Finally, the individual observations themselves are reduced, using the parameters from the extinction solution. This is a little tricky, as one needs color values to include the color terms in both extinction and transformation equations. If all colors are observed simultaneously, as is done with multichannel spectrometers, these colors can easily be extracted from the neighboring data. But a filter photometer or CCD must observe passbands sequentially. Again, this may not be a problem, if the star is constant in light.
However, for variable stars, the brightness is changing in all bands, and it is difficult to define the correct color to use in the reduction. In principle, one should construct the full light curve of the star, and then interpolate the colors to the time at which each passband was observed. In practice, a simpler compromise is used: the colors are simply interpolated linearly in time, if there are earlier and later observations of the star on the same night. For the first and last observations of a star on a given night, the nearest observations (in time) in the adjoining bands are used. In every case, the values of the colors used to reduce each magnitude are shown on the output.
Every observation is given individually, including observations of standard and extinction stars, together with the U.T. day fraction and heliocentric Julian Date. These data are of interest for standard and extinction stars if they later turn out to be variable. Note that the values of stellar parameters adopted in the extinction solution are to be preferred to values obtained by averaging these individual observations that have been corrected for extinction and transformation.
NOTE: Because colors are needed to transform the observations, first to outside the atmosphere and then to a standard system, only stars that were observed in all passbands on a given night can be reduced in this section of the program. This is in contrast to the treatment of constant stars in the general solution, where even single-passband measurements are useful, provided that (a) every star was observed at least once during the run in every passband; and (b) every night has some observations in all passbands.
That is, the extinction solution can in principle produce results if we observe extinction star 1 only in passband A and star 2 in passband B on night 1, and star 1 only in passband B and star 2 in passband A on night 2. This fulfills the requirements for data in both bands for each night, and for each star. Even if the passbands are linked by color terms in the extinction, the solution is possible, because we do obtain mean colors for each star. However, none of these data could be reduced individually, as there are no observations of either star in both bands on the same night.