Most of the observations presented in this paper were made during 80 scheduled observing nights at the Hale Observatories in the period 1975 March to 1976 April. The telescopes used were the 24-inch (61 cm), 60-inch (l.5 m), and 100-inch (2.5 m) reflectors on Mount Wilson, and the 200-inch (5 m) Hale reflector. The photometer employed had an offset guider and a star-sky chopper consisting of a rotating sector-wheel mirror near the focal plane. The measurements were made with an InSb detector cooled to 55° K. A similarly cooled iris diaphragm in the focal plane defined the aperture sizes. All of the filters were cooled to 77° K, and have effective wavelengths and full widths at half maxima as follows: 1.25, 0.24 µm; 1.65, 0.30 µm; 2.20, 0.40 µm; 2.20, 0.11 µm; 2.36, 0.08 µm. The narrow-band 2.20 and 2.36 µm filters define the strength of the 2.3 µm CO absorption band. The fact that these two filters were employed "cold", whereas previously they were mounted outside the dewar (Frogel et al. 1975b) loused a shift in the effective wavelength with a consequent decrease in the measured strength of the CO band in a late-type giant relative to that of Lyrae. This decrease arises almostentirely from the location of the 2.36 µ filter with respect to the CO absorption band which sets in very abruptly at 2.29 µm . 1 It has been found that the relative CO index is insensitive to small changes in the effective wavelength and bandwidth of the 2.20 µm filter (the continuum filter).
Supplemental infrared observations reported in this paper were made on the 0.9 m, 1.3 m, and 2.1 m telescopes of Kitt Peak National Observatory (KPNO), the 40-inch (1 m) telescope of Las Campanas Observatory, Chile, and the 60-inch (1.5 m) telescopes of Cerro Tololo Inter-American Observatory (CTIO) and of the Smithsonian Astrophysical Observatory on Mount Hopkins, Arizona. These observations were made with two additional and completely independent sets of photometers and filters. Transformations which relate the J - H and H - K colors on the different systems are discussed in the Appendix.
1 Examples of spectra of late-type stars which display the variations of the 2.3 µm CO and 1.9 µm H2O absorption bands as functions of luminosity and temperature are contained in Moroz (1966), Johnson and Mendez (1970), McCammon, Münch, and Neugebauer (1967), and Frogel (1971). Back.
b) Galaxy Observations
The only galaxies included in the present sample are those early-type systems which were bright enough so that the CO index within a 48" diameter aperture could be measured on a 1.5 m telescope to a statistical accuracy of better than 0.02 mag in 4-6 hours. With improvements in detector sensitivity, this requirement was met for objects as faint as K 9.5. The galaxies include most of the ellipticals and lenticulars for which detailed optical-line indices are available from the work of McClure and van den Bergh (1968), Faber (1973b), and O'Connell (1976a). Some bright galaxies from these lists and elsewhere were excluded because they lie too close to the galactic plane (|b| 20° ) or at too southerly a declination, ( -5°). Because of uncertainties in the K-correction to the CO index, galaxies with redshifts greater than that of the Coma cluster (z = 0.022) were not observed. The galaxies observed are listed in Table 1.
All galaxy measurements were made with the focal plane apertures centered on the optical nuclei. Centering was done visually and usually confirmed by maximizing the 1.6 µm signal. The separation of the "signal" and "reference" beams was typically 2 or 3 aperture diameters. The red Palomar Sky Survey prints were checked for the presence of stars in either of the beams. If any were present, they were measured and an appropriate correction was made in the few cases where the stellar signal was more than a few percent of the galaxian signal. Nearly all of the measurements were repeated on two nights, and the CO indices for the fainter galaxies were measured on as many as five nights on the 60-inch telescope. The night-to-night scatter in the CO index and the broad-band magnitudes was found to be consistent with that expected from the statistical and photometric errors associated with the individual measurements.
c) Sources of Error and Instrumental Corrections
In addition to statistical errors from thermal background radiation from the sky and telescope, and from detector noise, a purely photometric error of ± 0.01 mag at all wavelengths was found from repeated measurements of standard stars.
