My goal with this program was to collect large-aperture (UBV)J(RI)C Johnson/Cousins (hereinafter "UBVRI") photometry for galaxies with D25 larger than 5.0 arcminutes. My envisioned uses of such data included calibration of surface photometry, derivation of total magnitudes and color indices, studies of the luminosity function of nearby galaxies, and support of other programs needing photometric data. The final tally: 594 observations of 239 galaxies.
Along the way, I obtained aperture photometry of 18 Galactic globular clusters and more than 550 stars, and set up an internally consistent system of secondary standard stars around the equator. While these secondary standards are in principle independent of the usual Landolt equatorial standards -- at least in the R and I bands -- I've found that they represent essentially the same photometric system. So, the results I present here are, within the errors, on the "standard" UBVRI system.
The work has sat in my filing systems for decades, waiting for me to find the time to prepare it for "publication." My retirement has finally given me that time, and the World Wide Web and the Internet have made the presentation of the data possible.
However, as this work now depends on a technology more than half a century old, the discussions of the equipment and techniques I used will be mainly of historical interest. Nevertheless, the data here may still be useful for some work, particularly calibration of CCD images of the measured galaxies. Eventually, all of these data will be completely replaced by data collected from modern digital sky surveys. Those sky surveys are now (or soon will be; June 2018) in progress, and we can look forward to a photometric mapping of the extragalactic sky that we only dreamed of in 1980.
For a general introduction to galaxy photometry and the derivation of photometric parameters for galaxies, I can do no better than to refer you to the two papers by Ron Buta and his colleagues (AJ 109, 517, 1995a; and AJ 109, 543, 1995b) and the many references therein to earlier papers. What follows here is primarily a detailed example of the data-collection phase briefly described in those two papers. While I will also refer you to those papers for comparisons of the current data with other observers' data, I will give a brief summary of their findings below.
I've tried to keep this writeup readable, but have found myself slipping back far too easily into the passive voice I learned long ago for scientific writing. Please forgive whichever style you find least acceptable to your own sensibilities.
Following this short introduction, I've organized this write-up into several sections:
When I first envisioned the program in the late 1970s, a considerable amount of aperture photometry for large, nearby galaxies had been published, but at relatively small apertures. Digital surface photometry for galaxies was in its infancy, and the deep multi-color digital sky surveys that we now take for granted were still little more than dreams. Many of the large galaxies had poorly known total magnitudes and color indices, and the photometric data simply did not exist to derive these quantities.
While I realized that CCD or photographic surface photometry was the best way to find the photometric parameters for a galaxy, there still existed a need for calibration data at as large an aperture as possible. Surface photometry is also very time-consuming; few galaxies had been studied in the detail then because the technology available in the mid- to late twentieth century was barely adequate to deal quickly with the large data sets required.
The de Vaucouleurs and I had built the tools to derive total magnitudes and color indices from aperture photometry for all types of galaxies; these found their initial use in producing the Second Reference Catalogue of Bright Galaxies (RC2; Univ. of Texas Press, 1976). Thus, I chose to put together an observing list based on D25, the diameter in blue light at the 25th magnitude per square arcsecond isophote, the "standard" diameter of RC2.
RC2 itself provided a first approximation of such a list, supplemented by the UGC in the north, and the then-ongoing surveys of the southern sky: 1) by me at Edinburgh, collaborating with the de Vaucouleurs (SGC; UT Astronomy Department Monograph No. 4, 1985); and 2) by Andris Lauberts collaborating with a team at Uppsala and ESO (ESO-B; European Southern Observatory, 1982). The final observing list of large galaxies is in Table 1a.
I've updated positions, classifications, diameters/axis ratios, and redshifts as needed, usually from data in the Third Reference Catalogue (RC3; de Vaucouleurs et al., Springer-Verlag, 1991) as corrected by Corwin, Buta, and de Vaucouleurs (AJ 108, 2128, 1994). The D25 major diameters in particular changed considerably with the addition of much new data, primarily in the southern sky. Thus, there are several galaxies in the table which no longer meet my initial diameter requirement. Conversely, there are several galaxies now meeting that requirement that are not included here.
This list also predated the discoveries by e.g. 2MASS and SDSS of the large, tidally-distended star streams that are the remains of galaxies "cannibalized" by the Milky Way. These things are not appropriate for the single-channel aperture photometry that I had in mind, so would not have been included in the final observing list in any case. I excluded the Magellanic Clouds for that reason, but included M31 and M33 more or less whimsically -- they are, after all, accessible from McDonald Observatory.
My own observations at McDonald from 1982 to 1988 nominally covered the northern part of the list; those I observed are flagged with an "m" at the end of the name column. (Galaxies observed at Siding Spring Observatory by Robert Smyth -- see Section II, below -- are flagged with an "s".) I did finally observe most of the listed galaxies north of the equator. However, because of time and weather constraints, I could not get to several of the large galaxies, particularly in the heavily-populated main belt of the Local Supercluster between 11h and 13h. During less hectic observing runs, I picked up 63 other galaxies in both northern and southern hemispheres needing photometry to derive total magnitudes and color indices, or needing data for programs of colleagues. I've listed all of these additional galaxies in Table 1b.
All of the galaxy photometry helped yeild total magnitudes and color indices for inclusion in RC3. See the Buta et al 1995a,b AJ papers for details of the RC3 data reduction. The RC3 Introduction has a summary of the 1995a paper.
I also observed eighteen Galactic globular clusters (Table 1c; see also Corwin AJ 82, 193, 1977 for earlier UBV observations for seven globulars and general comments, most still applicable here), again mainly northern objects. I am interested in the total magnitudes of these objects for helping to build a luminosity function of Galactic globulars for comparison with similar luminosity functions for globulars in other galaxies. Charles Peterson's (PASP 98, 192, 1986 and CDS catalogue number II/117) collection of UBVRI photometry of globular clusters includes most of the present data, and the data were also used in William Harris's (AJ 112, 1487, 1996) collection of parameters for Galactic globular clusters. They also made a contribution to Brian Skiff's (QJ Webb Soc. No. 99, 1995 and The NGC/IC Project web site, 1999) globular cluster database.
