It should be fairly clear that we are likely to obtain clusters of sequentially occupied locations, if we use linear probing. Moreover, the larger a cluster becomes, the more likely we are to hit it when we come to insert a new item. This item will then be placed at the end of the cluster, making it larger still. This phenomenon is called primary clustering. Furthermore the larger a cluster becomes, the more comparisons we shall have to make on average to reach its end, where the insertion is to be made. The topic of primary clustering is discussed quantitatively in Chapter 20 of Weiss, Data Structures & Problem Solving Using Java.