the full set of supplied conditions even though every individual intensity is nonnegative.
Do not clip the result to one and then report perfect coherence. That would hide the inconsistency instead of resolving it. Preserve the observed values, the calculated bound and the assumptions used. The next step is to investigate the model or records, not to manufacture a permitted answer.
This example demonstrates why several checks are needed. Closure tests the midpoint, while the correlation bound tests the modulation relative to the individual beam strengths. A trace can pass one and fail the other. The checks answer different questions and should remain separately visible in a reproducible explanation.
CHAPTER 12 OF 26 · Challenge measurement assumptions
12. Correct background and imbalance in the right account
In the calibrated example, the raw visibility is 0.30. Correcting only for beam imbalance would divide by the balance factor 0.60 and give 0.50. That still misses the supplied coherence magnitude because the background dilution remains.
Subtract the phase-independent B = 2 from both extrema first. They become 13.6 and 6.4. Their visibility is 7.2/20 = 0.36. Dividing by the same balance factor 0.60 then gives 0.60.
The direct amplitude-and-individual-beam formula reaches the same answer without treating each operation as an unexplained correction. It is often easier to audit because the amplitude is unchanged by a common additive background, while the denominator is built from the matching individual contributions.
The order of bookkeeping matters when the quantities are named correctly. Background affects the total-intensity denominator in visibility but not a stable complete-sweep maximum-minus-minimum difference. Beam imbalance limits the possible modulation for a given total beam intensity. Combining those distinct effects into a single contrast-loss label makes it harder to determine which additional measurement would actually resolve the inference.
CHAPTER 13 OF 26 · Challenge measurement assumptions
13. Check gain and offset invariance
Suppose every fictional detector reading is multiplied by the same positive gain k. Both the fringe difference and the square-root denominator in the recovered-g formula acquire that factor, so it cancels. A common linear change of scale does not change the inferred dimensionless magnitude.
A common additive offset also cancels from I_max−I_min and from J₁−J₀ and J₂−J₀. The calibrated inference can therefore remain invariant when that offset is applied consistently to every relevant record. Raw visibility generally changes because its denominator contains the sum of the extrema.
This is a useful algebraic test of the formula. It does not mean gain and offset are irrelevant in a real measurement. Saturation, changing settings, different offsets between records or background subtraction errors can break the required consistency.
For an independent exercise, multiply the original five readings by ten and then add seven to each. Recover I₁ and I₂ in the new detector scale, then calculate g. It should still be 0.60. Explain why the unchanged result follows from common calibration transformations rather than from a claim that the detector has measured the original physical intensity units directly.
CHAPTER 14 OF 26 · Challenge measurement assumptions
14. Distinguish detector nonlinearity from a common gain
A nonlinear response does not preserve intensity addition in the way the calibration model requires. Consider a changed teaching record with true linear-scale extrema eight and four, and suppose a fictional device reports the square of every intensity.
The reported extrema become 64 and 16. Raw visibility changes from (8−4)/(8+4) = 1/3 to (64−16)/(64+16) = 0.60. The optical inputs have not changed; the reporting function has.
This is a constructed mathematical counterexample, not a description of a particular camera. It shows why linearity is an assumption that needs a basis. More decimal places in a nonlinear reading do not restore the relationship between field contributions and the reported contrast.
If a response curve is independently known and appropriately invertible over the relevant range, an analysis may transform readings back into a suitable scale before applying the model. That introduces its own calibration and uncertainty questions. Simply treating any monotonic detector output as intensity is insufficient: monotonicity preserves the order of maximum and minimum, but not their contrast ratio or the additive closure relation.
CHAPTER 15 OF 26 · Challenge measurement assumptions
15. Keep polarisation overlap separate from scalar coherence
The simple formula assumes suitable matching polarisation. If two otherwise related fields have different polarisation directions, their interference term depends on their overlap. For ideal linear polarisation directions separated by angle α, a scalar cross term acquires a factor related to cos α.
