1 Introduction

High redshift radio galaxies ($z\geq0.7$, HzRG) show optical regions of ionized gas that extend across tens (sometimes hundreds) of kiloparsecs. These structures are aligned with the radio axis (McCarthy et al. 1987; Chambers et al. 1987) and show very irregular and clumpy morphologies. The kinematics is extreme compared to the low redshift counterparts. Line widths (Ly$\alpha$ or [OII]$\lambda$3727) as broad as $FWHM\geq1000$ km s^-1are often measured (McCarthy et al. 1996; Villar-Martín et al. 1999; Baum & McCarthy 2000) versus $\sim300-400$ km s^-1 commonly observed at low redshift (e.g. Tadhunter et al. 1989). Velocity dispersions range from several hundreds up to 1600 km s^-1 and there are velocity shifts between different lines of up to $\sim 1000$ km s^-1 (Röttgering et al. 1997). Baum & McCarthy (2000) showed that the transition from predominantly quiescent systems to those with extreme motions occurs near $z\sim 0.6$.

While at low redshift the kinematics of the extended gas in radio galaxies is generally explained in terms of gravitational motions, the mechanism responsible for the extreme motions at higher redshifts is not well understood. The apparent connection between the size of the radio source and the emission line kinematics observed at $z\geq 2$ (van Ojik et al. 1997) suggests interactions between the radio and optical structures that perturb the kinematics. This is supported by the discovery in some HzRG of haloes of ionized gas extending beyond the radio structures that emit narrow Ly$\alpha$ ($\sim 250$ km s^-1) compared to the inner regions ($\sim1200$ km s^-1) (van Ojik et al. 1996).

Baum & McCarthy (2000) studied a much larger sample covering a wider range in redshift. They find no (or very weak) correlation between the emission line kinematics and the ratio of the radio to nebular size. The authors support a gravitational origin for the kinematics of HzRG. Bipolar outflows (possible consequence of a circumnuclear starburst) have also been proposed by some authors (e.g. Chambers 1998; Taniguchi & Shioya 2000).

Understanding the nature of the kinematics in HzRG will help to answer some open questions related to these galaxies:

• What is the nature of the alignment effect? Scattered light from the hidden AGN is an important contributor (e.g. Cimatti et al. 1997, 1998; Fosbury et al. 1999). However, the likely influence of radio jet/gas interactions is still not understood;

• The ionization processes in the extended gas of HzRG is clearly related to the nuclear activity, but what is the main mechanism: jet driven shocks or AGN illumination?

• How does the radio source interact with its environment and how does it influence the subsequent evolution of the system?

• Are very distant radio galaxies ($z\geq 2$) in proto clusters?

To date, the kinematic studies of the gas in HzRG have been done in the optical. The main limitation has been the need for a large collecting area to detect the emission lines with high signal/noise ratio (S/N) in the extended gas. With 3-4 m telescopes, only [OII]$\lambda$3727 (z<1.2) or Ly$\alpha$ ($z\geq 2$) could be used. At ($z\geq 2$) the uncertainties are important, since Ly$\alpha$ is highly sensitive to dust/neutral gas absorption. The strong optical (rest frame) emission lines ([OIII]$\lambda$5007, 4959, [OII]$\lambda$3727) are more reliable. However, they are redshifted into the NIR and the need for a large collecting area has constrained the NIR spectroscopic studies of HzRG (and quasars) to the spatially integrated properties (e.g. Jackson & Rawlings 1997; McIntosh et al. 1999; Larkin et al. 2000)

This is the epoch of the 8-10 m telescope generation. We are in the position for the first time to study the kinematics of the extended gas in HzRG:

• using UV rest frame lines (redshifted into the optical) other than Ly$\alpha$ such as HeII, CIV or CIII] (see Fig. 1);

• using the optical rest frame lines such us [OIII]$\lambda$5007, 4949, H$\alpha$, etc. (redshifted into the NIR). The development of NIR arrays (and large collecting areas) makes this study possible for the first time.

In spite of the opportunities opened by the new technological facilities, the study of the kinematic properties of HzRG is still complex. The goal of this paper is to assess the main sources of uncertainty and provide solutions to solve them.

Figure 1: KeckII+LRISp spectra of the HzRG B3 0731+438 (z=2.42). The new generation of 8-10 m telescopes allows for the first time the detection of emission lines other than Ly$\alpha$ in the extended gas of very distant radio galaxies ($z\geq 2$) with high S/N