The experiment described in this bulletin is the result of a previous study of experimental examinations in mathematics (Bulletin No. 2). A basic multifacet test was constructed with 21 mathematical situations in arithmetic, algebra, and geometry and presented in three separate forms: Type 1 Traditional; 2 multi-choice (with four choices); and 3 True/False/Don't Know. Candidates were some 917 sixteen year old students from 27 secondary schools. The students were separated into 3 groups with each group given one type of test. Analysis of the results indicate that the traditional form was the hardest and the true/false/don't know the easiest. The reliability and validity of each form were found high but there was evidence that they were highest for Type 1 and lowest for Type 3. The item analysis indicated that a shorter test with fewer situations might serve adequately. Of major concern was the assessing of each form of the test which operated at a different level of difficulty. Evidence suggests that Type 1 might have measured aptitude rather than attainment; Type 2 measured attainment rather than aptitude; and Type 3 measured mathematical abilities unevenly over the range of candidates. The appendices describe the structure of the experiment and evidence in support of the conclusions. (LS)