Vertical High Field Side (VHFS) pellet injection into H-mode plasmas on JET has been studied using a range of diagnostics providing data with sampling rates up to one MHz. The primary aim of this study is to identify threshold parameters required for a deuterium pellet to trigger an ELM in JET H-mode plasmas. By using signals with fine time resolution, unambiguous temporal correlations between pellets entering the plasma from the VHFS and ELM occurrence can be made. Once this threshold has been determined, an informed strategy can be developed to provide a mechanism for controlling the ELM frequency on ITER through the injection of a stream of pellets ­ a technique called pellet-ELM pacing. If the pellets are small enough then this could be implemented without increasing the plasma density. Da emission from the pellet ablation cloud as it enters the plasma from the VHFS is recorded at 250kHz or 1MHz. Correlations with measurements recorded at the same rate of changes in edge Te, edge SXR emission, and global Da can then be made. Changes in plasma density around the plasma edge are also measured at frequencies up to 100kHz. Preliminary results from these studies show for the first time in any tokamak that there is a threshold in the size of pellet required to trigger an ELM. In JET H-modes it is relatively small, typically 1mm3 or ~2 1019D. This value holds at least for isolated ELM triggering. These small pellets being only a few percent of the size of a fuelling pellet are not found to contribute to any net increase in plasma density. Whether independent control of ELM frequency and plasma density in ITER, through the corresponding injection of pacing and fuelling sized pellets would be possible, largely depends on how the pacing size threshold scales with triggered-ELM frequency, edge barrier parameters and entry port. This is to be studied in future JET operations. Estimates of pacing-pellet penetration into the edge barrier, their role in triggering ELM precursors and a comparison with ELM triggering by fuelling sized pellets will also be presented. Given the small size required for a pacing pellet it is proposed that noncryogenic materials other than D2 could be considered for this purpose.