76 KNUT LUNDMARK, GLOBULAR CLUSTBRS AND SPIRAL NEBULJE. 



they must also be regarded as illusory. Nevertheless the preceding calculations of 

 mean parallaxes of spiral nebula?, derived from their proper motions and radial 

 velocities, will have their value as fixing an upper limit of the parallaxes. 



Since the above paper was written, the discovery of four new stars in the An- 

 dromeda nebula has been communicated. If the M max . of the 15 nova3 known as yet 

 in this nebula is used for a determination of its parallax, we find a value only 

 slightly differring from that previously obtained. There seem to exist several reasons 

 for assuming that the novae also regarding the M m m. belong to both the giant and 

 dwarf types. In the Andromeda nebula only Hartwig's Nova would be of the giant 

 type and the 14 others dwarfs, whereas in the other nebula? it seems plausible 

 that it is only giant novse that we have been able to observe. With this assump- 

 tion, we obtain the parallax-values given under the head »k from novae» in the 

 table below. 



At determining the apparent står density on plates of the Andromeda nebula, 

 a greater average density seems to occur within the nebula than outside, provided 

 that sufficiently faint stars are included in the counts. This peculiar instance 

 probably ought not to be interpreted in any other way than as indicating an inci- 

 pient dissolution of the Andromeda system into separate stars. As all stars belong- 

 ing to this system are at the same distance from us, it is evident that we can see 

 only those stars dissolved that have the greatest luminosity. If we assume that the 

 absolute magnitude attains the same maximum in the Andromeda system and in 

 the other spirals as in our stellar system, we have herein, if this maximum can be 

 determined, a new means of finding the parallax of those spirals for which we have 

 been able to observe an incipient dissolution into separate stars. 



The direct parallax-measurements of stars indicate that the absolute magnitude 

 can attain an upper limit of — 7 m . Departing from the parallax determined for the 

 small Magellanic cloud from 3 Cephei-variables, cluster-diameters and Kapteyn's 

 integral-equation (3), and calculating the absolute magnitude of its most brilliant 

 stars, we find that they attain a maximal magnitude of —6'". The data given by 

 Russell 171 regarding the frequency of absolute står magnitudes also give an upper 

 limit aggreeing with this value. 



This quantity evidently can not at present be determined with any greater 

 accuracy, but we will provisionally assume the value — 6™. 



On plates taken with the 6-inch twin telescope of this Observatory with ex- 

 posures between 6 sec. and 9 Z U hours, står counts have been made, showing at a 

 provisional reduction that the dissolution of the Andromeda nebula into separate 

 stars is beginning at the star-magnitude 14. We thus obtain: 



*Andr. = 0",000010, 



a parallax only two times as great as that found from the novse. 



Professor von Zeipel kindly has put at my disposal a plate of Messier 33 

 taken on Mount Wilson at the 60-inch telescope with an exposure of 8,5 hours. On 



