van der Schaar, M., André, M. Demonstration of a Standard Method for Sperm Whale Coda Clustering Proceedings of the Sea Tech Week, Brest, France, Oct 2006 Abstract: Sperm whales produce, amongst other acoustic signals, short repetitive patterns of two or more pulses called codas. These codas have been recorded primarily when the whales are gathering at the surface and are thought to play a social role. To be able to study codas, and compare different results, it is first necessary to have a way to classify them. Therefore a naming protocol for normalized coda clusters is proposed, accurately describing the rhythm of a coda. Instead of counting the individual pulses, as is done with current methods, it focusses on the inter pulse intervals following a fixed protocol to decide whether consecutive intervals are similar or not. Additionally, a clustering approach is explained that allows clustering of codas when they are not spherically distributed. The method uses the k-means algorithm and, following g-means, it accepts clusters when they resemble a normal distribution in the direction of their principal component. The test for normality is performed using the Anderson-Darling statistic. Both the naming method and the clustering algorithm are demonstrated using codas from data recorded in the Canary Islands Project: Noise pollution effects on cetaceans