As the circumstances of the early electrical measurements of starlight in Ireland are not widely known we wish to take this opportunity to set down the facts, as far as they are known to us. Corrections or new information will be welcomed.

The observations made in Dublin in 1892 were the result of a collaboration between four graduates of Trinity College: George M. Minchin, William H.S. Monck, Stephen M. Dixon and George F. Fitzgerald. The observations in 1895 were made at Daramona Observatory, Co. Westmeath with a 24-inch reflector by Minchin, Fitzgerald and the owner of the telescope, William E. Wilson. In 1875 Minchin was appointed Professor of Applied Mechanics at the Royal Indian Engineering College at Coopers Hill, near Staines in London. In 1877 he started a long series of investigations of photoelectricity using a small optical laboratory at Coopers Hill and the laboratories of University College London. His initial aim was to transmit images electrically but he became skilled in making photovoltaic cells of selenium.

By September 1891 Minchin had succeeded in making some working cells and he wished “to test them on the stars”. He contacted his friend Monck who had recently set up an observatory in his back garden in Dublin. Monck in turn asked Fitzgerald for the loan of a galvanometer or electrometer and for advice in its use. As Fitzgerald did not have a suitable instrument, he ordered Clifton’s form of Thomson’s quadrant electrometer, which arrived near the end of the year. It seems that Minchin also sent some of his cells to Dr Boeddicker in the spring of 1892 for trials on the 72-inch telescope but no reports are available.

This historical review covers the first fifty years of the electrical measurement of starlight, which saw its beginnings in August 1892 when William Monck used a Minchin photovoltaic cell on his refractor in Dublin. The work of Stebbins using photoconductive cells from 1907 and of Guthnick, Stebbins and others using photoelectric cells from 1912 is also discussed. The advances brought about by thermionic amplification and the red-sensitive Cs-O-Ag photocathode in the early 1930s by respectively Whitford and Hall enabled new astrophysical problems to be tackled. Although many observers attempted photoelectric photometry, relatively few were successful during the 1920s and ‘30s. It was not until the introduction of the 1P21 photomultiplier tube that astronomers had a reliable and sensitive detector for photometric observations.

Multicolour photometry has historically been carried out within a variety of standard systems. With the advent of new detectors with different wavelength sensitivities to those of the original system and the use of subsets of secondary standards, many subtle and not so subtle changes have occurred to the original systems. However, by reverse engineering, the passbands of the modified standard systems can be determined which enables good realisation of theoretical colours and better passband matching for CCD-based photometry. In this paper, the various photometric systems will be discussed and compared and strategies will be outlined for ensuring that precise relative photometry is maintained in the future, in particular from area detectors.

There are two large sets of high precision standard star photometry which claim to represent the UBV(RI)c system, one in the E regions in the south established by Cousins and one at the equator due to Landolt. There appear to be systematic differences between them. Astrophysical interpretations which depend on photometric data may in consequence depend on which set of standards was used.

Many photometric systems for classification of stars have been proposed. However, not all of them are equally suitable for investigation of distant and reddened stars without additional information from their spectra. Earlier it was shown that the Vilnius photometric system is optimum for purely photometric classification of stars in spectral types, luminosities, metallicities and peculiarity types, when interstellar reddening is present. When realized with a CCD detector, the system permits to classify stars down to 17 mag with a 1.5 m class telescope or down to 20 mag with a 4 m telescope. A possibility of realizing the system with the WF camera of the Hubble Space Telescope is investigated. The problem of unification of broad-band and medium-band photometric systems is discussed.

The seven intermediate passband filters of the Vilnius photometric system enable a more comprehensive and discriminating classification of stellar objects than is possible in broadband systems. To date, a lack of suitable standard stars has inhibited its use in the Southern Hemisphere. We have an established programme for extending the network of northern Vilnius standards into the southern skies. This paper briefly summarizes the characteristics of the Vilnius system, provides a synopsis of the data obtained on over 200 standard stars so far and presents preliminary CM and CC diagrams for the open cluster Omicron Velorum (IC2391).

We present first efforts to establish a network of absolutely calibrated continuous infrared spectra of standard stars across the 1-35μm range in order to calibrate arbitrary broad and narrow passbands and low-resolution spectrometers from ground-based, airborne, balloon, and satellite-borne sensors. The value to photometry of such calibrated continuous spectra is that one can integrate arbitrary filters over the spectra and derive the stellar in-band flux, monochromatic flux density, and hence the magnitude, for any site. This work is based on new models of Sirius and Vega by Kurucz which were calculated by him, for the first time, with realistic stellar metallicities and a customized finely-gridded infrared wavelength scale. We have absolutely calibrated these two spectra and have calculated monochromatic flux densities for both stars, and isophotal wavelengths, for a number of infrared filters. Preliminarily, the current IRAS point source flux calibration is too high by 2, 6, 3, and 12% at 12, 25, 60, and 100μm, respectively.

The accuracy with which infrared photometry can be carried out is currently limited by poor definition of the instrumental system and by the accuracy of the available standard stars. We present new colour transformations between the CIT J, H, K photometric system and systems currently in use at Cerro Tololo, Mauna Kea and Kitt Peak. The precision of the J, H, K data for some of the stars observed by Elias et al. is improved and the system extended to fainter stars suitable for use with larger telescopes and/or infrared arrays. Evidence of infrared variability has been detected for one M dwarf star in our programme.

