that F. Paschen and myself have taken of the red spectrum of argon. Here also the gas was sparked with oxygen over a solution of potash, nevertheless the carbon was not removed. The cyanogen bands also made their appearance in the krypton vacuum tube and all the lines of the red spectrum of argon. Besides I noticed some other bands of which I do not know the origin. They have nothing to do with krypton as they were also observed in argon tubes. At first some of the nitrogen bands were to be seen; but they disappeared after I had run the tube for some time. With low pressure the carbon bands are greatly reduced in intensity; the krypton lines are also weakened and the lines of the blue spectrum of argon make their appearance. With a Leyden jar and a spark gap the spectrum of krypton changes as well as the spectrum of argon. The new lines are mostly in the blue part of the spectrum.

SPECTRUM OF KRYPTON EMITTED BY A VACUUM TUBE WITHOUT LEYDEN JAR AND SPARK GAP.

SPECTRUM OF KRYPTON EMITTED BY A VACUUM TUBE WITH LEYDEN JAR AND SPARK GAP.

To select the lines due to krypton I compared the photographs with photographs taken some years ago by Paschen and myself of the spectrum of argon. On each of the two plates to be compared I removed the gelatine on one side of a straight line, cutting the lines of the spectrum at right angles. The plates were then laid together in such a manner that the parts where the emulsion was left were in contact along the cut but on different sides. In this way the plates can be examined under the microscope and the lines that exist only on one of the plates are detected. None of the lines were measured visually. The

photographs cover the region from λ= 2400 A. U. to λ= 8000 X: A. U. The wave-lengths were interpolated, using the argon lines, some mercury lines and the D lines as standards. For the wave-lengths of the argon lines H. Kayser's1 measurements have been used and some unpublished measurements that F. Paschen and myself have made by comparison with iron lines. In the least refrangible part I used my own measurements of the argon lines, which are also based on Kayser's argon lines of the second order. The photographs were taken with a Rowland concave grating of one meter radius. The mounting has been described in the paper on the series spectra of oxygen, sulphur and selenium.3

2

These lists are, I presume, very far from complete. As long as krypton is so strongly diluted with argon the weaker krypton lines are likely to escape notice, or may even not appear at all.

The analogy of the first spectrum of krypton to the red spectrum of argon is further borne out by the fact that the wavenumbers of several pairs of lines show equal differences. For simplicity's sake I have in the following table not corrected to vacuo, as the correction does not affect the difference of wavenumbers to any appreciable extent.

The deviations from the mean are :0.22, 0.31, 0.10, which correspond to the differences in wave-length : 0.04, 0.06, 0.03. These differences are well within the limits of error.

H. KAYSER, this JOURNAL, 4, 1896.

C. RUNGE, this JOURNAL, 9, 281, 1899.

3C. RUNGE and F. PASCHEN, this JOURNAL, 8, 70, 1898.

RADIATION FROM A PERFECT RADIATOR.

By W. E. WILSON.

ALTHOUGH Kirchhoff introduced the conception of a perfectly "black" body in the deductions of his well-known law connecting the emissive and absorptive power of a body in regard to radiant heat, he seems not to have investigated the subject experimentally, but points out that the law of radiation for a truly "black body" must necessarily be of a simple character.1

During a conversation with Mr. Lanchester in the autumn of 1895, he pointed out to me that he thought if we took a hot enclosure into which there was only a small aperture and measured the radiation passing out through this aperture that the internal walls of the enclosure would behave as a perfect radiator, whether they were a bright metallic surface or coated with lampblack or any other substance.

About the same time Ch. E. St. John also pointed out that in a heated enclosure, such as a fire-clay furnace, metals raised to a red heat appeared of almost equal brightness whether their surfaces were polished or blackened with oxide.

As all investigations up to this time on the laws of radiation were made with the assumption that lampblack was a perfectly black body and therefore a perfect radiator, it seemed of interest to compare the radiation from it with that coming from a hot enclosure with a small aperture, and which would evidently behave as a perfect radiator.

We procured a half gallon tin T, and soldered it by the neck into a large biscuit box B. Some water was placed in the biscuit box and kept boiling with a Bunsen burner so that the tin enclosure was completely surrounded with steam at 100° C.

A Boys' radio-micrometer R was mounted in front of the aperture and suitable screens S were interposed so as to cut off all radiation except that coming from the enclosure through the 'G. KIRCHHOFF, Pogg. Ann., 109, 292, 1860.

aperture. The outside of the biscuit box near the aperture was coated with lampblack, and by slightly moving the box we could allow this blackened surface to radiate to the radio-micrometer instead of the enclosure.

The temperature of this blackened surface of the biscuit box must have been very nearly the same as that of the enclosure, but

we were astonished to find that if we represented the radiation from the enclosure as 100, the radiation from an equal area of the blackened surface was only about 60.

The result of this rough experiment was so interesting that I determined to investigate the law connecting the total radiation and temperature of such a theoretically perfect radiator.

An enclosure was formed of a large plumbago crucible with a cover of the same material. This stood in a Fletcher's gas furnace and could be raised to any desired temperature.

A hole was bored through the walls of the furnace and also through one side of the crucible. This hole was lined with a porcelain tube through which could be seen the interior of the crucible.

A second porcelain tube also passed into the crucible and was used to carry a thermo-electric junction, made of pure platinum and platinum-rhodium. The current from this was measured by a D'Arsonval galvanometer of low resistance, and its calibrating curve, which was practically a straight line, was obtained by inserting the junction in steam at 100° C., in pure lead freezing, and in pure gold freezing. The radio-micrometer was used to measure the radiation coming from the enclosure, but as this instrument was so sensitive as to give a considerable deflection before the crucible was even red hot, some means had to be devised to