The authors present a generalization of a 2-4 tree (also known as a symmetric binary B-tree and as a 2-3-4 tree): a 2^k-2^k+1 binary tree. This generalization allows a balanced n-node tree to be balanced as a factor of k, having a height of at most (1+1/k)lg(n+1). Thus searches can be made arbitrarily close to optimal by increasing k, with a concomitant increase in the cost of updates. To change the pseudo-node into a binary tree, the authors suggest the use of either a single bit as in a red-black tree, or counts of the maximum and minimum depths of the pseudo-node stored in each node. The advantage of their data structure is the need for only a few restructurings per update. They point out that it is possible to vary k in the data structure for improved performance. The explanations are clear, and the paper is complete with descriptive algorithms.