I noticed what I believe to be a basic experimental design flaw in Christopher Cuttle's article,“A Proposal to Reduce the Exposure to Light of Museum Objects without Reducing Illuminance or the Level of Visual Satisfaction of Museum Visitors,” which appeared in the Journal of the American Institute for Conservation, vol. 39, no. 2.

In his experiments, Mr. Cuttle used color “reproduction prints” of paintings to test a three-bandwidth lighting system. The lighting system employed 450 nm, 530 nm, and 610 nm (having the characteristic colors of blue, green, and red, respectively).

Although the author does not state the specific process by which the color reproduction prints he used were manufactured, they are most certainly standard offset lithographic halftone images (at least in the case of the Renoir portrait illustrated). Such reproductions (unlike paintings) are printed using dots of only three colors of ink: cyan, yellow, magenta (plus black, of course). And these three inks have maximum spectral reflectances at approximately the same three wavelengths produced by the light source tested.

Had test images made with real artists' paints been used—including paint colors having significant spectral reflectances at wavelengths where the spectral power distribution of the experimental light source was very low—the test subjects' responses would undoubtedly have been very different from those reported.

Contrary to Mr. Cuttle's conclusions, this is one of the main reasons museum professionals, including exhibition designers, avoid the use of triphosphor fluorescent lighting and prefer light sources with fairly even spectral power distributions that approximate natural light.

It also seems strange that Mr. Cuttle chose to test light sources with color temperatures of 2850�K and 4200�K, when testing conducted by Steven Weintraub and others has clearly shown that illumination having a color temperature of about 3500�K is generally preferable for viewing artwork. Why is it that the two color temperatures Mr. Cuttle decided to investigate avoid testing near a logical, established norm?

Unfortunately, Mr. Cuttle's extensive project has only shown (and apparently by pure coincidence) that if the spectral reflectance of a particular color of printing ink closely matches the spectral power distribution of a narrow-band illumination source, then that ink looks okay under that light. His proposed lighting system might be used to reduce light exposure on posters, book jackets, and postcards in a museum shop without annoying visitors too much, but he has not shown anything about how such lighting would work with real paintings (or even paint samples).

I hope other lighting researchers may find the above comments worthy of consideration, and thus avoid lighting tests based on the appearance of offset lithographic halftone reproductions. I invite Mr. Cuttle to refine his research in this area, and I wish him the best of luck in tests using the full spectrum of real artists' paints (don't forget oranges and violets).

Mr. van Gelder correctly points out that color reproduction prints are produced using three colors of ink: cyan, yellow, and magenta. However, I dispute his assertion that the spectral reflectances of these inks coincide with the peaks of the three 40 nm bandwidths of the experimental lighting system. These inks do not have similarly narrow spectral reflectance bands. They are characterized by much broader spectral reflectance curves, each covering at least two of the bandwidths of the lighting system, so that each of the inks reflects a combination of the incident wavebands.

The spectrum of light reflected by a pigment will differ substantially if illuminated by a broad-spectrum light source or a three-band source. However, human color vision is not based on spectrum analysis. The light focused onto a receptive field of the retina is sampled by three types of “cone” receptors, which respond maximally to short-, medium-, and long-wavelength regions of the visible spectrum, and the color sensation depends on the relative responses of these receptors. That we are able to discriminate more than 10 million color differences by this process is made all the more remarkable by the fact that we are unable to discriminate gross differences of light spectra that produce matching responses from the three types of cone receptor. Although color vision appears straightforward at this level, the processes between retinal stimulation and the sensation of color are complex. Our procedures for assessing color rendition are based on the CIE Color System, which in itself is a simplification of the process of color vision. We do not have a sound theoretical basis for comparing the acceptability, or the preferences, for color rendition of different light sources, and in this situation, the results of controlled observations must take preference over theoretical musings.

Mr. van Gelder points out that the experimental conditions are at variance with current practice, that that was the reason for conducting the experiment. It is well known that art museum directors tend to favor broad-spectrum light sources, and, as Mr. van Gelder suggests, they justify this on the basis of rendering the work in a way that is faithful to the conditions in which the artist produced the work. This assertion should not be accepted without some critical examination. The north sky light favored by artists (in the northern hemisphere) is indeed broad-spectrum source; it also has high color temperature, and typically, artists'studios provide it at high levels of illuminance. In museums, the artists' works are viewed under much lower illuminance, and where artificial lighting is used, with much lower color temperature. The appearance of the works is not as seen by the artist, yet these viewing conditions are found to be generally acceptable. The point of the experiment was to examine whether it might be acceptable also to depart from broad-spectrum light. Mr. van Gelder interprets the results as the viewed objects looking “okay” under the experimental source, but that is not an accurate comment. The subjects reported slightly more difference when comparing dissimilar light sources than when comparing identical sources, and that should hardly be surprising. What should surprise Mr. van Gelder is that they did not prefer the broad-spectrum source to the three-band experimental source.

I am well aware of Steven Weintraub's enthusiasm for 3500�K halogen light sources, and I enjoyed a visit to an exhibition at the National Gallery of Art in Washington, D. C., where this source was used. Even so, I do not accept that this source has been “clearly shown” to be “generally preferable” for all applications. However, the purpose of my experiment was not to examine a preferred or optimum color temperature, but to test whether color temperature is a variable that would influence the results. I ran the experimental procedure at a “low” and “high” color temperature, and the subjects' responses were essentially similar.