eruptions incessantes des protubtrances, doit augmenter in- cessamment, la pression de l’atmosphere doit aussi s’ac- croitre au-dessus de ces latitudes et peut atteindre une telle valeur, qui pourrait empCcher sensiblement les truptions ulttrieures des protubtrances. Alors arrivera un affaiblisse- ment progressif de leur activitt, qui ameoera P un mioimum des taches. Pendant ce temps l’atmosphere solaire au-dessus des taches, en recevant peu de mattriaux des eruptions, devra, par suite de l’incessante perte de la matibe dans I’espace, produite par la pression de la radiation, devenir peu B peu moins dense, la pression au-dessus des taches s’affaiblira, l’activitt des protubtrances augmentera et un maximum des taches arrivera de nouveau. Ensuite les pht- aomenes marcheront dans le mCme ordre.

Cette supposition peut &re confirmte par les recherches de M. Halm A Edinburg, qui a trouvt, que les raies ooires du spectre solaire se dtplacent vers le c6tt rouge du spectre pendant la ptriode entre un minimum et un maximum des taches. Mais ce phtnomene, comme l’explique M. Jewell, provient de l’augmentation de la pression, sous laquelle se trouve le gaz absorbant.

Lorsque, aux tpoques des maxima dcs taches, la pression de I’atmosphhre au-dessus des taches commence A s’augmenter, cette augmentation sera plus grande aux lati- tudes moyennes qu’aux latitudes plus proches de I’tquateur, par suite de l’influence de la force centrifuge, qui est plus grande P l’tquateur qu’aux latitudes moyennes. Alors, ap rb un maximum des taches, I’activitt solaire doit se dtplacer peu A peu vers l’tquateur, ce qu’on observe en rtalitt.

Les particules petites et ltghres, emporttes dans I’tspace Kichineff, le 5 Dtcembre 1908.

par la pression de radiation, comme il est dtjA dit, forment les rayons coronaux. Or, puisque ces rayons soot produits par l’activitt des protubtrances et que cette activitt aux tpoques des maxima des taches est concentrte aux latitudes moyennes du soleil et aux tpoques de leur minima se dt- place vers l’tquateur, ces rayons coronaux, eux aussi, doivent &re plus longs aux tpoques des maxima au-dessus des lati- tudes moyennes et aux tpoques des minima au-dessus de l’tquateur. Mais c’est le fait, constatt par les recherches de M. Hansky, qui comparait les formes de la couronne pendant les diverses eclipses du soleil, ayant eu lieu tant aux maxima des taches qu’h leur minima.

Les particules Itgkres, emporttes du soleil, peuvent former dans l’tspace des amas de matitre trts rarefite, comme la lumiere zodiacale, et la poussiere mtttorique.

Ces particules, par suite du frottement, auquel elles soot soumises, en s’tchappant P travers l’atmosphere solaire, doivent &re fortement tlectristes. En s’tloignant dans I’espace, ces particules emportent de grandes quantitts d’tlec- tricitd, qui pourrait influencer les champs magnttiques de la terre et des autres planetes et Ctre ainsi la cause des tem- pCtes magnttiques terrestres, des oscillations permanentes de l’aiguille aimantte et des aurores polaires.

Sans doute, elle ne peut pas prttendre P reprtsenter les phtnomenes solaires comme ils se passent en rtalitt. Cette rtalitt, nous ne la connai- trons jamais. ]’a; fait seulement une tentative de rtunir tous ces phtnomenes par une idte gtntrale pour leur donner une explication, du point de vue scientifique, admissible et logique.

Telle est mon hypothese.

A. Amaffoumky.

Distribution of the stars. (Harvard College Obse rva to ry Ci rcu lar No. 147).

A study is now being made of the results contained in the Revised Harvard Photometry, Harv. Ann. 50. The residuals from the various catalogues, compared in that work, are grouped according to their right ascension, declioatioo, magnitude, position with regard to the Milky Way, and class of spectrum. A similar study has been niade of the numbers of stars of different magnitude. It will be several weeks before these investigations are completed and printed. One result, however, seems of sufficient importance to bring at once to the attention of astronomers, since some of them are oow engaged in similar investigations, and the anoounce- ment at this time of the new law may save superfluous work, and enable them to modify their plans, if necessary.

If the stars are infinite in number, and distributed throughout space according to the laws of chance, we should ex- pect that, in any given class, the number, N, of stars, brighter than a given magnitude, M, would be represented by the formula l o g N = a M + b, in which a = 0.60, and b is a con- stant dependent on the number of bright stars in that class. The first accurate determination of a is contained in Harv. Ann. 14.483, where from a hundred thousand measures of four thousand bright stars, the value of a was found to be 0.52 instead of 0.60, with an uncertainty of less than & O . O I .

