Least squares adjustment of measurement data is a computationally expensive process that is used frequently throughout the field of Geodesy. Helmert Blocking is a known method of reducing the computational cost for the adjustment of geodetic networks. However, in order to apply the Helmert Blocking method, one is required to segment the network into a hierarchy of blocks that follow a strict set of rules. Here we introduce a new data structure, the HB Tree, which maintains an existing hierarchy of blocks as additional sets of measurements are added and can be used to automatically segment a network to facilitate Helmert Blocking. We have implemented both our data structure and the Helmert Blocking Adjustment method and have tested this implementation using real geodetic control network data and simulated land survey data. Our results indicate that the processing times for a Helmert Blocking Adjustment, using our data structure for automatic segmentation, are a significant improvement over the processing times for the standard adjustment method used in commercial network adjustment software. The performance of our implementation depends on the structure of the network and the structure of the input data, but in our largest trial we adjusted a simulated network with over 1,400,000 stations in under 6 minutes on a personal computer. We do not believe that there is any network adjustment software that uses the standard adjustment method that is able to adjust a network of this size in a practical time frame.