In order to measure ice deformation at frequent intervals across the Brunt Ice Shelf, Antarctica, 64 markers were planted to form a network of overlapping braced quadrilaterals. This network was triangulated in 1966 and, after a 10-month interval, again in 1967. Throughout the periods of observation the survey figures suffered continual distortion, and any computation method must allow for this. Making the assumption that ice movement at each marker is linear it is possible to interpolate the positions of observed stations and to process the triangulation scheme as a series of resections. To solve these a computer-oriented resection solution was developed, and adapted to accept measured lengths in addition to angles as input data. In order to correct for Earth-curvature, and to utilise observed true azimuths, a simple spheroid model was adopted. The survey yielded positions of all markers at the times of occupation; position accuracy relative to the origin of the scheme was 1 : 40,000. Values of ice velocity deduced from these positions ranged from 50 metres per year on the grounded inland ice sheet to 350m yr−1 at the front of the floating ice shelf.