shift of 0.538 A. The results of the comparison are given in the eighth column of Table II, where the AX for these plates is the difference Venus minus sky.

OXYGEN

Five spectrograms of Venus were taken in the region of the a band when the relative velocity of Venus and the earth was -12.8 km, the corresponding Doppler shift being 0.268 A to the violet, an amount sufficient to separate completely the terrestrial components from those of Venus. Although some solar lines of intensity ooo are faintly visible on these spectrograms, no lines are observable where they should appear if produced by oxygen in Venus' atmosphere. The measured wave-lengths of the oxygen lines present on the spectrograms of Venus and the sky and the magnitude of their deviations from the wave-lengths of Table I show, moreover, that the oxygen lines produced by the earth's atmosphere in the spectrum of the planet are not measurably shifted by blending with lines originating in the atmosphere of Venus and wide enough to overlap those of terrestrial origin.

The B band is producible by a much smaller quantity of oxygen than the a band and therefore furnishes a more sensitive test. King has recently shown that 39.4 m of air at 72 cm pressure, equivalent to 8 m of oxygen, give the lines of the B band faintly. His solar spectrograms indicate that the lines of the B band produced by 39.4 m of air are comparable in intensity with solar lines of intensity oo on the Rowland scale. On the spectrograms of Venus in this region lines of intensity I have about the same visibility as the limiting lines on King's spectrograms. To obtain an estimate of the length of an oxygen column which would produce the lines of the B band with an intensity of 1, we have made use of a valuable paper by Jewell' on the variation of the intensity of water-vapor lines with the quantity of water vapor traversed and of the change in intensity of the oxygen line A 6287.953 with zenith distance. Jewell determined the intensities of Fraunhofer lines-ooo to 6- in terms of the Fe line 5930.406. Graphs from these data show that the inten

Mt. Wilson Contr., No. 232; Astrophysical Journal, 55, 411, 1922.

2 Astrophysical Journal, 4, 324, 1896.

sity intervals from ooo to 6 are roughly equal, and that the intensity of the water-vapor line λ 5919.860 is strictly proportional to the amount of water vapor traversed (Figs. 1, 2). A similar proportionality is shown to hold for the oxygen line A 6287.953 and is assumed to hold for the oxygen lines of the B band. The interval from oo to I

0.0

FIG. 1.-Rowland intensities in terms of λ 5930 taken as unity

0.0

0.0

I.O

2.0

3.0

4.0

5.0 inches FIG. 2.-Intensity at the zenith of λ 5919 (wv) in terms of λ 5930 (Fe) for various amounts of precipitable water in the atmosphere expressed in inches.

is approximately 0.16 of the intensity of the Fe line λ 5930. This is equivalent to an increase in the length of path of 16 per cent. Hence the lines of the B band would probably appear on the spectrogram of Venus if the absorbing column contained the equivalent of 9.2 m of oxygen under ordinary conditions of pressure.

Two spectrograms of Venus showing the B band, March 24 and March 27, 1921, were taken when the relative velocity was -10.68 km, the equivalent Doppler effect being -0.245 A. A

similar spectrogram was taken August 14, when the relative velocity was +12.44 km, the Doppler displacement being +0.286 A. With the scale of 3 A per millimeter, the doublets of the B group are widely separated, but in the interspaces no lines are visible, where lines due to oxygen absorption in the atmosphere of Venus should appear, though solar lines of intensity o are present. Whether lines originating in the planet would be completely separated from the terrestrial lines by shifts of 0.264 A and 0.286 A depends on the widths of the lines. It is evident that at most only a small amount of oxygen is traversed in the atmosphere of Venus, and lines produced by it would be faint and narrow. On a plate of the B band taken by King with an air path of 39.4m, the mean width of the single lines is 0.19 A. The mean width of these lines on spectrogram V 359 is 0.27 A, and the Doppler shift +0.29 A, so that the edges of the lines would be separated by 0.06 A. Spectrograms V 301 and V 303 are not so suitable for this purpose, as the continuous background is fainter and the lines somewhat wider and the Doppler shift smaller than for V 359. When, however, the positions of the oxygen lines in these spectrograms are compared with their positions in the sky spectrograms, the measures show a displacement of 0.003 A to the red, whereas the Doppler shift would be 0.245 A to the violet. On V 359, the measured displacement is 0.002 A to the violet, with a possible Doppler shift of 0.286 A to the red. If, therefore, for any cause the lines of the B band produced by small amounts of oxygen on Venus are somewhat wider than those produced in the laboratory, the results show that there is no measurable influence of overlapping of the edges of the terrestrial lines by undetected Venus components.

