September 2000 Volume 21 Number 3
What is the color of white light? This may seem like an odd question. After all, white light is white -- the combination of all the colors of the rainbow. And yet, depending on the context, white light does not always appear to our eye as white. When an observer stands in an atrium filled with "white" daylight and looks at a gallery illuminated with "white" tungsten-halogen lamps, the artificial light appears yellow or unnaturally warm. When the visitor stands in an artificially lighted space where there is no reference to daylight, the light appears "white" and a small spot of natural light will appear blue or unnaturally cool compared to the primary tungsten-halogen light source. So, which light is "white"? The answer is -- both, even though they differ in the blue-to-red proportion of visible energy. In fact, there is a wide variety of "whites," ranging from cool to warm, determined by the varying proportion of blue to red.
When we compare different types of light sources such as incandescent lamps, tungsten-halogen lamps, various fluorescent lamps, and daylight during different times of the day, each source appears to transmit a different quality of "white" light, especially when these sources are seen side-by-side. The difference in white is defined by the "color temperature" of the source. The concept of color temperature is based on the observation that a substance heated to high temperatures emits visible radiation in a broad spectrum. At 2000° K, the emitted light looks orange-yellow, based on the high proportion of long "warm" wave lengths. As the temperature increases to 20,000° K, it appears blue, based on the high proportion of short "cool" wavelengths. Between these two extremes, the light appears to be "white" rather than yellow or blue, although the "white" light will range from warm to cool, depending on such factors as intensity and context. Since "white" light covers a broad range of color temperatures, how is our perception of a work of art affected by the choice of illumination? Is there an ideal white light, i.e. is there a preferred color temperature for viewing works of art?
The traditional response to the question of the optimum light source for viewing art has been to use the same type of light in which the object was either created or intended to be seen by the artist. Prior to the use of modern high color temperature sources like fluorescent lamps, this would have been either natural light (preferably northern light which has a very high color temperature), or sources such as a candle, gas light or an incandescent lamp which have very low color temperatures. Based on the fact that many artists preferred to work in daylight, it has been assumed that daylight is the best illumination source for viewing art. Many museums have spent enormous sums of money on systems that incorporate high color temperature natural light, especially for galleries where oil paintings are exhibited. Is this assumption about using high color temperature natural light valid? According to research published over half a century ago, the answer is an unequivocal "no".
In 1941, a Dutch researcher, A.A. Kruithof ("Tubular Luminescence Lamps for General Illumination," Philips Technical Review, vol.6, 65-96, 1941), published a graph (see illustration) summarizing the relationship between color temperature, intensity, and the "pleasant" quality of an illumination source. According to the Kruithof curve, an observer prefers lower color temperature lighting when the light level is lower, and prefers a higher color temperature when the light level is higher. In essence, Kruithof provided a quantitative basis for describing a phenomenon that we all experience. For example, a space uniformly illuminated at 20 footcandles by daylight at a color temperature above 6000° K appears gloomy and overcast, whereas the same space illuminated at the same 20 footcandles with tungsten-halogen lamps at 3000° K appears pleasant and comfortable. The conclusion drawn from Kruithof's curve is that our color-temperature preference changes based on the intensity of the light within a space.
When an artist paints outside or by an open window, the color temperature is very high and so is the intensity of the light. When the painting is exhibited inside a museum at 20-30 footcandles, it is illuminated at a much lower level of intensity. According to the Kruithof curve, if the painting is illuminated at the same color temperature under which it was created (for example, 10,000° K), but at a much lower light level (for example, at 20 footcandles), the condition of display would be unpleasant. Therefore, it is not desirable to illuminate a painting at the same color temperature under which it was created if it will be exhibited at a much lower level of light intensity. The appropriate color temperature must be selected based on the intensity of the ambient illumination.
Kruithof's curve describes the general experience of light as pleasant or unpleasant. This quality deals with general or ambient conditions. Although quality of ambiance is important, Kruithof's work provides no information on how color temperature affects the observer's perception of specific colors and color relationships. A review of the technological literature does not provide adequate answers to questions about the impact of color temperature on visual perception.
The relationship of color temperature to color perception is not obvious due to the remarkable ability of human vision to compensate for wide variations in the spectral distribution of light sources. This ability, referred to as "color constancy", is similar to the "white balance" adjustment on a video camera which takes into account the color temperature of the light source. Color constancy explains our ability to perceive colors in the same way under a wide variety of viewing conditions. When we look outside through a green tinted window, the scene appears natural because the brain compensates for the green tint and normalizes the view. However, when we see the same scene through a partially opened window where the eye compares the two views, we become aware of the tinted glass. The colors seen through the green tinted portion appear to be distorted because the brain normalizes colors and color relationships based on the untinted view and recognizes the distortions seen through the tinted glass. Although color constancy reduces the impact that a colored filter or a shift in color temperature has on color perception, does it fully compensate for such differences? The question remains--how important is color temperature in viewing a work of art?
