This report on model implementation during the third year (1971-72) of the Head Start Planned Variation experiment concentrates on the development of measures of implementation. Measures are examined individually in terms of design and use, strengths and weaknesses, and findings obtained, and agreement among raters on the extent of implementation is discussed. A comparative analysis of consultant checklists, sponsor checklists, sponsor overall ratings, teacher perceptions, and sponsor ratings is presented. Proposals to improve implementation measures include a contractual agreement between evaluators and those responsible for treatment implementation, field testing of implementation measures prior to their use, and systematic training of raters. The report concludes that substantial problems remain with implementation measurement, that some models may never fit into a quantified, easily observable measure, and that there may always be an element of subjectivity in assessing the extent to which a program is implemented. Appendices comprise site to site variation in levels of implementation and the 1971-72 implementation measures. (GO)