LECTURE XIII. PHYSIOLOGICAL PROBLEMS OF ILLUMINATING ENGINEERING. 123. The design of an illumination requires the solution of physiological as well as physical problems. Physical considera- tions, for instance, are the distribution of light-flux intensity throughout the illuminated space, as related to size, location and number of light sources, while the relation, to the satisfac- tory character of the illumination, of the direction of the light, its subdivision and diffusion, etc., are physiological questions. Very little, however, is known on the latter, although the entire field of the physiological effects of the physical methods of illumination is still largely unexplored. As result thereof, illuminating engineering is not yet an exact science, as is, for instance, apparatus design, but much further physiological investigation is needed to determine the requirements and conditions of satisfactory illumination. The physical side of illuminating engineering: — to produce a definite light flux density throughout the illuminated space, — is ah engineering problem, which can be solved with any desired degree of exactness, usually in a number of different ways. The solution of the physical problem of light distribution, however, does not yet complete the problem of illuminating engineering, does not yet assure a satisfactory illumination, but with the same distribution of light flux density throughout the illuminated surface, the illumination may be anything between entirely unsatisfactory and highly successful, depending on the ful- fillment or failure to fulfill numerous physiological requirements. Some of these are well understood and such that they can be taken into consideration in the physical design of the illumina- tion, and thus no excuse exists to fail in their fulfillment, though it is frequently done. Such, for instance, is the requirement of low intrinsic brilliancy in the field of vision, of the color of the light, etc. Other physiological requirements are still very little 277 278 RADIATION, LIGHT, AND ILLUMINATION. understood or entirely unknown, while on others not sufficient quantitative data are available for exact engineering calculation. Thus, for instance, the usual suburban street illumination, with arcs spaced at considerable distances from each other and located on fairly low posts, is very much inferior to the illumina- tion given by moonlight, even when allowing for the difference in intensity. Here the reason of the unsatisfactory character of the former illumination is mainly the almost horizontal direc- tion of the light flux. A perfectly vertical direction of the light flux again is unsatisfactory in many cases, and the most satis- factory results are given by a direction of the light flux which makes a considerable angle with the horizontal as well as the vertical direction. Thus, when dealing with directed light, the direction angle is of essential physiological importance. We have very little exact knowledge to guide in the determination of the proper angle in which to direct the light flux ; it is known that in general approximately horizontal and approximately vertical direction of the light flux are objectionable, and an in- termediary angle gives best results. However, the horizontal direction usually is objectionable by excessive contrasts, the vertical direction by flatness in the appearance of the illuminated objects, and, depending on the nature of the objects, sometimes the one, sometimes the other feature may be more objectionable. Hence, the best angle of incidence of the light depends on the nature, that is, the shape and location, of the illuminated objects, on the purpose of the illumination, etc., and thus is not con- stant, but is a function of the problem, which is still largely unknown. 124. Not represented by the physical distribution curve of illumination, but very marked in their physiological effect is the difference between directed light and diffused light. In most problems of illumination, either entirely directed light or entirely diffused light is unsatisfactory, and a combination of directed light and diffused light is required, as discussed in the preceding pages. No exact knowledge, however, exists on the proportion in which directed light and diffused light should be combined for satisfactory illumination, nor how this proportion varies with the nature, color, etc., of surrounding objects, with the purpose of the illumination, etc. That it varies is well known, as for some purposes, as a draughting room, entirely diffused light PHYSIOLOGICAL PROBLEMS. 