Orthogonal directions can therefore remove the observed modulation even when a scalar phase relationship is stable. A low-contrast record cannot automatically be attributed to temporal phase variability without checking that the assumed polarisation model applies.
MIT’s optics demonstrations separately examine intensity imbalance and polarisation difference. The separation is useful because changing either condition can affect contrast through a different mechanism.
For a paper task, hold the scalar correlation magnitude at one and compare matched directions with orthogonal directions. Ask what the observed absence of fringes identifies. It establishes the missing modulation under those conditions; it does not uniquely diagnose a source’s coherence in every arrangement. The calibrated g derived earlier should be interpreted as the parameter of its stated scalar, matching-polarisation model, rather than silently extended to an unknown vector-field configuration.
CHAPTER 16 OF 26 · Preserve scope and transfer
16. Interpret phase variability without overdiagnosing its cause
Averaging over changing relative phases can reduce the interference term. For an original discrete example, suppose an additional phase offset is equally likely to be +α or −α during the relevant averaging arrangement. The average complex phase factor is cos α.
Its magnitude is |cos α|. At α = 60 degrees, that magnitude is one half. The averaged modulation can therefore be reduced even if the instantaneous field contributions retain their individual strengths.
This supplied phase model illustrates one route to a reduced correlation magnitude. It does not prove that phase jitter caused a particular measured g below one. Other temporal, spatial or measurement conditions can affect the relevant correlation or violate the simple observation model.
A useful inference statement stops at what the calibrated record supports. It can report the recovered magnitude and the conditions under which it was obtained. To distinguish mechanisms, design an additional controlled comparison that changes a relevant variable while preserving the others. Repeating the same contrast measurement many times can improve precision but cannot, on its own, turn a shared observable into a unique causal diagnosis.
CHAPTER 17 OF 26 · Preserve scope and transfer
17. Check whether the phase extrema were actually sampled
For I(φ) = 12 + 3.6 cos φ, suppose the fictional record includes only φ = 0 and φ = π/2. The readings are 15.6 and 12. The second reading is the midpoint, not the full-cycle minimum of 8.4.
Calling those two readings maximum and minimum would give an apparent visibility of 3.6/27.6, approximately 0.1304. Their midpoint would also be 13.8, inconsistent with the calibrated midpoint of 12. A sampling limitation has changed the interpretation before any coherence calculation begins.
A sampled intensity range cannot exceed the true complete-cycle range under the unchanged model. With independently known individual beam intensities, it can provide a lower bound on the required modulation magnitude. It does not establish that the observed endpoints are the extrema.
The required phase coverage and measurement design belong to the actual method, not to a generic instruction to take more photographs. In this guide the complete-sweep condition is supplied explicitly for the main examples. In a real record, verify that condition or describe the narrower result supported by the sampled phases.
CHAPTER 18 OF 26 · Preserve scope and transfer
18. Carry stated intensity bounds through the inference
Suppose the individual corrected beam intensities are supplied exactly as nine and one, while the main example’s maximum and minimum each have a hard reading bound of ±0.2. The central difference is 7.2, and its conservative interval is [6.8,7.6].
The coherence denominator is twelve, so the resulting conservative magnitude interval is [6.8/12,7.6/12], approximately [0.5667,0.6333]. These are deterministic bounds under the stated conditions, not confidence limits inferred from a probability distribution.
The midpoint check also needs its uncertainty account. The extrema’s midpoint is centrally twelve with a conservative bound of ±0.2, compatible with the exactly supplied beam-and-background midpoint of twelve.
If beam calibration is uncertain too, its errors must enter the denominator and any shared dependencies must remain visible. Dividing unrelated worst cases can give a conservative result, but it should not be called the exact attainable interval without checking joint feasibility. An uncertainty calculation must obey the same physical and calibration relationships as the central reconstruction. More elaborate arithmetic is not permission to treat every input as independently adjustable.
