Skip to content
Sensmat

Lux and UV meters

Cosine correction, spectral response, and the two measurement habits that decide whether your readings mean anything.

Updated 14 August 2026

A light meter is the instrument that turns an argument about whether a gallery is too bright into a number. It is also the instrument most often used in a way that produces a number nobody should rely on.

What the specifications mean

Cosine correction. Illuminance on a surface depends on the angle the light arrives at. A correct meter has a diffusing head that weights angled light properly. Without it, light arriving obliquely, which in a gallery is most of it, is under-counted, and the meter flatters the installation. Instruments that state a cosine response error are the ones taking this seriously.

Spectral response. Lux is defined against the sensitivity curve of human vision. A raw silicon detector is more sensitive in the red and near-infrared than the eye is, so an unfiltered meter reads high under a halogen lamp and can be badly wrong under narrow-band LED. Since gallery lighting is now overwhelmingly LED, this has become the specification that matters most.

UV measurement. In conservation, UV content is conventionally quoted in microwatts per lumen, which expresses it relative to the visible light present. That is the form in which filtering decisions get made, so a meter that reads it directly saves a conversion.

Using it so the reading means something

Measure at the object. The number that matters is the illuminance falling on the surface at risk. Hold the sensor against the surface, in its plane, facing outward towards the light.

Do not shadow it. The commonest error by far. Standing between the window and the meter to read the display can halve the value. Use an extendable head, or read from the side.

Measure the worst case. The brightest point on the object, not the average of the wall. Fading happens where the light is.

Measure daylight more than once. A single reading on an overcast Tuesday in November says nothing about the same case in June. Where daylight contributes, either log across a day and a season or measure at the extremes and note both.

Record the conditions. Date, time, weather, which lamps were on, blind position. A reading without context cannot be compared with a later one, and comparison is the whole purpose.

What to do with the numbers

Two readings answer two different questions. The UV figure tells you whether filtering is needed, and if it is high the fix is cheap: UV filter film or filtering glazing removes a disproportionate share of the damage for very little money.

The lux figure feeds the exposure calculation, and on its own it is only half an answer, because damage depends on hours as much as on brightness. Light exposure budgets covers turning the reading into a display decision.

What to check on the specification

Criteria to hold an instrument against. We do not test meters, so this is what the data sheet has to answer.

  1. 01 · Optics

    Cosine correction

    Light arriving at an angle should contribute in proportion to the cosine of that angle. A meter without a properly diffusing head under-reads angled light, which is most of the light in a real gallery.

    Speaks for it

    • A stated cosine response, often given as an f2 error figure, indicates a serious instrument

    Worth weighing

    • A bare photodiode behind a flat window is the usual sign of a meter that will under-read
  2. 02 · Accuracy

    Spectral response matched to the photopic curve

    Lux is defined against human visual response. A silicon detector without proper filtering responds differently, and the error changes with the light source.

    Speaks for it

    • Meters quoting a V-lambda match figure can be compared meaningfully across LED, halogen and daylight

    Worth weighing

    • Uncorrected meters can be well out on narrow-band LED, which is now most gallery lighting
  3. 03 · UV

    UV expressed in microwatts per lumen

    UV content in collections is conventionally expressed relative to the visible light present, because that is what makes filtering decisions comparable.

    Speaks for it

    • Direct reading avoids a conversion step and a mistake

    Worth weighing

    • Some general-purpose meters give absolute UV irradiance only, which needs converting
  4. 04 · Handling

    A separate or extendable sensor head

    The reading has to be taken at the object surface, which is frequently inside a case, behind a barrier or above head height.

    Speaks for it

    • A head on a lead reaches places the body of a meter cannot, without the operator shadowing it

    Worth weighing

    • Integrated meters force awkward positions that change the reading
  5. 05 · Practicality

    Hold and logging

    Reading a display in an awkward position, or capturing daylight variation across a day, both need more than an instantaneous number.

    Speaks for it

    • A logging meter left in place for a day captures daylight variation properly

    Worth weighing

    • Logging models cost considerably more and are rarely needed for electric-only spaces

Frequently asked questions

Can I use a phone app?
For finding out whether a room is roughly 50 or roughly 500 lux, sometimes. For a measurement that goes into a record, no. Phone cameras have no cosine correction, an unknown spectral response and an automatic exposure system actively working against you.
Where exactly do I hold the meter?
At the object surface, in the plane of that surface, facing the light. Not in the middle of the room, not at eye height, and not with your body between the meter and the window.
How often do meters need calibrating?
Annually is the usual manufacturer recommendation. Detectors and filters both age, and UV sensors age faster than visible ones.

Related reading