Summary.Experimental tests in spoken English, comprising a ten‐minute test for pupils inprose readingandconversationand a twenty‐minute test for students in these two items and inpoetry‐speakingandspeech making, took place in six training colleges and six secondary schools (173 pupils, 90 students and 15 raters taking part).In one type of test each candidate was assessed on a single occasion by several raters, in another on two occasions, once by one rater and once by another, with ‘equal and parallel’ tests. Performances were rated on various aspects, and on ageneral impression, of each item and on atotal impressionof the whole.The results suggest: (1) That the types of speech test investigated have a reasonably high reliability and a reasonably high validity: i.e., that they are as valid and reliable as essay‐type written examinations. (2) That these tests appear to be well‐founded (i.e., that they comprise elements on which satisfactory judgments can be made). (3) That fatigue does not affect the reliability of rate