Radick: Well, as far as I'm concerned, that settles it [laughter], but maybe there a few dissenters...?

Garrison: With the kind of accuracy you're dealing with now with HIPPARCOS, it's a pity you're still using an extended source to compare with stellar sources. The Moon is not a good source; I wish you could use something else.

Cayrel: It is not a pity, because we have checked during several observing runs, at CFHT, ESO, and OHP, that the equivalent widths obtained from the Moon and from the asteroids Vesta and Ceres were undiscernable. As large-telescope time is very precious, and the time to get the desired S/N is at least a couple of hours, we are using moonlight.

de Mello: I have examined spectra of 18 Sco with those of Ganymede, at high signal-to-noise. All the spectra compare very well with one another. I have also examined the Moon and the sky, just a blue sky spectrum, at 30,000 resolution and again at very high signal-to-noise: they were the same, as far as I could tell.

Garrison: That depends on the spectrograph.

de Mello: Surely. It was a single-order spectrograph, well-behaved.

Cayrel: Because of this slight dependence of equivalent widths versus spectrograph (much reduced with solid state detectors in comparison with the old photographic spectra), we have always worked differentially, using the same spectrograph for program stars and comparison sunlight spectra.

Guinan: I just noticed that since 1980, the duration that you have a particular star ``lover'' decreases with time [laughter]. It's now just a year, or less.

Cayrel: You are certainly right. But this time, the time interval between the last but one (51 Peg) and the last best analog (18 Sco) was so short because of the Hipparcos event, which is responsible for this quick change of "lover." We are not done, and you [pointing at de Mello] are not now to think that you are sitting on yours [protracted laughter]!

Guinan: He has about six months [laughter].