The structure of the Tanim Hed (Amb Kanant), a courtship dance of the Medlpa, New Guinea, have been described and each of the recurring motor patterns (Cycle) broken down into 2 parts of 4 sections each. The dance for each pair of dancers seen was measured in terms of the duration of these sections to 1/25 of a second (one frame of a 25 f.p.s. film). On analysis of the relationships between the dancers the following points emerged: 1) The variation in speed (duration) for one dance cycle (and also for the sections) in each pair of dancers was very small. The mean duration was around 4 seconds with a 1-3/25 second standard deviation. 2) The duration seemed either not at all or very complexely controlled by the music, as i) the differences between pairs was large but within pairs very small and ii) in the same scene different pairs of dancers could be seen dancing with different rhythms. Dancing, however, only took place when the choir was singing, and thus probably serves as a basic stimulus and to control the rhythm of the dance within certain limits. 3) The mean duration seemed to be derived, within one pair, as a compromise between the individual dancers. 3 scenes from different pairs analysed over the first 35 cycles showed a marked logarithmic relation to time, gradually decreasing in speed with an obvious oscillation as the pair sought the compromise value. 4) The 2 parts of the dance, Turn and Bow, showed an unequal relation to the Cycle duration. As this shortened the Turn was shortened more than the Bow, thus showing the greatest lability. The former was also the part with the greater variability within one pair and lack of synchrony between the partners. 5) When the mean duration of the successive sections was plotted out as a graph, the form of this pattern showed the greatest similarity when the patterns of the male pair (i.e. the same man dancing with 2 different girls) were compared. Thus the man decides the exact form of the dance, in relation to the speed decided upon by the pair. 6) The beginning and end of the dance were more variable. The pre-dance was initiated by the man, the girl joining later but dancing to a more constant rhythm. Her speed correlated at 1.0 with the speed seen in the dance, within the one pair only. The end was more indefinite and occurred as a simple parting and only within certain sections (moving up from Bow Low or Forward from Turn Back). 7) Thus the different parts of the dance are seen to be controlled differentially between the partners, with the man initiating the dance and controlling the form, the girl controlling the fine variation in speed within the pair, and both contributing to the mean dance speed, which then has all influence on the form.