279 seems best suited, while for other purposes mainly directed light seems more satisfactory. Furthermore, the relations between directed and diffused light have in the illuminating engineering practice been obscured to some extent by the relation between high and low intrinsic bril- liancy and between direct and indirect lighting. Thus, to eliminate the objectionable feature of high intrinsic brilliancy of the illuminant, direct lighting by light sources of high brilliancy, which was largely directed lighting, has been replaced by indirect lighting, by reflection from ceilings, etc., which is diffused light- ing. Where such change has resulted in a great improvement of the illumination, it frequently has been attributed to the change from directed to diffused lighting, while in reality the improve- ment may have been due to the elimination of high brilliancy light sources from the field of vision, and engineers thereby led to the mistaken conclusion that perfectly diffused lighting is the preferable form. Again, in other instances such a change from direct to indirect lighting has not resulted in the expected im- provement, but the indirect lighting been found physiologically unsatisfactory, and the conclusion drawn that the elimination qf high brilliancy from the field of vision has not been beneficial, while in reality the dissatisfaction with the indirect light was due to the excess of diffused light and absence of directed light, and this improper proportion between directed and diffused light more than lost the advantage gained by eliminating the light sources of high brilliancy from the field of vision. In this case the proper arrangement would have been to reduce the brilliancy of the light sources, by diffusing or diffracting globes, to a suffi- ciently low value, but leave them in such position as to give the necessary directed light. Thus, in illuminating engineering, as in other sciences, it is very easy to draw erroneous conclusions from experience by attributing the results to a wrong cause. Any change in the arrangement usually involves other changes: as in the above instance, the change from high to low brilliancy commonly causes a change from directed to diffused light; by attributing the results to a wrong cause, serious mistakes thus may be made in basing further work on the results. 125. In discussing diffused light, we must realize that the meaning of " diffused light" is to some extent indefinite. To 280 RADIATION, LIGHT, AND ILLUMINATION. define diffused light as light which traverses the space in all direc- tions and thus casts no shadow, is not correct, since even diffused daylight casts shadows. For instance, if in Fig. 122 P is the sur- M/w///^^^^^ FIG. 122. face of the ground and A a flat circular shade at distance I above the ground, the intensity distribution of the light in plane P is as shown in Fig. 122 for I = 0.2 A, thus showing a fairly dark shadow beneath the center of A, but a shadow which blurs so very gradually that with most objects it is not marked. The light from a single point source is perfectly directed light; it traverses every point of space in one single direction only, as shown as A in Fig. 123. If we now enclose the point source in an opal globe, which then becomes the radiator, as discussed before, as diagrammatically shown as B in Fig. 123, the light flux traverses each point not in a single direction but in all directions within a narrow angle a, which is the angle subtended by the radiator L from the point P. With increasing size of the illuminant, and thus increasing angle a, C, Fig. 123, the pencil of rays, which traverses point P, gradually spreads, until, when a becomes 180 deg., we get perfectly diffused light, similar to daylight. Hence, with a gradual change of the diameter of the illum- inant, from a = 0 to a. = 180 deg., the light gradually changes from directed to diffused light. Thus, no sharp dividing line FIG. 123. PHYSIOLOGICAL PROBLEMS. 281 can be drawn between directed light, and diffused light, but the directed light from a light source of considerable diameter (that is, a diameter which is not neglible compared with the dis- tance of the illuminated objects from the light) already has to some extent the character of diffused light. Diffused light thus may be denned as light given by a radiator which subtends a spherical angle equal to a considerable part of the sphere. This makes the term "diffused light" a relative term. Near to a radiator of considerable size, the light given by this radiator thus is largely diffused light, while at considerable distance it is practically directed light, or, in other words, the light given by light sources of considerable size is directed light only at such distances from the radiator at which the law of inverse squares holds; but approaching the radiator so far that this law of inverse squares (flux density inverse proportional to the square of the distance) does not hold, the light approaches somewhat the character of diffused light. The physiological effects, however, during a gradual change from a = 0, or directed light, to a = 180 deg. , or diffused light, apparently do not change uniformly, but new effects appear and others disappear. 