Among the stars I picked up are hundreds needed for photometric sequences around several of the galaxies, particularly those having hosted supernovae (see e.g. de Vaucouleurs and Corwin, ApJ 295, 287, 1985; and de Vaucouleurs, Corwin, and Skiff PASP 106, 156, 1994) or around those in which I also observed the SNe themselves (see e.g. Corwin, IAU Circular No. 3794, 1983; Buta, Corwin, and Opal PASP 97, 229, 1985). I also observed stars around many other nearby galaxies represented in Thompson and Bryan's "Supernova Search Charts and Handbook" (Cambridge Univ. Press, 1990). Marian Frueh and Shireen Gonzaga also observed many stars for supernova sequences at McDonald; I present here their full UBV data for the first time (see Section IX-B below), though most of their V magnitudes are available on the Thompson/Bryan charts.
Many stars are superposed on the galaxies "contaminating" the aperture photometry. I subtracted their light from the galaxy photometry to yeild a closer representation of the galaxies' uncontaminated light. These stars may also be useful as additional comparison objects for supernovae, but many of them are on the bright galaxy backgrounds which makes their photometry difficult and uncertain. I recommend caution when using these superposed stars for any sort of differential photometry.
I also observed many additional "standard" stars filling in color or magnitude gaps in the list of equatorial standards that I used (see Section V, below). Finally, I observed several other stars and BL Lac objects at the request of colleagues at McDonald for their own programs (see e.g. J. Wray's "Color Atlas of Galaxies", CUP, 1988).
Dr. Robert Smyth made observations of most of the southern galaxies at Siding Spring Observatory with their 41-cm (primarily) and 61-cm telescopes in 1980 and 1981. His observations are given in the UBV and longer wavelength photometry collections of Guiseppe Longo and Antoinette de Vaucouleurs (1983, 1985, and 1988, Univ. of Texas Monographs in Astronomy, Numbers 3, 5, and 6). Because those collections are now difficult to find, I've included Robert's results here. I discuss these further in Section IX, below. I also obtained UBVRI data at McDonald for several stars superposed on galaxies that Robert observed; he was unable to measure the stars themselves because of time or weather constraints, or because the stars were simply too faint for the smaller telescopes he used.
Robert's data are also included (as are many of my data from the 1982 through 1987 runs) in the Prugniel and Heraudeau collection of 1998 (see A&AS, 128, 299, 1998) now available at CDS; this is an updating of the Longo and de Vaucouleurs lists. If there are differences between those lists and the ones I provide here, the ones here are to be preferred.
Robert also set up a belt of secondary UBVRI standard stars around the equator -- primarily in the Harvard equatorial Selected Areas -- which I refined into my own list of "standards". Robert tied about 100 equatorial stars to the Cousins standards in the -45 degree Harvard E-regions during his observing runs. I used these as one of the core data sets for the establishment of my own standards; I give the details in Section V.
My goal was to do a multi-color photometric survey, extending across the optical spectrum, of the 250+- largest galaxies, using the largest apertures readily available. The general service photometers at McDonald Observatory had one filter box ("No. 4") that had an aperture wheel that included a 12.6-mm aperture. At the plate scale of the 76-cm telescope (my preferred telescope; more below), this corresponded to 4.0 arcminutes (the other apertures used are listed below). While not as large as I would have liked, this aperture was large enough to include much of the light of a five-arcminute galaxy, and still small enough that it could be used with the general service photometers without modification of the equipment on the 76- and 91-cm reflectors at McDonald.
The apertures available in Filter Box No. 4 were:
Apertures 76-cm telescope 91-cm telescope Nominal Measured* (19.1"/mm) logA (16.8"/mm) logA mm mm arcsec (0.1') arcsec (0.1') 0.5 0.499 9.53 0.20 8.38 0.15 1.0 0.977 18.66 0.49 16.41 0.44 2.0 1.979 37.80 0.80 33.25 0.74 4.0 3.988 76.17 1.10 67.00 1.05 8.0 8.035 153.5 1.41 135.0 1.35 12.6 12.59 240.4 1.60 211.4 1.55 *Measured by W. Pence and B. Smith.
Photomultiplier tubes also existed at that time with extended red response that could cover the entire optical spectrum from about 3000A to 9000A. A commonly available tube was the EMI 9658R with an "end-on" configuration; at least two of these were available for general use at McDonald Observatory. These tubes have a reputation of being rather "noisy", even when cooled with dry ice. So, after some experimentation with them, and consultation with Tom Moffett and Tom Barnes -- experienced users of the equipment -- I adopted the strategy of using a somewhat lower voltage (+1200V to +1350V) than normal (+1500V to +1600V) for all but the first run (January 1982). I also had a "magnetic defocusing assembly" (a ring magnet with a large center opening) placed over the front of the tube for all observations from Run 5 (February 1983) on. These precautions had the effect of lowering the total signal by a factor of about 3.5, but the dark counts by a factor of about 16. Total dark counts using lower voltages and the ring magnet with the "L2" tube were typically 300, and with the "L4" tube around 500.
The ring magnets could have the additional effect of slightly reducing the effective size of the apertures. I checked for this at the beginning of each run by drift scans of moderately bright stars of high declination across the largest aperture. With the exception of the first run, all these drift scans showed acceptably "square" profiles with "flat" tops. The first run exception turned out to be due to the presence of a second(!) Fabry lens in the light path. Once that was removed after the first two nights, the aperture "flattened out" acceptably. I did not use the largest aperture on either of those nights in any case (the first night was mostly cloudy with no galaxy observations, an unfortunately common occurance during these runs).
The photomultiplier tubes were mounted at the end of a typical optical train that included (in addition to the telescope itself) an offset guider (the one I used for the entire program was labeled "Offset Guider No. 2") for target acquisition. In addition to the movable wide-field offset-guiding eyepiece, the guider included the aperture wheel in the focal plane that could be easily examined with another eyepiece mounted in a periscope -- this doubled as a "shutter" for the photomultiplier tube. After the first two nights, a (single!) Fabry lens was mounted at the front of a cold box containing the filters and the dry-ice cooled tube.
The filter wheel was driven by a computer-controlled stepping motor so that exactly-timed 10-second integrations could be made through the different filters automatically. A "standard source" of carbon-14 was also mounted in the filter wheel to monitor the stability of the photometer. The stepping motors, and all other aspects of data aquisition, were controlled by mini computers running programs written by UT Austin Astronomy Department programmer Sam Odoms; I'm grateful for his help over the several years of the observing program.
