On a bright afternoon in Clementi, a Secondary 1 Science class measures the light falling on a study table. The first group places a small light sensor flat on the desk and records a reading. The next pupil picks up the same sensor, tilts it towards a classroom window, and gets a different number. Nobody switched off the light. Nobody moved the window. The class wonders whether the instrument is unreliable—or whether one student has somehow measured the brightness of a different room.
A lux meter, illuminance sensor, or light-sensing phone app does not simply report an abstract quantity called “how bright the room is.” It responds to light arriving at its detector, and the measurement depends on the detector’s orientation, the directions from which light arrives, the light’s spectrum, the instrument’s optical design, and the measurement conditions. Tilt the receiving surface and you may change the amount of light intercepted per unit area, even when the source remains unchanged. With a suitable sensor, this behaviour is part of the physical quantity being measured, not necessarily a fault.
It is a wonderful classroom question because a small movement reveals a bigger principle: a number belongs to a measurement procedure. To compare light levels, we must specify what surface, what direction, what height, which instrument, and when the reading was taken. Once those details are controlled, the confusing numbers become evidence about geometry, measurement and how fair scientific comparisons work.
The 50-second answer
A light meter measuring illuminance reports how much visible light arrives on a surface per unit area, commonly in lux (lx). For a single directional source under appropriate ideal assumptions, illuminance on a surface changes with the cosine of the angle between the incoming light and the surface’s perpendicular direction. A surface facing the light receives more per unit area than the same surface tilted away from it. A properly designed illuminance meter aims to respond to this geometry correctly, although real sensors can have directional and spectral errors. Rooms with windows, ceiling lights, shadows and reflections do not follow the simple one-lamp example exactly. To compare readings, keep the sensor at the same place and orientation, record the angle, avoid shadows, and repeat measurements. A change after tilting may be a genuine change in what the sensor is measuring.
Quick route: A student can test flat-versus-tilted readings with a safe lamp. A parent can ask what the sensor was facing rather than compare two unexplained numbers. A teacher can turn the activity into a controlled investigation of orientation, light sources and measurement uncertainty.
What does a lux meter actually measure?
The name of the unit is a useful starting point. Illuminance describes visible-light flux incident on a surface per unit area. Lux is the SI derived unit: one lumen per square metre. The quantity depends on the receiving surface and on how the incoming light is distributed. In a classroom, illuminance may be considered on a desk plane, a book page, a whiteboard, or a sensor’s own receiving face. These are not automatically equivalent surfaces. A desk lies approximately horizontal; a whiteboard is vertical; a window may be at an angle; a student may hold a device upright while another lays it flat.
Luminous flux, measured in lumens, refers to the quantity of visible light flow weighted by human visual sensitivity. Luminous intensity, measured in candelas, concerns luminous flux per unit solid angle in a specified direction. Luminance concerns light emitted or reflected per unit projected area in a direction, and relates more closely to how bright a surface appears from a particular viewpoint. They are related but not interchangeable. A lux reading at a desk should not be labelled “the lumens of the lamp,” and a phone’s light-sensor reading is not a complete evaluation of how comfortable a room is to read in.
Why does the distinction matter for students? Suppose a shop sells two lamps with the same advertised lumens. Their light distributions can differ: one spreads light broadly, another concentrates it in a narrower beam. At a particular point on a table, the measured illuminance can differ even if their total luminous flux is similar. Likewise, a pupil turning a sensor toward a bright window changes the directions and proportions of incident light that reach its receiving surface. The total lighting installation has not necessarily changed, but the measured condition has.
The Illuminating Engineering Society defines a lux meter as a device for measuring illuminance on a plane. Its technical glossary distinguishes cosine correction, which concerns directional response, from colour correction, which concerns spectral response. This matters because a detector may respond to different wavelengths differently from the standard model of human daytime visual sensitivity. A bare photodiode and a calibrated illuminance meter may report values that diverge under certain lamps, even when aimed in the same direction. The question becomes not only “Where is the device?” but “What is this particular device built and calibrated to measure?”
The geometry: the same light spreads across a different area
Imagine a small bundle of light rays approaching a flat card. When the card faces the rays directly, that bundle lands on a relatively compact region. Rotate the card while leaving the incoming rays unchanged. The same projected bundle now spreads across a larger patch of the card. Per unit of the card’s actual surface area, the incident light is reduced. The relationship for an ideal directional source is described by a cosine factor. The angle is measured between the direction of incoming light and a line perpendicular to the receiving surface—its normal—not necessarily between the beam and the plane itself.
