February 1967 L E T T E R S TO T H E E D I T O R 279

Luminance and Induced Colors from Adaptation to 100-Millilambert Monochromatic Light

RALPH M. EVANS Photographic Technology Division, Eastman Kodak Company,

Rochester, N. Y. 14650 (Received 29 July 1966)

INDEX HEADINGS: Color; Vision.

IN the course of checking out a new instrument which was built for another purpose, I have run an exploratory experi­

ment that has turned up enough points of interest to justify letting others know about them.

280 L E T T E R S TO T H E E D I T O R Vol. 57

The instrument is, in effect, a visual transmission densitometer reading in the density, log(l/T), range from 0-4. The visual field is circular, subtending an angle of 10° with a central circular spot of 1°. The instrument is built so that interference filters may be placed in the path of the light for either the surround or the main beam or both, and both intensities are controlled by their own circular density wedges, calibrated from 0 to 4.4. The in­strument was designed by O. E. Miller and built by members of his staff.

The luminous transmittances of a set of narrow-band inter­ference filters have been determined by standard spectrophoto-metric procedures using a Cary recording spectrophotometer. The transmittance was integrated by computer using 2.5-nm intervals and was expressed in relation to a 3000°K tungsten lamp. The luminosity data for the CIE standard observer were used.

Using this calculated luminous transmittance and the cali­bration of the instrument, we placed each of these filters in the surround beam successively. The calculated luminance of the resulting surround was adjusted to 100 mL. The surround had, of course, a purity approaching 1.0 in each case. The center beam was then adjusted to a calculated luminance equality with the surround; the center beam had a color temperature of 3000°K when no filter was in that beam. The natural pupil of the eye was fully illuminated.

For each filter, after the appearance of both the center and the surround was recorded, adaptation was allowed to proceed until a maximum amount of color had been induced into the central field, and the hue and appearance of this color as well as of the surround was again noted. Adaptation usually took from 30 sec to 2 min. Since the eye was moved freely, it is assumed that adap­tation was due almost wholly to the surround.

The wedge controlling the central spot was then used to decrease the luminance of the central spot considerably. In all cases con­siderable gray was added to the appearance of the spot. The wedge was then readjusted until visually this gray just disap­peared. This was the sole criterion used; no attention was paid to the often very bright color that contained this gray.

FIG. 1. Solid lines are calculated density-wedge settings to make the central beam 100 mL and 15 mL, respectively. Dots are the actual settings to make gray just disappear from the central beam when the surround was set to a calculated 100 and 15 mL at the indicated wavelengths.

TABLE I. Colors induced by highly saturated surround colors.

In every case, without exception, this point was found to be at the previously calculated luminance equality with the sur­round, within the precision of the setting. This whole series was then repeated at a surround setting of 15 mL with the same result but poorer repeatability. The data are plotted in Fig. 1.

Several rather remarkable facts emerge from these results. First they confirm the precision of the spectrophotometry; the density of the lightest filter was 2.3 and that of the darkest 3.9. Second, they confirm the applicability of the CIE standard ob­server luminosity curve to this situation. Third, they indicate that after adaptation the proper criterion for luminance equality for this situation is the point at which gray just disappears, and not at the brightness-match point. Fourth, and perhaps most surprising, the results show that I myself am quite close to the CIE standard observer, for luminance.

The colors that were induced into the central 3000°K beam by the various monochromatic surrounds were equally surprising. Before describing them it is necessary to describe first the peculiar appearance of this spot after adaptation. No attempt was made to fixate the spot; in fact I deliberately looked around the field and back at the spot many times. When the eye moved, for most of the filters the spot flickered and often changed color. The effect was subjectively very similar to the retinal rivalry that is experi­enced when the two eyes view dissimilar objects or colors that do not fuse; but the new effect is not as slow or complete as retinal rivalry. Usually one color predominated, but a movement of the eye would bring in the other briefly in its place. There seemed to be no fusion of the colors except perhaps in the case of one or two of the filters as described below. I refer to the subdominant color alternating with the dominant, although alternating is too positive a word.

With the entire set of filters only jour hues were induced as either dominant or subdominant colors. These, to the best of my ability to decide, were the four psychological primaries blue, green, yellow, and red; (to me psychological red is a saturated pink and I call it pink in what follows). These appeared singly or in pairs, but only in certain combinations. With pink and green only yellow or blue, with blue only green or pink, and yellow was not dominant in any case. Fusion of the two colors was not possible except near 500 nm with which it was easy to see a fused orange, which separated easily into pink and yellow.

For all wavelengths shorter than 504 nm if the luminance of the spot was increased above a luminance match, the pink gradually disappeared and was replaced by pure nonalternating yellow whose saturation increased with luminance. For some of the shorter wavelengths this saturation increased until the lumi­nance of the spot was as much as 40 times that of the surround. For all other combinations in which this was tried, the saturation of the colors in the spot decreased and color disappeared shortly above the surround luminance.

In order to investigate the purple region, Wratten Filters, Numbers 36 (577c), 32 (547c), 97 (516c), and 33 (507c) were used in the surround; the range was extended toward the red by adding

February 1967 L E T T E R S T O T H E E D I T O R 281

CC yellow filters to the Number 33. Wratten Filters, Numbers 89B (718) and 87 (757), were used to extend the long wavelength range beyond 700 nm.

Four definite but not sharply defined regions of the spectrum were found in which the induced color had a single hue, (i.e., no color alternated with it). Each represented a transition point for the subdominant color at which it changed from blue to yellow or green to pink. These points are indicated in Table I. The transition point for induced green lay beyond the spectrum locus, in the purples near 700 nm.

Most of the residual colors of the surrounds also tended toward these four colors, the notable exception being the wavelengths longer than about 610 nm and on into the ir, all of which were orange red. The colors after adaptation, of course, were of much lower saturation than the induced hues, and a memory match between the larger area and the small one was difficult. In any case the colors were blue, green, yellow, and pink. There was a distinct orange color at 593 and 599 nm. Only one wavelength (583 nm) gave a pure yellow.

In all, six colors were seen in the surround after adaptation; the initial colors move toward these in either direction during adaptation. No violet was visible after adaptation.

For most of the filters the experiment was tried of adding a blue filter to the central beam, which raised its color temperature to 5600°K. This did not change the four induced hues but did change the wavelengths at which changes in the pairs took place. Table I applies only to the 3000°K beam.

These results suggest many further experiments along the same line. However, my present intentions are to proceed with the experiment for which the instrument was designed and to return to this only if it becomes pertinent to the new experiment.