The Crash of Comet SL9 on Jupiter Observed with the FTS

Observing conditions

This extraordinary event has been observed by almost all the ground-based telescopes available from the South Pole to the mid-northern latitudes. The CFH Telescope participated in this exceptional campaign by being scheduled from July 16 to July 27, with two instruments sharing the time. The FTS was on the telescope for the first 5 days and the French 10-micron camera for the rest of the time. With this partition, the FTS was covering most of the period of the impacts. The 20 final impacts were spread over 5 days and 12 hours, the first impact A taking place on July 16 around 10 a.m. (Hawaiian Time), and the last one W, at 10 p.m. July 21. The impacts occurring when Jupiter was visible from Hawaii were those created by the fragments named C, G, R, V and W. Given the nights allocated to the FTS we could expect to observe directly only the C, G and R impacts. The weather conditions made it impossible to observe G on the second day of the event. Taking no images, the circular entrance aperture of the FTS was placed on an impact site and kept accurately on this position by following the rapid rotation of the planet. A full revolution is only 10 hours. A precise guiding was obtained by placing the box of the autoguider on the image of a Galilean satellite and by changing continuously the offset of the satellite relative to the pointed position on Jupiter by an appropriate drift of the bonnette mirror. The selected aperture had a diameter of 2.5'', which, at the time of the observation(Jupiter's diameter was 38'') was corresponding to a projected diameter on the disk of Jupiter of 9400 km. Around the FTS, the observing team was composed, beside the author as P.I., of P. Drossart, B. Bezard from Meudon Observatory (France) and J. Caldwell from York University (Canada). But the proposal was associating with the observers a total of 12 people in an international team: T. Owen (IfA, Hawaii), C. de Bergh, E. Lellouch, A. Marten (Meudon Observatory, France), S. Atreya (U. of Michigan), S. Kim (Goddard), H. Waite (SWRI) and R. Yelle (U. of Arizona). In mid-July, Jupiter was more than 2 months and half passed the opposition, which means that each day the observations started at 2:30 p.m. in the afternoon to stop not much later than 10:30 p.m. at an airmass of already 2.5, Jupiter being at a south declination of -12 deg. Due to these particular observing conditions it had been decided to focus with the FTS on the 4.7-micron region. The sky is rather dark at this wavelength, making possible daytime observations.

Scientific goal

Other reasons led to this choice. A large majority of telescopes were planning to observe the event with cameras, from the visible to the thermal infrared. There were no serious guiding problems and the results are, for sure, more spectacular for the general public. However, spectroscopy is the necessary complement. Among the spectrometers available behind optical telescopes a small number covers the 4.7-micron window. The FTS is well adapted to observe in this spectral range and in addition is unique by giving access to a full coverage of the window at a high spectral resolution. In the study of Jupiter, this window presents a specific interest. The thermal emission from Jupiter is dominating. No absorption from CH4 is blocking the Jovian flux and the clouds which cover the Jupiter's surface are partially transparent. Therefore, this region is ideal to probe deep in the Jovian troposphere. In the past, several components were detected for the first time by analyzing spectra of this window. Fundamental molecules like CO, H2O, CH3D, and more exotic species like PH3 and GeH4 have been identified. By observing at high spectral resolution the impact sites in this spectral range, we were expecting to study the modifications of the troposphere generated by the heating of the deep atmospheric layers through the molecules cited above. Additional filters to cover the infrared domain from 1.4 to 4 micron, which is possible with the FTS, were also installed to be ready for some serendipitous search for eventual other molecular signatures. It was sure that the final observing strategy had to be adapted following the evolution the event.

Summary of the results

Table 1 sums up the impact sites which were observed over the 4 days of the run. The age of the impact at the time of the observation is indicated with the spectral range and the spectral resolution. As explained in the previous paragraph, a large amount of observing time was spent by using a filter centered at 4.7 micron (2080 - 2180 cm-1) particularly adapted to detect CO. It can be noticed that three other filters were also used, respectively at 3.3 micron (2850 - 3050 cm-1), 2.3 micron (4100 - 5200 cm-1) and 1.6 micron (5400 - 6800 cm-1).

the 4.7 micron region:

