Astron. Astrophys. 332, 93-101 (1998)

5. Discussion

From the previous analysis it turns out that the best estimate of the radio source position is almost equally distant from the three faint blue stars we have found in the PC1 field of view. None is closer than and consequently none is a realistic candidate for the optical counterpart. The nearest detected object is a faint feature in the F255W image (see Fig. 5), clearly visible in the V and I filters as a blend of two main sequence stars.

We have plotted (as filled circles) in the CMD of Fig. 2 and in the color-color diagram of Fig. 4 all the objects found in the error box centered on the position of the radio source. As it can be observed, they essentially occupy the expected positions for cluster stars, including the two faint main sequence stars found close to the best estimate of the radio source position shown in Fig. 5. However, looking at Fig. 1, it is interesting to note that the position of the radio source is nearly at the same distance () from two bright objects which require further comments: the bright HB star #7251 (AC29) and the bright BSS #7785.

star #7251

This is one of the brightest HB star in the PC chip and ). A straightforward identification with the catalogue of Aurière & Cordoni (1983) yields coincidence with star AC 29, which is however listed as a red star , rather than a blue object as we find here. The same object is also identified with stars #1247 and in the catalogue of GYBS and with #10849 ( and ) in Ferraro et al. (1997e). In the UV, U, V and I frames studied here, #7251 has a resolved fainter companion #7262 which is located in the faint yellow straggler-subgiant transition region in the UV as well as in the visible CMD's. If we merge the two components, the resulting composite system is consistent with the observations of GYBS and Ferraro et al. (1997e) done at lower angular resolutions. However, it would still deviate (by much more than the plausible uncertainty associated to the measures) from the observations carried out in 1978 and reported by Aurière & Cordoni (1983). Variability may be suspected as a possible cause for the large color variation. Alternatively, a different impact of crowding in the different ground-based observations may also change the derived colors.

Once deblended, #7251 is one of the brightest star in UV, while in the visible CMD , the star position is more ambiguous as it lies between the tip of the blue straggler sequence and the blue end of the HB. A conclusive explanation of this peculiar location can hardly be drawn at this stage.

star #7785

This star ) is listed as a bright BSS in the global sample of BSS in M 3 presented by Ferraro et al (1997a). It has also been detected by GYBS (#1196 in their catalogue) and similarly classified as BSS. This star is also present in the Bolte et al. (1993) survey, but it was not identified as BSS since it lies in the region between the BSS extension and the RGB. This is however probably due to the much lower resolution of ground-based observations rather than to variability.

5.1. Variability

The search for variables with periods of a fraction of a day and with amplitudes in the range of tenths of a magnitude is commonly used to identify RR Lyrae, cataclysmic variables or contact binaries in cluster cores (Guathakhurta et al. 1994 (M 3); Gilliland et al. 1995, (47 Tuc); Shara et al. 1995; Bailyn et al. 1996, (NGC 6752); Cool et al. 1995 (NGC 6397).

Within the present set of observations, since none of the selected filters is sensitive to the line, we have first searched for a short-term variability of the continuum by monitoring the sequence of measured magnitudes over the observing time for each star of interest. Then, using a different approach, we have also compared the magnitudes listed by two independent catalogues of observations for stars matched by their position.

Short term temporal variability was examined by plotting the residuals of instrumental magnitudes (with respect to the first acquired measure) for each available image in each filter versus time. The composite light curves which are thus derived span more than 5 hours and are displayed in Fig. 6.

Fig. 6. Composite light curves for possible variable objects. The relative instrumental magnitude variation with respect to the first acquired measure in each filter is plotted as a function of time. The error bars, calculated from the frame-to-frame scatter are also reported. Each filter is coded as follow: filled circles (U), filled squares (V), filled triangles (I) and asterisks (UV), respectively; crosses are for a nearby unblended star, slightly fainter in U,V and I than #7290 and sharing the same sky background. The zero-point reference of the time scale is assumed on 25 April 1995, .

