Photonic lanterns are an important enabling technology for astrophotonics with a wide range of potential applications including fibre Bragg grating OH suppression, integrated photonic spectrographs and fibre scramblers for high resolution spectroscopy. The behaviour of photonic lanterns differs in several important respects from the conventional fibre systems more frequently used in astronomical instruments and a detailed understanding of this behaviour is required in order to make the most effective use of this promising technology. To this end we have undertaken a laboratory study of photonic lanterns with the aim of developing an empirical model for the mapping from input to output illumination distributions. We have measured overall transmission and near field output light distributions as a function of input angle of incidence for photonic lanterns with between 19 and 61 cores. We present the results of this work, highlight the key differences between photonic lanterns and conventional fibres, and illustrate the implications for instrument design via a case study, the design of the PRAXIS spectrograph. The empirical photonic lantern model was incorporated into an end-to-end PRAXIS performance model which was used to optimise the design parameters of the instrument. We describe the methods used and the resulting conclusions. The details of photonic lantern behaviour proved particularly important in selecting the optimum on sky field of view per fibre and in modelling of the instrument thermal background.