In 1883 and 1884 examination of several types of commercial window-glass was made by Mr. Edward Atkinson, who was apparently the first to recognize the immense gain in effective light in rooms when lighted by windows which were roughened so that the smooth incident sky light or sun light did not pass directly through them, but was diffused in directions in which it otherwise did not go. In the summer of 1894 Mr. Atkinson laid before me the problem, requesting an examination of some dozen or more different kinds of glass. The hopelessness of trying to get something for nothing, that is, to get a sheet of window-glass to throw into a room more light than fell upon it, appeared so plain to me that I made all my preparations to measure not a gain but a loss of light in using Mr. Atkinson's samples. The results of my tests at that time, published in Circular No. 67 of the Boston Manufacturers Mutual Fire Insurance Company, showed how real was the gain in the amount of effective light in small rooms when glazed with a roughened or corrugated glass. The finely corrugated or ribbed glass, described more fully later, was found to give the greatest increase, and has since been widely used under the name of "factory ribbed" glass. Just as I was completing my tests the now common prismatic glass began to be put upon the market and I tested it, finding it slightly less effective as a diffusing medium than the "factory ribbed" glass. An examination of the prisms then used and those now furnished by the company show a considerable gain in efficiency in the newer ones.

The results of the tests on a score or more of different glasses may be stated briefly as follows:—

We may increase the light in a room 30 feet or more deep to from 3 to 15 times its present effect by using "factory-ribbed" glass instead of plane-glass in the upper sash. By using prisms we may, under certain conditions, increase the effective light to 50 times its present strength. The gain in effective light on substituting ribbed glass or prisms for plane-glass is much greater when the sky-angle is small, as in the case of windows opening upon light-shafts or narrow alleys. The increase in the strength of the light directly opposite a window in which ribbed glass or prisms have been substituted for plane-glass is at times such as to light a desk or table 50 feet from the window better than one 20 feet from the window had previously been lighted.

The samples of glass tested were of two distinct types: First, glasses which were roughened or ribbed, primarily to blur distinct vision, and which happened to be of service as diffusing media as well; second, the special prismatic forms, designed especially to divert light of windows from the original downward direction to one more nearly horizontal. The first group consists of comparatively inexpensive glass made in large sheets, and it has often a pronounced greenish tint, which varies with different makes, the white glass being the most desirable. The following samples were tested, their cross-section being shown in Plate I:—

¹ Continued from No. 1298, page 48.

Of these several specimens, one or two may be dismissed with brief mention. Ground-glass is of little value, except as a softening medium for bright sunlight. Its rapidly increasing opaqueness with moisture and dust makes it undesirable as a window-glass. The common rough plate has very little action as a diffusing medium, giving no perceptible change in the effective light. "Baird's hand made silver glass" has great value as a diffusing medium in small rooms with nearly open horizon. Of the ribbed glasses, the fine "factory ribbed," with 21 ribs to the inch, is distinctly the best, not in all probability because of the fineness, but because of the greater sharpness of the corrugation. The "ribbed wire" glass is about 20 per cent less effective than the ordinary "factory ribbed" glass. The addition of a second corrugation upon the back of the plate giving the "skylight" and "washboard" glass is of no apparent value. The raised pattern imprinted upon one surface of the glass, as in the case of the "Maze," gives the widest diffusion, especially in bright sunlight. A raised figure, when worked upon the back of the "ribbed" glass, renders it less offensive to the eye in bright sun light, but less effective in deep rooms. The only glasses of this group which it is worth while, then, to discuss further are the "factory ribbed" and the "Maze" glass.

The second group comprises the following glasses:—

The Luxfer prism shown in Plate I, Figure L consists of a plate smooth one side and deeply notched upon the other, the teeth or prisms being of very flat, smooth faces and of brilliant appearance. The glass being white, and the prisms used in canopies and in the major part of the vertical glazing are made in tiles or plates about 4 inches square. Tiles are built up in large sheets in frames of copper or brass, so made as to give to the sheets of tiles a strength and durability far in excess of a single sheet of the same size. The Luxfer prisms are now being made for factory use in large sheets, as well as in the small tiles. The Solar prisms are made in small tiles, which are held together in a metal frame to make large sheets. The main difference between the Solar and Luxfer prisms is that the under face of the former prism is curved instead of plane, as is shown in Plate I, Figure S. The Daylight prisms tested were made in large sheets and of approximately the same cross-section and

Fig. 4

general appearance as the Luxfer prisms for factory use. No tiles of Daylight prisms were tested, as none came to hand in time for the test. The Mississippi prism glass is much like the other prisms in cross-section, but the ridges or prisms do not run across the plate in a straight line, but in a wavy or sinuous line. I cannot detect any advantage arising from this over the straight-edge prism.