This definition prevents a common geometry error. If a pupil says that the sensor is “tilted by sixty degrees,” ask: sixty degrees relative to what? A sensor tilted sixty degrees from the direction in which it was directly facing a lamp is one case. A surface making sixty degrees with a light ray corresponds to an angle of thirty degrees between that ray and the surface normal. Draw a right-angle marker on the surface and a perpendicular arrow before applying the cosine rule. Clear diagrams are useful because the wrong angle can give a mathematically neat but physically incorrect result.
In an idealised one-source example with fixed distance, suppose the sensor receives 600 lux when its normal points straight toward the incoming beam. At an incidence angle of 60 degrees, the cosine is 0.5, so the ideal illuminance attributable to that beam on the tilted plane is approximately 300 lux. At 30 degrees, the cosine is about 0.866, giving about 520 lux. These values are worked illustrations, not predictions for a random phone or real classroom. They assume suitable directional geometry, an ideal cosine response, and no changing contribution from other lights or reflected surfaces.
At an incidence angle approaching ninety degrees, a perfectly flat, opaque plane oriented edge-on to a collimated beam receives an ideal contribution approaching zero on that face. Real devices may still display a nonzero value because of ambient light, optical housing, reflections, stray light, or other measurement imperfections. If the light comes from several directions at once, contributions from many angles combine. A room illuminated by windows and ceiling fixtures is more complicated than a single directional ray. The cosine model is a principle for understanding each contribution, not a universal conversion factor to apply blindly to every tilt result.
The physics has a useful everyday analogue. Solar panels, windows, and tilted roofs respond differently to light arriving at different angles. A child can see why a sheet facing a flashlight receives a concentrated patch, whereas a strongly tilted sheet receives an elongated one. But an analogue has limits: electrical output from a solar panel also depends on spectral response, temperature, electronics, shading and other factors. For the classroom lux experiment, keep the target on the measured illuminance and the controlled geometry. Do not smuggle unrelated effects into the conclusion.
Cosine correction is not a trick that makes orientation irrelevant
Here is a subtle point that often puzzles careful pupils. Manufacturers may describe a light meter as cosine corrected. Does that mean the device corrects away the effect of tilting and therefore displays the same number regardless of orientation? No. A well-designed illuminance meter seeks the correct dependence on incoming direction for the surface being measured. In simplified terms, its response should decrease with the cosine of incidence for an appropriate directional source, matching the illuminance on the meter’s receiving plane. It is not attempting to measure a room-wide, orientation-independent “brightness.”
The distinction separates physical geometry from instrument error. The geometry of an ideal surface requires a cosine relationship. A detector whose angular sensitivity departs from that ideal has directional-response error. A properly corrected instrument tries to reduce that error. NIST literature discusses directional response and how actual instruments can deviate from the ideal function. The lesson for students is that correction supports faithful measurement of the specified quantity; it does not abolish the way the quantity depends on the measurement plane.
Suppose two identical calibrated meters are placed at the same desk location, but one faces the ceiling and one faces a window. They are measuring illuminance on different oriented planes. Their different outputs need not contradict each other. By contrast, if two meters occupy the same place, face the same direction, and operate under steady conditions, a substantial disagreement raises questions about calibration, sensor characteristics, spectral response, range, timing, and instrument condition. In the first situation the measurands differ; in the second the measurands should be closer to the same and the instruments warrant comparison.
A smartphone app adds another uncertainty. Some phones have an ambient light sensor designed primarily to support screen-brightness adjustments rather than certified laboratory illuminance readings. The sensor may be located near a screen edge or hidden behind glass; its effective field of view, optical filtering and update behaviour depend on the device. An app cannot make hardware performance identical across phones. For an instructional demonstration, the device can still show directional variation, but an uncalibrated app should not be presented as a traceable reference instrument or used to certify compliance with a lighting standard.
