Table of contents

Contrary to historical expectation, stars within a given globular cluster often exhibit wide variations in the abundance of C, N, and O as well as certain light metals, particularly Na and Al. Owing to flux limitations, studies have been confined to evolved stars, especially to giants, but in few instances variations have been detected among main sequence stars. Variations in the Fe-peak elements exceeding ~0.1 dex are firmly established in the case of omega Centauri, the most massive cluster, and are strongly suspected in the case of M22, but in no other cluster. Among field halo giants of comparable Fe-peak metallicity, variations in the C, N, O group appear to be much less pronounced than in globular cluster giants. Among giants, the variations are of two kinds: (1) those related on the average to evolutionary state, and (2) variations among stars in the same apparent evolutionary state. In addition, clusters having the same Fe-peak abundances often contain stars with very different "signatures" of oxygen and CN-band strengths. The abundances of C and N are often anticorrelated, and in the limited number of cases in which both have been measured, O and N abundances have also often proved to be anticorrelated (Pilachowski 1988, Sneden et al. 1991, Brown et al. 1991, Kraft et al. 1992). Following pioneering work by Cohen (1978) and Peterson (1980), strong evidence has recently emerged for the existence of a significant global anticorrelation between O and Na abundances (Drake et al. 1992, Kraft et al. 1993). The observations are discussed in terms of contrasting hypotheses: evolutionary vs. primordial. In the former, the variations are attributed to the dredgeup of material that has been processed through the CNO cycle in the globular cluster stars themselves. In the latter, the variations are attributed to the primordial chemical inhomogeneities in the material out of which the cluster stars were formed, the composition of these "clumps" having been determined by nuclear processing in a prior generation of more massive stars. Observational evidence supporting each of these scenarios is cited. Recent studies of stellar rotation among horizontal branch stars in certain clusters (Peterson 1993) as well as new calculations of ²³Na and ²⁷Al production in te CNO processing regions of evolving low mass giants (Langer et al. 1993) lend fresh support to the evolutionary hypothesis. However, such calculations do not explain the variation of C and N abundances found among cluster main sequence stars (Suntzeff 1989, Briley et al. 1991) which seem explicable therefore only on the basis of a primordial scenario. Among mildly metal-poor giants, i.e., those in the range from solar metallicity to [Fe/H] ~-1, recent observational evidence suggesting the existence of a substructure in the [el/Fe]-ratios of the heavier alpha elements, e.g., Si, Mg, Ca and Ti, is discussed. The possible influence of this effect on the interpretation of the integrated spectra of extragalactic globular clusters and E galaxies is noted.

We have obtained new observations of both secondary and tertiary spectrophotometric standards with the RC spectrographs and CCD cameras on the 1.5-m and 4-m telescopes at CTIO in the wavelength range 6000-10500 A. We use the secondary spectrophotometric standards in order to calculate new fluxes for the tertiary standards of Stone and Baldwin (1983), as well as for three stars of the northern hemisphere from Stone (1977). We find that the synthetic magnitudes calculated from our spectra through the I_KC band agree extremely well the observed photometry, to better than 1% on average. For the monochromatic fluxes, we find an internal precision better than 0.01 mag blueward of 9000 A, and increasingly larger for longer wavelengths. We present also a fine grid of averaged monochromatic fluxes (at continuous steps of 16 A) for the ten secondary standards selected for our program, to be used in the flux calibration of high dispersion spectra. Finally, we provide combined fluxes for all secondary and tertiary standards in the range 3300-10500 A resulting from the calibration presented here with the one published in Paper I (Hamuy et al. 1992).

This paper begins by reviewing the nature and necessity of derived accidental errors, and it describes thier unsatisfactory traditional status in high-dispersion spectroscopy. The paper goes on to review the analysis which yeilds the [Fe/H] averages in subsequent papers of this series. Choices of stars and raw data are discussed first, and teh corrections applied to these data are then considered. These corrections include temperature corrections based on the results in Paper II of this series. The way in which the data are finally assembled is then described. The analysis yields accidental errors as well as means, and these errors are of interest in their own right. For recent data, a compromise decision about assigning the errors yields results which are modest--but usually not decisive--improvements over the errors of older data. The paper concludes with a brief review of some ways to obtain better accidental errors in high-dispersion work.

