3 Analysis and discussion of the data

3.1 Cluster membership and interstellar reddening

The observed colour-magnitude (CM) and colour-colour (CC) diagrams of NGC2354 with the stars listed in Table 2 are shown in Figs. 1 and 2. The selection process for cluster membership is here primarily based on the radial velocity data.

  

Figure 1: The colour-magnitude diagram for red giant stars in NGC2354. Confirmed or possible cluster members and red field stars are represented by filled and open circles, respectively. Spectroscopic binaries (underlined) are also indicated. The distance between the apparent ascending giant branch and the clump is much too large when compared to theoretical models

  

Figure 2: The UBV two-colour diagram for the same stars as in Fig. 1, with the same notation. Triangles represent photographic data from Dürbeck (1960). The continuous curve is the standard two-colour relationship for typical G and K giants as given by Fitzgerald (1970). Star #200 has a peculiar U-B colour

Fourteen stars with mean radial velocities larger than 40 km s^-1 or lower than 20 km s^-1 are undoubtedly non-members, four of them most probably spectroscopic binaries (see Table 3). The radial velocities of six non-SB obvious members from Table 4 (stars #66, 91, 125, 152, 183 and 205) fall within an interval of only 1.7 km s^-1. The mean radial velocity of these stars is 33.40 $\pm$ 0.27 km s^-1 (s.e. of the mean) and has been adopted for NGC2354.

When the mean radial velocities of some stars differ by some 2 - 2.5 km s^-1 from the cluster mean velocity, i.e. differences well larger than 3 $\sigma$, it is more difficult to derive the membership from the radial velocity only. The examination of the colour-magnitude diagram may provide further help to the decision. Generally, there is a very good agreement between the kinematic and photometric membership estimates, but in the case of NGC 2354, there appear to be some contradictions. Although it is formally possible to compute membership probabilities, practically the results do not bring much insight in the membership determination. The six stars listed above would have high membership probabilities, and all other would be close to 0. In fact it does not properly take into account the case of the binaries, because the rough mean velocities are not fully representative of the true mean values. Further observations, and not only radial velocities, but also proper motions, are needed to settle the point.

Stars #59 and 219 differ by 2.1 and 2.2 km s^-1 from this mean velocity so that they are possible cluster members. They could be long period, low amplitude binaries with an eccentric orbit. The position of both stars in the colour-magnitude diagram (Fig. 1) may also indicate that they are non-member. However, star #59 falls very close to star #152, which has a radial velocity (33.84 km s^-1) very close to the cluster mean. Star #219 has a high $V\sin i$, like #179 which is a binary. The status of star #184, a possible spectroscopic binary, is difficult to decide because it has a radial velocity close to the cluster mean but its position in Figs. 1 and 2 falls also too red, if one uses available isochrones to analyse the distribution of the red giants in the colour-magnitude diagram of NGC 2354. Finally, #200, the brightest red giant contributes to the definition of a plausible ascending giant branch.

The mean velocities of the confirmed spectroscopic binaries #113, 179 and 269 are based on observations not well distributed with respect to the mean cluster velocity as a result of the telescope time allocation. Accordingly, they are not yet representative of the real mean velocities of these stars. Although these mean values differ by about 5.3, 4.7 and 5.8 km s^-1 from the cluster mean, these stars are considered as probable members, because the observed individual radial velocities for each star do scatter around the cluster mean. A definitive statement about their membership will await more observations and the determination of the orbits.

With the exception of stars #59, 152, 184 and 200, all the remaining members form an elongated clump of stars near V $\simeq$ 11.5 in the CM diagram. The position of stars #179 and 219 in the CM diagram is due to their binary character, certain for star #179 and probable for 219. The morphology of the CM diagram will be discussed later.

Cluster membership was also examined by applying the photometric criteria A and B described by Clariá & Lapasset (1983). Taking into account the different combinations that might result from the application of both criteria, we decided to consider a star to have a high probability of being a cluster member if one (or both) of the criteria implies membership, while the other indicates that the star is a probable member. If one criterion (or both) suggests non-membership, the star is rejected as a cluster member. Finally, if both criteria simultaneously indicate probable membership, the star is then considered to be a probable member of the cluster.