I used standard Schott filters (see e.g. Bessell, PASP 91, 589, 1979):
|U:||1mm UG2 (+ CuS04 for red leak suppression)|
|B:||1mm BG12 + 1mm BG18 + 2mm GG385|
|V:||1mm BG18 + 2mm GG495|
|R:||2mm KG3 + 2mm OG 550 (Runs 1-12)|
|2mm KG3 + 3mm OG 570 (Runs 13-24)|
The copper sulfate red leak filter was a solid transparent block for most runs. Depending on its availability, however, a liquid solution in a "bottle" of optical glass was occasionally substituted. The exact shape of the I-band depended on the long-wavelength cutoff of the photomultiplier itself rather than on a filter. That this was an acceptable strategy was shown by the not-unreasonable transformation coefficients (see Section VII below). Bessell (PASP 88, 557, 1976; and PASP 102, 1181, 1990) shows transmission curves for a nominally-identical filter set.
Through a misunderstanding on my part, an incorrect blue-cutoff OG 550 filter was used for the R-band the first half of the program. However, reductions showed that the data are uncompromised: the transformations are still linear, though with a steeper slope than usual (again, see Section VII below).
I used the 76-cm reflector at McDonald as my primary telescope, though I did use the 91-cm for two runs in 1985. I preferred the 76-cm reflector for its shorter focal length (giving a larger plate scale), wider flat field, better internal light baffling, and greater ease of obtaining observing time. It is, however, a longer walk from the TQ (Transient Quarters, McDonald's "hotel" for visiting astronomers), so I had occasional encounters with the local wildlife (deer, skunks, peccaries, rattlesnakes, buzzards, house cats, other astronomers, etc).
I followed observing procedures developed over the decades at various observatories, and refined by de Vaucouleurs (see e.g. de Vaucouleurs and de Vaucouleurs, MemRAS 77, 1, 1972; and de Vaucouleurs, de Vaucouleurs, and Corwin, AJ 83, 1331, 1978). The summary I gave in Corwin (MNRAS 191, 1, 1980) is repeated here with the variations needed to accommodate the different location and equipment.
I developed a set of standard stars in the equatorial Selected Areas (see Section V below); those had a small enough air mass (~1.15) when on the meridian that I did not have to observe standards closer to the zenith. I also observed the equatorial standards 3-4 hours east or west of the meridian (air mass > 1.5) so that I could calculate nightly extinction coefficients. In general, I picked up at least one "red" and one "blue" star at low and high air mass; I often accompanied those with one or two stars of intermediate color. I observed standards three times a night (for nights that were photometric all night; as I noted above, this was unfortunately a rare occurance: my overall average of photometric weather was just 30% of the alloted observing time) -- just after evening twilight ended, at midnight, and just before dawn twilight began.
Each observation of a standard consisted of three groups of ten-second UBVRI integrations (plus a one-second integration of the carbon-14 "standard source") on the star itself, interspersed with two on the night sky north and south of the star, again with the standard source. I observed stars fainter than about V = 10 in five groups of ten-second integrations, interspersed with four sky fields in the cardinal directions. I checked sky fields at the eyepiece for significant intruding stars and moved the telescope to an empty field if the nominal sky field had interlopers. Similarly, a few stars had companions that I edged out of the aperture (though three companions were close enough that I either had to include them or measure them separately), though I could center most of the stardards without significant contamination.
I used the 2-mm (37.8-arcsecond) aperture for all standard stars. Stars that I observed at smaller apertures (1-mm/18.9-arcsec and 0.5-mm/9.5-arcsec) were corrected after reduction for scattering losses of 0.01 and 0.04 magnitudes, respectively. Color indices were not significantly affected by the aperture size.
For the program objects, whether galaxies, globulars, or stars, I typically used a set of five object-plus-sky UBVRI (plus the standard source) integrations, with four sky-alone UBVRI integrations in the cardinal directions interspersed (always with the standard source included). A few faint objects merited further integrations, sometimes with as many as nine object-plus-sky and eight sky-alone. As with standard stars, I checked the sky fields for stars or other intruding objects, and moved the telescope as necessary. Once in a while, I purposely slightly miscentered an aperture on a galaxy to exclude a star that would otherwise have intruded on the measurement. I detail all such observing irregularities in the notes appended to the reduced data (see Table 6).
I measured stars brighter than about V = 15.5 superposed on the galaxies separately, usually with the 1-mm aperture, and usually three integrations on the star, interspersed with two on the galaxy background, chosen by eye to be close to the same isophote as the superposed star. (Seeing was occasionally bad enough that faint stars could not be accurately centered in the smallest aperture. So, I rarely used that aperture except on nights of good seeing or for stars on backgrounds with steep light gradients.) I attempted to measure all stars down to about V ~ 15.5, but in a few cases of low-latitude galaxies (e.g. NGC 147, IC 342), there were too many superposed stars to measure in the time available. For these galaxies, I attempted to choose sky fields that included roughly as many stars of the same brightness as those superposed on the galaxy.
I observed the globular clusters in the same way as the galaxies, and I took the same precautions with field stars for the low-latitude globulars as I did with the galaxies. Two globulars, NGC 6934 and GCl 107, had superposed field stars that I measured and removed in the same manner as stars superposed on the galaxies.
The "standard" stars I used for the project are a set of 190 stars around the celestial equator, most in the Selected Areas, on the Johnson-Cousins UBVRI system. I had to establish the standards through my own observations at McDonald. There were several reasons for this: 1) I was using a photomultiplier with a long-wavelength cutoff of about 9000 Angstroms and with filters suitable for the Cousins R and I bands. Johnson's longer-wavelength R and I bands extended well past 1 micron; these long wavelengths could not be adequately accessed with the filters and photomultiplier I used. 2) Landolt's (RI)C standards in the Selected Areas along the equator (Landolt AJ 88, 439, 1983) had not yet been published. Finally, 3) there were therefore few Cousins RI system northern hemisphere stars accessible when I began the observations in 1982 that could be reasonably used as standards.
So, when Robert Smyth volunteered to devote his post-doctoral telescope time at Siding Spring observing the southern objects in the large-galaxy list, I asked if he could also help transfer the Cousins system to the equator. Fortunately, he was not only enthusiastic about the idea, but informed me that Mike Bessell at Siding Spring had already been working on that project. In addition, through correspondence with Alan Cousins and John Menzies, I found that they and their colleagues at the South African Astronomical Observatory (SAAO) had also set up a preliminary belt of standards around the celestial equator in the Harvard Selected Areas.