The small value of u is commonly supposed to be due to an absorbing medium in space, a view which is confirmed by the darkness of the sky. Otherwise, the brightness of the latter should approximately equal that of the surface of the Sun. In Harv. Ann. 48, No. V, a further study of the distribution of the stars was made, extending the work to the fainter stars. This confirmed the value of a for the bright stars, and showed that, for fainter stars, it diminished gradually, attaining the value, 0.35, for stars of the twelfth magnitude. In Harv. Ann. 56, No. I, a study was made of the proportion of stars having different spectra, in different parts of the sky, but without regard to brightness. Finally, Harv. Ann. 50, furnished so large an amount of material, more than a million observations of 91x0 stars, that the stars of each class of spectrum are sufficiently numerous to be grouped according to their brightness. They are ac- cordingly divided into the Classes B, A, F, G, K, and M, using the notation first employed by Mrs. Fleming in the Draper Catalogue, Harv. Ann. 27, and since used in all the work of this Observatory, except in Ham. Ann. 28, Part I. From a study of the number of stars of different magni- tudes, it appears that the spectra of Classes A and F, follow the same law. They have, therefore, been combined in the

1 49

first part of the following table. Spectra of Classes G, K, and M, are similarly combined, for the same reason. The Milky Way evidently introduces a disturbing element which affects the distribution. Accordingly, the regions north and south of the Milky Way, covering one half of the entire sky, have been combined. The first column of the table gives the limiting magnitudes of the groups. Thus, the first line relates to all stars brighter than the magnitude 4.25, The last line includes all stars in Harv. Ann. 50. The limiting magnitude is called 6.50, since all stars are included which have this magnitude, or brighter, in either of the photometric catalogues. Some fainter stars are thus included, but on the other hand, some slightly brighter stars have, doubtless, escaped observation in all the catalogues. Owing to this

4305

uncertainty, little weight should be given to the results de- rived from this line. The numbers of stars in the six groups into which the stars are divided are given in the second column, and the successive sums of these numbers are given in the third column. The logarithms of these numbers are given in the fourth column. The values computed by the formula 0.60M- 0.753, the residuals expressed in loga- rithms, and in magnitudes, are given in the next three co- lumns. The second part of the table gives the corresponding values for spectra of Classes G, K, and M, using the formula 0.60 M - 0.482. The residuals are here large and negative. The last three columns, give the computed values and the residuals using the formula 0.51 M - 0.093.

The average value of the residuals in the seventh column is fo.023 magnitudes. This quantity, small as it is, besides containing various accidental errors, is the sum of two systematic errors. These are, first, the deviation of the scale of the photometer from the truth, and secondly, the deviation of the law of distribution from that given by theory, assuming that there is no absorption. Of course, it is possible that these errors are equal and opposite, and thus neutralize each other, but this is so improbable that it appears more likely that both are insensible. In other words, that the scale of the meridian photometer is the true photo- metric scale within less than one part in a hundred; and that the stars of Classes A and F follow the theoretical law with the same degree of precision. This view is strengthened by the values given in the second part of the table. The residuals from theory differ nearly half a magnitude, the progression being nearly uniform. The residuals in the last column have an average deviation of f 0 .020 magnitudes, even including the last one, which should be rejected. Without it, they are insensible, and indicate that the law for stars of Classes G, K, and M is very nearly 0.51 M + b. It is difficult to understand how any absorption could affect un- equally the number of stars whose spectra are unlike. A colored medium would affect the distribution of light in the spectra of stars at different distances, and thus would furnish an admirable means of determining that most important ele- ment. Rays of the same wavelength would be affected equally in all stars whatever their spectrum. The effect would be that one or more dark bands would appear whose density

would increase with the distance, the change in the loga- rithm of the light of any portion of the spectrum, being proportional to the distance. It would not have any effect on the relative number of stars having different spectra if, as in the Draper Catalogue, the measures of brightness were all made of rays of the same wavelength. It is obvious that studies of the distribution of stars, will now have little value unless the stars are classified according to their spectra. Fortunately, the Henry Draper Memorial has not only fur- nished the means of doing this in the Revised Harvard Photometry, but by the plates enumerated in Harv. Ann. 56.8, it enables the work to be extended to much fainter stars. It furnishes also another example of the importance of con- sidering the class of spectrum when studying any stellar problem, like that of the motion of the Sun in space.

It will be noticed that stars of Class B are not in- cluded in the table. Contrary to expectation, these stars show no increase in number in the successive groups. The numbers of stars in the six groups are 135, 103, 163, 170, 146 and 105, respectively. This may be partly due’ to the difficulty, in faint stars, in distinguishing the helium lines characteristic of this class.

In conclusion, bright stars of the first type, Classes A and F, appear to follow closely the theoretical law, a = 0.60, as if there were no absorption in space. The number of stars of the second and third type, Classes G, I(, and M, are well represented by the coefficient a = 0.5 I. The Orion stars, Class B, show no increase in number.

Harvard College Observatory, Cambridge, Mass., 1909 Jan. 14. Edward C. Picketing.

(521) Brixia. Correzione all’effemeride (A. N. 4300): 1909 Gem. 28 +0?75 -315. 23. Bianihi.