5 Color-magnitude diagrams

5.1 Sequences in the V, B-V diagram

The $(V,\ B-V)$ diagram for NGC 6723 is presented in Fig.6. All its sequences are well populated. Plus signs (+) in Fig.6  and in other diagrams designate five RR Lyrae variables identified in our frames. Magnitudes of the variable stars in different bands were measured non-simultaneously, so the plus signs in the diagram have arbitrary positions. The signs of two other types, the cross ($\times$) and the asterisk, will be discussed below, when presenting Fig.8. Table2  summarizes UBVRI ridge points of the principal sequences.

  

Figure 6: The $(V,\, B-V)$ diagram for NGC 6723. Plus signs, RR Lyrae variables, with their magnitudes measured at arbitrary phases; cross and asterisk, see explanation for Fig. 8

  

Table 2: Ridge lines of sequences in NGC 6723

The cluster's horizontal branch in the $V,\ B-V$ diagram, in the color-index range 0.20<(B-V)<0.75, actually runs almost exactly horizontally. The stars closest to the instability region and to the blue of it have, on average, V=15.46, slightly brighter than the stars immediately to the red of the instability region which have, on average, V=15.50. So we can accept $V_{\rm HB}=15.48$ as the magnitude of the horizontal branch at the level of the instability strip. This value coincides exactly with that given by Menzies (1974) and does not differ significantly from the value 15.4 published by Martins & Fraquelli (1987).

  

Figure 7: The (B-V) color distribution of stars along the HB of NGC 6723

Figure 7  illustrates the distribution of stars along the horizontal branch. This histogram shows that, at the HB level, the blue edge of the instability strip is at $(B-V)_b=0.26\pm{}0.01$ and the red edge is at $(B-V)_r=0.50\pm{}0.01$. Menzies (1974) gives systematically bluer positions for the boundaries of the instability strip, namely (B-V)_b=0.17 and (B-V)_r=0.44, in agreement with the comparison of the photometries presented in Sect. 3.

To derive magnitudes of the stars in NGC 6723 used in Martins & Fraquelli (1987), "the photographic plates were first digitized using microdensitometer with a scanning aperture $40\ \mu$m square'' and then a PSF was derived and applied using the necessary software. However, Martins & Fraquelli do not directly present their (B-V)_b and (B-V)_r values. From the distribution of stars along the horizontal branch of NGC 6723 published in their paper (Fig.7), it follows that $(B-V)_{b}\approx0.30$ and $(B-V)_{r}\approx0.50$. However, these absolute values should be considered with caution. Figure 3b in Martins & Fraquelli (1987) shows that all their color indices are redder than in Menzies (1974), the difference at the blue edge of the instability strip being approximately $0\hbox{$.\!\!^{\rm m}$}15$ and at its red edge, $0\hbox{$.\!\!^{\rm m}$}10$. Martins & Fraquelli emphasize that their magnitudes of the photoelectric standards show good agreement with Menzies whereas the photographic magnitudes deviate considerably. We think that the discrepancy is the result of an insufficient account of systematic deviations between aperture and PSF photometry. Remarkably, within errors in our study and in studies of Menzies and of Martins & Fraquelli, the measured widths of the instability strip are in good agreement.

The conspicuous maximum at $(B-V)_1=0.69\pm{}0.005$ apparent in Fig.7 is noticeable as a clump of stars in the red horizontal branch in the diagrams published by Menzies and by Martins & Fraquelli.

It has long been known (Mironov & Samus 1974) that the dereddened color index of the most densely populated point on the horizontal branch, (B-V)_0,1, shows a good correlation with the difference of apparent (not dereddened) color indices of the same maximum and the blue edge of the instability strip, $\Delta_{b-1}=(B-V)_1-(B-V)_b$.This relation can be expressed as

$$\begin{eqnarray} \nonumber (B-V)_{0,1} = 0.154\pm{}0.008 - (1.006\pm{}0.034)\Delta_{b-1}, \nonumber\end{eqnarray}$$

and can be readily used for estimates of globular-cluster E(B-V) color excesses, since

$$\begin{eqnarray} \nonumber E(B-V)&=&(B-V)_1-(B-V)_{0,1} \\ \nonumber &=& (B-V)_1 - 0.154 + 1.006{}\Delta_{b-1}. \nonumber\end{eqnarray}$$

From our data for NGC 6723, $(B-V)_1=0.69\pm{}0.005$, and $\Delta_{b-1}=-0.43\pm{}0.012$. Thus, we obtain $E(B-V)=0.103{}\pm0.0=15$,in excellent agreement with the values found in Sects. 4.1 and 4.2. In subsequent analysis, we adopt $E(B-V) =0.11\pm{}0.01$.

The RGB color index at the level of the horizontal branch is $(B-V)_{\rm g}=0.97$; thus, for the adopted reddening, $(B-V)_{0,\rm g}=0.86$, and the standard color index (B-V)_1.4 becomes (B-V)=1.51. The RGB magnitude at the standard color index is V_1.4=13.15, and the elevation of the giant branch above the horizontal branch is $\Delta{}V_{1.4}=2.30$.

