The riddle of cosmic darkness, nowadays known as Olbers's paradox, asks, “Why is the sky at night dark?” The riddle is almost as old as the Copernican revolution and has inspired important contributions to the development of astronomy and cosmology. It climaxes in the twentieth century with the realization that the sky is covered not by unseen stars but by the infrared gloom of the big bang.

A historical review of integrated starlight, diffuse galactic light, and extragalactic light studies is presented. Together, these components compose the “background light.” Methods ranging from star counts to space-based photometric surveys have succeeded in quantifying the contribution of each component of the background. Integrated starlight is the dominant component. The contribution of diffuse galactic light in the general interstellar medium peaks slightly off the galactic plane and declines toward higher latitudes. The extragalactic light has been determined from both galaxy counts and photometric methods. The blue and red intensity and B–R color distribution of background light have been mapped. The relation between galactic structure and background light measurements is established. The distribution of interstellar extinction is the primary regulator of the brightness. However, spiral arm and stellar distribution effects are discerned in Carina and Sagittarius. The sun lies 13 pc north of the galactic plane as defined by brightness and dust distributions.

The near-infrared region of the spectrum is a particularly advantageous window for observing the distribution of old, evolved stars in the galactic disk and bulge. These stars are important because they provide an excellent tracer of the overall stellar mass distribution. At shorter wavelengths extinction is a serious problem, and at longer wavelengths the flux is dominated by dust emission. A summary of the large-scale diffuse near-infrared observations of the Galaxy is presented, as is a summary of the results obtained from these data on the structure of the galactic disk and bulge. The importance of combining CO and near-infrared maps of similar resolution to determine a three-dimensional model of galactic extinction is demonstrated. The Spacelab-2 Infrared Telescope (IRT) data are used in conjunction with a proposed galactic model to make preliminary measurements of the global scale parameters of the Galaxy.

The distribution of UV bright stars in the Galaxy is described. The integrated ultraviolet (UV) stellar radiation in the Galaxy is obtained from these data, and this flux of radiation is compared with observations and models of direct starlight in the Galaxy.

Most stars emit most of their radiation in the wavelength range between the far ultraviolet (FUV) and near-infrared (IR). Eddington (1926) first estimated for the solar vicinity a mean radiation intensity of the interstellar radiation field (ISRF) equivalent to that of blackbody radiation of ~3 K but with a spectral distribution that can be approximated by ~10,000 K blackbody radiation diluted by a factor ≈ 10−14.

This is a summary of the few observations available to date. Although there remain doubts concerning the reliability of most of the published results, a linear polarization of a few percent, perpendicular to the galactic plane, is indicated. Therefore, scattering of galactic light by interstellar dust appears to be the dominating source of galactic background polarization. There is a severe lack of model calculations, which would be needed for a physical interpretation of the data.

In this review I discuss some aspects of the luminosity distributions in our Galaxy and external spiral galaxies. The major conclusions are the following: (1) the radial scale length of the luminosity distribution in the disk of our Galaxy is 5.0 ± 0.5 kpc, (2) on this basis the Hubble constant needs to be at most 65 ± 10 km s−1 Mpc−1, if our Galaxy and M31 are among the largest spirals, as the Fisher-Tully relation suggests, (3) the probable Hubble type of the Galaxy is SbI–II, (4) the bi-modal distribution function of face-on, central surface brightness μ0 and radial scale length h of spirals shows a preferred value for μ0 of about 22 B-mag arcsec2 and a distribution of h that declines with one e-folding per kpc, (5) the Galaxy is a normal, fairly large Sb galaxy, and (6) galaxies similar to our own in terms of large-scale, nonmorphological properties are NGC 891 and NGC 5033.

An analysis of fluctuations in the brightness of the Milky Way using the concept that interstellar matter occurs in the form of discrete clouds was first applied by Ambarzumian (1940, 1944). This theory was formulated in a general way and discussed in great detail in a series of papers by Chandrasekhar and Münch (1950a, 1950b, 1951, 1952), by Münch and Chandrasekhar (1952), and by Limber (1953). More recently Peters (1970) presented an analysis of this kind based on extensive photographic observational material. Although the influence of clumpiness of the dust distribution on the mean integrated starlight was thoroughly discussed in these papers, it has not been properly included in most of the photometric models of the Galaxy. Only the models of Caplan and Grec (1979) and Mattila (1980a, 1980b) incorporate these effects.

