Introduction

Blazars are the most extreme variety of AGN known. Their signal properties include irregular, rapid variability; high optical polarization; core-dominant radio morphology; apparent superluminal motion; flat ($\alpha_{\rm r} < 0.5$) radio spectra; and a broad continuum extending from the radio through the gamma-rays (these properties are reviewed in detail by Urry & Padovani 1995). The broadband emission from blazars is dominated by non-thermal processes (most likely synchrotron and inverse-Compton radiation), likely emitted by a relativistic jet pointed close to our line of sight (as originally proposed by Blandford & Rees in 1978). The beaming hypothesis has been successful in reproducing the luminosity function of samples of blazars (e.g., Padovani & Urry 1990, 1991, 1992; Urry, Padovani & Stickel 1991), allowing the derivation of class properties such as the range of Lorentz factors $\Gamma$and opening angles in the radio band (and also in other bands for BL Lacs). However, the small size of these samples (30-50 objects) has prevented detailed modeling of the luminosity function, especially at low powers, yielding considerable uncertainties on the derived parameters.

Due to the rarity and low space-density of blazars, which make up considerably less than 5% of all AGN (Padovani 1997), ``pencil-beam'' surveys of the type carried out by, e.g., Castander et al. (1996), are poorly suited for finding blazars. As has already been noted by Perlman et al. (1996a), wide-angle surveys with appropriately restrictive search parameters are much more efficacious. Indeed, it is thanks to the advent of such surveys that complete samples of blazars exist today. With the advent of modern archival techniques and deeper wide-angle surveys, it is now possible to combine survey techniques and develop surveys which can sample the blazar luminosity function deeply and representatively. Several such projects are currently underway (§ 3).

This paper describes the DXRBS blazar survey and its goals, and presents our first 85 firm identifications. The main result presented here is that previous blazar samples are not representative of the blazar class, missing approximately half of the FSRQ population and a somewhat smaller part of the BL Lac population. The newly identified DXRBS blazars expand the range of L_X/L_R values found among blazars with emission lines by an order of magnitude, and for the first time samples the low-luminosity end of the luminosity function of FSRQs with reasonable statistics. In Section 2, we describe the methods used to find blazar candidates and prepare for optical observations. Section 3 contains a detailed discussion on the subject of classification of flat-spectrum radio sources. Section 4 describes the results of our optical spectroscopy and the makeup of the DXRBS blazar sample as of April 1997. Section 5 discusses the redshift distribution of both the BL Lac and FSRQ subsamples. Section 6 contains a discussion of the properties of the sample, and the implications of these first results for unified schemes and upon our picture of the blazar class. In Section 7, we discuss the topic of selection effects. The conclusions of this work are summarized in Section 8.