2004MNRAS.348..937C

We discuss the properties of the sources in the CLASS blazar survey, which aims at the selection of low radio power (P_5GHz< 10²⁵ W/Hz) blazars. We use Very Large Array (VLA) data from available catalogues and from our own observations to constrain the radio core-dominance of the sample which, together with the flat radio spectral index, is a signature of the blazar activity. X-ray data from the ROSAT All Sky Survey were also collected in order to constrain the radio-to-X-ray luminosity ratio (α_RX) of the sources. The data analysis shows that more than 30 per cent of sources at low radio power (P_5GHz< 10²⁵ W/Hz) have an α_RXsteeper than that expected in the framework of the `blazar sequence' recently put forward to unify the high- and low-power blazars. The possibility that this result is influenced by contaminating sources in the current sample is discussed. The conclusion is that, even if a number of non-blazars (typically CSO/GPS sources) are expected in the survey, it is unlikely that this constitutes the sole reason for the observed deviation. In particular, we show two examples for which the blazar nature is confirmed from VLBI data and for which the steep α<sub>RX</sub>(suggesting a synchrotron peak frequency below 10<sup>15.5</sup> Hz) and the low radio power (0.6-2x10<sup>24</sup> W/Hz) put these sources outside the `blazar sequence'. The results presented here show the importance of a correct and unbiased sampling of the low-power regime of the blazar population.