For Hyades and Coma dwarfs, [Fe/H] is found to be 0.107 ± 0.020 dex, respectively. At the 0.047-dex level, there is no sign of an [Fe/H] offset across the Hyades Li chasm. At the 0.064-dex level, there is no sign of a difference between [Fe/H] for the Hyades dwarfs and giants. At low resolution, the D strong-feature index of Taylor and Johnson (1987) is anomalously weak in the Hyades relative to field stars at 98% confidence or better. This result is from an analysis which excludes log g effects and allows for the effects of binaries. A similar analysis for B-V reveals no definite evidence of an anomaly in this index. In addition, if the Coma dwarfs are compared to their Hyades counterparts and the Coma-Hyades metallicty difference is allowed for, no B-V anomaly emerges.

We have carried out a survey of the photometric (V) variability of 61 "known" or suspected small-amplitude red variables, mostly M giants. Approximately two-thirds appear to be variable; several suspected variable comparison stars have also been identified. The incidence and average amplitude of variability increase rapidly from spectral type M0 III to M6 III.

The cataclysmic variable S193 was observed with the Ginga and ROSAT satellites along with ground-based optical observations. The bremsstrahlung temperatures and the column densities derived from these two observations are noticeably different. However, since the observations were separated by 3 years and took place at different optical magnitudes, it is not clear whether this is related to an intrinsic change in the system or to a two component source of X-rays. While the x-ray data are not sufficient to accomplish a detailed analysis for periodicities, the Ginga data place an upper limit of 40% on the amplitude of any sinusoidal modulation.

Observations of the planetary nebula Me2-1 were obtained with different set-ups, in order to analyze some effects of contamination produced by the second order ultraviolet spectrum in the first order red, for wavelengths longer than approximately 6000 A. This contamination problem will be discussed elsewhere. Here we present the observations of Me2-1, which include a wide wavelength range, from about 3100 to 10200 A. A comparison with previous results is shown for the wavelength intervals in common. From these data, the most relevant nebular parameters are derived, using lines in the optical and near IR regions. Some parameters corresponding to the central star are also determined. The results are compared with previously obtained values.

We have obtained CCD spectra of newly-discovered shell-like nebulae around the WN4 star Breysacher 13 and the WN1 star Breysacher 2 in the Large Magellanic Cloud. The shell around Br 13 shows definite signs of enrichment in both nitrogen and helium, having much stronger [N II] and He I emission lines than are seen in typical LMC H II regions. From the measured electron temperature of about 17,000 K in the shell, we derive He/H and N/O aubndance ratios which are factors of two and more than ten higher, respectively, than the average LMC interstellar values. The derived oxygen abundance in the Br 13 shell is down by a factor of eight compared to the local LMC ISM; however, the derived electron temperature is affected by the presence of an incomplete shock arising from the interaction of the stellar wind with photoionized material. This uncertainty does not affect the basic conclusion that the Br 13 shell is enriched by processed material from the Wolf-Rayet star. In contrast, the shell around Br 2 shows no clear evidence of enrichment. The nebular spectrum is characterized by extremely strong [O III] and He II emission and very weak [N II]. We derive normal He, O, and N abundances from our spectrum. This object therefore appears to be simply a wind-blown structure associated with a relatively dense cloud near the Wolf-Rayet star, although the very high ionization state of the gas is unusual for a nebula associated with a Wolf-Rayet star.

We present deep color-magnitude diagrams for two fields in the Large Magellanic Cloud, located about 6 arcminutes from 30 Doradus. The data are from images taken with the Hubble Space Telescope's Wide Field Camera, as part of the Medium Deep Survey Key Project. We describe a procedure for measuring stellar magnitudes in relatively crowded HST images based on aperture corrections. This procedure is used to obtain color-magnitude diagrams containing a total of ~7200 stars with 18 < V < 24. The color-magnitude diagrams shows a well populated main-sequence, and a pronounced giant branch concentrated at V ~ 20. There is tentative evidence for a population of low metallicity main-sequence stars, which may confirm the suggestion that the LMC has, like spiral galaxies, a metal-poor, spheroidal component.

We present new optical imaging in two broad-band colors ofthe z=0.29 bright QSO 1302-102, obtained with the Wide Field Camera of the Hubble Space Telescope. We have performed deconvolution on the HST data, previously published CFHT HRCam data, and the combined data. The structure and brightness are measured of an extended knot 2.3 arcsec from the QSO, and also of brighter structure in the inner 1 arcsec of the QSO. We compare the optical structure with new and existing maps of the radio structure from the VLA. The 5 arcsec radio structures lie along the directions of some of the optical components. We discuss the implications.