The stars in both lists were a subset of the stars that Landolt had published as UBV equatorial standards in 1973. The data, however, were tied to the fundamental Cousins E-region UBVRI standards at -45 degrees. All these observers generously shared these lists of preliminary standards with me, and I am extremely grateful for their help. The McDonald program would not have been possible in its present form without their work.
Robert Smyth also observed 17 of the Neckel and Chini (1980, A&AS 39, 411, 1980) secondary standards with the 61-cm Siding Spring reflector, but eventually discarded the observations. The nights were apparently unstable and highly sensitive to changes in seeing, sky transparency, and -- almost certainly -- instrumental stability, a persistent problem he faced, particularly on the 61-cm telescope. Robert tried discarding obviously bad observations and adjusting statistical weights of both the standard and the Neckel and Chini stars, but ultimately rejected his observations of all 17 of these stars. Thus, they unfortunately played no part in the establishment of the magnitudes and color indices of the standards that I used.
Two sets of (UBVRI)J equatorial standards in Johnson's system had at the time been recently published by Kunkel and Rydgren (AJ 84, 633, 1979) and Moffett and Barnes (all 1979: AJ 84, 627; PASP 91, 180 and 289). Through the stars in common with Smyth and Bessell, I converted the Johnson R and I to Cousins R and I, and used them to help strengthen the preliminary system I was developing. Consultation with Tom Moffett and Tom Barnes at McDonald early in the program was as valuable in this regard as it was in configuring the equipment (Section III above).
With these preliminary standards in place, I wrapped the system on itself around the sky several times with the McDonald equipment as I continued observing over the next six years. During that time, I did several interim reductions to refine the system, and I added new stars in several Selected Areas lacking sufficient numbers of red or blue stars.
After the observing was finished, I compared the resulting UBVRI system with those published by Landolt (AJ 88, 439, 1983) and by Menzies et al (MNRAS 248, 642, 1991, a refined version of the SAAO list I had used). Systematic differences are less than 0.01 magnitudes in V and in all colors, with standard deviations in those differences also less than 0.01 magnitudes, except in the U-B color index where the standard deviation approaches 0.02 magnitudes. Table 2 has the statisitics, and Figures 1-5 show the magnitude and color index comparisons. The figures are:
|Figure 1a:||V magnitudes, McDonald vs. Landolt|
|Figure 1b:||V magnitudes, McDonald vs. SAAO|
|Figure 2a:||B-V color indices, McDonald vs. Landolt|
|Figure 2b:||B-V color indices, McDonald vs. SAAO|
|Figure 3a:||U-B color indices, McDonald vs. Landolt|
|Figure 3b:||U-B color indices, McDonald vs. SAAO|
|Figure 4a:||V-R color indices, McDonald vs. Landolt|
|Figure 4b:||V-R color indices, McDonald vs. SAAO|
|Figure 5a:||V-I color indices, McDonald vs. Landolt|
|Figure 5b:||V-I color indices, McDonald vs. SAAO|
There are almost certainly the usual correlations (see e.g. Bessell PASP 102, 1181, 1990; and PASP 107, 672, 1995) among the differences with magnitude, color index, air mass, and perhaps right ascension (systematic differences from one equatorial Selected Area to another). Preliminary comparisons suggest these are small effects -- typically less than 0.02 magnitudes -- mostly occuring at the extremes of the ranges of magnitude and color index. I have little doubt that these correlations are present in the data at some level. As time allows over the next [interval TBD!], I'll explore these potential sources of systematic error.
From the Table 2 statistics, I concluded that the "standard" star system I built at McDonald is more than adequate for its intended purpose of calibrating the galaxy and globular cluster photometry. In particular, the range of color indices covered (B-V ~ -0.15 to B-V ~ 1.7) allows interpolation of the normal integrated color indices of nearby galaxies (B-V ~ 0.4 to B-V ~ 1.1); no extrapolation in color index is necessary. Most "normal" stars without significant emission, and with color indices in the same range, can also be handled well by this set of "standard" stars.
Nevertheless, I would recommend that UBVRI observers adopt one or another of the standard systems by Landolt or by Cousins, Menzies, and their colleagues. Landolt (AJ 104, 340, 1992 and AJ 137, 4186, 2009) has extended his system to fainter magnitudes and smaller fields suitable for CCD photometry. The Moffett and Barnes and Kunkel and Rydgren equatorial standards are similarly suitable for use with Johnson R- and I-system filters and photomultiplier tubes.
Table 3 lists my McDonald standard stars with J2000.0 positions and the adopted magnitude and color indices. I've generally adopted the primary name that appears in the published lists of standard stars (Selected Area numbers, and HD and BD numbers). A secondary name (HD or BD number) is also given where appropriate as a check. I collected J2000 positions from SIMBAD (March 2018); these come primarily from Hipparcos, Tycho 2, Gaia DR1, UCAC, or 2MASS PSC. Proper motions are now available in Gaia DR2, or with lower (but still adequate!) accuracy in Hipparcos, Tycho 2, UCAC, and Gaia DR1 (for Tycho 2 stars).
I inadvertantly included three stars as "standards" with nearby companions -- I discovered these at the eyepiece. I've shown data for the stars with and without the companions included in the apertures. In one case, HD 75138, I also give data for the companion alone. I do not recommend relying on any of these three stars as a standard unless the aperture you use is small enough to exclude the companion stars.
Finally, some of these "standard" stars are suggested as variable in the literature and noted as such by SIMBAD. While I've also noted this at the end of the table, my McDonald observations do not confirm variability for any of these stars. Nevertheless, I'd recommend not using these as standards even though they may appear in the other lists as well as this one.
I reduced the observations following the standard procedures set out by e.g. Robert Hardie, "Photoelectric Reductions" (page 178 in "Astronomical Techniques", ed. W.A. Hiltner, Univ. of Chicago Press, 1964; see also Hardie, ApJ 130, 663, 1959), though with some variations.
I stepped the data in sequence through four Fortran programs that I wrote to handle the reduction:
I performed the reduction at least twice for every observing run with preliminary UBVRI data for the standard stars, examining plots of the data to check for discordant stars. After discordant stars, or discordant counts leading to large discrepancies, were discarded and with stable standard star data in hand, I performed a final reduction of all nights, using nightly transformation as well as nightly extinction coefficients. The results of these final reduction runs are those that I report here.
