2. EARLY HISTORY

The existence of a diffuse XRB was discovered more than thirty years ago (Giacconi et al. 1962). Figure 1 shows data from the discovery flight. It is interesting to note that in these data both the diffuse emission and a strong source (i.e. the two elements which became the basis for the two main hypotheses for the production of the XRB) are already present. The first important step with respect to our knowledge of the XRB has been made with the first all-sky surveys (UHURU and ARIEL V) at the beginning of the seventies. The high degree of isotropy revealed by these surveys led immediately to realize that the origin of the XRB has to be mainly extragalactic. Moreover, under the discrete source hypothesis, the number of sources contributing to the XRB has to be very large (N > 10⁶ sr^-1; Schwartz 1980).

Figure 1. Data from the rocket flight in which the XRB was discovered.

In the same years a number of experiments were set up to measure the spectrum of the XRB over a large range of energy. It was found that over the energy range 3-1000 keV the XRB spectrum is reasonably well fitted with two power laws with slopes ₁ ~ 0.4 for E 25 keV and ₂ ~ 1.4 for E > 25 keV (see Figure 1 in Tanaka 1992).

At the beginning of the eighties two different sets of measurements led additional fire to the debate between supporters of the discrete source and diffuse hypotheses. On the one hand, the excellent HEAO-1 data showed that in the energy range 3-50 keV the shape of the XRB is very well fitted by an isothermal bremsstrahlung model corresponding to an optically thin, hot plasma with kT of the order of 40 keV (Marshall et al. 1980). Moreover, it was shown by Mushotzky (1984) that essentially all the Seyfert 1 galaxies with reliable 2-20 keV spectra (~ 30 objects, mostly from HEAO-1 data) were well fitted by a single power law with an average spectral index of the order of 0.65, significantly different from the slope of the XRB in the same energy range. These two observational facts were taken as clear "evidences" in favor of the diffuse thermal hypothesis. On the other hand, the results of the EINSTEIN deep surveys showed that about 20% of the soft XRB (1-3 keV) are resolved into discrete sources at fluxes of the order of a few x 10^-14erg cm^-2s^-1 (Giacconi et al. 1979, Griffiths et al. 1983, Primini et al. 1991, Hamilton et al. 1991). A large fraction of these faint X-ray sources have been identified with Active Galactic Nuclei (AGNs). Because of the difference between the spectra of the XRB and those of the few bright AGNs with good spectral data, the supporters of the diffuse, hot plasma hypothesis had to play down as much as possible the contribution of AGNs to the XRB to a limit which was close to be in conflict with an even mild extrapolation of the observed log N - log S. Actually, a number of papers were published in which it was "demonstrated" that even in the soft X-ray band AGNs could not contribute much more than what had already been detected at the EINSTEIN limit.

At that time I personally think that there were already evidences (for those who wanted to see them...) that the diffuse thermal emission as main contributor to the background was not tenable (see, for example, Setti 1985). Very simple arguments in this direction were given by Giacconi and Zamorani (1987). On the basis of reasonable extrapolations of the X-ray properties and the optical counts of known extragalactic X-ray sources (mainly AGNs and galaxies), they concluded that it is unlikely that their contribution to the soft X-ray background is smaller than 50%. Given this constraint, they then discussed two possibilities:

1. either faint AGNs have the so-called (at that time) "canonical" spectrum observed for brighter AGNs. In this case the residual XRB (i.e. the spectrum resulting after subtraction of the contribution from known sources) would not be fitted anymore by optically thin bremsstrahlung;

2. or spectral evolution for AGNs is allowed. In this case, in order not to destroy the excellent thermal fit in the 3-50 keV data, diffuse emission could still be accommodated only if discrete sources have essentially the same spectrum as the XRB. On this basis, they concluded that "since in this scenario we would already require that the average spectrum of faint sources yielding 50% of the soft XRB is essentially the same as the observed XRB, there is nothing that prevents us from concluding that the entire background may well be due to the same class of discrete sources, at even fainter fluxes."

In other words, reversing the usual line of thought, the excellent thermal fit of the 3-50 keV XRB spectrum was shown by these arguments to be a point in favor of the discrete source hypothesis, rather than of the hot gas hypothesis! These conclusions, however, were not well received in a large fraction of the X-ray community; probably, they had the defect of being too simple and direct...

Thus, the debate between the supporters of the two hypotheses continued, until the final resolution of the controversy came from the incredibly neat results obtained with the FIRAS instrument on board COBE: the absence of any detectable deviation from a pure black body of the cosmic microwave background set an upper limit to the comptonization parameter y < 10^-3 (Mather et al. 1990), more than ten times smaller than the value required by the hot intergalactic gas model. The most recent upper limit for the comptonization parameter is now y < 2.5 x 10^-5 (Mather et al. 1993). Discussing these data, Wright et al. (1993) conclude that a uniform, hot intergalactic gas produces at most 10^-4 of the observed XRB!