5 Summary and conclusions

We have discussed the possible uncertainties and proposed solutions for the interpretation of the kinematics of the extended gas in HzRG using the optical (UV rest frame) and NIR (optical rest frame) emission lines.

• The spectral decomposition of the emission lines will show multiple components. By studying the kinematic, flux and spatial properties of the individual components we can disentangle the mechanisms responsible for the extreme motions;

• The presence of several kinematic components (or absorption in the case of Ly$\alpha$) may produce apparent rotation curves;

• Ly$\alpha$ aborption by neutral gas/dust can change dramatically the emission line profiles. CIV$\lambda$1550 is also resonant, but less sensitive. The detection of non resonant lines with good S/N is important to understand the effects of resonant scattering;

• When scattered broad lines are present, they do not affect the emission line profiles of the brightest lines. However, they must be taken into account when studying different kinematic components, otherwise high velocity motions in the extended gas could be inferred. HeII$\lambda$1640 and the forbidden optical lines can help to determine the contribution of scattered light;

• The doublets (CIV, CIII], [OII]) are noticeably broader than simple lines (like HeII) when the gas motions have dispersion velocities similar to the separation between the doublet components;
• NIR observations are better suited for kinematic studies of HzRG since they map optical rest frame lines that are less sensitive to dust and the effects mentioned above. An additional advantage is that this spectral range allows the comparison with studies at low redshift so that we can use powerful diagnostics tools developed for nearby objects. However, the atmosphere sets limitations to this type of observations.

Acknowledgements

The authors thank the members of the "z2p5'' collaboration A. Cimatti, M. Cohen, B. Fosbury and B. Goodrich for the data presented in this paper for illustrative purposes. We thank Phil Lucas for useful discussions and Callum McCormick for providing Fig. 6.