CHAPTER 19 OF 26 · Preserve scope and transfer
19. Preserve shared calibration constraints
Imagine J₁ = 11, J₂ = 3 and the fringe midpoint M = 12 are exact in a new exercise, while the background-only record permits J₀ anywhere in [1.8,2.2]. The closure relation requires J₀ = J₁+J₂−M = 2.
Under the unchanged exact model, the other records narrow the allowed background to one value. It would be inconsistent to vary J₀ freely across its whole interval while simultaneously holding every exact midpoint relationship unchanged.
This example teaches a general habit: uncertainty intervals describe permitted inputs, but the model can add constraints linking them. The feasible combinations are the intersection of both. A list of independent-looking ranges is not necessarily the final set of possible records.
If the intervals fail to intersect at all, report the incompatibility rather than widen a convenient range silently. If the measurements are probabilistic or noisy rather than exact, use an appropriate statistical model and retain the distinction. The paper exercise isolates the logic of shared calibration; it does not prescribe how an actual instrument should reconcile discrepant observations or determine which record deserves more weight.
CHAPTER 20 OF 26 · Preserve scope and transfer
20. Solve an independent changed record
A new model has I₁ = 4, I₂ = 1, B = 1 and g = 0.50. Predict the midpoint and modulation amplitude before calculating the extrema.
The midpoint is six. The amplitude is 2√4 × 0.50 = 2. Thus the complete-sweep extrema are eight and four, with raw visibility 1/3. Background-only and individual-beam readings are J₀ = 1, J₁ = 5 and J₂ = 2.
Recovering the parameter gives g = (8−4)/[4√((5−1)(2−1))] = 4/8 = 0.50. Closure holds because (8+4)/2 = 5+2−1 = 6. The recovered value also respects the model bound.
Explain the result through the three factors: g = 0.50, balance factor 4/5 and background factor 5/6. Their product is 1/3. Then change only B to zero and predict what happens. The amplitude stays two, the midpoint falls to five and raw visibility rises to 0.40, even though the supplied correlation magnitude has not changed. That changed condition is the transfer check, not another substitution into an unchanged fraction.
CHAPTER 21 OF 26 · Preserve scope and transfer
21. Notice when a weak beam makes the inference sensitive
The recovered magnitude is g = A/[2√(I₁I₂)]. Holding the independently supplied beam intensities fixed, a small amplitude change ΔA produces a change Δg = ΔA/[2√(I₁I₂)]. A small denominator makes that inference sensitive.
For I₁ = 9 and I₂ = 1, the denominator is six. An amplitude uncertainty of 0.1 corresponds to a magnitude uncertainty of about 0.0167 under those exact beam assumptions. If I₂ instead equals 0.01, the denominator is 0.6 and the same amplitude uncertainty corresponds to about 0.1667.
The arithmetic does not recommend changing a real apparatus. It explains why weak individual contributions require an appropriate measurement and calibration account. Uncertainty in subtracting background can also be consequential when a corrected beam intensity is small.
At I₂ = 0, this normalisation is undefined and the two-beam interference amplitude vanishes. Absence of a fringe does not supply a measured zero correlation magnitude for a missing field. The distinction between a small identifiable parameter and an undefined inference matters. A report should acknowledge the denominator’s conditions before presenting a stable-looking percentage derived from nearly cancelling calibration readings.
CHAPTER 22 OF 26 · Practise and continue
22. Distinguish the correlation magnitude from its phase
The cross correlation is generally complex. Write it as a magnitude times a phase factor. In the intensity model, that phase can shift the argument of the cosine, producing I(φ) = M + A cos(φ+θ) for a fixed offset θ.
Changing θ moves the phase at which a maximum occurs, but leaves the complete-cycle maximum, minimum, midpoint and visibility unchanged. The five-record calibration used here therefore recovers the magnitude parameter without recovering this phase offset from the extrema alone.
For an original comparison, use the first calibrated model with θ = 0 and then θ = π/2. Both have extrema 15.6 and 8.4, yet their readings at the same supplied phase φ = 0 are 15.6 and twelve respectively. A phase-referenced intermediate measurement can distinguish them when its reference is known.
