Abstract
A complete count of 1379 ranges on one tree was made. Three different methods of selecting branches and three different branch sizes were employed in a sampling study. Selecting branches with probabilities proportional to their cross-sectional areas was found to be more efficient than equal probability sampling.