We have taken account of the the effect of metallicity on temperature
determination for all of the stars in the present study for which direct
measurements of [Fe/H] exist. Unfortunately, measures are available for
only 35 per cent of the total number of stars in the programme, and this
is reflected to some extent in the spread of the relations between
temperature and photometric indices such as 
.
Figure 4 (click here) plots derived temperatures
against the arbitarily selected parameter 
, using class V
stars of the higher class of quality, with known values of [Fe/H] in the
narrow range 
.
In these circumstances, the spread
about the regression line is very small, although it must be admitted
that there are only a few points.
Figure 4: Plot of temperatures of Class V stars against .
The stars have known metallicities in the range -0.59<[Fe/H]<-0.34
In order to derive the effect of metallicity on the representation
of determined temperature by functions of photometric indices, we have
evaluated the residuals in a regression as a function of measured
metallicity [Fe/H]. For stars of luminosity classes V and IV,
the residuals depend on metallicity according to the function
per cent. Accordingly, where metallicity values
exist, the polynomial relations derived in the previous sections should be
increased by this function. The full expression for temperature
determination using a photometric quantity x, where x can be
, or 
,
and metallicity [Fe/H] is now:
