next previous
Up: Photometrically determined membership

5. Discussion

5.1. Theoretical isochrones

It is a standard technique to fit theoretical isochrones to the observed colour-magnitude diagrams and to select candidate cluster members from the relative positions of the stars in the former. An evolutionary age of 36 Myr has previously been determined for IC 2391 by applying this technique to the photometry of the early-type members. Assuming that coeval star formation has occurred, then many of the low-mass members will still be contracting towards the zero age main-sequence (ZAMS) and so it is important to use pre-main-sequence isochrones in any analysis. The most comprehensive and up-to-date computation of evolutionary tracks for low-mass stars (tex2html_wrap_inline1407) are those of D'Antona & Mazzitelli (1994). We have used the set of pre-main-sequence evolutions that have been modelled using the Alexander opacities (Alexander et al. 1989) and the mixing length treatment of Canuto & Mazzitelli (1990), as D'Antona & Mazzitelli have shown these to be in good agreement with observations of nearby M-dwarfs.

However, relating observed quantities such as V,B-V and R,R-I to the stellar parameters tex2html_wrap_inline1413 and tex2html_wrap_inline1415 is a non-trivial problem. Temperature calibrations have, in many instances, been determined for a particular range of spectral types and these different calibrations are not necessarily consistent. For example, Johnson (1966) derived a temperature calibration for early-type stars; Mould & Hyland (1976) for K-stars; Bessell (1991) for late-K to mid-M spectral types; and Reid & Gilmore (1984) for M-stars. In order to ensure some consistency, we have transformed the theoretical isochrones of D'Antona & Mazzitelli (1994) to observed colours and magnitudes as follows. For the late-K and M-stars (0.53<(R-I)<2.38), we have converted tex2html_wrap_inline1423 to (R-I) and tex2html_wrap_inline1427 to MR using the revised relations of Bessell (1995). Additionally, the empirical relations of Caldwell et al. (1993) were used to transform (R-I) colour to (B-V). For the earlier spectral types (late-F to late-K), we used the theoretical colours and bolometric corrections of Kurucz computed by Wood & Bessell (private communication) and which are available via anonymous ftp from This temperature calibration is quite similar to the IR flux method temperature scale of Blackwell & Lynas-Gray (1994).

5.2. Observational errors

In the photometric analysis we have assumed an evolutionary age of 36 Myr for IC 2391 (Lynga 1987) and a distance modulus, (m-M)=6.05, which has been derived using both photometric and spectroscopic techniques (Becker & Fenkart 1974). Adopting these values, we transformed a 36 Myr theoretical isochrone to fit the different combinations of CMDs and reddened these by the appropriate amount corresponding to a reddening value of E(B-V)=0.04 (Becker & Fenkart 1974).

In order to implement selection criteria for cluster membership, it is necessary however to consider the various sources of error in matching the observations to the theoretical isochrones. These include an uncertainty of

  1. tex2html_wrap_inline143910 Myr in the cluster age, based on the dispersion of measurements found in the literature
  2. tex2html_wrap_inline14390.2 magnitudes in the distance modulus (Becker & Fenkart 1974)
  3. the magnitude dependent photometric uncertainties (see Table 3 (click here))
  4. an error in the reddening value.
The slope of the pre-main-sequence isochrones and the reddening line in both the (V,B-V) and (R,R-I) CMDs is such that reddening serves to effectively move the isochrone to redder colour at all magnitudes for the colour range under consideration. As the reddening value of Becker & Fenkart has an uncertainty of tex2html_wrap_inline1447 magnitudes, we adopted an upper limit of E(B-V)=0.06 for the bright error limits. However, for the faint error limits we have considered the possibility of negligible reddening and have adopted a value of E(B-V)=0.00.

5.3. Cluster membership

Candidate membership of the open cluster IC 2391 was based on the positions of stars in the (V,B-V) and (R,R-I) CMDs. We also consider the effect of binarity on the location of the theoretical isochrones (see, for example, Dabrowski & Beardsley 1977). Such an effect will obviously depend on the frequency of binaries and on the distribution of their mass ratios. However, assuming that an undetected companion has a lower mass and hence redder colour, its presence will cause its position to be shifted upwards in brightness and redwards in colour. Hence, the maximum increase in brightness allowing for a companion of equal mass would correspond to 0.75 magnitudes. This effect has been included in our bright error limit, and possible binary members were identified if they were situated between the single star, bright limit and the bright error limit. Objects were selected as possible members of IC 2391 if they were situated between the error limits as defined above. An object was then deemed a candidate member of IC 2391 if it fulfilled the selection criteria for both the (V,B-V) and (R,R-I) diagrams.

