willing to admit a sufficient range in density (much greater, however, than that postulated above) for the gases composing the nebula, then we might have refraction effects which would further complicate the problem. The observed widths of the lines seem to be opposed to an assumption of considerable densities for the gases emitting the light.

To what extent anomalous dispersion effects may play a part in this phenomenon is difficult to predict. Although the appearance of the lines is somewhat similar to that of lines exhibiting anomalous dispersion, the evidence does not seem to indicate its existence as an important factor. We look with favor upon the idea of the double lines being produced by a true reversal in virtue of an outer absorbing stratum, but our minds are open to the possibility of some other explanation.

GENERAL CONSIDERATIONS

The prevailing outlines and internal forms of symmetry of the planetary nebulae, as seen in projection on the photographic plates, are approximate ellipses whose eccentricities for the different objects lie between zero and one-half. Several are circular, or nearly so, and N.G.C. 7026 is the most elongated and irregular object thus far observed. The observing programme involves several scores of planetaries, distributed through all galactic longitudes. According to the laws of chance, if the planetaries were very flat rings, or other forms lying chiefly in two dimensions, we should have found a considerable number of long, slender nebular forms, yielding bright-line spectra. Not one such has been found. Therefore, we can scarcely doubt that the bright rings photographed in so many planetaries are the projected effects of ellipsoidal shells of radiating matter. The hypothesis of ellipsoidal shells is of course very old, and has been❤ suggested by many astronomers.57

Rotation effects observed are in general in harmony with the hypothesis that the rotation axes coincide, at least approximately, with the minor axis of figure of the projected images. This is what we should expect from rotating bodies in equilibrium.

It is a strongly marked characteristic of elongated planetary nebulae that the region of the major axis of the projected image is generally of reduced brightness in comparison with the average brightness. This is especially true at the ends of the major axis of figure. It suggests the presence of a dark or cooler stratum of matter prevailing in and near the equatorial plane of the nebula and outside of the visible image, which occults or absorbs strongly the radiations. from the visible ellipsoid. Such an interfering stratum should have its maximum effect at and near the ends of the major axis of visible figure, and this agrees with observed fact. The more elongated the figure, it would seem, the stronger should be the interference along the major axis, for the interfering stratum should be deep radially and relatively thin at right angles to the major axis. In N.G.C. 7026 the equatorial reduction in brightness is very strong across the entire diameter. The spectrographic observations of this nebula show that the two bright lobes parallel to the major axis of figure of the entire nebula are in rapid rotation about the minor axis of figure of the entire nebula. How can we account for the appearance of this object, and its apparent stability, if the dark lane along the major axis is really vacant and not merely apparently vacant?

A point of considerable strength against this hypothesis of the reduced brightness of the major axis zone, including the ends of the major axis, presented by Dr. Curtis on a preceding page (p. 65), is that the number of such nebulae showing the effect seems to be greater than is required by the doctrine of chance distribution of the equatorial planes with reference to an observer at one point in space; we appear to be in or near the equatorial planes of too great a proportion of the planetaries. If in many of those with outlines not strongly elliptical the interfering stratum is of great extent, both in and perpendicular to the equatorial plane, the ends of the major axis could still be appreciably reduced in their quantity of transmitted light, 57 Herschel's Outlines of Astronomy, p. 604, 1849.

and the other parts of the nebular images not be much reduced in brightness, even though the observer be 30° out of the equatorial plane. However, we do not consider that the objection has been fully met by this assumption.

For those planetaries whose spectra have doubled bright lines, giving the effect of a dark line between two bright lines, or of an absorption line superimposed upon a bright band, the absorption line is inclined but little, if any; and that is what we should expect from an outer absorbing stratum rotating more slowly than the visible structure which supplies the bright lines or bands.