Orientation is only one part of a fair light comparison
What if the pupil kept the sensor flat but moved it ten centimetres nearer the lamp? The reading might change because the source-to-detector distance changed. For an ideal point source in unobstructed space, the inverse-square law relates illuminance to distance: with geometry held fixed, doubling the distance can reduce the contribution to roughly one quarter. But ordinary desk lamps are extended sources at nearby distances, have reflectors and shades, and may be supplemented by other lights. The simple law becomes an approximation whose conditions must be examined.
What if a student’s hand moved above the sensor? The hand may cast a shadow or reflect additional light. What if the sun moved behind a cloud? Daylight contribution can change during the experiment. What if a glossy white sheet was placed nearby? Reflected light may alter the reading. What if a fluorescent or LED source flickers and the sensor updates intermittently? The meter’s sampling behaviour may influence the displayed values. Each factor can produce a plausible difference, and none can be separated from orientation merely by looking at one pair of numbers.
The experiment therefore needs a stable measurement location, reference direction, consistent source conditions, and repeated readings. Mark the position of the sensor’s centre. Mark the plane of its receiving surface or the device’s intended sensing face. Keep the same distance from a lamp while changing only orientation, ideally using a simple nonreflective support or protractor. Check that the support does not obstruct the incoming beam. Write down whether window light, room lights, or both were present. A note such as “500 lx on desk” is incomplete if another pupil must reproduce the setup.
Units alone cannot rescue poor measurement conditions. A precise-looking number such as 527.6 lx may be less trustworthy than a rounded 530 lx if the device is uncalibrated or the reading fluctuates. Significant figures should reflect justified precision, not the number of digits an app happens to display. A class can record a range or repeated readings and describe what varied. This teaches scientific caution: the quantity is physical, the measurement is conditional, and the confidence should reflect the method.
A reproducible Clementi classroom investigation
The following is a classroom investigation design, not a report of an experiment already conducted. Use a low-power, cool-running lamp approved by the teacher, a lux meter or suitable educational sensor, a stable support, a protractor, a ruler and a recording sheet. Do not use a laser, stare into a bright source, place electrical equipment near water, or modify mains-powered fixtures. Use adult supervision and the school’s normal laboratory rules. The aim is to study how sensor orientation affects measured illuminance—not to test how close a lamp can safely be placed to a child’s face.
Set the lamp in a fixed position and let its output stabilise if the device requires it. Reduce uncontrolled daylight where practical or record the ambient contribution separately. Put the sensor at a marked location. Identify the detector’s actual receiving face rather than assuming it is the phone’s display or the top of its case. Establish a reference direction in which the receiving face points toward the lamp. Define that as zero-degree incidence for the simple example. If the sensor is mounted on a rotating holder, make sure its centre does not travel nearer to or farther from the source while the angle changes.
Take three readings at the reference orientation, allowing the displayed value to settle according to the device’s documented behaviour. Rotate to a chosen angle such as 15°, 30°, 45° or 60° and repeat. Return to zero degrees between trials to see whether the baseline has drifted. If the zero-degree reading changes substantially, investigate daylight, source stability, moving shadows, sensor timing or position before interpreting an angle trend. Mark any reading affected by a passing person or other disruption rather than silently keeping it as a valid observation.
Record a table with angle, trial one, trial two, trial three, the mean for teaching purposes if appropriate, and a note about the environment. Do not calculate a mean from numbers taken under obviously different conditions simply to create a tidy column. Make a graph of reading against incidence angle, label lux correctly, and compare the shape with the expected qualitative decline for an ideal single-source model. The purpose is not to manufacture perfect cosine values. Real readings help pupils notice where a simple model works reasonably and where the instrument and room may add effects.
Consider an invented example using one lamp and a suitable meter. At 0° the readings are 596, 603 and 601 lx; at 30° they are 513, 520 and 517 lx; at 60° they are 292, 304 and 299 lx. The means are approximately 600, 517 and 298 lx. The values roughly follow a cosine-shaped decline and are close to the ideal illustration of 600, 520 and 300 lx. These made-up numbers demonstrate analysis, not a claimed successful classroom study. A genuine class should use the measurements it obtains and report departures from the model without erasing them.
Now add a second light source or reopen a window curtain, while keeping the sensor in the same position. Repeat a limited set of angles. The new pattern may differ from the first because the incoming light now arrives from more than one direction. A pupil might see a reading rise when tilting towards the window, even though the reference lamp’s own contribution declines. That does not violate the cosine law. It means the total reading is influenced by multiple components with different directions and perhaps different spectral qualities. The more realistic environment changes the problem being measured.