There are several standard fields for CCDs available in the literature. Some of these have been observed many times with the CCD cameras of the Isaac Newton Group of Telescopes on La Palma. These observations were made in order to establish the zero-points and colour equations for the U BV RI filters. Now that many observations are available in the data archive it is possible to combine them and search for inconsistencies in the published magnitudes and colours. This discussion will lead to standard fields of a higher quality. Furthermore astrometric plates have been taken of several of these fields which will be used to provide accurate positions as well as magnitudes and colours. These will provide the scale, orientation and distortion of any CCD camera.

The current precision of differential comparisons, within a fixed instrumental system, reaches about a millimagnitude. The current accuracy of transformations between systems is on the order of 0.01 mag, but is worse in some cases. The transformation errors are due to mis-matches between instrumental and standard systems, and cannot be transformed away without further information. A simple geometric model is proposed, which illuminates the transformation problem, and suggests means of minimizing it. Misplaced band edges are the main cause; tighter specifications on filters will help to reduce these errors. The ultimate solution is to use sampled systems in which some bands look like the derivatives of their neighbors. This automatically produces enough band overlap to comply reasonably well with the sampling theorem.

A consise overview of the errors related to transformations of data from one photometric system to another is given.

Traditional photoelectric techniques can achieve millimagnitude precision if a few simple rules are scrupulously followed. We illustrate the practical limits of differential photometry obtained with a dedicated telescope and photometer. Using a highly homogeneous set of measurements of bright F, G, and K solar analog stars, we present the distribution of observational noise and intrinsic stellar variability over a typical observing season and over the entire 8-year program. Finally, we list Strömgren uvby standards which we found to vary slightly.

The three-star technique for photoelectric photometers is discussed. The technique is used to study millimag variations of variable stars with periods in the range of 30 minutes to a few days. It is emphasized that for highest precision two comparison stars should be used and observed as often as the variable star. The different types of numbers used to express the precision of measurement are discussed together with possible misinterpretations.

Finally, the successful use of an extinction coefficient derived from differential extinction between the three stars is illustrated with measurements showing an extreme amount of instrumental sensitivity drift.

Stellar intensity scintillation on short and very short time scales (≃ 100 ms - 100 ns) was studied using an optical telescope on La Palma (Canary Islands). Photon counting detectors and real-time signal processing equipment were used to study atmospheric scintillation as function of telescope aperture size, degree of apodization, for single and double apertures, in different optical colors, at different zenith distances, times of night, and seasons of year. The statistics of temporal intensity variations can be adequately described by log-normal distributions, varying with time. The scintillation timescale (≃10 ms) decreases for smaller telescope apertures until ≃5 cm, where the atmospheric ‘shadow bands’ apparently are resolved. Some astrophysical sources may undergo very rapid intrinsic fluctuations. To detect such phenomena through the turbulent atmosphere requires optimized observing strategies.

The Sydney University Stellar Interferometer, SUSI, a modern version of a Michelson stellar interferometer, will be capable of measuring the angular diameters of stars down to V ≈ 8.0, for all spectral types. The resolution of the 640 m baseline — <0.1 milliarcsec at optical wavelengths — includes reasonable sample sizes of all spectral types and luminosity classes to this magnitude limit.

The angular diameter data require complementary photometric, spectrophotometric and also spectroscopic data for the determination of fundamental parameters of single and multiple stars. This paper discusses the accuracy required of these data for some of the main problems to be tackled by SUSI. A selection of the programmes planned for this unique instrument is listed below, where the need for complementary data is also indicated.

1. • The measurement of the changes of angular diameters of Cepheids and other pulsating stars such as Mira variables. Comparison with the linear changes determined from radial velocity curves will enable their distances and luminosities to be determined by an essentially geometric method.

2. • The measurement of the angular sizes of the orbits and of one or both components of spectroscopic binary stars. This effectively makes them ‘visual’ binaries so that the orbital inclination, i, may be determined. When these data are combined with the velocity curve solutions which include (mass)sin3i and (semi – majoraxis)sin i, we may determine masses and linear radii of one or both components and also the distances to the systems. Also, the light curve of an eclipsing binary provides information on the ellipticities and radii of the components and on the eccentricity of the orbit — information which may be used to aid the analysis of the SUSI data. In addition, the determination of the light curve (some bright systems have not been observed for up to one or two decades) is the quickest method of finding the current phase of its orbit.

3. • The establishment of the total-flux-based temperature scale of all spectral types and luminosity classes from O to M. This requires accurate photometry and spectrophotometry over as wide a wavelength range as possible; it therefore also requires observations from space observatories, as was also necessary for the analysis using the Narrabri intensity interferometer data (Code et al., 1976).

4. • The measurement of diameters of stars with shells or extended atmospheres, such as Be and Wolf Rayet stars, at different wavelengths — in particular, in the emission lines.

Three sets of spectrophotometric standards are presented: eight bright B and A stars of 00 - 265, fourteen B - A stars of 3m-4m from the catalogue of 238 secondary spectrophotometric standards and 24 A0-G0 stars of 7m-8m. For all of these stars UBV photometry was produced and brightness variations exceeding O01 were not found.