Clearly there is no significant variation of the measured magnitudes for the bright HB star #7251 and for the BSS #7785 which are both found near the expected position of the radio source nor for #7880 and #6359 (the two stars lying on the HB extension). The observed changes for star #7290 are more questionable. Variations in U and V with amplitudes mag are larger than the internal errors computed from the rms frame-to-frame scatter at the luminosity level of the star. No significant change is observed in I- or in the UV-band respectively. The negligible residual magnitudes observed for a nearby, fainter reference star, sharing the same local background (plotted as crosses in Fig. 6(#7290)) excludes the local background as a possible source of the observed variations in U and V. We conclude to a marginal evidence of variability for this star.

Variability was also explored by comparing the presently observed UVI magnitudes with those listed in the catalogue of GYBS, based on the pre-refurbishment WFPC1 HST images. The two catalogues have a significant overlapping region (40% of the field covered by our PC1). Observations are separated by about three years. A set of 346 stars matched by their position was then extracted and analysed. The faint star #7290 has no counterpart in the GYBS catalogue.

The residuals (WFPC2-WFPC1) in the I,U,V magnitudes respectively, for the 346 stars in common, are plotted in Fig. 7. Nine stars (labelled in the figure) have residuals larger than in, at least, one filter and are considered as variable candidates. They are plotted as full dots in all the panels of Fig. 7 and are listed in Table 2, where the GYBS identification number is also reported. Variability is confirmed for four out of seven RR Lyr as listed by GYBS (see their Table 3) which fall in the common region of observation. The stars #5439 and #7364 are additional candidate variables; their position within the instability gap in the visible CMD is consistent with RR Lyrae type variables. The two stars located near the radio source (#7251, #7785) are both well within the of the regression line and do not show evidence for variability over the time scale sampled by our observations.

Fig. 7. UVI magnitude residuals between the data presented in this paper (WFPC2) and the pre-refurbishment WFPC1 magnitudes presented by GYBS, for the 346 stars in common, plotted versus our magnitudes. The dashed lines represent the envelopes. Candidate variables are plotted as heavy dots superimposed to their error bars and they are labelled with their internal running number. Their position is indicated at all wavelengths. Star #1498 is the blue straggler #576 of GYBS: variability is not apparent in V but shows up in U and I. The position of the BSS #7785, close to the radio source position, is circled.

Table 2. List of the candidate variable stars. Variability in simultaneous bands is noted (UVI). The identification are from Ferraro et al. (1997a - F97) and GYBS.

We measure changes in the observed magnitudes (which may indicate variability) for three well established blue straggler stars. BSS #1498 is one of the two blue straggler stars (#576; P hr) found variable by Guhathakurta et al. (1994) and its variability is confirmed here in U- and in the I-bands. Interestingly, this blue straggler may show out-of-phase variations in the various bands as observed for RR Lyrae-type pulsating stars. #7069 (#1457 in GYBS) is a new candidate BSS variable.

The faint BSS #7570 is listed as a blue straggler in Ferraro et al. (1997a) and in GYBS (#1082), who noted that this candidate blue straggler was not detected by Bolte et al. (1993). While this could be interpreted as a further evidence for a strong variability supported by the present observations, a close look at the WFPC2 images indicates that this star is member of a barely resolved blend, best visible at F336W, with about equal components. It is likely that this blend was not angularly resolved with the unrepaired HST, resulting as a brighter object in the catalogue of GYBS. Hence, the suspected large variability of this blue straggler is probably an artifact.

The composite light curves for the three BSS are also displayed in Fig. 6. The most remarkable feature is the weakening of the UV magnitude of #7069. A trend seems also present in U. This behaviour is not seen in the visible bands. This BSS, like #1498, shows out-of-phase variation in UV and in the visible.

© European Southern Observatory (ESO) 1998

Online publication: March 10, 1998
helpdesk.link@springer.de