For a long time it has been usual to search for the objects that host the gas from which absorption line systems arise. However, the spatial distribution of metals around galaxies and more generally in the intergalactic medium is complex, and it is often very difficult to associate one absorption system with one galaxy. Except for possibly in the special case of DLA systems, it may be more appropriate to discuss the clustering properties of different classes of objects and to measure their correlation functions. I illustrate this approach with three examples: the distribution of metals around galaxies at intermediate redshift, the modelling of the clustering of C IV systems, and the determination of the transverse correlation function of the Lyman-$\alpha$ forest.