Five sites, B, C, G, L and R were observed, G being observed twice with one day interval. Respectively to the impact time, some were observed less than one hour to almost two days after the impact. From the observations around the world the B impact turns out to be almost undetectable. Actually, the fragments from the middle of the train created the biggest impacts. We have decided to consider this spectrum as a reference. A first result from these data is the detection of a strong emission of CO on the L site observed about 4 hours and half after the impact. This one was among the three biggest events. G was another energetic impact but no CO in emission was detected when observed one day later. This spectrum is quite comparable to the reference spectrum which may indicate that the deep troposphere was heated by these bolides and returned to its original status after more than 5 hours and less than 24 hours. On the contrary, for C and R observed shortly after the impact (see table) no CO in emission was detected. This is an indication that for these impacts the troposphere was not modified. They did not go deep enough to change noticeably the chemical composition. CO was observed also in emission in the millimetric range at IRAM. This CO is of stratospheric origin where it is normally not observed. It remained observable for two weeks. These millimetric observations have not enough spatial resolution to be able to attribute the emission to a particular site. This stratospheric CO may come from the tropospheric CO injected in the stratosphere by the major impacts, for which the glow was detected by the FTS few hours after the impact on one site.

3.3 micron region:

The 3.3 micron filter was chosen to detect the two directly observable impacts, C and R. This filter covers the strong 3 methane band which makes normally the planet totally dark (except at the poles) at this wavelength. At the instant of the impact, in the two cases, a strong emission was seen centered at 3000 cm-1, which corresponds to the Q branch of the band. The maximum of emission was detected 6 min after the impact for C and 11 min for R, but in both cases faded rapidly to become negligible 10 min later. This brief emission can be explained by the sudden heating of the stratospheric methane by the fireball following the impacts. However, no modified temperature profiles are able yet to reproduce properly the observed spectra. More work is needed. It must be emphasized that we are probably the only team who got a complete spectral coverage of this band at the time of impacts. Detection of an emission attributed to methane was reported by a team on board the Kuiper Airborne Observatory with a spectrometer covering another CH4 band at 7.8 micron, and by a team at ESO with a spectrometer covering a small spectral range at 3.53 micron, at the edge of the range we obtained with the FTS.

Figure 12

the 2.3 micron region:

From 4100 to 4650 cm-1, the solar reflected spectrum is totally absorbed by the methane above the clouds located at the 2-bar level. Beyond this limit, from 4650 to 5200 cm-1, the pressure dipole-induced spectrum of H2 is the major source of opacity. Two sites of impact were observed with the FTS on the whole range. For both, a continuum level is observed (fig. 12). The unusual presence of this continuum indicates that a stratospheric haze has been deposited by the impacts, high enough to limit the absorption of the solar flux by CH4 and H2. Actually, this layer is the origin of the bright spots seen in the numerous images of the southern hemisphere of Jupiter taken during the event by the telescopes equipped with infrared cameras able to observe in the same spectral range. The spectroscopy explains the cause of the brightness of the spots and should allow a determination of the altitude of the reflecting layer, probably close to the 10-mbar level. A bright zone at -43 deg. latitude was still present in mid-September.

the 1.6 micron region:

In this window a large portion of the solar reflected spectrum is absorbed by the 2 3 band of methane. In the spectra taken on 2 impact sites, the absorption remains total. It means that the methane absorption above the stratospheric layer clearly detected at 2.3 microns is still strong enough to block the solar reflected at this wavelength, which is another constraint on the altitude of the stratospheric cloud.


This crash of SL9 on Jupiter was a surprise for all the observers. However, nobody was able to obtain by himself a complete picture of the whole event. Images and spectroscopy at all wavelengths, from UV to millimetric and even centimetric and at different times are necessary to build a full understanding of it. At CFH, we have been participating to this effort. The results we have obtained need a more complete analysis which has to be conducted in close collaboration with other observations made around the world.

J.-P. Maillard, Institut d'Astrophysique de Paris CNRS

The Kinematics of Globular Clusters From SIS Fabry-Perot Observations

Surveys have found 20-25 globular clusters whose surface brightness profiles continue to increase with decreasing radius until the seeing disk is reached (Djorgovski and King 1986), rather than leveling off at a constant surface brightness. These ``cusps" in the profiles have been widely interpreted as evidence supporting the long-standing theoretical prediction (Henon 1961, 1971; Spitzer and Hart 1971) of core collapse. This evolution of the original cluster core towards higher and higher density is partly driven by the settling of heavy stars towards the center because of the equipartition of energy (mass segregation). However, the dominant process is the gravothermal instability, in which energy conducted outward from the hotter (higher velocity dispersion) core by two-body gravitational encounters causes the core to shrink and become still hotter because of the negative specific heat of self-gravitating stellar systems (for reviews of cluster evolution, see Elson etal. 1987 and Hut etal. 1992). This `core collapse' may play an important role in the formation of active galactic nuclei, as well as in globular cluster evolution.

