Edge Localised Modes (ELMs) are bursts of instabilities which deteriorate the confinement of H mode plasmas and can cause damage to the divertor of next generation of devices. On JET individual discharges can exhibit hundreds of ELMs but typically in the literature, mainly due to the lack of automatic analysis tools, single papers investigate only the behaviour of tens of individual ELMs. In this paper, an original tool, the Universal Event Locator (UMEL), is applied to the problem of automatically indentifying the time location of ELMs. With this approach, databases of hundreds of thousands of ELMs can be built with reasonable effort. The analysis has then been focussed on the investigation of the statistical distribution of the inter ELM intervals at steady state for type-I ELMs. Numerous probability distributions have been tested to address the data analysis and different distributions provided a best fit for sets of data from different experiments. This result constitutes robust experimental evidence that Type-I ELMs are not all the same type of instability. Moreover, the most likely distributions are not memoryless meaning that the waiting time, from a particular instant until the next ELM, does depend on the time elapsed from the previous event. These properties pose important new constraints on the models aimed at describing the ELM dynamics. This work also demonstrates the widespread applicability of the UMEL tool.