Aside from random errors, several sources of systematic error exist in the raw data. Orthogonal scans of a star across the focal plane aperture for each of the filters revealed small but measurable differences in the beam profiles for the J, H, and K filters. 2 By convolving a standard galaxy surface brightness profile (de Vaucouleurs and de Vaucouleurs 1964; hereafter RCBG) with an average of the beam profiles through each of the 3 filters we derived corrections to the J - H and H - K colors and to the K-magnitudes. The corrections to the colors never exceeded 0. 06 mag and were typically 0.02 to 0.03 mag. The corrections to the K-magnitudes were typically 0.00 to 0.01 mag and never exceeded 0.02 mag. The errors associated with these corrections and possible other wavelength-dependent irregularities in the beam profiles are thought to be not larger than ± 0.02 mag. No systematic errors of this type greater than ± 0.01 mag in the CO index could be detected either by scanning the beams or by measuring a standard star at several positions in the aperture. 3
Possible errors from other known instrumental sources are believed to be of the order of 0.01 mag in the mean. These include the correction to the K-magnitude for galaxian flux in the reference beam (Frogel et al. 1975c) and nonlinearity in the response of the system to bright and faint standards. An intercomparison of the same standards measured on the 0.6 m, 1.5 m, and 5m, telescopes was used as a monitor of possible; nonlinear response effects. A linearity problem in our previously published CO indices (Frogel et al. 1975b) is discussed in Section f) below.
The observed K-magnitudes and colors corrected for the instrumental effects discussed above are given in columns (6), (8), and (9) of Table 1. The CO indices in column (10) have had no instrumental corrections applied. The adopted errors are listed at the bottom of Table 1. These errors are consistent with the scatter in repeated measurements after the application of the instrumental corrections.
||Corrected for Reddening and
* The assigned nominal errors are
± 0.03, ± 0.10, ± 0.04, ± 0.03, and ± 0.02 for K,
(V-K), (J-H), (H-K), and CO, respectively. A colon indicates that the
error for that measurement may be as much as twice the nominal
value. CO index errors larger than ± 0.02 are indicated in the
body of the table.
2 The usual character of the profiles was that the K-filter beam was relatively flat while that of the J filter was peaked in the middle. This arises from the wavelength-dependent properties of the silicon field lens. Back.
3 The measurements made on the 1.3 m telescope at KPNO employed an off-axis field mirror rather than a field lens. Thus, the beam profiles were the same at all wavelengths, and no corrections to the colors were required. Corrections to the K-magnitudes were found to be of order -0.03 mag. Back.
d) Sources of V Data
The visual magnitudes used in column (7) of Table 1 are based largely upon a compilation from the literature by Sandage (1976). We have also included data from Tifft (1969, 1973). As described in Frogel et al. (1975c), values for V at apertures corresponding to those at which the infrared measurements were made were interpolated from the tabulated data. Errors in V - K, from uncertainties in combining measurements made with different aperture sizes, and from scatter in the published V data, are estimated to be ± 0.10 mag. A colon after a V - K color in Table 1 indicates that the scatter in the published V data is exceptionally large; for these cases we adopt an error in V - K of ± 0.2 mag. In addition, a systematic error could be present in the 200-inch V - K colors which depend on a relatively limited amount of published small-aperture V data. A few galaxies had no appropriate V measurements available from the literature. Data for some of these were taken from UBVR measurements made on the 0.9 m and 1.5 m telescopes: at CTIO (Paper III). Clearly, the use of such a heterogeneous body of V data will hinder our examination of variations in the V - K color within galaxies since these variations are small to begin with (cf. Frogel et al. 1975c).
e) Reddening and K-corrections
To obtain the reddening and extinction corrections, the absorption-free polar-cap model for AV given by Sandage (1973) was used together with the Van de Hulst reddening law (Johnson 1968). Values of EJ-H / AV, EH-K / AV, EV-K / AV, EU-V / AV, and ECO / AV used are 0.10, 0.06, 0.91, 0.56, and 0.00, respectively. Since the largest value of AV was only 0.21 mag, errors which may arise from the inapplicability of either the polar-cap model or the Van de Hulst reddening law cannot affect the conclusions of this paper.
The K-corrections to the V-magnitudes are from Schild and Oke (1971) and Whitford (1971). (For the limited redshift range of the galaxies considered here, the results of both authors are identical.) For the U - V colors obtained from the literature, the K-correction is from Table 3 of Sandage (1972). Heliocentric redshifts (column , Table 1) are from de Vaucouleurs and de Vaucouleurs (1964).
Separate procedures were followed for determining the K-correction for the infrared broad-band data and for the CO index. The stratoscope balloon scans of several late-type stars (Woolf, Schwarzchild, and Rose 1964) were convolved with the transmission curves of the JHK filters and redshifted. The resulting K-corrections as functions of z were quite similar for Tau (K5 III), µm Gem (M3 III), and Ori (M2 I) - the spread at z = 0.03 was only 0.01 to 0.02 mag in the J - H and H - K colors and K-magnitudes. Since the CO indices of nearby galaxies are most similar to those of a late K III star, the values for Tau were used to correct the broad-band galaxy measurements. The K-corrections are approximately linear with increasing z; for z = 0.03 they are -0.02, -0.10, and +0.10 for J - H, H - K , and K, respectively.