It has also become apparent that the high densities of stars resulting from core collapse produce stellar collisions that can alter the stellar population by creating such exotic objects as blue stragglers, x-ray binaries, and milli-second pulsars (eg, Hesser 1993). It may even influence such traditional cluster properties as the horizontal branch type (Fusi, Pecci etal. 1993), which has implications for the possible age spread in the globular cluster system.

Models of core collapse are making increasingly detailed predictions, but testing these has been hampered by the lack of velocity data in and around collapsed cores (eg, Grabhorn etal. 1992; Drukier 1994a,b). Unfortunately, dispersions measured by the broadening of the lines in integrated-light spectra, which is the traditional approach when individual stars cannot be resolved, have yielded contradictory results for the cusps of globular clusters (Peterson etal. 1989, Zaggia etal. 1992, Dubath etal. 1993). The last two of these papers showed that a dispersion estimated from an integrated-light spectrum of a small region of a globular cluster is uncertain because the light comes mostly from only a few bright stars.

To provide detailed kinematic data for the central regions of globular clusters, in 1992 we began a program of observations using the Rutgers Fabry-Perot on the CTIO 4-m telescope. Our procedure is to take narrow-band images of the cluster core in a sequence of wavelengths passing through a strong absorption line. Then we employ the DAOPHOT crowded-field photometry program (Stetson 1987) to construct a short spectrum and extract a radial velocity for each resolved star in the field (see Gebhardt etal. 1994a, hereafter G94a). This is the most efficient way to obtain velocities in the crowded centers of clusters, where fiber-fed spectrographs suffer from constraints on how close together fibers can be placed. On the CTIO 4-m, about four hours of observations yield velocities accurate to 1-5 km/s for stars as faint as V=18 throughout a 1 arcmin radius field. One field centered on 47 Tuc produced 548 velocities and five fields in M4 produced a total of 1896 velocities. These samples are among the largest ever obtained for globular clusters.

Though our CTIO program is very successful (G94a,b; Gebhardt & Fischer 1994, hereafter GF94), we desired better seeing than our typical 1.5 arcsec FWHM. The number of stars that we can measure in globular cluster cusps is set by crowding, so good seeing is crucial for making decisive tests of the theory of core collapse. Such observations can also constrain the number of heavy stellar remnants in the cluster, which are expected to be concentrated at small radii by mass segregation. Thus we arrived at CFHT in May of 1994 for a run with SISFP.

We used the Rutgers 0.8 ¸ resolution etalon because of concern about dust between the capacitors of the similar CFHT etalon. Our etalon also has broader coatings, allowing us to work as red as the Hà line, where cluster giants are brighter and CCD quantum efficiencies are greater. The Rutgers and CFHT etalons operate with the same Queensgate controller. A Rutgers-manufactured filter holder placed the order-selecting filter in the parallel beam and tilted it to prevent filter reflections from reaching the CCD. We were more comfortable with the filter in this location than the usual alternative, which had the filter nearly in focus. These measures were successful in preventing the recurrence of mysterious flat-fielding problems which had made our 1993 CFHT data unusable.

The set-up and check-out of this unusual combination of equipment proceeded smoothly with expert and enthusiastic support from the CFHT staff. We were particularly grateful for the installation of our deuterium calibration lamp at one of the comparison ports, for which it was not designed. This allowed the lamp to be computer controlled and allowed us to take projector flats from Hale Pohaku (we need flats for every one of the 20-30 wavelengths at which we work during a run!).

Unfortunately, the weather was less cooperative than the staff. We had varying amounts of clouds during the entire run. On our fourth and final night we were forced to dismantle hurriedly our equipment and abandon the mountain in a blizzard. We did, however, manage to get complete series of exposures (15, spaced by 0.33 ¸) on two clusters: M3 and M15.