126. The main objection to directed light from a single source results from the absence of light in the shadows. Using, how- ever, two or more illuminants, that is, combining directed light of several widely different directions, the shadow cast by one illuminant is illuminated by the other illuminants, and thus an effect produced very similar to diffusion. Thus with two light sources, at a point at which both light sources give the same illumination, the intensity in the shadow cast by one illuminant is still 50 per cent, that is, the illumination the same as if equal volumes of directed and of diffused light were combined, and to a considerable extent the physiological effect is the same. It is not completely so, however. In the illumination by equal volumes of diffused light and directed light from a single source, each object casts a single shadow, in which the illumination is reduced to half. When producing an equivalent diffusion by two light sources, an object casts two shadows, in which the illumination is reduced to half (if the two light sources give equal illumination), but, where the shadows overlap, a perfectly black and lightless shadow is produced. The more the two 282 RADIATION, LIGHT, AND ILLUMINATION. half shadows overlap to a complete shadow, the less the combina- tion of the two light sources is equivalent to diffusion. At the same time, occasionally the existence of two or more half shadows and of their compound shadows may assist distinction, and thereby be advantageous. In short, there is a vast and largely unexplored field in the physiology of illumination, which the illuminating engineer will have to study and investi- gate. While one point source of light gives directed light, two sources at distances from each other give an effect equivalent to diffusion, and three or more sources still more so, until in the theoretical case of an infinite number of point sources distributed through space — or, practically, a very large number of distrib- buted illuminants — we get perfect diffusion. With a change from a single to a very large number of illuminants, the illumi- nation thus changes from directed to diffused, and thus, for a moderate number of illuminants, is intermediate between directed and diffused, but nevertheless this intermediate state is physiologically of entirely different character from that given by a single illuminant of very large diameter, that is large angle a, as discussed above. 127. We thus have true diffused light, as daylight, the equiva- lent diffusion given by the combination of several light sources, which depends on their relative location, and the equivalent diffusion given by a large relative diameter of the light source. The latter again varies with the shape of the light source, and in extreme cases, as a linear straight radiator, as a Geissler tube (Moore tube), we may get an illumination which, at any point of space, is practically diffused in one direction, and practically directed in a direction at right angle to the former. In such cases we again get different physiological phenomena. For instance, a straight rod, held parallel to the radiator, casts a sharp black shadow — directed light — while, when held at right angles to the radiator, it casts no shadow — diffused light. With objects of more irregular shape, it can be seen that the shape and appearance of the shadows give a rather interesting problem, and the physiological impression made by such illumination thus is different again, from that of ordinary directed or diffused light or their combination. In general, wherever two or more illuminants are used, the PHYSIOLOGICAL PROBLEMS. 283 physiological effect depends on the relative position of the light sources to the illuminated objects, irrespective of the intensity of illumination. Thus, for instance, in the illumination shown in Fig. 117, on the same curve of equal illumination, the physiologi- cal effect is not constant, but varies from point to point. On the curve of 850 near the center of the room, an object casts four shadows of approximately equal intensity, in different direc- tions. The shadows are sufficiently marked to assist in seeing, and the illumination in the shadow is quite high ; thus the illumi- nation is very satisfactory. On the same curve 850, near the edge of the room, the four shadows fall in nearly the same direction, only one is marked, and by the overlap of the shadows a large compound shadow is formed, in which the illumination is very low, distinction difficult, and the illumination thus unsatisfactory. Thus with the same physical value of illumina- tion, on the same curve 850, the physiological effect in this case changes from a very satisfactory illumination at one place, to a quite unsatisfactory illumination at another place. Thus, in this instance, while the solution of the illuminating problem, given in Fig. 117, is physically perfect, that is, the illumination very uniform throughout the entire room, and the efficiency high, physiologically the illumination is satisfactory only in the middle of the room, but becomes more and more unsatisfac- tory the further we go outside of the square formed by the four light sources. Physiologically the illumination would probably be improved by locating the light sources in the four corners of the ceiling, or in the centers of the four sides of the ceiling. Physically, this arrangement of lamps in the corners of the room would greatly reduce the efficiency, thus require either more power, or lower the average illumination; the arrangement of the lamps at the sides would decrease the efficiency less, but would considerably impair the uniformity of illumination, giving a lower illumination near the corners of the room. Furthermore, in illuminating engineering, enters as an impor- tant and largely unknown factor, the effect on the physical