Redshift corrections to the CO index were first determined numerically by convolving the transmission curves of the two filters with the high resolution spectra of several stars observed by Frogel (1971). For K2, K5, and M2 giants, the corrections were similar out to a z of 0.012. Beyond this they diverged rapidly, the spread becoming nearly 0.04 mag at z = 0.022. Uncertainty in the choice of a realistic starting point (i.e., a model stellar population) therefore made it necessary to derive empirically the K-correction to the CO index at large redshift. We show below that the CO index depends only weakly on galaxian luminosity or radius for the brighter nearby galaxies. Thus the observed CO indices of the intrinsically brightest galaxies were plotted against redshift, and a straight-line fit to the data was made. The K-correction found in this way, and adopted for all galaxies, is given by KCO = + 4.8z. Out to z = 0.012 the empirical relation agreed with the mean computed relation for the three stars to better than ± 0.01 mag. Although the empirical relation is based on observations of the most luminous galaxies, no systematic error is introduced in the results for fainter galaxies, since they are all at small values of z. (See Lasker 1970 for a discussion of this point.) Note that for galaxies in the Coma cluster the K-correction is two-thirds of the final CO index.
f) Comparisons with Previous CO Measurements
The CO observations presented in this paper were made on the photometric system described above and defined by the standards listed in Table A1. By reobserving with the new ("cold") system. many of the stars used to define Figure 1 of Frogel et al. (1975b) (the old, "warm" system), and by making simultaneous observations on both old and new systems for 16 giant stars, the transformation of the CO index from the warm to the cold system was determined to be
|CO index:||Warm - Cold = 0.1 × Warm||Warm 0.2|
|Warm - Cold = 0.02||Warm > 0.2|
i.e., the sense of the transformation is that the CO indices for the giants and supergiants in Figure 1 of Frogel et al. (1975b) become smaller. The error in the transformation as determined from the scatter of stellar measurements is of the order of the size of the transformation. This transformation applies only to giants and supergiants.
All of the galaxies measured on the old system were remeasured with the new system. For the data obtained with the 5 m telescope, a comparison with the measurements reported in Table 1 of Frogel et al. (1975b) revealed that even after the old CO indices were transformed, the resulting values for the CO index were larger than the new values by 0.04 mag in the mean. Reexamination of the old data indicated that the source of most of this systematic difference was due to saturation of the continuum from bright standards. The continuum filter (2.20 µm) gave a signal nearly three times as strong as the CO filter, and the signal through the former was clipped by about 0.03 mag. The effect of this was to assign too strong a CO index to the galaxies. Old measurements made with the 1.5 m telescope (Frogel et al. 1975b plus unpublished data) were also transformed to the "cold" system and compared with measurements reported here. No statistically significant differences were found. We emphasize that the systematic differences in the 200-inch measurements in no way affect the main conclusion of the earlier paper. The one consequence of the error in these measurements was to give a false impression of the existence of a radial gradient in the CO index (Frogel et al. 1975a, b). The galaxy data presented here supersede these previously published CO data.
The K-magnitudes of all galaxies in Frogel et al. (1975c) were remeasured in the course of this work. There is no systematic difference larger than 0.01 mag in the mean between these two independent sets of data.
g) Globular Cluster Observations
The integrated infrared light of five globular clusters was observed with the KPNO 0.9 m and the CTIO 1.5 m telescopes. Reference beam corrections were derived by using the multiaperture photometry of Kron and Mayall (1960) with the assumption of no radial color dependence. For M13, the correction to the K-magnitude was -0.11 mag, and for the other clusters, it was between -0.03 and -0.05 mag. Corrections to the infrared measurements for nonuniformity in the beam profiles were calculated in a manner similar to that employed for the galaxy measurements, and amounted to a few hundredths of a magnitude. V - K colors were formed using V magnitudes interpolated from the data of Kron and Mayall (1960) and transformed to the Johnson V system (Peterson and King 1975). Values of E(B-V) are from van den Bergh (1967) for M69 and M15, and are those recommended by Sandage (1970) for the others. U - V values are also from van den Bergh (1967), although not at the same aperture sizes as the infrared measurements. Final values with all corrections applied are presented in Table 2.
|M69||66.4||> - 0.4||5.62||0.70||0.17||5.57||1.15||2.59||0.64||0.13||0.08|
|* The "observed" values have been corrected for beam profile and reference beam flux as discussed in the text. The CO values did not require any corrections.|
| For M3, M13, M15, and M92, [Fe/H] is from Hesser, Hartwick, and McClure (1976). For M69, [Fe/H1] is an estimate based on the results of Hartwick and Sandage (1968).|