This work is a preliminary study of methods to quantify alliteration. Ten pieces made up of poetry and prose (literary and non-literary) were used to create test sets. Three forms of each test set were examined: texts transcribed in IPA notational equivalent, in Chomsky and Halle features, and in Fromkin and Rodman features. Tests included the deletion of vowels, the weighting of the initial sounds, and the weighting of types according to their frequency in the population of the set. The various configurations were analyzed using a gap-recurrence method. Rankings were obtained by combining measures both of high frequency and of clustering properties. The resulting rankings compare not unfavorably with an intuitive ranking.