With nearly 1500 diffraction limited measurements of high accuracy to its credit, speckle interferometry must now be considered an established and productive method for performing binary star astrometry. This paper will attempt to give an explanation of the technique from the perspective of a binary star observer along with a general summary of the accomplishments that have given speckle interferometry its present respectability. Some recommendations aimed at increasing the effectiveness of the method will also be made. Speckle interferometry has provided valuable information on stellar diameters, and tantalizing results are beginning to appear for a variety of faint objects, but only the application to the measurement of binary stars will be considered here.

We discuss speckle interferometry and speckle holography measurements of double and triple stars. Speckle holography with the ESO 3.6m telescope yielded direct images with a resolution of 0.03 arc second (diffraction limit). Finally we discuss the speckle rotation method, which can be used to reconstruct high-resolution images from Space Telescope data.

Since 1977, the Digital Speckle Interferometer chiefly developed for the study of atmospheric structure in cool supergiant stars, has been used for observing multiple stars. It has provided binary star resolution down to 30 milliarcsec. separation with a precision from 3% to 10%.

The incertitude on position angle determination ranges from ± 1° to ± 5°.

Working in photon counting mode, we are able to estimate magnitude difference reaching Δ m ~ 3.0 with an uncertainty of about 0.2−0.3 magnitude.

On the basis of present experimentation, we think that 17th mag binaries can be resolved with a 3.6 meter telescope.

The CERGA interferometer is made of two 25 cm aperture telescopes with a variable north-south baseline, spanning from 5.5 to 35 meters. In addition to stellar diameters, it has provided binary stars measurements with a 0.5 milliarcsecs precision for separation down to 2.5 milliarcsoconds (Labeyrie 1971).

The visual limiting magnitude is presently 3.5. Large separation binaries will be observed with one milliarcsecond precision using a slightly different technique. This should provide a mean of detecting possible planetary induced orbital perturbations.

In the decade since its invention by Labeyrie in 1970, speckle interferometric techniques have evolved from simple optical processing of photographic images to high-speed digital processing of quantum-limited video data. This progress has been summarized in preceeding papers by McAlister and Weigelt in this colloquium.

The basic hardware of our system has been described in Hubbard, et al. (1979 – The basic speckle camera) and in Hege et al. (1980 – The intensified video system). These capabilities have been successfully applied to observations over a dynamic range of 16 magnitudes, from Betelgeuse (Goldberg, et al. 1982) and Capella (this paper), at one extreme, to Pluto/Charon (Hege, et al. 1981a) and the 16th magnitude resolved QSO system PG1115+080 (Hege, et al. 1981b) at the other. To accomodate this dynamic range two distinct data-recording/data-processing modes have been implemented. The Analoue mode records the image intensity for each pixel in the speckle interferogram. This is applicable for objects brighter than 7th to 10th magnitude, depending upon telescope aperture, observing band-pass and detector image scale. For fainter objects the Event mode records individual photoelectron addresses in the speckle interferogram.

Speckle interferometry is a powerful tool for close binary star research allowing angular resolutions as small as 20 milliarcsecond. A technique is proposed to resolve spectroscopic binaries with even smaller separations. It uses the fact that speckle images taken in one or the other of the Doppler shifted spectral lines give a different intensity weighting of the two components of the binary. The location of the speckles in the two speckle images is therefore different, the direction of the displacement being related to the position angle of the binary, the amount of the displacement being related to its separation. Since the location of speckles can be determined with a much higher precision than their diameter, this creates the possibility for submilliarcsecond observations. This paper describes an experiment being started now to use this so-called “Differential Speckle Interferometry” technique for the study of binaries, stellar rotation, stellar chromospheres, Ap stars and other objects.

A new technique called shift-and-add, which allows nearly diffraction-limited images to be formed of objects viewed through a randomly distorted, turbulent media, has been formulated. True images may be obtained through the earth’s atmosphere when the method is used. It has been shown that this can be achieved even though the telescope may have severe aberration and when using a much wider bandwidth than that used at present in stellar Speckle Interferometry. Results are given for an object of greater extent than simple binary or groups of point sources.

