3 Results

3.1 Colour-magnitude diagrams

Though not strictly calibrated, the stellar magnitudes reported in Table 3 allow us to place most of the stars in our database on a colour-magnitude diagram. In Fig. 2 we present Vversus (B-V) and V versus (V-I) colour-magnitude diagrams. The colours plotted have not been de-reddened (reddening towards M 67 is relatively small, E(B-V)=0.032-0.05, Nissen et al. 1987; Montgomery et al. 1993).

Figure 2: Colour-magnitude diagrams that show V versus (B-V) and Vversus (V-I) for all the stars in our observations. The variables listed in Table 5 are indicated with triangles, spectroscopic binaries with squares and X-ray sources with crosses. Periodic variables discussed in Sect. 3.3 are indicated.

The cluster main sequence is clearly present amid a field of apparent non-members, extending from the cluster turnoff at $V \sim 12.5$ down to the faint limit of our database at $V \sim 18.5$ (the range within which our database is roughly complete). The cluster binary sequence is also clearly apparent along this full range. The stars lying below and to the blue of the cluster main sequence have been noted in studies of M 67 before and are likely due to field stars in the halo (e.g. Richer et al. 1998). From the observing runs intended to study brighter sources (covering a few small areas in the cluster; see Table 1 and Fig. 1 in Paper I), some blue stragglers and a portion of the giant branch are also present for approximately 10<V<12.5.

   
3.2 Photometric variability - General

In Table 5 we present the 69 stars in our photometric database that meet the criteria for photometric variability discussed in Sect. 2.3. Table 5 also provides comments for most of the stars listed. These comments give additional information such as possible periodicities, evolutionary status, binarity, etc. Stars without comments are stars situated on the cluster main sequence that display only non-periodic variability in our observations.

Our criterion for identifying a star as variable is that the probability of its light curve being constant is smaller than 0.3%. Therefore, one expects that a small number of stars ($\sim$3) has been classified as a variable by chance. Statistically significant variability in more than one passband increases our confidence that the observed variability is real. In what follows, we refer to such stars as "high-confidence" variability candidates.

To give some clue as to the nature of the stars listed in Table 5 and, ultimately, to the physical origin of the observed photometric variability, we plot these stars in the colour-magnitude diagrams shown in Fig. 2. In addition to these variables (shown as triangles), we also indicate the known X-ray sources (shown as X's) and the known spectroscopic binaries (shown as squares). The paucity of spectroscopic binaries below $V\sim 14$ is a bias effect due to the sensitivity limit of present spectroscopic surveys in the cluster (e.g. Latham et al. 1992). We find photometric variables in all regions of the cluster colour-magnitude diagram; we discuss stars in each of the various regions in more detail in Sect. 4.

Figure 3: Light and colour curves for stars 2426, 2703 and 2976 (S 757) folded on the periods of 0.2888, 3.7 and 0.3600 days, respectively (see text, Sect. 3.3). Data from different observing runs are marked with different symbols: open circles for run 1, open triangles for run 5.

Of the 69 variables listed in Table 5, 38 are "high-confidence" variables (i.e. they show variability in more than one passband), and of these, 29 are known proper-motion members with a probability of 75% or greater (a total of 319 stars in our survey satisfy this membership criterion; Sanders 1977; Girard et al. 1989). Of these 29 high-confidence variable members, 16 are either known X-ray sources or binaries (or both). Another 2 stars (3780 and 4415) are situated on the binary sequence. Thus, among the proper-motion members of M 67 surveyed by us, there remain 11/319 (=3.4%) stars (all on the cluster main sequence) that exhibit "high-confidence" variability, the origin of which cannot presently be associated with binarity and/or X-ray activity.

Similar numbers are derived from the observations of Gilliland et al. (1991). Of the 124 stars monitored by them with V magnitudes within the limits of our study, 4 stars (not known to be X-ray sources or binaries) were observed to exhibit photometric variability at levels that would have been detected by us at the $3\sigma$ level or higher.

