1.0Probability 0.0 150 0.6 0.4 200 Figure 3. Distribution of stars in parallax in the NGC 7789 field. Stars nearest the proper motion locus (navy) and additional stars in a wider annulus (purple) form an obvious overdensity amongst the field stars (green). A two-Gaussian model is shown (red), with the broader component representing field stars. Figure 4. The resultant probability distribution after separate probabilities are combined for NGC 7789. The first bin contains about 20000 nonmember stars. 4. DISCUSSION AND CONCLUSION from Kharchenko et al. (2013) was converted from E(B-V) to E(BP RP ) using Casagrande & VandenBerg (2018). Results are summarized in Table 1 and shown in Fig. 6 with the H91 fit, transformed to logarithmic age, overlaid.

From Fig. 6 we find that the H91 line extrapolates poorly at the young end, and that many clusters have a larger clumpRGB color separation than the relation would predict. At log(age) = 9.25, or age = 1.8 Gyr, the samples overlap in age, but every cluster in the new sample shows a wider dB 3 R 4 CHRISS AND WORTHEY 2 2.5 0.8 8G(mag) 0 0.9Prob 0.0 3.0BPRP(mag) 0.7 1.0 4 4 0.5 2 6 1.5Gyr(BaSTI2004[Fe/H]=+0.06) 1.5849Gyr(Padova2008[Fe/H]=0.00) 2.0 1.5 Figure 5. Color-absolute magnitude diagram in Gaia passbands for probable members (P > 0.7) of NGC 7789 using the distance and reddening from Table 1. Symbols are color mapped by membership probability as indicated by the color bar. Example isochrones from Padova (blue, with carbon-rich stars excluded) and BaSTI (green) are shown. Table 2. Integrated light systematic color error if theoretical Z dependence is allowed to flourish.

Quantity Teff U V , MP, H91 U V , MR, H91 U V , MR, +theory Excess U V V K, MP, H91 V K, MR, H91 V K, MR, +theory Excess V K log age = 9 172 K 0.75 1.04 1.04 0.004 2.39 2.70 2.74 0.038 log age = 9.5 93 K 1.11 1.58 1.58 0.002 2.73 3.33 3.35 0.021 log age = 10 212 K 1.43 1.99 1.99 0.003 3.04 3.81 3.85 0.037 color separation. The magnitude of the effect exceeds any possible random or systematic uncertainties.

Even restricted to the new measurements, and further restricted to log age < 9.5, the scatter appears to be astrophysRP values for ical. A Kolmogorov-Smirnov test on our dBP clusters in our sample younger than 9.5 log years returned 5, indicating that the scatter is not drawn p from a normal distribution. The cause of the scatter cannot be age, or we would see a slope in Fig. 6. We sample metallicity span less well, but within that caveat, no metallicity dependence is evident. If neither age nor metallicity is the cause of the modulation in dB R, we are driven to consider more subtle causes such as variation in helium abundance, variation in the [/Fe] ratio, or variation in binary separation distributions. The investigation into the cause of the scatter must await future work. 1.62 10