For older students, the discussion can separate controlled variables, dependent variables, and sources of uncertainty. The changed variable is orientation. The dependent variable is the reported illuminance. Potential controls include distance, lamp output, room lighting, sensor model, measurement point and reading interval. Not all variables can be held perfectly fixed. A good report identifies the most consequential uncontrolled factors rather than claiming perfect experimental isolation. This is how a short Science practical becomes a lesson about experimental design rather than a demonstration staged to agree with a formula.
How the mathematics supports the experiment
For an approximately directional source, write the idealised relationship as E(θ) = E(0) × cos θ. Here E(θ) is illuminance on the receiving plane at incidence angle θ; E(0) is the reference value when the plane faces the incoming direction. This expression assumes the source geometry and distance do not change and excludes other light contributions. It is useful for teaching proportionality. At 0°, the cosine is one, so the ideal reading equals the reference. At 60°, the cosine is one half, so the expected contribution halves. At 90°, the ideal contribution to the plane approaches zero.
An important caution: do not plug the physical tilt of the phone into the equation unless that tilt equals the incidence angle between incoming light and the detector’s surface normal. A ceiling lamp located above and to one side already has an initial angle of incidence, even when the sensor is horizontal. Rotating the device by 30° does not necessarily change the incidence from 0° to 30°. Students should draw a line from the light source toward the sensor and another line perpendicular to the sensing surface. The angle between these lines is the one relevant to the simplest model.
The law for a point source at distance d is often expressed, under appropriate assumptions, as E = I cos θ / d², where I is luminous intensity in the relevant direction. This combines inverse-square and cosine effects. It does not mean a teacher should expect exact inverse-square behaviour from a nearby fluorescent panel or broad window. An extended source contains many points at different distances and directions. The pupil who notices that the model breaks down at very short distances is thinking scientifically, provided they can explain the model’s assumptions and inspect possible confounding effects.
The graph is another opportunity to teach precision. Plot θ on the horizontal axis with degrees specified, and illuminance on the vertical axis with lux specified. If the reading is normalised by dividing each measurement by the reference, label the new vertical axis as a ratio rather than lux. Compare the measured ratio at 60° with 0.5 and discuss differences. But be careful with arithmetic when the reference itself fluctuates; an unstable denominator can exaggerate apparent departures. For an introductory lesson, correctly labelled raw readings and a qualitative comparison may teach more than a forced calculation with false precision.
A student who understands this mathematics can apply it beyond the sensor. When light strikes a surface obliquely, the same incoming beam contributes less light per unit actual surface area. That relationship helps explain why the angle of illumination matters in photography, solar-energy applications and building lighting, although each practical field has additional variables. The bridge from a classroom measurement to a broader physical principle is strongest when the child knows both the connection and the limits of the analogy.
The school desk and the whiteboard are not the same measurement plane
Imagine a parent asks whether a child has enough light for homework. Someone places a meter flat on the desk and obtains one value. Another holds it facing the nearby lamp and obtains a higher value. Which one measures the child’s reading conditions? There is no universal answer from those values alone. If the question concerns illuminance on the page, the sensor should be oriented in an appropriate approximation to the page plane at the relevant location. If the question concerns a vertical noticeboard, a vertical measurement plane may be more relevant. In both cases, actual visibility also depends on contrast, glare, shadows, print size and the reader’s needs.
A phone app’s changing number may seem to suggest that moving a book is unnecessary because the room is objectively bright. Yet a glossy worksheet can catch glare from an overhead lamp, making text uncomfortable to read despite substantial incident light. A child’s body may shade the page when they lean forward. Lighting quality is not one scalar number. Illuminance helps quantify one part of a visual task, but it should be interpreted alongside the nature of the page, the location of the lamp, and the comfort and accessibility needs of the learner.
An explanation of illuminance is not a lighting inspection or a medical recommendation. Classrooms and workplaces may have formal design standards that specify measurement conditions and acceptable levels. Such assessments require the applicable standard and suitable instruments. The classroom activity here is about understanding why tilt changes a measured quantity, not certifying that a room meets regulatory requirements or diagnosing eye strain. If there is a genuine lighting concern, consult the school and an appropriately qualified professional.