   
3.3 Photometric variability - Periodic

Aside from the X-ray sources which are the subject of Paper I, we detected definitive periods in the light curves of only four of these variable stars. These are: star 2426 (P=0.29 days), star 2703 (P=3.7 days), star 2976 (P=0.36 days), and star 3665 (P=0.27 days).

Star 2426 is located below and to the blue of the cluster main sequence. While the best period identified by our period search (highest peak in the periodogram) is $0.1444 \pm 0.0002$ days, the data points at minimum light have sufficiently large errors that it is unclear whether or not the light curve consists of two dips with unequal depths. The (B-V) and (V-I) colour variations are not significant. However, additional spectroscopic and photometric observations by Orosz et al. (2002, in prep.) show this star to be an Algol-type binary (with a double-peaked light curve) with an A-star primary; the radial velocities indicate that it is not a member of M 67. In Fig. 3 the light curve for this star is shown folded on the period of 0.2888 days.

Star 2703 is located on the cluster binary sequence and exhibits a periodic light curve with $P = 3.7 \pm 0.3$ days in V, and $P = 3.4 \pm 0.2$days in B, both retrieved from data of run 1. The I data do not show a statistically significant periodicity. The V-band light curve is shown in Fig. 3. The (B-V) and (V-I) colour variations are not significant.

Figure 4: Light and colour curves for the W UMa 3665 (ET Cnc or III-79) folded on the photometric period of 0.2712 days.

Star 2976 (S 757) is located at the top of the cluster binary sequence. The period was determined by the combined data of runs 1 and 5 (B and V), that together cover a time span of 2 years. The best period determined by our period search is $0.18000 \pm 0.00005$ days, producing a single-peaked, roughly sinusoidal light-curve. In Fig. 3 the data are folded on the double period, as we believe this star is in fact a new W UMa system. This star was first noted to be a photometric variable by Rajamohan et al. (1988), although no periodicity was reported. The star shows no significant (B-V) and (V-I) colour variations. We discuss this star and our reasons for labelling it a W UMa system in Sect. 4.3.

Star 3665 (ET Cnc, or III-79 in the nomenclature of Eggen & Sandage 1964) is located on the cluster binary sequence and was identified as a W UMa system with a period of 6.49 hours by Gilliland et al. (1991), but no error in the period was specified. We searched for a period in a narrow window between 0.1 and 0.15 days (the power at half the period is much larger than at the full period) and find a best period from our photometry of run 1 of $0.1356 \pm 0.0004$ days (in B and V, $0.1351 \pm 0.0004$ days in I) which gives a full period of $6.51 \pm 0.02$ hours, compatible with Gilliland's measurement. The data are folded on this period in Fig. 4. The system is slightly redder during primary eclipse than during secondary eclipse. This star is discussed further in Sect. 4.3.

  

Table 5: List of variable stars. From left to right: identification number from Tables 1 and 2; identification number and proper-motion membership probability from Sanders (1977) or Girard (1989) if available; average Vmagnitude and B-V and V-I colours from our photometry as listed in Table 3; rms of the light curves in B, V and I, and corresponding $\chi ^2$ probabilities, rms values in italics are not significant at the 3$\sigma$-level; comments.

In addition to these four periodic variables, we indicate in Table 5 possible periods for another six stars (stars 2365, 3112, H, 4385, 4516, and 4712) whose light curves show some evidence for coherent variations on time scales longer than our observing windows. While our period searches did not reveal statistically significant periods for these stars, we include them here as candidate objects for follow-up study. We caution that these periods are based solely on our visual impression of the light curves, and should be taken only as a qualitative suggestion of periodicity on the timescale shown. Of course, these stars do still exhibit statistically significant variability, whether or not that variability is indeed periodic. All of these stars are on the cluster main sequence.

We also list in Table 5 tentative periods for star 2280 and star C. Star 2280 is notable in that it is located far to the red of the cluster main sequence in the colour-magnitude diagram. Its colours are those of an M dwarf, so its position in the colour-magnitude diagram suggests that it is a nearby star of the M spectral type. Star C does not have the requisite colours to be placed on the colour-magnitude diagram.