Bates and Cady (1980) proposed a method for astronomical image reconstruction that involves superimposing regions about the brightest pixels of each frame produced by speckle interferometric techniques. Unlike the method of Lynds, Worden, and Harvey (1976), no attempt is made to center apparent speckle images before superposition. This method is hereafter referred to as “shift-and-add” (SAA). The simplicity of this method makes it highly suitable for analysis and simulation.

This paper summarizes the application of extensions of SAA to the problem of determining reliable double star intensity ratios. Theoretical and simulation results presented in a previous paper (hereafter called “paper I”) by Bagnuolo (1982) will be summarized. In addition, the results of the application of SAA-type algorithms to Capella will be presented.

The achievements of speckle interferometry in recovering diffraction limited spatial information, primarily by optical processing of photographic data, have been well summarized by McAlister in a preceeding paper. In general, the recovery of intensity information has received less attention mainly because of complications such as the need for accurate deconvolutions and noise bias corrections.

Methods for producing image power spectra (or the equivalent image autocorrelation functions) from digitally recorded speckle interferograms, as well as methods for correcting these functions for seeing effects, are described in our previous paper, Cocke, et al. in this colloquium. In this paper we discuss effects of instrumental response and photon statistics, the so-called noise bias, and procedures for correcting these effects in order to recover binary star relative intensity information from speckle interferometric data. We find for Capella δmv = 0.48.

The observations of α Vir with the Narrabri Stellar Intensity Interferometer demonstrated the potential of long baseline interferometry for the determination of fundamental properties of double-lined spectroscopic binary systems. Since the completion of the programme with the Narrabri instrument the Chatterton Astronomy Department has been conducting a study aimed at developing a stellar interferometer with limiting magnitude V ≳ +8 and maximum baseline ≳ 1 km (resolution at 500 nm ≲ 7 × 10−5 seconds of arc). The way in which a long baseline interferometer may be used in the study of binary stars is outlined, the requirements for this work are discussed, and the current status and future plans of the Chatterton Astronomy Department’s programme to develop a new long baseline interferometer are summarised.

The basic goal of the Amplitude Interferometer Program at the University of Maryland has been to achieve very high angular resolution, far beyond the angular resolution which is normally permitted by the Earth’s atmosphere. This program is primarily directed to the observation of objects of astrophysical and astrometrical interest.

The initial general science objective has been the measurement of the angular diameters of various single stars at various wavelengths. As shall be discussed later, this has been achieved on the current equipment.

The second objective of the Amplitude Interferometer Program is the study of binary and multiple star systems. Some initial tests have been conducted to validate the approach and the expected accuracy using the existing Amplitude Interferometer (AI). However, detailed tests and/or a regular observing program require the multiplexing capablity of the next generation Amplitude Interferometer (the Multi-Aperture Amplitude Interferometer) in order to have a reasonable observing efficiency for the use of the telescopes which would be involved

A parallax program was started in July, 1980, using a CCD detector at the prime focus of the KPNO 4m telescope. The results obtained to date indicate that the centroids of stellar images can be measured to accuracies of 0.25 microns or better from each 2 minute exposure on stars of about 17th magnitude with a filter of 800 Angstroms width centered at 6500 Angstroms (the scale of the KPNO 4m with the doublet corrector is 19.60 milli-arc-seconds per micron). The current CCD, a Fairchild CCD 211, is of relatively low quantum efficiency, about 8%. The integration times become prohibitively short on 15th magnitude and brighter stars. This technology seems quite attractive for the observation of astrometric binaries, as accuracies of about 3 milli-arc-seconds are obtained with only 12 minutes of telescope time, and there is no special set-up required for each field. If time permits, some comments will be made about some preliminary astrometric tests of a 3-phase TI 800 × 800 CCD performed in January, 1981, in collaboration with Roger Lynds of the Sapce Telescope Wide Field Camera Team.