We present an analysis of CCD images of more than 100 quasars; many of the objects have been or will be observed in the HST Key Project Quasar Absorption Line Survey. The data set consists of B, g, and i photometry of 117 quasars and deeper g (limiting magnitude of ~22) images of 101 quasar fields. Positions accurate to ~1", measured with the Space Telescope Science Institute's Astrometric Support Program, are listed for all of the quasars in this study. Positions, magnitudes, and classifications of stars and galaxies located within ~100" of the quasars are given for all of the dep g data. The positions of the stars and galaxies relative to the quasar are accurate to ~0.5". The results presented here can be used to prepare spectroscopic programs designed to obtain redshifts of galaxies in the fields of these quasars.

A variety of specialized devices have been developed to aid the operation of an unattended observatory. We describe here a precipitation monitor and two types of cloud monitors. One cloud monitor uses a small telescope and CCD to continuously measure the brightness of Polaris. Other devices described include a simple altitude limit switch, an automated mirror cover, and a liquid nitrogen autofill system.

X-Ray point sources observed with ROSAT High Resolution Imager (HRI) often appear elongated over scales of ~5" - 10" from the image core. This elongation has been attributed to errors in the attitude correction as the satellite is wobbled during the observations, and affects sources with both soft and hard X-Ray spectra. In this paper, I report the results of an attempt to rid a high signal-to-noise observation of the soft X-Ray point source HZ 43 of its characteristic elongation. I divided the observation into 181 separate images, each containing photons from only a small region on the detector through which the source passed during the satellite's wobble. By measuring the positions of the individual image centroids, I found clear evidence for systematic offsets from a common mean by up to ~ ± 3" in both right ascention and declination as a function of phase in the satellite wobble. Shifting the sub-images to a common center and then restacking them into a single image measurably improved the symmetry of the point spread function. HRI observations are wobbled primarily to smooth out variations in the pixel-to-pixel sensitivity of the detector and also to extend the lifetime of the microchannel plates in the detector since these decay at a given location as a function of the number of photons detected. However, the elongations introduced by the aspect errors inhibit the identification of possible extended X-Ray emission associated with sources such as pulsars and active galactic nuclei. In light of these results, I suggest that until the aspect errors are understood, observations of compact sources, where this effect may be important, should not be wobbled.

Given a few determinations of the occurrence of a periodic event spanning a large interval of time, one can hope to make detailed determinations of the period and its behavior with time. This is accomplished by obtaining the correct cycle number for each observation: we show here that a modification of the Hough transform provides a robust technique for this detection. Very few observations of the event could then be all that is needed. An application of such a method to the eclipsing binary GW-Cep is given, showing that the most likely period change of this variable is opposite in sign and approximately one order of magnitude smaller than currently estimated.

This dissertation combines and instrument and a technique that have been studied separately for several years. The instrument is the multi-anode microchannel array (MAMA) detector. This device is a photon-counting imaging system that has been developed primarily for use in ultraviolet astrophysics. Current generation MAMA detectors have the ability to record the arrival time of every detected photon and can distinguish events which arrive only 140 nanoseconds apart in time. The technique is speckle imaging. This is an interferometric method of recovering high-resolution information about astronomical objects from ground-based telescopes, usually carried out in visible or infrared wavelengths. In conventional astronomy, fluctuations in the atmosphere above the telescope blur images so that the resolution is typically 1 to 2 arcseconds, much worse than the theoretical diffraction-limited resolution of large telescopes. Speckle imaging allows for the reconstruction of image features on a much smaller scale by taking many short (typically 0.01 second) exposure images of an object, where over the exposure time the atmospheric fluctuations are effectively frozen. Speckle imaging devices must therefore have the ability to read out images much faster than most other astronomical imaging devices. Since the MAMA detector offers outstanding timing abilities as well as good spatial linearity and is available in a visible light version, it is an excellent choice as the imaging device in a speckle imaging system. A new speckle system based on the MAMA detector has been built at Stanford University and a detailed study of the performance of the system forms a principle part of this work. Under certain observational conditions, the phenonomenon of channel saturation of the MAMA detector can significantly affect the signal-to-noise ratio of the speckle imaging results. In addition, channel saturation leads to a systematic error in the determination of the irradiance ratios (or equivalently, the magnitude differences) of interferometric binary stars. A correction algorithm is described which leads to better estimates of these ratios in simulated data. Four methods of computing the correlation functions necessary for producing high-resolution reconstructed images are compared using MAMA data. Two of these methods explicitly make use of the time-tagging ability of the MAMA detector and have not been previously studied in detail. The highest signal-to-noise ratio in the power spectrum and bispectrum is obtained with a weighted running-window method. Reconstructed images of several objects are presented. The MAMA-based speckle imaging system is shown to determine accurate position angles and separations of sub-arcsecond separation binary stars.