To apply criteria A and B, the colour excess E(B-V)$_{\rm MS}$ = 0.15 mag and true distance modulus (m-M)₀ = 10.80 both derived by Ahumada & Lapasset (1996) were adopted. The DDO colours were dereddened according to the reddening coefficients of McClure (1973) and the predicted luminosity class for each observed star was determined from the Schmidt-Kaler (1982) calibration assuming R = A_v/E(B-V) = 3.0.

Columns (6)-(9) of Table 4 contain the results from applying the photometric and kinematic criteria and the membership status finally adopted for each star (sb = spectroscopic binary, m = member, nm = non-member). Column (2) of Table 4 lists the E(B-V)$_{\rm GK}$ colour excesses derived from Janes's (1977) iterative procedure, which is abundance independent and valid over a wide range of luminosities for Population I stars. The standard deviation $\sigma$$_{\rm E}$, calculated from Clariá's (1985) Eq. (10) is given in Col. (3), while Cols. (4) and (5) include the predicted luminosity class (LC) and the MK spectral type inferred from the dereddened DDO colours.

  

Table 4: Red giant membership results

^a Not observed in the DDO system.
^b Star outside the range of the DDO calibration.

Although the results obtained from criterion A should be taken with caution because of the probable non-uniform reddening in the cluster field (Ahumada & Lapasset 1996), the agreement between the photometric analysis and the kinematic data is really excellent. This demonstrates once again that the photometric criteria A and B lead to reliable membership results provided the BV and DDO photometric data are of high quality. The only discrepant star (#184), an apparent radial velocity member, has a reddening significantly larger than those of the cluster giants (see Table 4), compatible with its position in the CM and CC diagrams. Since this star is located in an apparently obscured region in the cluster field and has a metal content nearly similar to that of the remaining red giants (see Sect. 3.2.1), we have retained it as a possible cluster member.

The interstellar reddening derived from Janes's (1977) method average to <E(B-V)$_{\rm GK}$> = 0.13 $\pm$ 0.03 mag, in very good agreement with the previous values derived by Dürbeck (1960) and Ahumada & Lapasset (1996). However, the individual E(B-V)$_{\rm GK}$ values listed in Table 4 were used to correct the DDO photometry for interstellar reddening.

3.2 Metal content

3.2.1 DDO and UBV abundance parameters

As a first abundance indicator, we have used the intrinsic DDO colour index C₀(41-42), which is an excellent abundance indicator measuring the strength of the $\lambda$4216 cyanogen band, such that the larger the index the greater the absorption by this band. Using this parameter we have computed for each cluster red giant the new cyanogen anomaly, $\Delta CN$,defined by Piatti et al. (1993) as the difference between the dereddened C₀(41-42) and the standard value of this index corresponding to a star with the same temperature and surface gravity, but not with the same C₀(42-45) and C₀(45-48) as the star in question. Column (3) of Table 5 lists the cyanogen anomaly $\Delta CN$ obtained for nine cluster giants. No value could be determined for stars #59, 179 and 219 which fall outside the range of Piatti et al. (1993) calibration. The mean cyanogen anomaly is <$\Delta CN$> = -0.035 $\pm$ 0.007 (m.e.), the negative sign indicating a weak cyanogen band compared with the mean for solar neigbourhood K giants. The cluster metallicity derived from the [Fe/H] versus $\Delta CN$ relation given by Piatti et al. (1993) is then [Fe/H] = -0.29 $\pm$ 0.10. We note that the DDO abundance derived for star #184 ([Fe/H] = -0.3) suggests again that this is a cluster giant.

  

Table 5: Abundance parameters of cluster giants

^a Not observed in the DDO system.
^b Outside the range of the DDO calibration.
^c Outside the range of the Washington calibration.