tex2html_wrap_inline1191 CCD photometry was determined for 1303 objects in the field of IC 2391. Using the afore-mentioned selection criteria, 100 objects were identified as being possible cluster members, of which 83 satisfied the constraint for the (V,B-V) CMD and 34 satisfied the constraint for the (R,R-I) CMD respectively. A subset of these objects (16 of the 83 and 10 of the 34) were located within the binary envelopes of the respective CMDs. Seventeen objects satisfied both selection criteria and have been classified as candidate members; these exhibit a range of colours between approximately 0.4tex2html_wrap_inline1467(R-I)tex2html_wrap_inline14671.7 which corresponds to spectral types between G8V and M4V. Identification of main-sequence members of earlier spectral types was not possible, as these stars were saturated on the CCD frames. The photometry and sky charts for all 17 candidate members are presented in Table 4 (click here) and Fig. 4 (click here) respectively.

Figure 2: a) The (V, B-V) and b) (R, R-I) colour-magnitude diagrams for the new photometric dataset. The solid line corresponds to the locus of a reddened 36 Myr isochrone and the dotted lines represent our estimates of upper and lower error limits for single stars having membership of IC 2391. The dashed line indicates the bright limit which accounts for the effect of binarity. Filled circles correspond to the candidate members identified using the selection criteria described in Sect. 5.3

Figure 3: The (R-I, B-V) colour-colour diagram for the subset of stars that have been selected from the individual (V, B-V) and (R, R-I) CMDs using the criteria discussed in Sect. 5.3. Objects that satisfied both membership criteria are shown as filled circles. The solid line corresponds to the reddened 36 Myr isochrone

  figure348  figure355
Figure 4: Identification charts for the candidate members of IC 2391

For a further 125 objects, it was only possible to determine two-colour photometry in either BV or RI. In many cases, this was the consequence of their faintness and of the different limiting colour sensitivities. Additionally, some of the stellar profiles were severely contaminated by under-lying bad pixels or cosmic ray events. However, two of these objects (1316 & 1428) were identified as having possible cluster membership (see Table 4 (click here)).


Id. tex2html_wrap_inline1219 tex2html_wrap_inline1249 V B-V V-R V-I
162 08 40 05.68 -53 05 59.971 13.065 1.252 0.677 1.329
302 08 40 31.18 -53 11 44.415 14.605 1.384 0.724 1.544
311 08 40 32.50 -53 10 32.571 12.295 0.902 0.466 0.980
314 08 40 32.99 -53 10 29.804 13.217 1.099 0.646 1.276
348 08 40 38.40 -53 06 29.594 14.405 1.310 0.792 1.570
362 08 40 39.75 -52 45 58.903 12.700 1.265 0.711 1.377
448 08 40 52.26 -52 46 46.917 14.426 1.316 0.798 1.566
480 08 40 53.89 -52 42 14.523 14.142 1.267 0.719 1.510
581 08 40 59.89 -52 44 27.479 13.752 1.386 0.722 1.524
586 08 40 59.99 -52 44 21.420 13.572 1.267 0.691 1.470
711 08 41 06.96 -52 51 04.148 13.880 1.237 0.705 1.385
729 08 41 08.02 -53 00 04.475 17.955 1.591 1.357 3.097
872 08 41 15.40 -53 00 14.300 13.040 1.050 0.591 1.182
950 08 41 18.30 -52 47 52.839 14.045 1.463 0.856 1.617
955 08 41 18.65 -52 58 54.409 18.718 1.460 1.312 3.005
1056 08 41 23.10 -52 59 57.538 13.657 1.391 0.747 1.503
1144 08 41 26.57 -53 04 26.286 14.401 1.463 0.760 1.602
Id. tex2html_wrap_inline1219 tex2html_wrap_inline1249 V B-V
1316 08 40 25.78 -53 11 15.093 18.654 1.641
Id. tex2html_wrap_inline1219 tex2html_wrap_inline1249 R R-I
1428 08 41 39.62 -52 59 34.208 12.613 0.780
Table 4: Photometry of the candidate members


The procedure discussed above for identifying cluster members is essentially the same as examining the (R-I, B-V) colour-colour diagram (see Stauffer et al. 1989). In Fig. 3 (click here), we have plotted this diagram for the subset of stars that have been selected from the independent CMDs. It is clear that many of the stars with B-V values redder than 1.5 and with R-I values less than 1.2 can be excluded from further consideration. These objects probably correspond to the reddened background giant population. However, stars that have B-V values bluer than 1.2 fall near the theoretical locus and even the two-colour diagram does not serve as a strong membership criterion. Hence, there may be a high contamination factor due to background objects in our candidate membership list. Unfortunately, due to the small image area of the RCA CCD, it would not have been observationally feasible to photometer a large number of offset fields in order to estimate the background contamination. Therefore, in the following section, we shall attempt to compare our results with the number density of low-mass objects found in the Pleiades.