A pupil route: the five-question check
When a light reading surprises you, ask what is the sensor measuring? If it measures illuminance, identify the receiving plane. Second, ask which direction is the sensor facing? Draw the surface normal and the important source directions. Third, ask what else changed? Check distance, shadows, nearby reflective objects and other light sources. Fourth, ask how trustworthy is the instrument for this purpose? A school lux meter, a general light sensor and a phone app may have different specifications. Fifth, ask what claim the data actually supports? Different readings with different orientations do not by themselves show that the lamp became brighter or dimmer.
Write a scientific conclusion in that order. For example: “When we tilted the receiving face away from the lamp, the measured illuminance decreased in our setup. This is consistent with the incident-angle model. The result does not prove that the lamp’s total luminous output changed. The readings may also be influenced by ambient light and sensor response.” This is more informative than “Tilting makes light weaker,” which confuses the interaction at a surface with a change in the source itself.
If you have a surprising result, keep it. Perhaps the value increases at one angle because the sensor has turned toward a window or because a nearby surface reflects light onto it. Write a possible mechanism and design a check that distinguishes it from another explanation. Close the curtain or move the reflective card while keeping orientation stable, if doing so is safe and permitted. Scientific reasoning often begins when real observations refuse to match an oversimplified drawing. Your goal is to improve the account of what happened, not to protect the first guess.
A parent route: useful questions without special equipment
You can explore the concept with a book and an ordinary torch in a safe setting. Shine the torch at a sheet of paper from a fixed direction, then gently tilt the sheet while keeping the beam aimed at it. Notice how the illuminated patch changes its shape and area. Ask whether the same beam is distributed across more or less of the surface. The point is to see why direction matters. This visual demonstration is not an accurate lux measurement, and it should not be treated as one.
Ask the child to describe the difference between “the lamp became less bright” and “less light landed on each square centimetre of the tilted page.” The first sentence concerns the source. The second concerns the receiving surface. That linguistic distinction is the heart of the mechanism. If the child can maintain it while describing several examples, they have gained a transferable reasoning skill useful in Science and Mathematics.
For homework, notice shadows and glare rather than chasing a magic number from a phone app. A tidy, comfortably lit study position with readable materials supports the practical aim. If the child is struggling to see or read, the cause may involve vision, accessibility, the page’s print quality, or a learning issue unrelated to the meter. Do not use a homemade lux test to rule those concerns in or out. Seek appropriate advice when necessary.
A teacher route: make the measurement contract visible
Before the practical, write a sentence on the board: “We will measure illuminance on the sensor’s receiving plane, at this position, under these lighting conditions.” Fill in the position and conditions with the class. This sentence acts as a measurement contract. It prevents students from treating readings obtained under different definitions as if they were repetitions of the same measurement. It also creates a natural reason to teach fair testing rather than merely reciting a list of controlled variables.
Use paired diagrams to distinguish the surface normal from the surface itself. Have students point to the angle of incidence before introducing a cosine calculation. If some pupils are not yet ready for trigonometry, the qualitative question is sufficient: does the same beam cover more of the tilted surface? Older students can compare measured ratios with expected values and discuss uncertainty. Differentiation need not split the class into unrelated tasks; everyone can investigate the same phenomenon at a level appropriate to their mathematical readiness.
End with a transfer question. “Two groups report different lux values at the same point in the room. What additional information do you need before deciding that one group made a mistake?” A strong answer asks about sensor orientation, measurement plane, instrument specifications, timing and nearby shadows. This is formative assessment of measurement literacy. The pupil who requests missing information is demonstrating a habit central to experimental Science: a numerical claim is incomplete until its conditions are known.
Where the ideal model can fail
The cosine law concerns ideal geometric incidence on a plane. Real devices are not perfectly cosine corrected, particularly at large angles. The detector may be recessed, shaded by its housing, or influenced by light reflecting inside the instrument. Its spectral correction may differ from the standard visual-response function; this can matter under sources with different spectral distributions. Temperature, electronic noise, sampling intervals, drift and sensor calibration may also influence readings. These issues do not invalidate the geometric principle. They explain why experiments and metrology require careful instrument characterisation.