and physiological illumination, of the objects in the illuminated space, and of the observer] that is, the light flux distribution and its physiological effect, as depending on the location of light sources and distribution of their light flux through the illuminated space, is not sufficient to solve the problem of 284 RADIATION, LIGHT, AND ILLUMINATION. illumination, but consideration must be given to the changes resulting from the use of the illumination. For instance, in the illumination shown in Fig. 117, and discussed above, the diffused light, 0.250, resulting from reflection from walls and ceiling, is quite considerable, and would be nearly sufficient for giving distinction in the compound shadow of all four illumi- nants, as it exists in a pronounced degree near the walls. Thus even there the illumination would be moderately fair. How- ever, when relying on this diffused illumination to see in the shadow of objects close to the walls, it may not be present, or largely reduced by the shadow of the observer, since, as seen above, diffused light also casts shadows, though the blur at the edges of these shadows is such as to make them very little noticeable. Thus, when approaching close to the walls to look at an object, we may find it shaded from the direct light and from most of the diffused light, thus giving unsatisfactory illumination. Locating the light sources in the corners or the centers of the sides of the room, we get pronounced shadows of the objects located against the walls of the room, and thereby again unsatisfactory illumination, although in this case, physio- logically, considering merely the room without the objects which may be located in it, the illumination would be satisfactory. Thus we may have to sacrifice uniformity of illumination still further, by arranging five light sources, four in the corners of centers of the sides of the room, and one, of larger light flux, in the center of the ceiling. Thus, occasionally, illuminations designed for uniform flux density are not satisfactory, even though the proportion of directed and of diffused light, and the direction of the directed light, is physiologically correct, because the changes resulting from the objects in the room, and the person of the user of the illumination, are not sufficiently considered. 129. The cause of most of these difficulties in dealing with illuminating problems is that, physiologically, light is not a vector quantity; that is, light flux densities cannot be combined by the parallelogram law. Two magnetomotive forces A and B, Fig. 124, acting on the same point P, combine by the parallelogram law to a resultant C; that is, the combined action of A and B is identical with the action of a single m.m.f. C. Thus the m.m.f. existing at any PHYSIOLOGICAL PROBLEMS. 285 point P of space is perfectly characterized by two quantities only — the resultant intensity, C, and its direction. If, however, in Fig. 125, A and B represent the two light flux densities produced at point P by two light sources I/t and L2, their physiological and also their physical action may be entirely different from that of one light flux C derived by combining A and B by the parallelogram law. FIG. 124. FIG. 125. In some respects the action of the two separate flux densities A and B is the same, or nearly the same, as that of a resultant flux density C; the illumination of an opaque plane a, located so that both light sources Ll and L2 are on the same side of the plane, is the same. If, however, the illuminated plane is trans- parent or translucent, and also in regard to the effects of polariza- tion, reflection etc., the effect of the two separate flux densities A and B differs from that of a single resultant C. Entirely different is the effect if the light sources Lt and L2 are on dif- ferent sides of the plane. Thus, with a plane c located in the direction C, the resultant flux density C would give no illumina- tion, while in reality by A and B both sides of the plane are fairly well illuminated. Thus, with the plane in any direction within the angle cu between PL2 and PA, it receives the same amount of light from A and B as it would receive from (7; but in any direction within the angle T = 7t — to, between PA and PB, it receives more light from A and B than it would receive 286 RADIATION, LIGHT, AND ILLUMINATION. from the resultant C, and receives infinitely more light in the direction c (that is, in this direction it receives no light from C). Within this angle T, both sides of the plane are illuminated by A and B, which obviously is never possible by a resultant vector C. In the illumination of a plane, the differences between the ac- tual illumination by A and B and the illumination which would result, if light were a vector quantity, by (7, are only those of intensity of illumination. With an object of different shape, however, the phenomenon becomes far more complex. Thus the illumination of a sphere S by the resultant C would be as shown in Fig. 126, — half the sphere dark, the other half light, and with a maximum intensity at c, shading off towards zero at the termi- nator mn. The actual illumination as shown in Fig. 127 gives a FIG. 126. FIG. 127. black segment of angle