PROPERTIES OF NEARBY STARS
So far we have used
ASTROMETRY to obtain the distances to stars out to (via HIPPARCOS) hundreds of parsec (this number was, until recently, several tens of parsce),
PHOTOMETRY to obtain the brightness of stars which, with distances, gives us luminosities,
BINARY STAR analysis (especially eclipsing) to obtain stellar masses and radii,
INTERFEROMETRY to obtain stellar radii (and turn more spectroscopic binaries into spectroscopic/visual binaries),
SPECTROSCOPY and spectral analysis to obtain effective temperature, surface gravity and composition.
Now we put the data (luminosity, mass, radius, temperature, surface gravity, composition) together to see what stars are like.
One of the first analysis involved plotting luminosity vs. spectral class, resulting in the Hertzsprung-Russell diagram:
The H-R diagram is a plot of absolute magnitude vs. spectral class, but often the ordinate is replaced by log(L/Lo) (Lo=luminosity of the sun) and often the abscissa is replaced by color (B-V) or effective temperature (Te). If you plot color vs. magnitude you should really call such a diagram a "clor-magnitude" diagram. And if the abscissa is Te note that the spectral class of a giant star with the same Te as a dwarf will appear "earlier" than the dwarf, ie, a G0I has the same Te as a G5V, about 5400K. So be a bit wary of "H-R diagrams" that show spectral class and temperature on the same abscissa.
Now what does the H-R diagram tell us? First we note that most stars with good parallax measurements, ie nearby stars, lie mostly in the lower main sequence with a fair sprinkling of white dwarfs (which are actually embers of stars but we include them anyway). Very few early stars or supergiants. But most of the stars that are bright are either early stars or giants!
The key to the strange feature can be found in another representation of our basic stellar data, the mass-luminosity diagram. If we plot log(L/Lo) vs. log(M/Mo) we find another "main sequence" (but don't let anyone hear you call it that)