Is lux the right measure in darkness?
When a lighting project is specified, calculated, and evaluated, it almost always happens in lux and lumens — units defined around how the eye responds to light under full daylight conditions, a state called photopic vision. In daylight that is a reasonable approximation. On a street-lit road at night, the assumption does not hold.
This is not a meaningful objection to using lux as a measure — lux is practical, well-established, and comparable. But there is a specific situation where the difference matters: when two light sources with the same measured lux but different spectral compositions are compared for visual performance in darkness. There, the comparison can become systematically misleading.
In short
Two luminaires with the same lux can be perceived as differently bright at night — depending on their spectrum. Lux is the correct measure under EN 13201, but it does not describe the full visual experience in darkness.
Three vision regimes
The eye continuously adapts to the ambient luminance level, and depending on that level it operates with different combinations of receptors. CIE distinguishes three regimes based on road surface luminance:
| Vision regime | Luminance range | Dominant receptors | Peak sensitivity |
|---|---|---|---|
| Photopic | Above ~3–5 cd/m² | Cones | 555 nm (green) |
| Mesopic | 0.005–5 cd/m² | Both cones and rods | 507–555 nm (level-dependent) |
| Scotopic | Below ~0.005 cd/m² | Rods | 507 nm (blue-green) |
Street lighting luminance class M is defined in EN 13201. Class M6 requires a mean luminance of 0.3 cd/m², class M1 requires 2.0 cd/m². The entire M-scale — the scale that governs street lighting design in Sweden and Europe — therefore falls within the mesopic range. This means that the photopic assumption underlying the units used in standard practice does not apply to the conditions actually being designed for.
The Purkinje shift
When the ambient luminance falls into the mesopic and scotopic range, the eye's sensitivity shifts toward shorter wavelengths — toward blue. This is called the Purkinje shift, described by Jan Evangelista Purkyně in the 19th century. Physiologically it occurs because rods, whose sensitivity peaks at approximately 507 nm, take over from cones, whose peak is at 555 nm.
The effect is observable in everyday life: a red object that appears bright in daylight becomes nearly dark at dusk, while a blue object remains relatively more visible. The same applies to road markings, vehicles, pedestrians, and obstacles on a night-lit street.
The consequence for lighting is that two sources with identical photopic lumens but different spectral compositions deliver different rod stimulation in darkness. A source rich in shorter wavelengths — higher colour temperature — activates rods more than the lux figure captures. A warm source with little short-wavelength content does the opposite. The lux comparison misses that difference.
The S/P ratio — what it measures
The S/P ratio (scotopic/photopic ratio) is the numerical measure of this spectral difference. It is defined as the ratio of a source's scotopic lumens to its photopic lumens — both measured from the same source but weighted with different sensitivity curves (V′(λ) and V(λ) respectively).
Typical S/P values
- High-pressure sodium (HPS/SON): S/P ≈ 0.4–0.6. Yellow spectrum dominated by sodium emission lines; little rod stimulation per photopic lumen.
- Warm white LED (2,700–3,000 K): S/P ≈ 0.9–1.3. Broader spectrum than HPS; somewhat more short-wavelength content.
- Neutral white LED (3,500–4,000 K): S/P ≈ 1.4–1.8. More pronounced blue component; higher rod stimulation per photopic lumen.
- Amber LED (monochromatic ~590–600 nm): S/P ≈ 0.3–0.5. Narrow emission near HPS; minimal rod stimulation.
| Colour temperature | Typical S/P ratio | Rod stimulation / lm |
|---|---|---|
| 1,900 K (amber) | ≈ 0.3–0.5 | Lowest |
| 2,200 K LED | ≈ 0.6–0.8 | Low |
| 3,000 K LED | ≈ 0.9–1.3 | Medium |
| 4,000 K LED | ≈ 1.4–1.8 | Highest |
Note: S/P values vary with the exact spectrum and source design. The values above are representative orders of magnitude from published source measurements, not universal constants.
A high S/P ratio means the source delivers more rod stimulation per photopic lumen. In mesopic conditions, this is a factor that affects actual visual performance — but one that the lux figure does not capture.
CIE 191:2010 — the mesopic system
In 2010, CIE published a system for mesopic photometry: CIE 191:2010 Recommended System for Mesopic Photometry Based on Visual Performance. The system defines an adaptation-dependent sensitivity function M(λ) that interpolates between V(λ) (photopic) and V′(λ) (scotopic) as a function of the luminance level and the source's S/P ratio.
In practice, this means that a mesopic correction factor can be calculated that compares sources' actual visual contribution in darkness rather than their photopic lumens. A source with a high S/P receives a more favourable correction than a source with a low S/P at the same luminance level.
The CIE 191 system is rarely used in standard procurement or calculation today. EN 13201 does not require mesopic correction, and most lighting calculation tools default to photopic weighting. The system exists as a technical framework, but its practical application in projects is limited.
Peripheral vision and motion detection
Rods are heavily concentrated in the periphery of the retina. Central (foveal) vision consists almost entirely of cones; peripheral vision in darkness is driven almost exclusively by rods. This is why in very low light, faint objects are sometimes more visible when looked at slightly off-centre — a phenomenon astronomers call off-axis viewing.
In traffic environments, peripheral vision is critical for detecting movement at the edge of the visual field: the pedestrian stepping off the kerb, the cyclist crossing from the left. Laboratory and simulator studies suggest that sources with higher S/P ratios can produce faster peripheral detection at the same photopic luminance in mesopic conditions. The magnitude of this effect in real traffic is difficult to quantify; field data is limited and laboratory results do not translate automatically into traffic safety metrics.