Note that whenever the weather allowed it, I always calculated and used nightly extinction coefficients, including second-order coefficients (though those were always small). I had to use mean extinction coefficients on only three nights of the total of 85 on which I observed. Using mean extinction coefficients was the normal practice of some observers at McDonald, but the observatory in the 1980s did not prove to be a very stable photometric site. In addition to the El Chichon eruptions in 1982, summer dust and thunder storms, and winter rain and snow storms all made the use of mean extinction coefficients, in my opinion, untenable. A strong El Nino in 1983 also made for unstable weather for at least a year. As noted earlier, only 30 percent of the time allocated for the program was photometric. Note also that the extinction coefficients were still elevated beyond their normal values at McDonald (kv ~ 0.15 - 0.20) for a year following the El Chichon eruptions (see Table 4).
Also, the reduction procedure I adopted was such that the nightly zero point was absorbed by the extinction calculation. The extinction reductions showed small but significant differences in the zero point from one night to another within a run, even though the instrumental setup was almost always unchanged from one night to the next.
Finally, because of the ease of running the four-step reduction procedure in its complete state, I also used nightly transformation coefficients. However, they were stable enough within a run that I could have adopted mean coefficients with no loss of accuracy. Only one night had observations for which I needed to use true mean coefficients for the final transformation to standard UBVRI data (see Table 5).
The extinction and transformation coefficients are collected in Table 4 and Table 5 , respectively. The tables are briefly explained in the headers in each table, but those explanations will refer you back to the next section for details.
The "dead time" or "coincidence" correction is nominally due to the overlapping arrival of photons at the phototube. We usually assume this to be a linear correction of the form
where N is the true count, n the observed count, and tau the dead time in seconds. The dead time is almost always a problem in the electronics of a photometer system rather than in the photomultiplier itself. For the system I used at McDonald, the dead time was found by the mountain staff to be 50 +- 25 nanoseconds. Thus, N can be determined with errors of less than 1% for all count rates less than about 3 MHz. For 7th magnitude stars at the 76-cm telescope, the count rate was 0.1 MHz with the ring magnets in place, so the correction was negligible for virtually all the observations aside from the very brightest stars and a few bright globular clusters. Nevertheless, I applied the correction to all the observations. (Fernie in PASP 88, 969, 1976, has words of wisdom if you need to deal with higher count rates.)
I approximated the air mass X of an object through
where lat is the observer's latitude (+30d 40m 18s for McDonald), Dec is the object's declination, and HA is its hour angle (sidereal time minus right ascension). See Andrew Young's article in "Methods of Experimental Physics" Vol. 12A (p. 123, ed. Carleton, Academic Press, 1974) for background on this approximation. Other approximations are available (see e.g. Hardie's "Photoelectric Reductions" article cited above).
I calculated these as standard Pogson magnitudes:
ci(c2 - c1)= 2.5*log[(cd)c1 - (cd)c2]
where v is the instrumental magnitude in the V band, (cd)v is the dead-time corrected number of counts through the v filter; ci(c2-c1) is the instrumental color index for filters "c1" and "c2", and (cd)c1 and (cd)c2 are the dead-time corrected counts through those filters.
I used standard equations of the form
where v0 and ci0 are the extinction-corrected magnitude and color indices in the instrumental system; k[v|ci]i, k[v|ci]s, and k[v|ci]2 are the extinction coefficients (zero point, slope, and second-order, respectively) for the instrumental v-magnitude and the color indices ci; X is the air mass; and
is the simple arithmetic mean of the instrumental color index and its value predicted through preliminary hand solutions.
Usual photometric reductions treat the second-order coefficients as identically equal to zero, except for kbv2 ~ -0.033. Nevertheless, I always solved for them as they sometimes proved to be significant, though always small.
I found solutions for each night (sometimes for parts of nights) using a standard least-squares program written many years ago by P. Jordahl and C. Michaelis (both then at McDonald Observatory), and revised by me for the requirements of the observing program and the various generations of Fortran compilers that I've used over the years.
Most photometric reductions assume linear transformation equations between the instrumental system and the standard system:
where a[v|ci]i and a[v|ci]s are the transformation coefficients for the magnitude and color indices; v0, (b-v)0, and ci0 are the extinction-corrected instrumental magnitude, (b-v) color index, and color indices. The upper case letters denote the standard UBVRI magnitude and color indices. I found solutions using the same least-squares program that I used for the extinction calculations.
As mentioned above, Robert Smyth (a fellow post-graduate student at the University of Edinburgh; he received his doctorate in 1980 a year before I did) did the southern observing during his 1980-81 post-doc at Siding Spring in New South Wales, Australia. He used the (now decommissioned) 41-cm and 61-cm telescopes there with the standard photometers and the same observational techniques I used at McDonald. As mentioned above, his UBVRI system is that of Cousins and is based on observations of standard stars in the Harvard "E-regions" at -45 degrees.
Robert had to deal with unusually bad weather and with balky equipment -- particularly on the 61-cm telescope -- at Siding Spring. The electronics in particular led to relatively unstable counts throughout many nights resulting in fluctuating zero points on the order of 0.05 magnitudes. The reduced color indices were apparently not much affected by this, but the V magnitudes were. Robert has clearly flagged the affected data in the final table of his results.
Robert used the filters listed above (exclusively the OG 570 R-band blue cutoff filter), though with three different EMI phototubes: two 9658R tubes, with one ambiguously-identified tube substituted on two nights when the first 9658R "died" unexpectedly. These were mounted in photometers (with aperture viewers) similar to the McDonald equipment. The apertures in these photometers were
41-cm telescope 61-cm telescope Ap log A Ap log A arcsec 0.1' arcsec 0.1' ( 15 0.39 ) ( 10 0.23 ) ( 28 0.67 ) ( 19 0.51 ) ( 42 0.84 ) ( 29 0.68 ) 55 0.96 38 0.80 88 1.17 61 1.01 134 1.35 92 1.19 203 1.53 139 1.37 264 1.64 182 1.48
There are aperture size and telescope scale disagreements (on the order of 5%) among the three lists of apertures that I have from Robert. The numbers that I have finally adopted in the above table are a compromise from the last of those lists, and from the "log A" apertures that Robert used in his final table of data (aside from superposed stars, he did not use the smallest three apertures for any observations, so those numbers are from his last aperture list). There is a single exception: one observation of NGC 1531 is listed at log A = 1.16. I assume it is -0.01 off in the log -- that should be corrected if you use the data for that galaxy.