The seeing was 0.5 arcsec-1.0 arcsec (typically about 0.8 arcsec) for our observations. During our first two nights a misalignment between the CCD and SIS guide-probe coordinates meant that the guide probe was not being placed on the chosen star (moonlight reflecting from the clouds provided enough counts that we did not at first realize that the star was absent). Once this problem was fixed, fast-guiding provided the expected 10-20 % improvement in the FWHM of the images. While this may seem a very modest gain, the reduction in the area of the stellar images goes as the square of this. Reducing this area by 40 % produces dramatic gains in the number of stars that we can resolve and, hence, measure.

These data have yielded velocities, accurate to better than 5 km/s, in a field 3 arcmin across for 409 stars in M3 and 483 in M15. These data represent a total of about 10 hours of observing in marginal conditions (half of the frames have only 50 % of the counts of the other half). We measured velocities for 27 stars within 6 arcsec of the center of M15 and 100 stars within 14 arcsec. These samples are five times larger than those of Peterson etal. (1989) in these regions. They are three times larger than our FP samples obtained at CTIO (G94a).

Our large sample in M3 contains more than a dozen stars with velocities as far from the cluster mean as the two "high-velocity'' stars reported by Gunn & Griffin (1979). Because we find a larger central velocity dispersion than Gunn & Griffin, 8 km/s vs 5 km/s, all of these stars are consistent with a normal distribution. This parallels what we have found for ``high-velocity'' stars seen in other clusters (G94b).

Figure 13: Projected velocity dispersion vs projected radius in M15 derived mostly from SISFP data. The filled circles are the absolute magnitude of the deviations of the individual measurements from the cluster mean velocity plotted vs radius from the cluster center. Points outside 1.5 arcmin are from Peterson et al. (1989). The solid line and the dashed lines are the estimate of the dispersion profile and the 90% confidence interval, respectively.

Figure 13, plots velocity dispersion vs radius in M15. The solid circles are the absolute magnitude of the deviation of the 537 stellar velocities from the cluster systemic velocity. Beyond a radius of 1.5 arcmin the velocities come from Peterson etal. (1989). The solid line and the dashed lines are the estimate of the dispersion profile and the 90 % confidence interval, respectively, calculated using LOWESS smoothing (see G94a,b). The profile's behavior inside about 0.3 arcmin strengthens G94a's conclusion that the profile flattens at small radii. We will analyze the integrated light from the central few arcseconds to look for the ~25 km/s dispersion seen by Peterson etal. (1989), but our stellar data suggests a central dispersion of 11.5 km/s.

Figure 14: Mass density vs spatial radius in M15, derived from the profile in figure 13, under the assumption of an isotropic velocity dispersion tensor. The solid line is the estimated density profile and the dotted lines show the 90% conficence interval. The dashed line is the profile predicted for cusps by the models of core collapse.

Our velocity dispersion profile and the cluster surface brightness profile both refer to the same dynamical population: the giants and turnoff stars. If it is assumed that this population has an isotropic velocity dispersion tensor, then combining the profiles with the Jeans equation yields a cluster mass density profile (GF94). Figure 14 shows the resulting density profile for M15. The solid line is the estimated density profile and the dotted lines show the 90 % confidence interval. The density profile extends to 10 times smaller radii than that of GF94 and shows the direct detection of densities as high as 106 M -pc-3. Inside a radius of 0.8 pc, the density profile is close to a power-law with an exponent slightly shallower than -2. The dashed line is the theoretically-predicted slope of -2.2, which agrees with the profile to within its uncertainties.

Figure 15: M/Lv vs spatial radius in M15, derived by dividing the mass density by the luminosity density profile. Note the increase at small and large radii, reflecting the segregation of feavy, dark remnants towards the center and low-mass, dim stars towards the outside.

Figure 15 shows the M15 visual-band mass-to-light ratio (M/LV) profile obtained by dividing the mass density profile by the luminosity density profile. The solid line is the estimate and the dotted lines show the 90 % confidence band. The increase in the M/L at small radii is presumably due to the concentration of dark, heavy stellar remnants towards the center by mass segregation. Similarly, the increase at large radii is probably due to the concentration of dim, low-mass stars there. More detailed analysis of these profiles is currently underway.

In summary, SISFP is an efficient, exciting instrument for obtaining two-dimensional velocity data with high angular resolution; we look forward to observing more clusters with better weather!