A classroom comparison is also affected by the environment. Sunlight varies, overhead fixtures contribute from different directions, and reflected light arrives from walls, desks and pupils’ clothing. A simple cosine curve is most credible when the source is approximately directional and other contributions are small or controlled. Under diffuse lighting from many directions, the relationship between device tilt and total illuminance is more complex. A measured increase at a certain orientation may be entirely legitimate because that orientation receives more light from a different source direction.
The word lux itself does not guarantee a high-quality reading. Two apps showing identical units may rely on different hardware and software, and the accuracy claim of a professional reference meter should not be assigned to an uncalibrated smartphone. When comparing readings from different devices, note which instrument was used and check its specifications where possible. For pupils, a consistent educational sensor and repeatable method are more useful than a screen full of unexplained digits.
Frequently asked questions
Why does my light sensor give different lux readings when I turn it?
Illuminance is defined for a receiving plane. Turning the sensor changes that plane’s orientation to incoming light. In a room with several sources, the contributions can change in complicated ways. A correctly designed sensor is expected to respond to the incident geometry, not display a fixed “room brightness” number at all angles.
Should a lux meter face the light or lie flat?
It should be oriented to measure illuminance on the plane relevant to the question. For a horizontal desk surface, that often means a horizontal receiving plane. For a vertical surface, the plane differs. Follow the instrument’s instructions and the appropriate measurement protocol; there is no universal orientation independent of purpose.
What is cosine correction?
It describes how an illuminance meter is designed to respond accurately to light arriving from different angles in accordance with the ideal cosine dependence. It does not remove the actual physical change in illuminance caused by rotating the receiving plane. Real meters can depart from the ideal response, and accuracy varies.
Can a phone measure lux accurately?
Some phones and apps can display useful approximate light information, but the hardware and calibration may not satisfy the requirements of a professional illuminance meter. Use a suitably specified instrument when the outcome requires documented accuracy. For a school demonstration of directional change, label the limitations clearly.
Does twice the lux mean a lamp is twice as bright to my eyes?
Not necessarily. Lux measures illuminance incident on a plane, while perceived visual brightness and comfort depend on other factors including source distribution, reflectance, glare, adaptation and the viewing task. Do not confuse the meter’s unit with a complete perceptual judgement.
Why didn’t my measured values follow the cosine formula?
Check angle definition, movement of the detector centre, other light sources, shadows, reflections, source geometry, device response, and measurement timing. A single ideal-source model is not expected to predict every real classroom setup exactly. The mismatch may reveal which assumption was not satisfied.
Does the inverse-square law apply to every classroom lamp?
It is an appropriate ideal relationship for a point-like source at distances where source size can be neglected and relevant conditions are controlled. Nearby extended fixtures and mixed indoor lighting require more careful treatment. Do not expect an exact quartering of lux whenever you move a school meter twice as far from a lamp.
Can students use this experiment to certify classroom lighting?
No. The exercise teaches measurement principles. Formal lighting assessment requires the applicable standard, defined measurement locations and planes, suitable calibrated instrumentation and competent interpretation. A classroom graph should not be represented as a professional certification.
The takeaway: say what plane the number belongs to
When two light readings disagree, the useful response is not automatically “the sensor is broken.” Ask what surface received the light, how that surface was oriented, which sources contributed, and what instrument did the measuring. At a Clementi desk, this turns a puzzling change in digits into a precise Science question. Students learn to connect geometry with observation, to test a model under stated conditions, and to describe uncertainty without embarrassment. That habit makes the next experiment stronger than the last.
Sources and further reading
- Illuminating Engineering Society — Illuminance (lux or footcandle) meter. Technical definition of illuminance meters, cosine response and spectral correction.
- Illuminating Engineering Society — Cosine law. The geometric relationship between incidence angle, distance and illuminance.
- NIST — Realization of the candela. Photometric quantities and calibrated measurement principles.
- NIST — Optical characterization of diffuser-input standard irradiance meters. Directional-response characterisation of reference instruments.
- NIST — Measurement Services: Photometric Calibrations. Metrological context for illuminance calibration and uncertainty.
- Australian Education Research Organisation — Explicit instruction practice guide. Support for sequencing concepts and worked examples during a Science investigation.
Explore eduKateSG
To connect this experiment to the way scientific explanations are checked and revised, visit How Science Works. For how a school lesson becomes evidence of learning, read How School Works. The wider parent-and-learner learning framework is in How Education Works.