The room in which the tests were made is the Physics Lecture-room in the Walker Building of the Massachusetts Institute of Technology. The room is 53 feet deep and 41 feet wide and has a sloping floor, as may be seen in the accompanying Illustrations, which were taken from near the window. In the middle of the west side, at a height of 8 feet above the floor, two openings 12 inches square were left in a large window. All other openings by which light might enter were closed. Over these two windows a shutter might be rapidly slid, allowing light through one or the other opening, as desired, and the effect of the difference in glass in the two windows noted.

A small screen of white paper was arranged upon a system of trolley-wires, so that it might be moved about and examined at points in a plane 6 feet below the centre of that window. A radial arm 15 feet long was also arranged to carry a white paper screen in a vertical or horizontal plane for examination of the intensity of the light at different angles. In a window some 15 feet from the one in which the test specimens were tried was placed the photometer, the great rapidity of change in the brightness of the daylight, at times, making it necessary to have for a standard light something which would vary directly with the light at the test window. I therefore made use of a modified Cornu "cat's-eye" photometer, the standard of light being a piece of white porcelain so set at the window as to be under the same conditions of light and shade as were test specimens. This enabled us to determine the intensity of light in percentages of light at the window. Examination of the data gathered on 16 days, between August 15 and September 15, shows that the probable error arising from uncertainty in the setting of the photometer is less than 5 per cent. In order that no error might arise from varying outside conditions which might favor one glass momentarily, the screen upon which the intensity measurement was made was kept in one position while all the glasses tested were placed in the window in rapid succession, the screen being then moved to a new position and the series repeated. When the intensity of light falling on the screen was very small, it was necessary to insert calibrated neutral-tint glasses between the porcelain plate and the telescope. In the diagram of the photometer, Figure 4, the window is shown at W, the screen at S, the porcelain plate at G, the "cat's-eye" at C, the eye-piece at E and the telescopes at T and T'. For the benefit of those who are not particularly familiar with photometers, it may be well to state that the apparatus

possible to restore the light by the use of ribbed glass, and to even improve its earlier condition by a prism canopy. The assumption is here made that the obstructing building is at least thirty feet distant. Figure 5 is an attempt to show this graphically by exhibiting sectional diagrams of two mills with a street thirty feet wide between them. On floor B we should receive as effective light with ribbed glass as floor A received with plane-glass. Or, again, the obstructing building shown at the right may be raised two stories in height and we could still have as good a light as we have at present with the plane-glass. The plotted curves show the great increase in efficiency of ribbed glass and prisms as the sky-angle diminishes. The intensity of the light is plotted vertically and the angular height of the obstruction horizontally. Rooms with windows opening upon light-shafts and narrow alleys with very limited sky, where the available light is now small, may have the light twenty feet back from the window increased ten or twenty times by using prisms; and, by using canopies of prisms, it is sometimes possible to strengthen the light from fifty to one hundred times. The data on the varying sky-angle was obtained by placing outside the window, at a distance of about five feet, a vertical curtain, which might be raised to cut off the sky-light at any desired angle, side curtains being provided to cut off the light at the ends of the main curtain.

The accompanying photographs show the Physics Lecture-room with its one window twelve inches square glazed in different ways. These were taken when bright sun was shining upon the window, a condition especially favorable to the glass of the "Maze" type. All were exposed and developed and printed under exactly the same conditions. Figure 7 shows the relative sizes of windows glazed

here used is merely a system of lenses and mirrors, by means of which the standard source of light, the white porcelain in the window, may be seen as if side by side with the screen whose brilliancy it is desired to determine. The standard is then obscured gradually, by closing the "cat's eye," until the two images appear of the same brightness. From the area of the opening of the "cat's eye" we may then compute the relative brightness of the standard light and the object examined — in this case, the white paper screen. No measurements were taken when the direct sunlight was upon the window, the slightest variation in the position of the sun causing such a change in the distribution of the light as to make all attempts at measurement hopeless. Moreover, the exceeding brightness of all the specimens except the "Maze" and ground-glass, when in the bright sunlight, is such as to forbid their use where they are constantly in sight, unless they be provided with shades of thin white cloth. This practice of using a thin white shade is to be commended, as it gives a maximum of light when it is needed, in dull weather, when the shade need not be drawn, and yet keeps the eyes protected from a painful glare in bright sunlight. The white cloth shade cuts off about 60 per cent of the light.

