JAIC , Volume 39, Number 2, Article 4 (pp. to )
JAIC online
Journal of the American Institute for Conservation
JAIC , Volume 39, Number 2, Article 4 (pp. to )




Current recommendations for museum lighting practice are broadly in agreement that ultraviolet (UV) should be severely restricted or eliminated and that exposure to light should be limited in both intensity and duration (CIBSE 1994; IESNA 1996). For exhibits that are categorized as highly susceptible to exposure damage, the recommended maximum illuminance is 50 lux, which is recognized as the lowest practical level for exhibits for which color discrimination is an important factor (Thomson 1986). Even when these recommendations are strictly applied, display lighting still causes permanent damage to exhibits (Feller 1967; Michalski 1987). While there are no specific standards for infrared (IR) control, dichroic reflector spotlights are generally recommended. The mirror that forms the beam for this type of spotlight has a wavelength-selective reflecting surface that directs light, but not IR, into the beam. Generally, current practice for museum display lighting utilizes incandescent filament light sources, such as the MR lamp (a tungsten halogen source with integral multifaceted dichroic reflector, usually 2 in. in diameter). The spectral power distribution for this type of illumination is characterized by a continuous, smooth curve throughout the visible spectrum, increasing toward the long-wavelength end.

It used to be supposed that incandescent light sources are safe because they are weak sources of short-wavelength radiation, particularly of UV, but also at the short-wave (blue) end of the visible spectrum. However, recent studies of exposure of artist's pigments to narrow wave bands within the visible spectrum have shown that the wavelength susceptibility of a pigment is largely determined by its spectral absorption characteristic (Saunders and Kirby 1994). An artist's palette that covers a full color range inevitably includes pigments that are selective absorbers for every wave band in the visible spectrum. When nonvisible radiation has been controlled, the damage potentials of the visible radiation provided by alternative light sources may be compared in terms of irradiances at equal illuminances (Michalski 1987).

Irradiance is the measure of incident radiant power density in watts per sq. m (W/m2), and illuminance is the measure of incident light density in lux, where one lux is one lumen per sq. m. The basis for comparison used in this report is to evaluate sources in terms of radiant luminous efficacy measured in lumens per radiant watt (lm/W(r)). This term should not be confused with the luminous efficacy values quoted in lamp catalogs, which are measures of lumens emitted per watt of electrical power input (lm/W). For the benefit of readers who are not familiar with lighting technology, suggestions for further reading follow the references.


Although radiant luminous efficacy provides a useful basis for comparing the damage potential of alternative light sources, it is far from being the whole story. A critical concern in museum lighting is how the illumination affects the appearance of colored materials. Consequently, this study was designed to relate the spectral power distribution of lighting to the responses of subjects viewing artworks in a simulated art gallery setting.

It is necessary to review briefly the metrics that are used to specify the color rendering properties of light sources. Comparison is made to a black body, which is a theoretical substance for which the spectral distribution of radiant power emission is defined by its temperature. The correlated color temperature (CCT) of a light source is the temperature in degrees Kelvin (K) of a black body that most closely matches the color appearance of the source. At a low CCT (<3000 K), the appearance is a warm, yellowish light reminiscent of sunlight or a candle flame; at an intermediate color temperature (≈4000 K), the color appearance is a more neutral, white light; and at a high color temperature (>5000 K), the appearance is a cool, bluish white light reminiscent of sky light.

The color rendering index (CRI) of a lamp is defined by a procedure that compares measurements relating to the color of the light reflected from a set of reference-colored samples illuminated by the lamp with the light reflected from the same samples when illuminated by a black body source having the same CCT as the lamp. (A different type of comparison source is used where CCT>5000 K.) If all of the samples match perfectly under both sources, the lamp is accorded a CRI of 100. Any departures from a perfect match reduce the CRI. This procedure assumes that, for low and intermediate color temperatures, the theoretical black body is the ideal color rendering source.