C. Pryor & T. Williams (Dept. of Physics & Astronomy, Rutgers, the State University of New Jersey), K. Gebhardt (Dept. of Astronomy, University of Michigan), J. E. Hesser, (DAO/NRC)


  1. Djorgovski, S., & King, I. R. 1986, ApJ Lett., 305, L61
  2. Drukier, G. A. 1994a,b, preprints.
  3. Dubath, P., Meylan, G., & Mayor, M. 1993, in Structure and Dynamics of Globular Clusters, ed. Djorgovski & Meylan (ASP), 69
  4. Elson, R., Hut, P., & Inagaki, S. 1987, ARAA, 25, 565
  5. Fusi Pecci, F., Ferraro, F.R., Bellazzini, M., Djorgovski, S., Piotto,G., & Buonanno, R. 1993, AJ, 105, 1145
  6. Gebhardt, K. & Fischer, P. 1994, AJ, in press (GF94)
  7. Gebhardt, K., Pryor, C., Williams, T., Hesser, J.E. 1994a, AJ, 107, 2067 (G94a)
  8. Gebhardt, K., Pryor, C., Williams, T., Hesser, J.E. 1994b, AJ
  9. , submitted (G94b)
  10. Grabhorn, R. P., Cohn, H. N., & Lugger, P. M. 1992, ApJ, 392, 86
  11. Gunn, J.E. & Griffin, R.F. 1979, ApJ, 84, 752
  12. H‚non, M. 1961, Annales d'Astrophys. 24, 369. & 1971, Astrophys. Space Sci. 13, 284
  13. Hesser, J. E. 1993, in Structure and Dynamics of Globular Clusters, ed. Djorgovski & Meylan (ASP), 15
  14. Hut, P. etal. 1992, PASP, 104, 981
  15. Peterson, R.C., Seitzer, P., and Cudworth, K.M. 1989, ApJ, 347, 251
  16. Spitzer, L. & Hart, M. H. 1971, ApJ 164, 399
  17. Stetson, P. B. 1987, PASP 99, 191
  18. Zaggia, S.R., Capaccioli, M., Piotto, G. & Stiavelli, M. 1992, A&A, 258, 302

Further Spectro-Imagery results on Planetary Nebulae with BEAR

During a successful observing session in September 1994, we obtained new spectro-imagery results on proto-planetary and planetary nebulae using the instrument BEAR at the 3.60-meter CFH telescope. With the aim to study the distribution and properties of the molecular hydrogen in the inner regions of the molecular envelopes of planetary nebulae (see first results reported in the CFH Bulletin n. 31, pp 2-3), we observed both in the K' and a narrow H2 filter three well-known sources. The results of these observations will be presented in detail elsewhere. Here we summarize near-infrared mapping results in the H2 filter obtained with BEAR towards the young planetary nebula BD+30ø3639 to further illustrate the unique capabilities of this instrument.

The color composite picture (Figure 1, cover) displays the velocity integrated emission of molecular hydrogen in the 1-0 S(1) transition at 4712.9 cm-1 towards BD+30ø3639 (the continuum emission has been subtracted). For comparison, the distribution of the ionized hydrogen in the Br line at 4616.6 cm-1 is shown as contours.

The contrast between the symmetrical, inner ring of ionized gas and the broken, asymmetrical outer distribution of molecular gas is particularly striking. The molecular hydrogen emission clearly delineates an outer, distinct shell about twice the size of the HII region. Its distribution is non-uniform and appears much brighter to the east than to the west. The inner regions are dominated by the continuum and line emission of the ionized gas including the Br (shown here) and the Helium lines. The overall morphology with the ionized gas abuting the outer molecular envelope is strongly reminiscent of a high-density photodissociation region. However the characteristics of this photodissociation region are rather unusual. First, the brightest lobes of the molecular hydrogen emission are aligned East-West, almost perpendicular to the direction defined by the brightest lobes of the Br emission. An East-West orientation is also evident in mid-infrared high resolution images in the dust continuum (Hora et al., 1990) or in the near-infrared dust emission bands (Bernard et al., 1994 - see also CFH Bulletin n. 21 pp. 19-21). Second, the expansion velocities of the molecular and ionized gas are significantly different: whereas the ionized gas expands at ~20 km/s, the expansion velocity of the molecular gas is as high as 50 km/s both in CO (Bachiller et al., 1991) and in H2 (present BEAR measurements). We note that an expansion velocity of about 50 km/s is much larger than the typical expansion velocities found in evolved planetary nebulae. A comprehensive study of the morphology and dynamics of this unusual molecular envelope around the planetary nebula BD+30ø3639 will be given elsewhere.

