Such a hypothesis was apparently confirmed by the first 1.2mm detection in February 1993 of BR1202-07, the brightest object of the sample and the third highest red shift known () (Mc Mahon et al. 1994). Its subsequent detection at JCMT at submillimeter wavelengths confirmed a very large spectral index (3.5) characteristic of dust (Isaak et al. 1994). However, during the last two years, the continuation of our programme had been repeatedly plagued by bad weather, until its decisive completion during our last run in early February 1995 with the remarkable MPIfR 7-channel bolometer array which allows a very efficient subtraction of the remaining sky background. The results, together with those of Chini and Krügel (1994), prove that the era of systematic millimeter studies at high z is already opened with the unique capacities provided by the 30m with bolometer arrays.
Five radio quiet QSOs with z>4 have been detected at a 5--sigma level in addition to the initial detection of BR1202-07 reported in McMahon et al. 1994. Their fluxes range from 2.5 to 10 mJy. All the detections were independently detected during at least three different nights. In addition 10 other sources of the Cambridge APM survey sample were searched for but not detected with rms levels 1.5 mJy i.e. with fluxes probably smaller than 3 mJy. We have also observed the QSO, PC2132+0126 and find a flux of 1.00.8 mJy, whereas Andreani et al. (1994) reported a flux of 11.51.7 mJy.
Accordingly, we have shown that a large fraction of the color identified z>4 radio quiet QSOs have a strong millimeter emission detectable with the 30 meter. Although the following point remains to be verified with the JCMT, it is very likely that the detected sources have a large spectral index characteristic of dust emission, similar to what is observed for BR1202-07. The very large amount of dust implied, 10, is comparable to what is found in the most luminous infrared starburst galaxies. It is quite possible that a giant starburst is taking place or has recently taken place in the host galaxies of these QSOs at a scale the central kiloparsec. However, it is possible that most of the heating of the observed dust is provided by the UV emitted by the central engine of the quasar.
It is crucial to complement our results by submillimeter observations at the JCMT and by sensitive far--infrared photometry with ISO in order to characterize the dust temperature and derive the far--infrared luminosity. In addition, it is important to extend our work at z>4 to lower red shifts i.e. 1 to 3, in order to determine the frequency among them of detectable millimeter emission and investigate the relationship between this emission in QSOs with the dust emission in ultraluminous IRAS galaxies and radio galaxies.