THE PHYSICAL PHOTOMETER IN THEORY AND PRACTICE.* BY W. W. COBLENTZ, Ph.D., Associate Physicist, Bureau of Standards. I. It is well known that the photometry of light sources which are different in color is subject to great errors. In order to obviate this difficulty some form of radiometer which, unlike the eye, functions independently of the frequency (wave-length) and the intensity of the stimulus, has been advocated. A bolometer or a thermopile is the simplest radiometer now available. In order to make it conform with the behavior of the average normal eye when subjected to a stimulus which does not deviate much from a certain specified intensity, it is necessary to interpose between the radiometer and the source of light (which is to be tested) an absorption screen which is opaque to all the infra-red and ultraviolet radiations and which transmits the visible radiations in proportion to the luminosity curve of the average normal eye. In other words, the spectral transmission curve of the absorbing medium must be an exact copy of the luminosity curve of the average normal eye, when subjected to a stimulus of a given intensity. All this sounds very enticing in theory, but in practice the radiometer has various limitations which, as will be shown presently, may restrict its use to special problems which can be investigated at leisure, when conditions are favorable for operating the radiometric outfit. This paper makes no pretence in giving the details of the development of the aforementioned ideas. In the earliest work by Féry1 an absorption cell containing a solution of copper acetate was used, and no attempt was made to have its transmission curve coincide with the luminosity curve. Recently Karrer 2 has made a series of radiant luminous efficiency measurements on various light sources by determining the ratio of energy transmitted by the luminosity curve solution to the total energy radiated. The solution used by Karrer has been modified, and the one now recommended has the following composition: Cupric chloride .... Cobalt ammonium sulphate Nitric acid, sp. gr. 1.05 Water to one litre of solution. 60.0 gr. 1.7 gr. 7.5 gr. 15.0 Cc. This solution is to be used in a parallel-walled glass tank I centimetre in thickness, but, as will be shown presently, this is not a sufficient thickness to be entirely opaque to infra radiations. Fortunately the solutions used by Karrer were in separate tanks, forming a total thickness of 4.2 cm., which, as will be shown on a subsequent page, is sufficient to eliminate all the infra-red. 5 The question of the mechanical valuation of visible radiation, "light," by using a radiometer and an absorbing screen whose transmission curve coincides with the luminosity curve of the eye has been discussed by Houstoun, Ives and others. Ives has recently discussed the subject of physical photometry, and in his instrument the visual luminosity medium is complete in one solution. Unfortunately no mention is made of the thickness of the cell, which, from the printed illustration, appears to be I cm. in thickness. The present inquiry into the merits of the physical photometer came about quite accidentally in connection with the question of the photometric determination of the transmission of a yellow and a blue solution for the luminous radiations from a "4-watt" carbon lamp. The yellow solution consists of 72 grammes of potassium dichromate per litre, and the blue solution contains 53 grammes of copper sulphate per litre; these being the concentrations specified by Ives and Kingsbury to give the same 3 Ives, Coblentz, Kingsbury, Phys. Review 5, p. 269, 1915. 4 The concentration of cupric chloride in the aforementioned luminosity solution amounts to only about 3 per cent. for a 2 cm. thickness. (Beers's law. Thickness and concentration are interchangeable.) In a previous paper (Bull. Bur. Standards, 9, p. 110, 1912) I have shown that a 3 per cent. solution is not sufficient to absorb all of the infra-red spectrum. 7 Ives and Kingsbury, Trans. Illum. Eng. Soc., 10, p. 203, 1915. transmission (for the light from a "4-watt 4-watt" carbon lamp) as determined with a "physical photometer" (to be described shortly), which presumably employed the aforementioned luminosity solution. The copper sulphate solution is opaque to radiations beyond 0.9μ, and the dichromate solution is opaque beyond 1.4. Consequently the copper sulphate solution will absorb most of the radiations transmitted (at 1.2μ) by the luminosity curve solution; but they will not be absorbed by the dichromate solution. Hence, in making up a copper sulphate solution to give the same radiometric transmission as that given by the dichromate solution, the sulphate solution is to be expected to be the more transparent Showing the region of transparency in the spectrum of the luminosity curve solution. when measured radiometrically by means of a physical photometer employing