As the result of the test made in a room fifty feet by forty feet, we are able to draw the following conclusions:—

First. — The conditions in a room less than fifteen feet deep are such that, except with a sky-angle of less than forty-five degrees, it is not advisable to alter the general course of the light by using a prismatic or ribbed glass. A nearly hemispherical diffusion, such as is given by the "Maze" or "silver ripple," is ordinarily preferable.

Second. — When a room is from twenty feet to sixty feet deep, or even more, and has a sky-angle of sixty degrees or less, the ribbed and prismatic glass gives a very great gain in effective light. The gain in brilliancy is such as to make a basement with prism canopies as light as a second story without them, and the first story with ribbed glass should be distinctly brighter than the second story with plane-glass. If the building which obscures the light from the sky at a window be increased two stories in height, it will be found

with plane, ribbed or prismatic glass, which would give the same effective light in a room of about fifty feet deep, with a sky-angle of sixty and thirty degrees.

The Luxfer factory-prism and the Daylight prisms for large sky-angle came to hand too late to be examined as carefully as were the other specimens. With sky-angles of thirty degrees or less, and in deep rooms, the relative efficiency of the prism-tile increases greatly.

Figure 6 shows the average distribution of light in a room glazed with plane, "factory ribbed," "Maze" or prismatic glass. It shows more clearly than would a lengthy description the rearrangement of the light resulting from the use of the diffusing or reflecting glasses. The interior of the room is shown in perspective, as if one side were removed.

The refraction of the incident ray in a case of the ribbed glass and prism is shown in Figures 2 and 3, Plate 1.

"Ribbed" and "Maze" glass are of very great value in softening the light, especially in the case of such windows as are exposed to the direct sun, aside from their effectiveness in strengthening the light at distant points. With the "Maze" glass, the artist may have, in all weather and in all directions, what is in effect the much desired "north light." The photographer may have in this way as well diffused a light as he now has with cloth screens or shades, with a much greater intensity. To be efficient in rooms of twenty feet deep or more, ribbed glass should be set with its ribs horizontal, and where the sunlight falls upon it, it should be provided with thin, white shades. All inferences drawn from the test are made upon the assumption that the windows are to be reglazed with diffusing glasses only in the upper half, which is the common practice. If the lower sash is to be reglazed as well, a further increase of about twenty-five per cent may be expected.

The question of the effect of whitening the ceiling cannot here be discussed, but it is certainly much more effective when the window is set with prisms or ribbed glass than when plane-glass is used.

The matter of the selection of one or another of the various types of glass for any particular window can be made in this report only

in a general way. Great difficulty in rolling sheets of sharp prisms has made the use of a small moulded tile imperative until very recently, and what small difference there is between a rolled sheet prism and the tile is due to the deforming of the prism in rolling, thereby causing some light to be thrown in directions not desired. Considering both expense and efficiency, the following general suggestions are given:—

Use "Maze" or "Baird's hand-made silver" glass in small rooms or offices not more than fifteen or twenty feet deep.

Use "factory-ribbed" glass in rooms thirty to fifty feet deep, with sky-angles of sixty degrees or more.

Use "Luxfer," "Daylight" or "Solar" prisms, or "factory-ribbed" glass, in sheets, in all vertical windows in rooms more than fifty to sixty feet deep, with sky-angle of less than forty-five degrees. With a sky-angle of less than thirty degrees, use "Luxfer," "Solar" or "Daylight" prisms in canopies.

It must be borne in mind that one factor, which can be merely hinted at in this report, may be the one which decides the matter of the selection of any one glass, that is, the cost. At present prices we may assume that the "factory-ribbed" glass costs but little more than ordinary double-thick plane-glass. The cost of rolled prismatic glass is not yet established. The cost of the cast prismatic glass is of necessity much greater than of either of the others. It is to be hoped that some systematic attempt will now be made by glass manufacturers to furnish a "prismatic ribbed glass," if the term may be used, consisting of a sort of prism with a rounded edge made at a cost commensurate with that of the ordinary ribbed glass. This glass should be made in large sheets, and have the ribs or prisms of several different angles for use with different sky-angles. Such glass would often have many advantages over the sharp-edge prism and the rounded corrugation.