The accuracy of the area scanning technique depends on observing conditions resulting from seeing, tracking irregularities of the telescope, photon and seeing statistics (integration time and the size of the telescope), and the atmospheric spectral dispersion of the stellar image in the spectral region used. In addition, the reduction method as well as the mechanical and optical quality of the telescope and photometer influence the accuracy of the results.

The tracking error of the telescope is usually larger than the observer believes. The amplitude of the periodic error of the worm gear is at best ± 0.5 arc sec with period length between 4 and 10 minutes, depending on the particular telescope. Figure 1 shows the tracking error for the ESO 50 cm and 100 cm telescopes. Other telescopes usually fit in between this two extremes.

Accurate magnitudes and colors of individual components of binaries are needed to provide valuable information on stellar evolution. Using an area scanner, accurate magnitudes, as well as good astrometric data, can be obtained for close pairs. A summary is given for the first major data set.

Classifying the spectral types of individual components of composite spectrum stars can be a difficult task. Markowitz (1969) discussed the problems of blended and veiled lines associated with the visual assessment of blue spectrograms. Although synthetic spectra (i.e. formed by the intentional superposition of two different spectra) can be used as a means of calibration, the classification of composites with slit spectrograms is a subjective process. Classifications based upon broad- and intermediate-band photometry (Young 1971) can be highly objective, but the lack of spectral resolution introduces other complications. For example, colors in the UBVR system measure the gross behavior of a star's energy distribution. Results based upon this information are not only affected by interstellar reddening but also are incapable of identifying common spectral peculiarities. The use of spectrophotometric indices (Beavers and Cook 1980) avoids the major difficulties of other techniques. The results are not influenced by the veiling or blending of lines, nor are the results influenced by interstellar reddening. If a sufficient number of features are observed, common types of spectral peculiarities can be identified.

The results of some first laboratory analysis of double star images recorded with a TV camera are presented.

The camera – already in use since a few years for occultation-timings has been described earlier. For double star observations, the equipment has been completed by a video disc memory, a video analyser with analogic-digital converter and a small programmable calculator.

Details are given about this equipment and its possibilities. The reduction method in its present state is described as well as some first results. These show a very good internal agreement leading to a high promising new observational technique.

The increase in accuracy of astrometric results has been very limited in the last half century, and improvements of more than about a factor of 3 at best for ground-based observations are not expected for the next two decades. This is largely due to well known error-sources connected with the Earth atmosphere. By moving the observing equipment to a platform aboard a satellite many of these errors would disappear completely, or to a large extent. Unfortunately till now no space astrometric project has been realised. Consequently there are no astrometric observations of visual binary stars from space. Of course, the astrophysical parameters of double stars are as important as their astrometric data. A large number of double stars has been observed from space by measuring their radiation and spectra in different wavelengths regions not attainable from ground based telescopes. A good example is HZ Hercules (Her X-1).

With the success of the Shuttle it seems now the Space Telescope will become a reality. The author wrote a study for the Langley Research Center in the summer of 1963 on the astronomical uses of a large space telescope which formed the basic document for the National Academy of Sciences Woods Hole study in 1965. This study recommended construction of a large space telescope scaled to use a Saturn V as the launch booster.

The initial study included some astrometric applications. Among these were various double star applications which is the subject of this symposium.

Although the Space Telescope Wide-Field/Planetary Camera is not primarily an astrometric instrument, it is expected to have some astrometric capability. Moreover, one of its possible astrometric applications is of unusually high scientific importance, namely, an attempt to detect the presence of planets around nearby stars. Let me therefore adopt that application as an example for discussing the anticipated astrometric performance of the camera system. It will be expeditious to make use of some diagrams that I have presented previously elsewhere, so part of the story may sound familiar (Baum 1979a, 1979b, 1980a, 1980b; Baum, Thomsen, and Kreidl 1981).