This illustration shows a hierarchical tree structure of how an index header block branches to root nodes and leaf nodes. From the header block, two brances point to instance one on the left, and instance two on the right. Instance one holds transactions for all records in the range "A" through "M", and Instance two hold transactions for all records in the range "N" through "Z". Each root node for each instance branches again to a leaf node to further subdivide the leaf node coverage of the transaction records. For example, the root node in instance one branches to one leaf node that holds transaction records "A" through "H", and another branches to a leaf node holding records "I" through "M", and so on for instance two's leaf nodes that hold transactions "N" through "S" and "T" through "Z". The bottom of the hierarchy shows unlabelled shaded partitions that house the actual records.