It is an interesting question: under the action of what forces is a rotating planetary nebula in equilibrium? If a nebula, such as N.G.C. 7662, consists chiefly of a massive nucleus and a surrounding ellipsoidal shell of matter in rotation, acted upon by gravity alone, how can the rotating shell retain its ellipsoidal form? What keeps the polar regions in their proper positions to comply with the figure of an ellipsoid? Why do the polar regions not "fall in" upon the nucleus in response to its attractive power, for there is no centrifugal force operating at the poles of figure to keep the materials out? Why should not the "meridian" cross-section of the nebular shell resemble the figure, or perhaps better the dumb-bell figure? Does radiation pressure between the nucleus and the shell act effectively to keep the polar regions of the shell from "caving in"? We are not accustomed to think of radiation pressure as capable of exactly counterbalancing gravitation in a gaseous figure in apparent equilibrium. Is the space between the nucleus and the well defined ellipsoidal shell surrounding it an approximate vacuum, or is it filled with an expansive and buoyant medium which plays a prominent part in counterbalancing gravity? It seems difficult to avoid viewing the latter hypothesis with great favor. The reasonableness of this hypothesis is emphasized by Mr. Campbell's observations in 1893 on peculiarities in the distribution of the N, and N, nebulium and the Hẞ hydrogen radiations in the Orion nebula58 and in N.G.C., 418,59 etc., and confirmed by Keeler's visual observations on N.G.C., 418 and color-screen photographs of the Orion nebula in 1899;60 by Mitchell's concave grating spectrogram of the Orion nebula, showing the great extent of area represented by the ultra-violet 3727 A radiations;" by Hartmann's color-screen photographs of the Orion nebula, confirming and extending Mitchell's results;62 by Wolf's and Burns's objective-prism spectrograms of the ring nebula in Lyra, which recorded the elemental and constituent rings as different in size and internal arrangement of intensities; and especially by Wright's recent investigations of the brighter northern planetary nebulae, with both slit and objective-prism spectrographs.65

If the bright-line nebulae are in fact "stratified" as to their more or less elemental gaseous and vapor constituents, it would be interesting to know the principal factors controlling the orderly arrangement. Are the strata ordered as functions of the atomic weights of their substances?

If the solar system has been developed from such a stratified planetary nebula, in harmony with a modified Laplacian hypothesis, we should certainly be justified in speculating upon the assumption that the planets in the solar system may differ greatly in chemical composition. The four outer planets on this hypothesis may well consist prevailingly of lighter elements than the four inner planets. Has such a factor been effective in controlling the evolutionary rates of the planets?

If the light radiations from a nebula are dependable indications of the quantities of materials existing at different radii from the nucleus, then the relatively brilliant shells contain the principal

58 Astron. and Astroph., 13, 384, 1893.

59 Publ. A. S. P., 5, 207, 1893. 60 Astroph. Jour., 9, 133, 1899.

61 Astroph. Jour., 10, 34, 1899.

62 Astroph. Jour., 21, 389, 1905.

63 V. J. S., Astr. Gesell., 43, 283, 1908.

64 Lick Obs. Bull., 6, 92, 1910.

65 Proc. N. A. S., 1, 590, 1915; Lick Obs. Bull., 9, 52, 1917; and in this volume.

masses of matter lying outside the nucleus. The evolution of such a nebula into central sun and attendant planets might naturally result in a very massive sun with small inner planets and large outer planets, as in our solar system.

RESULTS

1. The radial velocities of 125 bright-line nebulae have been determined. See pages 168-169, column "Observed Rad. Vel.," for adopted apparent velocities. The velocities corrected for motion of the solar system, toward apex at R.A. 270°, Decl.+30°, with speed -19.5 km./sec., are in column "V".

= =

2. The integrated radial velocity of the Great Nebula in Orion is +17.5 km./sec. (p. 107); in good agreement with Keeler's result (p. 96) for the bright region immediately preceding the Trapezium region. The integrated velocity with reference to the stellar system is -0.1 km./sec. 3. The radial velocities of 17 nebulae in the Greater Magellanic Cloud, observed by Wilson, lie between +251 km. and +309 km.; average +276 km. See Wilson's paper, following. The mean velocity corrected for solar motion is +261 km./sec.

4. The observed radial velocity of the only known bright-line nebula in the Lesser Magellanic Cloud is 168 km., and the velocity corrected for solar motion is +157 km./sec.

5. In addition to the 18 observed Magellanic nebulae, there are six observed planetary nebulae whose radial velocities, corrected for solar motion, are greater than 115 km./sec. These six are located in a small area of the sky, R.A.=151 to 192, and Decl.=-9° to -38°; see page 171. The exceptional character of the motion of the 24 objects is apparent, and they have been excluded from the statistical studies of the bright-line nebulae in general.

6. The remaining 101 nebulae yield a speed of 28.3 km./sec. for the solar motion with reference to them as a system, assuming the apex to lie at R.A.270°, Decl. +30°; see page 172.

7. Five of these objects are extended irregular nebulae. They yield a speed of -4.0 km./sec. for the solar motion; page 172.

8. The remaining 96 objects are planetary or small nebulae. They yield a speed of -29.6 km./sec. for the solar motion; page 172.