There exist a number of calibrations of metallicity [Fe/H] from $(B-V)_{0,\rm g}$and $\Delta{}V_{1.4}$. Recently Ferraro et al. (1996) presented a compilation of the relevant formulae. Table3  contains the values of metallicity for NGC 6723, estimated using these formulae. Almost all these values are considerably lower than the cluster's metallicity adopted earlier. The mean of these determinations is $\rm [Fe/H]=-1.22$, rather close to the value $\rm [Fe/H]=-1.26\pm{}0.09$, recently found by Fullton & Carney (1996) from high-dispersion spectra of three red giants.

  

Table 3: Metallicity ([Fe/H]) of NGC 6723 from RGB parameters

According to a recent detailed study (Sandquist et al. 1996), the RR Lyraes in the globular cluster M 5 are at the magnitude $V_{\rm HB,M\,5}=15.11$, and the apparent distance modulus of M 5 is $(m-M)_{V,{\rm M\,5}}=14.41$. Similarly, according to Buonanno et al. (1994) and Ferraro et al. (1996), for M 3 $V_{\rm HB,M\,3}=15.67$, and $(m-M)_{V,{\rm M\,3}}=14.94$. This leads to the absolute magnitude for RR Lyrae variables, M_V(RR)=0.70 or M_V(RR)=0.73, respectively. Different metallicities of the clusters do not significantly influence this absolute magnitude. Following Harris (1996), we adopt the slope $\partial M_V(RR)/\partial{\rm [Fe/H]}\approx0.2$. This means that, for a metallicity difference $\rm \Delta [Fe/H]=0.20$, the error in the distance modulus would be only $0\hbox{$.\!\!^{\rm m}$}04$. Adopting M_V(RR)=0.72, we obtain the apparent distance modulus (m-M)_V,6723=14.76 for NGC 6723. This value does not differ significantly from that quoted by Harris (1996), namely (m-M)_V,6723=14.82.

The main-sequence turnoff is at $V_{\rm TO}=18.95\pm{}0.05, (B-V)_{\rm TO}=0.585\pm{}0.005$. The turnoff and the horizontal branch are separated by $\Delta{}V_{\rm TO}^{\rm HB}=3.50$;the same value can be obtained from CMDs using different color indices along the abscissa, the corresponding turnoff colors can be easily retrieved from Table2.

  

Figure 8: The $(U, \, U-B)$ diagram for NGC 6723. Open circles, blue horizontal branch stars; open triangles, red horizontal branch stars; asterisk, a UV-bright star; plus signs (+), known RR Lyraes; cross ($\times$), new suspected variable, see text

5.2 The U, U-B diagram

Figure 8 shows the $(U,\ U-B)$ diagram for NGC 6723 and illustrates clearly that the U band is especially helpful for separating the stars of bright sequences that are practically overlapping in other diagrams. Note that the asymptotic giant branch is now well separated from the RGB and that the separation of red horizontal branch stars from the RGB has become clear.

The UV-bright star above the horizontal branch (asterisk) is of special interest; it was not clearly isolated in the $(V,\ B-V)$ diagram. Actually, the star is the standard H (see Table1); its equatorial coordinates are $18^{\rm h}59^{\rm m}43\hbox{$.\!\!^{\rm s}$}8$, $-36^\circ36'08''$ (2000.0). Its cluster membership needs confirmation.

Of special interest is the relative position of blue (open circles) and red (open triangles) horizontal branch stars. They overlap in (U-B) color and show a distinctly different U-magnitude level near the instability strip.

A star plotted as a cross ($\times$) sign in Figs.6, 8  and 9 appears a red horizontal branch star in Figs.6 and 9 but as a blue horizontal branch star in Fig.8. It might be an unknown RR Lyrae variable star, its behavior resembles that shown by known variables (plus signs). Its position, in the system of the catalog of variable stars in globular clusters (Sawyer Hogg 1973), is x=263'', y=17''.

  

Figure 9: The $(V, \, V-I)$ diagram for NGC 6723. Same symbols as in Fig. 6

5.3 The V, V-I diagram. Determination of E_V-I and [Fe/H] by Sarajedini's method

In Fig. 9, we show the $(V,\ V-I)$ diagram for NGC 6723. From this diagram, we can determine the parameters $(V-I)_{0,\rm g}$ and $\Delta{}V_{1.2}$,to be applied in the method suggested by Sarajedini (1994) for simultaneous determination of metallicity [Fe/H] and reddening E(V-I).

From our diagram, we find $(V-I)_{\rm g}=1.08\pm{}0.01$.Sarajedini's method gives $E_{V-I}=0.12\pm{}0.01$, corresponding to $E_{B-V}=0.09\pm{}0.01$, and $\rm [Fe/H] =-1.77\pm{}0.10$.This reddening agrees, within errors, with other determinations, but the [Fe/H] value is significantly lower than all previous determinations, including those reported in Sect. 5.1. Such a low metallicity value must be incorrect owing to the method's uncertainties. However, this result continues the trend of the recent studies to lower metallicity of NGC 6723. Indeed, the [Fe/H] value for the cluster must be lower than it was previously believed: as noted in Sect. 4.2, the RGBs of M 3 and NGC 6723 exhibit very similar shapes in two-color diagrams as well as in the $(V,\ B-V)$ diagram. We note that the metallicity for M 3 is -1.40 or lower.