Photoelectric photometry is inherently much more accurate than photographic work. But large time variations of airglow, inaccurate elimination of bright stars, and the problematic adding-up of different scans may destroy this advantage in the case of surface photometry. On the other hand, the low internal accuracy of photographic photometry and its problematic nonlinearity can be overcome by a sufficient number of plates and by careful photoelectric calibrations. Such measures allow us to take full advantage of the high homogeneity of super-wide-angle photographs over extremely large areas, their global calibration, and their direct imaging. When I started this work, no charge-coupled device or similar device was available. But even today, with such detectors problems would arise for surface photometry in flat-fielding, zero-points, and bright star elimination. Our camera (Schlosser and Schmidt-Kaler 1977) followed the Henyey-Greenstein design with a f/4.3 Zeiss UV-Sonnar lens (with constant focus from 2000 Å up to 10000 Å) and f/0.4 spherical mirror of 30 cm ⊘, giving an effective f=18 mm. The angular resolution, limited by optics and plate, is 4′ in the center, 7′ in the average, and about 20′ at the edges of the field of view (=140°).

Star-counts on the Palomar Sky Survey Atlas were made in 1968-87 in order to estimate the background brightness in zodiacal light and airglow observations. Measured diameters of blue and red images of each star were converted to the magnitude with empirical diameter-magnitude relations. The resultant blue and red brightnesses and colors of the integrated starlight are presented for 24 sky regions, including 4 polar (North Celestial, North Ecliptic, North and South Galactic) and 10 ecliptic regions.

We performed UBVRI photometry in a 4.7° field of diameter centered on the North Ecliptic Pole, covering all 94 SAO stars and one planetary nebula. This represents the first step of an attempt to improve the knowledge of integrated starlight in this important viewing direction. The contribution of these bright stars to the integrated starlight is considerable (30–40%) and quite unevenly distributed. Therefore, surface photometry with large fields of view should be individually corrected for this contribution.

Although many of the nearest dim and cool stars (the red dwarfs) were catalogued in the early sixties, the majority of the astronomers did not realize that these objects provide almost nine tenths of all the stars in our Galaxy. In fact 90-95% of the stars in the solar vicinity (r ≤ 25 pc) are main sequence stars, and at least 80% of them are M dwarfs. Outside the main sequence (MS) one can find a few subdwarfs and somewhat more white dwarfs, but the contribution of this latter type is not known precisely. Estimates range between 4% and 8%. The relative frequency of giants and supergiants can not be determined from the census of the local star population because they are not represented in a statistically meaningful number. But investigations of much greater cosmic volumes demonstrate that luminosity classes Ia, Ib, II, III, and IV altogether contribute fewer than 1% of the stellar content of the Galaxy.

We have constructed a detailed model for the infrared (IR) point source sky that comprises geometrically and physically realistic representations of the galactic disk, bulge, stellar halo, spiral arms (including the “local arm”), molecular ring, and the extragalactic sky. We represent each of the distinct galactic components by up to 87 types of galactic source, each fully characterized by scale heights, space densities, and absolute magnitudes at BVJHK, 12, and 25 μm, and a spectrum from the IRAS Low Resolution Spectrometer (LRS). Our model has been guided by a parallel Monte Carlo simulation of the Galaxy at 12 μm. The content of our galactic source table constitutes an excellent match to the 12 μm luminosity function in the simulation, as well as to the luminosity functions at V and K. We are able to predict differential and cumulative IR source counts for any bandpass lying fully within the IRAS LRS range (7.7–22.7 μm) as well as for the IRAS 12 and 25 μm bands. These source counts match the IRAS observations extremely well. The model can be used to predict the character of the point source sky expected for observations from future IR space experiments (e.g., ISO, SIRTF, LDR).

One of the components of the galactic infrared background (GIRB) radiation is emission by warm dust grains heated by OB stars embedded in molecular clouds. The main contributors are compact HII regions and comparatively radioquiet infrared (IR) point sources such as the Becklin-Neugebauer object. We present the average energy distribution between 1 and 1300 μm for a sample of BN-type objects. The average color temperature between 60 and 100 μm is ~40 K, which is very similar to the color temperature of the observed warm galactic dust emission (WGDE).

A new calibration has been developed for GEISHA (Groningen Exportable Infrared System for High-Resolution Analysis) employing the zodiacal emission that is clearly present in each long scan. We have assumed, because of several other fundamental IRAS studies, that the zodiacal emission as a whole (ZEM) can be approximated by the emission of a tilted oblate ellipsoid emitting at a certain effective brightness. Besides this zodiacal emission, there is always a contribution of the galactic emission (GEM). As a standard candle, we take the sum of ZEM and GEM. We allow the electronic baseline to change linearly over a scan. We assume that the slow exponential decay that occurs will be sufficiently approximated by a straight line. The nonlinear behavior of the gain is corrected by a model developed in Groningen.

OH/IR stars and carbon stars are examples of evolved stars known to lose mass rapidly. In a recent paper (Rose 1987) we have described calculations of infrared continuum radiation from dust in stellar winds from red giants. We compare calculated radiative models with recent infrared measurements and estimate the total contribution of evolved stars to the mid-infrared background radiation observed from the Milky Way and M31.