THE DIRECT CONCAVE GRATING SPECTROSCOPE.

THE ULTRA-VIOLET HYDROGEN SERIES.

THE GREAT NEBULA OF ORION.

By S. A. MITCHELL.

A NOTE has been published in this JOURNAL regarding the application of the concave grating to stellar photography.' The grating has been tried in several different ways, but the best results are obtained by using it as a direct grating spectroscope; that is, the light from the star falling directly on the grating, is diffracted and brought to a focus on the photographic plate.

This method was tried successfully at the Johns Hopkins University (loc. cit.). A small grating of one meter radius of curvature, with 15,000 lines to the inch, was employed. Although this grating had a ruled surface of only 1X2 inches, a photograph of Sirius was obtained which showed sixteen lines of the hydrogen series. This seemed so promising that a large grating was made, having a ruled surface of 2×534 inches. This grating has a radius of curvature of one meter, and is ruled with 7219 lines to the inch. It was mounted on the 91⁄2-inch Hastings refractor of the Johns Hopkins University. Experiments were continued, the results obtained being very good, considering the situation of the Observatory. The telescope is on the sixth floor of the Physical Laboratory, which is continually subject to the jars, the dust, and the glare of the city.

In November 1898 an opportunity was presented, through the kindness of Professor Hale, of working at the Yerkes Observatory. The spectroscope was mounted on the 12-inch Brashear refractor, and experiments have been continued through the past winter and spring.

The mounting for the grating is exceedingly simple, 'POOR and MITCHELL, "The Concave Grating for Stellar Photography," this JOURNAL, 8, 157, 1898; Monthly Notices, March 1898.

geocentric polar coördinates by R and 0. Then x, y, z, x', y', and 'may be expressed as follows:

Apply the proper subscripts and substitute these equations in (15). Then, for abbreviation, let

Expanding the determinant, and carrying out the computation in the original quantities, it is found, after rejecting common factors, that

μ P2+P1 {sin ẞ, sin ẞ, + cos ß, cos ß, cos (1, − d1 ) } + v p1⁄2

2

2

1

2

2

- R1 cos ẞ, cos (λ, — ¤ ̧) — μ R, cos ß, cos (λ, — © 2)

1

2

- v cos B, R, cos (λ, — o2) +R, sin (A, 0)

Let

V

2

m=- P2+

R

2

2

cos ß ̧ (R¡ cos (ì ̧ — © ̧)+R ̧ sin (1, − 0,) 0; } . (21)

2

2

Then (20) becomes,

P2 =m+Mp1

1

(22)

where m and M are known quantities. As was shown, Euler's equation, (2), depends upon P1 and P2 alone as unknowns. Therefore, (2) and (22) may be used for the determination of the two geocentric distances.

The equation (20) has been purposely reduced to the same form as that which arises in the theory of parabolic orbits, based on observations of position alone. The quantities m and M are entirely different in this case, but, because of the form of the equations, the solution would not be different from that given in Oppolzer's Bahnbestimmung der Kometen und Planeten, I, PP. 303-307. Therefore, it is not necessary to consider it here.

When this work is carried out, two positions of the comet are known, and the elements can at once be derived by familiar processes. Thus, the formulæ and references given in this paper are sufficient for the determination of the first approximation to the parabolic elements of a comet's orbit, when the computer has at his disposal two observations of position, and one of the motion in the line of sight.

THE UNIVERSITY OF CHICAGO,
May 20, 1899.

THE INFLUENCE OF THE PURKINJE PHENOMENON ON OBSERVATIONS OF FAINT SPECTRA.

By W. W. CAMPBELL.

THOSE who are familiar with the details of the spectrum observations of the Orion Nebula will, on reflection, easily convince themselves that the Purkinje phenomenon has very little to do with estimates of the relative intensities of the three principal lines. There remains the equally important question as to whether it enters the problem to any extent whatsoever.

This phenomenon seems to be an organic color effect, arising from the observer's effort to compare, numerically, two unlike objects. Professor Runge's laboratory observations showed a small Purkinje effect when he compared spectral regions of wavelengths AX 4862 and 5007; but for the regions AX 4862 and 4959, an effect which was at first apparent was later reduced to Professor Runge's experiments were based upon light which was initially of "medium brightness;" but his published account shows that he was fully aware of the one important condition essential to a laboratory solution of this problem, viz., that the initial intensities of the artificial lines should equal those of the nebular lines.

zero.

Now the absolute brightness of the nebular lines is surprisingly low. [For an approximate measure of the intensity of the Hẞ nebular line, see my paper in the May number of this JOURNAL.] The nebular lines are so faint that many-and I think all-observers are unable to distinguish between their colors. They lie so near the limit of color perception that some observers do not see them as blue-green, but as gray. In this and similar cases, does the Purkinje effect enter at all? Great numbers of observations are made under these conditions and it is important that the question should be settled.

In 1889, Hering and Hillebrand observed that the distribution of brightness in a faint colorless spectrum, as viewed by normal eyes, is identical with the distribution of brightness in a spectrum of every degree of intensity, as seen by totally colorblind observers. [Sitzungsb. d. Wien. Akad., 1889; Wied. Ann., 62, 17.]

About the same time, Professor A. König observed "dass die Vertheilung der Helligkeit im Spectrum in einzelnen Fällen auch dann ungeändert bleibt, wenn durch peripher oder central gelegene pathologische Processe die eigentliche Farbenempfindung völlig verloren geht und nur die Empfindungsreihe SchwarzGrau-Weiss bestehen bleibt." [Wied. Ann., 45, 607, 1892.]

These observations demonstrated that the Purkinje phenomenon is dependent upon color-perception; and that its effect upon observations of the Orion Nebula made by those who see the lines as gray is absolutely zero.

There remains the case of those who are able to perceive that the lines have a blue-green tinge, but who cannot distinguish differences in their colors. If the Purkinje phenomenon is a

differential-color effect - and so it seems to be-this case would be very closely related to the preceding one in which the effect

is zero.

Even if an observer were just able to distinguish differences in the colors of the lines-which I am strongly inclined to doubt the observations of König and others clearly indicate that the effect would be exceedingly slight, and safely negligible.

The critic who assumed that a twenty-five to thirtyfold observed variation "is nothing more nor less than the Purkinje phenomenon," wrote that "Although the lines in question differ but slightly in wave-length, their brightness is so close to the limit of visibility that great differences in relative intensity . . may be produced." [This JOURNAL, 7, 238.] It appears to me that this case of lines situated in the blue-green, which differ but slightly in wave-length, and whose brightness is so close to the limit of visibility, is precisely the case where the principles of color-vision would have justified the prediction of exceedingly