from VSSC103
The accurate photometry undertaken during the ESA Hipparcos Mission has been responsible for detecting 8,237 new variable stars. A high proportion of these new variables are red stars, and in particular red giants, as illustrated in the diagram on page 465 of volume 1 of the Hipparcos and Tycho Catalogues. This diagram shows that 70% of giant stars with V-I mags greater than 1.5 are variable stars.
Recently, I did a check on the stars that are between 0 hours and 12 hours in the Sky Catalogue 2000 and that are also listed as spectral class M0 or later, and I compared them with the Hipparcos data. Not surprisingly, out of 759 stars sampled, only 48 (mainly spectral class M0) were not listed as variable. Hipparcos made on average 110 photometric observations per star within the period November 1989 and March 1993, and we know from long-term monitoring of red variable stars that variations can be intermittent, so the risk remains that some of these 48 stars may still be variable outside of the 3 year period that Hipparcos was monitoring them. The above would appear to support what Stebbins and Huffer suspected as long ago as 1930 - that no red giant star is of really constant brightness.
Consequently, it would seem that unless it is a main sequence dwarf, red stars of spectral class M0 or later are unsafe to use as comparison stars. Since the vast majority of bright red stars are giants and not dwarfs, and considering that visual observers have different spectral responses (particularly with red stars), it would be sensible to avoid using stars of spectral class M0 or later as comparison stars in the future.
As accurate spectral data on stars is not always readily available, it is proposed that a limit is imposed, instead, on the basis of the star's B-V index. Upon review of Table 45 on page 142 of Nortons Star Atlas (18th Edition), a sensible limit would appear to be a B-V index of 1.5, which equates roughly to spectral class M2 for main sequence stars, K8 for giants and K2 for supergiants. So with immediate effect, no stars with a B-V index of greater than 1.5 will be selected as comparison stars on future sequences. The few that appear on existing sequences will be reviewed at the time that the chart is updated. Only where the sequence seems fine and we have a long series of data will existing comparison stars of B-V >1.5 be retained.