9. Of these 96 objects, 31 have diameters less than 5". The average radial velocity of the 31 with reference to the stellar system is 28 km./sec. The average radial velocity of the other 65 objects, with diameters greater than 5", is 31 km./sec. with reference to the stellar system; page 173. These velocities are about five times the average radial velocities of the Class B stars. 10. The inclusion of the group of 6 nebulae with abnormally high velocities would bring the average velocity of the 36 small nebulae up to 45 km./sec., and the average for the 31+ 65 + 6 =102 planetary or small nebulae up to 37 km./sec.

11. The average radial velocity of the 5 extended irregular nebulae with reference to the stellar system is 11 km./sec.

12. The so-called "stellar" nebulae are grouped in one quadrant of the Milky Way, without exception, and are probably farther from our region of the stellar system than the other planetaries are.

13. The evidence for a Kapteyn preferential motion of the bright-line nebulae is slight (p. 173), if we exclude the group of 6 high-velocity nebulae.

14. Nine nebulae known to have bright-line spectra are too faint for radial velocity determinations with exposures of practicable length at the present time; page 174.

15. Four spiral nebulae are known to have bright "nebular" bands in their spectra; page 174. 16. Of 54 miscellaneous nebulae (pp. 174-176), a very few probably have exceedingly faint bright-line spectra, but the great majority undoubtedly have continuous spectra, possibly with absorption lines.

17. The so-called "stellar" nebulae have appreciable angular diameters; page 176.

18. The radial velocities of a great many parts of the Orion nebula have been observed, in confirmation and extension of Buisson, Fabry, and Bourget's discovery that a considerable range of radial velocities exists. Our results range from +9.7 km. to +24.9 km.; see figures 16 and 17. The excesses and deficiencies of observed velocities with reference to average velocity +17.5 km. are plotted in figures 18 and 19. The results do not favor the hypothesis of a rotation of the great nebula as a whole; the observed differences appear to be local or regional in character. 19. In one region of the Orion nebula, at least, the spectral lines are not monochromatic; possibly the observed radiations proceed from various depths of nebular structure possessing different radial motions; page 99.

2

20. The estimated relative photographic intensities of the nebular lines recorded are set down, in the proper places, for nearly all the spectra observed. A wide variety of relative intensities prevails. The Hß bright line of hydrogen for N.G.C. 40, for example, is a great many times as bright as the N, and N, nebulium lines in the same spectrum, whereas for many nebulae the N1 line is at least ten times as bright as Hẞ. The relative intensities of N1, N2, and Hẞ are very different in many parts of the Orion nebula: the hydrogen lines are relatively strong in the fainter outlying areas-as observed visually in the 1890's. There is little or no evidence that the relative intensities of the N, and N. lines of nebulium vary.

21. Forty-three planetary nebulae were observed with 3-prism dispersion for evidences of rotation or internal motion, and 3 with low dispersion. Of these 46 objects, 25 showed internalmotion effects. Nineteen and possibly 21 of these are interpretable as rotations about axes approximately or roughly perpendicular to the line of sight, and 4 appear to be not so interpretable; page 176.

22. The most elongated planetary nebulae show the highest rotational speeds.

23. In general, the strata most distant from the centers of the nebulae show a reduced rotational speed with reference to the speeds of structures near the centers; a "lagging behind" of the outer strata.

24. The most probable order of mass is deduced for 3 nebulae (p. 177). They seem to be much more massive than the solar system. The probable masses of the other 16 or 18 planetaries seem to be greater than that of the solar system; page 177.

25. The spectrograms of 10 planetaries show the bright lines of nebulium-and the Hẞ line of hydrogen when it is bright enough to be recorded-as doubled lines, at least for the inner structures of the nebulae. The cause has been sought, but it is not known with certainty. In some cases, perhaps not in all, a reversal due to an outer, cooler, more slowly rotating stratum of nebulosity appears to be the most probable cause among the many causes considered; pages 178-180.

26. The forms of planetaries observed by us favor, in our opinion, the hypothesis that they are approximate ellipsoids of revolution, in general larger than they appear to be, especially in their equatorial regions, by virtue of invisible cooler strata lying outside the visible structures; pages 180-181.

27. The volumes between the nebular nuclei and the bright rings (assumed to be ellipsoidal shells in reality) are probably filled with expansive gases or vapors which play an important part in maintaining the ellipsoidal shells in equilibrium; page 181.

28. The scarcity and the high speeds of the planetary nebulae are unfavorable to the view that the stars in general have evolved from planetaries.

29. If the solar system has been evolved from a typical planetary nebula, we should perhaps have been justified in predicting that the planets near the Sun would be of small mass, and those far from the Sun of relatively great mass.

30. A further expected result might well be that the chemical composition of the minor planets would be different from that of the major planets.

February 1, 1918.