The various observations I made at McDonald Observatory are gathered in Tables 6 to 10 (links below). The general format for Tables 6-9 is
|Column 1:||Object name|
|Column 2:||Base-10 logarithm of the aperture in 1/10ths of arcminutes|
|Column 3:||V magnitude|
|Column 4:||B-V color index|
|Column 5:||U-B color index|
|Column 6:||V-R color index|
|Column 7:||V-I color index|
|Column 8:||Calendar date of observation (UT date minus 1 day), yy-mm-dd|
|Column 9:||References to Notes|
Though I give the photometric data to three decimal places, the last place is not significant for the galaxies and globulars, nor for the fainter stars (V >~ 10). I keep that last digit as a round-off guard. The standard star data, and the data for stars brighter than about V = 10, however, can be significant in the third decimal place, so those should be taken as I've presented them.
The individual tables are
|Table 6:||Galaxies with all measured superposed stars subtracted.|
|Table 7:||Galaxies with superposed stars included.|
|Table 8:||Stars and miscellaneous objects: stars superposed on, or near,|
|galaxies, including local comparison sequences; miscellaneous|
|stars; NGC 206 (M31 star cloud); HII regions; SNe; and so forth.|
|The positions are in Table 10.|
|Table 9:||Globular clusters (superposed stars are in Table 8).|
|Table 10:||Positions for all the stars (except the standard stars; see|
|Table 3 for those), and for the miscellaneous objects.|
Many of the positions for the stars were collected by Brian Skiff from various online catalogues; I'm grateful for Brian's help.
As a part of a larger program of galaxy photometry, Marian Frueh -- ably assisted on two nights by Shireen Gonzaga -- collected UBV data for stars around 23 bright galaxies. The data are intended as sequences for local comparisons with supernovae (coincidentally, two supernovae occured during the course of Marian's observations: SN 1986A in NGC 3367 and SN 1986G in NGC 5128. Marian also observed these SNe, of course).
We published the galaxy photometry (Frueh, Corwin, de Vaucouleurs, and Buta; AJ 111, 722, 1996) but not the sequence stars. Many of those stars' V magnitudes appeared on the Thompson/Bryan "Supernova Search Charts", but the entirety of the data appear here for the first time.
The paper with the galaxy photometry outlines the observing procedure; it is essentially identical to that that I used, described in Section IV above. Marian and Shireen used Landolt (1983) UBV standards, so the data are on the standard system. The standard deviations are expected to be similar to -- or perhaps even better than -- those I derived for the UBVRI data, as Marian and Shireen made the observations with the 76- and 91-cm telescopes and a photomultiplier (Amperex 56-DVP) with a less noisy output. The data are given in Table 11a (UBV data) and Table 11b (positions). Note that the dates in Table 11a are UT dates, not calendar dates as I've used for my own observations.
The observations of the stars around NGC 5128 deserve special mention as the field is only 16.5 degrees above McDonald's horizon when the galaxy is on the meridian. This means that the extinction corrections have to be extrapolated to air masses of approximately 3.5, making the absolute photometry quite uncertain. Marian also commented in her log that the "seeing looks really terrible down here" on 9 May 1986 when she estimated the seeing at the zenith to be 4-7 arcseconds. The seeing was probably not much better on the other nights that she observed SN 1986G.
Because of these problems, Brian Skiff suggested that we use the Tycho photometry of the brightest comparison star (A = HD 116647) as a zero point for B and V. At V = 9.14 +- 0.02 and B-V = 0.35 +- 0.02, the Tycho data agree to within the errors with the V magnitude found by the ASAS project, 9.167 +- 0.035. Making these zero-point corrections led to the revised BV data for the NGC 5128 supernova and comparison stars that I list in Table 11a. I've flagged these with colons as they are still obviously quite uncertain (especially the uncorrected U-B color indices), and I also include the uncorrected data when better photometry becomes available for the comparison stars.
At the moment, Robert Smyth's full observations are available only as a PDF file of his original typed table, circulated privately. It is shown here as Table 12. I have compared this with a preliminary hand-written version, again circulated privately, and have made a few additions and corrections to the typed table in my own hand. I have also added asterisks to the date if the observation was made with the Siding Spring 61-cm telescope rather than the 41-cm used for the bulk of the observations. The format is
|Column 1:||Object name|
|Column 2:||Numerical galaxy type (see RC2)|
|Column 3:||Base-10 logarithm of the aperture in 1/10s of arcminutes|
|Column 4:||V magnitude|
|Column 5:||B-V color index|
|Column 6:||U-B color index|
|Column 7:||V-R color index|
|Column 8:||V-I color index|
|Column 9:||Calendar date of observation (UT at Siding Spring), yy-mm-dd;|
|"*" indicates 61-cm observation.|
|Column 10:||References to superposed stars and to the Notes|
|Figure 6a:||U magnitudes|
|Figure 6b:||B magnitudes|
|Figure 6c:||V magnitudes|
|Figure 6d:||R magnitudes|
|Figure 6e:||I magnitudes|
The exponential fits shown are those calculated by the graphing program (DataGraph, version 4.3), and are listed in Table 14 . It's clear that this is an adequate representation of the internal errors in the galaxy photometry.
The only points rejected from Figures 7 through 10 are associated with the 4.0-arcminute aperture I-band measurements for NGC 3365. Note, too, that I have plotted ALL the points associated with the repeated measures. This accounts for the plethora of apparent double (and occasionally triple) points seen in the plots.
Stars and galaxies obviously have different observed radial luminosity profiles. Those for stars are essentially unresolved point sources convolved with the optics of the telescope and any scattering in the light path between the star and the photocathode in the photometer. Galaxy profiles are affected by the same optics and scattering, but of course begin as an areally extended image projected on the sky. So galaxies will usually have a lower surface brightness in the focal plane than a star at the same measured magnitude.