P. Cox (Observatoire de Marseille), J.-P. Maillard (Institut d'Astrophysique de Paris), F. Rigaut (CFHT)

The Hubble Constant from Cepheid Variables in the Virgo Cluster

Despite decades of effort a definitive value of the Hubble Constant (Ho) has been elusive. One of the primary reasons why Ho is so important is that a particular value may provide fundamental knowledge about the character of the universe itself. While significant and rapid progress is being made in establishing the extragalactic distance scale, the majority of modern "secondary'' methods indicate that Ho=80+/-10 km/s/Mpc (e.g. Jacoby et al. 1992; van den Bergh 1992). This value is troubling to many in that the corresponding expansion age for the universe is only 8 - 13 billion years (depending on the density parameter), which appears at odds with the estimated ages of the oldest globular clusters, which are thought to be 16+/-2 billion years old (e.g. VandenBerg 1991). Since many of the secondary methods of estimating distance are rather empirical with uncertainties in the underlying physics there remains considerable skepticism as to whether the extragalactic distance scale is well determined. As a result, one of the "key projects'' of the Hubble Space Telescope (HST) was to apply a "primary'' method of estimating extragalactic distances (the Period-Luminosity Relation of Cepheids) to galaxies at intermediate distance, as well as to galaxies in the Virgo Cluster, to also provide more definitive calibration of the secondary methods. A directly measured distance to the Virgo Cluster itself will be a byproduct of this key project. Unfortunately, the severe spherical aberration of HST greatly hampered its use until the repair mission in the fall of 1993. In the meantime, the success of HRCam on the CFHT appeared to indicate that a ground-based attempt at finding Cepheids in Virgo Cluster spirals was feasible. Therefore, we undertook a survey of the Virgo Cluster spiral NGC 4571 with HRCam on the CFHT beginning in 1991. We hoped to find Cepheids which would make it possible to establish a more secure basis for the extragalactic distance scale (see Pierce et al. 1994).

NGC 4571 was chosen as an ideal candidate for resolution into stars because it is a moderate-luminosity spiral (Sbc II) with a correspondingly lower disk surface brightness such that it should resolve relatively easily into stars. In addition, it is sufficiently luminous to contain a reasonable number of bright Cepheids. NGC 4571 is located less than 2.5 degrees from M 87 and shows signs of significant HI stripping (van der Hulst et al. 1987). This provides strong evidence for NGC 4571 being in, or close to, the "core'' of the Virgo Cluster, since spirals with significant stripping are within 3 degrees of M 87 (e.g. Cayatte et al. 1990).

Although the scheduling problems were severe we managed to obtain 13 epochs of imaging of NGC 4571 during the 1991 - 1993 seasons. Our median FWHM was 0.49 arcsec for 1 - 1.5 hr integrations in the R-band. The worst images had FWHM = 0.65 arcsec with the best having FWHM = 0.39 arcsec. The typical limiting mag was R ~ 26 enabling 10% photometry at R = 25. We also obtained two epochs in V in order to obtain colors for our candidate variables so as to distinguish Cepheids (with V-R~0.5) from Long-Period-Variables (having V-R ~1.0).

We obtained photometry of the resolved stars in NGC 4571 via DAOPHOT/ALLFRAME (Stetson 1987; 1994). ALLFRAME was ideal for our survey since it performs simultaneous point-spread function fitting photometry on multiple image frames using a "master'' list of program stars with fixed relative positions. This "master'' list was obtained by running DAOPHOT II on a median image using the 53 best (FWHM ~ 0.55 arcsec) individual exposures (Figure 16).

Figure 16: A combination of 53 15-20 min exposures of NGC 4571. The image has a FWHM=0.49 arcsec and is 150 arcsec on a side with 0.11arcsec/pixel. The image below is the same, after a smoothed version of the residual image (with fitted stars subtracted) has been subtracted. The limiting magnitude of these images is R~28. Note the numerous stars and star-forming regions.

The identification of variables was made by defining a variability index from the ALLFRAME photometry lists and formed using four sets of adjacent epochs (Welch & Stetson 1993). The non-photometric nights were referenced to those that were photometricly calibrated via a robust mean zero- point determination for each epoch (DAOMASTER). A plot of the variability index as a function of R magnitude appears in Figure 17. In the absence of additional factors which might correlate magnitudes from different epochs (e.g., systematic flat-fielding errors, close companions, etc.), the distribution should be symmetric about zero. Stars with unusually large positive variability indices are likely variable stars. Figure 17 also displays confidence intervals for the 5th-, 50th-, and 95th-percentile points of the distribution for stars within 0.5 mags of the abscissa. Several candidates clearly warrant further inspection. Stars above the 95th-percentile were examined for spurious "variability'' due to correlated photometry with close companions, background nebulosity, etc.