The black body source is luminous because it is incandescent, and its relative spectral power distribution almost exactly matches that of an electric incandescent lamp at the same CCT. Incandescent lamps are quoted to have CRI values of 99 or 100, and are widely perceived to be perfect color rendering lamps, but this perception needs to be qualified. They are perfect only in that the color appearances of illuminated surfaces match the appearances that they would have if illuminated by a black body of the same low color temperature.

It is well understood by lamp manufacturers that it is not necessary to match the spectral power distribution of a black body to achieve a high CRI value. Triphosphor fluorescent lamps concentrate their radiant power emission into three spectral bands, and there are lamps of this type that achieve high CRI values. The development of these lamps followed from basic research by William Thornton (1974; 1975; Thornton et al. 1975), who had identified three optimal wavelengths for matching the lamplight from broad spectrum sources, such as incandescent lamps, and achieving this result with high luminous efficacy and high CRI. The band center wavelengths are approximately 450 nm, 530 nm, and 610 nm, and the light of these wavelengths has the characteristic colors of blue, green, and red, respectively. Thornton also claimed that, compared with broad spectrum sources, lamplight composed of these “prime colors” has greater “visual clarity,” and, because of this characteristic, less illuminance is required for equal visual satisfaction (Thornton 1975, 38).

Although triphosphor fluorescent lamps have become widely used for many lighting applications with less demanding visual requirements, they have not been generally accepted for museum lighting. There probably are several good reasons for this. Exhibition designers like the compactness of the modern incandescent lamps, and the range of beam spreads available, and the ability to focus them, and the ease of dimming them … the list goes on. Exhibition designers are not likely to give up all of these advantages over fluorescent lamps even though the fluorescent lamps may be much more efficient. However, Garry Thomson has criticized the use of Thornton's “prime color” illumination for museum applications on the basis of color rendering and has commented that prime color illumination “amounts to distortion and so might be undesirable” (Thomson 1978, 178). As triphosphor lamps improved, he later added, “There will be certain museum situations not demanding the best color rendering where they will be the choice” (Thomson 1986, 207).

The author's own studies of spectral power distributions found that, at equal illuminances (lux) and the same color temperature, the irradiance (W/m2) for a three-band source may be as much as 40% less than for an incandescent lamp. This finding suggests that there is scope to significantly reduce the irradiance of exhibits without reducing the illuminance. If the irradiance can be reduced, it can be expected that the rate of damage will be correspondingly reduced, as it is irradiance that determines the damage potential for radiant energy within the visible spectrum. This theory leads to the first research question: Is the “distortion” of color rendering acceptable? If it can be shown that it is acceptable, at least in some situations, the second question becomes: Can a light source be produced that provides this conservation advantage and meets needs of exhibition designers?


The aim of the experiment was to address the first research question stated above. An experimental situation was constructed to obtain paired comparison subjective assessments of a range of artworks illuminated alternately by a tungsten halogen multifaceted reflector (MR) spotlight and by an experimental three-band source, both having the same correlated color temperature.

It was predictable that the CCT and the color appearance of an MR lamp could be matched by a three-band source that would have significantly higher radiant luminous efficacy (lm/W(r)) than the MR lamp. It was expected that if subjects were presented with an achromatic scene alternately lit to the same illuminance by these two types of light source, the subjects would not differentiate between the two presentations. What could not be predicted was how they would respond to a scene that involved colored materials.

The basis for the comparison was an illuminance of 50 lux on the artwork provided by an MR spotlight, as this the type of lighting and light level are widely adopted for display of susceptible exhibits. However, the CCT of the MR lamp is low: 3000 K at full voltage, and in practice often lower due to being dimmed to provide 50 lux. It was decided that the experiment should also include an intermediate color temperature source, and this was achieved by repeating the procedure using a new type of MR lamp that has a CCT of 4700 K at full voltage.