This will be manifest in the signal-to-noise ratios of the objects, and is clearly demonstrated by the plots of log (S/N) versus magnitude shown in Figure 7. The figures are:
|Figure 7a:||U magnitudes|
|Figure 7b:||B magnitudes|
|Figure 7c:||V magnitudes|
|Figure 7d:||R magnitudes|
|Figure 7e:||I magnitudes|
Two "sequences" are obvious, stars and galaxies, so I've separated them in the error analysis that follows. The only galaxy data that I've included in the star sequence are the smaller apertures for NGC 3315 (one of the brighter galaxies observed in this program), while the only stellar data in the galaxy sequence is star "10" superposed on the bulge of M31.
Calculating the signal-to-noise ratio depends on having access to the mean counts, not just the resulting reduced magnitudes. This makes it unnessarily difficult to find the statistical errors for any given observation as the counts themselves are not available beyond my raw data lists. So, I looked at parameters similar to signal-to-noise that would be available for all the REDUCED data which obviously are readily available. The mean surface brightness (already mentioned above) of the object plus sky within the aperture could well be such a number (again, see e.g. Buta et al, 1995a), so I plotted that versus signal-to-noise -- those plots, which include the stars with multiple observations as well as the galaxies, are in Figure 8. The figures are:
|Figure 8a:||U magnitudes|
|Figure 8b:||B magnitudes|
|Figure 8c:||V magnitudes|
|Figure 8d:||R magnitudes|
|Figure 8e:||I magnitudes|
These show that the correlation between surface brightness and log (S/N) is tight, and that surface brightness can be an adequate substitute for signal- to-noise. There is some evidence of "structure" in these plots, but I have assumed it to have little effect on this error analysis.
The standard deviation versus surface brightness plots are shown in Figure 9 along with the exponential fits from DataGraph, also listed in Table 14. The figures are:
|Figure 9a:||U magnitudes|
|Figure 9b:||B magnitudes|
|Figure 9c:||V magnitudes|
|Figure 9d:||R magnitudes|
|Figure 9e:||I magnitudes|
These lead to the calculated internal errors versus surface brightness listed in Table 15.
Errors in photometric observations of astronomical objects are more commonly related simply to magnitude. To check that possibility, I plotted the standard deviations in the repeated observations versus magnitude; the results are shown in Figure 10. The figures are:
|Figure 10a:||U magnitudes|
|Figure 10b:||B magnitudes|
|Figure 10c:||V magnitudes|
|Figure 10d:||R magnitudes|
|Figure 10e:||I magnitudes|
Because stars are essentially point sources, their "surface brightnesses" can be reasonably modelled by a "point-spread function". Because the aperture corrections I applied adequately address that problem, reducing all the star observations to the "standard" aperture (A = 38", log A = 0.80), I have simply plotted the standard deviations against magnitudes to determine the errors in the stars' magnitudes. The plots are shown in Figure 11. The figures are:
|Figure 11a:||U magnitudes|
|Figure 11b:||B magnitudes|
|Figure 11c:||V magnitudes|
|Figure 11d:||R magnitudes|
|Figure 11e:||I magnitudes|
Data rejected from these plots are:
|U:||IC 342 *K (sigma1 = 0.743)|
|B:||NGC 6946 *d (sigma1 = 0.777)|
|V:||NGC 6946 *d (sigma1 = 0.349), NGC 4615 *A (sigma1 = 0.266)|
|I:||NGC 6946 *d (sigma1 = 0.649)|
Other relatively discordant, but unrejected, points are labeled in the plots.
Stars superposed on the galaxies will be affected by the varying background of the galaxies themselves. That effect has undoubtedly led to the increased scatter that is seen in the Figure 11 plots, especially for the fainter stars.
I addressed the errors in the color indices in much the same way I did the magnitudes. I've simply plotted the standard deviations in the color indices versus different parameters. The plots of sigma1 versus V magnitude are shown in Figure 12; the figures are:
|Figure 12a:||Sigma1 (B-V) vs. V Magnitude|
|Figure 12b:||Sigma1 (U-B) vs. V Magnitude|
|Figure 12c:||Sigma1 (V-R) vs. V Magnitude|
|Figure 12d:||Sigma1 (V-I) vs. V Magnitude|
Similarly for surface brightness:
|Figure 13a:||Sigma1 (B-V) vs. V Surface Brightness|
|Figure 13b:||Sigma1 (U-B) vs. V Surface Brightness|
|Figure 13c:||Sigma1 (V-R) vs. V Surface Brightness|
|Figure 13d:||Sigma1 (V-I) vs. V Surface Brightness|
|U-B:||UGC 09749, log A = 1.60 (sigma1 = 0.340)|
Sigma1 (C.I.) vs. V surface brightness:
|U-B:||GCl 107, log A = 1.41 (sigma1 = 0.308)|
Outstanding data points identified in the plots have not been rejected from the exponential fits.
As I did with the magnitudes, I've chosen to treat the stars as simple point-spread functions in examining the errors in their color indices. So, I plot the standard deviations in their color indices simply as functions of V magnitude. These are shown in Figure 14:
|Figure 14a:||Sigma1 (B-V) vs. V Magnitude|
|Figure 14b:||Sigma1 (U-B) vs. V Magnitude|
|Figure 14c:||Sigma1 (V-R) vs. V Magnitude|
|Figure 14d:||Sigma1 (V-I) vs. V Magnitude|
|B-V:||NGC 6946 *d (sigma1 = 0.428)|
|U-B:||IC 342 *K (sigma1 = 0.774)|
|V-I:||NGC 3184 *B (sigma1 = 0.449)|
It's pretty obvious from the figures that the exponential fits are pretty general approximations -- step functions, especially for the surface brightness plots, would apparently be more appropriate. For now, I leave that to my readers and critics to work out if you really need to have a more representative estimate of the errors in the color indices.
Before I had done the analysis for the internal errors in the colors, I had introduced a zero-point shift of +0.02 magnitudes in the internal errors for the magnitudes. This seemed to me to better represent the true errors in the observations. However, while investigating the internal errors for the colors, I found that the standard deviations, for the brighter objects in particular, appear -- from these data -- to be very small, and are well-represented by the exponential fits described above. This should not have been a surprise as the errors in the standard stars in the same magnitude range are also very small. So, I decided to let the exponential error functions, as DataGraph reports them, act without modification; and removed the zero-point offset I had initially applied.
Comparison with observations from other observers will in any case give us a more accurate estimate of the true errors in these observations. This topic is discussed in the next section.