We also examine the cluster abundance by determining the ultraviolet excesses $\delta$(U-B) with respect to the field K giants. These quantities were derived using Janes's (1979) Eq. (7) and comparing the (U-B)₀ and (B-V)₀ intrinsic colours of the cluster giants with the standard class III two-colour line of Fitzgerald (1970) (see Fig. 2). The derived UV excesses are then directly comparable to $\Delta CN$, which is also based on typical field stars. The computed $\delta$(U-B) excesses are given in Table 5. The mean value <$\delta(U-B)$> = -0.03 $\pm$ 0.01 (m.e.) derived from 10 cluster giants implies [Fe/H] $\simeq$ -0.3, if Janes's (1979) Eq. (8) and Janes's (1975) Eq. (2) are used. We note that this value practically does not change if the three spectroscopic binaries #113, 179 and 269 are omitted. The resulting metallicity is then in excellent agreement with that found from the DDO data. Therefore, both $\Delta CN$ and $\delta$(U-B) values support the conclusion that NGC2354 is a moderately metal-poor open cluster.

3.2.2 Washington abundance parameters

The Washington photometric system provides several independent metallicity indicators. Geisler et al. (1991) have defined fiducial lines for solar abundance giants in the Washington colour-colour diagrams, including the $(C-T_1)_0\,/\,(T_1-T_2)_0$, $(C-M)_0\,/\,(M-T_2)_0$ and $(C-T_1)_0\,/\,(M-T_2)_0$ relations. The abundance-sensitive index $\Delta$ is the difference between the observed colour and the solar-abundance colour at the observed (T₁-T₂) [or (M-T₂)], where all colours refer to dereddened values. Geisler et al. (1991) described a method for correcting the decrease in abundance sensitivity with temperature and established new empirical calibrations of the abundance indices $\Delta^{\prime}_1-\Delta^{\prime}_5$ with [Fe/H], where $\Delta^{\prime}_1-\Delta^{\prime}_5$ refer respectively to $\Delta^{\prime}(C-M)_{T_1-T_2}$, $\Delta^{\prime}(M-T_1)_{T_1-T_2}$,$\Delta^{\prime}(C-T_1)_{T_1-T_2}$, $\Delta^{\prime}(C-M)_{M-T_2}$ and $\Delta^{\prime}(C-T_1)_{M-T_2}$. These $\Delta^{\prime}$ indices can be calculated from the $\Delta$ indices using Eq. (2) of Geisler et al. (1991). The derived Washington abundance indices for NGC2354 giants are given in Cols. (4)-(8) of Table 5. Stars #184 and 200 fall outside the range of the Washington calibration. The resulting mean values and standard deviations of the mean from ten cluster giants are:

<$\Delta^{\prime}_1$> = <$\Delta^{\prime}(C-M)_{T_1-T_2}$> = -0.14 $\pm$ 0.03,
<$\Delta^{\prime}_2$> = <$\Delta^{\prime}(M-T_1)_{T_1-T_2}$> = -0.04 $\pm$ 0.01,
<$\Delta^{\prime}_3$> = <$\Delta^{\prime}(C-T_1)_{T_1-T_2}$> = -0.18 $\pm$ 0.03,
<$\Delta^{\prime}_4$> = <$\Delta^{\prime}(C-M)_{M-T_2}$> = -0.09 $\pm$ 0.02,
<$\Delta^{\prime}_5$> = <$\Delta^{\prime}(C-T_1)_{M-T_2}$> = -0.10 $\pm$ 0.02.

These values practically do not change if the three spectroscopic binaries #113, 179 and 269 are omitted. Using the abundance calibration of Geisler et al. (1991), the above mean indices yield [Fe/H]₁ = -0.37 $\pm$ 0.06, [Fe/H]₂ = -0.33 $\pm$ 0.06, [Fe/H]₃ = -0.37 $\pm$ 0.06, [Fe/H]₄ = -0.32 $\pm$ 0.07 and [Fe/H]₅ = -0.32 $\pm$ 0.07. The average of the five Washington abundance estimates, [Fe/H] = -0.34 $\pm$ 0.02 (s.d.), is in very good agreement with the values derived from both the DDO data and the UV excesses.

NGC2354 is therefore on the metal-poor side of the distribution of the intermediate-age open clusters. Since this cluster is located about 1.4 kpc from the Sun at l = 238$\hbox{$^\circ$}$, its adopted metallicity ([Fe/H] = -0.30) is consistent with the existence of a radial metallicity gradient in the Galactic disk (see, e.g., Janes 1979; Piatti et al. 1995).