A reliable and accurate research is necessary to appraise the effectiveness of technical media in instruction. The same goes for the use of boxes in foreign language teaching, which are being more and more used as a means of making a greater number of pupils active simultaneously. For the purpose of finding out the effectiveness of language teaching boxes (the teacher is connected with the pupils’ boxes by a system of microphones and earphones, all the pupils can hear the teacher, but not one another, the teacher being able to listen to any of the pupils) the Brno Technical College carried out the first experiment using two sets of similar texts (one for experimenting, the other for testing) with one and the same group of students taking part in the experiment. In this way 167 comparisons were made showing the difference in the number of mistakes between the pupils taught in boxes and those taught without them. The result was in favour of the boxes, the number of mistakes being lower by one third. More accurate results were achieved in the second experiment by using the rotation method of research. Besides the two sets of texts for experimenting and testing (A and B), we also worked with two groups of students, one for experimenting, the other for testing (I and II), using two different methods of instruction both with and without boxes. We carried out rotations between the respective factors in the experiment so as to eliminate any possible distortion of results due to the inequality of texts and the differences between the groups. Group I worked first in boxes with text A, but Group II with text B; using the conventional method without boxes, Group I worked with text B and Group II with text A. In evaluating the results stress was laid not on arithmetical averages of the number of mistakes in each group as a whole, but on comparisons in pairs. Seventy such comparisons in pairs were made altogether, which showed that the method of teaching in boxes produced 52.8% favourable results, 32.8% unfavourable results, 14.2% results showing no difference. Out of sixteen groups (according to the abilities of the students taking part in the experiment) the results of ten groups were in favour of the boxes, five groups against, with one pair showing no difference. The method of teaching in boxes proved to be the most suitable for those students whose general achievement in the language was marked ≫Grade III≪ (pass), then for those marked ≫Grade 1≪ (very good) and ≫Grade IV≪ (fail), the least suitable being for those marked ≫Grade II≪ (good).