a luminosity curve solution which is (as it should be) entirely opaque to infra-red radiations. In order to determine the validity of this surmise, the spectral transmission of these various solutions was determined by means. of a fluorite prism and the spectro-radiometric apparatus used in various researches previously published. The galvanometer was steady to 0.1 mm. The Nernst glower used gave from 200 to 300 cm. deflection, so that it was an easy matter to make observations to 2 parts in 30,000 in the spectral region of 1 to 2μ. The luminosity curve solution was found more transparent (Fig. I, curve a) than anticipated, for a 1-cm. layer, and the greatest surprise was the observation of an additional transparent region at 1.7. This band is found in pure water (i.e., it is not caused by the coloring matter in solution), and is not entirely eliminated in a 1-cm. layer. This is shown more clearly in an examination of pure water, curves e and f, Fig. 1, where this transparent region is entirely eliminated (curve f) in the 2-cm. layer of water. The depression in the left-hand side of curve a, Fig. 1, was verified on a subsequent examination, from which it would appear that Beers's law (which states that the thickness and concentration are interchangeable) may not hold for highly-concentrated solutions. Curve a is entirely different from the transmission curves of cupric chloride previously published. Curve b, which gives the transmission of the luminosity solution diluted Showing the percentage of infra-red radiations as compared with the luminous radiations from a " 4-watt "carbon lamp which are transmitted by a layer of luminosity curve radiation, 1, 2, 3, etc., centimetres in thickness. one-half and examined in a 2-cm. layer (hence equivalent to a 3 per cent. solution), agrees exactly with the data of a 3 per cent. solution of cupric chloride previously published. Curves c and d, Fig. 1, give the transmissions of the above-mentioned luminosity solution diluted to one-third and one-fifth of its concentration, and examined in layers 3 and 5 cm., respectively, in thickness. This was done in order to determine what thickness of water (solution) would be necessary to eliminate the transparent band at 1.2 in the luminosity curve solution. This could be accomplished by using a greater concentration of cupric chloride, but this would necessitate working out a new luminosity curve solution. Using a cell 5 cm. in thickness, and a concentration of one-fifth of the prescribed ingredients of the above-mentioned luminosity curve solution, the amount of infra-red is from 0.6 to 0.8 per cent. of the visible radiations transmitted from a " 4-watt lamp. However, before using this solution it will be necessary to determine first whether the transmission curve has been appreciably modified, in the visible spectrum, by diluting the standard solution to one-fifth and then using it in a 5-cm. layer, instead of 1 cm. as prescribed. In Fig. 2 are given the percentages of infra-red radiations as compared with the visible radiations from a "4-watt" carbon lamp, which are transmitted by a layer of luminosity curve solution, 1, 2, etc., centimetres in thickness, but diluted as described. These data were obtained by integrating the spectral energy curves of a " 4-watt" lamp. In view of the fact that the luminosity solution, when diluted, became more transparent than was expected, at I to 1.2μ, an examination was made to determine what thickness of pure water when used, in a separate cell, with the original concentration of luminosity curve solution (used in a cell 1 cm. in thickness), would eliminate the transparent band at 1.3μ. As shown in curve g, Fig. 1, the combination of a cell of pure water, 3 cm. in thickness, with a cell of the undiluted luminosity curve solution, I cm. in thickness (the two cells forming an absorbing layer 4 cm. in thickness), is far more opaque at 1.2μ than is the 5-cm. layer of diluted luminosity solution. This is due to the great opacity of the luminosity curve solution at 1.1μ. In practice, however, it seems desirable to use a cell of water 4 cm. in thickness, combined with a cell 1 cm. in thickness containing the luminosity curve solution in its full concentration as originally prescribed. This should insure the elimination of all infra-red radiations. In Fig. 3 are given the transmission curves of various substances discussed in this paper, the cells being 1 cm. in thickness. Curve a shows the transmission through the undiluted luminosity curve solution, the region at 1 to 2μ being plotted to a larger scale in Fig. 1. Curve b gives the transmission of copper sulphate (53 grammes per litre); and curve c gives the transmission of potassium dichromate (72 grammes per litre) mentioned on a previous page. The transmission of cobalt ammonium sulphate (6 gr. in 94 Cc. of water; 6 per cent. solution) is given in VOL. CLXXX, No. 1077-25 |