5.4. Background contamination

Although the total number of identified candidate members is small, it should be noted that approximately 25% of these objects are located within the limits for binarity, which is compatible with the binary frequency found in the Pleiades (Bettis 1975).

In recent years, many studies have been directed at the Pleiades in an attempt to determine the luminosity and mass functions (see, for example, Stauffer et al. 1991b; Hambly et al. 1991b; Schilbach et al. 1995). Unfortunately, this photometric study does not permit the construction of a luminosity function for IC 2391 (due to the small number of members that have been found). However, it is worthwhile to consider the luminosity function that is currently accepted for the Pleiades and to estimate what the total number of members to be expected in the area of sky (0.06 sq. degrees) that was photometered for IC 2391.

Stauffer et al. (1991b) have identified 369 Pleiads in the magnitude range, tex2html_wrap_inline1581, distributed across 16 square degrees of sky. This magnitude range is similiar to that obtained in this study of IC 2391. As the Pleiades and IC 2391 open clusters have different distance moduli, we have applied a scaling factor in our calculation to account for the observed differences in their spatial extent. We have adopted angular diameters that have been based on the distribution of the early-type members in each cluster (Lynga 1987), and based on these observations made for the Pleiades field, approximately seven members are expected to be found in our sample for IC 2391.

However, Pleiads identified by Stauffer et al. (1991b) are distributed over a large area of sky, whereas our IC 2391 photometry has been obtained close to the cluster core where the star density may be expected to be greater. In fact, Hambly et al. (1991a) presented membership numbers as a function of radius which show that this scenario is true for the Pleiades. We have taken the star numbers for the inner 0.6tex2html_wrap_inline1583 radius of the Pleiades, and estimate that approximately 24 members are to be expected in our sky sample, which is the same order of magnitude as the number of selected candidate members. These statistics would suggest that the background contamination of candidates is not severe. However, it should be noted that these estimates implicitly assume that both IC 2391 and the Pleiades have similar star densities and mass functions, an assumption which we can not comment further upon here.

5.5. Luminosity function

From an inspection of the CMDs (see Fig. 2 (click here)), it would appear that very few members are to be found for spectral types later than M0V in IC 2391. This observation may be the result of having identified a small sample of candidate members and furthermore, a greater contamination of background objects may exist at GK-spectral types for the reasons discussed in Sect. 5.3. However, Foster et al. (1996) observed a sharp decline of cluster members in IC 2602 (tex2html_wrap_inline1585 Myr) at a spectral type of about M4V (tex2html_wrap_inline1587). This dataset contained photometry of an offset field and this clearly showed that the selection criteria identified a significant excess number of stars in the cluster field compared with the former. Foster et al. were confident that the contamination of the selected possible members due to background objects was small, implying that this observed decline in the late-type members is real. Intermediate-resolution spectra of our candidates would enable us to obtain spectral classifications which, combined with the derived radial velocities, would permit confirmation of cluster membership.

Reid & Hawley (1996) have found a similar result in the old open cluster, M 67. In this case they have attributed the effect to dynamical "boiling-off'' of the lower mass cluster members due to gravitational interaction of the cluster with passing massive objects. IC 2391, however, is too young for this mechanism to have operated.

5.6. Comparison with previous results

Previously, only one investigation has been directed at the lower main-sequence of IC 2391. Stauffer et al. (1989) identified 10 GKM pre-main-sequence stars using both photometric and spectroscopic techniques. However, we did not obtain photometric measurements for any of these objects. This is a result of several factors. First, our brighter magnitude limit is tex2html_wrap_inline12392 magnitudes fainter than that of Stauffer et al. Secondly, we avoided fields within several arcminutes of the bright stars (tex2html_wrap_inline1591) as their reflected starlight contaminated the CCD frame, and thirdly, our sky coverage was severely curtailed by bad weather experienced throughout the run. Therefore, a direct comparison is not possible.

next previous
Up: Photometrically determined membership

Copyright by the European Southern Observatory (ESO)