Letters
Shedding More Light on the Lumen
In his article "In the Dark about the Lumen " (IAEEL Newsletter 2/95) Mark Rea elucidated an important issue.The issue concerns the nature of the visual tasks, the participating retinal photoreceptors, and the overall light level under which the tasks are to be performed. The question arises as to whether the photopic lumen, as calculated with the V(lambda) function, is the sensible calibration of light spectrum when light levels are typical of roadway or nighttime security conditions (luminances on the order of 1 cd/m^2).
As Rea correctly argues, peripheral-field-of-view information is important for safe driving. However, peripheral information is not as important as the primary visual information contained in the small field of view defined by the fovea and the immediate para-fovea which covers about the central 10 degrees of the visual field. A person standing and viewed at 50 m subtends about 2 degrees of vertical visual angle, while a roadway of about 10 m width viewed at that distance subtends a visual angle of about 10 degrees.
The corresponding portions of the retina are mostly populated by cone receptors, with rod receptors being totally absent in the central 2 degrees. To the extent that the V(lambda) function represents an average cone spectral response, our primary concerns about its adequacy for roadway conditions should be the possible effects of lower levels of adaptation luminance and a slight increase in the field size in its determination. Fortunately, there is considerable information about these concerns, and although they are not negligible their quantitative effects are relatively small compared with the difference between a photopic lumen and a scotopic lumen for typical exterior light sources.
Some appreciation of the effect on spectral sensitivity of increasing the observers' field size from 2 degrees to 10 degrees can be gained by utilizing the function prescribed by the CIE for the 10-degree observer. For the four lamps listed in figure 2 by Rea, the ratio of 10 degree to 2 degree lumens per watt are respectively 1.056 for incandescnets, 1.035 for HPS, 1.093 for MH, and 1.023 for FSFL. The effects here are not large and are within the range of uncertainties in the spectral output of the lamps that would be expected due to manufacturer differences and to changes in the behavior of lamps over their lifetime.
The determination of the standard V(lambda) function is based on the responses of young adults to the procedures of flicker photometry (the method prescribed by the CIE). Typically, this procedure uses adaptation (surround) luminances of about 25 cd/m^2 or less with (central) test luminances generally less than 100 cd/m^2.
Measurements have also been carried out in completely dark adapted eyes over a wide range of adaptation conditions for central fields of view. There is a general tendency for the shape of the resultant V(lambda) functions to gradually become narrower as the adaptation luminance rises. However, the effects of changing the adaptation luminance by two orders of magnitude across subjects is often not greater than individual subject differences at a particular adaptation luminance.
Spectral sensitivities using the method of heterochromatic brightness matching have been determined for a few subjects over a wide range of adaptation conditions by Kokoschka. In terms of typical reductions in retinal illuminance occurring in conditions spanning interior to roadway light levels, the effect on lamp efficacy appears to be small.
Other visual functions, such as the contrast sensitivity function, also show only slight changes to varying adaptation luminances from 1 cd/m^2 to 100 cd/m^2. Perhaps even more significant for lighting practice is that these differences in the V(lambda) function resulting from a 2-log-unit change in adaptation luminance are comparable to the differences between the values of the V(lambda) function heterochromatic brightness or certain other alternate methods and that obtained using flicker photometry. Thus, it would appear that for present lighting practice when the task is foveal the V(lambda) function should be adequate.
However, this may not be the case, and Rea's concern about the eyes' peripheral response does need to be taken into account. This is because in the real world of lighting practice, rather than tasks generally being viewed monocularly through fixed size artificial pupils (the practice in V(lambda) determinations) they are viewed in binocular vision with natural pupils. It is mostly the scotopic luminance of the surrounding light that determines the size of the pupil, and to the extent that pupil size can affect vision of the task, the surrounding spectrum could be important.
At the lower light levels mentioned by Rea, a critical factor in visual performance could be the net retinal illuminance associated with the task which would be maximized by making pupils large (i.e., more light on the retina). This would be most efficiently achieved by supplying scene illumination with a scotopically deficient source, such as HPS, and would lead to a conclusion opposite to that drawn by Rea as to the choice of illuminants. On the other hand, the task could involve visual detail, and a smaller pupil, which provides a higher level of optical quality, might allow for improved vision.
As Mark Rea expresses in his final paragraph, the lighting profession needs to have a better understanding of the visual needs in the roadway and security lighting regime. We certainly agree.
Sam Berman
Senior Scientist
Lawrence Berkeley National Laboratory
Berkeley, CA, USA