The length of the path traversed through Venus' atmosphere in a layer of a given radial depth depends upon the phase angle of the planet and the position on the disk of the point from which the light is observed. For a point on the line perpendicular to the terminator and limb, whose distance from the limb in units of the length of that line is d, the zenith distances of the earth, z., and of the sun, z,, are given by

sin z=1-d-d cos i,

Zs=i-ze

where i is the phase angle. Table III gives the total length of the path traversed by the solar beam for different phase angles and different distances from the limb in units of the radial depth. The secant equation used here is of course not applicable just at the limb and terminator.

At the middle point of the visible disk, corresponding to the center of the spectral band, the path for plates V 301 and V 303 was 5.2 times the radial depth of the layer penetrated, and for V 359 it was 2.5 times the depth. On the basis that lines of intensity I would be produced by 9.2 m of oxygen, it follows for V 301 and V 303 that the quantity of oxygen traversed did not exceed the equivalent of a column 2 m long under standard conditions. For light from near the terminator the path traversed in Venus' atmosphere is much longer. With the slit normal to the terminator, as in these observations, one edge of the spectrum corresponds to this longer path. For spectrograms V 301 and V 303 the path for the light at the point halfway between the center and edge of the spectrum was 7.5 times the radial depth of the layer. The edge of the spectrum shows no trace of oxygen lines due to absorption in the planet's atmosphere. Absorption capable of producing lines of intensity I would be expected if an oxygen layer equivalent to I m under normal pressure had been traversed in the passage through the atmosphere of Venus. As the oxygen in the earth's

atmosphere is equivalent to a column 1500 m long, it follows that the oxygen in the path through Venus' atmosphere was less than one-thousandth of that in the terrestrial atmosphere.

WATER VAPOR

Five spectrograms of Venus taken when the relative velocity of Venus and the earth was 12.8 km, the corresponding Doppler shift being -0.252 A, form the basis of the investigation on water vapor. The relative velocity was sufficient to separate completely lines produced by absorbing gases common to both atmospheres. No traces of lines due to the planet's atmosphere are discernible on the spectrograms.

An upper limit for the amount of water vapor in the atmospheric layer on Venus penetrated by the solar beam may be found by the following considerations. Spectrogram V 282 was taken at mean zenith distance 76°. The average quantity of precipitable water above Mount Wilson is 0.69 cm3 (the mean for the days of observation was 0.72 cm). The water-vapor line X 5919.860 was of intensity about 5 on the Rowland scale, and telluric water-vapor lines of intensity ooo and oo can be identified on the spectrum of Venus. Had a like quantity of water vapor been present in the planet's atmosphere above the reflecting surface, the intensity of the component of X 5919.860 due to Venus should have been about 4 as against 5 for the terrestrial line, since the water-vapor masses traversed would have been 2.5 cm for Venus and 2.9 cm for the earth, the two respective paths being 3.6 and 4.1 times that for zenith distances o. For the point half-way between the center of the disk and the terminator the water mass traversed would have been 4.2 cm, corresponding to solar intensity 7 for the water-vapor line λ 5919.860. These results are deduced from the graphs of Jewell's data. Terrestrial water-vapor lines of oo intensity are easily seen on the plate, but no line is visible 0.25 A to the violet of the terrestrial line λ 5919.860, in which position 0.7 cm of water above the apparent surface of Venus would have produced a line of intensity 7. As the water vapor traversed in the atmosphere of Venus was not sufficient to produce the line λ 5919.860 with intensity oo, there must have been

Annals of the Astrophysical Observatory, 3, 189, 1913.