According to recent research by myself and colleagues in the fields of visual psychology and museum lighting, color temperature plays a very important role in how one views a work of art. A variety of experiments, conducted under controlled laboratory conditions and in museum galleries, are providing a new understanding of the interaction of color temperature and visual perception.
Studies were carried out utilizing an ingenious combination of equipment developed by Kevin McGuire, Tailored Lighting Inc., which allowed me to evaluate the effect of small incremental changes in color temperature. The test equipment utilized conventional 3000° K tungsten halogen MR-16 lamps, and special MR-16 lamps rated at 4700° K developed by McGuire (These lamps are commercially available under the brand name "SoLux" and will be described in greater detail in a future WAAC article). By altering the voltage in small steps with a programmed controller, the intensity and color temperature of each lamp is modified. When mixed, the combined output of both lamps produces a full-spectrum source at intermediate color temperatures between 3000° K and 4700° K with a minimal change in intensity.
A wide variety of color reproductions of paintings were examined under different color temperatures. The results were surprising. All the observers in the initial experiments preferred a similar, narrow color temperature range, regardless of the palette or subject of the painting. Further confirmation of these unexpected results took place in museum field tests. The first test took place in 1995 at the National Gallery of Art (NGA), during the special exhibition of paintings by Vermeer. Gordon Anson, Chief Lighting Designer at the NGA and an associate in much of my lighting research, arranged access to the Vermeer Exhibition and adjacent galleries containing the permanent collection of 19th century European paintings. Jay Kreuger, a painting conservator at the NGA, also participated in the study.
Each painting was examined at 20 footcandles utilizing a variety of color temperatures. We all agreed that there was a dramatic change in the appearance of all paintings as the color temperature varied. We all preferred the same color temperature for individual paintings within a narrow tolerance of about 200° K. Regardless of whether they were Impressionist paintings of bright, cool outdoor scenes, or dark, warm Dutch interiors, the color temperature preference was within 200° K for all paintings, The two exceptions, a painting by Turner and The Lace Maker by Vermeer, were unusual because of the yellowed condition of the varnish. In these two instances, we all preferred a slightly higher color temperature (300° K higher) which diminished the impact of the yellow varnish.
Subsequent tests in other museums yielded the same results. The most dramatic test took place at the Memorial Art Gallery of the University of Rochester, Rochester, N.Y.. The late 19th century French Painting gallery was permanently illuminated with 3500° K SoLux lamps under the supervision of Candace Adelson, Curator of European Art, who also played a key role in earlier studies. A visitor comment book was provided, along with a brief statement regarding the experimental nature of the gallery lighting. Visitors were asked to comment on the quality of the lighting compared to their memory of how the gallery had been illuminated, and compared to adjacent galleries, illuminated with incandescent (2700° K) and tungsten halogen (3000° K) lamps at the same level of intensity. The book was filled with observations about the increase in the saturation of colors, the greater sense of depth within the paintings, and the improved brightness and clarity of the paintings and of the gallery space in general. These comments were similar to those expressed by observers in other experiments when asked about how the paintings alter with changes in color temperature.
The fact that most observers chose the same preferred color temperature within a narrow range is further evidence that the choice of color temperature involves more than an arbitrary aesthetic preference. It is based on a fundamental property of human vision. To further understand the preference for a specific color temperature, additional studies were undertaken in a non-art context. A white reflective surface was illuminated at a fixed intensity as the color temperature was increased and decreased in small increments between 3000° K and 4700° K. Observers were asked to describe the light as warm, cool or intermediate. For a surface illuminated at 20 foot candles, a value around 3700° K was chosen as the intermediate value, measured with a Minolta photographic color temperature meter (Model II). At 20 foot candles, 3700° K appears as an achromatic white light compared to higher or lower color temperature sources. Coincidentally, the choice of 3700° K was the preferred color temperature chosen on aesthetic grounds when looking at paintings. This suggests that the aesthetic preference for a specific color temperature derives from a fundamental characteristic of human color perception.
These color temperature investigations are still at an early stage. Through the support of a two-year research grant from the National Center for Preservation Technology and Training further research will be conducted by myself and specialists in color vision research at the City University of New York. There are many questions that require more careful study. For example, the preferred color temperature readings taken with the Minolta meter must be reconfirmed with more accurately calibrated instruments. Another variable that requires further study is the relationship between the preferred color temperature and intensity. Based on the Kruithof curve and experiments in progress, it is probable that the preferred color temperature will shift, based on the intensity of the illumination source. Beyond the theoretical questions, there is a wide range of issues involving practical application that must be understood in order to utilize the results of these studies. A number of these application issues will be discussed in a subsequent WAAC Newsletter.