3.1. The colour-magnitude diagram
Fig. 1 shows the CMD of M55. From the original data set of 17269 stars, all objects within an inner radius of 320 have been taken. We selected according to a maximum error in V of mag for the stars brighter than . The narrow giant branch and the well defined blue horizontal branch (BHB) show the high internal precision of the selected subsample for the bright stars in M55. Below the photometric errors increase quickly and produce the large scatter near the turn-off point.
A number of blue-straggler stars populate the CMD in the region between the horizontal branch and the turn-off point. These objects will be discussed in the appendix.
3.2. The diagram
In the diagram of M55 (Fig. 2) stars with and have been plotted. The diagram shows a well defined sequence of the giant-branch stars and there is no evidence for a colour spread larger than the photometric errors. The diagram is reddening corrected with . We obtain this value from our photometric data, as we show in the following section.
Through the correlation between [Fe/H], and the diagram can be used to derive metallicities for individual giant-branch stars in globular clusters (Grebel & Richtler 1992). A new calibration of the diagram for the metallicity of red giants is given by Hilker (priv. comm.), in Fig. 2 indicated for the metallicities of 0.0,-1.0 and -2.0 dex. It is based on a sample of 60 stars with known Strömgren colours and spectroscopically determined [Fe/H]-values in the range -2.4 dex [Fe/H] -0.6 dex. 36 stars are taken from the work of Anthony-Twarog & Twarog (1998), 17 from a sample in Centauri and 7 stars of M22 from this work. The metallicity then is given by:
For stars more metal rich than [Fe/H] = -0.6 dex the new calibration has been adjusted to the older one of Grebel & Richtler (1992). Values for [Fe/H] derived by this method are sensitive to the adopted reddening. However, the CMD of a globular cluster also contains information about its reddening and metallicity. Gratton & Ortolani (1989) find the following correlation between [Fe/H] and the colour of the giant branch in globular clusters at the level of the horizontal branch, :
For these two equations reddening and metallicity are correlated in different ways, so that with the combination of both an unambiguous result for [Fe/H] and can be obtained.
For M55 we find , equivalent to (Grebel & Roberts, priv. comm.). Following Hilker's calibration of the diagram for stars with we find a reddening of and a mean Strömgren metallicity of [Fe/H] dex for M55. This result is, within the error range, in good agreement with the value of Zinn & West (1984; dex).
3.3. CN abundances
Richtler (1988) found (on the basis of aperture photometry) that the giant branch in the M55 diagram splits up in two sub-branches where the corresponding metallicities anticorrelate with the CN band strengths given by Smith & Norris (1982). Since in other clusters an anticorrelation between C and N is well established (e.g. Suntzeff 1993), the suspicion was that a continous absorption of CO, which affects the Strömgren v-filter, could be responsible for this effect.
For comparison, we took 27 stars from the sample of Briley et al. (1993) with known CN abundances (Table 4), identified by the identification numbers from Lee (1976). This sample also includes results from the measurements of Smith & Norris (1982). The range of cyanogen strengths is given by (Briley et al. 1993), where is the spectroscopic index for the CN band beginning at 3883 Å (see Norris & Freeman 1982 for a detailed description). In our diagram of M55 (Fig. 2) the scatter in is small and therefore no signs for significant CN variations are visible. The mean Strömgren metallicity of the selected 27 stars is -1.72 dex, derived with the calibration of Hilker. We define to be the vertical distance from the position of an individual star in the diagram to the calibration line of -1.72 dex. In Fig. 3, has been plotted versus the cyanogen index . No correlation between the values and the cyanogen strengths can be seen. The variations of the values are of similar size as the photometric errors of the index, so that the -CN anticorrelation found by Richtler (1988) might be below our detection limit. However, our measurements give no indication for a significant abundance spread in CN in the cluster giants of M55.
Table 4. CN strengths and Strömgren colours of red giants in M55. The values of S(3839) have been taken from Briley et al. (1993)
The comparison between the photometric study of Richtler (1988) and the present data reveals only slight differences in the measured Strömgren colours. From a comparison of 7 stars we find that our values of V, and show mean deviations of , and (in the sense of colour ). It is most likely that these shifts reflect the usage of different standard stars (see Richtler 1988). Anyway, the differences are of no importance for the qualitative comparison described above.
© European Southern Observatory (ESO) 1999
Online publication: October 4, 1999