Figure 17: The variability index of Welch & Stetson (1993) as a function of mean R magnitude. The three irregular curves indicate the 5th, 50th and 95th percentiles of the variability index distribution as derived from the uncorrelated magnitude residuals of stars of similar brightness and crowding. Only stars further than 22 arcsec from the nucleus of NGC 4571 were considered (see text). The three were identified from their variability index, while star 2114 (the open circle) was identified from a visual examination of the photometry of the most isolated stars.

Two stars were judged as genuine variables with light curves and periods (see Fig. 18) consistent with Cepheids (#1451 and #2645). In addition, we also performed an independent visual examination of the photometry of all stars with R ~ 25.0 which were considered isolated (no companion within twice the FWHM). Although subjective, this approach also identified star #2645 as a variable and produced one additional candidate Cepheid, #2114.

A form of the "string length'' method for finding likely periods, called the Lafler-Kinman technique (Lafler & Kinman, 1964), was then employed to search for likely periods. We present the Lafler-Kinman statistic (Theta), and best-fit lightcurve for our candidate variables in Figures 18a-c. While minima in the Lafler-Kinman statistic are associated with the true period, additional spurious minima resulting from chance phase alignments of points and aliasing also occur. The minima due to aliasing are generally present in all the plots, which permits them to be identified. Most of these additional spurious minima do not result in reasonable light curves when the data are phased with that trial period and can therefore be eliminated as well. This procedure produced three stars that we concluded to be Cepheids (Figure 18). The photometry of these stars and their periods are given in Table 1.

Table1: Cepheid Variables in NGC 4571

  1. ID____log P____R_____V-R__m-M
  2. 2114__1.696___24.33___0.2___30.93
  3. 2645__1.759___24.14___0.5___30.93
  4. 1451__1.954___23.48___0.0___30.86

We followed the usual procedure of adopting the slope and zero-point of the P-L Relation for Cepheids in the LMC and computed the best-fitting relative distance modulus. We adopt the calibration given by Madore & Freedman (1991), and assume a distance modulus of 18.5 mag for the LMC. Namely:

M_R = -3.04 [log(P) - 1.0] - 4.48+/-0.08 mag,

where the period (P) is in days. The resulting apparent distance modulus for each variable was computed and is also given in Table 1. We find a mean apparent modulus for NGC 4571 of (m-M)A = 30.91ñ0.15 mag, where the uncertainty resulting from the finite width of the instability strip (å= 0.2 mag; Madore & Freedman 1991) and the distance to the LMC contribute equally. This value must be reduced according to any extinction of the Cepheids in NGC 4571. The Galactic foreground extinction toward NGC 4571 given by Burstein & Heiles (1984) is A(B) = 0.07 mag, which corresponds to A(R)=0.04 mag, assuming a standard reddening law. While there is likely to be some additional internal extinction within NGC 4571 we cannot estimate this accurately because of the limited precision in our colors for the Cepheids. However, the color of the "plume'' of brightest blue supergiants suggests little reddening. Therefore, we assume no additional extinction for the Cepheids in NGC 4571 but allow for an additional ñ0.10 mag uncertainty and adopt:

(m-M)o = 30.87+/-0.18 mag,

The corresponding distance is 14.9+/-1.3 Mpc. While the effects of contamination by close companions will tend to bias the mean magnitudes too bright there is no evidence for a large amount of contamination. Given the periods of our candidate Cepheids the amplitudes are expected to be relatively small. Simulations indicate that our best Cepheid candidate has both an amplitude and a color consistent with no significant contamination. For the remaining two candidates we cannot rule out a contamination as high as 0.5 mag, but the data are also consistent with none.

Figure 18:The panels on the left show the Lafler-Kinman string length (Theta, see text) as a function of log (P) for each of the three candidate Cepheids. Note that aliases due to our limited number of epochs result in spurious deatures which generally repeat in each panel and can thus be eliminated from consideration. The adopted periods for each of the variables are indicated. The panels on the right illustrate the R-band light curves for the three candidate Cepheids with the periods given by the string length analysis. Each point is shown twice to allow continuity through zero phase to be assessed.

