Abstract-Ultra-wideband (UWB) systems have emerged as a possible solution for future wireless in-body communications. However, in-body channel characterization is complex. Animal experimentation is usually restricted. Furthermore, software simulations can be expensive and imply a high computational cost. Synthetic chemical solutions, known as phantoms, can be used to solve this issue. However, achieving a reliable UWB phantom can be challenging since UWB systems use a large bandwidth and the relative permittivity of human tissues are frequency dependent. In this paper, a measurement campaign within 3.1-8.5 GHz using a new UWB phantom is performed. Currently, this phantom achieves the best known approximation to the permittivity of human muscle in the whole UWB band. Measurements were performed in different spatial positions, in order to also investigate the diversity of the in-body channel in the spatial domain. Two experimental in-body to in-body (IB2IB) and in-body to on-body (IB2OB) scenarios are considered. From the measurements, new path loss models are obtained. Besides, the correlation in transmission and reception is computed for both scenarios. Our results show a highly uncorrelated channel in transmission for the IB2IB scenario at all locations. Nevertheless, for the IB2OB scenario, the correlation varies depending on the position of the receiver and transmitter.Index Terms-Body area network, in-body communications, path loss, ultra-wideband (UWB) phantom, uncorrelated channel.