This is a different missing variable from beam imbalance. Once I₁, I₂ and B are known, the extrema determine g under the model, but they still discard where the extrema occurred. The existing interferometry guide develops phase measurement and unwrapping. Keep that route separate from a claim that calibrated visibility reconstructs every feature of the underlying field relationship. Naming the recovered quantity precisely also names the information a different measurement must supply.
Name the observation point or region, phase variable, observation window and measurement scale. State whether the extrema come from a complete relevant sweep and whether individual beam contributions are assumed stable across the separate records.
Keep the five supplied readings together. Identify the background, two corrected beam intensities, fringe midpoint, modulation amplitude and inferred g. Show positivity, closure, the correlation bound and a forward reconstruction as separate checks.
Then record limits. Which polarisation and linearity assumptions apply? Does background remain phase independent? Could an additional coherent contribution exist? Are the uncertainty statements hard bounds or probability descriptions? What information would distinguish competing mechanisms if the recovered magnitude is below one?
The resulting note can be short because the reasoning has been made explicit. It should let another reader rebuild the conclusion without guessing what the word intensity means or why one reading was subtracted from another. A clear limited result is more useful than a single coherence percentage whose calibration, scope and observation conditions have disappeared from view.
For repair, distinguish midpoint from amplitude. Use a supplied sinusoid and read its maximum, minimum and zero-cosine level. Calculate visibility without yet interpreting its cause.
For stabilisation, compare the two models with identical 18-and-6 extrema. Ask why the same full fringe trace can conceal different coherence magnitudes. Add the separate beam and background records, then recover the first calibrated example with all four checks.
For extension, change background alone, introduce an incompatible modulation, compare common gain with nonlinear response and preserve a shared uncertainty constraint. The learner should identify which assumption changed before choosing an equation.
All of this can be done on paper or with a spreadsheet using the fictional records. No optical hardware is required. A completed explanation connects the observed contrast with the calibrated inference and states one quantity still unresolved. The useful learning outcome is the ability to choose the missing measurement in a changed case, rather than treating every weak fringe pattern as proof of the same source problem.
CHAPTER 25 OF 26 · Practise and continue
25. Questions about visibility and coherence
Does visibility always equal coherence magnitude? Only under suitable additional conditions, including the relevant balance and background assumptions in this scalar model.
Can unequal beams produce incomplete darkness despite full coherence? Yes. The two examples with identical extrema show how beam imbalance and reduced correlation can produce the same modulation.
Does subtracting background solve every contrast problem? No. Beam balance, polarisation, detector response, phase coverage and model scope remain separate conditions.
What does a recovered value above one mean? It is incompatible with the supplied model and records. Clipping it to one hides the discrepancy.
Can this recover the cause of reduced coherence? Not from these records alone. The calibrated magnitude is a more precise target than a complete causal explanation.
What is the best first check? Define which maximum and minimum were measured, then test midpoint closure against the matching individual-beam and background records before interpreting a parameter.
Check the optical mechanisms
MIT: Fringe Contrast and Intensity Ratio demonstrates the effect of unequal beam strengths. MIT: Fringe Contrast and Polarisation Difference provides a separate polarisation route. Sources were checked on 23 October 2026. The calibration records and reconstructions here are original paper examples, not instrument benchmarks.
Follow the existing specialist questions
Optical Interferometry, Fringe Counting and Phase Unwrapping already develops basic visibility and phase measurement. Finding Missing Orders concerns finite-aperture order coincidences. A faint or absent feature in a spatial pattern is not automatically the calibrated complete-sweep measurement used here.
Continue from model to evidence
Return to Wave Optics and Optics. Optical Engineering develops performance in real conditions. Continue through How X Works and Learning That Transfers into Life. Explain which separate record would distinguish the two identical-fringe examples; that is the final check that the measurement question has become clear.