Photometry

Sub-standard Lamps.

The sub-standards used for candle power measurement are usually tungsten filament lamps of spe cial construction such as that shown in fig. 7. Great care has to be taken in the manufacture of these lamps so as to avoid fluctua tions of candle-power due to loose contacts between the filament and its supporting hooks; the bulb, too, has to be specially selected so that there are no striations in the glass to cause slight altera tions of candle-power as the lamp is rotated through a very small angle on either side of the position adopted for the standardiza tion measurement. The candle-power of such a lamp is determined at a standardizing laboratory by comparison with the fundamental standards. The value of candle-power measured is that in the direction normal to the plane of the filament, when a given voltage is applied to the lamp contacts or when a given current is passing through the filament. These quantities have to be very accurately controlled since, in the case of a tungsten filament, a 1% change of voltage, or a o.5% change of current, produces a change of about 3.7% in candle-power.

Flux Measurement.

Owing to the fact that most light sources are now rated in lumens instead of in candles (see the British Standard Specification for Electric Lamps, issued by the Brit. Engineering Stds. Assn.), the measurement of luminous flux has now become an even more important operation in every-day photometry than the measurement of luminous intensity. The fundamental method of finding the flux output of a source is to measure its candle-power in a very large number of directions uniformly distributed in space, and thus to obtain the mean spherical candle-power which, when multi plied by 47r gives the flux in lumens (see p. 84o). This long and tedious process may be avoided in the case of certain sources, such as a tungsten filament vacuum lamp of ordinary filament form, when the light is distributed fairly symmetrically about the axis. Such a lamp can be spun while the photometric measurements are being made, so that the figure of candle-power obtained at any given angle with the axis is, in reality, the mean value in all directions making this angle with the axis. It is there fore only necessary to make measurements in a limited number of directions, say 20, which lie in a single plane passing through the axis of the lamp.

Apparatus suitable for this purpose is shown in fig. 8 which is almost self-explanatory. The lamp is mounted in a holder which can be rotated mechanically at a speed of, say, 15o revolutions per minute. The mirror system can be turned by hand about the horizontal line passing through the centre of the lamp, the photom eter head and the comparison lamp. It can be shown that, if the

directions in which the measurements are made be suitably chosen, the arithmetic mean of the 20 individual candle-powers gives the mean spherical candle-power with satisfactory accuracy. Such a lamp, then, may be used as a sub-standard of luminous flux, and all that is now required is a method for measuring the flux output from more irregular sources by comparison with this sub-standard.

The Sphere.—The method which is almost universally adopted to-day is based on the demonstrable fact that, if a source of light be placed inside a hollow sphere covered internally with some perfectly diffusing coating, the same amount of light is received at all parts of the sphere surface by reflection from all the other parts. The light received directly from the source naturally differs according to the candle-power distribution of the source, but the light which has been reflected once or more often from the walls of the sphere is distributed per fectly uniformly, no matter how unsymmetrical the original dis tribution from the source itself may be.

It will now be clear that, in order to compare the total flux from two sources of light, it is only necessary to place them in turn within a sphere of the kind described, to screen a small por tion of the sphere surface from direct light, and to measure the illumination of this screened area in the two cases. The ratio of the illuminations will then be strictly equal to the ratio of the flux outputs of the sources. There are, of course, certain precau tions which must be observed in the practical application of the method, owing to the unavoidable departures from the ideal con ditions assumed in the theory. The chief of these departures are (i.) lack of perfect diffusion by the internal coating of the sphere, and (ii.) the interference with the reflected flux brought about by the presence not only of the lamp and its accessories but also of the screen necessary to shield the measured part of the sphere surface from the direct light of the lamp.

The illumination of the screened area of the sphere surface may be measured by any convenient method. It is clearly unneces sary that this measurement should be absolute, since only a ratio is involved. One method is to have a small opening in the sphere and to cover this with a window of opal glass or some other translucent material. The luminous intensity of the outer surface of this window in the normal direction is then measured by means of a photometer head and comparison lamp moving on a bench attached to the sphere. Alternatively the sphere opening is left uncovered and the luminous intensity of that portion of the sphere opposite the opening is measured.