Buta et al (1995a) have included most of the UBV data reported here as sources COR-82, COR-84, and COR-87. Buta and Williams (AJ 109, 543, 1995) similarly include the VRI data from this work where the sources are listed as COR-84 and COR-93. Both papers provide a thorough error analysis of this McDonald galaxy data, comparing it to many published (and a few other unpublished) sources. I refer you to those two papers for the details on the comparisons.
Briefly, the current photometry -- once factors such as telescope size, aperture, and surface brightness are taken into account -- compares favorably with other "good" sources of UBVRI data for galaxies. The data can thus be used with some confidence for its intended purposes. Buta et al (1995a) and Buta and Williams (1995b) suggest the following mean errors in total magnitudes and colors derived from the current data:
Magnitude m.e. Color Index m.e. U 0.045 B-V 0.021 B 0.029 U-B 0.035 V 0.020 V-R 0.017 R 0.026 V-I 0.025 I 0.032
The mean magnitude errors for all but the B magnitude are calculated from the mean errors for the colors.
These are calculated from the mean values of the weights Buta et al calculate for the three sources (UBV) and two sources (VRI) to which they assign the current data, and from their mean errors corresponding to unit weight. So, while these are not directly comparable to the internal errors derived above, the approximate agreement is an indication of the general quality of the current data.
As I mentioned above, RC3 has total magnitudes and B-V and U-B colors for the galaxies, and Buta and Williams give V-R and V-I colors for galaxies with those observations in their analysis. The present data were included in both larger sets of data.
I must caution, however, that the internal errors become much larger at fainter magnitudes (the "step function" that I spoke of in the previous section). So, while the overall judgement of the data is that they are "acceptable", the data for the fainter galaxies, especially the those with lower surface brightnesses, must be used with considerable caution. Ron Buta took these factors into consideration in his work on the photometric data in RC3, so I can pretty confidently recommend the total magnitudes and colors, as well as the effective (half-light) colors, you find there. Similarly, I can recommend the V-R and V-I colors listed by Buta and Williams, should you find those useful for your work.
I have presented here three sets of Johnson/Cousins UBVRI and UBV data generally focussed on galaxies, or on stars near galaxies. I made the UBVRI observations of northern objects at McDonald Observatory in the Davis Mountains in west Texas, while Robert Smyth covered the southern objects at Siding Spring Observatory in New South Wales, Australia. The stellar data include supernova comparison sequences, many observed at McDonald by Marian Frueh and Shireen Gonzaga, as well as a set of secondary equatorial standard stars on the UBVRI Johnson/Cousins systems that I set up with help from Robert Smyth and Mike Bessell at Siding Spring, Alan Cousins and John Menzies at the SAAO, and Tom Moffett and Tom Barnes at McDonald.
I've given background information on the observations and reduction of the data, and have given at least preliminary estimates of the observational errors present in the data. The galaxy photometry may still be useful for calibration of CCD surface photometry, and the stellar magnitude/color sequences for determining at least preliminary values of supernovae or other objects in or near the galaxies. The set of secondary standard stars stands as an example of observational necessity given the lack of an accessible set of Cousins R and I standards in the northern hemisphere when I began observing in 1982. While it served adequately for the present program, I would not have taken the time to set it up had the accepted standards from Cousins and Landolt been generally available just a couple of years earlier than they were.
An observing program as big and sprawling as this one needs a lot of support from many people. I had that support from dozens of people whose names I recorded -- and from many others whose names have faded from memory over the years. My sincere apologies to all of you for not being able to include you by name here. This doesn't mean you didn't help, it just means that I was remiss in not making the detailed notes -- which I clearly need now! -- when I had the opportunity. Mea culpa -- mea maxima culpa!
In addition to the many astronomers mentioned in the text and tables here (whose names are engraved in the written record), I must also thank the McDonald Observatory Time Allocation Committee for their continued support of the program through generous amounts of dark time over the several years I observed. They also alloted similarly significant amounts of time for Marian Frueh's work on galaxies and stellar sequences. We were then, and are still, grateful.
The staff at McDonald was always helpful in details large and small while I was observing. Special thanks are due to at least Greg Henry, Dave Doss, Ed Dutchover, John Jordan, Tom Brown, and George Grubb for not only setting up the equipment prior to each run, but for occasional midnight heroics that kept the telescopes, domes, and other machinery in good shape. Sam Odoms saw to it that I always had a working data aquisition program for the computer that recorded the counts and spun the filter wheel. The several cooks and housekeepers at the TQ also took good care of their guests, doing their best to work within a restrictive budget to help keep us alert through the long nights. I would be remiss, too, if I did not acknowledge Marian Frueh and her "Movies at Marian's" cloudy-night entertainment. There were always several of us lined up at her door in bad weather looking for an hour or two of relief from the boredom of checking the sky once again, only to find it once again overcast, or fogged over, or raining, or thundering, or ...
In addition to Robert Smyth and Marian Frueh -- both of whom took an active part in the program -- Brian Skiff, James Bryan, Greg Thompson, Jim Wray, Alan Cousins, John Menzies, Mike Bessell, Rob Robinson, Don Winget, Rick Binzel, Ed Nather, Andris Lauberts, James Hilton, Ron Buta, Tom Barnes, Tom Moffett, Gerard and Antoinette de Vaucouleurs, Derek and Beverly Wills, Fritz Benedict, Andrew Young, Frank Bash, Steve Odewahn, Shyamal Mitra, Arlo Landolt, Guiseppe Longo, Chet Opal, Liz Bozyan, Shireen Gonzaga, Charles Peterson, William Harris, Greg Henry, and Mike Bode provided inspiration and observing opportunities for more than "just another galaxy" as one of them (unidentified here!) put it one day. Luis Campusano, visiting astronomer from Santiago, Chile, was a welcomed observing companion on two nights in November 1985; I hope he enjoyed his time at McDonald as much as I did.
After the observing was over, several people including Jim Schombert, Susan Gessner, Fred Chromey, and Lauren Jones -- as well as many in the preceding paragraph -- expressed an interest in larger or smaller pieces of the data. They pushed me to get some of it published, or at least sent off to them where it would do some good in their own programs.
Finally, my lovely wife Kathleen has been a stalwart companion through the several decades of this project, tolerating my obsession with galaxies with understanding and good humor. I cannot imagine how I could have done all this without her constant love and support.