As argued above, the estimated distance for NGC 4571 is very likely appropriate to the core of the Virgo Cluster itself. In fact, this estimate agrees with the majority of other methods currently in use (see Jacoby et al. 1992). The fact that HST has recently produced a Cepheid distance for M 100 (NGC 4321) which is in agreement with our own (Freedman et al. 1994) verifies our results and lends strong support to the view that the extragalactic distance scale is now established. Interestingly, the uncertainty in the HST Virgo distance is actually larger than ours. This is because the location of M100 within the cluster is poorly known due to the presence of considerable line-of-sight depth in the Virgo spiral population (e.g. Pierce & Tully 1988), whereas the HI stripping of NGC 4571 argues for placing it very close to the core of ellipticals. As a result, the two results are complementary despite the superior photometric data from HST.

Figure 19: The Period-Luminosity Relation for the Cepheids in NGC 4571 and the mean relation for those in the LMC assuming (M-M)lmc=18.5 (dashed line). The solid points indicate those variables identified from the variability index (see Fig. 18 and text), while the open point indicates the additional variable (#2114) found from an independent examination of the photometry of the most isolated stars. The adopted apparent distance modulus of (m-M)a=30.91+/-0.15 is derived from the mean of the three variables.

The implications of such a "short'' distance scale are well-known. Given that the relative distances between the Virgo Cluster and more distant clusters are in good agreement from a variety of methods, we can compute the Hubble Constant from the distance given above. We adopt a relative distance modulus between the Virgo and Coma Clusters of m = 3.70+/-0.05 mag (van den Bergh 1992) and a mean velocity for Coma of = 6888+/-86 km/s (Rood et al. 1972). With an additional correction of 258+/-10 km/s to place Coma in the reference frame of the CMB we find:

Ho = 87+/-8 km/s/Mpc.

This estimate assumes that no additional corrections to the velocity of the distant Coma Cluster are necessary due to peculiar motion. However, this value is likely to be representative over large scales because there is no evidence for very large streaming motions on scales of several hundred Mpc (e.g. Sandage & Tammann 1990; Lauer & Postman 1992). The corresponding expansion ages for the universe are 11.2+/-1.2 Gyr and 7.3+/-0.8 Gyr for Omega=0 and 1, respectively. These values are well-known to be in conflict with the estimated ages for metal-poor Galactic Globular Clusters (e.g. 16.5+/-2 Gyr; VandenBerg 1991) and an Inflationary cosmology with zero Cosmological Constant. The origin of this apparent paradox does not appear attributable to the observational data. A possible explanation is that the Cosmological Constant is non-zero.

M. J. Pierce (KPNO/Indiana Univ.), D. L. Welch (McMaster Univ.), R. D. McClure (NRC/DAO), S. van den Bergh (NRC/DAO), R. Racine (Univ. of Montreal), P. Stetson (NRC/DAO)


  1. Burstein, D. & Heiles, C. 1984, ApJS, 54, 33
  2. Cayatte, V. et al. 1990, AJ, 100, 604
  3. Freedman, W. et al. 1994, Nature, 371, 757
  4. Jacoby, G. H. et al. 1992, PASP, 104, 598
  5. Lafler, J. & Kinman, T. D. 1964, ApJS, 11, 216
  6. Lauer, T. R. & Postman, M. 1992, ApJ, 400, L47
  7. Madore, B.F. & Freedman, W. L. 1991, PASP, 103, 933
  8. Pierce, M. J. & Tully, R. B. 1988, ApJ, 330, 579
  9. Pierce, M. J. et al. 1994, Nature, 371, 385
  10. Rood, H. J. et al. 1972, ApJ, 175, 627
  11. Sandage, A. & Tammann, G. A. 1990, ApJ, 365, 1
  12. Stetson, P. B. 1994, PASP, 106, 250
  13. VandenBerg, D. A. 1991, in The Formation and Evolution of Star Clusters, PASP Conf. Ser. 13, ed K. Janes (San Francisco: ASP), p. 183
  14. van den Bergh, S. 1992, PASP, 104, 861
  15. van der Hulst et al. 1987, A&A, 177, 63
  16. Welch, D. L. & Stetson, P. B. 1993, AJ 105, 181

Return to Table of Contents