D O C . 1 4 M A R C H 1 9 1 9 1 3 the nebulae can be analyzed spectrally, cannot as yet be established, except in Orion. In the Pleiades the nebulous matter is so faint that its spectrum becomes vi- sible only after an exposure time of hours but it is not measurable. However, the following remarkable phenomenon is present in the fixed-star sys- tem. In the spectra of the B stars (they are those white, massive stars that also ag- gregate in Orion), 2 fine spectrum lines often appear where the H and K lines of calcium should occur. While these H & K lines are unusually broad and intense in stars of the type of the Sun, there they are very weak and sharp. What is striking about them, though, is the fact that in spectroscopic double stars they do not exhibit the periodic fluctuations of the lines or else not to the full amplitude value.[2] Hart- mann, who first noticed this fact with δ Orionis, believed that these lines could be generated by suspended calcium clouds in the universe.[3] In the meantime the same observation has been made in a variety of systems and the following hypoth- esis does not seem unlikely to me. The extraordinarily large B stars are frequently enveloped in a very faint but broad calcium atmosphere (hence the thin lines) in which both components are revolving around each other. Speaking for this assump- tion is also the fact that calcium plays the leading role in the outermost regions of the solar atmosphere. It seems to me very probable that they belong to the stars also because they yield the centroidal velocities approximately correctly. I have now derived the following quantities for those cases where I could get hold of adequately precise observational data: (1) Centroidal velocity of the system as mean value of the lines exhibiting the normal Doppler effect (2) radial velocity from the calcium lines. For the difference: (1) – (2) I find the values shown in the table: Except for the 2nd star, all show a marked red- shift, as it should be. Yet the values themselves are still uncertain, fre- quently by many kilometers, so that the disagreement of the star β Scorpii is perhaps only coincidental. These 6 stars lie in completely different locations in the sky. δ Orionis & ε Orio- nis belong within the Orion nebular cluster it is thus interesting to note whether the difference arising * Spect. (1) (2) 1 – 2 σ Aquilae B8 –5 km –10 km +5 km β Scorpii B1 –11 –9 –2 " o Persei B1 +18 +12 +6 δ Orionis B0 +19 +17 +2 VV Orionis B2 +21 +17 +4 ε Orionis B0 +24 16 +9 +6 km