The honest conflict: traffic safety versus ecology
An unresolved trade-off
Mesopic science argues — at a given photopic luminance — for sources with a higher S/P ratio, meaning higher colour temperature and more short-wavelength content. This benefits rod stimulation and may improve peripheral detection.
Ecological research and dark-sky concerns point in the opposite direction: short-wavelength (blue) light is the primary driver of light pollution, insect attraction, and disruption of nocturnal wildlife. The International Dark-Sky Association and national nature conservation requirements recommend colour temperatures below 3,000 K, or amber sources where justified.
Additionally, there are circadian physiological considerations: blue-rich outdoor light at night can affect melanopic stimulation in nearby residents and animals.
There is currently no consensus-based answer to where the net optimum lies when traffic safety, ecology, and energy efficiency are weighed together. In practice, the decision is made locally: is the installation in an urban core or in an ecologically sensitive area? What M-class requirement applies? Is there ecological vulnerability in the vicinity? Those answers drive the CCT choice more than any universal technical optimum.
The ageing eye
A factor rarely discussed alongside mesopic photometry: the human lens yellows progressively with age, transmitting less and less short-wavelength light. For older road users, the S/P benefit of a blue-rich source is partially damped by the aged lens, and contrast sensitivity generally declines regardless of source type.
The consequence is that the design point for optimal visual performance for an older driver may differ from that for a young driver — and that an ageing population as a whole shifts the socially optimal design point toward more light and better contrast rather than specifically more blue. These effects are documented in textbook literature (Boyce, Human Factors in Lighting, 2014) but are rarely quantified in procurement specifications.
What this means for specification
Lux is the correct measure under EN 13201 — and that will remain the case for as long as the standard applies. But lux describes incident light on a surface, weighted for daylight conditions. It does not describe the full visual experience at night. A few consequences of that difference are worth understanding:
- Lux comparisons are not spectrally neutral. Two luminaires with identical lux but different CCT can deliver different peripheral visual performance in mesopic conditions. Comparing "at the same lux" without accounting for the S/P difference compares photopic lumens, not actual visual performance.
- HPS to LED transition: Moving from high-pressure sodium to LED typically increases the S/P ratio, meaning "the same lux" provides more rod stimulation with modern LED. This is sometimes cited as a reason not to maintain identical lux levels after LED replacement — but quantifying the trade-off is complex.
- CCT choice in sensitive environments: In nature-sensitive areas, at coastlines and nature reserves, the ecological argument for low CCT or amber competes directly with the mesopic argument for higher S/P. This is a policy and environmental judgement, not a purely technical optimum.
- CIE 191 is available: If mesopic correction is desired, it can be calculated per CIE 191. This requires the source's S/P ratio, which should be available in technical documentation or can be computed from spectral power distribution data.
Next level of understanding
Mesopic theory assumes lux is measured correctly. But the calculation model has its own assumptions that break down when road surface properties change.
A DIALux calculation assumes a dry road surface, a fixed observer position, and a point source. On a wet November evening, none of that is true — and the uniformity the calculation predicts can fall substantially.
Lighting fundamentals
What is a lumen?
How lux, lumens, and candela are defined — and why all three are photopically weighted.
Frequently asked questions
Should S/P ratio be specified in procurement?
It is possible to specify, but uncommon in standard municipal procurement today. EN 13201 does not require it, and few suppliers report S/P ratio as standard data. The value can however be calculated from spectral data and requested where a project has specific requirements related to HPS replacement or sensitive environments where spectral composition matters.
Which colour temperature is safest for road environments?
There is no simple answer. Mesopic science argues for higher CCT at a given lux level due to the higher S/P ratio. But ecology, dark-sky concerns, and circadian physiology argue for lower CCT, particularly in sensitive areas. Where a tight budget forces a low lux level, the optimal CCT depends on the location, not on a universal technical preference. In urban areas with standard M-class requirements, 3,000–4,000 K is the most common choice. In nature-sensitive zones, amber (1,800–2,000 K) is increasingly selected despite the low S/P.
Does mesopic photometry also apply to pedestrian and cycle paths?
Yes, and arguably more so than for through traffic. Pedestrian paths are often designed to lower luminance levels than roads, meaning lower adaptation levels and a larger relative difference between photopic and mesopic performance. The peripheral detection that mesopic theory improves is also more relevant for a pedestrian spotting something moving at the edge of their vision, compared with a driver on a straight road. Area lighting standards for pedestrians and cyclists are handled partly separately in the EN 13201 series (P-classes) and in CIE 115.
VALDUR and colour temperature
VALDUR is available from 1,900 K amber to 4,000 K. Colour temperature selection should be based on the site requirements for traffic safety, nature conservation, and environmental impact — not on a universal technical optimum, as no such optimum currently exists.
Summary
Lux and lumens are photopically weighted units — built for daylight conditions. All street lighting operates in the mesopic range, where the eye's sensitivity shifts toward shorter wavelengths compared to the daylight model. This means that two sources with identical lux but different spectra — and therefore different S/P ratios — can deliver different actual visual performance in darkness.
CIE 191:2010 provides a quantitative system for mesopic correction, but it is rarely applied in standard procurement. EN 13201 uses photopic measures. This means that systematic differences in spectral composition can escape the standard process unnoticed.
The conflict between mesopic theory (which argues for higher S/P and higher CCT) and ecological concerns (which argue for lower CCT) is real and unresolved. The right answer depends on location, application, and which trade-offs weigh most heavily in the specific project.